nokogiri 1.0.3 → 1.0.4

data/History.txt

@@ -1,6 +1,15 @@
+=== 1.0.4
+
+* Bugfixes
+
+  * Changed memory mangement from weak refs to document refs
+  * Plugged some memory leaks
+  * extconf.rb now complains if your libxml2 is not between 2.6.13 and 2.6.99
+  * Builder blocks can call methods from surrounding contexts
+
 === 1.0.3
 
-* 5 Bigfixes
+* 5 Bugfixes
 
   * NodeSet now implements to_ary
   * XML::Document should not implement parent

data/Manifest.txt

@@ -119,3 +119,4 @@ test/xml/test_dtd.rb
 test/xml/test_node.rb
 test/xml/test_node_set.rb
 test/xml/test_text.rb
+vendor/hoe.rb

data/Rakefile

@@ -1,8 +1,8 @@
 # -*- ruby -*-
 
 require 'rubygems'
-require 'mkmf'
-require 'hoe'
+require 'rake'
+
 
 kind = Config::CONFIG['DLEXT']
 windows = RUBY_PLATFORM =~ /mswin/i ? true : false
@@ -10,6 +10,8 @@ windows = RUBY_PLATFORM =~ /mswin/i ? true : false
 LIB_DIR = File.expand_path(File.join(File.dirname(__FILE__), 'lib'))
 $LOAD_PATH << LIB_DIR
 
+require 'vendor/hoe'
+
 GENERATED_PARSER = "lib/nokogiri/css/generated_parser.rb"
 GENERATED_TOKENIZER = "lib/nokogiri/css/generated_tokenizer.rb"
 
@@ -41,7 +43,7 @@ namespace :gem do
   end
 
   namespace :win32 do
-    task :spec do
+    task :spec => ['build:win32'] do
       File.open("#{HOE.name}.gemspec", 'w') do |f|
         HOE.spec.files += Dir['ext/nokogiri/**.{dll,so}']
         if windows
@@ -55,6 +57,16 @@ namespace :gem do
     end
   end
 
+  namespace :jruby do
+    task :spec => ['build'] do
+      File.open("#{HOE.name}.gemspec", 'w') do |f|
+        HOE.spec.platform = 'jruby'
+        HOE.spec.extensions = []
+        f.write(HOE.spec.to_ruby)
+      end
+    end
+  end
+
   namespace :unix do
     task :spec do
       File.open("#{HOE.name}.gemspec", 'w') do |f|
@@ -73,17 +85,19 @@ task :coverage do
 end
 
 file GENERATED_PARSER => "lib/nokogiri/css/parser.y" do |t|
-  unless find_executable("racc")
+  begin
+    sh "racc -o #{t.name} #{t.prerequisites.first}"
+  rescue
     abort "need racc, get the tarball from http://i.loveruby.net/archive/racc/racc-1.4.5-all.tar.gz" 
   end
-  sh "racc -o #{t.name} #{t.prerequisites.first}"
 end
 
 file GENERATED_TOKENIZER => "lib/nokogiri/css/tokenizer.rex" do |t|
-  unless find_executable("frex")
+  begin
+    sh "frex -i --independent -o #{t.name} #{t.prerequisites.first}"
+  rescue
     abort "need frex, sudo gem install aaronp-frex -s http://gems.github.com"   
   end
-  sh "frex -i --independent -o #{t.name} #{t.prerequisites.first}"
 end
 
 task 'ext/nokogiri/Makefile' do
@@ -98,7 +112,11 @@ task EXT => 'ext/nokogiri/Makefile' do
   end
 end
 
-task :build => [EXT, GENERATED_PARSER, GENERATED_TOKENIZER]
+if RUBY_PLATFORM == 'java'
+  task :build => [GENERATED_PARSER, GENERATED_TOKENIZER]
+else
+  task :build => [EXT, GENERATED_PARSER, GENERATED_TOKENIZER]
+end
 
 namespace :build do
   namespace :win32 do
@@ -268,4 +286,12 @@ unless windows
   Rake::Task[:check_manifest].prerequisites << GENERATED_TOKENIZER
 end
 
