From: Eric Wong <e@80x24.org>
Date: Mon, 12 May 2025 20:45:01 +0000 (+0000)
Subject: www: extmsg: dedupe cross-posted Message-IDs
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=df0579765d8481607fcf580822f60b0224469ccb;p=thirdparty%2Fpublic-inbox.git

www: extmsg: dedupe cross-posted Message-IDs

Having redundant links to the same cross-posted message is
unlikely to be of help, especially when using an /all/ extindex.
So just grab the first matching inbox + Message-ID combo and
ignore subsequent URLs for Message-IDs which exist in multiple
inboxes/extindices.
---

diff --git a/lib/PublicInbox/ExtMsg.pm b/lib/PublicInbox/ExtMsg.pm
index 7453df1ef..9982e6729 100644
--- a/lib/PublicInbox/ExtMsg.pm
+++ b/lib/PublicInbox/ExtMsg.pm
@@ -36,10 +36,14 @@ sub partial_cb { # async_mset cb
 		++$ctx->{ext_msg_partial_fail};
 		warn($msg);
 	} else {
-		my $ibx = $ctx->{partial_ibx};
-		my @mid = map { $_->{mid} } @{$srch->mset_to_smsg($ibx, $mset)};
+		my $seen = $ctx->{partial_seen} //= {};
+		my (@mid, $mid);
+		for (@{$srch->mset_to_smsg($ctx->{partial_ibx}, $mset)}) {
+			$mid = $_->{mid};
+			$seen->{$mid} //= push @mid, $mid;
+		}
 		if (scalar @mid) {
-			push @{$ctx->{partial}}, [ $ibx, \@mid ];
+			push @{$ctx->{partial}}, [ $ctx->{partial_ibx}, \@mid ];
 			(($ctx->{n_partial} += scalar(@mid)) >= PARTIAL_MAX) and
 				delete $ctx->{again}; # done
 		}