nokogiri 1.6.7.2-java → 1.6.8-java

data/ext/nokogiri/html_document.c

@@ -18,8 +18,8 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   external_id = rb_ary_entry(rest, (long)1);
 
   doc = htmlNewDoc(
-      RTEST(uri) ? (const xmlChar *)StringValuePtr(uri) : NULL,
-      RTEST(external_id) ? (const xmlChar *)StringValuePtr(external_id) : NULL
+      RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
+      RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
   );
   rb_doc = Nokogiri_wrap_xml_document(klass, doc);
   rb_obj_call_init(rb_doc, argc, argv);
@@ -39,8 +39,8 @@ static VALUE read_io( VALUE klass,
                       VALUE encoding,
                       VALUE options )
 {
-  const char * c_url    = NIL_P(url)      ? NULL : StringValuePtr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
+  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
   VALUE error_list      = rb_ary_new();
   VALUE document;
   htmlDocPtr doc;
@@ -103,8 +103,8 @@ static VALUE read_memory( VALUE klass,
                           VALUE options )
 {
   const char * c_buffer = StringValuePtr(string);
-  const char * c_url    = NIL_P(url)      ? NULL : StringValuePtr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
+  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
   int len               = (int)RSTRING_LEN(string);
   VALUE error_list      = rb_ary_new();
   VALUE document;

data/ext/nokogiri/html_element_description.c

@@ -245,7 +245,7 @@ static VALUE name(VALUE self)
 static VALUE get_description(VALUE klass, VALUE tag_name)
 {
   const htmlElemDesc * description = htmlTagLookup(
-      (const xmlChar *)StringValuePtr(tag_name)
+      (const xmlChar *)StringValueCStr(tag_name)
   );
 
   if(NULL == description) return Qnil;

data/ext/nokogiri/html_entity_lookup.c

@@ -9,7 +9,7 @@
 static VALUE get(VALUE self, VALUE key)
 {
   const htmlEntityDesc * desc =
-    htmlEntityLookup((const xmlChar *)StringValuePtr(key));
+    htmlEntityLookup((const xmlChar *)StringValueCStr(key));
   VALUE klass, args[3];
 
   if(NULL == desc) return Qnil;

data/ext/nokogiri/html_sax_parser_context.c

@@ -31,12 +31,12 @@ parse_memory(VALUE klass, VALUE data, VALUE encoding)
     }
 
     if (RTEST(encoding)) {
-	xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValuePtr(encoding));
+	xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValueCStr(encoding));
 	if (enc != NULL) {
 	    xmlSwitchToEncoding(ctxt, enc);
 	    if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
 		rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
-			 StringValuePtr(encoding));
+			 StringValueCStr(encoding));
 	    }
 	}
     }
@@ -47,8 +47,8 @@ parse_memory(VALUE klass, VALUE data, VALUE encoding)
 static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
 {
   htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
-      StringValuePtr(filename),
-      StringValuePtr(encoding)
+      StringValueCStr(filename),
+      StringValueCStr(encoding)
   );
   return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
 }

data/ext/nokogiri/html_sax_push_parser.c

@@ -46,10 +46,10 @@ static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
 
   Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
 
-  if(_filename != Qnil) filename = StringValuePtr(_filename);
+  if(_filename != Qnil) filename = StringValueCStr(_filename);
 
   if (!NIL_P(encoding)) {
-    enc = xmlParseCharEncoding(StringValuePtr(encoding));
+    enc = xmlParseCharEncoding(StringValueCStr(encoding));
     if (enc == XML_CHAR_ENCODING_ERROR)
       rb_raise(rb_eArgError, "Unsupported Encoding");
   }

data/ext/nokogiri/nokogiri.c

@@ -28,17 +28,10 @@ int vasprintf (char **strp, const char *fmt, va_list ap)
 }
 #endif
 
-#ifdef USING_SYSTEM_ALLOCATOR_LIBRARY /* Ruby Enterprise Edition with tcmalloc */
-void vasprintf_free (void *p)
-{
-  system_free(p);
-}
-#else
 void vasprintf_free (void *p)
 {
   free(p);
 }
-#endif
 
 #ifdef HAVE_RUBY_UTIL_H
 #include "ruby/util.h"

data/ext/nokogiri/nokogiri.h

@@ -31,12 +31,8 @@ int vasprintf (char **strp, const char *fmt, va_list ap);
 #include <libxslt/extensions.h>
 #include <libxml/c14n.h>
 #include <ruby.h>
-
-#ifdef HAVE_RUBY_ENCODING_H
 #include <ruby/st.h>
-#else
-#include <st.h>
-#endif
+#include <ruby/encoding.h>
 
 #ifndef UNUSED
 # if defined(__GNUC__)
@@ -56,25 +52,12 @@ int vasprintf (char **strp, const char *fmt, va_list ap);
 # endif
 #endif
 
-#ifdef HAVE_RUBY_ENCODING_H
-
-#include <ruby/encoding.h>
-
 #define NOKOGIRI_STR_NEW2(str) \
   NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
 
 #define NOKOGIRI_STR_NEW(str, len) \
   rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
 
-#else
-
-#define NOKOGIRI_STR_NEW2(str) \
-  rb_str_new2((const char *)(str))
-
-#define NOKOGIRI_STR_NEW(str, len) \
-  rb_str_new((const char *)(str), (long)(len))
-#endif
-
 #define RBSTR_OR_QNIL(_str) \
   (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
 
@@ -135,22 +118,6 @@ void nokogiri_root_nsdef(xmlNsPtr, xmlDocPtr);
 
 #endif
 
-#ifndef RSTRING_PTR
-#define RSTRING_PTR(s) (RSTRING(s)->ptr)
-#endif
-
-#ifndef RSTRING_LEN
-#define RSTRING_LEN(s) (RSTRING(s)->len)
-#endif
-
-#ifndef RARRAY_PTR
-#define RARRAY_PTR(a) RARRAY(a)->ptr
-#endif
-
-#ifndef RARRAY_LEN
-#define RARRAY_LEN(a) RARRAY(a)->len
-#endif
-
 #ifndef __builtin_expect
 # if defined(__GNUC__)
 #  define __builtin_expect(expr, c) __builtin_expect((long)(expr), (long)(c))

data/ext/nokogiri/xml_attr.c

@@ -20,7 +20,7 @@ static VALUE set_value(VALUE self, VALUE content)
     xmlNode *tmp;
 
     /* Encode our content */
-    buffer = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValuePtr(content));
+    buffer = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValueCStr(content));
 
     attr->children = xmlStringGetNodeList(attr->doc, buffer);
     attr->last = NULL;
@@ -61,7 +61,7 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   node = xmlNewDocProp(
       xml_doc,
-      (const xmlChar *)StringValuePtr(name),
+      (const xmlChar *)StringValueCStr(name),
       NULL
   );
 

data/ext/nokogiri/xml_comment.c

@@ -34,7 +34,7 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   node = xmlNewDocComment(
       xml_doc,
-      (const xmlChar *)StringValuePtr(content)
+      (const xmlChar *)StringValueCStr(content)
   );
 
   rb_node = Nokogiri_wrap_xml_node(klass, node);

data/ext/nokogiri/xml_document.c

@@ -181,7 +181,7 @@ static VALUE set_encoding(VALUE self, VALUE encoding)
   if (doc->encoding)
       free((char *) doc->encoding); /* this may produce a gcc cast warning */
 
-  doc->encoding = xmlStrdup((xmlChar *)StringValuePtr(encoding));
+  doc->encoding = xmlStrdup((xmlChar *)StringValueCStr(encoding));
 
   return encoding;
 }
@@ -228,8 +228,8 @@ static VALUE read_io( VALUE klass,
                       VALUE encoding,
                       VALUE options )
 {
-  const char * c_url    = NIL_P(url)      ? NULL : StringValuePtr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
+  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
   VALUE error_list      = rb_ary_new();
   VALUE document;
   xmlDocPtr doc;
@@ -279,8 +279,8 @@ static VALUE read_memory( VALUE klass,
                           VALUE options )
 {
   const char * c_buffer = StringValuePtr(string);
-  const char * c_url    = NIL_P(url)      ? NULL : StringValuePtr(url);
-  const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
+  const char * c_url    = NIL_P(url)      ? NULL : StringValueCStr(url);
+  const char * c_enc    = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
   int len               = (int)RSTRING_LEN(string);
   VALUE error_list      = rb_ary_new();
   VALUE document;
@@ -322,22 +322,20 @@ static VALUE duplicate_document(int argc, VALUE *argv, VALUE self)
   xmlDocPtr doc, dup;
   VALUE copy;
   VALUE level;
-  VALUE error_list      = rb_ary_new();
+  VALUE error_list;
 
   if(rb_scan_args(argc, argv, "01", &level) == 0)
     level = INT2NUM((long)1);
 
   Data_Get_Struct(self, xmlDoc, doc);
 
-  xmlResetLastError();
-  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
   dup = xmlCopyDoc(doc, (int)NUM2INT(level));
-  xmlSetStructuredErrorFunc(NULL, NULL);
 
   if(dup == NULL) return Qnil;
 
   dup->type = doc->type;
   copy = Nokogiri_wrap_xml_document(rb_obj_class(self), dup);
+  error_list = rb_iv_get(self, "@errors");
   rb_iv_set(copy, "@errors", error_list);
   return copy ;
 }
@@ -357,7 +355,7 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   version = rb_ary_entry(rest, (long)0);
   if (NIL_P(version)) version = rb_str_new2("1.0");
 
-  doc = xmlNewDoc((xmlChar *)StringValuePtr(version));
+  doc = xmlNewDoc((xmlChar *)StringValueCStr(version));
   rb_doc = Nokogiri_wrap_xml_document(klass, doc);
   rb_obj_call_init(rb_doc, argc, argv);
   return rb_doc ;
@@ -385,13 +383,13 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
  *         </bicycle>
  *       </root>
  *       EOXML
- *    
+ *
  *    doc.xpath("//tire").to_s # => ""
  *    doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => "<part:tire>Michelin Model XGV</part:tire>"
  *    doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => "<part:tire>I'm a bicycle tire!</part:tire>"
- *    
+ *
  *    doc.remove_namespaces!
- *    
+ *
  *    doc.xpath("//tire").to_s # => "<tire>Michelin Model XGV</tire><tire>I'm a bicycle tire!</tire>"
  *    doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => ""
  *    doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => ""
@@ -438,11 +436,11 @@ static VALUE create_entity(int argc, VALUE *argv, VALUE self)
   xmlResetLastError();
   ptr = xmlAddDocEntity(
       doc,
-      (xmlChar *)(NIL_P(name)        ? NULL                        : StringValuePtr(name)),
+      (xmlChar *)(NIL_P(name)        ? NULL                        : StringValueCStr(name)),
       (int)      (NIL_P(type)        ? XML_INTERNAL_GENERAL_ENTITY : NUM2INT(type)),
-      (xmlChar *)(NIL_P(external_id) ? NULL                        : StringValuePtr(external_id)),
-      (xmlChar *)(NIL_P(system_id)   ? NULL                        : StringValuePtr(system_id)),
-      (xmlChar *)(NIL_P(content)     ? NULL                        : StringValuePtr(content))
+      (xmlChar *)(NIL_P(external_id) ? NULL                        : StringValueCStr(external_id)),
+      (xmlChar *)(NIL_P(system_id)   ? NULL                        : StringValueCStr(system_id)),
+      (xmlChar *)(NIL_P(content)     ? NULL                        : StringValueCStr(content))
     );
 
   if(NULL == ptr) {
@@ -486,9 +484,9 @@ static int block_caller(void * ctx, xmlNodePtr _node, xmlNodePtr _parent)
  *  doc.canonicalize { |obj, parent| ... }
  *
  * Canonicalize a document and return the results.  Takes an optional block
- * that takes two parameters: the +obj+ and that node's +parent+.  
+ * that takes two parameters: the +obj+ and that node's +parent+.
  * The  +obj+ will be either a Nokogiri::XML::Node, or a Nokogiri::XML::Namespace
- * The block must return a non-nil, non-false value if the +obj+ passed in 
+ * The block must return a non-nil, non-false value if the +obj+ passed in
  * should be included in the canonicalized document.
  */
 static VALUE canonicalize(int argc, VALUE* argv, VALUE self)
@@ -533,14 +531,14 @@ static VALUE canonicalize(int argc, VALUE* argv, VALUE self)
     ns = calloc((size_t)ns_len+1, sizeof(xmlChar *));
     for (i = 0 ; i < ns_len ; i++) {
       VALUE entry = rb_ary_entry(incl_ns, i);
-      const char * ptr = StringValuePtr(entry);
+      const char * ptr = StringValueCStr(entry);
       ns[i] = (xmlChar*) ptr;
     }
   }
 
 
-  xmlC14NExecute(doc, cb, ctx, 
-    (int)      (NIL_P(mode)        ? 0 : NUM2INT(mode)), 
+  xmlC14NExecute(doc, cb, ctx,
+    (int)      (NIL_P(mode)        ? 0 : NUM2INT(mode)),
     ns,
     (int)      RTEST(with_comments),
     buf);

data/ext/nokogiri/xml_encoding_handler.c

@@ -9,7 +9,7 @@ static VALUE get(VALUE klass, VALUE key)
 {
   xmlCharEncodingHandlerPtr handler;
 
-  handler = xmlFindCharEncodingHandler(StringValuePtr(key));
+  handler = xmlFindCharEncodingHandler(StringValueCStr(key));
   if(handler)
     return Data_Wrap_Struct(klass, NULL, NULL, handler);
 
@@ -23,7 +23,7 @@ static VALUE get(VALUE klass, VALUE key)
  */
 static VALUE delete(VALUE klass, VALUE name)
 {
-  if(xmlDelEncodingAlias(StringValuePtr(name))) return Qnil;
+  if(xmlDelEncodingAlias(StringValueCStr(name))) return Qnil;
 
   return Qtrue;
 }
@@ -35,7 +35,7 @@ static VALUE delete(VALUE klass, VALUE name)
  */
 static VALUE alias(VALUE klass, VALUE from, VALUE to)
 {
-  xmlAddEncodingAlias(StringValuePtr(from), StringValuePtr(to));
+  xmlAddEncodingAlias(StringValueCStr(from), StringValueCStr(to));
 
   return to;
 }

data/ext/nokogiri/xml_entity_reference.c

@@ -21,7 +21,7 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   node = xmlNewReference(
       xml_doc,
-      (const xmlChar *)StringValuePtr(name)
+      (const xmlChar *)StringValueCStr(name)
   );
 
   nokogiri_root_node(node);

data/ext/nokogiri/xml_namespace.c

@@ -2,6 +2,34 @@
 
 VALUE cNokogiriXmlNamespace ;
 
+static void dealloc_namespace(xmlNsPtr ns)
+{
+  /*
+   *
+   * this deallocator is only used for namespace nodes that are part of an xpath
+   * node set.
+   *
+   * see Nokogiri_wrap_xml_namespace() for more details.
+   *
+   */
+  NOKOGIRI_DEBUG_START(ns) ;
+  if (ns->href) {
+    xmlFree((xmlChar *)ns->href);
+  }
+  if (ns->prefix) {
+    xmlFree((xmlChar *)ns->prefix);
+  }
+  xmlFree(ns);
+  NOKOGIRI_DEBUG_END(ns) ;
+}
+
+
+int Nokogiri_namespace_eh(xmlNodePtr node)
+{
+  return (node->type == XML_NAMESPACE_DECL);
+}
+
+
 /*
  * call-seq:
  *  prefix
@@ -34,37 +62,48 @@ static VALUE href(VALUE self)
   return NOKOGIRI_STR_NEW2(ns->href);
 }
 
+static int part_of_an_xpath_node_set_eh(xmlNsPtr node)
+{
+  return (node->next && ! Nokogiri_namespace_eh(node->next));
+}
+
 VALUE Nokogiri_wrap_xml_namespace(xmlDocPtr doc, xmlNsPtr node)
 {
   VALUE ns, document, node_cache;
 
-  assert(doc->_private);
+  assert(doc->type == XML_DOCUMENT_NODE || doc->type == XML_HTML_DOCUMENT_NODE);
 
-  if(node->_private)
-    return (VALUE)node->_private;
+  if (node->_private) return (VALUE)node->_private;
 
-  ns = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, node);
+  if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
 
-  document = DOC_RUBY_OBJECT(doc);
-
-  node_cache = rb_iv_get(document, "@node_cache");
-  rb_ary_push(node_cache, ns);
+  ns = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, node);
 
-  rb_iv_set(ns, "@document", DOC_RUBY_OBJECT(doc));
+  if (DOC_RUBY_OBJECT_TEST(doc)) {
+    document = DOC_RUBY_OBJECT(doc);
+
+    if (part_of_an_xpath_node_set_eh(node)) {
+      /*
+       *  this is a duplicate returned as part of an xpath query node set, and so
+       *  we need to make sure we manage this memory.
+       *
+       *  see comments in xml_node_set.c for more details.
+       */
+      ns = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, dealloc_namespace, node);
+    } else {
+      ns = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, node);
+      node_cache = rb_iv_get(document, "@node_cache");
+      rb_ary_push(node_cache, ns);
+    }
+
+    rb_iv_set(ns, "@document", document);
+  }
 
   node->_private = (void *)ns;
 
   return ns;
 }
 
-VALUE Nokogiri_wrap_xml_namespace2(VALUE document, xmlNsPtr node)
-{
-  xmlDocPtr doc;
-  Data_Get_Struct(document, xmlDoc, doc) ;
-  return Nokogiri_wrap_xml_namespace(doc, node);
-}
-
-
 void init_xml_namespace()
 {
   VALUE nokogiri  = rb_define_module("Nokogiri");