Merge branch 'main' of https://github.com/Luna-OSS/nova-api

Refactor file operations

README.md

@@ -167,7 +167,7 @@ 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.
-`TEST_NOVA_KEY` is the API key the which is used in tests. It should be one with tons of credits.
+`NOVA_KEY` is the API key the which is used in tests. It should be one with tons of credits.
 
 ### Webhooks
 `DISCORD_WEBHOOK__USER_CREATED` is the Discord webhook URL for when a user is created.

api/backup_manager/main.py

@@ -1,6 +1,8 @@
 import os
 import json
 import asyncio
+import aiofiles
+import aiofiles.os
 
 from sys import argv
 from bson import json_util
@@ -18,8 +20,7 @@ async def main(output_dir: str):
 async def make_backup(output_dir: str):
     output_dir = os.path.join(FILE_DIR, '..', 'backups', output_dir)
 
-    if not os.path.exists(output_dir):
-        os.makedirs(output_dir)
+    await aiofiles.os.makedirs(output_dir, exist_ok=True)
 
     client = AsyncIOMotorClient(MONGO_URI)
     databases = await client.list_database_names()
@@ -29,22 +30,22 @@ async def make_backup(output_dir: str):
         if database == 'local':
             continue
 
-        if not os.path.exists(f'{output_dir}/{database}'):
-            os.mkdir(f'{output_dir}/{database}')
+        await aiofiles.os.makedirs(os.path.join(output_dir, database), exist_ok=True)
 
         for collection in databases[database]:
             print(f'Initiated database backup for {database}/{collection}')
             await make_backup_for_collection(database, collection, output_dir)
 
 async def make_backup_for_collection(database, collection, output_dir):
-    path = f'{output_dir}/{database}/{collection}.json'
+    path = os.path.join(output_dir, database, f'{collection}.json')
 
     client = AsyncIOMotorClient(MONGO_URI)
     collection = client[database][collection]
     documents = await collection.find({}).to_list(length=None)
 
-    with open(path, 'w') as f:
-        json.dump(documents, f, default=json_util.default)
+    async with aiofiles.open(path, 'w') as f:
+        for chunk in json.JSONEncoder(default=json_util.default).iterencode(documents):
+            await f.write(chunk)
 
 if __name__ == '__main__':
     if len(argv) < 2 or len(argv) > 2:

api/core.py

@@ -13,6 +13,7 @@ import json
 import hmac
 import httpx
 import fastapi
+import aiofiles
 import functools
 
 from dhooks import Webhook, Embed
@@ -130,7 +131,7 @@ async def run_checks(incoming_request: fastapi.Request):
         checks.client.test_chat_non_stream_gpt4,
         checks.client.test_chat_stream_gpt3,
         checks.client.test_function_calling,
-        checks.client.test_image_generation,
+        # checks.client.test_image_generation,
         # checks.client.test_speech_to_text,
         checks.client.test_models
     ]
@@ -148,11 +149,14 @@ async def run_checks(incoming_request: fastapi.Request):
 async def get_crypto_price(cryptocurrency: str) -> float:
     """Gets the price of a cryptocurrency using coinbase's API."""
 
-    if os.path.exists('cache/crypto_prices.json'):
-        with open('cache/crypto_prices.json', 'r') as f:
-            cache = json.load(f)
-    else:
+    cache_path = os.path.join('cache', 'crypto_prices.json')
+    try:
+        async with aiofiles.open(cache_path) as f:
+            content = await f.read()
+    except FileNotFoundError:
         cache = {}
+    else:
+        cache = json.loads(content)
 
     is_old = time.time() - cache.get('_last_updated', 0) > 60 * 60
 
@@ -164,8 +168,9 @@ async def get_crypto_price(cryptocurrency: str) -> float:
             cache[cryptocurrency] = usd_price
             cache['_last_updated'] = time.time()
 
