More Bugfixes (#2989)

This commit is contained in:
Joe Milazzo 2024-06-09 13:16:11 -05:00 committed by GitHub
parent 1ae723b405
commit a3e020fe17
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
49 changed files with 579 additions and 272 deletions

View file

@ -56,9 +56,9 @@ public class LibraryWatcher : ILibraryWatcher
/// <summary>
/// Counts within a time frame how many times the buffer became full. Is used to reschedule LibraryWatcher to start monitoring much later rather than instantly
/// </summary>
private int _bufferFullCounter;
private int _restartCounter;
private DateTime _lastErrorTime = DateTime.MinValue;
private static int _bufferFullCounter;
private static int _restartCounter;
private static DateTime _lastErrorTime = DateTime.MinValue;
/// <summary>
/// Used to lock buffer Full Counter
/// </summary>
@ -262,17 +262,19 @@ public class LibraryWatcher : ILibraryWatcher
return;
}
_taskScheduler.ScanFolder(fullPath, _queueWaitTime);
_taskScheduler.ScanFolder(fullPath, filePath, _queueWaitTime);
}
catch (Exception ex)
{
_logger.LogError(ex, "[LibraryWatcher] An error occured when processing a watch event");
}
_logger.LogDebug("[LibraryWatcher] ProcessChange completed in {ElapsedMilliseconds}ms", sw.ElapsedMilliseconds);
_logger.LogTrace("[LibraryWatcher] ProcessChange completed in {ElapsedMilliseconds}ms", sw.ElapsedMilliseconds);
}
private string GetFolder(string filePath, IEnumerable<string> libraryFolders)
{
// TODO: I can optimize this to avoid a library scan and instead do a Series Scan by finding the series that has a lowestFolderPath higher or equal to the filePath
var parentDirectory = _directoryService.GetParentDirectoryName(filePath);
_logger.LogTrace("[LibraryWatcher] Parent Directory: {ParentDirectory}", parentDirectory);
if (string.IsNullOrEmpty(parentDirectory)) return string.Empty;

