Update chatbot.py

chatbot.py

@@ -1,7 +1,6 @@
 import os
 import pandas as pd
 import requests
-# import PyPDF2  # ❌ PDF no longer needed
 from sentence_transformers import SentenceTransformer
 from pinecone import Pinecone
 from openai import OpenAI
@@ -11,9 +10,6 @@ import re
 PINECONE_API_KEY = os.getenv("PINECONE_API_KEY")
 OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
 
-if not PINECONE_API_KEY or not OPENAI_API_KEY:
-    raise RuntimeError("❌ API keys not loaded. Check .env file")
-
 # ================= CONSTANTS =================
 SERVICES_INDEX_NAME = "aaple-sarkar-services"
 BASICS_INDEX_NAME = "aaplesarkarbasics"
@@ -22,15 +18,14 @@ BASE_DIR = os.path.dirname(__file__)
 DATA_DIR = os.path.join(BASE_DIR, "data")
 
 # ================= INIT =================
-pc = Pinecone(api_key=PINECONE_API_KEY)
-client = OpenAI(api_key=OPENAI_API_KEY)
+pc = Pinecone(api_key=PINECONE_API_KEY) if PINECONE_API_KEY else None
+client = OpenAI(api_key=OPENAI_API_KEY) if OPENAI_API_KEY else None
 embedder = SentenceTransformer("all-MiniLM-L6-v2")
 
-services_index = pc.Index(SERVICES_INDEX_NAME)
-basics_index = pc.Index(BASICS_INDEX_NAME)
+services_index = pc.Index(SERVICES_INDEX_NAME) if pc else None
+basics_index = pc.Index(BASICS_INDEX_NAME) if pc else None
 
 # ================= LOAD DATA =================
-
 EXCEL_URL = "https://huggingface.co/datasets/PrathameshRaut/aaple-sarkar-data/resolve/main/Untitled%20spreadsheet-2.xlsx"
 LOCAL_EXCEL_PATH = os.path.join(DATA_DIR, "services.xlsx")
 
@@ -47,26 +42,6 @@ excel_df = pd.read_excel(
     sheet_name="Copy of Notified services (1212"
 )
 
-
-# ================= PDF REMOVED =================
-# PDF knowledge base is already embedded & stored in Pinecone
-# Hence no need to load, parse, or pass PDF text again
-
-# kb_pdf_path = os.path.join(
-#     DATA_DIR,
-#     "/Users/prathameshraut/Desktop/Code/updated/backend/data/chatbot_knowledge_base.pdf"
-# )
-
-# def extract_pdf_text(path):
-#     text = ""
-#     with open(path, "rb") as f:
-#         reader = PyPDF2.PdfReader(f)
-#         for page in reader.pages:
-#             text += (page.extract_text() or "") + "\n"
-#     return text
-
-# kb_text = extract_pdf_text(kb_pdf_path)[:3000]
-
 # ================= SEARCH =================
 def is_small_talk(text: str) -> bool:
     text = text.strip().lower()
@@ -74,8 +49,12 @@ def is_small_talk(text: str) -> bool:
         r"(hi|hello|hey|hii|hai|namaste|thanks|thank you|ok|okay|yes|no|namaskar|good morning|good afternoon|good evening|good night|how are you?|how are you|how are you.|who are you?|introduce yourself|introduce)",
         text
     ))
-    
+
+
 def search_services(query, min_score=0.75):
+    if not services_index:
+        return None
+
     emb = embedder.encode([query])[0].tolist()
     res = services_index.query(
         vector=emb,
@@ -88,13 +67,15 @@ def search_services(query, min_score=0.75):
 
     match = res.matches[0]
     if match.score < min_score:
-        return None   # ⛔ STOP RANDOM SERVICE MATCHING
+        return None
 
     return match.metadata["text"]
 
 
-
 def search_basics(query):
+    if not basics_index:
+        return ""
+
     emb = embedder.encode([query])[0].tolist()
     res = basics_index.query(
         vector=emb,
@@ -122,28 +103,28 @@ def get_service_info(dept, service):
 # ================= RESPONSE =================
 def generate_response(user_query, language="en"):
 
+    # ================= GUARD =================
+    if not client or not pc:
+        return "<p>❌ Server misconfiguration: API keys are missing. Please contact the administrator.</p>"
+
     # ================= SMALL TALK SHORT-CIRCUIT =================
     if is_small_talk(user_query):
         prompt = f"""
 You are Aaple Sarkar Services Chatbot (Maharashtra Government) named "Aapla Sahayak".
-
 Rules:
 - Respond politely, briefly, and naturally using some 1-2 emojis, also while greeting say Namaskar.
 - Respond with RAW, valid, render-ready HTML only.
 - Do not use markdown.
 - Do not add explanations outside HTML.
-
 User Query:
 {user_query}
 """
-
         response = client.chat.completions.create(
             model="gpt-4o-mini",
             messages=[{"role": "system", "content": prompt}],
             temperature=0.3,
             max_tokens=150,
         )
-
         return response.choices[0].message.content
 
     # ================= RAG PIPELINE =================
@@ -184,17 +165,13 @@ User Query:
     # ================= MAIN PROMPT =================
     prompt = f"""
 You are Aaple Sarkar Services Chatbot (Maharashtra Government).
-
 Rules:
 - Respond with RAW, valid, render-ready HTML only; do not use markdown, do not wrap the response in ``` or ```html, do not add explanations/comments/text outside HTML, do not add leading or trailing whitespace, and ensure the response starts directly with <html> or the first HTML tag.
 - If the user message is a generic, short, or social message (such as greetings, acknowledgements, or fillers), you must respond appropriately with a polite, natural, and context-independent reply.
-
 User Query:
 {user_query}
-
 Available Knowledge:
 {context}
-
 Provide complete guidance including:
 - Eligibility
 - Documents required
@@ -211,4 +188,4 @@ Provide complete guidance including:
         max_tokens=800,
     )
 
-    return response.choices[0].message.content
+    return response.choices[0].message.content