nip.code_validation.dataset_generation._extract_code_and_input

nip.code_validation.dataset_generation._extract_code_and_input#

nip.code_validation.dataset_generation._extract_code_and_input(model_output: str) tuple[str | None, list[str], bool][source]#

Extract the modified solution and problematic inputs from the model output.

This function uses regular expressions to parse the model output and extract:

  1. Problematic inputs.

  2. The modified solution.

  3. A safety check to ensure the modified solution does not contain dangerous functions.

Parameters:

model_output (str) – The output from the model containing problematic inputs and the modified solution.

Returns:

  • modified_solution (str | None) – The modified solution extracted from the model output, or None if not found.

  • problematic_inputs (list[str]) – A list of problematic inputs extracted from the model output.

  • flag_unsafe (bool) – A flag indicating if the modified solution contains dangerous functions, and is therefore unsafe.