Subquery with count problem

[Hasse1982]

Hi

I have a table KDOCUMENT with the columns
[DocID],
[Title],
[CreatedByUser],
[CreateDate],
[LastModifiedByU serName],
[LastModificatio nDate],
[Comment]

On this table I want to find the test documents so that they can be deleted, hence I have created the query to find documents with the words "test" and "delete" in [Title] and [Comment] looking like this:
SELECT [DocID], [Title],[CreatedByUser], [CreateDate],[LastModifiedByU serName], [LastModificatio nDate], [Comment]
FROM [DMSDB].[dbo].[KDOCUMENT]
WHERE [Title] LIKE '%test%'
OR [Comment] LIKE '%test%'
OR [Title] LIKE '%delete%'
OR [Comment] LIKE '%delete%'
OR [OnlineVersionsS ize] = 0.
This one is working fine returning 5070 rows with documents containing test and delete.

Problem is that we only want to list the documents belonging to users that have X documents or more (for instance 20), because we have to contact the people and ask them which documents actually are test documents. If someone only have 1 document we don't want to bother them, it is more important to clean up the "real heavy" user's documents.
So, I created a query

SELECT COUNT([CreatedByUser]) AS 'Number of docs', [CreatedByUser]
FROM [DMSDB].[dbo].[KDOCUMENT]
WHERE [Title] LIKE '%test%'
OR [Comment] LIKE '%test%'
OR [Title] LIKE '%delete%'
OR [Comment] LIKE '%delete%'
OR [OnlineVersionsS ize] = 0
GROUP BY [CreatedByUser]
HAVING COUNT([CreatedByUser]) > 20
ORDER BY COUNT([CreatedByUser]) DESC

returning correctly the number of documents for each user and the user name. It also limits itself to listing only users with a count of 20 or more.

Now, I want to combine these two, listing all documents with DocID and everything where a user has 20 documents or more. But when I do this:

SELECT [DocID], [Title],[CreatedByUser], [CreateDate], [LastModifiedByU serName], [LastModificatio nDate], [Comment]
FROM [DMSDB].[dbo].[KDOCUMENT]
WHERE [Title] LIKE '%delete%'
OR [Comment] LIKE '%test%'
OR [Title] LIKE '%test%'
OR [Comment] LIKE '%delete%'
OR [OnlineVersionsS ize] = 0
AND [CreatedByUser] IN (
SELECT [CreatedByUser]
FROM [DMSDB].[dbo].[KDOCUMENT]
GROUP BY [CreatedByUser]
HAVING COUNT([CreatedByUser]) > 20)

ORDER BY [CreatedByUser]

I lists all the documents as in the first query, returning 5070 rows while it should return about 4715 (I know this by adding up the numbers in the count query).

It has been a long time since I've been working with SQL, and I am struggling a bit here. Can someone help me? I get the same result using EXISTS, and when using = I of course get the error that it is returning more than one row.

[ck9663]

Hi

I have a table KDOCUMENT with the columns
[DocID],
[Title],
[CreatedByUser],
[CreateDate],
[LastModifiedByU serName],
[LastModificatio nDate],
[Comment]

On this table I want to find the test documents so that they can be deleted, hence I have created the query to find documents with the words "test" and "delete" in [Title] and [Comment] looking like this:
SELECT [DocID], [Title],[CreatedByUser], [CreateDate],[LastModifiedByU serName], [LastModificatio nDate], [Comment]
FROM [DMSDB].[dbo].[KDOCUMENT]
WHERE [Title] LIKE '%test%'
OR [Comment] LIKE '%test%'
OR [Title] LIKE '%delete%'
OR [Comment] LIKE '%delete%'
OR [OnlineVersionsS ize] = 0.
This one is working fine returning 5070 rows with documents containing test and delete.

Problem is that we only want to list the documents belonging to users that have X documents or more (for instance 20), because we have to contact the people and ask them which documents actually are test documents. If someone only have 1 document we don't want to bother them, it is more important to clean up the "real heavy" user's documents.
So, I created a query

SELECT COUNT([CreatedByUser]) AS 'Number of docs', [CreatedByUser]
FROM [DMSDB].[dbo].[KDOCUMENT]
WHERE [Title] LIKE '%test%'
OR [Comment] LIKE '%test%'
OR [Title] LIKE '%delete%'
OR [Comment] LIKE '%delete%'
OR [OnlineVersionsS ize] = 0
GROUP BY [CreatedByUser]
HAVING COUNT([CreatedByUser]) > 20
ORDER BY COUNT([CreatedByUser]) DESC

returning correctly the number of documents for each user and the user name. It also limits itself to listing only users with a count of 20 or more.

