Recursive CTE with hierarchy data - sql-server

Sample data:
declare #docs table (docid int, name varchar(10), isfolder int)
declare #hierarchy table (childid int, parent varchar(10))
insert #docs values (1, 'Doc1', 0)
insert #docs values (2, 'Doc2', 0 )
insert #docs values (3, 'Folder1', 1 )
insert #docs values (4, 'Folder2', 1 )
insert #docs values (5, 'SubFolderA', 1 )
insert #docs values (6, 'SubFolderB', 1 )
insert #hierarchy values (1, 5)
insert #hierarchy values (1, 6)
insert #hierarchy values (2, 6)
insert #hierarchy values (5, 3)
insert #hierarchy values (6, 4)
I want to list the data above so I end up with the documents (isfolder = 0) and the folder paths that they are in.
Expected output:
DocId Name Path
--------------------------------------
1 Doc1 Folder1\SubFolderA
1 Doc1 Folder2\SubFolderB
2 Doc2 Folder2\SubFolderB
CTE
I started to write this next CTE as it said this can be used to create my output but I am wrong somewhere
;WITH folderCTE (docid, name) AS
(
-- Anchor member
SELECT docid, CAST(name AS varchar(max)) AS name
FROM #docs
WHERE isfolder = 0
UNION ALL
-- Recursive member that references expression_name.
SELECT d.docid, CAST(folderCTE.Name + '\' + d.name AS varchar(max)) AS name
FROM folderCTE
INNER JOIN #docs d ON d.docid = folderCTE.docid
)
-- references expression name
SELECT *
FROM folderCTE
Can anyone show me how I should be doing a CTE for this?

If you use a CTE which just expands the hierarchy for folders, and a second one to build the hierarchy, you can simply join to this to get the folder for your docs:
declare #docs table (docid int, name varchar(10), isfolder int)
declare #hierarchy table (childid int, parentid varchar(10))
insert #docs values (1, 'Doc1', 0)
insert #docs values (2, 'Doc2', 0 )
insert #docs values (3, 'Folder1', 1 )
insert #docs values (4, 'Folder2', 1 )
insert #docs values (5, 'SubFolderA', 1 )
insert #docs values (6, 'SubFolderB', 1 )
insert #docs values (7, 'SuperFold1', 1 )
insert #hierarchy values (1, 5)
insert #hierarchy values (1, 6)
insert #hierarchy values (2, 6)
insert #hierarchy values (5, 3)
insert #hierarchy values (6, 4)
insert #hierarchy values (3, 7)
;WITH folderCTE
AS
(
select docid, cast(name as nvarchar(max)) as name, parentid
from #docs d
left join #hierarchy h on d.docid=h.childid
where d.isfolder = 1
),
folderHierarchyCTE
as
(
select docid, cast(name as nvarchar(max)) as name
from folderCTE where parentid is null
union all
select d.docid, cast(p.name + '/' + d.name as nvarchar(max))
from folderCTE d
inner join folderHierarchyCTE p on d.parentid = p.docid
)
SELECT d.docid, d.name, f.name
FROM #docs d
inner join #hierarchy h on h.childid=d.docid
inner join folderHierarchyCTE f on h.parentid=f.docid
where d.isfolder = 0

Related

Repeat the column value until value change in same column value

My table structure:
declare #TestTable as table
(
id int,
somedate date,
somevalue int
)
insert into #TestTable values
(1, '2019-01-01', 1000),
(2, '2019-01-02', null ),
(3, '2019-01-03', null),
(4, '2019-01-04', null ),
(5, '2019-01-05', 800),
(6, '2019-01-06', null),
(7, '2019-01-07', null),
(8, '2019-01-08', null),
(9, '2019-01-09', null),
(10, '2019-01-10', 700)
Repeat the column value until any change in value of column (somevalue).. Is it possible with window functions?
Required output:
You can achieve that by using window function as well.
try the following:
select id, somevalue, newvalue=max(somevalue) over (partition by c)
from
(
select id, somevalue
,c=count(somevalue) over (order by id)
from #testtable
) t
order by id;
Please find the demo here.
Try this below logic-
DEMO HERE
SELECT A.*,
(
SELECT somevalue
FROM #TestTable
WHERE id = (
SELECT MAX(id)
FROM #TestTable
WHERE id <= A.id
AND somevalue IS NOT NULL
)
) new_column
FROM #TestTable A
ORDER BY A.id
You can achieve this using COALESCE (Transact-SQL)
.
SELECT
[id]
, somedate
, COALESCE(somevalue,
(SELECT TOP (1) somevalue
FROM #TestTable AS p2
WHERE
p2.somevalue IS NOT NULL
AND p2.[id] <= p.[id] ORDER BY p2.[id] DESC))
FROM #TestTable AS p;
Here is the live db<>fiddle demo.

Join two columns and pick just one from the child results

I want to perform a JOIN on two tables to get the LEADTIME per ITEMNUM.
Both tables have common value ITEMNUM that I use for the JOIN operation.
The problem is that in the second table the ITEMNUM is not unique and can contain multiple. different LEADTIME values.
For example see ITEMNUM 2 in Table 2.
In case there are multiple LEADTIME values, I just want to get one of the LEADTIME values.
I don't care which one.
This is what I have so far, but it keeps returning multiple lines for ITEMNUM 2
SELECT ITEMNUM, LEADTIME
FROM TABLE1
LEFT JOIN TABLE2 on TABLE2.ITEMNUM = TABLE1.ITEMNUM
So what can I do to get just one LEADTIME for ITEMNUM 2? ( as mentioned, I don't care which value )
This approach assigns a row number to each row in #table2 resetting it for each ItemNum value. You need to have an order by clause (if you don't SQL Server raises an error) so I am ordering by NEWID() which should result in a randomized order. You will likely want to tweak what columns you are returning. Here is the dbfiddle to see it in action.
IF OBJECT_ID('tempdb.dbo.#table1', 'U') IS NOT NULL DROP TABLE #table1;
IF OBJECT_ID('tempdb.dbo.#table2', 'U') IS NOT NULL DROP TABLE #table2;
CREATE TABLE #table1
(
ID INT
, ItemNum INT
);
CREATE TABLE #table2
(
ID INT
, ItemNum INT
, LeadTime INT
);
INSERT INTO #table1 VALUES (1, 1)
INSERT INTO #table1 VALUES (2, 2)
INSERT INTO #table1 VALUES (3, 3)
INSERT INTO #table1 VALUES (4, 4)
INSERT INTO #table1 VALUES (5, 5)
INSERT INTO #table2 VALUES (1, 1, 6)
INSERT INTO #table2 VALUES (2, 2, 7)
INSERT INTO #table2 VALUES (3, 2, 2)
INSERT INTO #table2 VALUES (4, 3, 6)
INSERT INTO #table2 VALUES (5, 4, 3)
SELECT *
FROM #table1 AS t1
LEFT JOIN (
SELECT *
, ROW_NUMBER() OVER (PARTITION BY ItemNum ORDER BY NEWID()) AS rn
FROM #table2
) AS t2 ON t1.ItemNum = t2.ItemNum
AND t2.rn = 1;
There are several ways to get this done.I would use OUTER APPLY with TOP.
DROP TABLE IF EXISTS #Table1
CREATE TABLE #Table1
(
Id INT
, ItemNum INT
)
DROP TABLE IF EXISTS #Table2
CREATE TABLE #Table2
(
Id INT
, ItemNum INT
, LeadTime INT
)
INSERT INTO #Table1 VALUES
(1, 1)
, (2, 2)
, (3, 3)
, (4, 4)
, (5, 5)
INSERT INTO #Table2 VALUES
(1, 1, 6)
, (2, 2, 7)
, (3, 2, 2)
, (4, 3, 6)
, (5, 4, 3)
SELECT
*
FROM
#Table1 AS T1
OUTER APPLY
(
SELECT TOP 1 T2.LeadTime FROM #Table2 AS T2 WHERE T2.ItemNum = T1.ItemNum
) AS LT

I want to convert some data to pivot in SQL Server with join and dynamic

The table, at last, is my target.
This is my demo database
create database pvtestDb;
go
use pvtestDb;
go
create table custTransaction
(
id int,
custNum int,
value nvarchar(50)
)
go
create table customers
(
id int,
custName nvarchar(50)
)
insert into Customers(id, custName)
values (1, 'aaa'), (2, 'bbb'), (3, 'ccc'), (4, 'ddd'),
(5, 'eee'), (6, 'fff'), (7, 'ggg'), (8, 'hhh'), (9, 'iii')
insert into custTransaction (id, custNum, value)
values (1, 3, 'a'), (1, 4, 'b'), (1, 5, 'c'),
(2, 3, 'd'), (2, 4, 'e'), (2, 6, 'f'),
(3, 3, 'g'), (3, 8, 'h'), (3, 9, 'i')
select * from customers
select * from custTransaction
select custName, custNum, value
from customers
join custTransaction on custTransaction.id = customers.id
I tried code like this but not worked at all
SELECT
custNum, [a], [b], [c], [d]
FROM
customers
JOIN
custTransaction ON custTransaction.id = customers.id
PIVOT
(COUNT([custName])
FOR [custName] IN ([a], [b], [c], [d])) AS p
I need to join between the two tables in first.
Any hints would be appreciated as I am stuck with this situation
Here's approach with dynamic SQL
declare #customers varchar(8000)
declare #sql varchar(8000)
select #customers = stuff((
select ',' + quotename(custName)
from customers
for xml path('')
), 1, 1, '')
set #sql = 'select
id, ' + #customers + '
from (
select
ct.id, c.custName, ct.value
from
customers c
join custTransaction ct on ct.custNum = c.id
) t
pivot (
max(value) for custName in (' + #customers + ')
) p'
exec (#sql)
Output
id aaa bbb ccc ddd eee fff ggg hhh iii
----------------------------------------------------------------
1 NULL NULL a b c NULL NULL NULL NULL
2 NULL NULL d e NULL f NULL NULL NULL
3 NULL NULL g NULL NULL NULL NULL h i

SQL or Tableau Replace values with existing one with reference to another field

My problem with the data is
Id Name
--------
1 a
2 b
1 a
1 a
1 NULL
2 b
2 b
2 NULL
2 b
2 NULL
ID is unique to the name
I would like the NULL values to be replaced with existing Records.
Or if there are any other alternatives
I have about 2000 IDs so it needs to be automated.
The below query can use to replace the NULLs with its Id's Name, since Id is unique to the Name.
UPDATE T1 SET T1.Name = T2.ValidName
FROM TestTable T1
JOIN ( SELECT Id, MAX(Name) AS ValidName
FROM TestTable
WHERE Name IS NOT NULL
GROUP BY Id ) T2 ON T2.Id = T1.Id
WHERE T1.Name IS NULL
Sample execution in SQL-Server:
DECLARE #TestTable TABLE (Id INT, Name VARCHAR (20));
INSERT INTO #TestTable (Id, Name) VALUES
(1, 'a'),
(2, 'b'),
(1, 'a'),
(1, 'a'),
(1, NULL),
(2, 'b'),
(2, 'b'),
(2, NULL),
(2, 'b'),
(2, NULL);
UPDATE T1 SET T1.Name = T2.ValidName
FROM #TestTable T1
JOIN ( SELECT Id, MAX(Name) AS ValidName
FROM #TestTable
WHERE Name IS NOT NULL
GROUP BY Id ) T2 ON T2.Id = T1.Id
WHERE T1.Name IS NULL
SELECT * FROM #TestTable
try this: replace with your tablename
declare #id int, #name varchar(10)
declare c cursor for
select distinct id,name from <table> where id is not null and name is not null
open c
fetch next from c into #id,#name
while ##FETCH_STATUS=0
begin
update <table> set name=#name where id=#id
fetch next from c into #id,#name
end
close c
deallocate c

