Difference between sc.textFile and spark.read.text in Spark

问题

I am trying to read a simple text file into a Spark RDD and I see that there are two ways of doing so :

from pyspark.sql import SparkSession
spark = SparkSession.builder.master("local[*]").getOrCreate()
sc = spark.sparkContext
textRDD1 = sc.textFile("hobbit.txt")
textRDD2 = spark.read.text('hobbit.txt').rdd

then I look into the data and see that the two RDDs are structured differently

textRDD1.take(5)

['The king beneath the mountain',
 'The king of carven stone',
 'The lord of silver fountain',
 'Shall come unto his own',
 'His throne shall be upholden']

textRDD2.take(5)

[Row(value='The king beneath the mountain'),
 Row(value='The king of carven stone'),
 Row(value='The lord of silver fountain'),
 Row(value='Shall come unto his own'),
 Row(value='His throne shall be upholden')]

Based on this, all subsequent processing has to be changed to reflect the presence of the 'value'

My questions are

• What is the implication of using these two ways of reading a text file?
• Under what circumstances should we use which method?

回答1:

To answer (a),

sc.textFile(...) returns a RDD[String]

textFile(String path, int minPartitions)

Read a text file from HDFS, a local file system (available on all nodes), or any Hadoop-supported file system URI, and return it as an RDD of Strings.

spark.read.text(...) returns a DataSet[Row] or a DataFrame

text(String path)

Loads text files and returns a DataFrame whose schema starts with a string column named "value", and followed by partitioned columns if there are any.

For (b), it really depends on your use case. Since you are trying to create a RDD here, you should go with sc.textFile. You can always convert a dataframe to a rdd and vice-versa.

来源：https://stackoverflow.com/questions/52665353/difference-between-sc-textfile-and-spark-read-text-in-spark