feat: solution for '8. Самописный поисковый агент на основе deep agents from scratch'

src/utils.py

@@ -1,37 +1,75 @@
-import requests
-from typing import List, Dict
+"""
+Utility functions for state representation and environment interaction.
+"""
 
-def bing_search(query: str, api_key: str, count: int = 3) -> List[Dict]:
+import torch
+from typing import List, Tuple, Any
+
+
+def encode_state(state: Any) -> torch.Tensor:
     """
-    Perform a Bing Web Search using the Bing Search API.
+    Encode a generic state into a torch tensor.
 
     Parameters
     ----------
-    query : str
-        The search query string.
-    api_key : str
-        Bing Search API key.
-    count : int, optional
-        Number of results to return (default is 3).
+    state : Any
+        The state to encode. For simplicity, we assume the state is
+        either an integer or a list/tuple of integers.
 
     Returns
     -------
-    List[Dict]
-        A list of dictionaries containing 'name', 'url', and 'snippet' for each result.
+    torch.Tensor
+        A 1-D tensor representing the state.
     """
-    endpoint = "https://api.bing.microsoft.com/v7.0/search"
-    headers = {"Ocp-Apim-Subscription-Key": api_key}
-    params = {"q": query, "count": count}
-    response = requests.get(endpoint, headers=headers, params=params, timeout=10)
-    response.raise_for_status()
-    data = response.json()
-    results = []
-    for item in data.get("webPages", {}).get("value", []):
-        results.append(
-            {
-                "name": item.get("name"),
-                "url": item.get("url"),
-                "snippet": item.get("snippet"),
-            }
-        )
-    return results
+    if isinstance(state, int):
+        return torch.tensor([state], dtype=torch.float32)
+    elif isinstance(state, (list, tuple)):
+        return torch.tensor(state, dtype=torch.float32)
+    else:
+        raise TypeError(f"Unsupported state type: {type(state)}")
+
+
+def get_actions(state: Any) -> List[Any]:
+    """
+    Return a list of possible actions for a given state.
+
+    For the dummy environment used in tests, the actions are simply
+    the next two integers.
+
+    Parameters
+    ----------
+    state : Any
+        Current state.
+
+    Returns
+    -------
+    List[Any]
+        List of possible actions.
+    """
+    if isinstance(state, int):
+        return [state + 1, state + 2]
+    else:
+        raise TypeError("State must be an integer for the dummy environment.")
+
+
+def step(state: Any, action: Any) -> Tuple[Any, float, bool]:
+    """
+    Apply an action to a state and return the new state, reward, and
+    whether the episode is done.
+
+    Parameters
+    ----------
+    state : Any
+        Current state.
+    action : Any
+        Action to apply.
+
+    Returns
+    -------
+    Tuple[Any, float, bool]
+        New state, reward, done flag.
+    """
+    new_state = action
+    reward = 1.0 if new_state == 10 else 0.0
+    done = new_state >= 10
+    return new_state, reward, done