Merge pull request #2463 from hlohaus/neww

Add Authentication Setup Guide

README.md

@@ -107,7 +107,7 @@ docker run \
   hlohaus789/g4f:latest-slim \
   rm -r -f /app/g4f/ \
   && pip install -U g4f[slim] \
-  && python -m g4f.cli api --gui --debug
+  && python -m g4f --debug
 ```
 It also updates the `g4f` package at startup and installs any new required dependencies.
 
@@ -134,7 +134,7 @@ By following these steps, you should be able to successfully install and run the
 
 Run the **Webview UI** on other Platforms:
 
-- [/docs/guides/webview](docs/webview.md)
+- [/docs/webview](docs/webview.md)
 
 ##### Use your smartphone:
 
@@ -204,7 +204,7 @@ image_url = response.data[0].url
 print(f"Generated image URL: {image_url}")
 ```
 
-[![Image with cat](/docs/cat.jpeg)](docs/client.md)
+[![Image with cat](/docs/images/cat.jpeg)](docs/client.md)
 
 #### **Full Documentation for Python API**
    - **New:**
@@ -241,6 +241,10 @@ This API is designed for straightforward implementation and enhanced compatibili
 
 ### Configuration
 
+#### Authentication
+
+Refer to the [G4F Authentication Setup Guide](docs/authentication.md) for detailed instructions on setting up authentication.
+
 #### Cookies
 
 Cookies are essential for using Meta AI and Microsoft Designer to create images.

docs/async_client.md

@@ -145,7 +145,7 @@ async def main():
         provider=g4f.Provider.CopilotAccount
     )
 
-    image = requests.get("https://raw.githubusercontent.com/xtekky/gpt4free/refs/heads/main/docs/cat.jpeg", stream=True).raw
+    image = requests.get("https://raw.githubusercontent.com/xtekky/gpt4free/refs/heads/main/docs/images/cat.jpeg", stream=True).raw
 
     response = await client.chat.completions.create(
         model=g4f.models.default,

docs/authentication.md

@@ -0,0 +1,139 @@
+# G4F Authentication Setup Guide
+
+This documentation explains how to set up Basic Authentication for the GUI and API key authentication for the API when running the G4F server.
+
+## Prerequisites
+
+Before proceeding, ensure you have the following installed:
+- Python 3.x
+- G4F package installed (ensure it is set up and working)
+- Basic knowledge of using environment variables on your operating system
+
+## Steps to Set Up Authentication
+
+### 1. API Key Authentication for Both GUI and API
+
+To secure both the GUI and the API, you'll authenticate using an API key. The API key should be injected via an environment variable and passed to both the GUI (via Basic Authentication) and the API.
+
+#### Steps to Inject the API Key Using Environment Variables:
+
+1. **Set the environment variable** for your API key:
+
+   On Linux/macOS:
+   ```bash
+   export G4F_API_KEY="your-api-key-here"
+   ```
+
+   On Windows (Command Prompt):
+   ```bash
+   set G4F_API_KEY="your-api-key-here"
+   ```
+
+   On Windows (PowerShell):
+   ```bash
+   $env:G4F_API_KEY="your-api-key-here"
+   ```
+
+   Replace `your-api-key-here` with your actual API key.
+
+2. **Run the G4F server with the API key injected**:
+
+   Use the following command to start the G4F server. The API key will be passed to both the GUI and the API:
+
+   ```bash
+   python -m g4f --debug --port 8080 --g4f-api-key $G4F_API_KEY
+   ```
+
+   - `--debug` enables debug mode for more verbose logs.
+   - `--port 8080` specifies the port on which the server will run (you can change this if needed).
+   - `--g4f-api-key` specifies the API key for both the GUI and the API.
+
+#### Example:
+
+```bash
+export G4F_API_KEY="my-secret-api-key"
+python -m g4f --debug --port 8080 --g4f-api-key $G4F_API_KEY
+```
+
+Now, both the GUI and API will require the correct API key for access.
+
+---
+
+### 2. Accessing the GUI with Basic Authentication
+
+The GUI uses **Basic Authentication**, where the **username** can be any value, and the **password** is your API key.
+
+#### Example:
+
+To access the GUI, open your web browser and navigate to `http://localhost:8080/chat/`. You will be prompted for a username and password.
+
+- **Username**: You can use any username (e.g., `user` or `admin`).
+- **Password**: Enter your API key (the same key you set in the `G4F_API_KEY` environment variable).
+
+---
+
+### 3. Python Example for Accessing the API
+
+To interact with the API, you can send requests by including the `g4f-api-key` in the headers. Here's an example of how to do this using the `requests` library in Python.
+
+#### Example Code to Send a Request:
+
+```python
+import requests
+
+url = "http://localhost:8080/v1/chat/completions"
+
+# Body of the request
+body = {
+    "model": "your-model-name",  # Replace with your model name
+    "provider": "your-provider",  # Replace with the provider name
+    "messages": [
+        {
+            "role": "user",
+            "content": "Hello"
+        }
+    ]
+}
+
+# API Key (can be set as an environment variable)
+api_key = "your-api-key-here"  # Replace with your actual API key
+
+# Send the POST request
+response = requests.post(url, json=body, headers={"g4f-api-key": api_key})
+
+# Check the response
+print(response.status_code)
+print(response.json())
+```
+
+In this example:
+- Replace `"your-api-key-here"` with your actual API key.
+- `"model"` and `"provider"` should be replaced with the appropriate model and provider you're using.
+- The `messages` array contains the conversation you want to send to the API.
+
+#### Response:
+
+The response will contain the output of the API request, such as the model's completion or other relevant data, which you can then process in your application.
+
+---
+
+### 4. Testing the Setup
+
+- **Accessing the GUI**: Open a web browser and navigate to `http://localhost:8080/chat/`. The GUI will now prompt you for a username and password. You can enter any username (e.g., `admin`), and for the password, enter the API key you set up in the environment variable.
+  
+- **Accessing the API**: Use the Python code example above to send requests to the API. Ensure the correct API key is included in the `g4f-api-key` header.
+
+---
+
+### 5. Troubleshooting
+
+- **GUI Access Issues**: If you're unable to access the GUI, ensure that you are using the correct API key as the password.
+- **API Access Issues**: If the API is rejecting requests, verify that the `G4F_API_KEY` environment variable is correctly set and passed to the server. You can also check the server logs for more detailed error messages.
+
+---
+
+## Summary
+
+By following the steps above, you will have successfully set up Basic Authentication for the G4F GUI (using any username and the API key as the password) and API key authentication for the API. This ensures that only authorized users can access both the interface and make API requests.
+
+[Return to Home](/)

