garlic 🧄

cli and python interface for interacting with Snowflake

Features

• run queries against snowflake using simple UX
• auto-convert query results to polars dataframes
• convenience functions for:
  • formatting timestamps for use in SQL queries
  • setting the context (warehouse, database, schema, role)
  • listing catalog of databases, schemas, tables, and query history
  • creating tables from select queries

Installation

uv add paradigm_garlic

Credentials

To use garlic without giving your credentials each time, add your credentials the SNOWFLAKE_CREDENTIALS environment variable

export SNOWFLAKE_CREDENTIALS='<JSON_CREDENTIALS>'

Where <JSON_CREDENTIALS> is a JSON-encoded map of credentials

For example, if using a programmatic access key:

{
    "account": "abc12345.xy12345",
    "user": "my_username",
    "authenticator": "PROGRAMMATIC_ACCESS_KEY",
    "token": "<YOUR_PROGRAMMATIC_ACCESS_TOKEN>"
}

Alternatively, can set the credentials env var in a live python session:

os.environ['SNOWFLAKE_CREDENTIALS'] = json.dumps(credentials_dict)

Or specify the credentials in each function call:

import garlic

results = garlic.query('SELECT * FROM my_table', credentials=credentials_dict)

Example Usage

Simplest example

import garlic

dataframe = garlic.query('SELECT * FROM my_table')

Set different default warehouse:

import garlic

garlic.use_warehouse('BIG_WAREHOUSE')
dataframe = garlic.query('SELECT * FROM my_table')

Read from Snowflake management tables

import garlic

databases = garlic.list_databases()
schemas = garlic.list_schemas()
tables = garlic.list_tables()
query_history = garlic.list_query_history()

Use timestamps in CLI queries

import garlic
import datetime

sql = """
SELECT *
FROM my_table
WHERE
    block_timestamp >= {start_time}
    AND block_timestamp < {end_time}
""".format(
    start_time=garlic.format_timestamp('2024-01-01'),
    end_time=garlic.format_timestamp(datetime.datetime.now()),
)

dataframe = garlic.query(sql)

Set environment context

garlic.use_warehouse('BIG_WH')
garlic.use_schema('MY_SCHEMA')
garlic.use_database('MY_DB')
garlic.use_role('MY_ROLE')

dataframe = garlic.query('SELECT * FROM my_table')

Create table from select query

import garlic

garlic.create_table(
    target_table='new_table_name',
    select_sql='SELECT * FROM my_table WHERE some_column = some_value',
)