-            with open('cache/crypto_prices.json', 'w') as f:
-                json.dump(cache, f)
+            async with aiofiles.open(cache_path, 'w') as f:
+                for chunk in json.JSONEncoder().iterencode(cache):
+                    await f.write(chunk)
 
     return cache[cryptocurrency]
 

api/db/providerkeys.py

@@ -3,6 +3,8 @@ import time
 import random
 import asyncio
 
+import aiofiles
+import aiofiles.os
 from aiocache import cached
 from dotenv import load_dotenv
 from cachetools import TTLCache
@@ -72,10 +74,10 @@ class KeyManager:
         db = await self._get_collection('providerkeys')
         num = 0
 
-        for filename in os.listdir('api/secret'):
+        for filename in await aiofiles.os.listdir(os.path.join('api', 'secret')):
             if filename.endswith('.txt'):
-                with open(f'api/secret/{filename}') as f:
-                    for line in f.readlines():
+                async with aiofiles.open(os.path.join('api', 'secret', filename)) as f:
+                    async for line in f:
                         if not line.strip():
                             continue
 

api/db/users.py

@@ -14,7 +14,7 @@ except ImportError:
 
 load_dotenv()
 
-with open(helpers.root + '/api/config/config.yml', encoding='utf8') as f:
+with open(os.path.join(helpers.root, 'api', 'config', 'config.yml'), encoding='utf8') as f:
     credits_config = yaml.safe_load(f)
 
 ## MONGODB Setup

api/handler.py

@@ -19,10 +19,11 @@ from helpers import tokens, errors, network
 load_dotenv()
 
 users = UserManager()
-models_list = json.load(open('cache/models.json', encoding='utf8'))
+with open(os.path.join('cache', 'models.json'), encoding='utf8') as f:
+    models_list = json.load(f)
 models = [model['id'] for model in models_list['data']]
 
-with open('config/config.yml', encoding='utf8') as f:
+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')
@@ -38,7 +39,10 @@ async def handle(incoming_request: fastapi.Request):
 
     ip_address = await network.get_ip(incoming_request)
 
-    if '/models' in path:
+    if '/dashboard' in path:
+        return errors.error(404, 'You can\'t access /dashboard.', 'This is a private endpoint.')
+
+    if path.startswith('/v1/models'):
         return fastapi.responses.JSONResponse(content=models_list)
 
     try:
@@ -94,7 +98,6 @@ async def handle(incoming_request: fastapi.Request):
     if user['credits'] < cost:
         return await errors.error(429, 'Not enough credits.', 'Wait or earn more credits. Learn more on our website or Discord server.')
 
-
     if 'DISABLE_VARS' not in key_tags:
         payload_with_vars = json.dumps(payload)
 

api/main.py

@@ -7,10 +7,10 @@ from rich import print
 from dotenv import load_dotenv
 
 from bson.objectid import ObjectId
+from fastapi.middleware.cors import CORSMiddleware
 
 from slowapi.errors import RateLimitExceeded
 from slowapi.middleware import SlowAPIMiddleware
-from fastapi.middleware.cors import CORSMiddleware
 from slowapi.util import get_remote_address
 from slowapi import Limiter, _rate_limit_exceeded_handler
 

api/providers/__init__.py

@@ -1,12 +1,2 @@
-from . import \
-    azure \
-    # closed, \
-    # closed4
-    # closed432
-
-MODULES = [
-    azure,
-    # closed,
-    # closed4,
-    # closed432,
-]
+from . import azure, webraft
+MODULES =    [azure, webraft]

api/providers/__main__.py

@@ -3,14 +3,13 @@ import sys
 import aiohttp
 import asyncio
 import importlib
+import aiofiles.os
 
 from rich import print
 
 def remove_duplicate_keys(file):
     with open(file, 'r', encoding='utf8') as f:
-        lines = f.readlines()
-
-    unique_lines = set(lines)
+        unique_lines = set(f)
 
     with open(file, 'w', encoding='utf8') as f:
         f.writelines(unique_lines)
@@ -22,7 +21,7 @@ async def main():
     except IndexError:
         print('List of available providers:')
 
