superfeedr-nokogiri 1.4.0.20091116183308

data/lib/nokogiri/css/syntax_error.rb

@@ -0,0 +1,7 @@
+require 'nokogiri/syntax_error'
+module Nokogiri
+  module CSS
+    class SyntaxError < ::Nokogiri::SyntaxError
+    end
+  end
+end

data/lib/nokogiri/css/tokenizer.rb

@@ -0,0 +1,7 @@
+module Nokogiri
+  module CSS
+    class Tokenizer < GeneratedTokenizer
+      alias :scan :scan_setup
+    end
+  end
+end

data/lib/nokogiri/css/tokenizer.rex

@@ -0,0 +1,54 @@
+module Nokogiri
+module CSS
+class GeneratedTokenizer < GeneratedParser
+
+macro
+  nl        \n|\r\n|\r|\f
+  w         [\s\r\n\f]*
+  nonascii  [^\0-\177]
+  num       -?([0-9]+|[0-9]*\.[0-9]+)
+  unicode   \\[0-9A-Fa-f]{1,6}(\r\n|[\s\n\r\t\f])?
+
+  escape    {unicode}|\\[^\n\r\f0-9A-Fa-f]
+  nmchar    [_A-Za-z0-9-]|{nonascii}|{escape}
+  nmstart   [_A-Za-z]|{nonascii}|{escape}
+  ident     [-@]?({nmstart})({nmchar})*
+  name      ({nmchar})+
+  string1   "([^\n\r\f"]|{nl}|{nonascii}|{escape})*"
+  string2   '([^\n\r\f']|{nl}|{nonascii}|{escape})*'
+  string    {string1}|{string2}
+
+rule
+
+# [:state]  pattern  [actions]
+
+            {ident}\(\s*     { [:FUNCTION, text] }
+            {ident}          { [:IDENT, text] }
+            \#{name}         { [:HASH, text] }
+            {w}~={w}         { [:INCLUDES, text] }
+            {w}\|={w}        { [:DASHMATCH, text] }
+            {w}\^={w}        { [:PREFIXMATCH, text] }
+            {w}\$={w}        { [:SUFFIXMATCH, text] }
+            {w}\*={w}        { [:SUBSTRINGMATCH, text] }
+            {w}!={w}         { [:NOT_EQUAL, text] }
+            {w}={w}          { [:EQUAL, text] }
+            {w}\)            { [:RPAREN, text] }
+            {w}\[{w}         { [:LSQUARE, text] }
+            {w}\]            { [:RSQUARE, text] }
+            {w}\+{w}         { [:PLUS, text] }
+            {w}>{w}          { [:GREATER, text] }
+            {w},{w}          { [:COMMA, text] }
+            {w}~{w}          { [:TILDE, text] }
+            \:not\({w}       { [:NOT, text] }
+            {num}            { [:NUMBER, text] }
+            {w}\/\/{w}       { [:DOUBLESLASH, text] }
+            {w}\/{w}         { [:SLASH, text] }
+            
+            U\+[0-9a-f?]{1,6}(-[0-9a-f]{1,6})?  {[:UNICODE_RANGE, text] }
+            
+            [\s\t\r\n\f]+    { [:S, text] }
+            {string}         { [:STRING, text] }
+            .                { [text, text] }
+end
+end
+end

