如何比较清单 <long> 使用逗号分隔的ID（使用linq实体）

Question

I have Interest Ids column in user profile. & a List<long> interest ids . I want to fetch user profiles matching interest ids in List using Entity Framework.

Ex.

• Sam 1,5,9,13,4,8
• John 2,7,13,9
• Kettie 1,4,8,12,15

List: {4,8}

I want output as {Sam,Kettie}

Update: Db Structure-

public class UserProfile
{    
    public long UserId { get; set; }
    public string FullName { get; set; }
    public string Interests { get; set; }  //Store comma separated Interest Ids here 
}

public class Interest
{
    public long InterestId { get; set; }
    public string Name { get; set; }
}

I achieve this by

var interestIds = db.Interests.Where(i => i.Name.Contains(query))
        .Select(i=> i.InterestId)
        .ToList();
var profiles = new List<UserProfile>();
foreach (var id in interestIds)
{
    profiles.AddRange(db.UserProfiles
        .Where(p=> p.Interests.Contains(id.ToString()))
        .ToList());
}
return profiles;

But, when dealing with huge records it takes long time to execute so I want help to optimise this.

Answer 1

Apologies if the answer is not exactly what you are looking for as I'm unfamiliar with entity-framework, but looking at it purely from a C# point of view:

For every id in your interestIds you are filtering your entire database every time (and doing this by checking a collection on each entry) - I would imagine you are incurring a performance penalty here.

If possible, you could map each interest id to a list of user names. This only needs to be done once and may perform better - something like:

var dict = new Dictionary<long, List<string>>();
foreach(var user in userProfiles)
{
    foreach(var interest in user.Interests)
    {
        List<string> names;
        if(dict.TryGetValue(interest, out names))
            names.Add(user.Name);
        else
            dict.Add(interest, new[] { user.Name }.ToList());
    }
}

long[] interestIds = new[] { 4, 8 };

HashSet<string> profiles = new HashSet<string>();
foreach (var interestId in interestIds)
    profiles.UnionWith(dict[interestId]);

So iterating over every interest for every user, you then add each interest as a key and add the user to the list of users for that key.

You can then iterate over your list of interestIds and pull the matching list out user names out of the dictionary (note the HashSet - this will stop you getting duplicate names for users who match more than one interest).

Answer 2

This will help you to execute your query in sql server only once .. I checked it using profiler and it's better because when foreach loop iterate .. it open and close the sql the connection each time.. which leads to populate your desire result much slower..

        var interestIds = db.Interests.Where(i => i.Name.Contains("interest_name"))
                        .Select(i => i.InterestId)
                        .ToList().ConvertAll(i => i.ToString());
        //var profiles = new List<UserProfile>();

        var allProfiles = (from UserProfile up in db.UserProfiles
                           from i in interestIds
                           where up.Interests.Contains(i)
                           select up).ToList();
        return allProfiles;

        //string sId = null;
        //foreach (var id in interestIds)
        //{
        //    sId = id.ToString();
        //    profiles.AddRange(db.UserProfiles
        //        .Where(p => p.Interests.Contains(sId))
        //        .ToList());
        //}
        //return profiles;