From: ahoward <ahoward@...>
Date: 2002-11-15T05:20:54+09:00
Subject: Re: regex help

On Fri, 15 Nov 2002, Michael Campbell wrote:

> Those aren't (mathematically) "regular" then, are they?  (Granted,
> existing 'regular' expressions aren't either with backreferencing.)

in fact they are regular because the matching char is always known : it is
either the same, or a match to a known mate.  check out the oreily regex book,
it's a really good read and has many examples of using dfa and nfa regexes to
match, for example, matching comment paris (/* */) which is exactly like this
problem

now, it would be non-regular if you tried to make a single regular expression
which would match ALL occurances mentioned since backreferencing would be
required - maybe that's what you mean?  i do agree that this type of method
should not belong in a Regex class since a general method *would* appear to be
non-regular, even if the collection of regexs where individually regular

i think a String#delimscan method would be good, something like :

class String
  @@delimscan_chars =
    {
      '\'' => '\'',
      '"'  => '"',
      '('  => ')',
      '['  => ']',
      '{'  => '}',
      '<'  => '>',
    }
  def delimscan delim='"', escape='\\'
    o  = delim[0]
    c = @@delimscan_chars[delim]
    c = c ? c[0] : o
    e = escape[0]

    w = []
    b = 0
    a = 0
    n = 0

    while ((b = self[n])) do
      case b
	when e
	  n += 2 and next
	when o
	  a = n
	  n += 1
	  while ((b = self[n])) do
	    case b
	      when e
		n += 2 and next
	      when c
		w << self[a..n]
		break
	    end
	    n += 1
	  end
	else;
      end
      n += 1
    end

    return w
  end
end

if $0 == __FILE__
  strings = [
    %q("one"),
    %q("one" "two"),
    %q("o\"ne" "t\"wo"),
    %q("),
    %q(""),
    %q("""),
    %q(),
  ].each do |s|
    puts "TESTING #{s}"
    puts s.delimscan.inspect
  end

  strings = [
    %q(<one>),
    %q(<o!<ne> <t!>wo>),
    %q(<),
    %q(<>),
    %q(<>>),
    %q(),
  ].each do |s|
    puts "TESTING #{s}"
    puts s.delimscan('<', '!').inspect
  end
end


-a

-- 

 ====================================
 | Ara Howard
 | NOAA Forecast Systems Laboratory
 | Information and Technology Services
 | Data Systems Group
 | R/FST 325 Broadway
 | Boulder, CO 80305-3328
 | Email: ahoward@fsl.noaa.gov
 | Phone:  303-497-7238
 | Fax:    303-497-7259
 ====================================