Support "Think Deeper" in Copilot

Improve Documentation
2025-10-31 19:42:45 +08:00 · 2025-03-30 15:40:46 +02:00
parent c471c53671
commit 8c2c46b20f
14 changed files with 189 additions and 82 deletions
--- a/README.md
+++ b/README.md
@@ -97,31 +97,31 @@ Is your site on this repository and you want to take it down? Send an email to t
 1. **Install Docker:** [Download and install Docker](https://docs.docker.com/get-docker/).
 2. **Set Up Directories:** Before running the container, make sure the necessary data directories exist or can be created. For example, you can create and set ownership on these directories by running: 
 ```bash
-   mkdir -p ${PWD}/har_and_cookies ${PWD}/generated_images
-   sudo chown -R 1200:1201 ${PWD}/har_and_cookies ${PWD}/generated_images
+mkdir -p ${PWD}/har_and_cookies ${PWD}/generated_images
+sudo chown -R 1200:1201 ${PWD}/har_and_cookies ${PWD}/generated_images
 ```
 3. **Run the Docker Container:** Use the following commands to pull the latest image and start the container (Only x64):
 ```bash
-   docker pull hlohaus789/g4f
-   docker run -p 8080:8080 -p 7900:7900 \
-     --shm-size="2g" \
-     -v ${PWD}/har_and_cookies:/app/har_and_cookies \
-     -v ${PWD}/generated_images:/app/generated_images \
-     hlohaus789/g4f:latest
+docker pull hlohaus789/g4f
+docker run -p 8080:8080 -p 7900:7900 \
+  --shm-size="2g" \
+  -v ${PWD}/har_and_cookies:/app/har_and_cookies \
+  -v ${PWD}/generated_images:/app/generated_images \
+  hlohaus789/g4f:latest
 ```

 4. **Running the Slim Docker Image:** And use the following commands to run the Slim Docker image. This command also updates the `g4f` package at startup and installs any additional dependencies: (x64 and arm64)
 ```bash
-  mkdir -p ${PWD}/har_and_cookies ${PWD}/generated_images
-  chown -R 1000:1000 ${PWD}/har_and_cookies ${PWD}/generated_images
-	docker run \
-	  -p 1337:1337 \
-	  -v ${PWD}/har_and_cookies:/app/har_and_cookies \
-	  -v ${PWD}/generated_images:/app/generated_images \
-	  hlohaus789/g4f:latest-slim \
-	  rm -r -f /app/g4f/ \
-	  && pip install -U g4f[slim] \
-	  && python -m g4f --debug
+mkdir -p ${PWD}/har_and_cookies ${PWD}/generated_images
+chown -R 1000:1000 ${PWD}/har_and_cookies ${PWD}/generated_images
+docker run \
+  -p 1337:1337 \
+  -v ${PWD}/har_and_cookies:/app/har_and_cookies \
+  -v ${PWD}/generated_images:/app/generated_images \
+  hlohaus789/g4f:latest-slim \
+  rm -r -f /app/g4f/ \
+  && pip install -U g4f[slim] \
+  && python -m g4f --debug
 ```
 
 5. **Access the Client Interface:**
@@ -248,6 +248,7 @@ Run the Web UI on your smartphone for easy access on the go. Check out the dedic
   - **File API from G4F:** [/docs/file](docs/file.md)
   - **PydanticAI and LangChain Integration for G4F:** [/docs/pydantic_ai](docs/pydantic_ai.md)
   - **Legacy API with python modules:** [/docs/legacy](docs/legacy.md)
+   - **G4F - Media Documentation** [/docs/media](/docs/meda.md) *(New)*

 ---

--- a/docs/authentication.md
+++ b/docs/authentication.md
@@ -126,7 +126,7 @@ providers = {
 }

 for provider_name, api_key in providers.items():
-    client = Client(provider=f"g4f.Provider.{provider_name}", api_key=api_key)
+    client = Client(provider=provider_name, api_key=api_key)
    response = client.chat.completions.create(
        model="claude-3.5-sonnet",
        messages=[{"role": "user", "content": f"Hello from {provider_name}!"}]
@@ -144,16 +144,22 @@ for provider_name, api_key in providers.items():
   - Firefox: **Storage** → **Cookies**  

 ```python
+from g4f.client import Client
 from g4f.Provider import Gemini

-# Initialize with cookies
+# Using with cookies
 client = Client(
    provider=Gemini,
+)
+response = client.chat.completions.create(
+    model="", # Default model
+    messages="Hello Google",
    cookies={
        "__Secure-1PSID": "your_cookie_value_here",
        "__Secure-1PSIDTS": "timestamp_value_here"
    }
 )
+print(f"Gemini: {response.choices[0].message.content}")
 ```

