nonasciitoascii_Pyspark.py

#Create a Method to handle the Non Ascii to Ascii conversion
def nonasciitoascii(unicodestring):
  return unicodestring.encode("ascii","ignore")

#Create a Sample Dataframe
from pyspark.sql.window import Window
from pyspark.sql.functions import count, col
from pyspark.sql import Row
d=[ Row(coltype='regular', value="Happy Coding"),
    Row(coltype='non ascii', value="hello aåbäcö"),
    Row(coltype='non ascii',value="6Â 918Â 417Â 712"),
    Row(coltype='non ascii',value="SAN MATEO� �?A "),
    Row(coltype='non ascii',value="SAINT-LOUIS (CANADA� � AL)")]

data = sqlContext.createDataFrame(d)
#data.show()

data = sqlContext.createDataFrame(d)

#Apply this Conversion on the Dataframe
convertedudf = udf(nonasciitoascii)
converted = data.select('coltype','value').withColumn('converted',convertedudf(data.value))
converted.show()

Hey buddy,
How did you handle the belowone.
convertedudf = udf(nonasciitoascii)
NameError: name 'udf' is not defined

I was trying to find where did you define udf or is it a predefined and you loaded any package for it?

@duginivijay The udf should be imported like from pyspark.sql.functions import udf.

Hi , i get this as the output when i run your code . How do i get the correct value for converted ? Please can you help .

+---------+--------------------+-----------+
| coltype| value| converted|
+---------+--------------------+-----------+
| regular| Happy Coding| [B@2d5d8dd4|
|non ascii| hello aåbäcö| [B@1f1744a2|
|non ascii| 6Â 918Â 417Â 712| [B@7fbc580a|
|non ascii| SAN MATEO� �?A |[B@61ef3346|
|non ascii|SAINT-LOUIS (CANA...| [B@534ecdd5|
+---------+--------------------+-----------+

@Arpit9873 I know this is a very old thread, but for anyone else who might come across this one day - You just have to cast the return object as a string like this:

def nonasciitoascii(unicodestring):
    return str(unicodestring.encode("ascii", "replace"))