Firestore is a serverless document-oriented database that scales to meet any demand. Extend your database application to build AI-powered experiences leveraging Firestore's Langchain integrations.
This notebook goes over how to use Firestore to store vectors and query them using the FirestoreVectorStore class.
To run this notebook, you will need to do the following:
After confirmed access to database in the runtime environment of this notebook, filling the following values and run the cell before running example scripts.
🦜🔗 Library InstallationThe integration lives in its own langchain-google-firestore package, so we need to install it. For this notebook, we will also install langchain-google-genai to use Google Generative AI embeddings.
%pip install -upgrade --quiet langchain-google-firestore langchain-google-vertexai
Colab only: Uncomment the following cell to restart the kernel or use the button to restart the kernel. For Vertex AI Workbench you can restart the terminal using the button on top.
☁ Set Your Google Cloud ProjectSet your Google Cloud project so that you can leverage Google Cloud resources within this notebook.
If you don't know your project ID, try the following:
gcloud config list.gcloud projects list.
PROJECT_ID = "extensions-testing"
!gcloud config set project {PROJECT_ID}
Authenticate to Google Cloud as the IAM user logged into this notebook in order to access your Google Cloud Project.
from google.colab import auth
auth.authenticate_user()
FirestoreVectorStore allows you to store new vectors in a Firestore database. You can use it to store embeddings from any model, including those from Google Generative AI.
from langchain_google_firestore import FirestoreVectorStore
from langchain_google_vertexai import VertexAIEmbeddings
embedding = VertexAIEmbeddings(
model_name="textembedding-gecko@latest",
project=PROJECT_ID,
)
ids = ["apple", "banana", "orange"]
fruits_texts = ['{"name": "apple"}', '{"name": "banana"}', '{"name": "orange"}']
vector_store = FirestoreVectorStore(
collection="fruits",
embedding=embedding,
)
vector_store.add_texts(fruits_texts, ids=ids)
As a shorthand, you can initilize and add vectors in a single step using the from_texts and from_documents method.
vector_store = FirestoreVectorStore.from_texts(
collection="fruits",
texts=fruits_texts,
embedding=embedding,
)
from langchain_core.documents import Document
fruits_docs = [Document(page_content=fruit) for fruit in fruits_texts]
vector_store = FirestoreVectorStore.from_documents(
collection="fruits",
documents=fruits_docs,
embedding=embedding,
)
You can delete documents with vectors from the database using the delete method. You'll need to provide the document ID of the vector you want to delete. This will remove the whole document from the database, including any other fields it may have.
Updating vectors is similar to adding them. You can use the add method to update the vector of a document by providing the document ID and the new vector.
fruit_to_update = ['{"name": "apple","price": 12}']
apple_id = "apple"
vector_store.add_texts(fruit_to_update, ids=[apple_id])
You can use the FirestoreVectorStore to perform similarity searches on the vectors you have stored. This is useful for finding similar documents or text.
vector_store.similarity_search("I like fuji apples", k=3)
vector_store.max_marginal_relevance_search("fuji", 5)
You can add a pre-filter to the search by using the filters parameter. This is useful for filtering by a specific field or value.
from google.cloud.firestore_v1.base_query import FieldFilter
vector_store.max_marginal_relevance_search(
"fuji", 5, filters=FieldFilter("content", "==", "apple")
)
from google.api_core.client_options import ClientOptions
from google.cloud import firestore
from langchain_google_firestore import FirestoreVectorStore
client_options = ClientOptions()
client = firestore.Client(client_options=client_options)
vector_store = FirestoreVectorStore(
collection="fruits",
embedding=embedding,
client=client,
)
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4