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 $TYPE:/pathname */
65 CREATE TABLE IF NOT EXISTS blob2num (
66 oidbin VARBINARY NOT NULL,
67 fid INTEGER NOT NULL, /* folder ID */
68 uid INTEGER NOT NULL, /* NNTP article number, IMAP UID, MH number */
69 UNIQUE (oidbin, fid, uid)
72 # speeds up LeiImport->ck_update_kw (for "lei import") by 5-6x:
74 CREATE INDEX IF NOT EXISTS idx_fid_uid ON blob2num(fid,uid)
77 CREATE TABLE IF NOT EXISTS blob2name (
78 oidbin VARBINARY NOT NULL,
79 fid INTEGER NOT NULL, /* folder ID */
80 name VARBINARY NOT NULL, /* Maildir basename, JMAP blobId */
81 UNIQUE (oidbin, fid, name)
84 # speeds up LeiImport->pmdir_cb (for "lei import") by ~6x:
86 CREATE INDEX IF NOT EXISTS idx_fid_name ON blob2name(fid,name)
91 my ($self, $folder, $rw) = @_;
92 my $dbh = $self->{dbh} //= dbh_new($self, $rw);
93 my $sel = 'SELECT fid FROM folders WHERE loc = ? LIMIT 1';
94 my ($fid) = $dbh->selectrow_array($sel, undef, $folder);
95 return $fid if defined $fid;
97 # caller had trailing slash (LeiToMail)
98 if ($folder =~ s!\A((?:maildir|mh):.*?)/+\z!$1!i) {
99 ($fid) = $dbh->selectrow_array($sel, undef, $folder);
101 $dbh->do(<<EOM, undef, $folder, $fid) if $rw;
102 UPDATE folders SET loc = ? WHERE fid = ?
106 # sometimes we stored trailing slash..
107 } elsif ($folder =~ m!\A(?:maildir|mh):!i) {
108 ($fid) = $dbh->selectrow_array($sel, undef, "$folder/");
110 $dbh->do(<<EOM, undef, $folder, $fid) if $rw;
111 UPDATE folders SET loc = ? WHERE fid = ?
115 } elsif ($rw && $folder =~ m!\Aimaps?://!i) {
116 require PublicInbox::URIimap;
117 PublicInbox::URIimap->new($folder)->uidvalidity //
118 Carp::croak("BUG: $folder has no UIDVALIDITY");
122 ($fid) = $dbh->selectrow_array('SELECT MAX(fid) FROM folders');
125 # in case we're reusing, clobber existing stale refs:
126 $dbh->do('DELETE FROM blob2name WHERE fid = ?', undef, $fid);
127 $dbh->do('DELETE FROM blob2num WHERE fid = ?', undef, $fid);
129 my $sth = $dbh->prepare('INSERT INTO folders (fid, loc) VALUES (?, ?)');
130 $sth->execute($fid, $folder);
136 my ($self, @folders) = @_;
137 my $lk = $self->lock_for_scope;
138 for my $f (@folders) { $self->{fmap}->{$f} //= fid_for($self, $f, 1) }
142 my ($self, $oidbin, $folder, $id) = @_;
143 my $lk = $self->lock_for_scope;
144 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
146 if (ref($id)) { # scalar name
148 $sth = $self->{dbh}->prepare_cached(<<'');
149 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
151 } else { # numeric ID (IMAP UID, MH number)
152 $sth = $self->{dbh}->prepare_cached(<<'');
153 INSERT OR IGNORE INTO blob2num (oidbin, fid, uid) VALUES (?, ?, ?)
156 $sth->execute($oidbin, $fid, $id);
160 my ($self, $folder, $id) = @_;
161 my $lk = $self->lock_for_scope;
162 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
164 if (ref($id)) { # scalar name
166 $sth = $self->{dbh}->prepare_cached(<<'');
167 DELETE FROM blob2name WHERE fid = ? AND name = ?
170 $sth = $self->{dbh}->prepare_cached(<<'');
171 DELETE FROM blob2num WHERE fid = ? AND uid = ?
174 $sth->execute($fid, $id);
179 my ($self, $folder, $oidbin, $id, $newbn) = @_;
180 my $lk = $self->lock_for_scope;
181 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
182 my $sth = $self->{dbh}->prepare_cached(<<'');
183 UPDATE blob2name SET name = ? WHERE fid = ? AND oidbin = ? AND name = ?
185 my $nr = $sth->execute($newbn, $fid, $oidbin, $$id);
186 if ($nr == 0) { # may race with a clear_src, ensure new value exists
187 $sth = $self->{dbh}->prepare_cached(<<'');
188 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
190 $sth->execute($oidbin, $fid, $newbn);
194 # read-only, iterates every oidbin + UID or name for a given folder
196 my ($self, $folder, $cb, @args) = @_;
197 my $dbh = $self->{dbh} //= dbh_new($self);
199 if (ref($folder) eq 'HASH') {
200 $fid = $folder->{fid} // die "BUG: no `fid'";
202 $fid = $self->{fmap}->{$folder} //=
203 fid_for($self, $folder) // return;
206 # minimize implicit txn time to avoid blocking writers by
207 # batching SELECTs. This looks wonky but is necessary since
208 # $cb-> may access the DB on its own.
209 my $ary = $dbh->selectall_arrayref(<<'', undef, $fid);
210 SELECT _rowid_,oidbin,uid FROM blob2num WHERE fid = ?
211 ORDER BY _rowid_ ASC LIMIT 1000
213 my $min = @$ary ? $ary->[-1]->[0] : undef;
214 while (defined $min) {
215 for my $row (@$ary) { $cb->($row->[1], $row->[2], @args) }
217 $ary = $dbh->selectall_arrayref(<<'', undef, $fid, $min);
218 SELECT _rowid_,oidbin,uid FROM blob2num WHERE fid = ? AND _rowid_ > ?
219 ORDER BY _rowid_ ASC LIMIT 1000
221 $min = @$ary ? $ary->[-1]->[0] : undef;
224 $ary = $dbh->selectall_arrayref(<<'', undef, $fid);
225 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ?
226 ORDER BY _rowid_ ASC LIMIT 1000
228 $min = @$ary ? $ary->[-1]->[0] : undef;
229 while (defined $min) {
230 for my $row (@$ary) { $cb->($row->[1], \($row->[2]), @args) }
232 $ary = $dbh->selectall_arrayref(<<'', undef, $fid, $min);
233 SELECT _rowid_,oidbin,name FROM blob2name WHERE fid = ? AND _rowid_ > ?
234 ORDER BY _rowid_ ASC LIMIT 1000
236 $min = @$ary ? $ary->[-1]->[0] : undef;
241 my ($self, $folder) = @_;
242 my $dbh = $self->{dbh} //= dbh_new($self);
245 $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder) // return;
246 my ($row) = $dbh->selectrow_array(<<"", undef, $fid);
247 SELECT COUNT(name) FROM blob2name WHERE fid = ?
249 $ret->{'name.count'} = $row if $row;
250 my $ntype = ($folder =~ m!\A(?:nntps?|s?news)://!i) ? 'article' :
251 (($folder =~ m!\Aimaps?://!i) ? 'uid' : "TODO<$folder>");
252 for my $op (qw(count min max)) {
253 ($row) = $dbh->selectrow_array(<<"", undef, $fid);
254 SELECT $op(uid) FROM blob2num WHERE fid = ?
257 $ret->{"$ntype.$op"} = $row;
262 # returns a { location => [ list-of-ids-or-names ] } mapping
264 my ($self, $oidbin) = @_;
265 my ($fid, $sth, $id, %fid2id);
266 my $dbh = $self->{dbh} //= dbh_new($self);
267 $sth = $dbh->prepare('SELECT fid,uid FROM blob2num WHERE oidbin = ?');
268 $sth->execute($oidbin);
269 while (my ($fid, $uid) = $sth->fetchrow_array) {
270 push @{$fid2id{$fid}}, $uid;
272 $sth = $dbh->prepare('SELECT fid,name FROM blob2name WHERE oidbin = ?');
273 $sth->execute($oidbin);
274 while (my ($fid, $name) = $sth->fetchrow_array) {
275 push @{$fid2id{$fid}}, $name;
277 $sth = $dbh->prepare('SELECT loc FROM folders WHERE fid = ? LIMIT 1');
279 while (my ($fid, $ids) = each %fid2id) {
281 my ($loc) = $sth->fetchrow_array;
282 unless (defined $loc) {
283 my $oidhex = unpack('H*', $oidbin);
284 warn "E: fid=$fid for $oidhex unknown:\n", map {
285 'E: '.(ref() ? $$_ : "#$_")."\n";
291 scalar(keys %$ret) ? $ret : undef;
294 # returns a list of folders used for completion
296 my ($self, $pfx) = @_;
297 my $dbh = $self->{dbh} //= dbh_new($self);
298 my $sql = 'SELECT loc FROM folders';
301 $sql .= ' WHERE loc LIKE ? ESCAPE ?';
303 $pfx[0] =~ s/([%_\\])/\\$1/g; # glob chars
306 map { $_->[0] } @{$dbh->selectall_arrayref($sql, undef, @pfx)};
310 my ($self, $oidhex, $vrfy) = @_;
311 my $dbh = $self->{dbh} //= dbh_new($self);
312 my $b2n = $dbh->prepare(<<'');
313 SELECT f.loc,b.name FROM blob2name b
314 LEFT JOIN folders f ON b.fid = f.fid
317 $b2n->execute(pack('H*', $oidhex));
318 while (my ($d, $n) = $b2n->fetchrow_array) {
319 substr($d, 0, length('maildir:')) = '';
320 # n.b. both mbsync and offlineimap use ":2," as a suffix
321 # in "new/", despite (from what I understand of reading
322 # <https://cr.yp.to/proto/maildir.html>), the ":2," only
323 # applies to files in "cur/".
324 my @try = $n =~ /:2,[a-zA-Z]+\z/ ? qw(cur new) : qw(new cur);
327 open my $fh, '<', $f or next;
328 # some (buggy) Maildir writers are non-atomic:
332 if ($vrfy && git_sha(1, \$raw)->hexdigest ne $oidhex) {
333 warn "$f changed $oidhex\n";
343 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
344 $all //= [ $self->folders ];
345 require PublicInbox::URIimap;
346 my $want = PublicInbox::URIimap->new($url)->canonical;
347 my ($s, $h, $mb) = ($want->scheme, $want->host, $want->mailbox);
348 my @uri = map { PublicInbox::URIimap->new($_)->canonical }
349 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$mb\E\b!, @$all);
352 next if $x->mailbox ne $want->mailbox;
353 next if $x->host ne $want->host;
354 next if $x->port != $want->port;
355 my $x_uidval = $x->uidvalidity;
356 next if ($want->uidvalidity // $x_uidval) != $x_uidval;
358 # allow nothing in want to possibly match ";AUTH=ANONYMOUS"
359 if (defined($x->auth) && !defined($want->auth) &&
360 !defined($want->user)) {
362 # or maybe user was forgotten on CLI:
363 } elsif (defined($x->user) && !defined($want->user)) {
365 } elsif (($x->user//"\0") eq ($want->user//"\0")) {
369 return @match if wantarray;
370 scalar(@match) <= 1 ? $match[0] :
371 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
374 sub match_nntp_url ($$$) {
375 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
376 $all //= [ $self->folders ];
377 require PublicInbox::URInntps;
378 my $want = PublicInbox::URInntps->new($url)->canonical;
379 my ($s, $h, $p) = ($want->scheme, $want->host, $want->port);
380 my $ng = $want->group; # force scalar (no article ranges)
381 my @uri = map { PublicInbox::URInntps->new($_)->canonical }
382 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$ng\E\b!, @$all);
385 next if $x->group ne $ng || $x->host ne $h || $x->port != $p;
386 # maybe user was forgotten on CLI:
387 if (defined($x->userinfo) && !defined($want->userinfo)) {
389 } elsif (($x->userinfo//"\0") eq ($want->userinfo//"\0")) {
393 return @match if wantarray;
394 scalar(@match) <= 1 ? $match[0] :
395 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
398 # returns undef on failure, number on success
400 my ($self, $lei, $all, $folders) = @_;
401 return $lei->fail(<<EOM) if @$folders;
402 --all= not compatible with @$folders on command-line
404 my %x = map { $_ => $_ } split(/,/, $all);
405 my @ok = grep(defined, delete(@x{qw(local remote), ''}));
406 push(@ok, '') if $all eq '';
409 @no = (join(',', @no));
410 return $lei->fail(<<EOM);
411 --all=@no not accepted (must be `local' and/or `remote')
415 my @all = $self->folders;
417 if ($ok eq 'local') {
418 @inc = grep(!m!\A[a-z0-9\+]+://!i, @all);
419 } elsif ($ok eq 'remote') {
420 @inc = grep(m!\A[a-z0-9\+]+://!i, @all);
421 } elsif ($ok ne '') {
422 return $lei->fail("--all=$all not understood");
426 push(@$folders, (grep { !$seen{$_}++ } @inc));
428 scalar(@$folders) || $lei->fail(<<EOM);
429 no --mail-sync folders known to lei
433 # map CLI args to folder table entries, returns undef on failure
435 my ($self, $lei, $folders) = @_;
436 my @all = $self->folders;
437 my %all = map { $_ => 1 } @all;
440 next if $all{$_}; # ok
441 if (m!\A(maildir|mh):(.+)!i) {
443 my $d = "$type:".$lei->abs_path($2);
444 push(@no, $_) unless $all{$d};
446 } elsif (-d "$_/new" && -d "$_/cur") {
447 my $d = 'maildir:'.$lei->abs_path($_);
448 push(@no, $_) unless $all{$d};
450 } elsif (m!\Aimaps?://!i) {
452 my $res = match_imap_url($self, $orig, \@all);
455 push(@{$err->{qerr}}, <<EOM);
456 # using `$res' instead of `$orig'
459 $lei->err($res) if defined $res;
462 } elsif (m!\A(?:nntps?|s?news)://!i) {
464 my $res = match_nntp_url($self, $orig, \@all);
467 push(@{$err->{qerr}}, <<EOM);
468 # using `$res' instead of `$orig'
471 $lei->err($res) if defined $res;
479 my $no = join("\n\t", @no);
480 $err->{fail} = <<EOF;
481 No sync information for: $no
482 Run `lei ls-mail-sync' to display valid choices
489 my ($self, @folders) = @_;
490 my $lk = $self->lock_for_scope;
491 for my $folder (@folders) {
492 my $fid = delete($self->{fmap}->{$folder}) //
493 fid_for($self, $folder) // next;
494 for my $t (qw(blob2name blob2num folders)) {
495 $self->{dbh}->do("DELETE FROM $t WHERE fid = ?",
501 # only used for changing canonicalization errors
503 my ($self, $old, $new) = @_;
504 my $lk = $self->lock_for_scope;
505 my $ofid = delete($self->{fmap}->{$old}) //
506 fid_for($self, $old) // return;
508 $self->{dbh}->do(<<EOM, undef, $new, $ofid);
509 UPDATE folders SET loc = ? WHERE fid = ?
512 if ($@ =~ /\bunique\b/i) {
513 my $nfid = $self->{fmap}->{$new} // fid_for($self, $new);
514 for my $t (qw(blob2name blob2num)) {
515 $self->{dbh}->do(<<EOM, undef, $nfid, $ofid);
516 UPDATE OR REPLACE $t SET fid = ? WHERE fid = ?
519 $self->{dbh}->do(<<EOM, undef, $ofid);
520 DELETE FROM folders WHERE fid = ?
525 sub imap_oidbin ($$$) {
526 my ($self, $url, $uid) = @_; # $url MUST have UIDVALIDITY
527 my $fid = $self->{fmap}->{$url} //= fid_for($self, $url) // return;
528 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
529 SELECT oidbin FROM blob2num WHERE fid = ? AND uid = ?
531 $sth->execute($fid, $uid);
532 $sth->fetchrow_array;
535 sub name_oidbin ($$$) {
536 my ($self, $mdir, $nm) = @_;
537 my $fid = $self->{fmap}->{$mdir} //= fid_for($self, $mdir) // return;
538 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
539 SELECT oidbin FROM blob2name WHERE fid = ? AND name = ?
541 $sth->execute($fid, $nm);
542 $sth->fetchrow_array;
546 my ($self, $lei, $uid_uri) = @_;
547 my $mailbox_uri = $uid_uri->clone;
548 $mailbox_uri->uid(undef);
549 my $folders = [ $$mailbox_uri ];
550 if (my $err = $self->arg2folder($lei, $folders)) {
552 $lei->qerr("# no sync information for $mailbox_uri");
555 $lei->qerr(@{$err->{qerr}}) if $err->{qerr};
557 my $oidbin = imap_oidbin($self, $folders->[0], $uid_uri->uid);
558 $oidbin ? unpack('H*', $oidbin) : undef;