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};
16 my $creat = $rw && !-s $f;
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",'','', {
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] }
50 my $dbh = delete $self->{dbh};
51 eval { $dbh->do('PRAGMA optimize') } if $dbh;
55 my ($self, $dbh) = @_;
56 my $lk = $self->lock_for_scope;
59 CREATE TABLE IF NOT EXISTS folders (
60 fid INTEGER PRIMARY KEY,
61 loc VARBINARY NOT NULL, /* URL;UIDVALIDITY=$N or $TYPE:/pathname */
66 CREATE TABLE IF NOT EXISTS blob2num (
67 oidbin VARBINARY NOT NULL,
68 fid INTEGER NOT NULL, /* folder ID */
69 uid INTEGER NOT NULL, /* NNTP article number, IMAP UID, MH number */
70 /* not UNIQUE(fid, uid), since we may have broken servers */
71 UNIQUE (oidbin, fid, uid)
74 # speeds up LeiImport->ck_update_kw (for "lei import") by 5-6x:
76 CREATE INDEX IF NOT EXISTS idx_fid_uid ON blob2num(fid,uid)
79 CREATE TABLE IF NOT EXISTS blob2name (
80 oidbin VARBINARY NOT NULL,
81 fid INTEGER NOT NULL, /* folder ID */
82 name VARBINARY NOT NULL, /* Maildir basename, JMAP blobId */
83 /* not UNIQUE(fid, name), since we may have broken software */
84 UNIQUE (oidbin, fid, name)
87 # speeds up LeiImport->pmdir_cb (for "lei import") by ~6x:
89 CREATE INDEX IF NOT EXISTS idx_fid_name ON blob2name(fid,name)
93 # used to fixup pre-1.7.0 folders
94 sub update_fid ($$$) {
95 my ($dbh, $fid, $loc) = @_;
96 my $sth = $dbh->prepare(<<'');
97 UPDATE folders SET loc = ? WHERE fid = ?
99 $sth->bind_param(1, $loc, SQL_BLOB);
100 $sth->bind_param(2, $fid);
105 my ($sth, $folder, $dbh) = @_; # $dbh is set iff RW
106 $sth->bind_param(1, $folder, SQL_BLOB);
108 my ($fid) = $sth->fetchrow_array;
109 if (defined $fid) { # for downgrade+upgrade (1.8 -> 1.7 -> 1.8)
110 $dbh->do('DELETE FROM folders WHERE loc = ? AND fid != ?',
111 undef, $folder, $fid) if defined($dbh);
113 $sth->execute($folder); # fixup old stuff
114 ($fid) = $sth->fetchrow_array;
115 update_fid($dbh, $fid, $folder) if defined($fid) && $dbh;
121 my ($self, $folder, $rw) = @_;
122 my $dbh = $self->{dbh} //= dbh_new($self, $rw);
123 my $sth = $dbh->prepare_cached(<<'', undef, 1);
124 SELECT fid FROM folders WHERE loc = ? LIMIT 1
126 my $rw_dbh = $rw ? $dbh : undef;
127 my $fid = get_fid($sth, $folder, $rw_dbh);
128 return $fid if defined($fid);
130 # caller had trailing slash (LeiToMail)
131 if ($folder =~ s!\A((?:maildir|mh):.*?)/+\z!$1!i) {
132 $fid = get_fid($sth, $folder, $rw_dbh);
134 update_fid($dbh, $fid, $folder) if $rw;
137 # sometimes we stored trailing slash..
138 } elsif ($folder =~ m!\A(?:maildir|mh):!i) {
139 $fid = get_fid($sth, $folder, $rw_dbh);
141 update_fid($dbh, $fid, $folder) if $rw;
144 } elsif ($rw && $folder =~ m!\Aimaps?://!i) {
145 require PublicInbox::URIimap;
146 PublicInbox::URIimap->new($folder)->uidvalidity //
147 Carp::croak("BUG: $folder has no UIDVALIDITY");
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);
177 if (ref($id)) { # scalar name
179 $sth = $self->{dbh}->prepare_cached(<<'');
180 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
182 } else { # numeric ID (IMAP UID, MH number)
183 $sth = $self->{dbh}->prepare_cached(<<'');
184 INSERT OR IGNORE INTO blob2num (oidbin, fid, uid) VALUES (?, ?, ?)
187 $sth->execute($oidbin, $fid, $id);
191 my ($self, $folder, $id) = @_;
192 my $lk = $self->lock_for_scope;
193 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
195 if (ref($id)) { # scalar name
197 $sth = $self->{dbh}->prepare_cached(<<'');
