Refactor validation (incorrect file names)

.gitignore

@@ -273,3 +273,5 @@ devenv.lock
 
 # pre-commit
 .pre-commit-config.yaml
+
+.vscode

requirements.txt

@@ -17,3 +17,4 @@ urllib3==2.4.0
 pydantic==2.11.3
 langchain==0.3.23
 langchain-groq==0.3.2
+email-validator==2.2.0

tests/test_cross_validate.py

@@ -1,4 +1,9 @@
-from validation import FromAccount
+from datetime import date
+from validation.cross_validate import ExtractedData
+from validation.from_account import FromAccount
+from validation.from_description import FromDescription
+from validation.from_passport import FromPassport
+from validation.from_profile import FromProfile
 
 
 account_data = FromAccount(
@@ -15,5 +20,88 @@ account_data = FromAccount(
     country="Netherlands",
     name="Astrid Janneke Willems",
     phone_number="+31 06 34579996",
-    email="astrid.willems@upcmail.nl"
-)
+    email="astrid.willems@upcmail.nl",
+)
+
+description_data = FromDescription(
+    full_name="Astrid Janneke Willems",
+    age=28,
+    nationality="Netherlands",
+    marital_status="single",
+    has_children=False,
+    secondary_education_school="Pieter Nieuwland College Utrecht",
+    secondary_education_year=2016,
+    university_name="Webster University Leiden",
+    university_graduation_year=2020,
+    occupation_title="Art Dealer",
+    employer="Rijksmuseum Amsterdam",
+    start_year=2021,
+    annual_salary_eur=40000,
+    total_savings_eur=20000,
+    has_properties=False,
+    inheritance_amount_eur=1590000,
+    inheritance_year=2020,
+    inheritance_source="grandmother (Oil and Gas Executive)",
+)
+
+passport_data = FromPassport(
+    country="NLD",
+    passport_number="HW8642009",
+    surname="WILLEMS",
+    given_names="ASTRID JANNEKE",
+    birth_date=date(1997, 1, 19),
+    citizenship="Austrian/ÖSTERREICH",
+    sex="F",
+    issue_date=date(2016, 6, 4),
+    expiry_date=date(2026, 6, 3),
+    signature_present=True,
+    machine_readable_zone="P<NLDWILLEMS<<ASTRID<JANNEKE<<<<<<<<<<<<<<<<<<<HW8642009NLD970119",
+)
+
+profile_data = FromProfile(
+    first_name="Astrid Janneke",
+    last_name="Willems",
+    date_of_birth=date(1997, 1, 19),
+    nationality="Dutch",
+    country_of_domicile="Netherlands",
+    gender="Female",
+    passport_number="HW8642009",
+    id_type="passport",
+    id_issue_date=date(2016, 6, 4),
+    id_expiry_date=date(2026, 6, 3),
+    phone="+31 06 34579996",
+    email="astrid.willems@upcmail.nl",
+    address="Lijnbaan 18, 7523 05 Assen",
+    politically_exposed_person=False,
+    marital_status="Single",
+    highest_education="Tertiary",
+    education_history="Webster University Leiden (2020)",
+    employment_status="Employee",
+    employment_since=2021,
+    employer="Rijksmuseum Amsterdam",
+    position="Art Dealer",
+    annual_salary_eur=40000.0,
+    total_wealth_range="1.5m-5m",
+    origin_of_wealth=["Employment", "Inheritance"],
+    inheritance_details="Grandmother, 2020, Oil and Gas Executive",
+    business_assets_eur=20000.0,
+    estimated_annual_income="<250k",
+    income_country="Netherlands",
+    commercial_account=False,
+    investment_risk_profile="High",
+    mandate_type="Advisory",
+    investment_experience="Experienced",
+    investment_horizon="Medium",
+    preferred_markets=["Denmark", "Netherlands"],
+    total_aum=1610000.0,
+    aum_to_transfer=1320200.0,
+)
+
+
+def test_xval_name_account_description() -> None:
+    data = ExtractedData(
+        account=account_data,
+        description=description_data,
+        passport=passport_data,
+        profile=profile_data,
+    )

validation/cross_validate.py

@@ -2,7 +2,11 @@ from enum import StrEnum
 from typing import Any, Callable, Optional
 from pydantic import BaseModel
 
-from validation import FromAccount, FromDescription, FromPassport, FromProfile
+from validation.from_account import FromAccount
+from validation.from_description import FromDescription
+from validation.from_passport import FromPassport
+from validation.from_profile import FromProfile
+
 
 
 class ExtractedData(BaseModel):

validation/from_account.py

@@ -9,9 +9,9 @@ class FromAccount(BaseModel):
     model_config = ConfigDict(validate_assignment=True, str_strip_whitespace=True)
 
     # From account.pdf
-    account_name: str = Field(..., min_length=1)
-    account_holder_name: str = Field(..., min_length=1)
-    account_holder_surname: str = Field(..., min_length=1)
+    account_name: str = Field(min_length=1)
+    account_holder_name: str = Field(min_length=1)
+    account_holder_surname: str = Field(min_length=1)
 
     @model_validator(mode='after')
     def check_account_name_is_name_surname(self) -> Self:
@@ -20,17 +20,17 @@ class FromAccount(BaseModel):
             raise ValueError(f'Account name is not name + surname: {self.account_name} != {combined}')
         return self
 
-    passport_number: str = Field(..., min_length=5)
+    passport_number: str = Field(min_length=5)
 
     reference_currency: Literal["CHF", "EUR", "USD", "Other"]
     other_currency: Optional[str] = None
 
-    building_number: str = Field(..., min_length=1)
-    street_name: str = Field(..., min_length=1)
-    postal_code: str = Field(..., min_length=1)
-    city: str = Field(..., min_length=1)
-    country: str = Field(..., min_length=1)
+    building_number: str = Field(min_length=1)
+    street_name: str = Field(min_length=1)
+    postal_code: str = Field(min_length=1)
+    city: str = Field(min_length=1)
+    country: str = Field(min_length=1)
 
-    name: str = Field(..., min_length=1)
-    phone_number: str = Field(..., min_length=6)
+    name: str = Field(min_length=1)
+    phone_number: str = Field(min_length=6)
     email: EmailStr

validation/from_passport.py

@@ -10,7 +10,7 @@ class FromPassport(BaseModel):
     model_config = ConfigDict(validate_assignment=True, str_strip_whitespace=True)
     
     country: str = Field(..., min_length=3, max_length=3)  # ISO 3166-1 alpha-3
-    passport_number: str = Field(..., min_length=9, max_length=9, regex=r"^[A-Z0-9]{9}$")
+    passport_number: str = Field(..., min_length=9, max_length=9, pattern=r"^[A-Z0-9]{9}$")
 
     surname: str = Field(..., min_length=1)
     given_names: str = Field(..., min_length=1)

validation/from_profile.py

@@ -17,7 +17,7 @@ class FromProfile(BaseModel):
     gender: Literal["Female", "Male"]
 
     # ID information
-    passport_number: str = Field(..., min_length=9, max_length=9, regex=r"^[A-Z0-9]{9}$")
+    passport_number: str = Field(..., min_length=9, max_length=9, pattern=r"^[A-Z0-9]{9}$")
     id_type: Literal["passport"]
     id_issue_date: date
     id_expiry_date: date