Added playlist import (and sync)from m3u files, ref #391

This commit is contained in:
emeric
2024-12-18 14:15:25 +01:00
parent ec20c9bfa6
commit 3661eaccb6
95 changed files with 3439 additions and 1634 deletions
+20 -14
View File
@@ -1,20 +1,25 @@
add_library(lmsscanner SHARED
impl/FileScanQueue.cpp
impl/scanners/AudioFileScanner.cpp
impl/scanners/ImageFileScanner.cpp
impl/scanners/LyricsFileScanner.cpp
impl/scanners/PlayListFileScanner.cpp
impl/scanners/Utils.cpp
impl/steps/FileScanQueue.cpp
impl/steps/ScanStepAssociateArtistImages.cpp
impl/steps/ScanStepAssociateExternalLyrics.cpp
impl/steps/ScanStepAssociatePlayListTracks.cpp
impl/steps/ScanStepAssociateReleaseImages.cpp
impl/steps/ScanStepCheckForDuplicatedFiles.cpp
impl/steps/ScanStepCheckForRemovedFiles.cpp
impl/steps/ScanStepCompact.cpp
impl/steps/ScanStepComputeClusterStats.cpp
impl/steps/ScanStepDiscoverFiles.cpp
impl/steps/ScanStepOptimize.cpp
impl/steps/ScanStepRemoveOrphanedDbEntries.cpp
impl/steps/ScanStepScanFiles.cpp
impl/steps/ScanStepUpdateLibraryFields.cpp
impl/ScannerService.cpp
impl/ScannerStats.cpp
impl/ScanStepAssociateArtistImages.cpp
impl/ScanStepAssociateExternalLyrics.cpp
impl/ScanStepAssociateReleaseImages.cpp
impl/ScanStepCheckForDuplicatedFiles.cpp
impl/ScanStepCheckForRemovedFiles.cpp
impl/ScanStepCompact.cpp
impl/ScanStepComputeClusterStats.cpp
impl/ScanStepDiscoverFiles.cpp
impl/ScanStepOptimize.cpp
impl/ScanStepRemoveOrphanedDbEntries.cpp
impl/ScanStepScanFiles.cpp
impl/ScanStepUpdateLibraryFields.cpp
)
target_include_directories(lmsscanner INTERFACE
@@ -23,6 +28,7 @@ target_include_directories(lmsscanner INTERFACE
target_include_directories(lmsscanner PRIVATE
include
impl
)
target_link_libraries(lmsscanner PRIVATE
@@ -1,174 +0,0 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#include "FileScanQueue.hpp"
#include <fstream>
#include "core/ILogger.hpp"
#include "core/ITraceLogger.hpp"
#include "image/Exception.hpp"
#include "image/Image.hpp"
#include "metadata/Exception.hpp"
namespace lms::scanner
{
FileScanQueue::FileScanQueue(metadata::IParser& parser, std::size_t threadCount, bool& abort)
: _metadataParser{ parser }
, _scanContextRunner{ _scanContext, threadCount, "FileScan" }
, _abort{ abort }
{
}
void FileScanQueue::pushScanRequest(const std::filesystem::path& path, ScanRequestType type)
{
{
std::scoped_lock lock{ _mutex };
_ongoingScanCount += 1;
}
_scanContext.post([=, this] {
if (_abort)
{
std::scoped_lock lock{ _mutex };
_ongoingScanCount -= 1;
}
else
{
FileScanResult result;
result.path = path;
switch (type)
{
case ScanRequestType::AudioFile:
result.scanData = scanAudioFile(path);
break;
case ScanRequestType::ImageFile:
result.scanData = scanImageFile(path);
break;
case ScanRequestType::LyricsFile:
result.scanData = scanLyricsFile(path);
break;
}
{
std::scoped_lock lock{ _mutex };
_scanResults.emplace_back(std::move(result));
_ongoingScanCount -= 1;
}
}
_condVar.notify_all();
});
}
AudioFileScanData FileScanQueue::scanAudioFile(const std::filesystem::path& path)
{
LMS_SCOPED_TRACE_OVERVIEW("Scanner", "ScanAudioFile");
std::unique_ptr<metadata::Track> track;
try
{
track = _metadataParser.parse(path);
}
catch (const metadata::Exception& e)
{
LMS_LOG(DBUPDATER, INFO, "Failed to parse audio file '" << path.string() << "'");
}
return track;
}
ImageFileScanData FileScanQueue::scanImageFile(const std::filesystem::path& path)
{
LMS_SCOPED_TRACE_OVERVIEW("Scanner", "ScanImageFile");
std::optional<ImageInfo> optInfo;
try
{
std::unique_ptr<image::IRawImage> rawImage{ image::decodeImage(path) };
ImageInfo& imageInfo{ optInfo.emplace() };
imageInfo.width = rawImage->getWidth();
imageInfo.height = rawImage->getHeight();
}
catch (const image::Exception& e)
{
LMS_LOG(DBUPDATER, ERROR, "Cannot read image in file '" << path.string() << "': " << e.what());
}
return optInfo;
}
LyricsFileScanData FileScanQueue::scanLyricsFile(const std::filesystem::path& path)
{
LMS_SCOPED_TRACE_OVERVIEW("Scanner", "ScanLyricsFile");
LyricsFileScanData lyrics;
try
{
std::ifstream ifs{ path.string() };
if (!ifs)
LMS_LOG(DBUPDATER, ERROR, "Cannot open file '" << path.string() << "'");
else
lyrics = metadata::parseLyrics(ifs);
}
catch (const std::exception& e)
{
LMS_LOG(DBUPDATER, ERROR, "Cannot read lyrics in file '" << path.string() << "': " << e.what());
}
return lyrics;
}
std::size_t FileScanQueue::getResultsCount() const
{
std::scoped_lock lock{ _mutex };
return _scanResults.size();
}
size_t FileScanQueue::popResults(std::vector<FileScanResult>& results, std::size_t maxCount)
{
results.clear();
results.reserve(maxCount);
{
std::scoped_lock lock{ _mutex };
while (results.size() < maxCount && !_scanResults.empty())
{
results.push_back(std::move(_scanResults.front()));
_scanResults.pop_front();
}
}
return results.size();
}
void FileScanQueue::wait(std::size_t maxScanRequestCount)
{
LMS_SCOPED_TRACE_OVERVIEW("Scanner", "WaitParseResults");
std::unique_lock lock{ _mutex };
_condVar.wait(lock, [=, this] { return _ongoingScanCount <= maxScanRequestCount; });
}
} // namespace lms::scanner
@@ -1,86 +0,0 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include <condition_variable>
#include <deque>
#include <filesystem>
#include <mutex>
#include <variant>
#include <vector>
#include "core/IOContextRunner.hpp"
#include "metadata/IParser.hpp"
#include "metadata/Lyrics.hpp"
namespace lms::scanner
{
struct ImageInfo
{
std::size_t height{};
std::size_t width{};
};
using AudioFileScanData = std::unique_ptr<metadata::Track>;
using ImageFileScanData = std::optional<ImageInfo>;
using LyricsFileScanData = std::optional<metadata::Lyrics>;
struct FileScanResult
{
std::filesystem::path path;
std::variant<std::monostate, AudioFileScanData, ImageFileScanData, LyricsFileScanData> scanData;
};
class FileScanQueue
{
public:
FileScanQueue(metadata::IParser& parser, std::size_t threadCount, bool& abort);
std::size_t getThreadCount() const { return _scanContextRunner.getThreadCount(); }
enum ScanRequestType
{
AudioFile,
ImageFile,
LyricsFile,
};
void pushScanRequest(const std::filesystem::path& path, ScanRequestType type);
std::size_t getResultsCount() const;
size_t popResults(std::vector<FileScanResult>& results, std::size_t maxCount);
void wait(std::size_t maxScanRequestCount = 0); // wait until ongoing scan request count <= maxScanRequestCount
private:
AudioFileScanData scanAudioFile(const std::filesystem::path& path);
ImageFileScanData scanImageFile(const std::filesystem::path& path);
LyricsFileScanData scanLyricsFile(const std::filesystem::path& path);
metadata::IParser& _metadataParser;
boost::asio::io_context _scanContext;
core::IOContextRunner _scanContextRunner;
mutable std::mutex _mutex;
std::size_t _ongoingScanCount{};
std::deque<FileScanResult> _scanResults;
std::condition_variable _condVar;
bool& _abort;
};
} // namespace lms::scanner
@@ -0,0 +1,35 @@
/*
* 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include <filesystem>
#include "database/MediaLibraryId.hpp"
namespace lms::scanner
{
struct MediaLibraryInfo
{
db::MediaLibraryId id;
std::filesystem::path rootDirectory;
auto operator<=>(const MediaLibraryInfo& other) const = default;
};
} // namespace lms::scanner
@@ -0,0 +1,33 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include "services/scanner/ScannerOptions.hpp"
#include "services/scanner/ScannerStats.hpp"
namespace lms::scanner
{
struct ScanContext
{
ScanOptions scanOptions;
ScanStats stats;
ScanStepStats currentStepStats;
};
} // namespace lms::scanner
@@ -1,973 +0,0 @@
/*
* 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 <http://www.gnu.org/licenses/>.
*/
#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 "database/TrackLyrics.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<FileInfo> retrieveFileInfo(const std::filesystem::path& file, const std::filesystem::path& rootPath)
{
std::optional<FileInfo> 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 MediaLibrary::pointer& mediaLibrary)
{
Directory::pointer directory{ Directory::find(session, path) };
if (!directory)
{
Directory::pointer parentDirectory;
if (path != mediaLibrary->getPath())
parentDirectory = getOrCreateDirectory(session, path.parent_path(), mediaLibrary);
directory = session.create<Directory>(path);
directory.modify()->setParent(parentDirectory);
directory.modify()->setMediaLibrary(mediaLibrary);
}
// Don't update library if it does not match, will be updated elsewhere
return directory;
}
db::TrackLyrics::pointer createLyrics(Session& session, const metadata::Lyrics& lyricsInfo)
{
db::TrackLyrics::pointer lyrics{ session.create<db::TrackLyrics>() };
lyrics.modify()->setLanguage(!lyricsInfo.language.empty() ? lyricsInfo.language : "xxx");
lyrics.modify()->setOffset(lyricsInfo.offset);
lyrics.modify()->setDisplayArtist(lyricsInfo.displayArtist);
lyrics.modify()->setDisplayTitle(lyricsInfo.displayTitle);
if (!lyricsInfo.synchronizedLines.empty())
lyrics.modify()->setSynchronizedLines(lyricsInfo.synchronizedLines);
else
lyrics.modify()->setUnsynchronizedLines(lyricsInfo.unsynchronizedLines);
return lyrics;
}
Artist::pointer createArtist(Session& session, const metadata::Artist& artistInfo)
{
Artist::pointer artist{ session.create<Artist>(artistInfo.name) };
if (artistInfo.mbid)
artist.modify()->setMBID(artistInfo.mbid);
if (artistInfo.sortName)
artist.modify()->setSortName(*artistInfo.sortName);
return artist;
}
std::string optionalMBIDAsString(const std::optional<core::UUID>& uuid)
{
return uuid ? std::string{ uuid->getAsString() } : "<no MBID>";
}
void updateArtistIfNeeded(Artist::pointer artist, const metadata::Artist& artistInfo)
{
// Name may have been updated
if (artist->getName() != artistInfo.name)
{
LMS_LOG(DBUPDATER, DEBUG, "Artist [" << optionalMBIDAsString(artist->getMBID()) << "], updated name from '" << artist->getName() << "' to '" << artistInfo.name << "'");
artist.modify()->setName(artistInfo.name);
}
// Sortname may have been updated
// As the sort name is quite often not filled in, we update it only if already set (for now?)
if (artistInfo.sortName && *artistInfo.sortName != artist->getSortName())
{
LMS_LOG(DBUPDATER, DEBUG, "Artist [" << optionalMBIDAsString(artist->getMBID()) << "], updated sort name from '" << artist->getSortName() << "' to '" << *artistInfo.sortName << "'");
artist.modify()->setSortName(*artistInfo.sortName);
}
}
std::vector<Artist::pointer> getOrCreateArtists(Session& session, const std::vector<metadata::Artist>& artistsInfo, bool allowFallbackOnMBIDEntries)
{
std::vector<Artist::pointer> 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<ReleaseType>(name);
return releaseType;
}
Label::pointer getOrCreateLabel(Session& session, std::string_view name)
{
Label::pointer label{ Label::find(session, name) };
if (!label)
label = session.create<Label>(name);
return label;
}
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->isCompilation() != releaseInfo.isCompilation)
release.modify()->setCompilation(releaseInfo.isCompilation);
if (release->getBarcode() != releaseInfo.barcode)
release.modify()->setBarcode(releaseInfo.barcode);
if (release->getReleaseTypeNames() != releaseInfo.releaseTypes)
{
release.modify()->clearReleaseTypes();
for (std::string_view releaseType : releaseInfo.releaseTypes)
release.modify()->addReleaseType(getOrCreateReleaseType(session, releaseType));
}
if (release->getLabelNames() != releaseInfo.labels)
{
release.modify()->clearLabels();
for (std::string_view label : releaseInfo.labels)
release.modify()->addLabel(getOrCreateLabel(session, label));
}
}
// Compare release level info
bool isReleaseMatching(const Release::pointer& candidateRelease, const metadata::Release& releaseInfo)
{
// TODO: add more criterias?
