Split a row in a table into multiple rows from CSV - sql

Question

How do I split a row into multiple rows with SQL?
I have a .csv file that I've imported into a table in MS Azure Data Studio, and SQL Server Management Studio (trying both). It is a table of invoices with a header ("H") and details ("D").
I need to split each row within the table into multiple rows at a specific delimiter or column number.
I have a table that looks like this:
I need a table that looks like this, where each row is split at Column 8, and the data from the split row is inserted under the initial data.
Here is how to create the table:
CREATE TABLE InvoicesDemo2
(
[Column1] [nvarchar](1) NULL,
[Column2] [nvarchar](50) NULL,
[Column3] [nvarchar](50) NULL,
[Column4] [date] NULL,
[Column5] [date] NULL,
[Column6] [nvarchar](1) NULL,
[Column7] [money] NULL,
[Column8] [nvarchar](50) NULL,
[Column9] [nvarchar](50) NULL,
[Column10] [money] NULL
)
INSERT INTO InvoicesDemo2
VALUES ('H', 'VEN000002286', '27483387', '2023-02-14', '2023-02-14', NULL, 1840.60, 'D', '2501', 1840.60)
INSERT INTO InvoicesDemo2
VALUES ('H', 'VEN000002286', '27480005', '2023-02-13', '2023-02-13', NULL, 243.39, 'D', '2501', 243.39)
INSERT INTO InvoicesDemo2
VALUES ('H', 'VEN000002286', '27480003', '2023-02-13', '2023-02-13', NULL, 1601.64, 'D', '2501', 1601.64)
Thank you!

Answer 1

For getting match your desired format output you can try this query:
Note: This query is based on your given data and expected output.
SELECT Column1, Column2, Column3, Column4, Column5, Column6, Column7
FROM InvoicesDemo2
WHERE Column1 = 'H'
UNION ALL
SELECT 'D' as Column1,
Column9 as Column2,
Column10 as Column3,
NULL as Column4,
NULL as Column5,
NULL as Column6,
NULL as Column7
FROM InvoicesDemo2
WHERE Column1 = 'H'
ORDER BY Column2, Column1;

Answer 2

You need to union all, ordering by row number and row type e.g.
with cte as (
select Column1, Column2, Column3, Column4, Column5, Column6, Column7
, row_number () over (order by Column2, Column3) RowNumber,
, 0 RowOrder
from MyTable
union all
select Column8, Column9, Column10, null, null, null, null
, row_number () over (order by Column2, Column3) RowNumber
, 1 RowOrder
from MyTable
)
select Column1, Column2, Column3, Column4, Column5, Column6, Column7
from cte
order by RowNumber asc, RowOrder asc;
Note: I am assuming that order by Column2, Column3 gives the correct, unique, order of the rows.

Answer 3

This is just dreadfully awful and something that really should not be done this way. I would really work hard to find a solution to fix the source data so isn't so absurd. But sometimes that just isn't possible.
Here is one way to deal with this. First I added a ROW_NUMBER so we can number each row from the original data. We need that in order for the final sort. This works with the sample data that you posted. I would be careful here as the datatypes seem to be a little suspect. And I would avoid using the money datatype if at all possible.
with MergedData as
(
select h.Column1
, h.Column2
, h.Column3
, h.Column4
, h.Column5
, h.Column6
, h.Column7
, RowNum = ROW_NUMBER() over(order by Column1)
from InvoicesDemo2 h
union all
select d.Column8
, d.Column9
, d.Column10
, null
, null
, null
, null
, RowNum = ROW_NUMBER() over(order by Column1)
from InvoicesDemo2 d
)
select md.Column1
, md.Column2
, md.Column3
, md.Column4
, md.Column5
, md.Column6
, md.Column7
from MergedData md
order by md.RowNum
, md.Column1 desc

Answer 4

So, you’re going to get a lot of questions.
The main one being - why would anybody want to do this like this?
I’m having a hard time thinking up a scenario where anyone would want to start with your input and end up with your output.
Your data types for column 3 and column 10 are mismatched, which may add another wrinkle for you.
But try this - edited to account for good ideas from Dale K.
WITH LiamsCte
As (
Select
ROW_NUMBER() OVER (Order By Column2) as MyRowNumber,
Column1,
Column2,
Column3,
Column4,
Column5,
Column6,
Column7,
Column8,
Column9,
Column10
)
Select column1, column2, column3, column4, column5, column6, column7
From LiamsCte
Order by MyRowNumber
Union All
Select column8, column9, column10, null, null, null, null
From LiamsCte
Order by MyRowNumber
It may throw you an error about data types in which case you need to google how to convert column3 (the one that is a money data type) to nvarchar(50).