nokogiri 1.4.0-java → 1.4.1-java

data/ext/nokogiri/xml_namespace.h

@@ -8,5 +8,6 @@ void init_xml_namespace();
 extern VALUE cNokogiriXmlNamespace ;
 
 VALUE Nokogiri_wrap_xml_namespace(xmlDocPtr doc, xmlNsPtr node) ;
+VALUE Nokogiri_wrap_xml_namespace2(VALUE document, xmlNsPtr node) ;
 
 #endif

data/ext/nokogiri/xml_node.c

@@ -56,6 +56,17 @@ static void relink_namespace(xmlNodePtr reparented)
   }
 }
 
+/* :nodoc: */
+static xmlNodePtr xmlReplaceNodeWrapper(xmlNodePtr old, xmlNodePtr cur)
+{
+  xmlNodePtr retval ;
+  retval = xmlReplaceNode(old, cur) ;
+  if (retval == old) {
+    return cur ; // return semantics for reparent_node_with
+  }
+  return retval ;
+}
+
 /* :nodoc: */
 static VALUE reparent_node_with(VALUE node_obj, VALUE other_obj, node_other_func func)
 {
@@ -71,17 +82,6 @@ static VALUE reparent_node_with(VALUE node_obj, VALUE other_obj, node_other_func
   if(XML_DOCUMENT_NODE == node->type || XML_HTML_DOCUMENT_NODE == node->type)
     rb_raise(rb_eArgError, "cannot reparent a document node");
 
-  // 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->type == XML_TEXT_NODE) {
     NOKOGIRI_ROOT_NODE(node);
     node = xmlDocCopyNode(node, other->doc, 1);
@@ -370,17 +370,50 @@ static VALUE previous_sibling(VALUE self)
   return Nokogiri_wrap_xml_node(Qnil, sibling);
 }
 
-/* :nodoc: */
-static VALUE replace(VALUE self, VALUE _new_node)
+/*
+ * call-seq:
+ *  next_element
+ *
+ * Returns the next Nokogiri::XML::Element type sibling node.
+ */
+static VALUE next_element(VALUE self)
 {
-  xmlNodePtr node, new_node;
+  xmlNodePtr node, sibling;
   Data_Get_Struct(self, xmlNode, node);
-  Data_Get_Struct(_new_node, xmlNode, new_node);
 
-  xmlReplaceNode(node, new_node);
+  sibling = node->next;
+  if(!sibling) return Qnil;
 
-  // Appropriately link in namespaces
-  relink_namespace(new_node);
+  while(sibling && sibling->type != XML_ELEMENT_NODE)
+    sibling = sibling->next;
+
+  return sibling ? Nokogiri_wrap_xml_node(Qnil, sibling) : Qnil ;
+}
+
+/*
+ * call-seq:
+ *  previous_element
+ *
+ * Returns the previous Nokogiri::XML::Element type sibling node.
+ */
+static VALUE previous_element(VALUE self)
+{
+  xmlNodePtr node, sibling;
+  Data_Get_Struct(self, xmlNode, node);
+
+  sibling = node->prev;
+  if(!sibling) return Qnil;
+
+  while(sibling && sibling->type != XML_ELEMENT_NODE)
+    sibling = sibling->prev;
+
+  return sibling ? Nokogiri_wrap_xml_node(Qnil, sibling) : Qnil ;
+}
+
+/* :nodoc: */
+static VALUE replace(VALUE self, VALUE _new_node)
+{
+  reparent_node_with(_new_node, self, xmlReplaceNodeWrapper) ;
   return self ;
 }
 
@@ -667,12 +700,7 @@ static VALUE get_content(VALUE self)
   return Qnil;
 }
 
-/*
- * call-seq:
- *  add_child(node)
- *
- * Add +node+ as a child of this node. Returns the new child node.
- */
+/* :nodoc: */
 static VALUE add_child(VALUE self, VALUE child)
 {
   return reparent_node_with(child, self, xmlAddChild);
@@ -742,23 +770,13 @@ static VALUE path(VALUE self)
   return rval ;
 }
 
