Implement more error explanations #405
Conversation
| @@ -127,6 +130,25 @@ impl<'a> ExplainUnparsed<'a> { | |||
| } | |||
| None | |||
| } | |||
| fn is_redundant(&self, parsed: &[Name], unparsed: Name) -> Option<Error> { | |||
| let unparsed_info = self.all_names.get(&unparsed)?; | |||
| if unparsed_info.in_many { | |||
There was a problem hiding this comment.
This check makes sense. If parser is annotated with many in any way - it can be accepted several times so this diagnostic is no longer applies.
| if unparsed_info.in_many { | ||
| return None; | ||
| } | ||
| for p in parsed { |
There was a problem hiding this comment.
But I think the rest can simply be parsed.contains(&unparsed) of some sort - "if name can be parsed only once (first condition) and we already parsed it (this condition)" then we tell the user not to pass it twice...
There was a problem hiding this comment.
Do we have the information regarding only once stored right now?
There was a problem hiding this comment.
!info.in_many would be it I guess. Let me show an example.
#[derive(Bpaf, Clone, Debug)]
#[bpaf(options)]
struct Options {
a: usize,
b: usize,
#[bpaf(argument, many)]
c: Vec<usize>,
}So say we have this parser. It takes arguments -a=0 and -b=0 once, -c=0 multiple times.
If user passes -a0 -b3 -c1 -c2 -c3 -c4 -a1parser will run up to-c4and will give up on-a1`.
Visitor gets -a0 -b3 -c1 -c2 -c3 -c4in parsed,-a1` in unparsed (well, just names I guess)...
To explain that name -a can be used only once we only need to look at name -a that it is not inside of many context - it is not.
lu-zero
force-pushed
the
more-explanations
branch
from
ba244b2 to
f0018ee
Compare
|
Last push looks correct. Will merge in an hour or so. |
No description provided.