# data notes:
# * NNTP article numbers are UIDs and message sequence numbers (MSNs)
# * Message sequence numbers (MSNs) can be stable since we're read-only.
-# Most IMAP clients use UIDs (I hope), and we can return a dummy
-# message if a client requests a non-existent MSN.
+# Most IMAP clients use UIDs (I hope). We may return a dummy message
+# in the future if a client requests a non-existent MSN, but that seems
+# unecessary with mutt.
package PublicInbox::IMAP;
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::Syscall qw(EPOLLIN EPOLLONESHOT);
+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)) {
eval "require $mod" or next;
sub LINE_MAX () { 512 } # does RFC 3501 have a limit like RFC 977?
-my %FETCH_NEED_BLOB = ( # for future optimization
- 'BODY.PEEK[HEADER]' => 1,
- 'BODY.PEEK[TEXT]' => 1,
- 'BODY.PEEK[]' => 1,
- 'BODY[HEADER]' => 1,
- 'BODY[TEXT]' => 1,
- 'BODY[]' => 1,
- 'RFC822.HEADER' => 1,
- 'RFC822.SIZE' => 1, # needs CRLF conversion :<
- 'RFC822.TEXT' => 1,
- BODY => 1,
- BODYSTRUCTURE => 1,
- ENVELOPE => 1,
- FLAGS => 0,
- INTERNALDATE => 0,
- RFC822 => 1,
- UID => 0,
+# changing this will cause grief for clients which cache
+sub UID_BLOCK () { 50_000 }
+
+# these values area also used for sorting
+sub NEED_BLOB () { 1 }
+sub NEED_EML () { NEED_BLOB|2 }
+my $OP_EML_NEW = [ NEED_EML - 1, \&op_eml_new ];
+
+my %FETCH_NEED = ( # for future optimization
+ 'BODY[HEADER]' => [ NEED_EML, \&emit_rfc822_header ],
+ 'BODY[TEXT]' => [ NEED_EML, \&emit_rfc822_text ],
+ 'BODY[]' => [ NEED_BLOB, \&emit_rfc822 ],
+ 'RFC822.HEADER' => [ NEED_EML, \&emit_rfc822_header ],
+ 'RFC822.TEXT' => [ NEED_EML, \&emit_rfc822_text ],
+ 'RFC822.SIZE' => [ NEED_BLOB, \&emit_rfc822_size ],
+ RFC822 => [ NEED_BLOB, \&emit_rfc822 ],
+ BODY => [ NEED_EML, \&emit_body ],
+ BODYSTRUCTURE => [ NEED_EML, \&emit_bodystructure ],
+ ENVELOPE => [ NEED_EML, \&emit_envelope ],
+ FLAGS => [ 0, \&emit_flags ],
+ INTERNALDATE => [ 0, \&emit_internaldate ],
);
-my %FETCH_ATT = map { $_ => [ $_ ] } keys %FETCH_NEED_BLOB;
+my %FETCH_ATT = map { $_ => [ $_ ] } keys %FETCH_NEED;
# aliases (RFC 3501 section 6.4.5)
$FETCH_ATT{FAST} = [ qw(FLAGS INTERNALDATE RFC822.SIZE) ];
$FETCH_ATT{FULL} = [ @{$FETCH_ATT{ALL}}, 'BODY' ];
for my $att (keys %FETCH_ATT) {
- my %h = map { $_ => 1 } @{$FETCH_ATT{$att}};
+ my %h = map { $_ => $FETCH_NEED{$_} } @{$FETCH_ATT{$att}};
$FETCH_ATT{$att} = \%h;
}
+undef %FETCH_NEED;
+
+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) = @_;
login_success($self, $tag);
}
+sub cmd_close ($$) {
+ my ($self, $tag) = @_;
+ delete $self->{uid_min};
+ delete $self->{ibx} ? "$tag OK Close done\r\n"
+ : "$tag BAD No mailbox\r\n";
+}
+
sub cmd_logout ($$) {
my ($self, $tag) = @_;
delete $self->{logged_in};
- $self->write(\"* BYE logging out\r\n$tag OK logout completed\r\n");
+ $self->write(\"* BYE logging out\r\n$tag OK Logout done\r\n");
$self->shutdn; # PublicInbox::DS::shutdn
undef;
}
sub cmd_capability ($$) {
my ($self, $tag) = @_;
- '* '.capa($self)."\r\n$tag OK\r\n";
+ '* '.capa($self)."\r\n$tag OK Capability done\r\n";
}
-sub cmd_noop ($$) { "$_[1] OK NOOP completed\r\n" }
+sub cmd_noop ($$) { "$_[1] OK Noop done\r\n" }
# called by PublicInbox::InboxIdle
sub on_inbox_unlock {
return "$tag BAD internal bug\r\n";
};
$ibx->unsubscribe_unlock(fileno($self->{sock}));
- "$idle_tag OK Idle completed\r\n";
+ "$idle_tag OK Idle done\r\n";
+}
+
+sub ensure_ranges_exist ($$$) {
+ 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 complete\r\n";
}
sub _esc ($) {
join('', @hold);
}
-sub uid_fetch_cb { # called by git->cat_async
+sub requeue_once ($) {
+ my ($self) = @_;
+ # COMPRESS users all share the same DEFLATE context.
+ # Flush it here to ensure clients don't see
+ # each other's data
+ $self->zflush;
+
+ # no recursion, schedule another call ASAP,
+ # but only after all pending writes are done.
+ # autovivify wbuf:
+ my $new_size = push(@{$self->{wbuf}}, \&long_step);
+
+ # wbuf may be populated by $cb, no need to rearm if so:
+ $self->requeue if $new_size == 1;
+}
+
+sub uid_fetch_cb { # called by git->cat_async via git_async_cat
my ($bref, $oid, $type, $size, $fetch_m_arg) = @_;
- my ($self, undef, $ibx, undef, undef, $msgs, $want) = @$fetch_m_arg;
+ my ($self, undef, $msgs, undef, $ops, $partial) = @$fetch_m_arg;
my $smsg = shift @$msgs or die 'BUG: no smsg';
- $smsg->{blob} eq $oid or die "BUG: $smsg->{blob} != $oid";
+ if (!defined($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);
+ } else {
+ $smsg->{blob} eq $oid or die "BUG: $smsg->{blob} != $oid";
+ }
$$bref =~ s/(?<!\r)\n/\r\n/sg; # make strict clients happy
# fixup old bug from import (pre-a0c07cba0e5d8b6a)
$$bref =~ s/\A[\r\n]*From [^\r\n]*\r\n//s;
-
$self->msg_more("* $smsg->{num} FETCH (UID $smsg->{num}");
-
- $want->{'RFC822.SIZE'} and
- $self->msg_more(' RFC822.SIZE '.length($$bref));
- $want->{INTERNALDATE} and
- $self->msg_more(' INTERNALDATE "'.$smsg->internaldate.'"');
- $want->{FLAGS} and $self->msg_more(' FLAGS ()');
- for ('RFC822', 'BODY[]', 'BODY.PEEK[]') {
- next unless $want->{$_};
- $self->msg_more(" $_ {".length($$bref)."}\r\n");
- $self->msg_more($$bref);
+ my $eml;
+ for (my $i = 0; $i < @$ops;) {
+ my $k = $ops->[$i++];
+ $ops->[$i++]->($self, $k, $smsg, $bref, $eml);
}
+ partial_emit($self, $partial, $eml) if $partial;
+ $self->msg_more(")\r\n");
+ requeue_once($self);
+}
+
+sub emit_rfc822 {
+ my ($self, $k, undef, $bref) = @_;
+ $self->msg_more(" $k {" . length($$bref)."}\r\n");
+ $self->msg_more($$bref);
+}
+
+# Mail::IMAPClient::message_string cares about this by default
+# (->Ignoresizeerrors attribute)
+sub emit_rfc822_size {
+ my ($self, $k, undef, $bref) = @_;
+ $self->msg_more(' RFC822.SIZE ' . length($$bref));
+}
+
+sub emit_internaldate {
+ my ($self, undef, $smsg) = @_;
+ $self->msg_more(' INTERNALDATE "'.$smsg->internaldate.'"');
+}
+
+sub emit_flags { $_[0]->msg_more(' FLAGS ()') }
+
+sub emit_envelope {
+ my ($self, undef, undef, undef, $eml) = @_;
+ $self->msg_more(' ENVELOPE '.eml_envelope($eml));
+}
- my $eml = PublicInbox::Eml->new($bref);
+sub emit_rfc822_header {
+ my ($self, $k, undef, undef, $eml) = @_;
+ $self->msg_more(" $k {".length(${$eml->{hdr}})."}\r\n");
+ $self->msg_more(${$eml->{hdr}});
+}
+
+# n.b. this is sorted to be after any emit_eml_new ops
+sub emit_rfc822_text {
+ my ($self, $k, undef, $bref) = @_;
+ $self->msg_more(" $k {".length($$bref)."}\r\n");
+ $self->msg_more($$bref);
+}
+
+sub emit_bodystructure {
+ my ($self, undef, undef, undef, $eml) = @_;
+ $self->msg_more(' BODYSTRUCTURE '.fetch_body($eml, 1));
+}
+
+sub emit_body {
+ my ($self, undef, undef, undef, $eml) = @_;
+ $self->msg_more(' BODY '.fetch_body($eml));
+}
- $want->{ENVELOPE} and
- $self->msg_more(' ENVELOPE '.eml_envelope($eml));
+# set $eml once ($_[4] == $eml, $_[3] == $bref)
+sub op_eml_new { $_[4] = PublicInbox::Eml->new($_[3]) }
- for my $f ('RFC822.HEADER', 'BODY[HEADER]', 'BODY.PEEK[HEADER]') {
- next unless $want->{$f};
- $self->msg_more(" $f {".length(${$eml->{hdr}})."}\r\n");
- $self->msg_more(${$eml->{hdr}});
+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 ($self, $range_csv) = @_;
+ my ($beg, $end, $range);
+ if ($$range_csv =~ s/\A([^,]+),//) {
+ $range = $1;
+ } else {
+ $range = $$range_csv;
+ $$range_csv = undef;
}
- for my $f ('RFC822.TEXT', 'BODY[TEXT]') {
- next unless $want->{$f};
- $self->msg_more(" $f {".length($$bref)."}\r\n");
- $self->msg_more($$bref);
+ if ($range =~ /\A([0-9]+):([0-9]+)\z/) {
+ ($beg, $end) = ($1 + 0, $2 + 0);
+ } elsif ($range =~ /\A([0-9]+):\*\z/) {
+ $beg = $1 + 0;
+ $end = $self->{ibx}->mm->max // 0;
+ my $uid_end = ($self->{uid_min} // 1) - 1 + UID_BLOCK;
+ $end = $uid_end if $end > $uid_end;
+ $beg = $end if $beg > $end;
+ } elsif ($range =~ /\A[0-9]+\z/) {
+ $beg = $end = $range + 0;
+ undef $range;
+ } else {
+ return 'BAD fetch range';
}
- $want->{BODYSTRUCTURE} and
- $self->msg_more(' BODYSTRUCTURE '.fetch_body($eml, 1));
- $want->{BODY} and
- $self->msg_more(' BODY '.fetch_body($eml));
+ uid_clamp($self, \$beg, \$end) if defined($range);
+ [ $beg, $end, $$range_csv ];
+}
- $self->msg_more(")\r\n");
+sub refill_range ($$$) {
+ my ($self, $msgs, $range_info) = @_;
+ my ($beg, $end, $range_csv) = @$range_info;
+ if (scalar(@$msgs = @{$self->{ibx}->over->query_xover($beg, $end)})) {
+ $range_info->[0] = $msgs->[-1]->{num} + 1;
+ return;
+ }
+ return 'OK Fetch done' if !$range_csv;
+ my $next_range = range_step($self, \$range_csv);
+ return $next_range if !ref($next_range); # error
+ @$range_info = @$next_range;
+ undef; # keep looping
}
sub uid_fetch_m { # long_response
- my ($self, $tag, $ibx, $beg, $end, $msgs, $want) = @_;
- if (!@$msgs) { # refill
- @$msgs = @{$ibx->over->query_xover($$beg, $end)};
- if (!@$msgs) {
- $self->write(\"$tag OK Fetch done\r\n");
+ my ($self, $tag, $msgs, $range_info) = @_; # \@ops, \@partial
+ while (!@$msgs) { # rare
+ if (my $end = refill_range($self, $msgs, $range_info)) {
+ $self->write(\"$tag $end\r\n");
return;
}
- $$beg = $msgs->[-1]->{num} + 1;
}
- my $git = $ibx->git;
- $git->cat_async_begin; # TODO: actually make async
- $git->cat_async($msgs->[0]->{blob}, \&uid_fetch_cb, \@_);
- $git->cat_async_wait;
- 1;
+ git_async_cat($self->{ibx}->git, $msgs->[0]->{blob},
+ \&uid_fetch_cb, \@_);
}
sub cmd_status ($$$;@) {
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";
}
}
return "$tag BAD no items\r\n" if !@it;
"* STATUS $mailbox (".join(' ', @it).")\r\n" .
