/*
* Copyright (C) 2023 Emeric Poupon
*
* This file is part of LMS.
*
* LMS is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* LMS is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with LMS. If not, see .
*/
#include "ScanStepScanFiles.hpp"
#include "core/Exception.hpp"
#include "core/IConfig.hpp"
#include "core/ILogger.hpp"
#include "core/ITraceLogger.hpp"
#include "core/Path.hpp"
#include "database/Artist.hpp"
#include "database/Cluster.hpp"
#include "database/Db.hpp"
#include "database/Directory.hpp"
#include "database/Image.hpp"
#include "database/MediaLibrary.hpp"
#include "database/Release.hpp"
#include "database/Session.hpp"
#include "database/Track.hpp"
#include "database/TrackArtistLink.hpp"
#include "database/TrackFeatures.hpp"
#include "metadata/Exception.hpp"
#include "metadata/IParser.hpp"
namespace lms::scanner
{
using namespace db;
namespace
{
struct FileInfo
{
Wt::WDateTime lastWriteTime;
std::filesystem::path relativePath;
std::size_t fileSize{};
};
Wt::WDateTime retrieveFileGetLastWrite(const std::filesystem::path& file)
{
Wt::WDateTime res;
try
{
res = core::pathUtils::getLastWriteTime(file);
}
catch (core::LmsException& e)
{
LMS_LOG(DBUPDATER, ERROR, "Cannot get last write time: " << e.what());
}
return res;
}
std::optional retrieveFileInfo(const std::filesystem::path& file, const std::filesystem::path& rootPath)
{
std::optional res;
res.emplace();
res->lastWriteTime = retrieveFileGetLastWrite(file);
if (!res->lastWriteTime.isValid())
{
res.reset();
return res;
}
{
std::error_code ec;
res->relativePath = std::filesystem::relative(file, rootPath, ec);
if (ec)
{
LMS_LOG(DBUPDATER, ERROR, "Cannot get relative file path for '" << file.string() << "' from '" << rootPath.string() << "': " << ec.message());
res.reset();
return res;
}
}
{
std::error_code ec;
res->fileSize = std::filesystem::file_size(file, ec);
if (ec)
{
LMS_LOG(DBUPDATER, ERROR, "Cannot get file size for '" << file.string() << "': " << ec.message());
res.reset();
return res;
}
}
return res;
}
Directory::pointer getOrCreateDirectory(Session& session, const std::filesystem::path& path, const std::filesystem::path& rootPath)
{
Directory::pointer directory{ Directory::find(session, path) };
if (!directory)
{
Directory::pointer parentDirectory;
if (path != rootPath)
parentDirectory = getOrCreateDirectory(session, path.parent_path(), rootPath);
directory = session.create(path);
directory.modify()->setParent(parentDirectory);
}
return directory;
}
Artist::pointer createArtist(Session& session, const metadata::Artist& artistInfo)
{
Artist::pointer artist{ session.create(artistInfo.name) };
if (artistInfo.mbid)
artist.modify()->setMBID(*artistInfo.mbid);
if (artistInfo.sortName)
artist.modify()->setSortName(*artistInfo.sortName);
return artist;
}
void updateArtistIfNeeded(Artist::pointer artist, const metadata::Artist& artistInfo)
{
// Name may have been updated
if (artist->getName() != artistInfo.name)
{
artist.modify()->setName(artistInfo.name);
}
// Sortname may have been updated
if (artistInfo.sortName && *artistInfo.sortName != artist->getSortName())
{
artist.modify()->setSortName(*artistInfo.sortName);
}
}
std::vector getOrCreateArtists(Session& session, const std::vector& artistsInfo, bool allowFallbackOnMBIDEntries)
{
std::vector artists;
for (const metadata::Artist& artistInfo : artistsInfo)
{
Artist::pointer artist;
// First try to get by MBID
if (artistInfo.mbid)
{
artist = Artist::find(session, *artistInfo.mbid);
if (!artist)
artist = createArtist(session, artistInfo);
else
updateArtistIfNeeded(artist, artistInfo);
artists.emplace_back(std::move(artist));
continue;
}
// Fall back on artist name (collisions may occur)
if (!artistInfo.name.empty())
{
for (const Artist::pointer& sameNamedArtist : Artist::find(session, artistInfo.name))
{
// Do not fallback on artist that is correctly tagged
if (!allowFallbackOnMBIDEntries && sameNamedArtist->getMBID())
continue;
artist = sameNamedArtist;
break;
}
// No Artist found with the same name and without MBID -> creating
if (!artist)
artist = createArtist(session, artistInfo);
else
updateArtistIfNeeded(artist, artistInfo);
artists.emplace_back(std::move(artist));