+# Evil evil hack.  Do not run tests when gem installs
+if ENV['RUBYARCHDIR']
+  class << Rake::Task[:default]
+    attr_writer :prerequisites
+  end
+  Rake::Task[:default].prerequisites = [:build]
+end
+
 # vim: syntax=Ruby

data/ext/nokogiri/extconf.rb

@@ -50,6 +50,8 @@ else
   unless find_header('libxslt/xslt.h', INCLUDEDIR, '/usr/include')
     abort "need libxslt"
   end
+
+  version = try_constant('LIBXML_VERSION', 'libxml/xmlversion.h')
 end
 
 create_makefile('nokogiri/native')

data/ext/nokogiri/html_document.c

@@ -39,8 +39,11 @@ static VALUE read_memory( VALUE klass,
 
   htmlDocPtr doc = htmlReadMemory(c_buffer, len, c_url, c_enc, NUM2INT(options));
 
-  if(doc == NULL)
-    doc = htmlNewDoc((const xmlChar *)c_url, NULL);
+  if(doc == NULL) {
+    xmlFreeDoc(doc);
+    rb_raise(rb_eRuntimeError, "Couldn't create a document");
+    return Qnil;
+  }
 
   return Nokogiri_wrap_xml_document(klass, doc);
 }

data/ext/nokogiri/native.h

@@ -2,6 +2,7 @@
 #define NOKOGIRI_NATIVE
 
 #include <stdlib.h>
+#include <assert.h>
 #include <ruby.h>
 #include <libxml/parser.h>
 #include <libxml/xpath.h>

data/ext/nokogiri/xml_document.c

@@ -1,24 +1,9 @@
 #include <xml_document.h>
 
-/*
- *  note that xmlDocPtr is being cast as an xmlNodePtr, which is legal for the
- *  "common part" struct header which contains only node pointers.
- */
-static void gc_mark(xmlNodePtr node)
-{
-  VALUE rb_obj ;
-  xmlNodePtr child ;
-  /* mark children nodes */
-  for (child = node->children ; child ; child = child->next) {
-    if ((rb_obj = Nokogiri_xml_node2obj_get(child)) != Qnil)
-      rb_gc_mark(rb_obj);
-  }
-}
-
 static void dealloc(xmlDocPtr doc)
 {
   NOKOGIRI_DEBUG_START(doc);
-  Nokogiri_xml_node2obj_remove((xmlNodePtr)doc);
+  doc->_private = NULL;
   xmlFreeDoc(doc);
   NOKOGIRI_DEBUG_END(doc);
 }
@@ -57,7 +42,6 @@ static VALUE set_root(VALUE self, VALUE root)
   Data_Get_Struct(root, xmlNode, new_root);
 
   xmlDocSetRootElement(doc, new_root);
-  Nokogiri_xml_node_owned_set(new_root);
   return root;
 }
 
@@ -95,6 +79,7 @@ static VALUE read_memory( VALUE klass,
   if(doc == NULL) {
     xmlFreeDoc(doc);
     rb_raise(rb_eRuntimeError, "Couldn't create a document");
+    return Qnil;
   }
 
   return Nokogiri_wrap_xml_document(klass, doc);
@@ -156,11 +141,8 @@ VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
 {
   VALUE rb_doc = Qnil;
 
-  if ((rb_doc = Nokogiri_xml_node2obj_get((xmlNodePtr)doc)) != Qnil)
-    return rb_doc ;
-
-  rb_doc = Data_Wrap_Struct(klass ? klass : cNokogiriXmlDocument, gc_mark, dealloc, doc) ;
+  rb_doc = Data_Wrap_Struct(klass ? klass : cNokogiriXmlDocument, 0, dealloc, doc) ;
+  doc->_private = (void *)rb_doc;
 
