pybabel extract behaviour

[dmitriid]

On the heels of #71 and #296 I decdied to document some of the oddities I ecnountered when using pybabel CLI.

This is the test dist/gettext.js file I'm using

var x = __("sing")
var z = __n("sing1", "plu1")
var y = __n("sing2", "plu2", 10)
var zzz = _("aaa")

1. pybabel extract ignores babel.cfg (docs are incorrect?)

According to docs, babel.cfg should look like this:

[javascript: **.js]
extract_messages = __

Run:

$ pybabel extract -F ./babel.cfg -o dist/lang/en.pot ./dist/gettext.js
extracting messages from ./dist/gettext.js (extract_messages="__")
writing PO template file to dist/lang/en.pot

It will only extract messages surrounded by _() (single underscore)

2. Ignores keywords in babel.cfg

Google + friends suggest that this is the correct babel.cfg:

[javascript: **.js]
keywords = __

Run:

$ pybabel extract -F ./babel.cfg -o dist/lang/en.pot ./dist/gettext.js
extracting messages from ./dist/gettext.js (keywords="__")
writing PO template file to dist/lang/en.pot

Result: only messages surrounded by _() are extracted (single underscore)

3. pybabel extract ignores -keywords parameter

Run $ pybabel extract --keywords=__ -o dist/lang/en.pot ./dist/gettext.js writing PO template file to dist/lang/en.pot

Result: no strings are extracted

4. pybabel extract requires babel.cfg even though it ignores its settings

Given the babel.cfg from above, run:

$ pybabel extract -F ./babel.cfg --keywords=__ -o dist/lang/en.pot ./dist/gettext.js
extracting messages from ./dist/gettext.js (keywords="__")
writing PO template file to dist/lang/en.pot

Result: strings are extracted for both _() and __()

5. pybabel extract requires babel.cfg even though it ignores its settings and complains about keywords not being set

Basically an extension of the above. Remove --keywords and add --no-default-keywords:

$ pybabel extract -F ./babel.cfg --no-default-keywords -o dist/lang/en.pot ./dist/gettext.js
Usage: pybabel extract [options] <input-paths>

pybabel: error: you must specify new keywords if you disable the default ones

Summary:

Please make one of the following:

• make sure Babel CLI respects settings in babel.cfg, or
• add a command-line parameter to babel extract to let it extract from .js files without the need for babel.cfg

[dimaqq]

Also, pybabel attempts to extract from non-source files and crashes, e.g.:

extracting messages from project/subdir/._source_name.html.swp (extract_messages="gettext, ngettext, _", input_encoding="utf-8")
Traceback (most recent call last):
  ...
UnicodeDecodeError: 'utf-8' codec can't decode bytes in position 17-18: invalid continuation byte