Ufdjrw

I have the following data frame (called df) with columns item_name and item_level:

 item_name item_level 
----------------------------
 Item1 1
 Item2 2
 Item3 2
 Item4 3

I would like to create a new column which produces indentdation of the items, depending on their level. To do that, I would like to multiply the item_level by the string '---', with the idea that when I do that the string gets concatenated with itself as many times as the value of the integer I am multiplying the string with.

My desired result is something like this:

 item_name item_level new_column
------------------------------------------------
 Item1 1 ---Item1
 Item2 2 ------Item2
 Item3 2 ------Item3
 Item4 3 ---------Item4

In pyspark when I write the following command, the created column contains only null values:

from pyspark.sql import functions as F
df = df.withColumn('new_column',F.concat(F.lit(df.item_level*'---'),df.item_name))

The null values seem to come from the multiplication of the integers with the string. The concat function seems to work properly. For instance, the following works:

df = df.withColumn('new_column',F.concat(df.item_name,df.item_name))

I also tried few other things. If I use a constant number to multiply the string, the resulting string is displayed as wished:

number = 3
df = df.withColumn('new_column', F.lit(number*'---'))

Furthermore, adding the '---' string first in a column (with identical rows '---'), and then multiplying that column with the item_level column gives null values as well:

df = df.withColumn('padding',F.lit('---'))
df = df.withColumn('test',df.padding*df.item_name)

If I use pandas, however, this last piece of code does what I want. But I need to do this in pyspark.

There is a function pyspark.sql.functions.repeat that:

Repeats a string column n times, and returns it as a new string column.

Concatenate the result of repeat with the item_name as you were doing in your code. The only wrinkle is that you need to use pyspark.sql.functions.expr in order to pass a column value as an argument to a spark function.

from pyspark.sql.functions import concat, expr

df.withColumn(
 "new_column", 
 concat(expr("repeat('---', item_level)"), "item_name")
).show()
#+---------+----------+--------------+
#|item_name|item_level| new_column|
#+---------+----------+--------------+
#| Item1| 1| ---Item1|
#| Item2| 2| ------Item2|
#| Item3| 2| ------Item3|
#| Item4| 3|---------Item4|
#+---------+----------+--------------+

Note that show() will right justify the output that is displayed, but the underlying data is as you desired.