Release 0.3.12 (#852)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

.github/workflows/test.yml

@@ -69,15 +69,14 @@ jobs:
       - name: Install dependencies
         run: pnpm install
       - name: Build
-        run: pnpm run build --filter llamaindex
+        run: pnpm run build
       - name: Use Build For Examples
         run: pnpm link ../packages/core/
         working-directory: ./examples
       - name: Run Type Check
         run: pnpm run type-check
       - name: Run Circular Dependency Check
-        run: pnpm run circular-check
-        working-directory: ./packages/core
+        run: pnpm dlx turbo run circular-check
       - uses: actions/upload-artifact@v3
         if: failure()
         with:
@@ -105,7 +104,7 @@ jobs:
       - name: Install dependencies
         run: pnpm install
       - name: Build llamaindex
-        run: pnpm run build --filter llamaindex
+        run: pnpm run build
       - name: Build ${{ matrix.packages }}
         run: pnpm run build
         working-directory: packages/core/e2e/examples/${{ matrix.packages }}
@@ -124,7 +123,7 @@ jobs:
       - name: Install dependencies
         run: pnpm install
       - name: Build
-        run: pnpm run build --filter llamaindex
+        run: pnpm run build
       - name: Copy examples
         run: rsync -rv --exclude=node_modules ./examples ${{ runner.temp }}
       - name: Pack @llamaindex/env

.husky/pre-commit

@@ -1,3 +1,3 @@
-pnpm format:write
+pnpm format
 pnpm lint
 npx lint-staged

README.md

@@ -78,6 +78,17 @@ node --import tsx ./main.ts
 
 ### Next.js
 
+First, you will need to add a llamaindex plugin to your Next.js project.
+
+```js
+// next.config.js
+const withLlamaIndex = require("llamaindex/next");
+
+module.exports = withLlamaIndex({
+  // your next.js config
+});
+```
+
 You can combine `ai` with `llamaindex` in Next.js with RSC (React Server Components).
 
 ```tsx

apps/docs/CHANGELOG.md

@@ -1,5 +1,69 @@
 # docs
 
+## 0.0.20
+
+### Patch Changes
+
+- Updated dependencies [34fb1d8]
+  - llamaindex@0.3.12
+
+## 0.0.19
+
+### Patch Changes
+
+- Updated dependencies [e072c45]
+- Updated dependencies [9e133ac]
+- Updated dependencies [447105a]
+- Updated dependencies [320be3f]
+  - llamaindex@0.3.11
+
+## 0.0.18
+
+### Patch Changes
+
+- Updated dependencies [4aba02e]
+  - llamaindex@0.3.10
+
+## 0.0.17
+
+### Patch Changes
+
+- Updated dependencies [c3747d0]
+  - llamaindex@0.3.9
+
+## 0.0.16
+
+### Patch Changes
+
+- Updated dependencies [ce94780]
+  - llamaindex@0.3.8
+
+## 0.0.15
+
+### Patch Changes
+
+- Updated dependencies [b6a6606]
+- Updated dependencies [b6a6606]
+  - llamaindex@0.3.7
+
+## 0.0.14
+
+### Patch Changes
+
+- Updated dependencies [efa326a]
+  - llamaindex@0.3.6
+
+## 0.0.13
+
+### Patch Changes
+
+- Updated dependencies [bc7a11c]
+- Updated dependencies [2fe2b81]
+- Updated dependencies [5596e31]
+- Updated dependencies [e74fe88]
+- Updated dependencies [be5df5b]
+  - llamaindex@0.3.5
+
 ## 0.0.12
 
 ### Patch Changes

apps/docs/package.json

@@ -1,6 +1,6 @@
 {
   "name": "docs",
-  "version": "0.0.12",
+  "version": "0.0.20",
   "private": true,
   "scripts": {
     "docusaurus": "docusaurus",
@@ -15,29 +15,29 @@
     "typecheck": "tsc"
   },
   "dependencies": {
-    "@docusaurus/core": "^3.2.1",
-    "@docusaurus/remark-plugin-npm2yarn": "^3.2.1",
+    "@docusaurus/core": "^3.3.2",
+    "@docusaurus/remark-plugin-npm2yarn": "^3.3.2",
     "@llamaindex/examples": "workspace:*",
     "@mdx-js/react": "^3.0.1",
-    "clsx": "^2.1.0",
+    "clsx": "^2.1.1",
     "llamaindex": "workspace:*",
     "postcss": "^8.4.38",
     "prism-react-renderer": "^2.3.1",
     "raw-loader": "^4.0.2",
-    "react": "^18.2.0",
-    "react-dom": "^18.2.0"
+    "react": "^18.3.1",
+    "react-dom": "^18.3.1"
   },
   "devDependencies": {
-    "@docusaurus/module-type-aliases": "3.2.0",
-    "@docusaurus/preset-classic": "^3.2.1",
-    "@docusaurus/theme-classic": "^3.2.1",
-    "@docusaurus/types": "^3.2.1",
+    "@docusaurus/module-type-aliases": "3.3.2",
+    "@docusaurus/preset-classic": "^3.3.2",
+    "@docusaurus/theme-classic": "^3.3.2",
+    "@docusaurus/types": "^3.3.2",
     "@tsconfig/docusaurus": "^2.0.3",
-    "@types/node": "^20.12.7",
-    "docusaurus-plugin-typedoc": "^0.22.0",
+    "@types/node": "^20.12.11",
+    "docusaurus-plugin-typedoc": "^1.0.1",
     "typedoc": "^0.25.13",
-    "typedoc-plugin-markdown": "^3.17.1",
-    "typescript": "^5.4.4"
+    "typedoc-plugin-markdown": "^4.0.1",
+    "typescript": "^5.4.5"
   },
   "browserslist": {
     "production": [

examples/agent/query_openai_agent.ts

@@ -29,15 +29,16 @@ async function main() {
   // Create an OpenAIAgent with the function tools
   const agent = new OpenAIAgent({
     tools: [queryEngineTool],
+    verbose: true,
   });
 
   // Chat with the agent
   const response = await agent.chat({
-    message: "What was his salary?",
+    message: "What was his first salary?",
   });
 
   // Print the response
-  console.log(String(response));
+  console.log(response.response);
 }
 
 void main().then(() => {

examples/agent/retriever_openai_agent.ts

@@ -0,0 +1,65 @@
+import {
+  FunctionTool,
+  MetadataMode,
+  NodeWithScore,
+  OpenAIAgent,
+  SimpleDirectoryReader,
+  VectorStoreIndex,
+} from "llamaindex";
+
+async function main() {
+  // Load the documents
+  const documents = await new SimpleDirectoryReader().loadData({
+    directoryPath: "node_modules/llamaindex/examples",
+  });
+
+  // Create a vector index from the documents
+  const vectorIndex = await VectorStoreIndex.fromDocuments(documents);
+
+  const retriever = vectorIndex.asRetriever({ similarityTopK: 3 });
+
+  const retrieverTool = FunctionTool.from(
+    async ({ query }: { query: string }) => {
+      const nodesWithScores = await retriever.retrieve({
+        query,
+      });
+      return nodesWithScores
+        .map((nodeWithScore: NodeWithScore) =>
+          nodeWithScore.node.getContent(MetadataMode.NONE),
+        )
+        .join("\n");
+    },
+    {
+      name: "get_abramov_info",
+      description: "Get information about the Abramov documents",
+      parameters: {
+        type: "object",
+        properties: {
+          query: {
+            type: "string",
+            description: "The query about Abramov",
+          },
+        },
+        required: ["query"],
+      },
+    },
+  );
+
+  // Create an OpenAIAgent with the function tools
+  const agent = new OpenAIAgent({
+    tools: [retrieverTool],
+    verbose: true,
+  });
+
+  // Chat with the agent
+  const response = await agent.chat({
+    message: "What was his first salary?",
+  });
+
+  // Print the response
+  console.log(response.response);
+}
+
+void main().then(() => {
+  console.log("Done");
+});

examples/ollama.ts

@@ -2,7 +2,12 @@ import { OllamaEmbedding } from "llamaindex";
 import { Ollama } from "llamaindex/llm/ollama";
 
 (async () => {
-  const llm = new Ollama({ model: "llama3" });
+  const llm = new Ollama({
+    model: "llama3",
+    config: {
+      host: "http://localhost:11434",
+    },
+  });
   const embedModel = new OllamaEmbedding({ model: "nomic-embed-text" });
   {
     const response = await llm.chat({

examples/package.json

@@ -4,22 +4,22 @@
   "version": "0.0.4",
   "dependencies": {
     "@aws-crypto/sha256-js": "^5.2.0",
-    "@datastax/astra-db-ts": "^1.0.1",
+    "@datastax/astra-db-ts": "^1.1.0",
     "@notionhq/client": "^2.2.15",
-    "@pinecone-database/pinecone": "^1.1.3",
-    "@zilliz/milvus2-sdk-node": "^2.4.1",
-    "chromadb": "^1.8.1",
-    "commander": "^11.1.0",
+    "@pinecone-database/pinecone": "^2.2.0",
+    "@zilliz/milvus2-sdk-node": "^2.4.2",
+    "chromadb": "^1.7.3",
+    "commander": "^12.0.0",
     "dotenv": "^16.4.5",
     "js-tiktoken": "^1.0.11",
     "llamaindex": "*",
-    "mongodb": "^6.5.0",
+    "mongodb": "^6.6.1",
     "pathe": "^1.1.2"
   },
   "devDependencies": {
-    "@types/node": "^20.12.7",
+    "@types/node": "^20.12.11",
     "ts-node": "^10.9.2",
-    "tsx": "^4.7.2",
+    "tsx": "^4.9.3",
     "typescript": "^5.4.5"
   },
   "scripts": {

examples/readers/package.json

@@ -17,8 +17,8 @@
     "llamaindex": "*"
   },
   "devDependencies": {
-    "@types/node": "^20.12.7",
-    "tsx": "^4.7.2",
+    "@types/node": "^20.12.11",
+    "tsx": "^4.9.3",
     "typescript": "^5.4.5"
   }
 }

package.json

@@ -2,8 +2,8 @@
   "name": "@llamaindex/monorepo",
   "private": true,
   "scripts": {
-    "build": "turbo run build",
-    "build:release": "turbo run build lint test --filter=\"!docs\" --filter=\"!*-test\"",
+    "build": "turbo run build --filter=\"!docs\" --filter=\"!*-test\" --filter=\"!*-example\"",
+    "build:release": "turbo run build lint test --filter=\"!docs\" --filter=\"!*-test\" --filter=\"!*-example\"",
     "dev": "turbo run dev",
     "format": "prettier --ignore-unknown --cache --check .",
     "format:write": "prettier --ignore-unknown --write .",
@@ -15,22 +15,23 @@
     "release": "pnpm run check-minor-version && pnpm run build:release && changeset publish",
     "release-snapshot": "pnpm run check-minor-version && pnpm run build:release && changeset publish --tag snapshot",
     "check-minor-version": "node ./scripts/check-minor-version",
-    "new-version": "changeset version && pnpm run check-minor-version && pnpm run build:release",
+    "new-version": "changeset version && pnpm run check-minor-version && pnpm format:write && pnpm run build:release",
     "new-snapshot": "pnpm run build:release && changeset version --snapshot"
   },
   "devDependencies": {
     "@changesets/cli": "^2.27.1",
-    "@typescript-eslint/eslint-plugin": "^7.7.0",
+    "@typescript-eslint/eslint-plugin": "^7.8.0",
     "eslint": "^8.57.0",
-    "eslint-config-next": "^13.5.6",
-    "eslint-config-prettier": "^8.10.0",
-    "eslint-config-turbo": "^1.13.2",
-    "eslint-plugin-react": "7.28.0",
+    "eslint-config-next": "^14.2.3",
+    "eslint-config-prettier": "^9.1.0",
+    "eslint-config-turbo": "^1.13.3",
+    "eslint-plugin-react": "7.34.1",
     "husky": "^9.0.11",
     "lint-staged": "^15.2.2",
+    "madge": "^7.0.0",
     "prettier": "^3.2.5",
     "prettier-plugin-organize-imports": "^3.2.4",
-    "turbo": "^1.13.2",
+    "turbo": "^1.13.3",
     "typescript": "^5.4.5"
   },
   "packageManager": "pnpm@9.0.5",

packages/autotool/README.md

@@ -0,0 +1,83 @@
+# @llamaindex/autotool
+
+> Auto transpile your JS function to LLM Agent compatible
+
+## Usage
+
+First, Install the package
+
+```shell
+npm install @llamaindex/autotool
+pnpm add @llamaindex/autotool
+yarn add @llamaindex/autotool
+```
+
+Second, Add the plugin/loader to your configuration:
+
+### Next.js
+
+```javascript
+import { withNext } from "@llamaindex/autotool/next";
+
+/** @type {import('next').NextConfig} */
+const nextConfig = {};
+
+export default withNext(nextConfig);
+```
+
+### Node.js
+
+```shell
+node --import @llamaindex/autotool/node ./path/to/your/script.js
+```
+
+Third, add `"use tool"` on top of your tool file or change to `.tool.ts`.
+
+```typescript
+"use tool";
+
+export function getWeather(city: string) {
+  // ...
+}
+// ...
+```
+
+Finally, export a chat handler function to the frontend using `llamaindex` Agent
+
+```typescript
+"use server";
+
+// imports ...
+
+export async function chatWithAI(message: string): Promise<JSX.Element> {
+  const agent = new OpenAIAgent({
+    tools: convertTools("llamaindex"),
+  });
+  const uiStream = createStreamableUI();
+  agent
+    .chat({
+      stream: true,
+      message,
+    })
+    .then(async (responseStream) => {
+      return responseStream.pipeTo(
+        new WritableStream({
+          start: () => {
+            uiStream.append("\n");
+          },
+          write: async (message) => {
+            uiStream.append(message.response.delta);
+          },
+          close: () => {
+            uiStream.done();
+          },
+        }),
+      );
+    });
+  return uiStream.value;
+}
+```
+
+## License
+
+MIT

packages/autotool/examples/01_node/CHANGELOG.md

@@ -0,0 +1,36 @@
+# @llamaindex/autotool-01-node-example
+
+## null
+
+### Patch Changes
+
+- Updated dependencies [34fb1d8]
+  - llamaindex@0.3.12
+  - @llamaindex/autotool@0.0.1
+
+## null
+
+### Patch Changes
+
+- Updated dependencies [e072c45]
+- Updated dependencies [9e133ac]
+- Updated dependencies [447105a]
+- Updated dependencies [320be3f]
+  - llamaindex@0.3.11
+  - @llamaindex/autotool@0.0.1
+
+## null
+
+### Patch Changes
+
+- Updated dependencies [4aba02e]
+  - llamaindex@0.3.10
+  - @llamaindex/autotool@0.0.1
+
+## null
+
+### Patch Changes
+
+- Updated dependencies [c3747d0]
+  - llamaindex@0.3.9
+  - @llamaindex/autotool@0.0.1

packages/autotool/examples/01_node/package.json

@@ -0,0 +1,17 @@
+{
+  "name": "@llamaindex/autotool-01-node-example",
+  "private": true,
+  "type": "module",
+  "dependencies": {
+    "@llamaindex/autotool": "workspace:*",
+    "llamaindex": "workspace:*",
+    "openai": "^4.43.0"
+  },
+  "devDependencies": {
+    "tsx": "^4.9.3"
+  },
+  "scripts": {
+    "start": "node --import tsx --import @llamaindex/autotool/node ./src/index.ts"
+  },
+  "version": null
+}

packages/autotool/examples/01_node/src/index.tool.ts

@@ -0,0 +1,11 @@
+import { getWeather } from "./utils.js";
+
+/**
+ * Get current location
+ */
+export function getCurrentLocation() {
+  console.log("Getting current location");
+  return "London";
+}
+
+export { getWeather };

packages/autotool/examples/01_node/src/index.ts

@@ -0,0 +1,23 @@
+import { convertTools } from "@llamaindex/autotool";
+import { OpenAI } from "openai";
+import "./index.tool.js";
+
+const openai = new OpenAI();
+{
+  const response = await openai.chat.completions.create({
+    model: "gpt-3.5-turbo",
+    messages: [
+      {
+        role: "user",
+        content: "What's my current weather?",
+      },
+    ],
+    tools: convertTools("openai"),
+    stream: false,
+  });
+
+  const toolCalls = response.choices[0].message.tool_calls ?? [];
+  for (const toolCall of toolCalls) {
+    toolCall.function.name;
+  }
+}

packages/autotool/examples/01_node/src/utils.ts

@@ -0,0 +1,8 @@
+/**
+ * Get the weather for a city
+ * @param city The city to get the weather for
+ * @returns The weather for the city, e.g. "Sunny", "Rainy", etc.
+ */
+export function getWeather(city: string) {
+  return `The weather in ${city} is sunny!`;
+}

packages/autotool/examples/01_node/tsconfig.json

@@ -0,0 +1,9 @@
+{
+  "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    "outDir": "./lib",
+    "module": "node16",
+    "moduleResolution": "node16"
+  },
+  "include": ["./src"]
+}

packages/autotool/examples/02_nextjs/.env.example

@@ -0,0 +1,3 @@
+# Rename this file to `.env.local` to use environment variables locally with `next dev`
+# https://nextjs.org/docs/pages/building-your-application/configuring/environment-variables
+MY_HOST="example.com"

packages/autotool/examples/02_nextjs/.gitignore

@@ -0,0 +1,35 @@
+# See https://help.github.com/articles/ignoring-files/ for more about ignoring files.
+
+# dependencies
+/node_modules
+/.pnp
+.pnp.js
+
+# testing
+/coverage
+
+# next.js
+/.next/
+/out/
+
+# production
+/build
+
+# misc
+.DS_Store
+*.pem
+
+# debug
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+
+# local env files
+.env*.local
+
+# vercel
+.vercel
+
+# typescript
+*.tsbuildinfo
+next-env.d.ts

packages/autotool/examples/02_nextjs/CHANGELOG.md

@@ -0,0 +1,36 @@
+# @llamaindex/autotool-02-next-example
+
+## 0.1.4
+
+### Patch Changes
+
+- Updated dependencies [34fb1d8]
+  - llamaindex@0.3.12
+  - @llamaindex/autotool@0.0.1
+
+## 0.1.3
+
+### Patch Changes
+
+- Updated dependencies [e072c45]
+- Updated dependencies [9e133ac]
+- Updated dependencies [447105a]
+- Updated dependencies [320be3f]
+  - llamaindex@0.3.11
+  - @llamaindex/autotool@0.0.1
+
+## 0.1.2
+
+### Patch Changes
+
+- Updated dependencies [4aba02e]
+  - llamaindex@0.3.10
+  - @llamaindex/autotool@0.0.1
+
+## 0.1.1
+
+### Patch Changes
+
+- Updated dependencies [c3747d0]
+  - llamaindex@0.3.9
+  - @llamaindex/autotool@0.0.1

packages/autotool/examples/02_nextjs/README.md

@@ -0,0 +1,30 @@
+This is a [LlamaIndex](https://www.llamaindex.ai/) project using [Next.js](https://nextjs.org/) bootstrapped with [`create-llama`](https://github.com/run-llama/LlamaIndexTS/tree/main/packages/create-llama).
