]> Sergey Matveev's repositories - public-inbox.git/blobdiff - lib/PublicInbox/LeiXSearch.pm
lei: track seen messages to note duplicates
[public-inbox.git] / lib / PublicInbox / LeiXSearch.pm
index fba168613d96a2cdec8f8baec1d350e8dff76d40..6f8770191ee6a28b36b1f7f631cbb9ebc7b26e86 100644 (file)
@@ -1,4 +1,4 @@
-# Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
+# Copyright (C) all contributors <meta@public-inbox.org>
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
 
 # Combine any combination of PublicInbox::Search,
@@ -142,11 +142,11 @@ sub wait_startq ($) {
                                delete $lei->{opt}->{verbose};
                                delete $lei->{-progress};
                        } else {
-                               $lei->fail("$$ WTF `$do_augment_done'");
+                               die "BUG: do_augment_done=`$do_augment_done'";
                        }
                        return;
                }
-               return $lei->fail("$$ wait_startq: $!") unless $!{EINTR};
+               die "wait_startq: $!" unless $!{EINTR};
        }
 }
 
@@ -163,8 +163,9 @@ sub mset_progress {
 }
 
 sub l2m_progress {
-       my ($lei, $nr) = @_;
-       $lei->{-nr_write} += $nr;
+       my ($lei, $nr_write, $nr_seen) = @_;
+       $lei->{-nr_write} += $nr_write;
+       $lei->{-nr_seen} += $nr_seen;
 }
 
 sub query_one_mset { # for --threads and l2m w/o sort
@@ -282,23 +283,23 @@ sub each_remote_eml { # callback for MboxReader->mboxrd
        my $xoids = $lei->{ale}->xoids_for($eml, 1);
        my $smsg = bless {}, 'PublicInbox::Smsg';
        if ($self->{import_sto} && !$xoids) {
-               my $res = $self->{import_sto}->wq_do('add_eml', $eml);
+               my ($res, $kw) = $self->{import_sto}->wq_do('add_eml', $eml);
                if (ref($res) eq ref($smsg)) { # totally new message
                        $smsg = $res;
-                       $smsg->{kw} = []; # short-circuit xsmsg_vmd
+                       $self->{-sto_imported} = 1;
                }
+               $smsg->{kw} = $kw; # short-circuit xsmsg_vmd
        }
        $smsg->{blob} //= $xoids ? (keys(%$xoids))[0]
                                : $lei->git_oid($eml)->hexdigest;
        _smsg_fill($smsg, $eml);
        wait_startq($lei);
+       my $nr = ++$lei->{-nr_remote_eml}; # needed for lss->cfg_set
        if ($lei->{-progress}) {
-               ++$lei->{-nr_remote_eml};
                my $now = now();
                my $next = $lei->{-next_progress} //= ($now + 1);
                if ($now > $next) {
                        $lei->{-next_progress} = $now + 1;
-                       my $nr = $lei->{-nr_remote_eml};
                        mset_progress($lei, $lei->{-current_url}, $nr, '?');
                }
        }
@@ -342,7 +343,7 @@ sub query_remote_mboxrd {
        local $SIG{TERM} = sub { exit(0) }; # for DESTROY (File::Temp, $reap)
        my $lei = $self->{lei};
        my $opt = $lei->{opt};
-       my $qstr = $lei->{mset_opt}->{qstr};
+       chomp(my $qstr = $lei->{mset_opt}->{qstr});
        $qstr =~ s/[ \n\t]+/ /sg; # make URLs less ugly
        my @qform = (x => 'm');
        push(@qform, t => 1) if $opt->{threads};
@@ -376,13 +377,14 @@ sub query_remote_mboxrd {
                $fh = IO::Uncompress::Gunzip->new($fh, MultiStream => 1);
                PublicInbox::MboxReader->mboxrd($fh, \&each_remote_eml, $self,
                                                $lei, $each_smsg);
-               my $nr = $lei->{-nr_remote_eml};
-               my $wait = $lei->{sto}->wq_do('done') if $nr && $lei->{sto};
+               if (delete($self->{-sto_imported})) {
+                       my $wait = $self->{import_sto}->wq_do('done');
+               }
                $reap_curl->join;
                if ($? == 0) {
                        # don't update if no results, maybe MTA is down
-                       $key && $nr and
-                               $lei->{lss}->cfg_set($key, $start);
+                       my $nr = $lei->{-nr_remote_eml};
+                       $lei->{lss}->cfg_set($key, $start) if $key && $nr;
                        mset_progress($lei, $lei->{-current_url}, $nr, $nr);
                        next;
                }
@@ -408,72 +410,81 @@ sub git { $_[0]->{git} // die 'BUG: git uninitialized' }
 sub xsearch_done_wait { # dwaitpid callback
        my ($arg, $pid) = @_;
        my ($wq, $lei) = @$arg;
-       $lei->child_error($?, 'non-fatal error from '.ref($wq)) if $?;
+       return if !$?;
+       my $s = $? & 127;
+       return $lei->child_error($?) if $s == 13 || $s == 15;
+       $lei->child_error($?, 'non-fatal error from '.ref($wq)." \$?=$?");
 }
 
