This is a followup to the question here, where the answer seems to refer to an overly-complicated and overly-specific (EF, which I'm not using - not even using an ORM).
There has to be a more straightforward way around this common scenario than the smoke, mirrors, and sorcery hinted at in that answer.
Note: I encased "context" in parenthesis because I'm not using EF, so it is not a literal "dbcontext" that I'm talking about here.
So I got to wondering: Could I set a global variable for each session when the user is authenticated and authorized?
e.g., when the user is authenticated/authorized, I would know which database context/contents should be served up to him.
So it seems I could set a value in Global.asax.cs' Application_Start() method and then either alter the RepositoriesInstaller (implementing IWindsorInstaller) class to conditionally register different concrete Repositories based on the user and what data he should have OR place conditional code in the concrete Repository itself to use this or that database instance.
Is this feasible? Is one way (altering the RepositoriesInstaller / altering the concrete Repositories class) preferred?
Have a look at this answer which will show you how to resolve the correct repository based on a key or value.
If you want to store it with the authorized user, you need to simple serialize the data and store it on the authenticated cookie:
Create a Model to represent the logged in information:
public class AuthenticatedMember
{
public Guid Id { get; set; }
public string FirstName { get; set; }
public string LastName { get; set; }
public string Email { get; set; }
public int SiteNumber { get; set; }
}
Do something like a login in your controller:
var authenticatedMember = MembershipManager.ValidateLogin(model.Email, model.Password);
var cookie = FormsAuthentication.GetAuthCookie(authenticatedMember.Id.ToString(), false);
var ticket = FormsAuthentication.Decrypt(cookie.Value);
var newTicket = new FormsAuthenticationTicket(ticket.Version, ticket.Name, ticket.IssueDate, ticket.Expiration, ticket.IsPersistent, authenticatedMember.ToJson(), ticket.CookiePath);
cookie.Value = FormsAuthentication.Encrypt(newTicket);
Response.Cookies.Add(cookie);
Then use a model binder to deserialize the AuthenticatedMember when you require:
public class AuthenticatedMemberModelBinder : IModelBinder
{
#region IModelBinder Members
public object BindModel(ControllerContext controllerContext, ModelBindingContext bindingContext)
{
if (bindingContext.Model != null)
{
throw new InvalidOperationException("Cannot update instances");
}
if (controllerContext.RequestContext.HttpContext.Request.IsAuthenticated)
{
var cookie = controllerContext
.RequestContext
.HttpContext
.Request
.Cookies[FormsAuthentication.FormsCookieName];
if (null == cookie)
return null;
var decrypted = FormsAuthentication.Decrypt(cookie.Value);
if (!string.IsNullOrEmpty(decrypted.UserData))
{
return JsonConvert.DeserializeObject<AuthenticatedMember>(decrypted.UserData);
}
}
return null;
}
#endregion
}
Related
In our Asp.Net Core (2.2) MVC project we had to use an existing database (including all user & role related tables) from our previous Asp.Net Web app project.
Retrieving user data in asp.net web app (and having it available throughout the website) was preatty simple: upon login fill a custom user class/object with all the properties you need, save it as a Session variable and you call it wherever you need it (without going to the database).
This seems to me a lot harder to achieve in Asp.Net Core. What I have so far is:
ApplicationUser class:
public class ApplicationUser : IIdentity
{
public int Id { get; set; }
public Uporabnik Uporabnik { get; set; }
public string AuthenticationType { get; set; }
public bool IsAuthenticated { get; set; }
public string Name { get; set; }
}
Login form:
public IActionResult Prijava(PrijavaModel model)
{
// check user credentials
//
// ... validation code here ...
