#!/usr/bin/env python3
"""Benchmark 评估器 - 互动视频短剧选题"""

import json
import sys
import re

def evaluate(result_json, benchmark_file, testdata_dir):
    """评估选题结果"""

    try:
        result = json.loads(result_json)
    except:
        return {
            "score": 50,
            "dimensions": {},
            "reason": "无法解析输出JSON"
        }

    scores = {}

    # D1: 题材适配度 (20%)
    # 检查是否有冲突性、情感张力、讨论空间
    summary = result.get('summary', '')
    title = result.get('title', '')

    # 高冲突性特征：争吵、矛盾、冲突、出轨、离婚、分手
    conflict_indicators = ['矛盾', '冲突', '争吵', '分歧', '对立']
    has_conflict = any(word in summary or word in title for word in conflict_indicators)

    # 情感张力特征
    emotion_words = ['崩溃', '愤怒', '无奈', '绝望', '爆发', '哭泣']
    has_emotion = any(word in summary or word in title for word in emotion_words)

    if has_conflict and has_emotion:
        scores['D1'] = 90
    elif has_conflict:
        scores['D1'] = 75
    else:
        scores['D1'] = 55

    # D2: 剧情潜力 (20%)
    # 检查是否有角色、场景、分支点
    roles = result.get('roles', [])
    scenes = result.get('scenes', [])
    branch_points = result.get('branch_points', [])

    if len(roles) >= 2 and len(branch_points) >= 1:
        scores['D2'] = 90
    elif len(roles) >= 2:
        scores['D2'] = 75
    else:
        scores['D2'] = 55

    # D3: 可演绎性 (15%)
    # 检查推荐形式是否合理
    form = result.get('recommended_form', '')
    if form in ['对话形式', '对话形式为主', '对话+旁白混合']:
        scores['D3'] = 90
    elif form in ['旁白形式', '旁白为主']:
        scores['D3'] = 70
    else:
        scores['D3'] = 60

    # D4: 受众广度 (15%)
    # 检查目标人群是否广泛
    # 婆媳、职场类话题通常受众广泛
    if '婆媳' in summary or '婆媳' in title or '家庭' in result.get('recommended_interactive', ''):
        scores['D4'] = 85
    else:
        scores['D4'] = 70

    # D5: 互动空间 (20%)
    # 检查分支设计
    interactive = result.get('recommended_interactive', '')
    if interactive in ['AB选择', '投票PK', '多结局'] and len(branch_points) > 0:
        scores['D5'] = 90
    elif interactive:
        scores['D5'] = 75
    else:
        scores['D5'] = 50

    # D6: 时效周期 (10%)
    # 婆媳是永恒话题
    if '婆媳' in summary or '婆媳' in title or '家庭' in summary:
        scores['D6'] = 85
    else:
        scores['D6'] = 70

    # 计算总分
    total = (scores['D1'] * 0.20 + scores['D2'] * 0.20 +
             scores['D3'] * 0.15 + scores['D4'] * 0.15 +
             scores['D5'] * 0.20 + scores['D6'] * 0.10)

    result = {
        "score": round(total, 2),
        "dimensions": scores,
        "breakdown": {
            "D1_题材适配度": f"{scores['D1']} × 20% = {scores['D1'] * 0.20:.1f}",
            "D2_剧情潜力": f"{scores['D2']} × 20% = {scores['D2'] * 0.20:.1f}",
            "D3_可演绎性": f"{scores['D3']} × 15% = {scores['D3'] * 0.15:.1f}",
            "D4_受众广度": f"{scores['D4']} × 15% = {scores['D4'] * 0.15:.1f}",
            "D5_互动空间": f"{scores['D5']} × 20% = {scores['D5'] * 0.20:.1f}",
            "D6_时效周期": f"{scores['D6']} × 10% = {scores['D6'] * 0.10:.1f}"
        },
        "reason": "基于 benchmark_v1.md 6维度评估"
    }

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

if __name__ == '__main__':
    result_file = '/tmp/result.json'
    with open(result_file, 'r') as f:
        result_json = f.read()
    evaluate(result_json, '', '')