将具有多个分隔符的字符串解析为列

Question

我想将字符串拆分成列。

我的专栏应该是：

account_id, resource_type, resource_name

我有一个 JSON 文件源，我一直在尝试通过 ADF 数据流进行解析。这对我不起作用，因此我将数据展平并将其带入 SQL Server（如果有人可以告诉我如何通过 ADF 或 SQL 解析值，我愿意）。请检查底部的 JSON 文件。

使用此代码查询我正在处理的数据。

 CREATE TABLE test.test2
 (
     resource_type nvarchar(max) NULL
 )

 INSERT INTO test.test2 ([resource_type]) 
 VALUES 
     ('account_id:224526257458,resource_type:buckets,resource_name:camp-stage-artifactory'),
     ('account_id:535533456241,resource_type:buckets,resource_name:tni-prod-diva-backups'),
     ('account_id:369798452057,resource_type:buckets,resource_name:369798452057-s3-manifests'),
     ('account_id:460085747812,resource_type:buckets,resource_name:vessel-incident-report-nonprod-accesslogs')

我应该能够在 SQL Server 中查询的输出应该是这样的：

account_id	resource_type	resource_name
224526257458	buckets	camp-stage-artifactory
535533456241	buckets	tni-prod-diva-backups

等等。

如果需要，请帮助我并要求澄清。提前致谢。

编辑：

源 JSON 格式：

{
    "start_date": "2021-12-01 00:00:00+00:00",
    "end_date": "2021-12-31 23:59:59+00:00",
    "resource_type": "all",
    "records": [
        {
            "directconnect_connections": [
                "account_id:227148359287,resource_type:directconnect_connections,resource_name:'dxcon-fh40evn5'",
                "account_id:401311080156,resource_type:directconnect_connections,resource_name:'dxcon-ffxgf6kh'",
                "account_id:401311080156,resource_type:directconnect_connections,resource_name:'dxcon-fg5j5v6o'",
                "account_id:227148359287,resource_type:directconnect_connections,resource_name:'dxcon-fgvfo1ej'"
            ]
        },
        {
            "virtual_interfaces": [
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:'dxvif-fgvj25vt'",
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:'dxvif-fgbw5gs0'",
                "account_id:401311080156,resource_type:virtual_interfaces,resource_name:'dxvif-ffnosohr'",
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:'dxvif-fg18bdhl'",
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:'dxvif-ffmf6h64'",
                "account_id:390251991779,resource_type:virtual_interfaces,resource_name:'dxvif-fgkxjhcj'",
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:'dxvif-ffp6kl3f'"
            ]
        }
    ]
}

Answer 1

由于您没有有效的 JSON 字符串并且不想从事字符串操作的业务......也许这会有所帮助。

Select B.*
 From  test2 A
 Cross Apply ( Select account_id    = max(case when value like 'account_id:%'    then stuff(value,1,11,'') end )
                     ,resource_type = max(case when value like 'resource_type:%' then stuff(value,1,14,'') end )
                     ,resource_name = max(case when value like 'resource_name:%' then stuff(value,1,14,'') end )
                from  string_split(resource_type,',') 
             )B

结果

account_id      resource_type   resource_name
224526257458    buckets         camp-stage-artifactory
535533456241    buckets         tni-prod-diva-backups
369798452057    buckets         369798452057-s3-manifests
460085747812    buckets         vessel-incident-report-nonprod-accesslogs

Answer 2

很遗憾，数组中的值不是有效的 JSON。您可以通过在开头/结尾添加{} 并在: 和, 的任一侧添加" 来修补它们。

DECLARE @json nvarchar(max) = N'{
    "start_date": "2021-12-01 00:00:00+00:00",
    "end_date": "2021-12-31 23:59:59+00:00",
    "resource_type": "all",
    "records": [
        {
            "directconnect_connections": [
                "account_id:227148359287,resource_type:directconnect_connections,resource_name:''dxcon-fh40evn5''",
                "account_id:401311080156,resource_type:directconnect_connections,resource_name:''dxcon-ffxgf6kh''",
                "account_id:401311080156,resource_type:directconnect_connections,resource_name:''dxcon-fg5j5v6o''",
                "account_id:227148359287,resource_type:directconnect_connections,resource_name:''dxcon-fgvfo1ej''"
            ]
        },
        {
            "virtual_interfaces": [
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:''dxvif-fgvj25vt''",
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:''dxvif-fgbw5gs0''",
                "account_id:401311080156,resource_type:virtual_interfaces,resource_name:''dxvif-ffnosohr''",
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:''dxvif-fg18bdhl''",
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:''dxvif-ffmf6h64''",
                "account_id:390251991779,resource_type:virtual_interfaces,resource_name:''dxvif-fgkxjhcj''",
                "account_id:227148359287,resource_type:virtual_interfaces,resource_name:''dxvif-ffp6kl3f''"
            ]
        }
    ]
}';


SELECT
  j4.account_id,
  j4.resource_type,
  TRIM('''' FROM j4.resource_name) resource_name
FROM OPENJSON(@json, '$.records') j1
CROSS APPLY OPENJSON(j1.value) j2
CROSS APPLY OPENJSON(j2.value) j3
CROSS APPLY OPENJSON('{"' + REPLACE(REPLACE(j3.value, ':', '":"'), ',', '","') + '"}')
  WITH (
    account_id bigint,
    resource_type varchar(20),
    resource_name varchar(100)
  ) j4;

db<>fiddle

对OPENJSON 的前三个调用没有架构，因此结果集是三列：keyvalue 和type。对于数组（j1 和 j3），key 是数组的索引。对于单个对象（j2），key 是每个属性名称。