1 # Copyright (C) 2015-2020 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # bidirectional Message-ID <-> Article Number mapping for the NNTP
5 # and web interfaces. This is required for implementing stable article
6 # numbers for NNTP and allows prefix lookups for partial Message-IDs
7 # in case URLs get truncated from copy-n-paste errors by users.
9 # This is maintained by ::SearchIdx
10 package PublicInbox::Msgmap;
14 use PublicInbox::Over;
15 use PublicInbox::Spawn;
18 my ($class, $git_dir, $writable) = @_;
19 my $d = "$git_dir/public-inbox";
20 if ($writable && !-d $d && !mkdir $d) {
22 -d $d or die "$d not created: $err";
24 new_file($class, "$d/msgmap.sqlite3", $writable);
28 my ($class, $f, $rw) = @_;
29 return if !$rw && !-r $f;
31 my $self = bless { filename => $f }, $class;
32 my $dbh = $self->{dbh} = PublicInbox::Over::dbh_new($self, $rw);
34 # TRUNCATE reduces I/O compared to the default (DELETE)
35 $dbh->do('PRAGMA journal_mode = TRUNCATE');
39 $self->created_at(time) unless $self->created_at;
41 my $max = $self->max // 0;
42 $self->num_highwater($max);
48 # used to keep track of used numeric mappings for v2 reindex
50 my ($self, $dir) = @_;
52 my $tmp = "mm_tmp-$$-XXXXXX";
53 my ($fh, $fn) = File::Temp::tempfile($tmp, EXLOCK => 0, DIR => $dir);
54 PublicInbox::Spawn::nodatacow_fd(fileno($fh));
55 $self->{dbh}->sqlite_backup_to_file($fn);
56 $tmp = ref($self)->new_file($fn, 2);
57 $tmp->{dbh}->do('PRAGMA journal_mode = MEMORY');
62 # n.b. invoked directly by scripts/xhdr-num2mid
64 my ($self, $key, $value) = @_;
66 my $sql = 'SELECT val FROM meta WHERE key = ? LIMIT 1';
67 my $dbh = $self->{dbh};
69 defined $value or return $dbh->selectrow_array($sql, undef, $key);
71 $prev = $dbh->selectrow_array($sql, undef, $key);
74 $sql = 'UPDATE meta SET val = ? WHERE key = ?';
75 $dbh->do($sql, undef, $value, $key);
77 $sql = 'INSERT INTO meta (key,val) VALUES (?,?)';
78 $dbh->do($sql, undef, $key, $value);
84 my ($self, $commit) = @_;
85 $self->meta_accessor('last_commit', $commit);
88 # v2 uses this to keep track of how up-to-date Xapian is
89 # old versions may be automatically GC'ed away in the future,
90 # but it's a trivial amount of storage.
92 my ($self, $version, $i, $commit) = @_;
93 $self->meta_accessor("last_xap$version-$i", $commit);
97 my ($self, $second) = @_;
98 $self->meta_accessor('created_at', $second);
102 my ($self, $num) = @_;
103 my $high = $self->{num_highwater} ||=
104 $self->meta_accessor('num_highwater');
105 if (defined($num) && (!defined($high) || ($num > $high))) {
106 $self->{num_highwater} = $num;
107 $self->meta_accessor('num_highwater', $num);
109 $self->{num_highwater};
113 my ($self, $mid) = @_;
114 my $dbh = $self->{dbh};
115 my $sth = $dbh->prepare_cached(<<'');
116 INSERT INTO msgmap (mid) VALUES (?)
118 return unless eval { $sth->execute($mid) };
119 my $num = $dbh->last_insert_id(undef, undef, 'msgmap', 'num');
120 $self->num_highwater($num) if defined($num);
125 my ($self, $num) = @_;
126 my $dbh = $self->{dbh};
127 my $sth = $self->{mid_for} ||=
128 $dbh->prepare('SELECT mid FROM msgmap WHERE num = ? LIMIT 1');
129 $sth->bind_param(1, $num);
131 $sth->fetchrow_array;
135 my ($self, $mid) = @_;
136 my $dbh = $self->{dbh};
137 my $sth = $self->{num_for} ||=
138 $dbh->prepare('SELECT num FROM msgmap WHERE mid = ? LIMIT 1');
139 $sth->bind_param(1, $mid);
141 $sth->fetchrow_array;
145 my $sth = $_[0]->{dbh}->prepare_cached('SELECT MAX(num) FROM msgmap',
148 $sth->fetchrow_array;
152 # breaking MIN and MAX into separate queries speeds up from 250ms
153 # to around 700us with 2.7million messages.
