Started to restore clusters based recommendation engine

This commit is contained in:
emeric
2020-02-15 15:52:39 +01:00
parent 7f05a1d4e6
commit d1bdaf6203
26 changed files with 326 additions and 177 deletions
+15
View File
@@ -104,6 +104,21 @@ Artist::getAllOrphans(Session& session)
return std::vector<pointer>(res.begin(), res.end());
}
std::vector<IdType>
Artist::getAllIdsWithClusters(Session& session, std::optional<std::size_t> limit)
{
session.checkSharedLocked();
Wt::Dbo::collection<IdType> res = session.getDboSession().query<IdType>
("SELECT DISTINCT a.id FROM artist a"
" INNER JOIN track t ON t.id = t_a_l.track_id INNER JOIN track_artist_link t_a_l ON t_a_l.artist_id = a.id"
" INNER JOIN track_cluster t_c ON t_c.track_id = t.id")
.limit(limit ? static_cast<int>(*limit) : -1);
return std::vector<IdType>(res.begin(), res.end());
}
static
Wt::Dbo::Query<Artist::pointer>
getQuery(Session& session,
+15
View File
@@ -252,6 +252,21 @@ Release::getByFilter(Session& session,
return res;
}
std::vector<IdType>
Release::getAllIdsWithClusters(Session& session, std::optional<std::size_t> limit)
{
session.checkSharedLocked();
Wt::Dbo::collection<IdType> res = session.getDboSession().query<IdType>
("SELECT DISTINCT r.id 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")
.limit(limit ? static_cast<int>(*limit) : -1);
return std::vector<IdType>(res.begin(), res.end());
}
std::optional<std::size_t>
Release::getTotalTrackNumber(void) const
{
+1 -1
View File
@@ -118,7 +118,7 @@ Session::doDatabaseMigrationIfNeeded()
{
_session.execute("DROP TABLE similarity_settings");
_session.execute("DROP TABLE similarity_settings_feature");
_session.execute("ALTER TABLE scan_settings ADD similarity_engine_type INTEGER NOT NULL DEFAULT(" + std::to_string(static_cast<int>(ScanSettings::SimilarityEngineType::Clusters)) + ")");
_session.execute("ALTER TABLE scan_settings ADD similarity_engine_type INTEGER NOT NULL DEFAULT(" + std::to_string(static_cast<int>(ScanSettings::RecommendationEngineType::Clusters)) + ")");
}
else if (version == 8)
{
+14 -1
View File
@@ -163,6 +163,19 @@ Track::getAllIdsWithFeatures(Session& session, std::optional<std::size_t> limit)
return std::vector<IdType>(res.begin(), res.end());
}
std::vector<IdType>
Track::getAllIdsWithClusters(Session& session, std::optional<std::size_t> limit)
{
session.checkSharedLocked();
Wt::Dbo::collection<IdType> res = session.getDboSession().query<IdType>
("SELECT DISTINCT t.id FROM track t"
" INNER JOIN track_cluster t_c ON t_c.track_id = t.id")
.limit(limit ? static_cast<int>(*limit) : -1);
return std::vector<IdType>(res.begin(), res.end());
}
std::vector<Cluster::pointer>
Track::getClusters(void) const
{
@@ -263,7 +276,7 @@ Track::getByFilter(Session& session,
std::vector<Track::pointer>
Track::getSimilarTracks(Session& session,
const std::set<IdType>& tracks,
const std::unordered_set<IdType>& tracks,
std::optional<std::size_t> offset,
std::optional<std::size_t> size)
{
@@ -68,6 +68,7 @@ class Artist : public Wt::Dbo::Dbo<Artist>
static std::vector<IdType> getAllIds(Session& session);
static std::vector<pointer> getAllOrphans(Session& session); // No track related
static std::vector<pointer> getLastAdded(Session& session, Wt::WDateTime after, std::optional<std::size_t> size = {});
static std::vector<IdType> getAllIdsWithClusters(Session& session, std::optional<std::size_t> limit = {});
// Accessors
const std::string& getName(void) const { return _name; }
@@ -66,6 +66,7 @@ class Release : public Wt::Dbo::Dbo<Release>
std::optional<std::size_t> offset,
std::optional<std::size_t> size,
bool& moreExpected);
static std::vector<IdType> getAllIdsWithClusters(Session& session, std::optional<std::size_t> limit = {});
