Re: [bug-gettext] xgettext and messages from database

[Daiki Ueno]

Aljosha Papsch <address@hidden> writes:

> The remaining pieces are messages from a MySQL database and for that I
> would like to know your opinion about the best approach. On the web is
> an example[1] which writes translatable messages directly to a .po
> file.

Actually, the example does not seem to produce a .po file, but a .pot
file, which can be used as an input of xgettext.  That sounds a
reasonable solution, since neither YAML nor JSON has a standard
mechanism to mark strings translatable.

> That's certainly doable for me, though surely it's better to let
> xgettext be the sole creator of .po files instead of a home brew
> solution? Instead, I could imagine writing to a JSON file that simply
> lists the messages like this:
>
> [
>   "Greece",
>   "Nice and sunny",
>   "Spain",
>   "Just as nice and sunny"
> ]
>
> This file can be consumed by xgettext because JSON is a subset of
> JavaScript.

If you simply extract all strings in a JSON file, it is possible with
xgettext -L JavaScript -a.  However, if you want to extract certain
strings from a JSON file, I'd suggest to use an intermediate .pot file
as above, using something like JSONPath.

Regards,
-- 
Daiki Ueno