Enterprise fix

README.md

@@ -65,24 +65,31 @@ This one's code can be found in the following repository: [github.com/novaoss/no
 
 # Setup
 ## Requirements
-- Python 3.9+
-- pip
-- MongoDB database
-- `uvicorn`
+- newest **Python** version
+- newest Python **pip** version
+- **MongoDB** database
+- `uvicorn` in your system package manager
 
 ## Recommended
 - Setup of the other infrastructure
 - `git` (for updates)
 - `screen` (for production)
 - Cloudflare (for security, anti-DDoS, etc.) - we fully support Cloudflare
+- proxies, in case you need to protect your privacy from authorities (China, Iran, ...)
 
 ## Staging System
 This repository has an integrated staging system. It's a simple system that allows you to test the API server before deploying it to production.
 
 You should definitely set up two databases on MongoDB: `nova-core` and `nova-test`. Please note that `nova-core` is always used for `providerkeys`.
 
-Put your production `.env` file in `env/.prod.env`. Your test `.env` file should be in `.env`.
+Put your production env in `env/.prod.env` and modify the values from the test `.env` to your liking:
+- Set `MONGO_NAME` to `nova-core`, which is your database name for the production mode.
+- Set `CHECKS_ENDPOINT` to `http://localhost:2333` (or the production port you set for `nova-api`)
 
+**Warning -** always make sure to update your production `.env` (`env/.prod.env`), too!
+
+
+Your test `.env` file should be placed in here.
 Running `PUSH_TO_PRODUCTION.sh` will:
 - kill port `2333` (production)
 - remove all contents of the production directory, set to `/home/nova-prod/` (feel free to change it)
@@ -130,6 +137,8 @@ Create a `.env` file, make sure not to reveal any of its contents to anyone, and
 ### Database
 Set up a MongoDB database and set `MONGO_URI` to the MongoDB database connection URI. Quotation marks are definetly recommended here!
 
+Then set `MONGO_NAME` to `nova-test`, which is your database name for the tests.
+
 ### Proxy (optional)
 - `PROXY_TYPE` (optional, defaults to `socks.PROXY_TYPE_HTTP`): the type of proxy - can be `http`, `https`, `socks4`, `socks5`, `4` or `5`, etc... 
 - `PROXY_HOST`: the proxy host (host domain or IP address), without port!
@@ -182,13 +191,17 @@ You can also just add the *beginning* of an API address, like `12.123.` (without
 
 ### Core Keys
 `CORE_API_KEY` specifies the **very secret key** for  which need to access the entire user database etc.
+
+### Checks
 `NOVA_KEY` is the API key the which is used in tests. It should be one with tons of credits.
+`CHECKS_ENDPOINT` is the endpoint 
 
 ### Webhooks
 `DISCORD_WEBHOOK__USER_CREATED` is the Discord webhook URL for when a user is created.
 `DISCORD_WEBHOOK__API_ISSUE` is the Discord webhook URL for when an API issue occurs.
 
 ### Other
+`MODERATION_DEBUG_KEY` can be almost any string (avoid spaces or special characters) - users can add `#` + this key to their API key (e.g. `Bearer nv-123#modkey` as the `Authorization` header) to bypass the moderation checks. This is especially useful if the moderation is too sensitive and can be disabled for certain trusted users.
 `KEYGEN_INFIX` can be almost any string (avoid spaces or special characters) - this string will be put in the middle of every NovaAI API key which is generated. This is useful for identifying the source of the key using e.g. RegEx.
 
 ## Misc

api/handler.py

@@ -25,7 +25,7 @@ models = [model['id'] for model in models_list['data']]
 with open(os.path.join('config', 'config.yml'), encoding='utf8') as f:
     config = yaml.safe_load(f)
 
-moderation_debug_key_key = os.getenv('MODERATION_DEBUG_KEY')
+moderation_debug_key = os.getenv('MODERATION_DEBUG_KEY')
 
 async def handle(incoming_request: fastapi.Request):
     """Transfer a streaming response 
@@ -71,10 +71,10 @@ async def handle(incoming_request: fastapi.Request):
     if ban_reason:
         return await errors.error(403, f'Your NovaAI account has been banned. Reason: \'{ban_reason}\'.', 'Contact the staff for an appeal.')
 
-    # Checking for enterprise status
-    enterprise_keys = os.environ.get('ENTERPRISE_KEYS')
-    if path.startswith('/enterprise/v1') and user.get('api_key') not in enterprise_keys.split():
-        return await errors.error(403, 'Enterprise API is not available.', 'Contact the staff for an upgrade.')
+    is_enterprise_key = 'enterprise' in user.get('role', 'default')
+
+    if path.startswith('/enterprise/v1') and not is_enterprise_key:
+        return await errors.error(403, 'Enterprise API is not available for your API key.', 'Contact the staff for an upgrade.')
 
     if 'account/credits' in path:
         return fastapi.responses.JSONResponse({'credits': user['credits']})
@@ -87,6 +87,9 @@ async def handle(incoming_request: fastapi.Request):
 
     role = user.get('role', 'default')
 
+    if 'enterprise' in role:
+        role_cost_multiplier = 0.1
+    else:
         try:
             role_cost_multiplier = config['roles'][role]['bonus']
         except KeyError:
@@ -124,7 +127,7 @@ async def handle(incoming_request: fastapi.Request):
 
     policy_violation = False
 
-    if not (moderation_debug_key_key and moderation_debug_key_key in key_tags and 'gpt-3' in payload.get('model', '')):
+    if not (moderation_debug_key and moderation_debug_key in key_tags and 'gpt-3' in payload.get('model', '')):
         if '/moderations' not in path:
             inp = ''
 

api/providers/__init__.py

@@ -1,2 +1,2 @@
-from . import azure
-MODULES =    [azure]
+from . import ails, closed, closed4
+MODULES =    [ails, closed, closed4]

api/providers/ails.py

@@ -3,8 +3,10 @@ from .helpers import utils
 ORGANIC = False
 MODELS = [
     'gpt-3.5-turbo',
-    'gpt-3.5-turbo-0301',
-    'gpt-3.5-turbo-16k-0613'
+    'gpt-3.5-turbo-0613',
+    'gpt-4',
+    'gpt-3.5-turbo-16k',
+    'gpt-4-0613'
 ]
 
 async def chat_completion(**kwargs):

api/providers/azure.py

@@ -1,13 +1,14 @@
 from .helpers import utils
 
 ORGANIC = False # If all OpenAI endpoints should be used for the provider. If false, only a chat completions are used for this provider.
-ENDPOINT = 'https://nova-00001.openai.azure.com' # (Important: read below) The endpoint for the provider. 
+ENDPOINT = 'https://nova-00003.openai.azure.com' # (Important: read below) The endpoint for the provider. 
 #! IMPORTANT: If this is an ORGANIC provider, this should be the endpoint for the API with anything BEFORE the "/v1".
 MODELS = [
     'gpt-3.5-turbo',
-    'gpt-3.5-turbo-16k',
-    'gpt-4',
-    'gpt-4-32k'
+    # 'gpt-3.5-turbo-16k',
+    # 'gpt-3.5-turbo-instruct'
+    # 'gpt-4',
+    # 'gpt-4-32k'
 ]
 
 async def chat_completion(**payload):

api/providers/helpers/utils.py

@@ -31,7 +31,8 @@ async def random_secret_for(name: str) -> str:
 async def azure_chat_completion(endpoint: str, provider: str, payload: dict) -> dict:
     key = await random_secret_for(provider)
     model = payload['model']
-    deployment = model.replace('.', '').replace('-azure', '')
+    del payload['model']
+    deployment = model.replace('.', '')
 
     return {
         'method': 'POST',

api/proxies.py

@@ -122,7 +122,7 @@ def get_proxy() -> Proxy:
 
     return Proxy(
         proxy_type=os.getenv('PROXY_TYPE', 'http'),
-        host_or_ip=os.getenv('PROXY_HOST', '127.0.0.1'),
+        host_or_ip=os.environ['PROXY_HOST'],
         port=int(os.getenv('PROXY_PORT', '8080')),
         username=os.getenv('PROXY_USER'),
         password=os.getenv('PROXY_PASS')

api/responder.py

@@ -71,7 +71,7 @@ async def respond(
         'timeout': 0
     }
 
-    for _ in range(5):
+    for _ in range(10):
         try:
             if is_chat:
                 target_request = await load_balancing.balance_chat_request(payload)
@@ -107,7 +107,12 @@ async def respond(
         if target_request['method'] == 'GET' and not payload:
             target_request['payload'] = None
 
-        async with aiohttp.ClientSession(connector=proxies.get_proxy().connector) as session:
+        connector = None
+
+        if os.getenv('PROXY_HOST') or os.getenv('USE_PROXY_LIST', 'False').lower() == 'true':
+            connector = proxies.get_proxy().connector
+
+        async with aiohttp.ClientSession(connector=connector) as session:
             try:
                 async with session.request(
                     method=target_request.get('method', 'POST'),

checks/client.py

@@ -72,7 +72,11 @@ async def test_chat_non_stream_gpt4() -> float:
         )
         await _response_base_check(response)
 
+    try:
         assert '1337' in response.json()['choices'][0]['message']['content'], 'The API did not return a correct response.'
+    except json.decoder.JSONDecodeError:
+        return response.status_code
+
     return time.perf_counter() - request_start
 
 async def test_chat_stream_gpt3() -> float:
@@ -216,8 +220,9 @@ async def demo():
             raise ConnectionError('API Server is not running.')
 
         for func in [
+            test_chat_stream_gpt3,
             test_chat_non_stream_gpt4,
-            test_chat_stream_gpt3
+            test_function_calling,
         ]:
             print(f'[*] {func.__name__}')
             result = await func()