def launch_executable(file_name: str, args: List[str]) -> subprocess.Popen:
"""
Launches a Unity executable and returns the process handle for it.
:param file_name: the name of the executable
:param args: List of string that will be passed as command line arguments
when launching the executable.
"""
launch_string = validate_environment_path(file_name)
if launch_string is None:
raise UnityEnvironmentException(
f"Couldn't launch the {file_name} environment. Provided filename does not match any environments."
)
else:
get_logger(__name__).debug(
f"This is the launch string {launch_string}")
# Launch Unity environment
subprocess_args = [launch_string] + args
try:
return subprocess.Popen(
subprocess_args,
# start_new_session=True means that signals to the parent python process
# (e.g. SIGINT from keyboard interrupt) will not be sent to the new process on POSIX platforms.
# This is generally good since we want the environment to have a chance to shutdown,
# but may be undesirable in come cases; if so, we'll add a command-line toggle.
# Note that on Windows, the CTRL_C signal will still be sent.
start_new_session=True,
)
except PermissionError as perm:
# This is likely due to missing read or execute permissions on file.
raise UnityEnvironmentException(
f"Error when trying to launch environment - make sure "
f"permissions are set correctly. For example "
f'"chmod -R 755 {launch_string}"') from perm
def process_side_channel_message(self, data: bytes) -> None:
"""
Separates the data received from Python into individual messages for each
registered side channel and calls on_message_received on them.
:param data: The packed message sent by Unity
"""
offset = 0
while offset < len(data):
try:
channel_id = uuid.UUID(bytes_le=bytes(data[offset:offset +
16]))
offset += 16
message_len, = struct.unpack_from("<i", data, offset)
offset = offset + 4
message_data = data[offset:offset + message_len]
offset = offset + message_len
except (struct.error, ValueError, IndexError):
raise UnityEnvironmentException(
"There was a problem reading a message in a SideChannel. "
"Please make sure the version of MLAgents in Unity is "
"compatible with the Python version.")
if len(message_data) != message_len:
raise UnityEnvironmentException(
"The message received by the side channel {0} was "
"unexpectedly short. Make sure your Unity Environment "
"sending side channel data properly.".format(channel_id))
if channel_id in self._side_channels_dict:
incoming_message = IncomingMessage(message_data)
self._side_channels_dict[channel_id].on_message_received(
incoming_message)
else:
get_logger(__name__).warning(
f"Unknown side channel data received. Channel type: {channel_id}."
)
def validate_environment_path(env_path: str) -> Optional[str]:
"""
Strip out executable extensions of the env_path
:param env_path: The path to the executable
"""
env_path = (env_path.strip().replace(".app", "").replace(
".exe", "").replace(".x86_64", "").replace(".x86", ""))
true_filename = os.path.basename(os.path.normpath(env_path))
get_logger(__name__).debug(f"The true file name is {true_filename}")
if not (glob.glob(env_path) or glob.glob(env_path + ".*")):
return None
cwd = os.getcwd()
launch_string = None
true_filename = os.path.basename(os.path.normpath(env_path))
if get_platform() == "linux" or get_platform() == "linux2":
candidates = glob.glob(os.path.join(cwd, env_path) + ".x86_64")
if len(candidates) == 0:
candidates = glob.glob(os.path.join(cwd, env_path) + ".x86")
if len(candidates) == 0:
candidates = glob.glob(env_path + ".x86_64")
if len(candidates) == 0:
candidates = glob.glob(env_path + ".x86")
if len(candidates) > 0:
launch_string = candidates[0]
elif get_platform() == "darwin":
candidates = glob.glob(
os.path.join(cwd, env_path + ".app", "Contents", "MacOS",
true_filename))
if len(candidates) == 0:
candidates = glob.glob(
os.path.join(env_path + ".app", "Contents", "MacOS",
true_filename))
if len(candidates) == 0:
candidates = glob.glob(
os.path.join(cwd, env_path + ".app", "Contents", "MacOS", "*"))
if len(candidates) == 0:
candidates = glob.glob(
os.path.join(env_path + ".app", "Contents", "MacOS", "*"))
if len(candidates) > 0:
launch_string = candidates[0]
elif get_platform() == "win32":
candidates = glob.glob(os.path.join(cwd, env_path + ".exe"))
if len(candidates) == 0:
candidates = glob.glob(env_path + ".exe")
if len(candidates) == 0:
# Look for e.g. 3DBall\UnityEnvironment.exe
crash_handlers = set(
glob.glob(os.path.join(cwd, env_path,
"UnityCrashHandler*.exe")))
candidates = [
c for c in glob.glob(os.path.join(cwd, env_path, "*.exe"))
if c not in crash_handlers
]
if len(candidates) > 0:
launch_string = candidates[0]
return launch_string
def validate_environment_path(env_path: str) -> Optional[str]:
"""
Strip out executable extensions of the env_path
:param env_path: The path to the executable
"""
env_path = (
env_path.strip()
.replace(".app", "")
.replace(".exe", "")
.replace(".x86_64", "")
.replace(".x86", "")
)
true_filename = os.path.basename(os.path.normpath(env_path))
get_logger(__name__).debug("The true file name is {}".format(true_filename))
if not (glob.glob(env_path) or glob.glob(env_path + ".*")):
return None
cwd = os.getcwd()
launch_string = None
true_filename = os.path.basename(os.path.normpath(env_path))
if platform == "linux" or platform == "linux2":
candidates = glob.glob(os.path.join(cwd, env_path) + ".x86_64")
if len(candidates) == 0:
candidates = glob.glob(os.path.join(cwd, env_path) + ".x86")
if len(candidates) == 0:
candidates = glob.glob(env_path + ".x86_64")
if len(candidates) == 0:
candidates = glob.glob(env_path + ".x86")
if len(candidates) > 0:
launch_string = candidates[0]
elif platform == "darwin":
candidates = glob.glob(
os.path.join(cwd, env_path + ".app", "Contents", "MacOS", true_filename)
)
if len(candidates) == 0:
candidates = glob.glob(
os.path.join(env_path + ".app", "Contents", "MacOS", true_filename)
)
if len(candidates) == 0:
candidates = glob.glob(
os.path.join(cwd, env_path + ".app", "Contents", "MacOS", "*")
)
if len(candidates) == 0:
candidates = glob.glob(
os.path.join(env_path + ".app", "Contents", "MacOS", "*")
)
if len(candidates) > 0:
launch_string = candidates[0]
elif platform == "win32":
candidates = glob.glob(os.path.join(cwd, env_path + ".exe"))
if len(candidates) == 0:
candidates = glob.glob(env_path + ".exe")
if len(candidates) > 0:
launch_string = candidates[0]
return launch_string
def __init__(
self,
trainer_factory: TrainerFactory,
output_path: str,
run_id: str,
param_manager: EnvironmentParameterManager,
train: bool,
training_seed: int,
):
"""
:param output_path: Path to save the model.
:param summaries_dir: Folder to save training summaries.
:param run_id: The sub-directory name for model and summary statistics
:param param_manager: EnvironmentParameterManager object which stores information about all
environment parameters.
:param train: Whether to train model, or only run inference.
:param training_seed: Seed to use for Numpy and Tensorflow random number generation.
:param threaded: Whether or not to run trainers in a separate thread. Disable for testing/debugging.
"""
self.trainers: Dict[str, Trainer] = {}
self.brain_name_to_identifier: Dict[str, Set] = defaultdict(set)
self.trainer_factory = trainer_factory
self.output_path = output_path
self.logger = get_logger(__name__)
self.run_id = run_id
self.train_model = train
self.param_manager = param_manager
self.ghost_controller = self.trainer_factory.ghost_controller
self.registered_behavior_ids: Set[str] = set()
self.trainer_threads: List[threading.Thread] = []
self.kill_trainers = False
np.random.seed(training_seed)
tf.set_random_seed(training_seed)
def __init__(self, trainer_factory: TrainerFactory, output_path: str,
run_id: str, meta_curriculum: Optional[MetaCurriculum],
train: bool, training_seed: int,
sampler_manager: SamplerManager,
resampling_interval: Optional[int]):
"""
:param output_path: Path to save the model.
:param summaries_dir: Folder to save training summaries.
:param run_id: The sub-directory name for model and summary statistics
:param meta_curriculum: MetaCurriculum object which stores information about all curricula.
:param train: Whether to train model, or only run inference.
:param training_seed: Seed to use for Numpy and Tensorflow random number generation.
:param sampler_manager: SamplerManager object handles samplers for resampling the reset parameters.
:param resampling_interval: Specifies number of simulation steps after which reset parameters are resampled.
:param threaded: Whether or not to run trainers in a separate thread. Disable for testing/debugging.
