oga 0.1.3 → 0.2.0

data/lib/oga/version.rb

@@ -1,3 +1,3 @@
 module Oga
-  VERSION = '0.1.3'
+  VERSION = '0.2.0'
 end # Oga

data/lib/oga/xml/attribute.rb

@@ -87,7 +87,9 @@ module Oga
           full_name = name
         end
 
-        return %Q(#{full_name}="#{value}")
+        enc_value = value ? Entities.encode(value) : nil
+
+        return %Q(#{full_name}="#{enc_value}")
       end
 
       ##

data/lib/oga/xml/element.rb

@@ -138,6 +138,20 @@ module Oga
         end
       end
 
+      ##
+      # Removes an attribute from the element.
+      #
+      # @param [String] name The name (optionally including namespace prefix)
+      #  of the attribute to remove.
+      #
+      # @return [Oga::XML::Attribute]
+      #
+      def unset(name)
+        found = attribute(name)
+
+        return attributes.delete(found) if found
+      end
+
       ##
       # Returns the namespace of the element.
       #
@@ -273,7 +287,7 @@ module Oga
       # @return [Hash]
       #
       def available_namespaces
-        merged = namespaces
+        merged = namespaces.dup
         node   = parent
 
         while node && node.respond_to?(:namespaces)

data/lib/oga/xml/entities.rb

@@ -0,0 +1,60 @@
+module Oga
+  module XML
+    module Entities
+      ##
+      # Hash containing XML entities and the corresponding characters.
+      #
+      # The `&` mapping must come last to ensure proper conversion of non
+      # encoded to encoded forms (see {Oga::XML::Text#to_xml}).
+      #
+      # @return [Hash]
+      #
+      DECODE_MAPPING = {
+        '&lt;'  => '<',
+        '&gt;'  => '>',
+        '&amp;' => '&'
+      }
+
+      ##
+      # Hash containing characters and the corresponding XML entities.
+      #
+      # @return [Hash]
+      #
+      ENCODE_MAPPING = {
+        '&' => '&amp;',
+        '>' => '&gt;',
+        '<' => '&lt;'
+      }
+
+      ##
+      # Decodes XML entities.
+      #
+      # @param [String] input
+      # @return [String]
+      #
+      def self.decode(input)
+        if input.include?('&')
+          DECODE_MAPPING.each do |find, replace|
+            input = input.gsub(find, replace)
+          end
+        end
+
+        return input
+      end
+
+      ##
+      # Encodes special characters as XML entities.
+      #
+      # @param [String] input
+      # @return [String]
+      #
+      def self.encode(input)
+        ENCODE_MAPPING.each do |from, to|
+          input = input.gsub(from, to) if input.include?(from)
+        end
+
+        return input
+      end
+    end # Entities
+  end # XML
+end # Oga

data/lib/oga/xml/html_void_elements.rb

@@ -24,5 +24,7 @@ module Oga
       'track',
       'wbr'
     ])
+
+    HTML_VOID_ELEMENTS.merge(HTML_VOID_ELEMENTS.map { |name| name.upcase })
   end # XML
 end # Oga

data/lib/oga/xml/lexer.rb

@@ -66,7 +66,7 @@ module Oga
         @line     = 1
         @elements = []
 
-        @data.rewind if io_input?
+        @data.rewind if @data.respond_to?(:rewind)
 
         reset_native
       end
@@ -78,25 +78,18 @@ module Oga
       # @yieldparam [String]
       #
       def read_data
-        # We can't check for #each_line since String also defines that. Using
-        # String#each_line has no benefit over just lexing the String in one
-        # go.
-        if io_input?
-          @data.each_line do |line|
-            yield line
-          end
-        else
+        if @data.is_a?(String)
           yield @data
-        end
-      end
 
-      ##
-      # Returns `true` if the input is an IO like object, false otherwise.
-      #
-      # @return [TrueClass|FalseClass]
-      #
-      def io_input?
-        return @data.is_a?(IO) || @data.is_a?(StringIO)
+        # IO, StringIO, etc
+        # THINK: read(N) would be nice, but currently this screws up the C code
+        elsif @data.respond_to?(:each_line)
+          @data.each_line { |line| yield line }
+
+        # Enumerator, Array, etc
+        elsif @data.respond_to?(:each)
+          @data.each { |chunk| yield chunk }
+        end
       end
 
