如何将功能应用于Pyspark数据框列？

Question

I have a dataframe that looks like this:

+-----------+-------+-----------------+                                          
|A          |B      |              Num|
+-----------+-------+-----------------+
|      BAKEL|  BAKEL| 1 341 2323 01415|
|      BAKEL|  BAKEL| 2 272 7729 00307|
|      BAKEL|  BAKEL| 2 341 1224 00549|
|      BAKEL|  BAKEL| 2 341 1200 01194|
|      BAKEL|  BAKEL|1 845 0112 101159|
+-----------+-------+-----------------+

And I want an output like this:

+-----------+-------+---------------+                                          
|A          |B      |            Num|
+-----------+-------+---------------+
|      BAKEL|  BAKEL|  1341232301415|
|      BAKEL|  BAKEL|  2272772900307|
|      BAKEL|  BAKEL|  2341122400549|
|      BAKEL|  BAKEL|  2341120001194|
|      BAKEL|  BAKEL| 18450112101159|
+-----------+-------+---------------+

where the spaces in the values of the last column has been removed.

How can i do that with pyspark?

Answer 1

Use function regexp_replace() to solve this -

from pyspark.sql.functions import regexp_replace
myValues = [('BAKEL','BAKEL','1 341 2323 01415'),('BAKEL','BAKEL','2 272 7729 00307'),
            ('BAKEL','BAKEL','2 341 1224 00549'),('BAKEL','BAKEL','2 341 1200 01194'),
            ('BAKEL','BAKEL','1 845 0112 101159'),]
df = sqlContext.createDataFrame(myValues,['A','B','Num'])

df = df.withColumn('Num',regexp_replace('Num',' ',''))
#Convert String to Long (integral value)
df = df.withColumn('Num', df['Num'].cast("long"))
df.show()
+-----+-----+--------------+
|    A|    B|           Num|
+-----+-----+--------------+
|BAKEL|BAKEL| 1341232301415|
|BAKEL|BAKEL| 2272772900307|
|BAKEL|BAKEL| 2341122400549|
|BAKEL|BAKEL| 2341120001194|
|BAKEL|BAKEL|18450112101159|
+-----+-----+--------------+
df.printSchema()
root
 |-- A: string (nullable = true)
 |-- B: string (nullable = true)
 |-- Num: long (nullable = true)