nokogiri 1.6.1 → 1.6.2.rc1

data/lib/nokogiri/css/parser.y

@@ -28,17 +28,6 @@ rule
                     Node.new(:CONDITIONAL_SELECTOR, [val.first, val[1]])
                   end
       }
-    | element_name hcap_1toN negation {
-        result = Node.new(:CONDITIONAL_SELECTOR,
-          [
-            val.first,
-            Node.new(:COMBINATOR, [val[1], val.last])
-          ]
-        )
-      }
-    | element_name negation {
-        result = Node.new(:CONDITIONAL_SELECTOR, val)
-      }
     | function
     | function pseudo {
         result = Node.new(:CONDITIONAL_SELECTOR, val)
@@ -46,14 +35,6 @@ rule
     | function attrib {
         result = Node.new(:CONDITIONAL_SELECTOR, val)
       }
-    | hcap_1toN negation {
-        result = Node.new(:CONDITIONAL_SELECTOR,
-          [
-            Node.new(:ELEMENT_NAME, ['*']),
-            Node.new(:COMBINATOR, val)
-          ]
-        )
-      }
     | hcap_1toN {
         result = Node.new(:CONDITIONAL_SELECTOR,
           [Node.new(:ELEMENT_NAME, ['*']), val.first]
@@ -130,7 +111,7 @@ rule
     | FUNCTION expr RPAREN {
         result = Node.new(:FUNCTION, [val.first.strip, val[1]].flatten)
       }
-    | FUNCTION an_plus_b RPAREN {
+    | FUNCTION nth RPAREN {
         result = Node.new(:FUNCTION, [val.first.strip, val[1]].flatten)
       }
     | NOT expr RPAREN {
@@ -150,10 +131,10 @@ rule
       {
         if val[0] == 'even'
           val = ["2","n","+","0"]
-          result = Node.new(:AN_PLUS_B, val)
+          result = Node.new(:NTH, val)
         elsif val[0] == 'odd'
           val = ["2","n","+","1"]
-          result = Node.new(:AN_PLUS_B, val)
+          result = Node.new(:NTH, val)
         else
           # This is not CSS standard.  It allows us to support this:
           # assert_xpath("//a[foo(., @href)]", @parser.parse('a:foo(@href)'))
@@ -163,11 +144,11 @@ rule
         end
       }
     ;
-  an_plus_b
+  nth
     : NUMBER IDENT PLUS NUMBER          # 5n+3 -5n+3
       {
         if val[1] == 'n'
-          result = Node.new(:AN_PLUS_B, val)
+          result = Node.new(:NTH, val)
         else
           raise Racc::ParseError, "parse error on IDENT '#{val[1]}'"
         end
@@ -175,21 +156,27 @@ rule
     | IDENT PLUS NUMBER {               # n+3, -n+3
         if val[0] == 'n'
           val.unshift("1")
-          result = Node.new(:AN_PLUS_B, val)
+          result = Node.new(:NTH, val)
         elsif val[0] == '-n'
           val[0] = 'n'
           val.unshift("-1")
-          result = Node.new(:AN_PLUS_B, val)
+          result = Node.new(:NTH, val)
         else
           raise Racc::ParseError, "parse error on IDENT '#{val[1]}'"
         end
       }
-    | NUMBER IDENT                      # 5n, -5n
-      {
-        if val[1] == 'n'
+    | NUMBER IDENT {                    # 5n, -5n, 10n-1
+        n = val[1]
+        if n[0, 2] == 'n-'
+          val[1] = 'n'
+          val << "-"
+          # b is contained in n as n is the string "n-b"
+          val << n[2, n.size]
+          result = Node.new(:NTH, val)
+        elsif n == 'n'
           val << "+"
           val << "0"
-          result = Node.new(:AN_PLUS_B, val)
+          result = Node.new(:NTH, val)
         else
           raise Racc::ParseError, "parse error on IDENT '#{val[1]}'"
         end
@@ -218,10 +205,14 @@ rule
     | pseudo hcap_1toN {
         result = Node.new(:COMBINATOR, val)
       }
+    | negation hcap_1toN {
+        result = Node.new(:COMBINATOR, val)
+      }
     | attribute_id
     | class
     | attrib
     | pseudo
+    | negation
     ;
   attribute_id
     : HASH { result = Node.new(:ID, val) }

data/lib/nokogiri/css/xpath_visitor.rb

@@ -2,7 +2,7 @@ module Nokogiri
   module CSS
     class XPathVisitor # :nodoc:
       def visit_function node
-        #  note that nth-child and nth-last-child are preprocessed in css/node.rb.
+
         msg = :"visit_function_#{node.value.first.gsub(/[(]/, '')}"
         return self.send(msg, node) if self.respond_to?(msg)
 
