nokogiri 1.12.5 → 1.13.10

data/ext/nokogiri/nokogiri.h

@@ -170,6 +170,9 @@ int noko_io_read(void *ctx, char *buffer, int len);
 int noko_io_write(void *ctx, char *buffer, int len);
 int noko_io_close(void *ctx);
 
+#define Noko_Node_Get_Struct(obj,type,sval) ((sval) = (type*)DATA_PTR(obj))
+#define Noko_Namespace_Get_Struct(obj,type,sval) ((sval) = (type*)DATA_PTR(obj))
+
 VALUE noko_xml_node_wrap(VALUE klass, xmlNodePtr node) ;
 VALUE noko_xml_node_wrap_node_set_result(xmlNodePtr node, VALUE node_set) ;
 VALUE noko_xml_node_attrs(xmlNodePtr node) ;
@@ -200,6 +203,12 @@ NOKOPUBFUN VALUE Nokogiri_wrap_xml_document(VALUE klass,
 #define DISCARD_CONST_QUAL(t, v) ((t)(uintptr_t)(v))
 #define DISCARD_CONST_QUAL_XMLCHAR(v) DISCARD_CONST_QUAL(xmlChar *, v)
 
+#if HAVE_RB_CATEGORY_WARNING
+#  define NOKO_WARN_DEPRECATION(message) rb_category_warning(RB_WARN_CATEGORY_DEPRECATED, message)
+#else
+#  define NOKO_WARN_DEPRECATION(message) rb_warning(message)
+#endif
+
 void Nokogiri_structured_error_func_save(libxmlStructuredErrorHandlerState *handler_state);
 void Nokogiri_structured_error_func_save_and_set(libxmlStructuredErrorHandlerState *handler_state, void *user_data,
     xmlStructuredErrorFunc handler);

data/ext/nokogiri/xml_attr.c

@@ -16,7 +16,7 @@ set_value(VALUE self, VALUE content)
   xmlChar *value;
   xmlNode *cur;
 
-  Data_Get_Struct(self, xmlAttr, attr);
+  Noko_Node_Get_Struct(self, xmlAttr, attr);
 
   if (attr->children) {
     xmlFreeNodeList(attr->children);
@@ -68,7 +68,7 @@ new (int argc, VALUE *argv, VALUE klass)
     rb_raise(rb_eArgError, "parameter must be a Nokogiri::XML::Document");
   }
 
-  Data_Get_Struct(document, xmlDoc, xml_doc);
+  Noko_Node_Get_Struct(document, xmlDoc, xml_doc);
 
   node = xmlNewDocProp(
            xml_doc,

data/ext/nokogiri/xml_attribute_decl.c

@@ -12,7 +12,7 @@ static VALUE
 attribute_type(VALUE self)
 {
   xmlAttributePtr node;
-  Data_Get_Struct(self, xmlAttribute, node);
+  Noko_Node_Get_Struct(self, xmlAttribute, node);
   return INT2NUM((long)node->atype);
 }
 
@@ -26,7 +26,7 @@ static VALUE
 default_value(VALUE self)
 {
   xmlAttributePtr node;
-  Data_Get_Struct(self, xmlAttribute, node);
+  Noko_Node_Get_Struct(self, xmlAttribute, node);
 
   if (node->defaultValue) { return NOKOGIRI_STR_NEW2(node->defaultValue); }
   return Qnil;
@@ -45,7 +45,7 @@ enumeration(VALUE self)
   xmlEnumerationPtr enm;
   VALUE list;
 
-  Data_Get_Struct(self, xmlAttribute, node);
+  Noko_Node_Get_Struct(self, xmlAttribute, node);
 
   list = rb_ary_new();
   enm = node->tree;

data/ext/nokogiri/xml_cdata.c

@@ -25,7 +25,7 @@ new (int argc, VALUE *argv, VALUE klass)
 
   rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
 
-  Data_Get_Struct(doc, xmlDoc, xml_doc);
+  Noko_Node_Get_Struct(doc, xmlDoc, xml_doc);
 
   if (!NIL_P(content)) {
     content_str = (xmlChar *)StringValuePtr(content);

data/ext/nokogiri/xml_document.c

@@ -104,7 +104,11 @@ recursively_remove_namespaces_from_node(xmlNodePtr node)
        (node->type == XML_XINCLUDE_START) ||
        (node->type == XML_XINCLUDE_END)) &&
       node->nsDef) {
-    xmlFreeNsList(node->nsDef);
+    xmlNsPtr curr = node->nsDef;
+    while (curr) {
+      noko_xml_document_pin_namespace(curr, node->doc);
+      curr = curr->next;
+    }
     node->nsDef = NULL;
   }
 
@@ -161,7 +165,7 @@ rb_xml_document_root_set(VALUE self, VALUE rb_new_root)
                rb_obj_class(rb_new_root));
     }
 
-    Data_Get_Struct(rb_new_root, xmlNode, c_new_root);
+    Noko_Node_Get_Struct(rb_new_root, xmlNode, c_new_root);
 
     /* 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. */
@@ -533,59 +537,59 @@ block_caller(void *ctx, xmlNodePtr c_node, xmlNodePtr c_parent_node)
 static VALUE
 rb_xml_document_canonicalize(int argc, VALUE *argv, VALUE self)
 {
-  VALUE mode;
-  VALUE incl_ns;
-  VALUE with_comments;
-  xmlChar **ns;
-  long ns_len, i;
+  VALUE rb_mode;
+  VALUE rb_namespaces;
+  VALUE rb_comments_p;
+  xmlChar **c_namespaces;
 
-  xmlDocPtr doc;
-  xmlOutputBufferPtr buf;
-  xmlC14NIsVisibleCallback cb = NULL;
-  void *ctx = NULL;
+  xmlDocPtr c_doc;
+  xmlOutputBufferPtr c_obuf;
+  xmlC14NIsVisibleCallback c_callback_wrapper = NULL;
+  void *rb_callback = NULL;
 
   VALUE rb_cStringIO;
-  VALUE io;
+  VALUE rb_io;
 
-  rb_scan_args(argc, argv, "03", &mode, &incl_ns, &with_comments);
+  rb_scan_args(argc, argv, "03", &rb_mode, &rb_namespaces, &rb_comments_p);
+  if (!NIL_P(rb_mode)) { Check_Type(rb_mode, T_FIXNUM); }
+  if (!NIL_P(rb_namespaces)) { Check_Type(rb_namespaces, T_ARRAY); }
 
-  Data_Get_Struct(self, xmlDoc, doc);
+  Data_Get_Struct(self, xmlDoc, c_doc);
 
   rb_cStringIO = rb_const_get_at(rb_cObject, rb_intern("StringIO"));
-  io           = rb_class_new_instance(0, 0, rb_cStringIO);
-  buf          = xmlAllocOutputBuffer(NULL);
+  rb_io = rb_class_new_instance(0, 0, rb_cStringIO);
+  c_obuf = xmlAllocOutputBuffer(NULL);
 
-  buf->writecallback = (xmlOutputWriteCallback)noko_io_write;
-  buf->closecallback = (xmlOutputCloseCallback)noko_io_close;
-  buf->context       = (void *)io;
+  c_obuf->writecallback = (xmlOutputWriteCallback)noko_io_write;
+  c_obuf->closecallback = (xmlOutputCloseCallback)noko_io_close;
+  c_obuf->context = (void *)rb_io;
 
   if (rb_block_given_p()) {
-    cb = block_caller;
-    ctx = (void *)rb_block_proc();
+    c_callback_wrapper = block_caller;
+    rb_callback = (void *)rb_block_proc();
   }
 
-  if (NIL_P(incl_ns)) {
-    ns = NULL;
+  if (NIL_P(rb_namespaces)) {
+    c_namespaces = NULL;
   } else {
-    Check_Type(incl_ns, T_ARRAY);
-    ns_len = RARRAY_LEN(incl_ns);
-    ns = calloc((size_t)ns_len + 1, sizeof(xmlChar *));
-    for (i = 0 ; i < ns_len ; i++) {
-      VALUE entry = rb_ary_entry(incl_ns, i);
-      ns[i] = (xmlChar *)StringValueCStr(entry);
+    long ns_len = RARRAY_LEN(rb_namespaces);
+    c_namespaces = calloc((size_t)ns_len + 1, sizeof(xmlChar *));
+    for (int j = 0 ; j < ns_len ; j++) {
+      VALUE entry = rb_ary_entry(rb_namespaces, j);
+      c_namespaces[j] = (xmlChar *)StringValueCStr(entry);
     }
   }
 
