feat: add mocking for agent data (#1028)

* wip: add mocking for agent data

* chore: add tests; fix: various fixes

* chore: fix schema generation null handling

.pre-commit-config.yaml

@@ -34,7 +34,7 @@ repos:
     rev: v1.0.1
     hooks:
       - id: mypy
-        exclude: ^py/tests|^py/unit_tests
+        exclude: ^py/tests|^py/unit_tests|^examples
         additional_dependencies:
           [
             "types-requests",

examples-ts/classify/.gitignore

@@ -0,0 +1,21 @@
+node_modules
+package-lock.json
+yarn.lock
+
+.DS_Store
+.cache
+.env
+.vercel
+.output
+.nitro
+/build/
+/api/
+/server/build
+/public/build# Sentry Config File
+.env.sentry-build-plugin
+/test-results/
+/playwright-report/
+/blob-report/
+/playwright/.cache/
+.tanstack
+.vscode

examples-ts/classify/.prettierignore

@@ -0,0 +1,4 @@
+**/build
+**/public
+pnpm-lock.yaml
+routeTree.gen.ts

examples-ts/classify/README.md

@@ -0,0 +1,88 @@
+# LlamaClassify Demo
+
+A TypeScript demo application showcasing the power of **LlamaClassify** - an agentic documents classification service from [LlamaCloud](https://cloud.llamaindex.ai). This demo allows you to classify financial documents among three different types (Cash flow statement, Income Statement and Balance Sheet).
+
+## Table of Contents
+
+- [Features](#features)
+- [Prerequisites](#prerequisites)
+- [Installation](#installation)
+- [Usage](#usage)
+  - [Start the Demo](#start-the-demo)
+- [How It Works](#how-it-works)
+- [Troubleshooting](#troubleshooting)
+  - [Common Issues](#common-issues)
+- [License](#license)
+- [Contributing](#contributing)
+
+## Features
+
+- 📄 **Documemt Classification**: Classify files based on well-defined rules you can customized and play around with.
+- 🤖 **Reasoning-based Actionable Insights**: Get in-depth, reasoning based insights on the document classification, accompanied by confidence scores.
+- 🎨 **Beautiful UI**: [DaisyUI](https://daisyui.com)-based interface powered by [TanStack](https://tanstack.com)
+- ⚡ **Fast Development**: Hot reload support with development mode
+- 🛠️ **TypeScript**: Full TypeScript support with strict type checking
+
+## Prerequisites
+
+- Node.js (version 22 or higher)
+- pnpm package manager
+- LlamaCloud API key
+
+## Installation
+
+1. Clone the repository:
+
+```bash
+git clone https://github.com/run-llama/llama_cloud_services
+cd lama_cloud_services/examples-ts/classify/
+```
+
+2. Install dependencies:
+
+```bash
+npm install
+```
+
+3. Set up your environment variables:
+
+```bash
+# Add your API key to your environment
+export LLAMA_CLOUD_API_KEY="your-llamacloud-api-key"
+```
+
+## Usage
+
+### Start the Demo
+
+```bash
+npm run dev
+```
+
+The application will be up and running on http://localhost:3000
+
+## How It Works
+
+1. **Document Input**: Enter the path to your document when prompted
+2. **Parsing**: LlamaClassify, based on the rules you can find [here](./src/utils/classifier.ts), processes the document and classifies it
+3. **Results**: The classification outcome, as well as the reasoning behind it and the confidence score, are displayed in the UI.
+
+## Troubleshooting
+
+### Common Issues
+
+1. **Module Resolution Errors**: Ensure you're using Node.js 22+ and have all dependencies installed
+2. **API Key Issues**: Verify your LlamaCloud API key is correctly set
+3. **File Path Errors**: Use absolute paths or ensure relative paths are correct from the project root
+
+## License
+
+MIT License - see the [LICENSE](../../LICENSE) file for details.
+
+## Contributing
+
+1. Fork the repository
+2. Create a feature branch
+3. Make your changes
+4. Run `npm run format` and `npm run lint`
+5. Submit a pull request

examples-ts/classify/package.json

@@ -0,0 +1,34 @@
+{
+  "name": "tanstack-start-example-basic",
+  "private": true,
+  "sideEffects": false,
+  "type": "module",
+  "scripts": {
+    "dev": "vite dev",
+    "build": "vite build && tsc --noEmit",
+    "start": "node .output/server/index.mjs"
+  },
+  "dependencies": {
+    "@tanstack/react-router": "^1.133.22",
+    "@tanstack/react-router-devtools": "^1.133.22",
+    "@tanstack/react-start": "^1.133.22",
+    "llama-cloud-services": "file:../../ts/llama_cloud_services",
+    "react": "^19.0.0",
+    "react-dom": "^19.0.0",
+    "tailwind-merge": "^2.6.0",
+    "zod": "^3.24.2"
+  },
+  "devDependencies": {
+    "@tailwindcss/postcss": "^4.1.15",
+    "@types/node": "^22.5.4",
+    "@types/react": "^19.0.8",
+    "@types/react-dom": "^19.0.3",
+    "@vitejs/plugin-react": "^4.6.0",
+    "daisyui": "^5.3.7",
+    "postcss": "^8.5.1",
+    "tailwindcss": "^4.1.15",
+    "typescript": "^5.7.2",
+    "vite": "^7.1.7",
+    "vite-tsconfig-paths": "^5.1.4"
+  }
+}

examples-ts/classify/postcss.config.mjs

@@ -0,0 +1,5 @@
+export default {
+  plugins: {
+    '@tailwindcss/postcss': {},
+  },
+}

examples-ts/classify/public/site.webmanifest

@@ -0,0 +1,19 @@
+{
+  "name": "",
+  "short_name": "",
+  "icons": [
+    {
+      "src": "/android-chrome-192x192.png",
+      "sizes": "192x192",
+      "type": "image/png"
+    },
+    {
+      "src": "/android-chrome-512x512.png",
+      "sizes": "512x512",
+      "type": "image/png"
+    }
+  ],
+  "theme_color": "#ffffff",
+  "background_color": "#ffffff",
+  "display": "standalone"
+}

examples-ts/classify/src/components/DefaultCatchBoundary.tsx

@@ -0,0 +1,53 @@
+import {
+  ErrorComponent,
+  Link,
+  rootRouteId,
+  useMatch,
+  useRouter,
+} from '@tanstack/react-router'
+import type { ErrorComponentProps } from '@tanstack/react-router'
+
+export function DefaultCatchBoundary({ error }: ErrorComponentProps) {
+  const router = useRouter()
+  const isRoot = useMatch({
+    strict: false,
+    select: (state) => state.id === rootRouteId,
+  })
+
+  console.error('DefaultCatchBoundary Error:', error)
+
+  return (
+    <div className="min-w-0 flex-1 p-4 flex flex-col items-center justify-center gap-6">
+      <ErrorComponent error={error} />
+      <div className="flex gap-2 items-center flex-wrap">
+        <button
+          onClick={() => {
+            router.invalidate()
+          }}
+          className={`px-2 py-1 bg-gray-600 dark:bg-gray-700 rounded-sm text-white uppercase font-extrabold`}
+        >
+          Try Again
+        </button>
+        {isRoot ? (
+          <Link
+            to="/"
+            className={`px-2 py-1 bg-gray-600 dark:bg-gray-700 rounded-sm text-white uppercase font-extrabold`}
+          >
+            Home
+          </Link>
+        ) : (
+          <Link
+            to="/"
+            className={`px-2 py-1 bg-gray-600 dark:bg-gray-700 rounded-sm text-white uppercase font-extrabold`}
+            onClick={(e) => {
+              e.preventDefault()
+              window.history.back()
+            }}
+          >
+            Go Back
+          </Link>
+        )}
+      </div>
+    </div>
+  )
+}

examples-ts/classify/src/components/NotFound.tsx

@@ -0,0 +1,25 @@
+import { Link } from '@tanstack/react-router'
+
+export function NotFound({ children }: { children?: any }) {
+  return (
+    <div className="space-y-2 p-2">
+      <div className="text-gray-600 dark:text-gray-400">
+        {children || <p>The page you are looking for does not exist.</p>}
+      </div>
+      <p className="flex items-center gap-2 flex-wrap">
+        <button
+          onClick={() => window.history.back()}
+          className="bg-emerald-500 text-white px-2 py-1 rounded-sm uppercase font-black text-sm"
+        >
+          Go back
+        </button>
+        <Link
+          to="/"
+          className="bg-cyan-600 text-white px-2 py-1 rounded-sm uppercase font-black text-sm"
+        >
+          Start Over
+        </Link>
+      </p>
+    </div>
+  )
+}

examples-ts/classify/src/routeTree.gen.ts

@@ -0,0 +1,225 @@
+/* eslint-disable */
+
+// @ts-nocheck
+
+// noinspection JSUnusedGlobalSymbols
+
+// This file was automatically generated by TanStack Router.
+// You should NOT make any changes in this file as it will be overwritten.
+// Additionally, you should also exclude this file from your linter and/or formatter to prevent it from being checked or modified.
+
+import { Route as rootRouteImport } from './routes/__root'
+import { Route as UsersRouteImport } from './routes/users'
+import { Route as IndexRouteImport } from './routes/index'
+import { Route as UsersIndexRouteImport } from './routes/users.index'
+import { Route as PostsIndexRouteImport } from './routes/posts.index'
+import { Route as UsersUserIdRouteImport } from './routes/users.$userId'
+import { Route as PostsPostIdRouteImport } from './routes/posts.$postId'
+import { Route as ApiClassifyRouteImport } from './routes/api/classify'
+import { Route as PostsPostIdDeepRouteImport } from './routes/posts_.$postId.deep'
+
+const UsersRoute = UsersRouteImport.update({
+  id: '/users',
+  path: '/users',
+  getParentRoute: () => rootRouteImport,
+} as any)
+const IndexRoute = IndexRouteImport.update({
+  id: '/',
+  path: '/',
+  getParentRoute: () => rootRouteImport,
+} as any)
+const UsersIndexRoute = UsersIndexRouteImport.update({
+  id: '/',
+  path: '/',
+  getParentRoute: () => UsersRoute,
+} as any)
+const PostsIndexRoute = PostsIndexRouteImport.update({
+  id: '/posts/',
+  path: '/posts/',
+  getParentRoute: () => rootRouteImport,
+} as any)
+const UsersUserIdRoute = UsersUserIdRouteImport.update({
+  id: '/$userId',
+  path: '/$userId',
+  getParentRoute: () => UsersRoute,
+} as any)
+const PostsPostIdRoute = PostsPostIdRouteImport.update({
+  id: '/posts/$postId',
+  path: '/posts/$postId',
+  getParentRoute: () => rootRouteImport,
+} as any)
+const ApiClassifyRoute = ApiClassifyRouteImport.update({
+  id: '/api/classify',
+  path: '/api/classify',
+  getParentRoute: () => rootRouteImport,
+} as any)
+const PostsPostIdDeepRoute = PostsPostIdDeepRouteImport.update({
+  id: '/posts_/$postId/deep',
+  path: '/posts/$postId/deep',
+  getParentRoute: () => rootRouteImport,
+} as any)
+
+export interface FileRoutesByFullPath {
+  '/': typeof IndexRoute
+  '/users': typeof UsersRouteWithChildren
+  '/api/classify': typeof ApiClassifyRoute
+  '/posts/$postId': typeof PostsPostIdRoute
+  '/users/$userId': typeof UsersUserIdRoute
+  '/posts': typeof PostsIndexRoute
+  '/users/': typeof UsersIndexRoute
+  '/posts/$postId/deep': typeof PostsPostIdDeepRoute
+}
+export interface FileRoutesByTo {
+  '/': typeof IndexRoute
+  '/api/classify': typeof ApiClassifyRoute
+  '/posts/$postId': typeof PostsPostIdRoute
+  '/users/$userId': typeof UsersUserIdRoute
+  '/posts': typeof PostsIndexRoute
+  '/users': typeof UsersIndexRoute
+  '/posts/$postId/deep': typeof PostsPostIdDeepRoute
+}
+export interface FileRoutesById {
+  __root__: typeof rootRouteImport
+  '/': typeof IndexRoute
+  '/users': typeof UsersRouteWithChildren
+  '/api/classify': typeof ApiClassifyRoute
+  '/posts/$postId': typeof PostsPostIdRoute
+  '/users/$userId': typeof UsersUserIdRoute
+  '/posts/': typeof PostsIndexRoute
+  '/users/': typeof UsersIndexRoute
+  '/posts_/$postId/deep': typeof PostsPostIdDeepRoute
+}
+export interface FileRouteTypes {
+  fileRoutesByFullPath: FileRoutesByFullPath
+  fullPaths:
+    | '/'
+    | '/users'
+    | '/api/classify'
+    | '/posts/$postId'
+    | '/users/$userId'
+    | '/posts'
+    | '/users/'
+    | '/posts/$postId/deep'
+  fileRoutesByTo: FileRoutesByTo
+  to:
+    | '/'
+    | '/api/classify'
+    | '/posts/$postId'
+    | '/users/$userId'
+    | '/posts'
+    | '/users'
+    | '/posts/$postId/deep'
+  id:
+    | '__root__'
+    | '/'
+    | '/users'
+    | '/api/classify'
+    | '/posts/$postId'
+    | '/users/$userId'
+    | '/posts/'
+    | '/users/'
+    | '/posts_/$postId/deep'
+  fileRoutesById: FileRoutesById
+}
+export interface RootRouteChildren {
+  IndexRoute: typeof IndexRoute
+  UsersRoute: typeof UsersRouteWithChildren
+  ApiClassifyRoute: typeof ApiClassifyRoute
+  PostsPostIdRoute: typeof PostsPostIdRoute
+  PostsIndexRoute: typeof PostsIndexRoute
+  PostsPostIdDeepRoute: typeof PostsPostIdDeepRoute
+}
+
+declare module '@tanstack/react-router' {
+  interface FileRoutesByPath {
+    '/users': {
+      id: '/users'
+      path: '/users'
+      fullPath: '/users'
+      preLoaderRoute: typeof UsersRouteImport
+      parentRoute: typeof rootRouteImport
+    }
+    '/': {
+      id: '/'
+      path: '/'
+      fullPath: '/'
+      preLoaderRoute: typeof IndexRouteImport
+      parentRoute: typeof rootRouteImport
+    }
+    '/users/': {
+      id: '/users/'
+      path: '/'
+      fullPath: '/users/'
+      preLoaderRoute: typeof UsersIndexRouteImport
+      parentRoute: typeof UsersRoute
+    }
+    '/posts/': {
+      id: '/posts/'
+      path: '/posts'
+      fullPath: '/posts'
+      preLoaderRoute: typeof PostsIndexRouteImport
+      parentRoute: typeof rootRouteImport
+    }
+    '/users/$userId': {
+      id: '/users/$userId'
+      path: '/$userId'
+      fullPath: '/users/$userId'
+      preLoaderRoute: typeof UsersUserIdRouteImport
+      parentRoute: typeof UsersRoute
+    }
+    '/posts/$postId': {
+      id: '/posts/$postId'
+      path: '/posts/$postId'
+      fullPath: '/posts/$postId'
+      preLoaderRoute: typeof PostsPostIdRouteImport
+      parentRoute: typeof rootRouteImport
+    }
+    '/api/classify': {
+      id: '/api/classify'
+      path: '/api/classify'
+      fullPath: '/api/classify'
+      preLoaderRoute: typeof ApiClassifyRouteImport
+      parentRoute: typeof rootRouteImport
+    }
+    '/posts_/$postId/deep': {
+      id: '/posts_/$postId/deep'
+      path: '/posts/$postId/deep'
+      fullPath: '/posts/$postId/deep'
+      preLoaderRoute: typeof PostsPostIdDeepRouteImport
+      parentRoute: typeof rootRouteImport
+    }
+  }
+}
+
+interface UsersRouteChildren {
+  UsersUserIdRoute: typeof UsersUserIdRoute
+  UsersIndexRoute: typeof UsersIndexRoute
+}
+
+const UsersRouteChildren: UsersRouteChildren = {
+  UsersUserIdRoute: UsersUserIdRoute,
+  UsersIndexRoute: UsersIndexRoute,
+}
+
+const UsersRouteWithChildren = UsersRoute._addFileChildren(UsersRouteChildren)
+
+const rootRouteChildren: RootRouteChildren = {
+  IndexRoute: IndexRoute,
+  UsersRoute: UsersRouteWithChildren,
+  ApiClassifyRoute: ApiClassifyRoute,
+  PostsPostIdRoute: PostsPostIdRoute,
+  PostsIndexRoute: PostsIndexRoute,
+  PostsPostIdDeepRoute: PostsPostIdDeepRoute,
+}
+export const routeTree = rootRouteImport
+  ._addFileChildren(rootRouteChildren)
+  ._addFileTypes<FileRouteTypes>()
+
+import type { getRouter } from './router.tsx'
+import type { createStart } from '@tanstack/react-start'
+declare module '@tanstack/react-start' {
+  interface Register {
+    ssr: true
+    router: Awaited<ReturnType<typeof getRouter>>
+  }
+}

examples-ts/classify/src/router.tsx

@@ -0,0 +1,15 @@
+import { createRouter } from '@tanstack/react-router'
+import { routeTree } from './routeTree.gen'
+import { DefaultCatchBoundary } from './components/DefaultCatchBoundary'
+import { NotFound } from './components/NotFound'
+
+export function getRouter() {
+  const router = createRouter({
+    routeTree,
+    defaultPreload: 'intent',
+    defaultErrorComponent: DefaultCatchBoundary,
+    defaultNotFoundComponent: () => <NotFound />,
+    scrollRestoration: true,
+  })
+  return router
+}

examples-ts/classify/src/routes/__root.tsx

@@ -0,0 +1,128 @@
+/// <reference types="vite/client" />
+import {
+  HeadContent,
+  Scripts,
+  createRootRoute,
+} from '@tanstack/react-router'
+import * as React from 'react'
+import { DefaultCatchBoundary } from '~/components/DefaultCatchBoundary'
+import { NotFound } from '~/components/NotFound'
+import { seo } from '~/utils/seo'
+
+export const Route = createRootRoute({
+  head: () => ({
+    meta: [
+      {
+        charSet: 'utf-8',
+      },
+      {
+        name: 'viewport',
+        content: 'width=device-width, initial-scale=1',
+      },
+      ...seo({
+        title:
+          'Financial Documents Classification Agent',
+        description: `Classify financial documents as balance sheets, income statements and cash flow statemets. `,
+      }),
+    ],
+    links: [
+      { rel: 'stylesheet', href: "https://cdn.jsdelivr.net/npm/daisyui@5" },
+      {
+        rel: 'apple-touch-icon',
+        sizes: '180x180',
+        href: '/apple-touch-icon.png',
+      },
+      {
+        rel: 'icon',
+        type: 'image/png',
+        sizes: '32x32',
+        href: '/favicon-32x32.png',
+      },
+      {
+        rel: 'icon',
+        type: 'image/png',
+        sizes: '16x16',
+        href: '/favicon-16x16.png',
+      },
+      { rel: 'manifest', href: '/site.webmanifest', color: '#fffff' },
+      { rel: 'icon', href: '/favicon.ico' },
+    ],
+    scripts: [
+      {
+        src: '/customScript.js',
+        type: 'text/javascript',
+      },
+      {
+        src: "https://cdn.jsdelivr.net/npm/@tailwindcss/browser@4",
+        type: "text/javascript",
+      }
+    ],
+  }),
+  errorComponent: DefaultCatchBoundary,
+  notFoundComponent: () => <NotFound />,
+  shellComponent: RootDocument,
+})
+
+function RootDocument({ children }: { children: React.ReactNode }) {
+  return (
+    <html>
+      <head>
+        <HeadContent />
+      </head>
+      <body>
+        <div className="navbar bg-base-100 shadow-sm">
+          <div className="navbar-start">
+            <div className="dropdown">
+              <div tabIndex={0} role="button" className="btn btn-ghost btn-circle">
+                <svg
+                  xmlns="http://www.w3.org/2000/svg"
+                  className="h-5 w-5"
+                  fill="none"
+                  viewBox="0 0 24 24"
+                  stroke="currentColor"
+                >
+                  <path
+                    strokeLinecap="round"
+                    strokeLinejoin="round"
+                    strokeWidth="2"
+                    d="M4 6h16M4 12h16M4 18h7"
+                  />
+                </svg>
+              </div>
+              <ul
+                tabIndex={0}
+                className="menu menu-lg dropdown-content bg-base-100 rounded-box z-1 mt-3 w-80 p-2 shadow"
+              >
+                <li><a href="/">Home</a></li>
+                <li><a href="https://cloud.llamaindex.ai">Get Started with LlamaCloud</a></li>
+                <li><a href="https://developers.llamaindex.ai/python/cloud/llamaclassify/getting_started/">LlamaClassify Docs</a></li>
+              </ul>
+            </div>
+          </div>
+          <div className="navbar-center">
+            <a className="btn btn-ghost text-xl" href="/">Financial Documents Classification Agent</a>
+          </div>
+          <div className="navbar-end">
+            <a href="https://github.com/run-llama/llama_cloud_services/main/blob/examples-ts/classify">
+              <button className="btn btn-ghost btn-circle">
+                <div className="indicator">
+                  <svg
+                    xmlns="http://www.w3.org/2000/svg"
+                    className="h-10 w-10"
+                    fill="currentColor"
+                    viewBox="0 0 640 512"
+                  >
+                    <path d="M237.9 461.4C237.9 463.4 235.6 465 232.7 465C229.4 465.3 227.1 463.7 227.1 461.4C227.1 459.4 229.4 457.8 232.3 457.8C235.3 457.5 237.9 459.1 237.9 461.4zM206.8 456.9C206.1 458.9 208.1 461.2 211.1 461.8C213.7 462.8 216.7 461.8 217.3 459.8C217.9 457.8 216 455.5 213 454.6C210.4 453.9 207.5 454.9 206.8 456.9zM251 455.2C248.1 455.9 246.1 457.8 246.4 460.1C246.7 462.1 249.3 463.4 252.3 462.7C255.2 462 257.2 460.1 256.9 458.1C256.6 456.2 253.9 454.9 251 455.2zM316.8 72C178.1 72 72 177.3 72 316C72 426.9 141.8 521.8 241.5 555.2C254.3 557.5 258.8 549.6 258.8 543.1C258.8 536.9 258.5 502.7 258.5 481.7C258.5 481.7 188.5 496.7 173.8 451.9C173.8 451.9 162.4 422.8 146 415.3C146 415.3 123.1 399.6 147.6 399.9C147.6 399.9 172.5 401.9 186.2 425.7C208.1 464.3 244.8 453.2 259.1 446.6C261.4 430.6 267.9 419.5 275.1 412.9C219.2 406.7 162.8 398.6 162.8 302.4C162.8 274.9 170.4 261.1 186.4 243.5C183.8 237 175.3 210.2 189 175.6C209.9 169.1 258 202.6 258 202.6C278 197 299.5 194.1 320.8 194.1C342.1 194.1 363.6 197 383.6 202.6C383.6 202.6 431.7 169 452.6 175.6C466.3 210.3 457.8 237 455.2 243.5C471.2 261.2 481 275 481 302.4C481 398.9 422.1 406.6 366.2 412.9C375.4 420.8 383.2 435.8 383.2 459.3C383.2 493 382.9 534.7 382.9 542.9C382.9 549.4 387.5 557.3 400.2 555C500.2 521.8 568 426.9 568 316C568 177.3 455.5 72 316.8 72zM169.2 416.9C167.9 417.9 168.2 420.2 169.9 422.1C171.5 423.7 173.8 424.4 175.1 423.1C176.4 422.1 176.1 419.8 174.4 417.9C172.8 416.3 170.5 415.6 169.2 416.9zM158.4 408.8C157.7 410.1 158.7 411.7 160.7 412.7C162.3 413.7 164.3 413.4 165 412C165.7 410.7 164.7 409.1 162.7 408.1C160.7 407.5 159.1 407.8 158.4 408.8zM190.8 444.4C189.2 445.7 189.8 448.7 192.1 450.6C194.4 452.9 197.3 453.2 198.6 451.6C199.9 450.3 199.3 447.3 197.3 445.4C195.1 443.1 192.1 442.8 190.8 444.4zM179.4 429.7C177.8 430.7 177.8 433.3 179.4 435.6C181 437.9 183.7 438.9 185 437.9C186.6 436.6 186.6 434 185 431.7C183.6 429.4 181 428.4 179.4 429.7z" />
+                  </svg>
+                </div>
+              </button>
+            </a>
+          </div>
+        </div>
+        <hr />
+        {children}
+        <Scripts />
+      </body>
+    </html>
+  )
+}

examples-ts/classify/src/routes/api/classify.ts

@@ -0,0 +1,45 @@
+import { createFileRoute } from '@tanstack/react-router'
+import { classifier, classificationRules, parsingConfig } from '~/utils/classifier'
+
+export const Route = createFileRoute('/api/classify')({
+  component: RouteComponent,
+  server: {
+      handlers: {
+        POST: async ({ request }) => {
+          const body = await request.formData()
+          const fl = body.get("file") as File;
+          if (!fl) {
+                return new Response(JSON.stringify({"result": "you need to provide a file"}))
+            }
+          const buff = await fl.arrayBuffer()
+          const rawRes = await classifier.classify(
+            classificationRules,
+            parsingConfig,
+            { fileContents: [new Uint8Array(buff)] },
+          )
+          const results = rawRes.items
+          let classification = ""
+
+          for (const result of results) {
+            if ("result" in result && result.result) {
+                classification += `
+                    <div class="card bg-base-100 shadow-xl p-6 mb-4">
+                        <div class="space-y-3">
+                        <p><span class="font-semibold">📄 Document:</span> ${fl.name}</p>
+                        <p><span class="font-semibold">🏷️ Type:</span> <span class="badge badge-primary">${result.result.type}</span></p>
+                        <p><span class="font-semibold">📊 Confidence:</span> ${result.result.confidence*100}%</p>
+                        <p><span class="font-semibold">💭 Reasoning:</span> ${result.result.reasoning}</p>
+                        </div>
+                    </div>
+                `
+            }
+          }
+          return new Response(JSON.stringify({"result": classification}))
+        },
+      },
+    },
+})
+
+function RouteComponent() {
+  return
+}

examples-ts/classify/src/routes/index.tsx

@@ -0,0 +1,99 @@
+import { createFileRoute } from '@tanstack/react-router'
+import { useRef, useState } from 'react'
+
+export const Route = createFileRoute('/')({
+  component: Home,
+})
+
+function Home() {
+  const [file, setFile] = useState<null | File>(null)
+  const fileInputRef = useRef<HTMLInputElement>(null)
+  const [reply, setReply] = useState<null | string>(null)
+  const [loading, setLoading] = useState<boolean>(false)
+  const handleFileChange = (event: React.ChangeEvent<HTMLInputElement>) => {
+    const selectedFile = event.target.files?.[0]
+    if (selectedFile) {
+      setFile(selectedFile)
+    }
+  }
+  const handleClearFile = () => {
+    if (file) {
+      setFile(null)
+    }
+    if (fileInputRef.current) {
+      fileInputRef.current.value = ''
+    }
+    if (reply) {
+      setReply(null)
+    }
+  }
+
+  const handleClassify = async () => {
+    if (!file) return
+
+    if (reply) {
+      setReply(null)
+    }
+    setLoading(true)
+    try {
+      const formData = new FormData()
+      formData.append('file', file)
+
+      const res = await fetch('/api/classify', {
+        method: 'POST',
+        body: formData,
+      })
+
+      const data = await res.json()
+      setReply(data.result)
+    } catch (error) {
+      console.error('Error:', error)
+    } finally {
+      setLoading(false)
+    }
+  }
+
+  return (
+    <div className="flex flex-col justify-center items-center gap-y-8">
+        <br />
+        <h1 className="text-xl font-bold text-gray-700">AI-Powered finacial document classification</h1>
+        <h2 className="text-lg font-semibold text-gray-500">Need help sorting out the financial documents jungle? Let our classification agent handle it!</h2>
+        <fieldset className="fieldset bg-base-100 border-base-300 rounded-box w-200 border p-4">
+            <legend className="fieldset-legend text-lg">Upload your financial document here</legend>
+            <label className="label flex justify-center">
+                <input type="file" className="file-input" onChange={handleFileChange} accept='application/pdf' ref={fileInputRef} />
+            </label>
+        </fieldset>
+        {file && (
+          <div className="flex flex-col justify-center items-center gap-y-8">
+          <p className="text-sm text-gray-600">Selected file: {file.name}</p>
+          <div className='grid grid-cols-2 gap-x-6'>
+            <button
+                type="button"
+                className='btn bg-gray-500 text-white shadow-lg hover:bg-gray-600 hover:shadow-xl rounded'
+                onClick={handleClassify}
+            >
+            Classify
+            </button>
+            <button
+                  onClick={handleClearFile}
+                  type="button"
+                  className="px-4 py-2 bg-red-300 text-black rounded hover:bg-red-400 hover:shadow-xl shadow-lg"
+              >
+              Clear
+            </button>
+          </div>
+          </div>
+        )}
+        {loading && (
+          <span className="loading loading-spinner text-primary"></span>
+        )}
+        {reply && (
+          <div
+            className="max-w-2xl w-full"
+            dangerouslySetInnerHTML={{ __html: reply }}
+          />
+        )}
+    </div>
+  )
+}

examples-ts/classify/src/utils/classifier.ts

@@ -0,0 +1,23 @@
+import { LlamaClassify, ClassifierRule, ClassifyParsingConfiguration } from "llama-cloud-services"
+
+export const classifier = new LlamaClassify(process.env.LLAMA_CLOUD_API_KEY);
+
+export const classificationRules: ClassifierRule[] = [
+    {
+        description: "Shows a company's assets, liabilities, and shareholders' equity at a specific point in time, providing a snapshot of financial position.",
+        type: "balance_sheet"
+    },
+    {
+        description: "Reports cash inflows and outflows from operating, investing, and financing activities, highlighting liquidity and cash management.",
+        type: "cash_flow_statement"
+    },
+    {
+        description: "Summarizes revenues, expenses, and profits over a period, indicating financial performance and profitability.",
+        type: "income_statement"
+    },
+];
+
+export const parsingConfig: ClassifyParsingConfiguration = {
+    lang: "en",
+    max_pages: 20,
+}

examples-ts/classify/src/utils/seo.ts

@@ -0,0 +1,33 @@
+export const seo = ({
+  title,
+  description,
+  keywords,
+  image,
+}: {
+  title: string
+  description?: string
+  image?: string
+  keywords?: string
+}) => {
+  const tags = [
+    { title },
+    { name: 'description', content: description },
+    { name: 'keywords', content: keywords },
+    { name: 'twitter:title', content: title },
+    { name: 'twitter:description', content: description },
+    { name: 'twitter:creator', content: '@tannerlinsley' },
+    { name: 'twitter:site', content: '@tannerlinsley' },
+    { name: 'og:type', content: 'website' },
+    { name: 'og:title', content: title },
+    { name: 'og:description', content: description },
+    ...(image
+      ? [
+          { name: 'twitter:image', content: image },
+          { name: 'twitter:card', content: 'summary_large_image' },
+          { name: 'og:image', content: image },
+        ]
+      : []),
+  ]
+
+  return tags
+}

examples-ts/classify/tsconfig.json

@@ -0,0 +1,22 @@
+{
+  "include": ["**/*.ts", "**/*.tsx"],
+  "compilerOptions": {
+    "strict": true,
+    "esModuleInterop": true,
+    "jsx": "react-jsx",
+    "module": "ESNext",
+    "moduleResolution": "Bundler",
+    "lib": ["DOM", "DOM.Iterable", "ES2022"],
+    "isolatedModules": true,
+    "resolveJsonModule": true,
+    "skipLibCheck": true,
+    "target": "ES2022",
+    "allowJs": true,
+    "forceConsistentCasingInFileNames": true,
+    "baseUrl": ".",
+    "paths": {
+      "~/*": ["./src/*"]
+    },
+    "noEmit": true
+  }
+}

examples-ts/classify/vite.config.ts

@@ -0,0 +1,19 @@
+import { tanstackStart } from '@tanstack/react-start/plugin/vite'
+import { defineConfig } from 'vite'
+import tsConfigPaths from 'vite-tsconfig-paths'
+import viteReact from '@vitejs/plugin-react'
+
+export default defineConfig({
+  server: {
+    port: 3000,
+  },
+  plugins: [
+    tsConfigPaths({
+      projects: ['./tsconfig.json'],
+    }),
+    tanstackStart({
+      srcDirectory: 'src',
+    }),
+    viteReact(),
+  ],
+})

examples/extract/tabular_extraction.ipynb

@@ -0,0 +1,508 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "a7oq3cfnync",
+   "metadata": {},
+   "source": [
+    "# Extracting Repeating Entities from Documents\n",
+    "\n",
+    "This notebook demonstrates how to use the `PER_TABLE_ROW` extraction target to extract structured data from documents containing repeating entities like tables, lists, or catalogs.\n",
+    "\n",
+    "## Why Use the Tabular Extraction Target?\n",
+    "\n",
+    "`PER_DOC` (refer to the table below for a quick overview of the different extraction targets) is the default extraction target in LlamaExtract, which looks at the entire document's context when doing an extraction. When extracting lists of entities, LLM-based extraction has a critical failure mode — it often **only extracts the first few tens of entries** from a long list. This happens because LLMs have limited attention spans for repetitive data. Document-level extraction doesn't guarantee exhaustive coverage, and long lists lead to incomplete extractions.\n",
+    "\n",
+    "**The Solution**: `PER_TABLE_ROW` solves this by processing each entity individually or in smaller batches, ensuring **exhaustive extraction** of all entries regardless of list length.\n",
+    "\n",
+    "### Entity-Level Extraction\n",
+    "\n",
+    "When using `extraction_target=ExtractTarget.PER_TABLE_ROW`, you define a schema for a **single entity** (e.g., one hospital, one product, one invoice line item), not the full document. LlamaExtract automatically:\n",
+    "- Detects the formatting patterns that distinguish individual entities (table rows, list items, section headers, etc.)\n",
+    "- Applies your schema to each identified entity\n",
+    "- Returns a `list[YourSchema]` with one object per entity\n",
+    "\n",
+    "This approach is ideal when each entity locally contains all the information needed for your schema.\n",
+    "\n",
+    "### Choosing the Right Extraction Target\n",
+    "\n",
+    "| Extraction Target | Best For | Returns |\n",
+    "|-------------------|----------|---------|\n",
+    "| `PER_DOC` | Single-entity documents, summaries, or short lists | One JSON object for entire document |\n",
+    "| `PER_PAGE` | Multi-page documents where each page is independent | One JSON object per page |\n",
+    "| `PER_TABLE_ROW` | **Long lists, tables, catalogs with repeating entities** | List of JSON objects (one per entity) |\n",
+    "\n",
+    "📖 For more details, see the [Extraction Target documentation](https://developers.llamaindex.ai/python/cloud/llamaextract/features/concepts/#extraction-target)."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "9427d1de",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from dotenv import load_dotenv\n",
+    "from llama_cloud_services import LlamaExtract\n",
+    "\n",
+    "\n",
+    "# Load environment variables (put LLAMA_CLOUD_API_KEY in your .env file)\n",
+    "load_dotenv(override=True)\n",
+    "\n",
+    "# Optionally, add your project id/organization id\n",
+    "llama_extract = LlamaExtract()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "4426b360",
+   "metadata": {},
+   "source": [
+    "## Table of Hospitals by County and Insurance Plans\n",
+    "\n",
+    "We have a PDF document with a list of hospitals by county and different insurance plans offered by Blue Shield of California. \n",
+    "\n",
+    "\n",
+    "![First few entries from the PDF](./data/tables/bsc_page1.png)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c86sjymhn1r",
+   "metadata": {},
+   "source": [
+    "We want to extract each hospital from this table along with a list of applicable insurance plans. \n",
+    "\n",
+    "### Example 1: Structured Table\n",
+    "\n",
+    "This is an ideal use case for `PER_TABLE_ROW` extraction:\n",
+    "- **Clear structure**: The document has explicit table formatting with rows and columns\n",
+    "- **Repeating entities**: Each row represents one hospital with consistent attributes\n",
+    "- **Local information**: All data for each hospital (county, name, plans) is contained within its row\n",
+    "\n",
+    "Notice that our `Hospital` schema describes a **single hospital**, not the full document. LlamaExtract will return a `list[Hospital]` with one entry per table row."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "7c61a802",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from pydantic import BaseModel, Field\n",
+    "\n",
+    "\n",
+    "class Hospital(BaseModel):\n",
+    "    \"\"\"List of hospitals by county available for different BSC plans\"\"\"\n",
+    "\n",
+    "    county: str = Field(description=\"County name\")\n",
+    "    hospital_name: str = Field(description=\"Name of the hospital\")\n",
+    "    plan_names: list[str] = Field(\n",
+    "        description=\"List of plans available at the hospital. One of: Trio HMO, SaveNet, Access+ HMO, BlueHPN PPO, Tandem PPO, PPO\"\n",
+    "    )"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b8a69b7a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from llama_cloud_services.extract import ExtractConfig, ExtractMode, ExtractTarget\n",
+    "\n",
+    "\n",
+    "result = await llama_extract.aextract(\n",
+    "    data_schema=Hospital,\n",
+    "    files=\"./data/tables/BSC-Hospital-List-by-County.pdf\",\n",
+    "    config=ExtractConfig(\n",
+    "        extraction_mode=ExtractMode.PREMIUM,\n",
+    "        extraction_target=ExtractTarget.PER_TABLE_ROW,\n",
+    "        parse_model=\"anthropic-sonnet-4.5\",\n",
+    "    ),\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "43722cda",
+   "metadata": {},
+   "source": [
+    "### Results"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "95b5aca6",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "380"
+      ]
+     },
+     "execution_count": null,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "len(result.data)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1e355770",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "[{'county': 'Alameda',\n",
+       "  'hospital_name': 'Alameda Hospital',\n",
+       "  'plan_names': ['Trio HMO',\n",
+       "   'SaveNet',\n",
+       "   'Access+ HMO',\n",
+       "   'BlueHPN PPO',\n",
+       "   'Tandem PPO',\n",
+       "   'PPO']},\n",
+       " {'county': 'Alameda',\n",
+       "  'hospital_name': 'Alta Bates Med Ctr Herrick Campus',\n",
+       "  'plan_names': ['Trio HMO',\n",
+       "   'Access+ HMO',\n",
+       "   'BlueHPN PPO',\n",
+       "   'Tandem PPO',\n",
+       "   'PPO']},\n",
+       " {'county': 'Alameda',\n",
+       "  'hospital_name': 'Alta Bates Summit Med Ctr Alta Bates Campus',\n",
+       "  'plan_names': ['Trio HMO',\n",
+       "   'Access+ HMO',\n",
+       "   'BlueHPN PPO',\n",
+       "   'Tandem PPO',\n",
+       "   'PPO']},\n",
+       " {'county': 'Alameda',\n",
+       "  'hospital_name': 'Alta Bates Summit Med Ctr Summit Campus',\n",
+       "  'plan_names': ['Trio HMO',\n",
+       "   'Access+ HMO',\n",
+       "   'BlueHPN PPO',\n",
+       "   'Tandem PPO',\n",
+       "   'PPO']},\n",
+       " {'county': 'Alameda',\n",
+       "  'hospital_name': 'Alta Bates Summit Medical Center',\n",
+       "  'plan_names': ['Trio HMO',\n",
+       "   'Access+ HMO',\n",
+       "   'BlueHPN PPO',\n",
+       "   'Tandem PPO',\n",
+       "   'PPO']},\n",
+       " {'county': 'Alameda',\n",
+       "  'hospital_name': 'BHC Fremont Hospital',\n",
+       "  'plan_names': ['Trio HMO',\n",
+       "   'SaveNet',\n",
+       "   'Access+ HMO',\n",
+       "   'BlueHPN PPO',\n",
+       "   'Tandem PPO',\n",
+       "   'PPO']},\n",
+       " {'county': 'Alameda',\n",
+       "  'hospital_name': 'Centre For Neuro Skills San Francisco',\n",
+       "  'plan_names': ['Trio HMO',\n",
+       "   'SaveNet',\n",
+       "   'Access+ HMO',\n",
+       "   'BlueHPN PPO',\n",
+       "   'Tandem PPO',\n",
+       "   'PPO']},\n",
+       " {'county': 'Alameda',\n",
+       "  'hospital_name': 'Eden Medical Center',\n",
+       "  'plan_names': ['Trio HMO', 'Access+ HMO', 'PPO']},\n",
+       " {'county': 'Alameda',\n",
+       "  'hospital_name': 'Fairmont Hospital',\n",
+       "  'plan_names': ['Trio HMO',\n",
+       "   'SaveNet',\n",
+       "   'Access+ HMO',\n",
+       "   'BlueHPN PPO',\n",
+       "   'Tandem PPO',\n",
+       "   'PPO']},\n",
+       " {'county': 'Alameda',\n",
+       "  'hospital_name': 'Highland Hospital',\n",
+       "  'plan_names': ['Trio HMO',\n",
+       "   'SaveNet',\n",
+       "   'Access+ HMO',\n",
+       "   'BlueHPN PPO',\n",
+       "   'Tandem PPO',\n",
+       "   'PPO']}]"
+      ]
+     },
+     "execution_count": null,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "result.data[:10]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e28f0de8",
+   "metadata": {},
+   "source": [
+    "![](./data/tables/bsc_results.png)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "di156pb7s6j",
+   "metadata": {},
+   "source": [
+    "**Success!** We extracted all **380 hospitals** from the multi-page PDF. Each entity was correctly parsed with its county, hospital name, and applicable insurance plans. With `PER_DOC`, we would likely have only gotten the first 20-30 entries."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "gelvl6db268",
+   "metadata": {},
+   "source": [
+    "## Extracting from a Toy Catalog\n",
+    "\n",
+    "### Example 2: Semi-Structured List\n",
+    "\n",
+    "The `PER_TABLE_ROW` extraction target also works well for documents that aren't explicit tables but have similar properties:\n",
+    "- **Ordered listing**: The toys are listed sequentially with visual separation (section headers, spacing)\n",
+    "- **Repeating pattern**: Each toy entry has a consistent structure (code, name, specs, description)\n",
+    "- **Local information**: All attributes for each toy are grouped together in its entry\n",
+    "\n",
+    "Even though this isn't a traditional table format, each toy entity locally contains all the information needed for our schema. LlamaExtract detects the formatting patterns that distinguish each toy and extracts them as separate entities.\n",
+    "\n",
+    "![](./data/tables/toy_catalog_page.png)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "8cf0b2db",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from pydantic import BaseModel, Field\n",
+    "\n",
+    "\n",
+    "class ToyCatalog(BaseModel):\n",
+    "    \"\"\"Product information from a toy catalog.\"\"\"\n",
+    "\n",
+    "    section_name: str = Field(\n",
+    "        description=\"The name of the toy section (e.g. Table Toys, Active Toys).\"\n",
+    "    )\n",
+    "    product_code: str = Field(\n",
+    "        description=\"The unique product code for the toy (e.g., GA457).\"\n",
+    "    )\n",
+    "    toy_name: str = Field(description=\"The name of the toy.\")\n",
+    "    age_range: str = Field(\n",
+    "        description=\"The recommended age range for the toy (e.g., 6 +, 4 +).\",\n",
+    "    )\n",
+    "    player_range: str = Field(\n",
+    "        description=\"The number of players the toy is designed for (e.g., 2, 2-4, 1-6).\",\n",
+    "    )\n",
+    "    material: str = Field(\n",
+    "        description=\"The primary material(s) the toy is made of (e.g., wood, cardboard).\",\n",
+    "    )\n",
+    "    description: str = Field(\n",
+    "        description=\"A brief description of the toy and its components and dimensions.\",\n",
+    "    )"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "mysu1i2qo9e",
+   "metadata": {},
+   "source": [
+    "### Results\n",
+    "\n",
+    "Again, our schema represents a **single toy product**, not the entire catalog. The system will return a `list[ToyCatalog]` with one entry per toy."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "5b38b806",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "result = await llama_extract.aextract(\n",
+    "    data_schema=ToyCatalog,\n",
+    "    files=\"./data/tables/Click-BS-Toys-Catalogue-2024.pdf\",\n",
+    "    config=ExtractConfig(\n",
+    "        extraction_mode=ExtractMode.PREMIUM,\n",
+    "        extraction_target=ExtractTarget.PER_TABLE_ROW,\n",
+    "        parse_model=\"anthropic-sonnet-4.5\",\n",
+    "    ),\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "91aface0",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "153"
+      ]
+     },
+     "execution_count": null,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "len(result.data)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "51278736",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "[{'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA457',\n",
+       "  'toy_name': 'Dots and Boxes',\n",
+       "  'age_range': '6+',\n",
+       "  'player_range': '2',\n",
+       "  'material': 'wood',\n",
+       "  'description': 'base 17x17 cm\\n50 border pieces 4x1,2x0,3 cm\\n34 trees 2,6x1,4 cm'},\n",
+       " {'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA456',\n",
+       "  'toy_name': '3 In a Row',\n",
+       "  'age_range': '8+',\n",
+       "  'player_range': '2',\n",
+       "  'material': 'wood, pine, cardboard',\n",
+       "  'description': 'base 24x22,5x2,5 cm\\n30 cards 5,5x5 cm\\n6 chips'},\n",
+       " {'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA467',\n",
+       "  'toy_name': 'Which Cow am i?',\n",
+       "  'age_range': '6+',\n",
+       "  'player_range': '2',\n",
+       "  'material': 'wood, beech',\n",
+       "  'description': '2 cow bases 56x4x4,5 cm\\n16 cards 4x5 cm'},\n",
+       " {'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA460',\n",
+       "  'toy_name': 'Balance Bunnies',\n",
+       "  'age_range': '4+',\n",
+       "  'player_range': '2',\n",
+       "  'material': 'wood',\n",
+       "  'description': '1 base 35x12x25 cm\\n7 bunnies 7 foxes\\n1 dice 3 cm'},\n",
+       " {'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA462',\n",
+       "  'toy_name': 'Color Combination Race',\n",
+       "  'age_range': '4+',\n",
+       "  'player_range': '2-4',\n",
+       "  'material': 'wood, cardboard',\n",
+       "  'description': 'base 6,5x6,5x15 cm, rings 5,5x5,5x0,5 mm\\ncardholder 6x6x2 cm, cards 5,5x5,5 cm\\ncolor cards Ø 15,5 cm - Ø 7 cm'},\n",
+       " {'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA465',\n",
+       "  'toy_name': 'Plop It',\n",
+       "  'age_range': '6+',\n",
+       "  'player_range': '2-4',\n",
+       "  'material': 'wood, elastic, cardboard',\n",
+       "  'description': 'Catch the right balls and plop them in the net!\\n* 2 ploppers 8x5 cm\\n* 2 net holders Ø 5cm, length 55 cm\\n* 6 cards 1,5x2,5 cm, 30 balls Ø 2,5 cm\\n* 1 rope 120 cm'},\n",
+       " {'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA466',\n",
+       "  'toy_name': 'Whack a Shape',\n",
+       "  'age_range': '4+',\n",
+       "  'player_range': '2-4',\n",
+       "  'material': 'wood',\n",
+       "  'description': '* base 38,5x15,5 cm\\n* 2 stands 36 half balls, 4 hammers\\n* 1 dice 2,5 cm\\n* 4 cards'},\n",
+       " {'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA458',\n",
+       "  'toy_name': 'Sling Puck | Table Hockey',\n",
+       "  'age_range': '6+',\n",
+       "  'player_range': '2',\n",
+       "  'material': 'wood',\n",
+       "  'description': '* double sides base 39x21x3 cm\\n* 10 chips Ø 2,5 cm\\n* 2 pushers 4x4x3 cm'},\n",
+       " {'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA039',\n",
+       "  'toy_name': 'DIY Birdhouse',\n",
+       "  'age_range': '3+',\n",
+       "  'player_range': '1',\n",
+       "  'material': 'wood',\n",
+       "  'description': '* house 9x9x13 cm'},\n",
+       " {'section_name': 'Table Toys',\n",
+       "  'product_code': 'GA319',\n",
+       "  'toy_name': 'Triangle Domino',\n",
+       "  'age_range': '6+',\n",
+       "  'player_range': '2-4',\n",
+       "  'material': 'wood',\n",
+       "  'description': '* 35 triangles 10x10 x10 cm'}]"
+      ]
+     },
+     "execution_count": null,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "result.data[:10]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "d1810c0a",
+   "metadata": {},
+   "source": [
+    "![](./data/tables/toy_catalog_results.png)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "ezur9gnhmsb",
+   "metadata": {},
+   "source": [
+    "**Success!** Despite the semi-structured format, we extracted all **152 toy products** from the catalog (there's an extra repeated extracted toy from the Appendix section). LlamaExtract automatically detected the visual patterns separating each toy entry and applied our schema to each one."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "aeyr3io29u",
+   "metadata": {},
+   "source": [
+    "## Summary\n",
+    "\n",
+    "The `PER_TABLE_ROW` extraction target is powerful for extracting repeating structured entities from documents. Key takeaways:\n",
+    "\n",
+    "1. **Schema design**: Define your schema for a single entity, not the full document. The system returns `list[YourSchema]`.\n",
+    "\n",
+    "2. **Works with various formats**: Not just traditional tables—any document with distinguishable repeating entities (bullets, numbering, headers, visual separation, etc.). The common requirement is that each entity should contain all the necessary data for your schema within its local context.\n",
+    "\n",
+    "3. **Automatic pattern detection**: LlamaExtract identifies the formatting patterns that distinguish entities and applies your schema to each one."
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": ".venv",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