154 my $sth = $_[0]->{dbh}->prepare_cached('SELECT MIN(num) FROM msgmap',
157 ($sth->fetchrow_array, max($_[0]));
161 my ($self, $mid) = @_;
162 my $dbh = $self->{dbh};
163 my $sth = $dbh->prepare('DELETE FROM msgmap WHERE mid = ?');
164 $sth->bind_param(1, $mid);
169 my ($self, $num) = @_;
170 my $dbh = $self->{dbh};
171 my $sth = $dbh->prepare('DELETE FROM msgmap WHERE num = ?');
172 $sth->bind_param(1, $num);
180 $e = eval { $dbh->selectrow_array('EXPLAIN SELECT * FROM msgmap;') };
181 defined $e or $dbh->do('CREATE TABLE msgmap (' .
182 'num INTEGER PRIMARY KEY AUTOINCREMENT, '.
183 'mid VARCHAR(1000) NOT NULL, ' .
186 $e = eval { $dbh->selectrow_array('EXPLAIN SELECT * FROM meta') };
187 defined $e or $dbh->do('CREATE TABLE meta (' .
188 'key VARCHAR(32) PRIMARY KEY, '.
189 'val VARCHAR(255) NOT NULL)');
194 my ($self, $num) = @_;
195 my $ids = $self->{dbh}->selectcol_arrayref(<<'', undef, $$num);
196 SELECT num FROM msgmap WHERE num > ?
197 ORDER BY num ASC LIMIT 1000
199 $$num = $ids->[-1] if @$ids;
204 my ($self, $beg, $end, $cols) = @_;
206 my $dbh = $self->{dbh};
207 my $attr = { Columns => [] };
208 my $mids = $dbh->selectall_arrayref(<<"", $attr, $$beg, $end);
209 SELECT $cols FROM msgmap WHERE num >= ? AND num <= ?
210 ORDER BY num ASC LIMIT 1000
212 $$beg = $mids->[-1]->[0] + 1 if @$mids;
216 # only used for mapping external serial numbers (e.g. articles from gmane)
217 # see scripts/xhdr-num2mid or PublicInbox::Filter::RubyLang for usage
219 my ($self, $num, $mid) = @_;
220 my $sth = $self->{mid_set} ||= do {
221 $self->{dbh}->prepare(
222 'INSERT OR IGNORE INTO msgmap (num,mid) VALUES (?,?)');
224 my $result = $sth->execute($num, $mid);
225 $self->num_highwater($num) if (defined($result) && $result == 1);
231 my $dbh = $self->{dbh} or return;
232 if (($self->{pid} // 0) == $$) {
233 my $f = $dbh->sqlite_db_filename;
234 unlink $f or warn "failed to unlink $f: $!\n";
240 $self->{pid} or die 'BUG: not a temporary clone';
241 $self->{dbh} and die 'BUG: tmp_clone dbh not prepared for parent';
242 defined($self->{filename}) or die 'BUG: {filename} not defined';
243 $self->{dbh} = PublicInbox::Over::dbh_new($self, 2);
244 $self->{dbh}->do('PRAGMA journal_mode = MEMORY');
249 my $pid = $self->{pid} or die 'BUG: not a temporary clone';
250 $pid == $$ or die "BUG: atfork_prepare not called by $pid";
251 my $dbh = $self->{dbh} or die 'BUG: temporary clone not open';
253 # must clobber prepared statements
254 %$self = (filename => $dbh->sqlite_db_filename, pid => $pid);
258 my ($self, $skip_artnum) = @_;
259 return meta_accessor($self, 'skip_artnum') if !defined($skip_artnum);
261 my $cur = num_highwater($self) // 0;
262 if ($skip_artnum < $cur) {
263 die "E: current article number $cur ",
264 "exceeds --skip-artnum=$skip_artnum\n";
268 my $mid = 'skip'.rand.'@'.rand.'.example.com';
269 $ok = mid_set($self, $skip_artnum, $mid);
271 mid_delete($self, $mid);
275 $ok or die '--skip-artnum failed';
277 # in the future, the indexer may use this value for
278 # new messages in old epochs
279 meta_accessor($self, 'skip_artnum', $skip_artnum);
285 # no filename if in-:memory:
286 my $f = $self->{dbh}->sqlite_db_filename // return;
287 if (my @st = stat($f)) { # did st_dev, st_ino change?
288 my $st = pack('dd', $st[0], $st[1]);
289 if ($st ne ($self->{st} // $st)) {
290 my $tmp = eval { ref($self)->new_file($f) };
292 warn "E: DBI->connect($f): $@\n";
298 warn "W: stat $f: $!\n";