DMM Offline (#198)

* Process DMM all locally

single call to github to download the repo archive.
remove need for PAT
update RTN to 0.2.13
change to batch_parse for title parsing from RTN

* introduce concurrent dictionary, and parallelism

.gitignore

@@ -612,3 +612,7 @@ fabric.properties
 # Mac directory indexes
 .DS_Store
 deployment/docker/stack.env
+
+src/producer/src/python/
+src/debrid-collector/python/
+src/qbit-collector/python/

deployment/docker/docker-compose.yaml

@@ -94,7 +94,7 @@ services:
         condition: service_healthy
     env_file: stack.env
     hostname: knightcrawler-addon
-    image: gabisonfire/knightcrawler-addon:2.0.15
+    image: gabisonfire/knightcrawler-addon:2.0.20
     labels:
       logging: promtail
     networks:
@@ -117,7 +117,7 @@ services:
       redis:
         condition: service_healthy
     env_file: stack.env
-    image: gabisonfire/knightcrawler-consumer:2.0.15
+    image: gabisonfire/knightcrawler-consumer:2.0.20
     labels:
       logging: promtail
     networks:
@@ -138,7 +138,7 @@ services:
       redis:
         condition: service_healthy
     env_file: stack.env
-    image: gabisonfire/knightcrawler-debrid-collector:2.0.15
+    image: gabisonfire/knightcrawler-debrid-collector:2.0.20
     labels:
       logging: promtail
     networks:
@@ -152,7 +152,7 @@ services:
       migrator:
         condition: service_completed_successfully
     env_file: stack.env
-    image: gabisonfire/knightcrawler-metadata:2.0.15
+    image: gabisonfire/knightcrawler-metadata:2.0.20
     networks:
       - knightcrawler-network
     restart: "no"
@@ -163,7 +163,7 @@ services:
       postgres:
         condition: service_healthy
     env_file: stack.env
-    image: gabisonfire/knightcrawler-migrator:2.0.15
+    image: gabisonfire/knightcrawler-migrator:2.0.20
     networks:
       - knightcrawler-network
     restart: "no"
@@ -182,7 +182,7 @@ services:
       redis:
         condition: service_healthy
     env_file: stack.env
-    image: gabisonfire/knightcrawler-producer:2.0.15
+    image: gabisonfire/knightcrawler-producer:2.0.20
     labels:
       logging: promtail
     networks:
@@ -207,7 +207,7 @@ services:
     deploy:
       replicas: ${QBIT_REPLICAS:-0}
     env_file: stack.env
-    image: gabisonfire/knightcrawler-qbit-collector:2.0.15
+    image: gabisonfire/knightcrawler-qbit-collector:2.0.20
     labels:
       logging: promtail
     networks:

deployment/docker/src/components/knightcrawler.yaml

@@ -20,7 +20,7 @@ x-depends: &knightcrawler-app-depends
 
 services:
   metadata:
-    image: gabisonfire/knightcrawler-metadata:2.0.15
+    image: gabisonfire/knightcrawler-metadata:2.0.20
     env_file: ../../.env
     networks:
       - knightcrawler-network
@@ -30,7 +30,7 @@ services:
         condition: service_completed_successfully
 
   migrator:
-    image: gabisonfire/knightcrawler-migrator:2.0.15
+    image: gabisonfire/knightcrawler-migrator:2.0.20
     env_file: ../../.env
     networks:
       - knightcrawler-network
@@ -40,7 +40,7 @@ services:
         condition: service_healthy
 
   addon:
-    image: gabisonfire/knightcrawler-addon:2.0.15
+    image: gabisonfire/knightcrawler-addon:2.0.20
     <<: [*knightcrawler-app, *knightcrawler-app-depends]
     restart: unless-stopped
     hostname: knightcrawler-addon
@@ -48,22 +48,22 @@ services:
       - "7000:7000"
 
   consumer:
-    image: gabisonfire/knightcrawler-consumer:2.0.15
+    image: gabisonfire/knightcrawler-consumer:2.0.20
     <<: [*knightcrawler-app, *knightcrawler-app-depends]
     restart: unless-stopped
 
   debridcollector:
-    image: gabisonfire/knightcrawler-debrid-collector:2.0.15
+    image: gabisonfire/knightcrawler-debrid-collector:2.0.20
     <<: [*knightcrawler-app, *knightcrawler-app-depends]
     restart: unless-stopped
         
   producer:
-    image: gabisonfire/knightcrawler-producer:2.0.15
+    image: gabisonfire/knightcrawler-producer:2.0.20
     <<: [*knightcrawler-app, *knightcrawler-app-depends]
     restart: unless-stopped
 
   qbitcollector:
-    image: gabisonfire/knightcrawler-qbit-collector:2.0.15
+    image: gabisonfire/knightcrawler-qbit-collector:2.0.20
     <<: [*knightcrawler-app, *knightcrawler-app-depends]
     restart: unless-stopped
     depends_on:

deployment/docker/stack.env

@@ -38,6 +38,3 @@ QBIT_REPLICAS=0
 
 # Addon
 DEBUG_MODE=false
-
-# Producer
-GITHUB_PAT=

src/debrid-collector/DebridCollector.csproj

@@ -17,7 +17,6 @@
       <PackageReference Include="Microsoft.Extensions.Http" Version="8.0.0" />
       <PackageReference Include="Microsoft.Extensions.Http.Polly" Version="8.0.3" />
       <PackageReference Include="Polly" Version="8.3.1" />
-      <PackageReference Include="PromKnight.ParseTorrentTitle" Version="1.0.4" />
       <PackageReference Include="Serilog" Version="3.1.1" />
       <PackageReference Include="Serilog.AspNetCore" Version="8.0.1" />
       <PackageReference Include="Serilog.Sinks.Console" Version="5.0.1" />
@@ -29,10 +28,30 @@
       <None Include="Configuration\logging.json">
         <CopyToOutputDirectory>Always</CopyToOutputDirectory>
       </None>
+      <None Update="requirements.txt">
+        <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+      </None>
+      <Content Remove="eng\**" />
+      <None Remove="eng\**" />
+    </ItemGroup>
+
+    <ItemGroup Condition="'$(Configuration)' == 'Debug'">
+        <Content Remove="python\**" />
+        <None Include="python\**">
+            <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+        </None>
     </ItemGroup>
 
     <ItemGroup>
       <ProjectReference Include="..\shared\SharedContracts.csproj" />
     </ItemGroup>
 
+    <ItemGroup>
+      <Compile Remove="eng\**" />
+    </ItemGroup>
+
+    <ItemGroup>
+      <EmbeddedResource Remove="eng\**" />
+    </ItemGroup>
+
 </Project>

src/debrid-collector/DebridCollector.sln

@@ -6,6 +6,12 @@ Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "SharedContracts", "..\share
 EndProject
 Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "shared", "shared", "{2C0A0F53-28E6-404F-9EFE-DADFBEF8338B}"
 EndProject
+Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "eng", "eng", "{72A042C3-B4F3-45C5-AC20-041FE8F41EFC}"
+	ProjectSection(SolutionItems) = preProject
+		eng\install-python-reqs.ps1 = eng\install-python-reqs.ps1
+		eng\install-python-reqs.sh = eng\install-python-reqs.sh
+	EndProjectSection
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|Any CPU = Debug|Any CPU

src/debrid-collector/Dockerfile

@@ -9,12 +9,23 @@ RUN dotnet restore -a $TARGETARCH
 RUN dotnet publish -c Release --no-restore -o /src/out -a $TARGETARCH
 
 
-FROM mcr.microsoft.com/dotnet/aspnet:8.0-alpine
+FROM mcr.microsoft.com/dotnet/aspnet:8.0-alpine3.19
 
 WORKDIR /app
+
+ENV PYTHONUNBUFFERED=1
+
+RUN apk add --update --no-cache python3=~3.11.8-r0 py3-pip && ln -sf python3 /usr/bin/python
+
 COPY --from=build /src/out .
+
+RUN rm -rf /app/python && mkdir -p /app/python
+
+RUN pip3 install -r /app/requirements.txt -t /app/python
+
 RUN addgroup -S debrid && adduser -S -G debrid debrid
 USER debrid
 HEALTHCHECK --interval=30s --timeout=30s --start-period=5s --retries=3 \
   CMD pgrep -f dotnet || exit 1
+ENV PYTHONNET_PYDLL=/usr/lib/libpython3.11.so.1.0
 ENTRYPOINT ["dotnet", "DebridCollector.dll"]

src/debrid-collector/Extensions/ServiceCollectionExtensions.cs

@@ -1,5 +1,3 @@
-using DebridCollector.Features.Configuration;
-
 namespace DebridCollector.Extensions;
 
 public static class ServiceCollectionExtensions
@@ -17,7 +15,8 @@ public static class ServiceCollectionExtensions
         var serviceConfiguration = services.LoadConfigurationFromEnv<DebridCollectorConfiguration>();
         
         services.AddRealDebridClient(serviceConfiguration);
-        services.AddSingleton<IParseTorrentTitle, ParseTorrentTitle>();
+        services.RegisterPythonEngine();
+        services.AddSingleton<IRankTorrentName, RankTorrentName>();
         services.AddHostedService<DebridRequestProcessor>();
 
         return services;

src/debrid-collector/Features/Debrid/ServiceCollectionExtensions.cs

@@ -1,6 +1,4 @@
-using DebridCollector.Features.Configuration;
+namespace DebridCollector.Features.Debrid;
-
-namespace DebridCollector.Features.Debrid;
 
 public static class ServiceCollectionExtensions
 {

src/debrid-collector/Features/Worker/DebridMetaToTorrentMeta.cs

@@ -3,10 +3,11 @@ namespace DebridCollector.Features.Worker;
 public static class DebridMetaToTorrentMeta
 {
     public static IReadOnlyList<TorrentFile> MapMetadataToFilesCollection(
-        IParseTorrentTitle torrentTitle,
+        IRankTorrentName rankTorrentName,
         Torrent torrent,
         string ImdbId,
-        FileDataDictionary Metadata)
+        FileDataDictionary Metadata,
+        ILogger<WriteMetadataConsumer> logger)
     {
         try
         {
@@ -15,34 +16,42 @@ public static class DebridMetaToTorrentMeta
             foreach (var metadataEntry in Metadata.Where(m => Filetypes.VideoFileExtensions.Any(ext => m.Value.Filename.EndsWith(ext))))
             {
                 var validFileIndex = int.TryParse(metadataEntry.Key, out var fileIndex);
+                var fileIndexMinusOne = Math.Max(0, fileIndex - 1);
 
                 var file = new TorrentFile
                 {
                     ImdbId = ImdbId,
                     KitsuId = 0,
                     InfoHash = torrent.InfoHash,
-                    FileIndex = validFileIndex ? fileIndex : 0,
+                    FileIndex = validFileIndex ? fileIndexMinusOne : 0,
                     Title = metadataEntry.Value.Filename,
                     Size = metadataEntry.Value.Filesize.GetValueOrDefault(),
                 };
 
-                var parsedTitle = torrentTitle.Parse(file.Title);
+                var parsedTitle = rankTorrentName.Parse(file.Title, false);
 
-                file.ImdbSeason = parsedTitle.Seasons.FirstOrDefault();
+                if (!parsedTitle.Success)
-                file.ImdbEpisode = parsedTitle.Episodes.FirstOrDefault();
+                {
+                    logger.LogWarning("Failed to parse title {Title} for metadata mapping", file.Title);
+                    continue;
+                }
+
+                file.ImdbSeason = parsedTitle.Response?.Season?.FirstOrDefault() ?? 0;
+                file.ImdbEpisode = parsedTitle.Response?.Episode?.FirstOrDefault() ?? 0;
 
                 files.Add(file);
             }
 
             return files;
         }
-        catch (Exception)
+        catch (Exception ex)
         {
+            logger.LogWarning("Failed to map metadata to files collection: {Exception}", ex.Message);
             return [];
         }
     }
     
-    public static async Task<IReadOnlyList<SubtitleFile>> MapMetadataToSubtitlesCollection(IDataStorage storage, string InfoHash, FileDataDictionary Metadata)
+    public static async Task<IReadOnlyList<SubtitleFile>> MapMetadataToSubtitlesCollection(IDataStorage storage, string InfoHash, FileDataDictionary Metadata, ILogger<WriteMetadataConsumer> logger)
     {
         try
         {
@@ -58,13 +67,14 @@ public static class DebridMetaToTorrentMeta
             foreach (var metadataEntry in Metadata.Where(m => Filetypes.SubtitleFileExtensions.Any(ext => m.Value.Filename.EndsWith(ext))))
             {
                 var validFileIndex = int.TryParse(metadataEntry.Key, out var fileIndex);
+                var fileIndexMinusOne = Math.Max(0, fileIndex - 1);
                 var fileId = torrentFiles.FirstOrDefault(
                     t => Path.GetFileNameWithoutExtension(t.Title) == Path.GetFileNameWithoutExtension(metadataEntry.Value.Filename))?.Id ?? 0;
 
                 var file = new SubtitleFile
                 {
                     InfoHash = InfoHash,
-                    FileIndex = validFileIndex ? fileIndex : 0,
+                    FileIndex = validFileIndex ? fileIndexMinusOne : 0,
                     FileId = fileId,
                     Title = metadataEntry.Value.Filename,
                 };
@@ -74,8 +84,9 @@ public static class DebridMetaToTorrentMeta
 
             return files;
         }
-        catch (Exception)
+        catch (Exception ex)
         {
+            logger.LogWarning("Failed to map metadata to subtitles collection: {Exception}", ex.Message);
             return [];
         }
     }

src/debrid-collector/Features/Worker/InfohashMetadataSagaStateMachine.cs

@@ -53,6 +53,12 @@ public class InfohashMetadataSagaStateMachine : MassTransitStateMachine<Infohash
                 .Then(
                     context =>
                     {
+                        if (!context.Message.WithFiles)
+                        {
+                            logger.LogInformation("No files written for torrent {InfoHash} in Saga {SagaId}", context.Saga.Torrent.InfoHash, context.Saga.CorrelationId);
+                            return;
+                        }
+                        
                         logger.LogInformation("Metadata Written for torrent {InfoHash} in Saga {SagaId}", context.Saga.Torrent.InfoHash, context.Saga.CorrelationId);
                     })
                 .TransitionTo(Completed)

src/debrid-collector/Features/Worker/Requests.cs

@@ -16,7 +16,7 @@ public record WriteMetadata(Torrent Torrent, TorrentMetadataResponse Metadata, s
 }
 
 [EntityName("metadata-written-debrid-colloctor")]
-public record MetadataWritten(TorrentMetadataResponse Metadata) : CorrelatedBy<Guid>
+public record MetadataWritten(TorrentMetadataResponse Metadata, bool WithFiles) : CorrelatedBy<Guid>
 {
     public Guid CorrelationId { get; init; } = Metadata.CorrelationId;
 }

src/debrid-collector/Features/Worker/WriteMetadataConsumer.cs

@@ -1,25 +1,28 @@
 namespace DebridCollector.Features.Worker;
 
-public class WriteMetadataConsumer(IParseTorrentTitle parseTorrentTitle, IDataStorage dataStorage) : IConsumer<WriteMetadata>
+public class WriteMetadataConsumer(IRankTorrentName rankTorrentName, IDataStorage dataStorage, ILogger<WriteMetadataConsumer> logger) : IConsumer<WriteMetadata>
 {
     public async Task Consume(ConsumeContext<WriteMetadata> context)
     {
         var request = context.Message;
         
-        var torrentFiles = DebridMetaToTorrentMeta.MapMetadataToFilesCollection(parseTorrentTitle, request.Torrent, request.ImdbId, request.Metadata.Metadata);
+        var torrentFiles = DebridMetaToTorrentMeta.MapMetadataToFilesCollection(rankTorrentName, request.Torrent, request.ImdbId, request.Metadata.Metadata, logger);
 
-        if (torrentFiles.Any())
+        if (!torrentFiles.Any())
         {
-            await dataStorage.InsertFiles(torrentFiles);
+            await context.Publish(new MetadataWritten(request.Metadata, false));
-
+            return;
-            var subtitles = await DebridMetaToTorrentMeta.MapMetadataToSubtitlesCollection(dataStorage, request.Torrent.InfoHash, request.Metadata.Metadata);
-
-            if (subtitles.Any())
-            {
-                await dataStorage.InsertSubtitles(subtitles);
-            }
         }
         
