Bug#714297: Issue is with gen-guch-unicode-tables.pl

[Caliste Damien]

Hello,

   Thanks to the backtrace generated by Paul Wise, we can see that the
SIGSEG error appears because of strlen(NULL) at line 99. This is because
haystack_nfd is NULL, because the call to g_utf8_normalize() failed on
line 175.

   It failed because the nameslist_equals string is not UTF8 valid. One
can see it inside gucharmap, by looking for the character 0x00E6
( LATIN SMALL LETTER AE) there is an empty entry there.

   This string comes from unicode_nameslist.h (look for "ash (from"
there is a UTF8 character there) and is valid from the GNOME files. But
we regenerate this file using the Perl script
gen-guchar-unicode-tables.pl. And running this script
on /usr/share/unicode/NamesLists.txt generate a wrong
unicode_nameslist.h with not valid UTF8 characters inside.

   I'm not a Perl expert (already spent 20 minutes to understand what
was wrong with qw() syntax in this script) so I cannot go further, but
the solution for this issue seems clear: fix the Perl script to
generate valid UTF8 header files for gucharmap. Who's going for it ? ;)

   By the way, gucharmap may check every returns of g_utf8_normalize()
in case, even if in this case, strings should be UTF8 valid since they
are provided in headers files and not user provided.

Best regards,

Damien.