Switched to requests library to fix Render DNS bug and added wait_for_model flag

api.py

@@ -1,4 +1,5 @@
 import os
+import requests
 import httpx
 from fastapi import FastAPI
 from fastapi.responses import StreamingResponse
@@ -37,25 +38,37 @@ async def rag_query(query: SearchQuery):
     hf_url = "https://api-inference.huggingface.co/models/sentence-transformers/all-MiniLM-L6-v2"
     headers = {"Authorization": f"Bearer {HF_TOKEN}"}
     
-    # Fetch embeddings from HuggingFace
-    async with httpx.AsyncClient() as client:
-        hf_response = await client.post(
+    # NEW: Bulletproof synchronous network call (Bypasses Render DNS bugs)
+    hf_url = "https://api-inference.huggingface.co/pipeline/feature-extraction/sentence-transformers/all-MiniLM-L6-v2"
+    headers = {"Authorization": f"Bearer {HF_TOKEN}"}
+    
+    print("Pinging Hugging Face...")
+    try:
+        hf_response = requests.post(
             hf_url, 
             headers=headers, 
-            json={"inputs": [query.question]},
-            timeout=60.0
+            json={
+                "inputs": [query.question],
+                "options": {"wait_for_model": True} # Forces HF to wake up automatically
+            },
+            timeout=120.0 # Gives the model plenty of time to boot up
         )
+        hf_data = hf_response.json()
         
-    hf_data = hf_response.json()
-    
-    # NEW: Handle the Hugging Face "Cold Start" safely
-    if isinstance(hf_data, dict) and "error" in hf_data:
-        print(f"HuggingFace is sleeping: {hf_data}")
-        async def wakeup_message():
-            yield "The AI Brain is waking up from a deep sleep! Please wait 20 seconds and click Ask AI again."
-        return StreamingResponse(wakeup_message(), media_type="text/plain")
+        # Safety check if HF is completely overloaded
+        if isinstance(hf_data, dict) and "error" in hf_data:
+            print(f"HuggingFace Error: {hf_data}")
+            async def error_msg():
+                yield "The AI Brain is experiencing heavy traffic. Please try asking again in 15 seconds!"
+            return StreamingResponse(error_msg(), media_type="text/plain")
+            
+        query_numbers = hf_data[0]
         
-    query_numbers = hf_data[0]
+    except Exception as e:
+        print(f"Critical Network Error: {e}")
+        async def crash_msg():
+            yield "The AI Brain is temporarily disconnected from the network. Please try again."
+        return StreamingResponse(crash_msg(), media_type="text/plain")
     
     result = supabase.rpc("match_documents", {
         "query_embedding": query_numbers,