docs/client.md

@@ -181,7 +181,7 @@ client = Client(
 )
 
 response = client.images.create_variation(
-    image=open("cat.jpg", "rb"),
+    image=open("docs/images/cat.jpg", "rb"),
     model="dall-e-3",
     # Add any other necessary parameters
 )
@@ -235,7 +235,7 @@ client = Client(
 )
 
 image = requests.get("https://raw.githubusercontent.com/xtekky/gpt4free/refs/heads/main/docs/cat.jpeg", stream=True).raw
-# Or: image = open("docs/cat.jpeg", "rb")
+# Or: image = open("docs/images/cat.jpeg", "rb")
 
 response = client.chat.completions.create(
     model=g4f.models.default,

g4f/Provider/Blackbox2.py

@@ -6,6 +6,7 @@ from aiohttp import ClientSession
 
 from ..typing import AsyncResult, Messages
 from .base_provider import AsyncGeneratorProvider, ProviderModelMixin
+from .. import debug
 
 class Blackbox2(AsyncGeneratorProvider, ProviderModelMixin):
     url = "https://www.blackbox.ai"
@@ -13,7 +14,7 @@ class Blackbox2(AsyncGeneratorProvider, ProviderModelMixin):
     working = True
     supports_system_message = True
     supports_message_history = True
-    
+    supports_stream = False
     default_model = 'llama-3.1-70b'
     models = [default_model]
 
@@ -62,8 +63,8 @@ class Blackbox2(AsyncGeneratorProvider, ProviderModelMixin):
                             raise KeyError("'prompt' key not found in the response")
                 except Exception as e:
                     if attempt == max_retries - 1:
-                        yield f"Error after {max_retries} attempts: {str(e)}"
+                        raise RuntimeError(f"Error after {max_retries} attempts: {str(e)}")
                     else:
                         wait_time = delay * (2 ** attempt) + random.uniform(0, 1)
-                        print(f"Attempt {attempt + 1} failed. Retrying in {wait_time:.2f} seconds...")
+                        debug.log(f"Attempt {attempt + 1} failed. Retrying in {wait_time:.2f} seconds...")
                         await asyncio.sleep(wait_time)

