mirror of
https://github.com/NovaOSS/nova-api.git
synced 2024-11-25 16:13:58 +01:00
proxies have issues
This commit is contained in:
parent
72cea38d8d
commit
bf7a6b565a
|
@ -6,7 +6,7 @@ costs:
|
|||
|
||||
chat-models:
|
||||
gpt-3: 10
|
||||
gpt-4: 75
|
||||
gpt-4: 30
|
||||
gpt-4-32k: 100
|
||||
|
||||
# bonuses are multiplier for costs:
|
||||
|
@ -20,4 +20,4 @@ bonuses:
|
|||
# discord reward 0.99^lvl?
|
||||
|
||||
rewards:
|
||||
day: 1000
|
||||
day: 250
|
||||
|
|
|
@ -19,6 +19,7 @@ async def log_api_request(user: dict, incoming_request, target_url: str):
|
|||
last_prompt = payload['messages'][-1]['content']
|
||||
|
||||
model = payload.get('model')
|
||||
ip_address = await network.get_ip(incoming_request)
|
||||
|
||||
new_log_item = {
|
||||
'timestamp': time.time(),
|
||||
|
@ -26,7 +27,7 @@ async def log_api_request(user: dict, incoming_request, target_url: str):
|
|||
'path': incoming_request.url.path,
|
||||
'user_id': user['_id'],
|
||||
'security': {
|
||||
'ip': network.get_ip(incoming_request),
|
||||
'ip': ip_address,
|
||||
'useragent': incoming_request.headers.get('User-Agent')
|
||||
},
|
||||
'details': {
|
||||
|
|
58
api/helpers/chat.py
Normal file
58
api/helpers/chat.py
Normal file
|
@ -0,0 +1,58 @@
|
|||
import string
|
||||
import random
|
||||
import asyncio
|
||||
|
||||
from rich import print
|
||||
|
||||
class CompletionStart:
|
||||
"""Beinning of a chat"""
|
||||
|
||||
class CompletionStop:
|
||||
"""End of a chat"""
|
||||
|
||||
async def create_chat_id() -> str:
|
||||
chars = string.ascii_letters + string.digits
|
||||
chat_id = ''.join(random.choices(chars, k=32))
|
||||
|
||||
return f'chatcmpl-{chat_id}'
|
||||
|
||||
def create_chat_chunk(chat_id: str, model: str, content=None) -> dict:
|
||||
content = content or {}
|
||||
|
||||
delta = {}
|
||||
|
||||
if content:
|
||||
delta = {
|
||||
'content': content
|
||||
}
|
||||
|
||||
if not isinstance(content, str):
|
||||
delta = {
|
||||
'role': 'assistant'
|
||||
}
|
||||
|
||||
chunk = {
|
||||
'id': chat_id,
|
||||
'object': 'chat.completion.chunk',
|
||||
'created': 0,
|
||||
'model': model,
|
||||
'choices': [
|
||||
{
|
||||
'delta': delta,
|
||||
'index': 0,
|
||||
'finish_reason': None if not(isinstance(content, str)) else 'stop'
|
||||
}
|
||||
],
|
||||
}
|
||||
print(chunk)
|
||||
|
||||
return chunk
|
||||
|
||||
if __name__ == '__main__':
|
||||
demo_chat_id = asyncio.run(create_chat_id())
|
||||
print(demo_chat_id)
|
||||
print(asyncio.run(create_chat_chunk(
|
||||
model='gpt-4',
|
||||
content='Hello',
|
||||
chat_id=demo_chat_id,
|
||||
)))
|
|
@ -1,2 +1,20 @@
|
|||
import base64
|
||||
import asyncio
|
||||
|
||||
async def get_ip(request) -> str:
|
||||
return request.client.host
|
||||
|
||||
async def add_proxy_auth_to_headers(username: str, password: str, headers: dict) -> dict:
|
||||
proxy_auth = base64.b64encode(f'{username}:{password}'.encode()).decode()
|
||||
headers['Proxy-Authorization'] = f'Basic {proxy_auth}'
|
||||
return headers
|
||||
|
||||
if __name__ == '__main__':
|
||||
print(asyncio.run(add_proxy_auth_to_headers(
|
||||
'user',
|
||||
'pass',
|
||||
{
|
||||
'Authorization': 'Bearer demo',
|
||||
'Another-Header': '123'
|
||||
}
|
||||
)))
|
||||
|
|
|
@ -5,12 +5,18 @@ import chat_providers
|
|||
|
||||
provider_modules = [
|
||||
chat_providers.twa,
|
||||
chat_providers.quantum,
|
||||
chat_providers.churchless,
|
||||
chat_providers.closed,
|
||||
chat_providers.closed4
|
||||
# chat_providers.quantum,
|
||||
# chat_providers.churchless,
|
||||
# chat_providers.closed,
|
||||
# chat_providers.closed4
|
||||
]
|
||||
|
||||
def _get_module_name(module) -> str:
|
||||
name = module.__name__
|
||||
if '.' in name:
|
||||
