feat(organizations): migrate source storage to polymorphic records

src/organizations/source_identity.py

@@ -0,0 +1,52 @@
+"""Identity normalization helpers for organization source ingestion."""
+
+from __future__ import annotations
+
+import re
+
+VALID_INN_LENGTHS = frozenset({10, 12})
+VALID_KPP_LENGTH = 9
+VALID_OGRN_LENGTH = 13
+VALID_OGRIP_LENGTH = 15
+
+
+def digits(value: str | None) -> str:
+    """Return only decimal digits from a source identity value."""
+    return re.sub(r"\D+", "", str(value or ""))
+
+
+def normalize_identity_fields(
+    *,
+    inn: str | None = "",
+    kpp: str | None = "",
+    ogrn: str | None = "",
+    ogrip: str | None = "",
+) -> tuple[str, str, str, str]:
+    """Normalize parser identity values to canonical Organization constraints."""
+    inn_digits = digits(inn)
+    kpp_digits = digits(kpp)
+    ogrn_digits = digits(ogrn)
+    ogrip_digits = digits(ogrip)
+
+    normalized_ogrip = ""
+    if len(ogrip_digits) == VALID_OGRIP_LENGTH:
+        normalized_ogrip = ogrip_digits
+    elif len(ogrn_digits) == VALID_OGRIP_LENGTH:
+        normalized_ogrip = ogrn_digits
+
+    normalized_inn = inn_digits if len(inn_digits) in VALID_INN_LENGTHS else ""
+    normalized_kpp = (
+        ""
+        if normalized_ogrip
+        else kpp_digits
+        if len(kpp_digits) == VALID_KPP_LENGTH
+        else ""
+    )
+    normalized_ogrn = (
+        ""
+        if normalized_ogrip
+        else ogrn_digits
+        if len(ogrn_digits) == VALID_OGRN_LENGTH
+        else ""
+    )
+    return normalized_inn, normalized_kpp, normalized_ogrn, normalized_ogrip