Objects lost with an object stream that would not decode.
Pages the page tree declares (/Count) beyond those found in the file.
Reasons the reader could not do its job fully — each one drives an abstention.
All pages joined with a form feed, the form the extractors parse.
True when the file does not end with %%EOF: the download stopped early.
What was read may be partial, or an earlier revision of an object that a
later, cut-off update replaced.
Content streams refused outright — an encrypted or unsupported filter — plus pages refused because drawing them exceeded the reader's work budget.
Share of character codes no font mapping covered, 0..1.
Pages whose content the reader could not get — named but missing from the file, refused (an undecodable stream, the work budget) — in page order.
True when the document has no text-showing operators at all but draws images (a scan). A document with neither is not image-only: it has nothing to OCR.