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);
10 use PublicInbox::OverIdx;
11 use PublicInbox::LeiSearch;
12 use PublicInbox::Config;
13 use PublicInbox::Spawn qw(run_die);
14 use PublicInbox::ContentHash qw(git_sha);
15 use PublicInbox::MID qw(mids_for_index);
16 use Digest::SHA qw(sha256_hex);
17 my $LOCAL_PFX = qr!\A(?:maildir|mh|mbox.+|mmdf):!i; # TODO: put in LeiToMail?
19 # move this to PublicInbox::Config if other things use it:
20 my %cquote = ("\n" => '\\n', "\t" => '\\t', "\b" => '\\b');
21 sub cquote_val ($) { # cf. git-config(1)
23 $val =~ s/([\n\t\b])/$cquote{$1}/g;
27 sub ARRAY_FIELDS () { qw(only include exclude) }
29 qw(external local remote import-remote import-before threads)
32 sub lss_dir_for ($$;$) {
33 my ($lei, $dstref, $on_fs) = @_;
35 if ($$dstref =~ m,\Aimaps?://,i) { # already canonicalized
36 require PublicInbox::URIimap;
37 my $uri = PublicInbox::URIimap->new($$dstref)->canonical;
41 # can't use Cwd::abs_path since dirname($$dstref) may not exist
42 $$dstref = $lei->rel2abs($$dstref);
43 # Maildirs have trailing '/' internally
44 $$dstref .= '/' if -d $$dstref;
46 @n = ($$dstref =~ m{([^/]+)/*\z}); # basename
48 push @n, sha256_hex($$dstref);
49 my $lss_dir = $lei->share_path . '/saved-searches/';
50 my $d = $lss_dir . join('-', @n);
52 # fall-back to looking up by st_ino + st_dev in case we're in
53 # a symlinked or bind-mounted path
54 if ($on_fs && !-d $d && -e $$dstref) {
56 my $want = pack('dd', @cur[1,0]); # st_ino + st_dev
58 for my $g ("$n[0]-*", '*') {
59 my @maybe = glob("$lss_dir$g/lei.saved-search");
61 $c = PublicInbox::Config->git_config_dump($f);
62 $o = $c->{'lei.q.output'} // next;
63 $o =~ s!$LOCAL_PFX!! or next;
64 @st = stat($o) or next;
65 next if pack('dd', @st[1,0]) ne $want;
66 $f =~ m!\A(.+?)/[^/]+\z! and return $1;
75 my $lss_dir = $lei->share_path.'/saved-searches';
76 return () unless -d $lss_dir;
77 # TODO: persist the cache? Use another format?
78 my $f = $lei->cache_dir."/saved-tmp.$$.".time.'.config';
79 open my $fh, '>', $f or die "open $f: $!";
80 print $fh "[include]\n";
81 for my $p (glob("$lss_dir/*/lei.saved-search")) {
82 print $fh "\tpath = ", cquote_val($p), "\n";
84 close $fh or die "close $f: $!";
85 my $cfg = PublicInbox::Config->git_config_dump($f);
87 my $out = $cfg->get_all('lei.q.output') or return ();
94 sub translate_dedupe ($$$) {
95 my ($self, $lei, $dd) = @_;
97 return 1 if $dd eq 'content'; # the default
98 return $self->{"-dedupe_$dd"} = 1 if ($dd eq 'oid' || $dd eq 'mid');
99 $lei->fail("--dedupe=$dd unsupported with --save");
102 sub up { # updating existing saved search via "lei up"
103 my ($cls, $lei, $dst) = @_;
105 my $self = bless { ale => $lei->ale }, $cls;
107 output2lssdir($self, $lei, \$dir, \$f) or
108 return $lei->fail("--save was not used with $dst cwd=".
110 $self->{-cfg} = PublicInbox::Config->git_config_dump($f);
111 $self->{-ovf} = "$dir/over.sqlite3";
113 $self->{lock_path} = "$self->{-f}.flock";
117 sub new { # new saved search "lei q --save"
118 my ($cls, $lei) = @_;
119 my $self = bless { ale => $lei->ale }, $cls;
120 my $dst = $lei->{ovv}->{dst};
121 my $dir = lss_dir_for($lei, \$dst);
123 File::Path::make_path($dir); # raises on error
125 my $f = $self->{'-f'} = "$dir/lei.saved-search";
126 my $dd = $lei->{opt}->{dedupe};
127 translate_dedupe($self, $lei, $dd) or return;
128 open my $fh, '>', $f or return $lei->fail("open $f: $!");
129 my $sq_dst = PublicInbox::Config::squote_maybe($dst);
130 my $q = $lei->{mset_opt}->{q_raw} // die 'BUG: {q_raw} missing';
132 $q = join("\n", map { "\tq = ".cquote_val($_) } @$q);
134 $q = "\tq = ".cquote_val($q);
136 $dst = "$lei->{ovv}->{fmt}:$dst" if $dst !~ m!\Aimaps?://!i;
138 ; to refresh with new results, run: lei up $sq_dst
144 print $fh "\tdedupe = $dd\n" if $dd;
145 for my $k (ARRAY_FIELDS) {
146 my $ary = $lei->{opt}->{$k} // next;
148 print $fh "\t$k = ".cquote_val($x)."\n";
151 for my $k (BOOL_FIELDS) {
152 my $val = $lei->{opt}->{$k} // next;
153 print $fh "\t$k = ".($val ? 1 : 0)."\n";
155 close($fh) or return $lei->fail("close $f: $!");
156 $self->{lock_path} = "$self->{-f}.flock";
157 $self->{-ovf} = "$dir/over.sqlite3";
161 sub description { $_[0]->{qstr} } # for WWW
163 sub cfg_set { # called by LeiXSearch
164 my ($self, @args) = @_;
165 my $lk = $self->lock_for_scope; # git-config doesn't wait
166 run_die([qw(git config -f), $self->{'-f'}, @args]);
169 # drop-in for LeiDedupe API
171 my ($self, $eml, $smsg) = @_;
172 my $oidx = $self->{oidx} // die 'BUG: no {oidx}';
173 my $blob = $smsg ? $smsg->{blob} : undef;
174 my $lk = $self->lock_for_scope_fast;
175 return 1 if $blob && $oidx->blob_exists($blob);
176 if ($self->{-dedupe_mid}) {
177 for my $mid (@{mids_for_index($eml)}) {
179 return 1 if $oidx->next_by_mid($mid, \$id, \$prv);
182 if (my $xoids = PublicInbox::LeiSearch::xoids_for($self, $eml, 1)) {
183 for my $docid (values %$xoids) {
184 $oidx->add_xref3($docid, -1, $blob, '.');
187 if ($self->{-dedupe_oid}) {
188 $smsg->{blob} //= git_sha(1, $eml)->hexdigest;
189 exists $xoids->{$smsg->{blob}} ? 1 : undef;
194 # n.b. above xoids_for fills out eml->{-lei_fake_mid} if needed
196 $smsg = bless {}, 'PublicInbox::Smsg';
198 $smsg->populate($eml);
201 $smsg->{num} = $oidx->adj_counter('eidx_docid', '+');
202 $smsg->{blob} //= git_sha(1, $eml)->hexdigest;
203 $oidx->add_overview($eml, $smsg);
204 $oidx->add_xref3($smsg->{num}, -1, $smsg->{blob}, '.');
212 $self->{oidx} //= do {
213 my $creat = !-f $self->{-ovf};
214 my $lk = $self->lock_for_scope; # git-config doesn't wait
215 my $oidx = PublicInbox::OverIdx->new($self->{-ovf});
216 $oidx->{-no_fsync} = 1;
219 $oidx->{dbh}->do('PRAGMA journal_mode = WAL');
220 $oidx->eidx_prep; # for xref3
226 sub over { $_[0]->{oidx} } # for xoids_for
228 # don't use ale->git directly since is_dup is called inside
229 # ale->git->cat_async callbacks
230 sub git { $_[0]->{git} //= PublicInbox::Git->new($_[0]->{ale}->git->{git_dir}) }
235 my $lockfh = delete $self->{lockfh}; # from lock_for_scope_fast;
236 my $oidx = delete($self->{oidx}) // return;
246 # find existing directory containing a `lei.saved-search' file based on
247 # $dir_ref which is an output
249 my ($self, $lei, $dir_ref, $fn_ref) = @_;
250 my $dst = $$dir_ref; # imap://$MAILBOX, /path/to/maildir, /path/to/mbox
251 my $dir = lss_dir_for($lei, \$dst, 1);
252 my $f = "$dir/lei.saved-search";
254 $self->{-cfg} = PublicInbox::Config->git_config_dump($f);
263 *nntp_url = \&cloneurl;
264 *base_url = \&PublicInbox::Inbox::base_url;
265 *smsg_eml = \&PublicInbox::Inbox::smsg_eml;
266 *smsg_by_mid = \&PublicInbox::Inbox::smsg_by_mid;
267 *msg_by_mid = \&PublicInbox::Inbox::msg_by_mid;
268 *modified = \&PublicInbox::Inbox::modified;
269 *recent = \&PublicInbox::Inbox::recent;
270 *max_git_epoch = *nntp_usable = *msg_by_path = \&mm; # undef
271 *isrch = *search = \&mm; # TODO
272 *DESTROY = \&pause_dedupe;