improved bad charset handling in Mojo::DOM::HTML

mojolicious · Mar 17, 2013 · 4eb1364 · 4eb1364
1 parent 99dff6e
commit 4eb1364
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 5 deletions.
diff --git a/Changes b/Changes
@@ -1,5 +1,6 @@
 
 3.91  2013-03-17
+  - Improved bad charset handling in Mojo::DOM::HTML.
   - Improved tests.
   - Fixed support for RFC 2817 in Mojo::Message::Request.
   - Fixed whitespace bug in Mojo::DOM::HTML.

diff --git a/lib/Mojo/DOM/HTML.pm b/lib/Mojo/DOM/HTML.pm
@@ -76,8 +76,10 @@ my %INLINE = map { $_ => 1 } (
 sub parse {
   my ($self, $html) = @_;
 
-  my $charset = $self->charset;
-  $html = decode($charset, $html) // return $self->charset(undef) if $charset;
+  if (my $charset = $self->charset) {
+    if (my $chars = decode $charset, $html) { $html = $chars }
+    else                                    { $self->charset(undef) }
+  }
 
   my $tree    = ['root'];
   my $current = $tree;

diff --git a/t/mojo/dom.t b/t/mojo/dom.t
@@ -2144,9 +2144,15 @@ 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>});
-is $dom->at('#a'), undef, 'no result';
-is "$dom", '', 'right result';
+$dom->parse('<html><div id="a">A</div></html>');
+is $dom->charset, undef, 'no charset';
+is $dom->at('#a')->text, 'A', 'right text';
+is "$dom", '<html><div id="a">A</div></html>', 'right result';
+$dom = Mojo::DOM->new->charset('UTF-8');
+$dom->parse(qq{<div id="invalid">\x89</div>});
+is $dom->charset, undef, 'no charset';
+is $dom->at('#invalid')->text, "\x89", 'right text';
+is "$dom", qq{<div id="invalid">\x89</div>}, 'right result';
 
 # Comments
 $dom = Mojo::DOM->new(<<EOF);