Update app.py

app.py

@@ -36,7 +36,7 @@ import requests
 from requests.adapters import HTTPAdapter, Retry
 from langchain.llms.base import LLM
 
-
+'''
 class GitHubModelLLM(LLM):
 
     def __init__(
@@ -212,6 +212,61 @@ class GitHubModelLLM(LLM):
         raw = self._post_chat(body)
         self._last_raw = raw
         return raw
+'''
+from typing import Optional, List, Dict, Any
+from langchain.llms.base import LLM
+import requests, os, json
+from requests.adapters import HTTPAdapter, Retry
+
+class GitHubModelLLM(LLM):
+    """GitHub Models API 기반 LangChain LLM (Pydantic 호환)"""
+    model: str = "openai/gpt-4.1"
+    endpoint: str = "https://models.github.ai/inference"
+    token: Optional[str] = os.environ["token"]
+    system_prompt: Optional[str] = "너는 PIXAL(Primary Interactive X-ternal Assistant with multi Language)이야.너의 개발자는 정성윤 이라는 6학년 파이썬 프로그래머야."
+    request_timeout: float = 30.0
+    max_retries: int = 2
+    backoff_factor: float = 0.3
+
+    @property
+    def _llm_type(self) -> str:
+        return "github_models_api"
+
+    def _post_chat(self, body: Dict[str, Any]) -> Dict[str, Any]:
+        token = self.token or os.getenv("GITHUB_TOKEN") or os.getenv("token")
+        if not token:
+            raise ValueError("❌ GitHub token이 설정되지 않았습니다.")
+
+        session = requests.Session()
+        retries = Retry(total=self.max_retries, backoff_factor=self.backoff_factor,
+                        status_forcelist=[429, 500, 502, 503, 504])
+        session.mount("https://", HTTPAdapter(max_retries=retries))
+        session.headers.update({
+            "Content-Type": "application/json",
+            "Authorization": f"Bearer {token}"
+        })
+        resp = session.post(f"{self.endpoint}/chat/completions", json=body, timeout=self.request_timeout)
+        resp.raise_for_status()
+        return resp.json()
+
+    def _call(self, prompt: str, stop: Optional[List[str]] = None, **kwargs) -> str:
+        body = {
+            "model": self.model,
+            "messages": []
+        }
+        if self.system_prompt:
+            body["messages"].append({"role": "system", "content": self.system_prompt})
+        body["messages"].append({"role": "user", "content": prompt})
+
+        for key in ["temperature", "max_tokens", "functions", "function_call"]:
+            if key in kwargs:
+                body[key] = kwargs[key]
+        if stop:
+            body["stop"] = stop
+
+        res = self._post_chat(body)
+        msg = res.get("choices", [{}])[0].get("message", {})
+        return msg.get("content") or json.dumps(msg.get("function_call", {}))
 
 from langchain_community.retrievers import WikipediaRetriever
 from langchain.tools.retriever import create_retriever_tool