Works for BetterGPT

Dockerfile

@@ -0,0 +1,17 @@
+# 
+FROM python:3.10
+
+# 
+WORKDIR /code
+
+# 
+COPY ./requirements.txt /code/requirements.txt
+
+# 
+RUN pip install . pip install --no-cache-dir --upgrade -r /code/requirements.txt
+
+# 
+COPY ./app /code/app
+
+# 
+CMD ["uvicorn", "api.main:app", "--host", "0.0.0.0", "--port", "2333"]

api/keys.py

@@ -1,104 +0,0 @@
-import json
-import os
-import random
-import time
-from pymongo import MongoClient
-
-with open('./config.json', 'r') as file:
-    config = json.load(file)
-
-class Keys:
-    # --- START OF CONFIG ---
-    MONGO_URI = os.getenv('MONGO_URI') or config.get('MONGO_URI')
-    # --- END OF CONFIG ---
-
-    locked_keys = set()
-    cache = {}
-
-    # Initialize MongoDB
-    client = MongoClient(MONGO_URI)
-    db = client.get_database('keys_db')
-    collection = db['keys']
-
-    def __init__(self, key: str, model: str, provider_name: str, ratelimit: int, url: str):
-        self.key = key
-        self.model = model
-        self.provider_name = provider_name
-        self.ratelimit = ratelimit
-        self.url = url
-        if not Keys.cache:
-            self._load_keys()
-
-    def _load_keys(self) -> None:
-        cursor = Keys.collection.find({}, {'_id': 0, 'key_value': 1, 'model': 1, 'provider_name': 1, 'ratelimit': 1, 'url': 1, 'last_used': 1})
-        for doc in cursor:
-            key_value = doc['key_value']
-            model = doc['model']
-            provider_name = doc['provider_name']
-            ratelimit = doc['ratelimit']
-            url = doc['url']
-            last_used = doc.get('last_used', 0)
-            key_data = {'provider_name': provider_name, 'ratelimit': ratelimit, 'url': url, 'last_used': last_used}
-            Keys.cache.setdefault(model, {}).setdefault(key_value, key_data)
-
-    def lock(self) -> None:
-        self.locked_keys.add(self.key)
-
-    def unlock(self) -> None:
-        self.locked_keys.remove(self.key)
-
-    def is_locked(self) -> bool:
-        return self.key in self.locked_keys
-
-    @staticmethod
-    def get(model: str) -> str:
-        key_candidates = list(Keys.cache.get(model, {}).keys())
-        random.shuffle(key_candidates)
-
-        current_time = time.time()
-
-        for key_candidate in key_candidates:
-            key_data = Keys.cache[model][key_candidate]
-            key = Keys(key_candidate, model, key_data['provider_name'], key_data['ratelimit'], key_data['url'])
-            time_since_last_used = current_time - key_data.get('last_used', 0)
-
-            if not key.is_locked() and time_since_last_used >= key.ratelimit:
-                key.lock()
-                key_data['last_used'] = current_time  # Update last_used in the cache
-                Keys.collection.update_one(
-                    {'key_value': key.key, 'model': key.model},
-                    {'$set': {'last_used': current_time}}  # Update last_used in the database
-                )
-                return {
-                    'url': key.url,
-                    'key_value': key.key
-                }
-
-        print(f"[WARN] No unlocked keys found for model '{model}' in get keys request!")
-
-    def delete(self) -> None:
-        Keys.collection.delete_one({'key_value': self.key, 'model': self.model})
-        # Update cache
-        try:
-            del Keys.cache[self.model][self.key]
-        except KeyError:
-            print(f"[WARN] Tried to remove a key from cache which was not present: {self.key}")
-
-    def save(self) -> None:
-        key_data = {
-            'provider_name': self.provider_name,
-            'ratelimit': self.ratelimit,
-            'url': self.url,
-            'last_used': 0  # Initialize last_used to 0 when saving a new key
-        }
-        Keys.collection.insert_one({'key_value': self.key, 'model': self.model, **key_data})
-        # Update cache
-        Keys.cache.setdefault(self.model, {}).setdefault(self.key, key_data)
-
-# Usage example:
-# os.environ['MONGO_URI'] = "mongodb://localhost:27017"
-# key_instance = Keys("example_key", "gpt-4", "openai", "10", "https://whatever-openai-thing-is/chat/completions/")
-# key_instance.save()
-# key_value = Keys.get("gpt-4")
-
-

api/main.py

@@ -1,8 +1,8 @@
-import os
+"""FastAPI setup."""
-import fastapi
+
+import fastapi
+import asyncio
 
