Lexlynx - OpenAI Case Law Summarization 📚⚖️

Lexlynx is an OpenAI wrapper designed to summarize case law. It uses the fitz library to read PDF files, OpenAI's API to generate summaries, and tqdm for a progress bar during processing. Lexlynx makes legal case summarization easy and efficient! 🚀

Features ✨

• PDF Text Extraction: Extracts and summarizes case law from PDF files 📄.
• Chunking: Breaks large PDFs into manageable chunks for processing 📊.
• AI Summarization: Uses OpenAI's GPT-4 model to generate concise summaries, highlighting key legal issues, judgments, and precedents 🧠💡.
• Progress Bar: Displays a progress bar during the summarization process using tqdm ⏳.

Requirements 🛠️

pip install -r requirements.txt

This installs PyMuPDF (pymupdf), the OpenAI client and tqdm. Don't pip install fitz: that is an unrelated package that breaks PyMuPDF.

Usage 🏃‍♂️💨

With OpenAI (set your key in the environment, never in the code):

export OPENAI_API_KEY=sk-...        # PowerShell: $env:OPENAI_API_KEY = "sk-..."
python lexlynx.py judgment.pdf
python lexlynx.py judgment.pdf --model gpt-4o --output summary.md

Free and local, with Ollama or any other OpenAI-compatible server:

python lexlynx.py judgment.pdf --base-url http://localhost:11434/v1 --model qwen3:14b

Option	Default
--model	$LEXLYNX_MODEL, else gpt-4o-mini	Chat model
--base-url	$OPENAI_BASE_URL	OpenAI-compatible endpoint
--output		Also save the summary to a file
--chunk-size	8000	Characters per chunk

Long judgments are split at paragraph breaks, each chunk is summarised, and the chunk summaries are combined into one summary under the headings Case, Facts, Issues, Holding, Reasoning and Precedents.

Note ⚠️: a language model can misstate details (in testing, one summary reversed which party paid costs). Check every summary against the judgment before relying on it.