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 my $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
33 ReadOnly => !$writable,
34 sqlite_use_immediate_transaction => 1,
36 $dbh->do('PRAGMA case_sensitive_like = ON');
41 my ($class, $f, $writable) = @_;
42 return if !$writable && !-r $f;
44 my $dbh = dbh_new($f, $writable);
45 my $self = bless { dbh => $dbh }, $class;
50 $self->created_at(time) unless $self->created_at;
56 # used to keep track of used numeric mappings for v2 reindex
59 my ($fh, $fn) = tempfile('msgmap-XXXXXXXX', EXLOCK => 0, TMPDIR => 1);
60 $self->{dbh}->sqlite_backup_to_file($fn);
61 my $tmp = ref($self)->new_file($fn, 1);
62 $tmp->{dbh}->do('PRAGMA synchronous = OFF');
63 $tmp->{tmp_name} = $fn; # SQLite won't work if unlinked, apparently
65 close $fh or die "failed to close $fn: $!";
69 # n.b. invoked directly by scripts/xhdr-num2mid
71 my ($self, $key, $value) = @_;
73 my $sql = 'SELECT val FROM meta WHERE key = ? LIMIT 1';
74 my $dbh = $self->{dbh};
76 defined $value or return $dbh->selectrow_array($sql, undef, $key);
78 $prev = $dbh->selectrow_array($sql, undef, $key);
81 $sql = 'UPDATE meta SET val = ? WHERE key = ? LIMIT 1';
82 $dbh->do($sql, undef, $value, $key);
84 $sql = 'INSERT INTO meta (key,val) VALUES (?,?)';
85 $dbh->do($sql, undef, $key, $value);
91 my ($self, $commit) = @_;
92 $self->meta_accessor('last_commit', $commit);
95 # v2 uses this to keep track of how up-to-date Xapian is
96 # old versions may be automatically GC'ed away in the future,
97 # but it's a trivial amount of storage.
99 my ($self, $version, $i, $commit) = @_;
100 $self->meta_accessor("last_xap$version-$i", $commit);
104 my ($self, $second) = @_;
105 $self->meta_accessor('created_at', $second);
109 my ($self, $mid) = @_;
110 my $dbh = $self->{dbh};
111 my $sql = 'INSERT OR IGNORE INTO msgmap (mid) VALUES (?)';
112 my $sth = $self->{mid_insert} ||= $dbh->prepare($sql);
113 $sth->bind_param(1, $mid);
114 return if $sth->execute == 0;
115 $dbh->last_insert_id(undef, undef, 'msgmap', 'num');
119 my ($self, $num) = @_;
120 my $dbh = $self->{dbh};
121 my $sth = $self->{mid_for} ||=
122 $dbh->prepare('SELECT mid FROM msgmap WHERE num = ? LIMIT 1');
123 $sth->bind_param(1, $num);
125 $sth->fetchrow_array;
129 my ($self, $mid) = @_;
130 my $dbh = $self->{dbh};
131 my $sth = $self->{num_for} ||=
132 $dbh->prepare('SELECT num FROM msgmap WHERE mid = ? LIMIT 1');
133 $sth->bind_param(1, $mid);
135 $sth->fetchrow_array;
140 my $dbh = $self->{dbh};
141 # breaking MIN and MAX into separate queries speeds up from 250ms
142 # to around 700us with 2.7million messages.
143 my $sth = $dbh->prepare_cached('SELECT MIN(num) FROM msgmap', undef, 1);
145 my $min = $sth->fetchrow_array;
146 $sth = $dbh->prepare_cached('SELECT MAX(num) FROM msgmap', undef, 1);
148 ($min, $sth->fetchrow_array);
152 my ($self, $pfx, $limit) = @_;
154 die "No prefix given" unless (defined $pfx && $pfx ne '');
155 $pfx =~ s/([%_])/\\$1/g;
159 $limit += 0; # force to integer
162 $self->{dbh}->selectcol_arrayref('SELECT mid FROM msgmap ' .
163 'WHERE mid LIKE ? ESCAPE ? ' .
164 "ORDER BY num DESC LIMIT $limit",
169 my ($self, $mid) = @_;
170 my $dbh = $self->{dbh};
171 my $sth = $dbh->prepare('DELETE FROM msgmap WHERE mid = ?');
172 $sth->bind_param(1, $mid);
177 my ($self, $num) = @_;
178 my $dbh = $self->{dbh};
179 my $sth = $dbh->prepare('DELETE FROM msgmap WHERE num = ?');
180 $sth->bind_param(1, $num);
188 $e = eval { $dbh->selectrow_array('EXPLAIN SELECT * FROM msgmap;') };
189 defined $e or $dbh->do('CREATE TABLE msgmap (' .
190 'num INTEGER PRIMARY KEY AUTOINCREMENT, '.
191 'mid VARCHAR(1000) NOT NULL, ' .
194 $e = eval { $dbh->selectrow_array('EXPLAIN SELECT * FROM meta') };
195 defined $e or $dbh->do('CREATE TABLE meta (' .
196 'key VARCHAR(32) PRIMARY KEY, '.
197 'val VARCHAR(255) NOT NULL)');
202 my ($self, $num) = @_;
203 my $ids = $self->{dbh}->selectcol_arrayref(<<'', undef, $$num);
204 SELECT num FROM msgmap WHERE num > ?
205 ORDER BY num ASC LIMIT 1000
207 $$num = $ids->[-1] if @$ids;
212 my ($self, $beg, $end) = @_;
213 my $dbh = $self->{dbh};
214 my $attr = { Columns => [] };
215 my $mids = $dbh->selectall_arrayref(<<'', $attr, $$beg, $end);
216 SELECT num,mid FROM msgmap WHERE num >= ? AND num <= ?
219 $$beg = $mids->[-1]->[0] + 1 if @$mids;
223 # only used for mapping external serial numbers (e.g. articles from gmane)
224 # see scripts/xhdr-num2mid or PublicInbox::Filter::RubyLang for usage
226 my ($self, $num, $mid) = @_;
227 my $sth = $self->{mid_set} ||= do {
228 $self->{dbh}->prepare(
229 'INSERT OR IGNORE INTO msgmap (num,mid) VALUES (?,?)');
231 $sth->execute($num, $mid);
237 my $f = delete $self->{tmp_name};
238 if (defined $f && $self->{pid} == $$) {
239 unlink $f or warn "failed to unlink $f: $!\n";
245 my $f = $self->{tmp_name} or die "not a temporary clone\n";
246 delete $self->{dbh} and die "tmp_clone dbh not prepared for parent";
247 my $dbh = $self->{dbh} = dbh_new($f, 1);
248 $dbh->do('PRAGMA synchronous = OFF');
253 my $f = $self->{tmp_name} or die "not a temporary clone\n";
254 $self->{pid} == $$ or
255 die "BUG: atfork_prepare not called from $self->{pid}\n";
256 $self->{dbh} or die "temporary clone not open\n";
257 # must clobber prepared statements
258 %$self = (tmp_name => $f, pid => $$);