test: add XSS regression tests

lib/MarkdownParser.pm

@@ -379,16 +379,14 @@ s/\x01F$i\x02/<$part->{tag}>@{[escape_html($part->{content})]}<\/$part->{tag}>/;
 
 sub is_safe_url {
     my ($url) = @_;
-    my $normalized = $url // '';
+    my $normalized = decode_url_escapes($url);
 
     $normalized =~ s/^\s+//;
     $normalized =~ s/\s+$//;
-    $normalized =~ s/%([0-9A-Fa-f]{2})/chr(hex($1))/eg
+    my $scheme_check = $normalized;
-      while $normalized =~ /%[0-9A-Fa-f]{2}/;
+    $scheme_check =~ s/[\x00-\x20\x7f]+//g;
-    $normalized =~ s/&#x([0-9A-Fa-f]+);?/chr(hex($1))/eg;
-    $normalized =~ s/&#(\d+);?/chr($1)/eg;
 
-    if ( $normalized =~ /^([a-z][a-z0-9+\-.]*):/i ) {
+    if ( $scheme_check =~ /^([a-z][a-z0-9+\-.]*):/i ) {
         my $scheme = lc $1;
         return 1
           if $scheme eq 'http' || $scheme eq 'https' || $scheme eq 'mailto';
@@ -398,6 +396,28 @@ sub is_safe_url {
     return 1;
 }
 
+sub decode_url_escapes {
+    my ($value) = @_;
+    my $decoded = $value // '';
+
+    for ( 1 .. 8 ) {
+        my $before = $decoded;
+        $decoded =~ s/%([0-9A-Fa-f]{2})/chr(hex($1))/eg;
+        $decoded =~ s/&#x([0-9A-Fa-f]+);?/_safe_chr(hex($1))/eg;
+        $decoded =~ s/&#(\d+);?/_safe_chr($1)/eg;
+        last if $decoded eq $before;
+    }
+
+    return $decoded;
+}
+
+sub _safe_chr {
+    my ($codepoint) = @_;
+    return '' if !defined $codepoint;
+    return '' if $codepoint > 0x10FFFF;
+    return chr($codepoint);
+}
+
 sub escape_html {
     my ($text) = @_;
     $text =~ s/&/&amp;/g;

t/04-links-images.t

@@ -2,7 +2,7 @@
 use strict;
 use warnings;
 
-use Test::More tests => 10;
+use Test::More tests => 17;
 use MarkdownParser;
 
 my $parser = MarkdownParser->new();
@@ -48,3 +48,38 @@ is( $parser->parse("![Image](file:///etc/passwd)"),
     "<p>Image</p>\n", "File protocol blocked in images" );
 is( $parser->parse("![Image](javascript:%2f%2falert('XSS'))"),
     "<p>Image</p>\n", "Encoded JavaScript protocol blocked in images" );
+is(
+    $parser->parse("[Click me](javascript&#10;:alert('XSS'))"),
+    "<p>Click me</p>\n",
+    "JavaScript protocol with numeric newline entity blocked"
+);
+is(
+    $parser->parse("[Click me](java&#x0D;script:alert('XSS'))"),
+    "<p>Click me</p>\n",
+    "JavaScript protocol with hex carriage return entity blocked"
+);
+is(
+    $parser->parse("[Click me](javascr&#x69;pt%3Aalert('XSS'))"),
+    "<p>Click me</p>\n",
+    "Mixed encoded JavaScript protocol blocked"
+);
+is(
+    $parser->parse("![Image](java&#9;script:alert('XSS'))"),
+    "<p>Image</p>\n",
+    "JavaScript protocol with tab entity blocked in images"
+);
+is(
+    $parser->parse("[email](mailto:user\@example.com)"),
+    "<p><a href=\"mailto:user\@example.com\">email</a></p>\n",
+    "Mailto protocol remains allowed"
+);
+is(
+    $parser->parse("[safe](%68%74%74%70%73://example.com/path)"),
+    "<p><a href=\"%68%74%74%70%73://example.com/path\">safe</a></p>\n",
+    "Percent-encoded https scheme remains allowed"
+);
+is(
+    $parser->parse("[relative](/docs/java script:guide)"),
+    "<p><a href=\"/docs/java script:guide\">relative</a></p>\n",
+    "Relative URL with colon in path remains allowed"
+);