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 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 $pfx = $self->git_pfx;
92 $max //= $self->git_epoch_max;
94 my $latest = "$pfx/$max.git";
96 PublicInbox::Import::init_bare($latest);
97 my $git = PublicInbox::Git->new($latest);
99 $git->qx(qw(config core.sharedRepository 0600));
100 $self->done; # force eidx_init on next round
102 my $packed_bytes = $git->packed_bytes;
103 my $unpacked_bytes = $packed_bytes / $self->packing_factor;
104 if ($unpacked_bytes >= $self->rotate_bytes) {
108 my ($n, $e) = git_ident($git);
109 $self->{im} = $im = PublicInbox::Import->new($git, $n, $e);
110 $im->{bytes_added} = int($packed_bytes / $self->packing_factor);
111 $im->{lock_path} = undef;
112 $im->{path_type} = 'v2';
118 PublicInbox::LeiSearch->new($_[0]->{priv_eidx}->{topdir});
121 # follows the stderr file
124 print { $self->{-err_wr} } readline($self->{-tmp_err});
129 my $eidx = $self->{priv_eidx};
130 my $tl = wantarray && $self->{-err_wr} ?
131 PublicInbox::OnDestroy->new($$, \&_tail_err, $self) :
133 $eidx->idx_init({-private => 1});
134 wantarray ? ($eidx, $tl) : $eidx;
137 sub _docids_for ($$) {
138 my ($self, $eml) = @_;
140 my $eidx = $self->{priv_eidx};
141 my ($chash, $mids) = PublicInbox::LeiSearch::content_key($eml);
142 my $oidx = $eidx->{oidx};
143 my $im = $self->{im};
144 for my $mid (@$mids) {
146 while (my $cur = $oidx->next_by_mid($mid, \$id, \$prev)) {
147 next if $cur->{bytes} == 0; # external-only message
148 my $oid = $cur->{blob};
149 my $docid = $cur->{num};
150 my $bref = $im ? $im->cat_blob($oid) : undef;
151 $bref //= $eidx->git->cat_file($oid) // do {
152 warn "W: $oid (#$docid) <$mid> not found\n";
155 local $self->{current_info} = $oid;
156 my $x = PublicInbox::Eml->new($bref);
157 $docids{$docid} = $docid if content_hash($x) eq $chash;
160 sort { $a <=> $b } values %docids;
164 my ($self, $eml, $vmd, $docids) = @_;
165 my ($eidx, $tl) = eidx_init($self);
166 $docids //= [ _docids_for($self, $eml) ];
167 for my $docid (@$docids) {
168 $eidx->idx_shard($docid)->ipc_do('set_vmd', $docid, $vmd);
174 my ($self, $eml, $vmd) = @_;
175 my ($eidx, $tl) = eidx_init($self);
176 my @docids = _docids_for($self, $eml);
177 for my $docid (@docids) {
178 $eidx->idx_shard($docid)->ipc_do('add_vmd', $docid, $vmd);
184 my ($self, $eml, $vmd) = @_;
185 my ($eidx, $tl) = eidx_init($self);
186 my @docids = _docids_for($self, $eml);
187 for my $docid (@docids) {
188 $eidx->idx_shard($docid)->ipc_do('remove_vmd', $docid, $vmd);
193 sub set_sync_info ($$$) {
194 my ($self, $oidhex, $sync_info) = @_;
195 ($self->{lms} //= do {
196 require PublicInbox::LeiMailSync;
197 my $f = "$self->{priv_eidx}->{topdir}/mail_sync.sqlite3";
198 my $lms = PublicInbox::LeiMailSync->new($f);
201 })->set_src($oidhex, @$sync_info);
205 my ($self, $eml, $vmd, $xoids) = @_;
206 my $im = $self->importer; # may create new epoch
207 my ($eidx, $tl) = eidx_init($self); # updates/writes alternates file
208 my $oidx = $eidx->{oidx}; # PublicInbox::Import::add checks this
209 my $smsg = bless { -oidx => $oidx }, 'PublicInbox::Smsg';
210 my $im_mark = $im->add($eml, undef, $smsg);
