1 # Copyright (C) 2021 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);
10 use PublicInbox::ContentHash qw(git_sha);
15 my $f = $self->{filename};
16 my $creat = $rw && !-s $f;
18 require PublicInbox::Spawn;
19 open my $fh, '+>>', $f or Carp::croak "open($f): $!";
20 PublicInbox::Spawn::nodatacow_fd(fileno($fh));
22 my $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
27 sqlite_use_immediate_transaction => 1,
29 # no sqlite_unicode, here, all strings are binary
30 create_tables($self, $dbh) if $rw;
31 $dbh->do('PRAGMA journal_mode = WAL') if $creat;
32 $dbh->do('PRAGMA case_sensitive_like = ON');
41 lock_path => "$f.flock",
45 sub lms_write_prepare { ($_[0]->{dbh} //= dbh_new($_[0], 1)); $_[0] }
54 my ($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 $TYPE:/pathname */
65 CREATE TABLE IF NOT EXISTS blob2num (
66 oidbin VARBINARY NOT NULL,
67 fid INTEGER NOT NULL, /* folder ID */
68 uid INTEGER NOT NULL, /* NNTP article number, IMAP UID, MH number */
69 /* not UNIQUE(fid, uid), since we may have broken servers */
70 UNIQUE (oidbin, fid, uid)
73 # speeds up LeiImport->ck_update_kw (for "lei import") by 5-6x:
75 CREATE INDEX IF NOT EXISTS idx_fid_uid ON blob2num(fid,uid)
78 CREATE TABLE IF NOT EXISTS blob2name (
79 oidbin VARBINARY NOT NULL,
80 fid INTEGER NOT NULL, /* folder ID */
81 name VARBINARY NOT NULL, /* Maildir basename, JMAP blobId */
82 /* not UNIQUE(fid, name), since we may have broken software */
83 UNIQUE (oidbin, fid, name)
86 # speeds up LeiImport->pmdir_cb (for "lei import") by ~6x:
88 CREATE INDEX IF NOT EXISTS idx_fid_name ON blob2name(fid,name)
93 my ($self, $folder, $rw) = @_;
94 my $dbh = $self->{dbh} //= dbh_new($self, $rw);
95 my $sel = 'SELECT fid FROM folders WHERE loc = ? LIMIT 1';
96 my ($fid) = $dbh->selectrow_array($sel, undef, $folder);
97 return $fid if defined $fid;
99 # caller had trailing slash (LeiToMail)
100 if ($folder =~ s!\A((?:maildir|mh):.*?)/+\z!$1!i) {
101 ($fid) = $dbh->selectrow_array($sel, undef, $folder);
103 $dbh->do(<<EOM, undef, $folder, $fid) if $rw;
104 UPDATE folders SET loc = ? WHERE fid = ?
108 # sometimes we stored trailing slash..
109 } elsif ($folder =~ m!\A(?:maildir|mh):!i) {
110 ($fid) = $dbh->selectrow_array($sel, undef, "$folder/");
112 $dbh->do(<<EOM, undef, $folder, $fid) if $rw;
113 UPDATE folders SET loc = ? WHERE fid = ?
117 } elsif ($rw && $folder =~ m!\Aimaps?://!i) {
118 require PublicInbox::URIimap;
119 PublicInbox::URIimap->new($folder)->uidvalidity //
120 Carp::croak("BUG: $folder has no UIDVALIDITY");
124 ($fid) = $dbh->selectrow_array('SELECT MAX(fid) FROM folders');
127 # in case we're reusing, clobber existing stale refs:
128 $dbh->do('DELETE FROM blob2name WHERE fid = ?', undef, $fid);
129 $dbh->do('DELETE FROM blob2num WHERE fid = ?', undef, $fid);
131 my $sth = $dbh->prepare('INSERT INTO folders (fid, loc) VALUES (?, ?)');
132 $sth->execute($fid, $folder);
138 my ($self, @folders) = @_;
139 my $lk = $self->lock_for_scope;
140 for my $f (@folders) { $self->{fmap}->{$f} //= fid_for($self, $f, 1) }
144 my ($self, $oidbin, $folder, $id) = @_;
145 my $lk = $self->lock_for_scope;
146 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
148 if (ref($id)) { # scalar name
150 $sth = $self->{dbh}->prepare_cached(<<'');
151 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
153 } else { # numeric ID (IMAP UID, MH number)
154 $sth = $self->{dbh}->prepare_cached(<<'');
155 INSERT OR IGNORE INTO blob2num (oidbin, fid, uid) VALUES (?, ?, ?)
158 $sth->execute($oidbin, $fid, $id);
162 my ($self, $folder, $id) = @_;
163 my $lk = $self->lock_for_scope;
164 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
166 if (ref($id)) { # scalar name
168 $sth = $self->{dbh}->prepare_cached(<<'');
169 DELETE FROM blob2name WHERE fid = ? AND name = ?
172 $sth = $self->{dbh}->prepare_cached(<<'');
173 DELETE FROM blob2num WHERE fid = ? AND uid = ?
176 $sth->execute($fid, $id);
181 my ($self, $folder, $oidbin, $id, $newbn) = @_;
182 my $lk = $self->lock_for_scope;
183 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
184 my $sth = $self->{dbh}->prepare_cached(<<'');
185 UPDATE blob2name SET name = ? WHERE fid = ? AND oidbin = ? AND name = ?
187 my $nr = $sth->execute($newbn, $fid, $oidbin, $$id);
188 if ($nr == 0) { # may race with a clear_src, ensure new value exists
189 $sth = $self->{dbh}->prepare_cached(<<'');
190 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
192 $sth->execute($oidbin, $fid, $newbn);
196 # read-only, iterates every oidbin + UID or name for a given folder
198 my ($self, $folder, $cb, @args) = @_;
199 my $dbh = $self->{dbh} //= dbh_new($self);
201 if (ref($folder) eq 'HASH') {
202 $fid = $folder->{fid} // die "BUG: no `fid'";
204 $fid = $self->{fmap}->{$folder} //=
205 fid_for($self, $folder) // return;
208 # minimize implicit txn time to avoid blocking writers by
209 # batching SELECTs. This looks wonky but is necessary since
210 # $cb-> may access the DB on its own.
211 my $ary = $dbh->selectall_arrayref(<<'', undef, $fid);
212 SELECT _rowid_,oidbin,uid FROM blob2num WHERE fid = ?
213 ORDER BY _rowid_ ASC LIMIT 1000
215 my $min = @$ary ? $ary->[-1]->[0] : undef;
216 while (defined $min) {
217 for my $row (@$ary) { $cb->($row->[1], $row->[2], @args) }
219 $ary = $dbh->selectall_arrayref(<<'', undef, $fid, $min);
220 SELECT _rowid_,oidbin,uid FROM blob2num WHERE fid = ? AND _rowid_ > ?
221 ORDER BY _rowid_ ASC LIMIT 1000
223 $min = @$ary ? $ary->[-1]->[0] : undef;
226 $ary = $dbh->selectall_arrayref(<<'', undef, $fid);
227 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ?
228 ORDER BY _rowid_ ASC LIMIT 1000
230 $min = @$ary ? $ary->[-1]->[0] : undef;
231 while (defined $min) {
232 for my $row (@$ary) { $cb->($row->[1], \($row->[2]), @args) }
234 $ary = $dbh->selectall_arrayref(<<'', undef, $fid, $min);
235 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ? AND _rowid_ > ?
236 ORDER BY _rowid_ ASC LIMIT 1000
238 $min = @$ary ? $ary->[-1]->[0] : undef;
243 my ($self, $folder) = @_;
244 my $dbh = $self->{dbh} //= dbh_new($self);
247 $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder) // return;
248 my ($row) = $dbh->selectrow_array(<<"", undef, $fid);
249 SELECT COUNT(name) FROM blob2name WHERE fid = ?
251 $ret->{'name.count'} = $row if $row;
252 my $ntype = ($folder =~ m!\A(?:nntps?|s?news)://!i) ? 'article' :
253 (($folder =~ m!\Aimaps?://!i) ? 'uid' : "TODO<$folder>");
254 for my $op (qw(count min max)) {
255 ($row) = $dbh->selectrow_array(<<"", undef, $fid);
256 SELECT $op(uid) FROM blob2num WHERE fid = ?
259 $ret->{"$ntype.$op"} = $row;
264 # returns a { location => [ list-of-ids-or-names ] } mapping
266 my ($self, $oidbin) = @_;
267 my ($fid, $sth, $id, %fid2id);
268 my $dbh = $self->{dbh} //= dbh_new($self);
269 $sth = $dbh->prepare('SELECT fid,uid FROM blob2num WHERE oidbin = ?');
270 $sth->execute($oidbin);
271 while (my ($fid, $uid) = $sth->fetchrow_array) {
272 push @{$fid2id{$fid}}, $uid;
274 $sth = $dbh->prepare('SELECT fid,name FROM blob2name WHERE oidbin = ?');
275 $sth->execute($oidbin);
276 while (my ($fid, $name) = $sth->fetchrow_array) {
277 push @{$fid2id{$fid}}, $name;
279 $sth = $dbh->prepare('SELECT loc FROM folders WHERE fid = ? LIMIT 1');
281 while (my ($fid, $ids) = each %fid2id) {
283 my ($loc) = $sth->fetchrow_array;
284 unless (defined $loc) {
285 my $oidhex = unpack('H*', $oidbin);
286 warn "E: fid=$fid for $oidhex unknown:\n", map {
287 'E: '.(ref() ? $$_ : "#$_")."\n";
293 scalar(keys %$ret) ? $ret : undef;
296 # returns a list of folders used for completion
298 my ($self, $pfx) = @_;
299 my $dbh = $self->{dbh} //= dbh_new($self);
300 my $sql = 'SELECT loc FROM folders';
303 $sql .= ' WHERE loc LIKE ? ESCAPE ?';
305 $pfx[0] =~ s/([%_\\])/\\$1/g; # glob chars
308 map { $_->[0] } @{$dbh->selectall_arrayref($sql, undef, @pfx)};
312 my ($self, $oidhex, $vrfy) = @_;
313 my $dbh = $self->{dbh} //= dbh_new($self);
314 my $b2n = $dbh->prepare(<<'');
315 SELECT f.loc,b.name FROM blob2name b
316 LEFT JOIN folders f ON b.fid = f.fid
319 $b2n->execute(pack('H*', $oidhex));
320 while (my ($d, $n) = $b2n->fetchrow_array) {
321 substr($d, 0, length('maildir:')) = '';
322 # n.b. both mbsync and offlineimap use ":2," as a suffix
323 # in "new/", despite (from what I understand of reading
324 # <https://cr.yp.to/proto/maildir.html>), the ":2," only
325 # applies to files in "cur/".
326 my @try = $n =~ /:2,[a-zA-Z]+\z/ ? qw(cur new) : qw(new cur);
329 open my $fh, '<', $f or next;
330 # some (buggy) Maildir writers are non-atomic:
334 if ($vrfy && git_sha(1, \$raw)->hexdigest ne $oidhex) {
335 warn "$f changed $oidhex\n";
345 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
346 $all //= [ $self->folders ];
347 require PublicInbox::URIimap;
348 my $want = PublicInbox::URIimap->new($url)->canonical;
349 my ($s, $h, $mb) = ($want->scheme, $want->host, $want->mailbox);
350 my @uri = map { PublicInbox::URIimap->new($_)->canonical }
351 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$mb\E\b!, @$all);
354 next if $x->mailbox ne $want->mailbox;
355 next if $x->host ne $want->host;
356 next if $x->port != $want->port;
357 my $x_uidval = $x->uidvalidity;
358 next if ($want->uidvalidity // $x_uidval) != $x_uidval;
360 # allow nothing in want to possibly match ";AUTH=ANONYMOUS"
361 if (defined($x->auth) && !defined($want->auth) &&
362 !defined($want->user)) {
364 # or maybe user was forgotten on CLI:
365 } elsif (defined($x->user) && !defined($want->user)) {
367 } elsif (($x->user//"\0") eq ($want->user//"\0")) {
371 return @match if wantarray;
372 scalar(@match) <= 1 ? $match[0] :
373 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
376 sub match_nntp_url ($$$) {
377 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
378 $all //= [ $self->folders ];
379 require PublicInbox::URInntps;
380 my $want = PublicInbox::URInntps->new($url)->canonical;
381 my ($s, $h, $p) = ($want->scheme, $want->host, $want->port);
382 my $ng = $want->group; # force scalar (no article ranges)
383 my @uri = map { PublicInbox::URInntps->new($_)->canonical }
384 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$ng\E\b!, @$all);
387 next if $x->group ne $ng || $x->host ne $h || $x->port != $p;
388 # maybe user was forgotten on CLI:
389 if (defined($x->userinfo) && !defined($want->userinfo)) {
391 } elsif (($x->userinfo//"\0") eq ($want->userinfo//"\0")) {
395 return @match if wantarray;
396 scalar(@match) <= 1 ? $match[0] :
397 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
400 # returns undef on failure, number on success
402 my ($self, $lei, $all, $folders) = @_;
403 return $lei->fail(<<EOM) if @$folders;
404 --all= not compatible with @$folders on command-line
406 my %x = map { $_ => $_ } split(/,/, $all);
407 my @ok = grep(defined, delete(@x{qw(local remote), ''}));
408 push(@ok, '') if $all eq '';
411 @no = (join(',', @no));
412 return $lei->fail(<<EOM);
413 --all=@no not accepted (must be `local' and/or `remote')
417 my @all = $self->folders;
419 if ($ok eq 'local') {
420 @inc = grep(!m!\A[a-z0-9\+]+://!i, @all);
421 } elsif ($ok eq 'remote') {
422 @inc = grep(m!\A[a-z0-9\+]+://!i, @all);
423 } elsif ($ok ne '') {
424 return $lei->fail("--all=$all not understood");
428 push(@$folders, (grep { !$seen{$_}++ } @inc));
430 scalar(@$folders) || $lei->fail(<<EOM);
431 no --mail-sync folders known to lei
435 # map CLI args to folder table entries, returns undef on failure
437 my ($self, $lei, $folders) = @_;
438 my @all = $self->folders;
439 my %all = map { $_ => 1 } @all;
442 next if $all{$_}; # ok
443 if (m!\A(maildir|mh):(.+)!i) {
445 my $d = "$type:".$lei->abs_path($2);
446 push(@no, $_) unless $all{$d};
448 } elsif (-d "$_/new" && -d "$_/cur") {
449 my $d = 'maildir:'.$lei->abs_path($_);
450 push(@no, $_) unless $all{$d};
452 } elsif (m!\Aimaps?://!i) {
454 my $res = match_imap_url($self, $orig, \@all);
457 push(@{$err->{qerr}}, <<EOM);
458 # using `$res' instead of `$orig'
461 $lei->err($res) if defined $res;
464 } elsif (m!\A(?:nntps?|s?news)://!i) {
466 my $res = match_nntp_url($self, $orig, \@all);
469 push(@{$err->{qerr}}, <<EOM);
470 # using `$res' instead of `$orig'
473 $lei->err($res) if defined $res;
481 my $no = join("\n\t", @no);
482 $err->{fail} = <<EOF;
483 No sync information for: $no
484 Run `lei ls-mail-sync' to display valid choices
491 my ($self, @folders) = @_;
492 my $lk = $self->lock_for_scope;
493 for my $folder (@folders) {
494 my $fid = delete($self->{fmap}->{$folder}) //
495 fid_for($self, $folder) // next;
496 for my $t (qw(blob2name blob2num folders)) {
497 $self->{dbh}->do("DELETE FROM $t WHERE fid = ?",
503 # only used for changing canonicalization errors
505 my ($self, $old, $new) = @_;
506 my $lk = $self->lock_for_scope;
507 my $ofid = delete($self->{fmap}->{$old}) //
508 fid_for($self, $old) // return;
510 $self->{dbh}->do(<<EOM, undef, $new, $ofid);
511 UPDATE folders SET loc = ? WHERE fid = ?
514 if ($@ =~ /\bunique\b/i) {
515 my $nfid = $self->{fmap}->{$new} // fid_for($self, $new);
516 for my $t (qw(blob2name blob2num)) {
517 $self->{dbh}->do(<<EOM, undef, $nfid, $ofid);
518 UPDATE OR REPLACE $t SET fid = ? WHERE fid = ?
521 $self->{dbh}->do(<<EOM, undef, $ofid);
522 DELETE FROM folders WHERE fid = ?
527 sub num_oidbin ($$$) {
528 my ($self, $url, $uid) = @_; # $url MUST have UIDVALIDITY if IMAP
529 my $fid = $self->{fmap}->{$url} //= fid_for($self, $url) // return ();
530 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
531 SELECT oidbin FROM blob2num WHERE fid = ? AND uid = ? ORDER BY _rowid_
533 $sth->execute($fid, $uid);
534 map { $_->[0] } @{$sth->fetchall_arrayref};
537 sub name_oidbin ($$$) {
538 my ($self, $mdir, $nm) = @_;
539 my $fid = $self->{fmap}->{$mdir} //= fid_for($self, $mdir) // return;
540 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
541 SELECT oidbin FROM blob2name WHERE fid = ? AND name = ?
543 $sth->execute($fid, $nm);
544 map { $_->[0] } @{$sth->fetchall_arrayref};
548 my ($self, $lei, $uid_uri) = @_;
549 my $mailbox_uri = $uid_uri->clone;
550 $mailbox_uri->uid(undef);
551 my $folders = [ $$mailbox_uri ];
552 if (my $err = $self->arg2folder($lei, $folders)) {
554 $lei->qerr("# no sync information for $mailbox_uri");
556 $lei->qerr(@{$err->{qerr}}) if $err->{qerr};
558 map { unpack('H*',$_) } num_oidbin($self, $folders->[0], $uid_uri->uid)