KeyError '/Pages' is raised when pages are merged

[euter]

when I merge two pdf files with pdf version 1.3, the follow KeyError is raised.

Traceback (most recent call last):
  File "pdf_merge.py", line 56, in <module>
    merger.merge()
  File "pdf_merge.py", line 29, in merge
    self._merger.append(open(temp, "rb"))
  File "...\Python35\lib\site-packages\PyPDF2\merger.py", line 203, in append
    self.merge(len(self.pages), fileobj, bookmark, pages, import_bookmarks)
  File "...\Python35\lib\site-packages\PyPDF2\merger.py", line 139, in merge
    pages = (0, pdfr.getNumPages())
  File "...\Python35\lib\site-packages\PyPDF2\pdf.py", line 1155, in getNumPages
    self._flatten()
  File "...\Python35\lib\site-packages\PyPDF2\pdf.py", line 1506, in _flatten
    pages = catalog["/Pages"].getObject()
  File "...\Python35\lib\site-packages\PyPDF2\generic.py", line 516, in __getitem__
    return dict.__getitem__(self, key).getObject()
KeyError: '/Pages'

[b0g3r]

same error @euter any resolution?

[b0g3r]

Okey, my bad, i just called merger.append() after merger.write()

[joaopedroguimaraes]

Same error here! I could'nt find anything about this. Any solution?

[MartinThoma]

As it's been a long time and I don't have neither an example PDF nor example code, I'll close this. If anybody still runs into this issue with the latest PyPDF2 version, please let me know!

[red-shift]

I'm having the same issue. I have a copy of the PDF that's causing the issue.

UTA_OSHA_3115_Fall_Protection_Training_09162021_.pdf

For some context. I'm using requests to retrieve a PDF from a url, then I copy the the response content into a buffer and from there I attempt to merger the pdf using merger.append. Previously I was saving the file to a temporary location, later I refactored it. This works for 99% of the PDFs I'm merging, but there are 4 PDFs I'm working with that consistently trigger this error.

[MartinThoma]

Thanks for adding an example @red-shift !

MCVE: Code + PDF

UTA_OSHA_3115_Fall_Protection_Training_09162021_.pdf

from PyPDF2 import PdfReader

reader = PdfReader("UTA_OSHA_3115_Fall_Protection_Training_09162021_.pdf")

print(len(reader.pages))

[pubpub-zz]

Problem analyzed: the 1st xref is correctly read, but the "/Prev" one is starting at 1. when the first object are checked an offset in the index is detected and then applied to the whole table (by _zero_xref). This is damaging the 1st xref entries inducing the problem. The correction shall be applied object per object. PR in progress