nokogiri 1.6.7.2-java → 1.6.8-java

data/ext/nokogiri/xml_node_set.h

@@ -6,9 +6,8 @@ void init_xml_node_set();
 
 extern VALUE cNokogiriXmlNodeSet ;
 VALUE Nokogiri_wrap_xml_node_set(xmlNodeSetPtr node_set, VALUE document) ;
+VALUE Nokogiri_wrap_xml_node_set_node(xmlNodePtr node, VALUE node_set) ;
+VALUE Nokogiri_wrap_xml_node_set_namespace(xmlNsPtr node, VALUE node_set) ;
+int Nokogiri_namespace_eh(xmlNodePtr node) ;
 
-typedef struct _nokogiriNodeSetTuple {
-  xmlNodeSetPtr node_set;
-  st_table     *namespaces;
-} nokogiriNodeSetTuple;
 #endif

data/ext/nokogiri/xml_processing_instruction.c

@@ -23,8 +23,8 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   node = xmlNewDocPI(
       xml_doc,
-      (const xmlChar *)StringValuePtr(name),
-      (const xmlChar *)StringValuePtr(content)
+      (const xmlChar *)StringValueCStr(name),
+      (const xmlChar *)StringValueCStr(content)
   );
 
   nokogiri_root_node(node);

data/ext/nokogiri/xml_reader.c

@@ -218,20 +218,7 @@ static VALUE reader_attribute(VALUE self, VALUE name)
   if(NIL_P(name)) return Qnil;
   name = StringValue(name) ;
 
-  value = xmlTextReaderGetAttribute(reader, (xmlChar*)StringValuePtr(name));
-  if(value == NULL) {
-    /* this section is an attempt to workaround older versions of libxml that
-       don't handle namespaces properly in all attribute-and-friends functions */
-    xmlChar *prefix = NULL ;
-    xmlChar *localname = xmlSplitQName2((xmlChar*)StringValuePtr(name), &prefix);
-    if (localname != NULL) {
-      value = xmlTextReaderLookupNamespace(reader, localname);
-      xmlFree(localname) ;
-    } else {
-      value = xmlTextReaderLookupNamespace(reader, prefix);
-    }
-    xmlFree(prefix);
-  }
+  value = xmlTextReaderGetAttribute(reader, (xmlChar*)StringValueCStr(name));
   if(value == NULL) return Qnil;
 
   rb_value = NOKOGIRI_STR_NEW2(value);
@@ -546,8 +533,8 @@ static VALUE from_memory(int argc, VALUE *argv, VALUE klass)
   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_url)) c_url = StringValueCStr(rb_url);
+  if (RTEST(encoding)) c_encoding = StringValueCStr(encoding);
   if (RTEST(rb_options)) c_options = (int)NUM2INT(rb_options);
 
   reader = xmlReaderForMemory(
@@ -590,8 +577,8 @@ static VALUE from_io(int argc, VALUE *argv, VALUE klass)
   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_url)) c_url = StringValueCStr(rb_url);
+  if (RTEST(encoding)) c_encoding = StringValueCStr(encoding);
   if (RTEST(rb_options)) c_options = (int)NUM2INT(rb_options);
 
   reader = xmlReaderForIO(

data/ext/nokogiri/xml_sax_parser.c

@@ -9,9 +9,6 @@ static ID id_comment, id_characters, id_xmldecl, id_error, id_warning;
 static ID id_cdata_block, id_cAttribute;
 static ID id_processing_instruction;
 
-#define STRING_OR_NULL(str) \
-   (RTEST(str) ? StringValuePtr(str) : NULL)
-
 static void start_document(void * ctx)
 {
   VALUE self = NOKOGIRI_SAX_SELF(ctx);
@@ -21,16 +18,16 @@ static void start_document(void * ctx)
 
   if(NULL != ctxt && ctxt->html != 1) {
     if(ctxt->standalone != -1) {  /* -1 means there was no declaration */
-      VALUE encoding = ctxt->encoding ?
-        NOKOGIRI_STR_NEW2(ctxt->encoding) :
-        Qnil;
+      VALUE encoding = Qnil ;
+      if (ctxt->encoding) {
+        encoding = NOKOGIRI_STR_NEW2(ctxt->encoding) ;
+      } else if (ctxt->input && ctxt->input->encoding) {
+        encoding = NOKOGIRI_STR_NEW2(ctxt->input->encoding) ;
+      }
 
