provide some babel genshi extractor configuration to extract as msgids only strings in _()

[reuleaux@…]

In new projects for strings not to be translated one can mark them with xml:lang="..." like so

  <p>foo</p>
  <p>bar</p>
  <p xml:lang="en">don't translate me</p>

so pybabel will only pick up foo and bar as msgids.

However in older projects (that I want to convert to babel now) I have templates with explicit uses of _() like so

  <p>${_('foo')}</p>
  <p>${_('bar')}</p>
  <p>don't translate me</p>

pybabel extract extracts the msgids foo, bar and "don't translate me" however I only want foo and bar to be extracted (without editing the templates).

Christophers has already made some suggestions to my description of this problem on the babel mailing list, so I am just citing him here:

It might be best for us to add an additional option to the Genshi extractor for this use case, so that you could define:

   [genshi:**.html]
   extract_text = false

Or allow for a magic value of the "ignore_tag" option:

   [genshi:**.html]
   ignore_tag = *
   include_attr =

[cmlenz]

The extract_text option described above was added in [708], and ported to the 0.4.x branch in [709].