nokogiri 1.3.2 → 1.3.3

data/CHANGELOG.ja.rdoc

@@ -1,3 +1,24 @@
+=== 1.3.3 / ２００９年７月２６日
+
+* 新しい機能
+
+  * NodeSet#children 全ての子ノードを返すようになった
+
+* バグの修正
+
+  * libxml-ruby のグローバ ルエラー ハンドラー に優先するようになった
+  * ParseOption#strict を修正
+  * 空のstringを Node#inner_html= に与えた時に生じたSEGVを修正 GH #88
+  * Ruby 1.9 で返したStringのエンコードはUTF-8になった
+  * ドキュメントの根ノードから違うドキュメントの根ノードに移動した時に生じた
+    SEGVを修正 GH #91
+  * ノードをインスタンス化する時のO(n)のペナルティーを修正 GH #101
+  * XMLのドキュメントをHTMLのドキュメントととして出力出来るようになった
+
+* 廃棄予定
+
+  * Hpricotの互換性レイヤーがNokogiriの1.4.0で除去される予定
+
 === 1.3.2 / ２００９年６月２２日
 
 * 新しい機能
@@ -9,12 +30,12 @@
   * Nokogiri::XML::NodeSet#search はトップレベルのノードを検索するようになった
     GH #73
   * Nokogiri::XML::Documentからメソッドに関係する名前空間を取り除いた
-  * 同じ名前空間を２回目に入れた時に起るセグフォールトを修正した
-  * Snow Leopard で Nokogiri は動くようになった GH #79
-  * メーリングリストはGoogle Groupsの以下のアドレスに移動した
+  * 2回同じ名前空間が追加されたときSEGVする問題を修正した
+  * Snow Leopard で Nokogiri が動くようになった GH #79
+  * メーリングリストはGoogle Groupsの以下のURLに移動した
     http://groups.google.com/group/nokogiri-talk
   * HTML フラグメントはコメントとCDATAを正確に扱うようになった
-  * Nokogiri::XML::Document#cloneはdupのエイリアスと同様の働きをする
+  * Nokogiri::XML::Document#cloneはdupのaliasになった
 
 * 廃棄予定
 

data/CHANGELOG.rdoc

@@ -1,3 +1,23 @@
+=== 1.3.3 / 2009/07/26
+
+* New Features
+
+  * NodeSet#children returns all children of all nodes
+
+* Bugfixes
+
+  * Override libxml-ruby's global error handler
+  * ParseOption#strict fixed
+  * Fixed a segfault when sending an empty string to Node#inner_html= GH #88
+  * String encoding is now set to UTF-8 in Ruby 1.9
+  * Fixed a segfault when moving root nodes between documents. GH #91
+  * Fixed an O(n) penalty on node creation. GH #101
+  * Allowing XML documents to be output as HTML documents
+
+* Deprecations
+
+  * Hpricot compatibility layer will be removed in 1.4.0
+
 === 1.3.2 / 2009-06-22
 
 * New Features

data/Manifest.txt

@@ -176,6 +176,7 @@ test/css/test_parser.rb
 test/css/test_tokenizer.rb
 test/css/test_xpath_visitor.rb
 test/ffi/test_document.rb
+test/files/2ch.html
 test/files/address_book.rlx
 test/files/address_book.xml
 test/files/bar/bar.xsd
@@ -185,6 +186,7 @@ test/files/exslt.xslt
 test/files/foo/foo.xsd
 test/files/po.xml
 test/files/po.xsd
+test/files/shift_jis.xml
 test/files/snuggles.xml
 test/files/staff.xml
 test/files/staff.xslt

data/Rakefile

@@ -27,7 +27,7 @@ HOE = Hoe.spec 'nokogiri' do
   ]
 
   %w{ racc rexical rake-compiler }.each do |dep|
-    extra_dev_deps << dep
+    self.extra_dev_deps << dep
   end
 
   self.spec_extras = { :extensions => ["ext/nokogiri/extconf.rb"] }
@@ -52,30 +52,23 @@ unless java
       "--with-xslt-dir=#{File.join(cross_dir, 'libxslt')}"
   end
 
