bob_llama_cpp.prompt_tools
==========================

.. py:module:: bob_llama_cpp.prompt_tools


Attributes
----------

.. autoapisummary::

   bob_llama_cpp.prompt_tools.tool_calls
   bob_llama_cpp.prompt_tools.tool_functions
   bob_llama_cpp.prompt_tools.model_default
   bob_llama_cpp.prompt_tools.tokenizer


Functions
---------

.. autoapisummary::

   bob_llama_cpp.prompt_tools.generate_tool_results
   bob_llama_cpp.prompt_tools.parse_tool_calls
   bob_llama_cpp.prompt_tools.detect_and_process_tool_calls
   bob_llama_cpp.prompt_tools.import_module_from_path
   bob_llama_cpp.prompt_tools.apply_chat_template


Module Contents
---------------

.. py:data:: tool_calls
   :value: []


.. py:data:: tool_functions

.. py:data:: model_default
   :value: 'mistralai/Mistral-7B-Instruct-v0.3'


.. py:data:: tokenizer
   :value: None


.. py:function:: generate_tool_results() -> str

   Generate tool result response from tool_calls list.
   If global module var tokenizer is not initialized
   the tool results are generated in lama2 format

   :return: The string with generated tool results.
   Returns None if there were no tool calls to process
   :rtype: str


.. py:function:: parse_tool_calls(text: str, calls: list) -> str

   Parse text for json arrays, multiple arrays are merged together 
   into the calls argument.

   :param text: The string containing the json array
   :type text: str
   :param calls: The list where array items are added if found a json array
   :type calls: list
   :return: The remaing text without the python array
   :rtype: str


.. py:function:: detect_and_process_tool_calls(response: str) -> int

   Check if tools calls exists in the response. If one or more are found 
   and it matches a tool call it will try to execute it/them.
   If the execution fails an exception will be raised.
   The tool detection works with [] and <tool_call></tool_call> pairs.

   :param response: The text where to look for json function tool calls
   :type response: str
   :return: The count of called tool function
   :rtype: int


.. py:function:: import_module_from_path(module_name: str, path: str) -> Any

   Function to dynamically import a module from file path

   :param module_name: Name of the module
   :type module_name: str
   :param path: Path to python module file
   :type path: str
   :return: The loaded python module
   :rtype: Any


.. py:function:: apply_chat_template(conversation: list = None, model_id: str = None) -> str

   Generate a system prompt with the available tools using HF AutoTokenizer apply_chat_template function.
   It takes into account the model specific prompt format.

   :param conversation: Array with role/content dicts, defaults to None
   :type conversation: list, optional
   :param model_id: The Huggingface model_id to be used by the tokenizer, defaults to None
   :type model_id: str, optional
   :return: The generated prompt
   :rtype: str