fix: upgrade all Finnish SSN related code to support new format

refs YJDH-686

backend/benefit/requirements.in

@@ -25,7 +25,7 @@ pdfkit
 pillow
 psycopg2
 python-dateutil
-python-stdnum
+python-stdnum>=1.19
 pyyaml
 requests
 sentry-sdk

backend/kesaseteli/applications/tests/test_youth_applications_api.py

@@ -1156,6 +1156,16 @@ def test_youth_applications_reactivate_unexpired_inactive__vtj_disabled(
         "121212A899H",
         "111111-111C",
         "111111A111C",
+        "111111B111C",
+        "111111C111C",
+        "111111D111C",
+        "111111E111C",
+        "111111F111C",
+        "111111U111C",
+        "111111V111C",
+        "111111W111C",
+        "111111X111C",
+        "111111Y111C",
         # Django Rest Framework's serializers.CharField trims leading and trailing
         # whitespace by default, so we allow them here.
         "     111111-111C",
@@ -1853,6 +1863,13 @@ def test_youth_application_post_empty_required_field(
         # Not uppercase
         "111111-111c",
         "111111a111C",
+        # Invalid century character
+        "111111/111C",
+        "111111G111C",
+        "111111M111C",
+        "111111R111C",
+        "111111T111C",
+        "111111Z111C",
         # Invalid checksum
         "111111-111X",  # "111111-111C" would be valid
         "111111A111W",  # "111111A111C" would be valid

backend/kesaseteli/common/tests/test_utils.py

@@ -159,7 +159,7 @@ def test_utc_datetime(args):
     assert result.tzinfo == timezone.utc
 
 
-@pytest.mark.parametrize("year", [1800, 2022, 2023, 2099])
+@pytest.mark.parametrize("year", [1800, 1987, 2022, 2023, 2024, 2025, 2099])
 def test_get_random_social_security_number_for_year(year):
     for _ in range(1000):
         result = get_random_social_security_number_for_year(year)

backend/kesaseteli/common/tests/utils.py

@@ -21,6 +21,7 @@ def get_random_social_security_number_for_year(year: int) -> str:
             tzinfo=timezone.utc,
         )
         .date(),
+        century_variant=get_faker().pyint(0, 99),  # Inclusive range
         individual_number=get_faker().pyint(2, 899),  # Inclusive range
     )
 

backend/kesaseteli/requirements.in

@@ -18,7 +18,7 @@ jsonpath-ng
 mozilla-django-oidc
 pillow
 psycopg2
-python-stdnum
+python-stdnum>=1.19
 requests
 sentry-sdk
 xlsx-streaming

backend/kesaseteli/requirements.txt

@@ -152,7 +152,7 @@ python-dateutil==2.8.1
     #   azure-storage-common
     #   faker
     #   pysaml2
-python-stdnum==1.17
+python-stdnum==1.19
     # via
     #   -r requirements.in
     #   django-localflavor

backend/shared/shared/common/tests/test_utils.py

@@ -97,28 +97,88 @@ def test_valid_social_security_number_birthdate(test_value, expected_result):
 
 
 @pytest.mark.parametrize(
-    "expected_result,birthdate,individual_number",
+    "expected_result,birthdate,century_variant,individual_number",
     [
-        ("010100+002H", date(year=1800, month=1, day=1), 2),
-        ("010203-1230", date(year=1903, month=2, day=1), 123),
-        ("121212A899H", date(year=2012, month=12, day=12), 899),
-        ("111111-002V", date(year=1911, month=11, day=11), 2),
-        ("111111-111C", date(year=1911, month=11, day=11), 111),
-        ("111111A111C", date(year=2011, month=11, day=11), 111),
-        ("111111-900U", date(year=1911, month=11, day=11), 900),
-        ("111111-9991", date(year=1911, month=11, day=11), 999),
-        ("300522A0024", date(year=2022, month=5, day=30), 2),
-        ("311299A999E", date(year=2099, month=12, day=31), 999),
+        ("010100+002H", date(year=1800, month=1, day=1), 0, 2),
+        ("010100+002H", date(year=1800, month=1, day=1), 99, 2),
+        ("010203-1230", date(year=1903, month=2, day=1), 0, 123),
+        ("121212A899H", date(year=2012, month=12, day=12), 0, 899),
+        ("121212A899H", date(year=2012, month=12, day=12), 30, 899),
+        ("121212B899H", date(year=2012, month=12, day=12), 1, 899),
+        ("121212B899H", date(year=2012, month=12, day=12), 31, 899),
+        ("121212C899H", date(year=2012, month=12, day=12), 2, 899),
+        ("121212C899H", date(year=2012, month=12, day=12), 32, 899),
+        ("121212D899H", date(year=2012, month=12, day=12), 3, 899),
+        ("121212D899H", date(year=2012, month=12, day=12), 33, 899),
+        ("121212E899H", date(year=2012, month=12, day=12), 4, 899),
+        ("121212E899H", date(year=2012, month=12, day=12), 34, 899),
+        ("121212F899H", date(year=2012, month=12, day=12), 5, 899),
+        ("121212F899H", date(year=2012, month=12, day=12), 35, 899),
+        ("111111-002V", date(year=1911, month=11, day=11), 0, 2),
+        ("111111-111C", date(year=1911, month=11, day=11), 0, 111),
+        ("111111Y111C", date(year=1911, month=11, day=11), 1, 111),
+        ("111111Y111C", date(year=1911, month=11, day=11), 61, 111),
+        ("111111X111C", date(year=1911, month=11, day=11), 2, 111),
+        ("111111X111C", date(year=1911, month=11, day=11), 62, 111),
+        ("111111W111C", date(year=1911, month=11, day=11), 3, 111),
+        ("111111W111C", date(year=1911, month=11, day=11), 63, 111),
+        ("111111V111C", date(year=1911, month=11, day=11), 4, 111),
+        ("111111V111C", date(year=1911, month=11, day=11), 64, 111),
+        ("111111U111C", date(year=1911, month=11, day=11), 5, 111),
+        ("111111U111C", date(year=1911, month=11, day=11), 65, 111),
+        ("111111A111C", date(year=2011, month=11, day=11), 0, 111),
+        ("111111-900U", date(year=1911, month=11, day=11), 0, 900),
+        ("111111-9991", date(year=1911, month=11, day=11), 0, 999),
+        ("300522A0024", date(year=2022, month=5, day=30), 0, 2),
+        ("300522B0024", date(year=2022, month=5, day=30), 1, 2),
+        ("311299A999E", date(year=2099, month=12, day=31), 0, 999),
+        ("311299F999E", date(year=2099, month=12, day=31), 5, 999),
+        # Go through all the possible checksum characters & before first & beyond last
+        ("111111B098Y", date(year=2011, month=11, day=11), 1, 98),
+        ("111111B0990", date(year=2011, month=11, day=11), 1, 99),
+        ("111111B1001", date(year=2011, month=11, day=11), 1, 100),
+        ("111111B1012", date(year=2011, month=11, day=11), 1, 101),
+        ("111111B1023", date(year=2011, month=11, day=11), 1, 102),
+        ("111111B1034", date(year=2011, month=11, day=11), 1, 103),
+        ("111111B1045", date(year=2011, month=11, day=11), 1, 104),
+        ("111111B1056", date(year=2011, month=11, day=11), 1, 105),
+        ("111111B1067", date(year=2011, month=11, day=11), 1, 106),
+        ("111111B1078", date(year=2011, month=11, day=11), 1, 107),
+        ("111111B1089", date(year=2011, month=11, day=11), 1, 108),
+        ("111111B109A", date(year=2011, month=11, day=11), 1, 109),
+        ("111111B110B", date(year=2011, month=11, day=11), 1, 110),
+        ("111111B111C", date(year=2011, month=11, day=11), 1, 111),
+        ("111111B112D", date(year=2011, month=11, day=11), 1, 112),
+        ("111111B113E", date(year=2011, month=11, day=11), 1, 113),
+        ("111111B114F", date(year=2011, month=11, day=11), 1, 114),
+        ("111111B115H", date(year=2011, month=11, day=11), 1, 115),
+        ("111111B116J", date(year=2011, month=11, day=11), 1, 116),
+        ("111111B117K", date(year=2011, month=11, day=11), 1, 117),
+        ("111111B118L", date(year=2011, month=11, day=11), 1, 118),
+        ("111111B119M", date(year=2011, month=11, day=11), 1, 119),
+        ("111111B120N", date(year=2011, month=11, day=11), 1, 120),
+        ("111111B121P", date(year=2011, month=11, day=11), 1, 121),
+        ("111111B122R", date(year=2011, month=11, day=11), 1, 122),
+        ("111111B123S", date(year=2011, month=11, day=11), 1, 123),
+        ("111111B124T", date(year=2011, month=11, day=11), 1, 124),
+        ("111111B125U", date(year=2011, month=11, day=11), 1, 125),
+        ("111111B126V", date(year=2011, month=11, day=11), 1, 126),
+        ("111111B127W", date(year=2011, month=11, day=11), 1, 127),
+        ("111111B128X", date(year=2011, month=11, day=11), 1, 128),
+        ("111111B129Y", date(year=2011, month=11, day=11), 1, 129),
+        ("111111B1300", date(year=2011, month=11, day=11), 1, 130),
     ],
 )
 def test_valid_create_finnish_social_security_number(
-    expected_result: str, birthdate: date, individual_number: int
+    expected_result: str, birthdate: date, century_variant: int, individual_number: int
 ):
     assert validate_finnish_social_security_number(
         expected_result, allow_temporary=True
     )
     assert (
-        create_finnish_social_security_number(birthdate, individual_number)
+        create_finnish_social_security_number(
+            birthdate, century_variant, individual_number
+        )
         == expected_result
     )
 
@@ -138,31 +198,37 @@ def test_valid_consecutive_create_finnish_social_security_number(birthdate: date
     Testing consecutive social security numbers for validity to ensure all their parts
     are calculated correctly, including the checksum (i.e. the last character).
     """
-    for individual_number in range(2, 900):
-        result = create_finnish_social_security_number(birthdate, individual_number)
-        assert len(result) == 11
-        assert validate_finnish_social_security_number(result, allow_temporary=True)
-        individual_number_from_result: int = int(result[-4:-1])
-        assert individual_number_from_result == individual_number
-        assert social_security_number_birthdate(result) == birthdate
+    for century_variant in range(0, 6):
+        for individual_number in range(2, 900):
+            result = create_finnish_social_security_number(
+                birthdate, century_variant, individual_number
+            )
+            assert len(result) == 11
+            assert validate_finnish_social_security_number(result, allow_temporary=True)
+            individual_number_from_result: int = int(result[-4:-1])
+            assert individual_number_from_result == individual_number
+            assert social_security_number_birthdate(result) == birthdate
 
 
 @pytest.mark.parametrize(
-    "birthdate,individual_number",
+    "birthdate,century_variant,individual_number",
     [
-        (date(year=1799, month=12, day=31), 2),  # Birth year < 1800
-        (date(year=2100, month=1, day=1), 2),  # Birth year > 2099
-        (date(year=2000, month=1, day=1), -1),  # Individual number < 2
-        (date(year=2000, month=1, day=1), 0),  # Individual number < 2
-        (date(year=2000, month=1, day=1), 1),  # Individual number < 2
-        (date(year=2000, month=1, day=1), 1000),  # Individual number > 999
+        (date(year=1799, month=12, day=31), 0, 2),  # Birth year < 1800
+        (date(year=2100, month=1, day=1), 0, 2),  # Birth year > 2099
+        (date(year=2000, month=1, day=1), -1, 0),  # Century variant < 0
+        (date(year=2000, month=1, day=1), 0, -1),  # Individual number < 2
+        (date(year=2000, month=1, day=1), 0, 0),  # Individual number < 2
+        (date(year=2000, month=1, day=1), 0, 1),  # Individual number < 2
+        (date(year=2000, month=1, day=1), 0, 1000),  # Individual number > 999
     ],
 )
 def test_invalid_create_finnish_social_security_number(
-    birthdate: date, individual_number: int
+    birthdate: date, century_variant: int, individual_number: int
 ):
     with pytest.raises(ValueError):
-        assert create_finnish_social_security_number(birthdate, individual_number)
+        assert create_finnish_social_security_number(
+            birthdate, century_variant, individual_number
+        )
 
 
 @pytest.mark.django_db
@@ -175,6 +241,12 @@ def test_invalid_create_finnish_social_security_number(
         "111111 -111x",  # Invalid checksum, inner whitespace
         "320522A002T",  # Invalid date because no 32 days in any month
         "311322A002E",  # Invalid date because no 13 months in any year
+        "111111/111C",  # Invalid century character
+        "111111G111C",  # Invalid century character
+        "111111M111C",  # Invalid century character
+        "111111R111C",  # Invalid century character
+        "111111T111C",  # Invalid century character
+        "111111Z111C",  # Invalid century character
         # Invalid checksum
         "111111-111X",  # "111111-111C" would be valid
         "111111A111W",  # "111111A111C" would be valid

backend/shared/shared/common/tests/utils.py

@@ -5,31 +5,54 @@
 
 
 def create_finnish_social_security_number(
-    birthdate: date, individual_number: int
+    birthdate: date, century_variant: int, individual_number: int
 ) -> str:
     """
-    Create a Finnish social security number based on birthdate and individual number
+    Create a Finnish social security number based on birthdate, century variant and
+    individual number
 
     :param birthdate: Date of birth
+    :param century_variant: Non-negative integer value for deciding which century
+                            character variant to use, e.g. birthdates in 2000–2099 may
+                            have century character A, B, C, D, E or F. There's no upper
+                            limit as (century_variant % available_century_characters) is
+                            the value being used.
     :param individual_number: Integer value where 2 <= individual_number <= 999
     :return: Finnish social security number in format <ddmmyyciiis> where
              dd = day of birth with leading zeroes,
              mm = month of birth with leading zeroes,
              yy = year of birth modulo 100 with leading zeroes,
-             c = century of birth ("+" = 1800, "-" = 1900, "A" = 2000),
+             c = century of birth, where c is
+                 "+" = 1800,
+                 "-" = 1900 if (century_variant % 6) == 0,
+                 "Y" = 1900 if (century_variant % 6) == 1,
+                 "X" = 1900 if (century_variant % 6) == 2,
+                 "W" = 1900 if (century_variant % 6) == 3,
+                 "V" = 1900 if (century_variant % 6) == 4,
+                 "U" = 1900 if (century_variant % 6) == 5,
+                 "A" = 2000 if (century_variant % 6) == 0,
+                 "B" = 2000 if (century_variant % 6) == 1,
+                 "C" = 2000 if (century_variant % 6) == 2,
+                 "D" = 2000 if (century_variant % 6) == 3,
+                 "E" = 2000 if (century_variant % 6) == 4,
+                 "F" = 2000 if (century_variant % 6) == 5,
              iii = individual number in range 2–999 with leading zeroes,
              s = checksum value calculated from <ddmmyyiii>.
     :raises ValueError: if not (1800 <= birthdate.year <= 2099)
+    :raises ValueError: if century_variant < 0
     :raises ValueError: if not (2 <= individual_number <= 999)
     """
     if not (1800 <= birthdate.year <= 2099):
         raise ValueError("Invalid birthdate year, only years 1800–2099 are supported")
+    if century_variant < 0:
+        raise ValueError("Invalid century variant, must be non-negative")
     if not (2 <= individual_number <= 999):
         raise ValueError("Invalid individual number, must be in range 2–999")
     ddmmyy: str = f"{birthdate.day:02}{birthdate.month:02}{birthdate.year % 100:02}"
     iii: str = f"{individual_number:003}"
     ddmmyyiii: str = f"{ddmmyy}{iii}"
-    century_char: str = {18: "+", 19: "-", 20: "A"}[birthdate.year // 100]
+    century_chars: str = {18: "+", 19: "-YXWVU", 20: "ABCDEF"}[birthdate.year // 100]
+    century_char: str = century_chars[century_variant % len(century_chars)]
     checksum_char: str = "0123456789ABCDEFHJKLMNPRSTUVWXY"[int(ddmmyyiii) % 31]
     return f"{ddmmyy}{century_char}{iii}{checksum_char}"
 

backend/shared/shared/common/utils.py

@@ -15,6 +15,24 @@
 
 _ALWAYS_FALSE_Q_FILTER = Q(pk=None)  # A hack but works as primary keys can't be null
 
