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
8 # for xref3, the following are constant: $eidx_key = '.', $xnum = -1
9 package PublicInbox::LeiStore;
12 use parent qw(PublicInbox::Lock PublicInbox::IPC);
13 use PublicInbox::ExtSearchIdx;
14 use PublicInbox::Import;
15 use PublicInbox::InboxWritable qw(eml_from_path);
16 use PublicInbox::V2Writable;
17 use PublicInbox::ContentHash qw(content_hash);
18 use PublicInbox::MID qw(mids);
19 use PublicInbox::LeiSearch;
21 use List::Util qw(max);
24 my (undef, $dir, $opt) = @_;
25 my $eidx = PublicInbox::ExtSearchIdx->new($dir, $opt);
26 my $self = bless { priv_eidx => $eidx }, __PACKAGE__;
27 eidx_init($self)->done if $opt->{creat};
31 sub git { $_[0]->{priv_eidx}->git } # read-only
33 sub packing_factor { $PublicInbox::V2Writable::PACKING_FACTOR }
36 $_[0]->{rotate_bytes} // ((1024 * 1024 * 1024) / $_[0]->packing_factor)
39 sub git_pfx { "$_[0]->{priv_eidx}->{topdir}/local" };
43 if (opendir(my $dh, $self->git_pfx)) {
45 substr($_, 0, -4) + 0; # drop ".git" suffix
46 } grep(/\A[0-9]+\.git\z/, readdir($dh))) // 0;
48 $!{ENOENT} ? 0 : die("opendir ${\$self->git_pfx}: $!\n");
54 chomp(my $i = $git->qx(qw(var GIT_COMMITTER_IDENT)));
55 warn "$git->{git_dir} GIT_COMMITTER_IDENT failed\n" if $?;
56 $i =~ /\A(.+) <([^>]+)> [0-9]+ [-\+]?[0-9]+$/ ? ($1, $2) :
57 ('lei user', 'x@example.com')
65 return $im if $im->{bytes_added} < $self->rotate_bytes;
71 $max = $self->git_epoch_max + 1;
73 my $pfx = $self->git_pfx;
74 $max //= $self->git_epoch_max;
76 my $latest = "$pfx/$max.git";
78 PublicInbox::Import::init_bare($latest);
79 my $git = PublicInbox::Git->new($latest);
80 $git->qx(qw(config core.sharedRepository 0600)) if !$old;
81 my $packed_bytes = $git->packed_bytes;
82 my $unpacked_bytes = $packed_bytes / $self->packing_factor;
83 if ($unpacked_bytes >= $self->rotate_bytes) {
87 my ($n, $e) = git_ident($git);
88 $self->{im} = $im = PublicInbox::Import->new($git, $n, $e);
89 $im->{bytes_added} = int($packed_bytes / $self->packing_factor);
90 $im->{lock_path} = undef;
91 $im->{path_type} = 'v2';
97 PublicInbox::LeiSearch->new($_[0]->{priv_eidx}->{topdir});
102 my $eidx = $self->{priv_eidx};
103 $eidx->idx_init({-private => 1});
107 sub _docids_for ($$) {
108 my ($self, $eml) = @_;
110 my ($chash, $mids) = PublicInbox::LeiSearch::content_key($eml);
111 my $eidx = eidx_init($self);
112 my $oidx = $eidx->{oidx};
113 my $im = $self->{im};
114 for my $mid (@$mids) {
116 while (my $cur = $oidx->next_by_mid($mid, \$id, \$prev)) {
117 next if $cur->{bytes} == 0; # external-only message
118 my $oid = $cur->{blob};
119 my $docid = $cur->{num};
120 my $bref = $im ? $im->cat_blob($oid) : undef;
121 $bref //= $eidx->git->cat_file($oid) // do {
122 warn "W: $oid (#$docid) <$mid> not found\n";
125 local $self->{current_info} = $oid;
126 my $x = PublicInbox::Eml->new($bref);
127 $docids{$docid} = $docid if content_hash($x) eq $chash;
130 sort { $a <=> $b } values %docids;
134 my ($self, $eml, $vmd, $docids) = @_;
135 my $eidx = eidx_init($self);
136 $docids //= [ _docids_for($self, $eml) ];
137 for my $docid (@$docids) {
138 $eidx->idx_shard($docid)->ipc_do('set_vmd', $docid, $vmd);
144 my ($self, $eml, $vmd) = @_;
145 my $eidx = eidx_init($self);
146 my @docids = _docids_for($self, $eml);
147 for my $docid (@docids) {
148 $eidx->idx_shard($docid)->ipc_do('add_vmd', $docid, $vmd);
154 my ($self, $eml, $vmd) = @_;
155 my $eidx = eidx_init($self);
156 my @docids = _docids_for($self, $eml);
157 for my $docid (@docids) {
158 $eidx->idx_shard($docid)->ipc_do('remove_vmd', $docid, $vmd);
164 my ($self, $eml, $vmd, $xoids) = @_;
165 my $im = $self->importer; # may create new epoch
166 my $eidx = eidx_init($self); # writes ALL.git/objects/info/alternates
167 my $oidx = $eidx->{oidx}; # PublicInbox::Import::add checks this
168 my $smsg = bless { -oidx => $oidx }, 'PublicInbox::Smsg';
169 $im->add($eml, undef, $smsg) or return; # duplicate returns undef
171 local $self->{current_info} = $smsg->{blob};
172 my $vivify_xvmd = delete($smsg->{-vivify_xvmd}) // []; # exact matches
173 if ($xoids) { # fuzzy matches from externals in ale->xoids_for
174 delete $xoids->{$smsg->{blob}}; # added later
175 if (scalar keys %$xoids) {
176 my %docids = map { $_ => 1 } @$vivify_xvmd;
177 for my $oid (keys %$xoids) {
178 my @id = $oidx->blob_exists($oid);
181 @$vivify_xvmd = sort { $a <=> $b } keys(%docids);
186 $xoids->{$smsg->{blob}} = 1;
187 for my $docid (@$vivify_xvmd) {
188 my $cur = $oidx->get_art($docid);
189 my $idx = $eidx->idx_shard($docid);
190 if (!$cur || $cur->{bytes} == 0) { # really vivifying
191 $smsg->{num} = $docid;
192 $oidx->add_overview($eml, $smsg);
193 $smsg->{-merge_vmd} = 1;
194 $idx->index_eml($eml, $smsg);
195 } else { # lse fuzzy hit off ale
196 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
198 for my $oid (keys %$xoids) {
199 $oidx->add_xref3($docid, -1, $oid, '.');
201 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
204 } elsif (my @docids = _docids_for($self, $eml)) {
205 # fuzzy match from within lei/store
206 for my $docid (@docids) {
207 my $idx = $eidx->idx_shard($docid);
208 $oidx->add_xref3($docid, -1, $smsg->{blob}, '.');
209 # add_eidx_info for List-Id
210 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
211 $idx->ipc_do('add_vmd', $docid, $vmd) if $vmd;
214 } else { # totally new message
215 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
216 $oidx->add_overview($eml, $smsg);
217 $oidx->add_xref3($smsg->{num}, -1, $smsg->{blob}, '.');
218 my $idx = $eidx->idx_shard($smsg->{num});
219 $idx->index_eml($eml, $smsg);
220 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd) if $vmd;
226 my ($self, $eml, $vmd, $xoids) = @_;
227 add_eml($self, $eml, $vmd, $xoids) //
228 set_eml_vmd($self, $eml, $vmd);
231 sub _external_only ($$$) {
232 my ($self, $xoids, $eml) = @_;
233 my $eidx = $self->{priv_eidx};
234 my $oidx = $eidx->{oidx} // die 'BUG: {oidx} missing';
235 my $smsg = bless { blob => '' }, 'PublicInbox::Smsg';
236 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
237 # save space for an externals-only message
238 my $hdr = $eml->header_obj;
239 $smsg->populate($hdr); # sets lines == 0
241 delete @$smsg{qw(From Subject)};
