nokogiri 1.0.2 → 1.0.3

data/History.txt

@@ -1,3 +1,13 @@
+=== 1.0.3
+
+* 5 Bigfixes
+
+  * NodeSet now implements to_ary
+  * XML::Document should not implement parent
+  * More GC Bugs fixed.  (Mike is AWESOME!)
+  * Removed RARRAY_LEN for 1.8.5 compatibility.  Thanks Shane Hanna.
+  * inner_html fixed. (Thanks Yehuda!)
+
 === 1.0.2
 
 * 1 Bugfix

data/Manifest.txt

@@ -81,6 +81,7 @@ test/css/test_nthiness.rb
 test/css/test_parser.rb
 test/css/test_tokenizer.rb
 test/css/test_xpath_visitor.rb
+test/files/dont_hurt_em_why.xml
 test/files/staff.xml
 test/files/staff.xslt
 test/files/tlm.html
@@ -106,6 +107,7 @@ test/html/sax/test_parser.rb
 test/html/test_builder.rb
 test/html/test_document.rb
 test/test_convert_xpath.rb
+test/test_gc.rb
 test/test_nokogiri.rb
 test/test_reader.rb
 test/test_xslt_transforms.rb

data/README.ja.txt

@@ -14,10 +14,11 @@ Nokogiri はHTMLやXMLやSAXやXSLTやReaderのパーサーです。
 * CSS3 のセレクターで探せる
 * XML/HTMLのビルダーはある
 
-NokogiriはHpricotより早くパーサーし、検索出来たり、
-正確にCSS3とXPathをサポート出来たりする。
+検索出来たり、正確にCSS3とXPathをサポート出来たりする。
 
-  * http://gist.github.com/18533
+これはスピッドテストです：
+
+  * http://gist.github.com/22176
 
 NokogiriはHpricotの代わりに使用出来る。
 その互換性は簡単に正しいCSSとXPathを使用する事が出来る。

data/README.txt

@@ -13,12 +13,14 @@ Nokogiri (鋸) is an HTML, XML, SAX, and Reader parser.
 * XPath support for document searching
 * CSS3 selector support for document searching
 * XML/HTML builder
-* Drop in replacement for Hpricot
+* Drop in replacement for Hpricot (though not bug for bug)
 
-Nokogiri parses and searches XML/HTML faster than Hpricot, and also has
+Nokogiri parses and searches XML/HTML very quickly, and also has
 correctly implemented CSS3 selector support as well as XPath support.
 
-  * http://gist.github.com/18533
+Here is a speed test:
+
+  * http://gist.github.com/22176
 
 Nokogiri also features an Hpricot compatibility layer to help ease the change
 to using correct CSS and XPath.

data/ext/nokogiri/html_document.c

@@ -15,7 +15,7 @@ static VALUE serialize(VALUE self)
 
   htmlDocDumpMemory(doc, &buf, &size);
   VALUE rb_str = rb_str_new((char *)buf, (long)size);
-  free(buf);
+  xmlFree(buf);
   return rb_str;
 }
 

data/ext/nokogiri/native.h

@@ -34,15 +34,13 @@ extern VALUE mNokogiriXslt ;
 
 #ifdef DEBUG
 
-#define NOKOGIRI_DEBUG_START_NODE(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"\nnokogiri: %s:%d %p start node (%p %x %p/%s)\n", __FILE__, __LINE__, p, p->_private, p->type, p->name, p->name);
-#define NOKOGIRI_DEBUG_START_TEXT(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"\nnokogiri: %s:%d %p start node (%p %x %p/%s) '%s'\n", __FILE__, __LINE__, p, p->_private, p->type, p->name, p->name, p->content);
-#define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
+#define NOKOGIRI_DEBUG_START_NODE(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start node (%p %x %p/%s)\n", __FILE__, __LINE__, p, p->_private, p->type, p->name, p->name);
+#define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
 #define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
 
