2 # Copyright (C) 2015-2020 all contributors <meta@public-inbox.org>
3 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Basic tool to create a Xapian search index for a public-inbox.
5 # Usage with libeatmydata <https://www.flamingspork.com/projects/libeatmydata/>
6 # highly recommended: eatmydata public-inbox-index INBOX_DIR
10 use Getopt::Long qw(:config gnu_getopt no_ignore_case auto_abbrev);
11 my $usage = "public-inbox-index INBOX_DIR";
12 use PublicInbox::Admin;
13 PublicInbox::Admin::require_or_die('-index');
14 use PublicInbox::Xapcmd;
17 my $opt = { quiet => -1, compact => 0, maxsize => undef, fsync => 1 };
18 GetOptions($opt, qw(verbose|v+ reindex rethread compact|c+ jobs|j=i prune
19 fsync|sync! xapianonly|xapian-only
20 indexlevel|L=s maxsize|max-size=s batchsize|batch-size=s
21 sequentialshard|seq-shard|sequential-shard))
22 or die "bad command-line args\n$usage";
23 die "--jobs must be >= 0\n" if defined $opt->{jobs} && $opt->{jobs} < 0;
25 if ($opt->{compact}) {
26 require PublicInbox::Xapcmd;
27 PublicInbox::Xapcmd::check_compact();
28 $compact_opt = { -coarse_lock => 1, compact => 1 };
29 if (defined(my $jobs = $opt->{jobs})) {
30 $compact_opt->{jobs} = $jobs;
34 my $cfg = PublicInbox::Config->new;
35 my @ibxs = PublicInbox::Admin::resolve_inboxes(\@ARGV, undef, $cfg);
36 PublicInbox::Admin::require_or_die('-index');
37 unless (@ibxs) { print STDERR "Usage: $usage\n"; exit 1 }
39 my $max_size = $opt->{maxsize} // $cfg->{lc('publicInbox.indexMaxSize')};
40 if (defined $max_size) {
41 PublicInbox::Admin::parse_unsigned(\$max_size) or
42 die "`publicInbox.indexMaxSize=$max_size' not parsed\n";
45 if (my $bs = $opt->{batchsize} // $cfg->{lc('publicInbox.indexBatchSize')}) {
46 PublicInbox::Admin::parse_unsigned(\$bs) or
47 die "`publicInbox.indexBatchSize=$bs' not parsed\n";
48 $PublicInbox::SearchIdx::BATCH_BYTES = $bs;
51 my $s = $opt->{sequentialshard} //
52 $cfg->{lc('publicInbox.indexSequentialShard')};
54 my $v = $cfg->git_bool($s);
56 die "`publicInbox.indexSequentialShard=$s' not boolean\n";
57 $opt->{sequentialshard} = $v;
61 foreach my $ibx (@ibxs) {
62 # XXX: users can shoot themselves in the foot, with opt->{indexlevel}
63 $ibx->{indexlevel} //= $opt->{indexlevel} // ($opt->{xapianonly} ?
64 'full' : PublicInbox::Admin::detect_indexlevel($ibx));
65 $ibx->{index_max_size} = $max_size;
66 PublicInbox::Admin::scan_ibx_modules($mods, $ibx);
69 PublicInbox::Admin::require_or_die(keys %$mods);
70 PublicInbox::Admin::progress_prepare($opt);
72 $ibx = PublicInbox::InboxWritable->new($ibx);
73 if ($opt->{compact} >= 2) {
74 PublicInbox::Xapcmd::run($ibx, 'compact', $compact_opt);
76 $ibx->{-no_fsync} = 1 if !$opt->{fsync};
79 if (defined(my $s = $ibx->{indexsequentialshard})) {
80 defined(my $v = $cfg->git_bool($s)) or die <<EOL;
81 publicInbox.$ibx->{name}.indexSequentialShard not boolean
83 $ibx_opt = { %$opt, sequentialshard => $v };
85 PublicInbox::Admin::index_inbox($ibx, undef, $ibx_opt);
86 PublicInbox::Xapcmd::run($ibx, 'compact', $compact_opt) if $compact_opt;