uc-cdis · flashguerdon · Aug 20, 2024 · Aug 27, 2024 · Aug 28, 2024 · Aug 28, 2024
diff --git a/.gitignore b/.gitignore
@@ -108,3 +108,6 @@ tests/resources/keys/*.pem
 .DS_Store
 .vscode
 .idea
+
+# snyk
+.dccache
diff --git a/fence/__init__.py b/fence/__init__.py
@@ -470,6 +470,7 @@ def _setup_oidc_clients(app):
                 logger=logger,
                 HTTP_PROXY=config.get("HTTP_PROXY"),
                 idp=settings.get("name") or idp.title(),
+                arborist=app.arborist,
             )
             clean_idp = idp.lower().replace(" ", "")
             setattr(app, f"{clean_idp}_client", client)

diff --git a/fence/blueprints/login/base.py b/fence/blueprints/login/base.py
@@ -1,8 +1,12 @@
+import time
+import base64
+import json
+from urllib.parse import urlparse, urlencode, parse_qsl
+import jwt
+import requests
 import flask
 from cdislogging import get_logger
 from flask_restful import Resource
-from urllib.parse import urlparse, urlencode, parse_qsl
-
 from fence.auth import login_user
 from fence.blueprints.login.redirect import validate_redirect
 from fence.config import config
@@ -20,7 +24,7 @@ def __init__(self, idp_name, client):
         Args:
             idp_name (str): name for the identity provider
             client (fence.resources.openid.idp_oauth2.Oauth2ClientBase):
-                Some instaniation of this base client class or a child class
+                Some instantiation of this base client class or a child class
         """
         self.idp_name = idp_name
         self.client = client
@@ -92,8 +96,27 @@ def __init__(
         self.is_mfa_enabled = "multifactor_auth_claim_info" in config[
             "OPENID_CONNECT"
         ].get(self.idp_name, {})
+
+        # Config option to explicitly persist refresh tokens
+        self.persist_refresh_token = False
+
+        self.read_authz_groups_from_tokens = False
+
         self.app = app
 
+        # This block of code probably need to be made more concise
+        if "persist_refresh_token" in config["OPENID_CONNECT"].get(self.idp_name, {}):
+            self.persist_refresh_token = config["OPENID_CONNECT"][self.idp_name][
+                "persist_refresh_token"
+            ]
+
+        if "is_authz_groups_sync_enabled" in config["OPENID_CONNECT"].get(
+            self.idp_name, {}
+        ):
+            self.read_authz_groups_from_tokens = config["OPENID_CONNECT"][
+                self.idp_name
+            ]["is_authz_groups_sync_enabled"]
+
     def get(self):
         # Check if user granted access
         if flask.request.args.get("error"):
@@ -119,7 +142,11 @@ def get(self):
 
         code = flask.request.args.get("code")
         result = self.client.get_auth_info(code)
+
+        refresh_token = result.get("refresh_token")
+
         username = result.get(self.username_field)
+
         if not username:
             raise UserError(
                 f"OAuth2 callback error: no '{self.username_field}' in {result}"
@@ -129,11 +156,157 @@ def get(self):
         id_from_idp = result.get(self.id_from_idp_field)
 
         resp = _login(username, self.idp_name, email=email, id_from_idp=id_from_idp)
-        self.post_login(user=flask.g.user, token_result=result, id_from_idp=id_from_idp)
+
+        expires = self.extract_exp(refresh_token)
+
+        # if the access token is not a JWT, or does not carry exp,
+        # default to now + REFRESH_TOKEN_EXPIRES_IN
+        if expires is None:
+            expires = int(time.time()) + config["REFRESH_TOKEN_EXPIRES_IN"]
+
+        # Store refresh token in db
+        should_persist_token = (
+            self.persist_refresh_token or self.read_authz_groups_from_tokens
+        )
+        if should_persist_token:
+            # Ensure flask.g.user exists to avoid a potential AttributeError
+            if getattr(flask.g, "user", None):
+                self.client.store_refresh_token(flask.g.user, refresh_token, expires)
+            else:
+                logger.error(
+                    "User information is missing from flask.g; cannot store refresh token."
+                )
+
+        self.post_login(
+            user=flask.g.user,
+            token_result=result,
+            id_from_idp=id_from_idp,
+        )
+
         return resp
 
