Implement Extended Version Comparisons (Ranges, Hyphens, Tilde...)

[tomschr]

Situation

Currently, semver.VersionInfo semver.Version contains a match method. However, it does support only a limited number of comparisons. For example, ~ or ^ as known in NPM is unknown.

Proposal

Add an additional class which provides all the necessary functionality. How this is implemented is a matter of design. 😉

I'd suggest to add this after major 3 is released.

This could be the use cases (I use a class Spec here, but the naming is subject to change):

>>> import semver
>>> spec = semver.Spec(">=1.0 <= 1.7")
>>> ver = semver.Version.parse("1.2.0")

>>> spec in ver
True
>>> spec.match(ver)
True

Depending on how we want to design it, it could also influence, for example, semver.Version.match. Maybe we should (re)move that and move this functionality into a future Spec class? Wouldn't it a bit more appropriate there?

@python-semver/reviewers What do you think about this idea? Any preferences?

Grammar

range-set  ::= range ( logical-or range ) *
logical-or ::= ( ' ' ) * '||' ( ' ' ) *
range      ::= hyphen | simple ( ' ' simple ) * | ''
hyphen     ::= partial ' - ' partial
simple     ::= primitive | partial | tilde | caret
primitive  ::= ( '<' | '>' | '>=' | '<=' | '=' ) partial
partial    ::= xr ( '.' xr ( '.' xr qualifier ? )? )?
xr         ::= 'x' | 'X' | '*' | nr
nr         ::= '0' | ['1'-'9'] ( ['0'-'9'] ) *
tilde      ::= '~' partial
caret      ::= '^' partial
qualifier  ::= ( '-' pre )? ( '+' build )?
pre        ::= parts
build      ::= parts
parts      ::= part ( '.' part ) *
part       ::= nr | [-0-9A-Za-z]+

See Also

• https://docs.npmjs.com/cli/v6/using-npm/semver
• https://github.com/rbarrois/python-semanticversion/blob/master/semantic_version/base.py#L586
• http://instaparse.mojombo.com/
• Parsing a "simple" grammar