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;
9 use PublicInbox::ContentHash qw(git_sha);
14 my $f = $self->{filename};
15 my $creat = $rw && !-s $f;
16 my $dbh = DBI->connect("dbi:SQLite:dbname=$f",'','', {
21 sqlite_use_immediate_transaction => 1,
23 # no sqlite_unicode, here, all strings are binary
24 create_tables($dbh) if $rw;
25 $dbh->do('PRAGMA journal_mode = WAL') if $creat;
26 $dbh->do('PRAGMA case_sensitive_like = ON');
32 bless { filename => $f, fmap => {} }, $cls;
35 sub lms_commit { delete($_[0]->{dbh})->commit }
37 sub lms_begin { ($_[0]->{dbh} //= dbh_new($_[0], 1))->begin_work };
43 CREATE TABLE IF NOT EXISTS folders (
44 fid INTEGER PRIMARY KEY,
45 loc VARBINARY NOT NULL, /* URL;UIDVALIDITY=$N or $TYPE:/pathname */
50 CREATE TABLE IF NOT EXISTS blob2num (
51 oidbin VARBINARY NOT NULL,
52 fid INTEGER NOT NULL, /* folder ID */
53 uid INTEGER NOT NULL, /* NNTP article number, IMAP UID, MH number */
54 UNIQUE (oidbin, fid, uid)
57 # speeds up LeiImport->ck_update_kw (for "lei import") by 5-6x:
59 CREATE INDEX IF NOT EXISTS idx_fid_uid ON blob2num(fid,uid)
62 CREATE TABLE IF NOT EXISTS blob2name (
63 oidbin VARBINARY NOT NULL,
64 fid INTEGER NOT NULL, /* folder ID */
65 name VARBINARY NOT NULL, /* Maildir basename, JMAP blobId */
66 UNIQUE (oidbin, fid, name)
69 # speeds up LeiImport->pmdir_cb (for "lei import") by ~6x:
71 CREATE INDEX IF NOT EXISTS idx_fid_name ON blob2name(fid,name)
76 my ($self, $folder, $rw) = @_;
77 my $dbh = $self->{dbh} //= dbh_new($self, $rw);
78 my $sel = 'SELECT fid FROM folders WHERE loc = ? LIMIT 1';
79 my ($fid) = $dbh->selectrow_array($sel, undef, $folder);
80 return $fid if defined $fid;
82 if ($folder =~ s!\A((?:maildir|mh):.*?)/+\z!$1!i) {
83 warn "folder: $folder/ had trailing slash in arg\n";
84 ($fid) = $dbh->selectrow_array($sel, undef, $folder);
86 $dbh->do(<<EOM, undef, $folder, $fid) if $rw;
87 UPDATE folders SET loc = ? WHERE fid = ?
91 # sometimes we stored trailing slash..
92 } elsif ($folder =~ m!\A(?:maildir|mh):!i) {
93 ($fid) = $dbh->selectrow_array($sel, undef, "$folder/");
95 $dbh->do(<<EOM, undef, $folder, $fid) if $rw;
96 UPDATE folders SET loc = ? WHERE fid = ?
100 } elsif ($rw && $folder =~ m!\Aimaps?://!i) {
101 require PublicInbox::URIimap;
102 PublicInbox::URIimap->new($folder)->uidvalidity //
103 Carp::croak("BUG: $folder has no UIDVALIDITY");
107 ($fid) = $dbh->selectrow_array('SELECT MAX(fid) FROM folders');
110 # in case we're reusing, clobber existing stale refs:
111 $dbh->do('DELETE FROM blob2name WHERE fid = ?', undef, $fid);
112 $dbh->do('DELETE FROM blob2num WHERE fid = ?', undef, $fid);
114 my $sth = $dbh->prepare('INSERT INTO folders (fid, loc) VALUES (?, ?)');
115 $sth->execute($fid, $folder);
121 my ($self, $oidhex, $folder, $id) = @_;
122 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
124 if (ref($id)) { # scalar name
126 $sth = $self->{dbh}->prepare_cached(<<'');
127 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
129 } else { # numeric ID (IMAP UID, MH number)
130 $sth = $self->{dbh}->prepare_cached(<<'');
131 INSERT OR IGNORE INTO blob2num (oidbin, fid, uid) VALUES (?, ?, ?)
134 $sth->execute(pack('H*', $oidhex), $fid, $id);
138 my ($self, $folder, $id) = @_;
139 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
141 if (ref($id)) { # scalar name
143 $sth = $self->{dbh}->prepare_cached(<<'');
144 DELETE FROM blob2name WHERE fid = ? AND name = ?
147 $sth = $self->{dbh}->prepare_cached(<<'');
148 DELETE FROM blob2num WHERE fid = ? AND uid = ?
151 $sth->execute($fid, $id);
156 my ($self, $folder, $oidbin, $id, $newbn) = @_;
157 my $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder, 1);
158 my $sth = $self->{dbh}->prepare_cached(<<'');
159 UPDATE blob2name SET name = ? WHERE fid = ? AND oidbin = ? AND name = ?
161 my $nr = $sth->execute($newbn, $fid, $oidbin, $$id);
162 if ($nr == 0) { # may race with a clear_src, ensure new value exists
163 $sth = $self->{dbh}->prepare_cached(<<'');
164 INSERT OR IGNORE INTO blob2name (oidbin, fid, name) VALUES (?, ?, ?)
166 $sth->execute($oidbin, $fid, $newbn);
170 # read-only, iterates every oidbin + UID or name for a given folder
172 my ($self, $folder, $cb, @args) = @_;
173 my $dbh = $self->{dbh} //= dbh_new($self);
175 if (ref($folder) eq 'HASH') {
176 $fid = $folder->{fid} // die "BUG: no `fid'";
178 $fid = $self->{fmap}->{$folder} //=
179 fid_for($self, $folder) // return;
181 $sth = $dbh->prepare('SELECT oidbin,uid FROM blob2num WHERE fid = ?');
183 while (my ($oidbin, $id) = $sth->fetchrow_array) {
184 $cb->($oidbin, $id, @args);
186 $sth = $dbh->prepare('SELECT oidbin,name FROM blob2name WHERE fid = ?');
188 while (my ($oidbin, $id) = $sth->fetchrow_array) {
189 $cb->($oidbin, \$id, @args);
194 my ($self, $folder) = @_;
195 my $dbh = $self->{dbh} //= dbh_new($self);
198 $fid = $self->{fmap}->{$folder} //= fid_for($self, $folder) // return;
199 my ($row) = $dbh->selectrow_array(<<"", undef, $fid);
200 SELECT COUNT(name) FROM blob2name WHERE fid = ?
202 $ret->{'name.count'} = $row if $row;
203 for my $op (qw(count min max)) {
204 ($row) = $dbh->selectrow_array(<<"", undef, $fid);
205 SELECT $op(uid) FROM blob2num WHERE fid = ?
208 $ret->{"uid.$op"} = $row;
213 # returns a { location => [ list-of-ids-or-names ] } mapping
215 my ($self, $oidbin) = @_;
216 my ($fid, $sth, $id, %fid2id);
217 my $dbh = $self->{dbh} //= dbh_new($self);
218 $sth = $dbh->prepare('SELECT fid,uid FROM blob2num WHERE oidbin = ?');
219 $sth->execute($oidbin);
220 while (my ($fid, $uid) = $sth->fetchrow_array) {
221 push @{$fid2id{$fid}}, $uid;
223 $sth = $dbh->prepare('SELECT fid,name FROM blob2name WHERE oidbin = ?');
224 $sth->execute($oidbin);
225 while (my ($fid, $name) = $sth->fetchrow_array) {
226 push @{$fid2id{$fid}}, $name;
228 $sth = $dbh->prepare('SELECT loc FROM folders WHERE fid = ? LIMIT 1');
230 while (my ($fid, $ids) = each %fid2id) {
232 my ($loc) = $sth->fetchrow_array;
233 unless (defined $loc) {
234 my $oidhex = unpack('H*', $oidbin);
235 warn "E: fid=$fid for $oidhex unknown:\n", map {
236 'E: '.(ref() ? $$_ : "#$_")."\n";
242 scalar(keys %$ret) ? $ret : undef;
245 # returns a list of folders used for completion
247 my ($self, $pfx) = @_;
248 my $dbh = $self->{dbh} //= dbh_new($self);
249 my $sql = 'SELECT loc FROM folders';
252 $sql .= ' WHERE loc LIKE ? ESCAPE ?';
254 $pfx[0] =~ s/([%_\\])/\\$1/g; # glob chars
257 map { $_->[0] } @{$dbh->selectall_arrayref($sql, undef, @pfx)};
261 my ($self, $oidhex, $vrfy) = @_;
262 my $dbh = $self->{dbh} //= dbh_new($self);
263 my $b2n = $dbh->prepare(<<'');
264 SELECT f.loc,b.name FROM blob2name b
265 LEFT JOIN folders f ON b.fid = f.fid
268 $b2n->execute(pack('H*', $oidhex));
269 while (my ($d, $n) = $b2n->fetchrow_array) {
270 substr($d, 0, length('maildir:')) = '';
271 # n.b. both mbsync and offlineimap use ":2," as a suffix
272 # in "new/", despite (from what I understand of reading
273 # <https://cr.yp.to/proto/maildir.html>), the ":2," only
274 # applies to files in "cur/".
275 my @try = $n =~ /:2,[a-zA-Z]+\z/ ? qw(cur new) : qw(new cur);
278 open my $fh, '<', $f or next;
279 # some (buggy) Maildir writers are non-atomic:
283 if ($vrfy && git_sha(1, \$raw)->hexdigest ne $oidhex) {
284 warn "$f changed $oidhex\n";
294 my ($self, $url, $all) = @_; # $all = [ $lms->folders ];
295 $all //= [ $self->folders ];
296 require PublicInbox::URIimap;
297 my $want = PublicInbox::URIimap->new($url)->canonical;
298 my ($s, $h, $mb) = ($want->scheme, $want->host, $want->mailbox);
299 my @uri = map { PublicInbox::URIimap->new($_)->canonical }
300 grep(m!\A\Q$s\E://.*?\Q$h\E\b.*?/\Q$mb\E\b!, @$all);
303 next if $x->mailbox ne $want->mailbox;
304 next if $x->host ne $want->host;
305 next if $x->port != $want->port;
306 my $x_uidval = $x->uidvalidity;
307 next if ($want->uidvalidity // $x_uidval) != $x_uidval;
309 # allow nothing in want to possibly match ";AUTH=ANONYMOUS"
310 if (defined($x->auth) && !defined($want->auth) &&
311 !defined($want->user)) {
313 # or maybe user was forgotten on CLI:
314 } elsif (defined($x->user) && !defined($want->user)) {
316 } elsif (($x->user//"\0") eq ($want->user//"\0")) {
320 return @match if wantarray;
321 scalar(@match) <= 1 ? $match[0] :
322 "E: `$url' is ambiguous:\n\t".join("\n\t", @match)."\n";
325 # returns undef on failure, number on success
327 my ($self, $lei, $all, $folders) = @_;
328 return $lei->fail(<<EOM) if @$folders;
329 --all= not compatible with @$folders on command-line
331 my %x = map { $_ => $_ } split(/,/, $all);
332 my @ok = grep(defined, delete(@x{qw(local remote), ''}));
335 @no = (join(',', @no));
336 return $lei->fail(<<EOM);
337 --all=@no not accepted (must be `local' and/or `remote')
341 my @all = $self->folders;
343 if ($ok eq 'local') {
344 @inc = grep(!m!\A[a-z0-9\+]+://!i, @all);
345 } elsif ($ok eq 'remote') {
346 @inc = grep(m!\A[a-z0-9\+]+://!i, @all);
347 } elsif ($ok ne '') {
348 return $lei->fail("--all=$all not understood");
353 push(@$folders, $_) unless $seen{$_}++;
356 scalar(@$folders) || $lei->fail(<<EOM);
357 no --mail-sync folders known to lei
361 # map CLI args to folder table entries, returns undef on failure
363 my ($self, $lei, $folders) = @_;
364 my @all = $self->folders;
365 my %all = map { $_ => 1 } @all;
368 next if $all{$_}; # ok
369 if (m!\A(maildir|mh):(.+)!i) {
371 my $d = "$type:".$lei->abs_path($2);
372 push(@no, $_) unless $all{$d};
374 } elsif (-d "$_/new" && -d "$_/cur") {
375 my $d = 'maildir:'.$lei->abs_path($_);
376 push(@no, $_) unless $all{$d};
378 } elsif (m!\Aimaps?://!i) {
380 my $res = match_imap_url($self, $orig, \@all);
383 push(@{$err->{qerr}}, <<EOM);
384 # using `$res' instead of `$orig'
387 $lei->err($res) if defined $res;
395 my $no = join("\n\t", @no);
396 $err->{fail} = <<EOF;
397 No sync information for: $no
398 Run `lei ls-mail-sync' to display valid choices
405 my ($self, $folder) = @_;
407 $fid = delete($self->{fmap}->{$folder}) //
408 fid_for($self, $folder) // return;
409 my $dbh = $self->{dbh};
410 $dbh->do('DELETE FROM blob2name WHERE fid = ?', undef, $fid);
411 $dbh->do('DELETE FROM blob2num WHERE fid = ?', undef, $fid);
412 $dbh->do('DELETE FROM folders WHERE fid = ?', undef, $fid);
415 sub imap_oidbin ($$$) {
416 my ($self, $url, $uid) = @_; # $url MUST have UIDVALIDITY
417 my $fid = $self->{fmap}->{$url} //= fid_for($self, $url) // return;
418 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
419 SELECT oidbin FROM blob2num WHERE fid = ? AND uid = ?
421 $sth->execute($fid, $uid);
422 $sth->fetchrow_array;
425 sub name_oidbin ($$$) {
426 my ($self, $mdir, $nm) = @_;
427 my $fid = $self->{fmap}->{$mdir} //= fid_for($self, $mdir) // return;
428 my $sth = $self->{dbh}->prepare_cached(<<EOM, undef, 1);
429 SELECT oidbin FROM blob2name WHERE fid = ? AND name = ?
431 $sth->execute($fid, $nm);
432 $sth->fetchrow_array;
436 my ($self, $lei, $uid_uri) = @_;
437 my $mailbox_uri = $uid_uri->clone;
438 $mailbox_uri->uid(undef);
439 my $folders = [ $$mailbox_uri ];
440 if (my $err = $self->arg2folder($lei, $folders)) {
442 $lei->qerr("# no sync information for $mailbox_uri");
445 $lei->qerr(@{$err->{qerr}}) if $err->{qerr};
447 my $oidbin = imap_oidbin($self, $folders->[0], $uid_uri->uid);
448 $oidbin ? unpack('H*', $oidbin) : undef;
451 # FIXED? something with "lei <up|q>" is causing uncommitted transaction
455 my $dbh = delete($self->{dbh}) or return;
456 return if $dbh->{ReadOnly};
458 warn "BUG $$ $0 $self {dbh} OPEN ppid=".getppid.' '.Carp::longmess();