+
+## Getting Started
+
+First, install the dependencies:
+
+```
+npm install
+```
+
+Second, run the development server:
+
+```
+npm run dev
+```
+
+Open [http://localhost:3000](http://localhost:3000) with your browser to see the result.
+
+You can start editing the page by modifying `app/page.tsx`. The page auto-updates as you edit the file.
+
+This project uses [`next/font`](https://nextjs.org/docs/basic-features/font-optimization) to automatically optimize and load Inter, a custom Google Font.
+
+## Learn More
+
+To learn more about LlamaIndex, take a look at the following resources:
+
+- [LlamaIndex Documentation](https://docs.llamaindex.ai) - learn about LlamaIndex (Python features).
+- [LlamaIndexTS Documentation](https://ts.llamaindex.ai) - learn about LlamaIndex (Typescript features).
+
+You can check out [the LlamaIndexTS GitHub repository](https://github.com/run-llama/LlamaIndexTS) - your feedback and contributions are welcome!

packages/autotool/examples/02_nextjs/actions.ts

@@ -0,0 +1,38 @@
+"use server";
+import { OpenAIAgent } from "llamaindex";
+// import your tools on top, that's it
+import { runWithStreamableUI } from "@/context";
+import "@/tool";
+import { convertTools } from "@llamaindex/autotool";
+import { createStreamableUI } from "ai/rsc";
+import type { JSX } from "react";
+
+export async function chatWithAI(message: string): Promise<JSX.Element> {
+  const agent = new OpenAIAgent({
+    tools: convertTools("llamaindex"),
+  });
+  const uiStream = createStreamableUI();
+  runWithStreamableUI(uiStream, () =>
+    agent
+      .chat({
+        stream: true,
+        message,
+      })
+      .then(async (responseStream) => {
+        return responseStream.pipeTo(
+          new WritableStream({
+            start: () => {
+              uiStream.append("\n");
+            },
+            write: async (message) => {
+              uiStream.append(message.response.delta);
+            },
+            close: () => {
+              uiStream.done();
+            },
+          }),
+        );
+      }),
+  ).catch(uiStream.error);
+  return uiStream.value;
+}

packages/autotool/examples/02_nextjs/app/globals.css

@@ -0,0 +1,94 @@
+@tailwind base;
+@tailwind components;
+@tailwind utilities;
+
+@layer base {
+  :root {
+    --background: 0 0% 100%;
+    --foreground: 222.2 47.4% 11.2%;
+
+    --muted: 210 40% 96.1%;
+    --muted-foreground: 215.4 16.3% 46.9%;
+
+    --popover: 0 0% 100%;
+    --popover-foreground: 222.2 47.4% 11.2%;
+
+    --border: 214.3 31.8% 91.4%;
+    --input: 214.3 31.8% 91.4%;
+
+    --card: 0 0% 100%;
+    --card-foreground: 222.2 47.4% 11.2%;
+
+    --primary: 222.2 47.4% 11.2%;
+    --primary-foreground: 210 40% 98%;
+
+    --secondary: 210 40% 96.1%;
+    --secondary-foreground: 222.2 47.4% 11.2%;
+
+    --accent: 210 40% 96.1%;
+    --accent-foreground: 222.2 47.4% 11.2%;
+
+    --destructive: 0 100% 50%;
+    --destructive-foreground: 210 40% 98%;
+
+    --ring: 215 20.2% 65.1%;
+
+    --radius: 0.5rem;
+  }
+
+  .dark {
+    --background: 224 71% 4%;
+    --foreground: 213 31% 91%;
+
+    --muted: 223 47% 11%;
+    --muted-foreground: 215.4 16.3% 56.9%;
+
+    --accent: 216 34% 17%;
+    --accent-foreground: 210 40% 98%;
+
+    --popover: 224 71% 4%;
+    --popover-foreground: 215 20.2% 65.1%;
+
+    --border: 216 34% 17%;
+    --input: 216 34% 17%;
+
+    --card: 224 71% 4%;
+    --card-foreground: 213 31% 91%;
+
+    --primary: 210 40% 98%;
+    --primary-foreground: 222.2 47.4% 1.2%;
+
+    --secondary: 222.2 47.4% 11.2%;
+    --secondary-foreground: 210 40% 98%;
+
+    --destructive: 0 63% 31%;
+    --destructive-foreground: 210 40% 98%;
+
+    --ring: 216 34% 17%;
+
+    --radius: 0.5rem;
+  }
+}
+
+@layer base {
+  * {
+    @apply border-border;
+  }
+  body {
+    @apply bg-background text-foreground;
+    font-feature-settings:
+      "rlig" 1,
+      "calt" 1;
+  }
+  .background-gradient {
+    background-color: #fff;
+    background-image: radial-gradient(
+        at 21% 11%,
+        rgba(186, 186, 233, 0.53) 0,
+        transparent 50%
+      ),
+      radial-gradient(at 85% 0, hsla(46, 57%, 78%, 0.52) 0, transparent 50%),
+      radial-gradient(at 91% 36%, rgba(194, 213, 255, 0.68) 0, transparent 50%),
+      radial-gradient(at 8% 40%, rgba(251, 218, 239, 0.46) 0, transparent 50%);
+  }
+}

packages/autotool/examples/02_nextjs/app/layout.tsx

@@ -0,0 +1,26 @@
+import type { Metadata } from "next";
+import { Inter } from "next/font/google";
+import { Toaster } from "sonner";
+import "./globals.css";
+
+const inter = Inter({ subsets: ["latin"] });
+
+export const metadata: Metadata = {
+  title: "Create Llama App",
+  description: "Generated by create-llama",
+};
+
+export default function RootLayout({
+  children,
+}: {
+  children: React.ReactNode;
+}) {
+  return (
+    <html lang="en">
+      <body className={inter.className}>
+        <Toaster />
+        {children}
+      </body>
+    </html>
+  );
+}

packages/autotool/examples/02_nextjs/app/page.tsx

@@ -0,0 +1,11 @@
+import { ChatSection } from "@/components/chat-section";
+
+export const runtime = "edge";
+
+export default function Home() {
+  return (
+    <main className="flex min-h-screen flex-col items-center gap-10 p-24 background-gradient">
+      <ChatSection />
+    </main>
+  );
+}

packages/autotool/examples/02_nextjs/components/chat-section.tsx

@@ -0,0 +1,35 @@
+"use client";
+import { chatWithAI } from "@/actions";
+import { ReactNode, useActionState } from "react";
+import { toast } from "sonner";
+
+export function ChatSection() {
+  const [state, formAction] = useActionState<ReactNode | null, FormData>(
+    async (state, payload) => {
+      const input = payload.get("input") as string | null;
+      if (!input) {
+        toast.error("Please type a message");
+        return null;
+      }
+      return chatWithAI(input);
+    },
+    null,
+  );
+  return (
+    <form>
+      <div className="border border-gray-400 p-2 max-w-md">{state}</div>
+      <input
+        className="border border-gray-400 p-2"
+        type="text"
+        name="input"
+        placeholder="Type your message here"
+      />
+      <button
+        className="bg-blue-500 hover:bg-blue-700 text-white font-bold py-2 px-4 rounded"
+        formAction={formAction}
+      >
+        Chat
+      </button>
+    </form>
+  );
+}

packages/autotool/examples/02_nextjs/components/location-card.tsx

@@ -0,0 +1,9 @@
+export function LocationCard() {
+  return (
+    <div className="border border-gray-400 p-2 max-w-md">
+      <h1>Weather</h1>
+      <p>San Francisco, CA</p>
+      <p>Sunny</p>
+    </div>
+  );
+}

packages/autotool/examples/02_nextjs/components/spinner.tsx

@@ -0,0 +1,23 @@
+export function Spinner() {
+  return (
+    <div role="status">
+      <svg
+        aria-hidden="true"
+        className="w-8 h-8 text-gray-200 animate-spin dark:text-gray-600 fill-blue-600"
+        viewBox="0 0 100 101"
+        fill="none"
+        xmlns="http://www.w3.org/2000/svg"
+      >
+        <path
+          d="M100 50.5908C100 78.2051 77.6142 100.591 50 100.591C22.3858 100.591 0 78.2051 0 50.5908C0 22.9766 22.3858 0.59082 50 0.59082C77.6142 0.59082 100 22.9766 100 50.5908ZM9.08144 50.5908C9.08144 73.1895 27.4013 91.5094 50 91.5094C72.5987 91.5094 90.9186 73.1895 90.9186 50.5908C90.9186 27.9921 72.5987 9.67226 50 9.67226C27.4013 9.67226 9.08144 27.9921 9.08144 50.5908Z"
+          fill="currentColor"
+        />
+        <path
+          d="M93.9676 39.0409C96.393 38.4038 97.8624 35.9116 97.0079 33.5539C95.2932 28.8227 92.871 24.3692 89.8167 20.348C85.8452 15.1192 80.8826 10.7238 75.2124 7.41289C69.5422 4.10194 63.2754 1.94025 56.7698 1.05124C51.7666 0.367541 46.6976 0.446843 41.7345 1.27873C39.2613 1.69328 37.813 4.19778 38.4501 6.62326C39.0873 9.04874 41.5694 10.4717 44.0505 10.1071C47.8511 9.54855 51.7191 9.52689 55.5402 10.0491C60.8642 10.7766 65.9928 12.5457 70.6331 15.2552C75.2735 17.9648 79.3347 21.5619 82.5849 25.841C84.9175 28.9121 86.7997 32.2913 88.1811 35.8758C89.083 38.2158 91.5421 39.6781 93.9676 39.0409Z"
+          fill="currentFill"
+        />
+      </svg>
+      <span className="sr-only">Loading...</span>
+    </div>
+  );
+}

packages/autotool/examples/02_nextjs/context/index.ts

@@ -0,0 +1,14 @@
+import type { createStreamableUI } from "ai/rsc";
+import { AsyncLocalStorage } from "node:async_hooks";
+
+type StreamableUI = ReturnType<typeof createStreamableUI>;
+
+const streamUIAsyncLocalStorage = new AsyncLocalStorage<StreamableUI>();
+
+export function getCurrentStreamableUI() {
+  return streamUIAsyncLocalStorage.getStore();
+}
+
+export function runWithStreamableUI<T>(streamUI: StreamableUI, fn: () => T): T {
+  return streamUIAsyncLocalStorage.run(streamUI, fn);
+}

packages/autotool/examples/02_nextjs/next.config.mjs

@@ -0,0 +1,6 @@
+import { withNext } from "@llamaindex/autotool/next";
+
+/** @type {import('next').NextConfig} */
+const nextConfig = {};
+
+export default withNext(nextConfig);

packages/autotool/examples/02_nextjs/package.json

@@ -0,0 +1,37 @@
+{
+  "name": "@llamaindex/autotool-02-next-example",
+  "private": true,
+  "version": "0.1.4",
+  "scripts": {
+    "dev": "next dev",
+    "build": "next build",
+    "start": "next start"
+  },
+  "dependencies": {
+    "@llamaindex/autotool": "workspace:*",
+    "@radix-ui/react-slot": "^1.0.2",
+    "ai": "^3.1.3",
+    "class-variance-authority": "^0.7.0",
+    "dotenv": "^16.3.1",
+    "llamaindex": "workspace:*",
+    "lucide-react": "^0.378.0",
+    "next": "14.3.0-canary.51",
+    "react": "^18.3.1",
+    "react-dom": "^18.3.1",
+    "react-markdown": "^9.0.1",
+    "react-syntax-highlighter": "^15.5.0",
+    "sonner": "^1.4.41",
+    "tailwind-merge": "^2.1.0"
+  },
+  "devDependencies": {
+    "@types/node": "^20.12.11",
+    "@types/react": "^18.3.1",
+    "@types/react-dom": "^18.3.0",
+    "@types/react-syntax-highlighter": "^15.5.11",
+    "autoprefixer": "^10.4.16",
+    "cross-env": "^7.0.3",
+    "postcss": "^8.4.32",
+    "tailwindcss": "^3.3.6",
+    "typescript": "^5.4.5"
+  }
+}

packages/autotool/examples/02_nextjs/postcss.config.js

@@ -0,0 +1,6 @@
+module.exports = {
+  plugins: {
+    tailwindcss: {},
+    autoprefixer: {},
+  },
+};

packages/autotool/examples/02_nextjs/tailwind.config.ts

@@ -0,0 +1,78 @@
+import type { Config } from "tailwindcss";
+import { fontFamily } from "tailwindcss/defaultTheme";
+
+const config: Config = {
+  darkMode: ["class"],
+  content: ["app/**/*.{ts,tsx}", "components/**/*.{ts,tsx}"],
+  theme: {
+    container: {
+      center: true,
+      padding: "2rem",
+      screens: {
+        "2xl": "1400px",
+      },
+    },
+    extend: {
+      colors: {
+        border: "hsl(var(--border))",
+        input: "hsl(var(--input))",
+        ring: "hsl(var(--ring))",
+        background: "hsl(var(--background))",
+        foreground: "hsl(var(--foreground))",
+        primary: {
+          DEFAULT: "hsl(var(--primary))",
+          foreground: "hsl(var(--primary-foreground))",
+        },
+        secondary: {
+          DEFAULT: "hsl(var(--secondary))",
+          foreground: "hsl(var(--secondary-foreground))",
+        },
+        destructive: {
+          DEFAULT: "hsl(var(--destructive) / <alpha-value>)",
+          foreground: "hsl(var(--destructive-foreground) / <alpha-value>)",
+        },
+        muted: {
+          DEFAULT: "hsl(var(--muted))",
+          foreground: "hsl(var(--muted-foreground))",
+        },
+        accent: {
+          DEFAULT: "hsl(var(--accent))",
+          foreground: "hsl(var(--accent-foreground))",
+        },
+        popover: {
+          DEFAULT: "hsl(var(--popover))",
+          foreground: "hsl(var(--popover-foreground))",
+        },
+        card: {
+          DEFAULT: "hsl(var(--card))",
+          foreground: "hsl(var(--card-foreground))",
+        },
+      },
+      borderRadius: {
+        xl: `calc(var(--radius) + 4px)`,
+        lg: `var(--radius)`,
+        md: `calc(var(--radius) - 2px)`,
+        sm: "calc(var(--radius) - 4px)",
+      },
+      fontFamily: {
+        sans: ["var(--font-sans)", ...fontFamily.sans],
+      },
+      keyframes: {
+        "accordion-down": {
+          from: { height: "0" },
+          to: { height: "var(--radix-accordion-content-height)" },
+        },
+        "accordion-up": {
+          from: { height: "var(--radix-accordion-content-height)" },
+          to: { height: "0" },
+        },
+      },
+      animation: {
+        "accordion-down": "accordion-down 0.2s ease-out",
+        "accordion-up": "accordion-up 0.2s ease-out",
+      },
+    },
+  },
+  plugins: [],
+};
+export default config;

packages/autotool/examples/02_nextjs/tool/index.tsx

@@ -0,0 +1,27 @@
+"use tool";
+import { getCurrentStreamableUI } from "@/context";
+
+export async function getMyUserID() {
+  const ui = getCurrentStreamableUI()!;
+  ui.update("Getting user ID...");
+  await new Promise((resolve) => setTimeout(resolve, 2000));
+  return "12345";
+}
+
+export async function showUserInfo(userId: string) {
+  const ui = getCurrentStreamableUI()!;
+  ui.update("Getting user info...");
+  await new Promise((resolve) => setTimeout(resolve, 2000));
+  ui.update(
+    <div>
+      User ID: {userId}
+      <br />
+      Name: John Doe
+    </div>,
+  );
+  return `User ID: ${userId}\nName: John Doe\nEmail: alex@gmail.com\nPhone: 123-456-7890\nAddress: 123 Main St\nCity: San Francisco\nState: CA\nZip: 94105\nCountry: USA\n`;
+}
+
+export function getWeather(address: string) {
+  return `The weather in ${address} is sunny!`;
+}

packages/autotool/examples/02_nextjs/tsconfig.json

@@ -0,0 +1,28 @@
+{
+  "compilerOptions": {
+    "target": "es5",
+    "lib": ["dom", "dom.iterable", "esnext"],
+    "allowJs": true,
+    "skipLibCheck": true,
+    "strict": true,
+    "noEmit": true,
+    "esModuleInterop": true,
+    "module": "esnext",
+    "moduleResolution": "bundler",
+    "resolveJsonModule": true,
+    "isolatedModules": true,
+    "jsx": "preserve",
+    "incremental": true,
+    "plugins": [
+      {
+        "name": "next"
+      }
+    ],
+    "paths": {
+      "@/*": ["./*"]
+    },
+    "forceConsistentCasingInFileNames": true
+  },
+  "include": ["next-env.d.ts", "**/*.ts", "**/*.tsx", ".next/types/**/*.ts"],
+  "exclude": ["node_modules"]
+}

packages/autotool/package.json

@@ -0,0 +1,82 @@
+{
+  "name": "@llamaindex/autotool",
+  "type": "module",
+  "version": "0.0.1",
+  "description": "auto transpile your JS function to LLM Agent compatible",
+  "files": [
+    "dist",
+    "CHANGELOG.md"
+  ],
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js",
+      "require": "./dist/index.cjs",
+      "default": "./dist/index.js"
+    },
+    "./next": {
+      "types": "./dist/next.d.ts",
+      "import": "./dist/next.js",
+      "require": "./dist/next.cjs",
+      "default": "./dist/next.js"
+    },
+    "./webpack": {
+      "types": "./dist/webpack.d.ts",
+      "import": "./dist/webpack.js",
+      "require": "./dist/webpack.cjs",
+      "default": "./dist/webpack.js"
+    },
+    "./vite": {
+      "types": "./dist/vite.d.ts",
+      "import": "./dist/vite.js",
+      "require": "./dist/vite.cjs",
+      "default": "./dist/vite.js"
+    },
+    "./loader": {
+      "types": "./dist/loader.d.ts",
+      "import": "./dist/loader.js",
+      "require": "./dist/loader.cjs",
+      "default": "./dist/loader.js"
+    },
+    "./node": "./dist/node.js"
+  },
+  "scripts": {
+    "build": "bunchee",
+    "dev": "bunchee --watch"
+  },
+  "dependencies": {
+    "@swc/core": "^1.5.5",
+    "jotai": "^2.8.0",
+    "typedoc": "^0.25.13",
+    "unplugin": "^1.10.1"
+  },
+  "peerDependencies": {
+    "llamaindex": "^0.3.12",
+    "openai": "^4",
+    "typescript": "^4"
+  },
+  "peerDependenciesMeta": {
+    "openai": {
+      "optional": true
+    },
+    "llamaindex": {
+      "optional": true
+    },
+    "typescript": {
+      "optional": true
+    }
+  },
+  "devDependencies": {
+    "@swc/types": "^0.1.6",
+    "@types/json-schema": "^7.0.15",
+    "@types/node": "^20.12.11",
+    "bunchee": "^5.1.5",
+    "llamaindex": "workspace:*",
+    "next": "14.2.3",
+    "rollup": "^4.17.2",
+    "tsx": "^4.9.3",
+    "typescript": "^5.4.5",
+    "vitest": "^1.6.0",
+    "webpack": "^5.91.0"
+  }
+}

packages/autotool/src/compiler.ts

@@ -0,0 +1,103 @@
+import type {
+  JSONSchema7,
+  JSONSchema7Definition,
+  JSONSchema7TypeName,
+} from "json-schema";
+import type { ToolMetadata } from "llamaindex";
+import type { SourceMapInput } from "rollup";
+import td from "typedoc";
+import type { SourceMapCompact } from "unplugin";
+import type { InfoString } from "./internal";
+
+export const isToolFile = (url: string) => /tool\.[jt]sx?$/.test(url);
+export const isJSorTS = (url: string) => /\.m?[jt]sx?$/.test(url);
+
+async function parseRoot(entryPoint: string) {
+  const app = await td.Application.bootstrapWithPlugins(
+    {
+      entryPoints: [entryPoint],
+    },
+    [
+      new td.TypeDocReader(),
+      new td.PackageJsonReader(),
+      new td.TSConfigReader(),
+    ],
+  );
+  const project = await app.convert();
+
+  if (project) {
+    return app.serializer.projectToObject(project, process.cwd());
+  }
+  throw new Error("Failed to parse root");
+}
+
+export async function transformAutoTool(
+  code: string,
+  url: string,
+): Promise<{
+  code: string;
+  map?: SourceMapInput | SourceMapCompact | null;
+}> {
+  const json = await parseRoot(url);
+  const children = json.children;
+  if (Array.isArray(children)) {
+    const schema = {
+      type: "object",
+      properties: {} as {
+        [key: string]: JSONSchema7Definition;
+      },
+      additionalItems: false,
+      required: [] as string[],
+    } satisfies JSONSchema7;
+    const info: InfoString = {
+      originalFunction: undefined,
+      parameterMapping: {},
+    };
+    children.forEach((child) => {
+      // replace starting and ending quotes, to make it a function in the runtime
+      info.originalFunction = child.name;
+      const metadata: ToolMetadata = {
+        name: child.name,
+        description: "",
+        parameters: schema,
+      };
+      child.signatures?.forEach((signature) => {
+        const description = signature.comment?.summary
+          .map((x) => x.text)
+          .join("\n");
+        if (description) {
+          metadata.description += description;
+        }
+        signature.parameters?.map((parameter, idx) => {
+          if (parameter.type?.type === "intrinsic") {
+            // parameter.type.name
+            schema.properties[parameter.name as string] = {
+              type: parameter.type.name as JSONSchema7TypeName,
+              description: parameter.comment?.summary
+                .map((x) => x.text)
+                .join("\n"),
+            } as JSONSchema7Definition;
+            schema.required.push(parameter.name as string);
+            info.parameterMapping[parameter.name as string] = idx;
+          }
+        });
+      });
+      const infoJSON = JSON.stringify(info)
+        // remove quotes from `originalFunction` value
+        .replace(/"originalFunction":"(.*?)"/g, '"originalFunction":$1');
+      code =
+        code + `\ninjectMetadata(${JSON.stringify(metadata)}, ${infoJSON});`;
+    });
+  }
+  if (
+    !/^import\s+{\sinjectMetadata\s}\s+from\s+['"]@llamaindex\/tool['"]/.test(
+      code,
+    )
+  ) {
+    code = `import {injectMetadata} from '@llamaindex/autotool';\n${code}`;
+  }
+  return {
+    code,
+    map: null,
+  };
+}

packages/autotool/src/index.ts

@@ -0,0 +1,82 @@
+import { atom } from "jotai/vanilla";
+import type { BaseToolWithCall, ToolMetadata } from "llamaindex";
+import type { ChatCompletionTool } from "openai/resources/chat/completions";
+import { store, toolMetadataAtom, toolsAtom, type Info } from "./internal";
+
+export type { Info };
+
+/**
+ * @internal This function is used by the compiler to inject metadata into the source code.