211 if ($vmd && $vmd->{sync_info}) {
212 set_sync_info($self, $smsg->{blob}, $vmd->{sync_info});
214 $im_mark or return; # duplicate blob returns undef
216 local $self->{current_info} = $smsg->{blob};
217 my $vivify_xvmd = delete($smsg->{-vivify_xvmd}) // []; # exact matches
218 if ($xoids) { # fuzzy matches from externals in ale->xoids_for
219 delete $xoids->{$smsg->{blob}}; # added later
220 if (scalar keys %$xoids) {
221 my %docids = map { $_ => 1 } @$vivify_xvmd;
222 for my $oid (keys %$xoids) {
223 my @id = $oidx->blob_exists($oid);
226 @$vivify_xvmd = sort { $a <=> $b } keys(%docids);
231 $xoids->{$smsg->{blob}} = 1;
232 for my $docid (@$vivify_xvmd) {
233 my $cur = $oidx->get_art($docid);
234 my $idx = $eidx->idx_shard($docid);
235 if (!$cur || $cur->{bytes} == 0) { # really vivifying
236 $smsg->{num} = $docid;
237 $oidx->add_overview($eml, $smsg);
238 $smsg->{-merge_vmd} = 1;
239 $idx->index_eml($eml, $smsg);
240 } else { # lse fuzzy hit off ale
241 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
243 for my $oid (keys %$xoids) {
244 $oidx->add_xref3($docid, -1, $oid, '.');
246 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
249 } elsif (my @docids = _docids_for($self, $eml)) {
250 # fuzzy match from within lei/store
251 for my $docid (@docids) {
252 my $idx = $eidx->idx_shard($docid);
253 $oidx->add_xref3($docid, -1, $smsg->{blob}, '.');
254 # add_eidx_info for List-Id
255 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
256 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
259 } else { # totally new message
260 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
261 $oidx->add_overview($eml, $smsg);
262 $oidx->add_xref3($smsg->{num}, -1, $smsg->{blob}, '.');
263 my $idx = $eidx->idx_shard($smsg->{num});
264 $idx->index_eml($eml, $smsg);
265 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd) if $vmd;
271 my ($self, $eml, $vmd, $xoids) = @_;
272 add_eml($self, $eml, $vmd, $xoids) //
273 set_eml_vmd($self, $eml, $vmd);
276 sub _external_only ($$$) {
277 my ($self, $xoids, $eml) = @_;
278 my $eidx = $self->{priv_eidx};
279 my $oidx = $eidx->{oidx} // die 'BUG: {oidx} missing';
280 my $smsg = bless { blob => '' }, 'PublicInbox::Smsg';
281 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
282 # save space for an externals-only message
283 my $hdr = $eml->header_obj;
284 $smsg->populate($hdr); # sets lines == 0
286 delete @$smsg{qw(From Subject)};
287 $smsg->{to} = $smsg->{cc} = $smsg->{from} = '';
288 $oidx->add_overview($hdr, $smsg); # subject+references for threading
289 $smsg->{subject} = '';
290 for my $oid (keys %$xoids) {
291 $oidx->add_xref3($smsg->{num}, -1, $oid, '.');
293 my $idx = $eidx->idx_shard($smsg->{num});
294 $idx->index_eml(PublicInbox::Eml->new("\n\n"), $smsg);
299 my ($self, $xoids, $eml, $vmd_mod) = @_;
300 my ($eidx, $tl) = eidx_init($self);
301 my $oidx = $eidx->{oidx};
303 for my $oid (keys %$xoids) {
304 my @docids = $oidx->blob_exists($oid) or next;
305 scalar(@docids) > 1 and
306 warn "W: $oid indexed as multiple docids: @docids\n";
307 for my $docid (@docids) {
308 next if $seen{$docid}++;
309 my $idx = $eidx->idx_shard($docid);
