Support for parsing XML processing instructions.

lib/oga.rb

@@ -23,6 +23,7 @@ require_relative 'oga/xml/text'
 require_relative 'oga/xml/comment'
 require_relative 'oga/xml/cdata'
 require_relative 'oga/xml/xml_declaration'
+require_relative 'oga/xml/processing_instruction'
 require_relative 'oga/xml/doctype'
 require_relative 'oga/xml/attribute'
 require_relative 'oga/xml/namespace'

lib/oga/xml/parser.y

@@ -15,6 +15,7 @@ token T_DOCTYPE_INLINE
 token T_CDATA T_COMMENT
 token T_ELEM_START T_ELEM_NAME T_ELEM_NS T_ELEM_END T_ATTR T_ATTR_NS
 token T_XML_DECL_START T_XML_DECL_END
+token T_PROC_INS_START T_PROC_INS_NAME T_PROC_INS_END
 
 options no_result_var
 
@@ -40,6 +41,7 @@ rule
     | element
     | text
     | xmldecl
+    | proc_ins
     ;
 
   # Doctypes
@@ -95,6 +97,20 @@ rule
     : T_COMMENT { on_comment(val[0]) }
     ;
 
+  # Processing Instructions
+
+  proc_ins
+    # <?xml?>
+    : T_PROC_INS_START T_PROC_INS_NAME T_PROC_INS_END
+      {
+        on_proc_ins(val[1])
+      }
+    | T_PROC_INS_START T_PROC_INS_NAME T_TEXT T_PROC_INS_END
+      {
+        on_proc_ins(val[1], val[2])
+      }
+    ;
+
   # Elements
 
   element_open
@@ -155,6 +171,7 @@ rule
     ;
 
   # XML declarations
+
   xmldecl
     : T_XML_DECL_START attributes T_XML_DECL_END { on_xml_decl(val[1]) }
     ;
@@ -307,6 +324,15 @@ Unexpected #{name} with value #{value.inspect} on line #{@line}:
     return Comment.new(:text => text)
   end
 
+  ##
+  # @param [String] name
+  # @param [String] text
+  # @return [Oga::XML::ProcessingInstruction]
+  #
+  def on_proc_ins(name, text = nil)
+    return ProcessingInstruction.new(:name => name, :text => text)
+  end
+
   ##
   # @param [Array] attributes
   # @return [Oga::XML::XmlDeclaration]

lib/oga/xml/processing_instruction.rb

@@ -0,0 +1,39 @@
+module Oga
+  module XML
+    ##
+    # Class used for storing information about a single processing instruction.
+    #
+    # @!attribute [rw] name
+    #  @return [String]
+    #
+    class ProcessingInstruction < CharacterNode
+      attr_accessor :name
+
+      ##
+      # @param [Hash] options
+      #
+      # @option options [String] :name The name of the instruction.
+      # @see [Oga::XML::CharacterNode#initialize]
+      #
+      def initialize(options = {})
+        super
+
+        @name = options[:name]
+      end
+
+      ##
+      # @return [String]
+      #
+      def to_xml
+        return "<?#{name}#{text}?>"
+      end
+
+      ##
+      # @return [String]
+      #
+      def inspect
+        return "ProcessingInstruction(name: #{name.inspect} text: #{text.inspect})"
+      end
+    end # ProcessingInstruction
+  end # XML
+end # Oga

lib/oga/xml/pull_parser.rb

@@ -46,7 +46,8 @@ module Oga
       BLOCK_CALLBACKS = [
         :on_cdata,
         :on_comment,
-        :on_text
+        :on_text,
+        :on_proc_ins
       ]
 
       ##

spec/oga/xml/parser/processing_instruction_spec.rb

@@ -0,0 +1,35 @@
+require 'spec_helper'
+
+describe Oga::XML::Parser do
+  context 'empty processing instructions' do
+    before :all do
+      @node = parse('<?foo?>').children[0]
+    end
+
+    example 'return a ProcessingInstruction instance' do
+      @node.is_a?(Oga::XML::ProcessingInstruction).should == true
+    end
+
+    example 'set the name of the instruction' do
+      @node.name.should == 'foo'
+    end
+  end
+
+  context 'processing instructions with text' do
+    before :all do
+      @node = parse('<?foo bar ?>').children[0]
+    end
+
+    example 'return a ProcessingInstruction instance' do
+      @node.is_a?(Oga::XML::ProcessingInstruction).should == true
+    end
+
+    example 'set the name of the instruction' do
+      @node.name.should == 'foo'
+    end
+
+    example 'set the text of the instruction' do
+      @node.text.should == ' bar '
+    end
+  end
+end

spec/oga/xml/processing_instruction_spec.rb

@@ -0,0 +1,29 @@
+require 'spec_helper'
+
+describe Oga::XML::ProcessingInstruction do
+  context '#initialize' do
+    example 'set the name of the node' do
+      described_class.new(:name => 'foo').name.should == 'foo'
+    end
+
+    example 'set the text of the node' do
+      described_class.new(:text => 'foo').text.should == 'foo'
+    end
+  end
+
+  context '#to_xml' do
+    example 'conver the node into XML' do
+      node = described_class.new(:name => 'foo', :text => ' bar ')
+
+      node.to_xml.should == '<?foo bar ?>'
+    end
+  end
+
+  context '#inspect' do
+    example 'return the inspect value of the node' do
+      node = described_class.new(:name => 'foo', :text => ' bar ')
+
+      node.inspect.should == 'ProcessingInstruction(name: "foo" text: " bar ")'
+    end
+  end
+end

spec/oga/xml/pull_parser/processing_instruction_spec.rb

@@ -0,0 +1,25 @@
+require 'spec_helper'
+
+describe Oga::XML::PullParser do
+  context 'processing instructions' do
+    before :all do
+      @parser = described_class.new('<?foo bar ?>')
+
+      @node = nil
+
+      @parser.parse { |node| @node = node }
+    end
+
+    example 'return a ProcessingInstruction node' do
+      @node.is_a?(Oga::XML::ProcessingInstruction).should == true
+    end
+
+    example 'set the name of the node' do
+      @node.name.should == 'foo'
+    end
+
+    example 'set the text of the node' do
+      @node.text.should == ' bar '
+    end
+  end
+end