examples/misc/parse_classify_extract_workflow.ipynb

@@ -4,31 +4,19 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "# Complete Parse → Classify → Extract Workflow with LlamaCloud Services\n",
+    "# Document Classification + Extraction Workflow with LlamaCloud + LlamaIndex Workflows\n",
     "\n",
-    "This notebook demonstrates the complete workflow for processing documents using LlamaCloud services:\n",
-    "1. **Parse** - Extract and convert documents to markdown\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/misc/parse_classify_extract_workflow.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "\n",
+    "This notebook shows a multi-step agentic document workflow that uses the **parsing**, **classification** and **extraction** modules in LlamaCloud, orchestrated through **LlamaIndex Workflows**. The workflow can take in a complex input document, parse it into clean markdown, classify it according to its subtype, and extract data according to a specified schema for that subtype. This allows you to automate document extraction of various types within the same workflow instead of having to manually separate the data beforehand. \n",
+    "\n",
+    "This notebook uses the following modules:\n",
+    "1. **Parse (LlamaParse)** - Extract and convert documents to markdown\n",
     "2. **Classify** - Categorize documents based on their content\n",
-    "3. **Extract** - Extract structured data using the markdown as input via SourceText\n",
+    "3. **Extract (LlamaExtract)** - Extract structured data using the markdown as input via SourceText\n",
+    "4. **LlamaIndex Workflows** - Event-driven orchestration of the parse, classify and extract steps\n",
     "\n",
-    "## Overview of the Workflow\n",
-    "\n",
-    "### 1. Parse Phase\n",
-    "- Use `LlamaParse` to convert documents (PDFs, Word docs, etc.) into structured formats\n",
-    "- Extract markdown content that preserves document structure\n",
-    "- Get both raw text and markdown representations\n",
-    "\n",
-    "### 2. Classify Phase\n",
-    "- Use `ClassifyClient` to categorize documents based on content\n",
-    "- Apply classification rules to route documents appropriately\n",
-    "- Handle different document types with specific processing logic\n",
-    "\n",
-    "### 3. Extract Phase\n",
-    "- Use `LlamaExtract` with `SourceText` to extract structured data\n",
-    "- Pass the markdown content as input for more accurate extraction\n",
-    "- Define custom schemas for structured data extraction\n",
-    "\n",
-    "Let's walk through each step with practical examples."
+    "The workflow is implemented as a proper LlamaIndex Workflow with separate steps for parsing, classification, and extraction, connected by typed events. This provides modularity, observability, and type safety."
    ]
   },
   {
@@ -45,8 +33,8 @@
    "outputs": [],
    "source": [
     "# Install required packages\n",
-    "!pip install llama-cloud-services\n",
-    "!pip install python-dotenv"
+    "%pip install llama-cloud-services\n",
+    "%pip install python-dotenv"
    ]
   },
   {
@@ -73,7 +61,7 @@
     "nest_asyncio.apply()\n",
     "\n",
     "# Set up API key\n",
-    "os.environ[\"LLAMA_CLOUD_API_KEY\"] = \"\"  # edit it\n",
+    "# os.environ[\"LLAMA_CLOUD_API_KEY\"] = \"\"  # edit it\n",
     "\n",
     "# Setup Base URL\n",
     "# os.envrion[\"LLAMA_CLOUD_BASE_URL\"] = \"https://api.cloud.eu.llamaindex.ai/\" # update if necessay\n",
@@ -99,7 +87,8 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "📁 financial_report.pdf already exists\n",
+      "Downloading financial_report.pdf...\n",
+      "✅ Downloaded financial_report.pdf\n",
       "📁 technical_spec.pdf already exists\n",
       "\n",
       "📂 Sample documents ready!\n"
@@ -115,7 +104,7 @@
     "\n",
     "# Download sample documents\n",
     "docs_to_download = {\n",
-    "    \"financial_report.pdf\": \"https://raw.githubusercontent.com/run-llama/llama_index/main/docs/docs/examples/data/10k/uber_2021.pdf\",\n",
+    "    \"financial_report.pdf\": \"https://raw.githubusercontent.com/run-llama/llama_index/main/docs/examples/data/10k/uber_2021.pdf\",\n",
     "    \"technical_spec.pdf\": \"https://www.ti.com/lit/ds/symlink/lm317.pdf\",\n",
     "}\n",
     "\n",
@@ -155,10 +144,10 @@
      "output_type": "stream",
      "text": [
       "🔄 Parsing documents...\n",
-      "Started parsing the file under job_id 8a8c76f9-354d-4275-91d8-312ff1adc762\n",
-      "...✅ Parsed financial report (Job ID: 8a8c76f9-354d-4275-91d8-312ff1adc762)\n",
-      "Started parsing the file under job_id 7e603448-ed80-4d18-948b-6801ed51c41b\n",
-      "✅ Parsed technical spec (Job ID: 7e603448-ed80-4d18-948b-6801ed51c41b)\n",
+      "Started parsing the file under job_id 530c187a-bd2d-4eea-b38d-9e5738eab465\n",
+      ".✅ Parsed financial report (Job ID: 530c187a-bd2d-4eea-b38d-9e5738eab465)\n",
+      "Started parsing the file under job_id a6e27710-776b-4445-8b94-8d75959ff5db\n",
+      "✅ Parsed technical spec (Job ID: a6e27710-776b-4445-8b94-8d75959ff5db)\n",
       "\n",
       "📄 Parsing complete!\n"
      ]
@@ -246,23 +235,23 @@
       "\n",
       "## 1 Features\n",
       "\n",
-      "• Output voltage range:\n",
+      "- Output voltage range:\n",
       "  – Adjustable: 1.25V to 37V\n",
-      "• Output current: 1.5A\n",
-      "• Line regulation: 0.01%/V (typ)\n",
-      "• Load regulation: 0.1% (typ)\n",
-      "• Internal short-circuit current limiting\n",
-      "• Thermal overload protection\n",
-      "• Output safe-area compensation (new chip)\n",
-      "• PSRR: 80dB at 120Hz for CADJ = 10μF (new chip)\n",
-      "• Packages:\n",
+      "- Output current: 1.5A\n",
+      "- Line regulation: 0.01%/V (typ)\n",
+      "- Load regulation: 0.1% (typ)\n",
+      "- Internal short-circuit current limiting\n",
+      "- Thermal overload protection\n",
+      "- Output safe-area compensation (new chip)\n",
+      "- PSRR: 80dB at 120Hz for CADJ = 10μF (new chip)\n",
+      "- Packages:\n",
       "  – 4-pin, SOT-223 (DCY)\n",
       "  – 3-pin, TO-263 (KTT)\n",
       "  – 3-pin, TO-220 (KCS, KCT),\n",
       "...\n",
       "\n",
-      "📏 Financial report markdown length: 1348671 characters\n",
-      "📏 Technical spec markdown length: 90971 characters\n"
+      "📏 Financial report markdown length: 1338499 characters\n",
+      "📏 Technical spec markdown length: 92483 characters\n"
      ]
     }
    ],
@@ -291,7 +280,7 @@
    "source": [
     "## Phase 2: Document Classification\n",
     "\n",
-    "Next, let's classify our documents based on their content using the ClassifyClient."
+    "Next, let's classify our documents based on their content using `LlamaClassify`."
    ]
   },
   {
@@ -309,14 +298,14 @@
     }
    ],
    "source": [
-    "from llama_cloud_services.beta.classifier.client import ClassifyClient\n",
+    "from llama_cloud_services.beta.classifier.client import LlamaClassify\n",
     "from llama_cloud.types import ClassifierRule\n",
     "from llama_cloud_services.files.client import FileClient\n",
     "from llama_cloud.client import AsyncLlamaCloud\n",
     "\n",
     "# Initialize the classify client\n",
     "api_key = os.environ[\"LLAMA_CLOUD_API_KEY\"]\n",
-    "classify_client = ClassifyClient.from_api_key(api_key)\n",
+    "classify_client = LlamaClassify.from_api_key(api_key)\n",
     "\n",
     "print(\"🏷️  Setting up document classification...\")\n",
     "\n",
@@ -339,6 +328,72 @@
     "print(f\"📝 Created {len(classification_rules)} classification rules\")"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Try Classification Independently\n",
+    "\n",
+    "Let's test the classification on one of our parsed documents to see how it works:\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "🔍 Classifying financial document...\n",
+      "   Document length: 1,338,499 characters\n",
+      "\n",
+      "✅ Classification Result:\n",
+      "   Type: financial_document\n",
+      "   Confidence: 100.00%\n",
+      "   Reasoning: This document is a Form 10-K, which is an annual report required by the U.S. Securities and Exchange Commission (SEC) for publicly traded companies. It contains financial data, information about the c...\n",
+      "\n",
+      "======================================================================\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Let's classify the financial document\n",
+    "print(\"🔍 Classifying financial document...\")\n",
+    "print(f\"   Document length: {len(financial_markdown):,} characters\\n\")\n",
+    "\n",
+    "# Write to temp file for classification\n",
+    "import tempfile\n",
+    "from pathlib import Path\n",
+    "\n",
+    "with tempfile.NamedTemporaryFile(\n",
+    "    mode=\"w\", suffix=\".md\", delete=False, encoding=\"utf-8\"\n",
+    ") as tmp:\n",
+    "    tmp.write(financial_markdown)\n",
+    "    temp_financial_path = Path(tmp.name)\n",
+    "\n",
+    "# Classify the document\n",
+    "financial_classification = await classify_client.aclassify_file_path(\n",
+    "    rules=classification_rules, file_input_path=str(temp_financial_path)\n",
+    ")\n",
+    "\n",
+    "doc_type = financial_classification.items[0].result.type\n",
+    "confidence = financial_classification.items[0].result.confidence\n",
+    "reasoning = financial_classification.items[0].result.reasoning\n",
+    "\n",
+    "print(f\"✅ Classification Result:\")\n",
+    "print(f\"   Type: {doc_type}\")\n",
+    "print(f\"   Confidence: {confidence:.2%}\")\n",
+    "print(\n",
+    "    f\"   Reasoning: {reasoning[:200]}...\"\n",
+    "    if reasoning and len(reasoning) > 200\n",
+    "    else f\"   Reasoning: {reasoning}\"\n",
+    ")\n",
+    "\n",
+    "print(\"\\n\" + \"=\" * 70)"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -444,9 +499,31 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Complete Workflow Summary\n",
+    "## Building the Complete Workflow\n",
     "\n",
-    "Let's create a function that demonstrates the complete workflow:"
+    "Now that we've seen how parsing works, let's build a complete 3-step workflow (Parse → Classify → Extract) using LlamaIndex Workflows. We'll define the workflow structure here, and you can see it in action below where we also demonstrate the classification and extraction modules independently.\n",
+    "\n",
+    "### Install Workflows Package\n",
+    "\n",
+    "First, let's install the LlamaIndex workflows package:\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%pip install llama-index-workflows llama-index-utils-workflow"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Define the Workflow\n",
+    "\n",
+    "Let's restructure the document processing into a proper LlamaIndex Workflow with separate classification and extraction steps:\n"
    ]
   },
   {
@@ -458,7 +535,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "🔧 Workflow function defined!\n"
+      "🔧 Workflow defined!\n"
      ]
     }
    ],
@@ -466,81 +543,286 @@
     "import tempfile\n",
     "from pathlib import Path\n",
     "from llama_cloud import ExtractConfig\n",
+    "from workflows import Workflow, step, Context\n",
+    "from workflows.events import Event, StartEvent, StopEvent\n",
     "\n",
     "\n",
-    "async def complete_document_workflow(markdown_content: str):\n",
+    "# Define workflow events\n",
+    "class ParseEvent(Event):\n",
+    "    \"\"\"Event emitted after parsing\"\"\"\n",
+    "\n",
+    "    file_path: str\n",
+    "    markdown_content: str\n",
+    "    job_id: str\n",
+    "\n",
+    "\n",
+    "class ClassifyEvent(Event):\n",
+    "    \"\"\"Event emitted after classification\"\"\"\n",
+    "\n",
+    "    markdown_content: str\n",
+    "    temp_path: str\n",
+    "    doc_type: str\n",
+    "    confidence: float\n",
+    "\n",
+    "\n",
+    "class ExtractEvent(Event):\n",
+    "    \"\"\"Event emitted after extraction\"\"\"\n",
+    "\n",
+    "    doc_type: str\n",
+    "    confidence: float\n",
+    "    extracted_data: dict\n",
+    "    markdown_length: int\n",
+    "    temp_path: str\n",
+    "    markdown_sample: str\n",
+    "\n",
+    "\n",
+    "class DocumentWorkflow(Workflow):\n",
     "    \"\"\"\n",
-    "    Complete workflow: Parse → Classify → Extract\n",
+    "    Complete document processing workflow: Parse → Classify → Extract\n",
     "    \"\"\"\n",
-    "    print(f\"🚀 Starting complete workflow\")\n",
-    "    print(\"=\" * 60)\n",
     "\n",
-    "    # Step 1: Classify\n",
-    "    print(\"🏷️  Step 2: Classifying document...\")\n",
+    "    def __init__(\n",
+    "        self,\n",
+    "        parser,\n",
+    "        classify_client,\n",
+    "        classification_rules,\n",
+    "        llama_extract,\n",
+    "        financial_schema,\n",
+    "        technical_schema,\n",
+    "        **kwargs,\n",
+    "    ):\n",
+    "        super().__init__(**kwargs)\n",
+    "        self.parser = parser\n",
+    "        self.classify_client = classify_client\n",
+    "        self.classification_rules = classification_rules\n",
+    "        self.llama_extract = llama_extract\n",
+    "        self.financial_schema = financial_schema\n",
+    "        self.technical_schema = technical_schema\n",
     "\n",
-    "    with tempfile.NamedTemporaryFile(\n",
-    "        mode=\"w\", suffix=\".md\", delete=False, encoding=\"utf-8\"\n",
-    "    ) as tmp:\n",
-    "        tmp.write(markdown_content)\n",
-    "        temp_path = Path(tmp.name)\n",
+    "    @step\n",
+    "    async def parse_document(self, ctx: Context, ev: StartEvent) -> ParseEvent:\n",
+    "        \"\"\"\n",
+    "        Step 1: Parse the document to extract markdown\n",
+    "        \"\"\"\n",
+    "        file_path = ev.file_path\n",
+    "        print(f\"📄 Step 1: Parsing document: {file_path}...\")\n",
     "\n",
-    "    print(temp_path)\n",
+    "        # Parse the document\n",
+    "        parse_result = await self.parser.aparse(file_path)\n",
+    "        markdown_content = await parse_result.aget_markdown()\n",
+    "        job_id = parse_result.job_id\n",
     "\n",
-    "    classification = await classify_client.aclassify_file_path(\n",
-    "        rules=classification_rules, file_input_path=str(temp_path)\n",
-    "    )\n",
-    "    doc_type = classification.items[0].result.type\n",
-    "    confidence = classification.items[0].result.confidence\n",
-    "    print(f\"   ✅ Classified as: {doc_type} (confidence: {confidence:.2f})\")\n",
+    "        print(f\"   ✅ Parsed successfully (Job ID: {job_id})\")\n",
+    "        print(f\"   📝 Extracted {len(markdown_content):,} characters\")\n",
     "\n",
-    "    # Step 2: Extract based on classification\n",
-    "    print(\"🔍 Step 3: Extracting structured data using SourceText...\")\n",
-    "    source_text = SourceText(\n",
-    "        text_content=markdown_content,\n",
-    "        filename=f\"{os.path.basename(temp_path)}_markdown.md\",\n",
-    "    )\n",
+    "        # Write event to stream for monitoring\n",
+    "        parse_event = ParseEvent(\n",
+    "            file_path=file_path,\n",
+    "            markdown_content=markdown_content,\n",
+    "            job_id=job_id,\n",
+    "        )\n",
+    "        ctx.write_event_to_stream(parse_event)\n",
     "\n",
-    "    # Choose schema based on classification\n",
-    "    if \"financial\" in doc_type.lower():\n",
-    "        schema = FinancialMetrics\n",
-    "        print(\"   📊 Using FinancialMetrics schema\")\n",
-    "    elif \"technical\" in doc_type.lower():\n",
-    "        schema = TechnicalSpec\n",
-    "        print(\"   🔧 Using TechnicalSpec schema\")\n",
-    "    else:\n",
-    "        schema = FinancialMetrics  # Default fallback\n",
-    "        print(\"   📊 Using default FinancialMetrics schema\")\n",
+    "        return parse_event\n",
     "\n",
-    "    extract_config = ExtractConfig(\n",
-    "        extraction_mode=\"BALANCED\",\n",
-    "    )\n",
+    "    @step\n",
+    "    async def classify_document(self, ctx: Context, ev: ParseEvent) -> ClassifyEvent:\n",
+    "        \"\"\"\n",
+    "        Step 2: Classify the document based on its content\n",
+    "        \"\"\"\n",
+    "        markdown_content = ev.markdown_content\n",
+    "        print(\"🏷️  Step 2: Classifying document...\")\n",
     "\n",
-    "    extraction_result = llama_extract.extract(\n",
-    "        data_schema=schema, config=extract_config, files=source_text\n",
-    "    )\n",
+    "        # Write markdown to temp file for classification\n",
+    "        with tempfile.NamedTemporaryFile(\n",
+    "            mode=\"w\", suffix=\".md\", delete=False, encoding=\"utf-8\"\n",
+    "        ) as tmp:\n",
+    "            tmp.write(markdown_content)\n",
+    "            temp_path = Path(tmp.name)\n",
     "\n",
-    "    print(\"   ✅ Extraction complete!\")\n",
+    "        # Classify the document\n",
+    "        classification = await self.classify_client.aclassify_file_path(\n",
+    "            rules=self.classification_rules, file_input_path=str(temp_path)\n",
+    "        )\n",
+    "        doc_type = classification.items[0].result.type\n",
+    "        confidence = classification.items[0].result.confidence\n",
     "\n",
-    "    return {\n",
-    "        \"file_path\": temp_path,\n",
-    "        \"markdown_length\": len(markdown_content),\n",
-    "        \"classification\": doc_type,\n",
-    "        \"confidence\": confidence,\n",
-    "        \"extracted_data\": extraction_result.data,\n",
-    "        \"markdown_sample\": markdown_content[:200] + \"...\"\n",
-    "        if len(markdown_content) > 200\n",
-    "        else markdown_content,\n",
-    "    }\n",
+    "        print(f\"   ✅ Classified as: {doc_type} (confidence: {confidence:.2f})\")\n",
+    "\n",
+    "        # Write event to stream for monitoring\n",
+    "        classify_event = ClassifyEvent(\n",
+    "            markdown_content=markdown_content,\n",
+    "            temp_path=str(temp_path),\n",
+    "            doc_type=doc_type,\n",
+    "            confidence=confidence,\n",
+    "        )\n",
+    "        ctx.write_event_to_stream(classify_event)\n",
+    "\n",
+    "        return classify_event\n",
+    "\n",
+    "    @step\n",
+    "    async def extract_data(self, ctx: Context, ev: ClassifyEvent) -> ExtractEvent:\n",
+    "        \"\"\"\n",
+    "        Step 3: Extract structured data based on classification\n",
+    "        \"\"\"\n",
+    "        print(\"🔍 Step 3: Extracting structured data using SourceText...\")\n",
+    "\n",
+    "        # Choose schema based on classification\n",
+    "        if \"financial\" in ev.doc_type.lower():\n",
+    "            schema = self.financial_schema\n",
+    "            print(\"   📊 Using FinancialMetrics schema\")\n",
+    "        elif \"technical\" in ev.doc_type.lower():\n",
+    "            schema = self.technical_schema\n",
+    "            print(\"   🔧 Using TechnicalSpec schema\")\n",
+    "        else:\n",
+    "            schema = self.financial_schema  # Default fallback\n",
+    "            print(\"   📊 Using default FinancialMetrics schema\")\n",
+    "\n",
+    "        # Create SourceText from markdown content\n",
+    "        source_text = SourceText(\n",
+    "            text_content=ev.markdown_content,\n",
+    "            filename=f\"{os.path.basename(ev.temp_path)}_markdown.md\",\n",
+    "        )\n",
+    "\n",
+    "        # Configure extraction\n",
+    "        extract_config = ExtractConfig(\n",
+    "            extraction_mode=\"BALANCED\",\n",
+    "        )\n",
+    "\n",
+    "        # Perform extraction\n",
+    "        extraction_result = self.llama_extract.extract(\n",
+    "            data_schema=schema, config=extract_config, files=source_text\n",
+    "        )\n",
+    "\n",
+    "        print(\"   ✅ Extraction complete!\")\n",
+    "\n",
+    "        # Create markdown sample\n",
+    "        markdown_sample = (\n",
+    "            ev.markdown_content[:200] + \"...\"\n",
+    "            if len(ev.markdown_content) > 200\n",
+    "            else ev.markdown_content\n",
+    "        )\n",
+    "\n",
+    "        extract_event = ExtractEvent(\n",
+    "            doc_type=ev.doc_type,\n",
+    "            confidence=ev.confidence,\n",
+    "            extracted_data=extraction_result.data,\n",
+    "            markdown_length=len(ev.markdown_content),\n",
+    "            temp_path=ev.temp_path,\n",
+    "            markdown_sample=markdown_sample,\n",
+    "        )\n",
+    "        ctx.write_event_to_stream(extract_event)\n",
+    "\n",
+    "        return extract_event\n",
+    "\n",
+    "    @step\n",
+    "    async def finalize_results(self, ctx: Context, ev: ExtractEvent) -> StopEvent:\n",
+    "        \"\"\"\n",
+    "        Step 4: Finalize and return results\n",
+    "        \"\"\"\n",
+    "        result = {\n",
+    "            \"file_path\": ev.temp_path,\n",
+    "            \"markdown_length\": ev.markdown_length,\n",
+    "            \"classification\": ev.doc_type,\n",
+    "            \"confidence\": ev.confidence,\n",
+    "            \"extracted_data\": ev.extracted_data,\n",
+    "            \"markdown_sample\": ev.markdown_sample,\n",
+    "        }\n",
+    "\n",
+    "        return StopEvent(result=result)\n",
     "\n",
     "\n",
-    "print(\"🔧 Workflow function defined!\")"
+    "print(\"🔧 Workflow defined!\")"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Run Complete Workflow on Both Documents"
+    "### Workflow Structure\n",
+    "\n",
+    "The workflow consists of four steps connected by typed events:\n",
+    "\n",
+    "```\n",
+    "┌─────────────┐\n",
+    "│ StartEvent  │  (file_path)\n",
+    "└──────┬──────┘\n",
+    "       │\n",
+    "       ▼\n",
+    "┌──────────────────┐\n",
+    "│ parse_document   │  Step 1: Parse PDF to markdown\n",
+    "└──────┬───────────┘\n",
+    "       │\n",
+    "       ▼\n",
+    "┌─────────────┐\n",
+    "│  ParseEvent │  (markdown_content, job_id)\n",
+    "└──────┬──────┘\n",
+    "       │\n",
+    "       ▼\n",
+    "┌─────────────────────┐\n",
+    "│ classify_document   │  Step 2: Classification\n",
+    "└──────┬──────────────┘\n",
+    "       │\n",
+    "       ▼\n",
+    "┌──────────────┐\n",
+    "│ ClassifyEvent│  (doc_type, confidence, markdown_content)\n",
+    "└──────┬───────┘\n",
+    "       │\n",
+    "       ▼\n",
+    "┌──────────────┐\n",
+    "│ extract_data │  Step 3: Extraction with schema selection\n",
+    "└──────┬───────┘\n",
+    "       │\n",
+    "       ▼\n",
+    "┌──────────────┐\n",
+    "│ ExtractEvent │  (extracted_data, doc_type, confidence)\n",
+    "└──────┬───────┘\n",
+    "       │\n",
+    "       ▼\n",
+    "┌──────────────────┐\n",
+    "│ finalize_results │  Step 4: Format and return results\n",
+    "└──────┬───────────┘\n",
+    "       │\n",
+    "       ▼\n",
+    "┌─────────────┐\n",
+    "│  StopEvent  │  (final result dictionary)\n",
+    "└─────────────┘\n",
+    "```\n",
+    "\n",
+    "**Key Features:**\n",
+    "- **Step 1 (parse_document)**: Takes a file path and parses the document into clean markdown\n",
+    "- **Step 2 (classify_document)**: Takes markdown content and classifies it into document types\n",
+    "- **Step 3 (extract_data)**: Selects appropriate schema based on classification and extracts structured data\n",
+    "- **Step 4 (finalize_results)**: Packages all results into final output format\n",
+    "- Events are written to the stream for real-time monitoring\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Visualize the Workflow\n",
+    "\n",
+    "Let's visualize the workflow structure to see the flow of events:\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Initialize the workflow\n",
+    "workflow = DocumentWorkflow(\n",
+    "    parser=parser,\n",
+    "    classify_client=classify_client,\n",
+    "    classification_rules=classification_rules,\n",
+    "    llama_extract=llama_extract,\n",
+    "    financial_schema=FinancialMetrics,\n",
+    "    technical_schema=TechnicalSpec,\n",
+    "    timeout=300,\n",
+    "    verbose=True,\n",
+    ")"
    ]
   },
   {
@@ -552,53 +834,173 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "🚀 Starting complete workflow\n",
-      "============================================================\n",
-      "🏷️  Step 2: Classifying document...\n",
-      "/var/folders/g6/4b5lpp5974gcpr890ybhbw4r0000gn/T/tmpos3b62tm.md\n",
-      "   ✅ Classified as: financial_document (confidence: 1.00)\n",
-      "🔍 Step 3: Extracting structured data using SourceText...\n",
-      "   📊 Using FinancialMetrics schema\n",
-      "..   ✅ Extraction complete!\n",
-      "\n",
-      "============================================================\n",
-      "\n",
-      "🚀 Starting complete workflow\n",
-      "============================================================\n",
-      "🏷️  Step 2: Classifying document...\n",
-      "/var/folders/g6/4b5lpp5974gcpr890ybhbw4r0000gn/T/tmpppz9ub_m.md\n",
-      "   ✅ Classified as: technical_specification (confidence: 1.00)\n",
-      "🔍 Step 3: Extracting structured data using SourceText...\n",
-      "   🔧 Using TechnicalSpec schema\n",
-      "   ✅ Extraction complete!\n",
-      "\n",
-      "============================================================\n",
-      "\n",
-      "📋 Processed 2 documents successfully!\n"
+      "document_workflow.html\n"
      ]
     }
    ],
    "source": [
-    "# Process both documents through the complete workflow\n",
-    "results = []\n",
+    "# Draw the workflow visualization\n",
+    "from llama_index.utils.workflow import draw_all_possible_flows\n",
     "\n",
-    "for doc_text in document_texts:\n",
-    "    try:\n",
-    "        result = await complete_document_workflow(doc_text)\n",
-    "        results.append(result)\n",
-    "        print(\"\\n\" + \"=\" * 60 + \"\\n\")\n",
-    "    except Exception as e:\n",
-    "        print(f\"❌ Error processing {doc_path}: {str(e)}\")\n",
-    "        print(\"\\n\" + \"=\" * 60 + \"\\n\")\n",
-    "\n",
-    "print(f\"📋 Processed {len(results)} documents successfully!\")"
+    "draw_all_possible_flows(\n",
+    "    workflow,\n",
+    "    filename=\"document_workflow.html\",\n",
+    ")"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Final Results Summary"
+    "The workflow has been visualized and saved to `document_workflow.html`. You can open this file in a browser to see the interactive workflow diagram.\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "The workflow visualization shows:\n",
+    "1. **StartEvent** → **parse_document** step\n",
+    "2. **ParseEvent** → **classify_document** step\n",
+    "3. **ClassifyEvent** → **extract_data** step  \n",
+    "4. **ExtractEvent** → **finalize_results** step\n",
+    "5. **StopEvent** (final output)\n",
+    "\n",
+    "Each step is connected by typed events, allowing for clean separation of concerns and easy monitoring of the workflow execution.\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Run the Workflow on Both Documents\n",
+    "\n",
+    "Now let's run the workflow on both documents and monitor the events:\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "======================================================================\n",
+      "🚀 Processing Document 1: sample_docs/financial_report.pdf\n",
+      "======================================================================\n",
+      "\n",
+      "Running step parse_document\n",
+      "📄 Step 1: Parsing document: sample_docs/financial_report.pdf...\n",
+      "Started parsing the file under job_id bb53c6bf-79cc-4f63-9c97-16983d59f29d\n",
+      ".   ✅ Parsed successfully (Job ID: bb53c6bf-79cc-4f63-9c97-16983d59f29d)\n",
+      "   📝 Extracted 1,338,499 characters\n",
+      "Step parse_document produced event ParseEvent\n",
+      "📄 Parse Event: Extracted 1,338,499 characters\n",
+      "Running step classify_document\n",
+      "🏷️  Step 2: Classifying document...\n",
+      "   ✅ Classified as: financial_document (confidence: 1.00)\n",
+      "Step classify_document produced event ClassifyEvent\n",
+      "📊 Classification Event: financial_document (1.00)\n",
+      "Running step extract_data\n",
+      "🔍 Step 3: Extracting structured data using SourceText...\n",
+      "   📊 Using FinancialMetrics schema\n",
+      "..   ✅ Extraction complete!\n",
+      "Step extract_data produced event ExtractEvent\n",
+      "Running step finalize_results\n",
+      "Step finalize_results produced event StopEvent\n",
+      "✅ Extraction Event: 7 fields extracted\n",
+      "\n",
+      "✅ Document 1 processed successfully!\n",
+      "\n",
+      "======================================================================\n",
+      "🚀 Processing Document 2: sample_docs/technical_spec.pdf\n",
+      "======================================================================\n",
+      "\n",
+      "Running step parse_document\n",
+      "📄 Step 1: Parsing document: sample_docs/technical_spec.pdf...\n",
+      "Started parsing the file under job_id 944905c1-3c49-431a-ad86-4436d16f3d1c\n",
+      "   ✅ Parsed successfully (Job ID: 944905c1-3c49-431a-ad86-4436d16f3d1c)\n",
+      "   📝 Extracted 92,483 characters\n",
+      "Step parse_document produced event ParseEvent\n",
+      "📄 Parse Event: Extracted 92,483 characters\n",
+      "Running step classify_document\n",
+      "🏷️  Step 2: Classifying document...\n",
+      "   ✅ Classified as: technical_specification (confidence: 1.00)\n",
+      "Step classify_document produced event ClassifyEvent\n",
+      "📊 Classification Event: technical_specification (1.00)\n",
+      "Running step extract_data\n",
+      "🔍 Step 3: Extracting structured data using SourceText...\n",
+      "   🔧 Using TechnicalSpec schema\n",
+      "   ✅ Extraction complete!\n",
+      "Step extract_data produced event ExtractEvent\n",
+      "Running step finalize_results\n",
+      "Step finalize_results produced event StopEvent\n",
+      "✅ Extraction Event: 8 fields extracted\n",
+      "\n",
+      "✅ Document 2 processed successfully!\n",
+      "\n",
+      "\n",
+      "📋 Processed 2 documents successfully!\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Process both documents through the workflow\n",
+    "results = []\n",
+    "\n",
+    "# Define the document files to process\n",
+    "document_files = [\n",
+    "    \"sample_docs/financial_report.pdf\",\n",
+    "    \"sample_docs/technical_spec.pdf\",\n",
+    "]\n",
+    "\n",
+    "for i, file_path in enumerate(document_files, 1):\n",
+    "    print(f\"\\n{'='*70}\")\n",
+    "    print(f\"🚀 Processing Document {i}: {file_path}\")\n",
+    "    print(f\"{'='*70}\\n\")\n",
+    "\n",
+    "    try:\n",
+    "        # Run the workflow\n",
+    "        handler = workflow.run(file_path=file_path)\n",
+    "\n",
+    "        # Monitor events as they are emitted\n",
+    "        async for event in handler.stream_events():\n",
+    "            if isinstance(event, ParseEvent):\n",
+    "                print(\n",
+    "                    f\"📄 Parse Event: Extracted {len(event.markdown_content):,} characters\"\n",
+    "                )\n",
+    "            elif isinstance(event, ClassifyEvent):\n",
+    "                print(\n",
+    "                    f\"📊 Classification Event: {event.doc_type} ({event.confidence:.2f})\"\n",
+    "                )\n",
+    "            elif isinstance(event, ExtractEvent):\n",
+    "                print(\n",
+    "                    f\"✅ Extraction Event: {len(event.extracted_data)} fields extracted\"\n",
+    "                )\n",
+    "\n",
+    "        # Get final result\n",
+    "        result = await handler\n",
+    "        results.append(result)\n",
+    "\n",
+    "        print(f\"\\n✅ Document {i} processed successfully!\")\n",
+    "\n",
+    "    except Exception as e:\n",
+    "        print(f\"❌ Error processing document {i}: {str(e)}\")\n",
+    "        import traceback\n",
+    "\n",
+    "        traceback.print_exc()\n",
+    "\n",
+    "print(f\"\\n\\n📋 Processed {len(results)} documents successfully!\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Final Results Summary\n"
    ]
   },
   {
@@ -613,9 +1015,9 @@
       "📈 COMPLETE WORKFLOW RESULTS SUMMARY\n",
       "======================================================================\n",
       "\n",
-      "📄 Document 1: tmpos3b62tm.md\n",
+      "📄 Document 1: tmpuyxzpd3x.md\n",
       "   📊 Classification: financial_document (confidence: 1.00)\n",
-      "   📝 Markdown length: 1,348,671 characters\n",
+      "   📝 Markdown length: 1,338,499 characters\n",
       "   📋 Markdown sample: \n",
       "\n",
       "# UNITED STATES\n",
@@ -629,14 +1031,14 @@
       "   • company_name: Uber Technologies, Inc.\n",
       "   • document_type: Annual Report on Form 10-K\n",
       "   • fiscal_year: 2021\n",
-      "   • revenue_2021: $21,764\n",
-      "   • net_income_2021: $(496)\n",
-      "   • key_business_segments: ['Mobility', 'Delivery', 'Freight', 'All Other (including former New Mobility, e-bikes, e-scooters, Advanced Technologies Group and other technology programs)']\n",
-      "   • risk_factors: [\"The company faces numerous risk factors across its business operations and environment. The COVID-19 pandemic and related mitigation measures have adversely affected parts of the business, including reduced demand for Mobility offerings and creating ongoing uncertainties. The company's operational and financial performance is influenced by competitive pressure in the mobility, delivery, and logistics industries, characterized by well-established alternatives, low barriers to entry, and low switching costs. Driver classification risks exist if Drivers are deemed employees, workers, or quasi-employees rather than independent contractors, exposing the company to legal actions and financial liabilities globally. Competition challenges require the company to sometimes lower fares, offer incentives, and promotions, which impacts profitability. There are significant operating losses historically with substantial future operating expense increases anticipated, and the ability to achieve or maintain profitability is uncertain. Network value depends on maintaining critical mass among Drivers, consumers, merchants, shippers, and carriers, and failures to do so diminish platform attractiveness. Brand and reputation maintenance is critical, with exposure to negative publicity, media coverage, and risks from associated companies' brands or licensed brands in joint ventures.\\n\\nOperational risks include historical workplace culture and compliance challenges, management complexity due to rapid growth, technological infrastructure issues potentially causing disruptions or poor user experience, and security or data privacy breaches that could impact revenue and reputation. Platform users may engage in or be subjected to criminal, violent, or dangerous activity leading to safety incidents and legal actions. New offerings and technologies investments are inherently risky without guaranteed benefits. Economic conditions, inflation, and increased costs (fuel, food, labor, energy) may negatively impact results. Regulatory risks are extensive and global, involving payment and financial services compliance, licensing, anti-money laundering laws, data privacy (GDPR, CCPA, LGPD), and labor laws. Legal and regulatory investigations and inquiries, including antitrust, FCPA, labor classification, data protection, and intellectual property matters, pose risks of fines, penalties, operational changes, and increased costs.\\n\\nGeopolitical and jurisdictional risks include operating limitations or bans in some locations, currency exchange risk, and complex evolving regulations with the potential for fines and loss of licenses or permits. Insurance risks include potential inadequacy of reserves, liability exposure from accidents or impersonation, and insurer insolvency. Driver qualification requirements and background checks may increase costs or fail to expose all relevant information, with associated insurance cost risks and potential for courtroom or regulatory challenges to pricing models.\\n\\nFinancial risks comprise significant accumulated deficits, requirement for additional capital with uncertain availability, debt obligations, tax exposure including uncertain positions and observed changes in tax laws, and volatility in common stock price with no expected cash dividends. Accounting judgments and estimates involve critical assumptions affecting reported financial metrics related to goodwill, revenue recognition, incentive accruals, and stock-based compensation. Cybersecurity risks include exposures to malware, ransomware, phishing, and other cyberattacks. Climate change presents physical and transitional risks that may impact operations and costs, and failure to meet climate commitments may have operational and reputational consequences.\\n\\nOther risks include potential liability under anti-corruption and anti-terrorism laws, adverse effects from defaults under debt agreements, limitations in takeover actions due to corporate governance provisions, and the impact of non-GAAP financial measure limitations. Overall, these diverse and interconnected risk factors contribute to significant uncertainty regarding the company's future business prospects, operating results, and financial condition.\"]\n",
+      "   • revenue_2021: $17,455 and $21,764\n",
+      "   • net_income_2021: $(496) to (700)\n",
+      "   • key_business_segments: ['Borrower and the Restricted Subsidiaries', 'Holdings', 'Guarantors', 'Material Domestic Subsidiaries', 'Material Foreign Subsidiaries']\n",
+      "   • risk_factors: ['Indemnification obligations of the borrower for losses, claims, damages, liabilities, and out-of-pocket expenses incurred by agents, lenders, arrangers, and related parties in connection with the agreement or loans, except in certain cases such as gross negligence, bad faith, willful misconduct, or material breach by the indemnitee.', \"Borrower not required to indemnify any indemnitee for settlements entered into without the borrower's consent.\", 'Limitation of liability for special, indirect, consequential, or punitive damages, and for damages from unauthorized use of information, except for direct damages resulting from gross negligence, bad faith, or willful misconduct.', 'Obligation of the borrower to indemnify the administrative agent for liabilities arising from performance of duties, except in cases of gross negligence, bad faith, or willful misconduct.', 'Limitations and conditions on assignments and participations of lender rights, including restrictions on assignments to disqualified institutions, loan parties, affiliates of loan parties, defaulting lenders, and natural persons.', 'Setoff rights for lenders and issuing banks after an event of default, allowing them to apply borrower deposits toward obligations under the agreement.', 'Potential for increased obligations under the agreement as a result of changes in law affecting payment terms.', 'Requirement for the borrower and guarantors to provide information to comply with anti-money laundering rules and the USA PATRIOT Act.']\n",
       "\n",
-      "📄 Document 2: tmpppz9ub_m.md\n",
+      "📄 Document 2: tmp7ower2xm.md\n",
       "   📊 Classification: technical_specification (confidence: 1.00)\n",
-      "   📝 Markdown length: 90,971 characters\n",
+      "   📝 Markdown length: 92,483 characters\n",
       "   📋 Markdown sample: \n",
       "\n",
       "LM317\n",
@@ -648,20 +1050,14 @@
       "   🎯 Extracted fields: 8 fields\n",
       "   • component_name: LM317\n",
       "   • manufacturer: Texas Instruments\n",
-      "   • part_number: LM317\n",
-      "   • description: The LM317 is an adjustable three-pin, positive-voltage regulator capable of supplying up to 1.5A over an output voltage range of 1.25V to 37V. It features line and load regulation, internal current limiting, thermal overload protection, and safe operating area compensation.\n",
+      "   • part_number: LM317, SLVS044Z\n",
+      "   • description: The LM317 is an adjustable three-pin, positive-voltage regulator capable of supplying more than 1.5A (typically up to 1.5A) over an output voltage range of 1.25V to 37V. The device requires only two external resistors to set the output voltage. It features a typical line regulation of 0.01% and typical load regulation of 0.1%. The LM317 includes current limiting, thermal overload protection, and safe operating area protection. Overload protection remains functional even if the ADJUST pin is disconnected. The regulator is used in applications such as constant-current battery-charger circuits, slow turn-on 15V regulator circuits, AC voltage-regulator circuits, current-limited charger circuits, and high-current and adjustable regulator circuits. It is available in packages including SOT-223 (DCY), TO-220 (KCS), and TO-263 (KTT).\n",
       "   • operating_voltage: {'min_voltage': 1.25, 'max_voltage': 37.0, 'unit': 'V'}\n",
-      "   • maximum_current: 1.5\n",
-      "   • key_features: ['Adjustable output voltage: 1.25V to 37V', 'Output current up to 1.5A', 'Line regulation: 0.01%/V (typical)', 'Load regulation: 0.1% (typical)', 'Internal short-circuit current limiting', 'Thermal overload protection', 'Output safe-area compensation', 'High power supply rejection ratio (PSRR): 80dB at 120Hz (new chip)', 'Available in SOT-223, TO-263, and TO-220 packages']\n",
-      "   • applications: ['Multifunction printers', 'AC drive power stage modules', 'Electricity meters', 'Servo drive control modules', 'Merchant network and server power supply units']\n",
+      "   • maximum_current: 4.0\n",
+      "   • key_features: ['Adjustable output voltage range: 1.25V to 37V', 'Output current up to 1.5A (up to 4A with external pass elements)', 'Line regulation: typically 0.01%/V', 'Load regulation: typically 0.1%', 'Internal short-circuit current limiting / Current limiting', 'Thermal overload protection / Thermal shutdown', 'Output safe-area compensation / Safe operating area protection', 'PSRR: 80dB at 120Hz for CADJ = 10μF (new chip)', 'NPN Darlington output drive', 'Programmable feedback', 'Multiple package options (SOT-223, TO-220, TO-263)', 'Can be used in constant-current, battery-charging, and regulator applications']\n",
+      "   • applications: ['Multifunction printers, AC drive power stage modules, Electricity meters, Servo drive control modules, Merchant network and server PSU, Adjustable voltage regulator, 0V to 30V regulator circuit, Regulator circuit with improved ripple rejection, Precision current-limiter, Tracking preregulator, 1.25V to 20V regulator, Battery charger circuit, Constant-current battery charger circuits, Slow turn-on regulator, AC voltage-regulator, Current-limited charger circuits, High-current adjustable regulator circuits, General-purpose adjustable power supply']\n",
       "\n",
-      "✨ Workflow completed successfully!\n",
-      "\n",
-      "📚 Key Learnings:\n",
-      "   • Parse: Converted documents to clean markdown format\n",
-      "   • Classify: Automatically categorized document types\n",
-      "   • Extract: Used SourceText with markdown for structured data extraction\n",
-      "   • The markdown content provides much better context for extraction than raw PDFs\n"
+      "✨ Workflow completed successfully!\n"
      ]
     }
    ],
@@ -683,14 +1079,7 @@
     "    for key, value in extracted.items():\n",
     "        print(f\"   • {key}: {value}\")\n",
     "\n",
-    "print(\"\\n✨ Workflow completed successfully!\")\n",
-    "print(\"\\n📚 Key Learnings:\")\n",
-    "print(\"   • Parse: Converted documents to clean markdown format\")\n",
-    "print(\"   • Classify: Automatically categorized document types\")\n",
-    "print(\"   • Extract: Used SourceText with markdown for structured data extraction\")\n",
-    "print(\n",
-    "    \"   • The markdown content provides much better context for extraction than raw PDFs\"\n",
-    ")"
+    "print(\"\\n✨ Workflow completed successfully!\")"
    ]
   },
   {
@@ -699,54 +1088,33 @@
    "source": [
     "## Conclusion\n",
     "\n",
-    "This notebook demonstrated the complete **Parse → Classify → Extract** workflow using LlamaCloud services:\n",
+    "The notebook shows you how to build an e2e document **Classify → Extract** workflow using LlamaCloud. This uses some of our core building blocks around **classification** interleaved with **document extraction**.\n",
     "\n",
-    "### Key Components:\n",
+    "### Main Components:\n",
     "\n",
     "1. **LlamaParse** (`llama_cloud_services.parse.base.LlamaParse`):\n",
     "   - Converts documents to clean, structured markdown\n",
     "   - Preserves document structure and formatting\n",
     "   - Handles various file types (PDF, DOCX, etc.)\n",
     "\n",
-    "2. **ClassifyClient** (`llama_cloud_services.beta.classifier.client.ClassifyClient`):\n",
+    "2. **LlamaClassify** (`llama_cloud_services.beta.classifier.client.LlamaClassify`):\n",
     "   - Automatically categorizes documents based on content\n",
     "   - Uses customizable rules for classification\n",
     "   - Provides confidence scores for classifications\n",
     "\n",
     "3. **LlamaExtract with SourceText** (`llama_cloud_services.extract.extract.LlamaExtract`, `SourceText`):\n",
     "   - Extracts structured data using custom Pydantic schemas\n",
-    "   - **SourceText** allows using markdown content as input instead of raw files\n",
-    "   - Provides much better extraction accuracy when using processed markdown\n",
+    "   - You can either feed in the file directly (in which case parsing will happen under the hood), or the parsed text through the **SourceText** object (which is the case in this example) \n",
     "\n",
-    "### Workflow Benefits:\n",
-    "\n",
-    "- **Better Accuracy**: Using markdown from parsing provides cleaner, more structured input for extraction\n",
-    "- **Automatic Routing**: Classification allows different processing logic for different document types\n",
-    "- **Structured Output**: Custom schemas ensure consistent, structured data extraction\n",
-    "- **Flexible Input**: SourceText supports text content, file paths, and bytes\n",
-    "\n",
-    "### Key Insights:\n",
-    "\n",
-    "1. **SourceText is the bridge**: It allows you to pass the clean markdown content from parsing directly to extraction\n",
-    "2. **Markdown improves extraction**: Pre-processed markdown provides much better context than raw PDFs\n",
-    "3. **Classification enables smart routing**: Different document types can use different extraction schemas\n",
-    "4. **End-to-end automation**: The entire workflow can be automated for production use\n",
-    "\n",
-    "This approach is ideal for production document processing pipelines where you need to:\n",
-    "- Process various document types automatically\n",
-    "- Extract structured data consistently\n",
-    "- Maintain high accuracy and reliability\n",
-    "- Handle documents at scale\n",
-    "\n",
-    "The combination of these three services provides a powerful, flexible document processing pipeline that can handle complex, real-world document processing requirements."
+    "**Benefits of an e2e workflow**: The main benefit of doing Classify -> Extract, instead of only Extract, is the fact that you can handle documents of different types/different expected schemas within the same workflow, without having to separate out the data before and running separate extractions on each data subset. "
    ]
   }
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
+   "display_name": "llama_parse",
    "language": "python",
-   "name": "python3"
+   "name": "llama_parse"
   },
   "language_info": {
    "codemirror_mode": {

examples/sheets/coding-agent-analysis/CLAUDE.md

@@ -0,0 +1,73 @@
+This project uses LlamaSheets to extract data from spreadsheets for analysis.
+
+## Current Project Structure
+
+- `data/` - Contains extracted parquet files from LlamaSheets
+  - `{name}_region_{N}.parquet` - Table data files
+  - `{name}_metadata_{N}.parquet` - Cell metadata files
+  - `{name}_job_metadata.json` - Extraction job information
+- `scripts/` - Analysis and helper scripts
+- `reports/` - Your generated reports and outputs
+
+## Working with LlamaSheets Data
+
+### Understanding the Files
+
+When a spreadsheet is extracted, you'll find:
+
+1. **Table parquet files** (`region_*.parquet`): The actual table data
+   - Columns correspond to spreadsheet columns
+   - Data types are preserved (dates, numbers, strings, booleans)
+
+2. **Metadata parquet files** (`metadata_*.parquet`): Rich cell-level metadata
+   - Formatting: `font_bold`, `font_italic`, `font_size`, `background_color_rgb`
+   - Position: `row_number`, `column_number`, `coordinate` (e.g., "A1")
+   - Type detection: `data_type`, `is_date_like`, `is_percentage`, `is_currency`
+   - Layout: `is_in_first_row`, `is_merged_cell`, `horizontal_alignment`
+   - Content: `cell_value`, `raw_cell_value`
+
+3. **Job metadata JSON** (`job_metadata.json`): Overall extraction results
+   - `regions[]`: List of extracted regions with IDs, locations, and titles/descriptions
+   - `worksheet_metadata[]`: Generated titles and descriptions
+   - `status`: Success/failure status
+
+### Key Principles
+
+1. **Use metadata to understand structure**: Bold cells often indicate headers, colors indicate groupings
+2. **Validate before analysis**: Check data types, look for missing values
+3. **Preserve formatting context**: The metadata tells you what the spreadsheet author emphasized
+4. **Save intermediate results**: Store cleaned data as new parquet files
+
+### Common Patterns
+
+**Loading data:**
+```python
+import pandas as pd
+
+df = pd.read_parquet("data/region_1_Sheet1.parquet")
+meta_df = pd.read_parquet("data/metadata_1_Sheet1.parquet")
+```
+
+**Finding headers:**
+```python
+headers = meta_df[meta_df["font_bold"] == True]["cell_value"].tolist()
+```
+
+**Finding date columns:**
+```python
+date_cols = meta_df[meta_df["is_date_like"] == True]["column_number"].unique()
+```
+
+## Tools Available
+
+- **Python 3.11+**: For data analysis
+- **pandas**: DataFrame manipulation
+- **pyarrow**: Parquet file reading
+- **matplotlib**: Visualization (optional)
+
+## Guidelines
+
+- Always read the job_metadata.json first to understand what was extracted
+- Check both table data and metadata before making assumptions
+- Write reusable functions for common operations
+- Document any data quality issues discovered

examples/sheets/coding-agent-analysis/generate_sample_data.py

@@ -0,0 +1,278 @@
+"""
+Generate sample spreadsheets for LlamaSheets + Claude workflows.
+
+This script creates example Excel files that demonstrate different use cases:
+1. Simple data table (for Workflow 1)
+2. Regional sales data (for Workflow 2)
+3. Complex budget with formatting (for Workflow 3)
+4. Weekly sales report (for Workflow 4)
+
+Usage:
+    python generate_sample_data.py
+"""
+
+import random
+from datetime import datetime, timedelta
+from pathlib import Path
+
+import pandas as pd
+from openpyxl import Workbook
+from openpyxl.styles import Font, PatternFill, Alignment
+
+
+def generate_workflow_1_data(output_dir: Path) -> None:
+    """Generate simple financial report for Workflow 1."""
+    print("📊 Generating Workflow 1: financial_report_q1.xlsx")
+
+    # Create sample quarterly data
+    months = ["January", "February", "March"]
+    categories = ["Revenue", "Cost of Goods Sold", "Operating Expenses", "Net Income"]
+
+    data = []
+    for category in categories:
+        row: dict[str, str | int] = {"Category": category}
+        for month in months:
+            if category == "Revenue":
+                value = random.randint(80000, 120000)
+            elif category == "Cost of Goods Sold":
+                value = random.randint(30000, 50000)
+            elif category == "Operating Expenses":
+                value = random.randint(20000, 35000)
+            else:  # Net Income
+                value = int(
+                    int(row.get("January", 0))
+                    + int(row.get("February", 0))
+                    + int(row.get("March", 0))
+                )
+                value = random.randint(15000, 40000)
+            row[month] = value
+        data.append(row)
+
+    df = pd.DataFrame(data)
+
+    # Write to Excel
+    output_file = output_dir / "financial_report_q1.xlsx"
+    with pd.ExcelWriter(output_file, engine="openpyxl") as writer:
+        df.to_excel(writer, sheet_name="Q1 Summary", index=False)
+
+        # Format it nicely
+        worksheet = writer.sheets["Q1 Summary"]
+        for cell in worksheet[1]:  # Header row
+            cell.font = Font(bold=True)
+            cell.fill = PatternFill(
+                start_color="4F81BD", end_color="4F81BD", fill_type="solid"
+            )
+            cell.font = Font(color="FFFFFF", bold=True)
+
+    print(f"  ✅ Created {output_file}")
+
+
+def generate_workflow_2_data(output_dir: Path) -> None:
+    """Generate regional sales data for Workflow 2."""
+    print("\n📊 Generating Workflow 2: Regional sales data")
+
+    regions = ["northeast", "southeast", "west"]
+    products = ["Widget A", "Widget B", "Widget C", "Gadget X", "Gadget Y"]
+
+    for region in regions:
+        data = []
+        start_date = datetime(2024, 1, 1)
+
+        # Generate 90 days of sales data
+        for day in range(90):
+            date = start_date + timedelta(days=day)
+            # Random number of sales per day (3-8)
+            for _ in range(random.randint(3, 8)):
+                product = random.choice(products)
+                units_sold = random.randint(1, 20)
+                price_per_unit = random.randint(50, 200)
+                revenue = units_sold * price_per_unit
+
+                data.append(
+                    {
+                        "Date": date.strftime("%Y-%m-%d"),
+                        "Product": product,
+                        "Units_Sold": units_sold,
+                        "Revenue": revenue,
+                    }
+                )
+
+        df = pd.DataFrame(data)
+
+        # Write to Excel
+        output_file = output_dir / f"sales_{region}.xlsx"
+        df.to_excel(output_file, sheet_name="Sales", index=False)
+        print(f"  ✅ Created {output_file} ({len(df)} rows)")
+
+
+def generate_workflow_3_data(output_dir: Path) -> None:
+    """Generate complex budget spreadsheet with formatting for Workflow 3."""
+    print("\n📊 Generating Workflow 3: company_budget_2024.xlsx")
+
+    wb = Workbook()
+    ws = wb.active
+    ws.title = "Budget"
+
+    # Define departments with colors
+    departments = {
+        "Engineering": "C6E0B4",
+        "Marketing": "FFD966",
+        "Sales": "F4B084",
+        "Operations": "B4C7E7",
+    }
+
+    # Define categories
+    categories = {
+        "Personnel": ["Salaries", "Benefits", "Training"],
+        "Infrastructure": ["Office Rent", "Equipment", "Software Licenses"],
+        "Operations": ["Travel", "Supplies", "Miscellaneous"],
+    }
+
+    # Styles
+    header_font = Font(bold=True, size=12)
+    category_font = Font(bold=True, size=11)
+
+    row = 1
+
+    # Title
+    ws.merge_cells(f"A{row}:E{row}")
+    ws[f"A{row}"] = "2024 Annual Budget"
+    ws[f"A{row}"].font = Font(bold=True, size=14)
+    ws[f"A{row}"].alignment = Alignment(horizontal="center")
+    row += 2
+
+    # Headers
+    ws[f"A{row}"] = "Category"
+    ws[f"B{row}"] = "Item"
+    for i, dept in enumerate(departments.keys()):
+        ws.cell(row, 3 + i, dept)
+        ws.cell(row, 3 + i).font = header_font
+
+    for cell in ws[row]:
+        cell.font = header_font
+    row += 1
+
+    # Data
+    for category, items in categories.items():
+        # Category header (bold)
+        ws[f"A{row}"] = category
+        ws[f"A{row}"].font = category_font
+        row += 1
+
+        # Items with department budgets
+        for item in items:
+            ws[f"A{row}"] = ""
+            ws[f"B{row}"] = item
+
+            # Add budget amounts for each department (with color)
+            for i, (dept, color) in enumerate(departments.items()):
+                amount = random.randint(5000, 50000)
+                cell = ws.cell(row, 3 + i, amount)
+                cell.fill = PatternFill(
+                    start_color=color, end_color=color, fill_type="solid"
+                )
+                cell.number_format = "$#,##0"
+
+            row += 1
+
+        row += 1  # Blank row between categories
+
+    # Adjust column widths
+    ws.column_dimensions["A"].width = 20
+    ws.column_dimensions["B"].width = 25
+    for i in range(len(departments)):
+        ws.column_dimensions[chr(67 + i)].width = 15  # C, D, E, F
+
+    output_file = output_dir / "company_budget_2024.xlsx"
+    wb.save(output_file)
+    print(f"  ✅ Created {output_file}")
+    print("     • Bold categories, colored departments, merged title cell")
+
+
+def generate_workflow_4_data(output_dir: Path) -> None:
+    """Generate weekly sales report for Workflow 4."""
+    print("\n📊 Generating Workflow 4: sales_weekly.xlsx")
+
+    products = [
+        "Product A",
+        "Product B",
+        "Product C",
+        "Product D",
+        "Product E",
+        "Product F",
+        "Product G",
+        "Product H",
+    ]
+
+    # Generate one week of data
+    data = []
+    start_date = datetime(2024, 11, 4)  # Monday
+
+    for day in range(7):
+        date = start_date + timedelta(days=day)
+        # Each product has 3-10 transactions per day
+        for product in products:
+            for _ in range(random.randint(3, 10)):
+                units = random.randint(1, 15)
+                price = random.randint(20, 150)
+                revenue = units * price
+
+                data.append(
+                    {
+                        "Date": date.strftime("%Y-%m-%d"),
+                        "Product": product,
+                        "Units": units,
+                        "Revenue": revenue,
+                    }
+                )
+
+    df = pd.DataFrame(data)
+
+    # Write to Excel with some formatting
+    output_file = output_dir / "sales_weekly.xlsx"
+    with pd.ExcelWriter(output_file, engine="openpyxl") as writer:
+        df.to_excel(writer, sheet_name="Weekly Sales", index=False)
+
+        # Format header
+        worksheet = writer.sheets["Weekly Sales"]
+        for cell in worksheet[1]:
+            cell.font = Font(bold=True)
+
+    print(f"  ✅ Created {output_file} ({len(df)} rows)")
+
+
+def main() -> None:
+    """Generate all sample data files."""
+    print("=" * 60)
+    print("Generating Sample Data for LlamaSheets + Coding Agent Workflows")
+    print("=" * 60)
+
+    # Create output directory
+    output_dir = Path("input_data")
+    output_dir.mkdir(exist_ok=True)
+
+    # Generate data for each workflow
+    generate_workflow_1_data(output_dir)
+    generate_workflow_2_data(output_dir)
+    generate_workflow_3_data(output_dir)
+    generate_workflow_4_data(output_dir)
+
+    print("\n" + "=" * 60)
+    print("✅ All sample data generated!")
+    print("=" * 60)
+    print(f"\nFiles created in {output_dir.absolute()}:")
+    print("\nWorkflow 1 (Understanding a New Spreadsheet):")
+    print("  • financial_report_q1.xlsx")
+    print("\nWorkflow 2 (Generating Analysis Scripts):")
+    print("  • sales_northeast.xlsx")
+    print("  • sales_southeast.xlsx")
+    print("  • sales_west.xlsx")
+    print("\nWorkflow 3 (Using Cell Metadata):")
+    print("  • company_budget_2024.xlsx")
+    print("\nWorkflow 4 (Complete Automation):")
+    print("  • sales_weekly.xlsx")
+    print("\nYou can now use these files with the workflows in the documentation!")
+
+
+if __name__ == "__main__":
+    main()