-        await context.Publish(new MetadataWritten(request.Metadata));
+        await dataStorage.InsertFiles(torrentFiles);
+
+        var subtitles = await DebridMetaToTorrentMeta.MapMetadataToSubtitlesCollection(dataStorage, request.Torrent.InfoHash, request.Metadata.Metadata, logger);
+
+        if (subtitles.Any())
+        {
+            await dataStorage.InsertSubtitles(subtitles);
+        }
+        
+        await context.Publish(new MetadataWritten(request.Metadata, true));
     }
 }

src/debrid-collector/GlobalUsings.cs

@@ -4,17 +4,18 @@ global using System.Text.Json;
 global using System.Text.Json.Serialization;
 global using System.Threading.Channels;
 global using DebridCollector.Extensions;
+global using DebridCollector.Features.Configuration;
 global using DebridCollector.Features.Debrid;
 global using DebridCollector.Features.Worker;
 global using MassTransit;
-global using MassTransit.Mediator;
 global using Microsoft.AspNetCore.Builder;
 global using Microsoft.Extensions.DependencyInjection;
 global using Polly;
 global using Polly.Extensions.Http;
-global using PromKnight.ParseTorrentTitle;
 global using SharedContracts.Configuration;
 global using SharedContracts.Dapper;
 global using SharedContracts.Extensions;
 global using SharedContracts.Models;
+global using SharedContracts.Python;
+global using SharedContracts.Python.RTN;
 global using SharedContracts.Requests;

src/debrid-collector/eng/install-python-reqs.ps1

@@ -0,0 +1,2 @@
+mkdir -p ../python
+python -m pip install -r ../requirements.txt -t ../python/

src/debrid-collector/eng/install-python-reqs.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+rm -rf ../python
+mkdir -p ../python
+python3 -m pip install -r ../requirements.txt -t ../python/

src/debrid-collector/requirements.txt

@@ -0,0 +1 @@
+rank-torrent-name==0.2.13

src/producer/src/Configuration/scrapers.json

@@ -28,7 +28,7 @@
       },
       {
         "Name": "SyncDmmJob",
-        "IntervalSeconds": 1800,
+        "IntervalSeconds": 10800,
         "Enabled": true
       },
       {

src/producer/src/Features/Crawlers/Dmm/DMMFileDownloader.cs

@@ -0,0 +1,70 @@
+namespace Producer.Features.Crawlers.Dmm;
+
+public class DMMFileDownloader(HttpClient client, ILogger<DMMFileDownloader> logger) : IDMMFileDownloader
+{
+    private const string Filename = "main.zip";
+    private readonly IReadOnlyCollection<string> _filesToIgnore = [
+        "index.html",
+        "404.html",
+        "dedupe.sh",
+        "CNAME",
+    ];
+    
+    public const string ClientName = "DmmFileDownloader";
+    
+    public async Task<string> DownloadFileToTempPath(CancellationToken cancellationToken)
+    {
+        logger.LogInformation("Downloading DMM Hashlists");
+        
+        var response = await client.GetAsync(Filename, cancellationToken);
+
+        var tempDirectory = Path.Combine(Path.GetTempPath(), "DMMHashlists");
+
+        EnsureDirectoryIsClean(tempDirectory);
+
+        response.EnsureSuccessStatusCode();
+
+        await using var stream = await response.Content.ReadAsStreamAsync(cancellationToken);
+        using var archive = new ZipArchive(stream);
+        
+        logger.LogInformation("Extracting DMM Hashlists to {TempDirectory}", tempDirectory);
+
+        foreach (var entry in archive.Entries)
+        {
+            var entryPath = Path.Combine(tempDirectory, Path.GetFileName(entry.FullName));
+            if (!entry.FullName.EndsWith('/')) // It's a file
+            {
+                entry.ExtractToFile(entryPath, true);
+            }
+        }
+
+        foreach (var file in _filesToIgnore)
+        {
+            CleanRepoExtras(tempDirectory, file);    
+        }
+        
+        logger.LogInformation("Downloaded and extracted Repository to {TempDirectory}", tempDirectory);
+
+        return tempDirectory;
+    }
+
+    private static void CleanRepoExtras(string tempDirectory, string fileName)
+    {
+        var repoIndex = Path.Combine(tempDirectory, fileName);
+
+        if (File.Exists(repoIndex))
+        {
+            File.Delete(repoIndex);
+        }
+    }
+
+    private static void EnsureDirectoryIsClean(string tempDirectory)
+    {
+        if (Directory.Exists(tempDirectory))
+        {
+            Directory.Delete(tempDirectory, true);
+        }
+
+        Directory.CreateDirectory(tempDirectory);
+    }
+}

src/producer/src/Features/Crawlers/Dmm/DMMHttpClient.cs

@@ -0,0 +1,6 @@
+namespace Producer.Features.Crawlers.Dmm;
+
+public class DMMHttpClient
+{
+    
+}

src/producer/src/Features/Crawlers/Dmm/DebridMediaManagerCrawler.cs

@@ -1,64 +1,99 @@
 namespace Producer.Features.Crawlers.Dmm;
 
 public partial class DebridMediaManagerCrawler(
-    IHttpClientFactory httpClientFactory,
+    IDMMFileDownloader dmmFileDownloader,
     ILogger<DebridMediaManagerCrawler> logger,
     IDataStorage storage,
-    GithubConfiguration githubConfiguration,
     IRankTorrentName rankTorrentName,
     IDistributedCache cache) : BaseCrawler(logger, storage)
 {
     [GeneratedRegex("""<iframe src="https:\/\/debridmediamanager.com\/hashlist#(.*)"></iframe>""")]
     private static partial Regex HashCollectionMatcher();
-
+    protected override string Url => "";
-    private const string DownloadBaseUrl = "https://raw.githubusercontent.com/debridmediamanager/hashlists/main";
     protected override IReadOnlyDictionary<string, string> Mappings => new Dictionary<string, string>();
-    protected override string Url => "https://api.github.com/repos/debridmediamanager/hashlists/git/trees/main?recursive=1";
     protected override string Source => "DMM";
+
+    private const int ParallelismCount = 4;
     
     public override async Task Execute()
     {
-        var client = httpClientFactory.CreateClient("Scraper");
+        var tempDirectory = await dmmFileDownloader.DownloadFileToTempPath(CancellationToken.None);
-        client.DefaultRequestHeaders.Authorization = new("Bearer", githubConfiguration.PAT);
-        client.DefaultRequestHeaders.UserAgent.ParseAdd("curl");
 
-        var jsonBody = await client.GetStringAsync(Url);
+        var files = Directory.GetFiles(tempDirectory, "*.html", SearchOption.AllDirectories);
 
-        var json =  JsonDocument.Parse(jsonBody);
+        logger.LogInformation("Found {Files} files to parse", files.Length);
 
-        var entriesArray = json.RootElement.GetProperty("tree");
+        var options = new ParallelOptions { MaxDegreeOfParallelism = ParallelismCount };
-
+        
-        logger.LogInformation("Found {Entries} total DMM pages", entriesArray.GetArrayLength());
+        await Parallel.ForEachAsync(files, options, async (file, token) =>
-
-        foreach (var entry in entriesArray.EnumerateArray())
         {
-            await ParsePage(entry, client);
+            var fileName = Path.GetFileName(file);
-        }
+            var torrentDictionary = await ExtractPageContents(file, fileName);
+
+            if (torrentDictionary == null)
+            {
+                return;
+            }
+
+            await ParseTitlesWithRtn(fileName, torrentDictionary);
+            var results = await ParseTorrents(torrentDictionary);
+
+            if (results.Count <= 0)
+            {
+                return;
+            }
+
+            await InsertTorrents(results);
+            await Storage.MarkPageAsIngested(fileName, token);
+        });
     }
 
-    private async Task ParsePage(JsonElement entry, HttpClient client)
+    private async Task ParseTitlesWithRtn(string fileName, IDictionary<string, DmmContent> page)
     {
-        var (pageIngested, name) = await IsAlreadyIngested(entry);
+        logger.LogInformation("Parsing titles for {Page}", fileName);
 
-        if (string.IsNullOrEmpty(name) || pageIngested)
+        var batchProcessables = page.Select(value => new RtnBatchProcessable(value.Key, value.Value.Filename)).ToList();
+        var parsedResponses = rankTorrentName.BatchParse(
+            batchProcessables.Select<RtnBatchProcessable, string>(bp => bp.Filename).ToList(), trashGarbage: false);
+
+        // Filter out unsuccessful responses and match RawTitle to requesting title
+        var successfulResponses = parsedResponses
+            .Where(response => response != null && response.Success)
+            .GroupBy(response => response.Response.RawTitle!)
+            .ToDictionary(group => group.Key, group => group.First());
+
+        var options = new ParallelOptions { MaxDegreeOfParallelism = ParallelismCount };
+
+        await Parallel.ForEachAsync(batchProcessables.Select(t => t.InfoHash), options, (infoHash, _) =>
         {
-            return;
+            if (page.TryGetValue(infoHash, out var dmmContent) &&
-        }
+                successfulResponses.TryGetValue(dmmContent.Filename, out var parsedResponse))
-
+            {
-        var pageSource = await client.GetStringAsync($"{DownloadBaseUrl}/{name}");
+                page[infoHash] = dmmContent with {ParseResponse = parsedResponse};
-
+            }
-        await ExtractPageContents(pageSource, name);
+            
+            return ValueTask.CompletedTask;
+        });
     }
 
-    private async Task ExtractPageContents(string pageSource, string name)
+    private async Task<ConcurrentDictionary<string, DmmContent>?> ExtractPageContents(string filePath, string filenameOnly)
     {
+        var (pageIngested, name) = await IsAlreadyIngested(filenameOnly);
+
+        if (pageIngested)
+        {
+            return [];
+        }
+
+        var pageSource = await File.ReadAllTextAsync(filePath);
+        
         var match = HashCollectionMatcher().Match(pageSource);
 
         if (!match.Success)
         {
             logger.LogWarning("Failed to match hash collection for {Name}", name);
-            await Storage.MarkPageAsIngested(name);
+            await Storage.MarkPageAsIngested(filenameOnly);
-            return;
+            return [];
         }
 
         var encodedJson = match.Groups.Values.ElementAtOrDefault(1);
@@ -66,86 +101,92 @@ public partial class DebridMediaManagerCrawler(
         if (string.IsNullOrEmpty(encodedJson?.Value))
         {
             logger.LogWarning("Failed to extract encoded json for {Name}", name);
-            return;
+            return [];
         }
 
-        await ProcessExtractedContentsAsTorrentCollection(encodedJson.Value, name);
+        var decodedJson = LZString.DecompressFromEncodedURIComponent(encodedJson.Value);
-    }
-
-    private async Task ProcessExtractedContentsAsTorrentCollection(string encodedJson, string name)
-    {
-        var decodedJson = LZString.DecompressFromEncodedURIComponent(encodedJson);
 
         var json = JsonDocument.Parse(decodedJson);
+        
+        var torrents = await json.RootElement.EnumerateArray()
+            .ToAsyncEnumerable()
+            .Select(ParsePageContent)
+            .Where(t => t is not null)
+            .ToListAsync();
 
-        await InsertTorrentsForPage(json);
+        if (torrents.Count == 0)
-
-        var result = await Storage.MarkPageAsIngested(name);
-
-        if (!result.IsSuccess)
         {
-            logger.LogWarning("Failed to mark page as ingested: [{Error}]", result.Failure.ErrorMessage);
+            logger.LogWarning("No torrents found in {Name}", name);
-            return;
+            await Storage.MarkPageAsIngested(filenameOnly);
+            return [];
         }
+        
+        var torrentDictionary = torrents
+            .Where(x => x is not null)
+            .GroupBy(x => x.InfoHash)
+            .ToConcurrentDictionary(g => g.Key, g => new DmmContent(g.First().Filename, g.First().Bytes, null));
 
-        logger.LogInformation("Successfully marked page as ingested");
+        logger.LogInformation("Parsed {Torrents} torrents for {Name}", torrentDictionary.Count, name);
+
+        return torrentDictionary;
     }
 
-    private async Task<IngestedTorrent?> ParseTorrent(JsonElement item)
+    private async Task<List<IngestedTorrent>> ParseTorrents(IDictionary<string, DmmContent> page)
     {
+        var ingestedTorrents = new List<IngestedTorrent>();
 
-        if (!item.TryGetProperty("filename", out var filenameElement) ||
+        var options = new ParallelOptions { MaxDegreeOfParallelism = ParallelismCount };
-            !item.TryGetProperty("bytes", out var bytesElement) ||
+
-            !item.TryGetProperty("hash", out var hashElement))
+        await Parallel.ForEachAsync(page, options, async (kvp, ct) =>
         {
-            return null;
+            var (infoHash, dmmContent) = kvp;
-        }
+            var parsedTorrent = dmmContent.ParseResponse;
+            if (parsedTorrent is not {Success: true})
+            {
+                return;
+            }
 
-        var torrentTitle = filenameElement.GetString();
+            var torrentType = parsedTorrent.Response.IsMovie ? "movie" : "tvSeries";
+            var cacheKey = GetCacheKey(torrentType, parsedTorrent.Response.ParsedTitle, parsedTorrent.Response.Year);
+            var (cached, cachedResult) = await CheckIfInCacheAndReturn(cacheKey);
 
-        if (torrentTitle.IsNullOrEmpty())
+            if (cached)
-        {
+            {
-            return null;
+                logger.LogInformation("[{ImdbId}] Found cached imdb result for {Title}", cachedResult.ImdbId, parsedTorrent.Response.ParsedTitle);
-        }
+                lock (ingestedTorrents)
-        
+                {
-        var parsedTorrent = rankTorrentName.Parse(torrentTitle.CleanTorrentTitleForImdb());
+                    ingestedTorrents.Add(MapToTorrent(cachedResult, dmmContent.Bytes, infoHash, parsedTorrent));
-        
+                }
-        if (!parsedTorrent.Success)
+                return;
-        {
+            }
-            return null;
-        }
-        
-        var (cached, cachedResult) = await CheckIfInCacheAndReturn(parsedTorrent.Response.ParsedTitle);
-        
-        if (cached)
-        {
-            logger.LogInformation("[{ImdbId}] Found cached imdb result for {Title}", cachedResult.ImdbId, parsedTorrent.Response.ParsedTitle);
-            return MapToTorrent(cachedResult, bytesElement, hashElement, parsedTorrent);
-        }
 
-        int? year = parsedTorrent.Response.Year != 0 ? parsedTorrent.Response.Year : null;
+            int? year = parsedTorrent.Response.Year != 0 ? parsedTorrent.Response.Year : null;
-        var imdbEntry = await Storage.FindImdbMetadata(parsedTorrent.Response.ParsedTitle, parsedTorrent.Response.IsMovie ? "movies" : "tv", year);
+            var imdbEntry = await Storage.FindImdbMetadata(parsedTorrent.Response.ParsedTitle, torrentType, year, ct);
 
-        if (imdbEntry is null)
+            if (imdbEntry is null)
-        {
+            {
-            return null;
+                return;
-        }
+            }
-        
-        await AddToCache(parsedTorrent.Response.ParsedTitle.ToLowerInvariant(), imdbEntry);
-        
-        logger.LogInformation("[{ImdbId}] Found best match for {Title}: {BestMatch} with score {Score}", imdbEntry.ImdbId, parsedTorrent.Response.ParsedTitle, imdbEntry.Title, imdbEntry.Score);
 
-        return MapToTorrent(imdbEntry, bytesElement, hashElement, parsedTorrent);
+            await AddToCache(cacheKey, imdbEntry);
+            logger.LogInformation("[{ImdbId}] Found best match for {Title}: {BestMatch} with score {Score}", imdbEntry.ImdbId, parsedTorrent.Response.ParsedTitle, imdbEntry.Title, imdbEntry.Score);
+            lock (ingestedTorrents)
+            {
+                ingestedTorrents.Add(MapToTorrent(imdbEntry, dmmContent.Bytes, infoHash, parsedTorrent));
+            }
+        });
+
+        return ingestedTorrents;
     }
 
