nokogiri 1.6.8.rc1 → 1.6.8.rc2

data/ext/nokogiri/xml_node_set.c

@@ -321,25 +321,17 @@ static VALUE slice(int argc, VALUE *argv, VALUE self)
 static VALUE to_array(VALUE self, VALUE rb_node)
 {
   xmlNodeSetPtr node_set ;
-  VALUE *elts;
   VALUE list;
   int i;
 
   Data_Get_Struct(self, xmlNodeSet, node_set);
 
-  elts = (VALUE *)calloc((size_t)(node_set->nodeNr), sizeof(VALUE));
+  list = rb_ary_new2(node_set->nodeNr);
   for(i = 0; i < node_set->nodeNr; i++) {
-    elts[i] = Nokogiri_wrap_xml_node_set_node(node_set->nodeTab[i], self);
-    rb_gc_register_address(&elts[i]);
+    VALUE elt = Nokogiri_wrap_xml_node_set_node(node_set->nodeTab[i], self);
+    rb_ary_push( list, elt );
   }
 
-  list = rb_ary_new4((long)node_set->nodeNr, elts);
-
-  for(i = 0; i < node_set->nodeNr; i++) {
-    rb_gc_unregister_address(&elts[i]);
-  }
-  free(elts);
-
   return list;
 }
 

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

@@ -218,12 +218,12 @@ static VALUE reader_attribute(VALUE self, VALUE name)
   if(NIL_P(name)) return Qnil;
   name = StringValue(name) ;
 
-  value = xmlTextReaderGetAttribute(reader, (xmlChar*)StringValuePtr(name));
+  value = xmlTextReaderGetAttribute(reader, (xmlChar*)StringValueCStr(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);
+    xmlChar *localname = xmlSplitQName2((xmlChar*)StringValueCStr(name), &prefix);
     if (localname != NULL) {
       value = xmlTextReaderLookupNamespace(reader, localname);
       xmlFree(localname) ;
@@ -546,8 +546,8 @@ 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 +590,8 @@ 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_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_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
    );
 
@@ -108,7 +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:
@@ -200,7 +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. */

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/version.rb

@@ -1,6 +1,6 @@
 module Nokogiri
   # The version of Nokogiri you are using
-  VERSION = '1.6.8.rc1'
+  VERSION = '1.6.8.rc2'
 
   class VersionInfo # :nodoc:
     def jruby?

data/lib/nokogiri/xml/node.rb

@@ -1,3 +1,4 @@
+# encoding: UTF-8
 require 'stringio'
 require 'nokogiri/xml/node/save_options'
 
@@ -331,13 +332,13 @@ module Nokogiri
       ###
       # Get the attribute values for this Node.
       def values
-        attribute_nodes.map { |node| node.value }
+        attribute_nodes.map(&:value)
       end
 
       ###
       # Get the attribute names for this Node.
       def keys
-        attribute_nodes.map { |node| node.node_name }
+        attribute_nodes.map(&:node_name)
       end
 
       ###
@@ -439,12 +440,6 @@ module Nokogiri
       def namespaces
         Hash[namespace_scopes.map { |nd|
           key = ['xmlns', nd.prefix].compact.join(':')
-          if document.encoding
-            begin
-              key.force_encoding document.encoding
-            rescue ArgumentError
-            end
-          end
           [key, nd.href]
         }]
       end
@@ -546,9 +541,10 @@ module Nokogiri
         return NodeSet.new(document, parents) unless selector
 
         root = parents.last
+        search_results = root.search(selector)
 
         NodeSet.new(document, parents.find_all { |parent|
-          root.search(selector).include?(parent)
+          search_results.include?(parent)
         })
       end
 

data/lib/nokogiri/xml/node_set.rb

@@ -191,7 +191,7 @@ module Nokogiri
       ###
       # Get the inner text of all contained Node objects
       def inner_text
-        collect{|j| j.inner_text}.join('')
+        collect(&:inner_text).join('')
       end
       alias :text :inner_text
 
@@ -215,7 +215,7 @@ module Nokogiri
       ###
       # Convert this NodeSet to a string.
       def to_s
-        map { |x| x.to_s }.join
+        map(&:to_s).join
       end
 
