forked from phoenix/litellm-mirror
Merge pull request #5494 from BerriAI/litellm_allow_setting_route_controls
[Feat-Proxy] Enterprise - allow controlling allowed private, public, admin only routes
This commit is contained in:
commit
e9701ffa36
10 changed files with 563 additions and 240 deletions
|
@ -22,7 +22,7 @@ This covers:
|
|||
- ✅ [SSO for Admin UI](./proxy/ui#✨-enterprise-features)
|
||||
- ✅ [Audit Logs with retention policy](./proxy/enterprise#audit-logs)
|
||||
- ✅ [JWT-Auth](../docs/proxy/token_auth.md)
|
||||
- ✅ [Control available public, private routes](./proxy/enterprise#control-available-public-private-routes)
|
||||
- ✅ [Control available public, private routes (Restrict certain endpoints on proxy)](./proxy/enterprise#control-available-public-private-routes)
|
||||
- ✅ [**Secret Managers** AWS Key Manager, Google Secret Manager, Azure Key](./secret)
|
||||
- ✅ IP address‑based access control lists
|
||||
- ✅ Track Request IP Address
|
||||
|
|
|
@ -16,8 +16,9 @@ Features:
|
|||
- ✅ [SSO for Admin UI](./ui.md#✨-enterprise-features)
|
||||
- ✅ [Audit Logs with retention policy](#audit-logs)
|
||||
- ✅ [JWT-Auth](../docs/proxy/token_auth.md)
|
||||
- ✅ [Control available public, private routes (Restrict certain endpoints on proxy)](#control-available-public-private-routes)
|
||||
- ✅ [Control available public, private routes](#control-available-public-private-routes)
|
||||
- ✅ [**Secret Managers** AWS Key Manager, Google Secret Manager, Azure Key](../secret)
|
||||
- ✅ [[BETA] AWS Key Manager v2 - Key Decryption](#beta-aws-key-manager---key-decryption)
|
||||
- ✅ IP address‑based access control lists
|
||||
- ✅ Track Request IP Address
|
||||
- ✅ [Use LiteLLM keys/authentication on Pass Through Endpoints](pass_through#✨-enterprise---use-litellm-keysauthentication-on-pass-through-endpoints)
|
||||
|
@ -609,24 +610,35 @@ Expected Response
|
|||
|
||||
## Control available public, private routes
|
||||
|
||||
**Restrict certain endpoints of proxy**
|
||||
|
||||
:::info
|
||||
|
||||
❓ Use this when you want to make an existing private route -> public
|
||||
|
||||
Example - Make `/spend/calculate` a publicly available route (by default `/spend/calculate` on LiteLLM Proxy requires authentication)
|
||||
❓ Use this when you want to:
|
||||
- make an existing private route -> public
|
||||
- set certain routes as admin_only routes
|
||||
|
||||
:::
|
||||
|
||||
#### Usage - Define public routes
|
||||
#### Usage - Define public, admin only routes
|
||||
|
||||
**Step 1** - set allowed public routes on config.yaml
|
||||
**Step 1** - Set on config.yaml
|
||||
|
||||
|
||||
| Route Type | Optional | Requires Virtual Key Auth | Admin Can Access | All Roles Can Access | Description |
|
||||
|------------|----------|---------------------------|-------------------|----------------------|-------------|
|
||||
| `public_routes` | ✅ | ❌ | ✅ | ✅ | Routes that can be accessed without any authentication |
|
||||
| `admin_only_routes` | ✅ | ✅ | ✅ | ❌ | Routes that can only be accessed by [Proxy Admin](./self_serve#available-roles) |
|
||||
| `allowed_routes` | ✅ | ✅ | ✅ | ✅ | Routes are exposed on the proxy. If not set then all routes exposed. |
|
||||
|
||||
`LiteLLMRoutes.public_routes` is an ENUM corresponding to the default public routes on LiteLLM. [You can see this here](https://github.com/BerriAI/litellm/blob/main/litellm/proxy/_types.py)
|
||||
|
||||
```yaml
|
||||
general_settings:
|
||||
master_key: sk-1234
|
||||
public_routes: ["LiteLLMRoutes.public_routes", "/spend/calculate"]
|
||||
public_routes: ["LiteLLMRoutes.public_routes", "/spend/calculate"] # routes that can be accessed without any auth
|
||||
admin_only_routes: ["/key/generate"] # Optional - routes that can only be accessed by Proxy Admin
|
||||
allowed_routes: ["/chat/completions", "/spend/calculate", "LiteLLMRoutes.public_routes"] # Optional - routes that can be accessed by anyone after Authentication
|
||||
```
|
||||
|
||||
**Step 2** - start proxy
|
||||
|
@ -637,6 +649,10 @@ litellm --config config.yaml
|
|||
|
||||
**Step 3** - Test it
|
||||
|
||||
<Tabs>
|
||||
|
||||
<TabItem value="public" label="Test `public_routes`">
|
||||
|
||||
```shell
|
||||
curl --request POST \
|
||||
--url 'http://localhost:4000/spend/calculate' \
|
||||
|
@ -649,6 +665,97 @@ curl --request POST \
|
|||
|
||||
🎉 Expect this endpoint to work without an `Authorization / Bearer Token`
|
||||
|
||||
</TabItem>
|
||||
|
||||
<TabItem value="admin_only_routes" label="Test `admin_only_routes`">
|
||||
|
||||
|
||||
**Successfull Request**
|
||||
|
||||
```shell
|
||||
curl --location 'http://0.0.0.0:4000/key/generate' \
|
||||
--header 'Authorization: Bearer <your-master-key>' \
|
||||
--header 'Content-Type: application/json' \
|
||||
--data '{}'
|
||||
```
|
