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 content_ids ($) {
264 my @cids = ( content_id($mime) );
266 # Email::MIME->as_string doesn't always round-trip, so we may
267 # use a second content_id
268 my $rt = content_id(PublicInbox::MIME->new(\($mime->as_string)));
269 push @cids, $rt if $cids[0] ne $rt;
273 sub content_matches ($$) {
274 my ($cids, $existing) = @_;
275 my $cid = content_id($existing);
277 return 1 if $_ eq $cid
282 sub remove_internal {
283 my ($self, $mime, $cmt_msg, $purge) = @_;
285 my $im = $self->importer unless $purge;
286 my $over = $self->{over};
287 my $cids = content_ids($mime);
288 my $parts = $self->{idx_parts};
289 my $mm = $self->{mm};
291 my $mids = mids($mime->header_obj);
293 # We avoid introducing new blobs into git since the raw content
294 # can be slightly different, so we do not need the user-supplied
295 # message now that we have the mids and content_id
299 foreach my $mid (@$mids) {
302 while (my $smsg = $over->next_by_mid($mid, \$id, \$prev)) {
303 my $msg = get_blob($self, $smsg);
304 if (!defined($msg)) {
305 warn "broken smsg for $mid\n";
309 my $cur = PublicInbox::MIME->new($msg);
310 if (content_matches($cids, $cur)) {
311 $smsg->{mime} = $cur;
312 $gone{$smsg->{num}} = [ $smsg, \$orig ];
315 my $n = scalar keys %gone;
318 warn "BUG: multiple articles linked to <$mid>\n",
319 join(',', sort keys %gone), "\n";
321 foreach my $num (keys %gone) {
322 my ($smsg, $orig) = @{$gone{$num}};
323 $mm->num_delete($num);
324 # $removed should only be set once assuming
325 # no bugs in our deduplication code:
327 my $oid = $smsg->{blob};
331 ($mark, undef) = $im->remove($orig, $cmt_msg);
334 $self->unindex_oid_remote($oid, $mid);
339 my $cmt = $im->get_mark($mark);
340 $self->{last_commit}->[$self->{epoch_max}] = $cmt;
342 if ($purge && scalar keys %$purge) {
343 return purge_oids($self, $purge);
349 my ($self, $mime, $cmt_msg) = @_;
350 remove_internal($self, $mime, $cmt_msg);
354 my ($self, $mime) = @_;
355 my $purges = remove_internal($self, $mime, undef, {});
356 $self->idx_init if @$purges; # ->done is called on purges
357 for my $i (0..$#$purges) {
358 defined(my $cmt = $purges->[$i]) or next;
359 $self->{last_commit}->[$i] = $cmt;
364 sub last_commit_part ($$;$) {
365 my ($self, $i, $cmt) = @_;
366 my $v = PublicInbox::Search::SCHEMA_VERSION();
367 $self->{mm}->last_commit_xap($v, $i, $cmt);
370 sub set_last_commits ($) {
372 defined(my $epoch_max = $self->{epoch_max}) or return;
373 my $last_commit = $self->{last_commit};
374 foreach my $i (0..$epoch_max) {
375 defined(my $cmt = $last_commit->[$i]) or next;
376 $last_commit->[$i] = undef;
377 last_commit_part($self, $i, $cmt);
383 $self->{bnote} or return;
385 my $barrier = { map { $_ => 1 } (0..$n) };
389 my ($self, $barrier) = @_;
390 my $bnote = $self->{bnote} or return;
392 while (scalar keys %$barrier) {
393 defined(my $l = $r->getline) or die "EOF on barrier_wait: $!";
394 $l =~ /\Abarrier (\d+)/ or die "bad line on barrier_wait: $l";
395 delete $barrier->{$1} or die "bad part[$1] on barrier wait";
399 sub checkpoint ($;$) {
400 my ($self, $wait) = @_;
402 if (my $im = $self->{im}) {
409 my $parts = $self->{idx_parts};
411 my $dbh = $self->{mm}->{dbh};
