Couchbase Vectore Store Handler

This is the implementation of the Couchbase Vector store data handler for MindsDB.

Implementation

In order to make use of this handler and connect to a Couchbase server in MindsDB, the following syntax can be used. Note, that the example uses the default travel-sample bucket which can be enabled from the couchbase UI with pre-defined scope and documents.

CREATE DATABASE couchbase_vectorsource
WITH
engine='couchbasevector',
parameters={
    "connection_string": "couchbase://localhost",
    "bucket": "travel-sample",
    "user": "admin",
    "password": "password",
    "scope": "inventory"
};

This handler is implemented using the couchbase library, the Python driver for Couchbase.

The required arguments to establish a connection are as follows:

• connection_string: the connection string for the endpoint of the Couchbase server
• bucket: the bucket name to use when connecting with the Couchbase server
• user: the user to authenticate with the Couchbase server
• password: the password to authenticate the user with the Couchbase server
• scope: scopes are a level of data organization within a bucket. If omitted, will default to _default

Note: The connection string expects either the couchbases:// or couchbase:// protocol.

If you are using Couchbase Capella, you can find the connection_string under the Connect tab. It will also be required to whitelist the machine(s) that will be running MindsDB and database credentials will need to be created for the user. These steps can also be taken under the Connect tab.

Usage

Now, you can use the established connection to create a collection (or table in the context of MindsDB) in Couchbase and insert data into it:

Creating tables

Now, you can use the established connection to create a collection (or table in the context of MindsDB) in Couchbase and insert data into it:

CREATE TABLE couchbase_vectorsource.test_embeddings (
    SELECT embeddings
    FROM mysql_datasource.test_embeddings
);

`mysql_datasource` is another MindsDB data source that has been created by connecting to a MySQL database. The `test_embeddings` table in the `mysql_datasource` data source contains the embeddings that we want to store in Couchbase.

Querying and searching

You can query your collection (table) as shown below:

SELECT * 
FROM couchbase_vectorsource.test_embeddings;

To filter the data in your collection (table) by metadata, you can use the following query:

SELECT * 
FROM couchbase_vectorsource.test_embeddings
WHERE id = "some_id";

To perform a vector search, the following query can be used:

SELECT *
FROM couchbase_vectorsource.test_embeddings
WHERE embeddings = (
    SELECT embeddings
    FROM mysql_datasource.test_embeddings
    LIMIT 1
);

Deleting records

You can delete documents using DELETE just like in SQL.

DELETE FROM couchbase_vectorsource.test_embeddings
WHERE `metadata.test` = 'test1';

Dropping connection

To drop the connection, use this command

DROP DATABASE couchbase_vectorsource;