Merge pull request #170 from my-game-plan/feature/wyscout-v3-support

Add support for Wyscout V3
PySport · Mar 14, 2023 · d945f1e · d945f1e
2 parents 5c6c568 + 2df5b67
commit d945f1e
Show file tree

Hide file tree

Showing 7 changed files with 51,895 additions and 48,003 deletions.
diff --git a/kloppy/_providers/wyscout.py b/kloppy/_providers/wyscout.py
@@ -1,8 +1,10 @@
-from typing import Union
+import json
+from typing import Union, Type
 
 from kloppy.config import get_config
 from kloppy.infra.serializers.event.wyscout import (
-    WyscoutDeserializer,
+    WyscoutDeserializerV3,
+    WyscoutDeserializerV2,
     WyscoutInputs,
 )
 from kloppy.domain import EventDataset, Optional, List, EventFactory
@@ -14,6 +16,7 @@ def load(
     event_types: Optional[List[str]] = None,
     coordinates: Optional[str] = None,
     event_factory: Optional[EventFactory] = None,
+    data_version: Optional[str] = None,
 ) -> EventDataset:
     """
     Load Wyscout event data into a [`EventDataset`][kloppy.domain.models.event.EventDataset]
@@ -23,12 +26,21 @@ def load(
         event_types:
         coordinates:
         event_factory:
+        data_version:
     """
-    deserializer = WyscoutDeserializer(
+    if data_version == "V2":
+        deserializer_class = WyscoutDeserializerV2
+    elif data_version == "V3":
+        deserializer_class = WyscoutDeserializerV3
+    else:
+        deserializer_class = identify_deserializer(event_data)
+
+    deserializer = deserializer_class(
         event_types=event_types,
         coordinate_system=coordinates,
         event_factory=event_factory or get_config("event_factory"),
     )
+
     with open_as_file(event_data) as event_data_fp:
         return deserializer.deserialize(
             inputs=WyscoutInputs(event_data=event_data_fp),
@@ -47,3 +59,26 @@ def load_open_data(
         coordinates=coordinates,
         event_factory=event_factory,
     )
+
+
+def identify_deserializer(
+    event_data: FileLike,
+) -> Union[Type[WyscoutDeserializerV3], Type[WyscoutDeserializerV2]]:
+    with open_as_file(event_data) as event_data_fp:
+        events_with_meta = json.load(event_data_fp)
+
+    events = events_with_meta["events"]
+    first_event = events[0]
+
+    deserializer = None
+    if "eventName" in first_event:
+        deserializer = WyscoutDeserializerV2
+    elif "primary" in first_event.get("type", {}):
+        deserializer = WyscoutDeserializerV3
+
+    if deserializer is None:
+        raise ValueError(
+            "Wyscout data version could not be recognized, please specify"
+        )
+
+    return deserializer
diff --git a/kloppy/infra/serializers/event/wyscout/__init__.py b/kloppy/infra/serializers/event/wyscout/__init__.py
@@ -1 +1,2 @@
-from .deserializer import WyscoutDeserializer, WyscoutInputs
+from .deserializer_v3 import WyscoutDeserializerV3, WyscoutInputs
+from .deserializer_v2 import WyscoutDeserializerV2
diff --git a/...serializers/event/wyscout/deserializer.py → ...ializers/event/wyscout/deserializer_v2.py b/...serializers/event/wyscout/deserializer.py → ...ializers/event/wyscout/deserializer_v2.py
@@ -264,7 +264,7 @@ class WyscoutInputs(NamedTuple):
     event_data: IO[bytes]
 
 
-class WyscoutDeserializer(EventDataDeserializer[WyscoutInputs]):
+class WyscoutDeserializerV2(EventDataDeserializer[WyscoutInputs]):
     @property
     def provider(self) -> Provider:
         return Provider.WYSCOUT