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" command and provides internals for
   5 # several other sub-commands (up, lcat, ...)
   6 package PublicInbox::LeiQuery;
   7 use strict;
   8 use v5.10.1;
   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 { # used by "lei q" and "lei up"
  16         my ($self) = @_;
  17         require PublicInbox::LeiOverview;
  18         PublicInbox::LeiOverview->new($self) or return;
  19         my $opt = $self->{opt};
  20         my ($xj, $mj) = split(/,/, $opt->{jobs} // '');
  21         if (defined($xj) && $xj ne '' && $xj !~ /\A[1-9][0-9]*\z/) {
  22                 return $self->fail("`$xj' search jobs must be >= 1");
  23         }
  24         my $lxs = $self->{lxs};
  25         $xj ||= $lxs->concurrency($opt); # allow: "--jobs ,$WRITER_ONLY"
  26         my $nproc = $lxs->detect_nproc || 1; # don't memoize, schedtool(1) exists
  27         $xj = $nproc if $xj > $nproc;
  28         $lxs->{-wq_nr_workers} = $xj;
  29         if (defined($mj) && $mj !~ /\A[1-9][0-9]*\z/) {
  30                 return $self->fail("`$mj' writer jobs must be >= 1");
  31         }
  32         my $l2m = $self->{l2m};
  33         # we use \1 (a ref) to distinguish between default vs. user-supplied
  34         if ($l2m && grep { $opt->{$_} //= \1 } (qw(mail-sync import-remote
  35                                                         import-before))) {
  36                 $self->_lei_store(1)->write_prepare($self);
  37         }
  38         $l2m and $l2m->{-wq_nr_workers} = $mj // do {
  39                 $mj = int($nproc * 0.75 + 0.5); # keep some CPU for git
  40         };
  41
  42         # descending docid order is cheapest, MUA controls sorting order
  43         $self->{mset_opt}->{relevance} //= -2 if $l2m || $opt->{threads};
  44
  45         my $tot = $self->{mset_opt}->{total} //= $self->{opt}->{limit} // 10000;
  46         $self->{mset_opt}->{limit} = $tot > 10000 ? 10000 : $tot;
  47         $self->{mset_opt}->{offset} //= 0;
  48         $self->{mset_opt}->{threads} //= $opt->{threads};
  49
  50         if ($self->{net}) {
  51                 require PublicInbox::LeiAuth;
  52                 $self->{auth} = PublicInbox::LeiAuth->new
  53         }
  54         $lxs->do_query($self);
  55 }
  56
  57 sub qstr_add { # PublicInbox::InputPipe::consume callback for --stdin
  58         my ($self) = @_; # $_[1] = $rbuf
  59         if (defined($_[1])) {
  60                 $_[1] eq '' and return eval {
  61                         $self->fchdir or return;
  62                         $self->{mset_opt}->{q_raw} = $self->{mset_opt}->{qstr};
  63                         $self->{lse}->query_approxidate($self->{lse}->git,
  64                                                 $self->{mset_opt}->{qstr});
  65                         _start_query($self);
  66                 };
  67                 $self->{mset_opt}->{qstr} .= $_[1];
  68         } else {
  69                 $self->fail("error reading stdin: $!");
  70         }
  71 }
  72
  73 sub lxs_prepare {
  74         my ($self) = @_;
  75         require PublicInbox::LeiXSearch;
  76         # prepare any number of LeiXSearch || LeiSearch || Inbox || URL
  77         my $lxs = $self->{lxs} = PublicInbox::LeiXSearch->new;
  78         my $opt = $self->{opt};
  79         my @only = @{$opt->{only} // []};
  80         # --local is enabled by default unless --only is used
  81         # we'll allow "--only $LOCATION --local"
  82         my $sto = $self->_lei_store(1);
  83         $self->{lse} = $sto->search;
  84         if ($opt->{'local'} //= scalar(@only) ? 0 : 1) {
  85                 $lxs->prepare_external($self->{lse});
  86         }
  87         if (@only) {
  88                 for my $loc (@only) {
  89                         my @loc = $self->get_externals($loc) or return;
  90                         $lxs->prepare_external($_) for @loc;
  91                 }
  92         } else {
  93                 my (@ilocals, @iremotes);
  94                 for my $loc (@{$opt->{include} // []}) {
  95                         my @loc = $self->get_externals($loc) or return;
  96                         $lxs->prepare_external($_) for @loc;
  97                         @ilocals = @{$lxs->{locals} // []};
  98                         @iremotes = @{$lxs->{remotes} // []};
  99                 }
 100                 # --external is enabled by default, but allow --no-external
 101                 if ($opt->{external} //= 1) {
 102                         my %x;
 103                         for my $loc (@{$opt->{exclude} // []}) {
 104                                 my @l = $self->get_externals($loc, 1) or return;
 105                                 $x{$_} = 1 for @l;
 106                         }
 107                         my $ne = $self->externals_each(\&prep_ext, $lxs, \%x);
 108                         $opt->{remote} //= !($lxs->locals - $opt->{'local'});
 109                         if ($opt->{'local'}) {
 110                                 $lxs->{remotes} = \@iremotes if !$opt->{remote};
 111                         } else {
 112                                 $lxs->{locals} = \@ilocals;
 113                         }
 114                 }
 115         }
 116         ($lxs->locals || $lxs->remotes) ? ($self->{lxs} = $lxs) :
 117                 $self->fail('no local or remote inboxes to search');
 118 }
 119
 120 # the main "lei q SEARCH_TERMS" method
 121 sub lei_q {
 122         my ($self, @argv) = @_;
 123         PublicInbox::Config->json; # preload before forking
