1 # Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Writes PublicInbox::Eml objects atomically to a mbox variant or Maildir
5 package PublicInbox::LeiToMail;
8 use parent qw(PublicInbox::IPC);
10 use PublicInbox::Lock;
11 use PublicInbox::ProcessPipe;
12 use PublicInbox::Spawn qw(which spawn popen_rd);
13 use PublicInbox::LeiDedupe;
14 use PublicInbox::OnDestroy;
16 use PublicInbox::GitAsyncCat;
17 use Symbol qw(gensym);
18 use IO::Handle; # ->autoflush
19 use Fcntl qw(SEEK_SET SEEK_END O_CREAT O_EXCL O_WRONLY);
20 use Errno qw(EEXIST ESPIPE ENOENT EPIPE);
22 # struggles with short-lived repos, Gcf2Client makes little sense with lei;
23 # but we may use in-process libgit2 in the future.
24 $PublicInbox::GitAsyncCat::GCF2C = 0;
26 my %kw2char = ( # Maildir characters
34 flagged => [ 'X-Status' => 'F' ],
35 answered => [ 'X-Status' => 'A' ],
36 seen => [ 'Status' => 'R' ],
37 draft => [ 'X-Status' => 'T' ],
40 sub _mbox_hdr_buf ($$$) {
41 my ($eml, $type, $smsg) = @_;
42 $eml->header_set($_) for (qw(Lines Bytes Content-Length));
44 # Messages are always 'O' (non-\Recent in IMAP), it saves
45 # MUAs the trouble of rewriting the mbox if no other
47 my %hdr = (Status => [ 'O' ]); # set Status, X-Status
48 for my $k (@{$smsg->{kw} // []}) {
49 if (my $ent = $kw2status{$k}) {
50 push @{$hdr{$ent->[0]}}, $ent->[1];
52 warn "TODO: keyword `$k' not supported for mbox\n";
55 while (my ($name, $chars) = each %hdr) {
56 $eml->header_set($name, join('', sort @$chars));
58 my $buf = delete $eml->{hdr};
60 # fixup old bug from import (pre-a0c07cba0e5d8b6a)
61 $$buf =~ s/\A[\r\n]*From [^\r\n]*\r?\n//s;
62 my $ident = $smsg->{blob} // 'lei';
63 if (defined(my $pct = $smsg->{pct})) { $ident .= "=$pct" }
65 substr($$buf, 0, 0, # prepend From line
66 "From $ident\@$type Thu Jan 1 00:00:00 1970$eml->{crlf}");
70 sub atomic_append { # for on-disk destinations (O_APPEND, or O_EXCL)
72 if (defined(my $w = syswrite($lei->{1} // return, $$buf))) {
73 return if $w == length($$buf);
74 $buf = "short atomic write: $w != ".length($$buf);
75 } elsif ($! == EPIPE) {
76 return $lei->note_sigpipe(1);
78 $buf = "atomic write: $!";
83 sub eml2mboxrd ($;$) {
84 my ($eml, $smsg) = @_;
85 my $buf = _mbox_hdr_buf($eml, 'mboxrd', $smsg);
86 if (my $bdy = delete $eml->{bdy}) {
87 $$bdy =~ s/^(>*From )/>$1/gm;
88 $$buf .= $eml->{crlf};
89 substr($$bdy, 0, 0, $$buf); # prepend header
92 $$buf .= $eml->{crlf};
97 my ($eml, $smsg) = @_;
98 my $buf = _mbox_hdr_buf($eml, 'mboxo', $smsg);
99 if (my $bdy = delete $eml->{bdy}) {
100 $$bdy =~ s/^From />From /gm;
101 $$buf .= $eml->{crlf};
102 substr($$bdy, 0, 0, $$buf); # prepend header
105 $$buf .= $eml->{crlf};
109 sub _mboxcl_common ($$$) {
110 my ($buf, $bdy, $crlf) = @_;
111 # add Lines: so mutt won't have to add it on MUA close
112 my $lines = $$bdy =~ tr!\n!\n!;
113 $$buf .= 'Content-Length: '.length($$bdy).$crlf.