+ */
+export function injectMetadata(metadata: ToolMetadata, info: Info) {
+  store.get(toolMetadataAtom).push([metadata, info]);
+}
+
+const openaiToolsAtom = atom<ChatCompletionTool[]>((get) => {
+  const metadata = get(toolMetadataAtom);
+  return metadata.map(([metadata]) => ({
+    type: "function",
+    function: {
+      parameters: metadata.parameters,
+      name: metadata.name,
+      description: metadata.description,
+    },
+  }));
+});
+
+const llamaindexToolsAtom = atom<BaseToolWithCall[]>((get) => {
+  const metadata = get(toolMetadataAtom);
+  const fns = get(toolsAtom);
+  return metadata.map(([metadata, info]) => ({
+    call: (input: Record<string, unknown>) => {
+      const args = Object.entries(info.parameterMapping).reduce(
+        (arr, [name, idx]) => {
+          arr[idx] = input[name];
+          return arr;
+        },
+        [] as unknown[],
+      );
+      const fn = fns[metadata.name] ?? info.originalFunction;
+      if (!fn) {
+        throw new Error(`Cannot find function to call: ${metadata.name}`);
+      }
+      return fn(...args);
+    },
+    metadata,
+  }));
+});
+
+export function convertTools(format: "openai"): ChatCompletionTool[];
+export function convertTools(format: "llamaindex"): BaseToolWithCall[];
+export function convertTools(
+  format: string,
+): ChatCompletionTool[] | BaseToolWithCall[] {
+  switch (format) {
+    case "openai": {
+      return store.get(openaiToolsAtom);
+    }
+    case "llamaindex": {
+      return store.get(llamaindexToolsAtom);
+    }
+  }
+  throw new Error(`Unknown format: ${format}`);
+}
+
+/**
+ * Call a tool by name with the given input.
+ */
+export function callTool(
+  name: string,
+  input: string | Record<string, unknown>,
+): unknown | Promise<unknown> {
+  const tools = store.get(llamaindexToolsAtom);
+  const targetTool = tools.find((tool) => tool.metadata.name === name);
+  if (!targetTool) {
+    throw new Error(`Cannot find tool: ${name}`);
+  }
+  return targetTool.call(
+    // for OpenAI, input is a string
+    // for ClaudeAI, input is an object
+    typeof input === "string" ? JSON.parse(input) : input,
+  );
+}

packages/autotool/src/internal/index.ts

@@ -0,0 +1,26 @@
+import { atom, createStore } from "jotai/vanilla";
+import type { ToolMetadata } from "llamaindex";
+
+export type Info = {
+  originalFunction?: (...args: any[]) => any;
+  /**
+   * In current LLM, it doesn't support non-object parameter, so we mock arguments as object, and use this mapping to convert it back.
+   */
+  parameterMapping: Record<string, number>;
+};
+
+/**
+ * This is used in parser side to store the original function and parameter mapping.
+ *
+ * In the runtime, originalFunction is a JS function.
+ *
+ * @internal
+ */
+export type InfoString = {
+  originalFunction?: string;
+  parameterMapping: Record<string, number>;
+};
+
+export const store = createStore();
+export const toolMetadataAtom = atom<[ToolMetadata, Info][]>([]);
+export const toolsAtom = atom<Record<string, (...args: any[]) => any>>({});

packages/autotool/src/loader.ts

@@ -0,0 +1,38 @@
+/**
+ * This is a node module loader hook that injects metadata into the source code.
+ *
+ * @module
+ */
+import { parse } from "@swc/core";
+import type { ExpressionStatement } from "@swc/types";
+import type { LoadHook } from "node:module";
+import { fileURLToPath } from "node:url";
+import { isJSorTS, isToolFile, transformAutoTool } from "./compiler";
+
+export const load: LoadHook = async (url, context, nextLoad) => {
+  const output = await nextLoad(url, context);
+  if (typeof output.source === "string" && isJSorTS(url)) {
+    const isTool = isToolFile(url);
+    const hasToolDirective = (await parse(output.source)).body
+      .filter(
+        (node): node is ExpressionStatement =>
+          node.type === "ExpressionStatement",
+      )
+      .some(
+        (node) =>
+          node.expression.type === "StringLiteral" &&
+          node.expression.value === "use tool",
+      );
+    if (isTool || hasToolDirective) {
+      const { code } = await transformAutoTool(
+        output.source,
+        fileURLToPath(url),
+      );
+      return {
+        ...output,
+        source: code,
+      };
+    }
+  }
+  return output;
+};

packages/autotool/src/next.ts

@@ -0,0 +1,13 @@
+import type { NextConfig } from "next";
+import webpackPlugin from "./webpack";
+
+export function withNext(config: NextConfig) {
+  return {
+    ...config,
+    webpack: (webpackConfig: any, context: any) => {
+      webpackConfig = config.webpack?.(webpackConfig, context) ?? webpackConfig;
+      webpackConfig.plugins.push(webpackPlugin());
+      return webpackConfig;
+    },
+  };
+}

packages/autotool/src/node.ts

@@ -0,0 +1,16 @@
+/**
+ * @example
+ * ```shell
+ * node --import @llamaindex/autotool/node ./dist/index.js
+ * ```
+ *
+ * @example
+ * ```shell
+ * node --import tsx --import @llamaindex/autotool/node ./src/index.ts
+ * ```
+ *
+ * @module
+ */
+import { register } from "node:module";
+
+register("./loader.js", import.meta.url);

packages/autotool/src/plugin.ts

@@ -0,0 +1,35 @@
+import { parse } from "@swc/core";
+import type { ExpressionStatement } from "@swc/types";
+import { createUnplugin, type UnpluginFactory } from "unplugin";
+import { isJSorTS, isToolFile, transformAutoTool } from "./compiler";
+
+export interface Options {}
+
+const name = "llama-index-tool";
+
+export const unpluginFactory: UnpluginFactory<Options | undefined> = () => ({
+  name,
+  async transform(code, id) {
+    if (!isJSorTS(id)) {
+      return code;
+    }
+    const isTool = isToolFile(id);
+    const hasToolDirective = (await parse(code)).body
+      .filter(
+        (node): node is ExpressionStatement =>
+          node.type === "ExpressionStatement",
+      )
+      .some(
+        (node) =>
+          node.expression.type === "StringLiteral" &&
+          node.expression.value === "use tool",
+      );
+    if (isTool || hasToolDirective) {
+      return transformAutoTool(code, id);
+    }
+  },
+});
+
+export const unplugin = /* #__PURE__ */ createUnplugin(unpluginFactory);
+
+export default unplugin;

packages/autotool/src/vite.ts

@@ -0,0 +1,6 @@
+import { createVitePlugin } from "unplugin";
+import { unpluginFactory } from "./plugin";
+
+const vitePlugin = createVitePlugin(unpluginFactory);
+
+export default vitePlugin;

packages/autotool/src/webpack.ts

@@ -0,0 +1,6 @@
+import { createWebpackPlugin } from "unplugin";
+import { unpluginFactory } from "./plugin";
+
+const webpackPlugin = createWebpackPlugin(unpluginFactory);
+
+export default webpackPlugin;

packages/autotool/tsconfig.json

@@ -0,0 +1,19 @@
+{
+  "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    "target": "ESNext",
+    "module": "ESNext",
+    "moduleResolution": "bundler",
+    "outDir": "./lib",
+    "types": ["node"]
+  },
+  "include": ["./src"],
+  "references": [
+    {
+      "path": "../core/tsconfig.json"
+    },
+    {
+      "path": "../env/tsconfig.json"
+    }
+  ]
+}

packages/core/.madgerc

@@ -0,0 +1,7 @@
+{
+	"detectiveOptions": {
+		"ts": {
+			"skipTypeImports": true
+		}
+	}
+}

packages/core/CHANGELOG.md

@@ -1,5 +1,76 @@
 # llamaindex
 
+## 0.3.12
+
+### Patch Changes
+
+- 34fb1d8: fix: cloudflare dev
+
+## 0.3.11
+
+### Patch Changes
+
+- e072c45: fix: remove non-standard API `pipeline`
+- 9e133ac: refactor: remove `defaultFS` from parameters
+
+  We don't accept passing fs in the parameter since it's unnecessary for a determined JS environment.
+
+  This was a polyfill way for the non-Node.js environment, but now we use another way to polyfill APIs.
+
+- 447105a: Improve Gemini message and context preparation
+- 320be3f: Force ChromaDB version to 1.7.3 (to prevent NextJS issues)
+- Updated dependencies [e072c45]
+- Updated dependencies [9e133ac]
+  - @llamaindex/env@0.1.3
+
+## 0.3.10
+
+### Patch Changes
+
+- 4aba02e: feat: support gpt4-o
+
+## 0.3.9
+
+### Patch Changes
+
+- c3747d0: fix: import `@xenova/transformers`
+
+  For now, if you use llamaindex in next.js, you need to add a plugin from `llamaindex/next` to ensure some module resolutions are correct.
+
+## 0.3.8
+
+### Patch Changes
+
+- ce94780: Add page number to read PDFs and use generated IDs for PDF and markdown content
+
+## 0.3.7
+
+### Patch Changes
+
+- b6a6606: feat: allow change host of ollama
+- b6a6606: chore: export ollama in default js runtime
+
+## 0.3.6
+
+### Patch Changes
+
+- efa326a: chore: update package.json
+- Updated dependencies [efa326a]
+- Updated dependencies [efa326a]
+  - @llamaindex/env@0.1.2
+
+## 0.3.5
+
+### Patch Changes
+
+- bc7a11c: fix: inline ollama build
+- 2fe2b81: fix: filter with multiple filters in ChromaDB
+- 5596e31: feat: improve `@llamaindex/env`
+- e74fe88: fix: change <-> to <=> in the SELECT query
+- be5df5b: fix: anthropic agent on multiple chat
+- Updated dependencies [5596e31]
+  - @llamaindex/env@0.1.1
+
 ## 0.3.4
 
 ### Patch Changes

packages/core/e2e/CHANGELOG.md

@@ -1,5 +1,25 @@
 # @llamaindex/core-e2e
 
+## 0.0.6
+
+### Patch Changes
+
+- 34fb1d8: fix: cloudflare dev
+
+## 0.0.5
+
+### Patch Changes
+
+- c3747d0: fix: import `@xenova/transformers`
+
+  For now, if you use llamaindex in next.js, you need to add a plugin from `llamaindex/next` to ensure some module resolutions are correct.
+
+## 0.0.4
+
+### Patch Changes
+
+- be5df5b: fix: anthropic agent on multiple chat
+
 ## 0.0.3
 
 ### Patch Changes

packages/core/e2e/examples/cloudflare-worker-agent/CHANGELOG.md

@@ -1,5 +1,70 @@
 # @llamaindex/cloudflare-worker-agent-test
 
+## 0.0.13
+
+### Patch Changes
+
+- 34fb1d8: fix: cloudflare dev
+- Updated dependencies [34fb1d8]
+  - llamaindex@0.3.12
+
+## 0.0.12
+
+### Patch Changes
+
+- Updated dependencies [e072c45]
+- Updated dependencies [9e133ac]
+- Updated dependencies [447105a]
+- Updated dependencies [320be3f]
+  - llamaindex@0.3.11
+
+## 0.0.11
+
+### Patch Changes
+
+- Updated dependencies [4aba02e]
+  - llamaindex@0.3.10
+
+## 0.0.10
+
+### Patch Changes
+
+- Updated dependencies [c3747d0]
+  - llamaindex@0.3.9
+
+## 0.0.9
+
+### Patch Changes
+
+- Updated dependencies [ce94780]
+  - llamaindex@0.3.8
+
+## 0.0.8
+
+### Patch Changes
+
+- Updated dependencies [b6a6606]
+- Updated dependencies [b6a6606]
+  - llamaindex@0.3.7
+
+## 0.0.7
+
+### Patch Changes
+
+- Updated dependencies [efa326a]
+  - llamaindex@0.3.6
+
+## 0.0.6
+
+### Patch Changes
+
+- Updated dependencies [bc7a11c]
+- Updated dependencies [2fe2b81]
+- Updated dependencies [5596e31]
+- Updated dependencies [e74fe88]
+- Updated dependencies [be5df5b]
+  - llamaindex@0.3.5
+
 ## 0.0.5
 
 ### Patch Changes

packages/core/e2e/examples/cloudflare-worker-agent/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/cloudflare-worker-agent-test",
-  "version": "0.0.5",
+  "version": "0.0.13",
   "type": "module",
   "private": true,
   "scripts": {
@@ -12,11 +12,13 @@
     "cf-typegen": "wrangler types"
   },
   "devDependencies": {
-    "@cloudflare/vitest-pool-workers": "^0.2.3",
-    "@cloudflare/workers-types": "^4.20240423.0",
+    "@cloudflare/vitest-pool-workers": "^0.2.6",
+    "@cloudflare/workers-types": "^4.20240512.0",
+    "@vitest/runner": "1.3.0",
+    "@vitest/snapshot": "1.3.0",
     "typescript": "^5.4.5",
     "vitest": "1.3.0",
-    "wrangler": "^3.52.0"
+    "wrangler": "^3.56.0"
   },
   "dependencies": {
     "llamaindex": "workspace:*"

packages/core/e2e/examples/nextjs-agent/CHANGELOG.md

@@ -1,5 +1,73 @@
 # @llamaindex/next-agent-test
 
+## 0.1.13
+
+### Patch Changes
+
+- Updated dependencies [34fb1d8]
+  - llamaindex@0.3.12
+
+## 0.1.12
+
+### Patch Changes
+
+- Updated dependencies [e072c45]
+- Updated dependencies [9e133ac]
+- Updated dependencies [447105a]
+- Updated dependencies [320be3f]
+  - llamaindex@0.3.11
+
+## 0.1.11
+
+### Patch Changes
+
+- Updated dependencies [4aba02e]
+  - llamaindex@0.3.10
+
+## 0.1.10
+
+### Patch Changes
+
+- c3747d0: fix: import `@xenova/transformers`
+
+  For now, if you use llamaindex in next.js, you need to add a plugin from `llamaindex/next` to ensure some module resolutions are correct.
+
+- Updated dependencies [c3747d0]
+  - llamaindex@0.3.9
+
+## 0.1.9
+
+### Patch Changes
+
+- Updated dependencies [ce94780]
+  - llamaindex@0.3.8
+
+## 0.1.8
+
+### Patch Changes
+
+- Updated dependencies [b6a6606]
+- Updated dependencies [b6a6606]
+  - llamaindex@0.3.7
+
+## 0.1.7
+
+### Patch Changes
+
+- Updated dependencies [efa326a]
+  - llamaindex@0.3.6
+
+## 0.1.6
+
+### Patch Changes
+
+- Updated dependencies [bc7a11c]
+- Updated dependencies [2fe2b81]
+- Updated dependencies [5596e31]
+- Updated dependencies [e74fe88]
+- Updated dependencies [be5df5b]
+  - llamaindex@0.3.5
+
 ## 0.1.5
 
 ### Patch Changes

packages/core/e2e/examples/nextjs-agent/next.config.mjs

@@ -1,4 +1,6 @@
 /** @type {import('next').NextConfig} */
 const nextConfig = {};
 
-export default nextConfig;
+import withLlamaIndex from "llamaindex/next";
+
+export default withLlamaIndex(nextConfig);

packages/core/e2e/examples/nextjs-agent/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/next-agent-test",
-  "version": "0.1.5",
+  "version": "0.1.13",
   "private": true,
   "scripts": {
     "dev": "next dev",
@@ -9,20 +9,20 @@
     "lint": "next lint"
   },
   "dependencies": {
-    "ai": "^3.0.34",
+    "ai": "^3.1.3",
     "llamaindex": "workspace:*",
     "next": "14.2.3",
-    "react": "18.2.0",
-    "react-dom": "18.2.0"
+    "react": "18.3.1",
+    "react-dom": "18.3.1"
   },
   "devDependencies": {
-    "@types/node": "^20",
-    "@types/react": "^18.2.79",
-    "@types/react-dom": "^18.2.25",
-    "eslint": "^8",
+    "@types/node": "^20.12.11",
+    "@types/react": "^18.3.1",
+    "@types/react-dom": "^18.3.0",
+    "eslint": "^8.57.0",
     "eslint-config-next": "14.2.3",
     "postcss": "^8",
     "tailwindcss": "^3.4.1",
-    "typescript": "^5"
+    "typescript": "^5.4.5"
   }
 }

packages/core/e2e/examples/nextjs-edge-runtime/CHANGELOG.md

@@ -1,5 +1,52 @@
 # test-edge-runtime
 
+## 0.1.12
+
+### Patch Changes
+
+- Updated dependencies [34fb1d8]
+  - llamaindex@0.3.12
+
+## 0.1.11
+
+### Patch Changes
+
+- Updated dependencies [e072c45]
+- Updated dependencies [9e133ac]
+- Updated dependencies [447105a]
+- Updated dependencies [320be3f]
+  - llamaindex@0.3.11
+
+## 0.1.10
+
+### Patch Changes
+
+- Updated dependencies [4aba02e]
+  - llamaindex@0.3.10
+
+## 0.1.9
+
+### Patch Changes
+
+- c3747d0: fix: import `@xenova/transformers`
+
+  For now, if you use llamaindex in next.js, you need to add a plugin from `llamaindex/next` to ensure some module resolutions are correct.
