sherpa_ai.error_handling package

Overview

The error_handling package provides robust error management capabilities for Sherpa AI applications, ensuring graceful recovery from failures and maintaining system stability during agent operations.

Key Components

• AgentErrorHandler: Manages and handles errors occurring in agent operations

• Error Classification: Categorizes errors for appropriate handling strategies

• Recovery Mechanisms: Techniques for recovering from various error types

Example Usage

from sherpa_ai.error_handling.agent_error_handler import AgentErrorHandler
from sherpa_ai.agents import QAAgent

# Create an error handler
error_handler = AgentErrorHandler()

# Use with an agent in a try-except block
try:
    agent = QAAgent()
    response = agent.get_response("What is quantum computing?")
except Exception as e:
    # Handle the error
    handled_response = error_handler.handle(e, fallback_msg="I couldn't process your request.")
    print(handled_response)

Submodules

Module	Description
sherpa_ai.error_handling.agent_error_handler	Provides the AgentErrorHandler class for managing and recovering from errors.

sherpa_ai.error_handling.agent_error_handler module

class sherpa_ai.error_handling.agent_error_handler.AgentErrorHandler

Bases: object

Handles errors that occur during agent execution.

say

The function to call to send a message.

Type:

Callable[[str], None]

thread_ts

The thread timestamp of the message.

Type:

str

error_map

The map of errors to messages.

Type:

Dict[BaseException, str]

default_response

The default response to send if an error occurs.

Type:

str

Example

>>> handler = AgentErrorHandler()
>>> handler.run_with_error_handling(my_function, x=5)
'10'
>>> handler.run_with_error_handling(my_function, x="invalid")
'An unexpected error occurred. Please try again.'

say: Callable[[str], None]

thread_ts: str

error_map: Dict[BaseException, str]

run_with_error_handling(executable, **kwargs)

Run the executable with error handling. If an error occurs, return the error message.

Parameters:

• executable (Callable[[str, List[Dict]], Tuple[str, List[str]]]) – The executable to run with input (question, previous_messages) and returns a tuple of (response, log).

• **kwargs – The input arguments for the executable.

Returns:

The result of the function execution or an error message if an exception occurs.

Return type:

str

Example

>>> handler = AgentErrorHandler()
>>> result = handler.run_with_error_handling(my_function, x=5)
>>> print(result)
'10'
>>> result = handler.run_with_error_handling(my_function, x="invalid")
>>> print(result)
'An unexpected error occurred. Please try again.'

Module contents

Error handling module for Sherpa AI.

This module provides error handling functionality for the Sherpa AI system. It exports the AgentErrorHandler class which handles various types of errors that may occur during agent execution.

Example

>>> from sherpa_ai.error_handling import AgentErrorHandler
>>> handler = AgentErrorHandler()
>>> result = handler.run_with_error_handling(my_function, arg1="value1")
>>> print(result)
'Success' or error message if an error occurred

class sherpa_ai.error_handling.AgentErrorHandler

Bases: object

Handles errors that occur during agent execution.

say

The function to call to send a message.

Type:

Callable[[str], None]

thread_ts

The thread timestamp of the message.

Type:

str

error_map

The map of errors to messages.

Type:

Dict[BaseException, str]

default_response

The default response to send if an error occurs.

Type:

str

Example

>>> handler = AgentErrorHandler()
>>> handler.run_with_error_handling(my_function, x=5)
'10'
>>> handler.run_with_error_handling(my_function, x="invalid")
'An unexpected error occurred. Please try again.'

run_with_error_handling(executable, **kwargs)

Run the executable with error handling. If an error occurs, return the error message.

Parameters:

• executable (Callable[[str, List[Dict]], Tuple[str, List[str]]]) – The executable to run with input (question, previous_messages) and returns a tuple of (response, log).

• **kwargs – The input arguments for the executable.

Returns:

The result of the function execution or an error message if an exception occurs.

Return type:

str

Example

>>> handler = AgentErrorHandler()
>>> result = handler.run_with_error_handling(my_function, x=5)
>>> print(result)
'10'
>>> result = handler.run_with_error_handling(my_function, x="invalid")
>>> print(result)
'An unexpected error occurred. Please try again.'

say: Callable[[str], None]

thread_ts: str

error_map: Dict[BaseException, str]