114 'Lines: '.$lines.$crlf.$crlf;
115 substr($$bdy, 0, 0, $$buf); # prepend header
119 # mboxcl still escapes "From " lines
121 my ($eml, $smsg) = @_;
122 my $buf = _mbox_hdr_buf($eml, 'mboxcl', $smsg);
123 my $crlf = $eml->{crlf};
124 if (my $bdy = delete $eml->{bdy}) {
125 $$bdy =~ s/^From />From /gm;
126 _mboxcl_common($buf, $bdy, $crlf);
132 # mboxcl2 has no "From " escaping
134 my ($eml, $smsg) = @_;
135 my $buf = _mbox_hdr_buf($eml, 'mboxcl2', $smsg);
136 my $crlf = $eml->{crlf};
137 if (my $bdy = delete $eml->{bdy}) {
138 _mboxcl_common($buf, $bdy, $crlf);
144 sub git_to_mail { # git->cat_async callback
145 my ($bref, $oid, $type, $size, $arg) = @_;
146 if ($type ne 'blob') {
147 if ($type eq 'missing') {
148 warn "missing $oid\n";
150 warn "unexpected type=$type for $oid\n";
153 my ($write_cb, $smsg) = @$arg;
154 if ($smsg->{blob} ne $oid) {
155 die "BUG: expected=$smsg->{blob} got=$oid";
157 $write_cb->($bref, $smsg) if $size > 0;
160 sub reap_compress { # dwaitpid callback
161 my ($lei, $pid) = @_;
162 my $cmd = delete $lei->{"pid.$pid"};
164 $lei->fail("@$cmd failed", $? >> 8);
167 # all of these support -c for stdout and -d for decompression,
168 # mutt is commonly distributed with hooks for gz, bz2 and xz, at least
169 # { foo => '' } means "--foo" is passed to the command-line,
170 # otherwise { foo => '--bar' } passes "--bar"
172 gz => [ qw(GZIP pigz gzip), { rsyncable => '', threads => '-p' } ],
173 bz2 => [ 'bzip2', {} ],
174 xz => [ 'xz', { threads => '-T' } ],
175 # XXX does anybody care for these? I prefer zstd on entire FSes,
176 # so it's probably not necessary on a per-file basis
177 # zst => [ 'zstd', { -default => [ qw(-q) ], # it's noisy by default
178 # rsyncable => '', threads => '-T' } ],
179 # zz => [ 'pigz', { -default => [ '--zlib' ],
180 # rsyncable => '', threads => '-p' }],
181 # lzo => [ 'lzop', {} ],
182 # lzma => [ 'lzma', {} ],
186 my ($zsfx, $decompress, $lei) = @_;
187 my $x = $zsfx2cmd{$zsfx} // die "no support for suffix=.$zsfx";
189 my $cmd_opt = pop @info;
190 my @cmd = (undef, $decompress ? qw(-dc) : qw(-c));
191 for my $exe (@info) {
192 # I think respecting client's ENV{GZIP} is OK, not sure
193 # about ENV overrides for other, less-common compressors
194 if ($exe eq uc($exe)) {
195 $exe = $lei->{env}->{$exe} or next;
197 $cmd[0] = which($exe) and last;
199 $cmd[0] // die join(' or ', @info)." missing for .$zsfx";
200 # push @cmd, @{$cmd_opt->{-default}} if $cmd_opt->{-default};
201 for my $bool (qw(rsyncable)) {
202 my $switch = $cmd_opt->{rsyncable} // next;
203 push @cmd, '--'.($switch || $bool);
205 for my $key (qw(threads)) { # support compression level?
206 my $switch = $cmd_opt->{$key} // next;
207 my $val = $lei->{opt}->{$key} // next;
208 push @cmd, $switch, $val;
213 sub _post_augment_mbox { # open a compressor process
214 my ($self, $lei) = @_;
215 my $zsfx = $self->{zsfx} or return;
216 my $cmd = zsfx2cmd($zsfx, undef, $lei);
217 my ($r, $w) = @{delete $lei->{zpipe}};
218 my $rdr = { 0 => $r, 1 => $lei->{1}, 2 => $lei->{2} };
219 my $pid = spawn($cmd, $lei->{env}, $rdr);
221 my $dup = bless { "pid.$pid" => $cmd }, ref($lei);
222 $dup->{$_} = $lei->{$_} for qw(2 sock);
223 tie *$pp, 'PublicInbox::ProcessPipe', $pid, $w, \&reap_compress, $dup;
225 die 'BUG: unexpected {ovv}->{lock_path}' if $lei->{ovv}->{lock_path};
226 $lei->{ovv}->ovv_out_lk_init;
229 sub decompress_src ($$$) {
230 my ($in, $zsfx, $lei) = @_;
231 my $cmd = zsfx2cmd($zsfx, 1, $lei);
232 popen_rd($cmd, $lei->{env}, { 0 => $in, 2 => $lei->{2} });
237 open my $dup, '+>>&', $in or die "dup: $!";
