eml: pure-Perl replacement for Email::MIME

Email::MIME eats memory, wastes time parsing out all the
headers, and some problems can't be fixed without breaking
compatibility for other projects which depend on it.

Informal benchmarks show a ~2x improvement in general
stats gathering scripts and ~10% improvement in HTML
view rendering.

We also don't need the ability to create MIME messages, just
parse them and maybe drop an attachment.

While this isn't the zero-copy or streaming MIME parser of my
dreams; it's still an improvement in that it doesn't keep a
scalar copy of the raw body around along with subparts.  It also
doesn't parse subparts up front, so it can also replace our uses
of Email::Simple.

1 files changed, 363 insertions, 0 deletions

diff --git a/t/eml.t b/t/eml.t
new file mode 100644
index 00000000..43c735e7
--- /dev/null
+++ b/t/eml.t
@@ -0,0 +1,363 @@
+#!perl -w
+# Copyright (C) 2020 all contributors <meta@public-inbox.org>
+# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
+use strict;
+use Test::More;
+use PublicInbox::TestCommon;
+use PublicInbox::MsgIter qw(msg_part_text);
+my @classes = qw(PublicInbox::Eml);
+SKIP: {
+        require_mods('Email::MIME', 1);
+        push @classes, 'PublicInbox::MIME';
+};
+use_ok $_ for @classes;
+
+{
+        my $eml = PublicInbox::Eml->new(\(my $str = "a: b\n\nhi\n"));
+        is($str, "hi\n", '->new modified body like Email::Simple');
+        is($eml->body, "hi\n", '->body works');
+        is($eml->as_string, "a: b\n\nhi\n", '->as_string');
+}
+
+for my $cls (@classes) {
+        my $mime = $cls->new(my $orig = "From: x\n\nb");
+        is($mime->as_string, $orig, '->as_string works');
+        is($mime->header_obj->as_string, "From: x\n",
+                        'header ->as_string works');
+
+        # headers
+        is($mime->header_raw('From'), 'x', 'header_raw scalar context');
+        $mime = $cls->new("R:\n\tx\nR:\n 1\n");
+        is_deeply([$mime->header_raw('r')], [ 'x', '1' ], 'multi-value');
+        $mime = $cls->new("R:x\nR: 1\n");
+        is_deeply([$mime->header_raw('r')], [ 'x', '1' ], 'multi-value header');
+        $mime = $cls->new("R:x\n R: 1\nR:\n f\n");
+        is_deeply([$mime->header_raw('r')], [ 'x R: 1', 'f' ],
+                'multi-line, multi-value header');
+
+        $mime->header_set('r');
+        is_deeply([$mime->header_raw('r')], [], 'header_set clears');
+        $mime->header_set('r');
+        is_deeply([$mime->header_raw('r')], [], 'header_set clears idempotent');
+        $mime->header_set('r', 'h');
+        is_deeply([$mime->header_raw('r')], ['h'], 'header_set');
+        $mime->header_set('r', 'h', 'i');
+        is_deeply([$mime->header_raw('r')], ['h', 'i'], 'header_set ary');
+        $mime->header_set('rr', 'b');
+        is_deeply([$mime->header_raw('r')], ['h', 'i'],
+                                "header_set `rr' did not clobber `r'");
+        is($mime->header_raw('rr'), 'b', 'got set scalar');
+        $mime->header_set('rr', 'b'x100);
+        is($mime->header_raw('rr'), 'b'x100, 'got long set scalar');
+        if ($cls eq 'PublicInbox::Eml') {
+                like($mime->as_string, qr/^rr: b{100}\n(?:\n|\z)/sm,
+                        'single token not wrapped');
+        }
+        $mime->header_set('rr', ('b'x100) . ' wrap me');
+        if ($cls eq 'PublicInbox::Eml') {
+                like($mime->as_string, qr/^rr: b{100}\n\twrap me\n/sm,
+                        'wrapped after long token');
+        }
+        my $exp = "pre\tformatted\n with\n breaks";
+        $mime->header_set('r', $exp);
+        like($mime->as_string, qr/^r: \Q$exp\E/sm, 'preformatted preserved');
+} # for @classes
+
+for my $cls (@classes) { # make sure we don't add quotes if not needed
+        my $eml = $cls->new("From: John Smith <j\@example.com>\n\n");
+        is($eml->header('From'), 'John Smith <j@example.com>',
+                "name not unnecessarily quoted $cls");
+}
+
+for my $cls (@classes) {
+        my $eml = $cls->new("Subject: foo\n\n");
+        $eml->header_str_set('Subject', "\x{100}");
+        like($eml->header_raw('Subject'), qr/utf-8\?B\?/i,
+                'MIME-B encoded UTF-8 Subject');
+        is_deeply([$eml->header_str('Subject')], [ "\x{100}" ],
+                'got wide character back');
+}
+
+# linux-mips apparently got some messages injected w/o Message-ID
+# and long Subject: lines w/o leading whitespace.
