Snowpark for Python basics in Snowflake - Commands & Configuration

Introduction

Snowpark for Python lets you write Python code that runs directly inside Snowflake's data platform. This helps you process data close to where it lives, making data tasks faster and simpler without moving data around.

When you want to run Python data transformations directly inside Snowflake without exporting data.

When you need to build data pipelines that combine SQL and Python logic seamlessly.

When you want to use Python libraries for data processing but keep data secure inside Snowflake.

When you want to avoid managing separate compute resources for Python data jobs.

When you want to write reusable Python functions that run close to your data.

Commands

This command installs the Snowpark Python library so you can write Python code that connects to Snowflake and runs data operations.

Terminal

pip install snowflake-snowpark-python

Expected Output

Collecting snowflake-snowpark-python

  Downloading snowflake_snowpark_python-1.5.0-py3-none-any.whl (123 kB)

Installing collected packages: snowflake-snowpark-python

Successfully installed snowflake-snowpark-python-1.5.0

Runs a Python script that connects to Snowflake, creates a DataFrame, and shows data. This demonstrates basic Snowpark usage.

Terminal

python snowpark_example.py

Expected Output

   ID NAME

0   1 Alice

1   2 Bob

2   3 Carol

Key Concept

If you remember nothing else from Snowpark for Python, remember: it lets you run Python code directly inside Snowflake to process data efficiently without moving it.

Code Example

Snowflake

from snowflake.snowpark import Session

connection_parameters = {
    "account": "xy12345.us-east-1",
    "user": "MY_USER",
    "password": "MY_PASSWORD",
    "role": "SYSADMIN",
    "warehouse": "COMPUTE_WH",
    "database": "MY_DB",
    "schema": "PUBLIC"
}

session = Session.builder.configs(connection_parameters).create()

# Create a simple DataFrame
data = [(1, "Alice"), (2, "Bob"), (3, "Carol")]
df = session.create_dataframe(data, schema=["ID", "NAME"])

# Show the data
pandas_df = df.to_pandas()
print(pandas_df)

session.close()

Common Mistakes

Not installing the snowflake-snowpark-python package before running code.

The Python code will fail because the Snowpark library is missing.

Always run 'pip install snowflake-snowpark-python' first to install the required library.

Trying to run Snowpark code without setting up Snowflake connection parameters.

The code cannot connect to Snowflake and will error out.

Provide correct account, user, password, and warehouse details in the connection code.

Summary

Install the Snowpark Python library to enable Python data processing inside Snowflake.

Write Python scripts that connect to Snowflake using connection parameters.

Create DataFrames in Python and run operations that execute inside Snowflake.

Use to_pandas() to bring small results back to Python for viewing or further use.

Close the session to clean up the connection after your work is done.

Practice

1. What is the main purpose of Snowpark for Python in Snowflake?

easy

A. To run Python code directly inside Snowflake for data processing

B. To create visual dashboards from Snowflake data

C. To export Snowflake data to external Python scripts only

D. To replace SQL queries with Java code

Solution

1. Step 1: Understand Snowpark for Python's role

   Snowpark for Python allows running Python code inside Snowflake, enabling fast data processing close to the data.

2. Step 2: Compare options to the main purpose

   Creating visual dashboards, exporting Snowflake data to external Python scripts only, and replacing SQL queries with Java code describe other unrelated tasks or languages, while running Python code directly inside Snowflake for data processing correctly states the main purpose.

3. Final Answer:

   To run Python code directly inside Snowflake for data processing -> Option A

4. Quick Check:

   Snowpark runs Python inside Snowflake = A [OK]

Hint: Snowpark runs Python code inside Snowflake, not outside [OK]

Common Mistakes:

• Confusing Snowpark with visualization tools
• Thinking Snowpark exports data only
• Assuming Snowpark uses Java

2. Which of the following is the correct way to create a Snowpark session in Python?

easy

A. session = SnowflakeSession.connect(connection_parameters)

B. session = snowpark.Session.create(connection_parameters)

C. session = snowflake.snowpark.Session.builder.configs(connection_parameters).create()

D. session = snowflake.connect.snowpark(connection_parameters)

Solution

1. Step 1: Recall Snowpark session creation syntax

   The correct method uses the builder pattern: snowflake.snowpark.Session.builder.configs(...).create()

2. Step 2: Check each option's syntax

   Only session = snowflake.snowpark.Session.builder.configs(connection_parameters).create() matches the official Snowpark Python API syntax; others are incorrect or invalid.