examples/sheets/coding-agent-analysis/requirements.txt

@@ -0,0 +1,5 @@
+llama-cloud-services  # LlamaSheets SDK
+pandas>=2.0.0
+pyarrow>=12.0.0
+openpyxl>=3.0.0       # For Excel file support
+matplotlib>=3.7.0     # For visualizations (optional)

examples/sheets/coding-agent-analysis/scripts/extract.py

@@ -0,0 +1,100 @@
+"""Helper script to extract spreadsheets using LlamaSheets."""
+
+import asyncio
+import json
+import os
+import dotenv
+from pathlib import Path
+
+from llama_cloud_services.beta.sheets import LlamaSheets
+from llama_cloud_services.beta.sheets.types import (
+    SpreadsheetParsingConfig,
+    SpreadsheetResultType,
+)
+
+dotenv.load_dotenv()
+
+
+async def extract_spreadsheet(
+    file_path: str, output_dir: str = "data", generate_metadata: bool = True
+) -> dict:
+    """Extract a spreadsheet using LlamaSheets."""
+
+    client = LlamaSheets(
+        base_url="https://api.cloud.llamaindex.ai",
+        api_key=os.getenv("LLAMA_CLOUD_API_KEY"),
+    )
+
+    print(f"Extracting {file_path}...")
+
+    # Extract regions
+    config = SpreadsheetParsingConfig(
+        sheet_names=None,  # Extract all sheets
+        generate_additional_metadata=generate_metadata,
+    )
+
+    job_result = await client.aextract_regions(file_path, config=config)
+
+    print(f"Extracted {len(job_result.regions)} region(s)")
+
+    # Create output directory
+    output_path = Path(output_dir)
+    output_path.mkdir(parents=True, exist_ok=True)
+
+    # Get base name for files
+    base_name = Path(file_path).stem
+
+    # Save job metadata
+    job_metadata_path = output_path / f"{base_name}_job_metadata.json"
+    with open(job_metadata_path, "w") as f:
+        json.dump(job_result.model_dump(mode="json"), f, indent=2)
+    print(f"Saved job metadata to {job_metadata_path}")
+
+    # Download each region
+    for idx, region in enumerate(job_result.regions, 1):
+        sheet_name = region.sheet_name.replace(" ", "_")
+
+        # Download region data
+        region_bytes = await client.adownload_region_result(
+            job_id=job_result.id,
+            region_id=region.region_id,
+            result_type=region.region_type,
+        )
+
+        region_path = output_path / f"{base_name}_region_{idx}_{sheet_name}.parquet"
+        with open(region_path, "wb") as f:
+            f.write(region_bytes)
+        print(f"  Table {idx}: {region_path}")
+
+        # Download metadata
+        metadata_bytes = await client.adownload_region_result(
+            job_id=job_result.id,
+            region_id=region.region_id,
+            result_type=SpreadsheetResultType.CELL_METADATA,
+        )
+
+        metadata_path = output_path / f"{base_name}_metadata_{idx}_{sheet_name}.parquet"
+        with open(metadata_path, "wb") as f:
+            f.write(metadata_bytes)
+        print(f"  Metadata {idx}: {metadata_path}")
+
+    print(f"\nAll files saved to {output_path}/")
+
+    return job_result.model_dump(mode="json")
+
+
+if __name__ == "__main__":
+    import sys
+
+    if len(sys.argv) < 2:
+        print("Usage: python scripts/extract.py <spreadsheet_file>")
+        sys.exit(1)
+
+    file_path = sys.argv[1]
+
+    if not Path(file_path).exists():
+        print(f"❌ File not found: {file_path}")
+        sys.exit(1)
+
+    result = asyncio.run(extract_spreadsheet(file_path))
+    print(f"\n✅ Extraction complete! Job ID: {result['id']}")

examples/sheets/llama-index-agent/generate_sample_data.py

@@ -0,0 +1,278 @@
+"""
+Generate sample spreadsheets for LlamaSheets + LlamaIndex Agent workflows.
+
+This script creates example Excel files that demonstrate different use cases:
+1. Simple data table (for Workflow 1)
+2. Regional sales data (for Workflow 2)
+3. Complex budget with formatting (for Workflow 3)
+4. Weekly sales report (for Workflow 4)
+
+Usage:
+    python generate_sample_data.py
+"""
+
+import random
+from datetime import datetime, timedelta
+from pathlib import Path
+
+import pandas as pd
+from openpyxl import Workbook
+from openpyxl.styles import Font, PatternFill, Alignment
+
+
+def generate_workflow_1_data(output_dir: Path) -> None:
+    """Generate simple financial report for Workflow 1."""
+    print("📊 Generating Workflow 1: financial_report_q1.xlsx")
+
+    # Create sample quarterly data
+    months = ["January", "February", "March"]
+    categories = ["Revenue", "Cost of Goods Sold", "Operating Expenses", "Net Income"]
+
+    data = []
+    for category in categories:
+        row: dict[str, str | int] = {"Category": category}
+        for month in months:
+            if category == "Revenue":
+                value = random.randint(80000, 120000)
+            elif category == "Cost of Goods Sold":
+                value = random.randint(30000, 50000)
+            elif category == "Operating Expenses":
+                value = random.randint(20000, 35000)
+            else:  # Net Income
+                value = int(
+                    int(row.get("January", 0))
+                    + int(row.get("February", 0))
+                    + int(row.get("March", 0))
+                )
+                value = random.randint(15000, 40000)
+            row[month] = value
+        data.append(row)
+
+    df = pd.DataFrame(data)
+
+    # Write to Excel
+    output_file = output_dir / "financial_report_q1.xlsx"
+    with pd.ExcelWriter(output_file, engine="openpyxl") as writer:
+        df.to_excel(writer, sheet_name="Q1 Summary", index=False)
+
+        # Format it nicely
+        worksheet = writer.sheets["Q1 Summary"]
+        for cell in worksheet[1]:  # Header row
+            cell.font = Font(bold=True)
+            cell.fill = PatternFill(
+                start_color="4F81BD", end_color="4F81BD", fill_type="solid"
+            )
+            cell.font = Font(color="FFFFFF", bold=True)
+
+    print(f"  ✅ Created {output_file}")
+
+
+def generate_workflow_2_data(output_dir: Path) -> None:
+    """Generate regional sales data for Workflow 2."""
+    print("\n📊 Generating Workflow 2: Regional sales data")
+
+    regions = ["northeast", "southeast", "west"]
+    products = ["Widget A", "Widget B", "Widget C", "Gadget X", "Gadget Y"]
+
+    for region in regions:
+        data = []
+        start_date = datetime(2024, 1, 1)
+
+        # Generate 90 days of sales data
+        for day in range(90):
+            date = start_date + timedelta(days=day)
+            # Random number of sales per day (3-8)
+            for _ in range(random.randint(3, 8)):
+                product = random.choice(products)
+                units_sold = random.randint(1, 20)
+                price_per_unit = random.randint(50, 200)
+                revenue = units_sold * price_per_unit
+
+                data.append(
+                    {
+                        "Date": date.strftime("%Y-%m-%d"),
+                        "Product": product,
+                        "Units_Sold": units_sold,
+                        "Revenue": revenue,
+                    }
+                )
+
+        df = pd.DataFrame(data)
+
+        # Write to Excel
+        output_file = output_dir / f"sales_{region}.xlsx"
+        df.to_excel(output_file, sheet_name="Sales", index=False)
+        print(f"  ✅ Created {output_file} ({len(df)} rows)")
+
+
+def generate_workflow_3_data(output_dir: Path) -> None:
+    """Generate complex budget spreadsheet with formatting for Workflow 3."""
+    print("\n📊 Generating Workflow 3: company_budget_2024.xlsx")
+
+    wb = Workbook()
+    ws = wb.active
+    ws.title = "Budget"
+
+    # Define departments with colors
+    departments = {
+        "Engineering": "C6E0B4",
+        "Marketing": "FFD966",
+        "Sales": "F4B084",
+        "Operations": "B4C7E7",
+    }
+
+    # Define categories
+    categories = {
+        "Personnel": ["Salaries", "Benefits", "Training"],
+        "Infrastructure": ["Office Rent", "Equipment", "Software Licenses"],
+        "Operations": ["Travel", "Supplies", "Miscellaneous"],
+    }
+
+    # Styles
+    header_font = Font(bold=True, size=12)
+    category_font = Font(bold=True, size=11)
+
+    row = 1
+
+    # Title
+    ws.merge_cells(f"A{row}:E{row}")
+    ws[f"A{row}"] = "2024 Annual Budget"
+    ws[f"A{row}"].font = Font(bold=True, size=14)
+    ws[f"A{row}"].alignment = Alignment(horizontal="center")
+    row += 2
+
+    # Headers
+    ws[f"A{row}"] = "Category"
+    ws[f"B{row}"] = "Item"
+    for i, dept in enumerate(departments.keys()):
+        ws.cell(row, 3 + i, dept)
+        ws.cell(row, 3 + i).font = header_font
+
+    for cell in ws[row]:
+        cell.font = header_font
+    row += 1
+
+    # Data
+    for category, items in categories.items():
+        # Category header (bold)
+        ws[f"A{row}"] = category
+        ws[f"A{row}"].font = category_font
+        row += 1
+
+        # Items with department budgets
+        for item in items:
+            ws[f"A{row}"] = ""
+            ws[f"B{row}"] = item
+
+            # Add budget amounts for each department (with color)
+            for i, (dept, color) in enumerate(departments.items()):
+                amount = random.randint(5000, 50000)
+                cell = ws.cell(row, 3 + i, amount)
+                cell.fill = PatternFill(
+                    start_color=color, end_color=color, fill_type="solid"
+                )
+                cell.number_format = "$#,##0"
+
+            row += 1
+
+        row += 1  # Blank row between categories
+
+    # Adjust column widths
+    ws.column_dimensions["A"].width = 20
+    ws.column_dimensions["B"].width = 25
+    for i in range(len(departments)):
+        ws.column_dimensions[chr(67 + i)].width = 15  # C, D, E, F
+
+    output_file = output_dir / "company_budget_2024.xlsx"
+    wb.save(output_file)
+    print(f"  ✅ Created {output_file}")
+    print("     • Bold categories, colored departments, merged title cell")
+
+
+def generate_workflow_4_data(output_dir: Path) -> None:
+    """Generate weekly sales report for Workflow 4."""
+    print("\n📊 Generating Workflow 4: sales_weekly.xlsx")
+
+    products = [
+        "Product A",
+        "Product B",
+        "Product C",
+        "Product D",
+        "Product E",
+        "Product F",
+        "Product G",
+        "Product H",
+    ]
+
+    # Generate one week of data
+    data = []
+    start_date = datetime(2024, 11, 4)  # Monday
+
+    for day in range(7):
+        date = start_date + timedelta(days=day)
+        # Each product has 3-10 transactions per day
+        for product in products:
+            for _ in range(random.randint(3, 10)):
+                units = random.randint(1, 15)
+                price = random.randint(20, 150)
+                revenue = units * price
+
+                data.append(
+                    {
+                        "Date": date.strftime("%Y-%m-%d"),
+                        "Product": product,
+                        "Units": units,
+                        "Revenue": revenue,
+                    }
+                )
+
+    df = pd.DataFrame(data)
+
+    # Write to Excel with some formatting
+    output_file = output_dir / "sales_weekly.xlsx"
+    with pd.ExcelWriter(output_file, engine="openpyxl") as writer:
+        df.to_excel(writer, sheet_name="Weekly Sales", index=False)
+
+        # Format header
+        worksheet = writer.sheets["Weekly Sales"]
+        for cell in worksheet[1]:
+            cell.font = Font(bold=True)
+
+    print(f"  ✅ Created {output_file} ({len(df)} rows)")
+
+
+def main() -> None:
+    """Generate all sample data files."""
+    print("=" * 60)
+    print("Generating Sample Data for LlamaSheets + Coding Agent Workflows")
+    print("=" * 60)
+
+    # Create output directory
+    output_dir = Path("input_data")
+    output_dir.mkdir(exist_ok=True)
+
+    # Generate data for each workflow
+    generate_workflow_1_data(output_dir)
+    generate_workflow_2_data(output_dir)
+    generate_workflow_3_data(output_dir)
+    generate_workflow_4_data(output_dir)
+
+    print("\n" + "=" * 60)
+    print("✅ All sample data generated!")
+    print("=" * 60)
+    print(f"\nFiles created in {output_dir.absolute()}:")
+    print("\nWorkflow 1 (Understanding a New Spreadsheet):")
+    print("  • financial_report_q1.xlsx")
+    print("\nWorkflow 2 (Generating Analysis Scripts):")
+    print("  • sales_northeast.xlsx")
+    print("  • sales_southeast.xlsx")
+    print("  • sales_west.xlsx")
+    print("\nWorkflow 3 (Using Cell Metadata):")
+    print("  • company_budget_2024.xlsx")
+    print("\nWorkflow 4 (Complete Automation):")
+    print("  • sales_weekly.xlsx")
+    print("\nYou can now use these files with the workflows in the documentation!")
+
+
+if __name__ == "__main__":
+    main()

examples/sheets/llama-index-agent/llama_index_example.py

@@ -0,0 +1,308 @@
+"""
+LlamaSheets Agent with LlamaIndex
+
+This example shows how to build an agent that can work with spreadsheet data
+extracted by LlamaSheets using Python code execution.
+
+The agent has minimal tools but maximum flexibility - it can execute arbitrary
+pandas code against the extracted data, similar to a coding agent.
+
+NOTE: Code execution should be handled safely in a sandboxed environment for security.
+"""
+
+import io
+import json
+import sys
+from pathlib import Path
+from typing import Any, Dict, Optional
+
+import dotenv
+import pandas as pd
+from llama_index.core.agent import FunctionAgent, ToolCall, ToolCallResult, AgentStream
+from llama_index.llms.openai import OpenAI
+from workflows import Context
+
+dotenv.load_dotenv()
+
+# Global context for loaded dataframes
+_dataframe_context: Dict[str, Any] = {}
+
+
+# Helper function for initial agent context
+def list_extracted_data(data_dir: str = "data") -> str:
+    """
+    List all regions and metadata files extracted by LlamaSheets.
+
+    This helps discover what data is available to work with.
+
+    Args:
+        data_dir: Directory containing extracted parquet files (default: "data")
+
+    Returns:
+        JSON string with information about available files
+    """
+    data_path = Path(data_dir)
+
+    if not data_path.exists():
+        return json.dumps({"error": f"Data directory '{data_dir}' not found"})
+
+    # Find all parquet and metadata files
+    region_files = list(data_path.glob("*_region_*.parquet"))
+    job_metadata_files = list(data_path.glob("*_job_metadata.json"))
+
+    regions = []
+    for region_file in region_files:
+        # Quick peek at dimensions
+        df = pd.read_parquet(region_file)
+
+        # Find corresponding metadata file
+        base_name = region_file.stem.replace("_region_", "_metadata_")
+        metadata_path = region_file.parent / f"{base_name}.parquet"
+
+        regions.append(
+            {
+                "region_file": str(region_file),
+                "metadata_file": str(metadata_path) if metadata_path.exists() else None,
+                "shape": {"rows": len(df), "columns": len(df.columns)},
+                "columns": list(df.columns),
+            }
+        )
+
+    result = {
+        "data_directory": str(data_path.absolute()),
+        "num_regions": len(regions),
+        "regions": regions,
+        "job_metadata_files": [str(f) for f in job_metadata_files],
+    }
+
+    return json.dumps(result, indent=2)
+
+
+# Agent tool for code execution against dataframes
+def execute_dataframe_code(
+    code: str, load_files: Optional[Dict[str, str]] = None
+) -> str:
+    """
+        Execute Python pandas code against LlamaSheets extracted data.
+
+        This tool allows flexible data analysis by executing arbitrary pandas code.
+        You can load parquet files, manipulate dataframes, and return results.
+
+        The code executes in a context where:
+        - pandas is available as 'pd'
+        - json is available for formatting output
+        - Previously loaded dataframes are accessible by their variable names
+
+        Args:
+            code: Python code to execute. Any print() statements or stdout/stderr
+                  will be captured and returned. Optionally set a 'result' variable
+                  for structured output.
+            load_files: Optional dict mapping variable names to file paths to load
+                       Example: {"df": "data/sales_region_1.parquet",
+                                "meta": "data/sales_metadata_1.parquet"}
+
+        Returns:
+            String containing:
+            - Any stdout/stderr output from the code execution
+            - The 'result' variable if it was set (formatted appropriately)
+            - Error message if execution failed
+
+        Example usage:
+            code = '''
+    # Load and inspect data
+    df = pd.read_parquet("data/sales_region_1.parquet")
+    print(f"Loaded {len(df)} rows")
+
+    result = {
+        "shape": df.shape,
+        "columns": list(df.columns),
+        "sample": df.head(3).to_dict(orient="records")
+    }
+            '''
+    """
+    global _dataframe_context
+
+    # Capture stdout and stderr
+    stdout_capture = io.StringIO()
+    stderr_capture = io.StringIO()
+    old_stdout = sys.stdout
+    old_stderr = sys.stderr
+
+    try:
+        # Redirect stdout/stderr
+        sys.stdout = stdout_capture
+        sys.stderr = stderr_capture
+
+        # Create execution context with pandas, json, and previously loaded dfs
+        exec_context = {
+            "pd": pd,
+            "json": json,
+            "Path": Path,
+            **_dataframe_context,  # Include previously loaded dataframes
+        }
+
+        # Load any requested files into context
+        if load_files:
+            for var_name, file_path in load_files.items():
+                if file_path.endswith(".parquet"):
+                    exec_context[var_name] = pd.read_parquet(file_path)
+                    # Also save to global context for future calls
+                    _dataframe_context[var_name] = exec_context[var_name]
+                elif file_path.endswith(".json"):
+                    with open(file_path, "r") as f:
+                        exec_context[var_name] = json.load(f)
+                        _dataframe_context[var_name] = exec_context[var_name]
+
+        # Execute the code
+        exec(code, exec_context)
+
+        # Restore stdout/stderr
+        sys.stdout = old_stdout
+        sys.stderr = old_stderr
+
+        # Collect output
+        stdout_output = stdout_capture.getvalue()
+        stderr_output = stderr_capture.getvalue()
+
+        output_parts = []
+
+        # Add stdout if any
+        if stdout_output:
+            output_parts.append(f"<stdout>{stdout_output}</stdout>")
+
+        # Add stderr if any
+        if stderr_output:
+            output_parts.append(f"<stderr>{stderr_output}</stderr>")
+
+        # Try to get a result (if code set a 'result' variable)
+        if "result" in exec_context:
+            result = exec_context["result"]
+            result_str = None
+
+            if isinstance(result, pd.DataFrame):
+                # Convert DataFrame to readable format
+                result_str = result.to_string()
+            elif isinstance(result, (dict, list)):
+                result_str = json.dumps(result, indent=2, default=str)
+            else:
+                result_str = str(result)
+
+            if result_str:
+                output_parts.append(f"<result_var>{result_str}</result_var>")
+
+        # Return combined output or success message
+        if output_parts:
+            return "\n\n".join(output_parts)
+        else:
+            return "Code executed successfully (no output or result)"
+
+    except Exception as e:
+        # Restore stdout/stderr in case of error
+        sys.stdout = old_stdout
+        sys.stderr = old_stderr
+
+        # Get any partial output
+        stdout_output = stdout_capture.getvalue()
+        stderr_output = stderr_capture.getvalue()
+
+        error_parts = []
+        if stdout_output:
+            error_parts.append(f"=== STDOUT (before error) ===\n{stdout_output}")
+        if stderr_output:
+            error_parts.append(f"=== STDERR (before error) ===\n{stderr_output}")
+
+        error_parts.append(f"=== ERROR ===\n{str(e)}")
+        error_parts.append(f"\n=== CODE ===\n{code}")
+
+        return "\n\n".join(error_parts)
+
+
+def create_llamasheets_agent(
+    llm_model: str = "gpt-4.1", api_key: Optional[str] = None
+) -> FunctionAgent:
+    # Initialize LLM
+    llm = OpenAI(model=llm_model, api_key=api_key)
+
+    # Create tools - just 4 simple but powerful tools
+    tools = [execute_dataframe_code]
+
+    # System prompt to guide the agent
+    available_regions = list_extracted_data()
+    system_prompt = f"""You are an AI assistant that helps analyze spreadsheet data extracted by LlamaSheets.
+
+LlamaSheets extracts messy spreadsheets into clean parquet files with two types of outputs:
+1. Region files (*_region_*.parquet) - The actual data with columns and rows
+2. Metadata files (*_metadata_*.parquet) - Rich cell-level metadata including:
+   - Formatting: font_bold, font_italic, font_size, background_color_rgb
+   - Position: row_number, column_number, coordinate
+   - Type detection: data_type, is_date_like, is_percentage, is_currency
+   - Layout: is_in_first_row, is_merged_cell, horizontal_alignment
+
+Your approach:
+1. Use list_extracted_data() to discover available files
+2. Use execute_dataframe_code() to load and analyze data with pandas
+3. Use metadata to understand structure (bold = headers, colors = groups)
+4. Use save_dataframe() to export results
+
+Key tips:
+- Bold cells in metadata often indicate headers
+- Background colors often indicate groupings or departments
+- Load both region and metadata files for complete analysis
+- Write clear pandas code - you have full pandas functionality available
+- Store results in variables for reuse across multiple code executions
+
+Existing Processed Regions:
+{available_regions}
+"""
+
+    # Configure agent
+    return FunctionAgent(tools=tools, llm=llm, system_prompt=system_prompt)
+
+
+async def main():
+    """Example of using the LlamaSheets agent."""
+
+    # Create the agent
+    agent = create_llamasheets_agent()
+    ctx = Context(agent)
+
+    # Example queries the agent can handle:
+    queries = [
+        # Discovery
+        "What spreadsheet data is available?",
+        # Simple analysis
+        "Load the sales data and show me the first few rows with column info",
+        # Using metadata
+        "Find all bold cells in the metadata - these are likely headers",
+    ]
+
+    # Example: Run a query
+    for query in queries:
+        print(f"\n=== Query: {query} ===")
+        handler = agent.run(query, ctx=ctx)
+        async for ev in handler.stream_events():
+            if isinstance(ev, ToolCall):
+                tool_kwargs_str = (
+                    str(ev.tool_kwargs)[:500] + " ..."
+                    if len(str(ev.tool_kwargs)) > 500
+                    else str(ev.tool_kwargs)
+                )
+                print(f"\n[Tool Call] {ev.tool_name} with args:\n{tool_kwargs_str}\n\n")
+            elif isinstance(ev, ToolCallResult):
+                result_str = (
+                    str(ev.tool_output)[:500] + " ..."
+                    if len(str(ev.tool_output)) > 500
+                    else str(ev.tool_output)
+                )
+                print(f"\n[Tool Result] {ev.tool_name}:\n{result_str}\n\n")
+            elif isinstance(ev, AgentStream):
+                print(ev.delta, end="", flush=True)
+
+        _ = await handler
+        print("=== End Query ===\n")
+
+
+if __name__ == "__main__":
+    import asyncio
+
+    asyncio.run(main())

examples/sheets/llama-index-agent/requirements.txt

@@ -0,0 +1,7 @@
+llama-cloud-services  # LlamaSheets SDK
+llama-index-core
+llama-index-llms-openai
+pandas>=2.0.0
+pyarrow>=12.0.0
+openpyxl>=3.0.0       # For Excel file support
+matplotlib>=3.7.0     # For visualizations (optional)

examples/sheets/llama-index-agent/scripts/extract.py

@@ -0,0 +1,100 @@
+"""Helper script to extract spreadsheets using LlamaSheets."""
+
+import asyncio
+import json
+import os
+import dotenv
+from pathlib import Path
+
+from llama_cloud_services.beta.sheets import LlamaSheets
+from llama_cloud_services.beta.sheets.types import (
+    SpreadsheetParsingConfig,
+    SpreadsheetResultType,
+)
+
+dotenv.load_dotenv()
+
+
+async def extract_spreadsheet(
+    file_path: str, output_dir: str = "data", generate_metadata: bool = True
+) -> dict:
+    """Extract a spreadsheet using LlamaSheets."""
+
+    client = LlamaSheets(
+        base_url="https://api.cloud.llamaindex.ai",
+        api_key=os.getenv("LLAMA_CLOUD_API_KEY"),
+    )
+
+    print(f"Extracting {file_path}...")
+
+    # Extract regions
+    config = SpreadsheetParsingConfig(
+        sheet_names=None,  # Extract all sheets
+        generate_additional_metadata=generate_metadata,
+    )
+
+    job_result = await client.aextract_regions(file_path, config=config)
+
+    print(f"Extracted {len(job_result.regions)} region(s)")
+
+    # Create output directory
+    output_path = Path(output_dir)
+    output_path.mkdir(parents=True, exist_ok=True)
+
+    # Get base name for files
+    base_name = Path(file_path).stem
+
+    # Save job metadata
+    job_metadata_path = output_path / f"{base_name}_job_metadata.json"
+    with open(job_metadata_path, "w") as f:
+        json.dump(job_result.model_dump(mode="json"), f, indent=2)
+    print(f"Saved job metadata to {job_metadata_path}")
+
+    # Download each region
+    for idx, region in enumerate(job_result.regions, 1):
+        sheet_name = region.sheet_name.replace(" ", "_")
+
+        # Download region data
+        region_bytes = await client.adownload_region_result(
+            job_id=job_result.id,
+            region_id=region.region_id,
+            result_type=region.region_type,
+        )
+
+        region_path = output_path / f"{base_name}_region_{idx}_{sheet_name}.parquet"
+        with open(region_path, "wb") as f:
+            f.write(region_bytes)
+        print(f"  Table {idx}: {region_path}")
+
+        # Download metadata
+        metadata_bytes = await client.adownload_region_result(
+            job_id=job_result.id,
+            region_id=region.region_id,
+            result_type=SpreadsheetResultType.CELL_METADATA,
+        )
+
+        metadata_path = output_path / f"{base_name}_metadata_{idx}_{sheet_name}.parquet"
+        with open(metadata_path, "wb") as f:
+            f.write(metadata_bytes)
+        print(f"  Metadata {idx}: {metadata_path}")
+
+    print(f"\nAll files saved to {output_path}/")
+
+    return job_result.model_dump(mode="json")
+
+
+if __name__ == "__main__":
+    import sys
+
+    if len(sys.argv) < 2:
+        print("Usage: python scripts/extract.py <spreadsheet_file>")
+        sys.exit(1)
+
+    file_path = sys.argv[1]
+
+    if not Path(file_path).exists():
+        print(f"❌ File not found: {file_path}")
+        sys.exit(1)
+
+    result = asyncio.run(extract_spreadsheet(file_path))
+    print(f"\n✅ Extraction complete! Job ID: {result['id']}")

package.json

@@ -19,7 +19,7 @@
   "lint-staged": {
     "ts/llama_cloud_services/src/**/*.{ts,tsx,js,jsx}": [
       "pnpm --filter llama-cloud-services exec eslint --fix",
-      "pnpm --filter llama-cloud-services exec prettier --write"
+      "pnpm --filter llama-cloud-services exec prettier --write src/ tests/"
     ]
   },
   "packageManager": "pnpm@10.11.1+sha512.e519b9f7639869dc8d5c3c5dfef73b3f091094b0a006d7317353c72b124e80e1afd429732e28705ad6bfa1ee879c1fce46c128ccebd3192101f43dd67c667912"

py/CHANGELOG.md

@@ -1,5 +1,35 @@
 # llama-cloud-services-py
 
+## 0.6.81
+
+### Patch Changes
+
+- f3233de: Propagate retrieval metadata to retriever nodes
+
+## 0.6.80
+
+### Patch Changes
+
+- 0506c88: Moved ClassifyClient to LlamaClassify (backward compatible)
+
+## 0.6.79
+
+### Patch Changes
+
+- e020e3e: Remove unneeded organization_id param from beta classifier client
+
+## 0.6.78
+
+### Patch Changes
+
+- 9f1ef4e: Fix extract
+
+## 0.6.77
+
+### Patch Changes
+
+- 407292b: Now return partial results on job failure
+
 ## 0.6.76
 
 ### Patch Changes

py/llama_cloud_services/__init__.py

@@ -1,5 +1,7 @@
 from llama_cloud_services.parse import LlamaParse
-from llama_cloud_services.extract import LlamaExtract, ExtractionAgent, SourceText
+from llama_cloud_services.extract import LlamaExtract, ExtractionAgent
+from llama_cloud_services.testing_utils import FakeLlamaCloudServer
+from llama_cloud_services.utils import SourceText, FileInput
 from llama_cloud_services.constants import EU_BASE_URL
 from llama_cloud_services.index import (
     LlamaCloudCompositeRetriever,
@@ -12,8 +14,10 @@ __all__ = [
     "LlamaExtract",
     "ExtractionAgent",
     "SourceText",
+    "FileInput",
     "EU_BASE_URL",
     "LlamaCloudIndex",
     "LlamaCloudRetriever",
     "LlamaCloudCompositeRetriever",
+    "FakeLlamaCloudServer",
 ]

py/llama_cloud_services/beta/classifier/__init__.py

@@ -0,0 +1,11 @@
+from llama_cloud_services.beta.classifier.client import LlamaClassify, ClassifyClient
+from llama_cloud_services.beta.classifier.types import ClassifyJobResultsWithFiles
+from llama_cloud_services.utils import SourceText, FileInput
+
+__all__ = [
+    "LlamaClassify",
+    "ClassifyClient",
+    "ClassifyJobResultsWithFiles",
+    "SourceText",
+    "FileInput",
+]

py/llama_cloud_services/beta/classifier/client.py

@@ -1,6 +1,7 @@
 import asyncio
 import time
-from typing import Optional
+import warnings
+from typing import Optional, List, Union
 from pydantic import BaseModel
 from llama_cloud.client import AsyncLlamaCloud
 from llama_cloud.types import (
@@ -14,7 +15,11 @@ from llama_cloud.types import (
 from llama_cloud.resources.classifier.client import OMIT
 from llama_cloud_services.files.client import FileClient
 from llama_cloud_services.constants import POLLING_TIMEOUT_SECONDS
-from llama_cloud_services.utils import is_terminal_status, augment_async_errors
+from llama_cloud_services.utils import (
+    is_terminal_status,
+    augment_async_errors,
+    FileInput,
+)
 from llama_index.core.async_utils import DEFAULT_NUM_WORKERS, run_jobs
 from llama_cloud_services.beta.classifier.types import (
     ClassifyJobResultsWithFiles,
@@ -26,7 +31,7 @@ class ClassificationOutput(BaseModel):
     classification: str
 
 
-class ClassifyClient:
+class LlamaClassify:
     """
     Experimental - Client for interacting with the LlamaCloud Classifier API.
     The Classification API is currently in beta and may change in the future without notice.
@@ -34,7 +39,6 @@ class ClassifyClient:
     Args:
         client: The LlamaCloud client to use.
         project_id: The project ID to use.
-        organization_id: The organization ID to use.
         polling_interval: The interval to poll for job completion in seconds.
         polling_timeout: The timeout for the job to complete in seconds.
     """
@@ -43,15 +47,13 @@ class ClassifyClient:
         self,
         client: AsyncLlamaCloud,
         project_id: Optional[str] = None,
-        organization_id: Optional[str] = None,
         polling_interval: float = 1.0,
         polling_timeout: float = POLLING_TIMEOUT_SECONDS,
     ):
         self.client = client
         self.project_id = project_id
-        self.organization_id = organization_id
         self.polling_interval = polling_interval
-        self.file_client = FileClient(client, project_id, organization_id)
+        self.file_client = FileClient(client, project_id)
         self.polling_timeout = polling_timeout
 
     @classmethod
@@ -59,7 +61,6 @@ class ClassifyClient:
         cls,
         api_key: str,
         project_id: Optional[str] = None,
-        organization_id: Optional[str] = None,
         base_url: Optional[str] = None,
     ) -> "ClassifyClient":
         """
@@ -69,7 +70,6 @@ class ClassifyClient:
         return cls(
             client,
             project_id,
-            organization_id,
         )
 
     async def acreate_classify_job(
@@ -96,7 +96,6 @@ class ClassifyClient:
             file_ids=file_ids,
             parsing_configuration=parsing_configuration or OMIT,
             project_id=self.project_id,
-            organization_id=self.organization_id,
         )
 
     def create_classify_job(
@@ -147,7 +146,6 @@ class ClassifyClient:
         results = await self.client.classifier.get_classification_job_results(
             classify_job_with_status.id,
             project_id=self.project_id,
-            organization_id=self.organization_id,
         )
 
         return results
@@ -166,6 +164,98 @@ class ClassifyClient:
                 )
             )
 
+    async def aclassify(
+        self,
+        rules: list[ClassifierRule],
+        files: Union[FileInput, List[FileInput]],
+        parsing_configuration: Optional[ClassifyParsingConfiguration] = None,
+        raise_on_error: bool = True,
+        workers: int = DEFAULT_NUM_WORKERS,
+        show_progress: bool = False,
+    ) -> ClassifyJobResultsWithFiles:
+        """
+        Classify one or more files from various input types.
+
+        Args:
+            rules: The rules to use for classification.
+            files: The file(s) to classify. Can be a single file or list of files. Each can be:
+                - str/Path: File path
+                - SourceText: Text content or file with explicit filename
+                - File: Already uploaded file
+                - BufferedIOBase: File-like object
+            parsing_configuration: The parsing configuration to use for classification.
+            raise_on_error: Whether to raise an error if the classification job fails.
+            workers: Number of parallel workers for uploading files.
+            show_progress: Whether to show progress bars.
+
+        Returns:
+            The results of the classification job with file metadata.
+        """
+        # Normalize to list
+        if not isinstance(files, list):
+            files = [files]
+
+        # Upload all files
+        coroutines = [
+            self.file_client.upload_content(file_input) for file_input in files
+        ]
+        uploaded_files: List[File] = await run_jobs(
+            coroutines,
+            show_progress=show_progress,
+            workers=workers,
+            desc="Uploading files for classification",
+        )
+
+        # Classify
+        results = await self.aclassify_file_ids(
+            rules,
+            [file.id for file in uploaded_files],
+            parsing_configuration,
+            raise_on_error,
+        )
+        return ClassifyJobResultsWithFiles.from_classify_job_results(
+            results, uploaded_files
+        )
+
+    def classify(
+        self,
+        rules: list[ClassifierRule],
+        files: Union[FileInput, List[FileInput]],
+        parsing_configuration: Optional[ClassifyParsingConfiguration] = None,
+        raise_on_error: bool = True,
+        workers: int = DEFAULT_NUM_WORKERS,
+        show_progress: bool = False,
+    ) -> ClassifyJobResultsWithFiles:
+        """
+        Classify one or more files from various input types (synchronous version).
+
+        Args:
+            rules: The rules to use for classification.
+            files: The file(s) to classify. Can be a single file or list of files. Each can be:
+                - str/Path: File path
+                - SourceText: Text content or file with explicit filename
+                - File: Already uploaded file
+                - BufferedIOBase: File-like object
+            parsing_configuration: The parsing configuration to use for classification.
+            raise_on_error: Whether to raise an error if the classification job fails.
+            workers: Number of parallel workers for uploading files.
+            show_progress: Whether to show progress bars.
+
+        Returns:
+            The results of the classification job with file metadata.
+        """
+        with augment_async_errors():
+            return asyncio.run(
+                self.aclassify(
+                    rules,
+                    files,
+                    parsing_configuration,
+                    raise_on_error,
+                    workers,
+                    show_progress,
+                )
+            )
+
     async def aclassify_file_path(
         self,
         rules: list[ClassifierRule],
@@ -173,11 +263,17 @@ class ClassifyClient:
         parsing_configuration: Optional[ClassifyParsingConfiguration] = None,
         raise_on_error: bool = True,
     ) -> ClassifyJobResultsWithFiles:
-        file = await self.file_client.upload_file(file_input_path)
-        results = await self.aclassify_file_ids(
-            rules, [file.id], parsing_configuration, raise_on_error
+        """
+        Deprecated: Use aclassify() instead.
+        """
+        warnings.warn(
+            "aclassify_file_path is deprecated, use aclassify() instead",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        return await self.aclassify(
+            rules, file_input_path, parsing_configuration, raise_on_error
         )
-        return ClassifyJobResultsWithFiles.from_classify_job_results(results, [file])
 
     def classify_file_path(
         self,
@@ -186,12 +282,17 @@ class ClassifyClient:
         parsing_configuration: Optional[ClassifyParsingConfiguration] = None,
         raise_on_error: bool = True,
     ) -> ClassifyJobResultsWithFiles:
-        with augment_async_errors():
-            return asyncio.run(
-                self.aclassify_file_path(
-                    rules, file_input_path, parsing_configuration, raise_on_error
-                )
-            )
+        """
+        Deprecated: Use classify() instead.
+        """
+        warnings.warn(
+            "classify_file_path is deprecated, use classify() instead",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        return self.classify(
+            rules, file_input_path, parsing_configuration, raise_on_error
+        )
 
     async def aclassify_file_paths(
         self,
@@ -202,17 +303,22 @@ class ClassifyClient:
         workers: int = DEFAULT_NUM_WORKERS,
         show_progress: bool = False,
     ) -> ClassifyJobResultsWithFiles:
-        coroutines = [self.file_client.upload_file(path) for path in file_input_paths]
-        files: list[File] = await run_jobs(
-            coroutines,
-            show_progress=show_progress,
-            workers=workers,
-            desc="Uploading files for classification",
+        """
+        Deprecated: Use aclassify() instead.
+        """
+        warnings.warn(
+            "aclassify_file_paths is deprecated, use aclassify() instead",
+            DeprecationWarning,
+            stacklevel=2,
         )
-        results = await self.aclassify_file_ids(
-            rules, [file.id for file in files], parsing_configuration, raise_on_error
+        return await self.aclassify(
+            rules,
+            file_input_paths,
+            parsing_configuration,
+            raise_on_error,
+            workers,
+            show_progress,
         )
-        return ClassifyJobResultsWithFiles.from_classify_job_results(results, files)
 
     def classify_file_paths(
         self,
@@ -221,12 +327,17 @@ class ClassifyClient:
         parsing_configuration: Optional[ClassifyParsingConfiguration] = None,
         raise_on_error: bool = True,
     ) -> ClassifyJobResultsWithFiles:
-        with augment_async_errors():
-            return asyncio.run(
-                self.aclassify_file_paths(
-                    rules, file_input_paths, parsing_configuration, raise_on_error
-                )
-            )
+        """
+        Deprecated: Use classify() instead.
+        """
+        warnings.warn(
+            "classify_file_paths is deprecated, use classify() instead",
+            DeprecationWarning,
+            stacklevel=2,
+        )
+        return self.classify(
+            rules, file_input_paths, parsing_configuration, raise_on_error
+        )
 
     async def wait_for_job_completion(self, job_id: str) -> ClassifyJob:
         """
@@ -241,7 +352,7 @@ class ClassifyClient:
             The classify job with status.
         """
         job = await self.client.classifier.get_classify_job(
-            job_id, project_id=self.project_id, organization_id=self.organization_id
+            job_id, project_id=self.project_id
         )
         start_time = time.time()
         while not is_terminal_status(job.status):
@@ -252,6 +363,9 @@ class ClassifyClient:
                 )
             await asyncio.sleep(self.polling_interval)
             job = await self.client.classifier.get_classify_job(
-                job_id, project_id=self.project_id, organization_id=self.organization_id
+                job_id, project_id=self.project_id
             )
         return job
+
+
+ClassifyClient = LlamaClassify

py/llama_cloud_services/beta/sheets/__init__.py

@@ -0,0 +1,43 @@
+"""LlamaCloud Spreadsheet API SDK
+
+This module provides a Python SDK for the LlamaCloud Spreadsheet API.
+"""
+
+from llama_cloud_services.beta.sheets.client import (
+    LlamaSheets,
+    SpreadsheetAPIError,
+    SpreadsheetJobError,
+    SpreadsheetTimeoutError,
+)
+from llama_cloud_services.beta.sheets.types import (
+    ExtractedRegionSummary,
+    FileUploadResponse,
+    JobStatus,
+    PresignedUrlResponse,
+    SpreadsheetJob,
+    SpreadsheetJobResult,
+    SpreadsheetParseResult,
+    SpreadsheetParsingConfig,
+    SpreadsheetResultType,
+    WorksheetMetadata,
+)
+
+__all__ = [
+    # Client
+    "LlamaSheets",
+    # Exceptions
+    "SpreadsheetAPIError",
+    "SpreadsheetJobError",
+    "SpreadsheetTimeoutError",
+    # Types
+    "ExtractedRegionSummary",
+    "FileUploadResponse",
+    "JobStatus",
+    "PresignedUrlResponse",
+    "SpreadsheetJob",
+    "SpreadsheetJobResult",
+    "SpreadsheetParseResult",
+    "SpreadsheetParsingConfig",
+    "SpreadsheetResultType",
+    "WorksheetMetadata",
+]

py/llama_cloud_services/beta/sheets/client.py

@@ -0,0 +1,520 @@
+from __future__ import annotations
+
+import asyncio
+import io
+import os
+import time
+from typing import TYPE_CHECKING
+
+import httpx
+from llama_cloud.client import AsyncLlamaCloud
+from tenacity import (
+    AsyncRetrying,
+    retry_if_exception,
+    stop_after_attempt,
+    wait_exponential,
+)
+
+from llama_cloud_services.beta.sheets.types import (
+    FileUploadResponse,
+    JobStatus,
+    PresignedUrlResponse,
+    SpreadsheetJob,
+    SpreadsheetJobResult,
+    SpreadsheetParsingConfig,
+    SpreadsheetResultType,
+)
+from llama_cloud_services.constants import BASE_URL
+from llama_cloud_services.files.client import FileClient
+from llama_cloud_services.utils import (
+    augment_async_errors,
+    FileInput,
+)
+
+if TYPE_CHECKING:
+    import pandas as pd
+
+
+def _should_retry_exception(exception: BaseException) -> bool:
+    """Determine if an exception should be retried."""
+    if isinstance(exception, httpx.HTTPStatusError):
+        return exception.response.status_code in (429, 500, 502, 503, 504)
+    return False
+
+
+class SpreadsheetAPIError(Exception):
+    """Base exception for spreadsheet API errors"""
+
+    pass
+
+
+class SpreadsheetJobError(SpreadsheetAPIError):
+    """Exception raised when a spreadsheet job fails"""
+
+    pass
+
+
+class SpreadsheetTimeoutError(SpreadsheetAPIError):
+    """Exception raised when a job times out"""
+
+    pass
+
+
+class LlamaSheets:
+    """Client for the LlamaCloud Spreadsheet API"""
+
+    def __init__(
+        self,
+        api_key: str | None = None,
+        base_url: str | None = None,
+        max_timeout: int = 300,
+        poll_interval: int = 5,
+        max_retries: int = 3,
+        async_httpx_client: httpx.AsyncClient | None = None,
+    ) -> None:
+        """Initialize the LlamaSheets client.
+
+        Args:
+            api_key: API key for authentication. If not provided, will use LLAMA_CLOUD_API_KEY env var
+            base_url: Base URL for the API
+            max_timeout: Maximum time to wait for job completion in seconds
+            poll_interval: Interval between status checks in seconds
+            max_retries: Maximum number of retries for failed requests
+            async_httpx_client: Optional custom async httpx client
+        """
+        self.api_key = api_key or os.environ.get("LLAMA_CLOUD_API_KEY")
+        if not self.api_key:
+            raise ValueError(
+                "An API key must be provided either as an argument or via the LLAMA_CLOUD_API_KEY environment variable."
+            )
+
+        base_url = base_url or os.environ.get("LLAMA_CLOUD_BASE_URL", BASE_URL)
+        self.base_url = str(base_url).rstrip("/")
+
+        self.max_timeout = max_timeout
+        self.poll_interval = poll_interval
+        self.max_retries = max_retries
+
+        self._async_client: httpx.AsyncClient | None = async_httpx_client
+        self._files_client = FileClient(
+            AsyncLlamaCloud(
+                token=self.api_key,
+                base_url=self.base_url,
+                httpx_client=async_httpx_client,
+            )
+        )
+
+    def _get_async_client(self) -> httpx.AsyncClient:
+        """Get or create the async httpx client"""
+        if self._async_client is None:
+            self._async_client = httpx.AsyncClient(
+                timeout=httpx.Timeout(60.0),
+                follow_redirects=True,
+            )
+        return self._async_client
+
+    def _get_headers(self) -> dict[str, str]:
+        """Get common headers for API requests"""
+        return {
+            "Authorization": f"Bearer {self.api_key}",
+            "Content-Type": "application/json",
+        }
+
+    # Sync methods
+
+    def upload_file(
+        self, file_obj: FileInput, file_name: str | None = None
+    ) -> FileUploadResponse:
+        """Upload a file to the Files API.
+
+        Args:
+            file_obj: File to upload (path, bytes, or file-like object)
+            file_name: Optional name for the uploaded filename
+
+        Returns:
+            FileUploadResponse with the uploaded file ID
+        """
+        with augment_async_errors():
+            return asyncio.run(self.aupload_file(file_obj))
+
+    def create_job(
+        self,
+        file_id: str,
+        config: dict | SpreadsheetParsingConfig | None = None,
+    ) -> SpreadsheetJob:
+        """Create a new spreadsheet parsing job.
+
+        Args:
+            file_id: ID of the uploaded file
+            config: Parsing configuration
+
+        Returns:
+            SpreadsheetJob with job details
+        """
+        with augment_async_errors():
+            return asyncio.run(self.acreate_job(file_id, config))
+
+    def get_job(
+        self, job_id: str, include_results_metadata: bool = True
+    ) -> SpreadsheetJobResult:
+        """Get the status of a spreadsheet parsing job.
+
+        Args:
+            job_id: ID of the job
+            include_results_metadata: Whether to include results metadata in the response
+
+        Returns:
+            SpreadsheetJobResult with job status and optionally results
+        """
+        with augment_async_errors():
+            return asyncio.run(self.aget_job(job_id, include_results_metadata))
+
+    def wait_for_completion(self, job_id: str) -> SpreadsheetJobResult:
+        """Wait for a job to complete by polling.
+
+        Args:
+            job_id: ID of the job to wait for
+
+        Returns:
+            SpreadsheetJobResult when job is complete
+
+        Raises:
+            SpreadsheetTimeoutError: If job doesn't complete within max_timeout
+            SpreadsheetJobError: If job fails
+        """
+        with augment_async_errors():
+            return asyncio.run(self.await_for_completion(job_id))
+
+    def download_region_result(
+        self,
+        job_id: str,
+        region_id: str,
+        result_type: SpreadsheetResultType = SpreadsheetResultType.TABLE,
+    ) -> bytes:
+        """Download a region result (either region data or cell metadata).
+
+        Args:
+            job_id: ID of the job
+            region_id: ID of the region
+            result_type: Type of result to download (region or cell_metadata)
+
+        Returns:
+            Raw bytes of the parquet file
+        """
+        with augment_async_errors():
+            return asyncio.run(
+                self.adownload_region_result(job_id, region_id, result_type)
+            )
+
+    def download_region_as_dataframe(
+        self,
+        job_id: str,
+        region_id: str,
+        result_type: SpreadsheetResultType = SpreadsheetResultType.TABLE,
+    ) -> "pd.DataFrame":
+        """Download a region result as a pandas DataFrame.
+
+        Args:
+            job_id: ID of the job
+            region_id: ID of the region
+            result_type: Type of result to download (region or cell_metadata)
+
+        Returns:
+            pandas DataFrame
+        """
+        with augment_async_errors():
+            return asyncio.run(
+                self.adownload_region_as_dataframe(job_id, region_id, result_type)
+            )
+
+    def extract_regions(
+        self,
+        file_obj: FileInput,
+        config: dict | SpreadsheetParsingConfig | None = None,
+    ) -> SpreadsheetJobResult:
+        """High-level method to parse a spreadsheet file.
+
+        This method handles the entire workflow:
+        1. Upload the file
+        2. Create a parsing job
+        3. Wait for completion
+        4. Return results
+
+        Args:
+            file_obj: File to parse (path, bytes, or file-like object)
+            config: Parsing configuration
+
+        Returns:
+            SpreadsheetJobResult with parsing results
+        """
+        with augment_async_errors():
+            return asyncio.run(self.aextract_regions(file_obj, config))
+
+    # Async methods
+
+    async def aupload_file(
+        self, file_obj: FileInput, file_name: str | None = None
+    ) -> FileUploadResponse:
+        """Upload a file to the Files API.
+
+        Args:
+            file_obj: File to upload (path, bytes, or file-like object)
+            file_name: Optional name for the uploaded filename
+
+        Returns:
+            FileUploadResponse with the uploaded file ID
+        """
+        try:
+            async for attempt in AsyncRetrying(
+                stop=stop_after_attempt(self.max_retries),
+                wait=wait_exponential(multiplier=1, min=1, max=32),
+                retry=retry_if_exception(_should_retry_exception),
+                reraise=True,
+            ):
+                with attempt:
+                    return await self._files_client.upload_content(
+                        file_obj, external_file_id=file_name
+                    )
+        except Exception as e:
+            raise SpreadsheetAPIError(f"Failed to upload file: {e}") from e
+        raise RuntimeError("Tenacity did not execute")
+
+    async def acreate_job(
+        self,
+        file_id: str,
+        config: dict | SpreadsheetParsingConfig | None = None,
+    ) -> SpreadsheetJob:
+        """Create a new spreadsheet parsing job.
+
+        Args:
+            file_id: ID of the uploaded file
+            config: Parsing configuration
+
+        Returns:
+            SpreadsheetJob with job details
+        """
+        if config is None:
+            config = SpreadsheetParsingConfig()
+        elif isinstance(config, dict):
+            config = SpreadsheetParsingConfig.model_validate(config)
+
+        if not isinstance(config, SpreadsheetParsingConfig):
+            raise ValueError(
+                "config must be a dict or SpreadsheetParsingConfig instance"
+            )
+
+        payload = {
+            "file_id": file_id,
+            "config": config.model_dump(mode="json", exclude_none=True),
+        }
+
+        try:
+            async for attempt in AsyncRetrying(
+                stop=stop_after_attempt(self.max_retries),
+                wait=wait_exponential(multiplier=1, min=1, max=32),
+                retry=retry_if_exception(_should_retry_exception),
+                reraise=True,
+            ):
+                with attempt:
+                    client = self._get_async_client()
+                    response = await client.post(
+                        f"{self.base_url}/api/v1/beta/sheets/jobs",
+                        headers=self._get_headers(),
+                        json=payload,
+                    )
+                    response.raise_for_status()
+                    return SpreadsheetJob.model_validate(response.json())
+        except Exception as e:
+            raise SpreadsheetAPIError(f"Failed to create job: {e}") from e
+        raise RuntimeError("Tenacity did not execute")
+
+    async def aget_job(
+        self, job_id: str, include_results_metadata: bool = True
+    ) -> SpreadsheetJobResult:
+        """Get the status of a spreadsheet parsing job.
+
+        Args:
+            job_id: ID of the job
+            include_results_metadata: Whether to include results in the response
+
+        Returns:
+            SpreadsheetJobResult with job status and optionally results
+        """
+        try:
+            async for attempt in AsyncRetrying(
+                stop=stop_after_attempt(self.max_retries),
+                wait=wait_exponential(multiplier=1, min=1, max=32),
+                retry=retry_if_exception(_should_retry_exception),
+                reraise=True,
+            ):
+                with attempt:
+                    client = self._get_async_client()
+                    response = await client.get(
+                        f"{self.base_url}/api/v1/beta/sheets/jobs/{job_id}",
+                        headers=self._get_headers(),
+                        params={"include_results": include_results_metadata},
+                    )
+                    response.raise_for_status()
+                    return SpreadsheetJobResult.model_validate(response.json())
+        except Exception as e:
+            raise SpreadsheetAPIError(f"Failed to get job status: {e}") from e
+        raise RuntimeError("Tenacity did not execute")
+
+    async def await_for_completion(self, job_id: str) -> SpreadsheetJobResult:
+        """Wait for a job to complete by polling.
+
+        Args:
+            job_id: ID of the job to wait for
+
+        Returns:
+            SpreadsheetJobResult when job is complete
+
+        Raises:
+            SpreadsheetTimeoutError: If job doesn't complete within max_timeout
+            SpreadsheetJobError: If job fails
+        """
+        start_time = time.time()
+
+        while (time.time() - start_time) < self.max_timeout:
+            job_result = await self.aget_job(job_id, include_results_metadata=True)
+
+            if job_result.status in (
+                JobStatus.SUCCESS,
+                JobStatus.PARTIAL_SUCCESS,
+                JobStatus.ERROR,
+                JobStatus.FAILURE,
+            ):
+                if job_result.status in (JobStatus.SUCCESS, JobStatus.PARTIAL_SUCCESS):
+                    return job_result
+                else:
+                    error_msg = f"Job failed with status: {job_result.status}"
+                    if job_result.errors:
+                        error_msg += f"\nErrors: {', '.join(job_result.errors)}"
+                    raise SpreadsheetJobError(error_msg)
+
+            await asyncio.sleep(self.poll_interval)
+
+        raise SpreadsheetTimeoutError(
+            f"Job did not complete within {self.max_timeout} seconds"
+        )
+
+    async def adownload_region_result(
+        self,
+        job_id: str,
+        region_id: str,
+        result_type: SpreadsheetResultType = SpreadsheetResultType.TABLE,
+    ) -> bytes:
+        """Download a region result (either region data or cell metadata).
+
+        Args:
+            job_id: ID of the job
+            region_id: ID of the region
+            result_type: Type of result to download (region or cell_metadata)
+
+        Returns:
+            Raw bytes of the parquet file
+        """
+        # Get presigned URL
+        presigned_response = None
+        result_type_str = str(result_type)
+        try:
+            async for attempt in AsyncRetrying(
+                stop=stop_after_attempt(self.max_retries),
+                wait=wait_exponential(multiplier=1, min=1, max=32),
+                retry=retry_if_exception(_should_retry_exception),
+                reraise=True,
+            ):
+                with attempt:
+                    client = self._get_async_client()
+                    response = await client.get(
+                        f"{self.base_url}/api/v1/beta/sheets/jobs/{job_id}/regions/{region_id}/result/{result_type_str}",
+                        headers=self._get_headers(),
+                    )
+                    response.raise_for_status()
+                    presigned_response = PresignedUrlResponse.model_validate(
+                        response.json()
+                    )
+        except Exception as e:
+            raise SpreadsheetAPIError(f"Failed to get presigned URL: {e}") from e
+
+        # Download using presigned URL
+        if presigned_response is None:
+            raise SpreadsheetAPIError("Failed to obtain presigned URL.")
+
+        try:
+            async for attempt in AsyncRetrying(
+                stop=stop_after_attempt(self.max_retries),
+                wait=wait_exponential(multiplier=1, min=1, max=32),
+                retry=retry_if_exception(_should_retry_exception),
+                reraise=True,
+            ):
+                with attempt:
+                    download_response = await client.get(presigned_response.url)
+                    download_response.raise_for_status()
+                    return download_response.content
+        except Exception as e:
+            raise SpreadsheetAPIError(f"Failed to download result: {e}") from e
+        raise RuntimeError("Tenacity did not execute")
+
+    async def adownload_region_as_dataframe(
+        self,
+        job_id: str,
+        region_id: str,
+        result_type: SpreadsheetResultType = SpreadsheetResultType.TABLE,
+    ) -> "pd.DataFrame":
+        """Download a region result as a pandas DataFrame.
+
+        Args:
+            job_id: ID of the job
+            region_id: ID of the region
+            result_type: Type of result to download (region or cell_metadata)
+
+        Returns:
+            pandas DataFrame
+        """
+        import pandas as pd
+
+        parquet_bytes = await self.adownload_region_result(
+            job_id, region_id, result_type
+        )
+        return pd.read_parquet(io.BytesIO(parquet_bytes))
+
+    async def aextract_regions(
+        self,
+        file_obj: FileInput,
+        config: dict | SpreadsheetParsingConfig | None = None,
+    ) -> SpreadsheetJobResult:
+        """High-level method to parse a spreadsheet file.
+
+        This method handles the entire workflow:
+        1. Upload the file
+        2. Create a parsing job
+        3. Wait for completion
+        4. Return results
+
+        Args:
+            file_obj: File to parse (path, bytes, or file-like object)
+            config: Parsing configuration
+
+        Returns:
+            SpreadsheetJobResult with parsing results
+        """
+        # Upload file
+        file_response = await self.aupload_file(file_obj)
+
+        # Create job
+        job = await self.acreate_job(file_response.id, config)
+
+        # Wait for completion
+        return await self.await_for_completion(job.id)
+
+    async def aclose(self) -> None:
+        """Close all HTTP clients (async)"""
+        if self._async_client:
+            await self._async_client.aclose()
+
+    async def __aenter__(self) -> "LlamaSheets":
+        return self
+
+    async def __aexit__(self, _exc_type, _exc_val, _exc_tb) -> None:  # type: ignore
+        await self.aclose()

