I’m working on a method that retrieves a list of Student entities from a database based on several optional filtering and sorting criteria. The query is quite complex due to the following reasons:
Conditional Filtering:
Filtering by DivisionId, UserId, SiteId, StudentCategory, Status, and other fields depending on whether parameters are provided.
Some filters are combined using OR conditions, which complicates readability and maintainability.
Sorting:
Sorting by fields like EnrollmentNumber, StartDate, or Status, with support for ascending/descending order.
For EnrollmentNumber, I have a custom comparison logic that handles prefixed strings like “ABC-123”.
Pagination:
Skipping a specific number of records and limiting the results to a given page size.
Complex Joins and Includes:
The query involves several Include and ThenInclude calls to load related data such as StudentInfo, Statuses, and Enrollment.
My primary concerns are:
Performance: The query could be slow due to the complexity of filtering and sorting.
Readability: The code is getting hard to maintain and understand.
Optimization: I’m uncertain if I’m using the best practices for query building and executing.
public async Task<Result<List<Student>>> GetAll(GetStudentsQueryParametersDto queryParameters)
{
try
{
IQueryable<Student> query = _dbContext
.Students.Where(s => s.Enrollment != null && s.Enrollment.IsGroupEnrollment == false)
.Include(s => s.StudentInfo)
.Include(s => s.Enrollment);
if (queryParameters.ClassId.HasValue)
{
query = query.Where(s =>
(s.Enrollment != null && s.Enrollment.ClassId == queryParameters.ClassId)
|| (
s.StudentInfo != null
&& s.StudentInfo.StudentCategory != null
&& queryParameters.StudentCategories.Contains(s.StudentInfo.StudentCategory)
)
);
}
if (queryParameters.TeacherId.HasValue)
{
query = query.Where(s => s.Enrollment != null && s.Enrollment.TeacherId == queryParameters.TeacherId.Value);
}
if (queryParameters.SchoolId.HasValue)
{
query = query.Where(s => s.StudentInfo != null && s.StudentInfo.SchoolId == queryParameters.SchoolId.Value);
}
if (queryParameters.View != null)
{
query = FilterByDate(query, queryParameters.View);
}
if (queryParameters.EnrollmentType != null)
{
query = FilterByEnrollmentType(query, queryParameters.EnrollmentType);
}
query = query
.Include(s => s.StudentInfo)
.ThenInclude(i => i!.Statuses)
.ThenInclude(s => s.StatusReason)
.Include(s => s.StudentInfo)
.ThenInclude(i => i!.PreferredSchools)
.Include(s => s.FocusSubjects);
if (queryParameters.Status.Count != 0)
{
query = FilterByStatus(query, queryParameters.Status);
}
var studentsList = new List<Student>();
if (queryParameters.Sort != null && queryParameters.Sort.ToLower() != "enrollmentnumber")
{
var sortedQuery = ApplySorting(query, queryParameters.Sort, queryParameters.Direction);
var paginatedQuery = sortedQuery.Skip(queryParameters.Offset ?? 0).Take(queryParameters.Limit ?? 10);
studentsList = await paginatedQuery.ToListAsync();
}
else
{
studentsList = await query.ToListAsync();
var orderedStudents = SortByEnrollmentNumber(studentsList, queryParameters.Direction);
studentsList = orderedStudents.Skip(queryParameters.Offset ?? 0).Take(queryParameters.Limit ?? 10).ToList();
}
return Result<List<Student>>.Success(studentsList);
}
catch (BadRequestException e)
{
_logger.LogError(
e,
"Failed to get all students with pagination and parameters: {queryParameters}",
queryParameters
);
return Result<List<Student>>.Failure(Errors.BadRequest(e.Message));
}
catch (Exception e)
{
_logger.LogError(
e,
"Failed to get all students with pagination and parameters: {queryParameters}",
queryParameters
);
return Result<List<Student>>.Failure(Errors.InternalServerError);
}
}
These are the methods used:
private static IQueryable<Student> ApplySorting(IQueryable<Student> query, string? sortField, string? sortDirection)
{
var isDescending = sortDirection?.Equals("desc", StringComparison.OrdinalIgnoreCase) ?? false;
return sortField?.ToLower() switch
{
"startdate" => isDescending
? query.OrderByDescending(s => s.StudentInfo!.StartDate)
: query.OrderBy(s => s.StudentInfo!.StartDate),
"status" => isDescending
? query.OrderByDescending(s =>
s.StudentInfo!.Statuses.OrderByDescending(st => st.CreatedAt).FirstOrDefault()!.Status
)
: query.OrderBy(s =>
s.StudentInfo!.Statuses.OrderByDescending(st => st.CreatedAt).FirstOrDefault()!.Status
),
_ => isDescending
? query.OrderByDescending(s => s.StudentInfo!.StartDate)
