daveappadoo/mauritius-cabinet-parser

# Mauritius Cabinet Parser A Python pipeline for extracting and analysing decisions from Mauritius Cabinet Meeting PDFs. All PDFs are sourced from the official Prime Minister's Office website: **pmo.govmu.org The dataset covers Cabinet Meeting decisions from **10 November 2024 to 24 April 2026**. --- ## Project Structure ``` mauritius-cabinet-parser/ ├── pipeline.py # Orchestrates the core pipeline steps ├── requirements.txt ├── pdfs/ # Cabinet meeting PDFs (input) ├── scripts/ │ ├── extractCabinetDecisions.py # PDF → CSV extraction │ ├── generate_hugo_data.py # CSV → decisions.json │ └── classify.py # Keyword-based field classification ├── outputs/ │ ├── cabinet_decisions.csv # Raw extracted decisions (tabular) │ ├── decisions.json # All decisions as JSON │ └── decisions_classified.json # + keyword classification fields └── logs/ ├── pipeline.log # Pipeline run log └── parse_log_*.log # Per-run PDF extraction logs ``` --- ## Pipeline Run `pipeline.py` from the project root. It detects new PDFs and runs each step in sequence: ```bash python pipeline.py ``` | Step | Script | |------|--------| | Extract decisions from PDFs | `scripts/extractCabinetDecisions.py` | | Generate Hugo JSON | `scripts/generate_hugo_data.py` | | Classify decisions | `scripts/classify.py` | The pipeline tracks which PDFs have been processed in `outputs/.processed_pdfs.json` and only runs when new PDFs are detected. --- ## Output Files ### `outputs/decisions.json` Produced by `generate_hugo_data.py`. Each decision has: - `filename`, `date`, `decision_number`, `total_pages`, `text` ### `outputs/decisions_classified.json` Produced by `classify.py`. Adds: - `fields` — list of topic categories (e.g. `["Economy", "Diplomacy"]`) Available categories: ` …