Alan-Jowett / CoPilot-For-Consensus / 20177266589

82%

push

github

Implement Qdrant vector store adapter with UUID-based point IDs (#117)

* Initial plan

* Implement Qdrant vector store adapter with tests and documentation

Co-authored-by: Alan-Jowett <20480683+Alan-Jowett@users.noreply.github.com>

* Fix Qdrant unit tests - use correct mock path for QdrantClient

Co-authored-by: Alan-Jowett <20480683+Alan-Jowett@users.noreply.github.com>

* Fix Qdrant implementation - use UUID for point IDs and query_points API

Co-authored-by: Alan-Jowett <20480683+Alan-Jowett@users.noreply.github.com>

* Fix Qdrant unit test - update mock for query_points API

Co-authored-by: Alan-Jowett <20480683+Alan-Jowett@users.noreply.github.com>

* Improve exception handling in Qdrant store - remove fragile string matching

- Use explicit exception type checking instead of string matching
- Re-raise ValueError for duplicate detection
- Silently proceed on retrieve errors (allows upsert to work)
- More robust and maintainable error handling

Co-authored-by: Alan-Jowett <20480683+Alan-Jowett@users.noreply.github.com>

* Code quality improvements - move imports and simplify collection creation

- Move PointIdsList import to __init__ with other imports
- Remove try/except from _ensure_collection (simpler and clearer)
- All exception handling now uses explicit types, no string matching
- Maintains same functionality with cleaner code

Co-authored-by: Alan-Jowett <20480683+Alan-Jowett@users.noreply.github.com>

* Address PR review comments - improve error messages and code quality

- Show original string IDs in duplicate error messages instead of UUIDs
- Copy payload before mutation to avoid modifying Qdrant's cached objects
- Add better error handling for invalid environment variable values (port, batch_size)
- Remove unused imports: MagicMock, QdrantVectorStore in test files
- Remove unused variables in tests
- Rename test from test_qdrant_backend_not_implemented to test_qdrant_backend_requires_connection
- Fix ImportError test to only run when... (continued)

100 of 144 new or added lines in 3 files covered. (69.44%)

2384 of 3147 relevant lines covered (75.75%)

0.76 hits per line