From kaloian at doganov.org  Sun Feb  3 06:30:31 2008
From: kaloian at doganov.org (Kaloian Doganov)
Date: Sun, 03 Feb 2008 08:30:31 +0200
Subject: [Po4a-devel] Installing files outside PREFIX
Message-ID: <873asa4kmw.fsf@doganov.org>

The current version of Po4a from CVS by default installs files not only
in /usr/local, but also in /usr/share.  It looks like install procedure
does not honour PREFIX for all files.  This goes far beyond one's
expectation that no files outside the PREFIX tree should be created or
overwritten.



From nicolas.francois at centraliens.net  Thu Feb 14 21:30:54 2008
From: nicolas.francois at centraliens.net (Nicolas =?iso-8859-1?Q?Fran=E7ois?=)
Date: Thu, 14 Feb 2008 22:30:54 +0100
Subject: [Po4a-devel] Deformed XHTML comments spanning multiple lines
In-Reply-To: <87myqpgkgg.fsf@doganov.org>
References: <87myqpgkgg.fsf@doganov.org>
Message-ID: <20080214213054.GB13075@nekral.nekral.homelinux.net>

Hi,

On Tue, Jan 29, 2008 at 09:27:43AM +0200, kaloian at doganov.org wrote:
> I've noticed that XHTML comments that span multiple lines are deformed
> by po4a-translate.

This should now be fixed in the CVS version.

Thanks for reporting this.

Best Regards,
-- 
Nekral


From nicolas.francois at centraliens.net  Thu Feb 14 22:12:01 2008
From: nicolas.francois at centraliens.net (Nicolas =?iso-8859-1?Q?Fran=E7ois?=)
Date: Thu, 14 Feb 2008 23:12:01 +0100
Subject: [Po4a-devel] Turning off wrapping for a specific XHTML tag
In-Reply-To: <20080114004206.GB21896@nekral.nekral.homelinux.net>
References: <87d4sqljm3.fsf@doganov.org>
	<20080114004206.GB21896@nekral.nekral.homelinux.net>
Message-ID: <20080214221201.GC13075@nekral.nekral.homelinux.net>

On Mon, Jan 14, 2008 at 01:42:06AM +0100, nicolas.francois at centraliens.net wrote:
> Hello,
> 
> On Fri, Dec 28, 2007 at 09:11:48PM +0200, kaloian at doganov.org wrote:
> > Using po4a to process XHTML documents, I have to turn of wrapping for a
> > specific XHTML tag (<script>) still considering the containing text as
> > translatable.  Po4a's Xml module provides a way to specify this using
> > the `W' character in the `tags' option:
> > 
> >    -o "tags=W<script>"
> > 
> > Unfortunately this leads to skipping the containing text of <script> as
> > non-translatable.  I can avoid this by adding `tagsonly' option, but
> > this way I practically have to list all possible XHTML tags in `tags',
> > just to turn off wrapping for one of them.
> > 
> > It would be much easier if there was a way to specify this wrap-flag
> > individually for a tag, without having to deal with global
> > skipping/non-skipping of all mentioned tags.
> 
> I agree that the current options are not intuitive and probably do not
> permit what users want.
> 
> I will try to provide another set of options to configure the Xml module
> (and derivative).
> 
> The idea could be to support options like:
> 
>   -o translated="<tag1> w<tag2> W<tag3> <tag4><tag5> i<tag6> I<tag7>"
>   -o verbatim="<tag8> <tag9><tag10>"
> (And remove any tag from tags or inline lists if it is specified by one of
> these options)

That should be OK in the current CVS.
I forgot the name of the second option, so it is not "verbatim", but
"untranslated".

The tags are not removed from the tags and inline lists, but the new
options have an higher priority over the old ones.

i<tag6> and I<tag7> are currently not allowed, but that should not be too
difficult to add.

