Purrse/src/Purrse.Api/Services/DuplicateDetectionService.cs

using Microsoft.EntityFrameworkCore;
using Purrse.Core.DTOs;
using Purrse.Core.Helpers;
using Purrse.Core.Interfaces.Services;
using Purrse.Core.Models;
using Purrse.Data;

namespace Purrse.Api.Services;

public class DuplicateDetectionService : IDuplicateDetectionService
{
    private readonly PurrseDbContext _db;

    public DuplicateDetectionService(PurrseDbContext db)
    {
        _db = db;
    }

    public async Task<DuplicateMatch?> FindDuplicateAsync(Guid accountId, DateTime date, decimal amount, string? payeeName, string? fitId, string? checkNumber)
    {
        var fingerprint = TransactionFingerprint.Generate(accountId, date, amount, fitId);

        // Check exact fingerprint match
        var fpMatch = await _db.Transactions.FirstOrDefaultAsync(t => t.AccountId == accountId && t.Fingerprint == fingerprint);
        if (fpMatch != null)
            return new DuplicateMatch(fpMatch.Id, 1.0, "Exact fingerprint match");

        // Check FITID match
        if (!string.IsNullOrEmpty(fitId))
        {
            var fitIdMatch = await _db.Transactions.FirstOrDefaultAsync(t => t.AccountId == accountId && t.FitId == fitId);
            if (fitIdMatch != null)
                return new DuplicateMatch(fitIdMatch.Id, 0.95, "Matching FITID/Reference");
        }

        // Multi-factor scoring on nearby transactions
        var candidates = await _db.Transactions
            .Where(t => t.AccountId == accountId && t.Date >= date.AddDays(-3) && t.Date <= date.AddDays(3) && !t.IsVoid)
            .ToListAsync();

        DuplicateMatch? bestMatch = null;

        foreach (var existing in candidates)
        {
            var confidence = CalculateConfidence(existing, date, amount, payeeName, fitId, checkNumber);
            if (confidence >= 0.50 && (bestMatch == null || confidence > bestMatch.Confidence))
            {
                var reasons = new List<string>();
                if (existing.Amount == amount) reasons.Add("amount");
                if (existing.Date.Date == date.Date) reasons.Add("date");
                if (!string.IsNullOrEmpty(payeeName) && !string.IsNullOrEmpty(existing.PayeeName))
                    reasons.Add("payee");

                bestMatch = new DuplicateMatch(existing.Id, confidence, $"Matched on: {string.Join(", ", reasons)}");
            }
        }

        return bestMatch;
    }

    public double CalculateConfidence(Transaction existing, DateTime date, decimal amount, string? payeeName, string? fitId, string? checkNumber)
    {
        double score = 0;

        // Amount matching
        if (existing.Amount == amount)
            score += 0.40;
        else if (Math.Abs(existing.Amount - amount) <= 0.01m)
            score += 0.35;

        // Date matching
        var daysDiff = Math.Abs((existing.Date.Date - date.Date).Days);
        if (daysDiff == 0) score += 0.30;
        else if (daysDiff <= 1) score += 0.20;
        else if (daysDiff <= 2) score += 0.10;

        // Payee matching (Jaro-Winkler)
        if (!string.IsNullOrEmpty(payeeName) && !string.IsNullOrEmpty(existing.PayeeName))
        {
            var similarity = StringSimilarityHelper.Calculate(payeeName, existing.PayeeName);
            if (similarity >= 0.85) score += 0.20;
            else if (similarity >= 0.70) score += 0.10;
        }

        // Check number matching
        if (!string.IsNullOrEmpty(checkNumber) && existing.CheckNumber == checkNumber)
            score += 0.10;

        return score;
    }
}