 sub query_done { # EOF callback for main daemon
        my ($lei) = @_;
        local $PublicInbox::LEI::current_lei = $lei;
-       my $l2m = delete $lei->{l2m};
-       $l2m->wq_wait_old(\&xsearch_done_wait, $lei) if $l2m;
-       if (my $lxs = delete $lei->{lxs}) {
-               $lxs->wq_wait_old(\&xsearch_done_wait, $lei);
-       }
-       ($lei->{opt}->{'mail-sync'} && !$lei->{sto}) and
-               warn "BUG: {sto} missing with --mail-sync";
-       $lei->sto_done_request if $lei->{sto};
-       my $wait = $lei->{v2w} ? $lei->{v2w}->wq_do('done') : undef;
-       $lei->{ovv}->ovv_end($lei);
-       my $start_mua;
-       if ($l2m) { # close() calls LeiToMail reap_compress
-               if (my $out = delete $lei->{old_1}) {
-                       if (my $mbout = $lei->{1}) {
-                               close($mbout) or return $lei->fail(<<"");
-Error closing $lei->{ovv}->{dst}: $!
+       eval {
+               my $l2m = delete $lei->{l2m};
+               delete $lei->{lxs};
+               ($lei->{opt}->{'mail-sync'} && !$lei->{sto}) and
+                       warn "BUG: {sto} missing with --mail-sync";
+               $lei->sto_done_request if $lei->{sto};
+               if (my $v2w = delete $lei->{v2w}) {
+                       my $wait = $v2w->wq_do('done'); # may die
+                       $v2w->wq_close;
+               }
+               $lei->{ovv}->ovv_end($lei);
+               if ($l2m) { # close() calls LeiToMail reap_compress
+                       if (my $out = delete $lei->{old_1}) {
+                               if (my $mbout = $lei->{1}) {
+                                       close($mbout) or die <<"";
+Error closing $lei->{ovv}->{dst}: \$!=$! \$?=$?
 
