From: rpardee@...
Date: 2005-02-25T01:50:06+09:00
Subject: regexp help sought

Hey All,

I'm trying to parse lines from my text editor's config file, which look
like this (pls watch for line wrap--there is one line per language,
starting with /L<<digit>>):

/L1"SAS" Line Comment = * Block Comment On = /* Block Comment Off = */
Block Comment On Alt = * Block Comment Off Alt = ; Nocase File
Extensions = SAS
/L2"Visual Basic" Line Comment = ' File Extensions = BAS FRM CLS VBS
CTL WSF
/L4"HTML" Nocase Noquote HTML_LANG Block Comment On = <!-- Block
Comment Off = --> Block Comment On Alt = <% Block Comment Off Alt = %>
String Chars = "' File Extensions = HTM HTML ASP SHTML HTT HTX JSP
/L11"Ruby" Line Comment Num = 2# Block Comment On = =begin Block
Comment Off = =end String Chars='" Escape Char = \ File Extensions = RB
RBW

I'm trying to write a method for extracting the comment markers & their
types (line/block & on/off).  Regexps seemed the obvious tool, and I
eventually came up with this one:

   c = Regexp.new("(Line|Block) Comment (On |Off |On Alt |Off Alt)*=
([^\s\t\r\n\f]+) ")

This is working well so far, except that it only grabs out the first
type of comment in each line.  I'd hoped that I could make it get all
the comment types by putting an additional set of parens and a +
quantifier around the whole expression:

   c = Regexp.new("((Line|Block) Comment (On |Off |On Alt |Off Alt)*=
([^\s\t\r\n\f]+))+ ")

But that just seems to break it--that version doesn't capture anything.

Anybody got a clue for me?  I'm using v1.8 on windows.  My code is
below.  (And again, pls watch for line wrapping).

Thanks!

-Roy

def parse_comment_markers(line)
=begin

   There are line comments & (2 different kinds of) block comments.

   Line comments only have a start marker--EOL is the terminator.

   Comment types are:
      Line Comment          = <<mark>>
      Block Comment On      = <<mark>>
      Block Comment Off     = <<mark>>
      Block Comment On Alt  = <<mark>>
      Block Comment Off Alt = <<mark>>

   Where <<mark>> can be any contiguous set of non-whitespace chars.

   For Line comment marks, preceding digits specify the # of spaces
minus 1
   required after the nondigit portion of the marker.  So for ruby, the
line
   comment mark is 2#, signifying that # is a comment only if it is
followed by
   a space.  Ignore this for now.

   So--funky regexp time.  We want to grab sequences centered around
the string " Comment ".
   We want the single word prior to "Comment" and all words between
"Comment" and " = ", and
   then of course the contiguous nonwhitespace following " = ".

=end
   puts line
   # Why doesn't the \S char class work?
   # c = Regexp.new("(Line|Block) Comment (On |Off |On Alt |Off Alt)*=
(\S+)")
   c = Regexp.new("(Line|Block) Comment (On |Off |On Alt |Off Alt)*=
([^\s\t\r\n\f]+) ")
   cm = c.match(line)
   if cm.nil?
      puts "No match!"
   else
      puts cm.captures.join("     ||     ")
      puts "Comment type is \"" + cm.captures[0] + "\", and comment
marker is \"" + cm.captures[2] + "\""
   end
end

parse_comment_markers("/L2 \"Ruby\" Line Comment = # Block Comment On =
' File Extensions = RB RBW")
parse_comment_markers("/L2 \"Ruby\" Block Comment On = =begin Block
Comment Off = =end File Extensions = RB RBW")