Recode to UTF-8

recode_to_utf8.rb

#out = `find . -name "*.htm*" | xargs file -i`
out = (ARGV.size > 0 ? `file -i #{ARGV[0]}/*.htm*` : `file -i *.htm*`)
out.split("\n").each do |l|
  l =~ /(.*):(.*)charset=(.*)/ 
  filename = "#{$1}"
  charset = "#{$3}"
  charset = "windows-1252" if charset == "unknown-8bit"
  if (charset!= "binary" && charset != "us-ascii" && charset != "utf-8")
    succ = system("iconv -f #{charset} -t utf-8 #{filename} -o #{filename}.tmp")
    if !succ
      File.open("iconv_failed.txt", "a"){ |file| file.puts "#{filename}"}
      puts "#{filename}"
    end
    `mv #{filename}.tmp #{filename}`
   	#puts "#{filename} - #{charset}"
  end

  # append the html tag to force utf-8 encoding
  if (charset!= "binary")
    begin
      text = File.read("#{filename}")
      already_encoded = (text =~ /<META(.*)NAME="?GENERATOR"? CONTENT="(Microsoft FrontPage|Claris Home Page)(.*)">/i)
    rescue
      puts "@@ error with #{filename}!"
    end
    unless already_encoded.nil?
      replace = text.gsub!(/<META(.*)NAME="?GENERATOR"? CONTENT="(Microsoft FrontPage|Claris Home Page)(.*)">/i, "<meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\"/>")
      File.open("#{filename}", "w") { |file| file.puts replace }
    end

  #   begin
  #     text2 = File.read("#{filename}")
  #     utf_tag = (text2 =~ /<meta http-equiv="Content-Type" content="text\/html; charset=UTF-8"\/>/)
  #   rescue
  #     puts "@@@ error with #{filename}!"
  #   end
  #   if utf_tag.nil?
  #     puts "#{filename} does not have utf-8 tag!"
  #     `gedit #{filename} &`
  #   end
  end 
end