Mixed language documents/pages

[TinoDidriksen]

apertium/apertium-webext#2 made me think of how we can handle mixed language content.

In GramTrans, part of the translation pipe for documents (including websites) is a detection program that discards paragraphs not in the expected source language. Discarding means that the final "replace original with translation" step simply does nothing for that paragraph, which leaves the original untranslated. We should do the same in APy, when the user has chosen a specific source language.

But one can go further. When the user has chosen to autodetect input language, for each distinct detected source language, translate those chunks to the target language. Ensure that as much as possible of the document ends up as the desired target language.