Attachment #133178 for bug #195505




require 'rexml/parseexception'
require 'rexml/source'
require 'set'

module REXML
  module Parsers

    # Nat Price gave me some good ideas for the API.
    class BaseParser
      NCNAME_STR= '[\w:][\-\w\d.]*'
      NAME_STR= "(?:(#{NCNAME_STR}):)?(#{NCNAME_STR})"
      UNAME_STR= "(?:#{NCNAME_STR}:)?#{NCNAME_STR}"

      NAMECHAR = '[\-\w\d\.:]'
      NAME = "([\\w:]#{NAMECHAR}*)"


      DOCTYPE_START = /\A\s*<!DOCTYPE\s/um
      DOCTYPE_PATTERN = /\s*<!DOCTYPE\s+(.*?)(\[|>)/um
      ATTRIBUTE_PATTERN = /\s*(#{NAME_STR})\s*=\s*(["'])(.*?)\4/um
      COMMENT_START = /\A<!--/u
      COMMENT_PATTERN = /<!--(.*?)-->/um
      CDATA_START = /\A<!\[CDATA\[/u

      XMLDECL_PATTERN = /<\?xml\s+(.*?)\?>/um
      INSTRUCTION_START = /\A<\?/u
      INSTRUCTION_PATTERN = /<\?(.*?)(\s+.*?)?\?>/um
      TAG_MATCH = /^<((?>#{NAME_STR}))\s*((?>\s+#{UNAME_STR}\s*=\s*(["']).*?\5)*)\s*(\/)?>/um
      CLOSE_MATCH = /^\s*<\/(#{NAME_STR})\s*>/um

      VERSION = /\bversion\s*=\s*["'](.*?)['"]/um

        @tags = []
        @stack = []
        @entities = []
        @nsstack = []
      end

      def position

            return [ :processing_instruction, *@source.match(INSTRUCTION_PATTERN, true)[1,2] ]
          when DOCTYPE_START
            md = @source.match( DOCTYPE_PATTERN, true )
            @nsstack.unshift(curr_ns=Set.new)
            identity = md[1]
            close = md[2]
            identity =~ IDENTITY

                val = attdef[3]
                val = attdef[4] if val == "#FIXED "
                pairs[attdef[0]] = val
                if attdef[0] =~ /^xmlns:(.*)/
                  @nsstack[0] << $1
                end
              end
            end
            return [ :attlistdecl, element, pairs, contents ]

        begin
          if @source.buffer[0] == ?<
            if @source.buffer[1] == ?/
              @nsstack.shift
              last_tag = @tags.pop
              #md = @source.match_to_consume( '>', CLOSE_MATCH)
              md = @source.match( CLOSE_MATCH, true )

                raise REXML::ParseException.new("missing attribute quote", @source) if @source.match(MISSING_ATTRIBUTE_QUOTES )
                raise REXML::ParseException.new("malformed XML: missing tag start", @source) 
              end
              attributes = {}
              prefixes = Set.new
              prefixes << md[2] if md[2]
              @nsstack.unshift(curr_ns=Set.new)
              if md[4].size > 0
                attrs = md[4].scan( ATTRIBUTE_PATTERN )
                raise REXML::ParseException.new( "error parsing attributes: [#{attrs.join ', '}], excess = \"#$'\"", @source) if $' and $'.strip.size > 0
                attrs.each { |a,b,c,d,e| 
                  if b == "xmlns"
                    if c == "xml"
                      if d != "http://www.w3.org/XML/1998/namespace"
                        msg = "The 'xml' prefix must not be bound to any other namespace "+
                        "(http://www.w3.org/TR/REC-xml-names/#ns-decl)"
                        raise REXML::ParseException.new( msg, @source, self )
                      end
                    elsif c == "xmlns"
                      msg = "The 'xmlns' prefix must not be declared "+
                      "(http://www.w3.org/TR/REC-xml-names/#ns-decl)"
                      raise REXML::ParseException.new( msg, @source, self)
                    end
                    curr_ns << c
                  elsif b
                    prefixes << b unless b == "xml"
                  end
                  attributes[a] = e 
                }
              end
        
