RealCV/src/TrueCV.Infrastructure/Jobs/ProcessCVCheckJob.cs

using System.Text.Json;
using Microsoft.EntityFrameworkCore;
using Microsoft.Extensions.Logging;
using TrueCV.Application.Helpers;
using TrueCV.Application.Interfaces;
using TrueCV.Application.Models;
using TrueCV.Domain.Entities;
using TrueCV.Domain.Enums;
using TrueCV.Infrastructure.Data;

namespace TrueCV.Infrastructure.Jobs;

public sealed class ProcessCVCheckJob
{
    private readonly ApplicationDbContext _dbContext;
    private readonly IFileStorageService _fileStorageService;
    private readonly ICVParserService _cvParserService;
    private readonly ICompanyVerifierService _companyVerifierService;
    private readonly IEducationVerifierService _educationVerifierService;
    private readonly ITimelineAnalyserService _timelineAnalyserService;
    private readonly ILogger<ProcessCVCheckJob> _logger;

    private const int BaseScore = 100;
    private const int UnverifiedCompanyPenalty = 10;
    private const int GapMonthPenalty = 1;
    private const int MaxGapPenalty = 10;
    private const int OverlapMonthPenalty = 2;
    private const int DiplomaMillPenalty = 25;
    private const int SuspiciousInstitutionPenalty = 15;
    private const int UnverifiedEducationPenalty = 5;
    private const int EducationDatePenalty = 10;

    public ProcessCVCheckJob(
        ApplicationDbContext dbContext,
        IFileStorageService fileStorageService,
        ICVParserService cvParserService,
        ICompanyVerifierService companyVerifierService,
        IEducationVerifierService educationVerifierService,
        ITimelineAnalyserService timelineAnalyserService,
        ILogger<ProcessCVCheckJob> logger)
    {
        _dbContext = dbContext;
        _fileStorageService = fileStorageService;
        _cvParserService = cvParserService;
        _companyVerifierService = companyVerifierService;
        _educationVerifierService = educationVerifierService;
        _timelineAnalyserService = timelineAnalyserService;
        _logger = logger;
    }

    public async Task ExecuteAsync(Guid cvCheckId, CancellationToken cancellationToken)
    {
        _logger.LogInformation("Starting CV check processing for: {CheckId}", cvCheckId);

        var cvCheck = await _dbContext.CVChecks
            .FirstOrDefaultAsync(c => c.Id == cvCheckId, cancellationToken);

        if (cvCheck is null)
        {
            _logger.LogError("CV check not found: {CheckId}", cvCheckId);
            return;
        }

        try
        {
            // Step 1: Update status to Processing
            cvCheck.Status = CheckStatus.Processing;
            await _dbContext.SaveChangesAsync(cancellationToken);

            _logger.LogDebug("CV check {CheckId} status updated to Processing", cvCheckId);

            // Step 2: Download file from blob
            await using var fileStream = await _fileStorageService.DownloadAsync(cvCheck.BlobUrl);

            _logger.LogDebug("Downloaded CV file for check {CheckId}", cvCheckId);

            // Step 3: Parse CV
            var cvData = await _cvParserService.ParseAsync(fileStream, cvCheck.OriginalFileName, cancellationToken);

            _logger.LogDebug(
                "Parsed CV for check {CheckId}: {EmploymentCount} employment entries",
                cvCheckId, cvData.Employment.Count);

            // Step 4: Save extracted data
            cvCheck.ExtractedDataJson = JsonSerializer.Serialize(cvData, JsonDefaults.CamelCaseIndented);
            await _dbContext.SaveChangesAsync(cancellationToken);

            // Step 5: Verify each employment entry (parallelized with rate limiting)
            var verificationTasks = cvData.Employment.Select(async employment =>
            {
                var result = await _companyVerifierService.VerifyCompanyAsync(
                    employment.CompanyName,
                    employment.StartDate,
                    employment.EndDate);

                _logger.LogDebug(
                    "Verified {Company}: {IsVerified} (Score: {Score}%)",
                    employment.CompanyName, result.IsVerified, result.MatchScore);

                return result;
            });

            var verificationResults = (await Task.WhenAll(verificationTasks)).ToList();

            // Step 6: Verify education entries
            var educationResults = _educationVerifierService.VerifyAll(
                cvData.Education,
                cvData.Employment);

            _logger.LogDebug(
                "Education verification for check {CheckId}: {Count} entries verified ({Recognised} recognised, {DiplomaMill} diploma mills)",
                cvCheckId,
                educationResults.Count,
                educationResults.Count(e => e.IsVerified),
                educationResults.Count(e => e.IsDiplomaMill));

            // Step 7: Analyse timeline
            var timelineAnalysis = _timelineAnalyserService.Analyse(cvData.Employment);

            _logger.LogDebug(
                "Timeline analysis for check {CheckId}: {GapCount} gaps, {OverlapCount} overlaps",
                cvCheckId, timelineAnalysis.Gaps.Count, timelineAnalysis.Overlaps.Count);

            // Step 8: Calculate veracity score
            var (score, flags) = CalculateVeracityScore(verificationResults, educationResults, timelineAnalysis);

            _logger.LogDebug("Calculated veracity score for check {CheckId}: {Score}", cvCheckId, score);

            // Step 9: Create CVFlag records
            foreach (var flag in flags)
            {
                if (!Enum.TryParse<FlagCategory>(flag.Category, out var category))
                {
                    _logger.LogWarning("Unknown flag category: {Category}, defaulting to Timeline", flag.Category);
                    category = FlagCategory.Timeline;
                }

                if (!Enum.TryParse<FlagSeverity>(flag.Severity, out var severity))
                {
                    _logger.LogWarning("Unknown flag severity: {Severity}, defaulting to Info", flag.Severity);
                    severity = FlagSeverity.Info;
                }

                var cvFlag = new CVFlag
                {
                    Id = Guid.NewGuid(),
                    CVCheckId = cvCheckId,
                    Category = category,
                    Severity = severity,
                    Title = flag.Title,
                    Description = flag.Description,
                    ScoreImpact = flag.ScoreImpact
                };

                _dbContext.CVFlags.Add(cvFlag);
            }

            // Step 10: Generate veracity report
            var report = new VeracityReport
            {
                OverallScore = score,
                ScoreLabel = GetScoreLabel(score),
                EmploymentVerifications = verificationResults,
                EducationVerifications = educationResults,
                TimelineAnalysis = timelineAnalysis,
                Flags = flags,
                GeneratedAt = DateTime.UtcNow
            };

            cvCheck.ReportJson = JsonSerializer.Serialize(report, JsonDefaults.CamelCaseIndented);
            cvCheck.VeracityScore = score;

            // Step 11: Update status to Completed
            cvCheck.Status = CheckStatus.Completed;
            cvCheck.CompletedAt = DateTime.UtcNow;
            await _dbContext.SaveChangesAsync(cancellationToken);

            _logger.LogInformation(
                "CV check {CheckId} completed successfully with score {Score}",
                cvCheckId, score);
        }
        catch (Exception ex)
        {
            _logger.LogError(ex, "Error processing CV check {CheckId}", cvCheckId);

            cvCheck.Status = CheckStatus.Failed;
            // Use CancellationToken.None to ensure failure status is saved even if original token is cancelled
            await _dbContext.SaveChangesAsync(CancellationToken.None);

            throw;
        }
    }

    private static (int Score, List<FlagResult> Flags) CalculateVeracityScore(
        List<CompanyVerificationResult> verifications,
        List<EducationVerificationResult> educationResults,
        TimelineAnalysisResult timeline)
    {
        var score = BaseScore;
        var flags = new List<FlagResult>();

        // Penalty for unverified companies
        foreach (var verification in verifications.Where(v => !v.IsVerified))
        {
            score -= UnverifiedCompanyPenalty;

            flags.Add(new FlagResult
            {
                Category = FlagCategory.Employment.ToString(),
                Severity = FlagSeverity.Warning.ToString(),
                Title = "Unverified Company",
                Description = $"Could not verify employment at '{verification.ClaimedCompany}'. {verification.VerificationNotes}",
                ScoreImpact = -UnverifiedCompanyPenalty
            });
        }

        // Penalty for diploma mills (critical)
        foreach (var edu in educationResults.Where(e => e.IsDiplomaMill))
        {
            score -= DiplomaMillPenalty;

            flags.Add(new FlagResult
            {
                Category = FlagCategory.Education.ToString(),
                Severity = FlagSeverity.Critical.ToString(),
                Title = "Diploma Mill Detected",
                Description = $"'{edu.ClaimedInstitution}' is a known diploma mill. {edu.VerificationNotes}",
                ScoreImpact = -DiplomaMillPenalty
            });
        }

        // Penalty for suspicious institutions
        foreach (var edu in educationResults.Where(e => e.IsSuspicious && !e.IsDiplomaMill))
        {
            score -= SuspiciousInstitutionPenalty;

            flags.Add(new FlagResult
            {
                Category = FlagCategory.Education.ToString(),
                Severity = FlagSeverity.Warning.ToString(),
                Title = "Suspicious Institution",
                Description = $"'{edu.ClaimedInstitution}' has suspicious characteristics. {edu.VerificationNotes}",
                ScoreImpact = -SuspiciousInstitutionPenalty
            });
        }

        // Penalty for unverified education (not recognised, but not flagged as fake)
        foreach (var edu in educationResults.Where(e => !e.IsVerified && !e.IsDiplomaMill && !e.IsSuspicious && e.Status == "Unknown"))
        {
            score -= UnverifiedEducationPenalty;

            flags.Add(new FlagResult
            {
                Category = FlagCategory.Education.ToString(),
                Severity = FlagSeverity.Info.ToString(),
                Title = "Unverified Institution",
                Description = $"Could not verify '{edu.ClaimedInstitution}'. {edu.VerificationNotes}",
                ScoreImpact = -UnverifiedEducationPenalty
            });
        }

        // Penalty for implausible education dates
        foreach (var edu in educationResults.Where(e => !e.DatesArePlausible))
        {
            score -= EducationDatePenalty;

            flags.Add(new FlagResult
            {
                Category = FlagCategory.Education.ToString(),
                Severity = FlagSeverity.Warning.ToString(),
                Title = "Education Date Issues",
                Description = $"Date issues for '{edu.ClaimedInstitution}': {edu.DatePlausibilityNotes}",
                ScoreImpact = -EducationDatePenalty
            });
        }

        // Penalty for gaps (max -10 per gap)
        foreach (var gap in timeline.Gaps)
        {
            var gapPenalty = Math.Min(gap.Months * GapMonthPenalty, MaxGapPenalty);
            score -= gapPenalty;

            var severity = gap.Months >= 6 ? FlagSeverity.Warning : FlagSeverity.Info;

            flags.Add(new FlagResult
            {
                Category = FlagCategory.Timeline.ToString(),
                Severity = severity.ToString(),
                Title = "Employment Gap",
                Description = $"{gap.Months} month gap in employment from {gap.StartDate:MMM yyyy} to {gap.EndDate:MMM yyyy}",
                ScoreImpact = -gapPenalty
            });
        }

        // Penalty for overlaps (only if > 2 months)
        foreach (var overlap in timeline.Overlaps)
        {
            var excessMonths = Math.Max(0, overlap.Months - 2); // Allow 2 month transition, prevent negative
            var overlapPenalty = excessMonths * OverlapMonthPenalty;
            score -= overlapPenalty;

            var severity = overlap.Months >= 6 ? FlagSeverity.Critical : FlagSeverity.Warning;

            flags.Add(new FlagResult
            {
                Category = FlagCategory.Timeline.ToString(),
                Severity = severity.ToString(),
                Title = "Employment Overlap",
                Description = $"{overlap.Months} month overlap between '{overlap.Company1}' and '{overlap.Company2}' ({overlap.OverlapStart:MMM yyyy} to {overlap.OverlapEnd:MMM yyyy})",
                ScoreImpact = -overlapPenalty
            });
        }

        // Ensure score doesn't go below 0
        score = Math.Max(0, score);

        return (score, flags);
    }

    private static string GetScoreLabel(int score)
    {
        return score switch
        {
            >= 90 => "Excellent",
            >= 75 => "Good",
            >= 60 => "Fair",
            >= 40 => "Poor",
            _ => "Very Poor"
        };
    }
}