- "$tag OK Status complete\r\n";
+ "$tag OK Status done\r\n";
}
my %patmap = ('*' => '.*', '%' => '[^\.]*');
# 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 complete\r\n"));
+ \(join('', @$l, "$tag OK List done\r\n"));
}
-sub cmd_uid_fetch ($$$;@) {
- my ($self, $tag, $range, @want) = @_;
- my $ibx = $self->{ibx} or return "$tag BAD No mailbox selected\r\n";
- if ($want[0] =~ s/\A\(//s) {
- $want[-1] =~ s/\)\z//s or return "$tag BAD no rparen\r\n";
- }
- my %want = map {;
- my $x = $FETCH_ATT{uc($_)} or return "$tag BAD param: $_\r\n";
- %$x;
- } @want;
- my ($beg, $end);
- my $msgs = [];
- if ($range =~ /\A([0-9]+):([0-9]+)\z/s) {
- ($beg, $end) = ($1, $2);
- } elsif ($range =~ /\A([0-9]+):\*\z/s) {
- ($beg, $end) = ($1, $ibx->mm->max // 0);
- } elsif ($range =~ /\A[0-9]+\z/) {
- my $smsg = $ibx->over->get_art($range) or return "$tag OK\r\n";
- push @$msgs, $smsg;
- ($beg, $end) = ($range, 0);
- } else {
- return "$tag BAD\r\n";
+sub cmd_lsub ($$$$) {
+ my (undef, $tag) = @_; # same args as cmd_list
+ "$tag OK Lsub done\r\n";
+}
+
+sub eml_index_offs_i { # PublicInbox::Eml::each_part callback
+ my ($p, $all) = @_;
+ my ($eml, undef, $idx) = @$p;
+ if ($idx && lc($eml->ct->{type}) eq 'multipart') {
+ $eml->{imap_bdy} = $eml->{bdy} // \'';
}
- long_response($self, \&uid_fetch_m, $tag, $ibx,
- \$beg, $end, $msgs, \%want);
+ $all->{$idx} = $eml; # $idx => Eml
}
-sub uid_search_all { # long_response
- my ($self, $tag, $ibx, $num) = @_;
- my $uids = $ibx->mm->ids_after($num);
- if (scalar(@$uids)) {
- $self->msg_more(join(' ', '', @$uids));
+# prepares an index for BODY[$SECTION_IDX] fetches
+sub eml_body_idx ($$) {
+ my ($eml, $section_idx) = @_;
+ my $idx = $eml->{imap_all_parts} //= do {
+ my $all = {};
+ $eml->each_part(\&eml_index_offs_i, $all, 0, 1);
+ # top-level of multipart, BODY[0] not allowed (nz-number)
+ delete $all->{0};
+ $all;
+ };
+ $idx->{$section_idx};
+}
+
+# BODY[($SECTION_IDX)?(.$SECTION_NAME)?]<$offset.$bytes>
+sub partial_body {
+ my ($eml, $section_idx, $section_name) = @_;
+ if (defined $section_idx) {
+ $eml = eml_body_idx($eml, $section_idx) or return;
+ }
+ if (defined $section_name) {
+ if ($section_name eq 'MIME') {
+ # RFC 3501 6.4.5 states:
+ # The MIME part specifier MUST be prefixed
+ # by one or more numeric part specifiers
+ return unless defined $section_idx;
+ return $eml->header_obj->as_string . "\r\n";
+ }
+ my $bdy = $eml->{bdy} // $eml->{imap_bdy} // \'';
+ $eml = PublicInbox::Eml->new($$bdy);
+ if ($section_name eq 'TEXT') {
+ return $eml->body_raw;
+ } elsif ($section_name eq 'HEADER') {
+ return $eml->header_obj->as_string . "\r\n";
