Hi! I have the following two documents: The original: What is a Graph Database? ========================= The translation: Qu'est-ce qu'une base de données graphe? ======================================== Running po4a-gettextize on them gives: po4a gettextization: Structure disparity between original and translated files: msgid (at target/original/src/dummy.asciidoc:2) is of type 'Title =' while msgstr (at src/dummy.asciidoc:2) is of type 'Plain text'. Original text: What is a Graph Database? Translated text: Qu'est-ce qu'une base de données graphe?

Le 27/09/2012 07:55, D. Barbier a écrit : > Indeed, this is due to accented characters. > It seems that length() returns the number of bytes and not characters. > I looked at Unicode issues with Perl a very long time ago and do not > remember about its quirks; if anyone has a clue, please tell ;-) Thomas, CCed, helped us a lot for the DPNhtml2mail script [0], and managed to make that work.

On 2012/9/27 David Prévot wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA256 > > Hi, > > Le 27/09/2012 07:55, D. Barbier a écrit : > >> Indeed, this is due to accented characters. >> It seems that length() returns the number of bytes and not characters. >> I looked at Unicode issues with Perl a very long time ago and do not >> remember about its quirks; if anyone has a clue, please tell ;-) > > Thomas, CCed, helped us a lot for the DPNhtml2mail script [0], and > managed to make that work. > >> 0: http://anonscm.debian.org/viewvc/publicity/dpn/scripts/DPNhtml2mail.pl?vi... > > I guess the magic operates in the end of the following code: > > # number of column of a string > sub _columns { > my $str = scalar shift; > > return 0 if ( !defined $str || $str eq '' ); > > $str = decode_utf8($str) unless utf8::is_utf8($str); > return Unicode::GCString->new($str)->columns(); > } Thanks David, This seems to be different, you are computing the string width whereas I need the number of characters. I believe that all we need is to add some ":encoding(foo)" flag when opening file for reading, encoding must be specified and is thus known.

I installed it from CPAN, and still get the same message.

As you may recognize I'm not that used to Perl :-)

On 2012/10/1 Anders Nawroth wrote: > Hi! > >> I was wrong, we need text width; I checked in some commits to use >> Unicode::GCString if available, thanks to both of you for the help. >> The downside is that there is one new string in PO files. > > For me, it failed with: > po4a::text: Detection of two line titles failed at > target/original/src/introduction/index.asciidoc:3 > Install the Unicode::GCString module! > > (so "use if available" doesn't seem to work?) I was unclear; what I meant is that Unicode::GCString will be called only if it is needed. And it is needed if a title line contains non-ASCII characters. But if your titles contain only ASCII characters, po4a will work even if Unicode::GCString is not installed. > I installed it from CPAN, and still get the same message. > > I currently run po4a using PERLLIB pointing to my own copy pf po4a. > I tried copying the lib/ part of GCString into the same location, but > that didn't work either. > > As you may recognize I'm not that used to Perl :-) > > Any pointers? You must make sure that this module can be loaded: $ perl -MUnicode::GCString -e '' If there is an error, it has to be fixed first. Denis _______________________________________________ Po4a-devel mailing list Po4a-devel(a)lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/po4a-devel