How to append Psypark FOR loop output into single dataframe (spark.sql)-CodePudding

I have a pyspark for loop that uses a "customer" variable. I want to append the output of each loop iteration so that the final dataframe has all the rows output by the for loop. The code works except for the append portion. I have also tried using "union" but without success.

df = ""
df_output = []
customer=""

for customer in ['customer_abc', 'customer_xyz']:
  df = spark.sql(f"""
  SELECT sale, sum(amt) as total_sales
  FROM {customer}.salestable
  GROUP BY sale
  """).withColumn('Customer',lit(customer))
  df_output.append(df)
  
display(df_output)

CodePudding user response：

With your approach you can use a reduce :

from functools import reduce
unioned_df = reduce(lambda x,y: x.union(y) ,df_output)

Or instead of initiating df_output as a list, you can initiate it as a spark dataframe. and then keep unioning as mentioned by @Luiz

CodePudding user response：

use df = df.union(df_output) instead of df_output.append(df)