camel-ai 0.2.67__py3-none-any.whl → 0.2.80a2__py3-none-any.whl

camel/messages/base.py

@@ -64,12 +64,13 @@ class BaseMessage:
         content (str): The content of the message.
         video_bytes (Optional[bytes]): Optional bytes of a video associated
             with the message. (default: :obj:`None`)
-        image_list (Optional[List[Image.Image]]): Optional list of PIL Image
-            objects associated with the message. (default: :obj:`None`)
+        image_list (Optional[List[Union[Image.Image, str]]]): Optional list of
+            PIL Image objects or image URLs (strings) associated with the
+            message. (default: :obj:`None`)
         image_detail (Literal["auto", "low", "high"]): Detail level of the
             images associated with the message. (default: :obj:`auto`)
         video_detail (Literal["auto", "low", "high"]): Detail level of the
-            videos associated with the message. (default: :obj:`low`)
+            videos associated with the message. (default: :obj:`auto`)
         parsed: Optional[Union[Type[BaseModel], dict]]: Optional object which
             is parsed from the content. (default: :obj:`None`)
     """
@@ -80,9 +81,9 @@ class BaseMessage:
     content: str
 
     video_bytes: Optional[bytes] = None
-    image_list: Optional[List[Image.Image]] = None
+    image_list: Optional[List[Union[Image.Image, str]]] = None
     image_detail: Literal["auto", "low", "high"] = "auto"
-    video_detail: Literal["auto", "low", "high"] = "low"
+    video_detail: Literal["auto", "low", "high"] = "auto"
     parsed: Optional[Union[BaseModel, dict]] = None
 
     @classmethod
@@ -92,7 +93,7 @@ class BaseMessage:
         content: str,
         meta_dict: Optional[Dict[str, str]] = None,
         video_bytes: Optional[bytes] = None,
-        image_list: Optional[List[Image.Image]] = None,
+        image_list: Optional[List[Union[Image.Image, str]]] = None,
         image_detail: Union[
             OpenAIVisionDetailType, str
         ] = OpenAIVisionDetailType.AUTO,
@@ -109,8 +110,9 @@ class BaseMessage:
                 dictionary for the message.
             video_bytes (Optional[bytes]): Optional bytes of a video
                 associated with the message.
-            image_list (Optional[List[Image.Image]]): Optional list of PIL
-                Image objects associated with the message.
+            image_list (Optional[List[Union[Image.Image, str]]]): Optional list
+                of PIL Image objects or image URLs (strings) associated with
+                the message.
             image_detail (Union[OpenAIVisionDetailType, str]): Detail level of
                 the images associated with the message.
             video_detail (Union[OpenAIVisionDetailType, str]): Detail level of
@@ -137,7 +139,7 @@ class BaseMessage:
         content: str,
         meta_dict: Optional[Dict[str, str]] = None,
         video_bytes: Optional[bytes] = None,
-        image_list: Optional[List[Image.Image]] = None,
+        image_list: Optional[List[Union[Image.Image, str]]] = None,
         image_detail: Union[
             OpenAIVisionDetailType, str
         ] = OpenAIVisionDetailType.AUTO,
@@ -154,8 +156,9 @@ class BaseMessage:
                 dictionary for the message.
             video_bytes (Optional[bytes]): Optional bytes of a video
                 associated with the message.
-            image_list (Optional[List[Image.Image]]): Optional list of PIL
-                Image objects associated with the message.
+            image_list (Optional[List[Union[Image.Image, str]]]): Optional list
+                of PIL Image objects or image URLs (strings) associated with
+                the message.
             image_detail (Union[OpenAIVisionDetailType, str]): Detail level of
                 the images associated with the message.
             video_detail (Union[OpenAIVisionDetailType, str]): Detail level of
@@ -175,6 +178,32 @@ class BaseMessage:
             OpenAIVisionDetailType(video_detail).value,
         )
 
