204 lines
7.8 KiB
Python
204 lines
7.8 KiB
Python
# -*- coding: utf-8 -*-
|
|
"""第三轮4部短故事全面质量检查"""
|
|
import re, os
|
|
|
|
rr = re.compile(r'[\u4e00-\u9fff]')
|
|
|
|
repos = [
|
|
'r3-yue-tang-ping-yue-qiang',
|
|
'r3-shen-ming-zhong-jie',
|
|
'r3-man-ji-da-lao-hui-xin-shou-cun',
|
|
'r4-tun-jin-shou-xi-tong',
|
|
]
|
|
|
|
repo_names = {
|
|
'r3-yue-tang-ping-yue-qiang': '故事1：越躺平越强',
|
|
'r3-shen-ming-zhong-jie': '故事2：神明中介',
|
|
'r3-man-ji-da-lao-hui-xin-shou-cun': '故事3：满级大佬回新手村',
|
|
'r4-tun-jin-shou-xi-tong': '故事4：吞金兽系统',
|
|
}
|
|
|
|
results = {}
|
|
|
|
for repo in repos:
|
|
D = os.path.join('/app/working/workspaces/default/check_tmp', repo)
|
|
name = repo_names[repo]
|
|
|
|
print(f'\n{"="*60}')
|
|
print(f' {name}')
|
|
print(f' 仓库: http://git.qihuyun.com/qyyfw/{repo}')
|
|
print(f'{"="*60}')
|
|
|
|
# 1. 文件完整性
|
|
files = sorted([f for f in os.listdir(D) if f.endswith('.md') and '简介' in f])
|
|
ch_files = sorted([f for f in os.listdir(D) if f.endswith('.md') and f.startswith('第')])
|
|
|
|
print(f'\n 📋 章节完整性: {len(ch_files)}章')
|
|
if len(ch_files) != 15:
|
|
print(f' ❌ 应15章，实{len(ch_files)}章')
|
|
else:
|
|
print(f' ✅ 15章齐全')
|
|
|
|
# 2. 汉字字数
|
|
print(f'\n 📊 汉字字数:')
|
|
total_h = 0
|
|
all_ok = True
|
|
low_chs = []
|
|
for f in ch_files:
|
|
with open(os.path.join(D, f), 'r', encoding='utf-8') as fp:
|
|
h = len(rr.findall(fp.read()))
|
|
total_h += h
|
|
if h < 2000:
|
|
all_ok = False
|
|
low_chs.append((f, h))
|
|
print(f' ❌ {f}: {h} (不足2000)')
|
|
|
|
avg = total_h // len(ch_files) if ch_files else 0
|
|
if all_ok:
|
|
print(f' ✅ 全部≥2000，总计{total_h}汉字，平均{avg}字/章')
|
|
else:
|
|
print(f' ⚠️ {len(low_chs)}章不足2000，总计{total_h}汉字')
|
|
for f, h in low_chs:
|
|
print(f' 需补+{2000-h}: {f}')
|
|
|
|
# 3. 敏感词
|
|
print(f'\n 🔍 敏感词检查:')
|
|
sensitive_words = ['裁员', '被裁', '辞退', '集团', '企业']
|
|
# '公司'特殊处理：仅标记负面上下文
|
|
has_issues = False
|
|
for f in ch_files:
|
|
with open(os.path.join(D, f), 'r', encoding='utf-8') as fp:
|
|
c = fp.read()
|
|
for w in sensitive_words:
|
|
if w in c:
|
|
print(f' ❌ {f}: 含敏感词「{w}」')
|
|
has_issues = True
|
|
|
|
# 检查'公司'是否出现在负面上下文
|
|
for f in ch_files:
|
|
with open(os.path.join(D, f), 'r', encoding='utf-8') as fp:
|
|
lines = fp.readlines()
|
|
for i, l in enumerate(lines):
|
|
if '公司' in l:
|
|
# 简单检查上下文是否负面
|
|
negative_indicators = ['倒闭', '破产', '裁员', '压榨', '剥削', '违法', '纠纷', '官司', '欠薪', '跑路']
|
|
context = l + (lines[i+1] if i+1 < len(lines) else '') + (lines[i-1] if i > 0 else '')
|
|
for ni in negative_indicators:
|
|
if ni in context:
|
|
print(f' ⚠️ {f}: 「公司」在负面上下文(行{i+1}) - {ni}')
|
|
has_issues = True
|
|
break
|
|
|
|
if not has_issues:
|
|
print(f' ✅ 零敏感词问题')
|
|
|
|
# 4. AI味检查
|
|
print(f'\n 🤖 AI味检查:')
|
|
ai_issues = []
|
|
for f in ch_files:
|
|
with open(os.path.join(D, f), 'r', encoding='utf-8') as fp:
|
|
lines = fp.readlines()
|
|
c = ''.join(lines)
|
|
issues = []
|
|
|
|
cnt = c.count('燃点')
|
|
if cnt: issues.append(f'燃点×{cnt}')
|
|
|
|
cnt = len(re.findall('深吸[了一]*口气', c))
|
|
if cnt > 2: issues.append(f'深吸气×{cnt}(>2)')
|
|
elif cnt > 0: issues.append(f'深吸气×{cnt}')
|
|
|
|
cnt = len(re.findall('(握紧|攥紧)', c))
|
|
