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::ProcessPipe;
11 use PublicInbox::Spawn qw(spawn);
12 use Symbol qw(gensym);
13 use IO::Handle; # ->autoflush
14 use Fcntl qw(SEEK_SET SEEK_END O_CREAT O_EXCL O_WRONLY);
16 my %kw2char = ( # Maildir characters
19 forwarded => 'P', # passed
25 flagged => [ 'X-Status' => 'F' ],
26 answered => [ 'X-Status' => 'A' ],
27 seen => [ 'Status' => 'R' ],
28 draft => [ 'X-Status' => 'T' ],
31 sub _mbox_hdr_buf ($$$) {
32 my ($eml, $type, $smsg) = @_;
33 $eml->header_set($_) for (qw(Lines Bytes Content-Length));
35 my %hdr = (Status => []); # set Status, X-Status
36 for my $k (@{$smsg->{kw} // []}) {
37 if (my $ent = $kw2status{$k}) {
38 push @{$hdr{$ent->[0]}}, $ent->[1];
40 warn "# keyword `$k' not supported for mbox\n";
43 # When writing to empty mboxes, messages are always 'O'
44 # (not-\Recent in IMAP), it saves MUAs the trouble of
45 # rewriting the mbox if no other changes are made.
46 # We put 'O' at the end (e.g. "Status: RO") to match mutt(1) output.
47 # We only set smsg->{-recent} if augmenting existing stores.
48 my $status = join('', sort(@{$hdr{Status}}));
49 $status .= 'O' unless $smsg->{-recent};
50 $eml->header_set('Status', $status) if $status;
51 if (my $chars = delete $hdr{'X-Status'}) {
52 $eml->header_set('X-Status', join('', sort(@$chars)));
54 my $buf = delete $eml->{hdr};
56 # fixup old bug from import (pre-a0c07cba0e5d8b6a)
57 $$buf =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
58 my $ident = $smsg->{blob} // 'lei';
59 if (defined(my $pct = $smsg->{pct})) { $ident .= "=$pct" }
61 substr($$buf, 0, 0, # prepend From line
62 "From $ident\@$type Thu Jan 1 00:00:00 1970$eml->{crlf}");
66 sub atomic_append { # for on-disk destinations (O_APPEND, or O_EXCL)
68 if (defined(my $w = syswrite($lei->{1} // return, $$buf))) {
69 return if $w == length($$buf);
70 $buf = "short atomic write: $w != ".length($$buf);
72 return $lei->note_sigpipe(1);
74 $buf = "atomic write: $!";
79 sub eml2mboxrd ($;$) {
80 my ($eml, $smsg) = @_;
81 my $buf = _mbox_hdr_buf($eml, 'mboxrd', $smsg);
82 if (my $bdy = delete $eml->{bdy}) {
83 $$bdy =~ s/^(>*From )/>$1/gm;
84 $$buf .= $eml->{crlf};
85 substr($$bdy, 0, 0, $$buf); # prepend header
88 $$buf .= $eml->{crlf};
93 my ($eml, $smsg) = @_;
94 my $buf = _mbox_hdr_buf($eml, 'mboxo', $smsg);
95 if (my $bdy = delete $eml->{bdy}) {
96 $$bdy =~ s/^From />From /gm;
97 $$buf .= $eml->{crlf};
98 substr($$bdy, 0, 0, $$buf); # prepend header
101 $$buf .= $eml->{crlf};
105 sub _mboxcl_common ($$$) {
106 my ($buf, $bdy, $crlf) = @_;
107 # add Lines: so mutt won't have to add it on MUA close
108 my $lines = $$bdy =~ tr!\n!\n!;
109 $$buf .= 'Content-Length: '.length($$bdy).$crlf.
110 'Lines: '.$lines.$crlf.$crlf;
111 substr($$bdy, 0, 0, $$buf); # prepend header
116 # mboxcl still escapes "From " lines
118 my ($eml, $smsg) = @_;
119 my $buf = _mbox_hdr_buf($eml, 'mboxcl', $smsg);
120 my $bdy = delete($eml->{bdy}) // \(my $empty = '');
121 $$bdy =~ s/^From />From /gm;
122 _mboxcl_common($buf, $bdy, $eml->{crlf});
125 # mboxcl2 has no "From " escaping
127 my ($eml, $smsg) = @_;
128 my $buf = _mbox_hdr_buf($eml, 'mboxcl2', $smsg);
129 my $bdy = delete($eml->{bdy}) // \(my $empty = '');
130 _mboxcl_common($buf, $bdy, $eml->{crlf});
133 sub git_to_mail { # git->cat_async callback
134 my ($bref, $oid, $type, $size, $arg) = @_;
135 my ($write_cb, $smsg) = @$arg;
136 if ($type eq 'missing' && $smsg->{-lms_ro}) {
137 if ($bref = $smsg->{-lms_ro}->local_blob($oid, 1)) {
139 $size = length($$bref);
142 return warn("W: $oid is $type (!= blob)\n") if $type ne 'blob';
143 return warn("E: $oid is empty\n") unless $size;
144 die "BUG: expected=$smsg->{blob} got=$oid" if $smsg->{blob} ne $oid;
145 $write_cb->($bref, $smsg);
148 sub reap_compress { # dwaitpid callback
149 my ($lei, $pid) = @_;
150 my $cmd = delete $lei->{"pid.$pid"};
152 $lei->fail("@$cmd failed", $? >> 8);
155 sub _post_augment_mbox { # open a compressor process from top-level process
156 my ($self, $lei) = @_;
157 my $zsfx = $self->{zsfx} or return;
158 my $cmd = PublicInbox::MboxReader::zsfx2cmd($zsfx, undef, $lei);
159 my ($r, $w) = @{delete $lei->{zpipe}};
160 my $rdr = { 0 => $r, 1 => $lei->{1}, 2 => $lei->{2} };
161 my $pid = spawn($cmd, undef, $rdr);
163 my $dup = bless { "pid.$pid" => $cmd }, ref($lei);
164 $dup->{$_} = $lei->{$_} for qw(2 sock);
165 tie *$pp, 'PublicInbox::ProcessPipe', $pid, $w, \&reap_compress, $dup;
169 # --augment existing output destination, with deduplication
170 sub _augment { # MboxReader eml_cb
171 my ($eml, $lei) = @_;
172 # ignore return value, just populate the skv
173 $lei->{dedupe}->is_dup($eml);
176 sub _mbox_augment_kw_maybe {
177 my ($eml, $lei, $lse, $augment) = @_;
178 my $kw = PublicInbox::MboxReader::mbox_keywords($eml);
179 update_kw_maybe($lei, $lse, $eml, $kw);
180 _augment($eml, $lei) if $augment;
183 sub _mbox_write_cb ($$) {
184 my ($self, $lei) = @_;
185 my $ovv = $lei->{ovv};
186 my $m = 'eml2'.$ovv->{fmt};
187 my $eml2mbox = $self->can($m) or die "$self->$m missing";
188 $lei->{1} // die "no stdout ($m, $ovv->{dst})"; # redirected earlier
189 $lei->{1}->autoflush(1);
190 my $atomic_append = !defined($ovv->{lock_path});
191 my $dedupe = $lei->{dedupe};
192 $dedupe->prepare_dedupe;
193 my $lse = $lei->{lse}; # may be undef
194 my $set_recent = $dedupe->has_entries;
195 sub { # for git_to_mail
196 my ($buf, $smsg, $eml) = @_;
197 $eml //= PublicInbox::Eml->new($buf);
198 return if $dedupe->is_dup($eml, $smsg);
199 $lse->xsmsg_vmd($smsg) if $lse;
200 $smsg->{-recent} = 1 if $set_recent;
201 $buf = $eml2mbox->($eml, $smsg);
202 if ($atomic_append) {
203 atomic_append($lei, $buf);
205 my $lk = $ovv->lock_for_scope;
212 sub update_kw_maybe ($$$$) {
213 my ($lei, $lse, $eml, $kw) = @_;
215 my $c = $lse->kw_changed($eml, $kw, my $docids = []);
216 my $vmd = { kw => $kw };
217 if (scalar @$docids) { # already in lei/store
218 $lei->{sto}->wq_do('set_eml_vmd', undef, $vmd, $docids) if $c;
219 } elsif (my $xoids = $lei->{ale}->xoids_for($eml)) {
220 # it's in an external, only set kw, here
221 $lei->{sto}->wq_do('set_xvmd', $xoids, $eml, $vmd);
222 } else { # never-before-seen, import the whole thing
223 # XXX this is critical in protecting against accidental
224 # data loss without --augment
225 $lei->{sto}->wq_do('set_eml', $eml, $vmd);
229 sub _md_update { # maildir_each_eml cb
230 my ($f, $kw, $eml, $lei, $lse, $unlink) = @_;
231 update_kw_maybe($lei, $lse, $eml, $kw);
232 $unlink ? unlink($f) : _augment($eml, $lei);
235 # maildir_each_file callback, \&CORE::unlink doesn't work with it
236 sub _unlink { unlink($_[0]) }
240 sprintf('%x,%x,%x,%x', rand(0xffffffff), time, $$, ++$seq);
243 sub kw2suffix ($;@) {
245 join('', sort(map { $kw2char{$_} // () } @$kw, @_));
248 sub _buf2maildir ($$$$) {
249 my ($dst, $buf, $smsg, $dir) = @_;
250 my $kw = $smsg->{kw} // [];
251 my $rand = ''; # chosen by die roll :P
252 my ($tmp, $fh, $base, $ok);
253 my $common = $smsg->{blob} // _rand;
254 if (defined(my $pct = $smsg->{pct})) { $common .= "=$pct" }
256 $tmp = $dst.'tmp/'.$rand.$common;
257 } while (!($ok = sysopen($fh, $tmp, O_CREAT|O_EXCL|O_WRONLY)) &&
258 $!{EEXIST} && ($rand = _rand.','));
259 if ($ok && print $fh $$buf and close($fh)) {
260 $dst .= $dir; # 'new/' or 'cur/'
263 $base = $rand.$common.':2,'.kw2suffix($kw);
264 } while (!($ok = link($tmp, $dst.$base)) && $!{EEXIST} &&
265 ($rand = _rand.','));
266 die "link($tmp, $dst$base): $!" unless $ok;
267 unlink($tmp) or warn "W: failed to unlink $tmp: $!\n";
270 my $err = "Error writing $smsg->{blob} to $dst: $!\n";
271 $_[0] = undef; # clobber dst
277 sub _maildir_write_cb ($$) {
278 my ($self, $lei) = @_;
279 my $dedupe = $lei->{dedupe};
280 $dedupe->prepare_dedupe if $dedupe;
281 my $dst = $lei->{ovv}->{dst};
282 my $lse = $lei->{lse}; # may be undef
283 my $sto = $lei->{opt}->{'mail-sync'} ? $lei->{sto} : undef;
284 my $out = $sto ? 'maildir:'.$lei->abs_path($dst) : undef;
286 # Favor cur/ and only write to new/ when augmenting. This
287 # saves MUAs from having to do a mass rename when the initial
288 # search result set is huge.
289 my $dir = $dedupe && $dedupe->has_entries ? 'new/' : 'cur/';
290 sub { # for git_to_mail
291 my ($bref, $smsg, $eml) = @_;
292 $dst // return $lei->fail; # dst may be undef-ed in last run
293 return if $dedupe && $dedupe->is_dup($eml //
294 PublicInbox::Eml->new($$bref),
296 $lse->xsmsg_vmd($smsg) if $lse;
297 my $n = _buf2maildir($dst, $bref // \($eml->as_string),
299 $sto->wq_do('set_sync_info', $smsg->{blob}, $out, $n) if $sto;
304 sub _imap_write_cb ($$) {
305 my ($self, $lei) = @_;
306 my $dedupe = $lei->{dedupe};
307 $dedupe->prepare_dedupe if $dedupe;
308 my $append = $lei->{net}->can('imap_append');
309 my $uri = $self->{uri};
310 my $mic = $lei->{net}->mic_get($uri);
311 my $folder = $uri->mailbox;
312 $uri->uidvalidity($mic->uidvalidity($folder));
313 my $lse = $lei->{lse}; # may be undef
314 my $sto = $lei->{opt}->{'mail-sync'} ? $lei->{sto} : undef;
315 sub { # for git_to_mail
316 my ($bref, $smsg, $eml) = @_;
317 $mic // return $lei->fail; # mic may be undef-ed in last run
318 return if $dedupe && $dedupe->is_dup($eml //
319 PublicInbox::Eml->new($$bref),
321 $lse->xsmsg_vmd($smsg) if $lse;
322 my $uid = eval { $append->($mic, $folder, $bref, $smsg, $eml) };
327 # imap_append returns UID if IMAP server has UIDPLUS extension
328 ($sto && $uid =~ /\A[0-9]+\z/) and
329 $sto->wq_do('set_sync_info',
330 $smsg->{blob}, $$uri, $uid + 0);
335 sub _text_write_cb ($$) {
336 my ($self, $lei) = @_;
337 my $dedupe = $lei->{dedupe};
338 $dedupe->prepare_dedupe if $dedupe;
339 my $lvt = $lei->{lvt};
340 my $ovv = $lei->{ovv};
341 $lei->{1} // die "no stdout ($ovv->{dst})"; # redirected earlier
342 $lei->{1}->autoflush(1);
343 binmode $lei->{1}, ':utf8';
344 my $lse = $lei->{lse}; # may be undef
345 sub { # for git_to_mail
346 my ($bref, $smsg, $eml) = @_;
347 $lse->xsmsg_vmd($smsg) if $lse;
348 $eml //= PublicInbox::Eml->new($bref);
349 return if $dedupe && $dedupe->is_dup($eml, $smsg);
350 my $lk = $ovv->lock_for_scope;
351 $lei->out(${$lvt->eml_to_text($smsg, $eml)}, "\n");
355 sub _v2_write_cb ($$) {
356 my ($self, $lei) = @_;
357 my $dedupe = $lei->{dedupe};
358 $dedupe->prepare_dedupe if $dedupe;
359 sub { # for git_to_mail
360 my ($bref, $smsg, $eml) = @_;
361 $eml //= PublicInbox::Eml->new($bref);
362 return if $dedupe && $dedupe->is_dup($eml, $smsg);
363 $lei->{v2w}->wq_do('add', $eml); # V2Writable->add
368 sub write_cb { # returns a callback for git_to_mail
369 my ($self, $lei) = @_;
370 # _mbox_write_cb, _maildir_write_cb, _imap_write_cb, _v2_write_cb
371 my $m = "_$self->{base_type}_write_cb";
376 my ($cls, $lei) = @_;
377 my $fmt = $lei->{ovv}->{fmt};
378 my $dst = $lei->{ovv}->{dst};
379 my $self = bless {}, $cls;
381 if ($fmt eq 'maildir') {
382 require PublicInbox::MdirReader;
383 $self->{base_type} = 'maildir';
384 -e $dst && !-d _ and die
385 "$dst exists and is not a directory\n";
386 $lei->{ovv}->{dst} = $dst .= '/' if substr($dst, -1) ne '/';
387 $lei->{opt}->{save} //= \1 if $lei->{cmd} eq 'q';
388 } elsif (substr($fmt, 0, 4) eq 'mbox') {
389 require PublicInbox::MboxReader;
390 $self->can("eml2$fmt") or die "bad mbox format: $fmt\n";
391 $self->{base_type} = 'mbox';
392 if ($lei->{cmd} eq 'q' &&
393 (($lei->path_to_fd($dst) // -1) < 0) &&
394 (-f $dst || !-e _)) {
395 $lei->{opt}->{save} //= \1;
397 } elsif ($fmt =~ /\Aimaps?\z/) {
398 require PublicInbox::NetWriter;
399 require PublicInbox::URIimap;
400 # {net} may exist from "lei up" for auth
401 my $net = $lei->{net} // PublicInbox::NetWriter->new;
402 $net->{quiet} = $lei->{opt}->{quiet};
403 my $uri = PublicInbox::URIimap->new($dst)->canonical;
404 $net->add_url($$uri);
405 my $err = $net->errors($lei);
406 return $lei->fail($err) if $err;
407 $uri->mailbox or return $lei->fail("No mailbox: $dst");
409 $dst = $lei->{ovv}->{dst} = $$uri; # canonicalized
411 $self->{base_type} = 'imap';
412 $lei->{opt}->{save} //= \1 if $lei->{cmd} eq 'q';
413 } elsif ($fmt eq 'text' || $fmt eq 'reply') {
414 require PublicInbox::LeiViewText;
415 $lei->{lvt} = PublicInbox::LeiViewText->new($lei, $fmt);
416 $self->{base_type} = 'text';
417 @conflict = qw(mua save);
418 } elsif ($fmt eq 'v2') {
419 die "--dedupe=oid and v2 are incompatible\n" if
420 ($lei->{opt}->{dedupe}//'') eq 'oid';
421 $self->{base_type} = 'v2';
422 $lei->{opt}->{save} = \1;
423 $dst = $lei->{ovv}->{dst} = $lei->abs_path($dst);
424 @conflict = qw(mua sort);
426 die "bad mail --format=$fmt\n";
428 if ($self->{base_type} =~ /\A(?:text|mbox)\z/) {
429 (-d $dst || (-e _ && !-w _)) and die
430 "$dst exists and is not a writable file\n";
432 my @err = map { defined($lei->{opt}->{$_}) ? "--$_" : () } @conflict;
433 die "@err incompatible with $fmt\n" if @err;
435 $lei->{dedupe} = $lei->{lss} // do {
436 my $dd_cls = 'PublicInbox::'.
437 ($lei->{opt}->{save} ? 'LeiSavedSearch' : 'LeiDedupe');
438 eval "require $dd_cls";
439 die "$dd_cls: $@" if $@;
440 my $dd = $dd_cls->new($lei);
441 $lei->{lss} //= $dd if $dd && $dd->can('cfg_set');
447 sub _pre_augment_maildir {
448 my ($self, $lei) = @_;
449 my $dst = $lei->{ovv}->{dst};
450 for my $x (qw(tmp new cur)) {
454 File::Path::mkpath($d);
455 -d $d or die "$d is not a directory";
457 # for utime, so no opendir
458 open $self->{poke_dh}, '<', "${dst}cur" or die "open ${dst}cur: $!";
461 sub clobber_dst_prepare ($;$) {
463 if (my $lms = defined($f) ? $lei->lms : undef) {
464 $lms->lms_write_prepare;
465 $lms->forget_folders($f);
467 my $dedupe = $lei->{dedupe} or return;
468 $dedupe->reset_dedupe if $dedupe->can('reset_dedupe');
471 sub _do_augment_maildir {
472 my ($self, $lei) = @_;
473 return if $lei->{cmd} eq 'up';
474 my $dst = $lei->{ovv}->{dst};
475 my $lse = $lei->{opt}->{'import-before'} ? $lei->{lse} : undef;
476 my $mdr = PublicInbox::MdirReader->new;
477 if ($lei->{opt}->{augment}) {
478 my $dedupe = $lei->{dedupe};
479 if ($dedupe && $dedupe->prepare_dedupe) {
480 $mdr->{shard_info} = $self->{shard_info};
481 $mdr->maildir_each_eml($dst, \&_md_update, $lei, $lse);
482 $dedupe->pause_dedupe;
485 clobber_dst_prepare($lei, "maildir:$dst");
486 $mdr->{shard_info} = $self->{shard_info};
487 $mdr->maildir_each_eml($dst, \&_md_update, $lei, $lse, 1);
488 } else {# clobber existing Maildir
489 clobber_dst_prepare($lei, "maildir:$dst");
490 $mdr->maildir_each_file($dst, \&_unlink);
494 sub _imap_augment_or_delete { # PublicInbox::NetReader::imap_each cb
495 my ($uri, $uid, $kw, $eml, $lei, $lse, $delete_mic) = @_;
496 update_kw_maybe($lei, $lse, $eml, $kw);
498 $lei->{net}->imap_delete_1($uri, $uid, $delete_mic);
500 _augment($eml, $lei);
504 sub _do_augment_imap {
505 my ($self, $lei) = @_;
506 return if $lei->{cmd} eq 'up';
507 my $net = $lei->{net};
508 my $lse = $lei->{opt}->{'import-before'} ? $lei->{lse} : undef;
509 if ($lei->{opt}->{augment}) {
510 my $dedupe = $lei->{dedupe};
511 if ($dedupe && $dedupe->prepare_dedupe) {
512 $net->imap_each($self->{uri}, \&_imap_augment_or_delete,
514 $dedupe->pause_dedupe;
518 clobber_dst_prepare($lei, "$self->{uri}");
519 $net->imap_each($self->{uri}, \&_imap_augment_or_delete,
520 $lei, $lse, \$delete_mic);
521 $delete_mic->expunge if $delete_mic;
522 } elsif (!$self->{-wq_worker_nr}) { # undef or 0
523 # clobber existing IMAP folder
524 clobber_dst_prepare($lei, "$self->{uri}");
525 $net->imap_delete_all($self->{uri});
529 sub _pre_augment_text {
530 my ($self, $lei) = @_;
531 my $dst = $lei->{ovv}->{dst};
533 my $devfd = $lei->path_to_fd($dst) // die "bad $dst";
535 $out = $lei->{$devfd};
536 } else { # normal-looking path
538 open $out, '>', $dst or die "open($dst): $!";
539 } elsif (-f _ || !-e _) {
540 # text allows augment, HTML/Atom won't
541 my $mode = $lei->{opt}->{augment} ? '>>' : '>';
542 open $out, $mode, $dst or die "open($mode, $dst): $!";
544 die "$dst is not a file or FIFO\n";
547 $lei->{ovv}->ovv_out_lk_init if !$lei->{ovv}->{lock_path};
552 sub _pre_augment_mbox {
553 my ($self, $lei) = @_;
554 my $dst = $lei->{ovv}->{dst};
556 my $devfd = $lei->path_to_fd($dst) // die "bad $dst";
558 $out = $lei->{$devfd};
559 } else { # normal-looking path
561 open $out, '>', $dst or die "open($dst): $!";
562 } elsif (-f _ || !-e _) {
563 require PublicInbox::MboxLock;
564 my $m = $lei->{opt}->{'lock'} //
565 PublicInbox::MboxLock->defaults;
566 $self->{mbl} = PublicInbox::MboxLock->acq($dst, 1, $m);
567 $out = $self->{mbl}->{fh};
569 die "$dst is not a file or FIFO\n";
571 $lei->{old_1} = $lei->{1}; # keep for spawning MUA
573 # Perl does SEEK_END even with O_APPEND :<
574 $self->{seekable} = seek($out, 0, SEEK_SET);
575 if (!$self->{seekable} && !$!{ESPIPE} && !defined($devfd)) {
576 die "seek($dst): $!\n";
578 if (!$self->{seekable}) {
579 my $imp_before = $lei->{opt}->{'import-before'};
580 die "--import-before specified but $dst is not seekable\n"
581 if $imp_before && !ref($imp_before);
582 die "--augment specified but $dst is not seekable\n" if
583 $lei->{opt}->{augment};
584 die "cannot --save with unseekable $dst\n" if
585 $lei->{dedupe} && $lei->{dedupe}->can('reset_dedupe');
587 if ($self->{zsfx} = PublicInbox::MboxReader::zsfx($dst)) {
588 pipe(my ($r, $w)) or die "pipe: $!";
589 $lei->{zpipe} = [ $r, $w ];
590 $lei->{ovv}->{lock_path} and
591 die 'BUG: unexpected {ovv}->{lock_path}';
592 $lei->{ovv}->ovv_out_lk_init;
593 } elsif (!$self->{seekable} && !$lei->{ovv}->{lock_path}) {
594 $lei->{ovv}->ovv_out_lk_init;
600 sub _do_augment_mbox {
601 my ($self, $lei) = @_;
602 return unless $self->{seekable};
603 my $opt = $lei->{opt};
604 return if $lei->{cmd} eq 'up';
606 my ($fmt, $dst) = @{$lei->{ovv}}{qw(fmt dst)};
607 return clobber_dst_prepare($lei) unless -s $out;
608 unless ($opt->{augment} || $opt->{'import-before'}) {
609 truncate($out, 0) or die "truncate($dst): $!";
613 if (my $zsfx = $self->{zsfx}) {
614 $rd = PublicInbox::MboxReader::zsfxcat($out, $zsfx, $lei);
616 open($rd, '+>>&', $out) or die "dup: $!";
619 if ($opt->{augment}) {
620 $dedupe = $lei->{dedupe};
621 $dedupe->prepare_dedupe if $dedupe;
623 clobber_dst_prepare($lei);
625 if ($opt->{'import-before'}) { # the default
626 my $lse = $lei->{lse};
627 PublicInbox::MboxReader->$fmt($rd, \&_mbox_augment_kw_maybe,
628 $lei, $lse, $opt->{augment});
629 if (!$opt->{augment} and !truncate($out, 0)) {
630 die "truncate($dst): $!";
632 } else { # --augment --no-import-before
633 PublicInbox::MboxReader->$fmt($rd, \&_augment, $lei);
635 # maybe some systems don't honor O_APPEND, Perl does this:
636 seek($out, 0, SEEK_END) or die "seek $dst: $!";
637 $dedupe->pause_dedupe if $dedupe;
640 sub _pre_augment_v2 {
641 my ($self, $lei) = @_;
642 my $dir = $self->{dst};
643 require PublicInbox::InboxWritable;
646 my $opt = { -min_inbox_version => 2 };
647 require PublicInbox::Admin;
648 my @ibx = PublicInbox::Admin::resolve_inboxes([ $dir ], $opt);
649 $ibx = $ibx[0] or die "$dir is not a v2 inbox\n";
652 $ibx = PublicInbox::Inbox->new({
653 name => 'lei-result', # XXX configurable
656 address => [ 'lei@example.com' ],
659 PublicInbox::InboxWritable->new($ibx, @creat);
660 $ibx->init_inbox if @creat;
661 my $v2w = $ibx->importer;
662 $v2w->{-wq_no_bcast} = 1;
663 $v2w->wq_workers_start("lei/v2w $dir", 1, $lei->oldset, {lei => $lei});
665 return if !$lei->{opt}->{shared};
666 my $d = "$lei->{ale}->{git}->{git_dir}/objects";
667 my $al = "$dir/git/0.git/objects/info/alternates";
668 open my $fh, '+>>', $al or die "open($al): $!";
669 seek($fh, 0, SEEK_SET) or die "seek($al): $!";
670 grep(/\A\Q$d\E\n/, <$fh>) and return;
671 print $fh "$d\n" or die "print($al): $!";
672 close $fh or die "close($al): $!";
675 sub pre_augment { # fast (1 disk seek), runs in same process as post_augment
676 my ($self, $lei) = @_;
677 # _pre_augment_maildir, _pre_augment_mbox, _pre_augment_v2
678 my $m = $self->can("_pre_augment_$self->{base_type}") or return;
682 sub do_augment { # slow, runs in wq worker
683 my ($self, $lei) = @_;
684 # _do_augment_maildir, _do_augment_mbox, or _do_augment_imap
685 my $m = $self->can("_do_augment_$self->{base_type}") or return;
689 # fast (spawn compressor or mkdir), runs in same process as pre_augment
691 my ($self, $lei, @args) = @_;
692 $self->{-au_noted}++ and $lei->qerr("# writing to $self->{dst} ...");
694 my $wait = $lei->{opt}->{'import-before'} ?
695 $lei->{sto}->wq_do('checkpoint', 1) : 0;
697 my $m = $self->can("_post_augment_$self->{base_type}") or return;
698 $m->($self, $lei, @args);
701 # called by every single l2m worker process
704 my $lei = $self->{lei};
705 # lei_xsearch can start as soon as all l2m workers get here
706 $lei->{pkt_op_p}->pkt_do('incr_start_query') or
707 die "incr_start_query: $!";
709 if (lock_free($self)) { # all workers do_augment
710 my $mod = $self->{-wq_nr_workers};
711 my $shard = $self->{-wq_worker_nr};
712 if (my $net = $lei->{net}) {
713 $net->{shard_info} = [ $mod, $shard ];
715 $self->{shard_info} = [ $mod, $shard ];
717 $aug = 'incr_post_augment';
718 } elsif ($self->{-wq_worker_nr} == 0) { # 1st worker do_augment
719 $aug = 'do_post_augment';
722 local $0 = 'do_augment';
723 eval { do_augment($self, $lei) };
724 $lei->fail($@) if $@;
725 $lei->{pkt_op_p}->pkt_do($aug) or die "pkt_do($aug): $!";
727 # done augmenting, connect the compressor pipe for each worker
728 if (my $zpipe = delete $lei->{zpipe}) {
729 $lei->{1} = $zpipe->[1];
732 my $au_peers = delete $self->{au_peers};
733 if ($au_peers) { # wait for peer l2m to finish augmenting:
734 $au_peers->[1] = undef;
735 sysread($au_peers->[0], my $barrier1, 1);
737 $self->{wcb} = $self->write_cb($lei);
738 if ($au_peers) { # wait for peer l2m to set write_cb
739 $au_peers->[3] = undef;
740 sysread($au_peers->[2], my $barrier2, 1);
744 sub ipc_atfork_child {
746 my $lei = $self->{lei};
747 $lei->_lei_atfork_child;
748 if (my $lse = $lei->{lse}) {
749 $self->{-lms_ro} = $lse->{-lms_ro} //= $lse->lms;
751 $lei->{auth}->do_auth_atfork($self) if $lei->{auth};
752 $SIG{__WARN__} = PublicInbox::Eml::warn_ignore_cb();
753 $self->SUPER::ipc_atfork_child;
757 $_[0]->{base_type} =~ /\A(?:maildir|imap|jmap)\z/ ? 1 : 0;
760 # wakes up the MUA when complete so it can refresh messages list
763 if ($self->{base_type} eq 'maildir') {
765 utime($t, $t, $self->{poke_dh}) or warn "futimes: $!";
769 sub write_mail { # via ->wq_io_do
770 my ($self, $smsg, $eml) = @_;
771 return $self->{wcb}->(undef, $smsg, $eml) if $eml;
772 $smsg->{-lms_ro} = $self->{-lms_ro};
773 $self->{lei}->{ale}->git->cat_async($smsg->{blob}, \&git_to_mail,
774 [$self->{wcb}, $smsg]);
777 sub wq_atexit_child {
779 local $PublicInbox::DS::in_loop = 0; # waitpid synchronously
780 my $lei = $self->{lei};
782 $lei->{ale}->git->async_wait_all;
783 my $nr = delete($lei->{-nr_write}) or return;
784 return if $lei->{early_mua} || !$lei->{-progress} || !$lei->{pkt_op_p};
785 $lei->{pkt_op_p}->pkt_do('l2m_progress', $nr);
788 # runs on a 1s timer in lei-daemon
789 sub augment_inprogress {
790 my ($err, $opt, $dst, $au_noted) = @_;
792 return if $$au_noted++ || !$err || !defined(fileno($err));
793 print $err '# '.($opt->{'import-before'} ?
794 "importing non-external contents of $dst" : (
795 ($opt->{dedupe} // 'content') ne 'none') ?
796 "scanning old contents of $dst for dedupe" :
797 "removing old contents of $dst")." ...\n";
802 # called in top-level lei-daemon when LeiAuth is done
803 sub net_merge_all_done {
804 my ($self, $lei) = @_;
805 if ($PublicInbox::DS::in_loop &&
806 $self->can("_do_augment_$self->{base_type}") &&
807 !$lei->{opt}->{quiet}) {
808 $self->{-au_noted} = 0;
809 PublicInbox::DS::add_timer(1, \&augment_inprogress,
810 $lei->{2}, $lei->{opt},
811 $self->{dst}, \$self->{-au_noted});
813 $self->wq_broadcast('do_post_auth');