I am new to azure, tokens and so on...
I have "digged" microsoft documentation and google and stackoverflow, but still didn't get full understanding.
So I using openId with Owin library to connect to azure from web app(VS2013 .net 4.5.1). And I have next code to do it:
public void Configuration(IAppBuilder app)
{
app.SetDefaultSignInAsAuthenticationType(
CookieAuthenticationDefaults.AuthenticationType);
app.UseCookieAuthentication(new CookieAuthenticationOptions());
app.UseOpenIdConnectAuthentication(
new OpenIdConnectAuthenticationOptions
{
MetadataAddress = String.Format(aadInstance, tenant, policy),
AuthenticationType = policy,
ClientId = clientId,
RedirectUri = redirectUri,
PostLogoutRedirectUri = redirectUri,
Notifications = new OpenIdConnectAuthenticationNotifications
{
AuthenticationFailed = AuthenticationFailed
,SecurityTokenValidated = OnSecurityTokenValidated
,AuthorizationCodeReceived = OnAuthorizationCodeReceived
,SecurityTokenReceived = OnSecurityTokenReceived
},
Scope = "openid profile",
ResponseType = "id_token"
};
);
}
private Task OnSecurityTokenValidated(SecurityTokenValidatedNotification<OpenIdConnectMessage, OpenIdConnectAuthenticationOptions> notification)
{
var identity = notification.AuthenticationTicket.Identity;
var claims = notification.OwinContext.Authentication.User.Claims;
ClaimsPrincipal.Current.AddIdentity(identity);
return Task.FromResult(0);
}
And it is working, but in microsoft documentation I found next instruction "Currently, ID tokens are signed but not encrypted. When your app receives an ID token, it must validate the signature to prove the token's authenticity and validate a few claims in the token to prove its validity. The claims validated by an app vary depending on scenario requirements, but your app must perform some common claim validations in every scenario."
But there is SecurityTokenValidated-callback , which have AuthenticationTicket. So do I still need to somehow validate token/ticked or now it is handled automatically (I been tough in army that nothing happening automatically, but still)?
The library that you are using handles the validation for you.
It will check the signature is what it should be based on the keys provided by Azure AD.
So you don't need to do manual checks, other than your app's specific checks. For example, an app might allow only members of a certain group to access the app. You would need to do that check if that is the case.
Related
We are planning to move from Organization Service to Common Data Service Web API so we could utilize OAuth 2.0 authentication instead of a service account which customer has some security concerns.
Once we did some prototype, we discovered that the Web API authentication is a little different from typical Graph API authentication. It only supports Delegated Permission. Thus a user credential must be presented for acquiring the access token.
Here is the Azure AD Graph API permission for CRM Web API:
Here is the code in acquiring the access token for the sample code at Web API Global Discovery Service Sample (C#)
string GlobalDiscoUrl = "https://globaldisco.crm.dynamics.com/";
AuthenticationContext authContext = new AuthenticationContext("https://login.microsoftonline.com", false);
UserCredential cred = new UserCredential(username, password);
AuthenticationResult authResult = authContext.AcquireToken(GlobalDiscoUrl, clientId, cred);
Here is another similar post Connect to Dynamics 365 Customer Engagement web services using OAuth although it is more than one year old.
Do you know when MS would support Application permission to completely eliminate the user from authentication? Or there is any particular reason to keep the user here. Thanks for any insights.
[Update 1]
With below answer from James, I did the modification for the code, here is my code
string clientId = "3f4b24d8-61b4-47df-8efc-1232a72c8817";
string secret = "xxxxx";
ClientCredential cred = new ClientCredential(clientId, secret);
string GlobalDiscoUrl = "https://globaldisco.crm.dynamics.com/";
AuthenticationContext authContext = new AuthenticationContext("https://login.microsoftonline.com/common", false);
AuthenticationResult authResult = authContext.AcquireToken(GlobalDiscoUrl, cred);
HttpClient client = new HttpClient();
client.DefaultRequestHeaders.Authorization = new AuthenticationHeaderValue("Bearer", authResult.AccessToken);
client.Timeout = new TimeSpan(0, 2, 0);
client.BaseAddress = new Uri(GlobalDiscoUrl);
HttpResponseMessage response = client.GetAsync("api/discovery/v1.0/Instances", HttpCompletionOption.ResponseHeadersRead).Result;
if (response.IsSuccessStatusCode)
{
//Get the response content and parse it.
