Improve security when one tenant is comporomised

.github/workflows/code-quality.yml

@@ -120,3 +120,28 @@ jobs:
       - name: '[check] Run Ruff'
         working-directory: ./service
         run: make typecheck
+
+  service-test:
+    name: 'Service: Test'
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: '[setup] Check out repository'
+        uses: actions/checkout@v7
+
+      - name: '[setup] Install uv'
+        uses: astral-sh/setup-uv@v7
+        with:
+          python-version: ${{ env.PYTHON_VERSION }}
+
+      - name: '[setup] Set up Python'
+        working-directory: ./service
+        run: uv python install
+
+      - name: '[app] Install'
+        working-directory: ./service
+        run: make install
+
+      - name: '[check] Run pytest'
+        working-directory: ./service
+        run: make test

service/Makefile

@@ -10,6 +10,10 @@ lint:
 typecheck:
 	uv run ty check
 
+.PHONY: test
+test:
+	uv run pytest
+
 .PHONY: format
 format:
 	uv run ruff format

service/config.template.yaml

@@ -1,9 +1,14 @@
-# Specify urls which can be used for authenticating the user. The user will send the url in a header 'X-Dsw-Api-Url'
+# Specify urls and tenant ids which can be used for authenticating the user. The user will send the url in a header 'X-Dsw-Api-Url'
 # Use the base url for dsw api, not the frontend.
 # For example: https://researchers.dsw.elixir-europe.org/wizard-api
+# Get the tenant UUID from the JWT token
+# For debugging purposes, you can set url or tenant_uuid to * to allow all urls or tenant uuids.
 auth:
-  allowed_project_urls:
-    - "https://your-dsw-instance.example.com/wizard-api"
+  allowed_apis:
+    - url: "https://your-dsw-instance.example.com/wizard-api"
+      tenant_uuid: "123e4567-e89b-12d3-a456-426614174000"
+    - url: "*"
+      tenant_uuid: "*"
 
 logging:
   level: "INFO"

service/config.test.yaml

@@ -0,0 +1,20 @@
+auth:
+  allowed_apis:
+    - url: "https://your-dsw-instance.example.com/wizard-api"
+      tenant_uuid: "123e4567-e89b-12d3-a456-426614174000"
+
+logging:
+  level: "INFO"
+
+database:
+  host: "localhost"
+  port: "5432"
+  name: "ai_document_plugin"
+  user: "ai_document_plugin"
+  password: "ai_document_plugin"
+  schema: "public"
+
+files:
+  prompts_path: "prompts.yaml"
+
+max_parallel_executions: 2

service/pyproject.toml

@@ -44,8 +44,13 @@ dev = [
     "ruff",
     "ty",
     "pytest",
+    "pytest-asyncio"
 ]
 
+[tool.pytest.ini_options]
+asyncio_mode = "auto"
+asyncio_default_fixture_loop_scope = "function"
+
 [build-system]
 requires = ["uv_build>=0.11.23,<0.12.0"]
 build-backend = "uv_build"

service/src/ai_document_plugin_service/ai/common/config.py

@@ -1,12 +1,17 @@
+import logging
 import os
 import pathlib
 from dataclasses import dataclass
 from typing import Any
+from uuid import UUID
 
 import yaml
 
+logger = logging.getLogger(__name__)
+
 DEFAULT_CONFIG_PATH = 'config.yaml'
 CONFIG_PATH_ENV_VAR = 'AI_DOCUMENT_PLUGIN_CONFIG_PATH'
+WILDCARD = '*'
 
 
 @dataclass(frozen=True)
@@ -29,6 +34,12 @@ class FilePaths:
     prompts_path: str
 
 
+@dataclass(frozen=True)
+class AllowedApi:
+    url: str
+    tenant_uuid: str
+
+
 @dataclass(frozen=True)
 class DatabaseConfig:
     host: str
@@ -41,7 +52,7 @@ class DatabaseConfig:
 
 @dataclass(frozen=True)
 class Config:
-    allowed_project_urls: tuple[str, ...]
+    allowed_apis: tuple[AllowedApi, ...]
     log_level: str
     database: DatabaseConfig
     files: FilePaths
@@ -114,20 +125,49 @@ def normalize_project_url(url: str) -> str:
     return url.strip().rstrip('/')
 
 
-def _get_allowed_project_urls(config: dict[str, Any]) -> tuple[str, ...]:
-    raw_urls = _get(config, 'auth', 'allowed_project_urls')
-    if not isinstance(raw_urls, list) or not raw_urls:
-        msg = "Invalid config value: 'auth.allowed_project_urls' must be a non-empty list"
-        raise ValueError(msg)
+def _get_allowed_apis(config: dict[str, Any]) -> tuple[AllowedApi, ...]:
+    raw_apis = _get(config, 'auth', 'allowed_apis')
+    if not isinstance(raw_apis, list) or not raw_apis:
+        msg = "Invalid config value: 'auth.allowed_apis' must be a non-empty list"
+        raise TypeError(msg)
 
-    normalized: list[str] = []
-    for index, entry in enumerate(raw_urls):
-        if not isinstance(entry, str) or not entry.strip():
-            msg = f"Invalid config value: 'auth.allowed_project_urls[{index}]' must be a non-empty string"
+    allowed_apis: list[AllowedApi] = []
+    for index, entry in enumerate(raw_apis):
+        if not isinstance(entry, dict):
+            msg = f"Invalid config value: 'auth.allowed_apis[{index}]' must be a mapping"
+            raise TypeError(msg)
+
+        raw_url = entry.get('url')
+        raw_tenant_uuid = entry.get('tenant_uuid')
+        if not isinstance(raw_url, str) or not raw_url.strip():
+            msg = f"Invalid config value: 'auth.allowed_apis[{index}].url' must be a non-empty string"
+            raise ValueError(msg)
+        if not isinstance(raw_tenant_uuid, str) or not raw_tenant_uuid.strip():
+            msg = f"Invalid config value: 'auth.allowed_apis[{index}].tenant_uuid' must be a non-empty string"
             raise ValueError(msg)
-        normalized.append(normalize_project_url(entry))
 
-    return tuple(normalized)
+        url = raw_url.strip()
+        tenant_uuid = raw_tenant_uuid.strip()
+        normalized_url = url if url == WILDCARD else normalize_project_url(url)
+        if tenant_uuid != WILDCARD:
+            try:
+                tenant_uuid = str(UUID(tenant_uuid))
+            except ValueError as error:
+                msg = f"Invalid config value: 'auth.allowed_apis[{index}].tenant_uuid' must be a UUID or '*'"
+                raise ValueError(msg) from error
+
+        if WILDCARD in {normalized_url, tenant_uuid}:
+            logger.warning(
+                "Do not use in production! Config 'auth.allowed_apis[%d]' uses a wildcard (url=%s, tenant_uuid=%s) "
+                'which allows anyone to use this API.',
+                index,
+                normalized_url,
+                tenant_uuid,
+            )
+
+        allowed_apis.append(AllowedApi(url=normalized_url, tenant_uuid=tenant_uuid))
+
+    return tuple(allowed_apis)
 
 
 def _resolve_existing_path(path: str, *, base_dir: pathlib.Path | None = None) -> str:
@@ -182,7 +222,7 @@ def load_config(config_path: str | None = None) -> Config:
         raise TypeError(msg)
 
     return Config(
-        allowed_project_urls=_get_allowed_project_urls(config),
+        allowed_apis=_get_allowed_apis(config),
         log_level=_get_log_level(config),
         database=DatabaseConfig(
             host=_expand_env_vars(_get(config, 'database', 'host')),

service/src/ai_document_plugin_service/api/auth.py

@@ -5,7 +5,7 @@
 import fastapi
 import httpx
 
-from ai_document_plugin_service.ai.common.config import Config, normalize_project_url
+from ai_document_plugin_service.ai.common.config import WILDCARD, AllowedApi, Config, normalize_project_url
 from ai_document_plugin_service.api.jwt import extract_identity_from_token
 
 DSW_API_URL_HEADER = 'X-Dsw-Api-Url'
@@ -21,8 +21,17 @@ class AuthenticatedUser:
     tenant_uuid: UUID
 
 
-def is_allowed_project_url(api_url: str, allowed_project_urls: tuple[str, ...]) -> bool:
-    return normalize_project_url(api_url) in allowed_project_urls
+def is_allowed_request(api_url: str, tenant_uuid: UUID, allowed_apis: tuple[AllowedApi, ...]) -> bool:
+    normalized_api_url = normalize_project_url(api_url)
+    tenant_uuid_str = str(tenant_uuid)
+
+    for entry in allowed_apis:
+        url_matches = entry.url in {WILDCARD, normalized_api_url}
+        tenant_matches = entry.tenant_uuid in {WILDCARD, tenant_uuid_str}
+        if not (url_matches and tenant_matches):
+            continue
+        return True
+    return False
 
 
 def _parse_bearer_token(authorization: str | None) -> str | None:
@@ -65,15 +74,15 @@ def verify_authenticated(
     config: Config = request.app.state.config
     normalized_api_url = normalize_project_url(dsw_api_url)
 
-    if not is_allowed_project_url(normalized_api_url, config.allowed_project_urls):
-        raise fastapi.HTTPException(status_code=401, detail='Unauthorized')
-
-    if not _validate_dsw_user(normalized_api_url, token):
-        raise fastapi.HTTPException(status_code=401, detail='Unauthorized')
-
     try:
         user_uuid, tenant_uuid = extract_identity_from_token(token)
     except ValueError as error:
         raise fastapi.HTTPException(status_code=400, detail=str(error)) from error
 
+    if not is_allowed_request(normalized_api_url, tenant_uuid, config.allowed_apis):
+        raise fastapi.HTTPException(status_code=401, detail='Unauthorized')
+
+    if not _validate_dsw_user(normalized_api_url, token):
+        raise fastapi.HTTPException(status_code=401, detail='Unauthorized')
+
     return AuthenticatedUser(token=token, api_url=normalized_api_url, user_uuid=user_uuid, tenant_uuid=tenant_uuid)