llama_index_rag_pipeline_with_vector_store_persistence.py

1import os.path
2from llama_index.core import (
3    VectorStoreIndex,
4    SimpleDirectoryReader,
5    StorageContext,
6    load_index_from_storage,
7)
8
9# check if storage already exists
10PERSIST_DIR = "./storage"
11if not os.path.exists(PERSIST_DIR):
12    # load the documents and create the index
13    documents = SimpleDirectoryReader("data").load_data()
14    index = VectorStoreIndex.from_documents(documents)
15    # store it for later
16    index.storage_context.persist(persist_dir=PERSIST_DIR)
17else:
18    # load the existing index
19    storage_context = StorageContext.from_defaults(persist_dir=PERSIST_DIR)
20    index = load_index_from_storage(storage_context)
21
22# either way, we can now query the index
23query_engine = index.as_query_engine()
24response = query_engine.query("What did the author do growing up?")
25print(response)