continue;
}
}
return artists;
}
ReleaseType::pointer getOrCreateReleaseType(Session& session, std::string_view name)
{
ReleaseType::pointer releaseType{ ReleaseType::find(session, name) };
if (!releaseType)
releaseType = session.create(name);
return releaseType;
}
void updateReleaseIfNeeded(Session& session, Release::pointer release, const metadata::Release& releaseInfo)
{
if (release->getName() != releaseInfo.name)
release.modify()->setName(releaseInfo.name);
if (release->getSortName() != releaseInfo.sortName)
release.modify()->setSortName(releaseInfo.sortName);
if (release->getGroupMBID() != releaseInfo.groupMBID)
release.modify()->setGroupMBID(releaseInfo.groupMBID);
if (release->getTotalDisc() != releaseInfo.mediumCount)
release.modify()->setTotalDisc(releaseInfo.mediumCount);
if (release->getArtistDisplayName() != releaseInfo.artistDisplayName)
release.modify()->setArtistDisplayName(releaseInfo.artistDisplayName);
if (release->getReleaseTypeNames() != releaseInfo.releaseTypes)
{
release.modify()->clearReleaseTypes();
for (std::string_view releaseType : releaseInfo.releaseTypes)
release.modify()->addReleaseType(getOrCreateReleaseType(session, releaseType));
}
}
Release::pointer getOrCreateRelease(Session& session, const metadata::Release& releaseInfo, const std::filesystem::path& expectedReleaseDirectory)
{
Release::pointer release;
// First try to get by MBID
if (releaseInfo.mbid)
{
release = Release::find(session, *releaseInfo.mbid);
if (!release)
release = session.create(releaseInfo.name, releaseInfo.mbid);
updateReleaseIfNeeded(session, release, releaseInfo);
return release;
}
// Fall back on release name (collisions may occur), if and only if it is in the current directory
if (!releaseInfo.name.empty())
{
for (const Release::pointer& sameNamedRelease : Release::find(session, releaseInfo.name, expectedReleaseDirectory))
{
// do not fallback on properly tagged releases
if (sameNamedRelease->getMBID())
continue;
release = sameNamedRelease;
break;
}
// No release found with the same name and without MBID -> creating
if (!release)
release = session.create(releaseInfo.name);
updateReleaseIfNeeded(session, release, releaseInfo);
return release;
}
return Release::pointer{};
}
std::vector getOrCreateClusters(Session& session, const metadata::Track& track)
{
std::vector clusters;
auto getOrCreateClusters{ [&](std::string tag, std::span values) {
auto clusterType = ClusterType::find(session, tag);
if (!clusterType)
clusterType = session.create(tag);
for (const auto& value : values)
{
auto cluster{ clusterType->getCluster(value) };
if (!cluster)
cluster = session.create(clusterType, value);
clusters.push_back(cluster);
}
} };
// TODO: migrate these fields in dedicated tables in DB
getOrCreateClusters("GENRE", track.genres);
getOrCreateClusters("MOOD", track.moods);
getOrCreateClusters("LANGUAGE", track.languages);
getOrCreateClusters("GROUPING", track.groupings);
for (const auto& [tag, values] : track.userExtraTags)
getOrCreateClusters(tag, values);
return clusters;
}
metadata::ParserReadStyle getParserReadStyle()
{
std::string_view readStyle{ core::Service::get()->getString("scanner-parser-read-style", "average") };
if (readStyle == "fast")
return metadata::ParserReadStyle::Fast;
else if (readStyle == "average")
return metadata::ParserReadStyle::Average;
else if (readStyle == "accurate")
return metadata::ParserReadStyle::Accurate;
throw core::LmsException{ "Invalid value for 'scanner-parser-read-style'" };
}
std::size_t getScanMetaDataThreadCount()
{
std::size_t threadCount{ core::Service::get()->getULong("scanner-metadata-thread-count", 0) };
if (threadCount == 0)
threadCount = std::max(std::thread::hardware_concurrency() / 2, 1);
return threadCount;
}
} // namespace
ScanStepScanFiles::ScanStepScanFiles(InitParams& initParams)
: ScanStepBase{ initParams }
, _metadataParser{ metadata::createParser(metadata::ParserBackend::TagLib, getParserReadStyle()) } // For now, always use TagLib
, _fileScanQueue{ *_metadataParser, getScanMetaDataThreadCount(), _abortScan }
{
LMS_LOG(DBUPDATER, INFO, "Using " << _fileScanQueue.getThreadCount() << " thread(s) for scanning file metadata");
