Does anyone have any recommendation or procedures for repairing a corrupt PDF? When I open the file I get "There was an error opening this document. the file is damaged and cannot be repaired." There seems to be a myriad of tools out there but none that I could describe as reputable. Are there any opensource linux based solutions for this possibly?

  • Opensource PDF tools tend to be pretty crappy, I'm afraid. What are you using? – Satanicpuppy May 3 '11 at 14:38
  • Also see: superuser.com/questions/166999/… – slhck May 3 '11 at 14:39
  • didnt like the look of any of the tools as they looked like the myriad of "Registry Cleaners" out there that are useless. Have been trying Adobe Pro and have just started looking if Ghostscript or PDFForge have any repair switches. – Tim Alexander May 3 '11 at 14:48
  • Ghostscript is okay, but it's certainly not better than Acrobat. It's completely bare bones. – Satanicpuppy May 3 '11 at 18:41
  • 8
    @Satanicpuppy I disagree :: I use ghostscript to rebuild damaged or low-quality pdfs quite often and it performs very well. – Edward J Beckett Feb 5 '13 at 20:16

Ghostscript will repair your corrupted PDF automatically... if it can open it in the first place (that is, if it is not damaged beyond repair). But afterwards you'll still need to double-check the result...

On Linux, try this command:

 gs \
  -o repaired.pdf \
  -sDEVICE=pdfwrite \
  -dPDFSETTINGS=/prepress \

On Windows, try this one:

 gswin32c.exe ^
  -o repaired.pdf ^
  -sDEVICE=pdfwrite ^
  -dPDFSETTINGS=/prepress ^
  • 4
    Ghostscript does a fantastic job of rendering pdfs ... I regularly use gs to rebuild pdfs to improve font quality. – Edward J Beckett Feb 5 '13 at 20:14
  • 2
    The /prepress make the quality really good compared to /screen. Thanks. – Dolanor Sep 13 '15 at 22:17
  • I get "An error occurred while reading an XREF table." What does that mean? – Geremia Jun 18 '19 at 15:26
  • It means the internal table of contents (what PDFs have to contain as XREF table) had an error, pointing to a wrong byte offset for a PDF object. Ghostscript very likely repaired that error and inserted a correct XREF table into the output. You can check this by running the output through Ghostscript one more time and see if this message still appears. – Kurt Pfeifle Jun 18 '19 at 18:13
  • The quality of the book cover (which is a color image) got much worse, any idea why? – Yan King Yin Feb 12 at 8:53

I had a corrupted PDF file, print.pdf , that Ghostscript couldn't open, but the usual graphical Linux PDF viewers (Okular, Evince) opened fine. (In my case, the file had garbage at the start instead of a PDF header, when opened in a hex editor.)

These PDF viewers use Poppler as a back-end PDF renderer. So you can repair the PDF using Poppler's command-line tools. In Ubuntu these are in the poppler-utils package. I used:

pdftocairo -pdf print.pdf print_repaired.pdf

which generated a PDF file with correct headers, which tools like Ghostscript now accepted.

  • 4
    +1 this read my Quartz generated PDF without complaints, and immediately started generating output. Ghostscript, Adobe Acrobat Pro and others insisted on rebuilding my 120GB pdf first. – Orwellophile Dec 14 '13 at 14:17
  • This didn't work for at least one weird PDF I came across, but it seems like a good start. – Brian Peterson Nov 11 '14 at 20:00
  • 1
    Works perfectly on a PDF on which Ghostscript wanted to remove some arbitrary elements on pages. – Andrea Lazzarotto Nov 22 '14 at 16:14
  • Ghostscript failed to read the document but this worked like a charm. BTW I did this on Windows using the new linux subsystem, so cool! – HyLian Jun 5 '16 at 17:44

mutool (project page, manpage) will repair broken PDFs without printing them.

  • Installation e.g. on Ubuntu: sudo apt-get install mupdf-tools
  • Run it like this: mutool clean input.pdf output.pdf
mutool clean [options] input.pdf [output.pdf] [pages]

  The clean command pretty prints and rewrites the syntax of a PDF file.
   It can be used to repair broken files, expand compressed streams,
   filter out a range of pages, etc.
  If no output file is specified, it will write the cleaned PDF to
   "out.pdf" in the current directory.

Alternatively, there are a few tools and frameworks that can decompose/decompile PDFs into their components without rendering them. These could be useful for extracting text, scripts, and images. See this answer for a list of such tools: https://reverseengineering.stackexchange.com/q/1526/8210. E.g. you can try the current top answer Origami, it has a GTK-based viewer.

  • 5
    This solution works "better" than the solutions offered above or better ranked, as it does not "print" the PDF file and keeps active the links, clickable items, etc... To me, it sounds a more elegant solution than using ghostscript or cairo. – Speredenn Jun 5 '15 at 15:21
  • 1
    Unfortunately, mutool clean doesn't fix all possible errors. I have a file that has various errors in the font and content streams, and mutool will keep those errors. – Dominik Honnef Jun 9 '16 at 20:52
  • 1
    @DominikHonnef You can always try tools/frameworks that decompose the PDF and allow you to view all the parts without rendering them. That should enable you to get text, scripts, images, etc. directly. See this answer for a list of tools: reverseengineering.stackexchange.com/q/1526/8210 – jmiserez Jun 24 '16 at 10:29
  • 1
    Only thing that worked for me! – jamadagni Aug 22 '17 at 14:37
  • 1
    This worked better since this does not render the pdf it examinate the document. – riccs_0x Oct 4 '17 at 0:28

I had a corrupted pdf file, because the php file used to download it echoed some errors (in HTML) and NUL characters at the end.

The solution was to open the pdf with Notepad++ and remove all text after the line

  • had same, Adobe Reader didn't open but native Mac, Chrome and Firefox PDF plugin displayed PDF file fine. Reason was also extra "NUL" at last line added during the upload. – Tilo Apr 8 '14 at 19:23
  • I had a PDF with two %%EOF. I deleted everything after the first %%EOF using a hex editor. Now everything works fine. – adjan Jun 17 '17 at 8:21

Not the answer you're looking for? Browse other questions tagged or ask your own question.