Find a work-around for scanned pdf

[mvicenzi]

Scanned pdf documents are not read properly and operations fail.
Investigate this issue: is it a PyPDF2 limitation? is there a workaround?

Possible starting points:

• https://stackoverflow.com/questions/49939085/xref-table-not-zero-indexed-id-numbers-for-objects-will-be-corrected-wont-con
• PDFFileMerger : PdfReadWarning: Xref table not zero-indexed. ID numbers for objects will be corrected. [pdf.py:1736] py-pdf/pypdf#611

[pubpub-zz]

@mvicenzi, can you provide some pdf samples and a test code for analysis

[mvicenzi]

@pubpub-zz you can find the simplest code to reproduce the error here. The sample pdf file is this one, which was generated with a scanner.

[pubpub-zz]

I've tried with the dev version without any error.can you retry with the latest version of PyPDF2?

[mvicenzi]

I upgraded PyPDF2 from 1.27.3 to 1.27.12, but I still see errors.
I'm on python 3.10.2

Traceback (most recent call last):
  File "C:\Users\matte\Desktop\pdf_tools\debugging_scanned.py", line 7, in <module>
    merger.append(path1, bookmark=None, pages=None, import_bookmarks=True)
  File "C:\Users\matte\AppData\Local\Programs\Python\Python310\lib\site-packages\PyPDF2\merger.py", line 227, in append
    self.merge(len(self.pages), fileobj, bookmark, pages, import_bookmarks)
  File "C:\Users\matte\AppData\Local\Programs\Python\Python310\lib\site-packages\PyPDF2\merger.py", line 157, in merge
    pages = (0, pdfr.getNumPages())
  File "C:\Users\matte\AppData\Local\Programs\Python\Python310\lib\site-packages\PyPDF2\_reader.py", line 318, in getNumPages
    self._flatten()
  File "C:\Users\matte\AppData\Local\Programs\Python\Python310\lib\site-packages\PyPDF2\_reader.py", line 706, in _flatten
    catalog = self.trailer[TK.ROOT].getObject()
  File "C:\Users\matte\AppData\Local\Programs\Python\Python310\lib\site-packages\PyPDF2\generic.py", line 553, in __getitem__
    return dict.__getitem__(self, key).getObject()
  File "C:\Users\matte\AppData\Local\Programs\Python\Python310\lib\site-packages\PyPDF2\generic.py", line 198, in getObject
    return self.pdf.getObject(self).getObject()
  File "C:\Users\matte\AppData\Local\Programs\Python\Python310\lib\site-packages\PyPDF2\_reader.py", line 801, in getObject
    raise PdfReadError(
PyPDF2.errors.PdfReadError: Expected object ID (8 0) does not match actual (7 0); xref table not zero-indexed.

I also tried cloning again the repo to check if uploading the pdf had fixed the file somehow, but no.... Still the same error as above.

[pubpub-zz]

I've understood : your PDF does not respect the rule where the xref table should start at 0, inducing a PdfReadError if strict is asserted. PyPDF2 moved to strict = False as default, but PdfFileMerger has been forgotten. I will push the fix however meanwhile you can initialize with strict = False:

merger = PdfFileMerger( strict = False)