Allow authentication by JWT Bearer token

Dockerfile

@@ -230,8 +230,18 @@ EOF
 RUN echo "import os \nDEBUG = os.getenv('SP7_DEBUG', '').lower() == 'true'\n" \
         > settings/debug.py
 
-RUN echo "import os \nSECRET_KEY = os.environ['SECRET_KEY']\n" \
-        > settings/secret_key.py
+RUN cat <<EOF > settings/secret_key.py
+import os
+DEFAULT_KEY="$(cat /dev/urandom | tr -dc 'a-zA-Z0-9' | head -c 50)"
+CURRENT_KEY=os.getenv("SECRET_KEY")
+
+if CURRENT_KEY is None or CURRENT_KEY.strip() == "" or CURRENT_KEY.strip().replace(" ", "_") == "change_this_to_some_unique_random_string":
+    new_key = DEFAULT_KEY
+else:
+    new_key = CURRENT_KEY
+
+SECRET_KEY=new_key
+EOF
 
 ENV LC_ALL=C.UTF-8
 ENV LANG=C.UTF-8

specifyweb/backend/accounts/access_token_utils.py

@@ -0,0 +1,76 @@
+import uuid
+
+import jwt
+
+from datetime import datetime, timezone, timedelta
+from typing import Literal
+
+from django.conf import settings
+
+from specifyweb.backend.redis_cache.store import set_string, key_exists
+
+DEFAULT_AUTH_LIFESPAN_SECONDS = 1800
+
+# See https://pyjwt.readthedocs.io/en/latest/api.html#jwt.decode
+AUTH_JWT_DECODE_OPTIONS = {
+    "require": ["iat", "exp", "jti"],
+    "verify_signature": True,
+    "verify_iat": True,
+    "verify_exp": True
+}
+
+AUTH_TOKEN_ALGORITHMS = ["HS256"]
+
+
+def generate_access_token(user, collection_id: int, expires_in: int = DEFAULT_AUTH_LIFESPAN_SECONDS):
+    jti = str(uuid.uuid4())
+
+    jwt_payload = {
+        "sub": user.id,
+        "username": user.name,
+        "collection": collection_id,
+        "jti": jti,
+        "iat": datetime.now(timezone.utc),
+        "exp": datetime.now(timezone.utc) + timedelta(seconds=expires_in)
+    }
+    token = jwt.encode(jwt_payload, settings.SECRET_KEY,
+                       algorithm=AUTH_TOKEN_ALGORITHMS[0])
+    return token
+
+
+def revoke_access_token(token: dict):
+    """
+    Accepts and revokes a decoded JWT Auth Token.
+    Specifically, stores the token in a "blacklist" in Redis for the remaining
+    time of the token.
+    The JWT Auth Middleware checks to see if the token is blacklisted during
+    authorization
+    """
+    required_claims = ("jti", "exp")
+    if not all(k in token for k in required_claims):
+        raise ValueError(f"Token missing required claims: {required_claims}")
+    jti = token["jti"]
+    expires_at = token["exp"]
+    current_time = int(datetime.now(timezone.utc).timestamp())
+    blacklist_ttl = expires_at - current_time
+    set_string(f"revoked:{jti}", "true", time_to_live=blacklist_ttl)
+
+
+def get_token_from_request(request) -> Literal[False] | None | dict:
+    auth_header = request.headers.get("Authorization")
+    if auth_header is None or not auth_header.startswith("Bearer "):
+        return None
+
+    encoded_token = auth_header.split(" ")[1]
+
+    try:
+        token = jwt.decode(encoded_token, settings.SECRET_KEY,
+                           options=AUTH_JWT_DECODE_OPTIONS, algorithms=AUTH_TOKEN_ALGORITHMS)
+    except jwt.exceptions.InvalidTokenError:
+        return False
+    return token
+
+
+def token_is_revoked(token: dict):
+    token_identifier = token["jti"]
+    return key_exists(f"revoked:{token_identifier}")

