Added more HTML closing rules for colgroup/caption

lib/oga/xml/lexer.rb

@@ -66,8 +66,8 @@ module Oga
         'rp'       => NodeNameSet.new(%w{rb rt rtc rp}),
         'optgroup' => NodeNameSet.new(%w{optgroup}),
         'option'   => NodeNameSet.new(%w{option optgroup}),
-        'colgroup' => NodeNameSet.new(%w{thead tbody tfoot}),
+        'colgroup' => NodeNameSet.new(%w{thead tbody tfoot colgroup tr}),
-        'caption'  => NodeNameSet.new(%w{thead tbody tfoot}),
+        'caption'  => NodeNameSet.new(%w{thead tbody tfoot colgroup tr caption}),
         'thead'    => NodeNameSet.new(%w{thead tbody tfoot}),
         'tbody'    => NodeNameSet.new(%w{thead tbody tfoot}),
         'tfoot'    => NodeNameSet.new(%w{thead tbody tfoot}),

spec/oga/xml/lexer/html_closing_rules/caption_spec.rb

@@ -0,0 +1,27 @@
+require 'spec_helper'
+
+describe Oga::XML::Lexer do
+  describe 'HTML caption elements' do
+    it 'lexes an unclosed <caption> followed by another <caption>' do
+      lex_html('<caption>foo<caption>bar').should == [
+        [:T_ELEM_NAME, 'caption', 1],
+        [:T_TEXT, 'foo', 1],
+        [:T_ELEM_END, nil, 1],
+        [:T_ELEM_NAME, 'caption', 1],
+        [:T_TEXT, 'bar', 1],
+        [:T_ELEM_END, nil, 1]
+      ]
+    end
+
+    it 'lexes an unclosed <caption> followed by a <colgroup>' do
+      lex_html('<caption>foo<colgroup>bar').should == [
+        [:T_ELEM_NAME, 'caption', 1],
+        [:T_TEXT, 'foo', 1],
+        [:T_ELEM_END, nil, 1],
+        [:T_ELEM_NAME, 'colgroup', 1],
+        [:T_TEXT, 'bar', 1],
+        [:T_ELEM_END, nil, 1]
+      ]
+    end
+  end
+end

spec/oga/xml/lexer/html_closing_rules/colgroup_spec.rb

@@ -0,0 +1,16 @@
+require 'spec_helper'
+
+describe Oga::XML::Lexer do
+  describe 'HTML colgroup elements' do
+    it 'lexes an unclosed <colgroup> followed by another <colgroup>' do
+      lex_html('<colgroup>foo<colgroup>bar').should == [
+        [:T_ELEM_NAME, 'colgroup', 1],
+        [:T_TEXT, 'foo', 1],
+        [:T_ELEM_END, nil, 1],
+        [:T_ELEM_NAME, 'colgroup', 1],
+        [:T_TEXT, 'bar', 1],
+        [:T_ELEM_END, nil, 1]
+      ]
+    end
+  end
+end