-  ###
-  # To build the windows fat binary, do:
-  #
-  #   rake fat_binary native gem
-  #
-  # I keep my ruby in multiruby, so my command is like this:
-  #
-  #   RAKE19=~/.multiruby/install/1.9.1-p129/bin/rake \
-  #     rake fat_binary native gem
-  task 'fat_binary' do
-    rake19 = ENV['RAKE19'] || 'rake1.9'
-    system("rake clean cross compile RUBY_CC_VERSION=1.8.6 FAT_DIR=1.8")
-    system("#{rake19} clean cross compile RUBY_CC_VERSION=1.9.1 FAT_DIR=1.9")
+  file 'lib/nokogiri/nokogiri.rb' do
     File.open("lib/#{HOE.name}/#{HOE.name}.rb", 'wb') do |f|
       f.write <<-eoruby
 require "#{HOE.name}/\#{RUBY_VERSION.sub(/\\.\\d+$/, '')}/#{HOE.name}"
       eoruby
     end
-    HOE.spec.extensions = []
-    HOE.spec.platform = 'x86-mingw32'
-    HOE.spec.files += Dir["lib/#{HOE.name}/#{HOE.name}.rb"]
-    HOE.spec.files += Dir["lib/#{HOE.name}/1.{8,9}/*"]
-    HOE.spec.files += Dir["ext/nokogiri/*.dll"]
   end
+
+  namespace :cross do
+    task :file_list do
+      HOE.spec.platform = 'x86-mswin32'
+      HOE.spec.extensions = []
+      HOE.spec.files += Dir["lib/#{HOE.name}/#{HOE.name}.rb"]
+      HOE.spec.files += Dir["ext/nokogiri/*.dll"]
+    end
+  end
+
   CLOBBER.include("lib/nokogiri/nokogiri.{so,dylib,rb,bundle}")
   CLOBBER.include("lib/nokogiri/1.{8,9}")
 end
@@ -146,6 +139,7 @@ libs.each do |lib|
       system("wget #{url} || curl -O #{url}")
     end
   end
+
   file "tmp/cross/#{lib.split('-').first}" => ["tmp/stash/#{lib}.zip"] do |t|
     puts "unzipping #{lib}.zip"
     FileUtils.mkdir_p('tmp/cross')
@@ -157,8 +151,11 @@ libs.each do |lib|
       sh "touch #{lib.split('-').first}"
     end
   end
+
   if Rake::Task.task_defined?(:cross)
     Rake::Task[:cross].prerequisites << "tmp/cross/#{lib.split('-').first}"
+    Rake::Task[:cross].prerequisites << "lib/nokogiri/nokogiri.rb"
+    Rake::Task[:cross].prerequisites << "cross:file_list"
   end
 end
 
@@ -181,8 +178,10 @@ unless windows || java || ENV['NOKOGIRI_FFI']
   end
 
   Rake::Task[:test].prerequisites << :compile
-  ['valgrind', 'valgrind:mem', 'valgrind:mem0'].each do |task_name|
-    Rake::Task["test:#{task_name}"].prerequisites << :compile
+  if Hoe.plugins.include?(:debugging)
+    ['valgrind', 'valgrind:mem', 'valgrind:mem0'].each do |task_name|
+      Rake::Task["test:#{task_name}"].prerequisites << :compile
+    end
   end
 else
   [:test, :check_manifest].each do |task_name|

data/ext/nokogiri/extconf.rb

@@ -1,5 +1,7 @@
 ENV['RC_ARCHS'] = '' if RUBY_PLATFORM =~ /darwin/
 
+# :stopdoc:
+
 require 'mkmf'
 
 ROOT = File.expand_path(File.join(File.dirname(__FILE__), '..', '..'))
@@ -74,7 +76,11 @@ end
 CUSTOM_DASH_I = []
 
 def nokogiri_find_header header_file, *paths
-  message = checking_message(header_file, paths)
+  # mkmf in ruby 1.8.5 does not have the "checking_message" method
+  message = defined?(checking_message) ?
+    checking_message(header_file, paths) :
+    header_file
+
   header = cpp_include header_file
   checking_for message do
     found = false
@@ -152,3 +158,4 @@ if ENV['CPUPROFILE']
 end
 
 create_makefile('nokogiri/nokogiri')
+# :startdoc:

data/ext/nokogiri/html_document.c

@@ -40,7 +40,6 @@ static VALUE read_io( VALUE klass,
   const char * c_enc    = (encoding == Qnil) ? NULL : StringValuePtr(encoding);
   VALUE error_list      = rb_ary_new();
 
