def test_read(self):
base_dir = os.path.dirname(__file__)
with open(
f"{base_dir}/files/epts_metrica_metadata.xml", "rb"
) as metadata_fp:
metadata = load_metadata(metadata_fp)
with open(
f"{base_dir}/files/epts_metrica_tracking.txt", "rb"
) as raw_data:
iterator = read_raw_data(raw_data, metadata)
with performance_logging("load"):
assert list(iterator)
def test_skip_sensors(self):
base_dir = os.path.dirname(__file__)
with open(
f"{base_dir}/files/epts_metrica_metadata.xml", "rb"
) as metadata_fp, open(
f"{base_dir}/files/epts_metrica_tracking.txt", "rb"
) as raw_data:
metadata = load_metadata(metadata_fp)
records = read_raw_data(raw_data, metadata, sensor_ids=["speed"])
data_frame = DataFrame.from_records(records)
assert "player_Track_1_s" in data_frame.columns
assert "player_Track_1_x" not in data_frame.columns
def test_regex(self):
base_dir = os.path.dirname(__file__)
with open(
f"{base_dir}/files/epts_metrica_metadata.xml", "rb"
) as metadata_fp:
metadata = load_metadata(metadata_fp)
regex_str = build_regex(
metadata.data_format_specifications[0],
metadata.player_channels,
metadata.sensors,
)
regex = re.compile(regex_str)
result = regex.search(
"450:0.30602,0.97029,5,1.1127902269363403;0.48496,0.73308,1,5.2345757484436035;0.38851,0.97786,2,3.32438325881958;0.48414,0.92730,0,5.878201961517334;0.25661,0.60984,8,4.325275421142578;0.24170,0.49115,7,1.9475973844528198;0.38878,0.47540,4,1.8947187662124634;0.31854,0.89154,3,2.741175413131714;NaN,NaN,-1,NaN;0.25658,0.73095,9,3.418901205062866;NaN,NaN,10,NaN;0.32513,0.89532,19,0.6292267441749573;0.30286,0.98231,17,1.287971019744873;0.39954,0.93079,12,3.8761117458343506;0.23961,0.63742,20,1.6785365343093872;0.38319,0.81908,14,3.611333131790161;0.49984,0.91227,13,5.139825820922852;0.46733,0.49708,16,3.1186790466308594;0.26311,0.48218,18,2.309934139251709;0.53767,0.81373,11,6.402815818786621;0.39631,0.77277,15,3.338606357574463;NaN,NaN,21,NaN:0.52867,0.70690,NaN"
)
assert result is not None
def deserialize(self, inputs: MetricaJsonEventDataInputs) -> EventDataset:
with performance_logging("load data", logger=logger):
raw_events = json.load(inputs.event_data)
metadata = load_metadata(inputs.meta_data,
provider=Provider.METRICA)
transformer = self.get_transformer(
length=metadata.pitch_dimensions.length,
width=metadata.pitch_dimensions.width,
)
with performance_logging("parse data", logger=logger):
events = []
for i, raw_event in enumerate(raw_events["data"]):
if raw_event["team"]["id"] == metadata.teams[0].team_id:
team = metadata.teams[0]
elif raw_event["team"]["id"] == metadata.teams[1].team_id:
team = metadata.teams[1]
else:
raise DeserializationError(
f"Unknown team_id {raw_event['team']['id']}")
player = team.get_player_by_id(raw_event["from"]["id"])
event_type = raw_event["type"]["id"]
subtypes = _parse_subtypes(raw_event)
period = [
period for period in metadata.periods
if period.id == raw_event["period"]
][0]
previous_event = raw_events["data"][i - 1]
generic_event_kwargs = dict(
# from DataRecord
period=period,
timestamp=raw_event["start"]["time"],
ball_owning_team=_parse_ball_owning_team(event_type, team),
ball_state=BallState.ALIVE,
# from Event
event_id=None,
team=team,
player=player,
coordinates=(_parse_coordinates(raw_event["start"])),
raw_event=raw_event,
)
iteration_events = []
if event_type in MS_PASS_TYPES:
pass_event_kwargs = _parse_pass(
event=raw_event,
previous_event=previous_event,
subtypes=subtypes,
team=team,
)
event = PassEvent.create(
**pass_event_kwargs,
**generic_event_kwargs,
)
elif event_type == MS_EVENT_TYPE_SHOT:
shot_event_kwargs = _parse_shot(
event=raw_event,
previous_event=previous_event,
subtypes=subtypes,
)
event = ShotEvent.create(
**shot_event_kwargs,
**generic_event_kwargs,
)
elif subtypes and MS_EVENT_TYPE_DRIBBLE in subtypes:
take_on_event_kwargs = _parse_take_on(subtypes=subtypes)
event = TakeOnEvent.create(
qualifiers=None,
**take_on_event_kwargs,
**generic_event_kwargs,
)
elif event_type == MS_EVENT_TYPE_CARRY:
carry_event_kwargs = _parse_carry(event=raw_event, )
event = CarryEvent.create(
qualifiers=None,
**carry_event_kwargs,
**generic_event_kwargs,
)
elif event_type == MS_EVENT_TYPE_RECOVERY:
event = RecoveryEvent.create(
result=None,
qualifiers=None,
**generic_event_kwargs,
)
elif event_type == MS_EVENT_TYPE_FOUL_COMMITTED:
event = FoulCommittedEvent.create(
result=None,
qualifiers=None,
**generic_event_kwargs,
)
else:
event = GenericEvent.create(
result=None,
qualifiers=None,
event_name=raw_event["type"]["name"],
**generic_event_kwargs,
)
if self.should_include_event(event):
events.append(transformer.transform_event(event))
# Checks if the event ended out of the field and adds a synthetic out event
if event.result in OUT_EVENT_RESULTS:
generic_event_kwargs["ball_state"] = BallState.DEAD
if raw_event["end"]["x"]:
generic_event_kwargs[
"coordinates"] = _parse_coordinates(
raw_event["end"])
generic_event_kwargs["timestamp"] = raw_event["end"][
"time"]
event = BallOutEvent.create(
result=None,
qualifiers=None,
**generic_event_kwargs,
)
if self.should_include_event(event):
events.append(transformer.transform_event(event))
return EventDataset(
metadata=metadata,
records=events,
)