Update pyproject.toml

back out of version bump

.changeset/ninety-goats-look.md

@@ -0,0 +1,6 @@
+---
+"llama_parse": patch
+"llama-cloud-services-py": patch
+---
+
+Remove extension filter

extract.md

@@ -398,6 +398,6 @@ Another option (orthogonal to the above) is to break the document into smaller s
 ## Additional Resources
 
 - [Extract Documentation](https://docs.cloud.llamaindex.ai/llamaextract/getting_started) - Details on Extract features, API and examples.
-- [Example Notebook](docs/examples-py/extract/resume_screening.ipynb) - Detailed walkthrough of resume parsing
+- [Example Notebook](examples/extract/resume_screening.ipynb) - Detailed walkthrough of resume parsing
 - [Example Application with TypeScript](./examples-ts/extract/) - End-to-end examples using LlamaExtract TypeScript client.
 - [Discord Community](https://discord.com/invite/eN6D2HQ4aX) - Get help and share feedback

parse.md

@@ -97,7 +97,7 @@ for page in result.pages:
     print(page.structuredData)
 ```
 
-See more details about the result object in the [example notebook](./docs/examples-py/parse/demo_json_tour.ipynb).
+See more details about the result object in the [example notebook](./examples/parse/demo_json_tour.ipynb).
 
 ### Using with file object / bytes
 
@@ -153,10 +153,10 @@ Full documentation for `SimpleDirectoryReader` can be found on the [LlamaIndex D
 
 Several end-to-end indexing examples can be found in the examples folder
 
-- [Getting Started](docs/examples-py/parse/demo_basic.ipynb)
-- [Advanced RAG Example](docs/examples-py/parse/demo_advanced.ipynb)
-- [Raw API Usage](docs/examples-py/parse/demo_api.ipynb)
-- [Result Object Tour](docs/examples-py/parse/demo_json_tour.ipynb)
+- [Getting Started](examples/parse/demo_basic.ipynb)
+- [Advanced RAG Example](examples/parse/demo_advanced.ipynb)
+- [Raw API Usage](examples/parse/demo_api.ipynb)
+- [Result Object Tour](examples/parse/demo_json_tour.ipynb)
 
 ## Documentation
 

py/CHANGELOG.md

@@ -1,5 +1,11 @@
 # llama-cloud-services-py
 
+## 0.6.89
+
+### Patch Changes
+
+- b9b83c9: Parse bounding boxes from extract jobs results in agent data
+
 ## 0.6.88
 
 ### Patch Changes

py/llama_cloud_services/beta/agent_data/__init__.py

@@ -11,6 +11,9 @@ from .schema import (
     InvalidExtractionData,
     ExtractedFieldMetadata,
     ExtractedFieldMetaDataDict,
+    FieldCitation,
+    BoundingBox,
+    PageDimensions,
 )
 from .client import AsyncAgentDataClient
 
@@ -28,4 +31,7 @@ __all__ = [
     "InvalidExtractionData",
     "ExtractedFieldMetadata",
     "ExtractedFieldMetaDataDict",
+    "FieldCitation",
+    "BoundingBox",
+    "PageDimensions",
 ]

py/llama_cloud_services/beta/agent_data/schema.py

@@ -174,6 +174,22 @@ class TypedAgentDataItems(BaseModel, Generic[AgentDataT]):
     )
 
 
+class BoundingBox(BaseModel):
+    """Bounding box coordinates for a citation location on a page."""
+
+    x: float = Field(description="X coordinate of the bounding box origin")
+    y: float = Field(description="Y coordinate of the bounding box origin")
+    w: float = Field(description="Width of the bounding box")
+    h: float = Field(description="Height of the bounding box")
+
+
+class PageDimensions(BaseModel):
+    """Dimensions of a page in the source document."""
