nokogiri 1.2.3 → 1.3.0

data/ext/nokogiri/html_element_description.h

@@ -0,0 +1,10 @@
+#ifndef NOKOGIRI_HTML_ELEMENT_DESCRIPTION
+#define NOKOGIRI_HTML_ELEMENT_DESCRIPTION
+
+#include <nokogiri.h>
+
+void init_html_element_description();
+
+extern VALUE cNokogiriHtmlElementDescription ;
+
+#endif

data/ext/nokogiri/html_entity_lookup.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_HTML_ENTITY_LOOKUP
 #define NOKOGIRI_HTML_ENTITY_LOOKUP
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_html_entity_lookup();
 

data/ext/nokogiri/html_sax_parser.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_HTML_SAX_PARSER
 #define NOKOGIRI_HTML_SAX_PARSER
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_html_sax_parser();
 

data/ext/nokogiri/{native.c → nokogiri.c}

@@ -1,4 +1,4 @@
-#include <native.h>
+#include <nokogiri.h>
 
 VALUE mNokogiri ;
 VALUE mNokogiriXml ;
@@ -7,7 +7,7 @@ VALUE mNokogiriXslt ;
 VALUE mNokogiriXmlSax ;
 VALUE mNokogiriHtmlSax ;
 
-#ifdef XP_WIN
+#ifdef USE_INCLUDED_VASPRINTF
 /*
  * I srsly hate windows.  it doesn't have vasprintf.
  * Thank you Geoffroy Couprie for this implementation of vasprintf!
@@ -23,8 +23,12 @@ int vasprintf (char **strp, const char *fmt, va_list ap)
 }
 #endif
 
+int is_2_6_16(void)
+{
+  return (strcmp(xmlParserVersion, "20616") <= 0) ? 1 : 0 ;
+}
 
-void Init_native()
+void Init_nokogiri()
 {
   xmlMemSetup(
       (xmlFreeFunc)ruby_xfree,
@@ -66,8 +70,12 @@ void Init_native()
   init_xml_sax_push_parser();
   init_xml_reader();
   init_xml_dtd();
+  init_xml_namespace();
   init_html_sax_parser();
   init_xslt_stylesheet();
   init_xml_syntax_error();
   init_html_entity_lookup();
+  init_html_element_description();
+  init_xml_schema();
+  init_xml_relax_ng();
 }

data/ext/nokogiri/{native.h → nokogiri.h}

@@ -9,9 +9,23 @@
 #include <libxml/xpathInternals.h>
 #include <libxml/xmlreader.h>
 #include <libxml/xmlsave.h>
+#include <libxml/xmlschemas.h>
 #include <libxml/HTMLparser.h>
 #include <libxml/HTMLtree.h>
 
+#ifdef USE_INCLUDED_VASPRINTF
+int vasprintf (char **strp, const char *fmt, va_list ap);
+#else
+
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE
+#endif
+
+#  include <stdio.h>
+
+#endif
+
+int is_2_6_16(void) ;
 
 #ifndef UNUSED
 # if defined(__GNUC__)
@@ -62,10 +76,6 @@
   rb_str_new((const char *)str, (long)len)
 #endif
 
-#ifdef XP_WIN
-int vasprintf (char **strp, const char *fmt, va_list ap);
-#endif
-
 #include <xml_io.h>
 #include <xml_document.h>
 #include <html_entity_lookup.h>
@@ -88,6 +98,10 @@ int vasprintf (char **strp, const char *fmt, va_list ap);
 #include <html_sax_parser.h>
 #include <xslt_stylesheet.h>
 #include <xml_syntax_error.h>
+#include <xml_schema.h>
+#include <xml_relax_ng.h>
+#include <html_element_description.h>
+#include <xml_namespace.h>
 
 extern VALUE mNokogiri ;
 extern VALUE mNokogiriXml ;

data/ext/nokogiri/xml_attr.c

@@ -42,14 +42,20 @@ static VALUE set_value(VALUE self, VALUE content)
 
