Impossible to mark text in some .PDF's, primarily affecting home-copies of books.

[gevvan]

Hey guys!

Thanks for great software. There's only one little thing that gives me a headache right now. My subject line says it all. See the sample that I've linked here:

https://sharey.org/files/0X1cv1.pdf

In Chrome-based (or Firefox-based) browsers, I can easily mark text using my mouse. It is also doable without any problem in both ABBY and Adobe. But with Sumatra, when I try to mark text it becomes like a big weird square-formed thing that .. well .. it simply don't want to comply in markingletter by letter.

I guess that there is some formatting issues with the .pdf, but how come all other .PDF reading software (except for Sumatra pretty much) handles it? Could this be resolved easily you think? I have access to dozens of .pdf book copies with this exact issue, sigh.

[GitHubRulesOK]

You will need to use an Editor to correct the bad OCR fonts here I changed them into Times Roman and SumatraPDF has no problem with correctly sized fonts. so you can see it can with a good PDF select a single r
SumatraPDF is not a browser editor plug-in like I show it here. and when that file was first loaded I had the same problems it was only after edit the fonts in a PDF editor I can use it like any other web or local file.

The reason is the exceptionally big heights as shown by XHeight 84149251

<</Ascent 67306242/AvgWidth 640/CapHeight 100992260/Descent -10/Flags 262148/FontBBox[-10 -15 788 785]/FontFile3 22 0 R/FontName/*Minion#20Pro-Bold-6046/ItalicAngle 0/Lang/SV/MaxWidth 680/StemV 100/Type/FontDescriptor/XHeight 84149251

[gevvan]

Thanks for your response @GitHubRulesOK, appreciated.

I understand, but I still find this to be an issue. I mean, albeit not being an optimally formatted PDF document, I can mark text smoothly in all web browsers that I've tried, and also ABBY PDF Reader, and Acrobat PDF Reader. This being under Windows. In Linux environment I have been able to mark text without hassle in all the PDF readers that I've tried, so far Evince, Atril, Zathura, and Okular. A few of them aiming to be in the same league as Sumatra, i.e. light-weight readers.

So, all in all, it's like every reader software except for Sumatra manages to mark text in this file. So my question is, why is this? Also, I downloaded an old build of Sumatra (v2.something, the branch with the funny comic-like text in the splash screen) and opened the very same .pdf sample, and guess what, in there I could mark text.

So, it's like something has been changed in more recent builds of Sumatra that breaks the ability to mark text in files such as this. I'm not sure if Sumatra itself is to blame, or if the issue lies within the engine (MuPDF). Anyways, this has been a headache to me for a very long time. There's even (for Windows) an über-lightweight .PDF reader called "PDF Lite", which the creator states is a fork of Sumatra. In there, I'm able to mark text in the .pdf without any problems. This (PDF Lite) is, however, based on an ancient Sumatra (v2.x or even v1.x) so it feels obsolete, but right now just to have a minimalistic reader that supports marking text, I have both (recent) Sumatra and (ancient) PDFLite installed side-by-side. I am, of course, hoping to be able to ditch PDF Lite.

Hopefully some developer could take a look into this. Maybe it's a very simple fix. Again, every other reader that I've tried marks the text successfully, except for Sumatra.

[GitHubRulesOK]

PDF Lite was a viral knock off (though several variants) so beware some had at one time a poor VirusTotal high score.

Yes MuPDF is the core engine for SumatraPDF and varies (daily) from version to version due to "improvements"
Current version 1.26 pre-release would repeatedly (due to the overlapping text lines) just blat the main screen area
Note top line and bottom line behave different to body text

0X1cv1(TimesRoman).pdf
The page seems to have been subject to some scale manipulations as the source should have behaved like this variation

[gevvan]

Thanks again for your response! 🙏

Also, your "insight" in the sample is very interesting, and fascinating.

But yeah, again, seeing that pretty much every PDF viewer (i.e. out of those that actual has text-marking feature) handles this without having to change font and re-scale and stuff, perhaps developers could have a look at it. If Sumatra is more or less just a GUI then perhaps I should make a thread in the Issues page for MuPDF instead? I was thinking about it but I looked in there and it looked like the discussion was all about the actual library, and technical mumbo-jumbo about it, and I thought it might not even be welcome to address an issue that shows up under a GUI.

Hmmm, is there any other commonly known PDF viewer (other than Sumatra) that uses recent build MuPDF for engine? If so it would be interesting to load my sample in there, if it behaves the same way we could rule out the Sumatra "exterior" and pinpoint it to the engine beneath.

[GitHubRulesOK]

I showed it happens in MuPDF-GL the one with the Artifex/PDF.co MuCow and their multiple overlaid purples because they will say the font in each line says it is to be drawn as if gigantic letters.

Their call could naturally be it is a font problem, however if you raise the bug report based on their viewer has an issue, when others do not, they may revisit that part of their code (which probably depends on one of their 3rd party font handlers, such as freetype?)

[GitHubRulesOK]

This discussion is now covered by currently open issue #4919