我更新了扩展方法从@AminRostami返回IAsyncEnumerable(这样LINQ过滤可以应用)，它的映射模型列名的记录从DB返回到模型(EF Core 5测试):

扩展本身:

public static class QueryHelper
{
    private static string GetColumnName(this MemberInfo info)
    {
        List<ColumnAttribute> list = info.GetCustomAttributes<ColumnAttribute>().ToList();
        return list.Count > 0 ? list.Single().Name : info.Name;
    }
    /// <summary>
    /// Executes raw query with parameters and maps returned values to column property names of Model provided.
    /// Not all properties are required to be present in model (if not present - null)
    /// </summary>
    public static async IAsyncEnumerable<T> ExecuteQuery<T>(
        [NotNull] this DbContext db,
        [NotNull] string query,
        [NotNull] params SqlParameter[] parameters)
        where T : class, new()
    {
        await using DbCommand command = db.Database.GetDbConnection().CreateCommand();
        command.CommandText = query;
        command.CommandType = CommandType.Text;
        if (parameters != null)
        {
            foreach (SqlParameter parameter in parameters)
            {
                command.Parameters.Add(parameter);
            }
        }
        await db.Database.OpenConnectionAsync();
        await using DbDataReader reader = await command.ExecuteReaderAsync();
        List<PropertyInfo> lstColumns = new T().GetType()
            .GetProperties(BindingFlags.DeclaredOnly | BindingFlags.Instance | BindingFlags.Public | BindingFlags.NonPublic).ToList();
        while (await reader.ReadAsync())
        {
            T newObject = new();
            for (int i = 0; i < reader.FieldCount; i++)
            {
                string name = reader.GetName(i);
                PropertyInfo prop = lstColumns.FirstOrDefault(a => a.GetColumnName().Equals(name));
                if (prop == null)
                {
                    continue;
                }
                object val = await reader.IsDBNullAsync(i) ? null : reader[i];
                prop.SetValue(newObject, val, null);
            }
            yield return newObject;
        }
    }
}

使用的模型(注意列名与实际的属性名不同):

public class School
{
    [Key] [Column("SCHOOL_ID")] public int SchoolId { get; set; }

    [Column("CLOSE_DATE", TypeName = "datetime")]
    public DateTime? CloseDate { get; set; }

    [Column("SCHOOL_ACTIVE")] public bool? SchoolActive { get; set; }
}

实际的用法:

public async Task<School> ActivateSchool(int schoolId)
{
    // note that we're intentionally not returning "SCHOOL_ACTIVE" with select statement
    // this might be because of certain IF condition where we return some other data
    return await _context.ExecuteQuery<School>(
        "UPDATE SCHOOL SET SCHOOL_ACTIVE = 1 WHERE SCHOOL_ID = @SchoolId; SELECT SCHOOL_ID, CLOSE_DATE FROM SCHOOL",
        new SqlParameter("@SchoolId", schoolId)
    ).SingleAsync();
}

你也可以使用QueryFirst。和Dapper一样，这完全不在EF的范围内。与Dapper(或EF)不同的是，您不需要维护POCO，只需在真实环境中编辑sql sql，并根据DB不断对其进行重新验证。免责声明:我是QueryFirst的作者。

我找到了EntityFrameworkCore包。RawSQLExtensions在github上。要使用它，添加nuget包。

<PackageReference Include="EntityFrameworkCore.RawSQLExtensions" Version="1.2.0" />

这个库没有文档，但下面是我在。net 6 + EF Core 6 + Npgsql 6中使用它

public class DbResult
{
    public string Name { get; set; }
    public int Age { get; set; }
}

using EntityFrameworkCore.RawSQLExtensions.Extensions;

var results = await context.Database
    .SqlQuery<DbResult>(
        @"select name, age from ""users"" where age > @Age",
        new NpgsqlParameter("@Age", 15))
    .ToListAsync();

实际上，您可以创建一个通用的存储库并做类似的事情

public class GenericRepository<TEntity> : IGenericRepository<TEntity> where TEntity : BaseEntity
{
    private readonly DataContext context;
    private readonly DbSet<TEntity> dbSet;

    public GenericRepository(DataContext context)
    {
        this.context = context;
        this.dbSet = context.Set<TEntity>();
    }

   
    public IEnumerable<TEntity> ExecuteCommandQuery(string command)
        => dbSet.FromSqlRaw(command);

}

我的案例使用了存储过程而不是原始SQL

创建一个类

Public class School
{
    [Key]
    public Guid SchoolId { get; set; }
    public string Name { get; set; }
    public string Branch { get; set; }
    public int NumberOfStudents  { get; set; }
}

下面在我的DbContext类上添加

public DbSet<School> SP_Schools { get; set; }

执行存储过程。

var MySchools = _db.SP_Schools.FromSqlRaw("GetSchools @schoolId, @page, @size ",
              new SqlParameter("schoolId", schoolId),
              new SqlParameter("page", page),
              new SqlParameter("size", size)))
.IgnoreQueryFilters();

我使用Dapper绕过了Entity框架Core的这个约束。

IDbConnection.Query

正在处理带有多个参数的SQL查询或存储过程。 顺便说一下，它更快一些(参见基准测试)

Dapper很容易学。编写和运行带参数的存储过程需要15分钟。无论如何，你可以同时使用EF和Dapper。下面是一个例子:

 public class PodborsByParametersService
{
    string _connectionString = null;


    public PodborsByParametersService(string connStr)
    {
        this._connectionString = connStr;

    }

    public IList<TyreSearchResult> GetTyres(TyresPodborView pb,bool isPartner,string partnerId ,int pointId)
    {

        string sqltext  "spGetTyresPartnerToClient";

        var p = new DynamicParameters();
        p.Add("@PartnerID", partnerId);
        p.Add("@PartnerPointID", pointId);

        using (IDbConnection db = new SqlConnection(_connectionString))
        {
            return db.Query<TyreSearchResult>(sqltext, p,null,true,null,CommandType.StoredProcedure).ToList();
        }


        }
}