Update backend/chatbot.py

backend/chatbot.py

@@ -1,22 +1,53 @@
-import openai
+import os
+import requests
+from dotenv import load_dotenv
 from messages import krishna_blessings
 from ayush_messages import ayush_surprises
 
-openai.api_key = os.getenv("OPENAI_API_KEY")
+load_dotenv()
+REPLICATE_API_TOKEN = os.getenv("REPLICATE_API_TOKEN")
 
 def get_krishna_response(user_input):
+    # Check for predefined responses
     if "birthday" in user_input.lower():
         return ayush_surprises["birthday"]
     if "shy" in user_input.lower():
-        return "Hare Manavi! I hid from the gopis too—quiet time is golden like my butter!"
-    prompt = (
-        "You are Little Krishna, playful and wise. Respond to this in a fun, Krishna-like way: "
-        f"'{user_input}'"
-    )
-    response = openai.Completion.create(
-        engine="text-davinci-003",
-        prompt=prompt,
-        max_tokens=60,
-        temperature=0.9
+        return krishna_blessings["shy"]
+    if "hello" in user_input.lower() or "hi" in user_input.lower():
+        return krishna_blessings["greeting"]
+
+    # Use LLaMA-3 via Replicate API
+    headers = {
+        "Authorization": f"Token {REPLICATE_API_TOKEN}",
+        "Content-Type": "application/json"
+    }
+    payload = {
+        "input": {
+            "prompt": f"You are Little Krishna, playful and wise. Respond to this in a fun, Krishna-like way: '{user_input}'",
+            "max_tokens": 60,
+            "temperature": 0.9
+        }
+    }
+    response = requests.post(
+        "https://api.replicate.com/v1/predictions",
+        headers=headers,
+        json={
+            "version": "meta-llama/llama-3-8b",  # Replace with actual model version after access
+            "input": payload["input"]
+        }
     )
-    return response.choices[0].text.strip()
+    if response.status_code == 200:
+        prediction = response.json()
+        # Wait for prediction to complete (Replicate uses async predictions)
+        prediction_id = prediction["id"]
+        while True:
+            status_response = requests.get(
+                f"https://api.replicate.com/v1/predictions/{prediction_id}",
+                headers=headers
+            )
+            status_data = status_response.json()
+            if status_data["status"] == "succeeded":
+                return status_data["output"]
+            elif status_data["status"] == "failed":
+                return "Sorry, I couldn’t generate a response. Let’s try something else!"
+    return "Error connecting to LLaMA-3. Please try again later."