nokogiri 1.4.1 → 1.4.2

data/ext/nokogiri/html_document.c

@@ -9,12 +9,13 @@
 static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
   VALUE uri, external_id, rest, rb_doc;
+  htmlDocPtr doc;
 
   rb_scan_args(argc, argv, "0*", &rest);
   uri         = rb_ary_entry(rest, (long)0);
   external_id = rb_ary_entry(rest, (long)1);
 
-  htmlDocPtr doc = htmlNewDoc(
+  doc = htmlNewDoc(
       RTEST(uri) ? (const xmlChar *)StringValuePtr(uri) : NULL,
       RTEST(external_id) ? (const xmlChar *)StringValuePtr(external_id) : NULL
   );
@@ -39,11 +40,13 @@ static VALUE read_io( VALUE klass,
   const char * c_url    = NIL_P(url)      ? NULL : StringValuePtr(url);
   const char * c_enc    = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
   VALUE error_list      = rb_ary_new();
+  VALUE document;
+  htmlDocPtr doc;
 
   xmlResetLastError();
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 
-  htmlDocPtr doc = htmlReadIO(
+  doc = htmlReadIO(
       io_read_callback,
       io_close_callback,
       (void *)io,
@@ -54,9 +57,11 @@ static VALUE read_io( VALUE klass,
   xmlSetStructuredErrorFunc(NULL, NULL);
 
   if(doc == NULL) {
+    xmlErrorPtr error;
+
     xmlFreeDoc(doc);
 
-    xmlErrorPtr error = xmlGetLastError();
+    error = xmlGetLastError();
     if(error)
       rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
     else
@@ -65,7 +70,7 @@ static VALUE read_io( VALUE klass,
     return Qnil;
   }
 
-  VALUE document = Nokogiri_wrap_xml_document(klass, doc);
+  document = Nokogiri_wrap_xml_document(klass, doc);
   rb_iv_set(document, "@errors", error_list);
   return document;
 }
@@ -86,19 +91,23 @@ static VALUE read_memory( VALUE klass,
   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);
-  int len               = RSTRING_LEN(string);
+  int len               = (int)RSTRING_LEN(string);
   VALUE error_list      = rb_ary_new();
+  VALUE document;
+  htmlDocPtr doc;
 
   xmlResetLastError();
   xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
 
-  htmlDocPtr doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
+  doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
   xmlSetStructuredErrorFunc(NULL, NULL);
 
   if(doc == NULL) {
+    xmlErrorPtr error;
+
     xmlFreeDoc(doc);
 
-    xmlErrorPtr error = xmlGetLastError();
+    error = xmlGetLastError();
     if(error)
       rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
     else
@@ -107,7 +116,7 @@ static VALUE read_memory( VALUE klass,
     return Qnil;
   }
 
-  VALUE document = Nokogiri_wrap_xml_document(klass, doc);
+  document = Nokogiri_wrap_xml_document(klass, doc);
   rb_iv_set(document, "@errors", error_list);
   return document;
 }

data/ext/nokogiri/html_element_description.c

@@ -9,16 +9,17 @@
 static VALUE required_attributes(VALUE self)
 {
   htmlElemDesc * description;
+  VALUE list;
+  int i;
+
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  VALUE list = rb_ary_new();
+  list = rb_ary_new();
 
   if(NULL == description->attrs_req) return list;
 
-  int i = 0;
-  while(description->attrs_req[i]) {
+  for(i = 0; description->attrs_depr[i]; i++) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
-    i++;
   }
 
   return list;
@@ -33,16 +34,17 @@ static VALUE required_attributes(VALUE self)
 static VALUE deprecated_attributes(VALUE self)
 {
   htmlElemDesc * description;
+  VALUE list;
+  int i;
+
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  VALUE list = rb_ary_new();
+  list = rb_ary_new();
 
   if(NULL == description->attrs_depr) return list;
 
-  int i = 0;
-  while(description->attrs_depr[i]) {
+  for(i = 0; description->attrs_depr[i]; i++) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
-    i++;
   }
 
   return list;
@@ -57,16 +59,17 @@ static VALUE deprecated_attributes(VALUE self)
 static VALUE optional_attributes(VALUE self)
 {
   htmlElemDesc * description;
+  VALUE list;
+  int i;
+
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  VALUE list = rb_ary_new();
+  list = rb_ary_new();
 
   if(NULL == description->attrs_opt) return list;
 
-  int i = 0;
-  while(description->attrs_opt[i]) {
+  for(i = 0; description->attrs_opt[i]; i++) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
-    i++;
   }
 
   return list;
@@ -95,16 +98,17 @@ static VALUE default_sub_element(VALUE self)
 static VALUE sub_elements(VALUE self)
 {
   htmlElemDesc * description;
+  VALUE list;
+  int i;
+
   Data_Get_Struct(self, htmlElemDesc, description);
 
-  VALUE list = rb_ary_new();
+  list = rb_ary_new();
 
   if(NULL == description->subelts) return list;
 
-  int i = 0;
-  while(description->subelts[i]) {
+  for(i = 0; description->subelts[i]; i++) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
-    i++;
   }
 
   return list;

data/ext/nokogiri/html_entity_lookup.c

@@ -10,11 +10,11 @@ static VALUE get(VALUE self, VALUE key)
 {
   const htmlEntityDesc * desc =
     htmlEntityLookup((const xmlChar *)StringValuePtr(key));
+  VALUE klass, args[3];
 
   if(NULL == desc) return Qnil;
-  VALUE klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
+  klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
 
-  VALUE args[3];
   args[0] = INT2NUM((long)desc->value);
   args[1] = NOKOGIRI_STR_NEW2(desc->name);
   args[2] = NOKOGIRI_STR_NEW2(desc->desc);

data/ext/nokogiri/html_sax_parser_context.c

@@ -15,19 +15,21 @@ static void deallocate(xmlParserCtxtPtr ctxt)
 
 static VALUE parse_memory(VALUE klass, VALUE data, VALUE encoding)
 {
+  htmlParserCtxtPtr ctxt;
+
   if(NIL_P(data)) rb_raise(rb_eArgError, "data cannot be nil");
   if(!(int)RSTRING_LEN(data))
     rb_raise(rb_eRuntimeError, "data cannot be empty");
 
-  htmlParserCtxtPtr ctxt = htmlCreateMemoryParserCtxt(
+  ctxt = htmlCreateMemoryParserCtxt(
       StringValuePtr(data),
       (int)RSTRING_LEN(data)
   );
 
   if(RTEST(encoding)) {
-    xmlCharEncoding enc = xmlParseCharEncoding(StringValuePtr(encoding));
-    if(enc != XML_CHAR_ENCODING_ERROR) {
-      xmlSwitchEncoding(ctxt, enc);
+    xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValuePtr(encoding));
+    if(enc != NULL) {
+      xmlSwitchToEncoding(ctxt, enc);
       if(ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
         rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
             StringValuePtr(encoding));
@@ -49,16 +51,16 @@ static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
 
 static VALUE parse_with(VALUE self, VALUE sax_handler)
 {
+  htmlParserCtxtPtr ctxt;
+  htmlSAXHandlerPtr sax;
+
   if(!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
     rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
 
-  htmlParserCtxtPtr ctxt;
   Data_Get_Struct(self, htmlParserCtxt, ctxt);
-
-  htmlSAXHandlerPtr sax;
   Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
 
-  // Free the sax handler since we'll assign our own
+  /* Free the sax handler since we'll assign our own */
   if(ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
     xmlFree(ctxt->sax);
 

data/ext/nokogiri/nokogiri.c

@@ -73,7 +73,6 @@ void Init_nokogiri()
   init_xml_comment();
   init_xml_node_set();
   init_xml_xpath_context();
-  init_xml_xpath();
   init_xml_sax_parser_context();
   init_xml_sax_parser();
   init_xml_sax_push_parser();

data/ext/nokogiri/nokogiri.h

@@ -2,8 +2,22 @@
 #define NOKOGIRI_NATIVE
 
 #include <stdlib.h>
+#include <string.h>
 #include <assert.h>
+#include <stdarg.h>
+
+#ifdef USE_INCLUDED_VASPRINTF
+int vasprintf (char **strp, const char *fmt, va_list ap);
+#else
+
+#define _GNU_SOURCE
+#  include <stdio.h>
+#undef _GNU_SOURCE
+
+#endif
+
 #include <libxml/parser.h>
+#include <libxml/entities.h>
 #include <libxml/parserInternals.h>
 #include <libxml/xpath.h>
 #include <libxml/xpathInternals.h>
@@ -15,16 +29,10 @@
 #include <libxml/relaxng.h>
 #include <ruby.h>
 
-#ifdef USE_INCLUDED_VASPRINTF
-int vasprintf (char **strp, const char *fmt, va_list ap);
+#ifdef HAVE_RUBY_ENCODING_H
+#include <ruby/st.h>
 #else
-
-#ifndef _GNU_SOURCE
-#define _GNU_SOURCE
-#endif
-
-#  include <stdio.h>
-
+#include <st.h>
 #endif
 
 int is_2_6_16(void) ;
@@ -39,33 +47,31 @@ int is_2_6_16(void) ;
 # endif
 #endif
 
+#ifndef NORETURN
+# if defined(__GNUC__)
+#  define NORETURN(name) __attribute__((noreturn)) name
+# else
+#  define NORETURN(name) name
+# endif
+#endif
+
 #ifdef HAVE_RUBY_ENCODING_H
 
 #include <ruby/encoding.h>
 
 #define NOKOGIRI_STR_NEW2(str) \
-  ({ \
-    VALUE _string = rb_str_new2((const char *)str); \
-    int _enc = rb_enc_find_index("UTF-8"); \
-    rb_enc_associate_index(_string, _enc); \
-    _string; \
-  })
+  NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
 
 #define NOKOGIRI_STR_NEW(str, len) \
-  ({ \
-    VALUE _string = rb_str_new((const char *)str, (long)len); \
-    int _enc = rb_enc_find_index("UTF-8"); \
-    rb_enc_associate_index(_string, _enc); \
-    _string; \
-  })
+  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)
+  rb_str_new2((const char *)(str))
 
 #define NOKOGIRI_STR_NEW(str, len) \
-  rb_str_new((const char *)str, (long)len)
+  rb_str_new((const char *)(str), (long)(len))
 #endif
 
 #define RBSTR_OR_QNIL(_str) \
@@ -84,7 +90,6 @@ int is_2_6_16(void) ;
 #include <xml_document_fragment.h>
 #include <xml_comment.h>
 #include <xml_node_set.h>
-#include <xml_xpath.h>
 #include <xml_dtd.h>
 #include <xml_attribute_decl.h>
 #include <xml_element_decl.h>
@@ -112,10 +117,10 @@ extern VALUE mNokogiriHtmlSax ;
 extern VALUE mNokogiriXslt ;
 
 #define NOKOGIRI_ROOT_NODE(_node) \
-  ({ \
-    nokogiriTuplePtr tuple = (nokogiriTuplePtr)(_node->doc->_private);       \
-    st_insert(tuple->unlinkedNodes, (st_data_t)_node, (st_data_t)_node);     \
-  })
+  st_insert(((nokogiriTuplePtr)(_node)->doc->_private)->unlinkedNodes, (st_data_t)(_node), (st_data_t)(_node))
+
+#define NOKOGIRI_ROOT_NSDEF(_nsDef, _doc)     \
+  st_insert(((nokogiriTuplePtr)(_doc)->_private)->unlinkedNodes, (st_data_t)(_nsDef), (st_data_t)(_nsDef))
 
 #ifdef DEBUG
 

data/ext/nokogiri/xml_attr.c

@@ -19,21 +19,21 @@ static VALUE set_value(VALUE self, VALUE content)
     xmlChar *buffer;
     xmlNode *tmp;
 
-    // Encode our content
+    /* Encode our content */
     buffer = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValuePtr(content));
 
     attr->children = xmlStringGetNodeList(attr->doc, buffer);
     attr->last = NULL;
     tmp = attr->children;
 
-    // Loop through the children
+    /* Loop through the children */
     for(tmp = attr->children; tmp; tmp = tmp->next) {
       tmp->parent = (xmlNode *)attr;
       tmp->doc = attr->doc;
       if(tmp->next == NULL) attr->last = tmp;
     }
 
-    // Free up memory
+    /* Free up memory */
     xmlFree(buffer);
   }
 
@@ -52,12 +52,14 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   VALUE document;
   VALUE name;
   VALUE rest;
+  xmlAttrPtr node;
+  VALUE rb_node;
 
   rb_scan_args(argc, argv, "2*", &document, &name, &rest);
 
   Data_Get_Struct(document, xmlDoc, xml_doc);
 
-  xmlAttrPtr node = xmlNewDocProp(
+  node = xmlNewDocProp(
       xml_doc,
       (const xmlChar *)StringValuePtr(name),
       NULL
@@ -65,7 +67,7 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   NOKOGIRI_ROOT_NODE((xmlNodePtr)node);
 
-  VALUE rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
+  rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
   rb_obj_call_init(rb_node, argc, argv);
 
   if(rb_block_given_p()) rb_yield(rb_node);

data/ext/nokogiri/xml_attribute_decl.c

@@ -37,10 +37,13 @@ static VALUE default_value(VALUE self)
 static VALUE enumeration(VALUE self)
 {
   xmlAttributePtr node;
+  xmlEnumerationPtr enm;
+  VALUE list;
+
   Data_Get_Struct(self, xmlAttribute, node);
 
-  VALUE list = rb_ary_new();
-  xmlEnumerationPtr enm = node->tree;
+  list = rb_ary_new();
+  enm = node->tree;
 
   while(enm) {
     rb_ary_push(list, NOKOGIRI_STR_NEW2(enm->name));

data/ext/nokogiri/xml_cdata.c

@@ -9,15 +9,17 @@
 static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
+  xmlNodePtr node;
   VALUE doc;
   VALUE content;
   VALUE rest;
+  VALUE rb_node;
 
   rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
 
   Data_Get_Struct(doc, xmlDoc, xml_doc);
 
-  xmlNodePtr node = xmlNewCDataBlock(
+  node = xmlNewCDataBlock(
       xml_doc->doc,
       NIL_P(content) ? NULL : (const xmlChar *)StringValuePtr(content),
       NIL_P(content) ? 0 : (int)RSTRING_LEN(content)
@@ -25,7 +27,7 @@ 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_comment.c

@@ -9,20 +9,22 @@
 static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
   xmlDocPtr xml_doc;
+  xmlNodePtr node;
   VALUE document;
   VALUE content;
   VALUE rest;
+  VALUE rb_node;
 
   rb_scan_args(argc, argv, "2*", &document, &content, &rest);
 
   Data_Get_Struct(document, xmlDoc, xml_doc);
 
-  xmlNodePtr node = xmlNewDocComment(
+  node = xmlNewDocComment(
       xml_doc,
       (const xmlChar *)StringValuePtr(content)
   );
 
-  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);
 
   NOKOGIRI_ROOT_NODE(node);