X-Git-Url: http://wagnertech.de/git?a=blobdiff_plain;f=SL%2FHTML%2FUtil.pm;h=a54425d837c7b00f5d7370d18748d0aa5177b280;hb=f6191560e6db010cd9bf8b0854174d772b49b5c2;hp=72a7a7f57481a3522833d04bfc45740d3fccc27a;hpb=e09fbb1194ebfbb3b278bae40d31c67e416cf919;p=kivitendo-erp.git

diff --git a/SL/HTML/Util.pm b/SL/HTML/Util.pm
index 72a7a7f57..a54425d83 100644
--- a/SL/HTML/Util.pm
+++ b/SL/HTML/Util.pm
@@ -7,9 +7,11 @@ use HTML::Parser;
 
 my %stripper;
 my %entities = (
-  'lt'  => '<',
-  'gt'  => '>',
-  'amp' => '&',
+  'lt'   => '<',
+  'gt'   => '>',
+  'amp'  => '&',
+  'nbsp' => ' ',   # should be => "\x{00A0}", but this can lead to problems with
+                   # a non-visible character in csv-exports for example
 );
 
 sub strip {
@@ -17,10 +19,15 @@ sub strip {
 
   my $value = !ref($class_or_value) && (($class_or_value // '') eq 'SL::HTML::Util') ? $_[1] : $class_or_value;
 
+  return '' unless defined $value;
+
+  # Remove HTML comments.
+  $value =~ s{ <!-- .*? --> }{}gx;
+
   if (!%stripper) {
     %stripper = ( parser => HTML::Parser->new );
 
-    $stripper{parser}->handler(text => sub { $stripper{text} .= $_[1]; });
+    $stripper{parser}->handler(text => sub { $stripper{text} .= ' ' . $_[1]; });
   }
 
   $stripper{text} = '';
@@ -28,6 +35,8 @@ sub strip {
   $stripper{parser}->eof;
 
   $stripper{text} =~ s{\&([^;]+);}{ $entities{$1} || "\&$1;" }eg;
+  $stripper{text} =~ s{^ +| +$}{}g;
+  $stripper{text} =~ s{ {2,}}{ }g;
 
   return delete $stripper{text};
 }