+
+    width: float = Field(description="Width of the page")
+    height: float = Field(description="Height of the page")
+
+
 class FieldCitation(BaseModel):
     page: Optional[int] = Field(
         None, description="The page number that the field occurred on"
@@ -182,6 +198,14 @@ class FieldCitation(BaseModel):
         None,
         description="The original text this field's value was derived from",
     )
+    bounding_boxes: Optional[List[BoundingBox]] = Field(
+        None,
+        description="Bounding boxes indicating where the citation appears on the page",
+    )
+    page_dimensions: Optional[PageDimensions] = Field(
+        None,
+        description="Dimensions of the page containing the citation",
+    )
 
 
 class ExtractedFieldMetadata(BaseModel):
@@ -201,6 +225,10 @@ class ExtractedFieldMetadata(BaseModel):
         None,
         description="The confidence score for the field based on the extracted text only",
     )
+    parsing_confidence: Optional[float] = Field(
+        None,
+        description="The confidence score for the field based on the parsing/OCR quality",
+    )
     citation: Optional[List[FieldCitation]] = Field(
         None,
         description="The citation for the field, including page number and matching text",

py/llama_cloud_services/parse/base.py

@@ -751,11 +751,9 @@ class LlamaParse(BasePydanticReader):
             file_path = str(file_input)
             file_ext = os.path.splitext(file_path)[1].lower()
             if file_ext not in SUPPORTED_FILE_TYPES:
-                raise Exception(
-                    f"Currently, only the following file types are supported: {SUPPORTED_FILE_TYPES}\n"
-                    f"Current file type: {file_ext}"
-                )
-            mime_type = mimetypes.guess_type(file_path)[0]
+                mime_type = "application/octet-stream"
+            else:
+                mime_type = mimetypes.guess_type(file_path)[0]
             # Open the file here for the duration of the async context
             # load data, set the mime type
             fs = fs or get_default_fs()

py/llama_parse/CHANGELOG.md

@@ -1,5 +1,12 @@
 # llama_parse
 
+## 0.6.89
+
+### Patch Changes
+
+- Updated dependencies [b9b83c9]
+  - llama-cloud-services-py@0.6.89
+
 ## 0.6.88
 
 ### Patch Changes

py/llama_parse/README.md

@@ -146,9 +146,9 @@ Full documentation for `SimpleDirectoryReader` can be found on the [LlamaIndex D
 
 Several end-to-end indexing examples can be found in the examples folder
 
-- [Getting Started](/docs/examples-py/parse/demo_basic.ipynb)
-- [Advanced RAG Example](/docs/examples-py/parse/demo_advanced.ipynb)
-- [Raw API Usage](/docs/examples-py/parse/demo_api.ipynb)
+- [Getting Started](../../examples/parse/demo_basic.ipynb)
+- [Advanced RAG Example](../../examples/parse/demo_advanced.ipynb)
+- [Raw API Usage](../../examples/parse/demo_api.ipynb)
 
 ## Documentation
 

py/llama_parse/package.json

@@ -1,6 +1,6 @@
 {
   "name": "llama_parse",
-  "version": "0.6.88",
+  "version": "0.6.89",
   "description": "",
   "main": "index.js",
   "private": false,

py/llama_parse/pyproject.toml

@@ -11,13 +11,13 @@ dev = [
 
 [project]
 name = "llama-parse"
-version = "0.6.88"
+version = "0.6.89"
 description = "Parse files into RAG-Optimized formats."
 authors = [{name = "Logan Markewich", email = "logan@llamaindex.ai"}]
 requires-python = ">=3.9,<4.0"
 readme = "README.md"
 license = "MIT"
-dependencies = ["llama-cloud-services>=0.6.88"]
+dependencies = ["llama-cloud-services>=0.6.89"]
 
 [project.scripts]
 llama-parse = "llama_parse.cli.main:parse"

py/package.json

@@ -1,6 +1,6 @@
 {
   "name": "llama-cloud-services-py",
-  "version": "0.6.88",
+  "version": "0.6.89",
   "private": false,
   "license": "MIT",
   "scripts": {},

py/pyproject.toml

@@ -23,7 +23,7 @@ dev = [
 
 [project]
 name = "llama-cloud-services"
-version = "0.6.88"
+version = "0.6.89"
 description = "Tailored SDK clients for LlamaCloud services."
 authors = [{name = "Logan Markewich", email = "logan@runllama.ai"}]
 requires-python = ">=3.9,<4.0"

py/unit_tests/beta/agent/test_agent_data_schema.py

@@ -11,10 +11,12 @@ from llama_cloud.types.aggregate_group import AggregateGroup
 from pydantic import BaseModel, Field, ValidationError
 
 from llama_cloud_services.beta.agent_data.schema import (
+    BoundingBox,
     ExtractedData,
     ExtractedFieldMetadata,
     FieldCitation,
     InvalidExtractionData,
+    PageDimensions,
     TypedAgentData,
     TypedAggregateGroup,
     calculate_overall_confidence,
@@ -663,3 +665,69 @@ def test_field_conflict_in_schema():
     assert isinstance(
         extracted["majority_opinion"]["reasoning"], ExtractedFieldMetadata
     )
+
+
+def test_parse_extracted_field_metadata_with_bounding_boxes():
+    """Test parse_extracted_field_metadata with bounding boxes and page dimensions."""
