在SpringBoot中通过maven来做包管理构建，有几个地方需要注意一下的，需要解决包之间的冲突，否则运行时会报错：
（1）sparkSQL中需要先排除两个包：
1 <dependency> 2 <groupId>org.apache.spark</groupId> 3 <artifactId>spark-sql_2.11</artifactId> 4 <version>${spark.version}</version> 5 <exclusions> 6 <exclusion> 7 <groupId>org.codehaus.janino</groupId> 8 <artifactId>janino</artifactId> 9 </exclusion> 10 <exclusion> 11 <groupId>org.codehaus.janino</groupId> 12 <artifactId>commons-compiler</artifactId> 13 </exclusion> 14 </exclusions> 15 </dependency>
（2）重新引入:
1 <dependency> 2 <groupId>org.codehaus.janino</groupId> 3 <artifactId>commons-compiler</artifactId> 4 <version>2.7.8</version> 5 </dependency> 6 7 <dependency> 8 <groupId>org.codehaus.janino</groupId> 9 <artifactId>janino</artifactId> 10 <version>2.7.8</version> 11 </dependency>
ok,准备工作做完之后，开始代码层面得工作：
主要给大家演示的场景是将json字符串转换成临时表，然后通过sparkSQL操作临时表，非常简单方便：
1 public class SparkJsonSQL { 2 3 public void Exec(){ 4 SparkConf conf = new SparkConf(); 5 conf.setMaster("local[2]").setAppName("jsonRDD"); 6 JavaSparkContext sc = new JavaSparkContext(conf); 7 SQLContext sqlContext = new SQLContext(sc); 8 9 JavaRDD<String> nameRDD = sc.parallelize(Arrays.asList( 10 "{\"name\":\"zhangsan\",\"age\":\"18\"}", 11 "{\"name\":\"lisi\",\"age\":\"19\"}", 12 "{\"name\":\"wangwu\",\"age\":\"20\"}" 13 )); 14 JavaRDD<String> scoreRDD = sc.parallelize(Arrays.asList( 15 "{\"name\":\"zhangsan\",\"score\":\"100\"}", 16 "{\"name\":\"lisi\",\"score\":\"200\"}", 17 "{\"name\":\"wangwu\",\"score\":\"300\"}" 18 )); 19 20 Dataset<Row> namedf = sqlContext.read().json(nameRDD); 21 Dataset<Row> scoredf = sqlContext.read().json(scoreRDD); 22 namedf.registerTempTable("name"); 23 scoredf.registerTempTable("score"); 24 25 Dataset<Row> result = sqlContext.sql("select name.name,name.age,score.score from name,score where name.name = score.name"); 26 //Dataset<Row> result = sqlContext.sql("select * from name"); 27 result.show(); 28 result.foreach(x ->System.out.print(x)); 29 sc.stop(); 30 } 31 }
我们将程序运行起来看看效果：
