-- Conversations.
--
-- email_messages.thread_id has existed since the first migration and nothing
-- filled it, and in_reply_to was only ever copied, never written. A reply and
-- the message it answered were therefore unrelated rows, which is why the
-- webmail could neither show a conversation nor say whether it had been
-- answered.
--
-- The subject key lives in the database so delivery, the backfill below and
-- anything later share one definition of "the same subject": the prefixes mail
-- clients add in several languages are not part of it.
CREATE FUNCTION lm_subject_key(subject TEXT) RETURNS TEXT AS $$
    SELECT btrim(regexp_replace(
        lower(coalesce(subject, '')),
        '^(\s*(re|fw|fwd|enc|res|rv|aw|wg|tr|antw)(\[\d+\])?\s*:)+', ''))
$$ LANGUAGE SQL IMMUTABLE;

CREATE FUNCTION lm_has_reply_prefix(subject TEXT) RETURNS BOOLEAN AS $$
    SELECT lower(coalesce(subject, '')) ~ '^\s*(re|fw|fwd|enc|res|rv|aw|wg|tr|antw)(\[\d+\])?\s*:'
$$ LANGUAGE SQL IMMUTABLE;

-- Mail that arrived before replies carried their headers cannot be linked by
-- Message-ID, so it is grouped by subject, and only where a reply is actually
-- in the mailbox: two unrelated messages that merely share a subject stay apart.
-- Everything else keeps thread_id NULL, which the webmail reads as "a
-- conversation of its own" (COALESCE(thread_id, id)).
WITH groups AS (
    SELECT DISTINCT mailbox_id, lm_subject_key(subject) AS k
      FROM email_messages
     WHERE lm_has_reply_prefix(subject) AND lm_subject_key(subject) <> ''
)
UPDATE email_messages m
   SET thread_id = md5(m.mailbox_id::text || '|' || g.k)::uuid
  FROM groups g
 WHERE m.mailbox_id = g.mailbox_id
   AND lm_subject_key(m.subject) = g.k
   AND m.thread_id IS NULL;

-- The listing groups by this and filters by mailbox and folder.
CREATE INDEX idx_msg_mailbox_thread ON email_messages (mailbox_id, (COALESCE(thread_id, id)))
    WHERE expunged_at IS NULL;