py/llama_cloud_services/beta/sheets/types.py

@@ -0,0 +1,158 @@
+from __future__ import annotations
+
+from datetime import datetime
+from enum import Enum
+
+from pydantic import BaseModel, ConfigDict, Field, field_validator
+
+
+class SpreadsheetResultType(str, Enum):
+    TABLE = "table"
+    EXTRA = "extra"
+    CELL_METADATA = "cell_metadata"
+
+    def __str__(self) -> str:
+        return self.value
+
+
+class ExtractedRegionSummary(BaseModel):
+    """A summary of a single extracted region from a spreadsheet"""
+
+    region_id: str = Field(
+        ...,
+        description="Unique identifier for this region within the file",
+    )
+    sheet_name: str = Field(..., description="Worksheet name where region was found")
+    location: str = Field(..., description="Location of the region in the spreadsheet")
+    title: str | None = Field(None, description="Generated title for the region")
+    description: str | None = Field(
+        None, description="Generated description of the region"
+    )
+    region_type: SpreadsheetResultType = Field(
+        ..., description="Type of the extracted region"
+    )
+
+
+class WorksheetMetadata(BaseModel):
+    """Metadata about a worksheet in a spreadsheet"""
+
+    sheet_name: str = Field(..., description="Name of the worksheet")
+    title: str | None = Field(None, description="Generated title for the worksheet")
+    description: str | None = Field(
+        None, description="Generated description of the worksheet"
+    )
+
+
+class SpreadsheetParseResult(BaseModel):
+    """Result of parsing a single spreadsheet file"""
+
+    success: bool = Field(..., description="Whether parsing was successful")
+    file_name: str = Field(..., description="Original filename")
+
+    regions: list[ExtractedRegionSummary] = Field(
+        default_factory=list, description="All successfully extracted regions"
+    )
+    worksheet_metadata: list[WorksheetMetadata] = Field(
+        default_factory=list, description="Metadata for each processed worksheet"
+    )
+
+    # Error information
+    errors: list[str] = Field(
+        default_factory=list, description="Any errors encountered during parsing"
+    )
+
+
+class SpreadsheetParsingConfig(BaseModel):
+    """Configuration for spreadsheet parsing and region extraction"""
+
+    model_config = ConfigDict(extra="forbid")
+
+    sheet_names: list[str] | None = Field(
+        default=None,
+        description="The names of the sheets to extract regions from. If empty, the default sheet is extracted.",
+    )
+    include_hidden_cells: bool = Field(
+        default=True,
+        description="Whether to include hidden cells when extracting regions from the spreadsheet.",
+    )
+    extraction_range: str | None = Field(
+        default=None,
+        description="A1 notation of the range to extract a single region from. If None, the entire sheet is used.",
+    )
+    generate_additional_metadata: bool = Field(
+        default=True,
+        description="Whether to generate additional metadata (title, description) for each extracted region.",
+    )
+    use_experimental_processing: bool = Field(
+        default=False,
+        description="Enables experimental processing. Accuracy may be impacted.",
+    )
+
+
+class SpreadsheetJob(BaseModel):
+    """A spreadsheet parsing job"""
+
+    id: str = Field(..., description="The ID of the job")
+    user_id: str = Field(..., description="The ID of the user")
+    project_id: str = Field(..., description="The ID of the project")
+    file: dict = Field(..., description="The file object being parsed")
+    config: SpreadsheetParsingConfig = Field(
+        ..., description="Configuration for the parsing job"
+    )
+    status: str = Field(..., description="The status of the parsing job")
+    created_at: str = Field(..., description="When the job was created")
+    updated_at: str = Field(..., description="When the job was last updated")
+
+    @field_validator("created_at", "updated_at", mode="before")
+    def validate_dates(cls, v: str) -> str:
+        """Validate that the dates are in the correct format"""
+        if isinstance(v, datetime):
+            return v.isoformat()
+        else:
+            return v
+
+
+class SpreadsheetJobResult(SpreadsheetJob):
+    """A spreadsheet parsing job result."""
+
+    # Results are included when the job is complete
+    success: bool | None = Field(
+        None, description="Whether the job completed successfully"
+    )
+    regions: list[ExtractedRegionSummary] = Field(
+        default_factory=list,
+        description="All extracted regions (populated when job is complete)",
+    )
+    worksheet_metadata: list[WorksheetMetadata] = Field(
+        default_factory=list,
+        description="Metadata for each processed worksheet (populated when job is complete)",
+    )
+    errors: list[str] = Field(
+        default_factory=list, description="Any errors encountered"
+    )
+
+
+class JobStatus(str, Enum):
+    """Status of a spreadsheet parsing job"""
+
+    PENDING = "PENDING"
+    IN_PROGRESS = "IN_PROGRESS"
+    SUCCESS = "SUCCESS"
+    PARTIAL_SUCCESS = "PARTIAL_SUCCESS"
+    ERROR = "ERROR"
+    FAILURE = "FAILURE"
+
+
+class PresignedUrlResponse(BaseModel):
+    """Response containing a presigned URL for downloading results"""
+
+    url: str = Field(..., description="The presigned URL for downloading")
+
+
+class FileUploadResponse(BaseModel):
+    """Response from uploading a file"""
+
+    id: str = Field(..., description="The ID of the uploaded file")
+    name: str = Field(..., description="The name of the file")
+    project_id: str = Field(..., description="The project ID")
+    user_id: str = Field(..., description="The user ID")

py/llama_cloud_services/constants.py

@@ -1,2 +1,3 @@
+BASE_URL = "https://api.cloud.llamaindex.ai"
 EU_BASE_URL = "https://api.cloud.eu.llamaindex.ai"
 POLLING_TIMEOUT_SECONDS = 300.0

py/llama_cloud_services/extract/__init__.py

@@ -2,15 +2,16 @@ from llama_cloud_services.extract.extract import (
     LlamaExtract,
     ExtractConfig,
     ExtractionAgent,
-    SourceText,
     ExtractTarget,
     ExtractMode,
 )
+from llama_cloud_services.utils import SourceText, FileInput
 
 __all__ = [
     "LlamaExtract",
     "ExtractionAgent",
     "SourceText",
+    "FileInput",
     "ExtractConfig",
     "ExtractTarget",
     "ExtractMode",

py/llama_cloud_services/extract/extract.py

@@ -2,10 +2,9 @@ import asyncio
 import base64
 import os
 import time
-from io import BufferedIOBase, BufferedReader, BytesIO, TextIOWrapper
+from io import BufferedIOBase, TextIOWrapper
 from pathlib import Path
 from typing import List, Optional, Type, Union, Coroutine, Any, TypeVar
-import secrets
 import warnings
 import httpx
 from pydantic import BaseModel
@@ -33,7 +32,8 @@ from llama_cloud_services.extract.utils import (
     JSONObjectType,
     ExperimentalWarning,
 )
-from llama_cloud_services.utils import augment_async_errors
+from llama_cloud_services.utils import augment_async_errors, SourceText, FileInput
+from llama_cloud_services.files.client import FileClient
 from llama_index.core.schema import BaseComponent
 from llama_index.core.async_utils import run_jobs
 from llama_index.core.bridge.pydantic import Field, PrivateAttr
@@ -188,46 +188,6 @@ async def _wait_for_job_result(
             )
 
 
-class SourceText:
-    def __init__(
-        self,
-        *,
-        file: Union[bytes, BufferedIOBase, TextIOWrapper, str, Path, None] = None,
-        text_content: Optional[str] = None,
-        filename: Optional[str] = None,
-    ):
-        self.file = file
-        self.filename = filename
-        self.text_content = text_content
-        self._validate()
-
-    def _validate(self) -> None:
-        """Ensure filename is provided when needed."""
-        if not ((self.file is None) ^ (self.text_content is None)):
-            raise ValueError("Either file or text_content must be provided.")
-        if self.text_content is not None:
-            if not self.filename:
-                random_hex = secrets.token_hex(4)
-                self.filename = f"text_input_{random_hex}.txt"
-            return
-
-        if isinstance(self.file, (bytes, BufferedIOBase, TextIOWrapper)):
-            if not self.filename and hasattr(self.file, "name"):
-                self.filename = os.path.basename(str(self.file.name))
-            elif not hasattr(self.file, "name") and self.filename is None:
-                raise ValueError(
-                    "filename must be provided when file is bytes or a file-like object without a name"
-                )
-        elif isinstance(self.file, (str, Path)):
-            if not self.filename:
-                self.filename = os.path.basename(str(self.file))
-        else:
-            raise ValueError(f"Unsupported file type: {type(self.file)}")
-
-
-FileInput = Union[str, Path, BufferedIOBase, SourceText, File]
-
-
 def run_in_thread(
     coro: Coroutine[Any, Any, T],
     thread_pool: ThreadPoolExecutor,
@@ -320,6 +280,7 @@ class ExtractionAgent:
         self._thread_pool = ThreadPoolExecutor(
             max_workers=min(10, (os.cpu_count() or 1) + 4)
         )
+        self._file_client = FileClient(client, project_id, organization_id)
 
     @property
     def id(self) -> str:
@@ -369,65 +330,11 @@ class ExtractionAgent:
             ValueError: If filename is not provided for bytes input or for file-like objects
                        without a name attribute.
         """
-        file_contents: Optional[Union[BufferedIOBase, BytesIO]] = None
-        try:
-            if file_input.text_content is not None:
-                # Handle direct text content
-                file_contents = BytesIO(file_input.text_content.encode("utf-8"))
-            elif isinstance(file_input.file, TextIOWrapper):
-                # Handle text-based IO objects
-                file_contents = BytesIO(file_input.file.read().encode("utf-8"))
-            elif isinstance(file_input.file, (str, Path)):
-                # Handle file paths
-                file_contents = open(file_input.file, "rb")
-            elif isinstance(file_input.file, bytes):
-                # Handle bytes
-                file_contents = BytesIO(file_input.file)
-            elif isinstance(file_input.file, BufferedIOBase):
-                # Handle binary IO objects
-                file_contents = file_input.file
-            else:
-                raise ValueError(f"Unsupported file type: {type(file_input.file)}")
-
-            # Add name attribute to file object if needed
-            if not hasattr(file_contents, "name"):
-                file_contents.name = file_input.filename  # type: ignore
-
-            return await self._client.files.upload_file(
-                project_id=self._project_id, upload_file=file_contents
-            )
-        finally:
-            if file_contents is not None and isinstance(
-                file_contents, (BufferedReader, BytesIO)
-            ):
-                file_contents.close()
+        return await self._file_client.upload_content(file_input)
 
     async def _upload_file(self, file_input: FileInput) -> File:
-        source_text = None
-        if isinstance(file_input, File):
-            return file_input
-        if isinstance(file_input, SourceText):
-            source_text = file_input
-        elif isinstance(file_input, (str, Path)):
-            path = Path(file_input)
-            source_text = SourceText(file=path, filename=path.name)
-        else:
-            # Try to get filename from the file object if not provided
-            filename = None
-            if hasattr(file_input, "name"):
-                filename = os.path.basename(str(file_input.name))
-            if filename is None:
-                raise ValueError(
-                    "Use SourceText to provide filename when uploading bytes or file-like objects."
-                )
-
-            warnings.warn(
-                "Use SourceText instead of bytes or file-like objects",
-                DeprecationWarning,
-            )
-            source_text = SourceText(file=file_input, filename=filename)
-
-        return await self.upload_file(source_text)
+        """Upload a file from various input types using FileClient."""
+        return await self._file_client.upload_content(file_input)
 
     async def _wait_for_job_result(self, job_id: str) -> Optional[ExtractRun]:
         """Wait for and return the results of an extraction job."""

py/llama_cloud_services/files/client.py

@@ -1,9 +1,11 @@
 from io import BytesIO
 from typing import BinaryIO
 import os
+from pathlib import Path
 from llama_cloud.client import AsyncLlamaCloud
 from llama_cloud.types import File, FileCreate
 from typing import Optional
+from llama_cloud_services.utils import SourceText, FileInput
 
 
 class FileClient:
@@ -95,3 +97,83 @@ class FileClient:
                 project_id=self.project_id,
                 organization_id=self.organization_id,
             )
+
+    async def upload_content(
+        self, file_input: FileInput, external_file_id: Optional[str] = None
+    ) -> File:
+        """
+        Upload content from various input types or fetch an already-uploaded file.
+
+        Args:
+            file_input: The content to upload. Can be:
+                - File: Already uploaded file (returned as-is)
+                - str/Path: Path to a file on disk
+                - SourceText: Text content, file, or file_id with explicit filename
+                - BufferedIOBase: File-like binary object
+            external_file_id: Optional external identifier for the file
+
+        Returns:
+            File: The uploaded (or fetched) file object
+
+        Raises:
+            ValueError: If the input type is not supported or required info is missing
+        """
+        # If already a File object, return it
+        if isinstance(file_input, File):
+            return file_input
+
+        # Handle SourceText
+        if isinstance(file_input, SourceText):
+            # If file_id is provided, fetch the file object
+            if file_input.file_id is not None:
+                return await self.get_file(file_input.file_id)
+            elif file_input.text_content is not None:
+                # Handle direct text content
+                text_bytes = file_input.text_content.encode("utf-8")
+                return await self.upload_bytes(
+                    text_bytes, external_file_id or file_input.filename or "file"
+                )
+            elif isinstance(file_input.file, (str, Path)):
+                # Handle file paths using the existing upload_file method
+                return await self.upload_file(
+                    str(file_input.file), external_file_id or file_input.filename
+                )
+            elif isinstance(file_input.file, bytes):
+                # Handle bytes
+                return await self.upload_bytes(
+                    file_input.file, external_file_id or file_input.filename or "file"
+                )
+            elif hasattr(file_input.file, "read"):
+                # Handle any file-like object (TextIOWrapper, BytesIO, BufferedReader, BufferedIOBase, etc.)
+                content = file_input.file.read()  # type: ignore
+                if isinstance(content, str):
+                    content = content.encode("utf-8")
+                return await self.upload_bytes(
+                    content, external_file_id or file_input.filename or "file"
+                )
+            else:
+                raise ValueError(f"Unsupported file type: {type(file_input.file)}")
+
+        # Handle string/Path directly
+        elif isinstance(file_input, (str, Path)):
+            return await self.upload_file(str(file_input), external_file_id)
+
+        # Handle raw file-like objects
+        elif hasattr(file_input, "read"):
+            if hasattr(file_input, "name"):
+                filename = os.path.basename(str(file_input.name))
+            else:
+                filename = external_file_id or "file"
+
+            # Read content to determine size
+            content = file_input.read()
+            if isinstance(content, str):
+                content = content.encode("utf-8")
+
+            return await self.upload_bytes(content, external_file_id or filename)
+
+        else:
+            raise ValueError(
+                f"Unsupported file input type: {type(file_input)}. "
+                f"Supported types: str, Path, SourceText, BufferedIOBase, or File."
+            )

py/llama_cloud_services/index/api_utils.py

@@ -258,6 +258,7 @@ def page_screenshot_nodes_to_node_with_score(
     client: LlamaCloud,
     raw_image_nodes: Optional[List[PageScreenshotNodeWithScore]],
     project_id: str,
+    metadata: Optional[dict] = None,
 ) -> List[NodeWithScore]:
     if not raw_image_nodes:
         return []
@@ -273,6 +274,7 @@ def page_screenshot_nodes_to_node_with_score(
         image_base64 = base64.b64encode(image_bytes).decode("utf-8")
         image_node_metadata: Dict[str, Any] = {
             **(raw_image_node.node.metadata or {}),
+            **(metadata or {}),
             "file_id": raw_image_node.node.file_id,
             "page_index": raw_image_node.node.page_index,
         }
@@ -289,6 +291,7 @@ def image_nodes_to_node_with_score(
     client: LlamaCloud,
     raw_image_nodes: Optional[List[PageScreenshotNodeWithScore]],
     project_id: str,
+    metadata: Optional[dict] = None,
 ) -> List[NodeWithScore]:
     """
     Legacy method to alias page_screenshot_nodes_to_node_with_score.
@@ -297,7 +300,10 @@ def image_nodes_to_node_with_score(
         return []
 
     return page_screenshot_nodes_to_node_with_score(
-        client=client, raw_image_nodes=raw_image_nodes, project_id=project_id
+        client=client,
+        raw_image_nodes=raw_image_nodes,
+        project_id=project_id,
+        metadata=metadata,
     )
 
 
@@ -305,6 +311,7 @@ def page_figure_nodes_to_node_with_score(
     client: LlamaCloud,
     raw_figure_nodes: Optional[List[PageFigureNodeWithScore]],
     project_id: str,
+    metadata: Optional[dict] = None,
 ) -> List[NodeWithScore]:
     if not raw_figure_nodes:
         return []
@@ -321,6 +328,7 @@ def page_figure_nodes_to_node_with_score(
         figure_base64 = base64.b64encode(figure_bytes).decode("utf-8")
         figure_node_metadata: Dict[str, Any] = {
             **(raw_figure_node.node.metadata or {}),
+            **(metadata or {}),
             "file_id": raw_figure_node.node.file_id,
             "page_index": raw_figure_node.node.page_index,
             "figure_name": raw_figure_node.node.figure_name,
@@ -337,6 +345,7 @@ async def apage_screenshot_nodes_to_node_with_score(
     client: AsyncLlamaCloud,
     raw_image_nodes: Optional[List[PageScreenshotNodeWithScore]],
     project_id: str,
+    metadata: Optional[dict] = None,
 ) -> List[NodeWithScore]:
     if not raw_image_nodes:
         return []
@@ -357,6 +366,7 @@ async def apage_screenshot_nodes_to_node_with_score(
         image_base64 = base64.b64encode(image_bytes).decode("utf-8")
         image_node_metadata: Dict[str, Any] = {
             **(raw_image_node.node.metadata or {}),
+            **(metadata or {}),
             "file_id": raw_image_node.node.file_id,
             "page_index": raw_image_node.node.page_index,
         }
@@ -372,6 +382,7 @@ async def aimage_nodes_to_node_with_score(
     client: AsyncLlamaCloud,
     raw_image_nodes: Optional[List[PageScreenshotNodeWithScore]],
     project_id: str,
+    metadata: Optional[dict] = None,
 ) -> List[NodeWithScore]:
     """
     Legacy method to alias apage_screenshot_nodes_to_node_with_score.
@@ -380,7 +391,10 @@ async def aimage_nodes_to_node_with_score(
         return []
 
     return await apage_screenshot_nodes_to_node_with_score(
-        client=client, raw_image_nodes=raw_image_nodes, project_id=project_id
+        client=client,
+        raw_image_nodes=raw_image_nodes,
+        project_id=project_id,
+        metadata=metadata,
     )
 
 
@@ -388,6 +402,7 @@ async def apage_figure_nodes_to_node_with_score(
     client: AsyncLlamaCloud,
     raw_figure_nodes: Optional[List[PageFigureNodeWithScore]],
     project_id: str,
+    metadata: Optional[dict] = None,
 ) -> List[NodeWithScore]:
     if not raw_figure_nodes:
         return []
@@ -409,6 +424,7 @@ async def apage_figure_nodes_to_node_with_score(
         figure_base64 = base64.b64encode(figure_bytes).decode("utf-8")
         figure_node_metadata: Dict[str, Any] = {
             **(raw_figure_node.node.metadata or {}),
+            **(metadata or {}),
             "file_id": raw_figure_node.node.file_id,
             "page_index": raw_figure_node.node.page_index,
             "figure_name": raw_figure_node.node.figure_name,

py/llama_cloud_services/index/base.py

@@ -19,6 +19,7 @@ from llama_cloud import (
     PipelineCreate,
     PipelineCreateEmbeddingConfig,
     PipelineCreateTransformConfig,
+    PipelineFileCreateCustomMetadataValue,
     PipelineType,
     ProjectCreate,
     ManagedIngestionStatus,
@@ -333,7 +334,7 @@ class LlamaCloudIndex(BaseManagedIndex):
         if file_ids:
             self._wait_for_resources(
                 file_ids,
-                lambda fid: self._client.pipelines.get_pipeline_file_status(
+                lambda fid: self._client.pipeline_files.get_pipeline_file_status(
                     pipeline_id=self.pipeline.id, file_id=fid
                 ),
                 resource_name="file",
@@ -420,7 +421,7 @@ class LlamaCloudIndex(BaseManagedIndex):
         if file_ids:
             await self._await_for_resources(
                 file_ids,
-                lambda fid: self._aclient.pipelines.get_pipeline_file_status(
+                lambda fid: self._aclient.pipeline_files.get_pipeline_file_status(
                     pipeline_id=self.pipeline.id, file_id=fid
                 ),
                 resource_name="file",
@@ -905,6 +906,9 @@ class LlamaCloudIndex(BaseManagedIndex):
     def upload_file(
         self,
         file_path: str,
+        custom_metadata: Optional[
+            dict[str, Optional[PipelineFileCreateCustomMetadataValue]]
+        ] = None,
         verbose: bool = False,
         wait_for_ingestion: bool = True,
         raise_on_error: bool = False,
@@ -918,8 +922,10 @@ class LlamaCloudIndex(BaseManagedIndex):
                 print(f"Uploaded file {file.id} with name {file.name}")
 
         # Add file to pipeline
-        pipeline_file_create = PipelineFileCreate(file_id=file.id)
-        self._client.pipelines.add_files_to_pipeline_api(
+        pipeline_file_create = PipelineFileCreate(
+            file_id=file.id, custom_metadata=custom_metadata
+        )
+        self._client.pipeline_files.add_files_to_pipeline_api(
             pipeline_id=self.pipeline.id, request=[pipeline_file_create]
         )
 
@@ -932,6 +938,9 @@ class LlamaCloudIndex(BaseManagedIndex):
     async def aupload_file(
         self,
         file_path: str,
+        custom_metadata: Optional[
+            dict[str, Optional[PipelineFileCreateCustomMetadataValue]]
+        ] = None,
         verbose: bool = False,
         wait_for_ingestion: bool = True,
         raise_on_error: bool = False,
@@ -945,8 +954,10 @@ class LlamaCloudIndex(BaseManagedIndex):
                 print(f"Uploaded file {file.id} with name {file.name}")
 
         # Add file to pipeline
-        pipeline_file_create = PipelineFileCreate(file_id=file.id)
-        await self._aclient.pipelines.add_files_to_pipeline_api(
+        pipeline_file_create = PipelineFileCreate(
+            file_id=file.id, custom_metadata=custom_metadata
+        )
+        await self._aclient.pipeline_files.add_files_to_pipeline_api(
             pipeline_id=self.pipeline.id, request=[pipeline_file_create]
         )
 
@@ -961,6 +972,9 @@ class LlamaCloudIndex(BaseManagedIndex):
         self,
         file_name: str,
         url: str,
+        custom_metadata: Optional[
+            dict[str, Optional[PipelineFileCreateCustomMetadataValue]]
+        ] = None,
         proxy_url: Optional[str] = None,
         request_headers: Optional[Dict[str, str]] = None,
         verify_ssl: bool = True,
@@ -983,8 +997,10 @@ class LlamaCloudIndex(BaseManagedIndex):
             print(f"Uploaded file {file.id} with ID {file.id}")
 
         # Add file to pipeline
-        pipeline_file_create = PipelineFileCreate(file_id=file.id)
-        self._client.pipelines.add_files_to_pipeline_api(
+        pipeline_file_create = PipelineFileCreate(
+            file_id=file.id, custom_metadata=custom_metadata
+        )
+        self._client.pipeline_files.add_files_to_pipeline_api(
             pipeline_id=self.pipeline.id, request=[pipeline_file_create]
         )
 
@@ -998,6 +1014,9 @@ class LlamaCloudIndex(BaseManagedIndex):
         self,
         file_name: str,
         url: str,
+        custom_metadata: Optional[
+            dict[str, Optional[PipelineFileCreateCustomMetadataValue]]
+        ] = None,
         proxy_url: Optional[str] = None,
         request_headers: Optional[Dict[str, str]] = None,
         verify_ssl: bool = True,
@@ -1020,8 +1039,10 @@ class LlamaCloudIndex(BaseManagedIndex):
             print(f"Uploaded file {file.id} with ID {file.id}")
 
         # Add file to pipeline
-        pipeline_file_create = PipelineFileCreate(file_id=file.id)
-        await self._aclient.pipelines.add_files_to_pipeline_api(
+        pipeline_file_create = PipelineFileCreate(
+            file_id=file.id, custom_metadata=custom_metadata
+        )
+        await self._aclient.pipeline_files.add_files_to_pipeline_api(
             pipeline_id=self.pipeline.id, request=[pipeline_file_create]
         )
 

py/llama_cloud_services/index/retriever.py

@@ -129,11 +129,12 @@ class LlamaCloudRetriever(BaseRetriever):
         )
 
     def _result_nodes_to_node_with_score(
-        self, result_nodes: List[TextNodeWithScore]
+        self, result_nodes: List[TextNodeWithScore], metadata: Optional[dict] = None
     ) -> List[NodeWithScore]:
         nodes = []
         for res in result_nodes:
-            text_node = TextNode.parse_obj(res.node.dict())
+            text_node = TextNode.model_validate(res.node.dict())
+            text_node.metadata.update(metadata or {})
             nodes.append(NodeWithScore(node=text_node, score=res.score))
 
         return nodes
@@ -161,17 +162,25 @@ class LlamaCloudRetriever(BaseRetriever):
             search_filters_inference_schema=search_filters_inference_schema,
         )
 
-        result_nodes = self._result_nodes_to_node_with_score(results.retrieval_nodes)
+        result_nodes = self._result_nodes_to_node_with_score(
+            results.retrieval_nodes, metadata=results.metadata
+        )
         if self._retrieve_page_screenshot_nodes:
             result_nodes.extend(
                 page_screenshot_nodes_to_node_with_score(
-                    self._client, results.image_nodes, self.project.id
+                    self._client,
+                    results.image_nodes,
+                    self.project.id,
+                    metadata=results.metadata,
                 )
             )
         if self._retrieve_page_figure_nodes:
             result_nodes.extend(
                 page_figure_nodes_to_node_with_score(
-                    self._client, results.page_figure_nodes, self.project.id
+                    self._client,
+                    results.page_figure_nodes,
+                    self.project.id,
+                    metadata=results.metadata,
                 )
             )
 
@@ -200,17 +209,25 @@ class LlamaCloudRetriever(BaseRetriever):
             search_filters_inference_schema=search_filters_inference_schema,
         )
 
-        result_nodes = self._result_nodes_to_node_with_score(results.retrieval_nodes)
+        result_nodes = self._result_nodes_to_node_with_score(
+            results.retrieval_nodes, metadata=results.metadata
+        )
         if self._retrieve_page_screenshot_nodes:
             result_nodes.extend(
                 await apage_screenshot_nodes_to_node_with_score(
-                    self._aclient, results.image_nodes, self.project.id
+                    self._aclient,
+                    results.image_nodes,
+                    self.project.id,
+                    metadata=results.metadata,
                 )
             )
         if self._retrieve_page_figure_nodes:
             result_nodes.extend(
                 await apage_figure_nodes_to_node_with_score(
-                    self._aclient, results.page_figure_nodes, self.project.id
+                    self._aclient,
+                    results.page_figure_nodes,
+                    self.project.id,
+                    metadata=results.metadata,
                 )
             )
 

py/llama_cloud_services/parse/base.py

@@ -1146,6 +1146,25 @@ class LlamaParse(BasePydanticReader):
                     )
                 current_interval = self._calculate_backoff(current_interval)
 
+    async def _get_job_result_with_error_handling(
+        self, job_id: str, result_type: str, verbose: bool = False
+    ) -> Dict[str, Any]:
+        """Get job result with error handling based on ignore_errors setting."""
+        try:
+            return await self._get_job_result(job_id, result_type, verbose=verbose)
+        except JobFailedException as e:
+            if self.ignore_errors:
+                # Return error information when ignore_errors is True
+                return {
+                    "pages": [],
+                    "job_metadata": {},
+                    "error": f"{e.status}: {e.error_message or 'No error message'}",
+                    "error_code": e.error_code,
+                    "status": e.status,
+                }
+            else:
+                raise e
+
     async def _parse_one(
         self,
         file_path: FileInput,
@@ -1187,7 +1206,7 @@ class LlamaParse(BasePydanticReader):
         )
         if self.verbose:
             print("Started parsing the file under job_id %s" % job_id)
-        result = await self._get_job_result(
+        result = await self._get_job_result_with_error_handling(
             job_id, result_type or self.result_type.value, verbose=self.verbose
         )
         return job_id, result
@@ -1250,6 +1269,15 @@ class LlamaParse(BasePydanticReader):
                     result_type=ResultType.JSON.value,
                     partition_target_pages=f"{total}-{total + size - 1}",
                 )
+                # Check if the result is an error result (when ignore_errors=True)
+                if json_result.get("error_code") == "NO_DATA_FOUND_IN_FILE":
+                    raise JobFailedException(
+                        job_id=job_id,
+                        status=json_result.get("status", "ERROR"),
+                        error_code=json_result.get("error_code"),
+                        error_message=json_result.get("error"),
+                    )
+
                 result_type = result_type or self.result_type.value
                 if result_type == ResultType.JSON.value:
                     job_result = json_result
@@ -1775,7 +1803,7 @@ class LlamaParse(BasePydanticReader):
             JobResult object or list of JobResult objects if multiple job IDs were provided.
         """
         if isinstance(job_id, str):
-            result = await self._get_job_result(
+            result = await self._get_job_result_with_error_handling(
                 job_id, ResultType.JSON.value, verbose=self.verbose
             )
             return JobResult(
@@ -1790,7 +1818,9 @@ class LlamaParse(BasePydanticReader):
         elif isinstance(job_id, list):
             results = []
             jobs = [
-                self._get_job_result(id_, ResultType.JSON.value, verbose=self.verbose)
+                self._get_job_result_with_error_handling(
+                    id_, ResultType.JSON.value, verbose=self.verbose
+                )
                 for id_ in job_id
             ]
             results = await run_jobs(

py/llama_cloud_services/parse/types.py

@@ -269,6 +269,13 @@ class JobResult(SafeBaseModel):
     error: Optional[str] = Field(
         default=None, description="The error message if the job failed."
     )
+    error_code: Optional[str] = Field(
+        default=None, description="The error code if the job failed."
+    )
+    status: Optional[str] = Field(
+        default=None,
+        description="The job status (e.g., PENDING, SUCCESS, ERROR, CANCELED).",
+    )
 
     def __init__(
         self,

py/llama_cloud_services/testing_utils/__init__.py

@@ -0,0 +1,9 @@
+from .matchers import FileMatcher, RequestMatcher, SchemaMatcher
+from .server import FakeLlamaCloudServer
+
+__all__ = [
+    "FakeLlamaCloudServer",
+    "FileMatcher",
+    "SchemaMatcher",
+    "RequestMatcher",
+]

py/llama_cloud_services/testing_utils/_deterministic.py

@@ -0,0 +1,192 @@
+from __future__ import annotations
+
+import hashlib
+import json
+import random
+from datetime import datetime, timezone
+from typing import Any, Iterable, Mapping, MutableMapping
+
+
+def hash_chunks(chunks: Iterable[bytes]) -> str:
+    digest = hashlib.sha256()
+    for chunk in chunks:
+        digest.update(chunk)
+    return digest.hexdigest()
+
+
+def fingerprint_file(content: bytes, filename: str | None = None) -> str:
+    name_bytes = filename.encode("utf-8") if filename else b""
+    return hash_chunks((content, name_bytes))
+
+
+def hash_schema(schema: Any) -> str:
+    json_string = json.dumps(
+        _to_serializable(schema),
+        sort_keys=True,
+        separators=(",", ":"),
+    )
+    return hashlib.sha256(json_string.encode("utf-8")).hexdigest()
+
+
+def combined_seed(*parts: str) -> int:
+    digest = hash_chunks(tuple(part.encode("utf-8") for part in parts))
+    return int(digest[:16], 16)
+
+
+def generate_data_from_schema(schema: Any, seed: int) -> Any:
+    rng = random.Random(seed)
+    return _generate_value(schema, rng, depth=0)
+
+
+def generate_text_blob(seed: int, sentences: int = 3) -> str:
+    rng = random.Random(seed)
+    words = [
+        "aurora",
+        "copper",
+        "delta",
+        "ember",
+        "fable",
+        "glyph",
+        "harbor",
+        "iris",
+        "juniper",
+        "kepler",
+        "lumen",
+        "monarch",
+        "nylon",
+        "onyx",
+        "paragon",
+        "quartz",
+        "raptor",
+        "solstice",
+        "topaz",
+        "umbra",
+        "verdant",
+        "willow",
+        "xenon",
+        "yonder",
+        "zephyr",
+    ]
+    sentence_pieces = []
+    for _ in range(sentences):
+        length = rng.randint(6, 12)
+        chosen = rng.sample(words, k=length)
+        sentence = " ".join(chosen).capitalize() + "."
+        sentence_pieces.append(sentence)
+    return " ".join(sentence_pieces)
+
+
+def utcnow() -> datetime:
+    return datetime.now(timezone.utc)
+
+
+def _to_serializable(value: Any) -> Any:
+    if value is None:
+        return None
+    if isinstance(value, (str, int, float, bool)):
+        return value
+    if isinstance(value, bytes):
+        return value.decode("utf-8", errors="ignore")
+    if isinstance(value, Mapping):
+        return {key: _to_serializable(val) for key, val in value.items()}
+    if isinstance(value, MutableMapping):
+        return {key: _to_serializable(val) for key, val in value.items()}
+    if isinstance(value, (list, tuple, set)):
+        return [_to_serializable(item) for item in value]
+    if hasattr(value, "model_dump_json"):
+        return json.loads(value.model_dump_json())
+    if hasattr(value, "model_dump"):
+        return value.model_dump()
+    if hasattr(value, "dict"):
+        return value.dict()  # type: ignore[call-arg]
+    if hasattr(value, "model_json_schema"):
+        return value.model_json_schema()
+    return str(value)
+
+
+def _generate_value(schema: Any, rng: random.Random, depth: int) -> Any:
+    if depth > 8:
+        return rng.choice(
+            (
+                rng.randint(1, 999),
+                rng.random(),
+                generate_text_blob(rng.randint(0, 1_000_000), sentences=1),
+            )
+        )
+
+    if schema is None:
+        return generate_text_blob(rng.randint(0, 1_000_000), sentences=1)
+
+    if isinstance(schema, list):
+        return [_generate_value(item, rng, depth + 1) for item in schema]
+
+    if isinstance(schema, str):
+        return f"{schema}-{rng.randint(100, 999)}"
+
+    if isinstance(schema, Mapping):
+        if "enum" in schema:
+            options = schema["enum"]
+            if options:
+                index = rng.randint(0, len(options) - 1)
+                return options[index]
+
+        schema_type = schema.get("type")
+
+        if schema_type == "object":
+            properties = schema.get("properties", {})
+            result = {}
+            for key, subschema in properties.items():
+                result[key] = _generate_value(subschema, rng, depth + 1)
+            return result
+
+        if schema_type == "array":
+            items_schema = schema.get("items", {})
+            min_items = schema.get("minItems", 1)
+            max_items = schema.get("maxItems", max(3, min_items))
+            length = rng.randint(min_items, min(min_items + 2, max_items))
+            return [
+                _generate_value(items_schema, rng, depth + 1) for _ in range(length)
+            ]
+
+        if schema_type == "integer":
+            minimum = schema.get("minimum", 0)
+            maximum = schema.get("maximum", minimum + 500)
+            return rng.randint(int(minimum), int(maximum))
+
+        if schema_type == "number":
+            minimum = schema.get("minimum", 0.0)
+            maximum = schema.get("maximum", minimum + 500.0)
+            value = rng.uniform(float(minimum), float(maximum))
+            return round(value, 2)
+
+        if schema_type == "boolean":
+            return rng.choice((True, False))
+
+        if schema_type == "string":
+            fmt = schema.get("format")
+            if fmt == "date-time":
+                timestamp = utcnow().isoformat()
+                return timestamp
+            if fmt == "email":
+                return f"user{rng.randint(1000, 9999)}@example.com"
+            if fmt == "uri":
+                return f"https://example.com/{rng.randint(1000, 9999)}"
+            min_length = schema.get("minLength", 5)
+            max_length = schema.get("maxLength", max(10, min_length))
+            length = rng.randint(min_length, min(min_length + 5, max_length))
+            return generate_text_blob(
+                rng.randint(0, 1_000_000), sentences=max(1, length // 5)
+            )
+
+        if schema_type == "null":
+            return None
+
+        if "oneOf" in schema:
+            option = rng.choice(schema["oneOf"])
+            return _generate_value(option, rng, depth + 1)
+
+        if "anyOf" in schema:
+            option = rng.choice(schema["anyOf"])
+            return _generate_value(option, rng, depth + 1)
+
+    return generate_text_blob(rng.randint(0, 1_000_000), sentences=1)

py/llama_cloud_services/testing_utils/agent_data.py

@@ -0,0 +1,335 @@
+from __future__ import annotations
+
+import httpx
+import re
+from dataclasses import dataclass
+from typing import TYPE_CHECKING, Any, Dict
+from ._deterministic import utcnow, hash_schema
+
+if TYPE_CHECKING:
+    from .server import FakeLlamaCloudServer
+
+
+@dataclass
+class StoredAgentData:
+    data: dict[str, Any]
+    id: str
+    collection: str
+    deployment_name: str
+
+    def __getattr__(self, name: str) -> Any:
+        return self.data.get(name)
+
+    def __setattr__(self, name: str, value: Any) -> None:
+        if name in ("data", "id", "collection", "deployment_name"):
+            super().__setattr__(name, value)
+        else:
+            self.data[name] = value
+
+    @classmethod
+    def from_request_data(cls, data: dict[str, Any]) -> "StoredAgentData":
+        return cls(
+            data=data.get("data", {}),
+            collection=data.get("collection", "default"),
+            deployment_name=data.get("deployment_name", ""),
+            id=hash_schema(data.get("data", {}))[:7],
+        )
+
+
+def apply_filter(data: dict, filters: dict) -> bool:
+    """Check if data matches all filters"""
+    ops = {
+        "gt": lambda a, b: a > b,
+        "gte": lambda a, b: a >= b,
+        "lt": lambda a, b: a < b,
+        "lte": lambda a, b: a <= b,
+        "eq": lambda a, b: a == b,
+        "ne": lambda a, b: a != b,
+        "in": lambda a, b: a in b,
+        "nin": lambda a, b: a not in b,
+    }
+
+    for key, condition in filters.items():
+        if key not in data:
+            return False
+
+        if isinstance(condition, dict):
+            for op, value in condition.items():
+                if op in ops:
+                    if not ops[op](data[key], value):
+                        return False
+                else:
+                    return False
+        else:
+            if data[key] != condition:
+                return False
+
+    return True
+
+
+class FakeAgentDataNamespace:
+    def __init__(
+        self,
+        *,
+        server: "FakeLlamaCloudServer",
+    ) -> None:
+        self._server = server
+        self.stored: list[StoredAgentData] = []
+        self.routes: Dict[str, Any] = {}
+
+    def _create_data(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request=request)
+        data = StoredAgentData.from_request_data(payload)
+        self.stored.append(data)
+        response = {
+            "data": data.data,
+            "collection": data.collection,
+            "deployment_name": data.deployment_name,
+            "created_at": utcnow().isoformat(),
+            "updated_at": None,
+            "id": data.id,
+            "project_id": None,
+            "organization_id": None,
+        }
+        return self._server.json_response(response, status_code=200)
+
+    def _delete_data_by_query(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request=request)
+        delete_count = 0
+        if (filters := payload.get("filter")) is not None:
+            to_keep = []
+            for data in self.stored:
+                if data.collection == payload.get(
+                    "collection", "default"
+                ) and data.deployment_name == payload.get("deployment_name"):
+                    if not apply_filter(data.data, filters):
+                        to_keep.append(data)
+                    else:
+                        delete_count += 1
+            self.stored = to_keep
+        return self._server.json_response(
+            {"deleted_count": delete_count}, status_code=200
+        )
+
+    def _delete_data_by_id(self, request: httpx.Request) -> httpx.Response:
+        item_id = self._find_item_id(request=request)
+        if not item_id:
+            return self._server.json_response(
+                {
+                    "detail": "An item_id path parameter is required to perform this operation"
+                },
+                status_code=400,
+            )
+        if not any(data.id == item_id for data in self.stored):
+            return self._server.json_response(
+                {"detail": f"No data with ID: {item_id}"}, status_code=404
+            )
+        self.stored = [data for data in self.stored if data.id != item_id]
+        return self._server.json_response({}, status_code=200)
+
+    def _get_data_by_id(self, request: httpx.Request) -> httpx.Response:
+        item_id = self._find_item_id(request=request)
+        if not item_id:
+            return self._server.json_response(
+                {
+                    "detail": "An item_id path parameter is required to perform this operation"
+                },
+                status_code=400,
+            )
+        data = [data for data in self.stored if data.id == item_id]
+        if data:
+            response = {
+                "data": data[0].data,
+                "collection": data[0].collection,
+                "deployment_name": data[0].deployment_name,
+                "created_at": utcnow().isoformat(),
+                "updated_at": None,
+                "id": data[0].id,
+                "project_id": None,
+                "organization_id": None,
+            }
+            return self._server.json_response(response, status_code=200)
+        else:
+            return self._server.json_response(
+                {"detail": f"No data with ID: {item_id}"}, status_code=404
+            )
+
+    def _search_data(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request=request)
+        found = []
+        if (filters := payload.get("filter")) is not None:
+            for data in self.stored:
+                if data.collection == payload.get(
+                    "collection", "default"
+                ) and data.deployment_name == payload.get("deployment_name"):
+                    if apply_filter(data.data, filters):
+                        found.append(
+                            {
+                                "data": data.data,
+                                "collection": data.collection,
+                                "deployment_name": data.deployment_name,
+                                "created_at": utcnow().isoformat(),
+                                "updated_at": None,
+                                "id": data.id,
+                                "project_id": None,
+                                "organization_id": None,
+                            }
+                        )
+        else:
+            for data in self.stored:
+                if data.collection == payload.get(
+                    "collection", "default"
+                ) and data.deployment_name == payload.get("deployment_name"):
+                    found.append(
+                        {
+                            "data": data.data,
+                            "collection": data.collection,
+                            "deployment_name": data.deployment_name,
+                            "created_at": utcnow().isoformat(),
+                            "updated_at": None,
+                            "id": data.id,
+                            "project_id": None,
+                            "organization_id": None,
+                        }
+                    )
+        return self._server.json_response(
+            {"items": found, "next_page_token": None, "total_size": len(found)},
+            status_code=200,
+        )
+
+    def _update_data(self, request: httpx.Request) -> httpx.Response:
+        item_id = self._find_item_id(request=request)
+        payload = self._server.json(request=request)
+        if not item_id:
+            return self._server.json_response(
+                {
+                    "detail": "An item_id path parameter is required to perform this operation"
+                },
+                status_code=400,
+            )
+        updated = None
+        for i, data in enumerate(self.stored):
+            if data.id == item_id:
+                updated = data
+                updated.data = payload.get("data", data.data)
+                self.stored[i] = updated
+        print(updated)
+        if updated is not None:
+            response = {
+                "data": updated.data,
+                "collection": updated.collection,
+                "deployment_name": updated.deployment_name,
+                "created_at": None,
+                "updated_at": utcnow().isoformat(),
+                "id": updated.id,
+                "project_id": None,
+                "organization_id": None,
+            }
+            status_code = 200
+        else:
+            response = {"detail": f"Record with id {item_id} not found"}
+            status_code = 404
+        return self._server.json_response(response, status_code=status_code)
+
+    def _aggregate_data(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request=request)
+        add_count = payload.get("count", False)
+        group_bys: list[str] = payload.get("group_by", [])
+        groups: dict[str, dict[str, list[dict]]] = {key: {} for key in group_bys}
+        if (filters := payload.get("filter")) is not None:
+            for data in self.stored:
+                if data.collection == payload.get(
+                    "collection", "default"
+                ) and data.deployment_name == payload.get("deployment_name"):
+                    if apply_filter(data.data, filters):
+                        for key in group_bys:
+                            if key in data.data and data.data[key] in groups[key]:
+                                groups[key][data.data[key]].append(data.data)
+                            elif key in data.data and data.data[key] not in groups[key]:
+                                groups[key][data.data[key]] = [data.data]
+        else:
+            for data in self.stored:
+                if data.collection == payload.get(
+                    "collection", "default"
+                ) and data.deployment_name == payload.get("deployment_name"):
+                    for key in group_bys:
+                        if key in data.data and data.data[key] in groups[key]:
+                            groups[key][data.data[key]].append(data.data)
+                        elif key in data.data and data.data[key] not in groups[key]:
+                            groups[key][data.data[key]] = [data.data]
+
+        response: dict[str, Any] = {
+            "items": [],
+            "next_page_token": None,
+            "total_size": 0,
+        }
+        for k in groups:
+            if len(groups[k]) > 0:
+                for v in groups[k]:
+                    if groups[k][v]:
+                        first_element = groups[k][v][0]
+                    else:
+                        first_element = None
+                    response["items"].append(
+                        {
+                            "first_item": first_element,
+                            "count": len(groups[k][v]) if add_count else None,
+                            "group_key": {k: v},
+                        }
+                    )
+        response["total_size"] = len(response["items"])
+        return self._server.json_response(response, status_code=200)
+
+    def _find_item_id(self, request: httpx.Request) -> str | None:
+        matchgroups = re.search(r"/agent-data\/([^\/]+)$", request.url.path)
+        return matchgroups.group(1) if matchgroups is not None else None
+
+    def register(self) -> None:
+        server = self._server
+        route = server.add_route(
+            "POST",
+            "/api/v1/beta/agent-data",
+            self._create_data,
+            namespace="create_item",
+        )
+        self.routes["stateless_run"] = route
+        self.stateless_run = route
+        server.add_route(
+            "POST",
+            "/api/v1/beta/agent-data/:aggregate",
+            self._aggregate_data,
+            namespace="untyped_aggregate",
+            alias="aggregate",
+        )
+        server.add_route(
+            "POST",
+            "/api/v1/beta/agent-data/:delete",
+            self._delete_data_by_query,
+            namespace="delete",
+        )
+        server.add_route(
+            "POST",
+            "/api/v1/beta/agent-data/:search",
+            self._search_data,
+            namespace="untyped_search",
+            alias="search",
+        )
+        server.add_route(
+            "DELETE",
+            "/api/v1/beta/agent-data/{item_id}",
+            self._delete_data_by_id,
+            namespace="delete_item",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/beta/agent-data/{item_id}",
+            self._get_data_by_id,
+            namespace="untyped_get_item",
+            alias="get_item",
+        )
+        server.add_route(
+            "PUT",
+            "/api/v1/beta/agent-data/{item_id}",
+            self._update_data,
+            namespace="update_item",
+        )

py/llama_cloud_services/testing_utils/classify.py

@@ -0,0 +1,153 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from typing import TYPE_CHECKING, Dict, List
+
+import httpx
+from llama_cloud.types import (
+    ClassifierRule,
+    ClassifyJob,
+    ClassifyJobResults,
+    ClassificationResult,
+    FileClassification,
+    StatusEnum,
+)
+
+from ._deterministic import combined_seed, utcnow
+from .files import FakeFilesNamespace, StoredFile
+
+if TYPE_CHECKING:
+    from .server import FakeLlamaCloudServer
+
+
+@dataclass
+class ClassificationJobRecord:
+    job: ClassifyJob
+    results: ClassifyJobResults
+    files: List[StoredFile]
+
+
+class FakeClassifyNamespace:
+    def __init__(
+        self, *, server: "FakeLlamaCloudServer", files: FakeFilesNamespace
+    ) -> None:
+        self._server = server
+        self._files = files
+        self._jobs: Dict[str, ClassificationJobRecord] = {}
+
+    def register(self) -> None:
+        server = self._server
+        server.add_route(
+            "POST",
+            "/api/v1/classifier/jobs",
+            self._handle_create_job,
+            namespace="classify",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/classifier/jobs",
+            self._handle_list_jobs,
+            namespace="classify",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/classifier/jobs/{job_id}",
+            self._handle_get_job,
+            namespace="classify",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/classifier/jobs/{job_id}/results",
+            self._handle_get_results,
+            namespace="classify",
+        )
+
+    def _handle_create_job(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request)
+        file_ids = payload.get("file_ids", [])
+        rules_payload = payload.get("rules", [])
+        rules = [ClassifierRule.parse_obj(rule) for rule in rules_payload]
+        stored_files = []
+        for file_id in file_ids:
+            stored = self._files.get(file_id)
+            if not stored:
+                return self._server.json_response(
+                    {"detail": f"File {file_id} not found"}, status_code=404
+                )
+            stored_files.append(stored)
+
+        job_id = self._server.new_id("classify-job")
+        job = ClassifyJob(
+            id=job_id,
+            project_id=request.url.params.get(
+                "project_id", self._server.default_project_id
+            ),
+            user_id="fake-user",
+            rules=rules,
+            parsing_configuration=None,
+            status=StatusEnum.SUCCESS,
+            created_at=utcnow(),
+            updated_at=utcnow(),
+            effective_at=utcnow(),
+            error_message=None,
+            job_record_id=None,
+        )
+        results = self._build_results(job_id, stored_files, rules)
+        record = ClassificationJobRecord(job=job, results=results, files=stored_files)
+        self._jobs[job_id] = record
+        return self._server.json_response(job.dict())
+
+    def _handle_list_jobs(self, request: httpx.Request) -> httpx.Response:
+        return self._server.json_response(
+            [record.job.dict() for record in self._jobs.values()]
+        )
+
+    def _handle_get_job(self, request: httpx.Request) -> httpx.Response:
+        job_id = request.url.path.split("/")[-1]
+        record = self._jobs.get(job_id)
+        if not record:
+            return self._server.json_response(
+                {"detail": "Job not found"}, status_code=404
+            )
+        return self._server.json_response(record.job.dict())
+
+    def _handle_get_results(self, request: httpx.Request) -> httpx.Response:
+        job_id = request.url.path.split("/")[-2]
+        record = self._jobs.get(job_id)
+        if not record:
+            return self._server.json_response(
+                {"detail": "Results not found"}, status_code=404
+            )
+        return self._server.json_response(record.results.dict())
+
+    def _build_results(
+        self,
+        job_id: str,
+        stored_files: List[StoredFile],
+        rules: List[ClassifierRule],
+    ) -> ClassifyJobResults:
+        items: List[FileClassification] = []
+        for stored in stored_files:
+            seed = combined_seed(stored.sha256, job_id)
+            rule_index = seed % len(rules) if rules else 0
+            predicted_type = rules[rule_index].type if rules else "unlabeled"
+            confidence = 0.55 + (seed % 40) / 100
+            reasoning = (
+                f"Selected rule '{predicted_type}' using deterministic seed {seed}."
+            )
+            classification = FileClassification(
+                id=self._server.new_id("classification"),
+                file_id=stored.file.id,
+                classify_job_id=job_id,
+                created_at=utcnow(),
+                updated_at=utcnow(),
+                result=ClassificationResult(
+                    type=predicted_type,
+                    confidence=min(confidence, 0.95),
+                    reasoning=reasoning,
+                ),
+            )
+            items.append(classification)
+        return ClassifyJobResults(
+            items=items, next_page_token=None, total_size=len(items)
+        )

