Re: Multilingual support / was Re: Urgent: MsgIdExtract problem

RobertCZ Wed, 22 Dec 2004 03:53:22 -0800

Gerald Richter wrote:

Thanks for your feedback. I try to bring the things into the 2.0 release.

- MsgIdExtract ignores [- $epreq->gettext(expr) -] I don't 
know how to modify the syntax, but it's clearly a bug


MsgIdExtract at the moment only knows about [= =], maybe I can try to parse
it out gettext as well, but it will only work for literal strings. I will
think about a solution.

I believe literal strings would be enough, I was planning something like

sprintf $epreq->('%d records founds'), $found;

which would then generate "Nalezeno 32 zaznamu" or "32 records found" etc... so the gettext'ed string itself is just a literal string

PS What about that additional quotes after running extract and using deafult message if translated string is empty/undef instaed of missing key only?

PS2 I'm already using it in my two big projects and it's really REALLY nice ;-)

-----Original Message-----
From: RobertCZ [mailto:[EMAIL PROTECTED]] 
Sent: Wednesday, December 15, 2004 12:49 PM
To: [email protected]
Cc: Gerald Richter
Subject: Multilingual support / was Re: Urgent: MsgIdExtract problem


Gerald Richter wrote: 

		
'$Embperl::Syntax::MsgIdExtract::Ids{scalar(%#0%)} = q{} if (!exists
		($Embperl::Syntax::MsgIdExtract::Ids{scalar(%#0%)})) ;',
		    

	
	Try to change %#0% to %#\'0%
	  


OK, now I got it working and it's really cool (with the 
exception of the last couple of bugs). Here are my comments 
(relevant for 2.0RC2):


- I changed MsgIdExtract syntax from %#0% to %#\'0%, now it 
correctly extracts more complex expressions, but it adds 
superflous quotes and a space to the extracted expressions, 
it looks like this  '\'Test \'' => '', This can be hacked in 
the extract script with something like

foreach my $fn (@files) {
...
    Embperl::Execute ({use_env => 1, use_redirect_env => 1, 
syntax => 'MsgIdExtract', inputfile => $fn, output => \$out, 
errors => [EMAIL PROTECTED]) ; ...
}
my @correct_keys = map { /^\'\s*(.*?)\s*\'$/ } keys 
%Embperl::Syntax::MsgIdExtract::Ids;

but I think it's a bug


- MsgIdExtract ignores [- $epreq->gettext(expr) -] I don't 
know how to modify the syntax, but it's clearly a bug


- when messages are dumped with $Data::Dumper::Useqq  = 1 it 
break 8bit chars (dump as octal, but translators would go mad 
should they see it)


- I find it more practical to dump sorted messages with 
something like $Data::Dumper::Sortkeys  = \&{ sub {[ sort { 
$a cmp $b } keys %{$_[0]} ]} } 


- app file must delete empty (missing) translations to force 
EP use message in the default language, this is quite 
impractical because translators need it in the dictionary 
file and so during the development I need to modify the 
dictionary either for EP or for translators. Now I delete it 
in app file but it's stupid performance-wise

- Robert

-------------------------------------

Just in case somebody needs it, below are necessary files

_app.eo

-------------------------------------

@ISA = ('Embperl::App') ;

sub init {
        my $self = shift ;
        my $r = $self->curr_req ;
    my $fdat = $r->thread->form_hash;        
        my $lang = $fdat->{lang} || 'en' ;
    my %messages = %{ do '/home/www/germic/etc/dict/dict.dump' };

    foreach my $l (keys %messages) {
        foreach my $t (keys %{ $messages{$l} }) {
            delete $messages{$l}{$t} unless $messages{$l}{$t};
        }
    }

#use Data::Dumper; print STDERR "\n\n", Dumper(\%messages), "\n\n";
        
        push @{$r->messages}, $messages{$lang} ;
        push @{$r->default_messages}, $messages{'en'} if 
($lang ne 'en') ;

    return 0;

}
 
1 ;

-------------------------------------

extract

-------------------------------------

BEGIN 
    {
    %Embperl::initparam = (use_env => 1, use_redirect_env => 1) ;
    $ENV{EMBPERL_SESSION_HANDLER_CLASS} = 'no' ;
    }    
    
use Embperl;
use Data::Dumper ;

use strict;
use vars qw{$srcpath $dictpath @files @languages $msgids 
@correct_keys};

$srcpath    = '/home/www/germic';
$dictpath    = '/home/www/germic/etc/dict';
@files        = split ' ', `find $srcpath -name *.html 
-printf "%p " -or -name *.eo -printf "%p "`;
@languages    = qw(cz de); # en not listed, it's default!

$msgids = do "$dictpath/dict.dump" ;
die $@ if ($@) ;
#print Dumper($msgids);

foreach my $fn (@files) {
    my ( $out, @errors );
    Embperl::Execute ({use_env => 1, use_redirect_env => 1, 
syntax => 'MsgIdExtract', inputfile => $fn, output => \$out, 
errors => [EMAIL PROTECTED]) ;
    if (@errors) {
        print join ("\n", @errors) ;
        last ;
    }
}
#print Dumper(\%Embperl::Syntax::MsgIdExtract::Ids);

@correct_keys = map { /^\'\s*(.*?)\s*\'$/ } keys 
%Embperl::Syntax::MsgIdExtract::Ids;
#print Dumper([EMAIL PROTECTED]);

foreach my $lang (@languages) {
    foreach my $k (@correct_keys) {
        $msgids -> {en}{$k} = $k ;
        $msgids -> {$lang}{$k} = '' if (!exists $msgids -> 
{$lang}{$k}) ;
    }
}
#print Dumper($msgids);

rename "$dictpath/dict.dump", "$dictpath/dict.bak" ; open FH, 
">$dictpath/dict.dump" or die "Cannot open 
$dictpath/dict.dump ($!)" ; # $Data::Dumper::Indent = 1 ; # 
default is nicer # $Data::Dumper::Useqq  = 1 ; # breaks 8bit 
chars $Data::Dumper::Sortkeys  = \&{ sub {[ sort { $a cmp $b 
} keys %{$_[0]} ]} }  ; # more practical print FH 
Data::Dumper -> Dump  ([$msgids], ['msgids']) ; close FH ;

-------------------------------------

test

-------------------------------------

<h1>trans test</h1>

<p>Test: <b>[= Test =]</b>
<p>Hi!: <b>[= Hi! =]</b>
<p>Thank You!: <b>[= Thank You! =]</b>
<p>Athos, Porthos, Aramis & d'Artagnan: <b>[=  Athos, 
Porthos, Aramis & d'Artagnan =]</b> <p>How's life?: <b>[+ 
$epreq->gettext(qq{How's life?}) +]<b>

Re: Multilingual support / was Re: Urgent: MsgIdExtract problem

Reply via email to