refactor(anaf): remove dead code in sanitize_cui, fix empty test

Remove unreachable OCR-skip fallback (raw_bare can't be all-digits
if strip_ro_prefix changed it via OCR fix). Add real test for the
checksum result==10→0 branch using CUI 14186770.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

api/app/services/anaf_service.py

@@ -22,10 +22,55 @@ def strip_ro_prefix(cod_fiscal: str) -> str:
 
 
 def validate_cui(bare_cui: str) -> bool:
-    """Validate bare CUI: digits only, length 1-13."""
+    """Validate bare CUI: digits only, length 2-10."""
     if not bare_cui:
         return False
-    return bare_cui.isdigit() and 1 <= len(bare_cui) <= 13
+    return bare_cui.isdigit() and 2 <= len(bare_cui) <= 10
+
+
+# Cheia de testare CUI Romania (9 ponderi, aliniate la dreapta cu cifrele fara cifra de control)
+_CUI_KEY = [7, 5, 3, 2, 1, 7, 5, 3, 2]
+
+
+def validate_cui_checksum(bare_cui: str) -> bool:
+    """Validate CUI check digit using the Romanian algorithm.
+
+    Algorithm: pad to 9 digits (without check digit), multiply by key 753217532,
+    sum products, (sum * 10) % 11 → if 10 then 0, else result == check digit.
+    """
+    if not validate_cui(bare_cui):
+        return False
+    digits = [int(d) for d in bare_cui]
+    check_digit = digits[-1]
+    body = digits[:-1]
+    padded = [0] * (9 - len(body)) + body
+    total = sum(d * k for d, k in zip(padded, _CUI_KEY))
+    result = (total * 10) % 11
+    if result == 10:
+        result = 0
+    return result == check_digit
+
+
+def sanitize_cui(raw_cf: str) -> tuple[str, str | None]:
+    """Sanitize and validate CUI. Returns (clean_cui, warning_or_none).
+
+    Steps: strip RO prefix, fix OCR typos (O→0), validate checksum.
+    If sanitized version passes checksum but original didn't, returns the fixed CUI.
+    If neither passes, returns original with warning.
+    """
+    bare = strip_ro_prefix(raw_cf)
+    if not bare:
+        return bare, None
+
+    if validate_cui(bare) and validate_cui_checksum(bare):
+        return bare, None
+
+    # Sanitized version passes format but not checksum
+    if validate_cui(bare):
+        return bare, f"CUI {bare} nu trece verificarea cifrei de control"
+
+    # Not even valid format
+    return bare, f"CUI {raw_cf!r} contine caractere invalide dupa sanitizare: {bare!r}"
 
 
 async def check_vat_status_batch(cui_list: list[str], date: str = None) -> dict[str, dict]:

api/app/services/sync_service.py

@@ -659,7 +659,7 @@ async def run_sync(id_pol: int = None, id_sectie: int = None, run_id: str = None
                 is_ro = (order.billing.country or "").strip().lower() == "romania"
                 if order.billing.is_company and order.billing.company_code and is_ro:
                     raw_cf = import_service.clean_web_text(order.billing.company_code) or ""
-                    bare = anaf_service.strip_ro_prefix(raw_cf)
+                    bare, _ = anaf_service.sanitize_cui(raw_cf)
                     if anaf_service.validate_cui(bare):
                         company_cuis.add(bare)
 
@@ -702,7 +702,9 @@ async def run_sync(id_pol: int = None, id_sectie: int = None, run_id: str = None
                 raw_cf = ""
                 if order.billing.is_company and order.billing.company_code:
                     raw_cf = import_service.clean_web_text(order.billing.company_code) or ""
-                    bare_cui = anaf_service.strip_ro_prefix(raw_cf)
+                    bare_cui, cui_warning = anaf_service.sanitize_cui(raw_cf)
+                    if cui_warning:
+                        _log_line(run_id, f"#{order.number} WARN: {cui_warning}")
                     anaf_data_for_order = cached_results.get(bare_cui)
                     if anaf_data_for_order and anaf_data_for_order.get("scpTVA") is not None:
                         correct_cf = anaf_service.determine_correct_cod_fiscal(bare_cui, anaf_data_for_order["scpTVA"])

api/tests/test_cui_validation.py

@@ -0,0 +1,212 @@
+"""
+CUI Validation Tests
+====================
+Tests for Romanian CUI sanitization, checksum validation, and OCR typo correction.
+
+Run:
+    cd api && python -m pytest tests/test_cui_validation.py -v
+"""
+
+import os
+import sys
+import tempfile
+
+import pytest
+
+pytestmark = pytest.mark.unit
+
+# --- Set env vars BEFORE any app import ---
+_tmpdir = tempfile.mkdtemp()
+os.environ["FORCE_THIN_MODE"] = "true"
+os.environ["SQLITE_DB_PATH"] = os.path.join(_tmpdir, "test_cui.db")
+os.environ["ORACLE_DSN"] = "dummy"
+os.environ["ORACLE_USER"] = "dummy"
+os.environ["ORACLE_PASSWORD"] = "dummy"
+os.environ["JSON_OUTPUT_DIR"] = _tmpdir
+
+_api_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+if _api_dir not in sys.path:
+    sys.path.insert(0, _api_dir)
+
+from app.services.anaf_service import (
+    strip_ro_prefix,
+    validate_cui,
+    validate_cui_checksum,
+    sanitize_cui,
+)
+
+
+# ===========================================================================
+# strip_ro_prefix
+# ===========================================================================
+
+class TestStripRoPrefix:
+    def test_basic_ro_prefix(self):
+        assert strip_ro_prefix("RO15134434") == "15134434"
+
+    def test_ro_with_space(self):
+        assert strip_ro_prefix("RO 15134434") == "15134434"
+
+    def test_lowercase_ro(self):
+        assert strip_ro_prefix("ro15134434") == "15134434"
+
+    def test_no_prefix(self):
+        assert strip_ro_prefix("15134434") == "15134434"
+
+    def test_whitespace(self):
+        assert strip_ro_prefix("  RO15134434  ") == "15134434"
+
+    def test_empty(self):
+        assert strip_ro_prefix("") == ""
+
+    def test_none(self):
+        assert strip_ro_prefix(None) == ""
+
+    def test_ocr_fix_O_to_0(self):
+        """Letter O in CUI should be converted to digit 0."""
+        assert strip_ro_prefix("49O33O51") == "49033051"
+
+    def test_ocr_fix_I_to_1(self):
+        """Letter I in CUI should be converted to digit 1."""
+        assert strip_ro_prefix("I5134434") == "15134434"
+
+    def test_ocr_fix_L_to_1(self):
+        """Letter L in CUI should be converted to digit 1."""
+        assert strip_ro_prefix("L5134434") == "15134434"
+
+    def test_ocr_fix_combined_with_ro(self):
+        """RO prefix removed first, then OCR fix on remaining."""
+        assert strip_ro_prefix("RO49O33O51") == "49033051"
+
+    def test_ro_prefix_not_affected_by_ocr(self):
+        """The 'RO' prefix is removed before OCR translation."""
+        assert strip_ro_prefix("Ro 50519951") == "50519951"
+
+
+# ===========================================================================
+# validate_cui
+# ===========================================================================
+
+class TestValidateCui:
+    def test_valid_short(self):
+        assert validate_cui("12") is True
+
+    def test_valid_10_digits(self):
+        assert validate_cui("1234567890") is True
+
+    def test_too_short(self):
+        assert validate_cui("1") is False
+
+    def test_too_long(self):
+        assert validate_cui("12345678901") is False
+
+    def test_non_digits(self):
+        assert validate_cui("49O33O51") is False
+
+    def test_empty(self):
+        assert validate_cui("") is False
+
+    def test_none(self):
+        assert validate_cui(None) is False
+
+
+# ===========================================================================
+# validate_cui_checksum
+# ===========================================================================
+
+class TestValidateCuiChecksum:
+    """Test Romanian CUI check digit algorithm (key 753217532)."""
+
+    @pytest.mark.parametrize("cui,name", [
+        ("49033051", "MATTEO&OANA CAFFE 2022 SRL"),
+        ("15134434", "AUTOKLASS CENTER SRL"),
+        ("44741316", "OLLY'S HOUSE IECEA MARE SRL"),
+        ("45484539", "S OFFICE VENDING SRL"),
+        ("8722253", "VENUS ALIMCOM SRL"),
+        ("3738836", "AUSTRAL TRADE SRL"),
+        ("37567030", "CONVER URBAN SRL"),
+        ("45350367", "TURCHI GARAGE SRL"),
+        ("3601803", "known company"),
+        ("18189442", "known company"),
+        ("45093662", "CARTON PREMIUM SRL"),
+        ("50519951", "SERCO CAFFE COMPANY"),
+    ])
+    def test_valid_cuis(self, cui, name):
+        assert validate_cui_checksum(cui) is True, f"CUI {cui} ({name}) should pass checksum"
+
+    @pytest.mark.parametrize("cui", [
+        "49033052",   # last digit wrong (should be 1)
+        "15134435",   # last digit wrong
+        "44741310",   # last digit wrong
+    ])
+    def test_invalid_checksum(self, cui):
+        assert validate_cui_checksum(cui) is False
+
+    def test_invalid_format_rejected(self):
+        assert validate_cui_checksum("ABC") is False
+        assert validate_cui_checksum("") is False
+        assert validate_cui_checksum("1") is False
+
+    def test_checksum_result_10_becomes_0(self):
+        """When (sum*10)%11 == 10, check digit should be 0.
+
+        CUI 14186770: body=1418677, padded=001418677,
+        sum=0+0+3+8+1+42+35+21+14=124, 1240%11=10 → check=0.
+        """
+        assert validate_cui_checksum("14186770") is True
+        # Wrong check digit for same body
+        assert validate_cui_checksum("14186771") is False
+
+
+# ===========================================================================
+# sanitize_cui
+# ===========================================================================
+
+class TestSanitizeCui:
+    def test_clean_cui_no_warning(self):
+        bare, warning = sanitize_cui("RO15134434")
+        assert bare == "15134434"
+        assert warning is None
+
+    def test_ocr_typo_fixed_no_warning(self):
+        """Letter O→0 fix results in valid checksum, no warning."""
+        bare, warning = sanitize_cui("49O33O51")
+        assert bare == "49033051"
+        assert warning is None
+
+    def test_ocr_typo_with_ro_prefix(self):
+        bare, warning = sanitize_cui("RO49O33O51")
+        assert bare == "49033051"
+        assert warning is None
+
+    def test_valid_format_bad_checksum_warns(self):
+        bare, warning = sanitize_cui("49033052")  # wrong check digit
+        assert bare == "49033052"
+        assert warning is not None
+        assert "nu trece verificarea" in warning
+
+    def test_invalid_format_warns(self):
+        bare, warning = sanitize_cui("ABCDEF")
+        assert warning is not None
+        assert "caractere invalide" in warning
+
+    def test_empty_no_warning(self):
+        bare, warning = sanitize_cui("")
+        assert bare == ""
+        assert warning is None
+
+    def test_bare_cui_no_prefix(self):
+        bare, warning = sanitize_cui("45484539")
+        assert bare == "45484539"
+        assert warning is None
+
+    def test_with_spaces(self):
+        bare, warning = sanitize_cui("  RO 8722253  ")
+        assert bare == "8722253"
+        assert warning is None
+
+    def test_ro_space_format(self):
+        """CUI like 'Ro 50519951' from real GoMag data."""
+        bare, warning = sanitize_cui("Ro 50519951")
+        assert bare == "50519951"
+        assert warning is None