caml-list - the Caml user's mailing list
 help / color / mirror / Atom feed
From: Jean-Marie Gaillourdet <jm@gaillourdet.net>
To: Markus Mottl <markus.mottl@gmail.com>
Cc: ocaml <caml-list@inria.fr>
Subject: Re: [Caml-list] Partial parsing
Date: Fri, 28 Jul 2006 11:58:59 +0200	[thread overview]
Message-ID: <244C09B7-E152-4B9E-8B7E-A27AC225B5E7@gaillourdet.net> (raw)
In-Reply-To: <f8560b80607271537r29bf2ae1j850277cd4e1edafe@mail.gmail.com>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

On 28.07.2006, at 00:37, Markus Mottl wrote:

> Hi,
>
> one feature that I'd love to see with our lexing/parsing tools for
> OCaml would be the possibility to perform partial lexing / parsing.
>
> For example, imagine that you read a message from a socket into a
> string buffer, but it is not yet complete.  You may not be able to
> know that in advance, because that would require parsing it.
>
> What I'd like to be able to do is to e.g. call a lexer or parser
> function on this buffer, and it will recognize as much as possible
> with the option to continue parsing in another buffer.  For this we
> could e.g. define a type "parse_result":
>
>  type parse_result =
>    | Done of result * int
>    | Cont of parse_fun
>  and parse_fun = pos : int -> len : int -> string -> parse_result
>
> If a full parse was detected, it would return e.g. "Done (result,
> next_pos)", where "result" is the result of the successful parse, and
> "next_pos" is the position in the buffer that contains the character
> following the successfully parsed text.
>
> If the buffer did not contain enough data to finish the parse, the
> function would return e.g. "Cont f", where "f" can be called to
> continue parsing in a new buffer starting at position "pos" and
> reading at most "len" characters, e.g. in the case when the rest of
> the message arrives in the input buffer.
>
> So far this kind of parsing can only be implemented by hand, which is
> very tedious to do and inevitable if one wants to read self-delimiting
> messages of variable size from sources that return partial data.  I
> don't see any inherent problem extending the code generators for
> ocamllex and ocamlyacc to support this feature.
>
> Has anybody already implemented a generic solution to this kind of
> parsing problem?  Maybe Menhir, which seems like an ambitious parser
> project, could add this feature to support partial parsing (of course,
> this would require cooperation from the lexer, too)?

You could implement such a feature on top of the existing lexer and  
parser technologies by using thread. Dedicate one thread to the lexer  
and parser, make the input buffer blocking and  make every suitable  
intermediate result available to the main thread.

I think it should also be possible to construct such a feature with  
the resumable exceptions module posted on this list some weeks ago.

And I think it should be possible to implement that with the  
continuations posted here, too.

But I have not much experience with continuations. And I am not sure  
whether that really fits into the framework of ocamllex/ocamlyacc.

As Fermin Reig pointed out parser combinators provide that feature  
out of the box. AFAIK, they depend on lazy evaluation. I am not sure  
whether those libraries for OCaml are available to hide that or not.

Best regards,
   Jean-Marie
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)

iD8DBQFEyd/nNIUNP/I5YOgRAmWJAJ9WijR0B0f1FT/4smsum7Ln4qWXPACeLxkB
4Hp1UPcEseMzFYzGZfLhbYc=
=IqFi
-----END PGP SIGNATURE-----


  parent reply	other threads:[~2006-07-28  9:58 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-07-27 22:37 Markus Mottl
2006-07-27 22:57 ` [Caml-list] " Fermin Reig
2006-07-28 13:39   ` Sebastien Ferre
2006-07-28  9:58 ` Jean-Marie Gaillourdet [this message]
2006-07-28 14:01   ` Markus Mottl

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=244C09B7-E152-4B9E-8B7E-A27AC225B5E7@gaillourdet.net \
    --to=jm@gaillourdet.net \
    --cc=caml-list@inria.fr \
    --cc=markus.mottl@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).