Alan-Jowett / CoPilot-For-Consensus / 20258091399

78%

push

github

Embedding idempotency + Qdrant UUID fix (#285)

* Embedding: idempotent processing; update chunk status by Mongo _id; Vectorstore(Qdrant): deterministic UUIDv5 for point IDs; keep original id in payload; duplicate checks by UUID\n\nResolves: #284

Signed-off-by: GitHub Copilot <copilot@github.com>

* Fix test behavior

Signed-off-by: GitHub Copilot <copilot@github.com>

* Add comprehensive UUID conversion and idempotency tests

- Add test_uuid_conversion.py with determinism, uniqueness, and format tests
- Add test for original ID preservation in payload
- Add test for UUID usage in duplicate checks
- Update query test to verify original ID extraction from payload
- All embedding tests pass (20/20)
- All vectorstore tests pass (79/79 unit tests)
- Coverage: embedding 94%, vectorstore 85% for qdrant_store.py

Validates PR #285 embedding idempotency and Qdrant UUID compatibility.

Signed-off-by: GitHub Copilot <copilot@github.com>

* Address PR review: validate MongoDB _id before processing

Implements PR #285 review feedback to prevent inconsistent state:

- Add fail-fast validation that all chunks have MongoDB _id field
- Prevents embeddings being generated without status updates
- Removes conditional filtering that could skip status updates
- Adds test_process_chunks_missing_mongo_id_raises_error
- Updates existing tests with _id fields

Resolves copilot-pull-request-reviewer comment: chunks without
_id were silently filtered after embedding generation, causing
reprocessing and duplicate embeddings.

All 21 tests pass, coverage 95%.

Signed-off-by: GitHub Copilot <copilot@github.com>

* Address PR review: improve test code quality

Implements additional PR #285 review feedback:

- Move uuid import to module level for consistency
- Remove unused pytest import
- Fix chained inequality assertion (uuid1 != uuid2 != uuid3)
  to properly test all combinations
- Clarify test assertion uses MongoDB _id (not chunk_id)
  even though values are equal in test fix... (continued)

12 of 12 new or added lines in 2 files covered. (100.0%)

4091 of 4901 relevant lines covered (83.47%)

0.95 hits per line