 #else
 
 #define NOKOGIRI_DEBUG_START_NODE(p)
-#define NOKOGIRI_DEBUG_START_TEXT(p)
 #define NOKOGIRI_DEBUG_START(p)
 #define NOKOGIRI_DEBUG_END(p)
 

data/ext/nokogiri/xml_cdata.c

@@ -1,14 +1,5 @@
 #include <xml_cdata.h>
 
-static void dealloc(xmlNodePtr node)
-{
-  if (node->doc == NULL) {
-    NOKOGIRI_DEBUG_START_NODE(node);
-    xmlFreeNode(node);
-    NOKOGIRI_DEBUG_END(node);
-  }
-}
-
 /*
  * call-seq:
  *  new(document, content)
@@ -26,8 +17,7 @@ static VALUE new(VALUE klass, VALUE doc, VALUE content)
       NUM2INT(rb_funcall(content, rb_intern("length"), 0))
   );
 
-  VALUE rb_node = Data_Wrap_Struct(klass, NULL, dealloc, node);
-  node->_private = (void *)rb_node;
+  VALUE rb_node = Nokogiri_wrap_xml_node(node);
 
   if(rb_block_given_p()) rb_yield(rb_node);
 

data/ext/nokogiri/xml_document.c

@@ -6,17 +6,19 @@
  */
 static void gc_mark(xmlNodePtr node)
 {
+  VALUE rb_obj ;
   xmlNodePtr child ;
   /* mark children nodes */
   for (child = node->children ; child ; child = child->next) {
-    if (child->_private)
-      rb_gc_mark((VALUE)child->_private);
+    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);
   xmlFreeDoc(doc);
   NOKOGIRI_DEBUG_END(doc);
 }
@@ -36,7 +38,7 @@ static VALUE serialize(VALUE self)
 
   xmlDocDumpMemory(doc, &buf, &size);
   VALUE rb_str = rb_str_new((char *)buf, (long)size);
-  free(buf);
+  xmlFree(buf);
   return rb_str;
 }
 
@@ -145,15 +147,20 @@ void init_xml_document()
   rb_define_method(klass, "root", root, 0);
   rb_define_method(klass, "root=", set_root, 1);
   rb_define_method(klass, "serialize", serialize, 0);
+  rb_undef_method(klass, "parent");
 }
 
 
 /* this takes klass as a param because it's used for HtmlDocument, too. */
 VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
 {
-  if (doc->_private)
-    return (VALUE)doc->_private ;
-  VALUE rb_doc = Data_Wrap_Struct(klass ? klass : cNokogiriXmlDocument, gc_mark, dealloc, doc) ;
-  doc->_private = (void*)rb_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) ;
+
+  Nokogiri_xml_node2obj_set((xmlNodePtr)doc, rb_doc);
   return rb_doc ;
 }

data/ext/nokogiri/xml_node.c

@@ -16,7 +16,7 @@ static VALUE encode_special_chars(VALUE self, VALUE string)
   );
 
   VALUE encoded_str = rb_str_new2((const char *)encoded);
-  free(encoded);
+  xmlFree(encoded);
 
   return encoded_str;
 }
@@ -403,7 +403,7 @@ static VALUE document(VALUE self)
   Data_Get_Struct(self, xmlNode, node);
 
   if(!node->doc) return Qnil;
-  return (VALUE)node->doc->_private;
+  return Nokogiri_xml_node2obj_get(node->doc);
 }
 
 /*
@@ -510,6 +510,7 @@ static VALUE new_from_str(VALUE klass, VALUE xml)
 
 static void deallocate(xmlNodePtr node)
 {
+  Nokogiri_xml_node2obj_remove(node);
   if (! Nokogiri_xml_node_owned_get(node)) {
     NOKOGIRI_DEBUG_START_NODE(node);
     xmlFreeNode(node);
@@ -520,30 +521,31 @@ static void deallocate(xmlNodePtr node)
 static void gc_mark_node(xmlNodePtr node)
 {
   xmlNodePtr child ;
+  VALUE rb_obj ;
   /* mark document */
-  if (node && node->doc && node->doc->_private)
-    rb_gc_mark((VALUE)node->doc->_private);
+  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 && node->parent->_private)
-    rb_gc_mark((VALUE)node->parent->_private);
+  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 (child->_private)
-      rb_gc_mark((VALUE)child->_private);
+    if ((rb_obj = Nokogiri_xml_node2obj_get(child)) != Qnil)
+      rb_gc_mark(rb_obj);
   }
   /* mark sibling nodes */
-  if (node->next && node->next->_private)
-    rb_gc_mark((VALUE)node->next->_private) ;
-  if (node->prev && node->prev->_private)
-    rb_gc_mark((VALUE)node->prev->_private) ;
+  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);
 }
 
