//通过sqlContext读取json文件创建DataFrame

DataFrame dataFrame=sqlContext.read().json("src/main/resources/datafromcreate.txt");
//通过两种方式加载json文件
//sqlContext.read().json("src/main/resources/datafromcreate.txt");
sqlContext.read().format("json").load("src/main/resources/datafromcreate.txt"）；

//创建sqlContext

SQLContext sqlContext=new SQLContext(context);

//使用程序构建DataFrame的元数据

StructType structType=new StructType(new StructField[]{

                DataTypes.createStructField("id", DataTypes.IntegerType, false),

                DataTypes.createStructField("name", DataTypes.StringType, true),

                DataTypes.createStructField("age", DataTypes.IntegerType, true)

        });

//创建studentsRdd

JavaRDD<Row> studentsRdd=context.textFile("src/main/resources/students.txt").map(new Function<String, Row>() {

            private static final long serialVersionUID = 1L;

            public Row call(String line) throws Exception {

                String[] words=line.split(" ");

                return RowFactory.create(Integer.parseInt(words[]),words[],Integer.parseInt(words[]));

            }

        });

//使用动态构建的元数据创建DataFrame

DataFrame studentDataFrame= sqlContext.createDataFrame(studentsRdd, structType);

//封装为Student JavaRDD

JavaRDD<Student> students=context.textFile("src/main/resources/students.txt").map(new Function<String, Student>() {

            private static final long serialVersionUID = 1L;

            public Student call(String line) throws Exception {

                String[] words=line.split(" ");

                return new Student(Integer.parseInt(words[]), words[], Integer.parseInt(words[]));

            }

        });

//使用反射技术,将javaRdd转换为DataFrame，使用javabean的属性定义DataFrame的元数据

DataFrame studentDataFrame= sqlContext.createDataFrame(students, Student.class);

DataFrame goodStudentDF=hiveContext.table("sqark.good_student_info");

//加载parquet文件为DataFrame

 DataFrame usersDF=sqlContext.read().parquet("src/main/resources/parquet/users.parquet");

/**

 * root

    |-- name: string (nullable = true)

    |-- age: long (nullable = true)

*/

usersDF.printSchema();   

/**

* 加载区别表中的数据是会自动推断分区列，
　　users.parquet只有两个字段name,age;
　　female和coutry为分区字段

*/

usersDF=sqlContext.read().parquet("src/main/resources/parquet/female=male/coutry=US/users.parquet");

/**

* root

|-- name: string (nullable = true)

|-- age: long (nullable = true)

|-- female: string (nullable = true)

|-- coutry: string (nullable = true)

*/

usersDF.printSchema();

/**

* megerschema/idandage.txt 中的内容只有id和age两个属性

* megerschema/idandname.txt 中的内容只有id和name两个属性

* 合并以后的元素为id,name,age三个属性

*/

DataFrame personDF=sqlContext.read().option("mergeSchema", "true").format("json").load("src/main/resources/megerschema");

personDF.printSchema();

//将DataFrame,默认以parquet类型进行保存，可以使用format修改保存的文件格式

personDF.write().save("src/main/resources/output/persons");

//将DataFrame使用json格式保存
personDF.write().format("json").save("src/main/resources/output/persons");