X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FMsgmap.pm;h=feef8ba79ab9a657d8135b0c65ca0186dd7bed51;hb=74b92712fa7a21fe504b9908edebcf11bb9dc170;hp=2583ff47422c4e1d7c3d9e36344c453ea891b9b5;hpb=3a988efc28064954e64419bb0d368684fa7c7f20;p=public-inbox.git diff --git a/lib/PublicInbox/Msgmap.pm b/lib/PublicInbox/Msgmap.pm index 2583ff47..feef8ba7 100644 --- a/lib/PublicInbox/Msgmap.pm +++ b/lib/PublicInbox/Msgmap.pm @@ -1,5 +1,5 @@ -# Copyright (C) 2015 all contributors -# License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt) +# Copyright (C) 2015-2018 all contributors +# License: AGPL-3.0+ # bidirectional Message-ID <-> Article Number mapping for the NNTP # and web interfaces. This is required for implementing stable article @@ -12,6 +12,7 @@ use strict; use warnings; use DBI; use DBD::SQLite; +use File::Temp qw(tempfile); sub new { my ($class, $git_dir, $writable) = @_; @@ -20,7 +21,11 @@ sub new { my $err = $!; -d $d or die "$d not created: $err"; } - my $f = "$d/msgmap.sqlite3"; + new_file($class, "$d/msgmap.sqlite3", $writable); +} + +sub dbh_new { + my ($f, $writable) = @_; my $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', { AutoCommit => 1, RaiseError => 1, @@ -29,6 +34,14 @@ sub new { sqlite_use_immediate_transaction => 1, }); $dbh->do('PRAGMA case_sensitive_like = ON'); + $dbh; +} + +sub new_file { + my ($class, $f, $writable) = @_; + return if !$writable && !-r $f; + + my $dbh = dbh_new($f, $writable); my $self = bless { dbh => $dbh }, $class; if ($writable) { @@ -40,25 +53,36 @@ sub new { $self; } +# used to keep track of used numeric mappings for v2 reindex +sub tmp_clone { + my ($self) = @_; + my ($fh, $fn) = tempfile('msgmap-XXXXXXXX', EXLOCK => 0, TMPDIR => 1); + $self->{dbh}->sqlite_backup_to_file($fn); + my $tmp = ref($self)->new_file($fn, 1); + $tmp->{dbh}->do('PRAGMA synchronous = OFF'); + $tmp->{tmp_name} = $fn; # SQLite won't work if unlinked, apparently + $tmp->{pid} = $$; + close $fh or die "failed to close $fn: $!"; + $tmp; +} + +# n.b. invoked directly by scripts/xhdr-num2mid sub meta_accessor { my ($self, $key, $value) = @_; - use constant { - meta_select => 'SELECT val FROM meta WHERE key = ? LIMIT 1', - meta_update => 'UPDATE meta SET val = ? WHERE key = ? LIMIT 1', - meta_insert => 'INSERT INTO meta (key,val) VALUES (?,?)', - }; + my $sql = 'SELECT val FROM meta WHERE key = ? LIMIT 1'; my $dbh = $self->{dbh}; my $prev; - defined $value or - return $dbh->selectrow_array(meta_select, undef, $key); + defined $value or return $dbh->selectrow_array($sql, undef, $key); - $prev = $dbh->selectrow_array(meta_select, undef, $key); + $prev = $dbh->selectrow_array($sql, undef, $key); if (defined $prev) { - $dbh->do(meta_update, undef, $value, $key); + $sql = 'UPDATE meta SET val = ? WHERE key = ? LIMIT 1'; + $dbh->do($sql, undef, $value, $key); } else { - $dbh->do(meta_insert, undef, $key, $value); + $sql = 'INSERT INTO meta (key,val) VALUES (?,?)'; + $dbh->do($sql, undef, $key, $value); } $prev; } @@ -68,6 +92,14 @@ sub last_commit { $self->meta_accessor('last_commit', $commit); } +# v2 uses this to keep track of how up-to-date Xapian is +# old versions may be automatically GC'ed away in the future, +# but it's a trivial amount of storage. +sub last_commit_xap { + my ($self, $version, $i, $commit) = @_; + $self->meta_accessor("last_xap$version-$i", $commit); +} + sub created_at { my ($self, $second) = @_; $self->meta_accessor('created_at', $second); @@ -76,18 +108,18 @@ sub created_at { sub mid_insert { my ($self, $mid) = @_; my $dbh = $self->{dbh}; - use constant MID_INSERT => 'INSERT INTO msgmap (mid) VALUES (?)'; - my $sth = $self->{mid_insert} ||= $dbh->prepare(MID_INSERT); + my $sql = 'INSERT OR IGNORE INTO msgmap (mid) VALUES (?)'; + my $sth = $self->{mid_insert} ||= $dbh->prepare($sql); $sth->bind_param(1, $mid); - $sth->execute; + return if $sth->execute == 0; $dbh->last_insert_id(undef, undef, 'msgmap', 'num'); } sub mid_for { my ($self, $num) = @_; my $dbh = $self->{dbh}; - use constant MID_FOR => 'SELECT mid FROM msgmap WHERE num = ? LIMIT 1'; - my $sth = $self->{mid_for} ||= $dbh->prepare(MID_FOR); + my $sth = $self->{mid_for} ||= + $dbh->prepare('SELECT mid FROM msgmap WHERE num = ? LIMIT 1'); $sth->bind_param(1, $num); $sth->execute; $sth->fetchrow_array; @@ -96,8 +128,8 @@ sub mid_for { sub num_for { my ($self, $mid) = @_; my $dbh = $self->{dbh}; - use constant NUM_FOR => 'SELECT num FROM msgmap WHERE mid = ? LIMIT 1'; - my $sth = $self->{num_for} ||= $dbh->prepare(NUM_FOR); + my $sth = $self->{num_for} ||= + $dbh->prepare('SELECT num FROM msgmap WHERE mid = ? LIMIT 1'); $sth->bind_param(1, $mid); $sth->execute; $sth->fetchrow_array; @@ -106,10 +138,14 @@ sub num_for { sub minmax { my ($self) = @_; my $dbh = $self->{dbh}; - use constant NUM_MINMAX => 'SELECT MIN(num),MAX(num) FROM msgmap'; - my $sth = $self->{num_minmax} ||= $dbh->prepare(NUM_MINMAX); + # breaking MIN and MAX into separate queries speeds up from 250ms + # to around 700us with 2.7million messages. + my $sth = $dbh->prepare_cached('SELECT MIN(num) FROM msgmap', undef, 1); + $sth->execute; + my $min = $sth->fetchrow_array; + $sth = $dbh->prepare_cached('SELECT MAX(num) FROM msgmap', undef, 1); $sth->execute; - $sth->fetchrow_array; + ($min, $sth->fetchrow_array); } sub mid_prefixes { @@ -132,12 +168,19 @@ sub mid_prefixes { sub mid_delete { my ($self, $mid) = @_; my $dbh = $self->{dbh}; - use constant MID_DELETE => 'DELETE FROM msgmap WHERE mid = ?'; - my $sth = $dbh->prepare(MID_DELETE); + my $sth = $dbh->prepare('DELETE FROM msgmap WHERE mid = ?'); $sth->bind_param(1, $mid); $sth->execute; } +sub num_delete { + my ($self, $num) = @_; + my $dbh = $self->{dbh}; + my $sth = $dbh->prepare('DELETE FROM msgmap WHERE num = ?'); + $sth->bind_param(1, $num); + $sth->execute; +} + sub create_tables { my ($dbh) = @_; my $e; @@ -154,17 +197,65 @@ sub create_tables { 'val VARCHAR(255) NOT NULL)'); } -sub id_batch { - my ($self, $num, $cb) = @_; +# used by NNTP.pm +sub ids_after { + my ($self, $num) = @_; + my $ids = $self->{dbh}->selectcol_arrayref(<<'', undef, $$num); +SELECT num FROM msgmap WHERE num > ? +ORDER BY num ASC LIMIT 1000 + + $$num = $ids->[-1] if @$ids; + $ids; +} + +sub msg_range { + my ($self, $beg, $end) = @_; my $dbh = $self->{dbh}; - my $sth = $dbh->prepare('SELECT num FROM msgmap WHERE num > ? '. - 'ORDER BY num ASC LIMIT 1000'); - $sth->execute($num); - my $ary = $sth->fetchall_arrayref; - @$ary = map { $_->[0] } @$ary; - my $nr = scalar @$ary; - $cb->($ary) if $nr; - $nr; + my $attr = { Columns => [] }; + my $mids = $dbh->selectall_arrayref(<<'', $attr, $$beg, $end); +SELECT num,mid FROM msgmap WHERE num >= ? AND num <= ? +ORDER BY num ASC + + $$beg = $mids->[-1]->[0] + 1 if @$mids; + $mids +} + +# only used for mapping external serial numbers (e.g. articles from gmane) +# see scripts/xhdr-num2mid or PublicInbox::Filter::RubyLang for usage +sub mid_set { + my ($self, $num, $mid) = @_; + my $sth = $self->{mid_set} ||= do { + $self->{dbh}->prepare( + 'INSERT OR IGNORE INTO msgmap (num,mid) VALUES (?,?)'); + }; + $sth->execute($num, $mid); +} + +sub DESTROY { + my ($self) = @_; + delete $self->{dbh}; + my $f = delete $self->{tmp_name}; + if (defined $f && $self->{pid} == $$) { + unlink $f or warn "failed to unlink $f: $!\n"; + } +} + +sub atfork_parent { + my ($self) = @_; + my $f = $self->{tmp_name} or die "not a temporary clone\n"; + delete $self->{dbh} and die "tmp_clone dbh not prepared for parent"; + my $dbh = $self->{dbh} = dbh_new($f, 1); + $dbh->do('PRAGMA synchronous = OFF'); +} + +sub atfork_prepare { + my ($self) = @_; + my $f = $self->{tmp_name} or die "not a temporary clone\n"; + $self->{pid} == $$ or + die "BUG: atfork_prepare not called from $self->{pid}\n"; + $self->{dbh} or die "temporary clone not open\n"; + # must clobber prepared statements + %$self = (tmp_name => $f, pid => $$); } 1;