langchain/docs/extras/integrations/vectorstores/cassandra.ipynb

Cassandra

Apache Cassandra® is a NoSQL, row-oriented, highly scalable and highly available database.

Newest Cassandra releases natively support Vector Similarity Search.

To run this notebook you need either a running Cassandra cluster equipped with Vector Search capabilities (in pre-release at the time of writing) or a DataStax Astra DB instance running in the cloud (you can get one for free at datastax.com). Check cassio.org for more information.

In [ ]:

!pip install "cassio>=0.1.0"

Please provide database connection parameters and secrets:

In [ ]:

import os
import getpass

database_mode = (input("\n(C)assandra or (A)stra DB? ")).upper()

keyspace_name = input("\nKeyspace name? ")

if database_mode == "A":
    ASTRA_DB_APPLICATION_TOKEN = getpass.getpass('\nAstra DB Token ("AstraCS:...") ')
    #
    ASTRA_DB_SECURE_BUNDLE_PATH = input("Full path to your Secure Connect Bundle? ")
elif database_mode == "C":
    CASSANDRA_CONTACT_POINTS = input(
        "Contact points? (comma-separated, empty for localhost) "
    ).strip()

depending on whether local or cloud-based Astra DB, create the corresponding database connection "Session" object

In [ ]:

from cassandra.cluster import Cluster
from cassandra.auth import PlainTextAuthProvider

if database_mode == "C":
    if CASSANDRA_CONTACT_POINTS:
        cluster = Cluster(
            [cp.strip() for cp in CASSANDRA_CONTACT_POINTS.split(",") if cp.strip()]
        )
    else:
        cluster = Cluster()
    session = cluster.connect()
elif database_mode == "A":
    ASTRA_DB_CLIENT_ID = "token"
    cluster = Cluster(
        cloud={
            "secure_connect_bundle": ASTRA_DB_SECURE_BUNDLE_PATH,
        },
        auth_provider=PlainTextAuthProvider(
            ASTRA_DB_CLIENT_ID,
            ASTRA_DB_APPLICATION_TOKEN,
        ),
    )
    session = cluster.connect()
else:
    raise NotImplementedError

Please provide OpenAI access key

We want to use OpenAIEmbeddings so we have to get the OpenAI API Key.

In [ ]:

os.environ["OPENAI_API_KEY"] = getpass.getpass("OpenAI API Key:")

Creation and usage of the Vector Store

In [ ]:

from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.text_splitter import CharacterTextSplitter
from langchain.vectorstores import Cassandra
from langchain.document_loaders import TextLoader

In [ ]:

from langchain.document_loaders import TextLoader

SOURCE_FILE_NAME = "../../modules/state_of_the_union.txt"

loader = TextLoader(SOURCE_FILE_NAME)
documents = loader.load()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
docs = text_splitter.split_documents(documents)

embedding_function = OpenAIEmbeddings()

In [ ]:

table_name = "my_vector_db_table"

docsearch = Cassandra.from_documents(
    documents=docs,
    embedding=embedding_function,
    session=session,
    keyspace=keyspace_name,
    table_name=table_name,
)

query = "What did the president say about Ketanji Brown Jackson"
docs = docsearch.similarity_search(query)

In [ ]:

## if you already have an index, you can load it and use it like this:

# docsearch_preexisting = Cassandra(
#     embedding=embedding_function,
#     session=session,
#     keyspace=keyspace_name,
#     table_name=table_name,
# )

# docs = docsearch_preexisting.similarity_search(query, k=2)

In [ ]:

print(docs[0].page_content)

Maximal Marginal Relevance Searches

In addition to using similarity search in the retriever object, you can also use mmr as retriever.

In [ ]:

retriever = docsearch.as_retriever(search_type="mmr")
matched_docs = retriever.get_relevant_documents(query)
for i, d in enumerate(matched_docs):
    print(f"\n## Document {i}\n")
    print(d.page_content)

Or use max_marginal_relevance_search directly:

In [ ]:

found_docs = docsearch.max_marginal_relevance_search(query, k=2, fetch_k=10)
for i, doc in enumerate(found_docs):
    print(f"{i + 1}.", doc.page_content, "\n")

Metadata filtering

You can specify filtering on metadata when running searches in the vector store. By default, when inserting documents, the only metadata is the "source" (but you can customize the metadata at insertion time).

Since only one files was inserted, this is just a demonstration of how filters are passed:

In [ ]:

filter = {"source": SOURCE_FILE_NAME}
filtered_docs = docsearch.similarity_search(query, filter=filter, k=5)
print(f"{len(filtered_docs)} documents retrieved.")
print(f"{filtered_docs[0].page_content[:64]} ...")

In [ ]:

filter = {"source": "nonexisting_file.txt"}
filtered_docs2 = docsearch.similarity_search(query, filter=filter)
print(f"{len(filtered_docs2)} documents retrieved.")

Please visit the cassIO documentation for more on using vector stores with Langchain.