1 # Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Writes PublicInbox::Eml objects atomically to a mbox variant or Maildir
5 package PublicInbox::LeiToMail;
8 use parent qw(PublicInbox::IPC);
10 use PublicInbox::Lock;
11 use PublicInbox::ProcessPipe;
12 use PublicInbox::Spawn qw(which spawn popen_rd);
13 use PublicInbox::LeiDedupe;
15 use PublicInbox::GitAsyncCat;
16 use PublicInbox::PktOp qw(pkt_do);
17 use Symbol qw(gensym);
18 use IO::Handle; # ->autoflush
19 use Fcntl qw(SEEK_SET SEEK_END O_CREAT O_EXCL O_WRONLY);
20 use Errno qw(EEXIST ESPIPE ENOENT EPIPE);
21 use Digest::SHA qw(sha256_hex);
23 # struggles with short-lived repos, Gcf2Client makes little sense with lei;
24 # but we may use in-process libgit2 in the future.
25 $PublicInbox::GitAsyncCat::GCF2C = 0;
27 my %kw2char = ( # Maildir characters
35 flagged => [ 'X-Status' => 'F' ],
36 answered => [ 'X-Status' => 'A' ],
37 seen => [ 'Status' => 'R' ],
38 draft => [ 'X-Status' => 'T' ],
41 sub _mbox_hdr_buf ($$$) {
42 my ($eml, $type, $smsg) = @_;
43 $eml->header_set($_) for (qw(Lines Bytes Content-Length));
45 # Messages are always 'O' (non-\Recent in IMAP), it saves
46 # MUAs the trouble of rewriting the mbox if no other
48 my %hdr = (Status => [ 'O' ]); # set Status, X-Status
49 for my $k (@{$smsg->{kw} // []}) {
50 if (my $ent = $kw2status{$k}) {
51 push @{$hdr{$ent->[0]}}, $ent->[1];
53 warn "TODO: keyword `$k' not supported for mbox\n";
56 while (my ($name, $chars) = each %hdr) {
57 $eml->header_set($name, join('', sort @$chars));
59 my $buf = delete $eml->{hdr};
61 # fixup old bug from import (pre-a0c07cba0e5d8b6a)
62 $$buf =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
63 my $ident = $smsg->{blob} // 'lei';
64 if (defined(my $pct = $smsg->{pct})) { $ident .= "=$pct" }
66 substr($$buf, 0, 0, # prepend From line
67 "From $ident\@$type Thu Jan 1 00:00:00 1970$eml->{crlf}");
71 sub atomic_append { # for on-disk destinations (O_APPEND, or O_EXCL)
73 if (defined(my $w = syswrite($lei->{1} // return, $$buf))) {
74 return if $w == length($$buf);
75 $buf = "short atomic write: $w != ".length($$buf);
76 } elsif ($! == EPIPE) {
77 return $lei->note_sigpipe(1);
79 $buf = "atomic write: $!";
84 sub eml2mboxrd ($;$) {
85 my ($eml, $smsg) = @_;
86 my $buf = _mbox_hdr_buf($eml, 'mboxrd', $smsg);
87 if (my $bdy = delete $eml->{bdy}) {
88 $$bdy =~ s/^(>*From )/>$1/gm;
89 $$buf .= $eml->{crlf};
90 substr($$bdy, 0, 0, $$buf); # prepend header
93 $$buf .= $eml->{crlf};
98 my ($eml, $smsg) = @_;
99 my $buf = _mbox_hdr_buf($eml, 'mboxo', $smsg);
100 if (my $bdy = delete $eml->{bdy}) {
101 $$bdy =~ s/^From />From /gm;
102 $$buf .= $eml->{crlf};
103 substr($$bdy, 0, 0, $$buf); # prepend header
106 $$buf .= $eml->{crlf};
110 sub _mboxcl_common ($$$) {
111 my ($buf, $bdy, $crlf) = @_;
112 # add Lines: so mutt won't have to add it on MUA close
113 my $lines = $$bdy =~ tr!\n!\n!;
114 $$buf .= 'Content-Length: '.length($$bdy).$crlf.
