nokogiri 1.4.1 → 1.4.2

data/ext/nokogiri/xml_node_set.c

@@ -1,6 +1,8 @@
 #include <xml_node_set.h>
 #include <libxml/xpathInternals.h>
 
+static ID decorate ;
+
 /*
  * call-seq:
  *  dup
@@ -10,11 +12,13 @@
 static VALUE duplicate(VALUE self)
 {
   xmlNodeSetPtr node_set;
+  xmlNodeSetPtr dupl;
+
   Data_Get_Struct(self, xmlNodeSet, node_set);
 
-  xmlNodeSetPtr dupl = xmlXPathNodeSetMerge(NULL, node_set);
+  dupl = xmlXPathNodeSetMerge(NULL, node_set);
 
-  return Nokogiri_wrap_xml_node_set(dupl);
+  return Nokogiri_wrap_xml_node_set(dupl, rb_iv_get(self, "@document"));
 }
 
 /*
@@ -98,7 +102,7 @@ static VALUE intersection(VALUE self, VALUE rb_other)
   Data_Get_Struct(self, xmlNodeSet, node_set);
   Data_Get_Struct(rb_other, xmlNodeSet, other);
 
-  return Nokogiri_wrap_xml_node_set(xmlXPathIntersection(node_set, other));
+  return Nokogiri_wrap_xml_node_set(xmlXPathIntersection(node_set, other), rb_iv_get(self, "@document"));
 }
 
 
@@ -145,11 +149,7 @@ static VALUE set_union(VALUE self, VALUE rb_other)
   new = xmlXPathNodeSetMerge(NULL, node_set);
   new = xmlXPathNodeSetMerge(new, other);
 
-  VALUE new_set = Nokogiri_wrap_xml_node_set(new);
-
-  rb_iv_set(new_set, "@document", rb_iv_get(self, "@document"));
-
-  return new_set;
+  return Nokogiri_wrap_xml_node_set(new, rb_iv_get(self, "@document"));
 }
 
 /*
@@ -177,7 +177,7 @@ static VALUE minus(VALUE self, VALUE rb_other)
     xmlXPathNodeSetDel(new, other->nodeTab[j]);
   }
 
-  return Nokogiri_wrap_xml_node_set(new);
+  return Nokogiri_wrap_xml_node_set(new, rb_iv_get(self, "@document"));
 }
 
 
@@ -186,7 +186,7 @@ static VALUE index_at(VALUE self, long offset)
   xmlNodeSetPtr node_set;
   Data_Get_Struct(self, xmlNodeSet, node_set);
 
-  if(offset >= node_set->nodeNr || abs(offset) > node_set->nodeNr) return Qnil;
+  if(offset >= node_set->nodeNr || abs((int)offset) > node_set->nodeNr) return Qnil;
   if(offset < 0) offset = offset + node_set->nodeNr;
 
   if (XML_NAMESPACE_DECL == node_set->nodeTab[offset]->type)
@@ -196,7 +196,7 @@ static VALUE index_at(VALUE self, long offset)
 
 static VALUE subseq(VALUE self, long beg, long len)
 {
-  int j;
+  long j;
   xmlNodeSetPtr node_set;
   xmlNodeSetPtr new_set ;
 
@@ -205,11 +205,15 @@ static VALUE subseq(VALUE self, long beg, long len)
   if (beg > node_set->nodeNr) return Qnil ;
   if (beg < 0 || len < 0) return Qnil ;
 
+  if ((beg + len) > node_set->nodeNr) {
+    len = node_set->nodeNr - beg ;
+  }
+
   new_set = xmlXPathNodeSetCreate(NULL);
   for (j = beg ; j < beg+len ; ++j) {
-    xmlXPathNodeSetAdd(new_set, node_set->nodeTab[j]);
+    xmlXPathNodeSetAddUnique(new_set, node_set->nodeTab[j]);
   }
-  return Nokogiri_wrap_xml_node_set(new_set);
+  return Nokogiri_wrap_xml_node_set(new_set, rb_iv_get(self, "@document"));
 }
 