py/llama_cloud_services/testing_utils/extract.py

@@ -0,0 +1,673 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from typing import TYPE_CHECKING, Any, Dict, List, Optional
+
+import httpx
+from llama_cloud.types import (
+    ExtractAgent,
+    ExtractConfig,
+    ExtractJob,
+    ExtractRun,
+    ExtractState,
+    File as CloudFile,
+    PaginatedExtractRunsResponse,
+    StatusEnum,
+)
+
+from ._deterministic import (
+    combined_seed,
+    generate_data_from_schema,
+    hash_schema,
+    utcnow,
+)
+from ._deterministic import fingerprint_file
+from .files import FakeFilesNamespace, StoredFile
+from .matchers import RequestContext, RequestMatcher
+
+if TYPE_CHECKING:
+    from .server import FakeLlamaCloudServer
+
+
+@dataclass
+class ExtractRunStub:
+    matcher: Optional[RequestMatcher]
+    data: Optional[Any]
+    status: Optional[str]
+    metadata: Optional[Dict[str, Any]]
+    error: Optional[str]
+    job_status: Optional[str]
+    once: bool
+
+
+@dataclass
+class AgentRunStub:
+    agent_id: str
+    matcher: Optional[RequestMatcher]
+    job_status: Optional[str]
+    run_status: Optional[str]
+    error: Optional[str]
+    once: bool
+
+
+@dataclass
+class StoredRun:
+    job: ExtractJob
+    run: ExtractRun
+
+
+class FakeExtractNamespace:
+    def __init__(
+        self,
+        *,
+        server: "FakeLlamaCloudServer",
+        files: FakeFilesNamespace,
+    ) -> None:
+        self._server = server
+        self._files = files
+        self._jobs: Dict[str, StoredRun] = {}
+        self._runs: Dict[str, ExtractRun] = {}
+        self._agents: Dict[str, ExtractAgent] = {}
+        self._agents_by_name: Dict[str, str] = {}
+        self._run_stubs: List[ExtractRunStub] = []
+        self._agent_run_stubs: List[AgentRunStub] = []
+        self.routes: Dict[str, Any] = {}
+
+    # Public APIs ----------------------------------------------------
+    def stub_run(
+        self,
+        matcher: Optional[RequestMatcher],
+        *,
+        data: Optional[Any] = None,
+        status: Optional[str] = None,
+        job_status: Optional[str] = None,
+        metadata: Optional[Dict[str, Any]] = None,
+        error: Optional[str] = None,
+        once: bool = True,
+    ) -> None:
+        self._run_stubs.append(
+            ExtractRunStub(
+                matcher=matcher,
+                data=data,
+                status=status,
+                metadata=metadata,
+                error=error,
+                job_status=job_status,
+                once=once,
+            )
+        )
+
+    def stub_agent_run(
+        self,
+        *,
+        agent_id: str,
+        matcher: Optional[RequestMatcher],
+        job_status: Optional[str] = None,
+        run_status: Optional[str] = None,
+        error: Optional[str] = None,
+        once: bool = True,
+    ) -> None:
+        self._agent_run_stubs.append(
+            AgentRunStub(
+                agent_id=agent_id,
+                matcher=matcher,
+                job_status=job_status,
+                run_status=run_status,
+                error=error,
+                once=once,
+            )
+        )
+
+    # Route registration ---------------------------------------------
+    def register(self) -> None:
+        server = self._server
+        route = server.add_route(
+            "POST",
+            "/api/v1/extraction/run",
+            self._handle_stateless_run,
+            namespace="extract",
+            alias="extract_run",
+        )
+        self.routes["stateless_run"] = route
+        self.stateless_run = route
+        server.add_route(
+            "POST",
+            "/api/v1/extraction/extraction-agents",
+            self._handle_create_agent,
+            namespace="extract",
+        )
+        server.add_route(
+            "PATCH",
+            "/api/v1/extraction/extraction-agents/{agent_id}",
+            self._handle_update_agent,
+            namespace="extract",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/extraction/extraction-agents/{agent_id}",
+            self._handle_get_agent,
+            namespace="extract",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/extraction/extraction-agents/by-name/{name}",
+            self._handle_get_agent_by_name,
+            namespace="extract",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/extraction/extraction-agents",
+            self._handle_list_agents,
+            namespace="extract",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/extraction/extraction-agents/default",
+            self._handle_get_default_agent,
+            namespace="extract",
+        )
+        server.add_route(
+            "DELETE",
+            "/api/v1/extraction/extraction-agents/{agent_id}",
+            self._handle_delete_agent,
+            namespace="extract",
+        )
+        server.add_route(
+            "POST",
+            "/api/v1/extraction/extraction-agents/schema/validation",
+            self._handle_validate_schema,
+            namespace="extract",
+        )
+        agent_job_route = server.add_route(
+            "POST",
+            "/api/v1/extraction/jobs",
+            self._handle_agent_job,
+            namespace="extract",
+            alias="agent_job",
+        )
+        self.routes["agent_job"] = agent_job_route
+        self.agent_job = agent_job_route
+        server.add_route(
+            "POST",
+            "/api/v1/extraction/jobs/batch",
+            self._handle_agent_job_batch,
+            namespace="extract",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/extraction/jobs",
+            self._handle_list_jobs,
+            namespace="extract",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/extraction/jobs/{job_id}",
+            self._handle_get_job,
+            namespace="extract",
+        )
+        agent_run_route = server.add_route(
+            "GET",
+            "/api/v1/extraction/runs/by-job/{job_id}",
+            self._handle_get_run_by_job,
+            namespace="extract",
+            alias="agent_run",
+        )
+        self.routes["agent_run"] = agent_run_route
+        self.agent_run = agent_run_route
+        server.add_route(
+            "GET",
+            "/api/v1/extraction/runs/{run_id}",
+            self._handle_get_run,
+            namespace="extract",
+        )
+        server.add_route(
+            "DELETE",
+            "/api/v1/extraction/runs/{run_id}",
+            self._handle_delete_run,
+            namespace="extract",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/extraction/runs",
+            self._handle_list_runs,
+            namespace="extract",
+        )
+
+    # Handlers -------------------------------------------------------
+    def _handle_stateless_run(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request)
+        config = ExtractConfig.parse_obj(payload["config"])
+        data_schema = payload["data_schema"]
+        schema_hash = hash_schema(data_schema)
+
+        file_info = self._extract_file_info(payload, request)
+        agent = self._build_ephemeral_agent(
+            config, data_schema, file_info.file.project_id
+        )
+
+        context = RequestContext(
+            request=request,
+            json=payload,
+            file_id=file_info.file.id,
+            filename=file_info.file.name,
+            file_sha256=file_info.sha256,
+            schema_hash=schema_hash,
+            project_id=file_info.file.project_id,
+            organization_id=self._server.default_organization_id,
+        )
+
+        stub = self._pop_stub(self._run_stubs, context)
+        job_status = StatusEnum.SUCCESS
+        run_status = ExtractState.SUCCESS
+        metadata = {"deterministic": {"value": True}}
+        error = None
+        run_data = self._generate_run_data(data_schema, file_info.sha256)
+
+        if stub:
+            if stub.job_status:
+                job_status = StatusEnum(stub.job_status)
+            if stub.status:
+                run_status = ExtractState(stub.status)
+            if stub.metadata:
+                metadata = stub.metadata
+            if stub.error:
+                error = stub.error
+            if stub.data is not None:
+                if callable(stub.data):
+                    run_data = stub.data(payload)  # type: ignore[assignment]
+                else:
+                    run_data = stub.data
+
+        stored = self._create_job_and_run(
+            agent=agent,
+            config=config,
+            data_schema=data_schema,
+            file_info=file_info,
+            job_status=job_status,
+            run_status=run_status,
+            metadata=metadata,
+            data=run_data,
+            error=error,
+            project_id=file_info.file.project_id,
+        )
+        return self._server.json_response(stored.job.dict())
+
+    def _handle_create_agent(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request)
+        name = payload["name"]
+        config = ExtractConfig.parse_obj(payload["config"])
+        data_schema = payload["data_schema"]
+        agent_id = self._server.new_id("agent")
+        agent = ExtractAgent(
+            id=agent_id,
+            name=name,
+            config=config,
+            data_schema=data_schema,
+            project_id=request.url.params.get(
+                "project_id", self._server.default_project_id
+            ),
+            created_at=utcnow(),
+            updated_at=utcnow(),
+            custom_configuration=None,
+        )
+        self._agents[agent_id] = agent
+        self._agents_by_name[name] = agent_id
+        return self._server.json_response(agent.dict())
+
+    def _handle_update_agent(self, request: httpx.Request) -> httpx.Response:
+        agent_id = request.url.path.split("/")[-1]
+        if agent_id not in self._agents:
+            return self._server.json_response(
+                {"detail": "Agent not found"}, status_code=404
+            )
+        payload = self._server.json(request)
+        agent = self._agents[agent_id]
+        config = payload.get("config", agent.config)
+        data_schema = payload.get("data_schema", agent.data_schema)
+        updated = agent.copy(
+            update={
+                "config": ExtractConfig.parse_obj(config)
+                if isinstance(config, dict)
+                else config,
+                "data_schema": data_schema,
+                "updated_at": utcnow(),
+            }
+        )
+        self._agents[agent_id] = updated
+        return self._server.json_response(updated.dict())
+
+    def _handle_get_agent(self, request: httpx.Request) -> httpx.Response:
+        agent_id = request.url.path.split("/")[-1]
+        agent = self._agents.get(agent_id)
+        if not agent:
+            return self._server.json_response(
+                {"detail": "Agent not found"}, status_code=404
+            )
+        return self._server.json_response(agent.dict())
+
+    def _handle_get_agent_by_name(self, request: httpx.Request) -> httpx.Response:
+        name = request.url.path.split("/")[-1]
+        agent_id = self._agents_by_name.get(name)
+        if not agent_id:
+            return self._server.json_response(
+                {"detail": "Agent not found"}, status_code=404
+            )
+        return self._server.json_response(self._agents[agent_id].dict())
+
+    def _handle_list_agents(self, request: httpx.Request) -> httpx.Response:
+        include_default = (
+            request.url.params.get("include_default", "false").lower() == "true"
+        )
+        agents = list(self._agents.values())
+        if include_default and not agents:
+            default_agent = self._build_ephemeral_agent(
+                ExtractConfig(),
+                {"type": "object", "properties": {}},
+                self._server.default_project_id,
+            )
+            agents.append(default_agent)
+        return self._server.json_response([agent.dict() for agent in agents])
+
+    def _handle_get_default_agent(self, request: httpx.Request) -> httpx.Response:
+        if self._agents:
+            agent = next(iter(self._agents.values()))
+        else:
+            agent = self._build_ephemeral_agent(
+                ExtractConfig(),
+                {"type": "object", "properties": {}},
+                self._server.default_project_id,
+            )
+        return self._server.json_response(agent.dict())
+
+    def _handle_delete_agent(self, request: httpx.Request) -> httpx.Response:
+        agent_id = request.url.path.split("/")[-1]
+        agent = self._agents.pop(agent_id, None)
+        if agent:
+            self._agents_by_name.pop(agent.name, None)
+        return self._server.json_response({}, status_code=200)
+
+    def _handle_validate_schema(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request)
+        return self._server.json_response({"data_schema": payload["data_schema"]})
+
+    def _handle_agent_job(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request)
+        agent_id = payload["extraction_agent_id"]
+        agent = self._agents.get(agent_id)
+        if not agent:
+            return self._server.json_response(
+                {"detail": "Agent not found"}, status_code=404
+            )
+
+        file_id = payload["file_id"]
+        stored_file = self._files._files.get(file_id)
+        if not stored_file:
+            return self._server.json_response(
+                {"detail": "File not found"}, status_code=404
+            )
+
+        schema = payload.get("data_schema_override", agent.data_schema)
+        config_payload = payload.get("config_override", agent.config)
+        config = (
+            ExtractConfig.parse_obj(config_payload)
+            if isinstance(config_payload, dict)
+            else config_payload
+        )
+
+        stub = self._pop_agent_stub(
+            agent_id, RequestContext(request=request, json=payload)
+        )
+        job_status = StatusEnum.SUCCESS
+        run_status = ExtractState.SUCCESS
+        error = None
+        if stub:
+            if stub.job_status:
+                job_status = StatusEnum(stub.job_status)
+            if stub.run_status:
+                run_status = ExtractState(stub.run_status)
+            if stub.error:
+                error = stub.error
+
+        stored = self._create_job_and_run(
+            agent=agent,
+            config=config,
+            data_schema=schema,
+            file_info=stored_file,
+            job_status=job_status,
+            run_status=run_status,
+            metadata={"agent": {"value": agent.id}},
+            data=self._generate_run_data(schema, stored_file.sha256),
+            error=error,
+            project_id=agent.project_id,
+        )
+        return self._server.json_response(stored.job.dict())
+
+    def _handle_agent_job_batch(self, request: httpx.Request) -> httpx.Response:
+        payload = self._server.json(request)
+        file_ids = payload.get("file_ids", [])
+        jobs = []
+        for file_id in file_ids:
+            request_body = payload.copy()
+            request_body["file_id"] = file_id
+            fake_request = request.copy()
+            fake_request._content = self._server.encode_json(request_body)
+            response = self._handle_agent_job(fake_request)
+            if response.status_code != 200:
+                return response
+            jobs.append(response.json())
+        return self._server.json_response(jobs)
+
+    def _handle_list_jobs(self, request: httpx.Request) -> httpx.Response:
+        agent_id = request.url.params.get("extraction_agent_id")
+        items = []
+        for stored in self._jobs.values():
+            if agent_id and stored.job.extraction_agent.id != agent_id:
+                continue
+            items.append(stored.job.dict())
+        return self._server.json_response(items)
+
+    def _handle_get_job(self, request: httpx.Request) -> httpx.Response:
+        job_id = request.url.path.split("/")[-1]
+        stored = self._jobs.get(job_id)
+        if not stored:
+            return self._server.json_response(
+                {"detail": "Job not found"}, status_code=404
+            )
+        return self._server.json_response(stored.job.dict())
+
+    def _handle_get_run_by_job(self, request: httpx.Request) -> httpx.Response:
+        job_id = request.url.path.split("/")[-1]
+        stored = self._jobs.get(job_id)
+        if not stored:
+            return self._server.json_response(
+                {"detail": "Run not found"}, status_code=404
+            )
+        return self._server.json_response(stored.run.dict())
+
+    def _handle_get_run(self, request: httpx.Request) -> httpx.Response:
+        run_id = request.url.path.split("/")[-1]
+        run = self._runs.get(run_id)
+        if not run:
+            return self._server.json_response(
+                {"detail": "Run not found"}, status_code=404
+            )
+        return self._server.json_response(run.dict())
+
+    def _handle_delete_run(self, request: httpx.Request) -> httpx.Response:
+        run_id = request.url.path.split("/")[-1]
+        self._runs.pop(run_id, None)
+        to_delete = [
+            job_id for job_id, stored in self._jobs.items() if stored.run.id == run_id
+        ]
+        for job_id in to_delete:
+            self._jobs.pop(job_id, None)
+        return self._server.json_response({}, status_code=200)
+
+    def _handle_list_runs(self, request: httpx.Request) -> httpx.Response:
+        agent_id = request.url.params.get("extraction_agent_id")
+        skip = int(request.url.params.get("skip", "0"))
+        limit = int(request.url.params.get("limit", "50"))
+        filtered = [
+            stored.run
+            for stored in self._jobs.values()
+            if not agent_id or stored.job.extraction_agent.id == agent_id
+        ]
+        page = filtered[skip : skip + limit]
+        response = PaginatedExtractRunsResponse(
+            items=page,
+            skip=skip,
+            limit=limit,
+            total=len(filtered),
+        )
+        return self._server.json_response(response.dict())
+
+    # Internal helpers -----------------------------------------------
+    def _extract_file_info(
+        self, payload: Dict[str, Any], request: httpx.Request
+    ) -> StoredFile:
+        if "file_id" in payload:
+            file_id = payload["file_id"]
+            stored = self._files.get(file_id)
+            if not stored:
+                raise ValueError("file_id not found in fake store")
+            return stored
+        if "file" in payload:
+            content, filename = self._files.decode_file_data(payload)
+            file_id = self._server.new_id("file")
+            stored = StoredFile(
+                file=CloudFile(
+                    id=file_id,
+                    name=filename or f"inline-{file_id}",
+                    project_id=request.url.params.get(
+                        "project_id", self._server.default_project_id
+                    ),
+                    external_file_id=None,
+                    file_size=len(content),
+                    file_type=None,
+                    created_at=utcnow(),
+                    updated_at=utcnow(),
+                    data_source_id=None,
+                    permission_info=None,
+                    resource_info=None,
+                    last_modified_at=utcnow(),
+                ),
+                content=content,
+                sha256=fingerprint_file(content, filename),
+            )
+            return stored
+        if "text" in payload:
+            text_bytes = payload["text"].encode("utf-8")
+            file_id = self._server.new_id("file")
+            stored = StoredFile(
+                file=CloudFile(
+                    id=file_id,
+                    name=f"text-{file_id}.txt",
+                    project_id=self._server.default_project_id,
+                    external_file_id=None,
+                    file_size=len(text_bytes),
+                    file_type="text/plain",
+                    created_at=utcnow(),
+                    updated_at=utcnow(),
+                    data_source_id=None,
+                    permission_info=None,
+                    resource_info=None,
+                    last_modified_at=utcnow(),
+                ),
+                content=text_bytes,
+                sha256=fingerprint_file(text_bytes, None),
+            )
+            return stored
+        raise ValueError("file payload missing")
+
+    def _build_ephemeral_agent(
+        self,
+        config: ExtractConfig,
+        data_schema: Dict[str, Any],
+        project_id: str,
+    ) -> ExtractAgent:
+        return ExtractAgent(
+            id=self._server.new_id("agent"),
+            name="stateless-agent",
+            config=config,
+            data_schema=data_schema,
+            project_id=project_id,
+            created_at=utcnow(),
+            updated_at=utcnow(),
+            custom_configuration=None,
+        )
+
+    def _generate_run_data(self, schema: Dict[str, Any], file_hash: str) -> Any:
+        seed = combined_seed(file_hash, hash_schema(schema))
+        return generate_data_from_schema(schema, seed)
+
+    def _create_job_and_run(
+        self,
+        *,
+        agent: ExtractAgent,
+        config: ExtractConfig,
+        data_schema: Dict[str, Any],
+        file_info: StoredFile,
+        job_status: StatusEnum,
+        run_status: ExtractState,
+        metadata: Dict[str, Any],
+        data: Any,
+        error: Optional[str],
+        project_id: str,
+    ) -> StoredRun:
+        job_id = self._server.new_id("job")
+        run_id = self._server.new_id("run")
+        now = utcnow()
+
+        job = ExtractJob(
+            id=job_id,
+            file=file_info.file,
+            extraction_agent=agent,
+            status=job_status,
+            error=error,
+        )
+        run = ExtractRun(
+            id=run_id,
+            job_id=job_id,
+            file=file_info.file,
+            extraction_agent_id=agent.id,
+            status=run_status,
+            config=config,
+            data_schema=data_schema,
+            data=data,
+            extraction_metadata=metadata,
+            created_at=now,
+            updated_at=now,
+            from_ui=False,
+            error=error,
+            project_id=project_id,
+        )
+        stored = StoredRun(job=job, run=run)
+        self._jobs[job_id] = stored
+        self._runs[run_id] = run
+        return stored
+
+    def _pop_stub(
+        self,
+        stubs: List[ExtractRunStub],
+        context: RequestContext,
+    ) -> Optional[ExtractRunStub]:
+        for index, stub in enumerate(list(stubs)):
+            if context.matches(stub.matcher):
+                if stub.once:
+                    stubs.pop(index)
+                return stub
+        return None
+
+    def _pop_agent_stub(
+        self,
+        agent_id: str,
+        context: RequestContext,
+    ) -> Optional[AgentRunStub]:
+        for index, stub in enumerate(list(self._agent_run_stubs)):
+            if stub.agent_id != agent_id:
+                continue
+            if context.matches(stub.matcher):
+                if stub.once:
+                    self._agent_run_stubs.pop(index)
+                return stub
+        return None

py/llama_cloud_services/testing_utils/files.py

@@ -0,0 +1,324 @@
+from __future__ import annotations
+
+import base64
+from dataclasses import dataclass
+from pathlib import Path
+from typing import TYPE_CHECKING, Any, Dict, List, Optional
+from urllib.parse import urlencode
+
+import httpx
+import respx
+from llama_cloud.types import File as CloudFile
+from llama_cloud.types import FileIdPresignedUrl, PresignedUrl
+
+from ._deterministic import fingerprint_file, hash_chunks, utcnow
+from .matchers import RequestContext, RequestMatcher
+
+if TYPE_CHECKING:
+    from .server import FakeLlamaCloudServer
+
+
+@dataclass
+class StoredFile:
+    file: CloudFile
+    content: bytes
+    sha256: str
+
+
+@dataclass
+class PendingUpload:
+    file_id: str
+    filename: str
+    project_id: str
+    organization_id: str
+    external_file_id: Optional[str]
+    expected_size: Optional[int]
+
+
+class FakeFilesNamespace:
+    def __init__(
+        self,
+        *,
+        server: "FakeLlamaCloudServer",
+        upload_base_url: str,
+        download_base_url: str,
+    ) -> None:
+        self._server = server
+        self._upload_base_url = upload_base_url.rstrip("/")
+        self._download_base_url = download_base_url.rstrip("/")
+        self._files: Dict[str, StoredFile] = {}
+        self._pending: Dict[str, PendingUpload] = {}
+        self._upload_stubs: List[
+            tuple[RequestMatcher | None, int, Dict[str, Any], bool]
+        ] = []
+        self.routes: Dict[str, respx.Route] = {}
+
+    # Public helpers -------------------------------------------------
+    def preload(self, *, path: str | Path, filename: Optional[str] = None) -> str:
+        path = Path(path)
+        content = path.read_bytes()
+        file_id = self._server.new_id("file")
+        name = filename or path.name
+        stored = self._build_file(
+            file_id=file_id,
+            name=name,
+            project_id=self._server.default_project_id,
+            organization_id=self._server.default_organization_id,
+            content=content,
+            external_file_id=None,
+        )
+        self._files[file_id] = stored
+        return file_id
+
+    def read(self, file_id: str) -> bytes:
+        return self._files[file_id].content
+
+    def get(self, file_id: str) -> Optional[StoredFile]:
+        return self._files.get(file_id)
+
+    def stub_upload(
+        self,
+        matcher: Optional[RequestMatcher],
+        *,
+        status_code: int = 413,
+        json_body: Optional[Dict[str, Any]] = None,
+        once: bool = True,
+    ) -> None:
+        body = json_body or {"detail": "upload rejected by fake server"}
+        self._upload_stubs.append((matcher, status_code, body, once))
+
+    # Route registration ---------------------------------------------
+    def register(self) -> None:
+        server = self._server
+        server.add_route(
+            "PUT",
+            "/api/v1/files",
+            self._handle_generate_presigned_url,
+            namespace="files",
+            alias="generate_presigned_url",
+        )
+        upload_route = server.add_route(
+            "POST",
+            "/api/v1/files",
+            self._handle_direct_upload,
+            namespace="files",
+            alias="upload",
+        )
+        self.routes["upload"] = upload_route
+        get_route = server.add_route(
+            "GET",
+            "/api/v1/files/{file_id}",
+            self._handle_get_metadata,
+            namespace="files",
+            alias="get",
+        )
+        self.routes["get"] = get_route
+        server.add_route(
+            "DELETE",
+            "/api/v1/files/{file_id}",
+            self._handle_delete,
+            namespace="files",
+        )
+        server.add_route(
+            "GET",
+            "/api/v1/files/{file_id}/content",
+            self._handle_read_content,
+            namespace="files",
+            alias="read_content",
+        )
+        server.add_route(
+            "PUT",
+            "/upload/{file_id}",
+            self._handle_presigned_upload,
+            namespace="files",
+            base_urls=[self._upload_base_url],
+            alias="presigned_upload",
+        )
+        server.add_route(
+            "GET",
+            "/files/{file_id}",
+            self._handle_presigned_download,
+            namespace="files",
+            base_urls=[self._download_base_url],
+            alias="download",
+        )
+
+    # Handlers -------------------------------------------------------
+    def _handle_generate_presigned_url(self, request: httpx.Request) -> httpx.Response:
+        data = self._server.json(request)
+        now = utcnow()
+        file_id = self._server.new_id("file")
+        name = data.get("name") or f"upload-{file_id}.bin"
+        pending = PendingUpload(
+            file_id=file_id,
+            filename=name,
+            project_id=request.url.params.get(
+                "project_id", self._server.default_project_id
+            ),
+            organization_id=request.url.params.get(
+                "organization_id", self._server.default_organization_id
+            ),
+            external_file_id=data.get("external_file_id"),
+            expected_size=data.get("file_size"),
+        )
+        self._pending[file_id] = pending
+        presigned = FileIdPresignedUrl(
+            file_id=file_id,
+            url=f"{self._upload_base_url}/upload/{file_id}",
+            expires_at=now,
+            form_fields=None,
+        )
+        return self._server.json_response(presigned.dict())
+
+    def _handle_direct_upload(self, request: httpx.Request) -> httpx.Response:
+        file_bytes, filename = self._extract_multipart_file(request)
+        file_id = self._server.new_id("file")
+        stored = self._build_file(
+            file_id=file_id,
+            name=filename or f"upload-{file_id}.bin",
+            project_id=request.url.params.get(
+                "project_id", self._server.default_project_id
+            ),
+            organization_id=request.url.params.get(
+                "organization_id", self._server.default_organization_id
+            ),
+            content=file_bytes,
+            external_file_id=request.url.params.get("external_file_id"),
+        )
+        self._files[file_id] = stored
+        return self._server.json_response(stored.file.dict())
+
+    def _handle_get_metadata(self, request: httpx.Request) -> httpx.Response:
+        file_id = request.url.path.split("/")[-1]
+        if file_id not in self._files:
+            return self._server.json_response(
+                {"detail": "File not found"}, status_code=404
+            )
+        return self._server.json_response(self._files[file_id].file.dict())
+
+    def _handle_delete(self, request: httpx.Request) -> httpx.Response:
+        file_id = request.url.path.split("/")[-1]
+        self._files.pop(file_id, None)
+        self._pending.pop(file_id, None)
+        return self._server.json_response({}, status_code=200)
+
+    def _handle_read_content(self, request: httpx.Request) -> httpx.Response:
+        file_id = request.url.path.split("/")[-2]
+        if file_id not in self._files:
+            return self._server.json_response(
+                {"detail": "File not found"}, status_code=404
+            )
+        presigned = PresignedUrl(
+            url=f"{self._download_base_url}/files/{file_id}?{urlencode({'token': 'fake'})}",
+            expires_at=utcnow(),
+            form_fields=None,
+        )
+        return self._server.json_response(presigned.dict())
+
+    def _handle_presigned_upload(self, request: httpx.Request) -> httpx.Response:
+        file_id = request.url.path.split("/")[-1]
+        pending = self._pending.get(file_id)
+
+        context = RequestContext(
+            request=request,
+            json=None,
+            file_id=file_id,
+            filename=pending.filename if pending else None,
+            file_sha256=hash_chunks([request.content]),
+        )
+
+        for index, (matcher, status, body, once) in enumerate(list(self._upload_stubs)):
+            if context.matches(matcher):
+                if once:
+                    self._upload_stubs.pop(index)
+                return self._server.json_response(body, status_code=status)
+
+        if pending is None:
+            return self._server.json_response(
+                {"detail": "Unknown file"}, status_code=404
+            )
+
+        stored = self._build_file(
+            file_id=file_id,
+            name=pending.filename,
+            project_id=pending.project_id,
+            organization_id=pending.organization_id,
+            content=request.content,
+            external_file_id=pending.external_file_id,
+        )
+        self._files[file_id] = stored
+        self._pending.pop(file_id, None)
+        return httpx.Response(204)
+
+    def _handle_presigned_download(self, request: httpx.Request) -> httpx.Response:
+        file_id = request.url.path.split("/")[-1]
+        stored = self._files.get(file_id)
+        if not stored:
+            return httpx.Response(404, json={"detail": "File not found"})
+        return httpx.Response(200, content=stored.content)
+
+    # Internal helpers -----------------------------------------------
+    def _build_file(
+        self,
+        *,
+        file_id: str,
+        name: str,
+        project_id: str,
+        organization_id: str,
+        content: bytes,
+        external_file_id: Optional[str],
+    ) -> StoredFile:
+        sha256 = fingerprint_file(content, name)
+        now = utcnow()
+        cloud_file = CloudFile(
+            id=file_id,
+            name=name,
+            project_id=project_id,
+            external_file_id=external_file_id,
+            file_size=len(content),
+            file_type=Path(name).suffix or "application/octet-stream",
+            created_at=now,
+            updated_at=now,
+            data_source_id=None,
+            permission_info=None,
+            resource_info=None,
+            last_modified_at=now,
+        )
+        return StoredFile(file=cloud_file, content=content, sha256=sha256)
+
+    def _extract_multipart_file(
+        self, request: httpx.Request
+    ) -> tuple[bytes, Optional[str]]:
+        content_type = request.headers.get("content-type", "")
+        if "multipart/form-data" not in content_type:
+            raise ValueError("Expected multipart upload")
+
+        boundary = content_type.split("boundary=")[-1]
+        boundary_bytes = boundary.encode("utf-8")
+        body = request.content
+        delimiter = b"--" + boundary_bytes
+        parts = [
+            part
+            for part in body.split(delimiter)
+            if part.strip(b"\r\n") and part.strip(b"\r\n") != b"--"
+        ]
+        for part in parts:
+            headers, _, payload = part.partition(b"\r\n\r\n")
+            header_text = headers.decode("utf-8", errors="ignore")
+            if 'name="upload_file"' in header_text or 'name="file"' in header_text:
+                filename = None
+                if "filename=" in header_text:
+                    filename = (
+                        header_text.split("filename=")[-1].strip().strip('"').strip("'")
+                    )
+                return payload.rstrip(b"\r\n"), filename
+        raise ValueError("upload file part not found")
+
+    def decode_file_data(self, data: Dict[str, Any]) -> tuple[bytes, Optional[str]]:
+        if "file" not in data:
+            raise ValueError("file payload missing")
+        file_payload = data["file"]
+        encoded = file_payload["data"]
+        content = base64.b64decode(encoded)
+        filename = file_payload.get("filename")
+        return content, filename

py/llama_cloud_services/testing_utils/matchers.py

@@ -0,0 +1,100 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from typing import Any, Callable, Optional
+
+import httpx
+
+
+MatcherPredicate = Callable[[httpx.Request], bool]
+
+
+@dataclass
+class FileMatcher:
+    filename: Optional[str] = None
+    sha256: Optional[str] = None
+    file_id: Optional[str] = None
+
+
+@dataclass
+class SchemaMatcher:
+    model: Optional[type] = None
+    schema_hash: Optional[str] = None
+
+
+@dataclass
+class RequestMatcher:
+    file: Optional[FileMatcher | MatcherPredicate] = None
+    schema: Optional[SchemaMatcher] = None
+    agent_id: Optional[str] = None
+    project_id: Optional[str] = None
+    organization_id: Optional[str] = None
+    predicate: Optional[MatcherPredicate] = None
+
+
+@dataclass
+class RequestContext:
+    request: httpx.Request
+    json: Optional[dict[str, Any]]
+    file_id: Optional[str] = None
+    filename: Optional[str] = None
+    file_sha256: Optional[str] = None
+    schema_hash: Optional[str] = None
+    agent_id: Optional[str] = None
+    project_id: Optional[str] = None
+    organization_id: Optional[str] = None
+
+    def matches(self, matcher: Optional[RequestMatcher]) -> bool:
+        if matcher is None:
+            return True
+
+        if matcher.project_id and matcher.project_id != self.project_id:
+            return False
+
+        if matcher.organization_id and matcher.organization_id != self.organization_id:
+            return False
+
+        if matcher.agent_id and matcher.agent_id != self.agent_id:
+            return False
+
+        if matcher.file:
+            if isinstance(matcher.file, FileMatcher):
+                if matcher.file.filename and matcher.file.filename != self.filename:
+                    return False
+                if matcher.file.file_id and matcher.file.file_id != self.file_id:
+                    return False
+                if matcher.file.sha256 and matcher.file.sha256 != self.file_sha256:
+                    return False
+            else:
+                if not matcher.file(self.request):
+                    return False
+
+        if matcher.schema:
+            if (
+                matcher.schema.schema_hash
+                and matcher.schema.schema_hash != self.schema_hash
+            ):
+                return False
+            if matcher.schema.model and matcher.schema.schema_hash:
+                return matcher.schema.schema_hash == self.schema_hash
+            if matcher.schema.model and matcher.schema.schema_hash is None:
+                expected = _schema_hash_from_model(matcher.schema.model)
+                return expected == self.schema_hash
+
+        if matcher.predicate and not matcher.predicate(self.request):
+            return False
+
+        return True
+
+
+def _schema_hash_from_model(model: type) -> Optional[str]:
+    if hasattr(model, "model_json_schema"):
+        schema = model.model_json_schema()
+    elif hasattr(model, "schema"):
+        schema = model.schema()  # type: ignore[attr-defined]
+    else:
+        return None
+
+    from ._deterministic import hash_schema
+
+    return hash_schema(schema)

py/llama_cloud_services/testing_utils/parse.py

@@ -0,0 +1,153 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from typing import TYPE_CHECKING, Any, Dict
+
+import httpx
+
+from ._deterministic import generate_text_blob, hash_schema
+
+if TYPE_CHECKING:
+    from .server import FakeLlamaCloudServer
+
+
+@dataclass
+class ParseJobRecord:
+    job_id: str
+    file_name: str
+    status: str
+    result: Dict[str, Any]
+    content: bytes
+
+
+class FakeParseNamespace:
+    def __init__(self, *, server: "FakeLlamaCloudServer") -> None:
+        self._server = server
+        self._jobs: Dict[str, ParseJobRecord] = {}
+        self.routes: Dict[str, Any] = {}
+
+    def register(self) -> None:
+        server = self._server
+        server.add_route(
+            "POST",
+            "/api/parsing/upload",
+            self._handle_upload,
+            namespace="parse",
+        )
+        server.add_route(
+            "GET",
+            "/api/parsing/job/{job_id}",
+            self._handle_job_status,
+            namespace="parse",
+        )
+        server.add_route(
+            "GET",
+            "/api/parsing/job/{job_id}/result/{result_type}",
+            self._handle_job_result,
+            namespace="parse",
+        )
+
+    def _handle_upload(self, request: httpx.Request) -> httpx.Response:
+        file_bytes, filename, form_data = self._split_multipart(request)
+        job_id = self._server.new_id("parse-job")
+        seed_hash = hash_schema({"filename": filename, "form": form_data})
+        seed = int(seed_hash[:16], 16)
+        page_text = generate_text_blob(seed, sentences=3)
+        pages: list[Dict[str, Any]] = [
+            {
+                "page": index + 1,
+                "text": f"{page_text} (page {index + 1})",
+                "md": f"{page_text} (page {index + 1})",
+                "images": [],
+                "charts": [],
+                "tables": [],
+                "layout": [],
+                "items": [],
+                "status": "SUCCESS",
+                "links": [],
+                "width": 8.5,
+                "height": 11.0,
+                "parsingMode": "deterministic",
+                "structuredData": {},
+                "noStructuredContent": False,
+                "noTextContent": False,
+                "isAudioTranscript": False,
+                "durationInSeconds": None,
+                "slideSpeakerNotes": None,
+            }
+            for index in range(1)
+        ]
+        result = {
+            "job_id": job_id,
+            "status": "SUCCESS",
+            "file_name": filename,
+            "is_done": True,
+            "pages": pages,
+            "job_metadata": {"job_pages": len(pages)},
+            "text": "\n\n".join(str(page["text"]) for page in pages),
+            "markdown": "\n\n".join(str(page["md"]) for page in pages),
+            "json": {"pages": pages},
+        }
+        record = ParseJobRecord(
+            job_id=job_id,
+            file_name=filename,
+            status="SUCCESS",
+            result=result,
+            content=file_bytes,
+        )
+        self._jobs[job_id] = record
+        return self._server.json_response({"id": job_id})
+
+    def _handle_job_status(self, request: httpx.Request) -> httpx.Response:
+        job_id = request.url.path.split("/")[-1]
+        job = self._jobs.get(job_id)
+        if not job:
+            return self._server.json_response(
+                {"detail": "Job not found"}, status_code=404
+            )
+        return self._server.json_response({"id": job_id, "status": job.status})
+
+    def _handle_job_result(self, request: httpx.Request) -> httpx.Response:
+        job_id = request.url.path.split("/")[-3]
+        job = self._jobs.get(job_id)
+        if not job:
+            return self._server.json_response(
+                {"detail": "Result not found"}, status_code=404
+            )
+        return self._server.json_response(job.result)
+
+    def _split_multipart(
+        self, request: httpx.Request
+    ) -> tuple[bytes, str, Dict[str, str]]:
+        content_type = request.headers.get("content-type", "")
+        if "multipart/form-data" not in content_type:
+            raise ValueError("Expected multipart form data for parse upload")
+        boundary = content_type.split("boundary=")[-1]
+        delimiter = f"--{boundary}".encode()
+        closing = f"--{boundary}--".encode()
+        parts = []
+        body = request.content
+        for chunk in body.split(delimiter):
+            chunk = chunk.strip()
+            if not chunk or chunk == closing:
+                continue
+            parts.append(chunk)
+
+        file_bytes = b""
+        filename = "upload.pdf"
+        form_data: Dict[str, str] = {}
+        for part in parts:
+            header_blob, _, payload = part.partition(b"\r\n\r\n")
+            payload = payload.rstrip(b"\r\n")
+            header_text = header_blob.decode("utf-8", errors="ignore")
+            if "filename=" in header_text:
+                filename = (
+                    header_text.split("filename=")[-1].strip().strip('"').strip("'")
+                )
+                file_bytes = payload
+            else:
+                name = header_text.split('name="')[-1].split('"')[0].strip()
+                form_data[name] = payload.decode("utf-8", errors="ignore")
+        if not file_bytes:
+            raise ValueError("File part missing from multipart payload")
+        return file_bytes, filename, form_data

py/llama_cloud_services/testing_utils/server.py

@@ -0,0 +1,173 @@
+from __future__ import annotations
+
+import json
+import re
+import uuid
+from typing import Any, Callable, Dict, Optional, Sequence
+
+import httpx
+import respx
+
+from .classify import FakeClassifyNamespace
+from .extract import FakeExtractNamespace
+from .files import FakeFilesNamespace
+from .parse import FakeParseNamespace
+from .agent_data import FakeAgentDataNamespace
+
+Handler = Callable[[httpx.Request], httpx.Response]
+
+
+class FakeLlamaCloudServer:
+    DEFAULT_BASE_URL = "https://api.cloud.llamaindex.ai"
+    DEFAULT_UPLOAD_BASE = "https://uploads.fake-llama.test"
+    DEFAULT_DOWNLOAD_BASE = "https://downloads.fake-llama.test"
+
+    def __init__(
+        self,
+        *,
+        base_urls: Optional[Sequence[str]] = None,
+        namespaces: Optional[Sequence[str]] = None,
+        upload_base_url: Optional[str] = None,
+        download_base_url: Optional[str] = None,
+        default_project_id: str = "proj-test",
+        default_organization_id: str = "org-test",
+    ) -> None:
+        self.base_urls = tuple(base_urls or (self.DEFAULT_BASE_URL,))
+        selected = namespaces or ("files", "extract", "parse", "classify", "agent_data")
+        self._namespace_names = {name.lower() for name in selected}
+        self._upload_base_url = upload_base_url or self.DEFAULT_UPLOAD_BASE
+        self._download_base_url = download_base_url or self.DEFAULT_DOWNLOAD_BASE
+        self.default_project_id = default_project_id
+        self.default_organization_id = default_organization_id
+        self.router = respx.MockRouter(assert_all_called=False)
+        self._installed = False
+        self._registered = False
+
+        self.files = FakeFilesNamespace(
+            server=self,
+            upload_base_url=self._upload_base_url,
+            download_base_url=self._download_base_url,
+        )
+        self.extract = FakeExtractNamespace(server=self, files=self.files)
+        self.parse = FakeParseNamespace(server=self)
+        self.classify = FakeClassifyNamespace(server=self, files=self.files)
+        self.agent_data = FakeAgentDataNamespace(server=self)
+
+    # Context management ----------------------------------------------
+    def install(self) -> "FakeLlamaCloudServer":
+        if not self._registered:
+            self._register_namespaces()
+        if not self._installed:
+            self.router.__enter__()
+            self._installed = True
+        return self
+
+    def uninstall(self) -> None:
+        if self._installed:
+            self.router.__exit__(None, None, None)
+            self._installed = False
+
+    def __enter__(self) -> "FakeLlamaCloudServer":
+        return self.install()
+
+    def __exit__(self, exc_type: Any, exc: Any, tb: Any) -> None:
+        self.uninstall()
+
+    # Route utilities -------------------------------------------------
+    def add_route(
+        self,
+        method: str,
+        path: str,
+        handler: Handler,
+        *,
+        namespace: str,
+        alias: Optional[str] = None,
+        base_urls: Optional[Sequence[str]] = None,
+    ) -> respx.Route:
+        urls = base_urls or self.base_urls
+        first_route: Optional[respx.Route] = None
+        for base in urls:
+            route = self._register_route(method, base, path, handler)
+            if first_route is None:
+                first_route = route
+        if alias and first_route:
+            setattr(self, alias, first_route)
+        return first_route  # type: ignore[return-value]
+
+    def _register_route(
+        self,
+        method: str,
+        base: str,
+        path: str,
+        handler: Handler,
+    ) -> respx.Route:
+        url = self._build_url(base, path)
+        if "{" in path:
+            regex = self._compile_regex(base, path)
+            route = self.router.route(method=method, url__regex=regex)
+        else:
+            route = self.router.route(method=method, url=url)
+        route.mock(side_effect=lambda request, func=handler: func(request))
+        return route
+
+    def _build_url(self, base: str, path: str) -> str:
+        base = base.rstrip("/")
+        if not path.startswith("/"):
+            path = "/" + path
+        return f"{base}{path}"
+
+    def _compile_regex(self, base: str, path: str) -> re.Pattern[str]:
+        escaped = re.escape(base.rstrip("/"))
+        regex_path = re.sub(r"\{[^/]+\}", r"[^/]+", path)
+        pattern = f"^{escaped}{regex_path}$"
+        return re.compile(pattern)
+
+    # Helpers ---------------------------------------------------------
+    def json(self, request: httpx.Request) -> Dict[str, Any]:
+        if not request.content:
+            return {}
+        return json.loads(request.content.decode("utf-8"))
+
+    def encode_json(self, payload: Dict[str, Any]) -> bytes:
+        return json.dumps(payload).encode("utf-8")
+
+    def json_response(self, payload: Any, *, status_code: int = 200) -> httpx.Response:
+        body = json.dumps(payload, default=self._json_default).encode("utf-8")
+        headers = {"content-type": "application/json"}
+        return httpx.Response(status_code=status_code, headers=headers, content=body)
+
+    def new_id(self, prefix: str) -> str:
+        return f"{prefix}_{uuid.uuid4().hex[:8]}"
+
+    # Internal --------------------------------------------------------
+    def _json_default(self, value: Any) -> Any:
+        if hasattr(value, "model_dump"):
+            return value.model_dump()
+        if hasattr(value, "dict"):
+            return value.dict()
+        if isinstance(value, (set, frozenset)):
+            return list(value)
+        if isinstance(value, (bytes, bytearray)):
+            return value.decode("utf-8")
+        if hasattr(value, "isoformat"):
+            try:
+                return value.isoformat()  # datetime/date support
+            except Exception:
+                pass
+        raise TypeError(f"{value!r} is not JSON serializable")
+
+    def _register_namespaces(self) -> None:
+        if "files" in self._namespace_names:
+            self.files.register()
+        if "extract" in self._namespace_names:
+            self.extract.register()
+        if "parse" in self._namespace_names:
+            self.parse.register()
+        if "classify" in self._namespace_names:
+            self.classify.register()
+        if "agent_data" in self._namespace_names:
+            self.agent_data.register()
+        self._registered = True
+
+
+__all__ = ["FakeLlamaCloudServer"]

py/llama_cloud_services/utils.py

@@ -3,11 +3,14 @@ import importlib.metadata
 from contextlib import contextmanager
 from typing import Generator
 import difflib
-from llama_cloud.types import StatusEnum
+from llama_cloud.types import StatusEnum, File
 import httpx
 import packaging.version
 from pydantic import BaseModel
-from typing import Any, Dict, List, Tuple, Type
+from typing import Any, Dict, List, Tuple, Type, Union, Optional
+from io import BufferedIOBase, TextIOWrapper
+from pathlib import Path
+import secrets
 
 # Asyncio error messages
 nest_asyncio_err = "cannot be called from a running event loop"
@@ -104,3 +107,100 @@ def augment_async_errors() -> Generator[None, None, None]:
         if nest_asyncio_err in str(e):
             raise RuntimeError(nest_asyncio_msg)
         raise
+
+
+class SourceText:
+    """
+    A wrapper class for providing text or file input with optional filename specification.
+
+    This class allows you to provide input in multiple ways:
+    - Direct text content via text_content parameter
+    - File paths as strings or Path objects
+    - Raw bytes
+    - File-like objects (BufferedIOBase, TextIOWrapper)
+    - Already-uploaded file ID via file_id parameter
+
+    Args:
+        file: The file input (bytes, file-like object, str path, or Path).
+              Mutually exclusive with text_content and file_id.
+        text_content: Raw text content to process. Mutually exclusive with file and file_id.
+        file_id: ID of an already-uploaded file. Mutually exclusive with file and text_content.
+        filename: Optional filename. Required for bytes/file-like objects without names.
+                  If not provided, will be auto-generated for text_content or inferred from paths.
+
+    Examples:
+        # Direct text input
+        source = SourceText(text_content="Hello world")
+
+        # File path
+        source = SourceText(file="document.pdf")
+
+        # Bytes with filename
+        source = SourceText(file=b"...", filename="document.pdf")
+
+        # File-like object (will read from current position)
+        with open("document.pdf", "rb") as f:
+            source = SourceText(file=f)
+
+        # Already-uploaded file
+        source = SourceText(file_id="file_abc123")
+    """
+
+    def __init__(
+        self,
+        *,
+        file: Union[bytes, BufferedIOBase, TextIOWrapper, str, Path, None] = None,
+        text_content: Optional[str] = None,
+        file_id: Optional[str] = None,
+        filename: Optional[str] = None,
+    ):
+        self.file = file
+        self.filename = filename
+        self.text_content = text_content
+        self.file_id = file_id
+        self._validate()
+
+    def _validate(self) -> None:
+        """Ensure filename is provided when needed."""
+        # Check that exactly one of file, text_content, or file_id is provided
+        provided = sum(
+            [
+                self.file is not None,
+                self.text_content is not None,
+                self.file_id is not None,
+            ]
+        )
+
+        if provided == 0:
+            raise ValueError("One of file, text_content, or file_id must be provided.")
+        elif provided > 1:
+            raise ValueError(
+                "Only one of file, text_content, or file_id can be provided."
+            )
+
+        # If file_id is provided, we don't need filename validation
+        if self.file_id is not None:
+            return
+
+        if self.text_content is not None:
+            if not self.filename:
+                random_hex = secrets.token_hex(4)
+                self.filename = f"text_input_{random_hex}.txt"
+            return
+
+        if isinstance(self.file, (bytes, BufferedIOBase, TextIOWrapper)):
+            if not self.filename and hasattr(self.file, "name"):
+                self.filename = os.path.basename(str(self.file.name))
+            elif self.filename is None and not hasattr(self.file, "name"):
+                raise ValueError(
+                    "filename must be provided when file is bytes or a file-like object without a name"
+                )
+        elif isinstance(self.file, (str, Path)):
+            if not self.filename:
+                self.filename = os.path.basename(str(self.file))
+        else:
+            raise ValueError(f"Unsupported file type: {type(self.file)}")
+
+
+# Type alias for file input that can be used across services
+FileInput = Union[str, Path, BufferedIOBase, SourceText, File]

py/llama_parse/CHANGELOG.md

@@ -1,5 +1,41 @@
 # llama_parse
 
+## 0.6.81
+
+### Patch Changes
+
+- Updated dependencies [f3233de]
+  - llama-cloud-services-py@0.6.81
+
+## 0.6.80
+
+### Patch Changes
+
+- Updated dependencies [0506c88]
+  - llama-cloud-services-py@0.6.80
+
+## 0.6.79
+
+### Patch Changes
+
+- Updated dependencies [e020e3e]
+  - llama-cloud-services-py@0.6.79
+
+## 0.6.78
+
+### Patch Changes
+
+- 9f1ef4e: Fix extract
+- Updated dependencies [9f1ef4e]
+  - llama-cloud-services-py@0.6.78
+
+## 0.6.77
+
+### Patch Changes
+
+- Updated dependencies [407292b]
+  - llama-cloud-services-py@0.6.77
+
 ## 0.6.76
 
 ### Patch Changes

py/llama_parse/package.json

@@ -1,6 +1,6 @@
 {
   "name": "llama_parse",
-  "version": "0.6.76",
+  "version": "0.6.81",
   "description": "",
   "main": "index.js",
   "private": false,

py/llama_parse/pyproject.toml

@@ -11,13 +11,13 @@ dev = [
 
 [project]
 name = "llama-parse"
-version = "0.6.76"
+version = "0.6.81"
 description = "Parse files into RAG-Optimized formats."
 authors = [{name = "Logan Markewich", email = "logan@llamaindex.ai"}]
 requires-python = ">=3.9,<4.0"
 readme = "README.md"
 license = "MIT"
-dependencies = ["llama-cloud-services>=0.6.76"]
+dependencies = ["llama-cloud-services>=0.6.81"]
 
 [project.scripts]
 llama-parse = "llama_parse.cli.main:parse"

py/package.json

@@ -1,6 +1,6 @@
 {
   "name": "llama-cloud-services-py",
-  "version": "0.6.76",
+  "version": "0.6.81",
   "private": false,
   "license": "MIT",
   "scripts": {},

py/pyproject.toml

@@ -14,12 +14,15 @@ dev = [
   "ipython>=8.12.3,<9",
   "jupyter>=1.1.1,<2",
   "mypy>=1.14.1,<2",
-  "pydantic-settings>=2.10.1"
+  "pydantic-settings>=2.10.1",
+  "pandas",
+  "openpyxl",
+  "pyarrow"
 ]
 
 [project]
 name = "llama-cloud-services"
-version = "0.6.76"
+version = "0.6.81"
 description = "Tailored SDK clients for LlamaCloud services."
 authors = [{name = "Logan Markewich", email = "logan@runllama.ai"}]
 requires-python = ">=3.9,<4.0"
@@ -27,14 +30,15 @@ readme = "README.md"
 license = "MIT"
 dependencies = [
   "llama-index-core>=0.12.0",
-  "llama-cloud==0.1.43",
+  "llama-cloud==0.1.44",
   "pydantic>=2.8,!=2.10",
   "click>=8.1.7,<9",
   "python-dotenv>=1.0.1,<2",
   "eval-type-backport>=0.2.0,<0.3 ; python_version < '3.10'",
   "platformdirs>=4.3.7,<5",
   "tenacity>=8.5.0, <10.0",
-  "packaging>=23.0"
+  "packaging>=23.0",
+  "respx[tests]>=0.22.0"
 ]
 