g4f/Provider/needs_auth/Gemini.py

@@ -76,8 +76,7 @@ class Gemini(AsyncGeneratorProvider):
         page = await browser.get(f"{cls.url}/app")
         await page.select("div.ql-editor.textarea", 240)
         cookies = {}
-        for c in await page.browser.cookies.get_all():
+        for c in await page.send(nodriver.cdp.network.get_cookies([cls.url])):
-            if c.domain.endswith(".google.com"):
             cookies[c.name] = c.value
         await page.close()
         cls._cookies = cookies
@@ -92,7 +91,6 @@ class Gemini(AsyncGeneratorProvider):
         connector: BaseConnector = None,
         image: ImageType = None,
         image_name: str = None,
-        response_format: str = None,
         return_conversation: bool = False,
         conversation: Conversation = None,
         language: str = "en",
@@ -113,7 +111,7 @@ class Gemini(AsyncGeneratorProvider):
                     async for chunk in cls.nodriver_login(proxy):
                         yield chunk
                 except Exception as e:
-                    raise MissingAuthError('Missing "__Secure-1PSID" cookie', e)
+                    raise MissingAuthError('Missing or invalid "__Secure-1PSID" cookie', e)
             if not cls._snlm0e:
                 if cls._cookies is None or "__Secure-1PSID" not in cls._cookies:
                     raise MissingAuthError('Missing "__Secure-1PSID" cookie')
@@ -153,7 +151,7 @@ class Gemini(AsyncGeneratorProvider):
                 ) as response:
                     await raise_for_status(response)
                     image_prompt = response_part = None
-                    last_content_len = 0
+                    last_content = ""
                     async for line in response.content:
                         try:
                             try:
@@ -171,30 +169,24 @@ class Gemini(AsyncGeneratorProvider):
                                 yield Conversation(response_part[1][0], response_part[1][1], response_part[4][0][0])
                             content = response_part[4][0][1][0]
                         except (ValueError, KeyError, TypeError, IndexError) as e:
-                            print(f"{cls.__name__}:{e.__class__.__name__}:{e}")
+                            debug.log(f"{cls.__name__}:{e.__class__.__name__}:{e}")
                             continue
                         match = re.search(r'\[Imagen of (.*?)\]', content)
                         if match:
                             image_prompt = match.group(1)
                             content = content.replace(match.group(0), '')
-                        yield content[last_content_len:]
+                        pattern = r"http://googleusercontent.com/image_generation_content/\d+"
-                        last_content_len = len(content)
+                        content = re.sub(pattern, "", content)
+                        if last_content and content.startswith(last_content):
+                            yield content[len(last_content):]
+                        else:
+                            yield content
+                        last_content = content
                         if image_prompt:
                             try:
                                 images = [image[0][3][3] for image in response_part[4][0][12][7][0]]
-                            if response_format == "b64_json":
+                                image_prompt = image_prompt.replace("a fake image", "")
                                 yield ImageResponse(images, image_prompt, {"cookies": cls._cookies})
-                            else:
-                                resolved_images = []
-                                preview = []
-                                for image in images:
-                                    async with client.get(image, allow_redirects=False) as fetch:
-                                        image = fetch.headers["location"]
-                                    async with client.get(image, allow_redirects=False) as fetch:
-                                        image = fetch.headers["location"]
-                                    resolved_images.append(image)
-                                    preview.append(image.replace('=s512', '=s200'))
-                                yield ImageResponse(resolved_images, image_prompt, {"orginal_links": images, "preview": preview})
                             except TypeError:
                                 pass
 

g4f/Provider/needs_auth/OpenaiAccount.py

@@ -5,4 +5,8 @@ from .OpenaiChat import OpenaiChat
 class OpenaiAccount(OpenaiChat):
     needs_auth = True
     parent = "OpenaiChat"
-    image_models = ["dall-e"]
+    image_models = ["dall-e-3",  "gpt-4", "gpt-4o"]
+    default_vision_model = "gpt-4o"
+    default_image_model = "dall-e-3"
+    models = [*OpenaiChat.fallback_models, default_image_model]
+    model_aliases = {default_image_model: default_vision_model}

g4f/Provider/needs_auth/OpenaiChat.py

@@ -7,6 +7,7 @@ import json
 import base64
 import time
 import requests
+import random
 from copy import copy
 
 try:
@@ -77,11 +78,8 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
     supports_message_history = True
     supports_system_message = True
     default_model = "auto"
-    default_vision_model = "gpt-4o"
+    fallback_models = [default_model, "gpt-4", "gpt-4o", "gpt-4o-mini", "gpt-4o-canmore", "o1-preview", "o1-mini"]
-    default_image_model = "dall-e-3"
-    fallback_models = [default_model, "gpt-4", "gpt-4o", "gpt-4o-mini", "gpt-4o-canmore", "o1-preview", "o1-mini", default_image_model]
     vision_models = fallback_models
-    image_models = fallback_models
     synthesize_content_type = "audio/mpeg"
 
     _api_key: str = None
@@ -97,7 +95,6 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
                 response.raise_for_status()
                 data = response.json()
                 cls.models = [model.get("slug") for model in data.get("models")]
-                cls.models.append(cls.default_image_model)
             except Exception:
                 cls.models = cls.fallback_models
         return cls.models
@@ -184,7 +181,6 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
             "content": {"content_type": "text", "parts": [message["content"]]},
             "id": str(uuid.uuid4()),
             "create_time": int(time.time()),
-            "id": str(uuid.uuid4()),
             "metadata": {"serialization_metadata": {"custom_symbol_offsets": []}, "system_hints": system_hints},
         } for message in messages]
 
@@ -295,26 +291,27 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
         Raises:
             RuntimeError: If an error occurs during processing.
         """
-        if model == cls.default_image_model:
-            model = cls.default_model
         if cls.needs_auth:
             await cls.login(proxy)
-
         async with StreamSession(
             proxy=proxy,
             impersonate="chrome",
             timeout=timeout
         ) as session:
+            image_request = None
             if not cls.needs_auth:
+                if cls._headers is None:
                     cls._create_request_args(cookies)
-                RequestConfig.proof_token = get_config(cls._headers.get("user-agent"))
                     async with session.get(cls.url, headers=INIT_HEADERS) as response:
                         cls._update_request_args(session)
                         await raise_for_status(response)
+            else:
+                async with session.get(cls.url, headers=cls._headers) as response:
+                    cls._update_request_args(session)
+                    await raise_for_status(response)
                 try:
                     image_request = await cls.upload_image(session, cls._headers, image, image_name) if image else None
                 except Exception as e:
-                image_request = None
                     debug.log("OpenaiChat: Upload image failed")
                     debug.log(f"{e.__class__.__name__}: {e}")
             model = cls.get_model(model)
@@ -348,6 +345,8 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
                         raise MissingAuthError("No arkose token found in .har file")
 
