feat: validate cp1252 pydantic (#189)

* feat: add util for is_cp1252 encoding

* feat: add Cp1252Str type for main fields

* feat: test with not arabic chars desired fields

eox_nelp/pearson_vue/data_classes.py

@@ -1,19 +1,44 @@
 """
 Module to add data_classes related Pearson Vue Integration
 """
-from pydantic import BaseModel, Field
+from pydantic import AfterValidator, BaseModel, Field
+from typing_extensions import Annotated
+
+from eox_nelp.pearson_vue.utils import is_cp1252
+
+
+def validate_is_cp1252(text):
+    """Validate if a field is cp1252, raise exception if not.
+    Anyway return the same text.
+
+    Args:
+        text (str): input to be validated
+
+    Raises:
+        ValueError: if some char is not CP1252
+
+    Returns:
+        text (str)
+    """
+    if not is_cp1252(text):
+        raise ValueError(f"{text} -> some char is not CP1252")
+
+    return text
+
+
+Cp1252Str = Annotated[str, AfterValidator(validate_is_cp1252)]
 
 
 class Phone(BaseModel):
     """Phone data class model"""
-    phone_number: str = Field(alias="phoneNumber", min_length=1, max_length=20)
-    phone_country_code: str = Field(alias="phoneCountryCode", min_length=1, max_length=3)
+    phone_number: Cp1252Str = Field(alias="phoneNumber", min_length=1, max_length=20)
+    phone_country_code: Cp1252Str = Field(alias="phoneCountryCode", min_length=1, max_length=3)
 
 
 class Mobile(BaseModel):
     """Mobile data class model"""
-    mobile_number: str = Field(alias="mobileNumber", min_length=1, max_length=20)
-    mobile_country_code: str = Field(alias="mobileCountryCode", min_length=1, max_length=3)
+    mobile_number: Cp1252Str = Field(alias="mobileNumber", min_length=1, max_length=20)
+    mobile_country_code: Cp1252Str = Field(alias="mobileCountryCode", min_length=1, max_length=3)
 
 
 class NativeAddress(BaseModel):
@@ -28,9 +53,9 @@ class NativeAddress(BaseModel):
 
 class Address(BaseModel):
     """Address data class model"""
-    address1: str = Field(alias="address1", min_length=1, max_length=40)
-    city: str = Field(alias="city", min_length=1, max_length=32)
-    country: str = Field(alias="country", min_length=1, max_length=3)
+    address1: Cp1252Str = Field(alias="address1", min_length=1, max_length=40)
+    city: Cp1252Str = Field(alias="city", min_length=1, max_length=32)
+    country: Cp1252Str = Field(alias="country", min_length=1, max_length=3)
     phone: Phone = Field(alias="phone")
     mobile: Mobile = Field(alias="mobile")
     native_address: NativeAddress = Field(alias="nativeAddress")
@@ -46,33 +71,33 @@ class AlternateAddress(Address):
 
 class CandidateName(BaseModel):
     """CandidateName data class model"""
-    first_name: str = Field(alias="firstName", min_length=1, max_length=30)
-    last_name: str = Field(alias="lastName", min_length=1, max_length=50)
+    first_name: Cp1252Str = Field(alias="firstName", min_length=1, max_length=30)
+    last_name: Cp1252Str = Field(alias="lastName", min_length=1, max_length=50)
 
 
 class WebAccountInfo(BaseModel):
     """WebAccountInfo data class model"""
-    email: str = Field(alias="email", min_length=1, max_length=255)
+    email: Cp1252Str = Field(alias="email", min_length=1, max_length=255)
 
 
 class CddRequest(BaseModel):
     """CddRequest data class model"""
-    client_candidate_id: str = Field(alias="@clientCandidateID", min_length=1, max_length=50)
-    client_id: str = Field(alias="@clientID", min_length=1)
+    client_candidate_id: Cp1252Str = Field(alias="@clientCandidateID", min_length=1, max_length=50)
+    client_id: Cp1252Str = Field(alias="@clientID", min_length=1)
     candidate_name: CandidateName = Field(alias="candidateName")
-    last_update: str = Field(alias="lastUpdate", min_length=1)
+    last_update: Cp1252Str = Field(alias="lastUpdate", min_length=1)
     primary_address: PrimaryAddress = Field(alias="primaryAddress")
     web_account_info: WebAccountInfo = Field(alias="webAccountInfo")
 
 
 class EadRequest(BaseModel):
     """EadRequest data class model"""
-    client_id: str = Field(alias="@clientID", min_length=1)
-    authorization_transaction_type: str = Field(alias="@authorizationTransactionType", min_length=1)
-    client_authorization_id: str = Field(alias="@clientAuthorizationID", min_length=1, max_length=25)
-    client_candidate_id: str = Field(alias="clientCandidateID", min_length=1, max_length=50)
+    client_id: Cp1252Str = Field(alias="@clientID", min_length=1)
+    authorization_transaction_type: Cp1252Str = Field(alias="@authorizationTransactionType", min_length=1)
+    client_authorization_id: Cp1252Str = Field(alias="@clientAuthorizationID", min_length=1, max_length=25)
+    client_candidate_id: Cp1252Str = Field(alias="clientCandidateID", min_length=1, max_length=50)
     exam_authorization_count: int = Field(alias="examAuthorizationCount")
-    exam_series_code: str = Field(alias="examSeriesCode", min_length=1, max_length=20)
-    elegibility_appt_date_first: str = Field(alias="eligibilityApptDateFirst", min_length=1)
-    elegibility_appt_date_last: str = Field(alias="eligibilityApptDateLast", min_length=1)
-    last_update: str = Field(alias="lastUpdate", min_length=1)
+    exam_series_code: Cp1252Str = Field(alias="examSeriesCode", min_length=1, max_length=20)
+    elegibility_appt_date_first: Cp1252Str = Field(alias="eligibilityApptDateFirst", min_length=1)
+    elegibility_appt_date_last: Cp1252Str = Field(alias="eligibilityApptDateLast", min_length=1)
+    last_update: Cp1252Str = Field(alias="lastUpdate", min_length=1)