View file

@ -114,7 +114,6 @@ public class ParseScannedFiles
_eventHub = eventHub;
}
/// <summary>
/// This will Scan all files in a folder path. For each folder within the folderPath, FolderAction will be invoked for all files contained
/// </summary>
@ -122,48 +121,53 @@ public class ParseScannedFiles
/// <param name="seriesPaths">A dictionary mapping a normalized path to a list of <see cref="SeriesModified"/> to help scanner skip I/O</param>
/// <param name="folderPath">A library folder or series folder</param>
/// <param name="forceCheck">If we should bypass any folder last write time checks on the scan and force I/O</param>
public IList<ScanResult> ProcessFiles(string folderPath, bool scanDirectoryByDirectory,
public async Task<IList<ScanResult>> ProcessFiles(string folderPath, bool scanDirectoryByDirectory,
IDictionary<string, IList<SeriesModified>> seriesPaths, Library library, bool forceCheck = false)
{
string normalizedPath;
var result = new List<ScanResult>();
var fileExtensions = string.Join("|", library.LibraryFileTypes.Select(l => l.FileTypeGroup.GetRegex()));
var matcher = BuildMatcher(library);
var result = new List<ScanResult>();
if (scanDirectoryByDirectory)
{
// This is used in library scan, so we should check first for a ignore file and use that here as well
var matcher = new GlobMatcher();
foreach (var pattern in library.LibraryExcludePatterns.Where(p => !string.IsNullOrEmpty(p.Pattern)))
{
matcher.AddExclude(pattern.Pattern);
}
var directories = _directoryService.GetDirectories(folderPath, matcher).ToList();
var directories = _directoryService.GetDirectories(folderPath, matcher).Select(Parser.Parser.NormalizePath);
foreach (var directory in directories)
{
// Since this is a loop, we need a list return
normalizedPath = Parser.Parser.NormalizePath(directory);
if (HasSeriesFolderNotChangedSinceLastScan(seriesPaths, normalizedPath, forceCheck))
await _eventHub.SendMessageAsync(MessageFactory.NotificationProgress,
MessageFactory.FileScanProgressEvent(directory, library.Name, ProgressEventType.Updated));
if (HasSeriesFolderNotChangedSinceLastScan(seriesPaths, directory, forceCheck))
{
result.Add(new ScanResult()
if (result.Exists(r => r.Folder == directory))
{
Files = ArraySegment<string>.Empty,
Folder = directory,
LibraryRoot = folderPath,
HasChanged = false
});
continue;
}
result.Add(CreateScanResult(directory, folderPath, false, ArraySegment<string>.Empty));
}
else if (seriesPaths.TryGetValue(normalizedPath, out var series) && series.All(s => !string.IsNullOrEmpty(s.LowestFolderPath)))
else if (seriesPaths.TryGetValue(directory, out var series) && series.All(s => !string.IsNullOrEmpty(s.LowestFolderPath)))
{
// If there are multiple series inside this path, let's check each of them to see which was modified and only scan those
// This is very helpful for ComicVine libraries by Publisher
_logger.LogDebug("[ProcessFiles] {Directory} is dirty and has multiple series folders, checking if we can avoid a full scan", directory);
foreach (var seriesModified in series)
{
if (HasSeriesFolderNotChangedSinceLastScan(seriesModified, seriesModified.LowestFolderPath!))
var hasFolderChangedSinceLastScan = library.LastScanned.Truncate(TimeSpan.TicksPerSecond) <
_directoryService
.GetLastWriteTime(seriesModified.LowestFolderPath!)
.Truncate(TimeSpan.TicksPerSecond);
await _eventHub.SendMessageAsync(MessageFactory.NotificationProgress,
MessageFactory.FileScanProgressEvent(seriesModified.LowestFolderPath!, library.Name, ProgressEventType.Updated));
if (!hasFolderChangedSinceLastScan)
{
result.Add(CreateScanResult(directory, folderPath, false, ArraySegment<string>.Empty));
_logger.LogDebug("[ProcessFiles] {Directory} subfolder {Folder} did not change since last scan, adding entry to skip", directory, seriesModified.LowestFolderPath);
result.Add(CreateScanResult(seriesModified.LowestFolderPath!, folderPath, false, ArraySegment<string>.Empty));
}
else
{
_logger.LogDebug("[ProcessFiles] {Directory} subfolder {Folder} changed, adding folders", directory, seriesModified.LowestFolderPath);
result.Add(CreateScanResult(directory, folderPath, true,
_directoryService.ScanFiles(seriesModified.LowestFolderPath!, fileExtensions, matcher)));
}
@ -173,19 +177,22 @@ public class ParseScannedFiles
{
// For a scan, this is doing everything in the directory loop before the folder Action is called...which leads to no progress indication
result.Add(CreateScanResult(directory, folderPath, true,
_directoryService.ScanFiles(directory, fileExtensions)));
_directoryService.ScanFiles(directory, fileExtensions, matcher)));
}
}
return result;
}
normalizedPath = Parser.Parser.NormalizePath(folderPath);