413 # SQLite msgmap data is second in importance
416 # SQLite overview is third
417 $self->{over}->commit_lazy;
419 # Now deal with Xapian
421 my $barrier = $self->barrier_init(scalar @$parts);
423 # each partition needs to issue a barrier command
424 $_->remote_barrier for @$parts;
426 # wait for each Xapian partition
427 $self->barrier_wait($barrier);
429 $_->remote_commit for @$parts;
432 # last_commit is special, don't commit these until
433 # remote partitions are done:
435 set_last_commits($self);
440 $self->{transact_bytes} = 0;
443 # issue a write barrier to ensure all data is visible to other processes
444 # and read-only ops. Order of data importance is: git > SQLite > Xapian
445 sub barrier { checkpoint($_[0], 1) };
449 my $im = delete $self->{im};
450 $im->done if $im; # PublicInbox::Import::done
452 my $mm = delete $self->{mm};
453 $mm->{dbh}->commit if $mm;
454 my $parts = delete $self->{idx_parts};
456 $_->remote_close for @$parts;
458 $self->{over}->disconnect;
459 delete $self->{bnote};
460 $self->{transact_bytes} = 0;
461 $self->lock_release if $parts;
465 my ($self, $epoch) = @_;
466 my $pfx = "$self->{-inbox}->{mainrepo}/git";
467 my $git_dir = "$pfx/$epoch.git";
468 my @cmd = (qw(git init --bare -q), $git_dir);
469 PublicInbox::Import::run_die(\@cmd);
471 my $all = "$self->{-inbox}->{mainrepo}/all.git";
473 @cmd = (qw(git init --bare -q), $all);
474 PublicInbox::Import::run_die(\@cmd);
475 @cmd = (qw/git config/, "--file=$all/config",
476 'repack.writeBitmaps', 'true');
477 PublicInbox::Import::run_die(\@cmd);
480 @cmd = (qw/git config/, "--file=$git_dir/config",
481 'include.path', '../../all.git/config');
482 PublicInbox::Import::run_die(\@cmd);
484 my $alt = "$all/objects/info/alternates";
485 my $new_obj_dir = "../../git/$epoch.git/objects";
488 open(my $fh, '<', $alt) or die "open < $alt: $!\n";
489 %alts = map { chomp; $_ => 1 } (<$fh>);
491 return $git_dir if $alts{$new_obj_dir};
492 open my $fh, '>>', $alt or die "open >> $alt: $!\n";
493 print $fh "$new_obj_dir\n" or die "print >> $alt: $!\n";
494 close $fh or die "close $alt: $!\n";
499 my ($self, $max) = @_;
501 my $pfx = "$self->{-inbox}->{mainrepo}/git";
502 return unless -d $pfx;
504 opendir my $dh, $pfx or die "opendir $pfx: $!\n";
505 while (defined(my $git_dir = readdir($dh))) {
506 $git_dir =~ m!\A(\d+)\.git\z! or next;
509 $latest = "$pfx/$git_dir";
517 my $im = $self->{im};
519 if ($im->{bytes_added} < $self->{rotate_bytes}) {
526 my $git_dir = $self->git_init(++$self->{epoch_max});
527 my $git = PublicInbox::Git->new($git_dir);
528 return $self->import_init($git, 0);
533 my $latest = git_dir_latest($self, \$max);
534 if (defined $latest) {
535 my $git = PublicInbox::Git->new($latest);
536 my $packed_bytes = $git->packed_bytes;
537 if ($packed_bytes >= $self->{rotate_bytes}) {
540 $self->{epoch_max} = $max;
541 return $self->import_init($git, $packed_bytes);
544 $self->{epoch_max} = $epoch;
545 $latest = $self->git_init($epoch);
546 $self->import_init(PublicInbox::Git->new($latest), 0);
550 my ($self, $git, $packed_bytes, $tmp) = @_;
551 my $im = PublicInbox::Import->new($git, undef, undef, $self->{-inbox});
552 $im->{bytes_added} = int($packed_bytes / $PACKING_FACTOR);
