Glenn Adams commented on FOP-2409:

The basic problem is that you got to clever with xml:lang  when you set it to 
"ar_SA" instead of just "ar", or leave out entirely. At present, FOP does an 
exact match from a specified xml:lang to the font's tables, which are indexed 
by the tuple <feature, script, language>, and none of the fonts had a table 
matching a language of "ar_SA". Note that you really should use the XSL-FO 
"language" attribute/property, and not xml:lang. See [1] for more info.

However, even after fixing this, it turns out there is still a bug being 
tickled by this input related to <fo:page-number/> within a sequence of BIDI 
text. You would notice this if you had assertions enabled. I have created Issue 
FOP-2410 to track this problem.

Finally, I see you specified writing-mode="rl" on fo:root; however, it is not 
significant on fo:root, but is significant on fo:page-sequence. See [1] for 
further info.


> Arabic characters mess up on PDF
> --------------------------------
>                 Key: FOP-2409
>                 URL: https://issues.apache.org/jira/browse/FOP-2409
>             Project: Fop
>          Issue Type: Bug
>    Affects Versions: trunk
>         Environment: DITA-OT 1.8.4 (partially patched)
> FOP from trunk (07/18/2014)
>            Reporter: Red Shen
>         Attachments: basic_connector_instructions.xml, fop.pdf, topic.fo, 
> xep.pdf
> I don't understand Arabic. But just go look at the PDF output from FOP and 
> XEP, the title text looks different. The XEP one should be correct.

This message was sent by Atlassian JIRA

Reply via email to