bssoupblog

bssoupblog2.py

import re

wht_date = soup.select(".wht_date")
wht_date = [each.get_text() for each in wht_date[1:]]

wht_items = soup.select(".wht_items")
wht_items = [each.get_text().replace("\n", "_").replace("\t", "") for each in wht_items[1:]]

wht_type = soup.select(".wht_type")
wht_type = [each.get_text().replace("\n", "_").replace("\t", "") for each in wht_type[1:]]

wht_total = soup.select(".wht_total")
wht_total = [float(re.findall(r'\d+.\d+', each.get_text().strip().replace(',', ''))[0]) for each in wht_total[1:]]