Update app.py

app.py

@@ -272,6 +272,31 @@ def build_rag_prompt(query: str, contexts: list[dict]) -> str:
     prompt = RAG_PROMPT_TEMPLATE.format(context=context_text, question=query)
     return prompt
 
+def _call_model(cfg: dict, prompt: str) -> dict:
+    return call_llm_with_metrics(
+        prompt,
+        endpoint_url=cfg["endpoint_url"],
+        api_key=AZURE_API_KEY,
+        model=cfg["model"],
+        max_completion_tokens=cfg["max_completion_tokens"],
+        temperature=cfg["temperature"],
+        top_p=cfg["top_p"],
+    )
+
+
+def _parse_llm_json(raw_content: str):
+    json_str = raw_content.strip()
+    if json_str.startswith("```"):
+        json_str = json_str.split("\n", 1)[-1]
+        json_str = json_str.rsplit("```", 1)[0].strip()
+    try:
+        parsed = json.loads(json_str)
+        answer = parsed["answer"]
+        explanation = parsed.get("explanation", "")
+        confident = bool(parsed.get("confident", True))  # absent => on suppose confiant (pas d'escalade)
+        return answer, explanation, confident
+    except (json.JSONDecodeError, KeyError):
+        return raw_content, "LLM did not return a structured explanation.", True
 
 def rag_query(query: str, top_k: int = TOP_K_RESULTS) -> dict:
     start_time = time.perf_counter()