nokogiri 1.3.3-x86-mingw32 → 1.4.0-x86-mingw32

data/ext/nokogiri/xslt_stylesheet.c

@@ -64,7 +64,7 @@ static VALUE serialize(VALUE self, VALUE xmlobj)
     Data_Get_Struct(xmlobj, xmlDoc, xml);
     Data_Get_Struct(self, xsltStylesheet, ss);
     xsltSaveResultToString(&doc_ptr, &doc_len, xml, ss);
-    rval = NOKOGIRI_STR_NEW(doc_ptr, doc_len, xml->encoding);
+    rval = NOKOGIRI_STR_NEW(doc_ptr, doc_len);
     xmlFree(doc_ptr);
     return rval ;
 }
@@ -94,12 +94,12 @@ static VALUE transform(int argc, VALUE* argv, VALUE self)
     int param_len, j ;
 
     rb_scan_args(argc, argv, "11", &xmldoc, &paramobj);
-    if (paramobj == Qnil) { paramobj = rb_ary_new2(0) ; }
+    if (NIL_P(paramobj)) { paramobj = rb_ary_new2(0) ; }
 
     Data_Get_Struct(xmldoc, xmlDoc, xml);
     Data_Get_Struct(self, xsltStylesheet, ss);
 
-    param_len = NUM2INT(rb_funcall(paramobj, rb_intern("length"), 0));
+    param_len = RARRAY_LEN(paramobj);
     params = calloc((size_t)param_len+1, sizeof(char*));
     for (j = 0 ; j < param_len ; j++) {
       VALUE entry = rb_ary_entry(paramobj, j);

data/lib/nokogiri.rb

@@ -1,11 +1,12 @@
 # -*- coding: utf-8 -*-
 # Modify the PATH on windows so that the external DLLs will get loaded.
+
+require 'rbconfig'
 ENV['PATH'] = [File.expand_path(
   File.join(File.dirname(__FILE__), "..", "ext", "nokogiri")
-), ENV['PATH']].compact.join(';') if RUBY_PLATFORM =~ /(mswin|mingw)/i
+), ENV['PATH']].compact.join(';') if RbConfig::CONFIG['host_os'] =~ /(mswin|mingw)/i
 
 if ENV['NOKOGIRI_FFI'] || RUBY_PLATFORM =~ /java/
-  gem 'ffi', '>=0.4.0' unless RUBY_PLATFORM =~ /java/
   require 'ffi'
   require 'nokogiri/ffi/libxml'
 else
@@ -18,10 +19,9 @@ require 'nokogiri/syntax_error'
 require 'nokogiri/xml'
 require 'nokogiri/xslt'
 require 'nokogiri/html'
-require 'nokogiri/decorators'
+require 'nokogiri/decorators/slop'
 require 'nokogiri/css'
 require 'nokogiri/html/builder'
-require 'nokogiri/hpricot'
 
 # Nokogiri parses and searches XML/HTML very quickly, and also has
 # correctly implemented CSS3 selector support as well as XPath support.

data/lib/nokogiri/css/node.rb

@@ -36,15 +36,7 @@ module Nokogiri
           ]
         )
         matches.each do |match|
-          if match.value[1].value[0].value[0] =~ /^nth-child/
-            tag_name = match.value[0].value.first
-            match.value[0].value = ['*']
-            match.value[1] = Node.new(:COMBINATOR, [
-              match.value[1].value[0],
-              Node.new(:FUNCTION, ['self(', tag_name])
-            ])
-          end
-          if match.value[1].value[0].value[0] =~ /^nth-last-child/
+          if match.value[1].value[0].value[0] =~ /^nth-(last-)?child/
             tag_name = match.value[0].value.first
             match.value[0].value = ['*']
             match.value[1] = Node.new(:COMBINATOR, [

data/lib/nokogiri/css/xpath_visitor.rb

@@ -48,12 +48,6 @@ module Nokogiri
           ']'
       end
 
-      def visit_direct_adjacent_selector node
-        node.value.first.accept(self) +
-          "/following-sibling::*[1]/self::" +
-           node.value.last.accept(self)
-      end
-
       def visit_id node
         node.value.first =~ /^#(.*)$/
         "@id = '#{$1}'"
@@ -122,9 +116,17 @@ module Nokogiri
         "contains(concat(' ', @class, ' '), ' #{node.value.first} ')"
       end
 
