python-babel
Allow extractors to provide message flags
This adds a sixth value to the tuple returned by the extractor functions which should be a set of flags.
Which flags should be applied to a message should be determined by the extractor, as it depends on e.g. the file format being parsed. For example "%s" should have the python-format flag if it was parsed from a Python file and the c-format flag if it was extracted from a C file.
The logic of detecting python-format flags is also moved to the Python extractor in this PR.
NOTE:
This is partially a breaking change. Backwards compability is maintained with extractors that return 5-tuples instead of 6-tuples, but the interface Babel exposes for extracting messages always returns 6-tuples. I don't see a good way around this.
Removing the python-format detection from Message.init is also a breaking change, but that could potentially be put back for now.
Fixes #35
Let me know if this PR has any future and if so I'll clean it up, fix the last remaining tests and add a few new tests.
Based on this I also have working code fixing #759, support for the python-brace-format flag :)
Edit:
And I just noticed #80 🙈 That makes the same changes to the interface, but it doesn't appear to implement the flag detection bit in the Python extractor, which would be yet another breaking change when using the babel CLI.
Anyway, let me know if you want me to get this over the finish line. We might be switching over to Babel for a project at work, so I need these changes and will probably maintain a fork until this is fixed (we need python-brace-format detection as well, hence why I started on this).
I need this feature for custom extractor. Without it the custom extractor cannot set flags. The older version of babel had this feature, but it was removed.