+# Mapping from Finnish social security number century character to century number.
+# Should be the same as stdnum.fi.hetu._century_codes but that dictionary is private.
+_CENTURY_CHAR_TO_CENTURY_NUMBER = {
+    "+": 1800,
+    "-": 1900,
+    "A": 2000,
+    "B": 2000,
+    "C": 2000,
+    "D": 2000,
+    "E": 2000,
+    "F": 2000,
+    "U": 1900,
+    "V": 1900,
+    "W": 1900,
+    "X": 1900,
+    "Y": 1900,
+}
+
 
 def any_of_q_filter(**kwargs):
     """
@@ -123,5 +141,5 @@ def social_security_number_birthdate(social_security_number) -> date:
     month = int(compacted_social_security_number[2:4])
     year_mod_100 = int(compacted_social_security_number[4:6])
     century_char = compacted_social_security_number[6]
-    century = {"+": 1800, "-": 1900, "A": 2000}[century_char]
+    century = _CENTURY_CHAR_TO_CENTURY_NUMBER[century_char]
     return date(year=century + year_mod_100, month=month, day=day)

backend/tet/requirements.in

@@ -16,7 +16,7 @@ filetype
 mozilla-django-oidc
 pillow
 psycopg2
-python-stdnum
+python-stdnum>=1.19
 requests
 sentry-sdk
 xlsxwriter

backend/tet/requirements.txt

@@ -138,7 +138,7 @@ python-dateutil==2.8.1
     #   pysaml2
 python-jose==3.3.0
     # via django-helusers
-python-stdnum==1.17
+python-stdnum==1.19
     # via -r requirements.in
 pytz==2021.1
     # via

frontend/benefit/applicant/package.json

@@ -25,7 +25,7 @@
     "camelcase-keys": "^7.0.2",
     "date-fns": "^2.24.0",
     "dotenv": "^16.0.0",
-    "finnish-ssn": "^2.1.1",
+    "finnish-ssn": "^2.1.2",
     "formik": "^2.2.9",
     "hds-design-tokens": "^2.17.0",
     "hds-react": "^2.17.0",

frontend/benefit/handler/package.json

@@ -23,7 +23,7 @@
     "date-fns": "^2.24.0",
     "dotenv": "^16.0.0",
     "finnish-business-ids": "^3.1.1",
-    "finnish-ssn": "^2.1.1",
+    "finnish-ssn": "^2.1.2",
     "formik": "^2.2.9",
     "fuse.js": "^6.6.2",
     "hds-design-tokens": "^2.17.0",

frontend/kesaseteli/handler/package.json

@@ -23,7 +23,7 @@
     "@sentry/nextjs": "^7.16.0",
     "axios": "^0.27.2",
     "dotenv": "^16.0.0",
-    "finnish-ssn": "2.0.4",
+    "finnish-ssn": "^2.1.2",
     "hds-design-tokens": "^2.17.0",
     "hds-react": "^2.17.0",
     "lodash": "^4.17.21",

frontend/kesaseteli/shared/package.json

@@ -15,7 +15,7 @@
   "dependencies": {
     "@frontend/shared": "*",
     "axios": "^0.27.2",
-    "finnish-ssn": "2.0.4",
+    "finnish-ssn": "^2.1.2",
     "react-query": "^3.34.0"
   },
   "devDependencies": {

frontend/kesaseteli/youth/package.json

@@ -24,7 +24,7 @@
     "@sentry/nextjs": "^7.16.0",
     "axios": "^0.27.2",
     "dotenv": "^16.0.0",
-    "finnish-ssn": "2.0.4",
+    "finnish-ssn": "^2.1.2",
     "hds-design-tokens": "^2.17.0",
     "hds-react": "^2.17.0",
     "lodash": "^4.17.21",