 VALUE Nokogiri_wrap_xml_node(xmlNodePtr node)
 {
-  if (node->_private)
-    return (VALUE)node->_private ;
-
   VALUE rb_node = Qnil;
+
+  if ((rb_node = Nokogiri_xml_node2obj_get(node)) != Qnil)
+    return rb_node ;
   
   switch(node->type)
   {
@@ -573,7 +575,7 @@ VALUE Nokogiri_wrap_xml_node(xmlNodePtr node)
       rb_node = Data_Wrap_Struct(cNokogiriXmlNode, gc_mark_node, deallocate, node) ;
   }
 
-  node->_private = (void*)rb_node ;
+  Nokogiri_xml_node2obj_set(node, rb_node);
   rb_funcall(rb_node, rb_intern("decorate!"), 0);
   Nokogiri_xml_node_owned_set(node);
   return rb_node ;
@@ -632,6 +634,36 @@ 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"));

data/ext/nokogiri/xml_node.h

@@ -12,4 +12,7 @@ void Nokogiri_xml_node_properties(xmlNodePtr node, VALUE attr_hash) ;
 void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash) ;
 int Nokogiri_xml_node_owned_get(xmlNodePtr node) ;
 void Nokogiri_xml_node_owned_set(xmlNodePtr node) ;
+void Nokogiri_xml_node2obj_set(xmlNodePtr node, VALUE rb_obj) ;
+VALUE Nokogiri_xml_node2obj_get(xmlNodePtr node) ;
+void Nokogiri_xml_node2obj_remove(xmlNodePtr node) ;
 #endif

data/ext/nokogiri/xml_node_set.c

@@ -57,11 +57,12 @@ static VALUE index_at(VALUE self, VALUE number)
 
 static void gc_mark(xmlNodeSetPtr node_set)
 {
-    int j ;
-    for (j = 0 ; j < node_set->nodeNr ; ++j) {
-        if (node_set->nodeTab[j]->_private)
-            rb_gc_mark((VALUE)node_set->nodeTab[j]->_private);
-    }
+  VALUE rb_obj ;
+  int j ;
+  for (j = 0 ; j < node_set->nodeNr ; ++j) {
+    if ((rb_obj = Nokogiri_xml_node2obj_get(node_set->nodeTab[j])) != Qnil)
+      rb_gc_mark(rb_obj);
+  }
 }
 
 static void deallocate(xmlNodeSetPtr node_set)

data/ext/nokogiri/xml_reader.c

@@ -151,7 +151,7 @@ static VALUE reader_attribute(VALUE self, VALUE name)
     xmlChar *localname = xmlSplitQName2((xmlChar*)StringValuePtr(name), &prefix);
     if (localname != NULL) {
       value = xmlTextReaderLookupNamespace(reader, localname);
-      free(localname) ;
+      xmlFree(localname) ;
     } else {
       value = xmlTextReaderLookupNamespace(reader, prefix);
     }

data/ext/nokogiri/xml_syntax_error.c

@@ -3,7 +3,7 @@
 static void dealloc(xmlErrorPtr ptr)
 {
   NOKOGIRI_DEBUG_START(ptr);
-  free(ptr);
+  xmlFree(ptr);
   NOKOGIRI_DEBUG_END(ptr);
 }
 

data/ext/nokogiri/xml_text.c

@@ -1,19 +1,9 @@
 #include <xml_text.h>
 
-static void dealloc(xmlNodePtr node)
-{
-  if (node->doc == NULL) {
-    NOKOGIRI_DEBUG_START_TEXT(node);
-    xmlFreeNode(node);
-    NOKOGIRI_DEBUG_END(node);
-  }
-}
-
 static VALUE new(VALUE klass, VALUE string)
 {
   xmlNodePtr node = xmlNewText((xmlChar *)StringValuePtr(string));
-  VALUE rb_node = Data_Wrap_Struct(klass, NULL, dealloc, node);
-  node->_private = (void *)rb_node;
+  VALUE rb_node = Nokogiri_wrap_xml_node(node) ;
 
   if(rb_block_given_p()) rb_yield(rb_node);
 

data/ext/nokogiri/xml_xpath_context.c

@@ -7,6 +7,13 @@ static void deallocate(xmlXPathContextPtr ctx)
   NOKOGIRI_DEBUG_END(ctx);
 }
 
