Cleanup

[m1rl0k]

No description provided.

[augmentcode]

🤖 Augment PR Summary

Summary: This PR streamlines graph ingestion/query behavior and removes some expensive default work during indexing.

Changes:

• Neo4j backend: throttles/guards PageRank recomputation behind env flags (off by default) and adds a per-collection run interval with locking.
• Neo4j backend: uses asyncio.get_running_loop() (with a defensive fallback) when executing sync queries in an executor.
• Neo4j backend: ensures auto-backfill checks still run for new collections even if the hint DB was already initialized.
• Neo4j backend: broadens symbol matching in callers/callees/importers queries to include suffix matches for fully-qualified names.
• Ingest pipeline: tracks symbol_path -> point_id and passes caller_point_id into call/import edge extraction so edges can link back to source chunks.
• Smart reindexing: computes point IDs earlier to reuse them consistently across reused/new points and edge extraction.
• Graph backfill: switches to upsert-only during backfill (no per-path edge deletion) to avoid accidental data loss across multiple chunks per file.
• Graph backfill: adds runtime backend detection and adapts edge batches to Neo4j’s GraphEdge format when enabled.
• Context answering: removes redundant answer-cleanup/style helper code to simplify the module.

Technical Notes: PageRank refresh can be enabled via NEO4J_PAGERANK_ON_UPSERT=1, with optional throttling controlled by NEO4J_PAGERANK_ON_UPSERT_MIN_INTERVAL.

_{🤖 Was this summary useful? React with 👍 or 👎}

[augmentcode]

Review completed. 1 suggestions posted.

Comment augment review to trigger a new review at any time.

[augmentcode]

min_interval = float(os.environ.get(...)) will raise ValueError if NEO4J_PAGERANK_ON_UPSERT_MIN_INTERVAL is set to a non-numeric string, which would abort edge upserts; consider guarding the parse and falling back to a default.

_{🤖 Was this useful? React with 👍 or 👎}