Introduce max batch size, and configurable publish window

Still need to implement queue size limit
Also fixes env var consistency between addon and consumer

env/consumer.env

@@ -1,6 +1,6 @@
 TZ=London/Europe
 MONGODB_URI=mongodb://mongo:mongo@mongodb/selfhostio?tls=false&authSource=admin
-POSTGRES_DATABASE_URI=postgres://postgres:postgres@postgres/selfhostio
+DATABASE_URI=postgres://postgres:postgres@postgres/selfhostio
 RABBIT_URI=amqp://guest:guest@rabbitmq:5672/?heartbeat=30
 QUEUE_NAME=ingested
 JOB_CONCURRENCY=5

env/producer.env

@@ -1,4 +1,10 @@
 ScrapeConfiguration__StorageConnectionString=host=postgres;username=postgres;password=postgres;database=selfhostio;
 RabbitMqConfiguration__Host=rabbitmq
 RabbitMqConfiguration__QueueName=ingested
+RabbitMqConfiguration__Username=guest
+RabbitMqConfiguration__Password=guest
+RabbitMqConfiguration__Durable=true
+RabbitMqConfiguration__MaxQueueSize=1000
+RabbitMqConfiguration__MaxPublishBatchSize=100
+RabbitMqConfiguration__PublishIntervalInSeconds=10
 GithubSettings__PAT=

src/node/consumer/lib/config.js

@@ -10,7 +10,7 @@ export const cacheConfig = {
 }
 
 export const databaseConfig = {
-    DATABASE_URI: process.env.POSTGRES_DATABASE_URI || 'postgres://postgres:postgres@localhost:5432/selfhostio',
+    DATABASE_URI: process.env.DATABASE_URI || 'postgres://postgres:postgres@localhost:5432/selfhostio',
     ENABLE_SYNC: parseBool(process.env.ENABLE_SYNC, true)
 }
 

src/producer/Configuration/rabbitmq.json

@@ -4,6 +4,9 @@
     "Username": "guest",
     "Password": "guest",
     "QueueName": "test-queue",
-    "Durable": true
+    "Durable": true,
+    "MaxQueueSize": 1000,
+    "MaxPublishBatchSize": 100,
+    "PublishIntervalInSeconds": 10
   }
 }

src/producer/Crawlers/BaseCrawler.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Crawlers;
+namespace Producer.Crawlers;
 
 public abstract class BaseCrawler(ILogger<BaseCrawler> logger, IDataStorage storage) : ICrawler
 {

src/producer/Crawlers/BaseJsonCrawler.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Crawlers;
+namespace Producer.Crawlers;
 
 public abstract class BaseJsonCrawler(IHttpClientFactory httpClientFactory, ILogger<BaseJsonCrawler> logger, IDataStorage storage) : BaseCrawler(logger, storage)
 {

src/producer/Crawlers/BaseXmlCrawler.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Crawlers;
+namespace Producer.Crawlers;
 
 public abstract class BaseXmlCrawler(IHttpClientFactory httpClientFactory, ILogger<BaseXmlCrawler> logger, IDataStorage storage) : BaseCrawler(logger, storage)
 {

src/producer/Crawlers/CrawlerProvider.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Crawlers;
+namespace Producer.Crawlers;
 
 public class CrawlerProvider(IServiceProvider serviceProvider) : ICrawlerProvider
 {

src/producer/Crawlers/Sites/DebridMediaManagerCrawler.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Crawlers.Sites;
+namespace Producer.Crawlers.Sites;
 
 public partial class DebridMediaManagerCrawler(
     IHttpClientFactory httpClientFactory,
@@ -83,6 +83,7 @@ public partial class DebridMediaManagerCrawler(
         if (!result.Success)
         {
             logger.LogWarning("Failed to mark page as ingested: [{Error}]", result.ErrorMessage);
+            return;
         }
         
         logger.LogInformation("Successfully marked page as ingested");

src/producer/Crawlers/Sites/EzTvCrawler.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Crawlers.Sites;
+namespace Producer.Crawlers.Sites;
 
 public class EzTvCrawler(IHttpClientFactory httpClientFactory, ILogger<EzTvCrawler> logger, IDataStorage storage) : BaseXmlCrawler(httpClientFactory, logger, storage)
 {

src/producer/Crawlers/Sites/TgxCrawler.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Crawlers.Sites;
+namespace Producer.Crawlers.Sites;
 
 public partial class TgxCrawler(IHttpClientFactory httpClientFactory, ILogger<TgxCrawler> logger, IDataStorage storage) : BaseXmlCrawler(httpClientFactory, logger, storage)
 {

src/producer/Crawlers/Sites/TpbCrawler.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Crawlers.Sites;
+namespace Producer.Crawlers.Sites;
 
 public class TpbCrawler(IHttpClientFactory httpClientFactory, ILogger<TpbCrawler> logger, IDataStorage storage) : BaseJsonCrawler(httpClientFactory, logger, storage)
 {

src/producer/Crawlers/Sites/YtsCrawler.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Crawlers.Sites;
+namespace Producer.Crawlers.Sites;
 
 public class YtsCrawler(IHttpClientFactory httpClientFactory, ILogger<YtsCrawler> logger, IDataStorage storage) : BaseXmlCrawler(httpClientFactory, logger, storage)
 {

src/producer/Extensions/ConfigurationExtensions.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Extensions;
+namespace Producer.Extensions;
 
 public static class ConfigurationExtensions
 {

src/producer/Extensions/ConfigureHostBuilderExtensions.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Extensions;
+namespace Producer.Extensions;
 
 internal static class ConfigureHostBuilderExtensions
 {

src/producer/Extensions/ServiceCollectionExtensions.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Extensions;
+namespace Producer.Extensions;
 
 public static class ServiceCollectionExtensions
 {
@@ -53,6 +53,7 @@ public static class ServiceCollectionExtensions
     {
         var scrapeConfiguration = LoadScrapeConfiguration(services, configuration);
         var githubConfiguration = LoadGithubConfiguration(services, configuration);
+        var rabbitConfig = LoadRabbitMQConfiguration(services, configuration);
 
         services
             .AddTransient<SyncEzTvJob>()
@@ -75,7 +76,7 @@ public static class ServiceCollectionExtensions
                AddJobWithTrigger<SyncYtsJob>(quartz, SyncYtsJob.Key, SyncYtsJob.Trigger, scrapeConfiguration);
                AddJobWithTrigger<SyncTgxJob>(quartz, SyncTgxJob.Key, SyncTgxJob.Trigger, scrapeConfiguration);
                AddJobWithTrigger<IPJob>(quartz, IPJob.Key, IPJob.Trigger, 60 * 5);
-               AddJobWithTrigger<PublisherJob>(quartz, PublisherJob.Key, PublisherJob.Trigger, 10);
+               AddJobWithTrigger<PublisherJob>(quartz, PublisherJob.Key, PublisherJob.Trigger, rabbitConfig.PublishIntervalInSeconds);
 
                 if (!string.IsNullOrEmpty(githubConfiguration.PAT))
                 {
@@ -98,10 +99,21 @@ public static class ServiceCollectionExtensions
         
         ArgumentNullException.ThrowIfNull(githubConfiguration, nameof(githubConfiguration));
         
-        services.AddSingleton(githubConfiguration);
+        services.TryAddSingleton(githubConfiguration);
 
         return githubConfiguration;
     }
+    
+    private static RabbitMqConfiguration LoadRabbitMQConfiguration(IServiceCollection services, IConfiguration configuration)
+    {
+        var rabbitConfiguration = configuration.GetSection(RabbitMqConfiguration.SectionName).Get<RabbitMqConfiguration>();
+        
+        ArgumentNullException.ThrowIfNull(rabbitConfiguration, nameof(rabbitConfiguration));
+        
+        services.TryAddSingleton(rabbitConfiguration);
+
+        return rabbitConfiguration;
+    }
 
     private static ScrapeConfiguration LoadScrapeConfiguration(IServiceCollection services, IConfiguration configuration)
     {
@@ -109,7 +121,7 @@ public static class ServiceCollectionExtensions
         