+# What appears in the blobs was generated by V2Writable.
+for my $cls (@classes) {
+        my $eml = $cls->new(<<'EOF');
+Message-ID: <20101130193431@z>
+Subject: something really long
+and really wrong
+From: linux-mips archive injection
+Object-Id: 8c56b7abdd551b1264e6522ededbbed9890cccd0
+EOF
+        is_deeply([ $eml->header('Subject') ],
+                [ 'something really long and really wrong' ],
+                'continued long line w/o leading spaces '.$cls);
+        is_deeply([ $eml->header('From') ],
+                [ 'linux-mips archive injection' ],
+                'subsequent line not corrupted');
+        is_deeply([ $eml->header('Message-ID') ],
+                ['<20101130193431@z>'],
+                'preceding line readable');
+} # for @classes
+
+{
+        my $eml = eml_load 't/msg_iter-order.eml';
+        my @parts;
+        my $orig = $eml->as_string;
+        $eml->each_part(sub {
+                my ($part, $level, @ex) = @{$_[0]};
+                my $s = $part->body_str;
+                $s =~ s/\s+//sg;
+                push @parts, [ $s, $level, @ex ];
+        });
+        is_deeply(\@parts, [ [ qw(a 1 1) ], [ qw(b 1 2) ] ], 'order is fine');
+        is($eml->as_string, $orig, 'unchanged by ->each_part');
+        $eml->each_part(sub {}, undef, 1);
+        is(defined($eml) ? $eml->body_raw : '', # old msg_iter clobbers $eml
+                '', 'each_part can clobber body');
+}
+
+# body-less, boundary-less
+for my $cls (@classes) {
+        my $call = 0;
+        $cls->new(<<'EOF')->each_part(sub { $call++ }, 0, 1);
+Content-Type: multipart/mixed; boundary="body-less"
+
+EOF
+        is($call, 1, 'called on bodyless multipart');
+
+        my @tmp;
+        $cls->new(<<'EOF')->each_part(sub { push @tmp, \@_; }, 0, 1);
+Content-Type: multipart/mixed; boundary="boundary-less"
+
+hello world
+EOF
+        is(scalar(@tmp), 1, 'got one part even w/o boundary');
+        is($tmp[0]->[0]->[0]->body, "hello world\n", 'body preserved');
+        is($tmp[0]->[0]->[1], 0, '$depth is zero');
+        is($tmp[0]->[0]->[2], 0, '@idx is zero');
+}
+
+# I guess the following only worked in PI::M because of a happy accident
+# involving inheritance:
+for my $cls (@classes) {
+        my @tmp;
+        my $header_less = <<'EOF';
+Archived-At: <85k5su9k59.fsf_-_@lola.goethe.zz>
+Content-Type: multipart/mixed; boundary="header-less"
+
+--header-less
+
+this is the body
+
+--header-less
+i-haz: header
+
+something else
+
+--header-less--
+EOF
+        my $expect = "this is the body\n";
+        $cls->new($header_less)->each_part(sub { push @tmp, \@_  }, 0, 1);
+        my $body = $tmp[0]->[0]->[0]->body;
+        if ($cls eq 'PublicInbox::Eml') {
+                is($body, $expect, 'body-only subpart in '.$cls);
+        } elsif ($body ne $expect) {
+                diag "W: $cls `$body' != `$expect'";
+        }
+        is($tmp[1]->[0]->[0]->body, "something else\n");
+        is(scalar(@tmp), 2, 'two parts');
+}
+
+if ('one newline before headers') {
+        my $eml = PublicInbox::Eml->new("\nNewline: no Header \n");
+        my @v = $eml->header_raw('Newline');