242 $smsg->{to} = $smsg->{cc} = $smsg->{from} = '';
243 $oidx->add_overview($hdr, $smsg); # subject+references for threading
244 $smsg->{subject} = '';
245 for my $oid (keys %$xoids) {
246 $oidx->add_xref3($smsg->{num}, -1, $oid, '.');
248 my $idx = $eidx->idx_shard($smsg->{num});
249 $idx->index_eml(PublicInbox::Eml->new("\n\n"), $smsg);
254 my ($self, $xoids, $eml, $vmd_mod) = @_;
255 my $eidx = eidx_init($self);
256 my $oidx = $eidx->{oidx};
258 for my $oid (keys %$xoids) {
259 my @docids = $oidx->blob_exists($oid) or next;
260 scalar(@docids) > 1 and
261 warn "W: $oid indexed as multiple docids: @docids\n";
262 for my $docid (@docids) {
263 next if $seen{$docid}++;
264 my $idx = $eidx->idx_shard($docid);
265 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
267 delete $xoids->{$oid};
269 return unless scalar(keys(%$xoids));
271 # see if it was indexed, but with different OID(s)
272 if (my @docids = _docids_for($self, $eml)) {
273 for my $docid (@docids) {
274 next if $seen{$docid};
275 for my $oid (keys %$xoids) {
276 $oidx->add_xref3($docid, -1, $oid, '.');
278 my $idx = $eidx->idx_shard($docid);
279 $idx->ipc_do('update_vmd', $docid, $vmd_mod);
284 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
285 $idx->ipc_do('update_vmd', $smsg->{num}, $vmd_mod);
288 # set or update keywords for external message, called via ipc_do
290 my ($self, $xoids, $eml, $vmd) = @_;
292 my $eidx = eidx_init($self);
293 my $oidx = $eidx->{oidx};
296 # see if we can just update existing docs
297 for my $oid (keys %$xoids) {
298 my @docids = $oidx->blob_exists($oid) or next;
299 scalar(@docids) > 1 and
300 warn "W: $oid indexed as multiple docids: @docids\n";
301 for my $docid (@docids) {
302 next if $seen{$docid}++;
303 my $idx = $eidx->idx_shard($docid);
304 $idx->ipc_do('set_vmd', $docid, $vmd);
306 delete $xoids->{$oid}; # all done with this oid
308 return unless scalar(keys(%$xoids));
310 # see if it was indexed, but with different OID(s)
311 if (my @docids = _docids_for($self, $eml)) {
312 for my $docid (@docids) {
313 next if $seen{$docid};
314 for my $oid (keys %$xoids) {
315 $oidx->add_xref3($docid, -1, $oid, '.');
317 my $idx = $eidx->idx_shard($docid);
318 $idx->ipc_do('set_vmd', $docid, $vmd);
323 my ($smsg, $idx) = _external_only($self, $xoids, $eml);
324 $idx->ipc_do('add_vmd', $smsg->{num}, $vmd);
328 my ($self, $wait) = @_;
329 if (my $im = $self->{im}) {
330 $wait ? $im->barrier : $im->checkpoint;
332 $self->{priv_eidx}->checkpoint($wait);
338 if (my $im = delete($self->{im})) {
341 $err .= "import done: $@\n";
345 $self->{priv_eidx}->done;
349 sub ipc_atfork_child {
351 my $lei = $self->{lei};
352 $lei->_lei_atfork_child(1) if $lei;
353 $self->SUPER::ipc_atfork_child;
357 my ($self, $lei) = @_;
358 unless ($self->{-ipc_req}) {
359 my $d = $lei->store_path;
360 $self->ipc_lock_init("$d/ipc.lock");
361 substr($d, -length('/lei/store'), 10, '');
362 # Mail we import into lei are private, so headers filtered out
363 # by -mda for public mail are not appropriate
364 local @PublicInbox::MDA::BAD_HEADERS = ();
365 $self->ipc_worker_spawn("lei/store $d", $lei->oldset,