       ###
@@ -296,7 +296,7 @@ module Nokogiri
       ###
       # Return a nicely formated string representation
       def inspect
-        "[#{map { |c| c.inspect }.join ', '}]"
+        "[#{map(&:inspect).join ', '}]"
       end
 
       alias :+ :|

data/test/html/test_document.rb

@@ -97,7 +97,7 @@ module Nokogiri
         rescue Exception => e
           skip("This test needs the internet. Skips if no internet available. (#{e})")
         end
-        doc = Nokogiri::HTML html ,"http:/foobar.foobar/"
+        doc = Nokogiri::HTML html ,"http:/foobar.foobar/", 'UTF-8'
         refute_empty doc.to_s, "Document should not be empty"
       end
 
@@ -422,7 +422,7 @@ eohtml
         eohtml
         set = html.css('p, a')
         assert_equal(2, set.length)
-        assert_equal ['a tag', 'p tag'].sort, set.map { |x| x.content }.sort
+        assert_equal ['a tag', 'p tag'].sort, set.map(&:content).sort
       end
 
       def test_inner_text
@@ -600,7 +600,7 @@ eohtml
         eohtml
         list = doc.css('.red')
         assert_equal 2, list.length
-        assert_equal %w{ RED RED }, list.map { |x| x.text }
+        assert_equal %w{ RED RED }, list.map(&:text)
       end
 
       def test_parse_can_take_io

data/test/html/test_document_encoding.rb

@@ -107,16 +107,16 @@ module Nokogiri
       def test_document_xhtml_enc
         [ENCODING_XHTML_FILE, ENCODING_HTML_FILE].each { |file|
           doc_from_string_enc = Nokogiri::HTML(binread(file), nil, 'Shift_JIS')
-          ary_from_string_enc = doc_from_string_enc.xpath('//p/text()').map { |text| text.text }
+          ary_from_string_enc = doc_from_string_enc.xpath('//p/text()').map(&:text)
 
           doc_from_string = Nokogiri::HTML(binread(file))
-          ary_from_string = doc_from_string.xpath('//p/text()').map { |text| text.text }
+          ary_from_string = doc_from_string.xpath('//p/text()').map(&:text)
 
           doc_from_file_enc = Nokogiri::HTML(binopen(file), nil, 'Shift_JIS')
-          ary_from_file_enc = doc_from_file_enc.xpath('//p/text()').map { |text| text.text }
+          ary_from_file_enc = doc_from_file_enc.xpath('//p/text()').map(&:text)
 
           doc_from_file = Nokogiri::HTML(binopen(file))
-          ary_from_file = doc_from_file.xpath('//p/text()').map { |text| text.text }
+          ary_from_file = doc_from_file.xpath('//p/text()').map(&:text)
 
           title = 'たこ焼き仮面'
 

data/test/html/test_node_encoding.rb

@@ -4,6 +4,48 @@ require "helper"
 module Nokogiri
   module HTML
     class TestNodeEncoding < Nokogiri::TestCase
+      def setup
+        super
+        @html = Nokogiri::HTML(File.open(NICH_FILE, "rb"))
+      end
+
+      def test_get_attribute
+        node = @html.css('a').first
+        assert_equal 'UTF-8', node['href'].encoding.name
+      end
+
+      def test_text_encoding_is_utf_8
+        assert_equal 'UTF-8', @html.text.encoding.name
+      end
+
+      def test_serialize_encoding_html
+        assert_equal @html.encoding.downcase,
+          @html.serialize.encoding.name.downcase
+
+        @doc = Nokogiri::HTML(@html.serialize)
+        assert_equal @html.serialize, @doc.serialize
+      end
+
+      def test_encode_special_chars
+        foo = @html.css('a').first.encode_special_chars('foo')
+        assert_equal 'UTF-8', foo.encoding.name
+      end
+
+      def test_content
+        node = @html.css('a').first
+        assert_equal 'UTF-8', node.content.encoding.name
+      end
+
+      def test_name
+        node = @html.css('a').first
+        assert_equal 'UTF-8', node.name.encoding.name
+      end
+
+      def test_path
+        node = @html.css('a').first
+        assert_equal 'UTF-8', node.path.encoding.name
+      end
+
       def test_inner_html
         doc = Nokogiri::HTML File.open(SHIFT_JIS_HTML, 'rb')
 