eox_nelp/pearson_vue/tests/test_pipeline.py

@@ -845,6 +845,19 @@ def setUp(self):
         {"primaryAddress": {"phone": {"phoneCountryCode": ""}}},
         {"primaryAddress": {"phone": {"phoneNumber": ""}}},
         {"webAccountInfo": {"email": ""}},
+        {"@clientCandidateID": "فلان"},
+        {"@clientID": "فلان"},
+        {"candidateName": {"firstName": "فلان"}},
+        {"candidateName": {"lastName": "فلان"}},
+        {"lastUpdate": "فلان"},
+        {"primaryAddress": {"address1": "فلان"}},
+        {"primaryAddress": {"city": "فلان"}},
+        {"primaryAddress": {"country": "فلان"}},
+        {"primaryAddress": {"mobile": {"mobileCountryCode": "فلان"}}},
+        {"primaryAddress": {"mobile": {"mobileNumber": "فلان"}}},
+        {"primaryAddress": {"phone": {"phoneCountryCode": "فلان"}}},
+        {"primaryAddress": {"phone": {"phoneNumber": "فلان"}}},
+        {"webAccountInfo": {"email": "فلان"}},
     )
     def test_wrong_cdd_request(self, wrong_update):
         """Test validator with a wrong cdd_request updating with empty string
@@ -898,6 +911,15 @@ def setUp(self):
         {"examAuthorizationCount": ""},
         {"examSeriesCode": ""},
         {"lastUpdate": ""},
+        {"@authorizationTransactionType": "فلان"},
+        {"@clientAuthorizationID": "فلان"},
+        {"@clientID": "فلان"},
+        {"clientCandidateID": "فلان"},
+        {"eligibilityApptDateFirst": "فلان"},
+        {"eligibilityApptDateLast": "فلان"},
+        {"examAuthorizationCount": "فلان"},
+        {"examSeriesCode": "فلان"},
+        {"lastUpdate": "فلان"},
     )
     def test_wrong_ead_request(self, wrong_update):
         """Test validator with a wrong ead_request updating with empty string

eox_nelp/pearson_vue/tests/test_utils.py

@@ -13,7 +13,7 @@
 
 from eox_nelp.edxapp_wrapper.student import AnonymousUserId, CourseEnrollment
 from eox_nelp.pearson_vue.constants import PAYLOAD_CDD, PAYLOAD_EAD
-from eox_nelp.pearson_vue.utils import generate_client_authorization_id, update_xml_with_dict
+from eox_nelp.pearson_vue.utils import generate_client_authorization_id, is_cp1252, update_xml_with_dict
 
 User = get_user_model()
 
@@ -418,3 +418,36 @@ def test_generate_client_authorization_id(self):
         result = generate_client_authorization_id(self.user.id, self.course_id)
 
         self.assertEqual(expected_result, result)
+
+
+class TestIsCp1252(TestCase):
+    """Class to test is_cp1252 function"""
+    def test_english_true(self):
+        """Tests if a string with English characters is CP1252 encoded."""
+        text = "This is a test string"
+
+        self.assertTrue(is_cp1252(text))
+
+    def test_arabic_false(self):
+        """Tests if a string with Arabic characters is not CP1252 encoded."""
+        text = "هذا هو اختبار باللغة العربية"  # Arabic text
+
+        self.assertFalse(is_cp1252(text))
+
+    def test_mixed_chars(self):
+        """Tests if a string with mixed characters (English and Arabic) is not CP1252 encoded."""
+        text = "This is a test with عربية characters"
+
+        self.assertFalse(is_cp1252(text))
+
+    def test_empty_string(self):
+        """Tests if an empty string is considered CP1252 encoded."""
+        text = ""
+
+        self.assertTrue(is_cp1252(text))
+
+    def test_special_chars_false(self):
+        """Tests if a string with special characters is not CP1252 encoded."""
+        text = "This string has ©®€ symbols"
+
+        self.assertFalse(is_cp1252(text))

eox_nelp/pearson_vue/utils.py

@@ -2,6 +2,8 @@
 This includes xml helpers:
     - update_xml_with_dict
 """
+import re
+
 import xmltodict
 from pydantic.v1.utils import deep_update
 
@@ -43,3 +45,10 @@ def generate_client_authorization_id(user_id: int, course_id: str) -> str:
     anonymous_user_id_instance = AnonymousUserId.objects.get(anonymous_user_id=anonymous_user_id)
 
     return f"{course_enrollment.id}-{anonymous_user_id_instance.id}"
+
+
+def is_cp1252(text):
+    """Checks if the given text matchs the format CP1252"""
+    cp1252_regex = r'^[\x00-\x7F\x80-\x9F\xA0-\xFF]*$'
+
+    return re.match(cp1252_regex, text) is not None