Revision Parsing

[kalkin]

The goal of this issue is to implement parsing git revisions, according to
gitrevisions(7). Parsing
of ranges is not a goal of this issue, but the first step. See also #XXX.

To piggy back on the reply from @Byron from here.

Parsing would go into the git-revision crate…

Ok, but the main interaction point for the user of the library would still be
Repository::rev_parse()?

• parse plain bytes (or &BStr), not &str

Ok, I can adjust that.

• call functions to signal which item was parsed, that is it doesn't parse
  into a data structure

This design makes it possible to validate by making each callback a no-op and
to produce the final object id step by step otherwise if a Repository instance
is providing the callbacks.

I'am not sure I understand what you mean. My approach does not parse into a
data structure. It works like this (no ranges support):

1. Input contains @{ → parse_with_at() ([<refname>]@{reflog} spec NIY)
2. Input starts with :/ → search_by_regex() (NIY)
3. We have a revision of form <rev><nav>*<special>? where
   • <rev> is SHA1, refname, HEAD or @
   • <nav> is the ^ or ~ operator followed by an optional number
   • <special> is a type filter {tree}, a peel tag command {} or a regex `{/regex}'
4. Split in to <rev>, <nav> & <special> parts.
5. Convert <rev> to Id
6. If <nav> not empty navigate
7. Apply <specials> NIY
8. PROFIT!