EntityFramework - contains query of composite key

Question

given a list of ids, I can query all relevant rows by:

context.Table.Where(q => listOfIds.Contains(q.Id));

But how do you achieve the same functionality when the Table has a composite key?

Answer 1

You can use Union for each composite primary key:

var compositeKeys = new List<CK> 
{
    new CK { id1 = 1, id2 = 2 },
    new CK { id1 = 1, id2 = 3 },
    new CK { id1 = 2, id2 = 4 }
};

IQueryable<CK> query = null;

foreach(var ck in compositeKeys)
{
    var temp = context.Table.Where(x => x.id1 == ck.id1 && x.id2 == ck.id2);
    query = query == null ? temp : query.Union(temp);
}

var result = query.ToList();

Answer 2

I tried it on EFCore 7.0 and 8.0 with Npgsql provider, use Tuples:

var filter = new List<Table>
{
    new() { X = "A", Y = 3 }, 
    new() { X = "B", Y = 4 }
};

var pairs = filter.Select(p => Tuple.Create(p.X, p.Y)).ToArray();

var data = db.Table.Where(x => pairs.Contains(new Tuple<string?, int?>(x.X, x.Y))).ToListAsync();

Translates to:

-- @__pairs_0={ '(A, 3)', '(B, 4)' } (DbType = Object)
SELECT t."X", t."Y"
FROM "Table" AS t
WHERE (t."X", t."Y") = ANY (@__pairs_0)

Answer 3

Here's a method to encapsulate Gert Arnold's "solution 5" for tuples of size two.

This method starts by breaking the set you're searching for into two lists. For example, if you're looking for rows that match columns P1 and P2 against the set of pairs [ (1, 8), (2, 9) ], it will compute items1 = [ 1, 2 ] and items2 = [ 8, 9 ]. Next, it constructs and runs a query that matches each list separately: query.Where(row => items1.Contains(row.P1) && items2.Contains(row.P2)).

The result of this DB query is then refined on the C# side by excluding inappropriate rows, e.g. the database could return a row with P1 = 1 and P2 = 9, which is excluded.

/// <summary>
///   Given a set of tuples like `set = new[] { (1, 8), (2, 9) }`, EF Core 
///   does not support a query like `query.Where(row => set.Contains((row.P1, row.P2)))`.
///   This method compensates for the limitation by allowing you to write
///   `query.WhereTupleListContains(set, x => x.P1, x => x.P2)` instead.
/// </summary>
/// <param name="set">The set of tuples to find. If the list may be long,
///   a HashSet is recommended.</param>
public static IEnumerable<T> WhereTupleListContains<T, A, B>(this IQueryable<T> query,
    ICollection<(A, B)> set, Expression<Func<T, A>> getExp1, Expression<Func<T, B>> getExp2)
{
    var items1 = set.Select(p => p.Item1).ToList();
    var items2 = set.Select(p => p.Item2).ToList();

    var paramT = Expression.Parameter(typeof(T), "t");

    // `getAexp` typically represents something like `x => x.Prop`. Get the `x.Prop`
    // part of that and replace `x` with paramT (`t`), then do the same for `getB`.
    Expression getAbody = ExpressionReplacer.Replace(
        getExp1.Body, getExp1.Parameters[0], paramT);
    Expression getBbody = ExpressionReplacer.Replace(
        getExp2.Body, getExp2.Parameters[0], paramT);

    // Now construct expressions like `Bs.Contains(<getBbody>)` and `As.Contains(<getAbody>)`
    var contains1 = Expression.Call(Expression.Constant(items1), "Contains", null, getAbody);
    var contains2 = Expression.Call(Expression.Constant(items2), "Contains", null, getBbody);

    // Finally, combine the two expressions to get something like
    // `As.Contains(<getAbody>) && Bs.Contains(<getAbody>)`
    var combined = Expression.AndAlso(contains1, contains2);
    var lambda = Expression.Lambda<Func<T, bool>>(combined, paramT);

    // Imprecise database query
    IEnumerable<T> impreciseResult = query.Where(lambda);

    // Precisely refine the result
    var get1 = getExp1.Compile();
    var get2 = getExp2.Compile();
    var refined = impreciseResult.Where(row => set.Contains((get1(row), get2(row))));
    return refined;
}

Answer 4

Solution for Entity Framework Core with SQL Server

🎉 NEW! QueryableValues EF6 Edition has arrived!

