# 目标：需要提取文本中的手机号和其他所有数字
# 文本： QQ：123456扣扣5625231和12
# 结果：[123456],[5625231],[12]
import re
import os
# 需要处理的文本
filename1 = r'E:\.txt'
mobileblack = r"E:\.txt"
qqblack = r"E:\.txt"
if os.path.exists(mobileblack):
os.remove(mobileblack)
if os.path.exists(qqblack):
os.remove(qqblack)
f1 = open(filename1, 'r', encoding='utf-8')
def strreplace(stra):
str1 = stra.replace("一","1")
str2 = str1.replace("二", "2")
str3 = str2.replace("三", "3")
str4 = str3.replace("四", "4")
str5 = str4.replace("五", "5")
str6 = str5.replace("六", "6")
str7 = str6.replace("七", "7")
str8 = str7.replace("八", "8")
str9 = str8.replace("九", "9")
str10 = str9.replace("壹", "1")
str11 = str10.replace("贰", "2")
str12 = str11.replace("叁", "3")
str13 = str12.replace("肆", "4")
str14 = str13.replace("伍", "5")
str15 = str14.replace("陆", "6")
str16 = str15.replace("柒", "7")
str17 = str16.replace("捌", "8")
str18 = str17.replace("玖", "9")
str19 = str18.replace("幺", "1")
str20 = str19.replace("零", "0")
str21 = str20.replace("①", "1")
str22 = str21.replace("②", "2")
str23 = str22.replace("③", "3")
str24 = str23.replace("④", "4")
str25 = str24.replace("⑤", "5")
str26 = str25.replace("⑥", "6")
str27 = str26.replace("⑦", "7")
str28 = str27.replace("⑧", "8")
string = str28.replace("⑨", "9")
return string
def writemobile(item):
f = open(mobileblack, 'a', encoding='utf-8')
f.write(item)
f.write("\n")
f.close()
def writeQQ(item):
f = open(qqblack, 'a', encoding='utf-8')
f.write(item)
f.write("\n")
f.close()
reader1 = f1.readlines()
for line in reader1:
stra = re.sub("[\s+\.\!\/><_,$‘’·`%:：�♦^♥❤*.(+\"\']+|[+——！，。？、~@br#￥%……&*（）]+|[-]", "", line)
string = strreplace(stra)
pattern = re.compile(u"1[34578]\d{9}")
phone = pattern.findall(string)
if phone:
writemobile(str(phone))
try:
mode = re.compile(r'\d+')
qq = mode.findall(string)
if qq:
writeQQ(str(qq))
except Exception as e:
print(e)
print('空值')
pass