@@ -20,6 +62,24 @@ module Nokogiri
         contents = doc.at('h2').inner_html
         assert_match hello, contents
       end
+
+      def test_encoding_GH_1113
+        doc = Nokogiri::HTML::Document.new
+        hex = '<p>&#x1f340;</p>'
+        decimal = '<p>&#127808;</p>'
+        encoded = '<p>🍀</p>'
+
+        doc.encoding = 'UTF-8'
+        [hex, decimal, encoded].each do |document|
+          assert_equal encoded, doc.fragment(document).to_s
+        end
+
+        doc.encoding = 'US-ASCII'
+        expected = Nokogiri.jruby? ? hex : decimal
+        [hex, decimal].each do |document|
+          assert_equal expected, doc.fragment(document).to_s
+        end
+      end
     end
   end
 end

data/test/test_css_cache.rb

@@ -6,7 +6,7 @@ class TestCssCache < Nokogiri::TestCase
     super
     @css = "a1 > b2 > c3"
     @parse_result = Nokogiri::CSS.parse(@css)
-    @to_xpath_result = @parse_result.map {|ast| ast.to_xpath}
+    @to_xpath_result = @parse_result.map(&:to_xpath)
     Nokogiri::CSS::Parser.class_eval do
       class << @cache
         alias :old_bracket :[]

data/test/test_reader.rb

@@ -85,7 +85,7 @@ class TestReader < Nokogiri::TestCase
     reader = Nokogiri::XML::Reader.from_io(io)
     assert_equal false, reader.default?
     assert_equal [false, false, false, false, false, false, false],
-      reader.map { |x| x.default? }
+      reader.map(&:default?)
   end
 
   def test_io
@@ -93,7 +93,7 @@ class TestReader < Nokogiri::TestCase
     reader = Nokogiri::XML::Reader(io)
     assert_equal false, reader.default?
     assert_equal [false, false, false, false, false, false, false],
-      reader.map { |x| x.default? }
+      reader.map(&:default?)
   end
 
   def test_string_io
@@ -105,7 +105,7 @@ class TestReader < Nokogiri::TestCase
     reader = Nokogiri::XML::Reader(io)
     assert_equal false, reader.default?
     assert_equal [false, false, false, false, false, false, false],
-      reader.map { |x| x.default? }
+      reader.map(&:default?)
   end
   
   class ReallyBadIO
@@ -156,7 +156,7 @@ class TestReader < Nokogiri::TestCase
     eoxml
     assert_equal false, reader.default?
     assert_equal [false, false, false, false, false, false, false],
-      reader.map { |x| x.default? }
+      reader.map(&:default?)
   end
 
   def test_value?
@@ -167,7 +167,7 @@ class TestReader < Nokogiri::TestCase
     eoxml
     assert_equal false, reader.value?
     assert_equal [false, true, false, true, false, true, false],
-      reader.map { |x| x.value? }
+      reader.map(&:value?)
   end
 
   def test_read_error_document
@@ -191,7 +191,7 @@ class TestReader < Nokogiri::TestCase
     eoxml
     assert_equal false, reader.attributes?
     assert_equal [true, false, true, false, true, false, true],
-      reader.map { |x| x.attributes? }
+      reader.map(&:attributes?)
   end
 
   def test_attributes
@@ -208,7 +208,7 @@ class TestReader < Nokogiri::TestCase
                   {}, {"awesome"=>"true"}, {}, {"awesome"=>"true"}, {},
                   {'xmlns:tenderlove'=>'http://tenderlovemaking.com/',
                    'xmlns'=>'http://mothership.connection.com/'}],
-      reader.map { |x| x.attributes }
+      reader.map(&:attributes)
   end
 
