Skip to main content

Document Formats

The search engine accepts several document types and handles each one differently during parsing.

Supported file types

  • PDF (.pdf) – extracted page by page
  • Word (.docx) – treated as a single page
  • PowerPoint (.pptx) – one page per slide (titles, body, text boxes, tables, speaker notes)
  • Excel (.xlsx) – one page per visible worksheet, with the sheet name preserved for traceability
  • Legacy PowerPoint / Excel (.ppt, .xls) – converted to .pptx/.xlsx via LibreOffice when enabled (see below)
  • Plain text (.txt, .text) – treated as a single page
  • Markdown (.md) – treated as a single page
  • Images (.png, .jpg, .jpeg, .tif, .tiff) – processed with OCR

Legacy binary Office formats

.ppt and .xls are the older binary Office formats. They are supported only when OFFICE_LEGACY_CONVERSION_ENABLED=true and a LibreOffice (soffice) binary is available on the worker; the file is converted to its modern OOXML equivalent before parsing. When the feature is disabled or LibreOffice is missing, uploads of these formats fail with a clear error. The modern .pptx and .xlsx formats need no conversion and are always supported.

Limits

  • Maximum file size: 200MB
  • Validation happens before processing starts

Page handling

File TypeHow It ParsesWhat Page Number You Get
PDFExtracted page by pageActual page numbers preserved
DOCXTreated as single pageAlways 1
PPTX / PPTOne page per slideSlide number (1-indexed)
XLSX / XLSOne page per visible worksheetWorksheet ordinal (1-indexed)
TXT / TEXT / MDTreated as single pageAlways 1
ImagesOCR outputAlways 1
PDF page numbers

For PDF documents, the page field in search results reflects the actual page number from the source file. Presentations report the slide number and spreadsheets report the worksheet ordinal. For all other formats, the page is always reported as 1.