Copilot.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264
  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
  23. from ..requests.raise_for_status import raise_for_status
  24. from ..providers.response import BaseConversation, JsonConversation, RequestLogin, Parameters
  25. from ..providers.asyncio import get_running_loop
  26. from ..requests import get_nodriver
  27. from ..image import ImageResponse, to_bytes, is_accepted_format
  28. from .. import debug
  29. class Conversation(JsonConversation):
  30. conversation_id: str
  31. def __init__(self, conversation_id: str):
  32. self.conversation_id = conversation_id
  33. class Copilot(AbstractProvider, ProviderModelMixin):
  34. label = "Microsoft Copilot"
  35. url = "https://copilot.microsoft.com"
  36. working = True
  37. supports_stream = True
  38. default_model = "Copilot"
  39. models = [default_model]
  40. model_aliases = {
  41. "gpt-4": default_model,
  42. "gpt-4o": default_model,
  43. }
  44. websocket_url = "wss://copilot.microsoft.com/c/api/chat?api-version=2"
  45. conversation_url = f"{url}/c/api/conversations"
  46. _access_token: str = None
  47. _cookies: CookieJar = None
  48. @classmethod
  49. def create_completion(
  50. cls,
  51. model: str,
  52. messages: Messages,
  53. stream: bool = False,
  54. proxy: str = None,
  55. timeout: int = 900,
  56. prompt: str = None,
  57. images: ImagesType = None,
  58. conversation: BaseConversation = None,
  59. return_conversation: bool = False,
  60. api_key: str = None,
  61. web_search: bool = False,
  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. yield Parameters(**{"api_key": cls._access_token, "cookies": cls._cookies if isinstance(cls._cookies, dict) else {c.name: c.value for c in cls._cookies}})
  85. websocket_url = f"{websocket_url}&accessToken={quote(cls._access_token)}"
  86. headers = {"authorization": f"Bearer {cls._access_token}"}
  87. with Session(
  88. timeout=timeout,
  89. proxy=proxy,
  90. impersonate="chrome",
  91. headers=headers,
  92. cookies=cls._cookies,
  93. ) as session:
  94. if cls._access_token is not None:
  95. cls._cookies = session.cookies.jar
  96. # if cls._access_token is None:
  97. # try:
  98. # url = "https://copilot.microsoft.com/cl/eus-sc/collect"
  99. # headers = {
  100. # "Accept": "application/x-clarity-gzip",
  101. # "referrer": "https://copilot.microsoft.com/onboarding"
  102. # }
  103. # response = session.post(url, headers=headers, data=get_clarity())
  104. # clarity_token = json.loads(response.text.split(" ", maxsplit=1)[-1])[0]["value"]
  105. # debug.log(f"Copilot: Clarity Token: ...{clarity_token[-12:]}")
  106. # except Exception as e:
  107. # debug.log(f"Copilot: {e}")
  108. # else:
  109. # clarity_token = None
  110. response = session.get("https://copilot.microsoft.com/c/api/user")
  111. raise_for_status(response)
  112. user = response.json().get('firstName')
  113. if user is None:
  114. cls._access_token = None
  115. debug.log(f"Copilot: User: {user or 'null'}")
  116. if conversation is None:
  117. response = session.post(cls.conversation_url)
  118. raise_for_status(response)
  119. conversation_id = response.json().get("id")
  120. conversation = Conversation(conversation_id)
  121. if return_conversation:
  122. yield conversation
  123. if prompt is None:
  124. prompt = format_prompt_max_length(messages, 10000)
  125. debug.log(f"Copilot: Created conversation: {conversation_id}")
  126. else:
  127. conversation_id = conversation.conversation_id
  128. if prompt is None:
  129. prompt = messages[-1]["content"]
  130. debug.log(f"Copilot: Use conversation: {conversation_id}")
  131. uploaded_images = []
  132. if images is not None:
  133. for image, _ in images:
  134. data = to_bytes(image)
  135. response = session.post(
  136. "https://copilot.microsoft.com/c/api/attachments",
  137. headers={"content-type": is_accepted_format(data)},
  138. data=data
  139. )
  140. raise_for_status(response)
  141. uploaded_images.append({"type":"image", "url": response.json().get("url")})
