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(spawn);
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 ($$$) {
42 my ($eml, $type, $smsg) = @_;
43 $eml->header_set($_) for (qw(Lines Bytes Content-Length));
45 my %hdr = (Status => []); # set Status, X-Status
46 for my $k (@{$smsg->{kw} // []}) {
47 if (my $ent = $kw2status{$k}) {
48 push @{$hdr{$ent->[0]}}, $ent->[1];
50 warn "TODO: keyword `$k' not supported for mbox\n";
53 # Messages are always 'O' (non-\Recent in IMAP), it saves
54 # MUAs the trouble of rewriting the mbox if no other
55 # changes are made. We put 'O' at the end (e.g. "Status: RO")
56 # to match mutt(1) output.
57 $eml->header_set('Status', join('', sort(@{$hdr{Status}})). 'O');
58 if (my $chars = delete $hdr{'X-Status'}) {
59 $eml->header_set('X-Status', join('', sort(@$chars)));
61 my $buf = delete $eml->{hdr};
63 # fixup old bug from import (pre-a0c07cba0e5d8b6a)
64 $$buf =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
65 my $ident = $smsg->{blob} // 'lei';
66 if (defined(my $pct = $smsg->{pct})) { $ident .= "=$pct" }
68 substr($$buf, 0, 0, # prepend From line
69 "From $ident\@$type Thu Jan 1 00:00:00 1970$eml->{crlf}");
73 sub atomic_append { # for on-disk destinations (O_APPEND, or O_EXCL)
75 if (defined(my $w = syswrite($lei->{1} // return, $$buf))) {
76 return if $w == length($$buf);
77 $buf = "short atomic write: $w != ".length($$buf);
78 } elsif ($! == EPIPE) {
79 return $lei->note_sigpipe(1);
81 $buf = "atomic write: $!";
86 sub eml2mboxrd ($;$) {
87 my ($eml, $smsg) = @_;
88 my $buf = _mbox_hdr_buf($eml, 'mboxrd', $smsg);
89 if (my $bdy = delete $eml->{bdy}) {
90 $$bdy =~ s/^(>*From )/>$1/gm;
91 $$buf .= $eml->{crlf};
92 substr($$bdy, 0, 0, $$buf); # prepend header
95 $$buf .= $eml->{crlf};
100 my ($eml, $smsg) = @_;
101 my $buf = _mbox_hdr_buf($eml, 'mboxo', $smsg);
102 if (my $bdy = delete $eml->{bdy}) {
103 $$bdy =~ s/^From />From /gm;
104 $$buf .= $eml->{crlf};
105 substr($$bdy, 0, 0, $$buf); # prepend header
108 $$buf .= $eml->{crlf};
112 sub _mboxcl_common ($$$) {
113 my ($buf, $bdy, $crlf) = @_;
114 # add Lines: so mutt won't have to add it on MUA close
115 my $lines = $$bdy =~ tr!\n!\n!;
116 $$buf .= 'Content-Length: '.length($$bdy).$crlf.
117 'Lines: '.$lines.$crlf.$crlf;
118 substr($$bdy, 0, 0, $$buf); # prepend header
122 # mboxcl still escapes "From " lines
124 my ($eml, $smsg) = @_;
125 my $buf = _mbox_hdr_buf($eml, 'mboxcl', $smsg);
126 my $crlf = $eml->{crlf};
127 if (my $bdy = delete $eml->{bdy}) {
128 $$bdy =~ s/^From />From /gm;
129 _mboxcl_common($buf, $bdy, $crlf);
135 # mboxcl2 has no "From " escaping
137 my ($eml, $smsg) = @_;
138 my $buf = _mbox_hdr_buf($eml, 'mboxcl2', $smsg);
139 my $crlf = $eml->{crlf};
140 if (my $bdy = delete $eml->{bdy}) {
141 _mboxcl_common($buf, $bdy, $crlf);
147 sub git_to_mail { # git->cat_async callback
148 my ($bref, $oid, $type, $size, $arg) = @_;
149 if ($type ne 'blob') {
150 if ($type eq 'missing') {
151 warn "missing $oid\n";
153 warn "unexpected type=$type for $oid\n";
156 my ($write_cb, $smsg) = @$arg;
157 if ($smsg->{blob} ne $oid) {
158 die "BUG: expected=$smsg->{blob} got=$oid";
160 $write_cb->($bref, $smsg) if $size > 0;
163 sub reap_compress { # dwaitpid callback
164 my ($lei, $pid) = @_;
165 my $cmd = delete $lei->{"pid.$pid"};
167 $lei->fail("@$cmd failed", $? >> 8);
170 sub _post_augment_mbox { # open a compressor process
171 my ($self, $lei) = @_;
172 my $zsfx = $self->{zsfx} or return;
173 my $cmd = PublicInbox::MboxReader::zsfx2cmd($zsfx, undef, $lei);
174 my ($r, $w) = @{delete $lei->{zpipe}};
175 my $rdr = { 0 => $r, 1 => $lei->{1}, 2 => $lei->{2} };
176 my $pid = spawn($cmd, undef, $rdr);
178 my $dup = bless { "pid.$pid" => $cmd }, ref($lei);
179 $dup->{$_} = $lei->{$_} for qw(2 sock);
180 tie *$pp, 'PublicInbox::ProcessPipe', $pid, $w, \&reap_compress, $dup;
186 open my $dup, '+>>&', $in or die "dup: $!";
190 # --augment existing output destination, with deduplication
191 sub _augment { # MboxReader eml_cb
192 my ($eml, $lei) = @_;
193 # ignore return value, just populate the skv
194 $lei->{dedupe}->is_dup($eml);
197 sub _mbox_augment_kw_maybe {
198 my ($eml, $lei, $lse, $augment) = @_;
199 my $kw = PublicInbox::MboxReader::mbox_keywords($eml);
200 update_kw_maybe($lei, $lse, $eml, $kw);
201 _augment($eml, $lei) if $augment;
204 sub _mbox_write_cb ($$) {
205 my ($self, $lei) = @_;
206 my $ovv = $lei->{ovv};
207 my $m = 'eml2'.$ovv->{fmt};
208 my $eml2mbox = $self->can($m) or die "$self->$m missing";
209 $lei->{1} // die "no stdout ($m, $ovv->{dst})"; # redirected earlier
210 $lei->{1}->autoflush(1);
211 my $atomic_append = !defined($ovv->{lock_path});
212 my $dedupe = $lei->{dedupe};
213 $dedupe->prepare_dedupe;
214 my $lse = $lei->{lse}; # may be undef
215 sub { # for git_to_mail
216 my ($buf, $smsg, $eml) = @_;
217 $eml //= PublicInbox::Eml->new($buf);
218 return if $dedupe->is_dup($eml, $smsg->{blob});
219 $lse->xsmsg_vmd($smsg) if $lse;
220 $buf = $eml2mbox->($eml, $smsg);
221 return atomic_append($lei, $buf) if $atomic_append;
222 my $lk = $ovv->lock_for_scope;
227 sub update_kw_maybe ($$$$) {
228 my ($lei, $lse, $eml, $kw) = @_;
230 my $c = $lse->kw_changed($eml, $kw, my $docids = []);
231 my $vmd = { kw => $kw };
232 if (scalar @$docids) { # already in lei/store
233 $lei->{sto}->ipc_do('set_eml_vmd', undef, $vmd, $docids) if $c;
234 } elsif (my $xoids = $lei->{ale}->xoids_for($eml)) {
235 # it's in an external, only set kw, here
236 $lei->{sto}->ipc_do('set_xvmd', $xoids, $eml, $vmd);
237 } else { # never-before-seen, import the whole thing
238 # XXX this is critical in protecting against accidental
239 # data loss without --augment
240 $lei->{sto}->ipc_do('set_eml', $eml, $vmd);
244 sub _augment_or_unlink { # maildir_each_eml cb
245 my ($f, $kw, $eml, $lei, $lse, $mod, $shard, $unlink) = @_;
247 # can't get dirent.d_ino w/ pure Perl readdir, so we extract
248 # the OID if it looks like one instead of doing stat(2)
249 my $hex = $f =~ m!\b([a-f0-9]{40,})[^/]*\z! ?
251 my $recno = hex(substr($hex, 0, 8));
252 return if ($recno % $mod) != $shard;
253 update_kw_maybe($lei, $lse, $eml, $kw);
255 $unlink ? unlink($f) : _augment($eml, $lei);
258 # maildir_each_file callback, \&CORE::unlink doesn't work with it
259 sub _unlink { unlink($_[0]) }
263 sprintf('%x,%x,%x,%x', rand(0xffffffff), time, $$, ++$seq);
267 my ($dst, $buf, $smsg) = @_;
268 my $kw = $smsg->{kw} // [];
269 my $sfx = join('', sort(map { $kw2char{$_} // () } @$kw));
270 my $rand = ''; # chosen by die roll :P
271 my ($tmp, $fh, $final, $ok);
272 my $common = $smsg->{blob} // _rand;
273 if (defined(my $pct = $smsg->{pct})) { $common .= "=$pct" }
275 $tmp = $dst.'tmp/'.$rand.$common;
276 } while (!($ok = sysopen($fh, $tmp, O_CREAT|O_EXCL|O_WRONLY)) &&
277 $! == EEXIST && ($rand = _rand.','));
278 if ($ok && print $fh $$buf and close($fh)) {
279 # ignore new/ and write only to cur/, otherwise MUAs
280 # with R/W access to the Maildir will end up doing
281 # a mass rename which can take a while with thousands
286 $final = $dst.$rand.$common.':2,'.$sfx;
287 } while (!($ok = link($tmp, $final)) && $! == EEXIST &&
288 ($rand = _rand.','));
289 die "link($tmp, $final): $!" unless $ok;
290 unlink($tmp) or warn "W: failed to unlink $tmp: $!\n";
292 my $err = "Error writing $smsg->{blob} to $dst: $!\n";
293 $_[0] = undef; # clobber dst
299 sub _maildir_write_cb ($$) {
300 my ($self, $lei) = @_;
301 my $dedupe = $lei->{dedupe};
302 $dedupe->prepare_dedupe if $dedupe;
303 my $dst = $lei->{ovv}->{dst};
304 my $lse = $lei->{lse}; # may be undef
305 sub { # for git_to_mail
306 my ($buf, $smsg, $eml) = @_;
307 $dst // return $lei->fail; # dst may be undef-ed in last run
308 $buf //= \($eml->as_string);
309 $lse->xsmsg_vmd($smsg) if $lse;
310 return _buf2maildir($dst, $buf, $smsg) if !$dedupe;
311 $eml //= PublicInbox::Eml->new($$buf); # copy buf
312 return if $dedupe->is_dup($eml, $smsg->{blob});
314 _buf2maildir($dst, $buf, $smsg);
318 sub _imap_write_cb ($$) {
319 my ($self, $lei) = @_;
320 my $dedupe = $lei->{dedupe};
321 $dedupe->prepare_dedupe if $dedupe;
322 my $imap_append = $lei->{net}->can('imap_append');
323 my $mic = $lei->{net}->mic_get($self->{uri});
324 my $folder = $self->{uri}->mailbox;
325 my $lse = $lei->{lse}; # may be undef
326 sub { # for git_to_mail
327 my ($bref, $smsg, $eml) = @_;
328 $mic // return $lei->fail; # dst may be undef-ed in last run
330 $eml //= PublicInbox::Eml->new($$bref); # copy bref
331 return if $dedupe->is_dup($eml, $smsg->{blob});
333 $lse->xsmsg_vmd($smsg) if $lse;
334 eval { $imap_append->($mic, $folder, $bref, $smsg, $eml) };
342 sub write_cb { # returns a callback for git_to_mail
343 my ($self, $lei) = @_;
344 # _mbox_write_cb, _maildir_write_cb or _imap_write_cb
345 my $m = "_$self->{base_type}_write_cb";
350 my ($cls, $lei) = @_;
351 my $fmt = $lei->{ovv}->{fmt};
352 my $dst = $lei->{ovv}->{dst};
353 my $self = bless {}, $cls;
354 if ($fmt eq 'maildir') {
355 require PublicInbox::MdirReader;
356 $self->{base_type} = 'maildir';
357 -e $dst && !-d _ and die
358 "$dst exists and is not a directory\n";
359 $lei->{ovv}->{dst} = $dst .= '/' if substr($dst, -1) ne '/';
360 } elsif (substr($fmt, 0, 4) eq 'mbox') {
361 require PublicInbox::MboxReader;
362 (-d $dst || (-e _ && !-w _)) and die
363 "$dst exists and is not a writable file\n";
364 $self->can("eml2$fmt") or die "bad mbox format: $fmt\n";
365 $self->{base_type} = 'mbox';
366 } elsif ($fmt =~ /\Aimaps?\z/) { # TODO .onion support
367 require PublicInbox::NetWriter;
368 my $net = PublicInbox::NetWriter->new;
370 $net->{quiet} = $lei->{opt}->{quiet};
371 my $err = $net->errors($dst);
372 return $lei->fail($err) if $err;
373 require PublicInbox::URIimap; # TODO: URI cast early
374 $self->{uri} = PublicInbox::URIimap->new($dst);
375 $self->{uri}->mailbox or die "No mailbox: $dst";
377 $self->{base_type} = 'imap';
379 die "bad mail --format=$fmt\n";
382 $lei->{dedupe} = PublicInbox::LeiDedupe->new($lei);
386 sub _pre_augment_maildir {
387 my ($self, $lei) = @_;
388 my $dst = $lei->{ovv}->{dst};
389 for my $x (qw(tmp new cur)) {
393 File::Path::mkpath($d);
394 -d $d or die "$d is not a directory";
398 sub _do_augment_maildir {
399 my ($self, $lei) = @_;
400 my $dst = $lei->{ovv}->{dst};
401 my $lse = $lei->{opt}->{'import-before'} ? $lei->{lse} : undef;
402 my ($mod, $shard) = @{$self->{shard_info} // []};
403 if ($lei->{opt}->{augment}) {
404 my $dedupe = $lei->{dedupe};
405 if ($dedupe && $dedupe->prepare_dedupe) {
406 PublicInbox::MdirReader::maildir_each_eml($dst,
407 \&_augment_or_unlink,
408 $lei, $lse, $mod, $shard);
409 $dedupe->pause_dedupe;
412 PublicInbox::MdirReader::maildir_each_eml($dst,
413 \&_augment_or_unlink,
414 $lei, $lse, $mod, $shard, 1);
415 } else {# clobber existing Maildir
416 PublicInbox::MdirReader::maildir_each_file($dst, \&_unlink);
420 sub _imap_augment_or_delete { # PublicInbox::NetReader::imap_each cb
421 my ($url, $uid, $kw, $eml, $lei, $lse, $delete_mic) = @_;
422 update_kw_maybe($lei, $lse, $eml, $kw);
424 $lei->{net}->imap_delete_1($url, $uid, $delete_mic);
426 _augment($eml, $lei);
430 sub _do_augment_imap {
431 my ($self, $lei) = @_;
432 my $net = $lei->{net};
433 my $lse = $lei->{opt}->{'import-before'} ? $lei->{lse} : undef;
434 if ($lei->{opt}->{augment}) {
435 my $dedupe = $lei->{dedupe};
436 if ($dedupe && $dedupe->prepare_dedupe) {
437 $net->imap_each($self->{uri}, \&_imap_augment_or_delete,
439 $dedupe->pause_dedupe;
443 $net->imap_each($self->{uri}, \&_imap_augment_or_delete,
444 $lei, $lse, \$delete_mic);
445 $delete_mic->expunge if $delete_mic;
446 } elsif (!$self->{-wq_worker_nr}) { # undef or 0
447 # clobber existing IMAP folder
448 $net->imap_delete_all($self->{uri});
452 sub _pre_augment_mbox {
453 my ($self, $lei) = @_;
454 my $dst = $lei->{ovv}->{dst};
456 my $devfd = $lei->path_to_fd($dst) // die "bad $dst";
458 $out = $lei->{$devfd};
459 } else { # normal-looking path
461 open $out, '>', $dst or die "open($dst): $!";
462 } elsif (-f _ || !-e _) {
463 require PublicInbox::MboxLock;
464 my $m = $lei->{opt}->{'lock'} //
465 PublicInbox::MboxLock->defaults;
466 $self->{mbl} = PublicInbox::MboxLock->acq($dst, 1, $m);
467 $out = $self->{mbl}->{fh};
469 die "$dst is not a file or FIFO\n";
471 $lei->{old_1} = $lei->{1}; # keep for spawning MUA
473 # Perl does SEEK_END even with O_APPEND :<
474 $self->{seekable} = seek($out, 0, SEEK_SET);
475 if (!$self->{seekable} && $! != ESPIPE && !defined($devfd)) {
476 die "seek($dst): $!\n";
478 if (!$self->{seekable}) {
479 my $imp_before = $lei->{opt}->{'import-before'};
480 die "--import-before specified but $dst is not seekable\n"
481 if $imp_before && !ref($imp_before);
482 die "--augment specified but $dst is not seekable\n" if
483 $lei->{opt}->{augment};
485 if ($self->{zsfx} = PublicInbox::MboxReader::zsfx($dst)) {
486 pipe(my ($r, $w)) or die "pipe: $!";
487 $lei->{zpipe} = [ $r, $w ];
488 $lei->{ovv}->{lock_path} and
489 die 'BUG: unexpected {ovv}->{lock_path}';
490 $lei->{ovv}->ovv_out_lk_init;
491 } elsif (!$self->{seekable} && !$lei->{ovv}->{lock_path}) {
492 $lei->{ovv}->ovv_out_lk_init;
498 sub _do_augment_mbox {
499 my ($self, $lei) = @_;
500 return unless $self->{seekable};
501 my $opt = $lei->{opt};
503 my ($fmt, $dst) = @{$lei->{ovv}}{qw(fmt dst)};
504 return unless -s $out;
505 unless ($opt->{augment} || $opt->{'import-before'}) {
506 truncate($out, 0) or die "truncate($dst): $!";
509 my $zsfx = $self->{zsfx};
510 my $rd = $zsfx ? PublicInbox::MboxReader::zsfxcat($out, $zsfx, $lei)
513 if ($opt->{augment}) {
514 $dedupe = $lei->{dedupe};
515 $dedupe->prepare_dedupe if $dedupe;
517 if ($opt->{'import-before'}) { # the default
518 my $lse = $lei->{lse};
519 PublicInbox::MboxReader->$fmt($rd, \&_mbox_augment_kw_maybe,
520 $lei, $lse, $opt->{augment});
521 if (!$opt->{augment} and !truncate($out, 0)) {
522 die "truncate($dst): $!";
524 } else { # --augment --no-import-before
525 PublicInbox::MboxReader->$fmt($rd, \&_augment, $lei);
527 # maybe some systems don't honor O_APPEND, Perl does this:
528 seek($out, 0, SEEK_END) or die "seek $dst: $!";
529 $dedupe->pause_dedupe if $dedupe;
532 sub pre_augment { # fast (1 disk seek), runs in same process as post_augment
533 my ($self, $lei) = @_;
534 # _pre_augment_maildir, _pre_augment_mbox
535 my $m = $self->can("_pre_augment_$self->{base_type}") or return;
539 sub do_augment { # slow, runs in wq worker
540 my ($self, $lei) = @_;
541 # _do_augment_maildir, _do_augment_mbox, or _do_augment_imap
542 my $m = "_do_augment_$self->{base_type}";
546 # fast (spawn compressor or mkdir), runs in same process as pre_augment
548 my ($self, $lei, @args) = @_;
549 my $wait = $lei->{opt}->{'import-before'} ?
550 $lei->{sto}->ipc_do('checkpoint', 1) : 0;
552 my $m = $self->can("_post_augment_$self->{base_type}") or return;
553 $m->($self, $lei, @args);
558 my $lei = $self->{lei};
559 # lei_xsearch can start as soon as all l2m workers get here
560 pkt_do($lei->{pkt_op_p}, 'incr_start_query') or
561 die "incr_start_query: $!";
563 if (lock_free($self)) {
564 my $mod = $self->{-wq_nr_workers};
565 my $shard = $self->{-wq_worker_nr};
566 if (my $net = $lei->{net}) {
567 $net->{shard_info} = [ $mod, $shard ];
569 $self->{shard_info} = [ $mod, $shard ];
571 $aug = '+'; # incr_post_augment
572 } elsif ($self->{-wq_worker_nr} == 0) {
573 $aug = '.'; # do_post_augment
576 local $0 = 'do_augment';
577 eval { do_augment($self, $lei) };
578 $lei->fail($@) if $@;
579 pkt_do($lei->{pkt_op_p}, $aug) == 1 or
580 die "do_post_augment trigger: $!";
582 if (my $zpipe = delete $lei->{zpipe}) {
583 $lei->{1} = $zpipe->[1];
586 $self->{wcb} = $self->write_cb($lei);
589 sub ipc_atfork_child {
591 my $lei = $self->{lei};
592 $lei->_lei_atfork_child;
593 $lei->{auth}->do_auth_atfork($self) if $lei->{auth};
594 $SIG{__WARN__} = PublicInbox::Eml::warn_ignore_cb();
595 $self->SUPER::ipc_atfork_child;
599 $_[0]->{base_type} =~ /\A(?:maildir|imap|jmap)\z/ ? 1 : 0;
604 if ($self->{base_type} eq 'maildir') {
606 utime($t, $t, $self->{dst} . 'cur');
610 sub write_mail { # via ->wq_io_do
611 my ($self, $smsg) = @_;
612 git_async_cat($self->{lei}->{ale}->git, $smsg->{blob}, \&git_to_mail,
613 [$self->{wcb}, $smsg]);
616 sub wq_atexit_child {
619 $self->{lei}->{ale}->git->async_wait_all;
620 $SIG{__WARN__} = 'DEFAULT';
623 # called in top-level lei-daemon when LeiAuth is done
624 sub net_merge_complete {
626 $self->wq_broadcast('do_post_auth');
630 no warnings 'once'; # the following works even when LeiAuth is lazy-loaded
631 *net_merge_all = \&PublicInbox::LeiAuth::net_merge_all;