 /*
  * call-seq:
- *  new(document, content)
+ *  new(document, name)
  *
  * Create a new Attr element on the +document+ with +name+
  */
-static VALUE new(VALUE klass, VALUE doc, VALUE name)
+static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
-  Data_Get_Struct(doc, xmlDoc, xml_doc);
+  VALUE document;
+  VALUE name;
+  VALUE rest;
+
+  rb_scan_args(argc, argv, "2*", &document, &name, &rest);
+
+  Data_Get_Struct(document, xmlDoc, xml_doc);
 
   xmlAttrPtr node = xmlNewDocProp(
       xml_doc,
@@ -57,7 +63,10 @@ static VALUE new(VALUE klass, VALUE doc, VALUE name)
       NULL
   );
 
-  VALUE rb_node = Nokogiri_wrap_xml_node((xmlNodePtr)node);
+  NOKOGIRI_ROOT_NODE((xmlNodePtr)node);
+
+  VALUE rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
+  rb_funcall2(rb_node, rb_intern("initialize"), argc, argv);
 
   if(rb_block_given_p()) rb_yield(rb_node);
 
@@ -78,6 +87,6 @@ void init_xml_attr()
 
   cNokogiriXmlAttr = klass;
 
-  rb_define_singleton_method(klass, "new", new, 2);
+  rb_define_singleton_method(klass, "new", new, -1);
   rb_define_method(klass, "value=", set_value, 1);
 }

data/ext/nokogiri/xml_attr.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_XML_ATTR
 #define NOKOGIRI_XML_ATTR
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xml_attr();
 

data/ext/nokogiri/xml_cdata.c

@@ -6,18 +6,27 @@
  *
  * Create a new CData element on the +document+ with +content+
  */
-static VALUE new(VALUE klass, VALUE doc, VALUE content)
+static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
+  VALUE doc;
+  VALUE content;
+  VALUE rest;
+
+  rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
+
   Data_Get_Struct(doc, xmlDoc, xml_doc);
 
   xmlNodePtr node = xmlNewCDataBlock(
-      xml_doc,
-      (const xmlChar *)StringValuePtr(content),
-      RSTRING_LEN(content)
+      xml_doc->doc,
+      Qnil == content ? NULL : (const xmlChar *)StringValuePtr(content),
+      Qnil == content ? 0 : RSTRING_LEN(content)
   );
 
-  VALUE rb_node = Nokogiri_wrap_xml_node(node);
+  NOKOGIRI_ROOT_NODE(node);
+
+  VALUE rb_node = Nokogiri_wrap_xml_node(klass, node);
+  rb_funcall2(rb_node, rb_intern("initialize"), argc, argv);
 
   if(rb_block_given_p()) rb_yield(rb_node);
 
@@ -40,5 +49,5 @@ void init_xml_cdata()
 
   cNokogiriXmlCData = klass;
 
-  rb_define_singleton_method(klass, "new", new, 2);
+  rb_define_singleton_method(klass, "new", new, -1);
 }

data/ext/nokogiri/xml_cdata.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_XML_CDATA
 #define NOKOGIRI_XML_CDATA
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xml_cdata();
 

data/ext/nokogiri/xml_comment.c

@@ -6,17 +6,26 @@
  *
  * Create a new Comment element on the +document+ with +content+
  */
-static VALUE new(VALUE klass, VALUE doc, VALUE content)
+static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
-  Data_Get_Struct(doc, xmlDoc, xml_doc);
+  VALUE document;
+  VALUE content;
+  VALUE rest;
+
+  rb_scan_args(argc, argv, "2*", &document, &content, &rest);
+
+  Data_Get_Struct(document, xmlDoc, xml_doc);
 
   xmlNodePtr node = xmlNewDocComment(
       xml_doc,
       (const xmlChar *)StringValuePtr(content)
   );
 
