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 # Local storage (cache/memo) for lei(1), suitable for personal/private
5 # mail iff on encrypted device/FS. Based on v2, but only deduplicates
6 # git storage based on git OID (index deduplication is done in ContentHash)
8 # for xref3, the following are constant: $eidx_key = '.', $xnum = -1
10 # We rely on the synchronous IPC API for this in lei-daemon and
11 # multiple lei clients to write to it at once. This allows the
12 # lei/store IPC process to be decoupled from network latency in
14 package PublicInbox::LeiStore;
17 use parent qw(PublicInbox::Lock PublicInbox::IPC);
18 use PublicInbox::ExtSearchIdx;
20 use PublicInbox::Import;
21 use PublicInbox::InboxWritable qw(eml_from_path);
22 use PublicInbox::V2Writable;
23 use PublicInbox::ContentHash qw(content_hash);
24 use PublicInbox::MID qw(mids);
25 use PublicInbox::LeiSearch;
27 use PublicInbox::Spawn qw(spawn);
28 use PublicInbox::MdirReader;
29 use PublicInbox::LeiToMail;
32 use IO::Handle (); # ->autoflush
33 use Sys::Syslog qw(syslog openlog);
34 use Errno qw(EEXIST ENOENT);
37 my (undef, $dir, $opt) = @_;
38 my $eidx = PublicInbox::ExtSearchIdx->new($dir, $opt);
39 my $self = bless { priv_eidx => $eidx }, __PACKAGE__;
40 eidx_init($self)->done if $opt->{creat};
44 sub git { $_[0]->{priv_eidx}->git } # read-only
46 sub packing_factor { $PublicInbox::V2Writable::PACKING_FACTOR }
49 $_[0]->{rotate_bytes} // ((1024 * 1024 * 1024) / $_[0]->packing_factor)
55 open $rdr->{2}, '>', '/dev/null' or die "open /dev/null: $!";
56 chomp(my $i = $git->qx([qw(var GIT_COMMITTER_IDENT)], undef, $rdr));
57 $i =~ /\A(.+) <([^>]+)> [0-9]+ [-\+]?[0-9]+$/ and return ($1, $2);
58 my ($user, undef, undef, undef, undef, undef, $gecos) = getpwuid($<);
59 ($user) = (($user // $ENV{USER} // '') =~ /([\w\-\.\+]+)/);
61 ($gecos) = (($gecos // '') =~ /([\w\-\.\+ \t]+)/);
62 $gecos //= 'lei user';
63 require Sys::Hostname;
64 my ($host) = (Sys::Hostname::hostname() =~ /([\w\-\.]+)/);
65 $host //= 'localhost';
66 ($gecos, "$user\@$host")
74 return $im if $im->{bytes_added} < $self->rotate_bytes;
80 $max = $self->{priv_eidx}->{mg}->git_epochs + 1;
82 my (undef, $tl) = eidx_init($self); # acquire lock
83 $max //= $self->{priv_eidx}->{mg}->git_epochs;
85 my $latest = $self->{priv_eidx}->{mg}->add_epoch($max);
86 my $git = PublicInbox::Git->new($latest);
88 # re-acquire lock, update alternates for new epoch
89 (undef, $tl) = eidx_init($self);
90 my $packed_bytes = $git->packed_bytes;
91 my $unpacked_bytes = $packed_bytes / $self->packing_factor;
92 if ($unpacked_bytes >= $self->rotate_bytes) {
96 my ($n, $e) = git_ident($git);
97 $self->{im} = $im = PublicInbox::Import->new($git, $n, $e);
98 $im->{bytes_added} = int($packed_bytes / $self->packing_factor);
99 $im->{lock_path} = undef;
100 $im->{path_type} = 'v2';
106 PublicInbox::LeiSearch->new($_[0]->{priv_eidx}->{topdir});
109 # follows the stderr file
112 print { $self->{-err_wr} } readline($self->{-tmp_err});
117 my $eidx = $self->{priv_eidx};
118 my $tl = wantarray && $self->{-err_wr} ?
