def test_pitchdimensions_properties(self):
pitch_without_scale = PitchDimensions(x_dim=Dimension(-100, 100),
y_dim=Dimension(-50, 50))
assert pitch_without_scale.length is None
assert pitch_without_scale.width is None
pitch_with_scale = PitchDimensions(
x_dim=Dimension(-100, 100),
y_dim=Dimension(-50, 50),
x_per_meter=20 / 12,
y_per_meter=10 / 8,
)
assert pitch_with_scale.length == 120
assert pitch_with_scale.width == 80
def test_pitchdimensions_properties(self):
pitch_without_scale = PitchDimensions(
x_dim=Dimension(-100, 100), y_dim=Dimension(-50, 50)
)
assert pitch_without_scale.length is None
assert pitch_without_scale.width is None
pitch_with_scale = PitchDimensions(
x_dim=Dimension(-100, 100),
y_dim=Dimension(-50, 50),
length=120,
width=80,
)
assert pitch_with_scale.length == 120
assert pitch_with_scale.width == 80
def _get_tracking_dataset(self):
home_team = Team(team_id="home", name="home", ground=Ground.HOME)
away_team = Team(team_id="away", name="away", ground=Ground.AWAY)
teams = [home_team, away_team]
periods = [
Period(
id=1,
start_timestamp=0.0,
end_timestamp=10.0,
attacking_direction=AttackingDirection.HOME_AWAY,
),
Period(
id=2,
start_timestamp=15.0,
end_timestamp=25.0,
attacking_direction=AttackingDirection.AWAY_HOME,
),
]
metadata = Metadata(
flags=~(DatasetFlag.BALL_OWNING_TEAM | DatasetFlag.BALL_STATE),
pitch_dimensions=PitchDimensions(x_dim=Dimension(0, 100),
y_dim=Dimension(-50, 50)),
orientation=Orientation.HOME_TEAM,
frame_rate=25,
periods=periods,
teams=teams,
score=None,
provider=None,
)
tracking_data = TrackingDataset(
metadata=metadata,
records=[
Frame(
frame_id=1,
timestamp=0.1,
ball_owning_team=None,
ball_state=None,
period=periods[0],
players_coordinates={},
ball_coordinates=Point(x=100, y=-50),
),
Frame(
frame_id=2,
timestamp=0.2,
ball_owning_team=None,
ball_state=None,
period=periods[0],
players_coordinates={
Player(team=home_team, player_id="home_1", jersey_no=1):
Point(x=15, y=35)
},
ball_coordinates=Point(x=0, y=50),
),
],
)
return tracking_data
def _get_tracking_dataset(self):
periods = [
Period(
id=1,
start_timestamp=0.0,
end_timestamp=10.0,
attacking_direction=AttackingDirection.HOME_AWAY,
),
Period(
id=2,
start_timestamp=15.0,
end_timestamp=25.0,
attacking_direction=AttackingDirection.AWAY_HOME,
),
]
tracking_data = TrackingDataset(
flags=~(DatasetFlag.BALL_OWNING_TEAM | DatasetFlag.BALL_STATE),
pitch_dimensions=PitchDimensions(x_dim=Dimension(0, 100),
y_dim=Dimension(-50, 50)),
orientation=Orientation.HOME_TEAM,
frame_rate=25,
records=[
Frame(
frame_id=1,
timestamp=0.1,
ball_owning_team=None,
ball_state=None,
period=periods[0],
away_team_player_positions={},
home_team_player_positions={},
ball_position=Point(x=100, y=-50),
),
Frame(
frame_id=2,
timestamp=0.2,
ball_owning_team=None,
ball_state=None,
period=periods[0],
away_team_player_positions={"1": Point(x=10, y=20)},
home_team_player_positions={"1": Point(x=15, y=35)},
ball_position=Point(x=0, y=50),
),
],
periods=periods,
)
return tracking_data
def test_transform(self):
tracking_data = self._get_tracking_dataset()
# orientation change AND dimension scale
transformed_dataset = tracking_data.transform(
to_orientation="AWAY_TEAM",
to_pitch_dimensions=[[0, 1], [0, 1]],
)
assert transformed_dataset.frames[0].ball_coordinates == Point3D(x=0,
y=1,
z=0)
assert transformed_dataset.frames[1].ball_coordinates == Point3D(x=1,
y=0,
z=1)
assert (
transformed_dataset.metadata.orientation == Orientation.AWAY_TEAM)
assert transformed_dataset.metadata.coordinate_system is None
assert (
transformed_dataset.metadata.pitch_dimensions == PitchDimensions(
x_dim=Dimension(min=0, max=1), y_dim=Dimension(min=0, max=1)))
def _load_pitch_dimensions(
meta_data_elm, sensors: List[Sensor]) -> Union[None, PitchDimensions]:
normalized = False
for sensor in sensors:
if sensor.sensor_id == 'position':
if sensor.channels[0].unit == 'normalized':
normalized = True
break
field_size_path = objectify.ObjectPath("Metadata.Sessions.Session[0]")
field_size_elm = field_size_path.find(meta_data_elm).find('FieldSize')
if field_size_elm is not None and normalized:
return PitchDimensions(
x_dim=Dimension(0, 1),
y_dim=Dimension(0, 1),
x_per_meter=1 / int(field_size_elm.find('Width')),
y_per_meter=1 / int(field_size_elm.find('Height')))
else:
return None
def _load_pitch_dimensions(
metadata_elm, sensors: List[Sensor]) -> Union[None, PitchDimensions]:
normalized = False
for sensor in sensors:
if sensor.sensor_id == "position":
if sensor.channels[0].unit == "normalized":
normalized = True
break
field_size_path = objectify.ObjectPath("Metadata.Sessions.Session[0]")
field_size_elm = field_size_path.find(metadata_elm).find("FieldSize")
if field_size_elm is not None and normalized:
return PitchDimensions(
x_dim=Dimension(0, 1),
y_dim=Dimension(0, 1),
length=int(field_size_elm.find("Width")),
width=int(field_size_elm.find("Height")),
)
else:
return None
def deserialize(self,
inputs: Dict[str, Readable],
options: Dict = None) -> TrackingDataset:
"""
Deserialize Metrica tracking data into a `TrackingDataset`.
Parameters
----------
inputs : dict
input `raw_data_home` should point to a `Readable` object containing
the 'csv' formatted raw data for the home team. input `raw_data_away` should point
to a `Readable` object containing the 'csv' formatted raw data for the away team.
options : dict
Options for deserialization of the Metrica file. Possible options are
`sample_rate` (float between 0 and 1) to specify the amount of
frames that should be loaded, `limit` to specify the max number of
frames that will be returned.
