docling-bedrock-plugin

AWS Bedrock integration plugin for the Docling framework.

This plugin allows Docling users to leverage the powerful multimodal capabilities of AWS Bedrock models directly within their document processing workflows, enabling features like automated image description generation.

Table of Contents

• Installation
• AWS Configuration
• Usage
  • Examples
  • PictureDescriptionBedrockApiOptions
• Contributing
• Credits

Installation

Ensure you have Python >= 3.10 installed.

1. Install from Source (Development)

This is the recommended method for now, especially for development or using the latest code:

1. Clone the repository:

   # Replace with the actual repository URL when available
   git clone https://github.com/Slayer412/docling-bedrock-plugin.git
   cd docling-bedrock-plugin

2. Install in editable mode:

   pip install -e .

   • This installs the package and its dependencies (docling, boto3).
   • The -e flag means changes you make to the source code are immediately reflected without reinstalling.

2. Install from PyPI (Future)

Once published, you will be able to install directly using pip:

# Coming soon!
# pip install docling-bedrock-plugin

AWS Configuration

This plugin interacts with AWS Bedrock and requires AWS credentials configured for boto3. Ensure your environment is set up correctly.

Common methods include:

• Environment Variables: AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_REGION.
• Shared Credential File: ~/.aws/credentials.
• AWS Config File: ~/.aws/config.
• IAM Role: Attached to your EC2 instance or ECS task.

Refer to the official Boto3 Credentials Documentation for detailed configuration instructions. You must also have access granted to the specific Bedrock models you intend to use within your AWS account.

Usage

This plugin integrates seamlessly with Docling via its entry point mechanism. Once installed, you can configure and use the Bedrock functionalities within your Docling pipelines or scripts.

The primary use case currently implemented is generating descriptions for images within documents.

Examples

See the examples/ directory for practical, ready-to-run Python scripts:

• bedrock_image_description.py: Processes PDF, Word, or PowerPoint documents, extracts images, gets descriptions via Bedrock, and saves annotated documents.
• image_description.py: A simpler script to process a single image file directly with Bedrock and print the description.

📖 View Examples README for detailed usage instructions.

PictureDescriptionBedrockApiOptions Parameters

These options allow you to configure the Bedrock API call for image descriptions. They are typically passed when initializing the PictureDescriptionBedrockApiModel.

Note: Currently, this plugin specifically supports integration with Anthropic's Claude 3 models (e.g., Sonnet, Haiku) available via Bedrock. Support for other multimodal models may be added in the future.

Parameter	Type	Description	Default
model_id	str	Required. AWS Bedrock model ID (must be multimodal).	-
region_name	str	AWS region where Bedrock is available.	Uses boto3 default
profile_name	str	AWS profile name from credentials file.	Uses boto3 default
prompt	str	Prompt guiding the model's image description.	"Describe this image."
max_tokens	int	Max tokens in the generated description.	300
temperature	float	Controls randomness (0.0-1.0). Lower is more deterministic.	0.7
top_p	float	Nucleus sampling probability (0.0-1.0).	0.9
top_k	int	Top-k filtering diversity control.	None
max_workers	int	Max concurrent API calls when processing multiple images.	3
stop_sequences	List[str]	List of strings that will stop generation.	[]

Example Configuration:

from docling_bedrock_plugin.pipeline_options import PictureDescriptionBedrockApiOptions

bedrock_options = PictureDescriptionBedrockApiOptions(
    model_id="anthropic.claude-3-sonnet-20240229-v1:0",
    region_name="us-east-1",                     # Optional: Specify AWS region
    prompt="Provide a detailed accessibility description for this image.",
    max_tokens=400,
    temperature=0.4,
    max_workers=5
)

# ... then use bedrock_options when initializing the model or pipeline step

Contributing

Contributions are welcome! If you'd like to contribute, please follow these general steps:

1. Fork the repository on GitHub.
2. Clone your fork locally (git clone <your-fork-url>).
3. Create a new branch for your feature or fix (git checkout -b feature/your-new-feature).
4. Make your changes and commit them (git commit -am 'Add some feature').
5. Push your changes to your fork (git push origin feature/your-new-feature).
6. Create a Pull Request back to the main repository.

Please ensure your code includes tests where appropriate and follows existing coding style.

Credits

• Author: Shreyash Patel ([email protected])
• Docling: Docling Project on GitHub
• Boto3: AWS SDK for Python
• AWS Bedrock: Amazon Bedrock Documentation