Update app.py

app.py

@@ -68,25 +68,6 @@ landmark_data = {
     "Los Angeles": "Buckingham Palace"
 }
 
-# preloading the documents to make faster response as free hf space cpu is slow 
-PRELOAD_URLS = [
-    "https://hackrx.blob.core.windows.net/assets/Arogya%20Sanjeevani%20Policy%20-%20CIN%20-%20U10200WB1906GOI001713%201.pdf?sv=2023-01-03&st=2025-07-21T08%3A29%3A02Z&se=2025-09-22T08%3A29%3A00Z&sr=b&sp=r&sig=nzrz1K9Iurt%2BBXom%2FB%2BMPTFMFP3PRnIvEsipAX10Ig4%3D",
-    "https://hackrx.blob.core.windows.net/assets/Super_Splendor_(Feb_2023).pdf?sv=2023-01-03&st=2025-07-21T08%3A10%3A00Z&se=2025-09-22T08%3A10%3A00Z&sr=b&sp=r&sig=vhHrl63YtrEOCsAy%2BpVKr20b3ZUo5HMz1lF9%2BJh6LQ0%3D",
-    "https://hackrx.blob.core.windows.net/assets/Family%20Medicare%20Policy%20(UIN-%20UIIHLIP22070V042122)%201.pdf?sv=2023-01-03&st=2025-07-22T10%3A17%3A39Z&se=2025-08-23T10%3A17%3A00Z&sr=b&sp=r&sig=dA7BEMIZg3WcePcckBOb4QjfxK%2B4rIfxBs2%2F%2BNwoPjQ%3D",
-    "https://hackrx.blob.core.windows.net/assets/indian_constitution.pdf?sv=2023-01-03&st=2025-07-28T06%3A42%3A00Z&se=2026-11-29T06%3A42%3A00Z&sr=b&sp=r&sig=5Gs%2FOXqP3zY00lgciu4BZjDV5QjTDIx7fgnfdz6Pu24%3D",
-    "https://hackrx.blob.core.windows.net/assets/UNI%20GROUP%20HEALTH%20INSURANCE%20POLICY%20-%20UIIHLGP26043V022526%201.pdf?sv=2023-01-03&spr=https&st=2025-07-31T17%3A06%3A03Z&se=2026-08-01T17%3A06%3A00Z&sr=b&sp=r&sig=wLlooaThgRx91i2z4WaeggT0qnuUUEzIUKj42GsvMfg%3D",
-    "https://hackrx.blob.core.windows.net/assets/Happy%20Family%20Floater%20-%202024%20OICHLIP25046V062425%201.pdf?sv=2023-01-03&spr=https&st=2025-07-31T17%3A24%3A30Z&se=2026-08-01T17%3A24%3A00Z&sr=b&sp=r&sig=VNMTTQUjdXGYb2F4Di4P0zNvmM2rTBoEHr%2BnkUXIqpQ%3D",
-    "https://hackrx.blob.core.windows.net/assets/principia_newton.pdf?sv=2023-01-03&st=2025-07-28T07%3A20%3A32Z&se=2026-07-29T07%3A20%3A00Z&sr=b&sp=r&sig=V5I1QYyigoxeUMbnUKsdEaST99F5%2FDfo7wpKg9XXF5w%3D",
-    "https://hackrx.blob.core.windows.net/assets/hackrx_6/policies/HDFHLIP23024V072223.pdf?sv=2023-01-03&st=2025-07-30T06%3A46%3A49Z&se=2025-09-01T06%3A46%3A00Z&sr=c&sp=rl&sig=9szykRKdGYj0BVm1skP%2BX8N9%2FRENEn2k7MQPUp33jyQ%3D",
-    "https://hackrx.blob.core.windows.net/assets/Test%20/Test%20Case%20HackRx.pptx?sv=2023-01-03&spr=https&st=2025-08-04T18%3A36%3A56Z&se=2026-08-05T18%3A36%3A00Z&sr=b&sp=r&sig=v3zSJ%2FKW4RhXaNNVTU9KQbX%2Bmo5dDEIzwaBzXCOicJM%3D",
-    "https://hackrx.blob.core.windows.net/assets/Test%20/Mediclaim%20Insurance%20Policy.docx?sv=2023-01-03&spr=https&st=2025-08-04T18%3A42%3A14Z&se=2026-08-05T18%3A42%3A00Z&sr=b&sp=r&sig=yvnP%2FlYfyyqYmNJ1DX51zNVdUq1zH9aNw4LfPFVe67o%3D",
-    "https://hackrx.blob.core.windows.net/assets/Test%20/Salary%20data.xlsx?sv=2023-01-03&spr=https&st=2025-08-04T18%3A46%3A54Z&se=2026-08-05T18%3A46%3A00Z&sr=b&sp=r&sig=sSoLGNgznoeLpZv%2FEe%2FEI1erhD0OQVoNJFDPtqfSdJQ%3D",
-    "https://hackrx.blob.core.windows.net/assets/Test%20/Pincode%20data.xlsx?sv=2023-01-03&spr=https&st=2025-08-04T18%3A50%3A43Z&se=2026-08-05T18%3A50%3A00Z&sr=b&sp=r&sig=xf95kP3RtMtkirtUMFZn%2FFNai6sWHarZsTcvx8ka9mI%3D",
-    "https://hackrx.blob.core.windows.net/assets/Test%20/image.png?sv=2023-01-03&spr=https&st=2025-08-04T19%3A21%3A45Z&se=2026-08-05T19%3A21%3A00Z&sr=b&sp=r&sig=lAn5WYGN%2BUAH7mBtlwGG4REw5EwYfsBtPrPuB0b18M4%3D",
-    "https://hackrx.blob.core.windows.net/assets/Test%20/image.jpeg?sv=2023-01-03&spr=https&st=2025-08-04T19%3A29%3A01Z&se=2026-08-05T19%3A29%3A00Z&sr=b&sp=r&sig=YnJJThygjCT6%2FpNtY1aHJEZ%2F%2BqHoEB59TRGPSxJJBwo%3D",
-    "https://hackrx.blob.core.windows.net/assets/hackrx_pdf.zip?sv=2023-01-03&spr=https&st=2025-08-04T09%3A25%3A45Z&se=2027-08-05T09%3A25%3A00Z&sr=b&sp=r&sig=rDL2ZcGX6XoDga5%2FTwMGBO9MgLOhZS8PUjvtga2cfVk%3D",
-    "https://hackrx.blob.core.windows.net/assets/Test%20/Fact%20Check.docx?sv=2023-01-03&spr=https&st=2025-08-04T20%3A27%3A22Z&se=2028-08-05T20%3A27%3A00Z&sr=b&sp=r&sig=XB1%2FNzJ57eg52j4xcZPGMlFrp3HYErCW1t7k1fMyiIc%3D"
-]
 
 
 @asynccontextmanager
@@ -172,19 +153,6 @@ Step 3 – **Final Output**:
         print("❌ Lifespan error:", str(e))
         raise e
 
-        # Preloading vectorstores for all URLs
-    for url in PRELOAD_URLS:
-        doc_url = str(url)
-
-        if doc_url not in vector_cache:
-            print(f"📄 Processing new document: {doc_url}")
-            chunks = load_and_chunk(doc_url)
-    
-            # Building vectorstore & save to cache
-            vectorstore = await FAISS.afrom_documents(documents=chunks, embedding=ml_models["embedder"])
-            vector_cache[doc_url] = vectorstore  # store in memory cache
-            print(f"✅ Vectorstore cached for: {doc_url}")
-    
     yield
     print("🧹 Cleaning up.")
     ml_models.clear()