-  Nokogiri_xml_node2obj_set((xmlNodePtr)doc, rb_doc);
   return rb_doc ;
 }

data/ext/nokogiri/xml_node.c

@@ -1,5 +1,19 @@
 #include <xml_node.h>
 
+/*
+ * call-seq:
+ *  pointer_id
+ *
+ * Get the internal pointer number
+ */
+static VALUE pointer_id(VALUE self)
+{
+  xmlNodePtr node;
+  Data_Get_Struct(self, xmlNode, node);
+
+  return INT2NUM((int)(node));
+}
+
 /*
  * call-seq:
  *  encode_special_chars(string)
@@ -55,6 +69,10 @@ static VALUE duplicate_node(VALUE self)
 
   dup = xmlCopyNode(node, 1);
   if(dup == NULL) return Qnil;
+  dup->doc = node->doc;
+  assert(node->parent);
+
+  xmlAddChild(node->parent, dup);
 
   return Nokogiri_wrap_xml_node(dup);
 }
@@ -70,7 +88,6 @@ static VALUE unlink_node(VALUE self)
   xmlNodePtr node;
   Data_Get_Struct(self, xmlNode, node);
   xmlUnlinkNode(node);
-  Nokogiri_xml_node_owned_set(node);
   return self;
 }
 
@@ -136,8 +153,6 @@ static VALUE replace(VALUE self, VALUE _new_node)
   Data_Get_Struct(_new_node, xmlNode, new_node);
 
   xmlReplaceNode(node, new_node);
-  Nokogiri_xml_node_owned_set(node);
-  Nokogiri_xml_node_owned_set(new_node);
   return self ;
 }
 
@@ -324,7 +339,6 @@ static VALUE set_parent(VALUE self, VALUE parent_node)
   Data_Get_Struct(parent_node, xmlNode, parent);
 
   xmlAddChild(parent, node);
-  Nokogiri_xml_node_owned_set(node);
   return parent_node;
 }
 
@@ -391,21 +405,6 @@ static VALUE path(VALUE self)
   return rval ;
 }
 
-/*
- * call-seq:
- *  document
- *
- * Returns the Nokogiri::XML::Document associated with this Node
- */
-static VALUE document(VALUE self)
-{
-  xmlNodePtr node;
-  Data_Get_Struct(self, xmlNode, node);
-
-  if(!node->doc) return Qnil;
-  return Nokogiri_xml_node2obj_get(node->doc);
-}
-
 /*
  *  call-seq:
  *    add_next_sibling(node)
@@ -420,7 +419,6 @@ static VALUE add_next_sibling(VALUE self, VALUE rb_node)
   xmlAddNextSibling(node, new_sibling);
 
   rb_funcall(rb_node, rb_intern("decorate!"), 0);
-  Nokogiri_xml_node_owned_set(new_sibling);
 
   return rb_node;
 }
@@ -439,7 +437,6 @@ static VALUE add_previous_sibling(VALUE self, VALUE rb_node)
   xmlAddPrevSibling(node, new_sibling);
 
   rb_funcall(rb_node, rb_intern("decorate!"), 0);
-  Nokogiri_xml_node_owned_set(new_sibling);
 
   return rb_node;
 }
@@ -472,14 +469,18 @@ static VALUE to_xml(VALUE self)
  *
  * Create a new node with +name+
  */
-static VALUE new(VALUE klass, VALUE name)
+static VALUE new(VALUE klass, VALUE name, VALUE document)
 {
+  xmlDocPtr doc;
+
+  Data_Get_Struct(document, xmlDoc, doc);
+
   xmlNodePtr node = xmlNewNode(NULL, (xmlChar *)StringValuePtr(name));
-  VALUE rb_node = Nokogiri_wrap_xml_node(node) ;
+  node->doc = doc;
 
-  if(rb_block_given_p()) rb_yield(rb_node);
+  VALUE rb_node = Nokogiri_wrap_xml_node(node);
 
-  Nokogiri_xml_node_owned_set(node);
+  if(rb_block_given_p()) rb_yield(rb_node);
 
   return rb_node;
 }
@@ -505,79 +506,49 @@ static VALUE new_from_str(VALUE klass, VALUE xml)
                         xml, Qnil, Qnil, INT2NUM(0));
     Data_Get_Struct(rb_doc, xmlDoc, doc);
     node = xmlCopyNode(xmlDocGetRootElement(doc), 1); /* 1 => recursive */
-    return Nokogiri_wrap_xml_node(node);
-}
-
-static void deallocate(xmlNodePtr node)
-{
-  Nokogiri_xml_node2obj_remove(node);
-  if (! Nokogiri_xml_node_owned_get(node)) {
-    NOKOGIRI_DEBUG_START_NODE(node);
-    xmlFreeNode(node);
-    NOKOGIRI_DEBUG_END(node);
-  }
-}
+    node->doc = doc;
 
