JesusMan0529 opened a new pull request, #377: URL: https://github.com/apache/hugegraph-ai/pull/377
When more than 1,000 new vertices are indexed, embedding batches can finish out of order. `BuildSemanticIndex` currently flattens those results in completion order but pairs them with the original vertex IDs, so semantic search can return the wrong vertex. Collect batch results with `asyncio.gather` to retain input order. Keep the existing batch size, concurrency limit, synchronous embedding provider calls, and progress updates as each batch completes. The regression forces the trailing batch to finish before the first batch and checks real Faiss save/load/search for both PRIMARY_KEY and CUSTOMIZE IDs. It fails on the original code. Empty input and provider error propagation are also covered. Validation (Python 3.11): - `SKIP_EXTERNAL_SERVICES=true uv run pytest hugegraph-llm/src/tests/operators/index_op/ hugegraph-llm/src/tests/indices/ -q --tb=short`: 53 passed, 1 skipped (the existing live Ollama test). - `uv run ruff format --check .`: passed. - `uv run ruff check .`: passed. - `git diff --check`: passed. HugeGraph Server and live LLM/provider integration tests were not run locally. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