The following solution makes use of QueryableValues. This is a library that I wrote to primarily solve the problem of query plan cache pollution in SQL Server caused by queries that compose local values using the Contains LINQ method. It also allows you to compose values of complex types in your queries in a performant way, which will achieve what's being asked in this question.

First you will need to install and set up the library, after doing that you can use any of the following patterns that will allow you to query your entities using a composite key:

// Required to make the AsQueryableValues method available on the DbContext.
using BlazarTech.QueryableValues;

// Local data that will be used to query by the composite key
// of the fictitious OrderProduct table.
var values = new[]
{
    new { OrderId = 1, ProductId = 10 },
    new { OrderId = 2, ProductId = 20 },
    new { OrderId = 3, ProductId = 30 }
};

// Optional helper variable (needed by the second example due to CS0854)
var queryableValues = dbContext.AsQueryableValues(values);

// Example 1 - Using a Join (preferred).
var example1Results = dbContext
    .OrderProduct
    .Join(
        queryableValues,
        e => new { e.OrderId, e.ProductId },
        v => new { v.OrderId, v.ProductId },
        (e, v) => e
    )
    .ToList();

// Example 2 - Using Any (similar behavior as Contains).
var example2Results = dbContext
    .OrderProduct
    .Where(e => queryableValues
        .Where(v =>
            v.OrderId == e.OrderId &&
            v.ProductId == e.ProductId
        )
        .Any()
    )
    .ToList();

Useful Links

• Nuget Package
• GitHub Repository
• Benchmarks

QueryableValues is distributed under the MIT license.

Answer 5

This is a nasty problem for which I don't know any elegant solution.

Suppose you have these key combinations, and you only want to select the marked ones (*).

Id1  Id2
---  ---
1    2 *
1    3
1    6
2    2 *
2    3 *
... (many more)

How to do this is a way that Entity Framework is happy? Let's look at some possible solutions and see if they're any good.

Solution 1: Join (or Contains) with pairs

The best solution would be to create a list of the pairs you want, for instance Tuples, (List<Tuple<int,int>>) and join the database data with this list:

from entity in db.Table // db is a DbContext
join pair in Tuples on new { entity.Id1, entity.Id2 }
                equals new { Id1 = pair.Item1, Id2 = pair.Item2 }
select entity

In LINQ to objects this would be perfect, but, too bad, EF will throw an exception like

Unable to create a constant value of type 'System.Tuple`2 (...) Only primitive types or enumeration types are supported in this context.

which is a rather clumsy way to tell you that it can't translate this statement into SQL, because Tuples is not a list of primitive values (like int or string). For the same reason a similar statement using Contains (or any other LINQ statement) would fail.

Solution 2: In-memory

Of course we could turn the problem into simple LINQ to objects like so:

from entity in db.Table.AsEnumerable() // fetch db.Table into memory first
join pair Tuples on new { entity.Id1, entity.Id2 }
             equals new { Id1 = pair.Item1, Id2 = pair.Item2 }
select entity

Needless to say that this is not a good solution. db.Table could contain millions of records.

Solution 3: Two Contains statements (incorrect)

So let's offer EF two lists of primitive values, [1,2] for Id1 and [2,3] for Id2. We don't want to use join, so let's use Contains:

from entity in db.Table
where ids1.Contains(entity.Id1) && ids2.Contains(entity.Id2)
select entity

But now the results also contains entity {1,3}! Well, of course, this entity perfectly matches the two predicates. But let's keep in mind that we're getting closer. In stead of pulling millions of entities into memory, we now only get four of them.

