import requests,bs4,openpyxlfrom openpyxl.styles import Fontheader={'User-Agent':找自己浏览器的agent}n=0y=2wb=openpyxl.Workbook()ws=wb.activews.column_dimensions['A'].width=50ws.column_dimensions['B'].width=10ws['A1']='电影名称'ws['B1']='豆瓣评分'while n<251:#爬虫res = requests.get('/top250?start=%d&filter='%n, headers=header)res.raise_for_status()soup = bs4.BeautifulSoup(res.text, "html.parser")dx = soup.find_all('div', class_='hd')df=soup.select('.rating_num')for i in dx:ws['A'+str(y)]=i.a.span.textws['B'+str(y)]=df[dx.index(i)].texty+=1n += 25n=1while n<252:#设置单元格格式ws['A%d'%n].font = Font(name='微软雅黑', size=14)ws['B%d'%n].font = Font(name='微软雅黑', size=14,bold=True)n+=1wb.save('豆瓣top250.xlsx')
执行结果