       ##
@@ -189,12 +182,26 @@ module Oga
       end
 
       ##
-      # Called when processing single/double quoted strings.
+      # Called when processing a single quote.
+      #
+      def on_string_squote
+        add_token(:T_STRING_SQUOTE)
+      end
+
+      ##
+      # Called when processing a double quote.
+      #
+      def on_string_dquote
+        add_token(:T_STRING_DQUOTE)
+      end
+
+      ##
+      # Called when processing the body of a string.
       #
       # @param [String] value The data between the quotes.
       #
-      def on_string(value)
-        add_token(:T_STRING, value)
+      def on_string_body(value)
+        add_token(:T_STRING_BODY, Entities.decode(value))
       end
 
       ##
@@ -322,7 +329,12 @@ module Oga
       # Called on the closing `>` of the open tag of an element.
       #
       def on_element_open_end
-        if html? and HTML_VOID_ELEMENTS.include?(current_element.downcase)
+        return unless html?
+
+        # Only downcase the name if we can't find an all lower/upper version of
+        # the element name. This can save us a *lot* of String allocations.
+        if HTML_VOID_ELEMENTS.include?(current_element) \
+        or HTML_VOID_ELEMENTS.include?(current_element.downcase)
           add_token(:T_ELEM_END)
           @elements.pop
         end
@@ -343,13 +355,9 @@ module Oga
       # @param [String] value
       #
       def on_text(value)
-        unless value.empty?
-          add_token(:T_TEXT, value)
+        return if value.empty?
 
-          lines = value.count("\n")
-
-          advance_line(lines) if lines > 0
-        end
+        add_token(:T_TEXT, Entities.decode(value))
       end
 
       ##

data/lib/oga/xml/node_set.rb

@@ -197,6 +197,28 @@ module Oga
         return self.class.new(to_a | other.to_a)
       end
 
+      ##
+      # Returns `true` if the current node set and the one given in `other` are
+      # equal to each other.
+      #
+      # @param [Oga::XML::NodeSet] other
+      #
+      def ==(other)
+        return other.is_a?(NodeSet) && other.equal_nodes?(@nodes)
+      end
+
+      ##
+      # Returns `true` if the nodes given in `nodes` are equal to those
+      # specified in the current `@nodes` variable. This method allows two
+      # NodeSet instances to compare each other without the need of exposing
+      # `@nodes` to the public.
+      #
+      # @param [Array<Oga::XML::Node>] nodes
+      #
+      def equal_nodes?(nodes)
+        return @nodes == nodes
+      end
+
       ##
       # Adds the nodes of the given node set to the current node set.
       #

data/lib/oga/xml/parser.rb

@@ -81,43 +81,10 @@ module Oga
   # @raise [Racc::ParseError]
   #
   def on_error(type, value, stack)
-    name        = token_to_str(type)
-    name        = TOKEN_ERROR_MAPPING[name] || name
-    index       = @line - 1
-    index_range = (index - 5)..(index + 5)
-    code        = ''
-
-    # For IO we sadly have to re-read the input :<
-    if @data.respond_to?(:rewind)
-      @data.rewind
-    end
-
-    # Show up to 5 lines before and after the offending line (if they exist).
-    @data.each_line.with_index do |line, line_index|
-      next unless index_range.cover?(line_index)
-
-      number = line_index + 1
-
-      if line_index == index
-        prefix = '=> '
-      else
-        prefix = '   '
-      end
-
-      line = line.strip
+    name = token_to_str(type)
+    name = TOKEN_ERROR_MAPPING[name] || name
 
-      if line.length > 80
-        line = line[0..79] + ' (more)'
-      end
-
-      code << "#{prefix}#{number}: #{line}\n"
-    end
-
-    raise Racc::ParseError, <<-EOF.strip
-Unexpected #{name} on line #{@line}:
-
-#{code}
-    EOF
+    raise Racc::ParseError, "Unexpected #{name} on line #{@line}"
   end
 
