Desktop: Resolves #12400: Add option to recreate PDF with text transcription

[pedr]

Resolves #12400

Summary

I'm adding a new feature that combines the existing PDF with the transcription generate by the tesseract library.

Since the library output the bounding box of each word, it is possible to recreate the PDF by:

• splitting each page on its own image
• positioning the text over the image with css
• adding some other html wrapper
• sending the html to be "printed" as PDF

Images

When the resource is ready to be created

When the resource already has text annotation

After pressing "Reprocess file"

When the resource is not a PDF

Somethings to consider:

• The generate PDF is much bigger than the original, we probably should consider decreasing the image quality before printing
• Should we replace the original resource with the modified one, seems like it could be useful, but it could also be annoying, depending the user perspective.
• Right now I'm using a font-size of 8px for any text, maybe we could make some calculation about the real size of the text. This could be useful for when someone wants to copy/paste the content of a single line, for example.
  • Try the match the size of the font by counting pixels from the bounding box

Comparing outputs

Configuration	Size	File
original	590kB	scanned_doc_3.pdf
scaleFactor 1, JPG, quality 0.7	1.1MB	scale_factor_1.pdf
scaleFactor 1.5, JPG, quality 0.7	1.9MB	scale_factor_1_5.pdf
scaleFactor 2, JPG, quality 0.7	2.9MB	scale_factor_2.pdf
scaleFactor 2, webp, quality 0.1	1.9MB	webp_quality_01.pdf

Testing

To test it I'm using the PDF linked in the parent issue. I should look for other PDFs to ensure it is working with more cases.

[laurent22]

Sorry I'll close the PR for now as it's too complicated and it went way out of scope