-/*
- *  call-seq:
- *    add_next_sibling(node)
- *
- *  Insert +node+ after this node (as a sibling).
- */
+/* :nodoc: */
 static VALUE add_next_sibling(VALUE self, VALUE rb_node)
 {
   return reparent_node_with(rb_node, self, xmlAddNextSibling) ;
 }
 
-/*
- * call-seq:
- *  add_previous_sibling(node)
- *
- * Insert +node+ before this node (as a sibling).
- */
+/* :nodoc: */
 static VALUE add_previous_sibling(VALUE self, VALUE rb_node)
 {
   return reparent_node_with(rb_node, self, xmlAddPrevSibling) ;
@@ -1017,12 +1035,13 @@ void init_xml_node()
   rb_define_method(klass, "node_name", get_name, 0);
   rb_define_method(klass, "document", document, 0);
   rb_define_method(klass, "node_name=", set_name, 1);
-  rb_define_method(klass, "add_child", add_child, 1);
   rb_define_method(klass, "parent", get_parent, 0);
   rb_define_method(klass, "child", child, 0);
   rb_define_method(klass, "children", children, 0);
   rb_define_method(klass, "next_sibling", next_sibling, 0);
   rb_define_method(klass, "previous_sibling", previous_sibling, 0);
+  rb_define_method(klass, "next_element", next_element, 0);
+  rb_define_method(klass, "previous_element", previous_element, 0);
   rb_define_method(klass, "node_type", node_type, 0);
   rb_define_method(klass, "content", get_content, 0);
   rb_define_method(klass, "path", path, 0);
@@ -1035,8 +1054,6 @@ void init_xml_node()
   rb_define_method(klass, "attribute_with_ns", attribute_with_ns, 2);
   rb_define_method(klass, "namespace", namespace, 0);
   rb_define_method(klass, "namespace_definitions", namespace_definitions, 0);
-  rb_define_method(klass, "add_previous_sibling", add_previous_sibling, 1);
-  rb_define_method(klass, "add_next_sibling", add_next_sibling, 1);
   rb_define_method(klass, "encode_special_chars", encode_special_chars, 1);
   rb_define_method(klass, "dup", duplicate_node, -1);
   rb_define_method(klass, "unlink", unlink_node, 0);
@@ -1047,14 +1064,17 @@ void init_xml_node()
   rb_define_method(klass, "pointer_id", pointer_id, 0);
   rb_define_method(klass, "line", line, 0);
 
+  rb_define_private_method(klass, "add_child_node", add_child, 1);
+  rb_define_private_method(klass, "add_previous_sibling_node", add_previous_sibling, 1);
+  rb_define_private_method(klass, "add_next_sibling_node", add_next_sibling, 1);
+  rb_define_private_method(klass, "replace_node", replace, 1);
   rb_define_private_method(klass, "dump_html", dump_html, 0);
   rb_define_private_method(klass, "native_write_to", native_write_to, 4);
-  rb_define_private_method(klass, "replace_with_node", replace, 1);
   rb_define_private_method(klass, "native_content=", set_content, 1);
   rb_define_private_method(klass, "get", get, 1);
   rb_define_private_method(klass, "set_namespace", set_namespace, 1);
   rb_define_private_method(klass, "compare", compare, 1);
 
-  decorate = rb_intern("decorate");
+  decorate      = rb_intern("decorate");
   decorate_bang = rb_intern("decorate!");
 }

data/ext/nokogiri/xml_node_set.c

@@ -189,6 +189,8 @@ static VALUE index_at(VALUE self, long offset)
   if(offset >= node_set->nodeNr || abs(offset) > node_set->nodeNr) return Qnil;
   if(offset < 0) offset = offset + node_set->nodeNr;
 
+  if (XML_NAMESPACE_DECL == node_set->nodeTab[offset]->type)
+    return Nokogiri_wrap_xml_namespace2(rb_iv_get(self, "@document"), (xmlNsPtr)(node_set->nodeTab[offset]));
   return Nokogiri_wrap_xml_node(Qnil, node_set->nodeTab[offset]);
 }
 
@@ -280,15 +282,19 @@ static VALUE to_array(VALUE self, VALUE rb_node)
   VALUE *elts = calloc((size_t)set->nodeNr, sizeof(VALUE *));
   int i;
   for(i = 0; i < set->nodeNr; i++) {
-    xmlNodePtr node = set->nodeTab[i];
-
-    if(node->_private) {
-      if(node->type == XML_DOCUMENT_NODE || node->type == XML_HTML_DOCUMENT_NODE)
-        elts[i] = DOC_RUBY_OBJECT(node->doc);
-      else
-        elts[i] = (VALUE)node->_private;
+    if (XML_NAMESPACE_DECL == set->nodeTab[i]->type) {
+      elts[i] = Nokogiri_wrap_xml_namespace2(rb_iv_get(self, "@document"), (xmlNsPtr)(set->nodeTab[i]));
     } else {
-      elts[i] = Nokogiri_wrap_xml_node(Qnil, node);
+      xmlNodePtr node = set->nodeTab[i];
+
+      if(node->_private) {
+        if(node->type == XML_DOCUMENT_NODE || node->type == XML_HTML_DOCUMENT_NODE)
+          elts[i] = DOC_RUBY_OBJECT(node->doc);
+        else
+          elts[i] = (VALUE)node->_private;
+      } else {
+        elts[i] = Nokogiri_wrap_xml_node(Qnil, node);
+      }
     }
   }
 
@@ -313,12 +319,14 @@ static VALUE unlink_nodeset(VALUE self)
   Data_Get_Struct(self, xmlNodeSet, node_set);
   nodeNr = node_set->nodeNr ;
   for (j = 0 ; j < nodeNr ; j++) {
-    VALUE node ;
-    xmlNodePtr node_ptr;
-    node = Nokogiri_wrap_xml_node(Qnil, node_set->nodeTab[j]);
-    rb_funcall(node, rb_intern("unlink"), 0); /* modifies the C struct out from under the object */
-    Data_Get_Struct(node, xmlNode, node_ptr);
-    node_set->nodeTab[j] = node_ptr ;
+    if (XML_NAMESPACE_DECL != node_set->nodeTab[j]->type) {
+      VALUE node ;
+      xmlNodePtr node_ptr;
+      node = Nokogiri_wrap_xml_node(Qnil, node_set->nodeTab[j]);
+      rb_funcall(node, rb_intern("unlink"), 0); /* modifies the C struct out from under the object */
+      Data_Get_Struct(node, xmlNode, node_ptr);
+      node_set->nodeTab[j] = node_ptr ;
+    }
   }
   return self ;
 }

data/ext/nokogiri/xml_sax_parser.c

@@ -8,9 +8,6 @@ static ID id_cdata_block, id_cAttribute;
 #define STRING_OR_NULL(str) \
    (RTEST(str) ? StringValuePtr(str) : NULL)
 
-#define RBSTR_OR_QNIL(_str) \
-  (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
-
 static void start_document(void * ctx)
 {
   VALUE self = NOKOGIRI_SAX_SELF(ctx);

data/ext/nokogiri/xml_sax_parser_context.c

@@ -96,6 +96,8 @@ static VALUE parse_with(VALUE self, VALUE sax_handler)
   if(NULL != ctxt->myDoc) xmlFreeDoc(ctxt->myDoc);
 
   NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
+
+  return Qnil ;
 }
 
 /*

data/ext/nokogiri/xml_sax_push_parser.c

@@ -35,8 +35,10 @@ static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
   }
 
   if(xmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
-    xmlErrorPtr e = xmlCtxtGetLastError(ctx);
-    Nokogiri_error_raise(NULL, e);
+    if (!(ctx->options & XML_PARSE_RECOVER)) {
+      xmlErrorPtr e = xmlCtxtGetLastError(ctx);
+      Nokogiri_error_raise(NULL, e);
+    }
   }
 
   return self;
@@ -53,7 +55,7 @@ static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
   xmlSAXHandlerPtr sax;
 
   Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
-  
+
   const char * filename = NULL;
 
   if(_filename != Qnil) filename = StringValuePtr(_filename);
@@ -75,6 +77,25 @@ static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
   return self;
 }
 
+static VALUE get_options(VALUE self)
+{
+  xmlParserCtxtPtr ctx;
+  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  return INT2NUM(ctx->options);
+}
+
+static VALUE set_options(VALUE self, VALUE options)
+{
+  xmlParserCtxtPtr ctx;
+  Data_Get_Struct(self, xmlParserCtxt, ctx);
+
+  if (xmlCtxtUseOptions(ctx, (int)NUM2INT(options)) != 0)
+    rb_raise(rb_eRuntimeError, "Cannot set XML parser context options");
+
+  return Qnil;
+}
+
 VALUE cNokogiriXmlSaxPushParser ;
 void init_xml_sax_push_parser()
 {
@@ -88,4 +109,6 @@ void init_xml_sax_push_parser()
   rb_define_alloc_func(klass, allocate);
   rb_define_private_method(klass, "initialize_native", initialize_native, 2);
   rb_define_private_method(klass, "native_write", native_write, 2);
+  rb_define_method(klass, "options", get_options, 0);
+  rb_define_method(klass, "options=", set_options, 1);
 }

data/ext/nokogiri/xml_syntax_error.c

@@ -1,214 +1,5 @@
 #include <xml_syntax_error.h>
 
-static void dealloc(xmlErrorPtr ptr)
-{
-  NOKOGIRI_DEBUG_START(ptr);
-  xmlResetError(ptr);
-  xmlFree(ptr);
-  NOKOGIRI_DEBUG_END(ptr);
-}
-
-static VALUE allocate(VALUE klass)
-{
-  xmlErrorPtr error = xmlMalloc(sizeof(xmlError));
-  memset(error, 0, sizeof(xmlError));
-  return Data_Wrap_Struct(klass, NULL, dealloc, error);
-}
-
-/*
- * call-seq:
- *  column
- *
- * Column number or 0 if not available
- */
-static VALUE column(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->int2);
-}
-
-/*
- * call-seq:
- *  int1
- *
- * Extra number information
- */
-static VALUE int1(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->int1);
-}
-
-/*
- * call-seq:
- *  str3
- *
- * Extra string information
- */
-static VALUE str3(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  if(error->str3) return NOKOGIRI_STR_NEW2(error->str3);
-  return Qnil;
-}
-
-/*
- * call-seq:
- *  str2
- *
- * Extra string information
- */
-static VALUE str2(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  if(error->str2) return NOKOGIRI_STR_NEW2(error->str2);
-  return Qnil;
-}
-
-/*
- * call-seq:
- *  str1
- *
- * Extra string information
- */
-static VALUE str1(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  if(error->str1) return NOKOGIRI_STR_NEW2(error->str1);
-  return Qnil;
-}
-
-/*
- * call-seq:
- *  line
- *
- * Get the line number of the error
- */
-static VALUE line(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->line);
-}
-
-/*
- * call-seq:
- *  file
- *
- * Get the filename for the error
- */
-static VALUE file(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  if(error->file) return NOKOGIRI_STR_NEW2(error->file);
-
-  return Qnil;
-}
-
-/*
- * call-seq:
- *  level
- *
- * Get the error level
- */
-static VALUE level(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM((short)error->level);
-}
-
-/*
- * call-seq:
- *  code
- *
- * Get the error code
- */
-static VALUE code(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->code);
-}
-
-/*
- * call-seq:
- *  domain
- *
- * Get the part of the library that raised this exception
- */
-static VALUE domain(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  return INT2NUM(error->domain);
-}
-
-/*
- * call-seq:
- *  message
- *
- * Get the human readable message.
- */
-static VALUE message(VALUE self)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-  if(error->message) return NOKOGIRI_STR_NEW2(error->message);
-  return Qnil;
-}
-
-/*
- * call-seq:
- *  message=
- *
- * Set the human readable message.
- */
-static VALUE set_message(VALUE self, VALUE _message)
-{
-  xmlErrorPtr error;
-  Data_Get_Struct(self, xmlError, error);
-
-  if(error->message) {
-    xmlFree(error->message);
-    error->message = 0;
-  }
-
-  if(RTEST(_message)) {
-    error->message = xmlMalloc(RSTRING_LEN(_message) + 1);
-    memset(error->message, 0, RSTRING_LEN(_message) + 1);
-    memcpy(error->message, StringValuePtr(_message), RSTRING_LEN(_message));
-  }
-
-  return message;
-}
-
-/*
- * call-seq:
- *  initialize_copy(old_copy)
- *
- * Initialize a copy of the +old_copy+
- */
-static VALUE initialize_copy(VALUE self, VALUE _old_copy)
-{
-  if(!rb_obj_is_kind_of(_old_copy, cNokogiriXmlSyntaxError))
-    rb_raise(rb_eArgError, "node must be a Nokogiri::XML::SyntaxError");
-
-  xmlErrorPtr error, old_error;
-  Data_Get_Struct(self, xmlError, error);
-  Data_Get_Struct(_old_copy, xmlError, old_error);
-
-  xmlCopyError(old_error, error);
-
-  return message;
-}
-
 void Nokogiri_error_array_pusher(void * ctx, xmlErrorPtr error)
 {
   VALUE list = (VALUE)ctx;
@@ -224,10 +15,25 @@ VALUE Nokogiri_wrap_xml_syntax_error(VALUE klass, xmlErrorPtr error)
 {
   if(!klass) klass = cNokogiriXmlSyntaxError;
 
-  xmlErrorPtr ptr = calloc(1, sizeof(xmlError));
-  xmlCopyError(error, ptr);
+  VALUE msg = error->message ? NOKOGIRI_STR_NEW2(error->message) : Qnil;
+
+  VALUE e = rb_class_new_instance(
+      1,
+      &msg,
+      klass
+  );
+  rb_iv_set(e, "@domain", INT2NUM(error->domain));
+  rb_iv_set(e, "@code", INT2NUM(error->code));
+  rb_iv_set(e, "@level", INT2NUM((short)error->level));
+  rb_iv_set(e, "@file", RBSTR_OR_QNIL(error->file));
+  rb_iv_set(e, "@line", INT2NUM(error->line));
+  rb_iv_set(e, "@str1", RBSTR_OR_QNIL(error->str1));
+  rb_iv_set(e, "@str2", RBSTR_OR_QNIL(error->str2));
+  rb_iv_set(e, "@str3", RBSTR_OR_QNIL(error->str3));
+  rb_iv_set(e, "@int1", INT2NUM(error->int1));
+  rb_iv_set(e, "@column", INT2NUM(error->int2));
 
-  return Data_Wrap_Struct(klass, NULL, dealloc, ptr);
+  return e;
 }
 
 VALUE cNokogiriXmlSyntaxError;
@@ -243,19 +49,4 @@ void init_xml_syntax_error()
   VALUE klass = rb_define_class_under(xml, "SyntaxError", syntax_error_mommy);
   cNokogiriXmlSyntaxError = klass;
 
-  rb_define_alloc_func(klass, allocate);
-
-  rb_define_method(klass, "message", message, 0);
-  rb_define_method(klass, "message=", set_message, 1);
-  rb_define_method(klass, "initialize_copy", initialize_copy, 1);
-  rb_define_method(klass, "domain", domain, 0);
-  rb_define_method(klass, "code", code, 0);
-  rb_define_method(klass, "level", level, 0);
-  rb_define_method(klass, "file", file, 0);
-  rb_define_method(klass, "line", line, 0);
-  rb_define_method(klass, "str1", str1, 0);
-  rb_define_method(klass, "str2", str2, 0);
-  rb_define_method(klass, "str3", str3, 0);
-  rb_define_method(klass, "int1", int1, 0);
-  rb_define_method(klass, "column", column, 0);
 }