+
+- Updated dependencies [c3747d0]
+  - llamaindex@0.3.9
+
+## 0.1.8
+
+### Patch Changes
+
+- @llamaindex/edge@0.3.6
+
+## 0.1.7
+
+### Patch Changes
+
+- @llamaindex/edge@0.3.5
+
 ## 0.1.6
 
 ### Patch Changes

packages/core/e2e/examples/nextjs-edge-runtime/next.config.mjs

@@ -1,4 +1,6 @@
 /** @type {import('next').NextConfig} */
 const nextConfig = {};
 
-export default nextConfig;
+import withLlamaIndex from "llamaindex/next";
+
+export default withLlamaIndex(nextConfig);

packages/core/e2e/examples/nextjs-edge-runtime/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/nextjs-edge-runtime-test",
-  "version": "0.1.6",
+  "version": "0.1.12",
   "private": true,
   "scripts": {
     "dev": "next dev",
@@ -8,15 +8,15 @@
     "start": "next start"
   },
   "dependencies": {
-    "@llamaindex/edge": "workspace:*",
-    "next": "14.1.3",
-    "react": "^18",
-    "react-dom": "^18"
+    "llamaindex": "workspace:*",
+    "next": "14.2.3",
+    "react": "^18.3.1",
+    "react-dom": "^18.3.1"
   },
   "devDependencies": {
-    "@types/node": "^20.12.7",
-    "@types/react": "^18.2.79",
-    "@types/react-dom": "^18.2.25",
-    "typescript": "^5"
+    "@types/node": "^20.12.11",
+    "@types/react": "^18.3.1",
+    "@types/react-dom": "^18.3.0",
+    "typescript": "^5.4.5"
   }
 }

packages/core/e2e/examples/nextjs-edge-runtime/src/app/page.module.css

@@ -1,232 +0,0 @@
-.main {
-  display: flex;
-  flex-direction: column;
-  justify-content: space-between;
-  align-items: center;
-  padding: 6rem;
-  min-height: 100vh;
-}
-
-.description {
-  display: inherit;
-  justify-content: inherit;
-  align-items: inherit;
-  font-size: 0.85rem;
-  max-width: var(--max-width);
-  width: 100%;
-  z-index: 2;
-  font-family: var(--font-mono);
-}
-
-.description a {
-  display: flex;
-  justify-content: center;
-  align-items: center;
-  gap: 0.5rem;
-}
-
-.description p {
-  position: relative;
-  margin: 0;
-  padding: 1rem;
-  background-color: rgba(var(--callout-rgb), 0.5);
-  border: 1px solid rgba(var(--callout-border-rgb), 0.3);
-  border-radius: var(--border-radius);
-}
-
-.code {
-  font-weight: 700;
-  font-family: var(--font-mono);
-}
-
-.grid {
-  display: grid;
-  grid-template-columns: repeat(4, minmax(25%, auto));
-  max-width: 100%;
-  width: var(--max-width);
-}
-
-.card {
-  padding: 1rem 1.2rem;
-  border-radius: var(--border-radius);
-  background: rgba(var(--card-rgb), 0);
-  border: 1px solid rgba(var(--card-border-rgb), 0);
-  transition:
-    background 200ms,
-    border 200ms;
-}
-
-.card span {
-  display: inline-block;
-  transition: transform 200ms;
-}
-
-.card h2 {
-  font-weight: 600;
-  margin-bottom: 0.7rem;
-}
-
-.card p {
-  margin: 0;
-  opacity: 0.6;
-  font-size: 0.9rem;
-  line-height: 1.5;
-  max-width: 30ch;
-  text-wrap: balance;
-}
-
-.center {
-  display: flex;
-  justify-content: center;
-  align-items: center;
-  position: relative;
-  padding: 4rem 0;
-}
-
-.center::before {
-  background: var(--secondary-glow);
-  border-radius: 50%;
-  width: 480px;
-  height: 360px;
-  margin-left: -400px;
-}
-
-.center::after {
-  background: var(--primary-glow);
-  width: 240px;
-  height: 180px;
-  z-index: -1;
-}
-
-.center::before,
-.center::after {
-  content: "";
-  left: 50%;
-  position: absolute;
-  filter: blur(45px);
-  transform: translateZ(0);
-}
-
-.logo {
-  position: relative;
-}
-/* Enable hover only on non-touch devices */
-@media (hover: hover) and (pointer: fine) {
-  .card:hover {
-    background: rgba(var(--card-rgb), 0.1);
-    border: 1px solid rgba(var(--card-border-rgb), 0.15);
-  }
-
-  .card:hover span {
-    transform: translateX(4px);
-  }
-}
-
-@media (prefers-reduced-motion) {
-  .card:hover span {
-    transform: none;
-  }
-}
-
-/* Mobile */
-@media (max-width: 700px) {
-  .content {
-    padding: 4rem;
-  }
-
-  .grid {
-    grid-template-columns: 1fr;
-    margin-bottom: 120px;
-    max-width: 320px;
-    text-align: center;
-  }
-
-  .card {
-    padding: 1rem 2.5rem;
-  }
-
-  .card h2 {
-    margin-bottom: 0.5rem;
-  }
-
-  .center {
-    padding: 8rem 0 6rem;
-  }
-
-  .center::before {
-    transform: none;
-    height: 300px;
-  }
-
-  .description {
-    font-size: 0.8rem;
-  }
-
-  .description a {
-    padding: 1rem;
-  }
-
-  .description p,
-  .description div {
-    display: flex;
-    justify-content: center;
-    position: fixed;
-    width: 100%;
-  }
-
-  .description p {
-    align-items: center;
-    inset: 0 0 auto;
-    padding: 2rem 1rem 1.4rem;
-    border-radius: 0;
-    border: none;
-    border-bottom: 1px solid rgba(var(--callout-border-rgb), 0.25);
-    background: linear-gradient(
-      to bottom,
-      rgba(var(--background-start-rgb), 1),
-      rgba(var(--callout-rgb), 0.5)
-    );
-    background-clip: padding-box;
-    backdrop-filter: blur(24px);
-  }
-
-  .description div {
-    align-items: flex-end;
-    pointer-events: none;
-    inset: auto 0 0;
-    padding: 2rem;
-    height: 200px;
-    background: linear-gradient(
-      to bottom,
-      transparent 0%,
-      rgb(var(--background-end-rgb)) 40%
-    );
-    z-index: 1;
-  }
-}
-
-/* Tablet and Smaller Desktop */
-@media (min-width: 701px) and (max-width: 1120px) {
-  .grid {
-    grid-template-columns: repeat(2, 50%);
-  }
-}
-
-@media (prefers-color-scheme: dark) {
-  .vercelLogo {
-    filter: invert(1);
-  }
-
-  .logo {
-    filter: invert(1) drop-shadow(0 0 0.3rem #ffffff70);
-  }
-}
-
-@keyframes rotate {
-  from {
-    transform: rotate(360deg);
-  }
-  to {
-    transform: rotate(0deg);
-  }
-}

packages/core/e2e/examples/nextjs-edge-runtime/src/app/page.tsx

@@ -1,98 +1,20 @@
-import Image from "next/image";
-import "../utils/llm";
-import styles from "./page.module.css";
+import { tokenizerResultPromise } from "@/utils/llm";
+import { use } from "react";
 
 export const runtime = "edge";
 
 export default function Home() {
+  const result = use(tokenizerResultPromise);
   return (
-    <main className={styles.main}>
-      <div className={styles.description}>
-        <p>
-          Get started by editing&nbsp;
-          <code className={styles.code}>src/app/page.tsx</code>
-        </p>
+    <main>
+      <div>
+        <h1>Next.js Edge Runtime</h1>
         <div>
-          <a
-            href="https://vercel.com?utm_source=create-next-app&utm_medium=appdir-template&utm_campaign=create-next-app"
-            target="_blank"
-            rel="noopener noreferrer"
-          >
-            By{" "}
-            <Image
-              src="/vercel.svg"
-              alt="Vercel Logo"
-              className={styles.vercelLogo}
-              width={100}
-              height={24}
-              priority
-            />
-          </a>
+          {result.map((value, index) => (
+            <span key={index}>{value}</span>
+          ))}
         </div>
       </div>
-
-      <div className={styles.center}>
-        <Image
-          className={styles.logo}
-          src="/next.svg"
-          alt="Next.js Logo"
-          width={180}
-          height={37}
-          priority
-        />
-      </div>
-
-      <div className={styles.grid}>
-        <a
-          href="https://nextjs.org/docs?utm_source=create-next-app&utm_medium=appdir-template&utm_campaign=create-next-app"
-          className={styles.card}
-          target="_blank"
-          rel="noopener noreferrer"
-        >
-          <h2>
-            Docs <span>-&gt;</span>
-          </h2>
-          <p>Find in-depth information about Next.js features and API.</p>
-        </a>
-
-        <a
-          href="https://nextjs.org/learn?utm_source=create-next-app&utm_medium=appdir-template&utm_campaign=create-next-app"
-          className={styles.card}
-          target="_blank"
-          rel="noopener noreferrer"
-        >
-          <h2>
-            Learn <span>-&gt;</span>
-          </h2>
-          <p>Learn about Next.js in an interactive course with&nbsp;quizzes!</p>
-        </a>
-
-        <a
-          href="https://vercel.com/templates?framework=next.js&utm_source=create-next-app&utm_medium=appdir-template&utm_campaign=create-next-app"
-          className={styles.card}
-          target="_blank"
-          rel="noopener noreferrer"
-        >
-          <h2>
-            Templates <span>-&gt;</span>
-          </h2>
-          <p>Explore starter templates for Next.js.</p>
-        </a>
-
-        <a
-          href="https://vercel.com/new?utm_source=create-next-app&utm_medium=appdir-template&utm_campaign=create-next-app"
-          className={styles.card}
-          target="_blank"
-          rel="noopener noreferrer"
-        >
-          <h2>
-            Deploy <span>-&gt;</span>
-          </h2>
-          <p>
-            Instantly deploy your Next.js site to a shareable URL with Vercel.
-          </p>
-        </a>
-      </div>
     </main>
   );
 }

packages/core/e2e/examples/nextjs-edge-runtime/src/utils/llm.ts

@@ -1,9 +1,23 @@
-"use server";
 // test runtime
 import "llamaindex";
+import { ClipEmbedding } from "llamaindex/embeddings/ClipEmbedding";
 import "llamaindex/readers/SimpleDirectoryReader";
 
 // @ts-expect-error
 if (typeof EdgeRuntime !== "string") {
   throw new Error("Expected run in EdgeRuntime");
 }
+
+export const tokenizerResultPromise = new Promise<number[]>(
+  (resolve, reject) => {
+    const embedding = new ClipEmbedding();
+    //#region make sure @xenova/transformers is working in edge runtime
+    embedding
+      .getTokenizer()
+      .then((tokenizer) => {
+        resolve(tokenizer.encode("hello world"));
+      })
+      .catch(reject);
+    //#endregion
+  },
+);

packages/core/e2e/examples/nextjs-edge-runtime/tsconfig.json

@@ -1,5 +1,6 @@
 {
   "compilerOptions": {
+    "target": "ESNext",
     "lib": ["dom", "dom.iterable", "esnext"],
     "outDir": "./dist",
     "allowJs": true,

packages/core/e2e/examples/waku-query-engine/CHANGELOG.md

@@ -1,5 +1,69 @@
 # @llamaindex/waku-query-engine-test
 
+## 0.0.13
+
+### Patch Changes
+
+- Updated dependencies [34fb1d8]
+  - llamaindex@0.3.12
+
+## 0.0.12
+
+### Patch Changes
+
+- Updated dependencies [e072c45]
+- Updated dependencies [9e133ac]
+- Updated dependencies [447105a]
+- Updated dependencies [320be3f]
+  - llamaindex@0.3.11
+
+## 0.0.11
+
+### Patch Changes
+
+- Updated dependencies [4aba02e]
+  - llamaindex@0.3.10
+
+## 0.0.10
+
+### Patch Changes
+
+- Updated dependencies [c3747d0]
+  - llamaindex@0.3.9
+
+## 0.0.9
+
+### Patch Changes
+
+- Updated dependencies [ce94780]
+  - llamaindex@0.3.8
+
+## 0.0.8
+
+### Patch Changes
+
+- Updated dependencies [b6a6606]
+- Updated dependencies [b6a6606]
+  - llamaindex@0.3.7
+
+## 0.0.7
+
+### Patch Changes
+
+- Updated dependencies [efa326a]
+  - llamaindex@0.3.6
+
+## 0.0.6
+
+### Patch Changes
+
+- Updated dependencies [bc7a11c]
+- Updated dependencies [2fe2b81]
+- Updated dependencies [5596e31]
+- Updated dependencies [e74fe88]
+- Updated dependencies [be5df5b]
+  - llamaindex@0.3.5
+
 ## 0.0.5
 
 ### Patch Changes

packages/core/e2e/examples/waku-query-engine/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/waku-query-engine-test",
-  "version": "0.0.5",
+  "version": "0.0.13",
   "type": "module",
   "private": true,
   "scripts": {
@@ -16,10 +16,10 @@
     "waku": "0.20.1"
   },
   "devDependencies": {
-    "@types/react": "18.2.74",
-    "@types/react-dom": "18.2.24",
+    "@types/react": "18.3.1",
+    "@types/react-dom": "18.3.0",
     "autoprefixer": "10.4.19",
     "tailwindcss": "3.4.3",
-    "typescript": "5.4.4"
+    "typescript": "5.4.5"
   }
 }

packages/core/e2e/node/claude.e2e.ts

@@ -4,7 +4,7 @@ import { AnthropicAgent } from "llamaindex/agent/anthropic";
 import { extractText } from "llamaindex/llm/utils";
 import { ok, strictEqual } from "node:assert";
 import { beforeEach, test } from "node:test";
-import { sumNumbersTool } from "./fixtures/tools.js";
+import { getWeatherTool, sumNumbersTool } from "./fixtures/tools.js";
 import { mockLLMEvent } from "./utils.js";
 
 let llm: LLM;
@@ -118,14 +118,58 @@ await test("anthropic agent", async (t) => {
   });
 
   await t.test("sum numbers", async () => {
-    const openaiAgent = new AnthropicAgent({
+    const anthropicAgent = new AnthropicAgent({
       tools: [sumNumbersTool],
     });
 
-    const { response } = await openaiAgent.chat({
+    const { response } = await anthropicAgent.chat({
       message: "how much is 1 + 1?",
     });
 
     ok(extractText(response.message.content).includes("2"));
   });
 });
+
+await test("anthropic agent with multiple chat", async (t) => {
+  await mockLLMEvent(t, "anthropic-agent-multiple-chat");
+  await t.test("chat", async () => {
+    const agent = new AnthropicAgent({
+      tools: [getWeatherTool],
+    });
+    {
+      const { response } = await agent.chat({
+        message: 'Hello? Response to me "Yes"',
+      });
+      consola.debug("response:", response.message.content);
+      ok(extractText(response.message.content).includes("Yes"));
+    }
+    {
+      const { response } = await agent.chat({
+        message: 'Hello? Response to me "No"',
+      });
+      consola.debug("response:", response.message.content);
+      ok(extractText(response.message.content).includes("No"));
+    }
+    {
+      const { response } = await agent.chat({
+        message: 'Hello? Response to me "Maybe"',
+      });
+      consola.debug("response:", response.message.content);
+      ok(extractText(response.message.content).includes("Maybe"));
+    }
+    {
+      const { response } = await agent.chat({
+        message: "What is the weather in San Francisco?",
+      });
+      consola.debug("response:", response.message.content);
+      ok(extractText(response.message.content).includes("72"));
+    }
+    {
+      const { response } = await agent.chat({
+        message: "What is the weather in Shanghai?",
+      });
+      consola.debug("response:", response.message.content);
+      ok(extractText(response.message.content).includes("72"));
+    }
+  });
+});

packages/core/e2e/node/snapshot/anthropic-agent-multiple-chat.snap

@@ -0,0 +1,552 @@
+{
+  "llmEventStart": [
+    {
+      "id": "PRESERVE_0",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Yes\""
+        }
+      ]
+    },
+    {
+      "id": "PRESERVE_1",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Yes\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Yes\".\n</thinking>\n\nYes"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"No\""
+        }
+      ]
+    },
+    {
+      "id": "PRESERVE_2",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Yes\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Yes\".\n</thinking>\n\nYes"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"No\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"No\".\n</thinking>\n\nNo"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Maybe\""
+        }
+      ]
+    },
+    {
+      "id": "PRESERVE_3",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Yes\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Yes\".\n</thinking>\n\nYes"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"No\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"No\".\n</thinking>\n\nNo"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Maybe\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Maybe\".\n</thinking>\n\nMaybe"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "What is the weather in San Francisco?"
+        }
+      ]
+    },
+    {
+      "id": "PRESERVE_4",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Yes\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Yes\".\n</thinking>\n\nYes"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"No\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"No\".\n</thinking>\n\nNo"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Maybe\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Maybe\".\n</thinking>\n\nMaybe"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "What is the weather in San Francisco?"
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has asked for the weather in a specific city, San Francisco. The relevant tool to answer this is the getWeather function.\n\nLooking at the required parameters for getWeather:\ncity (string): The user directly provided the city \"San Francisco\"\n\nSince the required \"city\" parameter has been provided, we can proceed with the getWeather function call.\n</thinking>"
+            }
+          ],
+          "role": "assistant",
+          "options": {
+            "toolCall": {
+              "id": "toolu_01Gy7Gxbx7uGmjVncGH6pubL",
+              "name": "getWeather",
+              "input": {
+                "city": "San Francisco"
+              }
+            }
+          }
+        },
+        {
+          "content": "The weather in San Francisco is 72 degrees",
+          "role": "user",
+          "options": {
+            "toolResult": {
+              "result": "The weather in San Francisco is 72 degrees",
+              "isError": false,
+              "id": "toolu_01Gy7Gxbx7uGmjVncGH6pubL"
+            }
+          }
+        }
+      ]
+    },
+    {
+      "id": "PRESERVE_5",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Yes\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Yes\".\n</thinking>\n\nYes"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"No\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"No\".\n</thinking>\n\nNo"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Maybe\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Maybe\".\n</thinking>\n\nMaybe"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "What is the weather in San Francisco?"
