We need a script that retrieves message files from our Subversion repository, sends these files to [url removed, login to view] for translation and then puts the translated files back into subversion at the right place.
The translated files also require some automatic QA checking for possible issues introduced by the human translators.
The script should run daily, so that translations are automatically created if there are more strings for translation.
The issue is that mygengo does not provide a public API for their "String" translation interface, so you need to develop a scrapper/extraction script that goes through their user interface. It might be a good idea to use a Selenium script for interaction with mygengo, but of course you can also do it with curl.
Detailed requirements are in the attached document.