Markdown reader: taking an unusually long time during a certain kind of input

[DaedlyKitten]

The following is a 104 KB txt file of merely repeating the same paragraph that after some trial and error I managed to isolate from a lengthy novel I initially wanted to convert into EPUB,
when converting this sample txt to a EPUB file, it will take an exceedingly very long time, approximately several minutes on my PC, when usually any other ~100kb txt file conversion takes less than 1 second.

This is the file and command I used:

pandoc.exe "%homepath%\Downloads\sample.txt" -o %homepath%\downloads\11111.epub

sample.txt

version info:
pandoc.exe 2.5
Compiled with pandoc-types 1.17.5.4, texmath 0.11.1.2, skylighting 0.7.4
Default user data directory: C:\Users\konpo\AppData\Roaming\pandoc
Copyright (C) 2006-2018 John MacFarlane

[DaedlyKitten]

some further tries seem to suggest it has something to do with its particular length rather than specific characters? as the conversion time gets back to normal when i inserted random characters into the paragraph...

[mb21]

I can confirm even this with pandoc 2.5 takes way too long:

pandoc -f markdown -t native https://github.com/jgm/pandoc/files/2693913/sample.txt

Yet it works fine with pandoc 2.4

[jgm]

If this is something that occurs in 2.5 but not previous versions, then I suspect the following change in the 2.5 chalnegolg:

Fix parsing of citations, quotes, and underline emphasis after symbols. Starting with pandoc 2.4, citations, quoted inlines, and underline emphasis were no longer recognized after certain symbols, like parentheses (#5099, #5053).

This could be verified by using nightlies produced from before and after these changes.

[mb21]

@jgm, you are right. Reverting edc6510 fixes the performance degradation:

git revert edc651059ee617cf36b511de080a646d2e6513a4

[jgm]

@DaedlyKitten if you use -f markdown-smart, you will not experience any performance degradation.
It happens because pandoc is trying to match up quotes and things like

'去岁嵩山之会

are causing the problem. It would be good to fix things so you don't get the slowdown even with +smart, but this workaround should be fine for your purposes.

[DaedlyKitten]

@jgm thank you :)
I did the conversion with an earlier version of pandoc.
Thanks again for this awesome software!

[mb21]

It would be good to fix things so you don't get the slowdown even with +smart

therefore reopening.

[jgm]

A similar example from #7306, using pandoc -f textile:

<config>
    <section1>
        <a>
                <format identifier="1" offset="0" length="22" pn_format="%u-%u-%u" sn_format="%u-%u-%u-%u">%*02u%04u%04u%04u%02u%02u%03u%03u</format>
        </a>
    </section1>
    <section2>
        <b>
                <format identifier="2" offset="0" length="17" pn_format="%u-%u-%u" sn_format="%u-%u-%u-%u">%02u%03u%03u%02u%02u%02u%03u</format>
        </b>
    </section2>
    <section3>
        <c>
                <format identifier="2" offset="0" length="17" pn_format="%u-%u-%u" sn_format="%u-%u-%u-%u">%02u%03u%03u%02u%02u%02u%03u</format>
        </c>
    </section3>
</config>