RuntimeError: not a dict (null)

[doppelmarker]

Whenever I try to iterate over the pages of 2 pdf-documents attached and achieve the page with form I get this error:

The common thing between them is that both contain Signature fields, maybe the problem is here? Thanks.

failed_pdf1.pdf
failed_pdf2.pdf

[JorjMcKie]

As was to be expected: your files contain errors:

for page in doc:
    print("Page", page.number)

    
Page 0
Traceback (most recent call last):
  File "<pyshell#31>", line 1, in <module>
    for page in doc:
    <...>
    val = _fitz.Document_load_page(self, page_id)
RuntimeError: not a dict (null)
# see what MuPDF has to say:
print(fitz.TOOLS.mupdf_warnings())
invalid indirect reference (0 0 R)
... repeated 2 times...
not a dict (null)
cannot create appearance stream
invalid indirect reference (0 0 R)
... repeated 2 times...
not a dict (null)

If you look at the file directly with some text editor, you will see something like this:

By inspecting all xref sources with PyMuPDF, you could also confirm this:

for xref in range(1, doc.xref_length()):
    if " 0 0 R" in doc.xref_object(xref):
        print(f"Error '0 0 R' in object definition {xref}!")

        
Error '0 0 R' in object definition 27!
Error '0 0 R' in object definition 31!
Error '0 0 R' in object definition 32!
Error '0 0 R' in object definition 33!
Error '0 0 R' in object definition 34!
Error '0 0 R' in object definition 35!
Error '0 0 R' in object definition 39!
Error '0 0 R' in object definition 40!
Error '0 0 R' in object definition 41!
Error '0 0 R' in object definition 42!
Error '0 0 R' in object definition 44!

You must clean the file before you can use a page which refers to an illegal xref.

[JorjMcKie]

"Cleaning" works like so:

• either do a separate CLI step mutool clean -ggg error.pdf clean.pdf (use a garbage option that causes MuPDF to inspect all xrefs), then work with clean.pdf,
• or dynamically do the same in your script:

doc=fitz.open("error.pdf")
pdfbytes=doc.tobytes(garbage=3)
doc.close()
doc=fitz.open("pdf", pdfbytes)

[doppelmarker]

Thanks a lot, this fixed the issue! Actually, I was using 1.19.6 version whereas 1.20.0 was released 5 days ago. Updating to a new version resolves the problem as well.

[PasaOpasen]

@JorjMcKie is there some fast way to check if documents needs cleaning?

[JorjMcKie]

@JorjMcKie is there some fast way to check if documents needs cleaning?

Either an error is detectable at open, then this is fast per se, or it requires actually trying to read/interpret an object - then this is slow.
Even reading a PDF's pages from front to end is no guarantee to detect every error.