External Jupyter Notebook

Distinguishing between browser and non-browser requests

With OAuth2 enabled, when an unauthenticated user tries to access or perform an action against ModelOp Center using an external Jupyter notebook, they will not be allowed to do so until they authenticate. The trigger for the authentication logic, located in the jupyter-plugin repository, is a 401 Unauthorized status code. However, Spring Security configured as OAuth2Login is no longer sending a 401 Unauthorized, but instead a 302 Found status code. To override the 302 Found status code, we introduced the CustomAuthenticationEntryPoint class. The purpose of this class is to inspect all requests for protected resources coming from an unauthenticated user, and to determine if a 302 Found or 401 Unauthorized status code should be returned.
We identify external Jupyter notebook requests (non-browser requests) by checking for the Accept header and its value. If a request is coming from an external Jupyter notebook (non-browser request), then the Accept header will not contain the text/html value, in which case we return 401 Unauthorized. Otherwise, we return 302 Found and redirect to the login page.

...

http.exceptionHandling().authenticationEntryPoint(new CustomAuthenticationEntryPoint(URI.create(this.authenticationFailureRedirectUri)));

Accessing OAuth2 configuration

In order to retrieve the OAuth2 configuration information, a new endpoint was introduced: /api/oauth2/.well-known-configuration. This endpoint is used by all ModelOp OAuth2 clients to retrieve generic and client specific information such as authorization URI, client id, response type, scope, and redirect.

...