-      def visit_combinator node
-        node.value.first.accept(self) + ' and ' +
-        node.value.last.accept(self)
+      {
+        'combinator'                => ' and ',
+        'direct_adjacent_selector'  => "/following-sibling::*[1]/self::",
+        'descendant_selector'       => '//',
+        'child_selector'            => '/',
+      }.each do |k,v|
+        class_eval %{
+          def visit_#{k} node
+            "\#{node.value.first.accept(self)}#{v}\#{node.value.last.accept(self)}"
+          end
+        }
       end
 
       def visit_conditional_selector node
@@ -132,18 +134,6 @@ module Nokogiri
         node.value.last.accept(self) + ']'
       end
 
-      def visit_descendant_selector node
-        node.value.first.accept(self) +
-        '//' +
-        node.value.last.accept(self)
-      end
-
-      def visit_child_selector node
-        node.value.first.accept(self) +
-        '/' +
-        node.value.last.accept(self)
-      end
-
       def visit_element_name node
         node.value.first
       end

data/lib/nokogiri/ffi/html/document.rb

@@ -23,15 +23,6 @@ module Nokogiri
           LibXML.htmlReadMemory(string, string.length, url, encoding, options)
         end
       end
-
-      def meta_encoding=(encoding) # :nodoc:
-        LibXML.htmlSetMetaEncoding(cstruct, encoding)
-        encoding
-      end
-
-      def meta_encoding # :nodoc:
-        LibXML.htmlGetMetaEncoding(cstruct)
-      end
     end
   end
 end

data/lib/nokogiri/ffi/html/sax/parser_context.rb

@@ -0,0 +1,38 @@
+# :stopdoc:
+module Nokogiri
+  module HTML
+    module SAX
+      class ParserContext < Nokogiri::XML::SAX::ParserContext
+        attr_accessor :cstruct
+
+        def self.file filename, encoding
+          ctx = LibXML.htmlCreateFileParserCtxt filename, encoding
+          pc = allocate
+          pc.cstruct = LibXML::XmlParserContext.new ctx
+          pc
+        end
+
+        def self.memory data, encoding
+          raise ArgumentError unless data
+          raise "data cannot be empty" unless data.length > 0
+
+          ctx = LibXML.htmlCreateMemoryParserCtxt data, data.length
+          pc = allocate
+          pc.cstruct = LibXML::XmlParserContext.new ctx
+          if encoding
+            enc = LibXML.xmlParseCharEncoding(encoding)
+            if enc != LibXML::XML_CHAR_ENCODING_ERROR
+              LibXML.xmlSwitchEncoding(ctx, enc)
+            end
+          end
+          pc
+        end
+
+        def parse_with sax_handler, type = :html
+          super
+        end
+      end
+    end
+  end
+end
+# :startdoc:

data/lib/nokogiri/ffi/io_callbacks.rb

@@ -1,6 +1,7 @@
+# :stopdoc:
 module Nokogiri
-  module IoCallbacks # :nodoc:
-    
+  module IoCallbacks
+
     class << self
 
       def plain_old_reader(io)
@@ -38,3 +39,4 @@ module Nokogiri
 
   end
 end
+# :startdoc:

data/lib/nokogiri/ffi/libxml.rb

@@ -1,8 +1,13 @@
-
+# :stopdoc:
 module Nokogiri
-  module LibXML # :nodoc: all
+  module LibXML
     extend FFI::Library
-    ffi_lib 'xml2', 'xslt', 'exslt'
+    if RUBY_PLATFORM =~ /java/ && java.lang.System.getProperty('os.name') =~ /windows/i
+      raise(RuntimeError, "Nokogiri requires JRuby 1.4.0RC1 or later on Windows") if JRUBY_VERSION < "1.4.0RC1"
+      ffi_lib 'libxml2', 'libxslt', 'libexslt', 'msvcrt'
+    else
+      ffi_lib 'xml2', 'xslt', 'exslt'
+    end
 
     # globals.c
     attach_function :__xmlParserVersion, [], :pointer
@@ -28,7 +33,9 @@ else
 end
 
 module Nokogiri
-  module LibXML # :nodoc:
+  module LibXML
+    XML_CHAR_ENCODING_ERROR = -1
+
     # useful callback signatures
     callback :syntax_error_handler, [:pointer, :pointer], :void
     callback :generic_error_handler, [:pointer, :string], :void
@@ -50,10 +57,6 @@ module Nokogiri
     callback :start_element_ns_sax2_func, [:pointer, :pointer, :pointer, :pointer, :int, :pointer, :int, :int, :pointer], :void
     callback :end_element_ns_sax2_func, [:pointer, :pointer, :pointer, :pointer], :void
 
