Ibis Framework Documentation

After installation, you can enable interactive mode, fetch example data (like the penguins dataset), and perform data manipulations such as grouping and aggregating using the Ibis dataframe API.

import ibis
ibis.options.interactive = True
t = ibis.examples.penguins.fetch()
g = t.group_by("species", "island").agg(count=t.count()).order_by("count")

You can install Ibis along with the DuckDB backend and example datasets using pip. This is a quick way to get started with local data exploration.

pip install 'ibis-framework[duckdb,examples]'

You can create tables from an Ibis schema or directly from a Pandas DataFrame. Temporary tables created via create_table(..., temp=True) are automatically dropped when the connection closes.

import ibis
import pandas as pd

# Create table from schema
schema = ibis.schema([
    ('id', 'int64'),
    ('name', 'string'),
    ('price', 'float64'),
    ('created_at', 'timestamp')
])
tbl = ibis_con.create_table('new_products', schema=schema)

# Create temporary table from Pandas DataFrame
temp_data = pd.DataFrame({"id": [1, 2, 3], "value": [10, 20, 30]})
temp_table = ibis_con.create_table("temp_analysis", temp_data, temp=True)

If you already have an active connection using the singlestoredb Python SDK, you can wrap it in an Ibis client using ibis.singlestoredb.from_connection().

import singlestoredb as s2
import ibis

# Create connection using SingleStoreDB client directly
con = s2.connect(
    host="localhost",
    user="root",
    password="password",
    database="my_database"
)

# Create Ibis client from existing connection
ibis_con = ibis.singlestoredb.from_connection(con)

You can establish a connection to SingleStoreDB by passing connection parameters directly to ibis.singlestoredb.connect().

import ibis

con = ibis.singlestoredb.connect(
    host="localhost",
    port=3306,
    user="root",
    password="password",
    database="my_database"
)

Use ibis.connect() with a URI-style connection string. For passwords containing special characters, ensure you use URL encoding (e.g., via urllib.parse.quote_plus).

import ibis
from urllib.parse import quote_plus

# Basic connection string
con = ibis.connect("singlestoredb://user:password@host:port/database")

# With additional parameters
con = ibis.connect("singlestoredb://user:password@host:port/database?autocommit=true&local_infile=1")

# URL with special characters
password = "p@ssw0rd!"
encoded_password = quote_plus(password)
con = ibis.connect(f"singlestoredb://user:{encoded_password}@host:port/database")

The SingleStoreDB backend allows executing raw SQL queries. You can use raw_sql() which returns a cursor, or use it as a context manager for automatic cursor management.

# Using context manager for automatic cursor management
with ibis_con.raw_sql("SELECT COUNT(*) FROM users") as cursor:
    count = cursor.fetchone()[0]

# Manual cursor management
cursor = ibis_con.raw_sql("SHOW TABLES")
tables = [row[0] for row in cursor.fetchall()]
cursor.close()