Initial commit: TrueCV CV verification platform
Clean architecture solution with: - Domain: Entities (User, CVCheck, CVFlag, CompanyCache) and Enums - Application: Service interfaces, DTOs, and models - Infrastructure: EF Core, Identity, Hangfire, external API clients, services - Web: Blazor Server UI with pages and components Features: - CV upload and parsing (PDF/DOCX) using Claude API - Employment verification against Companies House API - Timeline analysis for gaps and overlaps - Veracity scoring algorithm - Background job processing with Hangfire - Azure Blob Storage for file storage Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
241
src/TrueCV.Infrastructure/Jobs/ProcessCVCheckJob.cs
Normal file
241
src/TrueCV.Infrastructure/Jobs/ProcessCVCheckJob.cs
Normal file
@@ -0,0 +1,241 @@
|
||||
using System.Text.Json;
|
||||
using Microsoft.EntityFrameworkCore;
|
||||
using Microsoft.Extensions.Logging;
|
||||
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 ITimelineAnalyserService _timelineAnalyserService;
|
||||
private readonly ILogger<ProcessCVCheckJob> _logger;
|
||||
|
||||
private static readonly JsonSerializerOptions JsonOptions = new()
|
||||
{
|
||||
PropertyNamingPolicy = JsonNamingPolicy.CamelCase,
|
||||
WriteIndented = true
|
||||
};
|
||||
|
||||
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;
|
||||
|
||||
public ProcessCVCheckJob(
|
||||
ApplicationDbContext dbContext,
|
||||
IFileStorageService fileStorageService,
|
||||
ICVParserService cvParserService,
|
||||
ICompanyVerifierService companyVerifierService,
|
||||
ITimelineAnalyserService timelineAnalyserService,
|
||||
ILogger<ProcessCVCheckJob> logger)
|
||||
{
|
||||
_dbContext = dbContext;
|
||||
_fileStorageService = fileStorageService;
|
||||
_cvParserService = cvParserService;
|
||||
_companyVerifierService = companyVerifierService;
|
||||
_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);
|
||||
|
||||
_logger.LogDebug(
|
||||
"Parsed CV for check {CheckId}: {EmploymentCount} employment entries",
|
||||
cvCheckId, cvData.Employment.Count);
|
||||
|
||||
// Step 4: Save extracted data
|
||||
cvCheck.ExtractedDataJson = JsonSerializer.Serialize(cvData, JsonOptions);
|
||||
await _dbContext.SaveChangesAsync(cancellationToken);
|
||||
|
||||
// Step 5: Verify each employment entry
|
||||
var verificationResults = new List<CompanyVerificationResult>();
|
||||
foreach (var employment in cvData.Employment)
|
||||
{
|
||||
var result = await _companyVerifierService.VerifyCompanyAsync(
|
||||
employment.CompanyName,
|
||||
employment.StartDate,
|
||||
employment.EndDate);
|
||||
|
||||
verificationResults.Add(result);
|
||||
|
||||
_logger.LogDebug(
|
||||
"Verified {Company}: {IsVerified} (Score: {Score}%)",
|
||||
employment.CompanyName, result.IsVerified, result.MatchScore);
|
||||
}
|
||||
|
||||
// Step 6: 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 7: Calculate veracity score
|
||||
var (score, flags) = CalculateVeracityScore(verificationResults, timelineAnalysis);
|
||||
|
||||
_logger.LogDebug("Calculated veracity score for check {CheckId}: {Score}", cvCheckId, score);
|
||||
|
||||
// Step 8: Create CVFlag records
|
||||
foreach (var flag in flags)
|
||||
{
|
||||
var cvFlag = new CVFlag
|
||||
{
|
||||
Id = Guid.NewGuid(),
|
||||
CVCheckId = cvCheckId,
|
||||
Category = Enum.Parse<FlagCategory>(flag.Category),
|
||||
Severity = Enum.Parse<FlagSeverity>(flag.Severity),
|
||||
Title = flag.Title,
|
||||
Description = flag.Description,
|
||||
ScoreImpact = flag.ScoreImpact
|
||||
};
|
||||
|
||||
_dbContext.CVFlags.Add(cvFlag);
|
||||
}
|
||||
|
||||
// Step 9: Generate veracity report
|
||||
var report = new VeracityReport
|
||||
{
|
||||
OverallScore = score,
|
||||
ScoreLabel = GetScoreLabel(score),
|
||||
EmploymentVerifications = verificationResults,
|
||||
TimelineAnalysis = timelineAnalysis,
|
||||
Flags = flags,
|
||||
GeneratedAt = DateTime.UtcNow
|
||||
};
|
||||
|
||||
cvCheck.ReportJson = JsonSerializer.Serialize(report, JsonOptions);
|
||||
cvCheck.VeracityScore = score;
|
||||
|
||||
// Step 10: 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;
|
||||
await _dbContext.SaveChangesAsync(cancellationToken);
|
||||
|
||||
throw;
|
||||
}
|
||||
}
|
||||
|
||||
private static (int Score, List<FlagResult> Flags) CalculateVeracityScore(
|
||||
List<CompanyVerificationResult> verifications,
|
||||
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 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 = overlap.Months - 2; // Allow 2 month transition
|
||||
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"
|
||||
};
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user