How do I select the max of several fields in the middle of a one to many to many query - sql

Question

All,
I have this query below which is a one to many to many relationship.
The relationship is as such, a story can have multiple tasks (1>M) each task can have multiple time entries (M-M)
I want to get the 1st occurrence of 3 fields in the tasks section and null the remainder for each story. So for example;
Story 1234
Tasks 12, 13, 14,15
Time entry 120, 121, 134, 135 etc....
There are 3 fields in tasks OptimisticEstimate, ProbableEstimate and Pessimistic Estimate that get repeated at the task level when Time entry table is linked in to the query (as it should be).
I want to get the 1st occurrence of those fields for each task and leave any other occurrence NULL so that when we sum it up the estimated don't get summed along with the hours from time entry.
Hopefully that is clear.
WITH
WorkItem_CTE AS
(-- Child Level (Task)
SELECT wi1.AreaID
,a.Name AS AreaName
,wi1.ProjectID
,wi1.IterationID
-- Work Items
,DATEADD(HOUR, DATEDIFF(HOUR, GETUTCDATE(), GETDATE()),wi1.CreatedDateUtc) AS WorkItemCreatedDTS
,wi1.WorkItemID
,wi1.[Type] AS WorkItemType
,wi1.Name AS WorkItemName
,wi1.[Status] AS WorkItemStatus
,wi1.[Priority] AS WorkItemPriority
,wi1.Complexity
,wi1.Estimate
-- Tasks
,DATEADD(HOUR, DATEDIFF(HOUR, GETUTCDATE(), GETDATE()),wi2.CreatedDateUtc) AS TaskCreatedDTS
,wi2.WorkItemID AS TaskID
,wi2.Name AS TaskName
,wi2.[Status] AS TaskStatus
,wi2.AssignedTo AS TaskAssignedTo
,wi2.EstimateOptimistic
,Case When wi2.EstimateProbable IS NULL Then (wi2.EstimateOptimistic + wi2.EstimatePessimistic)/2 Else wi2.EstimateProbable End As EstimateProbable
,wi2.EstimatePessimistic
,wi2.WorkCompleted
,wi2.WorkRemaining
FROM dbo.WorkItem wi1 WITH (NOLOCK)
INNER JOIN dbo.WorkItem wi2 WITH (NOLOCK)
ON wi1.WorkItemID = wi2.ParentID
AND wi2.[Status] <> 'Deleted'
LEFT OUTER JOIN dbo.Area a WITH (NOLOCK)
ON wi1.AreaID = a.AreaID
--LEFT OUTER JOIN dbo.Project p WITH (NOLOCK)
-- ON wi1.ProjectID = p.ProjectID
WHERE wi1.[Status] <> 'Deleted'
UNION ALL
-- Parent level (Story, Bug, Feedback)
SELECT wi1.AreaID
,a.Name AS AreaName
,wi1.ProjectID
,wi1.IterationID
-- Work Items
,DATEADD(HOUR, DATEDIFF(HOUR, GETUTCDATE(), GETDATE()),wi1.CreatedDateUtc) AS WorkItemCreatedDTS
,wi1.WorkItemID
,wi1.[Type] AS WorkItemType
,wi1.Name AS WorkItemName
,wi1.[Status] AS WorkItemStatus
,wi1.[Priority] AS WorkItemPriority
,wi1.Complexity
,wi1.Estimate
-- Tasks
,DATEADD(HOUR, DATEDIFF(HOUR, GETUTCDATE(), GETDATE()),wi2.CreatedDateUtc) AS TaskCreatedDTS
,wi2.WorkItemID AS TaskID
,wi2.Name AS TaskName
,wi2.[Status] AS TaskStatus
,wi2.AssignedTo AS TaskAssignedTo
,wi2.EstimateOptimistic
,Case When wi2.EstimateProbable IS NULL Then (wi2.EstimateOptimistic + wi2.EstimatePessimistic)/2 Else wi2.EstimateProbable End As EstimateProbable
,wi2.EstimatePessimistic
