Method implement_solution

optillm/plansearch.py:146–186

(self, problem: str, solution: str)

        return response.choices[0].message.content.strip()

    def implement_solution(self, problem: str, solution: str) -> str:
        prompt = f"""You are an expert Python programmer. You will be given a question (problem specification)
and a natural language solution/tutorial that describes how to solve the problem. You will
generate a correct Python program that matches said specification and tutorial and passes
all tests. You will NOT return anything except for the program inside markdown codeblocks.

Problem:
{problem}

Solution:
{solution}

Please implement the solution in Python."""

        # Prepare request for logging
        provider_request = {
            "model": self.model,
            "max_tokens": self.max_tokens,
            "messages": [
                {"role": "system", "content": self.system_prompt},
                {"role": "user", "content": prompt}
            ]
        }

        response = self.client.chat.completions.create(**provider_request)

        # Log provider call if conversation logging is enabled
        if hasattr(optillm, 'conversation_logger') and optillm.conversation_logger and self.request_id:
            response_dict = response.model_dump() if hasattr(response, 'model_dump') else response
            optillm.conversation_logger.log_provider_call(self.request_id, provider_request, response_dict)
        self.plansearch_completion_tokens += response.usage.completion_tokens

        # Check for valid response with None-checking
        if (response is None or
            not response.choices or
            response.choices[0].message.content is None or
            response.choices[0].finish_reason == "length"):
            logger.error("Implementation response truncated or empty. Consider increasing max_tokens.")
            return "Error: Response was truncated due to token limit. Please increase max_tokens or max_completion_tokens."

        return response.choices[0].message.content.strip()

    def solve(self, problem: str, num_initial_observations: int = 3, num_derived_observations: int = 2) -> Tuple[str, str]:
        logger.info("Generating initial observations")