I need to send an email to users from an ASP.NET Core 2 application, following some business rules. However, I need to ensure that the account the email is being sent to actually exists (for some reason, it may be that the account stopped being valid). The customer is using Azure Active Directory, so I need to query AAD somehow so it lets me know whether the account exists or not.
So far I have been looking for Microsoft Graph as a way to do this, however every example I have seen so far requires prior authentication and use a delegate authentication mechanism. I don't want my users having to authenticate nor to prompt the authentication screen.
Given this situation, what would you recommend using? If you can also point me to an example, that would be great. Thanks!
You don't really need to throw/catch exception for every invalid user as you're doing in current code. I have nothing against exception handling in general for other reasons but to see if the user exists or not you can try using Filter.
So your graph query could look like -
https://graph.microsoft.com/v1.0/users?$filter=startswith(userPrincipalName,'someuser#mytenant.onmicrosoft.com')
I have shown startswith here becuase eq didn't work for me in a quick trial. Although I would recommend two things:
Go through Microsoft documentation on Filters here and see what works best for your requirements - Use query parameters to customize responses with Microsoft Graph
Play a little bit with different queries in Microsoft Graph Explorer it's very simple and easy to use.
Here is a modified version for your code.
Note that I'm checking for the collection count to be > 0 and not checking for it to be null, as even in case user is not found the UsersCollectionPage was not null for my test run.
using Microsoft.Identity.Client;
using Microsoft.Graph.Auth;
using Microsoft.Graph;
...
private async Task<bool> ValidateAccounts(string accounts) {
var confidentialClientApplication = ConfidentialClientApplicationBuilder
.Create("clientId here")
.WithTenantId("tokenId here")
.WithClientSecret("secret here")
.Build();
var authProvider = new ClientCredentialProvider(confidentialClientApplication);
var graphClient = new GraphServiceClient(authProvider);
var valid = true;
try {
foreach (var account in accounts.Split(';')) {
var user = await graphClient.Users.Request().Filter("startswith(userPrincipalName, '" + account + "')").GetAsync();
if (user.Count <= 0) {
valid = false;
break;
}
}
} catch (ServiceException ex) {
valid = false;
}
return valid;
}
On a side note, I'm not not sure of your requirements but you could probably get creative by combining multiple user names in single query and then checking for result counts or other propertes. You could use or between multiple criteria or probably use any operator. I haven't really tried this out though.
Finally I came up with something workable. It's not nice, and it uses preview software. First, install Microsoft.Graph and Microsoft.Identity.Client packages. Then install Microsoft.Graph.Auth, which at the time of this writing, is in preview (v1.0.0-preview.1) so you'll need to tick "include prerelease" checkbox in nuget manager.
Then in your AAD, you need to get the ClientId, TenantId and SecretId. In my case, my app was already using AAD authentication so I already had ClientId and TenantId in my appsettings.json file. I only needed to create a new SecretId (in the Certificate & secrets section of my app registration). Then I needed to add permissions (in the API permissions section of my app registration) to include Microsoft.Graph with at least User.Read.All permission.
using Microsoft.Identity.Client;
using Microsoft.Graph.Auth;
using Microsoft.Graph;
...
private async Task<bool> ValidateAccounts(string accounts) {
var confidentialClientApplication = ConfidentialClientApplicationBuilder
.Create("clientId here")
.WithTenantId("tokenId here")
.WithClientSecret("secret here")
.Build();
var authProvider = new ClientCredentialProvider(confidentialClientApplication);
var graphClient = new GraphServiceClient(authProvider);
var valid = true;
try {
foreach (var account in accounts.Split(';')) {
var user = await graphClient.Users[account]
.Request()
.GetAsync();
if (user == null) {
valid = false;
break;
}
}
} catch (ServiceException ex) {
valid = false;
}
return valid;
}
Here, the function takes a semicolon-separated string for each account. The GetAsync method will throw a ServiceException if the user does not exist. I don't like that, but couldn't find another way. So that's about it. Hope this helps someone else, and hope someone could come up with a better solution eventually.
