Re: [Nuxeo-localizer] easily translate existing website

[Jan Nieuwenhuizen]

Juan David Ibáñez Palomar <address@hidden> writes:

> That's what I did with the EuroPython.org web site, it was
> already on production when its internationalization started.

Yes, I understand that.  But what technique was used?

> If you develop a multilingual application from scratch it's
> likely it will have a clean implementation.

Ok, but using what technique.  When do you translate whole html pages,
and when would you translate headings, titles, buttons and paragraphs
separately.

> If it's already in production you probably will prefer to avoid the
> most intrusive techniques (LocalContent/LocalPropertyManager) in
> favor of other approaches (MessageCatalog).
>
> Typically the message catalog should be used for the user
> interface (buttons, labels, etc..) and LocalContent for the
> data.

Ok.  But do you put entire pages into LocalContent, or build pages
from smaller LocalContent pieces?

> If you want a better answer try to describe how your web site
> is.

Part of the website will be redone from scratch.  But an important
part is a user manual, that is already written, and has source outside
of html.  From time to time, a new master version of the manual will
be produced.  Here's the first interesting page of that manual:

    
http://lilypond.org/src/zope/Documentation/user/out-www/lilypond/First-steps.html

The pages are fairly simplistic, static html, so we *could* translate
just page by page (putting every page in LocalContent, I guess?).  But
this raw html is no fun for a translator, and would make the
translation quite unmaintainable, if a new version of the manual is
produced.  The other way that I see, is to mark every heading, link
name, and paragraph.  Then you could simply send the translator a .po
file, and be better prepared for changes is the master document.

In the case of the above page, something like:

    ...
    <h3><dtml-var "msg.gettext ('First steps')"</h3>

    <dtml-gettext catalog=msg>
    <p>We start off by showing how very simple music is entered in
    LilyPond:  you get a note simply by typing its note name, from
    <code>a</code> through <code>g</code>. So if you enter
    </dtml-gettext>
    <blockquote>
    <br><pre>c d e f g a b
    </pre>
    </blockquote>
    <dtml-gettext catalog=msg>
    then the result looks like this:
    </dtml-gettext>
    <p>
    <a href="lily-1452586523.png">
    <img border=0 src="lily-1452586523.png" alt=[<dtml-var "msg.gettext 
('picture of music')">
    </a><p>
    </blockquote>
    <p></p><br><br>
    
    <dtml-gettext catalog=msg>
    <p>We will continue with this format: first we show a snippet of input,
    then the resulting output.
    </dtml-gettext>
    ...
        
But how are we going to get all those tags in, and where?  Is there a
handy tool for that?

Do you tag per paragraph, or use bigger chunk of texts?  For example,
look at this page:

   
http://lilypond.org/src/zope/Documentation/user/out-www/lilypond/About-this-manual.html

Where would you put the tags?

Greetings,
Jan.

-- 
Jan Nieuwenhuizen <address@hidden> | GNU LilyPond - The music typesetter
http://www.xs4all.nl/~jantien       | http://www.lilypond.org