2025

tests/test_utils.py

@@ -1,7 +1,9 @@
 import unittest
 from pathlib import Path
 # from convert_to_sportspress 
-from convert_to_sportspress.utils import validate_csv_header, normalize_header_key, read_csv, parse_score, is_visitor_home_order_reversed, process_data, aggregate_teams
+from src.utils.common import validate_csv_header, normalize_header_key, read_and_normalize_csv, parse_score, is_visitor_home_order_reversed, import_gamebygame, aggregate_teams
+from src.utils.normalize import normalize_value, normalize_header_key, load_config
+import toml
 
 class TestConvertToSportsPress(unittest.TestCase):
     def setUp(self):
@@ -23,11 +25,11 @@ class TestConvertToSportsPress(unittest.TestCase):
     def test_read_csv(self):
         # Assuming that the CSV file has a valid header
         with self.subTest("Read CSV data"):
-            data = read_csv(self.test_csv_path_2009)
+            data = read_and_normalize_csv(self.test_csv_path_2009)
             self.assertIsInstance(data, list)
             self.assertTrue(all(isinstance(row, dict) for row in data))
         with self.subTest("Normalized keys"):
-            normalized_data = read_csv(self.test_csv_path_2009)
+            normalized_data = read_and_normalize_csv(self.test_csv_path_2009)
             self.assertTrue(all("visitor" in row.keys() and "results" in row.keys() for row in normalized_data))
 
     def test_parse_score_visitor_first(self):
@@ -115,8 +117,8 @@ class TestConvertToSportsPress(unittest.TestCase):
 
     def test_process_data(self):
         # Assuming that the CSV file has a valid header and read_csv is good
-        data = read_csv(self.test_csv_path_2009)
-        processed_data = process_data(data)
+        data = read_and_normalize_csv(self.test_csv_path_2009)
+        processed_data = import_gamebygame(data)
         aggregate_team_data = aggregate_teams(processed_data)
         expected_result = [
             {"team": "Marlins", "gp": 28, "win": 23, "loss": 5, "tie": 0, "pts": 46, "runs_for": 249, "runs_against": 117},
@@ -139,7 +141,85 @@ class TestConvertToSportsPress(unittest.TestCase):
             with self.subTest(f'Results of "{team}"'):
                 self.assertDictContainsSubset(aggregate_team_data_dict, expected_dict)
 
-            
+class TestNormalization(unittest.TestCase):
+
+    def test_normalize_key(self):
+        header_key_normalization = {
+            "date": {"potential_keys": ["Date", "EventDate"]},
+            "time": {"potential_keys": ["Time", "EventTime"]},
+            "visitor": {"potential_keys": ["Away"]},
+            "field": 
+                            {
+                                "potential_keys": ["Field", "Location", "Venue"],
+                                "values": [{"original": ["Winnemac"], "normalized": "Winnemac Park"}],
+                            }
+                        
+             }
+        # Test cases for normalize_key function
+        self.assertEqual(normalize_header_key("Date", header_key_normalization), "date")
+        self.assertEqual(normalize_header_key("Time", header_key_normalization), "time")
+        self.assertEqual(normalize_header_key("Venue", header_key_normalization), "field")
+        self.assertEqual(normalize_header_key("Away", header_key_normalization), "visitor")
+
+    def test_load_config_file(self):
+        expected = {
+            "win": {"potential_keys": ["w", "wins"]},
+            "loss": {"potential_keys": ["l", "losses"]},
+            "tie": {"potential_keys": ["t", "ties"]},
+            "points": {"potential_keys": ["pts.", "pts", "pt"]},
+            "runs_for": {"potential_keys": ["rf", "rs"]},
+            "runs_against": {"potential_keys": ["ra"]},
+            "division": {"potential_keys": ["div"]},
+            "date": {"potential_keys": ["Date", "EventDate"]},
+            "time": {"potential_keys": ["Time", "EventTime"]},
+            "visitor": {"potential_keys": ["Away"]},
+            "field": {
+                "potential_keys": ["Field", "Location", "Venue"],
+                "values": [{"original": ["Winnemac"], "normalized": "Winnemac Park"}],
+            },
+            "results": {"potential_keys": ["Final Score", "Score", "Result", "Outcome"]},
+            "team": {
+                "values": [
+                    {
+                        "original": ["Hounds", "Chicago Hounds", "Winnemac Hounds", "Hound"],
+                        "normalized": "Hounds",
+                    },
+                    {"original": ["Chicago Red Sox"], "normalized": "Red Sox"},
+                ]
+            },
+        }
+
+
+        config_path = "normalization.toml"
+        config = load_config(config_path)
+
+        self.assertEqual(config, expected)
+
+
+    def test_normalize_value(self):
+        # Test cases for normalize_value function
+        team_normalization = {
+            "team": {
+                "values": [
+                    {
+                        "original": ["Hounds", "Chicago Hounds", "Winnemac Hounds", "Hound"],
+                        "normalized": "Hounds",
+                    },
+                    {"original": ["Chicago Red Sox"], "normalized": "Red Sox"},
+                ]
+            }
+        }
+
+        # Test case with normalization
+        self.assertEqual(normalize_value("Chicago Hounds", 'team', team_normalization), "Hounds")
+
+        # Test case without title case normalization
+        # self.assertEqual(normalize_value("red sox", team_normalization, 'team'), "Red Sox")
+
+        # Add more test cases for other values
+
+if __name__ == '__main__':
+    unittest.main()       
 
 if __name__ == "__main__":
     unittest.main()