string result = response.Content.ReadAsStringAsync().Result;
JObject body = JObject.Parse(result);
JArray values = (JArray)body.GetValue("value");
if (!values.HasValues)
{
return new List<Instance>();
}
return JsonConvert.DeserializeObject<List<Instance>>(values.ToString());
}
else
{
throw new Exception(response.ReasonPhrase);
}
}
so I am able to acquire the access token, but it still could not access the global discovery services.
Here is what the access token looks like:
{
"aud": "https://globaldisco.crm.dynamics.com/",
"iss": "https://sts.windows.net/f8cdef31-a31e-4b4a-93e4-5f571e91255a/",
"iat": 1565802457,
"nbf": 1565802457,
"exp": 1565806357,
"aio": "42FgYEj59uDNtwvxTLnprU0NYt49AA==",
"appid": "3f4b24d8-61b4-47df-8efc-1232a72c8817",
"appidacr": "1",
"idp": "https://sts.windows.net/f8cdef31-a31e-4b4a-93e4-5f571e91255a/",
"tid": "f8cdef31-a31e-4b4a-93e4-5f571e91255a",
"uti": "w8uwKBSPM0y7tdsfXtAgAA",
"ver": "1.0"
}
By the way, we did already create the application user inside CRM by following the instruction.
Anything I am missing here?
[Update 2]
For WhoAmI request, there are different results. If I am using latest MSAL and with authority "https://login.microsoftonline.com/AzureADDirectoryID/oauth2/authorize", I would be able to get the correct result. If I am using MSAL with "https://login.microsoftonline.com/common/oauth2/authorize", it won't work, I would get unauthorized error. If I am using ADAL 2.29, it is not working for both authority. Here is the working code:
IConfidentialClientApplication app = ConfidentialClientApplicationBuilder.Create("3f4b24d8-61b4-47df-8efc-1232a72cxxxx")
.WithClientSecret("xxxxxx")
// .WithAuthority("https://login.microsoftonline.com/common/oauth2/authorize", false)
.WithAuthority("https://login.microsoftonline.com/3a984a19-7f55-4ea3-a422-2d8771067f87/oauth2/authorize", false)
.Build();
var authResult = app.AcquireTokenForClient(new String[] { "https://crmxxxxx.crm5.dynamics.com/.default" }).ExecuteAsync().Result;
HttpClient client = new HttpClient();
client.DefaultRequestHeaders.Authorization = new AuthenticationHeaderValue("Bearer", authResult.AccessToken);
client.Timeout = new TimeSpan(0, 2, 0);
client.BaseAddress = new Uri("https://crm525842.api.crm5.dynamics.com/");
HttpResponseMessage response = client.GetAsync("api/data/v9.1/WhoAmI()", HttpCompletionOption.ResponseHeadersRead).Result;
if (response.IsSuccessStatusCode)
{
//Get the response content.
string result = response.Content.ReadAsStringAsync().Result;
Console.WriteLine(result);
}
else
{
throw new Exception(response.ReasonPhrase);
}
The documentation isn't the easiest to follow, but from what I understand you should start with Use OAuth with Common Data Service.
You then have two subtle options when registering your app. The second does not require the Access Dynamics 365/Common Data Service as organization users permission
Giving access to Common Data Service
If your app will be a client which allows the authenticated user to
perform operations, you must configure the application to have the
Access Dynamics 365 as organization users delegated permission.
Or
If your app will use Server-to-Server (S2S) authentication, this step
is not required. That configuration requires a specific system user
and the operations will be performed by that user account rather than
any user that must be authenticated.
This is elaborated further.
Connect as an app
Some apps you will create are not intended to be run interactively by
a user. ... In these cases you can create a special application user
which is bound to an Azure Active Directory registered application and
use either a key secret configured for the app or upload a X.509
certificate. Another benefit of this approach is that it doesn't
consume a paid license.