115 'Lines: '.$lines.$crlf.$crlf;
116 substr($$bdy, 0, 0, $$buf); # prepend header
120 # mboxcl still escapes "From " lines
122 my ($eml, $smsg) = @_;
123 my $buf = _mbox_hdr_buf($eml, 'mboxcl', $smsg);
124 my $crlf = $eml->{crlf};
125 if (my $bdy = delete $eml->{bdy}) {
126 $$bdy =~ s/^From />From /gm;
127 _mboxcl_common($buf, $bdy, $crlf);
133 # mboxcl2 has no "From " escaping
135 my ($eml, $smsg) = @_;
136 my $buf = _mbox_hdr_buf($eml, 'mboxcl2', $smsg);
137 my $crlf = $eml->{crlf};
138 if (my $bdy = delete $eml->{bdy}) {
139 _mboxcl_common($buf, $bdy, $crlf);
145 sub git_to_mail { # git->cat_async callback
146 my ($bref, $oid, $type, $size, $arg) = @_;
147 if ($type ne 'blob') {
148 if ($type eq 'missing') {
149 warn "missing $oid\n";
151 warn "unexpected type=$type for $oid\n";
154 my ($write_cb, $smsg) = @$arg;
155 if ($smsg->{blob} ne $oid) {
156 die "BUG: expected=$smsg->{blob} got=$oid";
158 $write_cb->($bref, $smsg) if $size > 0;
161 sub reap_compress { # dwaitpid callback
162 my ($lei, $pid) = @_;
163 my $cmd = delete $lei->{"pid.$pid"};
165 $lei->fail("@$cmd failed", $? >> 8);
168 # all of these support -c for stdout and -d for decompression,
169 # mutt is commonly distributed with hooks for gz, bz2 and xz, at least
170 # { foo => '' } means "--foo" is passed to the command-line,
171 # otherwise { foo => '--bar' } passes "--bar"
173 gz => [ qw(GZIP pigz gzip), { rsyncable => '' } ],
174 bz2 => [ 'bzip2', {} ],
176 # XXX does anybody care for these? I prefer zstd on entire FSes,
177 # so it's probably not necessary on a per-file basis
178 # zst => [ 'zstd', { -default => [ qw(-q) ], # it's noisy by default
179 # rsyncable => '', threads => '-T' } ],
180 # zz => [ 'pigz', { -default => [ '--zlib' ],
181 # rsyncable => '', threads => '-p' }],
182 # lzo => [ 'lzop', {} ],
183 # lzma => [ 'lzma', {} ],
187 my ($zsfx, $decompress, $lei) = @_;
188 my $x = $zsfx2cmd{$zsfx} // die "no support for suffix=.$zsfx";
190 my $cmd_opt = pop @info;
191 my @cmd = (undef, $decompress ? qw(-dc) : qw(-c));
192 for my $exe (@info) {
193 # I think respecting client's ENV{GZIP} is OK, not sure
194 # about ENV overrides for other, less-common compressors
195 if ($exe eq uc($exe)) {
196 $exe = $lei->{env}->{$exe} or next;
198 $cmd[0] = which($exe) and last;
200 $cmd[0] // die join(' or ', @info)." missing for .$zsfx";
201 # push @cmd, @{$cmd_opt->{-default}} if $cmd_opt->{-default};
202 for my $bool (qw(rsyncable)) {
203 my $switch = $cmd_opt->{rsyncable} // next;
204 push @cmd, '--'.($switch || $bool);
206 for my $key (qw(rsyncable)) { # support compression level?