+ } else {
+ die "BUG: bad section_name=$section_name";
+ }
+ }
+ ${$eml->{bdy} // $eml->{imap_bdy} // \''};
+}
+
+# similar to what's in PublicInbox::Eml::re_memo, but doesn't memoize
+# to avoid OOM with malicious users
+sub hdrs_regexp ($) {
+ my ($hdrs) = @_;
+ my $names = join('|', map { "\Q$_" } split(/[ \t]+/, $hdrs));
+ qr/^(?:$names):[ \t]*[^\n]*\r?\n # 1st line
+ # continuation lines:
+ (?:[^:\n]*?[ \t]+[^\n]*\r?\n)*
+ /ismx;
+}
+
+# BODY[($SECTION_IDX.)?HEADER.FIELDS.NOT ($HDRS)]<$offset.$bytes>
+sub partial_hdr_not {
+ my ($eml, $section_idx, $hdrs_re) = @_;
+ if (defined $section_idx) {
+ $eml = eml_body_idx($eml, $section_idx) or return;
+ }
+ my $str = $eml->header_obj->as_string;
+ $str =~ s/$hdrs_re//g;
+ $str .= "\r\n";
+}
+
+# BODY[($SECTION_IDX.)?HEADER.FIELDS ($HDRS)]<$offset.$bytes>
+sub partial_hdr_get {
+ my ($eml, $section_idx, $hdrs_re) = @_;
+ if (defined $section_idx) {
+ $eml = eml_body_idx($eml, $section_idx) or return;
+ }
+ my $str = $eml->header_obj->as_string;
+ join('', ($str =~ m/($hdrs_re)/g), "\r\n");
+}
+
+sub partial_prepare ($$$) {
+ my ($partial, $want, $att) = @_;
+
+ # recombine [ "BODY[1.HEADER.FIELDS", "(foo", "bar)]" ]
+ # back to: "BODY[1.HEADER.FIELDS (foo bar)]"
+ return unless $att =~ /\ABODY\[/s;
+ until (rindex($att, ']') >= 0) {
+ my $next = shift @$want or return;
+ $att .= ' ' . uc($next);
+ }
+ if ($att =~ /\ABODY\[([0-9]+(?:\.[0-9]+)*)? # 1 - section_idx
+ (?:\.(HEADER|MIME|TEXT))? # 2 - section_name
+ \](?:<([0-9]+)(?:\.([0-9]+))?>)?\z/sx) { # 3, 4
+ $partial->{$att} = [ \&partial_body, $1, $2, $3, $4 ];
+ } elsif ($att =~ /\ABODY\[(?:([0-9]+(?:\.[0-9]+)*)\.)? # 1 - section_idx
+ (?:HEADER\.FIELDS(\.NOT)?)\x20 # 2
+ \(([A-Z0-9\-\x20]+)\) # 3 - hdrs
+ \](?:<([0-9]+)(?:\.([0-9]+))?>)?\z/sx) { # 4 5
+ my $tmp = $partial->{$att} = [ $2 ? \&partial_hdr_not
+ : \&partial_hdr_get,
+ $1, undef, $4, $5 ];
+ $tmp->[2] = hdrs_regexp($3);
} else {
- $self->write(\"\r\n$tag OK\r\n");
undef;
}
}
+sub partial_emit ($$$) {
+ my ($self, $partial, $eml) = @_;
+ for (@$partial) {
+ my ($k, $cb, @args) = @$_;
+ my ($offset, $len) = splice(@args, -2);
+ # $cb is partial_body|partial_hdr_get|partial_hdr_not
+ my $str = $cb->($eml, @args) // '';
+ if (defined $offset) {
+ if (defined $len) {
+ $str = substr($str, $offset, $len);
+ $k =~ s/\.$len>\z/>/ or warn
+"BUG: unable to remove `.$len>' from `$k'";
+ } else {
+ $str = substr($str, $offset);
+ $len = length($str);
+ }
+ } else {
+ $len = length($str);
+ }
+ $self->msg_more(" $k {$len}\r\n");
+ $self->msg_more($str);
+ }
+}
+
+sub fetch_compile ($) {
+ my ($want) = @_;
