Spark Sql 映射问题

Question

Sparks2/Java8 Cassandra2 尝试从 Cassandra 读取一些数据，然后在 sparks 中通过查询运行一个组。我的 DF 中只有 2 列 转日期（日期），原点（字符串）

Dataset<Row> maxOrigindate = sparks.sql("SELECT origin, transdate, COUNT(*) AS cnt FROM origins  GROUP BY (origin,transdate) ORDER BY cnt DESC LIMIT 1"); `

获取错误：

 `Exception in thread "main" org.apache.spark.sql.AnalysisException: expression 'origins.`origin`' is neither present in the group by, nor is it an aggregate function. Add to group by or wrap in first() (or first_value)`

按问题分组已解决，删除分组中的（）如下

完整代码：（尝试获取某个来源/位置的最大翻译次数）

JavaRDD<TransByDate> originDateRDD = javaFunctions(sc).cassandraTable("trans", "trans_by_date", CassandraJavaUtil.mapRowTo(TransByDate.class))
                    .select(CassandraJavaUtil.column("origin"), CassandraJavaUtil.column("trans_date").as("transdate")).limit((long)100) ;
Dataset<Row> originDF = sparks.createDataFrame(originDateRDD, TransByDate.class);
String[] columns = originDF.columns();
System.out.println("originDF columns: "+columns[0]+" "+columns[1]) ; -> transdate origin
originDF.createOrReplaceTempView("origins");

Dataset<Row> maxOrigindate = sparks.sql("SELECT origin, transdate, COUNT(*) AS cnt FROM origins  GROUP BY origin,transdate ORDER BY cnt DESC LIMIT 1"); 
List list = maxOrigindate.collectAsList(); -> Exception here
int j = list.size();

originDF 列：转换原点

`public static class TransByDate implements Serializable {
        private String origin;
        private Date transdate;

        public TransByDate() { }

        public TransByDate (String origin, Date transdate) { 
            this.origin = origin;
            this.transdate= transdate;

        }

        public String getOrigin() { return origin; }
        public void setOrigin(String origin) { this.origin = origin; }

        public Date getTransdate() { return transdate; }
        public void setTransdate(Date transdate) { this.transdate = transdate; }

    }

架构

root
 |-- transdate: struct (nullable = true)
 |    |-- date: integer (nullable = false)
 |    |-- day: integer (nullable = false)
 |    |-- hours: integer (nullable = false)
 |    |-- minutes: integer (nullable = false)
 |    |-- month: integer (nullable = false)
 |    |-- seconds: integer (nullable = false)
 |    |-- time: long (nullable = false)
 |    |-- timezoneOffset: integer (nullable = false)
 |    |-- year: integer (nullable = false)
 |-- origin: string (nullable = true)

例外 错误执行程序：阶段 2.0 (TID 12) 中任务 0.0 中的异常 scala.MatchError: Sun Jan 01 00:00:00 PST 2012 (of class java.util.Date) 在 org.apache.spark.sql.catalyst.CatalystTypeConverters$StructConverter.toCatalystImpl(CatalystTypeConverters.scala:256) 在 org.apache.spark.sql.catalyst.CatalystTypeConverters$StructConverter.toCatalystImpl(CatalystTypeConverters.scala:251) 在 org.apache.spark.sql.catalyst.CatalystTypeConverters$CatalystTypeConverter.toCatalyst(CatalystTypeConverters.scala:103) …… 线程“主”org.apache.spark.SparkException 中的异常：作业因阶段失败而中止：阶段 2.0 中的任务 0 失败 1 次，最近一次失败：阶段 2.0 中丢失任务 0.0（TID 12，本地主机）：scala.MatchError : Sun Jan 01 00:00:00 PST 2012 (of class java.util.Date) 在 org.apache.spark.sql.catalyst.CatalystTypeConverters$StructConverter.toCatalystImpl(CatalystTypeConverters.scala:256) ... 驱动程序堆栈跟踪： 在 org.apache.spark.scheduler.DAGScheduler.org$apache$spark$scheduler$DAGScheduler$$failJobAndIndependentStages(DAGScheduler.scala:1454) 在 org.apache.spark.scheduler.DAGScheduler$$anonfun$abortStage$1.apply(DAGScheduler.scala:1442) 在 org.apache.spark.scheduler.DAGScheduler$$anonfun$abortStage$1.apply(DAGScheduler.scala:1441) ... 在 org.apache.spark.sql.Dataset$$anonfun$collectAsList$1.apply(Dataset.scala:2184) 在 org.apache.spark.sql.Dataset.withCallback(Dataset.scala:2559) 在 org.apache.spark.sql.Dataset.collectAsList(Dataset.scala:2184) 在 spark.SparkTest.sqlMaxCount(SparkTest.java:244) -> List list = maxOrigindate.collectAsList();

原因：scala.MatchError：Sun Jan 01 00:00:00 PST 2012（class java.util.Date） 在 org.apache.spark.sql.catalyst.CatalystTypeConverters$StructConverter.toCatalystImpl(CatalystTypeConverters.scala:256) 在 org.apache.spark.sql.catalyst.CatalystTypeConverters$StructConverter.toCatalystImpl(CatalystTypeConverters.scala:251)

Answer 1

您遇到了以下错误。

Caused by: scala.MatchError: Sun Jan 01 00:00:00 PST 2012 (of class java.util.Date) at 

这个错误是因为Spark sql 支持java.sql.Date 类型。请查看 Spark 文档here。也可以参考SPARK-2562。

Answer 2

将查询改为

Dataset<Row> maxOrigindate = sparks.sql("SELECT origin, 
transdate, 
COUNT(*) AS cnt FROM origins  GROUP BY origin,transdate 
ORDER BY cnt DESC LIMIT 1"); 

这会起作用。