docitup 0.1.1

docitup is a Python package designed to simplify document processing for LangChain. It provides various loaders to extract content from different file types and convert them into LangChain-compatible document classes, ready for storage in LangChain-supported vector stores.

Docitup

This package provides various document loaders that utilize different methods for processing and chunking documents. It is designed to facilitate the loading of documents in various formats into a structured format suitable for using them with langchain vector databases

Overview

The package includes the following loaders:

• PyMUPdf4LLMLoader: Loads and splits documents from files using the pymupdf4llm library.
• MarkitdownLoader: Loads documents using the MarkItDown library.
• LlamaparseLoader: Loads documents using the LlamaParse library and processes different file types.
• DoclingPDFLoader: Converts documents to text and splits them accordingly.

Installation

To install this package, simply run:

pip install docitup 

Usage

PyMUPdf4LLMLoader

from docitup import PyMUPdf4LLMLoader 
  
loader = PyMUPdf4LLMLoader(file_path='path/to/your/file.pdf')  
documents = loader.load()   

MarkitdownLoader

from docitup import MarkitDownLoader
  
loader = MarkitdownLoader(file_path='path/to/your/file.md')  
documents = loader.load()  

LlamaparseLoader

from docitup import LlamaparseLoader
from llama_parse.utils import ResultType
  
loader = LlamaparseLoader(file_path='path/to/your/directory', result_type=ResultType.MD, api_key='your_api_key')  
documents = loader.load()  

DoclingPDFLoader

from docitup import DoclingLoader
  
loader = DoclingLoader(file_path='path/to/your/file.pdf')  
documents = loader.load()

Configuration Options

Each loader can be configured with the following optional parameters:

splitter_type: The type of text splitter to use ("recursive" or other).

chunk_size: The size of each chunk (default is 1000).

chunk_overlap: The number of overlapping characters between chunks (default is 100).

Contributing

Contributions are welcome! Please feel free to submit issues or pull requests for improvements or bug fixes.

License

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

Acknowledgements

This package is made possible by the following libraries:

• pymupdf4llm
• MarkItDown
• LlamaParse
• Docling