553 $im->{want_object_info} = 1;
554 $im->{lock_path} = undef;
555 $im->{path_type} = 'v2';
556 $self->{im} = $im unless $tmp;
562 my ($mid, $cur, $new) = @_;
563 use File::Temp qw(tempfile);
564 use PublicInbox::Spawn qw(spawn);
566 my ($ah, $an) = tempfile('email-cur-XXXXXXXX', TMPDIR => 1);
567 print $ah $cur->as_string or die "print: $!";
568 close $ah or die "close: $!";
569 my ($bh, $bn) = tempfile('email-new-XXXXXXXX', TMPDIR => 1);
570 PublicInbox::Import::drop_unwanted_headers($new);
571 print $bh $new->as_string or die "print: $!";
572 close $bh or die "close: $!";
573 my $cmd = [ qw(diff -u), $an, $bn ];
574 print STDERR "# MID conflict <$mid>\n";
575 my $pid = spawn($cmd, undef, { 1 => 2 });
576 defined $pid or die "diff failed to spawn $!";
577 waitpid($pid, 0) == $pid or die "diff did not finish";
582 my ($self, $smsg) = @_;
583 if (my $im = $self->{im}) {
584 my $msg = $im->cat_blob($smsg->{blob});
587 # older message, should be in alternates
588 my $ibx = $self->{-inbox};
589 $ibx->msg_by_smsg($smsg);
593 my ($self, $mime, $mid) = @_;
594 my $over = $self->{over};
595 my $cids = content_ids($mime);
597 while (my $smsg = $over->next_by_mid($mid, \$id, \$prev)) {
598 my $msg = get_blob($self, $smsg);
599 if (!defined($msg)) {
600 warn "broken smsg for $mid\n";
603 my $cur = PublicInbox::MIME->new($msg);
604 if (content_matches($cids, $cur)) {
605 $smsg->{mime} = $cur;
610 # XXX DEBUG_DIFF is experimental and may be removed
611 diff($mid, $cur, $mime) if $ENV{DEBUG_DIFF};
618 my $fh = delete $self->{reindex_pipe};
620 if (my $parts = $self->{idx_parts}) {
621 $_->atfork_child foreach @$parts;
623 if (my $im = $self->{im}) {
626 die "unexpected mm" if $self->{mm};
627 close $self->{bnote}->[0] or die "close bnote[0]: $!\n";
632 my ($self, $D, $git, $oid) = @_;
633 my $msgref = $git->cat_file($oid);
634 my $mime = PublicInbox::MIME->new($$msgref);
635 my $mids = mids($mime->header_obj);
636 my $cid = content_id($mime);
637 foreach my $mid (@$mids) {
638 $D->{"$mid\0$cid"} = 1;
643 my ($self, $mm_tmp, $D, $git, $oid, $regen) = @_;
645 my $msgref = $git->cat_file($oid, \$len);
646 my $mime = PublicInbox::MIME->new($$msgref);
647 my $mids = mids($mime->header_obj);
648 my $cid = content_id($mime);
650 # get the NNTP article number we used before, highest number wins
651 # and gets deleted from mm_tmp;
655 foreach my $mid (@$mids) {
656 $del += (delete $D->{"$mid\0$cid"} || 0);
657 my $n = $mm_tmp->num_for($mid);
658 if (defined $n && $n > $num) {
663 if (!defined($mid0) && $regen && !$del) {
665 die "BUG: ran out of article numbers\n" if $num <= 0;
666 my $mm = $self->{mm};
667 foreach my $mid (reverse @$mids) {
668 if ($mm->mid_set($num, $mid) == 1) {
673 if (!defined($mid0)) {
674 my $id = '<' . join('> <', @$mids) . '>';
675 warn "Message-ID $id unusable for $num\n";
676 foreach my $mid (@$mids) {
677 defined(my $n = $mm->num_for($mid)) or next;
678 warn "#$n previously mapped for <$mid>\n";
683 if (!defined($mid0) || $del) {
684 if (!defined($mid0) && $del) { # expected for deletes
689 my $id = '<' . join('> <', @$mids) . '>';
691 warn "Skipping $id, no article number found\n";
692 if ($del && defined($mid0)) {
693 warn "$id was deleted $del " .
694 "time(s) but mapped to article #$num\n";
699 $mm_tmp->mid_delete($mid0) or
700 die "failed to delete <$mid0> for article #$num\n";
702 $self->{over}->add_overview($mime, $len, $num, $oid, $mid0);
703 my $nparts = $self->{partitions};
704 my $part = $num % $nparts;
705 my $idx = $self->idx_part($part);
706 $idx->index_raw($len, $msgref, $num, $oid, $mid0, $mime);
707 my $n = $self->{transact_bytes} += $len;
708 if ($n > (PublicInbox::SearchIdx::BATCH_BYTES * $nparts)) {
710 $mm_tmp->atfork_prepare;
711 $self->done; # release lock
712 # allow -watch or -mda to write...
713 $self->idx_init; # reacquire lock
714 $mm_tmp->atfork_parent;
718 # only update last_commit for $i on reindex iff newer than current
719 sub update_last_commit {
720 my ($self, $git, $i, $cmt) = @_;
721 my $last = last_commit_part($self, $i);
722 if (defined $last && is_ancestor($git, $last, $cmt)) {
723 my @cmd = (qw(rev-list --count), "$last..$cmt");
724 chomp(my $n = $git->qx(@cmd));
725 return if $n ne '' && $n == 0;
727 last_commit_part($self, $i, $cmt);
730 sub git_dir_n ($$) { "$_[0]->{-inbox}->{mainrepo}/git/$_[1].git" }
733 my ($self, $epoch_max) = @_;
735 for (my $i = $epoch_max; $i >= 0; $i--) {
736 $heads->[$i] = last_commit_part($self, $i);
741 sub is_ancestor ($$$) {
742 my ($git, $cur, $tip) = @_;
743 return 0 unless $git->check($cur);
744 my $cmd = [ 'git', "--git-dir=$git->{git_dir}",
745 qw(merge-base --is-ancestor), $cur, $tip ];
746 my $pid = spawn($cmd);
747 defined $pid or die "spawning ".join(' ', @$cmd)." failed: $!";
748 waitpid($pid, 0) == $pid or die join(' ', @$cmd) .' did not finish';
753 my ($self, $opts, $epoch_max, $ranges) = @_;
755 my $head = $self->{-inbox}->{ref_head} || 'refs/heads/master';
756 for (my $i = $epoch_max; $i >= 0; $i--) {
757 die "already indexing!\n" if $self->{index_pipe};
758 my $git_dir = git_dir_n($self, $i);
759 -d $git_dir or next; # missing parts are fine
760 my $git = PublicInbox::Git->new($git_dir);
761 chomp(my $tip = $git->qx('rev-parse', $head));
763 if (defined(my $cur = $ranges->[$i])) {
764 $range = "$cur..$tip";
765 if (is_ancestor($git, $cur, $tip)) { # common case
766 my $n = $git->qx(qw(rev-list --count), $range);
769 $ranges->[$i] = undef;
774 discontiguous range: $range
775 Rewritten history? (in $git_dir)
777 my $base = $git->qx('merge-base', $tip, $cur);
780 $range = "$base..$tip";
781 warn "found merge-base: $base\n"
785 discarding history at $cur
789 reindexing $git_dir starting at
792 $self->{"unindex-range.$i"} = "$base..$cur";
795 $range = $tip; # all of it
797 $ranges->[$i] = $range;
799 # can't use 'rev-list --count' if we use --diff-filter
800 my $fh = $git->popen(qw(log --pretty=tformat:%h
801 --no-notes --no-color --no-renames
802 --diff-filter=AM), $range, '--', 'm');
803 ++$regen_max while <$fh>;
808 sub unindex_oid_remote {
809 my ($self, $oid, $mid) = @_;
810 $_->remote_remove($oid, $mid) foreach @{$self->{idx_parts}};
