1 # Copyright (C) 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);
15 use PublicInbox::Syscall qw(rename_noreplace);
17 my %kw2char = ( # Maildir characters
20 forwarded => 'P', # passed
26 flagged => [ 'X-Status' => 'F' ],
27 answered => [ 'X-Status' => 'A' ],
28 seen => [ 'Status' => 'R' ],
29 draft => [ 'X-Status' => 'T' ],
32 sub _mbox_hdr_buf ($$$) {
33 my ($eml, $type, $smsg) = @_;
34 $eml->header_set($_) for (qw(Lines Bytes Content-Length));
36 my %hdr = (Status => []); # set Status, X-Status
37 for my $k (@{$smsg->{kw} // []}) {
38 if (my $ent = $kw2status{$k}) {
39 push @{$hdr{$ent->[0]}}, $ent->[1];
41 warn "# keyword `$k' not supported for mbox\n";
44 # When writing to empty mboxes, messages are always 'O'
45 # (not-\Recent in IMAP), it saves MUAs the trouble of
46 # rewriting the mbox if no other changes are made.
47 # We put 'O' at the end (e.g. "Status: RO") to match mutt(1) output.
48 # We only set smsg->{-recent} if augmenting existing stores.
49 my $status = join('', sort(@{$hdr{Status}}));
50 $status .= 'O' unless $smsg->{-recent};
51 $eml->header_set('Status', $status) if $status;
52 if (my $chars = delete $hdr{'X-Status'}) {
53 $eml->header_set('X-Status', join('', sort(@$chars)));
55 my $buf = delete $eml->{hdr};
57 # fixup old bug from import (pre-a0c07cba0e5d8b6a)
58 $$buf =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
59 my $ident = $smsg->{blob} // 'lei';
60 if (defined(my $pct = $smsg->{pct})) { $ident .= "=$pct" }
62 substr($$buf, 0, 0, # prepend From line
63 "From $ident\@$type Thu Jan 1 00:00:00 1970$eml->{crlf}");
67 sub atomic_append { # for on-disk destinations (O_APPEND, or O_EXCL)
69 if (defined(my $w = syswrite($lei->{1} // return, $$buf))) {
70 return if $w == length($$buf);
71 $buf = "short atomic write: $w != ".length($$buf);
73 return $lei->note_sigpipe(1);
75 $buf = "atomic write: $!";
80 sub eml2mboxrd ($;$) {
81 my ($eml, $smsg) = @_;
82 my $buf = _mbox_hdr_buf($eml, 'mboxrd', $smsg);
83 if (my $bdy = delete $eml->{bdy}) {
84 $$bdy =~ s/^(>*From )/>$1/gm;
85 $$buf .= $eml->{crlf};
86 substr($$bdy, 0, 0, $$buf); # prepend header
89 $$buf .= $eml->{crlf};
94 my ($eml, $smsg) = @_;
95 my $buf = _mbox_hdr_buf($eml, 'mboxo', $smsg);
96 if (my $bdy = delete $eml->{bdy}) {
97 $$bdy =~ s/^From />From /gm;
98 $$buf .= $eml->{crlf};
99 substr($$bdy, 0, 0, $$buf); # prepend header
102 $$buf .= $eml->{crlf};
106 sub _mboxcl_common ($$$) {
107 my ($buf, $bdy, $crlf) = @_;
108 # add Lines: so mutt won't have to add it on MUA close
109 my $lines = $$bdy =~ tr!\n!\n!;
110 $$buf .= 'Content-Length: '.length($$bdy).$crlf.
111 'Lines: '.$lines.$crlf.$crlf;
112 substr($$bdy, 0, 0, $$buf); # prepend header
117 # mboxcl still escapes "From " lines
119 my ($eml, $smsg) = @_;
120 my $buf = _mbox_hdr_buf($eml, 'mboxcl', $smsg);
121 my $bdy = delete($eml->{bdy}) // \(my $empty = '');
122 $$bdy =~ s/^From />From /gm;
123 _mboxcl_common($buf, $bdy, $eml->{crlf});
126 # mboxcl2 has no "From " escaping
128 my ($eml, $smsg) = @_;
129 my $buf = _mbox_hdr_buf($eml, 'mboxcl2', $smsg);
130 my $bdy = delete($eml->{bdy}) // \(my $empty = '');
131 _mboxcl_common($buf, $bdy, $eml->{crlf});
134 sub git_to_mail { # git->cat_async callback
135 my ($bref, $oid, $type, $size, $arg) = @_;
136 $type // return; # called by git->async_abort
137 my ($write_cb, $smsg) = @$arg;
138 if ($type eq 'missing' && $smsg->{-lms_rw}) {
139 if ($bref = $smsg->{-lms_rw}->local_blob($oid, 1)) {
141 $size = length($$bref);
144 return warn("W: $oid is $type (!= blob)\n") if $type ne 'blob';
145 return warn("E: $oid is empty\n") unless $size;
146 die "BUG: expected=$smsg->{blob} got=$oid" if $smsg->{blob} ne $oid;
147 $write_cb->($bref, $smsg);
150 sub reap_compress { # dwaitpid callback
151 my ($lei, $pid) = @_;
152 my $cmd = delete $lei->{"pid.$pid"};
154 $lei->fail("@$cmd failed", $? >> 8);
157 sub _post_augment_mbox { # open a compressor process from top-level process
158 my ($self, $lei) = @_;
159 my $zsfx = $self->{zsfx} or return;
160 my $cmd = PublicInbox::MboxReader::zsfx2cmd($zsfx, undef, $lei);
161 my ($r, $w) = @{delete $lei->{zpipe}};
162 my $rdr = { 0 => $r, 1 => $lei->{1}, 2 => $lei->{2}, pgid => 0 };
163 my $pid = spawn($cmd, undef, $rdr);
165 my $dup = bless { "pid.$pid" => $cmd }, ref($lei);
166 $dup->{$_} = $lei->{$_} for qw(2 sock);
167 tie *$pp, 'PublicInbox::ProcessPipe', $pid, $w, \&reap_compress, $dup;
171 # --augment existing output destination, with deduplication
172 sub _augment { # MboxReader eml_cb
173 my ($eml, $lei) = @_;
174 # ignore return value, just populate the skv
175 $lei->{dedupe}->is_dup($eml);
178 sub _mbox_augment_kw_maybe {
179 my ($eml, $lei, $lse, $augment) = @_;
180 my $kw = PublicInbox::MboxReader::mbox_keywords($eml);
181 update_kw_maybe($lei, $lse, $eml, $kw);
182 _augment($eml, $lei) if $augment;
185 sub _mbox_write_cb ($$) {
186 my ($self, $lei) = @_;
187 my $ovv = $lei->{ovv};
188 my $m = 'eml2'.$ovv->{fmt};
189 my $eml2mbox = $self->can($m) or die "$self->$m missing";
190 $lei->{1} // die "no stdout ($m, $ovv->{dst})"; # redirected earlier
191 $lei->{1}->autoflush(1);
192 my $atomic_append = !defined($ovv->{lock_path});
193 my $dedupe = $lei->{dedupe};
194 $dedupe->prepare_dedupe;
195 my $lse = $lei->{lse}; # may be undef
196 my $set_recent = $dedupe->has_entries;
197 sub { # for git_to_mail
198 my ($buf, $smsg, $eml) = @_;
199 $eml //= PublicInbox::Eml->new($buf);
201 return if $dedupe->is_dup($eml, $smsg);
202 $lse->xsmsg_vmd($smsg) if $lse;
203 $smsg->{-recent} = 1 if $set_recent;
204 $buf = $eml2mbox->($eml, $smsg);
