1 # Copyright (C) all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # for maintaining synchronization between lei/store <=> Maildir|MH|IMAP|JMAP
5 package PublicInbox::LeiMailSync;
8 use parent qw(PublicInbox::Lock);
9 use DBI qw(:sql_types); # SQL_BLOB
10 use PublicInbox::ContentHash qw(git_sha);
15 my $f = $self->{filename};
18 require PublicInbox::Syscall;
19 open my $fh, '+>>', $f or Carp::croak "open($f): $!";
20 PublicInbox::Syscall::nodatacow_fh($fh);
22 my $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
26 sqlite_use_immediate_transaction => 1,
28 # no sqlite_unicode, here, all strings are binary
29 create_tables($self, $dbh);
30 $dbh->do('PRAGMA journal_mode = WAL') if $creat;
31 $dbh->do('PRAGMA case_sensitive_like = ON');
40 lock_path => "$f.flock",
44 sub lms_write_prepare { ($_[0]->{dbh} //= dbh_new($_[0])); $_[0] }
49 my $dbh = delete $self->{dbh};
50 eval { $dbh->do('PRAGMA optimize') } if $dbh;
54 my ($self, $dbh) = @_;
55 my $lk = $self->lock_for_scope;
58 CREATE TABLE IF NOT EXISTS folders (
59 fid INTEGER PRIMARY KEY,
60 loc VARBINARY NOT NULL, /* URL;UIDVALIDITY=$N or $TYPE:/pathname */
65 CREATE TABLE IF NOT EXISTS blob2num (
66 oidbin VARBINARY NOT NULL,
67 fid INTEGER NOT NULL, /* folder ID */
68 uid INTEGER NOT NULL, /* NNTP article number, IMAP UID, MH number */
69 /* not UNIQUE(fid, uid), since we may have broken servers */
70 UNIQUE (oidbin, fid, uid)
73 # speeds up LeiImport->ck_update_kw (for "lei import") by 5-6x:
75 CREATE INDEX IF NOT EXISTS idx_fid_uid ON blob2num(fid,uid)
78 CREATE TABLE IF NOT EXISTS blob2name (
79 oidbin VARBINARY NOT NULL,
80 fid INTEGER NOT NULL, /* folder ID */
81 name VARBINARY NOT NULL, /* Maildir basename, JMAP blobId */
82 /* not UNIQUE(fid, name), since we may have broken software */
83 UNIQUE (oidbin, fid, name)
86 # speeds up LeiImport->pmdir_cb (for "lei import") by ~6x:
88 CREATE INDEX IF NOT EXISTS idx_fid_name ON blob2name(fid,name)
92 # used to fixup pre-1.7.0 folders
93 sub update_fid ($$$) {
94 my ($dbh, $fid, $loc) = @_;
95 my $sth = $dbh->prepare(<<'');
96 UPDATE folders SET loc = ? WHERE fid = ?
98 $sth->bind_param(1, $loc, SQL_BLOB);
99 $sth->bind_param(2, $fid);
104 my ($sth, $folder, $dbh) = @_;
105 $sth->bind_param(1, $folder, SQL_BLOB);
107 my ($fid) = $sth->fetchrow_array;
108 if (defined $fid) { # for downgrade+upgrade (1.8 -> 1.7 -> 1.8)
109 $dbh->do('DELETE FROM folders WHERE loc = ? AND fid != ?',
110 undef, $folder, $fid) if defined($dbh);
112 $sth->execute($folder); # fixup old stuff
113 ($fid) = $sth->fetchrow_array;
114 update_fid($dbh, $fid, $folder) if defined($fid) && $dbh;
120 my ($self, $folder, $creat) = @_;
121 my $dbh = $self->{dbh} //= dbh_new($self);
122 my $sth = $dbh->prepare_cached(<<'', undef, 1);
123 SELECT fid FROM folders WHERE loc = ? LIMIT 1
125 my $fid = get_fid($sth, $folder, $dbh);
126 return $fid if defined($fid);
128 # caller had trailing slash (LeiToMail)
129 if ($folder =~ s!\A((?:maildir|mh):.*?)/+\z!$1!i) {
130 $fid = get_fid($sth, $folder, $dbh);
132 update_fid($dbh, $fid, $folder);
135 # sometimes we stored trailing slash..
136 } elsif ($folder =~ m!\A(?:maildir|mh):!i) {
137 $fid = get_fid($sth, $folder, $dbh);
139 update_fid($dbh, $fid, $folder);
142 } elsif ($creat && $folder =~ m!\Aimaps?://!i) {
143 require PublicInbox::URIimap;
144 my $uri = PublicInbox::URIimap->new($folder);
146 Carp::croak("BUG: $folder has no UIDVALIDITY");
147 defined($uri->uid) and Carp::confess("BUG: $folder has UID");
150 return unless $creat;
151 ($fid) = $dbh->selectrow_array('SELECT MAX(fid) FROM folders');
154 # in case we're reusing, clobber existing stale refs:
155 $dbh->do('DELETE FROM blob2name WHERE fid = ?', undef, $fid);
156 $dbh->do('DELETE FROM blob2num WHERE fid = ?', undef, $fid);
158 $sth = $dbh->prepare('INSERT INTO folders (fid, loc) VALUES (?, ?)');
159 $sth->bind_param(1, $fid);
160 $sth->bind_param(2, $folder, SQL_BLOB);
167 my ($self, @folders) = @_;
168 my $lk = $self->lock_for_scope;
169 for my $f (@folders) { $self->{fmap}->{$f} //= fid_for($self, $f, 1) }
173 my ($self, $oidbin, $folder, $id) = @_;
174 my $lk = $self->lock_for_scope;
175 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
176 my $dbh = $self->{dbh};
177 my ($sth, @param3, $del_old);
178 if (ref($id)) { # scalar name
179 @param3 = ($$id, SQL_BLOB);
180 $sth = $dbh->prepare_cached(<<'');
181 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
183 $del_old = $dbh->prepare_cached(<<'');
184 DELETE FROM blob2name WHERE oidbin = ? AND fid = ? AND name = ?
186 } else { # numeric ID (IMAP UID, MH number)
188 $sth = $dbh->prepare_cached(<<'');
189 INSERT OR IGNORE INTO blob2num (oidbin, fid, uid) VALUES (?, ?, ?)
191 $del_old = $dbh->prepare_cached(<<'');
192 DELETE FROM blob2num WHERE oidbin = ? AND fid = ? AND uid = ?
195 $sth->bind_param(1, $oidbin, SQL_BLOB);
196 $sth->bind_param(2, $fid);
197 $sth->bind_param(3, @param3);
198 my $ret = $sth->execute;
199 $del_old->execute($oidbin, $fid, $param3[0]);
204 my ($self, $folder, $id) = @_;
205 my $lk = $self->lock_for_scope;
206 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
208 if (ref($id)) { # scalar name
209 @param3 = ($$id, SQL_BLOB);
210 $sth = $self->{dbh}->prepare_cached(<<'');
211 DELETE FROM blob2name WHERE fid = ? AND name = ?
215 $sth = $self->{dbh}->prepare_cached(<<'');
216 DELETE FROM blob2num WHERE fid = ? AND uid = ?
219 $sth->bind_param(1, $fid);
220 $sth->bind_param(2, @param3);
221 my $ret = $sth->execute;
223 # older versions may not have used SQL_BLOB:
224 if (defined($ret) && $ret == 0 && scalar(@param3) == 2) {
225 $sth->bind_param(1, $fid);
226 $sth->bind_param(2, $param3[0]);
227 $ret = $sth->execute;
234 my ($self, $folder, $oidbin, $id, $newbn) = @_;
235 my $lk = $self->lock_for_scope;
236 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
237 $self->{dbh}->begin_work;
238 my $sth = $self->{dbh}->prepare_cached(<<'');
239 UPDATE blob2name SET name = ? WHERE fid = ? AND oidbin = ? AND name = ?
241 # eval since unique constraint may fail due to race
242 $sth->bind_param(1, $newbn, SQL_BLOB);
243 $sth->bind_param(2, $fid);
244 $sth->bind_param(3, $oidbin, SQL_BLOB);
245 $sth->bind_param(4, $$id, SQL_BLOB);
246 my $nr = eval { $sth->execute };
247 if (!defined($nr) || $nr == 0) { # $nr may be `0E0'
248 # delete from old, pre-SQL_BLOB rows:
249 my $del_old = $self->{dbh}->prepare_cached(<<'');
250 DELETE FROM blob2name WHERE fid = ? AND oidbin = ? AND name = ?
252 $del_old->execute($fid, $oidbin, $$id); # missing-OK
253 $del_old->execute($fid, $oidbin, $newbn); # ditto
255 # may race with a clear_src, ensure new value exists
256 $sth = $self->{dbh}->prepare_cached(<<'');
257 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
259 $sth->bind_param(1, $oidbin, SQL_BLOB);
260 $sth->bind_param(2, $fid);
261 $sth->bind_param(3, $newbn, SQL_BLOB);
264 $self->{dbh}->commit;
267 # read-only, iterates every oidbin + UID or name for a given folder
269 my ($self, $folder, $cb, @args) = @_;
270 my $dbh = $self->{dbh} //= dbh_new($self);
273 if (ref($folder) eq 'HASH') {
274 $fid = $folder->{fid} // die "BUG: no `fid'";
275 @rng = grep(defined, @$folder{qw(min max)});
276 $and_ge_le = 'AND uid >= ? AND uid <= ?' if @rng;
278 $fid = $self->{fmap}->{$folder} //=
279 fid_for($self, $folder) // return;
282 # minimize implicit txn time to avoid blocking writers by
283 # batching SELECTs. This looks wonky but is necessary since
284 # $cb-> may access the DB on its own.
285 my $ary = $dbh->selectall_arrayref(<<"", undef, $fid, @rng);
286 SELECT _rowid_,oidbin,uid FROM blob2num WHERE fid = ? $and_ge_le
287 ORDER BY _rowid_ ASC LIMIT 1000
289 my $min = @$ary ? $ary->[-1]->[0] : undef;
290 while (defined $min) {
291 for my $row (@$ary) { $cb->($row->[1], $row->[2], @args) }
293 $ary = $dbh->selectall_arrayref(<<"", undef, $fid, @rng, $min);
294 SELECT _rowid_,oidbin,uid FROM blob2num
295 WHERE fid = ? $and_ge_le AND _rowid_ > ?
296 ORDER BY _rowid_ ASC LIMIT 1000
298 $min = @$ary ? $ary->[-1]->[0] : undef;
301 $ary = $dbh->selectall_arrayref(<<'', undef, $fid);
302 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ?
303 ORDER BY _rowid_ ASC LIMIT 1000
305 $min = @$ary ? $ary->[-1]->[0] : undef;
306 while (defined $min) {
307 for my $row (@$ary) { $cb->($row->[1], \($row->[2]), @args) }
309 $ary = $dbh->selectall_arrayref(<<'', undef, $fid, $min);
310 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ? AND _rowid_ > ?
311 ORDER BY _rowid_ ASC LIMIT 1000
313 $min = @$ary ? $ary->[-1]->[0] : undef;
318 my ($self, $folder) = @_;
319 my $dbh = $self->{dbh} //= dbh_new($self);
322 $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder) // return;
323 my ($row) = $dbh->selectrow_array(<<"", undef, $fid);
324 SELECT COUNT(name) FROM blob2name WHERE fid = ?
326 $ret->{'name.count'} = $row if $row;
327 my $ntype = ($folder =~ m!\A(?:nntps?|s?news)://!i) ? 'article' :
328 (($folder =~ m!\Aimaps?://!i) ? 'uid' : "TODO<$folder>");
329 for my $op (qw(count min max)) {
330 ($row) = $dbh->selectrow_array(<<"", undef, $fid);
331 SELECT $op(uid) FROM blob2num WHERE fid = ?
334 $ret->{"$ntype.$op"} = $row;
339 # returns a { location => [ list-of-ids-or-names ] } mapping
341 my ($self, $oidbin) = @_;
342 my ($fid, $sth, $id, %fid2id, %seen);
343 my $dbh = $self->{dbh} //= dbh_new($self);
344 $sth = $dbh->prepare('SELECT fid,uid FROM blob2num WHERE oidbin = ?');
345 $sth->bind_param(1, $oidbin, SQL_BLOB);
347 while (my ($fid, $uid) = $sth->fetchrow_array) {
348 push @{$fid2id{$fid}}, $uid;
349 $seen{"$uid.$fid"} = 1;
352 # deal with 1.7.0 DBs :<
353 $sth->execute($oidbin);
354 while (my ($fid, $uid) = $sth->fetchrow_array) {
355 next if $seen{"$uid.$fid"};
356 push @{$fid2id{$fid}}, $uid;
360 $sth = $dbh->prepare('SELECT fid,name FROM blob2name WHERE oidbin = ?');
361 $sth->bind_param(1, $oidbin, SQL_BLOB);
363 while (my ($fid, $name) = $sth->fetchrow_array) {
364 push @{$fid2id{$fid}}, $name;
365 $seen{"$fid.$name"} = 1;
368 # deal with 1.7.0 DBs :<
369 $sth->execute($oidbin);
370 while (my ($fid, $name) = $sth->fetchrow_array) {
371 next if $seen{"$fid.$name"};
372 push @{$fid2id{$fid}}, $name;
375 $sth = $dbh->prepare('SELECT loc FROM folders WHERE fid = ? LIMIT 1');
377 while (my ($fid, $ids) = each %fid2id) {
379 my ($loc) = $sth->fetchrow_array;
380 unless (defined $loc) {
381 my $oidhex = unpack('H*', $oidbin);
382 warn "E: fid=$fid for $oidhex unknown:\n", map {
383 'E: '.(ref() ? $$_ : "#$_")."\n";
389 scalar(keys %$ret) ? $ret : undef;
392 # returns a list of folders used for completion
394 my ($self, @pfx) = @_;
395 my $sql = 'SELECT loc FROM folders';
397 if (defined($pfx[0])) {
398 $sql .= ' WHERE loc REGEXP ?'; # DBD::SQLite uses perlre
399 $re = !!$pfx[1] ? '.*' : '';
400 $re .= quotemeta($pfx[0]);
403 my $sth = ($self->{dbh} //= dbh_new($self))->prepare($sql);
404 $sth->bind_param(1, $re) if defined($re);
406 map { $_->[0] } @{$sth->fetchall_arrayref};
410 my ($self, $oidhex, $vrfy) = @_;
411 my $dbh = $self->{dbh} //= dbh_new($self);
412 my $b2n = $dbh->prepare(<<'');
413 SELECT f.loc,b.name FROM blob2name b
414 LEFT JOIN folders f ON b.fid = f.fid
417 $b2n->bind_param(1, pack('H*', $oidhex), SQL_BLOB);
419 while (my ($d, $n) = $b2n->fetchrow_array) {
420 substr($d, 0, length('maildir:')) = '';
421 # n.b. both mbsync and offlineimap use ":2," as a suffix
422 # in "new/", despite (from what I understand of reading
423 # <https://cr.yp.to/proto/maildir.html>), the ":2," only
424 # applies to files in "cur/".
425 my @try = $n =~ /:2,[a-zA-Z]+\z/ ? qw(cur new) : qw(new cur);
428 open my $fh, '<', $f or next;
429 # some (buggy) Maildir writers are non-atomic:
434 my $got = git_sha(1, \$raw)->hexdigest;
435 if ($got ne $oidhex) {
436 warn "$f changed $oidhex => $got\n";
447 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
448 $all //= [ $self->folders ];
449 require PublicInbox::URIimap;
450 my $want = PublicInbox::URIimap->new($url)->canonical;
451 my ($s, $h, $mb) = ($want->scheme, $want->host, $want->mailbox);
452 my @uri = map { PublicInbox::URIimap->new($_)->canonical }
453 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$mb\E\b!, @$all);
456 next if $x->mailbox ne $want->mailbox;
457 next if $x->host ne $want->host;
458 next if $x->port != $want->port;
459 my $x_uidval = $x->uidvalidity;
460 next if ($want->uidvalidity // $x_uidval) != $x_uidval;
462 # allow nothing in want to possibly match ";AUTH=ANONYMOUS"
463 if (defined($x->auth) && !defined($want->auth) &&
464 !defined($want->user)) {
466 # or maybe user was forgotten on CLI:
467 } elsif (defined($x->user) && !defined($want->user)) {
469 } elsif (($x->user//"\0") eq ($want->user//"\0")) {
473 return @match if wantarray;
474 scalar(@match) <= 1 ? $match[0] :
475 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
478 sub match_nntp_url ($$$) {
479 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
480 $all //= [ $self->folders ];
481 require PublicInbox::URInntps;
482 my $want = PublicInbox::URInntps->new($url)->canonical;
483 my ($s, $h, $p) = ($want->scheme, $want->host, $want->port);
484 my $ng = $want->group; # force scalar (no article ranges)
485 my @uri = map { PublicInbox::URInntps->new($_)->canonical }
486 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$ng\E\b!, @$all);
489 next if $x->group ne $ng || $x->host ne $h || $x->port != $p;
490 # maybe user was forgotten on CLI:
491 if (defined($x->userinfo) && !defined($want->userinfo)) {
493 } elsif (($x->userinfo//"\0") eq ($want->userinfo//"\0")) {
497 return @match if wantarray;
498 scalar(@match) <= 1 ? $match[0] :
499 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
502 # returns undef on failure, number on success
504 my ($self, $lei, $all, $folders) = @_;
505 return $lei->fail(<<EOM) if @$folders;
506 --all= not compatible with @$folders on command-line
508 my %x = map { $_ => $_ } split(/,/, $all);
509 my @ok = grep(defined, delete(@x{qw(local remote), ''}));
510 push(@ok, '') if $all eq '';
513 @no = (join(',', @no));
514 return $lei->fail(<<EOM);
515 --all=@no not accepted (must be `local' and/or `remote')
519 my @all = $self->folders;
521 if ($ok eq 'local') {
522 @inc = grep(!m!\A[a-z0-9\+]+://!i, @all);
523 } elsif ($ok eq 'remote') {
524 @inc = grep(m!\A[a-z0-9\+]+://!i, @all);
525 } elsif ($ok ne '') {
526 return $lei->fail("--all=$all not understood");
530 push(@$folders, (grep { !$seen{$_}++ } @inc));
532 scalar(@$folders) || $lei->fail(<<EOM);
533 no --mail-sync folders known to lei
537 # map CLI args to folder table entries, returns undef on failure
539 my ($self, $lei, $folders) = @_;
540 my @all = $self->folders;
541 my %all = map { $_ => 1 } @all;
544 next if $all{$_}; # ok
545 if (m!\A(maildir|mh):(.+)!i) {
547 my $d = "$type:".$lei->abs_path($2);
548 push(@no, $_) unless $all{$d};
550 } elsif (-d "$_/new" && -d "$_/cur") {
551 my $d = 'maildir:'.$lei->abs_path($_);
552 push(@no, $_) unless $all{$d};
554 } elsif (m!\Aimaps?://!i) {
556 my $res = match_imap_url($self, $orig, \@all);
560 # using `$res' instead of `$orig'
563 warn($res, "\n") if defined $res;
566 } elsif (m!\A(?:nntps?|s?news)://!i) {
568 my $res = match_nntp_url($self, $orig, \@all);
572 # using `$res' instead of `$orig'
575 warn($res, "\n") if defined $res;
583 my $no = join("\n\t", @no);
585 No sync information for: $no
586 Run `lei ls-mail-sync' to display valid choices
592 my ($self, @folders) = @_;
593 my $lk = $self->lock_for_scope;
594 for my $folder (@folders) {
595 my $fid = delete($self->{fmap}->{$folder}) //
596 fid_for($self, $folder) // next;
597 for my $t (qw(blob2name blob2num folders)) {
598 $self->{dbh}->do("DELETE FROM $t WHERE fid = ?",
604 # only used for changing canonicalization errors
606 my ($self, $old, $new) = @_;
607 my $lk = $self->lock_for_scope;
608 my $ofid = delete($self->{fmap}->{$old}) //
609 fid_for($self, $old) // return;
611 $self->{dbh}->do(<<EOM, undef, $new, $ofid);
612 UPDATE folders SET loc = ? WHERE fid = ?
615 if ($@ =~ /\bunique\b/i) {
616 my $nfid = $self->{fmap}->{$new} // fid_for($self, $new);
617 for my $t (qw(blob2name blob2num)) {
618 $self->{dbh}->do(<<EOM, undef, $nfid, $ofid);
619 UPDATE OR REPLACE $t SET fid = ? WHERE fid = ?
622 $self->{dbh}->do(<<EOM, undef, $ofid);
623 DELETE FROM folders WHERE fid = ?
628 sub num_oidbin ($$$) {
629 my ($self, $url, $uid) = @_; # $url MUST have UIDVALIDITY if IMAP
630 my $fid = $self->{fmap}->{$url} //= fid_for($self, $url) // return ();
631 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
632 SELECT oidbin FROM blob2num WHERE fid = ? AND uid = ? ORDER BY _rowid_
634 $sth->execute($fid, $uid);
635 my %uniq; # for public-inbox <= 1.7.0
636 grep { !$uniq{$_}++ } map { $_->[0] } @{$sth->fetchall_arrayref};
639 sub name_oidbin ($$$) {
640 my ($self, $mdir, $nm) = @_;
641 my $fid = $self->{fmap}->{$mdir} //= fid_for($self, $mdir) // return;
642 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
643 SELECT oidbin FROM blob2name WHERE fid = ? AND name = ?
645 $sth->bind_param(1, $fid);
646 $sth->bind_param(2, $nm, SQL_BLOB);
648 my @bin = map { $_->[0] } @{$sth->fetchall_arrayref};
649 $sth->execute($fid, $nm);
650 my @old = map { $_->[0] } @{$sth->fetchall_arrayref};
651 my %uniq; # for public-inbox <= 1.7.0
652 grep { !$uniq{$_}++ } (@bin, @old);
656 my ($self, $lei, $uid_uri) = @_;
657 my $mailbox_uri = $uid_uri->clone;
658 $mailbox_uri->uid(undef);
659 my $folders = [ $$mailbox_uri ];
660 eval { $self->arg2folder($lei, $folders) };
661 $lei->qerr("# no sync information for $mailbox_uri") if $@;
662 map { unpack('H*',$_) } num_oidbin($self, $folders->[0], $uid_uri->uid)