Merge branch 'don/processencoding' of git+ssh://git.donarmstrong.com/srv/git/debbugs...

[debbugs.git] / Debbugs / MIME.pm

diff --git a/Debbugs/MIME.pm b/Debbugs/MIME.pm
index 0d35b0fb85e8b259fdf37d3faad373dc5f51087f..05534e3e054c47d2cb281f683aaacf1a0ac7a52e 100644 (file)
--- a/Debbugs/MIME.pm
+++ b/Debbugs/MIME.pm
@@ -10,6 +10,24 @@
 
 package Debbugs::MIME;
 
+=head1 NAME
+
+Debbugs::MIME -- Mime handling routines for debbugs
+
+=head1 SYNOPSIS
+
+ use Debbugs::MIME qw(parse decode_rfc1522);
+
+=head1 DESCRIPTION
+
+
+=head1 BUGS
+
+None known.
+
+=cut
+
+use warnings;
 use strict;
 
 use base qw(Exporter);
@@ -22,6 +40,7 @@ BEGIN {
 }
 
 use File::Path;
+use File::Temp qw();
 use MIME::Parser;
 
 use POSIX qw(strftime);
@@ -39,7 +58,7 @@ sub getmailbody
     my $entity = shift;
     my $type = $entity->effective_type;
     if ($type eq 'text/plain' or
-           ($type =~ m#text/# and $type ne 'text/html') or
+           ($type =~ m#text/?# and $type ne 'text/html') or
            $type eq 'application/pgp') {
        return $entity->bodyhandle;
     } elsif ($type eq 'multipart/alternative') {
@@ -65,8 +84,8 @@ sub parse
     my (@headerlines, @bodylines);
 
     my $parser = MIME::Parser->new();
-    mkdir "mime.tmp.$$", 0777;
-    $parser->output_under("mime.tmp.$$");
+    my $tempdir = File::Temp::tempdir();
+    $parser->output_under($tempdir);
     my $entity = eval { $parser->parse_data($_[0]) };
 
     if ($entity and $entity->head->tags) {
@@ -94,14 +113,19 @@ sub parse
        @bodylines = @msg[$i .. $#msg];
     }
 
-    rmtree "mime.tmp.$$", 0, 1;
+    rmtree $tempdir, 0, 1;
 
     # Remove blank lines.
     shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
 
     # Strip off RFC2440-style PGP clearsigning.
     if (@bodylines and $bodylines[0] =~ /^-----BEGIN PGP SIGNED/) {
-       shift @bodylines while @bodylines and length $bodylines[0];
+       shift @bodylines while @bodylines and
+           length $bodylines[0] and
+               # we currently don't strip \r; handle this for the
+               # time being, though eventually it should be stripped
+               # too, I think. [See #565981]
+               $bodylines[0] ne "\r";
        shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
        for my $findsig (0 .. $#bodylines) {
            if ($bodylines[$findsig] =~ /^-----BEGIN PGP SIGNATURE/) {
@@ -205,9 +229,7 @@ sub convert_to_utf8 {
      return $data if $charset eq 'raw' or is_utf8($data,1);
      my $result;
      eval {
-         # this encode/decode madness is to make sure that the data
-         # really is valid utf8 and that the is_utf8 flag is off.
-         $result = encode("utf8",decode($charset,$data))
+        $result = decode($charset,$data);
      };
      if ($@) {
          warn "Unable to decode charset; '$charset' and '$data': $@";
@@ -262,10 +284,15 @@ sub encode_rfc1522 {
 
      # handle being passed undef properly
      return undef if not defined $rawstr;
+     if (is_utf8($rawstr)) {
+        $rawstr= encode_utf8($rawstr);
+     }
      # We process words in reverse so we can preserve spacing between
      # encoded words. This regex splits on word|nonword boundaries and
-     # nonword|nonword boundaries.
-     my @words = reverse split /(?:(?<=[\s\n])|(?=[\s\n]))/m, $rawstr;
+     # nonword|nonword boundaries. We also consider parenthesis and "
+     # to be nonwords to avoid escaping them in comments in violation
+     # of RFC1522
+     my @words = reverse split /(?:(?<=[\s\n\)\(\"])|(?=[\s\n\)\(\"]))/m, $rawstr;
 
      my $previous_word_encoded = 0;
      my $string = '';
@@ -287,7 +314,7 @@ sub encode_rfc1522 {
               if (length $encoded > 75) {
                    # Turn utf8 into the internal perl representation
                    # so . is a character, not a byte.
-                   my $tempstr = decode_utf8($word,Encode::FB_DEFAULT);
+                   my $tempstr = is_utf8($word)?$word:decode_utf8($word,Encode::FB_DEFAULT);
                    my @encoded;
                    # Strip it into 10 character long segments, and encode
                    # the segments