Feature/long integer with underscores#923
Conversation
|
@davidism : how would you prefer to handle this? This feature is limited to Python 3..6 and up. Currently it doesn't degrade nicely but I could fix that. Or ignore it. I could make the regex dependent on the python version. I'm open to suggestions but I won't put any more work into this one until I get your feedback. |
|
We could support the underscore syntax for any version by removing them before calling literal_eval. Seems reasonable enough to do that. |
|
Good point and should be simple enough to do. |
|
Also of note, this PR contains the changes to #922 as they are literally on the same lines. |
|
Finally passes all tests. Let me know if it's acceptable, please. |
|
I've changed test_float and test_int but I'm thinking we should probably have a test_template that actually tests the regex. |
|
I'll get to this eventually! I'm just focused on Werkzeug right now. :-) |
add changelog clean up docs parametrize tests explain float regex
0ed6689 to
733851e
Compare
|
Rebased onto current master + #922 changes. Explained the regex with verbose mode. Allow '_' in any part, which matches Python's grammar. Moved the tests out of filters and into lexer, since that's what this was about, but cleaned up both sets of tests with |
It is now possible to use the '_' character to make long integers and floats more easily understandable by a user.
100_000_000_000 for example, or 100_000_000_000.000_000_001
I don't think this closes any issue but since I had my hands in regex, it was a good timing,