| import random |
| from typing import Dict, List |
|
|
| from .base_agent import BaseAgent |
|
|
| from autoresttest.graph.generate_graph import OperationGraph |
|
|
|
|
| class DataSourceAgent(BaseAgent): |
| def __init__( |
| self, |
| operation_graph: OperationGraph, |
| alpha: float = 0.1, |
| gamma: float = 0.9, |
| epsilon: float = 0.1, |
| ): |
| self.q_table: Dict[str, Dict[str, float]] = {} |
| self.operation_graph = operation_graph |
| self.alpha = alpha |
| self.gamma = gamma |
| self.epsilon = epsilon |
| self.available_data_sources: List[str] = ( |
| ["LLM", "DEPENDENCY", "DEFAULT"] |
| if operation_graph.operation_edges |
| else ["LLM", "DEFAULT"] |
| ) |
|
|
| def initialize_q_table(self) -> None: |
| for operation_id in self.operation_graph.operation_nodes.keys(): |
| if operation_id not in self.q_table: |
| self.q_table[operation_id] = { |
| data_source: 0 for data_source in self.available_data_sources |
| } |
|
|
| def initialize_dependency_source(self) -> None: |
| if "DEPENDENCY" not in self.available_data_sources: |
| for data_sources in self.q_table.values(): |
| data_sources["DEPENDENCY"] = 0 |
| self.available_data_sources.append("DEPENDENCY") |
|
|
| def get_action(self, operation_id: str) -> str: |
| if operation_id not in self.q_table: |
| raise ValueError(f"Operation '{operation_id}' not found in the Q-table for DataSourceAgent.") |
| if random.random() < self.epsilon: |
| return self.get_random_action(operation_id) |
| return self.get_best_action(operation_id) |
|
|
| def get_best_action(self, operation_id: str) -> str: |
| return max(self.q_table[operation_id].items(), key=lambda x: x[1])[0] |
|
|
| def get_random_action(self, operation_id: str) -> str: |
| return random.choice(list(self.q_table[operation_id].keys())) |
|
|
| def update_q_table(self, operation_id: str, action: str, reward: float) -> None: |
| if operation_id not in self.q_table: |
| return |
| if action not in self.q_table[operation_id]: |
| return |
| current_q = self.q_table[operation_id][action] |
| 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][action] = new_q |
|
|
| def get_Q_next(self, operation_id: str) -> float: |
| if operation_id not in self.q_table or not self.q_table[operation_id]: |
| return 0.0 |
| return max(self.q_table[operation_id].values()) |
|
|
| def get_Q_curr(self, operation_id: str, action: str) -> float: |
| if operation_id not in self.q_table: |
| return 0.0 |
| return self.q_table[operation_id].get(action, 0.0) |
|
|
| def update_Q_item(self, operation_id: str, action: str, td_error: float) -> None: |
| if operation_id not in self.q_table: |
| return |
| if action not in self.q_table[operation_id]: |
| return |
| self.q_table[operation_id][action] += self.alpha * td_error |
|
|
| def number_of_zeros(self, operation_id: str) -> int: |
| if operation_id not in self.q_table: |
| return 0 |
| zeros = 0 |
| for value in self.q_table[operation_id].values(): |
| if value == 0: |
| zeros += 1 |
| return zeros |
|
|