119 PublicInbox::OnDestroy->new($$, \&_tail_err, $self) :
121 $eidx->idx_init({-private => 1}); # acquires lock
122 wantarray ? ($eidx, $tl) : $eidx;
125 sub _docids_for ($$) {
126 my ($self, $eml) = @_;
128 my $eidx = $self->{priv_eidx};
129 my ($chash, $mids) = PublicInbox::LeiSearch::content_key($eml);
130 my $oidx = $eidx->{oidx};
131 my $im = $self->{im};
132 for my $mid (@$mids) {
134 while (my $cur = $oidx->next_by_mid($mid, \$id, \$prev)) {
135 next if $cur->{bytes} == 0; # external-only message
136 my $oid = $cur->{blob};
137 my $docid = $cur->{num};
138 my $bref = $im ? $im->cat_blob($oid) : undef;
139 $bref //= $eidx->git->cat_file($oid) //
140 _lms_rw($self)->local_blob($oid, 1) // do {
141 warn "W: $oid (#$docid) <$mid> not found\n";
144 local $self->{current_info} = $oid;
145 my $x = PublicInbox::Eml->new($bref);
146 $docids{$docid} = $docid if content_hash($x) eq $chash;
149 sort { $a <=> $b } values %docids;
152 # n.b. similar to LeiExportKw->export_kw_md, but this is for a single eml
153 sub export1_kw_md ($$$$$) {
154 my ($self, $mdir, $bn, $oidbin, $vmdish) = @_; # vmd/vmd_mod
156 my (@try, $unkn, $kw);
157 if ($bn =~ s/:2,([a-zA-Z]*)\z//) {
158 ($kw, $unkn) = PublicInbox::MdirReader::flags2kw($1);
159 if (my $set = $vmdish->{kw}) {
161 } elsif (my $add = $vmdish->{'+kw'}) {
163 } elsif (my $del = $vmdish->{-kw}) {
165 } # else no changes...
167 } else { # no keywords, yet, could be in new/
170 if (my $set = $vmdish->{kw}) {
172 } elsif (my $add = $vmdish->{'+kw'}) {
173 @$kw{@$add} = (); # auto-vivify
174 } else { # ignore $vmdish->{-kw}
178 $kw = [ keys %$kw ] if ref($kw) eq 'HASH';
179 $bn .= ':2,'. PublicInbox::LeiToMail::kw2suffix($kw, @$unkn);
180 return if $orig eq $bn; # no change
182 # we use link(2) + unlink(2) since rename(2) may
183 # inadvertently clobber if the "uniquefilename" part wasn't
185 my $dst = "$mdir/cur/$bn";
187 my $src = "$mdir/$d/$orig";
188 if (link($src, $dst)) {
189 if (!unlink($src) and $! != ENOENT) {
190 syslog('warning', "unlink($src): $!");
192 # TODO: verify oidbin?
193 $self->{lms}->mv_src("maildir:$mdir",
194 $oidbin, \$orig, $bn);
196 } elsif ($! == EEXIST) { # lost race with "lei export-kw"?
198 } elsif ($! != ENOENT) {
199 syslog('warning', "link($src -> $dst): $!");
202 for (@try) { return if -e "$mdir/$_/$orig" };
203 $self->{lms}->clear_src("maildir:$mdir", \$orig);
206 sub sto_export_kw ($$$) {
207 my ($self, $docid, $vmdish) = @_; # vmdish (vmd or vmd_mod)
208 my ($eidx, $tl) = eidx_init($self);
209 my $lms = _lms_rw($self) // return;
210 my $xr3 = $eidx->{oidx}->get_xref3($docid, 1);
211 for my $row (@$xr3) {
212 my (undef, undef, $oidbin) = @$row;
213 my $locs = $lms->locations_for($oidbin) // next;
214 while (my ($loc, $ids) = each %$locs) {
215 if ($loc =~ s!\Amaildir:!!i) {
217 export1_kw_md($self, $loc, $id,
226 # vmd = { kw => [ qw(seen ...) ], L => [ qw(inbox ...) ] }
228 my ($self, $eml, $vmd, $docids) = @_;
229 my ($eidx, $tl) = eidx_init($self);
230 $docids //= [ _docids_for($self, $eml) ];
231 for my $docid (@$docids) {
232 $eidx->idx_shard($docid)->ipc_do('set_vmd', $docid, $vmd);
233 sto_export_kw($self, $docid, $vmd);
239 my ($self, $eml, $vmd) = @_;
240 my ($eidx, $tl) = eidx_init($self);
241 my @docids = _docids_for($self, $eml);
242 for my $docid (@docids) {
243 $eidx->idx_shard($docid)->ipc_do('add_vmd', $docid, $vmd);
