You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Refine index configuration validation and typed helpers (#173)
## Summary
- add typed HNSW/fulltext configuration helpers and export them from
package entry points
- enforce stricter parameter validation and normalization for
dense/sparse/fulltext index configs while preserving forward-compatible
pass-through for unknown analyzers/properties
- update vector index SQL generation to honor configured `type`/`lib`
and extend unit tests for new defaults and validation behavior
## Test plan
- [x] `PYTHONPATH=src pytest tests/unit_tests/test_configuration.py
tests/unit_tests/test_sparse_vector_index_config.py`
- [ ] Run full unit/integration suite in CI
Made with [Cursor](https://cursor.com)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Added specialized fulltext configs (IK, SPACE, BENG, NGRAM, NGRAM2)
and BQ/SQ variants for HNSW vector indexes.
* **Improvements**
* Stronger validation, normalization and safer defaults for index and
sparse-vector configs (updated prune/refine/drop/refine_k defaults).
* HNSW config now exposes configurable type/lib and uses those values in
index generation.
* **Breaking Changes**
* Top-level export for the embedded client removed — import the client
from the client module instead.
* **Chores**
* Excluded a specific numpy release in dependency constraints.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
0 commit comments