+    raw_metadata = {
+        "document_type": {
+            "citation": [
+                {
+                    "page": 1,
+                    "matching_text": "FACTURE ORIGINALE",
+                    "bounding_boxes": [{"x": 77.28, "y": 615.12, "w": 70.6, "h": 7.2}],
+                    "page_dimensions": {"width": 222.24, "height": 736.56},
+                }
+            ],
+            "parsing_confidence": 1.0,
+            "extraction_confidence": 0.7252506422636493,
+            "confidence": 0.7252506422636493,
+        },
+        "summary": {
+            "citation": [
+                {
+                    "page": 1,
+                    "matching_text": "FACTURE ORIGINALE",
+                    "bounding_boxes": [{"x": 77.28, "y": 615.12, "w": 70.6, "h": 7.2}],
+                    "page_dimensions": {"width": 222.24, "height": 736.56},
+                },
+                {
+                    "page": 1,
+                    "matching_text": "Café filtre assiette — $1.90",
+                    "bounding_boxes": [
+                        {"x": 10.56, "y": 172.83, "w": 171.85, "h": 497.01}
+                    ],
+                    "page_dimensions": {"width": 222.24, "height": 736.56},
+                },
+            ],
+            "parsing_confidence": 1.0,
+            "extraction_confidence": 0.5700013128334419,
+            "confidence": 0.5700013128334419,
+        },
+    }
+
+    result = parse_extracted_field_metadata(raw_metadata)
+
+    # Verify document_type citation with bounding boxes
+    assert isinstance(result["document_type"], ExtractedFieldMetadata)
+    assert result["document_type"].parsing_confidence == 1.0
+    assert result["document_type"].extraction_confidence == 0.7252506422636493
+    assert result["document_type"].confidence == 0.7252506422636493
+    assert len(result["document_type"].citation) == 1
+
+    citation = result["document_type"].citation[0]
+    assert citation.page == 1
+    assert citation.matching_text == "FACTURE ORIGINALE"
+    assert len(citation.bounding_boxes) == 1
+    assert citation.bounding_boxes[0] == BoundingBox(x=77.28, y=615.12, w=70.6, h=7.2)
+    assert citation.page_dimensions == PageDimensions(width=222.24, height=736.56)
+
+    # Verify summary citation with multiple bounding boxes
+    assert isinstance(result["summary"], ExtractedFieldMetadata)
+    assert len(result["summary"].citation) == 2
+    assert result["summary"].citation[0].bounding_boxes[0].x == 77.28
+    assert result["summary"].citation[1].bounding_boxes[0].x == 10.56
+
+    # Verify round-trip serialization
+    result2 = parse_extracted_field_metadata(result)
+    assert result2 == result

py/uv.lock

@@ -1,5 +1,5 @@
 version = 1
-revision = 2
+revision = 3
 requires-python = ">=3.9, <4.0"
 resolution-markers = [
     "python_full_version >= '3.14'",
@@ -1609,7 +1609,7 @@ wheels = [
 
 [[package]]
 name = "llama-cloud-services"
-version = "0.6.85"
+version = "0.6.88"
 source = { editable = "." }
 dependencies = [
     { name = "click", version = "8.1.8", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.10'" },

ts/llama_cloud_services/CHANGELOG.md

@@ -1,5 +1,38 @@
 # llama-cloud-services
 
+## 0.5.3
+
+### Patch Changes
+
+- d7864af: bugfixes in retry logic for LlamaExtract and LlamaClassify
+
+## 0.5.2
+
+### Patch Changes
+
+- 997bcc8: Add types for bounding boxes
+
+## 0.5.1
+
+### Patch Changes
+
+- d5b18a0: Fix publishing
+
+## 0.5.0
+
+### Minor Changes
+
+- 576c3d9: feat: support zod v4 & v3
+
+  Adds support for zod v4 while maintaining backward compatibility with v3.