Register your app
When registering an app you follow many of the same steps ... with the
following exceptions:
You do not need to grant the Access Dynamics 365 as organization users permission.
You will still have a system user record in Dynamics to represent the application registration. This supports a range of basic Dynamics behaviours and allows you to apply Dynamics security to you app.
As opposed to a username and password you can then use the secret to connect.
string serviceUrl = "https://yourorg.crm.dynamics.com";
string clientId = "<your app id>";
string secret = "<your app secret>";
AuthenticationContext authContext = new AuthenticationContext("https://login.microsoftonline.com/common", false);
ClientCredential credential = new ClientCredential(clientId, secret);
AuthenticationResult result = authContext.AcquireToken(serviceUrl, credential);
string accessToken = result.AccessToken;
Or a certificate.
string CertThumbPrintId = "DC6C689022C905EA5F812B51F1574ED10F256FF6";
string AppID = "545ce4df-95a6-4115-ac2f-e8e5546e79af";
string InstanceUri = "https://yourorg.crm.dynamics.com";
string ConnectionStr = $#"AuthType=Certificate;
SkipDiscovery=true;url={InstanceUri};
thumbprint={CertThumbPrintId};
ClientId={AppID};
RequireNewInstance=true";
using (CrmServiceClient svc = new CrmServiceClient(ConnectionStr))
{
if (svc.IsReady)
{
...
}
}
You may also want to check out Build web applications using Server-to-Server (S2S) authentication which appears to be a similar (but different).
Use server-to-server (S2S) authentication to securely and seamlessly
communicate with Common Data Service with your web applications and
services. S2S authentication is the common way that apps registered on
Microsoft AppSource use to access the Common Data Service data of
their subscribers. ... Rather than user credentials, the application is authenticated based on a service principal identified by an Azure AD Object ID value which is stored in the application user record.
Aside; if you are currently using the Organization Service .NET object, that is being migrated to using the Web API internally.
Microsoft Dynamics CRM 2011 endpoint
The Dynamics 365 SDK assemblies will be updated to use the Web API.
This update will be fully transparent to you and any code written
using the SDK itself will be supported.
I'm trying to build a feature where a client application retrieves the graph resources via WebAPI layer. The scenario has following applications:
Angular5 Client application
ASP.Net Core Web API
The Angular5 client application uses MSAL to authenticate against application (resisted as Converged application via apps.dev.microsoft.com registration application; AAD v2 endpoint).
The authentication flow defines the Web API as scope while login or getting access token
constructor() {
var logger = new Msal.Logger((logLevel, message, piiEnabled) =>
{
console.log(message);
},
{ level: Msal.LogLevel.Verbose, correlationId: '12345' });
this.app = new Msal.UserAgentApplication(
CONFIGSETTINGS.clientId,
null,
this.authCallback,
{
redirectUri: window.location.origin,
cacheLocation: 'localStorage',
logger: logger
}
);
}
public getAPIAccessToken() {
return this.app.acquireTokenSilent(CONFIGSETTINGS.scopes).then(
accessToken => {
return accessToken;
},
error => {
return this.app.acquireTokenSilent(CONFIGSETTINGS.scopes).then(
accessToken => {
return accessToken;
},
err => {
console.error(err);
}
);
}
);
}
Here scope is defined as scopes: ['api://<<guid of application>>/readAccess']. This is the exact value which was generated when I've registered the Web API in registeration portal. Also, the client application id is added as Pre-authorized applications .