Cheers,
-- 
Nekral


From nori1 at dolphin.c.u-tokyo.ac.jp  Thu Feb 28 10:10:25 2008
From: nori1 at dolphin.c.u-tokyo.ac.jp (Kobayashi Noritada)
Date: Thu, 28 Feb 2008 19:10:25 +0900 (JST)
Subject: [Po4a-devel] [PATCH] Some patches for the Po4a's Wml module
Message-ID: <20080228.191025.50338403.nori1@dolphin.c.u-tokyo.ac.jp>

Hi,

I am working as a Japanese translator for Debian's web site and
considering whether we can handle translations in wml files with po
files.  I don't think the current way of maintaining web page
translations is good since all the translations must be hard-coded in
the wml files, linked with CVS revisions and block-level markups.  So,
I'm paying attention to the Po4a's Wml module.  I have used Po4a for
translations of release notes and aptitude, and like the idea of that
software. ;-)

Since the Wml module seems to be still immature and somewhat buggy, I
provide some patches here.  Could you please review and commit them?
# Or, shall I participate in this project and commit by myself? ;-)

Descriptions are below:

replacement.patch:
  Correct a mistake: do replacing instead of matching.
  Patch by: Kenshi Muto <kmuto at debian.org>

define-tag.patch:
  Enable handling define-tag tags.
  Patch by: Noritada Kobayashi <nori1 at dolphin.c.u-tokyo.ac.jp>

charset.patch:
  Enable handling input file charset correctly.
  Patch by: Noritada Kobayashi <nori1 at dolphin.c.u-tokyo.ac.jp>

I know these patches are not enough for making the Wml module usable.
I'd love to create more patches and try to make the Wml module
suitable for handling all pages in Debian's web site.

Thanks,

-nori
-------------- next part --------------
--- lib/Locale/Po4a/Wml.pm.orig	2008-02-28 18:48:23.000000000 +0900
+++ lib/Locale/Po4a/Wml.pm	2008-02-28 18:48:35.000000000 +0900
@@ -104,7 +104,7 @@
       }
 
       # Mask mp4h cruft         
-      while ($file =~ m|^#(.*)$|<!--PO4ASHARPBEGIN$1PO4ASHARPEND-->|m) {
+      while ($file =~ s|^#(.*)$|<!--PO4ASHARPBEGIN$1PO4ASHARPEND-->|m) {
         my $line = $1;
         print STDERR "PROTECT HEADER: $line\n";
         if ($line =~ m/title="([^"]*)"/) { #) {#"){
-------------- next part --------------
--- lib/Locale/Po4a/Wml.pm.orig	2008-02-28 18:49:37.000000000 +0900
+++ lib/Locale/Po4a/Wml.pm	2008-02-28 18:51:59.000000000 +0900
@@ -111,6 +111,9 @@
           warn "FIXME: We should translate the page title: $1\n";
         }          
       }
+
+      # Validate define-tag tag's argument
+      $file =~ s|(<define-tag\s+)([^\s>]+)|$1PO4ADUMMYATTR="$2"|g;
                 
       # Flush the result to disk          
       open OUTFILE,">$tmp_filename";
@@ -136,6 +139,7 @@
       $file = join("",@{$xmlizer->{TT}{doc_out}});
       $file =~ s/^<!--PO4ASHARPBEGIN(.*?)PO4ASHARPEND-->/#$1/mg;
       $file =~ s/<!--PO4ABEGINPERL(.*?)PO4AENDPERL-->/<:$1:>/msg;
+      $file =~ s|(<define-tag\s+)PO4ADUMMYATTR="([^"]*)"|$1$2|g;
       $file =~ s/PO4ALT/</msg;
       $file =~ s/PO4AGT/>/msg;
 
-------------- next part --------------
--- lib/Locale/Po4a/Wml.pm.orig	2008-02-28 18:51:59.000000000 +0900
+++ lib/Locale/Po4a/Wml.pm	2008-02-28 18:52:42.000000000 +0900
@@ -123,6 +123,7 @@
       
       # Build the XML TransTractor which will do the job for us
       my $xmlizer = Locale::Po4a::Chooser::new("xhtml");
+      $xmlizer->{TT}{'file_in_charset'}=$self->{TT}{'file_in_charset'};
       $xmlizer->{TT}{po_in}=$self->{TT}{po_in};
       $xmlizer->{TT}{po_out}=$self->{TT}{po_out};
       

From nicolas.francois at centraliens.net  Thu Feb 28 19:04:39 2008
From: nicolas.francois at centraliens.net (Nicolas =?iso-8859-1?Q?Fran=E7ois?=)
Date: Thu, 28 Feb 2008 20:04:39 +0100
Subject: [Po4a-devel] [PATCH] Some patches for the Po4a's Wml module
In-Reply-To: <20080228.191025.50338403.nori1@dolphin.c.u-tokyo.ac.jp>
References: <20080228.191025.50338403.nori1@dolphin.c.u-tokyo.ac.jp>
Message-ID: <20080228190439.GB13991@nekral.nekral.homelinux.net>

Hi,

On Thu, Feb 28, 2008 at 07:10:25PM +0900, nori1 at dolphin.c.u-tokyo.ac.jp wrote:
> 
> Since the Wml module seems to be still immature and somewhat buggy, I
> provide some patches here.  Could you please review and commit them?

I had a look at them, and committed them. Thanks!


> # Or, shall I participate in this project and commit by myself? ;-)

That would be even better, if you intend to make more changes in the Wml
module (or other parts).

Just send me your Alioth login, and I will add you to the po4a project.
(You can create one account on http://alioth.debian.org/account/register.php
if you don't have one already).

Thanks again for the patches,
-- 
Nekral


From nori1 at dolphin.c.u-tokyo.ac.jp  Fri Feb 29 10:59:55 2008
From: nori1 at dolphin.c.u-tokyo.ac.jp (Kobayashi Noritada)
Date: Fri, 29 Feb 2008 19:59:55 +0900 (JST)
Subject: [Po4a-devel] Dealing with series of comments in the XML module
Message-ID: <20080229.195955.106823246.nori1@dolphin.c.u-tokyo.ac.jp>

Hi,

Test data for the XML module contain

<!-- The content of this document is licensed under the XXX license -->
<!-- See bla bla -->

as an input data and

<!-- The content of this document is licensed under the XXX license 
 See bla bla -->

as its corresponding output data.  Does this mean that the conversion
of a series of comments into one comment is an expected behavior?

Since the Wml module temporarily converts '#'-headed comments into an
XML-syntax comments and uses the XML module for parsing, the
conversion of a series of comments results in a bug.  I'd like to know
whether this should be fixed in the XML module or in the Wml
module. ;-)

Many thanks,

-nori


From nicolas.francois at centraliens.net  Fri Feb 29 17:21:37 2008
From: nicolas.francois at centraliens.net (Nicolas =?iso-8859-1?Q?Fran=E7ois?=)
Date: Fri, 29 Feb 2008 18:21:37 +0100
Subject: [Po4a-devel] Dealing with series of comments in the XML module
In-Reply-To: <20080229.195955.106823246.nori1@dolphin.c.u-tokyo.ac.jp>
References: <20080229.195955.106823246.nori1@dolphin.c.u-tokyo.ac.jp>
Message-ID: <20080229172137.GC13991@nekral.nekral.homelinux.net>

Hi,

On Fri, Feb 29, 2008 at 07:59:55PM +0900, nori1 at dolphin.c.u-tokyo.ac.jp wrote:
> 
> Test data for the XML module contain
> 
> <!-- The content of this document is licensed under the XXX license -->
> <!-- See bla bla -->
> 
> as an input data and
> 
> <!-- The content of this document is licensed under the XXX license 
>  See bla bla -->
> 
> as its corresponding output data.  Does this mean that the conversion
> of a series of comments into one comment is an expected behavior?
> 
> Since the Wml module temporarily converts '#'-headed comments into an
> XML-syntax comments and uses the XML module for parsing, the
> conversion of a series of comments results in a bug.  I'd like to know
> whether this should be fixed in the XML module or in the Wml
> module. ;-)

