Updated app

app.py

@@ -1,3 +1,5 @@
+import os
+import chainlit as cl
 from datasets import load_dataset
 from langchain.embeddings import CacheBackedEmbeddings
 from langchain.storage import LocalFileStore
@@ -11,9 +13,13 @@ from langchain_community.vectorstores import FAISS
 from langchain_openai import OpenAIEmbeddings
 from langchain_openai import ChatOpenAI
 
-dataset = load_dataset('ShubhamChoksi/IMDB_Movies')
-dataset_dict = dataset
-dataset_dict["train"].to_csv('imdb.csv')
+
+imdb_csv_file="./imdb.csv"
+
+if not os.path.exists(imdb_csv_file):
+    dataset = load_dataset('ShubhamChoksi/IMDB_Movies')
+    dataset_dict = dataset
+    dataset_dict["train"].to_csv('imdb.csv')
 
 loader = CSVLoader(file_path='imdb.csv')
 data = loader.load()
@@ -24,8 +30,6 @@ text_splitter = RecursiveCharacterTextSplitter(
 )
 chunked_documents = text_splitter.split_documents(data)
 
-import os
-
 openai_api_key = os.getenv("openai_api_key")
 embedding_model = OpenAIEmbeddings(model="text-embedding-3-small", openai_api_key=openai_api_key)
 
@@ -33,8 +37,12 @@ store = LocalFileStore("./cache/")
 cached_embedder = CacheBackedEmbeddings.from_bytes_store(embedding_model, store, namespace=embedding_model.model)
 
 vector_file = "local_vector"
-vector_store = FAISS.from_documents(chunked_documents, cached_embedder)
-vector_store.save_local(vector_file)
+
+if not os.path.exists(imdb_csv_file):
+    vector_store = FAISS.from_documents(chunked_documents, cached_embedder)
+    vector_store.save_local(vector_file)
+else:
+    vector_store = FAISS.load_local(vector_file, cached_embedder)
 
 prompt_template = ChatPromptTemplate.from_template(
     "You are a movie recommendation system, for a given {query} find recommendations from {content}."
@@ -50,5 +58,19 @@ runnable_chain = (
     | StrOutputParser()
 )
 
-output_chunks = runnable_chain.invoke(query)
-print(''.join(output_chunks))
+@cl.on_chat_start
+async def on_chat_start():
+    await cl.Message(content="Hello, i am your movie recommender, how can i help you today?").send()
+    cl.user_session.set("chain", runnable_chain)
+
+
+@cl.on_message
+async def main(message):
+    chain = cl.user_session.get("chain")
+    cb = cl.AsyncLangchainCallbackHandler()
+    user_input = cl.Message(content="")
+
+    async for stream in chain.astream(user_input, cb):
+        await user_input.stream_token(stream)
+
+    await user_input.send()