feat(3p/public-inbox): always set list-id when ingesting in watch r/5775

There is a code path in public-inbox that will (under certain
conditions) set the expected list-id as derived from the recipient
mail address (so in our case depot@tvl.su -> `List-ID: depot.tvl.su`).

However, when/how this triggers seems to be particular to the code
path taken based on certain config settings. I couldn't fully figure
it out, and to be honest I don't really know Perl, and I don't think
this warrants a super-high-effort investigation.

For that reason, this patches the appropriate line in
public-inbox-watch to always trigger this code path regardless of what
is going on with the email.

I tested this locally with a public-inbox config that does *not* have
a `listid` setting set, as that just adds an additional filter which
would be a no-op in this case. All emails are ingested correctly with
List-ID set.

There might be a better place to put this in (e.g. right before the
actual mail ingestion), if a Perl expert is interested in figuring
this out and considers it relevant, feel free to send a CL.

Note that this will not update old emails. Probably.

Change-Id: I4a8a42653aa2f408a85c9301a1ee3545f0e74eed
Reviewed-on: https://cl.tvl.fyi/c/depot/+/7946
Tested-by: BuildkiteCI
Autosubmit: tazjin <tazjin@tvl.su>
Reviewed-by: flokli <flokli@flokli.de>

2 files changed, 39 insertions, 0 deletions

diff --git a/third_party/public-inbox/0001-feat-always-set-the-List-ID-header-even-in-watch.patch b/third_party/public-inbox/0001-feat-always-set-the-List-ID-header-even-in-watch.patch
new file mode 100644
index 000000000000..bff2d4c200e2
--- /dev/null
+++ b/third_party/public-inbox/0001-feat-always-set-the-List-ID-header-even-in-watch.patch
@@ -0,0 +1,30 @@
+From 1719e904acf19499209b16a8a008f55390a7b5e2 Mon Sep 17 00:00:00 2001
+From: Vincent Ambo <mail@tazj.in>
+Date: Sun, 29 Jan 2023 13:36:12 +0300
+Subject: [PATCH] feat: always set the List-ID header even in -watch
+
+Without bothering to figure out exactly how this code path is usually
+triggered, always set a list ID when ingesting new emails in
+public-inbox-watch.
+---
+ lib/PublicInbox/Watch.pm | 4 ++++
+ 1 file changed, 4 insertions(+)
+
+diff --git a/lib/PublicInbox/Watch.pm b/lib/PublicInbox/Watch.pm
+index 3f6fe21..147971c 100644
+--- a/lib/PublicInbox/Watch.pm
++++ b/lib/PublicInbox/Watch.pm
+@@ -188,6 +188,10 @@ sub _remove_spam {
+ sub import_eml ($$$) {
+ 	my ($self, $ibx, $eml) = @_;
+ 
++        # TVL-specific: always set the list-id header, regardless of
++        # any of the other logic below.
++        PublicInbox::MDA->set_list_headers($eml, $ibx);
++
+ 	# any header match means it's eligible for the inbox:
+ 	if (my $watch_hdrs = $ibx->{-watchheaders}) {
+ 		my $ok;
+-- 
+2.39.0
+
diff --git a/third_party/public-inbox/default.nix b/third_party/public-inbox/default.nix
new file mode 100644
index 000000000000..1a4b196f94b5
--- /dev/null
+++ b/third_party/public-inbox/default.nix
@@ -0,0 +1,9 @@
+{ pkgs, ... }:
+
+pkgs.public-inbox.overrideAttrs (old: {
+  patches = (old.patches or [ ]) ++ [
+    ./0001-feat-always-set-the-List-ID-header-even-in-watch.patch
+  ];
+
+  doCheck = false; # too slow, and nixpkgs already runs them
+})