+    @classmethod
+    def make_system_message(
+        cls,
+        content: str,
+        role_name: str = "System",
+        meta_dict: Optional[Dict[str, str]] = None,
+    ) -> "BaseMessage":
+        r"""Create a new system message.
+
+        Args:
+            content (str): The content of the system message.
+            role_name (str): The name of the system role.
+                (default: :obj:`"System"`)
+            meta_dict (Optional[Dict[str, str]]): Additional metadata
+                dictionary for the message.
+
+        Returns:
+            BaseMessage: The new system message.
+        """
+        return cls(
+            role_name,
+            RoleType.SYSTEM,
+            meta_dict,
+            content,
+        )
+
     def create_new_instance(self, content: str) -> "BaseMessage":
         r"""Create a new instance of the :obj:`BaseMessage` with updated
         content.
@@ -436,35 +465,64 @@ class BaseMessage:
         )
         if self.image_list and len(self.image_list) > 0:
             for image in self.image_list:
-                if image.format is None:
-                    raise ValueError(
-                        f"Image's `format` is `None`, please "
-                        f"transform the `PIL.Image.Image` to  one of "
-                        f"following supported formats, such as "
-                        f"{list(OpenAIImageType)}"
+                # Check if image is a URL string or PIL Image
+                if isinstance(image, str):
+                    # Image is a URL string
+                    hybrid_content.append(
+                        {
+                            "type": "image_url",
+                            "image_url": {
+                                "url": image,
+                                "detail": self.image_detail,
+                            },
+                        }
                     )
+                else:
+                    # Image is a PIL Image object
+                    if image.format is None:
+                        # Set default format to PNG as fallback
+                        image.format = 'PNG'
+
+                    image_type: str = image.format.lower()
+                    if image_type not in OpenAIImageType:
+                        raise ValueError(
+                            f"Image type {image.format} "
+                            f"is not supported by OpenAI vision model"
+                        )
+
+                    # Convert RGBA to RGB for formats that don't support
+                    # transparency or when the image has transparency channel
+                    img_to_save = image
+                    if image.mode in ('RGBA', 'LA', 'P') and image_type in (
+                        'jpeg',
+                        'jpg',
+                    ):
+                        # JPEG doesn't support transparency, convert to RGB
+                        img_to_save = image.convert('RGB')
+                    elif (
+                        image.mode in ('RGBA', 'LA', 'P')
+                        and image_type == 'png'
+                    ):
+                        # For PNG with transparency, convert to RGBA if needed
+                        if image.mode in ('LA', 'P'):
+                            img_to_save = image.convert('RGBA')
+                        # else: RGBA mode, keep as-is
 
-                image_type: str = image.format.lower()
-                if image_type not in OpenAIImageType:
-                    raise ValueError(
-                        f"Image type {image.format} "
-                        f"is not supported by OpenAI vision model"
-                    )
-                with io.BytesIO() as buffer:
-                    image.save(fp=buffer, format=image.format)
-                    encoded_image = base64.b64encode(buffer.getvalue()).decode(
-                        "utf-8"
+                    with io.BytesIO() as buffer:
+                        img_to_save.save(fp=buffer, format=image.format)
+                        encoded_image = base64.b64encode(
+                            buffer.getvalue()
+                        ).decode("utf-8")
+                    image_prefix = f"data:image/{image_type};base64,"
+                    hybrid_content.append(
+                        {
+                            "type": "image_url",
+                            "image_url": {
+                                "url": f"{image_prefix}{encoded_image}",
+                                "detail": self.image_detail,
+                            },
+                        }
                     )
-                image_prefix = f"data:image/{image_type};base64,"
-                hybrid_content.append(
-                    {
-                        "type": "image_url",
-                        "image_url": {
-                            "url": f"{image_prefix}{encoded_image}",
-                            "detail": self.image_detail,
-                        },
-                    }
-                )
 
         if self.video_bytes:
             import imageio.v3 as iio
@@ -537,7 +595,18 @@ class BaseMessage:
             OpenAIAssistantMessage: The converted :obj:`OpenAIAssistantMessage`
                 object.
         """
-        return {"role": "assistant", "content": self.content}
+        message_dict: Dict[str, Any] = {
+            "role": "assistant",
+            "content": self.content,
+        }
+
+        # Check if meta_dict contains tool_calls
+        if self.meta_dict and "tool_calls" in self.meta_dict:
+            tool_calls = self.meta_dict["tool_calls"]
+            if tool_calls:
+                message_dict["tool_calls"] = tool_calls
+
+        return message_dict  # type: ignore[return-value]
 
