1 # Copyright (C) 2018 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # This interface wraps and mimics PublicInbox::Import
5 package PublicInbox::V2Writable;
8 use base qw(PublicInbox::Lock);
9 use PublicInbox::SearchIdxPart;
10 use PublicInbox::MIME;
12 use PublicInbox::Import;
13 use PublicInbox::MID qw(mids);
14 use PublicInbox::ContentId qw(content_id content_digest);
15 use PublicInbox::Inbox;
16 use PublicInbox::OverIdx;
17 use PublicInbox::Msgmap;
18 use PublicInbox::Spawn;
21 # an estimate of the post-packed size to the raw uncompressed size
22 my $PACKING_FACTOR = 0.4;
24 # assume 2 cores if GNU nproc(1) is not available
26 my $n = int($ENV{NPROC} || `nproc 2>/dev/null` || 2);
27 # subtract for the main process and git-fast-import
32 sub count_partitions ($) {
35 my $xpfx = $self->{xpfx};
37 # always load existing partitions in case core count changes:
38 # Also, partition count may change while -watch is running
41 foreach my $part (<$xpfx/*>) {
42 -d $part && $part =~ m!/\d+\z! or next;
44 Search::Xapian::Database->new($part)->close;
53 my ($class, $v2ibx, $creat) = @_;
54 my $dir = $v2ibx->{mainrepo} or die "no mainrepo in inbox\n";
58 File::Path::mkpath($dir);
60 die "$dir does not exist\n";
64 $v2ibx = PublicInbox::InboxWritable->new($v2ibx);
66 my $xpfx = "$dir/xap" . PublicInbox::Search::SCHEMA_VERSION;
69 im => undef, # PublicInbox::Import
73 over => PublicInbox::OverIdx->new("$xpfx/over.sqlite3", 1),
74 lock_path => "$dir/inbox.lock",
75 # limit each git repo (epoch) to 1GB or so
76 rotate_bytes => int((1024 * 1024 * 1024) / $PACKING_FACTOR),
77 last_commit => [], # git repo -> commit
79 $self->{partitions} = count_partitions($self) || nproc_parts();
84 my ($self, $parallel) = @_;
85 $self->{parallel} = $parallel;
88 git_dir_latest($self, \$epoch_max);
89 $self->git_init($epoch_max >= 0 ? $epoch_max : 0);
93 # returns undef on duplicate or spam
94 # mimics Import::add and wraps it for v2
96 my ($self, $mime, $check_cb) = @_;
100 $mime = $check_cb->($mime) or return;
103 # All pipes (> $^F) known to Perl 5.6+ have FD_CLOEXEC set,
104 # as does SQLite 3.4.1+ (released in 2007-07-20), and
105 # Xapian 1.3.2+ (released 2015-03-15).
106 # For the most part, we can spawn git-fast-import without
107 # leaking FDs to it...
111 my $num = num_for($self, $mime, \$mid0);
112 defined $num or return; # duplicate
113 defined $mid0 or die "BUG: $mid0 undefined\n";
114 my $im = $self->importer;
115 my $cmt = $im->add($mime);
116 $cmt = $im->get_mark($cmt);
117 $self->{last_commit}->[$self->{epoch_max}] = $cmt;
119 my ($oid, $len, $msgref) = @{$im->{last_object}};
120 $self->{over}->add_overview($mime, $len, $num, $oid, $mid0);
121 my $nparts = $self->{partitions};
122 my $part = $num % $nparts;
123 my $idx = $self->idx_part($part);
124 $idx->index_raw($len, $msgref, $num, $oid, $mid0, $mime);
125 my $n = $self->{transact_bytes} += $len;
126 if ($n > (PublicInbox::SearchIdx::BATCH_BYTES * $nparts)) {
134 my ($self, $mime, $mid0) = @_;
135 my $mids = mids($mime->header_obj);
137 my $mid = $mids->[0];
138 my $num = $self->{mm}->mid_insert($mid);
139 if (defined $num) { # common case
144 # crap, Message-ID is already known, hope somebody just resent:
145 foreach my $m (@$mids) {
146 # read-only lookup now safe to do after above barrier
147 my $existing = $self->lookup_content($mime, $m);
148 # easy, don't store duplicates
149 # note: do not add more diagnostic info here since
150 # it gets noisy on public-inbox-watch restarts
155 warn "<$mid> reused for mismatched content\n";
157 # try the rest of the mids
158 for(my $i = $#$mids; $i >= 1; $i--) {
160 $num = $self->{mm}->mid_insert($m);
162 warn "alternative <$m> for <$mid> found\n";
168 # none of the existing Message-IDs are good, generate a new one:
169 num_for_harder($self, $mime, $mid0);
173 my ($self, $mime, $mid0) = @_;
175 my $hdr = $mime->header_obj;
176 my $dig = content_digest($mime);
177 $$mid0 = PublicInbox::Import::digest2mid($dig, $hdr);
178 my $num = $self->{mm}->mid_insert($$mid0);
179 unless (defined $num) {
180 # it's hard to spoof the last Received: header
181 my @recvd = $hdr->header_raw('Received');
182 $dig->add("Received: $_") foreach (@recvd);
183 $$mid0 = PublicInbox::Import::digest2mid($dig, $hdr);
184 $num = $self->{mm}->mid_insert($$mid0);
186 # fall back to a random Message-ID and give up determinism:
187 until (defined($num)) {
189 $$mid0 = PublicInbox::Import::digest2mid($dig, $hdr);
190 warn "using random Message-ID <$$mid0> as fallback\n";
191 $num = $self->{mm}->mid_insert($$mid0);
194 PublicInbox::Import::append_mid($hdr, $$mid0);
199 my ($self, $part) = @_;
200 $self->{idx_parts}->[$part];
206 return if $self->{idx_parts};
207 my $ibx = $self->{-inbox};
209 # do not leak read-only FDs to child processes, we only have these
210 # FDs for duplicate detection so they should not be
211 # frequently activated.
212 delete $ibx->{$_} foreach (qw(git mm search));
214 if ($self->{parallel}) {
215 pipe(my ($r, $w)) or die "pipe failed: $!";
216 $self->{bnote} = [ $r, $w ];
220 my $over = $self->{over};
222 $ibx->with_umask(sub {
226 # -compact can change partition count while -watch is idle
227 my $nparts = count_partitions($self);
228 if ($nparts && $nparts != $self->{partitions}) {
229 $self->{partitions} = $nparts;
232 # need to create all parts before initializing msgmap FD
233 my $max = $self->{partitions} - 1;
235 # idx_parts must be visible to all forked processes
236 my $idx = $self->{idx_parts} = [];
237 for my $i (0..$max) {
238 push @$idx, PublicInbox::SearchIdxPart->new($self, $i);
241 # Now that all subprocesses are up, we can open the FDs
243 my $mm = $self->{mm} = PublicInbox::Msgmap->new_file(
244 "$self->{-inbox}->{mainrepo}/msgmap.sqlite3", 1);
245 $mm->{dbh}->begin_work;
250 my ($self, $purge) = @_; # $purge = { $object_id => 1, ... }
252 my $pfx = "$self->{-inbox}->{mainrepo}/git";
254 foreach my $i (0..$self->{epoch_max}) {
255 my $git = PublicInbox::Git->new("$pfx/$i.git");
256 my $im = $self->import_init($git, 0, 1);
257 $purges->[$i] = $im->purge_oids($purge);
262 sub remove_internal {
263 my ($self, $mime, $cmt_msg, $purge) = @_;
265 my $im = $self->importer unless $purge;
266 my $over = $self->{over};
267 my $cid = content_id($mime);
268 my $parts = $self->{idx_parts};
269 my $mm = $self->{mm};
271 my $mids = mids($mime->header_obj);
273 # We avoid introducing new blobs into git since the raw content
274 # can be slightly different, so we do not need the user-supplied
275 # message now that we have the mids and content_id
279 foreach my $mid (@$mids) {
282 while (my $smsg = $over->next_by_mid($mid, \$id, \$prev)) {
283 my $msg = get_blob($self, $smsg);
284 if (!defined($msg)) {
285 warn "broken smsg for $mid\n";
289 my $cur = PublicInbox::MIME->new($msg);
290 if (content_id($cur) eq $cid) {
291 $smsg->{mime} = $cur;
292 $gone{$smsg->{num}} = [ $smsg, \$orig ];
295 my $n = scalar keys %gone;
298 warn "BUG: multiple articles linked to <$mid>\n",
299 join(',', sort keys %gone), "\n";