-        for file_name in os.listdir(os.path.dirname(__file__)):
+        for file_name in await aiofiles.os.listdir(os.path.dirname(__file__)):
             if file_name.endswith('.py') and not file_name.startswith('_'):
                 print(file_name.split('.')[0])
 

api/providers/azure.py

@@ -2,7 +2,6 @@ from .helpers import utils
 
 AUTH = True
 ORGANIC = False
-CONTEXT = True
 STREAMING = True
 MODERATIONS = False
 ENDPOINT = 'https://nova-00001.openai.azure.com'
@@ -12,7 +11,7 @@ MODELS = [
     'gpt-4',
     'gpt-4-32k'
 ]
-# MODELS = [f'{model}-azure' for model in MODELS]
+MODELS += [f'{model}-azure' for model in MODELS]
 
 AZURE_API = '2023-08-01-preview'
 

api/providers/closed.py

@@ -2,7 +2,6 @@ from .helpers import utils
 
 AUTH = True
 ORGANIC = True
-CONTEXT = True
 STREAMING = True
 MODERATIONS = True
 ENDPOINT = 'https://api.openai.com'

api/providers/closed4.py

@@ -2,7 +2,6 @@ from .helpers import utils
 
 AUTH = True
 ORGANIC = False
-CONTEXT = True
 STREAMING = True
 MODERATIONS = True
 ENDPOINT = 'https://api.openai.com'

api/providers/closed432.py

@@ -2,7 +2,6 @@ from .helpers import utils
 
 AUTH = True
 ORGANIC = False
-CONTEXT = True
 STREAMING = True
 MODERATIONS = False
 ENDPOINT = 'https://api.openai.com'

api/providers/helpers/utils.py

@@ -23,15 +23,5 @@ GPT_4_32K = GPT_4 + [
     'gpt-4-32k-0613',
 ]
 
-async def conversation_to_prompt(conversation: list) -> str:
-    text = ''
-
-    for message in conversation:
-        text += f'<|{message["role"]}|>: {message["content"]}\n'
-
-    text += '<|assistant|>:'
-
-    return text
-
 async def random_secret_for(name: str) -> str:
     return await providerkeys.manager.get_key(name)

api/providers/mandrill.py

@@ -2,7 +2,6 @@ from .helpers import utils
 
 AUTH = True
 ORGANIC = False
-CONTEXT = True
 STREAMING = True
 MODELS = ['llama-2-7b-chat']
 
@@ -12,7 +11,7 @@ async def chat_completion(**kwargs):
 
     return {
         'method': 'POST',
-        'url': f'https://api.mandrillai.tech/v1/chat/completions',
+        'url': 'https://api.mandrillai.tech/v1/chat/completions',
         'payload': payload,
         'headers': {
             'Authorization': f'Bearer {key}'

api/providers/webraft.py

@@ -0,0 +1,25 @@
+from .helpers import utils
+
+AUTH = True
+ORGANIC = False
+STREAMING = True
+MODELS = [
+    'gpt-3.5-turbo-0613',
+    'gpt-3.5-turbo-0301',
+    'gpt-3.5-turbo-16k-0613'
+]
+
+async def chat_completion(**kwargs):
+    payload = kwargs
+    key = await utils.random_secret_for('webraft')
+
+    return {
+        'method': 'POST',
+        'url': 'https://thirdparty.webraft.in/v1/chat/completions',
+        'payload': payload,
+        'headers': {
+            'Content-Type': 'application/json',
+            'Authorization': f'Bearer {key}'
+        },
+        'provider_auth': f'webraft>{key}'
+    }

api/proxies.py

@@ -96,7 +96,7 @@ proxies_in_files = []
 
 for proxy_type in ['http', 'socks4', 'socks5']:
     try:
-        with open(f'secret/proxies/{proxy_type}.txt') as f:
+        with open(os.path.join('secret', 'proxies', f'{proxy_type}.txt')) as f:
             for line in f:
                 clean_line = line.split('#', 1)[0].strip()
                 if clean_line:

api/responder.py

@@ -63,7 +63,14 @@ async def respond(
         'Content-Type': 'application/json'
     }
 
-    for i in range(5):
+    skipped_errors = {
+        'insufficient_quota': 0,
+        'billing_not_active': 0,
+        'critical_provider_error': 0,
+        'timeout': 0
+    }
+
+    for _ in range(5):
         try:
             if is_chat:
                 target_request = await load_balancing.balance_chat_request(payload)
@@ -130,11 +137,13 @@ async def respond(
                         if error_code == 'insufficient_quota':
                             print('[!] insufficient quota')
                             await keymanager.rate_limit_key(provider_name, provider_key, 86400)
+                            skipped_errors['insufficient_quota'] += 1
                             continue
 
                         if error_code == 'billing_not_active':
                             print('[!] billing not active')
                             await keymanager.deactivate_key(provider_name, provider_key, 'billing_not_active')
+                            skipped_errors['billing_not_active'] += 1
                             continue
 
                         critical_error = False
@@ -142,25 +151,16 @@ async def respond(
                             if error in str(client_json_response):
                                 await keymanager.deactivate_key(provider_name, provider_key, error)
                                 critical_error = True
-                        
+
                         if critical_error:
-                            print('[!] critical error')
+                            print('[!] critical provider error')
+                            skipped_errors['critical_provider_error'] += 1
                             continue
 
                         if response.ok:
                             server_json_response = client_json_response
 
-                        else:
-                            continue
-
                     if is_stream:
-                        try:
-                            response.raise_for_status()
-                        except Exception as exc:
-                            if 'Too Many Requests' in str(exc):
-                                print('[!] too many requests')
-                                continue
-
                         chunk_no = 0
                         buffer = ''
 
@@ -170,7 +170,7 @@ async def respond(
                             chunk = chunk.decode('utf8')
 
                             if 'azure' in provider_name:
-                                chunk = chunk.replace('data: ', '')
+                                chunk = chunk.replace('data: ', '', 1)
 
                                 if not chunk or chunk_no == 1:
                                     continue
@@ -178,19 +178,26 @@ async def respond(
                             subchunks = chunk.split('\n\n')
                             buffer += subchunks[0]
 
-                            yield buffer + '\n\n'
-                            buffer = subchunks[-1]
+                            for subchunk in [buffer] + subchunks[1:-1]:
+                                if not subchunk.startswith('data: '):
+                                    subchunk = 'data: ' + subchunk
 
-                            for subchunk in subchunks[1:-1]:
                                 yield subchunk + '\n\n'
 
+                            buffer = subchunks[-1]
                     break
 
             except aiohttp.client_exceptions.ServerTimeoutError:
+                skipped_errors['timeout'] += 1
                 continue
 
     else:
-        yield await errors.yield_error(500, 'Sorry, our API seems to have issues connecting to our provider(s).', 'This most likely isn\'t your fault. Please try again later.')
+        skipped_errors = {k: v for k, v in skipped_errors.items() if v > 0}
+        skipped_errors = ujson.dumps(skipped_errors, indent=4)
+        yield await errors.yield_error(500,
+            'Sorry, our API seems to have issues connecting to our provider(s).',
+            f'Please send this info to support: {skipped_errors}'
+        )
         return
 
     if (not is_stream) and server_json_response:

checks/client.py

@@ -100,7 +100,7 @@ async def test_chat_stream_gpt3() -> float:
 
     async for chunk in response.aiter_text():
         for subchunk in chunk.split('\n\n'):
-            chunk = subchunk.replace('data: ', '').strip()
+            chunk = subchunk.replace('data: ', '', 1).strip()
 
             if chunk == '[DONE]':
                 break

requirements.txt

@@ -1,3 +1,4 @@
+aiofiles==23.2.1
 aiohttp==3.8.5
 aiohttp_socks==0.8.0
 dhooks==1.1.4

rewards/main.py

@@ -51,7 +51,7 @@ async def update_roles():
 def launch():
     asyncio.run(main())
 
-    with open('rewards/last_update.txt', 'w', encoding='utf8') as f:
+    with open(os.path.join('rewards', 'last_update.txt'), 'w', encoding='utf8') as f:
         f.write(str(time.time()))
 
 if __name__ == '__main__':

run/__main__.py

@@ -14,7 +14,6 @@ Runs for production on the speicified port.
 
 import os
 import sys
-import time
 
 port = sys.argv[1] if len(sys.argv) > 1 else 2332
 dev = True