T-SQL SELECT query to return combined result of multiple tables

I am wondering if it is possible and more efficient to do something that I am presently doing in code, to do in T-SQL instead.
I have a database with courses. Each course can have different offerings which are variations of the course at different locations and at different awards.
Here's my (simplified) database structure and some sample data:
CREATE TABLE tblCourse (CourseId int, CourseName varchar(50))
CREATE TABLE tblOffering (OfferingId int, CourseId int, LocationId int, AwardId int)
CREATE TABLE tblLocation (LocationId int, LocationName varchar(50))
CREATE TABLE tblAward (AwardId int, AwardName varchar(50))
INSERT INTO tblCourse VALUES (1, 'Course A')
INSERT INTO tblCourse VALUES (2, 'Course B')
INSERT INTO tblOffering VALUES (1, 1, 1, 1)
INSERT INTO tblOffering VALUES (2, 1, 2, 1)
INSERT INTO tblOffering VALUES (3, 1, 3, 1)
INSERT INTO tblOffering VALUES (4, 1, 1, 2)
INSERT INTO tblOffering VALUES (5, 2, 3, 1)
INSERT INTO tblLocation VALUES (1, 'Location A')
INSERT INTO tblLocation VALUES (2, 'Location B')
INSERT INTO tblLocation VALUES (3, 'Location C')
INSERT INTO tblAward VALUES (1, 'Award A')
INSERT INTO tblAward VALUES (2, 'Award B')
What I want to retrieve from SQL is a single row for each course/award combination. Each row would have columns for each location and whether a course of that CourseId/AwardId combination was available. There would be now rows for course/award combinations that have no offerings.
The required result, from the sample data, would be a recordset like this:
CourseId | CourseName | AwardId | AwardName | LocationA | LocationB | LocationC
---------+------------+---------+-----------+-----------+-----------+----------
1 | Course A | 1 | Award A | True | True | True
1 | Course A | 2 | Award B | True | NULL | NULL
2 | Course B | 1 | Award A | NULL | NULL | True
(NULL could also be False)
At present I am doing a simple SELECT statement with various JOINS which gives me multiple rows for each course/award combination, then I loop through all rows in my code and build the required result. However, I don't think this is so efficient as I also need to page results.
I think I could do this fairly easily in a stored procedure by creating a temporary table and a bunch of separate queries, but I don't think that would be too efficient. Wondering if there is a better way of doing it in T-SQL???
So to clarify, what I am looking for is a T-SQL query or stored procedure that will produce the above sample recordset, and which I could adapt paging to.
NB. I am using SQL Server 2008
For Dynamic columns:
DECLARE #COLUMNS VARCHAR(max)
,#query varchar(1024)
,#True varchar(6)
SELECT #COLUMNS =
COALESCE(
#Columns + ',[' + L.LocationName + ']',
'[' + L.LocationName +']'
)
FROM tblLocation L
SELECT #True = '''True'''
SELECT #QUERY = 'SELECT C.CourseName
,A.AwardName
, pvt.*
FROM (SELECT O.OfferingID AS OID
,O.AwardID AS AID
,O.CourseID AS CID
,L.LocationName AS LID
FROM tblOffering O Inner Join tblLocation L on L.LocationID = O.LocationID) AS S
PIVOT
(
count(oID) For LID IN (' +#COLUMNS+ ')
) As pvt
inner join tblCourse C on C.CourseID = CID
inner join tblAward A on A.AwardID = pvt.AID'
EXEC (#QUERY)
GO
This will produce a paginated version of your example results:
declare #tblCourse as table (CourseId int, CourseName varchar(50))
declare #tblOffering as table (OfferingId int, CourseId int, LocationId int, AwardId int)
declare #tblLocation as table (LocationId int, LocationName varchar(50))
declare #tblAward as table (AwardId int, AwardName varchar(50))
INSERT INTO #tblCourse VALUES (1, 'Course A')
INSERT INTO #tblCourse VALUES (2, 'Course B')
INSERT INTO #tblOffering VALUES (1, 1, 1, 1)
INSERT INTO #tblOffering VALUES (2, 1, 2, 1)
INSERT INTO #tblOffering VALUES (3, 1, 3, 1)
INSERT INTO #tblOffering VALUES (4, 1, 1, 2)
INSERT INTO #tblOffering VALUES (5, 2, 3, 1)
INSERT INTO #tblLocation VALUES (1, 'Location A')
INSERT INTO #tblLocation VALUES (2, 'Location B')
INSERT INTO #tblLocation VALUES (3, 'Location C')
INSERT INTO #tblAward VALUES (1, 'Award A')
INSERT INTO #tblAward VALUES (2, 'Award B') -- This had id 1 in your example.
-- Set the following parameters to control paging:
declare #PageSize as Int = 5
declare #PageNumber as Int = 1
; with CourseAwardSummary as (
select distinct C.CourseId, C.CourseName, A.AwardId, A.AwardName,
case when exists ( select 42 from #tblOffering where CourseId = C.CourseId and AwardId = A.AwardId and LocationId = 1 ) then 'True' end as LocationA,
case when exists ( select 42 from #tblOffering where CourseId = C.CourseId and AwardId = A.AwardId and LocationId = 2 ) then 'True' end as LocationB,
case when exists ( select 42 from #tblOffering where CourseId = C.CourseId and AwardId = A.AwardId and LocationId = 3 ) then 'True' end as LocationC
from #tblCourse as C inner join
#tblOffering as O on O.CourseId = C.CourseId inner join
#tblAward as A on A.AwardId = O.AwardId
),
CourseAwardSummaryRows as (
select *, Row_Number() over ( order by CourseName, AwardName ) as RowNumber
from CourseAwardSummary
)
select CourseId, CourseName, AwardId, AwardName, LocationA, LocationB, LocationC
from CourseAwardSummaryRows
where ( #PageNumber - 1 ) * #PageSize + 1 <= RowNumber and RowNumber <= #PageNumber * #PageSize
order by CourseName, AwardName
The following query does this by joining and aggregating the offering table, and then joining the result to the course and award tables:
select c.CourseId, c.CourseName, oa.AwardId, oa.AwardName,
oa.LocationA, oa.LocationB, oa.LocationC
from tblCourse c left outer join
(select o.CourseId, o.AwardId, a.awardName
max(case when LocationName = 'Location A' then 'true' end) as LocationA,
max(case when LocationName = 'Location B' then 'true' end) as LocationB,
max(case when LocationName = 'Location C' then 'true' end) as LocationC
from tblOffering o join
tblLocation l
on o.LocationId = l.LocationId join
tblAward a
on a.awardID = o.AwardId
group by o.CourseId, o.AwardId, a.awardName
) oa
on oa.CourseId = c.CourseId

Resources