python解析时间格式脚本

对于这种时间格式：發表於: 星期一 五月 28, 2012 6:59 am

import re

INPUT = "發表於: 星期一 五月 28, 2012 6:59 am    文章主題: 對《大話新聞》改組的誠心思考／蔬菜麵"

pattern = re.compile(r'[\d]+')

b = re.findall(pattern, INPUT)

a = INPUT.split(' ')

monthdict = {"一月": "","二月": "", "三月": "", "四月": "", "五月": "", "六月": "",

                "七月": "",  "八月": "",  "九月": "",  "十月": "",  "十一月": "",  "十二月": ""}

year = a[4]

month = monthdict[a[2]]

day = b[0]

if a[6] == 'pm':

    hour = int(b[2].encode('utf-8')) + 12

hour= b[2]

min = b[3]

OUTPUT = "%s-%s-%s %s:%s:00"% (year, month, day, hour, min)

print OUTPUT

对于这种正常的时间格式   http://www.cdnews.com.tw　2015-11-02 17:33:55

import re

INPUT="http://www.cdnews.com.tw　2015-11-02 17:33:55"

pattern = re.compile(r'[\d]+')

a = re.findall(pattern, INPUT)

year = a[0]

month = a[1]

day = a[2]

hour = a[3]

minute = a[4]

second = a[5]

OUTPUT = "%s-%s-%s %s:%s:%s" % (year,month,day,hour,minute,second)

print OUTPUT

对于这种时间格式  發表於: 星期三 十二月 14, 2016 6:45 pm

import re

INPUT = "發表於: 星期三 十二月 14, 2016 6:45 pm"

pattern = re.compile(r'[\d]+')

b = re.findall(pattern, INPUT)

a = INPUT.split(' ')

monthdict = {"一月": "","二月": "", "三月": "", "四月": "", "五月": "", "六月": "","七月": "",  "八月": "",  "九月": "",  "十月": "",  "十一月": "",  "十二月": ""}

year = a[4]

month = monthdict[a[2]]

day = b[0]

if a[6] == 'pm':

    hour = int(b[2].encode('utf-8')) + 12

elif a[6] == 'am':

    h = int(b[2])

    if h >= 10:

        hour = h

    elif h<10:

        hour= ""+b[2]

min = b[3]

OUTPUT = "%s-%s-%s %s:%s:00"% (year, month, day, hour, min)

print OUTPUT