better html_unescape tests

mojolicious · Apr 23, 2012 · 82b9f76 · 82b9f76
1 parent 59fccf1
commit 82b9f76
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 26 deletions.
diff --git a/Changes b/Changes
@@ -2,6 +2,7 @@ This file documents the revision history for Perl extension Mojolicious.
 
 2.88  2012-04-24
   - Improved documentation.
+  - Improved tests.
 
 2.87  2012-04-23
   - Improved html_escape performance and added pattern support.

diff --git a/lib/Mojo/Util.pm b/lib/Mojo/Util.pm
@@ -2375,9 +2375,6 @@ $REVERSE{$ENTITIES{$_}} //= $_ for reverse sort keys %ENTITIES;
 # "apos"
 $ENTITIES{'apos;'} = "\x{0027}";
 
-# Entities regex for html_unescape
-my $ENTITIES_RE = qr/&(?:\#((?:\d{1,7}|x[0-9A-Fa-f]{1,6}));|([\w\.]+;?))/;
-
 # Encode cache
 my %ENCODE;
 
@@ -2475,15 +2472,16 @@ sub html_escape {
   my ($string, $pattern) = @_;
   $pattern ||= '^\n\r\t !\#\$%\(-;=?-~';
   return $string unless $string =~ /[^$pattern]/;
-  $string =~ s/([$pattern])/_escape($1)/ge;
+  $string =~ s/([$pattern])/_encode($1)/ge;
   return $string;
 }
 
 # "Daddy, I'm scared. Too scared to even wet my pants.
 #  Just relax and it'll come, son."
 sub html_unescape {
   my $string = shift;
-  $string =~ s/$ENTITIES_RE/_unescape($1, $2)/ge;
+  $string
+    =~ s/&(?:\#((?:\d{1,7}|x[0-9A-Fa-f]{1,6}));|([\w\.]+;?))/_decode($1,$2)/ge;
   return $string;
 }
 
@@ -2698,10 +2696,25 @@ sub _adapt {
   return $k + (((PC_BASE - PC_TMIN + 1) * $delta) / ($delta + PC_SKEW));
 }
 
+# Helper for html_unescape
+sub _decode {
+
+  # Numeric
+  return substr($_[0], 0, 1) eq 'x' ? chr(hex $_[0]) : chr($_[0]) unless $_[1];
+
+  # Find entity name
+  my $rest   = '';
+  my $entity = $_[1];
+  while (length $entity) {
+    return "$ENTITIES{$entity}$rest" if exists $ENTITIES{$entity};
+    $rest = chop($entity) . $rest;
+  }
+  return "&$_[1]";
+}
+
 # Helper for html_escape
-sub _escape {
-  return "&$REVERSE{$_[0]}" if exists $REVERSE{$_[0]};
-  return '&#' . ord($_[0]) . ';';
+sub _encode {
+  return exists $REVERSE{$_[0]} ? "&$REVERSE{$_[0]}" : "&#@{[ord($_[0])]};";
 }
 
 sub _hmac {
@@ -2720,22 +2733,6 @@ sub _hmac {
   return unpack 'H*', $hash->($opad . $hash->($ipad . $string));
 }
 
-# Helper for html_unescape
-sub _unescape {
-
-  # Numeric
-  return substr($_[0], 0, 1) eq 'x' ? chr(hex $_[0]) : chr($_[0]) unless $_[1];
-
-  # Find entity name
-  my $rest   = '';
-  my $entity = $_[1];
-  while (length $entity) {
-    return "$ENTITIES{$entity}$rest" if exists $ENTITIES{$entity};
-    $rest = chop($entity) . $rest;
-  }
-  return "&$_[1]";
-}
-
 1;
 __END__
 

diff --git a/t/mojo/bytestream.t b/t/mojo/bytestream.t
@@ -203,8 +203,8 @@ $stream = b('foobar');
 is $stream->html_escape, 'foobar', 'right html escaped result';
 
 # html_unescape
-$stream = b('foobar&lt;baz&gt;&#x26;&#34;');
-is $stream->html_unescape, "foobar<baz>&\"", 'right html unescaped result';
+$stream = b('&#x3c;foo&#x3E;bar&lt;baz&gt;&#x26;&#34;');
+is $stream->html_unescape, "<foo>bar<baz>&\"", 'right html unescaped result';
 
 # html_unescape (special entities)
 $stream = b('foo &CounterClockwiseContourIntegral; bar &b.Sigma; &sup1baz');