Issue metadata
Sign in to add a comment
|
PDF equation/math rendering broken
Reported by
emil.fih...@gmail.com,
Jan 17 2018
|
||||||||||||||||||||||||
Issue descriptionUserAgent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36 Example URL: Steps to reproduce the problem: 1. Open PDF in browser 2. Look at equation What is the expected behavior? Equations rendered correctly What went wrong? Equations have missing/invisible symbols Does it occur on multiple sites: N/A Is it a problem with a plugin? N/A Did this work before? N/A Does this work in other browsers? Yes Chrome version: 63.0.3239.132 Channel: stable OS Version: 10.0 Flash Version: Firefox renders okay Page 14-15 in the longer pdf
,
Jan 17 2018
My guess would be that this is a font issue. Is the font embedded in the PDF or are we doing substitution? npm@ would you have time to take a look?
,
Jan 17 2018
According to aphirst on #chromium-support @ Freenode there is a lot of font substitution going on.
,
Jan 18 2018
Maybe this can be fixed today before the M65 branch cut to avoid needing to merge? With ITP_2018_Laskut_viikko_2.pdf: - This likely regressed in r505649. Bisect says 505642 good / 505660 bad. https://pdfium.googlesource.com/pdfium.git/+/10e1f05a9e644cd954792bcd40ef787551cbd209 or https://pdfium.googlesource.com/pdfium.git/+/dd0e6e1eba14c76dedd4b4e55ab47406856c9a76 ? - Edge displays it correctly. - libpoppler on Linux has trouble, possible due to bad substitution
,
Jan 18 2018
Also, in ITP_2018_Laskut_viikko_2.pdf, the font that's not displaying correctly is "CambriaMath".
,
Jan 18 2018
These files both render correctly with pdfium_test at tip-of-tree on Linux. Going to try in tip-of-tree Chromium on Linux and see what happens.
,
Jan 18 2018
This does repro in a Chrome tip-of-tree build on Linux.
,
Jan 18 2018
I reverted chrome to before 505649 and the issue still repros for me, so I think this happened before that. I also tried reverting the two suspect CLs and the issue still repro'd in Chrome.
,
Jan 18 2018
To answer the question in #2, according to Evince there is a font TrueType CID font CambriaMath that is embedded. But there is also a TrueType non-embedded font CambriaMath that is not embedded, so that must be the cause of the problem (and no, we do not substitute fonts using fonts embedded in the PDF - maybe we should start thinking about doing that?). What OS the bisect was on? On Linux, the issue is present before the 'good' of bisect from #4.
,
Jan 18 2018
I've reverted back to the M61 branch and this bug existed at that point so I think this is just an existing issue, not a regression.
,
Jul 3
re: comment 10 - is that on Linux? I was trying on Windows. Try bisecting on Windows between 505642 and 505660.
,
Jul 3
This is reproducible with pdfium_test on Windows. It bisects to https://pdfium.googlesource.com/pdfium.git/+/10e1f05a9e644cd954792bcd40ef787551cbd209
,
Jul 3
Hmm, that CL looks familiar. Looks like this is already fixed on Chrome Canary. |
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by torne@chromium.org
, Jan 17 2018