-      VALUE version = ctxt->version ?
-        NOKOGIRI_STR_NEW2(ctxt->version) :
-        Qnil;
+      VALUE version = ctxt->version ? NOKOGIRI_STR_NEW2(ctxt->version) : Qnil;
 
       VALUE standalone = Qnil;
-
       switch(ctxt->standalone)
       {
         case 0:
@@ -160,7 +157,7 @@ start_element_ns (
 }
 
 /**
- * end_element_ns was borrowed heavily from libxml-ruby. 
+ * end_element_ns was borrowed heavily from libxml-ruby.
  */
 static void
 end_element_ns (
@@ -172,7 +169,7 @@ end_element_ns (
   VALUE self = NOKOGIRI_SAX_SELF(ctx);
   VALUE doc = rb_iv_get(self, "@document");
 
-  rb_funcall(doc, id_end_element_namespace, 3, 
+  rb_funcall(doc, id_end_element_namespace, 3,
     NOKOGIRI_STR_NEW2(localname),
     RBSTR_OR_QNIL(prefix),
     RBSTR_OR_QNIL(uri)

data/ext/nokogiri/xml_sax_parser_context.c

@@ -45,7 +45,7 @@ parse_io(VALUE klass, VALUE io, VALUE encoding)
  */
 static VALUE parse_file(VALUE klass, VALUE filename)
 {
-  xmlParserCtxtPtr ctxt = xmlCreateFileParserCtxt(StringValuePtr(filename));
+  xmlParserCtxtPtr ctxt = xmlCreateFileParserCtxt(StringValueCStr(filename));
   return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
 }
 

data/ext/nokogiri/xml_sax_push_parser.c

@@ -59,7 +59,7 @@ 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);
 
   ctx = xmlCreatePushParserCtxt(
       sax,

data/ext/nokogiri/xml_schema.c

@@ -61,7 +61,7 @@ static VALUE validate_file(VALUE self, VALUE rb_filename)
   VALUE errors;
 
   Data_Get_Struct(self, xmlSchema, schema);
-  filename = (const char*)StringValuePtr(rb_filename) ;
+  filename = (const char*)StringValueCStr(rb_filename) ;
 
   errors = rb_ary_new();
 

data/ext/nokogiri/xml_syntax_error.c

@@ -7,10 +7,6 @@ void Nokogiri_error_array_pusher(void * ctx, xmlErrorPtr error)
   rb_ary_push(list,  Nokogiri_wrap_xml_syntax_error(error));
 }
 
-void Nokogiri_error_silencer(void * ctx, xmlErrorPtr error)
-{
-}
-
 void Nokogiri_error_raise(void * ctx, xmlErrorPtr error)
 {
   rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));

data/ext/nokogiri/xml_syntax_error.h

@@ -6,7 +6,6 @@
 void init_xml_syntax_error();
 VALUE Nokogiri_wrap_xml_syntax_error(xmlErrorPtr error);
 void Nokogiri_error_array_pusher(void * ctx, xmlErrorPtr error);
-void Nokogiri_error_silencer(void * ctx, xmlErrorPtr error);
 NORETURN(void Nokogiri_error_raise(void * ctx, xmlErrorPtr error));
 
 extern VALUE cNokogiriXmlSyntaxError;

data/ext/nokogiri/xml_text.c

@@ -19,7 +19,7 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
 
   Data_Get_Struct(document, xmlDoc, doc);
 