+        is_deeply(\@v, ['no Header'], 'no header');
+        is($eml->crlf, "\n", 'got CRLF as "\n"');
+        is($eml->body, "");
+}
+
+for my $cls (@classes) { # XXX: matching E::M, but not sure about this
+        my $s = <<EOF;
+Content-Type: multipart/mixed; boundary="b"
+
+--b
+header: only
+--b--
+EOF
+        my $eml = $cls->new(\$s);
+        my $nr = 0;
+        my @v;
+        $eml->each_part(sub {
+                @v = $_[0]->[0]->header_raw('Header');
+                $nr++;
+        });
+        is($nr, 1, 'only one part');
+        is_deeply(\@v, [], "nothing w/o body $cls");
+}
+
+for my $cls (@classes) {
+        my $s = <<EOF; # double epilogue, double the fun
+Content-Type: multipart/mixed; boundary="b"
+
+--b
+should: appear
+
+yes
+
+--b--
+
+--b
+should: not appear
+
+nope
+--b--
+EOF
+        my $eml = $cls->new(\$s);
+        my $nr = 0;
+        $eml->each_part(sub {
+                my $part = $_[0]->[0];
+                is_deeply([$part->header_raw('should')], ['appear'],
+                        'only got one header');
+                is($part->body, "yes\n", 'got expected body');
+                $nr++;
+        });
+        is($nr, 1, 'only one part');
+}
+
+for my $cls (@classes) {
+        my $s = <<EOF; # buggy git-send-email versions, again?
+Content-Type: text/plain; =?ISO-8859-1?Q?=20charset=3D=1BOF?=
+Content-Transfer-Encoding: 8bit
+Object-Id: ab0440d8cd6d843bee9a27709a459ce3b2bdb94d (lore/kvm)
+
+\xc4\x80
+EOF
+        my $eml = $cls->new(\$s);
+        my ($str, $err) = msg_part_text($eml, $eml->content_type);
+        is($str, "\x{100}\n", "got wide character by assuming utf-8");
+}
+
+if ('we differ from Email::MIME with final "\n" on missing epilogue') {
+        my $s = <<EOF;
+Content-Type: multipart/mixed; boundary="b"
+
+--b
+header: but
+
+no epilogue
+EOF
+        my $eml = PublicInbox::Eml->new(\$s);
+        is(($eml->subparts)[-1]->body, "no epilogue\n",
+                'final "\n" preserved on missing epilogue');
+}
+
+if ('maxparts is a feature unique to us') {
+        my $eml = eml_load 't/psgi_attach.eml';
+        my @orig;
+        $eml->each_part(sub { push @orig, $_[0]->[0] });
+
+        local $PublicInbox::Eml::MAXPARTS = scalar(@orig);
+        my $i = 0;
+        $eml->each_part(sub {
+                my $cur = $_[0]->[0];
+                my $prv = $orig[$i++];
+                is($cur->body_raw, $prv->body_raw, "part #$i matches");
+        });
+        is($i, scalar(@orig), 'maxparts honored');
+        $PublicInbox::Eml::MAXPARTS--;
+        my @ltd;
+        $eml->each_part(sub { push @ltd, $_[0]->[0] });
+        for ($i = 0; $i <= $#ltd; $i++) {
+                is($ltd[$i]->body_raw, $orig[$i]->body_raw,
+                        "part[$i] matches");
+        }
+        is(scalar(@ltd), scalar(@orig) - 1, 'maxparts honored');
+}
+
+SKIP: {
+        require_mods('PublicInbox::MIME', 1);
+        my $eml = eml_load 't/utf8.eml';
+        my $mime = mime_load 't/utf8.eml';
+        for my $h (qw(Subject From To)) {
+                my $v = $eml->header($h);
+                my $m = $mime->header($h);
+                is($v, $m, "decoded -8 $h matches Email::MIME");