The Web API layer (built in dotnet core -- and uses JwtBearer to validate the authentication), defines the API which internally fetches the graph resources (using HttpClient). To get the access token, I've used following code
public async Task<string> GetAccesToken(string resourceName)
{
var userAssertion = this.GetUserAssertion();
string upn = GetLoggedInUpn();
var userTokenCache = new SessionTokenCache(upn, new Microsoft.Extensions.Caching.Memory.MemoryCache(new MemoryCacheOptions())).GetCacheInstance();
string msGraphScope = "https://graph.microsoft.com/User.Read";
string authority = string.Format("https://login.microsoftonline.com/{0}/v2.0", this.authConfig.TenantId);
ConfidentialClientApplication clientApplication = new ConfidentialClientApplication(this.authConfig.ClientId, authority, new ClientCredential(this.authConfig.AppKey), userTokenCache, null);
var result = await clientApplication.AcquireTokenOnBehalfOfAsync(new string[] { msGraphScope }, userAssertion);
return result != null ? result.AccessToken : null;
}
private UserAssertion GetUserAssertion()
{
string token = this.httpContextAccessor.HttpContext.Request.Headers["Authorization"];
string upn = GetLoggedInUpn();
if (token.StartsWith("Bearer", true, CultureInfo.InvariantCulture))
{
token = token.Trim().Substring("Bearer".Length).Trim();
return new UserAssertion(token, "urn:ietf:params:oauth:grant-type:jwt-bearer");
}
else
{
throw new Exception($"ApiAuthService.GetUserAssertion() failed: Invalid Authorization token");
}
}
Note here, The method AcquireTokenOnBehalfOfAsync is used to get the access token using graph scope. However it throws the following exception:
AADSTS65001: The user or administrator has not consented to use the application with ID '<>' named '<>'. Send an interactive authorization request for this user and resource.
I'm not sure why the of-behalf flow for AAD v2 is not working even when client application uses the Web API as scope while fetching access token and Web API registers the client application as the pre-authorized application.
Note - I've tried using the other methods of ConfidentialClientApplication but even those did not work.
Can someone please point out how the above flow can work without providing the admin consent on Web API?
I've been trying to figure this out for weeks! My solution isn't great (it requires the user to go through the consent process again for the Web API), but I'm not sure that's entirely unexpected. After all, either the Admin has to give consent for the Web API to access the graph for the user, or the user has to give consent.
Anyway, the key was getting consent from the user, which of course the Web API can't do since it has no UI. However, ConfidentialClientApplication will tell you the URL that the user has to visit with GetAuthorizationRequestUrlAsync.
Here's a snippet of the code that I used to get it working (I'm leaving out all the details of propagating the url back to the webapp, but you can check out https://github.com/rlittletht/msal-s2s-ref for a working example.)
async Task<string> GetAuthenticationUrlForConsent(ConfidentialClientApplication cca, string []graphScopes)
{
// if this throws, just let it throw
Uri uri = await cca.GetAuthorizationRequestUrlAsync(graphScopes, "", null);
return uri.AbsoluteUri;
}
async Task<string> GetAccessTokenForGraph()
{
// (be sure to use the redirectUri here that matches the Web platform
// that you added to your WebApi
ConfidentialClientApplication cca =
new ConfidentialClientApplication(Startup.clientId,
"http://localhost/webapisvc/auth.aspx",
new ClientCredential(Startup.appKey), null, null);
string[] graphScopes = {"https://graph.microsoft.com/.default"};
UserAssertion userAssertion = GetUserAssertion();
AuthenticationResult authResult = null;
try
{
authResult = await cca.AcquireTokenOnBehalfOfAsync(graphScopes, userAssertion);
}
catch (Exception exc)
{
if (exc is Microsoft.Identity.Client.MsalUiRequiredException
|| exc.InnerException is Microsoft.Identity.Client.MsalUiRequiredException)
{
// We failed because we don't have consent from the user -- even
// though they consented for the WebApp application to access
// the graph, they also need to consent to this WebApi to grant permission
string sUrl = await GetAuthenticationUrlForConsent(cca, graphScopes);
// you will need to implement this exception and handle it in the callers
throw new WebApiExceptionNeedConsent(sUrl, "WebApi does not have consent from the user to access the graph on behalf of the user", exc);
}
// otherwise, just rethrow
throw;
}
return authResult.AccessToken;
}
One of the things that I don't like about my solution is that it requires that I add a "Web" platform to my WebApi for the sole purpose of being able to give it a redirectUri when I create the ConfidentialClientApplication. I wish there was some way to just launch the consent workflow, get the user consent, and then just terminate the flow (since I don't need a token to be returned to me -- all I want is consent to be granted).
