避免LINQ中的记录重复

Question

I have a situation where records are being duplicated and I don't know how to deal with it. Here is the LINQ statement:

theData = (from urls in this.ObjectContext.Activities.AsExpandable()
            .Where(predicateA)
            .Where(r => (r.StartDate >= beginDate && r.StartDate <= endDate) ||
                  (r.EndDate >= beginDate && r.EndDate <= endDate) ||
                  (r.StartDate <= beginDate && r.EndDate >= endDate))
                join idGroups in this.ObjectContext.IdentityGroups 
                    on urls.IdentityID equals idGroups.IdentityID
                join groupSup in this.ObjectContext.GroupSupervisors
                    .Where(r => r.SupervisorID == loggedInID) 
                    on idGroups.GroupID equals groupSup.GroupID
                join programs in progs 
                    on urls.ProcessName.ToUpper() equals programs.ProcessName.ToUpper() 
            into jt
            from jt1 in jt.DefaultIfEmpty()
                .Where(r => r == null || r.Ignore == false)
            group urls by new 
                        { urls.ProcessName, 
                          urls.ContextID, 
                          jt1.CustomCategory, 
                          jt1.Name, 
                          groupSup.SupervisorID 
                        } 
            into groupedTable
            select new ActivityInfoSummary_DTO
            {
               recId = Guid.NewGuid(),
               Context = groupedTable.Key.ProcessName,
               ContextId = groupedTable.Key.ContextID,
               SupervisorId = groupedTable.Key.SupervisorID,
               FocusCount = groupedTable.Sum(r => r.FocusCount),
               many more fields....
            }).ToList();

The dilemma is: urls.identityId is the ID of the person who created the record.

The person creating the record can belong in more than one group

Each group has a single supervisor

Each supervisor can be the supervisor of multiple groups

A person can belong to multiple groups

The linq statements are trying to filter down the records created by a person based on the fact that the person is a member of a group the supervisor manages (supervisor ID is the loggedInID field in the groupSup filter).

If a person is a memberof of multiple groups the supervisor manges, the record is being reported multiple times and the numbers are being inflated.

That is one of my test cases :( My question is how do I restructure this so if the supervisor manages multiple groups, all the people reporting to them are recorded only once - so a person belonging to 2 or more groups managed by the same supervisor only has their data reported once?

Thanks in advance!!

Answer 1

You maybe able to do this by just introducing another ID into the object model. So during your first piece of the statement:

from urls in this.ObjectContext.Activities.AsExpandable().Where(predicateA)
      .Where(x => x.GroupID == groupSelectionID)
      .Where(r => (r.StartDate >= beginDate && r.StartDate <= endDate) ||
            (r.EndDate >= beginDate && r.EndDate <= endDate) ||
            (r.StartDate <= beginDate && r.EndDate >= endDate))

In that solution, x is just the group that has some type of context. That would mean the urls collection should have objects that have a member property of GroupID.

Answer 2

This is what happens when you work too closely with something for too long. I finally figured it out - I broke the query down into 2 steps, first step gets a list of distinct identities and then uses that list to filter the query by identity.

So:

            var theIDsToInclude = (from id in this.ObjectContext.Identities
                                join idGroups in this.ObjectContext.IdentityGroups
                                    on id.IdentityID equals idGroups.IdentityID
                                join groupSup in this.ObjectContext.GroupSupervisors.Where(r => r.SupervisorID == loggedInID)
                                    on idGroups.GroupID equals groupSup.GroupID
                                select id.IdentityID).Distinct().ToList();


            theData = (from urls in this.ObjectContext.Activities.AsExpandable().Where(predicateA)
                                  .Where(r => (r.StartDate >= beginDate && r.StartDate <= endDate) ||
                                        (r.EndDate >= beginDate && r.EndDate <= endDate) ||
                                        (r.StartDate <= beginDate && r.EndDate >= endDate))
                                        .Where(r=> theIDsToInclude.Contains(r.IdentityID))
                       //join idGroups in this.ObjectContext.IdentityGroups on urls.IdentityID equals idGroups.IdentityID
                       //join groupSup in this.ObjectContext.GroupSupervisors.Where(r => r.SupervisorID == loggedInID) on idGroups.GroupID equals groupSup.GroupID
                       join programs in progs on urls.ProcessName.ToUpper() equals programs.ProcessName.ToUpper() into jt
                       from jt1 in jt.DefaultIfEmpty().Where(r => r == null || r.Ignore == false)
                       group urls by new { urls.ProcessName, urls.ContextID, jt1.CustomCategory, jt1.Name } into groupedTable
                       select new ActivityInfoSummary_DTO
                       {
                           recId = Guid.NewGuid(),
                           Context = groupedTable.Key.ProcessName,
                           ContextId = groupedTable.Key.ContextID,
                           FocusCount = groupedTable.Sum(r => r.FocusCount),
                           many more fields...
                       }).ToList();