,wi2.WorkCompleted
,wi2.WorkRemaining
FROM dbo.WorkItem wi1 WITH (NOLOCK)
LEFT OUTER JOIN dbo.WorkItem wi2 WITH (NOLOCK)
ON wi1.WorkItemID = wi2.ParentID
AND wi2.[Status] <> 'Deleted'
LEFT OUTER JOIN dbo.Area a WITH (NOLOCK)
ON wi1.AreaID = a.AreaID
WHERE wi1.[Type] <> 'Task'
AND wi1.[Status] <> 'Deleted'
AND wi2.ParentID IS NULL
),
--Time Entry for Tasks
TimeEntry_CTE As
(SELECT te.WorkItemID
,te.ProjectID
,te.Date
,te.Hours
,te.CreatedBy
,te.CreatedDateUtc
,te.LastModifiedBy
,te.LastModifiedDateUtc
,te.Note
FROM TimeEntry te
)
SELECT wi.AreaID
,wi.AreaName
,wi.ProjectID
,p.Name AS ProjectName
,wi.WorkItemCreatedDTS
,wi.WorkItemID
,wi.WorkItemType
,wi.WorkItemName
,wi.WorkItemStatus
,wi.WorkItemPriority
,wi.Complexity
,wi.Estimate AS StoryEstimate
,wi.TaskCreatedDTS
,wi.TaskID
,wi.TaskName
,wi.EstimateOptimistic
,wi.EstimateProbable
,wi.EstimatePessimistic
,wi.TaskStatus
,wi.TaskAssignedTo
,wi.WorkCompleted
,wi.WorkRemaining
,te.WorkItemID as TimeWorkItemID
,te.ProjectID
,te.Date
,te.Hours
,te.CreatedBy
,te.CreatedDateUtc
,te.LastModifiedBy
,te.LastModifiedDateUtc
,te.Note
FROM WorkItem_CTE wi WITH (NOLOCK)
LEFT OUTER JOIN dbo.Project p WITH (NOLOCK)
ON wi.ProjectID = p.ProjectID
inner Join TimeEntry_CTE te With (nolock)
ON wi.TaskID = te.WorkItemID
Thanks in Advance for any help.
Revised Code based on suggestions. Closer but not quite right.
WITH
WorkItem_CTE AS
(-- Child Level (Task)
SELECT wi1.AreaID
,a.Name AS AreaName
,wi1.ProjectID
,wi1.IterationID
-- Work Items
,DATEADD(HOUR, DATEDIFF(HOUR, GETUTCDATE(), GETDATE()),wi1.CreatedDateUtc) AS WorkItemCreatedDTS
,wi1.WorkItemID
,wi1.[Type] AS WorkItemType
,wi1.Name AS WorkItemName
,wi1.[Status] AS WorkItemStatus
,wi1.[Priority] AS WorkItemPriority
,wi1.Complexity
,wi1.Estimate
-- Tasks
,DATEADD(HOUR, DATEDIFF(HOUR, GETUTCDATE(), GETDATE()),wi2.CreatedDateUtc) AS TaskCreatedDTS
,wi2.WorkItemID AS TaskID
,wi2.Name AS TaskName
,wi2.[Status] AS TaskStatus
,wi2.AssignedTo AS TaskAssignedTo
,wi2.EstimateOptimistic
,Case When wi2.EstimateProbable IS NULL Then (wi2.EstimateOptimistic + wi2.EstimatePessimistic)/2 Else wi2.EstimateProbable End As EstimateProbable
,wi2.EstimatePessimistic
,wi2.WorkCompleted
,wi2.WorkRemaining
FROM dbo.WorkItem wi1 WITH (NOLOCK)
INNER JOIN dbo.WorkItem wi2 WITH (NOLOCK)
ON wi1.WorkItemID = wi2.ParentID
AND wi2.[Status] <> 'Deleted'
LEFT OUTER JOIN dbo.Area a WITH (NOLOCK)
ON wi1.AreaID = a.AreaID
--LEFT OUTER JOIN dbo.Project p WITH (NOLOCK)
-- ON wi1.ProjectID = p.ProjectID
WHERE wi1.[Status] <> 'Deleted'
UNION ALL
-- Parent level (Story, Bug, Feedback)
SELECT wi1.AreaID
,a.Name AS AreaName
,wi1.ProjectID
,wi1.IterationID
-- Work Items
,DATEADD(HOUR, DATEDIFF(HOUR, GETUTCDATE(), GETDATE()),wi1.CreatedDateUtc) AS WorkItemCreatedDTS
,wi1.WorkItemID
,wi1.[Type] AS WorkItemType
,wi1.Name AS WorkItemName
,wi1.[Status] AS WorkItemStatus
