Merge pull request #50 from ai-cfia/41-correct-phone-number-and-url-f…

…ormat

Issue #41: correct phone number and url format in FertilizerInspection

.gitignore

@@ -171,5 +171,6 @@ cython_debug/
 
 # Testing artifacts
 end_to_end_pipeline_artifacts
-reports
-logs
+
+logs/
+reports/

pipeline/inspection.py

@@ -1,6 +1,9 @@
 import re
 from typing import List, Optional
-from pydantic import BaseModel, Field, field_validator, model_validator
+
+import phonenumbers
+from pydantic import BaseModel, ConfigDict, Field, field_validator, model_validator
+
 
 class npkError(ValueError):
     pass
@@ -58,10 +61,10 @@ def replace_none_with_empty_list(cls, v):
         if v is None:
             v = []
         return v
-    
+
     @model_validator(mode="after")
     def set_is_minimal(self):
-        pattern = r'\bminim\w*\b'
+        pattern = r"\bminim\w*\b"
         if self.title:
             self.is_minimal = re.search(pattern, self.title, re.IGNORECASE) is not None
         return self
@@ -87,11 +90,15 @@ class FertilizerInspection(BaseModel):
     company_name: Optional[str] = None
     company_address: Optional[str] = None
     company_website: Optional[str] = None
-    company_phone_number: Optional[str] = None
+    company_phone_number: Optional[str] = Field(
+        None, description="The distributor's primary phone number. Return only one."
+    )
     manufacturer_name: Optional[str] = None
     manufacturer_address: Optional[str] = None
     manufacturer_website: Optional[str] = None
-    manufacturer_phone_number: Optional[str] = None
+    manufacturer_phone_number: Optional[str] = Field(
+        None, description="The manufacturer's primary phone number. Return only one."
+    )
     fertiliser_name: Optional[str] = None
     registration_number: Optional[str] = None
     lot_number: Optional[str] = None
@@ -107,6 +114,7 @@ class FertilizerInspection(BaseModel):
     instructions_fr: List[str] = []
     ingredients_en: List[NutrientValue] = []
     ingredients_fr: List[NutrientValue] = []
+    model_config = ConfigDict(populate_by_name=True)
 
     @field_validator("npk", mode="before")
     def validate_npk(cls, v):
@@ -130,14 +138,28 @@ def replace_none_with_empty_list(cls, v):
         if v is None:
             v = []
         return v
-    
+
     @field_validator("registration_number", mode="before")
     def check_registration_number_format(cls, v):
         if v is not None:
-            pattern = r'^\d{7}[A-Z]$'
+            pattern = r"^\d{7}[A-Z]$"
             if re.match(pattern, v):
                 return v
         return None
 
-    class Config:
-        populate_by_name = True
+    @field_validator("company_phone_number", "manufacturer_phone_number", mode="before")
+    def check_phone_number_format(cls, v):
+        if v is None:
+            return
+
+        try:
+            phone_number = phonenumbers.parse(v, "CA")
+            if not phonenumbers.is_valid_number(phone_number):
+                return
+            phone_number = phonenumbers.format_number(
+                phone_number, phonenumbers.PhoneNumberFormat.E164
+            )
+            return phone_number
+
+        except phonenumbers.phonenumberutil.NumberParseException:
+            return

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "fertiscan_pipeline"
-version = "0.0.4"
+version = "0.0.5"
 description = "A pipeline for the FertiScan project"
 authors = [
     { name = "Albert Bryan Ndjeutcha", email = "[email protected]" }

requirements.txt

@@ -7,5 +7,7 @@ openinference-instrumentation-dspy
 python-dotenv
 reportlab
 setuptools
+phonenumbers
 # Test dependencies
 Levenshtein
+pytest

tests/test_inspection.py

@@ -2,10 +2,10 @@
 
 from pipeline.inspection import (
     FertilizerInspection,
+    GuaranteedAnalysis,
     NutrientValue,
     Specification,
     Value,
-    GuaranteedAnalysis,
 )
 
 
@@ -206,26 +206,27 @@ def test_invalid_npk(self):
                     inspection.npk, f"Expected None for npk with input {npk}"
                 )
 
+
 class TestGuaranteedAnalysis(unittest.TestCase):
-
     def setUp(self):
         self.nutrient_1 = NutrientValue(nutrient="Nitrogen", value="2", unit="mg/L")
-        self.nutrient_2 = NutrientValue(nutrient="Organic matter", value="15", unit="mg/L")
+        self.nutrient_2 = NutrientValue(
+            nutrient="Organic matter", value="15", unit="mg/L"
+        )
 
     def test_set_is_minimal(self):
         guaranteed_analysis = GuaranteedAnalysis(
             title="Guaranteed minimum analysis",
-            nutrients=[self.nutrient_1, self.nutrient_2]
+            nutrients=[self.nutrient_1, self.nutrient_2],
         )
         self.assertTrue(guaranteed_analysis.is_minimal)
 
     def test_set_is_not_minimal(self):
         guaranteed_analysis = GuaranteedAnalysis(
-            title="Guaranteed analysis",
-            nutrients=[self.nutrient_1, self.nutrient_2]
+            title="Guaranteed analysis", nutrients=[self.nutrient_1, self.nutrient_2]
         )
         self.assertFalse(guaranteed_analysis.is_minimal)
-    
+
     def test_is_minimal_in_none(self):
         guaranteed_analysis = GuaranteedAnalysis(
             nutrients=[self.nutrient_1, self.nutrient_2]
@@ -268,7 +269,8 @@ def test_replace_none_with_empty_list(self):
         self.assertEqual(inspection.ingredients_en, [])
         self.assertEqual(inspection.ingredients_fr, [])
         self.assertEqual(inspection.weight, [])
-
+
+
 class TestFertilizerInspectionRegistrationNumber(unittest.TestCase):
     def test_registration_number_with_less_digits(self):
         instance = FertilizerInspection(registration_number="1234")
@@ -299,5 +301,44 @@ def test_registration_number_mixed_format(self):
         self.assertIsNone(instance.registration_number)
 
 
+class TestFertilizerInspectionPhoneNumberFormat(unittest.TestCase):
+    def test_valid_phone_number_with_country_code(self):
+        instance = FertilizerInspection(company_phone_number="+1 800 640 9605")
+        self.assertEqual(instance.company_phone_number, "+18006409605")
+
+    def test_valid_phone_number_without_country_code(self):
+        instance = FertilizerInspection(company_phone_number="800 640 9605")
+        self.assertEqual(instance.company_phone_number, "+18006409605")
+
+    def test_phone_number_with_parentheses(self):
+        instance = FertilizerInspection(company_phone_number="(757) 321-4567")
+        self.assertEqual(instance.company_phone_number, "+17573214567")
+
+    def test_phone_number_with_extra_characters(self):
+        instance = FertilizerInspection(company_phone_number="+1 800 321-9605 FAX")
+        self.assertIsNone(instance.company_phone_number)
+
+    def test_phone_number_with_multiple_numbers(self):
+        instance = FertilizerInspection(
+            company_phone_number="(757) 123-4567 (800) 456-7890, 1234567890"
+        )
+        self.assertIsNone(instance.company_phone_number)
+
+    def test_phone_number_from_other_country(self):
+        instance = FertilizerInspection(manufacturer_phone_number="+44 20 7946 0958")
+        self.assertEqual(instance.manufacturer_phone_number, "+442079460958")
+
+    def test_invalid_phone_number(self):
+        instance = FertilizerInspection(company_phone_number="invalid phone")
+        self.assertIsNone(instance.company_phone_number)
+
+    def test_phone_number_with_invalid_format(self):
+        instance = FertilizerInspection(company_phone_number="12345")
+        self.assertIsNone(instance.company_phone_number)
+
+
+if __name__ == "__main__":
+    unittest.main()
+
 if __name__ == "__main__":
     unittest.main()