#
# Rubyのインストール場所がC:\RUBYならばC:\RUBY\lib\ruby\site_ruby
#
# サンプルプログラム
# require 'jcode'
# require 'html'
# $KCODE = 's'
# 
# html = HTMLtokens.new
# Dir.glob("*.htm*").each { |name|
#   html.load(name)
#   print "タイトル =\n"
#   p html.between("title")
#   print "コメント =\n"
#   p html.find("!--")
#   print "壁紙 =\n"
#   p html.findattr("body", "background")
#   print "画像 =\n"
#   p html.findattr("img", "src")
#   print "リンク =\n"
#   p html.findattr("a", "href")
# }

require 'jcode'
$KCODE = 's' if not $KCODE

class HTMLtokens < Struct.new(:tokens, :map)
  def initialize
    self.tokens = nil
    self.map = nil
  end
  def fetch(s)
    self.tokens = Array.new(0)
    stt = 0
    txt = ""
    tag = ""
    cmt = ""
    (0..(s.size - 1)).each { |i|
      c = s.slice(i, 1)
      # printf("%s", c) # DEBUG
      if stt == 0 then
        if c == "<" then
          stt = 1
  	self.tokens.push txt if txt.size > 0
  	tag = ""
        else
          txt = txt + c
        end
      elsif stt == 1 then
        if (tag == "") && (c == "!") then
          stt = 3
        elsif c == ">" then
          stt = 0
  	self.tokens.push "<" + tag.strip + ">"
  	txt = ""
        elsif c == '"'
          stt = 2
          tag = tag + '"'
        elsif c =~ /[ \t\n　]/
          if tag == "/"
  	  # nop
  	elsif (tag.size > 0) && (tag.slice(tag.size - 1, 1) != " ") then
  	  tag = tag + " "
  	end
        else
          tag = tag + c.downcase
        end
      elsif stt == 2 then
        if c == '"' then
          stt = 1
          tag = tag + '"'
        else
          tag = tag + c
        end
      elsif stt == 3 then # '<!'
        if c == "-" then
          stt = 4
        else # <!DOCTYPE>, etc.
          stt = 8
  	  tag = c
        end
      elsif stt == 4 then # '<!-'
        if c == "-" then
          stt = 5
          cmt = ""
        else # syntax error(comment)???
          stt = 8
  	  tag = "!-" + c
        end
      elsif stt == 5 then # '<!--'
        if c == "-" then
          stt = 6
        else # comment-body
          cmt = cmt + c
        end
      elsif stt == 6 then # '-'
        if c == "-" then
          stt = 7
        else # comment-body(?)
          stt = 5
          cmt = cmt + "-" + c
        end
      elsif stt == 7 then # '--'
        if c == ">" then
          stt = 0 
  	  self.tokens.push "<!--" + cmt + "-->"
        else # comment-body
          stt = 5
          cmt = cmt + "--" + c
        end
      elsif stt == 8 then # '<!DOCTYPE', etc.
        if c == ">" then
          stt = 0
          self.tokens.push "<!" + tag + ">"
        elsif c == '"'
          stt = 9
  	tag = tag + '"'
        else
          tag = tag + c
        end
      elsif stt == 9 then # '<!... "':quote
        if c == '"' then
          stt = 8
          tag = tag + '"'
        else
          tag = tag + c
        end
      end
    }
    return nil if stt != 0 # syntax error
    self.tokens.push txt if txt.size > 0 
    return true
  end
  def parse
    self.map = Hash.new(nil)
    stt = 0
    (0..(self.tokens.size - 1)).each { |i|
      if self.tokens[i] =~ /^<!--/ then
        tname = "!--"
  	self.map[tname] = Array.new(0) if not self.map[tname]
        self.map[tname].push [i, nil]
      elsif self.tokens[i] =~ /^<([^ >]*)/ then
        tname = Regexp.last_match(1)
        if tname.slice(0, 1) != "/" then
  	  self.map[tname] = Array.new(0) if not self.map[tname]
          self.map[tname].push [i, nil]
        else # </>
          tname = tname.slice(1, tname.size - 1)
  	  if self.map[tname] then
	    # printf("tname=%s:", tname); p self.map[tname] # DEBUG
  	    m = nil
  	    j = self.map[tname].size - 1
  	    j.downto(0) { |k|
	      if self.map[tname][k][0] && !self.map[tname][k][1] then
	        m = k
		break
              end
	    }
	    if m then
  	      self.map[tname][m][1] = i
	    else # ペアになっていない余分な</>
	      self.map[tname].push [nil, i]
	    end
  	  else # 一個だけ</>がある
	    self.map[tname] = Array.new(0)
	    self.map[tname].push [nil, i]
  	  end
	end
      end
    }
  end
  def read(html)
    return nil if not fetch(html)
    parse
    return true
  end
  def load(name)
    html = ""
    open(name, "r").each { |line|
      html = html + line.chomp
      html = html + "\n"
    }
    return nil if not fetch(html)
    parse
    return true
  end
  def find(tname)
    if block_given?
      proc = Proc.new
      self.map[tname].each { |cel|
        proc.call(cel[0], cel[1])
      } if self.map[tname]
    else
      v = Array.new(0)
      self.map[tname].each { |cel|
        if cel[0] && cel[1] && (cel[1] > cel[0]) then
	  ((cel[0])..(cel[1])).each { |i| v.push self.tokens[i] }
        elsif cel[0] && cel[1] then
	  v.push self.tokens[cel[0]]
	  v.push self.tokens[cel[1]]
	elsif cel[0] then
	  v.push self.tokens[cel[0]]
	elsif cel[1] then
	  v.push self.tokens[cel[1]]
	end
      } if self.map[tname]
      return v
    end
  end
  def between(tname)
    if block_given?
      proc = Proc.new
      self.map[tname].each { |cel|
        if cel[0] && cel[1] && ((cel[1] - cel[0]) > 1) then
	  ((cel[0] + 1)..(cel[1] - 1)).each { |i| proc.call(i) }
	end
      } if self.map[tname]
    else
      v = Array.new(0)
      self.map[tname].each { |cel|
        if cel[0] && cel[1] && ((cel[1] - cel[0]) > 1) then
	  ((cel[0] + 1)..(cel[1] - 1)).each { |i| v.push self.tokens[i] }
	end
      } if self.map[tname]
      return v
    end
  end
  def findattr(tname, aname)
    if block_given?
      proc = Proc.new
      find(tname) { |b, e|
        if b then
	  tag = self.tokens[b]
	  if tag =~ /^(.* #{aname}?=\")([^\"]*)(\".*)$/ then
	    l = Regexp.last_match(1)
	    m = Regexp.last_match(2)
	    r = Regexp.last_match(3)
	    m = proc.call(m)
	    self.tokens[b] = l + m + r
          end
	end
      }
    else
      v = Array.new(0)
      find(tname) { |b, e|
        if b then
	  tag = self.tokens[b]
	  tag =~ /^(.* #{aname}?=\")([^\"]*)(\".*)$/
	  m = Regexp.last_match(2)
	  v.push m
	end
      }
      return v
    end
  end
  def to_s
    s = ""
    self.tokens.each { |v| s = s + v }
    return s
  end
end