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 git_sha);
24 use PublicInbox::MID qw(mids);
25 use PublicInbox::LeiSearch;
27 use PublicInbox::Spawn qw(spawn);
28 use List::Util qw(max);
33 my (undef, $dir, $opt) = @_;
34 my $eidx = PublicInbox::ExtSearchIdx->new($dir, $opt);
35 my $self = bless { priv_eidx => $eidx }, __PACKAGE__;
36 eidx_init($self)->done if $opt->{creat};
40 sub git { $_[0]->{priv_eidx}->git } # read-only
42 sub packing_factor { $PublicInbox::V2Writable::PACKING_FACTOR }
45 $_[0]->{rotate_bytes} // ((1024 * 1024 * 1024) / $_[0]->packing_factor)
48 sub git_pfx { "$_[0]->{priv_eidx}->{topdir}/local" };
52 if (opendir(my $dh, $self->git_pfx)) {
54 substr($_, 0, -4) + 0; # drop ".git" suffix
55 } grep(/\A[0-9]+\.git\z/, readdir($dh))) // 0;
57 $!{ENOENT} ? 0 : die("opendir ${\$self->git_pfx}: $!\n");
64 open $rdr->{2}, '>', '/dev/null' or die "open /dev/null: $!";
65 chomp(my $i = $git->qx([qw(var GIT_COMMITTER_IDENT)], undef, $rdr));
66 $i =~ /\A(.+) <([^>]+)> [0-9]+ [-\+]?[0-9]+$/ and return ($1, $2);
67 my ($user, undef, undef, undef, undef, undef, $gecos) = getpwuid($<);
68 ($user) = (($user // $ENV{USER} // '') =~ /([\w\-\.\+]+)/);
70 ($gecos) = (($gecos // '') =~ /([\w\-\.\+ \t]+)/);
71 $gecos //= 'lei user';
72 require Sys::Hostname;
73 my ($host) = (Sys::Hostname::hostname() =~ /([\w\-\.]+)/);
74 $host //= 'localhost';
75 ($gecos, "$user\@$host")
83 return $im if $im->{bytes_added} < $self->rotate_bytes;
89 $max = $self->git_epoch_max + 1;
91 my (undef, $tl) = eidx_init($self); # acquire lock
92 my $pfx = $self->git_pfx;
93 $max //= $self->git_epoch_max;
95 my $latest = "$pfx/$max.git";
97 PublicInbox::Import::init_bare($latest);
98 my $git = PublicInbox::Git->new($latest);
100 $git->qx(qw(config core.sharedRepository 0600));
101 $self->done; # unlock
102 # re-acquire lock, update alternates for new epoch
103 (undef, $tl) = eidx_init($self);
105 my $packed_bytes = $git->packed_bytes;
106 my $unpacked_bytes = $packed_bytes / $self->packing_factor;
107 if ($unpacked_bytes >= $self->rotate_bytes) {
111 my ($n, $e) = git_ident($git);
112 $self->{im} = $im = PublicInbox::Import->new($git, $n, $e);
113 $im->{bytes_added} = int($packed_bytes / $self->packing_factor);
114 $im->{lock_path} = undef;
115 $im->{path_type} = 'v2';
121 PublicInbox::LeiSearch->new($_[0]->{priv_eidx}->{topdir});
124 # follows the stderr file
127 print { $self->{-err_wr} } readline($self->{-tmp_err});
132 my $eidx = $self->{priv_eidx};
133 my $tl = wantarray && $self->{-err_wr} ?
