Merge pull request #1713 from sparklemotion/flavorjones-1238-segfault…

…-reparented-entities

issue #1238 segfault on predefined entities

CHANGELOG.md

@@ -15,6 +15,7 @@
 ## Bug fixes
 
 * Node#serialize once again returns UTF-8-encoded strings. [#1659]
+* [JRuby] made SAX parsing of characters consistent with C implementation [#1676] (Thanks, @andrew-aladev!)
 
 
 # 1.8.1 / 2017-09-19

Manifest.txt

@@ -217,6 +217,7 @@ lib/nokogiri/xml/dtd.rb
 lib/nokogiri/xml/element_content.rb
 lib/nokogiri/xml/element_decl.rb
 lib/nokogiri/xml/entity_decl.rb
+lib/nokogiri/xml/entity_reference.rb
 lib/nokogiri/xml/namespace.rb
 lib/nokogiri/xml/node.rb
 lib/nokogiri/xml/node/save_options.rb
@@ -297,8 +298,8 @@ test/files/xinclude.xml
 test/helper.rb
 test/html/sax/test_parser.rb
 test/html/sax/test_parser_context.rb
-test/html/sax/test_push_parser.rb
 test/html/sax/test_parser_text.rb
+test/html/sax/test_push_parser.rb
 test/html/test_builder.rb
 test/html/test_document.rb
 test/html/test_document_encoding.rb
@@ -325,8 +326,8 @@ test/xml/node/test_save_options.rb
 test/xml/node/test_subclass.rb
 test/xml/sax/test_parser.rb
 test/xml/sax/test_parser_context.rb
-test/xml/sax/test_push_parser.rb
 test/xml/sax/test_parser_text.rb
+test/xml/sax/test_push_parser.rb
 test/xml/test_attr.rb
 test/xml/test_attribute_decl.rb
 test/xml/test_builder.rb

lib/nokogiri/xml.rb

@@ -23,6 +23,7 @@
 require 'nokogiri/xml/reader'
 require 'nokogiri/xml/notation'
 require 'nokogiri/xml/entity_decl'
+require 'nokogiri/xml/entity_reference'
 require 'nokogiri/xml/schema'
 require 'nokogiri/xml/relax_ng'
 

lib/nokogiri/xml/entity_reference.rb

@@ -0,0 +1,18 @@
+module Nokogiri
+  module XML
+    class EntityReference < Nokogiri::XML::Node
+      def children
+        # libxml2 will create a malformed child node for predefined
+        # entities. because any use of that child is likely to cause a
+        # segfault, we shall pretend that it doesn't exist.
+        #
+        # see https://github.com/sparklemotion/nokogiri/issues/1238 for details
+        NodeSet.new(document)
+      end
+
+      def inspect_attributes
+        [:name]
+      end
+    end
+  end
+end

test/xml/test_entity_reference.rb

@@ -28,6 +28,17 @@ def test_newline_node
         doc.xpath('/item').first.add_child(lf_node)
         assert_match(/
/, doc.to_xml)
       end
+
+      def test_children_should_always_be_empty
+        # https://github.com/sparklemotion/nokogiri/issues/1238
+        #
+        # libxml2 will create a malformed child node for predefined
+        # entities. because any use of that child is likely to cause a
+        # segfault, we shall pretend that it doesn't exist.
+        entity = Nokogiri::XML::EntityReference.new(@xml, "amp")
+        assert_equal 0, entity.children.length
+        entity.inspect # should not segfault
+      end
     end
 
     module Common