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 POSIX ();
   9
  10 sub prep_ext { # externals_each callback
  11         my ($lxs, $exclude, $loc) = @_;
  12         $lxs->prepare_external($loc) unless $exclude->{$loc};
  13 }
  14
  15 sub _start_query {
  16         my ($self) = @_;
  17         if ($self->{net}) {
  18                 require PublicInbox::LeiAuth;
  19                 $self->{auth} = PublicInbox::LeiAuth->new
  20         }
  21         $self->{lxs}->do_query($self);
  22 }
  23
  24 sub qstr_add { # PublicInbox::InputPipe::consume callback for --stdin
  25         my ($self) = @_; # $_[1] = $rbuf
  26         if (defined($_[1])) {
  27                 $_[1] eq '' and return eval {
  28                         $self->{lse}->query_approxidate($self->{lse}->git,
  29                                                 $self->{mset_opt}->{qstr});
  30                         _start_query($self);
  31                 };
  32                 $self->{mset_opt}->{qstr} .= $_[1];
  33         } else {
  34                 $self->fail("error reading stdin: $!");
  35         }
  36 }
  37
  38 sub lxs_prepare {
  39         my ($self) = @_;
  40         require PublicInbox::LeiXSearch;
  41         # prepare any number of LeiXSearch || LeiSearch || Inbox || URL
  42         my $lxs = $self->{lxs} = PublicInbox::LeiXSearch->new;
  43         my $opt = $self->{opt};
  44         my @only = @{$opt->{only} // []};
  45         # --local is enabled by default unless --only is used
  46         # we'll allow "--only $LOCATION --local"
  47         my $sto = $self->_lei_store(1); # FIXME: should not create
  48         $self->{lse} = $sto->search;
  49         if ($opt->{'local'} //= scalar(@only) ? 0 : 1) {
  50                 $lxs->prepare_external($self->{lse});
  51         }
  52         if (@only) {
  53                 for my $loc (@only) {
  54                         my @loc = $self->get_externals($loc) or return;
  55                         $lxs->prepare_external($_) for @loc;
  56                 }
  57         } else {
  58                 my (@ilocals, @iremotes);
  59                 for my $loc (@{$opt->{include} // []}) {
  60                         my @loc = $self->get_externals($loc) or return;
  61                         $lxs->prepare_external($_) for @loc;
  62                         @ilocals = @{$lxs->{locals} // []};
  63                         @iremotes = @{$lxs->{remotes} // []};
  64                 }
  65                 # --external is enabled by default, but allow --no-external
  66                 if ($opt->{external} //= 1) {
  67                         my %x;
  68                         for my $loc (@{$opt->{exclude} // []}) {
  69                                 my @l = $self->get_externals($loc, 1) or return;
  70                                 $x{$_} = 1 for @l;
  71                         }
  72                         my $ne = $self->externals_each(\&prep_ext, $lxs, \%x);
  73                         $opt->{remote} //= !($lxs->locals - $opt->{'local'});
  74                         if ($opt->{'local'}) {
  75                                 $lxs->{remotes} = \@iremotes if !$opt->{remote};
  76                         } else {
  77                                 $lxs->{locals} = \@ilocals;
  78                         }
  79                 }
  80         }
  81         ($lxs->locals || $lxs->remotes) ? ($self->{lxs} = $lxs) :
  82                 $self->fail('no local or remote inboxes to search');
  83 }
  84
  85 # the main "lei q SEARCH_TERMS" method
  86 sub lei_q {
  87         my ($self, @argv) = @_;
  88         require PublicInbox::LeiOverview;
  89         require PublicInbox::V2Writable;
  90         PublicInbox::Config->json; # preload before forking
  91         PublicInbox::LeiOverview->new($self) or return;
  92         my $lxs = lxs_prepare($self) or return;
  93         $self->ale->refresh_externals($lxs);
  94         my $opt = $self->{opt};
  95         my ($xj, $mj) = split(/,/, $opt->{jobs} // '');
  96         if (defined($xj) && $xj ne '' && $xj !~ /\A[1-9][0-9]*\z/) {
  97                 return $self->fail("`$xj' search jobs must be >= 1");
  98         }
  99         $xj ||= $lxs->concurrency($opt); # allow: "--jobs ,$WRITER_ONLY"
 100         my $nproc = $lxs->detect_nproc // 1; # don't memoize, schedtool(1) exists
 101         $xj = $nproc if $xj > $nproc;
 102         $lxs->{-wq_nr_workers} = $xj;
 103         if (defined($mj) && $mj !~ /\A[1-9][0-9]*\z/) {
 104                 return $self->fail("`$mj' writer jobs must be >= 1");
 105         }
 106         if ($self->{l2m} && ($opt->{'import-remote'} //= 1) |
 107                                 # we use \1 (a ref) to distinguish between
 108                                 # user-supplied and default value
 109                                 (($opt->{'import-before'} //= \1) ? 1 : 0)) {
 110                 $self->_lei_store(1)->write_prepare($self);
 111         }
 112         $self->{l2m} and $self->{l2m}->{-wq_nr_workers} = $mj // do {
 113                 $mj = POSIX::lround($nproc * 3 / 4); # keep some CPU for git
 114                 $mj <= 0 ? 1 : $mj;
 115         };
 116
 117         my %mset_opt = map { $_ => $opt->{$_} } qw(threads limit offset);
 118         $mset_opt{asc} = $opt->{'reverse'} ? 1 : 0;
 119         $mset_opt{limit} //= 10000;
 120         if (defined(my $sort = $opt->{'sort'})) {
 121                 if ($sort eq 'relevance') {
 122                         $mset_opt{relevance} = 1;
 123                 } elsif ($sort eq 'docid') {
 124                         $mset_opt{relevance} = $mset_opt{asc} ? -1 : -2;
 125                 } elsif ($sort =~ /\Areceived(?:-?[aA]t)?\z/) {
 126                         # the default
 127                 } else {
 128                         die "unrecognized --sort=$sort\n";
 129                 }
 130         }
 131         # descending docid order
 132         $mset_opt{relevance} //= -2 if $opt->{threads};
 133         $self->{mset_opt} = \%mset_opt;
 134
 135         if ($opt->{stdin}) {
 136                 return $self->fail(<<'') if @argv;
 137 no query allowed on command-line with --stdin
 138
 139                 require PublicInbox::InputPipe;
 140                 PublicInbox::InputPipe::consume($self->{0}, \&qstr_add, $self);
 141                 return;
 142         }
 143         $mset_opt{qstr} =
 144                 $self->{lse}->query_argv_to_string($self->{lse}->git, \@argv);
 145         _start_query($self);
 146 }
 147
 148 # shell completion helper called by lei__complete
 149 sub _complete_q {
 150         my ($self, @argv) = @_;
 151         my $ext = qr/\A(?:-I|(?:--(?:include|exclude|only)))\z/;
 152         my @cur;
 153         while (@argv) {
 154                 if ($argv[-1] =~ $ext) {
 155                         my @c = $self->_complete_forget_external(@cur);
 156                         # try basename match:
 157                         if (scalar(@cur) == 1 && index($cur[0], '/') < 0) {
 158                                 my $all = $self->externals_each;
 159                                 my %bn;
 160                                 for my $loc (keys %$all) {
 161                                         my $bn = (split(m!/!, $loc))[-1];
 162                                         ++$bn{$bn};
 163                                 }
 164                                 push @c, grep {
 165                                         $bn{$_} == 1 && /\A\Q$cur[0]/
 166                                 } keys %bn;
 167                         }
 168                         return @c if @c;
 169                 }
 170                 unshift(@cur, pop @argv);
 171         }
 172         ();
 173 }
 174
 175 # Stuff we may pass through to curl (as of 7.64.0), see curl manpage for
 176 # details, so most options which make sense for HTTP/HTTPS (including proxy
 177 # support for Tor and other methods of getting past weird networks).
 178 # Most of these are untested by us, some may not make sense for our use case
 179 # and typos below are likely.
 180 # n.b. some short options (-$NUMBER) are not supported since they conflict
 181 # with other "lei q" switches.
 182 # FIXME: Getopt::Long doesn't easily let us support support options with
 183 # '.' in them (e.g. --http1.1)
 184 # TODO: should we depend on "-c http.*" options for things which have
 185 # analogues in git(1)? that would reduce likelyhood of conflicts with
 186 # our other CLI options
 187 # Note: some names are renamed to avoid potential conflicts,
 188 # see %lei2mail in lib/PublicInbox/LeiCurl.pm
 189 sub curl_opt { qw(
 190         curl-config=s@
 191         abstract-unix-socket=s anyauth basic cacert=s capath=s
 192         cert-status cert-type cert=s ciphers=s
 193         connect-timeout=s connect-to=s cookie-jar=s cookie=s crlfile=s
 194         digest disable dns-interface=s dns-ipv4-addr=s dns-ipv6-addr=s
 195         dns-servers=s doh-url=s egd-file=s engine=s false-start
 196         happy-eyeballs-timeout-ms=s haproxy-protocol header=s@
 197         http2-prior-knowledge http2 insecure
 198         interface=s ipv4 ipv6 junk-session-cookies
 199         key-type=s key=s limit-rate=s local-port=s location-trusted location
 200         max-redirs=i max-time=s negotiate netrc-file=s netrc-optional netrc
 201         no-alpn no-buffer no-npn no-sessionid noproxy=s ntlm-wb ntlm
 202         pass=s pinnedpubkey=s post301 post302 post303 preproxy=s
 203         proxy-anyauth proxy-basic proxy-cacert=s proxy-capath=s
 204         proxy-cert-type=s proxy-cert=s proxy-ciphers=s proxy-crlfile=s
 205         proxy-digest proxy-header=s@ proxy-insecure
 206         proxy-key-type=s proxy-key proxy-negotiate proxy-ntlm proxy-pass=s
 207         proxy-pinnedpubkey=s proxy-service-name=s proxy-ssl-allow-beast
 208         proxy-tls13-ciphers=s proxy-tlsauthtype=s proxy-tlspassword=s
 209         proxy-tlsuser=s proxy-tlsv1 proxy-user=s proxy=s
 210         proxytunnel=s pubkey=s random-file=s referer=s resolve=s
 211         retry-connrefused retry-delay=s retry-max-time=s retry=i
 212         sasl-ir service-name=s socks4=s socks4a=s socks5-basic
 213         socks5-gssapi-service-name=s socks5-gssapi socks5-hostname=s socks5=s
 214         speed-limit speed-type ssl-allow-beast sslv2 sslv3
 215         suppress-connect-headers tcp-fastopen tls-max=s
 216         tls13-ciphers=s tlsauthtype=s tlspassword=s tlsuser=s
 217         tlsv1 trace-ascii=s trace-time trace=s
 218         unix-socket=s user-agent=s user=s
 219 )
 220 }
 221
 222 1;