+  - Updated zod peer dependency to accept both v3 and v4: `^3.25.76 || ^4.0.0`
+  - Migrated all import statements to use `zod/v4` import path for compatibility
+
+### Patch Changes
+
+- c8321d2: Improve parse results polling
+- 576c3d9: Support zod v3 an v4
+
 ## 0.4.3
 
 ### Patch Changes

ts/llama_cloud_services/package.json

@@ -1,12 +1,12 @@
 {
   "name": "llama-cloud-services",
-  "version": "0.4.3",
+  "version": "0.5.3",
   "type": "module",
   "license": "MIT",
   "scripts": {
     "get-openapi": "node ./scripts/get-openapi.js",
     "generate": "./node_modules/.bin/openapi-ts",
-    "build": "pnpm run generate && bunchee",
+    "build": "bunchee",
     "dev": "bunchee --watch",
     "lint": "eslint src/ --ignore-pattern client/*.ts --no-warn-ignored",
     "format": "prettier --write ./src/ tests/",
@@ -116,9 +116,9 @@
     "@eslint/js": "^9.32.0",
     "@hey-api/client-fetch": "^0.10.1",
     "@hey-api/openapi-ts": "^0.67.5",
-    "@llamaindex/core": "^0.6.19",
+    "@llamaindex/core": "^0.6.22",
     "@llamaindex/env": "^0.1.30",
-    "@llamaindex/workflow-core": "^0.4.1",
+    "@llamaindex/workflow-core": "^1.3.3",
     "@types/node": "^20.19.9",
     "@typescript-eslint/eslint-plugin": "^8.38.0",
     "@typescript-eslint/parser": "^8.38.0",
@@ -131,18 +131,19 @@
     "turbo": "^2.5.5",
     "typescript": "^5.8.3",
     "typescript-eslint": "^8.38.0",
-    "vitest": "^2.0.0"
+    "vitest": "^2.0.0",
+    "zod": "^4.1.13"
   },
   "peerDependencies": {
     "@llamaindex/core": "^0.6.19",
     "@llamaindex/env": "^0.1.30",
-    "@llamaindex/workflow-core": "^0.4.1"
+    "@llamaindex/workflow-core": "^1.3.3",
+    "zod": "^3.25.0 || ^4.0.0"
   },
   "dependencies": {
     "ajv": "^8.17.1",
     "file-type": "^21.0.0",
-    "p-retry": "^6.2.1",
-    "zod": "^3.25.76"
+    "p-retry": "^6.2.1"
   },
   "packageManager": "pnpm@10.8.1"
 }

ts/llama_cloud_services/src/beta/agent/index.ts

@@ -2,11 +2,14 @@ export { AgentClient, createAgentDataClient } from "./client";
 
 export type {
   AggregateAgentDataOptions,
+  BoundingBox,
   ComparisonOperator,
   ExtractedData,
   ExtractedFieldMetadata,
   ExtractedFieldMetadataDict,
+  FieldCitation,
   FilterOperation,
+  PageDimensions,
   SearchAgentDataOptions,
   StatusType,
   TypedAgentData,

ts/llama_cloud_services/src/beta/agent/types.ts

@@ -28,6 +28,44 @@ export type ComparisonOperator =
  */
 export type FilterOperation = RawFilterOperation;
 
+/**
+ * Bounding box coordinates for a citation location on a page
+ */
+export interface BoundingBox {
+  /** X coordinate of the bounding box origin */
+  x: number;
+  /** Y coordinate of the bounding box origin */
+  y: number;
+  /** Width of the bounding box */
+  w: number;
+  /** Height of the bounding box */
+  h: number;
+}
+
+/**
+ * Dimensions of a page in the source document
+ */
+export interface PageDimensions {
+  /** Width of the page */
+  width: number;
+  /** Height of the page */
+  height: number;
+}
+
+/**
+ * Citation information for an extracted field
+ */
+export interface FieldCitation {
+  /** The page number that the field occurred on */
+  page?: number;
+  /** The original text this field's value was derived from */
+  matching_text?: string;
+  /** Bounding boxes indicating where the citation appears on the page */
+  bounding_boxes?: BoundingBox[];
+  /** Dimensions of the page containing the citation */
+  page_dimensions?: PageDimensions;
+}
+
 /**
  * Metadata for an extracted field, including confidence and citation information
  */
@@ -38,16 +76,11 @@ export interface ExtractedFieldMetadata {
   confidence?: number;
   /** The confidence score for the field based on the extracted text only */
   extraction_confidence?: number;
+  /** The confidence score for the field based on the parsing/OCR quality */
+  parsing_confidence?: number;
   citation?: FieldCitation[];
 }
 
-export interface FieldCitation {
-  /** The page number that the field occurred on */
-  page?: number;
-  /** The original text this field's value was derived from */
-  matching_text?: string;
-}
-
 /**
  * Dictionary mapping field names to their metadata
  * Values can be ExtractedFieldMetadata objects, nested dictionaries, or arrays

ts/llama_cloud_services/src/classify.ts

@@ -108,20 +108,19 @@ async function pollForJobCompletion({
     }
     const response =
       await getClassifyJobApiV1ClassifierJobsClassifyJobIdGet(jobOptions);
-    if (!response.response.ok) {
-      numIterations++;
-    }
     if (typeof response.data != "undefined") {
       status = response.data.status as StatusEnum;
       if (status == StatusEnum.CANCELLED || status == StatusEnum.ERROR) {
-        throw new Error("There was an error during the classification job.");
-      } else if (status == StatusEnum.SUCCESS) {
+        throw new Error("There was an error extracting data from your file.");
+      } else if (
+        status == StatusEnum.SUCCESS ||
+        status == StatusEnum.PARTIAL_SUCCESS
+      ) {
         return true;
-      } else {
-        numIterations++;
-        await sleep(interval * 1000);
       }
     }
+    numIterations++;
+    await sleep(interval * 1000);
   }
 }
 
@@ -169,7 +168,7 @@ async function getJobResult({
       retries++;
       await sleep(retryInterval * 1000);
     }
-    if (typeof response.data != "undefined") {
+    if (response.response.ok && typeof response.data != "undefined") {
       return response.data as ClassifyJobResults;
     } else {
       throw new Error(

ts/llama_cloud_services/src/client/zod.gen.ts

@@ -1,6 +1,6 @@
 // This file is auto-generated by @hey-api/openapi-ts
 
-import { z } from "zod";
+import { z } from "zod/v4";
 
 export const zNoneSegmentationConfig = z.object({
   mode: z.literal("none").optional().default("none"),
@@ -770,7 +770,7 @@ export const zMetadataFilter = z.object({
 
 export const zFilterCondition: z.ZodTypeAny = z.enum(["and", "or", "not"]);
 
-export const zMetadataFilters: z.AnyZodObject = z.object({
+export const zMetadataFilters: z.ZodObject<z.ZodRawShape> = z.object({
   filters: z.array(z.unknown()),
   condition: z.union([zFilterCondition, z.null()]).optional(),
 });
@@ -782,7 +782,7 @@ export const zRetrievalMode: z.ZodTypeAny = z.enum([
   "auto_routed",
 ]);
 
-export const zPresetRetrievalParams: z.AnyZodObject = z.object({
+export const zPresetRetrievalParams: z.ZodObject<z.ZodRawShape> = z.object({
   dense_similarity_top_k: z
     .union([z.number().int().gte(1).lte(100), z.null()])
     .optional(),
@@ -825,7 +825,7 @@ export const zSupportedLlmModelNames: z.ZodTypeAny = z.enum([
   "VERTEX_AI_CLAUDE_3_5_SONNET_V2",
 ]);
 
-export const zLlmParameters: z.AnyZodObject = z.object({
+export const zLlmParameters: z.ZodObject<z.ZodRawShape> = z.object({
   model_name: zSupportedLlmModelNames.optional(),
   system_prompt: z.union([z.string().max(3000), z.null()]).optional(),
   temperature: z.union([z.number(), z.null()]).optional(),
@@ -834,7 +834,7 @@ export const zLlmParameters: z.AnyZodObject = z.object({
   class_name: z.string().optional().default("base_component"),
 });
 
-export const zChatData: z.AnyZodObject = z.object({
+export const zChatData: z.ZodObject<z.ZodRawShape> = z.object({
   retrieval_parameters: zPresetRetrievalParams.optional(),
   llm_parameters: z.union([zLlmParameters, z.null()]).optional(),
   class_name: z.string().optional().default("base_component"),
@@ -2141,7 +2141,7 @@ export const zTextItem = z.object({
   value: z.string(),
 });
 
-export const zListItem: z.AnyZodObject = z.object({
+export const zListItem: z.ZodObject<z.ZodRawShape> = z.object({
   type: z.literal("list").optional().default("list"),
   bBox: z.union([z.unknown(), z.null()]).optional(),
   items: z.array(z.unknown()),

ts/llama_cloud_services/src/events.ts

@@ -1,6 +1,6 @@
 import { workflowEvent } from "@llamaindex/workflow-core";
 import { zodEvent } from "@llamaindex/workflow-core/util/zod";
-import { z } from "zod";
+import { z } from "zod/v4";
 import { parseFormSchema } from "./schema";
 
 export const uploadEvent = zodEvent(

ts/llama_cloud_services/src/extract.ts

@@ -296,20 +296,16 @@ async function pollForJobCompletion(
       return false;
     }
     const response = await getJobApiV1ExtractionJobsJobIdGet(jobOptions);
-    if (!response.response.ok) {
-      numIterations++;
-    }
     if (typeof response.data != "undefined") {
       status = response.data.status as StatusEnum;
       if (status == StatusEnum.CANCELLED || status == StatusEnum.ERROR) {
         throw new Error("There was an error extracting data from your file.");
       } else if (status == StatusEnum.SUCCESS) {
         return true;
-      } else {
-        numIterations++;
-        await sleep(interval * 1000);
       }
     }
+    numIterations++;
+    await sleep(interval * 1000);
   }
 }
 
@@ -350,7 +346,7 @@ async function getJobResult(
       retries++;
       await sleep(retryInterval * 1000);
     }
-    if (typeof response.data != "undefined") {
+    if (response.response.ok && typeof response.data != "undefined") {
       return {
         data: response.data.data,
         extractionMetadata: response.data.extraction_metadata,

ts/llama_cloud_services/src/query-tool.ts

@@ -2,7 +2,7 @@ import type { JSONValue } from "@llamaindex/core/global";
 import type { ToolMetadata } from "@llamaindex/core/llms";
 import type { BaseQueryEngine } from "@llamaindex/core/query-engine";
 import { tool } from "@llamaindex/core/tools";
-import { z } from "zod";
+import { z } from "zod/v4";
 
 const DEFAULT_NAME = "llama_cloud_index_tool";
 const DEFAULT_DESCRIPTION =
@@ -21,9 +21,7 @@ export function createQueryEngineTool(
     name: metadata?.name ?? DEFAULT_NAME,
     description: metadata?.description ?? DEFAULT_DESCRIPTION,
     parameters: z.object({
-      query: z.string({
-        description: "The query to search for",
-      }),
+      query: z.string().describe("The query to search for"),
     }),
     execute: async ({ query }) => {
       const response = await queryEngine.query({ query });

ts/llama_cloud_services/src/reader.ts

@@ -1,9 +1,10 @@
 /* eslint-disable @typescript-eslint/no-explicit-any */
 import { type Client, createClient, createConfig } from "@hey-api/client-fetch";
+import { type FailedAttemptError } from "p-retry";
 import { Document, FileReader } from "@llamaindex/core/schema";
 import { fs, getEnv, path } from "@llamaindex/env";
 import {
-  type BodyUploadFileApiParsingUploadPost,
+  type BodyUploadFileApiV1ParsingUploadPost,
   type FailPageMode,
   type ParserLanguages,
   type ParsingMode,
@@ -32,6 +33,33 @@ type WriteStream = {
 // eslint-disable-next-line no-var
 var process: any;
 
+function handleFailedAttempt(
+  error: FailedAttemptError,
+  jobId: string,
+  verbose: boolean,
+) {
+  // Retry only on 5XX or socket errors.
+  const status = (error.cause as any)?.response?.status;
+  if (
+    !(
+      (status && status >= 500 && status < 600) ||
+      ((error.cause as any)?.code &&
+        ((error.cause as any).code === "ECONNRESET" ||
+          (error.cause as any).code === "ETIMEDOUT" ||
+          (error.cause as any).code === "ECONNREFUSED")) ||
+      (status && status === 404)
+    )
+  ) {
+    throw error;
+  }
+
+  if (verbose) {
+    console.warn(
+      `Attempting to get job ${jobId} result (attempt ${error.attemptNumber}) failed. Retrying...`,
+    );
+  }
+}
+
 /**
  * Represents a reader for parsing files using the LlamaParse API.
  * See https://github.com/run-llama/llama_parse
@@ -188,6 +216,16 @@ export class LlamaParseReader extends FileReader {
   extract_printed_page_number?: boolean | undefined;
   tier?: string | undefined;
   version?: string | undefined;
+  layout_aware?: boolean | undefined;
+  line_level_bounding_box?: boolean | undefined;
+  specialized_image_parsing?: boolean | undefined;
+  aggressive_table_extraction?: boolean | undefined;
+  preserve_very_small_text?: boolean | undefined;
+  spreadsheet_force_formula_computation?: boolean | undefined;
+  inline_images_in_markdown?: boolean | undefined;
+  keep_page_separator_when_merging_tables?: boolean | undefined;
+  remove_hidden_text?: boolean | undefined;
+  presentation_out_of_bounds_content?: boolean | undefined;
 
   constructor(
     params: Partial<Omit<LlamaParseReader, "language" | "apiKey">> & {
@@ -387,11 +425,25 @@ export class LlamaParseReader extends FileReader {
       extract_printed_page_number: this.extract_printed_page_number,
       tier: this.tier,
       version: this.version,
+      layout_aware: this.layout_aware,
+      line_level_bounding_box: this.line_level_bounding_box,
+      specialized_image_parsing: this.specialized_image_parsing,
+      aggressive_table_extraction: this.aggressive_table_extraction,
+      preserve_very_small_text: this.preserve_very_small_text,
+      spreadsheet_force_formula_computation:
+        this.spreadsheet_force_formula_computation,
+      inline_images_in_markdown: this.inline_images_in_markdown,
+      webhook_configurations: undefined,
+      keep_page_separator_when_merging_tables:
+        this.keep_page_separator_when_merging_tables,
+      remove_hidden_text: this.remove_hidden_text,
+      presentation_out_of_bounds_content:
+        this.presentation_out_of_bounds_content,
     } satisfies {
-      [Key in keyof BodyUploadFileApiParsingUploadPost]-?:
-        | BodyUploadFileApiParsingUploadPost[Key]
+      [Key in keyof BodyUploadFileApiV1ParsingUploadPost]-?:
+        | BodyUploadFileApiV1ParsingUploadPost[Key]
         | undefined;
-    } as unknown as BodyUploadFileApiParsingUploadPost;
+    } as unknown as BodyUploadFileApiV1ParsingUploadPost;
 
     const response = await uploadFileApiV1ParsingUploadPost({
       client: this.#client,
@@ -443,26 +495,8 @@ export class LlamaParseReader extends FileReader {
             }),
           {
             retries: this.maxErrorCount,
-            onFailedAttempt: (error) => {
-              // Retry only on 5XX or socket errors.
-              const status = (error.cause as any)?.response?.status;
-              if (
-                !(
-                  (status && status >= 500 && status < 600) ||
-                  ((error.cause as any)?.code &&
-                    ((error.cause as any).code === "ECONNRESET" ||
-                      (error.cause as any).code === "ETIMEDOUT" ||
-                      (error.cause as any).code === "ECONNREFUSED"))
-                )
-              ) {
-                throw error;
-              }
-              if (this.verbose) {
-                console.warn(
-                  `Attempting to get job ${jobId} result (attempt ${error.attemptNumber}) failed. Retrying...`,
-                );
-              }
-            },
+            onFailedAttempt: (error) =>
+              handleFailedAttempt(error, jobId, this.verbose),
           },
         );
       } catch (e: any) {
@@ -475,49 +509,69 @@ export class LlamaParseReader extends FileReader {
       const status = (data as Record<string, unknown>)["status"];
 
       if (status === "SUCCESS") {
-        let resultData;
         switch (resultType) {
           case "json": {
-            resultData =
-              await getJobJsonResultApiV1ParsingJobJobIdResultJsonGet({
-                client: this.#client,
-                throwOnError: true,
-                path: { job_id: jobId },
-                query: {
-                  organization_id: this.organization_id ?? null,
-                },
-                signal: AbortSignal.timeout(this.maxTimeout * 1000),
-              });
-            break;
+            const resultData = await pRetry(
+              () =>
+                getJobJsonResultApiV1ParsingJobJobIdResultJsonGet({
+                  client: this.#client,
+                  throwOnError: true,
+                  path: { job_id: jobId },
+                  query: {
+                    organization_id: this.organization_id ?? null,
+                  },
+                  signal: AbortSignal.timeout(this.maxTimeout * 1000),
+                }),
+              {
+                retries: this.maxErrorCount,
+                onFailedAttempt: (error) =>
+                  handleFailedAttempt(error, jobId, this.verbose),
+              },
+            );
+            return resultData.data;
           }
           case "markdown": {
-            resultData =
-              await getJobResultApiV1ParsingJobJobIdResultMarkdownGet({
-                client: this.#client,
-                throwOnError: true,
-                path: { job_id: jobId },
-                query: {
-                  organization_id: this.organization_id ?? null,
-                },
-                signal: AbortSignal.timeout(this.maxTimeout * 1000),
-              });
-            break;
+            const resultData = await pRetry(
+              () =>
+                getJobResultApiV1ParsingJobJobIdResultMarkdownGet({
+                  client: this.#client,
+                  throwOnError: true,
+                  path: { job_id: jobId },
+                  query: {
+                    organization_id: this.organization_id ?? null,
+                  },
+                  signal: AbortSignal.timeout(this.maxTimeout * 1000),
+                }),
+              {
+                retries: this.maxErrorCount,
+                onFailedAttempt: (error) =>
+                  handleFailedAttempt(error, jobId, this.verbose),
+              },
+            );
+            return resultData.data;
           }
           case "text": {
-            resultData =
-              await getJobTextResultApiV1ParsingJobJobIdResultTextGet({
-                client: this.#client,
-                throwOnError: true,
-                path: { job_id: jobId },
-                query: {
-                  organization_id: this.organization_id ?? null,
-                },
-                signal: AbortSignal.timeout(this.maxTimeout * 1000),
-              });
-            break;
+            const resultData = await pRetry(
+              () =>
+                getJobTextResultApiV1ParsingJobJobIdResultTextGet({
+                  client: this.#client,
+                  throwOnError: true,
+                  path: { job_id: jobId },
+                  query: {
+                    organization_id: this.organization_id ?? null,
+                  },
+                  signal: AbortSignal.timeout(this.maxTimeout * 1000),
+                }),
+              {
+                retries: this.maxErrorCount,
+                onFailedAttempt: (error) =>
+                  handleFailedAttempt(error, jobId, this.verbose),
+              },
+            );
+
+            return resultData.data;
           }
         }
-        return resultData.data;
       } else if (status === "PENDING") {
         if (this.verbose && tries % 10 === 0) {
           this.stdout?.write(".");

ts/llama_cloud_services/src/schema.ts

@@ -1,6 +1,6 @@
 import { FailPageMode, ParserLanguages, ParsingMode } from "./client";
 
-import { z } from "zod";
+import { z } from "zod/v4";
 
 type Language = ParserLanguages;
 const VALUES: [Language, ...Language[]] = [
@@ -52,9 +52,10 @@ export const parseFormSchema = z.object({
   html_remove_navigation_elements: z.boolean().optional(),
   http_proxy: z
     .string()
-    .url(
-      'Set a valid URL for the HTTP proxy, e.g., "http://proxy.example.com:8080"',
-    )
+    .url({
+      error:
+        'Set a valid URL for the HTTP proxy, e.g., "http://proxy.example.com:8080"',
+    })
     .refine(
       (url) => {
         try {
@@ -67,7 +68,7 @@ export const parseFormSchema = z.object({
         }
       },
       {
-        message: "Invalid HTTP proxy URL",
+        error: "Invalid HTTP proxy URL",
       },
     )
     .optional(),
@@ -100,7 +101,7 @@ export const parseFormSchema = z.object({
   vendor_multimodal_model_name: z.string().optional(),
   model: z.string().optional(),
   webhook_url: z.string().url().optional(),
-  parse_mode: z.nativeEnum(ParsingMode).nullable().optional(),
+  parse_mode: z.enum(ParsingMode).nullable().optional(),
   system_prompt: z.string().optional(),
   system_prompt_append: z.string().optional(),
   user_prompt: z.string().optional(),
@@ -129,7 +130,7 @@ export const parseFormSchema = z.object({
   compact_markdown_table: z.boolean().optional(),
   markdown_table_multiline_header_separator: z.string().optional(),
   page_error_tolerance: z.number().min(0).max(1).optional(),
-  replace_failed_page_mode: z.nativeEnum(FailPageMode).nullable().optional(),
+  replace_failed_page_mode: z.enum(FailPageMode).nullable().optional(),
   replace_failed_page_with_error_message_prefix: z.string().optional(),
   replace_failed_page_with_error_message_suffix: z.string().optional(),
 });