Merge branch 'main' into update-dependencies

* main:
  🔏 Requires valid access tokens for API calls (#173)

.env.sample

@@ -0,0 +1,13 @@
+# Postgres URL for Chowda DB.
+# Format: postgresql://{user}:{password}@{host}:{port}/{database}
+DB_URL=''
+
+# Env vars beginning with "AUTH0_" are used for authentication and authorization with Auth0.
+# These values can be obtained by logging into our Auth0 account.
+AUTH0_DOMAIN=''
+AUTH0_CLIENT_ID=''
+AUTH0_CLIENT_SECRET=''
+AUTH0_API_AUDIENCE=''
+
+# Secret value for securing session data using SessionMiddleware
+CHOWDA_SECRET=''

.gitguardian.yaml

@@ -0,0 +1,5 @@
+secret:
+  ignored-matches:
+  - match: 73f13c96260784cee71e44a63798a1a6716f19d6581a2ce28b391c4d8449bd19
+    name: Default jwt.io Bearer Token, for auth tests - tests/routers/test_events.py
+version: 2

.github/workflows/CI.yml

@@ -2,9 +2,6 @@ name: 🧪 Integration Tests
 
 on: [push, pull_request, workflow_dispatch]
 
-env:
-  DB_URL: postgresql://postgres:postgres@postgres:5432/chowda-test
-
 jobs:
   tests:
     name: ⚗️ Application Tests
@@ -13,7 +10,7 @@ jobs:
     with:
       pdm_args: -G test,ci
       pytest_args: -n auto --nbmake -ra -s
-      pg_db: chowda-test
+      pg_db: chowda
 
   lint:
     name: 👕 Lint

.gitignore

@@ -23,4 +23,9 @@ ci.toml
 
 # Metaflow generated files
 metaflow.s3.*
-.metaflow
+.metaflow
+
+.cache_ggshield
+
+.env.*
+!.env.sample

.vscode/launch.json

@@ -18,20 +18,6 @@
       "jinja": false,
       "justMyCode": false
     },
-    {
-      "name": "Chowda: tests",
-      "type": "python",
-      "request": "launch",
-      "cwd": "${workspaceFolder}",
-      "module": "pytest",
-      "args": [
-        "-v",
-        "-n",
-        "auto"
-      ],
-      "jinja": false,
-      "justMyCode": false
-    },
     {
       "name": "Metaflow: Run",
       "type": "python",

chowda/app.py

@@ -2,7 +2,7 @@
 
 Main Chowda application"""
 
-from fastapi import FastAPI
+from fastapi import Depends, FastAPI
 from fastapi.staticfiles import StaticFiles
 from starlette.middleware import Middleware
 from starlette.middleware.sessions import SessionMiddleware
@@ -13,6 +13,7 @@
 from chowda.admin import Admin
 from chowda.api import api
 from chowda.auth import OAuthProvider
+from chowda.auth.utils import get_admin_user, verified_access_token
 from chowda.config import SECRET, STATIC_DIR, TEMPLATES_DIR
 from chowda.db import engine
 from chowda.models import (
@@ -53,8 +54,10 @@
 )
 app.mount('/static', StaticFiles(directory=STATIC_DIR), name='static')
 
-app.include_router(api, prefix='/api')
-app.include_router(dashboard, prefix='/dashboard')
+app.include_router(api, prefix='/api', dependencies=[Depends(verified_access_token)])
+app.include_router(
+    dashboard, prefix='/dashboard', dependencies=[Depends(get_admin_user)]
+)
 
 
 # Create admin

chowda/auth/oauth.py

@@ -10,7 +10,7 @@
 from starlette_admin.auth import AdminUser, AuthMiddleware, AuthProvider
 
 from chowda.config import (
-    API_AUDIENCE,
+    AUTH0_API_AUDIENCE,
     AUTH0_CLIENT_ID,
     AUTH0_CLIENT_SECRET,
     AUTH0_DOMAIN,
@@ -25,7 +25,7 @@
         'scope': 'openid profile email',
     },
     server_metadata_url=f'https://{AUTH0_DOMAIN}/.well-known/openid-configuration',
-    authorize_params={'audience': API_AUDIENCE},
+    authorize_params={'audience': AUTH0_API_AUDIENCE},
 )
 
 

chowda/auth/utils.py

@@ -1,17 +1,35 @@
-from typing import Annotated
+from typing import Annotated, List, Set
 
 from fastapi import Depends, HTTPException, Request, status
 from pydantic import BaseModel, Field
 
