1 # Copyright (C) 2015-2018 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;
15 use File::Temp qw(tempfile);
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", $writable);
28 my ($f, $writable) = @_;
29 if ($writable && !-f $f) { # SQLite defaults mode to 0644, we want 0666
30 open my $fh, '+>>', $f or die "failed to open $f: $!";
32 my $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
36 ReadOnly => !$writable,
37 sqlite_use_immediate_transaction => 1,
43 my ($class, $f, $writable) = @_;
44 return if !$writable && !-r $f;
46 my $dbh = dbh_new($f, $writable);
47 my $self = bless { dbh => $dbh }, $class;
52 $self->created_at(time) unless $self->created_at;
58 # used to keep track of used numeric mappings for v2 reindex
61 my ($fh, $fn) = tempfile('msgmap-XXXXXXXX', EXLOCK => 0, TMPDIR => 1);
62 $self->{dbh}->sqlite_backup_to_file($fn);
63 my $tmp = ref($self)->new_file($fn, 1);
64 $tmp->{dbh}->do('PRAGMA synchronous = OFF');
65 $tmp->{tmp_name} = $fn; # SQLite won't work if unlinked, apparently
67 close $fh or die "failed to close $fn: $!";
71 # n.b. invoked directly by scripts/xhdr-num2mid
73 my ($self, $key, $value) = @_;
75 my $sql = 'SELECT val FROM meta WHERE key = ? LIMIT 1';
76 my $dbh = $self->{dbh};
78 defined $value or return $dbh->selectrow_array($sql, undef, $key);
80 $prev = $dbh->selectrow_array($sql, undef, $key);
83 $sql = 'UPDATE meta SET val = ? WHERE key = ?';
84 $dbh->do($sql, undef, $value, $key);
86 $sql = 'INSERT INTO meta (key,val) VALUES (?,?)';
87 $dbh->do($sql, undef, $key, $value);
93 my ($self, $commit) = @_;
94 $self->meta_accessor('last_commit', $commit);
97 # v2 uses this to keep track of how up-to-date Xapian is
98 # old versions may be automatically GC'ed away in the future,
99 # but it's a trivial amount of storage.
100 sub last_commit_xap {
101 my ($self, $version, $i, $commit) = @_;
102 $self->meta_accessor("last_xap$version-$i", $commit);
106 my ($self, $second) = @_;
107 $self->meta_accessor('created_at', $second);
111 my ($self, $mid) = @_;
112 my $dbh = $self->{dbh};
113 my $sth = $dbh->prepare_cached(<<'');
114 INSERT OR IGNORE INTO msgmap (mid) VALUES (?)
116 return if $sth->execute($mid) == 0;
117 $dbh->last_insert_id(undef, undef, 'msgmap', 'num');
121 my ($self, $num) = @_;
122 my $dbh = $self->{dbh};
123 my $sth = $self->{mid_for} ||=
124 $dbh->prepare('SELECT mid FROM msgmap WHERE num = ? LIMIT 1');
125 $sth->bind_param(1, $num);
127 $sth->fetchrow_array;
131 my ($self, $mid) = @_;
132 my $dbh = $self->{dbh};
133 my $sth = $self->{num_for} ||=
134 $dbh->prepare('SELECT num FROM msgmap WHERE mid = ? LIMIT 1');
135 $sth->bind_param(1, $mid);
137 $sth->fetchrow_array;
142 my $dbh = $self->{dbh};
143 # breaking MIN and MAX into separate queries speeds up from 250ms
144 # to around 700us with 2.7million messages.
145 my $sth = $dbh->prepare_cached('SELECT MIN(num) FROM msgmap', undef, 1);
147 my $min = $sth->fetchrow_array;
148 $sth = $dbh->prepare_cached('SELECT MAX(num) FROM msgmap', undef, 1);
150 ($min, $sth->fetchrow_array);
154 my ($self, $mid) = @_;
155 my $dbh = $self->{dbh};
156 my $sth = $dbh->prepare('DELETE FROM msgmap WHERE mid = ?');
157 $sth->bind_param(1, $mid);
162 my ($self, $num) = @_;
163 my $dbh = $self->{dbh};
164 my $sth = $dbh->prepare('DELETE FROM msgmap WHERE num = ?');
165 $sth->bind_param(1, $num);
173 $e = eval { $dbh->selectrow_array('EXPLAIN SELECT * FROM msgmap;') };
174 defined $e or $dbh->do('CREATE TABLE msgmap (' .
175 'num INTEGER PRIMARY KEY AUTOINCREMENT, '.
176 'mid VARCHAR(1000) NOT NULL, ' .
179 $e = eval { $dbh->selectrow_array('EXPLAIN SELECT * FROM meta') };
180 defined $e or $dbh->do('CREATE TABLE meta (' .
181 'key VARCHAR(32) PRIMARY KEY, '.
182 'val VARCHAR(255) NOT NULL)');
187 my ($self, $num) = @_;
188 my $ids = $self->{dbh}->selectcol_arrayref(<<'', undef, $$num);
189 SELECT num FROM msgmap WHERE num > ?
190 ORDER BY num ASC LIMIT 1000
192 $$num = $ids->[-1] if @$ids;
197 my ($self, $beg, $end) = @_;
198 my $dbh = $self->{dbh};
199 my $attr = { Columns => [] };
200 my $mids = $dbh->selectall_arrayref(<<'', $attr, $$beg, $end);
201 SELECT num,mid FROM msgmap WHERE num >= ? AND num <= ?
202 ORDER BY num ASC LIMIT 1000
204 $$beg = $mids->[-1]->[0] + 1 if @$mids;
208 # only used for mapping external serial numbers (e.g. articles from gmane)
209 # see scripts/xhdr-num2mid or PublicInbox::Filter::RubyLang for usage
211 my ($self, $num, $mid) = @_;
212 my $sth = $self->{mid_set} ||= do {
213 $self->{dbh}->prepare(
214 'INSERT OR IGNORE INTO msgmap (num,mid) VALUES (?,?)');
216 $sth->execute($num, $mid);
222 my $f = delete $self->{tmp_name};
223 if (defined $f && $self->{pid} == $$) {
224 unlink $f or warn "failed to unlink $f: $!\n";
230 my $f = $self->{tmp_name} or die "not a temporary clone\n";
231 delete $self->{dbh} and die "tmp_clone dbh not prepared for parent";
232 my $dbh = $self->{dbh} = dbh_new($f, 1);
233 $dbh->do('PRAGMA synchronous = OFF');
238 my $f = $self->{tmp_name} or die "not a temporary clone\n";
239 $self->{pid} == $$ or
240 die "BUG: atfork_prepare not called from $self->{pid}\n";
241 $self->{dbh} or die "temporary clone not open\n";
242 # must clobber prepared statements
243 %$self = (tmp_name => $f, pid => $$);