+                               }
+                               $lei->{1} = $out;
+                       }
+                       if ($l2m->lock_free) {
+                               $l2m->poke_dst;
+                               $lei->poke_mua;
+                       } else { # mbox users
+                               delete $l2m->{mbl}; # drop dotlock
                        }
-                       $lei->{1} = $out;
-               }
-               if ($l2m->lock_free) {
-                       $l2m->poke_dst;
-                       $lei->poke_mua;
-               } else { # mbox users
-                       delete $l2m->{mbl}; # drop dotlock
-                       $start_mua = 1;
                }
-       }
-       if ($lei->{-progress}) {
-               my $tot = $lei->{-mset_total} // 0;
-               my $nr = $lei->{-nr_write} // 0;
-               if ($l2m) {
-                       my $m = "# $nr written to " .
-                               "$lei->{ovv}->{dst} ($tot matches)";
-                       $nr ? $lei->qfin($m) : $lei->qerr($m);
-               } else {
-                       $lei->qerr("# $tot matches");
+               if ($lei->{-progress}) {
+                       my $tot = $lei->{-mset_total} // 0;
+                       my $nr_w = $lei->{-nr_write} // 0;
+                       my $d = ($lei->{-nr_seen} // 0) - $nr_w;
+                       my $x = "$tot matches";
+                       $x .= ", $d duplicates" if $d;
+                       if ($l2m) {
+                               my $m = "# $nr_w written to " .
+                                       "$lei->{ovv}->{dst} ($x)";
+                               $nr_w ? $lei->qfin($m) : $lei->qerr($m);
+                       } else {
+                               $lei->qerr("# $x");
+                       }
                }
-       }
-       $lei->start_mua if $start_mua;
-       $lei->dclose;
+               $lei->start_mua if $l2m && !$l2m->lock_free;
+               $lei->dclose;
+       };
+       $lei->fail($@) if $@;
 }
 
 sub do_post_augment {
        my ($lei) = @_;
        local $PublicInbox::LEI::current_lei = $lei;
        my $l2m = $lei->{l2m} or return; # client disconnected
-       $lei->fchdir or return;
-       my $err;
-       eval { $l2m->post_augment($lei) };
-       $err = $@;
+       eval {
+               $lei->fchdir;
+               $l2m->post_augment($lei);
+       };
+       my $err = $@;
        if ($err) {
                if (my $lxs = delete $lei->{lxs}) {
                        $lxs->wq_kill('-TERM');
-                       $lxs->wq_close(0, undef, $lei);
+                       $lxs->wq_close;
                }
                $lei->fail("$err");
        }
        if (!$err && delete $lei->{early_mua}) { # non-augment case
-               $lei->start_mua;
+               eval { $lei->start_mua };
+               $lei->fail($@) if $@;
        }
        close(delete $lei->{au_done}); # triggers wait_startq in lei_xsearch
 }
@@ -516,7 +527,7 @@ sub start_query ($$) { # always runs in main (lei-daemon) process
        if ($self->{-do_lcat}) {
                $self->wq_io_do('lcat_dump', []);
        }
-       $self->wq_close(1); # lei_xsearch workers stop when done
+       $self->wq_close; # lei_xsearch workers stop when done
 }
 
 sub incr_start_query { # called whenever an l2m shard starts do_post_auth
@@ -547,7 +558,7 @@ sub do_query {
                'child_error' => [ $lei ],
                'incr_start_query' => [ $self, $lei ],
        };
-       $lei->{auth}->op_merge($ops, $l2m) if $l2m && $lei->{auth};
+       $lei->{auth}->op_merge($ops, $l2m, $lei) if $l2m && $lei->{auth};
        my $end = $lei->pkt_op_pair;
        $lei->{1}->autoflush(1);
        $lei->start_pager if delete $lei->{need_pager};
@@ -571,12 +582,14 @@ sub do_query {
                }
                $l2m->wq_workers_start('lei2mail', undef,
                                        $lei->oldset, { lei => $lei });
+               $l2m->wq_wait_async(\&xsearch_done_wait, $lei);
                pipe($lei->{startq}, $lei->{au_done}) or die "pipe: $!";
                fcntl($lei->{startq}, $F_SETPIPE_SZ, 4096) if $F_SETPIPE_SZ;
                delete $l2m->{au_peers};
        }
        $self->wq_workers_start('lei_xsearch', undef,
                                $lei->oldset, { lei => $lei });
+       $self->wq_wait_async(\&xsearch_done_wait, $lei);
        my $op_c = delete $lei->{pkt_op_c};
        delete $lei->{pkt_op_p};
        @$end = ();