Display Customer and all of their order dates on single row - sql

Question

I have a customers table and an orders table. I want to display the customer and all of his/her order dates on one row, rather than multiple rows. Here is what I have and what I'm looking for:
Basic code to get results:
select customerid, name, orderdate
from customer_table c inner join
order_table o
on c.customerid = o.customerid

Answer 1

this will work at the most you cant show it on different columns having nulls:
select customer_id,name,LISTAGG(orderdate, ', ') WITHIN GROUP (ORDER BY orderdate)
from(select customerid, name, orderdate
from customer_table c inner join
order_table o
on c.customerid = o.customerid );

Answer 2

You can use the following because you're bound by the 12 order limit. If you expand to an unknown number of orders with no upper limit, then you would need to use dynamic SQL and even then it would be tricky, because you would also need to dynamically create unique column names.
This query batches by customer and sets the order values. They will be in dated order and set as NULL if there's no more orders. It kinda assumes that at least one customer has 12 orders. You'll get a column of all NULLS if that's not the case
IF OBJECT_ID('tempdb..#Results') IS NOT NULL DROP TABLE #Results
IF OBJECT_ID('tempdb..#sortedRows') IS NOT NULL DROP TABLE #SortedRows
DECLARE #CustomerList TABLE(CustomerID INT, RowNo INT);
INSERT INTO #CustomerList SELECT DISTINCT CustomerID, ROW_NUMBER() OVER(ORDER BY CustomerID) RowNo FROM Customer_Table (NOLOCK)
DECLARE #Count INT = (SELECT COUNT(DISTINCT CustomerID) RowNumber FROM #CustomerList)
DECLARE #Counter INT = 0
DECLARE #CustToProcess INT
CREATE TABLE #Results(CustomerID INT, [Name] VARCHAR(50), OrderDate1 DATETIME,
OrderDate2 DATETIME, OrderDate3 DATETIME, OrderDate4 DATETIME, OrderDate5 DATETIME,
OrderDate6 DATETIME, OrderDate7 DATETIME, OrderDate8 DATETIME, OrderDate9 DATETIME,
OrderDate10 DATETIME, OrderDate11 DATETIME, OrderDate12 DATETIME)
INSERT INTO #Results(CustomerID, Name) SELECT DISTINCT CustomerID, Name FROM Customer_Table
SELECT ROW_NUMBER() OVER(PARTITION BY c.CustomerID ORDER BY OrderDate) RowNo,
c.CustomerID, c.Name, t.OrderDate INTO #SortedRows
FROM Customer_Table c (NOLOCK) JOIN Order_Table t ON c.CustomerID = t.CustomerID
WHILE #Counter < #Count
BEGIN
SET #Counter += 1
SET #CustToProcess = (SELECT CustomerID FROM #CustomerList WHERE RowNo = #Counter)
PRINT #CustToProcess
SELECT * INTO #RowsForProcessing FROM #SortedRows WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate1 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 1) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate2 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 2) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate3 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 3) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate4 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 4) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate5 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 5) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate6 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 6) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate7 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 7) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate8 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 8) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate9 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 9) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate10 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 10) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate11 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 11) WHERE CustomerID = #CustToProcess
UPDATE #Results SET OrderDate12 = (SELECT OrderDate FROM #RowsForProcessing WHERE Rowno = 12) WHERE CustomerID = #CustToProcess
DROP Table #RowsForProcessing
END
SELECT * FROM #Results

Answer 3

Try this if you use MS SQL Server:
-- 1st, get the number of columns
declare #columnnumber int = 1
select #columnnumber = max(a.count)
from (
select c.cid,c.name, count(o.orderdate) as count
from
customer_table c
join order_table o
on c.cid = o.cid
group by c.cid,c.name)a
print #columnnumber
-- Compose the column names for Pivot
declare #columnname varchar(max) = ''
declare #int int = 1
while #int <= #columnnumber
begin
set #columnname = #columnname + '[date' + cast(#int as varchar(10))+ '],'
set #int = #int + 1
end
set #columnname = '('+left(#columnname,len(#columnname)-1)+')'
print #columnname
--Pivot !!! + Dynamic SQL
declare #str varchar(max)
set #str =
'SELECT *
FROM
(SELECT c.cid,c.name, o.orderdate,concat(''date'',row_number() over (partition by c.cid,c.name order by o.orderdate)) rnk
FROM customer_table c
join order_table o
on c.cid = o.cid) AS s
PIVOT
(
min(s.orderdate)
FOR s.rnk IN '+ #columnname+
' ) AS PivotTable'
print #str
execute (#str)
Please change the column name. I used cid as your customerid.
Output:
cid name date1 date2 date3
12 John 2017-03-04 2017-05-26 2017-12-01
4 Nancy 2017-02-01 NULL NULL

Answer 4

Like others said, it's in principle impossible (in pure SQL) to generate this not knowing how many orders you have for a single customer.
I like #nikhil-sugandh's answer, it works great if you are OK with having all orders comma-separated in a single column.
If you insist on having multiple columns, you can build on that answer, by replacing LISTAGG with ARRAY_AGG and postprocessing it. It will be MUCH more efficient than e.g. the proposed solution with multiple joins. You can also use ARRAY_SLICE to handle cases when there are more orders than you are prepared for.
Example (note, I added an extra order to demonstrate handling more-than-expected orders
create or replace table customer_table(customerId int, name varchar)
as select * from values
(12,'John'),(4,'Nancy');
create or replace table order_table(orderId int, customerId int, orderDate date)
as select * from values
(1,12,'3/4/2017'),(2,12,'5/26/2017'),(3,12,'12/1/2017'),(4,4,'2/1/2017'),(5,12,'1/1/2019');
with subq as (
select c.customerid, name,
array_agg(orderdate) within group (order by orderdate) as orders
from customer_table c
inner join order_table o on c.customerid = o.customerid
group by c.customerid, c.name
)
select customerid, name,
orders[0]::date AS order1, orders[1]::date AS order2,
array_to_string(array_slice(orders, 2, 999), ' , ') AS overflow
from subq;
------------+-------+------------+------------+-------------------------+
CUSTOMERID | NAME | ORDER1 | ORDER2 | OVERFLOW |
------------+-------+------------+------------+-------------------------+
4 | Nancy | 2017-02-01 | [NULL] | |
12 | John | 2017-03-04 | 2017-05-26 | 2017-12-01 , 2019-01-01 |
------------+-------+------------+------------+-------------------------+

Answer 5

first create a view like this:
create view order_view as
select
count(*) over (partition by customerId order by orderDate) as ord,
CustomerId,
orderdate
from order_table
then you can use this query:
select c.customerid,
o1.orderdate,
o2.orderdate
o3.orderdate
.
.
.
o12.orderdate
from customer_table c
left join order_view o1
on c.customerid = o1.customerid and ord = 1
left join order_view o2
on c.customerid = o2.customerid and ord = 2
left join order_view o3
on c.customerid = o3.customerid and ord = 3
.
.
.
left join order_view o12
on c.customerid = o12.customerid and ord = 12