134 PublicInbox::OnDestroy->new($$, \&_tail_err, $self) :
136 $eidx->idx_init({-private => 1}); # acquires lock
137 wantarray ? ($eidx, $tl) : $eidx;
140 sub _docids_for ($$) {
141 my ($self, $eml) = @_;
143 my $eidx = $self->{priv_eidx};
144 my ($chash, $mids) = PublicInbox::LeiSearch::content_key($eml);
145 my $oidx = $eidx->{oidx};
146 my $im = $self->{im};
147 for my $mid (@$mids) {
149 while (my $cur = $oidx->next_by_mid($mid, \$id, \$prev)) {
150 next if $cur->{bytes} == 0; # external-only message
151 my $oid = $cur->{blob};
152 my $docid = $cur->{num};
153 my $bref = $im ? $im->cat_blob($oid) : undef;
154 $bref //= $eidx->git->cat_file($oid) //
155 _lms_rw($self)->local_blob($oid, 1) // do {
156 warn "W: $oid (#$docid) <$mid> not found\n";
159 local $self->{current_info} = $oid;
160 my $x = PublicInbox::Eml->new($bref);
161 $docids{$docid} = $docid if content_hash($x) eq $chash;
164 sort { $a <=> $b } values %docids;
168 my ($self, $eml, $vmd, $docids) = @_;
169 my ($eidx, $tl) = eidx_init($self);
170 $docids //= [ _docids_for($self, $eml) ];
171 for my $docid (@$docids) {
172 $eidx->idx_shard($docid)->ipc_do('set_vmd', $docid, $vmd);
178 my ($self, $eml, $vmd) = @_;
179 my ($eidx, $tl) = eidx_init($self);
180 my @docids = _docids_for($self, $eml);
181 for my $docid (@docids) {
182 $eidx->idx_shard($docid)->ipc_do('add_vmd', $docid, $vmd);
187 sub remove_eml_vmd { # remove just the VMD
188 my ($self, $eml, $vmd) = @_;
189 my ($eidx, $tl) = eidx_init($self);
190 my @docids = _docids_for($self, $eml);
191 for my $docid (@docids) {
192 $eidx->idx_shard($docid)->ipc_do('remove_vmd', $docid, $vmd);
199 my ($eidx, $tl) = eidx_init($self);
200 $self->{lms} //= do {
201 require PublicInbox::LeiMailSync;
202 my $f = "$self->{priv_eidx}->{topdir}/mail_sync.sqlite3";
203 my $lms = PublicInbox::LeiMailSync->new($f);
210 my ($self, $folder, $id) = @_;
211 _lms_rw($self)->clear_src($folder, $id);
215 my ($self, $folder, $oidbin, $id, $newbn) = @_;
216 _lms_rw($self)->mv_src($folder, $oidbin, $id, $newbn);
219 sub lms_forget_folders {
220 my ($self, @folders) = @_;
221 my $lms = _lms_rw($self);
222 for my $f (@folders) { $lms->forget_folder($f) }
225 sub lms_rename_folder {
226 my ($self, $old, $new) = @_;
227 _lms_rw($self)->rename_folder($old, $new);
231 my ($self, $oidhex, $folder, $id) = @_;
232 _lms_rw($self)->set_src($oidhex, $folder, $id);
235 sub _remove_if_local { # git->cat_async arg
236 my ($bref, $oidhex, $type, $size, $self) = @_;
237 $self->{im}->remove($bref) if $bref;
240 sub remove_docids ($;@) {
241 my ($self, @docids) = @_;
242 my $eidx = eidx_init($self);
243 for my $docid (@docids) {
244 $eidx->idx_shard($docid)->ipc_do('xdb_remove', $docid);
245 $self->{oidx}->delete_by_num($docid);
246 $self->{oidx}->{dbh}->do(<<EOF, undef, $docid);
247 DELETE FROM xref3 WHERE docid = ?
252 # remove the entire message from the index, does not touch mail_sync.sqlite3
254 my ($self, $eml) = @_;
255 my $im = $self->importer; # may create new epoch
256 my ($eidx, $tl) = eidx_init($self);
257 my $oidx = $eidx->{oidx};
258 my @docids = _docids_for($self, $eml);
259 my $git = $eidx->git;
260 for my $docid (@docids) {
261 my $xr3 = $oidx->get_xref3($docid, 1);
262 for my $row (@$xr3) {
263 my (undef, undef, $oidbin) = @$row;
264 my $oidhex = unpack('H*', $oidbin);
265 $git->cat_async($oidhex, \&_remove_if_local, $self);
268 $git->cat_async_wait;
269 remove_docids($self, @docids);
274 my ($self, $oid) = @_;
275 my $eidx = eidx_init($self);
276 my ($docid, @cull) = $eidx->{oidx}->blob_exists($oid);
277 if (@cull) { # fixup old bugs...