205 if ($atomic_append) {
206 atomic_append($lei, $buf);
208 my $lk = $ovv->lock_for_scope;
215 sub update_kw_maybe ($$$$) {
216 my ($lei, $lse, $eml, $kw) = @_;
218 my $c = $lse->kw_changed($eml, $kw, my $docids = []);
219 my $vmd = { kw => $kw };
220 if (scalar @$docids) { # already in lei/store
221 $lei->{sto}->wq_do('set_eml_vmd', undef, $vmd, $docids) if $c;
222 } elsif (my $xoids = $lei->{ale}->xoids_for($eml)) {
223 # it's in an external, only set kw, here
224 $lei->{sto}->wq_do('set_xvmd', $xoids, $eml, $vmd);
225 } else { # never-before-seen, import the whole thing
226 # XXX this is critical in protecting against accidental
227 # data loss without --augment
228 $lei->{sto}->wq_do('set_eml', $eml, $vmd);
232 sub _md_update { # maildir_each_eml cb
233 my ($f, $kw, $eml, $lei, $lse, $unlink) = @_;
234 update_kw_maybe($lei, $lse, $eml, $kw);
235 $unlink ? unlink($f) : _augment($eml, $lei);
238 # maildir_each_file callback, \&CORE::unlink doesn't work with it
239 sub _unlink { unlink($_[0]) }
243 sprintf('%x,%x,%x,%x', rand(0xffffffff), time, $$, ++$seq);
246 sub kw2suffix ($;@) {
248 join('', sort(map { $kw2char{$_} // () } @$kw, @_));
251 sub _buf2maildir ($$$$) {
252 my ($dst, $buf, $smsg, $dir) = @_;
253 my $kw = $smsg->{kw} // [];
254 my $rand = ''; # chosen by die roll :P
255 my ($tmp, $fh, $base, $ok);
256 my $common = $smsg->{blob} // _rand;
257 if (defined(my $pct = $smsg->{pct})) { $common .= "=$pct" }
259 $tmp = $dst.'tmp/'.$rand.$common;
260 } while (!($ok = sysopen($fh, $tmp, O_CREAT|O_EXCL|O_WRONLY)) &&
261 $!{EEXIST} && ($rand = _rand.','));
262 if ($ok && print $fh $$buf and close($fh)) {
263 $dst .= $dir; # 'new/' or 'cur/'
266 $base = $rand.$common.':2,'.kw2suffix($kw);
267 } while (!($ok = rename_noreplace($tmp, $dst.$base)) &&
268 $!{EEXIST} && ($rand = _rand.','));
271 my $err = "Error writing $smsg->{blob} to $dst: $!\n";
272 $_[0] = undef; # clobber dst
278 sub _maildir_write_cb ($$) {
279 my ($self, $lei) = @_;
280 my $dedupe = $lei->{dedupe};
281 $dedupe->prepare_dedupe if $dedupe;
282 my $dst = $lei->{ovv}->{dst};
283 my $lse = $lei->{lse}; # may be undef
284 my $lms = $self->{-lms_rw};
285 my $out = $lms ? 'maildir:'.$lei->abs_path($dst) : undef;
286 $lms->lms_write_prepare if $lms;
288 # Favor cur/ and only write to new/ when augmenting. This
289 # saves MUAs from having to do a mass rename when the initial
290 # search result set is huge.
291 my $dir = $dedupe && $dedupe->has_entries ? 'new/' : 'cur/';
292 sub { # for git_to_mail
293 my ($bref, $smsg, $eml) = @_;
294 $dst // return $lei->fail; # dst may be undef-ed in last run
297 return if $dedupe && $dedupe->is_dup($eml //
298 PublicInbox::Eml->new($$bref),
300 $lse->xsmsg_vmd($smsg) if $lse;
301 my $n = _buf2maildir($dst, $bref // \($eml->as_string),
303 $lms->set_src($smsg->oidbin, $out, $n) if $lms;
308 sub _imap_write_cb ($$) {
309 my ($self, $lei) = @_;
310 my $dedupe = $lei->{dedupe};
311 $dedupe->prepare_dedupe if $dedupe;