-  node = xmlNewText((xmlChar *)StringValuePtr(string));
+  node = xmlNewText((xmlChar *)StringValueCStr(string));
   node->doc = doc->doc;
 
   nokogiri_root_node(node);

data/ext/nokogiri/xml_xpath_context.c

@@ -21,8 +21,8 @@ static VALUE register_ns(VALUE self, VALUE prefix, VALUE uri)
   Data_Get_Struct(self, xmlXPathContext, ctx);
 
   xmlXPathRegisterNs( ctx,
-                      (const xmlChar *)StringValuePtr(prefix),
-                      (const xmlChar *)StringValuePtr(uri)
+                      (const xmlChar *)StringValueCStr(prefix),
+                      (const xmlChar *)StringValueCStr(uri)
   );
   return self;
 }
@@ -39,10 +39,10 @@ static VALUE register_variable(VALUE self, VALUE name, VALUE value)
    xmlXPathObjectPtr xmlValue;
    Data_Get_Struct(self, xmlXPathContext, ctx);
 
-   xmlValue = xmlXPathNewCString(StringValuePtr(value));
+   xmlValue = xmlXPathNewCString(StringValueCStr(value));
 
    xmlXPathRegisterVariable( ctx,
-      (const xmlChar *)StringValuePtr(name),
+      (const xmlChar *)StringValueCStr(name),
       xmlValue
    );
 
@@ -57,7 +57,6 @@ void Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr c
   VALUE node_set = Qnil;
   xmlNodeSetPtr xml_node_set = NULL;
   xmlXPathObjectPtr obj;
-  nokogiriNodeSetTuple *node_set_tuple;
 
   assert(ctx->context->doc);
   assert(DOC_RUBY_OBJECT_TEST(ctx->context->doc));
@@ -109,7 +108,7 @@ void Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr c
     case T_STRING:
       xmlXPathReturnString(
           ctx,
-          xmlCharStrdup(StringValuePtr(result))
+          xmlCharStrdup(StringValueCStr(result))
       );
       break;
     case T_TRUE:
@@ -123,25 +122,23 @@ void Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr c
     case T_ARRAY:
       {
         VALUE args[2];
-	args[0] = doc;
-	args[1] = result;
+        args[0] = doc;
+        args[1] = result;
         node_set = rb_class_new_instance(2, args, cNokogiriXmlNodeSet);
-        Data_Get_Struct(node_set, nokogiriNodeSetTuple, node_set_tuple);
-	xml_node_set = node_set_tuple->node_set;
+        Data_Get_Struct(node_set, xmlNodeSet, xml_node_set);
         xmlXPathReturnNodeSet(ctx, xmlXPathNodeSetMerge(NULL, xml_node_set));
       }
-      break;
+    break;
     case T_DATA:
       if(rb_obj_is_kind_of(result, cNokogiriXmlNodeSet)) {
-        Data_Get_Struct(result, nokogiriNodeSetTuple, node_set_tuple);
-	xml_node_set = node_set_tuple->node_set;
+        Data_Get_Struct(result, xmlNodeSet, xml_node_set);
         /* Copy the node set, otherwise it will get GC'd. */
         xmlXPathReturnNodeSet(ctx, xmlXPathNodeSetMerge(NULL, xml_node_set));
         break;
       }
     default:
       rb_raise(rb_eRuntimeError, "Invalid return type");
-  }
+    }
 }
 
 static void ruby_funcall(xmlXPathParserContextPtr ctx, int nargs)
@@ -203,7 +200,7 @@ static VALUE evaluate(int argc, VALUE *argv, VALUE self)
   if(rb_scan_args(argc, argv, "11", &search_path, &xpath_handler) == 1)
     xpath_handler = Qnil;
 
-  query = (xmlChar *)StringValuePtr(search_path);
+  query = (xmlChar *)StringValueCStr(search_path);
 