-    # libc
-    attach_function :calloc, [:int, :int], :pointer
-    attach_function :free, [:pointer], :void
-
     # HTMLparser.c
     attach_function :htmlReadMemory, [:string, :int, :string, :string, :int], :pointer
     attach_function :htmlReadIO, [:io_read_callback, :io_close_callback, :pointer, :string, :string, :int], :pointer
@@ -62,6 +65,9 @@ module Nokogiri
     attach_function :htmlEntityLookup, [:string], :pointer
     attach_function :htmlSAXParseFile, [:string, :pointer, :pointer, :pointer], :pointer # second arg 'encoding' should be a string, but we assign it as a pointer elsewhere
     attach_function :htmlSAXParseDoc, [:pointer, :pointer, :pointer, :pointer], :pointer # second arg 'encoding' should be a string, but we assign it as a pointer elsewhere
+    attach_function :htmlCreateMemoryParserCtxt, [:pointer, :int], :pointer
+    attach_function :htmlCreateFileParserCtxt, [:pointer, :pointer], :pointer
+    attach_function :htmlParseDocument, [:pointer], :int
 
     # HTMLtree.c
     attach_function :htmlDocDumpMemory, [:pointer, :pointer, :pointer], :void
@@ -74,6 +80,8 @@ module Nokogiri
     attach_function :xmlInitParser, [], :void
     attach_function :xmlReadIO, [:io_read_callback, :io_close_callback, :pointer, :string, :string, :int], :pointer
     attach_function :xmlCreateIOParserCtxt, [:pointer, :pointer, :io_read_callback, :io_close_callback, :pointer, :int], :pointer
+    attach_function :xmlCreateMemoryParserCtxt, [:pointer, :int], :pointer
+    attach_function :xmlCreateFileParserCtxt, [:pointer], :pointer
     attach_function :xmlSAXUserParseMemory, [:pointer, :pointer, :string, :int], :int
     attach_function :xmlSAXUserParseFile, [:pointer, :pointer, :string], :int
     attach_function :xmlParseDocument, [:pointer], :int
@@ -126,6 +134,8 @@ module Nokogiri
     attach_function :xmlSplitQName2, [:string, :pointer], :pointer # returns char* that must be freed
     attach_function :xmlNewDocProp, [:pointer, :string, :string], :pointer
     attach_function :xmlFreePropList, [:pointer], :void
+    attach_function :xmlCreateIntSubset, [:pointer] * 4, :pointer
+    attach_function :xmlNewDtd, [:pointer] * 4, :pointer
 
     # valid.c
     attach_function :xmlNewValidCtxt, [], :pointer
@@ -200,6 +210,7 @@ module Nokogiri
     attach_function :xmlResetLastError, [], :void
     attach_function :xmlCopyError, [:pointer, :pointer], :int
     attach_function :xmlGetLastError, [], :pointer
+    attach_function :xmlCtxtGetLastError, [:pointer], :pointer
 
     # hash.c
     attach_function :xmlHashScan, [:pointer, :hash_copier_callback, :pointer], :void
@@ -210,6 +221,8 @@ module Nokogiri
     attach_function :xmlTextReaderGetAttributeNo, [:pointer, :int], :pointer # returns char* that must be freed
     attach_function :xmlTextReaderLookupNamespace, [:pointer, :string], :pointer # returns char* that must be freed
     attach_function :xmlTextReaderRead, [:pointer], :int
+    attach_function :xmlTextReaderReadInnerXml, [:pointer], :pointer
+    attach_function :xmlTextReaderReadOuterXml, [:pointer], :pointer
     attach_function :xmlTextReaderAttributeCount, [:pointer], :int
     attach_function :xmlTextReaderCurrentNode, [:pointer], :pointer
     attach_function :xmlTextReaderExpand, [:pointer], :pointer
@@ -226,6 +239,7 @@ module Nokogiri
     attach_function :xmlTextReaderHasValue, [:pointer], :int
     attach_function :xmlFreeTextReader, [:pointer], :void
     attach_function :xmlReaderForIO, [:io_read_callback, :io_close_callback, :pointer, :string, :string, :int], :pointer
+    attach_function :xmlTextReaderNodeType, [:pointer], :int
 
     # xslt.c
     attach_function :xsltParseStylesheetDoc, [:pointer], :pointer
@@ -259,6 +273,13 @@ module Nokogiri
     attach_function :xmlRelaxNGFreeParserCtxt, [:pointer], :void
     attach_function :xmlRelaxNGNewDocParserCtxt, [:pointer], :pointer
 
