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 # front-end for the "lei import" sub-command
5 package PublicInbox::LeiImport;
8 use parent qw(PublicInbox::IPC PublicInbox::LeiInput);
10 use PublicInbox::PktOp qw(pkt_do);
12 sub _import_eml { # MboxReader callback
13 my ($eml, $lei, $mbox_keywords) = @_;
16 my $kw = $mbox_keywords->($eml);
17 $vmd = { kw => $kw } if scalar(@$kw);
19 my $xoids = $lei->{ale}->xoids_for($eml);
20 $lei->{sto}->ipc_do('set_eml', $eml, $vmd, $xoids);
23 sub import_done_wait { # dwaitpid callback
25 my ($imp, $lei) = @$arg;
26 $lei->child_error($?, 'non-fatal errors during import') if $?;
27 my $sto = delete $lei->{sto};
28 my $wait = $sto->ipc_do('done') if $sto; # PublicInbox::LeiStore::done
32 sub import_done { # EOF callback for main daemon
34 my $imp = delete $lei->{imp} or return;
35 $imp->wq_wait_old(\&import_done_wait, $lei);
38 sub net_merge_complete { # callback used by LeiAuth
40 for my $input (@{$self->{inputs}}) {
41 $self->wq_io_do('import_path_url', [], $input);
48 my $self = $lei->{imp};
50 my $j = $lei->{opt}->{jobs} // scalar(@{$self->{inputs}}) || 1;
51 if (my $net = $lei->{net}) {
52 # $j = $net->net_concurrency($j); TODO
54 my $nproc = $self->detect_nproc;
55 $j = $nproc if $j > $nproc;
57 my $ops = { '' => [ \&import_done, $lei ] };
58 $lei->{auth}->op_merge($ops, $self) if $lei->{auth};
59 $self->{-wq_nr_workers} = $j // 1; # locked
60 my $op = $lei->workers_start($self, 'lei_import', undef, $ops);
61 $self->wq_io_do('import_stdin', []) if $self->{0};
62 net_merge_complete($self) unless $lei->{auth};
63 while ($op && $op->{sock}) { $op->event_step }
66 sub lei_import { # the main "lei import" method
67 my ($lei, @inputs) = @_;
68 my $sto = $lei->_lei_store(1);
69 $sto->write_prepare($lei);
70 $lei->{opt}->{kw} //= 1;
71 my $self = $lei->{imp} = bless {}, __PACKAGE__;
72 $self->prepare_inputs($lei, \@inputs) or return;
76 sub ipc_atfork_child {
78 my $lei = $self->{lei};
79 delete $lei->{imp}; # drop circular ref
80 $lei->lei_atfork_child;
81 $self->SUPER::ipc_atfork_child;
82 $lei->{auth}->do_auth_atfork($self) if $lei->{auth};
87 my ($lei, $fh, $input, $ifmt) = @_;
88 my $kw = $lei->{opt}->{kw} ?
89 PublicInbox::MboxReader->can('mbox_keywords') : undef;
92 my $buf = do { local $/; <$fh> } //
93 return $lei->child_error(1 << 8, <<"");
94 error reading $input: $!
96 my $eml = PublicInbox::Eml->new(\$buf);
97 _import_eml($eml, $lei, $kw);
98 } else { # some mbox (->can already checked in call);
99 my $cb = PublicInbox::MboxReader->reads($ifmt) //
100 die "BUG: bad fmt=$ifmt";
101 $cb->(undef, $fh, \&_import_eml, $lei, $kw);
104 $lei->child_error(1 << 8, "$input: $@") if $@;
107 sub _import_maildir { # maildir_each_eml cb
108 my ($f, $kw, $eml, $sto, $set_kw) = @_;
109 $sto->ipc_do('set_eml', $eml, $set_kw ? { kw => $kw }: ());
112 sub _import_net { # imap_each, nntp_each cb
113 my ($url, $uid, $kw, $eml, $sto, $set_kw) = @_;
114 $sto->ipc_do('set_eml', $eml, $set_kw ? { kw => $kw } : ());
117 sub import_path_url {
118 my ($self, $input) = @_;
119 my $lei = $self->{lei};
120 my $ifmt = lc($lei->{opt}->{'in-format'} // '');
122 if ($input =~ m!\Aimaps?://!i) {
123 $lei->{net}->imap_each($input, \&_import_net, $lei->{sto},
126 } elsif ($input =~ m!\A(?:nntps?|s?news)://!i) {
127 $lei->{net}->nntp_each($input, \&_import_net, $lei->{sto}, 0);
129 } elsif ($input =~ s!\A([a-z0-9]+):!!i) {
133 my $m = $lei->{opt}->{'lock'} // ($ifmt eq 'eml' ? ['none'] :
134 PublicInbox::MboxLock->defaults);
135 my $mbl = PublicInbox::MboxLock->acq($input, 0, $m);
136 _import_fh($lei, $mbl->{fh}, $input, $ifmt);
137 } elsif (-d _ && (-d "$input/cur" || -d "$input/new")) {
138 return $lei->fail(<<EOM) if $ifmt && $ifmt ne 'maildir';
139 $input appears to a be a maildir, not $ifmt
141 PublicInbox::MdirReader::maildir_each_eml($input,
143 $lei->{sto}, $lei->{opt}->{kw});
145 $lei->fail("$input unsupported (TODO)");
151 my $lei = $self->{lei};
152 my $in = delete $self->{0};
153 _import_fh($lei, $in, '<stdin>', $lei->{opt}->{'in-format'});
156 no warnings 'once'; # the following works even when LeiAuth is lazy-loaded
157 *net_merge_all = \&PublicInbox::LeiAuth::net_merge_all;