+    def extract_exp(self, refresh_token):
+        """
+        Extract the expiration time (`exp`) from a refresh token.
+
+        This function attempts to retrieve the expiration time from the provided
+        refresh token using three methods:
+
+        1. Using PyJWT to decode the token (without signature verification).
+        2. Introspecting the token (if supported by the identity provider).
+        3. Manually base64 decoding the token's payload (if it's a JWT).
+
+        **Disclaimer:** This function assumes that the refresh token is valid and
+        does not perform any JWT validation. For JWTs from an OpenID Connect (OIDC)
+        provider, validation should be done using the public keys provided by the
+        identity provider (from the JWKS endpoint) before using this function to
+        extract the expiration time. Without validation, the token's integrity and
+        authenticity cannot be guaranteed, which may expose your system to security
+        risks. Ensure validation is handled prior to calling this function,
+        especially in any public or production-facing contexts.
+
+        Args:
+            refresh_token (str): The JWT refresh token from which to extract the expiration.
+
+        Returns:
+            int or None: The expiration time (`exp`) in seconds since the epoch,
+            or None if extraction fails.
+        """
+
+        # Method 1: PyJWT
+        try:
+            # Skipping keys since we're not verifying the signature
+            decoded_refresh_token = jwt.decode(
+                refresh_token,
+                options={
+                    "verify_aud": False,
+                    "verify_at_hash": False,
+                    "verify_signature": False,
+                },
+                algorithms=["RS256", "HS512"],
+            )
+            exp = decoded_refresh_token.get("exp")
+
+            if exp is not None:
+                return exp
+        except Exception as e:
+            logger.info(f"Refresh token expiry: Method (PyJWT) failed: {e}")
+
+        # Method 2: Introspection
+        try:
+            introspection_response = self.introspect_token(refresh_token)
+            exp = introspection_response.get("exp")
+
+            if exp is not None:
+                return exp
+        except Exception as e:
+            logger.info(f"Refresh token expiry: Method Introspection failed: {e}")
+
+        # Method 3: Manual base64 decoding
+        try:
+            # Assuming the token is a JWT (header.payload.signature)
+            payload_encoded = refresh_token.split(".")[1]
+            # Add necessary padding for base64 decoding
+            payload_encoded += "=" * (4 - len(payload_encoded) % 4)
+            payload_decoded = base64.urlsafe_b64decode(payload_encoded)
+            payload_json = json.loads(payload_decoded)
+            exp = payloa
F438
d_json.get("exp")
+
+            if exp is not None:
+                return exp
+        except Exception as e:
+            logger.info(f"Method 3 (Manual decoding) failed: {e}")
+
+        # If all methods fail, return None
+        return None
+
+    def introspect_token(self, token):
+        """Introspects an access token to determine its validity and retrieve associated metadata.
+
+        This method sends a POST request to the introspection endpoint specified in the OpenID
+        discovery document. The request includes the provided token and client credentials,
+        allowing verification of the token's validity and retrieval of any additional metadata
+        (e.g., token expiry, scopes, or user information).
+
+        Args:
+            token (str): The access token to be introspected.
+
+        Returns:
+            dict or None: A dictionary containing the token's introspection data if the request
+            is successful and the response status code is 200. If the introspection fails or an
+            exception occurs, returns None.
+
+        Raises:
+            Exception: Logs an error message if an error occurs during the introspection process.
+        """
+        try:
+            introspect_endpoint = self.client.get_value_from_discovery_doc(
+                "introspection_endpoint", ""
+            )
+
+            # Headers and payload for the introspection request
+            headers = {"Content-Type": "application/x-www-form-urlencoded"}
+            data = {
+                "token": token,
+                "client_id": flask.session.get("client_id"),
+                "client_secret": flask.session.get("client_secret"),
+            }
+
+            response = requests.post(introspect_endpoint, headers=headers, data=data)
+
+            if response.status_code == 200:
+                return response.json()
+            else:
+                logger.info(f"Error introspecting token: {response.status_code}")
+                return None
+
+        except Exception as e:
+            logger.info(f"Error introspecting token: {e}")
+            return None
+
     def post_login(self, user=None, token_result=None, **kwargs):
         prepare_login_log(self.idp_name)
+
         metrics.add_login_event(
             user_sub=flask.g.user.id,
             idp=self.idp_name,
@@ -142,6 +315,13 @@ def post_login(self, user=None, token_result=None, **kwargs):
             client_id=flask.session.get("client_id"),
         )
 
+        # this attribute is only applicable to some OAuth clients
+        # (e.g., not all clients need is_read_authz_groups_from_tokens_enabled)
+        if self.read_authz_groups_from_tokens:
+            self.client.update_user_authorization(
+                user=user, pkey_cache=None, db_session=None, idp_name=self.idp_name
+            )
+
         if token_result:
             username = token_result.get(self.username_field)
             if self.is_mfa_enabled:

diff --git a/fence/config-default.yaml b/fence/config-default.yaml
@@ -94,6 +94,7 @@ DB_MIGRATION_POSTGRES_LOCK_KEY: 100
 #   - WARNING: Be careful changing the *_ALLOWED_SCOPES as you can break basic
 #              and optional functionality
 # //////////////////////////////////////////////////////////////////////////////////////
+
 OPENID_CONNECT:
   # any OIDC IDP that does not differ from the generic implementation can be
   # configured without code changes
@@ -115,6 +116,24 @@ OPENID_CONNECT:
     multifactor_auth_claim_info: # optional, include if you're using arborist to enforce mfa on a per-file level
       claim: '' # claims field that indicates mfa, either the acr or acm claim.
       values: [ "" ] # possible values that indicate mfa was used. At least one value configured here is required to be in the token
+    # When true, it allows refresh tokens to be stored even if is_authz_groups_sync_enabled is set false.
+    # When false, the system will only store refresh tokens if is_authz_groups_sync_enabled is enabled
+    persist_refresh_token: false
+    # is_authz_groups_sync_enabled: A configuration flag that determines whether the application should
+    # verify and synchronize user group memberships between the identity provider (IdP)
+    # and the local authorization system (Arborist). When enabled, the refresh token is stored, the system retrieves
+    # the user's group information from their token issued by the IdP and compares it against
+    # the groups defined in the local system. Based on the comparison, the user is added to
+    # or removed from relevant groups in the local system to ensure their group memberships
+    # remain up-to-date. If this flag is disabled, no group synchronization occurs
+    is_authz_groups_sync_enabled: true
+    authz_groups_sync:
+      # This defines the prefix used to identify authorization groups.
+      group_prefix: "some_prefix"
+    # This flag indicates whether the audience (aud) claim in the JWT should be verified during token validation.
+    verify_aud: true
+    # This specifies the expected audience (aud) value for the JWT, ensuring that the token is intended for use with the 'fence' service.
+    audience: fence
   # These Google values must be obtained from Google's Cloud Console
   # Follow: https://developers.google.com/identity/protocols/OpenIDConnect
   #

diff --git a/fence/config.py b/fence/config.py
@@ -145,6 +145,13 @@ def post_process(self):
                     f"IdP '{idp_id}' is using multifactor_auth_claim_info '{mfa_info['claim']}', which is neither AMR or ACR. Unable to determine if a user used MFA. Fence will continue and assume they have not used MFA."
                 )
 
+            groups_sync_enabled = idp.get("is_authz_groups_sync_enabled", False)
+            # when is_authz_groups_sync_enabled, then you must provide authz_groups_sync, with group prefix
+            if groups_sync_enabled and not idp.get("authz_groups_sync"):
+                error = f"Error: is_authz_groups_sync_enabled is enabled, required values not configured, for idp: {idp_id}"
+                logger.error(error)
+                raise Exception(error)
+
         self._validate_parent_child_studies(self._configs["dbGaP"])
 
     @staticmethod

diff --git a/fence/error_handler.py b/fence/error_handler.py
@@ -8,12 +8,13 @@
 
 from fence.errors import APIError
 from fence.config import config
+import traceback
 
 
 logger = get_logger(__name__)
 
 
-def get_error_response(error):
+def get_error_response(error: Exception):
     details, status_code = get_error_details_and_status(error)
     support_email = config.get("SUPPORT_EMAIL_FOR_ERRORS")
     app_name = config.get("APP_NAME", "Gen3 Data Commons")
@@ -27,6 +28,11 @@ def get_error_response(error):
         )
     )
 
+    # TODO: Issue: Error messages are obfuscated, the line below needs be
+    #  uncommented when troubleshooting errors.
+    #  Breaks tests if not commented out / removed. We need a fix for this.
+    # raise error
+
     # don't include internal details in the public error message
     # to do this, only include error messages for known http status codes
     # that are less that 500
-Original file line number
+Diff line change
@@ Expand Up / @@ -108,3 +108,6 @@ tests/resources/keys/*.pem @@
     .DS_Store
     .vscode
     .idea
+    # snyk
+    .dccache