-  xmlInitParser();
   xmlResetLastError();
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 
@@ -92,7 +91,6 @@ static VALUE read_memory( VALUE klass,
   int len               = RSTRING_LEN(string);
   VALUE error_list      = rb_ary_new();
 
-  xmlInitParser();
   xmlResetLastError();
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 

data/ext/nokogiri/nokogiri.c

@@ -53,6 +53,8 @@ void Init_nokogiri()
                 NOKOGIRI_STR_NEW2(xmlParserVersion, "UTF-8")
               );
 
+  xmlInitParser();
+
   init_xml_document();
   init_html_document();
   init_xml_node();

data/ext/nokogiri/nokogiri.h

@@ -46,7 +46,7 @@ int is_2_6_16(void) ;
   ({ \
     VALUE _string = rb_str_new2((const char *)str); \
     if(NULL != encoding) { \
-      int _enc = rb_enc_find_index(encoding); \
+      int _enc = rb_enc_find_index("UTF-8"); \
       if(_enc == -1) \
         rb_enc_associate_index(_string, rb_enc_find_index("ASCII")); \
       else \
@@ -59,7 +59,7 @@ int is_2_6_16(void) ;
   ({ \
     VALUE _string = rb_str_new((const char *)str, (long)len); \
     if(NULL != encoding) { \
-      int _enc = rb_enc_find_index(encoding); \
+      int _enc = rb_enc_find_index("UTF-8"); \
       if(_enc == -1) \
         rb_enc_associate_index(_string, rb_enc_find_index("ASCII")); \
       else \
@@ -114,8 +114,7 @@ extern VALUE mNokogiriXslt ;
 #define NOKOGIRI_ROOT_NODE(_node) \
   ({ \
     nokogiriTuplePtr tuple = (nokogiriTuplePtr)(_node->doc->_private);       \
-    xmlNodeSetPtr node_set = (xmlNodeSetPtr)(tuple->unlinkedNodes);     \
-    xmlXPathNodeSetAdd(node_set, _node); \
+    st_insert(tuple->unlinkedNodes, (st_data_t)_node, (st_data_t)_node);     \
     _node; \
   })
 

data/ext/nokogiri/xml_document.c

@@ -1,32 +1,29 @@
 #include <xml_document.h>
 
+static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
+{
+  switch(node->type) {
+  case XML_ATTRIBUTE_NODE:
+    xmlFreePropList((xmlAttrPtr)node);
+    break;
+  default:
+    if(node->parent == NULL) {
+      xmlAddChild((xmlNodePtr)doc, node);
+    }
+  }
+  return ST_CONTINUE;
+}
+
 static void dealloc(xmlDocPtr doc)
 {
   NOKOGIRI_DEBUG_START(doc);
 
-  nokogiriTuplePtr tuple = doc->_private;
-  xmlNodeSetPtr node_set = tuple->unlinkedNodes;
+  st_table *node_hash = DOC_UNLINKED_NODE_HASH(doc);
 
   xmlDeregisterNodeFunc func = xmlDeregisterNodeDefault(NULL);
 
-  int j ;
-  for(j = 0 ; j < node_set->nodeNr ; j++) {
-    xmlNodePtr node = node_set->nodeTab[j];
-    switch(node->type)
-    {
-      case XML_ATTRIBUTE_NODE:
-        xmlFreePropList((xmlAttrPtr)node);
-        break;
-      default:
-        if(node->parent == NULL) {
-          xmlAddChild((xmlNodePtr)doc, node);
-        }
-    }
-  }
-
-  if (node_set->nodeTab != NULL)
-    xmlFree(node_set->nodeTab);
-  xmlFree(node_set);
+  st_foreach(node_hash, dealloc_node_i, (st_data_t)doc);
+  st_free_table(node_hash);
 
   free(doc->_private);
   doc->_private = NULL;
@@ -68,7 +65,19 @@ static VALUE set_root(VALUE self, VALUE root)
   Data_Get_Struct(self, xmlDoc, doc);
   Data_Get_Struct(root, xmlNode, new_root);
 
