Add tests

socceraction/data/opta/parsers/ma12_json.py

@@ -158,7 +158,7 @@ def extract_players(self) -> Dict[Tuple[str, str], Dict[str, Any]]:  # noqa: C90
                 "player_id": player.player_id,
                 "player_name": player.player_name,
                 "is_starter": is_starter,
-                "minutes_played": player.minutes_played,
+                "minutes_played": int(player.minutes_played),
                 "jersey_number": player.jersey_number,
                 # Fields required by the opta schema
                 "starting_position": player.starting_position,

socceraction/data/opta/parsers/ma5_json.py

@@ -32,16 +32,6 @@ def _get_live_data(self) -> Dict[str, Any]:
             return self.root["liveData"]
         raise MissingDataError
 
-    """
-    game_id: Series[Object] = pa.Field()
-    season_id: Series[Object] = pa.Field()
-    competition_id: Series[Object] = pa.Field()
-    game_day: Series[pd.Int64Dtype] = pa.Field(nullable=True)
-    game_date: Series[DateTime] = pa.Field()
-    home_team_id: Series[Object] = pa.Field()
-    away_team_id: Series[Object] = pa.Field()
-    """
-
     def extract_games(self) -> Dict[str, Dict[str, Any]]:
         """Return a dictionary with all available games.
 
@@ -81,7 +71,7 @@ def extract_games(self) -> Dict[str, Dict[str, Any]]:
                 game_date=datetime.strptime(game_datetime, "%Y-%m-%dT%H:%M:%S"),
                 home_team_id=home_team_id,
                 away_team_id=away_team_id,
-                home_possession=home_possession,
-                away_possession=away_possession,
+                home_possession=float(home_possession),
+                away_possession=float(away_possession),
             )
         }

socceraction/data/opta/schema.py

@@ -48,11 +48,9 @@ class OptaPlayerSchema(PlayerSchema):
 
     starting_position: Series[str]
     """The starting position of the player."""
-
     position_side: Optional[Series[str]]
     """The side of the pitch where the player started the game"""
-
-    xG_non_penalty: Optional[Series[int]]
+    xG_non_penalty: Optional[Series[float]]
     """The xG of the player without taking penalties into account"""
 
     # minutes_played: Optional[Series[int]]

tests/data/opta/parsers/test_ma12_json.py

@@ -0,0 +1,46 @@
+import os
+
+import pandas as pd
+from pytest import fixture
+
+from socceraction.data.opta import OptaPlayerSchema
+from socceraction.data.opta.parsers import MA12JSONParser
+
+
+@fixture()
+def ma12json_parser() -> MA12JSONParser:
+    path = os.path.join(
+        os.path.dirname(__file__),
+        os.pardir,
+        os.pardir,
+        os.pardir,
+        "datasets",
+        "opta",
+        "ma12_drnl1etmyzjz088y0bb4kh4pg.json",
+    )
+    return MA12JSONParser(str(path))
+
+
+def test_extract_players(ma12json_parser: MA12JSONParser) -> None:
+    players = ma12json_parser.extract_players()
+    assert len(players) == 40
+    assert players[("drnl1etmyzjz088y0bb4kh4pg", "4iijb6llnz28unsz4rirr3umt")] == {
+        "game_id": "drnl1etmyzjz088y0bb4kh4pg",
+        "player_id": "4iijb6llnz28unsz4rirr3umt",
+        "player_name": "David Raya",
+        "team_id": "7yx5dqhhphyvfisohikodajhv",
+        "jersey_number": 1,
+        "minutes_played": 90,
+        "starting_position": "Goalkeeper",
+        "is_starter": True,
+        "position_side": "Centre",
+        "xG_non_penalty": 0.0,
+    }
+    # red card
+    assert (
+        players[("drnl1etmyzjz088y0bb4kh4pg", "ds1wjqejslhsbcvzaufxubey2")]["minutes_played"] == 10
+    )
+    assert (
+        players[("drnl1etmyzjz088y0bb4kh4pg", "2o5etqath4k9qaliho4op0hd5")]["minutes_played"] == 0
+    )
+    OptaPlayerSchema.validate(pd.DataFrame.from_dict(players, orient="index"))

tests/data/opta/parsers/test_ma5_json.py

@@ -0,0 +1,39 @@
+import os
+from datetime import datetime
+
+import pandas as pd
+from pytest import fixture
+
+from socceraction.data.opta import OptaGameSchema
+from socceraction.data.opta.parsers import MA5JSONParser
+
+
+@fixture()
+def ma5json_parser() -> MA5JSONParser:
+    path = os.path.join(
+        os.path.dirname(__file__),
+        os.pardir,
+        os.pardir,
+        os.pardir,
+        "datasets",
+        "opta",
+        "ma5_drnl1etmyzjz088y0bb4kh4pg.json",
+    )
+    return MA5JSONParser(str(path))
+
+
+def test_extract_games(ma5json_parser: MA5JSONParser) -> None:
+    games = ma5json_parser.extract_games()
+    assert len(games) == 1
+    assert games["drnl1etmyzjz088y0bb4kh4pg"] == {
+        "game_id": "drnl1etmyzjz088y0bb4kh4pg",
+        "season_id": "8l3o9v8n8tva0bb2cds2dhatw",
+        "competition_id": "2kwbbcootiqqgmrzs6o5inle5",
+        "game_day": 1,
+        "game_date": datetime(2021, 8, 13, 19, 0),
+        "home_team_id": "7yx5dqhhphyvfisohikodajhv",
+        "away_team_id": "4dsgumo7d4zupm2ugsvm4zm4d",
+        "away_possession": 64.7,
+        "home_possession": 35.3,
+    }
+    OptaGameSchema.validate(pd.DataFrame.from_dict(games, orient="index"))