std::vector<Wt::Dbo::ptr<Track>> getTracks(const std::set<IdType>& clusters = std::set<IdType>()) const;
std::size_t getTracksCount() const;
@@ -43,7 +43,7 @@ class ScanSettings : public Wt::Dbo::Dbo<ScanSettings>
};
// Do not modify values (just add)
enum class SimilarityEngineType
enum class RecommendationEngineType
{
Clusters = 0,
Features,
@@ -60,7 +60,7 @@ class ScanSettings : public Wt::Dbo::Dbo<ScanSettings>
UpdatePeriod getUpdatePeriod() const { return _updatePeriod; }
std::vector<Wt::Dbo::ptr<ClusterType>> getClusterTypes() const;
std::set<std::filesystem::path> getAudioFileExtensions() const;
SimilarityEngineType getSimilarityEngineType() const { return _similarityEngineType; }
RecommendationEngineType getRecommendationEngineType() const { return _recommendationEngineType; }
// Setters
void addAudioFileExtension(const std::filesystem::path& ext);
@@ -68,7 +68,7 @@ class ScanSettings : public Wt::Dbo::Dbo<ScanSettings>
void setUpdateStartTime(Wt::WTime t) { _startTime = t; }
void setUpdatePeriod(UpdatePeriod p) { _updatePeriod = p; }
void setClusterTypes(Session& session, const std::set<std::string>& clusterTypeNames);
void setSimilarityEngineType(SimilarityEngineType type) { _similarityEngineType = type; }
void setRecommendationEngineType(RecommendationEngineType type) { _recommendationEngineType = type; }
void incScanVersion();
template<class Action>
@@ -79,7 +79,7 @@ class ScanSettings : public Wt::Dbo::Dbo<ScanSettings>
Wt::Dbo::field(a, _startTime, "start_time");
Wt::Dbo::field(a, _updatePeriod, "update_period");
Wt::Dbo::field(a, _audioFileExtensions, "audio_file_extensions");
Wt::Dbo::field(a, _similarityEngineType,"similarity_engine_type");
Wt::Dbo::field(a, _recommendationEngineType,"similarity_engine_type");
Wt::Dbo::hasMany(a, _clusterTypes, Wt::Dbo::ManyToOne, "scan_settings");
}
@@ -89,7 +89,7 @@ class ScanSettings : public Wt::Dbo::Dbo<ScanSettings>
std::string _mediaDirectory;
Wt::WTime _startTime = Wt::WTime {0,0,0};
UpdatePeriod _updatePeriod {UpdatePeriod::Never};
SimilarityEngineType _similarityEngineType {SimilarityEngineType::Clusters};
RecommendationEngineType _recommendationEngineType {RecommendationEngineType::Clusters};
std::string _audioFileExtensions {".alac .mp3 .ogg .oga .aac .m4a .m4b .flac .wav .wma .aif .aiff .ape .mpc .shn .opus"};
Wt::Dbo::collection<Wt::Dbo::ptr<ClusterType>> _clusterTypes;
};
+4 -2
View File
@@ -22,8 +22,9 @@
#include <chrono>
#include <filesystem>
#include <optional>
#include <vector>
#include <string>
#include <unordered_set>
#include <vector>
#include <Wt/Dbo/Dbo.h>
#include <Wt/WDateTime.h>
@@ -58,7 +59,7 @@ class Track : public Wt::Dbo::Dbo<Track>
static pointer getById(Session& session, IdType id);
static pointer getByMBID(Session& session, const UUID& MBID);
static std::vector<pointer> getSimilarTracks(Session& session,
const std::set<IdType>& trackIds,
const std::unordered_set<IdType>& trackIds,
std::optional<std::size_t> offset = {},
std::optional<std::size_t> size = {});
static std::vector<pointer> getByClusters(Session& session,
@@ -78,6 +79,7 @@ class Track : public Wt::Dbo::Dbo<Track>
static std::vector<pointer> getLastAdded(Session& session, const Wt::WDateTime& after, std::optional<std::size_t> size = 1);
static std::vector<pointer> getAllWithMBIDAndMissingFeatures(Session& session);
static std::vector<IdType> getAllIdsWithFeatures(Session& session, std::optional<std::size_t> limit = {});
static std::vector<IdType> getAllIdsWithClusters(Session& session, std::optional<std::size_t> limit = {});
// Create utility
static pointer create(Session& session, const std::filesystem::path& p);