+                ok(utf8::is_utf8($v), "$h is UTF-8");
+                ok(utf8::valid($v), "UTF-8 valid $h");
+        }
+        my $s = $eml->body_str;
+        ok(utf8::is_utf8($s), 'body_str is UTF-8');
+        ok(utf8::valid($s), 'UTF-8 valid body_str');
+        my $ref = \(my $x = 'ref');
+        for my $msg ($eml, $mime) {
+                $msg->body_str_set($s .= "\nHI\n");
+                ok(!utf8::is_utf8($msg->body_raw),
+                                'raw octets after body_str_set');
+                $s = $msg->body_str;
+                ok(utf8::is_utf8($s), 'body_str is UTF-8 after set');
+                ok(utf8::valid($s), 'UTF-8 valid body_str after set');
+                $msg->body_set($ref);
+                is($msg->body_raw, $$ref, 'body_set worked on scalar ref');
+                $msg->body_set($$ref);
+                is($msg->body_raw, $$ref, 'body_set worked on scalar');
+        }
+        $eml = eml_load 't/iso-2202-jp.eml';
+        $mime = mime_load 't/iso-2202-jp.eml';
+        $s = $eml->body_str;
+        is($s, $mime->body_str, 'ISO-2202-JP body_str');
+        ok(utf8::is_utf8($s), 'ISO-2202-JP => UTF-8 body_str');
+        ok(utf8::valid($s), 'UTF-8 valid body_str');
+
+        $eml = eml_load 't/psgi_attach.eml';
+        $mime = mime_load 't/psgi_attach.eml';
+        is_deeply([ map { $_->body_raw } $eml->subparts ],
+                [ map { $_->body_raw } $mime->subparts ],
+                'raw ->subparts match deeply');
+        is_deeply([ map { $_->body } $eml->subparts ],
+                [ map { $_->body } $mime->subparts ],
+                '->subparts match deeply');
+        for my $msg ($eml, $mime) {
+                my @old = $msg->subparts;
+                $msg->parts_set([]);
+                is_deeply([$msg->subparts], [], 'parts_set can clear');
+                $msg->parts_set([$old[-1]]);
+                is(scalar $msg->subparts, 1, 'only last remains');
+        }
+        is($eml->as_string, $mime->as_string,
+                'as_string matches after parts_set');
+}
+
+for my $cls (@classes) {
+        my $s = <<'EOF';
+Content-Type: text/x-patch; name="=?utf-8?q?vtpm-fakefile.patch?="
+Content-Disposition: attachment; filename="=?utf-8?q?vtpm-makefile.patch?="
+
+EOF
+        is($cls->new($s)->filename, 'vtpm-makefile.patch', 'filename decoded');
+        $s =~ s/^Content-Disposition:.*$//sm;
+        is($cls->new($s)->filename, 'vtpm-fakefile.patch', 'filename fallback');
+        is($cls->new($s)->content_type,
+                'text/x-patch; name="vtpm-fakefile.patch"',
+                'matches Email::MIME output, "correct" or not');
+
+        $s = <<'EOF';
+Content-Type: multipart/foo; boundary=b
+
+--b
+Content-Disposition: attachment; filename="=?utf-8?q?vtpm-makefile.patch?="
+
+a
+--b
+Content-Type: text/x-patch; name="=?utf-8?q?vtpm-fakefile.patch?="
+
+b
+--b--
+EOF
+        my @tmp;
+        $cls->new($s)->each_part(sub { push @tmp, $_[0]->[0]->filename });
+        is_deeply(['vtpm-makefile.patch', 'vtpm-fakefile.patch'], \@tmp,
+                'got filename for both attachments');
+}
+
+done_testing;