use strict;
use base qw(PublicInbox::DS);
use fields qw(imapd logged_in ibx long_cb -login_tag
- -idle_tag -idle_max);
+ uid_min -idle_tag -idle_max);
use PublicInbox::Eml;
use PublicInbox::EmlContentFoo qw(parse_content_disposition);
use PublicInbox::DS qw(now);
use PublicInbox::GitAsyncCat;
use Text::ParseWords qw(parse_line);
use Errno qw(EAGAIN);
+use Time::Local qw(timegm);
+use POSIX qw(strftime);
my $Address;
for my $mod (qw(Email::Address::XS Mail::Address)) {
sub LINE_MAX () { 512 } # does RFC 3501 have a limit like RFC 977?
+# changing this will cause grief for clients which cache
+sub UID_BLOCK () { 50_000 }
+
my %FETCH_NEED_BLOB = ( # for future optimization
'BODY[HEADER]' => 1,
'BODY[TEXT]' => 1,
my $valid_range = '[0-9]+|[0-9]+:[0-9]+|[0-9]+:\*';
$valid_range = qr/\A(?:$valid_range)(?:,(?:$valid_range))*\z/;
+my @MoY = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
+my %MoY;
+@MoY{@MoY} = (0..11);
+
sub greet ($) {
my ($self) = @_;
my $capa = capa($self);
sub cmd_close ($$) {
my ($self, $tag) = @_;
+ delete $self->{uid_min};
delete $self->{ibx} ? "$tag OK Close done\r\n"
: "$tag BAD No mailbox\r\n";
}
"$idle_tag OK Idle done\r\n";
}
+sub ensure_ranges_exist ($$$) {
+ my ($imapd, $ibx, $max) = @_;
+ my $mailboxes = $imapd->{mailboxes};
+ my $mb_top = $ibx->{newsgroup};
+ my @created;
+ for (my $i = int($max/UID_BLOCK); $i >= 0; --$i) {
+ my $sub_mailbox = "$mb_top.$i";
+ last if exists $mailboxes->{$sub_mailbox};
+ $mailboxes->{$sub_mailbox} = $ibx;
+ push @created, $sub_mailbox;
+ }
+ return unless @created;
+ my $l = $imapd->{inboxlist} or return;
+ push @$l, map { qq[* LIST (\\HasNoChildren) "." $_\r\n] } @created;
+}
+
+sub inbox_lookup ($$) {
+ my ($self, $mailbox) = @_;
+ my ($ibx, $exists, $uidnext);
+ if ($mailbox =~ /\A(.+)\.([0-9]+)\z/) {
+ # old mail: inbox.comp.foo.$uid_block_idx
+ my ($mb_top, $uid_min) = ($1, $2 * UID_BLOCK + 1);
+
+ $ibx = $self->{imapd}->{mailboxes}->{lc $mailbox} or return;
+ $exists = $ibx->mm->max // 0;
+ $self->{uid_min} = $uid_min;
+ ensure_ranges_exist($self->{imapd}, $ibx, $exists);
+ my $uid_end = $uid_min + UID_BLOCK - 1;
+ $exists = $uid_end if $exists > $uid_end;
+ $uidnext = $exists + 1;
+ } else { # check for dummy inboxes
+ $ibx = $self->{imapd}->{mailboxes}->{lc $mailbox} or return;
+ delete $self->{uid_min};
+ $exists = 0;
+ $uidnext = 1;
+ }
+ ($ibx, $exists, $uidnext);
+}
+
sub cmd_examine ($$$) {
my ($self, $tag, $mailbox) = @_;
- my $ibx = $self->{imapd}->{groups}->{$mailbox} or
- return "$tag NO Mailbox doesn't exist: $mailbox\r\n";
- my $mm = $ibx->mm;
- my $max = $mm->max // 0;
- # RFC 3501 2.3.1.1 - "A good UIDVALIDITY value to use in
- # this case is a 32-bit representation of the creation
- # date/time of the mailbox"
- my $uidvalidity = $mm->created_at or return "$tag BAD UIDVALIDITY\r\n";
- my $uidnext = $max + 1;
+ my ($ibx, $exists, $uidnext) = inbox_lookup($self, $mailbox);
+ return "$tag NO Mailbox doesn't exist: $mailbox\r\n" if !$ibx;
# XXX: do we need this? RFC 5162/7162
my $ret = $self->{ibx} ? "* OK [CLOSED] previous closed\r\n" : '';
$self->{ibx} = $ibx;
$ret .= <<EOF;
-* $max EXISTS\r
-* $max RECENT\r
+* $exists EXISTS\r
+* $exists RECENT\r
* FLAGS (\\Seen)\r
* OK [PERMANENTFLAGS ()] Read-only mailbox\r
+* OK [UNSEEN $exists]\r
+* OK [UIDNEXT $uidnext]\r
+* OK [UIDVALIDITY $ibx->{uidvalidity}]\r
+$tag OK [READ-ONLY] EXAMINE/SELECT done\r
EOF
- $ret .= "* OK [UNSEEN $max]\r\n" if $max;
- $ret .= "* OK [UIDNEXT $uidnext]\r\n" if defined $uidnext;
- $ret .= "* OK [UIDVALIDITY $uidvalidity]\r\n" if defined $uidvalidity;
- $ret .= "$tag OK [READ-ONLY] EXAMINE/SELECT done\r\n";
}
sub _esc ($) {
}
sub dummy_message ($$) {
- my ($seqno, $ibx) = @_;
+ my ($self, $seqno) = @_;
my $ret = <<EOF;
From: nobody\@localhost\r
To: nobody\@localhost\r
Date: Thu, 01 Jan 1970 00:00:00 +0000\r
-Message-ID: <dummy-$seqno\@$ibx->{newsgroup}>\r
+Message-ID: <dummy-$seqno\@$self->{ibx}->{newsgroup}>\r
Subject: dummy message #$seqno\r
\r
You're seeing this message because your IMAP client didn't use UIDs.\r
sub uid_fetch_cb { # called by git->cat_async via git_async_cat
my ($bref, $oid, $type, $size, $fetch_m_arg) = @_;
- my ($self, undef, $ibx, $msgs, undef, $want) = @$fetch_m_arg;
+ my ($self, undef, $msgs, undef, $want) = @$fetch_m_arg;
my $smsg = shift @$msgs or die 'BUG: no smsg';
if (!defined($oid)) {
# it's possible to have TOCTOU if an admin runs
# public-inbox-(edit|purge), just move onto the next message
return requeue_once($self) unless defined $want->{-seqno};
- $bref = dummy_message($smsg->{num}, $ibx);
+ $bref = dummy_message($self, $smsg->{num});
} else {
$smsg->{blob} eq $oid or die "BUG: $smsg->{blob} != $oid";
}
requeue_once($self);
}
+sub uid_clamp ($$$) {
+ my ($self, $beg, $end) = @_;
+ my $uid_min = $self->{uid_min} or return;
+ my $uid_end = $uid_min + UID_BLOCK - 1;
+ $$beg = $uid_min if $$beg < $uid_min;
+ $$end = $uid_end if $$end > $uid_end;
+}
+
sub range_step ($$) {
- my ($ibx, $range_csv) = @_;
+ my ($self, $range_csv) = @_;
my ($beg, $end, $range);
if ($$range_csv =~ s/\A([^,]+),//) {
$range = $1;
($beg, $end) = ($1 + 0, $2 + 0);
} elsif ($range =~ /\A([0-9]+):\*\z/) {
$beg = $1 + 0;
- $end = $ibx->mm->max // 0;
+ $end = $self->{ibx}->mm->max // 0;
+ my $uid_end = ($self->{uid_min} // 1) - 1 + UID_BLOCK;
+ $end = $uid_end if $end > $uid_end;
$beg = $end if $beg > $end;
} elsif ($range =~ /\A[0-9]+\z/) {
$beg = $end = $range + 0;
+ undef $range;
} else {
return 'BAD fetch range';
}
+ uid_clamp($self, \$beg, \$end) if defined($range);
[ $beg, $end, $$range_csv ];
}
sub refill_range ($$$) {
- my ($ibx, $msgs, $range_info) = @_;
+ my ($self, $msgs, $range_info) = @_;
my ($beg, $end, $range_csv) = @$range_info;
- if (scalar(@$msgs = @{$ibx->over->query_xover($beg, $end)})) {
+ if (scalar(@$msgs = @{$self->{ibx}->over->query_xover($beg, $end)})) {
$range_info->[0] = $msgs->[-1]->{num} + 1;
return;
}
return 'OK Fetch done' if !$range_csv;
- my $next_range = range_step($ibx, \$range_csv);
+ my $next_range = range_step($self, \$range_csv);
return $next_range if !ref($next_range); # error
@$range_info = @$next_range;
undef; # keep looping
}
sub uid_fetch_m { # long_response
- my ($self, $tag, $ibx, $msgs, $range_info, $want) = @_;
+ my ($self, $tag, $msgs, $range_info, $want) = @_;
while (!@$msgs) { # rare
- if (my $end = refill_range($ibx, $msgs, $range_info)) {
+ if (my $end = refill_range($self, $msgs, $range_info)) {
$self->write(\"$tag $end\r\n");
return;
}
}
- git_async_cat($ibx->git, $msgs->[0]->{blob}, \&uid_fetch_cb, \@_);
+ git_async_cat($self->{ibx}->git, $msgs->[0]->{blob},
+ \&uid_fetch_cb, \@_);
}
sub cmd_status ($$$;@) {
my ($self, $tag, $mailbox, @items) = @_;
- my $ibx = $self->{imapd}->{groups}->{$mailbox} or
- return "$tag NO Mailbox doesn't exist: $mailbox\r\n";
return "$tag BAD no items\r\n" if !scalar(@items);
($items[0] !~ s/\A\(//s || $items[-1] !~ s/\)\z//s) and
return "$tag BAD invalid args\r\n";
-
- my $mm = $ibx->mm;
- my ($max, @it);
+ my ($ibx, $exists, $uidnext) = inbox_lookup($self, $mailbox);
+ return "$tag NO Mailbox doesn't exist: $mailbox\r\n" if !$ibx;
+ my @it;
for my $it (@items) {
$it = uc($it);
push @it, $it;
if ($it =~ /\A(?:MESSAGES|UNSEEN|RECENT)\z/) {
- push(@it, ($max //= $mm->max // 0));
+ push @it, $exists;
} elsif ($it eq 'UIDNEXT') {
- push(@it, ($max //= $mm->max // 0) + 1);
+ push @it, $uidnext;
} elsif ($it eq 'UIDVALIDITY') {
- push(@it, $mm->created_at //
- return("$tag BAD UIDVALIDITY\r\n"));
+ push @it, $ibx->{uidvalidity};
} else {
return "$tag BAD invalid item\r\n";
}
# request for hierarchy delimiter
$l = [ qq[* LIST (\\Noselect) "." ""\r\n] ];
} elsif ($refname ne '' || $wildcard ne '*') {
- $wildcard =~ s!([^a-z0-9_])!$patmap{$1} // "\Q$1"!eig;
+ $wildcard = lc $wildcard;
+ $wildcard =~ s!([^a-z0-9_])!$patmap{$1} // "\Q$1"!eg;
$l = [ grep(/ \Q$refname\E$wildcard\r\n\z/s, @$l) ];
}
\(join('', @$l, "$tag OK List done\r\n"));
} sort keys %partial ];
}
$range_csv = 'bad' if $range_csv !~ $valid_range;
- my $range_info = range_step($ibx, \$range_csv);
+ my $range_info = range_step($self, \$range_csv);
return "$tag $range_info\r\n" if !ref($range_info);
- [ $tag, $ibx, [], $range_info, \%want ];
+ [ $tag, [], $range_info, \%want ];
}
sub cmd_uid_fetch ($$$;@) {
}
sub seq_fetch_m { # long_response
- my ($self, $tag, $ibx, $msgs, $range_info, $want) = @_;
+ my ($self, $tag, $msgs, $range_info, $want) = @_;
while (!@$msgs) { # rare
- if (my $end = refill_range($ibx, $msgs, $range_info)) {
+ if (my $end = refill_range($self, $msgs, $range_info)) {
$self->write(\"$tag $end\r\n");
return;
}
my $seq = $want->{-seqno}++;
my $cur_num = $msgs->[0]->{num};
if ($cur_num == $seq) { # as expected
- git_async_cat($ibx->git, $msgs->[0]->{blob},
+ git_async_cat($self->{ibx}->git, $msgs->[0]->{blob},
\&uid_fetch_cb, \@_);
} elsif ($cur_num > $seq) {
# send dummy messages until $seq catches up to $cur_num
my $smsg = bless { num => $seq, ts => 0 }, 'PublicInbox::Smsg';
unshift @$msgs, $smsg;
- my $bref = dummy_message($seq, $ibx);
+ my $bref = dummy_message($self, $seq);
uid_fetch_cb($bref, undef, undef, undef, \@_);
$smsg; # blessed response since uid_fetch_cb requeues
} else { # should not happen
my $args = fetch_common($self, $tag, $range_csv, \@want);
ref($args) eq 'ARRAY' ? do {
my $want = $args->[-1];
- $want->{-seqno} = $args->[3]->[0]; # $beg == $range_info->[0];
+ $want->{-seqno} = $args->[2]->[0]; # $beg == $range_info->[0];
long_response($self, \&seq_fetch_m, @$args)
} : $args; # error
}
-sub uid_search_all { # long_response
- my ($self, $tag, $ibx, $num) = @_;
- my $uids = $ibx->mm->ids_after($num);
- if (scalar(@$uids)) {
- $self->msg_more(join(' ', '', @$uids));
- } else {
- $self->write(\"\r\n$tag OK Search done\r\n");
- undef;
- }
+
+sub parse_date ($) { # 02-Oct-1993
+ my ($date_text) = @_;
+ my ($dd, $mon, $yyyy) = split(/-/, $_[0], 3);
+ defined($yyyy) or return;
+ my $mm = $MoY{$mon} // return;
+ $dd =~ /\A[0123]?[0-9]\z/ or return;
+ $yyyy =~ /\A[0-9]{4,}\z/ or return; # Y10K-compatible!