"""
self.trainers: Dict[str, Trainer] = {}
self.brain_name_to_identifier: Dict[str, Set] = defaultdict(set)
self.trainer_factory = trainer_factory
self.output_path = output_path
self.logger = get_logger(__name__)
self.run_id = run_id
self.train_model = train
self.meta_curriculum = meta_curriculum
self.sampler_manager = sampler_manager
self.resampling_interval = resampling_interval
self.ghost_controller = self.trainer_factory.ghost_controller
self.trainer_threads: List[threading.Thread] = []
self.kill_trainers = False
np.random.seed(training_seed)
tf.set_random_seed(training_seed)
def __init__(
self,
trainer_factory: TrainerFactory,
model_path: str,
summaries_dir: str,
run_id: str,
save_freq: int,
meta_curriculum: Optional[MetaCurriculum],
train: bool,
training_seed: int,
sampler_manager: SamplerManager,
resampling_interval: Optional[int],
):
"""
:param model_path: Path to save the model.
:param summaries_dir: Folder to save training summaries.
:param run_id: The sub-directory name for model and summary statistics
:param save_freq: Frequency at which to save model
:param meta_curriculum: MetaCurriculum object which stores information about all curricula.
:param train: Whether to train model, or only run inference.
:param training_seed: Seed to use for Numpy and Tensorflow random number generation.
:param sampler_manager: SamplerManager object handles samplers for resampling the reset parameters.
:param resampling_interval: Specifies number of simulation steps after which reset parameters are resampled.
"""
self.trainers: Dict[str, Trainer] = {}
self.brain_name_to_identifier: Dict[str, Set] = defaultdict(set)
self.trainer_factory = trainer_factory
self.model_path = model_path
self.summaries_dir = summaries_dir
self.logger = get_logger(__name__)
self.run_id = run_id
self.save_freq = save_freq
self.train_model = train
self.meta_curriculum = meta_curriculum
self.sampler_manager = sampler_manager
self.resampling_interval = resampling_interval
np.random.seed(training_seed)
tf.set_random_seed(training_seed)
from mlagents_envs.communicator_objects.unity_rl_input_pb2 import UnityRLInputProto
from mlagents_envs.communicator_objects.unity_rl_output_pb2 import UnityRLOutputProto
from mlagents_envs.communicator_objects.agent_action_pb2 import AgentActionProto
from mlagents_envs.communicator_objects.unity_output_pb2 import UnityOutputProto
from mlagents_envs.communicator_objects.capabilities_pb2 import UnityRLCapabilitiesProto
from mlagents_envs.communicator_objects.unity_rl_initialization_input_pb2 import (
UnityRLInitializationInputProto,
)
from mlagents_envs.communicator_objects.unity_input_pb2 import UnityInputProto
from .rpc_communicator import RpcCommunicator
import signal
logger = get_logger(__name__)
class UnityEnvironment(BaseEnv):
# Communication protocol version.
# When connecting to C#, this must be compatible with Academy.k_ApiVersion.
# We follow semantic versioning on the communication version, so existing
# functionality will work as long the major versions match.
# This should be changed whenever a change is made to the communication protocol.
API_VERSION = "1.0.0"
# Default port that the editor listens on. If an environment executable
# isn't specified, this port will be used.
DEFAULT_EDITOR_PORT = 5004
# Default base port for environments. Each environment will be offset from this
from typing import Set, Dict, Any, TextIO
import os
import yaml
from mlagents.trainers.exception import TrainerConfigError
from mlagents_envs.environment import UnityEnvironment
import argparse
from mlagents_envs import logging_util
logger = logging_util.get_logger(__name__)
class RaiseRemovedWarning(argparse.Action):
"""
Internal custom Action to raise warning when argument is called.
"""
def __init__(self, nargs=0, **kwargs):
super().__init__(nargs=nargs, **kwargs)
def __call__(self, arg_parser, namespace, values, option_string=None):
logger.warning(
f"The command line argument {option_string} was removed.")
class DetectDefault(argparse.Action):
"""
Internal custom Action to help detect arguments that aren't default.
"""
non_default_args: Set[str] = set()
def __call__(self, arg_parser, namespace, values, option_string=None):
def test_set_logging_level():
for level in [INFO, ERROR, FATAL, CRITICAL, DEBUG]:
set_log_level(level)
assert get_logger("test").level == level
