Extracting Figures from Documents

LlamaCloud provides several API endpoints to help you extract and work with figures (images) from your documents, including charts, tables, and other visual elements. This guide will show you how to use these endpoints effectively.

These figures can be used for a variety of purposes, such as creating visual summaries, generating reports, chatbot responses, and more.

Note

To extract figures from documents, you need to create an index with a file and enable Extract Layout option when creating the index on Parse Settings -> Text and images handling -> Extract Layout. This will allow you to extract and work with figures from your documents.

Available Endpoints

1. List All Figures: /v1/files/{id}/page-figures
2. List Figures on a Specific Page: /v1/files/{id}/page-figures/{page_index}
3. Get a Specific Figure: /v1/files/{id}/page-figures/{page_index}/{figure_name}

How to Use

App setup

Python Sync Client

Install API client package

pip install llama-cloud

Import and configure client

from llama_cloud.client import LlamaCloud

client = LlamaCloud(token='<llama-cloud-api-key>')

Python Async Client

Install API client package

pip install llama-cloud

Import and configure client

from llama_cloud.client import AsyncLlamaCloud

async_client = AsyncLlamaCloud(token='<llama-cloud-api-key>')

TypeScript Client

Install API client package

npm install llama-cloud-services

Import and configure client

import { LlamaCloudApiClient } from 'llama-cloud-services';

const client = new LlamaCloudApiClient({
  token: apiKey
});

1. List All Figures in a Document

To get a list of all figures across all pages in a document:

Python Sync Client

# Get all figures from a document
figures = client.files.list_file_pages_figures(id="your-file-id")

Python Async Client

# Get all figures from a document
figures = await async_client.files.list_file_pages_figures(id="your-file-id")

TypeScript Client

// Get all figures from a document
const figures = await client.files.listFilePagesFigures({
  fileId: "your-file-id"
});

output:

[
  {
    "figure_name": "page_1_figure_1.jpg",
    "file_id": "71370e55-0f32-4977-b347-460735079386",
    "page_index": 1,
    "figure_size": 87724,
    "is_likely_noise": true,
    "confidence": 0.423
  },
  {
    "figure_name": "page_2_figure_1.jpg",
    "file_id": "71370e55-0f32-4977-b347-460735079386",
    "page_index": 2,
    "figure_size": 87724,
    "is_likely_noise": true,
    "confidence": 0.423
  },
]

2. List Figures on a Specific Page

To get figures from a specific page in your document:

Python Sync Client

# Get figures from a specific page
page_figures = client.files.list_file_page_figures(
    id="your-file-id",
    page_index=1  # Page numbers start at 0
)

Python Async Client

# Get figures from a specific page
page_figures = await async_client.files.list_file_page_figures(
    id="your-file-id",
    page_index=1  # Page numbers start at 0
)

TypeScript Client

// Get figures from a specific page
const pageFigures = await client.files.listFilePageFigures({
  fileId: "your-file-id",
  pageIndex: 1  // Page numbers start at 0
});

output:

[
  {
    "figure_name": "page_1_figure_1.jpg",
    "file_id": "71370e55-0f32-4977-b347-460735079386",
    "page_index": 1,
    "figure_size": 87724,
    "is_likely_noise": true,
    "confidence": 0.423
  },
  {
    "figure_name": "page_1_figure_2.jpg",
    "file_id": "71370e55-0f32-4977-b347-460735079386",
    "page_index": 1,
    "figure_size": 47724,
    "is_likely_noise": true,
    "confidence": 0.423
  }
]

3. Get a Specific Figure

To retrieve a specific figure from your document:

Python Sync Client

# Get a specific figure
figure = client.files.get_file_page_figure(
    id="your-file-id",
    page_index=1,
    figure_name="figure1"
)

Python Async Client

# Get a specific figure
figure = await async_client.files.get_file_page_figure(
    id="your-file-id",
    page_index=1,
    figure_name="figure1"
)

TypeScript Client

// Get a specific figure
const figure = await client.files.getFilePageFigure({
  fileId: "your-file-id",
  pageIndex: 1,
  figureName: "figure1"
});

output:

the base64 encoded image