if cnt > 4: issues.append(f'握紧/攥紧×{cnt}(>4)')
|
|
elif cnt > 0: issues.append(f'握紧/攥紧×{cnt}')
|
|
|
|
cnt = sum(1 for l in lines if l.strip() == '——' and len(l.strip()) == 2)
|
|
if cnt: issues.append(f'独立——行×{cnt}')
|
|
|
|
cnt = c.count('（全书完）')
|
|
if cnt > 1: issues.append(f'（全书完）重复×{cnt}')
|
|
|
|
cnt = len(re.findall(r'第.*章.*完|本.*章.*完|本章完', c))
|
|
if cnt: issues.append(f'章末标注×{cnt}')
|
|
|
|
cnt = len(re.findall('(在我看来|不得不说|毫不夸张地说|可想而知|毋庸置疑)', c))
|
|
if cnt > 3: issues.append(f'meta自嘲×{cnt}(>3)')
|
|
elif cnt > 0: issues.append(f'meta自嘲×{cnt}')
|
|
|
|
if issues:
|
|
ai_issues.append(f'{f}: {" | ".join(issues)}')
|
|
|
|
if ai_issues:
|
|
for s in ai_issues:
|
|
print(f' ⚠️ {s}')
|
|
else:
|
|
print(f' ✅ 零AI味问题')
|
|
|
|
# 5. 开篇核弹 (第1章前500字)
|
|
print(f'\n 💣 开篇核弹V2.0检查:')
|
|
if ch_files:
|
|
with open(os.path.join(D, ch_files[0]), 'r', encoding='utf-8') as fp:
|
|
ch1 = fp.read()
|
|
|
|
first_100 = ch1[:300] # about 100 hanzi
|
|
first_300 = ch1[:900] # about 300 hanzi
|
|
first_500 = ch1[:1500] # about 500 hanzi
|
|
|
|
h_100 = len(rr.findall(first_100))
|
|
h_300 = len(rr.findall(first_300))
|
|
h_500 = len(rr.findall(first_500))
|
|
|
|
print(f' 前100字区域({h_100}汉字): {"✅" if h_100 >= 50 else "❌"}')
|
|
print(f' 前300字区域({h_300}汉字): {"✅" if h_300 >= 200 else "❌"}')
|
|
print(f' 前500字区域({h_500}汉字): {"✅" if h_500 >= 400 else "❌"}')
|
|
|
|
# 检查前100字是否有冲突
|
|
conflict_keywords = ['分手', '被甩', '完蛋', '完了', '被裁', '开除', '被骂', '破产', '欠债', '被嘲', '被笑', '被打了', '冲突', '矛盾', '决裂', '背叛', '被坑', '被骗']
|
|
has_conflict = any(kw in first_100 for kw in conflict_keywords)
|
|
print(f' 前100字出冲突: {"✅ 有" if has_conflict else "⚠️ 未检测到明确冲突"}')
|
|
|
|
# 检查前300字是否兑现书名
|
|
print(f' 前300字兑现书名: 需人工判断标题核心要素是否出现')
|
|
|
|
# 差异化钩子
|
|
spec_hooks = ['系统', '觉醒', '金手指', '绑定', '激活', '穿越', '重生', '特殊能力', '不一样', '别人不知道']
|
|
has_hook = any(kw in first_500 for kw in spec_hooks)
|
|
print(f' 前500字差异化钩子: {"✅ 有" if has_hook else "⚠️ 未检测到明确钩子"}')
|
|
|
|
# 6. 剧情连贯性抽样
|
|
print(f'\n 🔗 剧情连贯性(抽样):')
|
|
sample_gaps = []
|
|
for i in range(1, len(ch_files)):
|
|
with open(os.path.join(D, ch_files[i-1]), 'r', encoding='utf-8') as fp:
|
|
prev_end = fp.readlines()[-5:]
|
|
with open(os.path.join(D, ch_files[i]), 'r', encoding='utf-8') as fp:
|
|
curr_start = fp.readlines()[:5]
|
|
|
|
prev_last = ''.join(prev_end)[-200:]
|
|
curr_first = ''.join(curr_start)[:200]
|
|
|
|
# 简单检查是否有明显断裂（主题跳跃）
|
|
# 仅做提示
|
|
print(f' 章与章之间剧情连续(需人工通读确认)')
|
|
|
|
# 综合评级
|
|
score = 0
|
|
if len(ch_files) == 15: score += 20
|
|
if all_ok: score += 30
|
|
if not has_issues: score += 20
|
|
if not ai_issues: score += 20
|
|
|
|
grade = 'S' if score >= 85 else ('A' if score >= 70 else 'B')
|
|
print(f'\n 🏆 综合评分: {score}/90 → {grade}级')
|
|
results[repo] = {'score': score, 'grade': grade, 'total_h': total_h, 'avg': avg}
|
|
|
|
print(f'\n{"="*60}')
|
|
print(f' 第三轮4部短故事全面检查报告')
|
|
print(f'{"="*60}')
|
|
print(f'{"故事":<20} {"汉字":<8} {"平均":<8} {"评分":<8} {"等级":<8}')
|
|
print(f'{"-"*52}')
|
|
for repo, r in results.items():
|
|
n = repo_names[repo]
|
|
print(f'{n:<20} {r["total_h"]:<8} {r["avg"]:<8} {r["score"]:<8} {r["grade"]:<8}')
|
|
print(f'{"-"*52}')
|