Re: Generalized parser without generation

vidar@hokstad.name (Vidar Hokstad)
4 Feb 2004 21:56:48 -0500

          From comp.compilers

Related articles
Generalized parser without generation moughanj@tcd.ie (2004-02-01)
Re: Generalized parser without generation joachim.durchholz@web.de (Joachim Durchholz) (2004-02-04)
Re: Generalized parser without generation derkgwen@HotPOP.com (Derk Gwen) (2004-02-04)
Re: Generalized parser without generation jesjones@mindspring.com (Jesse Jones) (2004-02-04)
Re: Generalized parser without generation pete@restall.net (Peter Restall) (2004-02-04)
Re: Generalized parser without generation cfc@shell01.TheWorld.com (Chris F Clark) (2004-02-04)
Re: Generalized parser without generation vidar@hokstad.name (2004-02-04)
Re: Generalized parser without generation bear@sonic.net (Ray Dillinger) (2004-02-08)
| List of all articles for this month |
From: vidar@hokstad.name (Vidar Hokstad)
Newsgroups: comp.compilers
Date: 4 Feb 2004 21:56:48 -0500
Organization: http://groups.google.com
References: 04-02-031
Keywords: parse
Posted-Date: 04 Feb 2004 21:56:48 EST

moughanj@tcd.ie (James Moughan) wrote
> A friend of mine has been given a final year project for which his
> supervisor wants the above; that is, a program which can read in a
> description of a grammar then parse a file using it directly. Users
> must be able to add new grammars at any time without compiling. ...


I can't point you to a specific publicly available tool, but having
written tools like this myself. It's a fairly straightforward thing to
do, as you are basically creating an interpreter that is specialized
for a very small set of operations.


Personally I've built configurable parsers that used an EBNF variation
to specify the parser. Instead of generating a parser all you have to
do is execute the corresponding actions on an input stream instead.


Recursive descent for instance is trivial to do this way, as you only
build a dictionary of productions, and build parse trees for each of
the productions, and then the recursive descent is basically a matter
of descending the parse tree for the grammar until you come across
terminals and match them against the input stream.


I used it for a tool that would take an arbitrary grammar and output
an XML document representing the parse tree for further
transformation. It was quite useful as a way of quickly prototyping
source code transformation tools.


Vidar


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.