 [project.scripts]

py/test_utils_plan.md

@@ -0,0 +1,101 @@
+# Testing Utils Implementation Plan
+
+## Goal
+Build a `FakeLlamaCloudServer` that intercepts all SDK HTTP traffic (extract, parse, classify, files, etc.) and returns deterministic responses so offline tests behave like production, per `testing_utils_spec.md`.
+
+## High-Level Phases
+1. **Router + lifecycle scaffold**
+   - Implement `FakeLlamaCloudServer` with `respx.Router` that can be used as a context manager or via explicit `install()` / `uninstall()`.
+   - Support multiple base URLs and namespace filtering so only selected APIs are intercepted.
+2. **State stores + deterministic generators**
+   - Create in-memory stores for files, jobs, runs, parse results, and classification predictions.
+   - Implement deterministic data generation seeded by (file hash, schema hash, namespace) as described in the spec.
+3. **Namespace handlers**
+   - Extract: stub `/api/v1/extraction/extraction-agents/*`, `/api/v1/extraction/run`, `/api/v1/extraction/jobs*`, `/api/v1/extraction/runs/by-job/{id}`.
+   - Files: stub `/api/v1/files/**` plus presigned upload/download workflows.
+   - Parse: stub `/api/parsing/upload`, `/api/parsing/job/{id}`, `/api/parsing/job/{id}/result/{result_type}`.
+   - Classify: stub `/api/v1/classifier/*` (job creation, polling, results).
+4. **Matcher + override system**
+   - Implement `RequestMatcher`, `FileMatcher`, `SchemaMatcher`, etc., and expose helper APIs like `fake.extract.stub_run`.
+5. **Ergonomic utilities**
+   - Provide helper shortcuts (`fake.extract.stateless_run`) and spy APIs (call count assertions).
+6. **Docs + tests**
+   - Document usage in `testing_utils_spec.md`.
+   - Add tests that demonstrate end-to-end flows using the fake.
+
+## Detailed Steps & Considerations
+
+### 1. Router & Lifecycle
+- Create a single `FakeLlamaCloudServer` class that holds a `respx.Router` configured for each base URL.
+- Provide `__enter__/__exit__` plus `install()/uninstall()` to attach/detach the router.
+- Complexity: need to handle both sync and async clients since `LlamaParse` uses raw `httpx.AsyncClient` instances constructed on the fly. Ensure `respx.mock(assert_all_called=False)` works for both.
+
+### 2. State Management & Determinism
+- Implement a `FileStore` that tracks uploaded file bytes, metadata, generated IDs, and seeded RNG values.
+- Implement `ExtractStore`, `ParseStore`, `ClassifyStore` to track job lifecycles and generated runs.
+- Deterministic generator design:
+  - Compute SHA256 of (file bytes + filename) and of normalized schema JSON.
+  - Combine into a seed (e.g., `seed = sha256(file_hash + schema_hash)`).
+  - Use that seed for namespace-specific RNG (extract uses schema walk, parse uses layout heuristics, classify uses label sets).
+- Complexity: schema normalization requires Pydantic `model_json_schema()` ordering; ensure we match production ordering to avoid drift.
+
+### 3. Namespace Handlers
+#### Extract
+- Stub endpoints listed under `LlamaExtract` usage (`create_extraction_agent`, `run_job`, stateless run, poll job/run).
+- Mirror response bodies (`ExtractJob`, `ExtractRun`, `PaginatedExtractRunsResponse`) so the SDK’s type deserialization works.
+- Manage transitions `PENDING → SUCCESS/FAILED` with realistic timestamps.
+#### Files
+- Implement both presigned workflow and direct upload fallback:
+  - `POST /api/v1/files` (or equivalent) returns a fake presigned URL (e.g., `https://fake-upload.local/{file_id}`) that our router also intercepts.
+  - The subsequent `PUT` should store the bytes and mark upload complete.
+  - `GET /api/v1/files/{id}` returns stored metadata; `read_file_content` returns presigned download URLs or raw bytes.
+- Complexity: need to intercept arbitrary presigned hostnames (e.g., AWS S3). Spec does not clarify if presigned URLs live on the same base; we may need to whitelist custom domains or provide a fake S3 host.
+#### Parse
+- Because `LlamaParse` manually constructs `/api/parsing/*` URLs, ensure the fake registers these exact routes against every provided base URL.
+- Store job configs, return deterministic `JobResult` payloads (text, markdown, JSON), and support partitioned jobs.
+#### Classify
+- Stub job creation/polling/responses, ensuring statuses transition according to `StatusEnum`.
+- Return deterministically chosen labels based on input payload + rules (seed derived from contents).
+
+### 4. Matcher / Override System
+- Provide dataclasses from the spec (`FileMatcher`, `SchemaMatcher`, `RequestMatcher`).
+- Implement matcher evaluation order with `once=True` behavior to remove one-time overrides.
+- Expose helper APIs:
+  - `fake.extract.stub_run(...)`
+  - `fake.parse.stub_parse(...)`
+  - `fake.classify.stub_prediction(...)`
+  - `fake.files.stub_upload(...)`, etc.
+- Complexity: Need to ensure matcher evaluation can inspect raw `httpx.Request` bodies/headers for both sync and async flows without consuming the stream twice.
+
+### 5. Assertions & Spies
+- Expose convenience attributes pointing to `respx.Route` objects for frequently used paths (e.g., `fake.extract.stateless_run`).
+- Provide helper methods for call counts, captured requests, etc.
+- Ensure naming stays stable to avoid brittle tests.
+
+### 6. Testing Strategy
+- Add pytest fixtures to install the fake server globally for integration tests.
+- Cover scenarios:
+  - Stateless extract returns deterministic payload.
+  - Agent-backed extract polls job/runs.
+  - Files API handles presigned upload and retrieval.
+  - Parse job lifecycle for both success and failure.
+  - Classification job with deterministic label output.
+  - Matcher overrides injection & once-only behavior.
+  - Mixed namespace configurations (e.g., intercept extract only, let parse hit real network).
+
+## Extra Complexity & Spec Concerns
+- **Presigned URL scope**: Spec assumes presigned uploads can be intercepted the same way as SaaS APIs, but actual presigned URLs often point to AWS domains outside `base_urls`. Need a strategy (e.g., generate fake host names the SDK will call, or rewrite responses to use local URLs).
+- **Async client coverage**: LlamaParse builds new `httpx.AsyncClient` objects; the spec’s install/uninstall story must ensure respx patches all clients, not just the global one.
+- **Deterministic generators**: The spec outlines hashing inputs but doesn’t define exact algorithms. Without mirroring production generator logic, fixtures might diverge. We may need to document any intentional differences.
+- **Job state timelines**: The spec expects transitions (`PENDING → SUCCESS`) with realistic timestamps. Need to ensure we schedule async updates or respond with multi-step polling; otherwise, tests relying on delays may behave differently.
+- **Namespace toggling**: Clarify behavior when a namespace is disabled—should unmatched routes fall through automatically or raise? Current spec implies fall-through to real network, but that could be surprising in CI.
+- **Schema handling**: `_validate_schema` currently calls production for dict schemas. The fake must emulate validation; otherwise tests will still hit SaaS. Spec doesn’t detail validation logic, so we must decide on a simplified validator or deterministic echo.
+- **Parse partitioning**: `LlamaParse` can partition jobs and expects consistent pagination semantics. Need to ensure deterministic results respect `target_pages`, `partition_pages`, etc., or document limitations.
+
+## Next Actions
+1. Prototype router + lifecycle with namespace toggles.
+2. Implement FileStore + presigned workflow since other namespaces depend on file IDs.
+3. Build deterministic generators and stores for extract/parse/classify.
+4. Layer matcher/override system on top of the stores.
+5. Write initial tests per namespace and refine spec gaps (presigned host, validation behavior).
+6. Update `testing_utils_spec.md` with any clarifications discovered above.

py/test_utils_research.md

@@ -0,0 +1,199 @@
+# Testing Utils Research
+
+## Objective
+
+The `FakeLlamaCloudServer` proposal aims to intercept raw HTTP traffic for extract, parse, classify, and files APIs so local tests behave like SaaS without per-test stubbing. The mock must respect base URLs, support context-manager or long-lived install modes, and deterministically synthesize payloads from file + schema hashes while still allowing targeted overrides.
+
+```1:67:py/testing_utils_spec.md
+## Local Testing Utilities 2.0 (Spec Draft)
+
+- **Everything mocked by default** …
+- **Context manager optional** …
+- **Pydantic-first ergonomics** …
+- **API-only contract** …
+```
+
+## Python Hand-Written SDK Surfaces
+
+These modules sit on top of the generated `llama_cloud` client and are what most application tests exercise. `FakeLlamaCloudServer` must satisfy the HTTP contracts they rely on.
+
+### Extract flow (`py/llama_cloud_services/extract/extract.py`)
+
+- The class imports resource types from `llama_cloud` and owns an `AsyncLlamaCloud` client shared across both stateless and agent-backed flows.
+
+```17:37:py/llama_cloud_services/extract/extract.py
+from llama_cloud import (
+    ExtractAgent as CloudExtractAgent,
+    ExtractConfig,
+    …
+)
+from llama_cloud.client import AsyncLlamaCloud
+```
+
+- Schema validation always calls `POST /api/v1/extraction/extraction-agents/schema/validation` through `client.llama_extract.validate_extraction_schema`, so the fake must mimic that endpoint.
+
+```65:82:py/llama_cloud_services/extract/extract.py
+async def _validate_schema(...):
+    …
+    validated_schema = await client.llama_extract.validate_extraction_schema(
+        data_schema=processed_schema
+    )
+```
+
+- Agent creation, listing, and run management use the `llama_extract` namespace. These calls surface the same request/response bodies as the SaaS API, so tests that interact through agents expect consistent metadata (IDs, status enums, etc.).
+
+```636:738:py/llama_cloud_services/extract/extract.py
+def create_agent(...):
+    agent = self._run_in_thread(
+        self._async_client.llama_extract.create_extraction_agent(
+            project_id=self._project_id,
+            …
+        )
+    )
+    return ExtractionAgent(...)
+```
+
+- Stateless extraction queues work by converting file inputs into either `file_id`, inline text, or base64 payloads and forwarding them to `POST /api/v1/extraction/run`. Deterministic responses from the fake should be keyed off `processed_schema` + whichever file representation the SDK sent.
+
+```921:1018:py/llama_cloud_services/extract/extract.py
+async def queue_extraction(...):
+    processed_schema = await _validate_schema(...)
+    …
+    job = await self._async_client.llama_extract.extract_stateless(
+        project_id=self._project_id,
+        organization_id=self._organization_id,
+        data_schema=processed_schema,
+        config=config,
+        **file_args,
+    )
+```
+
+### File uploads (`py/llama_cloud_services/files/client.py`)
+
+- All higher-level services route file uploads/downloads through `FileClient`. When `use_presigned_url` is enabled, the client first calls `POST /api/v1/files` (generate URL), then performs the PUT upload directly, then fetches metadata via `GET /api/v1/files/{id}`. The fake must intercept both the API calls and the presigned PUT hops to return consistent `file_id`s and stored bytes.
+
+```63:140:py/llama_cloud_services/files/client.py
+presigned_url = await self.client.files.generate_presigned_url(...)
+upload_response = await httpx_client.put(presigned_url.url, data=buffer.read())
+…
+return await self.client.files.get_file(presigned_url.file_id, …)
+```
+
+### Parse reader (`py/llama_cloud_services/parse/base.py`)
+
+- `LlamaParse` is a bespoke reader that talks straight to HTTP routes defined in the module (e.g., `/api/parsing/upload`, `/api/parsing/job/{id}`). Unlike `LlamaExtract`, it does not go through the generated client; instead it builds URLs manually and uses `httpx`/`make_api_request`. A fake server must therefore implement these exact paths.
+
+```49:66:py/llama_cloud_services/parse/base.py
+JOB_RESULT_URL = "/api/parsing/job/{job_id}/result/{result_type}"
+JOB_STATUS_ROUTE = "/api/parsing/job/{job_id}"
+JOB_UPLOAD_ROUTE = "/api/parsing/upload"
+```
+
+```1056:1070:py/llama_cloud_services/parse/base.py
+url = build_url(JOB_UPLOAD_ROUTE, self.organization_id, self.project_id)
+resp = await make_api_request(self.aclient, "POST", url, …, files=files, data=data)
+```
+
+### Classifier beta client (`py/llama_cloud_services/beta/classifier/client.py`)
+
+- Classification flows reuse `FileClient` for uploads and then call `AsyncLlamaCloud.classifier` endpoints (`create_classify_job`, `get_classify_job`, `get_classification_job_results`). Long-running tests poll until status becomes terminal, so mocking needs to cover both the enqueue POST and the follow-up GETs.
+
+```75:151:py/llama_cloud_services/beta/classifier/client.py
+return await self.client.classifier.create_classify_job(...)
+…
+results = await self.client.classifier.get_classification_job_results(
+    classify_job_with_status.id,
+    project_id=self.project_id,
+)
+```
+
+## Generated Python Client
+
+- The repo depends on the published `llama_cloud` package (currently 0.1.44) which is itself generated from the OpenAPI spec. All hand-written modules import types and service clients from this package, so the fake server may need to mirror whatever transport settings `AsyncLlamaCloud` expects (headers, pagination, etc.).
+
+```1605:1608:py/uv.lock
+sdist = { … "llama_cloud-0.1.44.tar.gz", … }
+wheels = [{ … "llama_cloud-0.1.44-py3-none-any.whl", … }]
+```
+
+- Since `AsyncLlamaCloud` handles auth headers and base URLs, integrating the fake server typically means pointing `LLAMA_CLOUD_BASE_URL` at the mock and letting the generated client continue to build resource paths.
+
+## TypeScript + OpenAPI Assets
+
+- The canonical OpenAPI document lives in `ts/llama_cloud_services/openapi.json`. It defines every path/operation used by both the generated TypeScript SDK and the Python client. For example, the stateless extract endpoint is captured as `POST /api/v1/extraction/run`.
+
+```13825:13872:ts/llama_cloud_services/openapi.json
+"/api/v1/extraction/run": {
+  "post": {
+    "summary": "Extract Stateless",
+    "description": "… Requires data_schema, config, and either file_id, text, or base64 encoded file data.",
+    …
+  }
+}
+```
+
+- The OpenAPI document is downloaded from production via `scripts/download.mjs` and then fed into `@hey-api/openapi-ts` to regenerate the TypeScript client and schema wrappers. Keeping the fake server in sync with the spec means you can diff regenerated clients when the API evolves.
+
+```3:21:ts/llama_cloud_services/scripts/download.mjs
+const response = await fetch('https://api.cloud.llamaindex.ai/api/openapi.json');
+…
+fs.writeFileSync('openapi.json', JSON.stringify(data, null, 2));
+```
+
+```1:24:ts/llama_cloud_services/openapi-ts.config.ts
+export default defineConfig({
+  input: "./openapi.json",
+  output: { path: "./src/client", format: "prettier", lint: "eslint" },
+  plugins: [ … "@hey-api/sdk", "@hey-api/typescript" ],
+});
+```
+
+- The public TypeScript surface (`src/LlamaClassify.ts`, etc.) already consumes the generated client by injecting auth headers and delegating to `classify(...)`. If Python tests eventually need to reuse the same fake server, TypeScript examples provide another reference for how SDK consumers expect responses to look.
+
+```12:74:ts/llama_cloud_services/src/LlamaClassify.ts
+export class LlamaClassify {
+  constructor(apiKey?: string, baseUrl?: string, region?: string) {
+    …
+    this.client = createClient(createConfig({ baseUrl: url, headers: { Authorization: `Bearer ${key}` }}));
+  }
+
+  async classify(rules, configuration, { fileContents, filePaths, projectId, … }) {
+    const result = await classify(rules, configuration, {
+      fileContents,
+      filePaths,
+      projectId: projectId ?? undefined,
+      client: this.client,
+      …
+    });
+    return result;
+  }
+}
+```
+
+## Endpoint Map to Stub First
+
+Cross-referencing the Python call-sites with the OpenAPI spec yields the minimum set of HTTP routes the fake server must implement:
+
+1. `/api/v1/extraction/run` for stateless jobs, plus the agent CRUD endpoints under `/api/v1/extraction/extraction-agents`, `/api/v1/extraction/jobs`, and `/api/v1/extraction/runs/by-job/{id}` (see `LlamaExtract` usage above).
+2. `/api/v1/files/**` for upload/generate-presigned/list/get/delete, plus any presigned `PUT` destinations (`FileClient`).
+3. `/api/parsing/upload`, `/api/parsing/job/{job_id}`, `/api/parsing/job/{job_id}/result/{result_type}` (direct HTTPX calls in `LlamaParse`).
+4. `/api/v1/classifier/**` for job creation, polling, and result retrieval (`LlamaClassify` and `ClassifyClient`).
+
+Having deterministic handlers for these routes unlocks end-to-end coverage of extract/parse/classify flows without touching live SaaS.
+
+## Implementation Reminders from the Spec
+
+- Namespace toggles let tests intercept a subset of APIs while letting others fall through—mirror this by allowing `FakeLlamaCloudServer(namespaces=[...])` to selectively register respx routes.
+- Deterministic payloads should derive from uploaded file bytes + schema hashes for extract, layout characteristics for parse, and label sets for classify so that rerunning the same test yields identical responses (reducing fixture churn).
+- Keep the matcher system (`RequestMatcher`, `FileMatcher`, etc.) flexible so individual tests can stub failures (e.g., presigned upload errors, job timeouts) without reconfiguring the entire fake.
+
+```44:210:py/testing_utils_spec.md
+with FakeLlamaCloudServer() as fake:
+    extractor = LlamaExtract(...)
+    parser = LlamaParse(...)
+    classifier = LlamaClassify(...)
+…
+fake.extract.stub_run(... RequestMatcher ...)
+```
+
+Armed with the file map above, a new developer can trace any SDK call from the hand-written layers down to the generated client and the authoritative OpenAPI route, making it clear where the fake server needs to hook in.

py/testing_utils_spec.md

@@ -0,0 +1,329 @@
+## Local Testing Utilities 2.0 (Spec Draft)
+
+Offline testing should feel identical to calling the public LlamaCloud API. The new utilities center a single `FakeLlamaCloudServer` that intercepts HTTP traffic at the API boundary, deterministically generates responses from real files + schemas, and only requires overrides when a test needs to exercise edge cases.
+
+### Design goals
+- **Everything mocked by default**: instantiating `FakeLlamaCloudServer()` wires up every public LlamaCloud namespace (extract, parse, classify, files, etc.) so the SDK behaves as if it were talking to production. Deterministic responses are returned without any additional wiring.
+- **Context manager optional**: `FakeLlamaCloudServer` still supports `with ...` for pytest isolation, but you can call `install()` / `uninstall()` to keep the mock server active inside a long-running process (e.g., a FastAPI dev server that proxies to the fake).
+- **Pydantic-first ergonomics**: all documentation and helpers assume schemas are declared as `BaseModel` subclasses. JSON Schema dictionaries are still accepted for compatibility.
+- **API-only contract**: handlers talk raw HTTP (request dicts, status codes, JSON payloads) so we can reuse the mock in future SDKs or other languages without depending on `LlamaExtract`.
+
+### Quick start (pytest-friendly, deterministic by default)
+
+```python
+from pathlib import Path
+from pydantic import BaseModel, Field
+from llama_cloud import ExtractConfig, ExtractMode
+from llama_cloud_services.extract import LlamaExtract
+from llama_cloud_services.testing_utils import FakeLlamaCloudServer
+
+
+class Receipt(BaseModel):
+    merchant: str = Field(description="Vendor name")
+    total: float = Field(description="Grand total in USD")
+
+
+config = ExtractConfig(extraction_mode=ExtractMode.FAST)
+pdf_path = Path("tests/fixtures/receipt.pdf")
+
+
+with FakeLlamaCloudServer() as fake:
+    extractor = LlamaExtract(
+        api_key="test-key",
+        verify=False,
+    )
+    run = extractor.extract(Receipt, config, pdf_path)
+    assert run.status.value == "SUCCESS"
+    assert run.data["total"] > 0  # generated entirely from file + schema
+```
+
+Key points:
+- No manual stubbing required. The fake server hashes the uploaded file bytes + schema JSON to derive a deterministic seed and walks the schema to produce stable mock data.
+- `FakeLlamaCloudServer` automatically intercepts the default SaaS URL (`https://api.cloud.llamaindex.ai`). If your tests point at another host (e.g., BYOC), pass it via `FakeLlamaCloudServer(base_urls=["https://byoc.dev/api"])`; otherwise, keep using your normal SDK base URL.
+
+### Works across extract, parse, classify
+
+```python
+from llama_cloud_services.testing_utils import FakeLlamaCloudServer
+from llama_cloud_services.extract import LlamaExtract
+from llama_cloud_services.parse import LlamaParse
+from llama_cloud_services.classify import LlamaClassify
+
+
+with FakeLlamaCloudServer() as fake:
+    extractor = LlamaExtract(api_key="test-key")
+    parser = LlamaParse(api_key="test-key")
+    classifier = LlamaClassify(api_key="test-key")
+
+    run = extractor.extract(
+        Receipt, config, "noisebridge.pdf"
+    )  # reuse quick-start schema/config
+    parse_result = parser.parse("noisebridge.pdf")
+    classification = classifier.classify({"text": "foo"})
+
+    assert run.status.value == "SUCCESS"
+    assert parse_result.documents[0].text  # deterministically generated
+    assert classification.prediction in {
+        "A",
+        "B",
+    }  # stable RNG driven by payload
+```
+
+Every namespace uses its own deterministic generator (schema-driven for extract, layout-driven for parse, label-driven for classify) but shares the same matcher/override system described below.
+
+### Limiting intercepted APIs
+
+If you only need a subset of APIs (e.g., extract + files during early bring-up), pass `namespaces` explicitly. Anything omitted will fall through to the real network, which is handy for hybrid tests.
+
+```python
+fake = FakeLlamaCloudServer(
+    namespaces=["extract", "files"],
+    base_urls=["https://api.cloud.llamaindex.ai"],  # optionally point at BYOC
+)
+
+with fake:
+    extractor = LlamaExtract(api_key="test-key")
+    extractor.extract(Receipt, config, "noisebridge.pdf")
+```
+
+### Long-lived install for iterative development
+
+```python
+import os
+from contextlib import asynccontextmanager
+from fastapi import FastAPI
+
+fake = FakeLlamaCloudServer(
+    namespaces=["extract"],
+    base_urls=[
+        os.environ.get(
+            "LLAMA_CLOUD_BASE_URL", "https://api.cloud.llamaindex.ai"
+        )
+    ],
+)
+
+
+@asynccontextmanager
+async def lifespan(app):
+    fake.install()
+    app.state.extractor = LlamaExtract(
+        api_key="dev",
+        verify=False,
+    )
+    try:
+        yield
+    finally:
+        fake.uninstall()
+
+
+app = FastAPI(lifespan=lifespan)
+```
+
+`install()`/`uninstall()` simply wrap the respx router lifecycle so you can keep the mock server hot for REPLs, background workers, or manual QA sessions without relying on a context manager.
+
+### Files API behavior
+
+`FakeLlamaCloudServer` ships with a first-class fake for `/api/v1/files/*` because uploads sit on the critical path for both extract and downstream workflows that pre-stage files.
+
+- Every call to `POST /files/generate-presigned-url`, the subsequent `PUT` upload, and the follow-up `GET /files/{file_id}` is intercepted and stored in-memory. The response objects mirror the real API so `FileClient` keeps working unchanged.
+- `fake.files.preload(path="tests/fixtures/plan.pdf", filename="plan.pdf")` ingests local fixtures ahead of time and returns a reusable `file_id`, which is useful when tests pass `SourceText(file_id=...)`.
+- `fake.files.stub_upload(...)` lets you simulate storage failures (e.g., 413 "file too large") using the same matcher system as extract.
+- You can download what the SDK uploaded via `fake.files.read(file_id)` to assert on the bytes or to feed downstream mocks.
+
+```python
+from llama_cloud_services.extract import SourceText
+
+with FakeLlamaCloudServer() as fake:
+    file_id = fake.files.preload(path="tests/fixtures/noisebridge.pdf")
+    extractor = LlamaExtract(api_key="test-key")
+    run = extractor.extract(Receipt, config, SourceText(file_id=file_id))
+    assert fake.files.read(file_id).startswith(b"%PDF")
+```
+
+### Deterministic response generation
+1. Files uploaded via `/api/v1/files` (or inlined via `extract_stateless`) are fingerprinted using SHA256 (file content bytes + filename).
+2. Schemas are normalized (Pydantic `model_json_schema()` plus sorted keys) and hashed.
+3. A seed derived from `sha256(file_fingerprint + schema_digest)` feeds a tiny RNG that walks the schema to synthesize values (numbers, strings, arrays) while respecting field metadata (descriptions hint names, numeric ranges, etc.).
+4. Runs transition through the same states as production (`PENDING` → `SUCCESS`) and return realistic timestamps, metadata, and config echoes.
+
+Because the seed is stable, rerunning the same schema/file pair yields identical mock payloads without stubbing.
+
+### Stubbing, spying, and assertions
+Most tests only need the deterministic defaults, but the fake server provides a layered set of helpers for overriding responses, asserting call counts, and finally dropping down to raw `respx` when necessary.
+
+#### Matcher API
+
+```python
+from dataclasses import dataclass
+from typing import Callable, Optional
+from httpx import Request
+
+
+@dataclass
+class FileMatcher:
+    filename: str | None = None
+    sha256: str | None = None
+    file_id: str | None = None
+
+
+@dataclass
+class SchemaMatcher:
+    model: type[BaseModel] | None = None
+    schema_hash: str | None = None
+
+
+@dataclass
+class RequestMatcher:
+    file: FileMatcher | Callable[[Request], bool] | None = None
+    schema: SchemaMatcher | None = None
+    agent_id: str | None = None
+    project_id: str | None = None
+    organization_id: str | None = None
+    predicate: Callable[[Request], bool] | None = None
+```
+
+Every callable matcher receives the raw `httpx.Request` object that respx captured, so you can inspect headers, cookies, bodies, etc., without learning another wrapper type. The helper dataclasses (`FileMatcher`, `SchemaMatcher`) just cover the common cases; mix and match as needed. Stubs are evaluated in registration order, and `once=True` removes the stub after the first match.
+
+#### Stateless extraction example
+
+```python
+fake.extract.stub_run(
+    matcher=RequestMatcher(file=FileMatcher(filename="noisebridge.pdf")),
+    data={"merchant": "Noisebridge", "total": 42.0},
+    status="SUCCESS",  # defaults to deterministic timeline when omitted
+    metadata={"source": "unit-test"},
+    once=True,
+)
+
+run = extractor.extract(Receipt, config, "noisebridge.pdf")
+assert run.data["merchant"] == "Noisebridge"
+```
+
+`data` accepts dictionaries, Pydantic models, or callables (`Callable[[Request], dict]`). If you omit `status`, the stub only replaces the payload while preserving the deterministic job/run lifecycle.
+
+#### Agent extraction example
+
+```python
+agent = extractor.create_agent(
+    name="tests", data_schema=Receipt, config=config
+)
+
+fake.extract.stub_agent_run(
+    agent_id=agent.id,
+    matcher=RequestMatcher(file=FileMatcher(filename="bad.pdf")),
+    job_status="FAILED",  # overrides POST /extraction/jobs
+    run_status="FAILED",  # overrides GET /runs/by-job
+    error={"message": "Schema mismatch"},
+)
+
+with pytest.raises(ApiError):
+    agent.extract("bad.pdf")
+```
+
+`stub_agent_run` targets the stateful job pipeline (`/extraction/jobs`, `/extraction/jobs/{id}`, `/extraction/runs/by-job/{id}`) so you can mimic long-running failures, retries, or partial completions without hand-writing multiple HTTP handlers.
+
+#### Parse, classify, and files stubs
+
+- `fake.parse.stub_parse(...)` lets you override document splits, token counts, or even return structured HTML for specific file IDs.
+- `fake.classify.stub_prediction(...)` accepts label sets and score distributions so you can test downstream logic that inspects confidences.
+- `fake.files.stub_upload(...)` / `fake.files.stub_download(...)` simulate storage edge cases such as timeouts or corrupted content.
+
+Because every namespace uses the same matcher primitives, you can coordinate multi-API scenarios (e.g., stub the file upload and the subsequent extract run) without duplicating predicates.
+
+#### Assertions without extra abstractions
+
+Since everything runs through the same `respx.MockRouter` you already use elsewhere, assertions stay lightweight:
+
+```python
+with FakeLlamaCloudServer() as fake:
+    route = fake.router["POST", "/api/v1/extraction/run"]
+    extractor = LlamaExtract(api_key="test-key")
+    extractor.extract(Receipt, config, "noisebridge.pdf")
+
+assert route.called
+assert route.call_count == 1
+req = route.calls[0].request  # this is httpx.Request
+assert req.headers["authorization"].startswith("Bearer ")
+```
+
+For friendlier names, every frequently used route is also pinned to a stable attribute:
+
+- `fake.extract.stateless_run` (alias: `fake.extract_run`) → `POST /api/v1/extraction/run`
+- `fake.extract.agent_job` → `POST /api/v1/extraction/jobs`
+- `fake.extract.agent_run` → `GET /api/v1/extraction/runs/by-job/{job_id}`
+- `fake.files.upload` → `POST /api/v1/files/upload` (or the presigned PUT hop, depending on mode)
+- `fake.files.get` → `GET /api/v1/files/{file_id}`
+
+Each attribute is the underlying `respx.Route`, so assertions feel natural:
+
+```python
+assert fake.extract.stateless_run.called
+fake.extract.stateless_run.assert_called_once()
+assert fake.files.upload.call_count == 1
+assert fake.extract_run.called  # global alias for the same route
+```
+
+If you ever need the full mapping, `fake.extract.routes["stateless_run"] is fake.extract.stateless_run`.
+
+#### Advanced (respx-level) overrides
+
+When you need total control, drop straight into respx:
+
+```python
+route = fake.router["POST", "/api/v1/extraction/run"]
+route.mock(side_effect=lambda request: (418, {"detail": "I'm a teapot"}))
+```
+
+Or use the attribute shortcuts:
+
+```python
+fake.extract.stateless_run.mock(
+    side_effect=lambda request: (500, {"detail": "boom"})
+)
+```
+
+Either way you're dealing with the canonical respx objects, so regex paths, call assertions, and other ecosystem tools keep working. The only convention is that handlers should return `(status_code, json_body | bytes)` so logging and deterministic fallbacks remain consistent.
+
+### API-layer implementation hints
+- Route decorators such as `server.add_handler("POST", "/api/v1/extraction/run")` install handlers for **every** registered base URL declared in the constructor, keeping the mock independent of SDK client classes.
+- Request objects passed to handlers expose method, URL, headers, query params, JSON body, and raw bytes—everything needed to mirror production without importing internal models.
+- Namespaces self-register via the constructor (e.g., `namespaces=["extract"]`) so no additional attach helpers are required; future SDKs can opt into the same HTTP contracts by toggling the namespaces they care about.
+
+## Research: Extract SDK surface map
+
+The current Python SDK (`py/llama_cloud_services/extract/extract.py`) is a thin wrapper over the HTTP API exposed in `ts/llama_cloud_services/openapi.json`. Understanding this mapping helps ensure the fake server mirrors the real contract.
+
+### Core classes
+- `LlamaExtract`: factory that owns an `AsyncLlamaCloud` client, manages thread pools, and exposes both stateless extraction (`extract`, `aextract`, `queue_extraction`) and agent CRUD helpers.
+- `ExtractionAgent`: wraps an existing agent returned by the API and provides methods for queuing files, polling jobs, listing runs, updating schemas/configs, and deleting runs.
+- `FileClient`: abstracts the `/api/v1/files` upload + download flow, including presigned URL handling for uploads.
+
+### Stateless extraction flow
+1. `LlamaExtract.queue_extraction(data_schema, config, files)` validates schemas via `POST /api/v1/extraction/extraction-agents/schema/validation`, converts input files into either `file_id`, `file` (base64 body), or inline `text`.
+2. For each file the SDK calls `POST /api/v1/extraction/run` with the processed schema + config + file payload. The API responds with an `ExtractJob`.
+3. `LlamaExtract.aextract` waits for completion by polling `_wait_for_job_result`, which hits `GET /api/v1/extraction/jobs/{job_id}` until the job is `SUCCESS`/`FAILED`, then fetches the run via `GET /api/v1/extraction/runs/by-job/{job_id}`. The synchronous `extract` just wraps this coroutine in a worker thread.
+
+### Agent-backed flow
+1. `create_agent` issues `POST /api/v1/extraction/extraction-agents` with name, schema, and config; responses seed `ExtractionAgent`.
+2. `ExtractionAgent.queue_extraction` uploads files via `FileClient`, then enqueues jobs with `POST /api/v1/extraction/jobs` (or `/jobs/file` for multipart uploads). Returned job IDs are polled via `_wait_for_job_result` just like the stateless path.
+3. `ExtractionAgent.list_extraction_runs` and `delete_extraction_run` map to `GET /api/v1/extraction/runs` (with pagination) and `DELETE /api/v1/extraction/runs/{run_id}` respectively.
+4. Manual inspection helpers (`get_extraction_job`, `get_extraction_run_for_job`, `get_extraction_run`) call `GET /api/v1/extraction/jobs/{job_id}` and `GET /api/v1/extraction/runs/by-job/{job_id}` / `GET /api/v1/extraction/runs/{run_id}`.
+
+### Files API touch points
+- Uploads default to presigned URLs: the SDK first calls `POST /api/v1/files/generate-presigned-url`, then performs an HTTP PUT to the returned URL, and finally fetches the file metadata via `GET /api/v1/files/{file_id}`.
+- When BYOC deployments disable presigned uploads, `FileClient` falls back to `POST /api/v1/files/upload`.
+
+### Implications for the fake server
+- **API-level parity**: mocking should happen at the HTTP layer (matching the endpoints listed above) so new SDKs can reuse the fake by simply pointing their base URL at it.
+- **State surfaces**: to emulate production, the fake needs in-memory stores for files, jobs, and runs keyed by UUIDs, plus schema validation stubs that mimic the `/schema/validation` endpoint.
+- **Deterministic generators**: since `ExtractRun.data` is derived from schema + file, implementing the generator once at the API layer ensures consistency across SDKs.
+- **Error simulation hooks**: overrides should let us short-circuit any endpoint (jobs, runs, schema validation) without changing SDK code, mirroring how the real API might fail.
+
+This map should serve as the checklist when we implement the mock: if an SDK method calls a certain path, our fake server must expose the same path with compatible request/response bodies so we can eventually lift these utilities into a standalone package.
+
+## Implementation status
+
+- `FakeLlamaCloudServer` now lives in `llama_cloud_services.testing_utils` and registers the files, extract, parse, and classify namespaces by default. It can be used as `with FakeLlamaCloudServer(): ...` or by calling `install()` / `uninstall()` explicitly.
+- Deterministic payloads derive from file fingerprints plus schema hashes for extraction, seeded layout information for parse, and rule/file hashes for classification. The helper exposes matcher-driven overrides (`stub_run`, `stub_agent_run`, `files.stub_upload`) so tests can simulate errors.
+- Common `respx.Route` handles are exposed via namespaces (`fake.extract.stateless_run`, `fake.extract.agent_job`, `fake.files.routes["upload"]`, etc.) for assertions that mirror the examples in this spec.
+- End-to-end usage is covered in `py/unit_tests/testing_utils/test_fake_server.py`, which exercises stateless extraction, agent-backed uploads, and LlamaParse readers entirely against the fake server without external network calls.

py/tests/beta/sheets/test_sheets.py

@@ -0,0 +1,171 @@
+import os
+import tempfile
+import pytest
+import pandas as pd
+
+from llama_cloud_services.beta.sheets import LlamaSheets
+from llama_cloud_services.beta.sheets.types import SpreadsheetParsingConfig
+
+
+@pytest.fixture
+def sheets_client():
+    """Create a LlamaSheets client for testing."""
+    api_key = os.getenv(
+        "LLAMA_CLOUD_API_KEY", "llx-3AEorIw5v0lnJPzEOI9xSl0N8yFx3fguw0Zn8QJHzGWmwg5r"
+    )
+    base_url = os.getenv("LLAMA_CLOUD_BASE_URL", "https://api.staging.llamaindex.ai")
+
+    client = LlamaSheets(
+        api_key=api_key,
+        base_url=base_url,
+        max_timeout=300,
+        poll_interval=2,
+    )
+    return client
+
+
+@pytest.fixture
+def sample_excel_file():
+    """Create a temporary Excel file with sample data."""
+    # Create a simple dataframe with various data types
+    data = {
+        "Name": ["Alice", "Bob", "Charlie", "David", "Eve"],
+        "Age": [25, 30, 35, 40, 45],
+        "City": ["New York", "Los Angeles", "Chicago", "Houston", "Phoenix"],
+        "Salary": [50000.50, 75000.75, 100000.00, 125000.25, 150000.50],
+    }
+    df = pd.DataFrame(data)
+
+    # Create a temporary file
+    with tempfile.NamedTemporaryFile(suffix=".xlsx", delete=False) as tmp:
+        tmp_path = tmp.name
+        df.to_excel(tmp_path, index=False, sheet_name="TestSheet")
+
+    yield tmp_path
+
+    # Cleanup
+    try:
+        os.unlink(tmp_path)
+    except Exception:
+        pass
+
+
+@pytest.mark.skipif(
+    os.environ.get(
+        "LLAMA_CLOUD_API_KEY", "llx-3AEorIw5v0lnJPzEOI9xSl0N8yFx3fguw0Zn8QJHzGWmwg5r"
+    )
+    == "",
+    reason="LLAMA_CLOUD_API_KEY not set",
+)
+@pytest.mark.asyncio
+async def test_spreadsheet_extraction_e2e(
+    sheets_client: LlamaSheets, sample_excel_file: str
+):
+    """End-to-end test for spreadsheet extraction.
+
+    This test:
+    1. Creates a temporary Excel file with sample data
+    2. Uploads and extracts tables from the file
+    3. Downloads the extracted table as a DataFrame
+    4. Verifies the extracted data matches the original data
+    """
+    # Extract tables from the spreadsheet
+    result = await sheets_client.aextract_regions(sample_excel_file)
+
+    # Verify job completed successfully
+    assert result.status in ("SUCCESS", "PARTIAL_SUCCESS")
+    assert result.success is True
+
+    # Verify we extracted at least one table
+    assert len(result.regions) > 0, "Expected at least one table to be extracted"
+
+    # Get the first table
+    first_table = result.regions[0]
+    assert first_table.sheet_name == "TestSheet"
+
+    # Download the table as a DataFrame
+    extracted_df = await sheets_client.adownload_region_as_dataframe(
+        job_id=result.id,
+        region_id=first_table.region_id,
+        result_type=first_table.region_type,
+    )
+
+    # Load the original dataframe for comparison
+    original_df = pd.read_excel(sample_excel_file)
+
+    # Verify the extracted DataFrame has the expected shape
+    assert extracted_df.shape[0] == original_df.shape[0], (
+        f"Row count mismatch: extracted {extracted_df.shape[0]}, "
+        f"original {original_df.shape[0]}"
+    )
+    assert extracted_df.shape[1] == original_df.shape[1], (
+        f"Column count mismatch: extracted {extracted_df.shape[1]}, "
+        f"original {original_df.shape[1]}"
+    )
+
+    # Verify column names match
+    assert list(extracted_df.columns) == list(original_df.columns), (
+        f"Column names mismatch: extracted {list(extracted_df.columns)}, "
+        f"original {list(original_df.columns)}"
+    )
+
+    # Verify data types are preserved (at least numerically)
+    for col in original_df.columns:
+        if original_df[col].dtype in ["int64", "float64"]:
+            assert extracted_df[col].dtype in ["int64", "float64"], (
+                f"Column {col} type mismatch: extracted {extracted_df[col].dtype}, "
+                f"original {original_df[col].dtype}"
+            )
+
+    # Verify the data values match (allowing for minor type conversions)
+    for col in original_df.columns:
+        original_values = original_df[col].tolist()
+        extracted_values = extracted_df[col].tolist()
+
+        # Convert both to strings for comparison to handle type differences
+        original_str = [str(v) for v in original_values]
+        extracted_str = [str(v) for v in extracted_values]
+
+        assert original_str == extracted_str, (
+            f"Column {col} values mismatch:\n"
+            f"Original: {original_str}\n"
+            f"Extracted: {extracted_str}"
+        )
+
+
+@pytest.mark.skipif(
+    os.environ.get(
+        "LLAMA_CLOUD_API_KEY", "llx-3AEorIw5v0lnJPzEOI9xSl0N8yFx3fguw0Zn8QJHzGWmwg5r"
+    )
+    == "",
+    reason="LLAMA_CLOUD_API_KEY not set",
+)
+@pytest.mark.asyncio
+async def test_spreadsheet_extraction_with_config(
+    sheets_client: LlamaSheets, sample_excel_file: str
+):
+    """Test spreadsheet extraction with custom configuration."""
+    # Create a config with specific settings
+    config = SpreadsheetParsingConfig(
+        sheet_names=["TestSheet"],
+        include_hidden_cells=True,
+        generate_additional_metadata=True,
+    )
+
+    # Extract tables with the config
+    result = await sheets_client.aextract_regions(sample_excel_file, config=config)
+
+    # Verify job completed successfully
+    assert result.status in ("SUCCESS", "PARTIAL_SUCCESS")
+    assert result.success is True
+
+    # Verify that additional metadata was generated
+    assert len(result.worksheet_metadata) > 0
+    assert result.worksheet_metadata[0].title is not None
+    assert result.worksheet_metadata[0].description is not None
+
+    # Verify we extracted at least one table
+    assert len(result.regions) > 0
+
+    # Verify the sheet name matches
+    assert result.regions[0].sheet_name == "TestSheet"

py/tests/classifier/test_client.py

@@ -44,7 +44,6 @@ def classify_client(
     return ClassifyClient(
         async_llama_cloud_client,
         project_id=project.id,
-        organization_id=project.organization_id,
         polling_interval=1,
     )
 
@@ -56,7 +55,6 @@ def file_client(
     return FileClient(
         async_llama_cloud_client,
         project_id=project.id,
-        organization_id=project.organization_id,
         use_presigned_url=False,
     )
 
@@ -148,7 +146,6 @@ async def test_classify_file_ids_from_api_key(
         api_key=e2e_test_settings.LLAMA_CLOUD_API_KEY.get_secret_value(),
         base_url=e2e_test_settings.LLAMA_CLOUD_BASE_URL,
         project_id=pdf_file.project_id,
-        organization_id=e2e_test_settings.LLAMA_CLOUD_ORGANIZATION_ID,
     )
 
     # Classify the uploaded files

py/tests/conftest.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import pytest
 from llama_index.core.constants import DEFAULT_BASE_URL
 from pydantic import Field, SecretStr

py/tests/extract/test_extract_e2e.py

@@ -58,6 +58,8 @@ def get_test_cases():
         settings = [
             ExtractConfig(extraction_mode=ExtractMode.FAST),
             ExtractConfig(extraction_mode=ExtractMode.BALANCED),
+            ExtractConfig(extraction_mode=ExtractMode.MULTIMODAL),
+            ExtractConfig(extraction_mode=ExtractMode.PREMIUM),
         ]
 
         for input_file in sorted(input_files):

py/tests/index/test_index.py

@@ -44,7 +44,7 @@ def index_name() -> Generator[str, None, None]:
         client = LlamaCloud(token=api_key, base_url=base_url)
         pipeline = client.pipelines.search_pipelines(project_name=name)
         if pipeline:
-            client.pipelines.delete(pipeline_id=pipeline[0].id)
+            client.pipelines.delete_pipeline(pipeline_id=pipeline[0].id)
 
 
 @pytest.fixture()
@@ -83,7 +83,7 @@ def _setup_index_with_file(
 
     # add file to pipeline
     pipeline_file_create = PipelineFileCreate(file_id=file.id)
-    client.pipelines.add_files_to_pipeline_api(
+    client.pipeline_files.add_files_to_pipeline_api(
         pipeline_id=pipeline.id, request=[pipeline_file_create]
     )
 
@@ -170,6 +170,43 @@ def test_upload_file(index_name: str):
         os.remove(temp_file_path)
 
 
+@pytest.mark.skipif(
+    not base_url or not api_key, reason="No platform base url or api key set"
+)
+def test_upload_file_with_custom_metadata(index_name: str):
+    index = LlamaCloudIndex.create_index(
+        name=index_name,
+        project_name=project_name,
+        organization_id=organization_id,
+        api_key=api_key,
+        base_url=base_url,
+    )
+
+    # Create a temporary file to upload
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".txt") as temp_file:
+        temp_file.write(b"Sample content for testing upload.")
+        temp_file_path = temp_file.name
+    custom_metadata = {"foo": "bar"}
+
+    try:
+        # Upload the file
+        file_id = index.upload_file(
+            temp_file_path, custom_metadata=custom_metadata, verbose=True
+        )
+        assert file_id is not None
+
+        # Verify the file is part of the index
+        docs = index.ref_doc_info
+        temp_file_name = os.path.basename(temp_file_path)
+        assert any(
+            temp_file_name == doc.metadata.get("file_name") for doc in docs.values()
+        )
+
+    finally:
+        # Clean up the temporary file
+        os.remove(temp_file_path)
+
+
 @pytest.mark.skipif(
     not base_url or not api_key, reason="No platform base url or api key set"
 )
@@ -196,6 +233,38 @@ def test_upload_file_from_url(remote_file: Tuple[str, str], index_name: str):
     assert any(test_file_name == doc.metadata.get("file_name") for doc in docs.values())
 
 
+@pytest.mark.skipif(
+    not base_url or not api_key, reason="No platform base url or api key set"
+)
+def test_upload_file_from_url_with_custom_metadata(
+    remote_file: Tuple[str, str], index_name: str
+):
+    index = LlamaCloudIndex.create_index(
+        name=index_name,
+        project_name=project_name,
+        organization_id=organization_id,
+        api_key=api_key,
+        base_url=base_url,
+    )
+
+    # Define a URL to a file for testing
+    custom_metadata = {"foo": "bar"}
+    test_file_url, test_file_name = remote_file
+
+    # Upload the file from the URL
+    file_id = index.upload_file_from_url(
+        file_name=test_file_name,
+        url=test_file_url,
+        custom_metadata=custom_metadata,
+        verbose=True,
+    )
+    assert file_id is not None
+
+    # Verify the file is part of the index
+    docs = index.ref_doc_info
+    assert any(test_file_name == doc.metadata.get("file_name") for doc in docs.values())
+
+
 @pytest.mark.skipif(
     not base_url or not api_key, reason="No platform base url or api key set"
 )
@@ -507,6 +576,33 @@ async def test_async_upload_file_from_url(
     await index.await_for_completion()
 
 
+@pytest.mark.skipif(
+    not base_url or not api_key, reason="No platform base url or api key set"
+)
+@pytest.mark.asyncio
+async def test_async_upload_file_from_url_with_custom_metadata(
+    remote_file: Tuple[str, str], index_name: str
+):
+    index = await LlamaCloudIndex.acreate_index(
+        name=index_name,
+        project_name=project_name,
+        api_key=api_key,
+        base_url=base_url,
+    )
+
+    custom_metadata = {"foo": "bar"}
+    test_file_url, test_file_name = remote_file
+    file_id = await index.aupload_file_from_url(
+        file_name=test_file_name,
+        url=test_file_url,
+        custom_metadata=custom_metadata,
+        verbose=True,
+    )
+    assert file_id is not None
+
+    await index.await_for_completion()
+
+
 @pytest.mark.skipif(
     not base_url or not api_key, reason="No platform base url or api key set"
 )
@@ -525,6 +621,29 @@ async def test_async_index_from_file(index_name: str, local_file: str):
     await index.await_for_completion()
 
 
+@pytest.mark.skipif(
+    not base_url or not api_key, reason="No platform base url or api key set"
+)
+@pytest.mark.asyncio
+async def test_async_index_from_file_with_custom_metadata(
+    index_name: str, local_file: str
+):
+    index = await LlamaCloudIndex.acreate_index(
+        name=index_name,
+        project_name=project_name,
+        api_key=api_key,
+        base_url=base_url,
+    )
+
+    custom_metadata = {"foo": "bar"}
+    file_id = await index.aupload_file(
+        file_path=local_file, custom_metadata=custom_metadata, verbose=True
+    )
+    assert file_id is not None
+
+    await index.await_for_completion()
+
+
 class DummySchema(BaseModel):
     source: str
 

py/unit_tests/beta/agent/test_agent_data_schema.py

@@ -2,6 +2,7 @@ from datetime import datetime
 import json
 from pathlib import Path
 from typing import Any, Dict, Optional
+import uuid
 
 import pytest
 from llama_cloud import ExtractRun, File
@@ -434,6 +435,7 @@ def create_extract_run(
             "extraction_agent_id": "extraction-agent-123",
             "config": {},
             "status": "SUCCESS",
+            "project_id": str(uuid.uuid4()),
             "from_ui": False,
         }
     )

py/unit_tests/data/capacitor.json

@@ -112,5 +112,6 @@
       "num_output_tokens": 3440
     }
   },
+  "project_id": "77bdc79f-fb69-49ae-a783-fcc573eec7ce",
   "from_ui": false
 }

py/unit_tests/testing_utils/test_fake_server.py

@@ -0,0 +1,311 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+import pytest
+from llama_cloud import ExtractConfig
+from llama_cloud.types import ExtractMode
+from llama_cloud.core.api_error import ApiError
+from llama_cloud_services.extract import LlamaExtract
+from llama_cloud_services.parse import LlamaParse
+from llama_cloud_services.beta.agent_data import AsyncAgentDataClient
+from llama_cloud_services.testing_utils import FakeLlamaCloudServer
+from llama_cloud_services.testing_utils._deterministic import hash_schema
+from pydantic import BaseModel, Field
+
+
+class Receipt(BaseModel):
+    merchant: str = Field(description="Vendor name")
+    total: float = Field(description="Grand total")
+
+
+@pytest.fixture(autouse=True)
+def _fake_env(monkeypatch: pytest.MonkeyPatch) -> None:
+    monkeypatch.setenv("LLAMA_CLOUD_API_KEY", "unit-test-key")
+    monkeypatch.setenv("LLAMA_CLOUD_BASE_URL", FakeLlamaCloudServer.DEFAULT_BASE_URL)
+
+
+@pytest.fixture
+def fake_server() -> FakeLlamaCloudServer:
+    with FakeLlamaCloudServer() as server:
+        yield server
+
+
+def _write_sample_file(tmp_path: Path, name: str, content: str) -> Path:
+    target = tmp_path / name
+    target.write_text(content)
+    return target
+
+
+def test_stateless_extract_is_deterministic(
+    fake_server: FakeLlamaCloudServer, tmp_path: Path
+) -> None:
+    extractor = LlamaExtract(api_key="unit-test-key", verify=False)
+    config = ExtractConfig(extraction_mode=ExtractMode.FAST)
+    sample_path = _write_sample_file(
+        tmp_path, "receipt.txt", "Merchant: Lunar Bistro\nTotal: 123.45"
+    )
+
+    first_run = extractor.extract(Receipt, config, sample_path)
+    second_run = extractor.extract(Receipt, config, sample_path)
+
+    assert first_run.status.value == "SUCCESS"
+    assert second_run.data == first_run.data
+    assert "merchant" in first_run.data
+    assert fake_server.extract.stateless_run.called
+
+
+def test_agent_flow_uploads_and_processes_files(
+    fake_server: FakeLlamaCloudServer, tmp_path: Path
+) -> None:
+    extractor = LlamaExtract(api_key="unit-test-key", verify=False)
+    config = ExtractConfig(extraction_mode=ExtractMode.FAST)
+    agent = extractor.create_agent(
+        name="unit-test-agent", data_schema=Receipt, config=config
+    )
+
+    sample_path = _write_sample_file(
+        tmp_path, "contract.pdf", "Agreement between parties."
+    )
+    run = agent.extract(sample_path)
+
+    assert run.status.value == "SUCCESS"
+    assert "merchant" in run.data
+
+    uploaded_bytes = fake_server.files.read(run.file.id)
+    assert uploaded_bytes.startswith(b"Agreement")
+    assert fake_server.extract.agent_job.called
+    assert fake_server.extract.agent_run.called
+
+
+def test_parse_load_data_returns_documents(
+    fake_server: FakeLlamaCloudServer, tmp_path: Path
+) -> None:
+    parser = LlamaParse(
+        api_key="unit-test-key", base_url=FakeLlamaCloudServer.DEFAULT_BASE_URL
+    )
+    sample_path = _write_sample_file(
+        tmp_path, "report.pdf", "Executive summary of quarterly goals."
+    )
+
+    documents = parser.load_data(sample_path)
+
+    assert documents
+    assert "(page 1)" in documents[0].text
+
+
+@pytest.mark.asyncio
+async def test_agent_data_create(fake_server: FakeLlamaCloudServer):
+    with fake_server as _:
+        client = AsyncAgentDataClient(
+            Receipt,
+            collection="extracted_data",
+            deployment_name="extraction_agent",
+            token="fake-api-key",
+        )
+        data = Receipt(merchant="Test Inc", total=1000)
+        item = await client.create_item(data)
+        assert item.id == hash_schema(data)[:7]
+        assert item.data.merchant == data.merchant and item.data.total == data.total
+        assert item.collection == "extracted_data"
+        assert item.deployment_name == "extraction_agent"
+
+
+@pytest.mark.asyncio
+async def test_agent_data_update(fake_server: FakeLlamaCloudServer):
+    with fake_server as _:
+        client = AsyncAgentDataClient(
+            Receipt,
+            collection="extracted_data",
+            deployment_name="extraction_agent",
+            token="fake-api-key",
+        )
+        data = Receipt(merchant="Test Inc", total=1000)
+        item = await client.create_item(data)
+        assert item.id is not None
+        updated_data = Receipt(merchant="Testing Inc", total=1100)
+        updated_item = await client.update_item(item_id=item.id, data=updated_data)
+        # ensure that the data actually changed
+        assert (
+            updated_item.data.merchant == updated_data.merchant
+            and updated_item.data.total == updated_data.total
+        )
+        # make sure nothing else changed
+        assert updated_item.id == item.id
+        assert updated_item.collection == item.collection
+        assert updated_item.deployment_name == item.deployment_name
+
+
+@pytest.mark.asyncio
+async def test_agent_data_search(fake_server: FakeLlamaCloudServer):
+    with fake_server as _:
+        client = AsyncAgentDataClient(
+            Receipt,
+            collection="extracted_data",
+            deployment_name="extraction_agent",
+            token="fake-api-key",
+        )
+        data1 = Receipt(merchant="Test Inc", total=1000)
+        data2 = Receipt(merchant="Test Inc", total=1300)
+        data3 = Receipt(merchant="Testing Inc", total=1100)
+        item1 = await client.create_item(data1)
+        item2 = await client.create_item(data2)
+        item3 = await client.create_item(data3)
+        result = await client.search(filter={"merchant": {"eq": "Test Inc"}})
+        assert result.total == 2
+        assert any(item.id == item1.id for item in result.items) and any(
+            item.id == item2.id for item in result.items
+        )
+        assert all(item.data.merchant == "Test Inc" for item in result.items)
+        result1 = await client.search(filter={"total": {"lt": 1200}})
+        assert result.total == 2
+        assert any(item.id == item1.id for item in result1.items) and any(
+            item.id == item3.id for item in result1.items
+        )
+        assert all(item.data.total < 1200 for item in result1.items)
+
+
+@pytest.mark.asyncio
+async def test_agent_data_aggregate(fake_server: FakeLlamaCloudServer):
+    with fake_server as _:
+        client = AsyncAgentDataClient(
+            Receipt,
+            collection="extracted_data",
+            deployment_name="extraction_agent",
+            token="fake-api-key",
+        )
+        data1 = Receipt(merchant="Test Inc", total=1000)
+        data2 = Receipt(merchant="Test Inc", total=1300)
+        data3 = Receipt(merchant="Testing Inc", total=1100)
+        await client.create_item(data1)
+        await client.create_item(data2)
+        await client.create_item(data3)
+        result = await client.aggregate(
+            filter={"merchant": {"eq": "Test Inc"}},
+            group_by=["merchant"],
+            count=True,
+        )
+        # filtering for 'Test Inc' on merchant means that only data with 'Test Inc' are left, meaning that there is only one group of data for merchant, i.e. the 'Test Inc' group
+        assert len(result.items) == 1
+        assert result.items[0].count == 2
+        assert result.items[0].first_item is not None
+        assert result.items[0].first_item.merchant == data1.merchant
+        assert result.items[0].first_item.total == data1.total
+        assert result.items[0].group_key == {"merchant": "Test Inc"}
+        result = await client.aggregate(
+            group_by=["merchant"],
+            count=True,
+        )
+        assert len(result.items) == 2
+        assert result.items[0].count == 2
+        assert result.items[0].first_item is not None
+        assert result.items[0].first_item.merchant == data1.merchant
+        assert result.items[0].first_item.total == data1.total
+        assert result.items[0].group_key == {"merchant": "Test Inc"}
+        assert result.items[1].count == 1
+        assert result.items[1].first_item is not None
+        assert result.items[1].first_item.merchant == data3.merchant
+        assert result.items[1].first_item.total == data3.total
+        assert result.items[1].group_key == {"merchant": "Testing Inc"}
+
+
+@pytest.mark.asyncio
+async def test_agent_data_get(fake_server: FakeLlamaCloudServer):
+    with fake_server as _:
+        client = AsyncAgentDataClient(
+            Receipt,
+            collection="extracted_data",
+            deployment_name="extraction_agent",
+            token="fake-api-key",
+        )
+        data1 = Receipt(merchant="Test Inc", total=1000)
+        data2 = Receipt(merchant="Test Inc", total=1300)
+        item1 = await client.create_item(data1)
+        assert item1.id is not None
+        item2 = await client.create_item(data2)
+        assert item2.id is not None
+        item = await client.get_item(item1.id)
+        assert item.collection == item1.collection
+        assert item.deployment_name == item1.deployment_name
+        assert item.data.merchant == data1.merchant
+        assert item.data.total == data1.total
+        # using this pattern instead of `with pytest.raise` for more granual control over the error itself
+        try:
+            notitem = await client.get_item(item2.id + "thisdoesnotexist")
+            e = None
+        except ApiError as err:
+            e = err
+            notitem = None
+        assert notitem is None
+        assert e is not None
+        assert e.status_code == 404
+        assert e.body == {"detail": f"No data with ID: {item2.id+'thisdoesnotexist'}"}
+
+
+@pytest.mark.asyncio
+async def test_agent_data_delete_by_id(fake_server: FakeLlamaCloudServer):
+    with fake_server as _:
+        client = AsyncAgentDataClient(
+            Receipt,
+            collection="extracted_data",
+            deployment_name="extraction_agent",
+            token="fake-api-key",
+        )
+        data = Receipt(merchant="Test Inc", total=1300)
+        item = await client.create_item(data)
+        assert item.id is not None
+        await client.delete_item(item.id)
+        # using this pattern instead of `with pytest.raise` for more granual control over the error itself
+        try:
+            notitem = await client.get_item(item.id)
+            e = None
+        except ApiError as err:
+            e = err
+            notitem = None
+        assert notitem is None
+        assert e is not None
+        assert e.status_code == 404
+        assert e.body == {"detail": f"No data with ID: {item.id}"}
+        # using this pattern instead of `with pytest.raise` for more granual control over the error itself
+        try:
+            await client.delete_item(item.id)
+            e = None
+        except ApiError as err:
+            e = err
+        assert e is not None
+        assert e.status_code == 404
+        assert e.body == {"detail": f"No data with ID: {item.id}"}
+
+
+@pytest.mark.asyncio
+async def test_agent_data_delete_by_query(fake_server: FakeLlamaCloudServer):
+    with fake_server as _:
+        client = AsyncAgentDataClient(
+            Receipt,
+            collection="extracted_data",
+            deployment_name="extraction_agent",
+            token="fake-api-key",
+        )
+        data1 = Receipt(merchant="Test Inc", total=1000)
+        data2 = Receipt(merchant="Test Inc", total=1300)
+        data3 = Receipt(merchant="Testing Inc", total=1100)
+        item1 = await client.create_item(data1)
+        item2 = await client.create_item(data2)
+        item3 = await client.create_item(data3)
+        result = await client.delete(filter={"merchant": {"eq": "Test Inc"}})
+        assert result == 2
+        for item in (item1, item2):
+            assert item.id is not None
+            try:
+                notitem = await client.get_item(item.id)
+                e = None
+            except ApiError as err:
+                e = err
+                notitem = None
+            assert notitem is None
+            assert e is not None
+            assert e.status_code == 404
+            assert e.body == {"detail": f"No data with ID: {item.id}"}
+        assert item3.id is not None
+        itemfound = await client.get_item(item3.id)
+        assert itemfound.id == item3.id

py/uv.lock

@@ -3,7 +3,8 @@ revision = 3
 requires-python = ">=3.9, <4.0"
 resolution-markers = [
     "python_full_version >= '3.14'",
-    "python_full_version >= '3.11' and python_full_version < '3.14'",
+    "python_full_version >= '3.12' and python_full_version < '3.14'",
+    "python_full_version == '3.11.*'",
     "python_full_version == '3.10.*'",
     "python_full_version < '3.10'",
 ]
@@ -220,7 +221,8 @@ name = "argon2-cffi-bindings"
 version = "25.1.0"
 source = { registry = "https://pypi.org/simple" }
 resolution-markers = [
-    "python_full_version >= '3.11' and python_full_version < '3.14'",
+    "python_full_version >= '3.12' and python_full_version < '3.14'",
+    "python_full_version == '3.11.*'",
     "python_full_version == '3.10.*'",
     "python_full_version < '3.10'",
 ]
@@ -589,7 +591,8 @@ version = "8.2.1"
 source = { registry = "https://pypi.org/simple" }
 resolution-markers = [
     "python_full_version >= '3.14'",
-    "python_full_version >= '3.11' and python_full_version < '3.14'",
+    "python_full_version >= '3.12' and python_full_version < '3.14'",
+    "python_full_version == '3.11.*'",
     "python_full_version == '3.10.*'",
 ]
 dependencies = [
@@ -720,6 +723,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/33/6b/e0547afaf41bf2c42e52430072fa5658766e3d65bd4b03a563d1b6336f57/distlib-0.4.0-py2.py3-none-any.whl", hash = "sha256:9659f7d87e46584a30b5780e43ac7a2143098441670ff0a49d5f9034c54a6c16", size = 469047, upload-time = "2025-07-17T16:51:58.613Z" },
 ]
 
