fixed whitespace bug in Mojo::DOM::HTML (closes #467)

mojolicious · Mar 15, 2013 · 4d90541 · 4d90541
1 parent bc58398
commit 4d90541
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 0 deletions.
diff --git a/Changes b/Changes
@@ -1,6 +1,7 @@
 
 3.91  2013-03-15
   - Improved tests.
+  - Fixed whitespace bug in Mojo::DOM::HTML.
 
 3.90  2013-03-14
   - Added direct array access for parsed parameters to Mojo::Parameters.

diff --git a/lib/Mojo/DOM/HTML.pm b/lib/Mojo/DOM/HTML.pm
@@ -42,6 +42,7 @@ my $TOKEN_RE = qr/
     <(
       \s*
       [^>\s]+                                       # Tag
+      \s*
       (?:$ATTR_RE)*                                 # Attributes
     )>
   )??

diff --git a/t/mojo/dom.t b/t/mojo/dom.t
@@ -2135,6 +2135,13 @@ 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';
 
+# Extra whitespace
+$dom = Mojo::DOM->new('< span>a< /span><b >b</b><span >c</ span>');
+is $dom->at('span')->text,     'a', 'right text';
+is $dom->at('span + b')->text, 'b', 'right text';
+is $dom->at('b + span')->text, 'c', 'right text';
+is "$dom", '<span>a</span><b>b</b><span>c</span>', 'right result';
+
 # Bad charset
 $dom = Mojo::DOM->new->charset('doesnotexist');
 $dom->parse(qq{<html><div id="a">A</div></html>});