//
if (uporabnik != null)
{
//Create the identity for the user
var identity = new ClaimsIdentity(new[] {
new Claim("Email", model.Email),
new Claim("Id", uporabnik.IdWebUser.ToString()),
new Claim("Name", uporabnik.ImeInPriimek),
new Claim(ClaimTypes.Name, uporabnik.ImeInPriimek),
new Claim(ClaimTypes.PrimarySid, uporabnik.IdWebUser.ToString())
}, CookieAuthenticationDefaults.AuthenticationScheme);
var principal = new ClaimsPrincipal(identity);
var login = HttpContext.SignInAsync(CookieAuthenticationDefaults.AuthenticationScheme, principal);
return RedirectToAction("Index", "Home");
}
return View();
}
Now to retrieve the data in a controller I have to do something like this:
// id
int idUser = int.Parse(#User.Claims.FirstOrDefault(x => x.Type == "Id").Value);
// or email
string email = #User.Claims.FirstOrDefault(x => x.Type == "Email").Value;
Well, this all works fine, but it's definitely not practical. To access any other user data I can go to the database (by "ID") and retrieve it, but I don't think this is the right way to do it!?!
Can I expand the identity class in such a way that I can set the extra properties I need at login time and retrive in a fashion similar to something like this:
var property1 = #User.Property1;
var property2 = #User.Property2;
// or
var property1 = #User.MyExtraProperties.Property1;
var property2 = #User.MyExtraProperties.Property2;
Is it possible (and also keeping it simple)?
EDIT: since there are no answers/suggestions, can I do the same thing with a different approach?
Look like you only want to call your properties in a better way?
public class ApplicationUser : IdentityUser
{
public string CustomName { get; set; }
}
Assuming you have done adding your extra properties, you could create an extension method for your properties, so you can later call them like User.Identity.GetCustomName().
namespace Project.Extensions
{
public static class IdentityExtensions
{
public static string GetCustomName(this IIdentity identity)
{
var claim = ((ClaimsIdentity)identity).FindFirst("CustomName");
return (claim != null) ? claim.Value : string.Empty;
}
}
}
Note that I didn't include the part where you add the claims, because you already have it. In this case, you should have CustomName claim.
Also, #Dementic is right about the session. If a user is removed/disabled, he would still have access to. So, having a db call each time you need to fetch information is correct.
I'm trying to implement authorization in my ASP.NET Core 2.0 Web app.
This app has like 20 models, each with a controller implementing at least a CRUD. I found these two pages and I liked the idea of using a handler to authorize requisitions. I would like initially to implement authorization by user, i.e., a user has only permission to see/edit his own entities. All my database entities have an OwnerId field.
These examples I found seem to only work for one specific controller.
So, my question is: is it possible to create one authorization handler for all controllers?
Have you found a solution or workaround yet that works with the authorization handler or authorization attributes? I have the exact same setup as you do.
I was trying to create a generic attribute to serve all may Entity CRUD owner checks, but generic attributes are not allowed by design.
The only two (unsatisfying) solutions that I came up with are:
Within the controller action, get the ownerId from the User, forward it all the way to your CRUD and include there a check for the ownerId. However, the code must be duplicated for every action in every controller.
[HttpGet("{id}"]
public async Task<IActionResult> GetById(int id)
{
var stringGuid = User.Claims.FirstOrDefault(c => c.Type == "sub")?.Value;
if (String.IsNullOrWhiteSpace(stringGuid)) return Unauthorized();
var ownerGuid = new Guid(stringGuid);
var entity = _yourCrudInstance.GetById(id, ownerGuid);
return Ok(entity);
}
Add a method to your CRUD repository like bool IsOwner(Guid ownerId) and use this method when creating the custom authorization handler (by creating a custom requirement together with a custom handler). This eliminates code duplication in the controller, because you can create a new policy with this custom authorization handler and consequently you can simply decorate every action with a [Authorize(Policy = "yourOwnershipPolicy")]. But still, there must be a service created for each and every controller. Moreover, the IsOwner(...) method adds an additional database call compared to solution 1 - one db call for checking the ownership (during authorization check) and one db call for actually getting the entity (by working through the controller action).
