improved Mojo::DOM::HTML to handle slash between attributes more grac…

…efully
mojolicious · Feb 12, 2014 · 551ebfe · 551ebfe
1 parent b2170e8
commit 551ebfe
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 5 deletions.
diff --git a/Changes b/Changes
@@ -4,7 +4,8 @@
   - Added next_sibling and previous_sibling methods to Mojo::DOM.
   - Added last method to Mojo::Collection.
   - Improved many methods in Mojo::DOM to work with all node types.
-  - Improved Mojo::DOM::HTML performance.
+  - Improved Mojo::DOM::HTML to handle slash between attributes more
+    gracefully.
   - Fixed list parsing bug in Mojo::DOM::HTML.
 
 4.79  2014-02-11

diff --git a/lib/Mojo/DOM/HTML.pm b/lib/Mojo/DOM/HTML.pm
@@ -8,15 +8,15 @@ has 'xml';
 has tree => sub { ['root'] };
 
 my $ATTR_RE = qr/
-  ([^<>=\s]+)      # Key
+  ([^<>=\s\/]+|\/)   # Key
   (?:
     \s*=\s*
     (?:
-      "([^"]*?)"   # Quotation marks
+      "([^"]*?)"     # Quotation marks
     |
-      '([^']*?)'   # Apostrophes
+      '([^']*?)'     # Apostrophes
     |
-      ([^>\s]*)    # Unquoted
+      ([^>\s\/]*)    # Unquoted
     )
   )?
   \s*

diff --git a/t/mojo/dom.t b/t/mojo/dom.t
@@ -2295,6 +2295,14 @@ is $dom->find('div > ul li')->[2], undef, 'no result';
 is $dom->find('div > ul ul')->[0]->text, 'C', 'right text';
 is $dom->find('div > ul ul')->[1], undef, 'no result';
 
+# Slash between attributes
+$dom = Mojo::DOM->new('<input /type=checkbox/ value="/a/" checked/><br/>');
+is $dom->at('input')->{type},  'checkbox', 'right attribute';
+is $dom->at('input')->{value}, '/a/',      'right attribute';
+ok exists $dom->at('input')->{checked}, 'attribute exists';
+is $dom->at('input')->{checked}, undef, 'right attribute';
+is "$dom", '<input checked type="checkbox" value="/a/"><br>', 'right result';
+
 # Extra whitespace
 $dom = Mojo::DOM->new('< span>a< /span><b >b</b><span >c</ span>');
 is $dom->at('span')->text,     'a', 'right text';