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 # /^input_/ subs are used by (or override) PublicInbox::LeiInput superclass
12 sub input_eml_cb { # used by PublicInbox::LeiInput::input_fh
13 my ($self, $eml, $vmd) = @_;
14 my $xoids = $self->{lei}->{ale}->xoids_for($eml);
15 if (my $all_vmd = $self->{all_vmd}) {
16 @$vmd{keys %$all_vmd} = values %$all_vmd;
18 $self->{lei}->{sto}->ipc_do('set_eml', $eml, $vmd, $xoids);
21 sub input_mbox_cb { # MboxReader callback
22 my ($eml, $self) = @_;
24 if ($self->{-import_kw}) {
25 my $kw = PublicInbox::MboxReader::mbox_keywords($eml);
26 $vmd = { kw => $kw } if scalar(@$kw);
28 input_eml_cb($self, $eml, $vmd);
31 sub input_maildir_cb { # maildir_each_eml cb
32 my ($f, $kw, $eml, $self) = @_;
33 my $vmd = $self->{-import_kw} ? { kw => $kw } : undef;
34 if ($self->{-mail_sync}) {
35 if ($f =~ m!\A(.+?)/(?:new|cur)/([^/]+)\z!) { # ugh...
36 $vmd->{sync_info} = [ "maildir:$1", \(my $n = $2) ];
38 warn "E: $f was not from a Maildir?\n";
41 input_eml_cb($self, $eml, $vmd);
44 sub input_imap_cb { # imap_each
45 my ($url, $uid, $kw, $eml, $self) = @_;
46 my $vmd = $self->{-import_kw} ? { kw => $kw } : undef;
47 $vmd->{sync_info} = [ $url, $uid ] if $self->{-mail_sync};
48 input_eml_cb($self, $eml, $vmd);
51 sub input_nntp_cb { # nntp_each
52 my ($url, $num, $kw, $eml, $self) = @_;
53 input_eml_cb($self, $eml, $self->{-import_kw} ? { kw => $kw } : undef);
56 sub net_merge_complete { # callback used by LeiAuth
58 $self->wq_io_do('process_inputs');
62 sub lei_import { # the main "lei import" method
63 my ($lei, @inputs) = @_;
64 my $sto = $lei->_lei_store(1);
65 $sto->write_prepare($lei);
66 my $self = bless {}, __PACKAGE__;
67 $self->{-import_kw} = $lei->{opt}->{kw} // 1;
68 my $vmd_mod = $self->vmd_mod_extract(\@inputs);
69 return $lei->fail(join("\n", @{$vmd_mod->{err}})) if $vmd_mod->{err};
70 $self->{all_vmd} = $vmd_mod if scalar keys %$vmd_mod;
71 $self->prepare_inputs($lei, \@inputs) or return;
72 $self->{-mail_sync} = $lei->{opt}->{sync} // 1;
74 $lei->ale; # initialize for workers to read
75 my $j = $lei->{opt}->{jobs} // scalar(@{$self->{inputs}}) || 1;
76 if (my $net = $lei->{net}) {
77 # $j = $net->net_concurrency($j); TODO
78 if ($lei->{opt}->{incremental} // 1) {
79 $net->{incremental} = 1;
80 $net->{itrk_fn} = $lei->store_path .
84 my $nproc = $self->detect_nproc;
85 $j = $nproc if $j > $nproc;
88 $lei->{auth}->op_merge($ops, $self) if $lei->{auth};
89 $self->{-wq_nr_workers} = $j // 1; # locked
90 $lei->{-eml_noisy} = 1;
91 (my $op_c, $ops) = $lei->workers_start($self, 'lei-import', $j, $ops);
93 $lei->{-err_type} = 'non-fatal';
94 net_merge_complete($self) unless $lei->{auth};
95 $op_c->op_wait_event($ops);
99 *ipc_atfork_child = \&PublicInbox::LeiInput::input_only_atfork_child;
101 # the following works even when LeiAuth is lazy-loaded
102 *net_merge_all = \&PublicInbox::LeiAuth::net_merge_all;