Question

I have the following table and data in SQL Server 2005:

create table LogEntries (   ID int identity,   LogEntry varchar(100) )  insert into LogEntries values ('beans') insert into LogEntries values ('beans') insert into LogEntries values ('beans') insert into LogEntries values ('cabbage') insert into LogEntries values ('cabbage') insert into LogEntries values ('beans') insert into LogEntries values ('beans') 

I would like to group repeated LogEntries so that I have the following results:

LogEntry  EntryCount beans     3 cabbage   2 beans     2 

Can you think of any way to do this in TSQL outside of using a cursor?

Answer 1

This is a set-based solution for the problem. The performance will probably suck, but it works 🙂

CREATE TABLE #LogEntries (   ID INT IDENTITY,   LogEntry VARCHAR(100) )  INSERT INTO #LogEntries VALUES ('beans') INSERT INTO #LogEntries VALUES ('beans') INSERT INTO #LogEntries VALUES ('beans') INSERT INTO #LogEntries VALUES ('cabbage') INSERT INTO #LogEntries VALUES ('cabbage') INSERT INTO #LogEntries VALUES ('carrots') INSERT INTO #LogEntries VALUES ('beans') INSERT INTO #LogEntries VALUES ('beans') INSERT INTO #LogEntries VALUES ('carrots')  SELECT logentry, COUNT(*) FROM (     SELECT logentry,      ISNULL((SELECT MAX(id) FROM #logentries l2 WHERE l1.logentry<>l2.logentry AND l2.id < l1.id), 0) AS id     FROM #LogEntries l1 ) AS a GROUP BY logentry, id   DROP TABLE #logentries  

Results:

beans   3 cabbage 2 carrots 1 beans   2 carrots 1 

The ISNULL() is required for the first set of beans.