var normalizedPath = Parser.Parser.NormalizePath(folderPath);
var libraryRoot =
library.Folders.FirstOrDefault(f =>
Parser.Parser.NormalizePath(folderPath).Contains(Parser.Parser.NormalizePath(f.Path)))?.Path ??
normalizedPath.Contains(Parser.Parser.NormalizePath(f.Path)))?.Path ??
folderPath;
await _eventHub.SendMessageAsync(MessageFactory.NotificationProgress,
MessageFactory.FileScanProgressEvent(normalizedPath, library.Name, ProgressEventType.Updated));
if (HasSeriesFolderNotChangedSinceLastScan(seriesPaths, normalizedPath, forceCheck))
{
result.Add(CreateScanResult(folderPath, libraryRoot, false, ArraySegment<string>.Empty));
@ -193,13 +200,24 @@ public class ParseScannedFiles
else
{
result.Add(CreateScanResult(folderPath, libraryRoot, true,
_directoryService.ScanFiles(folderPath, fileExtensions)));
_directoryService.ScanFiles(folderPath, fileExtensions, matcher)));
}
return result;
}
private static GlobMatcher BuildMatcher(Library library)
{
var matcher = new GlobMatcher();
foreach (var pattern in library.LibraryExcludePatterns.Where(p => !string.IsNullOrEmpty(p.Pattern)))
{
matcher.AddExclude(pattern.Pattern);
}
return matcher;
}
private static ScanResult CreateScanResult(string folderPath, string libraryRoot, bool hasChanged,
IList<string> files)
{
@ -243,7 +261,7 @@ public class ParseScannedFiles
NormalizedName = normalizedSeries
};
scannedSeries.AddOrUpdate(existingKey, new List<ParserInfo>() {info}, (_, oldValue) =>
scannedSeries.AddOrUpdate(existingKey, [info], (_, oldValue) =>
{
oldValue ??= new List<ParserInfo>();
if (!oldValue.Contains(info))
@ -338,7 +356,7 @@ public class ParseScannedFiles
{
try
{
var scanResults = ProcessFiles(folderPath, isLibraryScan, seriesPaths, library, forceCheck);
var scanResults = await ProcessFiles(folderPath, isLibraryScan, seriesPaths, library, forceCheck);
foreach (var scanResult in scanResults)
{
@ -414,15 +432,19 @@ public class ParseScannedFiles
/// <param name="library"></param>
private async Task ProcessScanResult(ScanResult result, IDictionary<string, IList<SeriesModified>> seriesPaths, Library library)
{
// TODO: This should return the result as we are modifying it as a side effect
// If the folder hasn't changed, generate fake ParserInfos for the Series that were in that folder.
var normalizedFolder = Parser.Parser.NormalizePath(result.Folder);
if (!result.HasChanged)
{
var normalizedFolder = Parser.Parser.NormalizePath(result.Folder);
result.ParserInfos = seriesPaths[normalizedFolder].Select(fp => new ParserInfo()
{
Series = fp.SeriesName,
Format = fp.Format,
}).ToList();
result.ParserInfos = seriesPaths[normalizedFolder]
.Select(fp => new ParserInfo()
{
Series = fp.SeriesName,
Format = fp.Format,
})
.ToList();
_logger.LogDebug("[ScannerService] Skipped File Scan for {Folder} as it hasn't changed since last scan", normalizedFolder);
await _eventHub.SendMessageAsync(MessageFactory.NotificationProgress,
@ -431,25 +453,24 @@ public class ParseScannedFiles
}
var files = result.Files;
var folder = result.Folder;
var libraryRoot = result.LibraryRoot;
// When processing files for a folder and we do enter, we need to parse the information and combine parser infos
// NOTE: We might want to move the merge step later in the process, like return and combine.
_logger.LogDebug("[ScannerService] Found {Count} files for {Folder}", files.Count, folder);
await _eventHub.SendMessageAsync(MessageFactory.NotificationProgress,
MessageFactory.FileScanProgressEvent($"{files.Count} files in {folder}", library.Name, ProgressEventType.Updated));
if (files.Count == 0)
{
_logger.LogInformation("[ScannerService] {Folder} is empty, no longer in this location, or has no file types that match Library File Types", folder);
_logger.LogInformation("[ScannerService] {Folder} is empty, no longer in this location, or has no file types that match Library File Types", normalizedFolder);
result.ParserInfos = ArraySegment<ParserInfo>.Empty;
return;
}
_logger.LogDebug("[ScannerService] Found {Count} files for {Folder}", files.Count, normalizedFolder);
await _eventHub.SendMessageAsync(MessageFactory.NotificationProgress,
MessageFactory.FileScanProgressEvent($"{files.Count} files in {normalizedFolder}", library.Name, ProgressEventType.Updated));
// Multiple Series can exist within a folder. We should instead put these infos on the result and perform merging above
IList<ParserInfo> infos = files
.Select(file => _readingItemService.ParseFile(file, folder, libraryRoot, library.Type))
.Select(file => _readingItemService.ParseFile(file, normalizedFolder, result.LibraryRoot, library.Type))
.Where(info => info != null)
.ToList()!;

View file

@ -26,7 +26,7 @@ public class BasicParser(IDirectoryService directoryService, IDefaultParser imag
{
Filename = Path.GetFileName(filePath),
Format = Parser.ParseFormat(filePath),
Title = Parser.RemoveExtensionIfSupported(fileName),
Title = Parser.RemoveExtensionIfSupported(fileName)!,
FullFilePath = Parser.NormalizePath(filePath),
Series = string.Empty,
ComicInfo = comicInfo
@ -76,6 +76,9 @@ public class BasicParser(IDirectoryService directoryService, IDefaultParser imag
ret.Chapters = Parser.DefaultChapter;
ret.Volumes = Parser.SpecialVolume;
// NOTE: This uses rootPath. LibraryRoot works better for manga, but it's not always that way.
// It might be worth writing some logic if the file is a special, to take the folder above the Specials/
// if present
ParseFromFallbackFolders(filePath, rootPath, type, ref ret);
}

View file

@ -101,7 +101,7 @@ public abstract class DefaultParser(IDirectoryService directoryService) : IDefau
}
}
protected void UpdateFromComicInfo(ParserInfo info)
protected static void UpdateFromComicInfo(ParserInfo info)
{
if (info.ComicInfo == null) return;
@ -109,6 +109,10 @@ public abstract class DefaultParser(IDirectoryService directoryService) : IDefau
{
info.Volumes = info.ComicInfo.Volume;
}
if (!string.IsNullOrEmpty(info.ComicInfo.Number))
{
info.Chapters = info.ComicInfo.Number;
}
if (!string.IsNullOrEmpty(info.ComicInfo.Series))
{
info.Series = info.ComicInfo.Series.Trim();
@ -125,16 +129,6 @@ public abstract class DefaultParser(IDirectoryService directoryService) : IDefau
info.Volumes = Parser.SpecialVolume;
}
if (!string.IsNullOrEmpty(info.ComicInfo.Number))
{
info.Chapters = info.ComicInfo.Number;
if (info.IsSpecial && Parser.DefaultChapter != info.Chapters)
{
info.IsSpecial = false;
info.Volumes = Parser.SpecialVolume;
}
}
// Patch is SeriesSort from ComicInfo
if (!string.IsNullOrEmpty(info.ComicInfo.TitleSort))
{

View file

@ -103,7 +103,11 @@ public static class Parser
private static readonly Regex CoverImageRegex = new Regex(@"(?<![[a-z]\d])(?:!?)(?<!back)(?<!back_)(?<!back-)(cover|folder)(?![\w\d])",
MatchOptions, RegexTimeout);
private static readonly Regex NormalizeRegex = new Regex(@"[^\p{L}0-9\+!\*]",
/// <summary>
/// Normalize everything within Kavita. Some characters don't fall under Unicode, like full-width characters and need to be
/// added on a case-by-case basis.
/// </summary>
private static readonly Regex NormalizeRegex = new Regex(@"[^\p{L}0-9\+!]",
MatchOptions, RegexTimeout);
/// <summary>

View file

@ -45,7 +45,7 @@ public interface IScannerService
[AutomaticRetry(Attempts = 3, OnAttemptsExceeded = AttemptsExceededAction.Delete)]
Task ScanSeries(int seriesId, bool bypassFolderOptimizationChecks = true);
Task ScanFolder(string folder);
Task ScanFolder(string folder, string originalPath);
Task AnalyzeFiles();
}
@ -135,30 +135,35 @@ public class ScannerService : IScannerService
/// Given a generic folder path, will invoke a Series scan or Library scan.
/// </summary>
/// <remarks>This will Schedule the job to run 1 minute in the future to allow for any close-by duplicate requests to be dropped</remarks>
/// <param name="folder"></param>
public async Task ScanFolder(string folder)
/// <param name="folder">Normalized folder</param>
/// <param name="originalPath">If invoked from LibraryWatcher, this maybe a nested folder and can allow for optimization</param>
public async Task ScanFolder(string folder, string originalPath)
{
Series? series = null;
try
{
series = await _unitOfWork.SeriesRepository.GetSeriesByFolderPath(folder, SeriesIncludes.Library);
series = await _unitOfWork.SeriesRepository.GetSeriesThatContainsLowestFolderPath(originalPath,
SeriesIncludes.Library) ??
await _unitOfWork.SeriesRepository.GetSeriesByFolderPath(originalPath, SeriesIncludes.Library) ??
await _unitOfWork.SeriesRepository.GetSeriesByFolderPath(folder, SeriesIncludes.Library);
}
catch (InvalidOperationException ex)
{
if (ex.Message.Equals("Sequence contains more than one element."))
{
_logger.LogCritical(ex, "[ScannerService] Multiple series map to this folder. Library scan will be used for ScanFolder");
_logger.LogCritical(ex, "[ScannerService] Multiple series map to this folder or folder is at library root. Library scan will be used for ScanFolder");
}
}
// TODO: Figure out why we have the library type restriction here
if (series != null && (series.Library.Type != LibraryType.Book || series.Library.Type != LibraryType.LightNovel))
if (series != null && series.Library.Type is not (LibraryType.Book or LibraryType.LightNovel))
{
if (TaskScheduler.HasScanTaskRunningForSeries(series.Id))
{
_logger.LogInformation("[ScannerService] Scan folder invoked for {Folder} but a task is already queued for this series. Dropping request", folder);
return;
}
_logger.LogInformation("[ScannerService] Scan folder invoked for {Folder}, Series matched to folder and ScanSeries enqueued for 1 minute", folder);
BackgroundJob.Schedule(() => ScanSeries(series.Id, true), TimeSpan.FromMinutes(1));
return;
}

View file

@ -181,7 +181,9 @@ public class StatsService : IStatsService
{
InstallId = serverSettings.InstallId,
KavitaVersion = serverSettings.InstallVersion,
IsDocker = OsInfo.IsDocker
IsDocker = OsInfo.IsDocker,
FirstInstallDate = serverSettings.FirstInstallDate,
FirstInstallVersion = serverSettings.FirstInstallVersion
};
}