But, I'm willing to live with the extra clunky step since it actually gets consent granted and now the API can call the graph on behalf of the user.
If someone has a better, cleaner, solution, PLEASE let us know! This was incredibly frustrating to research.
I have a Web API and a UI application that used ADAL library to call the Web API.
I already gave DELEGATED PERMISSIONS (Read directory data) for both Web API and UI application while registering the apps to Azure AD.
I have below code in Web API to save Token for log-In user,
private void ConfigureAuthentication(IAppBuilder app)
{
app.UseWindowsAzureActiveDirectoryBearerAuthentication(
new WindowsAzureActiveDirectoryBearerAuthenticationOptions
{
Tenant = ConfigurationManager.AppSettings["ida:Tenant"],
TokenValidationParameters = new TokenValidationParameters { SaveSigninToken = true, ValidAudience = ConfigurationManager.AppSettings["ida:Audience"] }
});
}
Now in Web API controllers, I am trying to get token to access Microsoft AD Graph API using below code,
var bootstrapContext = ClaimsPrincipal.Current.Identities.First().BootstrapContext as System.IdentityModel.Tokens.BootstrapContext;
string userName = "test#onmicrosoft.com";
string userAccessToken = bootstrapContext.Token;
UserAssertion userAssertion = new UserAssertion(bootstrapContext.Token, "urn:ietf:params:oauth:grant-type:jwt-bearer", userName);
var authContext = new AuthenticationContext(string.Format(CultureInfo.InvariantCulture, aadInstance, tenant));
var clientCred = new ClientCredential(clientId, appKey);
var result = await authContext.AcquireTokenAsync("https://graph.windows.net", clientCred, userAssertion);
accessToken = result.AccessToken;
The above code giving me token back but the scope values is below,
`scp: "User.Read"`
Question - Why the token not giving directory access (Directory.Read.All) as I already set Directory access?
`scp: "Directory.Read.All User.Read"`
Update:
I am missing Grant Permission for Read Directory Data under DELEGATED PERMISSIONS. After giving Grant Permission I am able to get token with scope scp: "Directory.Read.All User.Read"
If I understand correctly, you want to use Microsoft Graph API ,not Azure AD Graph API.
However, based on that the screenshot you post in this question is a v1 enpoint AAD Application, it does nothing about Microsoft Graph API which you're trying to approaching. So, whatever you changed on this application, the result shouldn be same. I suggest you register v2 enpoint Application in https://apps.dev.microsoft.com/
Here is a document which shows how to get auth tokens for using Microsoft Graph.
Hope this helps!
I have a mobile app that talks to a backend web API (core 2.0). Presently I have the API configured to use Opendidict with Facebook integration based on the configuration listed below.
public static IServiceCollection AddAuthentication(this IServiceCollection services, AppSettings settings)
{
services.AddOpenIddict<int>(options =>
{
options.AddEntityFrameworkCoreStores<RouteManagerContext>();
options.AddMvcBinders();
options.EnableAuthorizationEndpoint("/auth/authorize");
options.EnableTokenEndpoint("/auth/token");
options.AllowAuthorizationCodeFlow();
options.AllowImplicitFlow();
options.AllowPasswordFlow();
options.AllowRefreshTokenFlow();
options.SetAccessTokenLifetime(TimeSpan.FromMinutes(1));
options.SetRefreshTokenLifetime(TimeSpan.FromMinutes(20160));
options.DisableHttpsRequirement();
options.AddEphemeralSigningKey();
});
services.AddAuthentication()
.AddFacebook(o => { o.ClientId = settings.FacebookAppID; o.ClientSecret = settings.FacebookAppSecret; })
.AddOAuthValidation();
return services;
}
The password flow works perfectly when they want to use local account. What I'm struggling with is how to return the access/refresh token after successfully authenticating with Facebook. I have the standard account controller with ExternalLogin and ExternalLoginCallback which also works perfectly as I'm able to successfully login and get the local user account it's tied to and signed in.
