lib/rexml/parsers/treeparser.rb

   1 require 'rexml/validation/validationexception'
   2 require 'rexml/undefinednamespaceexception'
   3
   4 module REXML
   5   module Parsers
   6     class TreeParser
   7       def initialize( source, build_context = Document.new )
   8         @build_context = build_context
   9         @parser = Parsers::BaseParser.new( source )
  10       end
  11
  12       def add_listener( listener )
  13         @parser.add_listener( listener )
  14       end
  15
  16       def parse
  17         tag_stack = []
  18         in_doctype = false
  19         entities = nil
  20         begin
  21           while true
  22             event = @parser.pull
  23             #STDERR.puts "TREEPARSER GOT #{event.inspect}"
  24             case event[0]
  25             when :end_document
  26               unless tag_stack.empty?
  27                 #raise ParseException.new("No close tag for #{tag_stack.inspect}")
  28                 raise ParseException.new("No close tag for #{@build_context.xpath}")
  29               end
  30               return
  31             when :start_element
  32               tag_stack.push(event[1])
  33               el = @build_context = @build_context.add_element( event[1], event[2] )
  34             when :end_element
  35               tag_stack.pop
  36               @build_context = @build_context.parent
  37             when :text
  38               if not in_doctype
  39                 if @build_context[-1].instance_of? Text
  40                   @build_context[-1] << event[1]
  41                 else
  42                   @build_context.add(
  43                     Text.new(event[1], @build_context.whitespace, nil, true)
  44                   ) unless (
  45                     @build_context.ignore_whitespace_nodes and
  46                     event[1].strip.size==0
  47                   )
  48                 end
  49               end
  50             when :comment
  51               c = Comment.new( event[1] )
  52               @build_context.add( c )
  53             when :cdata
  54               c = CData.new( event[1] )
  55               @build_context.add( c )
  56             when :processing_instruction
  57               @build_context.add( Instruction.new( event[1], event[2] ) )
  58             when :end_doctype
  59               in_doctype = false
  60               entities.each { |k,v| entities[k] = @build_context.entities[k].value }
  61               @build_context = @build_context.parent
  62             when :start_doctype
  63               doctype = DocType.new( event[1..-1], @build_context )
  64               @build_context = doctype
  65               entities = {}
  66               in_doctype = true
  67             when :attlistdecl
  68               n = AttlistDecl.new( event[1..-1] )
  69               @build_context.add( n )
  70             when :externalentity
  71               n = ExternalEntity.new( event[1] )
  72               @build_context.add( n )
  73             when :elementdecl
  74               n = ElementDecl.new( event[1] )
  75               @build_context.add(n)
  76             when :entitydecl
  77               entities[ event[1] ] = event[2] unless event[2] =~ /PUBLIC|SYSTEM/
  78               @build_context.add(Entity.new(event))
  79             when :notationdecl
  80               n = NotationDecl.new( *event[1..-1] )
  81               @build_context.add( n )
  82             when :xmldecl
  83               x = XMLDecl.new( event[1], event[2], event[3] )
  84               @build_context.add( x )
  85             end
  86           end
  87         rescue REXML::Validation::ValidationException
  88           raise
  89         rescue REXML::UndefinedNamespaceException
  90           raise
  91         rescue
  92           raise ParseException.new( $!.message, @parser.source, @parser, $! )
  93         end
  94       end
  95     end
  96   end
  97 end