Comparing RFC1123 based Dates

Douglas Wells see at signature.invalid
Sun Aug 5 09:22:02 EDT 2007


In article <1186284911.988459.324690 at e16g2000pri.googlegroups.com>,
  Phoe6 <orsenthil at gmail.com> writes:
> Phoe6 wrote:
> > I would like to parse RFC 1123 date format and compare two dates. I
> > find that
> > datetime module does not specifically confirms to any RFC. Any
> > suggestions as how I can handle the RFC 1123 date format using
> > standard libraries before I go to re based parsing?
> 
> Well,
> >>> import time
> >>> timeobj = time.strptime("Thu, 01 Dec 1994 16:00:00 GMT","%a, %d %b %Y %H:%M:%S %Z")
> 
> was easy.

Well, it might have been easy, but it's got several gotchas (in
both Python and C), including:

  - The zone field (%Z) only corresponds for the GMT and UT timezones,
    which are obsolete (see RFC 2822).  There is no support for
    the recommended +/-time-offset form.

  - The day-of-week (%a) and month (%b) fields in strptime and
    strftime are subject to the process's locale, whereas the RFC
    time forms are not.  Those are hardwired to names that happen
    to correspond to the C, POSIX, and probably most of the en_*
    locales, but not to others that would be used by billions of
    people.  Thus anyone using your program who doesn't happen to
    reside in one of the English-speaking countries (or does and
    is using a native locale) is likely to encounter problems when
    using your program.

  - The day-of-week field is optional.

  - Comments are allowed (but deprecated) in the whitespace fields
    of the time format.  (On the other hand, I've never seen this
    is normal e-mail.)

I find the use of strptime and strftime difficult enough to manage
with Internet date/times that I restrict my use of them to programs
that are limited to processing date/times.  Even then, I then
explicitly set the locale (LC_TIME) to the "C" locale.  Otherwise,
I use ad hoc code that explicitly recognizes the RFC-defined forms.

-- 
.   Douglas Wells             .  Connection Technologies      .
.   Internet:  -sp9804- -at - contek.com-                     .



More information about the Python-list mailing list