-    private IngestedTorrent MapToTorrent(ImdbEntry result, JsonElement bytesElement, JsonElement hashElement, ParseTorrentTitleResponse parsedTorrent) =>
+    private IngestedTorrent MapToTorrent(ImdbEntry result, long size, string infoHash, ParseTorrentTitleResponse parsedTorrent) =>
         new()
         {
             Source = Source,
             Name = result.Title,
             Imdb = result.ImdbId,
-            Size = bytesElement.GetInt64().ToString(),
+            Size = size.ToString(),
-            InfoHash = hashElement.ToString(),
+            InfoHash = infoHash,
             Seeders = 0,
             Leechers = 0,
             Category = AssignCategory(result),
@@ -153,19 +194,19 @@ public partial class DebridMediaManagerCrawler(
         };
     
 
-    private Task AddToCache(string lowerCaseTitle, ImdbEntry best)
+    private Task AddToCache(string cacheKey, ImdbEntry best)
     {
         var cacheOptions = new DistributedCacheEntryOptions
         {
             AbsoluteExpirationRelativeToNow = TimeSpan.FromDays(1),
         };
         
-        return cache.SetStringAsync(lowerCaseTitle, JsonSerializer.Serialize(best), cacheOptions);
+        return cache.SetStringAsync(cacheKey, JsonSerializer.Serialize(best), cacheOptions);
     }
 
-    private async Task<(bool Success, ImdbEntry? Entry)> CheckIfInCacheAndReturn(string title)
+    private async Task<(bool Success, ImdbEntry? Entry)> CheckIfInCacheAndReturn(string cacheKey)
     {
-        var cachedImdbId = await cache.GetStringAsync(title.ToLowerInvariant());
+        var cachedImdbId = await cache.GetStringAsync(cacheKey);
         
         if (!string.IsNullOrEmpty(cachedImdbId))
         {
@@ -175,48 +216,36 @@ public partial class DebridMediaManagerCrawler(
         return (false, null);
     }
 
-    private async Task InsertTorrentsForPage(JsonDocument json)
+    private async Task<(bool Success, string? Name)> IsAlreadyIngested(string filename)
     {
-        var torrents = await json.RootElement.EnumerateArray()
+        var pageIngested = await Storage.PageIngested(filename);
-            .ToAsyncEnumerable()
-            .SelectAwait(async x => await ParseTorrent(x))
-            .Where(t => t is not null)
-            .ToListAsync();
 
-        if (torrents.Count == 0)
+        return (pageIngested, filename);
-        {
-            logger.LogWarning("No torrents found in {Source} response", Source);
-            return;
-        }
-
-        await InsertTorrents(torrents!);
-    }
-
-    private async Task<(bool Success, string? Name)> IsAlreadyIngested(JsonElement entry)
-    {
-        var name = entry.GetProperty("path").GetString();
-
-        if (string.IsNullOrEmpty(name))
-        {
-            return (false, null);
-        }
-
-        var pageIngested = await Storage.PageIngested(name);
-
-        return (pageIngested, name);
     }
     
     private static string AssignCategory(ImdbEntry entry) =>
-        entry.Category switch
+        entry.Category.ToLower() switch
         {
-            "movie" => "movies",
+            var category when string.Equals(category, "movie", StringComparison.OrdinalIgnoreCase) => "movies",
-            "tvMovie" => "movies",
+            var category when string.Equals(category, "tvSeries", StringComparison.OrdinalIgnoreCase) => "tv",
-            "tvSeries" => "tv",
-            "tvEpisode" => "tv",
-            "tvSpecial" => "tv",
-            "tvMiniSeries" => "tv",
-            "tv" => "tv",
-            "short" => "tv",
             _ => "unknown",
         };
+    
+    private static string GetCacheKey(string category, string title, int year) => $"{category.ToLowerInvariant()}:{year}:{title.ToLowerInvariant()}";
+    
+    private static ExtractedDMMContent? ParsePageContent(JsonElement item)
+    {
+        if (!item.TryGetProperty("filename", out var filenameElement) ||
+            !item.TryGetProperty("bytes", out var bytesElement) ||
+            !item.TryGetProperty("hash", out var hashElement))
+        {
+            return null;
+        }
+        
+        return new(filenameElement.GetString(), bytesElement.GetInt64(), hashElement.GetString());
+    }
+    
+    private record DmmContent(string Filename, long Bytes, ParseTorrentTitleResponse? ParseResponse);
+    private record ExtractedDMMContent(string Filename, long Bytes, string InfoHash);
+    private record RtnBatchProcessable(string InfoHash, string Filename);
 }

src/producer/src/Features/Crawlers/Dmm/GithubConfiguration.cs

@@ -1,9 +0,0 @@
-namespace Producer.Features.Crawlers.Dmm;
-
-public class GithubConfiguration
-{
-    private const string Prefix = "GITHUB";
-    private const string PatVariable = "PAT";
-
-    public string? PAT { get; init; } =  Prefix.GetOptionalEnvironmentVariableAsString(PatVariable);
-}

src/producer/src/Features/Crawlers/Dmm/IDMMFileDownloader.cs

@@ -0,0 +1,6 @@
+namespace Producer.Features.Crawlers.Dmm;
+
+public interface IDMMFileDownloader
+{
+    Task<string> DownloadFileToTempPath(CancellationToken cancellationToken);
+}

src/producer/src/Features/Crawlers/Dmm/ServiceCollectionExtensions.cs

@@ -0,0 +1,16 @@
+namespace Producer.Features.Crawlers.Dmm;
+
+public static class ServiceCollectionExtensions
+{
+    public static IServiceCollection AddDmmSupport(this IServiceCollection services)
+    {
+        services.AddHttpClient<IDMMFileDownloader, DMMFileDownloader>(DMMFileDownloader.ClientName, client =>
+        {
+            client.BaseAddress = new("https://github.com/debridmediamanager/hashlists/zipball/main/");
+            client.DefaultRequestHeaders.Add("Accept-Encoding", "gzip");
+            client.DefaultRequestHeaders.UserAgent.ParseAdd("curl");
+        });
+
+        return services;
+    }
+}

src/producer/src/Features/JobSupport/ServiceCollectionExtensions.cs

@@ -5,7 +5,6 @@ internal static class ServiceCollectionExtensions
     internal static IServiceCollection AddQuartz(this IServiceCollection services, IConfiguration configuration)
     {
         var scrapeConfiguration = services.LoadConfigurationFromConfig<ScrapeConfiguration>(configuration, ScrapeConfiguration.SectionName);
-        var githubConfiguration = services.LoadConfigurationFromEnv<GithubConfiguration>();
         var rabbitConfiguration = services.LoadConfigurationFromEnv<RabbitMqConfiguration>();
 
         var jobTypes = Assembly.GetAssembly(typeof(BaseJob))
@@ -19,18 +18,13 @@ internal static class ServiceCollectionExtensions
             services.AddTransient(type);
         }
 
-        if (!string.IsNullOrEmpty(githubConfiguration.PAT))
-        {
-            services.AddTransient<SyncDmmJob>();
-        }
-
         var openMethod = typeof(ServiceCollectionExtensions).GetMethod(nameof(AddJobWithTrigger), BindingFlags.NonPublic | BindingFlags.Static | BindingFlags.Instance);
 
         services.AddQuartz(
             quartz =>
             {
                 RegisterAutomaticRegistrationJobs(jobTypes, openMethod, quartz, scrapeConfiguration);
-                RegisterDmmJob(githubConfiguration, quartz, scrapeConfiguration);
+                RegisterDmmJob(quartz, scrapeConfiguration);
                 RegisterTorrentioJob(services, quartz, configuration, scrapeConfiguration);
                 RegisterPublisher(quartz, rabbitConfiguration);
             });
@@ -64,13 +58,8 @@ internal static class ServiceCollectionExtensions
         }
     }
 
-    private static void RegisterDmmJob(GithubConfiguration githubConfiguration, IServiceCollectionQuartzConfigurator quartz, ScrapeConfiguration scrapeConfiguration)
+    private static void RegisterDmmJob(IServiceCollectionQuartzConfigurator quartz, ScrapeConfiguration scrapeConfiguration) => 
-    {
+        AddJobWithTrigger<SyncDmmJob>(quartz, SyncDmmJob.Key, SyncDmmJob.Trigger, scrapeConfiguration);
-        if (!string.IsNullOrEmpty(githubConfiguration.PAT))
-        {
-            AddJobWithTrigger<SyncDmmJob>(quartz, SyncDmmJob.Key, SyncDmmJob.Trigger, scrapeConfiguration);
-        }
-    }
 
     private static void RegisterTorrentioJob(
         IServiceCollection services,

src/producer/src/GlobalUsings.cs

@@ -1,12 +1,12 @@
 // Global using directives
 
+global using System.Collections.Concurrent;
+global using System.IO.Compression;
 global using System.Reflection;
 global using System.Text;
 global using System.Text.Json;
 global using System.Text.RegularExpressions;
 global using System.Xml.Linq;
-global using FuzzySharp;
-global using FuzzySharp.Extractor;
 global using FuzzySharp.PreProcess;
 global using FuzzySharp.SimilarityRatio.Scorer;
 global using FuzzySharp.SimilarityRatio.Scorer.StrategySensitive;

src/producer/src/Program.cs

@@ -12,7 +12,8 @@ builder.Services
     .RegisterMassTransit()
     .AddDataStorage()
     .AddCrawlers()
+    .AddDmmSupport()
     .AddQuartz(builder.Configuration);
 
 var app = builder.Build();
-app.Run();
+app.Run();

src/producer/src/requirements.txt

@@ -1 +1 @@
-rank-torrent-name==0.1.9
+rank-torrent-name==0.2.13

src/qbit-collector/Dockerfile

@@ -9,12 +9,23 @@ RUN dotnet restore -a $TARGETARCH
 RUN dotnet publish -c Release --no-restore -o /src/out -a $TARGETARCH
 
 
-FROM mcr.microsoft.com/dotnet/aspnet:8.0-alpine
+FROM mcr.microsoft.com/dotnet/aspnet:8.0-alpine3.19
 
 WORKDIR /app
+
+ENV PYTHONUNBUFFERED=1
+
+RUN apk add --update --no-cache python3=~3.11.8-r0 py3-pip && ln -sf python3 /usr/bin/python
+
 COPY --from=build /src/out .
+
+RUN rm -rf /app/python && mkdir -p /app/python
+
+RUN pip3 install -r /app/requirements.txt -t /app/python
+
 RUN addgroup -S qbit && adduser -S -G qbit qbit
 USER qbit
 HEALTHCHECK --interval=30s --timeout=30s --start-period=5s --retries=3 \
   CMD pgrep -f dotnet || exit 1
+ENV PYTHONNET_PYDLL=/usr/lib/libpython3.11.so.1.0
 ENTRYPOINT ["dotnet", "QBitCollector.dll"]

src/qbit-collector/Extensions/ServiceCollectionExtensions.cs

@@ -13,11 +13,13 @@ public static class ServiceCollectionExtensions
     internal static IServiceCollection AddServiceConfiguration(this IServiceCollection services)
     {
         services.AddQBitTorrentClient();
-        services.AddSingleton<IParseTorrentTitle, ParseTorrentTitle>();
+        services.RegisterPythonEngine();
+        services.AddSingleton<IRankTorrentName, RankTorrentName>();
         services.AddSingleton<QbitRequestProcessor>();
         services.AddHttpClient();
         services.AddSingleton<ITrackersService, TrackersService>();
         services.AddHostedService<TrackersBackgroundService>();
+        services.AddHostedService<HousekeepingBackgroundService>();
 
         return services;
     }

src/qbit-collector/Features/Qbit/HousekeepingBackgroundService.cs

@@ -0,0 +1,52 @@
+namespace QBitCollector.Features.Qbit;
+
+public class HousekeepingBackgroundService(IQBittorrentClient client, ILogger<HousekeepingBackgroundService> logger) : BackgroundService
+{
+    protected override async Task ExecuteAsync(CancellationToken stoppingToken)
+    {
+        logger.LogInformation("Service is Running.");
+
+        await DoWork();
+
+        using PeriodicTimer timer = new(TimeSpan.FromMinutes(2));
+
+        try
+        {
+            while (await timer.WaitForNextTickAsync(stoppingToken))
+            {
+                await DoWork();
+            }
+        }
+        catch (OperationCanceledException)
+        {
+            logger.LogInformation("Service stopping.");
+        }
+    }
+
+    private async Task DoWork()
+    {
+        try
+        {
+            logger.LogInformation("Cleaning Stale Entries in Qbit...");
+
+            var torrents = await client.GetTorrentListAsync();
+
+            foreach (var torrentInfo in torrents)
+            {
+                if (!(torrentInfo.AddedOn < DateTimeOffset.UtcNow.AddMinutes(-1)))
+                {
+                    continue;
+                }
+                
+                logger.LogInformation("Torrent [{InfoHash}] Identified as stale because was added at {AddedOn}", torrentInfo.Hash, torrentInfo.AddedOn);
+
+                await client.DeleteAsync(new[] {torrentInfo.Hash}, deleteDownloadedData: true);
+                logger.LogInformation("Cleaned up stale torrent: [{InfoHash}]", torrentInfo.Hash);
+            }
+        }
+        catch (Exception e)
+        {
+            logger.LogError(e, "Error cleaning up stale torrents this interval.");
+        }
+    }
+}

src/qbit-collector/Features/Worker/QbitMetaToTorrentMeta.cs

@@ -3,10 +3,11 @@ namespace QBitCollector.Features.Worker;
 public static class QbitMetaToTorrentMeta
 {
     public static IReadOnlyList<TorrentFile> MapMetadataToFilesCollection(
-        IParseTorrentTitle torrentTitle,
+        IRankTorrentName rankTorrentName,
         Torrent torrent,
         string ImdbId,
-        IReadOnlyList<TorrentContent> Metadata)
+        IReadOnlyList<TorrentContent> Metadata,
+        ILogger<WriteQbitMetadataConsumer> logger)
     {
         try
         {
@@ -24,23 +25,31 @@ public static class QbitMetaToTorrentMeta
                     Size = metadataEntry.Size,
                 };
             
-                var parsedTitle = torrentTitle.Parse(file.Title);
+                var parsedTitle = rankTorrentName.Parse(file.Title, false);
+
+                if (!parsedTitle.Success)
+                {
+                    logger.LogWarning("Failed to parse title {Title} for metadata mapping", file.Title);
+                    continue;
+                }
             
-                file.ImdbSeason = parsedTitle.Seasons.FirstOrDefault();
+                file.ImdbSeason = parsedTitle.Response?.Season?.FirstOrDefault() ?? 0;
-                file.ImdbEpisode = parsedTitle.Episodes.FirstOrDefault();
+                file.ImdbEpisode = parsedTitle.Response?.Episode?.FirstOrDefault() ?? 0;
             
                 files.Add(file);
             }
 
             return files;
         }
-        catch (Exception)
+        catch (Exception ex)
         {
+            logger.LogWarning("Failed to map metadata to files collection: {Exception}", ex.Message);
             return [];
         }
     }
     
-    public static async Task<IReadOnlyList<SubtitleFile>> MapMetadataToSubtitlesCollection(IDataStorage storage, string InfoHash, IReadOnlyList<TorrentContent> Metadata)
+    public static async Task<IReadOnlyList<SubtitleFile>> MapMetadataToSubtitlesCollection(IDataStorage storage, string InfoHash, IReadOnlyList<TorrentContent> Metadata, 
+        ILogger<WriteQbitMetadataConsumer> logger)
     {
         try
         {
@@ -70,8 +79,9 @@ public static class QbitMetaToTorrentMeta
 
             return files;
         }
-        catch (Exception)
+        catch (Exception ex)
         {
+            logger.LogWarning("Failed to map metadata to subtitles collection: {Exception}", ex.Message);
             return [];
         }
     }

src/qbit-collector/Features/Worker/QbitMetadataSagaStateMachine.cs

@@ -53,6 +53,12 @@ public class QbitMetadataSagaStateMachine : MassTransitStateMachine<QbitMetadata
                 .Then(
                     context =>
                     {
+                        if (!context.Message.WithFiles)
+                        {
+                            logger.LogInformation("No files written for torrent {InfoHash} in Saga {SagaId}", context.Saga.Torrent.InfoHash, context.Saga.CorrelationId);
+                            return;
+                        }
+                        
                         logger.LogInformation("Metadata Written for torrent {InfoHash} in Saga {SagaId}", context.Saga.Torrent.InfoHash, context.Saga.CorrelationId);
                     })
                 .TransitionTo(Completed)

src/qbit-collector/Features/Worker/Requests.cs

@@ -16,7 +16,7 @@ public record WriteQbitMetadata(Torrent Torrent, QBitMetadataResponse Metadata,
 }
 
