Docs
Services
Openai.service

OpenAI Service

services.openai_service.chat_ta(assistant_id, thread_id, user_input)

Sends a message to an OpenAI assistant and manages the conversation within a specific thread, counting the tokens used.

  • Parameters:
    • assistant_id (str) – The unique identifier for the AI assistant.
    • thread_id (str) – The identifier for the conversation thread.
    • user_input (str) – The user’s message to the AI assistant.
  • Returns: A dictionary with either a ‘success’ key containing the assistant’s response or an ‘error’ key with an error message.
  • Return type: dict

Notes

  • Uses tiktoken library to count tokens for both the user’s input and the assistant’s response.
  • Manages messages and interactions via client.beta.threads.messages.create and client.beta.threads.runs.create.
  • Handles timeouts, returning an error if response time exceeds 55 seconds.

Examples

>>> response = chat_ta('asst_abc123', 'thread_abc123', 'Hello, assistant!')
>>> print(response)
{'success': "Assistant's response text here."}
>>> response = chat_ta('asst_abc123', '', 'Hello, assistant!')
>>> print(response)
{'error': 'missing thread_id'}

services.openai_service.create_agent(agent_id)

Creates a new agent on the OpenAI server or retrieves details of an existing agent based on the provided ID.

  • Parameters: agent_id (str) – The unique identifier for the agent, used to store or retrieve the agent’s data locally.
  • Returns: The unique identifier of the created or retrieved agent, or None if creation failed.
  • Return type: str or None
  • Raises:
    • OSError – If necessary directories or files cannot be created.
    • IOError – If there’s an error in reading from or writing to the local JSON file.
    • ApiError – If there’s an issue with the OpenAI client during agent creation or file upload.

Notes

  • Uses client.beta.assistants.create to create an agent with properties like name and instructions.
  • Manages file uploads via client.files.create and stores agent details locally in /tmp/agents/{agent_id}.json.
  • Ensures the /tmp/agents directory exists for storing agent data.

Example

>>> agent_id = create_agent('1234')
>>> print(agent_id)
'unique-agent-id'

services.openai_service.delete_agent(agent_id: str)

Deletes an agent from the OpenAI server and its associated local file based on the provided agent ID.

  • Parameters: agent_id (str) – A unique identifier for the agent to be deleted.
  • Returns: The unique identifier of the deleted agent or None if the deletion failed.
  • Return type: str or None
  • Raises:
    • OSError – If unable to delete the local file.
    • ApiError – If the OpenAI client fails to delete the agent.

Notes

  • Utilizes the client.beta.assistants.delete API call to delete the agent from the server.
  • Checks for the agent’s existence by looking for a corresponding local file.
Voiceflow.routesSession.service