   ##
@@ -254,118 +221,138 @@ Unexpected #{name} on line #{@line}:
 ##### State transition tables begin ###
 
 racc_action_table = [
-    40,    20,    12,    25,    26,    20,    12,    13,    14,    16,
-    41,    13,    14,    16,    21,    19,    35,    15,    39,    19,
-    43,    15,    20,    12,    52,    31,    32,    51,    13,    14,
-    16,    48,    36,    37,    47,    38,    19,    44,    15,    31,
-    32,    31,    32,    45,    46,    24,    49,    50,    23,    53 ]
+    40,    20,    47,    46,    49,    12,    60,    64,    50,    63,
+    13,    14,    16,    20,    48,    47,    46,    12,    19,    51,
+    15,    39,    13,    14,    16,    20,    58,    53,    57,    12,
+    19,    54,    15,    41,    13,    14,    16,    36,    37,    35,
+    38,    24,    19,    23,    15,    31,    32,    62,    63,    31,
+    32,    55,    57,    25,    26,    47,    46,    31,    32,    21,
+    65 ]
 
 racc_action_check = [
-    24,     0,     0,    16,    16,     3,     3,     0,     0,     0,
-    26,     3,     3,     3,     1,     0,    21,     0,    24,     3,
-    30,     3,    18,    18,    48,    19,    19,    48,    18,    18,
-    18,    37,    23,    23,    37,    23,    18,    32,    18,    28,
-    28,    17,    17,    33,    34,    15,    38,    40,    12,    52 ]
+    24,     0,    52,    52,    33,     0,    52,    59,    34,    59,
+     0,     0,     0,     3,    32,    37,    37,     3,     0,    37,
+     0,    24,     3,     3,     3,    18,    47,    38,    47,    18,
+     3,    40,     3,    26,    18,    18,    18,    23,    23,    21,
+    23,    15,    18,    12,    18,    19,    19,    56,    56,    17,
+    17,    46,    46,    16,    16,    30,    30,    28,    28,     1,
+    61 ]
 
 racc_action_pointer = [
-    -2,    14,   nil,     2,   nil,   nil,   nil,   nil,   nil,   nil,
-   nil,   nil,    41,   nil,   nil,    25,    -9,    26,    19,    10,
-   nil,    16,   nil,    27,    -3,   nil,    -2,   nil,    24,   nil,
-    18,   nil,    22,    29,    26,   nil,   nil,    29,    41,   nil,
-    26,   nil,   nil,   nil,   nil,   nil,   nil,   nil,    22,   nil,
-   nil,   nil,    44,   nil ]
+    -1,    59,   nil,    11,   nil,   nil,   nil,   nil,   nil,   nil,
+   nil,   nil,    34,   nil,   nil,    19,    39,    32,    23,    28,
+   nil,    39,   nil,    30,    -2,   nil,    19,   nil,    40,   nil,
+    52,   nil,    -3,   -12,   -12,   nil,   nil,    12,    20,   nil,
+     8,   nil,   nil,   nil,   nil,   nil,    47,    23,   nil,   nil,
+   nil,   nil,    -1,   nil,   nil,   nil,    43,   nil,   nil,     4,
+   nil,    53,   nil,   nil,   nil,   nil ]
 
 racc_action_default = [
-    -3,   -36,    -1,    -2,    -5,    -6,    -7,    -8,    -9,   -10,
-   -11,   -12,   -36,   -18,   -19,   -36,   -36,   -27,    -3,   -27,
-   -35,   -36,    -4,   -36,   -36,   -22,   -36,   -24,   -26,   -29,
-   -30,   -32,   -36,   -36,   -36,    54,   -13,   -36,   -36,   -20,
-   -36,   -23,   -28,   -31,   -33,   -25,   -34,   -14,   -36,   -17,
-   -21,   -15,   -36,   -16 ]
+    -3,   -44,    -1,    -2,    -5,    -6,    -7,    -8,    -9,   -10,
+   -11,   -12,   -44,   -18,   -19,   -44,   -44,   -27,    -3,   -27,
+   -35,   -44,    -4,   -44,   -44,   -22,   -44,   -24,   -26,   -29,
+   -30,   -32,   -44,   -44,   -44,    66,   -13,   -44,   -44,   -20,
+   -44,   -23,   -28,   -31,   -36,   -37,   -44,   -44,   -33,   -25,
+   -34,   -14,   -44,   -17,   -21,   -38,   -44,   -42,   -40,   -44,
+   -15,   -44,   -39,   -43,   -41,   -16 ]
 
 racc_goto_table = [
-     2,    27,    22,    34,     1,    42,   nil,   nil,   nil,   nil,
-   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,    33 ]
+    43,     2,    27,    22,    34,    56,    59,    52,    42,     1,
+   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,    33,
+   nil,   nil,    61 ]
 
 racc_goto_check = [
-     2,    14,     4,    14,     1,    16,   nil,   nil,   nil,   nil,
-   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,     2 ]
+    12,     2,    15,     4,    15,    21,    21,    12,    17,     1,
+   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,   nil,     2,
+   nil,   nil,    12 ]
 
 racc_goto_pointer = [
-   nil,     4,     0,   nil,    -1,   nil,   nil,   nil,   nil,   nil,
-   nil,   nil,   nil,   nil,   -16,   nil,   -23,   nil ]
+   nil,     9,     1,   nil,     0,   nil,   nil,   nil,   nil,   nil,
+   nil,   nil,   -30,   nil,   nil,   -15,   nil,   -20,   nil,   nil,
+   nil,   -41 ]
 
 racc_goto_default = [
    nil,   nil,   nil,     3,     4,     5,     6,     7,     8,     9,
-    10,    11,    17,    18,   nil,    28,    29,    30 ]
+    10,    11,   nil,    17,    18,   nil,    28,    29,    30,    44,
+    45,   nil ]
 
 racc_reduce_table = [
   0, 0, :racc_error,
-  1, 23, :_reduce_1,
-  1, 24, :_reduce_2,
-  0, 24, :_reduce_3,
-  2, 25, :_reduce_4,
-  1, 25, :_reduce_5,
-  1, 26, :_reduce_none,
-  1, 26, :_reduce_none,
-  1, 26, :_reduce_none,
-  1, 26, :_reduce_none,
-  1, 26, :_reduce_none,
-  1, 26, :_reduce_none,
-  1, 26, :_reduce_none,
-  3, 27, :_reduce_13,
-  4, 27, :_reduce_14,
-  5, 27, :_reduce_15,
-  6, 27, :_reduce_16,
-  4, 27, :_reduce_17,
-  1, 28, :_reduce_18,
-  1, 29, :_reduce_19,
-  3, 33, :_reduce_20,
-  4, 33, :_reduce_21,
-  2, 34, :_reduce_22,
-  3, 34, :_reduce_23,
-  2, 35, :_reduce_24,
-  3, 30, :_reduce_25,
-  1, 36, :_reduce_26,
-  0, 36, :_reduce_27,
-  2, 37, :_reduce_28,
-  1, 37, :_reduce_29,
-  1, 38, :_reduce_30,
-  2, 38, :_reduce_31,
-  1, 39, :_reduce_32,
-  2, 39, :_reduce_33,
-  3, 32, :_reduce_34,
-  1, 31, :_reduce_35 ]
-
-racc_reduce_n = 36
-
-racc_shift_n = 54
+  1, 25, :_reduce_1,
+  1, 26, :_reduce_2,
+  0, 26, :_reduce_3,
+  2, 27, :_reduce_4,
+  1, 27, :_reduce_5,
+  1, 28, :_reduce_none,
+  1, 28, :_reduce_none,
+  1, 28, :_reduce_none,
+  1, 28, :_reduce_none,
+  1, 28, :_reduce_none,
+  1, 28, :_reduce_none,
+  1, 28, :_reduce_none,
+  3, 29, :_reduce_13,
+  4, 29, :_reduce_14,
+  5, 29, :_reduce_15,
+  6, 29, :_reduce_16,
+  4, 29, :_reduce_17,
+  1, 30, :_reduce_18,
+  1, 31, :_reduce_19,
+  3, 35, :_reduce_20,
+  4, 35, :_reduce_21,
+  2, 37, :_reduce_22,
+  3, 37, :_reduce_23,
+  2, 38, :_reduce_24,
+  3, 32, :_reduce_25,
+  1, 39, :_reduce_26,
+  0, 39, :_reduce_27,
+  2, 40, :_reduce_28,
+  1, 40, :_reduce_29,
+  1, 41, :_reduce_30,
+  2, 41, :_reduce_31,
+  1, 42, :_reduce_32,
+  2, 42, :_reduce_33,
+  3, 34, :_reduce_34,
+  1, 33, :_reduce_35,
+  1, 36, :_reduce_none,
+  1, 36, :_reduce_none,
+  2, 43, :_reduce_38,
+  3, 43, :_reduce_39,
+  2, 44, :_reduce_40,
+  3, 44, :_reduce_41,
+  1, 45, :_reduce_42,
+  2, 45, :_reduce_43 ]
+
+racc_reduce_n = 44
+
+racc_shift_n = 66
 