 ---
--- a/docs/gui.md
+++ b/docs/gui.md
@@ -69,9 +69,9 @@ The G4F GUI is a self-contained, user-friendly interface designed for interactin
   - **Basic Authentication**  
     You can set a password for Basic Authentication using the `--g4f-api-key` argument when starting the web server.

-### 9. **Continue Button (ChatGPT & HuggingChat)**
+### 9. **Continue Button**
   - **Automatic Detection of Truncated Responses**  
-     When using **ChatGPT** or **HuggingChat** providers, responses may occasionally be cut off or truncated.  
+     When using providers, responses may occasionally be cut off or truncated.  
   - **Continue Button**  
     If the GUI detects that the response ended abruptly, a **Continue** button appears directly below the truncated message. Clicking this button sends a follow-up request to the same provider and model, retrieving the rest of the message.  
   - **Seamless Conversation Flow**  
@@ -154,7 +154,7 @@ http://localhost:8080/chat/
   - **Text/Code:** The generated response appears in the conversation window.  
   - **Images:** Generated images are displayed as thumbnails. Click on any thumbnail to view it in full size within the lightbox.

-5. **Continue Button (ChatGPT & HuggingChat)**  
+5. **Continue Button**  
   - If a response is truncated, a **Continue** button will appear under the last message. Clicking it asks the same provider to continue the response from where it ended.

 6. **Manage Conversations**  
--- a/docs/media.md
+++ b/docs/media.md
@@ -175,7 +175,7 @@ from g4f.Provider import HuggingFaceMedia
 async def main():
    client = AsyncClient(
        provider=HuggingFaceMedia,
-        api_key="hf_***"  # Your API key here
+        api_key=os.getenv("HF_TOKEN") # Your API key here
    )

    video_models = client.models.get_video()
@@ -214,7 +214,7 @@ from g4f.Provider import HuggingFaceMedia
 async def main():
    client = AsyncClient(
        provider=HuggingFaceMedia,
-        api_key=os.getenv("HUGGINGFACE_API_KEY")  # Your API key here
+        api_key=os.getenv("HF_TOKEN")  # Your API key here
    )

    video_models = client.models.get_video()
--- a/g4f/Provider/Cloudflare.py
+++ b/g4f/Provider/Cloudflare.py
@@ -83,9 +83,9 @@ class Cloudflare(AsyncGeneratorProvider, ProviderModelMixin, AuthFileMixin):
            pass
        data = {
            "messages": [{
-                "role":"user",
+                **message,
                "content": message["content"] if isinstance(message["content"], str) else "",
-                "parts": [{"type":"text", "text":message["content"]}] if isinstance(message["content"], str) else message["content"]} for message in messages],
+                "parts": [{"type":"text", "text":message["content"]}] if isinstance(message["content"], str) else message} for message in messages],
            "lora": None,
            "model": model,
            "max_tokens": max_tokens,
--- a/g4f/Provider/Copilot.py
+++ b/g4f/Provider/Copilot.py
@@ -24,7 +24,7 @@ from .openai.har_file import get_headers, get_har_files
 from ..typing import CreateResult, Messages, MediaListType
 from ..errors import MissingRequirementsError, NoValidHarFileError, MissingAuthError
 from ..requests.raise_for_status import raise_for_status
-from ..providers.response import BaseConversation, JsonConversation, RequestLogin, ImageResponse
+from ..providers.response import BaseConversation, JsonConversation, RequestLogin, ImageResponse, FinishReason, SuggestedFollowups
 from ..providers.asyncio import get_running_loop
 from ..tools.media import merge_media
 from ..requests import get_nodriver
@@ -46,10 +46,13 @@ class Copilot(AbstractProvider, ProviderModelMixin):
    supports_stream = True
    
    default_model = "Copilot"
-    models = [default_model]
+    models = [default_model, "Think Deeper"]
    model_aliases = {
        "gpt-4": default_model,
-        "o1": default_model,
+        "gpt-4o": default_model,
+        "o1": "Think Deeper",
+        "reasoning": "Think Deeper",
+        "dall-e-3": default_model
    }

    websocket_url = "wss://copilot.microsoft.com/c/api/chat?api-version=2"
@@ -75,10 +78,10 @@ class Copilot(AbstractProvider, ProviderModelMixin):
    ) -> CreateResult:
        if not has_curl_cffi:
            raise MissingRequirementsError('Install or update "curl_cffi" package | pip install -U curl_cffi')
-
+        model = cls.get_model(model)
        websocket_url = cls.websocket_url
        headers = None
-        if cls.needs_auth or media is not None:
+        if cls._access_token:
            if api_key is not None:
                cls._access_token = api_key
            if cls._access_token is None:
@@ -163,6 +166,7 @@ class Copilot(AbstractProvider, ProviderModelMixin):
            #         "token": clarity_token,
            #         "method":"clarity"
            #     }).encode(), CurlWsFlag.TEXT)
+            wss.send(json.dumps({"event":"setOptions","supportedCards":["weather","local","image","sports","video","ads","finance"],"ads":{"supportedTypes":["multimedia","product","tourActivity","propertyPromotion","text"]}}));
            wss.send(json.dumps({
                "event": "send",
                "conversationId": conversation_id,
@@ -170,7 +174,7 @@ class Copilot(AbstractProvider, ProviderModelMixin):
                    "type": "text",
                    "text": prompt,
                }],
-                "mode": "chat"
+                "mode": "reasoning" if "Think" in model else "chat",
            }).encode(), CurlWsFlag.TEXT)

            is_started = False
@@ -193,6 +197,10 @@ class Copilot(AbstractProvider, ProviderModelMixin):
                    elif msg.get("event") == "imageGenerated":
                        yield ImageResponse(msg.get("url"), image_prompt, {"preview": msg.get("thumbnailUrl")})
                    elif msg.get("event") == "done":
+                        yield FinishReason("stop")
+                        break
+                    elif msg.get("event") == "suggestedFollowups":
+                        yield SuggestedFollowups(msg.get("suggestions"))
                        break
                    elif msg.get("event") == "replaceText":
                        yield msg.get("text")
--- a/g4f/Provider/DuckDuckGo.py
+++ b/g4f/Provider/DuckDuckGo.py
@@ -0,0 +1,85 @@
+from __future__ import annotations
+
+import asyncio
+
+try:
+    from duckduckgo_search import DDGS
+    from duckduckgo_search.exceptions import DuckDuckGoSearchException, RatelimitException, ConversationLimitException
+    has_requirements = True
+except ImportError:
+    has_requirements = False
+try:
+    import nodriver
+    has_nodriver = True
+except ImportError:
+    has_nodriver = False
+
+from ..typing import AsyncResult, Messages
+from ..requests import get_nodriver
+from .base_provider import AsyncGeneratorProvider, ProviderModelMixin
+from .helper import get_last_user_message
+
+class DuckDuckGo(AsyncGeneratorProvider, ProviderModelMixin):
+    label = "Duck.ai (duckduckgo_search)"
+    url = "https://duckduckgo.com/aichat"
+    api_base = "https://duckduckgo.com/duckchat/v1/"
+    
+    working = False
+    supports_stream = True
+    supports_system_message = True
+    supports_message_history = True
+    
+    default_model = "gpt-4o-mini"
+    models = [default_model, "meta-llama/Llama-3.3-70B-Instruct-Turbo", "claude-3-haiku-20240307", "o3-mini", "mistralai/Mistral-Small-24B-Instruct-2501"]
+
+    ddgs: DDGS = None
+
+    model_aliases = {
+        "gpt-4": "gpt-4o-mini",
+        "llama-3.3-70b": "meta-llama/Llama-3.3-70B-Instruct-Turbo",
+        "claude-3-haiku": "claude-3-haiku-20240307",
+        "mixtral-small-24b": "mistralai/Mistral-Small-24B-Instruct-2501",
+    }
+
+    @classmethod
+    async def create_async_generator(
+        cls,
+        model: str,
+        messages: Messages,
+        proxy: str = None,
+        timeout: int = 60,
+        **kwargs
+    ) -> AsyncResult:
+        if not has_requirements:
+            raise ImportError("duckduckgo_search is not installed. Install it with `pip install duckduckgo-search`.")
+        if cls.ddgs is None:
+            cls.ddgs = DDGS(proxy=proxy, timeout=timeout)
+            if has_nodriver:
+                await cls.nodriver_auth(proxy=proxy)
+        model = cls.get_model(model)
+        for chunk in cls.ddgs.chat_yield(get_last_user_message(messages), model, timeout):
+            yield chunk
+
+    @classmethod
+    async def nodriver_auth(cls, proxy: str = None):
+        browser, stop_browser = await get_nodriver(proxy=proxy)
+        try:
+            page = browser.main_tab
+            def on_request(event: nodriver.cdp.network.RequestWillBeSent, page=None):
+                if cls.api_base in event.request.url:
+                    if "X-Vqd-4" in event.request.headers:
+                        cls.ddgs._chat_vqd = event.request.headers["X-Vqd-4"]
+                    if "X-Vqd-Hash-1" in event.request.headers:
+                        cls.ddgs._chat_vqd_hash = event.request.headers["X-Vqd-Hash-1"]
+                    if "F-Fe-Version" in event.request.headers:
+                        cls.ddgs._chat_xfe = event.request.headers["F-Fe-Version" ]
+            await page.send(nodriver.cdp.network.enable())
+            page.add_handler(nodriver.cdp.network.RequestWillBeSent, on_request)
+            page = await browser.get(cls.url)
+            while True:
+                if cls.ddgs._chat_vqd:
+                    break
+                await asyncio.sleep(1)
+            await page.close()
+        finally:
+            stop_browser()
--- a/g4f/Provider/init.py
+++ b/g4f/Provider/init.py
@@ -40,6 +40,7 @@ try:
    from .Copilot              import Copilot
    from .DDG                  import DDG
    from .DeepInfraChat        import DeepInfraChat
+    from .DuckDuckGo           import DuckDuckGo
    from .Dynaspark            import Dynaspark
 except ImportError as e:
    debug.error("Providers not loaded (A-D):", e)
--- a/g4f/Provider/needs_auth/CopilotAccount.py
+++ b/g4f/Provider/needs_auth/CopilotAccount.py
@@ -19,11 +19,6 @@ class CopilotAccount(AsyncAuthedProvider, Copilot):
    parent = "Copilot"
    default_model = "Copilot"
    default_vision_model = default_model
-    models = [default_model]
-    image_models = models
-    model_aliases = {
-        "dall-e-3": default_model
-    }

    @classmethod
    async def on_auth_async(cls, proxy: str = None, **kwargs) -> AsyncIterator:
--- a/g4f/api/stubs.py
+++ b/g4f/api/stubs.py
@@ -68,6 +68,7 @@ class ImageGenerationConfig(BaseModel):
    aspect_ratio: Optional[str] = None
    n: Optional[int] = None
    negative_prompt: Optional[str] = None
+    resolution: Optional[str] = None

 class ProviderResponseModel(BaseModel):
    id: str
--- a/g4f/gui/client/static/js/chat.v1.js
+++ b/g4f/gui/client/static/js/chat.v1.js
@@ -1325,7 +1325,12 @@ const delete_conversation = async (conversation_id) => {
            }
        }
    }
-
+    if (window.share_id && conversation_id == window.start_id) {
+        const url = `${window.share_url}/backend-api/v2/files/${window.share_id}`;
+        await fetch(url, {
+            method: 'DELETE'
+        });
+    }
    appStorage.removeItem(`conversation:${conversation_id}`);
    const item = document.getElementById(`convo-${conversation_id}`);
    item.remove();
@@ -2622,7 +2627,7 @@ function connectToSSE(url, do_refine, bucket_id) {
        } else if (data.action == "media") {
            inputCount.innerText = `File: ${data.filename}`;
            const url = `/files/${bucket_id}/media/${data.filename}`;
-            const media = [{bucket_id: bucket_id, url: url}];
+            const media = [{bucket_id: bucket_id, url: url, name: data.filename}];
            await handle_ask(false, media);
        } else if (data.action == "load") {
            inputCount.innerText = `Read data: ${formatFileSize(data.size)}`;
--- a/g4f/image/init.py
+++ b/g4f/image/init.py
@@ -37,7 +37,7 @@ EXTENSIONS_MAP: dict[str, str] = {
    "flac": "audio/flac",
    "opus": "audio/opus",
    "ogg": "audio/ogg",
-    "m4a": "audio/mp4",
+    "m4a": "audio/m4a",
     # Video
    "mkv": "video/x-matroska",
    "webm": "video/webm",
--- a/g4f/providers/response.py
+++ b/g4f/providers/response.py
@@ -284,6 +284,10 @@ class SynthesizeData(HiddenResponse, JsonMixin):
        self.provider = provider
        self.data = data

+class SuggestedFollowups(HiddenResponse):
+    def __init__(self, suggestions: list[str]):
+        self.suggestions = suggestions
+
 class RequestLogin(HiddenResponse):
    def __init__(self, label: str, login_url: str) -> None:
        """Initialize with label and login URL."""
--- a/g4f/tools/web_search.py
+++ b/g4f/tools/web_search.py
@@ -13,6 +13,7 @@ try:
    from duckduckgo_search import DDGS
    from duckduckgo_search.exceptions import DuckDuckGoSearchException
    from bs4 import BeautifulSoup
+    ddgs = DDGS()
    has_requirements = True
 except ImportError:
    has_requirements = False
@@ -164,49 +165,49 @@ async def fetch_and_scrape(session: ClientSession, url: str, max_words: int = No
 async def search(query: str, max_results: int = 5, max_words: int = 2500, backend: str = "auto", add_text: bool = True, timeout: int = 5, region: str = "wt-wt") -> SearchResults:
    if not has_requirements:
        raise MissingRequirementsError('Install "duckduckgo-search" and "beautifulsoup4" package | pip install -U g4f[search]')
-    with DDGS() as ddgs:
-        results = []
-        for result in ddgs.text(
-                query,
-                region=region,
-                safesearch="moderate",
-                timelimit="y",
-                max_results=max_results,
-                backend=backend,
-            ):
-            if ".google." in result["href"]:
-                continue
-            results.append(SearchResultEntry(
-                result["title"],
-                result["href"],
-                result["body"]
-            ))

+    results = []
+    for result in ddgs.text(
+            query,
+            region=region,
+            safesearch="moderate",
+            timelimit="y",
+            max_results=max_results,
+            backend=backend,
+        ):
+        if ".google." in result["href"]:
+            continue
+        results.append(SearchResultEntry(
+            result["title"],
+            result["href"],
+            result["body"]
+        ))
+
+    if add_text:
+        requests = []
+        async with ClientSession(timeout=ClientTimeout(timeout)) as session:
+            for entry in results:
+                requests.append(fetch_and_scrape(session, entry.url, int(max_words / (max_results - 1)), False))
+            texts = await asyncio.gather(*requests)
+
+    formatted_results = []
+    used_words = 0
+    left_words = max_words
+    for i, entry in enumerate(results):
        if add_text:
-            requests = []
-            async with ClientSession(timeout=ClientTimeout(timeout)) as session:
-                for entry in results:
-                    requests.append(fetch_and_scrape(session, entry.url, int(max_words / (max_results - 1)), False))
-                texts = await asyncio.gather(*requests)
+            entry.text = texts[i]
+        if max_words:
+            left_words -= entry.title.count(" ") + 5
+            if entry.text:
+                left_words -= entry.text.count(" ")
+            else:
+                left_words -= entry.snippet.count(" ")
+            if 0 > left_words:
+                break
+        used_words = max_words - left_words
+        formatted_results.append(entry)

-        formatted_results = []
-        used_words = 0
-        left_words = max_words
-        for i, entry in enumerate(results):
-            if add_text:
-                entry.text = texts[i]
-            if max_words:
-                left_words -= entry.title.count(" ") + 5
-                if entry.text:
-                    left_words -= entry.text.count(" ")
-                else:
-                    left_words -= entry.snippet.count(" ")
-                if 0 > left_words:
-                    break
-            used_words = max_words - left_words
-            formatted_results.append(entry)
-
-        return SearchResults(formatted_results, used_words)
+    return SearchResults(formatted_results, used_words)

 async def do_search(prompt: str, query: str = None, instructions: str = DEFAULT_INSTRUCTIONS, **kwargs) -> tuple[str, Sources]:
    if instructions and instructions in prompt: