【大数据】字符串、文件操作，英文词频统计预处理

code = input(‘请输入身份证号码:‘)
province = code[0:2]
city = code[2:4]
#area = code[4:6]
sz_area = {‘03‘: ‘罗湖区‘, ‘04‘: ‘福田区‘, ‘05‘: ‘南山区‘, ‘06‘: ‘宝安区‘, ‘07‘: ‘龙岗区‘, ‘08‘: ‘盐田区‘}
year = code[6:10]
month = code[10:12]
day = code[12:14]
sex = code[16]
if int(sex) % 2 == 0:
    sex = ‘女‘
else:
    sex = ‘男‘
if len(code) == 18:
    print(‘你输入的身份证号码为‘ + code)
    if province != ‘44‘:
        print(‘该身份证所属的地区在广东省外‘)
    else:
        if city != ‘03‘:
            print(‘该身份证所属的地区在广东省深圳市以外‘)
        else:
            print(‘该身份证所属的地区为广东省深圳市{}‘.format(sz_area[code[4:6]]))
    print(‘出生日期是‘ + year + ‘年‘ + month + ‘月‘ + day + ‘日‘ + ‘ 性别为‘ + sex)
else:
    print(‘你输入的身份证号码有误‘)

text = input(‘请输入明文: ‘)
k = int(input(‘请输入位移值: ‘))
s = ord("a")
e = ord("z")
choose = input("编码(1) 解码(2):")
print("凯撒密码编码：", end="")
for i in text:
    if s <= ord(i) <= e:
        if choose == "1":
            print(chr(s+(ord(i)-s+int(k)) % 26), end="")
        elif choose == "2":
            print("凯撒密码解码：", end="")
            print(chr(s + (ord(i)-s-int(k)) % 26), end="")
        else:
            print("你的选择有误")
    else:
        print(i, end="")

for i in range(2, 8):
    url = ‘http://news.gzcc.cn/html/xiaoyuanxinwen/{}.html‘.format(i)
    print(url) 

f = open(‘If I Die Young.txt‘, ‘r‘)
text = f.read()
f.close()
text = text.lower()
print(text)
sep = ‘,?.!-:_‘
for s in sep:
    text = text.replace(s, ‘ ‘)
print(text.split())
print(text.count(‘die‘), text.count(‘time‘), text.count(‘on‘))