 124         my $lxs = lxs_prepare($self) or return;
 125         $self->ale->refresh_externals($lxs, $self);
 126         my $opt = $self->{opt};
 127         my %mset_opt;
 128         $mset_opt{asc} = $opt->{'reverse'} ? 1 : 0;
 129         if (defined(my $sort = $opt->{'sort'})) {
 130                 if ($sort eq 'relevance') {
 131                         $mset_opt{relevance} = 1;
 132                 } elsif ($sort eq 'docid') {
 133                         $mset_opt{relevance} = $mset_opt{asc} ? -1 : -2;
 134                 } elsif ($sort =~ /\Areceived(?:-?[aA]t)?\z/) {
 135                         # the default
 136                 } else {
 137                         die "unrecognized --sort=$sort\n";
 138                 }
 139                 $opt->{save} and return
 140                         $self->fail('--save and --sort are incompatible');
 141         }
 142         $self->{mset_opt} = \%mset_opt;
 143
 144         if ($opt->{stdin}) {
 145                 return $self->fail(<<'') if @argv;
 146 no query allowed on command-line with --stdin
 147
 148                 require PublicInbox::InputPipe;
 149                 PublicInbox::InputPipe::consume($self->{0}, \&qstr_add, $self);
 150                 return;
 151         }
 152         $mset_opt{q_raw} = [ @argv ]; # copy
 153         $mset_opt{qstr} =
 154                 $self->{lse}->query_argv_to_string($self->{lse}->git, \@argv);
 155         _start_query($self);
 156 }
 157
 158 # shell completion helper called by lei__complete
 159 sub _complete_q {
 160         my ($self, @argv) = @_;
 161         my $ext = qr/\A(?:-I|(?:--(?:include|exclude|only)))\z/;
 162         my @cur;
 163         while (@argv) {
 164                 if ($argv[-1] =~ $ext) {
 165                         my @c = $self->_complete_forget_external(@cur);
 166                         # try basename match:
 167                         if (scalar(@cur) == 1 && index($cur[0], '/') < 0) {
 168                                 my $all = $self->externals_each;
 169                                 my %bn;
 170                                 for my $loc (keys %$all) {
 171                                         my $bn = (split(m!/!, $loc))[-1];
 172                                         ++$bn{$bn};
 173                                 }
 174                                 push @c, grep {
 175                                         $bn{$_} == 1 && /\A\Q$cur[0]/
 176                                 } keys %bn;
 177                         }
 178                         return @c if @c;
 179                 }
 180                 unshift(@cur, pop @argv);
 181         }
 182         ();
 183 }
 184
 185 # Stuff we may pass through to curl (as of 7.64.0), see curl manpage for
 186 # details, so most options which make sense for HTTP/HTTPS (including proxy
 187 # support for Tor and other methods of getting past weird networks).
 188 # Most of these are untested by us, some may not make sense for our use case
 189 # and typos below are likely.
 190 # n.b. some short options (-$NUMBER) are not supported since they conflict
 191 # with other "lei q" switches.
 192 # FIXME: Getopt::Long doesn't easily let us support support options with
 193 # '.' in them (e.g. --http1.1)
 194 # TODO: should we depend on "-c http.*" options for things which have
 195 # analogues in git(1)? that would reduce likelyhood of conflicts with
 196 # our other CLI options
 197 # Note: some names are renamed to avoid potential conflicts,
 198 # see %lei2curl in lib/PublicInbox/LeiCurl.pm
 199 sub curl_opt { qw(
 200         curl-config=s@
 201         abstract-unix-socket=s anyauth basic cacert=s capath=s
 202         cert-status cert-type cert=s ciphers=s
 203         connect-timeout=s connect-to=s cookie-jar=s cookie=s crlfile=s
 204         digest disable dns-interface=s dns-ipv4-addr=s dns-ipv6-addr=s
 205         dns-servers=s doh-url=s egd-file=s engine=s false-start
 206         happy-eyeballs-timeout-ms=s haproxy-protocol header=s@
 207         http2-prior-knowledge http2 insecure
 208         interface=s ipv4 ipv6 junk-session-cookies
 209         key-type=s key=s limit-rate=s local-port=s location-trusted location
 210         max-redirs=i max-time=s negotiate netrc-file=s netrc-optional netrc
 211         no-alpn no-buffer no-npn no-sessionid noproxy=s ntlm-wb ntlm
 212         pass=s pinnedpubkey=s post301 post302 post303 preproxy=s
 213         proxy-anyauth proxy-basic proxy-cacert=s proxy-capath=s
 214         proxy-cert-type=s proxy-cert=s proxy-ciphers=s proxy-crlfile=s
 215         proxy-digest proxy-header=s@ proxy-insecure
 216         proxy-key-type=s proxy-key proxy-negotiate proxy-ntlm proxy-pass=s
 217         proxy-pinnedpubkey=s proxy-service-name=s proxy-ssl-allow-beast
 218         proxy-tls13-ciphers=s proxy-tlsauthtype=s proxy-tlspassword=s
 219         proxy-tlsuser=s proxy-tlsv1 proxy-user=s proxy=s
 220         proxytunnel=s pubkey=s random-file=s referer=s resolve=s
 221         retry-connrefused retry-delay=s retry-max-time=s retry=i
 222         sasl-ir service-name=s socks4=s socks4a=s socks5-basic
 223         socks5-gssapi-service-name=s socks5-gssapi socks5-hostname=s socks5=s
 224         speed-limit speed-type ssl-allow-beast sslv2 sslv3
 225         suppress-connect-headers tcp-fastopen tls-max=s
 226         tls13-ciphers=s tlsauthtype=s tlspassword=s tlsuser=s
 227         tlsv1 trace-ascii=s trace-time trace=s
 228         unix-socket=s user-agent=s user=s
 229 )
 230 }
 231
 232 1;