实体框架避免多对多关系中的重复

Question

I've been stuck a little with the EF 6 database-first approach.

I have 3 tables in SQL Server:

• Members
• Teams
• JT_Teams_Members (junction table)

I would like to insert to data to this relations.

From the code, I get a list of teams that I iterated

Inserting:

foreach (var team in teams.Teams)
{
    var teamEntity = new Contexts.Team
            {
                TeamName = team.Name,
                ....
            };

    foreach (var member in team.Members)
    {
        teamEntity.Members.Add(
                new Contexts.Member()
                {
                    MemberName = member.Name,
                    EmailAddress = member.Email,  
                ...
                });
    }

    DbUtility.Context.Teams.Add(teamEntity);
}

My problem is I found a lot of duplication in the members table, which is not so good.

How can I manage the inserting to avoid the duplication and redirecting the ids to the properly place?

Answer 1

When dealing with associations, you will want to use the same reference for related entities rather than new-ing up a new reference for each related entity.

If I'm inserting a number of teams that contain members and some of those members may be new and shared across teams, while others could be existing members already in the database:

1. For all associated entities, always pass PKs in your view models. For members, if you can select from existing members, even if you only display a name and e-Mail address, pass the Member ID in your view model for later lookups.

From there we can start to look at something like below:

var memberIds = teams.Teams
    .SelectMany(t => t.Members)
    .Select(m => m.MemberId)
    .Where(m => m.MemberId != 0)
    .ToList();

var existingMembers = Context.Members
    .Where(m => memberIds.Contains(m.MemberId))
    .ToList();


foreach (var team in teams.Teams)
{
    var teamEntity = new Contexts.Team
    {
        TeamName = team.Name,
        ....
    };

    foreach (var member in team.Members)
    {   //We assume that all e-mail addresses are unique.
        var existingMember = existingMembers.SingleOrDefault(m => m.Email == member.Email);
        if (existingMember == null)
        {
            existingMember = new Member()
            {
                MemberName = member.Name,
                EmailAddress = member.Email,  
            ...
            });
            existingMembers.Add(existingMember); // For future reference.
        }
        teamEntity.Members.Add(existingMember);
    }

    DbUtility.Context.Teams.Add(teamEntity);
}

The first step is to look for any existing members for the teams. If you know that there will be no existing members then you can skip that load, and just use an initially empty collection. As we go through our teams we first check the existing collection for a member matching the e-mail address. If we find one, we use it, otherwise we create a new member and add it to the collection. That way when another team with the same member is added, it is found in the collection and the same reference will be used.

When EF inserts data, it respects references, not necessarily IDs/constraints to associate entities to the same reference rather than creating duplicates. If you have 2 entities that reference the same member (joe.smith@gmail.com) then you want to ensure that both entities reference the same member entity for joe.smith, not 2 different entities with the same e-mail address. When dealing with bulk operations (inserting many teams) you will want to keep a local cache or read from the DbContext to leverage its cache. For single operations you should always go to the DbContext to retrieve an existing entity if available before attempting to create a new reference. (To avoid duplicate data or constraint violations/duplicate key exceptions.)