207 my $switch = $cmd_opt->{$key} // next;
208 my $val = $lei->{opt}->{$key} // next;
209 push @cmd, $switch, $val;
214 sub _post_augment_mbox { # open a compressor process
215 my ($self, $lei) = @_;
216 my $zsfx = $self->{zsfx} or return;
217 my $cmd = zsfx2cmd($zsfx, undef, $lei);
218 my ($r, $w) = @{delete $lei->{zpipe}};
219 my $rdr = { 0 => $r, 1 => $lei->{1}, 2 => $lei->{2} };
220 my $pid = spawn($cmd, undef, $rdr);
222 my $dup = bless { "pid.$pid" => $cmd }, ref($lei);
223 $dup->{$_} = $lei->{$_} for qw(2 sock);
224 tie *$pp, 'PublicInbox::ProcessPipe', $pid, $w, \&reap_compress, $dup;
228 sub decompress_src ($$$) {
229 my ($in, $zsfx, $lei) = @_;
230 my $cmd = zsfx2cmd($zsfx, 1, $lei);
231 popen_rd($cmd, undef, { 0 => $in, 2 => $lei->{2} });
236 open my $dup, '+>>&', $in or die "dup: $!";
240 # --augment existing output destination, with deduplication
241 sub _augment { # MboxReader eml_cb
242 my ($eml, $lei) = @_;
243 # ignore return value, just populate the skv
244 $lei->{dedupe}->is_dup($eml);
247 sub _mbox_augment_kw_maybe {
248 my ($eml, $lei, $lse, $augment) = @_;
249 my $kw = PublicInbox::MboxReader::mbox_keywords($eml);
250 update_kw_maybe($lei, $lse, $eml, $kw);
251 _augment($eml, $lei) if $augment;
254 sub _mbox_write_cb ($$) {
255 my ($self, $lei) = @_;
256 my $ovv = $lei->{ovv};
257 my $m = 'eml2'.$ovv->{fmt};
258 my $eml2mbox = $self->can($m) or die "$self->$m missing";
259 $lei->{1} // die "no stdout ($m, $ovv->{dst})"; # redirected earlier
260 $lei->{1}->autoflush(1);
261 my $atomic_append = !defined($ovv->{lock_path});
262 my $dedupe = $lei->{dedupe};
263 $dedupe->prepare_dedupe;
264 sub { # for git_to_mail
265 my ($buf, $smsg, $eml) = @_;
266 $eml //= PublicInbox::Eml->new($buf);
267 return if $dedupe->is_dup($eml, $smsg->{blob});
268 $buf = $eml2mbox->($eml, $smsg);
269 return atomic_append($lei, $buf) if $atomic_append;
270 my $lk = $ovv->lock_for_scope;
275 sub update_kw_maybe ($$$$) {
276 my ($lei, $lse, $eml, $kw) = @_;
278 my $x = $lse->kw_changed($eml, $kw);
280 $lei->{sto}->ipc_do('set_eml', $eml, @$kw);
281 } elsif (!defined($x)) {
282 $lei->{sto}->ipc_do('set_xkw', $eml, $kw);
286 sub _augment_or_unlink { # maildir_each_eml cb
287 my ($f, $kw, $eml, $lei, $lse, $mod, $shard, $unlink) = @_;
289 # can't get dirent.d_ino w/ pure Perl, so we extract the OID
290 # if it looks like one:
291 my $hex = $f =~ m!\b([a-f0-9]{40,})[^/]*\z! ?
293 my $recno = hex(substr($hex, 0, 8));
294 return if ($recno % $mod) != $shard;
295 update_kw_maybe($lei, $lse, $eml, $kw);
297 $unlink ? unlink($f) : _augment($eml, $lei);
300 # maildir_each_file callback, \&CORE::unlink doesn't work with it
301 sub _unlink { unlink($_[0]) }
305 sprintf('%x,%x,%x,%x', rand(0xffffffff), time, $$, ++$seq);
309 my ($dst, $buf, $smsg) = @_;
310 my $kw = $smsg->{kw} // [];
311 my $sfx = join('', sort(map { $kw2char{$_} // () } @$kw));
312 my $rand = ''; # chosen by die roll :P
313 my ($tmp, $fh, $final, $ok);
314 my $common = $smsg->{blob} // _rand;
315 if (defined(my $pct = $smsg->{pct})) { $common .= "=$pct" }
317 $tmp = $dst.'tmp/'.$rand.$common;
318 } while (!($ok = sysopen($fh, $tmp, O_CREAT|O_EXCL|O_WRONLY)) &&
319 $! == EEXIST && ($rand = _rand.','));
320 if ($ok && print $fh $$buf and close($fh)) {
321 # ignore new/ and write only to cur/, otherwise MUAs
322 # with R/W access to the Maildir will end up doing
323 # a mass rename which can take a while with thousands
328 $final = $dst.$rand.$common.':2,'.$sfx;
329 } while (!($ok = link($tmp, $final)) && $! == EEXIST &&
330 ($rand = _rand.','));
331 die "link($tmp, $final): $!" unless $ok;
332 unlink($tmp) or warn "W: failed to unlink $tmp: $!\n";
334 my $err = "Error writing $smsg->{blob} to $dst: $!\n";
335 $_[0] = undef; # clobber dst
341 sub _maildir_write_cb ($$) {
342 my ($self, $lei) = @_;
343 my $dedupe = $lei->{dedupe};
344 $dedupe->prepare_dedupe if $dedupe;
345 my $dst = $lei->{ovv}->{dst};
346 sub { # for git_to_mail
347 my ($buf, $smsg, $eml) = @_;
348 $dst // return $lei->fail; # dst may be undef-ed in last run
349 $buf //= \($eml->as_string);
350 return _buf2maildir($dst, $buf, $smsg) if !$dedupe;
351 $eml //= PublicInbox::Eml->new($$buf); # copy buf
352 return if $dedupe->is_dup($eml, $smsg->{blob});
354 _buf2maildir($dst, $buf, $smsg);
358 sub _imap_write_cb ($$) {
359 my ($self, $lei) = @_;
360 my $dedupe = $lei->{dedupe};
361 $dedupe->prepare_dedupe if $dedupe;
362 my $imap_append = $lei->{net}->can('imap_append');
363 my $mic = $lei->{net}->mic_get($self->{uri});
364 my $folder = $self->{uri}->mailbox;
365 sub { # for git_to_mail
366 my ($bref, $smsg, $eml) = @_;
367 $mic // return $lei->fail; # dst may be undef-ed in last run
369 $eml //= PublicInbox::Eml->new($$bref); # copy bref
370 return if $dedupe->is_dup($eml, $smsg->{blob});
372 eval { $imap_append->($mic, $folder, $bref, $smsg, $eml) };
380 sub write_cb { # returns a callback for git_to_mail
381 my ($self, $lei) = @_;
382 # _mbox_write_cb, _maildir_write_cb or _imap_write_cb
383 my $m = "_$self->{base_type}_write_cb";
388 my ($cls, $lei) = @_;
389 my $fmt = $lei->{ovv}->{fmt};
390 my $dst = $lei->{ovv}->{dst};
391 my $self = bless {}, $cls;
392 if ($fmt eq 'maildir') {
393 require PublicInbox::MdirReader;
394 $self->{base_type} = 'maildir';
395 -e $dst && !-d _ and die
396 "$dst exists and is not a directory\n";
397 $lei->{ovv}->{dst} = $dst .= '/' if substr($dst, -1) ne '/';
398 } elsif (substr($fmt, 0, 4) eq 'mbox') {
399 require PublicInbox::MboxReader;
400 (-d $dst || (-e _ && !-w _)) and die
401 "$dst exists and is not a writable file\n";
402 $self->can("eml2$fmt") or die "bad mbox format: $fmt\n";
403 $self->{base_type} = 'mbox';
404 } elsif ($fmt =~ /\Aimaps?\z/) { # TODO .onion support
405 require PublicInbox::NetWriter;
406 my $net = PublicInbox::NetWriter->new;
408 $net->{quiet} = $lei->{opt}->{quiet};
409 my $err = $net->errors($dst);
410 return $lei->fail($err) if $err;
411 require PublicInbox::URIimap; # TODO: URI cast early
412 $self->{uri} = PublicInbox::URIimap->new($dst);
413 $self->{uri}->mailbox or die "No mailbox: $dst";
415 $self->{base_type} = 'imap';
417 die "bad mail --format=$fmt\n";
420 $lei->{dedupe} = PublicInbox::LeiDedupe->new($lei);
424 sub _pre_augment_maildir {
425 my ($self, $lei) = @_;
426 my $dst = $lei->{ovv}->{dst};
427 for my $x (qw(tmp new cur)) {
431 File::Path::mkpath($d);
432 -d $d or die "$d is not a directory";