              # Verify that all of the prefixes have been defined
              for prefix in prefixes
                unless @nsstack.find{|k| k.member?(prefix)}
                  raise UndefinedNamespaceException.new(prefix,@source,self)
                end
              end

              if md[6]
                @closed = md[1]
                @nsstack.shift
              else
                @tags.push( md[1] )
              end
              attributes = {}
              attrs.each { |a,b,c| attributes[a] = c }
              return [ :start_element, md[1], attributes ]
            end
          else

            # return PullEvent.new( :text, md[1], unnormalized )
            return [ :text, md[1] ]
          end
        rescue REXML::UndefinedNamespaceException
          raise
        rescue REXML::ParseException
          raise
        rescue Exception, NameError => error




              return
            when :start_element
              tag_stack.push(event[1])
              el = @build_context = @build_context.add_element( event[1], event[2] )
              @build_context = @build_context.add_element( event[1], event[2] )
            when :end_element
              tag_stack.pop
              @build_context = @build_context.parent

          end
        rescue REXML::Validation::ValidationException
          raise
        rescue REXML::UndefinedNamespaceException
          raise
        rescue
          raise ParseException.new( $!.message, @parser.source, @parser, $! )
        end

Lines 66-71 Link Here

(-)lib/rexml/document.rb (-1 / +2 lines)
66	def add( child )	66	def add( child )
67	if child.kind_of? XMLDecl	67	if child.kind_of? XMLDecl
68	@children.unshift child	68	@children.unshift child
		69	child.parent = self
69	elsif child.kind_of? DocType	70	elsif child.kind_of? DocType
70	# Find first Element or DocType node and insert the decl right	71	# Find first Element or DocType node and insert the decl right
71	# before it. If there is no such node, just insert the child at the	72	# before it. If there is no such node, just insert the child at the
Lines 183-189 Link Here
183	output = Output.new( output, xml_decl.encoding )	184	output = Output.new( output, xml_decl.encoding )
184	end	185	end
185	formatter = if indent > -1	186	formatter = if indent > -1
186	if transitive	187	if trans
187	REXML::Formatters::Transitive.new( indent, ie_hack )	188	REXML::Formatters::Transitive.new( indent, ie_hack )
188	else	189	else
189	REXML::Formatters::Pretty.new( indent, ie_hack )	190	REXML::Formatters::Pretty.new( indent, ie_hack )




    def attribute( name, namespace=nil )
      prefix = nil
      prefix = namespaces.index(namespace) if namespace
      prefix = nil if prefix == 'xmlns'
      attributes.get_attribute( "#{prefix ? prefix + ':' : ''}#{name}" )
    end


    #   Source (see Element.initialize).  If not supplied or nil, a
    #   new, default Element will be constructed
    # Returns:: the added Element
    #  a = Element.new('a')
    #  a.elements.add(Element.new('b'))  #-> <a><b/></a>
    #  a.elements.add('c')               #-> <a><b/><c/></a>
    def add element=nil
      rv = nil
      if element.nil?
        Element.new("", self, @element.context)
      elsif not element.kind_of?(Element)
        Element.new(element, self, @element.context)
      else
        @element << element
        element.context = @element.context

Line 0 Link Here

(-)lib/rexml/undefinednamespaceexception.rb (+8 lines)
	1	require 'rexml/parseexception'
	2	module REXML
	3	class UndefinedNamespaceException < ParseException
	4	def initialize( prefix, source, parser )
	5	super( "Undefined prefix #{prefix} found" )
	6	end
	7	end
	8	end