Returns
-------
dataset : TrackingDataset
Raises
------
ValueError when both input files don't seem to belong to each other
See Also
--------
Examples
--------
>>> serializer = MetricaTrackingSerializer()
>>> with open("Sample_Game_1_RawTrackingData_Away_Team.csv", "rb") as raw_home, \
>>> open("Sample_Game_1_RawTrackingData_Home_Team.csv", "rb") as raw_away:
>>>
>>> dataset = serializer.deserialize(
>>> inputs={
>>> 'raw_data_home': raw_home,
>>> 'raw_data_away': raw_away
>>> },
>>> options={
>>> 'sample_rate': 1/12
>>> }
>>> )
"""
self.__validate_inputs(inputs)
if not options:
options = {}
sample_rate = float(options.get('sample_rate', 1.0))
limit = int(options.get('limit', 0))
# consider reading this from data
frame_rate = 25
with performance_logging("prepare", logger=logger):
home_iterator = self.__create_iterator(inputs['raw_data_home'],
sample_rate, frame_rate)
away_iterator = self.__create_iterator(inputs['raw_data_away'],
sample_rate, frame_rate)
partial_frames = zip(home_iterator, away_iterator)
with performance_logging("loading", logger=logger):
frames = []
periods = []
partial_frame_type = self.__PartialFrame
home_partial_frame: partial_frame_type
away_partial_frame: partial_frame_type
for n, (home_partial_frame,
away_partial_frame) in enumerate(partial_frames):
self.__validate_partials(home_partial_frame,
away_partial_frame)
period: Period = home_partial_frame.period
frame_id: int = home_partial_frame.frame_id
frame = Frame(frame_id=frame_id,
timestamp=frame_id / frame_rate -
period.start_timestamp,
ball_position=home_partial_frame.ball_position,
home_team_player_positions=home_partial_frame.
player_positions,
away_team_player_positions=away_partial_frame.
player_positions,
period=period,
ball_state=None,
ball_owning_team=None)
frames.append(frame)
if not periods or period.id != periods[-1].id:
periods.append(period)
if not period.attacking_direction_set:
period.set_attacking_direction(
attacking_direction=attacking_direction_from_frame(
frame))
n += 1
if limit and n >= limit:
break
orientation = (
Orientation.FIXED_HOME_AWAY if periods[0].attacking_direction
== AttackingDirection.HOME_AWAY else Orientation.FIXED_AWAY_HOME)
return TrackingDataset(
flags=~(DatasetFlag.BALL_STATE | DatasetFlag.BALL_OWNING_TEAM),
frame_rate=frame_rate,
orientation=orientation,
pitch_dimensions=PitchDimensions(x_dim=Dimension(0, 1),
y_dim=Dimension(0, 1)),
periods=periods,
records=frames)
def deserialize(self,
inputs: Dict[str, Readable],
options: Dict = None) -> EventDataset:
"""
Deserialize StatsBomb event data into a `EventDataset`.
Parameters
----------
inputs : dict
input `event_data` should point to a `Readable` object containing
the 'json' formatted event data. input `lineup_data` should point
to a `Readable` object containing the 'json' formatted lineup data.
options : dict
Options for deserialization of the StatsBomb 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 = StatsBombSerializer()
>>> with open("events/12312312.json", "rb") as event_data, \
>>> open("lineups/123123123.json", "rb") as lineup_data:
>>>
>>> dataset = serializer.deserialize(
>>> inputs={
>>> 'event_data': event_data,
>>> 'lineup_data': lineup_data
>>> },
>>> 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['event_data'])
home_lineup, away_lineup = json.load(inputs['lineup_data'])
shot_fidelity_version, xy_fidelity_version = _determine_xy_fidelity_versions(
raw_events)
logger.info(
f"Determined Fidelity versions: shot v{shot_fidelity_version} / XY v{xy_fidelity_version}"
)
with performance_logging("parse data", logger=logger):
home_player_map = {
player['player_id']: str(player['jersey_number'])
for player in home_lineup['lineup']
}
away_player_map = {
player['player_id']: str(player['jersey_number'])
for player in away_lineup['lineup']
}
wanted_event_types = [
EventType[event_type.upper()]
for event_type in options.get('event_types', [])
]
periods = []
period = None
events = []
for raw_event in raw_events:
if raw_event['team']['id'] == home_lineup['team_id']:
team = Team.HOME
current_team_map = home_player_map
elif raw_event['team']['id'] == away_lineup['team_id']:
team = Team.AWAY
current_team_map = away_player_map
else:
raise Exception(
f"Unknown team_id {raw_event['team']['id']}")
if raw_event['possession_team']['id'] == home_lineup[
'team_id']:
possession_team = Team.HOME
elif raw_event['possession_team']['id'] == away_lineup[
'team_id']:
possession_team = Team.AWAY
else:
raise Exception(
f"Unknown possession_team_id: {raw_event['possession_team']}"
)
timestamp = parse_str_ts(raw_event['timestamp'])
period_id = int(raw_event['period'])
if not period or period.id != period_id:
period = Period(id=period_id,
start_timestamp=timestamp if not period
else timestamp + period.end_timestamp,
end_timestamp=None)
periods.append(period)
else:
period.end_timestamp = period.start_timestamp + timestamp
player_jersey_no = None
if 'player' in raw_event:
player_jersey_no = current_team_map[raw_event['player']
['id']]
event_type = raw_event['type']['id']
if event_type == SB_EVENT_TYPE_SHOT:
fidelity_version = shot_fidelity_version
elif event_type in (SB_EVENT_TYPE_CARRY, SB_EVENT_TYPE_DRIBBLE,
SB_EVENT_TYPE_PASS):
fidelity_version = xy_fidelity_version
else:
# TODO: Uh ohhhh.. don't know which one to pick
fidelity_version = xy_fidelity_version
generic_event_kwargs = dict(
# from DataRecord
period=period,
timestamp=timestamp,
ball_owning_team=possession_team,
ball_state=BallState.ALIVE,
# from Event
event_id=raw_event['id'],
team=team,
player_jersey_no=player_jersey_no,
position=(_parse_position(raw_event.get('location'),
fidelity_version)
if 'location' in raw_event else None),
raw_event=raw_event)
if event_type == SB_EVENT_TYPE_PASS:
pass_event_kwargs = _parse_pass(
pass_dict=raw_event['pass'],
current_team_map=current_team_map,
fidelity_version=fidelity_version)
event = PassEvent(
# TODO: Consider moving this to _parse_pass
receive_timestamp=timestamp + raw_event['duration'],
**pass_event_kwargs,
**generic_event_kwargs)
elif event_type == SB_EVENT_TYPE_SHOT:
shot_event_kwargs = _parse_shot(
shot_dict=raw_event['shot'])
event = ShotEvent(**shot_event_kwargs,
**generic_event_kwargs)
# For dribble and carry the definitions
# are flipped between Statsbomb and kloppy
elif event_type == SB_EVENT_TYPE_DRIBBLE:
take_on_event_kwargs = _parse_take_on(
take_on_dict=raw_event['dribble'])
event = TakeOnEvent(**take_on_event_kwargs,
**generic_event_kwargs)
elif event_type == SB_EVENT_TYPE_CARRY:
carry_event_kwargs = _parse_carry(
carry_dict=raw_event['carry'],
fidelity_version=fidelity_version)
event = CarryEvent(
# TODO: Consider moving this to _parse_carry
end_timestamp=timestamp + raw_event['duration'],
**carry_event_kwargs,
**generic_event_kwargs)
else:
event = GenericEvent(result=None, **generic_event_kwargs)
if not wanted_event_types or event.event_type in wanted_event_types:
events.append(event)
return EventDataset(flags=DatasetFlag.BALL_OWNING_TEAM,
orientation=Orientation.ACTION_EXECUTING_TEAM,
pitch_dimensions=PitchDimensions(
x_dim=Dimension(0, 120),
y_dim=Dimension(0, 80)),
periods=periods,
records=events)
def deserialize(self,
inputs: Dict[str, Readable],
options: Dict = None) -> TrackingDataSet:
"""
Deserialize TRACAB tracking data into a `TrackingDataSet`.
Parameters
----------
inputs : dict
input `raw_data` should point to a `Readable` object containing
the 'csv' formatted raw data. input `meta_data` should point to
the xml metadata data.
options : dict
Options for deserialization of the TRACAB file. Possible options are
`only_alive` (boolean) to specify that only frames with alive ball state
should be loaded, or `sample_rate` (float between 0 and 1) to specify
the amount of frames that should be loaded.