||||
|
||||
|
||||
**Un-successfull Request**
|
||||
|
||||
```shell
|
||||
curl --location 'http://0.0.0.0:4000/key/generate' \
|
||||
--header 'Authorization: Bearer <virtual-key-from-non-admin>' \
|
||||
--header 'Content-Type: application/json' \
|
||||
--data '{"user_role": "internal_user"}'
|
||||
```
|
||||
|
||||
**Expected Response**
|
||||
|
||||
```json
|
||||
{
|
||||
"error": {
|
||||
"message": "user not allowed to access this route. Route=/key/generate is an admin only route",
|
||||
"type": "auth_error",
|
||||
"param": "None",
|
||||
"code": "403"
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
|
||||
</TabItem>
|
||||
|
||||
<TabItem value="allowed_routes" label="Test `allowed_routes`">
|
||||
|
||||
|
||||
**Successfull Request**
|
||||
|
||||
```shell
|
||||
curl http://localhost:4000/chat/completions \
|
||||
-H "Content-Type: application/json" \
|
||||
-H "Authorization: Bearer sk-1234" \
|
||||
-d '{
|
||||
"model": "fake-openai-endpoint",
|
||||
"messages": [
|
||||
{"role": "user", "content": "Hello, Claude"}
|
||||
]
|
||||
}'
|
||||
```
|
||||
|
||||
|
||||
**Un-successfull Request**
|
||||
|
||||
```shell
|
||||
curl --location 'http://0.0.0.0:4000/embeddings' \
|
||||
--header 'Content-Type: application/json' \
|
||||
-H "Authorization: Bearer sk-1234" \
|
||||
--data ' {
|
||||
"model": "text-embedding-ada-002",
|
||||
"input": ["write a litellm poem"]
|
||||
}'
|
||||
```
|
||||
|
||||
**Expected Response**
|
||||
|
||||
```json
|
||||
{
|
||||
"error": {
|
||||
"message": "Route /embeddings not allowed",
|
||||
"type": "auth_error",
|
||||
"param": "None",
|
||||
"code": "403"
|
||||
}
|
||||
}
|
||||
```
|
||||
|
||||
|
||||
</TabItem>
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
</Tabs>
|
||||
|
||||
## Guardrails - Secret Detection/Redaction
|
||||
❓ Use this to REDACT API Keys, Secrets sent in requests to an LLM.
|
||||
|
|
|
@ -26,7 +26,7 @@ from litellm.proxy._types import (
|
|||
LitellmUserRoles,
|
||||
UserAPIKeyAuth,
|
||||
)
|
||||
from litellm.proxy.auth.auth_utils import is_llm_api_route
|
||||
from litellm.proxy.auth.route_checks import is_llm_api_route
|
||||
from litellm.proxy.utils import PrismaClient, ProxyLogging, log_to_opentelemetry
|
||||
from litellm.types.services import ServiceLoggerPayload, ServiceTypes
|
||||
|
||||
|
@ -40,22 +40,6 @@ else:
|
|||
all_routes = LiteLLMRoutes.openai_routes.value + LiteLLMRoutes.management_routes.value
|
||||
|
||||
|
||||
def is_request_body_safe(request_body: dict) -> bool:
|
||||
"""
|
||||
Check if the request body is safe.
|
||||
|
||||
A malicious user can set the api_base to their own domain and invoke POST /chat/completions to intercept and steal the OpenAI API key.
|
||||
Relevant issue: https://huntr.com/bounties/4001e1a2-7b7a-4776-a3ae-e6692ec3d997
|
||||
"""
|
||||
banned_params = ["api_base", "base_url"]
|
||||
|
||||
for param in banned_params:
|
||||
if param in request_body:
|
||||
raise ValueError(f"BadRequest: {param} is not allowed in request body")
|
||||
|
||||
return True
|
||||
|
||||
|
||||
def common_checks(
|
||||
request_body: dict,
|
||||
team_object: Optional[LiteLLM_TeamTable],
|
||||
|
|
|
@ -1,13 +1,123 @@
|
|||
import re
|
||||
import sys
|
||||
import traceback
|
||||
from typing import List, Optional, Tuple
|
||||
|
||||
from fastapi import Request
|
||||
from fastapi import HTTPException, Request, status
|
||||
|
||||
from litellm._logging import verbose_proxy_logger
|
||||
from litellm.proxy._types import *
|
||||
|
||||
|
||||
def _get_request_ip_address(
|
||||
request: Request, use_x_forwarded_for: Optional[bool] = False
|
||||
) -> Optional[str]:
|
||||
|
||||
client_ip = None
|
||||
if use_x_forwarded_for is True and "x-forwarded-for" in request.headers:
|
||||
client_ip = request.headers["x-forwarded-for"]
|
||||
elif request.client is not None:
|
||||
client_ip = request.client.host
|
||||
else:
|
||||
client_ip = ""
|
||||
|
||||
return client_ip
|
||||
|
||||
|
||||
def _check_valid_ip(
|
||||
allowed_ips: Optional[List[str]],
|
||||
request: Request,
|
||||
use_x_forwarded_for: Optional[bool] = False,
|
||||
) -> Tuple[bool, Optional[str]]:
|
||||
"""
|
||||
Returns if ip is allowed or not
|
||||
"""
|
||||
if allowed_ips is None: # if not set, assume true
|
||||
return True, None
|
||||
|
||||
# if general_settings.get("use_x_forwarded_for") is True then use x-forwarded-for
|
||||
client_ip = _get_request_ip_address(
|
||||
request=request, use_x_forwarded_for=use_x_forwarded_for
|
||||
)
|
||||
|
||||
# Check if IP address is allowed
|
||||
if client_ip not in allowed_ips:
|
||||
return False, client_ip
|
||||
|
||||
return True, client_ip
|
||||
|
||||
|
||||
def is_request_body_safe(request_body: dict) -> bool:
|
||||
"""
|
||||
Check if the request body is safe.