   def test_attribute_roundtrip
@@ -257,7 +257,7 @@ class TestReader < Nokogiri::TestCase
     </x>
     eoxml
     assert_equal 0, reader.attribute_count
-    assert_equal [1, 0, 1, 0, 0, 0, 0], reader.map { |x| x.attribute_count }
+    assert_equal [1, 0, 1, 0, 0, 0, 0], reader.map(&:attribute_count)
   end
 
   def test_depth
@@ -267,7 +267,7 @@ class TestReader < Nokogiri::TestCase
     </x>
     eoxml
     assert_equal 0, reader.depth
-    assert_equal [0, 1, 1, 2, 1, 1, 0], reader.map { |x| x.depth }
+    assert_equal [0, 1, 1, 2, 1, 1, 0], reader.map(&:depth)
   end
 
   def test_encoding
@@ -278,7 +278,7 @@ class TestReader < Nokogiri::TestCase
     </awesome>
     eoxml
     reader = Nokogiri::XML::Reader.from_memory(string, nil, 'UTF-8')
-    assert_equal ['UTF-8'], reader.map { |x| x.encoding }.uniq
+    assert_equal ['UTF-8'], reader.map(&:encoding).uniq
   end
 
   def test_xml_version
@@ -288,7 +288,7 @@ class TestReader < Nokogiri::TestCase
     </x>
     eoxml
     assert_nil reader.xml_version
-    assert_equal ['1.0'], reader.map { |x| x.xml_version }.uniq
+    assert_equal ['1.0'], reader.map(&:xml_version).uniq
   end
 
   def test_lang
@@ -300,7 +300,7 @@ class TestReader < Nokogiri::TestCase
     eoxml
     assert_nil reader.lang
     assert_equal [nil, nil, "en", "en", "en", nil, "ja", "ja", "ja", nil, nil],
-      reader.map { |x| x.lang }
+      reader.map(&:lang)
   end
 
   def test_value
@@ -311,7 +311,7 @@ class TestReader < Nokogiri::TestCase
     eoxml
     assert_nil reader.value
     assert_equal [nil, "\n      ", nil, "snuggles!", nil, "\n    ", nil],
-      reader.map { |x| x.value }
+      reader.map(&:value)
   end
 
   def test_prefix
@@ -322,7 +322,7 @@ class TestReader < Nokogiri::TestCase
     eoxml
     assert_nil reader.prefix
     assert_equal [nil, nil, "edi", nil, "edi", nil, nil],
-      reader.map { |n| n.prefix }
+      reader.map(&:prefix)
   end
 
   def test_node_type
@@ -332,7 +332,7 @@ class TestReader < Nokogiri::TestCase
     </x>
     eoxml
     assert_equal 0, reader.node_type
-    assert_equal [1, 14, 1, 3, 15, 14, 15], reader.map { |n| n.node_type }
+    assert_equal [1, 14, 1, 3, 15, 14, 15], reader.map(&:node_type)
   end
 
   def test_inner_xml
@@ -383,7 +383,7 @@ class TestReader < Nokogiri::TestCase
                   "http://ecommerce.example.org/schema",
                   nil,
                   nil],
-                  reader.map { |n| n.namespace_uri })
+                  reader.map(&:namespace_uri))
   end
 
   def test_namespaced_attributes
@@ -413,7 +413,7 @@ class TestReader < Nokogiri::TestCase
     eoxml
     assert_nil reader.local_name
     assert_equal(["x", "#text", "foo", "#text", "foo", "#text", "x"],
-                 reader.map { |n| n.local_name })
+                 reader.map(&:local_name))
   end
 
   def test_name
@@ -424,7 +424,7 @@ class TestReader < Nokogiri::TestCase
     eoxml
     assert_nil reader.name
     assert_equal(["x", "#text", "edi:foo", "#text", "edi:foo", "#text", "x"],
-                 reader.map { |n| n.name })
+                 reader.map(&:name))
   end
 
   def test_base_uri
@@ -452,7 +452,7 @@ class TestReader < Nokogiri::TestCase
                   "http://base.example.org/base/relative",
                   "http://base.example.org/base/",
                   "http://base.example.org/base/"],
-                  reader.map {|n| n.base_uri })
+                  reader.map(&:base_uri))
   end
 
   def test_xlink_href_without_base_uri