Issue 115618 - [regression] Pdf export: in some fonts Hebrew becomes Gibberish
Summary: [regression] Pdf export: in some fonts Hebrew becomes Gibberish
Status: CLOSED FIXED
Alias: None
Product: gsl
Classification: Code
Component: code (show other issues)
Version: OOO330m15
Hardware: All All
: P3 Trivial (vote)
Target Milestone: OOo 3.3
Assignee: h.ilter
QA Contact: issues@gsl
URL:
Keywords: regression
Depends on:
Blocks:
 
Reported: 2010-11-16 22:29 UTC by netanel
Modified: 2017-05-20 10:22 UTC (History)
5 users (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments
origianl odt file with hebrew strings of the fonts (19.19 KB, application/vnd.oasis.opendocument.text)
2010-11-16 22:31 UTC, netanel
no flags Details
export to pdf by 3.3-rc5 - unreadable (236.58 KB, application/pdf)
2010-11-16 22:32 UTC, netanel
no flags Details
export to pdf by 3.2.1 (oo320m18) - OK (237.09 KB, application/pdf)
2010-11-16 22:33 UTC, netanel
no flags Details

Note You need to log in before you can comment on or make changes to this issue.
Description netanel 2010-11-16 22:29:39 UTC
Export to pdf of some fonts from Culmus family causes the pdf to be unreadable.
This is a regression, as in 3.2.1 (ooo320m18) the export is OK.
This happens only in some fonts from Culmus (not all of them), which is
available only for LInux, so this bug is marked as Linux issue, although it
might happen also in Win, if the fonts are installed there.
http://culmus.sourceforge.net/

wrong exported fonts are: Aharoni CLM, Drugulin CLM, Ellinia CLM, Frank Ruehl
CLM, Miriam Mono CLM, Nachlieli CLM and Yehuda CLM.
Comment 1 netanel 2010-11-16 22:31:18 UTC
Created attachment 75022 [details]
origianl odt file with hebrew strings of the fonts
Comment 2 netanel 2010-11-16 22:32:44 UTC
Created attachment 75023 [details]
export to pdf by 3.3-rc5 - unreadable
Comment 3 netanel 2010-11-16 22:33:56 UTC
Created attachment 75024 [details]
export to pdf by 3.2.1 (oo320m18) - OK
Comment 4 Olaf Felka 2010-11-17 07:21:19 UTC
@ hi: Please have a look.
Comment 5 h.ilter 2010-11-17 15:39:51 UTC
Confirmed, but too late for 3.3
Thereofore set to 3.4
Comment 6 iorsh 2010-11-17 21:26:39 UTC
It seems that only Type 1 fonts in Culmus package are affected by the issue,
while TrueType fonts perform ok.
Comment 7 hdu@apache.org 2010-11-18 10:21:09 UTC
The first few hours of debugging indicate that this regression was probably introduced by issue 107254. 
Especially its variable pUnicodesPerGlyphs is suspicious as it gets calculated as being negative in RTL runs.
Comment 8 hdu@apache.org 2010-11-19 13:44:42 UTC
Fixed in CWS.
Comment 9 iorsh 2010-11-19 15:19:21 UTC
@hdu: is the fix safe enough to be backported to 3.3? Broken Hebrew PDF
production is a rather grave issue, and on my opinion it could severely affect
Hebrew users.

@kaplan: what do you think? Frank Ruehl is still Type1, and I think people use
it extensively. There are several other fonts too.
Comment 10 kaplanlior 2010-11-19 16:37:45 UTC
Well, I do agree that having this included in 3.3 will be the best option for
the Hebrew users. 

Otherwise, I'd like to add this issue to the release notes (or errata to be exact).
Comment 11 kaplanlior 2010-11-20 19:49:07 UTC
@hdu - to which CWS did you commit the fix? I couldn't find it.
Comment 12 hdu@apache.org 2010-11-22 12:18:59 UTC
@hi: please verify in CWS pdffix03
testing hints: Make sure you have the Culmus fonts installed as Type1 fonts
Comment 13 netanel 2010-11-22 12:46:11 UTC
BTW, the CWS isn't looked to be in the Mercurial Repository:
http://hg.services.openoffice.org/cws/pdffix03 
links to the main page of OpenOffice Mercurial Repository.
Comment 14 h.ilter 2010-11-22 13:48:57 UTC
Verified with cws pdffix03 = OK
Comment 15 hdu@apache.org 2010-11-22 14:17:34 UTC
Adjusting target
Comment 16 nyh 2010-11-28 21:31:10 UTC
I ran across the same problem - text in Ellinia CLM font (a stylized Hebrew
heading font) comes out gibberish when exported to PDF. I'm happy you already
have a fix for this. 
In the meantime, until the fix is included, I found a workaround, sort of - you
can "print to file" and generate a correct Postscript file. You can then convert
this Postscript file to PDF if you wish (for me, Postscript is good enough anyway).