Solution 4: One Contains with computed values

Solution 3 failed because the two separate Contains statements don't only filter the combinations of their values. What if we create a list of combinations first and try to match these combinations? We know from solution 1 that this list should contain primitive values. For instance:

var computed = ids1.Zip(ids2, (i1,i2) => i1 * i2); // [2,6]

and the LINQ statement:

from entity in db.Table
where computed.Contains(entity.Id1 * entity.Id2)
select entity

There are some problems with this approach. First, you'll see that this also returns entity {1,6}. The combination function (a*b) does not produce values that uniquely identify a pair in the database. Now we could create a list of strings like ["Id1=1,Id2=2","Id1=2,Id2=3]" and do

from entity in db.Table
where computed.Contains("Id1=" + entity.Id1 + "," + "Id2=" + entity.Id2)
select entity

(This would work in EF6, not in earlier versions).

This is getting pretty messy. But a more important problem is that this solution is not sargable, which means: it bypasses any database indexes on Id1 and Id2 that could have been used otherwise. This will perform very very poorly.

Solution 5: Best of 2 and 3

So the most viable solution I can think of is a combination of Contains and a join in memory: First do the contains statement as in solution 3. Remember, it got us very close to what we wanted. Then refine the query result by joining the result as an in-memory list:

var rawSelection = from entity in db.Table
                   where ids1.Contains(entity.Id1) && ids2.Contains(entity.Id2)
                   select entity;

var refined = from entity in rawSelection.AsEnumerable()
              join pair in Tuples on new { entity.Id1, entity.Id2 }
                              equals new { Id1 = pair.Item1, Id2 = pair.Item2 }
              select entity;

It's not elegant, messy all the same maybe, but so far it's the only scalable¹ solution to this problem I found, and applied in my own code.

Solution 6: Build a query with OR clauses

Using a Predicate builder like Linqkit or alternatives, you can build a query that contains an OR clause for each element in the list of combinations. This could be a viable option for really short lists. With a couple of hundreds of elements, the query will start performing very poorly. So I don't consider this a good solution unless you can be 100% sure that there will always be a small number of elements. One elaboration of this option can be found here.

Solution 7: Unions

There's also a solution using UNIONs that I posted later here.

---

¹As far as the Contains statement is scalable: Scalable Contains method for LINQ against a SQL backend

Answer 6

For EF Core I use a slightly modified version of the bucketized IN method by EricEJ to map composite keys as tuples. It performs pretty well for small sets of data.

Sample usage

List<(int Id, int Id2)> listOfIds = ...
context.Table.In(listOfIds, q => q.Id, q => q.Id2);

Implementation

public static IQueryable<TQuery> In<TKey1, TKey2, TQuery>(
            this IQueryable<TQuery> queryable,
            IEnumerable<(TKey1, TKey2)> values,
            Expression<Func<TQuery, TKey1>> key1Selector,
            Expression<Func<TQuery, TKey2>> key2Selector)
        {
            if (values is null)
            {
                throw new ArgumentNullException(nameof(values));
            }

            if (key1Selector is null)
            {
                throw new ArgumentNullException(nameof(key1Selector));
            }

            if (key2Selector is null)
            {
                throw new ArgumentNullException(nameof(key2Selector));
            }

            if (!values.Any())
            {
                return queryable.Take(0);
            }

            var distinctValues = Bucketize(values);

            if (distinctValues.Length > 1024)
            {
                throw new ArgumentException("Too many parameters for SQL Server, reduce the number of parameters", nameof(values));
            }

            var predicates = distinctValues
                .Select(v =>
                {
                    // Create an expression that captures the variable so EF can turn this into a parameterized SQL query
                    Expression<Func<TKey1>> value1AsExpression = () => v.Item1;
                    Expression<Func<TKey2>> value2AsExpression = () => v.Item2;
                    var firstEqual = Expression.Equal(key1Selector.Body, value1AsExpression.Body);

                    var visitor = new ReplaceParameterVisitor(key2Selector.Parameters[0], key1Selector.Parameters[0]);

                    var secondEqual = Expression.Equal(visitor.Visit(key2Selector.Body), value2AsExpression.Body);

                    return Expression.AndAlso(firstEqual, secondEqual);
                })
                .ToList();

