#
# xmlscan.rb
#
#   Copyright (C) Ueno Katsuhiro 2000,2001
#
# $Id: xmlscan.rb,v 1.8 2001/01/04 09:56:56 katsu Exp $
#

class XMLScanner

  class ParseError < StandardError; end

  class PrivateArray < Array
    private(*superclass.instance_methods)
  end

  class XMLSource < PrivateArray

    class PortWrapper

      def gets ; @port.gets ; end
      def lineno ; 0 ; end
      def path ; '-' ; end

      def initialize(port)
        @port = port
        unless port.respond_to? :gets then
          if port.is_a? Array then
            @n = -1
            def self.gets ; @port[@n += 1] ; end
            def self.lineno ; @n + 1 ; end
          else
            @port = port.to_s
            def self.gets ; s = @port ; @port = nil ; s ; end
          end
        end
        if port.respond_to? :lineno then
          def self.lineno ; @port.lineno ; end
        end
        if port.respond_to? :path then
          def self.path ; @port.path ; end
        end
      end

      def send_port(method, *args)
        @port.send(method, *args)
      end

      def self.wrap(port)
        if instance_methods.find { |i| not port.respond_to? i } then
          new port
        else
          port
        end
      end

    end


    module DummyPort
      def self.gets ; nil ; end
      def self.lineno ; 0 ; end
      def self.path ; '-' ; end
    end


    def initialize(port = nil)
      super()
      feed port
    end

    public

    def feed(port)
      if port then
        @port = PortWrapper.wrap(port)
      else
        @port = DummyPort
      end
      @eof = false
      @lineno_size, @lineno_count = -1, 0
      self
    end

    def abort
      @eof = true
      clear
      self
    end

    def pop
      if at(1) or first == '>' or @eof then  # at(1) == (size > 1)
        super
      else
        begin
          src = @port.gets
          unless src then
            @eof = true
            break
          end
          a = src.split(/(?=>?<)|>/, -1)
          a[0] = super << a[0] if last and (c = a[0][0]) != ?< and c != ?>
          concat a
        end until at(1)  # at(1) == (size > 1)
        reverse!
        @lineno_size = size - 1
        @lineno_count = super
      end
    end

    def tag_end?
      s = last and s[0] != ?<
    end

    def tag_start?
      s = last and s[0] == ?<
    end

    def eof?
      @eof and empty?
    end

    def lineno
      unless size == @lineno_size then
        @port.lineno
      else
        unless @lineno_count.is_a? Integer then
          if @lineno_count then
            @lineno_count = @lineno_count.sub(/\A\s+/, '').scan(/^/).size - 1
          else
            @lineno_count = 0
          end
        end
        @port.lineno - @lineno_count
      end
    end

    def path
      @port.path
    end

    def send_port(method, *args)
      if @port.is_a? PortWrapper then
        @port.send_port(method, *args)
      else
        @port.send(method, *args)
      end
    end

  end




  def initialize(port = nil)
    @src = XMLSource.new
    @prolog = false
    feed port if port
  end


  def feed(port)
    @src.feed port
    @prolog = true
    self
  end


  attr_reader :prolog
  alias in_prolog? prolog
  undef prolog


  private

  def on_error(path, lineno, msg)
    raise ParseError, sprintf('%s:%d: %s', path, lineno, msg)
  end

  def on_xmldecl(version, encoding, standalone)
  end

  def on_doctype(root, pubid, sysid)
  end

  def on_comment(strs)
  end

  def on_pi(target, pi)
  end

  def on_chardata(str)
  end

  def on_etag(name)
  end

  def on_stag(name, attr)
  end

  def on_emptyelem(name, attr)
  end

  def on_entityref(ref)
    s = entityref_literal(ref)
    on_chardata s if s
  end