+[[package]]
+name = "et-xmlfile"
+version = "2.0.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/d3/38/af70d7ab1ae9d4da450eeec1fa3918940a5fafb9055e934af8d6eb0c2313/et_xmlfile-2.0.0.tar.gz", hash = "sha256:dab3f4764309081ce75662649be815c4c9081e88f0837825f90fd28317d4da54", size = 17234, upload-time = "2024-10-25T17:25:40.039Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/c1/8b/5fe2cc11fee489817272089c4203e679c63b570a5aaeb18d852ae3cbba6a/et_xmlfile-2.0.0-py3-none-any.whl", hash = "sha256:7a91720bc756843502c3b7504c77b8fe44217c85c537d85037f0f536151b2caa", size = 18059, upload-time = "2024-10-25T17:25:39.051Z" },
+]
+
 [[package]]
 name = "eval-type-backport"
 version = "0.2.2"
@@ -1120,7 +1132,8 @@ version = "8.37.0"
 source = { registry = "https://pypi.org/simple" }
 resolution-markers = [
     "python_full_version >= '3.14'",
-    "python_full_version >= '3.11' and python_full_version < '3.14'",
+    "python_full_version >= '3.12' and python_full_version < '3.14'",
+    "python_full_version == '3.11.*'",
     "python_full_version == '3.10.*'",
 ]
 dependencies = [
@@ -1582,21 +1595,21 @@ wheels = [
 
 [[package]]
 name = "llama-cloud"
-version = "0.1.43"
+version = "0.1.44"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "certifi" },
     { name = "httpx" },
     { name = "pydantic" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/9b/33/33a8bd3a617c071caf450ca2627969f8b28272d0692f122997c10a32247e/llama_cloud-0.1.43.tar.gz", hash = "sha256:00429f05aea515449d90cde91ef3ed3687fcd93e46f6246d08cbea02f9b397a9", size = 112992, upload-time = "2025-10-02T21:55:38.355Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/54/eb/16e31fb0fc4df91b08fa19cc3f28ac6e3c7d4df0bcbb71dd2bf596e9586f/llama_cloud-0.1.44.tar.gz", hash = "sha256:276a2b4f94463da037431ca3063331b3b6be398bbfb003113ee76b7c2a873b53", size = 120502, upload-time = "2025-11-04T00:51:58.578Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/2b/54/559a67542396d5660a71115b29e0160e9dd784e570e1f4ef55ad22bf5b39/llama_cloud-0.1.43-py3-none-any.whl", hash = "sha256:540605d4dd13c6536a3b75cd4d04b211f29b16d17faee9381e3793a651f1dec1", size = 311460, upload-time = "2025-10-02T21:55:37.282Z" },
+    { url = "https://files.pythonhosted.org/packages/69/0a/fabe54c21d5927d626550cb9560a20e51e42468355f5f0fb300f84806e28/llama_cloud-0.1.44-py3-none-any.whl", hash = "sha256:dfdcc4932353711fc8639f14261cbb54a88139b7790ebdd3ed4fde29bbbc0b88", size = 332779, upload-time = "2025-11-04T00:51:57.371Z" },
 ]
 
 [[package]]
 name = "llama-cloud-services"
-version = "0.6.73"
+version = "0.6.81"
 source = { editable = "." }
 dependencies = [
     { name = "click", version = "8.1.8", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.10'" },
@@ -1608,6 +1621,7 @@ dependencies = [
     { name = "platformdirs" },
     { name = "pydantic" },
     { name = "python-dotenv" },
+    { name = "respx" },
     { name = "tenacity" },
 ]
 
@@ -1620,7 +1634,11 @@ dev = [
     { name = "ipython", version = "8.37.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.10'" },
     { name = "jupyter" },
     { name = "mypy" },
+    { name = "openpyxl" },
+    { name = "pandas" },
     { name = "pre-commit" },
+    { name = "pyarrow", version = "21.0.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.10'" },
+    { name = "pyarrow", version = "22.0.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.10'" },
     { name = "pydantic-settings" },
     { name = "pytest" },
     { name = "pytest-asyncio" },
@@ -1631,12 +1649,13 @@ dev = [
 requires-dist = [
     { name = "click", specifier = ">=8.1.7,<9" },
     { name = "eval-type-backport", marker = "python_full_version < '3.10'", specifier = ">=0.2.0,<0.3" },
-    { name = "llama-cloud", specifier = "==0.1.43" },
+    { name = "llama-cloud", specifier = "==0.1.44" },
     { name = "llama-index-core", specifier = ">=0.12.0" },
     { name = "packaging", specifier = ">=23.0" },
     { name = "platformdirs", specifier = ">=4.3.7,<5" },
     { name = "pydantic", specifier = ">=2.8,!=2.10" },
     { name = "python-dotenv", specifier = ">=1.0.1,<2" },
+    { name = "respx", extras = ["tests"], specifier = ">=0.22.0" },
     { name = "tenacity", specifier = ">=8.5.0,<10.0" },
 ]
 
@@ -1648,7 +1667,10 @@ dev = [
     { name = "ipython", specifier = ">=8.12.3,<9" },
     { name = "jupyter", specifier = ">=1.1.1,<2" },
     { name = "mypy", specifier = ">=1.14.1,<2" },
+    { name = "openpyxl" },
+    { name = "pandas" },
     { name = "pre-commit", specifier = "==3.2.0" },
+    { name = "pyarrow" },
     { name = "pydantic-settings", specifier = ">=2.10.1" },
     { name = "pytest", specifier = ">=8.0.0,<9" },
     { name = "pytest-asyncio" },
@@ -2098,7 +2120,8 @@ version = "3.5"
 source = { registry = "https://pypi.org/simple" }
 resolution-markers = [
     "python_full_version >= '3.14'",
-    "python_full_version >= '3.11' and python_full_version < '3.14'",
+    "python_full_version >= '3.12' and python_full_version < '3.14'",
+    "python_full_version == '3.11.*'",
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/6c/4f/ccdb8ad3a38e583f214547fd2f7ff1fc160c43a75af88e6aec213404b96a/networkx-3.5.tar.gz", hash = "sha256:d4c6f9cf81f52d69230866796b82afbccdec3db7ae4fbd1b65ea750feed50037", size = 2471065, upload-time = "2025-05-29T11:35:07.804Z" }
 wheels = [
@@ -2284,7 +2307,8 @@ version = "2.3.2"
 source = { registry = "https://pypi.org/simple" }
 resolution-markers = [
     "python_full_version >= '3.14'",
-    "python_full_version >= '3.11' and python_full_version < '3.14'",
+    "python_full_version >= '3.12' and python_full_version < '3.14'",
+    "python_full_version == '3.11.*'",
 ]
 sdist = { url = "https://files.pythonhosted.org/packages/37/7d/3fec4199c5ffb892bed55cff901e4f39a58c81df9c44c280499e92cad264/numpy-2.3.2.tar.gz", hash = "sha256:e0486a11ec30cdecb53f184d496d1c6a20786c81e55e41640270130056f8ee48", size = 20489306, upload-time = "2025-07-24T21:32:07.553Z" }
 wheels = [
@@ -2363,6 +2387,18 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/78/e3/6690b3f85a05506733c7e90b577e4762517404ea78bab2ca3a5cb1aeb78d/numpy-2.3.2-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:6936aff90dda378c09bea075af0d9c675fe3a977a9d2402f95a87f440f59f619", size = 12977811, upload-time = "2025-07-24T21:29:18.234Z" },
 ]
 
+[[package]]
+name = "openpyxl"
+version = "3.1.5"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "et-xmlfile" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/3d/f9/88d94a75de065ea32619465d2f77b29a0469500e99012523b91cc4141cd1/openpyxl-3.1.5.tar.gz", hash = "sha256:cf0e3cf56142039133628b5acffe8ef0c12bc902d2aadd3e0fe5878dc08d1050", size = 186464, upload-time = "2024-06-28T14:03:44.161Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/c0/da/977ded879c29cbd04de313843e76868e6e13408a94ed6b987245dc7c8506/openpyxl-3.1.5-py2.py3-none-any.whl", hash = "sha256:5282c12b107bffeef825f4617dc029afaf41d0ea60823bbb665ef3079dc79de2", size = 250910, upload-time = "2024-06-28T14:03:41.161Z" },
+]
+
 [[package]]
 name = "orderly-set"
 version = "5.5.0"
@@ -2390,6 +2426,76 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/20/12/38679034af332785aac8774540895e234f4d07f7545804097de4b666afd8/packaging-25.0-py3-none-any.whl", hash = "sha256:29572ef2b1f17581046b3a2227d5c611fb25ec70ca1ba8554b24b0e69331a484", size = 66469, upload-time = "2025-04-19T11:48:57.875Z" },
 ]
 
+[[package]]
+name = "pandas"
+version = "2.3.3"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "numpy", version = "2.0.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.10'" },
+    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version == '3.10.*'" },
+    { name = "numpy", version = "2.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
+    { name = "python-dateutil" },
+    { name = "pytz" },
+    { name = "tzdata" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/33/01/d40b85317f86cf08d853a4f495195c73815fdf205eef3993821720274518/pandas-2.3.3.tar.gz", hash = "sha256:e05e1af93b977f7eafa636d043f9f94c7ee3ac81af99c13508215942e64c993b", size = 4495223, upload-time = "2025-09-29T23:34:51.853Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/3d/f7/f425a00df4fcc22b292c6895c6831c0c8ae1d9fac1e024d16f98a9ce8749/pandas-2.3.3-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:376c6446ae31770764215a6c937f72d917f214b43560603cd60da6408f183b6c", size = 11555763, upload-time = "2025-09-29T23:16:53.287Z" },
+    { url = "https://files.pythonhosted.org/packages/13/4f/66d99628ff8ce7857aca52fed8f0066ce209f96be2fede6cef9f84e8d04f/pandas-2.3.3-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:e19d192383eab2f4ceb30b412b22ea30690c9e618f78870357ae1d682912015a", size = 10801217, upload-time = "2025-09-29T23:17:04.522Z" },
+    { url = "https://files.pythonhosted.org/packages/1d/03/3fc4a529a7710f890a239cc496fc6d50ad4a0995657dccc1d64695adb9f4/pandas-2.3.3-cp310-cp310-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5caf26f64126b6c7aec964f74266f435afef1c1b13da3b0636c7518a1fa3e2b1", size = 12148791, upload-time = "2025-09-29T23:17:18.444Z" },
+    { url = "https://files.pythonhosted.org/packages/40/a8/4dac1f8f8235e5d25b9955d02ff6f29396191d4e665d71122c3722ca83c5/pandas-2.3.3-cp310-cp310-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:dd7478f1463441ae4ca7308a70e90b33470fa593429f9d4c578dd00d1fa78838", size = 12769373, upload-time = "2025-09-29T23:17:35.846Z" },
+    { url = "https://files.pythonhosted.org/packages/df/91/82cc5169b6b25440a7fc0ef3a694582418d875c8e3ebf796a6d6470aa578/pandas-2.3.3-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:4793891684806ae50d1288c9bae9330293ab4e083ccd1c5e383c34549c6e4250", size = 13200444, upload-time = "2025-09-29T23:17:49.341Z" },
+    { url = "https://files.pythonhosted.org/packages/10/ae/89b3283800ab58f7af2952704078555fa60c807fff764395bb57ea0b0dbd/pandas-2.3.3-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:28083c648d9a99a5dd035ec125d42439c6c1c525098c58af0fc38dd1a7a1b3d4", size = 13858459, upload-time = "2025-09-29T23:18:03.722Z" },
+    { url = "https://files.pythonhosted.org/packages/85/72/530900610650f54a35a19476eca5104f38555afccda1aa11a92ee14cb21d/pandas-2.3.3-cp310-cp310-win_amd64.whl", hash = "sha256:503cf027cf9940d2ceaa1a93cfb5f8c8c7e6e90720a2850378f0b3f3b1e06826", size = 11346086, upload-time = "2025-09-29T23:18:18.505Z" },
+    { url = "https://files.pythonhosted.org/packages/c1/fa/7ac648108144a095b4fb6aa3de1954689f7af60a14cf25583f4960ecb878/pandas-2.3.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:602b8615ebcc4a0c1751e71840428ddebeb142ec02c786e8ad6b1ce3c8dec523", size = 11578790, upload-time = "2025-09-29T23:18:30.065Z" },
+    { url = "https://files.pythonhosted.org/packages/9b/35/74442388c6cf008882d4d4bdfc4109be87e9b8b7ccd097ad1e7f006e2e95/pandas-2.3.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:8fe25fc7b623b0ef6b5009149627e34d2a4657e880948ec3c840e9402e5c1b45", size = 10833831, upload-time = "2025-09-29T23:38:56.071Z" },
+    { url = "https://files.pythonhosted.org/packages/fe/e4/de154cbfeee13383ad58d23017da99390b91d73f8c11856f2095e813201b/pandas-2.3.3-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b468d3dad6ff947df92dcb32ede5b7bd41a9b3cceef0a30ed925f6d01fb8fa66", size = 12199267, upload-time = "2025-09-29T23:18:41.627Z" },
+    { url = "https://files.pythonhosted.org/packages/bf/c9/63f8d545568d9ab91476b1818b4741f521646cbdd151c6efebf40d6de6f7/pandas-2.3.3-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b98560e98cb334799c0b07ca7967ac361a47326e9b4e5a7dfb5ab2b1c9d35a1b", size = 12789281, upload-time = "2025-09-29T23:18:56.834Z" },
+    { url = "https://files.pythonhosted.org/packages/f2/00/a5ac8c7a0e67fd1a6059e40aa08fa1c52cc00709077d2300e210c3ce0322/pandas-2.3.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:1d37b5848ba49824e5c30bedb9c830ab9b7751fd049bc7914533e01c65f79791", size = 13240453, upload-time = "2025-09-29T23:19:09.247Z" },
+    { url = "https://files.pythonhosted.org/packages/27/4d/5c23a5bc7bd209231618dd9e606ce076272c9bc4f12023a70e03a86b4067/pandas-2.3.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:db4301b2d1f926ae677a751eb2bd0e8c5f5319c9cb3f88b0becbbb0b07b34151", size = 13890361, upload-time = "2025-09-29T23:19:25.342Z" },
+    { url = "https://files.pythonhosted.org/packages/8e/59/712db1d7040520de7a4965df15b774348980e6df45c129b8c64d0dbe74ef/pandas-2.3.3-cp311-cp311-win_amd64.whl", hash = "sha256:f086f6fe114e19d92014a1966f43a3e62285109afe874f067f5abbdcbb10e59c", size = 11348702, upload-time = "2025-09-29T23:19:38.296Z" },
+    { url = "https://files.pythonhosted.org/packages/9c/fb/231d89e8637c808b997d172b18e9d4a4bc7bf31296196c260526055d1ea0/pandas-2.3.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:6d21f6d74eb1725c2efaa71a2bfc661a0689579b58e9c0ca58a739ff0b002b53", size = 11597846, upload-time = "2025-09-29T23:19:48.856Z" },
+    { url = "https://files.pythonhosted.org/packages/5c/bd/bf8064d9cfa214294356c2d6702b716d3cf3bb24be59287a6a21e24cae6b/pandas-2.3.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:3fd2f887589c7aa868e02632612ba39acb0b8948faf5cc58f0850e165bd46f35", size = 10729618, upload-time = "2025-09-29T23:39:08.659Z" },
+    { url = "https://files.pythonhosted.org/packages/57/56/cf2dbe1a3f5271370669475ead12ce77c61726ffd19a35546e31aa8edf4e/pandas-2.3.3-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:ecaf1e12bdc03c86ad4a7ea848d66c685cb6851d807a26aa245ca3d2017a1908", size = 11737212, upload-time = "2025-09-29T23:19:59.765Z" },
+    { url = "https://files.pythonhosted.org/packages/e5/63/cd7d615331b328e287d8233ba9fdf191a9c2d11b6af0c7a59cfcec23de68/pandas-2.3.3-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:b3d11d2fda7eb164ef27ffc14b4fcab16a80e1ce67e9f57e19ec0afaf715ba89", size = 12362693, upload-time = "2025-09-29T23:20:14.098Z" },
+    { url = "https://files.pythonhosted.org/packages/a6/de/8b1895b107277d52f2b42d3a6806e69cfef0d5cf1d0ba343470b9d8e0a04/pandas-2.3.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:a68e15f780eddf2b07d242e17a04aa187a7ee12b40b930bfdd78070556550e98", size = 12771002, upload-time = "2025-09-29T23:20:26.76Z" },
+    { url = "https://files.pythonhosted.org/packages/87/21/84072af3187a677c5893b170ba2c8fbe450a6ff911234916da889b698220/pandas-2.3.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:371a4ab48e950033bcf52b6527eccb564f52dc826c02afd9a1bc0ab731bba084", size = 13450971, upload-time = "2025-09-29T23:20:41.344Z" },
+    { url = "https://files.pythonhosted.org/packages/86/41/585a168330ff063014880a80d744219dbf1dd7a1c706e75ab3425a987384/pandas-2.3.3-cp312-cp312-win_amd64.whl", hash = "sha256:a16dcec078a01eeef8ee61bf64074b4e524a2a3f4b3be9326420cabe59c4778b", size = 10992722, upload-time = "2025-09-29T23:20:54.139Z" },
+    { url = "https://files.pythonhosted.org/packages/cd/4b/18b035ee18f97c1040d94debd8f2e737000ad70ccc8f5513f4eefad75f4b/pandas-2.3.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:56851a737e3470de7fa88e6131f41281ed440d29a9268dcbf0002da5ac366713", size = 11544671, upload-time = "2025-09-29T23:21:05.024Z" },
+    { url = "https://files.pythonhosted.org/packages/31/94/72fac03573102779920099bcac1c3b05975c2cb5f01eac609faf34bed1ca/pandas-2.3.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:bdcd9d1167f4885211e401b3036c0c8d9e274eee67ea8d0758a256d60704cfe8", size = 10680807, upload-time = "2025-09-29T23:21:15.979Z" },
+    { url = "https://files.pythonhosted.org/packages/16/87/9472cf4a487d848476865321de18cc8c920b8cab98453ab79dbbc98db63a/pandas-2.3.3-cp313-cp313-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e32e7cc9af0f1cc15548288a51a3b681cc2a219faa838e995f7dc53dbab1062d", size = 11709872, upload-time = "2025-09-29T23:21:27.165Z" },
+    { url = "https://files.pythonhosted.org/packages/15/07/284f757f63f8a8d69ed4472bfd85122bd086e637bf4ed09de572d575a693/pandas-2.3.3-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:318d77e0e42a628c04dc56bcef4b40de67918f7041c2b061af1da41dcff670ac", size = 12306371, upload-time = "2025-09-29T23:21:40.532Z" },
+    { url = "https://files.pythonhosted.org/packages/33/81/a3afc88fca4aa925804a27d2676d22dcd2031c2ebe08aabd0ae55b9ff282/pandas-2.3.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:4e0a175408804d566144e170d0476b15d78458795bb18f1304fb94160cabf40c", size = 12765333, upload-time = "2025-09-29T23:21:55.77Z" },
+    { url = "https://files.pythonhosted.org/packages/8d/0f/b4d4ae743a83742f1153464cf1a8ecfafc3ac59722a0b5c8602310cb7158/pandas-2.3.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:93c2d9ab0fc11822b5eece72ec9587e172f63cff87c00b062f6e37448ced4493", size = 13418120, upload-time = "2025-09-29T23:22:10.109Z" },
+    { url = "https://files.pythonhosted.org/packages/4f/c7/e54682c96a895d0c808453269e0b5928a07a127a15704fedb643e9b0a4c8/pandas-2.3.3-cp313-cp313-win_amd64.whl", hash = "sha256:f8bfc0e12dc78f777f323f55c58649591b2cd0c43534e8355c51d3fede5f4dee", size = 10993991, upload-time = "2025-09-29T23:25:04.889Z" },
+    { url = "https://files.pythonhosted.org/packages/f9/ca/3f8d4f49740799189e1395812f3bf23b5e8fc7c190827d55a610da72ce55/pandas-2.3.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:75ea25f9529fdec2d2e93a42c523962261e567d250b0013b16210e1d40d7c2e5", size = 12048227, upload-time = "2025-09-29T23:22:24.343Z" },
+    { url = "https://files.pythonhosted.org/packages/0e/5a/f43efec3e8c0cc92c4663ccad372dbdff72b60bdb56b2749f04aa1d07d7e/pandas-2.3.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:74ecdf1d301e812db96a465a525952f4dde225fdb6d8e5a521d47e1f42041e21", size = 11411056, upload-time = "2025-09-29T23:22:37.762Z" },
+    { url = "https://files.pythonhosted.org/packages/46/b1/85331edfc591208c9d1a63a06baa67b21d332e63b7a591a5ba42a10bb507/pandas-2.3.3-cp313-cp313t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6435cb949cb34ec11cc9860246ccb2fdc9ecd742c12d3304989017d53f039a78", size = 11645189, upload-time = "2025-09-29T23:22:51.688Z" },
+    { url = "https://files.pythonhosted.org/packages/44/23/78d645adc35d94d1ac4f2a3c4112ab6f5b8999f4898b8cdf01252f8df4a9/pandas-2.3.3-cp313-cp313t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:900f47d8f20860de523a1ac881c4c36d65efcb2eb850e6948140fa781736e110", size = 12121912, upload-time = "2025-09-29T23:23:05.042Z" },
+    { url = "https://files.pythonhosted.org/packages/53/da/d10013df5e6aaef6b425aa0c32e1fc1f3e431e4bcabd420517dceadce354/pandas-2.3.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:a45c765238e2ed7d7c608fc5bc4a6f88b642f2f01e70c0c23d2224dd21829d86", size = 12712160, upload-time = "2025-09-29T23:23:28.57Z" },
+    { url = "https://files.pythonhosted.org/packages/bd/17/e756653095a083d8a37cbd816cb87148debcfcd920129b25f99dd8d04271/pandas-2.3.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:c4fc4c21971a1a9f4bdb4c73978c7f7256caa3e62b323f70d6cb80db583350bc", size = 13199233, upload-time = "2025-09-29T23:24:24.876Z" },
+    { url = "https://files.pythonhosted.org/packages/04/fd/74903979833db8390b73b3a8a7d30d146d710bd32703724dd9083950386f/pandas-2.3.3-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:ee15f284898e7b246df8087fc82b87b01686f98ee67d85a17b7ab44143a3a9a0", size = 11540635, upload-time = "2025-09-29T23:25:52.486Z" },
+    { url = "https://files.pythonhosted.org/packages/21/00/266d6b357ad5e6d3ad55093a7e8efc7dd245f5a842b584db9f30b0f0a287/pandas-2.3.3-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:1611aedd912e1ff81ff41c745822980c49ce4a7907537be8692c8dbc31924593", size = 10759079, upload-time = "2025-09-29T23:26:33.204Z" },
+    { url = "https://files.pythonhosted.org/packages/ca/05/d01ef80a7a3a12b2f8bbf16daba1e17c98a2f039cbc8e2f77a2c5a63d382/pandas-2.3.3-cp314-cp314-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6d2cefc361461662ac48810cb14365a365ce864afe85ef1f447ff5a1e99ea81c", size = 11814049, upload-time = "2025-09-29T23:27:15.384Z" },
+    { url = "https://files.pythonhosted.org/packages/15/b2/0e62f78c0c5ba7e3d2c5945a82456f4fac76c480940f805e0b97fcbc2f65/pandas-2.3.3-cp314-cp314-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:ee67acbbf05014ea6c763beb097e03cd629961c8a632075eeb34247120abcb4b", size = 12332638, upload-time = "2025-09-29T23:27:51.625Z" },
+    { url = "https://files.pythonhosted.org/packages/c5/33/dd70400631b62b9b29c3c93d2feee1d0964dc2bae2e5ad7a6c73a7f25325/pandas-2.3.3-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:c46467899aaa4da076d5abc11084634e2d197e9460643dd455ac3db5856b24d6", size = 12886834, upload-time = "2025-09-29T23:28:21.289Z" },
+    { url = "https://files.pythonhosted.org/packages/d3/18/b5d48f55821228d0d2692b34fd5034bb185e854bdb592e9c640f6290e012/pandas-2.3.3-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:6253c72c6a1d990a410bc7de641d34053364ef8bcd3126f7e7450125887dffe3", size = 13409925, upload-time = "2025-09-29T23:28:58.261Z" },
+    { url = "https://files.pythonhosted.org/packages/a6/3d/124ac75fcd0ecc09b8fdccb0246ef65e35b012030defb0e0eba2cbbbe948/pandas-2.3.3-cp314-cp314-win_amd64.whl", hash = "sha256:1b07204a219b3b7350abaae088f451860223a52cfb8a6c53358e7948735158e5", size = 11109071, upload-time = "2025-09-29T23:32:27.484Z" },
+    { url = "https://files.pythonhosted.org/packages/89/9c/0e21c895c38a157e0faa1fb64587a9226d6dd46452cac4532d80c3c4a244/pandas-2.3.3-cp314-cp314t-macosx_10_13_x86_64.whl", hash = "sha256:2462b1a365b6109d275250baaae7b760fd25c726aaca0054649286bcfbb3e8ec", size = 12048504, upload-time = "2025-09-29T23:29:31.47Z" },
+    { url = "https://files.pythonhosted.org/packages/d7/82/b69a1c95df796858777b68fbe6a81d37443a33319761d7c652ce77797475/pandas-2.3.3-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:0242fe9a49aa8b4d78a4fa03acb397a58833ef6199e9aa40a95f027bb3a1b6e7", size = 11410702, upload-time = "2025-09-29T23:29:54.591Z" },
+    { url = "https://files.pythonhosted.org/packages/f9/88/702bde3ba0a94b8c73a0181e05144b10f13f29ebfc2150c3a79062a8195d/pandas-2.3.3-cp314-cp314t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:a21d830e78df0a515db2b3d2f5570610f5e6bd2e27749770e8bb7b524b89b450", size = 11634535, upload-time = "2025-09-29T23:30:21.003Z" },
+    { url = "https://files.pythonhosted.org/packages/a4/1e/1bac1a839d12e6a82ec6cb40cda2edde64a2013a66963293696bbf31fbbb/pandas-2.3.3-cp314-cp314t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2e3ebdb170b5ef78f19bfb71b0dc5dc58775032361fa188e814959b74d726dd5", size = 12121582, upload-time = "2025-09-29T23:30:43.391Z" },
+    { url = "https://files.pythonhosted.org/packages/44/91/483de934193e12a3b1d6ae7c8645d083ff88dec75f46e827562f1e4b4da6/pandas-2.3.3-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:d051c0e065b94b7a3cea50eb1ec32e912cd96dba41647eb24104b6c6c14c5788", size = 12699963, upload-time = "2025-09-29T23:31:10.009Z" },
+    { url = "https://files.pythonhosted.org/packages/70/44/5191d2e4026f86a2a109053e194d3ba7a31a2d10a9c2348368c63ed4e85a/pandas-2.3.3-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:3869faf4bd07b3b66a9f462417d0ca3a9df29a9f6abd5d0d0dbab15dac7abe87", size = 13202175, upload-time = "2025-09-29T23:31:59.173Z" },
+    { url = "https://files.pythonhosted.org/packages/56/b4/52eeb530a99e2a4c55ffcd352772b599ed4473a0f892d127f4147cf0f88e/pandas-2.3.3-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:c503ba5216814e295f40711470446bc3fd00f0faea8a086cbc688808e26f92a2", size = 11567720, upload-time = "2025-09-29T23:33:06.209Z" },
+    { url = "https://files.pythonhosted.org/packages/48/4a/2d8b67632a021bced649ba940455ed441ca854e57d6e7658a6024587b083/pandas-2.3.3-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:a637c5cdfa04b6d6e2ecedcb81fc52ffb0fd78ce2ebccc9ea964df9f658de8c8", size = 10810302, upload-time = "2025-09-29T23:33:35.846Z" },
+    { url = "https://files.pythonhosted.org/packages/13/e6/d2465010ee0569a245c975dc6967b801887068bc893e908239b1f4b6c1ac/pandas-2.3.3-cp39-cp39-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:854d00d556406bffe66a4c0802f334c9ad5a96b4f1f868adf036a21b11ef13ff", size = 12154874, upload-time = "2025-09-29T23:33:49.939Z" },
+    { url = "https://files.pythonhosted.org/packages/1f/18/aae8c0aa69a386a3255940e9317f793808ea79d0a525a97a903366bb2569/pandas-2.3.3-cp39-cp39-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:bf1f8a81d04ca90e32a0aceb819d34dbd378a98bf923b6398b9a3ec0bf44de29", size = 12790141, upload-time = "2025-09-29T23:34:05.655Z" },
+    { url = "https://files.pythonhosted.org/packages/f7/26/617f98de789de00c2a444fbe6301bb19e66556ac78cff933d2c98f62f2b4/pandas-2.3.3-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:23ebd657a4d38268c7dfbdf089fbc31ea709d82e4923c5ffd4fbd5747133ce73", size = 13208697, upload-time = "2025-09-29T23:34:21.835Z" },
+    { url = "https://files.pythonhosted.org/packages/b9/fb/25709afa4552042bd0e15717c75e9b4a2294c3dc4f7e6ea50f03c5136600/pandas-2.3.3-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:5554c929ccc317d41a5e3d1234f3be588248e61f08a74dd17c9eabb535777dc9", size = 13879233, upload-time = "2025-09-29T23:34:35.079Z" },
+    { url = "https://files.pythonhosted.org/packages/98/af/7be05277859a7bc399da8ba68b88c96b27b48740b6cf49688899c6eb4176/pandas-2.3.3-cp39-cp39-win_amd64.whl", hash = "sha256:d3e28b3e83862ccf4d85ff19cf8c20b2ae7e503881711ff2d534dc8f761131aa", size = 11359119, upload-time = "2025-09-29T23:34:46.339Z" },
+]
+
 [[package]]
 name = "pandocfilters"
 version = "1.5.1"
@@ -2735,6 +2841,122 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/8e/37/efad0257dc6e593a18957422533ff0f87ede7c9c6ea010a2177d738fb82f/pure_eval-0.2.3-py3-none-any.whl", hash = "sha256:1db8e35b67b3d218d818ae653e27f06c3aa420901fa7b081ca98cbedc874e0d0", size = 11842, upload-time = "2024-07-21T12:58:20.04Z" },
 ]
 