            while (predicates.Count > 1)
            {
                predicates = PairWise(predicates).Select(p => Expression.OrElse(p.Item1, p.Item2)).ToList();
            }

            var body = predicates.Single();

            var clause = Expression.Lambda<Func<TQuery, bool>>(body, key1Selector.Parameters[0]);

            return queryable.Where(clause);
        }

        class ReplaceParameterVisitor : ExpressionVisitor
        {
            private ParameterExpression _oldParameter;
            private ParameterExpression _newParameter;

            public ReplaceParameterVisitor(ParameterExpression oldParameter, ParameterExpression newParameter)
            {
                _oldParameter = oldParameter;
                _newParameter = newParameter;
            }

            protected override Expression VisitParameter(ParameterExpression node)
            {
                if (ReferenceEquals(node, _oldParameter))
                    return _newParameter;

                return base.VisitParameter(node);
            }
        }

        /// <summary>
        /// Break a list of items tuples of pairs.
        /// </summary>
        private static IEnumerable<(T, T)> PairWise<T>(this IEnumerable<T> source)
        {
            var sourceEnumerator = source.GetEnumerator();
            while (sourceEnumerator.MoveNext())
            {
                var a = sourceEnumerator.Current;
                sourceEnumerator.MoveNext();
                var b = sourceEnumerator.Current;

                yield return (a, b);
            }
        }

        private static TKey[] Bucketize<TKey>(IEnumerable<TKey> values)
        {
            var distinctValueList = values.Distinct().ToList();

            // Calculate bucket size as 1,2,4,8,16,32,64,...
            var bucket = 1;
            while (distinctValueList.Count > bucket)
            {
                bucket *= 2;
            }

            // Fill all slots.
            var lastValue = distinctValueList.Last();
            for (var index = distinctValueList.Count; index < bucket; index++)
            {
                distinctValueList.Add(lastValue);
            }

            var distinctValues = distinctValueList.ToArray();
            return distinctValues;
        }

Answer 7

I tried this on EF Core 5.0.3 with the Postgres provider.

context.Table
    .Select(entity => new
    {
        Entity = entity,
        CompositeKey = entity.Id1 + entity.Id2,
    })
    .Where(x => compositeKeys.Contains(x.CompositeKey))
    .Select(x => x.Entity);

This produced SQL like:

SELECT *
FROM table AS t
WHERE t.Id1 + t.Id2 IN (@__compositeKeys_0)), 

Caveats

• this should only be used where the combination of Id1 and Id2 will always produce a unique result (e.g., they're both UUIDs)
• this cannot use indexes, though you could save the composite key to the db with an index

Answer 8

Ran into this issue as well and needed a solution that both did not perform a table scan and also provided exact matches.

This can be achieved by combining Solution 3 and Solution 4 from Gert Arnold's Answer

var firstIds = results.Select(r => r.FirstId);
var secondIds = results.Select(r => r.SecondId);
var compositeIds = results.Select(r => $"{r.FirstId}:{r.SecondId}");
var query = from e in dbContext.Table
 
            //first check the indexes to avoid a table scan
            where firstIds.Contains(e.FirstId) && secondIds.Contains(e.SecondId))
 
            //then compare the compositeId for an exact match
            //ToString() must be called unless using EF Core 5+
            where compositeIds.Contains(e.FirstId.ToString() + ":" + e.SecondId.ToString()))
            select e;
var entities = await query.ToListAsync();

Answer 9

I ended up writing a helper for this problem that relies on System.Linq.Dynamic.Core;

Its a lot of code and don't have time to refactor at the moment but input / suggestions appreciated.

        public static IQueryable<TEntity> WhereIsOneOf<TEntity, TSource>(this IQueryable<TEntity> dbSet, 
            IEnumerable<TSource> source, 
            Expression<Func<TEntity, TSource,bool>> predicate) where TEntity : class
        {
            var (where, pDict) = GetEntityPredicate(predicate, source);
            return dbSet.Where(where, pDict);