  def on_charref(ref)
    s = parse_charref(ref)
    on_chardata s if s
  end

  def on_eof
  end


  private

  def parse_error(msg)
    on_error @src.path, @src.lineno, msg
  end


  PredefinedEntity = {
    'lt' => '<',     'gt' => '>',   'quot' => '"',
    'apos' => '\'',  'amp' => '&',
  }

  def charref_literal(code)
    [code].pack('N').sub(/\A\000+/mn, '')
  end

  def entityref_literal(ref)
    PredefinedEntity[ref] or
      begin
        parse_error "undefined general entity `#{ref}'"
        nil
      end
  end


  def parse_charref(ref)
    if /\A#(\d+)\z/ =~ ref then
      charref_literal $1.to_i
    elsif /\A#x([\dA-Fa-f]+)\z/ =~ ref then
      charref_literal $1.hex
    else
      parse_error "parse error at `&#{ref}'"
      nil
    end
  end


  def on_attribute_value(key, val)
    inc = 0
    val.gsub!(/&([^;\s<>]+)?;?/) { |m|
      if m[-1] == ?; and (s = $1) then
        if s[0] == ?\# then
          rep = parse_charref(s)
        else
          rep = entityref_literal(s)
        end
        unless rep then
          @unexpanded_entityrefs = [] unless defined? @unexpanded_entityrefs
          @unexpanded_entityrefs.push [ key.dup, $~.begin(0) + inc, s ]
          rep = ''
        end
        inc += rep.size - m.size
        rep
      else
        parse_error "parse error at `#{m}'"
        m
      end
    }
    true
  end


  def scan_content(s)
    while true
      unless /&/ =~ s then
        on_chardata s
      else
        on_chardata s unless (s = $`).empty?
        $'.split(/&/, -1).each { |i|
          unless /;/ =~ i then
            parse_error "parse error at `&#{i.split(/\b|\s/,2)[0]}'"
            on_chardata('&' << i)
            next
          end
          e, i = $`, $'
          if /\s/ =~ e then
            parse_error "parse error at `&#{$`}'"
            on_chardata('&' << i)
            next
          end
          if e[0] == ?\# then
            parse_charref e
          else
            on_entityref e
          end
          on_chardata i unless i.empty?
        }
      end
      break if @src.tag_start?
      s = @src.pop
      break unless s
      s[0,0] = '>' unless s == '>'
    end
  end


  def scan_comment(s)
    s[0,4] = ''   # <!--
    comm = [ s ]
    until /--/ =~ s
      s = @src.pop
      unless s then
        parse_error "unterminated comment meets EOF"
        return on_comment(comm)
      end
      comm.push '>' if (c = s[0]) != ?< and c != ?>
      comm.push s
    end
    if @src.tag_end? and $'.strip.empty? then
      comm[-1] = $`
    else
      parse_error "comment is not terminated by `-->'"
      until @src.tag_end? and (s = @src.pop)
        comm.push '>' if (c = s[0]) != ?< and c != ?>
        comm.push s
      end
    end
    on_comment comm
  end


  def scan_pi(s)
    unless /\A<\?(\S+)(?:\s+|(?=\?\z))/ =~ s then
      parse_error "parse error at `<?'"
      on_chardata s
    else
      target, pi = $1, $'
      until @src.tag_end? and pi[-1] == ??
        s = @src.pop
        unless s then
          parse_error "unterminated PI meets EOF"
          return on_pi(name, pi)
        end
        pi << '>' if (c = s[0]) != ?< and c != ?>
          pi << s
      end
      pi.chop!
      on_pi target, pi
    end
  end


  def scan_cdata(s)
    cdata = [ s ]
    until @src.tag_end? and s[-1] == ?] and s[-2] == ?]
      s = @src.pop
      unless s then
        parse_error "unterminated CDATA section meets EOF"
        return on_chardata(cdata.join)
      end
      cdata.push '>' if (c = s[0]) != ?< and c != ?>
      cdata.push s
    end
    s.chop!.chop!
    on_chardata cdata.join
  end