|
||||
|
||||
A malicious user can set the api_base to their own domain and invoke POST /chat/completions to intercept and steal the OpenAI API key.
|
||||
Relevant issue: https://huntr.com/bounties/4001e1a2-7b7a-4776-a3ae-e6692ec3d997
|
||||
"""
|
||||
banned_params = ["api_base", "base_url"]
|
||||
|
||||
for param in banned_params:
|
||||
if param in request_body:
|
||||
raise ValueError(f"BadRequest: {param} is not allowed in request body")
|
||||
|
||||
return True
|
||||
|
||||
|
||||
async def pre_db_read_auth_checks(
|
||||
request: Request,
|
||||
request_data: dict,
|
||||
route: str,
|
||||
):
|
||||
"""
|
||||
1. Checks if request size is under max_request_size_mb (if set)
|
||||
2. Check if request body is safe (example user has not set api_base in request body)
|
||||
3. Check if IP address is allowed (if set)
|
||||
4. Check if request route is an allowed route on the proxy (if set)
|
||||
|
||||
Returns:
|
||||
- True
|
||||
|
||||
Raises:
|
||||
- HTTPException if request fails initial auth checks
|
||||
"""
|
||||
from litellm.proxy.proxy_server import general_settings, premium_user
|
||||
|
||||
# Check 1. request size
|
||||
await check_if_request_size_is_safe(request=request)
|
||||
|
||||
# Check 2. Request body is safe
|
||||
is_request_body_safe(request_body=request_data)
|
||||
|
||||
# Check 3. Check if IP address is allowed
|
||||
is_valid_ip, passed_in_ip = _check_valid_ip(
|
||||
allowed_ips=general_settings.get("allowed_ips", None),
|
||||
use_x_forwarded_for=general_settings.get("use_x_forwarded_for", False),
|
||||
request=request,
|
||||
)
|
||||
|
||||
if not is_valid_ip:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"Access forbidden: IP address {passed_in_ip} not allowed.",
|
||||
)
|
||||
|
||||
# Check 4. Check if request route is an allowed route on the proxy
|
||||
if "allowed_routes" in general_settings:
|
||||
_allowed_routes = general_settings["allowed_routes"]
|
||||
if premium_user is not True:
|
||||
verbose_proxy_logger.error(
|
||||
f"Trying to set allowed_routes. This is an Enterprise feature. {CommonProxyErrors.not_premium_user.value}"
|
||||
)
|
||||
if route not in _allowed_routes:
|
||||
verbose_proxy_logger.error(
|
||||
f"Route {route} not in allowed_routes={_allowed_routes}"
|
||||
)
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"Access forbidden: Route {route} not allowed",
|
||||
)
|
||||
|
||||
|
||||
def route_in_additonal_public_routes(current_route: str):
|
||||
"""
|
||||
Helper to check if the user defined public_routes on config.yaml
|
||||
|
@ -50,48 +160,6 @@ def route_in_additonal_public_routes(current_route: str):
|
|||
return False
|
||||
|
||||
|
||||
def is_llm_api_route(route: str) -> bool:
|
||||
"""
|
||||
Helper to checks if provided route is an OpenAI route
|
||||
|
||||
|
||||
Returns:
|
||||
- True: if route is an OpenAI route
|
||||
- False: if route is not an OpenAI route
|
||||
"""
|
||||
|
||||
if route in LiteLLMRoutes.openai_routes.value:
|
||||
return True
|
||||
|
||||
if route in LiteLLMRoutes.anthropic_routes.value:
|
||||
return True
|
||||
|
||||
# fuzzy match routes like "/v1/threads/thread_49EIN5QF32s4mH20M7GFKdlZ"
|
||||
# Check for routes with placeholders
|
||||
for openai_route in LiteLLMRoutes.openai_routes.value:
|
||||
# Replace placeholders with regex pattern
|
||||
# placeholders are written as "/threads/{thread_id}"
|
||||
if "{" in openai_route:
|
||||
pattern = re.sub(r"\{[^}]+\}", r"[^/]+", openai_route)
|
||||
# Anchor the pattern to match the entire string
|
||||
pattern = f"^{pattern}$"
|
||||
if re.match(pattern, route):
|
||||
return True
|
||||
|
||||
# Pass through Bedrock, VertexAI, and Cohere Routes
|
||||
if "/bedrock/" in route:
|
||||
return True
|
||||
if "/vertex-ai/" in route:
|
||||
return True
|
||||
if "/gemini/" in route:
|
||||
return True
|
||||
if "/cohere/" in route:
|
||||
return True
|
||||
if "/langfuse/" in route:
|
||||
return True
|
||||
return False
|
||||
|
||||
|
||||
def get_request_route(request: Request) -> str:
|
||||
"""
|
||||
Helper to get the route from the request
|
||||
|
@ -259,3 +327,21 @@ def should_run_auth_on_pass_through_provider_route(route: str) -> bool:
|
|||
"""
|
||||
# by default we do not run virtual key auth checks on /vertex-ai/{endpoint} routes
|
||||
return False
|
||||
|
||||
|
||||
def _has_user_setup_sso():
|
||||
"""
|
||||
Check if the user has set up single sign-on (SSO) by verifying the presence of Microsoft client ID, Google client ID, and UI username environment variables.