   if(Qnil != xpath_handler) {
     /* FIXME: not sure if this is the correct place to shove private data. */
@@ -236,13 +233,8 @@ static VALUE evaluate(int argc, VALUE *argv, VALUE self)
       xmlFree(xpath->stringval);
       break;
     case XPATH_NODESET:
-      if(NULL == xpath->nodesetval) {
-        thing = Nokogiri_wrap_xml_node_set(xmlXPathNodeSetCreate(NULL),
-          DOC_RUBY_OBJECT(ctx->doc));
-      } else {
-        thing = Nokogiri_wrap_xml_node_set(xpath->nodesetval,
-            DOC_RUBY_OBJECT(ctx->doc));
-      }
+      thing = Nokogiri_wrap_xml_node_set(xpath->nodesetval,
+                                         DOC_RUBY_OBJECT(ctx->doc));
       break;
     case XPATH_NUMBER:
       thing = rb_float_new(xpath->floatval);
@@ -251,8 +243,7 @@ static VALUE evaluate(int argc, VALUE *argv, VALUE self)
       thing = xpath->boolval == 1 ? Qtrue : Qfalse;
       break;
     default:
-      thing = Nokogiri_wrap_xml_node_set(xmlXPathNodeSetCreate(NULL),
-        DOC_RUBY_OBJECT(ctx->doc));
+      thing = Nokogiri_wrap_xml_node_set(NULL, DOC_RUBY_OBJECT(ctx->doc));
   }
 
   xmlXPathFreeNodeSetList(xpath);

data/ext/nokogiri/xslt_stylesheet.c

@@ -22,7 +22,7 @@ static void dealloc(nokogiriXsltStylesheetTuple *wrapper)
     NOKOGIRI_DEBUG_START(doc);
     xsltFreeStylesheet(doc); /* commented out for now. */
     NOKOGIRI_DEBUG_END(doc);
-    
+
     free(wrapper);
 }
 
@@ -47,7 +47,7 @@ VALUE Nokogiri_wrap_xslt_stylesheet(xsltStylesheetPtr ss)
 
   self = Data_Make_Struct(cNokogiriXsltStylesheet, nokogiriXsltStylesheetTuple,
                           mark, dealloc, wrapper);
-  
+
   ss->_private = (void *)self;
   wrapper->ss = ss;
   wrapper->func_instances = rb_ary_new();
@@ -122,7 +122,7 @@ static void swallow_superfluous_xml_errors(void * userdata, xmlErrorPtr error, .
  *  returns Nokogiri::XML::Document
  *
  *  Example:
- * 
+ *
  *    doc   = Nokogiri::XML(File.read(ARGV[0]))
  *    xslt  = Nokogiri::XSLT(File.read(ARGV[1]))
  *    puts xslt.transform(doc, ['key', 'value'])
@@ -158,7 +158,7 @@ static VALUE transform(int argc, VALUE* argv, VALUE self)
     params = calloc((size_t)param_len+1, sizeof(char*));
     for (j = 0 ; j < param_len ; j++) {
       VALUE entry = rb_ary_entry(paramobj, j);
-      const char * ptr = StringValuePtr(entry);
+      const char * ptr = StringValueCStr(entry);
       params[j] = ptr;
     }
     params[param_len] = 0 ;
@@ -211,7 +211,7 @@ static void * initFunc(xsltTransformContextPtr ctxt, const xmlChar *uri)
     for(i = 0; i < RARRAY_LEN(methods); i++) {
 	VALUE method_name = rb_obj_as_string(rb_ary_entry(methods, i));
 	xsltRegisterExtFunction(ctxt,
-          (unsigned char *)StringValuePtr(method_name), uri, method_caller);
+          (unsigned char *)StringValueCStr(method_name), uri, method_caller);
     }
 
     Data_Get_Struct(ctxt->style->_private, nokogiriXsltStylesheetTuple,
@@ -245,7 +245,7 @@ static VALUE registr(VALUE self, VALUE uri, VALUE obj)
     if(NIL_P(modules)) rb_raise(rb_eRuntimeError, "wtf! @modules isn't set");
 
     rb_hash_aset(modules, uri, obj);
-    xsltRegisterExtModule((unsigned char *)StringValuePtr(uri), initFunc, shutdownFunc);
+    xsltRegisterExtModule((unsigned char *)StringValueCStr(uri), initFunc, shutdownFunc);
     return self;
 }
 

data/lib/nokogiri.rb

@@ -19,6 +19,9 @@ if defined?(RUBY_ENGINE) && RUBY_ENGINE == "jruby"
     require 'nekohtml.jar'
     require 'nekodtd.jar'
     require 'xercesImpl.jar'
+    require 'serializer.jar'
+    require 'xalan.jar'
+    require 'xml-apis.jar'
   end
 end
 
@@ -111,15 +114,19 @@ module Nokogiri
     def Slop(*args, &block)
       Nokogiri(*args, &block).slop!
     end
+
+    def install_default_aliases
+      # Make sure to support some popular encoding aliases not known by
+      # all iconv implementations.
+      {
+        'Windows-31J' => 'CP932',	# Windows-31J is the IANA registered name of CP932.
+      }.each { |alias_name, name|
+        EncodingHandler.alias(name, alias_name) if EncodingHandler[alias_name].nil?
+      }
+    end
   end
 
-  # Make sure to support some popular encoding aliases not known by
-  # all iconv implementations.
-  {
-    'Windows-31J' => 'CP932',	# Windows-31J is the IANA registered name of CP932.
-  }.each { |alias_name, name|
-    EncodingHandler.alias(name, alias_name) if EncodingHandler[alias_name].nil?
-  }
+  Nokogiri.install_default_aliases
 end
 
 ###

data/lib/nokogiri/css/parser.rb

@@ -8,9 +8,15 @@ require 'racc/parser.rb'
 
 
 require 'nokogiri/css/parser_extras'
+
 module Nokogiri
   module CSS
     class Parser < Racc::Parser
+
+
+def unescape_css_identifier(identifier)
+  identifier.gsub(/\\(?:([^0-9a-fA-F])|([0-9a-fA-F]{1,6})\s?)/){ |m| $1 || [$2.hex].pack('U') }
+end
 ##### State transition tables begin ###
 
 racc_action_table = [
@@ -406,7 +412,7 @@ end
 # reduce 17 omitted
 
 def _reduce_18(val, _values, result)
- result = Node.new(:CLASS_CONDITION, [val[1]]) 
+ result = Node.new(:CLASS_CONDITION, [unescape_css_identifier(val[1])]) 
     result
 end
 
@@ -650,7 +656,7 @@ end
 # reduce 57 omitted
 
 def _reduce_58(val, _values, result)
- result = Node.new(:ID, val) 
+ result = Node.new(:ID, [unescape_css_identifier(val.first)]) 
     result
 end
 

data/lib/nokogiri/css/parser.y

@@ -55,7 +55,7 @@ rule
     | simple_selector
     ;
   class
-    : '.' IDENT { result = Node.new(:CLASS_CONDITION, [val[1]]) }
+    : '.' IDENT { result = Node.new(:CLASS_CONDITION, [unescape_css_identifier(val[1])]) }
     ;
   element_name
     : namespaced_ident
@@ -218,7 +218,7 @@ rule
     | negation
     ;
   attribute_id
-    : HASH { result = Node.new(:ID, val) }
+    : HASH { result = Node.new(:ID, [unescape_css_identifier(val.first)]) }
     ;
   attrib_val_0or1
     : eql_incl_dash IDENT { result = [val.first, val[1]] }
@@ -254,3 +254,8 @@ end
 
 require 'nokogiri/css/parser_extras'
 
+---- inner
+
+def unescape_css_identifier(identifier)
+  identifier.gsub(/\\(?:([^0-9a-fA-F])|([0-9a-fA-F]{1,6})\s?)/){ |m| $1 || [$2.hex].pack('U') }
+end

data/lib/nokogiri/html/document.rb

@@ -173,7 +173,7 @@ module Nokogiri
 
           if string_or_io.respond_to?(:read)
             url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
-            if !encoding
+            unless encoding
               # Libxml2's parser has poor support for encoding
               # detection.  First, it does not recognize the HTML5
               # style meta charset declaration.  Secondly, even if it
@@ -196,7 +196,9 @@ module Nokogiri
           end
 
           # read_memory pukes on empty docs
-          return new if string_or_io.nil? or string_or_io.empty?
+          if string_or_io.nil? or string_or_io.empty?
+            return encoding ? new.tap { |i| i.encoding = encoding } : new
+          end
 
           encoding ||= EncodingReader.detect_encoding(string_or_io)