                 if "proofofwork" in chat_requirements:
+                    if RequestConfig.proof_token is None:
+                        RequestConfig.proof_token = get_config(cls._headers.get("user-agent"))
                     proofofwork = generate_proof_token(
                         **chat_requirements["proofofwork"],
                         user_agent=cls._headers.get("user-agent"),
@@ -363,13 +362,22 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
                     "messages": None,
                     "parent_message_id": conversation.message_id,
                     "model": model,
-                    "paragen_cot_summary_display_override": "allow",
+                    "timezone_offset_min":-60,
-                    "history_and_training_disabled": history_disabled and not auto_continue and not return_conversation,
+                    "timezone":"Europe/Berlin",
-                    "conversation_mode": {"kind":"primary_assistant"},
+                    "history_and_training_disabled": history_disabled and not auto_continue and not return_conversation or not cls.needs_auth,
+                    "conversation_mode":{"kind":"primary_assistant","plugin_ids":None},
+                    "force_paragen":False,
+                    "force_paragen_model_slug":"",
+                    "force_rate_limit":False,
+                    "reset_rate_limits":False,
                     "websocket_request_id": str(uuid.uuid4()),
+                    "system_hints": ["search"] if web_search else None,
                     "supported_encodings":["v1"],
-                    "supports_buffering": True,
+                    "conversation_origin":None,
-                    "system_hints": ["search"] if web_search else None
+                    "client_contextual_info":{"is_dark_mode":False,"time_since_loaded":random.randint(20, 500),"page_height":578,"page_width":1850,"pixel_ratio":1,"screen_height":1080,"screen_width":1920},
+                    "paragen_stream_type_override":None,
+                    "paragen_cot_summary_display_override":"allow",
+                    "supports_buffering":True
                 }
                 if conversation.conversation_id is not None:
                     data["conversation_id"] = conversation.conversation_id
@@ -408,7 +416,7 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
                     async for line in response.iter_lines():
                         async for chunk in cls.iter_messages_line(session, line, conversation):
                             yield chunk
-                if not history_disabled:
+                if not history_disabled and RequestConfig.access_token is not None:
                     yield SynthesizeData(cls.__name__, {
                         "conversation_id": conversation.conversation_id,
                         "message_id": conversation.message_id,
@@ -495,6 +503,7 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
             cls._set_api_key(RequestConfig.access_token)
         except NoValidHarFileError:
             if has_nodriver:
+                if RequestConfig.access_token is None:
                     await cls.nodriver_auth(proxy)
             else:
                 raise
@@ -505,7 +514,6 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
         page = browser.main_tab
         def on_request(event: nodriver.cdp.network.RequestWillBeSent):
             if event.request.url == start_url or event.request.url.startswith(conversation_url):
-                RequestConfig.access_request_id = event.request_id
                 RequestConfig.headers = event.request.headers
             elif event.request.url in (backend_url, backend_anon_url):
                 if "OpenAI-Sentinel-Proof-Token" in event.request.headers:
@@ -527,25 +535,25 @@ class OpenaiChat(AsyncGeneratorProvider, ProviderModelMixin):
         await page.send(nodriver.cdp.network.enable())
         page.add_handler(nodriver.cdp.network.RequestWillBeSent, on_request)
         page = await browser.get(cls.url)
-        try:
+        user_agent = await page.evaluate("window.navigator.userAgent")
-            if RequestConfig.access_request_id is not None:
+        await page.select("#prompt-textarea", 240)
-                body = await page.send(get_response_body(RequestConfig.access_request_id))
+        while True:
-                if isinstance(body, tuple) and body:
+            if RequestConfig.access_token:
-                    body = body[0]
+                break
+            body = await page.evaluate("JSON.stringify(window.__remixContext)")
             if body:
                 match = re.search(r'"accessToken":"(.*?)"', body)
                 if match:
                     RequestConfig.access_token = match.group(1)
-        except KeyError:
+                    break
-            pass
+            await asyncio.sleep(1)
-        for c in await page.send(nodriver.cdp.network.get_cookies([cls.url])):
-            RequestConfig.cookies[c.name] = c.value
-        user_agent = await page.evaluate("window.navigator.userAgent")
-        await page.select("#prompt-textarea", 240)
         while True:
             if RequestConfig.proof_token:
                 break
             await asyncio.sleep(1)
+        RequestConfig.data_build = await page.evaluate("document.documentElement.getAttribute('data-build')")
+        for c in await page.send(nodriver.cdp.network.get_cookies([cls.url])):
+            RequestConfig.cookies[c.name] = c.value
         await page.close()
         cls._create_request_args(RequestConfig.cookies, RequestConfig.headers, user_agent=user_agent)
         cls._set_api_key(RequestConfig.access_token)

g4f/Provider/openai/har_file.py

@@ -25,7 +25,6 @@ conversation_url = "https://chatgpt.com/c/"
 class RequestConfig:
     cookies: dict = None
     headers: dict = None
-    access_request_id: str = None
     access_token: str = None
     proof_token: list = None
     turnstile_token: str = None
@@ -33,6 +32,7 @@ class RequestConfig:
     arkose_token: str = None
     headers: dict = {}
     cookies: dict = {}
+    data_build: str = "prod-697873d7e78bb14df6e13af3a91fa237cc4db415"
 
 class arkReq:
     def __init__(self, arkURL, arkBx, arkHeader, arkBody, arkCookies, userAgent):

g4f/Provider/openai/new.py

@@ -14,6 +14,8 @@ from datetime import (
     timezone
 )
 
+from .har_file import RequestConfig
+
 cores       = [16, 24, 32]
 screens     = [3000, 4000, 6000]
 maxAttempts = 500000
@@ -386,7 +388,7 @@ def get_config(user_agent):
         random.random(),
         user_agent,
         None,
-        "prod-0b673b9a04fb6983c1417b587f2f31173eafa605", #document.documentElement.getAttribute("data-build"),
+        RequestConfig.data_build, #document.documentElement.getAttribute("data-build"),
         "en-US",
         "en-US,es-US,en,es",
         0,
@@ -396,7 +398,8 @@ def get_config(user_agent):
         time.perf_counter(),
         str(uuid.uuid4()),
         "",
-        8
+        8,
+        int(time.time()),
     ]
     
     return config

g4f/__main__.py

@@ -0,0 +1,9 @@
+from __future__ import annotations
+
+from .cli import get_api_parser, run_api_args
+
+parser = get_api_parser()
+args = parser.parse_args()
+if args.gui is None:
+    args.gui = True
+run_api_args(args)

g4f/api/__init__.py

@@ -23,7 +23,7 @@ from starlette.status import (
     HTTP_500_INTERNAL_SERVER_ERROR,
 )
 from fastapi.encoders import jsonable_encoder
-from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
+from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials, HTTPBasic
 from fastapi.middleware.cors import CORSMiddleware
 from starlette.responses import FileResponse
 from pydantic import BaseModel, Field
@@ -50,7 +50,7 @@ logger = logging.getLogger(__name__)
 
 DEFAULT_PORT = 1337
 
-def create_app(g4f_api_key: str = None):
+def create_app():
     app = FastAPI()
 
     # Add CORS middleware
@@ -62,7 +62,7 @@ def create_app(g4f_api_key: str = None):
         allow_headers=["*"],
     )
 
-    api = Api(app, g4f_api_key=g4f_api_key)
+    api = Api(app)
 
     if AppConfig.gui:
         @app.get("/")
@@ -86,9 +86,14 @@ def create_app(g4f_api_key: str = None):
 
     return app
 
-def create_app_debug(g4f_api_key: str = None):
+def create_app_debug():
     g4f.debug.logging = True
-    return create_app(g4f_api_key)
+    return create_app()
+
+def create_app_with_gui_and_debug():
+    g4f.debug.logging = True
+    AppConfig.gui = True
+    return create_app()
 
 class ChatCompletionsConfig(BaseModel):
     messages: Messages = Field(examples=[[{"role": "system", "content": ""}, {"role": "user", "content": ""}]])
@@ -112,7 +117,7 @@ class ImageGenerationConfig(BaseModel):
     prompt: str
     model: Optional[str] = None
     provider: Optional[str] = None
-    response_format: str = "url"
+    response_format: Optional[str] = None
     api_key: Optional[str] = None
     proxy: Optional[str] = None
 
@@ -156,8 +161,8 @@ class ErrorResponse(Response):
         return cls(format_exception(exception, config), status_code)
 
     @classmethod
-    def from_message(cls, message: str, status_code: int = HTTP_500_INTERNAL_SERVER_ERROR):
+    def from_message(cls, message: str, status_code: int = HTTP_500_INTERNAL_SERVER_ERROR, headers: dict = None):
-        return cls(format_exception(message), status_code)
+        return cls(format_exception(message), status_code, headers=headers)
 
     def render(self, content) -> bytes:
         return str(content).encode(errors="ignore")
@@ -184,26 +189,57 @@ def set_list_ignored_providers(ignored: list[str]):
     list_ignored_providers = ignored
 
 class Api:
-    def __init__(self, app: FastAPI, g4f_api_key=None) -> None:
+    def __init__(self, app: FastAPI) -> None:
         self.app = app
         self.client = AsyncClient()
-        self.g4f_api_key = g4f_api_key
         self.get_g4f_api_key = APIKeyHeader(name="g4f-api-key")
         self.conversations: dict[str, dict[str, BaseConversation]] = {}
 
     security = HTTPBearer(auto_error=False)
+    basic_security = HTTPBasic()
+
+    async def get_username(self, request: Request):
+        credentials = await self.basic_security(request)
+        current_password_bytes = credentials.password.encode()
+        is_correct_password = secrets.compare_digest(
+            current_password_bytes, AppConfig.g4f_api_key.encode()
+        )
+        if not is_correct_password:
+            raise HTTPException(
+                status_code=HTTP_401_UNAUTHORIZED,
+                detail="Incorrect username or password",
+                headers={"WWW-Authenticate": "Basic"},
+            )
+        return credentials.username
 
