myAi/Jobs/cv-search-job/Tasks/CvSearchJobTask.cs

using System.Text.Json;
using CvMatcher.Models.Requests;
using CvSearch.Data;
using CvSearch.Data.Entities;
using CvMatcher.Models.Settings;
using CvSearchJob.Clients;
using CvSearchJob.Services;
using JobScheduler.Tasks;
using Microsoft.EntityFrameworkCore;
using Microsoft.Extensions.Configuration;
using Microsoft.Extensions.DependencyInjection;
using Microsoft.Extensions.Logging;
using Microsoft.Extensions.Options;

namespace CvSearchJob.Tasks;

/// <summary>
/// Background job task that processes pending job search sessions: scrapes providers,
/// scores each URL against the CV via the matcher API, persists results, and sends the results email.
/// </summary>
public sealed class CvSearchJobTask : IJobTask
{
    private readonly IServiceScopeFactory _scopeFactory;
    private readonly JobSearchSettings _settings;
    private readonly HtmlJobSearcher _searcher;
    private readonly ICvMatcherInternalApi _matcherApi;
    private readonly CvSearchEmailSender _emailSender;
    private readonly ILogger<CvSearchJobTask> _logger;

    public string TaskType => "CvSearch";

    public CvSearchJobTask(
        IServiceScopeFactory scopeFactory,
        IOptions<JobSearchSettings> settings,
        HtmlJobSearcher searcher,
        ICvMatcherInternalApi matcherApi,
        CvSearchEmailSender emailSender,
        ILogger<CvSearchJobTask> logger)
    {
        _scopeFactory = scopeFactory;
        _settings = settings.Value;
        _searcher = searcher;
        _matcherApi = matcherApi;
        _emailSender = emailSender;
        _logger = logger;
    }

    /// <summary>
    /// Called by the scheduler on each tick. Resets orphaned sessions, picks the oldest pending session,
    /// runs the full search pipeline, and sends the results email.
    /// Does nothing when <c>JobSearch:Enabled</c> is <c>false</c>.
    /// </summary>
    public async Task ExecuteAsync(IConfiguration parametersSection, CancellationToken cancellationToken)
    {
        if (!_settings.Enabled) return;

        using var scope = _scopeFactory.CreateScope();
        var db = scope.ServiceProvider.GetRequiredService<CvSearchDbContext>();

        // Recover orphaned Processing sessions (container crashed mid-run)
        var stuckCutoff = DateTime.UtcNow.AddMinutes(-10);
        var stuckSessions = await db.JobSearchSessions
            .Where(s => s.Status == JobSearchStatus.Processing && s.CreatedAt < stuckCutoff)
            .ToListAsync(cancellationToken);
        foreach (var stuck in stuckSessions)
        {
            stuck.Status = JobSearchStatus.Pending;
            _logger.LogWarning("Reset stuck session {SessionId} back to Pending", stuck.Id);
        }
        if (stuckSessions.Count > 0)
            await db.SaveChangesAsync(cancellationToken);

        var pending = await db.JobSearchSessions
            .Where(s => s.Status == JobSearchStatus.Pending)
            .OrderBy(s => s.CreatedAt)
            .Take(1)
            .FirstOrDefaultAsync(cancellationToken);

        if (pending is null) return;

        _logger.LogInformation("Processing job search session {SessionId}", pending.Id);
        pending.Status = JobSearchStatus.Processing;
        await db.SaveChangesAsync(cancellationToken);

        try
        {
            var cvKeywords = pending.Keywords
                .Split(',', StringSplitOptions.RemoveEmptyEntries)
                .Select(k => k.Trim())
                .Where(k => k.Length > 0)
                .ToList();

            var providers = GetProviders(pending.ProviderConfigJson);

            _logger.LogInformation(
                "Session {SessionId}: keywords=[{Keywords}] | providers=[{Providers}]",
                pending.Id,
                cvKeywords.Count > 0 ? string.Join(", ", cvKeywords) : "(none)",
                providers.Count > 0 ? string.Join(", ", providers.Select(p => p.Name)) : "(none)");

            var results = await RunSearchAsync(pending, cvKeywords, providers, db, cancellationToken);

            pending.Status = JobSearchStatus.Done;
            await db.SaveChangesAsync(cancellationToken);

            var attachmentFileName = BuildCvFileName(pending.CvDocumentId);
            await _emailSender.SendResultsAsync(
                pending.Email,
                attachmentFileName,
                results,
                cvKeywords,
                providers.Select(p => p.Name).ToList(),
                pending.Language,
                cancellationToken);

            _logger.LogInformation("Session {SessionId} done. {Count} results sent.", pending.Id, results.Count);
        }
        catch (Exception ex)
        {
            _logger.LogError(ex, "Session {SessionId} failed.", pending.Id);
            pending.Status = JobSearchStatus.Failed;
            await db.SaveChangesAsync(cancellationToken);
        }
    }

