SQL Queries
Entity Framework Core allows you to drop down to SQL queries when working with a relational database. SQL queries are useful if the query you want can't be expressed using LINQ, or if a LINQ query causes EF to generate inefficient SQL. SQL queries can return regular entity types or keyless entity types that are part of your model.
Tip
You can view this article's sample on GitHub.
Basic SQL queries
You can use FromSql to begin a LINQ query based on a SQL query:
var blogs = context.Blogs
.FromSql($"SELECT * FROM dbo.Blogs")
.ToList();
Note
FromSql was introduced in EF Core 7.0. When using older versions, use FromSqlInterpolated instead.
SQL queries can be used to execute a stored procedure which returns entity data:
var blogs = context.Blogs
.FromSql($"EXECUTE dbo.GetMostPopularBlogs")
.ToList();
Note
FromSql can only be used directly on a DbSet
. It cannot be composed over an arbitrary LINQ query.
Passing parameters
Warning
Pay close attention to parameterization when using SQL queries
When introducing any user-provided values into a SQL query, care must be taken to avoid SQL injection attacks. SQL injection occurs when a program integrates a user-provided string value into a SQL query, and the user-provided value is crafted to terminate the string and perform another malicious SQL operation. To learn more about SQL injection, see this page.
The FromSql and FromSqlInterpolated methods are safe against SQL injection, and always integrate parameter data as a separate SQL parameter. However, the FromSqlRaw method can be vulnerable to SQL injection attacks, if improperly used. See below for more details.
The following example passes a single parameter to a stored procedure by including a parameter placeholder in the SQL query string and providing an additional argument:
var user = "johndoe";
var blogs = context.Blogs
.FromSql($"EXECUTE dbo.GetMostPopularBlogsForUser {user}")
.ToList();
While this syntax may look like regular C# string interpolation, the supplied value is wrapped in a DbParameter
and the generated parameter name inserted where the {0}
placeholder was specified. This makes FromSql safe from SQL injection attacks, and sends the value efficiently and correctly to the database.
When executing stored procedures, it can be useful to use named parameters in the SQL query string, especially when the stored procedure has optional parameters:
var user = new SqlParameter("user", "johndoe");
var blogs = context.Blogs
.FromSql($"EXECUTE dbo.GetMostPopularBlogsForUser @filterByUser={user}")
.ToList();
If you need more control over the database parameter being sent, you can also construct a DbParameter
and supply it as a parameter value. This allows you to set the precise database type of the parameter, or facets such as its size, precision or length:
var user = new SqlParameter("user", "johndoe");
var blogs = context.Blogs
.FromSql($"EXECUTE dbo.GetMostPopularBlogsForUser {user}")
.ToList();
Note
The parameters you pass must exactly match the stored procedure definition. Pay special attention to the ordering of the parameters, taking care to not miss or misplace any of them - or consider using named parameter notation. Also, make sure the parameter types correspond, and that their facets (size, precision, scale) are set as needed.
Dynamic SQL and parameters
FromSql and its parameterization should be used wherever possible. However, there are certain scenarios where SQL needs to be dynamically pieced together, and database parameters cannot be used. For example, let's assume that a C# variable holds the name of the a property to be filtered by. It may be tempting to use a SQL query such as the following:
var propertyName = "User";
var propertyValue = "johndoe";
var blogs = context.Blogs
.FromSql($"SELECT * FROM [Blogs] WHERE {propertyName} = {propertyValue}")
.ToList();
This code doesn't work, since databases do not allow parameterizing column names (or any other part of the schema).
First, it's important to consider the implications of dynamically constructing a query - via SQL or otherwise. Accepting a column name from a user may allow them to choose a column that isn't indexed, making the query run extremely slowly and overload your database; or it may allow them to choose a column containing data you don't want exposed. Except for truly dynamic scenarios, it's usually better to have two queries for two column names, rather than using parameterization to collapse them into a single query.
If you've decided you do want to dynamically construct your SQL, you'll have to use FromSqlRaw, which allows interpolating variable data directly into the SQL string, instead of using a database parameter:
var columnName = "Url";
var columnValue = new SqlParameter("columnValue", "http://SomeURL");
var blogs = context.Blogs
.FromSqlRaw($"SELECT * FROM [Blogs] WHERE {columnName} = @columnValue", columnValue)
.ToList();
In the above code, the column name is inserted directly into the SQL, using C# string interpolation. It is your responsibility to make sure this string value is safe, sanitizing it if it comes from an unsafe origin; this means detecting special characters such as semicolons, comments, and other SQL constructs, and either escaping them properly or rejecting such inputs.
On the other hand, the column value is sent via a DbParameter
, and is therefore safe in the face of SQL injection.
Warning
Be very careful when using FromSqlRaw, and always make sure values are either from a safe origin, or are properly sanitized. SQL injection attacks can have disasterous consequences for your application.
