xyc
2026-08-14 1f4a6cc6045f47861c3e8d49a782afc3e11e3843
1
2
3
4
5
6
7
8
9
10
11
# -*- coding: utf-8 -*-
import zipfile, os, re
tmp = r'C:\Users\jcxiong\AppData\Local\Temp\ta_e2e_524cecf1'
zf = zipfile.ZipFile(os.path.join(tmp, 'gen_freight_rank.xlsx'))
xml = zf.read('xl/worksheets/sheet1.xml').decode('utf-8')
rows = re.findall(r'<row r="(\d+)"[^>]*>(.*?)</row>|<row r="(\d+)"[^>]*/>', xml)
print('total <row> elements:', len(rows))
for m in rows:
    r = m[0] or m[2]
    cells = len(re.findall(r'<c ', m[1] if m[1] else ''))
    print('row', r, 'cells:', cells)