3 files changed, 48 insertions, 40 deletions

diff --git a/etc/testing/test_async.py b/etc/testing/test_async.py
index 2c15f6b0..12ec732e 100644
--- a/etc/testing/test_async.py
+++ b/etc/testing/test_async.py
@@ -6,7 +6,7 @@ sys.path.append(str(Path(__file__).parent.parent))
 sys.path.append(str(Path(__file__).parent.parent.parent))
 
 import g4f
-from testing.test_providers import get_providers
+from testing._providers import get_providers
 from testing.log_time import log_time_async
 
 async def create_async(provider):
diff --git a/g4f/Provider/ChatgptAi.py b/g4f/Provider/ChatgptAi.py
index cf45909c..d5fd5bff 100644
--- a/g4f/Provider/ChatgptAi.py
+++ b/g4f/Provider/ChatgptAi.py
@@ -1,36 +1,34 @@
 from __future__ import annotations
 
-import re
+import re, html, json, string, random
 from aiohttp import ClientSession
 
-from ..typing import Messages
-from .base_provider import AsyncProvider, format_prompt
+from ..typing import Messages, AsyncResult
+from .base_provider import AsyncGeneratorProvider
 
 
-class ChatgptAi(AsyncProvider):
-    url: str = "https://chatgpt.ai/"
+class ChatgptAi(AsyncGeneratorProvider):
+    url: str = "https://chatgpt.ai"
     working = True
     supports_gpt_35_turbo  = True
-    _nonce = None
-    _post_id = None
-    _bot_id = None
+    _system = None
 
     @classmethod
-    async def create_async(
+    async def create_async_generator(
         cls,
         model: str,
         messages: Messages,
         proxy: str = None,
         **kwargs
-    ) -> str:
+    ) -> AsyncResult:
         headers = {
             "authority"          : "chatgpt.ai",
             "accept"             : "*/*",
-            "accept-language"    : "en,fr-FR;q=0.9,fr;q=0.8,es-ES;q=0.7,es;q=0.6,en-US;q=0.5,am;q=0.4,de;q=0.3",
+            "accept-language"    : "en-US",
             "cache-control"      : "no-cache",
-            "origin"             : "https://chatgpt.ai",
+            "origin"             : cls.url,
             "pragma"             : "no-cache",
-            "referer"            : cls.url,
+            "referer"            : f"{cls.url}/",
             "sec-ch-ua"          : '"Not.A/Brand";v="8", "Chromium";v="114", "Google Chrome";v="114"',
             "sec-ch-ua-mobile"   : "?0",
             "sec-ch-ua-platform" : '"Windows"',
@@ -42,34 +40,40 @@ class ChatgptAi(AsyncProvider):
         async with ClientSession(
             headers=headers
         ) as session:
-            if not cls._nonce:
+            if not cls._system:
                 async with session.get(cls.url, proxy=proxy) as response:
                     response.raise_for_status()
                     text = await response.text()
-                result = re.search(r'data-nonce="(.*?)"', text)
+                result = re.search(r"data-system='(.*?)'", text)
                 if result:
-                    cls._nonce = result.group(1)
-                result = re.search(r'data-post-id="(.*?)"', text)
-                if result:
-                    cls._post_id = result.group(1)
-                result = re.search(r'data-bot-id="(.*?)"', text)
-                if result:
-                    cls._bot_id = result.group(1)
-                if not cls._nonce or not cls._post_id or not cls._bot_id:
-                    raise RuntimeError("Nonce, post-id or bot-id not found")
-
+                    cls._system = json.loads(html.unescape(result.group(1)))
+                if not cls._system:
+                    raise RuntimeError("System args not found")
+                
             data = {
-                "_wpnonce": cls._nonce,
-                "post_id": cls._post_id,
-                "url": "https://chatgpt.ai",
-                "action": "wpaicg_chat_shortcode_message",
-                "message": format_prompt(messages),
-                "bot_id": cls._bot_id
+                "botId": cls._system["botId"],
+                "customId": cls._system["customId"],
+                "session": cls._system["sessionId"],
+                "chatId": "".join(random.choices(f"{string.ascii_lowercase}{string.digits}", k=11)),
+                "contextId": cls._system["contextId"],
+                "messages": messages,
+                "newMessage": messages[-1]["content"],
+                "stream": True
             }
             async with session.post(
-                "https://chatgpt.ai/wp-admin/admin-ajax.php",
+               f"{cls.url}/wp-json/mwai-ui/v1/chats/submit",
                 proxy=proxy,
-                data=data
+                json=data
             ) as response:
                 response.raise_for_status()
-                return (await response.json())["data"]
\ No newline at end of file
+                async for line in response.content:
+                    if line.startswith(b"data: "):
+                        try:
+                            line = json.loads(line[6:])
+                            assert "type" in line
+                        except:
+                            raise RuntimeError(f"Broken line: {line.decode()}")
+                        if line["type"] == "live":
+                            yield line["data"]
+                        elif line["type"] == "end":
+                            break
\ No newline at end of file
diff --git a/g4f/Provider/Geekgpt.py b/g4f/Provider/Geekgpt.py
index 3bbc0e75..3c577cf8 100644
--- a/g4f/Provider/Geekgpt.py
+++ b/g4f/Provider/Geekgpt.py
@@ -14,11 +14,15 @@ class GeekGpt(BaseProvider):
     supports_gpt_4        = True
 
     @classmethod
-    def create_completion(cls,
-                        model: str,
-                        messages: Messages,
-                        stream: bool, **kwargs) -> CreateResult:
-
+    def create_completion(
+        cls,
+        model: str,
+        messages: Messages,
+        stream: bool,
+        **kwargs
+    ) -> CreateResult:
+        if not model:
+            model = "gpt-3.5-turbo"
         json_data = {
             'messages': messages,
             'model': model,