How would the following SQL-query translate to LINQ:
SELECT TOP 2 A.* FROM UserInterests AS UIa
INNER JOIN UserInterests AS UIb ON UIb.InterestId = UIa.InterestId
INNER JOIN AspNetUsers AS A ON A.Id = UIb.ApplicationUserId
WHERE UIa.ApplicationUserId = {userId}
AND NOT UIb.ApplicationUserId = {userId}
AND (
A.Location LIKE {locationSubString "%"} OR
A.Location LIKE {neighbours[0] "%"} OR
A.Location LIKE {neighbours[1] "%"} OR
A.Location LIKE {neighbours[2] "%"} OR
A.Location LIKE {neighbours[3] "%"} OR
A.Location LIKE {neighbours[4] "%"} OR
A.Location LIKE {neighbours[5] "%"} OR
A.Location LIKE {neighbours[6] "%"} OR
A.Location LIKE {neighbours[7] "%"}
)
ORDER BY NEWID()
I am selecting two random users who are near me and who have the same interests. The query above is executed with DbSet.FromSqlInterpolated. The userId-variable is my id and the locationSubString neighbours-array are geohashes.
I have the following, but I am not sure how I would match the interests:
dbContext.Users.Where(u => (
u.Location.StartsWith(locationSubString) ||
u.Location.StartsWith(neighbours[0]) ||
u.Location.StartsWith(neighbours[1]) ||
u.Location.StartsWith(neighbours[2]) ||
u.Location.StartsWith(neighbours[3]) ||
u.Location.StartsWith(neighbours[4]) ||
u.Location.StartsWith(neighbours[5]) ||
u.Location.StartsWith(neighbours[6]) ||
u.Location.StartsWith(neighbours[7])
) && u.Id != userId
).Include(u => u.Interests)
Can this be done in one query? Or do I need to query my own interests first and then compare it with that list like so (myInterests is a list):
...
.Include(u => u.Interests)
.Where(u => u.Interests.Any(i => myInterests.Contains(i)))
CodePudding user response:
I have come up with the following:
await dbContext.Users
.Where(u =>
u.Location != null && (
u.Location.StartsWith(locationSubString) ||
u.Location.StartsWith(neighbours[0]) ||
u.Location.StartsWith(neighbours[1]) ||
u.Location.StartsWith(neighbours[2]) ||
u.Location.StartsWith(neighbours[3]) ||
u.Location.StartsWith(neighbours[4]) ||
u.Location.StartsWith(neighbours[5]) ||
u.Location.StartsWith(neighbours[6]) ||
u.Location.StartsWith(neighbours[7])
) && u.Id != userId
)
.Include(u => u.Interests)
.Where(u => u.Interests
.Any(i => applicationUser.Interests
.Select(i => i.InterestId)
.Contains(i.InterestId)
)
)
.OrderBy(u => Guid.NewGuid())
.Take(2)
I haven't found a way to get all the info in one query, so I looked up the user's interests before executing this query and then check all.
Not necessarily an answer to the question, but I also made Location a non-clustered index, to speed up the query.
CodePudding user response:
Here is the matching code I came up with:
var currentPersonId = 1;
var data = context.UserInterests
.Join(context.UserInterests, x => x.InterestID, x => x.InterestID, (a, b) => new { CurrentInterest = a, MatchedInterest = b })
.Where(x => x.CurrentInterest.UserID == currentPersonId && x.MatchedInterest.UserID != currentPersonId)
.Select(x => x.MatchedInterest.User)
.Distinct()
.Where(x => x.Location.StartsWith("a"))
.OrderBy(u => Guid.NewGuid())
.Take(2)
.ToArray();
You can see it work on sample data here.