extractflow-parser 0.1.0

Parse PDF documents into markdown formatted content using Vision LLMs

ExtractFlow Parse

🚀 Getting Started

Prerequisites

• 🐍 Python >= 3.9
• 🖥️ Ollama (if you want to use local models)
• 🤖 API Key for OpenAI (OpenAI compatible models) or Google Gemini (if you want to use OpenAI or Google Gemini)

Models Support

ExtractFlow supports various vision language models from different providers:

OpenAI Compatible Models

• You can use any model that is compatible with OpenAI API.
• By passing openai_compatible=True, and base_url to the parser, the parser will use the OpenAI API to process the image.

Ollama Models

• llama3.2-vision:11b
• llama3.2-vision:70b
• llava:13b
• llava:34b

OpenAI Models

• gpt-4o
• gpt-4o-mini

Gemini Models

• gemini-1.5-flash
• gemini-2.0-flash-exp
• gemini-1.5-pro

Installation

Install the package using pip (Recommended):

pip install extractflow_parser

Install the optional dependencies for OpenAI or Gemini:

pip install 'extractflow_parser[openai]'

pip install 'extractflow_parser[gemini]'

Setting up Ollama (Optional)

See examples/ollama_setup.md on how to setup Ollama locally.

⌛️ Usage

Basic Example Usage

from extractflow_parser import ExtractFlowParser

# Initialize parser
parser = ExtractFlowParser(
    openai_compatible=True,
    base_url="https://api.fireworks.ai/inference/v1",
    api_key="your_api_key",
    model_name="accounts/fireworks/models/qwen2-vl-72b-instruct",
    temperature=0.2,
    top_p=0.3,
    extraction_complexity=True # Set to True for more detailed extraction
)

# Convert PDF to markdown
pdf_path = "path/to/your/document.pdf"
markdown_pages = parser.convert_pdf(pdf_path)

# Process results
for i, page_content in enumerate(markdown_pages):
    print(f"\n--- Page {i+1} ---\n{page_content}")

PDF Page Configuration

from extractflow_parser import ExtractFlowParser, PDFPageConfig

# Configure PDF processing settings
page_config = PDFPageConfig(
    dpi=400,
    color_space="RGB",
    include_annotations=True,
    preserve_transparency=False
)

# Initialize parser with custom page config
parser = ExtractFlowParser(
    model_name="llama3.2-vision:11b",
    temperature=0.7,
    top_p=0.4,
    extraction_complexity=False,
    page_config=page_config
)

# Convert PDF to markdown
pdf_path = "path/to/your/document.pdf"
markdown_pages = parser.convert_pdf(pdf_path)

OpenAI or Gemini Model Usage

from extractflow_parser import ExtractFlowParser

# Initialize parser with OpenAI model
parser = ExtractFlowParser(
    base_url="https://api.openai.com/v1",  # Optional: Custom OpenAI API endpoint
    model_name="gpt-4o",
    api_key="your-openai-api-key", # Get the OpenAI API key from https://platform.openai.com/api-keys
    temperature=0.7,
    top_p=0.4,
    extraction_complexity=True # Set to True for more detailed extraction
)

# Initialize parser with Google Gemini model
parser = ExtractFlowParser(
    model_name="gemini-1.5-flash",
    api_key="your-gemini-api-key", # Get the Gemini API key from https://aistudio.google.com/app/apikey
    temperature=0.7,
    top_p=0.4,
    extraction_complexity=True # Set to True for more detailed extraction
)

📄 License

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