When you run a Python function in a Python cell, it executes in the local Python environment of the notebook. However, when you call a Python function from a SQL cell, it runs as a UDF within the Spark execution environment. 

You need to define the function as a UDF explicitly if you want to use it within SQL cells. This involves using the pyspark.sql.functions.udf decorator to register the function

# Register the UDF with Spark
spark.udf.register("get_useragent_string", get_useragent_string_udf)

After registering the function as a UDF, you can call it from a SQL cell:

%sql
SELECT get_useragent_string('xxx', 49738, '104.16.184.241', 1730235533)