+ if ($want->[0] =~ s/\A\(//s) {
+ $want->[-1] =~ s/\)\z//s or return 'BAD no rparen';
+ }
+ my (%partial, %seen, @op);
+ my $need = 0;
+ while (defined(my $att = shift @$want)) {
+ $att = uc($att);
+ next if $att eq 'UID'; # always returned
+ $att =~ s/\ABODY\.PEEK\[/BODY\[/; # we're read-only
+ my $x = $FETCH_ATT{$att};
+ if ($x) {
+ while (my ($k, $fl_cb) = each %$x) {
+ next if $seen{$k}++;
+ $need |= $fl_cb->[0];
+
+ # insert a special op to convert $bref to $eml
+ # the first time we need it
+ if ($need == NEED_EML && !$seen{$need}++) {
+ push @op, $OP_EML_NEW;
+ }
+ # $fl_cb = [ flags, \&emit_foo ]
+ push @op, [ @$fl_cb , $k ];
+ }
+ } elsif (!partial_prepare(\%partial, $want, $att)) {
+ return "BAD param: $att";
+ }
+ }
+ my @r;
+
+ # stabilize partial order for consistency and ease-of-debugging:
+ if (scalar keys %partial) {
+ $need = NEED_EML;
+ push @op, $OP_EML_NEW if !$seen{$need}++;
+ $r[2] = [ map { [ $_, @{$partial{$_}} ] } sort keys %partial ];
+ }
+
+ $r[0] = $need;
+
+ # r[1] = [ $key1, $cb1, $key2, $cb2, ... ]
+ use sort 'stable'; # makes output more consistent
+ $r[1] = [ map { ($_->[2], $_->[1]) } sort { $a->[0] <=> $b->[0] } @op ];
+ @r;
+}
+
+sub cmd_uid_fetch ($$$;@) {
+ my ($self, $tag, $range_csv, @want) = @_;
+ my $ibx = $self->{ibx} or return "$tag BAD No mailbox selected\r\n";
+ my ($need, $ops, $partial) = fetch_compile(\@want);
+ return "$tag $need\r\n" unless $ops;
+
+ $range_csv = 'bad' if $range_csv !~ $valid_range;
+ my $range_info = range_step($self, \$range_csv);
+ return "$tag $range_info\r\n" if !ref($range_info);
+
+ long_response($self, \&uid_fetch_m,
+ $tag, [], $range_info, $ops, $partial);
+}
+
+sub parse_date ($) { # 02-Oct-1993
+ my ($date_text) = @_;
+ my ($dd, $mon, $yyyy) = split(/-/, $_[0], 3);
+ defined($yyyy) or return;
+ my $mm = $MoY{$mon} // return;
+ $dd =~ /\A[0123]?[0-9]\z/ or return;
+ $yyyy =~ /\A[0-9]{4,}\z/ or return; # Y10K-compatible!
+ timegm(0, 0, 0, $dd, $mm, $yyyy);
+}
+
sub uid_search_uid_range { # long_response
- my ($self, $tag, $ibx, $beg, $end) = @_;
- my $uids = $ibx->mm->msg_range($beg, $end, 'num');
+ my ($self, $tag, $beg, $end, $sql) = @_;
+ my $uids = $self->{ibx}->over->uid_range($$beg, $end, $sql);
if (@$uids) {
- $self->msg_more(join('', map { " $_->[0]" } @$uids));
+ $$beg = $uids->[-1] + 1;
+ $self->msg_more(join(' ', '', @$uids));
} else {
- $self->write(\"\r\n$tag OK\r\n");
+ $self->write(\"\r\n$tag OK Search done\r\n");
undef;
}
}
+sub date_search {
+ my ($q, $k, $d) = @_;
+ my $sql = $q->{sql};
+
+ # Date: header
+ if ($k eq 'SENTON') {
+ my $end = $d + 86399; # no leap day...
