spark-sql集合的“条件过滤”，“合并”，“动态类型映射DataFrame”，“存储”

  List<String> basicList = new ArrayList<String>();
        basicList.add("{\"name\": \"zzq\",\"age\": 15}");
        basicList.add("{\"name\": \"zzq1\",\"age\": 25}");
        basicList.add("{\"name\": \"zzq2\",\"age\": 35}");

        List<String> scoreList = new ArrayList<String>();
        scoreList.add("{\"name\": \"zzq\",\"sex\": \"男\",\"score\": 110}");
        scoreList.add("{\"name\": \"zzq1\",\"sex\": \"女\",\"score\": 90}");
        scoreList.add("{\"name\": \"zzq2\",\"sex\": \"男\",\"score\": 70}");

        SparkConf sparkConf = new SparkConf()
                .setAppName("StudentsScore")
                .setMaster("local");

        JavaSparkContext javaSparkContext = new JavaSparkContext(sparkConf);
        SQLContext sqlContext = new SQLContext(javaSparkContext);


        JavaRDD<String> rdd_basicList = javaSparkContext.parallelize(basicList);
        JavaRDD<String> rdd_scoreList = javaSparkContext.parallelize(scoreList);


        DataFrame df_scoreList = sqlContext.read().json(rdd_scoreList);
        JavaRDD<Row> rdd_filter_score = df_scoreList.filter(df_scoreList.col("score").geq(90)).javaRDD();

        //Pair默认返回一个Tuple2，如果更多属性值的话可以在第二个参数下使用TupleX，例子如下
        JavaPairRDD<String, Tuple2<String, Long>> rdd_pair_score = rdd_filter_score.mapToPair(new PairFunction<Row, String, Tuple2<String, Long>>() {
            @Override
            public Tuple2<String, Tuple2<String, Long>> call(Row row) throws Exception {
                return new Tuple2<String, Tuple2<String, Long>>(row.getString(0), new Tuple2<String, Long>(row.getString(2), row.getLong(1)));
            }
        });

        DataFrame df_basicList = sqlContext.read().json(rdd_basicList);
        df_basicList.registerTempTable("df_basicList_table");
        StringBuilder sqlStrB = new StringBuilder();
        sqlStrB.append("select name,age from df_basicList_table where name in ( ");
        List<Tuple2<String, Tuple2<String, Long>>> local_rdd_pair_score = rdd_pair_score.collect();
        Iterator<Tuple2<String, Tuple2<String, Long>>> itr = local_rdd_pair_score.iterator();
        for (; itr.hasNext(); ) {
            Tuple2<String, Tuple2<String, Long>> currItem = itr.next();
            sqlStrB.append("\"");
            sqlStrB.append(currItem._1());
            sqlStrB.append("\"");
            if (itr.hasNext())
                sqlStrB.append(",");
        }
        sqlStrB.append(" ) ");

        DataFrame df_filter_basicList = sqlContext.sql(sqlStrB.toString());
        JavaRDD<Row> rdd_filter_basic = df_filter_basicList.javaRDD();
        JavaPairRDD<String, Long> rdd_pair_basic = rdd_filter_basic.mapToPair(new PairFunction<Row, String, Long>() {
            @Override
            public Tuple2<String, Long> call(Row row) throws Exception {
                return new Tuple2<String, Long>(row.getString(0), row.getLong(1));
            }
        });

        JavaPairRDD<String, Tuple2<Tuple2<String, Long>, Long>> all_studentsInfo = rdd_pair_score.join(rdd_pair_basic);


        //存储-------------------------------start----------------------------------
        JavaRDD<Row> row_all_studentsInfo = all_studentsInfo.map(new Function<Tuple2<String, Tuple2<Tuple2<String, Long>, Long>>, Row>() {
            @Override
            public Row call(Tuple2<String, Tuple2<Tuple2<String, Long>, Long>> v1) throws Exception {
                return RowFactory.create(v1._1(), v1._2()._1()._1(), v1._2()._1()._2(), v1._2()._2());
            }
        });

        List<StructField> fieldList = new ArrayList<StructField>();
        fieldList.add(DataTypes.createStructField("name", DataTypes.StringType, true));
        fieldList.add(DataTypes.createStructField("sex", DataTypes.StringType, true));
        fieldList.add(DataTypes.createStructField("score", DataTypes.LongType, true));
        fieldList.add(DataTypes.createStructField("age", DataTypes.LongType, true));
        StructType temp = DataTypes.createStructType(fieldList);

        DataFrame df_save = sqlContext.createDataFrame(row_all_studentsInfo, temp);

        df_save.write().save("hdfs://xxxx..........parquet");//将文件存储
        //存储-------------------------------end----------------------------------

        all_studentsInfo.foreach(new VoidFunction<Tuple2<String, Tuple2<Tuple2<String, Long>, Long>>>() {
            @Override
            public void call(Tuple2<String, Tuple2<Tuple2<String, Long>, Long>> stringTuple2Tuple2) throws Exception {
                System.out.println(">>>>>>>>>>>>" + stringTuple2Tuple2._1() + "  -- " + stringTuple2Tuple2._2()._1()._1() + "  --  " + stringTuple2Tuple2._2()._1()._2() + "  --  " + stringTuple2Tuple2._2()._2());
            }
        });