about summary refs log tree commit homepage
path: root/lib/PublicInbox/Eml.pm
diff options
context:
space:
mode:
Diffstat (limited to 'lib/PublicInbox/Eml.pm')
-rw-r--r--lib/PublicInbox/Eml.pm100
1 files changed, 79 insertions, 21 deletions
diff --git a/lib/PublicInbox/Eml.pm b/lib/PublicInbox/Eml.pm
index 571edc5c..d59d7c3f 100644
--- a/lib/PublicInbox/Eml.pm
+++ b/lib/PublicInbox/Eml.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2020 all contributors <meta@public-inbox.org>
+# Copyright (C) all contributors <meta@public-inbox.org>
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
 #
 # Lazy MIME parser, it still slurps the full message but keeps short
@@ -28,7 +28,7 @@ package PublicInbox::Eml;
 use strict;
 use v5.10.1;
 use Carp qw(croak);
-use Encode qw(find_encoding decode encode); # stdlib
+use Encode qw(find_encoding); # stdlib
 use Text::Wrap qw(wrap); # stdlib, we need Perl 5.6+ for $huge
 use MIME::Base64 3.05; # Perl 5.10.0 / 5.9.2
 use MIME::QuotedPrint 3.05; # ditto
@@ -51,7 +51,9 @@ $MIME_ENC{quotedprint} = $MIME_ENC{'quoted-printable'} = $MIME_ENC{qp};
 $MIME_DEC{quotedprint} = $MIME_DEC{'quoted-printable'} = $MIME_DEC{qp};
 $MIME_ENC{$_} = \&identity_codec for qw(7bit 8bit binary);
 
-my %DECODE_ADDRESS = map { $_ => 1 } qw(From To Cc Sender Reply-To);
+my %DECODE_ADDRESS = map {
+        ($_ => 1, "Resent-$_" => 1)
+} qw(From To Cc Sender Reply-To Bcc);
 my %DECODE_FULL = (
         Subject => 1,
         'Content-Description' => 1,
@@ -120,9 +122,10 @@ sub new {
                 my $hdr = substr($$ref, 0, $header_size_limit + 1);
                 hdr_truncate($hdr) if length($hdr) > $header_size_limit;
                 bless { hdr => \$hdr, crlf => $1 }, __PACKAGE__;
-        } else { # nothing useful
-                my $hdr = $$ref = '';
-                bless { hdr => \$hdr, crlf => "\n" }, __PACKAGE__;
+        } else { # just a body w/o header?
+                my $hdr = '';
+                my $eol = ($$ref =~ /(\r?\n)/) ? $1 : "\n";
+                bless { hdr => \$hdr, crlf => $eol, bdy => $ref }, __PACKAGE__;
         }
 }
 
@@ -141,6 +144,7 @@ sub header_raw {
         my $re = re_memo($_[1]);
         my @v = (${ $_[0]->{hdr} } =~ /$re/g);
         for (@v) {
+                utf8::decode($_); # SMTPUTF8
                 # for compatibility w/ Email::Simple::Header,
                 s/\s+\z//s;
                 s/\A\s+//s;
@@ -217,7 +221,7 @@ sub mp_descend ($$) {
                 # There's also a case where quoted text showed up in the
                 # preamble
                 # <20060515162817.65F0F1BBAE@citi.umich.edu>
-                unshift(@parts, new_sub(undef, \$pre)) if $pre =~ /:/s;
+                unshift(@parts, new_sub(undef, \$pre)) if index($pre, ':') >= 0;
                 return \@parts;
         }
         # "multipart", but no boundary found, treat as single part
@@ -234,6 +238,7 @@ sub mp_descend ($$) {
 # $cb - user-supplied callback sub
 # $arg - user-supplied arg (think pthread_create)
 # $once - unref body scalar during iteration
+# $all - used by IMAP server, only
 sub each_part {
         my ($self, $cb, $arg, $once, $all) = @_;
         my $p = mp_descend($self, $once // 0) or
@@ -329,11 +334,24 @@ sub body_set {
         undef;
 }
 
+# workaround https://rt.cpan.org/Public/Bug/Display.html?id=139622
+# Encode 2.87..3.12 leaks on croak, so we defer and croak ourselves
+our @enc_warn;
+my $enc_warn = sub { push @enc_warn, @_ };
+
 sub body_str_set {
-        my ($self, $body_str) = @_;
-        my $charset = ct($self)->{attributes}->{charset} or
-                Carp::confess('body_str was given, but no charset is defined');
-        body_set($self, \(encode($charset, $body_str, Encode::FB_CROAK)));
+        my ($self, $str) = @_;
+        my $cs = ct($self)->{attributes}->{charset} //
+                croak('body_str was given, but no charset is defined');
+        my $enc = find_encoding($cs) // croak "unknown encoding `$cs'";
+        my $tmp;
+        {
+                local @enc_warn;
+                local $SIG{__WARN__} = $enc_warn;
+                $tmp = $enc->encode($str, Encode::FB_WARN);
+                croak(@enc_warn) if @enc_warn;
+        };
+        body_set($self, \$tmp);
 }
 
 sub content_type { scalar header($_[0], 'Content-Type') }
@@ -347,14 +365,15 @@ sub header_set {
         $pfx .= ': ';
         my $len = 78 - length($pfx);
         @vals = map {;
+                utf8::encode(my $v = $_); # to bytes, support SMTPUTF8
                 # folding differs from Email::Simple::Header,
                 # we favor tabs for visibility (and space savings :P)
                 if (length($_) >= $len && (/\n[^ \t]/s || !/\n/s)) {
                         local $Text::Wrap::columns = $len;
                         local $Text::Wrap::huge = 'overflow';
-                        $pfx . wrap('', "\t", $_) . $self->{crlf};
+                        $pfx . wrap('', "\t", $v) . $self->{crlf};
                 } else {
-                        $pfx . $_ . $self->{crlf};
+                        $pfx . $v . $self->{crlf};
                 }
         } @vals;
         $$hdr =~ s!$re!shift(@vals) // ''!ge; # replace current headers, first
@@ -378,7 +397,9 @@ sub header_str_set {
         header_set($self, $name, @vals);
 }
 
-sub mhdr_decode ($) { eval { $MIME_Header->decode($_[0]) } // $_[0] }
+sub mhdr_decode ($) {
+        eval { $MIME_Header->decode($_[0], Encode::FB_DEFAULT) } // $_[0];
+}
 
 sub filename {
         my $dis = header_raw($_[0], 'Content-Disposition');
@@ -447,15 +468,19 @@ sub body {
 sub body_str {
         my ($self) = @_;
         my $ct = ct($self);
-        my $charset = $ct->{attributes}->{charset};
-        if (!$charset) {
-                if ($STR_TYPE{$ct->{type}} && $STR_SUBTYPE{$ct->{subtype}}) {
+        my $cs = $ct->{attributes}->{charset} // do {
+                ($STR_TYPE{$ct->{type}} && $STR_SUBTYPE{$ct->{subtype}}) and
                         return body($self);
-                }
-                Carp::confess("can't get body as a string for ",
+                croak("can't get body as a string for ",
                         join("\n\t", header_raw($self, 'Content-Type')));
-        }
-        decode($charset, body($self), Encode::FB_CROAK);
+        };
+        my $enc = find_encoding($cs) or croak "unknown encoding `$cs'";
+        my $ret = body($self);
+        local @enc_warn;
+        local $SIG{__WARN__} = $enc_warn;
+        $ret = $enc->decode($ret, Encode::FB_WARN);
+        croak(@enc_warn) if @enc_warn;
+        $ret;
 }
 
 sub as_string {
@@ -475,9 +500,42 @@ sub charset_set {
 
 sub crlf { $_[0]->{crlf} // "\n" }
 
+sub raw_size {
+        my ($self) = @_;
+        my $len = length(${$self->{hdr}});
+        defined($self->{bdy}) and
+                $len += length(${$self->{bdy}}) + length($self->{crlf});
+        $len;
+}
+
+# warnings to ignore when handling spam mailboxes and maybe other places
+sub warn_ignore {
+        my $s = "@_";
+        # Email::Address::XS warnings
+        $s =~ /^Argument contains empty /
+        || $s =~ /^Element at index [0-9]+.*? contains /
+        # PublicInbox::MsgTime
+        || $s =~ /^bogus TZ offset: .+?, ignoring and assuming \+0000/
+        || $s =~ /^bad Date: .+? in /
+        # Encode::Unicode::UTF7
+        || $s =~ /^Bad UTF7 data escape at /
+}
+
+# this expects to be RHS in this assignment: "local $SIG{__WARN__} = ..."
+sub warn_ignore_cb {
+        my $cb = $SIG{__WARN__} // \&CORE::warn;
+        sub { $cb->(@_) unless warn_ignore(@_) }
+}
+
 sub willneed { re_memo($_) for @_ }
 
 willneed(qw(From To Cc Date Subject Content-Type In-Reply-To References
                 Message-ID X-Alt-Message-ID));
 
+# This fixes an old bug from import (pre-a0c07cba0e5d8b6a)
+# mutt also pipes single RFC822 messages with a "From " line,
+# but no Content-Length or "From " escaping.
+# "git format-patch" also generates such files by default.
+sub strip_from { $_[0] =~ s/\A[\r\n]*From [^\n]*\n//s }
+
 1;