docviz-python 0.11.0

[!Caution] This project is in active development. The API is subject to change and breaking changes may occur. Package may not work until first stable release (1.0.0).

Overview

Extract content from documents easily with Python.

• Extract from PDFs (other formats are coming soon)
• Streaming extraction for large documents and real-time results
• Process one or many files using batch extraction
• Choose what to extract (tables, text, images, etc.)
• Export results to JSON, CSV, Excel and others
• Simple and flexible API with high configurability

📦 Installation

• Using uv:

  uv add docviz-python
  

  Upgrading from previous version:

  uv pip install docviz-python --upgrade
  

• Using pip:

  pip install docviz-python --upgrade
  

• Directly from source:

  git clone https://github.com/privateai-com/docviz.git
  cd docviz
  pip install -e .
  

Quick Start

Basic Usage

import asyncio
import docviz

async def main():
    # Create a document instance (can be a local file or a URL)
    document = docviz.Document("path/to/your/document.pdf")
    
    # Extract all content asynchronously
    extractions = await document.extract_content()
    
    # Save results (file name without extension, it will be inherited from chosen format)
    extractions.save("results", save_format=docviz.SaveFormat.JSON)

asyncio.run(main())

Synchronous Usage

import docviz

document = docviz.Document("path/to/your/document.pdf")
extractions = document.extract_content_sync()
extractions.save("results", save_format=docviz.SaveFormat.JSON)

Code Examples

Batch Processing

import docviz
from pathlib import Path

# Process all PDF files in a directory
pdf_directory = Path("data/papers/")
output_dir = Path("output/")
output_dir.mkdir(exist_ok=True)

pdfs = pdf_directory.glob("*.pdf")
documents = [docviz.Document(str(pdf)) for pdf in pdfs]
extractions = docviz.batch_extract(documents)

for ext in extractions:
    ext.save(output_dir, save_format=[docviz.SaveFormat.JSON, docviz.SaveFormat.CSV])

Selective Extraction

import docviz

document = docviz.Document("path/to/document.pdf")

# Extract only specific types of content
extractions = document.extract_content(
    includes=[
        docviz.ExtractionType.TABLE,
        docviz.ExtractionType.TEXT,
        docviz.ExtractionType.FIGURE,
        docviz.ExtractionType.EQUATION,
    ]
)

extractions.save("selective_results", save_format=docviz.SaveFormat.JSON)

Custom Configuration

import docviz

document = docviz.Document("path/to/document.pdf")
extractions = document.extract_content(
    extraction_config=docviz.ExtractionConfig(page_limit=30),
    llm_config=LLMConfig(
        model="gpt-4o-mini",
        api_key=os.getenv("OPENAI_API_KEY"),
        base_url="https://api.openai.com/v1",
    )
)
extractions.save("configured_results", save_format=docviz.SaveFormat.JSON)

Streaming Processing

import docviz

document = docviz.Document("path/to/large_document.pdf")

# Process document in pages to save memory
for page_result in document.extract_streaming():
    # Process each page
    page_result.save(f"page_{page_result.page_number}", save_format=docviz.SaveFormat.JSON)

Progress Tracking

import docviz
from tqdm import tqdm

document = docviz.Document("path/to/document.pdf")

# Extract with progress bar
with tqdm(total=document.page_count, desc="Extracting content") as pbar:
    extractions = document.extract_content(progress_callback=pbar.update)

extractions.save("progress_results", save_format=docviz.SaveFormat.JSON)

Docs

Project has a static site with docs and examples on almost all of its functionality. You can find it at GitHub Pages or build it on your own using sphinx locally. All the dependencies are included in pyproject.toml under the docs group.

Examples

• Basic Usage with 3 different approaches: simple, passing url to document, streaming example and custom configuration using OpenAI key.
• Streaming Processing with progress tracking and generator API.
• OpenAI API Example with custom configuration using OpenAI key.

Pipeline Visualization

Original page with chart

Chart region extracted by Page Parser

Gemma3 output

Contributing

Refer to CONTRIBUTING.md for more information.

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.