  def unclosed_tag(t)
    if @src.eof? then      # <tag[EOF]
      parse_error "unclosed #{t} meets EOF"
    else                   # <tag <tag
      parse_error "unclosed #{t} meets another tag"
    end
  end


  def read_until(re, dst, t)
    begin
      s = @src.pop
      unless s then
        parse_error "unterminated #{t} meets EOF"
        return ''
      end
      dst << '>' if (c = s[0]) != ?< and c != ?>
      v, s = s.split(re, 2)
      dst << v
    end until s
    s
  end


  def scan_etag(s)
    s[0,2] = ''  # </
    if /\s/ =~ s then
      s1, s2 = $`, $'
      if s1.empty? then    # </ tag
        parse_error "parse error at `</'"
        return on_chardata('</' + s)
      elsif not s2.strip.empty? then     # </ta g
        parse_error "parse error at `</#{s1}'"
      end
      s = s1
    elsif s.empty? then
      if @src.tag_end? then
        parse_error "found an empty end tag `</>'"
      else
        parse_error "parse error at `</'"
        return on_chardata('</')
      end
    end
    unclosed_tag 'end tag' unless @src.tag_end?
    on_etag s
  end


  def scan_stag(s)
    attr = {}
    method = :on_stag
    unless /(?=[\/\s])/ =~ s then
      name = s
      name[0,1] = ''
      if name.empty? then   # << or <>
        if @src.tag_end? then
          parse_error "found an empty start tag `<>'"
        else
          parse_error "parse error at `<'"
          return on_chardata('<' + s)
        end
      end
    else
      name = $`
      s = $'
      name[0,1] = ''
      if name.empty? then   # < tag
        parse_error "parse error at `<'"
        return on_chardata('<' + s)
      end
      begin
        complete = true
        s.scan(/\s+(?:([^\s\/=]+)\s*=\s*('[^']*'?|"[^"]*"?)|\z)|\s*(\/\z)|\s*(.[^='"\s]*)/m
               ) { |key,val,e,err|
          if key then
            qmark = val.slice!(0,1)
            if val[-1] == qmark[0] then
              val.chop!
            else
              complete = false
              re = /#{qmark}/
              begin
                s = @src.pop
                if not s then
                  parse_error "unterminated #{t} meets EOF"
                  complete = true
                  break
                elsif (c = s[0]) == ?< then
                  parse_error "`<' is found in attribute value"
                elsif c != ?> then
                  val << '>'
                end
                v, s = s.split(re, 2)
                val << v
              end until s
              # always break here.
            end
            if on_attribute_value(key, val) then
              parse_error "doubled attribute `#{key}'" if attr.key? key
              attr[key] = val
            end
          elsif e then
            method = :on_emptyelem
          elsif err then
            parse_error "parse error at `#{err.split(/\b|\s/,2)[0]}'"
          end
        }
      end until complete
    end
    unclosed_tag 'start tag' unless @src.tag_end?
    send method, name, attr
  end


  def parse_internal_dtd(s)
    parse_error "internal DTD subset is not supported"
  end


  DOCTYPEPattern =
    /\A([^\s\["']+)(?:\s+(?:SYSTEM|PUBLIC\s+("[^"]*"|'[^"']*'))\s+("[^"]*"?|'[^']*'?))\s*/

  def scan_doctype(s)
    unless DOCTYPEPattern =~ s then
      parse_error "parse error in DOCTYPE"
      return
    end
    root, pubid, sysid, s = $1, $2, $3, $'
    if pubid then
      pubid.chop!
      pubid[0,1] = ''
    end
    if sysid then
      c = sysid.slice!(0,1)
      if c[0] == sysid[-1] then
        sysid.chop!
      else
        s = read_until(/#{c}\s*/, sysid, 'DOCTYPE')
      end
    end
    if s[0] == ?[ then
      s[0,1] = ''
      parse_internal_dtd s
    elsif not s.empty? then
      parse_error "parse error at `#{s.split(/\b|\s/,2)[0]}'"
    end
    unclosed_tag 'DOCTYPE' unless @src.tag_end?
    on_doctype root, pubid, sysid
  end