[Authorize(Policy = "yourOwnershipPolicy")]
public async Task<IActionResult> GetById(int id)
{
var entity = _yourCrudInstance.GetById(id);
return Ok(entity);
}
I am going with the first solution until I found a way to create a generic authorization handling for my generic CRUD repository, because one may forget creating the required authorization policy for a new entity, but one cannot forget to supply the parameter ownerId to .GetById(id, ownerGuid), provided there is no overload method, or the code doesn't compile.
Update:
I found a third solution in which was able to create a kind of generic authorization attribute. The trick was to use the type of concrete repository as input parameter in the authorization attribute. Yet, there is still a limitation: The authorization attribute must be copied for every type of Id, for example int Id, Guid id, etc. But still, this reduces repeated code to the types of ids. In most cases, people only have one type of id, probably int or Guid.
Here some code that demonstrates my architecture. It is heavily summarized and redacted, but should compile successfully. My original code is working and in production:
using System;
using System.Linq;
using System.Threading.Tasks;
using Microsoft.AspNetCore.Authorization;
using Microsoft.AspNetCore.Mvc;
using Microsoft.AspNetCore.Mvc.Filters;
using Microsoft.EntityFrameworkCore;
using Microsoft.Extensions.DependencyInjection;
[Route("api/yourcontroller")]
public class YourApiController : Controller
{
private readonly YourEntityXYZRepository _repo;
public YourApiController(YourDbContext yourDbContext)
{
_repo = new YourEntityXYZRepository(yourDbContext);
}
[HttpGet("{id}")]
[AuthorizeOwnerIntId(typeof(YourEntityXYZRepository), Policy = "YourCustomPolicy")]
public async Task<IActionResult> GetById(int id)
{
var entity = _repo.GetById(id);
return Ok(entity);
}
}
// The "generic" authorization attribute for type int id
// Similar authorization attributes for every type of id must be created additionally, for example Guid
[AttributeUsage(AttributeTargets.Method, AllowMultiple = true, Inherited = true)]
public class AuthorizeOwnerIntIdAttribute : AuthorizeAttribute, IAuthorizationFilter
{
private object _entityRepositoryObject;
private IAsyncOwnerIntId _entityRepository;
private readonly Type _TCrudRepository;
public AuthorizeOwnerIntIdAttribute(Type TCrudRepository)
{
_TCrudRepository = TCrudRepository;
}
public void OnAuthorization(AuthorizationFilterContext context)
{
var yourDbContext = context.HttpContext.RequestServices.GetService<YourDbContext>();
_entityRepositoryObject = Activator.CreateInstance(_TCrudRepository, yourDbContext);
_entityRepository = _entityRepositoryObject as IAsyncOwnerIntId;
var user = context.HttpContext.User;
if (!user.Identity.IsAuthenticated)
{
// it isn't needed to set unauthorized result
// as the base class already requires the user to be authenticated
// this also makes redirect to a login page work properly
// context.Result = new UnauthorizedResult();
return;
}
// get entityId from uri
var idString = context.RouteData.Values["id"].ToString();
if (!int.TryParse(idString, out var entityId))
{
context.Result = new UnauthorizedResult();
return;
}
// get subjectId from user claims
var ownerIdString = context.HttpContext.User.Claims.FirstOrDefault(c => c.Type == "sub")?.Value;
if (!Guid.TryParse(ownerIdString, out var ownerGuid))
{
context.Result = new UnauthorizedResult();
return;
}
if (!_entityRepository.IsEntityOwner(entityId, ownerGuid))
{
context.Result = new UnauthorizedResult();
}
}
}
// Your concrete repository
public class YourEntityXYZRepository : AsyncCrud<YourEntityXYZ, int>,
IAsyncOwnerIntId // Note that type concrete IAsyncOwnerIntId is only implemented in concrete repository
{
public YourEntityXYZRepository(YourDbContext yourDbContext) : base(yourDbContext)
{
}
}
// Your generic Crud repository
