[Poppler-bugs] [Bug 93299] Poppler fails to display pdf correctly

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=93299

--- Comment #7 from Jason Crain <jason at aquaticape.us> ---
Created attachment 121099
  --> https://bugs.freedesktop.org/attachment.cgi?id=121099&action=edit
example rendering

(In reply to Albert Astals Cid from comment #6)
> I'm a bit hesitant on the ifdef for ICU, makes it for hard testing for the
> two codepaths. How much does it actually improve things?

You would prefer ICU be a mandatory dependency?  Or hesitant on having ICU be a
dependency at all?

Completely removing the block of ICU/FcCharSet code, the majority of the text
in this PDF (and a couple others I found) is probably going to be missing
because there's a good chance that fontconfig is going to choose a font without
Cyrillic glyphs.

If I keep the FcCharSet code but remove the ICU IsAlphabetic check, some
documents will have regressions as fontconfig chooses different fonts.  Most of
the differences are benign, replacing a font with another similar font, but
some are not.  Examples in the attached image.

* http://www.epson.com/cmc_upload/0/000/044/561/10000XL_CatSheet.pdf
*
http://upload.wikimedia.org/wikipedia/de/7/77/Wikipedia_2005_Inlaycard_small.pdf
Light font is replaced with a regular font because the light variant is missing
some symbols like bullet, trademark, notequal, and others that are in the
default encodings.

* https://www.ftb.ca.gov/forms/2013/13_540.pdf - numbers along the bottom of
the page
* http://bugs.ghostscript.com/attachment.cgi?id=7435 - ACPIspec40a.pdf - code
samples such as top of page 32
6
And a few other documents.  First choice font is missing a euro symbol so
fontconfig picks a bold oblique font.  I'm starting to suspect this is a
fontconfig bug because there are non-bold non-oblique fonts available further
down the substitution list.

*
http://acroeng.adobe.com/Test_Files/weblinks/Links_Bookmarks/FinalLinkTest.pdf
On page 3, a Bookman font is replaced with a sans-serif font because it's
missing some math symbols.

If those aren't a problem I can do a simpler check, skip ICU/IsAlphabetic and
only filter out Unicode private use area characters.  Or if it is a problem but
you don't want to add ICU for something relatively simple like this, make my
own blacklist of certain characters.  That might be better anyway because I
think the ligatures (ff ffi ffl) are also causing some minor oddities and those
are still alphabetic letters.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/poppler-bugs/attachments/20160118/f4cf7785/attachment.html>