Bases: BaseTool[ChatCompletionMessageToolCall]
A base class for easy use of tools with the OpenAI Chat client.
OpenAITool
internally handles the logic that allows you to use tools with simple
calls such as OpenAICallResponse.tool
or OpenAITool.fn
, as seen in the
examples below.
Example:
from mirascope.openai import OpenAICall, OpenAICallParams
def animal_matcher(fav_food: str, fav_color: str) -> str:
"""Tells you your most likely favorite animal from personality traits.
Args:
fav_food: your favorite food.
fav_color: your favorite color.
Returns:
The animal most likely to be your favorite based on traits.
"""
return "Your favorite animal is the best one, a frog."
class AnimalMatcher(OpenAICall):
prompt_template = """
Tell me my favorite animal if my favorite food is {food} and my
favorite color is {color}.
"""
food: str
color: str
call_params = OpenAICallParams(tools=[animal_matcher])
response = AnimalMatcher(food="pizza", color="red").call
tool = response.tool
print(tool.fn(**tool.args))
#> Your favorite animal is the best one, a frog.
Source code in mirascope/openai/tools.py
| class OpenAITool(BaseTool[ChatCompletionMessageToolCall]):
'''A base class for easy use of tools with the OpenAI Chat client.
`OpenAITool` internally handles the logic that allows you to use tools with simple
calls such as `OpenAICallResponse.tool` or `OpenAITool.fn`, as seen in the
examples below.
Example:
```python
from mirascope.openai import OpenAICall, OpenAICallParams
def animal_matcher(fav_food: str, fav_color: str) -> str:
"""Tells you your most likely favorite animal from personality traits.
Args:
fav_food: your favorite food.
fav_color: your favorite color.
Returns:
The animal most likely to be your favorite based on traits.
"""
return "Your favorite animal is the best one, a frog."
class AnimalMatcher(OpenAICall):
prompt_template = """
Tell me my favorite animal if my favorite food is {food} and my
favorite color is {color}.
"""
food: str
color: str
call_params = OpenAICallParams(tools=[animal_matcher])
response = AnimalMatcher(food="pizza", color="red").call
tool = response.tool
print(tool.fn(**tool.args))
#> Your favorite animal is the best one, a frog.
```
'''
@classmethod
def tool_schema(cls) -> ChatCompletionToolParam:
"""Constructs a tool schema for use with the OpenAI Chat client.
A Mirascope `OpenAITool` is deconstructed into a JSON schema, and relevant keys
are renamed to match the OpenAI `ChatCompletionToolParam` schema used to make
function/tool calls in OpenAI API.
Returns:
The constructed `ChatCompletionToolParam` schema.
"""
fn = super().tool_schema()
return cast(ChatCompletionToolParam, {"type": "function", "function": fn})
@classmethod
def from_tool_call(
cls,
tool_call: ChatCompletionMessageToolCall,
allow_partial: bool = False,
) -> OpenAITool:
"""Extracts an instance of the tool constructed from a tool call response.
Given `ChatCompletionMessageToolCall` from an OpenAI chat completion response,
takes its function arguments and creates an `OpenAITool` instance from it.
Args:
tool_call: The `ChatCompletionMessageToolCall` to extract the tool from.
Returns:
An instance of the tool constructed from the tool call.
Raises:
ValidationError: if the tool call doesn't match the tool schema.
"""
if allow_partial:
model_json = from_json(tool_call.function.arguments, allow_partial=True)
else:
try:
model_json = json.loads(tool_call.function.arguments)
except json.JSONDecodeError as e:
raise ValueError() from e
model_json["tool_call"] = tool_call.model_dump()
return cls.model_validate(model_json)
@classmethod
def from_model(cls, model: Type[BaseModel]) -> Type[OpenAITool]:
"""Constructs a `OpenAITool` type from a `BaseModel` type."""
return convert_base_model_to_tool(model, OpenAITool)
@classmethod
def from_fn(cls, fn: Callable) -> Type[OpenAITool]:
"""Constructs a `OpenAITool` type from a function."""
return convert_function_to_tool(fn, OpenAITool)
@classmethod
def from_base_type(cls, base_type: Type[BaseType]) -> Type[OpenAITool]:
"""Constructs a `OpenAITool` type from a `BaseType` type."""
return convert_base_type_to_tool(base_type, OpenAITool)
|
Constructs a OpenAITool
type from a BaseType
type.
Source code in mirascope/openai/tools.py
| @classmethod
def from_base_type(cls, base_type: Type[BaseType]) -> Type[OpenAITool]:
"""Constructs a `OpenAITool` type from a `BaseType` type."""
return convert_base_type_to_tool(base_type, OpenAITool)
|
Constructs a OpenAITool
type from a function.
Source code in mirascope/openai/tools.py
| @classmethod
def from_fn(cls, fn: Callable) -> Type[OpenAITool]:
"""Constructs a `OpenAITool` type from a function."""
return convert_function_to_tool(fn, OpenAITool)
|
Constructs a OpenAITool
type from a BaseModel
type.
Source code in mirascope/openai/tools.py
| @classmethod
def from_model(cls, model: Type[BaseModel]) -> Type[OpenAITool]:
"""Constructs a `OpenAITool` type from a `BaseModel` type."""
return convert_base_model_to_tool(model, OpenAITool)
|
Extracts an instance of the tool constructed from a tool call response.
Given ChatCompletionMessageToolCall
from an OpenAI chat completion response,
takes its function arguments and creates an OpenAITool
instance from it.
Parameters:
Name |
Type |
Description |
Default |
tool_call |
ChatCompletionMessageToolCall
|
The ChatCompletionMessageToolCall to extract the tool from.
|
required
|
Returns:
Type |
Description |
OpenAITool
|
An instance of the tool constructed from the tool call.
|
Raises:
Type |
Description |
ValidationError
|
if the tool call doesn't match the tool schema.
|
Source code in mirascope/openai/tools.py
| @classmethod
def from_tool_call(
cls,
tool_call: ChatCompletionMessageToolCall,
allow_partial: bool = False,
) -> OpenAITool:
"""Extracts an instance of the tool constructed from a tool call response.
Given `ChatCompletionMessageToolCall` from an OpenAI chat completion response,
takes its function arguments and creates an `OpenAITool` instance from it.
Args:
tool_call: The `ChatCompletionMessageToolCall` to extract the tool from.
Returns:
An instance of the tool constructed from the tool call.
Raises:
ValidationError: if the tool call doesn't match the tool schema.
"""
if allow_partial:
model_json = from_json(tool_call.function.arguments, allow_partial=True)
else:
try:
model_json = json.loads(tool_call.function.arguments)
except json.JSONDecodeError as e:
raise ValueError() from e
model_json["tool_call"] = tool_call.model_dump()
return cls.model_validate(model_json)
|
Constructs a tool schema for use with the OpenAI Chat client.
A Mirascope OpenAITool
is deconstructed into a JSON schema, and relevant keys
are renamed to match the OpenAI ChatCompletionToolParam
schema used to make
function/tool calls in OpenAI API.
Returns:
Type |
Description |
ChatCompletionToolParam
|
The constructed ChatCompletionToolParam schema.
|
Source code in mirascope/openai/tools.py
| @classmethod
def tool_schema(cls) -> ChatCompletionToolParam:
"""Constructs a tool schema for use with the OpenAI Chat client.
A Mirascope `OpenAITool` is deconstructed into a JSON schema, and relevant keys
are renamed to match the OpenAI `ChatCompletionToolParam` schema used to make
function/tool calls in OpenAI API.
Returns:
The constructed `ChatCompletionToolParam` schema.
"""
fn = super().tool_schema()
return cast(ChatCompletionToolParam, {"type": "function", "function": fn})
|