@@ -13,19 +13,31 @@ module Nokogiri
           "self::#{node.value[1]}"
         when /^eq\(/
           "position() = #{node.value[1]}"
-        when /^(nth|nth-of-type|nth-child)\(/
-          if node.value[1].is_a?(Nokogiri::CSS::Node) and node.value[1].type == :AN_PLUS_B
-            an_plus_b(node.value[1])
+        when /^(nth|nth-of-type)\(/
+          if node.value[1].is_a?(Nokogiri::CSS::Node) and node.value[1].type == :NTH
+            nth(node.value[1])
           else
             "position() = #{node.value[1]}"
           end
-        when /^(nth-last-child|nth-last-of-type)\(/
-          if node.value[1].is_a?(Nokogiri::CSS::Node) and node.value[1].type == :AN_PLUS_B
-            an_plus_b(node.value[1], :last => true)
+        when /^nth-child\(/
+          if node.value[1].is_a?(Nokogiri::CSS::Node) and node.value[1].type == :NTH
+            nth(node.value[1], :child => true)
+          else
+            "count(preceding-sibling::*) = #{node.value[1].to_i-1}"
+          end
+        when /^nth-last-of-type\(/
+          if node.value[1].is_a?(Nokogiri::CSS::Node) and node.value[1].type == :NTH
+            nth(node.value[1], :last => true)
           else
             index = node.value[1].to_i - 1
             index == 0 ? "position() = last()" : "position() = last() - #{index}"
           end
+        when /^nth-last-child\(/
+          if node.value[1].is_a?(Nokogiri::CSS::Node) and node.value[1].type == :NTH
+            nth(node.value[1], :last => true, :child => true)
+          else
+            "count(following-sbiling::*) = #{node.value[1].to_i-1}"
+          end
         when /^(first|first-of-type)\(/
           "position() = 1"
         when /^(last|last-of-type)\(/
@@ -105,10 +117,13 @@ module Nokogiri
           return self.send(msg, node) if self.respond_to?(msg)
 
           case node.value.first
-          when "first", "first-child" then "position() = 1"
-          when "last", "last-child" then "position() = last()"
+          when "first" then "position() = 1"
+          when "first-child" then "count(preceding-sibling::*) = 0"
+          when "last" then "position() = last()"
+          when "last-child" then "count(following-sibling::*) = 0"
           when "first-of-type" then "position() = 1"
           when "last-of-type" then "position() = last()"
+          when "only-child" then "count(preceding-sibling::*) = 0 and count(following-sibling::*) = 0"
           when "only-of-type" then "last() = 1"
           when "empty" then "not(node())"
           when "parent" then "node()"
@@ -123,8 +138,15 @@ module Nokogiri
         "contains(concat(' ', normalize-space(@class), ' '), ' #{node.value.first} ')"
       end
 
+      def visit_combinator node
+        if is_of_type_pseudo_class?(node.value.last)
+          "#{node.value.first.accept(self) if node.value.first}][#{node.value.last.accept(self)}"
+        else
+          "#{node.value.first.accept(self) if node.value.first} and #{node.value.last.accept(self)}"
+        end
+      end
+      
       {
-        'combinator'                => ' and ',
         'direct_adjacent_selector'  => "/following-sibling::*[1]/self::",
         'following_selector'        => "/following-sibling::",
         'descendant_selector'       => '//',
@@ -151,12 +173,15 @@ module Nokogiri
       end
 
     private
-      def an_plus_b node, options={}
+      def nth node, options={}
         raise ArgumentError, "expected an+b node to contain 4 tokens, but is #{node.value.inspect}" unless node.value.size == 4
 