specifyweb/backend/accounts/middleware.py

@@ -0,0 +1,64 @@
+from django.utils.functional import SimpleLazyObject
+from django.core.exceptions import PermissionDenied
+from django.http import HttpResponse
+
+from specifyweb.specify.models import Collection, Specifyuser, Agent
+from specifyweb.specify.api.filter_by_col import filter_by_collection
+from specifyweb.backend.accounts.access_token_utils import get_token_from_request, token_is_revoked
+from specifyweb.backend.context.views import has_collection_access
+
+
+def get_agent(request):
+    try:
+        return filter_by_collection(Agent.objects, request.specify_collection) \
+            .select_related('specifyuser') \
+            .get(specifyuser=request.specify_user)
+    except Agent.DoesNotExist:
+        return None
+
+
+class JWTAuthMiddleware:
+    def __init__(self, get_response):
+        self.get_response = get_response
+
+    def __call__(self, request):
+        token = get_token_from_request(request)
+        # The request doesn't have an access token, so pass through
+        if token is None:
+            return self.get_response(request)
+
+        # There was an access token in the request, but it was invalid or
+        # revoked. Stop here and return a 401 Unauthorized
+        if token == False or token_is_revoked(token):
+            response = HttpResponse('Invalid access token', status=401)
+            response["WWW-Authenticate"] = 'error=\"invalid_token\", error_description=\"The access token is expired, revoked, or invalid\"'
+            return response
+
+        user_id = token["sub"]
+        collection_id = token["collection"]
+
+        # This shouldn't happen often in practice as this is also enforced when
+        # the tokens are generated, but just in case a token is forged or the
+        # user's collection access was revoked since the token was generated,
+        # this prevents users from accessing Collections they shouldn't
+        if not has_collection_access(collection_id, user_id):
+            raise PermissionDenied()
+
+        request.specify_collection = SimpleLazyObject(
+            lambda: Collection.objects.get(id=collection_id))
+        lazy_user = SimpleLazyObject(
+            lambda: Specifyuser.objects.get(id=user_id))
+        request.specify_user = lazy_user
+        request.user = lazy_user
+        request.specify_user_agent = SimpleLazyObject(
+            lambda: get_agent(request))
+
+        # We can disable CSRF checks with users authenticated via JWT.
+        # This is ONLY because the end user must explicitly pass the auth token
+        # as a header, and is not stored within the session, cookies, etc.
+        # Essentially, with CSRF protection disabled for users authenticated
+        # via token, we have to be careful not to store any auth information in
+        # a stateful way within the session
+        # e.g., avoid calling django.contrib.auth.login
+        request._dont_enforce_csrf_checks = True
+        return self.get_response(request)

specifyweb/backend/accounts/urls.py

@@ -13,6 +13,9 @@
     # OpenId Connect callback endpoint:
     path('oic_callback/', views.oic_callback),
 
+    path('token/', views.acquire_access_token),
+    path('token/revoke/', views.revoke_access_token),
+
     path(
         'logout/',
         skip_collection_access_check(auth_views.LogoutView.as_view(next_page='/accounts/login/'))

specifyweb/backend/accounts/views.py

@@ -6,11 +6,10 @@
 import logging
 import requests
 import time
-from urllib.parse import unquote_plus
 from django import forms
 from django import http
 from django.conf import settings
-from django.contrib.auth import login
+from django.contrib.auth import login, authenticate
 from django.contrib.auth.models import AbstractBaseUser
 from django.db import connection
 from django.db.models import Max
@@ -19,6 +18,7 @@
 from django.utils import crypto
 from django.utils.http import url_has_allowed_host_and_scheme, urlencode
 from django.views.decorators.cache import never_cache
+from django.views.decorators.csrf import csrf_exempt
 from typing import cast
 
 from specifyweb.backend.accounts.account_utils import check_collection_access_against_agents, is_provider_info
@@ -27,15 +27,17 @@
 from specifyweb.backend.accounts.permissions_types import InviteLinkPT, SetPasswordPT, SetUserAgentsPT, Sp6AdminPT, UserOICProvidersPT
 from specifyweb.backend.accounts.types import ExternalUser, InviteToken, OAuthLogin, ProviderConf, ProviderInfo
 from specifyweb.middleware.general import require_GET, require_http_methods
+from specifyweb.backend.context.views import has_collection_access, set_collection_cookie, users_collections_for_sp7
 
 from specifyweb.backend.permissions.permissions import check_permission_targets
 from specifyweb.specify import models as spmodels
 from specifyweb.specify.views import login_maybe_required, openapi
 from .models import Spuserexternalid
-from specifyweb.specify.models import Specifyuser
+from specifyweb.specify.models import Specifyuser, Collection
 from django.views.decorators.http import require_POST
 from specifyweb.backend.permissions.permissions import check_permission_targets
 from specifyweb.specify.auth.support_login import b64_url_to_bytes
+from specifyweb.backend.accounts.access_token_utils import DEFAULT_AUTH_LIFESPAN_SECONDS, generate_access_token, revoke_access_token as revoke_token, AUTH_JWT_DECODE_OPTIONS, AUTH_TOKEN_ALGORITHMS
 from django.db import transaction, connection
 
 logger = logging.getLogger(__name__)
@@ -356,7 +358,6 @@
     if not settings.ALLOW_SUPPORT_LOGIN:
         return http.HttpResponseForbidden()
 
-    from django.contrib.auth import login, authenticate
     token = request.GET["token"]
     key = b64_url_to_bytes(request.GET["key"])
 
@@ -582,3 +583,147 @@
     else:
         user.clear_admin()
         return http.HttpResponse('false', content_type='text/plain')
+
+
+@openapi(schema={
+    'post': {
+        "requestBody": {
+            "required": True,
+            "description": "Obtain an access token that can be used with the API",
+            "content": {
+                "application/x-www-form-urlencoded": {
+                    "schema": {
+                        "type": "object",
+                        "properties": {
+                            "username": {
+                                "type": "string"
+                            },
+                            "password": {
+                                "type": "string"
+                            },
+                            "collectionid": {
+                                "type": "integer"
+                            },
+                            "expires": {
+                                "type": "integer",
+                                "description": f"The number of seconds this token should be valid for. Defaults to {DEFAULT_AUTH_LIFESPAN_SECONDS / 60} minutes if not specified",
+                                "default": DEFAULT_AUTH_LIFESPAN_SECONDS
+                            }
+                        },
+                        "required": ['username', 'password', 'collectionid'],
+                        'additionalProperties': False
+                    }
+                }
+            }
+        },
+        "responses": {
+            "200": {
+                "description": "The access token was successfully generated",
+                "content": {
+                    "application/json": {
+                        "schema": {
+                            "type": "object",
+                            "properties": {
+                                "access_token": {
+                                    "type": "string"
+                                },
+                                "expires_in": {
+                                    "type": "integer",
+                                    "description": "The number of seconds the access token is live for."
+                                }
+                            }
+                        }
+                    }
+                }
+            },
+            "400": {"description": "One of the required keys was not supplied, or one or more keys were supplied incorrectly"},
+            "403": {"description": "The provided credentials were incorrect, or the user does not have access to the collection"},
+            "405": {"description": "A non-POST method was made to the endpoint. Only POST is supported"}
+        }
+    },
+})
+@require_POST
+@csrf_exempt
+def acquire_access_token(request):
+    username = request.POST.get("username")
+    password = request.POST.get("password")
+    collection_id = request.POST.get("collectionid")
+    raw_expires_in = request.POST.get("expires", DEFAULT_AUTH_LIFESPAN_SECONDS)
+
+    if None in (username, password, collection_id):
+        return http.HttpResponseBadRequest(f"username, password, and collection are required")
+
+    try:
+        expires_in = int(raw_expires_in)
+        if expires_in <= 0:
+            return http.HttpResponseBadRequest(f"Invalid expiry time")
+    except ValueError:
+        return http.HttpResponseBadRequest(f"Expiry time could be parsed as integer")
+
+
+    try:
+        collection = Collection.objects.get(id=collection_id)
+    except Collection.DoesNotExist:
+        return http.HttpResponseBadRequest(f'collection {collection_id} does not exist')
+
+    user = authenticate(username=username, password=password)
+
+    if user is None or not has_collection_access(collection.id, user.id):
+        return http.HttpResponseForbidden()
+
+    token = generate_access_token(user, collection.id, expires_in=expires_in)
+
+    # TODO: lower default expiry time and also issue refresh tokens that can be
+    # used to re-issue auth tokens
+    # Issue refresh tokens as HTTP-only cookies? Need to keep CSRF protection
+    # and replay attacks in mind. Maybe just store them in Redis...
+    response = {
+        "access_token": token,
+        "expires_in": expires_in
+    }
+    return http.JsonResponse(response)
+
+
+@openapi(schema={
+    'post': {
+        "requestBody": {
+            "required": True,
+            "description": "Revoke a previously granted access token",
+            "content": {
+                "application/x-www-form-urlencoded": {
+                    "schema": {
+                        "type": "object",
+                        "properties": {
+                            "access_token": {
+                                "type": "string",
+                                "description": "The JWT Access token to revoke"
+                            },
+                        },
+                        "required": ['access_token'],
+                        'additionalProperties': False
+                    }
+                }
+            }
+        },
+        "responses": {
+            "204": {"description": "The access token was revoked"},
+            "400": {"description": "The provided token is already invalid or could not be validated"},
+            "403": {"description": "The user was not logged-in when making a request to the endpoint"},
+            "405": {"description": "A non-POST method was made to the endpoint. Only POST is supported"}
+        }
+    },
+})
+@require_POST
+@login_maybe_required
+def revoke_access_token(request):
+    encoded_token = request.POST.get("access_token")
+
+    try:
+        token = jwt.decode(encoded_token, settings.SECRET_KEY,
+                           options=AUTH_JWT_DECODE_OPTIONS, algorithms=AUTH_TOKEN_ALGORITHMS)
+    except jwt.exceptions.InvalidTokenError:
+        return http.HttpResponseBadRequest()
+
+    revoke_token(token)
+
+    return http.HttpResponse('', status=204)

specifyweb/backend/cache/redis/store.py

@@ -1,4 +1,4 @@
-from .utils import _set_string, _get_string, _delete_key, _add_to_set, _remove_from_set, _set_elements, _redis_type, format_key
+from .utils import _set_string, _get_string, _delete_key, _add_to_set, _remove_from_set, _set_elements, _redis_type, format_key, _key_exists
 
 # REFACTOR: Replace these with RedisConnection adapters
 
@@ -36,3 +36,7 @@ def delete_key(key: str | bytes):
 
 def redis_type(key: str | bytes):
     return _redis_type(format_key(key))
+
+
+def key_exists(key: str) -> bool:
+    return _key_exists(key)