使用strptime，一天超出了一个月的范围

Question

I have long list of date in dataframe that need to be converted into datetime, the date is in the form "%d%m%Y" , and I apply datetime.strptime(x,'%d%m%Y') , which works fine until meet the date "3122012" , which should be datetime.datetime(2012, 12, 3, 0, 0) , but instead it throw the error message:

day is out of range for month

which is obviously because the program consider it as 31/2/2012, which doesn't exist. any suggestions to avoid such problems?Thanks!

Answer 1

This is just part of a much larger problem: Your format is inherently ambiguous. ^*

For example, both 1-11-2012 and 11-1-2012 will format as 1112012 . So, how are you going to parse that? You can invent and implement a disambiguation rule for that, but whichever one you choose, the other date can no longer be represented by your system.

If that's OK with you, then whatever rule you implement for disambiguating 1112012 will automatically handle 3122012 as well. For example:

def parse_dmy(s):
    if len(s) == 6: s = '0' + s[0] + '0' + s[1:]
    elif len(s) == 7: s = '0' + s # or s[0] + '0' + s[1:]
    return datetime.strptime(x, '%d%m%Y')

I picked the '0' + s rule rather than the s[0] + '0' + s[1:] rule because the former gives you 03-12-2012 for this example, while the latter gives you 31-02-2012 . But, again, neither rule works for all values.

---

_{* Also, %d and %m explicitly mean "0-padded", so your strings are technically invalid… but most platforms are happy with non-padded numbers in, eg, %d-%m-%Y —I believe POSIX requires them to be happy, and Windows is happy, and what other non-POSIX platform do you care about? Anyway, that would just be a minor problem if not for the larger problem that there is no way to unambiguously convert your strings into a valid format.}