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 $i =~ /\A(.+) <([^>]+)> [0-9]+ [-\+]?[0-9]+$/ ? ($1, $2) :
67 ('lei user', 'x@example.com')
75 return $im if $im->{bytes_added} < $self->rotate_bytes;
81 $max = $self->git_epoch_max + 1;
83 my $pfx = $self->git_pfx;
84 $max //= $self->git_epoch_max;
86 my $latest = "$pfx/$max.git";
88 PublicInbox::Import::init_bare($latest);
89 my $git = PublicInbox::Git->new($latest);
91 $git->qx(qw(config core.sharedRepository 0600));
92 $self->done; # force eidx_init on next round
94 my $packed_bytes = $git->packed_bytes;
95 my $unpacked_bytes = $packed_bytes / $self->packing_factor;
96 if ($unpacked_bytes >= $self->rotate_bytes) {
100 my ($n, $e) = git_ident($git);
101 $self->{im} = $im = PublicInbox::Import->new($git, $n, $e);
102 $im->{bytes_added} = int($packed_bytes / $self->packing_factor);
103 $im->{lock_path} = undef;
104 $im->{path_type} = 'v2';
110 PublicInbox::LeiSearch->new($_[0]->{priv_eidx}->{topdir});
113 # follows the stderr file
116 print { $self->{-err_wr} } readline($self->{-tmp_err});
121 my $eidx = $self->{priv_eidx};
122 my $tl = wantarray && $self->{-err_wr} ?
123 PublicInbox::OnDestroy->new($$, \&_tail_err, $self) :
125 $eidx->idx_init({-private => 1});
126 wantarray ? ($eidx, $tl) : $eidx;
129 sub _docids_for ($$) {
130 my ($self, $eml) = @_;
132 my $eidx = $self->{priv_eidx};
133 my ($chash, $mids) = PublicInbox::LeiSearch::content_key($eml);
134 my $oidx = $eidx->{oidx};
135 my $im = $self->{im};
136 for my $mid (@$mids) {
138 while (my $cur = $oidx->next_by_mid($mid, \$id, \$prev)) {
139 next if $cur->{bytes} == 0; # external-only message
140 my $oid = $cur->{blob};
141 my $docid = $cur->{num};
142 my $bref = $im ? $im->cat_blob($oid) : undef;
143 $bref //= $eidx->git->cat_file($oid) // do {
144 warn "W: $oid (#$docid) <$mid> not found\n";
147 local $self->{current_info} = $oid;
148 my $x = PublicInbox::Eml->new($bref);
149 $docids{$docid} = $docid if content_hash($x) eq $chash;
152 sort { $a <=> $b } values %docids;
156 my ($self, $eml, $vmd, $docids) = @_;
157 my ($eidx, $tl) = eidx_init($self);
158 $docids //= [ _docids_for($self, $eml) ];
159 for my $docid (@$docids) {
160 $eidx->idx_shard($docid)->ipc_do('set_vmd', $docid, $vmd);
166 my ($self, $eml, $vmd) = @_;
167 my ($eidx, $tl) = eidx_init($self);
168 my @docids = _docids_for($self, $eml);
169 for my $docid (@docids) {
170 $eidx->idx_shard($docid)->ipc_do('add_vmd', $docid, $vmd);
176 my ($self, $eml, $vmd) = @_;
177 my ($eidx, $tl) = eidx_init($self);
178 my @docids = _docids_for($self, $eml);
179 for my $docid (@docids) {
180 $eidx->idx_shard($docid)->ipc_do('remove_vmd', $docid, $vmd);
186 my ($self, $eml, $vmd, $xoids) = @_;
187 my $im = $self->importer; # may create new epoch
188 my ($eidx, $tl) = eidx_init($self); # updates/writes alternates file
189 my $oidx = $eidx->{oidx}; # PublicInbox::Import::add checks this
190 my $smsg = bless { -oidx => $oidx }, 'PublicInbox::Smsg';
191 $im->add($eml, undef, $smsg) or return; # duplicate returns undef
193 local $self->{current_info} = $smsg->{blob};
194 my $vivify_xvmd = delete($smsg->{-vivify_xvmd}) // []; # exact matches
195 if ($xoids) { # fuzzy matches from externals in ale->xoids_for
196 delete $xoids->{$smsg->{blob}}; # added later
197 if (scalar keys %$xoids) {
198 my %docids = map { $_ => 1 } @$vivify_xvmd;
199 for my $oid (keys %$xoids) {
200 my @id = $oidx->blob_exists($oid);
203 @$vivify_xvmd = sort { $a <=> $b } keys(%docids);
208 $xoids->{$smsg->{blob}} = 1;
209 for my $docid (@$vivify_xvmd) {
210 my $cur = $oidx->get_art($docid);
211 my $idx = $eidx->idx_shard($docid);
212 if (!$cur || $cur->{bytes} == 0) { # really vivifying
213 $smsg->{num} = $docid;
214 $oidx->add_overview($eml, $smsg);
215 $smsg->{-merge_vmd} = 1;
216 $idx->index_eml($eml, $smsg);
217 } else { # lse fuzzy hit off ale
218 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
220 for my $oid (keys %$xoids) {
221 $oidx->add_xref3($docid, -1, $oid, '.');
223 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
226 } elsif (my @docids = _docids_for($self, $eml)) {
227 # fuzzy match from within lei/store
228 for my $docid (@docids) {
229 my $idx = $eidx->idx_shard($docid);
230 $oidx->add_xref3($docid, -1, $smsg->{blob}, '.');
231 # add_eidx_info for List-Id
232 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
233 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
236 } else { # totally new message
237 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
238 $oidx->add_overview($eml, $smsg);
239 $oidx->add_xref3($smsg->{num}, -1, $smsg->{blob}, '.');
240 my $idx = $eidx->idx_shard($smsg->{num});
241 $idx->index_eml($eml, $smsg);
242 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd) if $vmd;
248 my ($self, $eml, $vmd, $xoids) = @_;
249 add_eml($self, $eml, $vmd, $xoids) //
250 set_eml_vmd($self, $eml, $vmd);
253 sub _external_only ($$$) {
254 my ($self, $xoids, $eml) = @_;
255 my $eidx = $self->{priv_eidx};
256 my $oidx = $eidx->{oidx} // die 'BUG: {oidx} missing';
257 my $smsg = bless { blob => '' }, 'PublicInbox::Smsg';
258 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
259 # save space for an externals-only message
260 my $hdr = $eml->header_obj;
261 $smsg->populate($hdr); # sets lines == 0
263 delete @$smsg{qw(From Subject)};
264 $smsg->{to} = $smsg->{cc} = $smsg->{from} = '';
265 $oidx->add_overview($hdr, $smsg); # subject+references for threading
266 $smsg->{subject} = '';
267 for my $oid (keys %$xoids) {
268 $oidx->add_xref3($smsg->{num}, -1, $oid, '.');
270 my $idx = $eidx->idx_shard($smsg->{num});
271 $idx->index_eml(PublicInbox::Eml->new("\n\n"), $smsg);
276 my ($self, $xoids, $eml, $vmd_mod) = @_;
277 my ($eidx, $tl) = eidx_init($self);
278 my $oidx = $eidx->{oidx};
280 for my $oid (keys %$xoids) {
281 my @docids = $oidx->blob_exists($oid) or next;
282 scalar(@docids) > 1 and
283 warn "W: $oid indexed as multiple docids: @docids\n";
284 for my $docid (@docids) {
285 next if $seen{$docid}++;
286 my $idx = $eidx->idx_shard($docid);
287 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
289 delete $xoids->{$oid};
291 return unless scalar(keys(%$xoids));
293 # see if it was indexed, but with different OID(s)
294 if (my @docids = _docids_for($self, $eml)) {
295 for my $docid (@docids) {
296 next if $seen{$docid};
297 for my $oid (keys %$xoids) {
298 $oidx->add_xref3($docid, -1, $oid, '.');
300 my $idx = $eidx->idx_shard($docid);
301 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
306 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
307 $idx->ipc_do('update_vmd', $smsg->{num}, $vmd_mod);
310 # set or update keywords for external message, called via ipc_do
312 my ($self, $xoids, $eml, $vmd) = @_;
314 my ($eidx, $tl) = eidx_init($self);
315 my $oidx = $eidx->{oidx};
318 # see if we can just update existing docs
319 for my $oid (keys %$xoids) {
320 my @docids = $oidx->blob_exists($oid) or next;
321 scalar(@docids) > 1 and
322 warn "W: $oid indexed as multiple docids: @docids\n";
323 for my $docid (@docids) {
324 next if $seen{$docid}++;
325 my $idx = $eidx->idx_shard($docid);
326 $idx->ipc_do('set_vmd', $docid, $vmd);
328 delete $xoids->{$oid}; # all done with this oid
330 return unless scalar(keys(%$xoids));
332 # n.b. we don't do _docids_for here, we expect the caller
333 # already checked $lse->kw_changed before calling this sub
335 return unless (@{$vmd->{kw} // []}) || (@{$vmd->{L} // []});
337 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
338 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd);
342 my ($self, $wait) = @_;
343 if (my $im = $self->{im}) {
344 $wait ? $im->barrier : $im->checkpoint;
346 $self->{priv_eidx}->checkpoint($wait);
351 _tail_err($self) if $self->{-err_wr};
352 my $dir = $self->{priv_eidx}->{topdir};
353 return unless -e $dir;
354 my $old = delete $self->{-tmp_err};
355 my $pfx = POSIX::strftime('%Y%m%d%H%M%S', gmtime(time));
356 my $err = File::Temp->new(TEMPLATE => "$pfx.$$.lei_storeXXXX",
357 SUFFIX => '.err', DIR => $dir);
358 open STDERR, '>>', $err->filename or die "dup2: $!";
359 STDERR->autoflush(1); # shared with shard subprocesses
360 $self->{-tmp_err} = $err; # separate file description for RO access
367 if (my $im = delete($self->{im})) {
370 $err .= "import done: $@\n";
374 $self->{priv_eidx}->done; # V2Writable::done
379 sub ipc_atfork_child {
381 my $lei = $self->{lei};
382 $lei->_lei_atfork_child(1) if $lei;
384 if (my $err = delete($self->{err_pipe})) {
386 $self->{-err_wr} = $err->[1];
388 $SIG{__WARN__} = PublicInbox::Eml::warn_ignore_cb();
389 $self->SUPER::ipc_atfork_child;
393 my ($self, $lei) = @_;
394 unless ($self->{-ipc_req}) {
395 my $d = $lei->store_path;
396 $self->ipc_lock_init("$d/ipc.lock");
397 substr($d, -length('/lei/store'), 10, '');
399 unless ($lei->{oneshot}) {
400 pipe(my ($r, $w)) or die "pipe: $!";
401 $err_pipe = [ $r, $w ];
403 # Mail we import into lei are private, so headers filtered out
404 # by -mda for public mail are not appropriate
405 local @PublicInbox::MDA::BAD_HEADERS = ();
406 $self->ipc_worker_spawn("lei/store $d", $lei->oldset,
407 { lei => $lei, err_pipe => $err_pipe });
409 require PublicInbox::LeiStoreErr;
410 PublicInbox::LeiStoreErr->new($err_pipe->[0], $lei);