]> Sergey Matveev's repositories - public-inbox.git/blobdiff - lib/PublicInbox/LeiQuery.pm
lei_to_mail: limit workers for text, reply and v2 outputs
[public-inbox.git] / lib / PublicInbox / LeiQuery.pm
index 3a437bf0030842c291c572af7301ff6c11b28a57..352ee60131aaf6bc9baabfbd4ae735c4a0f06568 100644 (file)
@@ -1,38 +1,72 @@
 # Copyright (C) 2021 all contributors <meta@public-inbox.org>
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
 
-# handles lei <q|ls-query|rm-query|mv-query> commands
+# handles "lei q" command and provides internals for
+# several other sub-commands (up, lcat, ...)
 package PublicInbox::LeiQuery;
 use strict;
 use v5.10.1;
-use POSIX ();
 
 sub prep_ext { # externals_each callback
        my ($lxs, $exclude, $loc) = @_;
        $lxs->prepare_external($loc) unless $exclude->{$loc};
 }
 
-sub _start_query {
+sub _start_query { # used by "lei q" and "lei up"
        my ($self) = @_;
+       require PublicInbox::LeiOverview;
+       PublicInbox::LeiOverview->new($self) or return;
+       my $opt = $self->{opt};
+       my ($xj, $mj) = split(/,/, $opt->{jobs} // '');
+       (defined($xj) && $xj ne '' && $xj !~ /\A[1-9][0-9]*\z/) and
+               die "`$xj' search jobs must be >= 1\n";
+       my $lxs = $self->{lxs};
+       $xj ||= $lxs->concurrency($opt); # allow: "--jobs ,$WRITER_ONLY"
+       my $nproc = $lxs->detect_nproc || 1; # don't memoize, schedtool(1) exists
+       $xj = $nproc if $xj > $nproc;
+       $lxs->{-wq_nr_workers} = $xj;
+       (defined($mj) && $mj !~ /\A[1-9][0-9]*\z/) and
+               die "`$mj' writer jobs must be >= 1\n";
+       my $l2m = $self->{l2m};
+       # we use \1 (a ref) to distinguish between default vs. user-supplied
+       if ($l2m && grep { $opt->{$_} //= \1 } (qw(mail-sync import-remote
+                                                       import-before))) {
+               $self->_lei_store(1)->write_prepare($self);
+               if ($opt->{'mail-sync'}) {
+                       my $lms = $l2m->{-lms_rw} = $self->lms(1);
+                       $lms->lms_write_prepare->lms_pause; # just create
+               }
+       }
+       $l2m and $l2m->{-wq_nr_workers} //= $mj //
+               int($nproc * 0.75 + 0.5); # keep some CPU for git
+
+       # descending docid order is cheapest, MUA controls sorting order
+       $self->{mset_opt}->{relevance} //= -2 if $l2m || $opt->{threads};
+
+       my $tot = $self->{mset_opt}->{total} //= $self->{opt}->{limit} // 10000;
+       $self->{mset_opt}->{limit} = $tot > 10000 ? 10000 : $tot;
+       $self->{mset_opt}->{offset} //= 0;
+       $self->{mset_opt}->{threads} //= $opt->{threads};
+
        if ($self->{net}) {
                require PublicInbox::LeiAuth;
                $self->{auth} = PublicInbox::LeiAuth->new
        }
-       $self->{lxs}->do_query($self);
+       $lxs->do_query($self);
 }
 
 sub qstr_add { # PublicInbox::InputPipe::consume callback for --stdin
-       my ($self) = @_; # $_[1] = $rbuf
-       if (defined($_[1])) {
-               $_[1] eq '' and return eval {
-                       $self->{lse}->query_approxidate($self->{lse}->git,
-                                               $self->{mset_opt}->{qstr});
-                       _start_query($self);
-               };
-               $self->{mset_opt}->{qstr} .= $_[1];
-       } else {
-               $self->fail("error reading stdin: $!");
-       }
+       my ($lei) = @_; # $_[1] = $rbuf
+       $_[1] // $lei->fail("error reading stdin: $!");
+       return $lei->{mset_opt}->{qstr} .= $_[1] if $_[1] ne '';
+       eval {
+               $lei->fchdir;
+               $lei->{mset_opt}->{q_raw} = $lei->{mset_opt}->{qstr};
+               $lei->{lse}->query_approxidate($lei->{lse}->git,
+                                               $lei->{mset_opt}->{qstr});
+               _start_query($lei);
+       };
+       $lei->fail($@) if $@;
 }
 