    /// <summary>
    /// Runs the full search pipeline for a session: scrapes all providers, deduplicates URLs,
    /// scores each candidate via the matcher API, and persists results that meet the minimum score threshold.
    /// </summary>
    private async Task<List<JobSearchResultEntity>> RunSearchAsync(
        JobSearchSessionEntity session,
        List<string> cvKeywords,
        List<JobProviderConfig> providers,
        CvSearchDbContext db,
        CancellationToken ct)
    {
        if (cvKeywords.Count == 0)
            _logger.LogWarning("Session {SessionId}: keyword list is empty — scraper will rely on provider InitialKeywords only", session.Id);

        var jobUrls = new HashSet<string>(StringComparer.OrdinalIgnoreCase);

        foreach (var provider in providers)
        {
            var urls = await _searcher.SearchJobUrlsAsync(provider, cvKeywords, ct);
            _logger.LogInformation("Session {SessionId}: provider {Provider} returned {Count} URLs", session.Id, provider.Name, urls.Count);
            foreach (var url in urls) jobUrls.Add(url);
        }

        var candidates = jobUrls.Take(_settings.MaxJobsToMatch).ToList();
        _logger.LogInformation(
            "Session {SessionId}: {Total} unique URLs across all providers, scoring {Scoring} (cap={Cap})",
            session.Id, jobUrls.Count, candidates.Count, _settings.MaxJobsToMatch);

        var results = new List<JobSearchResultEntity>();

        foreach (var url in candidates)
        {
            try
            {
                var matchRequest = new MatchJobRequest
                {
                    CvDocumentId = session.CvDocumentId,
                    JobUrl = url,
                    // User already gave GDPR consent when they clicked the one-time job search link
                    GdprConsent = true
                };

                var matchResult = await _matcherApi.MatchJobAsync(matchRequest, ct);

                _logger.LogInformation(
                    "Session {SessionId}: {Url} → score={Score}% (threshold={Threshold}%) {Verdict}",
                    session.Id, url, matchResult.Score, _settings.MinMatchScore,
                    matchResult.Score >= _settings.MinMatchScore ? "ACCEPTED" : "rejected");

                if (matchResult.Score < _settings.MinMatchScore)
                    continue;

                var entity = new JobSearchResultEntity
                {
                    Id = Guid.NewGuid().ToString("N"),
                    SessionId = session.Id,
                    ProviderName = GuessProvider(url, providers),
                    JobUrl = url,
                    JobTitle = matchResult.Summary.Split('.').FirstOrDefault()?.Trim() ?? "Job",
                    JobText = string.Empty,
                    Score = matchResult.Score,
                    ResultJson = JsonSerializer.Serialize(matchResult, new JsonSerializerOptions(JsonSerializerDefaults.Web)),
                    CreatedAt = DateTime.UtcNow
                };

                db.JobSearchResults.Add(entity);
                await db.SaveChangesAsync(ct);
                results.Add(entity);
            }
            catch (Exception ex)
            {
                _logger.LogWarning(ex, "Session {SessionId}: match failed for {Url}", session.Id, url);
            }
        }

        results.Sort((a, b) => b.Score.CompareTo(a.Score));
        return results;
    }

    /// <summary>
    /// Deserialises the provider configuration snapshot stored on the session.
    /// Falls back to the current live config when the snapshot is absent or unparseable.
    /// </summary>
    private List<JobProviderConfig> GetProviders(string? providerConfigJson)
    {
        if (string.IsNullOrWhiteSpace(providerConfigJson)) return _settings.Providers.Where(p => p.Enabled).ToList();
        try
        {
            return JsonSerializer.Deserialize<List<JobProviderConfig>>(providerConfigJson,
                       new JsonSerializerOptions(JsonSerializerDefaults.Web))
                   ?? _settings.Providers.Where(p => p.Enabled).ToList();
        }
        catch
        {
            return _settings.Providers.Where(p => p.Enabled).ToList();
        }
    }

    /// <summary>
    /// Infers the provider name from the job URL by matching against each provider's <c>JobLinkContains</c> pattern.
    /// Falls back to the URL hostname when no provider matches.
    /// </summary>
    private static string GuessProvider(string url, List<JobProviderConfig> providers)
    {
        foreach (var p in providers)
        {
            if (!string.IsNullOrWhiteSpace(p.JobLinkContains) &&
                url.Contains(p.JobLinkContains, StringComparison.OrdinalIgnoreCase))
                return p.Name;
        }

        return Uri.TryCreate(url, UriKind.Absolute, out var uri) ? uri.Host : "unknown";
    }

    /// <summary>
    /// Constructs the CV PDF filename from the document ID.
    /// </summary>
    private static string BuildCvFileName(string cvDocumentId)
    {
        // Strip non-alphanumeric characters so the filename is safe for all OS/email clients.
        var safeId = string.Concat(cvDocumentId.Where(char.IsLetterOrDigit));
        if (string.IsNullOrWhiteSpace(safeId)) safeId = "cv";
        return $"{safeId}.pdf";
    }
}