|
||||
Returns a boolean indicating whether SSO has been set up.
|
||||
"""
|
||||
microsoft_client_id = os.getenv("MICROSOFT_CLIENT_ID", None)
|
||||
google_client_id = os.getenv("GOOGLE_CLIENT_ID", None)
|
||||
ui_username = os.getenv("UI_USERNAME", None)
|
||||
|
||||
sso_setup = (
|
||||
(microsoft_client_id is not None)
|
||||
or (google_client_id is not None)
|
||||
or (ui_username is not None)
|
||||
)
|
||||
|
||||
return sso_setup
|
||||
|
|
180
litellm/proxy/auth/route_checks.py
Normal file
180
litellm/proxy/auth/route_checks.py
Normal file
|
@ -0,0 +1,180 @@
|
|||
import re
|
||||
from typing import Optional
|
||||
|
||||
from fastapi import HTTPException, Request, status
|
||||
|
||||
from litellm._logging import verbose_proxy_logger
|
||||
from litellm.proxy._types import (
|
||||
CommonProxyErrors,
|
||||
LiteLLM_UserTable,
|
||||
LiteLLMRoutes,
|
||||
LitellmUserRoles,
|
||||
UserAPIKeyAuth,
|
||||
)
|
||||
from litellm.proxy.utils import hash_token
|
||||
|
||||
from .auth_utils import _has_user_setup_sso
|
||||
|
||||
|
||||
def non_admin_allowed_routes_check(
|
||||
user_obj: Optional[LiteLLM_UserTable],
|
||||
_user_role: Optional[LitellmUserRoles],
|
||||
route: str,
|
||||
request: Request,
|
||||
valid_token: UserAPIKeyAuth,
|
||||
api_key: str,
|
||||
request_data: dict,
|
||||
):
|
||||
"""
|
||||
Checks if Non-Admin User is allowed to access the route
|
||||
"""
|
||||
|
||||
# Check user has defined custom admin routes
|
||||
custom_admin_only_route_check(
|
||||
route=route,
|
||||
)
|
||||
|
||||
if is_llm_api_route(route=route):
|
||||
pass
|
||||
elif (
|
||||
route in LiteLLMRoutes.info_routes.value
|
||||
): # check if user allowed to call an info route
|
||||
if route == "/key/info":
|
||||
# check if user can access this route
|
||||
query_params = request.query_params
|
||||
key = query_params.get("key")
|
||||
if key is not None and hash_token(token=key) != api_key:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail="user not allowed to access this key's info",
|
||||
)
|
||||
elif route == "/user/info":
|
||||
# check if user can access this route
|
||||
query_params = request.query_params
|
||||
user_id = query_params.get("user_id")
|
||||
verbose_proxy_logger.debug(
|
||||
f"user_id: {user_id} & valid_token.user_id: {valid_token.user_id}"
|
||||
)
|
||||
if user_id != valid_token.user_id:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail="key not allowed to access this user's info. user_id={}, key's user_id={}".format(
|
||||
user_id, valid_token.user_id
|
||||
),
|
||||
)
|
||||
elif route == "/model/info":
|
||||
# /model/info just shows models user has access to
|
||||
pass
|
||||
elif route == "/team/info":
|
||||
pass # handled by function itself
|
||||
elif _has_user_setup_sso() and route in LiteLLMRoutes.sso_only_routes.value:
|
||||
pass
|
||||
elif (
|
||||
route in LiteLLMRoutes.global_spend_tracking_routes.value
|
||||
and getattr(valid_token, "permissions", None) is not None
|
||||
and "get_spend_routes" in getattr(valid_token, "permissions", [])
|
||||
):
|
||||
|
||||
pass
|
||||
elif _user_role == LitellmUserRoles.PROXY_ADMIN_VIEW_ONLY.value:
|
||||
if is_llm_api_route(route=route):
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"user not allowed to access this OpenAI routes, role= {_user_role}",
|
||||
)
|
||||
if route in LiteLLMRoutes.management_routes.value:
|
||||
# the Admin Viewer is only allowed to call /user/update for their own user_id and can only update
|
||||
if route == "/user/update":
|
||||
|
||||
# Check the Request params are valid for PROXY_ADMIN_VIEW_ONLY
|
||||
if request_data is not None and isinstance(request_data, dict):
|
||||
_params_updated = request_data.keys()
|
||||
for param in _params_updated:
|
||||
if param not in ["user_email", "password"]:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"user not allowed to access this route, role= {_user_role}. Trying to access: {route} and updating invalid param: {param}. only user_email and password can be updated",
|
||||
)
|
||||
else:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"user not allowed to access this route, role= {_user_role}. Trying to access: {route}",
|
||||
)
|
||||
|
||||
elif (
|
||||
_user_role == LitellmUserRoles.INTERNAL_USER.value
|
||||
and route in LiteLLMRoutes.internal_user_routes.value
|
||||
):
|
||||
pass
|
||||
elif (
|
||||
route in LiteLLMRoutes.self_managed_routes.value
|
||||
): # routes that manage their own allowed/disallowed logic
|
||||
pass
|
||||
else:
|
||||
user_role = "unknown"
|
||||
user_id = "unknown"
|
||||
if user_obj is not None:
|
||||
user_role = user_obj.user_role or "unknown"