436 sub _do_augment_maildir {
437 my ($self, $lei) = @_;
438 my $dst = $lei->{ovv}->{dst};
439 my $lse = $lei->{sto}->search if $lei->{opt}->{'import-before'};
440 my ($mod, $shard) = @{$self->{shard_info} // []};
441 if ($lei->{opt}->{augment}) {
442 my $dedupe = $lei->{dedupe};
443 if ($dedupe && $dedupe->prepare_dedupe) {
444 PublicInbox::MdirReader::maildir_each_eml($dst,
445 \&_augment_or_unlink,
446 $lei, $lse, $mod, $shard);
447 $dedupe->pause_dedupe;
450 PublicInbox::MdirReader::maildir_each_eml($dst,
451 \&_augment_or_unlink,
452 $lei, $lse, $mod, $shard, 1);
453 } else {# clobber existing Maildir
454 PublicInbox::MdirReader::maildir_each_file($dst, \&_unlink);
458 sub _imap_augment_or_delete { # PublicInbox::NetReader::imap_each cb
459 my ($url, $uid, $kw, $eml, $lei, $lse, $delete_mic) = @_;
460 update_kw_maybe($lei, $lse, $eml, $kw);
462 $lei->{net}->imap_delete_1($url, $uid, $delete_mic);
464 _augment($eml, $lei);
468 sub _do_augment_imap {
469 my ($self, $lei) = @_;
470 my $net = $lei->{net};
471 my $lse = $lei->{sto}->search if $lei->{opt}->{'import-before'};
472 if ($lei->{opt}->{augment}) {
473 my $dedupe = $lei->{dedupe};
474 if ($dedupe && $dedupe->prepare_dedupe) {
475 $net->imap_each($self->{uri}, \&_imap_augment_or_delete,
477 $dedupe->pause_dedupe;
481 $net->imap_each($self->{uri}, \&_imap_augment_or_delete,
482 $lei, $lse, \$delete_mic);
483 $delete_mic->expunge if $delete_mic;
484 } elsif (!$self->{-wq_worker_nr}) { # undef or 0
485 # clobber existing IMAP folder
486 $net->imap_delete_all($self->{uri});
490 sub _pre_augment_mbox {
491 my ($self, $lei) = @_;
492 my $dst = $lei->{ovv}->{dst};
494 if ($dst ne '/dev/stdout') {
497 open $out, '>', $dst or die "open($dst): $!";
498 } elsif (-f _ || !-e _) {
499 require PublicInbox::MboxLock;
500 my $m = $lei->{opt}->{'lock'} //
501 PublicInbox::MboxLock->defaults;
502 $self->{mbl} = PublicInbox::MboxLock->acq($dst, 1, $m);
503 $out = $self->{mbl}->{fh};
505 $lei->{old_1} = $lei->{1}; # keep for spawning MUA
507 # Perl does SEEK_END even with O_APPEND :<
508 $self->{seekable} = seek($out, 0, SEEK_SET);
509 if (!$self->{seekable} && $! != ESPIPE && $dst ne '/dev/stdout') {
510 die "seek($dst): $!\n";
512 if (!$self->{seekable}) {
513 my $ia = $lei->{opt}->{'import-before'};
514 die "--import-before specified but $dst is not seekable\n"
516 die "--augment specified but $dst is not seekable\n" if
517 $lei->{opt}->{augment};
519 state $zsfx_allow = join('|', keys %zsfx2cmd);
520 if (($self->{zsfx}) = ($dst =~ /\.($zsfx_allow)\z/)) {
521 pipe(my ($r, $w)) or die "pipe: $!";
522 $lei->{zpipe} = [ $r, $w ];
523 $lei->{ovv}->{lock_path} and
524 die 'BUG: unexpected {ovv}->{lock_path}';
525 $lei->{ovv}->ovv_out_lk_init;
526 } elsif (!$self->{seekable} && !$lei->{ovv}->{lock_path}) {
527 $lei->{ovv}->ovv_out_lk_init;
533 sub _do_augment_mbox {
534 my ($self, $lei) = @_;
535 return unless $self->{seekable};
536 my $opt = $lei->{opt};
538 my ($fmt, $dst) = @{$lei->{ovv}}{qw(fmt dst)};
539 return unless -s $out;
540 unless ($opt->{augment} || $opt->{'import-before'}) {
541 truncate($out, 0) or die "truncate($dst): $!";
544 my $zsfx = $self->{zsfx};
545 my $rd = $zsfx ? decompress_src($out, $zsfx, $lei) : dup_src($out);