-static void gc_mark_node(xmlNodePtr node)
-{
-  xmlNodePtr child ;
-  VALUE rb_obj ;
-  /* mark document */
-  if (node && node->doc && (rb_obj = Nokogiri_xml_node2obj_get((xmlNodePtr)node->doc)) != Qnil)
-    rb_gc_mark(rb_obj);
-  /* mark parent node */
-  if (node && node->parent && (rb_obj = Nokogiri_xml_node2obj_get(node->parent)) != Qnil)
-    rb_gc_mark(rb_obj);
-  /* mark children nodes */
-  for (child = node->children ; child ; child = child->next) {
-    if ((rb_obj = Nokogiri_xml_node2obj_get(child)) != Qnil)
-      rb_gc_mark(rb_obj);
-  }
-  /* mark sibling nodes */
-  if (node->next && (rb_obj = Nokogiri_xml_node2obj_get(node->next)) != Qnil)
-    rb_gc_mark(rb_obj);
-  if (node->prev && (rb_obj = Nokogiri_xml_node2obj_get(node->prev)) != Qnil)
-    rb_gc_mark(rb_obj);
+    return Nokogiri_wrap_xml_node(node);
 }
 
 VALUE Nokogiri_wrap_xml_node(xmlNodePtr node)
 {
   VALUE rb_node = Qnil;
 
-  if ((rb_node = Nokogiri_xml_node2obj_get(node)) != Qnil)
-    return rb_node ;
-  
   switch(node->type)
   {
     VALUE klass;
 
     case XML_TEXT_NODE:
       klass = rb_const_get(mNokogiriXml, rb_intern("Text"));
-      rb_node = Data_Wrap_Struct(klass, gc_mark_node, deallocate, node) ;
+      rb_node = Data_Wrap_Struct(klass, 0, 0, node) ;
       break;
     case XML_ELEMENT_NODE:
       klass = rb_const_get(mNokogiriXml, rb_intern("Element"));
-      rb_node = Data_Wrap_Struct(klass, gc_mark_node, deallocate, node) ;
+      rb_node = Data_Wrap_Struct(klass, 0, 0, node) ;
       break;
     case XML_ENTITY_DECL:
       klass = rb_const_get(mNokogiriXml, rb_intern("EntityDeclaration"));
-      rb_node = Data_Wrap_Struct(klass, gc_mark_node, deallocate, node) ;
+      rb_node = Data_Wrap_Struct(klass, 0, 0, node) ;
       break;
     case XML_CDATA_SECTION_NODE:
       klass = rb_const_get(mNokogiriXml, rb_intern("CDATA"));
-      rb_node = Data_Wrap_Struct(klass, gc_mark_node, deallocate, node) ;
+      rb_node = Data_Wrap_Struct(klass, 0, 0, node) ;
       break;
     case XML_DTD_NODE:
       klass = rb_const_get(mNokogiriXml, rb_intern("DTD"));
-      rb_node = Data_Wrap_Struct(klass, gc_mark_node, deallocate, node) ;
+      rb_node = Data_Wrap_Struct(klass, 0, 0, node) ;
       break;
     default:
-      rb_node = Data_Wrap_Struct(cNokogiriXmlNode, gc_mark_node, deallocate, node) ;
+      rb_node = Data_Wrap_Struct(cNokogiriXmlNode, 0, 0, node) ;
   }
 
-  Nokogiri_xml_node2obj_set(node, rb_node);
+  assert(node);
+  assert(node->doc);
+  assert(node->doc->_private);
+
+  rb_iv_set(rb_node, "@document",(VALUE)node->doc->_private);
   rb_funcall(rb_node, rb_intern("decorate!"), 0);
-  Nokogiri_xml_node_owned_set(node);
   return rb_node ;
 }
 