301 foreach my $num (keys %gone) {
302 my ($smsg, $orig) = @{$gone{$num}};
303 $mm->num_delete($num);
304 # $removed should only be set once assuming
305 # no bugs in our deduplication code:
307 my $oid = $smsg->{blob};
311 ($mark, undef) = $im->remove($orig, $cmt_msg);
314 $self->unindex_oid_remote($oid, $mid);
319 my $cmt = $im->get_mark($mark);
320 $self->{last_commit}->[$self->{epoch_max}] = $cmt;
322 if ($purge && scalar keys %$purge) {
323 return purge_oids($self, $purge);
329 my ($self, $mime, $cmt_msg) = @_;
330 remove_internal($self, $mime, $cmt_msg);
334 my ($self, $mime) = @_;
335 my $purges = remove_internal($self, $mime, undef, {});
336 $self->idx_init if @$purges; # ->done is called on purges
337 for my $i (0..$#$purges) {
338 defined(my $cmt = $purges->[$i]) or next;
339 $self->{last_commit}->[$i] = $cmt;
344 sub last_commit_part ($$;$) {
345 my ($self, $i, $cmt) = @_;
346 my $v = PublicInbox::Search::SCHEMA_VERSION();
347 $self->{mm}->last_commit_xap($v, $i, $cmt);
350 sub set_last_commits ($) {
352 defined(my $epoch_max = $self->{epoch_max}) or return;
353 my $last_commit = $self->{last_commit};
354 foreach my $i (0..$epoch_max) {
355 defined(my $cmt = $last_commit->[$i]) or next;
356 $last_commit->[$i] = undef;
357 last_commit_part($self, $i, $cmt);
363 $self->{bnote} or return;
365 my $barrier = { map { $_ => 1 } (0..$n) };
369 my ($self, $barrier) = @_;
370 my $bnote = $self->{bnote} or return;
372 while (scalar keys %$barrier) {
373 defined(my $l = $r->getline) or die "EOF on barrier_wait: $!";
374 $l =~ /\Abarrier (\d+)/ or die "bad line on barrier_wait: $l";
375 delete $barrier->{$1} or die "bad part[$1] on barrier wait";
379 sub checkpoint ($;$) {
380 my ($self, $wait) = @_;
382 if (my $im = $self->{im}) {
389 my $parts = $self->{idx_parts};
391 my $dbh = $self->{mm}->{dbh};
393 # SQLite msgmap data is second in importance
396 # SQLite overview is third
397 $self->{over}->commit_lazy;
399 # Now deal with Xapian
401 my $barrier = $self->barrier_init(scalar @$parts);
403 # each partition needs to issue a barrier command
404 $_->remote_barrier for @$parts;
406 # wait for each Xapian partition
407 $self->barrier_wait($barrier);
409 $_->remote_commit for @$parts;
412 # last_commit is special, don't commit these until
413 # remote partitions are done:
415 set_last_commits($self);
420 $self->{transact_bytes} = 0;
423 # issue a write barrier to ensure all data is visible to other processes
424 # and read-only ops. Order of data importance is: git > SQLite > Xapian
425 sub barrier { checkpoint($_[0], 1) };
429 my $im = delete $self->{im};
430 $im->done if $im; # PublicInbox::Import::done
432 my $mm = delete $self->{mm};
433 $mm->{dbh}->commit if $mm;
434 my $parts = delete $self->{idx_parts};
436 $_->remote_close for @$parts;
438 $self->{over}->disconnect;
439 delete $self->{bnote};
440 $self->{transact_bytes} = 0;
441 $self->lock_release if $parts;
445 my ($self, $epoch) = @_;
446 my $pfx = "$self->{-inbox}->{mainrepo}/git";
447 my $git_dir = "$pfx/$epoch.git";
448 my @cmd = (qw(git init --bare -q), $git_dir);
449 PublicInbox::Import::run_die(\@cmd);
451 my $all = "$self->{-inbox}->{mainrepo}/all.git";
453 @cmd = (qw(git init --bare -q), $all);
454 PublicInbox::Import::run_die(\@cmd);
455 @cmd = (qw/git config/, "--file=$all/config",
456 'repack.writeBitmaps', 'true');
457 PublicInbox::Import::run_die(\@cmd);
460 @cmd = (qw/git config/, "--file=$git_dir/config",
461 'include.path', '../../all.git/config');
462 PublicInbox::Import::run_die(\@cmd);