,wi1.[Priority] AS WorkItemPriority
,wi1.Complexity
,wi1.Estimate
-- Tasks
,DATEADD(HOUR, DATEDIFF(HOUR, GETUTCDATE(), GETDATE()),wi2.CreatedDateUtc) AS TaskCreatedDTS
,wi2.WorkItemID AS TaskID
,wi2.Name AS TaskName
,wi2.[Status] AS TaskStatus
,wi2.AssignedTo AS TaskAssignedTo
,wi2.EstimateOptimistic
,Case When wi2.EstimateProbable IS NULL Then (wi2.EstimateOptimistic + wi2.EstimatePessimistic)/2 Else wi2.EstimateProbable End As EstimateProbable
,wi2.EstimatePessimistic
,wi2.WorkCompleted
,wi2.WorkRemaining
FROM dbo.WorkItem wi1 WITH (NOLOCK)
LEFT OUTER JOIN dbo.WorkItem wi2 WITH (NOLOCK)
ON wi1.WorkItemID = wi2.ParentID
AND wi2.[Status] <> 'Deleted'
LEFT OUTER JOIN dbo.Area a WITH (NOLOCK)
ON wi1.AreaID = a.AreaID
WHERE wi1.[Type] <> 'Task'
AND wi1.[Status] <> 'Deleted'
AND wi2.ParentID IS NULL
),
--Time Entry for Tasks
TimeEntry_CTE As
(SELECT te.WorkItemID
,te.ProjectID
,te.Date
,te.Hours
,te.CreatedBy
,te.CreatedDateUtc
,te.LastModifiedBy
,te.LastModifiedDateUtc
,te.Note
FROM TimeEntry te
),
wi_CTE As
(SELECT wi.AreaID
,wi.AreaName
,wi.ProjectID
,p.Name AS ProjectName
,wi.WorkItemCreatedDTS
,wi.WorkItemID
,wi.WorkItemType
,wi.WorkItemName
,wi.WorkItemStatus
,wi.WorkItemPriority
,wi.Complexity
,wi.Estimate AS StoryEstimate
,wi.TaskCreatedDTS
,wi.TaskID
,wi.TaskName
,wi.EstimateOptimistic
,wi.EstimateProbable
,wi.EstimatePessimistic
,wi.TaskStatus
,wi.TaskAssignedTo
,wi.WorkCompleted
,wi.WorkRemaining
,te.WorkItemID as TimeWorkItemID
,te.ProjectID TimeProjectID
,te.Date
,Sum(te.Hours) Hours
,te.CreatedBy
,te.CreatedDateUtc
,te.LastModifiedBy
,te.LastModifiedDateUtc
,te.Note
FROM WorkItem_CTE wi WITH (NOLOCK)
LEFT OUTER JOIN dbo.Project p WITH (NOLOCK)
ON wi.ProjectID = p.ProjectID
inner Join TimeEntry_CTE te With (nolock)
ON wi.TaskID = te.WorkItemID
--where wi.WorkItemID =94
Group By
wi.AreaID
,wi.AreaName
,wi.ProjectID
,p.Name
,wi.WorkItemCreatedDTS
,wi.WorkItemID
,wi.WorkItemType
,wi.WorkItemName
,wi.WorkItemStatus
,wi.WorkItemPriority
,wi.Complexity
,wi.Estimate
,wi.TaskCreatedDTS
,wi.TaskID
,wi.TaskName
,wi.EstimateOptimistic
,wi.EstimateProbable
,wi.EstimatePessimistic
,wi.TaskStatus
,wi.TaskAssignedTo
,wi.WorkCompleted
,wi.WorkRemaining
,te.WorkItemID
,te.ProjectID
,te.Date
,te.Hours
,te.CreatedBy
,te.CreatedDateUtc
,te.LastModifiedBy
,te.LastModifiedDateUtc
,te.Note),
cte AS (SELECT wi_cte.*, ROW_NUMBER() OVER(PARTITION BY WorkItemID, TaskID ORDER BY TimeWorkItemID) RN
FROM wi_CTE
)
SELECT
AreaID
,AreaName
,ProjectID
,ProjectName
,WorkItemCreatedDTS
,WorkItemID
,WorkItemType
,WorkItemName
,WorkItemStatus
,WorkItemPriority
,Complexity
,StoryEstimate
,TaskCreatedDTS
,TaskID
,TaskName
,EstimateOptimistic
,EstimateProbable
,EstimatePessimistic
,TaskStatus
,TaskAssignedTo
,WorkCompleted
,WorkRemaining
,WorkItemID
,ProjectID
,Date
,Hours
,CreatedBy
,CreatedDateUtc
,LastModifiedBy
,LastModifiedDateUtc
,Note
,RN