+ timegm(0, 0, 0, $dd, $mm, $yyyy);
}
sub uid_search_uid_range { # long_response
- my ($self, $tag, $ibx, $beg, $end) = @_;
- my $uids = $ibx->mm->msg_range($beg, $end, 'num');
+ my ($self, $tag, $beg, $end) = @_;
+ my $uids = $self->{ibx}->mm->msg_range($beg, $end, 'num');
if (@$uids) {
$self->msg_more(join('', map { " $_->[0]" } @$uids));
} else {
}
}
+sub date_search {
+ my ($q, $k, $d) = @_;
+ my $sql = $q->{sql};
+
+ # Date: header
+ if ($k eq 'SENTON') {
+ my $end = $d + 86399; # no leap day...
+ my $da = strftime('%Y%m%d%H%M%S', gmtime($d));
+ my $db = strftime('%Y%m%d%H%M%S', gmtime($end));
+ $q->{xap} .= " dt:$da..$db";
+ $$sql .= " AND ds >= $d AND ds <= $end" if defined($sql);
+ } elsif ($k eq 'SENTBEFORE') {
+ $q->{xap} .= ' d:..'.strftime('%Y%m%d', gmtime($d));
+ $$sql .= " AND ds <= $d" if defined($sql);
+ } elsif ($k eq 'SENTSINCE') {
+ $q->{xap} .= ' d:'.strftime('%Y%m%d', gmtime($d)).'..';
+ $$sql .= " AND ds >= $d" if defined($sql);
+
+ # INTERNALDATE (Received)
+ } elsif ($k eq 'ON') {
+ my $end = $d + 86399; # no leap day...
+ $q->{xap} .= " ts:$d..$end";
+ $$sql .= " AND ts >= $d AND ts <= $end" if defined($sql);
+ } elsif ($k eq 'BEFORE') {
+ $q->{xap} .= " ts:..$d";
+ $$sql .= " AND ts <= $d" if defined($sql);
+ } elsif ($k eq 'SINCE') {
+ $q->{xap} .= " ts:$d..";
+ $$sql .= " AND ts >= $d" if defined($sql);
+ } else {
+ die "BUG: $k not recognized";
+ }
+}
+
+# IMAP to Xapian search key mapping
+my %I2X = (
+ SUBJECT => 's:',
+ BODY => 'b:',
+ FROM => 'f:',
+ TEXT => '', # n.b. does not include all headers
+ TO => 't:',
+ CC => 'c:',
+ # BCC => 'bcc:', # TODO
+ # KEYWORD # TODO ? dfpre,dfpost,...
+);
+
+sub parse_query {
+ my ($self, $rest) = @_;
+ if (uc($rest->[0]) eq 'CHARSET') {
+ shift @$rest;
+ defined(my $c = shift @$rest) or return 'BAD missing charset';
+ $c =~ /\A(?:UTF-8|US-ASCII)\z/ or return 'NO [BADCHARSET]';
+ }
+
+ my $sql = ''; # date conditions, {sql} deleted if Xapian is needed
+ my $q = { xap => '', sql => \$sql };
+ while (@$rest) {
+ my $k = uc(shift @$rest);
+ # default criteria
+ next if $k =~ /\A(?:ALL|RECENT|UNSEEN|NEW)\z/;
+ next if $k eq 'AND'; # the default, until we support OR
+ if ($k =~ $valid_range) { # sequence numbers == UIDs
+ push @{$q->{uid}}, $k;
+ } elsif ($k eq 'UID') {
+ $k = shift(@$rest) // '';
+ $k =~ $valid_range or return 'BAD UID range';
+ push @{$q->{uid}}, $k;
+ } elsif ($k =~ /\A(?:SENT)?(?:SINCE|ON|BEFORE)\z/) {
+ my $d = parse_date(shift(@$rest) // '');
+ defined $d or return "BAD $k date format";
+ date_search($q, $k, $d);
+ } elsif ($k =~ /\A(?:SMALLER|LARGER)\z/) {
+ delete $q->{sql}; # can't use over.sqlite3
+ my $bytes = shift(@$rest) // '';
+ $bytes =~ /\A[0-9]+\z/ or return "BAD $k not a number";
+ $q->{xap} .= ' bytes:' . ($k eq 'SMALLER' ?
+ '..'.(--$bytes) :
+ (++$bytes).'..');
+ } elsif (defined(my $xk = $I2X{$k})) {
+ delete $q->{sql}; # can't use over.sqlite3
+ my $arg = shift @$rest;
+ defined($arg) or return "BAD $k no arg";
+
+ # Xapian can't handle [*"] in probabilistic terms
+ $arg =~ tr/*"//d;
+ $q->{xap} .= qq[ $xk:"$arg"];
+ } else {
+ # TODO: parentheses, OR, NOT ...
+ return "BAD $k not supported (yet?)";
+ }
+ }
+
+ # favor using over.sqlite3 if possible, since Xapian is optional
+ if (exists $q->{sql}) {
+ delete($q->{xap});
+ delete($q->{sql}) if $sql eq '';
+ } elsif (!$self->{ibx}->search) {
+ return 'BAD Xapian not configured for mailbox';
+ }
+
+ if (my $uid = $q->{uid}) {
+ ((@$uid > 1) || $uid->[0] =~ /,/) and
+ return 'BAD multiple ranges not supported, yet';
+ ($q->{sql} // $q->{xap}) and
+ return 'BAD ranges and queries do not mix, yet';
+ $q->{uid} = join(',', @$uid); # TODO: multiple ranges
+ }
+ $q;
+}
+
sub cmd_uid_search ($$$;) {
- my ($self, $tag, $arg, @rest) = @_;
+ my ($self, $tag) = splice(@_, 0, 2);
my $ibx = $self->{ibx} or return "$tag BAD No mailbox selected\r\n";
- $arg = uc($arg);
- if ($arg eq 'ALL' && !@rest) {
+ my $q = parse_query($self, \@_);
+ return "$tag $q\r\n" if !ref($q);
+
+ if (!scalar(keys %$q)) {
$self->msg_more('* SEARCH');
- my $num = 0;
- long_response($self, \&uid_search_all, $tag, $ibx, \$num);
- } elsif ($arg eq 'UID' && scalar(@rest) == 1) {
- if ($rest[0] =~ /\A([0-9]+):([0-9]+|\*)\z/s) {
+ my $beg = $self->{uid_min} // 1;
+ my $end = $ibx->mm->max;
+ uid_clamp($self, \$beg, \$end);
+ long_response($self, \&uid_search_uid_range, $tag, \$beg, $end);
+ } elsif (my $uid = $q->{uid}) {
+ if ($uid =~ /\A([0-9]+):([0-9]+|\*)\z/s) {
my ($beg, $end) = ($1, $2);
$end = $ibx->mm->max if $end eq '*';
+ uid_clamp($self, \$beg, \$end);
$self->msg_more('* SEARCH');
long_response($self, \&uid_search_uid_range,
- $tag, $ibx, \$beg, $end);
- } elsif ($rest[0] =~ /\A[0-9]+\z/s) {
- my $uid = $rest[0];
+ $tag, \$beg, $end);
+ } elsif ($uid =~ /\A[0-9]+\z/s) {
$uid = $ibx->over->get_art($uid) ? " $uid" : '';
"* SEARCH$uid\r\n$tag OK Search done\r\n";
} else {
cmd_done($self, $tag);
} else { # this is weird
auth_challenge_ok($self) //
- "$tag BAD Error in IMAP command $req: ".
- "Unknown command\r\n";
+ ($tag // '*') .
+ ' BAD Error in IMAP command '.
+ ($req // '(???)').
+ ": Unknown command\r\n";
}
};
my $err = $@;
if ($err && $self->{sock}) {
$l =~ s/\r?\n//s;
err($self, 'error from: %s (%s)', $l, $err);
+ $tag //= '*';
$res = "$tag BAD program fault - command not performed\r\n";
}
return 0 unless defined $res;