Issues
When Unstructured tries to partition or chunk a document, you get one of the following error types:TooManyPageFailuresExceptionControllerException: For example, an error wrapping partitioning or chunking errors such as code512.
Possible causes
- There are issues with the document’s underlying content or structure.
- Unstructured encountered too many errors while processing individual pages or elements within a specific document, exceeding some internal failure threshold.
- The document is corrupted or malformed. This is especially the case for some complex PDFs.
- The document has a highly unusual underlying structure or has highly unusual content, which the specified partitioning model cannot handle.
- The document is encrypted or password-protected.
- Some underlying issues are causing page-level failures. This could sometimes be related to quotas if a per-page VLM is used.
Possible solutions
- Inspect the document: Examine the specific document. Make sure it opens correctly, does not seem to look unusual, and does not appear to be corrupted or encrypted.
- Test a simpler document: Try processing a known-good, simple document of the same type, to see if the error is document-specific.
- Check quotas (if VLM related): If the underlying errors mention quotas, see Quota, billing, and rate limiting issues.
- Report the problematic document to Unstructured Support: If the issue seems specific to an otherwise seemingly valid document, report it to Unstructured Support. If possible, provide the document for Unstructured to further investigate.