+  xmlC14NExecute(c_doc, c_callback_wrapper, rb_callback,
+                 (int)(NIL_P(rb_mode) ? 0 : NUM2INT(rb_mode)),
+                 c_namespaces,
+                 (int)RTEST(rb_comments_p),
+                 c_obuf);
 
-  xmlC14NExecute(doc, cb, ctx,
-                 (int)(NIL_P(mode)        ? 0 : NUM2INT(mode)),
-                 ns,
-                 (int)      RTEST(with_comments),
-                 buf);
-
-  xmlOutputBufferClose(buf);
+  free(c_namespaces);
+  xmlOutputBufferClose(c_obuf);
 
-  return rb_funcall(io, rb_intern("string"), 0);
+  return rb_funcall(rb_io, rb_intern("string"), 0);
 }
 
 VALUE

data/ext/nokogiri/xml_document_fragment.c

@@ -28,8 +28,6 @@ new (int argc, VALUE *argv, VALUE klass)
   rb_node = noko_xml_node_wrap(klass, node);
   rb_obj_call_init(rb_node, argc, argv);
 
-  if (rb_block_given_p()) { rb_yield(rb_node); }
-
   return rb_node;
 }
 

data/ext/nokogiri/xml_dtd.c

@@ -44,7 +44,7 @@ entities(VALUE self)
   xmlDtdPtr dtd;
   VALUE hash;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->entities) { return Qnil; }
 
@@ -57,9 +57,9 @@ entities(VALUE self)
 
 /*
  * call-seq:
- *   notations
+ *   notations() → Hash<name(String)⇒Notation>
  *
- * Get a hash of the notations for this DTD.
+ * [Returns] All the notations for this DTD in a Hash of Notation +name+ to Notation.
  */
 static VALUE
 notations(VALUE self)
@@ -67,7 +67,7 @@ notations(VALUE self)
   xmlDtdPtr dtd;
   VALUE hash;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->notations) { return Qnil; }
 
@@ -90,7 +90,7 @@ attributes(VALUE self)
   xmlDtdPtr dtd;
   VALUE hash;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   hash = rb_hash_new();
 
@@ -113,7 +113,7 @@ elements(VALUE self)
   xmlDtdPtr dtd;
   VALUE hash;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->elements) { return Qnil; }
 
@@ -138,8 +138,8 @@ validate(VALUE self, VALUE document)
   xmlValidCtxtPtr ctxt;
   VALUE error_list;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
-  Data_Get_Struct(document, xmlDoc, doc);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(document, xmlDoc, doc);
   error_list = rb_ary_new();
 
   ctxt = xmlNewValidCtxt();
@@ -165,7 +165,7 @@ static VALUE
 system_id(VALUE self)
 {
   xmlDtdPtr dtd;
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->SystemID) { return Qnil; }
 
@@ -182,7 +182,7 @@ static VALUE
 external_id(VALUE self)
 {
   xmlDtdPtr dtd;
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->ExternalID) { return Qnil; }
 

data/ext/nokogiri/xml_element_decl.c

@@ -14,7 +14,7 @@ static VALUE
 element_type(VALUE self)
 {
   xmlElementPtr node;
-  Data_Get_Struct(self, xmlElement, node);
+  Noko_Node_Get_Struct(self, xmlElement, node);
   return INT2NUM((long)node->etype);
 }
 
@@ -28,7 +28,7 @@ static VALUE
 content(VALUE self)
 {
   xmlElementPtr node;
-  Data_Get_Struct(self, xmlElement, node);
+  Noko_Node_Get_Struct(self, xmlElement, node);
 
   if (!node->content) { return Qnil; }
 
@@ -48,7 +48,7 @@ static VALUE
 prefix(VALUE self)
 {
   xmlElementPtr node;
-  Data_Get_Struct(self, xmlElement, node);
+  Noko_Node_Get_Struct(self, xmlElement, node);
 
   if (!node->prefix) { return Qnil; }
 

data/ext/nokogiri/xml_encoding_handler.c

@@ -3,70 +3,82 @@
 VALUE cNokogiriEncodingHandler;
 
 
+static void
+_xml_encoding_handler_dealloc(xmlCharEncodingHandlerPtr c_handler)
+{
+  /* make sure iconv handlers are cleaned up and freed */
+  xmlCharEncCloseFunc(c_handler);
+}
+
+
 /*
  * call-seq: Nokogiri::EncodingHandler.[](name)
  *
  * Get the encoding handler for +name+
  */
 static VALUE
-get(VALUE klass, VALUE key)
+rb_xml_encoding_handler_s_get(VALUE klass, VALUE key)
 {
   xmlCharEncodingHandlerPtr handler;
 
   handler = xmlFindCharEncodingHandler(StringValueCStr(key));
   if (handler) {
-    return Data_Wrap_Struct(klass, NULL, NULL, handler);
+    return Data_Wrap_Struct(klass, NULL, _xml_encoding_handler_dealloc, handler);
   }
 
   return Qnil;
 }
 
+
 /*
  * call-seq: Nokogiri::EncodingHandler.delete(name)
  *
  * Delete the encoding alias named +name+
  */
 static VALUE
-delete (VALUE klass, VALUE name)
+rb_xml_encoding_handler_s_delete(VALUE klass, VALUE name)
 {
   if (xmlDelEncodingAlias(StringValueCStr(name))) { return Qnil; }
 
   return Qtrue;
 }
 
+
 /*
  * call-seq: Nokogiri::EncodingHandler.alias(from, to)
  *
  * Alias encoding handler with name +from+ to name +to+
  */
 static VALUE
-alias(VALUE klass, VALUE from, VALUE to)
+rb_xml_encoding_handler_s_alias(VALUE klass, VALUE from, VALUE to)
 {
   xmlAddEncodingAlias(StringValueCStr(from), StringValueCStr(to));
 
   return to;
 }
 
+
 /*
  * call-seq: Nokogiri::EncodingHandler.clear_aliases!
  *
  * Remove all encoding aliases.
  */
 static VALUE
-clear_aliases(VALUE klass)
+rb_xml_encoding_handler_s_clear_aliases(VALUE klass)
 {
   xmlCleanupEncodingAliases();
 
   return klass;
 }
 
+
 /*
  * call-seq: name
  *
  * Get the name of this EncodingHandler
  */
 static VALUE
-name(VALUE self)
+rb_xml_encoding_handler_name(VALUE self)
 {
   xmlCharEncodingHandlerPtr handler;
 
@@ -75,6 +87,7 @@ name(VALUE self)
   return NOKOGIRI_STR_NEW2(handler->name);
 }
 
+
 void
 noko_init_xml_encoding_handler()
 {
@@ -82,9 +95,10 @@ noko_init_xml_encoding_handler()
 
   rb_undef_alloc_func(cNokogiriEncodingHandler);
 
-  rb_define_singleton_method(cNokogiriEncodingHandler, "[]", get, 1);
-  rb_define_singleton_method(cNokogiriEncodingHandler, "delete", delete, 1);
-  rb_define_singleton_method(cNokogiriEncodingHandler, "alias", alias, 2);
-  rb_define_singleton_method(cNokogiriEncodingHandler, "clear_aliases!", clear_aliases, 0);
-  rb_define_method(cNokogiriEncodingHandler, "name", name, 0);
+  rb_define_singleton_method(cNokogiriEncodingHandler, "[]", rb_xml_encoding_handler_s_get, 1);
+  rb_define_singleton_method(cNokogiriEncodingHandler, "delete", rb_xml_encoding_handler_s_delete, 1);
+  rb_define_singleton_method(cNokogiriEncodingHandler, "alias", rb_xml_encoding_handler_s_alias, 2);
+  rb_define_singleton_method(cNokogiriEncodingHandler, "clear_aliases!", rb_xml_encoding_handler_s_clear_aliases, 0);
+
+  rb_define_method(cNokogiriEncodingHandler, "name", rb_xml_encoding_handler_name, 0);
 }

data/ext/nokogiri/xml_entity_decl.c

@@ -12,7 +12,7 @@ static VALUE
 original_content(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   if (!node->orig) { return Qnil; }
 
@@ -29,7 +29,7 @@ static VALUE
 get_content(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   if (!node->content) { return Qnil; }
 
@@ -46,7 +46,7 @@ static VALUE
 entity_type(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   return INT2NUM((int)node->etype);
 }
@@ -61,7 +61,7 @@ static VALUE
 external_id(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   if (!node->ExternalID) { return Qnil; }
 
@@ -78,7 +78,7 @@ static VALUE
 system_id(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   if (!node->SystemID) { return Qnil; }
 

data/ext/nokogiri/xml_namespace.c

@@ -25,13 +25,15 @@
 VALUE cNokogiriXmlNamespace ;
 
 static void
-dealloc_namespace(xmlNsPtr ns)
+_xml_namespace_dealloc(void *ptr)
 {
   /*
    * this deallocator is only used for namespace nodes that are part of an xpath
    * node set. see noko_xml_namespace_wrap().
    */
+  xmlNsPtr ns = ptr;
   NOKOGIRI_DEBUG_START(ns) ;
+
   if (ns->href) {
     xmlFree(DISCARD_CONST_QUAL_XMLCHAR(ns->href));
   }
@@ -42,6 +44,36 @@ dealloc_namespace(xmlNsPtr ns)
   NOKOGIRI_DEBUG_END(ns) ;
 }
 
+#ifdef HAVE_RB_GC_LOCATION
+static void
+_xml_namespace_update_references(void *ptr)
+{
+  xmlNsPtr ns = ptr;
+  if (ns->_private) {
+    ns->_private = (void *)rb_gc_location((VALUE)ns->_private);
+  }
+}
+#else
+#  define _xml_namespace_update_references 0
+#endif
+
+static const rb_data_type_t nokogiri_xml_namespace_type_with_dealloc = {
+  "Nokogiri/XMLNamespace/WithDealloc",
+  {0, _xml_namespace_dealloc, 0, _xml_namespace_update_references},
+  0, 0,
+#ifdef RUBY_TYPED_FREE_IMMEDIATELY
+  RUBY_TYPED_FREE_IMMEDIATELY,
+#endif
+};
+
+static const rb_data_type_t nokogiri_xml_namespace_type_without_dealloc = {
+  "Nokogiri/XMLNamespace/WithoutDealloc",
+  {0, 0, 0, _xml_namespace_update_references},
+  0, 0,
+#ifdef RUBY_TYPED_FREE_IMMEDIATELY
+  RUBY_TYPED_FREE_IMMEDIATELY,
+#endif
+};
 
 /*
  * call-seq:
@@ -54,7 +86,7 @@ prefix(VALUE self)
 {
   xmlNsPtr ns;
 
-  Data_Get_Struct(self, xmlNs, ns);
+  Noko_Namespace_Get_Struct(self, xmlNs, ns);
   if (!ns->prefix) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(ns->prefix);
@@ -71,7 +103,7 @@ href(VALUE self)
 {
   xmlNsPtr ns;
 
-  Data_Get_Struct(self, xmlNs, ns);
+  Noko_Namespace_Get_Struct(self, xmlNs, ns);
   if (!ns->href) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(ns->href);
@@ -87,14 +119,18 @@ noko_xml_namespace_wrap(xmlNsPtr c_namespace, xmlDocPtr c_document)
   }
 
   if (c_document) {
-    rb_namespace = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, c_namespace);
+    rb_namespace = TypedData_Wrap_Struct(cNokogiriXmlNamespace,
+                                         &nokogiri_xml_namespace_type_without_dealloc,
+                                         c_namespace);
 
     if (DOC_RUBY_OBJECT_TEST(c_document)) {
       rb_iv_set(rb_namespace, "@document", DOC_RUBY_OBJECT(c_document));
       rb_ary_push(DOC_NODE_CACHE(c_document), rb_namespace);
     }
   } else {
-    rb_namespace = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, dealloc_namespace, c_namespace);
+    rb_namespace = TypedData_Wrap_Struct(cNokogiriXmlNamespace,
+                                         &nokogiri_xml_namespace_type_with_dealloc,
+                                         c_namespace);
   }
 
   c_namespace->_private = (void *)rb_namespace;