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;
17 my $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
22 sqlite_use_immediate_transaction => 1,
24 # no sqlite_unicode, here, all strings are binary
25 create_tables($self, $dbh) if $rw;
26 $dbh->do('PRAGMA journal_mode = WAL') if $creat;
27 $dbh->do('PRAGMA case_sensitive_like = ON');
36 lock_path => "$f.flock",
40 sub lms_write_prepare { ($_[0]->{dbh} //= dbh_new($_[0], 1)); $_[0] }
49 my ($self, $dbh) = @_;
50 my $lk = $self->lock_for_scope;
53 CREATE TABLE IF NOT EXISTS folders (
54 fid INTEGER PRIMARY KEY,
55 loc VARBINARY NOT NULL, /* URL;UIDVALIDITY=$N or $TYPE:/pathname */
60 CREATE TABLE IF NOT EXISTS blob2num (
61 oidbin VARBINARY NOT NULL,
62 fid INTEGER NOT NULL, /* folder ID */
63 uid INTEGER NOT NULL, /* NNTP article number, IMAP UID, MH number */
64 UNIQUE (oidbin, fid, uid)
67 # speeds up LeiImport->ck_update_kw (for "lei import") by 5-6x:
69 CREATE INDEX IF NOT EXISTS idx_fid_uid ON blob2num(fid,uid)
72 CREATE TABLE IF NOT EXISTS blob2name (
73 oidbin VARBINARY NOT NULL,
74 fid INTEGER NOT NULL, /* folder ID */
75 name VARBINARY NOT NULL, /* Maildir basename, JMAP blobId */
76 UNIQUE (oidbin, fid, name)
79 # speeds up LeiImport->pmdir_cb (for "lei import") by ~6x:
81 CREATE INDEX IF NOT EXISTS idx_fid_name ON blob2name(fid,name)
86 my ($self, $folder, $rw) = @_;
87 my $dbh = $self->{dbh} //= dbh_new($self, $rw);
88 my $sel = 'SELECT fid FROM folders WHERE loc = ? LIMIT 1';
89 my ($fid) = $dbh->selectrow_array($sel, undef, $folder);
90 return $fid if defined $fid;
92 # caller had trailing slash (LeiToMail)
93 if ($folder =~ s!\A((?:maildir|mh):.*?)/+\z!$1!i) {
94 ($fid) = $dbh->selectrow_array($sel, undef, $folder);
96 $dbh->do(<<EOM, undef, $folder, $fid) if $rw;
97 UPDATE folders SET loc = ? WHERE fid = ?
101 # sometimes we stored trailing slash..
102 } elsif ($folder =~ m!\A(?:maildir|mh):!i) {
103 ($fid) = $dbh->selectrow_array($sel, undef, "$folder/");
105 $dbh->do(<<EOM, undef, $folder, $fid) if $rw;
106 UPDATE folders SET loc = ? WHERE fid = ?
110 } elsif ($rw && $folder =~ m!\Aimaps?://!i) {
111 require PublicInbox::URIimap;
112 PublicInbox::URIimap->new($folder)->uidvalidity //
113 Carp::croak("BUG: $folder has no UIDVALIDITY");
117 ($fid) = $dbh->selectrow_array('SELECT MAX(fid) FROM folders');
120 # in case we're reusing, clobber existing stale refs:
121 $dbh->do('DELETE FROM blob2name WHERE fid = ?', undef, $fid);
122 $dbh->do('DELETE FROM blob2num WHERE fid = ?', undef, $fid);
124 my $sth = $dbh->prepare('INSERT INTO folders (fid, loc) VALUES (?, ?)');
125 $sth->execute($fid, $folder);
131 my ($self, @folders) = @_;
132 my $lk = $self->lock_for_scope;
133 for my $f (@folders) { $self->{fmap}->{$f} //= fid_for($self, $f, 1) }
137 my ($self, $oidbin, $folder, $id) = @_;
138 my $lk = $self->lock_for_scope;
139 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
141 if (ref($id)) { # scalar name
143 $sth = $self->{dbh}->prepare_cached(<<'');
144 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
146 } else { # numeric ID (IMAP UID, MH number)
147 $sth = $self->{dbh}->prepare_cached(<<'');
