Method parse

evaluation/dataset_parser.py:35–77

Parse XML evaluation file and return list of evaluation tasks. Supports both and tags. Prefers if both exist. Args: file_path: Path to XML evaluation file Returns: List of evaluation tasks wi

(self, file_path: Path)

    """Parser for XML evaluation dataset files."""

    def parse(self, file_path: Path) -> List[Dict[str, Any]]:
        """
        Parse XML evaluation file and return list of evaluation tasks.

        Supports both <response> and <response_pattern> tags.
        Prefers <response> if both exist.

        Args:
            file_path: Path to XML evaluation file

        Returns:
            List of evaluation tasks with "prompt" and "response" keys
        """
        try:
            tree = ET.parse(file_path)
            root = tree.getroot()
            evaluations = []

            tasks = root.findall(".//task")
            for task in tasks:
                prompt_elem = task.find("prompt")
                response_elem = task.find("response")
                response_pattern_elem = task.find("response_pattern")

                # Support both <response> and <response_pattern>
                # Prefer <response> if both exist
                expected_response = None
                if response_elem is not None:
                    expected_response = (response_elem.text or "").strip()
                elif response_pattern_elem is not None:
                    expected_response = (response_pattern_elem.text or "").strip()

                if prompt_elem is not None and expected_response is not None:
                    eval_dict = {
                        "prompt": (prompt_elem.text or "").strip(),
                        "response": expected_response,
                    }
                    evaluations.append(eval_dict)

            return evaluations
        except Exception as e:
            print(f"Error parsing evaluation file {file_path}: {e}")
            return []