Tagging 0.5.3

git-svn-id: http://svn.caldersphere.net/svn/main/rubyforge/jrexml/tags/0.5.3@194 b03c2d0b-2f10-0410-a2f9-fc8001506dfa

History.txt

@@ -1,3 +1,7 @@
+= 0.5.3
+
+* Take advantage of the fact that the XPP parser expands entities for us, so that we don't have to use the ridiculously slow REXML::Text::unnormalize method.
+
 = 0.5.2
 
 * Raise REXML::ParseException on parse errors, instead of a custom error.

Rakefile

@@ -3,9 +3,12 @@ require 'spec/rake/spectask'
 MANIFEST = FileList["History.txt", "Manifest.txt", "README.txt", "LICENSE.txt", "Rakefile",
   "lib/**/*.rb", "lib/xpp*", "spec/**/*.rb", "spec/*.xml"]
 
+File.open("Manifest.txt", "w") {|f| MANIFEST.each {|l| f.puts l } }
+
+require './lib/jrexml/version'
 begin
   require 'hoe'
-  hoe = Hoe.new("jrexml", "0.5.2") do |p|
+  hoe = Hoe.new("jrexml", JREXML::Version) do |p|
     p.rubyforge_name = "caldersphere"
     p.url = "http://caldersphere.rubyforge.org/jrexml"
     p.author = "Nick Sieger"

lib/jrexml.rb

@@ -1,13 +1,3 @@
-require 'rexml/parsers/baseparser'
 require 'jrexml/java_pull_parser'
-
-class REXML::Parsers::BaseParser #:nodoc:
-  # Extend every REXML base parser with a version that uses a Java pull parser 
-  # library
-  def self.new(*args)
-    obj = allocate
-    obj.extend(JREXML::JavaPullParser)
-    obj.send :initialize, *args
-    obj
-  end
-end
+require 'jrexml/ext/base_parser'
+require 'jrexml/ext/no_unnormalize'

lib/jrexml/ext/base_parser.rb

@@ -0,0 +1,26 @@
+require 'rexml/parsers/baseparser'
+
+class REXML::Parsers::BaseParser #:nodoc:
+  class << self
+    # Set to true to disable JREXML (default nil/unset means use JREXML)
+    attr_accessor :default_parser
+
+    def new_default_parser(*args)
+      prev = self.default_parser
+      self.default_parser = true
+      new(*args)
+    ensure
+      self.default_parser = prev
+    end
+
+    # Extend every REXML base parser with a version that uses a Java pull parser
+    # library
+    def new(*args)
+      obj = allocate
+      obj.extend(JREXML::JavaPullParser) unless self.default_parser
+      class << obj; public :initialize; end
+      obj.initialize *args
+      obj
+    end
+  end
+end

lib/jrexml/ext/no_unnormalize.rb

@@ -0,0 +1,25 @@
+require 'rexml/text'
+require 'jrexml/ext/base_parser'
+
+module REXML
+  class Text
+    alias_method :original_initialize, :initialize
+
+    # Redefine text initialize to receive the expanded value, since this is done
+    # by JREXML.
+    #
+    # Original arity/args is:
+    # def initialize arg, respect_whitespace=false, parent=nil, raw=nil, entity_filter=nil, illegal=ILLEGAL
+    def initialize(value, *args)
+      # Text.new is always called with raw = true from treeparser.rb
+      if !REXML::Parsers::BaseParser.default_parser && args[2]
+        args[2] = nil
+        original_initialize(value, *args)
+        # Set the 'unnormalized' ivar up front, since it's already expanded
+        @unnormalized = value
+      else
+        original_initialize(value, *args)
+      end
+    end
+  end
+end