  def scan_bang_tag(s)
    parse_error "parse error at `<!'"
    on_chardata s
  end


  def scan_text(s)
    if (c = s[0]) == ?< then
      if (c = s[1]) == ?/ then
        scan_etag s
      elsif c == ?! then
        if s[2] == ?- and s[3] == ?- then
          scan_comment s
        elsif /\A<!\[CDATA\[/ =~ s then
          scan_cdata $'
        else
          scan_bang_tag s
        end
      elsif c == ?? then
        scan_pi s
      else
        scan_stag s
      end
    elsif c == ?> then
      scan_text @src.pop
    else
      scan_content s
    end
  end


  XMLDeclPattern, TextDeclPattern = instance_eval {
    version = '\\s+version\\s*=\\s*("[^"\']+"|\'[^"\']+\')'
    encoding = '\\s+encoding\\s*=\\s*("[^"\']+"|\'[^"\']+\')'
    standalone = '\\s+standalone\\s*=\\s*("[^"]+"|\'[^\']+\')'
    [ /\A<\?xml#{version}(?:#{encoding}(?:#{standalone})?)?\s*\?\z/,
      /\A<\?xml(?:#{version})?#{encoding}\s*\?\z/
    ]
  }

  def scan_prolog
    s = @src.pop
    if s and /\A<\?xml\b/ =~ s then
      unless XMLDeclPattern =~ s then
        parse_error 'parse error in XML declaration'
      else
        version, encoding, standalone = $1, $2, $3
        version.chop!
        version[0,1] = ''
        if encoding then
          encoding.chop!
          encoding[0,1] = ''
          encoding.downcase!
        end
        if standalone then
          standalone.chop!
          standalone[0,1] = ''
          if standalone == 'yes' then
            standalone = true
          elsif standalone == 'no' then
            standalone = false
          else
            parse_error 'invalid standalone document declaration'
            standalone = nil
          end
        end
        on_xmldecl version, encoding, standalone
        unclosed_tag 'XML declaration' unless @src.tag_end?
        s = @src.pop
      end
    end
    while s
      if s[0] == ?< then
        if (c = s[1]) == ?! then
          if s[2] == ?- and s[3] == ?- then
            scan_comment s
          elsif /\A<!DOCTYPE\s+/ =~ s then
            scan_doctype $'
          else
            break
          end
        elsif c == ?? then
          scan_pi s
        else
          break
        end
      elsif s.strip.empty? and @src.tag_start? then
        on_chardata s
      else
        break
      end
      s = @src.pop
    end
    @prolog = false
    s and scan_text(s)
  end


  public

  def step
    if @prolog then
      scan_prolog
    elsif @src.eof? then
      ret = on_eof
      @src.feed nil
      ret
    elsif s = @src.pop then
      scan_text s
    else
      nil
    end
  end


  def parse(src = nil)
    feed src if src
    scan_prolog if @prolog
    while s = @src.pop
      scan_text s
    end
    on_eof
    self
  end




  module Loose

    def parse_error(msg)
      STDERR.printf "parse error:%s:%d: %s\n", path, lineno, msg if $VERBOSE
    end
    private :parse_error

  end



  module Recoverable

    def feed(*args)
      @errors = []
      super
    end

    private

    def on_error(path, lineno, msg)
      @errors.push sprintf('%s:%d: %s', path, lineno, msg)
    end

    def error?
      @errors.empty?
    end

    def scan_text(s)
      raise ParseError, @errors.shift unless @errors.empty?
      super
    end

    public

    def step
      ret = super
      raise ParseError, @errors.shift unless @errors.empty?
      ret
    end

