因此,如果我先执行df = sql_context.read.csv("test_data_2019-01-01.csv", header=False)
,然后执行df.write.parquet("test_data_2019-01-01.parquet")
,则一切正常,但是如果我在header=True
中设置了read.csv
,然后尝试编写,则会出现以下错误:
An error occurred while calling o522.parquet.
: org.apache.spark.sql.AnalysisException: Attribute name " M6_Debt_Review_Ind" contains invalid character(s) among " ,;{}()\n\t=". Please use alias to rename it.
我需要那些标题,否则列名称如下所示:
[Row(_c0='foo', _c1='bar', _c2='bla', _c3='bla2', _c4='blabla', _c5='bla3', _c6=' bla4'),
Row(_c0='1161057', _c1='57793622', _c2='6066807', _c3='2017-01-31', _c4='2017-01-31', _c5='1', _c6='0'),
Row(_c0='1177047', _c1='58973984', _c2='4938603', _c3='2017-02-28', _c4='2017-02-28', _c5='0', _c6='0')]
而不是
[Row(foo='1161057', bar='57793622', bla='6066807', bla2='2017-01-31', blabla='2017-01-31', bla3='1', M6_Debt_Review_Ind='0'),
Row(foo='1177047', bar='58973984', bla='4938603', bla2='2017-02-28', blabla='2017-02-28', bla3='0', bla4='0')]
谢谢您的任何建议。
没关系,愚蠢的错误。列名称中有一个空格。