+  xmlNodePtr old_root = NULL;
+
+  /* If the new root's document is not the same as the current document,
+   * then we need to dup the node in to this document. */
+  if(new_root->doc != doc) {
+    old_root = xmlDocGetRootElement(doc);
+    if (!(new_root = xmlDocCopyNode(new_root, doc, 1))) {
+      rb_raise(rb_eRuntimeError, "Could not reparent node (xmlDocCopyNode)");
+    }
+  }
+
   xmlDocSetRootElement(doc, new_root);
+  if(old_root) NOKOGIRI_ROOT_NODE(old_root);
   return root;
 }
 
@@ -136,7 +145,6 @@ static VALUE read_io( VALUE klass,
   const char * c_enc    = (encoding == Qnil) ? NULL : StringValuePtr(encoding);
   VALUE error_list      = rb_ary_new();
 
-  xmlInitParser();
   xmlResetLastError();
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 
@@ -187,7 +195,6 @@ static VALUE read_memory( VALUE klass,
   int len               = RSTRING_LEN(string);
   VALUE error_list      = rb_ary_new();
 
-  xmlInitParser();
   xmlResetLastError();
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
   xmlDocPtr doc = xmlReadMemory(c_buffer, len, c_url, c_enc, NUM2INT(options));
@@ -303,7 +310,7 @@ VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
   rb_funcall(rb_doc, rb_intern("initialize"), 0);
 
   tuple->doc = (void *)rb_doc;
-  tuple->unlinkedNodes = xmlXPathNodeSetCreate(NULL);
+  tuple->unlinkedNodes = st_init_numtable_with_size(128);
   tuple->node_cache = cache;
   doc->_private = tuple ;
 

data/ext/nokogiri/xml_document.h

@@ -2,10 +2,11 @@
 #define NOKOGIRI_XML_DOCUMENT
 
 #include <nokogiri.h>
+#include "st.h"
 
 struct _nokogiriTuple {
   xmlDocPtr     doc;
-  xmlNodeSetPtr unlinkedNodes;
+  st_table     *unlinkedNodes;
   VALUE         node_cache;
 };
 typedef struct _nokogiriTuple nokogiriTuple;
@@ -16,7 +17,7 @@ VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc);
 
 #define DOC_RUBY_OBJECT_TEST(x) ((nokogiriTuplePtr)(x->_private))
 #define DOC_RUBY_OBJECT(x) ((VALUE)((nokogiriTuplePtr)(x->_private))->doc)
-#define DOC_UNLINKED_NODE_SET(x) ((xmlNodeSetPtr)((nokogiriTuplePtr)(x->_private))->unlinkedNodes)
+#define DOC_UNLINKED_NODE_HASH(x) (((nokogiriTuplePtr)(x->_private))->unlinkedNodes)
 #define DOC_NODE_CACHE(x) ((VALUE)((nokogiriTuplePtr)(x->_private))->node_cache)
 
 extern VALUE cNokogiriXmlDocument ;

data/ext/nokogiri/xml_dtd.c

@@ -113,6 +113,8 @@ static VALUE validate(VALUE self, VALUE document)
   return error_list;
 }
 
+VALUE cNokogiriXmlDtd;
+
 void init_xml_dtd()
 {
   VALUE nokogiri = rb_define_module("Nokogiri");
@@ -124,6 +126,8 @@ void init_xml_dtd()
    */
   VALUE klass = rb_define_class_under(xml, "DTD", node);
 
+  cNokogiriXmlDtd = klass;
+
   rb_define_method(klass, "notations", notations, 0);
   rb_define_method(klass, "elements", elements, 0);
   rb_define_method(klass, "entities", entities, 0);

data/ext/nokogiri/xml_dtd.h

@@ -3,6 +3,8 @@
 
 #include <nokogiri.h>
 
+extern VALUE cNokogiriXmlDtd;
+
 void init_xml_dtd();
 
 #endif

data/ext/nokogiri/xml_node.c

@@ -12,7 +12,10 @@ static void debug_node_dealloc(xmlNodePtr x)
 
 static void mark(xmlNodePtr node)
 {
-  rb_gc_mark(DOC_RUBY_OBJECT(node->doc));
+  // it's OK if the document isn't fully realized (as in XML::Reader).
+  // see http://github.com/tenderlove/nokogiri/issues/closed/#issue/95
+  if (DOC_RUBY_OBJECT_TEST(node->doc) && DOC_RUBY_OBJECT(node->doc))
+    rb_gc_mark(DOC_RUBY_OBJECT(node->doc));
 }
 
 /* :nodoc: */
@@ -60,6 +63,17 @@ static VALUE reparent_node_with(VALUE node_obj, VALUE other_obj, node_other_func
   Data_Get_Struct(node_obj, xmlNode, node);
   Data_Get_Struct(other_obj, xmlNode, other);
 
+  // If a document fragment is added, we need to reparent all of it's children
+  if(node->type == XML_DOCUMENT_FRAG_NODE)
+  {
+    xmlNodePtr child = node->children;
+    while(NULL != child) {
+      reparent_node_with(Nokogiri_wrap_xml_node((VALUE)NULL, child), other_obj, func);
+      child = child->next;
+    }
+    return node_obj;
+  }
+
   if (node->doc == other->doc) {
     xmlUnlinkNode(node) ;
     if ( node->type == XML_TEXT_NODE
@@ -673,6 +687,8 @@ static VALUE native_write_to(
 
   xmlIndentTreeOutput = 1;
 
+  const char * before_indent = xmlTreeIndentString;
+
   xmlTreeIndentString = StringValuePtr(indent_string);
 
   xmlSaveCtxtPtr savectx = xmlSaveToIO(
@@ -685,6 +701,8 @@ static VALUE native_write_to(
 
   xmlSaveTree(savectx, node);
   xmlSaveClose(savectx);
+
+  xmlTreeIndentString = before_indent;
   return io;
 }
 
@@ -754,7 +772,10 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   node->doc = doc->doc;
   NOKOGIRI_ROOT_NODE(node);
 
-  VALUE rb_node = Nokogiri_wrap_xml_node(klass, node);
+  VALUE rb_node = Nokogiri_wrap_xml_node(
+      klass == cNokogiriXmlNode ? (VALUE)NULL : klass,
+      node
+  );
   rb_funcall2(rb_node, rb_intern("initialize"), argc, argv);
 
   if(rb_block_given_p()) rb_yield(rb_node);
@@ -774,9 +795,6 @@ static VALUE dump_html(VALUE self)
   xmlNodePtr node ;
   Data_Get_Struct(self, xmlNode, node);
 
-  if(node->doc->type == XML_DOCUMENT_NODE)
-    return rb_funcall(self, rb_intern("to_xml"), 0);
-
   buf = xmlBufferCreate() ;
   htmlNodeDump(buf, node->doc, node);
   VALUE html = NOKOGIRI_STR_NEW2(buf->content, node->doc->encoding);
@@ -845,7 +863,7 @@ VALUE Nokogiri_wrap_xml_node(VALUE klass, xmlNodePtr node)
       klass = cNokogiriXmlCData;
       break;
     case XML_DTD_NODE:
-      klass = rb_const_get(mNokogiriXml, rb_intern("DTD"));
+      klass = cNokogiriXmlDtd;
       break;
     default:
       klass = cNokogiriXmlNode;
@@ -856,13 +874,14 @@ VALUE Nokogiri_wrap_xml_node(VALUE klass, xmlNodePtr node)
   node->_private = (void *)rb_node;
 
   if (DOC_RUBY_OBJECT_TEST(node->doc) && DOC_RUBY_OBJECT(node->doc)) {
+    // it's OK if the document isn't fully realized (as in XML::Reader).
+    // see http://github.com/tenderlove/nokogiri/issues/closed/#issue/95
     document = DOC_RUBY_OBJECT(node->doc);
     node_cache = DOC_NODE_CACHE(node->doc);
+    rb_ary_push(node_cache, rb_node);
+    rb_funcall(document, rb_intern("decorate"), 1, rb_node);
   }
 
-  rb_ary_push(node_cache, rb_node);
-  rb_funcall(document, rb_intern("decorate"), 1, rb_node);
-
   return rb_node ;
 }