|
||||
user_id = user_obj.user_id or "unknown"
|
||||
raise Exception(
|
||||
f"Only proxy admin can be used to generate, delete, update info for new keys/users/teams. Route={route}. Your role={user_role}. Your user_id={user_id}"
|
||||
)
|
||||
|
||||
|
||||
def custom_admin_only_route_check(route: str):
|
||||
from litellm.proxy.proxy_server import general_settings, premium_user
|
||||
|
||||
if "admin_only_routes" in general_settings:
|
||||
if premium_user is not True:
|
||||
verbose_proxy_logger.error(
|
||||
f"Trying to use 'admin_only_routes' this is an Enterprise only feature. {CommonProxyErrors.not_premium_user.value}"
|
||||
)
|
||||
return
|
||||
if route in general_settings["admin_only_routes"]:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"user not allowed to access this route. Route={route} is an admin only route",
|
||||
)
|
||||
pass
|
||||
|
||||
|
||||
def is_llm_api_route(route: str) -> bool:
|
||||
"""
|
||||
Helper to checks if provided route is an OpenAI route
|
||||
|
||||
|
||||
Returns:
|
||||
- True: if route is an OpenAI route
|
||||
- False: if route is not an OpenAI route
|
||||
"""
|
||||
|
||||
if route in LiteLLMRoutes.openai_routes.value:
|
||||
return True
|
||||
|
||||
if route in LiteLLMRoutes.anthropic_routes.value:
|
||||
return True
|
||||
|
||||
# fuzzy match routes like "/v1/threads/thread_49EIN5QF32s4mH20M7GFKdlZ"
|
||||
# Check for routes with placeholders
|
||||
for openai_route in LiteLLMRoutes.openai_routes.value:
|
||||
# Replace placeholders with regex pattern
|
||||
# placeholders are written as "/threads/{thread_id}"
|
||||
if "{" in openai_route:
|
||||
pattern = re.sub(r"\{[^}]+\}", r"[^/]+", openai_route)
|
||||
# Anchor the pattern to match the entire string
|
||||
pattern = f"^{pattern}$"
|
||||
if re.match(pattern, route):
|
||||
return True
|
||||
|
||||
# Pass through Bedrock, VertexAI, and Cohere Routes
|
||||
if "/bedrock/" in route:
|
||||
return True
|
||||
if "/vertex-ai/" in route:
|
||||
return True
|
||||
if "/gemini/" in route:
|
||||
return True
|
||||
if "/cohere/" in route:
|
||||
return True
|
||||
if "/langfuse/" in route:
|
||||
return True
|
||||
return False
|
|
@ -54,19 +54,20 @@ from litellm.proxy.auth.auth_checks import (
|
|||
get_org_object,
|
||||
get_team_object,
|
||||
get_user_object,
|
||||
is_request_body_safe,
|
||||
log_to_opentelemetry,
|
||||
)
|
||||
from litellm.proxy.auth.auth_utils import (
|
||||
check_if_request_size_is_safe,
|
||||
_get_request_ip_address,
|
||||
_has_user_setup_sso,
|
||||
get_request_route,
|
||||
is_llm_api_route,
|
||||
is_pass_through_provider_route,
|
||||
pre_db_read_auth_checks,
|
||||
route_in_additonal_public_routes,
|
||||
should_run_auth_on_pass_through_provider_route,
|
||||
)
|
||||
from litellm.proxy.auth.oauth2_check import check_oauth2_token
|
||||
from litellm.proxy.auth.oauth2_proxy_hook import handle_oauth2_proxy_request
|
||||
from litellm.proxy.auth.route_checks import non_admin_allowed_routes_check
|
||||
from litellm.proxy.common_utils.http_parsing_utils import _read_request_body
|
||||
from litellm.proxy.utils import _to_ns
|
||||
|
||||
|
@ -128,25 +129,11 @@ async def user_api_key_auth(
|
|||
route: str = get_request_route(request=request)
|
||||
# get the request body
|
||||
request_data = await _read_request_body(request=request)
|
||||
is_request_body_safe(request_body=request_data)
|
||||
|
||||
### LiteLLM Enterprise Security Checks
|
||||
# Check 1. Check if request size is under max_request_size_mb
|
||||
# Check 2. FILTER IP ADDRESS
|
||||
await check_if_request_size_is_safe(request=request)
|
||||
|
||||
is_valid_ip, passed_in_ip = _check_valid_ip(
|
||||
allowed_ips=general_settings.get("allowed_ips", None),
|
||||
use_x_forwarded_for=general_settings.get("use_x_forwarded_for", False),
|
||||
await pre_db_read_auth_checks(
|
||||
request_data=request_data,
|
||||
request=request,
|
||||
route=route,
|
||||
)
|
||||
|
||||
if not is_valid_ip:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"Access forbidden: IP address {passed_in_ip} not allowed.",
|
||||
)
|
||||
|
||||
pass_through_endpoints: Optional[List[dict]] = general_settings.get(
|
||||
"pass_through_endpoints", None
|
||||
)
|
||||
|
@ -200,6 +187,7 @@ async def user_api_key_auth(
|
|||
```
|
||||
"""
|
||||
|
||||
######## Route Checks Before Reading DB / Cache for "token" ################
|
||||
if (
|
||||
route in LiteLLMRoutes.public_routes.value
|
||||
or route_in_additonal_public_routes(current_route=route)
|
||||
|
@ -211,6 +199,9 @@ async def user_api_key_auth(
|
|||
return UserAPIKeyAuth(
|
||||
user_role=LitellmUserRoles.INTERNAL_USER_VIEW_ONLY
|
