Date/Time parsing in C++

Question

While doing the data/time parsing in c++ (converting a string in any format to a date), i found the following useful methods

1) strptime() - here the %d, %m etc can have either 1 or 2 characters. The function will take care of that. As a result of this it will enforce that we use a separator between two conversion specifiers. Ex: Its not valid to give %d%m it has be to %d/%m or any other separator. Also this does not support timezones.

2) Boost date IO - here the %d, %m has to have 2 characters. Now, the input string i get is not guaranteed to have this. As a result, its not possible to use this successfully. However, this does seem to support timezone, but not sure. because it says for inputs it does support timezone

So i am planning to use both in conjunction to determine the date. But i would like to get one where i can take into account the timezone as well. But none seems to be supporting that.

Does anybody have a suggestion? Rgds, AJ

Answer 1

As you noticed strptime doesn't directly support timezones. All it does is read formatted timezone-less data into a tm structure. You'll have to parse an optional timezone indicator yourself (unless it's fixed), and then set TZ and use mktime to convert into a time_t. If you set tm.is_dst to -1 you can even ask that mktime try to figure out the DST for you automatically.

Alternately you could construct your own parser using steams (boost supports a few formats, but may not be general enough). Again like above you can use mktime to compose a time_t.

Answer 2

Depends on the libc version, I'd say, and of course what you mean by `taking into account': Is it supposed to read the additional info and then ignore it, is it supposed to store the parsed info in the tm struct, or is it meant to convert to system/environment time?

glibc 2.10.1's strptime() does time zones, from the info page

`%z'
      The offset from GMT in ISO 8601/RFC822 format.

`%Z'
      The timezone name.

      _Note:_ Currently, this is not fully implemented.  The format
      is recognized, input is consumed but no field in TM is set.

Answer 3

@AJ: Added another answer so the code gets formatted

#include <stdio.h>
#include <sys/time.h>
#include <time.h>

int
main(void)
{
        struct tm tm[1] = {{0}};

        strptime("Tue, 19 Feb 2008 20:47:53 +0530", "%a, %d %b %Y %H:%M:%S %z", tm);
        fprintf(stdout, "off %ld\n", tm->tm_gmtoff);
        return 0;
}

And a run looks like (glibc 2.10.1):

freundt@clyde:pts/28:~/temp> ./test
off 19800