    def parse(*args)
      ret = super
      raise ParseError, @errors.shift unless @errors.empty?
      ret
    end

  end



  ## for external general parsed entities

  module ExternalEntity

    def scan_prolog
      s = @src.pop
      if s and /\A<\?xml\b/ =~ s then
        unless TextDeclPattern =~ s then
          parse_error 'parse error in text declaration'
        else
          version, encoding = $1, $2
          if encoding then
            encoding.chop!
            encoding[0,1] = ''
            encoding.downcase!
          end
          ret = on_xmldecl(version, encoding, nil)
          unclosed_tag 'text declaration' unless @src.tag_end?
          @prolog = false
          return ret
        end
      end
      @prolog = false
      s and scan_text(s)
    end

  end

end



class LooseXMLScanner < XMLScanner
  include Loose
end

class RecoverableXMLScanner < XMLScanner
  include Recoverable
end




class WellFormedXMLScanner < XMLScanner

  class ElementStack < PrivateArray

    def no_element?
      first.nil?       # empty? or first.nil?
    end

    def root_found?
      not empty?
    end

    def push_element(name)
      if first or empty? then
        push name
      else
        pop
        push name
        nil
      end
    end

    def pop_element(name)
      if name == last then
        pop
        push nil if empty?
        self
      else
        nil
      end
    end

    alias current last
    public :current

    def each
      reverse_each { |i| yield i if i }
    end

  end



  def feed(*args)
    @elemstack = type::ElementStack.new
    @standalone = false
    super
  end


  private

  def on_start_element(name, attr)
  end

  def on_end_element(name)
  end

  def on_text(str)
  end


  def name_in_errmsg(name)
    name
  end


  def on_xmldecl(version, encoding, standalone)
    @standalone = standalone
  end

  def on_doctype(root, pubid, sysid)
    @standalone = false if @standalone.nil? and (pubid or sysid)
  end

  def scan_content(s)
    @__xml_text__ = ''
    super
    on_text @__xml_text__
  end

  def on_chardata(str)
    unless @elemstack.no_element? then
      @__xml_text__ << str
    else
      unless str.strip.empty? then
        parse_error "character data are found out of root element"
      end
    end
  end

  # def on_entityref(ref)
  # def on_charref(ref)

  def on_stag(name, attr)
    if @elemstack.push_element name then
      on_start_element name, attr
    else
      parse_error "another root element `#{name_in_errmsg(name)}'"
      @src.abort
    end
  end

  def on_etag(name)
    unless @elemstack.pop_element name then
      parse_error "element type `#{name_in_errmsg(name)}' is not matched"
    else
      on_end_element name
    end
  end

  def on_emptyelem(name, attr)
    if @elemstack.push_element name then
      on_start_element name, attr
      on_end_element name
      @elemstack.pop_element name
    else
      parse_error "another root element `#{name_in_errmsg(name)}'"
      @src.abort
    end
  end

  def on_eof
    if not @elemstack.root_found? then
      parse_error "no root element was found"
    elsif not @elemstack.no_element? then
      @elemstack.dup.each { |name|
        parse_error "unclosed element `#{name_in_errmsg(name)}' meets EOF"
        on_end_element name
        @elemstack.pop_element name
      }
    end
  end


  def entityref_literal(ref)
    PredefinedEntity[ref] or (@standalone and super)
  end

  def on_attribute_value(key, val)
    val.gsub!(/\r\n|\s/, ' ')
    super
  end

end




class XMLScannerWithNamespace < WellFormedXMLScanner

  PredefinedNamespace = {
    'xml' => 'http://www.w3.org/XML/1998/namespace',
  }

  class ElementStack < superclass::ElementStack

    def initialize
      super
      @namespace = {}
    end

    attr_reader :namespace

    def default_namespace
      @namespace[:default]
    end

    def get_namespace(name)
      @namespace[name] or PredefinedNamespace[name]
    end

    def set_namespace(name, uri)
      push [ name, @namespace[name] ]
      if uri.empty? then
        @namespace.delete name
      else
        @namespace[name] = uri
      end
    end

    def pop_element(name)
      if name == last then
        pop
        while log = last and not log[2]
          pop
          @namespace[log[0]] = log[1]
        end
        push nil unless log
        self
      else
        nil
      end
    end

    def each
      reverse_each { |i| yield i if i[2] }
    end

  end



  private

  def expand_qualified_name(name, default = nil)
    unless /:/ =~ name then
      [ default, nil, name ]
    else
      prefix, localpart = $`, $'
      if localpart.empty? then
        parse_error "parse error at `:'"
        return [ nil, nil, name ]
      elsif /:/ =~ localpart then
        parse_error "localpart `#{localpart}' includes a colon"
      end
      unless namespace = @elemstack.get_namespace(prefix) then
        parse_error "undeclared namespace `#{prefix}'"
        namespace = nil
      end
      [ namespace, prefix, localpart ]
    end
  end

  def expand_attr_namespace(attr)
    dst = {}
    attr.each { |key,val|
      namespace, prefix, name = expand_qualified_name(key)
      h = dst[namespace]
      dst[namespace] = h = {} unless h
      h[name] = val
    }
    dst
  end

  def expand_attr_namespace_2(attr)
    attr.keys.each { |key|
      namespace, prefix, name = expand_qualified_name(key)
      if namespace then
        k = namespace + ' ' + name
      else
        k = ' ' + name
      end
      attr[k] = attr.delete(key)
    }
    attr
  end

  def name_in_errmsg(name)
    if name[1] then
      "#{name[1]}:#{name[2]}"
    else
      name[2]
    end
  end

  def on_stag(name, attr)
    super(expand_qualified_name(name, @elemstack.default_namespace),
          expand_attr_namespace(attr))
  end

  def on_emptyelem(name, attr)
    super(expand_qualified_name(name, @elemstack.default_namespace),
          expand_attr_namespace(attr))
  end

  def on_etag(name)
    super expand_qualified_name(name, @elemstack.default_namespace)
  end

  def on_pi(target, pi)
    parse_error "PI target must not include `:'" if /:/ =~ target
  end

  def on_attribute_value(key, val)
    super
    f = nil
    if key == 'xmlns' or f = (key[0,6] == 'xmlns:') then
      if f then
        name = key[6..-1]
        if name.empty? then
          parse_error "parse error at `:'"
        elsif /:/ =~ name then
          parse_error "namespace name `#{name}' includes a colon"
        elsif name[0,3].downcase == 'xml' then
          parse_error "prefix `#{name}' is reserved"
        elsif /\s/ =~ val then
          parse_error "invalid namespace `#{val}'"
        elsif val.empty? then
          parse_error "null namespace is declared as `#{name}'"
        else
          @elemstack.set_namespace name, val
        end
      else
        @elemstack.set_namespace :default, val
      end
      false
    else
      true
    end
  end

end


=begin

supported by XMLScanner

Well-Formedness Constraint: 属性指定の一意性
開始タグ又は空要素タグでは，同一の属性名が二回以上出現してはならない。

Well-Formedness Constraint: 属性値に<を含まないこと
属性値内で直接的又は間接的に参照する実体の置換テキストには，<を
含んではならない。


supported by WellFormedXMLScanner

Well-Formedness Constraint: 要素型のマッチ
要素の終了タグの名前は，その要素の開始タグにおける要素型（の名前）とマッチしな
ければならない。

Well-Formedness Constraint: 実体が宣言されていること
DTDをもたない文書，パラメタ実体参照を含まない内部DTDサブセットだけをもつ文書，
又は "standalone='yes'" をもつ文書において，実体参照で用いる Name は，
外部サブセット及びパラメタ実体内以外に現れる実体宣言に含まれる名前と
マッチしなければならない。ただし，整形式の文書は，実体amp,
lt, gt, apos, quot を宣言する必要はない。一般実体の場合は，属性リスト宣言の
デフォルト値内での参照より先に，宣言が現れなければならない。外部サブセット又は
外部パラメタ実体で実体を宣言するとき，妥当性を検証しないプロセサが，宣言を
読み，処理することを義務づけないことに注意。それらの文書では，実体は
宣言されなければならないという規則は，standalone='yes'の場合のみ，
整形式制約となる。


supported by XMLScanner with xmldtd.rb

Well-Formedness Constraint: 内部サブセット内のパラメタ実体
DTDの内部サブセットでは，パラメタ実体参照は，マーク付け宣言が出現可能な場所だけ
に出現できる。マーク付け宣言の一部としては出現できない。この制約は，外部パラメ
タ実体又は外部サブセットでの参照には適用しない。

Well-Formedness Constraint: DTDの中
パラメタ実体参照は，DTD内にだけ，出現してよい。


supported by XMLParsedEntity

Well-Formedness Constraint: 再帰なし
解析対象実体は，それ自体への参照を，直接にも間接にも含んではならない。




Well-Formedness Constraint: 外部実体への参照がないこと
属性値には，外部実体への直接的又は間接的な参照を含むことはできない。

Well-Formedness Constraint: 使用できる文字
文字参照で参照する文字は，Charの生成規則にマッチしなければならない。

Well-Formedness Constraint: 解析対象実体
実体参照は，解析対象外実体の名前を含んでいてはならない。解析対象外実体は，
ENTITY型又はENTITIES 型として宣言した属性値としてだけ参照できる。

=end






## for internal general parsed entities

class XMLParsedEntity < WellFormedXMLScanner

  def initialize(name, src)
    @name = name
    super src
  end

  #undef feed

  private

end




## for external general parsed entities

class ExtXMLScanner < XMLScanner

  def scan_prolog
    s = @src.pop
    if s and /\A<\?xml\b/ =~ s then
      unless TextDeclPattern =~ s then
        parse_error 'parse error in text declaration'
      else
        version, encoding = $1, $2
        if encoding then
          encoding.chop!
          encoding[0,1] = ''
          encoding.downcase!
        end
        ret = on_xmldecl(version, encoding, nil)
        unclosed_tag 'text declaration' unless @src.tag_end?
        @prolog = false
        return ret
      end
    end
    @prolog = false
    s and scan_text(s)
  end

end





if __FILE__ == $0 then
  #class TestScanner < XMLScanner
  #class TestScanner < WellFormedXMLScanner
  class TestScanner < XMLScannerWithNamespace
    def on_error(path, lineno, msg)
      STDERR.printf "%s:%d: %s\n", path, lineno, msg
    end
  end
  STDOUT.sync = STDERR.sync = true

  if /\A--?\z/ === ARGV[0] then
    if ARGV.shift == '--' and ARGV.size == 1 then
      p = IO.popen("diff -u #{ARGV[0]} -", 'w')
      STDOUT.reopen p
      class Hash
        def []=(k,v)
          (@a ||= []).push [ k, v ]
        end
        def each(&b)
          @a.each(&b) if defined? @a
        end
      end
    end
    class TestScanner
      $".push 'xmlscan.rb'
      require 'xmltoken'
      def self.def_handler(*name)
        name.each { |i|
          eval %{
            def on_#{i.downcase}(*a)
              super
              print Tokenizer::#{i}.new(*a).to_s
            end
          }
        }
      end
      def_handler 'CharData', 'Comment', 'PI', 'XMLDecl', 'Doctype'
      def_handler 'ETag', 'STag', 'EmptyElem'
    end
  end

  src = ARGF.read
  scan = TestScanner.new
  t1 = Time.times.utime
  scan.parse(src)
  t2 = Time.times.utime
  STDERR.printf "%2.3f sec\n", t2 - t1
end