From 2910be8a9a85ecf343d7594b6170c6edb724cdc4 Mon Sep 17 00:00:00 2001
From: emeric <emeric.poupon@laposte.net>
Date: Wed, 1 Nov 2023 23:55:24 +0100
Subject: [PATCH] Similarity by clusters: removed duplicates in the list of
 clusters to search

---
 src/libs/services/database/impl/Artist.cpp    | 2 +-
 src/libs/services/database/impl/Release.cpp   | 8 +++++++-
 src/libs/services/database/impl/Track.cpp     | 2 +-
 src/libs/services/database/impl/TrackList.cpp | 2 +-
 4 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/src/libs/services/database/impl/Artist.cpp b/src/libs/services/database/impl/Artist.cpp
index a1b33294..a5e9e6fc 100644
--- a/src/libs/services/database/impl/Artist.cpp
+++ b/src/libs/services/database/impl/Artist.cpp
@@ -223,7 +223,7 @@ Artist::findSimilarArtists(EnumSet<TrackArtistLinkType> artistLinkTypes, Range r
 			" INNER JOIN track t ON t.id = t_a_l.track_id"
 			" INNER JOIN track_cluster t_c ON t_c.track_id = t.id"
 				" WHERE "
-					" t_c.cluster_id IN (SELECT c.id from cluster c"
+					" t_c.cluster_id IN (SELECT DISTINCT c.id from cluster c"
 											" INNER JOIN track t ON c.id = t_c.cluster_id"
 											" INNER JOIN track_cluster t_c ON t_c.track_id = t.id"
 											" INNER JOIN artist a ON a.id = t_a_l.artist_id"
diff --git a/src/libs/services/database/impl/Release.cpp b/src/libs/services/database/impl/Release.cpp
index a04bf9ab..fbd25c59 100644
--- a/src/libs/services/database/impl/Release.cpp
+++ b/src/libs/services/database/impl/Release.cpp
@@ -382,12 +382,18 @@ namespace Database
     {
         assert(session());
 
+        // Select the similar releases using the 5 most used clusters of the release
         auto res{ session()->query<Wt::Dbo::ptr<Release>>(
                 "SELECT r FROM release r"
                 " INNER JOIN track t ON t.release_id = r.id"
                 " INNER JOIN track_cluster t_c ON t_c.track_id = t.id"
                     " WHERE "
-                        " t_c.cluster_id IN (SELECT c.id from cluster c INNER JOIN track t ON c.id = t_c.cluster_id INNER JOIN track_cluster t_c ON t_c.track_id = t.id INNER JOIN release r ON r.id = t.release_id WHERE r.id = ?)"
+                        " t_c.cluster_id IN "
+                        "(SELECT DISTINCT c.id FROM cluster c"
+                         " INNER JOIN track t ON c.id = t_c.cluster_id"
+                         " INNER JOIN track_cluster t_c ON t_c.track_id = t.id"
+                         " INNER JOIN release r ON r.id = t.release_id"
+                         " WHERE r.id = ?)"
                         " AND r.id <> ?"
                     )
             .bind(getId())
diff --git a/src/libs/services/database/impl/Track.cpp b/src/libs/services/database/impl/Track.cpp
index 5f3933e5..4d2b8612 100644
--- a/src/libs/services/database/impl/Track.cpp
+++ b/src/libs/services/database/impl/Track.cpp
@@ -327,7 +327,7 @@ Track::findSimilarTracks(Session& session, const std::vector<TrackId>& tracks, R
 	auto query {session.getDboSession().query<TrackId>(
 			"SELECT t.id FROM track t"
 			" INNER JOIN track_cluster t_c ON t_c.track_id = t.id"
-					" AND t_c.cluster_id IN (SELECT c.id FROM cluster c INNER JOIN track_cluster t_c ON t_c.cluster_id = c.id WHERE t_c.track_id IN (" + oss.str() + "))"
+					" AND t_c.cluster_id IN (SELECT DISTINCT c.id FROM cluster c INNER JOIN track_cluster t_c ON t_c.cluster_id = c.id WHERE t_c.track_id IN (" + oss.str() + "))"
 					" AND t.id NOT IN (" + oss.str() + ")")
 		.groupBy("t.id")
 		.orderBy("COUNT(*) DESC, RANDOM()")};
diff --git a/src/libs/services/database/impl/TrackList.cpp b/src/libs/services/database/impl/TrackList.cpp
index 227eaa4c..490fc06a 100644
--- a/src/libs/services/database/impl/TrackList.cpp
+++ b/src/libs/services/database/impl/TrackList.cpp
@@ -538,7 +538,7 @@ TrackList::getSimilarTracks(std::optional<std::size_t> offset, std::optional<std
 			"SELECT t FROM track t"
 			" INNER JOIN track_cluster t_c ON t_c.track_id = t.id"
 				" WHERE "
-					" (t_c.cluster_id IN (SELECT c.id from cluster c INNER JOIN track t ON c.id = t_c.cluster_id INNER JOIN track_cluster t_c ON t_c.track_id = t.id INNER JOIN tracklist_entry p_e ON p_e.track_id = t.id INNER JOIN tracklist p ON p.id = p_e.tracklist_id WHERE p.id = ?)"
+					" (t_c.cluster_id IN (SELECT DISTINCT c.id from cluster c INNER JOIN track t ON c.id = t_c.cluster_id INNER JOIN track_cluster t_c ON t_c.track_id = t.id INNER JOIN tracklist_entry p_e ON p_e.track_id = t.id INNER JOIN tracklist p ON p.id = p_e.tracklist_id WHERE p.id = ?)"
 					" AND t.id NOT IN (SELECT tracklist_t.id FROM track tracklist_t INNER JOIN tracklist_entry t_e ON t_e.track_id = tracklist_t.id WHERE t_e.tracklist_id = ?))"
 				)
 		.bind(getId())