# -*- coding: utf-8 -*- import io, sys for f in sys.argv[1:]: c = io.open(f, encoding='utf-8').read() print('===', f.replace('\\', '/').split('/')[-1]) for label, needle in [ ('CSS .prob-detail', '.prob-detail{border'), ('CSS .prob-body', '.prob-body{display:none'), ('CSS .prob-code', '.prob-code{background'), ('📋 题目', '📋 题目'), ('👦 孩子的代码', '👦 孩子的代码'), ('✅ 参考答案', '✅ 参考答案'), ('💡 解析', '💡 解析'), ('题目讲解标题', '题目讲解'), ]: print(' %-18s count=%d' % (label, c.count(needle))) # show a real panel sample (find first prob-body, strip tags for a peek) import re m = re.search(r'
(.*?)
', c, re.S) if m: txt = re.sub(r'<[^>]+>', ' ', m.group(1)) txt = re.sub(r'\s+', ' ', txt) print(' first-panel text snippet:', txt[:180])