Update app.py

app.py

@@ -1,3 +1,9 @@
+from gevent import monkey
+monkey.patch_all()
+
+import nltk
+nltk.download('punkt_tab')
+
 import os
 from dotenv import load_dotenv
 import asyncio
@@ -15,6 +21,8 @@ from pinecone_text.sparse import BM25Encoder
 from langchain_huggingface import HuggingFaceEmbeddings
 from langchain_community.retrievers import PineconeHybridSearchRetriever
 from langchain_groq import ChatGroq
+from langchain.retrievers.document_compressors import FlashrankRerank
+from langchain_community.chat_models import ChatPerplexity
 
 # Load environment variables
 load_dotenv(".env")
@@ -32,7 +40,7 @@ os.environ["TOKENIZERS_PARALLELISM"] = 'true'
 # Initialize Flask app and SocketIO with CORS
 app = Flask(__name__)
 CORS(app)
-socketio = SocketIO(app, cors_allowed_origins="*")
+socketio = SocketIO(app, async_mode='gevent', cors_allowed_origins="*")
 app.config['SESSION_COOKIE_SECURE'] = True  # Use HTTPS
 app.config['SESSION_COOKIE_HTTPONLY'] = True
 app.config['SESSION_COOKIE_SAMESITE'] = 'Lax'
@@ -56,27 +64,10 @@ def initialize_pinecone(index_name: str):
 pinecone_index = initialize_pinecone("traveler-demo-website-vectorstore")
 bm25 = BM25Encoder().load("./bm25_traveler_website.json")
 
-### This is for UAE Legislation Website
-# pinecone_index = initialize_pinecone("uae-legislation-site-data")
-# bm25 = BM25Encoder().load("./bm25_uae_legislation_data.json")
-
-
-### This is for u.ae Website
-# pinecone_index = initialize_pinecone("vector-store-index")
-# bm25 = BM25Encoder().load("./bm25_u.ae.json")
-
-
-# #### This is for UAE Economic Department Website
-# pinecone_index = initialize_pinecone("uae-department-of-economics-site-data")
-# bm25 = BM25Encoder().load("./bm25_uae_department_of_economics_data.json")
-
-
 
 ##################################################
 ##################################################
 
-old_embed_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-base-v2")
-
 # Initialize models and retriever
 embed_model = HuggingFaceEmbeddings(model_name="Alibaba-NLP/gte-large-en-v1.5", model_kwargs={"trust_remote_code":True})
 retriever = PineconeHybridSearchRetriever(
@@ -88,7 +79,8 @@ retriever = PineconeHybridSearchRetriever(
 )
 
 # Initialize LLM
-llm = ChatGroq(model="llama-3.1-70b-versatile", temperature=0, max_tokens=1024, max_retries=2)
+# llm = ChatGroq(model="llama-3.1-70b-versatile", temperature=0, max_tokens=1024, max_retries=2)
+llm = ChatPerplexity(temperature=0, pplx_api_key=GROQ_API_KEY, model="llama-3.1-sonar-large-128k-online", max_tokens=1024, max_retries=2)
 
 # Contextualization prompt and retriever
 contextualize_q_system_prompt = """Given a chat history and the latest user question \