||||
)
|
||||
|
||||
########## End of Route Checks Before Reading DB / Cache for "token" ########
|
||||
|
||||
if general_settings.get("enable_oauth2_auth", False) is True:
|
||||
# return UserAPIKeyAuth object
|
||||
# helper to check if the api_key is a valid oauth2 token
|
||||
|
@ -985,96 +976,15 @@ async def user_api_key_auth(
|
|||
_user_role = _get_user_role(user_obj=user_obj)
|
||||
|
||||
if not _is_user_proxy_admin(user_obj=user_obj): # if non-admin
|
||||
if is_llm_api_route(route=route):
|
||||
pass
|
||||
elif (
|
||||
route in LiteLLMRoutes.info_routes.value
|
||||
): # check if user allowed to call an info route
|
||||
if route == "/key/info":
|
||||
# check if user can access this route
|
||||
query_params = request.query_params
|
||||
key = query_params.get("key")
|
||||
if key is not None and hash_token(token=key) != api_key:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail="user not allowed to access this key's info",
|
||||
)
|
||||
elif route == "/user/info":
|
||||
# check if user can access this route
|
||||
query_params = request.query_params
|
||||
user_id = query_params.get("user_id")
|
||||
verbose_proxy_logger.debug(
|
||||
f"user_id: {user_id} & valid_token.user_id: {valid_token.user_id}"
|
||||
)
|
||||
if user_id != valid_token.user_id:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail="key not allowed to access this user's info. user_id={}, key's user_id={}".format(
|
||||
user_id, valid_token.user_id
|
||||
),
|
||||
)
|
||||
elif route == "/model/info":
|
||||
# /model/info just shows models user has access to
|
||||
pass
|
||||
elif route == "/team/info":
|
||||
pass # handled by function itself
|
||||
elif (
|
||||
_has_user_setup_sso()
|
||||
and route in LiteLLMRoutes.sso_only_routes.value
|
||||
):
|
||||
pass
|
||||
elif (
|
||||
route in LiteLLMRoutes.global_spend_tracking_routes.value
|
||||
and getattr(valid_token, "permissions", None) is not None
|
||||
and "get_spend_routes" in getattr(valid_token, "permissions", [])
|
||||
):
|
||||
|
||||
pass
|
||||
elif _user_role == LitellmUserRoles.PROXY_ADMIN_VIEW_ONLY.value:
|
||||
if is_llm_api_route(route=route):
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"user not allowed to access this OpenAI routes, role= {_user_role}",
|
||||
)
|
||||
if route in LiteLLMRoutes.management_routes.value:
|
||||
# the Admin Viewer is only allowed to call /user/update for their own user_id and can only update
|
||||
if route == "/user/update":
|
||||
|
||||
# Check the Request params are valid for PROXY_ADMIN_VIEW_ONLY
|
||||
if request_data is not None and isinstance(
|
||||
request_data, dict
|
||||
):
|
||||
_params_updated = request_data.keys()
|
||||
for param in _params_updated:
|
||||
if param not in ["user_email", "password"]:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"user not allowed to access this route, role= {_user_role}. Trying to access: {route} and updating invalid param: {param}. only user_email and password can be updated",
|
||||
)
|
||||
else:
|
||||
raise HTTPException(
|
||||
status_code=status.HTTP_403_FORBIDDEN,
|
||||
detail=f"user not allowed to access this route, role= {_user_role}. Trying to access: {route}",
|
||||
)
|
||||
|
||||
elif (
|
||||
_user_role == LitellmUserRoles.INTERNAL_USER.value
|
||||
and route in LiteLLMRoutes.internal_user_routes.value
|
||||
):
|
||||
pass
|
||||
elif (
|
||||
route in LiteLLMRoutes.self_managed_routes.value
|
||||
): # routes that manage their own allowed/disallowed logic
|
||||
pass
|
||||
else:
|
||||
user_role = "unknown"
|
||||
user_id = "unknown"
|
||||
if user_obj is not None:
|
||||
user_role = user_obj.user_role or "unknown"
|
||||
user_id = user_obj.user_id or "unknown"
|
||||
raise Exception(
|
||||
f"Only proxy admin can be used to generate, delete, update info for new keys/users/teams. Route={route}. Your role={user_role}. Your user_id={user_id}"
|
||||
)
|
||||
non_admin_allowed_routes_check(
|
||||
user_obj=user_obj,
|
||||
_user_role=_user_role,
|
||||
route=route,
|
||||
request=request,
|
||||
request_data=request_data,
|
||||
api_key=api_key,
|
||||
valid_token=valid_token,
|
||||
)
|
||||
|
||||
# check if token is from litellm-ui, litellm ui makes keys to allow users to login with sso. These keys can only be used for LiteLLM UI functions
|
||||
# sso/login, ui/login, /key functions and /user functions
|
||||
|
@ -1228,24 +1138,6 @@ def _return_user_api_key_auth_obj(
|
|||
)
|
||||
|
||||
|
||||
def _has_user_setup_sso():
|
||||
"""
|
||||
Check if the user has set up single sign-on (SSO) by verifying the presence of Microsoft client ID, Google client ID, and UI username environment variables.