+ my $da = strftime('%Y%m%d%H%M%S', gmtime($d));
+ my $db = strftime('%Y%m%d%H%M%S', gmtime($end));
+ $q->{xap} .= " dt:$da..$db";
+ $$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);
+ my $sql = delete $q->{sql};
+
+ if (!scalar(keys %$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, $sql);
+ } 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, $sql);
+ } elsif ($uid =~ /\A[0-9]+\z/s) {
$uid = $ibx->over->get_art($uid) ? " $uid" : '';
- "* SEARCH$uid\r\n$tag OK\r\n";
+ "* SEARCH$uid\r\n$tag OK Search done\r\n";
} else {
- "$tag BAD\r\n";
+ "$tag BAD Error\r\n";
}
} else {
- "$tag BAD\r\n";
+ "$tag BAD Error\r\n";
}
}
cmd_done($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 $res;
} elsif ($more) { # $self->{wbuf}:
$self->update_idle_time;
- # COMPRESS users all share the same DEFLATE context.
- # Flush it here to ensure clients don't see
- # each other's data
- $self->zflush;
-
- # no recursion, schedule another call ASAP, but only after
- # all pending writes are done. autovivify wbuf:
- my $new_size = push(@{$self->{wbuf}}, \&long_step);
-
- # wbuf may be populated by $cb, no need to rearm if so:
- $self->requeue if $new_size == 1;
+ # control passed to $more may be a GitAsyncCat object
+ requeue_once($self) if !ref($more);
} else { # all done!
delete $self->{long_cb};
my $elapsed = now() - $t0;
sub event_step {
my ($self) = @_;
- return unless $self->flush_write && $self->{sock};
+ return unless $self->flush_write && $self->{sock} && !$self->{long_cb};
$self->update_idle_time;
# only read more requests if we've drained the write buffer,
# otherwise we can be buffering infinitely w/o backpressure
- my $rbuf = $self->{rbuf} // (\(my $x = ''));
- my $r = 1;
-
- if (index($$rbuf, "\n") < 0) {
- my $off = length($$rbuf);
- $r = $self->do_read($rbuf, LINE_MAX, $off) or return;
- }
- while ($r > 0 && $$rbuf =~ s/\A[ \t]*([^\n]*?)\r?\n//) {
- my $line = $1;
- return $self->close if $line =~ /[[:cntrl:]]/s;
- my $t0 = now();
- my $fd = fileno($self->{sock});
- $r = eval { process_line($self, $line) };
- my $pending = $self->{wbuf} ? ' pending' : '';
- out($self, "[$fd] %s - %0.6f$pending", $line, now() - $t0);
+ my $rbuf = $self->{rbuf} // \(my $x = '');
+ my $line = index($$rbuf, "\n");
+ while ($line < 0) {
+ return $self->close if length($$rbuf) >= LINE_MAX;
+ $self->do_read($rbuf, LINE_MAX, length($$rbuf)) or return;
+ $line = index($$rbuf, "\n");
}
+ $line = substr($$rbuf, 0, $line + 1, '');
+ $line =~ s/\r?\n\z//s;
+ return $self->close if $line =~ /[[:cntrl:]]/s;
+ my $t0 = now();
+ my $fd = fileno($self->{sock});
+ my $r = eval { process_line($self, $line) };
+ my $pending = $self->{wbuf} ? ' pending' : '';
+ out($self, "[$fd] %s - %0.6f$pending - $r", $line, now() - $t0);
return $self->close if $r < 0;
- my $len = length($$rbuf);
- return $self->close if ($len >= LINE_MAX);
$self->rbuf_idle($rbuf);
$self->update_idle_time;
# maybe there's more pipelined data, or we'll have
# to register it for socket-readiness notifications
- $self->requeue unless $self->{wbuf};
+ $self->requeue unless $pending;
}
sub compressed { undef }
# we're read-only, so SELECT and EXAMINE do the same thing
no warnings 'once';
*cmd_select = \&cmd_examine;
+*cmd_fetch = \&cmd_uid_fetch;
1;