Translating ODF files on Pootle

l10n
#1

Hi all,

As promised, I'm going on with the tests concerning the ability to use Pootle in the translation workflow for documentation. As you may know, Pootle is able to handle very large projects now thanks to the great work the team has provided. There is all the tools needed for the translation process, but also you can assign task and objectives to a member or a team.

I've worked on the first chapter of the OOoAuthors guide for beginners. I used WordForge filters [1] to convert the file from .odt (ODF file format) to .xlf (XLIFF file format). Both are standards, so it should be easier to have a good conversion. Several tools manage off line translation for Xliff file format, like WordForge or OmegaT, there may be others I don't know too.

Rimas has kindly uploaded the file on Pootle [2], creating a new project for several languages for the beginning. Currently, en_GB fr hu ko nl oc pt_BR ru zh_CN . Those are the 7 languages for which libo33 project is fully localized plus hu and nl.
The issue we meet is that the strings appears as needing review instead of being empty.

If other team wants to give a try, please follow the same process as for the extensions project, post on the mailing list and Rimas (big thanks to him :slight_smile: will take care of your request.

Please, follow up on the l10n list for the moment, we will settle a more precise workflow once we have solved the issued that may raised and that we are sure most (I know we miss an offline tool currently for Mac translators) of the needs to work quietly and efficiently on the documentation translation.

Thanks for your help in improving the process :slight_smile:

[1] https://sourceforge.net/projects/wordforge2/
[2] http://translations.documentfoundation.org/projects/doc_test/

Kind regards
Sophie

#2

Hi Sophie, Rimas,

Hi all,

As promised, I'm going on with the tests concerning the ability to use
Pootle in the translation workflow for documentation. As you may know,
Pootle is able to handle very large projects now thanks to the great work
the team has provided. There is all the tools needed for the translation
process, but also you can assign task and objectives to a member or a team.

I've worked on the first chapter of the OOoAuthors guide for beginners. I
used WordForge filters [1] to convert the file from .odt (ODF file format)
to .xlf (XLIFF file format). Both are standards, so it should be easier to
have a good conversion. Several tools manage off line translation for Xliff
file format, like WordForge or OmegaT, there may be others I don't know too.

Rimas has kindly uploaded the file on Pootle [2], creating a new project for
several languages for the beginning. Currently, en_GB  fr  hu  ko nl  oc
 pt_BR  ru  zh_CN . Those are the 7 languages for which libo33 project is
fully localized plus hu and nl.

I would like to test the process for de - if that is possible. :slight_smile:

Thanks.

Sigrid

#3

Hi,

Pootle does not handle placeables. I translated a segment in Pootle and
placeables are not there in the translated segment. It will cause problems
when you convert the translation back to ODT.

      <trans-unit xml:space="preserve"
id="office:document-content[0]/office:body[0]/office:text[0]/text:p[2]"
approved="yes">
        <source><g id="1"/>Chapter <g id="2">1</g><g id="3"/> <g
id="4"/>Introducing LibreOffice</source>
      <target state="translated">1. fejezet Bevezetés a
LibreOffice-ba</target></trans-unit>

See the <g> tags are missing.

When I translated this segment offline and uploaded the translation to
Pootle, it inserted an extra \n at the beginning of the segment.

Best regards,
Andras

#4

Hi Sigrid,

2010.12.26 19:40, Sigrid Carrera rašė:

I would like to test the process for de - if that is possible. :slight_smile:

de enabled.

Rimas

#5

2010.12.26 19:40, Andras Timar rašė:

Hi all,

As promised, I'm going on with the tests concerning the ability to use
Pootle in the translation workflow for documentation. As you may know,
Pootle is able to handle very large projects now thanks to the great work
the team has provided. There is all the tools needed for the translation
process, but also you can assign task and objectives to a member or a team.

I've worked on the first chapter of the OOoAuthors guide for beginners. I
used WordForge filters [1] to convert the file from .odt (ODF file format)
to .xlf (XLIFF file format). Both are standards, so it should be easier to
have a good conversion. Several tools manage off line translation for Xliff
file format, like WordForge or OmegaT, there may be others I don't know too.

Rimas has kindly uploaded the file on Pootle [2], creating a new project
for several languages for the beginning. Currently, en_GB fr hu ko nl oc
  pt_BR ru zh_CN . Those are the 7 languages for which libo33 project is
fully localized plus hu and nl.
The issue we meet is that the strings appears as needing review instead of
being empty.

If other team wants to give a try, please follow the same process as for
the extensions project, post on the mailing list and Rimas (big thanks to
him :slight_smile: will take care of your request.

Please, follow up on the l10n list for the moment, we will settle a more
precise workflow once we have solved the issued that may raised and that we
are sure most (I know we miss an offline tool currently for Mac translators)
of the needs to work quietly and efficiently on the documentation
translation.

Thanks for your help in improving the process :slight_smile:

[1] https://sourceforge.net/projects/wordforge2/
[2] http://translations.documentfoundation.org/projects/doc_test/

Hi,

Pootle does not handle placeables. I translated a segment in Pootle and
placeables are not there in the translated segment. It will cause problems
when you convert the translation back to ODT.

       <trans-unit xml:space="preserve"
id="office:document-content[0]/office:body[0]/office:text[0]/text:p[2]"
approved="yes">
         <source><g id="1"/>Chapter<g id="2">1</g><g id="3"/> <g
id="4"/>Introducing LibreOffice</source>
       <target state="translated">1. fejezet Bevezetés a
LibreOffice-ba</target></trans-unit>

See the<g> tags are missing.

When I translated this segment offline and uploaded the translation to
Pootle, it inserted an extra \n at the beginning of the segment.

Hm, that seems like a bug in Pootle. I guess it should be reported to the authors...

Rimas

#6

Hi Rimas,

Hi Sigrid,

2010.12.26 19:40, Sigrid Carrera rašė:

I would like to test the process for de - if that is possible. :slight_smile:

de enabled.

Thank you.

Sigrid

#7

Hi Andras,

2010.12.26 19:52, Rimas Kudelis rašė:

2010.12.26 19:40, Andras Timar rašė:

Pootle does not handle placeables. I translated a segment in Pootle and
placeables are not there in the translated segment. It will cause problems
when you convert the translation back to ODT.

<trans-unit xml:space="preserve"
id="office:document-content[0]/office:body[0]/office:text[0]/text:p[2]"
approved="yes">
<source><g id="1"/>Chapter<g id="2">1</g><g id="3"/> <g
id="4"/>Introducing LibreOffice</source>
<target state="translated">1. fejezet Bevezetés a
LibreOffice-ba</target></trans-unit>

See the<g> tags are missing.

When I translated this segment offline and uploaded the translation to
Pootle, it inserted an extra \n at the beginning of the segment.

Hm, that seems like a bug in Pootle. I guess it should be reported to the authors...

http://bugs.locamotion.org/show_bug.cgi?id=1811

Feel free to add your comments.

Rimas

#8

Hello Sophie, Rimas, Andras

I am glad to see that we are making good use of our pootle server.

I have a concern however.

I think it is time to load all our UI and HelpContent strings into Pootle. Not only because we, as translators, are always trying to improve our work, but also if we will use pootle to translate documentation, it may be interesting to use or peek on the help and UI strings.

That's it.

Kind Regards
Olivier

#9

Hi Sophie,

The ODF2XLIFF filters in both translate toolkit and WordForge are broken. They ignore some strings from the source (they do not go into the XLIFF files) in which there is mark-up. They can be used only for plain text without any markup, and even them segmentation is not very good. There are also other problems with them. I-tools does not have enough working tools either.

Our testing shows that the Okapi framework does a good job for the round trip, I don't know about other tools.

This is a very important issue, but not easy to solve. We will be putting some time to fix it, but it will not happen immediately.

Cheers,

Javier

Sophie Gautier wrote:

#10

Hi Javier,

Hi Sophie,

The ODF2XLIFF filters in both translate toolkit and WordForge are
broken. They ignore some strings from the source (they do not go into
the XLIFF files) in which there is mark-up. They can be used only for
plain text without any markup, and even them segmentation is not very
good. There are also other problems with them. I-tools does not have
enough working tools either.

Ok, I didn't test enough to be aware of the errors, the round trip I made seems to be ok, but thanks a lot for your feedback.

Our testing shows that the Okapi framework does a good job for the round
trip, I don't know about other tools.

Ok, good to know, thanks, I'll test it now for the conversion.

This is a very important issue, but not easy to solve. We will be
putting some time to fix it, but it will not happen immediately.

I understand. Thanks a lot for the work you have already done for us :slight_smile:

Kind regards
Sophie

#11

Hi Sophie, all,

Hi Javier,
> Hi Sophie,
>
> The ODF2XLIFF filters in both translate toolkit and WordForge are
> broken. They ignore some strings from the source (they do not go
> into the XLIFF files) in which there is mark-up. They can be used
> only for plain text without any markup, and even them segmentation
> is not very good. There are also other problems with them. I-tools
> does not have enough working tools either.

Ok, I didn't test enough to be aware of the errors, the round trip I
made seems to be ok, but thanks a lot for your feedback.
>
> Our testing shows that the Okapi framework does a good job for the
> round trip, I don't know about other tools.

Ok, good to know, thanks, I'll test it now for the conversion.
>
> This is a very important issue, but not easy to solve. We will be
> putting some time to fix it, but it will not happen immediately.
>
I understand. Thanks a lot for the work you have already done for
us :slight_smile:

Sorry to jump in late, but would this platform help?
http://www.hforge.org/odf-i18n-tests

Best,

#12

Hi Charles,

Hi Sophie, all,

Hi Javier,

Hi Sophie,

The ODF2XLIFF filters in both translate toolkit and WordForge are
broken. They ignore some strings from the source (they do not go
into the XLIFF files) in which there is mark-up. They can be used
only for plain text without any markup, and even them segmentation
is not very good. There are also other problems with them. I-tools
does not have enough working tools either.

Ok, I didn't test enough to be aware of the errors, the round trip I
made seems to be ok, but thanks a lot for your feedback.

Our testing shows that the Okapi framework does a good job for the
round trip, I don't know about other tools.

Ok, good to know, thanks, I'll test it now for the conversion.

This is a very important issue, but not easy to solve. We will be
putting some time to fix it, but it will not happen immediately.

I understand. Thanks a lot for the work you have already done for
us :slight_smile:

Sorry to jump in late, but would this platform help?
http://www.hforge.org/odf-i18n-tests

Thank you for the link Charles, I was not aware of this site.
From what Javier said, the filter of the translate toolkit is also buggy and I think this is the one they use. But I'll check later. In the mean time I'll test this evening the Okapi converter.
Kind regards
Sophie

#13

Hi,

2010.12.27 10:37, Javier Sola rašė:

The ODF2XLIFF filters in both translate toolkit and WordForge are broken. They ignore some strings from the source (they do not go into the XLIFF files) in which there is mark-up. They can be used only for plain text without any markup, and even them segmentation is not very good. There are also other problems with them. I-tools does not have enough working tools either.

Are you sure? There were plenty of placeholders in the file exported by Sophie. To me, it looks like most of the problem is actually in Pootle itself, because it simply ignores XML tags in the translatable string.

By the way, I haven't really seen the original document before conversion, but I wonder how odf2xliff actually split the file into strings. In my opinion, splitting by anything less than a paragraph (or maybe even a chapter) reduces flexibility and may be undesired, no? IMO, localization of documentation and help, unlike that of UI, is a bit more creative task, and to better convey the idea, it may sometimes be desirable to shuffle sentences, and perhaps even entire paragraphs, around. Could anyone comment on that?

Rimas

#14

Please also enable zh_TW => Chinese (Taiwan).

Thanks!

#15

Hi,

2010.12.27 16:10, Tseng, Cheng-Chia rašė:

Please also enable zh_TW => Chinese (Taiwan).

zh_TW enabled. Though the problems found make me think we won't be using Pootle for documentation translation this time.

May I suggest to work on the Extensions project first? It's a mere 100 words, and would definately be used. :slight_smile:

Rimas

#16

Hi Rimas,

We have been looking into this for a while. We tried to use it to translate some text books.

We were considering integrating it Odf2Xliff in WordForge, but in the toolkit it has old (2009) known bugs that have not been solved. We need to find another solution.

We have looked into it, same as other filters that are not 100% correct, but in this case it is not easy to solve, given the way ODF files are built. You might have an in-line tag that is a reference to an internal tag of the document, that finally might come out to be just marking text as "bold", but you have to do a recursive analysis to get there.. not obvious... and then you have to build it back when returning to the original translated text.

Cheers,

Javier

Rimas Kudelis wrote:

#17

Hi Sophie,

I have just one question regarding the document you used:

Hi all,

[...]

I've worked on the first chapter of the OOoAuthors guide for beginners. I
used WordForge filters [1] to convert the file from .odt (ODF file format)
to .xlf (XLIFF file format). Both are standards, so it should be easier to
have a good conversion. Several tools manage off line translation for Xliff
file format, like WordForge or OmegaT, there may be others I don't know too.

Did you use the OOo version or the file, that was already adapted to
LibO? The latter can be found in the tdf wiki on this page:
http://wiki.documentfoundation.org/Documentation

Chapter 1 is considered to be finished, so it is already offered as
pdf as well. :wink:

I just noticed some paragraphs, that weren't in the LibO version like
the explanation that OpenOffice is a trademark owned by someone else
and so on.... So, if we were to use Pootle for translation I think it
would be useful to use the already adapted file as our sourcefile.

Sigrid

#18

Hi Javier,

Is there a chance, that you can offer either the Source code of
WordForge (so that I can use configure, make, make install) or a rpm
package of WordForge? I was looking into installing alien to convert
deb into rpm but with all the dependencies (I don't have my computer
set up for any development stuff) the download will take a few days (I
have a really crappy internet connection).

Thanks in advance.

Sigrid

#19

Hi Sigrid,

Hi Sophie,

I have just one question regarding the document you used:

Hi all,

[...]

I've worked on the first chapter of the OOoAuthors guide for beginners. I
used WordForge filters [1] to convert the file from .odt (ODF file format)
to .xlf (XLIFF file format). Both are standards, so it should be easier to
have a good conversion. Several tools manage off line translation for Xliff
file format, like WordForge or OmegaT, there may be others I don't know too.

Did you use the OOo version or the file, that was already adapted to
LibO? The latter can be found in the tdf wiki on this page:
http://wiki.documentfoundation.org/Documentation

Chapter 1 is considered to be finished, so it is already offered as
pdf as well. :wink:

I just noticed some paragraphs, that weren't in the LibO version like
the explanation that OpenOffice is a trademark owned by someone else
and so on.... So, if we were to use Pootle for translation I think it
would be useful to use the already adapted file as our sourcefile.

I'm really sorry but I think I missed it when I begin the tests last month. I'm working very slowly on this because it needs comparison at several stages and my time is quite short. So I may have begin before you upload the updated files.
Of course for the real translation, we will get in synch with OOoAuthors production (or the new team name :), currently it's really for testing because as you see, we have to make sure that both the converter and Pootle handle the file correctly. The most important for the moment is to see how the xliff format is handled and converted back.

But thank for you help and your support Sigrid and again sorry for not noticing the last version of your work.

Kind regards
Sophie

#20

Hi!

I guess it's time for asturian team to take care of this one :slight_smile:

Please, Rimas, can you activate ast language for us?

Best regards,