def test_read(self):
base_dir = os.path.dirname(__file__)
with open(f"{base_dir}/files/epts_meta.xml", "rb") as metadata_fp:
metadata = load_metadata(metadata_fp)
with open(f"{base_dir}/files/epts_raw.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_meta.xml",
"rb") as metadata_fp, open(f"{base_dir}/files/epts_raw.txt",
"rb") as raw_data:
metadata = load_metadata(metadata_fp)
records = read_raw_data(raw_data,
metadata,
sensor_ids=["heartbeat"])
data_frame = DataFrame.from_records(records)
assert "player_1_max_heartbeat" in data_frame.columns
assert "player_1_x" not in data_frame.columns
def test_regex(self):
base_dir = os.path.dirname(__file__)
with open(f"{base_dir}/files/epts_meta.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)
# NOTE: use broken example of FIFA
result = regex.search(
"1779143:,-2.013,-500,100,9.63,9.80,4,5,177,182;-461,-615,-120,99,900,9.10,4,5,170,179;-2638,3478,120,110,1.15,5.20,3,4,170,175;:-2656,367,100:"
)
assert result is not None
def deserialize(self,
inputs: Dict[str, Readable],
options: Dict = None) -> EventDataset:
"""
Deserialize Metrica Sports event data json format into a `EventDataset`.
Parameters
----------
inputs : dict
input `raw_data` should point to a `Readable` object containing
the 'json' formatted event data. input `metadata` should point
to a `Readable` object containing the `xml` metadata file.
options : dict
Options for deserialization of the Metrica Sports event json file.
Possible options are `event_types` (list of event types) to specify
the event types that should be returned. Valid types: "shot", "pass",
"carry", "take_on" and "generic". Generic is everything other than
the first 4. Those events are barely parsed. This type of event can
be used to do the parsing yourself.
Every event has a 'raw_event' attribute which contains the original
dictionary.
Returns
-------
dataset : EventDataset
Raises
------
See Also
--------
Examples
--------
>>> serializer = MetricaEventsJsonSerializer()
>>> with open("events.json", "rb") as raw_data, \
>>> open("metadata.xml", "rb") as metadata:
>>>
>>> dataset = serializer.deserialize(
>>> inputs={
>>> 'raw_data': raw_data,
>>> 'metadata': metadata
>>> },
>>> options={
>>> 'event_types': ["pass", "take_on", "carry", "shot"]
>>> }
>>> )
"""
self.__validate_inputs(inputs)
if not options:
options = {}
with performance_logging("load data", logger=logger):
raw_events = json.load(inputs["raw_data"])
metadata = load_metadata(inputs["metadata"],
provider=Provider.METRICA)
with performance_logging("parse data", logger=logger):
wanted_event_types = [
EventType[event_type.upper()]
for event_type in options.get("event_types", [])
]
events = []
for raw_event in 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 Exception(
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]
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,
)
if event_type in MS_PASS_TYPES:
pass_event_kwargs = _parse_pass(
event=raw_event,
subtypes=subtypes,
team=team,
)
event = PassEvent(
**pass_event_kwargs,
**generic_event_kwargs,
)
elif event_type == MS_EVENT_TYPE_SHOT:
shot_event_kwargs = _parse_shot(event=raw_event,
subtypes=subtypes)
event = ShotEvent(**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(**take_on_event_kwargs,
**generic_event_kwargs)
elif event_type == MS_EVENT_TYPE_CARRY:
carry_event_kwargs = _parse_carry(event=raw_event, )
event = CarryEvent(
**carry_event_kwargs,
**generic_event_kwargs,
)
else:
event = GenericEvent(
result=None,
event_name=raw_event["type"]["name"],
**generic_event_kwargs,
)
if (not wanted_event_types
or event.event_type in wanted_event_types):
events.append(event)
return EventDataset(
metadata=metadata,
records=events,
)