@@ -634,64 +605,6 @@ void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash)
 }
 
 
-void Nokogiri_xml_node2obj_set(xmlNodePtr node, VALUE rb_obj)
-{
-  VALUE hash = rb_cvar_get(cNokogiriXmlNode, rb_intern("@@node2obj"));
-  VALUE weakref = rb_obj_id(rb_obj) ; /* so GC won't mark the object. sneaky
-                                         sneaky. */
-  rb_hash_aset(hash, INT2NUM((long)node), weakref);
-}
-
-VALUE Nokogiri_xml_node2obj_get(xmlNodePtr node)
-{
-  static VALUE ObjectSpace = 0 ;
-  static ID id2ref = 0 ;
-
-  if (! ObjectSpace) ObjectSpace = rb_const_get(rb_cObject, rb_intern("ObjectSpace"));
-  if (! id2ref) id2ref = rb_intern("_id2ref");
-
-  VALUE hash = rb_cvar_get(cNokogiriXmlNode, rb_intern("@@node2obj"));
-  VALUE weakref = rb_hash_aref(hash, INT2NUM((long)node)) ;
-  if (weakref == Qnil)
-    return Qnil ;
-  return rb_funcall(ObjectSpace, id2ref, 1, weakref); /* translate weakref to
-                                                       * the object */
-}
-
-void Nokogiri_xml_node2obj_remove(xmlNodePtr node)
-{
-  VALUE hash = rb_cvar_get(cNokogiriXmlNode, rb_intern("@@node2obj"));
-  rb_hash_delete(hash, INT2NUM((long)node));
-}
-
-void Nokogiri_xml_node_owned_set(xmlNodePtr node)
-{
-  VALUE hash = rb_cvar_get(cNokogiriXmlNode, rb_intern("@@owned"));
-  rb_hash_aset(hash, INT2NUM((long)node), node->parent ? Qtrue : Qfalse) ;
-}
-
-int Nokogiri_xml_node_owned_get(xmlNodePtr node)
-{
-  VALUE hash = rb_cvar_get(cNokogiriXmlNode, rb_intern("@@owned"));
-  VALUE q = rb_hash_aref(hash, INT2NUM((long)node)) ;
-  return q == Qtrue ? Qtrue : Qfalse ;
-}
-
-/*
- * call-seq:
- *   owned?
- *
- * Is this node owned by a document?
- */
-static VALUE owned_eh(VALUE self)
-{
-  xmlNodePtr node ;
-  VALUE hash = rb_cvar_get(cNokogiriXmlNode, rb_intern("@@owned"));
-  Data_Get_Struct(self, xmlNode, node);
-  return rb_hash_aref(hash, INT2NUM((long)node)) == Qtrue ? Qtrue : Qfalse ;
-}
-
-
 VALUE cNokogiriXmlNode ;
 void init_xml_node()
 {
@@ -706,10 +619,9 @@ void init_xml_node()
 
   VALUE klass = cNokogiriXmlNode = rb_const_get(mNokogiriXml, rb_intern("Node"));
 
-  rb_define_singleton_method(klass, "new", new, 1);
+  rb_define_singleton_method(klass, "new", new, 2);
   rb_define_singleton_method(klass, "new_from_str", new_from_str, 1);
 
-  rb_define_method(klass, "document", document, 0);
   rb_define_method(klass, "name", get_name, 0);
   rb_define_method(klass, "name=", set_name, 1);
   rb_define_method(klass, "parent=", set_parent, 1);
@@ -734,8 +646,8 @@ void init_xml_node()
   rb_define_method(klass, "dup", duplicate_node, 0);
   rb_define_method(klass, "unlink", unlink_node, 0);
   rb_define_method(klass, "internal_subset", internal_subset, 0);
+  rb_define_method(klass, "pointer_id", pointer_id, 0);
 
   rb_define_private_method(klass, "native_content=", set_content, 1);
   rb_define_private_method(klass, "get", get, 1);
-  rb_define_private_method(klass, "owned?", owned_eh, 0);
 }