sparksql操作csv排序时如何进行Int类型的排序, 而不是String类型的排序

Question

Dataset<Row> df = spark.read().format("csv").load("C:\\develop\\intellij-workspace\\SparkSqlDemos\\resources\\down.csv");
df.createOrReplaceTempView("down");
Dataset<Row> dfSQL = spark.sql("SELECT * FROM down order by _c3");
//无法使用CAST关键字
//Dataset<Row> dfSQL = spark.sql("SELECT * FROM down order by CAST(_c3 as SIGNED)");

如何实现按照Int类型排序

Answer 1

.option("inferSchema", true)

见: org.apache.spark.sql.DataFrameReader 文档, 通过修改option方法中的inferSchema选项为true, 推到数据类型为数值, 而不是一律默认为String