Same English phrase - different Swedish phrases

[philip@…]

The same English word has two different translations, depending on context. Is this possible to fix somehow?

A specific example:

The English word "March" is used twice in the file utils/dates.py has two different translations when translated to Swedish.
On line 18 "March" is marked for translation in the context of being the full month name.
Next, on line 31 "March" is in the context of being in the Associated Press style.
For a Swedish translation, I'd like to have the AP style months abbreviated, but not the full month names.
I don't know how to solve this.

Another example:
The word "one" in contrib/humanize/templatetags/humanize.py on line 68 is translated to "en" or "ett" (Swedish) depending context.
This is probably pretty hard to solve.

[Ludvig Ericson <ludvig.ericson@…>]

As for numbers, the number one is 'en etta,' or 'ett' and I'd bet you know this. But yes, there's an issue when translating to languages which have neutrum and utrum.

[Malcolm Tredinnick]

There's nothing we can really do about these. MO file lookup works by byte matching, so the same string of bytes/characters cannot be translated in two different ways.

All you can do in those cases, unfortunately, is recommend that people don't use the Associated Press versions in their non-English text output (which kind of makes sense anyway -- AP's style guide is very North American-centric).

Particularly the apnumber filter is something we can't solve. Filters can't really look ahead and change their behaviour based on the following text (which they won't even have access to, in many cases). Again, sites requiring localisation into languages with gender-specific word endings are going to have to use more specialised filters in those cases.