547 if ($opt->{augment}) {
548 $dedupe = $lei->{dedupe};
549 $dedupe->prepare_dedupe if $dedupe;
551 if ($opt->{'import-before'}) { # the default
552 my $lse = $lei->{sto}->search;
553 PublicInbox::MboxReader->$fmt($rd, \&_mbox_augment_kw_maybe,
554 $lei, $lse, $opt->{augment});
555 if (!$opt->{augment} and !truncate($out, 0)) {
556 die "truncate($dst): $!";
558 } else { # --augment --no-import-before
559 PublicInbox::MboxReader->$fmt($rd, \&_augment, $lei);
561 # maybe some systems don't honor O_APPEND, Perl does this:
562 seek($out, 0, SEEK_END) or die "seek $dst: $!";
563 $dedupe->pause_dedupe if $dedupe;
566 sub pre_augment { # fast (1 disk seek), runs in same process as post_augment
567 my ($self, $lei) = @_;
568 # _pre_augment_maildir, _pre_augment_mbox
569 my $m = $self->can("_pre_augment_$self->{base_type}") or return;
573 sub do_augment { # slow, runs in wq worker
574 my ($self, $lei) = @_;
575 # _do_augment_maildir, _do_augment_mbox, or _do_augment_imap
576 my $m = "_do_augment_$self->{base_type}";
580 # fast (spawn compressor or mkdir), runs in same process as pre_augment
582 my ($self, $lei, @args) = @_;
583 my $wait = $lei->{opt}->{'import-before'} ?
584 $lei->{sto}->ipc_do('checkpoint', 1) : 0;
586 my $m = $self->can("_post_augment_$self->{base_type}") or return;
587 $m->($self, $lei, @args);
592 my $lei = $self->{lei};
593 # lei_xsearch can start as soon as all l2m workers get here
594 pkt_do($lei->{pkt_op_p}, 'incr_start_query') or
595 die "incr_start_query: $!";
597 if (lock_free($self)) {
598 my $mod = $self->{-wq_nr_workers};
599 my $shard = $self->{-wq_worker_nr};
600 if (my $net = $lei->{net}) {
601 $net->{shard_info} = [ $mod, $shard ];
602 } else { # Maildir (MH?)
603 $self->{shard_info} = [ $mod, $shard ];
605 $aug = '+'; # incr_post_augment
606 } elsif ($self->{-wq_worker_nr} == 0) {
607 $aug = '.'; # do_post_augment
610 local $0 = 'do_augment';
611 eval { do_augment($self, $lei) };
612 $lei->fail($@) if $@;
613 pkt_do($lei->{pkt_op_p}, $aug) == 1 or
614 die "do_post_augment trigger: $!";
616 if (my $zpipe = delete $lei->{zpipe}) {
617 $lei->{1} = $zpipe->[1];
620 $self->{wcb} = $self->write_cb($lei);
623 sub ipc_atfork_child {
625 my $lei = $self->{lei};
626 $lei->lei_atfork_child;
627 $lei->{auth}->do_auth_atfork($self) if $lei->{auth};
628 $SIG{__WARN__} = PublicInbox::Eml::warn_ignore_cb();
629 $self->SUPER::ipc_atfork_child;
633 $_[0]->{base_type} =~ /\A(?:maildir|mh|imap|jmap)\z/ ? 1 : 0;
638 if ($self->{base_type} eq 'maildir') {
640 utime($t, $t, "$self->{dst}/cur");
644 sub write_mail { # via ->wq_io_do
645 my ($self, $git_dir, $smsg) = @_;
646 my $git = $self->{"$$\0$git_dir"} //= PublicInbox::Git->new($git_dir);
647 git_async_cat($git, $smsg->{blob}, \&git_to_mail,
648 [$self->{wcb}, $smsg]);
651 sub wq_atexit_child {
654 for my $git (delete @$self{grep(/\A$$\0/, keys %$self)}) {
655 $git->async_wait_all;
657 $SIG{__WARN__} = 'DEFAULT';
660 # called in top-level lei-daemon when LeiAuth is done
661 sub net_merge_complete {
663 $self->wq_broadcast('do_post_auth');
667 no warnings 'once'; # the following works even when LeiAuth is lazy-loaded
668 *net_merge_all = \&PublicInbox::LeiAuth::net_merge_all;