241 # --augment existing output destination, with deduplication
242 sub _augment { # MboxReader eml_cb
243 my ($eml, $lei) = @_;
244 # ignore return value, just populate the skv
245 $lei->{dedupe}->is_dup($eml);
248 sub _mbox_write_cb ($$) {
249 my ($self, $lei) = @_;
250 my $ovv = $lei->{ovv};
251 my $m = 'eml2'.$ovv->{fmt};
252 my $eml2mbox = $self->can($m) or die "$self->$m missing";
253 $lei->{1} // die "no stdout ($m, $ovv->{dst})"; # redirected earlier
254 $lei->{1}->autoflush(1);
255 my $atomic_append = !defined($ovv->{lock_path});
256 my $dedupe = $lei->{dedupe};
257 $dedupe->prepare_dedupe;
258 sub { # for git_to_mail
259 my ($buf, $smsg, $eml) = @_;
260 $eml //= PublicInbox::Eml->new($buf);
261 return if $dedupe->is_dup($eml, $smsg->{blob});
262 $buf = $eml2mbox->($eml, $smsg);
263 return atomic_append($lei, $buf) if $atomic_append;
264 my $lk = $ovv->lock_for_scope;
269 sub _maildir_each_file ($$;@) {
270 my ($dir, $cb, @arg) = @_;
271 for my $d (qw(new/ cur/)) {
273 opendir my $dh, $pfx or next;
274 while (defined(my $fn = readdir($dh))) {
275 $cb->($pfx.$fn, @arg) if $fn =~ /:2,[A-Za-z]*\z/;
280 sub _augment_file { # _maildir_each_file cb
282 my $eml = PublicInbox::InboxWritable::eml_from_path($f) or return;
283 _augment($eml, $lei);
286 # _maildir_each_file callback, \&CORE::unlink doesn't work with it
287 sub _unlink { unlink($_[0]) }
291 sprintf('%x,%x,%x,%x', rand(0xffffffff), time, $$, ++$seq);
295 my ($dst, $buf, $smsg) = @_;
296 my $kw = $smsg->{kw} // [];
297 my $sfx = join('', sort(map { $kw2char{$_} // () } @$kw));
298 my $rand = ''; # chosen by die roll :P
299 my ($tmp, $fh, $final, $ok);
300 my $common = $smsg->{blob} // _rand;
301 if (defined(my $pct = $smsg->{pct})) { $common .= "=$pct" }
303 $tmp = $dst.'tmp/'.$rand.$common;
304 } while (!($ok = sysopen($fh, $tmp, O_CREAT|O_EXCL|O_WRONLY)) &&
305 $! == EEXIST && ($rand = _rand.','));
306 if ($ok && print $fh $$buf and close($fh)) {
307 # ignore new/ and write only to cur/, otherwise MUAs
308 # with R/W access to the Maildir will end up doing
309 # a mass rename which can take a while with thousands
314 $final = $dst.$rand.$common.':2,'.$sfx;
315 } while (!link($tmp, $final) && $! == EEXIST &&
316 ($rand = _rand.','));
317 unlink($tmp) or warn "W: failed to unlink $tmp: $!\n";
319 my $err = "Error writing $smsg->{blob} to $dst: $!\n";
320 $_[0] = undef; # clobber dst
326 sub _maildir_write_cb ($$) {
327 my ($self, $lei) = @_;
328 my $dedupe = $lei->{dedupe};
329 $dedupe->prepare_dedupe if $dedupe;
330 my $dst = $lei->{ovv}->{dst};
331 sub { # for git_to_mail
332 my ($buf, $smsg, $eml) = @_;
333 $dst // return $lei->fail; # dst may be undef-ed in last run
334 $buf //= \($eml->as_string);
335 return _buf2maildir($dst, $buf, $smsg) if !$dedupe;
336 $eml //= PublicInbox::Eml->new($$buf); # copy buf
337 return if $dedupe->is_dup($eml, $smsg->{blob});
339 _buf2maildir($dst, $buf, $smsg);
343 sub write_cb { # returns a callback for git_to_mail
344 my ($self, $lei) = @_;
345 # _mbox_write_cb or _maildir_write_cb
346 my $m = "_$self->{base_type}_write_cb";
351 my ($cls, $lei) = @_;
352 my $fmt = $lei->{ovv}->{fmt};
353 my $dst = $lei->{ovv}->{dst};
354 my $self = bless {}, $cls;
355 if ($fmt eq 'maildir') {
356 $self->{base_type} = 'maildir';
357 -e $dst && !-d _ and die
358 "$dst exists and is not a directory\n";
359 $lei->{ovv}->{dst} = $dst .= '/' if substr($dst, -1) ne '/';
360 } elsif (substr($fmt, 0, 4) eq 'mbox') {
361 (-d $dst || (-e _ && !-w _)) and die
362 "$dst exists and is not a writable file\n";
363 $self->can("eml2$fmt") or die "bad mbox --format=$fmt\n";