public abstract class AsyncCrud<TEntity, TId> : IAsyncCrud<TEntity, TId>
where TEntity : class, IEntityUniqueIdentifier<TId>, IEntityOwner
where TId : struct
{
protected YourDbContext YourDbContext;
public AsyncCrud(YourDbContext yourDbContext)
{
YourDbContext = yourDbContext;
}
// Note that the following single concrete implementation satisfies both interface members
// bool IsEntityOwner(TId id, Guid ownerGuid); from IAsyncCrud<TEntity, TId> and
// bool IsEntityOwner(int id, Guid ownerGuid); from IAsyncOwnerIntId
public bool IsEntityOwner(TId id, Guid ownerGuid)
{
var entity = YourDbContext.Set<TEntity>().Find(id);
if (entity != null && entity.OwnerGuid == ownerGuid)
{
return true;
}
return false;
}
// Further implementations (redacted)
public Task<bool> SaveContext() { throw new NotImplementedException(); }
public Task<TEntity> Update(TEntity entity){ throw new NotImplementedException(); }
public Task<TEntity> Create(TEntity entity, Guid ownerGuid) { throw new NotImplementedException(); }
public Task<bool> Delete(TId id) { throw new NotImplementedException(); }
public Task<bool> DoesEntityExist(TId id) { throw new NotImplementedException(); }
public virtual Task<TEntity> GetById(TId id) { throw new NotImplementedException(); }
}
// The interface for the Crud operations
public interface IAsyncCrud<TEntity, TId>
where TEntity : class, IEntityUniqueIdentifier<TId>
where TId : struct
{
bool IsEntityOwner(TId id, Guid ownerGuid);
Task<bool> DoesEntityExist(TId id);
Task<TEntity> GetById(TId id);
Task<TEntity> Create(TEntity entity, Guid ownerGuid);
Task<TEntity> Update(TEntity entity);
Task<bool> Delete(TId id);
Task<bool> SaveContext();
}
// The interface for the concrete type method for int id
// Similar interfaces for every type of id must be created additionally, for example Guid
public interface IAsyncOwnerIntId
{
bool IsEntityOwner(int id, Guid ownerGuid);
}
// Typical db context
public class YourDbContext : DbContext
{
public YourDbContext(DbContextOptions<YourDbContext> options) : base(options)
{
}
public DbSet<YourEntityXYZ> YourEntityXYZ { get; set; }
}
public class YourEntityXYZ : IEntityUniqueIdentifier<int>, IEntityOwner
{
public int Id { get; set; }
public Guid? OwnerGuid { get; set; }
// ... Additonal custom properties
}
public interface IEntityUniqueIdentifier<TId>
where TId : struct
{
TId Id { get; set; }
}
public interface IEntityOwner
{
Guid? OwnerGuid { get; set; }
}
I have following tables:
Users, Groups, GroupUsers
I have my class like:
public class User
{
List<int> GroupIds;
}
I have a method in my Data Access Layer which returns a user after successful login with the list of all the group ids. How can I override or do something similar in identity to get information from that method?
I have searched the Internet a lot, watched YouTube tutorials, but no one is actually explaining it the way I need it. Can any one help?
First, use claims based security instead of role based: http://visualstudiomagazine.com/articles/2013/08/01/leveraging-claims-based-security-in-aspnet-45.aspx
I've also been retro-fitting an existing SQL based login system to work with identity management. Most of the work you're going to have to do lies within IdentityModel.cs. Specifically how you define ApplicationUser
public class ApplicationUser : IdentityUser<string, ApplicationUserLogin, ApplicationUserRole, ApplicationUserClaim>
{
public async Task<ClaimsIdentity> GenerateUserIdentityAsync(ApplicationUserManager manager)
{
// Note the authenticationType must match the one defined in CookieAuthenticationOptions.AuthenticationType
var userIdentity = await manager.CreateIdentityAsync(this, DefaultAuthenticationTypes.ApplicationCookie);
// Add custom user claims here
return userIdentity;
}
}
I've overridden all 4 type parameters on the IdentityUser base class for my implementation, you may not need to do so, depends on how different your retrofit is from how Identity expects things to be.
The other main place you'll most likely have to customize things is within ApplicationDbContext where you'll have to set up where your user, group, and claims/role definitions reside with SQL.
I found these articles on typecastexception to be very useful:
http://typecastexception.com/post/2014/04/20/ASPNET-MVC-and-Identity-20-Understanding-the-Basics.aspx
http://typecastexception.com/post/2014/06/22/ASPNET-Identity-20-Customizing-Users-and-Roles.aspx
http://typecastexception.com/post/2014/04/20/ASPNET-Identity-20-Setting-Up-Account-Validation-and-Two-Factor-Authorization.aspx
http://typecastexception.com/post/2014/07/13/ASPNET-Identity-20-Extending-Identity-Models-and-Using-Integer-Keys-Instead-of-Strings.aspx
Overall there going to be a lot of trial and error in the process as you figure out what pieces of Identity you can utilize as is and where you need to plug in your own code.
Something else you may run into if your passwords aren't stored is having to provide your own implementation of PasswordHasher and plugging that in:
Asp.net Identity password hashing
I did' t get your question, if you want to override you have to mark the method virtual and inherit the class like this:
public class User
{
public virtual void YourMethod()
{
}
}
public class YourClass : User
{
public override void YourMethod()
{
}
}
If you want to separate the class to add some more mothods you can go like this:
partial class User
{
public static void YourMethod()
{
}
}
Create a UserInfo object
public class ApplicationUser : IdentityUser
{
public virtual UserInfo UserInfo { get; set; }
}
public class UserInfo : ComparableEntity
{
public string Email { get; set; }
public string FullName { get; set; }
public string KidName { get; set; }
public string MobilePhone { get; set; }
}
Then create a database context
public class DatabaseContext : IdentityDbContext<ApplicationUser>, IDatabaseContext
{
public IDbSet<UserInfo> UserInfos { get; set; }
}
I would like to pass information from the action filter (database) to the Action function.
Is it secure to use ActionContext Request.Properties.Add to store the data?
is there any chance that the information will be seen by the WEBAPI client or its safe as it safe to store information in the Cache\Session?
Is it a better way to do it?
The client will not see request properties unless you explicitly serialize them. They completely remain on the server side.
To answer your followup question here are two other ways to do it. There is no "Best" way per se. It all depends on how far you want the information to flow, and how generic you want your filter to be. My personal preference is using the controller object, but again it is just a preference.
For the sample here is a simple values controller and a POCO class:
[MyActionfilter]
public class ValuesController : ApiController
{
public string Foo { get; set; }
public User Get(User user)
{
if (Foo != null && user != null)
{
user.FamilyName = Foo;
}
return user;
}
}
public class User
{
public string FirstName { get; set; }
public string FamilyName { get; set; }
}
The action filter below is naively implementing access to the controller object or the method parameters. Note that it's up to you to either apply the filter sparingly or do type checks/dictionary checks.
public class MyActionfilter : ActionFilterAttribute
{
public override void OnActionExecuting(System.Web.Http.Controllers.HttpActionContext actionContext)
{
controller = actionContext.ControllerContext.Controller;
// Not safe unless applied only to controllers deriving
// from ValuesController
((ValuesController)controller).Foo = "From filter";
// Not safe unless you know the user is on the signature
// of the action method.
actionContext.ActionArguments["user"] = new User()
{
FirstName = "From filter"
};
}
}
This is probably quite straight forward for some, however I'm a bit confused and can't find a decent example. Say I'm using view models and my POST action takes in that view model. Typically I would do something along the following lines:
[HttpPost]
public ActionResult Update(UserViewModel uvm)
{
User user = Mapper.Map<UserViewModel, User>(uvm);
_repository.Update(user);
return RedirectToAction("Index");
}
Although this isn't the full picture. The mapping would work fine, however if I were to just update what I've mapped then it'd get rid of valuable data in the database because of course in this case I'm not updating the password or other details.
My repository looks something like this:
public void Update(User user)
{
User u = Session.QueryOver<User>().Where(x => x.UserName == user.UserName).SingleOrDefault();
if (u == null)
throw new Exception("User not found");
u.Forename = user.Forename;
u.Surname = user.Surname;
u.EmailAddress = user.EmailAddress;
}
[I'm using NHibernate so it'll save the object back to the DB once the session is closed (after the request has finished) automatically for me.]
So my question is, in my repository should I load the "User" entity, then update the values I want, and then save it back, or is there another method to do this? The reason I ask is because it seems a bit... "manual" if you see what I mean? Perhaps it is correct, but I just wanted to see opinions of those with more experience in this area.
Cheers
I use the following approach:
[HttpPost]
public ActionResult Update(UserViewModel uvm)
{
User user = _userRepository.FindById(uvm.Id);
user.Forename = uvm.Forename;
user.Surname = uvm.Surname;
user.EmailAddress = uvm.EmailAddress;
_userRepository.Update(user);
return RedirectToAction("Index");
}
UPDATE:
To address the comments about AutoMapper here's how to proceed:
Let's take for example the following classes:
public class UserViewModel
{
public string Forename { get; set; }
public string Surname { get; set; }
public string EmailAddress { get; set; }
}
public class User
{
public string Forename { get; set; }
public string Surname { get; set; }
public string EmailAddress { get; set; }
public string Password { get; set; }
}
We don't want to modify the user password in the UI. So we express our intention to AutoMapper:
Mapper
.CreateMap<UserViewModel, User>()
.ForMember(dest => dest.Password, opt => opt.Ignore());
and then:
[HttpPost]
public ActionResult Update(UserViewModel uvm)
{
// Fetch the original model we would like to update
User user = _userRepository.FindById(uvm.Id);
Mapper.Map(uvm, user);
// At this stage the user model will have its
// Forename, Surname and EmailAddress properties
// updated from the view model and its Password property
// will remain the one we got from the repository
_userRepository.Update(user);
return RedirectToAction("Index");
}
UPDATE 2:
To address the question in the comments about configuring AutoMapper I usually use Profiles:
public class UsersProfile : Profile
{
protected override void Configure()
{
Mapper
.CreateMap<UserViewModel, User>()
.ForMember(dest => dest.Password, opt => opt.Ignore());
Mapper
.CreateMap<User, UserViewModel>();
}
}
and then have a registry class which registers all the mappers:
public class MappingsRegistry
{
public static void Configure()
{
Mapper.AddProfile(new UsersProfile());
Mapper.AddProfile(new SomeOtherProfile());
...
}
}
which is called in Application_Start:
MappingsRegistry.Configure();
Finally my controllers have a reference to the mapping engine:
public class UsersController : Controller
{
private readonly IUsersRepository _repository;
private readonly IMappingEngine _mappingEngine;
public ContratsFCController(IUsersRepository repository, IMappingEngine mapperEngine)
{
_repository = repository;
_mapperEngine = mapperEngine;
}
[AutoMap(typeof(User), typeof(UserViewModel))]
public ActionResult Update(int id)
{
var user = _repository.FindById(id);
return View(user);
}
[HttpPost]
public ActionResult Update(UserViewModel uvm)
{
if (!ModelState.IsValid)
{
return View(uvm);
}
var user = _repository.FindById(uvm.Id);
_mapperEngine.Map(uvm, user);
_repository.Update(user);
return RedirectToAction("Index");
}
}
Now all that's left is to instruct your DI framework to pass the Mapper.Engine property to the constructor and in your unit tests obviously substitute them with an appropriate mock.