#!/usr/bin/env python3
"""Benchmark 评估脚本 - 互动剧编剧"""

import json
import sys

def evaluate(skill_content, output_content, benchmark_def):
    """评估互动剧编剧结果"""

    # 默认弱分
    scores = {
        "completeness": 60,
        "practicality": 60,
        "tool_support": 60,
        "methodology": 60,
        "source_quality": 60,
        "platform_coverage": 60
    }

    # 解析 skill 内容检查关键要素
    skill_lower = skill_content.lower()
    output_lower = output_content.lower()

    # completeness: 知识完整性 - 检查是否覆盖方法论、工具、平台特性
    completeness_checks = [
        ("六段式" in skill_content or "六段" in skill_content, 15),
        ("开场" in skill_content and "冲突" in skill_content and "选择" in skill_content, 10),
        ("分支" in skill_content and ("结局" in skill_content or "高潮" in skill_content), 10),
        ("工具" in skill_content and ("twine" in skill_lower or "ink" in skill_lower or "ren'py" in skill_lower), 10),
    ]
    completeness_score = sum(score for check, score in completeness_checks if check)
    scores["completeness"] = min(100, 60 + completeness_score)

    # practicality: 实用性 - 检查是否有可直接指导创作实践的内容
    practicality_checks = [
        ("示例" in skill_content or "例子" in skill_content, 20),
        ("格式" in skill_content or "模板" in skill_content or "json" in skill_lower, 15),
        ("检查清单" in skill_content or "checklist" in skill_lower, 10),
    ]
    practicality_score = sum(score for check, score in practicality_checks if check)
    scores["practicality"] = min(100, 60 + practicality_score)

    # tool_support: 工具支持 - 检查是否有工具推荐、语法示例
    tool_checks = [
        ("twine" in skill_lower, 15),
        ("ink" in skill_lower, 15),
        ("[[" in skill_content or "==" in skill_content, 10),  # 语法标记
        ("语法" in skill_content or "syntax" in skill_lower, 10),
    ]
    tool_score = sum(score for check, score in tool_checks if check)
    scores["tool_support"] = min(100, 60 + tool_score)

    # methodology: 方法论 - 检查分支设计、选择机制、剧本结构
    method_checks = [
        ("选择点" in skill_content, 15),
        ("分支" in skill_content and "深度" in skill_content, 15),
        ("菱形" in skill_content or "树形" in skill_content or "结构" in skill_content, 10),
        ("变量" in skill_content and ("状态" in skill_content or "condition" in skill_lower), 10),
    ]
    method_score = sum(score for check, score in method_checks if check)
    scores["methodology"] = min(100, 60 + method_score)

    # source_quality: 信源质量 - 检查是否有官方文档 vs 社区分享
    source_checks = [
        ("官方" in skill_content or "文档" in skill_content or "documentation" in skill_lower, 20),
        ("github" in skill_lower, 15),
        ("star" in skill_lower, 10),
    ]
    source_score = sum(score for check, score in source_checks if check)
    scores["source_quality"] = min(100, 60 + source_score)

    # platform_coverage: 平台覆盖 - 检查抖音/快手等平台特性
    platform_checks = [
        ("抖音" in skill_content, 15),
        ("快手" in skill_content, 15),
        ("时长" in skill_content and "平台" in skill_content, 10),
        ("3秒" in skill_content or "开头" in skill_content, 10),
    ]
    platform_score = sum(score for check, score in platform_checks if check)
    scores["platform_coverage"] = min(100, 60 + platform_score)

    # 计算总分 (按权重)
    weights = {
        "completeness": 0.20,
        "practicality": 0.20,
        "tool_support": 0.15,
        "methodology": 0.20,
        "source_quality": 0.10,
        "platform_coverage": 0.15
    }

    total = sum(scores[k] * weights[k] for k in weights)

    result = {
        "score": round(total, 2),
        "dimensions": scores,
        "passed": total >= 90,
        "reason": "基于 SKILL.md 内容评估"
    }

    print(json.dumps(result, indent=2, ensure_ascii=False))
    return result

if __name__ == '__main__':
    # 读取参数
    if len(sys.argv) < 3:
        print("Usage: evaluator.py <skill_file> <output_file>")
        sys.exit(1)

    skill_file = sys.argv[1]
    output_file = sys.argv[2]

    try:
        with open(skill_file, 'r') as f:
            skill_content = f.read()

        # 如果输出文件存在则读取，否则用 skill 作为输出
        try:
            with open(output_file, 'r') as f:
                output_content = f.read()
        except:
            output_content = skill_content

        result = evaluate(skill_content, output_content, None)
        print(json.dumps(result, indent=2, ensure_ascii=False))

    except Exception as e:
        print(json.dumps({"score": 50, "dimensions": {}, "error": str(e)}))