return candidateRelease->getName() == releaseInfo.name
&& candidateRelease->getSortName() == releaseInfo.sortName
&& candidateRelease->getTotalDisc() == releaseInfo.mediumCount
&& candidateRelease->isCompilation() == releaseInfo.isCompilation
&& candidateRelease->getLabelNames() == releaseInfo.labels
&& candidateRelease->getBarcode() == releaseInfo.barcode;
}
Release::pointer getOrCreateRelease(Session& session, const metadata::Release& releaseInfo, const Directory::pointer& currentDirectory)
{
Release::pointer release;
// First try to get by MBID: fastest, safest
if (releaseInfo.mbid)
{
release = Release::find(session, *releaseInfo.mbid);
if (!release)
release = session.create<Release>(releaseInfo.name, releaseInfo.mbid);
}
else if (releaseInfo.name.empty())
{
// No release name (only mbid) -> nothing to do
return release;
}
// Fall back on release name (collisions may occur)
// First try using all sibling directories (case for Album/DiscX), only if the disc number is set
const DirectoryId parentDirectoryId{ currentDirectory->getParentDirectoryId() };
if (!release && releaseInfo.mediumCount && *releaseInfo.mediumCount > 1 && parentDirectoryId.isValid())
{
Release::FindParameters params;
params.setParentDirectory(parentDirectoryId);
params.setName(releaseInfo.name);
Release::find(session, params, [&](const Release::pointer& candidateRelease) {
// Already found a candidate
if (release)
return;
// Do not fallback on properly tagged releases
if (candidateRelease->getMBID().has_value())
return;
if (!isReleaseMatching(candidateRelease, releaseInfo))
return;
release = candidateRelease;
});
}
// Lastly try in the current directory: we do this at last to have
// opportunities to merge releases in case of migration / rescan
if (!release)
{
Release::FindParameters params;
params.setDirectory(currentDirectory->getId());
params.setName(releaseInfo.name);
Release::find(session, params, [&](const Release::pointer& candidateRelease) {
// Already found a candidate
if (release)
return;
// Do not fallback on properly tagged releases
if (candidateRelease->getMBID().has_value())
return;
if (!isReleaseMatching(candidateRelease, releaseInfo))
return;
release = candidateRelease;
});
}
if (!release)
release = session.create<Release>(releaseInfo.name);
updateReleaseIfNeeded(session, release, releaseInfo);
return release;
}
std::vector<Cluster::pointer> getOrCreateClusters(Session& session, const metadata::Track& track)
{
std::vector<Cluster::pointer> clusters;
auto getOrCreateClusters{ [&](std::string_view tag, std::span<const std::string> values) {
auto clusterType = ClusterType::find(session, tag);
if (!clusterType)
clusterType = session.create<ClusterType>(tag);
for (const auto& value : values)
{
auto cluster{ clusterType->getCluster(value) };
if (!cluster)
cluster = session.create<Cluster>(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<core::IConfig>::get()->getString("scanner-parser-read-style", "average") };
if (readStyle == "fast")
return metadata::ParserReadStyle::Fast;
if (readStyle == "average")
return metadata::ParserReadStyle::Average;
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<core::IConfig>::get()->getULong("scanner-metadata-thread-count", 0) };
if (threadCount == 0)
threadCount = std::max<std::size_t>(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<std::string> 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<FileScanResult> 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; // stop iterating
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 fileMatched{};
if (core::pathUtils::hasFileAnyExtension(path, _settings.supportedAudioFileExtensions))
{
fileMatched = true;
if (checkAudioFileNeedScan(context, path, mediaLibrary))
_fileScanQueue.pushScanRequest(path, FileScanQueue::ScanRequestType::AudioFile);
}
else if (core::pathUtils::hasFileAnyExtension(path, _settings.supportedImageFileExtensions))
{
fileMatched = true;
if (checkImageFileNeedScan(context, path))
_fileScanQueue.pushScanRequest(path, FileScanQueue::ScanRequestType::ImageFile);
}
else if (core::pathUtils::hasFileAnyExtension(path, _settings.supportedLyricsFileExtensions))
{
fileMatched = true;
if (checkLyricsFileNeedScan(context, path))
_fileScanQueue.pushScanRequest(path, FileScanQueue::ScanRequestType::LyricsFile);
}
if (fileMatched)
{
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;
}
if (context.scanOptions.fullScan)
return true;
bool needUpdateLibrary{};
db::Session& dbSession{ _db.getTLSSession() };
{
auto transaction{ dbSession.createReadTransaction() };
// Skip file if last write is the same
const Track::pointer track{ Track::findByPath(dbSession, file) };
if (track
&& track->getLastWriteTime() == lastWriteTime
&& 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)
{
auto transaction{ dbSession.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
}
bool ScanStepScanFiles::checkLyricsFileNeedScan(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::TrackLyrics::pointer lyrics{ db::TrackLyrics::find(dbSession, file) };
if (lyrics && lyrics->getLastWriteTime() == lastWriteTime)
{
stats.skips++;
return false;
}
}
return true; // need to scan
}
void ScanStepScanFiles::processFileScanResults(ScanContext& context, std::span<const FileScanResult> 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<AudioFileScanData>(&scanResult.scanData) })
{
context.stats.scans++;
processAudioFileScanData(context, scanResult.path, scanData->get(), libraryInfo);
}
else if (const ImageFileScanData * scanData{ std::get_if<ImageFileScanData>(&scanResult.scanData) })
{
context.stats.scans++;
processImageFileScanData(context, scanResult.path, scanData->has_value() ? &scanData->value() : nullptr, libraryInfo);
}
else if (const LyricsFileScanData * scanData{ std::get_if<LyricsFileScanData>(&scanResult.scanData) })
{
context.stats.scans++;
processLyricsFileScanData(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> 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<Track::pointer> 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<Track>();
track.modify()->setAbsoluteFilePath(file);
added = true;
}
// Track related data
assert(track);
// Audio properties
track.modify()->setBitrate(trackMetadata->audioProperties.bitrate);
track.modify()->setBitsPerSample(trackMetadata->audioProperties.bitsPerSample);
track.modify()->setChannelCount(trackMetadata->audioProperties.channelCount);
track.modify()->setDuration(trackMetadata->audioProperties.duration);
track.modify()->setSampleRate(trackMetadata->audioProperties.sampleRate);
track.modify()->setRelativeFilePath(fileInfo->relativePath);
track.modify()->setFileSize(fileInfo->fileSize);
track.modify()->setLastWriteTime(fileInfo->lastWriteTime);
MediaLibrary::pointer mediaLibrary{ MediaLibrary::find(dbSession, libraryInfo.id) }; // may be null if settings are updated in // => next scan will correct this
track.modify()->setMediaLibrary(mediaLibrary);
Directory::pointer directory{ getOrCreateDirectory(dbSession, file.parent_path(), mediaLibrary) };
track.modify()->setDirectory(directory);
track.modify()->clearArtistLinks();
// Do not fallback on artists with the same name but having a MBID for artist and releaseArtists, as it may be corrected by properly tagging files
for (const Artist::pointer& artist : getOrCreateArtists(dbSession, trackMetadata->artists, false))
track.modify()->addArtistLink(TrackArtistLink::create(dbSession, track, artist, TrackArtistLinkType::Artist));
if (trackMetadata->medium && trackMetadata->medium->release)
{
for (const Artist::pointer& releaseArtist : getOrCreateArtists(dbSession, trackMetadata->medium->release->artists, false))
track.modify()->addArtistLink(TrackArtistLink::create(dbSession, track, releaseArtist, TrackArtistLinkType::ReleaseArtist));
}
// Allow fallbacks on artists with the same name even if they have MBID, since there is no tag to indicate the MBID of these artists
// We could ask MusicBrainz to get all the information, but that would heavily slow down the import process
for (const Artist::pointer& conductor : getOrCreateArtists(dbSession, trackMetadata->conductorArtists, true))
track.modify()->addArtistLink(TrackArtistLink::create(dbSession, track, conductor, TrackArtistLinkType::Conductor));
for (const Artist::pointer& composer : getOrCreateArtists(dbSession, trackMetadata->composerArtists, true))
track.modify()->addArtistLink(TrackArtistLink::create(dbSession, track, composer, TrackArtistLinkType::Composer));
for (const Artist::pointer& lyricist : getOrCreateArtists(dbSession, trackMetadata->lyricistArtists, true))
track.modify()->addArtistLink(TrackArtistLink::create(dbSession, track, lyricist, TrackArtistLinkType::Lyricist));
for (const Artist::pointer& mixer : getOrCreateArtists(dbSession, trackMetadata->mixerArtists, true))
track.modify()->addArtistLink(TrackArtistLink::create(dbSession, track, mixer, TrackArtistLinkType::Mixer));
for (const auto& [role, performers] : trackMetadata->performerArtists)
{
for (const Artist::pointer& performer : getOrCreateArtists(dbSession, performers, true))
track.modify()->addArtistLink(TrackArtistLink::create(dbSession, track, performer, TrackArtistLinkType::Performer, role));
}
for (const Artist::pointer& producer : getOrCreateArtists(dbSession, trackMetadata->producerArtists, true))
track.modify()->addArtistLink(TrackArtistLink::create(dbSession, track, producer, TrackArtistLinkType::Producer));
for (const Artist::pointer& remixer : getOrCreateArtists(dbSession, trackMetadata->remixerArtists, true))
track.modify()->addArtistLink(TrackArtistLink::create(dbSession, track, remixer, TrackArtistLinkType::Remixer));
track.modify()->setScanVersion(_settings.scanVersion);
if (trackMetadata->medium && trackMetadata->medium->release)
track.modify()->setRelease(getOrCreateRelease(dbSession, *trackMetadata->medium->release, directory));
else
track.modify()->setRelease({});
track.modify()->setTotalTrack(trackMetadata->medium ? trackMetadata->medium->trackCount : std::nullopt);
track.modify()->setReleaseReplayGain(trackMetadata->medium ? trackMetadata->medium->replayGain : std::nullopt);
track.modify()->setDiscSubtitle(trackMetadata->medium ? trackMetadata->medium->name : "");
track.modify()->setClusters(getOrCreateClusters(dbSession, *trackMetadata));
track.modify()->setName(title);
track.modify()->setAddedTime(Wt::WDateTime::currentDateTime());
track.modify()->setTrackNumber(trackMetadata->position);
track.modify()->setDiscNumber(trackMetadata->medium ? trackMetadata->medium->position : std::nullopt);
track.modify()->setDate(trackMetadata->date);
track.modify()->setYear(trackMetadata->year);
track.modify()->setOriginalDate(trackMetadata->originalDate);
track.modify()->setOriginalYear(trackMetadata->originalYear);
// If a file has an OriginalDate but no date, set it to ease filtering
if (!trackMetadata->date.isValid() && trackMetadata->originalDate.isValid())
track.modify()->setDate(trackMetadata->originalDate);
// If a file has an OriginalYear but no Year, set it to ease filtering
if (!trackMetadata->year && trackMetadata->originalYear)
track.modify()->setYear(trackMetadata->originalYear);
track.modify()->setRecordingMBID(trackMetadata->recordingMBID);
track.modify()->setTrackMBID(trackMetadata->mbid);
if (auto trackFeatures{ TrackFeatures::find(dbSession, track->getId()) })
trackFeatures.remove(); // TODO: only if MBID changed?