248 sub remove_eml_vmd { # remove just the VMD
249 my ($self, $eml, $vmd) = @_;
250 my ($eidx, $tl) = eidx_init($self);
251 my @docids = _docids_for($self, $eml);
252 for my $docid (@docids) {
253 $eidx->idx_shard($docid)->ipc_do('remove_vmd', $docid, $vmd);
258 sub _lms_rw ($) { # it is important to have eidx processes open before lms
260 my ($eidx, $tl) = eidx_init($self);
261 $self->{lms} //= do {
262 require PublicInbox::LeiMailSync;
263 my $f = "$self->{priv_eidx}->{topdir}/mail_sync.sqlite3";
264 my $lms = PublicInbox::LeiMailSync->new($f);
265 $lms->lms_write_prepare;
271 my ($self, $oidhex, $folder, $id) = @_;
272 _lms_rw($self)->set_src(pack('H*', $oidhex), $folder, $id);
275 sub _remove_if_local { # git->cat_async arg
276 my ($bref, $oidhex, $type, $size, $self) = @_;
277 $self->{im}->remove($bref) if $bref;
280 sub remove_docids ($;@) {
281 my ($self, @docids) = @_;
282 my $eidx = eidx_init($self);
283 for my $docid (@docids) {
284 $eidx->remove_doc($docid);
285 $eidx->{oidx}->{dbh}->do(<<EOF, undef, $docid);
286 DELETE FROM xref3 WHERE docid = ?
291 # remove the entire message from the index, does not touch mail_sync.sqlite3
293 my ($self, $eml) = @_;
294 my $im = $self->importer; # may create new epoch
295 my ($eidx, $tl) = eidx_init($self);
296 my $oidx = $eidx->{oidx};
297 my @docids = _docids_for($self, $eml);
298 my $git = $eidx->git;
299 for my $docid (@docids) {
300 my $xr3 = $oidx->get_xref3($docid, 1);
301 for my $row (@$xr3) {
302 my (undef, undef, $oidbin) = @$row;
303 my $oidhex = unpack('H*', $oidbin);
304 $git->cat_async($oidhex, \&_remove_if_local, $self);
307 $git->async_wait_all;
308 remove_docids($self, @docids);
313 my ($self, $oid) = @_;
314 my $eidx = eidx_init($self);
315 my ($docid, @cull) = $eidx->{oidx}->blob_exists($oid);
316 if (@cull) { # fixup old bugs...
318 W: $oid indexed as multiple docids: $docid @cull, culling to fixup old bugs
320 remove_docids($self, @cull);
325 sub _add_vmd ($$$$) {
326 my ($self, $idx, $docid, $vmd) = @_;
327 $idx->ipc_do('add_vmd', $docid, $vmd);
328 sto_export_kw($self, $docid, $vmd);
331 sub _docids_and_maybe_kw ($$) {
332 my ($self, $docids) = @_;
333 return $docids unless wantarray;
335 for my $num (@$docids) { # likely only 1, unless ContentHash changes
336 # can't use ->search->msg_keywords on uncommitted docs
337 my $idx = $self->{priv_eidx}->idx_shard($num);
338 my $tmp = eval { $idx->ipc_do('get_terms', 'K', $num) };
339 if ($@) { warn "#$num get_terms: $@" }
340 else { @$kw{keys %$tmp} = values(%$tmp) };
342 ($docids, [ sort keys %$kw ]);
346 my ($self, $eml, $vmd, $xoids) = @_;
347 my $im = $self->{-fake_im} // $self->importer; # may create new epoch
348 my ($eidx, $tl) = eidx_init($self);
349 my $oidx = $eidx->{oidx}; # PublicInbox::Import::add checks this
350 my $smsg = bless { -oidx => $oidx }, 'PublicInbox::Smsg';
351 $smsg->{-eidx_git} = $eidx->git if !$self->{-fake_im};
352 my $im_mark = $im->add($eml, undef, $smsg);
353 if ($vmd && $vmd->{sync_info}) {
354 set_sync_info($self, $smsg->{blob}, @{$vmd->{sync_info}});
356 unless ($im_mark) { # duplicate blob returns undef
357 return unless wantarray;
358 my @docids = $oidx->blob_exists($smsg->{blob});
359 return _docids_and_maybe_kw $self, \@docids;
362 local $self->{current_info} = $smsg->{blob};
363 my $vivify_xvmd = delete($smsg->{-vivify_xvmd}) // []; # exact matches
