fixed bug where Mojo::DOM::HTML could not handle certain broken tags

mojolicious · Jun 2, 2014 · dfd89c6 · dfd89c6
1 parent e8f341a
commit dfd89c6
Show file tree

Hide file tree

Showing 3 changed files with 8 additions and 2 deletions.
diff --git a/Changes b/Changes
@@ -1,5 +1,6 @@
 
-5.03  2014-06-01
+5.03  2014-06-02
+  - Fixed bug where Mojo::DOM::HTML could not handle certain broken tags.
 
 5.02  2014-05-31
   - Added multi-name support to cookie and signed_cookie methods in

diff --git a/lib/Mojo/DOM/HTML.pm b/lib/Mojo/DOM/HTML.pm
@@ -42,7 +42,7 @@ my $TOKEN_RE = qr/
       \s*
       [^<>\s]+                                      # Tag
       \s*
-      (?:(?:$ATTR_RE){0,32766})*                    # Attributes
+      (?:(?:$ATTR_RE){0,32766}){0,32766}+           # Attributes
     )>
   |
     (<)                                             # Runaway "<"

diff --git a/t/mojo/dom.t b/t/mojo/dom.t
@@ -2363,4 +2363,9 @@ my $huge = '<div ' . ('a=b ' x 32768) . '>Test</div>';
 $dom = Mojo::DOM->new($huge);
 is $dom->at('div[a=b]')->text, 'Test', 'right text';
 
+# Broken tag
+$huge = '<br< abc abc abc abc abc abc abc abc<p>Test</p>';
+$dom = Mojo::DOM->new($huge);
+is $dom->at('p')->text, 'Test', 'right text';
+
 done_testing();