Translating documentation requires some special considerations which do not apply when translating programs. This page contains instructions for translators and an explanation of the toolchain used to convert the documentation into translation templates and back again.

Ubuntu documentation uses Rosetta as a translation portal, as with the rest of Ubuntu. For explanations of how to use Rosetta, see the Rosetta page.

The documentation can be found in the following packages: 'ubuntu-docs', 'kubuntu-docs', 'edubuntu-docs' and 'xubuntu-docs'.

There are a number of specific considerations to bear in mind when translating documentation.

Failure to observe these instructions will result in the translated document breaking.

1. Tags should not be translated. Where you see a tag in the translation (for example: «'<xref linkend=«free-software»/>'»), the tag and anything inside the tag itself should not be translated. It is very important to copy and paste these tags exactly as they appear in this English. The exception to this is where a tag contains a URL where there is an equivalent translated URL. In those cases the translator should use discretion about whether to localise the URL.
2. Entities should not be translated. Where you see a phrase expressed like «'&gt;'», do not change this.
3. Always maintain the order of tags. When you see different tags nested inside each other (for example: «'<menuchoice><guimenu>System</guimenu><guimenuitem>Administration</guimenuitem><guimenuitem>Users and Groups</guimenuitem></menuchoice>'»), you must always preserve that order carefully - copying and pasting from the English is the best way to ensure that you do this correctly. Only translate the words inside the tags.
4. For full English copies of the various guides to give you context while translating, click System → Help and Support in Ubuntu, or Khelpcenter in Kubuntu. Alternatively, all the guides are available to browse at http://doc.ubuntu.com
5. The documents are not automatically imported into Ubuntu (like the translation of programs). They need to be exported and uploaded manually to the packages. The documentation team will take care of this on occasions.

If you have any questions, please ask on the ubuntu-translators mailing list: or the ubuntu-doc mailing list.

You can now test the translations you've done in Rosetta by inserting them into our repository and viewing them in Yelp.

1. Download the LANG.po file for a particular document from Rosetta. For example the relevant templates for Intrepid can be found here. These instructions use the example of the french (fr) translation of the 'internet' document.
2. Get the Intrepid branch of our repository:

sudo apt-get install bzr xmlto gettext gnome-doc-utils
bzr checkout --lightweight lp:ubuntu-doc/intrepid ubuntu-docs-intrepid

1. Browse to the documentation directory

cd ubuntu-docs-intrepid

1. Insert the new po file in the relevant document folder, under the 'po' subfolder using the naming scheme LANG.po:

mv ~/Desktop/internet-fr.po internet/po/fr.po

1. Run the translate script and choose the relevant document and language:

./scripts/translate.sh -d internet -l fr

1. The script will update the docbook xml for your document and language and will check the file created for validity. It will print any errors as output.
2. Open the file in the GNOME help viewer, called yelp:

yelp file://${PWD}/internet/fr/internet.xml

Yelp always requires the full path to the xml document.

To build all documents with the latest translation run:

for doc in $(cat libs/shipped-docs); do ./scripts/translate.sh -d $doc -l LANG; done

A html version be created by running:

mkdir -p build/html
for doc in $(cat libs/shipped-docs); do find ${doc}/LANG -name "*.xml" -exec xmlto -o build/html html-nochunks {} \;; done

A nice formated html version (similar to help.ubuntu.com) be created by running:

for doc in $(cat libs/shipped-docs); do
     xsltproc --xinclude -o build/html/$doc/$lang/index.html \
	libs/ubuntu-html-chunk-cust.xsl $doc/$lang/$doc.xml 
done

Create a copy of libs/ubuntu-html-chunk-cust.xsl into libs/ubuntu-html-chunk-cust-$lang.xsl to customize/localise it's content. Also check ubuntu-banner.xsl.

LANG should be replaced by your language code.

NOTE: if you wish to make any changes to the translation, you should do so in Rosetta.

The rest of this page is a description of how the Documentation Team work gets translated and inserted into Ubuntu. It is primarily intended for reference of the Documentation Team: translators do not need to know all this information.

To install all packages required to build ubuntu-docs, including translations, run:

apt-get build-dep ubuntu-docs

There are three steps to getting documentation translated:

1. generating a translation template
2. letting the translators work their magic
3. importing the translations back into our repository

These are taken in turn below.

• When an English XML document is ready for translation, we make a POT file, which is the template that translators use. To make this, we use the command:

xml2po -e -o output.pot input.xml

• For documents spread over more than one file, these can be included as follows:

xml2po -e -o output.pot chapter1.xml chapter2.xml

• IMPORTANT: For documents which have an omf file, these should also be included in the pot file.
• As a result, generally the appropriate command is:

xml2po -e -o output.pot C/*.xml C/*.omf

NOTE: We have an automatic script to generate POT files, called 'get-pot.sh'. Running this script refreshes the POT files for all documents.

• The pot file should then be uploaded to rosetta. This is done automatically if the pot is in a source deb package uploaded into the current version of Ubuntu, if not, it must be done manually at the relevant template page in Rosetta.
• Translators then do the rest!

• Automatic exporting from Rosetta into the docteam tree is not implemented. So here is what happens.
• When a good time comes to import the translations, the translations can be downloaded from rosetta (for example, the «Download Translations link here). You can either download translations for all documents or for documents individually.
• The translations from the tarball are placed in the relevant document's directory.
• The translation script (translate.sh) is then used to convert the translations back to xml in the appropriate directory.
• The script also checks the resulting xml for validity, and there are generally errors that must be corrected which are displayed onscreen. This must be done manually.

That's it!

На главную страницу переводчиков