- carp if the a log text record is in perl's internal encoding instead of octets

[debbugs.git] / Debbugs / Log.pm

diff --git a/Debbugs/Log.pm b/Debbugs/Log.pm
index 89b14d5d04d90642761dbe9ae3c8e7561ff50179..de639494f087788f11f48304fcdf271d494dab4e 100644 (file)
--- a/Debbugs/Log.pm
+++ b/Debbugs/Log.pm
@@ -51,6 +51,11 @@ The Debbugs::Log module provides a convenient way for scripts to read and
 write the .log files used by debbugs to store the complete textual records
 of all bug transactions.
 
+Debbugs::Log does not decode utf8 into perl's internal encoding or
+encode into utf8 from perl's internal encoding. For html records and
+all recips, this should probably be done. For other records, this should
+not be needed.
+
 =head2 The .log File Format
 
 .log files consist of a sequence of records, of one of the following four
@@ -201,7 +206,6 @@ sub new
              die "Unable to open bug log $bug_log for reading: $!";
     }
 
-    binmode($self->{logfh},':utf8');
     $self->{state} = 'kill-init';
     $self->{linenum} = 0;
     return $self;
@@ -386,8 +390,12 @@ sub write_log_records
        croak "record type '$type' with no text field" unless defined $record->{text};
        # I am not sure if we really want to croak here; but this is
        # almost certainly a bug if is_utf8 is on.
-       # croak "probably wrong encoding" if is_utf8($record->{text});
-       my ($text) = escape_log($record->{text});
+        my $text = $record->{text};
+        if (is_utf8($text)) {
+            carp('Record text was in the wrong encoding (perl internal instead of utf8 octets)');
+            $text = encode_utf8($text)
+        }
+       ($text) = escape_log($text);
        if ($type eq 'autocheck') {
            print {$logfh} "\01\n$text\03\n" or
                die "Unable to write to logfile: $!";