add opta blocked pass as interception and for wyscout-v2 touch and du…

…el as interception.
PySport · Nov 14, 2023 · bdc6d33 · bdc6d33
1 parent 471da05
commit bdc6d33
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 17 deletions.
diff --git a/kloppy/infra/serializers/event/opta/deserializer.py b/kloppy/infra/serializers/event/opta/deserializer.py
@@ -1,4 +1,4 @@
-from typing import Tuple, Dict, List, NamedTuple, IO
+from typing import Tuple, Dict, List, NamedTuple, IO, Optional
 import logging
 from datetime import datetime
 import pytz
@@ -78,6 +78,7 @@
 EVENT_TYPE_RECOVERY = 49
 EVENT_TYPE_FORMATION_CHANGE = 40
 EVENT_TYPE_BALL_TOUCH = 61
+EVENT_TYPE_BLOCKED_PASS = 74
 
 EVENT_TYPE_SAVE = 10
 EVENT_TYPE_CLAIM = 11
@@ -391,7 +392,7 @@ def _parse_interception(
     qualifiers = _get_event_qualifiers(raw_qualifiers)
     result = InterceptionResult.SUCCESS
 
-    if next_event:
+    if next_event is not None:
         next_event_type_id = int(next_event.attrib["type_id"])
         if next_event_type_id in BALL_OUT_EVENTS:
             result = InterceptionResult.OUT
@@ -642,7 +643,11 @@ def deserialize(self, inputs: OptaInputs) -> EventDataset:
             ]
             possession_team = None
             events = []
-            events_list = list(game_elm.iterchildren("Event"))
+            events_list = [
+                event
+                for event in list(game_elm.iterchildren("Event"))
+                if int(event.attrib["type_id"]) != EVENT_TYPE_DELETED_EVENT
+            ]
             for idx, event_elm in enumerate(events_list):
                 next_event_elm = (
                     events_list[idx + 1]
@@ -672,11 +677,6 @@ def deserialize(self, inputs: OptaInputs) -> EventDataset:
                         f"Set end of period {period.id} to {timestamp}"
                     )
                     period.end_timestamp = timestamp
-                elif type_id == EVENT_TYPE_DELETED_EVENT:
-                    logger.debug(
-                        f"Skipping event {event_id} because it is a deleted event (type id - {type_id})"
-                    )
-                    continue
                 else:
                     if not period.start_timestamp:
                         # not started yet
@@ -792,7 +792,10 @@ def deserialize(self, inputs: OptaInputs) -> EventDataset:
                             **duel_event_kwargs,
                             **generic_event_kwargs,
                         )
-                    elif type_id == EVENT_TYPE_INTERCEPTION:
+                    elif type_id in (
+                        EVENT_TYPE_INTERCEPTION,
+                        EVENT_TYPE_BLOCKED_PASS,
+                    ):
                         interception_event_kwargs = _parse_interception(
                             raw_qualifiers, team, next_event_elm
                         )

diff --git a/kloppy/infra/serializers/event/wyscout/deserializer_v2.py b/kloppy/infra/serializers/event/wyscout/deserializer_v2.py
@@ -510,7 +510,8 @@ def deserialize(self, inputs: WyscoutInputs) -> EventDataset:
                         **generic_event_args,
                     )
 
-                # If also an interception event, add this before other event (It is a tag on multiple wyscout_events)
+                # Since Interception is not an event in wyscout v2 but a tag for pass, touch and duel. Therefore,
+                # we convert those duels and touch events to an interception. And insert interception before passes.
                 if _has_tag(raw_event, wyscout_tags.INTERCEPTION):
                     interception_event_args = _parse_interception(
                         raw_event, next_event
@@ -520,13 +521,16 @@ def deserialize(self, inputs: WyscoutInputs) -> EventDataset:
                         **generic_event_args,
                     )
 
-                    if event.event_type.name != "RECOVERY":
+                    if event.event_type.name == "DUEL":
+                        # when DuelEvent is interception, we need to overwrite this and the previous DuelEvent
+                        events = events[:-1]
+                        event = interception_event
+                    elif event.event_name in ["recovery", "miscontrol"]:
+                        event = interception_event
+                    elif event.event_name in ["pass", "clearance"]:
                         events.append(
                             transformer.transform_event(interception_event)
                         )
-                    else:
-                        # overwrite recovery event
-                        event = interception_event
 
                 if event and self.should_include_event(event):
                     events.append(transformer.transform_event(event))

diff --git a/kloppy/tests/test_wyscout.py b/kloppy/tests/test_wyscout.py
@@ -70,7 +70,7 @@ def test_correct_v2_deserialization(self, event_v2_data: Path):
         assert dataset.records[2].coordinates == Point(29.0, 6.0)
         assert dataset.events[11].event_type == EventType.MISCONTROL
         assert dataset.events[34].event_type == EventType.INTERCEPTION
-        assert dataset.events[144].event_type == EventType.CLEARANCE
+        assert dataset.events[143].event_type == EventType.CLEARANCE
 
         assert (
             dataset.events[39].get_qualifier_value(DuelQualifier)
@@ -81,11 +81,11 @@ def test_correct_v2_deserialization(self, event_v2_data: Path):
             == DuelType.AERIAL
         )
         assert (
-            dataset.events[269].get_qualifier_values(DuelQualifier)[2].value
+            dataset.events[268].get_qualifier_values(DuelQualifier)[2].value
             == DuelType.SLIDING_TACKLE
         )
         assert (
-            dataset.events[302].get_qualifier_value(GoalkeeperQualifier)
+            dataset.events[301].get_qualifier_value(GoalkeeperQualifier)
             == GoalkeeperActionType.SAVE
         )