-  VALUE rb_node = Nokogiri_wrap_xml_node(node);
+  VALUE rb_node = Nokogiri_wrap_xml_node(klass, node);
+  rb_funcall2(rb_node, rb_intern("initialize"), argc, argv);
+
+  NOKOGIRI_ROOT_NODE(node);
 
   if(rb_block_given_p()) rb_yield(rb_node);
 
@@ -38,5 +47,5 @@ void init_xml_comment()
 
   cNokogiriXmlComment = klass;
 
-  rb_define_singleton_method(klass, "new", new, 2);
+  rb_define_singleton_method(klass, "new", new, -1);
 }

data/ext/nokogiri/xml_comment.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_XML_COMMENT
 #define NOKOGIRI_XML_COMMENT
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xml_comment();
 

data/ext/nokogiri/xml_document.c

@@ -7,9 +7,21 @@ static void dealloc(xmlDocPtr doc)
   nokogiriTuplePtr tuple = doc->_private;
   xmlNodeSetPtr node_set = tuple->unlinkedNodes;
 
-  int i;
-  for(i = 0; i < node_set->nodeNr; i++) {
-    xmlAddChild((xmlNodePtr)doc, node_set->nodeTab[i]);
+  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)
@@ -20,6 +32,8 @@ static void dealloc(xmlDocPtr doc)
   doc->_private = NULL;
   xmlFreeDoc(doc);
 
+  xmlDeregisterNodeDefault(func);
+
   NOKOGIRI_DEBUG_END(doc);
 }
 
@@ -72,7 +86,23 @@ static VALUE root(VALUE self)
   xmlNodePtr root = xmlDocGetRootElement(doc);
 
   if(!root) return Qnil;
-  return Nokogiri_wrap_xml_node(root) ;
+  return Nokogiri_wrap_xml_node(Qnil, root) ;
+}
+
+/*
+ * call-seq:
+ *  encoding= encoding
+ *
+ * Set the encoding string for this Document
+ */
+static VALUE set_encoding(VALUE self, VALUE encoding)
+{
+  xmlDocPtr doc;
+  Data_Get_Struct(self, xmlDoc, doc);
+
+  doc->encoding = xmlStrdup((xmlChar *)StringValuePtr(encoding));
+
+  return encoding;
 }
 
 /*
@@ -203,50 +233,29 @@ static VALUE duplicate_node(int argc, VALUE *argv, VALUE self)
   if(dup == NULL) return Qnil;
 
   dup->type = doc->type;
-  if(dup->type == XML_DOCUMENT_NODE)
-    return Nokogiri_wrap_xml_document(cNokogiriXmlDocument, dup);
-  else
-    return Nokogiri_wrap_xml_document(cNokogiriHtmlDocument, dup);
+  return Nokogiri_wrap_xml_document(RBASIC(self)->klass, dup);
 }
 
 /*
  * call-seq:
- *  new
+ *  new(version = default)
  *
- * Create a new document
+ * Create a new document with +version+ (defaults to "1.0")
  */
 static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
-  VALUE version;
-  if(rb_scan_args(argc, argv, "01", &version) == 0)
+  VALUE version, rest, rb_doc ;
+
+  rb_scan_args(argc, argv, "0*", &rest);
+  version = rb_ary_entry(rest, 0);
+  if (version == Qnil) {
     version = rb_str_new2("1.0");
+  }
 
   xmlDocPtr doc = xmlNewDoc((xmlChar *)StringValuePtr(version));
-  return Nokogiri_wrap_xml_document(klass, doc);
-}
-
-/*
- *  call-seq:
- *    substitute_entities=(boolean)
- *
- *  Set the global XML default for substitute entities.
- */
-static VALUE substitute_entities_set(VALUE klass, VALUE value)
-{
-    xmlSubstituteEntitiesDefault(NUM2INT(value));
-    return Qnil ;
-}
-
-/*
- *  call-seq:
- *    load_external_subsets=(boolean)
- *
- *  Set the global XML default for load external subsets.
- */
-static VALUE load_external_subsets_set(VALUE klass, VALUE value)
-{
-    xmlLoadExtDtdDefaultValue = NUM2INT(value);
-    return Qnil ;
+  rb_doc = Nokogiri_wrap_xml_document(klass, doc);
+  rb_funcall2(rb_doc, rb_intern("initialize"), argc, argv);
+  return rb_doc ;
 }
 
