xandaros / abnf Goto Github PK

Thanks for your great library - I was looking for a Haskell equivalent to abnf.py, and this looks good!

Is it possible for you to please publish an updated version to Hackage? Version 0.4.1.0 has a base bound of >=4.8 && <4.10. base-4.10 was bundled with GHC 8.2.1 in 2017, so it's quite old now. However, I noticed that this repo has a much more recent bound, which would make it more widely usable.

Document is missing Applicative and Monad instances and the Applicative instance for Content needs to be made explicit.

All those instances (incl. Functor) need their respective laws proven.

…to prevent people from accidentally using uncanonicalised rules

Provide a usage example in the Text.ABNF module, as that is likely the first module a potential user clicks on.

There are still two parsers using partial function - these will crash on certain invalid input

The library should use linear (or close to) memory. Find out if that is the case and if it isn't - why.

These will probably break the tests, so the actual functions need to go into an Internal module, which gets selectively re-exported by the modules that currently exists.

Generated parser doesn't work. Needs tests.

(I suppose the ABNF parser can get some, too. While we're at it)

I was looking at using this project to make a parser for BCP47 Language-Tags, but I can't seem to make even simpler cases work.

I've started by using the following simplified rule,

Language-Tag = 2*3ALPHA

To parse a simple value,

en

But I always get

Left "Rule > Sum: Failed reading: empty."

I can tell something is working, because if I change the ABNF to,

Language-Tag = "en"

Then I get correctly,

Right (Document "Language-Tag" [Terminal "en"])

Am I doing something wrong? Shouldn't 2*3ALPHA produce a parser that succeeds on 2-3 alpha characters?

I have my doubts many people will immediately understand what this library is about.