commit 2d547bcc6c9af8a65973dae1bc804b08445abaf1
author Akron <nils@diewald-online.de> Sat Jul 04 10:34:35 2020 +0200
committer Akron <diewald@ids-mannheim.de> Sat Jul 04 11:58:23 2020 +0200
tree cdb6346a80432894224a5e420ac174afda8e70ff
parent 5ca6efc5fee7874b599841685f03c723a5fe6d32

Fix a bug in delHTMLcom where comments were left open

Change-Id: I424ac394fa7eaad9a2b62c61761d2de6720870a2

.gitignore

diff --git a/.gitignore b/.gitignore
index 1c83e3e..9071e73 100644
--- a/.gitignore
+++ b/.gitignore
@@ -2,3 +2,7 @@
 MYMETA*
 Makefile
 pm_to_blib
+\#*
+*~
+.*
+!.gitignore

lib/KorAP/XML/TEI.pm

diff --git a/lib/KorAP/XML/TEI.pm b/lib/KorAP/XML/TEI.pm
index 12131e7..7a2174d 100644
--- a/lib/KorAP/XML/TEI.pm
+++ b/lib/KorAP/XML/TEI.pm
@@ -9,7 +9,9 @@
   # sometimes is not desirable (e.g.: '...<!-- comment -->\n<w>token</w>...' would lead to '... <w>token</w>...' in $buf_in).
   # removing comments before processing the line, prevents this situation.
 
-  my ( $pfx, $sfx );
+  my ( $pfx, $sfx ) = ('','');
+
+ CHECK:
 
   while ( $html =~ s/<!--.*?-->//g ){}; # remove all comments in actual line
 
@@ -26,7 +28,7 @@
     }
 
     $html = "$pfx$sfx";
-
+    goto CHECK;
   }
 
   if ( $html =~ s/^\s*$// ){ # get next line and feed it also to this sub, if actual line is empty or only contains whitespace

t/tei.t

diff --git a/t/tei.t b/t/tei.t
index 292ed19..fcfd32e 100644
--- a/t/tei.t
+++ b/t/tei.t
@@ -27,4 +27,19 @@
 
 is(KorAP::XML::TEI::delHTMLcom($fh, '<!--'), "Test\n");
 
+seek($fh, 0, 0);
+
+print $fh <<'HTML';
+mehrzeiliger
+Kommentar
+  --><!-- Versuch
+-->ist <!-- a --><!-- b --> ein Test
+HTML
+
+seek($fh, 0, 0);
+
+is(KorAP::XML::TEI::delHTMLcom($fh, 'Dies <!--'), "Dies ist  ein Test");
+
+close($fh);
+
 done_testing;

xt/benchmark.pl

diff --git a/xt/benchmark.pl b/xt/benchmark.pl
index 35c0c65..ddd17a2 100644
--- a/xt/benchmark.pl
+++ b/xt/benchmark.pl
@@ -4,7 +4,7 @@
 use Dumbbench;
 use File::Basename 'dirname';
 use File::Spec::Functions qw/catfile rel2abs/;
-use File::Temp ':POSIX';
+use File::Temp 'tempfile';
 use FindBin;
 use Getopt::Long;
 
@@ -41,6 +41,15 @@
 );
 
 my $result;
+my ($fh, $filename) = tempfile();
+
+print $fh <<'HTML';
+mehrzeiliger
+Kommentar
+  --><!-- Versuch
+-->ist <!-- a --><!-- b --> ein Test
+HTML
+
 
 # Add benchmark instances
 $bench->add_instances(
@@ -61,11 +70,26 @@
       };
     }
   ),
+  Dumbbench::Instance::PerlSub->new(
+    name => 'delHTMLcom-long',
+    code => sub {
+      for (1..10_000) {
+        $result = KorAP::XML::TEI::delHTMLcom(
+          $fh,
+          "This <!--" . $_
+        );
+        seek($fh, 0, 0);
+      };
+    }
+  ),
 );
 
 # Run benchmarks
 $bench->run;
 
+# Clean up
+close($fh);
+
 # Output in a single row
 if ($columns) {
   unless ($no_header) {