# higher-education-cost — pinned to what this project actually uses. # (playwright, geopandas, python-dotenv dropped: no JS pages, no maps, and the # optional BLS_API_KEY is read from the plain environment, not via dotenv.) # --- core data --- pandas==3.0.5 duckdb==1.5.5 pyarrow==25.0.0 # Parquet engine # --- web / file ingestion --- requests==2.32.3 # BLS public API + file downloads beautifulsoup4==4.12.3 # static HTML parsing lxml==5.4.0 # parser backend for bs4 / pandas.read_html # --- data packaging --- openpyxl==3.1.5 # Excel export for the sellable datasets # --- viz (exploration drafts in src/viz.py) --- matplotlib==3.10.3 # imported by src/viz.py (social/web PNGs render via the shared Pillow factory) Pillow==12.3.0 # chart rendering (shared Pillow pipeline) # --- config --- pyyaml==6.0.2 # config.yaml loader # --- notebooks / reproduce pipeline --- nbformat==5.11.1 # scripts/_build_nb_*.py build the notebooks nbconvert==7.17.1 # scripts/reproduce.py executes the pipeline notebooks ipykernel==6.29.5 jupyterlab==4.4.3