data/lib/nokogiri/css/xpath_visitor.rb

@@ -0,0 +1,162 @@
+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)
+
+        case node.value.first
+        when /^text\(/
+          'child::text()'
+        when /^self\(/
+          "self::#{node.value[1]}"
+        when /^(eq|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])
+          else
+            "position() = " + node.value[1]
+          end
+        when /^(first|first-of-type)\(/
+          "position() = 1"
+        when /^(last|last-of-type)\(/
+          "position() = last()"
+        when /^(nth-last-child|nth-last-of-type)\(/
+          "position() = last() - #{node.value[1]}"
+        when /^contains\(/
+          "contains(., #{node.value[1]})"
+        when /^gt\(/
+          "position() > #{node.value[1]}"
+        when /^only-child\(/
+          "last() = 1"
+        when /^comment\(/
+          "comment()"
+        else
+          args = ['.'] + node.value[1..-1]
+          "#{node.value.first}#{args.join(', ')})"
+        end
+      end
+
+      def visit_not node
+        'not(' + node.value.first.accept(self) + ')'
+      end
+
+      def visit_preceding_selector node
+        node.value.last.accept(self) +
+          '[preceding-sibling::' +
+          node.value.first.accept(self) +
+          ']'
+      end
+
+      def visit_id node
+        node.value.first =~ /^#(.*)$/
+        "@id = '#{$1}'"
+      end
+
+      def visit_attribute_condition node
+         attribute = if (node.value.first.type == :FUNCTION) or (node.value.first.value.first =~ /::/)
+                       ''
+                     else
+                       '@'
+                     end
+        attribute += node.value.first.accept(self)
+
+        # Support non-standard css
+        attribute.gsub!(/^@@/, '@')
+
+        return attribute unless node.value.length == 3
+
+        value = node.value.last
+        value = "'#{value}'" if value !~ /^['"]/
+
+        case node.value[1]
+        when :equal
+          attribute + " = " + "#{value}"
+        when :not_equal
+          attribute + " != " + "#{value}"
+        when :substring_match
+          "contains(#{attribute}, #{value})"
+        when :prefix_match
+          "starts-with(#{attribute}, #{value})"
+        when :dash_match
+          "#{attribute} = #{value} or starts-with(#{attribute}, concat(#{value}, '-'))"
+        when :includes
+          "contains(concat(\" \", #{attribute}, \" \"),concat(\" \", #{value}, \" \"))"
+        when :suffix_match
+          "substring(#{attribute}, string-length(#{attribute}) - " +
+            "string-length(#{value}) + 1, string-length(#{value})) = #{value}"
+        else
+          attribute + " #{node.value[1]} " + "#{value}"
+        end
+      end
+
+      def visit_pseudo_class node
+        if node.value.first.is_a?(Nokogiri::CSS::Node) and node.value.first.type == :FUNCTION
+          node.value.first.accept(self)
+        else
+          msg = :"visit_pseudo_class_#{node.value.first.gsub(/[(]/, '')}"
+          return self.send(msg, node) if self.respond_to?(msg)
+
+          case node.value.first
+          when "first" then "position() = 1"
+          when "last" then "position() = last()"
+          when "first-of-type" then "position() = 1"
+          when "last-of-type" then "position() = last()"
+          when "only-of-type" then "last() = 1"
+          when "empty" then "not(node())"
+          when "parent" then "node()"
+          when "root" then "not(parent::*)"
+          else
+            node.value.first + "(.)"
+          end
+        end
+      end
+
+      def visit_class_condition node
+        "contains(concat(' ', @class, ' '), ' #{node.value.first} ')"
+      end
+
+      {
+        '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
+        node.value.first.accept(self) + '[' +
+        node.value.last.accept(self) + ']'
+      end
+
+      def visit_element_name node
+        node.value.first
+      end
+
+      def accept node
+        node.accept(self)
+      end
+
+    private
+      def an_plus_b node
+        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
+
+        if (b == 0)
+          return "(position() mod #{a}) = 0"
+        else
+          compare = (a < 0) ? "<=" : ">="
+          return "(position() #{compare} #{b}) and (((position()-#{b}) mod #{a.abs}) = 0)"
+        end
+      end
+
+    end
+  end
+end

data/lib/nokogiri/decorators/slop.rb

@@ -0,0 +1,33 @@
+module Nokogiri
+  module Decorators
+    ###
+    # The Slop decorator implements method missing such that a methods may be
+    # used instead of XPath or CSS.  See Nokogiri.Slop
+    module Slop
+      ###
+      # look for node with +name+.  See Nokogiri.Slop
+      def method_missing name, *args, &block
+        if args.empty?
+          list = xpath("./#{name}")
+        elsif args.first.is_a? Hash
+          hash = args.first
+          if hash[:css]
+            list = css("#{name}#{hash[:css]}")
+          elsif hash[:xpath]
+            conds = Array(hash[:xpath]).join(' and ')
+            list = xpath("./#{name}[#{conds}]")
+          end
+        else
+          CSS::Parser.without_cache do
+            list = xpath(
+              *CSS.xpath_for("#{name}#{args.first}", :prefix => "./")
+            )
+          end
+        end
+
+        super if list.empty?
+        list.length == 1 ? list.first : list
+      end
+    end
+  end
+end

