redis correction in question.py

routes/question.py

@@ -1,12 +1,14 @@
 import os
-import requests
-import datetime
 import json
+import datetime
+from typing import List, Dict
+
+import requests
 from fastapi import APIRouter
 from pydantic import BaseModel
-from clients.redis_client import redis_client as r
 from dotenv import load_dotenv
 
+from clients.redis_client import redis_client as _r
 from models_initialization.mistral_registry import mistral_generate
 from nuse_modules.classifier import classify_question, REVERSE_MAP
 from nuse_modules.keyword_extracter import keywords_extractor
@@ -16,63 +18,75 @@ load_dotenv()
 
 askMe = APIRouter()
 
+# ──────────────────────────────────────────────────────────────
+# Pydantic schema
+# ──────────────────────────────────────────────────────────────
 class QuestionInput(BaseModel):
     question: str
 
 
+# ──────────────────────────────────────────────────────────────
+# Helper functions
+# ──────────────────────────────────────────────────────────────
+
 def should_extract_keywords(type_id: int) -> bool:
+    """Map the intent id to whether we need keyword extraction."""
     return type_id in {1, 2, 3, 4, 5, 6, 7, 10, 11, 12}
 
 
 def extract_answer_after_label(text: str) -> str:
-    """
-    Extracts everything after the first 'Answer:' label.
-    Assumes 'Answer:' appears once and is followed by the relevant content.
-    """
+    """Extracts everything after the first 'Answer:' label."""
     if "Answer:" in text:
         return text.split("Answer:", 1)[1].strip()
     return text.strip()
 
 
+# ──────────────────────────────────────────────────────────────
+# FastAPI route
+# ──────────────────────────────────────────────────────────────
+
 @askMe.post("/ask")
 async def ask_question(input: QuestionInput):
-    question = input.question
+    question = input.question.strip()
 
-    # Step 1: Classify question intent
+    # 1️⃣ Classify intent
     qid = classify_question(question)
     print("Intent ID:", qid)
     print("Category:", REVERSE_MAP.get(qid, "unknown"))
 
-    context = ""
-    sources = []
-
+    # Special case: ID 13 → return cached headlines
     if qid == 13:
         date_str = datetime.datetime.utcnow().strftime("%Y-%m-%d")
         categories = ["world", "india", "finance", "sports", "entertainment"]
-        all_headlines = []
+        all_headlines: List[Dict] = []
 
         for cat in categories:
-            key = f"headlines:{date_str}:{cat}"
-            cached = _r.get(key)
+            redis_key = f"headlines:{date_str}:{cat}"
+            cached = _r.get(redis_key)
             if cached:
-                articles = json.loads(cached)
+                try:
+                    articles = json.loads(cached)
+                except json.JSONDecodeError:
+                    continue
                 for art in articles:
                     all_headlines.append({
-                        "title": art["title"],
-                        "summary": art["summary"],
-                        "url": art["url"],
-                        "image": art.get("image"),
+                        "title":   art.get("title"),
+                        "summary": art.get("summary"),
+                        "url":     art.get("url"),
+                        "image":   art.get("image"),
                         "category": cat,
                     })
 
         return {
             "question": question,
             "answer": "Here are today’s top headlines:",
-            "headlines": all_headlines  # include this for frontend/bot formatting
+            "headlines": all_headlines,
         }
-        
 
-    # Step 2: Keyword extraction and news search (if needed)
+    # 2️⃣ Keyword‑based flow for other intents
+    context = ""
+    sources: List[Dict] = []
+
     if should_extract_keywords(qid):
         keywords = keywords_extractor(question)
         print("Raw extracted keywords:", keywords)
@@ -80,60 +94,45 @@ async def ask_question(input: QuestionInput):
         if not keywords:
             return {"error": "Keyword extraction failed."}
 
-        # Search Google News
+        # Google News search
         results = search_google_news(keywords)
         print("Found articles:", results)
 
-        # for r in results:
-        #     print(r["title"], r["link"])
-
-        # Build context from snippet/description
         context = "\n\n".join([
-            r.get("snippet") or r.get("description", "")
-            for r in results
+            r.get("snippet") or r.get("description", "") for r in results
         ])[:15000]
 
-        sources = [
-            {"title": r["title"], "url": r["link"]}
-            for r in results
-        ]
+        sources = [{"title": r["title"], "url": r["link"]} for r in results]
 
         if not context.strip():
             return {
                 "question": question,
                 "answer": "Cannot answer – no relevant context found.",
-                "sources": sources
+                "sources": sources,
             }
-        
-         # Step 3: Ask Mistral to answer
+
         answer_prompt = (
-            f"You are a concise news assistant. Answer the user's question clearly using the context below if relevant. "
-            f"Make sure you are precise, accurate and to the point. Do not assume that the reader has any prerequisite understanding of the subject."
-            f"If the context is not helpful, you may rely on your own knowledge, but do not mention the context or question again.\n\n"
+            "You are a concise news assistant. Answer the user's question clearly using the provided context if relevant. "
+            "If the context is not helpful, rely on your own knowledge but do not mention the context.\n\n"
             f"Context:\n{context}\n\n"
-            f"Question: {question}\n\n"
-            f"Answer:"
+            f"Question: {question}\n\nAnswer:"
         )
         answer_raw = mistral_generate(answer_prompt, max_new_tokens=256)
-        
-    else:
 
+    else:
         answer_prompt = (
-            f"You are a concise news assistant. Answer the user's question clearly using the context below if relevant. "
-            f"Make sure you are precise, accurate and to the point. Do not assume that the reader has any prerequisite understanding of the subject."
-            f"Question: {question}\n\n"
-            f"Answer:"
+            "You are a concise news assistant. Answer the user's question clearly and accurately.\n\n"
+            f"Question: {question}\n\nAnswer:"
         )
-
         answer_raw = mistral_generate(answer_prompt, max_new_tokens=256)
 
-    if not answer_raw:
-        final_answer = "Cannot answer – model did not return a valid response."
-    else:
-        final_answer = extract_answer_after_label(answer_raw)
+    # 3️⃣ Post‑process model output
+    final_answer = extract_answer_after_label(answer_raw or "") or (
+        "Cannot answer – model did not return a valid response."
+    )
 
     return {
         "question": question,
         "answer": final_answer.strip(),
-        "sources": sources
+        "sources": sources,
     }