PySpark数据类型转换异常分析
温馨提示：要看高清无码套图，请使用手机打开并单击图片放大查看。
1.问题描述
在使用PySpark的SparkSQL读取HDFS的文本文件创建DataFrame时，在做数据类型转换时会出现一些异常，如下：
1.在设置Schema字段类型为DoubleType，抛“name 'DoubleType' is not defined”异常；
2.将读取的数据字段转换为DoubleType类型时抛“Double Type can not accept object u'23' in type <type 'unicode'>”异常；
3.将字段定义为StringType类型，SparkSQL也可以对数据进行统计如sum求和，非数值的数据不会被统计。
具体异常如下：
异常一：
NameError: name 'DoubleType' is not defined
NameErrorTraceback (most recent call last)
in engine
1 schema = StructType([StructField("person_name", StringType(), False),
----> 2 StructField("person_age", DoubleType(), False)])
NameError: name 'DoubleType' is not defined
异常二：
Py4JJavaError: An error occurred while calling o152.showString.
: org.apache.spark.SparkException: Job aborted due to stage failure: Task 0 in stage 0.0 failed 4 times, most recent failure: Lost task 0.3 in stage 0.0 (TID 3, ip-172-31-26-102.ap-southeast-1.compute.internal, executor 1): org.apache.spark.api.python.PythonException: Traceback (most recent call last):
File "/opt/cloudera/parcels/SPARK2-2.1.0.cloudera1-1.cdh5.7.0.p0.120904/lib/spark2/python/lib/pyspark.zip/pyspark/worker.py", line 174, in main
process()
File "/opt/cloudera/parcels/SPARK2-2.1.0.cloudera1-1.cdh5.7.0.p0.120904/lib/spark2/python/lib/pyspark.zip/pyspark/worker.py", line 169, in process
serializer.dump_stream(func(split_index, iterator), outfile)
File "/opt/cloudera/parcels/SPARK2-2.1.0.cloudera1-1.cdh5.7.0.p0.120904/lib/spark2/python/lib/pyspark.zip/pyspark/serializers.py", line 268, in dump_stream
vs = list(itertools.islice(iterator, batch))
File "/opt/cloudera/parcels/SPARK2/lib/spark2/python/pyspark/sql/session.py", line 509, in prepare
verify_func(obj, schema)
File "/opt/cloudera/parcels/SPARK2-2.1.0.cloudera1-1.cdh5.7.0.p0.120904/lib/spark2/python/lib/pyspark.zip/pyspark/sql/types.py", line 1360, in _verify_type
_verify_type(v, f.dataType, f.nullable)
File "/opt/cloudera/parcels/SPARK2-2.1.0.cloudera1-1.cdh5.7.0.p0.120904/lib/spark2/python/lib/pyspark.zip/pyspark/sql/types.py", line 1324, in _verify_type
raise TypeError("%s can not accept object %r in type %s" % (dataType, obj, type(obj)))
TypeError: DoubleType can not accept object u'23' in type <type 'unicode'>
at org.apache.spark.api.python.PythonRunner$$anon$1.read(PythonRDD.scala:193)
at org.apache.spark.api.python.PythonRunner$$anon$1.<init>(PythonRDD.scala:234)
at org.apache.spark.api.python.PythonRunner.compute(PythonRDD.scala:152)
at org.apache.spark.api.python.PythonRDD.compute(PythonRDD.scala:63)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:323)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:287)
at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:323)
2.解决方法
- 异常一：
NameError: name 'DoubleType' is not defined问题原因：
由于在Python代码中未引入pyspark.sql.types为DoubleType的数据类型导致
解决方法：
from pyspark.sql.types import *或者
from pyspark.sql.types import Row, StructField, StructType, StringType, IntegerType, DoubleType
- 异常二：
TypeError: DoubleType can not accept object u'23' in type <type 'unicode'>
at org.apache.spark.api.python.PythonRunner$$anon$1.read(PythonRDD.scala:193)
at org.apache.spark.api.python.PythonRunner$$anon$1.<init>(PythonRDD.scala:234)
at org.apache.spark.api.python.PythonRunner.compute(PythonRDD.scala:152)
at org.apache.spark.api.python.PythonRDD.compute(PythonRDD.scala:63)问题原因：
由于Python默认的字符编码集为unicode，如果需要将字段转为Double类型，则需要进行转换。
解决方法：
# Schema with two fields - person_name and person_age
schema = StructType([StructField("person_name", StringType(), False),
StructField("person_age", DoubleType(), False)])
lines = spark.read.text("/tmp/test/").rdd \
.map(lambda x:x[0].split(",")) \
.map(lambda x: (x[0], float(x[1])))
增加标红部分代码，将需要转换的字段转换为float类型。
转换完成后代码正常运行。
SparkSQL和DataFrame支持的数据类型参考官网： http://spark.apache.org/docs/latest/sql-programming-guide.html#data-types
3.总结
1.在上述测试代码中，如果x1列的数据中有空字符串或者非数字字符串则会导致转换失败，因此在指定字段数据类型的时候，如果数据中存在“非法数据”则需要对数据进行剔除，否则不能正常执行。
测试数据如下：
代码执行报错如下：
Py4JJavaError: An error occurred while calling o291.showString.
: org.apache.spark.SparkException: Job aborted due to stage failure: Task 0 in stage 0.0 failed 4 times, most recent failure: Lost task 0.3 in stage 0.0 (TID 3, ip-172-31-26-102.ap-southeast-1.compute.internal, executor 1): org.apache.spark.api.python.PythonException: Traceback (most recent call last):
File "/opt/cloudera/parcels/SPARK2-2.1.0.cloudera1-1.cdh5.7.0.p0.120904/lib/spark2/python/lib/pyspark.zip/pyspark/worker.py", line 174, in main
process()
File "/opt/cloudera/parcels/SPARK2-2.1.0.cloudera1-1.cdh5.7.0.p0.120904/lib/spark2/python/lib/pyspark.zip/pyspark/worker.py", line 169, in process
serializer.dump_stream(func(split_index, iterator), outfile)
File "/opt/cloudera/parcels/SPARK2-2.1.0.cloudera1-1.cdh5.7.0.p0.120904/lib/spark2/python/lib/pyspark.zip/pyspark/serializers.py", line 268, in dump_stream
vs = list(itertools.islice(iterator, batch))
File "<ipython-input-1-29d7e609de57>", line 1, in <lambda>
ValueError: invalid literal for float(): 23q
at org.apache.spark.api.python.PythonRunner$$anon$1.read(PythonRDD.scala:193)