Skip to content

v1.2.49

Choose a tag to compare

@suguanYang suguanYang released this 09 Nov 10:52
ebc0076

Changes

  • Recover text during progressive PDF ingestion by running OCR on pages whose native text is detected as corrupt. Recovered text is available to search and later enrichment. See #523.
  • The initial parse uses a bounded OCR budget. Pages that exceed the budget retain their page image and may lack searchable text. Existing parse snapshots require reprocessing to pick up the change.