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

@@ -0,0 +1,28 @@
+module Nokogiri
+  module HTML
+    class Document < XML::Document
+
+      attr_accessor :cstruct # :nodoc:
+
+      def self.new(*args) # :nodoc:
+        uri         = args[0]
+        external_id = args[1]
+        doc = wrap(LibXML.htmlNewDoc(uri, external_id))
+        doc.send :initialize, *args
+        doc
+      end
+
+      def self.read_io(io, url, encoding, options) # :nodoc:
+        wrap_with_error_handling do
+          LibXML.htmlReadIO(IoCallbacks.reader(io), nil, nil, url, encoding, options)
+        end
+      end
+
+      def self.read_memory(string, url, encoding, options) # :nodoc:
+        wrap_with_error_handling do
+          LibXML.htmlReadMemory(string, string.length, url, encoding, options)
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,85 @@
+module Nokogiri
+  module HTML
+    class ElementDescription
+
+      attr_accessor :cstruct # :nodoc:
+
+      def required_attributes # :nodoc:
+        get_string_array_from :attrs_req
+      end
+
+      def deprecated_attributes # :nodoc:
+        get_string_array_from :attrs_depr
+      end
+
+      def optional_attributes # :nodoc:
+        get_string_array_from :attrs_opt
+      end
+
+      def default_sub_element # :nodoc:
+        cstruct[:defaultsubelt]
+      end
+
+      def sub_elements # :nodoc:
+        get_string_array_from :subelts
+      end
+
+      def description # :nodoc:
+        cstruct[:desc]
+      end
+
+      def inline? # :nodoc:
+        cstruct[:isinline] != 0
+      end
+
+      def deprecated? # :nodoc:
+        cstruct[:depr] != 0
+      end
+
+      def empty? # :nodoc:
+        cstruct[:empty] != 0
+      end
+
+      def save_end_tag? # :nodoc:
+        cstruct[:saveEndTag] != 0
+      end
+
+      def implied_end_tag? # :nodoc:
+        cstruct[:endTag] != 0
+      end
+
+      def implied_start_tag? # :nodoc:
+        cstruct[:startTag] != 0
+      end
+
+      def name # :nodoc:
+        cstruct[:name]
+      end
+
+      def self.[](tag_name) # :nodoc:
+        ptr = LibXML.htmlTagLookup(tag_name)
+        return nil if ptr.null?
+
+        desc = allocate
+        desc.cstruct = LibXML::HtmlElemDesc.new(ptr)
+        desc
+      end
+
+      private
+
+      def get_string_array_from(sym) # :nodoc:
+        list = []
+        return list if cstruct[sym].null?
+
+        j = 0
+        while (ptr = cstruct[sym].get_pointer(j * FFI.type_size(:pointer))) && ! ptr.null?
+          list << ptr.read_string
+          j += 1
+        end
+
+        list
+      end
+
+    end
+  end
+end

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

@@ -0,0 +1,16 @@
+module Nokogiri
+  module HTML
+    class EntityLookup
+
+      def get(key) # :nodoc:
+        ptr = LibXML.htmlEntityLookup(key.to_s)
+        return nil if ptr.null?
+
+        cstruct = LibXML::HtmlEntityDesc.new(ptr)
+        EntityDescription.new cstruct[:value], cstruct[:name], cstruct[:desc]
+      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

@@ -0,0 +1,42 @@
+# :stopdoc:
+module Nokogiri
+  module IoCallbacks
+
+    class << self
+
+      def plain_old_reader(io)
+        lambda do |ctx, buffer, len|
+          string = io.read(len)
+          return 0 if string.nil?
+          buffer.put_bytes(0, string, 0, string.length)
+          string.length
+        end
+      end
+
+      if defined?(FFI::IO.native_read)
+        def ffi_io_native_reader(io)
+          if io.is_a?(StringIO)
+            plain_old_reader(io)
+          else
+            lambda do |ctx, buffer, len|
+              rcode = FFI::IO.native_read(io, buffer, len)
+              (rcode < 0) ? 0 : rcode
+            end
+          end
+        end
+        alias :reader :ffi_io_native_reader
+      else
+        alias :reader :plain_old_reader
+      end
+
+      def writer(io)
+        lambda do |context, buffer, len|
+          io.write buffer
+          len
+        end
+      end
+    end
+
+  end
+end
+# :startdoc: