[#XWIKI-49] PDF generation failed with vietnamese characters encoded in UTF-8

XWiki Core

PDF generation failed with vietnamese characters encoded in UTF-8

Created: 16/Jun/05 11:27 Updated: 14/Aug/07 22:29

Component/s:

None

Affects Version/s:

0.9.793

Fix Version/s:

1.0 RC2

Environment:

Linux Debian 3.0, JVM Sun 1.5.03, Tomcat 5.5.9

Issue Links:

Duplicate

This issue is duplicated by:

~~XWIKI-237~~ PDF export fails if there are some special characters

Related

This issue relates to:

~~XWIKI-1609~~ Some UTF-8 characters replaced with '#' when generating PDF

Date of First Response:	13/Sep/05 19:51
Resolution Date:	14/Aug/07 20:18

Description

« Hide

I've got a xwiki installation that contains vietnamese characters encoded in UTF-8.
I can see the characters without problem in my browser. The characters are encoded in UTF-8 int the file ApplicationResouces_vn.properties (with \uXXXX).

However it seems I canno't generate PDF when there are vietnamese characters. I've got the below exception in Tomcat. You can test it by your self on
http://www.cosgov.org

All

Comments

Change History

FishEye

Sort Order:

[ Permlink | « Hide ]

Artem Melentev added a comment - 13/Sep/05 19:51

I found in com.xpn.xwiki.pdf.impl.PdfExportImpl:
..
public PdfExportImpl() {
...
tidy.setCharEncoding(Configuration.LATIN1); // need .UTF8
}
Now PDF generate.
But utf chars in result pdf was replaced by '#'. Maybe special fonts needed?

[ Permlink | « Hide ]

Michael Reinsch added a comment - 13/Mar/06 08:56

[ Permlink | « Hide ]

Mustafa Tan added a comment - 12/Oct/06 14:51

i have the same problem with turkish characters...

[ Permlink | « Hide ]

Sergiu Dumitriu added a comment - 24/Apr/07 04:30

Almost fixed in 2973. Indeed, the problem was at

tidy.setCharEncoding(Configuration.LATIN1);

. I upgraded to the latest JTidy (0.8-SNAPSHOT), which allows setting an encoding by name.

Some characters are replaced by # because FOP does not have a font containing those chars. We need to add such a font and configure JTidy to use it.

Another problem is when XWiki is configured to ISO-8859-1, and the document contains xml char references (&#xyz;). It seems that these references are lost somewhere in the process.

[ Permlink | « Hide ]

Vincent Massol added a comment - 24/Apr/07 09:12

Sergiu, are we going to close this issue for RC2?

I'm worried that we've commmitted code for it in the 1.0 branch and if we don't close it we won't have any jira issue representing the code we have committed.

Thanks
-Vincent

[ Permlink | « Hide ]

Sergiu Dumitriu added a comment - 24/Apr/07 13:06

Well, it does output a PDF (no more stack traces). But the non-latin1 characters are missing from the output. We could close this bug as fixed, and make a new one regarding the missing chars.

[ Permlink | « Hide ]

Vincent Massol added a comment - 24/Apr/07 13:22

We could close this bug as fixed, and make a new one regarding the missing chars.

Sounds the best to me.

[ Permlink | « Hide ]

Catalin Hritcu added a comment - 14/Aug/07 20:18

This should have been closed earlier. New issue is ~~XWIKI-1609~~.