 [EntityName("metadata-written-qbit-collector")]
-public record QbitMetadataWritten(QBitMetadataResponse Metadata) : CorrelatedBy<Guid>
+public record QbitMetadataWritten(QBitMetadataResponse Metadata, bool WithFiles) : CorrelatedBy<Guid>
 {
     public Guid CorrelationId { get; init; } = Metadata.CorrelationId;
 

src/qbit-collector/Features/Worker/WriteQbitMetadataConsumer.cs

@@ -1,25 +1,30 @@
 namespace QBitCollector.Features.Worker;
 
-public class WriteQbitMetadataConsumer(IParseTorrentTitle parseTorrentTitle, IDataStorage dataStorage) : IConsumer<WriteQbitMetadata>
+public class WriteQbitMetadataConsumer(IRankTorrentName rankTorrentName, IDataStorage dataStorage, ILogger<WriteQbitMetadataConsumer> logger) : IConsumer<WriteQbitMetadata>
 {
     public async Task Consume(ConsumeContext<WriteQbitMetadata> context)
     {
         var request = context.Message;
-        
-        var torrentFiles = QbitMetaToTorrentMeta.MapMetadataToFilesCollection(parseTorrentTitle, request.Torrent, request.ImdbId, request.Metadata.Metadata);
 
-        if (torrentFiles.Any())
+        var torrentFiles = QbitMetaToTorrentMeta.MapMetadataToFilesCollection(
+            rankTorrentName, request.Torrent, request.ImdbId, request.Metadata.Metadata, logger);
+
+        if (!torrentFiles.Any())
         {
-            await dataStorage.InsertFiles(torrentFiles);
+            await context.Publish(new QbitMetadataWritten(request.Metadata, false));
-
+            return;
-            var subtitles = await QbitMetaToTorrentMeta.MapMetadataToSubtitlesCollection(dataStorage, request.Torrent.InfoHash, request.Metadata.Metadata);
-
-            if (subtitles.Any())
-            {
-                await dataStorage.InsertSubtitles(subtitles);
-            }
         }
-        
+
-        await context.Publish(new QbitMetadataWritten(request.Metadata));
+        await dataStorage.InsertFiles(torrentFiles);
+
+        var subtitles = await QbitMetaToTorrentMeta.MapMetadataToSubtitlesCollection(
+            dataStorage, request.Torrent.InfoHash, request.Metadata.Metadata, logger);
+
+        if (subtitles.Any())
+        {
+            await dataStorage.InsertSubtitles(subtitles);
+        }
+
+        await context.Publish(new QbitMetadataWritten(request.Metadata, true));
     }
 }

src/qbit-collector/GlobalUsings.cs

@@ -1,17 +1,11 @@
 // Global using directives
 
 global using System.Text.Json;
-global using System.Text.Json.Serialization;
-global using System.Threading.Channels;
 global using MassTransit;
-global using MassTransit.Mediator;
 global using Microsoft.AspNetCore.Builder;
 global using Microsoft.Extensions.Caching.Distributed;
 global using Microsoft.Extensions.Caching.Memory;
 global using Microsoft.Extensions.DependencyInjection;
-global using Polly;
-global using Polly.Extensions.Http;
-global using PromKnight.ParseTorrentTitle;
 global using QBitCollector.Extensions;
 global using QBitCollector.Features.Qbit;
 global using QBitCollector.Features.Trackers;
@@ -21,4 +15,6 @@ global using SharedContracts.Configuration;
 global using SharedContracts.Dapper;
 global using SharedContracts.Extensions;
 global using SharedContracts.Models;
+global using SharedContracts.Python;
+global using SharedContracts.Python.RTN;
 global using SharedContracts.Requests;

src/qbit-collector/QBitCollector.csproj

@@ -18,7 +18,6 @@
       <PackageReference Include="Microsoft.Extensions.Http" Version="8.0.0" />
       <PackageReference Include="Microsoft.Extensions.Http.Polly" Version="8.0.3" />
       <PackageReference Include="Polly" Version="8.3.1" />
-      <PackageReference Include="PromKnight.ParseTorrentTitle" Version="1.0.4" />
       <PackageReference Include="QBittorrent.Client" Version="1.9.23349.1" />
       <PackageReference Include="Serilog" Version="3.1.1" />
       <PackageReference Include="Serilog.AspNetCore" Version="8.0.1" />
@@ -31,10 +30,30 @@
       <None Include="Configuration\logging.json">
         <CopyToOutputDirectory>Always</CopyToOutputDirectory>
       </None>
+      <Content Remove="eng\**" />
+      <None Remove="eng\**" />
+      <None Update="requirements.txt">
+        <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+      </None>
     </ItemGroup>
 
     <ItemGroup>
       <ProjectReference Include="..\shared\SharedContracts.csproj" />
     </ItemGroup>
 
+    <ItemGroup Condition="'$(Configuration)' == 'Debug'">
+        <Content Remove="python\**" />
+        <None Include="python\**">
+            <CopyToOutputDirectory>Always</CopyToOutputDirectory>
+        </None>
+    </ItemGroup>
+
+    <ItemGroup>
+      <Compile Remove="eng\**" />
+    </ItemGroup>
+
+    <ItemGroup>
+      <EmbeddedResource Remove="eng\**" />
+    </ItemGroup>
+
 </Project>

src/qbit-collector/QBitCollector.sln

@@ -6,6 +6,12 @@ Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "shared", "shared", "{2C0A0F
 EndProject
 Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "QBitCollector", "QBitCollector.csproj", "{1EF124BE-6EBE-4D9E-846C-FFF814999F3B}"
 EndProject
+Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "eng", "eng", "{2F2EA33A-1303-405D-939B-E9394D262BC9}"
+	ProjectSection(SolutionItems) = preProject
+		eng\install-python-reqs.ps1 = eng\install-python-reqs.ps1
+		eng\install-python-reqs.sh = eng\install-python-reqs.sh
+	EndProjectSection
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|Any CPU = Debug|Any CPU

src/qbit-collector/eng/install-python-reqs.ps1

@@ -0,0 +1,3 @@
+Remove-Item -Recurse -Force ../python
+mkdir -p ../python
+python -m pip install -r ../requirements.txt -t ../python/

src/qbit-collector/eng/install-python-reqs.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+rm -rf ../python
+mkdir -p ../python
+python3 -m pip install -r ../requirements.txt -t ../python/

src/qbit-collector/requirements.txt

@@ -0,0 +1 @@
+rank-torrent-name==0.2.13

src/shared/Dapper/DapperDataStorage.cs

@@ -118,7 +118,7 @@ public class DapperDataStorage(PostgresConfiguration configuration, RabbitMqConf
     public async Task<ImdbEntry?> FindImdbMetadata(string? parsedTorrentTitle, string torrentType, int? year, CancellationToken cancellationToken = default) =>
         await ExecuteCommandAsync(async connection =>
         {
-            var query = $"select \"imdb_id\" as \"ImdbId\", \"title\" as \"Title\", \"year\" as \"Year\", \"score\" as Score from search_imdb_meta('{parsedTorrentTitle.Replace("'", "").Replace("\"", "")}', '{(torrentType.Equals("movie", StringComparison.OrdinalIgnoreCase) ? "movie" : "tvSeries")}'";
+            var query = $"select \"imdb_id\" as \"ImdbId\", \"title\" as \"Title\", \"year\" as \"Year\", \"score\" as Score, \"category\" as Category from search_imdb_meta('{parsedTorrentTitle.Replace("'", "").Replace("\"", "")}', '{torrentType}'";
             query += year is not null ? $", {year}" : ", NULL";
             query += ", 1)";
             
@@ -167,12 +167,7 @@ public class DapperDataStorage(PostgresConfiguration configuration, RabbitMqConf
                     INSERT INTO subtitles
                         ("infoHash", "fileIndex", "fileId", "title")
                     VALUES
-                        (@InfoHash, @FileIndex, @FileId, @Title)
+                        (@InfoHash, @FileIndex, @FileId, @Title);
-                    ON CONFLICT
-                        ("infoHash", "fileIndex")
-                    DO UPDATE SET
-                        "fileId" = COALESCE(subtitles."fileId", EXCLUDED."fileId"),
-                        "title" = COALESCE(subtitles."title", EXCLUDED."title");
                     """;
                 
                 await connection.ExecuteAsync(query, subtitles);

src/shared/Extensions/DictionaryExtensions.cs

@@ -0,0 +1,19 @@
+namespace SharedContracts.Extensions;
+
+public static class DictionaryExtensions
+{
+    public static ConcurrentDictionary<TKey, TValue> ToConcurrentDictionary<TSource, TKey, TValue>(
+        this IEnumerable<TSource> source,
+        Func<TSource, TKey> keySelector,
+        Func<TSource, TValue> valueSelector) where TKey : notnull
+    {
+        var concurrentDictionary = new ConcurrentDictionary<TKey, TValue>();
+
+        foreach (var element in source)
+        {
+            concurrentDictionary.TryAdd(keySelector(element), valueSelector(element));
+        }
+
+        return concurrentDictionary;
+    }
+}

src/shared/Extensions/StringExtensions.cs

@@ -1,5 +1,3 @@
-using System.Text.RegularExpressions;
-
 namespace SharedContracts.Extensions;
 
 public static partial class StringExtensions

src/shared/GlobalUsings.cs

@@ -1,7 +1,9 @@
 // Global using directives
 
+global using System.Collections.Concurrent;
 global using System.Text.Json;
 global using System.Text.Json.Serialization;
+global using System.Text.RegularExpressions;
 global using Dapper;
 global using MassTransit;
 global using Microsoft.AspNetCore.Builder;

src/shared/Python/RTN/IRankTorrentName.cs

@@ -2,5 +2,6 @@ namespace SharedContracts.Python.RTN;
 
 public interface IRankTorrentName
 {
-    ParseTorrentTitleResponse Parse(string title);
+    ParseTorrentTitleResponse Parse(string title, bool trashGarbage = true, bool logErrors = false, bool throwOnErrors = false);
+    List<ParseTorrentTitleResponse?> BatchParse(IReadOnlyCollection<string> titles, int chunkSize = 500, int workers = 20, bool trashGarbage = true, bool logErrors = false, bool throwOnErrors = false);
 }

src/shared/Python/RTN/RankTorrentName.cs

@@ -12,33 +12,102 @@ public class RankTorrentName : IRankTorrentName
         _pythonEngineService = pythonEngineService;
         InitModules();
     }
-   
-    public ParseTorrentTitleResponse Parse(string title) =>
-        _pythonEngineService.ExecutePythonOperationWithDefault(
-            () =>
-            {
-                var result = _rtn?.parse(title);
-                return ParseResult(result);
-            }, new ParseTorrentTitleResponse(false, null), nameof(Parse), throwOnErrors: false, logErrors: false);
 
-
+    public ParseTorrentTitleResponse Parse(string title, bool trashGarbage = true, bool logErrors = false, bool throwOnErrors = false)
-    private static ParseTorrentTitleResponse ParseResult(dynamic result)
     {
-        if (result == null)
+        try
         {
+            using var gil = Py.GIL();
+            var result = _rtn?.parse(title, trashGarbage);
+            return ParseResult(result);
+        }
+        catch (Exception ex)
+        {
+            if (logErrors)
+            {
+                _pythonEngineService.Logger.LogError(ex, "Python Error: {Message} ({OperationName})", ex.Message, nameof(Parse));
+            }
+            
+            if (throwOnErrors)
+            {
+                throw;
+            }
+
             return new(false, null);
         }
+    }
+    
+    public List<ParseTorrentTitleResponse?> BatchParse(IReadOnlyCollection<string> titles, int chunkSize = 500, int workers = 20, bool trashGarbage = true, bool logErrors = false, bool throwOnErrors = false)
+    {
+        var responses = new List<ParseTorrentTitleResponse?>();
 
-        var json = result.model_dump_json()?.As<string?>();
+        try
-
-        if (json is null || string.IsNullOrEmpty(json))
         {
-            return new(false, null);
+            if (titles.Count == 0)
+            {
+                return responses;
+            }
+
+            using var gil = Py.GIL();
+            var pythonList = new PyList(titles.Select(x => new PyString(x).As<PyObject>()).ToArray());
+            PyList results = _rtn?.batch_parse(pythonList, trashGarbage, chunkSize, workers);
+
+            if (results == null)
+            {
+                return responses;
+            }
+
+            responses.AddRange(results.Select(ParseResult));
         }
+        catch (Exception ex)
+        {
+            if (logErrors)
+            {
+                _pythonEngineService.Logger.LogError(ex, "Python Error: {Message} ({OperationName})", ex.Message, nameof(Parse));
+            }
+            
+            if (throwOnErrors)
+            {
+                throw;
+            }
+        }
+
+        return responses;
+    }
+
+    private static ParseTorrentTitleResponse? ParseResult(dynamic result)
+    {
+        try
+        {
+            if (result == null)
+            {
+                return new(false, null);
+            }
+
+            var json = result.model_dump_json()?.As<string?>();
+
+            if (json is null || string.IsNullOrEmpty(json))
+            {
+                return new(false, null);
+            }
+
+            var mediaType = result.GetAttr("type")?.As<string>();
         
-        var response = JsonSerializer.Deserialize<RtnResponse>(json);
+            if (string.IsNullOrEmpty(mediaType))
-
+            {
-        return new(true, response);
+                return new(false, null);
+            }
+        
+            var response = JsonSerializer.Deserialize<RtnResponse>(json);
+        
+            response.IsMovie = mediaType.Equals("movie", StringComparison.OrdinalIgnoreCase);
+        
+            return new(true, response);
+        }
+        catch
+        {
+            return new(false, null);
+        }
     }
 
     private void InitModules() => 

src/shared/Python/RTN/RtnResponse.cs

@@ -76,8 +76,8 @@ public class RtnResponse
     
     [JsonPropertyName("extended")]
     public bool Extended { get; set; }
-    
+
-    public bool IsMovie => (Season == null && Episode == null) || (Season?.Count == 0 && Episode?.Count == 0);
+    public bool IsMovie { get; set; }
     
     public string ToJson() => this.AsJson();
 }

src/torrent-consumer/Extensions/ServiceCollectionExtensions.cs

@@ -82,11 +82,4 @@ public static class ServiceCollectionExtensions
 
         x.AddConsumer<PerformIngestionConsumer>();
     }
-
-    internal static IServiceCollection AddServiceConfiguration(this IServiceCollection services)
-    {
-        services.AddSingleton<IParseTorrentTitle, ParseTorrentTitle>();
-        
-        return services;
-    }
 }

src/torrent-consumer/GlobalUsings.cs

@@ -5,7 +5,6 @@ global using MassTransit;
 global using MassTransit.Mediator;
 global using Microsoft.AspNetCore.Builder;
 global using Microsoft.Extensions.DependencyInjection;
-global using PromKnight.ParseTorrentTitle;
 global using SharedContracts.Configuration;
 global using SharedContracts.Dapper;
 global using SharedContracts.Extensions;

src/torrent-consumer/Program.cs

@@ -10,7 +10,6 @@ builder.Host
 
 builder.Services
     .RegisterMassTransit()
-    .AddServiceConfiguration()
     .AddDatabase();
 
 var app = builder.Build();

src/torrent-consumer/TorrentConsumer.csproj

@@ -16,7 +16,6 @@
       <PackageReference Include="Microsoft.Extensions.Http" Version="8.0.0" />
       <PackageReference Include="Microsoft.Extensions.Http.Polly" Version="8.0.3" />
       <PackageReference Include="Polly" Version="8.3.1" />
-      <PackageReference Include="PromKnight.ParseTorrentTitle" Version="1.0.4" />
       <PackageReference Include="Serilog" Version="3.1.1" />
       <PackageReference Include="Serilog.AspNetCore" Version="8.0.1" />
       <PackageReference Include="Serilog.Sinks.Console" Version="5.0.1" />