from rl_coach.agents.clipped_ppo_agent import ClippedPPOAgentParameters
from rl_coach.architectures.layers import Dense
from rl_coach.base_parameters import PresetValidationParameters, VisualizationParameters
from rl_coach.core_types import (
    EnvironmentEpisodes,
    EnvironmentSteps,
    MaxDumpFilter,
    RunPhase,
    SelectedPhaseOnlyDumpFilter,
    TrainingSteps,
)
from rl_coach.environments.gym_environment import GymVectorEnvironment
from rl_coach.exploration_policies.e_greedy import EGreedyParameters
from rl_coach.filters.observation.observation_normalization_filter import (
    ObservationNormalizationFilter,
)
from rl_coach.graph_managers.basic_rl_graph_manager import BasicRLGraphManager
from rl_coach.graph_managers.graph_manager import ScheduleParameters
from rl_coach.schedules import LinearSchedule

####################
# Graph Scheduling #
####################

schedule_params = ScheduleParameters()
schedule_params.improve_steps = EnvironmentEpisodes(100)
schedule_params.steps_between_evaluation_periods = EnvironmentEpisodes(10)
schedule_params.evaluation_steps = EnvironmentEpisodes(1)
schedule_params.heatup_steps = EnvironmentEpisodes(10)

#########
# Agent #
#########
agent_params = ClippedPPOAgentParameters()

agent_params.network_wrappers["main"].learning_rate = 0.001
agent_params.network_wrappers["main"].input_embedders_parameters[
    "observation"
].activation_function = "tanh"
agent_params.network_wrappers["main"].input_embedders_parameters["observation"].scheme = [Dense(32)]
agent_params.network_wrappers["main"].middleware_parameters.scheme = [Dense(32)]
agent_params.network_wrappers["main"].middleware_parameters.activation_function = "tanh"
agent_params.network_wrappers["main"].batch_size = 256
agent_params.network_wrappers["main"].optimizer_epsilon = 1e-5
agent_params.network_wrappers["main"].adam_optimizer_beta2 = 0.999

agent_params.algorithm.clip_likelihood_ratio_using_epsilon = 0.3
agent_params.algorithm.clipping_decay_schedule = LinearSchedule(0.5, 0.1, 10000 * 50)
agent_params.algorithm.beta_entropy = 0
agent_params.algorithm.gae_lambda = 0.95
agent_params.algorithm.discount = 0.999
agent_params.algorithm.estimate_state_value_using_gae = True

agent_params.algorithm.num_steps_between_copying_online_weights_to_target = EnvironmentEpisodes(10)
agent_params.algorithm.num_episodes_in_experience_replay = 100
agent_params.algorithm.num_consecutive_playing_steps = EnvironmentEpisodes(10)
agent_params.algorithm.optimization_epochs = 10

agent_params.pre_network_filter.add_observation_filter(
    "observation",
    "normalize_observation",
    ObservationNormalizationFilter(name="normalize_observation"),
)

###############
# Environment #
###############
env_params = GymVectorEnvironment(level="patient_envs:PatientContinuousMountainCar")

#################
# Visualization #
#################
vis_params = VisualizationParameters()
vis_params.dump_gifs = True
vis_params.video_dump_filters = [SelectedPhaseOnlyDumpFilter(RunPhase.TEST), MaxDumpFilter()]

########
# Test #
########
preset_validation_params = PresetValidationParameters()
preset_validation_params.test = True
preset_validation_params.min_reward_threshold = 150
preset_validation_params.max_episodes_to_achieve_reward = 250

graph_manager = BasicRLGraphManager(
    agent_params=agent_params,
    env_params=env_params,
    schedule_params=schedule_params,
    vis_params=vis_params,
    preset_validation_params=preset_validation_params,
)