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 my $oid = $cur->{blob};
118 my $docid = $cur->{num};
119 my $bref = $im ? $im->cat_blob($oid) : undef;
120 $bref //= $eidx->git->cat_file($oid) // do {
121 warn "W: $oid (#$docid) <$mid> not found\n";
124 local $self->{current_info} = $oid;
125 my $x = PublicInbox::Eml->new($bref);
126 $docids{$docid} = $docid if content_hash($x) eq $chash;
129 sort { $a <=> $b } values %docids;
132 sub set_eml_keywords {
133 my ($self, $eml, @kw) = @_;
134 my $eidx = eidx_init($self);
135 my @docids = _docids_for($self, $eml);
136 for my $docid (@docids) {
137 $eidx->idx_shard($docid)->ipc_do('set_keywords', $docid, @kw);
142 sub add_eml_keywords {
143 my ($self, $eml, @kw) = @_;
144 my $eidx = eidx_init($self);
145 my @docids = _docids_for($self, $eml);
146 for my $docid (@docids) {
147 $eidx->idx_shard($docid)->ipc_do('add_keywords', $docid, @kw);
152 sub remove_eml_keywords {
153 my ($self, $eml, @kw) = @_;
154 my $eidx = eidx_init($self);
155 my @docids = _docids_for($self, $eml);
156 for my $docid (@docids) {
157 $eidx->idx_shard($docid)->ipc_do('remove_keywords', $docid, @kw)
163 my ($self, $eml, @kw) = @_;
164 my $im = $self->importer; # may create new epoch
165 my $eidx = eidx_init($self); # writes ALL.git/objects/info/alternates
166 my $oidx = $eidx->{oidx};
167 my $smsg = bless { -oidx => $oidx }, 'PublicInbox::Smsg';
168 $im->add($eml, undef, $smsg) or return; # duplicate returns undef
170 local $self->{current_info} = $smsg->{blob};
171 if (my @docids = _docids_for($self, $eml)) {
172 for my $docid (@docids) {
173 my $idx = $eidx->idx_shard($docid);
174 $oidx->add_xref3($docid, -1, $smsg->{blob}, '.');
175 # add_eidx_info for List-Id
176 $idx->ipc_do('add_eidx_info', $docid, '.', $eml);
177 $idx->ipc_do('add_keywords', $docid, @kw) if @kw;
181 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
182 $oidx->add_overview($eml, $smsg);
183 $oidx->add_xref3($smsg->{num}, -1, $smsg->{blob}, '.');
184 my $idx = $eidx->idx_shard($smsg->{num});
185 $idx->index_eml($eml, $smsg);
186 $idx->ipc_do('add_keywords', $smsg->{num}, @kw) if @kw;
192 my ($self, $eml, @kw) = @_;
193 add_eml($self, $eml, @kw) // set_eml_keywords($self, $eml, @kw);
197 my ($self, $eml) = @_;
198 my $lxs = $self->{lxs_all_local} // die 'BUG: no {lxs_all_local}';
199 return if $lxs->xids_for($eml, 1);
200 add_eml($self, $eml);
203 # set or update keywords for external message, called via ipc_do
205 my ($self, $eml, $kw) = @_;
206 my $lxs = $self->{lxs_all_local} // die 'BUG: no {lxs_all_local}';
207 if ($lxs->xids_for($eml, 1)) { # is it in a local external?
208 # TODO: index keywords only
210 set_eml($self, $eml, @$kw);
215 my ($self, $wait) = @_;
216 if (my $im = $self->{im}) {
217 $wait ? $im->barrier : $im->checkpoint;
219 $self->{priv_eidx}->checkpoint($wait);
225 if (my $im = delete($self->{im})) {
228 $err .= "import done: $@\n";
232 $self->{priv_eidx}->done;
236 sub ipc_atfork_child {
238 my $lei = $self->{lei};
239 $lei->lei_atfork_child(1) if $lei;
240 $self->SUPER::ipc_atfork_child;
243 sub refresh_local_externals {
245 my $cfg = $self->{lei}->_lei_cfg or return;
246 my $cur_cfg = $self->{cur_cfg} // -1;
247 my $lxs = $self->{lxs_all_local};
248 if ($cfg != $cur_cfg || !$lxs) {
249 $lxs = PublicInbox::LeiXSearch->new;
250 my @loc = $self->{lei}->externals_each;
251 for my $loc (@loc) { # locals only
252 $lxs->prepare_external($loc) if -d $loc;
254 $self->{lxs_all_local} = $lxs;
255 $self->{cur_cfg} = $cfg;
257 ($lxs->{git_tmp} //= $lxs->git_tmp)->{git_dir};
261 my ($self, $lei) = @_;
262 unless ($self->{-ipc_req}) {
263 require PublicInbox::LeiXSearch;
264 $self->ipc_lock_init;
265 # Mail we import into lei are private, so headers filtered out
266 # by -mda for public mail are not appropriate
267 local @PublicInbox::MDA::BAD_HEADERS = ();
268 $self->ipc_worker_spawn('lei_store', $lei->oldset,
271 $lei->{all_ext_git_dir} = $self->ipc_do('refresh_local_externals');