Re: [libreoffice-l10n] Question about odd use of msgctxt

On Tue, Jul 10, 2012 at 2:37 AM, Rimas Kudelis <rq@akl.lt> wrote:

> I believe there have been thoughts about migrating to gettext properly, but
> considering the amount of work it would take and the impact, no wonder it
> hasn't happened (or perhaps even started) yet.
>
> Rimas

Thanks for the answer Rimas.  I do understand the challenges of
working solely in gettext with a project as large and diverse as
LibreOffice.

It is nonetheless unfortunate that it is necessary to hack the msgctxt
comment to achieve your L10n.  Not so much for the extra burden it
places on localizers, which can largely be mitigated by working in
Virtaal with it's translation memory (for instance); but for it's
impact on other string comparison tools like Translate Toolkit.

I came across this during a process of doing cross-project consistency
checks for conflicting translations.  I'm a scientist by training and
data pipelining quality checks is just one of those things I feel
compelled to do. I realized that one of my usual tricks was less
effective than it usually is, however, it still yields some useful
results.

Here is the trick.

1) Download the zip of the entire LibreOffice 3.6 – UI project for a
given language as a "Zip of Directory" from the top of the Translate
tab for a given language.

2) Place it in it's own folder, unzip it, place the collapse36.sh file
in that folder, at the same level as the large group of folders from
the zip file.

3) Run the shell script, which collapses the folder hierarchy, moving
all PO files into a new folder called po and deletes the other folders
as it goes along to clean up.  it renames files with common names by
adding numbers to make them all uniquely named.

sh collapse36.sh

4) You need the Translate toolit for this.  Run pocompendium on the
collected PO files.

pocompendium -out.po -d po

5) Then run posplit.

posplit out.po

6) Examine the compendium conflicts in the out-fuzzy.po file and you
will often find errors in need of correction.  Not all cases will
genuine problems, but most are worth investigating. It may reveal the
need for a msgctxt comment (as traditionally used) to distinguish an
ambiguous English word that could use more context ot achieve optimal
translation.

If you find references to files with numerical additions (e.g.
src3.po) use the text of collapse36.sh to map back to it's original
name and location in the file hierarchy.

Unfortunately because pocompendium respects the location-based msgctxt
comments LO uses extensively, this is less effective than it could be,
but it still identifies opportunities for improving consistency of
translation across the project.

Regards,

cjl

-- 
Unsubscribe instructions: E-mail to l10n+help@global.libreoffice.org
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://listarchives.libreoffice.org/global/l10n/
All messages sent to this list will be publicly archived and cannot be deleted