cloudy-sql 0.0.0.4

A Jupyter magics extension designed for seamless interaction with Snowflake

cloudy_sql

Snowflake Support

The cloudy_sql is an Ipython Magics library and pandas extension that currently supports:

• a Magics function that allows users to easily execute SQL Queries in Snowflake
• writing to an existing Snowflake table from a pandas dataframe
• creating a new Snowflake table from a pandas dataframe

How To Use cloudy_sql

Installation

pip install cloudy-sql

Configuration

Upon installation, open a jupyter notebook and run the following cell.

%load_ext cloudy_sql

After you run the cell, a configuration file will be created in your HOME directory. The path to the configuration file is: $HOME/.cloudy_sql/configuration_profiles.yml

For Windows user: use $USERPROFILE instead of $HOME variable

The configuration file is a YAML file with the following format

profiles:
  snowflake:
    user: <your snowflake username>
    pass: <your snowflake password>
    acct: <your snowflake account>
    role: <your snowflake role>
    warehouse: <your snowflake warehouse>
    database: <your snowflake database>
    schema: <your snowflake schema>

The user, pass, acct, database, schema, values all should be filled in with your desired Snowflake credentials and connection details. The variables in this file serve as default arguments when calling a cloudy_sql method. Role and warehouse can be filled in as well, but they are optional arguments when connecting to Snowflake.

API

The intent has been to keep the API as simple as possible by minimally extending the pandas and IPython Magics APIs.

Optional Arguments

There are two methods for passing optional arguments into a method.

1. The configuration file
2. Directly pass in the arguments when calling the method

The variables saved in the configuration file serve as default arguments for the methods to use. However, you tell the method to use different credentials by passing in arguments directly. The method will use the passed in arguments instead of the default arguments saved in configuration_profiles.yml.

For example, if I had the database variable saved in the configuration_profiles.yml as database_1, but passed in database = database_2 directly into the method, the method would use database_2 instead of database_1.

However, if I choose to not directly pass a database argument in, the method will use the database_1 because it is the default. The passed in arguments take priority over the default variables saved in configuration_profiles.yml.

IPython Magics

%%sql_to_snowflake

IPython Cell Magic to seamlessly connect to Snowflake and run a query in Snowflake and return a pandas DataFrame as the result.

%%sql_to_snowflake [<destination_var>] [--username <snowflake_username>]
                   [--password <snowflake_password>] [--account <snowflake_account>]
                   [--role <snowflake_role>] [--warehouse <snowflake_warehouse>]
<SQL query>

Parameters

* <destination_var> (Optional [IPython line argument]): 
    Variable to store the query results. If none is given, the magic will return
    the first 10 rows of the pandas DataFrame if applicable.

* --params <params> (Optional [IPython line argument]):
    Parameters to be used in the SQL Query. Params must be passed in as a 
    dictionary string in the format {"param_name": "param_value"} or reference a 
    dictionary string defined in a previous cell. The use of the parameter in the 
    query should be indicated with {{param_name}}.

* --username <username> (Optional [IPython line argument]):
    If provided, the called method will connect to Snowflake with this username 
    instead of the default in the configuration file.

* --password <password> (Optional [IPython line argument]):
    If provided, the called method will connect to Snowflake with this password
    instead of the default in the configuration file.

* --account <account> (Optional [IPython line argument]):
    If provided, the called method will connect to Snowflake with this account 
    instead of the default in the configuration file.

* --role <role> (Optional [IPython line argument]):
    If provided, the called method will connect to Snowflake with this role 
    instead of the default in the configuration file.