+static void gc_mark_context(xmlXPathContextPtr ctx)
+{
+  VALUE rb_obj ;
+  if (ctx && ctx->node && (rb_obj = Nokogiri_xml_node2obj_get(ctx->node)) != Qnil)
+    rb_gc_mark(rb_obj);
+}
+
 /*
  * call-seq:
  *  register_ns(prefix, uri)
@@ -59,7 +66,7 @@ static VALUE new(VALUE klass, VALUE nodeobj)
 
   xmlXPathContextPtr ctx = xmlXPathNewContext(node->doc);
   ctx->node = node ;
-  return Data_Wrap_Struct(klass, 0, deallocate, ctx);
+  return Data_Wrap_Struct(klass, gc_mark_context, deallocate, ctx);
 }
 
 VALUE cNokogiriXmlXpathContext;

data/ext/nokogiri/xslt_stylesheet.c

@@ -45,7 +45,7 @@ static VALUE serialize(VALUE self, VALUE xmlobj)
     Data_Get_Struct(self, xsltStylesheet, ss);
     xsltSaveResultToString(&doc_ptr, &doc_len, xml, ss);
     rval = rb_str_new((char*)doc_ptr, doc_len);
-    free(doc_ptr);
+    xmlFree(doc_ptr);
     return rval ;
 }
 
@@ -73,7 +73,7 @@ static VALUE apply_to(int argc, VALUE* argv, VALUE self)
     Data_Get_Struct(xmldoc, xmlDoc, xml);
     Data_Get_Struct(self, xsltStylesheet, ss);
 
-    param_len = RARRAY_LEN(paramobj);
+    param_len = RARRAY(paramobj)->len;
     params = calloc((size_t)param_len+1, sizeof(char*));
     for (j = 0 ; j < param_len ; j++) {
       VALUE entry = rb_ary_entry(paramobj, j);

data/lib/nokogiri/version.rb

@@ -1,3 +1,3 @@
 module Nokogiri
-  VERSION = '1.0.2'
+  VERSION = '1.0.3'
 end

data/lib/nokogiri/xml/document.rb

@@ -22,6 +22,10 @@ module Nokogiri
         serialize
       end
 
+      def inner_html
+        serialize
+      end
+
       def namespaces
         root ? root.collect_namespaces : {}
       end

data/lib/nokogiri/xml/node.rb

@@ -15,6 +15,7 @@ module Nokogiri
       DOCB_DOCUMENT_NODE = 21
 
       @@owned = {}
+      @@node2obj = {}
 
       ###
       # Decorate this node with the decorators set up in this node's Document
@@ -168,7 +169,10 @@ module Nokogiri
         to_xml
       end
       alias :to_s :to_html
-      alias :inner_html :to_html
+
+      def inner_html
+        children.map { |x| x.to_html }.join
+      end
 
       def css_path
         path.split(/\//).map { |part|

data/lib/nokogiri/xml/node_set.rb

@@ -178,6 +178,10 @@ module Nokogiri
       def size
         length
       end
+
+      def to_ary
+        to_a
+      end
     end
   end
 end

data/test/files/dont_hurt_em_why.xml

@@ -0,0 +1,422 @@
+<statuses type="array"> 
+<status> 
+  <created_at>Sat Aug 09 05:38:12 +0000 2008</created_at> 
+  <id>882281424</id> 
+  <text>I so just thought the guy lighting the Olympic torch was falling when he began to run on the wall. Wow that would have been catastrophic.</text> 
+  <source>web</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Sat Aug 09 02:04:56 +0000 2008</created_at> 
+  <id>882145663</id> 
+  <text>@ijonas - wow that stuff sounds sweet</text> 
+  <source>web</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id>882005142</in_reply_to_status_id> 
+  <in_reply_to_user_id>1000471</in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Sat Aug 09 01:36:41 +0000 2008</created_at> 
+  <id>882126691</id> 
+  <text>Steph is driving Sally for the first time. I'm proud of her.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Fri Aug 08 22:21:21 +0000 2008</created_at> 
+  <id>881987762</id> 
+  <text>@ijonas - what are you making with couchdb, ruby and httparty?</text> 
+  <source>web</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id>881947237</in_reply_to_status_id> 
+  <in_reply_to_user_id>1000471</in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Fri Aug 08 14:20:14 +0000 2008</created_at> 
+  <id>881535796</id> 
+  <text>@oaknd1 - delete it off phone and iTunes.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id>881526234</in_reply_to_status_id> 
+  <in_reply_to_user_id>3038211</in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Fri Aug 08 14:07:29 +0000 2008</created_at> 
+  <id>881522394</id> 
+  <text>Listening to U2 "beautiful day" in honor of it being such a beautiful day.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Fri Aug 08 14:06:44 +0000 2008</created_at> 
+  <id>881521592</id> 
+  <text>@lizsmc1 - hi! (just passed her on the road)</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id>881519558</in_reply_to_status_id> 
+  <in_reply_to_user_id>11485452</in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Fri Aug 08 13:59:35 +0000 2008</created_at> 
+  <id>881514030</id> 
+  <text>Beautiful day for a motorcycle ride.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Fri Aug 08 13:45:21 +0000 2008</created_at> 
+  <id>881499439</id> 
+  <text>@lizamc1 - no way! Politos will be missed. I remember eating a whole garlic pizza there with Joe.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Fri Aug 08 13:41:50 +0000 2008</created_at> 
+  <id>881496024</id> 
+  <text>Riding my motorcyle to campus. Using the library basement for a meeting.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 22:52:20 +0000 2008</created_at> 
+  <id>880896190</id> 
+  <text>Scraping super glue off my finger with a knife.</text> 
+  <source>web</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 22:14:35 +0000 2008</created_at> 
+  <id>880866160</id> 
+  <text>So cold...Starbucks north side, you win the day, but I shall bring a sweatshirt to our next battle!</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 17:25:40 +0000 2008</created_at> 
+  <id>880610064</id> 
+  <text>Headed home for a bit to get my headphones and then to the north side to meet @orderedlist.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>485</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 17:15:59 +0000 2008</created_at> 
+  <id>880600278</id> 
+  <text>Panera wifi, why do you hate me?</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>485</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 15:46:25 +0000 2008</created_at> 
+  <id>880509577</id> 
+  <text>At panera. Turned my alarm off this morning and woke up late. Oh well.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>485</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 15:01:35 +0000 2008</created_at> 
+  <id>880463746</id> 
+  <text>@kloh I remember days like that. Exhausting. Also, NullRiver said to sync again, uninstall app from phone and then resync app. Evidently ...</text> 
+  <source>web</source> 
+  <truncated>true</truncated> 
+  <in_reply_to_status_id>880432701</in_reply_to_status_id> 
+  <in_reply_to_user_id>10193732</in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>485</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 02:52:38 +0000 2008</created_at> 
+  <id>879986739</id> 
+  <text>@kloh - I haven't updated my OS so I'm wondering how apple could have made the app stop working. I did contact NullRiver support just no ...</text> 
+  <source>web</source> 
+  <truncated>true</truncated> 
+  <in_reply_to_status_id>879980813</in_reply_to_status_id> 
+  <in_reply_to_user_id>10193732</in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 02:38:58 +0000 2008</created_at> 
+  <id>879976045</id> 
+  <text>@jerry - i went with the pdf download.</text> 
+  <source>web</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id>879878196</in_reply_to_status_id> 
+  <in_reply_to_user_id>613</in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 02:31:26 +0000 2008</created_at> 
+  <id>879969851</id> 
+  <text>@kloh - it worked at home for multiple tests and just stopped when I acyltually needed it.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id>879968483</in_reply_to_status_id> 
+  <in_reply_to_user_id>10193732</in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+<status> 
+  <created_at>Thu Aug 07 01:18:28 +0000 2008</created_at> 
+  <id>879913748</id> 
+  <text>Netshare will no longer start up for me. Of course it borks the first time I actually want to use it.</text> 
+  <source>&lt;a href="http://help.twitter.com/index.php?pg=kb.page&amp;id=75"&gt;txt&lt;/a&gt;</source> 
+  <truncated>false</truncated> 
+  <in_reply_to_status_id></in_reply_to_status_id> 
+  <in_reply_to_user_id></in_reply_to_user_id> 
+  <favorited></favorited> 
+  <user> 
+    <id>4243</id> 
+    <name>John Nunemaker</name> 
+    <screen_name>jnunemaker</screen_name> 
+    <location>Mishawaka, IN, US</location> 
+    <description>Loves his wife, ruby, notre dame football and iu basketball</description> 
+    <profile_image_url>http://s3.amazonaws.com/twitter_production/profile_images/53781608/Photo_75_normal.jpg</profile_image_url> 
+    <url>http://addictedtonew.com</url> 
+    <protected>false</protected> 
+    <followers_count>486</followers_count> 
+  </user> 
+</status> 
+</statuses>

data/test/helper.rb

@@ -16,7 +16,10 @@ module Nokogiri
     undef :default_test
 
     def teardown
-      GC.start if ENV['NOKOGIRI_GC']
+      if ENV['NOKOGIRI_GC']
+        STDOUT.putc '!'
+        GC.start 
+      end
     end
   end
 

data/test/html/test_document.rb

@@ -12,6 +12,38 @@ module Nokogiri
         assert html.html?
       end
 
+      def test_inner_text
+        html = Nokogiri::HTML(<<-eohtml)
+        <html>
+          <body>
+            <div>
+              <p>
+                Hello world!
+              </p>
+            </div>
+          </body>
+        </html>
+        eohtml
+        node = html.xpath('//div').first
+        assert_equal('Hello world!', node.inner_text.strip)
+      end
+
+      def test_inner_html
+        html = Nokogiri::HTML(<<-eohtml)
+        <html>
+          <body>
+            <div>
+              <p>
+                Hello world!
+              </p>
+            </div>
+          </body>
+        </html>
+        eohtml
+        node = html.xpath('//div').first
+        assert_equal('<p>Helloworld!</p>', node.inner_html.gsub(/\s/, ''))
+      end
+
       def test_fragment
         node_set = Nokogiri::HTML.fragment(<<-eohtml)
           <div>

data/test/test_gc.rb

@@ -0,0 +1,15 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+
+class TestGc < Nokogiri::TestCase
+
+  def test_dont_hurt_em_why
+    content = File.open("#{File.dirname(__FILE__)}/files/dont_hurt_em_why.xml").read
+    ndoc = Nokogiri::XML(content)
+    2.times do
+      info = ndoc.search('status text').first.inner_text
+      url = ndoc.search('user name').first.inner_text
+      GC.start
+    end
+  end
+
+end

data/test/xml/test_document.rb

@@ -12,6 +12,13 @@ module Nokogiri
         assert xml.xml?
       end
 
+      def test_document_parent
+        xml = Nokogiri::XML(File.read(XML_FILE), XML_FILE)
+        assert_raises(NoMethodError) {
+          xml.parent
+        }
+      end
+
       def test_document_name
         xml = Nokogiri::XML(File.read(XML_FILE), XML_FILE)
         assert_equal 'document', xml.name

data/test/xml/test_node_set.rb

@@ -17,6 +17,13 @@ module Nokogiri
         assert_equal node_set.first, node_set.at(0)
       end
 
+      def test_to_ary
+        assert node_set = @xml.search('//employee')
+        foo = []
+        foo += node_set
+        assert_equal node_set.length, foo.length
+      end
+
       def test_push
         node = Nokogiri::XML::Node.new('foo')
         node.content = 'bar'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: nokogiri
 version: !ruby/object:Gem::Version 
-  version: 1.0.2
+  version: 1.0.3
 platform: ruby
 authors: 
 - Aaron Patterson
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2008-10-31 00:00:00 -07:00
+date: 2008-11-04 00:00:00 -08:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -118,6 +118,7 @@ files:
 - test/css/test_parser.rb
 - test/css/test_tokenizer.rb
 - test/css/test_xpath_visitor.rb
+- test/files/dont_hurt_em_why.xml
 - test/files/staff.xml
 - test/files/staff.xslt
 - test/files/tlm.html
@@ -143,6 +144,7 @@ files:
 - test/html/test_builder.rb
 - test/html/test_document.rb
 - test/test_convert_xpath.rb
+- test/test_gc.rb
 - test/test_nokogiri.rb
 - test/test_reader.rb
 - test/test_xslt_transforms.rb
@@ -197,6 +199,7 @@ test_files:
 - test/html/test_builder.rb
 - test/html/test_document.rb
 - test/test_convert_xpath.rb
+- test/test_gc.rb
 - test/test_nokogiri.rb
 - test/test_reader.rb
 - test/test_xslt_transforms.rb