SQL-如何从表的XML字段中提取多个属性

如何解决SQL-如何从表的XML字段中提取多个属性

我正在尝试在一个表上运行SQL查询，该表具有一个包含XML数据的字段，但是该XML包含多个值，这些值需要转换为一个字段。请注意，该字段是XML内容，但实际的字段类型设置为nvarchar（max），而不是xml。

编辑：版本为 SQL Server 2014 Express Edition

我有一个这样的表： [有市场的客户清单]

我想在同一行（以逗号分隔）中提取“ marketCode”值：

|CompanyCode|CompanyName|MarketCode,MarketCode,etc.|Phone|

示例的预期输出（请参见屏幕截图）：

|ABC123|JOHN DEERE|AA,BB,CC,DD|555-123-000|
|DEF456|NEW HOLLLAND|AA,FF,GG,HH,KK|555-456-0000|

解决方法

样本数据

请下次提供文字而不是图片；-）

列Markets被定义为nvarchar(max)。数据以unicode（带有N前缀）插入。

create table Company2
(
    Code nvarchar(6),Name nvarchar(11),Markets nvarchar(max),Phone nvarchar(12)
);

insert into Company2 (Code,Name,Markets,Phone) values
(N'ABC123',N'JOHN DEERE',N'<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
  <license>
    <company companyCode="ABC123">
      <markets>
        <market marketCode="AA"/>
        <market marketCode="BB"/>
        <market marketCode="CC"/>
        <market marketCode="DD"/>
      </markets>
    </company>
  </license>',N'555-123-0000'),(N'DEF456',N'NEW HOLLAND',N'<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
  <license>
    <company companyCode="DEF456">
      <markets>
        <market marketCode="AA"/>
        <market marketCode="FF"/>
        <market marketCode="GG"/>
        <market marketCode="HH"/>
        <market marketCode="KK"/>
      </markets>
    </company>
  </license>',N'555-456-0000');

解决方案

无法将Markets直接投射到XML，因为nvarchar(max)编码与数据中的“ utf-8”冲突。我将转换移动到一个单独的公用表表达式（CTE，cte_convert）中，从nvarchar(max)到varchar(max)到XML。

下一个CTE（cte_parse）现在可以使用c.MarketsXML.nodes()从XML提取<market>节点到新列m.Market中。从该列中提取@marketCode属性作为所需值。

然后使用带有for xml path('')的子查询来连接值。

with cte_convert as
(
    select c.Code,c.Name,convert(XML,convert(varchar(max),c.Markets)) as MarketsXML,c.Phone
    from Company2 c
),cte_parse as
(
    select c.Code,m.Market.value('@marketCode','nvarchar(10)') as MarketCode,c.Phone
    from cte_convert c
    outer apply c.MarketsXML.nodes('/license/company/markets/market') as m(Market)
)
select  cp.Code,cp.Name,stuff(( select ',' + cp2.MarketCode as MC
                from cte_parse cp2
                where cp2.Code = cp.Code
                for xml path(''),type).value('.','nvarchar(max)'),1,'') as MarketCodes,cp.Phone
from cte_parse cp
group by cp.Code,cp.Phone;

结果

Code   Name        MarketCodes    Phone
------ ----------- -------------- ------------
ABC123 JOHN DEERE  AA,BB,CC,DD    555-123-0000
DEF456 NEW HOLLAND AA,FF,GG,HH,KK 555-456-0000

---

此原始解决方案使用了string_agg()函数，该函数从SQL Server 2017开始可用。

样本数据

注释：Markets列被定义为XML以反映其内容。

create table Company
(
    Code nvarchar(6),Markets XML,Phone nvarchar(12)
);

insert into Company (Code,Phone) values
('ABC123','JOHN DEERE','<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
  <license>
    <company companyCode="ABC123">
      <markets>
        <market marketCode="AA"/>
        <market marketCode="BB"/>
        <market marketCode="CC"/>
        <market marketCode="DD"/>
      </markets>
    </company>
  </license>','555-123-0000'),('DEF456','NEW HOLLAND','<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
  <license>
    <company companyCode="DEF456">
      <markets>
        <market marketCode="AA"/>
        <market marketCode="FF"/>
        <market marketCode="GG"/>
        <market marketCode="HH"/>
        <market marketCode="KK"/>
      </markets>
    </company>
  </license>','555-456-0000');

解决方案

with cte_parse as
(
    select c.Code,c.Phone
    from Company c
    outer apply c.Markets.nodes('/license/company/markets/market') as m(Market)
)
select cp.Code,string_agg(cp.MarketCode,',') as MarketCodes,cp.Phone;

Fiddle

,

declare @t table(CompanyMarkets nvarchar(max));
insert into @t(CompanyMarkets)
values(N'<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<license>
<company companyCode="ABC123">
<markets>
<market marketCode="AA"/>
<market marketCode="BB"/>
<market marketCode="CC"/>
<market marketCode="DD"/>
</markets>
</company>
</license>');

select
    --remove encoding ...
    try_cast(replace(CompanyMarkets,'encoding="UTF-8"','') as xml),--... or transform any prolog to a weird&harmless processing instruction 
    try_cast(concat(case when CompanyMarkets like N'<?xml%' then '<?x ' end,CompanyMarkets) as xml) 
from @t;


select *,try_cast(concat(case when CompanyMarkets like N'<?xml%' then '<?x ' end,CompanyMarkets) as xml).query('
    for $i in (data(/license/company/markets/market/@marketCode)[1],(for $k in data(/license/company/markets/market/@marketCode)[position()>1] return concat(",",$k)))
    return text {$i}
    ').value('.',--?? no spaces in marketCodes
    replace(
    try_cast(concat(case when CompanyMarkets like N'<?xml%' then '<?x ' end,CompanyMarkets) as xml).query('data(/license/company/markets/market/@marketCode)').value('.',' ','),stuff(
    try_cast(concat(case when CompanyMarkets like N'<?xml%' then '<?x ' end,CompanyMarkets) as xml).query('
    for $i in data(/license/company/markets/market/@marketCode) return text {concat(",$i)}
    ').value('.','')

from @t;