 VALUE cNokogiriXmlDocument ;
@@ -266,12 +275,11 @@ void init_xml_document()
   rb_define_singleton_method(klass, "read_memory", read_memory, 4);
   rb_define_singleton_method(klass, "read_io", read_io, 4);
   rb_define_singleton_method(klass, "new", new, -1);
-  rb_define_singleton_method(klass, "substitute_entities=", substitute_entities_set, 1);
-  rb_define_singleton_method(klass, "load_external_subsets=", load_external_subsets_set, 1);
 
   rb_define_method(klass, "root", root, 0);
   rb_define_method(klass, "root=", set_root, 1);
   rb_define_method(klass, "encoding", encoding, 0);
+  rb_define_method(klass, "encoding=", set_encoding, 1);
   rb_define_method(klass, "dup", duplicate_node, -1);
   rb_define_method(klass, "url", url, 0);
 }
@@ -280,16 +288,17 @@ void init_xml_document()
 /* this takes klass as a param because it's used for HtmlDocument, too. */
 VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
 {
-  VALUE rb_doc = Qnil;
   nokogiriTuplePtr tuple = (nokogiriTuplePtr)malloc(sizeof(nokogiriTuple));
 
-  rb_doc = Data_Wrap_Struct(
+  VALUE rb_doc = Data_Wrap_Struct(
       klass ? klass : cNokogiriXmlDocument,
       0,
       dealloc,
       doc
   );
   rb_iv_set(rb_doc, "@decorators", Qnil);
+  rb_iv_set(rb_doc, "@node_cache", rb_ary_new());
+  rb_funcall(rb_doc, rb_intern("initialize"), 0);
 
   tuple->doc = (void *)rb_doc;
   tuple->unlinkedNodes = xmlXPathNodeSetCreate(NULL);

data/ext/nokogiri/xml_document.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_XML_DOCUMENT
 #define NOKOGIRI_XML_DOCUMENT
 
-#include <native.h>
+#include <nokogiri.h>
 
 struct _nokogiriTuple {
   xmlDocPtr     doc;

data/ext/nokogiri/xml_document_fragment.c

@@ -6,16 +6,24 @@
  *
  * Create a new DocumentFragment element on the +document+
  */
-static VALUE new(VALUE klass, VALUE doc)
+static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
-  Data_Get_Struct(doc, xmlDoc, xml_doc);
+  VALUE document;
+  VALUE rest;
+
+  rb_scan_args(argc, argv, "1*", &document, &rest);
+
+  Data_Get_Struct(document, xmlDoc, xml_doc);
 
   xmlNodePtr node = xmlNewDocFragment(xml_doc->doc);
   if(node->doc->children)
     node->ns = node->doc->children->ns;
 
-  VALUE rb_node = Nokogiri_wrap_xml_node(node);
+  NOKOGIRI_ROOT_NODE(node);
+
+  VALUE rb_node = Nokogiri_wrap_xml_node(klass, node);
+  rb_funcall2(rb_node, rb_intern("initialize"), argc, argv);
 
   if(rb_block_given_p()) rb_yield(rb_node);
 
@@ -36,5 +44,5 @@ void init_xml_document_fragment()
 
   cNokogiriXmlDocumentFragment = klass;
 
-  rb_define_singleton_method(klass, "new", new, 1);
+  rb_define_singleton_method(klass, "new", new, -1);
 }