hydromodpy.results.parquet_schemas#

PyArrow schemas for every per-simulation Parquet view.

One declared pa.Schema per file under simulations/<basename>.parquet/. These schemas are the single source of truth for column names, Arrow types, nullability, primary keys, and field-level metadata (units, descriptions, allowed values). The catalog write path casts incoming records to the matching schema before handing them to hydromodpy.results.parquet_io.write_table_atomic().

Schema-level metadata embeds hmp.schema, hmp.schema_version, hmp.pk and CF Conventions. Per-field metadata documents the physical unit and a short description. PARQUET_SCHEMA_VERSION is bumped to v2 and rejected at read time when a Parquet file ships an older marker.

Module attributes

`PARQUET_SCHEMA_VERSION`	Current Parquet schema generation.
`CF_CONVENTIONS`	CF convention version embedded in schema metadata.
`TIMESERIES_SCHEMA`	Per-simulation timeseries observations and outputs.
`BUDGETS_SCHEMA`	Per-simulation water budget per zone and component.
`MASS_BALANCE_SCHEMA`	Per-simulation global mass balance summary.
`METRICS_SCHEMA`	Per-simulation calibration and validation metrics.
`PROVENANCE_SCHEMA`	Per-simulation provenance records for inputs and derived arrays.
`GEOGRAPHIC_FEATURES_SCHEMA`	Logical schema of the GeoParquet 1.1 vector files.
`VIEW_SCHEMAS`	Map of per-simulation view name to declared `pa.Schema`.

Functions

`check_schema_version`(metadata)	Raise `ParquetSchemaVersionError` if `hmp.schema_version` differs.
`schema_for`(view_name)	Return the declared `pa.Schema` for `view_name`.

Exceptions

ParquetSchemaVersionError

Raised when a Parquet file ships an unsupported hmp.schema_version.