364 if ($xoids) { # fuzzy matches from externals in ale->xoids_for
365 delete $xoids->{$smsg->{blob}}; # added later
366 if (scalar keys %$xoids) {
367 my %docids = map { $_ => 1 } @$vivify_xvmd;
368 for my $oid (keys %$xoids) {
369 my $docid = oid2docid($self, $oid);
370 $docids{$docid} = $docid if defined($docid);
372 @$vivify_xvmd = sort { $a <=> $b } keys(%docids);
375 if (@$vivify_xvmd) { # docids list
377 $xoids->{$smsg->{blob}} = 1;
378 for my $docid (@$vivify_xvmd) {
379 my $cur = $oidx->get_art($docid);
380 my $idx = $eidx->idx_shard($docid);
381 if (!$cur || $cur->{bytes} == 0) { # really vivifying
382 $smsg->{num} = $docid;
383 $oidx->add_overview($eml, $smsg);
384 $smsg->{-merge_vmd} = 1;
385 $idx->index_eml($eml, $smsg);
386 } else { # lse fuzzy hit off ale
387 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
389 for my $oid (keys %$xoids) {
390 $oidx->add_xref3($docid, -1, $oid, '.');
392 _add_vmd($self, $idx, $docid, $vmd) if $vmd;
394 _docids_and_maybe_kw $self, $vivify_xvmd;
395 } elsif (my @docids = _docids_for($self, $eml)) {
396 # fuzzy match from within lei/store
397 for my $docid (@docids) {
398 my $idx = $eidx->idx_shard($docid);
399 $oidx->add_xref3($docid, -1, $smsg->{blob}, '.');
400 # add_eidx_info for List-Id
401 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
402 _add_vmd($self, $idx, $docid, $vmd) if $vmd;
404 _docids_and_maybe_kw $self, \@docids;
405 } else { # totally new message, no keywords
406 delete $smsg->{-oidx}; # for IPC-friendliness
407 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
408 $oidx->add_overview($eml, $smsg);
409 $oidx->add_xref3($smsg->{num}, -1, $smsg->{blob}, '.');
410 my $idx = $eidx->idx_shard($smsg->{num});
411 $idx->index_eml($eml, $smsg);
412 _add_vmd($self, $idx, $smsg->{num}, $vmd) if $vmd;
413 wantarray ? ($smsg, []) : $smsg;
418 my ($self, $eml, $vmd, $xoids) = @_;
419 add_eml($self, $eml, $vmd, $xoids) //
420 set_eml_vmd($self, $eml, $vmd);
424 my ($self, $eml, $vmd, $xoids) = @_;
425 require PublicInbox::FakeImport;
426 local $self->{-fake_im} = PublicInbox::FakeImport->new;
427 set_eml($self, $eml, $vmd, $xoids);
430 # store {kw} / {L} info for a message which is only in an external
431 sub _external_only ($$$) {
432 my ($self, $xoids, $eml) = @_;
433 my $eidx = $self->{priv_eidx};
434 my $oidx = $eidx->{oidx} // die 'BUG: {oidx} missing';
435 my $smsg = bless { blob => '' }, 'PublicInbox::Smsg';
436 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
437 # save space for an externals-only message
438 my $hdr = $eml->header_obj;
439 $smsg->populate($hdr); # sets lines == 0
441 delete @$smsg{qw(From Subject)};
442 $smsg->{to} = $smsg->{cc} = $smsg->{from} = '';
443 $oidx->add_overview($hdr, $smsg); # subject+references for threading
444 $smsg->{subject} = '';
445 for my $oid (keys %$xoids) {
446 $oidx->add_xref3($smsg->{num}, -1, $oid, '.');
448 my $idx = $eidx->idx_shard($smsg->{num});
449 $idx->index_eml(PublicInbox::Eml->new("\n\n"), $smsg);
454 my ($self, $xoids, $eml, $vmd_mod) = @_;
455 my ($eidx, $tl) = eidx_init($self);
456 my $oidx = $eidx->{oidx};
458 for my $oid (keys %$xoids) {
459 my $docid = oid2docid($self, $oid) // next;
460 delete $xoids->{$oid};
461 next if $seen{$docid}++;
462 my $idx = $eidx->idx_shard($docid);
463 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
