textilewc.rb

#!/bin/env ruby
############################################################################
# textilewc: Count words in a Textile file, if somewhat crudely.
# By Tammy Cravit, [email protected]
#
# $Revision$ $Date$
#
# If a single file is provided on the command line, it displays the number
# of words in the file. If multiple files are given on the command line, it
# generates a listing similar to the output of wc(1), except that directories
# and non-Textile files are silently skipped with no warning.
#
# And yes, this is more complicated than it probably needs to be
############################################################################
# IMPORTANT NOTE:
# 
# In my directories, I use the prefix character _ to denote a template file,
# such as "_Character Template.textile". The definition of textile_file?
# will SILENTLY ignore Textile files whose names start with an underscore.
#
# You can override this behavior on the command line with the -a option. The
# default behavior can be changed by modifying the initialize method to set
# @hide_leading_underscores to false. If you call the TextileWordCounter
# class from your own code, you can change this setting with the method
# hide_leading_underscores!(boolValue)
############################################################################

require 'rubygems'
gem 'RedCloth', ">= 0"
require 'redcloth'

############################################################################
# The main Textile word counter class
############################################################################

class TextileWordCounter
  def initialize
    @hide_leading_underscores = true
  end
  
  # Return true to exclude files with leading _ characters in their names
  # from the count. I use a leading _ to denote a template file, so I want
  # them excluded when I run textilewc with a wildcard.  
  def hide_leading_underscores?
    @hide_leading_underscores
  end
  
  def hide_leading_underscores!(val)
    @hide_leading_underscores = val
  end

  # Helper method to determine if a file is a Textile document.
    def textile_file?(file)
      if hide_leading_underscores?
        file =~ /\.textile$/i && file !~ /^_/
    else
      file =~ /\.textile$/i
    end        
    end    

  # Count the words in a Textile string. Used by count_words_file and (by
  # extension) count_words_files
  def count_words(filecontent)
    html_content = RedCloth.new(filecontent).to_html
        s = html_content.gsub(/<\/?[^>]*>/, "") # Strip HTML tags
        s = s.gsub(/\&\#[0123456789]+\;/, "")   # Strip HTML entity chars
        s.split.length
  end
  
    # Count the words in a single file. This can be invoked directly, or
    # from within count_words_multifiles.
    def count_words_file(file)
        if File.exists?(file)
            buff = IO.readlines(file).join("\n")
            fsize = count_words(buff)
            printf "%7d %s\n", fsize, file
            fsize
        else
            0
        end
    end
        
    # Count words in multiple files
    def count_words_files(filelist)
        total_count = 0

        filelist.each { 
          |file| 
            unless File.directory?(file)
                if textile_file?(file)
                    fsize = count_words_file(file)
                    total_count += fsize
                end
            end
        }
        if filelist.count > 1
            printf "%7d total\n", total_count
        end
    end
    
    def CommandLineDriver(args)
    if (args.count == 0)
      puts <<END
***************************************************************
#{$0}: Count words in one or more Textile files.
By Tammy Cravit, [email protected]
$Revision$
***************************************************************

Usage: #{$0} [-a] <file> [file] ...

The default behavior of the script is to ignore files whose 
names begin with an underscore. To include these files in the 
count, include the -a option.
END
    else
      if args.count == 1
        if File.exists?(args[0])
          count_words_file(args[0])
        else
          printf "%s: not found\n", args[0]
        end
      else
        if args[0] == "-a"
          hide_leading_underscores!(false)
          count_words_files(args.slice(1, args.count - 1))
        else
            count_words_files(args)
          end
      end
    end
  end
end

############################################################################
# Driver routine - display a message if no files are specified; 
# otherwise, count the words in the subset of specified files which
# are Textile files.
############################################################################

module Kernel
  
  # Runs the passed block of code, if the calling file is the one being executed.
  def on_execute
    calling_file = caller.first.split(':').first
    if File.expand_path(calling_file) == File.expand_path($0)
      yield
    end
  end
  
end

on_execute do
  TextileWordCounter.new.CommandLineDriver(ARGV)
end