Forbid unparseable facts

[ederag]

Make the save button active only if the fact entered can be parsed (and thus edited) from the cmdline.
This hopefully answers #461 (comment) favorably, while waiting for #465.

The first two commit are from PR #475, only a544d5e is added by this PR.

[GeraldJansen]

Thanks! Yes this solves #461 (comment) reasonably well, as well as the two issues closed by the previous two commits.

I noticed that the parser changes allow "#" in any position in the activity name, in addition to the first, which is fine with me. Playing with this a bit, I noticed some quirky behaviour. For example, typing #123 and #456@bugs, description slowly in the cmdline, 456@bugs is considered a tag up until the ', d' is entered and then flips to be part of the description and category. I guess this is a corner case that can be addressed later under #465.

(ps. very glad you're back!)

[ederag]

@GeraldJansen

56@bugs is considered a tag up until the ', d' is entered and then flips to be part of the description and category

Thanks for the feedback, glad you're still there !
That kind of interpretation change seems inevitable since
activity #456 was interpreted as an activity with 456 tag,
and we want to be fully backward compatible.

Just noticed: adding a comma in the middle of a tag in the gui cmdline wreaks havoc. 😞

[GeraldJansen]

This change would fix the strange behaviour with #456@cat:

    ([^#@\s,]+)  # the tag (anything but #, @, spaces or comma)

[ederag]

Wouldn't that forbid '@' in tags and be a regression ?
IMO the behavior is certainly puzzling at first sight, but clearly understandable
if the rules can be kept simple.

[GeraldJansen]

Ah, so we also need to allow @ in the middle of tags. I never would have imagined that. Okay.

[GeraldJansen]

Looking at the Input help page (eg. here), it seems that "#tag with spaces" is explicitly mentioned. This no longer works with this PR (or maybe earlier). For cmdline input abc@cat, desc #tag with spaces the description becomes "desc #tag with spaces" and there are no tags. A regression?

[ederag]

Perfect, that used to work in v2.0 (c3e5fb7), indeed;
it would have been a pity not to allow spaces in tags as well 😄

edit:
Well, this is not compatible with 9fe4518

hamster/tests/stuff_test.py

Lines 63 to 66 in 9fe4518

	activity = Fact("case, with added #de description #and, #some #tags")
	self.assertEquals(activity.activity, "case")
	self.assertEquals(activity.description, "with added #de description")
	self.assertEquals(set(activity.tags), set(["and", "some", "tags"]))

So parser backward compatibility seems impossible to achieve.

[ederag]

superseded by #481