+    # libc
+    attach_function :calloc, [:int, :int], :pointer
+    attach_function :free, [:pointer], :void
+
+    attach_function :xmlParseCharEncoding, [:string], :int
+    attach_function :xmlSwitchEncoding, [:pointer, :int], :void
+
     # helpers
     def self.pointer_offset(n)
       n * FFI.type_size(:pointer) # byte offset of nth pointer in an array of pointers
@@ -266,6 +287,8 @@ module Nokogiri
   end
 end
 
+# :startdoc:
+
 require 'nokogiri/syntax_error'
 require 'nokogiri/xml/syntax_error'
 
@@ -289,13 +312,23 @@ require 'nokogiri/xml/syntax_error'
   "structs/xml_text_reader",
   "structs/xml_sax_handler",
   "structs/xml_sax_push_parser_context",
+  "structs/xml_enumeration",
   "structs/html_elem_desc",
   "structs/html_entity_desc",
   "structs/xslt_stylesheet",
+  "structs/xml_parser_context",
+  "structs/xml_attribute",
+  "structs/xml_element",
+  "structs/xml_entity",
+  "structs/xml_element_content",
   "xml/node",
   "xml/namespace",
   "xml/dtd",
   "xml/attr",
+  "xml/attribute_decl",
+  "xml/element_decl",
+  "xml/element_content",
+  "xml/entity_decl",
   "xml/document",
   "xml/document_fragment",
   "xml/schema",
@@ -310,13 +343,14 @@ require 'nokogiri/xml/syntax_error'
   "xml/syntax_error",
   "xml/reader",
   "xml/entity_reference",
+  "xml/sax/parser_context",
   "xml/sax/parser",
   "xml/sax/push_parser",
   "html/document",
   "html/element_description",
   "html/entity_lookup",
-  "html/sax/parser",
+  "html/sax/parser_context",
   "xslt/stylesheet",
 ].each do |file|
-  require File.join(File.dirname(__FILE__), file)
+  require "nokogiri/ffi/#{file}"
 end

data/lib/nokogiri/ffi/structs/common_node.rb

@@ -11,7 +11,7 @@ module Nokogiri
         self[:_private] != 0 ? ObjectSpace._id2ref(self[:_private]) : nil
       end
 
-      def ruby_node=(object)
+      def ruby_node= object
         self[:_private] = object.object_id
       end
 

data/lib/nokogiri/ffi/structs/xml_attribute.rb

@@ -0,0 +1,27 @@
+module Nokogiri
+  module LibXML # :nodoc:
+    class XmlAttribute < FFI::Struct # :nodoc:
+      include CommonNode
+
+      layout(
+        :_private,      :long,
+        :type,          :int,
+        :name,          :string,
+        :children,      :pointer,
+        :last,          :pointer,
+        :parent,        :pointer,
+        :next,          :pointer,
+        :prev,          :pointer,
+        :doc,           :pointer,
+
+        :nexth,         :pointer,
+        :atype,         :int,
+        :def,           :int,
+        :default_value, :string,
+        :tree,          :pointer,
+        :prefix,        :string,
+        :elem,          :string
+        )
+    end
+  end
+end

data/lib/nokogiri/ffi/structs/xml_dtd.rb

@@ -16,7 +16,9 @@ module Nokogiri
         :notations,     :pointer,
         :elements,      :pointer,
         :attributes,    :pointer,
-        :entities,      :pointer
+        :entities,      :pointer,
+        :external_id,   :string,
+        :system_id,     :string
         )
 
       include CommonNode

data/lib/nokogiri/ffi/structs/xml_element.rb

@@ -0,0 +1,26 @@
+module Nokogiri
+  # :stopdoc:
+  module LibXML
+    class XmlElement < FFI::Struct
+      include CommonNode
+
+      layout(
+        :_private,      :long,
+        :type,          :int,
+        :name,          :string,
+        :children,      :pointer,
+        :last,          :pointer,
+        :parent,        :pointer,
+        :next,          :pointer,
+        :prev,          :pointer,
+        :doc,           :pointer,
+
+        :etype,         :int,
+        :content,       :pointer,
+        :properties,    :pointer,
+        :prefix,        :string
+      )
+    end
+  end
+  # :startdoc:
+end

data/lib/nokogiri/ffi/structs/xml_element_content.rb

@@ -0,0 +1,17 @@
+# :stopdoc:
+module Nokogiri
+  module LibXML
+    class XmlElementContent < FFI::Struct
+      layout(
+        :type,  :int,
+        :ocur,  :int, # This is misspelled in the header file
+        :name,  :string,
+        :c1,    :pointer,
+        :c2,    :pointer,
+        :parent,:pointer,
+        :prefix,:string
+      )
+    end
+  end
+end
+# :startdoc: