Error during updating metadata

[DrChiper]

Hi Kovid, everytime I update the metadata of an ebook, of which multiple files are present I get an error concerning the impossibility of updating the PDF-file, while the other files are correctly ignored. Can the PDF-file somehow be marked to be ignored too?

[kovidgoyal]

The other files are not ignored there is just no error setting their metadata. Click the copy to clipboard button and paste the full error

[DrChiper]

Running on Windows 11, this is the full message:

calibre, version 6.27.0
WARNING: Failed for some files: Failed to embed metadata into some book files. Click "Show details" for details.

The PDF format of  xxxxx:

Traceback (most recent call last):
  File "calibre\utils\podofo\__init__.py", line 36, in set_metadata
  File "calibre\utils\ipc\simple_worker.py", line 255, in fork_job
  File "calibre\utils\ipc\simple_worker.py", line 180, in run_job
  File "calibre\utils\ipc\simple_worker.py", line 115, in communicate
calibre.utils.ipc.simple_worker.WorkerError: Failed to communicate with worker process

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "calibre\customize\ui.py", line 488, in set_file_type_metadata
  File "calibre\customize\builtins.py", line 525, in set_metadata
  File "calibre\ebooks\metadata\pdf.py", line 185, in set_metadata
  File "calibre\utils\podofo\__init__.py", line 40, in set_metadata
Exception: Failed to set PDF metadata in (xxxxxx): Traceback (most recent call last):
  File "multiprocessing\connection.py", line 317, in _recv_bytes
BrokenPipeError: [WinError 109] The pipe has been ended

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "calibre\utils\ipc\simple_worker.py", line 53, in run
  File "calibre\utils\ipc\__init__.py", line 24, in eintr_retry_call
  File "multiprocessing\connection.py", line 255, in recv
  File "multiprocessing\connection.py", line 326, in _recv_bytes
EOFError

[kovidgoyal]

That looks like something on your computer is interfering with calibre, most likley an antivirus, try rebooting in safe mode and see if the errors stop

[DrChiper]

I get the same error in "Safe Mode". The PDF itself seems ok. Strangely enough not all ebooks with additional PDFs show errors. Those ebooks were added with calibre 6.25 or earlier.
I try to see whether I can find file property differences.

[DrChiper]

It narrows down to the PDF-file: the file itself is perfectly genuine (PDF version 1.5), but somehow still not appreciated by calibre for updating the metadata. I'll investigate further.

[BetterRed]

Originally Posted by DrChiper

It narrows down to the PDF-file: the file itself is perfectly genuine (PDF version 1.5), but somehow still not appreciated by calibre for updating the metadata. I'll investigate further.

Check the PDF security settings, a PDF can be protected against editing - without password protection. So it can be read but it cannot be changed.

BR

[ownedbycats]

I'd also check whether there might be a read-only flag on it.

[DrChiper]

@BetterRed & @ownedbycats: That were the first things I checked: negative.

I found the cause: its in the PDF file handling.
A working PDF-file has the following ending:

<</ID[<B43E3B0286664B420FA6AC3FE4CBFF8B><B43E3B0286664B420FA6AC3FE4CBFF8B>]
/Info 1935 0 R
/Root 1937 0 R
/Size 1945>>
startxref
55999427
%%EOF

My failing PDF-file (which is however perfectly genuine) shows:

<</Size 1127
/ID [(Vؘr„g‘X%ÃVCÙÂ) (Vؘr„g‘X%ÃVCÙÂ) ] 
/Root 1625 0 R >> 
startxref
37312594
%%EOF

Calibre did show as an error "EOFError", so I concluded it was parsing to something it expected, but did not find before hitting %%EOF.

As can be seen, my PDF-file is missing the (optional) entry "/info xxxx x R", so I manually added that and now calibre does not complain anymore.

My PDF-file was generated by using the well-known Poppler tool "pdfunite" (I tried multiple versions and the most recent 23.08.0 one for windows) , which never seems to include the "/info" entry.

First conclusion therefor is that the calibre parsing-tool is expecting something which not necessary needs to be present.
PDF-files are (and will be) a pain in the xxx

[kovidgoyal]

calibre uses podofo to parse PDF. I doubt that requires the /Info dictionary but opena bug report and attach your PDF and I will take a look, podofo recently underwent a rewrite which broke quite a few things, maybe it broke this.

[DrChiper]

Ah, that figures.
Please find my 2 test PDF-files attached.

[kovidgoyal]

OK I'll look into it when I have a moment

[kovidgoyal]

https://github.com/kovidgoyal/calibr...16519551f5b4b5

[DrChiper]

Thanks Kovid, will give it a try on the next release.

[DrChiper]

The PDF-files seem to pass-on fine now using calibre 6.28.