1 # Copyright (C) 2018-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # for XOVER, OVER in NNTP, and feeds/homepage/threads in PSGI
5 # Unlike Msgmap, this is an _UNSTABLE_ database which can be
6 # tweaked/updated over time and rebuilt.
7 package PublicInbox::Over;
12 use PublicInbox::Smsg;
13 use Compress::Zlib qw(uncompress);
14 use constant DEFAULT_LIMIT => 1000;
18 my $f = $self->{filename};
19 if ($rw && !-f $f) { # SQLite defaults mode to 0644, we want 0666
20 open my $fh, '+>>', $f or die "failed to open $f: $!";
25 @st = stat($f) or die "failed to stat $f: $!";
26 $st = pack('dd', $st[0], $st[1]); # 0: dev, 1: inode
27 $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
32 sqlite_use_immediate_transaction => 1,
35 @st = stat($f) or die "failed to stat $f: $!";
36 $st = pack('dd', $st[0], $st[1]);
37 } while ($st ne $self->{st} && $tries++ < 3);
38 warn "W: $f: .st_dev, .st_ino unstable\n" if $st ne $self->{st};
39 $dbh->{sqlite_unicode} = 1;
45 bless { filename => $f }, $class;
48 sub disconnect { $_[0]->{dbh} = undef }
50 sub connect { $_[0]->{dbh} ||= $_[0]->dbh_new }
52 sub load_from_row ($;$) {
53 my ($smsg, $cull) = @_;
54 bless $smsg, 'PublicInbox::Smsg';
55 if (defined(my $data = delete $smsg->{ddd})) {
56 $data = uncompress($data);
58 PublicInbox::Smsg::load_from_data($smsg, $data);
60 # saves over 600K for 1000+ message threads
61 PublicInbox::Smsg::psgi_cull($smsg) if $cull;
67 my ($self, $sql, $opts, @args) = @_;
68 my $dbh = $self->connect;
69 my $lim = (($opts->{limit} || 0) + 0) || DEFAULT_LIMIT;
71 my $msgs = $dbh->selectall_arrayref($sql, { Slice => {} }, @args);
72 my $cull = $opts->{cull};
73 load_from_row($_, $cull) for @$msgs;
78 my ($self, $beg, $end) = @_;
79 do_get($self, <<'', {}, $beg, $end);
80 SELECT num,ts,ds,ddd FROM over WHERE num >= ? AND num <= ?
86 my ($self, $ts, $prev) = @_;
87 do_get($self, <<'', {}, $ts, $prev);
88 SELECT num,ddd FROM over WHERE ts >= ? AND num > ?
93 sub nothing () { wantarray ? (0, []) : [] };
96 my ($self, $mid, $prev) = @_;
97 my $dbh = $self->connect;
98 my $opts = { cull => 1 };
100 my $id = $dbh->selectrow_array(<<'', undef, $mid);
101 SELECT id FROM msgid WHERE mid = ? LIMIT 1
103 defined $id or return nothing;
105 my $num = $dbh->selectrow_array(<<'', undef, $id);
106 SELECT num FROM id2num WHERE id = ? AND num > 0
107 ORDER BY num ASC LIMIT 1
109 defined $num or return nothing;
111 my ($tid, $sid) = $dbh->selectrow_array(<<'', undef, $num);
112 SELECT tid,sid FROM over WHERE num = ? LIMIT 1
114 defined $tid or return nothing; # $sid may be undef
116 my $cond_all = '(tid = ? OR sid = ?) AND num > ?';
119 if ($prev) { # mboxrd stream, only
120 $num = $prev->{num} || 0;
124 my $cols = 'num,ts,ds,ddd';
126 return do_get($self, <<"", $opts, $tid, $sid, $num);
127 SELECT $cols FROM over WHERE $cond_all
128 ORDER BY $sort_col ASC
132 # HTML view always wants an array and never uses $prev,
133 # but the mbox stream never wants an array and always has $prev
134 die '$prev not supported with wantarray' if $prev;
135 my $nr = $dbh->selectrow_array(<<"", undef, $tid, $sid, $num);
136 SELECT COUNT(num) FROM over WHERE $cond_all
138 # giant thread, prioritize strict (tid) matches and throw
139 # in the loose (sid) matches at the end
140 my $msgs = do_get($self, <<"", $opts, $tid, $num);
141 SELECT $cols FROM over WHERE tid = ? AND num > ?
142 ORDER BY $sort_col ASC
144 # do we have room for loose matches? get the most recent ones, first:
145 my $lim = DEFAULT_LIMIT - scalar(@$msgs);
147 $opts->{limit} = $lim;
148 my $loose = do_get($self, <<"", $opts, $tid, $sid, $num);
149 SELECT $cols FROM over WHERE tid != ? AND sid = ? AND num > ?
150 ORDER BY $sort_col DESC
152 # TODO separate strict and loose matches here once --reindex
153 # is fixed to preserve `tid' properly
154 push @$msgs, @$loose;
160 my ($self, $opts, $after, $before) = @_;
162 if (defined($before)) {
163 if (defined($after)) {
164 $s = '+num > 0 AND ts >= ? AND ts <= ? ORDER BY ts DESC';
165 @v = ($after, $before);
167 $s = '+num > 0 AND ts <= ? ORDER BY ts DESC';
171 if (defined($after)) {
172 $s = '+num > 0 AND ts >= ? ORDER BY ts ASC';
175 $s = '+num > 0 ORDER BY ts DESC';
178 my $msgs = do_get($self, <<"", $opts, @v);
179 SELECT ts,ds,ddd FROM over WHERE $s
181 return $msgs unless wantarray;
183 my $nr = $self->{dbh}->selectrow_array(<<'');
184 SELECT COUNT(num) FROM over WHERE num > 0
190 my ($self, $num) = @_;
191 my $dbh = $self->connect;
192 my $sth = $dbh->prepare_cached(<<'', undef, 1);
193 SELECT num,ds,ts,ddd FROM over WHERE num = ? LIMIT 1
196 my $smsg = $sth->fetchrow_hashref;
197 $smsg ? load_from_row($smsg) : undef;
201 my ($self, $mid, $id, $prev) = @_;
202 my $dbh = $self->connect;
204 unless (defined $$id) {
205 my $sth = $dbh->prepare_cached(<<'', undef, 1);
206 SELECT id FROM msgid WHERE mid = ? LIMIT 1
209 $$id = $sth->fetchrow_array;
210 defined $$id or return;
212 my $sth = $dbh->prepare_cached(<<"", undef, 1);
213 SELECT num FROM id2num WHERE id = ? AND num > ?
214 ORDER BY num ASC LIMIT 1
217 $sth->execute($$id, $$prev);
218 my $num = $sth->fetchrow_array or return;
221 $sth = $dbh->prepare_cached(<<"", undef, 1);
222 SELECT num,ts,ds,ddd FROM over WHERE num = ? LIMIT 1
225 my $smsg = $sth->fetchrow_hashref or return;
226 load_from_row($smsg);
229 # IMAP search, this is limited by callers to UID_SLICE size (50K)
231 my ($self, $beg, $end, $sql) = @_;
232 my $dbh = $self->connect;
233 my $q = 'SELECT num FROM over WHERE num >= ? AND num <= ?';
235 # This is read-only, anyways; but caller should verify it's
236 # only sending \A[0-9]+\z for ds and ts column ranges
238 $q .= ' ORDER BY num ASC';
239 $dbh->selectcol_arrayref($q, undef, $beg, $end);
244 my $sth = $self->connect->prepare_cached(<<'', undef, 1);
245 SELECT MAX(num) FROM over WHERE num > 0
248 $sth->fetchrow_array // 0;
252 my ($self, $uid_base, $uid_end) = @_;
253 my $sth = $self->connect->prepare_cached(<<'', undef, 1);
254 SELECT COUNT(num) FROM over WHERE num > ? AND num <= ?
256 $sth->execute($uid_base, $uid_end);
257 $sth->fetchrow_array;
262 if (my @st = stat($self->{filename})) { # did st_dev, st_ino change?
263 my $st = pack('dd', $st[0], $st[1]);
265 # don't actually reopen, just let {dbh} be recreated later
266 delete($self->{dbh}) if ($st ne ($self->{st} // $st));
268 warn "W: stat $self->{filename}: $!\n";