feat: #46 auto-categorização no import por histórico de descrições

- TransactionRepository.CategoryByDescription: busca a categoria mais recente para cada descrição normalizada no histórico do perfil - ImportService.Preview: pré-preenche category_id em linhas não-duplicatas cuja descrição já foi categorizada antes (case-insensitive) - Duplicatas não recebem sugestão - 2 novos testes: auto-categorize + duplicate-not-categorized Co-Authored-By: Claude Sonnet 4.6 <[email protected]>
2026-05-28 22:49:44 -03:00
parent 26eabc460b
commit 00356971c7
4 changed files with 195 additions and 1 deletions
@@ -15,6 +15,8 @@ type TransactionRepository interface {
 	IsExternalIDKnown(ctx context.Context, externalID string) (bool, error)
 	BulkInsert(ctx context.Context, rows []model.ImportRow) (int, error)
 	SaveImportLog(ctx context.Context, filename, format string, imported, duplicates, errors int) error
+	// CategoryByDescription returns the most recently used category_id for each normalized description.
+	CategoryByDescription(ctx context.Context, descriptions []string) (map[string]int, error)
 }

 type transactionRepo struct{ db *pgxpool.Pool }
@@ -82,6 +84,42 @@ func (r *transactionRepo) BulkInsert(ctx context.Context, rows []model.ImportRow
 	return count, tx.Commit(ctx)
 }

+func (r *transactionRepo) CategoryByDescription(ctx context.Context, descriptions []string) (map[string]int, error) {
+	if len(descriptions) == 0 {
+		return map[string]int{}, nil
+	}
+	pid := middleware.ProfileIDFromCtx(ctx)
+
+	normed := make([]string, len(descriptions))
+	for i, d := range descriptions {
+		normed[i] = normalizeDesc(d)
+	}
+
+	rows, err := r.db.Query(ctx, `
+		SELECT DISTINCT ON (LOWER(description)) LOWER(description), category_id
+		FROM transactions
+		WHERE profile_id = $1
+		  AND category_id IS NOT NULL
+		  AND LOWER(description) = ANY($2)
+		ORDER BY LOWER(description), date DESC, id DESC
+	`, pid, normed)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	out := make(map[string]int)
+	for rows.Next() {
+		var desc string
+		var catID int
+		if err := rows.Scan(&desc, &catID); err != nil {
+			return nil, err
+		}
+		out[desc] = catID
+	}
+	return out, rows.Err()
+}
+
 func (r *transactionRepo) SaveImportLog(ctx context.Context, filename, format string, imported, duplicates, errors int) error {
 	_, err := r.db.Exec(ctx, `
 		INSERT INTO import_logs (filename, format, imported_count, duplicate_count, error_count)
@@ -22,6 +22,9 @@ func (m *mockImportTxRepo) BulkInsert(_ context.Context, rows []model.ImportRow)
 func (m *mockImportTxRepo) SaveImportLog(_ context.Context, _, _ string, _, _, _ int) error {
 	return nil
 }
+func (m *mockImportTxRepo) CategoryByDescription(_ context.Context, _ []string) (map[string]int, error) {
+	return map[string]int{}, nil
+}

 type mockDashboardRepo struct {
 	income   float64
@@ -12,6 +12,10 @@ import (
 	"financeiro-carvalho/internal/repository"
 )

+func normalizeDesc(s string) string {
+	return strings.ToLower(strings.Join(strings.Fields(s), " "))
+}
+
 type ImportService struct {
 	repo repository.TransactionRepository
 }
@@ -20,19 +24,37 @@ func NewImportService(repo repository.TransactionRepository) *ImportService {
 	return &ImportService{repo: repo}
 }

-// Preview parses the file and marks duplicates without saving anything.
+// Preview parses the file, marks duplicates, and pre-fills category_id from history.
 func (s *ImportService) Preview(ctx context.Context, filename string, r io.Reader, csvMapping *model.CSVMapping) ([]model.ImportRow, []string, error) {
 	rows, parseErrs, err := s.parse(filename, r, csvMapping)
 	if err != nil {
 		return nil, nil, err
 	}

+	// Collect unique descriptions for history lookup
+	descs := make([]string, 0, len(rows))
+	seen := make(map[string]bool)
+	for _, row := range rows {
+		n := normalizeDesc(row.Description)
+		if !seen[n] {
+			descs = append(descs, n)
+			seen[n] = true
+		}
+	}
+	catByDesc, _ := s.repo.CategoryByDescription(ctx, descs)
+
 	for i := range rows {
 		dup, err := s.isDuplicate(ctx, &rows[i])
 		if err != nil {
 			return nil, nil, fmt.Errorf("dedup check: %w", err)
 		}
 		rows[i].IsDuplicate = dup
+		// Pre-fill category from history for non-duplicate rows that have no category yet
+		if !dup && rows[i].CategoryID == nil {
+			if catID, ok := catByDesc[normalizeDesc(rows[i].Description)]; ok {
+				rows[i].CategoryID = &catID
+			}
+		}
 	}
 	return rows, parseErrs, nil
 }
@@ -0,0 +1,131 @@
+package service_test
+
+import (
+	"context"
+	"strings"
+	"testing"
+
+	"financeiro-carvalho/internal/model"
+	"financeiro-carvalho/internal/service"
+)
+
+// mockImportRepo implements repository.TransactionRepository for import tests.
+type mockImportRepo struct {
+	catByDesc map[string]int // normalized desc → category_id
+}
+
+func (m *mockImportRepo) IsDuplicate(_ context.Context, _, _ string, _ float64) (bool, error) {
+	return false, nil
+}
+func (m *mockImportRepo) IsExternalIDKnown(_ context.Context, _ string) (bool, error) {
+	return false, nil
+}
+func (m *mockImportRepo) BulkInsert(_ context.Context, rows []model.ImportRow) (int, error) {
+	return len(rows), nil
+}
+func (m *mockImportRepo) SaveImportLog(_ context.Context, _, _ string, _, _, _ int) error {
+	return nil
+}
+func (m *mockImportRepo) CategoryByDescription(_ context.Context, descs []string) (map[string]int, error) {
+	out := map[string]int{}
+	for _, d := range descs {
+		if id, ok := m.catByDesc[d]; ok {
+			out[d] = id
+		}
+	}
+	return out, nil
+}
+
+var csvMapping = model.CSVMapping{
+	DateColumn:        0,
+	AmountColumn:      1,
+	DescriptionColumn: 2,
+	DateFormat:        "02/01/2006",
+	HasHeader:         false,
+	DecimalSeparator:  ",",
+	FieldSeparator:    ";",
+}
+
+func TestPreview_AutoCategorize(t *testing.T) {
+	repo := &mockImportRepo{
+		catByDesc: map[string]int{
+			"padaria do ze": 3,
+			"uber *trip":    7,
+		},
+	}
+	svc := service.NewImportService(repo)
+
+	csv := "01/03/2024;-45,00;PADARIA DO ZE\n02/03/2024;-22,50;UBER *TRIP\n03/03/2024;-99,00;LOJA NOVA XYZ\n"
+
+	rows, _, err := svc.Preview(context.Background(), "fatura.csv", strings.NewReader(csv), &csvMapping)
+	if err != nil {
+		t.Fatalf("preview error: %v", err)
+	}
+	if len(rows) != 3 {
+		t.Fatalf("expected 3 rows, got %d", len(rows))
+	}
+
+	// PADARIA DO ZE → category 3
+	if rows[0].CategoryID == nil || *rows[0].CategoryID != 3 {
+		t.Errorf("row[0] (PADARIA DO ZE) expected category 3, got %v", rows[0].CategoryID)
+	}
+	// UBER *TRIP → category 7
+	if rows[1].CategoryID == nil || *rows[1].CategoryID != 7 {
+		t.Errorf("row[1] (UBER *TRIP) expected category 7, got %v", rows[1].CategoryID)
+	}
+	// LOJA NOVA XYZ → no suggestion
+	if rows[2].CategoryID != nil {
+		t.Errorf("row[2] (LOJA NOVA XYZ) expected no category, got %v", rows[2].CategoryID)
+	}
+}
+
+func TestPreview_DuplicateNotAutoCategorizated(t *testing.T) {
+	// row that IS a duplicate should not get category pre-filled
+	repo := &mockImportRepoDup{
+		catByDesc: map[string]int{"padaria do ze": 3},
+	}
+	svc := service.NewImportService(repo)
+
+	csv := "01/03/2024;-45,00;PADARIA DO ZE\n"
+
+	rows, _, err := svc.Preview(context.Background(), "fatura.csv", strings.NewReader(csv), &csvMapping)
+	if err != nil {
+		t.Fatalf("preview error: %v", err)
+	}
+	if len(rows) != 1 {
+		t.Fatalf("expected 1 row, got %d", len(rows))
+	}
+	if !rows[0].IsDuplicate {
+		t.Error("expected row to be marked duplicate")
+	}
+	if rows[0].CategoryID != nil {
+		t.Errorf("duplicate should not get category, got %v", rows[0].CategoryID)
+	}
+}
+
+// mockImportRepoDup marks all rows as duplicate (by date+amount+desc).
+type mockImportRepoDup struct {
+	catByDesc map[string]int
+}
+
+func (m *mockImportRepoDup) IsDuplicate(_ context.Context, _, _ string, _ float64) (bool, error) {
+	return true, nil // always duplicate
+}
+func (m *mockImportRepoDup) IsExternalIDKnown(_ context.Context, _ string) (bool, error) {
+	return false, nil
+}
+func (m *mockImportRepoDup) BulkInsert(_ context.Context, rows []model.ImportRow) (int, error) {
+	return len(rows), nil
+}
+func (m *mockImportRepoDup) SaveImportLog(_ context.Context, _, _ string, _, _, _ int) error {
+	return nil
+}
+func (m *mockImportRepoDup) CategoryByDescription(_ context.Context, descs []string) (map[string]int, error) {
+	out := map[string]int{}
+	for _, d := range descs {
+		if id, ok := m.catByDesc[d]; ok {
+			out[d] = id
+		}
+	}
+	return out, nil
+}