Updated on 2022-09-23 GMT+08:00

Complex Data Types

Spark SQL supports complex data types, as shown in Table 1.

Table 1 Complex data types

Data Type

Description

Syntax

ARRAY

A set of ordered fields that construct an ARRAY with the specified values. The value can be of any type and the data type of all fields must be the same.

array(<value>,<value>[, ...])

For details, see Example of ARRAY.

MAP

A group of unordered key/value pairs used to generate a MAP. The key must be native data type, but the value can be either native data type or complex data type. The type of the same MAP key, as well as the MAP value, must be the same.

map(K <key1>, V <value1>, K <key2>, V <value2>[, ...])

For details, see Example of Map.

STRUCT

Indicates a group of named fields. The data types of the fields can be different.

struct(<value1>,<value2>[, ...])

For details, see Example of STRUCT.

Restrictions

  • When a table containing fields of the complex data type is created, the storage format of this table cannot be CSV (txt).
  • If a table contains fields of the complex data type, data in CSV (txt) files cannot be imported to the table.
  • When creating a table of the MAP data type, you must specify the schema and do not support the date, short, and timestamp data types.
  • For the OBS table in JSON format, the key type of the MAP supports only the STRING type.
  • The key of the MAP type cannot be NULL. Therefore, the MAP key does not support implicit conversion between inserted data formats where NULL values are allowed. For example, the STRING type cannot be converted to other native types, the FLOAT type cannot be converted to the TIMESTAMP type, and other native types cannot be converted to the DECIMAL type.
  • Values of the double or boolean data type cannot be included in the STRUCT data type does not support the.

Example of ARRAY

Create an array_test table, set id to ARRAY<INT>, and name to STRING. After the table is created, insert test data into array_test. The procedure is as follows:

  1. Create a table.

    CREATE TABLE array_test(name STRING, id ARRAY < INT >) USING PARQUET;

  2. Run the following statements to insert test data:

    INSERT INTO array_test VALUES ('test',array(1,2,3,4));

    INSERT INTO array_test VALUES ('test2',array(4,5,6,7))

    INSERT INTO array_test VALUES ('test3',array(7,8,9,0));

  3. Query the result.

    To query all data in the array_test table, run the following statement:

    SELECT * FROM array_test;

    test3	[7,8,9,0]
    test2	[4,5,6,7]
    test	[1,2,3,4]

    To query the data of element 0 in the id array in the array_test table, run the following statement:

    SELECT id[0] FROM array_test;

    7
    4
    1

Example of Map

Create the map_test table and set score to map<STRING,INT>. The key is of the STRING type and the value is of the INT type. After the table is created, insert test data to map_test. The procedure is as follows:

  1. Create a table.

    CREATE TABLE map_test(id STRING, score map<STRING,INT>) USING PARQUET;

  2. Run the following statements to insert test data:

    INSERT INTO map_test VALUES ('test4',map('math',70,'chemistry',84));

    INSERT INTO map_test VALUES ('test5',map('math',85,'chemistry',97));

    INSERT INTO map_test VALUES ('test6',map('math',88,'chemistry',80));

  3. Query the result.

    To query all data in the map_test table, run the following statement:

    SELECT * FROM map_test;

    test6	{"chemistry":80,"math":88}
    test5	{"chemistry":97,"math":85}
    test4	{"chemistry":84,"math":70}

    To query the math score in the map_test table, run the following statement:

    SELECT id, score['Math'] FROM map_test;

    test6	88
    test5	85
    test4	70

Example of STRUCT

Create a struct_test table and set info to the STRUCT<name:STRING, age:INT> data type (the field consists of name and age, where the type of name is STRING and age is INT). After the table is created, insert test data into the struct_test table. The procedure is as follows:

  1. Create a table.

    CREATE TABLE struct_test(id INT, info STRUCT<name:STRING,age:INT>) USING PARQUET;

  2. Run the following statements to insert test data:

    INSERT INTO struct_test VALUES (8, struct('zhang',23));

    INSERT INTO struct_test VALUES (9, struct('li',25));

    INSERT INTO struct_test VALUES (10, struct('wang',26));

  3. Query the result.

    To query all data in the struct_test table, run the following statement:

    SELECT * FROM struct_test;

    8	{"name":"zhang","age":23}
    10	{"name":"wang","age":26}
    9	{"name":"li","age":25}

    Query name and age in the struct_test table.

    SELECT id,info.name,info.age FROM struct_test;

    8	zhang	23
    10	wang	26
    9	li	25