Original and Converted Document Dimentions Don't Match

[deeplow]

Comment by @j75 from another issue:

the original document (https://s1.q4cdn.com/806093406/files/doc_downloads/test.pdf) has Page size: 612 x 792 pts (letter) the test-safe.pdf document has Page size: 1275 x 1650 pts

I had just come across this as well:

	original size	converted size
width (inches)	8.27	17.73
height (inches)	11.7	25.06
width (points)	595.28	1276.46
height (points)	841.89	1804.11

[deeplow]

It turns out that this is an issue with the mismatch of the pixel density when converting into pixels and when reassembling it back into a PDF. This is further conformed by the difference in document size when doing OCR (uses tesseract under the hood) and no OCR (which uses GraphicsMagic).

document	width x height
original	612 x 792 pts (letter)
with OCR	1311.43 x 1697.14 pts
no OCR	1275 x 1650 pts

In Doc to Pixels, it uses pdftoppm with its default configuration (dpi=150).

However, when reassembling it, with OCR it uses dpi=70 and it uses GraphicsMagick with default params, which I assume is using dpi=72).

To fix this I've had success in setting the tesseract-ocr to dpi=150. However, for the non-OCR option I don't have a clue how to fix since with GraphicsMagick the -density param isn't making a difference. But it's not too critical until we decide on whether or not we want #627.

[j75]

uses tesseract under the hood - well, in this case there might be another issue (at least on Ubuntu) - the package does not depends on tesseract!

Depends: python3:any (>= 3.6~), podman, python3, python3-pyside2.qtcore, python3-pyside2.qtgui,
         python3-pyside2.qtwidgets, python3-pyside2.qtsvg, python3-appdirs, python3-click,
         python3-xdg, python3-colorama, python3-requests, python3-markdown, python3-packaging

In my case fortunately I already have this package installed, however in the general case I think it should be part of deb's dependencies.

[deeplow]

Oh, that part is fine because tesseract comes pre-installed in the container image we use. So even if one's system doesn't have tesseract, it will still work.