lei: saved searches support --dedupe=<mid|oid>

This is less surprising in case users are used to using --dedupe=
without --save.

diff --git a/lib/PublicInbox/LeiSavedSearch.pm b/lib/PublicInbox/LeiSavedSearch.pm
index cd9effce43ac7cc9dc52034a6315031e48c4cf0e..ed217cf2f47d4cd840c5b441e8f680a2d7453f77 100644 (file)
--- a/lib/PublicInbox/LeiSavedSearch.pm
+++ b/lib/PublicInbox/LeiSavedSearch.pm
@@ -11,6 +11,7 @@ use PublicInbox::LeiSearch;
 use PublicInbox::Config;
 use PublicInbox::Spawn qw(run_die);
 use PublicInbox::ContentHash qw(git_sha);
+use PublicInbox::MID qw(mids_for_index);
 use Digest::SHA qw(sha256_hex);
 
 # move this to PublicInbox::Config if other things use it:
@@ -65,6 +66,14 @@ sub list {
        } @$out
 }
 
+sub translate_dedupe ($$$) {
+       my ($self, $lei, $dd) = @_;
+       $dd //= 'content';
+       return 1 if $dd eq 'content'; # the default
+       return $self->{"-dedupe_$dd"} = 1 if ($dd eq 'oid' || $dd eq 'mid');
+       $lei->fail("--dedupe=$dd unsupported with --save");
+}
+
 sub up { # updating existing saved search via "lei up"
        my ($cls, $lei, $dst) = @_;
        my $f;
@@ -89,6 +98,8 @@ sub new { # new saved search "lei q --save"
        File::Path::make_path($dir); # raises on error
        $self->{-cfg} = {};
        my $f = $self->{'-f'} = "$dir/lei.saved-search";
+       my $dd = $lei->{opt}->{dedupe};
+       translate_dedupe($self, $lei, $dd) or return;
        open my $fh, '>', $f or return $lei->fail("open $f: $!");
        my $sq_dst = PublicInbox::Config::squote_maybe($dst);
        my $q = $lei->{mset_opt}->{q_raw} // die 'BUG: {q_raw} missing';
@@ -105,6 +116,7 @@ sub new { # new saved search "lei q --save"
 [lei "q"]
        output = $dst
 EOM
+       print $fh "\tdedupe = $dd\n" if $dd;
        for my $k (ARRAY_FIELDS) {
                my $ary = $lei->{opt}->{$k} // next;
                for my $x (@$ary) {
@@ -134,14 +146,25 @@ sub is_dup {
        my ($self, $eml, $smsg) = @_;
        my $oidx = $self->{oidx} // die 'BUG: no {oidx}';
        my $blob = $smsg ? $smsg->{blob} : undef;
-       return 1 if $blob && $oidx->blob_exists($blob);
        my $lk = $self->lock_for_scope_fast;
+       return 1 if $blob && $oidx->blob_exists($blob);
+       if ($self->{-dedupe_mid}) {
+               for my $mid (@{mids_for_index($eml)}) {
+                       my ($id, $prv);
+                       return 1 if $oidx->next_by_mid($mid, \$id, \$prv);
+               }
+       }
        if (my $xoids = PublicInbox::LeiSearch::xoids_for($self, $eml, 1)) {
                for my $docid (values %$xoids) {
                        $oidx->add_xref3($docid, -1, $blob, '.');
                }
                $oidx->commit_lazy;
-               1;
+               if ($self->{-dedupe_oid}) {
+                       $smsg->{blob} //= git_sha(1, $eml)->hexdigest;
+                       exists $xoids->{$smsg->{blob}} ? 1 : undef;
+               } else {
+                       1;
+               }
        } else {
                # n.b. above xoids_for fills out eml->{-lei_fake_mid} if needed
                unless ($smsg) {

diff --git a/lib/PublicInbox/LeiUp.pm b/lib/PublicInbox/LeiUp.pm
index 0fb9698b0394c3b535d353e638ef7eb1ed300983..f4ff070be8bb925c0902035de32077223c55403e 100644 (file)
--- a/lib/PublicInbox/LeiUp.pm
+++ b/lib/PublicInbox/LeiUp.pm
@@ -25,6 +25,10 @@ sub up1 ($$) {
        my $o = $lei->{opt}->{output} = $lss->{-cfg}->{'lei.q.output'} //
                return $lei->fail("lei.q.output unset in $f");
        ref($o) and return $lei->fail("multiple values of lei.q.output in $f");
+       if (defined(my $dd = $lss->{-cfg}->{'lei.q.dedupe'})) {
+               $lss->translate_dedupe($lei, $dd) or return;
+               $lei->{opt}->{dedupe} = $dd;
+       }
        for my $k (qw(only include exclude)) {
                my $v = $lss->{-cfg}->get_all("lei.q.$k") // next;
                $lei->{opt}->{$k} = $v;

diff --git a/t/lei-q-save.t b/t/lei-q-save.t
index 5a2f7fff5f54612c2c177b3746bb31ba1338ba13..26ea5cb82859f204f6b1ac1d88490a72c527813c 100644 (file)
--- a/t/lei-q-save.t
+++ b/t/lei-q-save.t
@@ -121,5 +121,42 @@ test_lei(sub {
        unlike($lei_out, qr/mbrd-aug/,
                'forget-search completion cleared after forget');
        ok(!lei('up', "$home/mbrd-aug"), 'lei up fails after forget');
+
+       # dedupe=mid
+       my $o = "$home/dd-mid";
+       $in = $doc2->as_string . "\n-------\nappended list sig\n";
+       lei_ok [qw(import -q -F eml -)], undef, { 0 => \$in, %$lei_opt };
+       lei_ok(qw(q --dedupe=mid --save m:testmessage@example.com -o), $o);
+       my @m = glob("$o/cur/*");
+       is(scalar(@m), 1, '--dedupe=mid w/ --save');
+       $in = $doc2->as_string . "\n-------\nanother list sig\n";
+       lei_ok [qw(import -q -F eml -)], undef, { 0 => \$in, %$lei_opt };
+       lei_ok 'up', $o;
+       is_deeply([glob("$o/cur/*")], \@m, 'lei up dedupe=mid works');
+
+       for my $dd (qw(content)) {
+               $o = "$home/dd-$dd";
+               lei_ok(qw(q --save m:testmessage@example.com -o), $o,
+                               "--dedupe=$dd");
+               @m = glob("$o/cur/*");
+               is(scalar(@m), 3, 'all 3 matches with dedupe='.$dd);
+       }
+
+       # dedupe=oid
+       $o = "$home/dd-oid";
+       my $ibx = create_inbox 'ibx', indexlevel => 'medium',
+                       tmpdir => "$home/v1", sub {};
+       lei_ok(qw(q --save --dedupe=oid m:qp@example.com -o), $o,
+               '-I', $ibx->{inboxdir});
+       @m = glob("$o/cur/*");
+       is(scalar(@m), 1, 'got first result');
+
+       my $im = $ibx->importer(0);
+       my $diff = "X-Insignificant-Header: x\n".$doc1->as_string;
+       $im->add(PublicInbox::Eml->new($diff));
+       $im->done;
+       lei_ok('up', $o);
+       @m = glob("$o/cur/*");
+       is(scalar(@m), 2, 'got 2nd result due to different OID');
 });
 done_testing;