312 my $append = $lei->{net}->can('imap_append');
313 my $uri = $self->{uri};
314 my $mic = $lei->{net}->mic_get($uri);
315 my $folder = $uri->mailbox;
316 $uri->uidvalidity($mic->uidvalidity($folder));
317 my $lse = $lei->{lse}; # may be undef
318 my $lms = $self->{-lms_rw};
319 $lms->lms_write_prepare if $lms;
320 sub { # for git_to_mail
321 my ($bref, $smsg, $eml) = @_;
322 $mic // return $lei->fail; # mic may be undef-ed in last run
325 return if $dedupe && $dedupe->is_dup($eml //
326 PublicInbox::Eml->new($$bref),
328 $lse->xsmsg_vmd($smsg) if $lse;
329 my $uid = eval { $append->($mic, $folder, $bref, $smsg, $eml) };
334 # imap_append returns UID if IMAP server has UIDPLUS extension
335 ($lms && $uid =~ /\A[0-9]+\z/) and
336 $lms->set_src($smsg->oidbin, $$uri, $uid + 0);
341 sub _text_write_cb ($$) {
342 my ($self, $lei) = @_;
343 my $dedupe = $lei->{dedupe};
344 $dedupe->prepare_dedupe if $dedupe;
345 my $lvt = $lei->{lvt};
346 my $ovv = $lei->{ovv};
347 $lei->{1} // die "no stdout ($ovv->{dst})"; # redirected earlier
348 $lei->{1}->autoflush(1);
349 binmode $lei->{1}, ':utf8';
350 my $lse = $lei->{lse}; # may be undef
351 sub { # for git_to_mail
352 my ($bref, $smsg, $eml) = @_;
353 $lse->xsmsg_vmd($smsg) if $lse;
354 $eml //= PublicInbox::Eml->new($bref);
355 return if $dedupe && $dedupe->is_dup($eml, $smsg);
356 my $lk = $ovv->lock_for_scope;
357 $lei->out(${$lvt->eml_to_text($smsg, $eml)}, "\n");
361 sub _v2_write_cb ($$) {
362 my ($self, $lei) = @_;
363 my $dedupe = $lei->{dedupe};
364 $dedupe->prepare_dedupe if $dedupe;
365 sub { # for git_to_mail
366 my ($bref, $smsg, $eml) = @_;
367 $eml //= PublicInbox::Eml->new($bref);
369 return if $dedupe && $dedupe->is_dup($eml, $smsg);
370 $lei->{v2w}->wq_do('add', $eml); # V2Writable->add
375 sub write_cb { # returns a callback for git_to_mail
376 my ($self, $lei) = @_;
377 # _mbox_write_cb, _maildir_write_cb, _imap_write_cb, _v2_write_cb
378 my $m = "_$self->{base_type}_write_cb";
383 my ($cls, $lei) = @_;
384 my $fmt = $lei->{ovv}->{fmt};
385 my $dst = $lei->{ovv}->{dst};
386 my $self = bless {}, $cls;
388 if ($fmt eq 'maildir') {
389 require PublicInbox::MdirReader;
390 $self->{base_type} = 'maildir';
391 -e $dst && !-d _ and die
392 "$dst exists and is not a directory\n";
393 $lei->{ovv}->{dst} = $dst .= '/' if substr($dst, -1) ne '/';
394 $lei->{opt}->{save} //= \1 if $lei->{cmd} eq 'q';
395 } elsif (substr($fmt, 0, 4) eq 'mbox') {
396 require PublicInbox::MboxReader;
397 $self->can("eml2$fmt") or die "bad mbox format: $fmt\n";
398 $self->{base_type} = 'mbox';
399 if ($lei->{cmd} eq 'q' &&
400 (($lei->path_to_fd($dst) // -1) < 0) &&
401 (-f $dst || !-e _)) {
402 $lei->{opt}->{save} //= \1;
404 } elsif ($fmt =~ /\Aimaps?\z/) {
405 require PublicInbox::NetWriter;
406 require PublicInbox::URIimap;
407 # {net} may exist from "lei up" for auth
408 my $net = $lei->{net} // PublicInbox::NetWriter->new;
409 $net->{quiet} = $lei->{opt}->{quiet};
410 my $uri = PublicInbox::URIimap->new($dst)->canonical;
411 $net->add_url($$uri);
412 my $err = $net->errors($lei);
413 return $lei->fail($err) if $err;
414 $uri->mailbox or return $lei->fail("No mailbox: $dst");
416 $dst = $lei->{ovv}->{dst} = $$uri; # canonicalized
418 $self->{base_type} = 'imap';
419 $lei->{opt}->{save} //= \1 if $lei->{cmd} eq 'q';
420 } elsif ($fmt eq 'text' || $fmt eq 'reply') {
421 require PublicInbox::LeiViewText;
422 $lei->{lvt} = PublicInbox::LeiViewText->new($lei, $fmt);
423 $self->{base_type} = 'text';
424 $self->{-wq_nr_workers} = 1; # for pager
425 @conflict = qw(mua save);
426 } elsif ($fmt eq 'v2') {
427 die "--dedupe=oid and v2 are incompatible\n" if
428 ($lei->{opt}->{dedupe}//'') eq 'oid';
429 $self->{base_type} = 'v2';
430 $self->{-wq_nr_workers} = 1; # v2 has shards
431 $lei->{opt}->{save} = \1;
432 $dst = $lei->{ovv}->{dst} = $lei->abs_path($dst);
433 @conflict = qw(mua sort);
435 die "bad mail --format=$fmt\n";
437 if ($self->{base_type} =~ /\A(?:text|mbox)\z/) {
438 (-d $dst || (-e _ && !-w _)) and die
439 "$dst exists and is not a writable file\n";
441 my @err = map { defined($lei->{opt}->{$_}) ? "--$_" : () } @conflict;
442 die "@err incompatible with $fmt\n" if @err;
444 $lei->{dedupe} = $lei->{lss} // do {
445 my $dd_cls = 'PublicInbox::'.
446 ($lei->{opt}->{save} ? 'LeiSavedSearch' : 'LeiDedupe');
447 eval "require $dd_cls";
448 die "$dd_cls: $@" if $@;
449 my $dd = $dd_cls->new($lei);
450 $lei->{lss} //= $dd if $dd && $dd->can('cfg_set');
456 sub _pre_augment_maildir {
457 my ($self, $lei) = @_;
458 my $dst = $lei->{ovv}->{dst};
459 for my $x (qw(tmp new cur)) {
463 File::Path::mkpath($d);
464 -d $d or die "$d is not a directory";
466 # for utime, so no opendir
467 open $self->{poke_dh}, '<', "${dst}cur" or die "open ${dst}cur: $!";
470 sub clobber_dst_prepare ($;$) {
472 if (my $lms = defined($f) ? $lei->lms : undef) {
473 $lms->lms_write_prepare;
474 $lms->forget_folders($f);
476 my $dedupe = $lei->{dedupe} or return;
477 $dedupe->reset_dedupe if $dedupe->can('reset_dedupe');
480 sub _do_augment_maildir {
481 my ($self, $lei) = @_;
482 return if $lei->{cmd} eq 'up';
483 my $dst = $lei->{ovv}->{dst};
484 my $lse = $lei->{opt}->{'import-before'} ? $lei->{lse} : undef;
485 my $mdr = PublicInbox::MdirReader->new;
486 if ($lei->{opt}->{augment}) {
487 my $dedupe = $lei->{dedupe};
488 if ($dedupe && $dedupe->prepare_dedupe) {
489 $mdr->{shard_info} = $self->{shard_info};
490 $mdr->maildir_each_eml($dst, \&_md_update, $lei, $lse);
491 $dedupe->pause_dedupe;
494 clobber_dst_prepare($lei, "maildir:$dst");
495 $mdr->{shard_info} = $self->{shard_info};
496 $mdr->maildir_each_eml($dst, \&_md_update, $lei, $lse, 1);
497 } else {# clobber existing Maildir
498 clobber_dst_prepare($lei, "maildir:$dst");
499 $mdr->maildir_each_file($dst, \&_unlink);
503 sub _imap_augment_or_delete { # PublicInbox::NetReader::imap_each cb
504 my ($uri, $uid, $kw, $eml, $lei, $lse, $delete_mic) = @_;
505 update_kw_maybe($lei, $lse, $eml, $kw);
507 $lei->{net}->imap_delete_1($uri, $uid, $delete_mic);
509 _augment($eml, $lei);
513 sub _do_augment_imap {
514 my ($self, $lei) = @_;
515 return if $lei->{cmd} eq 'up';
516 my $net = $lei->{net};
517 my $lse = $lei->{opt}->{'import-before'} ? $lei->{lse} : undef;
518 if ($lei->{opt}->{augment}) {
519 my $dedupe = $lei->{dedupe};
520 if ($dedupe && $dedupe->prepare_dedupe) {
521 $net->imap_each($self->{uri}, \&_imap_augment_or_delete,
523 $dedupe->pause_dedupe;
527 clobber_dst_prepare($lei, "$self->{uri}");
528 $net->imap_each($self->{uri}, \&_imap_augment_or_delete,
529 $lei, $lse, \$delete_mic);
530 $delete_mic->expunge if $delete_mic;
531 } elsif (!$self->{-wq_worker_nr}) { # undef or 0
532 # clobber existing IMAP folder
533 clobber_dst_prepare($lei, "$self->{uri}");
534 $net->imap_delete_all($self->{uri});
538 sub _pre_augment_text {
539 my ($self, $lei) = @_;
540 my $dst = $lei->{ovv}->{dst};
542 my $devfd = $lei->path_to_fd($dst) // die "bad $dst";
544 $out = $lei->{$devfd};
545 } else { # normal-looking path
547 open $out, '>', $dst or die "open($dst): $!";
548 } elsif (-f _ || !-e _) {
549 # text allows augment, HTML/Atom won't
550 my $mode = $lei->{opt}->{augment} ? '>>' : '>';
551 open $out, $mode, $dst or die "open($mode, $dst): $!";
553 die "$dst is not a file or FIFO\n";
556 $lei->{ovv}->ovv_out_lk_init if !$lei->{ovv}->{lock_path};
561 sub _pre_augment_mbox {
562 my ($self, $lei) = @_;
563 my $dst = $lei->{ovv}->{dst};
565 my $devfd = $lei->path_to_fd($dst) // die "bad $dst";
567 $out = $lei->{$devfd};
568 } else { # normal-looking path
570 open $out, '>', $dst or die "open($dst): $!";
571 } elsif (-f _ || !-e _) {
572 require PublicInbox::MboxLock;
573 my $m = $lei->{opt}->{'lock'} //
574 PublicInbox::MboxLock->defaults;
575 $self->{mbl} = PublicInbox::MboxLock->acq($dst, 1, $m);
576 $out = $self->{mbl}->{fh};
578 die "$dst is not a file or FIFO\n";
580 $lei->{old_1} = $lei->{1}; # keep for spawning MUA
582 # Perl does SEEK_END even with O_APPEND :<
583 $self->{seekable} = seek($out, 0, SEEK_SET);
584 if (!$self->{seekable} && !$!{ESPIPE} && !defined($devfd)) {
585 die "seek($dst): $!\n";
587 if (!$self->{seekable}) {
588 my $imp_before = $lei->{opt}->{'import-before'};
589 die "--import-before specified but $dst is not seekable\n"
590 if $imp_before && !ref($imp_before);
591 die "--augment specified but $dst is not seekable\n" if
592 $lei->{opt}->{augment};
593 die "cannot --save with unseekable $dst\n" if
594 $lei->{dedupe} && $lei->{dedupe}->can('reset_dedupe');
596 if ($self->{zsfx} = PublicInbox::MboxReader::zsfx($dst)) {
597 pipe(my ($r, $w)) or die "pipe: $!";
598 $lei->{zpipe} = [ $r, $w ];
599 $lei->{ovv}->{lock_path} and
600 die 'BUG: unexpected {ovv}->{lock_path}';
601 $lei->{ovv}->ovv_out_lk_init;
602 } elsif (!$self->{seekable} && !$lei->{ovv}->{lock_path}) {
603 $lei->{ovv}->ovv_out_lk_init;
609 sub _do_augment_mbox {
610 my ($self, $lei) = @_;
611 return unless $self->{seekable};
612 my $opt = $lei->{opt};
613 return if $lei->{cmd} eq 'up';
615 my ($fmt, $dst) = @{$lei->{ovv}}{qw(fmt dst)};
616 return clobber_dst_prepare($lei) unless -s $out;
617 unless ($opt->{augment} || $opt->{'import-before'}) {
618 truncate($out, 0) or die "truncate($dst): $!";
622 if (my $zsfx = $self->{zsfx}) {
623 $rd = PublicInbox::MboxReader::zsfxcat($out, $zsfx, $lei);
625 open($rd, '+>>&', $out) or die "dup: $!";
628 if ($opt->{augment}) {
629 $dedupe = $lei->{dedupe};
630 $dedupe->prepare_dedupe if $dedupe;
632 clobber_dst_prepare($lei);
634 if ($opt->{'import-before'}) { # the default
635 my $lse = $lei->{lse};
636 PublicInbox::MboxReader->$fmt($rd, \&_mbox_augment_kw_maybe,
637 $lei, $lse, $opt->{augment});
638 if (!$opt->{augment} and !truncate($out, 0)) {
639 die "truncate($dst): $!";
641 } else { # --augment --no-import-before
642 PublicInbox::MboxReader->$fmt($rd, \&_augment, $lei);
644 # maybe some systems don't honor O_APPEND, Perl does this:
645 seek($out, 0, SEEK_END) or die "seek $dst: $!";
646 $dedupe->pause_dedupe if $dedupe;
649 sub v2w_done_wait { # dwaitpid callback
650 my ($arg, $pid) = @_;
651 my ($v2w, $lei) = @$arg;
652 $lei->child_error($?, "error for $v2w->{ibx}->{inboxdir}") if $?;
655 sub _pre_augment_v2 {
656 my ($self, $lei) = @_;
657 my $dir = $self->{dst};
658 require PublicInbox::InboxWritable;
661 my $opt = { -min_inbox_version => 2 };
662 require PublicInbox::Admin;
663 my @ibx = PublicInbox::Admin::resolve_inboxes([ $dir ], $opt);
664 $ibx = $ibx[0] or die "$dir is not a v2 inbox\n";
667 $ibx = PublicInbox::Inbox->new({
668 name => 'lei-result', # XXX configurable
671 address => [ 'lei@example.com' ],
674 PublicInbox::InboxWritable->new($ibx, @creat);
675 $ibx->init_inbox if @creat;
676 my $v2w = $ibx->importer;
677 $v2w->wq_workers_start("lei/v2w $dir", 1, $lei->oldset, {lei => $lei});
678 $v2w->wq_wait_async(\&v2w_done_wait, $lei);
680 return if !$lei->{opt}->{shared};
681 my $d = "$lei->{ale}->{git}->{git_dir}/objects";
682 my $al = "$dir/git/0.git/objects/info/alternates";
683 open my $fh, '+>>', $al or die "open($al): $!";
684 seek($fh, 0, SEEK_SET) or die "seek($al): $!";
685 grep(/\A\Q$d\E\n/, <$fh>) and return;
686 print $fh "$d\n" or die "print($al): $!";
687 close $fh or die "close($al): $!";
690 sub pre_augment { # fast (1 disk seek), runs in same process as post_augment
691 my ($self, $lei) = @_;
692 # _pre_augment_maildir, _pre_augment_mbox, _pre_augment_v2
693 my $m = $self->can("_pre_augment_$self->{base_type}") or return;
697 sub do_augment { # slow, runs in wq worker
698 my ($self, $lei) = @_;
699 # _do_augment_maildir, _do_augment_mbox, or _do_augment_imap
700 my $m = $self->can("_do_augment_$self->{base_type}") or return;
704 # fast (spawn compressor or mkdir), runs in same process as pre_augment
706 my ($self, $lei, @args) = @_;
707 $self->{-au_noted}++ and $lei->qerr("# writing to $self->{dst} ...");
709 my $wait = $lei->{opt}->{'import-before'} ?