811 $self->{over}->remove_oid($oid, $mid);
815 my ($self, $git, $oid) = @_;
816 my $msgref = $git->cat_file($oid);
817 my $mime = PublicInbox::MIME->new($msgref);
818 my $mids = mids($mime->header_obj);
819 $mime = $msgref = undef;
820 my $over = $self->{over};
821 foreach my $mid (@$mids) {
824 while (my $smsg = $over->next_by_mid($mid, \$id, \$prev)) {
825 $gone{$smsg->{num}} = 1 if $oid eq $smsg->{blob};
828 my $n = scalar keys %gone;
831 warn "BUG: multiple articles linked to $oid\n",
832 join(',',sort keys %gone), "\n";
834 $self->{unindexed}->{$_}++ foreach keys %gone;
835 $self->unindex_oid_remote($oid, $mid);
839 my $x40 = qr/[a-f0-9]{40}/;
841 my ($self, $opts, $git, $unindex_range) = @_;
842 my $un = $self->{unindexed} ||= {}; # num => removal count
843 my $before = scalar keys %$un;
844 my @cmd = qw(log --raw -r
845 --no-notes --no-color --no-abbrev --no-renames);
846 my $fh = $self->{reindex_pipe} = $git->popen(@cmd, $unindex_range);
848 /\A:\d{6} 100644 $x40 ($x40) [AM]\tm$/o or next;
849 $self->unindex_oid($git, $1);
851 delete $self->{reindex_pipe};
854 return unless $opts->{prune};
855 my $after = scalar keys %$un;
856 return if $before == $after;
858 # ensure any blob can not longer be accessed via dumb HTTP
859 PublicInbox::Import::run_die(['git', "--git-dir=$git->{git_dir}",
860 qw(-c gc.reflogExpire=now gc --prune=all)]);
864 my ($self, $opts) = @_;
867 my $latest = git_dir_latest($self, \$epoch_max);
868 return unless defined $latest;
869 $self->idx_init; # acquire lock
870 my $mm_tmp = $self->{mm}->tmp_clone;
871 my $ranges = $opts->{reindex} ? [] : $self->last_commits($epoch_max);
873 my ($min, $max) = $mm_tmp->minmax;
874 my $regen = $self->index_prepare($opts, $epoch_max, $ranges);
875 $$regen += $max if $max;
877 my @cmd = qw(log --raw -r --pretty=tformat:%h
878 --no-notes --no-color --no-abbrev --no-renames);
880 # work backwards through history
881 my $last_commit = [];
882 for (my $i = $epoch_max; $i >= 0; $i--) {
883 my $git_dir = git_dir_n($self, $i);
884 die "already reindexing!\n" if delete $self->{reindex_pipe};
885 -d $git_dir or next; # missing parts are fine
886 my $git = PublicInbox::Git->new($git_dir);
887 my $unindex = delete $self->{"unindex-range.$i"};
888 $self->unindex($opts, $git, $unindex) if $unindex;
889 defined(my $range = $ranges->[$i]) or next;
890 my $fh = $self->{reindex_pipe} = $git->popen(@cmd, $range);
893 if (/\A$x40$/o && !defined($cmt)) {
895 } elsif (/\A:\d{6} 100644 $x40 ($x40) [AM]\tm$/o) {
896 $self->reindex_oid($mm_tmp, $D, $git, $1,
898 } elsif (/\A:\d{6} 100644 $x40 ($x40) [AM]\td$/o) {
899 $self->mark_deleted($D, $git, $1);
903 delete $self->{reindex_pipe};
904 $self->update_last_commit($git, $i, $cmt) if defined $cmt;
906 my @d = sort keys %$D;
908 warn "BUG: ", scalar(@d)," unseen deleted messages marked\n";
910 my ($mid, undef) = split(/\0/, $_, 2);