Skip to content

RockBrentwood/RegEx

Repository files navigation

These are software demos and sources for regular expressions -> finite automata conversion and regular language accquisition.
The programs are in ISO C99.

The sources:
	nfa.c, dfa.c, fsc.c, rex.c
are respectively for RE→NFA conversion, RE→DFA conversion, a "finite state classifier" solution to the regular language acquisition problem, and the rex regular expression text filter.

The programs are small: at present, 154 lines (fsc.c), 365 lines (nfa.c), 483 lines (dfa.c), 615 lines (rex.c).

Included are descriptions of the various software and (for comparison with rex), the manual section for GNU GREP.

Sample inputs are in the files *.in; outputs in *.ex listed under the Ref directory.

The files fa7.in, nfa7.ex illustrate the full syntax of dfa and nfa, and show their limitations relative to a prospective CFG parsing utility (while also showing the potential for use as the basis of one).

The files fa8.in, nfa8.ex illustrate the general solution in 4 variables corresponding to the problem of converting a 4-state finite state automaton into a regular expression.
Running dfa on it will recover the original automaton back.
The size of the expression is of the order Q³ in the number of state Q of the dfa, with the following detailed count:
	⅓ Q(Q-1)(Q+1) bars, ½ Q(Q-1) stars, ⅙ Q(Q-1)(2Q+5) + 1 concatenations, and Q(Q+1) equations.
illustrated for Q = 4.

The files dfa0.in, dfa1.in and dfa2.in illustrate the first four levels in the approximation series respectively for the languages given as the least fixed point solutions to the following inequalities:
	dfa0.in: S → [S ^ (a b c)*]	-- a context-sensitive 2-counter language
	dfa1.in: S → [S ^ (a b)*]	-- the context-free Dyck Language S -> 1 + a S b S
	dfa2.in: S → [a S ^ (b a)*]	-- (I'm not sure what this one is)
The automaton for dfa0.in is a *quiver* that has the form triangular grid, with arrows going in the 3 principal directions labelled by a, b and c, such that each a-b-c sequence traverses a loop.

If nfa and dfa were compiled into executables, named nfa and dfa respectively, then the files nfa1.ex and dfa1.ex would be generated as follows:
                           nfa <fa1.in >nfa1.ex
                           dfa <fa1.in >dfa1.ex
The files dfa0.in, dfa1.in and dfa2.in contain extended regular expression operators that can only be processed by the dfa program.

The file rex0.in lists a regular expression representing all lines consisting of 3 a's and 5 b's.
The file rex1.in lists an anagram of a former president a previous decade.
Run it on the sample files provided (rex0.txt, rex1.txt) as follows:
                        rex -f rex0.in rex0.txt
                        rex -f rex1.in rex1.txt
This utility is so powerful that it can find anagrams, as these examples illustrate.

Compiling the Programs:
To compile this package, you will need a C99 C compiler that produces code targeted for hosted environments.
That includes most contemporary C compilers for most mainframes, desktops, laptops and mobile systems.

If using the Makefile, make the appropriate changes to the topmost lines of the file before using it.
Each program is self-contained and may be compiled separately as a stand-alone program file.

A host-dependency exists for rex.c in order to support wildcards on the command-line.
It is set to compile for a POSIX system.
If working on DOS or Windows, you will need to #define SYSTEM flag in the rex.c program appropriately before compiling it.
The supported systems are SYS_DOS, SYS_WIN and SYS_NIX, respectively for DOS, Windows and UNIX.
Alternatively, you might need to rewrite the MatchCount() routine in the rex.c program to suit your system and/or compiler.
Its definition is as follows:
	MatchCount(P, E) --
		E == 0: The number of files whose name matches pattern P.
		E != 0: The number of files with names matching pattern P whose contents match the regular expression E.
For E != 0, MatchCount also applies rex with expression E to each file whose name matches the wildcard pattern P.

Locations:
The Site Locations, as of 2007 January 27, were:
∙	FTP via pardo@cs.washington.edu -- ftp.cs.washington.edu::pub/pardo/regex.tar.Z
∙	FTP via comp.compilers archive -- iecc.com:pub/files/regex.tar.gz
∙	FTP via colas.nahaboo@sophia.inria.fr -- avahi.inria.fr:/pub/regex.tar.gz
As of 2020, they may no longer be valid.

Related References:
[1]	Released on 1997 March 29 by Mark Leisher <mleisher@crl.nmsu.edu>
	Copyright 1997, 1998, 1999 Computing Research Labs, New Mexico State University
	Version: 0.5; 1999 September 21

	A filter for Unicode and Regular Expressions:
	"This is a simple regular expression package for matching against Unicode text in UCS2 form.
	The implementation of this URE package is a variation on the RE->DFA algorithm done by Mark Hopkins (markh@csd4.csd.uwm.edu).
	Mark Hopkins' algorithm had the virtue of being very simple, so it was used as a model."
[2]	TREX-Q: A query language based on XML Schema
	Brad Penoff, Chris Brew
	http://xml.coverpages.org/PenoffIRCS.pdf

About

Regular expressions and finite state automata using Kleene algebraic techniques. Formerly archived in the comp.compilers archive

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published