问题
I am trying to join two tables by first converting a comma separated values from a column "SupplierId" which I am doing successfully. However, the issue comes in when I try joining to another table 'Vendors' with the Supplier names via a foreign key 'DCLink'.
This is what I mean:
The select statement for the Original Table,
SELECT InquiryId, SupplierId FROM Procure_InquiryDetails
Gives this result
InquiryId SupplierId
1 2,3
2 175
3 170,280
5
7 12
8 5,9
I am able to split the columns from SupplierId using this sql statement
;WITH CTE
AS
(
SELECT InquiryId,
[xml_val] = CAST('<t>' + REPLACE(SupplierId,',','</t><t>') + '</t>' AS XML)
FROM Procure_InquiryDetails
)
SELECT InquiryId,
[SupplierId] = col.value('.','VARCHAR(100)')
FROM CTE
CROSS APPLY [xml_val].nodes('/t') CA(col)
and get these results
InquiryId SupplierId
1 2
1 3
2 175
3 170
3 280
5
7 12
8 5
8 9
When I apply this bit of code to join the InquiryDetails table to the Vendor Table on supplier Name however,
;WITH CTE
AS
(
SELECT InquiryId,
[xml_val] = CAST('<t>' + REPLACE(SupplierId,',','</t><t>') + '</t>' AS XML),
Vendor.Name
FROM Procure_InquiryDetails inner join Vendor
on ',' + Procure_InquiryDetails.SupplierId + ',' like '%,' + cast(Vendor.DCLink as nvarchar(20)) + ',%'
)
SELECT InquiryId, Name,
[SupplierId] = col.value('.','VARCHAR(100)')
FROM CTE
CROSS APPLY [xml_val].nodes('/t') CA(col)
It gives me this very inconvenient result:
InquiryId Name SupplierId
----------- ------------------------------------------------------------------------------------------------------------------------------------------------------ ----------------------------------------------------------------------------------------------------
1 Accesskenya Group Ltd 2
1 Accesskenya Group Ltd 3
1 Aquisana Ltd 2
1 Aquisana Ltd 3
2 TOYOTA KENYA 175
3 Institute of Chartered Shipbrokers ICS-USD 170
3 Institute of Chartered Shipbrokers ICS-USD 280
7 CMA CGM Kenya Ltd 12
8 Aon Kenya Insurance Brokers Ltd 5
8 Aon Kenya Insurance Brokers Ltd 9
8 Bill investments ltd 5
8 Bill investments ltd
I would wish for the join statement to show and flow as the original select statement.
I am stuck and I cannot seem to figure out where I am going wrong. Any pointers in the right direction?
回答1:
Assuming you're using SQL Server 2016, you can use string_split() to parse out your CSV column (aside: comma-separated values in a field is a sign of a poor data model) without resorting to a CTE or XML methods.
select I.inquiry_id, sup.value,V.Name
from Procure_InquiryDetails I
CROSS APPLY string_split(I.supplier_value,',') sup
join Vendor v on v.DCLink = sup.value
回答2:
You've forgotten to supply expected results, so this is a stab in the dark, however, what's wrong with splitting your string and using the results with a JOIN:
SELECT {Needed Columns}
FROM dbo.Procure_InquiryDetails PID
CROSS APPLY STRING_SPLIT(PID.SupplierId,',') SS
JOIN dbo.Vendor V ON SS.[value] = V.SupplierID;
Ideally, however, you shouldn't be storing delimited data in your RDBMS. Considering switching to a proper normalised many-to-many relationship structure.
If you're still on SQL Server 2008 (to which I would highly recommend you upgrade), then you can use delimitedsplit8k, or on 2012/2014 you can use delimitedsplit8k_lead.
来源：https://stackoverflow.com/questions/54065989/how-to-join-comma-separated-column-values-with-another-table-as-rows