documentation on is_approximate and is_uncertain

[rettinghaus]

for a date of classes UncertainOrApproximate and PartialUncertainOrApproximate return 'false' on is_approximate and is_uncertain. I guess that this is a bug, but couldn't find documentation on that.

[azaroth42]

Just ran into this as well.

[ColeDCrawford]

@aweakley there is an easy fix for UncertainOrApproximate and Unspecified, but want to check on the intended behavior. Right now we have three different properties - is_uncertain, is_approximate, and is_approximate_and_uncertain. is_uncertain and is_approximate are False if is_approximate_and_uncertain is True - we're treating these as mutually exclusive. Does that make sense?

After setting the properties in UncertainOrApproximate and Unspecified based on ua:

>>> from edtf import parse_edtf, text_to_edtf
>>> e = parse_edtf("1666%")
>>> e.is_uncertain
False
>>> e.is_approximate
False
>>> e.is_uncertain_and_approximate
True
>>> f = parse_edtf("1666~")
>>> f.is_approximate
True

For PartialUncertainOrApproximate, we have uncertainty/approximation at sublevels, but not currently at top level properties:

>>> parse_edtf("2004-06~-11")
PartialUncertainOrApproximate: '2004-06~-11'
>>> e = parse_edtf("2004-06~-11")
>>> e.is_approximate
False
>>> e.year_month_ua
UA: '~'
>>> e.year_ua
False

I think it makes sense to have is_approximate, is_uncertain, and is_uncertain_and_approximate available here too. The question is whether those resolve to True if any part of the date is approximate / uncertain? I think they must, because otherwise fully qualified (qualification on the far right) would just parse as L1 UncertainOrApproximate.

[ColeDCrawford]

Any thoughts on this one @aweakley ?

[aweakley]

Sorry. Yes I agree, I think if any part of the date is approximate or uncertain then the appropriate one should resolve to True.

we're treating these as mutually exclusive. Does that make sense?

I think it does, because each alternative has different implications for our output.

[ColeDCrawford]

Great. Should Intervals also have these properties? Right now they do (since they inherit from EDTFObject) but they are not useful as they don't return actual results (just False):

>>> from edtf import parse_edtf
>>> e = parse_edtf("2004-06-~01/2004-06-~20")
>>> e
Level2Interval: '2004-06-~01/2004-06-~20'
>>> e.is_approximate
False
>>> e.is_uncertain
False
>>> e.is_uncertain_and_approximate
False
>>> f = parse_edtf("1984~/2004-06")
>>> f
Level1Interval: '1984~/2004-06'
>>> f.is_approximate
False
>>> f.is_uncertain_and_approximate
False
>>> f.is_uncertain
False

I think we should probably set those properties based on whether either of the interval sections has .is_approximate, etc?

[aweakley]

This is resolved in #56