Creating a simple test generator for our tests (#642)

* Generate patient data

* Add data generation

* Added folder prefix

tests/linkage/test_linkage.py

@@ -47,6 +47,9 @@
 from random import seed
 from math import log
 from json.decoder import JSONDecodeError
+from tests.test_data_generator import (
+    generate_list_patients_contact,
+)
 
 
 def _set_up_postgres_client():
@@ -367,20 +370,7 @@ def test_block_parquet_data():
 
 
 def test_compile_match_lists():
-    data = [
-        ["11-7-2153", "John", "Shepard", "", "", "", "", "90909", 1],
-        ["11-7-2153", "Jhon", "Sheperd", "", "", "", "", "90909", 5],
-        ["11-7-2153", "Jon", "Shepherd", "", "", "", "", "90909", 11],
-        ["11-7-2153", "Johnathan", "Shepard", "", "", "", "", "90909", 12],
-        ["11-7-2153", "Nathan", "Shepard", "", "", "", "", "90909", 13],
-        ["01-10-1986", "Jane", "Smith", "", "", "", "", "12345", 14],
-        ["12-12-1992", "Daphne", "Walker", "", "", "", "", "23456", 18],
-        ["1-1-1980", "Alejandro", "Villanueve", "", "", "", "", "15935", 23],
-        ["1-1-1980", "Alejandro", "Villanueva", "", "", "", "", "15935", 24],
-        ["2-2-1990", "Philip", "", "", "", "", "", "64873", 27],
-        ["1-1-1980", "Alejandr", "Villanueve", "", "", "", "", "15935", 31],
-        ["1-1-1980", "Aelxdrano", "Villanueve", "", "", "", "", "15935", 32],
-    ]
+    data = generate_list_patients_contact()
     data = pd.DataFrame(
         data,
         columns=[
@@ -434,20 +424,7 @@ def test_feature_match_four_char():
 
 
 def test_map_matches_to_ids():
-    data = [
-        ["11-7-2153", "John", "Shepard", "", "", "", "", "90909", 1],
-        ["11-7-2153", "Jhon", "Sheperd", "", "", "", "", "90909", 5],
-        ["11-7-2153", "Jon", "Shepherd", "", "", "", "", "90909", 11],
-        ["11-7-2153", "Johnathan", "Shepard", "", "", "", "", "90909", 12],
-        ["11-7-2153", "Nathan", "Shepard", "", "", "", "", "90909", 13],
-        ["01-10-1986", "Jane", "Smith", "", "", "", "", "12345", 14],
-        ["12-12-1992", "Daphne", "Walker", "", "", "", "", "23456", 18],
-        ["1-1-1980", "Alejandro", "Villanueve", "", "", "", "", "15935", 23],
-        ["1-1-1980", "Alejandro", "Villanueva", "", "", "", "", "15935", 24],
-        ["2-2-1990", "Philip", "", "", "", "", "", "64873", 27],
-        ["1-1-1980", "Alejandr", "Villanueve", "", "", "", "", "15935", 31],
-        ["1-1-1980", "Aelxdrano", "Villanueve", "", "", "", "", "15935", 32],
-    ]
+    data = generate_list_patients_contact()
     data = pd.DataFrame(
         data,
         columns=[

tests/test_data_generator.py

@@ -0,0 +1,33 @@
+def generate_list_patients_contact():
+    patient_data = [
+        ["11-7-2153", "John", "Shepard", "", "", "", "", "90909", 1],
+        ["11-7-2153", "Jhon", "Sheperd", "", "", "", "", "90909", 5],
+        ["11-7-2153", "Jon", "Shepherd", "", "", "", "", "90909", 11],
+        ["11-7-2153", "Johnathan", "Shepard", "", "", "", "", "90909", 12],
+        ["11-7-2153", "Nathan", "Shepard", "", "", "", "", "90909", 13],
+        ["01-10-1986", "Jane", "Smith", "", "", "", "", "12345", 14],
+        ["12-12-1992", "Daphne", "Walker", "", "", "", "", "23456", 18],
+        ["1-1-1980", "Alejandro", "Villanueve", "", "", "", "", "15935", 23],
+        ["1-1-1980", "Alejandro", "Villanueva", "", "", "", "", "15935", 24],
+        ["2-2-1990", "Philip", "", "", "", "", "", "64873", 27],
+        ["1-1-1980", "Alejandr", "Villanueve", "", "", "", "", "15935", 31],
+        ["1-1-1980", "Aelxdrano", "Villanueve", "", "", "", "", "15935", 32],
+    ]
+    return patient_data
+
+
+def generate_eicr_results():
+    eicr_result = {
+        "root": "2.16.840.1.113883.9.9.9.9.9",
+        "extension": "db734647-fc99-424c-a864-7e3cda82e704",
+    }
+    return eicr_result
+
+
+def generate_ecr_msg_ids():
+    msg_ids = {
+        "internal_message": {"id1": ["ID1"], "id2": "SecondID"},
+        "external_id": ["extID"],
+        "last_id": "123",
+    }
+    return msg_ids

tests/validation/test_validation.py

@@ -1,6 +1,7 @@
 import pathlib
 
 import yaml
+from tests.test_data_generator import generate_eicr_results
 from phdi.validation.validation import validate_ecr
 
 
@@ -211,10 +212,7 @@ def test_custom_error_messages():
 
 
 def test_validate_good_with_rr_data():
-    eicr_result = {
-        "root": "2.16.840.1.113883.9.9.9.9.9",
-        "extension": "db734647-fc99-424c-a864-7e3cda82e704",
-    }
+    eicr_result = generate_eicr_results()
     rr_result = {
         "root": "4efa0e5c-c34c-429f-b5de-f1a13aef4a28",
         "extension": None,
@@ -238,10 +236,7 @@ def test_validate_good_with_rr_data():
 
 
 def test_validate_with_rr_data_missing_rr():
-    eicr_result = {
-        "root": "2.16.840.1.113883.9.9.9.9.9",
-        "extension": "db734647-fc99-424c-a864-7e3cda82e704",
-    }
+    eicr_result = generate_eicr_results()
 
     expected_response = {
         "message_valid": False,

tests/validation/test_validation_helpers.py

@@ -7,7 +7,7 @@
     _clear_all_errors_and_ids,
     ERROR_MESSAGES,
 )
-
+from tests.test_data_generator import generate_ecr_msg_ids
 
 test_include_errors = ["fatal", "errors", "warnings", "information"]
 
@@ -211,11 +211,7 @@ def test_append_error_message():
 def test_add_message_ids():
     _clear_all_errors_and_ids()
     # first let's test with a proper dict
-    msg_ids = {
-        "internal_message": {"id1": ["ID1"], "id2": "SecondID"},
-        "external_id": ["extID"],
-        "last_id": "123",
-    }
+    msg_ids = generate_ecr_msg_ids()
     expected_result = {
         "fatal": [],
         "errors": [],
@@ -243,11 +239,7 @@ def test_add_message_ids():
 def test_clear_all_errors_and_ids():
     # first let's load up all errors and msg ids
     #  and clear them out and make sure they are clear
-    msg_ids = {
-        "internal_message": {"id1": ["ID1"], "id2": "SecondID"},
-        "external_id": ["extID"],
-        "last_id": "123",
-    }
+    msg_ids = generate_ecr_msg_ids()
     fatal = ["foo"]
     errors = ["my error1", "my_error2"]
     warns = ["my warn1"]