Error - “UNION operator must have an equal number of expressions” when using CTE for recursive selection

Question

At this moment I have a table tblLocation with columns ID, Location, PartOfID.

The table is recursively connected to itself: PartOfID -> ID

My goal is to have a select output as followed:

> France > Paris > AnyCity >

Explanation: AnyCity is located in Paris, Paris is located in France.

My solution that I found until now was this:

; with q as (
select ID,Location,PartOf_LOC_id from tblLocatie t
where t.ID = 1 -- 1 represents an example
union all
select t.Location + '>' from tblLocation t
inner join q parent on parent.ID = t.LOC_PartOf_ID
)
select * from q

Unfortunately I get the following error:

All queries combined using a UNION, INTERSECT or EXCEPT operator must have an equal number of expressions in their target lists.

If you have any idea how I could fix my output it would be great.

Answer 1

The problem lays here:

--This result set has 3 columns
select LOC_id,LOC_locatie,LOC_deelVan_LOC_id from tblLocatie t
where t.LOC_id = 1 -- 1 represents an example

union all

--This result set has 1 columns   
select t.LOC_locatie + '>' from tblLocatie t
inner join q parent on parent.LOC_id = t.LOC_deelVan_LOC_id

In order to use union or union all number of columns and their types should be identical cross all result sets.

I guess you should just add the column LOC_deelVan_LOC_id to your second result set

Answer 2

Then number of columns must match between both parts of the union.

In order to build the full path, you need to "aggregate" all values of the Location column. You still need to select the id and other columns inside the CTE in order to be able to join properly. You get "rid" of them by simply not selecting them in the outer select:

with q as 
(
   select ID, PartOf_LOC_id, Location, ' > ' + Location as path
   from tblLocation 
   where ID = 1 

   union all

   select child.ID, child.PartOf_LOC_id, Location, parent.path + ' > ' + child.Location 
   from tblLocation child
     join q parent on parent.ID = t.LOC_PartOf_ID
)
select path
from q;

Answer 3

The second result set have only one column but it should have 3 columns for it to be contented to the first result set

(columns must match when you use UNION)

Try to add ID as first column and PartOf_LOC_id to your result set, so you can do the UNION.

;
WITH    q AS ( SELECT   ID ,
                    Location ,
                    PartOf_LOC_id
           FROM     tblLocation t
           WHERE    t.ID = 1 -- 1 represents an example
           UNION ALL
           SELECT   t.ID ,
                    parent.Location + '>' + t.Location ,
                    t.PartOf_LOC_id
           FROM     tblLocation t
                    INNER JOIN q parent ON parent.ID = t.LOC_PartOf_ID
         )
SELECT  *
FROM    q

Answer 4

You could use a recursive scalar function:-

set nocount on

create table location (
    id int,
    name varchar(50),
    parent int
)
insert into location values
    (1,'france',null),
    (2,'paris',1),
    (3,'belleville',2),
    (4,'lyon',1),
    (5,'vaise',4),
    (6,'united kingdom',null),
    (7,'england',6),
    (8,'manchester',7),
    (9,'fallowfield',8),
    (10,'withington',8)
go
create function dbo.breadcrumb(@child int)
returns varchar(1024)
as begin
    declare @returnValue varchar(1024)=''
    declare @parent int
    select @returnValue+=' > '+name,@parent=parent
    from location
    where id=@child
    if @parent is not null
        set @returnValue=dbo.breadcrumb(@parent)+@returnValue
    return @returnValue
end
go

declare @location int=1
while @location<=10 begin
    print dbo.breadcrumb(@location)+' >'
    set @location+=1
end

produces:-

 > france >
 > france > paris >
 > france > paris > belleville >
 > france > lyon >
 > france > lyon > vaise >
 > united kingdom >
 > united kingdom > england >
 > united kingdom > england > manchester >
 > united kingdom > england > manchester > fallowfield >
 > united kingdom > england > manchester > withington >

Answer 5

Although this an old post, I am sharing another working example.

"COLUMN COUNT AS WELL AS EACH COLUMN DATATYPE MUST MATCH WHEN 'UNION' OR 'UNION ALL' IS USED"

Let us take an example:

1:

In SQL if we write - SELECT 'column1', 'column2' (NOTE: remember to specify names in quotes) In a result set, it will display empty columns with two headers - column1 and column2

2: I share one simple instance I came across.

I had seven columns with few different datatypes in SQL. I.e. uniqueidentifier, datetime, nvarchar

My task was to retrieve comma separated result set with column header. So that when I export the data to CSV I have comma separated rows with first row as header and has respective column names.

SELECT CONVERT(NVARCHAR(36), 'Event ID') + ', ' + 
'Last Name' + ', ' + 
'First Name' + ', ' + 
'Middle Name' + ', ' + 
CONVERT(NVARCHAR(36), 'Document Type') + ', ' + 
'Event Type' + ', ' + 
CONVERT(VARCHAR(23), 'Last Updated', 126)

UNION ALL

SELECT CONVERT(NVARCHAR(36), inspectionid) + ', ' + 
       individuallastname + ', ' + 
       individualfirstname + ', ' + 
       individualmiddlename + ', ' +
       CONVERT(NVARCHAR(36), documenttype) + ', ' + 
       'I' + ', ' +
       CONVERT(VARCHAR(23), modifiedon, 126)
FROM Inspection

Above, columns 'inspectionid' & 'documenttype' has uniqueidentifer datatype and so applied CONVERT(NVARCHAR(36)). column 'modifiedon' is datetime and so applied CONVERT(NVARCHAR(23), 'modifiedon', 126).

Parallel to above 2nd SELECT query matched 1st SELECT query as per datatype of each column.