,SUM(CASE WHEN RN = 1 THEN Hours END) AS SumTime
FROM cte
Where RN=1
GROUP BY
AreaID
,AreaName
,ProjectID
,ProjectName
,WorkItemCreatedDTS
,WorkItemID
,WorkItemType
,WorkItemName
,WorkItemStatus
,WorkItemPriority
,Complexity
,StoryEstimate
,TaskCreatedDTS
,TaskID
,TaskName
,EstimateOptimistic
,EstimateProbable
,EstimatePessimistic
,TaskStatus
,TaskAssignedTo
,WorkCompleted
,WorkRemaining
,WorkItemID
,ProjectID
,Date
,Hours
,CreatedBy
,CreatedDateUtc
,LastModifiedBy
,LastModifiedDateUtc
,Note
,RN
What I get
ProjectID ProjectName WorkItemID WorkItemType WorkItemName EstimateOptimistic EstimateProbable EstimatePessimistic RN SumTime
11 Data Group 94 Story Backlog Report with new logic 2 3 4 1 8
11 Data Group 94 Story Backlog Report with new logic 2 3 4 2 4
11 Data Group 94 Story Backlog Report with new logic 2 3 4 3 4
11 Data Group 94 Story Backlog Report with new logic 2 3 4 4 4
11 Data Group 94 Story Backlog Report with new logic 2 3 4 1 4
11 Data Group 94 Story Backlog Report with new logic 2 3 4 2 4
11 Data Group 94 Story Backlog Report with new logic 2 3 4 3 4
11 Data Group 94 Story Backlog Report with new logic 2 3 4 4 8
11 Data Group 94 Story Backlog Report with new logic 2 3 4 5 3
What I'm Looking for
ProjectID ProjectName WorkItemID WorkItemType WorkItemName EstimateOptimistic EstimateProbable EstimatePessimistic RN SumTime
11 Data Group 94 Story Backlog Report with new logic 2 3 4 1 8
11 Data Group 94 Story Backlog Report with new logic NULL NULL NULL 2 4
11 Data Group 94 Story Backlog Report with new logic NULL NULL NULL 3 4
11 Data Group 94 Story Backlog Report with new logic NULL NULL NULL 4 4
11 Data Group 94 Story Backlog Report with new logic 2 3 4 1 4
11 Data Group 94 Story Backlog Report with new logic NULL NULL NULL 2 4
11 Data Group 94 Story Backlog Report with new logic NULL NULL NULL 3 4
11 Data Group 94 Story Backlog Report with new logic NULL NULL NULL 4 8
11 Data Group 94 Story Backlog Report with new logic NULL NULL NULL 5 3
Hopefully that helps clarify what I'm trying for.

Answer 1

I suggest using a conditional SUM() and the ROW_NUMBER() function.
Let's say the query above outputs data like this:
Story, Task, TimeEntry
1234, 12, 120
1234, 12, 121
1234, 12, 134
1234, 13, 135
1234, 13, 120
1234, 13, 121
1234, 13, 134
1234, 14, 135
1234, 14, 120
1234, 14, 121
1234, 14, 134
1234, 14, 135
You could use this:
;WITH cte AS (SELECT *, ROW_NUMBER() OVER(PARTITION BY Story, Task ORDER BY TimeEntry) RN
FROM Table1
)
SELECT Story, Task, SUM(CASE WHEN RN = 1 THEN TimeEntry END) AS SumTime
FROM cte
GROUP BY Story, Task
Demo: SQL Fiddle
Each line will be numbered, starting at 1 for each combination of Story and Task, allowing you to use the condition of RN = 1 to avoid aggregating duplicate lines. Since you said the values repeat it doesn't matter what you put in the ORDER BY clause for your ROW_NUMBER().