]> Sergey Matveev's repositories - public-inbox.git/blob - lib/PublicInbox/LeiQuery.pm
lei q: do not leave temporary files after oneshot exit
[public-inbox.git] / lib / PublicInbox / LeiQuery.pm
1 # Copyright (C) 2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
3
4 # handles lei <q|ls-query|rm-query|mv-query> commands
5 package PublicInbox::LeiQuery;
6 use strict;
7 use v5.10.1;
8 use PublicInbox::DS qw(dwaitpid);
9
10 sub prep_ext { # externals_each callback
11         my ($lxs, $exclude, $loc) = @_;
12         $lxs->prepare_external($loc) unless $exclude->{$loc};
13 }
14
15 # the main "lei q SEARCH_TERMS" method
16 sub lei_q {
17         my ($self, @argv) = @_;
18         require PublicInbox::LeiXSearch;
19         require PublicInbox::LeiOverview;
20         PublicInbox::Config->json; # preload before forking
21         my $opt = $self->{opt};
22         # prepare any number of LeiXSearch || LeiSearch || Inbox || URL
23         my $lxs = $self->{lxs} = PublicInbox::LeiXSearch->new;
24         my @only = @{$opt->{only} // []};
25         # --local is enabled by default unless --only is used
26         # we'll allow "--only $LOCATION --local"
27         if ($opt->{'local'} //= scalar(@only) ? 0 : 1) {
28                 my $sto = $self->_lei_store(1);
29                 $lxs->prepare_external($sto->search);
30         }
31         if (@only) {
32                 for my $loc (@only) {
33                         $lxs->prepare_external($self->ext_canonicalize($loc));
34                 }
35         } else {
36                 for my $loc (@{$opt->{include} // []}) {
37                         $lxs->prepare_external($self->ext_canonicalize($loc));
38                 }
39                 # --external is enabled by default, but allow --no-external
40                 if ($opt->{external} //= 1) {
41                         my %x = map {;
42                                 ($self->ext_canonicalize($_), 1)
43                         } @{$self->{exclude} // []};
44                         my $ne = $self->externals_each(\&prep_ext, $lxs, \%x);
45                         $opt->{remote} //= !($lxs->locals - $opt->{'local'});
46                         if ($opt->{'local'}) {
47                                 delete($lxs->{remotes}) if !$opt->{remote};
48                         } else {
49                                 delete($lxs->{locals});
50                         }
51                 }
52         }
53         unless ($lxs->locals || $lxs->remotes) {
54                 return $self->fail('no local or remote inboxes to search');
55         }
56         my $xj = $lxs->concurrency($opt);
57         PublicInbox::LeiOverview->new($self) or return;
58         $self->atfork_prepare_wq($lxs);
59         $lxs->wq_workers_start('lei_xsearch', $xj, $self->oldset);
60         delete $lxs->{-ipc_atfork_child_close};
61         if (my $l2m = $self->{l2m}) {
62                 my $mj = 4; # TODO: configurable
63                 $self->atfork_prepare_wq($l2m);
64                 $l2m->wq_workers_start('lei2mail', $mj, $self->oldset);
65                 delete $l2m->{-ipc_atfork_child_close};
66         }
67
68         # no forking workers after this
69
70         my %mset_opt = map { $_ => $opt->{$_} } qw(thread limit offset);
71         $mset_opt{asc} = $opt->{'reverse'} ? 1 : 0;
72         $mset_opt{limit} //= 10000;
73         $mset_opt{qstr} = join(' ', map {;
74                 # Consider spaces in argv to be for phrase search in Xapian.
75                 # In other words, the users should need only care about
76                 # normal shell quotes and not have to learn Xapian quoting.
77                 /\s/ ? (s/\A(\w+:)// ? qq{$1"$_"} : qq{"$_"}) : $_
78         } @argv);
79         if (defined(my $sort = $opt->{'sort'})) {
80                 if ($sort eq 'relevance') {
81                         $mset_opt{relevance} = 1;
82                 } elsif ($sort eq 'docid') {
83                         $mset_opt{relevance} = $mset_opt{asc} ? -1 : -2;
84                 } elsif ($sort =~ /\Areceived(?:-?[aA]t)?\z/) {
85                         # the default
86                 } else {
87                         die "unrecognized --sort=$sort\n";
88                 }
89         }
90         # descending docid order
91         $mset_opt{relevance} //= -2 if $opt->{thread};
92         $self->{mset_opt} = \%mset_opt;
93         $self->{ovv}->ovv_begin($self);
94         $lxs->do_query($self);
95 }
96
97 # shell completion helper called by lei__complete
98 sub _complete_q {
99         my ($self, @argv) = @_;
100         my $ext = qr/\A(?:-I|(?:--(?:include|exclude|only)))\z/;
101         # $argv[-1] =~ $ext and return $self->_complete_forget_external;
102         my @cur;
103         while (@argv) {
104                 if ($argv[-1] =~ $ext) {
105                         my @c = $self->_complete_forget_external(@cur);
106                         return @c if @c;
107                 }
108                 unshift(@cur, pop @argv);
109         }
110         ();
111 }
112
113 # Stuff we may pass through to curl (as of 7.64.0), see curl manpage for
114 # details, so most options which make sense for HTTP/HTTPS (including proxy
115 # support for Tor and other methods of getting past weird networks).
116 # Most of these are untested by us, some may not make sense for our use case
117 # and typos below are likely.
118 # n.b. some short options (-$NUMBER) are not supported since they conflict
119 # with other "lei q" switches.
120 # FIXME: Getopt::Long doesn't easily let us support support options with
121 # '.' in them (e.g. --http1.1)
122 sub curl_opt { qw(
123         abstract-unix-socket=s anyauth basic cacert=s capath=s
124         cert-status cert-type cert|E=s ciphers=s config|K=s@
125         connect-timeout=s connect-to=s cookie-jar|c=s cookie|b=s crlfile=s
126         digest disable dns-interface=s dns-ipv4-addr=s dns-ipv6-addr=s
127         dns-servers=s doh-url=s egd-file=s engine=s false-start
128         happy-eyeballs-timeout-ms=s haproxy-protocol header|H=s@
129         http2-prior-knowledge http2 insecure|k
130         interface=s ipv4 ipv6 junk-session-cookies
131         key-type=s key=s limit-rate=s local-port=s location-trusted location|L
132         max-redirs=i max-time=s negotiate netrc-file=s netrc-optional netrc
133         no-alpn no-buffer|N no-npn no-sessionid noproxy=s ntlm-wb ntlm
134         pass=s pinnedpubkey=s post301 post302 post303 preproxy=s
135         proxy-anyauth proxy-basic proxy-cacert=s proxy-capath=s
136         proxy-cert-type=s proxy-cert=s proxy-ciphers=s proxy-crlfile=s
137         proxy-digest proxy-header=s@ proxy-insecure
138         proxy-key-type=s proxy-key proxy-negotiate proxy-ntlm proxy-pass=s
139         proxy-pinnedpubkey=s proxy-service-name=s proxy-ssl-allow-beast
140         proxy-tls13-ciphers=s proxy-tlsauthtype=s proxy-tlspassword=s
141         proxy-tlsuser=s proxy-tlsv1 proxy-user|U=s proxy=s
142         proxytunnel=s pubkey=s random-file=s referer=s resolve=s
143         retry-connrefused retry-delay=s retry-max-time=s retry=i
144         sasl-ir service-name=s socks4=s socks4a=s socks5-basic
145         socks5-gssapi-service-name=s socks5-gssapi socks5-hostname=s socks5=s
146         speed-limit|Y speed-type|y ssl-allow-beast sslv2 sslv3
147         suppress-connect-headers tcp-fastopen tls-max=s
148         tls13-ciphers=s tlsauthtype=s tlspassword=s tlsuser=s
149         tlsv1 trace-ascii=s trace-time trace=s
150         unix-socket=s user-agent|A=s user|u=s
151 )
152 }
153
154 1;