148 INSERT OR IGNORE INTO blob2num (oidbin, fid, uid) VALUES (?, ?, ?)
151 $sth->execute($oidbin, $fid, $id);
155 my ($self, $folder, $id) = @_;
156 my $lk = $self->lock_for_scope;
157 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
159 if (ref($id)) { # scalar name
161 $sth = $self->{dbh}->prepare_cached(<<'');
162 DELETE FROM blob2name WHERE fid = ? AND name = ?
165 $sth = $self->{dbh}->prepare_cached(<<'');
166 DELETE FROM blob2num WHERE fid = ? AND uid = ?
169 $sth->execute($fid, $id);
174 my ($self, $folder, $oidbin, $id, $newbn) = @_;
175 my $lk = $self->lock_for_scope;
176 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
177 my $sth = $self->{dbh}->prepare_cached(<<'');
178 UPDATE blob2name SET name = ? WHERE fid = ? AND oidbin = ? AND name = ?
180 my $nr = $sth->execute($newbn, $fid, $oidbin, $$id);
181 if ($nr == 0) { # may race with a clear_src, ensure new value exists
182 $sth = $self->{dbh}->prepare_cached(<<'');
183 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
185 $sth->execute($oidbin, $fid, $newbn);
189 # read-only, iterates every oidbin + UID or name for a given folder
191 my ($self, $folder, $cb, @args) = @_;
192 my $dbh = $self->{dbh} //= dbh_new($self);
194 if (ref($folder) eq 'HASH') {
195 $fid = $folder->{fid} // die "BUG: no `fid'";
197 $fid = $self->{fmap}->{$folder} //=
198 fid_for($self, $folder) // return;
201 # minimize implicit txn time to avoid blocking writers by
202 # batching SELECTs. This looks wonky but is necessary since
203 # $cb-> may access the DB on its own.
204 my $ary = $dbh->selectall_arrayref(<<'', undef, $fid);
205 SELECT _rowid_,oidbin,uid FROM blob2num WHERE fid = ?
206 ORDER BY _rowid_ ASC LIMIT 1000
208 my $min = @$ary ? $ary->[-1]->[0] : undef;
209 while (defined $min) {
210 for my $row (@$ary) { $cb->($row->[1], $row->[2], @args) }
212 $ary = $dbh->selectall_arrayref(<<'', undef, $fid, $min);
213 SELECT _rowid_,oidbin,uid FROM blob2num WHERE fid = ? AND _rowid_ > ?
214 ORDER BY _rowid_ ASC LIMIT 1000
216 $min = @$ary ? $ary->[-1]->[0] : undef;
219 $ary = $dbh->selectall_arrayref(<<'', undef, $fid);
220 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ?
221 ORDER BY _rowid_ ASC LIMIT 1000
223 $min = @$ary ? $ary->[-1]->[0] : undef;
224 while (defined $min) {
225 for my $row (@$ary) { $cb->($row->[1], \($row->[2]), @args) }
227 $ary = $dbh->selectall_arrayref(<<'', undef, $fid, $min);
228 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ? AND _rowid_ > ?
229 ORDER BY _rowid_ ASC LIMIT 1000
231 $min = @$ary ? $ary->[-1]->[0] : undef;
236 my ($self, $folder) = @_;
237 my $dbh = $self->{dbh} //= dbh_new($self);
240 $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder) // return;
241 my ($row) = $dbh->selectrow_array(<<"", undef, $fid);
242 SELECT COUNT(name) FROM blob2name WHERE fid = ?
244 $ret->{'name.count'} = $row if $row;
245 for my $op (qw(count min max)) {
246 ($row) = $dbh->selectrow_array(<<"", undef, $fid);
247 SELECT $op(uid) FROM blob2num WHERE fid = ?
250 $ret->{"uid.$op"} = $row;
255 # returns a { location => [ list-of-ids-or-names ] } mapping
257 my ($self, $oidbin) = @_;
258 my ($fid, $sth, $id, %fid2id);
259 my $dbh = $self->{dbh} //= dbh_new($self);
260 $sth = $dbh->prepare('SELECT fid,uid FROM blob2num WHERE oidbin = ?');
261 $sth->execute($oidbin);
262 while (my ($fid, $uid) = $sth->fetchrow_array) {
263 push @{$fid2id{$fid}}, $uid;
265 $sth = $dbh->prepare('SELECT fid,name FROM blob2name WHERE oidbin = ?');
266 $sth->execute($oidbin);
267 while (my ($fid, $name) = $sth->fetchrow_array) {
268 push @{$fid2id{$fid}}, $name;
270 $sth = $dbh->prepare('SELECT loc FROM folders WHERE fid = ? LIMIT 1');
272 while (my ($fid, $ids) = each %fid2id) {
274 my ($loc) = $sth->fetchrow_array;
275 unless (defined $loc) {
276 my $oidhex = unpack('H*', $oidbin);
277 warn "E: fid=$fid for $oidhex unknown:\n", map {
278 'E: '.(ref() ? $$_ : "#$_")."\n";
284 scalar(keys %$ret) ? $ret : undef;
287 # returns a list of folders used for completion
289 my ($self, $pfx) = @_;
290 my $dbh = $self->{dbh} //= dbh_new($self);
291 my $sql = 'SELECT loc FROM folders';
294 $sql .= ' WHERE loc LIKE ? ESCAPE ?';
296 $pfx[0] =~ s/([%_\\])/\\$1/g; # glob chars
299 map { $_->[0] } @{$dbh->selectall_arrayref($sql, undef, @pfx)};
303 my ($self, $oidhex, $vrfy) = @_;
304 my $dbh = $self->{dbh} //= dbh_new($self);
305 my $b2n = $dbh->prepare(<<'');
306 SELECT f.loc,b.name FROM blob2name b
307 LEFT JOIN folders f ON b.fid = f.fid
310 $b2n->execute(pack('H*', $oidhex));
311 while (my ($d, $n) = $b2n->fetchrow_array) {
312 substr($d, 0, length('maildir:')) = '';
313 # n.b. both mbsync and offlineimap use ":2," as a suffix
314 # in "new/", despite (from what I understand of reading
315 # <https://cr.yp.to/proto/maildir.html>), the ":2," only
316 # applies to files in "cur/".
317 my @try = $n =~ /:2,[a-zA-Z]+\z/ ? qw(cur new) : qw(new cur);
320 open my $fh, '<', $f or next;
321 # some (buggy) Maildir writers are non-atomic:
325 if ($vrfy && git_sha(1, \$raw)->hexdigest ne $oidhex) {
326 warn "$f changed $oidhex\n";
336 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
337 $all //= [ $self->folders ];
338 require PublicInbox::URIimap;
339 my $want = PublicInbox::URIimap->new($url)->canonical;
340 my ($s, $h, $mb) = ($want->scheme, $want->host, $want->mailbox);
341 my @uri = map { PublicInbox::URIimap->new($_)->canonical }
342 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$mb\E\b!, @$all);
345 next if $x->mailbox ne $want->mailbox;
346 next if $x->host ne $want->host;
347 next if $x->port != $want->port;
348 my $x_uidval = $x->uidvalidity;
349 next if ($want->uidvalidity // $x_uidval) != $x_uidval;
351 # allow nothing in want to possibly match ";AUTH=ANONYMOUS"
352 if (defined($x->auth) && !defined($want->auth) &&
353 !defined($want->user)) {
355 # or maybe user was forgotten on CLI:
356 } elsif (defined($x->user) && !defined($want->user)) {
358 } elsif (($x->user//"\0") eq ($want->user//"\0")) {
362 return @match if wantarray;
363 scalar(@match) <= 1 ? $match[0] :
364 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
367 # returns undef on failure, number on success
369 my ($self, $lei, $all, $folders) = @_;
370 return $lei->fail(<<EOM) if @$folders;
371 --all= not compatible with @$folders on command-line
373 my %x = map { $_ => $_ } split(/,/, $all);
374 my @ok = grep(defined, delete(@x{qw(local remote), ''}));
375 push(@ok, '') if $all eq '';
378 @no = (join(',', @no));
379 return $lei->fail(<<EOM);
380 --all=@no not accepted (must be `local' and/or `remote')
384 my @all = $self->folders;
386 if ($ok eq 'local') {
387 @inc = grep(!m!\A[a-z0-9\+]+://!i, @all);
388 } elsif ($ok eq 'remote') {
389 @inc = grep(m!\A[a-z0-9\+]+://!i, @all);
390 } elsif ($ok ne '') {
391 return $lei->fail("--all=$all not understood");
395 push(@$folders, (grep { !$seen{$_}++ } @inc));
397 scalar(@$folders) || $lei->fail(<<EOM);
398 no --mail-sync folders known to lei
402 # map CLI args to folder table entries, returns undef on failure
404 my ($self, $lei, $folders) = @_;
405 my @all = $self->folders;
406 my %all = map { $_ => 1 } @all;
409 next if $all{$_}; # ok
410 if (m!\A(maildir|mh):(.+)!i) {
412 my $d = "$type:".$lei->abs_path($2);
413 push(@no, $_) unless $all{$d};
415 } elsif (-d "$_/new" && -d "$_/cur") {
416 my $d = 'maildir:'.$lei->abs_path($_);
417 push(@no, $_) unless $all{$d};
419 } elsif (m!\Aimaps?://!i) {
421 my $res = match_imap_url($self, $orig, \@all);
424 push(@{$err->{qerr}}, <<EOM);
425 # using `$res' instead of `$orig'
428 $lei->err($res) if defined $res;
436 my $no = join("\n\t", @no);
437 $err->{fail} = <<EOF;
438 No sync information for: $no
439 Run `lei ls-mail-sync' to display valid choices
446 my ($self, @folders) = @_;
447 my $lk = $self->lock_for_scope;
448 for my $folder (@folders) {
449 my $fid = delete($self->{fmap}->{$folder}) //
450 fid_for($self, $folder) // next;
451 for my $t (qw(blob2name blob2num folders)) {
452 $self->{dbh}->do("DELETE FROM $t WHERE fid = ?",
458 # only used for changing canonicalization errors
460 my ($self, $old, $new) = @_;
461 my $lk = $self->lock_for_scope;
462 my $ofid = delete($self->{fmap}->{$old}) //
463 fid_for($self, $old) // return;
465 $self->{dbh}->do(<<EOM, undef, $new, $ofid);
466 UPDATE folders SET loc = ? WHERE fid = ?
469 if ($@ =~ /\bunique\b/i) {
470 my $nfid = $self->{fmap}->{$new} // fid_for($self, $new);
471 for my $t (qw(blob2name blob2num)) {
472 $self->{dbh}->do(<<EOM, undef, $nfid, $ofid);
473 UPDATE OR REPLACE $t SET fid = ? WHERE fid = ?
476 $self->{dbh}->do(<<EOM, undef, $ofid);
477 DELETE FROM folders WHERE fid = ?
482 sub imap_oidbin ($$$) {
483 my ($self, $url, $uid) = @_; # $url MUST have UIDVALIDITY
484 my $fid = $self->{fmap}->{$url} //= fid_for($self, $url) // return;
485 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
486 SELECT oidbin FROM blob2num WHERE fid = ? AND uid = ?
488 $sth->execute($fid, $uid);
489 $sth->fetchrow_array;
492 sub name_oidbin ($$$) {
493 my ($self, $mdir, $nm) = @_;
494 my $fid = $self->{fmap}->{$mdir} //= fid_for($self, $mdir) // return;
495 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
496 SELECT oidbin FROM blob2name WHERE fid = ? AND name = ?
498 $sth->execute($fid, $nm);
499 $sth->fetchrow_array;
503 my ($self, $lei, $uid_uri) = @_;
504 my $mailbox_uri = $uid_uri->clone;
505 $mailbox_uri->uid(undef);
506 my $folders = [ $$mailbox_uri ];
507 if (my $err = $self->arg2folder($lei, $folders)) {
509 $lei->qerr("# no sync information for $mailbox_uri");
512 $lei->qerr(@{$err->{qerr}}) if $err->{qerr};
514 my $oidbin = imap_oidbin($self, $folders->[0], $uid_uri->uid);
515 $oidbin ? unpack('H*', $oidbin) : undef;