从逗号分隔的 GUID 中提取逗号分隔的值

Question

I have a column in table T1 named Categories , which contains GUIDs in XML. I am able to extract the GUIDs in a comma-separated form using the below query.

SELECT
  Row, ID, Name, City,
  Category = STUFF(
    (
      SELECT ',' + t.c.value('.', 'nvarchar(max)') 
      FROM dbo.T1 t1  
      OUTER APPLY t1.Categories.nodes('root/string') as t(c)
      WHERE t1.ID = t2.ID FOR XML PATH('')
    ), 1, 1, ''
  )
FROM
  dbo.T1 t2

I have another table T2 , which contains the names of the Categories. I now want to use these comma-separated GUIDs to go and fetch their corresponding Name from T2 .

What changes do I need to make in my SELECT statement to write a LEFT OUTER JOIN which takes this comma-separated GUIDs and returns comma-separated names from T2 .

T2 looks something like this:

Answer 1

I would join the category name table before concatenating the values to avoid another iteration of splitting and concatenating.

Sample data

create table xmlData
(
  id int,
  data xml
);

insert into xmlData (id, data) values
(1,'
<root>
  <guid>5d8547aa-e1e7-4f69-88a2-655879531582</guid>
  <guid>78555c5d-e39f-48f3-a148-30161b0fb995</guid>
</root>
'),
(2,'
<root>
  <guid>5d8547aa-e1e7-4f69-88a2-655879531582</guid>
  <guid>f58177f6-63c8-4985-baa8-2db05248f13f</guid>
</root>
'),
(3,'
<root>
  <guid>5d8547aa-e1e7-4f69-88a2-655879531582</guid>
  <guid>d8f9b789-6d60-4688-9d91-c0f8b1df5319</guid>
</root>
');

create table categoryName
(
  guid uniqueidentifier,
  name nvarchar(20)
);

insert into categoryName (guid, name) values
('5d8547aa-e1e7-4f69-88a2-655879531582', 'Alpha'),
('78555c5d-e39f-48f3-a148-30161b0fb995', 'Beta'),
('f58177f6-63c8-4985-baa8-2db05248f13f', 'Gamma'),
('d8f9b789-6d60-4688-9d91-c0f8b1df5319', 'Delta');

Solution

Two versions because the SQL Server version is not specified in the question tags... The string_agg() function is available starting from SQL Server 2017.

With string_agg()

select xd.id,
       string_agg(cn.name, ',') as 'category_names'
from xmlData xd
cross apply xd.data.nodes('root/guid') g(guid)
join categoryName cn
  on cn.guid = g.guid.value('.', 'nvarchar(36)')
group by xd.id
order by xd.id;

Without string_agg()

select xd.id,
       stuff( ( select ',' + cn.name
                from xmlData xd2
                cross apply xd.data.nodes('root/guid') g(guid)
                join categoryName cn
                  on cn.guid = g.guid.value('.', 'nvarchar(36)')
                where xd2.id = xd.id
                for xml path('') ), 1, 1, '' ) as 'category_names'
from xmlData xd
order by xd.id;

Result

id  category_names
--  --------------
1   Alpha,Beta
2   Alpha,Gamma
3   Alpha,Delta

Fiddle to see things in action.