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has asked for the weather in a specific city, San Francisco. The relevant tool to answer this is the getWeather function.\n\nLooking at the required parameters for getWeather:\ncity (string): The user directly provided the city \"San Francisco\"\n\nSince the required \"city\" parameter has been provided, we can proceed with the getWeather function call.\n</thinking>"
+            }
+          ],
+          "role": "assistant",
+          "options": {
+            "toolCall": {
+              "id": "toolu_01Gy7Gxbx7uGmjVncGH6pubL",
+              "name": "getWeather",
+              "input": {
+                "city": "San Francisco"
+              }
+            }
+          }
+        },
+        {
+          "content": "The weather in San Francisco is 72 degrees",
+          "role": "user",
+          "options": {
+            "toolResult": {
+              "result": "The weather in San Francisco is 72 degrees",
+              "isError": false,
+              "id": "toolu_01Gy7Gxbx7uGmjVncGH6pubL"
+            }
+          }
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "The current weather in San Francisco is 72 degrees."
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "What is the weather in Shanghai?"
+        }
+      ]
+    },
+    {
+      "id": "PRESERVE_6",
+      "messages": [
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Yes\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Yes\".\n</thinking>\n\nYes"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"No\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"No\".\n</thinking>\n\nNo"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "Hello? Response to me \"Maybe\""
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Maybe\".\n</thinking>\n\nMaybe"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "What is the weather in San Francisco?"
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has asked for the weather in a specific city, San Francisco. The relevant tool to answer this is the getWeather function.\n\nLooking at the required parameters for getWeather:\ncity (string): The user directly provided the city \"San Francisco\"\n\nSince the required \"city\" parameter has been provided, we can proceed with the getWeather function call.\n</thinking>"
+            }
+          ],
+          "role": "assistant",
+          "options": {
+            "toolCall": {
+              "id": "toolu_01Gy7Gxbx7uGmjVncGH6pubL",
+              "name": "getWeather",
+              "input": {
+                "city": "San Francisco"
+              }
+            }
+          }
+        },
+        {
+          "content": "The weather in San Francisco is 72 degrees",
+          "role": "user",
+          "options": {
+            "toolResult": {
+              "result": "The weather in San Francisco is 72 degrees",
+              "isError": false,
+              "id": "toolu_01Gy7Gxbx7uGmjVncGH6pubL"
+            }
+          }
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "The current weather in San Francisco is 72 degrees."
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        },
+        {
+          "role": "user",
+          "content": "What is the weather in Shanghai?"
+        },
+        {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has asked for the weather in a specific city, Shanghai. The relevant tool to answer this is the getWeather function.\n\nLooking at the required parameters for getWeather:\ncity (string): The user directly provided the city \"Shanghai\"\n\nSince the required \"city\" parameter has been provided, we can proceed with the getWeather function call.\n</thinking>"
+            }
+          ],
+          "role": "assistant",
+          "options": {
+            "toolCall": {
+              "id": "toolu_01NHyahSUqrPjxQk9mvCvvGe",
+              "name": "getWeather",
+              "input": {
+                "city": "Shanghai"
+              }
+            }
+          }
+        },
+        {
+          "content": "The weather in Shanghai is 72 degrees",
+          "role": "user",
+          "options": {
+            "toolResult": {
+              "result": "The weather in Shanghai is 72 degrees",
+              "isError": false,
+              "id": "toolu_01NHyahSUqrPjxQk9mvCvvGe"
+            }
+          }
+        }
+      ]
+    }
+  ],
+  "llmEventEnd": [
+    {
+      "id": "PRESERVE_0",
+      "response": {
+        "raw": null,
+        "message": {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Yes\".\n</thinking>\n\nYes"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        }
+      }
+    },
+    {
+      "id": "PRESERVE_1",
+      "response": {
+        "raw": null,
+        "message": {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"No\".\n</thinking>\n\nNo"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        }
+      }
+    },
+    {
+      "id": "PRESERVE_2",
+      "response": {
+        "raw": null,
+        "message": {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has not asked a question that requires using any of the available tools. They have simply requested that I respond with the word \"Maybe\".\n</thinking>\n\nMaybe"
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        }
+      }
+    },
+    {
+      "id": "PRESERVE_3",
+      "response": {
+        "raw": null,
+        "message": {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has asked for the weather in a specific city, San Francisco. The relevant tool to answer this is the getWeather function.\n\nLooking at the required parameters for getWeather:\ncity (string): The user directly provided the city \"San Francisco\"\n\nSince the required \"city\" parameter has been provided, we can proceed with the getWeather function call.\n</thinking>"
+            }
+          ],
+          "role": "assistant",
+          "options": {
+            "toolCall": {
+              "id": "toolu_01Gy7Gxbx7uGmjVncGH6pubL",
+              "name": "getWeather",
+              "input": {
+                "city": "San Francisco"
+              }
+            }
+          }
+        }
+      }
+    },
+    {
+      "id": "PRESERVE_4",
+      "response": {
+        "raw": null,
+        "message": {
+          "content": [
+            {
+              "type": "text",
+              "text": "The current weather in San Francisco is 72 degrees."
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        }
+      }
+    },
+    {
+      "id": "PRESERVE_5",
+      "response": {
+        "raw": null,
+        "message": {
+          "content": [
+            {
+              "type": "text",
+              "text": "<thinking>\nThe user has asked for the weather in a specific city, Shanghai. The relevant tool to answer this is the getWeather function.\n\nLooking at the required parameters for getWeather:\ncity (string): The user directly provided the city \"Shanghai\"\n\nSince the required \"city\" parameter has been provided, we can proceed with the getWeather function call.\n</thinking>"
+            }
+          ],
+          "role": "assistant",
+          "options": {
+            "toolCall": {
+              "id": "toolu_01NHyahSUqrPjxQk9mvCvvGe",
+              "name": "getWeather",
+              "input": {
+                "city": "Shanghai"
+              }
+            }
+          }
+        }
+      }
+    },
+    {
+      "id": "PRESERVE_6",
+      "response": {
+        "raw": null,
+        "message": {
+          "content": [
+            {
+              "type": "text",
+              "text": "The current weather in Shanghai is 72 degrees."
+            }
+          ],
+          "role": "assistant",
+          "options": {}
+        }
+      }
+    }
+  ],
+  "llmEventStream": []
+}

packages/core/e2e/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/core-e2e",
   "private": true,
-  "version": "0.0.3",
+  "version": "0.0.6",
   "type": "module",
   "scripts": {
     "e2e": "node --import tsx --import ./mock-register.js --test ./node/*.e2e.ts",
@@ -10,9 +10,9 @@
   },
   "devDependencies": {
     "@faker-js/faker": "^8.4.1",
-    "@types/node": "^20.12.7",
+    "@types/node": "^20.12.11",
     "consola": "^3.2.3",
     "llamaindex": "workspace:*",
-    "tsx": "^4.7.2"
+    "tsx": "^4.9.3"
   }
 }

packages/core/jsr.json

@@ -1,8 +1,11 @@
 {
   "name": "@llamaindex/core",
-  "version": "0.3.4",
+  "version": "0.3.12",
   "exports": "./src/index.ts",
   "imports": {
-    "@llamaindex/env": "jsr:@llamaindex/env@0.0.6"
+    "@llamaindex/env": "jsr:@llamaindex/env@0.1.3"
+  },
+  "publish": {
+    "include": ["LICENSE", "README.md", "src/**/*", "jsr.json"]
   }
 }

packages/core/package.json

@@ -1,49 +1,65 @@
 {
   "name": "llamaindex",
-  "version": "0.3.4",
+  "version": "0.3.12",
   "expectedMinorVersion": "3",
   "license": "MIT",
   "type": "module",
+  "keywords": [
+    "llm",
+    "llama",
+    "openai",
+    "gpt",
+    "data science",
+    "prompt",
+    "prompt engineering",
+    "chatgpt",
+    "machine learning",
+    "ml",
+    "embedding",
+    "vectorstore",
+    "data framework",
+    "llamaindex"
+  ],
   "dependencies": {
-    "@anthropic-ai/sdk": "^0.20.6",
+    "@anthropic-ai/sdk": "^0.20.9",
     "@aws-crypto/sha256-js": "^5.2.0",
-    "@datastax/astra-db-ts": "^1.0.1",
-    "@google/generative-ai": "^0.8.0",
-    "@grpc/grpc-js": "^1.10.6",
+    "@datastax/astra-db-ts": "^1.1.0",
+    "@google/generative-ai": "^0.11.0",
+    "@grpc/grpc-js": "^1.10.7",
     "@huggingface/inference": "^2.6.7",
     "@llamaindex/cloud": "0.0.5",
     "@llamaindex/env": "workspace:*",
-    "@mistralai/mistralai": "^0.1.3",
+    "@mistralai/mistralai": "^0.2.0",
     "@pinecone-database/pinecone": "^2.2.0",
-    "@qdrant/js-client-rest": "^1.8.2",
-    "@types/lodash": "^4.17.0",
+    "@qdrant/js-client-rest": "^1.9.0",
+    "@types/lodash": "^4.17.1",
     "@types/papaparse": "^5.3.14",
-    "@types/pg": "^8.11.5",
+    "@types/pg": "^8.11.6",
     "@xenova/transformers": "^2.17.1",
-    "@zilliz/milvus2-sdk-node": "^2.4.1",
-    "ajv": "^8.12.0",
-    "assemblyai": "^4.4.1",
+    "@zilliz/milvus2-sdk-node": "^2.4.2",
+    "ajv": "^8.13.0",
+    "assemblyai": "^4.4.2",
     "chromadb": "~1.7.3",
     "cohere-ai": "^7.9.5",
     "js-tiktoken": "^1.0.11",
     "lodash": "^4.17.21",
     "magic-bytes.js": "^1.10.0",
-    "mammoth": "^1.7.1",
+    "mammoth": "^1.7.2",
     "md-utils-ts": "^2.0.0",
-    "mongodb": "^6.5.0",
+    "mongodb": "^6.6.1",
     "notion-md-crawler": "^1.0.0",
-    "ollama": "^0.5.0",
-    "openai": "^4.38.0",
+    "openai": "^4.46.0",
     "papaparse": "^5.4.1",
     "pathe": "^1.1.2",
-    "pdf2json": "^3.0.5",
+    "pdf2json": "3.0.5",
     "pg": "^8.11.5",
     "pgvector": "^0.1.8",
     "portkey-ai": "^0.1.16",
     "rake-modified": "^1.0.8",
+    "std-env": "^3.7.0",
     "string-strip-html": "^13.4.8",
     "wikipedia": "^2.1.2",
-    "wink-nlp": "^1.14.3"
+    "wink-nlp": "^2.2.2"
   },
   "peerDependencies": {
     "@notionhq/client": "^2.2.15"
@@ -51,10 +67,9 @@
   "devDependencies": {
     "@notionhq/client": "^2.2.15",
     "@swc/cli": "^0.3.12",
-    "@swc/core": "^1.4.16",
+    "@swc/core": "^1.5.5",
     "concurrently": "^8.2.2",
     "glob": "^10.3.12",
-    "madge": "^7.0.0",
     "typescript": "^5.4.5"
   },
   "engines": {

packages/core/src/Node.ts

@@ -1,5 +1,5 @@
 import { createSHA256, path, randomUUID } from "@llamaindex/env";
-import _ from "lodash";
+import { chunkSizeCheck, lazyInitHash } from "./internal/decorator/node.js";
 
 export enum NodeRelationship {
   SOURCE = "SOURCE",
@@ -37,6 +37,16 @@ export type RelatedNodeType<T extends Metadata = Metadata> =
   | RelatedNodeInfo<T>
   | RelatedNodeInfo<T>[];
 
+export type BaseNodeParams<T extends Metadata = Metadata> = {
+  id_?: string;
+  metadata?: T;
+  excludedEmbedMetadataKeys?: string[];
+  excludedLlmMetadataKeys?: string[];
+  relationships?: Partial<Record<NodeRelationship, RelatedNodeType<T>>>;
+  hash?: string;
+  embedding?: number[];
+};
+
 /**
  * Generic abstract class for retrievable nodes
  */
@@ -47,21 +57,37 @@ export abstract class BaseNode<T extends Metadata = Metadata> {
    *
    * Set to a UUID by default.
    */
-  id_: string = randomUUID();
+  id_: string;
   embedding?: number[];
 
   // Metadata fields
-  metadata: T = {} as T;
-  excludedEmbedMetadataKeys: string[] = [];
-  excludedLlmMetadataKeys: string[] = [];
-  relationships: Partial<Record<NodeRelationship, RelatedNodeType<T>>> = {};
-  hash: string = "";
+  metadata: T;
+  excludedEmbedMetadataKeys: string[];
+  excludedLlmMetadataKeys: string[];
+  relationships: Partial<Record<NodeRelationship, RelatedNodeType<T>>>;
 
-  constructor(init?: Partial<BaseNode<T>>) {
-    Object.assign(this, init);
+  @lazyInitHash
+  accessor hash: string = "";
+
+  protected constructor(init?: BaseNodeParams<T>) {
+    const {
+      id_,
+      metadata,
+      excludedEmbedMetadataKeys,
+      excludedLlmMetadataKeys,
+      relationships,
+      hash,
+      embedding,
+    } = init || {};
+    this.id_ = id_ ?? randomUUID();
+    this.metadata = metadata ?? ({} as T);
+    this.excludedEmbedMetadataKeys = excludedEmbedMetadataKeys ?? [];
+    this.excludedLlmMetadataKeys = excludedLlmMetadataKeys ?? [];
+    this.relationships = relationships ?? {};
+    this.embedding = embedding;
   }
 
-  abstract getType(): ObjectType;
+  abstract get type(): ObjectType;
 
   abstract getContent(metadataMode: MetadataMode): string;
   abstract getMetadataStr(metadataMode: MetadataMode): string;
@@ -146,7 +172,12 @@ export abstract class BaseNode<T extends Metadata = Metadata> {
    * @see toMutableJSON - use to return a mutable JSON instead
    */
   toJSON(): Record<string, any> {
-    return { ...this, type: this.getType() };
+    return {
+      ...this,
+      type: this.type,
+      // hash is an accessor property, so it's not included in the rest operator
+      hash: this.hash,
+    };
   }
 
   clone(): BaseNode {
@@ -159,32 +190,43 @@ export abstract class BaseNode<T extends Metadata = Metadata> {
    * @return {Record<string, any>} - The JSON representation of the object.
    */
   toMutableJSON(): Record<string, any> {
-    return _.cloneDeep(this.toJSON());
+    return structuredClone(this.toJSON());
   }
 }
 
+export type TextNodeParams<T extends Metadata = Metadata> =
+  BaseNodeParams<T> & {
+    text?: string;
+    textTemplate?: string;
+    startCharIdx?: number;
+    endCharIdx?: number;
+    metadataSeparator?: string;
+  };
+
 /**
  * TextNode is the default node type for text. Most common node type in LlamaIndex.TS
  */
 export class TextNode<T extends Metadata = Metadata> extends BaseNode<T> {
-  text: string = "";
-  textTemplate: string = "";
+  text: string;
+  textTemplate: string;
 
   startCharIdx?: number;
   endCharIdx?: number;
   // textTemplate: NOTE write your own formatter if needed
   // metadataTemplate: NOTE write your own formatter if needed
-  metadataSeparator: string = "\n";
+  metadataSeparator: string;
 
-  constructor(init?: Partial<TextNode<T>>) {
+  constructor(init: TextNodeParams<T> = {}) {
     super(init);
-    Object.assign(this, init);
-
-    if (new.target === TextNode) {
-      // Don't generate the hash repeatedly so only do it if this is
-      // constructing the derived class
-      this.hash = init?.hash ?? this.generateHash();
+    const { text, textTemplate, startCharIdx, endCharIdx, metadataSeparator } =
+      init;
+    this.text = text ?? "";
+    this.textTemplate = textTemplate ?? "";
+    if (startCharIdx) {
+      this.startCharIdx = startCharIdx;
     }
+    this.endCharIdx = endCharIdx;
+    this.metadataSeparator = metadataSeparator ?? "\n";
   }
 
   /**
@@ -194,7 +236,7 @@ export class TextNode<T extends Metadata = Metadata> extends BaseNode<T> {
    */
   generateHash() {
     const hashFunction = createSHA256();
-    hashFunction.update(`type=${this.getType()}`);
+    hashFunction.update(`type=${this.type}`);
     hashFunction.update(
       `startCharIdx=${this.startCharIdx} endCharIdx=${this.endCharIdx}`,
     );
@@ -202,10 +244,11 @@ export class TextNode<T extends Metadata = Metadata> extends BaseNode<T> {
     return hashFunction.digest();
   }
 
-  getType(): ObjectType {
+  get type() {
     return ObjectType.TEXT;
   }
 
+  @chunkSizeCheck
   getContent(metadataMode: MetadataMode = MetadataMode.NONE): string {
     const metadataStr = this.getMetadataStr(metadataMode).trim();
     return `${metadataStr}\n\n${this.text}`.trim();
@@ -246,19 +289,21 @@ export class TextNode<T extends Metadata = Metadata> extends BaseNode<T> {
   }
 }
 
+export type IndexNodeParams<T extends Metadata = Metadata> =
+  TextNodeParams<T> & {
+    indexId: string;
+  };
+
 export class IndexNode<T extends Metadata = Metadata> extends TextNode<T> {
-  indexId: string = "";
+  indexId: string;
 
-  constructor(init?: Partial<IndexNode<T>>) {
+  constructor(init?: IndexNodeParams<T>) {
     super(init);
-    Object.assign(this, init);
-
-    if (new.target === IndexNode) {
-      this.hash = init?.hash ?? this.generateHash();
-    }
+    const { indexId } = init || {};
+    this.indexId = indexId ?? "";
   }
 
-  getType(): ObjectType {
+  get type() {
     return ObjectType.INDEX;
   }
 }
@@ -267,16 +312,11 @@ export class IndexNode<T extends Metadata = Metadata> extends TextNode<T> {
  * A document is just a special text node with a docId.
  */
 export class Document<T extends Metadata = Metadata> extends TextNode<T> {
-  constructor(init?: Partial<Document<T>>) {
+  constructor(init?: TextNodeParams<T>) {
     super(init);
-    Object.assign(this, init);
-
-    if (new.target === Document) {
-      this.hash = init?.hash ?? this.generateHash();
-    }
   }
 
-  getType() {
+  get type() {
     return ObjectType.DOCUMENT;
   }
 }
@@ -303,21 +343,21 @@ export function jsonToNode(json: any, type?: ObjectType) {
 
 export type ImageType = string | Blob | URL;
 
-export type ImageNodeConstructorProps<T extends Metadata> = Pick<
-  ImageNode<T>,
-  "image" | "id_"
-> &
-  Partial<ImageNode<T>>;
+export type ImageNodeParams<T extends Metadata = Metadata> =
+  TextNodeParams<T> & {
+    image: ImageType;
+  };
 
 export class ImageNode<T extends Metadata = Metadata> extends TextNode<T> {
   image: ImageType; // image as blob
 
-  constructor(init: ImageNodeConstructorProps<T>) {
+  constructor(init: ImageNodeParams<T>) {
     super(init);
-    this.image = init.image;
+    const { image } = init;
+    this.image = image;
   }
 
-  getType(): ObjectType {
+  get type() {
     return ObjectType.IMAGE;
   }
 
@@ -360,15 +400,11 @@ export class ImageNode<T extends Metadata = Metadata> extends TextNode<T> {
 }
 
 export class ImageDocument<T extends Metadata = Metadata> extends ImageNode<T> {
-  constructor(init: ImageNodeConstructorProps<T>) {
+  constructor(init: ImageNodeParams<T>) {
     super(init);
-
-    if (new.target === ImageDocument) {
-      this.hash = init?.hash ?? this.generateHash();
-    }
   }
 
-  getType() {
+  get type() {
     return ObjectType.IMAGE_DOCUMENT;
   }
 }

packages/core/src/Prompt.ts

@@ -371,6 +371,7 @@ export function messagesToHistoryStr(messages: ChatMessage[]) {
 }
 
 export const defaultContextSystemPrompt = ({ context = "" }) => {
+  if (!context) return "";
   return `Context information is below.