     def to_dict(self) -> Dict:
         r"""Converts the message to a dictionary.
@@ -545,9 +614,66 @@ class BaseMessage:
         Returns:
             dict: The converted dictionary.
         """
-        return {
+        result = {
             "role_name": self.role_name,
-            "role_type": self.role_type.name,
+            "role_type": self.role_type.value,
             **(self.meta_dict or {}),
             "content": self.content,
         }
+
+        # Include image/video fields if present
+        if self.image_list is not None:
+            # Handle both PIL Images and URL strings
+            import base64
+            from io import BytesIO
+
+            image_data_list = []
+            for img in self.image_list:
+                if isinstance(img, str):
+                    # Image is a URL string, store as-is
+                    image_data_list.append({"type": "url", "data": img})
+                else:
+                    # Image is a PIL Image, convert to base64
+                    # Preserve format, default to PNG if not set
+                    img_format = img.format if img.format else "PNG"
+
+                    # Handle transparency for different formats
+                    img_to_save = img
+                    if img.mode in (
+                        'RGBA',
+                        'LA',
+                        'P',
+                    ) and img_format.upper() in ('JPEG', 'JPG'):
+                        # JPEG doesn't support transparency, convert to RGB
+                        img_to_save = img.convert('RGB')
+                    elif (
+                        img.mode in ('LA', 'P') and img_format.upper() == 'PNG'
+                    ):
+                        # For PNG with transparency, convert to RGBA if needed
+                        img_to_save = img.convert('RGBA')
+                    # else: keep as-is for other combinations
+
+                    buffered = BytesIO()
+                    img_to_save.save(buffered, format=img_format)
+                    img_str = base64.b64encode(buffered.getvalue()).decode()
+                    image_data_list.append(
+                        {
+                            "type": "base64",
+                            "data": img_str,
+                            "format": img_format,  # Preserve format
+                        }
+                    )
+            result["image_list"] = image_data_list
+
+        if self.video_bytes is not None:
+            import base64
+
+            result["video_bytes"] = base64.b64encode(self.video_bytes).decode()
+
+        if self.image_detail is not None:
+            result["image_detail"] = self.image_detail
+
+        if self.video_detail is not None:
+            result["video_detail"] = self.video_detail
+
+        return result

camel/messages/func_message.py

@@ -47,12 +47,16 @@ class FunctionCallingMessage(BaseMessage):
             (default: :obj:`None`)
         tool_call_id (Optional[str]): The ID of the tool call, if available.
             (default: :obj:`None`)
+        mask_output (Optional[bool]): Whether to return a sanitized placeholder
+            instead of the raw tool output.
+            (default: :obj:`False`)
     """
 
     func_name: Optional[str] = None
     args: Optional[Dict] = None
     result: Optional[Any] = None
     tool_call_id: Optional[str] = None
+    mask_output: Optional[bool] = False
 
     def to_openai_message(
         self,
@@ -105,10 +109,13 @@ class FunctionCallingMessage(BaseMessage):
             # This is a function response
             # TODO: Allow for more flexible setting of tool role,
             #  optionally to be the same as assistant messages
-            content = function_format.format_tool_response(
-                self.func_name,  # type: ignore[arg-type]
-                self.result,  # type: ignore[arg-type]
-            )
+            if self.mask_output:
+                content = "[MASKED]"
+            else:
+                content = function_format.format_tool_response(
+                    self.func_name,  # type: ignore[arg-type]
+                    self.result,  # type: ignore[arg-type]
+                )
             return ShareGPTMessage(from_="tool", value=content)  # type: ignore[call-arg]
 
     def to_openai_assistant_message(self) -> OpenAIAssistantMessage:
@@ -154,10 +161,30 @@ class FunctionCallingMessage(BaseMessage):
                 " due to missing function name."
             )
 
-        result_content = str(self.result)
+        if self.mask_output:
+            result_content = "[MASKED]"
+        else:
+            result_content = str(self.result)
 
         return {
             "role": "tool",
             "content": result_content,
             "tool_call_id": self.tool_call_id or "null",
         }
+
+    def to_dict(self) -> Dict:
+        r"""Converts the message to a dictionary.
+
+        Returns:
+            dict: The converted dictionary.
+        """
+        base = super().to_dict()
+        base["func_name"] = self.func_name
+        if self.args is not None:
+            base["args"] = self.args
+        if self.result is not None:
+            base["result"] = self.result
+        if self.tool_call_id is not None:
+            base["tool_call_id"] = self.tool_call_id
+        base["mask_output"] = self.mask_output
+        return base

