fixed Mojo::DOM to support escaped selectors

mojolicious · Jul 14, 2010 · 8473569 · 8473569
1 parent 44a9909
commit 8473569
Show file tree

Hide file tree

Showing 3 changed files with 60 additions and 15 deletions.
diff --git a/Changes b/Changes
@@ -18,6 +18,7 @@ This file documents the revision history for Perl extension Mojolicious.
         - Improved Mojo::Template error messages.
         - Improved generated multipart messages to be 2 bytes shorter.
           (John Kingsley)
+        - Fixed Mojo::DOM to support escaped selectors.
         - Fixed json/data rendering with layouts in MojoX::Renderer.
         - Fixed Mojo::IOLoop to not stop unexpectedly.
         - Fixed graceful shutdown in Mojo::Server::Daemon::Prefork.

diff --git a/lib/Mojo/DOM.pm b/lib/Mojo/DOM.pm
@@ -19,11 +19,11 @@ __PACKAGE__->attr(tree => sub { ['root'] });
 # Regex
 my $CSS_ATTR_RE = qr/
     \[
-    (\w+)       # Key
+    (\w+)               # Key
     (?:
-    (\W)?       # Operator
+    (\W)?               # Operator
     =
-    "([^"]+)"   # Value
+    "((?:[^"]|\\")+)"   # Value
     )?
     \]
 /x;
@@ -32,13 +32,13 @@ my $CSS_ELEMENT_RE      = qr/^([^\.\#]+)/;
 my $CSS_ID_RE           = qr/\#([^\#]+)/;
 my $CSS_PSEUDO_CLASS_RE = qr/(?:\:(\w+)(?:\(([^\)]+)\))?)/;
 my $CSS_TOKEN_RE        = qr/
-    (\s*,\s*)?                        # Separator
-    ([\w\.\*\#]+)?                    # Element
-    ((?:\:\w+(?:\([^\)]+\))?)*)?      # Pseudo Class
-    ((?:\[\w+(?:\W?="[^"]+")?\])*)?   # Attributes
+    (\s*,\s*)?                                               # Separator
+    ((?:[\w\.\*\#]|\\[^0-9a-fA-F]|\\[0-9a-fA-F]{1,6}\s?)+)?  # Element
+    ((?:\:\w+(?:\([^\)]+\))?)*)?                             # Pseudo Class
+    ((?:\[\w+(?:\W?="(?:[^"]|\\")+")?\])*)?                  # Attributes
     (?:
     \s*
-    ([\>\+\~])                        # Combinator
+    ([\>\+\~])                                               # Combinator
     )?
 /x;
 my $XML_ATTR_RE = qr/
@@ -278,6 +278,21 @@ sub _compare {
     return 1;
 }
 
+sub _css_unescape {
+    my ($self, $value) = @_;
+
+    # Remove escaped newlines
+    $value =~ s/\\\n//g;
+
+    # Unescape unicode characters
+    $value =~ s/\\([0-9a-fA-F]{1,6})\s?/pack('U', hex $1)/gex;
+
+    # Remove backslash
+    $value =~ s/\\//g;
+
+    return $value;
+}
+
 sub _doctype {
     my ($self, $doctype, $current) = @_;
 
@@ -415,12 +430,15 @@ sub _parse_css {
 
         # Classes
         while ($element =~ /$CSS_CLASS_RE/g) {
-            push @$selector, ['attribute', 'class', qr/(?:^|\W+)$1(?:\W+|$)/];
+            my $class = $self->_css_unescape($1);
+            push @$selector,
+              ['attribute', 'class', qr/(?:^|\W+)$class(?:\W+|$)/];
         }
 
         # ID
         if ($element =~ /$CSS_ID_RE/) {
-            push @$selector, ['attribute', 'id', qr/^$1$/];
+            my $id = $self->_css_unescape($1);
+            push @$selector, ['attribute', 'id', qr/^$id$/];
         }
 
         # Pseudo classes
@@ -441,7 +459,7 @@ sub _parse_css {
             if ($value) {
 
                 # Quote
-                $value = quotemeta $value;
+                $value = quotemeta $self->_css_unescape($value);
 
                 # "^=" (begins with)
                 if ($op eq '^') { $regex = qr/^$value/ }

diff --git a/t/mojo/dom.t b/t/mojo/dom.t
@@ -7,7 +7,7 @@ use warnings;
 
 use utf8;
 
-use Test::More tests => 59;
+use Test::More tests => 81;
 
 # Homer gave me a kidney: it wasn't his, I didn't need it,
 # and it came postage due- but I appreciated the gesture!
@@ -163,6 +163,32 @@ is($dom->at('[foo="bar"]')->text, 'works', 'right text');
 is($dom->at('[foo="ba"]'),        undef,   'no result');
 is($dom->at('.tset')->text,       'works', 'right text');
 
-# Already decoded unicode snowman
-$dom->charset(undef)->parse('<div id="snowman">☃</div>');
-is($dom->at('#snowman')->text, '☃', 'right text');
+# Already decoded unicode snowman and quotes in selector
+$dom->charset(undef)->parse('<div id="sno&quot;wman">☃</div>');
+is($dom->at('[id="sno\"wman"]')->text, '☃', 'right text');
+
+# Unicode and escaped id selectors
+$dom->parse(
+    qq/<p><div id="☃x">Snowman<\/div><div class="x ♥">Heart<\/div><\/p>/);
+is($dom->at("#\\\n\\002603x")->text,               'Snowman', 'right text');
+is($dom->at('#\\2603 x')->text,                    'Snowman', 'right text');
+is($dom->at("#\\\n\\2603 x")->text,                'Snowman', 'right text');
+is($dom->at(qq/[id="\\\n\\2603 x"]/)->text,        'Snowman', 'right text');
+is($dom->at(qq/[id="\\\n\\002603x"]/)->text,       'Snowman', 'right text');
+is($dom->at(qq/[id="\\\\2603 x"]/)->text,          'Snowman', 'right text');
+is($dom->at("p #\\\n\\002603x")->text,             'Snowman', 'right text');
+is($dom->at('p #\\2603 x')->text,                  'Snowman', 'right text');
+is($dom->at("p #\\\n\\2603 x")->text,              'Snowman', 'right text');
+is($dom->at(qq/p [id="\\\n\\2603 x"]/)->text,      'Snowman', 'right text');
+is($dom->at(qq/p [id="\\\n\\002603x"]/)->text,     'Snowman', 'right text');
+is($dom->at(qq/p [id="\\\\2603 x"]/)->text,        'Snowman', 'right text');
+is($dom->at(".\\\n\\002665")->text,                'Heart',   'right text');
+is($dom->at('.\\2665')->text,                      'Heart',   'right text');
+is($dom->at("p .\\\n\\002665")->text,              'Heart',   'right text');
+is($dom->at('p .\\2665')->text,                    'Heart',   'right text');
+is($dom->at(qq/p [class\$="\\\n\\002665"]/)->text, 'Heart',   'right text');
+is($dom->at(qq/p [class\$="\\2665"]/)->text,       'Heart',   'right text');
+is($dom->at(qq/[class\$="\\\n\\002665"]/)->text,   'Heart',   'right text');
+is($dom->at(qq/[class\$="\\2665"]/)->text,         'Heart',   'right text');
+is($dom->at('.x')->text,                           'Heart',   'right text');
+is($dom->at('p .x')->text,                         'Heart',   'right text');