Extract text from "flattened" PDF?

[jasonharitou]

Hi,

I'm attempting to extract text from a PDF, but it's not being extracted. I've tried with multiple other PDFs and the text extracts fine, so there's something particular about this PDF in question.

Could anyone suggest some ways I can debug why the text is not being extracted? For reference the PDF is a single page, version 1.4, Author is Canva, and Security is None.

File itself is 1.8MB.

Any help would be appreciated. Thanks

[jasonharitou]

Did a bit more testing with Canva - when a document is "flattened" during the export process, it becomes unreadable by pypdfium2. Please see the two attached files.
canva-flattened.pdf
canva-full.pdf

[mara004]

AFAICS, the canva_flattened PDF just doesn't have a text layer, so there is nothing for a pdf engine to extract. You could use an OCR tool like ocrmypdf to change that, though.
Also note that canva_full is a vector PDF, whereas canva_flattened merely contains a rasterized image of the former.
For this reason, I recommend just not using this "flattening", as it evidently rasterizes, which is inelegant (maybe you've noticed the file size of canva_flattened is way higher than of canva_full).

[jasonharitou]

Yeah I noticed the full version was massive! Thanks for your answer, OCR tools here I come…

[mara004]

(note for readers: canva's flattening should not be confused with PDF form field flattening as provided by pdfium)