Import following namespaces (You needs to install relevant packages using nuget):
using Microsoft.Graph;
using Microsoft.IdentityModel.Clients.ActiveDirectory;
Set your Azure AD app values:
private string _tenant => "your_tenant_id";
private string _appId => "your_ad_app_client_id";
private string _appSecret => "your_app_client_secret";
Create Graph Service Client using this:
public static GraphServiceClient CreateGraphServiceClient()
{
var clientCredential = new ClientCredential(_appId, _appSecret);
var authenticationContext = new AuthenticationContext("https://login.microsoftonline.com/{_tenant}");
var authenticationResult = authenticationContext.AcquireTokenAsync("https://graph.microsoft.com", clientCredential).Result;
var delegateAuthProvider = new DelegateAuthenticationProvider((requestMessage) =>
{
requestMessage.Headers.Authorization = new AuthenticationHeaderValue("bearer", authenticationResult.AccessToken);
return Task.FromResult(0);
});
return new GraphServiceClient(delegateAuthProvider);
}
var graphServiceClient = GraphServiceClientHelper.CreateGraphServiceClient();
Then call graph api and filter users by Email Address as follows:
var user = await graphServiceClient.Users.Request().Filter("mail eq '" + UserEmailAddress + "'").GetAsync();
if (user.Count == 0) {
//user not exist
}
Below code worked for me.
using Microsoft.Graph;
using Microsoft.Graph.Auth;
using Microsoft.Identity.Client;
private static async Task<bool> ValidateAccounts(string accounts)
{
var confidentialClientApplication = ConfidentialClientApplicationBuilder
.Create("client id")
.WithTenantId("tenant id")
.WithClientSecret("client secret")
.Build();
var authProvider = new ClientCredentialProvider(confidentialClientApplication);
var graphClient = new GraphServiceClient(authProvider);
var valid = true;
try
{
foreach (var account in accounts.Split(';'))
{
var user = await
graphClient.Users.Request().Filter($"identities/any(c:c/issuerAssignedId eq
'{account}' and c/issuer eq 'xyz.onmicrosoft.com')").GetAsync();
if (user.Count <= 0)
{
valid = false;
break;
}
}
}
catch(Exception ex)
{
valid = false;
}
return valid;
}
Related
I'm running .NET Core v3.1 and Blazor and have implemented authorization using Google limited to our domain in Google G Suite as mentioned here: https://www.jerriepelser.com/blog/forcing-users-sign-in-gsuite-domain-account/
Login/logout is working fine, but when the user who logged in is blocked or removed in Google G Suite the user stays logged in into my application until he logs out from the application. When he doesn't logs out he can keep using the application.
I'm looking for a refresh every hour.
This is my login.cshtml.cs:
public async Task<IActionResult> OnGetCallbackAsync(string returnUrl = null, string remoteError = null)
{
// Get the information about the user from the external login provider
var GoogleUser = User.Identities.FirstOrDefault();
if (GoogleUser.IsAuthenticated)
{
var authProperties = new AuthenticationProperties
{
IsPersistent = true,
RedirectUri = Request.Host.Value,
IssuedUtc = System.DateTime.UtcNow,
ExpiresUtc = System.DateTime.UtcNow.AddHours(1)
};
await HttpContext.SignInAsync(CookieAuthenticationDefaults.AuthenticationScheme, new ClaimsPrincipal(GoogleUser), authProperties);
}
return LocalRedirect("/");
}
I already added IssuedUtc and ExpiresUtc but that didn't change anything.
You have to enable the ability to call Google APIs (https://www.googleapis.com/auth/admin.directory.user, https://www.googleapis.com/auth/admin.directory.group) to get this information, but, before you can do that, the G-Suite Domain Admin has to authorize that access using https://developers.google.com/admin-sdk/directory/v1/guides/authorizing
This explains the process:
https://developers.google.com/admin-sdk/directory/v1/guides/delegation
You will want to see this GitHub repo for code samples:
https://github.com/googleapis/google-api-dotnet-client
Here is some psudo code:
string[] Scopes = {
DirectoryService.Scope.AdminDirectoryGroup,
DirectoryService.Scope.AdminDirectoryUser
};
GoogleCredential credential;
//redirectUrl = this.Request.Host.Value;
string keyfilepath = "yourKeyFile.json";
using (var stream = new FileStream(keyfilepath, FileMode.Open, FileAccess.Read))
{
// As we are using admin SDK, we need to still impersonate user who has admin access
// https://developers.google.com/admin-sdk/directory/v1/guides/delegation
credential = GoogleCredential.FromStream(stream)
.CreateScoped(Scopes).CreateWithUser(EmailOfGoogleDomainAdmin);
}
// Create Directory API service.