310 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
312 delete $xoids->{$oid};
314 return unless scalar(keys(%$xoids));
316 # see if it was indexed, but with different OID(s)
317 if (my @docids = _docids_for($self, $eml)) {
318 for my $docid (@docids) {
319 next if $seen{$docid};
320 for my $oid (keys %$xoids) {
321 $oidx->add_xref3($docid, -1, $oid, '.');
323 my $idx = $eidx->idx_shard($docid);
324 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
329 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
330 $idx->ipc_do('update_vmd', $smsg->{num}, $vmd_mod);
333 # set or update keywords for external message, called via ipc_do
335 my ($self, $xoids, $eml, $vmd) = @_;
337 my ($eidx, $tl) = eidx_init($self);
338 my $oidx = $eidx->{oidx};
341 # see if we can just update existing docs
342 for my $oid (keys %$xoids) {
343 my @docids = $oidx->blob_exists($oid) or next;
344 scalar(@docids) > 1 and
345 warn "W: $oid indexed as multiple docids: @docids\n";
346 for my $docid (@docids) {
347 next if $seen{$docid}++;
348 my $idx = $eidx->idx_shard($docid);
349 $idx->ipc_do('set_vmd', $docid, $vmd);
351 delete $xoids->{$oid}; # all done with this oid
353 return unless scalar(keys(%$xoids));
355 # n.b. we don't do _docids_for here, we expect the caller
356 # already checked $lse->kw_changed before calling this sub
358 return unless (@{$vmd->{kw} // []}) || (@{$vmd->{L} // []});
360 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
361 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd);
365 my ($self, $wait) = @_;
366 if (my $im = $self->{im}) {
367 $wait ? $im->barrier : $im->checkpoint;
369 $self->{priv_eidx}->checkpoint($wait);
374 _tail_err($self) if $self->{-err_wr};
375 my $dir = $self->{priv_eidx}->{topdir};
376 return unless -e $dir;
377 my $old = delete $self->{-tmp_err};
378 my $pfx = POSIX::strftime('%Y%m%d%H%M%S', gmtime(time));
379 my $err = File::Temp->new(TEMPLATE => "$pfx.$$.lei_storeXXXX",
380 SUFFIX => '.err', DIR => $dir);
381 open STDERR, '>>', $err->filename or die "dup2: $!";
382 STDERR->autoflush(1); # shared with shard subprocesses
383 $self->{-tmp_err} = $err; # separate file description for RO access
390 if (my $im = delete($self->{im})) {
393 $err .= "import done: $@\n";
397 if (my $lms = delete $self->{lms}) {
400 $self->{priv_eidx}->done; # V2Writable::done
405 sub ipc_atfork_child {
407 my $lei = $self->{lei};
408 $lei->_lei_atfork_child(1) if $lei;
410 if (my $err = delete($self->{err_pipe})) {
412 $self->{-err_wr} = $err->[1];
414 $self->SUPER::ipc_atfork_child;
418 my ($self, $lei) = @_;
419 unless ($self->{-ipc_req}) {
420 my $d = $lei->store_path;
421 $self->ipc_lock_init("$d/ipc.lock");
422 substr($d, -length('/lei/store'), 10, '');
424 unless ($lei->{oneshot}) {
425 pipe(my ($r, $w)) or die "pipe: $!";
426 $err_pipe = [ $r, $w ];
428 # Mail we import into lei are private, so headers filtered out
429 # by -mda for public mail are not appropriate
430 local @PublicInbox::MDA::BAD_HEADERS = ();
431 $self->ipc_worker_spawn("lei/store $d", $lei->oldset,
432 { lei => $lei, err_pipe => $err_pipe });
434 require PublicInbox::LeiStoreErr;
435 PublicInbox::LeiStoreErr->new($err_pipe->[0], $lei);