Text selection causes unnecessary line breaks

[gracile-fr]

PDFs created with "Print Pages to Pdf" addon based upon the opensource library wkhtmltopdf have an annoying issue when opened with pdf.js: if you copy and paste the textcontent of the PDF, each character is separated by a line break.

E.g.: With this PDF, if you copy and paste "The Mozilla Manifesto", you'll get:
T
h
e
M
o
z
i
l
l
a
M
a
n
i
f
e
s
t
o

[timvandermeij]

Another PDF with the same issue: http://archive.cs.uu.nl/mirror/CTAN/macros/latex/contrib/chessboard/chessboard.pdf#page=43&zoom=auto,0,770. Also, spaces are not copied in your PDF if you copy 'The Mozilla Manifesto'.

[timvandermeij]

Another example: http://sundoc.bibliothek.uni-halle.de/diss-online/05/05H152/t4.pdf

[ReporterX]

Well the top post states it is the issue with wkhtmltopdf, but in the case above, it has nothing to do with it.
The example above is a PDF created by Adobe Reader.
It seems the unintentional line breaks could happen to any PDF.

The bug has been around more than 1 year.
So what is the real culprit? Any clue?

[timvandermeij]

@ReporterX I have updated the title of this issue to reflect the current state. The problem is that each character is put in a separate div. When you copy that, you get the abovementioned behaviour. IIRC work is being done to refactor the text layer, i.e., to reduce the amount of divs (combining them into one).

[abilashs90]

Is there any solution to this?

[jasonparallel]

@timvandermeij Does it still look like the current refactoring work will alleviate this issue?

[timvandermeij]

@jasonparallel I'm afraid not. There are some PRs open for text layer alignment, but combining divs is not yet being looked into. If anyone is willing to do that, we welcome PRs anytime.

[rugk]

Is this related to #2989?
And was there any process made regarding this issue? because copy & pasting currently often works badly in pdf.js.

In any case here is another example: https://www.ietf.org/proceedings/82/slides/rtcweb-13.pdf
When copying the normal text everything works (expect that some characters like { and � seem to be inserted in some places).
But when you copy a text which is displayed non-horizontally like the HTTPS (ROAP?) on site 6 you get many line breaks:

HTTPS
(R
OA
P?)

[yurydelendik]

The original PDF issue was fixed by #6590 (?). Rest of the problematic PDFs might have different problems -- the new issues shall be created for them. Closing as fixed.

[rugk]

So the HTTPS (ROAP?) thing is fixed there?

And what about #2989? Is this also closed by that PR?

And thirdly I've created a separate issue for the character problem: #6658

[yurydelendik]

So the HTTPS (ROAP?) thing is fixed there?

No, it was not. The issue is closed once the reporter's issue is resolved.

And what about #2989? Is this also closed by that PR?

The #2989 talks about using span elements vs div and/or using events to replace the clipboard.

And thirdly I've created a separate issue for the character problem: #6658

Thank you.

[rugk]

So the HTTPS (ROAP?) thing is fixed there?

No, it was not. The issue is closed once the reporter's issue is resolved.

Okay, I've created a new issue for this: #6659