+[[package]]
+name = "pyarrow"
+version = "21.0.0"
+source = { registry = "https://pypi.org/simple" }
+resolution-markers = [
+    "python_full_version < '3.10'",
+]
+sdist = { url = "https://files.pythonhosted.org/packages/ef/c2/ea068b8f00905c06329a3dfcd40d0fcc2b7d0f2e355bdb25b65e0a0e4cd4/pyarrow-21.0.0.tar.gz", hash = "sha256:5051f2dccf0e283ff56335760cbc8622cf52264d67e359d5569541ac11b6d5bc", size = 1133487, upload-time = "2025-07-18T00:57:31.761Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/17/d9/110de31880016e2afc52d8580b397dbe47615defbf09ca8cf55f56c62165/pyarrow-21.0.0-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:e563271e2c5ff4d4a4cbeb2c83d5cf0d4938b891518e676025f7268c6fe5fe26", size = 31196837, upload-time = "2025-07-18T00:54:34.755Z" },
+    { url = "https://files.pythonhosted.org/packages/df/5f/c1c1997613abf24fceb087e79432d24c19bc6f7259cab57c2c8e5e545fab/pyarrow-21.0.0-cp310-cp310-macosx_12_0_x86_64.whl", hash = "sha256:fee33b0ca46f4c85443d6c450357101e47d53e6c3f008d658c27a2d020d44c79", size = 32659470, upload-time = "2025-07-18T00:54:38.329Z" },
+    { url = "https://files.pythonhosted.org/packages/3e/ed/b1589a777816ee33ba123ba1e4f8f02243a844fed0deec97bde9fb21a5cf/pyarrow-21.0.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:7be45519b830f7c24b21d630a31d48bcebfd5d4d7f9d3bdb49da9cdf6d764edb", size = 41055619, upload-time = "2025-07-18T00:54:42.172Z" },
+    { url = "https://files.pythonhosted.org/packages/44/28/b6672962639e85dc0ac36f71ab3a8f5f38e01b51343d7aa372a6b56fa3f3/pyarrow-21.0.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:26bfd95f6bff443ceae63c65dc7e048670b7e98bc892210acba7e4995d3d4b51", size = 42733488, upload-time = "2025-07-18T00:54:47.132Z" },
+    { url = "https://files.pythonhosted.org/packages/f8/cc/de02c3614874b9089c94eac093f90ca5dfa6d5afe45de3ba847fd950fdf1/pyarrow-21.0.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:bd04ec08f7f8bd113c55868bd3fc442a9db67c27af098c5f814a3091e71cc61a", size = 43329159, upload-time = "2025-07-18T00:54:51.686Z" },
+    { url = "https://files.pythonhosted.org/packages/a6/3e/99473332ac40278f196e105ce30b79ab8affab12f6194802f2593d6b0be2/pyarrow-21.0.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:9b0b14b49ac10654332a805aedfc0147fb3469cbf8ea951b3d040dab12372594", size = 45050567, upload-time = "2025-07-18T00:54:56.679Z" },
+    { url = "https://files.pythonhosted.org/packages/7b/f5/c372ef60593d713e8bfbb7e0c743501605f0ad00719146dc075faf11172b/pyarrow-21.0.0-cp310-cp310-win_amd64.whl", hash = "sha256:9d9f8bcb4c3be7738add259738abdeddc363de1b80e3310e04067aa1ca596634", size = 26217959, upload-time = "2025-07-18T00:55:00.482Z" },
+    { url = "https://files.pythonhosted.org/packages/94/dc/80564a3071a57c20b7c32575e4a0120e8a330ef487c319b122942d665960/pyarrow-21.0.0-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:c077f48aab61738c237802836fc3844f85409a46015635198761b0d6a688f87b", size = 31243234, upload-time = "2025-07-18T00:55:03.812Z" },
+    { url = "https://files.pythonhosted.org/packages/ea/cc/3b51cb2db26fe535d14f74cab4c79b191ed9a8cd4cbba45e2379b5ca2746/pyarrow-21.0.0-cp311-cp311-macosx_12_0_x86_64.whl", hash = "sha256:689f448066781856237eca8d1975b98cace19b8dd2ab6145bf49475478bcaa10", size = 32714370, upload-time = "2025-07-18T00:55:07.495Z" },
+    { url = "https://files.pythonhosted.org/packages/24/11/a4431f36d5ad7d83b87146f515c063e4d07ef0b7240876ddb885e6b44f2e/pyarrow-21.0.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:479ee41399fcddc46159a551705b89c05f11e8b8cb8e968f7fec64f62d91985e", size = 41135424, upload-time = "2025-07-18T00:55:11.461Z" },
+    { url = "https://files.pythonhosted.org/packages/74/dc/035d54638fc5d2971cbf1e987ccd45f1091c83bcf747281cf6cc25e72c88/pyarrow-21.0.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:40ebfcb54a4f11bcde86bc586cbd0272bac0d516cfa539c799c2453768477569", size = 42823810, upload-time = "2025-07-18T00:55:16.301Z" },
+    { url = "https://files.pythonhosted.org/packages/2e/3b/89fced102448a9e3e0d4dded1f37fa3ce4700f02cdb8665457fcc8015f5b/pyarrow-21.0.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:8d58d8497814274d3d20214fbb24abcad2f7e351474357d552a8d53bce70c70e", size = 43391538, upload-time = "2025-07-18T00:55:23.82Z" },
+    { url = "https://files.pythonhosted.org/packages/fb/bb/ea7f1bd08978d39debd3b23611c293f64a642557e8141c80635d501e6d53/pyarrow-21.0.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:585e7224f21124dd57836b1530ac8f2df2afc43c861d7bf3d58a4870c42ae36c", size = 45120056, upload-time = "2025-07-18T00:55:28.231Z" },
+    { url = "https://files.pythonhosted.org/packages/6e/0b/77ea0600009842b30ceebc3337639a7380cd946061b620ac1a2f3cb541e2/pyarrow-21.0.0-cp311-cp311-win_amd64.whl", hash = "sha256:555ca6935b2cbca2c0e932bedd853e9bc523098c39636de9ad4693b5b1df86d6", size = 26220568, upload-time = "2025-07-18T00:55:32.122Z" },
+    { url = "https://files.pythonhosted.org/packages/ca/d4/d4f817b21aacc30195cf6a46ba041dd1be827efa4a623cc8bf39a1c2a0c0/pyarrow-21.0.0-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:3a302f0e0963db37e0a24a70c56cf91a4faa0bca51c23812279ca2e23481fccd", size = 31160305, upload-time = "2025-07-18T00:55:35.373Z" },
+    { url = "https://files.pythonhosted.org/packages/a2/9c/dcd38ce6e4b4d9a19e1d36914cb8e2b1da4e6003dd075474c4cfcdfe0601/pyarrow-21.0.0-cp312-cp312-macosx_12_0_x86_64.whl", hash = "sha256:b6b27cf01e243871390474a211a7922bfbe3bda21e39bc9160daf0da3fe48876", size = 32684264, upload-time = "2025-07-18T00:55:39.303Z" },
+    { url = "https://files.pythonhosted.org/packages/4f/74/2a2d9f8d7a59b639523454bec12dba35ae3d0a07d8ab529dc0809f74b23c/pyarrow-21.0.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:e72a8ec6b868e258a2cd2672d91f2860ad532d590ce94cdf7d5e7ec674ccf03d", size = 41108099, upload-time = "2025-07-18T00:55:42.889Z" },
+    { url = "https://files.pythonhosted.org/packages/ad/90/2660332eeb31303c13b653ea566a9918484b6e4d6b9d2d46879a33ab0622/pyarrow-21.0.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:b7ae0bbdc8c6674259b25bef5d2a1d6af5d39d7200c819cf99e07f7dfef1c51e", size = 42829529, upload-time = "2025-07-18T00:55:47.069Z" },
+    { url = "https://files.pythonhosted.org/packages/33/27/1a93a25c92717f6aa0fca06eb4700860577d016cd3ae51aad0e0488ac899/pyarrow-21.0.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:58c30a1729f82d201627c173d91bd431db88ea74dcaa3885855bc6203e433b82", size = 43367883, upload-time = "2025-07-18T00:55:53.069Z" },
+    { url = "https://files.pythonhosted.org/packages/05/d9/4d09d919f35d599bc05c6950095e358c3e15148ead26292dfca1fb659b0c/pyarrow-21.0.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:072116f65604b822a7f22945a7a6e581cfa28e3454fdcc6939d4ff6090126623", size = 45133802, upload-time = "2025-07-18T00:55:57.714Z" },
+    { url = "https://files.pythonhosted.org/packages/71/30/f3795b6e192c3ab881325ffe172e526499eb3780e306a15103a2764916a2/pyarrow-21.0.0-cp312-cp312-win_amd64.whl", hash = "sha256:cf56ec8b0a5c8c9d7021d6fd754e688104f9ebebf1bf4449613c9531f5346a18", size = 26203175, upload-time = "2025-07-18T00:56:01.364Z" },
+    { url = "https://files.pythonhosted.org/packages/16/ca/c7eaa8e62db8fb37ce942b1ea0c6d7abfe3786ca193957afa25e71b81b66/pyarrow-21.0.0-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:e99310a4ebd4479bcd1964dff9e14af33746300cb014aa4a3781738ac63baf4a", size = 31154306, upload-time = "2025-07-18T00:56:04.42Z" },
+    { url = "https://files.pythonhosted.org/packages/ce/e8/e87d9e3b2489302b3a1aea709aaca4b781c5252fcb812a17ab6275a9a484/pyarrow-21.0.0-cp313-cp313-macosx_12_0_x86_64.whl", hash = "sha256:d2fe8e7f3ce329a71b7ddd7498b3cfac0eeb200c2789bd840234f0dc271a8efe", size = 32680622, upload-time = "2025-07-18T00:56:07.505Z" },
+    { url = "https://files.pythonhosted.org/packages/84/52/79095d73a742aa0aba370c7942b1b655f598069489ab387fe47261a849e1/pyarrow-21.0.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:f522e5709379d72fb3da7785aa489ff0bb87448a9dc5a75f45763a795a089ebd", size = 41104094, upload-time = "2025-07-18T00:56:10.994Z" },
+    { url = "https://files.pythonhosted.org/packages/89/4b/7782438b551dbb0468892a276b8c789b8bbdb25ea5c5eb27faadd753e037/pyarrow-21.0.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:69cbbdf0631396e9925e048cfa5bce4e8c3d3b41562bbd70c685a8eb53a91e61", size = 42825576, upload-time = "2025-07-18T00:56:15.569Z" },
+    { url = "https://files.pythonhosted.org/packages/b3/62/0f29de6e0a1e33518dec92c65be0351d32d7ca351e51ec5f4f837a9aab91/pyarrow-21.0.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:731c7022587006b755d0bdb27626a1a3bb004bb56b11fb30d98b6c1b4718579d", size = 43368342, upload-time = "2025-07-18T00:56:19.531Z" },
+    { url = "https://files.pythonhosted.org/packages/90/c7/0fa1f3f29cf75f339768cc698c8ad4ddd2481c1742e9741459911c9ac477/pyarrow-21.0.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:dc56bc708f2d8ac71bd1dcb927e458c93cec10b98eb4120206a4091db7b67b99", size = 45131218, upload-time = "2025-07-18T00:56:23.347Z" },
+    { url = "https://files.pythonhosted.org/packages/01/63/581f2076465e67b23bc5a37d4a2abff8362d389d29d8105832e82c9c811c/pyarrow-21.0.0-cp313-cp313-win_amd64.whl", hash = "sha256:186aa00bca62139f75b7de8420f745f2af12941595bbbfa7ed3870ff63e25636", size = 26087551, upload-time = "2025-07-18T00:56:26.758Z" },
+    { url = "https://files.pythonhosted.org/packages/c9/ab/357d0d9648bb8241ee7348e564f2479d206ebe6e1c47ac5027c2e31ecd39/pyarrow-21.0.0-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:a7a102574faa3f421141a64c10216e078df467ab9576684d5cd696952546e2da", size = 31290064, upload-time = "2025-07-18T00:56:30.214Z" },
+    { url = "https://files.pythonhosted.org/packages/3f/8a/5685d62a990e4cac2043fc76b4661bf38d06efed55cf45a334b455bd2759/pyarrow-21.0.0-cp313-cp313t-macosx_12_0_x86_64.whl", hash = "sha256:1e005378c4a2c6db3ada3ad4c217b381f6c886f0a80d6a316fe586b90f77efd7", size = 32727837, upload-time = "2025-07-18T00:56:33.935Z" },
+    { url = "https://files.pythonhosted.org/packages/fc/de/c0828ee09525c2bafefd3e736a248ebe764d07d0fd762d4f0929dbc516c9/pyarrow-21.0.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:65f8e85f79031449ec8706b74504a316805217b35b6099155dd7e227eef0d4b6", size = 41014158, upload-time = "2025-07-18T00:56:37.528Z" },
+    { url = "https://files.pythonhosted.org/packages/6e/26/a2865c420c50b7a3748320b614f3484bfcde8347b2639b2b903b21ce6a72/pyarrow-21.0.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:3a81486adc665c7eb1a2bde0224cfca6ceaba344a82a971ef059678417880eb8", size = 42667885, upload-time = "2025-07-18T00:56:41.483Z" },
+    { url = "https://files.pythonhosted.org/packages/0a/f9/4ee798dc902533159250fb4321267730bc0a107d8c6889e07c3add4fe3a5/pyarrow-21.0.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:fc0d2f88b81dcf3ccf9a6ae17f89183762c8a94a5bdcfa09e05cfe413acf0503", size = 43276625, upload-time = "2025-07-18T00:56:48.002Z" },
+    { url = "https://files.pythonhosted.org/packages/5a/da/e02544d6997037a4b0d22d8e5f66bc9315c3671371a8b18c79ade1cefe14/pyarrow-21.0.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:6299449adf89df38537837487a4f8d3bd91ec94354fdd2a7d30bc11c48ef6e79", size = 44951890, upload-time = "2025-07-18T00:56:52.568Z" },
+    { url = "https://files.pythonhosted.org/packages/e5/4e/519c1bc1876625fe6b71e9a28287c43ec2f20f73c658b9ae1d485c0c206e/pyarrow-21.0.0-cp313-cp313t-win_amd64.whl", hash = "sha256:222c39e2c70113543982c6b34f3077962b44fca38c0bd9e68bb6781534425c10", size = 26371006, upload-time = "2025-07-18T00:56:56.379Z" },
+    { url = "https://files.pythonhosted.org/packages/3e/cc/ce4939f4b316457a083dc5718b3982801e8c33f921b3c98e7a93b7c7491f/pyarrow-21.0.0-cp39-cp39-macosx_12_0_arm64.whl", hash = "sha256:a7f6524e3747e35f80744537c78e7302cd41deee8baa668d56d55f77d9c464b3", size = 31211248, upload-time = "2025-07-18T00:56:59.7Z" },
+    { url = "https://files.pythonhosted.org/packages/1f/c2/7a860931420d73985e2f340f06516b21740c15b28d24a0e99a900bb27d2b/pyarrow-21.0.0-cp39-cp39-macosx_12_0_x86_64.whl", hash = "sha256:203003786c9fd253ebcafa44b03c06983c9c8d06c3145e37f1b76a1f317aeae1", size = 32676896, upload-time = "2025-07-18T00:57:03.884Z" },
+    { url = "https://files.pythonhosted.org/packages/68/a8/197f989b9a75e59b4ca0db6a13c56f19a0ad8a298c68da9cc28145e0bb97/pyarrow-21.0.0-cp39-cp39-manylinux_2_28_aarch64.whl", hash = "sha256:3b4d97e297741796fead24867a8dabf86c87e4584ccc03167e4a811f50fdf74d", size = 41067862, upload-time = "2025-07-18T00:57:07.587Z" },
+    { url = "https://files.pythonhosted.org/packages/fa/82/6ecfa89487b35aa21accb014b64e0a6b814cc860d5e3170287bf5135c7d8/pyarrow-21.0.0-cp39-cp39-manylinux_2_28_x86_64.whl", hash = "sha256:898afce396b80fdda05e3086b4256f8677c671f7b1d27a6976fa011d3fd0a86e", size = 42747508, upload-time = "2025-07-18T00:57:13.917Z" },
+    { url = "https://files.pythonhosted.org/packages/3b/b7/ba252f399bbf3addc731e8643c05532cf32e74cebb5e32f8f7409bc243cf/pyarrow-21.0.0-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:067c66ca29aaedae08218569a114e413b26e742171f526e828e1064fcdec13f4", size = 43345293, upload-time = "2025-07-18T00:57:19.828Z" },
+    { url = "https://files.pythonhosted.org/packages/ff/0a/a20819795bd702b9486f536a8eeb70a6aa64046fce32071c19ec8230dbaa/pyarrow-21.0.0-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:0c4e75d13eb76295a49e0ea056eb18dbd87d81450bfeb8afa19a7e5a75ae2ad7", size = 45060670, upload-time = "2025-07-18T00:57:24.477Z" },
+    { url = "https://files.pythonhosted.org/packages/10/15/6b30e77872012bbfe8265d42a01d5b3c17ef0ac0f2fae531ad91b6a6c02e/pyarrow-21.0.0-cp39-cp39-win_amd64.whl", hash = "sha256:cdc4c17afda4dab2a9c0b79148a43a7f4e1094916b3e18d8975bfd6d6d52241f", size = 26227521, upload-time = "2025-07-18T00:57:29.119Z" },
+]
+
+[[package]]
+name = "pyarrow"
+version = "22.0.0"
+source = { registry = "https://pypi.org/simple" }
+resolution-markers = [
+    "python_full_version >= '3.14'",
+    "python_full_version >= '3.12' and python_full_version < '3.14'",
+    "python_full_version == '3.11.*'",
+    "python_full_version == '3.10.*'",
+]
+sdist = { url = "https://files.pythonhosted.org/packages/30/53/04a7fdc63e6056116c9ddc8b43bc28c12cdd181b85cbeadb79278475f3ae/pyarrow-22.0.0.tar.gz", hash = "sha256:3d600dc583260d845c7d8a6db540339dd883081925da2bd1c5cb808f720b3cd9", size = 1151151, upload-time = "2025-10-24T12:30:00.762Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/d9/9b/cb3f7e0a345353def531ca879053e9ef6b9f38ed91aebcf68b09ba54dec0/pyarrow-22.0.0-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:77718810bd3066158db1e95a63c160ad7ce08c6b0710bc656055033e39cdad88", size = 34223968, upload-time = "2025-10-24T10:03:31.21Z" },
+    { url = "https://files.pythonhosted.org/packages/6c/41/3184b8192a120306270c5307f105b70320fdaa592c99843c5ef78aaefdcf/pyarrow-22.0.0-cp310-cp310-macosx_12_0_x86_64.whl", hash = "sha256:44d2d26cda26d18f7af7db71453b7b783788322d756e81730acb98f24eb90ace", size = 35942085, upload-time = "2025-10-24T10:03:38.146Z" },
+    { url = "https://files.pythonhosted.org/packages/d9/3d/a1eab2f6f08001f9fb714b8ed5cfb045e2fe3e3e3c0c221f2c9ed1e6d67d/pyarrow-22.0.0-cp310-cp310-manylinux_2_28_aarch64.whl", hash = "sha256:b9d71701ce97c95480fecb0039ec5bb889e75f110da72005743451339262f4ce", size = 44964613, upload-time = "2025-10-24T10:03:46.516Z" },
+    { url = "https://files.pythonhosted.org/packages/46/46/a1d9c24baf21cfd9ce994ac820a24608decf2710521b29223d4334985127/pyarrow-22.0.0-cp310-cp310-manylinux_2_28_x86_64.whl", hash = "sha256:710624ab925dc2b05a6229d47f6f0dac1c1155e6ed559be7109f684eba048a48", size = 47627059, upload-time = "2025-10-24T10:03:55.353Z" },
+    { url = "https://files.pythonhosted.org/packages/3a/4c/f711acb13075c1391fd54bc17e078587672c575f8de2a6e62509af026dcf/pyarrow-22.0.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:f963ba8c3b0199f9d6b794c90ec77545e05eadc83973897a4523c9e8d84e9340", size = 47947043, upload-time = "2025-10-24T10:04:05.408Z" },
+    { url = "https://files.pythonhosted.org/packages/4e/70/1f3180dd7c2eab35c2aca2b29ace6c519f827dcd4cfeb8e0dca41612cf7a/pyarrow-22.0.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:bd0d42297ace400d8febe55f13fdf46e86754842b860c978dfec16f081e5c653", size = 50206505, upload-time = "2025-10-24T10:04:15.786Z" },
+    { url = "https://files.pythonhosted.org/packages/80/07/fea6578112c8c60ffde55883a571e4c4c6bc7049f119d6b09333b5cc6f73/pyarrow-22.0.0-cp310-cp310-win_amd64.whl", hash = "sha256:00626d9dc0f5ef3a75fe63fd68b9c7c8302d2b5bbc7f74ecaedba83447a24f84", size = 28101641, upload-time = "2025-10-24T10:04:22.57Z" },
+    { url = "https://files.pythonhosted.org/packages/2e/b7/18f611a8cdc43417f9394a3ccd3eace2f32183c08b9eddc3d17681819f37/pyarrow-22.0.0-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:3e294c5eadfb93d78b0763e859a0c16d4051fc1c5231ae8956d61cb0b5666f5a", size = 34272022, upload-time = "2025-10-24T10:04:28.973Z" },
+    { url = "https://files.pythonhosted.org/packages/26/5c/f259e2526c67eb4b9e511741b19870a02363a47a35edbebc55c3178db22d/pyarrow-22.0.0-cp311-cp311-macosx_12_0_x86_64.whl", hash = "sha256:69763ab2445f632d90b504a815a2a033f74332997052b721002298ed6de40f2e", size = 35995834, upload-time = "2025-10-24T10:04:35.467Z" },
+    { url = "https://files.pythonhosted.org/packages/50/8d/281f0f9b9376d4b7f146913b26fac0aa2829cd1ee7e997f53a27411bbb92/pyarrow-22.0.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:b41f37cabfe2463232684de44bad753d6be08a7a072f6a83447eeaf0e4d2a215", size = 45030348, upload-time = "2025-10-24T10:04:43.366Z" },
+    { url = "https://files.pythonhosted.org/packages/f5/e5/53c0a1c428f0976bf22f513d79c73000926cb00b9c138d8e02daf2102e18/pyarrow-22.0.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:35ad0f0378c9359b3f297299c3309778bb03b8612f987399a0333a560b43862d", size = 47699480, upload-time = "2025-10-24T10:04:51.486Z" },
+    { url = "https://files.pythonhosted.org/packages/95/e1/9dbe4c465c3365959d183e6345d0a8d1dc5b02ca3f8db4760b3bc834cf25/pyarrow-22.0.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:8382ad21458075c2e66a82a29d650f963ce51c7708c7c0ff313a8c206c4fd5e8", size = 48011148, upload-time = "2025-10-24T10:04:59.585Z" },
+    { url = "https://files.pythonhosted.org/packages/c5/b4/7caf5d21930061444c3cf4fa7535c82faf5263e22ce43af7c2759ceb5b8b/pyarrow-22.0.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:1a812a5b727bc09c3d7ea072c4eebf657c2f7066155506ba31ebf4792f88f016", size = 50276964, upload-time = "2025-10-24T10:05:08.175Z" },
+    { url = "https://files.pythonhosted.org/packages/ae/f3/cec89bd99fa3abf826f14d4e53d3d11340ce6f6af4d14bdcd54cd83b6576/pyarrow-22.0.0-cp311-cp311-win_amd64.whl", hash = "sha256:ec5d40dd494882704fb876c16fa7261a69791e784ae34e6b5992e977bd2e238c", size = 28106517, upload-time = "2025-10-24T10:05:14.314Z" },
+    { url = "https://files.pythonhosted.org/packages/af/63/ba23862d69652f85b615ca14ad14f3bcfc5bf1b99ef3f0cd04ff93fdad5a/pyarrow-22.0.0-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:bea79263d55c24a32b0d79c00a1c58bb2ee5f0757ed95656b01c0fb310c5af3d", size = 34211578, upload-time = "2025-10-24T10:05:21.583Z" },
+    { url = "https://files.pythonhosted.org/packages/b1/d0/f9ad86fe809efd2bcc8be32032fa72e8b0d112b01ae56a053006376c5930/pyarrow-22.0.0-cp312-cp312-macosx_12_0_x86_64.whl", hash = "sha256:12fe549c9b10ac98c91cf791d2945e878875d95508e1a5d14091a7aaa66d9cf8", size = 35989906, upload-time = "2025-10-24T10:05:29.485Z" },
+    { url = "https://files.pythonhosted.org/packages/b4/a8/f910afcb14630e64d673f15904ec27dd31f1e009b77033c365c84e8c1e1d/pyarrow-22.0.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:334f900ff08ce0423407af97e6c26ad5d4e3b0763645559ece6fbf3747d6a8f5", size = 45021677, upload-time = "2025-10-24T10:05:38.274Z" },
+    { url = "https://files.pythonhosted.org/packages/13/95/aec81f781c75cd10554dc17a25849c720d54feafb6f7847690478dcf5ef8/pyarrow-22.0.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:c6c791b09c57ed76a18b03f2631753a4960eefbbca80f846da8baefc6491fcfe", size = 47726315, upload-time = "2025-10-24T10:05:47.314Z" },
+    { url = "https://files.pythonhosted.org/packages/bb/d4/74ac9f7a54cfde12ee42734ea25d5a3c9a45db78f9def949307a92720d37/pyarrow-22.0.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:c3200cb41cdbc65156e5f8c908d739b0dfed57e890329413da2748d1a2cd1a4e", size = 47990906, upload-time = "2025-10-24T10:05:58.254Z" },
+    { url = "https://files.pythonhosted.org/packages/2e/71/fedf2499bf7a95062eafc989ace56572f3343432570e1c54e6599d5b88da/pyarrow-22.0.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:ac93252226cf288753d8b46280f4edf3433bf9508b6977f8dd8526b521a1bbb9", size = 50306783, upload-time = "2025-10-24T10:06:08.08Z" },
+    { url = "https://files.pythonhosted.org/packages/68/ed/b202abd5a5b78f519722f3d29063dda03c114711093c1995a33b8e2e0f4b/pyarrow-22.0.0-cp312-cp312-win_amd64.whl", hash = "sha256:44729980b6c50a5f2bfcc2668d36c569ce17f8b17bccaf470c4313dcbbf13c9d", size = 27972883, upload-time = "2025-10-24T10:06:14.204Z" },
+    { url = "https://files.pythonhosted.org/packages/a6/d6/d0fac16a2963002fc22c8fa75180a838737203d558f0ed3b564c4a54eef5/pyarrow-22.0.0-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:e6e95176209257803a8b3d0394f21604e796dadb643d2f7ca21b66c9c0b30c9a", size = 34204629, upload-time = "2025-10-24T10:06:20.274Z" },
+    { url = "https://files.pythonhosted.org/packages/c6/9c/1d6357347fbae062ad3f17082f9ebc29cc733321e892c0d2085f42a2212b/pyarrow-22.0.0-cp313-cp313-macosx_12_0_x86_64.whl", hash = "sha256:001ea83a58024818826a9e3f89bf9310a114f7e26dfe404a4c32686f97bd7901", size = 35985783, upload-time = "2025-10-24T10:06:27.301Z" },
+    { url = "https://files.pythonhosted.org/packages/ff/c0/782344c2ce58afbea010150df07e3a2f5fdad299cd631697ae7bd3bac6e3/pyarrow-22.0.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:ce20fe000754f477c8a9125543f1936ea5b8867c5406757c224d745ed033e691", size = 45020999, upload-time = "2025-10-24T10:06:35.387Z" },
+    { url = "https://files.pythonhosted.org/packages/1b/8b/5362443737a5307a7b67c1017c42cd104213189b4970bf607e05faf9c525/pyarrow-22.0.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:e0a15757fccb38c410947df156f9749ae4a3c89b2393741a50521f39a8cf202a", size = 47724601, upload-time = "2025-10-24T10:06:43.551Z" },
+    { url = "https://files.pythonhosted.org/packages/69/4d/76e567a4fc2e190ee6072967cb4672b7d9249ac59ae65af2d7e3047afa3b/pyarrow-22.0.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:cedb9dd9358e4ea1d9bce3665ce0797f6adf97ff142c8e25b46ba9cdd508e9b6", size = 48001050, upload-time = "2025-10-24T10:06:52.284Z" },
+    { url = "https://files.pythonhosted.org/packages/01/5e/5653f0535d2a1aef8223cee9d92944cb6bccfee5cf1cd3f462d7cb022790/pyarrow-22.0.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:252be4a05f9d9185bb8c18e83764ebcfea7185076c07a7a662253af3a8c07941", size = 50307877, upload-time = "2025-10-24T10:07:02.405Z" },
+    { url = "https://files.pythonhosted.org/packages/2d/f8/1d0bd75bf9328a3b826e24a16e5517cd7f9fbf8d34a3184a4566ef5a7f29/pyarrow-22.0.0-cp313-cp313-win_amd64.whl", hash = "sha256:a4893d31e5ef780b6edcaf63122df0f8d321088bb0dee4c8c06eccb1ca28d145", size = 27977099, upload-time = "2025-10-24T10:08:07.259Z" },
+    { url = "https://files.pythonhosted.org/packages/90/81/db56870c997805bf2b0f6eeeb2d68458bf4654652dccdcf1bf7a42d80903/pyarrow-22.0.0-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:f7fe3dbe871294ba70d789be16b6e7e52b418311e166e0e3cba9522f0f437fb1", size = 34336685, upload-time = "2025-10-24T10:07:11.47Z" },
+    { url = "https://files.pythonhosted.org/packages/1c/98/0727947f199aba8a120f47dfc229eeb05df15bcd7a6f1b669e9f882afc58/pyarrow-22.0.0-cp313-cp313t-macosx_12_0_x86_64.whl", hash = "sha256:ba95112d15fd4f1105fb2402c4eab9068f0554435e9b7085924bcfaac2cc306f", size = 36032158, upload-time = "2025-10-24T10:07:18.626Z" },
+    { url = "https://files.pythonhosted.org/packages/96/b4/9babdef9c01720a0785945c7cf550e4acd0ebcd7bdd2e6f0aa7981fa85e2/pyarrow-22.0.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:c064e28361c05d72eed8e744c9605cbd6d2bb7481a511c74071fd9b24bc65d7d", size = 44892060, upload-time = "2025-10-24T10:07:26.002Z" },
+    { url = "https://files.pythonhosted.org/packages/f8/ca/2f8804edd6279f78a37062d813de3f16f29183874447ef6d1aadbb4efa0f/pyarrow-22.0.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:6f9762274496c244d951c819348afbcf212714902742225f649cf02823a6a10f", size = 47504395, upload-time = "2025-10-24T10:07:34.09Z" },
+    { url = "https://files.pythonhosted.org/packages/b9/f0/77aa5198fd3943682b2e4faaf179a674f0edea0d55d326d83cb2277d9363/pyarrow-22.0.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:a9d9ffdc2ab696f6b15b4d1f7cec6658e1d788124418cb30030afbae31c64746", size = 48066216, upload-time = "2025-10-24T10:07:43.528Z" },
+    { url = "https://files.pythonhosted.org/packages/79/87/a1937b6e78b2aff18b706d738c9e46ade5bfcf11b294e39c87706a0089ac/pyarrow-22.0.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:ec1a15968a9d80da01e1d30349b2b0d7cc91e96588ee324ce1b5228175043e95", size = 50288552, upload-time = "2025-10-24T10:07:53.519Z" },
+    { url = "https://files.pythonhosted.org/packages/60/ae/b5a5811e11f25788ccfdaa8f26b6791c9807119dffcf80514505527c384c/pyarrow-22.0.0-cp313-cp313t-win_amd64.whl", hash = "sha256:bba208d9c7decf9961998edf5c65e3ea4355d5818dd6cd0f6809bec1afb951cc", size = 28262504, upload-time = "2025-10-24T10:08:00.932Z" },
+    { url = "https://files.pythonhosted.org/packages/bd/b0/0fa4d28a8edb42b0a7144edd20befd04173ac79819547216f8a9f36f9e50/pyarrow-22.0.0-cp314-cp314-macosx_12_0_arm64.whl", hash = "sha256:9bddc2cade6561f6820d4cd73f99a0243532ad506bc510a75a5a65a522b2d74d", size = 34224062, upload-time = "2025-10-24T10:08:14.101Z" },
+    { url = "https://files.pythonhosted.org/packages/0f/a8/7a719076b3c1be0acef56a07220c586f25cd24de0e3f3102b438d18ae5df/pyarrow-22.0.0-cp314-cp314-macosx_12_0_x86_64.whl", hash = "sha256:e70ff90c64419709d38c8932ea9fe1cc98415c4f87ea8da81719e43f02534bc9", size = 35990057, upload-time = "2025-10-24T10:08:21.842Z" },
+    { url = "https://files.pythonhosted.org/packages/89/3c/359ed54c93b47fb6fe30ed16cdf50e3f0e8b9ccfb11b86218c3619ae50a8/pyarrow-22.0.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:92843c305330aa94a36e706c16209cd4df274693e777ca47112617db7d0ef3d7", size = 45068002, upload-time = "2025-10-24T10:08:29.034Z" },
+    { url = "https://files.pythonhosted.org/packages/55/fc/4945896cc8638536ee787a3bd6ce7cec8ec9acf452d78ec39ab328efa0a1/pyarrow-22.0.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:6dda1ddac033d27421c20d7a7943eec60be44e0db4e079f33cc5af3b8280ccde", size = 47737765, upload-time = "2025-10-24T10:08:38.559Z" },
+    { url = "https://files.pythonhosted.org/packages/cd/5e/7cb7edeb2abfaa1f79b5d5eb89432356155c8426f75d3753cbcb9592c0fd/pyarrow-22.0.0-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:84378110dd9a6c06323b41b56e129c504d157d1a983ce8f5443761eb5256bafc", size = 48048139, upload-time = "2025-10-24T10:08:46.784Z" },
+    { url = "https://files.pythonhosted.org/packages/88/c6/546baa7c48185f5e9d6e59277c4b19f30f48c94d9dd938c2a80d4d6b067c/pyarrow-22.0.0-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:854794239111d2b88b40b6ef92aa478024d1e5074f364033e73e21e3f76b25e0", size = 50314244, upload-time = "2025-10-24T10:08:55.771Z" },
+    { url = "https://files.pythonhosted.org/packages/3c/79/755ff2d145aafec8d347bf18f95e4e81c00127f06d080135dfc86aea417c/pyarrow-22.0.0-cp314-cp314-win_amd64.whl", hash = "sha256:b883fe6fd85adad7932b3271c38ac289c65b7337c2c132e9569f9d3940620730", size = 28757501, upload-time = "2025-10-24T10:09:59.891Z" },
+    { url = "https://files.pythonhosted.org/packages/0e/d2/237d75ac28ced3147912954e3c1a174df43a95f4f88e467809118a8165e0/pyarrow-22.0.0-cp314-cp314t-macosx_12_0_arm64.whl", hash = "sha256:7a820d8ae11facf32585507c11f04e3f38343c1e784c9b5a8b1da5c930547fe2", size = 34355506, upload-time = "2025-10-24T10:09:02.953Z" },
+    { url = "https://files.pythonhosted.org/packages/1e/2c/733dfffe6d3069740f98e57ff81007809067d68626c5faef293434d11bd6/pyarrow-22.0.0-cp314-cp314t-macosx_12_0_x86_64.whl", hash = "sha256:c6ec3675d98915bf1ec8b3c7986422682f7232ea76cad276f4c8abd5b7319b70", size = 36047312, upload-time = "2025-10-24T10:09:10.334Z" },
+    { url = "https://files.pythonhosted.org/packages/7c/2b/29d6e3782dc1f299727462c1543af357a0f2c1d3c160ce199950d9ca51eb/pyarrow-22.0.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:3e739edd001b04f654b166204fc7a9de896cf6007eaff33409ee9e50ceaff754", size = 45081609, upload-time = "2025-10-24T10:09:18.61Z" },
+    { url = "https://files.pythonhosted.org/packages/8d/42/aa9355ecc05997915af1b7b947a7f66c02dcaa927f3203b87871c114ba10/pyarrow-22.0.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:7388ac685cab5b279a41dfe0a6ccd99e4dbf322edfb63e02fc0443bf24134e91", size = 47703663, upload-time = "2025-10-24T10:09:27.369Z" },
+    { url = "https://files.pythonhosted.org/packages/ee/62/45abedde480168e83a1de005b7b7043fd553321c1e8c5a9a114425f64842/pyarrow-22.0.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:f633074f36dbc33d5c05b5dc75371e5660f1dbf9c8b1d95669def05e5425989c", size = 48066543, upload-time = "2025-10-24T10:09:34.908Z" },
+    { url = "https://files.pythonhosted.org/packages/84/e9/7878940a5b072e4f3bf998770acafeae13b267f9893af5f6d4ab3904b67e/pyarrow-22.0.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:4c19236ae2402a8663a2c8f21f1870a03cc57f0bef7e4b6eb3238cc82944de80", size = 50288838, upload-time = "2025-10-24T10:09:44.394Z" },
+    { url = "https://files.pythonhosted.org/packages/7b/03/f335d6c52b4a4761bcc83499789a1e2e16d9d201a58c327a9b5cc9a41bd9/pyarrow-22.0.0-cp314-cp314t-win_amd64.whl", hash = "sha256:0c34fe18094686194f204a3b1787a27456897d8a2d62caf84b61e8dfbc0252ae", size = 29185594, upload-time = "2025-10-24T10:09:53.111Z" },
+]
+
 [[package]]
 name = "pycparser"
 version = "2.22"
@@ -2969,6 +3191,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/08/20/0f2523b9e50a8052bc6a8b732dfc8568abbdc42010aef03a2d750bdab3b2/python_json_logger-3.3.0-py3-none-any.whl", hash = "sha256:dd980fae8cffb24c13caf6e158d3d61c0d6d22342f932cb6e9deedab3d35eec7", size = 15163, upload-time = "2025-03-07T07:08:25.627Z" },
 ]
 
+[[package]]
+name = "pytz"
+version = "2025.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/f8/bf/abbd3cdfb8fbc7fb3d4d38d320f2441b1e7cbe29be4f23797b4a2b5d8aac/pytz-2025.2.tar.gz", hash = "sha256:360b9e3dbb49a209c21ad61809c7fb453643e048b38924c765813546746e81c3", size = 320884, upload-time = "2025-03-25T02:25:00.538Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/81/c4/34e93fe5f5429d7570ec1fa436f1986fb1f00c3e0f43a589fe2bbcd22c3f/pytz-2025.2-py2.py3-none-any.whl", hash = "sha256:5ddf76296dd8c44c26eb8f4b6f35488f3ccbf6fbbd7adee0b7262d43f0ec2f00", size = 509225, upload-time = "2025-03-25T02:24:58.468Z" },
+]
+
 [[package]]
 name = "pywin32"
 version = "311"
@@ -3360,6 +3591,18 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/7c/e4/56027c4a6b4ae70ca9de302488c5ca95ad4a39e190093d6c1a8ace08341b/requests-2.32.4-py3-none-any.whl", hash = "sha256:27babd3cda2a6d50b30443204ee89830707d396671944c998b5975b031ac2b2c", size = 64847, upload-time = "2025-06-09T16:43:05.728Z" },
 ]
 
+[[package]]
+name = "respx"
+version = "0.22.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "httpx" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/f4/7c/96bd0bc759cf009675ad1ee1f96535edcb11e9666b985717eb8c87192a95/respx-0.22.0.tar.gz", hash = "sha256:3c8924caa2a50bd71aefc07aa812f2466ff489f1848c96e954a5362d17095d91", size = 28439, upload-time = "2024-12-19T22:33:59.374Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/8e/67/afbb0978d5399bc9ea200f1d4489a23c9a1dad4eee6376242b8182389c79/respx-0.22.0-py2.py3-none-any.whl", hash = "sha256:631128d4c9aba15e56903fb5f66fb1eff412ce28dd387ca3a81339e52dbd3ad0", size = 25127, upload-time = "2024-12-19T22:33:57.837Z" },
+]
+
 [[package]]
 name = "rfc3339-validator"
 version = "0.1.4"
@@ -3860,6 +4103,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/17/69/cd203477f944c353c31bade965f880aa1061fd6bf05ded0726ca845b6ff7/typing_inspection-0.4.1-py3-none-any.whl", hash = "sha256:389055682238f53b04f7badcb49b989835495a96700ced5dab2d8feae4b26f51", size = 14552, upload-time = "2025-05-21T18:55:22.152Z" },
 ]
 
+[[package]]
+name = "tzdata"
+version = "2025.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/95/32/1a225d6164441be760d75c2c42e2780dc0873fe382da3e98a2e1e48361e5/tzdata-2025.2.tar.gz", hash = "sha256:b60a638fcc0daffadf82fe0f57e53d06bdec2f36c4df66280ae79bce6bd6f2b9", size = 196380, upload-time = "2025-03-23T13:54:43.652Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/5c/23/c7abc0ca0a1526a0774eca151daeb8de62ec457e77262b66b359c3c7679e/tzdata-2025.2-py2.py3-none-any.whl", hash = "sha256:1a403fada01ff9221ca8044d701868fa132215d84beb92242d9acd2147f667a8", size = 347839, upload-time = "2025-03-23T13:54:41.845Z" },
+]
+
 [[package]]
 name = "uri-template"
 version = "1.3.0"

ts/llama_cloud_services/CHANGELOG.md

@@ -1,5 +1,23 @@
 # llama-cloud-services
 
+## 0.4.1
+
+### Patch Changes
+
+- f3233de: Propagate retrieval metadata to retriever nodes
+
+## 0.4.0
+
+### Minor Changes
+
+- f293547: Switch to keyword arguments rather than positional args
+
+## 0.3.10
+
+### Patch Changes
+
+- fee516d: Adding LlamaClassify among the available LlamaCloud services
+
 ## 0.3.9
 
 ### Patch Changes

ts/llama_cloud_services/classify/package.json

@@ -0,0 +1,8 @@
+{
+  "type": "module",
+  "main": "./dist/index.cjs",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": "./dist/index.js",
+  "private": true
+}

ts/llama_cloud_services/package.json

@@ -1,6 +1,6 @@
 {
   "name": "llama-cloud-services",
-  "version": "0.3.9",
+  "version": "0.4.1",
   "type": "module",
   "license": "MIT",
   "scripts": {
@@ -9,8 +9,8 @@
     "build": "pnpm run generate && bunchee",
     "dev": "bunchee --watch",
     "lint": "eslint src/ --ignore-pattern client/*.ts --no-warn-ignored",
-    "format": "prettier --write ./src/",
-    "format:check": "prettier --check ./src/",
+    "format": "prettier --write ./src/ tests/",
+    "format:check": "prettier --check ./src/ tests/",
     "test": "vitest run --testTimeout=60000",
     "test:watch": "vitest --watch",
     "test:ui": "vitest --ui",
@@ -24,7 +24,8 @@
     "./reader",
     "./parse",
     "./beta/agent",
-    "./extract"
+    "./extract",
+    "./classify"
   ],
   "exports": {
     "./openapi.json": "./openapi.json",
@@ -83,6 +84,17 @@
       },
       "default": "./extract/dist/index.js"
     },
+    "./classify": {
+      "require": {
+        "types": "./classify/dist/index.d.cts",
+        "default": "./classify/dist/index.cjs"
+      },
+      "import": {
+        "types": "./classify/dist/index.d.ts",
+        "default": "./classify/dist/index.js"
+      },
+      "default": "./classify/dist/index.js"
+    },
     ".": {
       "require": {
         "types": "./dist/index.d.cts",

ts/llama_cloud_services/src/LlamaClassify.ts

@@ -0,0 +1,75 @@
+import { createClient, createConfig, type Client } from "@hey-api/client-fetch";
+import {
+  classify,
+  type ClassifyParsingConfiguration,
+  type ClassifierRule,
+  type ClassifyJobResults,
+} from "./classify";
+import { getUrl } from "./utils";
+import { getEnv } from "@llamaindex/env";
+import { File } from "buffer";
+
+export class LlamaClassify {
+  private client: Client;
+
+  constructor(
+    apiKey: string | undefined = undefined,
+    baseUrl: string | undefined = undefined,
+    region: string | undefined = undefined,
+  ) {
+    const key = apiKey ?? getEnv("LLAMA_CLOUD_API_KEY");
+    if (typeof key === "undefined") {
+      throw new Error(
+        "No API key provided and no API key found in environment. Please pass the API key or set `LLAMA_CLOUD_API_KEY` as an environment variable.",
+      );
+    }
+    const url = getUrl(baseUrl, region);
+    this.client = createClient(
+      createConfig({
+        baseUrl: url,
+        headers: {
+          Authorization: `Bearer ${key}`,
+        },
+      }),
+    );
+  }
+
+  async classify(
+    rules: ClassifierRule[],
+    configuration: ClassifyParsingConfiguration,
+    {
+      fileContents,
+      filePaths,
+      projectId,
+      pollingInterval = 1,
+      maxPollingIterations = 1800,
+      maxRetriesOnError = 10,
+      retryInterval = 0.5,
+    }: {
+      fileContents?:
+        | Buffer<ArrayBufferLike>[]
+        | File[]
+        | Uint8Array<ArrayBuffer>[]
+        | string[]
+        | undefined;
+      filePaths?: string[] | undefined;
+      projectId?: string;
+      pollingInterval?: number;
+      maxPollingIterations?: number;
+      maxRetriesOnError?: number;
+      retryInterval?: number;
+    },
+  ): Promise<ClassifyJobResults> {
+    const result = await classify(rules, configuration, {
+      fileContents,
+      filePaths,
+      projectId: projectId ?? undefined,
+      client: this.client,
+      pollingInterval,
+      maxPollingIterations,
+      maxRetriesOnError,
+      retryInterval,
+    });
+    return result;
+  }
+}

ts/llama_cloud_services/src/LlamaCloudRetriever.ts

@@ -34,12 +34,15 @@ export class LlamaCloudRetriever extends BaseRetriever {
 
   private resultNodesToNodeWithScore(
     nodes: TextNodeWithScore[],
+    metadata: Record<string, string> | undefined,
   ): NodeWithScore[] {
     return nodes.map((node: TextNodeWithScore) => {
       const textNode = jsonToNode(node.node, ObjectType.TEXT);
+      const extra_metadata = metadata || {};
       textNode.metadata = {
         ...textNode.metadata,
         ...node.node.extra_info, // append LlamaCloud extra_info to node metadata (file_name, pipeline_id, etc.)
+        ...extra_metadata, // append retrieval-level metadata
       };
       return {
         // Currently LlamaCloud only supports text nodes
@@ -63,6 +66,7 @@ export class LlamaCloudRetriever extends BaseRetriever {
   private async pageScreenshotNodesToNodeWithScore(
     nodes: PageScreenshotNodeWithScore[] | undefined,
     projectId: string,
+    metadata: Record<string, string> | undefined,
   ): Promise<NodeWithScore[]> {
     if (!nodes || nodes.length === 0) return [];
 
@@ -87,6 +91,7 @@ export class LlamaCloudRetriever extends BaseRetriever {
           image: base64,
           metadata: {
             ...(n.node.metadata ?? {}),
+            ...(metadata || {}),
             file_id: n.node.file_id,
             page_index: n.node.page_index,
           },
@@ -101,6 +106,7 @@ export class LlamaCloudRetriever extends BaseRetriever {
   private async pageFigureNodesToNodeWithScore(
     nodes: PageFigureNodeWithScore[] | undefined,
     projectId: string,
+    metadata: Record<string, string> | undefined,
   ): Promise<NodeWithScore[]> {
     if (!nodes || nodes.length === 0) return [];
 
@@ -126,6 +132,7 @@ export class LlamaCloudRetriever extends BaseRetriever {
           image: base64,
           metadata: {
             ...(n.node.metadata ?? {}),
+            ...(metadata || {}),
             file_id: n.node.file_id,
             page_index: n.node.page_index,
             figure_name: n.node.figure_name,
@@ -222,7 +229,10 @@ export class LlamaCloudRetriever extends BaseRetriever {
         },
       });
 
-    const textNodes = this.resultNodesToNodeWithScore(results.retrieval_nodes);
+    const textNodes = this.resultNodesToNodeWithScore(
+      results.retrieval_nodes,
+      results.metadata,
+    );
 
     const needScreenshots = (this.retrieveParams as RetrievalParams)
       .retrieve_page_screenshot_nodes;
@@ -240,12 +250,14 @@ export class LlamaCloudRetriever extends BaseRetriever {
         ? this.pageScreenshotNodesToNodeWithScore(
             results.image_nodes,
             projectId,
+            results.metadata,
           )
         : Promise.resolve([] as NodeWithScore[]),
       needFigures
         ? this.pageFigureNodesToNodeWithScore(
             results.page_figure_nodes,
             projectId,
+            results.metadata,
           )
         : Promise.resolve([] as NodeWithScore[]),
     ]);

ts/llama_cloud_services/src/LlamaExtract.ts

@@ -4,25 +4,7 @@ import * as extract from "./extract";
 import type { ExtractAgent, ExtractConfig } from "./extract";
 import { getEnv } from "@llamaindex/env";
 import type { ExtractResult } from "./type";
-
-const URLS = {
-  us: "https://api.cloud.llamaindex.ai",
-  eu: "https://api.cloud.eu.llamaindex.ai",
-  "us-staging": "https://api.staging.llamaindex.ai",
-} as const;
-
-function getUrl(baseUrl: string | undefined, region: string | undefined) {
-  if (typeof baseUrl != "undefined") {
-    return baseUrl;
-  }
-  if (typeof region === "undefined") {
-    return URLS["us"];
-  } else if (region === "us" || region === "eu" || region === "us-staging") {
-    return URLS[region];
-  } else {
-    throw new Error(`Unsupported region: ${region}`);
-  }
-}
+import { getUrl } from "./utils";
 
 export class LlamaExtractAgent {
   private agent: ExtractAgent;

ts/llama_cloud_services/src/classify.ts

@@ -0,0 +1,307 @@
+import type {
+  Options,
+  CreateClassifyJobApiV1ClassifierJobsPostData,
+  ClassifyJobCreate,
+  ClassifierRule,
+  ClassifyParsingConfiguration,
+  GetClassifyJobApiV1ClassifierJobsClassifyJobIdGetData,
+  GetClassificationJobResultsApiV1ClassifierJobsClassifyJobIdResultsGetData,
+  ClassifyJobResults,
+} from "./api";
+import {
+  StatusEnum,
+  createClassifyJobApiV1ClassifierJobsPost,
+  getClassifyJobApiV1ClassifierJobsClassifyJobIdGet,
+  getClassificationJobResultsApiV1ClassifierJobsClassifyJobIdResultsGet,
+} from "./api";
+import type { Client } from "@hey-api/client-fetch";
+import { sleep } from "./utils";
+import { uploadFile } from "./fileUpload";
+import { File } from "buffer";
+
+async function createClassifyJob({
+  fileIds,
+  rules,
+  parsingConfiguration,
+  projectId,
+  client,
+  maxRetriesOnError = 10,
+  retryInterval = 0.5,
+}: {
+  fileIds: string[];
+  rules: ClassifierRule[];
+  parsingConfiguration: ClassifyParsingConfiguration;
+  projectId?: string | undefined;
+  client?: Client | undefined;
+  maxRetriesOnError?: number;
+  retryInterval?: number;
+}): Promise<string> {
+  const rawData = {
+    file_ids: fileIds,
+    rules: rules,
+    parsing_configuration: parsingConfiguration,
+  } as ClassifyJobCreate;
+  const data = {
+    body: rawData,
+    query: {
+      project_id: projectId,
+    },
+  } as CreateClassifyJobApiV1ClassifierJobsPostData;
+  const options = data as Options<CreateClassifyJobApiV1ClassifierJobsPostData>;
+  if (typeof client != "undefined") {
+    options.client = client;
+  }
+  let retries = 0;
+  while (true) {
+    if (retries > maxRetriesOnError) {
+      throw new Error(
+        "Error while creating the classify job: Exceeded maximum number of retries, the API keeps returning errors.",
+      );
+    }
+    const response = await createClassifyJobApiV1ClassifierJobsPost(options);
+    if (!response.response.ok) {
+      if ("error" in response) {
+        console.log(
+          `An error occurred while creating the classification job.\nDetails:\n\n${JSON.stringify(
+            response.error,
+          )}\n\nRetrying...`,
+        );
+      }
+      retries++;
+      await sleep(retryInterval * 1000);
+    } else {
+      if (typeof response.data != "undefined") {
+        return response.data.id;
+      } else {
+        throw new Error(
+          "Error while creating the classify job: the job creation succeeded but no data where returned",
+        );
+      }
+    }
+  }
+}
+
+async function pollForJobCompletion({
+  jobId,
+  interval = 1,
+  maxIterations = 1800,
+  client,
+}: {
+  jobId: string;
+  interval?: number;
+  maxIterations?: number;
+  client?: Client | undefined;
+}): Promise<boolean> {
+  let status: StatusEnum | undefined = undefined;
+  const jobData = {
+    path: { classify_job_id: jobId },
+  } as GetClassifyJobApiV1ClassifierJobsClassifyJobIdGetData;
+  const jobOptions =
+    jobData as Options<GetClassifyJobApiV1ClassifierJobsClassifyJobIdGetData>;
+  if (typeof client != "undefined") {
+    jobOptions.client = client;
+  }
+  let numIterations: number = 0;
+  while (true) {
+    if (numIterations > maxIterations) {
+      return false;
+    }
+    const response =
+      await getClassifyJobApiV1ClassifierJobsClassifyJobIdGet(jobOptions);
+    if (!response.response.ok) {
+      numIterations++;
+    }
+    if (typeof response.data != "undefined") {
+      status = response.data.status as StatusEnum;
+      if (status == StatusEnum.CANCELLED || status == StatusEnum.ERROR) {
+        throw new Error("There was an error during the classification job.");
+      } else if (status == StatusEnum.SUCCESS) {
+        return true;
+      } else {
+        numIterations++;
+        await sleep(interval * 1000);
+      }
+    }
+  }
+}
+
+async function getJobResult({
+  jobId,
+  client,
+  projectId,
+  maxRetriesOnError = 10,
+  retryInterval = 0.5,
+}: {
+  jobId: string;
+  client?: Client | undefined;
+  projectId?: string | undefined;
+  maxRetriesOnError?: number;
+  retryInterval?: number;
+}): Promise<ClassifyJobResults> {
+  const jobData = {
+    path: { classify_job_id: jobId },
+    query: { project_id: projectId },
+  } as GetClassificationJobResultsApiV1ClassifierJobsClassifyJobIdResultsGetData;
+  const jobOptions =
+    jobData as Options<GetClassificationJobResultsApiV1ClassifierJobsClassifyJobIdResultsGetData>;
+  if (typeof client != "undefined") {
+    jobOptions.client = client;
+  }
+  let retries: number = 0;
+  while (true) {
+    if (retries > maxRetriesOnError) {
+      throw new Error(
+        "Error while getting the result of the classification job: Exceeded maximum number of retries, the API keeps returning errors.",
+      );
+    }
+    const response =
+      await getClassificationJobResultsApiV1ClassifierJobsClassifyJobIdResultsGet(
+        jobOptions,
+      );
+    if (!response.response.ok) {
+      if ("error" in response) {
+        console.log(
+          "An error occurred: ",
+          JSON.stringify(response.error),
+          "\nRetrying...",
+        );
+      }
+      retries++;
+      await sleep(retryInterval * 1000);
+    }
+    if (typeof response.data != "undefined") {
+      return response.data as ClassifyJobResults;
+    } else {
+      throw new Error(
+        "Error while retrieving results for the classify job: the result was successfully obtained but no data were returned",
+      );
+    }
+  }
+}
+
+export async function classify(
+  rules: ClassifierRule[],
+  parsingConfiguration: ClassifyParsingConfiguration,
+  {
+    fileContents,
+    filePaths,
+    projectId,
+    client,
+    pollingInterval = 1,
+    maxPollingIterations = 1800,
+    maxRetriesOnError = 10,
+    retryInterval = 0.5,
+  }: {
+    fileContents?:
+      | Buffer<ArrayBufferLike>[]
+      | File[]
+      | Uint8Array<ArrayBuffer>[]
+      | string[]
+      | undefined;
+    filePaths?: string[] | undefined;
+    projectId?: string | undefined;
+    client?: Client | undefined;
+    pollingInterval?: number;
+    maxPollingIterations?: number;
+    maxRetriesOnError?: number;
+    retryInterval?: number;
+  },
+): Promise<ClassifyJobResults> {
+  const fileIds: string[] = [];
+  if (!filePaths && !fileContents) {
+    throw new Error(
+      "One between filePath and fileContent needs to be provided",
+    );
+  }
+
+  if (filePaths) {
+    const uploadPromises = filePaths.map(async (name) => {
+      try {
+        const fileId = await uploadFile({
+          filePath: name,
+          maxRetriesOnError,
+          retryInterval: retryInterval,
+          project_id: projectId,
+          client: client,
+        });
+        if (fileId) {
+          return fileId;
+        } else {
+          console.error(`Unable to upload ${name}, skipping...`);
+          return null;
+        }
+      } catch (error) {
+        console.error(`Error uploading ${name}:`, error);
+        return null;
+      }
+    });
+
+    const results = await Promise.all(uploadPromises);
+    fileIds.push(...results.filter((id) => id !== null));
+  }
+
+  if (fileContents) {
+    const uploadPromises = fileContents.map(async (content) => {
+      try {
+        const fileId = await uploadFile({
+          fileContent: content,
+          ...(projectId ? { project_id: projectId } : {}),
+          ...(client ? { client: client } : {}),
+          maxRetriesOnError,
+          retryInterval,
+        });
+        if (fileId) {
+          return fileId;
+        } else {
+          console.error(`Unable to upload file (content), skipping...`);
+          return null;
+        }
+      } catch (error) {
+        console.error(`Error uploading file (content):`, error);
+        return null;
+      }
+    });
+
+    const results = await Promise.all(uploadPromises);
+    fileIds.push(...results.filter((id) => id !== null));
+  }
+
+  if (fileIds.length == 0) {
+    throw new Error(
+      "None of the provided files was successfully uploaded, it is not possible to create a classification job.",
+    );
+  }
+
+  const jobId = await createClassifyJob({
+    fileIds,
+    rules,
+    parsingConfiguration,
+    ...(projectId ? { projectId: projectId } : {}),
+    ...(client ? { client: client } : {}),
+    maxRetriesOnError,
+    retryInterval,
+  });
+  const success = await pollForJobCompletion({
+    jobId,
+    interval: pollingInterval,
+    maxIterations: maxPollingIterations,
+    client,
+  });
+  if (!success) {
+    throw new Error("Your job is taking longer than 10 minutes, timing out...");
+  } else {
+    return (await getJobResult({
+      jobId,
+      client,
+      projectId,
+      maxRetriesOnError,
+      retryInterval,
+    })) as ClassifyJobResults;
+  }
+}
+
+export {
+  type ClassifierRule,
+  type ClassifyJobResults,
+  type ClassifyParsingConfiguration,
+};

ts/llama_cloud_services/src/extract.ts

@@ -1,9 +1,5 @@
 import { emitWarning } from "process";
-import fs from "fs/promises";
-import { Blob } from "buffer";
-import * as path from "path";
 import type { ExtractResult } from "./type";
-import { randomUUID } from "@llamaindex/env";
 import { File } from "buffer";
 import {
   type Options,
@@ -19,7 +15,6 @@ import {
   type GetJobApiV1ExtractionJobsJobIdGetData,
   type GetJobResultApiV1ExtractionJobsJobIdResultGetData,
   StatusEnum,
-  type UploadFileApiV1FilesPostData,
   type StatelessExtractionRequest,
   type ExtractStatelessApiV1ExtractionRunPostData,
   type DeleteExtractionAgentApiV1ExtractionExtractionAgentsExtractionAgentIdDeleteData,
@@ -29,17 +24,12 @@ import {
   runJobApiV1ExtractionJobsPost,
   getJobApiV1ExtractionJobsJobIdGet,
   getJobResultApiV1ExtractionJobsJobIdResultGet,
-  uploadFileApiV1FilesPost,
   extractStatelessApiV1ExtractionRunPost,
   deleteExtractionAgentApiV1ExtractionExtractionAgentsExtractionAgentIdDelete,
 } from "./api";
 import type { Client } from "@hey-api/client-fetch";
 import { sleep } from "./utils";
-import { fileTypeFromBuffer } from "file-type";
-
-type BodyUploadFileApiV1FilesPost = {
-  upload_file: Blob | File;
-};
+import { uploadFile } from "./fileUpload";
 
 export async function createAgent(
   name: string,
@@ -221,95 +211,6 @@ export async function getAgent(
   }
 }
 
-function textToFile(text: string, fileName: string | null = null) {
-  return new File(
-    [text],
-    fileName ?? "uploadedFile_" + randomUUID().replaceAll("-", "_") + ".txt",
-  );
-}
-
-async function uploadFile(
-  filePath: string | undefined = undefined,
-  fileContent:
-    | Buffer<ArrayBufferLike>
-    | File
-    | Uint8Array<ArrayBuffer>
-    | string
-    | undefined = undefined,
-  fileName: string | undefined = undefined,
-  project_id: string | null = null,
-  organization_id: string | null = null,
-  client: Client | undefined = undefined,
-  maxRetriesOnError: number = 10,
-  retryInterval: number = 0.5,
-): Promise<string | undefined> {
-  let file: File | undefined = undefined;
-  if (typeof filePath === "undefined" && typeof fileContent === "undefined") {
-    throw new Error(
-      "One between filePath and fileContent needs to be provided",
-    );
-  } else if (typeof filePath != "undefined") {
-    const buffer = await fs.readFile(filePath);
-    const actualFileName = fileName ?? path.basename(filePath);
-    const uint8Array = new Uint8Array(buffer);
-    file = new File([uint8Array], actualFileName);
-  } else if (typeof fileContent != "undefined") {
-    if (fileContent instanceof File) {
-      file = fileContent;
-    } else if (fileContent instanceof Buffer) {
-      const fileType = await fileTypeFromBuffer(fileContent);
-      const ext = fileType?.ext ?? "pdf";
-      const uint8Array = new Uint8Array(fileContent);
-      file = new File(
-        [uint8Array],
-        fileName ??
-          "uploadedFile_" + randomUUID().replaceAll("-", "_") + "." + ext,
-      );
-    } else if (fileContent instanceof Uint8Array) {
-      const fileType = await fileTypeFromBuffer(fileContent);
-      const ext = fileType?.ext ?? "pdf";
-      file = new File(
-        [fileContent],
-        fileName ??
-          "uploadedFile_" + randomUUID().replaceAll("-", "_") + "." + ext,
-      );
-    } else if (typeof fileContent === "string") {
-      file = textToFile(fileContent, fileName);
-    } else {
-      throw new Error("Unsupported fileContent type");
-    }
-  }
-  const fileToUpload = {
-    upload_file: file,
-  } as BodyUploadFileApiV1FilesPost;
-  const uploadData = {
-    body: fileToUpload,
-    query: { organization_id: organization_id, project_id: project_id },
-  } as UploadFileApiV1FilesPostData;
-  const uploadOptions = uploadData as Options<UploadFileApiV1FilesPostData>;
-  if (typeof client != "undefined") {
-    uploadOptions.client = client;
-  }
-  let retries: number = 0;
-  while (true) {
-    if (retries > maxRetriesOnError) {
-      throw new Error(
-        "Error while processing your file: Exceeded maximum number of retries, the API keeps returning errors.",
-      );
-    }
-    const uploadResponse = await uploadFileApiV1FilesPost(uploadOptions);
-    let fileId: string | undefined = undefined;
-    if (!uploadResponse.response.ok) {
-      retries++;
-      await sleep(retryInterval * 1000);
-    }
-    if (typeof uploadResponse.data != "undefined") {
-      fileId = uploadResponse.data.id as string;
-      return fileId;
-    }
-  }
-}
-
 async function createExtractJob(
   options:
     | Options<RunJobApiV1ExtractionJobsPostData>
@@ -477,16 +378,16 @@ export async function extract(
   maxRetriesOnError: number = 10,
   retryInterval: number = 0.5,
 ): Promise<ExtractResult | undefined> {
-  const fileId = (await uploadFile(
+  const fileId = (await uploadFile({
     filePath,
     fileContent,
     fileName,
-    project_id,
-    organization_id,
+    project_id: project_id ?? undefined,
+    organization_id: organization_id ?? undefined,
     client,
     maxRetriesOnError,
     retryInterval,
-  )) as string;
+  })) as string;
   const extractJobCreate = {
     extraction_agent_id: agentId,
     file_id: fileId,
@@ -556,16 +457,16 @@ export async function extractStateless(
   maxRetriesOnError: number = 10,
   retryInterval: number = 0.5,
 ): Promise<ExtractResult | undefined> {
-  const fileId = (await uploadFile(
+  const fileId = (await uploadFile({
     filePath,
     fileContent,
     fileName,
-    project_id,
-    organization_id,
+    project_id: project_id ?? undefined,
+    organization_id: organization_id ?? undefined,
     client,
     maxRetriesOnError,
     retryInterval,
-  )) as string;
+  })) as string;
   const extractStatetelessCreate = {
     data_schema: dataSchema,
     file_id: fileId,

ts/llama_cloud_services/src/fileUpload.ts

@@ -0,0 +1,120 @@
+import fs from "fs/promises";
+import { Blob } from "buffer";
+import * as path from "path";
+import { randomUUID } from "@llamaindex/env";
+import { File } from "buffer";
+import {
+  type Options,
+  type UploadFileApiV1FilesPostData,
+  uploadFileApiV1FilesPost,
+} from "./api";
+import type { Client } from "@hey-api/client-fetch";
+import { sleep } from "./utils";
+import { fileTypeFromBuffer } from "file-type";
+
+type BodyUploadFileApiV1FilesPost = {
+  upload_file: Blob | File;
+};
+
+function textToFile(text: string, fileName: string | null = null) {
+  return new File(
+    [text],
+    fileName ?? "uploadedFile_" + randomUUID().replaceAll("-", "_") + ".txt",
+  );
+}
+
+export async function uploadFile({
+  filePath,
+  fileContent,
+  fileName,
+  project_id,
+  organization_id,
+  client,
+  maxRetriesOnError = 10,
+  retryInterval = 0.5,
+}: {
+  filePath?: string | undefined;
+  fileContent?:
+    | Buffer<ArrayBufferLike>
+    | File
+    | Uint8Array<ArrayBuffer>
+    | string
+    | undefined;
+  fileName?: string | undefined;
+  project_id?: string | undefined;
+  organization_id?: string | undefined;
+  client?: Client | undefined;
+  maxRetriesOnError?: number;
+  retryInterval?: number;
+}): Promise<string | undefined> {
+  let file: File | undefined = undefined;
+  if (typeof filePath === "undefined" && typeof fileContent === "undefined") {
+    throw new Error(
+      "One between filePath and fileContent needs to be provided",
+    );
+  } else if (typeof filePath != "undefined") {
+    const buffer = await fs.readFile(filePath);
+    const actualFileName = fileName ?? path.basename(filePath);
+    const uint8Array = new Uint8Array(buffer);
+    file = new File([uint8Array], actualFileName);
+  } else if (typeof fileContent != "undefined") {
+    if (fileContent instanceof File) {
+      file = fileContent;
+    } else if (fileContent instanceof Buffer) {
+      const fileType = await fileTypeFromBuffer(fileContent);
+      const ext = fileType?.ext ?? "pdf";
+      const uint8Array = new Uint8Array(fileContent);
+      file = new File(
+        [uint8Array],
+        fileName ??
+          "uploadedFile_" + randomUUID().replaceAll("-", "_") + "." + ext,
+      );
+    } else if (fileContent instanceof Uint8Array) {
+      const fileType = await fileTypeFromBuffer(fileContent);
+      const ext = fileType?.ext ?? "pdf";
+      file = new File(
+        [fileContent],
+        fileName ??
+          "uploadedFile_" + randomUUID().replaceAll("-", "_") + "." + ext,
+      );
+    } else if (typeof fileContent === "string") {
+      file = textToFile(fileContent, fileName);
+    } else {
+      throw new Error("Unsupported fileContent type");
+    }
+  }
+  const fileToUpload = {
+    upload_file: file,
+  } as BodyUploadFileApiV1FilesPost;
+  const uploadData = {
+    body: fileToUpload,
+    query: { project_id: project_id, organization_id: organization_id },
+  } as UploadFileApiV1FilesPostData;
+  const uploadOptions = uploadData as Options<UploadFileApiV1FilesPostData>;
+  if (typeof client != "undefined") {
+    uploadOptions.client = client;
+  }
+  let retries: number = 0;
+  while (true) {
+    if (retries > maxRetriesOnError) {
+      throw new Error(
+        "Error while processing your file: Exceeded maximum number of retries, the API keeps returning errors.",
+      );
+    }
+    const uploadResponse = await uploadFileApiV1FilesPost(uploadOptions);
+    let fileId: string | undefined = undefined;
+    if (!uploadResponse.response.ok) {
+      const error = await uploadResponse.response.text();
+      console.error("Error while uploading file: ", error);
+      retries++;
+      await sleep(retryInterval * 1000);
+    }
+    if (
+      uploadResponse.response.ok &&
+      typeof uploadResponse.data != "undefined"
+    ) {
+      fileId = uploadResponse.data.id as string;
+      return fileId;
+    }
+  }
+}

ts/llama_cloud_services/src/index.ts

@@ -8,3 +8,9 @@ export type { CloudConstructorParams } from "./type.js";
 export { LlamaParseReader } from "./reader.js";
 export { LlamaExtract, LlamaExtractAgent } from "./LlamaExtract.js";
 export type { ExtractConfig } from "./extract.js";
+export { LlamaClassify } from "./LlamaClassify.js";
+export type {
+  ClassifierRule,
+  ClassifyJobResults,
+  ClassifyParsingConfiguration,
+} from "./classify.js";

ts/llama_cloud_services/src/utils.ts

@@ -117,3 +117,25 @@ export function getSavePath(downloadPath: string, i: number): string {
 
   return savePath;
 }
+
+const URLS = {
+  us: "https://api.cloud.llamaindex.ai",
+  eu: "https://api.cloud.eu.llamaindex.ai",
+  "us-staging": "https://api.staging.llamaindex.ai",
+} as const;
+
+export function getUrl(
+  baseUrl: string | undefined,
+  region: string | undefined,
+) {
+  if (typeof baseUrl != "undefined") {
+    return baseUrl;
+  }
+  if (typeof region === "undefined") {
+    return URLS["us"];
+  } else if (region === "us" || region === "eu" || region === "us-staging") {
+    return URLS[region];
+  } else {
+    throw new Error(`Unsupported region: ${region}`);
+  }
+}