Finlytic/FinlyticSentiment/Services/SentimentBackgroundService.cs

using System.Collections.Concurrent;
using FinlyticCore.Dtos.News;
using FinlyticSentiment.Util;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Hosting;
using Microsoft.Extensions.Logging;

namespace FinlyticSentiment.Services;

/// <summary>
/// Background hosted worker executing periodic sentiment analysis sweeps on pending news articles
/// and processing real-time article broadcasts.
/// </summary>
public class SentimentBackgroundService : BackgroundService
{
    private readonly SentimentMqttClient _mqttClient;
    private readonly IFinBertAnalyzerService _analyzer;
    private readonly ISentimentStorageService _storage;
    private readonly IServiceScopeFactory _scopeFactory;
    private readonly ILogger<SentimentBackgroundService> _logger;

    // In-Memory Mutex / Cache zur Vermeidung doppelter Verarbeitung (Race Conditions zwischen Broadcast & Sweep)
    private static readonly ConcurrentDictionary<Guid, byte> ProcessingArticles = new();

    /// <summary>
    /// Initializes a new instance of the <see cref="SentimentBackgroundService"/> class.
    /// </summary>
    public SentimentBackgroundService(
        SentimentMqttClient mqttClient,
        IFinBertAnalyzerService analyzer,
        ISentimentStorageService storage,
        IServiceScopeFactory scopeFactory,
        ILogger<SentimentBackgroundService> logger)
    {
        _mqttClient = mqttClient;
        _analyzer = analyzer;
        _storage = storage;
        _scopeFactory = scopeFactory;
        _logger = logger;
    }

    /// <inheritdoc />
    protected override async Task ExecuteAsync(CancellationToken stoppingToken)
    {
        _logger.LogInformation("[{Channel}] FinlyticSentiment Background Service started.", "SentimentChannel");

        // Realtime Broadcast Event registrieren (Echtzeit-Artikel)
        _mqttClient.OnArticleReceived += async (article) =>
        {
            await ProcessSingleArticleAsync(article, stoppingToken);
        };

        // Kurze Initialisierungs-Verzögerung für die MQTT-Verbindung
        await Task.Delay(3000, stoppingToken);

        while (!stoppingToken.IsCancellationRequested)
        {
            int maxBatchSize;
            int sweepIntervalMinutes;

            using (var scope = _scopeFactory.CreateScope())
            {
                var settingsDb = scope.ServiceProvider.GetRequiredService<ISettingsDbService>();
                var settings = await settingsDb.GetSettingsAsync();
                maxBatchSize = settings.MaxBatchSize;
                sweepIntervalMinutes = settings.SweepIntervalMinutes;
            }

            var interval = TimeSpan.FromMinutes(Math.Max(1, sweepIntervalMinutes));

            try
            {
                await PerformSentimentSweepAsync(maxBatchSize, stoppingToken);
            }
            catch (OperationCanceledException) when (stoppingToken.IsCancellationRequested)
            {
                // Normales Beenden beim Stoppen des Hosts
                break;
            }
            catch (Exception ex)
            {
                _logger.LogError(ex, "[{Channel}] Unhandled exception encountered during sentiment sweep cycle.",
                    "SentimentChannel");
            }

            _logger.LogInformation("[{Channel}] Waiting {Minutes} minute(s) until next sentiment sweep...",
                "SentimentChannel", interval.TotalMinutes);

            // Verwendet PeriodicTimer oder CancellationToken-resistenten Delay
            using var timer = new PeriodicTimer(interval);
            try
            {
                await timer.WaitForNextTickAsync(stoppingToken);
            }
            catch (OperationCanceledException)
            {
                break;
            }
        }

        _logger.LogInformation("[{Channel}] FinlyticSentiment Background Service is shutting down gracefully.",
            "SentimentChannel");
    }

    /// <summary>
    /// Performs a single batch sweep of pending news articles.
    /// </summary>
    private async Task PerformSentimentSweepAsync(int maxBatchSize, CancellationToken cancellationToken)
    {
        _logger.LogInformation("[{Channel}] Starting sentiment sweep for pending news articles (Limit: {Limit})...",
            "SentimentChannel", maxBatchSize);

        List<NewsArticleDto> pendingArticles = await _mqttClient.GetPendingArticlesAsync(limit: maxBatchSize);
        if (pendingArticles.Count == 0)
        {
            _logger.LogInformation("[{Channel}] No pending news articles found in FinlyticNews.", "SentimentChannel");
            return;
        }

        _logger.LogInformation("[{Channel}] Retrieved {Count} pending article(s) for FinBERT sentiment evaluation.",
            "SentimentChannel", pendingArticles.Count);

        foreach (var article in pendingArticles)
        {
            if (cancellationToken.IsCancellationRequested) break;
            await ProcessSingleArticleAsync(article, cancellationToken);
        }
    }

    /// <summary>
    /// Processes FinBERT sentiment analysis for a single article, updates two-stage JSON summaries, and notifies FinlyticNews of status update.
    /// </summary>
    private async Task ProcessSingleArticleAsync(NewsArticleDto article, CancellationToken cancellationToken = default)
    {
        if (article == null || article.Id == Guid.Empty) return;

        // Deduplizierung: Verhindert, dass derselbe Artikel zeitgleich im Sweep & im Broadcast verarbeitet wird
        if (!ProcessingArticles.TryAdd(article.Id, 0))
        {
            _logger.LogDebug("[{Channel}] Article {Id} is already being processed. Skipping duplicate run.",
                "SentimentChannel", article.Id);
            return;
        }

        try
        {
            _logger.LogInformation("[{Channel}] Evaluating FinBERT sentiment for article: '{Title}' (ID: {Id})",
                "SentimentChannel", article.Title, article.Id);

            var finbert = await _analyzer.AnalyzeArticleAsync(article);

            if (finbert == null)
            {
                _logger.LogWarning(
                    "[{Channel}] FinBERT analysis returned NULL for article {Id} ('{Title}'). Aborting processing for this run.",
                    "SentimentChannel", article.Id, article.Title);
                return;
            }

            if (cancellationToken.IsCancellationRequested) return;

            // 1. Artikel-Level Sentiment speichern
            await _storage.SaveArticleSentimentAsync(article, finbert);

            // 2. ISIN- & Sektor-Summaries aktualisieren
            if (article.MatchedAssets != null && article.MatchedAssets.Count > 0)
            {
                foreach (var asset in article.MatchedAssets)
                {
                    if (cancellationToken.IsCancellationRequested) break;
                    if (string.IsNullOrWhiteSpace(asset.Isin)) continue;

                    await _storage.UpdateIsinSummaryAsync(
                        asset.Isin,
                        asset.Name,
                        "General",
                        article,
                        finbert);

                    await _storage.UpdateSectorSummaryAsync(
                        "General",
                        asset.Isin,
                        article.Id.ToString(),
                        finbert);
                }
            }

            if (cancellationToken.IsCancellationRequested) return;

            // 3. FinlyticNews über Erfolg informieren
            bool updated = await _mqttClient.UpdateArticleStatusAsync(article.Id, "Analyzed");
            if (updated)
            {
                _logger.LogInformation(
                    "[{Channel}] Article sentiment processed and status set to 'Analyzed' in FinlyticNews: {Title} (ID: {Id}) -> {Label}",
                    "SentimentChannel", article.Title, article.Id, finbert.Label);
            }
            else
            {
                _logger.LogWarning(
                    "[{Channel}] Failed to confirm status update to 'Analyzed' in FinlyticNews for article: {Id}",
                    "SentimentChannel", article.Id);
            }
        }
        catch (Exception ex)
        {
            _logger.LogError(ex, "[{Channel}] Error processing sentiment for article: {Id} ({Title})",
                "SentimentChannel", article.Id, article.Title);
        }
        finally
        {
            // Lock nach der Verarbeitung immer freigeben
            ProcessingArticles.TryRemove(article.Id, out _);
        }
    }
}