 /*
@@ -240,8 +244,6 @@ static VALUE slice(int argc, VALUE *argv, VALUE self)
     if (beg < 0) {
       beg += node_set->nodeNr ;
     }
-    if (len > node_set->nodeNr) len = node_set->nodeNr;
-
     return subseq(self, beg, len);
   }
 
@@ -277,10 +279,13 @@ static VALUE slice(int argc, VALUE *argv, VALUE self)
 static VALUE to_array(VALUE self, VALUE rb_node)
 {
   xmlNodeSetPtr set;
+  VALUE *elts;
+  VALUE list;
+  int i;
+
   Data_Get_Struct(self, xmlNodeSet, set);
 
-  VALUE *elts = calloc((size_t)set->nodeNr, sizeof(VALUE *));
-  int i;
+  elts = calloc((size_t)set->nodeNr, sizeof(VALUE *));
   for(i = 0; i < set->nodeNr; i++) {
     if (XML_NAMESPACE_DECL == set->nodeTab[i]->type) {
       elts[i] = Nokogiri_wrap_xml_namespace2(rb_iv_get(self, "@document"), (xmlNsPtr)(set->nodeTab[i]));
@@ -298,9 +303,9 @@ static VALUE to_array(VALUE self, VALUE rb_node)
     }
   }
 
-  VALUE list = rb_ary_new4((long)set->nodeNr, elts);
+  list = rb_ary_new4((long)set->nodeNr, elts);
 
-  //free(elts);
+  /*free(elts); */
 
   return list;
 }
@@ -373,12 +378,18 @@ static void deallocate(xmlNodeSetPtr node_set)
 
 static VALUE allocate(VALUE klass)
 {
-  return Nokogiri_wrap_xml_node_set(xmlXPathNodeSetCreate(NULL));
+  return Nokogiri_wrap_xml_node_set(xmlXPathNodeSetCreate(NULL), Qnil);
 }
 
-VALUE Nokogiri_wrap_xml_node_set(xmlNodeSetPtr node_set)
+VALUE Nokogiri_wrap_xml_node_set(xmlNodeSetPtr node_set, VALUE document)
 {
-  return Data_Wrap_Struct(cNokogiriXmlNodeSet, 0, deallocate, node_set);
+  VALUE new_set ;
+  new_set = Data_Wrap_Struct(cNokogiriXmlNodeSet, 0, deallocate, node_set);
+  if (document != Qnil) {
+    rb_iv_set(new_set, "@document", document);
+    rb_funcall(document, decorate, 1, new_set);
+  }
+  return new_set ;
 }
 
 VALUE cNokogiriXmlNodeSet ;
@@ -402,4 +413,6 @@ void init_xml_node_set(void)
   rb_define_method(klass, "delete", delete, 1);
   rb_define_method(klass, "&", intersection, 1);
   rb_define_method(klass, "include?", include_eh, 1);
+
+  decorate      = rb_intern("decorate");
 }

data/ext/nokogiri/xml_node_set.h

@@ -5,5 +5,5 @@
 void init_xml_node_set();
 
 extern VALUE cNokogiriXmlNodeSet ;
-VALUE Nokogiri_wrap_xml_node_set(xmlNodeSetPtr node_set) ;
+VALUE Nokogiri_wrap_xml_node_set(xmlNodeSetPtr node_set, VALUE document) ;
 #endif

data/ext/nokogiri/xml_processing_instruction.c

@@ -10,16 +10,18 @@
 static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
+  xmlNodePtr node;
   VALUE document;
   VALUE name;
   VALUE content;
   VALUE rest;
+  VALUE rb_node;
 
   rb_scan_args(argc, argv, "3*", &document, &name, &content, &rest);
 
   Data_Get_Struct(document, xmlDoc, xml_doc);
 
-  xmlNodePtr node = xmlNewDocPI(
+  node = xmlNewDocPI(
       xml_doc,
       (const xmlChar *)StringValuePtr(name),
       (const xmlChar *)StringValuePtr(content)
@@ -27,7 +29,7 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   NOKOGIRI_ROOT_NODE(node);
 
-  VALUE rb_node = Nokogiri_wrap_xml_node(klass, node);
+  rb_node = Nokogiri_wrap_xml_node(klass, node);
   rb_obj_call_init(rb_node, argc, argv);
 
   if(rb_block_given_p()) rb_yield(rb_node);

data/ext/nokogiri/xml_reader.c

@@ -74,8 +74,10 @@ static void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash)
 static VALUE default_eh(VALUE self)
 {
   xmlTextReaderPtr reader;
+  int eh;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  int eh = xmlTextReaderIsDefault(reader);
+  eh = xmlTextReaderIsDefault(reader);
   if(eh == 0) return Qfalse;
   if(eh == 1) return Qtrue;
 
@@ -91,8 +93,10 @@ static VALUE default_eh(VALUE self)
 static VALUE value_eh(VALUE self)
 {
   xmlTextReaderPtr reader;
+  int eh;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  int eh = xmlTextReaderHasValue(reader);
+  eh = xmlTextReaderHasValue(reader);
   if(eh == 0) return Qfalse;
   if(eh == 1) return Qtrue;
 
@@ -108,8 +112,10 @@ static VALUE value_eh(VALUE self)
 static VALUE attributes_eh(VALUE self)
 {
   xmlTextReaderPtr reader;
+  int eh;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  int eh = has_attributes(reader);
+  eh = has_attributes(reader);
   if(eh == 0) return Qfalse;
   if(eh == 1) return Qtrue;
 
@@ -125,6 +131,7 @@ static VALUE attributes_eh(VALUE self)
 static VALUE namespaces(VALUE self)
 {
   xmlTextReaderPtr reader;
+  xmlNodePtr ptr;
   VALUE attr ;
 
   Data_Get_Struct(self, xmlTextReader, reader);
@@ -134,7 +141,7 @@ static VALUE namespaces(VALUE self)
   if (! has_attributes(reader))
     return attr ;
 
-  xmlNodePtr ptr = xmlTextReaderExpand(reader);
+  ptr = xmlTextReaderExpand(reader);
   if(ptr == NULL) return Qnil;
 
   Nokogiri_xml_node_namespaces(ptr, attr);
@@ -151,6 +158,7 @@ static VALUE namespaces(VALUE self)
 static VALUE attribute_nodes(VALUE self)
 {
   xmlTextReaderPtr reader;
+  xmlNodePtr ptr;
   VALUE attr ;
 
   Data_Get_Struct(self, xmlTextReader, reader);
@@ -160,7 +168,7 @@ static VALUE attribute_nodes(VALUE self)
   if (! has_attributes(reader))
     return attr ;
 
-  xmlNodePtr ptr = xmlTextReaderExpand(reader);
+  ptr = xmlTextReaderExpand(reader);
   if(ptr == NULL) return Qnil;
 
   Nokogiri_xml_node_properties(ptr, attr);
@@ -177,18 +185,21 @@ static VALUE attribute_nodes(VALUE self)
 static VALUE attribute_at(VALUE self, VALUE index)
 {
   xmlTextReaderPtr reader;
+  xmlChar *value;
+  VALUE rb_value;
+
   Data_Get_Struct(self, xmlTextReader, reader);
 
   if(NIL_P(index)) return Qnil;
   index = rb_Integer(index);
 
-  xmlChar * value = xmlTextReaderGetAttributeNo(
+  value = xmlTextReaderGetAttributeNo(
       reader,
-      NUM2INT(index)
+      (int)NUM2INT(index)
   );
   if(value == NULL) return Qnil;
 
-  VALUE rb_value = NOKOGIRI_STR_NEW2(value);
+  rb_value = NOKOGIRI_STR_NEW2(value);
   xmlFree(value);
   return rb_value;
 }
@@ -203,6 +214,8 @@ static VALUE reader_attribute(VALUE self, VALUE name)
 {
   xmlTextReaderPtr reader;
   xmlChar *value ;
+  VALUE rb_value;
+
   Data_Get_Struct(self, xmlTextReader, reader);
 
   if(NIL_P(name)) return Qnil;
@@ -224,7 +237,7 @@ static VALUE reader_attribute(VALUE self, VALUE name)
   }
   if(value == NULL) return Qnil;
 
-  VALUE rb_value = NOKOGIRI_STR_NEW2(value);
+  rb_value = NOKOGIRI_STR_NEW2(value);
   xmlFree(value);
   return rb_value;
 }
@@ -238,8 +251,10 @@ static VALUE reader_attribute(VALUE self, VALUE name)
 static VALUE attribute_count(VALUE self)
 {
   xmlTextReaderPtr reader;
+  int count;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  int count = xmlTextReaderAttributeCount(reader);
+  count = xmlTextReaderAttributeCount(reader);
   if(count == -1) return Qnil;
 
   return INT2NUM((long)count);
@@ -254,8 +269,10 @@ static VALUE attribute_count(VALUE self)
 static VALUE depth(VALUE self)
 {
   xmlTextReaderPtr reader;
+  int depth;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  int depth = xmlTextReaderDepth(reader);
+  depth = xmlTextReaderDepth(reader);
   if(depth == -1) return Qnil;
 
   return INT2NUM((long)depth);
@@ -270,8 +287,10 @@ static VALUE depth(VALUE self)
 static VALUE xml_version(VALUE self)
 {
   xmlTextReaderPtr reader;
+  const char *version;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  const char * version = (const char *)xmlTextReaderConstXmlVersion(reader);
+  version = (const char *)xmlTextReaderConstXmlVersion(reader);
   if(version == NULL) return Qnil;
 
   return NOKOGIRI_STR_NEW2(version);
@@ -286,8 +305,10 @@ static VALUE xml_version(VALUE self)
 static VALUE lang(VALUE self)
 {
   xmlTextReaderPtr reader;
+  const char *lang;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  const char * lang = (const char *)xmlTextReaderConstXmlLang(reader);
+  lang = (const char *)xmlTextReaderConstXmlLang(reader);
   if(lang == NULL) return Qnil;
 
   return NOKOGIRI_STR_NEW2(lang);
@@ -297,13 +318,15 @@ static VALUE lang(VALUE self)
  * call-seq:
  *   value
  *
- * Get the text value of the node if present
+ * Get the text value of the node if present. Returns a utf-8 encoded string.
  */
 static VALUE value(VALUE self)
 {
   xmlTextReaderPtr reader;
+  const char *value;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  const char * value = (const char *)xmlTextReaderConstValue(reader);
+  value = (const char *)xmlTextReaderConstValue(reader);
   if(value == NULL) return Qnil;
 
   return NOKOGIRI_STR_NEW2(value);
@@ -318,8 +341,10 @@ static VALUE value(VALUE self)
 static VALUE prefix(VALUE self)
 {
   xmlTextReaderPtr reader;
+  const char *prefix;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  const char * prefix = (const char *)xmlTextReaderConstPrefix(reader);
+  prefix = (const char *)xmlTextReaderConstPrefix(reader);
   if(prefix == NULL) return Qnil;
 
   return NOKOGIRI_STR_NEW2(prefix);
@@ -334,8 +359,10 @@ static VALUE prefix(VALUE self)
 static VALUE namespace_uri(VALUE self)
 {
   xmlTextReaderPtr reader;
+  const char *uri;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  const char * uri = (const char *)xmlTextReaderConstNamespaceUri(reader);
+  uri = (const char *)xmlTextReaderConstNamespaceUri(reader);
   if(uri == NULL) return Qnil;
 
   return NOKOGIRI_STR_NEW2(uri);
@@ -350,8 +377,10 @@ static VALUE namespace_uri(VALUE self)
 static VALUE local_name(VALUE self)
 {
   xmlTextReaderPtr reader;
+  const char *name;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  const char * name = (const char *)xmlTextReaderConstLocalName(reader);
+  name = (const char *)xmlTextReaderConstLocalName(reader);
   if(name == NULL) return Qnil;
 
   return NOKOGIRI_STR_NEW2(name);
@@ -361,18 +390,38 @@ static VALUE local_name(VALUE self)
  * call-seq:
  *   name
  *
- * Get the name of the node
+ * Get the name of the node. Returns a utf-8 encoded string.
  */
 static VALUE name(VALUE self)
 {
   xmlTextReaderPtr reader;
+  const char *name;
+
   Data_Get_Struct(self, xmlTextReader, reader);
-  const char * name = (const char *)xmlTextReaderConstName(reader);
+  name = (const char *)xmlTextReaderConstName(reader);
   if(name == NULL) return Qnil;
 
   return NOKOGIRI_STR_NEW2(name);
 }
 
+/*
+ * call-seq:
+ * base_uri
+ *
+ * Get the xml:base of the node
+ */
+static VALUE base_uri(VALUE self)
+{
+  xmlTextReaderPtr reader;
+  const char * base_uri;
+
+  Data_Get_Struct(self, xmlTextReader, reader);
+  base_uri = (const char *)xmlTextReaderBaseUri(reader);
+  if (base_uri == NULL) return Qnil;
+
+  return NOKOGIRI_STR_NEW2(base_uri);
+}
+
 /*
  * call-seq:
  *   state
@@ -408,18 +457,22 @@ static VALUE node_type(VALUE self)
 static VALUE read_more(VALUE self)
 {
   xmlTextReaderPtr reader;
+  xmlErrorPtr error;
+  VALUE error_list;
+  int ret;
+
   Data_Get_Struct(self, xmlTextReader, reader);
 
-  VALUE error_list = rb_funcall(self, rb_intern("errors"), 0);
+  error_list = rb_funcall(self, rb_intern("errors"), 0);
 
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
-  int ret = xmlTextReaderRead(reader);
+  ret = xmlTextReaderRead(reader);
   xmlSetStructuredErrorFunc(NULL, NULL);
 
   if(ret == 1) return self;
   if(ret == 0) return Qnil;
 
-  xmlErrorPtr error = xmlGetLastError();
+  error = xmlGetLastError();
   if(error)
     rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
   else
@@ -433,18 +486,25 @@ static VALUE read_more(VALUE self)
  *   inner_xml
  *
  * Read the contents of the current node, including child nodes and markup.
+ * Returns a utf-8 encoded string.
  */
 static VALUE inner_xml(VALUE self)
 {
   xmlTextReaderPtr reader;
+  xmlChar* value;
+  VALUE str;
+
   Data_Get_Struct(self, xmlTextReader, reader);
 
-  const char * value = (const char *)xmlTextReaderReadInnerXml(reader);
+  value = xmlTextReaderReadInnerXml(reader);
 
-  if(value == NULL)
-    return Qnil;
-  else
-    return NOKOGIRI_STR_NEW2(value);
+  str = Qnil;
+  if(value) {
+    str = NOKOGIRI_STR_NEW2((char*)value);
+    xmlFree(value);
+  }
+
+  return str;
 }
 
 /*
@@ -452,18 +512,23 @@ static VALUE inner_xml(VALUE self)
  *   outer_xml
  *
  * Read the current node and its contents, including child nodes and markup.
+ * Returns a utf-8 encoded string.
  */
 static VALUE outer_xml(VALUE self)
 {
   xmlTextReaderPtr reader;
-  Data_Get_Struct(self, xmlTextReader, reader);
+  xmlChar *value;
+  VALUE str = Qnil;
 
-  const char * value = (const char *)xmlTextReaderReadOuterXml(reader);
+  Data_Get_Struct(self, xmlTextReader, reader);
 
-  if(value == NULL)
-    return Qnil;
-  else
-    return NOKOGIRI_STR_NEW2(value);
+  value = xmlTextReaderReadOuterXml(reader);
+	
+  if(value) {
+    str = NOKOGIRI_STR_NEW2((char*)value);
+    xmlFree(value);
+  }
+  return str;
 }
 
 /*
@@ -475,21 +540,22 @@ static VALUE outer_xml(VALUE self)
 static VALUE from_memory(int argc, VALUE *argv, VALUE klass)
 {
   VALUE rb_buffer, rb_url, encoding, rb_options;
-
+  xmlTextReaderPtr reader;
   const char * c_url      = NULL;
   const char * c_encoding = NULL;
-  int c_options           = 0; 
+  int c_options           = 0;
+  VALUE rb_reader, args[3];
 
   rb_scan_args(argc, argv, "13", &rb_buffer, &rb_url, &encoding, &rb_options);
 
   if (!RTEST(rb_buffer)) rb_raise(rb_eArgError, "string cannot be nil");
   if (RTEST(rb_url)) c_url = StringValuePtr(rb_url);
   if (RTEST(encoding)) c_encoding = StringValuePtr(encoding);
-  if (RTEST(rb_options)) c_options = NUM2INT(rb_options);
+  if (RTEST(rb_options)) c_options = (int)NUM2INT(rb_options);
 
-  xmlTextReaderPtr reader = xmlReaderForMemory(
+  reader = xmlReaderForMemory(
       StringValuePtr(rb_buffer),
-      RSTRING_LEN(rb_buffer),
+      (int)RSTRING_LEN(rb_buffer),
       c_url,
       c_encoding,
       c_options
@@ -500,8 +566,10 @@ static VALUE from_memory(int argc, VALUE *argv, VALUE klass)
     rb_raise(rb_eRuntimeError, "couldn't create a parser");
   }
 
-  VALUE rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
-  VALUE args[3] = {rb_buffer, rb_url, encoding};
+  rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
+  args[0] = rb_buffer;
+  args[1] = rb_url;
+  args[2] = encoding;
   rb_obj_call_init(rb_reader, 3, args);
 
   return rb_reader;
@@ -516,19 +584,20 @@ static VALUE from_memory(int argc, VALUE *argv, VALUE klass)
 static VALUE from_io(int argc, VALUE *argv, VALUE klass)
 {
   VALUE rb_io, rb_url, encoding, rb_options;
-
+  xmlTextReaderPtr reader;
   const char * c_url      = NULL;
   const char * c_encoding = NULL;
-  int c_options           = 0; 
+  int c_options           = 0;
+  VALUE rb_reader, args[3];
 
   rb_scan_args(argc, argv, "13", &rb_io, &rb_url, &encoding, &rb_options);
 
   if (!RTEST(rb_io)) rb_raise(rb_eArgError, "io cannot be nil");
   if (RTEST(rb_url)) c_url = StringValuePtr(rb_url);
   if (RTEST(encoding)) c_encoding = StringValuePtr(encoding);
-  if (RTEST(rb_options)) c_options = NUM2INT(rb_options);
+  if (RTEST(rb_options)) c_options = (int)NUM2INT(rb_options);
 
-  xmlTextReaderPtr reader = xmlReaderForIO(
+  reader = xmlReaderForIO(
       (xmlInputReadCallback)io_read_callback,
       (xmlInputCloseCallback)io_close_callback,
       (void *)rb_io,
@@ -542,8 +611,10 @@ static VALUE from_io(int argc, VALUE *argv, VALUE klass)
     rb_raise(rb_eRuntimeError, "couldn't create a parser");
   }
 
-  VALUE rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
-  VALUE args[3] = {rb_io, rb_url, encoding};
+  rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
+  args[0] = rb_io;
+  args[1] = rb_url;
+  args[2] = encoding;
   rb_obj_call_init(rb_reader, 3, args);
 
   return rb_reader;
@@ -588,6 +659,7 @@ void init_xml_reader()
   rb_define_method(klass, "attributes?", attributes_eh, 0);
   rb_define_method(klass, "value?", value_eh, 0);
   rb_define_method(klass, "default?", default_eh, 0);
+  rb_define_method(klass, "base_uri", base_uri, 0);
 
   rb_define_private_method(klass, "attr_nodes", attribute_nodes, 0);
 }