* --warehouse <warehouse> (Optional [IPython line argument]):
    If provided, the called method will use this warehouse instead of the 
    default in the configuration file.

write_snowflake

pd.DataFrame.cloudy_sql.write_snowflake(table: str, 
                                        database: str = None, 
                                        schema: str = None, 
                                        overwrite: bool = False, 
                                        username: str = None,
                                        password: str = None, 
                                        account: str = None, 
                                        role: str = None, 
                                        warehouse: str = None
                                       )

This method writes to a Snowflake table and informs you on success. This method works when writing to either an existing Snowflake table or a previously non-existing Snowflake table. If the table that you provide does not exist, this method creates a new Snowflake table and writes to it. If the table already exists, the DataFrame data is appended to the existing table by default. If you would like to replace the table with the pandas DataFrame set overwrite = True when calling the method. You can configure your database and schema default values in the configuration file. You can also pass them in directly when calling the method. The passed in values are used instead of the defaults in the configuration file.

The goal of this method is to be used in tandem with %%sql_to_snowflake. You use the magic function to run your SQL query that returns a pandas DataFrame. Then, you can transform the DataFrame and write the DataFrame to a Snowflake using this method.

Examples

Using %%sql_to_snowflake magic and write_snowflake method

In [1]:  %load_ext cloudy_sql

In [2]:  %%sql_to_snowflake df 
         SELECT * from db.schema.table
         
         Query successfully ran and results were stored to the 'df' destination variable.

In [3]:  df.head()
Out[3]:  
         CUSTOMER_ID	FIRST_NAME	LAST_NAME	FIRST_ORDER_DATE	MOST_RECENT_ORDER_DATE	NUMBER_OF_ORDERS
     0   	   1	   Michael	       P.	      2018-01-01	            2018-02-10	               2
     1   	   2	     Shawn	       M.	      2018-01-11	            2018-01-11	               1
     2   	   3	  Kathleen	       P.	      2018-01-02	            2018-03-11	               3
     3   	   4	     Jimmy	       C.	            None	                  None	               0
     4   	   5	 Katherine	       R.	            None	                  None	               0
     
In [4]: df.cloudy_sql.write_snowflake(table="test_cloudy_sql")

        Successfully wrote to the test_cloudy_sql Snowflake table

In [5]: %%sql_to_snowflake
        drop table if exists db.schema.test_cloudy_sql
        
        Successfully ran SQL Query in Snowflake
        
In [6]: %close_connection

The above example runs a SQL query with %%sql_to_snowflake and saves the results as a pandas DataFrame by passing in the destination variable df. The example then shows how to easily write that df to a Snowflake table In [4]. Also, the created table is dropped In [5]. The connection is then closed by calling the %close_connection magic In [6].

Using %%sql_to_snowflake magic with the --params inline argument

In [1]:  %load_ext cloudy_sql

In [2]:  parameters = {'firstname': 'Michael', 'orders': '2'}
         
In [3]:  %%sql_to_snowflake --params $parameters
         SELECT * from db.schema.table
         WHERE FIRST_NAME = {{firstname}} and NUMBER_OF_ORDERS = {{orders}}
Out[3]:  
         CUSTOMER_ID	FIRST_NAME	LAST_NAME	FIRST_ORDER_DATE	MOST_RECENT_ORDER_DATE	NUMBER_OF_ORDERS
     0   	   1	   Michael	       P.	      2018-01-01	            2018-02-10	               2
     

The above example runs a SQL query with passed-in variables. The variables are used directly in the SQL query by placing each one inside {{ }}. A dictionary string parameters is passed in when the magic is called by including a --params inline argument and placing a $ to reference the dictionary string creating in the previous cell. There is no specified destination variable. Therefore, the magic prints out the resulting pandas dataframe.

Using %%sql_to_snowflake magic with the --params inline argument

In [1]:  %load_ext cloudy_sql
         
In [2]:  %%sql_to_snowflake --params {'firstname': 'Michael', 'orders': '2'}
         SELECT * from db.schema.table
         WHERE FIRST_NAME = {{firstname}} and NUMBER_OF_ORDERS = {{orders}}
Out[2]:  
         CUSTOMER_ID	FIRST_NAME	LAST_NAME	FIRST_ORDER_DATE	MOST_RECENT_ORDER_DATE	NUMBER_OF_ORDERS
     0   	   1	   Michael	       P.	      2018-01-01	            2018-02-10	               2
     

This example directly passes in the dictionary string when calling the --params inline argument.