  142. break
  143. wss = session.ws_connect(cls.websocket_url)
  144. # if clarity_token is not None:
  145. # wss.send(json.dumps({
  146. # "event": "challengeResponse",
  147. # "token": clarity_token,
  148. # "method":"clarity"
  149. # }).encode(), CurlWsFlag.TEXT)
  150. wss.send(json.dumps({
  151. "event": "send",
  152. "conversationId": conversation_id,
  153. "content": [*uploaded_images, {
  154. "type": "text",
  155. "text": prompt,
  156. }],
  157. "mode": "chat"
  158. }).encode(), CurlWsFlag.TEXT)
  159. is_started = False
  160. msg = None
  161. image_prompt: str = None
  162. last_msg = None
  163. try:
  164. while True:
  165. try:
  166. msg = wss.recv()[0]
  167. msg = json.loads(msg)
  168. except:
  169. break
  170. last_msg = msg
  171. if msg.get("event") == "appendText":
  172. is_started = True
  173. yield msg.get("text")
  174. elif msg.get("event") == "generatingImage":
  175. image_prompt = msg.get("prompt")
  176. elif msg.get("event") == "imageGenerated":
  177. yield ImageResponse(msg.get("url"), image_prompt, {"preview": msg.get("thumbnailUrl")})
  178. elif msg.get("event") == "done":
  179. break
  180. elif msg.get("event") == "replaceText":
  181. yield msg.get("text")
  182. elif msg.get("event") == "error":
  183. raise RuntimeError(f"Error: {msg}")
  184. elif msg.get("event") not in ["received", "startMessage", "citation", "partCompleted"]:
  185. debug.log(f"Copilot Message: {msg}")
  186. if not is_started:
  187. raise RuntimeError(f"Invalid response: {last_msg}")
  188. finally:
  189. yield Parameters(**{"conversation": conversation.get_dict(), "user": user, "prompt": prompt})
  190. yield Parameters(**{"cookies": {c.name: c.value for c in session.cookies.jar}})
  191. async def get_access_token_and_cookies(url: str, proxy: str = None, target: str = "ChatAI",):
  192. browser = await get_nodriver(proxy=proxy, user_data_dir="copilot")
  193. try:
  194. page = await browser.get(url)
  195. access_token = None
  196. while access_token is None:
  197. access_token = await page.evaluate("""
  198. (() => {
  199. for (var i = 0; i < localStorage.length; i++) {
  200. try {
  201. item = JSON.parse(localStorage.getItem(localStorage.key(i)));
  202. if (item.credentialType == "AccessToken"
  203. && item.expiresOn > Math.floor(Date.now() / 1000)
  204. && item.target.includes("target")) {
  205. return item.secret;
  206. }
  207. } catch(e) {}
  208. }
  209. })()
  210. """.replace('"target"', json.dumps(target)))
  211. if access_token is None:
  212. await asyncio.sleep(1)
  213. cookies = {}
  214. for c in await page.send(nodriver.cdp.network.get_cookies([url])):
  215. cookies[c.name] = c.value
  216. await page.close()
  217. return access_token, cookies
  218. finally:
  219. browser.stop()
  220. def readHAR(url: str):
  221. api_key = None
  222. cookies = None
  223. for path in get_har_files():
  224. with open(path, 'rb') as file:
  225. try:
  226. harFile = json.loads(file.read())
  227. except json.JSONDecodeError:
  228. # Error: not a HAR file!
  229. continue
  230. for v in harFile['log']['entries']:
  231. if v['request']['url'].startswith(url):
  232. v_headers = get_headers(v)
  233. if "authorization" in v_headers:
  234. api_key = v_headers["authorization"].split(maxsplit=1).pop()
  235. if v['request']['cookies']:
  236. cookies = {c['name']: c['value'] for c in v['request']['cookies']}
  237. if api_key is None:
  238. raise NoValidHarFileError("No access token found in .har files")
  239. return api_key, cookies
  240. def get_clarity() -> bytes:
  241. #{"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]]]}
  242. body = base64.b64decode("H4sIAAAAAAAAA23RwU7DMAwG4HfJ2aqS2E5ibjxH1cMOnQYqYZvUTQPx7vyJRGGAemj01XWcP+9udg+j80MetDhSyrEISc5GrqrtZnmaTydHbrdUnSsWYT2u+8Obo0Ce/IQvaDBmjkwhUlKKIRNHmQgosqEArWPRDQMx90rxeUMPzB1j+UJvwNIxhTvsPcXyX1T+rizE4juK3mEEhpAUg/JvzW1/+U/tB1LATmhqotoiweMea50PLy2vui4LOY3XfD1dwnkor5fn/e18XBFgm6fHjSzZmCyV7d3aRByAEYextaTHEH3i5pgKGVP/s+DScE5PuLKIpW6FnCi1gY3Rbpqmj0/DI/+L7QEAAA==")
  243. return body