Added an option to skip duplicated recording MBID. Fixes #180

src/libs/services/database/impl/Track.cpp

@@ -186,12 +186,12 @@ Track::findPaths(Session& session, Range range)
 }
 
 RangeResults<TrackId>
-Track::findMBIDDuplicates(Session& session, Range range)
+Track::findRecordingMBIDDuplicates(Session& session, Range range)
 {
 	session.checkSharedLocked();
 
-	auto query {session.getDboSession().query<TrackId>( "SELECT track.id FROM track WHERE mbid in (SELECT mbid FROM track WHERE mbid <> '' GROUP BY mbid HAVING COUNT (*) > 1)")
-		.orderBy("track.release_id,track.disc_number,track.track_number,track.mbid")};
+	auto query {session.getDboSession().query<TrackId>( "SELECT track.id FROM track WHERE recording_mbid in (SELECT recording_mbid FROM track WHERE recording_mbid <> '' GROUP BY recording_mbid HAVING COUNT (*) > 1)")
+		.orderBy("track.release_id,track.disc_number,track.track_number,track.recording_mbid")};
 
 	return execQuery(query, range);
 }

src/libs/services/database/include/services/database/Track.hpp

@@ -92,7 +92,7 @@ class Track : public Object<Track, TrackId>
 		static RangeResults<TrackId>	find(Session& session, const FindParameters& parameters);
 		static RangeResults<TrackId>	findByNameAndReleaseName(Session& session, std::string_view trackName, std::string_view releaseName);
 		static RangeResults<PathResult>	findPaths(Session& session, Range range);
-		static RangeResults<TrackId>	findMBIDDuplicates(Session& session, Range range);
+		static RangeResults<TrackId>	findRecordingMBIDDuplicates(Session& session, Range range);
 		static RangeResults<TrackId>	findWithRecordingMBIDAndMissingFeatures(Session& session, Range range);
 
 		// Create utility

src/libs/services/scanner/impl/ScannerService.cpp

@@ -34,6 +34,7 @@
 #include "metadata/TagLibParser.hpp"
 #include "services/recommendation/IRecommendationService.hpp"
 #include "utils/Exception.hpp"
+#include "utils/IConfig.hpp"
 #include "utils/Logger.hpp"
 #include "utils/Path.hpp"
 #include "utils/UUID.hpp"
@@ -256,8 +257,11 @@ createScannerService(Db& db, Recommendation::IRecommendationService& recommendat
 
 ScannerService::ScannerService(Db& db, Recommendation::IRecommendationService& recommendationService)
 : _recommendationService {recommendationService}
+, _skipDuplicateRecordingMBID {Service<IConfig>::get()->getBool("scanner-skip-duplicate-recording-mbid", false)}
 , _dbSession {db}
 {
+	LMS_LOG(DBUPDATER, INFO) << "skipDuplicateRecordingMBID = " << _skipDuplicateRecordingMBID;
+
 	// For now, always use TagLib
 	_metadataParser = std::make_unique<MetaData::TagLibParser>();
 
@@ -719,6 +723,25 @@ ScannerService::scanAudioFile(const std::filesystem::path& file, bool forceScan,
 
 	Track::pointer track {Track::findByPath(_dbSession, file) };
 
+	// Skip duplicate recording MBID
+	if (trackInfo->recordingMBID && _skipDuplicateRecordingMBID)
+	{
+		for (Track::pointer otherTrack : Track::findByRecordingMBID(_dbSession, *trackInfo->recordingMBID))
+		{
+			if (track && track->getId() == otherTrack->getId())
+				continue;
+
+			LMS_LOG(DBUPDATER, DEBUG) << "Skipped '" << file.string() << "' (similar recording MBID in '" << otherTrack->getPath().string() << "')";
+			// This recording MBID already exists, just remove what we just scanned
+			if (track)
+			{
+				track.remove();
+				stats.deletions++;
+			}
+			return;
+		}
+	}
+
 	// We estimate this is an audio file if:
 	// - we found a least one audio stream
 	// - the duration is not null
@@ -1026,14 +1049,14 @@ ScannerService::checkDuplicatedAudioFiles(ScanStats& stats)
 
 	auto transaction {_dbSession.createSharedTransaction()};
 
-	const RangeResults<TrackId> tracks = Track::findMBIDDuplicates(_dbSession, Range {});
+	const RangeResults<TrackId> tracks = Track::findRecordingMBIDDuplicates(_dbSession, Range {});
 	for (const TrackId trackId : tracks.results)
 	{
 		const Track::pointer track {Track::find(_dbSession, trackId)};
-		if (auto trackMBID {track->getTrackMBID()})
+		if (auto recordingMBID {track->getRecordingMBID()})
 		{
-			LMS_LOG(DBUPDATER, INFO) << "Found duplicated Track MBID [" << trackMBID->getAsString() << "], file: " << track->getPath().string() << " - " << track->getName();
-			stats.duplicates.emplace_back(ScanDuplicate {track->getId(), DuplicateReason::SameMBID});
+			LMS_LOG(DBUPDATER, INFO) << "Found duplicated recording MBID [" << recordingMBID->getAsString() << "], file: " << track->getPath().string() << " - " << track->getName();
+			stats.duplicates.emplace_back(ScanDuplicate {track->getId(), DuplicateReason::SameRecordingMBID});
 		}
 	}
 

src/libs/services/scanner/impl/ScannerService.hpp

@@ -98,6 +98,7 @@ namespace Scanner
 			std::atomic<bool>						_abortScan {};
 			Wt::WIOService							_ioService;
 			boost::asio::system_timer				_scheduleTimer {_ioService};
+			const bool								_skipDuplicateRecordingMBID {};
 			Events									_events;
 			std::chrono::system_clock::time_point	_lastScanInProgressEmit {};
 			Database::Session						_dbSession;

src/libs/services/scanner/include/services/scanner/ScannerStats.hpp

@@ -30,16 +30,16 @@ namespace Scanner
 {
 	enum class ScanErrorType
 	{
-		CannotReadFile,		// cannot read file
-		CannotParseFile,	// cannot parse file
-		NoAudioTrack,		// no audio track found
-		BadDuration,		// bad duration
+		CannotReadFile,			// cannot read file
+		CannotParseFile,		// cannot parse file
+		NoAudioTrack,			// no audio track found
+		BadDuration,			// bad duration
 	};
 
 	enum class DuplicateReason
 	{
 		SameHash,
-		SameMBID,
+		SameRecordingMBID,
 	};
 
 	struct ScanError

src/lms/ui/admin/ScannerController.cpp

@@ -96,7 +96,7 @@ class ReportResource : public Wt::WResource
 
 					response.out() << track->getPath().string();
 					if (auto mbid {track->getTrackMBID()})
-						response.out() << " (Track MBID " << mbid->getAsString() << ")";
+						response.out() << " (Recording MBID " << mbid->getAsString() << ")";
 
 					response.out() << " - " << duplicateReasonToWString(duplicate.reason).toUTF8() << '\n';
 				}
@@ -122,7 +122,7 @@ class ReportResource : public Wt::WResource
 			switch (reason)
 			{
 				case Scanner::DuplicateReason::SameHash: return Wt::WString::tr("Lms.Admin.ScannerController.same-hash");
-				case Scanner::DuplicateReason::SameMBID: return Wt::WString::tr("Lms.Admin.ScannerController.same-mbid");
+				case Scanner::DuplicateReason::SameRecordingMBID: return Wt::WString::tr("Lms.Admin.ScannerController.same-mbid");
 			}
 			return "?";
 		}