printSchema.py

#   Read the source tables in Parquet format
sales_table = spark.read.parquet("./data/sales_parquet")

'''
-- Create a temporary table with a few renamings
CREATE TABLE temp_1 AS
SELECT seller_id AS the_seller,
       num_pieces_sold AS pieces,
       product_id
FROM sales_table;

-- Do aggregation on the new table
CREATE TABLE temp_2 AS
SELECT product_id,
       SUM(pieces) AS total_pieces
FROM temp_1
GROUP BY product_id;

-- Add a further column
SELECT a.*,
       1 AS fake_column
FROM temp2 a;
'''
sales_table_execution_plan = sales_table. \
    withColumnRenamed("seller_id", "the_seller"). \
    withColumnRenamed("num_pieces_sold", "pieces").\
groupBy(
    col("product_id")
).agg(
    sum("pieces").alias("total_pieces")
).withColumn("fake_column", lit(1))

#   Print Schema
sales_table_execution_plan.printSchema()