The purpose of this bot is to import (and later, export) translated texts. It uses the Pywikipediabot framework - in particular, the pagefromfile.py script (the first runs will just consist of invoking this script).
The background: Under the Translate Extension that was activated last year here on Meta, translated texts are split into translation units, each represented by a subpage (like this: 1, 2, 3, ...), which are automatically integrated into a page representing the whole translation (example). In some cases such as the soon to be launched new editor survey, which is going to be provided in at least 11 languages, we need to import translations from other sources, both existing translations here on Meta (many questions and answers have already been translated for the last survey) and external translations which are to be reviewed by the community using the Translate extension. I have talked to some of the extension's developer team and they agree that such a bot seems currently the only viable solution for this kind of job (at least until a possible future enabling of support for the gettext format).
In short, the bot will read translated texts from a file and create a subpage for each translation unit. Later, it will also export translated texts by reading out those subpages sequentially.
The bot has completed a small test run on testwiki. In the beginning, it is going to operate automated but supervised; if things go well I'd like to switch to automated unsupervised.
Happy to answer any further questions.
Regards, Tbayer (WMF) (talk) 07:23, 16 August 2012 (UTC)
This bot will no doubt be extremely useful if it is successful in importing as you have suggested, especially with potentially moving the old translations using the old system onto Meta-Wiki. As such, I'm more than happy to give you a trial run for some bot-assisted importing. As you have mentioned this year's editor survey, I think that would be a good trial to see the effectiveness of this bot and to find any potential problems that it may have. Please keep us updated with your progress, I'm sure many people participating in translations would be interested to see how you get on. TheHelpfulOne 07:37, 16 August 2012 (UTC)
I don't fully understand the third paragraph (In short...). Translation units are already in the Translations: namespace; so why do we need a subpage? Regards, -- MarcoAurelio (talk) 16:23, 16 August 2012 (UTC)
I was referring to the usual format in which the Translate extension stores translated texts (in the Translations: namespace, as you say), i.e. pages of the form 1, 2, 3. Apologies if calling them "subpages" was confusing. Regards, Tbayer (WMF) (talk) 17:33, 16 August 2012 (UTC)
Fine - No problem & thanks for the explanation. I'm also in favor of a test run before final approval. -- MarcoAurelio (talk) 18:59, 16 August 2012 (UTC)
I see the bot has been running (test?) and does not have bot flag yet. Before it runs more, it needs bot flag so it doesn't flood recent changes. (and I support the bot request and approval, other than bot flag it looks good) Aude (talk) 07:01, 17 August 2012 (UTC)
Yes, the bot has now done a test run without bot flag, as suggested above. There haven't been any problems during this trial and the result (importing one part of this and this translation) seems fine. I can do some more testing and checking if required, but I'm already confident that the bot's code won't cause any problems. Regards, Tbayer (WMF) (talk) 07:30, 17 August 2012 (UTC)
I just did a quick check of the two pages and noticed a minor error. Looking at the French version, if you look at what the "Priorities" looked like before translation and now after translation, you can see that there appears to be double letters "b b", "c c", "d d" etc, although this didn't happen with the "a" part of that list. Actually, looking at the latest Spanish version the same problem appears to have occurred, again only for the "Priorities" section, but this time there is a double "a a". TheHelpfulOne 10:48, 17 August 2012 (UTC)
Thanks for spotting this! But it doesn't have anything to do with the bot itself. As one can see from the <translate> ...<translate> tags in the source of the English original of the translated text, the "a ", "b ", "c " prefixes had been excluded from translation. But they somehow slipped back into the text that was prepared (manually) as the basis for these translations, so they now show up twice - once in the untranslated part, once in the translated part (e.g. this page added by the bot contains only one "b "). Fixing such issues is after all the purpose of importing the translations ;)
In other words, the bot itself appears to have worked fine with those pages too. Regards, Tbayer (WMF) (talk) 19:27, 17 August 2012 (UTC)
Ah, thanks, that explains it, so it was a problem with the translation, not the bot code and as such I am happy to approve the bot and set the flag, which I will do now. TheHelpfulOne 22:06, 17 August 2012 (UTC)
Successful request: Test run showed no errors in the bot's code, will be run by a trusted bot operator for an invaluable task. Also flagging as a bot to avoid spamming the RC feed. TheHelpfulOne 22:10, 17 August 2012 (UTC)