数据类型

概述

数据类型是数据的一个基本属性，用于区分不同类别的数据。不同的数据类型所占的存储空间不同，能够进行的操作也不相同。数据库中的数据存储在数据表中。数据表中的每一列都定义了数据类型，用户存储数据时，须遵从这些数据类型的属性，否则可能会出错。

大数据平台的Flink SQL与开源社区相同，支持原生数据类型、复杂数据类型和复杂类型嵌套。

原生数据类型

Flink SQL支持原生数据类型，请参见表1。

复杂数据类型

Flink SQL支持复杂数据类型和复杂类型嵌套。复杂数据类型如表2所示。

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

CREATE SOURCE STREAM car_infos (
  car_id STRING,
  address ROW<city STRING, province STRING, country STRING>,
  average_speed MAP[STRING, LONG],
  speeds ARRAY[LONG]
) 
  WITH (
    type = "dis",
    region =  "xxxx",
    channel = "dliinput",
    encode = "json"
);

CREATE SINK STREAM car_speed_infos (
  car_id STRING,
  province STRING,
  average_speed LONG,
  start_speed LONG
);

INSERT INTO car_speed_infos SELECT
   car_id,
   address.province,
   average_speed[address.city],
   speeds[1]
FROM car_infos;

复杂类型嵌套

• Json格式增强

  以Source为例进行说明，Sink的使用方法相同。

  • 支持配置Json_schema
    配置了json_schema后，可以不声明DDL中的字段，自动从json_schema中生成。使用示例如下：

    CREATE SOURCE STREAM data_with_schema WITH (
           type = "dis",
           region =  "xxxx",
           channel = "dis-in",
           encode = "json",
           json_schema = '{"definitions":{"address":{"type":"object","properties":{"street_address":{"type":"string"},"city":{"type":"string"},"state":{"type":"string"}},"required":["street_address","city","state"]}},"type":"object","properties":{"billing_address":{"$ref":"#/definitions/address"},"shipping_address":{"$ref":"#/definitions/address"},"optional_address":{"oneOf":[{"type":"null"},{"$ref":"#/definitions/address"}]}}}'
         );
    
         CREATE SINK STREAM buy_infos (
           billing_address_city STRING,
           shipping_address_state string
         ) WITH (
           type = "obs",
           encode = "csv",
           region =  "xxxx" ,
           field_delimiter = ",",
           row_delimiter = "\n",
           obs_dir = "bucket/car_infos",
           file_prefix = "over",
           rolling_size = "100m"
         );
    
         insert into buy_infos select billing_address.city, shipping_address.state from data_with_schema;

    示例数据：

    {
     "billing_address":
      {
       "street_address":"binjiang",
       "city":"hangzhou",
       "state":"zhejiang"
       },
     "shipping_address":
      {
       "street_address":"tianhe",
       "city":"guangzhou",
       "state":"guangdong"
      }
    }

  • 支持不配置json_schema也不配置json_config
    这种情况下默认用ddl中属性名当做json key来进行解析。使用示例如下：

    CREATE SOURCE STREAM car_info_data (
           billing_address Row<street_address string, city string, state string>,
           shipping_address Row<street_address string, city string, state string>,
           optional_address Row<street_address string, city string, state string>
         ) WITH (
           type = "dis",
           region =  "xxxx",
           channel = "dis-in",
           encode = "json"
    	 );
    	
    	 CREATE SINK STREAM buy_infos (
           billing_address_city STRING,
           shipping_address_state string
         ) WITH (
           type = "obs",
           encode = "csv",
           region =  "xxxx" ,
           field_delimiter = ",",
           row_delimiter = "\n",
           obs_dir = "bucket/car_infos",
           file_prefix = "over",
           rolling_size = "100m"
         );
    
         insert into buy_infos select billing_address.city, shipping_address.state from car_info_data;   

• Sink序列化支持复杂类型
  • 目前只有CSV、Json两种格式支持复杂类型。
  • ​ Json请参考•Json格式增强。
  • ​ 由于CSV没有标准的格式，所以目前暂不支持source解析，只支持sink。
  • ​ 输出格式：尽量和flink原生保持一致。

    ​ Map: {key1=Value1, key2=Value2}

    ​ Row: 平摊用逗号分隔属性，如Row(1, '2') => 1,'2'