: query.OrderBy(s => s.StudentInfo!.StartDate)
};
}
private static IOrderedEnumerable<Student> SortByEnrollmentNumber(
IEnumerable<Student> studentsList,
string? queryParametersDirection
)
{
if (!string.IsNullOrEmpty(queryParametersDirection) && queryParametersDirection.ToLower() == "desc")
{
return studentsList.OrderByDescending(s => s.Enrollment!.EnrollmentNumber, new EnrollmentNumberComparer());
}
return studentsList.OrderBy(s => s.Enrollment!.EnrollmentNumber, new EnrollmentNumberComparer());
}
public class EnrollmentNumberComparer : IComparer<string>
{
public int Compare(string? enrollmentNumber1, string? enrollmentNumber2)
{
ArgumentNullException.ThrowIfNull(enrollmentNumber1);
ArgumentNullException.ThrowIfNull(enrollmentNumber2);
if (enrollmentNumber1 == enrollmentNumber2)
return 0;
var enrollmentNumber1Parts = enrollmentNumber1.Split('-');
var enrollmentNumber2Parts = enrollmentNumber2.Split('-');
var enrollmentNumber1Prefix = enrollmentNumber1Parts[0];
var enrollmentNumber2Prefix = enrollmentNumber2Parts[0];
var prefixComparisonResult = enrollmentNumber1Prefix.CompareTo(enrollmentNumber2Prefix);
if (prefixComparisonResult != 0)
{
return prefixComparisonResult;
}
else
{
return int.Parse(enrollmentNumber1Parts[1]).CompareTo(int.Parse(enrollmentNumber2Parts[1]));
}
}
}
Overall, the approach is good. Conditionally adding filters etc. to the query if they are needed is far, far better than writing the conditional code into the Linq expression itself.
The first improvement to consider for search read operations like this is: “Do I need the entire entity graph?” Most of the time you will have a search result view that displays a row per result possibly with a summary of some related data like a count, or “flattening” information from a related entity. An improvement is to define a summary view model or DTO for the search result containing just the fields you intend to display, then use EF to project to that DTO using .Select()
. This results in a more lightweight block of data to read from the server.
If you do need entities then consider just setting up the .Include()
statements one time rather than some at the start and then others later on. To reference navigation properties in query expressions you do not need to eager load the entities.
The only real mistake I see is the sorting:
if (queryParameters.Sort != null && queryParameters.Sort.ToLower() != "enrollmentnumber")
{
var sortedQuery = ApplySorting(query, queryParameters.Sort, queryParameters.Direction);
var paginatedQuery = sortedQuery.Skip(queryParameters.Offset ?? 0).Take(queryParameters.Limit ?? 10);
studentsList = await paginatedQuery.ToListAsync();
}
else
{
studentsList = await query.ToListAsync();
var orderedStudents = SortByEnrollmentNumber(studentsList, queryParameters.Direction);
studentsList = orderedStudents.Skip(queryParameters.Offset ?? 0).Take(queryParameters.Limit ?? 10).ToList();
}
… specifically the else
condition which is materializing the entire list (bad) before applying a default sort and paginating. This conditional code can be re-factored to:
bool customSort = queryParameters.Sort != null && queryParameters.Sort.ToLower() != "enrollmentnumber";
var sortedQuery = customSort
? ApplySorting(query, queryParameters.Sort, queryParameters.Direction);
: SortByEnrollmentNumber(studentsList, queryParameters.Direction);
var studentList = await sortedQuery
.Skip(queryParameters.Offset ?? 0)
.Take(queryParameters.Limit ?? 10)
.ToListAsync();
This ensures the materialization does not occur prematurely.
Lastly, when inspecting performance use database analysis tools for query performance against live-like conditions to tune indexing. It likely won’t be possible to create indexes for all scenarios, but look at what is used most often and stands to get the most benefit. Overall the crux of searching is to aim to provide users with what they need rather than the most flexible option you can provide. The more flexibility a search feature provides, the more potential performance issues you can let in if users start abusing the search. Good examples to watch out for are things like using string.Contains()
by default for string searches rather than .Equals()
or .StartsWith()
. More forgiving search options can be provided where they are really needed, but keep those to situations where users explicitly need them, defaulting to more narrow search options being the default.