* before

[Pitometsu]

BNF from RFC https://www.ietf.org/rfc/rfc2396.txt throw lot of errors.

[arrdem]

Yeah, Instaparse doesn't support this RFC's strange leading n*.... notation for N or more repetitions. The traditional notation for this would be ....{n,}. It actually doesn't look like Instaparse supports that either. Skimming the RFC however it looks like most of the leading * notation could be refactored using *? or +?.

[Pitometsu]

Yes, I replaced *() by ()* and 1*() by ()+. I hope, it's correct. Just put it here in case you want to add that notation as well.

[Engelberg]

The n* notation is supported by ABNF, but not EBNF.

…

On Fri, Mar 10, 2017 at 7:36 PM, Netsu ***@***.***> wrote: Yes, I replaced *() by ()* and 1*() by ()+. I hope, it's correct. Just put it here in case you want to add that notation as well. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#156 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAIdS7dcQis0cF5ubyxUk1YGSWxPjWuDks5rkhbTgaJpZM4MaEUJ> .

[Pitometsu]

But with :input-format :abnf I have error with that bnf from rfc:

   Error parsing grammar specification: Parse error at line 1, column 31:
   URI-reference = [ absoluteURI | relativeURI ] [ '#' fragment ] ^ Expected one
   of: ] / #"#\"[^\"\\]*(?:\\.[^\"\\]*)*\"(?x) #Double-quoted regexp"
   #"#'[^'\\]*(?:\\.[^'\\]*)*'(?x) #Single-quoted regexp" !  & % ' " [ < (
   #"[a-zA-Z][-a-zA-Z0-9]*(?x) #identifier" * #"[0-9]"

I there type in rfc abnf?

[Engelberg]

ABNF doesn't support the | character, which is in the grammar.

There are a lot of subtle variations in grammars, and I think the problems you are running into stem from the fact that this particular spec uses a hybrid of notations from EBNF and ABNF.

If you're looking for ABNF for URI, you might get farther with this as your starting point:
https://tools.ietf.org/html/rfc3986#page-49

[Engelberg]

Actually, there's already a tested URI spec in the test/data directory of instaparse.

[Engelberg]

If you're working on ABNF extensively, see issue #136

As one user pointed out, a lot of ABNF grammars are actually ambiguous and assume left-to-right processing, but instaparse process alternatives right-to-left. I'll likely be changing this in the next major point release of instaparse for better handling of these grammars that make this assumption.

[Pitometsu]

@Engelberg thanks a lot for suggestions. So, closing the issue for now.