 sub lxs_prepare {
@@ -44,7 +78,7 @@ sub lxs_prepare {
        my @only = @{$opt->{only} // []};
        # --local is enabled by default unless --only is used
        # we'll allow "--only $LOCATION --local"
-       my $sto = $self->_lei_store(1); # FIXME: should not create
+       my $sto = $self->_lei_store(1);
        $self->{lse} = $sto->search;
        if ($opt->{'local'} //= scalar(@only) ? 0 : 1) {
                $lxs->prepare_external($self->{lse});
@@ -64,59 +98,26 @@ sub lxs_prepare {
                }
                # --external is enabled by default, but allow --no-external
                if ($opt->{external} //= 1) {
-                       my %x;
-                       for my $loc (@{$opt->{exclude} // []}) {
-                               my @l = $self->get_externals($loc, 1) or return;
-                               $x{$_} = 1 for @l;
-                       }
-                       my $ne = $self->externals_each(\&prep_ext, $lxs, \%x);
+                       my $ex = $self->canonicalize_excludes($opt->{exclude});
+                       $self->externals_each(\&prep_ext, $lxs, $ex);
                        $opt->{remote} //= !($lxs->locals - $opt->{'local'});
-                       if ($opt->{'local'}) {
-                               $lxs->{remotes} = \@iremotes if !$opt->{remote};
-                       } else {
-                               $lxs->{locals} = \@ilocals;
-                       }
+                       $lxs->{locals} = \@ilocals if !$opt->{'local'};
+                       $lxs->{remotes} = \@iremotes if !$opt->{remote};
                }
        }
        ($lxs->locals || $lxs->remotes) ? ($self->{lxs} = $lxs) :
-               $self->fail('no local or remote inboxes to search');
+               die("no local or remote inboxes to search\n");
 }
 
 # the main "lei q SEARCH_TERMS" method
 sub lei_q {
        my ($self, @argv) = @_;
-       require PublicInbox::LeiOverview;
-       require PublicInbox::V2Writable;
        PublicInbox::Config->json; # preload before forking
-       PublicInbox::LeiOverview->new($self) or return;
        my $lxs = lxs_prepare($self) or return;
-       $self->ale->refresh_externals($lxs);
+       $self->ale->refresh_externals($lxs, $self);
        my $opt = $self->{opt};
-       my ($xj, $mj) = split(/,/, $opt->{jobs} // '');
-       if (defined($xj) && $xj ne '' && $xj !~ /\A[1-9][0-9]*\z/) {
-               return $self->fail("`$xj' search jobs must be >= 1");
-       }
-       $xj ||= $lxs->concurrency($opt); # allow: "--jobs ,$WRITER_ONLY"
-       my $nproc = $lxs->detect_nproc // 1; # don't memoize, schedtool(1) exists
-       $xj = $nproc if $xj > $nproc;
-       $lxs->{-wq_nr_workers} = $xj;
-       if (defined($mj) && $mj !~ /\A[1-9][0-9]*\z/) {
-               return $self->fail("`$mj' writer jobs must be >= 1");
-       }
-       if ($self->{l2m} && ($opt->{'import-remote'} //= 1) |
-                               # we use \1 (a ref) to distinguish between
-                               # user-supplied and default value
-                               (($opt->{'import-before'} //= \1) ? 1 : 0)) {
-               $self->_lei_store(1)->write_prepare($self);
-       }
-       $self->{l2m} and $self->{l2m}->{-wq_nr_workers} = $mj // do {
-               $mj = POSIX::lround($nproc * 3 / 4); # keep some CPU for git
-               $mj <= 0 ? 1 : $mj;
-       };
-
-       my %mset_opt = map { $_ => $opt->{$_} } qw(threads limit offset);
+       my %mset_opt;
        $mset_opt{asc} = $opt->{'reverse'} ? 1 : 0;
-       $mset_opt{limit} //= 10000;
        if (defined(my $sort = $opt->{'sort'})) {
                if ($sort eq 'relevance') {
                        $mset_opt{relevance} = 1;
@@ -127,9 +128,9 @@ sub lei_q {
                } else {
                        die "unrecognized --sort=$sort\n";
                }
+               $opt->{save} and return
+                       $self->fail('--save and --sort are incompatible');
        }
-       # descending docid order is cheapest, MUA controls sorting order
-       $mset_opt{relevance} //= -2 if $self->{l2m} || $opt->{threads};
        $self->{mset_opt} = \%mset_opt;
 
        if ($opt->{stdin}) {
@@ -140,6 +141,7 @@ no query allowed on command-line with --stdin
                PublicInbox::InputPipe::consume($self->{0}, \&qstr_add, $self);
                return;
        }
+       $mset_opt{q_raw} = [ @argv ]; # copy
        $mset_opt{qstr} =
                $self->{lse}->query_argv_to_string($self->{lse}->git, \@argv);
        _start_query($self);
@@ -148,11 +150,11 @@ no query allowed on command-line with --stdin
 # shell completion helper called by lei__complete
 sub _complete_q {
        my ($self, @argv) = @_;
-       my $ext = qr/\A(?:-I|(?:--(?:include|exclude|only)))\z/;
        my @cur;
+       my $cb = $self->lazy_cb(qw(forget-external _complete_));
        while (@argv) {
-               if ($argv[-1] =~ $ext) {
-                       my @c = $self->_complete_forget_external(@cur);
+               if ($argv[-1] =~ /\A(?:-I|(?:--(?:include|exclude|only)))\z/) {
+                       my @c = $cb->($self, @cur);
                        # try basename match:
                        if (scalar(@cur) == 1 && index($cur[0], '/') < 0) {
                                my $all = $self->externals_each;
@@ -185,7 +187,7 @@ sub _complete_q {
 # analogues in git(1)? that would reduce likelyhood of conflicts with
 # our other CLI options
 # Note: some names are renamed to avoid potential conflicts,
-# see %lei2mail in lib/PublicInbox/LeiCurl.pm
+# see %lei2curl in lib/PublicInbox/LeiCurl.pm
 sub curl_opt { qw(
        curl-config=s@
        abstract-unix-socket=s anyauth basic cacert=s capath=s