| import random | |
| from .base_agent import BaseAgent | |
| from autoresttest.graph import OperationGraph | |
| class OperationAgent(BaseAgent): | |
| def __init__( | |
| self, | |
| operation_graph: OperationGraph, | |
| alpha: float = 0.1, | |
| gamma: float = 0.9, | |
| epsilon: float = 0.1, | |
| ): | |
| self.alpha = alpha | |
| self.gamma = gamma | |
| self.epsilon = epsilon | |
| self.operation_graph = operation_graph | |
| self.q_table: dict[str, float] = {} | |
| def initialize_q_table(self) -> None: | |
| operation_ids = self.operation_graph.operation_nodes.keys() | |
| self.q_table = {operation_id: 0 for operation_id in operation_ids} | |
| def get_action(self) -> str: | |
| if random.random() < self.epsilon: | |
| return self.get_random_action() | |
| return self.get_best_action() | |
| def get_best_action(self) -> str: | |
| if not self.q_table: | |
| raise ValueError( | |
| "No operations were parsed from the specification for OperationAgent." | |
| ) | |
| return max(self.q_table.items(), key=lambda x: x[1])[0] | |
| def get_random_action(self) -> str: | |
| if not self.q_table: | |
| raise ValueError( | |
| "No operations were parsed from the specification for OperationAgent." | |
| ) | |
| return random.choice(list(self.q_table.keys())) | |
| def update_q_table(self, operation_id: str, reward: float) -> None: | |
| if operation_id not in self.q_table: | |
| return | |
| current_q = self.q_table[operation_id] | |
| best_next_q = self.get_Q_next(operation_id) | |
| new_q = current_q + self.alpha * (reward + self.gamma * best_next_q - current_q) | |
| self.q_table[operation_id] = new_q | |
| def get_Q_next(self, operation_id: str) -> float: | |
| return max(self.q_table.values()) if self.q_table else 0.0 | |
| def get_Q_curr(self, operation_id: str) -> float: | |
| return self.q_table.get(operation_id, 0.0) | |
| def update_Q_item(self, operation_id: str, td_error: float) -> None: | |
| if operation_id not in self.q_table: | |
| return | |
| self.q_table[operation_id] = ( | |
| self.q_table.get(operation_id, 0.0) + self.alpha * td_error | |
| ) | |