This was currently not an issue in the Xml module, but if it is a limitation
for the Wml module, it can be changed (I don't think any module or user
rely on it).
I think it can be fixed in the translate_paragraph subroutine of the Xml
module (before the first pushline)

I will add a test to the Xml testsuite to check how it evolves for
something like:

    <p>
	<!-- first comment block -->
	<!-- second comment block
	     multiline -->
	paragraph foo
	bar
	<!-- comment inside the paragraph -->
	baz qux
	<!-- comment after the paragraph -->
    </p>

The goal is just to check how it evolves, but if the resulting translated
document changes, and if it looks more like the original comments, then
it's better.
Just update the testsuite when you update the Xml module.



BTW, I will soon make a new release (I hope this weekend), to fix an
important bug (#459451 - po4a failures with perl 5.10).

The Wml module is currently not released in the po4a tarballs, so I will
not really care about this module during this release.
Just let me know when you think it is of an acceptable quality for a
release.  We can plan a later release with Wml (and also translations).

(I also need to translate / remove the French IRC session extract in the
TODO session of the Wml manpage)

Kind Regards,
-- 
Nekral


From nori1 at dolphin.c.u-tokyo.ac.jp  Fri Feb 29 18:30:13 2008
From: nori1 at dolphin.c.u-tokyo.ac.jp (Kobayashi Noritada)
Date: Sat, 01 Mar 2008 03:30:13 +0900 (JST)
Subject: [Po4a-devel] Dealing with series of comments in the XML module
In-Reply-To: <20080229172137.GC13991@nekral.nekral.homelinux.net>
References: <20080229.195955.106823246.nori1@dolphin.c.u-tokyo.ac.jp>
	<20080229172137.GC13991@nekral.nekral.homelinux.net>
Message-ID: <20080301.033013.96920753.nori1@dolphin.c.u-tokyo.ac.jp>

Hi,

From: Nicolas Fran?ois <nicolas.francois at centraliens.net>
Subject: Re: [Po4a-devel] Dealing with series of comments in the XML module
Date: Fri, 29 Feb 2008 18:21:37 +0100

> > Test data for the XML module contain
> > 
> > <!-- The content of this document is licensed under the XXX license -->
> > <!-- See bla bla -->
> > 
> > as an input data and
> > 
> > <!-- The content of this document is licensed under the XXX license 
> >  See bla bla -->
> > 
> > as its corresponding output data.  Does this mean that the conversion
> > of a series of comments into one comment is an expected behavior?
> > 
> > Since the Wml module temporarily converts '#'-headed comments into an
> > XML-syntax comments and uses the XML module for parsing, the
> > conversion of a series of comments results in a bug.  I'd like to know
> > whether this should be fixed in the XML module or in the Wml
> > module. ;-)
> 
> This was currently not an issue in the Xml module, but if it is a limitation
> for the Wml module, it can be changed (I don't think any module or user
> rely on it).
> I think it can be fixed in the translate_paragraph subroutine of the Xml
> module (before the first pushline)

Thank you for polite suggestions!  I'll try to fix.

> The goal is just to check how it evolves, but if the resulting translated
> document changes, and if it looks more like the original comments, then
> it's better.
> Just update the testsuite when you update the Xml module.

All right.

> The Wml module is currently not released in the po4a tarballs, so I will
> not really care about this module during this release.
> Just let me know when you think it is of an acceptable quality for a
> release.  We can plan a later release with Wml (and also translations).
> 
> (I also need to translate / remove the French IRC session extract in the
> TODO session of the Wml manpage)

Sure.  I'm looking forward to give you good news. ;-)

Thanks,

-nori