camel/models/__init__.py

@@ -11,13 +11,16 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========
+from .aihubmix_model import AihubMixModel
 from .aiml_model import AIMLModel
+from .amd_model import AMDModel
 from .anthropic_model import AnthropicModel
 from .aws_bedrock_model import AWSBedrockModel
 from .azure_openai_model import AzureOpenAIModel
 from .base_audio_model import BaseAudioModel
 from .base_model import BaseModelBackend
 from .cohere_model import CohereModel
+from .cometapi_model import CometAPIModel
 from .crynux_model import CrynuxModel
 from .deepseek_model import DeepSeekModel
 from .fish_audio_model import FishAudioModel
@@ -26,11 +29,13 @@ from .groq_model import GroqModel
 from .internlm_model import InternLMModel
 from .litellm_model import LiteLLMModel
 from .lmstudio_model import LMStudioModel
+from .minimax_model import MinimaxModel
 from .mistral_model import MistralModel
 from .model_factory import ModelFactory
 from .model_manager import ModelManager, ModelProcessingError
 from .modelscope_model import ModelScopeModel
 from .moonshot_model import MoonshotModel
+from .nebius_model import NebiusModel
 from .nemotron_model import NemotronModel
 from .netmind_model import NetmindModel
 from .novita_model import NovitaModel
@@ -61,11 +66,13 @@ __all__ = [
     'OpenRouterModel',
     'AzureOpenAIModel',
     'AnthropicModel',
+    'AMDModel',
     'MistralModel',
     'GroqModel',
     'StubModel',
     'ZhipuAIModel',
     'CohereModel',
+    'CometAPIModel',
     'ModelFactory',
     'ModelManager',
     'LiteLLMModel',
@@ -87,6 +94,7 @@ __all__ = [
     'QwenModel',
     'AWSBedrockModel',
     'ModelProcessingError',
+    'NebiusModel',
     'DeepSeekModel',
     'FishAudioModel',
     'InternLMModel',
@@ -97,7 +105,9 @@ __all__ = [
     'SiliconFlowModel',
     'VolcanoModel',
     'LMStudioModel',
+    'MinimaxModel',
     'WatsonXModel',
     'QianfanModel',
     'CrynuxModel',
+    'AihubMixModel',
 ]

camel/models/aihubmix_model.py

@@ -0,0 +1,83 @@
+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========
+import os
+from typing import Any, Dict, Optional, Union
+
+from camel.models.openai_compatible_model import OpenAICompatibleModel
+from camel.types import ModelType
+from camel.utils import (
+    BaseTokenCounter,
+    api_keys_required,
+)
+
+
+class AihubMixModel(OpenAICompatibleModel):
+    r"""AihubMix API in a unified OpenAICompatibleModel interface.
+
+    Args:
+        model_type (Union[ModelType, str]): Model for which a backend is
+            created.
+        model_config_dict (Optional[Dict[str, Any]], optional): A dictionary
+            that will be fed into OpenAI client. If :obj:`None`,
+            :obj:`{}` will be used.
+            (default: :obj:`None`)
+        api_key (Optional[str], optional): The API key for authenticating with
+            AihubMix service. (default: :obj:`None`)
+        url (Optional[str], optional): The URL to AihubMix service. If
+            not provided, :obj:`https://aihubmix.com/v1` will be used.
+            (default: :obj:`None`)
+        token_counter (Optional[BaseTokenCounter], optional): Token counter to
+            use for the model. If not provided, :obj:`OpenAITokenCounter(
+            ModelType.GPT_4O_MINI)` will be used.
+            (default: :obj:`None`)
+        timeout (Optional[float], optional): The timeout value in seconds for
+            API calls. If not provided, will fall back to the MODEL_TIMEOUT
+            environment variable or default to 180 seconds.
+            (default: :obj:`None`)
+        max_retries (int, optional): Maximum number of retries for API calls.
+            (default: :obj:`3`)
+        **kwargs (Any): Additional arguments to pass to the client
+            initialization.
+    """
+
+    @api_keys_required([("api_key", "AIHUBMIX_API_KEY")])
+    def __init__(
+        self,
+        model_type: Union[ModelType, str],
+        model_config_dict: Optional[Dict[str, Any]] = None,
+        api_key: Optional[str] = None,
+        url: Optional[str] = None,
+        token_counter: Optional[BaseTokenCounter] = None,
+        timeout: Optional[float] = None,
+        max_retries: int = 3,
+        **kwargs: Any,
+    ) -> None:
+        if model_config_dict is None:
+            model_config_dict = {}
+        api_key = api_key or os.environ.get("AIHUBMIX_API_KEY")
+        url = url or os.environ.get(
+            "AIHUBMIX_API_BASE_URL",
+            "https://aihubmix.com/v1",
+        )
+        timeout = timeout or float(os.environ.get("MODEL_TIMEOUT", 180))
+        super().__init__(
+            model_type=model_type,
+            model_config_dict=model_config_dict,
+            api_key=api_key,
+            url=url,
+            token_counter=token_counter,
+            timeout=timeout,
+            max_retries=max_retries,
+            **kwargs,
+        )

camel/models/aiml_model.py

@@ -14,7 +14,7 @@
 import os
 from typing import Any, Dict, Optional, Union
 
-from camel.configs import AIML_API_PARAMS, AIMLConfig
+from camel.configs import AIMLConfig
 from camel.models.openai_compatible_model import OpenAICompatibleModel
 from camel.types import ModelType
 from camel.utils import (
@@ -82,18 +82,3 @@ class AIMLModel(OpenAICompatibleModel):
             max_retries=max_retries,
             **kwargs,
         )
-
-    def check_model_config(self):
-        r"""Check whether the model configuration contains any
-        unexpected arguments to AIML API.
-
-        Raises:
-            ValueError: If the model configuration dictionary contains any
-                unexpected arguments to AIML API.
-        """
-        for param in self.model_config_dict:
-            if param not in AIML_API_PARAMS:
-                raise ValueError(
-                    f"Unexpected argument `{param}` is "
-                    "input into AIML model backend."
-                )

camel/models/amd_model.py

@@ -0,0 +1,101 @@
+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ========= Copyright 2023-2024 @ CAMEL-AI.org. All Rights Reserved. =========
+
+import os
+from typing import Any, Dict, Optional, Union
+
+from camel.configs import AMD_API_PARAMS, AMDConfig
+from camel.models.openai_compatible_model import OpenAICompatibleModel
+from camel.types import ModelType
+from camel.utils import BaseTokenCounter, api_keys_required
+
+
+class AMDModel(OpenAICompatibleModel):
+    r"""AMD API in a unified OpenAICompatibleModel interface.
+
+    Args:
+        model_type (Union[ModelType, str]): Model for which a backend is
+            created, one of AMD series.
+        model_config_dict (Optional[Dict[str, Any]], optional): A dictionary
+            that will be fed into:obj:`openai.ChatCompletion.create()`. If
+            :obj:`None`, :obj:`AMDConfig().as_dict()` will be used.
+            (default: :obj:`None`)
+        api_key (Optional[str], optional): The API key for authenticating with
+            the AMD service. (default: :obj:`None`)
+        url (Optional[str], optional): The url to the AMD service.
+            (default: :obj:`None`)
+        token_counter (Optional[BaseTokenCounter], optional): Token counter to
+            use for the model. If not provided, :obj:`OpenAITokenCounter(
+            ModelType.GPT_4)` will be used.
+            (default: :obj:`None`)
+        timeout (Optional[float], optional): The timeout value in seconds for
+            API calls. If not provided, will fall back to the MODEL_TIMEOUT
+            environment variable or default to 180 seconds.
+            (default: :obj:`None`)
+        max_retries (int, optional): Maximum number of retries for API calls.
+            (default: :obj:`3`)
+        **kwargs (Any): Additional arguments to pass to the client
+            initialization.
+    """
+
+    @api_keys_required(
+        [
+            ("api_key", "AMD_API_KEY"),
+        ]
+    )
+    def __init__(
+        self,
+        model_type: Union[ModelType, str],
+        model_config_dict: Optional[Dict[str, Any]] = None,
+        api_key: Optional[str] = None,
+        url: Optional[str] = None,
+        token_counter: Optional[BaseTokenCounter] = None,
+        timeout: Optional[float] = None,
+        max_retries: int = 3,
+        **kwargs: Any,
+    ) -> None:
+        if model_config_dict is None:
+            model_config_dict = AMDConfig().as_dict()
+        api_key = api_key or os.environ.get("AMD_API_KEY")
+        url = url or os.environ.get(
+            "AMD_API_BASE_URL", "https://llm-api.amd.com"
+        )
+        headers = {'Ocp-Apim-Subscription-Key': api_key}
+        kwargs["default_headers"] = headers
+        timeout = timeout or float(os.environ.get("MODEL_TIMEOUT", 180))
+        super().__init__(
+            model_type=model_type,
+            model_config_dict=model_config_dict,
+            api_key=api_key,
+            url=url,
+            token_counter=token_counter,
+            timeout=timeout,
+            max_retries=max_retries,
+            **kwargs,
+        )
+
+    def check_model_config(self):
+        r"""Check whether the model configuration contains any
+        unexpected arguments to AMD API.
+
+        Raises:
+            ValueError: If the model configuration dictionary contains any
+                unexpected arguments to AMD API.
+        """
+        for param in self.model_config_dict:
+            if param not in AMD_API_PARAMS:
+                raise ValueError(
+                    f"Unexpected argument `{param}` is "
+                    "input into AMD model backend."
+                )