plinder.data.docs#
- plinder.data.docs.get_cluster_column_descriptions(plindex: DataFrame) list[tuple[str, str | None, str | None]][source]#
- plinder.data.docs.get_table_column_descriptions(*, table_name: str, schema: Schema) DataFrame[source]#
Build ordered descriptions for exactly the columns in one release table.
The Arrow schema supplies the published names, order, and data types. A missing prose description is an error so documentation cannot silently lag behind a release table.
- plinder.data.docs.get_column_descriptions(table_name: str, *, description_dir: Path = PosixPath('/home/runner/work/plinder/plinder/src/plinder/data/column_descriptions/tables')) DataFrame[source]#
Read the checked-in descriptions for one release parquet table.
- plinder.data.docs.write_column_descriptions(*, data_dir: Path, output_dir: Path = PosixPath('/home/runner/work/plinder/plinder/src/plinder/data/column_descriptions/tables')) None[source]#
Write one complete description TSV per published parquet table.
TODO: the checked-in TSVs are hand-edited; nothing regenerates them or checks their prose against the model descriptions. Discuss: build from the static schemas already in code, add a parity test, run as a pre-commit hook.