3. Final Answer:

   session = snowflake.snowpark.Session.builder.configs(connection_parameters).create() -> Option C

4. Quick Check:

   Session builder pattern = B [OK]

Hint: Use Session.builder.configs(...).create() to start Snowpark session [OK]

Common Mistakes:

• Using incorrect method names like connect()
• Wrong module or class names
• Missing builder pattern

3. Given the following Snowpark Python code, what will df.show() display?

df = session.table('EMPLOYEES').filter(col('AGE') > 30).select('NAME', 'AGE')

medium

A. A table showing NAME and AGE of employees older than 30

B. A table showing all columns of EMPLOYEES

C. An error because filter syntax is wrong

D. A table showing only AGE column for employees younger than 30

Solution

1. Step 1: Understand the filter condition

   The filter selects rows where AGE > 30, so only employees older than 30 remain.

2. Step 2: Understand the select columns

   Only NAME and AGE columns are selected to display.

3. Final Answer:

   A table showing NAME and AGE of employees older than 30 -> Option A

4. Quick Check:

   Filter AGE > 30 and select NAME, AGE = A [OK]

Hint: Filter then select columns before show() to see filtered data [OK]

Common Mistakes:

• Confusing filter condition syntax
• Expecting all columns to show
• Thinking filter selects AGE < 30

4. What is wrong with this Snowpark Python code snippet?

df = session.table('SALES').filter(col('AMOUNT') < 1000).select('DATE', 'AMOUNT')
df.show()

medium

A. The filter condition should use > instead of <

B. The table name SALES is invalid

C. The show() method does not exist

D. The col function is not imported or defined

Solution

1. Step 1: Check usage of col function

   The col function must be imported from snowflake.snowpark.functions to be used.

2. Step 2: Verify other parts

   Table name SALES can be valid, filter condition < 1000 is valid, and show() exists on DataFrame.

3. Final Answer:

   The col function is not imported or defined -> Option D

4. Quick Check:

   Missing import for col() causes error = C [OK]

Hint: Always import col() from snowflake.snowpark.functions before filtering [OK]

Common Mistakes:

• Forgetting to import col()
• Assuming show() is missing
• Confusing filter operators

5. You want to create a new DataFrame with employee names in uppercase from the EMPLOYEES table using Snowpark for Python. Which code snippet correctly does this?

hard

A. df = session.table('EMPLOYEES').select(upper('NAME'))

B. df = session.table('EMPLOYEES').select(upper(col('NAME')).alias('NAME_UPPER'))

C. df = session.table('EMPLOYEES').select('NAME').map(str.upper)

D. df = session.table('EMPLOYEES').select(col('NAME').upper())

Solution

1. Step 1: Use upper() function with col()

   To convert a column to uppercase, use upper(col('NAME')) and alias it for clarity.

2. Step 2: Check each option's correctness

   df = session.table('EMPLOYEES').select(upper(col('NAME')).alias('NAME_UPPER')) uses correct syntax with upper() and col(). df = session.table('EMPLOYEES').select(col('NAME').upper()) tries to call upper() on a column object, which is invalid. df = session.table('EMPLOYEES').select('NAME').map(str.upper) uses map on DataFrame, which is not supported. df = session.table('EMPLOYEES').select(upper('NAME')) passes string directly to upper(), which is incorrect.

3. Final Answer:

   df = session.table('EMPLOYEES').select(upper(col('NAME')).alias('NAME_UPPER')) -> Option B

4. Quick Check:

   Use upper(col('NAME')) with alias = D [OK]

Hint: Use upper(col('COLUMN')) and alias() to transform columns [OK]

Common Mistakes:

• Calling upper() as a method on column
• Using map() on DataFrame
• Passing column name as string to upper()