[R] Warning object has offset 0. PDF file

[Jean-Louis Abitbol]

Thanks so much Ivan, JL

On Mon, Oct 14, 2024, at 1:33 PM, Ivan Krylov wrote:
> В Sun, 13 Oct 2024 12:45:36 +0200
> "Jean-Louis Abitbol" <abitbol using sent.com> пишет:
>
>> WARNING:
>> /Users/jla/Library/CloudStorage/Dropbox/7cordas/Caio/record/90 NEW
>> RODA SONG BOOK.pdf (object 254 0): object has offset 0
>> 
>> This does not prevent from getting the pages extracted properly and
>> written to a pdf file which is readable.
>
>> How can I detect and potentially correct using appropriate open
>> source software what is wrong with the pdf file. I am on macos.
>
> Here's a couple of links, [1] where people investigate and find the
> root cause of the same warning and [2] where a number of possible ways
> to corrupt PDF (and warning messages from various software) is
> investigated, including this one.
>
> Ironically, this is almost in scope for R-package-devel, because the
> message comes from "qpdf", a tool used by R CMD check (and
> tools::compactPDF() in particular). The internal "table of contents" at
> the end of the PDF file contains nonsensical references, which "qpdf"
> complains about. If you run the following command line:
>
> qpdf "/path/to/original/file.pdf" "/path/to/reconstructed/file.pdf"
>
> ...qpdf will complain about the file once again, but it will also
> produce a reconstructed file, which won't have this particular problem.
> "qpdf" tries very hard to be "loss-less" about its transformations, so
> once you verify that the reconstructed file is fine, you can replace
> the original with it.
>
> -- 
> Best regards,
> Ivan
>
> [1]
> https://github.com/OpenPrinting/cups/issues/321#issuecomment-1246741326
>
> [2]
> https://openpreservation.org/blogs/trouble-shooting-pdf-validation-errors-a-case-of-pdf-hul-38/

-- 
Verif04062023