|
||||
Returns a boolean indicating whether SSO has been set up.
|
||||
"""
|
||||
microsoft_client_id = os.getenv("MICROSOFT_CLIENT_ID", None)
|
||||
google_client_id = os.getenv("GOOGLE_CLIENT_ID", None)
|
||||
ui_username = os.getenv("UI_USERNAME", None)
|
||||
|
||||
sso_setup = (
|
||||
(microsoft_client_id is not None)
|
||||
or (google_client_id is not None)
|
||||
or (ui_username is not None)
|
||||
)
|
||||
|
||||
return sso_setup
|
||||
|
||||
|
||||
def _is_user_proxy_admin(user_obj: Optional[LiteLLM_UserTable]):
|
||||
if user_obj is None:
|
||||
return False
|
||||
|
@ -1282,44 +1174,6 @@ def _get_user_role(
|
|||
return role
|
||||
|
||||
|
||||
def _get_request_ip_address(
|
||||
request: Request, use_x_forwarded_for: Optional[bool] = False
|
||||
) -> Optional[str]:
|
||||
|
||||
client_ip = None
|
||||
if use_x_forwarded_for is True and "x-forwarded-for" in request.headers:
|
||||
client_ip = request.headers["x-forwarded-for"]
|
||||
elif request.client is not None:
|
||||
client_ip = request.client.host
|
||||
else:
|
||||
client_ip = ""
|
||||
|
||||
return client_ip
|
||||
|
||||
|
||||
def _check_valid_ip(
|
||||
allowed_ips: Optional[List[str]],
|
||||
request: Request,
|
||||
use_x_forwarded_for: Optional[bool] = False,
|
||||
) -> Tuple[bool, Optional[str]]:
|
||||
"""
|
||||
Returns if ip is allowed or not
|
||||
"""
|
||||
if allowed_ips is None: # if not set, assume true
|
||||
return True, None
|
||||
|
||||
# if general_settings.get("use_x_forwarded_for") is True then use x-forwarded-for
|
||||
client_ip = _get_request_ip_address(
|
||||
request=request, use_x_forwarded_for=use_x_forwarded_for
|
||||
)
|
||||
|
||||
# Check if IP address is allowed
|
||||
if client_ip not in allowed_ips:
|
||||
return False, client_ip
|
||||
|
||||
return True, client_ip
|
||||
|
||||
|
||||
def get_api_key_from_custom_header(
|
||||
request: Request, custom_litellm_key_header_name: str
|
||||
):
|
||||
|
|
|
@ -15,6 +15,6 @@ litellm_settings:
|
|||
success_callback: ["prometheus"]
|
||||
failure_callback: ["prometheus"]
|
||||
|
||||
general_settings:
|
||||
master_key: sk-1234
|
||||
key_management_system: "google_secret_manager"
|
||||
general_settings:
|
||||
master_key: sk-1234
|
||||
|
||||
|
|
|
@ -3092,3 +3092,71 @@ async def test_team_tags(prisma_client):
|
|||
|
||||
print("team_info_response", team_info_response)
|
||||
assert team_info_response["team_info"].metadata["tags"] == ["teamA", "teamB"]
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_admin_only_routes(prisma_client):
|
||||
"""
|
||||
Tests if setting admin_only_routes works
|
||||
|
||||
only an admin should be able to access admin only routes
|
||||
"""
|
||||
litellm.set_verbose = True
|
||||
setattr(litellm.proxy.proxy_server, "prisma_client", prisma_client)
|
||||
setattr(litellm.proxy.proxy_server, "master_key", "sk-1234")
|
||||
await litellm.proxy.proxy_server.prisma_client.connect()
|
||||
general_settings = {
|
||||
"allowed_routes": ["/embeddings", "/key/generate"],
|
||||
"admin_only_routes": ["/key/generate"],
|
||||
}
|
||||
from litellm.proxy import proxy_server
|
||||
|
||||
initial_general_settings = getattr(proxy_server, "general_settings")
|
||||
|
||||
setattr(proxy_server, "general_settings", general_settings)
|
||||
|
||||
admin_user = await new_user(
|
||||
data=NewUserRequest(
|
||||
user_name="admin",
|
||||
user_role=LitellmUserRoles.PROXY_ADMIN,
|
||||
),
|
||||
user_api_key_dict=UserAPIKeyAuth(user_role=LitellmUserRoles.PROXY_ADMIN),
|
||||
)
|
||||
|
||||
non_admin_user = await new_user(
|
||||
data=NewUserRequest(
|
||||
user_name="non-admin",
|
||||
user_role=LitellmUserRoles.INTERNAL_USER,
|
||||
),
|
||||
user_api_key_dict=UserAPIKeyAuth(user_role=LitellmUserRoles.PROXY_ADMIN),
|
||||
)
|
||||
|
||||
admin_user_key = admin_user.key
|
||||
non_admin_user_key = non_admin_user.key
|
||||
|
||||
assert admin_user_key is not None
|
||||
assert non_admin_user_key is not None
|
||||
|
||||
# assert non-admin can not access admin routes
|
||||
request = Request(scope={"type": "http"})
|
||||
request._url = URL(url="/key/generate")
|
||||
await user_api_key_auth(
|
||||
request=request,
|
||||
api_key="Bearer " + admin_user_key,
|
||||
)
|
||||
|
||||
# this should pass
|
||||
|
||||
try:
|
||||
await user_api_key_auth(
|
||||
request=request,
|
||||
api_key="Bearer " + non_admin_user_key,
|
||||
)
|
||||
pytest.fail("Expected this call to fail. User is over limit.")
|
||||
except Exception as e:
|
||||
print("error str=", str(e.message))
|
||||
error_str = str(e.message)
|
||||
assert "Route" in error_str and "admin only route" in error_str
|
||||
pass
|
||||
|
||||
setattr(proxy_server, "general_settings", initial_general_settings)
|
||||
|
|
|
@ -21,7 +21,8 @@ from starlette.datastructures import URL, Headers, QueryParams
|
|||
|
||||
import litellm
|
||||
from litellm.proxy._types import LiteLLMRoutes
|
||||
from litellm.proxy.auth.auth_utils import get_request_route, is_llm_api_route
|
||||
from litellm.proxy.auth.auth_utils import get_request_route
|
||||
from litellm.proxy.auth.route_checks import is_llm_api_route
|
||||
from litellm.proxy.proxy_server import app
|
||||
|
||||
# Configure logging
|
||||
|
|
|
@ -42,7 +42,7 @@ class Request:
|
|||
def test_check_valid_ip(
|
||||
allowed_ips: Optional[List[str]], client_ip: Optional[str], expected_result: bool
|
||||
):
|
||||
from litellm.proxy.auth.user_api_key_auth import _check_valid_ip
|
||||
from litellm.proxy.auth.auth_utils import _check_valid_ip
|
||||
|
||||
request = Request(client_ip)
|
||||
|
||||
|
@ -70,7 +70,7 @@ def test_check_valid_ip(
|
|||
def test_check_valid_ip_sent_with_x_forwarded_for(
|
||||
allowed_ips: Optional[List[str]], client_ip: Optional[str], expected_result: bool
|
||||
):
|
||||
from litellm.proxy.auth.user_api_key_auth import _check_valid_ip
|
||||
from litellm.proxy.auth.auth_utils import _check_valid_ip
|
||||
|
||||
request = Request(client_ip, headers={"X-Forwarded-For": client_ip})
|
||||
|
||||
|
@ -246,3 +246,46 @@ async def test_user_api_key_auth_fails_with_prohibited_params(prohibited_param):
|
|||
error_message = str(e.message)
|
||||
print("error message=", error_message)
|
||||
assert "is not allowed in request body" in error_message
|
||||
|
||||
|
||||
@pytest.mark.asyncio()
|
||||
@pytest.mark.parametrize(
|
||||
"route, should_raise_error",
|
||||
[
|
||||
("/embeddings", False),
|
||||
("/chat/completions", True),
|
||||
("/completions", True),
|
||||
("/models", True),
|
||||
("/v1/embeddings", True),
|
||||
],
|
||||
)
|
||||
async def test_auth_with_allowed_routes(route, should_raise_error):
|
||||
# Setup
|
||||
user_key = "sk-1234"
|
||||
|
||||
general_settings = {"allowed_routes": ["/embeddings"]}
|
||||
from fastapi import Request
|
||||
|
||||
from litellm.proxy import proxy_server
|
||||
|
||||
initial_general_settings = getattr(proxy_server, "general_settings")
|
||||
|
||||
setattr(proxy_server, "master_key", "sk-1234")
|
||||
setattr(proxy_server, "general_settings", general_settings)
|
||||
|
||||
request = Request(scope={"type": "http"})
|
||||
request._url = URL(url=route)
|
||||
|
||||
if should_raise_error:
|
||||
try:
|
||||
await user_api_key_auth(request=request, api_key="Bearer " + user_key)
|
||||
pytest.fail("Expected this call to fail. User is over limit.")
|
||||
except Exception as e:
|
||||
print("error str=", str(e.message))
|
||||
error_str = str(e.message)
|
||||
assert "Route" in error_str and "not allowed" in error_str
|
||||
pass
|
||||
else:
|
||||
await user_api_key_auth(request=request, api_key="Bearer " + user_key)
|
||||
|
||||
setattr(proxy_server, "general_settings", initial_general_settings)
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue