nokogiri 1.5.10 → 1.10.4

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;
@@ -77,7 +77,7 @@ static VALUE read_io( VALUE klass,
 
     error = xmlGetLastError();
     if(error)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
     else
       rb_raise(rb_eRuntimeError, "Could not parse document");
 
@@ -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;
@@ -123,7 +123,7 @@ static VALUE read_memory( VALUE klass,
 
     error = xmlGetLastError();
     if(error)
-      rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
+      rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
     else
       rb_raise(rb_eRuntimeError, "Could not parse document");
 

data/ext/nokogiri/html_element_description.c

@@ -8,7 +8,7 @@
  */
 static VALUE required_attributes(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   VALUE list;
   int i;
 
@@ -33,7 +33,7 @@ static VALUE required_attributes(VALUE self)
  */
 static VALUE deprecated_attributes(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   VALUE list;
   int i;
 
@@ -58,7 +58,7 @@ static VALUE deprecated_attributes(VALUE self)
  */
 static VALUE optional_attributes(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   VALUE list;
   int i;
 
@@ -83,7 +83,7 @@ static VALUE optional_attributes(VALUE self)
  */
 static VALUE default_sub_element(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if (description->defaultsubelt)
@@ -100,7 +100,7 @@ static VALUE default_sub_element(VALUE self)
  */
 static VALUE sub_elements(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   VALUE list;
   int i;
 
@@ -125,7 +125,7 @@ static VALUE sub_elements(VALUE self)
  */
 static VALUE description(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   return NOKOGIRI_STR_NEW2(description->desc);
@@ -139,7 +139,7 @@ static VALUE description(VALUE self)
  */
 static VALUE inline_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->isinline) return Qtrue;
@@ -154,7 +154,7 @@ static VALUE inline_eh(VALUE self)
  */
 static VALUE deprecated_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->depr) return Qtrue;
@@ -169,7 +169,7 @@ static VALUE deprecated_eh(VALUE self)
  */
 static VALUE empty_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->empty) return Qtrue;
@@ -184,7 +184,7 @@ static VALUE empty_eh(VALUE self)
  */
 static VALUE save_end_tag_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->saveEndTag) return Qtrue;
@@ -199,7 +199,7 @@ static VALUE save_end_tag_eh(VALUE self)
  */
 static VALUE implied_end_tag_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->endTag) return Qtrue;
@@ -214,7 +214,7 @@ static VALUE implied_end_tag_eh(VALUE self)
  */
 static VALUE implied_start_tag_eh(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(description->startTag) return Qtrue;
@@ -229,7 +229,7 @@ static VALUE implied_start_tag_eh(VALUE self)
  */
 static VALUE name(VALUE self)
 {
-  htmlElemDesc * description;
+  const htmlElemDesc * description;
   Data_Get_Struct(self, htmlElemDesc, description);
 
   if(NULL == description->name) return Qnil;
@@ -245,11 +245,11 @@ 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;
-  return Data_Wrap_Struct(klass, 0, 0, (void *)description);
+  return Data_Wrap_Struct(klass, 0, 0, (void *)(uintptr_t)description);
 }
 
 VALUE cNokogiriHtmlElementDescription ;

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

@@ -14,7 +14,12 @@ VALUE mNokogiriHtmlSax ;
  */
 int vasprintf (char **strp, const char *fmt, va_list ap)
 {
-  int len = vsnprintf (NULL, 0, fmt, ap) + 1;
+  /* Mingw32/64 have a broken vsnprintf implementation that fails when
+   * using a zero-byte limit in order to retrieve the required size for malloc.
+   * So we use a one byte buffer instead.
+   */
+  char tmp[1];
+  int len = vsnprintf (tmp, 1, fmt, ap) + 1;
   char *res = (char *)malloc((unsigned int)len);
   if (res == NULL)
       return -1;
@@ -23,25 +28,16 @@ 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"
 #else
-#ifndef __MACRUBY__
 #include "util.h"
 #endif
-#endif
 
 void nokogiri_root_node(xmlNodePtr node)
 {
@@ -65,14 +61,12 @@ void nokogiri_root_nsdef(xmlNsPtr ns, xmlDocPtr doc)
 
 void Init_nokogiri()
 {
-#ifndef __MACRUBY__
   xmlMemSetup(
       (xmlFreeFunc)ruby_xfree,
       (xmlMallocFunc)ruby_xmalloc,
       (xmlReallocFunc)ruby_xrealloc,
       ruby_strdup
   );
-#endif
 
   mNokogiri         = rb_define_module("Nokogiri");
   mNokogiriXml      = rb_define_module_under(mNokogiri, "XML");
@@ -90,6 +84,20 @@ void Init_nokogiri()
                 NOKOGIRI_STR_NEW2(xmlParserVersion)
               );
 
+#ifdef NOKOGIRI_USE_PACKAGED_LIBRARIES
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_USE_PACKAGED_LIBRARIES"), Qtrue);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATH"), NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATH));
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATH"), NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATH));
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATCHES), " "));
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATCHES), " "));
+#else
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_USE_PACKAGED_LIBRARIES"), Qfalse);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATH"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATH"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXML2_PATCHES"), Qnil);
+  rb_const_set(mNokogiri, rb_intern("NOKOGIRI_LIBXSLT_PATCHES"), Qnil);
+#endif
+
 #ifdef LIBXML_ICONV_ENABLED
   rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qtrue);
 #else

data/ext/nokogiri/nokogiri.h

@@ -31,22 +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
-
-#ifndef UNUSED
-# if defined(__GNUC__)
-#  define MAYBE_UNUSED(name) name __attribute__((unused))
-#  define UNUSED(name) MAYBE_UNUSED(UNUSED_ ## name)
-# else
-#  define MAYBE_UNUSED(name) name
-#  define UNUSED(name) name
-# endif
-#endif
+#include <ruby/encoding.h>
 
 #ifndef NORETURN
 # if defined(__GNUC__)
@@ -56,25 +42,12 @@ 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 +108,6 @@ 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

@@ -4,37 +4,40 @@
  * call-seq:
  *  value=(content)
  *
- * Set the value for this Attr to +content+
+ * Set the value for this Attr to +content+. Use `nil` to remove the value
+ * (e.g., a HTML boolean attribute).
  */
 static VALUE set_value(VALUE self, VALUE content)
 {
   xmlAttrPtr attr;
-  Data_Get_Struct(self, xmlAttr, attr);
+  xmlChar *value;
+  xmlNode *cur;
 
-  if(attr->children) xmlFreeNodeList(attr->children);
+  Data_Get_Struct(self, xmlAttr, attr);
 
+  if (attr->children) {
+    xmlFreeNodeList(attr->children);
+  }
   attr->children = attr->last = NULL;
 
-  if(content) {
-    xmlChar *buffer;
-    xmlNode *tmp;
-
-    /* Encode our content */
-    buffer = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValuePtr(content));
+  if (content == Qnil) {
+    return content;
+  }
 
-    attr->children = xmlStringGetNodeList(attr->doc, buffer);
-    attr->last = NULL;
-    tmp = attr->children;
+  value = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValueCStr(content));
+  if (xmlStrlen(value) == 0) {
+    attr->children = xmlNewDocText(attr->doc, value);
+  } else {
+    attr->children = xmlStringGetNodeList(attr->doc, value);
+  }
+  xmlFree(value);
 
-    /* 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;
+  for (cur = attr->children; cur; cur = cur->next) {
+    cur->parent = (xmlNode *)attr;
+    cur->doc = attr->doc;
+    if (cur->next == NULL) {
+      attr->last = cur;
     }
-
-    /* Free up memory */
-    xmlFree(buffer);
   }
 
   return content;
@@ -57,20 +60,26 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   rb_scan_args(argc, argv, "2*", &document, &name, &rest);
 
+  if (! rb_obj_is_kind_of(document, cNokogiriXmlDocument)) {
+    rb_raise(rb_eArgError, "parameter must be a Nokogiri::XML::Document");
+  }
+
   Data_Get_Struct(document, xmlDoc, xml_doc);
 
   node = xmlNewDocProp(
-      xml_doc,
-      (const xmlChar *)StringValuePtr(name),
-      NULL
-  );
+           xml_doc,
+           (const xmlChar *)StringValueCStr(name),
+           NULL
+         );
 
   nokogiri_root_node((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);
+  if (rb_block_given_p()) {
+    rb_yield(rb_node);
+  }
 
   return rb_node;
 }

data/ext/nokogiri/xml_cdata.c

@@ -5,6 +5,9 @@
  *  new(document, content)
  *
  * Create a new CDATA element on the +document+ with +content+
+ *
+ * If +content+ cannot be implicitly converted to a string, this method will
+ * raise a TypeError exception.
  */
 static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
@@ -14,23 +17,26 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
   VALUE content;
   VALUE rest;
   VALUE rb_node;
+  xmlChar *content_str = NULL;
+  int content_str_len = 0;
 
   rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
 
   Data_Get_Struct(doc, xmlDoc, xml_doc);
 
-  node = xmlNewCDataBlock(
-      xml_doc->doc,
-      NIL_P(content) ? NULL : (const xmlChar *)StringValuePtr(content),
-      NIL_P(content) ? 0 : (int)RSTRING_LEN(content)
-  );
+  if (!NIL_P(content)) {
+    content_str = (xmlChar *)StringValuePtr(content);
+    content_str_len = RSTRING_LEN(content);
+  }
+
+  node = xmlNewCDataBlock(xml_doc->doc, content_str, content_str_len);
 
   nokogiri_root_node(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);
+  if(rb_block_given_p()) { rb_yield(rb_node); }
 
   return rb_node;
 }

data/ext/nokogiri/xml_comment.c

@@ -1,10 +1,13 @@
 #include <xml_comment.h>
 
+static ID document_id ;
+
 /*
  * call-seq:
- *  new(document, content)
+ *  new(document_or_node, content)
  *
- * Create a new Comment element on the +document+ with +content+
+ * Create a new Comment element on the +document+ with +content+.
+ * Alternatively, if a +node+ is passed, the +node+'s document is used.
  */
 static VALUE new(int argc, VALUE *argv, VALUE klass)
 {
@@ -17,11 +20,21 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   rb_scan_args(argc, argv, "2*", &document, &content, &rest);
 
+  if (rb_obj_is_kind_of(document, cNokogiriXmlNode))
+  {
+    document = rb_funcall(document, document_id, 0);
+  }
+  else if (   !rb_obj_is_kind_of(document, cNokogiriXmlDocument)
+           && !rb_obj_is_kind_of(document, cNokogiriXmlDocumentFragment))
+  {
+    rb_raise(rb_eArgError, "first argument must be a XML::Document or XML::Node");
+  }
+
   Data_Get_Struct(document, xmlDoc, xml_doc);
 
   node = xmlNewDocComment(
       xml_doc,
-      (const xmlChar *)StringValuePtr(content)
+      (const xmlChar *)StringValueCStr(content)
   );
 
   rb_node = Nokogiri_wrap_xml_node(klass, node);
@@ -51,4 +64,6 @@ void init_xml_comment()
   cNokogiriXmlComment = klass;
 
   rb_define_singleton_method(klass, "new", new, -1);
+
+  document_id = rb_intern("document");
 }