279 W: $oid indexed as multiple docids: $docid @cull, culling to fixup old bugs
281 remove_docids($self, @cull);
287 my ($self, $eml, $vmd, $xoids) = @_;
288 my $im = $self->{-fake_im} // $self->importer; # may create new epoch
289 my ($eidx, $tl) = eidx_init($self);
290 my $oidx = $eidx->{oidx}; # PublicInbox::Import::add checks this
291 my $smsg = bless { -oidx => $oidx }, 'PublicInbox::Smsg';
292 $smsg->{-eidx_git} = $eidx->git if !$self->{-fake_im};
293 my $im_mark = $im->add($eml, undef, $smsg);
294 if ($vmd && $vmd->{sync_info}) {
295 set_sync_info($self, $smsg->{blob}, @{$vmd->{sync_info}});
297 $im_mark or return; # duplicate blob returns undef
299 local $self->{current_info} = $smsg->{blob};
300 my $vivify_xvmd = delete($smsg->{-vivify_xvmd}) // []; # exact matches
301 if ($xoids) { # fuzzy matches from externals in ale->xoids_for
302 delete $xoids->{$smsg->{blob}}; # added later
303 if (scalar keys %$xoids) {
304 my %docids = map { $_ => 1 } @$vivify_xvmd;
305 for my $oid (keys %$xoids) {
306 my $docid = oid2docid($self, $oid);
307 $docids{$docid} = $docid if defined($docid);
309 @$vivify_xvmd = sort { $a <=> $b } keys(%docids);
314 $xoids->{$smsg->{blob}} = 1;
315 for my $docid (@$vivify_xvmd) {
316 my $cur = $oidx->get_art($docid);
317 my $idx = $eidx->idx_shard($docid);
318 if (!$cur || $cur->{bytes} == 0) { # really vivifying
319 $smsg->{num} = $docid;
320 $oidx->add_overview($eml, $smsg);
321 $smsg->{-merge_vmd} = 1;
322 $idx->index_eml($eml, $smsg);
323 } else { # lse fuzzy hit off ale
324 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
326 for my $oid (keys %$xoids) {
327 $oidx->add_xref3($docid, -1, $oid, '.');
329 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
332 } elsif (my @docids = _docids_for($self, $eml)) {
333 # fuzzy match from within lei/store
334 for my $docid (@docids) {
335 my $idx = $eidx->idx_shard($docid);
336 $oidx->add_xref3($docid, -1, $smsg->{blob}, '.');
337 # add_eidx_info for List-Id
338 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
339 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
342 } else { # totally new message
343 delete $smsg->{-oidx}; # for IPC-friendliness
344 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
345 $oidx->add_overview($eml, $smsg);
346 $oidx->add_xref3($smsg->{num}, -1, $smsg->{blob}, '.');
347 my $idx = $eidx->idx_shard($smsg->{num});
348 $idx->index_eml($eml, $smsg);
349 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd) if $vmd;
355 my ($self, $eml, $vmd, $xoids) = @_;
356 add_eml($self, $eml, $vmd, $xoids) //
357 set_eml_vmd($self, $eml, $vmd);
361 my ($self, $eml, $vmd, $xoids) = @_;
362 require PublicInbox::FakeImport;
363 local $self->{-fake_im} = PublicInbox::FakeImport->new;
364 set_eml($self, $eml, $vmd, $xoids);
367 sub _external_only ($$$) {
368 my ($self, $xoids, $eml) = @_;
369 my $eidx = $self->{priv_eidx};
370 my $oidx = $eidx->{oidx} // die 'BUG: {oidx} missing';
371 my $smsg = bless { blob => '' }, 'PublicInbox::Smsg';
372 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
373 # save space for an externals-only message
374 my $hdr = $eml->header_obj;
375 $smsg->populate($hdr); # sets lines == 0
377 delete @$smsg{qw(From Subject)};
378 $smsg->{to} = $smsg->{cc} = $smsg->{from} = '';
379 $oidx->add_overview($hdr, $smsg); # subject+references for threading
380 $smsg->{subject} = '';
381 for my $oid (keys %$xoids) {
382 $oidx->add_xref3($smsg->{num}, -1, $oid, '.');
384 my $idx = $eidx->idx_shard($smsg->{num});
385 $idx->index_eml(PublicInbox::Eml->new("\n\n"), $smsg);
390 my ($self, $xoids, $eml, $vmd_mod) = @_;
391 my ($eidx, $tl) = eidx_init($self);
392 my $oidx = $eidx->{oidx};
394 for my $oid (keys %$xoids) {
395 my $docid = oid2docid($self, $oid) // next;
