GNU bug report logs - #73188
PEG parser does not support full PEG grammar

Previous Next

Full log

Message #23 received at 73188 <at> debbugs.gnu.org (full text, mbox):

From: Ekaitz Zarraga <ekaitz <at> elenq.tech>
To: Ludovic Courtès <ludo <at> gnu.org>
Cc: 73188 <at> debbugs.gnu.org
Subject: Re: bug#73188: PEG parser does not support full PEG grammar
Date: Mon, 14 Oct 2024 16:00:02 +0200

[Message part 1 (text/plain, inline)]

Hi,

So I reviewed the suggestions and this what I have:

- Added the comment

- Added my copyright line

- Nothing in Guix is affected by the changes proposed here, because 
there's no use of `string-peg` and I didn't change how the S-Expression 
based PEG works.

- The changes applied here only affect to the PEG written as PEG strings 
and their processing, which was introduced in the documentation as "read 
the wikipedia", but our PEG definition wasn't implementing the real PEG 
explained in that wikipedia link. Our documentation claimed that we were 
able to do things we were not able to do.

- I added a link to the paper we are following below the link to the 
wikipedia page just in case, to let our users know exactly what does our 
PEG support. The paper says:

	Literals and character classes can contain C-like escape codes

That should be enough for anyone. Also we support `-` in identifiers, 
but our documentation already said that. I replaced the description of 
what normal PEG supports. Our docs said PEG only supported alphabetic 
characters but it has never been true, the paper explains PEG supports 
alpha, numeric and `_`, but the numeric cannot be the first character 
(like C identifiers).

- What it is true is if people were using `peg-as-peg` for their things, 
which isn't even exported by the module, their code would have problems. 
But we can't predict that.

- In summary, what I implemented here is a PEG string to PEG Sexp 
compiler. So it shouldn't affect anyone that uses the PEG Sexp directly 
and if I did my work properly it shouldn't affect those that used their 
PEG patterns in strings either, because I only made it work closely to 
what the paper said, adding a couple of missing features.

EXTRA:

I rebased the support for the `not-in-range` part, that does touch the 
PEG sexp part, but only adds more functionality without affecting 
anything that already existed.

In following days I'll review Error reporting and captures that were 
added in a different thread in guile-devel and see if we can integrate 
everything to create a production ready PEG parser.

Thanks!
Ekaitz

I attached both patches.

[v3-0001-PEG-Add-full-support-for-PEG-some-extensions.patch (text/x-patch, attachment)]

[v3-0002-PEG-Add-support-for-not-in-range-and.patch (text/x-patch, attachment)]

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.