nokogiri 1.2.3 → 1.3.0

data/ext/nokogiri/xml_text.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_XML_TEXT
 #define NOKOGIRI_XML_TEXT
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xml_text();
 

data/ext/nokogiri/xml_xpath.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_XML_XPATH
 #define NOKOGIRI_XML_XPATH
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xml_xpath();
 VALUE Nokogiri_wrap_xml_xpath(xmlXPathObjectPtr xpath);

data/ext/nokogiri/xml_xpath_context.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_XML_XPATH_CONTEXT
 #define NOKOGIRI_XML_XPATH_CONTEXT
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xml_xpath_context();
 

data/ext/nokogiri/xslt_stylesheet.c

@@ -12,6 +12,18 @@ static void dealloc(xsltStylesheetPtr doc)
     NOKOGIRI_DEBUG_END(doc);
 }
 
+static void xslt_generic_error_handler(void * ctx, const char *msg, ...)
+{
+  char * message;
+
+  va_list args;
+  va_start(args, msg);
+  vasprintf(&message, msg, args);
+  va_end(args);
+
+  rb_raise(rb_eRuntimeError, message);
+}
+
 /*
  * call-seq:
  *   parse_stylesheet_doc(document)
@@ -24,7 +36,13 @@ static VALUE parse_stylesheet_doc(VALUE klass, VALUE xmldocobj)
     xsltStylesheetPtr ss ;
     Data_Get_Struct(xmldocobj, xmlDoc, xml);
     exsltRegisterAll();
+
+    xsltSetGenericErrorFunc(NULL, xslt_generic_error_handler);
+
     ss = xsltParseStylesheetDoc(xmlCopyDoc(xml, 1)); /* 1 => recursive */
+
+    xsltSetGenericErrorFunc(NULL, NULL);
+
     return Data_Wrap_Struct(klass, NULL, dealloc, ss);
 }
 
@@ -50,13 +68,21 @@ static VALUE serialize(VALUE self, VALUE xmlobj)
     xmlFree(doc_ptr);
     return rval ;
 }
+
 /*
  *  call-seq:
- *    transform(document, params)
+ *    transform(document, params = [])
  *
  *  Apply an XSLT stylesheet to an XML::Document.
  *  +params+ is an array of strings used as XSLT parameters.
  *  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'])
+ *
  */
 static VALUE transform(int argc, VALUE* argv, VALUE self)
 {
@@ -85,21 +111,9 @@ static VALUE transform(int argc, VALUE* argv, VALUE self)
     result = xsltApplyStylesheet(ss, xml, params);
     free(params);
 
-    return Nokogiri_wrap_xml_document(0, result) ;
-}
+    if (!result) rb_raise(rb_eRuntimeError, "could not perform xslt transform on document");
 
-/*
- *  call-seq:
- *    apply_to(document, params)
- *
- *  Apply an XSLT stylesheet to an XML::Document.
- *  +params+ is an array of strings used as XSLT parameters.
- *  returns serialized document
- */
-static VALUE apply_to(int argc, VALUE* argv, VALUE self)
-{
-  return rb_funcall(self, rb_intern("serialize"), 1, 
-                    transform(argc, argv, self));
+    return Nokogiri_wrap_xml_document(0, result) ;
 }
 
 VALUE cNokogiriXsltStylesheet ;
@@ -113,6 +127,5 @@ void init_xslt_stylesheet()
     
   rb_define_singleton_method(klass, "parse_stylesheet_doc", parse_stylesheet_doc, 1);
   rb_define_method(klass, "serialize", serialize, 1);
-  rb_define_method(klass, "apply_to", apply_to, -1);
   rb_define_method(klass, "transform", transform, -1);
 }

data/ext/nokogiri/xslt_stylesheet.h

@@ -1,7 +1,7 @@
 #ifndef NOKOGIRI_XSLT_STYLESHEET
 #define NOKOGIRI_XSLT_STYLESHEET
 
-#include <native.h>
+#include <nokogiri.h>
 
 void init_xslt_stylesheet();
 

data/lib/action-nokogiri.rb

@@ -12,8 +12,11 @@ require 'nokogiri'
 #  end
 #
 module ActionController
-  module TestResponseBehavior
+  module TestResponseBehavior # :nodoc:
 
+    ###
+    # Get your response as a Nokogiri::XML::Document using the
+    # Nokogiri.HTML parser
     def html(flavor=nil)
       if flavor == :hpricot
         @_nokogiri_html_hpricot ||= Nokogiri::Hpricot(body)
@@ -22,6 +25,9 @@ module ActionController
       end
     end
 
+    ###
+    # Get your response as a Nokogiri::XML::Document using the
+    # Nokogiri.XML parser
     def xml
       @_nokogiri_xml ||= Nokogiri::XML(body)
     end

data/lib/nokogiri.rb

@@ -1,11 +1,19 @@
+# -*- coding: utf-8 -*-
 # Modify the PATH on windows so that the external DLLs will get loaded.
 ENV['PATH'] = [File.expand_path(
   File.join(File.dirname(__FILE__), "..", "ext", "nokogiri")
-), ENV['PATH']].compact.join(';') if RUBY_PLATFORM =~ /mswin/i
+), ENV['PATH']].compact.join(';') if RUBY_PLATFORM =~ /(mswin|mingw)/i
 
-require 'nokogiri/native' unless RUBY_PLATFORM =~ /java/
+if ENV['NOKOGIRI_FFI'] || RUBY_PLATFORM =~ /java/
+  gem 'ffi', '>=0.4.0' unless RUBY_PLATFORM =~ /java/
+  require 'ffi'
+  require 'nokogiri/ffi/libxml'
+else
+  require 'nokogiri/nokogiri'
+end
 
 require 'nokogiri/version'
+require 'nokogiri/version_warning'
 require 'nokogiri/syntax_error'
 require 'nokogiri/xml'
 require 'nokogiri/xslt'
@@ -47,14 +55,16 @@ module Nokogiri
     def parse string, url = nil, encoding = nil, options = nil
       doc =
         if string =~ /^\s*<[^Hh>]*html/i # Probably html
-          Nokogiri::HTML.parse(string, url, encoding, options || 2145)
+          Nokogiri::HTML::Document.parse(string, url, encoding, options || 2145)
         else
-          Nokogiri::XML.parse(string, url, encoding, options || 2159)
+          Nokogiri::XML::Document.parse(string, url, encoding, options || 2159)
         end
       yield doc if block_given?
       doc
     end
 
+    ###
+    # Create a new Nokogiri::XML::DocumentFragment
     def make input = nil, opts = {}, &blk
       if input
         Nokogiri::HTML.fragment(input).children.first
@@ -62,7 +72,7 @@ module Nokogiri
         Nokogiri(&blk)
       end
     end
-    
+
     ###
     # Parse a document and add the Slop decorator.  The Slop decorator
     # implements method_missing such that methods may be used instead of CSS
@@ -84,6 +94,12 @@ module Nokogiri
   end
 end
 
+###
+# Parser a document contained in +args+.  Nokogiri will try to guess what
+# type of document you are attempting to parse.  For more information, see
+# Nokogiri.parse
+#
+# To specify the type of document, use Nokogiri.XML or Nokogiri.HTML.
 def Nokogiri(*args, &block)
   if block_given?
     builder = Nokogiri::HTML::Builder.new(&block)

data/lib/nokogiri/css/generated_parser.rb

@@ -147,13 +147,13 @@ racc_reduce_table = [
   2, 43, :_reduce_53,
   2, 43, :_reduce_54,
   0, 43, :_reduce_none,
-  1, 48, :_reduce_none,
-  1, 48, :_reduce_none,
-  1, 48, :_reduce_none,
-  1, 48, :_reduce_none,
-  1, 48, :_reduce_none,
-  1, 48, :_reduce_none,
-  1, 48, :_reduce_none,
+  1, 48, :_reduce_56,
+  1, 48, :_reduce_57,
+  1, 48, :_reduce_58,
+  1, 48, :_reduce_59,
+  1, 48, :_reduce_60,
+  1, 48, :_reduce_61,
+  1, 48, :_reduce_62,
   3, 38, :_reduce_63,
   1, 49, :_reduce_none ]
 
@@ -680,19 +680,54 @@ module_eval(<<'.,.,', 'parser.y', 203)
 
 # reduce 55 omitted
 
-# reduce 56 omitted
+module_eval(<<'.,.,', 'parser.y', 207)
+  def _reduce_56(val, _values, result)
+     result = :equal 
+    result
+  end
+.,.,
 
-# reduce 57 omitted
+module_eval(<<'.,.,', 'parser.y', 208)
+  def _reduce_57(val, _values, result)
+     result = :prefix_match 
+    result
+  end
+.,.,
 
-# reduce 58 omitted
+module_eval(<<'.,.,', 'parser.y', 209)
+  def _reduce_58(val, _values, result)
+     result = :suffix_match 
+    result
+  end
+.,.,
 
-# reduce 59 omitted
+module_eval(<<'.,.,', 'parser.y', 210)
+  def _reduce_59(val, _values, result)
+     result = :substring_match 
+    result
+  end
+.,.,
 
-# reduce 60 omitted
+module_eval(<<'.,.,', 'parser.y', 211)
+  def _reduce_60(val, _values, result)
+     result = :not_equal 
+    result
+  end
+.,.,
 
-# reduce 61 omitted
+module_eval(<<'.,.,', 'parser.y', 212)
+  def _reduce_61(val, _values, result)
+     result = :includes 
+    result
+  end
+.,.,
 
-# reduce 62 omitted
+module_eval(<<'.,.,', 'parser.y', 213)
+  def _reduce_62(val, _values, result)
+     result = :dash_match 
+    result
+  end
+.,.,
 
 module_eval(<<'.,.,', 'parser.y', 217)
   def _reduce_63(val, _values, result)

data/lib/nokogiri/css/generated_tokenizer.rb

@@ -1,8 +1,8 @@
-#--
+#
 # DO NOT MODIFY!!!!
 # This file is automatically generated by rex 1.0.1
 # from lexical definition file "lib/nokogiri/css/tokenizer.rex".
-#++
+#
 
 module Nokogiri
 module CSS

data/lib/nokogiri/css/node.rb

@@ -1,12 +1,18 @@
 module Nokogiri
   module CSS
     class Node
-      attr_accessor :type, :value
+      # Get the type of this node
+      attr_accessor :type
+      # Get the value of this node
+      attr_accessor :value
+
+      # Create a new Node with +type+ and +value+
       def initialize type, value
         @type = type
         @value = value
       end
 
+      # Accept +visitor+
       def accept visitor
         visitor.send(:"visit_#{type.to_s.downcase}", self)
       end
@@ -18,6 +24,7 @@ module Nokogiri
         prefix + visitor.accept(self)
       end
 
+      # Preprocess this node tree
       def preprocess!
         ### Deal with nth-child
         matches = find_by_type(
@@ -67,14 +74,15 @@ module Nokogiri
             match.value[1] = Node.new(:COMBINATOR, [
               Node.new(:FUNCTION, ["#{match.value[1].value.first}("]),
               Node.new(:FUNCTION, ['self(', tag_name])
-            ])                                        
+            ])
           end
         end
 
         self
       end
 
-      def find_by_type(types)
+      # Find a node by type using +types+
+      def find_by_type types
         matches = []
         matches << self if to_type == types
         @value.each do |v|
@@ -83,12 +91,14 @@ module Nokogiri
         matches
       end
 
+      # Convert to_type
       def to_type
         [@type] + @value.map { |n|
           n.to_type if n.respond_to?(:to_type)
         }.compact
       end
 
+      # Convert to array
       def to_a
         [@type] + @value.map { |n| n.respond_to?(:to_a) ? n.to_a : [n] }
       end

data/lib/nokogiri/css/parser.rb

@@ -8,24 +8,29 @@ module Nokogiri
       @mutex    = Mutex.new
 
       class << self
+        # Turn on CSS parse caching
         attr_accessor :cache_on
         alias :cache_on? :cache_on
         alias :set_cache :cache_on=
 
+        # Get the css selector in +string+ from the cache
         def [] string
           return unless @cache_on
           @mutex.synchronize { @cache[string] }
         end
 
+        # Set the css selector in +string+ in the cache to +value+
         def []= string, value
           return value unless @cache_on
           @mutex.synchronize { @cache[string] = value }
         end
 
+        # Clear the cache
         def clear_cache
           @mutex.synchronize { @cache = {} }
         end
 
+        # Execute +block+ without cache
         def without_cache &block
           tmp = @cache_on
           @cache_on = false
@@ -45,12 +50,14 @@ module Nokogiri
         end
       end
 
+      # Create a new CSS parser with respect to +namespaces+
       def initialize namespaces = {}
         @namespaces = namespaces
         super()
       end
       alias :parse :scan_str
 
+      # Get the xpath for +string+ using +options+
       def xpath_for string, options={}
         key = string + options[:ns].to_s
         v = self.class[key]
@@ -65,6 +72,7 @@ module Nokogiri
         }
       end
 
+      # On CSS parser error, raise an exception
       def on_error error_token_id, error_value, value_stack
         after = value_stack.compact.last
         raise SyntaxError.new("unexpected '#{error_value}' after '#{after}'")

data/lib/nokogiri/css/parser.y

@@ -205,13 +205,13 @@ rule
     |
     ;
   eql_incl_dash
-    : EQUAL
-    | PREFIXMATCH
-    | SUFFIXMATCH
-    | SUBSTRINGMATCH
-    | NOT_EQUAL
-    | INCLUDES
-    | DASHMATCH
+    : EQUAL           { result = :equal }
+    | PREFIXMATCH     { result = :prefix_match }
+    | SUFFIXMATCH     { result = :suffix_match }
+    | SUBSTRINGMATCH  { result = :substring_match }
+    | NOT_EQUAL       { result = :not_equal }
+    | INCLUDES        { result = :includes }
+    | DASHMATCH       { result = :dash_match }
     ;
   negation
     : NOT negation_arg RPAREN {

data/lib/nokogiri/css/tokenizer.rb

@@ -1,6 +1,8 @@
 module Nokogiri
   module CSS
     class Tokenizer < GeneratedTokenizer
+      ###
+      # Scan and tokenize +str+
       def scan(str)
         scan_evaluate(str)
       end