464 my $alt = "$all/objects/info/alternates";
465 my $new_obj_dir = "../../git/$epoch.git/objects";
468 open(my $fh, '<', $alt) or die "open < $alt: $!\n";
469 %alts = map { chomp; $_ => 1 } (<$fh>);
471 return $git_dir if $alts{$new_obj_dir};
472 open my $fh, '>>', $alt or die "open >> $alt: $!\n";
473 print $fh "$new_obj_dir\n" or die "print >> $alt: $!\n";
474 close $fh or die "close $alt: $!\n";
479 my ($self, $max) = @_;
481 my $pfx = "$self->{-inbox}->{mainrepo}/git";
482 return unless -d $pfx;
484 opendir my $dh, $pfx or die "opendir $pfx: $!\n";
485 while (defined(my $git_dir = readdir($dh))) {
486 $git_dir =~ m!\A(\d+)\.git\z! or next;
489 $latest = "$pfx/$git_dir";
497 my $im = $self->{im};
499 if ($im->{bytes_added} < $self->{rotate_bytes}) {
506 my $git_dir = $self->git_init(++$self->{epoch_max});
507 my $git = PublicInbox::Git->new($git_dir);
508 return $self->import_init($git, 0);
513 my $latest = git_dir_latest($self, \$max);
514 if (defined $latest) {
515 my $git = PublicInbox::Git->new($latest);
516 my $packed_bytes = $git->packed_bytes;
517 if ($packed_bytes >= $self->{rotate_bytes}) {
520 $self->{epoch_max} = $max;
521 return $self->import_init($git, $packed_bytes);
524 $self->{epoch_max} = $epoch;
525 $latest = $self->git_init($epoch);
526 $self->import_init(PublicInbox::Git->new($latest), 0);
530 my ($self, $git, $packed_bytes, $tmp) = @_;
531 my $im = PublicInbox::Import->new($git, undef, undef, $self->{-inbox});
532 $im->{bytes_added} = int($packed_bytes / $PACKING_FACTOR);
533 $im->{want_object_info} = 1;
534 $im->{lock_path} = undef;
535 $im->{path_type} = 'v2';
536 $self->{im} = $im unless $tmp;
542 my ($mid, $cur, $new) = @_;
543 use File::Temp qw(tempfile);
544 use PublicInbox::Spawn qw(spawn);
546 my ($ah, $an) = tempfile('email-cur-XXXXXXXX', TMPDIR => 1);
547 print $ah $cur->as_string or die "print: $!";
548 close $ah or die "close: $!";
549 my ($bh, $bn) = tempfile('email-new-XXXXXXXX', TMPDIR => 1);
550 PublicInbox::Import::drop_unwanted_headers($new);
551 print $bh $new->as_string or die "print: $!";
552 close $bh or die "close: $!";
553 my $cmd = [ qw(diff -u), $an, $bn ];
554 print STDERR "# MID conflict <$mid>\n";
555 my $pid = spawn($cmd, undef, { 1 => 2 });
556 defined $pid or die "diff failed to spawn $!";
557 waitpid($pid, 0) == $pid or die "diff did not finish";
562 my ($self, $smsg) = @_;
563 if (my $im = $self->{im}) {
564 my $msg = $im->cat_blob($smsg->{blob});
567 # older message, should be in alternates
568 my $ibx = $self->{-inbox};
569 $ibx->msg_by_smsg($smsg);
573 my ($self, $mime, $mid) = @_;
574 my $over = $self->{over};
575 my $cid = content_id($mime);
578 while (my $smsg = $over->next_by_mid($mid, \$id, \$prev)) {
579 my $msg = get_blob($self, $smsg);
580 if (!defined($msg)) {
581 warn "broken smsg for $mid\n";
584 my $cur = PublicInbox::MIME->new($msg);
585 if (content_id($cur) eq $cid) {
586 $smsg->{mime} = $cur;
591 # XXX DEBUG_DIFF is experimental and may be removed
592 diff($mid, $cur, $mime) if $ENV{DEBUG_DIFF};
599 my $fh = delete $self->{reindex_pipe};
601 if (my $parts = $self->{idx_parts}) {
602 $_->atfork_child foreach @$parts;
604 if (my $im = $self->{im}) {
607 die "unexpected mm" if $self->{mm};
608 close $self->{bnote}->[0] or die "close bnote[0]: $!\n";
613 my ($self, $D, $git, $oid) = @_;
614 my $msgref = $git->cat_file($oid);
615 my $mime = PublicInbox::MIME->new($$msgref);
616 my $mids = mids($mime->header_obj);
617 my $cid = content_id($mime);
618 foreach my $mid (@$mids) {