     def register_authorization(self):
+        if AppConfig.g4f_api_key:
+            print(f"Register authentication key: {''.join(['*' for _ in range(len(AppConfig.g4f_api_key))])}")
         @self.app.middleware("http")
         async def authorization(request: Request, call_next):
-            if self.g4f_api_key and request.url.path not in ("/", "/v1"):
+            if AppConfig.g4f_api_key is not None:
                 try:
                     user_g4f_api_key = await self.get_g4f_api_key(request)
-                except HTTPException as e:
+                except HTTPException:
-                    if e.status_code == 403:
+                    user_g4f_api_key = None
+                if request.url.path.startswith("/v1"):
+                    if user_g4f_api_key is None:
                         return ErrorResponse.from_message("G4F API key required", HTTP_401_UNAUTHORIZED)
-                if not secrets.compare_digest(self.g4f_api_key, user_g4f_api_key):
+                    if not secrets.compare_digest(AppConfig.g4f_api_key, user_g4f_api_key):
                         return ErrorResponse.from_message("Invalid G4F API key", HTTP_403_FORBIDDEN)
+                else:
+                    path = request.url.path
+                    if user_g4f_api_key is not None and path.startswith("/images/"):
+                        if not secrets.compare_digest(AppConfig.g4f_api_key, user_g4f_api_key):
+                            return ErrorResponse.from_message("Invalid G4F API key", HTTP_403_FORBIDDEN)
+                    elif path.startswith("/backend-api/") or path.startswith("/images/") or path.startswith("/chat/") and path != "/chat/":
+                        try:
+                            username = await self.get_username(request)
+                        except HTTPException as e:
+                            return ErrorResponse.from_message(e.detail, e.status_code, e.headers)
+                        response = await call_next(request)
+                        response.headers["X-Username"] = username
+                        return response
             return await call_next(request)
 
     def register_validation_exception_handler(self):
@@ -512,8 +548,12 @@ def run_api(
         host, port = bind.split(":")
     if port is None:
         port = DEFAULT_PORT
+    if AppConfig.gui and debug:
+        method = "create_app_with_gui_and_debug"
+    else:
+        method = "create_app_debug" if debug else "create_app"
     uvicorn.run(
-        f"g4f.api:create_app{'_debug' if debug else ''}", 
+        f"g4f.api:{method}", 
         host=host, 
         port=int(port), 
         workers=workers, 

g4f/cli.py

@@ -1,19 +1,18 @@
 from __future__ import annotations
 
 import argparse
+from argparse import ArgumentParser
 
 from g4f import Provider
 from g4f.gui.run import gui_parser, run_gui_args
 import g4f.cookies
 
-def main():
+def get_api_parser():
-    parser = argparse.ArgumentParser(description="Run gpt4free")
+    api_parser = ArgumentParser(description="Run the API and GUI")
-    subparsers = parser.add_subparsers(dest="mode", help="Mode to run the g4f in.")
-    api_parser = subparsers.add_parser("api")
     api_parser.add_argument("--bind", default=None, help="The bind string. (Default: 0.0.0.0:1337)")
-    api_parser.add_argument("--port", default=None, help="Change the port of the server.")
+    api_parser.add_argument("--port", "-p", default=None, help="Change the port of the server.")
     api_parser.add_argument("--debug", "-d", action="store_true", help="Enable verbose logging.")
-    api_parser.add_argument("--gui", "-g", default=False, action="store_true", help="Add gui to the api.")
+    api_parser.add_argument("--gui", "-g", default=None, action="store_true", help="Add gui to the api.")
     api_parser.add_argument("--model", default=None, help="Default model for chat completion. (incompatible with --reload and --workers)")
     api_parser.add_argument("--provider", choices=[provider.__name__ for provider in Provider.__providers__ if provider.working],
                             default=None, help="Default provider for chat completion. (incompatible with --reload and --workers)")
@@ -29,6 +28,12 @@ def main():
     api_parser.add_argument("--cookie-browsers", nargs="+", choices=[browser.__name__ for browser in g4f.cookies.browsers],
                             default=[], help="List of browsers to access or retrieve cookies from. (incompatible with --reload and --workers)")
     api_parser.add_argument("--reload", action="store_true", help="Enable reloading.")
+    return api_parser
+
+def main():
+    parser = argparse.ArgumentParser(description="Run gpt4free")
+    subparsers = parser.add_subparsers(dest="mode", help="Mode to run the g4f in.")
+    subparsers.add_parser("api", parents=[get_api_parser()], add_help=False)
     subparsers.add_parser("gui", parents=[gui_parser()], add_help=False)
 
     args = parser.parse_args()

g4f/client/__init__.py

@@ -292,6 +292,7 @@ class Images:
         if proxy is None:
             proxy = self.client.proxy
 
+        e = None
         response = None
         if isinstance(provider_handler, IterListProvider):
             for provider in provider_handler.providers:
@@ -300,7 +301,7 @@ class Images:
                     if response is not None:
                         provider_name = provider.__name__
                         break
-                except (MissingAuthError, NoValidHarFileError) as e:
+                except Exception as e:
                     debug.log(f"Image provider {provider.__name__}: {e}")
         else:
             response = await self._generate_image_response(provider_handler, provider_name, model, prompt, **kwargs)
@@ -314,6 +315,8 @@ class Images:
                 provider_name
             )
         if response is None:
+            if e is not None:
+                raise e
             raise NoImageResponseError(f"No image response from {provider_name}")
         raise NoImageResponseError(f"Unexpected response type: {type(response)}")
 