 ---------------------
 ${context}

packages/core/src/Settings.ts

@@ -13,6 +13,11 @@ import {
   setCallbackManager,
   withCallbackManager,
 } from "./internal/settings/CallbackManager.js";
+import {
+  getChunkSize,
+  setChunkSize,
+  withChunkSize,
+} from "./internal/settings/chunk-size.js";
 import type { LLM } from "./llm/types.js";
 import type { NodeParser } from "./nodeParsers/types.js";
 
@@ -41,14 +46,12 @@ class GlobalSettings implements Config {
   #promptHelper: PromptHelper | null = null;
   #embedModel: BaseEmbedding | null = null;
   #nodeParser: NodeParser | null = null;
-  #chunkSize?: number;
   #chunkOverlap?: number;
 
   #llmAsyncLocalStorage = new AsyncLocalStorage<LLM>();
   #promptHelperAsyncLocalStorage = new AsyncLocalStorage<PromptHelper>();
   #embedModelAsyncLocalStorage = new AsyncLocalStorage<BaseEmbedding>();
   #nodeParserAsyncLocalStorage = new AsyncLocalStorage<NodeParser>();
-  #chunkSizeAsyncLocalStorage = new AsyncLocalStorage<number>();
   #chunkOverlapAsyncLocalStorage = new AsyncLocalStorage<number>();
   #promptAsyncLocalStorage = new AsyncLocalStorage<PromptConfig>();
 
@@ -115,8 +118,8 @@ class GlobalSettings implements Config {
   get nodeParser(): NodeParser {
     if (this.#nodeParser === null) {
       this.#nodeParser = new SimpleNodeParser({
-        chunkSize: this.#chunkSize,
-        chunkOverlap: this.#chunkOverlap,
+        chunkSize: this.chunkSize,
+        chunkOverlap: this.chunkOverlap,
       });
     }
 
@@ -147,15 +150,15 @@ class GlobalSettings implements Config {
   }
 
   set chunkSize(chunkSize: number | undefined) {
-    this.#chunkSize = chunkSize;
+    setChunkSize(chunkSize);
   }
 
   get chunkSize(): number | undefined {
-    return this.#chunkSizeAsyncLocalStorage.getStore() ?? this.#chunkSize;
+    return getChunkSize();
   }
 
   withChunkSize<Result>(chunkSize: number, fn: () => Result): Result {
-    return this.#chunkSizeAsyncLocalStorage.run(chunkSize, fn);
+    return withChunkSize(chunkSize, fn);
   }
 
   get chunkOverlap(): number | undefined {

packages/core/src/agent/base.ts

@@ -1,9 +1,4 @@
-import {
-  ReadableStream,
-  TransformStream,
-  pipeline,
-  randomUUID,
-} from "@llamaindex/env";
+import { ReadableStream, TransformStream, randomUUID } from "@llamaindex/env";
 import { Settings } from "../Settings.js";
 import {
   type ChatEngine,
@@ -21,7 +16,6 @@ import type {
   LLM,
   MessageContent,
 } from "../llm/index.js";
-import { extractText } from "../llm/utils.js";
 import type { BaseToolWithCall, ToolOutput } from "../types.js";
 import type {
   AgentTaskContext,
@@ -169,7 +163,7 @@ export abstract class AgentWorker<
   abstract taskHandler: TaskHandler<AI, Store, AdditionalMessageOptions>;
 
   public createTask(
-    query: string,
+    query: MessageContent,
     context: AgentTaskContext<AI, Store, AdditionalMessageOptions>,
   ): ReadableStream<TaskStepOutput<AI, Store, AdditionalMessageOptions>> {
     context.store.messages.push({
@@ -305,7 +299,7 @@ export abstract class AgentRunner<
         });
       }
     }
-    return this.#runner.createTask(extractText(message), {
+    return this.#runner.createTask(message, {
       stream,
       toolCallCount: 0,
       llm: this.#llm,
@@ -340,45 +334,36 @@ export abstract class AgentRunner<
     | ReadableStream<AgentStreamChatResponse<AdditionalMessageOptions>>
   > {
     const task = this.createTask(params.message, !!params.stream);
-    const stepOutput = await pipeline(
-      task,
-      async (
-        iter: AsyncIterable<
-          TaskStepOutput<AI, Store, AdditionalMessageOptions>
-        >,
-      ) => {
-        for await (const stepOutput of iter) {
-          if (stepOutput.isLast) {
-            return stepOutput;
-          }
-        }
-        throw new Error("Task did not complete");
-      },
-    );
-    const { output, taskStep } = stepOutput;
-    this.#chatHistory = [...taskStep.context.store.messages];
-    if (isAsyncIterable(output)) {
-      return output.pipeThrough<
-        AgentStreamChatResponse<AdditionalMessageOptions>
-      >(
-        new TransformStream({
-          transform(chunk, controller) {
-            controller.enqueue({
-              response: chunk,
-              get sources() {
-                return [...taskStep.context.store.toolOutputs];
+    for await (const stepOutput of task) {
+      // update chat history for each round
+      this.#chatHistory = [...stepOutput.taskStep.context.store.messages];
+      if (stepOutput.isLast) {
+        const { output, taskStep } = stepOutput;
+        if (isAsyncIterable(output)) {
+          return output.pipeThrough<
+            AgentStreamChatResponse<AdditionalMessageOptions>
+          >(
+            new TransformStream({
+              transform(chunk, controller) {
+                controller.enqueue({
+                  response: chunk,
+                  get sources() {
+                    return [...taskStep.context.store.toolOutputs];
+                  },
+                });
               },
-            });
-          },
-        }),
-      );
-    } else {
-      return {
-        response: output,
-        get sources() {
-          return [...taskStep.context.store.toolOutputs];
-        },
-      } satisfies AgentChatResponse<AdditionalMessageOptions>;
+            }),
+          );
+        } else {
+          return {
+            response: output,
+            get sources() {
+              return [...taskStep.context.store.toolOutputs];
+            },
+          } satisfies AgentChatResponse<AdditionalMessageOptions>;
+        }
+      }
     }
+    throw new Error("Task ended without a last step.");
   }
 }

packages/core/src/agent/openai.ts

@@ -1,4 +1,5 @@
-import { pipeline, ReadableStream } from "@llamaindex/env";
+import { ReadableStream } from "@llamaindex/env";
+import { Settings } from "../Settings.js";
 import { stringifyJSONToMessageContent } from "../internal/utils.js";
 import type {
   ChatResponseChunk,
@@ -8,7 +9,6 @@ import type {
 } from "../llm/index.js";
 import { OpenAI } from "../llm/openai.js";
 import { ObjectRetriever } from "../objects/index.js";
-import { Settings } from "../Settings.js";
 import type { BaseToolWithCall } from "../types.js";
 import { AgentRunner, AgentWorker, type AgentParamsBase } from "./base.js";
 import type { TaskHandler } from "./types.js";
@@ -130,22 +130,14 @@ export class OpenAIAgent extends AgentRunner<OpenAI> {
 
       if (hasToolCall) {
         // you need to consume the response to get the full toolCalls
-        const toolCalls = await pipeline(
-          pipStream,
-          async (
-            iter: AsyncIterable<ChatResponseChunk<ToolCallLLMMessageOptions>>,
-          ) => {
-            const toolCalls = new Map<string, ToolCall | PartialToolCall>();
-            for await (const chunk of iter) {
-              if (chunk.options && "toolCall" in chunk.options) {
-                const toolCall = chunk.options.toolCall;
-                toolCalls.set(toolCall.id, toolCall);
-              }
-            }
-            return [...toolCalls.values()];
-          },
-        );
-        for (const toolCall of toolCalls) {
+        const toolCalls = new Map<string, ToolCall | PartialToolCall>();
+        for await (const chunk of pipStream) {
+          if (chunk.options && "toolCall" in chunk.options) {
+            const toolCall = chunk.options.toolCall;
+            toolCalls.set(toolCall.id, toolCall);
+          }
+        }
+        for (const toolCall of toolCalls.values()) {
           const targetTool = tools.find(
             (tool) => tool.metadata.name === toolCall.name,
           );

packages/core/src/callbacks/CallbackManager.ts

@@ -12,6 +12,8 @@ import type {
   LLMStreamEvent,
   LLMToolCallEvent,
   LLMToolResultEvent,
+  RetrievalEndEvent,
+  RetrievalStartEvent,
 } from "../llm/types.js";
 
 export class LlamaIndexCustomEvent<T = any> extends CustomEvent<T> {
@@ -45,6 +47,8 @@ export interface LlamaIndexEventMaps {
    * @deprecated
    */
   retrieve: CustomEvent<RetrievalCallbackResponse>;
+  "retrieve-start": RetrievalStartEvent;
+  "retrieve-end": RetrievalEndEvent;
   /**
    * @deprecated
    */

packages/core/src/embeddings/ClipEmbedding.ts

@@ -1,12 +1,17 @@
 import _ from "lodash";
 import type { ImageType } from "../Node.js";
+import { lazyLoadTransformers } from "../internal/deps/transformers.js";
 import { MultiModalEmbedding } from "./MultiModalEmbedding.js";
+// only import type, to avoid bundling error
+import type {
+  CLIPTextModelWithProjection,
+  CLIPVisionModelWithProjection,
+  PreTrainedTokenizer,
+  Processor,
+} from "@xenova/transformers";
 
 async function readImage(input: ImageType) {
-  const { RawImage } = await import(
-    /* webpackIgnore: true */
-    "@xenova/transformers"
-  );
+  const { RawImage } = await lazyLoadTransformers();
   if (input instanceof Blob) {
     return await RawImage.fromBlob(input);
   } else if (_.isString(input) || input instanceof URL) {
@@ -25,39 +30,30 @@ export class ClipEmbedding extends MultiModalEmbedding {
   modelType: ClipEmbeddingModelType =
     ClipEmbeddingModelType.XENOVA_CLIP_VIT_BASE_PATCH16;
 
-  private tokenizer: any;
-  private processor: any;
-  private visionModel: any;
-  private textModel: any;
+  private tokenizer: PreTrainedTokenizer | null = null;
+  private processor: Processor | null = null;
+  private visionModel: CLIPVisionModelWithProjection | null = null;
+  private textModel: CLIPTextModelWithProjection | null = null;
 
   async getTokenizer() {
+    const { AutoTokenizer } = await lazyLoadTransformers();
     if (!this.tokenizer) {
-      const { AutoTokenizer } = await import(
-        /* webpackIgnore: true */
-        "@xenova/transformers"
-      );
       this.tokenizer = await AutoTokenizer.from_pretrained(this.modelType);
     }
     return this.tokenizer;
   }
 
   async getProcessor() {
+    const { AutoProcessor } = await lazyLoadTransformers();
     if (!this.processor) {
-      const { AutoProcessor } = await import(
-        /* webpackIgnore: true */
-        "@xenova/transformers"
-      );
       this.processor = await AutoProcessor.from_pretrained(this.modelType);
     }
     return this.processor;
   }
 
   async getVisionModel() {
+    const { CLIPVisionModelWithProjection } = await lazyLoadTransformers();
     if (!this.visionModel) {
-      const { CLIPVisionModelWithProjection } = await import(
-        /* webpackIgnore: true */
-        "@xenova/transformers"
-      );
       this.visionModel = await CLIPVisionModelWithProjection.from_pretrained(
         this.modelType,
       );
@@ -67,11 +63,8 @@ export class ClipEmbedding extends MultiModalEmbedding {
   }
 
   async getTextModel() {
+    const { CLIPTextModelWithProjection } = await lazyLoadTransformers();
     if (!this.textModel) {
-      const { CLIPTextModelWithProjection } = await import(
-        /* webpackIgnore: true */
-        "@xenova/transformers"
-      );
       this.textModel = await CLIPTextModelWithProjection.from_pretrained(
         this.modelType,
       );

packages/core/src/embeddings/HuggingFaceEmbedding.ts

@@ -1,3 +1,4 @@
+import { lazyLoadTransformers } from "../internal/deps/transformers.js";
 import { BaseEmbedding } from "./types.js";
 
 export enum HuggingFaceEmbeddingModelType {
@@ -31,7 +32,7 @@ export class HuggingFaceEmbedding extends BaseEmbedding {
 
   async getExtractor() {
     if (!this.extractor) {
-      const { pipeline } = await import("@xenova/transformers");
+      const { pipeline } = await lazyLoadTransformers();
       this.extractor = await pipeline("feature-extraction", this.modelType, {
         quantized: this.quantized,
       });

packages/core/src/embeddings/index.ts

@@ -2,6 +2,7 @@ export * from "./GeminiEmbedding.js";
 export * from "./JinaAIEmbedding.js";
 export * from "./MistralAIEmbedding.js";
 export * from "./MultiModalEmbedding.js";
+export { OllamaEmbedding } from "./OllamaEmbedding.js";
 export * from "./OpenAIEmbedding.js";
 export { FireworksEmbedding } from "./fireworks.js";
 export { TogetherEmbedding } from "./together.js";

packages/core/src/embeddings/utils.ts

@@ -1,4 +1,4 @@
-import { defaultFS } from "@llamaindex/env";
+import { fs } from "@llamaindex/env";
 import _ from "lodash";
 import { filetypemime } from "magic-bytes.js";
 import type { ImageType } from "../Node.js";
@@ -243,7 +243,7 @@ export async function imageToDataUrl(input: ImageType): Promise<string> {
     _.isString(input)
   ) {
     // string or file URL
-    const dataBuffer = await defaultFS.readFile(
+    const dataBuffer = await fs.readFile(
       input instanceof URL ? input.pathname : input,
     );
     input = new Blob([dataBuffer]);

packages/core/src/index.ts

@@ -10,5 +10,3 @@ export {
   HuggingFaceEmbedding,
   HuggingFaceEmbeddingModelType,
 } from "./embeddings/HuggingFaceEmbedding.js";
-export { OllamaEmbedding } from "./embeddings/OllamaEmbedding.js";
-export { Ollama, type OllamaParams } from "./llm/ollama.js";

packages/core/src/indices/vectorStore/index.ts

@@ -1,14 +1,8 @@
-import type {
-  BaseNode,
-  Document,
-  Metadata,
-  NodeWithScore,
-} from "../../Node.js";
+import type { BaseNode, Document, NodeWithScore } from "../../Node.js";
 import { ImageNode, ObjectType, splitNodesByType } from "../../Node.js";
 import type { BaseRetriever, RetrieveParams } from "../../Retriever.js";
 import type { ServiceContext } from "../../ServiceContext.js";
 import {
-  Settings,
   embedModelFromSettingsOrContext,
   nodeParserFromSettingsOrContext,
 } from "../../Settings.js";
@@ -25,6 +19,7 @@ import {
   createDocStoreStrategy,
 } from "../../ingestion/strategies/index.js";
 import { wrapEventCaller } from "../../internal/context/EventCaller.js";
+import { getCallbackManager } from "../../internal/settings/CallbackManager.js";
 import type { BaseNodePostprocessor } from "../../postprocessors/types.js";
 import type { StorageContext } from "../../storage/StorageContext.js";
 import { storageContextFromDefaults } from "../../storage/StorageContext.js";
@@ -316,7 +311,7 @@ export class VectorStoreIndex extends BaseIndex<IndexDict> {
     // NOTE: if the vector store keeps text,
     // we only need to add image and index nodes
     for (let i = 0; i < nodes.length; ++i) {
-      const type = nodes[i].getType();
+      const { type } = nodes[i];
       if (
         !vectorStore.storesText ||
         type === ObjectType.INDEX ||
@@ -411,10 +406,16 @@ export class VectorIndexRetriever implements BaseRetriever {
     this.imageSimilarityTopK = imageSimilarityTopK ?? DEFAULT_SIMILARITY_TOP_K;
   }
 
+  @wrapEventCaller
   async retrieve({
     query,
     preFilters,
   }: RetrieveParams): Promise<NodeWithScore[]> {
+    getCallbackManager().dispatchEvent("retrieve-start", {
+      payload: {
+        query,
+      },
+    });
     let nodesWithScores = await this.textRetrieve(
       query,
       preFilters as MetadataFilters,
@@ -422,7 +423,17 @@ export class VectorIndexRetriever implements BaseRetriever {
     nodesWithScores = nodesWithScores.concat(
       await this.textToImageRetrieve(query, preFilters as MetadataFilters),
     );
-    this.sendEvent(query, nodesWithScores);
+    getCallbackManager().dispatchEvent("retrieve-end", {
+      payload: {
+        query,
+        nodes: nodesWithScores,
+      },
+    });
+    // send deprecated event
+    getCallbackManager().dispatchEvent("retrieve", {
+      query,
+      nodes: nodesWithScores,
+    });
     return nodesWithScores;
   }
 
@@ -459,17 +470,6 @@ export class VectorIndexRetriever implements BaseRetriever {
     return this.buildNodeListFromQueryResult(result);
   }
 
-  @wrapEventCaller
-  protected sendEvent(
-    query: string,
-    nodesWithScores: NodeWithScore<Metadata>[],
-  ) {
-    Settings.callbackManager.dispatchEvent("retrieve", {
-      query,
-      nodes: nodesWithScores,
-    });
-  }
-
   protected async buildVectorStoreQuery(
     embedModel: BaseEmbedding,
     query: string,

packages/core/src/internal/decorator/node.ts

@@ -0,0 +1,60 @@
+import { getEnv } from "@llamaindex/env";
+import type { BaseNode } from "../../Node.js";
+import { getChunkSize } from "../settings/chunk-size.js";
+
+const emitOnce = false;
+
+export function chunkSizeCheck(
+  contentGetter: () => string,
+  _context: ClassMethodDecoratorContext | ClassGetterDecoratorContext,
+) {
+  return function <Node extends BaseNode>(this: Node) {
+    const content = contentGetter.call(this);
+    const chunkSize = getChunkSize();
+    const enableChunkSizeCheck = getEnv("ENABLE_CHUNK_SIZE_CHECK") === "true";
+    if (
+      enableChunkSizeCheck &&
+      chunkSize !== undefined &&
+      content.length > chunkSize
+    ) {
+      console.warn(
+        `Node (${this.id_}) is larger than chunk size: ${content.length}`,
+      );
+      if (!emitOnce) {
+        console.warn(
+          "Will truncate the content if it is larger than chunk size",
+        );
+        console.warn("If you want to disable this behavior:");
+        console.warn("  1. Set Settings.chunkSize = undefined");
+        console.warn("  2. Set Settings.chunkSize to a larger value");
+        console.warn(
+          "  3. Change the way of splitting content into smaller chunks",
+        );
+      }
+      return content.slice(0, chunkSize);
+    }
+    return content;
+  };
+}
+
+export function lazyInitHash(
+  value: ClassAccessorDecoratorTarget<BaseNode, string>,
+  _context: ClassAccessorDecoratorContext,
+): ClassAccessorDecoratorResult<BaseNode, string> {
+  return {
+    get() {
+      const oldValue = value.get.call(this);
+      if (oldValue === "") {
+        const hash = this.generateHash();
+        value.set.call(this, hash);
+      }
+      return value.get.call(this);
+    },
+    set(newValue: string) {
+      value.set.call(this, newValue);
+    },
+    init(value: string): string {
+      return value;
+    },
+  };
+}

packages/core/src/internal/deps/ollama.d.ts

@@ -0,0 +1,264 @@
+type Fetch = typeof fetch;
+interface Config {
+  host: string;
+  fetch?: Fetch;
+  proxy?: boolean;
+}
+interface Options {
+  numa: boolean;
+  num_ctx: number;
+  num_batch: number;
+  main_gpu: number;
+  low_vram: boolean;
+  f16_kv: boolean;
+  logits_all: boolean;
+  vocab_only: boolean;
+  use_mmap: boolean;
+  use_mlock: boolean;
+  embedding_only: boolean;
+  num_thread: number;
+  num_keep: number;
+  seed: number;
+  num_predict: number;
+  top_k: number;
+  top_p: number;
+  tfs_z: number;
+  typical_p: number;
+  repeat_last_n: number;
+  temperature: number;
+  repeat_penalty: number;
+  presence_penalty: number;
+  frequency_penalty: number;
+  mirostat: number;
+  mirostat_tau: number;
+  mirostat_eta: number;
+  penalize_newline: boolean;
+  stop: string[];
+}
+interface GenerateRequest {
+  model: string;
+  prompt: string;
+  system?: string;
+  template?: string;
+  context?: number[];
+  stream?: boolean;
+  raw?: boolean;
+  format?: string;
+  images?: Uint8Array[] | string[];
+  keep_alive?: string | number;
+  options?: Partial<Options>;
+}
+interface Message {
+  role: string;
+  content: string;
+  images?: Uint8Array[] | string[];
+}
+interface ChatRequest {
+  model: string;
+  messages?: Message[];
+  stream?: boolean;
+  format?: string;
+  keep_alive?: string | number;
+  options?: Partial<Options>;
+}
+interface PullRequest {
+  model: string;
+  insecure?: boolean;
+  stream?: boolean;
+}
+interface PushRequest {
+  model: string;
+  insecure?: boolean;
+  stream?: boolean;
+}
+interface CreateRequest {
+  model: string;
+  path?: string;
+  modelfile?: string;
+  stream?: boolean;
+}
+interface DeleteRequest {
+  model: string;
+}
+interface CopyRequest {
+  source: string;
+  destination: string;
+}
+interface ShowRequest {
+  model: string;
+  system?: string;
+  template?: string;
+  options?: Partial<Options>;
+}
+interface EmbeddingsRequest {
+  model: string;
+  prompt: string;
+  keep_alive?: string | number;
+  options?: Partial<Options>;
+}
+interface GenerateResponse {
+  model: string;
+  created_at: Date;
+  response: string;
+  done: boolean;
+  context: number[];
+  total_duration: number;
+  load_duration: number;
+  prompt_eval_count: number;
+  prompt_eval_duration: number;
+  eval_count: number;
+  eval_duration: number;
+}
+interface ChatResponse {
+  model: string;
+  created_at: Date;
+  message: Message;
+  done: boolean;
+  total_duration: number;
+  load_duration: number;
+  prompt_eval_count: number;
+  prompt_eval_duration: number;
+  eval_count: number;
+  eval_duration: number;
+}
+interface EmbeddingsResponse {
+  embedding: number[];
+}
+interface ProgressResponse {
+  status: string;
+  digest: string;
+  total: number;
+  completed: number;
+}
+interface ModelResponse {
+  name: string;
+  modified_at: Date;
+  size: number;
+  digest: string;
+  details: ModelDetails;
+}
+interface ModelDetails {
+  parent_model: string;
+  format: string;
+  family: string;
+  families: string[];
+  parameter_size: string;
+  quantization_level: string;
+}
+interface ShowResponse {
+  license: string;
+  modelfile: string;
+  parameters: string;
+  template: string;
+  system: string;
+  details: ModelDetails;
+  messages: Message[];
+}
+interface ListResponse {
+  models: ModelResponse[];
+}
+interface ErrorResponse {
+  error: string;
+}
+interface StatusResponse {
+  status: string;
+}
+
+declare class Ollama {
+  protected readonly config: Config;
+  protected readonly fetch: Fetch;
+  private abortController;
+  constructor(config?: Partial<Config>);
+  abort(): void;
+  protected processStreamableRequest<T extends object>(
+    endpoint: string,
+    request: {
+      stream?: boolean;
+    } & Record<string, any>,
+  ): Promise<T | AsyncGenerator<T>>;
+  encodeImage(image: Uint8Array | string): Promise<string>;
+  generate(
+    request: GenerateRequest & {
+      stream: true;
+    },
+  ): Promise<AsyncGenerator<GenerateResponse>>;
+  generate(
+    request: GenerateRequest & {
+      stream?: false;
+    },
+  ): Promise<GenerateResponse>;
+  chat(
+    request: ChatRequest & {
+      stream: true;
+    },
+  ): Promise<AsyncGenerator<ChatResponse>>;
+  chat(
+    request: ChatRequest & {
+      stream?: false;
+    },
+  ): Promise<ChatResponse>;
+  create(
+    request: CreateRequest & {
+      stream: true;
+    },
+  ): Promise<AsyncGenerator<ProgressResponse>>;
+  create(
+    request: CreateRequest & {
+      stream?: false;
+    },
+  ): Promise<ProgressResponse>;
+  pull(
+    request: PullRequest & {
+      stream: true;
+    },
+  ): Promise<AsyncGenerator<ProgressResponse>>;
+  pull(
+    request: PullRequest & {
+      stream?: false;
+    },
+  ): Promise<ProgressResponse>;
+  push(
+    request: PushRequest & {
+      stream: true;
+    },
+  ): Promise<AsyncGenerator<ProgressResponse>>;
+  push(
+    request: PushRequest & {
+      stream?: false;
+    },
+  ): Promise<ProgressResponse>;
+  delete(request: DeleteRequest): Promise<StatusResponse>;
+  copy(request: CopyRequest): Promise<StatusResponse>;
+  list(): Promise<ListResponse>;
+  show(request: ShowRequest): Promise<ShowResponse>;
+  embeddings(request: EmbeddingsRequest): Promise<EmbeddingsResponse>;
+}
+declare const _default: Ollama;
+
+export {
+  Ollama,
+  _default as default,
+  type ChatRequest,
+  type ChatResponse,
+  type Config,
+  type CopyRequest,
+  type CreateRequest,
+  type DeleteRequest,
+  type EmbeddingsRequest,
+  type EmbeddingsResponse,
+  type ErrorResponse,
+  type Fetch,
+  type GenerateRequest,
+  type GenerateResponse,
+  type ListResponse,
+  type Message,
+  type ModelDetails,
+  type ModelResponse,
+  type Options,
+  type ProgressResponse,
+  type PullRequest,
+  type PushRequest,
+  type ShowRequest,
+  type ShowResponse,
+  type StatusResponse,
+};

packages/core/src/internal/deps/ollama.js

@@ -0,0 +1,462 @@
+// generate from "tsup ./src/browser.js --format esm --dts"
+var __defProp = Object.defineProperty;
+var __getOwnPropSymbols = Object.getOwnPropertySymbols;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __propIsEnum = Object.prototype.propertyIsEnumerable;
+var __knownSymbol = (name, symbol) => {
+  return (symbol = Symbol[name]) ? symbol : Symbol.for("Symbol." + name);
+};
+var __defNormalProp = (obj, key, value) =>
+  key in obj
+    ? __defProp(obj, key, {
+        enumerable: true,
+        configurable: true,
+        writable: true,
+        value,
+      })
+    : (obj[key] = value);
+var __spreadValues = (a, b) => {
+  for (var prop in b || (b = {}))
+    if (__hasOwnProp.call(b, prop)) __defNormalProp(a, prop, b[prop]);
+  if (__getOwnPropSymbols)
+    for (var prop of __getOwnPropSymbols(b)) {
+      if (__propIsEnum.call(b, prop)) __defNormalProp(a, prop, b[prop]);
+    }
+  return a;
+};
+var __async = (__this, __arguments, generator) => {
+  return new Promise((resolve, reject) => {
+    var fulfilled = (value) => {
+      try {
+        step(generator.next(value));
+      } catch (e) {
+        reject(e);
+      }
+    };
+    var rejected = (value) => {
+      try {
+        step(generator.throw(value));
+      } catch (e) {
+        reject(e);
+      }
+    };
+    var step = (x) =>
+      x.done
+        ? resolve(x.value)
+        : Promise.resolve(x.value).then(fulfilled, rejected);
+    step((generator = generator.apply(__this, __arguments)).next());
+  });
+};
+var __await = function (promise, isYieldStar) {
+  this[0] = promise;
+  this[1] = isYieldStar;
+};
+var __asyncGenerator = (__this, __arguments, generator) => {
+  var resume = (k, v, yes, no) => {
+    try {
+      var x = generator[k](v),
+        isAwait = (v = x.value) instanceof __await,
+        done = x.done;
+      Promise.resolve(isAwait ? v[0] : v)
+        .then((y) =>
+          isAwait
+            ? resume(
+                k === "return" ? k : "next",
+                v[1] ? { done: y.done, value: y.value } : y,
+                yes,
+                no,
+              )
+            : yes({ value: y, done }),
+        )
+        .catch((e) => resume("throw", e, yes, no));
+    } catch (e) {
+      no(e);
+    }
+  };
+  var method = (k) =>
+    (it[k] = (x) => new Promise((yes, no) => resume(k, x, yes, no)));
+  var it = {};
+  return (
+    (generator = generator.apply(__this, __arguments)),
+    (it[__knownSymbol("asyncIterator")] = () => it),
+    method("next"),
+    method("throw"),
+    method("return"),
+    it
+  );
+};
+var __forAwait = (obj, it, method) =>
+  (it = obj[__knownSymbol("asyncIterator")])
+    ? it.call(obj)
+    : ((obj = obj[__knownSymbol("iterator")]()),
+      (it = {}),
+      (method = (key, fn) =>
+        (fn = obj[key]) &&
+        (it[key] = (arg) =>
+          new Promise(
+            (yes, no, done) => (
+              (arg = fn.call(obj, arg)),
+              (done = arg.done),
+              Promise.resolve(arg.value).then(
+                (value) => yes({ value, done }),
+                no,
+              )
+            ),
+          ))),
+      method("next"),
+      method("return"),
+      it);
+
+// src/version.ts
+var version = "0.0.0";
+
+// src/utils.ts
+var ResponseError = class _ResponseError extends Error {
+  constructor(error, status_code) {
+    super(error);
+    this.error = error;
+    this.status_code = status_code;
+    this.name = "ResponseError";
+    if (Error.captureStackTrace) {
+      Error.captureStackTrace(this, _ResponseError);
+    }
+  }
+};
+var checkOk = (response) =>
+  __async(void 0, null, function* () {
+    var _a;
+    if (!response.ok) {
+      let message = `Error ${response.status}: ${response.statusText}`;
+      let errorData = null;
+      if (
+        (_a = response.headers.get("content-type")) == null
+          ? void 0
+          : _a.includes("application/json")
+      ) {
+        try {
+          errorData = yield response.json();
+          message = errorData.error || message;
+        } catch (error) {
+          console.log("Failed to parse error response as JSON");
+        }
+      } else {
+        try {
+          console.log("Getting text from response");
+          const textResponse = yield response.text();
+          message = textResponse || message;
+        } catch (error) {
+          console.log("Failed to get text from error response");
+        }
+      }
+      throw new ResponseError(message, response.status);
+    }
+  });
+function getPlatform() {
+  if (typeof window !== "undefined" && window.navigator) {
+    return `${window.navigator.platform.toLowerCase()} Browser/${navigator.userAgent};`;
+  } else if (typeof process !== "undefined") {
+    return `${process.arch} ${process.platform} Node.js/${process.version}`;
+  }
+  return "";
+}
+var fetchWithHeaders = (_0, _1, ..._2) =>
+  __async(void 0, [_0, _1, ..._2], function* (fetch2, url, options = {}) {
+    const defaultHeaders = {
+      "Content-Type": "application/json",
+      Accept: "application/json",
+      "User-Agent": `ollama-js/${version} (${getPlatform()})`,
+    };
+    if (!options.headers) {
+      options.headers = {};
+    }
+    options.headers = __spreadValues(
+      __spreadValues({}, defaultHeaders),
+      options.headers,
+    );
+    return fetch2(url, options);
+  });
+var get = (fetch2, host) =>
+  __async(void 0, null, function* () {
+    const response = yield fetchWithHeaders(fetch2, host);
+    yield checkOk(response);
+    return response;
+  });
+var post = (fetch2, host, data, options) =>
+  __async(void 0, null, function* () {
+    const isRecord = (input) => {
+      return (
+        input !== null && typeof input === "object" && !Array.isArray(input)
+      );
+    };
+    const formattedData = isRecord(data) ? JSON.stringify(data) : data;
+    const response = yield fetchWithHeaders(fetch2, host, {
+      method: "POST",
+      body: formattedData,
+      signal: options == null ? void 0 : options.signal,
+    });
+    yield checkOk(response);
+    return response;
+  });
+var del = (fetch2, host, data) =>
+  __async(void 0, null, function* () {
+    const response = yield fetchWithHeaders(fetch2, host, {
+      method: "DELETE",
+      body: JSON.stringify(data),
+    });
+    yield checkOk(response);
+    return response;
+  });
+var parseJSON = function (itr) {
+  return __asyncGenerator(this, null, function* () {
+    var _a;
+    const decoder = new TextDecoder("utf-8");
+    let buffer = "";
+    const reader = itr.getReader();
+    while (true) {
+      const { done, value: chunk } = yield new __await(reader.read());
+      if (done) {
+        break;
+      }
+      buffer += decoder.decode(chunk);
+      const parts = buffer.split("\n");
+      buffer = (_a = parts.pop()) != null ? _a : "";
+      for (const part of parts) {
+        try {
+          yield JSON.parse(part);
+        } catch (error) {
+          console.warn("invalid json: ", part);
+        }
+      }
+    }
+    for (const part of buffer.split("\n").filter((p) => p !== "")) {
+      try {
+        yield JSON.parse(part);
+      } catch (error) {
+        console.warn("invalid json: ", part);
+      }
+    }
+  });
+};
+var formatHost = (host) => {
+  if (!host) {
+    return "http://127.0.0.1:11434";
+  }
+  let isExplicitProtocol = host.includes("://");
+  if (host.startsWith(":")) {
+    host = `http://127.0.0.1${host}`;
+    isExplicitProtocol = false;
+  }
+  if (!isExplicitProtocol) {
+    host = `http://${host}`;
+  }
+  const url = new URL(host);
+  let port = url.port;
+  if (!port) {
+    if (!isExplicitProtocol) {
+      port = "11434";
+    } else {
+      port = url.protocol === "https:" ? "443" : "80";
+    }
+  }
+  let formattedHost = `${url.protocol}//${url.hostname}:${port}${url.pathname}`;
+  if (formattedHost.endsWith("/")) {
+    formattedHost = formattedHost.slice(0, -1);
+  }
+  return formattedHost;
+};
+
+// src/browser.ts
+// import "whatwg-fetch";
+var Ollama = class {
+  constructor(config) {
+    var _a;
+    this.config = {
+      host: "",
+    };
+    if (!(config == null ? void 0 : config.proxy)) {
+      this.config.host = formatHost(
+        (_a = config == null ? void 0 : config.host) != null
+          ? _a
+          : "http://127.0.0.1:11434",
+      );
+    }
+    this.fetch = fetch;
+    if ((config == null ? void 0 : config.fetch) != null) {
+      this.fetch = config.fetch;
+    }
+    this.abortController = new AbortController();
+  }
+  // Abort any ongoing requests to Ollama
+  abort() {
+    this.abortController.abort();
+    this.abortController = new AbortController();
+  }
+  processStreamableRequest(endpoint, request) {
+    return __async(this, null, function* () {
+      var _a;
+      request.stream = (_a = request.stream) != null ? _a : false;
+      const response = yield post(
+        this.fetch,
+        `${this.config.host}/api/${endpoint}`,
+        __spreadValues({}, request),
+        { signal: this.abortController.signal },
+      );
+      if (!response.body) {
+        throw new Error("Missing body");
+      }
+      const itr = parseJSON(response.body);
+      if (request.stream) {
+        return (function () {
+          return __asyncGenerator(this, null, function* () {
+            try {
+              for (
+                var iter = __forAwait(itr), more, temp, error;
+                (more = !(temp = yield new __await(iter.next())).done);
+                more = false
+              ) {
+                const message = temp.value;
+                if ("error" in message) {
+                  throw new Error(message.error);
+                }
+                yield message;
+                if (message.done || message.status === "success") {
+                  return;
+                }
+              }
+            } catch (temp) {
+              error = [temp];
+            } finally {
+              try {
+                more &&
+                  (temp = iter.return) &&
+                  (yield new __await(temp.call(iter)));
+              } finally {
+                if (error) throw error[0];
+              }
+            }
+            throw new Error(
+              "Did not receive done or success response in stream.",
+            );
+          });
+        })();
+      } else {
+        const message = yield itr.next();
+        if (!message.value.done && message.value.status !== "success") {
+          throw new Error("Expected a completed response.");
+        }
+        return message.value;
+      }
+    });
+  }
+  encodeImage(image) {
+    return __async(this, null, function* () {
+      if (typeof image !== "string") {
+        const uint8Array = new Uint8Array(image);
+        const numberArray = Array.from(uint8Array);
+        const base64String = btoa(String.fromCharCode.apply(null, numberArray));
+        return base64String;
+      }
+      return image;
+    });
+  }
+  generate(request) {
+    return __async(this, null, function* () {
+      if (request.images) {
+        request.images = yield Promise.all(
+          request.images.map(this.encodeImage.bind(this)),
+        );
+      }
+      return this.processStreamableRequest("generate", request);
+    });
+  }
+  chat(request) {
+    return __async(this, null, function* () {
+      if (request.messages) {
+        for (const message of request.messages) {
+          if (message.images) {
+            message.images = yield Promise.all(
+              message.images.map(this.encodeImage.bind(this)),
+            );
+          }
+        }
+      }
+      return this.processStreamableRequest("chat", request);
+    });
+  }
+  create(request) {
+    return __async(this, null, function* () {
+      return this.processStreamableRequest("create", {
+        name: request.model,
+        stream: request.stream,
+        modelfile: request.modelfile,
+      });
+    });
+  }
+  pull(request) {
+    return __async(this, null, function* () {
+      return this.processStreamableRequest("pull", {
+        name: request.model,
+        stream: request.stream,
+        insecure: request.insecure,
+      });
+    });
+  }
+  push(request) {
+    return __async(this, null, function* () {
+      return this.processStreamableRequest("push", {
+        name: request.model,
+        stream: request.stream,
+        insecure: request.insecure,
+      });
+    });
+  }
+  delete(request) {
+    return __async(this, null, function* () {
+      yield del(this.fetch, `${this.config.host}/api/delete`, {
+        name: request.model,
+      });
+      return { status: "success" };
+    });
+  }
+  copy(request) {
+    return __async(this, null, function* () {
+      yield post(
+        this.fetch,
+        `${this.config.host}/api/copy`,
+        __spreadValues({}, request),
+      );
+      return { status: "success" };
+    });
+  }
+  list() {
+    return __async(this, null, function* () {
+      const response = yield get(this.fetch, `${this.config.host}/api/tags`);
+      const listResponse = yield response.json();
+      return listResponse;
+    });
+  }
+  show(request) {
+    return __async(this, null, function* () {
+      const response = yield post(
+        this.fetch,
+        `${this.config.host}/api/show`,
+        __spreadValues({}, request),
+      );
+      const showResponse = yield response.json();
+      return showResponse;
+    });
+  }
+  embeddings(request) {
+    return __async(this, null, function* () {
+      const response = yield post(
