1 # Copyright (C) 2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # pretends to be like LeiDedupe and also PublicInbox::Inbox
5 package PublicInbox::LeiSavedSearch;
8 use parent qw(PublicInbox::Lock);
9 use PublicInbox::OverIdx;
10 use PublicInbox::LeiSearch;
11 use PublicInbox::Config;
12 use PublicInbox::Spawn qw(run_die);
13 use PublicInbox::ContentHash qw(git_sha);
14 use PublicInbox::MID qw(mids_for_index);
15 use Digest::SHA qw(sha256_hex);
17 # move this to PublicInbox::Config if other things use it:
18 my %cquote = ("\n" => '\\n', "\t" => '\\t', "\b" => '\\b');
19 sub cquote_val ($) { # cf. git-config(1)
21 $val =~ s/([\n\t\b])/$cquote{$1}/g;
25 sub ARRAY_FIELDS () { qw(only include exclude) }
27 qw(external local remote import-remote import-before threads)
30 sub lss_dir_for ($$) {
31 my ($lei, $dstref) = @_;
33 if ($$dstref =~ m,\Aimaps?://,i) { # already canonicalized
34 require PublicInbox::URIimap;
35 my $uri = PublicInbox::URIimap->new($$dstref)->canonical;
39 $$dstref = $lei->rel2abs($$dstref);
40 $$dstref .= '/' if -d $$dstref;
42 @n = ($$dstref =~ m{([^/]+)/*\z});
44 push @n, sha256_hex($$dstref);
45 $lei->share_path . '/saved-searches/' . join('-', @n);
50 my $lss_dir = $lei->share_path.'/saved-searches/';
51 return () unless -d $lss_dir;
52 # TODO: persist the cache? Use another format?
53 my $f = $lei->cache_dir."/saved-tmp.$$.".time.'.config';
54 open my $fh, '>', $f or die "open $f: $!";
55 print $fh "[include]\n";
56 for my $p (glob("$lss_dir/*/lei.saved-search")) {
57 print $fh "\tpath = ", cquote_val($p), "\n";
59 close $fh or die "close $f: $!";
60 my $cfg = PublicInbox::Config->git_config_dump($f);
62 my $out = $cfg->get_all('lei.q.output') or return ();
64 s!\A(?:maildir|mh|mbox.+|mmdf):!!i;
69 sub translate_dedupe ($$$) {
70 my ($self, $lei, $dd) = @_;
72 return 1 if $dd eq 'content'; # the default
73 return $self->{"-dedupe_$dd"} = 1 if ($dd eq 'oid' || $dd eq 'mid');
74 $lei->fail("--dedupe=$dd unsupported with --save");
77 sub up { # updating existing saved search via "lei up"
78 my ($cls, $lei, $dst) = @_;
80 my $self = bless { ale => $lei->ale }, $cls;
82 output2lssdir($self, $lei, \$dir, \$f) or
83 return $lei->fail("--save was not used with $dst cwd=".
85 $self->{-cfg} = PublicInbox::Config->git_config_dump($f);
86 $self->{-ovf} = "$dir/over.sqlite3";
88 $self->{lock_path} = "$self->{-f}.flock";
92 sub new { # new saved search "lei q --save"
94 my $self = bless { ale => $lei->ale }, $cls;
95 my $dst = $lei->{ovv}->{dst};
96 my $dir = lss_dir_for($lei, \$dst);
98 File::Path::make_path($dir); # raises on error
100 my $f = $self->{'-f'} = "$dir/lei.saved-search";
101 my $dd = $lei->{opt}->{dedupe};
102 translate_dedupe($self, $lei, $dd) or return;
103 open my $fh, '>', $f or return $lei->fail("open $f: $!");
104 my $sq_dst = PublicInbox::Config::squote_maybe($dst);
105 my $q = $lei->{mset_opt}->{q_raw} // die 'BUG: {q_raw} missing';
107 $q = join("\n", map { "\tq = ".cquote_val($_) } @$q);
109 $q = "\tq = ".cquote_val($q);
111 $dst = "$lei->{ovv}->{fmt}:$dst" if $dst !~ m!\Aimaps?://!i;
113 ; to refresh with new results, run: lei up $sq_dst
119 print $fh "\tdedupe = $dd\n" if $dd;
120 for my $k (ARRAY_FIELDS) {
121 my $ary = $lei->{opt}->{$k} // next;
123 print $fh "\t$k = ".cquote_val($x)."\n";
126 for my $k (BOOL_FIELDS) {
127 my $val = $lei->{opt}->{$k} // next;
128 print $fh "\t$k = ".($val ? 1 : 0)."\n";
130 close($fh) or return $lei->fail("close $f: $!");
131 $self->{lock_path} = "$self->{-f}.flock";
132 $self->{-ovf} = "$dir/over.sqlite3";
136 sub description { $_[0]->{qstr} } # for WWW
138 sub cfg_set { # called by LeiXSearch
139 my ($self, @args) = @_;
140 my $lk = $self->lock_for_scope; # git-config doesn't wait
141 run_die([qw(git config -f), $self->{'-f'}, @args]);
144 # drop-in for LeiDedupe API
146 my ($self, $eml, $smsg) = @_;
147 my $oidx = $self->{oidx} // die 'BUG: no {oidx}';
148 my $blob = $smsg ? $smsg->{blob} : undef;
149 my $lk = $self->lock_for_scope_fast;
150 return 1 if $blob && $oidx->blob_exists($blob);
151 if ($self->{-dedupe_mid}) {
152 for my $mid (@{mids_for_index($eml)}) {
154 return 1 if $oidx->next_by_mid($mid, \$id, \$prv);
157 if (my $xoids = PublicInbox::LeiSearch::xoids_for($self, $eml, 1)) {
158 for my $docid (values %$xoids) {
159 $oidx->add_xref3($docid, -1, $blob, '.');
162 if ($self->{-dedupe_oid}) {
163 $smsg->{blob} //= git_sha(1, $eml)->hexdigest;
164 exists $xoids->{$smsg->{blob}} ? 1 : undef;
169 # n.b. above xoids_for fills out eml->{-lei_fake_mid} if needed
171 $smsg = bless {}, 'PublicInbox::Smsg';
173 $smsg->populate($eml);
176 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
177 $smsg->{blob} //= git_sha(1, $eml)->hexdigest;
178 $oidx->add_overview($eml, $smsg);
179 $oidx->add_xref3($smsg->{num}, -1, $smsg->{blob}, '.');
187 $self->{oidx} //= do {
188 my $creat = !-f $self->{-ovf};
189 my $lk = $self->lock_for_scope; # git-config doesn't wait
190 my $oidx = PublicInbox::OverIdx->new($self->{-ovf});
191 $oidx->{-no_fsync} = 1;
194 $oidx->{dbh}->do('PRAGMA journal_mode = WAL');
195 $oidx->eidx_prep; # for xref3
201 sub over { $_[0]->{oidx} } # for xoids_for
203 sub git { $_[0]->{ale}->git }
207 $self->{ale}->git->cleanup;
208 my $lockfh = delete $self->{lockfh}; # from lock_for_scope_fast;
209 my $oidx = delete($self->{oidx}) // return;
219 # find existing directory containing a `lei.saved-search' file based on
220 # $dir_ref which is an output
222 my ($self, $lei, $dir_ref, $fn_ref) = @_;
223 my $dst = $$dir_ref; # imap://$MAILBOX, /path/to/maildir, /path/to/mbox
224 my $dir = lss_dir_for($lei, \$dst);
225 my $f = "$dir/lei.saved-search";
227 $self->{-cfg} = PublicInbox::Config->git_config_dump($f);
236 *nntp_url = \&cloneurl;
237 *base_url = \&PublicInbox::Inbox::base_url;
238 *smsg_eml = \&PublicInbox::Inbox::smsg_eml;
239 *smsg_by_mid = \&PublicInbox::Inbox::smsg_by_mid;
240 *msg_by_mid = \&PublicInbox::Inbox::msg_by_mid;
241 *modified = \&PublicInbox::Inbox::modified;
242 *recent = \&PublicInbox::Inbox::recent;
243 *max_git_epoch = *nntp_usable = *msg_by_path = \&mm; # undef
244 *isrch = *search = \&mm; # TODO
245 *DESTROY = \&pause_dedupe;