ruby/lib/rexml/parsers/treeparser.rb
ser 9acf1749fb r1366@bean: ser | 2007-10-01 21:24:33 -0400
r1352@bean:  ser | 2007-07-29 11:33:07 -0400
  Implements namespace validation in the baseparser.  This means that, as per
  the XML namespace spec, unbound prefixes generate UndefinedNamespaceException.
  Also, as per the namespace spec, the 'xml' prefix must be bound to 
  http://www.w3.org/XML/1998/namespace, and the 'xmlns' prefix must not be declared. 
  in the XML.
 


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@13595 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2007-10-02 01:46:32 +00:00

96 lines
3.3 KiB
Ruby

require 'rexml/validation/validationexception'
module REXML
module Parsers
class TreeParser
def initialize( source, build_context = Document.new )
@build_context = build_context
@parser = Parsers::BaseParser.new( source )
end
def add_listener( listener )
@parser.add_listener( listener )
end
def parse
tag_stack = []
in_doctype = false
entities = nil
begin
while true
event = @parser.pull
#STDERR.puts "TREEPARSER GOT #{event.inspect}"
case event[0]
when :end_document
unless tag_stack.empty?
#raise ParseException.new("No close tag for #{tag_stack.inspect}")
raise ParseException.new("No close tag for #{@build_context.xpath}")
end
return
when :start_element
tag_stack.push(event[1])
el = @build_context = @build_context.add_element( event[1], event[2] )
when :end_element
tag_stack.pop
@build_context = @build_context.parent
when :text
if not in_doctype
if @build_context[-1].instance_of? Text
@build_context[-1] << event[1]
else
@build_context.add(
Text.new(event[1], @build_context.whitespace, nil, true)
) unless (
@build_context.ignore_whitespace_nodes and
event[1].strip.size==0
)
end
end
when :comment
c = Comment.new( event[1] )
@build_context.add( c )
when :cdata
c = CData.new( event[1] )
@build_context.add( c )
when :processing_instruction
@build_context.add( Instruction.new( event[1], event[2] ) )
when :end_doctype
in_doctype = false
entities.each { |k,v| entities[k] = @build_context.entities[k].value }
@build_context = @build_context.parent
when :start_doctype
doctype = DocType.new( event[1..-1], @build_context )
@build_context = doctype
entities = {}
in_doctype = true
when :attlistdecl
n = AttlistDecl.new( event[1..-1] )
@build_context.add( n )
when :externalentity
n = ExternalEntity.new( event[1] )
@build_context.add( n )
when :elementdecl
n = ElementDecl.new( event[1] )
@build_context.add(n)
when :entitydecl
entities[ event[1] ] = event[2] unless event[2] =~ /PUBLIC|SYSTEM/
@build_context.add(Entity.new(event))
when :notationdecl
n = NotationDecl.new( *event[1..-1] )
@build_context.add( n )
when :xmldecl
x = XMLDecl.new( event[1], event[2], event[3] )
@build_context.add( x )
end
end
rescue REXML::Validation::ValidationException
raise
rescue REXML::UndefinedNamespaceException
raise
rescue
raise ParseException.new( $!.message, @parser.source, @parser, $! )
end
end
end
end
end