CupidDB

---

Running CupidDB Server

The easiest way to run a CupidDB server is by using Docker. You can simply start a server with the following command:

docker run --rm -p 5995:5995 cupiddb/cupiddb:latest

---

Installing Python Client

The CupidDB Python client is available on PyPI under the name pycupiddb.

pip install pycupiddb

---

Using CupidDB to Store DataFrame

CupidDB is ideal for storing DataFrames. When you store a DataFrame, CupidDB saves the data in Apache Arrow format. This allows the client to request the DataFrame with filter logic, which is executed server-side, reducing data transfer over the network. More importantly, data serialization and deserialization are avoided since CupidDB natively stores and transfers data in Apache Arrow format.

    from pycupiddb import CupidClient, RowFilter
    import pandas as pd

    df = pd.DataFrame({'a': [1, 2, 3], 'b': [4, 5, 6]})

    cupid_client = CupidClient(host='localhost', port=5995)

    cupid_client.set(key='key', value=df)

    df = cupid_client.get_dataframe(key='key')

    # Or using filters

    filters = [
        RowFilter(column='b', logic='gte', value=5, data_type='int')
    ]
    df = cupid_client.get_dataframe(
        key='key',
        columns=['a'],
        filters=filters,
    )
    print(df)
    # You should see
    #   a
    # 0 2
    # 1 3
  

---

Using CupidDB as a Cache Server

CupidDB can be used as a cache server. It follows the get and set API you might already be familiar with.

    cupid_client.set(key='key', value='value', timeout=60)

    result = cupid_client.get(key='key')
    print(result)
    # This should print out "value"

    cupid_client.set(key='key', value=1, timeout=60)
    result = cupid_client.incr(key='key')
    print(result) # This should print out 2

    result = cupid_client.incr(key='key', delta=5)
    print(result) # This should print out 7