remove leading and trailing spaces in references and replyto

[debbugs.git] / Debbugs / MIME.pm

diff --git a/Debbugs/MIME.pm b/Debbugs/MIME.pm
index 909191549ca13b380a2f815ea5da8d97c1ee5495..aae3b58d9b543329655433c20cda808333bf06c1 100644 (file)
--- a/Debbugs/MIME.pm
+++ b/Debbugs/MIME.pm
@@ -10,6 +10,8 @@
 
 package Debbugs::MIME;
 
+=encoding utf8
+
 =head1 NAME
 
 Debbugs::MIME -- Mime handling routines for debbugs
@@ -47,21 +49,18 @@ BEGIN {
     $EXPORT_TAGS{all} = [@EXPORT_OK];
 }
 
-use File::Path;
-use File::Temp qw();
+use File::Path qw(remove_tree);
+use File::Temp qw(tempdir);
 use MIME::Parser;
 
 use POSIX qw(strftime);
 use List::MoreUtils qw(apply);
 
 # for convert_to_utf8
-use Debbugs::UTF8 qw(convert_to_utf8 encode_utf8_safely);
+use Debbugs::UTF8 qw(convert_to_utf8);
 
-# for decode_rfc1522
-use MIME::WordDecoder qw();
+# for decode_rfc1522 and encode_rfc1522
 use Encode qw(decode encode encode_utf8 decode_utf8 is_utf8);
-
-# for encode_rfc1522
 use MIME::Words qw();
 
 sub getmailbody
@@ -95,7 +94,7 @@ sub parse
     my (@headerlines, @bodylines);
 
     my $parser = MIME::Parser->new();
-    my $tempdir = File::Temp::tempdir();
+    my $tempdir = tempdir(CLEANUP => 1);
     $parser->output_under($tempdir);
     my $entity = eval { $parser->parse_data($_[0]) };
 
@@ -118,6 +117,9 @@ sub parse
        my @msg = split /\n/, $_[0];
        my $i;
 
+        # assume us-ascii unless charset is set; probably bad, but we
+        # really shouldn't get to this point anyway
+        my $charset = 'us-ascii';
        for ($i = 0; $i <= $#msg; ++$i) {
            $_ = $msg[$i];
            last unless length;
@@ -125,13 +127,15 @@ sub parse
                ++$i;
                $_ .= "\n" . $msg[$i];
            }
+            if (/charset=\"([^\"]+)\"/) {
+                $charset = $1;
+            }
            push @headerlines, $_;
        }
-
-       @bodylines = @msg[$i .. $#msg];
+       @bodylines = map {convert_to_utf8($_,$charset)} @msg[$i .. $#msg];
     }
 
-    rmtree $tempdir, 0, 1;
+    remove_tree($tempdir,{verbose => 0, safe => 1});
 
     # Remove blank lines.
     shift @bodylines while @bodylines and $bodylines[0] !~ /\S/;
@@ -189,7 +193,7 @@ sub create_mime_message{
      die "The third argument to create_mime_message must be an arrayref" unless ref($attachments) eq 'ARRAY';
 
      if ($include_date) {
-        my %headers = apply {lc($_)} @{$headers};
+        my %headers = apply {defined $_ ? lc($_) : ''} @{$headers};
         if (not exists $headers{date}) {
             push @{$headers},
                 ('Date',
@@ -202,8 +206,8 @@ sub create_mime_message{
      # MIME::Entity is stupid, and doesn't rfc1522 encode its headers, so we do it for it.
      my $msg = MIME::Entity->build('Content-Type' => 'text/plain; charset=utf-8',
                                   'Encoding'     => 'quoted-printable',
-                                  (map{encode_rfc1522(encode_utf8_safely($_))} @{$headers}),
-                                  Data    => encode_utf8_safely($body),
+                                  (map{encode_rfc1522(encode_utf8(defined $_ ? $_:''))} @{$headers}),
+                                  Data    => encode_utf8($body),
                                  );
 
      # Attach the attachments
@@ -248,24 +252,27 @@ Turn RFC-1522 names into the UTF-8 equivalent.
 
 =cut
 
-BEGIN {
-    # Set up the default RFC1522 decoder, which turns all charsets that
-    # are supported into the appropriate UTF-8 charset.
-    MIME::WordDecoder->default(new MIME::WordDecoder(
-       ['*' => \&convert_to_utf8,
-       ]));
-}
-
 sub decode_rfc1522 {
     my ($string) = @_;
 
     # this is craptacular, but leading space is hacked off by unmime.
     # Save it.
     my $leading_space = '';
-    $leading_space = $1 if $string =~ s/^(\s+)//;
-    # unmime calls the default MIME::WordDecoder handler set up at
-    # initialization time.
-    return $leading_space . MIME::WordDecoder::unmime($string);
+    $leading_space = $1 if $string =~ s/^(\ +)//;
+    # we must do this to switch off the utf8 flag before calling decode_mimewords
+    $string = encode_utf8($string);
+    my @mime_words = MIME::Words::decode_mimewords($string);
+    my $tmp = $leading_space .
+        join('',
+             (map {
+                 if (@{$_} > 1) {
+                     convert_to_utf8(${$_}[0],${$_}[1]);
+                 } else {
+                     decode_utf8(${$_}[0]);
+                 }
+             } @mime_words)
+            );
+    return $tmp;
 }
 
 =head2 encode_rfc1522
@@ -285,8 +292,10 @@ sub encode_rfc1522 {
 
      # handle being passed undef properly
      return undef if not defined $rawstr;
-     # convert to octets if is_utf8 is set
-     $rawstr= encode_utf8_safely($rawstr);
+
+     # convert to octets if we are given a string in perl's internal
+     # encoding
+     $rawstr= encode_utf8($rawstr) if is_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. We also consider parenthesis and "