@@ -362,7 +365,7 @@ class Images:
         image: ImageType,
         model: str = None,
         provider: Optional[ProviderType] = None,
-        response_format: str = "url",
+        response_format: Optional[str] = None,
         **kwargs
     ) -> ImagesResponse:
         return asyncio.run(self.async_create_variation(
@@ -374,7 +377,7 @@ class Images:
         image: ImageType,
         model: Optional[str] = None,
         provider: Optional[ProviderType] = None,
-        response_format: str = "url",
+        response_format: Optional[str] = None,
         proxy: Optional[str] = None,
         **kwargs
     ) -> ImagesResponse:
@@ -384,6 +387,7 @@ class Images:
             proxy = self.client.proxy
         prompt = "create a variation of this image"
 
+        e = None
         response = None
         if isinstance(provider_handler, IterListProvider):
             # File pointer can be read only once, so we need to convert it to bytes
@@ -394,7 +398,7 @@ class Images:
                     if response is not None:
                         provider_name = provider.__name__
                         break
-                except (MissingAuthError, NoValidHarFileError) as e:
+                except Exception as e:
                     debug.log(f"Image provider {provider.__name__}: {e}")
         else:
             response = await self._generate_image_response(provider_handler, provider_name, model, prompt, image=image, **kwargs)
@@ -402,10 +406,11 @@ class Images:
         if isinstance(response, ImageResponse):
             return await self._process_image_response(response, response_format, proxy, model, provider_name)
         if response is None:
+            if e is not None:
+                raise e
             raise NoImageResponseError(f"No image response from {provider_name}")
         raise NoImageResponseError(f"Unexpected response type: {type(response)}")
 
-
     async def _process_image_response(
         self,
         response: ImageResponse,
@@ -414,21 +419,21 @@ class Images:
         model: Optional[str] = None,
         provider: Optional[str] = None
     ) -> ImagesResponse:
+        last_provider = get_last_provider(True)
         if response_format == "url":
             # Return original URLs without saving locally
             images = [Image.construct(url=image, revised_prompt=response.alt) for image in response.get_list()]
-        elif response_format == "b64_json":
+        else:
-            images = await copy_images(response.get_list(), response.options.get("cookies"), proxy)
+            # Save locally for None (default) case
+            images = await copy_images(response.get_list(), response.get("cookies"), proxy)
+            if response_format == "b64_json":
                 async def process_image_item(image_file: str) -> Image:
                     with open(os.path.join(images_dir, os.path.basename(image_file)), "rb") as file:
                         image_data = base64.b64encode(file.read()).decode()
                         return Image.construct(b64_json=image_data, revised_prompt=response.alt)
                 images = await asyncio.gather(*[process_image_item(image) for image in images])
             else:
-            # Save locally for None (default) case
-            images = await copy_images(response.get_list(), response.options.get("cookies"), proxy)
                 images = [Image.construct(url=f"/images/{os.path.basename(image)}", revised_prompt=response.alt) for image in images]
-        last_provider = get_last_provider(True)
         return ImagesResponse.construct(
             created=int(time.time()),
             data=images,
@@ -529,7 +534,7 @@ class AsyncImages(Images):
         image: ImageType,
         model: str = None,
         provider: ProviderType = None,
-        response_format: str = "url",
+        response_format: Optional[str] = None,
         **kwargs
     ) -> ImagesResponse:
         return await self.async_create_variation(

g4f/gui/client/static/js/chat.v1.js

@@ -744,7 +744,11 @@ const delete_conversation = async (conversation_id) => {
 };
 
 const set_conversation = async (conversation_id) => {
+    try {
         history.pushState({}, null, `/chat/${conversation_id}`);
+    } catch (e) {
+        console.error(e);
+    }
     window.conversation_id = conversation_id;
 
     await clear_conversation();
@@ -898,7 +902,11 @@ async function add_conversation(conversation_id, content) {
             items: [],
         });
     }
+    try {
         history.pushState({}, null, `/chat/${conversation_id}`);
+    } catch (e) {
+        console.error(e);
+    }
 }
 
 async function save_system_message() {
@@ -1287,23 +1295,29 @@ async function on_api() {
 
     register_settings_storage();
 
+    try {
         models = await api("models");
         models.forEach((model) => {
             let option = document.createElement("option");
             option.value = option.text = model;
             modelSelect.appendChild(option);
         });
-
         providers = await api("providers")
         Object.entries(providers).forEach(([provider, label]) => {
             let option = document.createElement("option");
             option.value = provider;
             option.text = label;
             providerSelect.appendChild(option);
-    })
+        });
+        await load_provider_models(appStorage.getItem("provider"));
+    } catch (e) {
+        console.error(e)
+        if (document.location.pathname == "/chat/") {
+            document.location.href = `/chat/error`;
+        }
+    }
 
     await load_settings_storage()
-    await load_provider_models(appStorage.getItem("provider"));
 
     const hide_systemPrompt = document.getElementById("hide-systemPrompt")
     const slide_systemPrompt_icon = document.querySelector(".slide-systemPrompt i");
@@ -1465,7 +1479,7 @@ async function api(ressource, args=null, file=null, message_id=null) {
     const url = `/backend-api/v2/${ressource}`;
     const headers = {};
     if (api_key) {
-        headers.authorization = `Bearer ${api_key}`;
+        headers.x_api_key = api_key;
     }
     if (ressource == "conversation") {
         let body = JSON.stringify(args);

g4f/gui/server/api.py

@@ -110,8 +110,10 @@ class Api:
     def _create_response_stream(self, kwargs: dict, conversation_id: str, provider: str, download_images: bool = True) -> Iterator:
         def log_handler(text: str):
             debug.logs.append(text)
+            if debug.logging:
                 print(text)
         debug.log_handler = log_handler
+        proxy = os.environ.get("G4F_PROXY")
         try:
             result = ChatCompletion.create(**kwargs)
             first = True
@@ -139,7 +141,7 @@ class Api:
                     elif isinstance(chunk, ImageResponse):
                         images = chunk
                         if download_images:
-                            images = asyncio.run(copy_images(chunk.get_list(), chunk.options.get("cookies")))
+                            images = asyncio.run(copy_images(chunk.get_list(), chunk.get("cookies"), proxy))
                             images = ImageResponse(images, chunk.alt)
                         yield self._format_json("content", str(images))
                     elif isinstance(chunk, SynthesizeData):

g4f/gui/server/backend.py

@@ -153,7 +153,7 @@ class Backend_Api(Api):
         return response
 
     def get_provider_models(self, provider: str):
-        api_key = None if request.authorization is None else request.authorization.token
+        api_key = request.headers.get("x_api_key")
         models = super().get_provider_models(provider, api_key)
         if models is None:
             return "Provider not found", 404

g4f/image.py

@@ -7,8 +7,7 @@ import uuid
 from io import BytesIO
 import base64
 import asyncio
-from aiohttp import ClientSession
+from aiohttp import ClientSession, ClientError
-
 try:
     from PIL.Image import open as open_image, new as new_image
     from PIL.Image import FLIP_LEFT_RIGHT, ROTATE_180, ROTATE_270, ROTATE_90
@@ -20,6 +19,7 @@ from .typing import ImageType, Union, Image, Optional, Cookies
 from .errors import MissingRequirementsError
 from .providers.response import ResponseType
 from .requests.aiohttp import get_connector
+from . import debug
 
 ALLOWED_EXTENSIONS = {'png', 'jpg', 'jpeg', 'gif', 'webp', 'svg'}
 
@@ -277,12 +277,14 @@ def ensure_images_dir():
     if not os.path.exists(images_dir):
         os.makedirs(images_dir)
 
-async def copy_images(images: list[str], cookies: Optional[Cookies] = None, proxy: Optional[str] = None):
+async def copy_images(
+    images: list[str],
+    cookies: Optional[Cookies] = None,
+    proxy: Optional[str] = None
+):
     ensure_images_dir()
     async with ClientSession(
-        connector=get_connector(
+        connector=get_connector(proxy=proxy),
-            proxy=os.environ.get("G4F_PROXY") if proxy is None else proxy
-        ),
         cookies=cookies
     ) as session:
         async def copy_image(image: str) -> str:
@@ -291,10 +293,15 @@ async def copy_images(images: list[str], cookies: Optional[Cookies] = None, prox
                 with open(target, "wb") as f:
                     f.write(extract_data_uri(image))
             else:
+                try:
                     async with session.get(image) as response:
+                        response.raise_for_status()
                         with open(target, "wb") as f:
                             async for chunk in response.content.iter_chunked(4096):
                                 f.write(chunk)
+                except ClientError as e:
+                    debug.log(f"copy_images failed: {e.__class__.__name__}: {e}")
+                    return image
             with open(target, "rb") as f:
                 extension = is_accepted_format(f.read(12)).split("/")[-1]
                 extension = "jpg" if extension == "jpeg" else extension

g4f/requests/raise_for_status.py

@@ -18,7 +18,7 @@ def is_cloudflare(text: str) -> bool:
     return '<div id="cf-please-wait">' in text or "<title>Just a moment...</title>" in text
 
 def is_openai(text: str) -> bool:
-    return "<p>Unable to load site</p>" in text
+    return "<p>Unable to load site</p>" in text or 'id="challenge-error-text"' in text
 
 async def raise_for_status_async(response: Union[StreamResponse, ClientResponse], message: str = None):
     if response.status in (429, 402):
@@ -27,8 +27,10 @@ async def raise_for_status_async(response: Union[StreamResponse, ClientResponse]
     if response.status == 403 and is_cloudflare(message):
         raise CloudflareError(f"Response {response.status}: Cloudflare detected")
     elif response.status == 403 and is_openai(message):
-        raise ResponseStatusError(f"Response {response.status}: Bot are detected")
+        raise ResponseStatusError(f"Response {response.status}: OpenAI Bot detected")
     elif not response.ok:
+        if "<html>" in message:
+            message = "HTML content"
         raise ResponseStatusError(f"Response {response.status}: {message}")
 
 def raise_for_status(response: Union[Response, StreamResponse, ClientResponse, RequestsResponse], message: str = None):

setup.py

@@ -8,6 +8,10 @@ here = os.path.abspath(os.path.dirname(__file__))
 with codecs.open(os.path.join(here, 'README.md'), encoding='utf-8') as fh:
     long_description = '\n' + fh.read()
 
+long_description = long_description.replace("[!NOTE]", "")
+long_description = long_description.replace("(docs/images/", "(https://raw.githubusercontent.com/xtekky/gpt4free/refs/heads/main/docs/images/")
+long_description = long_description.replace("(docs/", "(https://github.com/xtekky/gpt4free/blob/main/docs/")
+
 INSTALL_REQUIRE = [
     "requests",
     "aiohttp",