Google Cloud SQL Vector Store Example

This example demonstrates how to use Cloud SQL for Postgres for vector similarity search with LangChain in Go.

What This Example Does

1. Creates a Cloud SQL VectorStore:

   • Initializes the cloudsql.PostgresEngine object to establish a connection to the Cloud SQL database.
   • Initializes a new table to store embeddings.
   • Initializes a cloudsql.VectorStore object using a VertexAI model for embeddings.

2. Initializes VertexAI Embeddings:

   • Creates an embeddings client using the VertexAI API.

3. Adds Sample Documents:

   • Inserts several documents (cities) with metadata into the vector store.
   • Each document includes the city name, population, and area.

4. Performs Similarity Searches:

   • Basic search for documents similar to "Japan".
   • Customized search for documents using filters by metadata.

How to Run the Example

1. Set the following environment variables:

   Copy

   export PROJECT_ID=<your project Id>
   export GOOGLE_CLOUD_LOCATION=<your cloud location>
   export POSTGRES_USERNAME=<your user>
   export POSTGRES_PASSWORD=<your password>
   export POSTGRES_REGION=<your region>
   export POSTGRES_INSTANCE=<your instance>
   export POSTGRES_DATABASE=<your database>
   export POSTGRES_TABLE=<your tablename>
   

2. Run the Go example:

   Copy

   go run google_cloudsql_vectorstore_example.go
   

Key Features

• This example demonstrates how to use cloudsql.PostgresEngine for connection pooling.
• It shows how to integrate with VertexAI embeddings models.
• Run the code to add documents and perform a similarity search with cloudsql.VectorStore.
• Demonstrates how to filter through the metadata added by using key value pairs.

This example provides a practical demonstration of using vector databases for semantic search and similarity matching, which can be incredibly useful for various AI and machine learning applications.