From: Sven Schöling <s.schoeling@linet-services.de>
Date: Mon, 21 Jul 2014 09:08:44 +0000 (+0200)
Subject: HTML::Util: Tests, und Bugfixes für ein paar Randbedingungen
X-Git-Tag: release-3.2.0beta~392
X-Git-Url: http://wagnertech.de/git?a=commitdiff_plain;h=06a0f32d8602a6b79e51b0fcd51edb67e780a003;p=kivitendo-erp.git

HTML::Util: Tests, und Bugfixes für ein paar Randbedingungen
---

diff --git a/SL/HTML/Util.pm b/SL/HTML/Util.pm
index ed5ffdec9..c62fd9202 100644
--- a/SL/HTML/Util.pm
+++ b/SL/HTML/Util.pm
@@ -17,7 +17,10 @@ sub strip {
 
   my $value = !ref($class_or_value) && (($class_or_value // '') eq 'SL::HTML::Util') ? $_[1] : $class_or_value;
 
-  return '' unless $value;
+  return '' unless defined $value;
+
+  # Remove HTML comments.
+  $value =~ s{ <!-- .*? --> }{}gx;
 
   if (!%stripper) {
     %stripper = ( parser => HTML::Parser->new );
diff --git a/t/html/util.t b/t/html/util.t
new file mode 100644
index 000000000..1b9bb80ab
--- /dev/null
+++ b/t/html/util.t
@@ -0,0 +1,24 @@
+use Test::More;
+
+use_ok 'SL::HTML::Util';
+
+sub test {
+  is(SL::HTML::Util::strip($_[0]), $_[1], "$_[2] (direct invocation)");
+  is(SL::HTML::Util->strip($_[0]), $_[1], "$_[2] (class invocation)");
+}
+
+test undef, '', 'undef';
+test 0, '0', '0';
+test '0 but true', '0 but true', 'zero but true';
+test '<h1>title</h1>', 'title', 'standard case';
+test '<h1>title</h2>', 'title', 'imbalanced html';
+test 'walter &amp; walter', 'walter & walter', 'known entities';
+test 'Walter&Walter; Chicago', 'Walter&Walter; Chicago', 'unknown entities';
+test '<h1>title</h1', 'title', 'invalid html 1';
+
+# This happens when someone copies a block from MS Word. The HTML that is
+# generated contains style information in comments. These styles can contain
+# sgml. HTML::Parser does not handle these properly.
+test '<!-- style: <>  -->title', 'title', 'nested stuff in html comments';
+
+done_testing;