Lines 17-24 Link Here

(-)lib/rexml/source.rb (-2 / +2 lines)
17	elsif arg.kind_of? Source	17	elsif arg.kind_of? Source
18	arg	18	arg
19	else	19	else
20	raise "#{source.class} is not a valid input stream. It must walk \n"+	20	raise "#{arg.class} is not a valid input stream. It must walk \n"+
21	"like either a String, IO, or Source."	21	"like either a String, an IO, or a Source."
22	end	22	end
23	end	23	end
24	end	24	end

Lines 50-56 Link Here

(-)lib/rexml/attribute.rb (-1 / +1 lines)
50	@element = first.element	50	@element = first.element
51	end	51	end
52	elsif first.kind_of? String	52	elsif first.kind_of? String
53	@element = parent if parent.kind_of? Element	53	@element = parent
54	self.name = first	54	self.name = first
55	@normalized = second.to_s	55	@normalized = second.to_s
56	else	56	else

Lines 1-3 Link Here

(-)lib/rexml/rexml.rb (-6 / +8 lines)
		1	# -- encoding: utf-8 --
1	# REXML is an XML toolkit for Ruby[http://www.ruby-lang.org], in Ruby.	2	# REXML is an XML toolkit for Ruby[http://www.ruby-lang.org], in Ruby.
2	#	3	#
3	# REXML is a _pure_ Ruby, XML 1.0 conforming,	4	# REXML is a _pure_ Ruby, XML 1.0 conforming,
Lines 10-17 Link Here
10	#	11	#
11	# Main page:: http://www.germane-software.com/software/rexml	12	# Main page:: http://www.germane-software.com/software/rexml
12	# Author:: Sean Russell <serATgermaneHYPHENsoftwareDOTcom>	13	# Author:: Sean Russell <serATgermaneHYPHENsoftwareDOTcom>
13	# Version:: 3.1.7.1	14	# Version:: 3.1.7.2
14	# Date:: 2007/209	15	# Date:: 2007/275
		16	# Revision:: $Revision$
15	#	17	#
16	# This API documentation can be downloaded from the REXML home page, or can	18	# This API documentation can be downloaded from the REXML home page, or can
17	# be accessed online[http://www.germane-software.com/software/rexml_doc]	19	# be accessed online[http://www.germane-software.com/software/rexml_doc]
Lines 20-29 Link Here
20	# or can be accessed	22	# or can be accessed
21	# online[http://www.germane-software.com/software/rexml/docs/tutorial.html]	23	# online[http://www.germane-software.com/software/rexml/docs/tutorial.html]
22	module REXML	24	module REXML
23	COPYRIGHT = "Copyright © 2001-2007 Sean Russell <ser@germane-software.com>"	25	COPYRIGHT = "Copyright \xC2\xA9 2001-2006 Sean Russell <ser@germane-software.com>"
24	DATE = "2007/209"	26	VERSION = "3.1.7.2"
25	VERSION = "3.1.7.1"	27	DATE = "2007/275"
26	REVISION = "$Revision: 1270$".gsub(/\$Revision:\|\$/,'').strip	28	REVISION = "$Revision$".gsub(/\$Revision:\|\$/,'').strip
27		29
28	Copyright = COPYRIGHT	30	Copyright = COPYRIGHT
29	Version = VERSION	31	Version = VERSION
30	+ 3a3bbbf4-582e-0410-a78b-8bf2211dae1f:/local/ruby-trunk/dtd/dtd.rb:1298	32	+ 3a3bbbf4-582e-0410-a78b-8bf2211dae1f:/local/ruby-trunk/dtd/dtd.rb:1298
31	- 3a3bbbf4-582e-0410-a78b-8bf2211dae1f:/local/rexml-trunk/src/rexml:1287	33	- 3a3bbbf4-582e-0410-a78b-8bf2211dae1f:/local/rexml-trunk/src/rexml:1287
Line 4 Link Here
Line 4 Link Here
4	+ 3a3bbbf4-582e-0410-a78b-8bf2211dae1f:/local/rexml-trunk/src/rexml:1287	3	+ 3a3bbbf4-582e-0410-a78b-8bf2211dae1f:/local/rexml-trunk/src/rexml:1287
Line 4 Link Here
Line 4 Link Here
Line 4 Link Here

Return to bug 195505

Lines 553-558 Link Here

(-)lib/rexml/element.rb (-5 / +6 lines)
553	def attribute( name, namespace=nil )	553	def attribute( name, namespace=nil )
554	prefix = nil	554	prefix = nil
555	prefix = namespaces.index(namespace) if namespace	555	prefix = namespaces.index(namespace) if namespace
		556	prefix = nil if prefix == 'xmlns'
556	attributes.get_attribute( "#{prefix ? prefix + ':' : ''}#{name}" )	557	attributes.get_attribute( "#{prefix ? prefix + ':' : ''}#{name}" )
557	end	558	end
558		559
Lines 854-868 Link Here
854	# Source (see Element.initialize). If not supplied or nil, a	855	# Source (see Element.initialize). If not supplied or nil, a
855	# new, default Element will be constructed	856	# new, default Element will be constructed
856	# Returns:: the added Element	857	# Returns:: the added Element
857	# a = Element.new 'a'	858	# a = Element.new('a')
858	# a.elements.add Element.new 'b' #-> <a><b/></a>	859	# a.elements.add(Element.new('b')) #-> <a><b/></a>
859	# a.elements.add 'c' #-> <a><b/><c/></a>	860	# a.elements.add('c') #-> <a><b/><c/></a>
860	def add element=nil	861	def add element=nil
861	rv = nil	862	rv = nil
862	if element.nil?	863	if element.nil?
863	Element.new "", self, @element.context	864	Element.new("", self, @element.context)
864	elsif not element.kind_of?(Element)	865	elsif not element.kind_of?(Element)
865	Element.new element, self, @element.context	866	Element.new(element, self, @element.context)
866	else	867	else
867	@element << element	868	@element << element
868	element.context = @element.context	869	element.context = @element.context

Lines 1-5 Link Here

(-)lib/rexml/parsers/baseparser.rb (-9 / +47 lines)
1	require 'rexml/parseexception'	1	require 'rexml/parseexception'
2	require 'rexml/source'	2	require 'rexml/source'
		3	require 'set'
3		4
4	module REXML	5	module REXML
5	module Parsers	6	module Parsers
Lines 24-30 Link Here
24	# Nat Price gave me some good ideas for the API.	25	# Nat Price gave me some good ideas for the API.
25	class BaseParser	26	class BaseParser
26	NCNAME_STR= '[\w:][\-\w\d.]*'	27	NCNAME_STR= '[\w:][\-\w\d.]*'
27	NAME_STR= "(?:#{NCNAME_STR}:)?#{NCNAME_STR}"	28	NAME_STR= "(?:(#{NCNAME_STR}):)?(#{NCNAME_STR})"
		29	UNAME_STR= "(?:#{NCNAME_STR}:)?#{NCNAME_STR}"
28		30
29	NAMECHAR = '[\-\w\d\.:]'	31	NAMECHAR = '[\-\w\d\.:]'
30	NAME = "([\\w:]#{NAMECHAR}*)"	32	NAME = "([\\w:]#{NAMECHAR}*)"
Lines 35-41 Link Here
35		37
36	DOCTYPE_START = /\A\s*<!DOCTYPE\s/um	38	DOCTYPE_START = /\A\s*<!DOCTYPE\s/um
37	DOCTYPE_PATTERN = /\s<!DOCTYPE\s+(.?)(\[\|>)/um	39	DOCTYPE_PATTERN = /\s<!DOCTYPE\s+(.?)(\[\|>)/um
38	ATTRIBUTE_PATTERN = /\s(#{NAME_STR})\s=\s(["'])(.?)\2/um	40	ATTRIBUTE_PATTERN = /\s(#{NAME_STR})\s=\s(["'])(.?)\4/um
39	COMMENT_START = /\A<!--/u	41	COMMENT_START = /\A<!--/u
40	COMMENT_PATTERN = /<!--(.*?)-->/um	42	COMMENT_PATTERN = /<!--(.*?)-->/um
41	CDATA_START = /\A<!\[CDATA\[/u	43	CDATA_START = /\A<!\[CDATA\[/u
Lines 45-51 Link Here
45	XMLDECL_PATTERN = /<\?xml\s+(.*?)\?>/um	47	XMLDECL_PATTERN = /<\?xml\s+(.*?)\?>/um
46	INSTRUCTION_START = /\A<\?/u	48	INSTRUCTION_START = /\A<\?/u
47	INSTRUCTION_PATTERN = /<\?(.?)(\s+.?)?\?>/um	49	INSTRUCTION_PATTERN = /<\?(.?)(\s+.?)?\?>/um
48	TAG_MATCH = /^<((?>#{NAME_STR}))\s((?>\s+#{NAME_STR}\s=\s(["']).?\3))\s(\/)?>/um	50	TAG_MATCH = /^<((?>#{NAME_STR}))\s((?>\s+#{UNAME_STR}\s=\s(["']).?\5))\s(\/)?>/um
49	CLOSE_MATCH = /^\s<\/(#{NAME_STR})\s>/um	51	CLOSE_MATCH = /^\s<\/(#{NAME_STR})\s>/um
50		52
51	VERSION = /\bversion\s=\s["'](.*?)['"]/um	53	VERSION = /\bversion\s=\s["'](.*?)['"]/um
Lines 133-138 Link Here
133	@tags = []	135	@tags = []
134	@stack = []	136	@stack = []
135	@entities = []	137	@entities = []
		138	@nsstack = []
136	end	139	end
137		140
138	def position	141	def position
Lines 213-218 Link Here
213	return [ :processing_instruction, *@source.match(INSTRUCTION_PATTERN, true)[1,2] ]	216	return [ :processing_instruction, *@source.match(INSTRUCTION_PATTERN, true)[1,2] ]
214	when DOCTYPE_START	217	when DOCTYPE_START
215	md = @source.match( DOCTYPE_PATTERN, true )	218	md = @source.match( DOCTYPE_PATTERN, true )
		219	@nsstack.unshift(curr_ns=Set.new)
216	identity = md[1]	220	identity = md[1]
217	close = md[2]	221	close = md[2]
218	identity =~ IDENTITY	222	identity =~ IDENTITY
Lines 288-293 Link Here
288	val = attdef[3]	292	val = attdef[3]
289	val = attdef[4] if val == "#FIXED "	293	val = attdef[4] if val == "#FIXED "
290	pairs[attdef[0]] = val	294	pairs[attdef[0]] = val
		295	if attdef[0] =~ /^xmlns:(.*)/
		296	@nsstack[0] << $1
		297	end
291	end	298	end
292	end	299	end
293	return [ :attlistdecl, element, pairs, contents ]	300	return [ :attlistdecl, element, pairs, contents ]
Lines 312-317 Link Here
312	begin	319	begin
313	if @source.buffer[0] == ?<	320	if @source.buffer[0] == ?<
314	if @source.buffer[1] == ?/	321	if @source.buffer[1] == ?/
		322	@nsstack.shift
315	last_tag = @tags.pop	323	last_tag = @tags.pop
316	#md = @source.match_to_consume( '>', CLOSE_MATCH)	324	#md = @source.match_to_consume( '>', CLOSE_MATCH)
317	md = @source.match( CLOSE_MATCH, true )	325	md = @source.match( CLOSE_MATCH, true )
Lines 345-363 Link Here