Now, I want to combine these two, listing all documents with DocID and everything where a user has 20 documents or more. But when I do this:

SELECT [DocID], [Title],[CreatedByUser], [CreateDate], [LastModifiedByU serName], [LastModificatio nDate], [Comment]
FROM [DMSDB].[dbo].[KDOCUMENT]
WHERE [Title] LIKE '%delete%'
OR [Comment] LIKE '%test%'
OR [Title] LIKE '%test%'
OR [Comment] LIKE '%delete%'
OR [OnlineVersionsS ize] = 0
AND [CreatedByUser] IN (
SELECT [CreatedByUser]
FROM [DMSDB].[dbo].[KDOCUMENT]
GROUP BY [CreatedByUser]
HAVING COUNT([CreatedByUser]) > 20)

ORDER BY [CreatedByUser]

I lists all the documents as in the first query, returning 5070 rows while it should return about 4715 (I know this by adding up the numbers in the count query).

It has been a long time since I've been working with SQL, and I am struggling a bit here. Can someone help me? I get the same result using EXISTS, and when using = I of course get the error that it is returning more than one row.

Use parenthesis to group your conditions

 
WHERE ([Title] LIKE '%delete%'                                                
    OR [Comment] LIKE '%test%'                                                
    OR [Title] LIKE '%test%'                                                
    OR [Comment] LIKE '%delete%'                                            
    OR [OnlineVersionsSize] = 0)
   AND [CreatedByUser] IN (
               SELECT [CreatedByUser]
                             FROM [DMSDB].[dbo].[KDOCUMENT]
                            GROUP BY [CreatedByUser]
                           HAVING COUNT([CreatedByUser]) > 20)
 
 

Conditions are evaluated left to right. If one of those ORs is true, your whole condition is true.

Similar to math formula:

2 + 3 * 5 = 25

2 + (3 * 5) = 17

Happy coding!

-- CK

[Hasse1982]

Use parenthesis to group your conditions

 
WHERE ([Title] LIKE '%delete%'                                                
    OR [Comment] LIKE '%test%'                                                
    OR [Title] LIKE '%test%'                                                
    OR [Comment] LIKE '%delete%'                                            
    OR [OnlineVersionsSize] = 0)
   AND [CreatedByUser] IN (
               SELECT [CreatedByUser]
                             FROM [DMSDB].[dbo].[KDOCUMENT]
                            GROUP BY [CreatedByUser]
                           HAVING COUNT([CreatedByUser]) > 20)
 
 

Conditions are evaluated left to right. If one of those ORs is true, your whole condition is true.

Similar to math formula:

2 + 3 * 5 = 25

2 + (3 * 5) = 17

Happy coding!

-- CK

Wow, thank you!

It works like a charm, and the best part is that I was not completely off, it was just some paranthesis missing.
I now just discovered one thing you might be able to help me with:
This query: SELECT [CreatedByUser]
FROM [DMSDB].[dbo].[KDOCUMENT]
GROUP BY [CreatedByUser]
HAVING COUNT([CreatedByUser]) > 20)
counts ALL users in the entire table, but I only want it to count the ones with the criteria listed in the first query (Title and Comment equals "test" and/or "delete". When I try to put:

WHERE ([Title] LIKE '%delete%'
OR [Comment] LIKE '%test%'
OR [Title] LIKE '%test%'
OR [Comment] LIKE '%delete%'
OR [OnlineVersionsS ize] = 0

into the second query like this:

SELECT [DocID], [Title],[CreatedByUser], [CreateDate], [LastModifiedByU serName], [LastModificatio nDate], [Comment], [OnlineVersionsS ize]
FROM [DMSDB].[dbo].[KDOCUMENT]
WHERE [CreatedByUser] IN (
SELECT [CreatedByUser]
FROM [DMSDB].[dbo].[KDOCUMENT]
WHERE ([Title] LIKE '%delete%'
OR [Comment] LIKE '%test%'
OR [Title] LIKE '%test%'
OR [Comment] LIKE '%delete%'
OR [OnlineVersionsS ize] = 0)
GROUP BY [CreatedByUser]
HAVING COUNT([CreatedByUser]) > 20)

it shows 0 results. Any ideas?
There might be another way to do this, so I am very open for suggestions :)

[ck9663]

Do you have NULL in CreatedByUser ?

try :

 
SELECT DocID, Title,CreatedByUser, CreateDate, LastModifiedByUserName, 
   LastModificationDate, Comment, OnlineVersionsSize
FROM DMSDB.dbo.KDOCUMENT
WHERE CreatedByUser IN 
   (
      SELECT CreatedByUser
      FROM DMSDB.dbo.KDOCUMENT
      WHERE (Title LIKE '%delete%' OR Comment LIKE '%test%' OR Title LIKE '%test%' OR Comment LIKE '%delete%' OR OnlineVersionsSize = 0)
         and CreatedByUser IS NOT NULL
      GROUP BY CreatedByUser
      HAVING COUNT(*) > 20
   )
WHERE CreatedByUser IS NOT NULL
 
 

You should be careful with IN. Read more about it here

-- CK