396 delete $xoids->{$oid};
397 next if $seen{$docid}++;
398 my $idx = $eidx->idx_shard($docid);
399 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
401 return unless scalar(keys(%$xoids));
403 # see if it was indexed, but with different OID(s)
404 if (my @docids = _docids_for($self, $eml)) {
405 for my $docid (@docids) {
406 next if $seen{$docid};
407 for my $oid (keys %$xoids) {
408 $oidx->add_xref3($docid, -1, $oid, '.');
410 my $idx = $eidx->idx_shard($docid);
411 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
416 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
417 $idx->ipc_do('update_vmd', $smsg->{num}, $vmd_mod);
420 # set or update keywords for external message, called via ipc_do
422 my ($self, $xoids, $eml, $vmd) = @_;
424 my ($eidx, $tl) = eidx_init($self);
425 my $oidx = $eidx->{oidx};
428 # see if we can just update existing docs
429 for my $oid (keys %$xoids) {
430 my $docid = oid2docid($self, $oid) // next;
431 delete $xoids->{$oid}; # all done with this oid
432 next if $seen{$docid}++;
433 my $idx = $eidx->idx_shard($docid);
434 $idx->ipc_do('set_vmd', $docid, $vmd);
436 return unless scalar(keys(%$xoids));
438 # n.b. we don't do _docids_for here, we expect the caller
439 # already checked $lse->kw_changed before calling this sub
441 return unless (@{$vmd->{kw} // []}) || (@{$vmd->{L} // []});
443 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
444 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd);
448 my ($self, $wait) = @_;
449 if (my $im = $self->{im}) {
450 $wait ? $im->barrier : $im->checkpoint;
452 if (my $lms = delete $self->{lms}) {
455 $self->{priv_eidx}->checkpoint($wait);
460 _tail_err($self) if $self->{-err_wr};
461 my $dir = $self->{priv_eidx}->{topdir};
462 return unless -e $dir;
463 my $old = delete $self->{-tmp_err};
464 my $pfx = POSIX::strftime('%Y%m%d%H%M%S', gmtime(time));
465 my $err = File::Temp->new(TEMPLATE => "$pfx.$$.lei_storeXXXX",
466 SUFFIX => '.err', DIR => $dir);
467 open STDERR, '>>', $err->filename or die "dup2: $!";
468 STDERR->autoflush(1); # shared with shard subprocesses
469 $self->{-tmp_err} = $err; # separate file description for RO access
474 my ($self, $sock_ref) = @_;
476 if (my $im = delete($self->{im})) {
479 $err .= "import done: $@\n";
483 if (my $lms = delete $self->{lms}) {
486 $self->{priv_eidx}->done; # V2Writable::done
490 # notify clients ->done has been issued
491 defined($sock_ref) and
492 $self->{s2d_op_p}->pkt_do('sto_done_complete', $sock_ref);
495 sub ipc_atfork_child {
497 my $lei = $self->{lei};
498 $lei->_lei_atfork_child(1) if $lei;
500 if (my $to_close = delete($self->{to_close})) {
501 close($_) for @$to_close;
503 $self->SUPER::ipc_atfork_child;
507 my ($self, $lei) = @_;
508 $lei // die 'BUG: $lei not passed';
509 unless ($self->{-ipc_req}) {
510 # s2d => store-to-daemon messages
511 require PublicInbox::PktOp;
512 my ($s2d_op_c, $s2d_op_p) = PublicInbox::PktOp->pair;
513 my $dir = $lei->store_path;
514 $self->ipc_lock_init("$dir/ipc.lock");
515 substr($dir, -length('/lei/store'), 10, '');
516 pipe(my ($r, $w)) or die "pipe: $!";
517 # Mail we import into lei are private, so headers filtered out
518 # by -mda for public mail are not appropriate
519 local @PublicInbox::MDA::BAD_HEADERS = ();
520 $self->ipc_worker_spawn("lei/store $dir", $lei->oldset, {
523 to_close => [ $r, $s2d_op_c->{sock} ],
524 s2d_op_p => $s2d_op_p,
526 require PublicInbox::LeiStoreErr;
527 PublicInbox::LeiStoreErr->new($r, $lei);
529 sto_done_complete => [ $lei->can('sto_done_complete') ]
535 # TODO: support SHA-256
536 sub git_blob_id { # called via LEI->git_blob_id
537 my ($self, $eml) = @_;
538 $eml->header_set($_) for @PublicInbox::Import::UNWANTED_HEADERS;
539 git_sha(1, $eml)->hexdigest;