 racc_token_table = {
   false => 0,
   :error => 1,
-  :T_STRING => 2,
-  :T_TEXT => 3,
-  :T_DOCTYPE_START => 4,
-  :T_DOCTYPE_END => 5,
-  :T_DOCTYPE_TYPE => 6,
-  :T_DOCTYPE_NAME => 7,
-  :T_DOCTYPE_INLINE => 8,
-  :T_CDATA => 9,
-  :T_COMMENT => 10,
-  :T_ELEM_START => 11,
-  :T_ELEM_NAME => 12,
-  :T_ELEM_NS => 13,
-  :T_ELEM_END => 14,
-  :T_ATTR => 15,
-  :T_ATTR_NS => 16,
-  :T_XML_DECL_START => 17,
-  :T_XML_DECL_END => 18,
-  :T_PROC_INS_START => 19,
-  :T_PROC_INS_NAME => 20,
-  :T_PROC_INS_END => 21 }
-
-racc_nt_base = 22
+  :T_TEXT => 2,
+  :T_STRING_SQUOTE => 3,
+  :T_STRING_DQUOTE => 4,
+  :T_STRING_BODY => 5,
+  :T_DOCTYPE_START => 6,
+  :T_DOCTYPE_END => 7,
+  :T_DOCTYPE_TYPE => 8,
+  :T_DOCTYPE_NAME => 9,
+  :T_DOCTYPE_INLINE => 10,
+  :T_CDATA => 11,
+  :T_COMMENT => 12,
+  :T_ELEM_START => 13,
+  :T_ELEM_NAME => 14,
+  :T_ELEM_NS => 15,
+  :T_ELEM_END => 16,
+  :T_ATTR => 17,
+  :T_ATTR_NS => 18,
+  :T_XML_DECL_START => 19,
+  :T_XML_DECL_END => 20,
+  :T_PROC_INS_START => 21,
+  :T_PROC_INS_NAME => 22,
+  :T_PROC_INS_END => 23 }
+
+racc_nt_base = 24
 
 racc_use_result_var = false
 
@@ -388,8 +375,10 @@ Racc_arg = [
 Racc_token_to_s_table = [
   "$end",
   "error",
-  "T_STRING",
   "T_TEXT",
+  "T_STRING_SQUOTE",
+  "T_STRING_DQUOTE",
+  "T_STRING_BODY",
   "T_DOCTYPE_START",
   "T_DOCTYPE_END",
   "T_DOCTYPE_TYPE",
@@ -420,12 +409,16 @@ Racc_token_to_s_table = [
   "text",
   "xmldecl",
   "proc_ins",
+  "string",
   "element_open",
   "element_start",
   "attributes",
   "attributes_",
   "attribute",
-  "attribute_name" ]
+  "attribute_name",
+  "string_dquote",
+  "string_squote",
+  "string_body" ]
 
 Racc_debug_parser = false
 
@@ -579,6 +572,34 @@ def _reduce_35(val, _values)
  on_text(val[0]) 
 end
 
+# reduce 36 omitted
+
+# reduce 37 omitted
+
+def _reduce_38(val, _values)
+ '' 
+end
+
+def _reduce_39(val, _values)
+ val[1] 
+end
+
+def _reduce_40(val, _values)
+ '' 
+end
+
+def _reduce_41(val, _values)
+ val[1] 
+end
+
+def _reduce_42(val, _values)
+ val[0] 
+end
+
+def _reduce_43(val, _values)
+ val[0] + val[1] 
+end
+
 def _reduce_none(val, _values)
   val[0]
 end