camel-ai 0.2.75a6__py3-none-any.whl → 0.2.76__py3-none-any.whl

camel/toolkits/hybrid_browser_toolkit_py/hybrid_browser_toolkit.py

@@ -92,11 +92,12 @@ class HybridBrowserToolkit(BaseToolkit, RegisteredAgentToolkit):
         user_data_dir: Optional[str] = None,
         stealth: bool = False,
         web_agent_model: Optional[BaseModelBackend] = None,
-        cache_dir: str = "tmp/",
+        cache_dir: Optional[str] = None,
         enabled_tools: Optional[List[str]] = None,
         browser_log_to_file: bool = False,
+        log_dir: Optional[str] = None,
         session_id: Optional[str] = None,
-        default_start_url: str = "https://google.com/",
+        default_start_url: Optional[str] = None,
         default_timeout: Optional[int] = None,
         short_timeout: Optional[int] = None,
         navigation_timeout: Optional[int] = None,
@@ -144,6 +145,8 @@ class HybridBrowserToolkit(BaseToolkit, RegisteredAgentToolkit):
                 and page loading times.
                 Logs are saved to an auto-generated timestamped file.
                 Defaults to `False`.
+            log_dir (Optional[str]): Custom directory path for log files.
+                If None, defaults to "browser_log". Defaults to `None`.
             session_id (Optional[str]): A unique identifier for this browser
                 session. When multiple HybridBrowserToolkit instances are
                 used
@@ -199,9 +202,10 @@ class HybridBrowserToolkit(BaseToolkit, RegisteredAgentToolkit):
         self._user_data_dir = user_data_dir
         self._stealth = stealth
         self._web_agent_model = web_agent_model
-        self._cache_dir = cache_dir
+        self._cache_dir = cache_dir or "tmp/"
         self._browser_log_to_file = browser_log_to_file
-        self._default_start_url = default_start_url
+        self._log_dir = log_dir
+        self._default_start_url = default_start_url or "https://google.com/"
         self._session_id = session_id or "default"
         self._viewport_limit = viewport_limit
 
@@ -237,7 +241,7 @@ class HybridBrowserToolkit(BaseToolkit, RegisteredAgentToolkit):
         # Set up log file if needed
         if self.log_to_file:
             # Create log directory if it doesn't exist
-            log_dir = "browser_log"
+            log_dir = self._log_dir if self._log_dir else "browser_log"
             os.makedirs(log_dir, exist_ok=True)
 
             timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")

camel/toolkits/{openai_image_toolkit.py → image_generation_toolkit.py}

@@ -15,7 +15,7 @@
 import base64
 import os
 from io import BytesIO
-from typing import List, Literal, Optional, Union
+from typing import ClassVar, List, Literal, Optional, Tuple, Union
 
 from openai import OpenAI
 from PIL import Image
@@ -29,21 +29,32 @@ logger = get_logger(__name__)
 
 
 @MCPServer()
-class OpenAIImageToolkit(BaseToolkit):
-    r"""A class toolkit for image generation using OpenAI's
-    Image Generation API.
-    """
-
-    @api_keys_required(
-        [
-            ("api_key", "OPENAI_API_KEY"),
-        ]
-    )
+class ImageGenToolkit(BaseToolkit):
+    r"""A class toolkit for image generation using Grok and OpenAI models."""
+
+    GROK_MODELS: ClassVar[List[str]] = [
+        "grok-2-image",
+        "grok-2-image-latest",
+        "grok-2-image-1212",
+    ]
+    OPENAI_MODELS: ClassVar[List[str]] = [
+        "gpt-image-1",
+        "dall-e-3",
+        "dall-e-2",
+    ]
+
     def __init__(
         self,
         model: Optional[
-            Literal["gpt-image-1", "dall-e-3", "dall-e-2"]
-        ] = "gpt-image-1",
+            Literal[
+                "gpt-image-1",
+                "dall-e-3",
+                "dall-e-2",
+                "grok-2-image",
+                "grok-2-image-latest",
+                "grok-2-image-1212",
+            ]
+        ] = "dall-e-3",
         timeout: Optional[float] = None,
         api_key: Optional[str] = None,
         url: Optional[str] = None,
@@ -72,12 +83,12 @@ class OpenAIImageToolkit(BaseToolkit):
         # NOTE: Some arguments are set in the constructor to prevent the agent
         # from making invalid API calls with model-specific parameters. For
         # example, the 'style' argument is only supported by 'dall-e-3'.
-        r"""Initializes a new instance of the OpenAIImageToolkit class.
+        r"""Initializes a new instance of the ImageGenToolkit class.
 
         Args:
             api_key (Optional[str]): The API key for authenticating
-                with the OpenAI service. (default: :obj:`None`)
-            url (Optional[str]): The url to the OpenAI service.
+                with the image model service. (default: :obj:`None`)
+            url (Optional[str]): The url to the image model service.
                 (default: :obj:`None`)
             model (Optional[str]): The model to use.
                 (default: :obj:`"dall-e-3"`)
@@ -103,9 +114,23 @@ class OpenAIImageToolkit(BaseToolkit):
                 image.(default: :obj:`"image_save"`)
         """
         super().__init__(timeout=timeout)
-        api_key = api_key or os.environ.get("OPENAI_API_KEY")
-        url = url or os.environ.get("OPENAI_API_BASE_URL")
-        self.client = OpenAI(api_key=api_key, base_url=url)
+        if model not in self.GROK_MODELS + self.OPENAI_MODELS:
+            available_models = sorted(self.OPENAI_MODELS + self.GROK_MODELS)
+            raise ValueError(
+                f"Unsupported model: {model}. "
+                f"Supported models are: {available_models}"
+            )
+
+        # Set default url for Grok models
+        url = "https://api.x.ai/v1" if model in self.GROK_MODELS else url
+
+        api_key, base_url = (
+            self.get_openai_credentials(url, api_key)
+            if model in self.OPENAI_MODELS
+            else self.get_grok_credentials(url, api_key)
+        )
+
+        self.client = OpenAI(api_key=api_key, base_url=base_url)
         self.model = model
         self.size = size
         self.quality = quality
@@ -139,7 +164,7 @@ class OpenAIImageToolkit(BaseToolkit):
             return None
 
     def _build_base_params(self, prompt: str, n: Optional[int] = None) -> dict:
-        r"""Build base parameters dict for OpenAI API calls.
+        r"""Build base parameters dict for Image Model API calls.
 
         Args:
             prompt (str): The text prompt for the image operation.
@@ -153,6 +178,10 @@ class OpenAIImageToolkit(BaseToolkit):
         # basic parameters supported by all models
         if n is not None:
             params["n"] = n  # type: ignore[assignment]
+
+        if self.model in self.GROK_MODELS:
+            return params
+
         if self.size is not None:
             params["size"] = self.size
 
@@ -179,16 +208,18 @@ class OpenAIImageToolkit(BaseToolkit):
                 params["quality"] = self.quality
             if self.background is not None:
                 params["background"] = self.background
-
         return params
 
     def _handle_api_response(
-        self, response, image_name: Union[str, List[str]], operation: str
+        self,
+        response,
+        image_name: Union[str, List[str]],
+        operation: str,
     ) -> str:
-        r"""Handle API response from OpenAI image operations.
+        r"""Handle API response from image operations.
 
         Args:
-            response: The response object from OpenAI API.
+            response: The response object from image model API.
             image_name (Union[str, List[str]]): Name(s) for the saved image
                 file(s). If str, the same name is used for all images (will
                 cause error for multiple images). If list, must have exactly
@@ -198,8 +229,9 @@ class OpenAIImageToolkit(BaseToolkit):
         Returns:
             str: Success message with image path/URL or error message.
         """
+        source = "Grok" if self.model in self.GROK_MODELS else "OpenAI"
         if response.data is None or len(response.data) == 0:
-            error_msg = "No image data returned from OpenAI API."
+            error_msg = f"No image data returned from {source} API."
             logger.error(error_msg)
             return error_msg
 
@@ -283,7 +315,7 @@ class OpenAIImageToolkit(BaseToolkit):
         image_name: Union[str, List[str]] = "image.png",
         n: int = 1,
     ) -> str:
-        r"""Generate an image using OpenAI's Image Generation models.
+        r"""Generate an image using image models.
         The generated image will be saved locally (for ``b64_json`` response
         formats) or an image URL will be returned (for ``url`` response
         formats).
@@ -309,15 +341,50 @@ class OpenAIImageToolkit(BaseToolkit):
             logger.error(error_msg)
             return error_msg
 
+    @api_keys_required([("api_key", "XAI_API_KEY")])
+    def get_grok_credentials(self, url, api_key) -> Tuple[str, str]:  # type: ignore[return-value]
+        r"""Get API credentials for the specified Grok model.
+
+        Args:
+            url (str): The base URL for the Grok API.
+            api_key (str): The API key for the Grok API.
+
+        Returns:
+            tuple: (api_key, base_url)
+        """
+
+        # Get credentials based on model type
+        api_key = api_key or os.getenv("XAI_API_KEY")
+        return api_key, url
+
+    @api_keys_required([("api_key", "OPENAI_API_KEY")])
+    def get_openai_credentials(self, url, api_key) -> Tuple[str, str | None]:  # type: ignore[return-value]
+        r"""Get API credentials for the specified OpenAI model.
+
+        Args:
+            url (str): The base URL for the OpenAI API.
+            api_key (str): The API key for the OpenAI API.
+
+        Returns:
+            Tuple[str, str | None]: (api_key, base_url)
+        """
+
+        api_key = api_key or os.getenv("OPENAI_API_KEY")
+        base_url = url or os.getenv("OPENAI_API_BASE_URL")
+        return api_key, base_url
+
     def get_tools(self) -> List[FunctionTool]:
-        r"""Returns a list of FunctionTool objects representing the
-        functions in the toolkit.
+        r"""Returns a list of FunctionTool objects representing the functions
+            in the toolkit.
 
         Returns:
-            List[FunctionTool]: A list of FunctionTool objects
-                representing the functions in the toolkit.
+            List[FunctionTool]: A list of FunctionTool objects representing the
+                functions in the toolkit.
         """
         return [
             FunctionTool(self.generate_image),
-            # could add edit_image function later
         ]
+
+
+# Backward compatibility alias
+OpenAIImageToolkit = ImageGenToolkit

camel/toolkits/markitdown_toolkit.py

@@ -25,12 +25,38 @@ logger = get_logger(__name__)
 
 @MCPServer()
 class MarkItDownToolkit(BaseToolkit):
-    r"""A class representing a toolkit for MarkItDown."""
+    r"""A class representing a toolkit for MarkItDown.
+
+    .. deprecated::
+        MarkItDownToolkit is deprecated. Use FileToolkit instead, which now
+        includes the same functionality through its read_file method that
+        supports both single files and multiple files.
+
+        Example migration:
+            # Old way
+            from camel.toolkits import MarkItDownToolkit
+            toolkit = MarkItDownToolkit()
+            content = toolkit.read_files(['file1.pdf', 'file2.docx'])
+
+            # New way
+            from camel.toolkits import FileToolkit
+            toolkit = FileToolkit()
+            content = toolkit.read_file(['file1.pdf', 'file2.docx'])
+    """
 
     def __init__(
         self,
         timeout: Optional[float] = None,
     ):
+        import warnings
+
+        warnings.warn(
+            "MarkItDownToolkit is deprecated and will be removed in a future "
+            "version. Please use FileToolkit instead, which now includes "
+            "read_file method that supports both single and multiple files.",
+            DeprecationWarning,
+            stacklevel=2,
+        )
         super().__init__(timeout=timeout)
 
     def read_files(self, file_paths: List[str]) -> Dict[str, str]: