#==============================================================================
# ■ htmlcheck.rb  ver 0.90
#------------------------------------------------------------------------------
# 　開き忘れたり閉じ忘れたりしたタグを無効にします。
#   簡易版なので、要素の破壊は行えません。
#==============================================================================

#==============================================================================
# ■ CGIsystem
#==============================================================================
module CGIsystem
  #--------------------------------------------------------------------------
  # ● 使用可能タグ指定
  #--------------------------------------------------------------------------
  def self.can_use_tag(tag = "")
    # ここで使えるタグの指定をします。初期状態ではほぼフルに使えます。
    base = ["p", "strong", "em", "del", "tt", "span", "div", "pre", 
    "font", "u", "sup", "sub", "h1", "h2", "h3", "h4", "h5", "h6", "ins", "q", 
    "blockquote", "a", "ul", "ol", "dl", "table"]
    case tag.downcase
    when "table"
      return (base + ["th", "tr"])
    when "tr"
      return (base + ["td"])
    when "ul", "ol"
      return (base + ["li"])
    when "dl"
      return (base + ["dt", "dd"])
    else
      return base
    end
  end
end

#==============================================================================
# ■ String
#==============================================================================
class String
  #--------------------------------------------------------------------------
  # ● 欠損タグの破壊
  #--------------------------------------------------------------------------
  def htmlcheck(tag)
    @text = self
    @text.gsub!(/(\r\n|\r|\n)/){"<br>"}
    @text.gsub!(/\\/){"\\\\"}
    if tag
      @stack = []
      @tag = []
      @save = []
      @tagset = ""
      @text.gsub!(/([^<]\/)\//im){""}
      @text.gsub!(/<(br([ \n]+?clear=\"(left|right|all)\")?)>/im){"\000#{$1}\001"}
      @text.gsub!(/<(nobr)>/im){"\000#{$1}\001"}
      @text.gsub!(/<(\?php[ \n].+[ \n]\?)>/im){"\000#{$1}\001"}
      @text.gsub!(/<(img(([ \n][^<>]+?=\"[^<>]+\")+)?)>/im){"\000#{$1}\001"}
      while @text =~ /.*?<([^\/<>]+?(([ \n].[^<>]+?=\".[^<>]+\")+)? *?|\/[^\/<>]+?>)/im
        self.stackset
      end
      for s in @save
        @text.sub!(/#{s}\000(([^\/<>]+?)(([ \n][^<>]+?=\"[^<>]+\")+)? *?)\001/im){"#{s}<#{$1}>"}
      end
    end
    @text.gsub!(/&/){"&amp;"}
    @text.gsub!(/( |&nbsp;) /i){"#{$1}&nbsp;"}
    @text.gsub!(/</){"&lt;"}
    @text.gsub!(/(\002)/){"&lt;"}
    @text.gsub!(/>/){"&gt;"}
    @text.gsub!(/(\003)/){"&gt;"}
    @text.gsub!(/(\000)/){"<"}
    @text.gsub!(/(\001)/){">"}
    return @text
  end
  #--------------------------------------------------------------------------
  # ● 欠損タグの破壊(ループ部分)
  #--------------------------------------------------------------------------
  def stackset
    if @text =~ /(<([^\/<>]+)(([ \n][^<>]+?=\"[^<>]+\")+)?.*?>|<\/([^<>]+?)>)/im
      if $5 != nil
        if @stack.last == $5.downcase
          @text.sub!(/(#{@save.last})\000(#{@tag.last})\001(.*?)<\/(#{$5})>/im){"#{$1}\000#{$2}\001#{$3}\000/#{$4}\001"}
          @tagset = "#{$1}\000#{$2}\001#{$3}\000/#{$4}\001"
          @save.pop
          @stack.pop
          @tag.pop
        elsif @text =~ /(#{@save.last})\000(#{@tag.last})\001<\/(#{@stack.last})>/im
          @text.sub!(/(#{$1})\000(#{$2})\001<\/(#{$3})>/im){"#{$1}\002#{$2}\003\002\/#{$3}\003"}
        else
          @text.sub!(/(.*?)<\/([^<>]+?)>/im){"#{$1}\002/#{$2}\003"}
        end
      else
        if CGIsystem.can_use_tag("#{@stack.last}").include?($2.downcase)
          @text.sub!(/(.*?)<(([^\/<>]+?)(([ \n][^<>]+?=\"[^<>]+\")+)?)>/im){"#{$1}\000#{$2}\001"}
          @save.push($1)
          @stack.push($3.downcase)
          @tag.push($2)
        else
          @text.sub!(/(.*?)<(([^\/<>]+?)(([ \n][^<>]+?=\"[^<>]+\")+)?)>/im){"#{$1}\002#{$2}\003"}
        end
      end
    else
      @text.sub!(/(.*?)<((.+?)(([ \n].+?=\".+\")+)?)>/im){"#{$1}\002#{$2}\003"}
    end
  end
end