def reduce_log()

in project/paperbench/paperbench/judge/utils.py [0:0]

• 25 lines of code
• 7 McCabe index (conditional complexity)

def reduce_log(input_string: str) -> str:
    """
    Reduce a multi-line log string to a filtered version with repeated lines collapsed.
    """
    template_miner = TemplateMiner()
    output_lines = []

    previous_cluster_id = None
    repeat_count = 1

    for raw_line in input_string.splitlines():
        original_line = raw_line
        sanitized = sanitize_line(original_line)

        result = template_miner.add_log_message(sanitized)
        cluster_id = result["cluster_id"]

        if previous_cluster_id is None:
            # First line
            output_lines.append(original_line)
            previous_cluster_id = cluster_id
            continue

        if cluster_id == previous_cluster_id:
            repeat_count += 1
        else:
            if repeat_count > 1:
                output_lines.append(f"  (repeated {repeat_count} times)")
            output_lines.append(original_line)
            repeat_count = 1
            previous_cluster_id = cluster_id

    if previous_cluster_id is not None and repeat_count > 1:
        output_lines.append(f"  (repeated {repeat_count} times)")

    return "\n".join(output_lines)