Copilot.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265
  1. from __future__ import annotations
  2. import os
  3. import json
  4. import asyncio
  5. import base64
  6. from http.cookiejar import CookieJar
  7. from urllib.parse import quote
  8. try:
  9. from curl_cffi.requests import Session, CurlWsFlag
  10. has_curl_cffi = True
  11. except ImportError:
  12. has_curl_cffi = False
  13. try:
  14. import nodriver
  15. has_nodriver = True
  16. except ImportError:
  17. has_nodriver = False
  18. from .base_provider import AbstractProvider, ProviderModelMixin
  19. from .helper import format_prompt_max_length
  20. from .openai.har_file import get_headers, get_har_files
  21. from ..typing import CreateResult, Messages, ImagesType
  22. from ..errors import MissingRequirementsError, NoValidHarFileError, MissingAuthError
  23. from ..requests.raise_for_status import raise_for_status
  24. from ..providers.response import BaseConversation, JsonConversation, RequestLogin, Parameters, ImageResponse
  25. from ..providers.asyncio import get_running_loop
  26. from ..requests import get_nodriver
  27. from ..image import to_bytes, is_accepted_format
  28. from .helper import get_last_user_message
  29. from .. import debug
  30. class Conversation(JsonConversation):
  31. conversation_id: str
  32. def __init__(self, conversation_id: str):
  33. self.conversation_id = conversation_id
  34. class Copilot(AbstractProvider, ProviderModelMixin):
  35. label = "Microsoft Copilot"
  36. url = "https://copilot.microsoft.com"
  37. working = True
  38. supports_stream = True
  39. default_model = "Copilot"
  40. models = [default_model]
  41. model_aliases = {
  42. "gpt-4": default_model,
  43. "gpt-4o": default_model,
  44. }
  45. websocket_url = "wss://copilot.microsoft.com/c/api/chat?api-version=2"
  46. conversation_url = f"{url}/c/api/conversations"
  47. _access_token: str = None
  48. _cookies: CookieJar = None
  49. @classmethod
  50. def create_completion(
  51. cls,
  52. model: str,
  53. messages: Messages,
  54. stream: bool = False,
  55. proxy: str = None,
  56. timeout: int = 900,
  57. prompt: str = None,
  58. images: ImagesType = None,
  59. conversation: BaseConversation = None,
  60. return_conversation: bool = False,
  61. api_key: str = None,
  62. **kwargs
  63. ) -> CreateResult:
  64. if not has_curl_cffi:
  65. raise MissingRequirementsError('Install or update "curl_cffi" package | pip install -U curl_cffi')
  66. websocket_url = cls.websocket_url
  67. headers = None
  68. if cls.needs_auth or images is not None:
  69. if api_key is not None:
  70. cls._access_token = api_key
  71. if cls._access_token is None:
  72. try:
  73. cls._access_token, cls._cookies = readHAR(cls.url)
  74. except NoValidHarFileError as h:
  75. debug.log(f"Copilot: {h}")
  76. if has_nodriver:
  77. login_url = os.environ.get("G4F_LOGIN_URL")
  78. if login_url:
  79. yield RequestLogin(cls.label, login_url)
  80. get_running_loop(check_nested=True)
  81. cls._access_token, cls._cookies = asyncio.run(get_access_token_and_cookies(cls.url, proxy))
  82. else:
  83. raise h
  84. websocket_url = f"{websocket_url}&accessToken={quote(cls._access_token)}"
  85. headers = {"authorization": f"Bearer {cls._access_token}"}
  86. with Session(
  87. timeout=timeout,
  88. proxy=proxy,
  89. impersonate="chrome",
  90. headers=headers,
  91. cookies=cls._cookies,
  92. ) as session:
  93. if cls._access_token is not None:
  94. cls._cookies = session.cookies.jar
  95. # if cls._access_token is None:
  96. # try:
  97. # url = "https://copilot.microsoft.com/cl/eus-sc/collect"
  98. # headers = {
  99. # "Accept": "application/x-clarity-gzip",
  100. # "referrer": "https://copilot.microsoft.com/onboarding"
  101. # }
  102. # response = session.post(url, headers=headers, data=get_clarity())
  103. # clarity_token = json.loads(response.text.split(" ", maxsplit=1)[-1])[0]["value"]
  104. # debug.log(f"Copilot: Clarity Token: ...{clarity_token[-12:]}")
  105. # except Exception as e:
  106. # debug.log(f"Copilot: {e}")
  107. # else:
  108. # clarity_token = None
  109. response = session.get("https://copilot.microsoft.com/c/api/user")
  110. if response.status_code == 401:
  111. raise MissingAuthError("Status 401: Invalid access token")
  112. raise_for_status(response)
  113. user = response.json().get('firstName')
  114. if user is None:
  115. cls._access_token = None
  116. debug.log(f"Copilot: User: {user or 'null'}")
  117. if conversation is None:
  118. response = session.post(cls.conversation_url)
  119. raise_for_status(response)
  120. conversation_id = response.json().get("id")
  121. conversation = Conversation(conversation_id)
  122. if return_conversation:
  123. yield conversation
  124. if prompt is None:
  125. prompt = format_prompt_max_length(messages, 10000)
  126. debug.log(f"Copilot: Created conversation: {conversation_id}")
  127. else:
  128. conversation_id = conversation.conversation_id
  129. if prompt is None:
  130. prompt = get_last_user_message(messages)
  131. debug.log(f"Copilot: Use conversation: {conversation_id}")
  132. uploaded_images = []
  133. if images is not None:
  134. for image, _ in images:
  135. data = to_bytes(image)
  136. response = session.post(
  137. "https://copilot.microsoft.com/c/api/attachments",
  138. headers={"content-type": is_accepted_format(data)},
  139. data=data
  140. )
  141. raise_for_status(response)
  142. uploaded_images.append({"type":"image", "url": response.json().get("url")})
  143. break
  144. wss = session.ws_connect(cls.websocket_url)
  145. # if clarity_token is not None:
  146. # wss.send(json.dumps({
  147. # "event": "challengeResponse",
  148. # "token": clarity_token,
  149. # "method":"clarity"
  150. # }).encode(), CurlWsFlag.TEXT)
  151. wss.send(json.dumps({
  152. "event": "send",
  153. "conversationId": conversation_id,
  154. "content": [*uploaded_images, {
  155. "type": "text",
  156. "text": prompt,
  157. }],
  158. "mode": "chat"
  159. }).encode(), CurlWsFlag.TEXT)
  160. is_started = False
  161. msg = None
  162. image_prompt: str = None
  163. last_msg = None
  164. try:
  165. while True:
  166. try:
  167. msg = wss.recv()[0]
  168. msg = json.loads(msg)
  169. except:
  170. break
  171. last_msg = msg
  172. if msg.get("event") == "appendText":
  173. is_started = True
  174. yield msg.get("text")
  175. elif msg.get("event") == "generatingImage":
  176. image_prompt = msg.get("prompt")
  177. elif msg.get("event") == "imageGenerated":
  178. yield ImageResponse(msg.get("url"), image_prompt, {"preview": msg.get("thumbnailUrl")})
  179. elif msg.get("event") == "done":
  180. break
  181. elif msg.get("event") == "replaceText":
  182. yield msg.get("text")
  183. elif msg.get("event") == "error":
  184. raise RuntimeError(f"Error: {msg}")
  185. elif msg.get("event") not in ["received", "startMessage", "citation", "partCompleted"]:
  186. debug.log(f"Copilot Message: {msg}")
  187. if not is_started:
  188. raise RuntimeError(f"Invalid response: {last_msg}")
  189. finally:
  190. yield Parameters(**{"conversation": conversation.get_dict(), "user": user, "prompt": prompt})
  191. yield Parameters(**{"cookies": {c.name: c.value for c in session.cookies.jar}})
  192. async def get_access_token_and_cookies(url: str, proxy: str = None, target: str = "ChatAI",):
  193. browser, stop_browser = await get_nodriver(proxy=proxy, user_data_dir="copilot")
  194. try:
  195. page = await browser.get(url)
  196. access_token = None
  197. while access_token is None:
  198. access_token = await page.evaluate("""
  199. (() => {
  200. for (var i = 0; i < localStorage.length; i++) {
  201. try {
  202. item = JSON.parse(localStorage.getItem(localStorage.key(i)));
  203. if (item.credentialType == "AccessToken"
  204. && item.expiresOn > Math.floor(Date.now() / 1000)
  205. && item.target.includes("target")) {
  206. return item.secret;
  207. }
  208. } catch(e) {}
  209. }
  210. })()
  211. """.replace('"target"', json.dumps(target)))
  212. if access_token is None:
  213. await asyncio.sleep(1)
  214. cookies = {}
  215. for c in await page.send(nodriver.cdp.network.get_cookies([url])):
  216. cookies[c.name] = c.value
  217. await page.close()
  218. return access_token, cookies
  219. finally:
  220. stop_browser()
  221. def readHAR(url: str):
  222. api_key = None
  223. cookies = None
  224. for path in get_har_files():
  225. with open(path, 'rb') as file:
  226. try:
  227. harFile = json.loads(file.read())
  228. except json.JSONDecodeError:
  229. # Error: not a HAR file!
  230. continue
  231. for v in harFile['log']['entries']:
  232. if v['request']['url'].startswith(url):
  233. v_headers = get_headers(v)
  234. if "authorization" in v_headers:
  235. api_key = v_headers["authorization"].split(maxsplit=1).pop()
  236. if v['request']['cookies']:
  237. cookies = {c['name']: c['value'] for c in v['request']['cookies']}
  238. if api_key is None:
  239. raise NoValidHarFileError("No access token found in .har files")
  240. return api_key, cookies
  241. def get_clarity() -> bytes:
  242. #{"e":["0.7.58",5,7284,4779,"n59ae4ieqq","aln5en","1upufhz",1,0,0],"a":[[7323,12,65,217,324],[7344,12,65,214,329],[7385,12,65,211,334],[7407,12,65,210,337],[7428,12,65,209,338],[7461,12,65,209,339],[7497,12,65,209,339],[7531,12,65,208,340],[7545,12,65,208,342],[11654,13,65,208,342],[11728,14,65,208,342],[11728,9,65,208,342,17535,19455,0,0,0,"Annehmen",null,"52w7wqv1r.8ovjfyrpu",1],[7284,4,1,393,968,393,968,0,0,231,310,939,0],[12063,0,2,147,3,4,4,18,5,1,10,79,25,15],[12063,36,6,[11938,0]]]}
  243. body = base64.b64decode("H4sIAAAAAAAAA23RwU7DMAwG4HfJ2aqS2E5ibjxH1cMOnQYqYZvUTQPx7vyJRGGAemj01XWcP+9udg+j80MetDhSyrEISc5GrqrtZnmaTydHbrdUnSsWYT2u+8Obo0Ce/IQvaDBmjkwhUlKKIRNHmQgosqEArWPRDQMx90rxeUMPzB1j+UJvwNIxhTvsPcXyX1T+rizE4juK3mEEhpAUg/JvzW1/+U/tB1LATmhqotoiweMea50PLy2vui4LOY3XfD1dwnkor5fn/e18XBFgm6fHjSzZmCyV7d3aRByAEYextaTHEH3i5pgKGVP/s+DScE5PuLKIpW6FnCi1gY3Rbpqmj0/DI/+L7QEAAA==")
  244. return body