var service = new DirectoryService(new BaseClientService.Initializer()
{
HttpClientInitializer = credential,
ApplicationName = "ApplicationName",
});
// G Suite User to get information about
// This test user should be suspended
var gs_email = UserToCHeck;
var request = service.Users.Get(gs_email);
var result = request.Execute();
Console.WriteLine("Full Name: {0}", result.Name.FullName);
Console.WriteLine("Email: {0}", result.PrimaryEmail);
Console.WriteLine("ID: {0}", result.Id);
Console.WriteLine("Is Admin: {0}", result.IsAdmin);
Console.WriteLine("Is Suspended: {0}", result.Suspended);
I am trying to build a service client to simplify calling my microservices in .net core.
Here is a service client sample:
public ProductServiceClient(SystemEnvironment.MachineEnvironment? environment = null)
{
this.url = ServiceEnvironment.Urls.GetUrl(ServiceEnvironment.Service.Product, environment);
}
private RestClient GetClient(string method)
{
return new RestClient(url + "/api/" + method);
}
private RestRequest GetRestRequest(Method method)
{
var restRequest = new RestRequest(method);
restRequest.RequestFormat = DataFormat.Json;
restRequest.AddHeader("Content-Type", "application/json");
return restRequest;
}
public FindProductsResponse FindProducts(FindProductsRequest request)
{
var restRequest = GetRestRequest(Method.GET);
restRequest.AddJsonBody(request);
var client = this.GetClient("Products");
var restResponse = client.Get(restRequest);
return new JsonDeserializer().Deserialize<FindProductsResponse>(restResponse);
}
public void Dispose()
{
}
And here is how I am trying to read it in my .net core api:
[HttpGet]
public ActionResult<FindProductsResponse> Get()
{
var request = "";
using (StreamReader reader = new StreamReader(Request.Body, Encoding.UTF8))
{
request = reader.ReadToEnd();
}
var buildRequest = JsonConvert.DeserializeObject<FindProductsRequest>(request);
var products = _service.FindProducts(buildRequest);
if (products != null && products.Any())
{
return new FindProductsResponse()
{
Products = products
};
}
return BadRequest("Not found");
}
However the request variable is always empty after Request.Body has been processed by the StreamReader.
If I make the same request from Postman (also using GET), I get the body just fine.
What am I doing wrong here?
EDIT: This is the unit test calling the api:
[Test]
public void Test1()
{
using (var productServiceClient = new ProductServiceClient())
{
var products = productServiceClient.FindProducts(new FindProductsRequest()
{
Id = 50
}).Products;
}
}
It can be your Request.Body has been already consumed.
Try to call Request.EnableRewind() before to open the StreamReader.
I'm not sure why you are manually doing it. It looks like you are reinventing the wheel. ASP.NET Core already does that for you.
This is what your service should look like:
[HttpGet] // oops, GET requests will not allow Bodies, this won't work
public ActionResult<FindProductsResponse> Get([FromBody]FindProductsRequest buildRequest)
{
// skip all the serialization stuff, the framework does that for you
var products = _service.FindProducts(buildRequest);
if (products != null && products.Any())
{
return new FindProductsResponse()
{
Products = products
};
}
return BadRequest("Not found");
}
And if you don't want to redo all the busy work that is retyping all the code on the client side, I suggest you read up on swagger (probably in the form of Swashbuckle). Client code can be generated. Even from within Visual Studio, if you right-click on the project and in the context menu pick "Add REST API Client...". Please don't erroneously hand-code what can be generated flawlessly by a machine instead. I don't really know what went wrong in your specific case, but searching bugs that could be avoided altogether is just busywork, that time should be spent on other parts of the program.
I just realized this is a GET request. ASP.NET will not recognize bodies for GET-Requests. You will need to make it a PUT or POST request or put your parameters in the query string.
If you happen to make that mistake as often as I did, you might want to write some unit tests that cover this. Because .NET is not helping you there. Been there, done that..
I have an ASP.NET 5 beta 8 application which integrates with Azure Active Directory using OpenIdConnect. I tried updating the application to RC1 and changed the openid nuget package to "Microsoft.AspNet.Authentication.OpenIdConnect": "1.0.0-rc1-final". Everything seems to be correct, but the OnAuthenticationComplete method which I used to add roles to my ClaimsIdentity is no longer in the Events object, and I can't find the alternative. How can I add claims to my Identity with the new version?
UPDATE: Changing to OnAuthenticationValidated still not works for me. I must be doing something wrong in my event's code:
OnAuthenticationValidated = async notification =>
{
var claimsIdentity = notification.AuthenticationTicket.Principal.Identity as ClaimsIdentity;
var userRepository=new UserRepository(Configuration);
var userId = claimsIdentity?.FindFirst(ClaimTypes.NameIdentifier)?.Value;
var user = await userRepository.FindAsync(userId);
if (user == null)
{
await userRepository.AddAsync(new UserDto
{
UserId = userId,
Username = claimsIdentity?.FindFirst(ClaimTypes.Name)?.Value,
DisplayName = claimsIdentity?.FindFirst(ClaimTypes.Name)?.Value
});
}
claimsIdentity?.AddClaim(new Claim(ClaimTypes.Role, "super"));
}
Also, the code I use to do the login is:
[HttpGet]
public IActionResult Login()
{
if (HttpContext.User == null || !HttpContext.User.Identity.IsAuthenticated)
return new ChallengeResult(OpenIdConnectDefaults.AuthenticationScheme, new AuthenticationProperties { RedirectUri = "/" });
return RedirectToAction("Index", "Home");
}
The AuthenticationCompleted event has been renamed to AuthenticationValidated.
You can find more information on this ticket: https://github.com/aspnet/Security/pull/442.
NOTE: Using the Sign-in button is NOT an option
A year ago I was having a problem creating a moment. Back then I was using version 1.2 of the Google+ API .Net client. As I described in this post, I had it working although the code failed to insert a moment from time to time. I was hoping that the process is more stable and easier to implement now, and it seems like it as can be seen in the example that you can download here - the current version as of this writing is v1.8. So I created a simple project following the SimpleOAuth2 sample in the download, but implementing Google+. This is the code I came up:
public partial class _Default : System.Web.UI.Page
{
private PlusService service;
// Application logic should manage users authentication.
// This sample works with only one user. You can change
// it by retrieving data from the session.
private const string UserId = "user-id";
protected void Page_Load(object sender, EventArgs e)
{
GoogleAuthorizationCodeFlow flow;
var assembly = Assembly.GetExecutingAssembly();
using (var stream = assembly.GetManifestResourceStream(
"GPlusSample.client_secrets.json"))
{
flow = new GoogleAuthorizationCodeFlow(
new GoogleAuthorizationCodeFlow.Initializer
{
DataStore = new FileDataStore("GPlusSample.Store"),
ClientSecretsStream = stream,
//
// Tried only this scope but it did not work
//Scopes = new[] { PlusService.Scope.PlusMe }
//
// I tried the following: but did not work either
//Scopes = new[] { PlusService.Scope.PlusMe,
// "https://www.googleapis.com/auth/plus.moments.write" }
//
// I tried this as well and it failed
//Scopes = new[] { PlusService.Scope.PlusLogin }
//
// Maybe this... but still no joy
Scopes = new[] { PlusService.Scope.PlusLogin,
PlusService.Scope.PlusMe }
});
}
var uri = Request.Url.ToString();
var code = Request["code"];
if (code != null)
{
var token = flow.ExchangeCodeForTokenAsync(UserId, code,
uri.Substring(0, uri.IndexOf("?")), CancellationToken.None).Result;
// Extract the right state.
var oauthState = AuthWebUtility.ExtracRedirectFromState(
flow.DataStore, UserId, Request["state"]).Result;
Response.Redirect(oauthState);
}
else
{
var result = new AuthorizationCodeWebApp(flow, uri, uri)
.AuthorizeAsync(UserId, CancellationToken.None).Result;
if (result.RedirectUri != null)
{
// Redirect the user to the authorization server.
Response.Redirect(result.RedirectUri);
}
else
{
// The data store contains the user credential,
// so the user has been already authenticated.
service = new PlusService(new BaseClientService.Initializer
{
ApplicationName = "Plus API Sample",
HttpClientInitializer = result.Credential
});
}
}
}
/// <summary>Gets the TasksLists of the user.</summary>
public async System.Threading.Tasks.Task InsertMoment()
{
try
{
var me = service.People.Get("me").Execute();
var request = service.Moments.Insert(new Moment()
{
Target = new ItemScope {
Id=Guid.NewGuid().ToString(),
Image="http://www.google.com/s2/static/images/GoogleyEyes.png",
Type="",
Name = "test message",
Description="test",
Text="test message",
},
Type = "http://schemas.google.com/AddActivity",
}, me.Id, MomentsResource.InsertRequest.CollectionEnum.Vault);
var response =await request.ExecuteAsync();
output.Text = "<h1>" + response.Id + "</h1>";
}
catch (Exception ex)
{
var str = ex.ToString();
str = str.Replace(Environment.NewLine, Environment.NewLine + "<br/>");
str = str.Replace(" ", " ");
output.Text = string.Format("<font color=\"red\">{0}</font>", str);
}
}
protected async void createMomentButton_Click(object sender, EventArgs e)
{
await InsertMoment();
}
}
That code always give me a 401 Unauthorized error, even if I have the Google+ API turned on for my project. Here's the actual error I got:
The service plus has thrown an exception: Google.GoogleApiException:
Google.Apis.Requests.RequestError Unauthorized [401] Errors [
Message[Unauthorized] Location[ - ] Reason[unauthorized]
Domain[global] ]
It's interesting to see that the insert moment is failing even though the call to People.Get("me") works - get("me") works with all of the scope combinations I listed above. It's important to note that each time I try a new scope, I first log out of my Google account and delete the access token that is stored in GPlusSample.Store.
EDIT
I tried setting just the Url instead of individual items as suggested by Ian and I got the exact same error.
var request = service.Moments.Insert(new Moment()
{
Target = new ItemScope {
Url = "https://developers.google.com/+/web/snippet/examples/thing"
},
Type = "http://schemas.google.com/AddActivity",
}, me.Id, MomentsResource.InsertRequest.CollectionEnum.Vault);
var response =await request.ExecuteAsync();
https://www.googleapis.com/auth/plus.login is the right scope for writing moments, but you need to have requested the specific app activity types you want to write as well. The parameter for this is request_visible_actions, and it takes a space separated list of arguments of the types (Listed on https://developers.google.com/+/api/moment-types/ - e.g. http://schemas.google.com/AddActivity).
The client library may not have a method for adding request_visible_actions, so you may have to add it on to the auth URL you redirect the user to manually (remember to URLencode the app activity type URLs!)
I've spent several days attempting to get to grips with ServiceStack and it seems great. Only issue is with authentication which seems to be a lot of friction, hard work and tears.
I want MonoTouch to register users, authenticate against ServiceStack/authenticate against OAuth and generally minimise calls to the database when authenticating.
So far I have got this:
var client = new JsonServiceClient(newbaseUri);
// register a new user:
var registration = new Registration {
FirstName = "john"
UserName = "user" ,
Password = "pass",
Email = "john#john.com",
};
var registerResponse = client.Send<RegistrationResponse>(registration);
--------------------------------
// user registered...later on I authenticate:
var authResponse = client.Send<AuthResponse>(new Auth {
UserName = "user",
Password = "pass",
RememberMe = true
});
var authResponse = clientlogin.Send<AuthResponse>(auth);
--------------------------------
// somehow I need to store 'authresponse' for later calls, not sure how without a browser
// tried manually setting the cookies and credentials parameters later but no joy
// but now I need to call a secured ([Authenticate] decorated) service method:
var client = new JsonServiceClient(newbaseUri);
var response = client.Send<HelloResponse>(new Hello { Name = "World!" });
return response.Result;
-----------------------------------------
// heres the configuration
var appSettings = new AppSettings();
//Default route: /auth/{provider}
Plugins.Add(new AuthFeature(() => new CustomUserSession(),
new IAuthProvider[] {
new CredentialsAuthProvider(appSettings), // never seems to get called
//new FacebookAuthProvider(appSettings), // not sure how to get this to work on monotouch
//new TwitterAuthProvider(appSettings), // same issue as facebook
new BasicAuthProvider(appSettings) // works but what about caching/tokens/cookies?
}));
//Default route: /register
Plugins.Add(new RegistrationFeature()); // how do i send extra params to this as created in mongodb collection
var mongoClient = new MongoClient("mongodb://localhost");
var server = mongoClient.GetServer();
var db = server.GetDatabase("users");
container.Register<ICacheClient>(new MemoryCacheClient());
container.Register<IUserAuthRepository>(new MongoDBAuthRepository(db, true));
My question is:
1) How do I enable extra fields to be passed in along with registration (as the mongodb [Servicestack.Authentication.Mongodb] has lots of empty fields i.e. birthdate, firstline, city, timezone, etc) that are not present in ServiceStack.Common.ServiceClient.Web.Registration object?
2) How can I transfer the cookie (or even maybe a token system) sent in the 'authresponse' to subsequent calls in order to allow ServiceStack to match against the session for ongoing authentication rather than more ongoing database calls that what seems to be issue with 'basic authentication' method (i.e CredentialsAuthProvider doesnt get called on server side)?
Please help...I've read documentation, run tests, examined social bootstrap and now I'm seriously losing days over this and thinking of integrating SS with simplemembership or even throwing ServiceStack away completely for old skool soap/wcf which is far easier to implement by the looks of it :(
1) If you want to use the Registration Plugin I don't think you can add additional fields since the Registration request/class is already defined. You could make your own registration Service and call into the RegistrationService/Plugin. Also, this post might be helpful.
[Route("/myregistration")]
public class MyRegistration : Registration //Add Additional fields for registration
{
public DateTime? BirthDate { get; set; }
public string Gender { get; set; }
}
public class MyRegisterService : Service
{
public IUserAuthRepository UserAuthRepo { get; set; }
public object Post(MyRegistration request)
{
using (var registrationService = base.ResolveService<RegistrationService>())
{
//handle the registration
var response = registrationService.Post(request.TranslateTo<Registration>());
}
//save the additional data
var userAuth = request.TranslateTo<UserAuth>();
UserAuthRepo.SaveUserAuth(userAuth);
//can make your own response or grab response from RegistrationService above
return new MyRegistrationResponse();
}
}
2) You can authenticate your JsonServiceClient and reuse it to make multiple requests.
var client = new JsonServiceClient(newbaseUri);
var authResponse = client.Send<AuthResponse>(new Auth {
UserName = "user",
Password = "pass",
RememberMe = true
}); //if successful your 'client' will have a populated CookieContainer with 'ss-id' and 'ss-pid' values
//reusing 'client' (after successful authentication) to make a request
//to a service requiring authentication
var response = client.Send<HelloResponse>(new Hello { Name = "World!" });
If reusing your 'client' is not an option you can try to store the ss-id. I don't know much about MonoTouch and how it stores 'browser sessions' so I'm not sure how you would accomplish this. After you authenticate and store the ss-id you can add it to the client using a Request Filter
//Get ss-id value
foreach(Cookie cookie in previousAuthenticatedClient.GetCookies(new Uri(newbaseUri)))
{
if (cookie.Name == "ss-id")
{
//store ss-id
}
}
var newClient = new JsonServiceClient(newbaseUri)
{
LocalHttpWebRequestFilter = (req) =>
{
req.CookieContainer.Add(new Uri("http://localhost:56006"), new System.Net.Cookie("ss-id", ssId));
}
};