198 DELETE FROM blob2name WHERE fid = ? AND name = ?
201 $sth = $self->{dbh}->prepare_cached(<<'');
202 DELETE FROM blob2num WHERE fid = ? AND uid = ?
205 $sth->execute($fid, $id);
210 my ($self, $folder, $oidbin, $id, $newbn) = @_;
211 my $lk = $self->lock_for_scope;
212 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
213 $self->{dbh}->begin_work;
214 my $sth = $self->{dbh}->prepare_cached(<<'');
215 UPDATE blob2name SET name = ? WHERE fid = ? AND oidbin = ? AND name = ?
217 # eval since unique constraint may fail due to race
218 my $nr = eval { $sth->execute($newbn, $fid, $oidbin, $$id) };
219 if (!defined($nr) || $nr == 0) { # $nr may be `0E0'
220 # may race with a clear_src, ensure new value exists
221 $sth = $self->{dbh}->prepare_cached(<<'');
222 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
224 $sth->execute($oidbin, $fid, $newbn);
226 $self->{dbh}->commit;
229 # read-only, iterates every oidbin + UID or name for a given folder
231 my ($self, $folder, $cb, @args) = @_;
232 my $dbh = $self->{dbh} //= dbh_new($self);
235 if (ref($folder) eq 'HASH') {
236 $fid = $folder->{fid} // die "BUG: no `fid'";
237 @rng = grep(defined, @$folder{qw(min max)});
238 $and_ge_le = 'AND uid >= ? AND uid <= ?' if @rng;
240 $fid = $self->{fmap}->{$folder} //=
241 fid_for($self, $folder) // return;
244 # minimize implicit txn time to avoid blocking writers by
245 # batching SELECTs. This looks wonky but is necessary since
246 # $cb-> may access the DB on its own.
247 my $ary = $dbh->selectall_arrayref(<<"", undef, $fid, @rng);
248 SELECT _rowid_,oidbin,uid FROM blob2num WHERE fid = ? $and_ge_le
249 ORDER BY _rowid_ ASC LIMIT 1000
251 my $min = @$ary ? $ary->[-1]->[0] : undef;
252 while (defined $min) {
253 for my $row (@$ary) { $cb->($row->[1], $row->[2], @args) }
255 $ary = $dbh->selectall_arrayref(<<"", undef, $fid, @rng, $min);
256 SELECT _rowid_,oidbin,uid FROM blob2num
257 WHERE fid = ? $and_ge_le AND _rowid_ > ?
258 ORDER BY _rowid_ ASC LIMIT 1000
260 $min = @$ary ? $ary->[-1]->[0] : undef;
263 $ary = $dbh->selectall_arrayref(<<'', undef, $fid);
264 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ?
265 ORDER BY _rowid_ ASC LIMIT 1000
267 $min = @$ary ? $ary->[-1]->[0] : undef;
268 while (defined $min) {
269 for my $row (@$ary) { $cb->($row->[1], \($row->[2]), @args) }
271 $ary = $dbh->selectall_arrayref(<<'', undef, $fid, $min);
272 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ? AND _rowid_ > ?
273 ORDER BY _rowid_ ASC LIMIT 1000
275 $min = @$ary ? $ary->[-1]->[0] : undef;
280 my ($self, $folder) = @_;
281 my $dbh = $self->{dbh} //= dbh_new($self);
284 $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder) // return;
285 my ($row) = $dbh->selectrow_array(<<"", undef, $fid);
286 SELECT COUNT(name) FROM blob2name WHERE fid = ?
288 $ret->{'name.count'} = $row if $row;
289 my $ntype = ($folder =~ m!\A(?:nntps?|s?news)://!i) ? 'article' :
290 (($folder =~ m!\Aimaps?://!i) ? 'uid' : "TODO<$folder>");
291 for my $op (qw(count min max)) {
292 ($row) = $dbh->selectrow_array(<<"", undef, $fid);