345	raise REXML::ParseException.new("missing attribute quote", @source) if @source.match(MISSING_ATTRIBUTE_QUOTES )	353	raise REXML::ParseException.new("missing attribute quote", @source) if @source.match(MISSING_ATTRIBUTE_QUOTES )
346	raise REXML::ParseException.new("malformed XML: missing tag start", @source)	354	raise REXML::ParseException.new("malformed XML: missing tag start", @source)
347	end	355	end
348	attrs = []	356	attributes = {}
349	if md[2].size > 0	357	prefixes = Set.new
350	attrs = md[2].scan( ATTRIBUTE_PATTERN )	358	prefixes << md[2] if md[2]
		359	@nsstack.unshift(curr_ns=Set.new)
		360	if md[4].size > 0
		361	attrs = md[4].scan( ATTRIBUTE_PATTERN )
351	raise REXML::ParseException.new( "error parsing attributes: [#{attrs.join ', '}], excess = \"#$'\"", @source) if $' and $'.strip.size > 0	362	raise REXML::ParseException.new( "error parsing attributes: [#{attrs.join ', '}], excess = \"#$'\"", @source) if $' and $'.strip.size > 0
		363	attrs.each { \|a,b,c,d,e\|
		364	if b == "xmlns"
		365	if c == "xml"
		366	if d != "http://www.w3.org/XML/1998/namespace"
		367	msg = "The 'xml' prefix must not be bound to any other namespace "+
		368	"(http://www.w3.org/TR/REC-xml-names/#ns-decl)"
		369	raise REXML::ParseException.new( msg, @source, self )
		370	end
		371	elsif c == "xmlns"
		372	msg = "The 'xmlns' prefix must not be declared "+
		373	"(http://www.w3.org/TR/REC-xml-names/#ns-decl)"
		374	raise REXML::ParseException.new( msg, @source, self)
		375	end
		376	curr_ns << c
		377	elsif b
		378	prefixes << b unless b == "xml"
		379	end
		380	attributes[a] = e
		381	}
352	end	382	end
353		383
354	if md[4]	384	# Verify that all of the prefixes have been defined
		385	for prefix in prefixes
		386	unless @nsstack.find{\|k\| k.member?(prefix)}
		387	raise UndefinedNamespaceException.new(prefix,@source,self)
		388	end
		389	end
		390
		391	if md[6]
355	@closed = md[1]	392	@closed = md[1]
		393	@nsstack.shift
356	else	394	else
357	@tags.push( md[1] )	395	@tags.push( md[1] )
358	end	396	end
359	attributes = {}
360	attrs.each { \|a,b,c\| attributes[a] = c }
361	return [ :start_element, md[1], attributes ]	397	return [ :start_element, md[1], attributes ]
362	end	398	end
363	else	399	else
Lines 371-376 Link Here
371	# return PullEvent.new( :text, md[1], unnormalized )	407	# return PullEvent.new( :text, md[1], unnormalized )
372	return [ :text, md[1] ]	408	return [ :text, md[1] ]
373	end	409	end
		410	rescue REXML::UndefinedNamespaceException
		411	raise
374	rescue REXML::ParseException	412	rescue REXML::ParseException
375	raise	413	raise
376	rescue Exception, NameError => error	414	rescue Exception, NameError => error

Lines 29-36 Link Here

(-)lib/rexml/parsers/treeparser.rb (-2 / +3 lines)
29	return	29	return
30	when :start_element	30	when :start_element
31	tag_stack.push(event[1])	31	tag_stack.push(event[1])
32	# find the observers for namespaces	32	el = @build_context = @build_context.add_element( event[1], event[2] )
33	@build_context = @build_context.add_element( event[1], event[2] )
34	when :end_element	33	when :end_element
35	tag_stack.pop	34	tag_stack.pop
36	@build_context = @build_context.parent	35	@build_context = @build_context.parent
Lines 86-91 Link Here
86	end	85	end
87	rescue REXML::Validation::ValidationException	86	rescue REXML::Validation::ValidationException
88	raise	87	raise
		88	rescue REXML::UndefinedNamespaceException
		89	raise
89	rescue	90	rescue
90	raise ParseException.new( $!.message, @parser.source, @parser, $! )	91	raise ParseException.new( $!.message, @parser.source, @parser, $! )
91	end	92	end