using System.Text.Json; using System.Text.RegularExpressions; using Api.Clients.Api.Contracts; using Api.Services.Contracts; using CvMatcher.Models.Responses; using CvSearch.Data; using CvSearch.Data.Entities; using CvMatcher.Models.Settings; using Microsoft.EntityFrameworkCore; using Microsoft.Extensions.Options; namespace Api.Services; public sealed class JobTokenService : IJobTokenService { private readonly CvSearchDbContext _db; private readonly IRagApiClient _rag; private readonly JobSearchSettings _settings; private readonly ILogger _logger; public JobTokenService( CvSearchDbContext db, IRagApiClient rag, IOptions settings, ILogger logger) { _db = db; _rag = rag; _settings = settings.Value; _logger = logger; } public async Task CreateTokenAsync(string cvDocumentId, string email, string language, CancellationToken ct) { var token = new JobSearchTokenEntity { Id = Guid.NewGuid().ToString("N"), CvDocumentId = cvDocumentId, Email = email, Language = language, ExpiresAt = DateTime.UtcNow.AddDays(_settings.TokenExpiryDays), Used = false, CreatedAt = DateTime.UtcNow }; _db.JobSearchTokens.Add(token); await _db.SaveChangesAsync(ct); _logger.LogInformation("Job search token created. TokenId={TokenId}, CvDocumentId={CvDocumentId}", token.Id, cvDocumentId); return token.Id; } public async Task TriggerStartAsync(string tokenId, CancellationToken ct) { var token = await _db.JobSearchTokens.FirstOrDefaultAsync(x => x.Id == tokenId, ct); if (token is null) return StartJobSearchStatus.NotFound; if (token.Used) return StartJobSearchStatus.AlreadyUsed; if (token.ExpiresAt <= DateTime.UtcNow) return StartJobSearchStatus.Expired; token.Used = true; await _db.SaveChangesAsync(ct); var cv = await _rag.GetDocumentAsync(token.CvDocumentId, ct); var keywords = cv is not null ? ExtractKeywords(cv.Text) : string.Empty; var providerConfigJson = JsonSerializer.Serialize( _settings.Providers.Where(p => p.Enabled).ToList(), new JsonSerializerOptions(JsonSerializerDefaults.Web)); var session = new JobSearchSessionEntity { Id = Guid.NewGuid().ToString("N"), TokenId = token.Id, CvDocumentId = token.CvDocumentId, Email = token.Email, Language = token.Language, Status = JobSearchStatus.Pending, Keywords = keywords, ProviderConfigJson = providerConfigJson, CreatedAt = DateTime.UtcNow }; _db.JobSearchSessions.Add(session); await _db.SaveChangesAsync(ct); _logger.LogInformation("Job search session created. SessionId={SessionId}, Keywords={Keywords}", session.Id, keywords); return StartJobSearchStatus.Started; } private static string ExtractKeywords(string cvText) { var lines = cvText .Split(['\n', '\r'], StringSplitOptions.RemoveEmptyEntries) .Select(l => l.Trim()) .Where(l => l.Length > 5 && l.Length < 200) // Skip lines that are purely digits, spaces, and phone/contact punctuation (phone numbers, emails, etc.) .Where(l => !Regex.IsMatch(l, @"^[\d\s\+\-\(\)\@\.]+$")) .Take(5) .ToList(); var words = lines .SelectMany(l => l.Split(' ', StringSplitOptions.RemoveEmptyEntries)) .Select(w => Regex.Replace(w, @"[^\w\-]", "")) .Where(w => w.Length > 2) .Distinct(StringComparer.OrdinalIgnoreCase) .Take(10) .ToList(); return string.Join(",", words); } }