-from chowda.config import API_AUDIENCE
+from chowda.config import AUTH0_API_AUDIENCE, AUTH0_JWKS_URL
+
+unauthorized_redirect = HTTPException(
+    status_code=status.HTTP_303_SEE_OTHER,
+    detail='Not Authorized',
+    headers={'Location': '/admin'},
+)
+
+unauthorized = HTTPException(
+    status_code=status.HTTP_401_UNAUTHORIZED,
+    detail='Not Authorized',
+)
 
 
-class UserToken(BaseModel):
+class OAuthAccessToken(BaseModel):
+    """OAuth Authorization token"""
+
+    sub: str
+    permissions: List[str] = []
+
+
+class OAuthUser(BaseModel):
     """ID Token model for authorization."""
 
     name: str
     email: str | None = None
-    roles: set[str] = Field(set(), alias=f'{API_AUDIENCE}/roles')
+    roles: set[str] = Field(set(), alias=f'{AUTH0_API_AUDIENCE}/roles')
 
     @property
     def is_admin(self) -> bool:
@@ -24,29 +42,107 @@ def is_clammer(self) -> bool:
         return 'clammer' in self.roles
 
 
-unauthorized = HTTPException(
-    status_code=status.HTTP_303_SEE_OTHER,
-    detail='Not Authorized',
-    headers={'Location': '/admin'},
-)
-
-
-def get_user(request: Request) -> UserToken:
+def get_oauth_user(request: Request) -> OAuthUser:
     """Get the user token from the session."""
     user = request.session.get('user', None)
     if not user:
         request.session['error'] = 'Not Logged In'
-        raise unauthorized
-    return UserToken(**user)
+        raise unauthorized_redirect
+    return OAuthUser(**user)
 
 
-def admin_user(
-    request: Request, user: Annotated[UserToken, Depends(get_user)]
-) -> UserToken:
+def get_admin_user(
+    request: Request, user: Annotated[OAuthUser, Depends(get_oauth_user)]
+) -> OAuthUser:
     """Check if the user has the admin role using FastAPI Depends.
     If not, sets a session error and raises an HTTPException."""
     if not user.is_admin:
         request.session['error'] = 'Not Authorized'
-        raise unauthorized
+        raise unauthorized_redirect
 
     return user
+
+
+def unverified_access_token(request: Request) -> str:
+    """Extract and return the unverified access token from the Authorization header.
+    Raises an HTTPUnauthorizedException if the header is missing or malformed."""
+    auth_header = request.headers.get('Authorization', None)
+
+    if not auth_header:
+        raise HTTPException(status_code=401, detail="Missing Authorization header")
+    if not auth_header.startswith('Bearer '):
+        raise HTTPException(
+            status_code=401,
+            detail="Bearer token malformed or missing in Authorization header",
+        )
+
+    # Return the access token from the Authorization header,
+    # i.e. the string without the "Bearer " prefix
+    return auth_header.replace('Bearer ', '')
+
+
+def jwt_signing_key(
+    unverified_access_token: Annotated[str, Depends(unverified_access_token)]
+) -> str:
+    """Get the JWT signing key from the JWKS URL."""
+    from jwt import PyJWKClient
+
+    jwks_client = PyJWKClient(AUTH0_JWKS_URL)
+    try:
+        signing_key = jwks_client.get_signing_key_from_jwt(unverified_access_token)
+        return signing_key.key
+
+    except Exception as exc:
+        raise HTTPException(
+            status_code=status.HTTP_401_UNAUTHORIZED, detail=str(exc)
+        ) from exc
+
+
+def verified_access_token(
+    request: Request,
+    unverified_access_token: Annotated[str, Depends(unverified_access_token)],
+    jwt_signing_key: Annotated[str, Depends(jwt_signing_key)],
+) -> OAuthAccessToken:
+    """Decodes and verifies an access token. If any exceptions occur, an
+    HTTPUnauthorizedException is raised from the original exception."""
+    from jwt import decode
+
+    try:
+        decoded_token = decode(
+            unverified_access_token,
+            jwt_signing_key,
+            algorithms=['RS256', 'HS256'],
+            audience='https://chowda.wgbh-mla.org/api',
+        )
+        return OAuthAccessToken(**decoded_token)
+
+    except Exception as exc:
+        raise HTTPException(
+            status_code=status.HTTP_401_UNAUTHORIZED, detail=str(exc)
+        ) from exc
+
+
+def permissions(permissions: str | List[str] | Set[str]) -> None:
+    """Dependency function to check if token has required permissions.
+
+    Args:
+        permissions (str, List, Set): Required permissions. Can be a str, list, or set.
+    Examples:
+        @app.get('/users/', dependencies=[Depends(permissions('read:users'))])
+    """
+    if isinstance(permissions, (str, list)):
+        permissions: set = {permissions}
+
+    def _permissions(
+        token: Annotated[OAuthAccessToken, Depends(verified_access_token)],
+    ) -> None:
+        """Verify token has all required permissions, or raise a 403 Forbidden exception
+        with the missing permissions in the detail message."""
+        missing_permissions = permissions - set(token.permissions)
+        if missing_permissions:
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail=f'Missing required permissions: {missing_permissions}',
+            )
+
+    return _permissions

chowda/config.py

@@ -16,7 +16,9 @@
 AUTH0_CLIENT_ID = environ.get('AUTH0_CLIENT_ID')
 AUTH0_CLIENT_SECRET = environ.get('AUTH0_CLIENT_SECRET')
 AUTH0_DOMAIN = environ.get('AUTH0_DOMAIN')
+AUTH0_JWKS_URL = f'https://{AUTH0_DOMAIN}/.well-known/jwks.json'
+AUTH0_API_AUDIENCE = environ.get(
+    'AUTH0_API_AUDIENCE', 'https://chowda.wgbh-mla.org/api'
+)
 
 SECRET = environ.get('CHOWDA_SECRET')
-
-API_AUDIENCE = environ.get('API_AUDIENCE', 'https://chowda.wgbh-mla.org/api')

chowda/routers/dashboard.py

@@ -1,10 +1,8 @@
-from fastapi import APIRouter, Depends, Request, status
+from fastapi import APIRouter, Request, status
 from metaflow.integrations import ArgoEvent
 from starlette.responses import RedirectResponse, Response
 
-from chowda.auth.utils import admin_user
-
-dashboard = APIRouter(dependencies=[Depends(admin_user)])
+dashboard = APIRouter()
 
 
 @dashboard.post('/sync')

chowda/routers/events.py

@@ -1,36 +1,54 @@
-from json import loads
+from json import JSONDecodeError, loads
 
-from fastapi import APIRouter, HTTPException
+from fastapi import APIRouter, Depends, HTTPException, status
 from metaflow import Run, namespace
 from sqlmodel import Session
 
+from chowda.auth.utils import permissions
 from chowda.db import engine
 from chowda.models import MetaflowRun
 
 events = APIRouter()
 
 
-@events.post('/')
-def event(event: dict):
+@events.post('/', dependencies=[Depends(permissions('create:event'))])
+async def event(event: dict):
     """Receive an event from Argo Events."""
     print('Chowda event received', event)
-    if not event.get('body'):
-        raise HTTPException(400, 'No body')
-    body = loads(event['body'])
-    if body['name'] == 'pipeline':
+    body = event.get('body')
+    if not body:
+        raise HTTPException(
+            status.HTTP_400_BAD_REQUEST,
+            'Argo Event must include a `body` key as a string',
+        )
+    try:
+        body = loads(event['body'])
+    except JSONDecodeError as e:
+        raise HTTPException(
+            status.HTTP_400_BAD_REQUEST, 'Argo Event body must be valid JSON'
+        ) from e
+    name = body.get('name')
+    if not name:
+        raise HTTPException(
+            status.HTTP_400_BAD_REQUEST,
+            "Argo Event body string must include a `name` key",
+        )
+    if name == 'pipeline':
         print('new pipeline event!')
         # FIXME: Ideally, we would add the run to the database here,
         # but the run_id won't be minted until metaflow gets this event.
         # For now, we can continue creating the db row inside the running flow,
         # then update metaflow status events, as they come in.
-        return
+        return None
     if body['name'].startswith('metaflow.Pipeline'):
         print('Found event!', body['name'])
         payload = body['payload']
         with Session(engine) as db:
             row = db.get(MetaflowRun, payload['run_id'])
             if not row:
-                raise HTTPException(404, 'MetaflowRun row not found!')
+                raise HTTPException(
+                    status.HTTP_404_NOT_FOUND, 'MetaflowRun row not found!'
+                )
             namespace(None)
             run = Run(f"{payload['flow_name']}/{payload['run_id']}")
             row.finished = run.finished
@@ -41,5 +59,5 @@ def event(event: dict):
             db.add(row)
             db.commit()
             print('Successfully updated MetaflowRun row!', row)
-            return
-    raise HTTPException(400, 'Unknown event')
+            return None
+    return 'Event successfully processed, but did not match known event'