         ArgumentNullException.ThrowIfNull(scrapeConfiguration, nameof(scrapeConfiguration));
 
-        services.AddSingleton(scrapeConfiguration);
+        services.TryAddSingleton(scrapeConfiguration);
 
         return scrapeConfiguration;
     }

src/producer/GlobalUsings.cs

@@ -9,14 +9,14 @@ global using LZStringCSharp;
 global using MassTransit;
 global using Microsoft.AspNetCore.Builder;
 global using Microsoft.Extensions.DependencyInjection;
+global using Microsoft.Extensions.DependencyInjection.Extensions;
 global using Microsoft.Extensions.Logging;
 global using Npgsql;
 global using Quartz;
-global using Scraper.Crawlers;
+global using Producer.Crawlers;
-global using Scraper.Crawlers.Sites;
+global using Producer.Crawlers.Sites;
-global using Scraper.Extensions;
+global using Producer.Interfaces;
-global using Scraper.Interfaces;
+global using Producer.Jobs;
-global using Scraper.Jobs;
+global using Producer.Models;
-global using Scraper.Models;
+global using Producer.Services;
-global using Scraper.Services;
 global using Serilog;

src/producer/Interfaces/ICrawler.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Interfaces;
+namespace Producer.Interfaces;
 
 public interface ICrawler
 {

src/producer/Interfaces/ICrawlerProvider.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Interfaces;
+namespace Producer.Interfaces;
 
 public interface ICrawlerProvider
 {

src/producer/Interfaces/IDataStorage.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Interfaces;
+namespace Producer.Interfaces;
 
 public interface IDataStorage
 {

src/producer/Interfaces/IIpService.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Interfaces;
+namespace Producer.Interfaces;
 
 public interface IIpService
 {

src/producer/Interfaces/IMessagePublisher.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Interfaces;
+namespace Producer.Interfaces;
 
 public interface IMessagePublisher
 {

src/producer/Jobs/BaseJob.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Jobs;
+namespace Producer.Jobs;
 
 public abstract class BaseJob(ICrawlerProvider crawlerProvider) : IJob
 {

src/producer/Jobs/ICrawlerJob.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Jobs;
+namespace Producer.Jobs;
 
 public interface ICrawlerJob<out TCrawler> : IJob
     where TCrawler : ICrawler

src/producer/Jobs/IPJob.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Jobs;
+namespace Producer.Jobs;
 
 [DisallowConcurrentExecution]
 public class IPJob(IIpService ipService) : IJob

src/producer/Jobs/PublisherJob.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Jobs;
+namespace Producer.Jobs;
 
 [DisallowConcurrentExecution]
 public class PublisherJob(IMessagePublisher publisher, IDataStorage storage, ILogger<PublisherJob> logger) : IJob

src/producer/Jobs/SyncDmmJob.cs

@@ -1,4 +1,6 @@
-namespace Scraper.Jobs;
+using DebridMediaManagerCrawler = Producer.Crawlers.Sites.DebridMediaManagerCrawler;
+
+namespace Producer.Jobs;
 
 [DisallowConcurrentExecution]
 public class SyncDmmJob(ICrawlerProvider crawlerProvider) : BaseJob(crawlerProvider)

src/producer/Jobs/SyncEzTvJob.cs

@@ -1,4 +1,6 @@
-namespace Scraper.Jobs;
+using Producer.Crawlers.Sites;
+
+namespace Producer.Jobs;
 
 [DisallowConcurrentExecution]
 public class SyncEzTvJob(ICrawlerProvider crawlerProvider) : BaseJob(crawlerProvider)

src/producer/Jobs/SyncTgxJob.cs

@@ -1,4 +1,6 @@
-namespace Scraper.Jobs;
+using TgxCrawler = Producer.Crawlers.Sites.TgxCrawler;
+
+namespace Producer.Jobs;
 
 [DisallowConcurrentExecution]
 public class SyncTgxJob(ICrawlerProvider crawlerProvider) : BaseJob(crawlerProvider)

src/producer/Jobs/SyncTpbJob.cs

@@ -1,4 +1,6 @@
-namespace Scraper.Jobs;
+using Producer.Crawlers.Sites;
+
+namespace Producer.Jobs;
 
 [DisallowConcurrentExecution]
 public class SyncTpbJob(ICrawlerProvider crawlerProvider) : BaseJob(crawlerProvider)

src/producer/Jobs/SyncYtsJob.cs

@@ -1,4 +1,6 @@
-namespace Scraper.Jobs;
+using Producer.Crawlers.Sites;
+
+namespace Producer.Jobs;
 
 [DisallowConcurrentExecution]
 public class SyncYtsJob(ICrawlerProvider crawlerProvider) : BaseJob(crawlerProvider)

src/producer/Models/GithubConfiguration.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Models;
+namespace Producer.Models;
 
 public class GithubConfiguration
 {

src/producer/Models/RabbitMqConfiguration.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Models;
+namespace Producer.Models;
 
 public class RabbitMqConfiguration
 {
@@ -10,4 +10,7 @@ public class RabbitMqConfiguration
     public string? Password { get; set; }
     public string? QueueName { get; set; }
     public bool Durable { get; set; }
+    public int MaxQueueSize { get; set; } = 1000;
+    public int MaxPublishBatchSize { get; set; } = 100;
+    public int PublishIntervalInSeconds { get; set; } = 1000 * 10;
 }

src/producer/Models/Results.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Models;
+namespace Producer.Models;
 
 public record InsertTorrentResult(bool Success, int InsertedCount = 0, string? ErrorMessage = null);
 public record UpdatedTorrentResult(bool Success, int UpdatedCount = 0, string? ErrorMessage = null);

src/producer/Models/ScrapeConfiguration.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Models;
+namespace Producer.Models;
 
 public class ScrapeConfiguration
 {

src/producer/Models/Scraper.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Models;
+namespace Producer.Models;
 
 public class Scraper
 {

src/producer/Models/Torrent.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Models;
+namespace Producer.Models;
 
 // Torrent represents a crawled torrent from one of our
 // supported sources.

src/producer/Program.cs

@@ -1,4 +1,6 @@
-var builder = WebApplication.CreateBuilder(args);
+using Producer.Extensions;
+
+var builder = WebApplication.CreateBuilder(args);
 
 builder.Configuration
     .AddScrapeConfiguration();

src/producer/Services/DapperDataStorage.cs

@@ -1,6 +1,6 @@
-namespace Scraper.Services;
+namespace Producer.Services;
 
-public class DapperDataStorage(ScrapeConfiguration configuration, ILogger<DapperDataStorage> logger) : IDataStorage
+public class DapperDataStorage(ScrapeConfiguration configuration, RabbitMqConfiguration rabbitConfig, ILogger<DapperDataStorage> logger) : IDataStorage
 {
     private const string InsertTorrentSql =
         """
@@ -65,7 +65,7 @@ public class DapperDataStorage(ScrapeConfiguration configuration, ILogger<Dapper
             await using var connection = new NpgsqlConnection(configuration.StorageConnectionString);
             await connection.OpenAsync(cancellationToken);
             var torrents = await connection.QueryAsync<Torrent>(GetMovieAndSeriesTorrentsNotProcessedSql);
-            return torrents.ToList();
+            return torrents.Take(rabbitConfig.MaxPublishBatchSize).ToList();
         }
         catch (Exception e)
         {

src/producer/Services/IpService.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Services;
+namespace Producer.Services;
 
 public class IpService(ILogger<IpService> logger, IHttpClientFactory httpClientFactory) : IIpService
 {

src/producer/Services/TorrentPublisher.cs

@@ -1,4 +1,4 @@
-namespace Scraper.Services;
+namespace Producer.Services;
 
 public class TorrentPublisher(ISendEndpointProvider sendEndpointProvider, RabbitMqConfiguration configuration) : IMessagePublisher
 {