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 (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) // do {
155 warn "W: $oid (#$docid) <$mid> not found\n";
158 local $self->{current_info} = $oid;
159 my $x = PublicInbox::Eml->new($bref);
160 $docids{$docid} = $docid if content_hash($x) eq $chash;
163 sort { $a <=> $b } values %docids;
167 my ($self, $eml, $vmd, $docids) = @_;
168 my ($eidx, $tl) = eidx_init($self);
169 $docids //= [ _docids_for($self, $eml) ];
170 for my $docid (@$docids) {
171 $eidx->idx_shard($docid)->ipc_do('set_vmd', $docid, $vmd);
177 my ($self, $eml, $vmd) = @_;
178 my ($eidx, $tl) = eidx_init($self);
179 my @docids = _docids_for($self, $eml);
180 for my $docid (@docids) {
181 $eidx->idx_shard($docid)->ipc_do('add_vmd', $docid, $vmd);
187 my ($self, $eml, $vmd) = @_;
188 my ($eidx, $tl) = eidx_init($self);
189 my @docids = _docids_for($self, $eml);
190 for my $docid (@docids) {
191 $eidx->idx_shard($docid)->ipc_do('remove_vmd', $docid, $vmd);
196 sub set_sync_info ($$$) {
197 my ($self, $oidhex, $sync_info) = @_;
198 ($self->{lms} //= do {
199 require PublicInbox::LeiMailSync;
200 my $f = "$self->{priv_eidx}->{topdir}/mail_sync.sqlite3";
201 my $lms = PublicInbox::LeiMailSync->new($f);
204 })->set_src($oidhex, @$sync_info);
208 my ($self, $eml, $vmd, $xoids) = @_;
209 my $im = $self->importer; # may create new epoch
210 my ($eidx, $tl) = eidx_init($self);
211 my $oidx = $eidx->{oidx}; # PublicInbox::Import::add checks this
212 my $smsg = bless { -oidx => $oidx }, 'PublicInbox::Smsg';
213 my $im_mark = $im->add($eml, undef, $smsg);
214 if ($vmd && $vmd->{sync_info}) {
215 set_sync_info($self, $smsg->{blob}, $vmd->{sync_info});
217 $im_mark or return; # duplicate blob returns undef
219 local $self->{current_info} = $smsg->{blob};
220 my $vivify_xvmd = delete($smsg->{-vivify_xvmd}) // []; # exact matches
221 if ($xoids) { # fuzzy matches from externals in ale->xoids_for
222 delete $xoids->{$smsg->{blob}}; # added later
223 if (scalar keys %$xoids) {
224 my %docids = map { $_ => 1 } @$vivify_xvmd;
225 for my $oid (keys %$xoids) {
226 my @id = $oidx->blob_exists($oid);
229 @$vivify_xvmd = sort { $a <=> $b } keys(%docids);
234 $xoids->{$smsg->{blob}} = 1;
235 for my $docid (@$vivify_xvmd) {
236 my $cur = $oidx->get_art($docid);
237 my $idx = $eidx->idx_shard($docid);
238 if (!$cur || $cur->{bytes} == 0) { # really vivifying
239 $smsg->{num} = $docid;
240 $oidx->add_overview($eml, $smsg);
241 $smsg->{-merge_vmd} = 1;
242 $idx->index_eml($eml, $smsg);
243 } else { # lse fuzzy hit off ale
244 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
246 for my $oid (keys %$xoids) {
247 $oidx->add_xref3($docid, -1, $oid, '.');
249 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
252 } elsif (my @docids = _docids_for($self, $eml)) {
253 # fuzzy match from within lei/store
254 for my $docid (@docids) {
255 my $idx = $eidx->idx_shard($docid);
256 $oidx->add_xref3($docid, -1, $smsg->{blob}, '.');
257 # add_eidx_info for List-Id
258 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
259 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
262 } else { # totally new message
263 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
264 $oidx->add_overview($eml, $smsg);
265 $oidx->add_xref3($smsg->{num}, -1, $smsg->{blob}, '.');
266 my $idx = $eidx->idx_shard($smsg->{num});
267 $idx->index_eml($eml, $smsg);
268 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd) if $vmd;
274 my ($self, $eml, $vmd, $xoids) = @_;
275 add_eml($self, $eml, $vmd, $xoids) //
276 set_eml_vmd($self, $eml, $vmd);
279 sub _external_only ($$$) {
280 my ($self, $xoids, $eml) = @_;
281 my $eidx = $self->{priv_eidx};
282 my $oidx = $eidx->{oidx} // die 'BUG: {oidx} missing';
283 my $smsg = bless { blob => '' }, 'PublicInbox::Smsg';
284 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
285 # save space for an externals-only message
286 my $hdr = $eml->header_obj;
287 $smsg->populate($hdr); # sets lines == 0
289 delete @$smsg{qw(From Subject)};
290 $smsg->{to} = $smsg->{cc} = $smsg->{from} = '';
291 $oidx->add_overview($hdr, $smsg); # subject+references for threading
292 $smsg->{subject} = '';
293 for my $oid (keys %$xoids) {
294 $oidx->add_xref3($smsg->{num}, -1, $oid, '.');
296 my $idx = $eidx->idx_shard($smsg->{num});
297 $idx->index_eml(PublicInbox::Eml->new("\n\n"), $smsg);
302 my ($self, $xoids, $eml, $vmd_mod) = @_;
303 my ($eidx, $tl) = eidx_init($self);
304 my $oidx = $eidx->{oidx};
306 for my $oid (keys %$xoids) {
307 my @docids = $oidx->blob_exists($oid) or next;
308 scalar(@docids) > 1 and
309 warn "W: $oid indexed as multiple docids: @docids\n";
310 for my $docid (@docids) {
311 next if $seen{$docid}++;
312 my $idx = $eidx->idx_shard($docid);
313 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
315 delete $xoids->{$oid};
317 return unless scalar(keys(%$xoids));
319 # see if it was indexed, but with different OID(s)
320 if (my @docids = _docids_for($self, $eml)) {
321 for my $docid (@docids) {
322 next if $seen{$docid};
323 for my $oid (keys %$xoids) {
324 $oidx->add_xref3($docid, -1, $oid, '.');
326 my $idx = $eidx->idx_shard($docid);
327 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
332 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
333 $idx->ipc_do('update_vmd', $smsg->{num}, $vmd_mod);
336 # set or update keywords for external message, called via ipc_do
338 my ($self, $xoids, $eml, $vmd) = @_;
340 my ($eidx, $tl) = eidx_init($self);
341 my $oidx = $eidx->{oidx};
344 # see if we can just update existing docs
345 for my $oid (keys %$xoids) {
346 my @docids = $oidx->blob_exists($oid) or next;
347 scalar(@docids) > 1 and
348 warn "W: $oid indexed as multiple docids: @docids\n";
349 for my $docid (@docids) {
350 next if $seen{$docid}++;
351 my $idx = $eidx->idx_shard($docid);
352 $idx->ipc_do('set_vmd', $docid, $vmd);
354 delete $xoids->{$oid}; # all done with this oid
356 return unless scalar(keys(%$xoids));
358 # n.b. we don't do _docids_for here, we expect the caller
359 # already checked $lse->kw_changed before calling this sub
361 return unless (@{$vmd->{kw} // []}) || (@{$vmd->{L} // []});
363 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
364 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd);
368 my ($self, $wait) = @_;
369 if (my $im = $self->{im}) {
370 $wait ? $im->barrier : $im->checkpoint;
372 $self->{priv_eidx}->checkpoint($wait);
377 _tail_err($self) if $self->{-err_wr};
378 my $dir = $self->{priv_eidx}->{topdir};
379 return unless -e $dir;
380 my $old = delete $self->{-tmp_err};
381 my $pfx = POSIX::strftime('%Y%m%d%H%M%S', gmtime(time));
382 my $err = File::Temp->new(TEMPLATE => "$pfx.$$.lei_storeXXXX",
383 SUFFIX => '.err', DIR => $dir);
384 open STDERR, '>>', $err->filename or die "dup2: $!";
385 STDERR->autoflush(1); # shared with shard subprocesses
386 $self->{-tmp_err} = $err; # separate file description for RO access
393 if (my $im = delete($self->{im})) {
396 $err .= "import done: $@\n";
400 if (my $lms = delete $self->{lms}) {
403 $self->{priv_eidx}->done; # V2Writable::done
408 sub ipc_atfork_child {
410 my $lei = $self->{lei};
411 $lei->_lei_atfork_child(1) if $lei;
413 if (my $err = delete($self->{err_pipe})) {
415 $self->{-err_wr} = $err->[1];
417 $self->SUPER::ipc_atfork_child;
421 my ($self, $lei) = @_;
422 unless ($self->{-ipc_req}) {
423 my $d = $lei->store_path;
424 $self->ipc_lock_init("$d/ipc.lock");
425 substr($d, -length('/lei/store'), 10, '');
427 unless ($lei->{oneshot}) {
428 pipe(my ($r, $w)) or die "pipe: $!";
429 $err_pipe = [ $r, $w ];
431 # Mail we import into lei are private, so headers filtered out
432 # by -mda for public mail are not appropriate
433 local @PublicInbox::MDA::BAD_HEADERS = ();
434 $self->ipc_worker_spawn("lei/store $d", $lei->oldset,
435 { lei => $lei, err_pipe => $err_pipe });
437 require PublicInbox::LeiStoreErr;
438 PublicInbox::LeiStoreErr->new($err_pipe->[0], $lei);