Lexing/parsing of XML declaration tags.

This closes #12.

lib/oga/lexer.rl

@@ -376,6 +376,34 @@ module Oga
         any;
       *|;
 
+      # XML declaration tags
+      #
+      # http://www.w3.org/TR/REC-xml/#sec-prolog-dtd
+      #
+      xml_decl_start = '<?xml';
+      xml_decl_end   = '?>';
+
+      action start_xml_decl {
+        emit_buffer
+        add_token(:T_XML_DECL_START, nil)
+
+        start_buffer
+
+        fcall xml_decl;
+      }
+
+      # Machine that processes the contents of an XML declaration tag.
+      xml_decl := |*
+        xml_decl_end => {
+          emit_buffer
+          add_token(:T_XML_DECL_END, nil)
+
+          fret;
+        };
+
+        any;
+      *|;
+
       # Elements
       #
       # http://www.w3.org/TR/html-markup/syntax.html#syntax-elements
@@ -433,10 +461,11 @@ module Oga
       *|;
 
       main := |*
-        element_start => start_element;
-        doctype_start => start_doctype;
-        cdata_start   => start_cdata;
-        comment_start => start_comment;
+        element_start  => start_element;
+        doctype_start  => start_doctype;
+        cdata_start    => start_cdata;
+        comment_start  => start_comment;
+        xml_decl_start => start_xml_decl;
 
         # Enter the body of the tag. If HTML mode is enabled and the current
         # element is a void element we'll close it and bail out.

lib/oga/parser.y

@@ -14,6 +14,7 @@ token T_DOCTYPE_START T_DOCTYPE_END T_DOCTYPE_TYPE
 token T_CDATA_START T_CDATA_END
 token T_COMMENT_START T_COMMENT_END
 token T_ELEM_START T_ELEM_NAME T_ELEM_NS T_ELEM_END T_ATTR
+token T_XML_DECL_START T_XML_DECL_END
 
 options no_result_var
 
@@ -35,6 +36,7 @@ rule
     | comment
     | element
     | text
+    | xmldecl
     ;
 
   # Doctypes
@@ -119,6 +121,11 @@ rule
     | T_ATTR T_STRING { s(:attribute, val[0], val[1]) }
     ;
 
+  # XML declarations
+  xmldecl
+    : T_XML_DECL_START T_XML_DECL_END      { s(:xml_decl) }
+    | T_XML_DECL_START text T_XML_DECL_END { s(:xml_decl, val[1]) }
+
   # Plain text
 
   text

spec/oga/lexer/xml_declaration_spec.rb

@@ -0,0 +1,24 @@
+require 'spec_helper'
+
+describe Oga::Lexer do
+  context 'XML declaration tags' do
+    example 'lex a start tag' do
+      lex('<?xml').should == [[:T_XML_DECL_START, nil, 1]]
+    end
+
+    example 'lex a start and end tag' do
+      lex('<?xml?>').should == [
+        [:T_XML_DECL_START, nil, 1],
+        [:T_XML_DECL_END, nil, 1]
+      ]
+    end
+
+    example 'lex a tag with text inside it' do
+      lex('<?xml version="1.0" ?>').should == [
+        [:T_XML_DECL_START, nil, 1],
+        [:T_TEXT, ' version="1.0" ', 1],
+        [:T_XML_DECL_END, nil, 1]
+      ]
+    end
+  end
+end

spec/oga/parser/xml_declaration_spec.rb

@@ -0,0 +1,12 @@
+require 'spec_helper'
+
+describe Oga::Parser do
+  context 'XML declaration tags' do
+    example 'lex an XML declaration tag' do
+      parse('<?xml hello ?>').should == s(
+        :document,
+        s(:xml_decl, s(:text, ' hello '))
+      )
+    end
+  end
+end