Returns
-------
data_set : TrackingDataSet
Raises
------
-
See Also
--------
Examples
--------
>>> serializer = TRACABSerializer()
>>> with open("metadata.xml", "rb") as meta, \
>>> open("raw.dat", "rb") as raw:
>>> data_set = serializer.deserialize(
>>> inputs={
>>> 'meta_data': meta,
>>> 'raw_data': raw
>>> },
>>> options={
>>> 'only_alive': True,
>>> 'sample_rate': 1/12
>>> }
>>> )
"""
self.__validate_inputs(inputs)
if not options:
options = {}
sample_rate = float(options.get('sample_rate', 1.0))
only_alive = bool(options.get('only_alive', True))
with performance_logging("Loading metadata"):
match = objectify.fromstring(inputs['meta_data'].read()).match
frame_rate = int(match.attrib['iFrameRateFps'])
pitch_size_width = float(match.attrib['fPitchXSizeMeters'])
pitch_size_height = float(match.attrib['fPitchYSizeMeters'])
periods = []
for period in match.iterchildren(tag='period'):
start_frame_id = int(period.attrib['iStartFrame'])
end_frame_id = int(period.attrib['iEndFrame'])
if start_frame_id != 0 or end_frame_id != 0:
periods.append(
Period(id=int(period.attrib['iId']),
start_timestamp=start_frame_id / frame_rate,
end_timestamp=end_frame_id / frame_rate))
with performance_logging("Loading data"):
def _iter():
n = 0
sample = 1. / sample_rate
for line in inputs['raw_data'].readlines():
line = line.strip().decode("ascii")
if not line:
continue
frame_id = int(line[:10].split(":", 1)[0])
if only_alive and not line.endswith("Alive;:"):
continue
for period in periods:
if period.contains(frame_id / frame_rate):
if n % sample == 0:
yield period, line
n += 1
frames = []
for period, line in _iter():
frame = self._frame_from_line(period, line, frame_rate)
frames.append(frame)
if not period.attacking_direction_set:
period.set_attacking_direction(
attacking_direction=attacking_direction_from_frame(
frame))
orientation = (
Orientation.FIXED_HOME_AWAY if periods[0].attacking_direction
== AttackingDirection.HOME_AWAY else Orientation.FIXED_AWAY_HOME)
return TrackingDataSet(flags=DataSetFlag.BALL_OWNING_TEAM
| DataSetFlag.BALL_STATE,
frame_rate=frame_rate,
orientation=orientation,
pitch_dimensions=PitchDimensions(
x_dim=Dimension(-1 * pitch_size_width / 2,
pitch_size_width / 2),
y_dim=Dimension(-1 * pitch_size_height / 2,
pitch_size_height / 2),
x_per_meter=100,
y_per_meter=100),
periods=periods,
records=frames)
def deserialize(
self, inputs: Dict[str, Readable], options: Dict = None
) -> TrackingDataset:
"""
Deserialize TRACAB tracking data into a `TrackingDataset`.
Parameters
----------
inputs : dict
input `raw_data` should point to a `Readable` object containing
the 'csv' formatted raw data. input `metadata` should point to
the xml metadata data.
options : dict
Options for deserialization of the TRACAB file. Possible options are
`only_alive` (boolean) to specify that only frames with alive ball state
should be loaded, or `sample_rate` (float between 0 and 1) to specify
the amount of frames that should be loaded, `limit` to specify the max number of
frames that will be returned.
Returns
-------
dataset : TrackingDataset
Raises
------
-
See Also
--------
Examples
--------
>>> serializer = TRACABSerializer()
>>> with open("metadata.xml", "rb") as meta, \
>>> open("raw.dat", "rb") as raw:
>>> dataset = serializer.deserialize(
>>> inputs={
>>> 'metadata': meta,
>>> 'raw_data': raw
>>> },
>>> options={
>>> 'only_alive': True,
>>> 'sample_rate': 1/12
>>> }
>>> )
"""
self.__validate_inputs(inputs)
if not options:
options = {}
sample_rate = float(options.get("sample_rate", 1.0))
limit = int(options.get("limit", 0))
only_alive = bool(options.get("only_alive", True))
# TODO: also used in Metrica, extract to a method
home_team = Team(team_id="home", name="home", ground=Ground.HOME)
away_team = Team(team_id="away", name="away", ground=Ground.AWAY)
teams = [home_team, away_team]
with performance_logging("Loading metadata", logger=logger):
match = objectify.fromstring(inputs["metadata"].read()).match
frame_rate = int(match.attrib["iFrameRateFps"])
pitch_size_width = float(match.attrib["fPitchXSizeMeters"])
pitch_size_height = float(match.attrib["fPitchYSizeMeters"])
periods = []
for period in match.iterchildren(tag="period"):
start_frame_id = int(period.attrib["iStartFrame"])
end_frame_id = int(period.attrib["iEndFrame"])
if start_frame_id != 0 or end_frame_id != 0:
periods.append(
Period(
id=int(period.attrib["iId"]),
start_timestamp=start_frame_id / frame_rate,
end_timestamp=end_frame_id / frame_rate,
)
)
with performance_logging("Loading data", logger=logger):
def _iter():
n = 0
sample = 1.0 / sample_rate
for line_ in inputs["raw_data"].readlines():
line_ = line_.strip().decode("ascii")
if not line_:
continue
frame_id = int(line_[:10].split(":", 1)[0])
if only_alive and not line_.endswith("Alive;:"):
continue
for period_ in periods:
if period_.contains(frame_id / frame_rate):
if n % sample == 0:
yield period_, line_
n += 1
frames = []
for n, (period, line) in enumerate(_iter()):
frame = self._frame_from_line(teams, period, line, frame_rate)
frames.append(frame)
if not period.attacking_direction_set:
period.set_attacking_direction(
attacking_direction=attacking_direction_from_frame(
frame
)
)
if limit and n >= limit:
break
orientation = (
Orientation.FIXED_HOME_AWAY
if periods[0].attacking_direction == AttackingDirection.HOME_AWAY
else Orientation.FIXED_AWAY_HOME
)
metadata = Metadata(
teams=teams,
periods=periods,
pitch_dimensions=PitchDimensions(
x_dim=Dimension(
-1 * pitch_size_width / 2, pitch_size_width / 2
),
y_dim=Dimension(
-1 * pitch_size_height / 2, pitch_size_height / 2
),
x_per_meter=100,
y_per_meter=100,
),
score=None,
frame_rate=frame_rate,
orientation=orientation,
provider=Provider.TRACAB,
flags=DatasetFlag.BALL_OWNING_TEAM | DatasetFlag.BALL_STATE,
)
return TrackingDataset(
records=frames,
metadata=metadata,
)
def deserialize(
self, inputs: Dict[str, Readable], options: Dict = None
) -> EventDataset:
WyscoutSerializer.__validate_inputs(inputs)
if not options:
options = {}
wanted_event_types = [
EventType[event_type.upper()]
for event_type in options.get("event_types", [])
]
with performance_logging("load data", logger=logger):
raw_events = json.load(inputs["event_data"])
periods = []
with performance_logging("parse data", logger=logger):
home_team_id, away_team_id = raw_events["teams"].keys()
home_team = _parse_team(raw_events, home_team_id, Ground.HOME)
away_team = _parse_team(raw_events, away_team_id, Ground.AWAY)
teams = {home_team_id: home_team, away_team_id: away_team}
players = dict(
[
(wyId, _players_to_dict(team.players))
for wyId, team in teams.items()
]
)
events = []
for idx, raw_event in enumerate(raw_events["events"]):
next_event = None
if (idx + 1) < len(raw_events["events"]):
next_event = raw_events["events"][idx + 1]
team_id = str(raw_event["teamId"])
player_id = str(raw_event["playerId"])
if (
len(periods) == 0
or periods[-1].id != raw_event["matchPeriod"]
):
periods.append(
Period(
id=raw_event["matchPeriod"],
start_timestamp=0,
end_timestamp=0,
)
)
generic_event_args = {
"event_id": raw_event["id"],
"raw_event": raw_event,
"coordinates": Point(
x=float(raw_event["positions"][0]["x"]),
y=float(raw_event["positions"][0]["y"]),
),
"team": teams[team_id],
"player": players[team_id][player_id]
if player_id != INVALID_PLAYER
else None,
"ball_owning_team": None,
"ball_state": None,
"period": periods[-1],
"timestamp": raw_event["eventSec"],
}
event = None
if raw_event["eventName"] == wyscout_events.SHOT.EVENT:
shot_event_args = _parse_shot(raw_event, next_event)
event = ShotEvent.create(
**shot_event_args, **generic_event_args
)
elif raw_event["eventName"] == wyscout_events.PASS.EVENT:
pass_event_args = _parse_pass(raw_event, next_event)
event = PassEvent.create(
**pass_event_args, **generic_event_args
)
elif raw_event["eventName"] == wyscout_events.FOUL.EVENT:
foul_event_args = _parse_foul(raw_event)
event = FoulCommittedEvent.create(
**foul_event_args, **generic_event_args
)
if any(
(_has_tag(raw_event, tag) for tag in wyscout_tags.CARD)
):
card_event_args = _parse_card(raw_event)
event = CardEvent.create(
**card_event_args, **generic_event_args
)
elif (
raw_event["eventName"] == wyscout_events.INTERRUPTION.EVENT
):
ball_out_event_args = _parse_ball_out(raw_event)
event = BallOutEvent.create(
**ball_out_event_args, **generic_event_args
)
elif raw_event["eventName"] == wyscout_events.FREE_KICK.EVENT:
set_piece_event_args = _parse_set_piece(
raw_event, next_event
)
if (
raw_event["subEventName"]
in wyscout_events.FREE_KICK.PASS_TYPES
):
event = PassEvent.create(
**set_piece_event_args, **generic_event_args
)
elif (
raw_event["subEventName"]
in wyscout_events.FREE_KICK.SHOT_TYPES
):
event = ShotEvent.create(
**set_piece_event_args, **generic_event_args
)
elif (
raw_event["eventName"]
== wyscout_events.OTHERS_ON_BALL.EVENT
):
recovery_event_args = _parse_recovery(raw_event)
event = RecoveryEvent.create(
**recovery_event_args, **generic_event_args
)
elif raw_event["eventName"] == wyscout_events.DUEL.EVENT:
takeon_event_args = _parse_takeon(raw_event)
event = TakeOnEvent.create(
**takeon_event_args, **generic_event_args
)
elif raw_event["eventName"] not in [
wyscout_events.SAVE.EVENT,
wyscout_events.OFFSIDE.EVENT,
]:
# The events SAVE and OFFSIDE are already merged with PASS and SHOT events
qualifiers = _generic_qualifiers(raw_event)
event = GenericEvent.create(
result=None,
qualifiers=qualifiers,
**generic_event_args
)
if event and _include_event(event, wanted_event_types):
events.append(event)
metadata = Metadata(
teams=[home_team, away_team],
periods=periods,
pitch_dimensions=PitchDimensions(
x_dim=Dimension(0, 100), y_dim=Dimension(0, 100)
),
score=None,
frame_rate=None,
orientation=Orientation.BALL_OWNING_TEAM,
flags=None,
provider=Provider.WYSCOUT,
)
return EventDataset(metadata=metadata, records=events)
def deserialize(self,
inputs: Dict[str, Readable],
options: Dict = None) -> EventDataset:
"""
Deserialize Opta event data into a `EventDataset`.
Parameters
----------
inputs : dict
input `f24_data` should point to a `Readable` object containing
the 'xml' formatted event data. input `f7_data` should point
to a `Readable` object containing the 'xml' formatted f7 data.
options : dict
Options for deserialization of the Opta 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 = OptaSerializer()
>>> with open("123_f24.xml", "rb") as f24_data, \
>>> open("123_f7.xml", "rb") as f7_data:
>>>
>>> dataset = serializer.deserialize(
>>> inputs={
>>> 'f24_data': f24_data,
>>> 'f7_data': f7_data
>>> },
>>> options={
>>> 'event_types': ["pass", "take_on", "carry", "shot"]
>>> }
>>> )
"""
self.__validate_inputs(inputs)
if not options:
options = {}
with performance_logging("load data", logger=logger):
f7_root = objectify.fromstring(inputs["f7_data"].read())
f24_root = objectify.fromstring(inputs["f24_data"].read())
wanted_event_types = [
EventType[event_type.upper()]
for event_type in options.get("event_types", [])
]
with performance_logging("parse data", logger=logger):
matchdata_path = objectify.ObjectPath(
"SoccerFeed.SoccerDocument.MatchData")
team_elms = list(
matchdata_path.find(f7_root).iterchildren("TeamData"))
home_score = None
away_score = None
for team_elm in team_elms:
if team_elm.attrib["Side"] == "Home":
home_score = team_elm.attrib["Score"]
home_team = _team_from_xml_elm(team_elm, f7_root)
elif team_elm.attrib["Side"] == "Away":
away_score = team_elm.attrib["Score"]
away_team = _team_from_xml_elm(team_elm, f7_root)
else:
raise Exception(f"Unknown side: {team_elm.attrib['Side']}")
score = Score(home=home_score, away=away_score)
teams = [home_team, away_team]
if len(home_team.players) == 0 or len(away_team.players) == 0:
raise Exception("LineUp incomplete")
game_elm = f24_root.find("Game")
periods = [
Period(
id=1,
start_timestamp=None,
end_timestamp=None,
),
Period(
id=2,
start_timestamp=None,
end_timestamp=None,
),
]
possession_team = None
events = []
for event_elm in game_elm.iterchildren("Event"):
event_id = event_elm.attrib["id"]
type_id = int(event_elm.attrib["type_id"])
timestamp = _parse_f24_datetime(event_elm.attrib["timestamp"])
period_id = int(event_elm.attrib["period_id"])
for period in periods:
if period.id == period_id:
break
else:
logger.debug(
f"Skipping event {event_id} because period doesn't match {period_id}"
)
continue
if type_id == EVENT_TYPE_START_PERIOD:
logger.debug(
f"Set start of period {period.id} to {timestamp}")
period.start_timestamp = timestamp
elif type_id == EVENT_TYPE_END_PERIOD:
logger.debug(
f"Set end of period {period.id} to {timestamp}")
period.end_timestamp = timestamp
else:
if not period.start_timestamp:
# not started yet
continue
if event_elm.attrib["team_id"] == home_team.team_id:
team = teams[0]
elif event_elm.attrib["team_id"] == away_team.team_id:
team = teams[1]
else:
raise Exception(
f"Unknown team_id {event_elm.attrib['team_id']}")
x = float(event_elm.attrib["x"])
y = float(event_elm.attrib["y"])
outcome = int(event_elm.attrib["outcome"])
raw_qualifiers = {
int(qualifier_elm.attrib["qualifier_id"]):
qualifier_elm.attrib.get("value")
for qualifier_elm in event_elm.iterchildren("Q")
}
player = None
if "player_id" in event_elm.attrib:
player = team.get_player_by_id(
event_elm.attrib["player_id"])
if type_id in BALL_OWNING_EVENTS:
possession_team = team
generic_event_kwargs = dict(
# from DataRecord
period=period,
timestamp=timestamp - period.start_timestamp,
ball_owning_team=possession_team,
ball_state=BallState.ALIVE,
# from Event
event_id=event_id,
team=team,
player=player,
coordinates=Point(x=x, y=y),
raw_event=event_elm,
)
if type_id == EVENT_TYPE_PASS:
pass_event_kwargs = _parse_pass(
raw_qualifiers, outcome)
event = PassEvent.create(
**pass_event_kwargs,
**generic_event_kwargs,
)
elif type_id == EVENT_TYPE_OFFSIDE_PASS:
pass_event_kwargs = _parse_offside_pass(raw_qualifiers)
event = PassEvent.create(
**pass_event_kwargs,
**generic_event_kwargs,
)
elif type_id == EVENT_TYPE_TAKE_ON:
take_on_event_kwargs = _parse_take_on(outcome)
event = TakeOnEvent.create(
qualifiers=None,
**take_on_event_kwargs,
**generic_event_kwargs,
)
elif type_id in (
EVENT_TYPE_SHOT_MISS,
EVENT_TYPE_SHOT_POST,
EVENT_TYPE_SHOT_SAVED,
EVENT_TYPE_SHOT_GOAL,
):
shot_event_kwargs = _parse_shot(
raw_qualifiers,
type_id,
coordinates=generic_event_kwargs["coordinates"],
)
kwargs = {}
kwargs.update(generic_event_kwargs)
kwargs.update(shot_event_kwargs)
event = ShotEvent.create(**kwargs)
elif type_id == EVENT_TYPE_RECOVERY:
event = RecoveryEvent.create(
result=None,
qualifiers=None,
**generic_event_kwargs,
)
elif type_id == EVENT_TYPE_FOUL_COMMITTED:
event = FoulCommittedEvent.create(
result=None,
qualifiers=None,
**generic_event_kwargs,
)
elif type_id in BALL_OUT_EVENTS:
generic_event_kwargs["ball_state"] = BallState.DEAD
event = BallOutEvent.create(
result=None,
qualifiers=None,
**generic_event_kwargs,
)
else:
event = GenericEvent.create(
**generic_event_kwargs,
result=None,
qualifiers=None,
event_name=_get_event_type_name(type_id),
)
if (not wanted_event_types
or event.event_type in wanted_event_types):
events.append(event)
metadata = Metadata(
teams=teams,
periods=periods,
pitch_dimensions=PitchDimensions(x_dim=Dimension(0, 100),
y_dim=Dimension(0, 100)),
score=score,
frame_rate=None,
orientation=Orientation.ACTION_EXECUTING_TEAM,
flags=DatasetFlag.BALL_OWNING_TEAM,
provider=Provider.OPTA,
)
return EventDataset(
metadata=metadata,
records=events,
)
def deserialize(self,
inputs: Dict[str, Readable],
options: Dict = None) -> TrackingDataset:
"""
Deserialize SkillCorner tracking data into a `TrackingDataset`.
Parameters
----------
inputs : dict
input `raw_data` should point to a `Readable` object containing
the 'json' formatted raw data. input `metadata` should point to
the json metadata data.
options : dict
Options for deserialization of the TRACAB file. Possible options are:
`include_empty_frames` (boolean): default = False to specify whether frames without
any players_coordinates or the ball_coordinates should be loaded
`sample_rate` (float between 0 and 1) to specify the amount of frames that should be loaded
and `limit` (int) to specify the max number of frames that will be returned.
Returns
-------
dataset : TrackingDataset
Raises
------
-
See Also
--------
Examples
--------
>>> serializer = SkillCornerSerializer()
>>> with open("match_data.json", "rb") as meta, \
>>> open("structured_data.json", "rb") as raw:
>>> dataset = serializer.deserialize(
>>> inputs={
>>> 'metadata': meta,
>>> 'raw_data': raw
>>> },
>>> options={
>>> }
>>> )
"""
self.__validate_inputs(inputs)
metadata = self.__load_json(inputs["metadata"])
raw_data = self.__load_json(inputs["raw_data"])
if not options:
options = {}
sample_rate = float(options.get("sample_rate", 1.0))
limit = int(options.get("limit", 0))
include_empty_frames = bool(options.get("include_empty_frames", False))
with performance_logging("Loading metadata", logger=logger):
periods = self.__get_periods(raw_data)
teamdict = {
metadata["home_team"].get("id"): "home_team",
metadata["away_team"].get("id"): "away_team",
}
player_id_to_team_dict = {
player["trackable_object"]: player["team_id"]
for player in metadata["players"]
}
player_dict = {
player["trackable_object"]: player
for player in metadata["players"]
}
referee_dict = {
ref["trackable_object"]: "referee"
for ref in metadata["referees"]
}
ball_id = metadata["ball"]["trackable_object"]
# there are different pitch_sizes in SkillCorner
pitch_size_width = metadata["pitch_width"]
pitch_size_length = metadata["pitch_length"]
home_team_id = metadata["home_team"]["id"]
away_team_id = metadata["away_team"]["id"]
players = {"HOME": {}, "AWAY": {}}
home_team = Team(
team_id=home_team_id,
name=metadata["home_team"]["name"],
ground=Ground.HOME,
)
self.home_team = home_team
away_team = Team(
team_id=away_team_id,
name=metadata["away_team"]["name"],
ground=Ground.AWAY,
)
self.away_team = away_team
teams = [home_team, away_team]
for player_id in player_dict.keys():
player = player_dict.get(player_id)
team_id = player["team_id"]
if team_id == home_team_id:
team_string = "HOME"
team = home_team
elif team_id == away_team_id:
team_string = "AWAY"
team = away_team
players[team_string][player_id] = Player(
player_id=f"{team.ground}_{player['number']}",
team=team,
jersey_no=player["number"],
name=f"{player['first_name']} {player['last_name']}",
first_name=player["first_name"],
last_name=player["last_name"],
starting=player["start_time"] == "00:00:00",
position=Position(
position_id=player["player_role"].get("id"),
name=player["player_role"].get("name"),
coordinates=None,
),
attributes={},
)
home_team.players = list(players["HOME"].values())
away_team.players = list(players["AWAY"].values())
anon_players = {"HOME": {}, "AWAY": {}}
with performance_logging("Loading data", logger=logger):
def _iter():
n = 0
sample = 1.0 / sample_rate
for frame in raw_data:
frame_period = frame["period"]
if frame_period is not None:
if n % sample == 0:
yield frame
n += 1
frames = []
n_frames = 0
for _frame in _iter():
# include frame if there is any tracking data, players or ball.
# or if include_empty_frames == True
if include_empty_frames or len(_frame["data"]) > 0:
frame = self._get_frame_data(
teams,
teamdict,
players,
player_id_to_team_dict,
periods,
player_dict,
anon_players,
ball_id,
referee_dict,
_frame,
)
frames.append(frame)
n_frames += 1
if limit and n_frames >= limit:
break
self._set_skillcorner_attacking_directions(frames, periods)
frame_rate = 10
orientation = (Orientation.HOME_TEAM if periods[1].attacking_direction
== AttackingDirection.HOME_AWAY else
Orientation.AWAY_TEAM)
metadata = Metadata(
teams=teams,
periods=periods,
pitch_dimensions=PitchDimensions(
x_dim=Dimension(-(pitch_size_length / 2),
(pitch_size_length / 2)),
y_dim=Dimension(-(pitch_size_width / 2),
(pitch_size_width / 2)),
x_per_meter=1,
y_per_meter=1,
),
score=Score(
home=metadata["home_team_score"],
away=metadata["away_team_score"],
),
frame_rate=frame_rate,
orientation=orientation,
provider=Provider.SKILLCORNER,
flags=~(DatasetFlag.BALL_STATE | DatasetFlag.BALL_OWNING_TEAM),
)
return TrackingDataset(
records=frames,
metadata=metadata,
)
def deserialize(
self, inputs: Dict[str, Readable], options: Dict = None
) -> EventDataset:
"""
Deserialize StatsBomb event data into a `EventDataset`.
Parameters
----------
inputs : dict
input `event_data` should point to a `Readable` object containing
the 'json' formatted event data. input `lineup_data` should point
to a `Readable` object containing the 'json' formatted lineup data.
options : dict
Options for deserialization of the StatsBomb 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 = StatsBombSerializer()
>>> with open("events/12312312.json", "rb") as event_data, \
>>> open("lineups/123123123.json", "rb") as lineup_data:
>>>
>>> dataset = serializer.deserialize(
>>> inputs={
>>> 'event_data': event_data,
>>> 'lineup_data': lineup_data
>>> },
>>> 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["event_data"])
home_lineup, away_lineup = json.load(inputs["lineup_data"])
(
shot_fidelity_version,
xy_fidelity_version,
) = _determine_xy_fidelity_versions(raw_events)
logger.info(
f"Determined Fidelity versions: shot v{shot_fidelity_version} / XY v{xy_fidelity_version}"
)
with performance_logging("parse data", logger=logger):
home_team = Team(
team_id=str(home_lineup["team_id"]),
name=home_lineup["team_name"],
ground=Ground.HOME,
)
home_team.players = [
Player(
player_id=str(player["player_id"]),
team=home_team,
name=player["player_name"],
jersey_no=int(player["jersey_number"]),
)
for player in home_lineup["lineup"]
]
away_team = Team(
team_id=str(away_lineup["team_id"]),
name=away_lineup["team_name"],
ground=Ground.AWAY,
)
away_team.players = [
Player(
player_id=str(player["player_id"]),
team=away_team,
name=player["player_name"],
jersey_no=int(player["jersey_number"]),
)
for player in away_lineup["lineup"]
]
teams = [home_team, away_team]
wanted_event_types = [
EventType[event_type.upper()]
for event_type in options.get("event_types", [])
]
periods = []
period = None
events = []
for raw_event in raw_events:
if raw_event["team"]["id"] == home_lineup["team_id"]:
team = teams[0]
elif raw_event["team"]["id"] == away_lineup["team_id"]:
team = teams[1]
else:
raise Exception(
f"Unknown team_id {raw_event['team']['id']}"
)
if (
raw_event["possession_team"]["id"]
== home_lineup["team_id"]
):
possession_team = teams[0]
elif (
raw_event["possession_team"]["id"]
== away_lineup["team_id"]
):
possession_team = teams[1]
else:
raise Exception(
f"Unknown possession_team_id: {raw_event['possession_team']}"
)
timestamp = parse_str_ts(raw_event["timestamp"])
period_id = int(raw_event["period"])
if not period or period.id != period_id:
period = Period(
id=period_id,
start_timestamp=(
timestamp
if not period
# period = [start, end], add millisecond to prevent overlapping
else timestamp + period.end_timestamp + 0.001
),
end_timestamp=None,
)
periods.append(period)
else:
period.end_timestamp = period.start_timestamp + timestamp
player = None
if "player" in raw_event:
player = team.get_player_by_id(raw_event["player"]["id"])
event_type = raw_event["type"]["id"]
if event_type == SB_EVENT_TYPE_SHOT:
fidelity_version = shot_fidelity_version
elif event_type in (
SB_EVENT_TYPE_CARRY,
SB_EVENT_TYPE_DRIBBLE,
SB_EVENT_TYPE_PASS,
):
fidelity_version = xy_fidelity_version
else:
# TODO: Uh ohhhh.. don't know which one to pick
fidelity_version = xy_fidelity_version
generic_event_kwargs = dict(
# from DataRecord
period=period,
timestamp=timestamp,
ball_owning_team=possession_team,
ball_state=BallState.ALIVE,
# from Event
event_id=raw_event["id"],
team=team,
player=player,
coordinates=(
_parse_coordinates(
raw_event.get("location"), fidelity_version
)
if "location" in raw_event
else None
),
raw_event=raw_event,
)
if event_type == SB_EVENT_TYPE_PASS:
pass_event_kwargs = _parse_pass(
pass_dict=raw_event["pass"],
team=team,
fidelity_version=fidelity_version,
)
event = PassEvent(
# TODO: Consider moving this to _parse_pass
receive_timestamp=timestamp + raw_event["duration"],
**pass_event_kwargs,
**generic_event_kwargs,
)
elif event_type == SB_EVENT_TYPE_SHOT:
shot_event_kwargs = _parse_shot(
shot_dict=raw_event["shot"]
)
event = ShotEvent(
**shot_event_kwargs, **generic_event_kwargs
)
# For dribble and carry the definitions
# are flipped between Statsbomb and kloppy
elif event_type == SB_EVENT_TYPE_DRIBBLE:
take_on_event_kwargs = _parse_take_on(
take_on_dict=raw_event["dribble"]
)
event = TakeOnEvent(
**take_on_event_kwargs, **generic_event_kwargs
)
elif event_type == SB_EVENT_TYPE_CARRY:
carry_event_kwargs = _parse_carry(
carry_dict=raw_event["carry"],
fidelity_version=fidelity_version,
)
event = CarryEvent(
# TODO: Consider moving this to _parse_carry
end_timestamp=timestamp + raw_event["duration"],
**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)
metadata = Metadata(
teams=teams,
periods=periods,
pitch_dimensions=PitchDimensions(
x_dim=Dimension(0, 120), y_dim=Dimension(0, 80)
),
frame_rate=None,
orientation=Orientation.ACTION_EXECUTING_TEAM,
flags=DatasetFlag.BALL_OWNING_TEAM,
score=None,
)
return EventDataset(metadata=metadata, records=events,)
def deserialize(self,
inputs: Dict[str, Readable],
options: Dict = None) -> EventDataSet:
self.__validate_inputs(inputs)
periods = []
period = None
events = []
game_state = self.__GameState(ball_state=BallState.DEAD,
ball_owning_team=None)
reader = csv.DictReader(
map(lambda x: x.decode('utf-8'), inputs['raw_data']))
for event_id, record in enumerate(reader):
event_type = event_type_map[record['Type']]
subtypes = record['Subtype'].split('-')
start_timestamp = float(record['Start Time [s]'])
end_timestamp = float(record['End Time [s]'])
period_id = int(record['Period'])
if not period or period.id != period_id:
period = Period(id=period_id,
start_timestamp=start_timestamp,
end_timestamp=end_timestamp)
periods.append(period)
else:
period.end_timestamp = end_timestamp
if record['Team'] == 'Home':
team = Team.HOME
elif record['Team'] == 'Away':
team = Team.AWAY
else:
raise ValueError(f'Unknown team: {record["team"]}')
event_kwargs = dict(
# From DataRecord:
timestamp=start_timestamp,
ball_owning_team=None, ## todo
ball_state=None, # todo
period=period,
# From Event:
event_id=event_id,
team=team,
end_timestamp=end_timestamp,
player_jersey_no=record['From'][6:],
position=Point(x=float(record['Start X']),
y=1 - float(record['Start Y']))
if record['Start X'] != 'NaN' else None,
)
secondary_position = None
if record['End X'] != 'NaN':
secondary_position = Point(x=float(record['End X']),
y=1 - float(record['End Y']))
secondary_jersey_no = None
if record['To']:
secondary_jersey_no = record['To'][6:]
event = None
if event_type == EventType.SET_PIECE:
set_piece, fk_attempt, retaken = \
build_subtypes(subtypes, [SetPiece, FKAttempt, Retaken])
event = SetPieceEvent(**event_kwargs)
elif event_type == EventType.RECOVERY:
interference1, interference2 = \
build_subtypes(subtypes, [Interference1, Interference2])
event = RecoveryEvent(**event_kwargs)
elif event_type == EventType.PASS:
body_part, attempt, deflection, offside = \
build_subtypes(subtypes, [BodyPart, Attempt, Deflection, Offside])
event = PassEvent(
receiver_position=secondary_position,
receiver_player_jersey_no=secondary_jersey_no,
**event_kwargs)
elif event_type == EventType.BALL_LOST:
body_part, attempt, interference1, intervention, deflection, offside = \
build_subtypes(subtypes, [
BodyPart, Attempt, Interference1, Intervention,
Deflection, Offside
])
event = BallLossEvent(**event_kwargs)
elif event_type == EventType.BALL_OUT:
body_part, attempt, intervention, deflection, offside, own_goal = \
build_subtypes(subtypes, [
BodyPart, Attempt, Intervention, Deflection,
Offside, OwnGoal
])
event = BallOutEvent(**event_kwargs)
elif event_type == EventType.SHOT:
body_part, deflection, shot_direction, shot_result, offside = \
build_subtypes(subtypes, [
BodyPart, Deflection, ShotDirection,
ShotResult, Offside
])
event = ShotEvent(shot_result=shot_result, **event_kwargs)
elif event_type == EventType.FAULT_RECEIVED:
event = FaultReceivedEvent(**event_kwargs)
elif event_type == EventType.CHALLENGE:
challenge, fault, challenge_result = \
build_subtypes(subtypes, [Challenge, Fault, ChallengeResult])
event = ChallengeEvent(**event_kwargs)
elif event_type == EventType.CARD:
card, = build_subtypes(subtypes, [Card])
event = CardEvent(**event_kwargs)
else:
raise NotImplementedError(
f"EventType {event_type} not implemented")
# We want to attach the game_state after the event to the event
game_state = self.__reduce_game_state(event=event,
game_state=game_state)
event.ball_state = game_state.ball_state
event.ball_owning_team = game_state.ball_owning_team
events.append(event)
orientation = (
Orientation.FIXED_HOME_AWAY if periods[0].attacking_direction
== AttackingDirection.HOME_AWAY else Orientation.FIXED_AWAY_HOME)
return EventDataSet(
flags=DataSetFlag.BALL_STATE | DataSetFlag.BALL_OWNING_TEAM,
orientation=orientation,
pitch_dimensions=PitchDimensions(x_dim=Dimension(0, 1),
y_dim=Dimension(0, 1)),
periods=periods,
records=events)
def deserialize(self,
inputs: Dict[str, Readable],
options: Dict = None) -> EventDataset:
"""
Deserialize Opta event data into a `EventDataset`.
Parameters
----------
inputs : dict
input `f24_data` should point to a `Readable` object containing
the 'xml' formatted event data. input `f7_data` should point
to a `Readable` object containing the 'xml' formatted f7 data.
options : dict
Options for deserialization of the Opta 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 = OptaSerializer()
>>> with open("123_f24.xml", "rb") as f24_data, \
>>> open("123_f7.xml", "rb") as f7_data:
>>>
>>> dataset = serializer.deserialize(
>>> inputs={
>>> 'f24_data': f24_data,
>>> 'f7_data': f7_data
>>> },
>>> options={
>>> 'event_types': ["pass", "take_on", "carry", "shot"]
>>> }
>>> )
"""
self.__validate_inputs(inputs)
if not options:
options = {}
with performance_logging("load data", logger=logger):
f7_root = objectify.fromstring(inputs["f7_data"].read())
f24_root = objectify.fromstring(inputs["f24_data"].read())
wanted_event_types = [
EventType[event_type.upper()]
for event_type in options.get("event_types", [])
]
with performance_logging("parse data", logger=logger):
matchdata_path = objectify.ObjectPath(
"SoccerFeed.SoccerDocument.MatchData")
team_elms = list(
matchdata_path.find(f7_root).iterchildren("TeamData"))
away_player_map = {}
home_player_map = {}
home_team_id = None
away_team_id = None
for team_elm in team_elms:
player_map = {
player_elm.attrib["PlayerRef"].lstrip("p"):
player_elm.attrib["ShirtNumber"]
for player_elm in team_elm.find(
"PlayerLineUp").iterchildren("MatchPlayer")
}
team_id = team_elm.attrib["TeamRef"].lstrip("t")
if team_elm.attrib["Side"] == "Home":
home_player_map = player_map
home_team_id = team_id
elif team_elm.attrib["Side"] == "Away":
away_player_map = player_map
away_team_id = team_id
else:
raise Exception(f"Unknown side: {team_elm.attrib['Side']}")
if not away_player_map or not home_player_map:
raise Exception("LineUp incomplete")
game_elm = f24_root.find("Game")
periods = [
Period(
id=1,
start_timestamp=None,
end_timestamp=None,
),
Period(
id=2,
start_timestamp=None,
end_timestamp=None,
),
]
events = []
for event_elm in game_elm.iterchildren("Event"):
event_id = event_elm.attrib["id"]
type_id = int(event_elm.attrib["type_id"])
timestamp = _parse_f24_datetime(event_elm.attrib["timestamp"])
period_id = int(event_elm.attrib["period_id"])
for period in periods:
if period.id == period_id:
break
else:
logger.debug(
f"Skipping event {event_id} because period doesn't match {period_id}"
)
continue
if type_id == EVENT_TYPE_START_PERIOD:
logger.debug(
f"Set start of period {period.id} to {timestamp}")
period.start_timestamp = timestamp
elif type_id == EVENT_TYPE_END_PERIOD:
logger.debug(
f"Set end of period {period.id} to {timestamp}")
period.end_timestamp = timestamp
else:
if not period.start_timestamp:
# not started yet
continue
if event_elm.attrib["team_id"] == home_team_id:
team = Team.HOME
current_team_map = home_player_map
elif event_elm.attrib["team_id"] == away_team_id:
team = Team.AWAY
current_team_map = away_player_map
else:
raise Exception(
f"Unknown team_id {event_elm.attrib['team_id']}")
x = float(event_elm.attrib["x"])
y = float(event_elm.attrib["y"])
outcome = int(event_elm.attrib["outcome"])
qualifiers = {
int(qualifier_elm.attrib["qualifier_id"]):
qualifier_elm.attrib.get("value")
for qualifier_elm in event_elm.iterchildren("Q")
}
player_jersey_no = None
if "player_id" in event_elm.attrib:
player_jersey_no = current_team_map[
event_elm.attrib["player_id"]]
generic_event_kwargs = dict(
# from DataRecord
period=period,
timestamp=timestamp - period.start_timestamp,
ball_owning_team=None,
ball_state=BallState.ALIVE,
# from Event
event_id=event_id,
team=team,
player_jersey_no=player_jersey_no,
position=Point(x=x, y=y),
raw_event=event_elm,
)
if type_id == EVENT_TYPE_PASS:
pass_event_kwargs = _parse_pass(qualifiers, outcome)
event = PassEvent(
**pass_event_kwargs,
**generic_event_kwargs,
)
elif type_id == EVENT_TYPE_OFFSIDE_PASS:
pass_event_kwargs = _parse_offside_pass()
event = PassEvent(
**pass_event_kwargs,
**generic_event_kwargs,
)
elif type_id == EVENT_TYPE_TAKE_ON:
take_on_event_kwargs = _parse_take_on(outcome)
event = TakeOnEvent(
**take_on_event_kwargs,
**generic_event_kwargs,
)
elif type_id in (
EVENT_TYPE_SHOT_MISS,
EVENT_TYPE_SHOT_POST,
EVENT_TYPE_SHOT_SAVED,
EVENT_TYPE_SHOT_GOAL,
):
shot_event_kwargs = _parse_shot(
qualifiers,
type_id,
position=generic_event_kwargs["position"],
)
kwargs = {}
kwargs.update(generic_event_kwargs)
kwargs.update(shot_event_kwargs)
event = ShotEvent(**kwargs)
else:
event = GenericEvent(**generic_event_kwargs,
result=None)
if (not wanted_event_types
or event.event_type in wanted_event_types):
events.append(event)
return EventDataset(
flags=DatasetFlag.BALL_OWNING_TEAM,
orientation=Orientation.ACTION_EXECUTING_TEAM,
pitch_dimensions=PitchDimensions(x_dim=Dimension(0, 100),
y_dim=Dimension(0, 100)),
periods=periods,
records=events,
)
def deserialize(self,
inputs: Dict[str, Readable],
options: Dict = None) -> EventDataset:
self.__validate_inputs(inputs)
if not options:
options = {}
with performance_logging("load data", logger=logger):
match_root = objectify.fromstring(inputs["match_data"].read())
event_root = objectify.fromstring(inputs["event_data"].read())
wanted_event_types = [
EventType[event_type.upper()]
for event_type in options.get("event_types", [])
]
with performance_logging("parse data", logger=logger):
x_max = float(
match_root.MatchInformation.Environment.attrib["PitchX"])
y_max = float(
match_root.MatchInformation.Environment.attrib["PitchY"])
team_path = objectify.ObjectPath(
"PutDataRequest.MatchInformation.Teams")
team_elms = list(team_path.find(match_root).iterchildren("Team"))
for team_elm in team_elms:
if team_elm.attrib["Role"] == "home":
home_team = _team_from_xml_elm(team_elm)
elif team_elm.attrib["Role"] == "guest":
away_team = _team_from_xml_elm(team_elm)
else:
raise Exception(f"Unknown side: {team_elm.attrib['Role']}")
(
home_score,
away_score,
) = match_root.MatchInformation.General.attrib["Result"].split(":")
score = Score(home=int(home_score), away=int(away_score))
teams = [home_team, away_team]
if len(home_team.players) == 0 or len(away_team.players) == 0:
raise Exception("LineUp incomplete")
periods = []
period_id = 0
events = []
for event_elm in event_root.iterchildren("Event"):
event_chain = _event_chain_from_xml_elm(event_elm)
timestamp = _parse_datetime(event_chain["Event"]["EventTime"])
if (SPORTEC_EVENT_NAME_KICKOFF in event_chain and "GameSection"
in event_chain[SPORTEC_EVENT_NAME_KICKOFF]):
period_id += 1
period = Period(
id=period_id,
start_timestamp=timestamp,
end_timestamp=None,
)
if period_id == 1:
team_left = event_chain[SPORTEC_EVENT_NAME_KICKOFF][
"TeamLeft"]
if team_left == home_team.team_id:
# goal of home team is on the left side.
# this means they attack from left to right
orientation = Orientation.FIXED_HOME_AWAY
period.set_attacking_direction(
AttackingDirection.HOME_AWAY)
else:
orientation = Orientation.FIXED_AWAY_HOME
period.set_attacking_direction(
AttackingDirection.AWAY_HOME)
else:
last_period = periods[-1]
period.set_attacking_direction(
AttackingDirection.AWAY_HOME if last_period.
attacking_direction == AttackingDirection.
HOME_AWAY else AttackingDirection.HOME_AWAY)
periods.append(period)
elif SPORTEC_EVENT_NAME_FINAL_WHISTLE in event_chain:
period.end_timestamp = timestamp
continue
team = None
player = None
flatten_attributes = dict()
# reverse because top levels are more important
for event_attributes in reversed(event_chain.values()):
flatten_attributes.update(event_attributes)
if "Team" in flatten_attributes:
team = (home_team if flatten_attributes["Team"]
== home_team.team_id else away_team)
if "Player" in flatten_attributes:
if not team:
raise ValueError("Player set while team is not set")
player = team.get_player_by_id(
flatten_attributes["Player"])
generic_event_kwargs = dict(
# from DataRecord
period=period,
timestamp=timestamp - period.start_timestamp,
ball_owning_team=None,
ball_state=BallState.ALIVE,
# from Event
event_id=event_chain["Event"]["EventId"],
coordinates=_parse_coordinates(event_chain["Event"]),
raw_event=flatten_attributes,
team=team,
player=player,
)
event_name, event_attributes = event_chain.popitem()
if event_name in SPORTEC_SHOT_EVENT_NAMES:
shot_event_kwargs = _parse_shot(event_name=event_name,
event_chain=event_chain)
event = ShotEvent.create(
**shot_event_kwargs,
**generic_event_kwargs,
)
elif event_name in SPORTEC_PASS_EVENT_NAMES:
pass_event_kwargs = _parse_pass(event_chain=event_chain,
team=team)
event = PassEvent.create(
**pass_event_kwargs,
**generic_event_kwargs,
receive_timestamp=None,
receiver_coordinates=None,
)
elif event_name == SPORTEC_EVENT_NAME_BALL_CLAIMING:
event = RecoveryEvent.create(
result=None,
qualifiers=None,
**generic_event_kwargs,
)
elif event_name == SPORTEC_EVENT_NAME_SUBSTITUTION:
substitution_event_kwargs = _parse_substitution(
event_attributes=event_attributes, team=team)
generic_event_kwargs["player"] = substitution_event_kwargs[
"player"]
del substitution_event_kwargs["player"]
event = SubstitutionEvent.create(
result=None,
qualifiers=None,
**substitution_event_kwargs,
**generic_event_kwargs,
)
elif event_name == SPORTEC_EVENT_NAME_CAUTION:
card_kwargs = _parse_caution(event_attributes)
event = CardEvent.create(
result=None,
qualifiers=None,
**card_kwargs,
**generic_event_kwargs,
)
elif event_name == SPORTEC_EVENT_NAME_FOUL:
foul_kwargs = _parse_foul(event_attributes, teams=teams)
generic_event_kwargs.update(foul_kwargs)
event = FoulCommittedEvent.create(
result=None,
qualifiers=None,
**generic_event_kwargs,
)
else:
event = GenericEvent.create(
result=None,
qualifiers=None,
event_name=event_name,
**generic_event_kwargs,
)
if events:
previous_event = events[-1]
if (previous_event.event_type == EventType.PASS
and previous_event.result == PassResult.COMPLETE):
if "X-Source-Position" in event_chain["Event"]:
previous_event.receiver_coordinates = Point(
x=float(
event_chain["Event"]["X-Source-Position"]),
y=float(
event_chain["Event"]["Y-Source-Position"]),
)
if (event.event_type == EventType.PASS
and event.get_qualifier_value(SetPieceQualifier) in (
SetPieceType.THROW_IN,
SetPieceType.GOAL_KICK,
SetPieceType.CORNER_KICK,
)):
# 1. update previous pass
if events[-1].event_type == EventType.PASS:
events[-1].result = PassResult.OUT
# 2. add synthetic out event
decision_timestamp = _parse_datetime(event_chain[list(
event_chain.keys())[1]]["DecisionTimestamp"])
out_event = BallOutEvent.create(
period=period,
timestamp=decision_timestamp - period.start_timestamp,
ball_owning_team=None,
ball_state=BallState.DEAD,
# from Event
event_id=event_chain["Event"]["EventId"] + "-ball-out",
team=events[-1].team,
player=events[-1].player,
coordinates=None,
raw_event={},
result=None,
qualifiers=None,
)
events.append(out_event)
events.append(event)
events = list(
filter(
lambda _event: _include_event(_event, wanted_event_types),
events,
))
metadata = Metadata(
teams=teams,
periods=periods,
pitch_dimensions=PitchDimensions(x_dim=Dimension(0, x_max),
y_dim=Dimension(0, y_max)),
score=score,
frame_rate=None,
orientation=orientation,
flags=~(DatasetFlag.BALL_STATE | DatasetFlag.BALL_OWNING_TEAM),
provider=Provider.SPORTEC,
)
return EventDataset(
metadata=metadata,
records=events,
)