            (string WhereStr, IDictionary<string, object> paramDict) GetEntityPredicate(Expression<Func<TEntity, TSource, bool>> func, IEnumerable<TSource> source)
            {
                var firstP = func.Parameters[0];
                var binaryExpressions = RecurseBinaryExpressions((BinaryExpression)func.Body);

                var i = 0;
                var paramDict = new Dictionary<string, object>();
                var res = new List<string>();
                foreach (var sourceItem in source)
                {
                    var innerRes = new List<string>();
                    foreach (var bExp in binaryExpressions)
                    {
                        var emp = ToEMemberPredicate(firstP, bExp);
                        var val = emp.GetKeyValue(sourceItem);
                        var pName = $"@{i++}";
                        paramDict.Add(pName, val);
                        var str = $"{emp.EntityMemberName} {emp.SQLOperator} {pName}";
                        innerRes.Add(str);
                    }

                    res.Add( "(" + string.Join(" and ", innerRes) + ")");
                }

                var sRes = string.Join(" || ", res);

                return (sRes, paramDict);
            }
            
            EMemberPredicate ToEMemberPredicate(ParameterExpression firstP, BinaryExpression bExp)
            {
                var lMember = (MemberExpression)bExp.Left;
                var rMember = (MemberExpression)bExp.Right;

                var entityMember = lMember.Expression == firstP ? lMember : rMember;
                var keyMember = entityMember == lMember ? rMember : lMember;

                return new EMemberPredicate(entityMember, keyMember, bExp.NodeType);
            }
            List<BinaryExpression> RecurseBinaryExpressions(BinaryExpression e, List<BinaryExpression> runningList = null)
            {
                if (runningList == null) runningList = new List<BinaryExpression>();

                if (e.Left is BinaryExpression lbe)
                {
                    var additions = RecurseBinaryExpressions(lbe);
                    runningList.AddRange(additions);
                }
                
                if (e.Right is BinaryExpression rbe)
                {
                    var additions = RecurseBinaryExpressions(rbe);
                    runningList.AddRange(additions);
                }

                if (e.Left is MemberExpression && e.Right is MemberExpression)
                {
                    runningList.Add(e);
                }

                return runningList;
            }
        }

Helper class:

    public class EMemberPredicate
    {
        public readonly MemberExpression EntityMember;
        public readonly MemberExpression KeyMember;
        public readonly PropertyInfo KeyMemberPropInfo;
        public readonly string EntityMemberName;
        public readonly string SQLOperator;

        public EMemberPredicate(MemberExpression entityMember, MemberExpression keyMember, ExpressionType eType)
        {
            EntityMember = entityMember;
            KeyMember = keyMember;
            KeyMemberPropInfo = (PropertyInfo)keyMember.Member;
            EntityMemberName = entityMember.Member.Name;
            SQLOperator = BinaryExpressionToMSSQLOperator(eType);
        }

        public object GetKeyValue(object o)
        {
            return KeyMemberPropInfo.GetValue(o, null);
        }

        private string BinaryExpressionToMSSQLOperator(ExpressionType eType)
        {
            switch (eType)
            {
                case ExpressionType.Equal:
                    return "==";
                case ExpressionType.GreaterThan:
                    return ">";
                case ExpressionType.GreaterThanOrEqual:
                    return ">=";
                case ExpressionType.LessThan:
                    return "<";
                case ExpressionType.LessThanOrEqual:
                    return "<=";
                case ExpressionType.NotEqual:
                    return "<>";
                default:
                    throw new ArgumentException($"{eType} is not a handled Expression Type.");
            }
        }
    }

Use Like so:

// This can be a Tuple or whatever..  If Tuple, then y below would be .Item1, etc.
// This data structure is up to you but is what I use.
[FromBody] List<CustomerAddressPk> cKeys

            var res = await dbCtx.CustomerAddress
                .WhereIsOneOf(cKeys, (x, y) => y.CustomerId == x.CustomerId 
                   && x.AddressId == y.AddressId)
                .ToListAsync();

Hope this helps others.

Answer 10

I tried this solution and it worked with me and the output query was perfect without any parameters

using LinqKit; // nuget     
   var customField_Ids = customFields?.Select(t => new CustomFieldKey { Id = t.Id, TicketId = t.TicketId }).ToList();

    var uniqueIds1 = customField_Ids.Select(cf => cf.Id).Distinct().ToList();
    var uniqueIds2 = customField_Ids.Select(cf => cf.TicketId).Distinct().ToList();
    var predicate = PredicateBuilder.New<CustomFieldKey>(false); //LinqKit
    var lambdas = new List<Expression<Func<CustomFieldKey, bool>>>();
    foreach (var cfKey in customField_Ids)
    {
        var id = uniqueIds1.Where(uid => uid == cfKey.Id).Take(1).ToList();
        var ticketId = uniqueIds2.Where(uid => uid == cfKey.TicketId).Take(1).ToList();
        lambdas.Add(t => id.Contains(t.Id) && ticketId.Contains(t.TicketId));
    }

    predicate = AggregateExtensions.AggregateBalanced(lambdas.ToArray(), (expr1, expr2) =>
     {
         var invokedExpr = Expression.Invoke(expr2, expr1.Parameters.Cast<Expression>());
         return Expression.Lambda<Func<CustomFieldKey, bool>>
               (Expression.OrElse(expr1.Body, invokedExpr), expr1.Parameters);
     });


    var modifiedCustomField_Ids = repository.GetTable<CustomFieldLocal>()
         .Select(cf => new CustomFieldKey() { Id = cf.Id, TicketId = cf.TicketId }).Where(predicate).ToArray();

Answer 11

In the absence of a general solution, I think there are two things to consider:

1. Avoid multi-column primary keys (will make unit testing easier too).
2. But if you have to, chances are that one of them will reduce the query result size to O(n) where n is the size of the ideal query result. From here, its Solution 5 from Gerd Arnold above.

For example, the problem leading me to this question was querying order lines, where the key is order id + order line number + order type, and the source had the order type being implicit. That is, the order type was a constant, order ID would reduce the query set to order lines of relevant orders, and there would usually be 5 or less of these per order.

To rephrase: If you have a composite key, changes are that one of them have very few duplicates. Apply Solution 5 from above with that.

Answer 12

You can create a collection of strings with both keys like this (I am assuming that your keys are int type):

var id1id2Strings = listOfIds.Select(p => p.Id1+ "-" + p.Id2);

Then you can just use "Contains" on your db:

using (dbEntities context = new dbEntities())
            {
                var rec = await context.Table1.Where(entity => id1id2Strings .Contains(entity.Id1+ "-" + entity.Id2));
                return rec.ToList();
            }

Answer 13

You need a set of objects representing the keys you want to query.

class Key
{
    int Id1 {get;set;}
    int Id2 {get;set;}

If you have two lists and you simply check that each value appears in their respective list then you are getting the cartesian product of the lists - which is likely not what you want. Instead you need to query the specific combinations required

List<Key> keys = // get keys;

context.Table.Where(q => keys.Any(k => k.Id1 == q.Id1 && k.Id2 == q.Id2)); 

I'm not completely sure that this is valid use of Entity Framework; you may have issues with sending the Key type to the database. If that happens then you can be creative:

var composites = keys.Select(k => p1 * k.Id1 + p2 * k.Id2).ToList();
context.Table.Where(q => composites.Contains(p1 * q.Id1 + p2 * q.Id2)); 

You can create an isomorphic function (prime numbers are good for this), something like a hashcode, which you can use to compare the pair of values. As long as the multiplicative factors are co-prime this pattern will be isomorphic (one-to-one) - i.e. the result of p1*Id1 + p2*Id2 will uniquely identify the values of Id1 and Id2 as long as the prime numbers are correctly chosen.

But then you end up in a situation where you're implementing complex concepts and someone is going to have to support this. Probably better to write a stored procedure which takes the valid key objects.

Answer 14

in Case of composite key you can use another idlist and add a condition for that in your code

context.Table.Where(q => listOfIds.Contains(q.Id) && listOfIds2.Contains(q.Id2));

or you can use one another trick create a list of your keys by adding them

listofid.add(id+id1+......)
context.Table.Where(q => listOfIds.Contains(q.Id+q.id1+.......));