+        this.fetch,
+        `${this.config.host}/api/embeddings`,
+        __spreadValues({}, request),
+      );
+      const embeddingsResponse = yield response.json();
+      return embeddingsResponse;
+    });
+  }
+};
+var browser_default = new Ollama();
+export { Ollama, browser_default as default };

packages/core/src/internal/deps/ollama.license

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 Saul
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

packages/core/src/internal/deps/transformers.ts

@@ -0,0 +1,23 @@
+/**
+ * Copy/paste from `@xenova/transformers` dist and types folders
+ *
+ * @module
+ */
+import { runtime } from "std-env";
+// @ts-expect-error
+let transformer: typeof import("./transformers/transformers.js") | null = null;
+
+export async function lazyLoadTransformers(): Promise<
+  typeof import("@xenova/transformers")
+> {
+  if (!transformer) {
+    // @ts-expect-error
+    transformer = await import("./transformers/transformers.js");
+  }
+
+  if (runtime !== "node" && runtime !== "deno" && runtime !== "bun") {
+    // there is no local file system for such runtimes
+    transformer.env.allowLocalModels = false;
+  }
+  return transformer;
+}

packages/core/src/internal/settings/chunk-size.ts

@@ -0,0 +1,19 @@
+import { AsyncLocalStorage } from "@llamaindex/env";
+
+const chunkSizeAsyncLocalStorage = new AsyncLocalStorage<number | undefined>();
+const globalChunkSize: number | null = null;
+
+export function getChunkSize(): number | undefined {
+  return globalChunkSize ?? chunkSizeAsyncLocalStorage.getStore();
+}
+
+export function setChunkSize(chunkSize: number | undefined) {
+  chunkSizeAsyncLocalStorage.enterWith(chunkSize);
+}
+
+export function withChunkSize<Result>(
+  embeddedModel: number,
+  fn: () => Result,
+): Result {
+  return chunkSizeAsyncLocalStorage.run(embeddedModel, fn);
+}

packages/core/src/llm/anthropic.ts

@@ -156,58 +156,104 @@ export class Anthropic extends ToolCallLLM<AnthropicAdditionalChatOptions> {
   formatMessages<Beta = false>(
     messages: ChatMessage<ToolCallLLMMessageOptions>[],
   ): Beta extends true ? ToolsBetaMessageParam[] : MessageParam[] {
-    return messages.map<any>((message) => {
-      if (message.role !== "user" && message.role !== "assistant") {
-        throw new Error("Unsupported Anthropic role");
-      }
-      const options = message.options ?? {};
-      if ("toolResult" in options) {
-        const { id, isError } = options.toolResult;
-        return {
-          role: "user",
-          content: [
-            {
-              type: "tool_result",
-              is_error: isError,
-              content: [
-                {
-                  type: "text",
-                  text: extractText(message.content),
-                },
-              ],
-              tool_use_id: id,
-            },
-          ] satisfies ToolResultBlockParam[],
-        } satisfies ToolsBetaMessageParam;
-      } else if ("toolCall" in options) {
-        const aiThinkingText = extractText(message.content);
-        return {
-          role: "assistant",
-          content: [
-            // this could be empty when you call two tools in one query
-            ...(aiThinkingText.trim()
-              ? [
+    const result: ToolsBetaMessageParam[] = messages
+      .filter(
+        (message) => message.role === "user" || message.role === "assistant",
+      )
+      .map((message) => {
+        const options = message.options ?? {};
+        if ("toolResult" in options) {
+          const { id, isError } = options.toolResult;
+          return {
+            role: "user",
+            content: [
+              {
+                type: "tool_result",
+                is_error: isError,
+                content: [
                   {
                     type: "text",
-                    text: aiThinkingText,
-                  } satisfies TextBlockParam,
-                ]
-              : []),
-            {
-              type: "tool_use",
-              id: options.toolCall.id,
-              name: options.toolCall.name,
-              input: options.toolCall.input,
-            } satisfies ToolUseBlockParam,
-          ] satisfies ToolsBetaContentBlock[],
-        } satisfies ToolsBetaMessageParam;
-      }
+                    text: extractText(message.content),
+                  },
+                ],
+                tool_use_id: id,
+              },
+            ] satisfies ToolResultBlockParam[],
+          } satisfies ToolsBetaMessageParam;
+        } else if ("toolCall" in options) {
+          const aiThinkingText = extractText(message.content);
+          return {
+            role: "assistant",
+            content: [
+              // this could be empty when you call two tools in one query
+              ...(aiThinkingText.trim()
+                ? [
+                    {
+                      type: "text",
+                      text: aiThinkingText,
+                    } satisfies TextBlockParam,
+                  ]
+                : []),
+              {
+                type: "tool_use",
+                id: options.toolCall.id,
+                name: options.toolCall.name,
+                input: options.toolCall.input,
+              } satisfies ToolUseBlockParam,
+            ] satisfies ToolsBetaContentBlock[],
+          } satisfies ToolsBetaMessageParam;
+        }
 
-      return {
-        content: extractText(message.content),
-        role: message.role,
-      } satisfies MessageParam;
-    });
+        return {
+          content: extractText(message.content),
+          role: message.role as "user" | "assistant",
+        } satisfies MessageParam;
+      });
+    // merge messages with the same role
+    // in case of 'messages: roles must alternate between "user" and "assistant", but found multiple "user" roles in a row'
+    const realResult: ToolsBetaMessageParam[] = [];
+    for (let i = 0; i < result.length; i++) {
+      if (i === 0) {
+        realResult.push(result[i]);
+        continue;
+      }
+      const current = result[i];
+      const previous = result[i - 1];
+      if (current.role === previous.role) {
+        // merge two messages with the same role
+        if (Array.isArray(previous.content)) {
+          if (Array.isArray(current.content)) {
+            previous.content.push(...current.content);
+          } else {
+            previous.content.push({
+              type: "text",
+              text: current.content,
+            });
+          }
+        } else {
+          if (Array.isArray(current.content)) {
+            previous.content = [
+              {
+                type: "text",
+                text: previous.content,
+              },
+              ...current.content,
+            ];
+          } else {
+            previous.content += `\n${current.content}`;
+          }
+        }
+        // no need to push the message
+      }
+      // if the roles are different, just push the message
+      else {
+        realResult.push(current);
+      }
+    }
+
+    return realResult as Beta extends true
+      ? ToolsBetaMessageParam[]
+      : MessageParam[];
   }
 
   chat(

packages/core/src/llm/gemini.ts

@@ -139,7 +139,7 @@ class GeminiHelper {
   > = {
     user: "user",
     system: "user",
-    assistant: "user",
+    assistant: "model",
     memory: "user",
   };
 
@@ -152,38 +152,26 @@ class GeminiHelper {
   };
 
   public static mergeNeighboringSameRoleMessages(
-    messages: ChatMessage[],
-  ): ChatMessage[] {
-    // Gemini does not support multiple messages of the same role in a row, so we merge them
-    const mergedMessages: ChatMessage[] = [];
-    let i: number = 0;
-
-    while (i < messages.length) {
-      const currentMessage: ChatMessage = messages[i];
-      // Initialize merged content with current message content
-      const mergedContent: MessageContent[] = [currentMessage.content];
-
-      // Check if the next message exists and has the same role
-      while (
-        i + 1 < messages.length &&
-        this.ROLES_TO_GEMINI[messages[i + 1].role] ===
-          this.ROLES_TO_GEMINI[currentMessage.role]
-      ) {
-        i++;
-        const nextMessage: ChatMessage = messages[i];
-        mergedContent.push(nextMessage.content);
-      }
-
-      // Create a new ChatMessage object with merged content
-      const mergedMessage: ChatMessage = {
-        role: currentMessage.role,
-        content: mergedContent.join("\n"),
-      };
-      mergedMessages.push(mergedMessage);
-      i++;
-    }
-
-    return mergedMessages;
+    messages: GeminiMessageContent[],
+  ): GeminiMessageContent[] {
+    return messages.reduce(
+      (
+        result: GeminiMessageContent[],
+        current: GeminiMessageContent,
+        index: number,
+      ) => {
+        if (index > 0 && messages[index - 1].role === current.role) {
+          result[result.length - 1].parts = [
+            ...result[result.length - 1].parts,
+            ...current.parts,
+          ];
+        } else {
+          result.push(current);
+        }
+        return result;
+      },
+      [],
+    );
   }
 
   public static messageContentToGeminiParts(content: MessageContent): Part[] {
@@ -214,8 +202,8 @@ class GeminiHelper {
     message: ChatMessage,
   ): GeminiMessageContent {
     return {
-      role: this.ROLES_TO_GEMINI[message.role],
-      parts: this.messageContentToGeminiParts(message.content),
+      role: GeminiHelper.ROLES_TO_GEMINI[message.role],
+      parts: GeminiHelper.messageContentToGeminiParts(message.content),
     };
   }
 }
@@ -260,22 +248,20 @@ export class Gemini extends ToolCallLLM<GeminiAdditionalChatOptions> {
     chat: ChatSession;
     messageContent: Part[];
   } {
-    const { messages } = params;
-    const mergedMessages =
-      GeminiHelper.mergeNeighboringSameRoleMessages(messages);
-    const history = mergedMessages.slice(0, -1);
-    const nextMessage = mergedMessages[mergedMessages.length - 1];
-    const messageContent = GeminiHelper.chatMessageToGemini(nextMessage).parts;
+    const messages = GeminiHelper.mergeNeighboringSameRoleMessages(
+      params.messages.map(GeminiHelper.chatMessageToGemini),
+    );
+
+    const history = messages.slice(0, -1);
 
     const client = this.session.gemini.getGenerativeModel(this.metadata);
 
     const chat = client.startChat({
-      history: history.map(GeminiHelper.chatMessageToGemini),
+      history,
     });
-
     return {
       chat,
-      messageContent,
+      messageContent: messages[messages.length - 1].parts,
     };
   }
 

packages/core/src/llm/index.ts

@@ -5,7 +5,7 @@ export {
   Anthropic,
 } from "./anthropic.js";
 export { FireworksLLM } from "./fireworks.js";
-export { GEMINI_MODEL, Gemini } from "./gemini.js";
+export { GEMINI_MODEL, Gemini, GeminiSession } from "./gemini.js";
 export { Groq } from "./groq.js";
 export { HuggingFaceInferenceAPI } from "./huggingface.js";
 export {
@@ -17,6 +17,7 @@ export * from "./openai.js";
 export { Portkey } from "./portkey.js";
 export * from "./replicate_ai.js";
 // Note: The type aliases for replicate are to simplify usage for Llama 2 (we're using replicate for Llama 2 support)
+export { Ollama, type OllamaParams } from "./ollama.js";
 export {
   ALL_AVAILABLE_REPLICATE_MODELS,
   DeuceChatStrategy,

packages/core/src/llm/ollama.ts

@@ -1,12 +1,24 @@
-import ollama, {
+import { BaseEmbedding } from "../embeddings/types.js";
+import {
+  Ollama as OllamaBase,
+  type Config,
+  type CopyRequest,
   type CreateRequest,
+  type DeleteRequest,
+  type EmbeddingsRequest,
+  type EmbeddingsResponse,
+  type GenerateRequest,
+  type ListResponse,
   type ChatResponse as OllamaChatResponse,
   type GenerateResponse as OllamaGenerateResponse,
   type Options,
   type ProgressResponse,
+  type PullRequest,
+  type PushRequest,
   type ShowRequest,
-} from "ollama/browser";
-import { BaseEmbedding } from "../embeddings/types.js";
+  type ShowResponse,
+  type StatusResponse,
+} from "../internal/deps/ollama.js";
 import type {
   ChatResponse,
   ChatResponseChunk,
@@ -35,15 +47,21 @@ const completionAccessor = (
 
 export type OllamaParams = {
   model: string;
+  config?: Partial<Config>;
   options?: Partial<Options>;
 };
 
 /**
  * This class both implements the LLM and Embedding interfaces.
  */
-export class Ollama extends BaseEmbedding implements LLM {
+export class Ollama
+  extends BaseEmbedding
+  implements LLM, Omit<OllamaBase, "chat">
+{
   readonly hasStreaming = true;
 
+  ollama: OllamaBase;
+
   // https://ollama.ai/library
   model: string;
 
@@ -57,6 +75,7 @@ export class Ollama extends BaseEmbedding implements LLM {
   constructor(params: OllamaParams) {
     super();
     this.model = params.model;
+    this.ollama = new OllamaBase(params.config);
     if (params.options) {
       this.options = {
         ...this.options,
@@ -97,7 +116,7 @@ export class Ollama extends BaseEmbedding implements LLM {
       },
     };
     if (!stream) {
-      const chatResponse = await ollama.chat({
+      const chatResponse = await this.ollama.chat({
         ...payload,
         stream: false,
       });
@@ -110,7 +129,7 @@ export class Ollama extends BaseEmbedding implements LLM {
         raw: chatResponse,
       };
     } else {
-      const stream = await ollama.chat({
+      const stream = await this.ollama.chat({
         ...payload,
         stream: true,
       });
@@ -137,7 +156,7 @@ export class Ollama extends BaseEmbedding implements LLM {
       },
     };
     if (!stream) {
-      const response = await ollama.generate({
+      const response = await this.ollama.generate({
         ...payload,
         stream: false,
       });
@@ -146,7 +165,7 @@ export class Ollama extends BaseEmbedding implements LLM {
         raw: response,
       };
     } else {
-      const stream = await ollama.generate({
+      const stream = await this.ollama.generate({
         ...payload,
         stream: true,
       });
@@ -162,7 +181,7 @@ export class Ollama extends BaseEmbedding implements LLM {
         ...this.options,
       },
     };
-    const response = await ollama.embeddings({
+    const response = await this.ollama.embeddings({
       ...payload,
     });
     return response.embedding;
@@ -176,104 +195,63 @@ export class Ollama extends BaseEmbedding implements LLM {
     return this.getEmbedding(query);
   }
 
-  // ollama specific methods, inherited from the ollama library
-  static async list() {
-    const { models } = await ollama.list();
-    return models;
-  }
+  // Inherited from OllamaBase
 
-  static async detail(modelName: string, options?: Omit<ShowRequest, "model">) {
-    return ollama.show({
-      model: modelName,
-      ...options,
-    });
-  }
-
-  static async create(
-    modelName: string,
-    options?: Omit<CreateRequest, "model"> & {
-      stream: false;
-    },
+  push(
+    request: PushRequest & { stream: true },
+  ): Promise<AsyncGenerator<ProgressResponse, any, unknown>>;
+  push(
+    request: PushRequest & { stream?: false | undefined },
   ): Promise<ProgressResponse>;
-  static async create(
-    modelName: string,
-    options: Omit<CreateRequest, "model"> & {
-      stream: true;
-    },
+  push(request: any): any {
+    return this.ollama.push(request);
+  }
+  abort(): void {
+    return this.ollama.abort();
+  }
+  encodeImage(image: string | Uint8Array): Promise<string> {
+    return this.ollama.encodeImage(image);
+  }
+  generate(
+    request: GenerateRequest & { stream: true },
+  ): Promise<AsyncGenerator<OllamaGenerateResponse>>;
+  generate(
+    request: GenerateRequest & { stream?: false | undefined },
+  ): Promise<OllamaGenerateResponse>;
+  generate(request: any): any {
+    return this.ollama.generate(request);
+  }
+  create(
+    request: CreateRequest & { stream: true },
   ): Promise<AsyncGenerator<ProgressResponse>>;
-  static async create(
-    modelName: string,
-    options?: Omit<CreateRequest, "model"> & {
-      stream: boolean;
-    },
-  ) {
-    return ollama.create({
-      model: modelName,
-      ...options,
-      stream: (options ? !!options.stream : false) as never,
-    }) as Promise<ProgressResponse> | Promise<AsyncGenerator<ProgressResponse>>;
-  }
-
-  static async delete(modelName: string) {
-    return ollama.delete({
-      model: modelName,
-    });
-  }
-
-  static async copy(source: string, destination: string) {
-    return ollama.copy({
-      source,
-      destination,
-    });
-  }
-
-  static async pull(
-    modelName: string,
-    options?: Omit<CreateRequest, "model"> & {
-      stream: false;
-    },
+  create(
+    request: CreateRequest & { stream?: false | undefined },
   ): Promise<ProgressResponse>;
-  static async pull(
-    modelName: string,
-    options: Omit<CreateRequest, "model"> & {
-      stream: true;
-    },
-  ): Promise<AsyncGenerator<ProgressResponse>>;
-  static async pull(
-    modelName: string,
-    options?: Omit<CreateRequest, "model"> & {
-      stream: boolean;
-    },
-  ) {
-    return ollama.pull({
-      model: modelName,
-      ...options,
-      stream: (options ? !!options.stream : false) as never,
-    }) as Promise<ProgressResponse> | Promise<AsyncGenerator<ProgressResponse>>;
+  create(request: any): any {
+    return this.ollama.create(request);
   }
-
-  static async push(
-    modelName: string,
-    options?: Omit<CreateRequest, "model"> & {
-      stream: false;
-    },
-  ): Promise<ProgressResponse>;
-  static async push(
-    modelName: string,
-    options: Omit<CreateRequest, "model"> & {
-      stream: true;
-    },
+  pull(
+    request: PullRequest & { stream: true },
   ): Promise<AsyncGenerator<ProgressResponse>>;
-  static async push(
-    modelName: string,
-    options?: Omit<CreateRequest, "model"> & {
-      stream: boolean;
-    },
-  ) {
-    return ollama.push({
-      model: modelName,
-      ...options,
-      stream: (options ? !!options.stream : false) as never,
-    }) as Promise<ProgressResponse> | Promise<AsyncGenerator<ProgressResponse>>;
+  pull(
+    request: PullRequest & { stream?: false | undefined },
+  ): Promise<ProgressResponse>;
+  pull(request: any): any {
+    return this.ollama.pull(request);
+  }
+  delete(request: DeleteRequest): Promise<StatusResponse> {
+    return this.ollama.delete(request);
+  }
+  copy(request: CopyRequest): Promise<StatusResponse> {
+    return this.ollama.copy(request);
+  }
+  list(): Promise<ListResponse> {
+    return this.ollama.list();
+  }
+  show(request: ShowRequest): Promise<ShowResponse> {
+    return this.ollama.show(request);
+  }
+  embeddings(request: EmbeddingsRequest): Promise<EmbeddingsResponse> {
+    return this.ollama.embeddings(request);
   }
 }