Fix for lexing HTML quoted attrs followed by "/>"

This ensures that when using input such as <a href="foo"/> the "/" is
not part of the attribute value.

ext/ragel/base_lexer.rl

@@ -52,11 +52,6 @@
         if ( fc == '\n' ) lines++;
     }
 
-    action hold_and_return {
-        fhold;
-        fret;
-    }
-
     whitespace = [ \t];
     ident_char = [a-zA-Z0-9\-_];
     identifier = ident_char+;
@@ -375,6 +370,11 @@
         };
     *|;
 
+    action hold_start_element_head {
+        fhold;
+        fnext element_head;
+    }
+
     # Characters that can be used for unquoted HTML attribute values.
     # See https://html.spec.whatwg.org/multipage/introduction.html#intro-early-example
     # for more info.
@@ -384,8 +384,10 @@
 
     # Machine used for processing HTML attribute values.
     html_attribute_value := |*
-        squote => start_string_squote;
+        squote | dquote => {
-        dquote => start_string_dquote;
+            fhold;
+            fnext html_attribute_value_quoted;
+        };
 
         # Unquoted attribute values are lexed as if they were single quoted
         # strings.
@@ -397,14 +399,23 @@
             callback_simple(id_on_string_squote);
         };
 
-        any => hold_and_return;
+        any => hold_start_element_head;
+    *|;
+
+    # Machine specifically used when dealing with quoted HTML attributes. This
+    # ensures that input such as <a href="foo"/> doesn't result in "/" being
+    # considered part of the attribute value.
+    html_attribute_value_quoted := |*
+        squote => start_string_squote;
+        dquote => start_string_dquote;
+        any    => hold_start_element_head;
     *|;
 
     # Machine used for processing XML attribute values.
     xml_attribute_value := |*
         squote => start_string_squote;
         dquote => start_string_dquote;
-        any    => hold_and_return;
+        any    => hold_start_element_head;
     *|;
 
     # Machine used for processing the contents of an element's starting tag.
@@ -429,11 +440,11 @@
         '=' => {
             if ( html_p )
             {
-                fcall html_attribute_value;
+                fnext html_attribute_value;
             }
             else
             {
-                fcall xml_attribute_value;
+                fnext xml_attribute_value;
             }
         };
 

spec/oga/xml/lexer/elements_spec.rb

@@ -171,6 +171,28 @@ describe Oga::XML::Lexer do
         [:T_ELEM_END, nil, 1]
       ]
     end
+
+    describe 'without a space before the closing tag' do
+      it 'lexes a void element' do
+        lex('<br/>').should == [
+          [:T_ELEM_START, nil, 1],
+          [:T_ELEM_NAME, 'br', 1],
+          [:T_ELEM_END, nil, 1]
+        ]
+      end
+
+      it 'lexes a void element with an attribute' do
+        lex('<br class="foo"/>').should == [
+          [:T_ELEM_START, nil, 1],
+          [:T_ELEM_NAME, 'br', 1],
+          [:T_ATTR, 'class', 1],
+          [:T_STRING_DQUOTE, nil, 1],
+          [:T_STRING_BODY, 'foo', 1],
+          [:T_STRING_DQUOTE, nil, 1],
+          [:T_ELEM_END, nil, 1]
+        ]
+      end
+    end
   end
 
   describe 'elements with namespaces' do

spec/oga/xml/lexer/html_attributes_spec.rb

@@ -50,5 +50,17 @@ describe Oga::XML::Lexer do
         [:T_ELEM_END, nil, 1]
       ]
     end
+
+    it 'lexes an attribute with an unquoted value containing a slash' do
+      lex_html('<a href=foo/></a>').should == [
+        [:T_ELEM_START, nil, 1],
+        [:T_ELEM_NAME, 'a', 1],
+        [:T_ATTR, 'href', 1],
+        [:T_STRING_SQUOTE, nil, 1],
+        [:T_STRING_BODY, 'foo/', 1],
+        [:T_STRING_SQUOTE, nil, 1],
+        [:T_ELEM_END, nil, 1]
+      ]
+    end
   end
 end

spec/oga/xml/lexer/html_void_elements_spec.rb

@@ -3,7 +3,7 @@ require 'spec_helper'
 describe Oga::XML::Lexer do
   describe 'HTML void elements' do
     it 'lexes a void element that omits the closing /' do
-      lex('<link>', :html => true).should == [
+      lex_html('<link>').should == [
         [:T_ELEM_START, nil, 1],
         [:T_ELEM_NAME, 'link', 1],
         [:T_ELEM_END, nil, 1]
@@ -11,7 +11,7 @@ describe Oga::XML::Lexer do
     end
 
     it 'lexes a upper case void element' do
-      lex('<BR>', :html => true).should == [
+      lex_html('<BR>').should == [
         [:T_ELEM_START, nil, 1],
         [:T_ELEM_NAME, "BR", 1],
         [:T_ELEM_END, nil, 1]
@@ -19,7 +19,7 @@ describe Oga::XML::Lexer do
     end
 
     it 'lexes text after a void element' do
-      lex('<link>foo', :html => true).should == [
+      lex_html('<link>foo').should == [
         [:T_ELEM_START, nil, 1],
         [:T_ELEM_NAME, 'link', 1],
         [:T_ELEM_END, nil, 1],
@@ -28,7 +28,7 @@ describe Oga::XML::Lexer do
     end
 
     it 'lexes a void element inside another element' do
-      lex('<head><link></head>', :html => true).should == [
+      lex_html('<head><link></head>').should == [
         [:T_ELEM_START, nil, 1],
         [:T_ELEM_NAME, 'head', 1],
         [:T_ELEM_START, nil, 1],
@@ -39,7 +39,7 @@ describe Oga::XML::Lexer do
     end
 
     it 'lexes a void element inside another element with whitespace' do
-      lex("<head><link>\n</head>", :html => true).should == [
+      lex_html("<head><link>\n</head>").should == [
         [:T_ELEM_START, nil, 1],
         [:T_ELEM_NAME, 'head', 1],
         [:T_ELEM_START, nil, 1],
@@ -49,5 +49,51 @@ describe Oga::XML::Lexer do
         [:T_ELEM_END, nil, 2]
       ]
     end
+
+    it 'lexes a void element with an unquoted attribute value' do
+      lex_html('<br class=foo />').should == [
+        [:T_ELEM_START, nil, 1],
+        [:T_ELEM_NAME, 'br', 1],
+        [:T_ATTR, 'class', 1],
+        [:T_STRING_SQUOTE, nil, 1],
+        [:T_STRING_BODY, 'foo', 1],
+        [:T_STRING_SQUOTE, nil, 1],
+        [:T_ELEM_END, nil, 1]
+      ]
+    end
+
+    describe 'without a space before the closing tag' do
+      it 'lexes a void element' do
+        lex_html('<br/>').should == [
+          [:T_ELEM_START, nil, 1],
+          [:T_ELEM_NAME, 'br', 1],
+          [:T_ELEM_END, nil, 1]
+        ]
+      end
+
+      it 'lexes a void element with an attribute' do
+        lex_html('<br class="foo"/>').should == [
+          [:T_ELEM_START, nil, 1],
+          [:T_ELEM_NAME, 'br', 1],
+          [:T_ATTR, 'class', 1],
+          [:T_STRING_DQUOTE, nil, 1],
+          [:T_STRING_BODY, 'foo', 1],
+          [:T_STRING_DQUOTE, nil, 1],
+          [:T_ELEM_END, nil, 1]
+        ]
+      end
+
+      it 'lexes a void element with an unquoted attribute value' do
+        lex_html('<br class=foo/>').should == [
+          [:T_ELEM_START, nil, 1],
+          [:T_ELEM_NAME, 'br', 1],
+          [:T_ATTR, 'class', 1],
+          [:T_STRING_SQUOTE, nil, 1],
+          [:T_STRING_BODY, 'foo/', 1],
+          [:T_STRING_SQUOTE, nil, 1],
+          [:T_ELEM_END, nil, 1]
+        ]
+      end
+    end
   end
 end