How to trim pyspark schema output

My pyspark dataframe has the following schema...

DataFrame[ExternalData: struct<provider:string,data:string,modality:array<string>>]

If I write (where sdf is my pyspark dataframe)..

sdf.schema

I get...

StructType([StructField('ExternalData', StructType([StructField('provider', StringType(), True), StructField('data', StringType(), True), StructField('modality', ArrayType(StringType(), True), True)]), True)])

How can I get just the below?

StructType([StructField('provider', StringType(), True), StructField('data', StringType(), True), StructField('modality', ArrayType(StringType(), True), True)])

There is a subtle difference in that the ExternalData StructType and StructField has been removed. The reason I need to do this is because the system I'm integrating parquet with expects parquet schema in this format, where ExternalData field and struct is passed elsewhere.

Does anyone have any advice?

Try this:

Your DataFrame schema:

root
 |-- ExternalData: struct (nullable = true)
 |    |-- provider: string (nullable = true)
 |    |-- data: string (nullable = true)
 |    |-- modality: array (nullable = true)
 |    |    |-- element: string (containsNull = true)

Selecting all the sub-columns of ExternalData to get the desired output

sdf = sdf.select("ExternalData.*")
sdf.printSchema()

Output:

root
 |-- provider: string (nullable = true)
 |-- data: string (nullable = true)
 |-- modality: array (nullable = true)
 |    |-- element: string (containsNull = true)