Query to select a list of values from a table based on this multiple criteria

Question

I have a table with 2 columns, Ex_Id and Term_Id, both int type. My table will have many Term Ids for one Exercise Id.

     Table would look like this:
      Ex_Id Term_Id
         1     2
         1     3
         1     4
         1     5
         2     2
         3     2
         3     4

etc. Getting a list of Ex_Id is the primary requirement. My function would be like this.

List<int> Get_ExId_List(List<int> lst_TermId)
{
    // return a list of Ex_Id <int>
}

That is, I'll be passing a list of Term Ids and I need to get a list of Exercise Ids back matching some criteria. The criteria to select can be better explained with this pseudo-code: SELECT such Ex_Ids FROM table Exercise_Term WHERE Ex_Id has all the corresponding Term_Ids in the lst_TermId

For eg, from the sample table I provided above,

List<int> Get_ExId_List([2])
{
    // return [1,2,3]
}

List<int> Get_ExId_List([2,4])
{
    // return [1,3]
}

List<int> Get_ExId_List([2,3,4])
{
    // return [1]
}

Query part is my confusion. What would be the query in this condition like? Rest I can manage. Hope question is clear. Thanks..

ypercubeᵀᴹ · Accepted Answer · 2011-07-26T07:01:07.763

SELECT Ex_ID 
FROM TableName 
WHERE Term_ID IN (?, ?, ?)                --- (2, 3, 4)
GROUP BY Ex_ID
HAVING COUNT(DISTINCT Term_ID) = 3        --- number of terms in the above list

If the combination (Ex_ID, Term_ID) is unique in the table, you can replace COUNT(DISTINCT Term_ID) with COUNT(*)

This is a relational division problem. The "standard" solution would be using two negatives (NOT EXISTS):

SELECT DISTINCT Ex_ID
FROM TableName e
WHERE NOT EXISTS
        ( SELECT *
          FROM TableName t
          WHERE t.Term_ID IN (?, ?, ?)           --- the list of terms
            AND NOT EXISTS
                  ( SELECT *
                    FROM TableName a
                    WHERE a.Term_ID = t.Term_ID
                      AND a.Ex_ID = e.Ex_ID
                  )
        )

or better in your case:

SELECT DISTINCT Ex_ID
FROM TableName e
WHERE NOT EXISTS
        ( SELECT *
          FROM
            ( SELECT ? AS Term_ID  
            UNION
              SELECT ?
            UNION 
              SELECT ?
            ) AS t
          WHERE NOT EXISTS
                  ( SELECT *
                    FROM TableName a
                    WHERE a.Term_ID = t.Term_ID
                      AND a.Ex_ID = e.Ex_ID
                  )
        )

Gimme some time to test this. Thanks. This query is indeed tricky — nawfal, Jul 26 '11 at 06:40

score 1 · Answer 2 · answered Jul 26 '11 at 06:50

You can use LINQ. Get the whole table into an IEnumerable of some sort and then use LINQ. Here is an example:

static IEnumerable<int> Get_ExId_List(ICollection<int> lst_TermId)
{
    //this is just for the example - get the real data instead
    var data = new[] {
        new { Ex_Id = 1, Term_Id = 2},
        new { Ex_Id = 1, Term_Id = 3},
        new { Ex_Id = 1, Term_Id = 4},
        new { Ex_Id = 1, Term_Id = 5},
        new { Ex_Id = 2, Term_Id = 2},
        new { Ex_Id = 3, Term_Id = 2},
        new { Ex_Id = 3, Term_Id = 4},
    };

    return data
        .Where(row => lst_TermId.Contains(row.Term_Id))
        .GroupBy(row => row.Ex_Id)
        .Where(group => group.Count() == lst_TermId.Count())
        .Select(group => group.Key);
}

static void Main(string[] args)
{
    HashSet<int> lst_TermId = new HashSet<int>();
    lst_TermId.Add(2);

    Console.WriteLine();
    var result = Get_ExId_List(lst_TermId);
    foreach (var exid in result)
        Console.WriteLine(exid);

    lst_TermId.Add(4);

    Console.WriteLine();
    result = Get_ExId_List(lst_TermId);
    foreach (var exid in result)
        Console.WriteLine(exid);

    lst_TermId.Add(3);

    Console.WriteLine();
    result = Get_ExId_List(lst_TermId);
    foreach (var exid in result)
        Console.WriteLine(exid);
}

Note that you'll get better performance if your lst_TermId is a HashSet<int>, because the contains method will be O(1) instead of O(n).

But anyway this is useful for me, considering I at home use .Net 4 :) — nawfal, Jul 26 '11 at 08:02

Query to select a list of values from a table based on this multiple criteria

2 Answers2