lib/jrexml/java_pull_parser.rb

@@ -1,3 +1,5 @@
+require 'rexml/parseexception'
+
 module JREXML
   begin
     XmlPullParser = Java::org.xmlpull.v1.XmlPullParser
@@ -98,7 +100,12 @@ def convert_event(event)
           when TEXT
             text << @source.text
           when ENTITY_REF
-            text << "&#{@source.name};"
+            val = @source.text
+            if val
+              text << val
+            else
+              text << "&#{@source.name};"
+            end
           end
           event = event_stack.shift
           break unless event
@@ -108,7 +115,7 @@ def convert_event(event)
           end
         end
       end
-      convert_event_without_text_or_entityref(event)      
+      convert_event_without_text_or_entityref(event)
     end
 
     def convert_event_without_text_or_entityref(event)
@@ -168,5 +175,9 @@ def get_bytes(src)
     def debug_event(event)
       "XmlPullParser::#{XmlPullParser::TYPES[event]}" if event
     end
+
+    def using_jrexml?
+      true
+    end
   end
 end

lib/jrexml/version.rb

@@ -0,0 +1,3 @@
+module JREXML
+  Version = "0.5.3"
+end

spec/document_spec.rb

@@ -0,0 +1,17 @@
+require File.dirname(__FILE__) + '/spec_helper'
+require 'rexml/document'
+
+describe JREXML do
+  def document
+    REXML::Document.new %q(<document>text &lt; other &gt;&#x20;text</document>)
+  end
+
+  it "should not need REXML's unnormalize method" do
+    REXML::Parsers::BaseParser.default_parser = true
+    document.root.text.should == %q(text < other > text)
+
+    require 'jrexml/ext/no_unnormalize'
+    REXML::Parsers::BaseParser.default_parser = false
+    document.root.text.should == %q(text < other > text)
+  end
+end

spec/java_pull_parser_spec.rb

@@ -3,11 +3,10 @@
 describe JREXML::JavaPullParser do
   def parse(source)
     @parser = REXML::Parsers::BaseParser.new(source)
-    @parser.extend(JREXML::JavaPullParser)
     @parser.stream = source
     (class << @parser; self; end).send :define_method, "base_events" do
       events = []
-      baseparser = REXML::Parsers::BaseParser.new(source)
+      baseparser = REXML::Parsers::BaseParser.new_default_parser(source)
       loop do
         event = baseparser.pull
         events << event
@@ -20,6 +19,8 @@ def parse(source)
 
   def verify_events
     @parser.base_events.each do |evt|
+      # still need to expand entities to compare to REXML's base parser
+      evt[1] = REXML::Text::unnormalize(evt[1]) if evt[0] == :text
       @parser.pull.should == evt
     end
     @parser.should be_empty
@@ -30,6 +31,10 @@ def parse_and_verify(source)
     verify_events
   end
 
+  it "should use JREXML by default once it's loaded" do
+    REXML::Parsers::BaseParser.new("<doc/>").should be_using_jrexml
+  end
+
   it "should parse a document consisting of a single empty element" do
     parse_and_verify %q(<document/>)
   end
@@ -91,6 +96,15 @@ def parse_and_verify(source)
     parse_and_verify %q(<document>text &lt; other &gt;&#x20;text</document>)
   end
 
+  it "should not expand extended (e.g., HTML) entities" do
+    parse "<doc>&eacute;</doc>"
+    events = @parser.all_events
+
+    events[0].should == [:start_element, "doc", {}]
+    events[1].should == [:text, "&eacute;"]
+    events[2].should == [:end_element, "doc"]
+  end
+
   it "should handle a longer, more complex document (50+K atom feed)" do
     File.open(File.dirname(__FILE__) + "/atom_feed.xml") do |f|
       parse_and_verify f.read

spec/spec_helper.rb

@@ -1,6 +1,6 @@
 $LOAD_PATH.unshift File.dirname(__FILE__) + '/../lib'
-require 'rexml/parsers/baseparser'
-require 'jrexml'
+require 'jrexml/java_pull_parser'
+require 'jrexml/ext/base_parser'
 
 Spec::Runner.configure do |config|
   config.before :all do