Composing with LINQ
You can compose on top of the initial SQL query using LINQ operators; EF Core will treat your SQL as a subquery and compose over it in the database. The following example uses a SQL query that selects from a Table-Valued Function (TVF). And then composes on it using LINQ to do filtering and sorting.
var searchTerm = "Lorem ipsum";
var blogs = context.Blogs
.FromSql($"SELECT * FROM dbo.SearchBlogs({searchTerm})")
.Where(b => b.Rating > 3)
.OrderByDescending(b => b.Rating)
.ToList();
The above query generates the following SQL:
SELECT [b].[BlogId], [b].[OwnerId], [b].[Rating], [b].[Url]
FROM (
SELECT * FROM dbo.SearchBlogs(@p0)
) AS [b]
WHERE [b].[Rating] > 3
ORDER BY [b].[Rating] DESC
Including related data
The Include
operator can be used to load related data, just like with any other LINQ query:
var searchTerm = "Lorem ipsum";
var blogs = context.Blogs
.FromSql($"SELECT * FROM dbo.SearchBlogs({searchTerm})")
.Include(b => b.Posts)
.ToList();
Composing with LINQ requires your SQL query to be composable, since EF Core will treat the supplied SQL as a subquery. Composable SQL queries generally begin with the SELECT
keyword, and cannot contain SQL features that aren't valid in a subquery, such as:
- A trailing semicolon
- On SQL Server, a trailing query-level hint (for example,
OPTION (HASH JOIN)
) - On SQL Server, an
ORDER BY
clause that isn't used withOFFSET 0
ORTOP 100 PERCENT
in theSELECT
clause
SQL Server doesn't allow composing over stored procedure calls, so any attempt to apply additional query operators to such a call will result in invalid SQL. Use AsEnumerable or AsAsyncEnumerable right after FromSql or FromSqlRaw to make sure that EF Core doesn't try to compose over a stored procedure.
Change Tracking
Queries that use FromSql or FromSqlRaw follow the exact same change tracking rules as any other LINQ query in EF Core. For example, if the query projects entity types, the results are tracked by default.
The following example uses a SQL query that selects from a Table-Valued Function (TVF), then disables change tracking with the call to AsNoTracking
:
var searchTerm = "Lorem ipsum";
var blogs = context.Blogs
.FromSql($"SELECT * FROM dbo.SearchBlogs({searchTerm})")
.AsNoTracking()
.ToList();
Querying scalar (non-entity) types
Note
This feature was introduced in EF Core 7.0.
While FromSql is useful for querying entities defined in your model, SqlQuery allows you to easily query for scalar, non-entity types via SQL, without needing to drop down to lower-level data access APIs. For example, the following query fetches all the IDs from the Blogs
table:
var ids = context.Database
.SqlQuery<int>($"SELECT [BlogId] FROM [Blogs]")
.ToList();
You can also compose LINQ operators over your SQL query. However, since your SQL becomes a subquery whose output column needs to be referenced by the SQL EF adds, you must name the output column Value
. For example, the following query returns the IDs which are above the ID average:
var overAverageIds = context.Database
.SqlQuery<int>($"SELECT [BlogId] AS [Value] FROM [Blogs]")
.Where(id => id > context.Blogs.Average(b => b.BlogId))
.ToList();
FromSql can be used with any scalar type supported by your database provider. If you'd like to use a type not supported by your database provider, you can use pre-convention configuration to define a value conversion for it.
SqlQueryRaw allows for dynamic construction of SQL queries, just like FromSqlRaw does for entity types.
Executing non-querying SQL
In some scenarios, it may be necessary to execute SQL which does not return any data, typically for modifying data in the database or calling a stored procedure which doesn't return any result sets. This can be done via ExecuteSql:
using (var context = new BloggingContext())
{
var rowsModified = context.Database.ExecuteSql($"UPDATE [Blogs] SET [Url] = NULL");
}
This executes the provided SQL and returns the number of rows modified. ExecuteSql protects against SQL injection by using safe parameterization, just like FromSql, and ExecuteSqlRaw allows for dynamic construction of SQL queries, just like FromSqlRaw does for queries.
Note
Prior to EF Core 7.0, it was sometimes necessary to use the ExecuteSql
APIs to perform a "bulk update" on the database, as above; this is considerably more efficient than querying for all matching rows and then using SaveChanges
to modify them. EF Core 7.0 introduced ExecuteUpdate and ExecuteDelete, which made it possible to express efficient bulk update operations via LINQ. It's recommended to use those APIs whenever possible, instead of ExecuteSql
.
Limitations
There are a few limitations to be aware of when returning entity types from SQL queries:
- The SQL query must return data for all properties of the entity type.
- The column names in the result set must match the column names that properties are mapped to. Note that this behavior is different from EF6; EF6 ignored property-to-column mapping for SQL queries, and result set column names had to match those property names.
- The SQL query can't contain related data. However, in many cases you can compose on top of the query using the
Include
operator to return related data (see Including related data).