293 SELECT $op(uid) FROM blob2num WHERE fid = ?
296 $ret->{"$ntype.$op"} = $row;
301 # returns a { location => [ list-of-ids-or-names ] } mapping
303 my ($self, $oidbin) = @_;
304 my ($fid, $sth, $id, %fid2id);
305 my $dbh = $self->{dbh} //= dbh_new($self);
306 $sth = $dbh->prepare('SELECT fid,uid FROM blob2num WHERE oidbin = ?');
307 $sth->execute($oidbin);
308 while (my ($fid, $uid) = $sth->fetchrow_array) {
309 push @{$fid2id{$fid}}, $uid;
311 $sth = $dbh->prepare('SELECT fid,name FROM blob2name WHERE oidbin = ?');
312 $sth->execute($oidbin);
313 while (my ($fid, $name) = $sth->fetchrow_array) {
314 push @{$fid2id{$fid}}, $name;
316 $sth = $dbh->prepare('SELECT loc FROM folders WHERE fid = ? LIMIT 1');
318 while (my ($fid, $ids) = each %fid2id) {
320 my ($loc) = $sth->fetchrow_array;
321 unless (defined $loc) {
322 my $oidhex = unpack('H*', $oidbin);
323 warn "E: fid=$fid for $oidhex unknown:\n", map {
324 'E: '.(ref() ? $$_ : "#$_")."\n";
330 scalar(keys %$ret) ? $ret : undef;
333 # returns a list of folders used for completion
335 my ($self, @pfx) = @_;
336 my $sql = 'SELECT loc FROM folders';
338 if (defined($pfx[0])) {
339 $sql .= ' WHERE loc REGEXP ?'; # DBD::SQLite uses perlre
340 $re = !!$pfx[1] ? '.*' : '';
341 $re .= quotemeta($pfx[0]);
344 my $sth = ($self->{dbh} //= dbh_new($self))->prepare($sql);
345 $sth->bind_param(1, $re) if defined($re);
347 map { $_->[0] } @{$sth->fetchall_arrayref};
351 my ($self, $oidhex, $vrfy) = @_;
352 my $dbh = $self->{dbh} //= dbh_new($self);
353 my $b2n = $dbh->prepare(<<'');
354 SELECT f.loc,b.name FROM blob2name b
355 LEFT JOIN folders f ON b.fid = f.fid
358 $b2n->execute(pack('H*', $oidhex));
359 while (my ($d, $n) = $b2n->fetchrow_array) {
360 substr($d, 0, length('maildir:')) = '';
361 # n.b. both mbsync and offlineimap use ":2," as a suffix
362 # in "new/", despite (from what I understand of reading
363 # <https://cr.yp.to/proto/maildir.html>), the ":2," only
364 # applies to files in "cur/".
365 my @try = $n =~ /:2,[a-zA-Z]+\z/ ? qw(cur new) : qw(new cur);
368 open my $fh, '<', $f or next;
369 # some (buggy) Maildir writers are non-atomic:
374 my $got = git_sha(1, \$raw)->hexdigest;
375 if ($got ne $oidhex) {
376 warn "$f changed $oidhex => $got\n";
387 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
388 $all //= [ $self->folders ];
389 require PublicInbox::URIimap;
390 my $want = PublicInbox::URIimap->new($url)->canonical;
391 my ($s, $h, $mb) = ($want->scheme, $want->host, $want->mailbox);
392 my @uri = map { PublicInbox::URIimap->new($_)->canonical }
393 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$mb\E\b!, @$all);
396 next if $x->mailbox ne $want->mailbox;
397 next if $x->host ne $want->host;
398 next if $x->port != $want->port;
399 my $x_uidval = $x->uidvalidity;
400 next if ($want->uidvalidity // $x_uidval) != $x_uidval;
402 # allow nothing in want to possibly match ";AUTH=ANONYMOUS"
403 if (defined($x->auth) && !defined($want->auth) &&
404 !defined($want->user)) {
406 # or maybe user was forgotten on CLI:
407 } elsif (defined($x->user) && !defined($want->user)) {
409 } elsif (($x->user//"\0") eq ($want->user//"\0")) {
413 return @match if wantarray;
414 scalar(@match) <= 1 ? $match[0] :
415 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
418 sub match_nntp_url ($$$) {
419 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
420 $all //= [ $self->folders ];
421 require PublicInbox::URInntps;
422 my $want = PublicInbox::URInntps->new($url)->canonical;
423 my ($s, $h, $p) = ($want->scheme, $want->host, $want->port);
424 my $ng = $want->group; # force scalar (no article ranges)
425 my @uri = map { PublicInbox::URInntps->new($_)->canonical }
426 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$ng\E\b!, @$all);
429 next if $x->group ne $ng || $x->host ne $h || $x->port != $p;
430 # maybe user was forgotten on CLI:
431 if (defined($x->userinfo) && !defined($want->userinfo)) {
433 } elsif (($x->userinfo//"\0") eq ($want->userinfo//"\0")) {
437 return @match if wantarray;
438 scalar(@match) <= 1 ? $match[0] :
439 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
442 # returns undef on failure, number on success
444 my ($self, $lei, $all, $folders) = @_;
445 return $lei->fail(<<EOM) if @$folders;
446 --all= not compatible with @$folders on command-line
448 my %x = map { $_ => $_ } split(/,/, $all);
449 my @ok = grep(defined, delete(@x{qw(local remote), ''}));
450 push(@ok, '') if $all eq '';
453 @no = (join(',', @no));
454 return $lei->fail(<<EOM);
455 --all=@no not accepted (must be `local' and/or `remote')
459 my @all = $self->folders;
461 if ($ok eq 'local') {
462 @inc = grep(!m!\A[a-z0-9\+]+://!i, @all);
463 } elsif ($ok eq 'remote') {
464 @inc = grep(m!\A[a-z0-9\+]+://!i, @all);
465 } elsif ($ok ne '') {
466 return $lei->fail("--all=$all not understood");
470 push(@$folders, (grep { !$seen{$_}++ } @inc));
472 scalar(@$folders) || $lei->fail(<<EOM);
473 no --mail-sync folders known to lei
477 # map CLI args to folder table entries, returns undef on failure
479 my ($self, $lei, $folders) = @_;
480 my @all = $self->folders;
481 my %all = map { $_ => 1 } @all;
484 next if $all{$_}; # ok
485 if (m!\A(maildir|mh):(.+)!i) {
487 my $d = "$type:".$lei->abs_path($2);
488 push(@no, $_) unless $all{$d};
490 } elsif (-d "$_/new" && -d "$_/cur") {
491 my $d = 'maildir:'.$lei->abs_path($_);
492 push(@no, $_) unless $all{$d};
494 } elsif (m!\Aimaps?://!i) {
496 my $res = match_imap_url($self, $orig, \@all);
500 # using `$res' instead of `$orig'
503 warn($res, "\n") if defined $res;
506 } elsif (m!\A(?:nntps?|s?news)://!i) {
508 my $res = match_nntp_url($self, $orig, \@all);
512 # using `$res' instead of `$orig'
515 warn($res, "\n") if defined $res;
523 my $no = join("\n\t", @no);
525 No sync information for: $no
526 Run `lei ls-mail-sync' to display valid choices
532 my ($self, @folders) = @_;
533 my $lk = $self->lock_for_scope;
534 for my $folder (@folders) {
535 my $fid = delete($self->{fmap}->{$folder}) //
536 fid_for($self, $folder) // next;
537 for my $t (qw(blob2name blob2num folders)) {
538 $self->{dbh}->do("DELETE FROM $t WHERE fid = ?",
544 # only used for changing canonicalization errors
546 my ($self, $old, $new) = @_;
547 my $lk = $self->lock_for_scope;
548 my $ofid = delete($self->{fmap}->{$old}) //
549 fid_for($self, $old) // return;
551 $self->{dbh}->do(<<EOM, undef, $new, $ofid);
552 UPDATE folders SET loc = ? WHERE fid = ?
555 if ($@ =~ /\bunique\b/i) {
556 my $nfid = $self->{fmap}->{$new} // fid_for($self, $new);
557 for my $t (qw(blob2name blob2num)) {
558 $self->{dbh}->do(<<EOM, undef, $nfid, $ofid);
559 UPDATE OR REPLACE $t SET fid = ? WHERE fid = ?
562 $self->{dbh}->do(<<EOM, undef, $ofid);
563 DELETE FROM folders WHERE fid = ?
568 sub num_oidbin ($$$) {
569 my ($self, $url, $uid) = @_; # $url MUST have UIDVALIDITY if IMAP
570 my $fid = $self->{fmap}->{$url} //= fid_for($self, $url) // return ();
571 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
572 SELECT oidbin FROM blob2num WHERE fid = ? AND uid = ? ORDER BY _rowid_
574 $sth->execute($fid, $uid);
575 map { $_->[0] } @{$sth->fetchall_arrayref};
578 sub name_oidbin ($$$) {
579 my ($self, $mdir, $nm) = @_;
580 my $fid = $self->{fmap}->{$mdir} //= fid_for($self, $mdir) // return;
581 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
582 SELECT oidbin FROM blob2name WHERE fid = ? AND name = ?
584 $sth->execute($fid, $nm);
585 map { $_->[0] } @{$sth->fetchall_arrayref};
589 my ($self, $lei, $uid_uri) = @_;
590 my $mailbox_uri = $uid_uri->clone;
591 $mailbox_uri->uid(undef);
592 my $folders = [ $$mailbox_uri ];
593 eval { $self->arg2folder($lei, $folders) };
594 $lei->qerr("# no sync information for $mailbox_uri") if $@;
595 map { unpack('H*',$_) } num_oidbin($self, $folders->[0], $uid_uri->uid)