return name.split('.')[-1]
|
||||
return name
|
||||
|
||||
async def balance_chat_request(payload: dict) -> dict:
|
||||
providers_available = []
|
||||
|
||||
|
@ -24,7 +30,10 @@ async def balance_chat_request(payload: dict) -> dict:
|
|||
providers_available.append(provider_module)
|
||||
|
||||
provider = random.choice(providers_available)
|
||||
return provider.chat_completion(**payload)
|
||||
target = provider.chat_completion(**payload)
|
||||
target['module'] = _get_module_name(provider)
|
||||
|
||||
return target
|
||||
|
||||
async def balance_organic_request(request: dict) -> dict:
|
||||
providers_available = []
|
||||
|
@ -34,8 +43,10 @@ async def balance_organic_request(request: dict) -> dict:
|
|||
providers_available.append(provider_module)
|
||||
|
||||
provider = random.choice(providers_available)
|
||||
target = provider.organify(request)
|
||||
target['module'] = _get_module_name(provider)
|
||||
|
||||
return provider.organify(request)
|
||||
return target
|
||||
|
||||
if __name__ == '__main__':
|
||||
req = asyncio.run(balance_chat_request(payload={'model': 'gpt-3.5-turbo', 'stream': True}))
|
||||
|
|
|
@ -2,6 +2,7 @@
|
|||
|
||||
import os
|
||||
import socket
|
||||
import random
|
||||
import asyncio
|
||||
import aiohttp
|
||||
import aiohttp_socks
|
||||
|
@ -14,21 +15,34 @@ class Proxy:
|
|||
"""Represents a proxy. The type can be either http, https, socks4 or socks5."""
|
||||
|
||||
def __init__(self,
|
||||
url: str=None,
|
||||
proxy_type: str='http',
|
||||
host_or_ip: str='127.0.0.1',
|
||||
port: int=8080,
|
||||
username: str=None,
|
||||
password: str=None
|
||||
):
|
||||
if url:
|
||||
proxy_type = url.split('://')[0]
|
||||
url = url.split('://')[1]
|
||||
|
||||
if '@' in url:
|
||||
username = url.split('@')[1].split(':')[0]
|
||||
password = url.split('@')[1].split(':')[1]
|
||||
|
||||
host_or_ip = url.split(':')[0]
|
||||
port = url.split(':')[1]
|
||||
|
||||
self.proxy_type = proxy_type
|
||||
self.host_or_ip = host_or_ip
|
||||
self.ip_address = socket.gethostbyname(self.host_or_ip) if host_or_ip[0].isdigit() else host_or_ip
|
||||
self.ip_address = socket.gethostbyname(self.host_or_ip) #if host_or_ip.replace('.', '').isdigit() else host_or_ip
|
||||
self.host = self.host_or_ip
|
||||
self.port = port
|
||||
self.username = username
|
||||
self.password = password
|
||||
|
||||
self.url = f'{self.proxy_type}://{self.username}:{self.password}@{self.host}:{self.port}'
|
||||
self.url_ip = f'{self.proxy_type}://{self.username}:{self.password}@{self.ip_address}:{self.port}'
|
||||
self.urls = {
|
||||
'http': self.url,
|
||||
'https': self.url
|
||||
|
@ -37,21 +51,6 @@ class Proxy:
|
|||
self.urls_httpx = {k + '://' :v for k, v in self.urls.items()}
|
||||
self.proxies = self.url
|
||||
|
||||
async def initialize_connector(self, connector):
|
||||
async with aiohttp.ClientSession(
|
||||
connector=connector,
|
||||
timeout=aiohttp.ClientTimeout(total=10),
|
||||
raise_for_status=True
|
||||
) as session:
|
||||
async with session.request(
|
||||
method='get',
|
||||
url='https://checkip.amazonaws.com',
|
||||
headers={'Content-Type': 'application/json'}
|
||||
) as response:
|
||||
detected_ip = await response.text()
|
||||
print(f'Detected IP: {detected_ip}')
|
||||
return detected_ip.strip()
|
||||
|
||||
@property
|
||||
def connector(self):
|
||||
proxy_types = {
|
||||
|
@ -63,13 +62,33 @@ class Proxy:
|
|||
|
||||
return aiohttp_socks.ProxyConnector(
|
||||
proxy_type=proxy_types[self.proxy_type],
|
||||
host=self.host,
|
||||
host=self.ip_address,
|
||||
port=self.port,
|
||||
rdns=False,
|
||||
username=self.username,
|
||||
password=self.password
|
||||
)
|
||||
|
||||
proxies_in_files = []
|
||||
|
||||
for proxy_type in ['http', 'socks4', 'socks5']:
|
||||
with open(f'secret/proxies/{proxy_type}.txt') as f:
|
||||
for line in f.readlines():
|
||||
if line.strip() and not line.strip().startswith('#'):
|
||||
if '#' in line:
|
||||
line = line.split('#')[0]
|
||||
|
||||
proxies_in_files.append(f'{proxy_type}://{line.strip()}')
|
||||
|
||||
class ProxyChain:
|
||||
def __init__(self):
|
||||
random_proxy = random.choice(proxies_in_files)
|
||||
|
||||
self.get_random = Proxy(url=random_proxy)
|
||||
self.connector = aiohttp_socks.ChainProxyConnector.from_urls(proxies_in_files)
|
||||
|
||||
default_chain = ProxyChain()
|
||||
|
||||
default_proxy = Proxy(
|
||||
proxy_type=os.getenv('PROXY_TYPE', 'http'),
|
||||
host_or_ip=os.getenv('PROXY_HOST', '127.0.0.1'),
|
||||
|
@ -78,6 +97,8 @@ default_proxy = Proxy(
|
|||
password=os.getenv('PROXY_PASS')
|
||||
)
|
||||
|
||||
random_proxy = ProxyChain().get_random
|
||||
|
||||
def test_httpx_workaround():
|
||||
import httpx
|
||||
|
||||
|
@ -106,7 +127,41 @@ async def test_aiohttp_socks():
|
|||
html = await response.text()
|
||||
return html.strip()
|
||||
|
||||
async def streaming_aiohttp_socks():
|
||||
async with aiohttp.ClientSession(connector=default_proxy.connector) as session:
|
||||
async with session.post(
|
||||
'https://free.churchless.tech/v1/chat/completions',
|
||||
json={
|
||||
"model": "gpt-3.5-turbo",
|
||||
"messages": [
|
||||
{
|
||||
"role": "user",
|
||||
"content": "Hi"
|
||||
}
|
||||
],
|
||||
"stream": True
|
||||
},
|
||||
# headers={
|
||||
# 'Authorization': 'Bearer MyDiscord'
|
||||
# }
|
||||
) as response:
|
||||
html = await response.text()
|
||||
return html.strip()
|
||||
|
||||
async def text_httpx_socks():
|
||||
import httpx
|
||||
from httpx_socks import AsyncProxyTransport
|
||||
|
||||
print(default_proxy.url_ip)
|
||||
|
||||
transport = AsyncProxyTransport.from_url(default_proxy.url_ip)
|
||||
async with httpx.AsyncClient(transport=transport) as client:
|
||||
res = await client.get('https://checkip.amazonaws.com')
|
||||
return res.text
|
||||
|
||||
if __name__ == '__main__':
|
||||
# print(test_httpx())
|
||||
# print(test_requests())
|
||||
print(asyncio.run(test_aiohttp_socks()))
|
||||
# print(asyncio.run(test_aiohttp_socks()))
|
||||
# print(asyncio.run(streaming_aiohttp_socks()))
|
||||
print(asyncio.run(text_httpx_socks()))
|
||||
|
|
119
api/streaming.py
119
api/streaming.py
|
@ -1,17 +1,18 @@
|
|||
import os
|
||||
import yaml
|
||||
import json
|
||||
import asyncio
|
||||
import aiohttp
|
||||
import starlette
|
||||
|
||||
from rich import print
|
||||
from dotenv import load_dotenv
|
||||
|
||||
import proxies
|
||||
import load_balancing
|
||||
|
||||
from db import logs, users, stats
|
||||
from rich import print
|
||||
from helpers import network
|
||||
from helpers import network, chat
|
||||
|
||||
load_dotenv()
|
||||
|
||||
|
@ -38,21 +39,38 @@ async def stream(
|
|||
incoming_request: starlette.requests.Request=None,
|
||||
):
|
||||
payload = payload or DEMO_PAYLOAD
|
||||
is_chat = False
|
||||
|
||||
if 'chat/completions' in path: # is a chat endpoint
|
||||
target_request = await load_balancing.balance_chat_request(payload)
|
||||
else:
|
||||
target_request = await load_balancing.balance_organic_request(payload)
|
||||
if 'chat/completions' in path:
|
||||
is_chat = True
|
||||
chat_id = await chat.create_chat_id()
|
||||
model = payload['model']
|
||||
|
||||
headers = {
|
||||
'Content-Type': 'application/json'
|
||||
}
|
||||
chat_chunk = chat.create_chat_chunk(
|
||||
chat_id=chat_id,
|
||||
model=model,
|
||||
content=chat.CompletionStart
|
||||
)
|
||||
data = json.dumps(chat_chunk)
|
||||
|
||||
for k, v in target_request.get('headers', {}).items():
|
||||
headers[k] = v
|
||||
chunk = f'data: {data}'
|
||||
|
||||
yield chunk
|
||||
|
||||
for _ in range(5):
|
||||
async with aiohttp.ClientSession(connector=proxies.default_proxy.connector) as session:
|
||||
if is_chat:
|
||||
target_request = await load_balancing.balance_chat_request(payload)
|
||||
else:
|
||||
target_request = await load_balancing.balance_organic_request(payload)
|
||||
|
||||
headers = {
|
||||
'Content-Type': 'application/json'
|
||||
}
|
||||
|
||||
for k, v in target_request.get('headers', {}).items():
|
||||
headers[k] = v
|
||||
|
||||
async with aiohttp.ClientSession(connector=proxies.random_proxy.connector) as session:
|
||||
async with session.request(
|
||||
method=target_request.get('method', 'POST'),
|
||||
url=target_request['url'],
|
||||
|
@ -70,38 +88,63 @@ async def stream(
|
|||
try:
|
||||
await response.raise_for_status()
|
||||
except Exception as exc:
|
||||
if 'Too Many Requests' in str(exc):
|
||||
continue
|
||||
else:
|
||||
break
|
||||
continue
|
||||
# if 'Too Many Requests' in str(exc):
|
||||
|
||||
if user and incoming_request:
|
||||
await logs.log_api_request(
|
||||
user=user,
|
||||
incoming_request=incoming_request,
|
||||
target_url=target_request['url']
|
||||
)
|
||||
if user and incoming_request:
|
||||
await logs.log_api_request(
|
||||
user=user,
|
||||
incoming_request=incoming_request,
|
||||
target_url=target_request['url']
|
||||
)
|
||||
|
||||
if credits_cost and user:
|
||||
await users.update_by_id(user['_id'], {
|
||||
'$inc': {'credits': -credits_cost}
|
||||
})
|
||||
if credits_cost and user:
|
||||
await users.update_by_id(user['_id'], {
|
||||
'$inc': {'credits': -credits_cost}
|
||||
})
|
||||
|
||||
if not demo_mode:
|
||||
await stats.add_date()
|
||||
await stats.add_ip_address(network.get_ip(incoming_request))
|
||||
await stats.add_model(payload.get('model', '_non-chat'))
|
||||
await stats.add_path(path)
|
||||
await stats.add_target(target_request['url'])
|
||||
await stats.add_tokens(input_tokens)
|
||||
if not demo_mode:
|
||||
ip_address = await network.get_ip(incoming_request)
|
||||
|
||||
async for chunk in response.content.iter_chunks():
|
||||
# chunk = f'{chunk.decode("utf8")}\n\n'
|
||||
await stats.add_date()
|
||||
await stats.add_ip_address(ip_address)
|
||||
await stats.add_path(path)
|
||||
await stats.add_target(target_request['url'])
|
||||
|
||||
if demo_mode:
|
||||
print(chunk)
|
||||
if is_chat:
|
||||
await stats.add_model(model)
|
||||
await stats.add_tokens(input_tokens, model)
|
||||
|
||||
yield chunk
|
||||
async for chunk in response.content.iter_any():
|
||||
chunk = f'{chunk.decode("utf8")}\n\n'
|
||||
|
||||
if chunk.strip():
|
||||
if is_chat:
|
||||
if target_request['module'] == 'twa':
|
||||
data = json.loads(chunk.split('data: ')[1])
|
||||
|
||||
if data.get('text'):
|
||||
chat_chunk = chat.create_chat_chunk(
|
||||
chat_id=chat_id,
|
||||
model=model,
|
||||
content=['text']
|
||||
)
|
||||
data = json.dumps(chat_chunk)
|
||||
|
||||
chunk = f'data: {data}'
|
||||
|
||||
yield chunk
|
||||
break
|
||||
if is_chat:
|
||||
chat_chunk = chat.create_chat_chunk(
|
||||
chat_id=chat_id,
|
||||
model=model,
|
||||
content=chat.CompletionStop
|
||||
)
|
||||
data = json.dumps(chat_chunk)
|
||||
|
||||
yield f'data: {data}'
|
||||
yield 'data: [DONE]'
|
||||
|
||||
if __name__ == '__main__':
|
||||
asyncio.run(stream())
|
||||
|
|
|
@ -22,8 +22,6 @@ logging.basicConfig(
|
|||
format='%(asctime)s %(levelname)s %(name)s %(message)s'
|
||||
)
|
||||
|
||||
logging.info('API started')
|
||||
|
||||
with open('config/credits.yml', encoding='utf8') as f:
|
||||
credits_config = yaml.safe_load(f)
|
||||
|
||||
|
|
Loading…
Reference in a new issue