]> Sergey Matveev's repositories - public-inbox.git/blobdiff - lib/PublicInbox/Import.pm
lei index: new command to index mail w/o git storage
[public-inbox.git] / lib / PublicInbox / Import.pm
index 3adf9dec21785ee9865a22bcdd8e7a338051dcac..362cdc47375db4b7947c41747c6d15bf91583d7d 100644 (file)
@@ -413,19 +413,19 @@ sub add {
                $smsg->{blob} = $self->get_mark(":$blob");
                $smsg->set_bytes($raw_email, $n);
                if (my $oidx = delete $smsg->{-oidx}) { # used by LeiStore
-                       my @docids = $oidx->blob_exists($smsg->{blob});
-                       my @vivify_xvmd;
-                       for my $id (@docids) {
-                               if (my $cur = $oidx->get_art($id)) {
-                                       # already imported if bytes > 0
-                                       return if $cur->{bytes} > 0;
-                                       push @vivify_xvmd, $id;
-                               } else {
-                                       warn "W: $smsg->{blob} ",
-                                               "#$id gone (bug?)\n";
-                               }
-                       }
-                       $smsg->{-vivify_xvmd} = \@vivify_xvmd;
+                       my $eidx_git = delete $smsg->{-eidx_git};
+
+                       # we need this sharedkv to dedupe blobs added in the
+                       # same fast-import transaction
+                       my $u = $self->{uniq_skv} //= do {
+                               require PublicInbox::SharedKV;
+                               my $x = PublicInbox::SharedKV->new;
+                               $x->dbh;
+                               $x;
+                       };
+                       return if !$u->set_maybe(pack('H*', $smsg->{blob}), 1);
+                       return if (!$oidx->vivify_xvmd($smsg) &&
+                                       $eidx_git->check($smsg->{blob}));
                }
        }
        my $ref = $self->{ref};