619 $D->{"$mid\0$cid"} = 1;
624 my ($self, $mm_tmp, $D, $git, $oid, $regen) = @_;
626 my $msgref = $git->cat_file($oid, \$len);
627 my $mime = PublicInbox::MIME->new($$msgref);
628 my $mids = mids($mime->header_obj);
629 my $cid = content_id($mime);
631 # get the NNTP article number we used before, highest number wins
632 # and gets deleted from mm_tmp;
636 foreach my $mid (@$mids) {
637 $del += (delete $D->{"$mid\0$cid"} || 0);
638 my $n = $mm_tmp->num_for($mid);
639 if (defined $n && $n > $num) {
644 if (!defined($mid0) && $regen && !$del) {
646 die "BUG: ran out of article numbers\n" if $num <= 0;
647 my $mm = $self->{mm};
648 foreach my $mid (reverse @$mids) {
649 if ($mm->mid_set($num, $mid) == 1) {
654 if (!defined($mid0)) {
655 my $id = '<' . join('> <', @$mids) . '>';
656 warn "Message-ID $id unusable for $num\n";
657 foreach my $mid (@$mids) {
658 defined(my $n = $mm->num_for($mid)) or next;
659 warn "#$n previously mapped for <$mid>\n";
664 if (!defined($mid0) || $del) {
665 if (!defined($mid0) && $del) { # expected for deletes
670 my $id = '<' . join('> <', @$mids) . '>';
672 warn "Skipping $id, no article number found\n";
673 if ($del && defined($mid0)) {
674 warn "$id was deleted $del " .
675 "time(s) but mapped to article #$num\n";
680 $mm_tmp->mid_delete($mid0) or
681 die "failed to delete <$mid0> for article #$num\n";
683 $self->{over}->add_overview($mime, $len, $num, $oid, $mid0);
684 my $nparts = $self->{partitions};
685 my $part = $num % $nparts;
686 my $idx = $self->idx_part($part);
687 $idx->index_raw($len, $msgref, $num, $oid, $mid0, $mime);
688 my $n = $self->{transact_bytes} += $len;
689 if ($n > (PublicInbox::SearchIdx::BATCH_BYTES * $nparts)) {
691 $mm_tmp->atfork_prepare;
692 $self->done; # release lock
693 # allow -watch or -mda to write...
694 $self->idx_init; # reacquire lock
695 $mm_tmp->atfork_parent;
699 # only update last_commit for $i on reindex iff newer than current
700 sub update_last_commit {
701 my ($self, $git, $i, $cmt) = @_;
702 my $last = last_commit_part($self, $i);
703 if (defined $last && is_ancestor($git, $last, $cmt)) {
704 my @cmd = (qw(rev-list --count), "$last..$cmt");
705 chomp(my $n = $git->qx(@cmd));
706 return if $n ne '' && $n == 0;
708 last_commit_part($self, $i, $cmt);
711 sub git_dir_n ($$) { "$_[0]->{-inbox}->{mainrepo}/git/$_[1].git" }
714 my ($self, $epoch_max) = @_;
716 for (my $i = $epoch_max; $i >= 0; $i--) {
717 $heads->[$i] = last_commit_part($self, $i);
722 sub is_ancestor ($$$) {
723 my ($git, $cur, $tip) = @_;
724 return 0 unless $git->check($cur);
725 my $cmd = [ 'git', "--git-dir=$git->{git_dir}",
726 qw(merge-base --is-ancestor), $cur, $tip ];
727 my $pid = spawn($cmd);
728 defined $pid or die "spawning ".join(' ', @$cmd)." failed: $!";
729 waitpid($pid, 0) == $pid or die join(' ', @$cmd) .' did not finish';
734 my ($self, $opts, $epoch_max, $ranges) = @_;
736 my $head = $self->{-inbox}->{ref_head} || 'refs/heads/master';
737 for (my $i = $epoch_max; $i >= 0; $i--) {
738 die "already indexing!\n" if $self->{index_pipe};
739 my $git_dir = git_dir_n($self, $i);
740 -d $git_dir or next; # missing parts are fine
741 my $git = PublicInbox::Git->new($git_dir);
742 chomp(my $tip = $git->qx('rev-parse', $head));
744 if (defined(my $cur = $ranges->[$i])) {
745 $range = "$cur..$tip";
746 if (is_ancestor($git, $cur, $tip)) { # common case
747 my $n = $git->qx(qw(rev-list --count), $range);
750 $ranges->[$i] = undef;
755 discontiguous range: $range
756 Rewritten history? (in $git_dir)
