requirements.txt

# Core runtime dependencies for the chronology pipeline.
# Developed and tested on Python 3.12.3 (any 3.12.x works); the pinned dependencies require Python 3.11 or newer.
# Install into a virtualenv:  python3 -m venv .venv && .venv/bin/pip install -r requirements.txt

PyMuPDF==1.26.4       # PDF text extraction and page rendering (imported as `fitz`)
requests==2.31.0      # HTTP client for the cloud LLM, Anthropic, and Google Vision REST APIs
python-dotenv==1.2.2  # config.py auto-loads API keys from a local .env (no manual sourcing needed)

# Optional: only needed when LLM_PROVIDER = "ollama" (local LLM backend) in config.py.
ollama==0.6.1

# Charts & evaluation analysis (tools/scientific_report/generate_charts/ and the evaluation scripts).
matplotlib==3.11.0  # renders the thesis chart figures (PNG)
numpy==2.4.6        # numeric arrays used in the chart statistics
pandas==3.0.3       # reads/aggregates the granular evaluation summaries

# Roman-villa maps (tools/scientific_report/generate_maps_roman_villas/). Pulls in shapely + pyproj.
geopandas==1.1.3    # geospatial dataframes for the province boundaries + villa points
shapely==2.1.2      # geometry ops (study-area box / intersection in the inset)
pyproj==3.7.2       # CRS handling (RD New / EPSG:28992)

# Aoristic & Monte Carlo case study (docs/research/aoristic_monte_carlo/).
seaborn==0.13.2     # statistical plotting for the aoristic/Monte Carlo figures

# Build-time only: regenerating the ABR standard-vocabulary maps (tools/build_abr_maps.py) from the
# frozen RDF snapshot. The pipeline runtime reads the generated CSVs and never imports rdflib.
rdflib==7.6.0