710 $lei->{sto}->wq_do('checkpoint', 1) : 0;
712 my $m = $self->can("_post_augment_$self->{base_type}") or return;
713 $m->($self, $lei, @args);
716 # called by every single l2m worker process
719 my $lei = $self->{lei};
720 # lei_xsearch can start as soon as all l2m workers get here
721 $lei->{pkt_op_p}->pkt_do('incr_start_query') or
722 die "incr_start_query: $!";
724 if (lock_free($self)) { # all workers do_augment
725 my $mod = $self->{-wq_nr_workers};
726 my $shard = $self->{-wq_worker_nr};
727 if (my $net = $lei->{net}) {
728 $net->{shard_info} = [ $mod, $shard ];
730 $self->{shard_info} = [ $mod, $shard ];
732 $aug = 'incr_post_augment';
733 } elsif ($self->{-wq_worker_nr} == 0) { # 1st worker do_augment
734 $aug = 'do_post_augment';
737 local $0 = 'do_augment';
738 eval { do_augment($self, $lei) };
739 $lei->fail($@) if $@;
740 $lei->{pkt_op_p}->pkt_do($aug) or die "pkt_do($aug): $!";
742 # done augmenting, connect the compressor pipe for each worker
743 if (my $zpipe = delete $lei->{zpipe}) {
744 $lei->{1} = $zpipe->[1];
747 my $au_peers = delete $self->{au_peers};
748 if ($au_peers) { # wait for peer l2m to finish augmenting:
749 $au_peers->[1] = undef;
750 sysread($au_peers->[0], my $barrier1, 1);
752 $self->{wcb} = $self->write_cb($lei);
753 if ($au_peers) { # wait for peer l2m to set write_cb
754 $au_peers->[3] = undef;
755 sysread($au_peers->[2], my $barrier2, 1);
759 sub ipc_atfork_child {
761 my $lei = $self->{lei};
762 $lei->_lei_atfork_child;
763 $lei->{auth}->do_auth_atfork($self) if $lei->{auth};
764 $SIG{__WARN__} = PublicInbox::Eml::warn_ignore_cb();
765 $self->{git} = $self->{lei}->{ale}->git;
766 $SIG{TERM} = sub { # avoid ->DESTROY ordering problems
767 my $git = delete $self->{git};
768 $git->async_wait_all if $git;
771 $self->SUPER::ipc_atfork_child;
775 $_[0]->{base_type} =~ /\A(?:maildir|imap|jmap)\z/ ? 1 : 0;
778 # wakes up the MUA when complete so it can refresh messages list
781 if ($self->{base_type} eq 'maildir') {
783 utime($t, $t, $self->{poke_dh}) or warn "futimes: $!";
787 sub write_mail { # via ->wq_io_do
788 my ($self, $smsg, $eml) = @_;
789 return $self->{wcb}->(undef, $smsg, $eml) if $eml;
790 $smsg->{-lms_rw} = $self->{-lms_rw};
791 $self->{git}->cat_async($smsg->{blob}, \&git_to_mail,
792 [$self->{wcb}, $smsg]);
795 sub wq_atexit_child {
797 local $PublicInbox::DS::in_loop = 0; # waitpid synchronously
798 my $lei = $self->{lei};
800 $lei->{ale}->git->async_wait_all;
801 my ($nr_w, $nr_s) = delete(@$lei{qw(-nr_write -nr_seen)});
803 return if $lei->{early_mua} || !$lei->{-progress} || !$lei->{pkt_op_p};
804 $lei->{pkt_op_p}->pkt_do('l2m_progress', $nr_w, $nr_s);
807 # runs on a 1s timer in lei-daemon
808 sub augment_inprogress {
809 my ($err, $opt, $dst, $au_noted) = @_;
811 return if $$au_noted++ || !$err || !defined(fileno($err));
812 print $err '# '.($opt->{'import-before'} ?
813 "importing non-external contents of $dst" : (
814 ($opt->{dedupe} // 'content') ne 'none') ?
815 "scanning old contents of $dst for dedupe" :
816 "removing old contents of $dst")." ...\n";
818 warn "E: $@ ($dst)" if $@;
821 # called in top-level lei-daemon when LeiAuth is done
822 sub net_merge_all_done {
823 my ($self, $lei) = @_;
824 if ($PublicInbox::DS::in_loop &&
825 $self->can("_do_augment_$self->{base_type}") &&
826 !$lei->{opt}->{quiet}) {
827 $self->{-au_noted} = 0;
828 PublicInbox::DS::add_timer(1, \&augment_inprogress,
829 $lei->{2}, $lei->{opt},
830 $self->{dst}, \$self->{-au_noted});
832 $self->wq_broadcast('do_post_auth');