364 $self->{base_type} = 'mbox';
366 die "bad mail --format=$fmt\n";
369 $lei->{dedupe} = PublicInbox::LeiDedupe->new($lei);
373 sub _pre_augment_maildir {} # noop
375 sub _do_augment_maildir {
376 my ($self, $lei) = @_;
377 my $dst = $lei->{ovv}->{dst};
378 if ($lei->{opt}->{augment}) {
379 my $dedupe = $lei->{dedupe};
380 if ($dedupe && $dedupe->prepare_dedupe) {
381 require PublicInbox::InboxWritable; # eml_from_path
382 _maildir_each_file($dst, \&_augment_file, $lei);
383 $dedupe->pause_dedupe;
385 } else { # clobber existing Maildir
386 _maildir_each_file($dst, \&_unlink);
390 sub _post_augment_maildir {
391 my ($self, $lei) = @_;
392 my $dst = $lei->{ovv}->{dst};
393 for my $x (qw(tmp new cur)) {
397 File::Path::mkpath($d);
398 -d $d or die "$d is not a directory";
402 sub _pre_augment_mbox {
403 my ($self, $lei) = @_;
404 my $dst = $lei->{ovv}->{dst};
405 if ($dst ne '/dev/stdout') {
406 my $mode = -p $dst ? '>' : '+>>';
407 if (-f _ && !$lei->{opt}->{augment} and !unlink($dst)) {
408 $! == ENOENT or die "unlink($dst): $!";
410 open my $out, $mode, $dst or die "open($dst): $!";
411 $lei->{old_1} = $lei->{1}; # keep for spawning MUA
414 # Perl does SEEK_END even with O_APPEND :<
415 $self->{seekable} = seek($lei->{1}, 0, SEEK_SET);
416 if (!$self->{seekable} && $! != ESPIPE && $dst ne '/dev/stdout') {
417 die "seek($dst): $!\n";
419 state $zsfx_allow = join('|', keys %zsfx2cmd);
420 ($self->{zsfx}) = ($dst =~ /\.($zsfx_allow)\z/) or return;
421 pipe(my ($r, $w)) or die "pipe: $!";
422 $lei->{zpipe} = [ $r, $w ];
425 sub _do_augment_mbox {
426 my ($self, $lei) = @_;
427 return if !$lei->{opt}->{augment};
428 my $dedupe = $lei->{dedupe};
429 my $dst = $lei->{ovv}->{dst};
430 die "cannot augment $dst, not seekable\n" if !$self->{seekable};
432 if (-s $out && $dedupe && $dedupe->prepare_dedupe) {
433 my $zsfx = $self->{zsfx};
434 my $rd = $zsfx ? decompress_src($out, $zsfx, $lei) :
436 my $fmt = $lei->{ovv}->{fmt};
437 require PublicInbox::MboxReader;
438 PublicInbox::MboxReader->$fmt($rd, \&_augment, $lei);
440 # maybe some systems don't honor O_APPEND, Perl does this:
441 seek($out, 0, SEEK_END) or die "seek $dst: $!";
442 $dedupe->pause_dedupe if $dedupe;
445 sub pre_augment { # fast (1 disk seek), runs in main daemon
446 my ($self, $lei) = @_;
447 # _pre_augment_maildir, _pre_augment_mbox
448 my $m = "_pre_augment_$self->{base_type}";
452 sub do_augment { # slow, runs in wq worker
453 my ($self, $lei) = @_;
454 # _do_augment_maildir, _do_augment_mbox
455 my $m = "_do_augment_$self->{base_type}";
459 sub post_augment { # fast (spawn compressor or mkdir), runs in main daemon
460 my ($self, $lei, @args) = @_;
461 # _post_augment_maildir, _post_augment_mbox
462 my $m = "_post_augment_$self->{base_type}";
463 $self->$m($lei, @args);
466 sub ipc_atfork_child {
468 my $lei = delete $self->{lei};
469 $lei->lei_atfork_child;
470 if (my $zpipe = delete $lei->{zpipe}) {
471 $lei->{1} = $zpipe->[1];
474 $self->{wcb} = $self->write_cb($lei);
475 $SIG{__WARN__} = PublicInbox::Eml::warn_ignore_cb();
476 $self->SUPER::ipc_atfork_child;
480 $_[0]->{base_type} =~ /\A(?:maildir|mh|imap|jmap)\z/ ? 1 : 0;
485 if ($self->{base_type} eq 'maildir') {
487 utime($t, $t, "$self->{dst}/cur");
491 sub write_mail { # via ->wq_io_do
492 my ($self, $git_dir, $smsg) = @_;
493 my $git = $self->{"$$\0$git_dir"} //= PublicInbox::Git->new($git_dir);
494 git_async_cat($git, $smsg->{blob}, \&git_to_mail,
495 [$self->{wcb}, $smsg]);
498 sub wq_atexit_child {
501 for my $git (delete @$self{grep(/\A$$\0/, keys %$self)}) {
502 $git->async_wait_all;
504 $SIG{__WARN__} = 'DEFAULT';