-from starlette.responses import StreamingResponse
-from starlette.requests import Request
 from fastapi.middleware.cors import CORSMiddleware
 
 from dotenv import load_dotenv
@@ -25,9 +25,7 @@ app.add_middleware(
 @app.on_event('startup')
 async def startup_event():
     """Read up the API server."""
-
+    # await security.ip_protection_check()
-    # security.enable_proxy()
-    # security.ip_protection_check()
 
 @app.get('/')
 async def root():
@@ -35,8 +33,15 @@ async def root():
 
     return {
         'status': 'ok',
-        'discord': 'https://discord.gg/85gdcd57Xr',
+        'readme': 'https://nova-oss.com'
-        'github': 'https://github.com/Luna-OSS'
+    }
+
+@app.route('/v1')
+async def api_root():
+    """Returns the API root endpoint."""
+
+    return {
+        'status': 'ok',
     }
 
 app.add_route('/{path:path}', transfer.transfer_streaming_response, ['GET', 'POST', 'PUT', 'DELETE', 'PATCH'])

api/proxies.py

@@ -1,12 +1,9 @@
-"""This module contains the Proxy class, which represents a proxy."""
+"""This module makes it easy to implement proxies by providing a class.."""
 
 import os
-import httpx
 import socket
 import asyncio
 
-from sockslib import socks
-from rich import print
 from dotenv import load_dotenv
 
 load_dotenv()
@@ -28,43 +25,6 @@ class Proxy:
         self.username = username
         self.password = password
 
-    @property
-    def auth(self):
-        """Returns the authentication part of the proxy URL, if the proxy has a username and password."""
-        return f'{self.username}:{self.password}@' if all([self.username, self.password]) else ''
-
-    @property
-    def protocol(self):
-        """Makes sure the hostnames are resolved correctly using the proxy.
-        See https://stackoverflow.com/a/43266186
-        """
-        return self.proxy_type# + 'h' if self.proxy_type.startswith('socks') else self.proxy_type
-
-    @property
-    def proxies(self):
-        """Returns a dictionary of proxies, ready to be used with the requests library or httpx.
-        """
-
-        url = f'{self.protocol}://{self.auth}{self.host}:{self.port}'
-
-        proxies_dict = {
-            'http://': url.replace('https', 'http') if self.proxy_type == 'https' else url,
-            'https://': url.replace('http', 'https') if self.proxy_type == 'http' else url
-        }
-
-        return proxies_dict
-
-    @property
-    def url(self):
-        """Returns the proxy URL."""
-        return f'{self.protocol}://{self.auth}{self.host}:{self.port}'
-
-    def __str__(self):
-        return f'{self.proxy_type}://{len(self.auth) * "*"}{self.host}:{self.port}'
-
-    def __repr__(self):
-        return f'<Proxy type={self.proxy_type} host={self.host} port={self.port} username={self.username} password={len(self.password) * "*"}>'
-
 active_proxy = Proxy(
     proxy_type=os.getenv('PROXY_TYPE', 'http'),
     host=os.getenv('PROXY_HOST', '127.0.0.1'),
@@ -72,54 +32,3 @@ active_proxy = Proxy(
     username=os.getenv('PROXY_USER'),
     password=os.getenv('PROXY_PASS')
 )
-
-def activate_proxy() -> None:
-    socks.set_default_proxy(
-        proxy_type=socks.PROXY_TYPES[active_proxy.proxy_type.upper()],
-        addr=active_proxy.host,
-        port=active_proxy.port,
-        username=active_proxy.username,
-        password=active_proxy.password
-    )
-    socket.socket = socks.socksocket
-
-def check_proxy():
-    """Checks if the proxy is working."""
-
-    resp = httpx.get(
-        url='https://echo.hoppscotch.io/',
-        timeout=20,
-        proxies=active_proxy.proxies
-    )
-    resp.raise_for_status()
-
-    return resp.ok
-
-async def check_api():
-    model = 'gpt-3.5-turbo'
-    messages = [
-        {
-            'role': 'user',
-            'content': 'Explain what a wormhole is.'
-        },
-    ]
-
-    headers = {
-        'Content-Type': 'application/json',
-        'Authorization': 'Bearer '  + os.getenv('CLOSEDAI_KEY')
-    }
-
-    json_data = {
-        'model': model,
-        'messages': messages,
-        'stream': True
-    }
-
-    async with httpx.AsyncClient(timeout=20) as client:
-        async with client.stream("POST", 'https://api.openai.com/v1/chat/completions', headers=headers, json=json_data) as response:
-            response.raise_for_status()
-            async for chunk in response.aiter_text():
-                print(chunk.strip())
-
-if __name__ == '__main__':
-    asyncio.run(check_api())

api/security.py

@@ -1,47 +1,45 @@
 """Security checks for the API. Checks if the IP is masked etc."""
 
 import os
-import httpx
+import asyncio
 
 from rich import print
-
+from tests import check_proxy
-import proxies
-
 from dotenv import load_dotenv
 
 load_dotenv()
-is_proxy_enabled = False
+is_proxy_enabled = bool(os.getenv('PROXY_HOST', None))
 
 class InsecureIPError(Exception):
     """Raised when the IP address of the server is not secure."""
 
-def ip_protection_check():
+async def ip_protection_check():
     """Makes sure that the actual server IP address is not exposed to the public."""
 
+    if not is_proxy_enabled:
+        print('[yellow]WARN:     The proxy is not enabled. \
+Skipping IP protection check.[/yellow]')
+        return True
+
     actual_ips = os.getenv('ACTUAL_IPS', '').split()
 
     if actual_ips:
-        echo_response = httpx.get(
+        # run the async function check_proxy() and get its result
-            url='https://echo.hoppscotch.io/',
+        response_ip = await check_proxy()
-            timeout=15
-        )
-
-        response_data = echo_response.json()
-        response_ip = response_data['headers']['x-forwarded-for']
 
         for actual_ip in actual_ips:
-            if actual_ip in response_data:
+            if actual_ip in response_ip:
                 raise InsecureIPError(f'IP pattern "{actual_ip}" is in the values of ACTUAL_IPS of the\
 .env file. Enable a VPN or proxy to continue.')
 
         if is_proxy_enabled:
-            print(f'[green]SUCCESS: The IP "{response_ip}" was detected, which seems to be a proxy. Great![/green]')
+            print(f'[green]GOOD:     The IP "{response_ip}" was detected, which seems to be a proxy. Great![/green]')
+            return True
 
     else:
-        print('[yellow]WARNING: ACTUAL_IPS is not set in the .env file or empty.\
+        print('[yellow]WARN:     ACTUAL_IPS is not set in the .env file or empty.\
 This means that the real IP of the server could be exposed. If you\'re using something\
 like Cloudflare or Repl.it, you can ignore this warning.[/yellow]')
 
 if __name__ == '__main__':
-    enable_proxy()
     ip_protection_check()

api/transfer.py

@@ -1,6 +1,8 @@
 """Module for transferring requests to ClosedAI API"""
 
 import os
+import json
+import logging
 import aiohttp
 import aiohttp_socks
 
@@ -8,11 +10,21 @@ import proxies
 
 from dotenv import load_dotenv
 
+import starlette
 from starlette.responses import StreamingResponse
 from starlette.background import BackgroundTask
 
 load_dotenv()
 
+# log to "api.log" file
+logging.basicConfig(
+    filename='api.log',
+    level=logging.DEBUG,
+    format='%(asctime)s %(levelname)s %(name)s %(message)s'
+)
+
+logging.info('API started')
+
 EXCLUDED_HEADERS = [
     'content-encoding',
     'content-length',
@@ -33,7 +45,16 @@ proxy_connector = aiohttp_socks.ProxyConnector(
 async def transfer_streaming_response(incoming_request, target_endpoint: str='https://api.openai.com/v1'):
     """Transfer a streaming response from the incoming request to the target endpoint"""
 
+    if incoming_request.headers.get('Authorization') != f'Bearer {os.getenv("DEMO_AUTH")}':
+        return starlette.responses.Response(
+            status_code=403,
+            content='Invalid API key'
+        )
+
+    try:
         incoming_json_payload = await incoming_request.json()
+    except json.decoder.JSONDecodeError:
+        incoming_json_payload = {}
 
     async def receive_target_stream():
         connector = aiohttp_socks.ProxyConnector(
@@ -46,12 +67,15 @@ async def transfer_streaming_response(incoming_request, target_endpoint: str='ht
         )
         async with aiohttp.ClientSession(
             connector=connector,
-            timeout=aiohttp.ClientTimeout(total=120),
+            timeout=aiohttp.ClientTimeout(total=int(os.getenv('TRANSFER_TIMEOUT', '120'))),
             raise_for_status=True
         ) as session:
+            target_url = f'{target_endpoint}{incoming_request.url.path}'.replace('/v1/v1', '/v1')
+            logging.info('TRANSFER %s -> %s', incoming_request.url.path, target_url)
+
             async with session.request(
                 method=incoming_request.method,
-                url=f'{target_endpoint}/{incoming_request.url.path}',
+                url=target_url,
                 json=incoming_json_payload,
                 headers={
                     'Content-Type': 'application/json',
@@ -60,6 +84,7 @@ async def transfer_streaming_response(incoming_request, target_endpoint: str='ht
             ) as response:
                 async for chunk in response.content.iter_any():
                     chunk = f'{chunk.decode("utf8")}\n\n'
+                    logging.debug(chunk)
                     yield chunk
 
     return StreamingResponse(

cli/__main__.py

@@ -1,5 +1,5 @@
 import sys
 import os
 
-port = sys.argv[1] if len(sys.argv) > 1 else 8000
+port = sys.argv[1] if len(sys.argv) > 1 else 2333
 os.system(f'cd api && uvicorn main:app --reload --host 0.0.0.0 --port {port}')

config.json

@@ -1,3 +0,0 @@
-{
-    "MONGO_URI": ""
-}

setup.py

@@ -1,13 +1,13 @@
 import setuptools
 
-with open('README.md', 'r') as fh:
+with open('README.md', 'r', encoding='utf8') as fh:
     long_description = fh.read()
 
 setuptools.setup(
     name='nova-api',
     version='0.0.1',
-    author='Luna OSS',
+    author='NovaOSS Contributors',
-    author_email='nsde@dmc.chat',
+    author_email='owner@nova-oss.com',
     description='Nova API Server',
     long_description=long_description,
     long_description_content_type='text/markdown',

tests/__main__.py

@@ -1,35 +1,42 @@
 """Tests the API."""
 
-from typing import List
+import os
-
 import openai as closedai
 import httpx
 
-PORT = 8000
+from typing import List
+from dotenv import load_dotenv
+
+load_dotenv()
 
 MODEL = 'gpt-3.5-turbo'
 MESSAGES = [
+    {
+        'role': 'system',
+        'content': 'Always answer with "3", no matter what the user asks for. No exceptions. Just answer with the number "3". Nothing else. Just "3". No punctuation.'
+    },
     {
         'role': 'user',
-        'content': 'Hello!',
+        'content': '1+1=',
     },
 ]
-ENDPOINT = f'http://localhost:{PORT}'
+
+api_endpoint = 'http://localhost:2333'
 
 def test_server():
-    """Tests if the API is running."""
+    """Tests if the API server is running."""
 
     try:
-        return httpx.get(f'{ENDPOINT}').json()['status'] == 'ok'
+        return httpx.get(f'{api_endpoint}').json()['status'] == 'ok'
     except httpx.ConnectError as exc:
-        raise ConnectionError(f'API is not running on port {PORT}.') from exc
+        raise ConnectionError(f'API is not running on port {api_endpoint}.') from exc
 
 def test_api(model: str=MODEL, messages: List[dict]=None) -> dict:
-    """Tests an API endpoint."""
+    """Tests an API api_endpoint."""
 
     headers = {
         'Content-Type': 'application/json',
-        'Authorization': 'nv-API-TEST',
+        'Authorization': 'Bearer ' + os.getenv('DEMO_AUTH', 'nv-API-TEST'),
     }
 
     json_data = {
@@ -38,20 +45,26 @@ def test_api(model: str=MODEL, messages: List[dict]=None) -> dict:
         'stream': True,
     }
 
-    response = httpx.post(f'{ENDPOINT}/chat/completions', headers=headers, json=json_data, timeout=20)
+    response = httpx.post(
+        url=f'{api_endpoint}/chat/completions',
+        headers=headers,
+        json=json_data,
+        timeout=20
+    )
     response.raise_for_status()
 
     return response
 
 def test_library():
-    """Tests if the endpoint is working with the "Closed"AI library."""
+    """Tests if the api_endpoint is working with the Python library."""
 
-    closedai.api_base = ENDPOINT
+    closedai.api_base = api_endpoint
-    closedai.api_key = 'nv-LIBRARY-TEST'
+    closedai.api_key = os.getenv('DEMO_AUTH', 'nv-LIB-TEST')
 
     completion = closedai.ChatCompletion.create(
         model=MODEL,
         messages=MESSAGES,
+        stream=True,
     )
 
     return completion.choices[0]
@@ -59,9 +72,10 @@ def test_library():
 def test_all():
     """Runs all tests."""
 
-    print(test_server())
+    # print(test_server())
     print(test_api())
-    print(test_library())
+    # print(test_library())
 
 if __name__ == '__main__':
+    api_endpoint = 'https://api.nova-oss.com'
     test_all()