-        a = node.value[0].to_i
-        b = node.value[3].to_i
-        position = options[:last] ? "(last()-position()+1)" : "position()"
+        a, b = read_a_and_positive_b node.value
+        position = if options[:child]
+          options[:last] ? "(count(following-sibling::*) + 1)" : "(count(preceding-sibling::*) + 1)"
+        else
+          options[:last] ? "(last()-position()+1)" : "position()"
+        end
 
         if (b == 0)
           return "(#{position} mod #{a}) = 0"
@@ -166,6 +191,29 @@ module Nokogiri
         end
       end
 
+      def read_a_and_positive_b values
+        op = values[2]
+        if op == "+"
+          a = values[0].to_i
+          b = values[3].to_i
+        elsif op == "-"
+          a = values[0].to_i
+          b = a - (values[3].to_i % a)
+        else
+          raise ArgumentError, "expected an+b node to have either + or - as the operator, but is #{op.inspect}"
+        end
+        [a, b]
+      end
+      
+      def is_of_type_pseudo_class? node
+        if node.type==:PSEUDO_CLASS
+          if node.value[0].is_a?(Nokogiri::CSS::Node) and node.value[0].type == :FUNCTION
+            node.value[0].value[0]
+          else
+            node.value[0]
+          end =~ /(nth|first|last|only)-of-type(\()?/
+        end   
+      end
     end
   end
 end

data/lib/nokogiri/html/document.rb

@@ -5,24 +5,59 @@ module Nokogiri
       # Get the meta tag encoding for this document.  If there is no meta tag,
       # then nil is returned.
       def meta_encoding
-        meta = meta_content_type and
-          match = /charset\s*=\s*([\w-]+)/i.match(meta['content']) and
-          match[1]
+        case
+        when meta = at('//meta[@charset]')
+          meta[:charset]
+        when meta = meta_content_type
+          meta['content'][/charset\s*=\s*([\w-]+)/i, 1]
+        end
       end
 
       ###
-      # Set the meta tag encoding for this document.  If there is no meta
-      # content tag, the encoding is not set.
+      # Set the meta tag encoding for this document.
+      #
+      # If an meta encoding tag is already present, its content is
+      # replaced with the given text.
+      #
+      # Otherwise, this method tries to create one at an appropriate
+      # place supplying head and/or html elements as necessary, which
+      # is inside a head element if any, and before any text node or
+      # content element (typically <body>) if any.
+      #
+      # The result when trying to set an encoding that is different
+      # from the document encoding is undefined.
+      #
+      # Beware in CRuby, that libxml2 automatically inserts a meta tag
+      # into a head element.
       def meta_encoding= encoding
-        meta = meta_content_type and
-          meta['content'] = "text/html; charset=%s" % encoding
+        case
+        when meta = meta_content_type
+          meta['content'] = 'text/html; charset=%s' % encoding
+          encoding
+        when meta = at('//meta[@charset]')
+          meta['charset'] = encoding
+        else
+          meta = XML::Node.new('meta', self)
+          if dtd = internal_subset and dtd.html5_dtd?
+            meta['charset'] = encoding
+          else
+            meta['http-equiv'] = 'Content-Type'
+            meta['content'] = 'text/html; charset=%s' % encoding
+          end
+
+          case
+          when head = at('//head')
+            head.prepend_child(meta)
+          else
+            set_metadata_element(meta)
+          end
+          encoding
+        end
       end
 
       def meta_content_type
-        css('meta[@http-equiv]').find { |node|
-          node['http-equiv'] =~ /\AContent-Type\z/i and
-            !node['content'].nil? and
-            !node['content'].empty?
+        xpath('//meta[@http-equiv and boolean(@content)]').find { |node|
+          node['http-equiv'] =~ /\AContent-Type\z/i
         }
       end
       private :meta_content_type
@@ -31,20 +66,64 @@ module Nokogiri
       # Get the title string of this document.  Return nil if there is
       # no title tag.
       def title
-        title = at('title') and title.inner_text
+        title = at('//title') and title.inner_text
       end
 
       ###
-      # Set the title string of this document.  If there is no head
-      # element, the title is not set.
+      # Set the title string of this document.
+      #
+      # If a title element is already present, its content is replaced
+      # with the given text.
+      #
+      # Otherwise, this method tries to create one at an appropriate
+      # place supplying head and/or html elements as necessary, which
+      # is inside a head element if any, right after a meta
+      # encoding/charset tag if any, and before any text node or
+      # content element (typically <body>) if any.
       def title=(text)
-        unless title = at('title')
-          head = at('head') or return nil
-          title = Nokogiri::XML::Node.new('title', self)
+        tnode = XML::Text.new(text, self)
+        if title = at('//title')
+          title.children = tnode
+          return text
+        end
+
+        title = XML::Node.new('title', self) << tnode
+        case
+        when head = at('//head')
           head << title
+        when meta = at('//meta[@charset]') || meta_content_type
+          # better put after charset declaration
+          meta.add_next_sibling(title)
+        else
+          set_metadata_element(title)
+        end
+        text
+      end
+
+      def set_metadata_element(element)
+        case
+        when head = at('//head')
+          head << element
+        when html = at('//html')
+          head = html.prepend_child(XML::Node.new('head', self))
+          head.prepend_child(element)
+        when first = children.find { |node|
+            case node
+            when XML::Element, XML::Text
+              true
+            end
+          }
+          # We reach here only if the underlying document model
+          # allows <html>/<head> elements to be omitted and does not
+          # automatically supply them.
+          first.add_previous_sibling(element)
+        else
+          html = add_child(XML::Node.new('html', self))
+          head = html.add_child(XML::Node.new('head', self))
+          head.prepend_child(element)
         end
-        title.children = XML::Text.new(text, self)
       end
+      private :set_metadata_element
 
       ####
       # Serialize Node using +options+.  Save options can also be set using a

data/lib/nokogiri/html/sax/parser.rb

@@ -22,7 +22,7 @@ module Nokogiri
       #   end
       #
       #   parser = Nokogiri::HTML::SAX::Parser.new(MyDoc.new)
-      #   parser.parse(File.read(ARGV[0], 'rb'))
+      #   parser.parse(File.read(ARGV[0], mode: 'rb'))
       #
       # For more information on SAX parsers, see Nokogiri::XML::SAX
       class Parser < Nokogiri::XML::SAX::Parser
@@ -40,7 +40,7 @@ module Nokogiri
         # Parse a file with +filename+
         def parse_file filename, encoding = 'UTF-8'
           raise ArgumentError unless filename
-          raise Errno::ENOENT unless File.exists?(filename)
+          raise Errno::ENOENT unless File.exist?(filename)
           raise Errno::EISDIR if File.directory?(filename)
           ctx = ParserContext.file(filename, encoding)
           yield ctx if block_given?

data/lib/nokogiri/version.rb

@@ -1,6 +1,6 @@
 module Nokogiri
   # The version of Nokogiri you are using
-  VERSION = '1.6.1'
+  VERSION = '1.6.2.rc1'
 
   class VersionInfo # :nodoc:
     def jruby?

data/lib/nokogiri/xml/builder.rb

@@ -380,7 +380,7 @@ module Nokogiri
       ###
       # Insert +node+ as a child of the current Node
       def insert(node, &block)
-        node.parent = @parent
+        node = @parent.add_child(node)
         if block_given?
           old_parent = @parent
           @parent    = node

data/lib/nokogiri/xml/document.rb

@@ -45,7 +45,7 @@ module Nokogiri
         # Give the options to the user
         yield options if block_given?
 
-        return new if empty_doc?(string_or_io)
+        return new if !options.strict? && empty_doc?(string_or_io)
 
         doc = if string_or_io.respond_to?(:read)
           url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
@@ -233,7 +233,7 @@ module Nokogiri
       undef_method :namespace_definitions, :line, :add_namespace
 
       def add_child node_or_tags
-        raise "Document already has a root node" if root
+        raise "Document already has a root node" if root && root.name != 'nokogiri_text_wrapper'
         node_or_tags = coerce(node_or_tags)
         if node_or_tags.is_a?(XML::NodeSet)
           raise "Document cannot have multiple root nodes" if node_or_tags.size > 1

data/lib/nokogiri/xml/dtd.rb

@@ -17,6 +17,16 @@ module Nokogiri
           block.call([key, value])
         }
       end
+
+      def html_dtd?
+        name.casecmp('html').zero?
+      end
+
+      def html5_dtd?
+        html_dtd? &&
+          external_id.nil? &&
+          (system_id.nil? || system_id == 'about:legacy-compat')
+      end
     end
   end
 end