Centaur Client Python Library

Centaur provides fully-managed, serverless, scalable vector-database service for building various machine learning applications. The Centaur client SDK is your gateway to access the Centaur service.

Installation

To install the Centaur client Python SDK, simply run:

pip install centaur-client

QuickStart

You can use Client api to communicate with Centaur service.

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')

Sample Code

Create a Collection

Client host various Collection APIs for interacting with Centaur service.

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
client.create('YOUR-COLLECTION-NAME', dimension=4)

List Collections

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
collections = client.list()
print(collections)

Describe Collection

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
rsp = client.describe('YOUR-COLLECTION-NAME')
print(rsp)

Delete Collection

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
client.delete('YOUR-COLLECTION-NAME')

Get a Collection Instance

Collection provides APIs for accessing Doc and Partition

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
collection = client.get('YOUR-COLLECTION-NAME')
if collection:
    print('collection:', collection)

Describe Collection Statistics

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
collection = client.get('YOUR-COLLECTION-NAME')
rsp = collection.stats()
print(rsp)

Insert/Update/Upsert Docs

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
collection = client.get('YOUR-COLLECTION-NAME')
collection.upsert(('YOUR-DOC-ID', [0.1, 0.2], {'price': 100, 'type': 'dress'}))

Query a Collection

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
collection = client.get('YOUR-COLLECTION-NAME')
match_docs = collection.query([0.1, 0.2, 0.3, 0.4], 
                       topk=100, 
                       include_vector=True)
if match_docs:
    for doc in match_docs:
        print(doc)

Delete Docs

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
collection = client.get('YOUR-COLLECTION-NAME')
collection.delete('YOUR-DOC-ID')

Fetch Docs

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
collection = client.get('YOUR-COLLECTION-NAME')
fetch_docs = collection.fetch(['YOUR-DOC-ID1', 'YOUR-DOC-ID2'])
if fetch_docs:
    for doc_id in fetch_docs:
        print(fetch_docs[doc_id])

Create a Collection Partition

import centaur

client = centaur.Client(api_key='YOUR-CENTAUR-API-KEY')
collection = client.get('YOUR-COLLECTION-NAME')
collection.create_partition('YOUR-PARTITION-NAME')

Centaur Response

from centaur import CentaurCode

@dataclasses
class CentaurResponse(object):
    code: CentaurCode
    message: str
    request_id: str
    output: Any

License

This project is licensed under the Apache License (Version 2.0).