464 sto_export_kw($self, $docid, $vmd_mod);
466 return unless scalar(keys(%$xoids));
468 # see if it was indexed, but with different OID(s)
469 if (my @docids = _docids_for($self, $eml)) {
470 for my $docid (@docids) {
471 next if $seen{$docid};
472 for my $oid (keys %$xoids) {
473 $oidx->add_xref3($docid, -1, $oid, '.');
475 my $idx = $eidx->idx_shard($docid);
476 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
477 sto_export_kw($self, $docid, $vmd_mod);
482 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
483 $idx->ipc_do('update_vmd', $smsg->{num}, $vmd_mod);
484 sto_export_kw($self, $smsg->{num}, $vmd_mod);
487 # set or update keywords for external message, called via ipc_do
489 my ($self, $xoids, $eml, $vmd) = @_;
491 my ($eidx, $tl) = eidx_init($self);
492 my $oidx = $eidx->{oidx};
495 # see if we can just update existing docs
496 for my $oid (keys %$xoids) {
497 my $docid = oid2docid($self, $oid) // next;
498 delete $xoids->{$oid}; # all done with this oid
499 next if $seen{$docid}++;
500 my $idx = $eidx->idx_shard($docid);
501 $idx->ipc_do('set_vmd', $docid, $vmd);
502 sto_export_kw($self, $docid, $vmd);
504 return unless scalar(keys(%$xoids));
506 # n.b. we don't do _docids_for here, we expect the caller
507 # already checked $lse->kw_changed before calling this sub
509 return unless (@{$vmd->{kw} // []}) || (@{$vmd->{L} // []});
511 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
512 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd);
513 sto_export_kw($self, $smsg->{num}, $vmd);
517 my ($self, $wait) = @_;
518 if (my $im = $self->{im}) {
519 $wait ? $im->barrier : $im->checkpoint;
522 $self->{priv_eidx}->checkpoint($wait);
527 _tail_err($self) if $self->{-err_wr};
528 my $dir = $self->{priv_eidx}->{topdir};
529 return unless -e $dir;
530 my $old = delete $self->{-tmp_err};
531 my $pfx = POSIX::strftime('%Y%m%d%H%M%S', gmtime(time));
532 my $err = File::Temp->new(TEMPLATE => "$pfx.$$.err-XXXX",
533 SUFFIX => '.err', DIR => $dir);
534 open STDERR, '>>', $err->filename or die "dup2: $!";
535 STDERR->autoflush(1); # shared with shard subprocesses
536 $self->{-tmp_err} = $err; # separate file description for RO access
541 my ($self, $sock_ref) = @_;
543 if (my $im = delete($self->{im})) {
546 $err .= "import done: $@\n";
551 $self->{priv_eidx}->done; # V2Writable::done
556 sub ipc_atfork_child {
558 my $lei = $self->{lei};
559 $lei->_lei_atfork_child(1) if $lei;
561 if (my $to_close = delete($self->{to_close})) {
562 close($_) for @$to_close;
564 openlog('lei/store', 'pid,nowait,nofatal,ndelay', 'user');
565 $self->SUPER::ipc_atfork_child;
569 my ($self, @args) = @_;
570 local $PublicInbox::DS::in_loop = 0; # waitpid synchronously
571 $self->SUPER::recv_and_run(@args);
575 my ($self, $lei) = @_;
576 $lei // die 'BUG: $lei not passed';
577 unless ($self->{-ipc_req}) {
578 my $dir = $lei->store_path;
579 substr($dir, -length('/lei/store'), 10, '');
580 pipe(my ($r, $w)) or die "pipe: $!";
582 # Mail we import into lei are private, so headers filtered out
583 # by -mda for public mail are not appropriate
584 local @PublicInbox::MDA::BAD_HEADERS = ();
585 $self->wq_workers_start("lei/store $dir", 1, $lei->oldset, {
590 $self->wq_wait_async; # outlives $lei
591 require PublicInbox::LeiStoreErr;
592 PublicInbox::LeiStoreErr->new($r, $lei);