In my mind, the user clicks facebook login, which calls ExternalLogincallBack, which logs in the user. After that all I want to do is return the access/refresh token just like the password flow.
When I try to use the ImplicitFlow by providing the implicit flow arguments in the redirect (/auth/authorize?...) from ExternalLoginCallback, I can get the access token, but no refresh token even if I specify the offline_scope. From what I read, it seems the implicit flow doesn't support refresh so I tried code flow.
When using the CodeFlow, I can get the code token from the redirect to "/auth/authorize" but can't figure out how to call into the token endpoint from the authorize endpoint to return the access/refresh token directly to the client app. Do I just need to return the code to the client and have them make another call to post to the token endpoint to get access/refresh tokens?
This doesn't feel correct and I'm stumped. Seems like I should be able to just return the access/refresh token after I've signed in externally just like what happens with password flow. Any help would be greatly appreciated as I've been struggling with this for several days.
[HttpGet("~/auth/authorize")]
public async Task<IActionResult> Authorize(OpenIdConnectRequest request)
{
if (!User.Identity.IsAuthenticated)
{
// If the client application request promptless authentication,
// return an error indicating that the user is not logged in.
if (request.HasPrompt(OpenIdConnectConstants.Prompts.None))
{
var properties = new AuthenticationProperties(new Dictionary<string, string>
{
[OpenIdConnectConstants.Properties.Error] = OpenIdConnectConstants.Errors.LoginRequired,
[OpenIdConnectConstants.Properties.ErrorDescription] = "The user is not logged in."
});
// Ask OpenIddict to return a login_required error to the client application.
return Forbid(properties, OpenIdConnectServerDefaults.AuthenticationScheme);
}
return Challenge();
}
// Retrieve the profile of the logged in user.
var user = await _userManager.GetUserAsync(User);
if (user == null)
{
return BadRequest(new
{
Error = OpenIdConnectConstants.Errors.ServerError,
ErrorDescription = "An internal error has occurred"
});
}
// Create a new authentication ticket.
var ticket = await CreateTicketAsync(request, user);
// Returning a SignInResult will ask OpenIddict to issue the appropriate access/identity tokens.
return SignIn(ticket.Principal, ticket.Properties, ticket.AuthenticationScheme);
}
private async Task<AuthenticationTicket> CreateTicketAsync(OpenIdConnectRequest request, ApplicationUser user, AuthenticationProperties properties = null)
{
// Create a new ClaimsPrincipal containing the claims that will be used to create an id_token, a token or a code.
var principal = await _signInManager.CreateUserPrincipalAsync(user);
// Create a new authentication ticket holding the user identity.
var ticket = new AuthenticationTicket(principal, properties, OpenIdConnectServerDefaults.AuthenticationScheme);
if (!request.IsRefreshTokenGrantType())
{
// Set the list of scopes granted to the client application.
// Note: the offline_access scope must be granted to allow OpenIddict to return a refresh token.
ticket.SetScopes(new[]
{
OpenIdConnectConstants.Scopes.OpenId,
OpenIdConnectConstants.Scopes.Email,
OpenIdConnectConstants.Scopes.Profile,
OpenIdConnectConstants.Scopes.OfflineAccess,
OpenIddictConstants.Scopes.Roles
}.Intersect(request.GetScopes()));
}
ticket.SetResources("RouteManagerAPI");
// Note: by default, claims are NOT automatically included in the access and identity tokens.
// To allow OpenIddict to serialize them, you must attach them to a destination, that specifies
// whether they should be included in access tokens, in identity tokens or in both.
foreach (var claim in ticket.Principal.Claims)
{
// Never include the security stamp in the access and identity tokens, as it's a secret value.
if (claim.Type == _identityOptions.Value.ClaimsIdentity.SecurityStampClaimType)
{
continue;
}
var destinations = new List<string>
{
OpenIdConnectConstants.Destinations.AccessToken
};
// Only add the iterated claim to the id_token if the corresponding scope was granted to the client application.
// The other claims will only be added to the access_token, which is encrypted when using the default format.
if ((claim.Type == OpenIdConnectConstants.Claims.Name && ticket.HasScope(OpenIdConnectConstants.Scopes.Profile)) ||
(claim.Type == OpenIdConnectConstants.Claims.Email && ticket.HasScope(OpenIdConnectConstants.Scopes.Email)) ||
(claim.Type == OpenIdConnectConstants.Claims.Role && ticket.HasScope(OpenIddictConstants.Claims.Roles)))
{
destinations.Add(OpenIdConnectConstants.Destinations.IdentityToken);
}
claim.SetDestinations(destinations);
}
return ticket;
}
When I try to use the CodeFlow, I can get the code token but can't figure out how to call into the token endpoint from the authorize endpoint to return the access/refresh token directly to the client app. Do I just need to return the code to the client and have them make another call to post to the token endpoint to get access/refresh tokens?
That's exactly what you're supposed to do as the code flow is a 2-part process: once your mobile apps has an authorization code, it must redeem it using a simple HTTP call to the token endpoint to get an access token and a refresh token.
OK I am now days into this and have made significant progress but am still completely stumped about the fundamentals.
My application uses Cognito User Pools for creating and managing users - these are identified on S3 it seems by their IdentityId. Each of my users has their own S3 folder, and AWS automatically gives them a folder name that is equal to the user's IdentityId.
I need to relate the IdentityId to the other Cognito user information but cannot work out how.
The key thing I need is to be able to identify the username plus other cognito user attributes for a given IdentityId - and it's insanely hard.
So the first battle was to work out how to get the IdentityId when a Cognito user does a request via the AWS API Gateway. Finally I got that worked out, and now I have a Cognito user, who does a request to the API Gateway, and my Lambda function behind that now has the IdentityId. That bit works.
But I am completely stumped as to how to now access the Cognito user's information that is stored in the user pool. I can't find any clear information, and certainly no code, that shows how to use the IdentityId to get the Cognito user's attributes, username etc.
It appears that if I use a "Cognito user pool" to authorize my method in API Gateway, then the body mapping template can be used to put Cognito User information such as the sub and the username and email address into the context, BUT I do NOT get the IdentityId.
BUT if I use the AWS_IAM to authorize my method in the API gateway then the body mapping template does the inverse - it gives me the IdentityId but not the Cognito user fields such as sub and username and email.
It's driving me crazy - how can I get the IdentityId and all the Cognito users fields and attributes together into one data structure? The fact that I seem to be only able to get one or the other just makes no sense.
It turns out that to get the IdentityId AND user details at the same time using AWS Lambda/Cognito/API Gateway, you need to have a Lambda function that is authenticated using AWS_IAM (NOT COGNITO_USER_POOLS), you must send your request the AWS API Gateway, BUT it MUST be a signed request, you must then modify the integration request body mapping templates so that you are given the IdentityId in the event (maybe the context? can't remember). Now you have the IdentityId. Phew. Now you must submit the client's Cognito ID token from the front end to the back end. It is important to validate the token - you cannot trust that it has not been tampered with if you do not validate it. To decode and validate the token you must get the keys from your userpool, put them into your script, ensure that you have jwt decoding libraries plus signature validation libraries included in your AWS lambda zipfile. Now your script must validate the token submitted from the front end and then you can get the user details out of the token. Voila! Now you have both IdentityId plus user details such as their sub, username and email address. So easy.
The above is what is takes to get the username associated with an IdentityId using AWS Cognito/Lambda/API Gateway. This took me days to get working.
Can I please say to any Amazon employees who wander across this ........ well it's WAY too hard to get the user details associated with an IdentityId. You need to fix this. It made me angry that this was so hard and burned so much of my time.
The solution:
I did this by modifying an Amazon employees custom authorizer here:
https://s3.amazonaws.com/cup-resources/cup_custom_authorizer_lambda_function_blueprint.zip
as found and described here:
https://aws.amazon.com/blogs/mobile/integrating-amazon-cognito-user-pools-with-api-gateway/
use strict';
let util = require('util');
var jwt = require('jsonwebtoken');
var jwkToPem = require('jwk-to-pem');
var userPoolId = 'YOUR USERPOOL ID';
var region = 'YOUR REGION'; //e.g. us-east-1
var iss = 'https://cognito-idp.' + region + '.amazonaws.com/' + userPoolId;
//https://docs.aws.amazon.com/cognito/latest/developerguide/amazon-cognito-user-pools-using-tokens-with-identity-providers.html
// DOWNLOAD FROM https://cognito-idp.{region}.amazonaws.com/{userPoolId}/.well-known/jwks.json
let userPoolKeys = {PUT YOUR DOWNLOADED USER POOL KEYS JSON HERE};
var pems = {};
let convertKeysToPems = () => {
var keys = userPoolKeys['keys'];
for(var i = 0; i < keys.length; i++) {
//Convert each key to PEM
var key_id = keys[i].kid;
var modulus = keys[i].n;
var exponent = keys[i].e;
var key_type = keys[i].kty;
var jwk = { kty: key_type, n: modulus, e: exponent};
var pem = jwkToPem(jwk);
pems[key_id] = pem;
}
}
exports.handler = function(event, context) {
convertKeysToPems()
console.log(event);
let token = event['body-json'].cognitoUserToken;
console.log(event['body-json'].cognitoUserToken);
ValidateToken(pems, event, context, token);
};
let ValidateToken = (pems, event, context, token) => {
//Fail if the token is not jwt
var decodedJwt = jwt.decode(token, {complete: true});
console.log(decodedJwt)
if (!decodedJwt) {
console.log("Not a valid JWT token");
context.fail("Unauthorized");
return;
}
//Fail if token is not from your UserPool
if (decodedJwt.payload.iss != iss) {
console.log("invalid issuer");
context.fail("Unauthorized");
return;
}
//Reject the jwt if it's not an 'Access Token'
if (decodedJwt.payload.token_use != 'id') {
console.log("Not an id token");
context.fail("Unauthorized");
return;
}
//Get the kid from the token and retrieve corresponding PEM
var kid = decodedJwt.header.kid;
var pem = pems[kid];
if (!pem) {
console.log(pems, 'pems');
console.log(kid, 'kid');
console.log('Invalid token');
context.fail("Unauthorized");
return;
}
//Verify the signature of the JWT token to ensure it's really coming from your User Pool
jwt.verify(token, pem, { issuer: iss }, function(err, payload) {
if(err) {
context.fail("Unauthorized");
} else {
let x = decodedJwt.payload
x.identityId = context.identity.cognitoIdentityId
//let x = {'identityId': context['cognito-identity-id'], 'decodedJwt': decodedJwt}
console.log(x);
context.succeed(x);
}
});
}
This problem -- the problem of using the user's sub instead of their identityId in S3 paths and how to set that up -- is 100% solved for me thanks to the help of #JesseDavda's solution to this problem in this issue: https://github.com/aws-amplify/amplify-js/issues/54
For all of you developers who have been trying to get the identityId in lambdas so that your Amplify default paths in S3 work - this solution simply ends up ignoring identityId altogether - it is a solution that sets up the paths in S3 based on sub instead of the identityId. At the end of this solution, you will never have to deal with more than one id for your users, you will never have to deal with identityId (hopefully) ever again.
If I'm understanding this correctly you want the CognitoIdentityId and the User attributes in the same place. How we do it is the following way:
From the event request context we get the IdentityId:
event.requestContext.identity.cognitoIdentityId
Also from the request context we get the user's sub:
event.requestContext.identity.cognitoAuthenticationProvider.split(':CognitoSignIn:')[1]
Then with the sub you can request the rest of the attributes the following way:
const AWS = require('aws-sdk');
let cognito = new AWS.CognitoIdentityServiceProvider();
let request = {
Username: userSub,
UserPoolId: process.env.userPoolId,
};
let result = await cognito.adminGetUser(request).promise();
const userAttributes = result.UserAttributes.reduce((acc, attribute) => {
const { Name, Value } = attribute;
acc[Name] = Value;
return acc;
}, {});
return userAttributes;