track.modify()->setHasCover(trackMetadata->hasCover);
track.modify()->setCopyright(trackMetadata->copyright);
track.modify()->setCopyrightURL(trackMetadata->copyrightURL);
track.modify()->setComment(!trackMetadata->comments.empty() ? trackMetadata->comments.front() : ""); // only take the first one for now
track.modify()->setTrackReplayGain(trackMetadata->replayGain);
track.modify()->setArtistDisplayName(trackMetadata->artistDisplayName);
track.modify()->clearEmbeddedLyrics();
for (const metadata::Lyrics& lyricsInfo : trackMetadata->lyrics)
{
db::TrackLyrics::pointer lyrics{ createLyrics(dbSession, lyricsInfo) };
track.modify()->addLyrics(lyrics);
}
if (added)
{
LMS_LOG(DBUPDATER, DEBUG, "Added audio file '" << file.string() << "'");
stats.additions++;
}
else
{
LMS_LOG(DBUPDATER, DEBUG, "Updated audio file '" << file.string() << "'");
stats.updates++;
}
}
void ScanStepScanFiles::processImageFileScanData(ScanContext& context, const std::filesystem::path& file, const ImageInfo* imageInfo, const ScannerSettings::MediaLibraryInfo& libraryInfo)
{
LMS_SCOPED_TRACE_DETAILED("Scanner", "ProcessImageScanData");
ScanStats& stats{ context.stats };
const std::optional<FileInfo> fileInfo{ retrieveFileInfo(file, libraryInfo.rootDirectory) };
if (!fileInfo)
{
stats.skips++;
return;
}
db::Session& dbSession{ _db.getTLSSession() };
db::Image::pointer image{ db::Image::find(dbSession, file) };
if (!imageInfo)
{
if (image)
{
image.remove();
stats.deletions++;
}
context.stats.errors.emplace_back(file, ScanErrorType::CannotReadImageFile);
return;
}
const bool added{ !image };
if (!image)
image = dbSession.create<db::Image>(file);
image.modify()->setLastWriteTime(fileInfo->lastWriteTime);
image.modify()->setFileSize(fileInfo->fileSize);
image.modify()->setHeight(imageInfo->height);
image.modify()->setWidth(imageInfo->width);
MediaLibrary::pointer mediaLibrary{ MediaLibrary::find(dbSession, libraryInfo.id) }; // may be null if settings are updated in // => next scan will correct this
image.modify()->setDirectory(getOrCreateDirectory(dbSession, file.parent_path(), mediaLibrary));
if (added)
{
LMS_LOG(DBUPDATER, DEBUG, "Added image '" << file.string() << "'");
stats.additions++;
}
else
{
LMS_LOG(DBUPDATER, DEBUG, "Updated image '" << file.string() << "'");
stats.updates++;
}
}
void ScanStepScanFiles::processLyricsFileScanData(ScanContext& context, const std::filesystem::path& file, const metadata::Lyrics* lyricsInfo, const ScannerSettings::MediaLibraryInfo& libraryInfo)
{
LMS_SCOPED_TRACE_DETAILED("Scanner", "ProcessImageScanData");
ScanStats& stats{ context.stats };
const std::optional<FileInfo> fileInfo{ retrieveFileInfo(file, libraryInfo.rootDirectory) };
if (!fileInfo)
{
stats.skips++;
return;
}
db::Session& dbSession{ _db.getTLSSession() };
db::TrackLyrics::pointer trackLyrics{ db::TrackLyrics::find(dbSession, file) };
if (!lyricsInfo)
{
if (trackLyrics)
{
trackLyrics.remove();
stats.deletions++;
}
context.stats.errors.emplace_back(file, ScanErrorType::CannotReadLyricsFile);
return;
}
const bool added{ !trackLyrics };
if (!trackLyrics)
{
trackLyrics = dbSession.create<db::TrackLyrics>();
trackLyrics.modify()->setAbsoluteFilePath(file);
}
trackLyrics.modify()->setLastWriteTime(fileInfo->lastWriteTime);
trackLyrics.modify()->setFileSize(fileInfo->fileSize);
trackLyrics.modify()->setLanguage(!lyricsInfo->language.empty() ? lyricsInfo->language : "xxx");
trackLyrics.modify()->setOffset(lyricsInfo->offset);
trackLyrics.modify()->setDisplayTitle(lyricsInfo->displayTitle);
trackLyrics.modify()->setDisplayArtist(lyricsInfo->displayArtist);
if (!lyricsInfo->synchronizedLines.empty())
trackLyrics.modify()->setSynchronizedLines(lyricsInfo->synchronizedLines);
else
trackLyrics.modify()->setUnsynchronizedLines(lyricsInfo->unsynchronizedLines);
MediaLibrary::pointer mediaLibrary{ MediaLibrary::find(dbSession, libraryInfo.id) }; // may be null if settings are updated in // => next scan will correct this
trackLyrics.modify()->setDirectory(getOrCreateDirectory(dbSession, file.parent_path(), mediaLibrary));
if (added)
{
LMS_LOG(DBUPDATER, DEBUG, "Added external lyrics '" << file.string() << "'");
stats.additions++;
}
else
{
LMS_LOG(DBUPDATER, DEBUG, "Updated external lyrics '" << file.string() << "'");
stats.updates++;
}
}
} // namespace lms::scanner
@@ -21,26 +21,32 @@
#include <ctime>
#include <Wt/WDate.h>
#include "core/IConfig.hpp"
#include "core/ILogger.hpp"
#include "core/ITraceLogger.hpp"
#include "database/MediaLibrary.hpp"
#include "database/ScanSettings.hpp"
#include "database/TrackFeatures.hpp"
#include "image/Image.hpp"
#include "ScanStepAssociateArtistImages.hpp"
#include "ScanStepAssociateExternalLyrics.hpp"
#include "ScanStepAssociateReleaseImages.hpp"
#include "ScanStepCheckForDuplicatedFiles.hpp"
#include "ScanStepCheckForRemovedFiles.hpp"
#include "ScanStepCompact.hpp"
#include "ScanStepComputeClusterStats.hpp"
#include "ScanStepDiscoverFiles.hpp"
#include "ScanStepOptimize.hpp"
#include "ScanStepRemoveOrphanedDbEntries.hpp"
#include "ScanStepScanFiles.hpp"
#include "ScanStepUpdateLibraryFields.hpp"
#include "scanners/AudioFileScanner.hpp"
#include "scanners/ImageFileScanner.hpp"
#include "scanners/LyricsFileScanner.hpp"
#include "scanners/PlayListFileScanner.hpp"
#include "steps/ScanStepAssociateArtistImages.hpp"
#include "steps/ScanStepAssociateExternalLyrics.hpp"
#include "steps/ScanStepAssociatePlayListTracks.hpp"
#include "steps/ScanStepAssociateReleaseImages.hpp"
#include "steps/ScanStepCheckForDuplicatedFiles.hpp"
#include "steps/ScanStepCheckForRemovedFiles.hpp"
#include "steps/ScanStepCompact.hpp"
#include "steps/ScanStepComputeClusterStats.hpp"
#include "steps/ScanStepDiscoverFiles.hpp"
#include "steps/ScanStepOptimize.hpp"
#include "steps/ScanStepRemoveOrphanedDbEntries.hpp"
#include "steps/ScanStepScanFiles.hpp"
#include "steps/ScanStepUpdateLibraryFields.hpp"
namespace lms::scanner
{
@@ -270,7 +276,7 @@ namespace lms::scanner
refreshScanSettings();
IScanStep::ScanContext scanContext;
ScanContext scanContext;
scanContext.scanOptions = scanOptions;
ScanStats& stats{ scanContext.stats };
stats.startTime = Wt::WDateTime::currentDateTime();
@@ -341,27 +347,38 @@ namespace lms::scanner
notifyInProgressIfNeeded(stats);
} };
_fileScanners.clear();
_fileScanners.emplace_back(std::make_unique<AudioFileScanner>(_db, _settings));
_fileScanners.emplace_back(std::make_unique<ImageFileScanner>(_db));
_fileScanners.emplace_back(std::make_unique<LyricsFileScanner>(_db));
_fileScanners.emplace_back(std::make_unique<PlayListFileScanner>(_db));
std::vector<IFileScanner*> fileScanners;
std::transform(std::cbegin(_fileScanners), std::cend(_fileScanners), std::back_inserter(fileScanners), [](const std::unique_ptr<IFileScanner>& scanner) { return scanner.get(); });
ScanStepBase::InitParams params{
_settings,
cbFunc,
_abortScan,
_db
.settings = _settings,
.progressCallback = cbFunc,
.abortScan = _abortScan,
.db = _db,
.fileScanners = fileScanners,
};
// Order is important: steps are sequential
_scanSteps.clear();
_scanSteps.push_back(std::make_unique<ScanStepDiscoverFiles>(params));
_scanSteps.push_back(std::make_unique<ScanStepScanFiles>(params));
_scanSteps.push_back(std::make_unique<ScanStepCheckForRemovedFiles>(params));
_scanSteps.push_back(std::make_unique<ScanStepUpdateLibraryFields>(params));
_scanSteps.push_back(std::make_unique<ScanStepAssociateArtistImages>(params));
_scanSteps.push_back(std::make_unique<ScanStepAssociateReleaseImages>(params));
_scanSteps.push_back(std::make_unique<ScanStepAssociateExternalLyrics>(params));
_scanSteps.push_back(std::make_unique<ScanStepRemoveOrphanedDbEntries>(params));
_scanSteps.push_back(std::make_unique<ScanStepCompact>(params));
_scanSteps.push_back(std::make_unique<ScanStepOptimize>(params));
_scanSteps.push_back(std::make_unique<ScanStepComputeClusterStats>(params));
_scanSteps.push_back(std::make_unique<ScanStepCheckForDuplicatedFiles>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepDiscoverFiles>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepScanFiles>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepCheckForRemovedFiles>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepAssociatePlayListTracks>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepUpdateLibraryFields>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepAssociateArtistImages>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepAssociateReleaseImages>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepAssociateExternalLyrics>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepRemoveOrphanedDbEntries>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepCompact>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepOptimize>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepComputeClusterStats>(params));
_scanSteps.emplace_back(std::make_unique<ScanStepCheckForDuplicatedFiles>(params));
}
ScannerSettings ScannerService::readSettings()
@@ -378,29 +395,8 @@ namespace lms::scanner
newSettings.startTime = scanSettings->getUpdateStartTime();
newSettings.updatePeriod = scanSettings->getUpdatePeriod();
{
const auto audioFileExtensions{ scanSettings->getAudioFileExtensions() };
newSettings.supportedAudioFileExtensions.reserve(audioFileExtensions.size());
std::transform(std::cbegin(audioFileExtensions), std::cend(audioFileExtensions), std::back_inserter(newSettings.supportedAudioFileExtensions),
[](const std::filesystem::path& extension) { return std::filesystem::path{ core::stringUtils::stringToLower(extension.string()) }; });
}
{
const auto imageFileExtensions{ image::getSupportedFileExtensions() };
newSettings.supportedImageFileExtensions.reserve(imageFileExtensions.size());
std::transform(std::cbegin(imageFileExtensions), std::cend(imageFileExtensions), std::back_inserter(newSettings.supportedImageFileExtensions),
[](const std::filesystem::path& extension) { return std::filesystem::path{ core::stringUtils::stringToLower(extension.string()) }; });
}
{
const auto lyricsFileExtensions{ metadata::getSupportedLyricsFileExtensions() };
newSettings.supportedLyricsFileExtensions.reserve(lyricsFileExtensions.size());
std::transform(std::cbegin(lyricsFileExtensions), std::cend(lyricsFileExtensions), std::back_inserter(newSettings.supportedLyricsFileExtensions),
[](const std::filesystem::path& extension) { return std::filesystem::path{ core::stringUtils::stringToLower(extension.string()) }; });
}
MediaLibrary::find(_db.getTLSSession(), [&](const MediaLibrary::pointer& mediaLibrary) {
newSettings.mediaLibraries.push_back(ScannerSettings::MediaLibraryInfo{ mediaLibrary->getId(), mediaLibrary->getPath().lexically_normal() });
newSettings.mediaLibraries.push_back(MediaLibraryInfo{ .id = mediaLibrary->getId(), .rootDirectory = mediaLibrary->getPath().lexically_normal() });
});
{
@@ -29,14 +29,16 @@
#include <Wt/WSignal.h>
#include <boost/asio/system_timer.hpp>
#include "IScanStep.hpp"
#include "ScannerSettings.hpp"
#include "database/Db.hpp"
#include "database/Session.hpp"
#include "services/scanner/IScannerService.hpp"
#include "steps/IScanStep.hpp"
namespace lms::scanner
{
class IFileScanner;
class ScannerService : public IScannerService
{
public:
@@ -73,6 +75,7 @@ namespace lms::scanner
void notifyInProgressIfNeeded(const ScanStepStats& stats);
void notifyInProgress(const ScanStepStats& stats);
std::vector<std::unique_ptr<IFileScanner>> _fileScanners;
std::vector<std::unique_ptr<IScanStep>> _scanSteps;
std::mutex _controlMutex;
@@ -25,31 +25,23 @@
#include <Wt/WDateTime.h>
#include "database/MediaLibraryId.hpp"
#include "database/ScanSettings.hpp"
#include "MediaLibraryInfo.hpp"
namespace lms::scanner
{
static inline const std::filesystem::path excludeDirFileName{ ".lmsignore" };
struct ScannerSettings
{
std::size_t scanVersion{};
Wt::WTime startTime;
db::ScanSettings::UpdatePeriod updatePeriod{ db::ScanSettings::UpdatePeriod::Never };
std::vector<std::filesystem::path> supportedAudioFileExtensions;
std::vector<std::filesystem::path> supportedImageFileExtensions;
std::vector<std::filesystem::path> supportedLyricsFileExtensions;
bool skipDuplicateMBID{};
std::vector<std::string> extraTags;
std::vector<std::string> artistTagDelimiters;
std::vector<std::string> defaultTagDelimiters;
struct MediaLibraryInfo
{
db::MediaLibraryId id;
std::filesystem::path rootDirectory;
auto operator<=>(const MediaLibraryInfo& other) const = default;
};
std::vector<MediaLibraryInfo> mediaLibraries;
bool operator==(const ScannerSettings& rhs) const = default;
@@ -0,0 +1,668 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#include "AudioFileScanner.hpp"
#include "core/IConfig.hpp"
#include "core/ILogger.hpp"
#include "core/ITraceLogger.hpp"
#include "core/Path.hpp"
#include "core/Service.hpp"
#include "database/Artist.hpp"
#include "database/Cluster.hpp"
#include "database/Db.hpp"
#include "database/Directory.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 "database/TrackLyrics.hpp"
#include "metadata/Exception.hpp"
#include "metadata/IParser.hpp"
#include "IFileScanOperation.hpp"
#include "ScanContext.hpp"
#include "ScannerSettings.hpp"
#include "Utils.hpp"
namespace lms::scanner
{
namespace
{
db::Artist::pointer createArtist(db::Session& session, const metadata::Artist& artistInfo)
{
db::Artist::pointer artist{ session.create<db::Artist>(artistInfo.name) };
if (artistInfo.mbid)
artist.modify()->setMBID(artistInfo.mbid);
if (artistInfo.sortName)
artist.modify()->setSortName(*artistInfo.sortName);
return artist;
}
std::string optionalMBIDAsString(const std::optional<core::UUID>& uuid)
{
return uuid ? std::string{ uuid->getAsString() } : "<no MBID>";
}
void updateArtistIfNeeded(db::Artist::pointer artist, const metadata::Artist& artistInfo)
{
// Name may have been updated
if (artist->getName() != artistInfo.name)
{
LMS_LOG(DBUPDATER, DEBUG, "Artist [" << optionalMBIDAsString(artist->getMBID()) << "], updated name from '" << artist->getName() << "' to '" << artistInfo.name << "'");
artist.modify()->setName(artistInfo.name);
}
// Sortname may have been updated
// As the sort name is quite often not filled in, we update it only if already set (for now?)
if (artistInfo.sortName && *artistInfo.sortName != artist->getSortName())
{
LMS_LOG(DBUPDATER, DEBUG, "Artist [" << optionalMBIDAsString(artist->getMBID()) << "], updated sort name from '" << artist->getSortName() << "' to '" << *artistInfo.sortName << "'");
artist.modify()->setSortName(*artistInfo.sortName);
}
}
std::vector<db::Artist::pointer> getOrCreateArtists(db::Session& session, const std::vector<metadata::Artist>& artistsInfo, bool allowFallbackOnMBIDEntries)
{
std::vector<db::Artist::pointer> artists;
for (const metadata::Artist& artistInfo : artistsInfo)
{
db::Artist::pointer artist;
// First try to get by MBID
if (artistInfo.mbid)
{
artist = db::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 db::Artist::pointer& sameNamedArtist : db::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;
}
db::ReleaseType::pointer getOrCreateReleaseType(db::Session& session, std::string_view name)
{
db::ReleaseType::pointer releaseType{ db::ReleaseType::find(session, name) };
if (!releaseType)
releaseType = session.create<db::ReleaseType>(name);
return releaseType;
}
db::Label::pointer getOrCreateLabel(db::Session& session, std::string_view name)
{
db::Label::pointer label{ db::Label::find(session, name) };
if (!label)
label = session.create<db::Label>(name);
return label;
}
void updateReleaseIfNeeded(db::Session& session, db::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->isCompilation() != releaseInfo.isCompilation)
release.modify()->setCompilation(releaseInfo.isCompilation);
if (release->getBarcode() != releaseInfo.barcode)
release.modify()->setBarcode(releaseInfo.barcode);
if (release->getReleaseTypeNames() != releaseInfo.releaseTypes)
{
release.modify()->clearReleaseTypes();
for (std::string_view releaseType : releaseInfo.releaseTypes)
release.modify()->addReleaseType(getOrCreateReleaseType(session, releaseType));
}
if (release->getLabelNames() != releaseInfo.labels)
{
release.modify()->clearLabels();
for (std::string_view label : releaseInfo.labels)
release.modify()->addLabel(getOrCreateLabel(session, label));
}
}
// Compare release level info
bool isReleaseMatching(const db::Release::pointer& candidateRelease, const metadata::Release& releaseInfo)
{
// TODO: add more criterias?
return candidateRelease->getName() == releaseInfo.name
&& candidateRelease->getSortName() == releaseInfo.sortName
&& candidateRelease->getTotalDisc() == releaseInfo.mediumCount
&& candidateRelease->isCompilation() == releaseInfo.isCompilation
&& candidateRelease->getLabelNames() == releaseInfo.labels
&& candidateRelease->getBarcode() == releaseInfo.barcode;
}
db::Release::pointer getOrCreateRelease(db::Session& session, const metadata::Release& releaseInfo, const db::Directory::pointer& currentDirectory)
{
db::Release::pointer release;
// First try to get by MBID: fastest, safest
if (releaseInfo.mbid)
{
release = db::Release::find(session, *releaseInfo.mbid);
if (!release)
release = session.create<db::Release>(releaseInfo.name, releaseInfo.mbid);
}
else if (releaseInfo.name.empty())
{
// No release name (only mbid) -> nothing to do
return release;
}
// Fall back on release name (collisions may occur)
// First try using all sibling directories (case for Album/DiscX), only if the disc number is set
const db::DirectoryId parentDirectoryId{ currentDirectory->getParentDirectoryId() };
if (!release && releaseInfo.mediumCount && *releaseInfo.mediumCount > 1 && parentDirectoryId.isValid())
{
db::Release::FindParameters params;
params.setParentDirectory(parentDirectoryId);
params.setName(releaseInfo.name);
db::Release::find(session, params, [&](const db::Release::pointer& candidateRelease) {
// Already found a candidate
if (release)
return;
// Do not fallback on properly tagged releases
if (candidateRelease->getMBID().has_value())
return;
if (!isReleaseMatching(candidateRelease, releaseInfo))
return;
release = candidateRelease;
});
}
// Lastly try in the current directory: we do this at last to have
// opportunities to merge releases in case of migration / rescan
if (!release)
{
db::Release::FindParameters params;
params.setDirectory(currentDirectory->getId());
params.setName(releaseInfo.name);
db::Release::find(session, params, [&](const db::Release::pointer& candidateRelease) {
// Already found a candidate
if (release)
return;
// Do not fallback on properly tagged releases
if (candidateRelease->getMBID().has_value())
return;
if (!isReleaseMatching(candidateRelease, releaseInfo))
return;
release = candidateRelease;
});
}
if (!release)
release = session.create<db::Release>(releaseInfo.name);
updateReleaseIfNeeded(session, release, releaseInfo);
return release;
}
std::vector<db::Cluster::pointer> getOrCreateClusters(db::Session& session, const metadata::Track& track)
{
std::vector<db::Cluster::pointer> clusters;
auto getOrCreateClusters{ [&](std::string_view tag, std::span<const std::string> values) {
auto clusterType = db::ClusterType::find(session, tag);
if (!clusterType)
clusterType = session.create<db::ClusterType>(tag);
for (const auto& value : values)
{
auto cluster{ clusterType->getCluster(value) };
if (!cluster)
cluster = session.create<db::Cluster>(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;
}
db::TrackLyrics::pointer createLyrics(db::Session& session, const metadata::Lyrics& lyricsInfo)
{
db::TrackLyrics::pointer lyrics{ session.create<db::TrackLyrics>() };
lyrics.modify()->setLanguage(!lyricsInfo.language.empty() ? lyricsInfo.language : "xxx");
lyrics.modify()->setOffset(lyricsInfo.offset);
lyrics.modify()->setDisplayArtist(lyricsInfo.displayArtist);
lyrics.modify()->setDisplayTitle(lyricsInfo.displayTitle);
if (!lyricsInfo.synchronizedLines.empty())
lyrics.modify()->setSynchronizedLines(lyricsInfo.synchronizedLines);
else
lyrics.modify()->setUnsynchronizedLines(lyricsInfo.unsynchronizedLines);
return lyrics;
}
class AudioFileScanOperation : public IFileScanOperation
{
public:
AudioFileScanOperation(const FileToScan& fileToScan, db::Db& db, metadata::IParser& parser, const ScannerSettings& settings)
: _file{ fileToScan.file }
, _mediaLibrary{ fileToScan.mediaLibrary }
, _db{ db }
, _parser{ parser }
, _settings{ settings }
{
}
~AudioFileScanOperation() override = default;
AudioFileScanOperation(const AudioFileScanOperation&) = delete;
AudioFileScanOperation& operator=(const AudioFileScanOperation&) = delete;
private:
core::LiteralString getName() const override { return "ScanAudioFile"; }
void scan() override;
void processResult(ScanContext& context) override;
const std::filesystem::path _file;
const MediaLibraryInfo _mediaLibrary;
db::Db& _db;
metadata::IParser& _parser;
const ScannerSettings& _settings;
std::unique_ptr<metadata::Track> _parsedTrack;
};
void AudioFileScanOperation::scan()
{
LMS_SCOPED_TRACE_OVERVIEW("Scanner", "ScanAudioFile");
std::unique_ptr<metadata::Track> track;
try
{
_parsedTrack = _parser.parse(_file);
}
catch (const metadata::Exception& e)
{
LMS_LOG(DBUPDATER, INFO, "Failed to parse audio file '" << _file.string() << "'");
}
}
void AudioFileScanOperation::processResult(ScanContext& context)
{
LMS_SCOPED_TRACE_DETAILED("Scanner", "ProcessAudioScanData");
ScanStats& stats{ context.stats };
const std::optional<FileInfo> fileInfo{ utils::retrieveFileInfo(_file, _mediaLibrary.rootDirectory) };
if (!fileInfo)
{
stats.skips++;
return;
}
db::Session& dbSession{ _db.getTLSSession() };
db::Track::pointer track{ db::Track::findByPath(dbSession, _file) };
if (!_parsedTrack)
{
if (track)
{
track.remove();
stats.deletions++;
}
context.stats.errors.emplace_back(_file, ScanErrorType::CannotReadAudioFile);
return;
}
if (_parsedTrack->mbid && (!track || _settings.skipDuplicateMBID))
{
std::vector<db::Track::pointer> duplicateTracks{ db::Track::findByMBID(dbSession, *_parsedTrack->mbid) };
// find for an existing track MBID as the file may have just been moved
if (!track && duplicateTracks.size() == 1)
{
db::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 (db::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 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 (_parsedTrack->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 (!_parsedTrack->title.empty())
title = _parsedTrack->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<db::Track>();
track.modify()->setAbsoluteFilePath(_file);
added = true;
}
// Track related data
assert(track);
// Audio properties
track.modify()->setBitrate(_parsedTrack->audioProperties.bitrate);
track.modify()->setBitsPerSample(_parsedTrack->audioProperties.bitsPerSample);
track.modify()->setChannelCount(_parsedTrack->audioProperties.channelCount);
track.modify()->setDuration(_parsedTrack->audioProperties.duration);
track.modify()->setSampleRate(_parsedTrack->audioProperties.sampleRate);
track.modify()->setRelativeFilePath(fileInfo->relativePath);
track.modify()->setFileSize(fileInfo->fileSize);
track.modify()->setLastWriteTime(fileInfo->lastWriteTime);
db::MediaLibrary::pointer mediaLibrary{ db::MediaLibrary::find(dbSession, _mediaLibrary.id) }; // may be null if settings are updated in // => next scan will correct this
track.modify()->setMediaLibrary(mediaLibrary);
db::Directory::pointer directory{ utils::getOrCreateDirectory(dbSession, _file.parent_path(), mediaLibrary) };
track.modify()->setDirectory(directory);
track.modify()->clearArtistLinks();
// Do not fallback on artists with the same name but having a MBID for artist and releaseArtists, as it may be corrected by properly tagging files
for (const db::Artist::pointer& artist : getOrCreateArtists(dbSession, _parsedTrack->artists, false))
track.modify()->addArtistLink(db::TrackArtistLink::create(dbSession, track, artist, db::TrackArtistLinkType::Artist));
if (_parsedTrack->medium && _parsedTrack->medium->release)
{
for (const db::Artist::pointer& releaseArtist : getOrCreateArtists(dbSession, _parsedTrack->medium->release->artists, false))
track.modify()->addArtistLink(db::TrackArtistLink::create(dbSession, track, releaseArtist, db::TrackArtistLinkType::ReleaseArtist));
}
// Allow fallbacks on artists with the same name even if they have MBID, since there is no tag to indicate the MBID of these artists
// We could ask MusicBrainz to get all the information, but that would heavily slow down the import process
for (const db::Artist::pointer& conductor : getOrCreateArtists(dbSession, _parsedTrack->conductorArtists, true))
track.modify()->addArtistLink(db::TrackArtistLink::create(dbSession, track, conductor, db::TrackArtistLinkType::Conductor));
for (const db::Artist::pointer& composer : getOrCreateArtists(dbSession, _parsedTrack->composerArtists, true))
track.modify()->addArtistLink(db::TrackArtistLink::create(dbSession, track, composer, db::TrackArtistLinkType::Composer));
for (const db::Artist::pointer& lyricist : getOrCreateArtists(dbSession, _parsedTrack->lyricistArtists, true))
track.modify()->addArtistLink(db::TrackArtistLink::create(dbSession, track, lyricist, db::TrackArtistLinkType::Lyricist));
for (const db::Artist::pointer& mixer : getOrCreateArtists(dbSession, _parsedTrack->mixerArtists, true))
track.modify()->addArtistLink(db::TrackArtistLink::create(dbSession, track, mixer, db::TrackArtistLinkType::Mixer));
for (const auto& [role, performers] : _parsedTrack->performerArtists)
{
for (const db::Artist::pointer& performer : getOrCreateArtists(dbSession, performers, true))
track.modify()->addArtistLink(db::TrackArtistLink::create(dbSession, track, performer, db::TrackArtistLinkType::Performer, role));
}
for (const db::Artist::pointer& producer : getOrCreateArtists(dbSession, _parsedTrack->producerArtists, true))
track.modify()->addArtistLink(db::TrackArtistLink::create(dbSession, track, producer, db::TrackArtistLinkType::Producer));
for (const db::Artist::pointer& remixer : getOrCreateArtists(dbSession, _parsedTrack->remixerArtists, true))
track.modify()->addArtistLink(db::TrackArtistLink::create(dbSession, track, remixer, db::TrackArtistLinkType::Remixer));
track.modify()->setScanVersion(_settings.scanVersion);
if (_parsedTrack->medium && _parsedTrack->medium->release)
track.modify()->setRelease(getOrCreateRelease(dbSession, *_parsedTrack->medium->release, directory));
else
track.modify()->setRelease({});
track.modify()->setTotalTrack(_parsedTrack->medium ? _parsedTrack->medium->trackCount : std::nullopt);
track.modify()->setReleaseReplayGain(_parsedTrack->medium ? _parsedTrack->medium->replayGain : std::nullopt);
track.modify()->setDiscSubtitle(_parsedTrack->medium ? _parsedTrack->medium->name : "");
track.modify()->setClusters(getOrCreateClusters(dbSession, *_parsedTrack));
track.modify()->setName(title);
track.modify()->setAddedTime(Wt::WDateTime::currentDateTime());
track.modify()->setTrackNumber(_parsedTrack->position);
track.modify()->setDiscNumber(_parsedTrack->medium ? _parsedTrack->medium->position : std::nullopt);
track.modify()->setDate(_parsedTrack->date);
track.modify()->setYear(_parsedTrack->year);
track.modify()->setOriginalDate(_parsedTrack->originalDate);
track.modify()->setOriginalYear(_parsedTrack->originalYear);
// If a file has an OriginalDate but no date, set it to ease filtering
if (!_parsedTrack->date.isValid() && _parsedTrack->originalDate.isValid())
track.modify()->setDate(_parsedTrack->originalDate);
// If a file has an OriginalYear but no Year, set it to ease filtering
if (!_parsedTrack->year && _parsedTrack->originalYear)
track.modify()->setYear(_parsedTrack->originalYear);
track.modify()->setRecordingMBID(_parsedTrack->recordingMBID);
track.modify()->setTrackMBID(_parsedTrack->mbid);
if (auto trackFeatures{ db::TrackFeatures::find(dbSession, track->getId()) })
trackFeatures.remove(); // TODO: only if MBID changed?
track.modify()->setHasCover(_parsedTrack->hasCover);
track.modify()->setCopyright(_parsedTrack->copyright);
track.modify()->setCopyrightURL(_parsedTrack->copyrightURL);
track.modify()->setComment(!_parsedTrack->comments.empty() ? _parsedTrack->comments.front() : ""); // only take the first one for now
track.modify()->setTrackReplayGain(_parsedTrack->replayGain);
track.modify()->setArtistDisplayName(_parsedTrack->artistDisplayName);
track.modify()->clearEmbeddedLyrics();
for (const metadata::Lyrics& lyricsInfo : _parsedTrack->lyrics)
{
db::TrackLyrics::pointer lyrics{ createLyrics(dbSession, lyricsInfo) };
track.modify()->addLyrics(lyrics);
}
if (added)
{
LMS_LOG(DBUPDATER, DEBUG, "Added audio file '" << _file.string() << "'");
stats.additions++;
}
else
{
LMS_LOG(DBUPDATER, DEBUG, "Updated audio file '" << _file.string() << "'");
stats.updates++;
}
}
metadata::ParserReadStyle getParserReadStyle()
{
std::string_view readStyle{ core::Service<core::IConfig>::get()->getString("scanner-parser-read-style", "average") };
if (readStyle == "fast")
return metadata::ParserReadStyle::Fast;
if (readStyle == "average")
return metadata::ParserReadStyle::Average;
if (readStyle == "accurate")
return metadata::ParserReadStyle::Accurate;
throw core::LmsException{ "Invalid value for 'scanner-parser-read-style'" };
}
} // namespace
AudioFileScanner::AudioFileScanner(db::Db& db, const ScannerSettings& settings)
: _db{ db }
, _settings{ settings }
, _metadataParser{ metadata::createParser(metadata::ParserBackend::TagLib, getParserReadStyle()) } // For now, always use TagLib
{
std::vector<std::string> 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);
}
AudioFileScanner::~AudioFileScanner() = default;
std::span<const std::filesystem::path> AudioFileScanner::getSupportedExtensions() const
{
return _metadataParser->getSupportedExtensions();
}
bool AudioFileScanner::needsScan(ScanContext& context, const FileToScan& file) const
{
ScanStats& stats{ context.stats };
const Wt::WDateTime lastWriteTime{ utils::retrieveFileGetLastWrite(file.file) };
// Should rarely fail as we are currently iterating it
if (!lastWriteTime.isValid())
{
stats.skips++;
return false;
}
if (context.scanOptions.fullScan)
return true;
bool needUpdateLibrary{};
db::Session& dbSession{ _db.getTLSSession() };
{
auto transaction{ dbSession.createReadTransaction() };
// Skip file if last write is the same
const db::Track::pointer track{ db::Track::findByPath(dbSession, file.file) };
if (track
&& track->getLastWriteTime() == lastWriteTime
&& 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() == file.mediaLibrary.id)
{
stats.skips++;
return false;
}
needUpdateLibrary = true;
}
}
if (needUpdateLibrary)
{
auto transaction{ dbSession.createWriteTransaction() };
db::Track::pointer track{ db::Track::findByPath(dbSession, file.file) };
assert(track);
track.modify()->setMediaLibrary(db::MediaLibrary::find(dbSession, file.mediaLibrary.id)); // may be null, will be handled in the next scan anyway
stats.updates++;
return false;
}
return true; // need to scan
}
std::unique_ptr<IFileScanOperation> AudioFileScanner::createScanOperation(const FileToScan& fileToScan) const
{
return std::make_unique<AudioFileScanOperation>(fileToScan, _db, *_metadataParser, _settings);
}
} // namespace lms::scanner
@@ -0,0 +1,62 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include <string>
#include <vector>
#include "IFileScanner.hpp"
namespace lms
{
namespace db
{
class Db;
}
namespace metadata
{
class IParser;
}
} // namespace lms
namespace lms::scanner
{
struct ScannerSettings;
class AudioFileScanner : public IFileScanner
{
public:
AudioFileScanner(db::Db& db, const ScannerSettings& settings);
~AudioFileScanner() override;
AudioFileScanner(const AudioFileScanner&) = delete;
AudioFileScanner& operator=(const AudioFileScanner&) = delete;
private:
std::span<const std::filesystem::path> getSupportedExtensions() const override;
bool needsScan(ScanContext& context, const FileToScan& file) const override;
std::unique_ptr<IFileScanOperation> createScanOperation(const FileToScan& fileToScan) const override;
db::Db& _db;
const ScannerSettings& _settings;
std::unique_ptr<metadata::IParser> _metadataParser;
const std::vector<std::string> _extraTagsToParse;
};
} // namespace lms::scanner
@@ -0,0 +1,37 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include "core/LiteralString.hpp"
namespace lms::scanner
{
class ScanContext;
class IFileScanOperation
{
public:
virtual ~IFileScanOperation() = default;
virtual core::LiteralString getName() const = 0;
virtual void scan() = 0;
virtual void processResult(ScanContext& context) = 0;
};
} // namespace lms::scanner
@@ -0,0 +1,48 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include <filesystem>
#include <span>
#include "MediaLibraryInfo.hpp"
namespace lms::scanner
{
class ScanContext;
class IFileScanOperation;
struct ScannerSettings;
struct FileToScan
{
std::filesystem::path file;
MediaLibraryInfo mediaLibrary;
};
class IFileScanner
{
public:
virtual ~IFileScanner() = default;
virtual std::span<const std::filesystem::path> getSupportedExtensions() const = 0;
virtual bool needsScan(ScanContext& context, const FileToScan& file) const = 0;
virtual std::unique_ptr<IFileScanOperation> createScanOperation(const FileToScan& fileToScan) const = 0;
};
} // namespace lms::scanner
@@ -0,0 +1,171 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#include "ImageFileScanner.hpp"
#include <optional>
#include "core/ILogger.hpp"
#include "database/Db.hpp"
#include "database/Image.hpp"
#include "database/MediaLibrary.hpp"
#include "database/Session.hpp"
#include "image/Exception.hpp"
#include "image/Image.hpp"
#include "IFileScanOperation.hpp"
#include "ScanContext.hpp"
#include "Utils.hpp"
namespace lms::scanner
{
namespace
{
class ImageFileScanOperation : public IFileScanOperation
{
public:
ImageFileScanOperation(const FileToScan& file, db::Db& db)
: _file{ file.file }
, _mediaLibrary{ file.mediaLibrary }
, _db{ db } {}
private:
core::LiteralString getName() const override { return "ScanImageFile"; }
void scan() override;
void processResult(ScanContext& context) override;
const std::filesystem::path _file;
const MediaLibraryInfo _mediaLibrary;
db::Db& _db;
struct ImageInfo
{
std::size_t height{};
std::size_t width{};
};
std::optional<ImageInfo> _parsedImageInfo;
};
void ImageFileScanOperation::scan()
{
try
{
std::unique_ptr<image::IRawImage> rawImage{ image::decodeImage(_file) };
ImageInfo& imageInfo{ _parsedImageInfo.emplace() };
imageInfo.width = rawImage->getWidth();
imageInfo.height = rawImage->getHeight();
}
catch (const image::Exception& e)
{
LMS_LOG(DBUPDATER, ERROR, "Cannot read image in file '" << _file.string() << "': " << e.what());
}
}
void ImageFileScanOperation::processResult(ScanContext& context)
{
ScanStats& stats{ context.stats };
const std::optional<FileInfo> fileInfo{ utils::retrieveFileInfo(_file, _mediaLibrary.rootDirectory) };
if (!fileInfo)
{
stats.skips++;
return;
}
db::Session& dbSession{ _db.getTLSSession() };
db::Image::pointer image{ db::Image::find(dbSession, _file) };
if (!_parsedImageInfo)
{
if (image)
{
image.remove();
stats.deletions++;
}
context.stats.errors.emplace_back(_file, ScanErrorType::CannotReadImageFile);
return;
}
const bool added{ !image };
if (!image)
image = dbSession.create<db::Image>(_file);
image.modify()->setLastWriteTime(fileInfo->lastWriteTime);
image.modify()->setFileSize(fileInfo->fileSize);
image.modify()->setHeight(_parsedImageInfo->height);
image.modify()->setWidth(_parsedImageInfo->width);
db::MediaLibrary::pointer mediaLibrary{ db::MediaLibrary::find(dbSession, _mediaLibrary.id) }; // may be null if settings are updated in // => next scan will correct this
image.modify()->setDirectory(utils::getOrCreateDirectory(dbSession, _file.parent_path(), mediaLibrary));
if (added)
{
LMS_LOG(DBUPDATER, DEBUG, "Added image '" << _file.string() << "'");
stats.additions++;
}
else
{
LMS_LOG(DBUPDATER, DEBUG, "Updated image '" << _file.string() << "'");
stats.updates++;
}
}
} // namespace
ImageFileScanner::ImageFileScanner(db::Db& db)
: _db{ db }
{
}
std::span<const std::filesystem::path> ImageFileScanner::getSupportedExtensions() const
{
return image::getSupportedFileExtensions();
}
bool ImageFileScanner::needsScan(ScanContext& context, const FileToScan& file) const
{
ScanStats& stats{ context.stats };
const Wt::WDateTime lastWriteTime{ utils::retrieveFileGetLastWrite(file.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.file) };
if (image && image->getLastWriteTime() == lastWriteTime)
{
stats.skips++;
return false;
}
}
return true; // need to scan
}
std::unique_ptr<IFileScanOperation> ImageFileScanner::createScanOperation(const FileToScan& fileToScan) const
{
return std::make_unique<ImageFileScanOperation>(fileToScan, _db);
}
} // namespace lms::scanner
@@ -0,0 +1,51 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include "IFileScanner.hpp"
namespace lms
{
namespace db
{
class Db;
}
} // namespace lms
namespace lms::scanner
{
struct ScannerSettings;
class ImageFileScanner : public IFileScanner
{
public:
ImageFileScanner(db::Db& db);
~ImageFileScanner() override = default;
ImageFileScanner(const ImageFileScanner&) = delete;
ImageFileScanner& operator=(const ImageFileScanner&) = delete;
private:
std::span<const std::filesystem::path> getSupportedExtensions() const override;
bool needsScan(ScanContext& context, const FileToScan& file) const override;
std::unique_ptr<IFileScanOperation> createScanOperation(const FileToScan& fileToScan) const override;
db::Db& _db;
};
} // namespace lms::scanner
@@ -0,0 +1,177 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#include "LyricsFileScanner.hpp"
#include <fstream>
#include <optional>
#include "core/ILogger.hpp"
#include "database/Db.hpp"
#include "database/MediaLibrary.hpp"
#include "database/Session.hpp"
#include "database/TrackLyrics.hpp"
#include "metadata/Lyrics.hpp"
#include "IFileScanOperation.hpp"
#include "ScanContext.hpp"
#include "Utils.hpp"
namespace lms::scanner
{
namespace
{
class LyricsFileScanOperation : public IFileScanOperation
{
public:
LyricsFileScanOperation(const FileToScan& file, db::Db& db)
: _file{ file.file }
, _mediaLibrary{ file.mediaLibrary }
, _db{ db } {}
private:
core::LiteralString getName() const override { return "ScanLyricsFile"; }
void scan() override;
void processResult(ScanContext& context) override;
const std::filesystem::path _file;
const MediaLibraryInfo _mediaLibrary;
db::Db& _db;
std::optional<metadata::Lyrics> _parsedLyrics;
};
void LyricsFileScanOperation::scan()
{
try
{
std::ifstream ifs{ _file.string() };
if (!ifs)
LMS_LOG(DBUPDATER, ERROR, "Cannot open file '" << _file.string() << "'");
else
_parsedLyrics = metadata::parseLyrics(ifs);
}
catch (const metadata::Exception& e)
{
LMS_LOG(DBUPDATER, ERROR, "Cannot read lyrics in file '" << _file.string() << "': " << e.what());
}
}
void LyricsFileScanOperation::processResult(ScanContext& context)
{
ScanStats& stats{ context.stats };
const std::optional<FileInfo> fileInfo{ utils::retrieveFileInfo(_file, _mediaLibrary.rootDirectory) };
if (!fileInfo)
{
stats.skips++;
return;
}
db::Session& dbSession{ _db.getTLSSession() };
db::TrackLyrics::pointer trackLyrics{ db::TrackLyrics::find(dbSession, _file) };
if (!_parsedLyrics)
{
if (trackLyrics)
{
trackLyrics.remove();
stats.deletions++;
}
context.stats.errors.emplace_back(_file, ScanErrorType::CannotReadLyricsFile);
return;
}
const bool added{ !trackLyrics };
if (!trackLyrics)
{
trackLyrics = dbSession.create<db::TrackLyrics>();
trackLyrics.modify()->setAbsoluteFilePath(_file);
}
trackLyrics.modify()->setLastWriteTime(fileInfo->lastWriteTime);
trackLyrics.modify()->setFileSize(fileInfo->fileSize);
trackLyrics.modify()->setLanguage(!_parsedLyrics->language.empty() ? _parsedLyrics->language : "xxx");
trackLyrics.modify()->setOffset(_parsedLyrics->offset);
trackLyrics.modify()->setDisplayTitle(_parsedLyrics->displayTitle);
trackLyrics.modify()->setDisplayArtist(_parsedLyrics->displayArtist);
if (!_parsedLyrics->synchronizedLines.empty())
trackLyrics.modify()->setSynchronizedLines(_parsedLyrics->synchronizedLines);
else
trackLyrics.modify()->setUnsynchronizedLines(_parsedLyrics->unsynchronizedLines);
db::MediaLibrary::pointer mediaLibrary{ db::MediaLibrary::find(dbSession, _mediaLibrary.id) }; // may be null if settings are updated in // => next scan will correct this
trackLyrics.modify()->setDirectory(utils::getOrCreateDirectory(dbSession, _file.parent_path(), mediaLibrary));
if (added)
{
LMS_LOG(DBUPDATER, DEBUG, "Added external lyrics '" << _file.string() << "'");
stats.additions++;
}
else
{
LMS_LOG(DBUPDATER, DEBUG, "Updated external lyrics '" << _file.string() << "'");
stats.updates++;
}
}
} // namespace
LyricsFileScanner::LyricsFileScanner(db::Db& db)
: _db{ db }
{
}
std::span<const std::filesystem::path> LyricsFileScanner::getSupportedExtensions() const
{
return metadata::getSupportedLyricsFileExtensions();
}
bool LyricsFileScanner::needsScan(ScanContext& context, const FileToScan& file) const
{
ScanStats& stats{ context.stats };
const Wt::WDateTime lastWriteTime{ utils::retrieveFileGetLastWrite(file.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::TrackLyrics::pointer lyrics{ db::TrackLyrics::find(dbSession, file.file) };
if (lyrics && lyrics->getLastWriteTime() == lastWriteTime)
{
stats.skips++;
return false;
}
}
return true; // need to scan
}
std::unique_ptr<IFileScanOperation> LyricsFileScanner::createScanOperation(const FileToScan& fileToScan) const
{
return std::make_unique<LyricsFileScanOperation>(fileToScan, _db);
}
} // namespace lms::scanner
@@ -0,0 +1,51 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include "IFileScanner.hpp"
namespace lms
{
namespace db
{
class Db;
}
} // namespace lms
namespace lms::scanner
{
struct ScannerSettings;
class LyricsFileScanner : public IFileScanner
{
public:
LyricsFileScanner(db::Db& db);
~LyricsFileScanner() override = default;
LyricsFileScanner(const LyricsFileScanner&) = delete;
LyricsFileScanner& operator=(const LyricsFileScanner&) = delete;
private:
std::span<const std::filesystem::path> getSupportedExtensions() const override;
bool needsScan(ScanContext& context, const FileToScan& file) const override;
std::unique_ptr<IFileScanOperation> createScanOperation(const FileToScan& fileToScan) const override;
db::Db& _db;
};
} // namespace lms::scanner
@@ -0,0 +1,176 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#include "PlayListFileScanner.hpp"
#include <fstream>
#include <optional>
#include "core/ILogger.hpp"
#include "database/Db.hpp"
#include "database/MediaLibrary.hpp"
#include "database/PlayListFile.hpp"
#include "database/Session.hpp"
#include "metadata/Exception.hpp"
#include "metadata/PlayList.hpp"
#include "IFileScanOperation.hpp"
#include "ScanContext.hpp"
#include "Utils.hpp"
namespace lms::scanner
{
namespace
{
class PlayListFileScanOperation : public IFileScanOperation
{
public:
PlayListFileScanOperation(const FileToScan& file, db::Db& db)
: _file{ file.file }
, _mediaLibrary{ file.mediaLibrary }
, _db{ db } {}
~PlayListFileScanOperation() override = default;
PlayListFileScanOperation(const PlayListFileScanOperation&) = delete;
PlayListFileScanOperation& operator=(const PlayListFileScanOperation&) = delete;
private:
core::LiteralString getName() const override { return "ScanPlayListFile"; }
void scan() override;
void processResult(ScanContext& context) override;
const std::filesystem::path _file;
const MediaLibraryInfo _mediaLibrary;
db::Db& _db;
std::optional<metadata::PlayList> _parsedPlayList;
};
void PlayListFileScanOperation::scan()
{
try
{
std::ifstream ifs{ _file.string() };
if (!ifs)
LMS_LOG(DBUPDATER, ERROR, "Cannot open file '" << _file.string() << "'");
else
_parsedPlayList = metadata::parsePlayList(ifs);
}
catch (const metadata::Exception& e)
{
LMS_LOG(DBUPDATER, ERROR, "Cannot read playlist in file '" << _file.string() << "': " << e.what());
}
}
void PlayListFileScanOperation::processResult(ScanContext& context)
{
ScanStats& stats{ context.stats };
const std::optional<FileInfo> fileInfo{ utils::retrieveFileInfo(_file, _mediaLibrary.rootDirectory) };
if (!fileInfo)
{
stats.skips++;
return;
}
db::Session& dbSession{ _db.getTLSSession() };
db::PlayListFile::pointer playList{ db::PlayListFile::find(dbSession, _file) };
if (!_parsedPlayList)
{
if (playList)
{
playList.remove();
stats.deletions++;
}
context.stats.errors.emplace_back(_file, ScanErrorType::CannotReadPlayListFile);
return;
}
const bool added{ !playList };
if (!playList)
playList = dbSession.create<db::PlayListFile>(_file);
playList.modify()->setLastWriteTime(fileInfo->lastWriteTime);
playList.modify()->setFileSize(fileInfo->fileSize);
if (!_parsedPlayList->name.empty())
playList.modify()->setName(_parsedPlayList->name);
else
playList.modify()->setName(_file.stem().string());
playList.modify()->setFiles(_parsedPlayList->files);
db::MediaLibrary::pointer mediaLibrary{ db::MediaLibrary::find(dbSession, _mediaLibrary.id) }; // may be null if settings are updated in // => next scan will correct this
playList.modify()->setDirectory(utils::getOrCreateDirectory(dbSession, _file.parent_path(), mediaLibrary));
if (added)
{
LMS_LOG(DBUPDATER, DEBUG, "Added playlist file '" << _file.string() << "'");
LMS_LOG(DBUPDATER, DEBUG, "db playlist file = '" << playList->getAbsoluteFilePath().string() << "'");
stats.additions++;
}
else
{
LMS_LOG(DBUPDATER, DEBUG, "Updated playlist file '" << _file.string() << "'");
stats.updates++;
}
}
} // namespace
PlayListFileScanner::PlayListFileScanner(db::Db& db)
: _db{ db }
{
}
std::span<const std::filesystem::path> PlayListFileScanner::getSupportedExtensions() const
{
return metadata::getSupportedPlayListFileExtensions();
}
bool PlayListFileScanner::needsScan(ScanContext& context, const FileToScan& file) const
{
ScanStats& stats{ context.stats };
const Wt::WDateTime lastWriteTime{ utils::retrieveFileGetLastWrite(file.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::PlayListFile::pointer playList{ db::PlayListFile::find(dbSession, file.file) };
if (playList && playList->getLastWriteTime() == lastWriteTime)
{
stats.skips++;
return false;
}
}
return true; // need to scan
}
std::unique_ptr<IFileScanOperation> PlayListFileScanner::createScanOperation(const FileToScan& fileToScan) const
{
return std::make_unique<PlayListFileScanOperation>(fileToScan, _db);
}
} // namespace lms::scanner
@@ -0,0 +1,51 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include "IFileScanner.hpp"
namespace lms
{
namespace db
{
class Db;
}
} // namespace lms
namespace lms::scanner
{
struct ScannerSettings;
class PlayListFileScanner : public IFileScanner
{
public:
PlayListFileScanner(db::Db& db);
~PlayListFileScanner() override = default;
PlayListFileScanner(const PlayListFileScanner&) = delete;
PlayListFileScanner& operator=(const PlayListFileScanner&) = delete;
private:
std::span<const std::filesystem::path> getSupportedExtensions() const override;
bool needsScan(ScanContext& context, const FileToScan& file) const override;
std::unique_ptr<IFileScanOperation> createScanOperation(const FileToScan& fileToScan) const override;
db::Db& _db;
};
} // namespace lms::scanner
@@ -0,0 +1,100 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#include "Utils.hpp"
#include "core/ILogger.hpp"
#include "core/Path.hpp"
#include "database/Directory.hpp"
#include "database/MediaLibrary.hpp"
#include "database/Session.hpp"
namespace lms::scanner::utils
{
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<FileInfo> retrieveFileInfo(const std::filesystem::path& file, const std::filesystem::path& rootPath)
{
std::optional<FileInfo> 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;
}
db::Directory::pointer getOrCreateDirectory(db::Session& session, const std::filesystem::path& path, const db::MediaLibrary::pointer& mediaLibrary)
{
db::Directory::pointer directory{ db::Directory::find(session, path) };
if (!directory)
{
db::Directory::pointer parentDirectory;
if (path != mediaLibrary->getPath())
parentDirectory = getOrCreateDirectory(session, path.parent_path(), mediaLibrary);
directory = session.create<db::Directory>(path);
directory.modify()->setParent(parentDirectory);
directory.modify()->setMediaLibrary(mediaLibrary);
}
// Don't update library if it does not match, will be updated elsewhere
return directory;
}
} // namespace lms::scanner::utils
@@ -0,0 +1,52 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include <filesystem>
#include <optional>
#include <Wt/WDateTime.h>
#include "database/Object.hpp"
namespace lms::db
{
class Directory;
class MediaLibrary;
class Session;
} // namespace lms::db
namespace lms::scanner
{
struct FileInfo
{
Wt::WDateTime lastWriteTime;
std::filesystem::path relativePath;
std::size_t fileSize{};
};
namespace utils
{
Wt::WDateTime retrieveFileGetLastWrite(const std::filesystem::path& file);
std::optional<FileInfo> retrieveFileInfo(const std::filesystem::path& file, const std::filesystem::path& rootPath);
db::ObjectPtr<db::Directory> getOrCreateDirectory(db::Session& session, const std::filesystem::path& path, const db::ObjectPtr<db::MediaLibrary>& mediaLibrary);
} // namespace utils
} // namespace lms::scanner
@@ -0,0 +1,109 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#include "FileScanQueue.hpp"
#include <boost/asio/post.hpp>
#include "core/ITraceLogger.hpp"
#include "scanners/IFileScanOperation.hpp"
namespace lms::scanner
{
FileScanQueue::FileScanQueue(std::size_t threadCount, bool& abort)
: _scanContextRunner{ _scanIoContext, threadCount, "FileScan" }
, _abort{ abort }
{
}
FileScanQueue::~FileScanQueue() = default;
void FileScanQueue::pushScanRequest(std::unique_ptr<IFileScanOperation> operation)
{
{
std::scoped_lock lock{ _mutex };
_ongoingScanCount += 1;
}
auto operationHandler{ [operation = std::move(operation), this]() mutable {
if (_abort)
{
std::scoped_lock lock{ _mutex };
_ongoingScanCount -= 1;
}
else
{
{
LMS_SCOPED_TRACE_OVERVIEW("Scanner", operation->getName());
operation->scan();
}
{
std::scoped_lock lock{ _mutex };
_scanResults.emplace_back(std::move(operation));
_ongoingScanCount -= 1;
}
}
_condVar.notify_all();
} };
boost::asio::post(_scanIoContext, std::move(operationHandler));
}
std::size_t FileScanQueue::getResultsCount() const
{
std::scoped_lock lock{ _mutex };
return _scanResults.size();
}
size_t FileScanQueue::popResults(std::vector<std::unique_ptr<IFileScanOperation>>& results, std::size_t maxCount)
{
results.clear();
results.reserve(maxCount);
{
std::scoped_lock lock{ _mutex };
while (results.size() < maxCount && !_scanResults.empty())
{
results.push_back(std::move(_scanResults.front()));
_scanResults.pop_front();
}
}
return results.size();
}
void FileScanQueue::wait(std::size_t maxScanRequestCount)
{
if (_ongoingScanCount <= maxScanRequestCount)
return;
{
LMS_SCOPED_TRACE_OVERVIEW("Scanner", "WaitParseResults");
std::unique_lock lock{ _mutex };
_condVar.wait(lock, [=, this] { return _ongoingScanCount <= maxScanRequestCount; });
}
}
} // namespace lms::scanner
@@ -0,0 +1,62 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include <atomic>
#include <condition_variable>
#include <deque>
#include <memory>
#include <mutex>
#include <vector>
#include "core/IOContextRunner.hpp"
namespace lms::scanner
{
class IFileScanOperation;
class FileScanQueue
{
public:
FileScanQueue(std::size_t threadCount, bool& abort);
~FileScanQueue();
FileScanQueue(const FileScanQueue&) = delete;
FileScanQueue& operator=(const FileScanQueue&) = delete;
std::size_t getThreadCount() const { return _scanContextRunner.getThreadCount(); }
void pushScanRequest(std::unique_ptr<IFileScanOperation> operation);
std::size_t getResultsCount() const;
size_t popResults(std::vector<std::unique_ptr<IFileScanOperation>>& results, std::size_t maxCount);
void wait(std::size_t maxScanRequestCount = 0); // wait until ongoing scan request count <= maxScanRequestCount
private:
boost::asio::io_context _scanIoContext;
core::IOContextRunner _scanContextRunner;
mutable std::mutex _mutex;
std::atomic<std::size_t> _ongoingScanCount{};
std::deque<std::unique_ptr<IFileScanOperation>> _scanResults;
std::condition_variable _condVar;
bool& _abort;
};
} // namespace lms::scanner
@@ -20,8 +20,8 @@
#pragma once
#include "core/LiteralString.hpp"
#include "services/scanner/ScannerOptions.hpp"
#include "services/scanner/ScannerStats.hpp"
#include "ScanContext.hpp"
namespace lms::scanner
{
@@ -32,13 +32,6 @@ namespace lms::scanner
virtual ScanStep getStep() const = 0;
virtual core::LiteralString getStepName() const = 0;
struct ScanContext
{
ScanOptions scanOptions;
ScanStats stats;
ScanStepStats currentStepStats;
};
virtual void process(ScanContext& context) = 0;
};
} // namespace lms::scanner
@@ -59,9 +59,9 @@ namespace lms::scanner
assert(!lyrics->getFileStem().empty());
params.setDirectory(lyrics->getDirectory()->getId());
params.setStem(stem);
params.setFileStem(stem);
db::Track::find(session, params, [&](const db::Track::pointer track) {
db::Track::find(session, params, [&](const db::Track::pointer& track) {
if (matchingTrack)
LMS_LOG(DBUPDATER, DEBUG, "External lyrics '" << lyrics->getAbsoluteFilePath() << "' already matched with '" << matchingTrack->getAbsoluteFilePath() << "', replaced by '" << track->getAbsoluteFilePath() << "'");
@@ -0,0 +1,202 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#include "ScanStepAssociatePlayListTracks.hpp"
#include <deque>
#include <filesystem>
#include "core/ILogger.hpp"
#include "database/Db.hpp"
#include "database/Directory.hpp"
#include "database/PlayListFile.hpp"
#include "database/Session.hpp"
#include "database/Track.hpp"
#include "database/TrackList.hpp"
namespace lms::scanner
{
namespace
{
constexpr std::size_t readBatchSize{ 20 };
constexpr std::size_t writeBatchSize{ 5 };
struct PlayListFileAssociation
{
db::PlayListFileId playListFileIdId;
std::vector<db::TrackId> trackIds;
};
using PlayListFileAssociationContainer = std::deque<PlayListFileAssociation>;
struct SearchPlayListFileContext
{
db::Session& session;
db::PlayListFileId lastRetrievedPlayListFileId;
std::size_t processedPlayListFileCount{};
};
db::Track::pointer getMatchingTrack(db::Session& session, const std::filesystem::path& filePath, const db::Directory::pointer& playListDirectory)
{
db::Track::pointer matchingTrack;
if (filePath.is_absolute())
{
matchingTrack = db::Track::findByPath(session, filePath);
}
else
{
const std::filesystem::path absolutePath{ playListDirectory->getAbsolutePath() / filePath };
matchingTrack = db::Track::findByPath(session, absolutePath.lexically_normal());
}
return matchingTrack;
}
bool trackListNeedsUpdate(db::Session& session, std::string_view name, std::span<const db::TrackId> trackIds, const db::TrackList::pointer& trackList)
{
if (trackList->getName() != name)
return true;
db::TrackListEntry::FindParameters params;
params.setTrackList(trackList->getId());
bool needUpdate{};
std::size_t currentIndex{};
db::TrackListEntry::find(session, params, [&](const db::TrackListEntry::pointer& entry) {
if (currentIndex > trackIds.size() || trackIds[currentIndex] != entry->getTrackId())
needUpdate = true;
currentIndex += 1;
});
if (currentIndex != trackIds.size())
needUpdate = true;
return needUpdate;
}
bool fetchNextPlayListFilesToUpdate(SearchPlayListFileContext& searchContext, PlayListFileAssociationContainer& playListFileAssociations)
{
const db::PlayListFileId playListFileIdId{ searchContext.lastRetrievedPlayListFileId };
{
auto transaction{ searchContext.session.createReadTransaction() };
db::PlayListFile::find(searchContext.session, searchContext.lastRetrievedPlayListFileId, readBatchSize, [&](const db::PlayListFile::pointer& playListFile) {
PlayListFileAssociation playListAssociation;
playListAssociation.playListFileIdId = playListFile->getId();
const auto files{ playListFile->getFiles() };
for (const std::filesystem::path& file : files)
{
// TODO optim: no need to fetch the whole track
db::Track::pointer track{ getMatchingTrack(searchContext.session, file, playListFile->getDirectory()) };
if (track)
playListAssociation.trackIds.push_back(track->getId());
else
LMS_LOG(DBUPDATER, DEBUG, "Track '" << file.string() << "' not found in playlist '" << playListFile->getAbsoluteFilePath().string() << "'");
}
bool needUpdate{ true };
if (const db::TrackList::pointer trackList{ playListFile->getTrackList() })
needUpdate = trackListNeedsUpdate(searchContext.session, playListFile->getName(), playListAssociation.trackIds, trackList);
if (needUpdate)
{
LMS_LOG(DBUPDATER, DEBUG, "Updating PlayList '" << playListFile->getAbsoluteFilePath().string() << "' (" << playListAssociation.trackIds.size() << " files)");
playListFileAssociations.emplace_back(std::move(playListAssociation));
}
searchContext.processedPlayListFileCount++;
});
}
return playListFileIdId != searchContext.lastRetrievedPlayListFileId;
}
void updatePlayListFile(db::Session& session, const PlayListFileAssociation& playListFileAssociation)
{
db::PlayListFile::pointer playListFile{ db::PlayListFile::find(session, playListFileAssociation.playListFileIdId) };
assert(playListFile);
db::TrackList::pointer trackList{ playListFile->getTrackList() };
if (!trackList)
{
trackList = session.create<db::TrackList>(playListFile->getName(), db::TrackListType::PlayList);
playListFile.modify()->setTrackList(trackList);
}
trackList.modify()->setVisibility(db::TrackList::Visibility::Public);
trackList.modify()->setLastModifiedDateTime(playListFile->getLastWriteTime());
trackList.modify()->setName(playListFile->getName());
trackList.modify()->clear();
for (const db::TrackId trackId : playListFileAssociation.trackIds)
{
if (db::Track::pointer track{ db::Track::find(session, trackId) })
session.create<db::TrackListEntry>(track, trackList, playListFile->getLastWriteTime());
}
}
void updatePlayListFiles(db::Session& session, PlayListFileAssociationContainer& playListFileAssociations)
{
while (!playListFileAssociations.empty())
{
auto transaction{ session.createWriteTransaction() };
for (std::size_t i{}; !playListFileAssociations.empty() && i < writeBatchSize; ++i)
{
updatePlayListFile(session, playListFileAssociations.front());
playListFileAssociations.pop_front();
}
}
}
} // namespace
void ScanStepAssociatePlayListTracks::process(ScanContext& context)
{
if (_abortScan)
return;
if (context.stats.nbChanges() == 0)
return;
auto& session{ _db.getTLSSession() };
{
auto transaction{ session.createReadTransaction() };
context.currentStepStats.totalElems = db::PlayListFile::getCount(session);
}
SearchPlayListFileContext searchContext{
.session = session,
.lastRetrievedPlayListFileId = {},
};
PlayListFileAssociationContainer playListFileAssociations;
while (fetchNextPlayListFilesToUpdate(searchContext, playListFileAssociations))
{
if (_abortScan)
return;
updatePlayListFiles(session, playListFileAssociations);
context.currentStepStats.processedElems = searchContext.processedPlayListFileCount;
_progressCallback(context.currentStepStats);
}
}
} // namespace lms::scanner
@@ -0,0 +1,36 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#pragma once
#include "ScanStepBase.hpp"
namespace lms::scanner
{
class ScanStepAssociatePlayListTracks : public ScanStepBase
{
public:
using ScanStepBase::ScanStepBase;
private:
ScanStep getStep() const override { return ScanStep::AssociatePlayListTracks; }
core::LiteralString getStepName() const override { return "Associate playlist tracks"; }
void process(ScanContext& context) override;
};
} // namespace lms::scanner
@@ -20,11 +20,10 @@
#pragma once
#include <functional>
#include "services/scanner/ScannerStats.hpp"
#include <span>
#include <vector>
#include "IScanStep.hpp"
#include "ScannerSettings.hpp"
namespace lms::db
{
@@ -33,10 +32,13 @@ namespace lms::db
namespace lms::scanner
{
class IFileScanner;
struct ScannerSettings;
struct ScanStepStats;
class ScanStepBase : public IScanStep
{
public:
static inline const std::filesystem::path excludeDirFileName{ ".lmsignore" };
using ProgressCallback = std::function<void(const ScanStepStats& stats)>;
struct InitParams
@@ -45,21 +47,26 @@ namespace lms::scanner
ProgressCallback progressCallback;
bool& abortScan;
db::Db& db;
std::span<IFileScanner*> fileScanners;
};
ScanStepBase(InitParams& initParams)
: _settings{ initParams.settings }
, _progressCallback{ initParams.progressCallback }
, _abortScan{ initParams.abortScan }
, _db{ initParams.db }
, _fileScanners(std::cbegin(initParams.fileScanners), std::cend(initParams.fileScanners))
{
}
protected:
~ScanStepBase() override = default;
ScanStepBase(const ScanStepBase&) = delete;
ScanStepBase& operator=(const ScanStepBase&) = delete;
const ScannerSettings& _settings;
ProgressCallback _progressCallback;
bool& _abortScan;
db::Db& _db;
std::vector<IFileScanner*> _fileScanners;
};
} // namespace lms::scanner
@@ -19,13 +19,18 @@
#include "ScanStepCheckForRemovedFiles.hpp"
#include <vector>
#include "ScannerSettings.hpp"
#include "core/ILogger.hpp"
#include "core/Path.hpp"
#include "database/Db.hpp"
#include "database/Image.hpp"
#include "database/PlayListFile.hpp"
#include "database/Session.hpp"
#include "database/Track.hpp"
#include "database/TrackLyrics.hpp"
#include "scanners/IFileScanner.hpp"
namespace lms::scanner
{
@@ -47,12 +52,21 @@ namespace lms::scanner
context.currentStepStats.totalElems += db::Track::getCount(session);
context.currentStepStats.totalElems += db::Image::getCount(session);
context.currentStepStats.totalElems += db::TrackLyrics::getExternalLyricsCount(session);
context.currentStepStats.totalElems += db::PlayListFile::getCount(session);
}
LMS_LOG(DBUPDATER, DEBUG, context.currentStepStats.totalElems << " files to be checked...");
checkForRemovedFiles<db::Track>(context, _settings.supportedAudioFileExtensions);
checkForRemovedFiles<db::Image>(context, _settings.supportedImageFileExtensions);
checkForRemovedFiles<db::TrackLyrics>(context, _settings.supportedLyricsFileExtensions);
std::vector<std::filesystem::path> supportedFileExtensions;
for (IFileScanner* scanner : _fileScanners)
{
for (const std::filesystem::path& extension : scanner->getSupportedExtensions())
supportedFileExtensions.emplace_back(extension);
}
checkForRemovedFiles<db::Track>(context, supportedFileExtensions);
checkForRemovedFiles<db::Image>(context, supportedFileExtensions);
checkForRemovedFiles<db::TrackLyrics>(context, supportedFileExtensions);
checkForRemovedFiles<db::PlayListFile>(context, supportedFileExtensions);
}
template<typename Object>
@@ -124,7 +138,7 @@ namespace lms::scanner
}
if (std::none_of(std::cbegin(_settings.mediaLibraries), std::cend(_settings.mediaLibraries),
[&](const ScannerSettings::MediaLibraryInfo& libraryInfo) {
[&](const MediaLibraryInfo& libraryInfo) {
return core::pathUtils::isPathInRootPath(p, libraryInfo.rootDirectory, &excludeDirFileName);
}))
{
@@ -22,21 +22,24 @@
#include "core/ILogger.hpp"
#include "core/Path.hpp"
#include "MediaLibraryInfo.hpp"
#include "ScannerSettings.hpp"
#include "scanners/IFileScanner.hpp"
namespace lms::scanner
{
void ScanStepDiscoverFiles::process(ScanContext& context)
{
context.stats.totalFileCount = 0;
std::vector<std::filesystem::path> supportedExtensions;
for (const auto& extension : _settings.supportedAudioFileExtensions)
supportedExtensions.emplace_back(extension);
for (const auto& extension : _settings.supportedImageFileExtensions)
supportedExtensions.emplace_back(extension);
for (const auto& extension : _settings.supportedLyricsFileExtensions)
supportedExtensions.emplace_back(extension);
std::vector<std::filesystem::path> supportedFileExtensions;
for (IFileScanner* scanner : _fileScanners)
{
for (const std::filesystem::path& extension : scanner->getSupportedExtensions())
supportedFileExtensions.emplace_back(extension);
}
for (const ScannerSettings::MediaLibraryInfo& mediaLibrary : _settings.mediaLibraries)
for (const MediaLibraryInfo& mediaLibrary : _settings.mediaLibraries)
{
std::size_t currentDirectoryProcessElemsCount{};
core::pathUtils::exploreFilesRecursive(
@@ -44,7 +47,7 @@ namespace lms::scanner
if (_abortScan)
return false;
if (!ec && core::pathUtils::hasFileAnyExtension(path, supportedExtensions))
if (!ec && core::pathUtils::hasFileAnyExtension(path, supportedFileExtensions))
{
context.currentStepStats.processedElems++;
currentDirectoryProcessElemsCount++;
@@ -29,7 +29,7 @@ namespace lms::scanner
{
ScanStats& stats{ context.stats };
if (context.scanOptions.forceOptimize || (stats.nbChanges() > (stats.nbFiles() / 5)))
if (context.scanOptions.forceOptimize || (stats.nbChanges() > (stats.nbFiles() / 10)))
{
LMS_LOG(DBUPDATER, INFO, "Database analyze started");
@@ -84,7 +84,7 @@ namespace lms::scanner
}
template<typename T>
void ScanStepRemoveOrphanedDbEntries::removeOrphanedEntries(ScanStepRemoveOrphanedDbEntries::ScanContext& context)
void ScanStepRemoveOrphanedDbEntries::removeOrphanedEntries(ScanContext& context)
{
constexpr std::size_t batchSize = 100;
@@ -42,6 +42,6 @@ namespace lms::scanner
void removeOrphanedDirectories(ScanContext& context);
template<typename T>
void removeOrphanedEntries(ScanStepRemoveOrphanedDbEntries::ScanContext& context);
void removeOrphanedEntries(ScanContext& context);
};
} // namespace lms::scanner
@@ -0,0 +1,146 @@
/*
* Copyright (C) 2024 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 <http://www.gnu.org/licenses/>.
*/
#include "ScanStepScanFiles.hpp"
#include "FileScanQueue.hpp"
#include "ScannerSettings.hpp"
#include "core/IConfig.hpp"
#include "core/ILogger.hpp"
#include "core/ITraceLogger.hpp"
#include "core/Path.hpp"
#include "database/Db.hpp"
#include "database/Session.hpp"
#include "scanners/IFileScanOperation.hpp"
#include "scanners/IFileScanner.hpp"
namespace lms::scanner
{
using namespace db;
namespace
{
std::size_t getScanMetaDataThreadCount()
{
std::size_t threadCount{ core::Service<core::IConfig>::get()->getULong("scanner-metadata-thread-count", 0) };
if (threadCount == 0)
threadCount = std::max<std::size_t>(std::thread::hardware_concurrency() / 2, 1);
return threadCount;
}
} // namespace
ScanStepScanFiles::ScanStepScanFiles(InitParams& initParams)
: ScanStepBase{ initParams }
, _fileScanQueue{ getScanMetaDataThreadCount(), _abortScan }
{
for (IFileScanner* scanner : _fileScanners)
{
for (const std::filesystem::path& extension : scanner->getSupportedExtensions())
{
[[maybe_unused]] auto [it, inserted]{ _scannerByExtension.emplace(extension, scanner) };
assert(inserted);
}
}
LMS_LOG(DBUPDATER, INFO, "Using " << _fileScanQueue.getThreadCount() << " thread(s) for scanning file metadata");
}
void ScanStepScanFiles::process(ScanContext& context)
{
context.currentStepStats.totalElems = context.stats.totalFileCount;
for (const MediaLibraryInfo& mediaLibrary : _settings.mediaLibraries)
process(context, mediaLibrary);
}
void ScanStepScanFiles::process(ScanContext& context, const MediaLibraryInfo& mediaLibrary)
{
const std::size_t scanQueueMaxScanRequestCount{ 100 * _fileScanQueue.getThreadCount() };
const std::size_t processFileResultsBatchSize{ 5 };
std::vector<std::unique_ptr<IFileScanOperation>> scanOperations;
core::pathUtils::exploreFilesRecursive(
mediaLibrary.rootDirectory, [&](std::error_code ec, const std::filesystem::path& path) {
LMS_SCOPED_TRACE_DETAILED("Scanner", "OnExploreFile");
if (_abortScan)
return false; // stop iterating
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
{
auto itScanner{ _scannerByExtension.find(core::stringUtils::stringToLower(path.extension().string())) };
if (itScanner != std::cend(_scannerByExtension))
{
IFileScanner& scanner{ *itScanner->second };
FileToScan fileToScan{ .file = path, .mediaLibrary = mediaLibrary };
if (scanner.needsScan(context, fileToScan))
{
auto scanOperation{ scanner.createScanOperation(fileToScan) };
_fileScanQueue.pushScanRequest(std::move(scanOperation));
}
context.currentStepStats.processedElems++;
_progressCallback(context.currentStepStats);
}
}
while (_fileScanQueue.getResultsCount() > (scanQueueMaxScanRequestCount / 2))
{
_fileScanQueue.popResults(scanOperations, processFileResultsBatchSize);
processFileScanResults(context, scanOperations);
}
_fileScanQueue.wait(scanQueueMaxScanRequestCount);
return true;
},
&excludeDirFileName);
_fileScanQueue.wait();
while (!_abortScan && _fileScanQueue.popResults(scanOperations, processFileResultsBatchSize) > 0)
processFileScanResults(context, scanOperations);
}
void ScanStepScanFiles::processFileScanResults(ScanContext& context, std::span<std::unique_ptr<IFileScanOperation>> scanOperations)
{
LMS_SCOPED_TRACE_OVERVIEW("Scanner", "ProcessScanResults");
db::Session& dbSession{ _db.getTLSSession() };
auto transaction{ dbSession.createWriteTransaction() };
for (auto& scanOperation : scanOperations)
{
if (_abortScan)
return;
scanOperation->processResult(context);
context.stats.scans++;
}
}
} // namespace lms::scanner
@@ -21,16 +21,15 @@
#include <filesystem>
#include <span>
#include <string>
#include <vector>
#include "metadata/IParser.hpp"
#include "FileScanQueue.hpp"
#include "ScanStepBase.hpp"
namespace lms::scanner
{
class IFileScanner;
class MediaLibraryInfo;
class ScanStepScanFiles : public ScanStepBase
{
public:
@@ -40,20 +39,10 @@ namespace lms::scanner
ScanStep getStep() const override { return ScanStep::ScanFiles; }
core::LiteralString getStepName() const override { return "Scan files"; }
void process(ScanContext& context) override;
void process(ScanContext& context, const ScannerSettings::MediaLibraryInfo& mediaLibrary);
bool checkAudioFileNeedScan(ScanContext& context, const std::filesystem::path& file, const ScannerSettings::MediaLibraryInfo& libraryInfo);
bool checkImageFileNeedScan(ScanContext& context, const std::filesystem::path& file);
bool checkLyricsFileNeedScan(ScanContext& context, const std::filesystem::path& file);
void processFileScanResults(ScanContext& context, std::span<const FileScanResult> scanResults, const ScannerSettings::MediaLibraryInfo& libraryInfo);
void processAudioFileScanData(ScanContext& context, const std::filesystem::path& file, const metadata::Track* trackMetadata, const ScannerSettings::MediaLibraryInfo& libraryInfo);
void processImageFileScanData(ScanContext& context, const std::filesystem::path& file, const ImageInfo* imageInfo, const ScannerSettings::MediaLibraryInfo& libraryInfo);
void processLyricsFileScanData(ScanContext& context, const std::filesystem::path& file, const metadata::Lyrics* lyrics, const ScannerSettings::MediaLibraryInfo& libraryInfo);
std::unique_ptr<metadata::IParser> _metadataParser;
const std::vector<std::string> _extraTagsToParse;
void process(ScanContext& context, const MediaLibraryInfo& mediaLibrary);
void processFileScanResults(ScanContext& context, std::span<std::unique_ptr<IFileScanOperation>> scanOperations);
FileScanQueue _fileScanQueue;
std::unordered_map<std::filesystem::path, IFileScanner*> _scannerByExtension;
};
} // namespace lms::scanner
@@ -24,6 +24,9 @@
#include "database/MediaLibrary.hpp"
#include "database/Session.hpp"
#include "MediaLibraryInfo.hpp"
#include "ScannerSettings.hpp"
namespace lms::scanner
{
@@ -34,7 +37,7 @@ namespace lms::scanner
void ScanStepUpdateLibraryFields::processDirectories(ScanContext& context)
{
for (const ScannerSettings::MediaLibraryInfo& mediaLibrary : _settings.mediaLibraries)
for (const MediaLibraryInfo& mediaLibrary : _settings.mediaLibraries)
{
if (_abortScan)
break;
@@ -43,7 +46,7 @@ namespace lms::scanner
}
}
void ScanStepUpdateLibraryFields::processDirectory(ScanContext& context, const ScannerSettings::MediaLibraryInfo& mediaLibrary)
void ScanStepUpdateLibraryFields::processDirectory(ScanContext& context, const MediaLibraryInfo& mediaLibrary)
{
db::Session& session{ _db.getTLSSession() };
@@ -23,6 +23,8 @@
namespace lms::scanner
{
class MediaLibraryInfo;
class ScanStepUpdateLibraryFields : public ScanStepBase
{
public:
@@ -34,6 +36,6 @@ namespace lms::scanner
void process(ScanContext& context) override;
void processDirectories(ScanContext& context);
void processDirectory(ScanContext& context, const ScannerSettings::MediaLibraryInfo& mediaLibrary);
void processDirectory(ScanContext& context, const MediaLibraryInfo& mediaLibrary);
};
} // namespace lms::scanner
@@ -30,12 +30,13 @@ namespace lms::scanner
{
enum class ScanErrorType
{
CannotReadFile, // cannot read file
CannotReadAudioFile, // cannot parse audio file
CannotReadImageFile, // cannot parse image file
CannotReadLyricsFile, // cannot parse lyrics file
NoAudioTrack, // no audio track found
BadDuration, // bad duration
CannotReadFile,
CannotReadAudioFile,
CannotReadImageFile,
CannotReadLyricsFile,
CannotReadPlayListFile,
NoAudioTrack,
BadDuration,
};
enum class DuplicateReason
@@ -64,6 +65,7 @@ namespace lms::scanner
{
AssociateArtistImages,
AssociateExternalLyrics,
AssociatePlayListTracks,
AssociateReleaseImages,
CheckForDuplicatedFiles,
CheckForRemovedFiles,