758 my $base = $git->qx('merge-base', $tip, $cur);
761 $range = "$base..$tip";
762 warn "found merge-base: $base\n"
766 discarding history at $cur
770 reindexing $git_dir starting at
773 $self->{"unindex-range.$i"} = "$base..$cur";
776 $range = $tip; # all of it
778 $ranges->[$i] = $range;
780 # can't use 'rev-list --count' if we use --diff-filter
781 my $fh = $git->popen(qw(log --pretty=tformat:%h
782 --no-notes --no-color --no-renames
783 --diff-filter=AM), $range, '--', 'm');
784 ++$regen_max while <$fh>;
789 sub unindex_oid_remote {
790 my ($self, $oid, $mid) = @_;
791 $_->remote_remove($oid, $mid) foreach @{$self->{idx_parts}};
792 $self->{over}->remove_oid($oid, $mid);
796 my ($self, $git, $oid) = @_;
797 my $msgref = $git->cat_file($oid);
798 my $mime = PublicInbox::MIME->new($msgref);
799 my $mids = mids($mime->header_obj);
800 $mime = $msgref = undef;
801 my $over = $self->{over};
802 foreach my $mid (@$mids) {
805 while (my $smsg = $over->next_by_mid($mid, \$id, \$prev)) {
806 $gone{$smsg->{num}} = 1 if $oid eq $smsg->{blob};
809 my $n = scalar keys %gone;
812 warn "BUG: multiple articles linked to $oid\n",
813 join(',',sort keys %gone), "\n";
815 $self->{unindexed}->{$_}++ foreach keys %gone;
816 $self->unindex_oid_remote($oid, $mid);
820 my $x40 = qr/[a-f0-9]{40}/;
822 my ($self, $opts, $git, $unindex_range) = @_;
823 my $un = $self->{unindexed} ||= {}; # num => removal count
824 my $before = scalar keys %$un;
825 my @cmd = qw(log --raw -r
826 --no-notes --no-color --no-abbrev --no-renames);
827 my $fh = $self->{reindex_pipe} = $git->popen(@cmd, $unindex_range);
829 /\A:\d{6} 100644 $x40 ($x40) [AM]\tm$/o or next;
830 $self->unindex_oid($git, $1);
832 delete $self->{reindex_pipe};
835 return unless $opts->{prune};
836 my $after = scalar keys %$un;
837 return if $before == $after;
839 # ensure any blob can not longer be accessed via dumb HTTP
840 PublicInbox::Import::run_die(['git', "--git-dir=$git->{git_dir}",
841 qw(-c gc.reflogExpire=now gc --prune=all)]);
845 my ($self, $opts) = @_;
848 my $latest = git_dir_latest($self, \$epoch_max);
849 return unless defined $latest;
850 $self->idx_init; # acquire lock
851 my $mm_tmp = $self->{mm}->tmp_clone;
852 my $ranges = $opts->{reindex} ? [] : $self->last_commits($epoch_max);
854 my ($min, $max) = $mm_tmp->minmax;
855 my $regen = $self->index_prepare($opts, $epoch_max, $ranges);
856 $$regen += $max if $max;
858 my @cmd = qw(log --raw -r --pretty=tformat:%h
859 --no-notes --no-color --no-abbrev --no-renames);
861 # work backwards through history
862 my $last_commit = [];
863 for (my $i = $epoch_max; $i >= 0; $i--) {
864 my $git_dir = git_dir_n($self, $i);
865 die "already reindexing!\n" if delete $self->{reindex_pipe};
866 -d $git_dir or next; # missing parts are fine
867 my $git = PublicInbox::Git->new($git_dir);
868 my $unindex = delete $self->{"unindex-range.$i"};
869 $self->unindex($opts, $git, $unindex) if $unindex;
870 defined(my $range = $ranges->[$i]) or next;
871 my $fh = $self->{reindex_pipe} = $git->popen(@cmd, $range);
874 if (/\A$x40$/o && !defined($cmt)) {
876 } elsif (/\A:\d{6} 100644 $x40 ($x40) [AM]\tm$/o) {
877 $self->reindex_oid($mm_tmp, $D, $git, $1,
879 } elsif (/\A:\d{6} 100644 $x40 ($x40) [AM]\td$/o) {
880 $self->mark_deleted($D, $git, $1);
884 delete $self->{reindex_pipe};
885 $self->update_last_commit($git, $i, $cmt) if defined $cmt;
887 my @d = sort keys %$D;
889 warn "BUG: ", scalar(@d)," unseen deleted messages marked\n";
891 my ($mid, undef) = split(/\0/, $_, 2);