}
void ScanStepScanFiles::process(ScanContext& context)
{
{
std::vector tagsToParse{ _extraTagsToParse };
tagsToParse.insert(std::end(tagsToParse), std::cbegin(_settings.extraTags), std::cend(_settings.extraTags));
_metadataParser->setUserExtraTags(tagsToParse);
_metadataParser->setArtistTagDelimiters(_settings.artistTagDelimiters);
_metadataParser->setDefaultTagDelimiters(_settings.defaultTagDelimiters);
}
context.currentStepStats.totalElems = context.stats.totalFileCount;
for (const ScannerSettings::MediaLibraryInfo& mediaLibrary : _settings.mediaLibraries)
process(context, mediaLibrary);
}
void ScanStepScanFiles::process(ScanContext& context, const ScannerSettings::MediaLibraryInfo& mediaLibrary)
{
const std::size_t scanQueueMaxScanRequestCount{ 100 * _fileScanQueue.getThreadCount() };
const std::size_t processFileResultsBatchSize{ 5 };
std::vector scanResults;
core::pathUtils::exploreFilesRecursive(
mediaLibrary.rootDirectory, [&](std::error_code ec, const std::filesystem::path& path) {
LMS_SCOPED_TRACE_DETAILED("Scanner", "OnExploreFile");
if (_abortScan)
return false;
if (ec)
{
LMS_LOG(DBUPDATER, ERROR, "Cannot scan file '" << path.string() << "': " << ec.message());
context.stats.errors.emplace_back(ScanError{ path, ScanErrorType::CannotReadFile, ec.message() });
}
else
{
bool fileToProcess{};
if (core::pathUtils::hasFileAnyExtension(path, _settings.supportedAudioFileExtensions))
{
fileToProcess = true;
if (checkAudioFileNeedScan(context, path, mediaLibrary))
_fileScanQueue.pushScanRequest(path, FileScanQueue::ScanRequestType::AudioFile);
}
else if (core::pathUtils::hasFileAnyExtension(path, _settings.supportedImageFileExtensions))
{
fileToProcess = true;
if (checkImageFileNeedScan(context, path))
_fileScanQueue.pushScanRequest(path, FileScanQueue::ScanRequestType::ImageFile);
}
if (fileToProcess)
{
context.currentStepStats.processedElems++;
_progressCallback(context.currentStepStats);
}
}
while (_fileScanQueue.getResultsCount() > (scanQueueMaxScanRequestCount / 2))
{
_fileScanQueue.popResults(scanResults, processFileResultsBatchSize);
processFileScanResults(context, scanResults, mediaLibrary);
}
_fileScanQueue.wait(scanQueueMaxScanRequestCount);
return true;
},
&excludeDirFileName);
_fileScanQueue.wait();
while (!_abortScan && _fileScanQueue.popResults(scanResults, processFileResultsBatchSize) > 0)
processFileScanResults(context, scanResults, mediaLibrary);
}
bool ScanStepScanFiles::checkAudioFileNeedScan(ScanContext& context, const std::filesystem::path& file, const ScannerSettings::MediaLibraryInfo& libraryInfo)
{
ScanStats& stats{ context.stats };
const Wt::WDateTime lastWriteTime{ retrieveFileGetLastWrite(file) };
// Should rarely fail as we are currently iterating it
if (!lastWriteTime.isValid())
{
stats.skips++;
return false;
}
bool needUpdateLibrary{};
if (!context.scanOptions.fullScan)
{
// Skip file if last write is the same
db::Session& dbSession{ _db.getTLSSession() };
auto transaction{ _db.getTLSSession().createReadTransaction() };
const Track::pointer track{ Track::findByPath(dbSession, file) };
if (track
&& track->getLastWriteTime().toTime_t() == lastWriteTime.toTime_t()
&& track->getScanVersion() == _settings.scanVersion)
{
// this file may have been moved from one library to another, then we just need to update the media library id instead of a full rescan
const auto trackMediaLibrary{ track->getMediaLibrary() };
if (trackMediaLibrary && trackMediaLibrary->getId() == libraryInfo.id)
{
stats.skips++;
return false;
}
needUpdateLibrary = true;
}
}
if (needUpdateLibrary)
{
db::Session& dbSession{ _db.getTLSSession() };
auto transaction{ _db.getTLSSession().createWriteTransaction() };
Track::pointer track{ Track::findByPath(dbSession, file) };
assert(track);
track.modify()->setMediaLibrary(db::MediaLibrary::find(dbSession, libraryInfo.id)); // may be null, will be handled in the next scan anyway
stats.updates++;
return false;
}
return true; // need to scan
}
bool ScanStepScanFiles::checkImageFileNeedScan(ScanContext& context, const std::filesystem::path& file)
{
ScanStats& stats{ context.stats };
const Wt::WDateTime lastWriteTime{ retrieveFileGetLastWrite(file) };
// Should rarely fail as we are currently iterating it
if (!lastWriteTime.isValid())
{
stats.skips++;
return false;
}
if (!context.scanOptions.fullScan)
{
db::Session& dbSession{ _db.getTLSSession() };
auto transaction{ _db.getTLSSession().createReadTransaction() };
const db::Image::pointer image{ db::Image::find(dbSession, file) };
if (image && image->getLastWriteTime() == lastWriteTime)
{
stats.skips++;
return false;
}
}
return true; // need to scan
}
void ScanStepScanFiles::processFileScanResults(ScanContext& context, std::span scanResults, const ScannerSettings::MediaLibraryInfo& libraryInfo)
{
LMS_SCOPED_TRACE_OVERVIEW("Scanner", "ProcessScanResults");
db::Session& dbSession{ _db.getTLSSession() };
auto transaction{ dbSession.createWriteTransaction() };
for (const FileScanResult& scanResult : scanResults)
{
if (_abortScan)
return;
if (const AudioFileScanData * scanData{ std::get_if(&scanResult.scanData) })
{
context.stats.scans++;
processAudioFileScanData(context, scanResult.path, scanData->get(), libraryInfo);
}
else if (const ImageFileScanData * scanData{ std::get_if(&scanResult.scanData) })
{
context.stats.scans++;
processImageFileScanData(context, scanResult.path, scanData->has_value() ? &scanData->value() : nullptr, libraryInfo);
}
}
}
void ScanStepScanFiles::processAudioFileScanData(ScanContext& context, const std::filesystem::path& file, const metadata::Track* trackMetadata, const ScannerSettings::MediaLibraryInfo& libraryInfo)
{
LMS_SCOPED_TRACE_DETAILED("Scanner", "ProcessAudioScanData");
ScanStats& stats{ context.stats };
const std::optional fileInfo{ retrieveFileInfo(file, libraryInfo.rootDirectory) };
if (!fileInfo)
{
stats.skips++;
return;
}
db::Session& dbSession{ _db.getTLSSession() };
Track::pointer track{ Track::findByPath(dbSession, file) };
if (!trackMetadata)
{
if (track)
{
track.remove();
stats.deletions++;
}
context.stats.errors.emplace_back(file, ScanErrorType::CannotReadAudioFile);
return;
}
if (trackMetadata->mbid && (!track || _settings.skipDuplicateMBID))
{
std::vector duplicateTracks{ Track::findByMBID(dbSession, *trackMetadata->mbid) };
// find for an existing track MBID as the file may have just been moved
if (!track && duplicateTracks.size() == 1)
{
Track::pointer otherTrack{ duplicateTracks.front() };
std::error_code ec;
if (!std::filesystem::exists(otherTrack->getAbsoluteFilePath(), ec))
{
LMS_LOG(DBUPDATER, DEBUG, "Considering track '" << file.string() << "' moved from '" << otherTrack->getAbsoluteFilePath() << "'");
track = otherTrack;
track.modify()->setAbsoluteFilePath(file);
}
}
// Skip duplicate track MBID
if (_settings.skipDuplicateMBID)
{
for (Track::pointer& otherTrack : duplicateTracks)
{
// Skip ourselves
if (track && track->getId() == otherTrack->getId())
continue;
// Skip if duplicate files no longer in media root: as it will be removed later, we will end up with no file
if (std::none_of(std::cbegin(_settings.mediaLibraries), std::cend(_settings.mediaLibraries),
[&](const ScannerSettings::MediaLibraryInfo& libraryInfo) {
return core::pathUtils::isPathInRootPath(file, libraryInfo.rootDirectory, &excludeDirFileName);
}))
{
continue;
}
LMS_LOG(DBUPDATER, DEBUG, "Skipped '" << file.string() << "' (similar MBID in '" << otherTrack->getAbsoluteFilePath().string() << "')");
// As this MBID already exists, just remove what we just scanned
if (track)
{
track.remove();
stats.deletions++;
}
return;
}
}
}
// We estimate this is an audio file if the duration is not null
if (trackMetadata->audioProperties.duration == std::chrono::milliseconds::zero())
{
LMS_LOG(DBUPDATER, DEBUG, "Skipped '" << file.string() << "' (duration is 0)");
// If Track exists here, delete it!
if (track)
{
track.remove();
stats.deletions++;
}
stats.errors.emplace_back(file, ScanErrorType::BadDuration);
return;
}
// ***** Title
std::string title;
if (!trackMetadata->title.empty())
title = trackMetadata->title;
else
{
// TODO parse file name guess track etc.
// For now juste use file name as title
title = file.filename().string();
}
// If file already exists, update its data
// Otherwise, create it
bool added{};
if (!track)
{
track = dbSession.create