# -*- coding:utf-8 -*-
from datetime import datetime, timedelta
import re
import time
def tz_offset(tz):
res = (re.search(r'(?P<F>[-+])(?P<H>\d{2}):?(?P<M>\d{2})', tz) or re.search('', '')).groupdict()
offset = (1 if res.get('F', '+')=='+' else -1) * timedelta(
hours = int(res.get('H', 0)),
minutes = int(res.get('M', 0)))
return offset
def parse_date(data, fmt, tz):
"""
时间匹配模块，可转化为固定格式
返回时间字符串 0000-00-00 00:00:00
可匹配结构 |今天~前天， 几天前，分钟秒前等 | 2017-1-4 12:10 | 2017/1/4 12:10 | 2018年4月2日 12:12
| 2018年4月2日 | 2017-1-4 | 2017/1/4 | 1/4 |
"""
offset = tz_offset(tz)
if fmt == 'auto':
now = (datetime.utcnow() + timedelta(hours=8)).replace(microsecond=0) + offset
now_1 = now - timedelta(days=1)
now_2 = now - timedelta(days=2)
# 几/刚/今天/昨天/前天
x = data.strip()
x = x.replace(u'几', ' 0 ')
x = x.replace(u'刚[刚才]', now.strftime(' %Y-%m-%d %H:%M:%S '))
x = x.replace(u'今天', now.strftime(' %Y-%m-%d '))
x = x.replace(u'昨天', now_1.strftime(' %Y-%m-%d '))
x = x.replace(u'前天', now_2.strftime(' %Y-%m-%d '))
x = re.sub(r'[年月]', '/', x)
x = re.sub(r'[日]', ' ', x)
x = re.sub(r'\s{2,}', r' ', x)
# XX前
res = (re.search(r'(?P<S>\d+)\s*秒钟?前', x) \
or re.search(r'(?P<M>\d+)\s*分钟前', x) \
or re.search(r'(?P<H>\d+)\s*小时前', x) \
or re.search(r'(?P<d>\d+)\s*天前', x) \
or re.search('', '')).groupdict()
if res:
dt = now - timedelta(
days=int(res.get('d', 0)),
hours=int(res.get('H', 0)),
minutes=int(res.get('M', 0)),
seconds=int(res.get('S', 0))
)
# 不是几天前分钟前的形式
else:
# XX-XX-XX XX:XX:XX
res = (re.search(r'(?P<Y>\d+)[/-](?P<m>\d+)[/-](?P<d>\d+)(\s+(?P<H>\d{1,2}):(?P<M>\d{2})(:(?P<S>\d{2}))?)?',
x) or re.search('', '')).groupdict()
if res == dict():
# 匹配没有年份的时候，格式 XX-XX XX:XX:XX 月-日 时：分：秒 或 17年10月10日 时：分：秒
res = (re.search(
r'(?P<m>\d{1,2})[/-](?P<d>\d+)(\s+(?P<H>\d{2}):(?P<M>\d{2})(:(?P<S>\d{2}))?)?',
x) or re.search('', '')).groupdict()
if res:
Y = res.get('Y', now.year)
Y = "20" + Y if len(str(Y)) == 2 else Y
m = res.get('m', now.month)
d = res.get('d', now.day)
H = res.get('H', now.hour)
M = res.get('M', now.minute)
S = res.get('S', 0)
dt = datetime(
year=int(Y) if Y != None and 1987 <= int(Y) <= now.year else now.year,
month=int(m) if m != None else now.month,
day=int(d) if d != None else now.day,
# 如果没有时分秒，则被认定为00:00:00
hour=int(H) if H != None else 0,
minute=int(M) if M != None else 0,
second=int(S) if S != None else 0
)
else:
# 1970-01-01 00:00:00
# dt = datetime.utcfromtimestamp(0)+offset
return ""
# 时间可能超过当前时间，若超过则减去一年
if int(time.mktime((dt - offset).timetuple())) > int(time.time()):
# 时间超过当前时间，减去一年
delta = timedelta(days=-365)
real_time = (dt - offset) + delta
real_time = real_time.strftime("%Y-%m-%d %H:%M:%S")
else:
real_time = (dt - offset).strftime("%Y-%m-%d %H:%M:%S")
return real_time
if __name__ == '__main__':
print(parse_date('2秒前', 'auto', ''))
print(parse_date('2分钟前', 'auto', ''))
print(parse_date('2小时前', 'auto', ''))
print(parse_date('昨天 00:30', 'auto', ''))
print(parse_date('07-20', 'auto', ''))
|
