Dear Abiword team,
Here I'm forwarding to you information and the patch partially fixing crash on
unicode character(s) paste. (2.8.6 and the trunk versions are affected).
Patch is only fixing this problem in 'Paste' but 'Paste Unformatted' is still
affected. I'm sure you would know better how to fix.
Just to remind, this important problem can be easily reproduced with the
following steps:
* Go to http://en.wikipedia.org/wiki/Linear_B#Unicode
* Copy one or more character(s) from the Unicode table
* Paste in a new Abiword document.
I also provided this information in remotely related ticket #13280 but I hope
it will have better visibility here.
Thank you.
Regards,
Dmitry.
---------- Forwarded Message ----------
Subject: Bug#681060: abiword: Crash when copying Unicode text from Firefox
Date: Mon, 13 Aug 2012, 03:40:23
From: "Ying-Chun Liu (PaulLiu)" <paulliu@debian.org>
To: 681060@bugs.debian.org
Hi,
I think I found the root cause of the bug.
It is because the character is not in UCS-2 but in UTF-16LE.
I first make a patch to add an error message to the g_convert function call.
As attachment add_debug_message_of_g_convert.patch
And then it prints
"ut_iconv.cpp:379 UT_convert() Cannot convert str fffe00d801dc(6) from
UCS-2LE to UTF-8 because Invalid byte sequence in conversion input"
And because the conversion is failed, thus the result string is NULL.
And the NULL passed to memmove() at ut_bytebuf.cpp:96 which causes the
CRASH.
And then I just reorder the encoding preference order in
src/af/xap/xp/xap_EncodingManager.cpp. Please see attachment
reorderencoding.patch.
It works fine. The character is just appeared onto abiword.
So my suggestion is we either do the proper handle of the conversion
failure (when NULL is returned, we ignore the paste maybe). Or we can
make "UTF-16LE" as preferred value rather than UCS-2.
Yours,
Paul
-- PaulLiu (劉穎駿) E-mail: Ying-Chun Liu (PaulLiu) <paulliu@debian.org> -----------------------------------------
This archive was generated by hypermail 2.1.8 : Mon Aug 13 2012 - 16:10:45 CEST