mirror of
https://github.com/lancedb/lancedb.git
synced 2026-08-29 09:28:27 +00:00
302b21aa94
## Summary - add an end-to-end Node regression matching LangChain PDFLoader metadata - verify create/query round trips rich nested `loc` and `pdf.info` fields against the currently configured Apache Arrow peer ## Root cause LanceDB v0.14 delegated nested object inference to Apache Arrow. Nested strings were dictionary-encoded with colliding dictionary IDs, so serializing query results as an IPC file failed with a dictionary-replacement error. Current `main` recursively infers nested fields and avoids those invalid dictionaries, but the reported LangChain path had no end-to-end regression coverage. ## Validation - `pnpm build` - `pnpm lint` - `pnpm run docs` - `pnpm test --runInBand` (678 passed, 5 skipped) Fixes #1963 <!-- lance-gatekeeper-fix:v1 agent=bf8d489db7db2e17678b143f9f0a36d2 generation=1 --> --------- Co-authored-by: Gatefixer <313497061+lancedb-gatefixer[bot]@users.noreply.github.com> Co-authored-by: Xuanwo <github@xuanwo.io>