v0.3.0

Breaking Changes

• SegmentSpec.entity_id is now required. Add entity_id: <dim_pk_column> to every segment spec YAML. Specs without it raise SpecValidationError at load time.
• SegmentSpec.source is now the DIM table URI. Previously parsed but ignored; it is now used to generate a JOIN in the SQL query. Ensure it points to the DIM table, not the fact table.
• segments in compute() no longer accepts list[str]. The parameter type changed from str | list[str] | None to dict[str, str] | str | None. Only one segment per compute() call is supported; call compute() once per segment.

Added

• SegmentSpec.entity_id — required field identifying the primary key column on the DIM table.
• SegmentSpec.join_keys — optional whitelist of fact-table FK columns accepted as join keys. When non-empty, the key supplied at compute() time must appear in this list.
• DIM-table JOIN in generated SQL — segments now generate FROM fact t JOIN dim _dim ON t.<join_key> = _dim.<entity_id>. Unqualified column references in values[].where are automatically qualified with _dim. via sqlglot AST rewriting.
• segments dict form — segments={"customer_value": "buyer_id"} pins the fact-side FK column explicitly, enabling the same segment spec to join via different columns (e.g. buyer_id vs seller_id).
• referenced_columns for segment specs — ValidationResult.referenced_columns now includes entity_id, join_keys (when non-empty), and values[i].where keys.
• __version__ synced — aitaem.__version__ was stale at "0.2.1"; now correctly reflects the installed version.

Documentation

• User guide: rewritten SegmentSpec section with DIM-join model, example YAML, and field table.
• Computing metrics guide: updated segments parameter docs and error table.
• API reference: added segment spec keys to ValidationResult.referenced_columns.
• Changelog: this entry.

Full changelog: https://chaturv3di.github.io/aitaem/changelog/