Supported file types

Stored upload

The upload URL and confirm routes (POST /api/collections/:id/documents/upload-url and /confirm) share the same validation rules (documentUploadRejection) in the browser and on the server, so a file accepted before upload is never rejected afterwards for its name or size:

• Formats: pdf, docx (including .doc), txt (.txt, .md, .text, or extension-less text), image (.png, .jpg, .jpeg, .gif, .webp), csv, xlsx (including .xls), and other (reserved for archive packages that unpack into supported files). Any other unknown extension is rejected before upload (unsupportedType) rather than read as plain text.
• Limits: Filename length 1–500 characters (nameTooLong), positive byte size (empty), and maximum size 500 MiB (500 * 1024 * 1024 bytes, tooLarge).

Text extraction

Discovery and extraction read text for txt, pdf, docx, csv, and xlsx. Images can be stored and viewed, and other binaries are catalogued with unsupported_format:<format>. A PDF with no usable text is catalogued with no_extractable_text. Password-protected files are not unlocked by Bundata.

For Hebrew PDFs where layout elements were extracted in visual (left-to-right) order, PDF extraction and section-map builds (document-topology/6) automatically repair Hebrew lines, numbers, dates, clause codes, and Latin runs into logical reading order.

Non-retaining

The separate non-retaining path accepts only txt, csv, and docx, with a 1,000,000 byte file limit. It does not accept PDF or images.

Removed claims

This site no longer says Bundata supports 65 or more file types. Email containers, presentations, and HTML are not in the upload format enum.