1 import xlrd
2 import xlwt
3 import re
4 def dataDump():
5 workbook=xlrd.open_workbook("C:\\Users\\AQB\\Downloads\\网站基本信息20201013 (2).xls")
6 table=workbook.sheet_by_index(0)
7 urllist=[]
8 namelist=[]
9 punc = '[;|,\']'
10 for i in range(1,table.nrows):
11 url=table.cell(i,1).value
12 name=table.cell(i,0).value
13 s=re.split(punc,url)
14 for i in s:
15 if i is not None:
16 namelist.append(name)
17 url="http://"+i
18 #print(url)
19 urllist.append(url)
20 #namelist.append(name)
21 #print(namelist)
22 data_write("C:\\Users\\AQB\\Desktop\\白名单提取.xlsx",namelist,urllist)
23
24 def data_write(file_path, datas,urls):
25 f = xlwt.Workbook()
26 sheet1 = f.add_sheet(u'sheet1',cell_overwrite_ok=True) #创建sheet
27 #将数据写入第 i 行,第 j 列
28 i = 0
29 j=0
30 for data in datas:
31 print(data)
32 sheet1.write(i,j,data)
33 i = i + 1
34 i = 0
35 j=1
36 for data in urls:
37 print(data)
38 sheet1.write(i,j,data)
39 i = i + 1
40 f.save(file_path) #保存文件
41
42 dataDump()