Simple timestamp generation using POSIX strftime in Perl
Very often I need to add a timestamp to some log file, or create a file based on a the current date. There are plenty of really good modules on CPAN to handle dates and times, that can help creating these timestamps, but this time I am going to use the strftime function of the standard POSIX module.
The basic usage of strftime, as also explained in the documentation looks like this:
strftime(format, sec, min, hour, mday, mon, year, wday = -1, yday = -1, isdst = -1)
Where the "format" can use various placeholder such as "%d" from this list aAbBcdHIjmMpSUwWxXyYZ%, but unfortunately it is not explained in the documentation what each of these mean.
The sec, min, etc values are the seconds, minutes, etc. values of the time that we would like to use.
These numbers can come from any place, but they are actually corresponding to the values returned by either the localtime or the gmtime functions in LIST context.
A simple example would be:
use warnings; use 5.010; use POSIX qw(strftime); say strftime '%Y-%m-%d', gmtime(); # 2014-11-09
say prints) a string based on the years, month and days as it is in Greenwich at the time of running the script.
If you are creating logfiles, or basically any other timestamp, it is probably much better to use the YYYY-MM-DD format instead of the British DD/MM/YYYY or the American MM/DD/YYYY. Neither of those can be easily sorted and because both exists they are extremely confusing. The YYYY-MM-DD is how Hungarians write it and therefore it is much better. (Remember other Hungarian Notations, right?). It also happens to correspond to the ISO standard.
Another very useful format looks like this:
say strftime '%Y-%m-%d-%H-%M-%S', gmtime(); # 2014-11-09-15-31-13
This is the YYYY-MM-DD-HH-MM-SS (though I know the duplicating of MM in this example is a bit confusing so you might want to think about it as YYYY-mm-dd-HH-MM-SS that will correspond to both the format and the letters used to create that format.
In a nutshell, it is Year, Month, Day, Hour, Minute, Second with 0 padding where necessary.
It is very useful to create timestamps even though probably less useful as filename. (Do you really want a new file every second?)
Oh and if you hope that this will be unique, then don't do it. With any luck after an hour of no file generation there will be two at the exact same second and then all the "uniqueness" is gone.
Of course when you are creating a file you don't necessarily need to use dashes to separate the parts. This would be great as well:
say strftime '%Y%m%d', gmtime(); # 20141109
In the following script you'll see examples for the various placeholder that can be used in the format string. I've tried to group them according to the level of usefulness,. At least what I see of the level of usefulness.
In this example I also used a fixed timestamp which then was converted to the appropriate array using the gmtime function. This was only necessary so the the values I show will really correspond to the same time and one that will show the values in an interesting way. (0 padded where necessary.)
use strict; use warnings; use 5.010; use POSIX qw(strftime); my $time = 1415520307; my @time = gmtime($time); # Very useful: say strftime '%Y', @time; # 2014 say strftime '%m', @time; # 11 say strftime '%d', @time; # 09 say strftime '%H', @time; # 08 say strftime '%M', @time; # 05 say strftime '%S', @time; # 07 # Sometimes useful: say strftime '%j', @time; # 313 say strftime '%U', @time; # 45 say strftime '%W', @time; # 44 say strftime '%w', @time; # 0 say strftime '%%', @time; # % say strftime '%c', @time; # Sun Nov 9 08:05:07 2014 # IMHO not very useful because they depend on the current locale. say strftime '%I', @time; # 08 say strftime '%p', @time; # AM say strftime '%X', @time; # 08:05:07 say strftime '%x', @time; # 11/09/2014 say strftime '%y', @time; # 14 say strftime '%Z', @time; # IST (I have no idea why) say strftime '%A', @time; # Sunday say strftime '%b', @time; # Nov say strftime '%B', @time; # November
Explanation Example %Y Year YYYY 2014 %m Month mm 11 %d Day dd 09 %H Hour HH 08 %M Min MM 05 %S Seconds SS 07
Explanation Example %j Day of the year as a zero-padded decimal number. 313 On January 1 this will be 001 %U Week number of the year as a zero padded decimal number. 45 (Sunday as the first day of the week) All days in a new year preceding the first Sunday are considered to be in week 0 %W Week number of the year as a decimal number. 44 (Monday as the first day of the week) All days in a new year preceding the first Monday are considered to be in week 0. %w Weekday as a decimal number, where 0 is Sunday and 6 is Saturday. 0 %% Literal % % %c Locale’s appropriate date and time representation. Sun Nov 9 08:05:07 2014 I think this is what gmtime() would return in scalar context.
IMHO not very useful because they depend on the current locale.
Explanation Example %I Hour (12-hour clock) 08 %p Locale’s equivalent of either AM or PM. AM %X Locale’s appropriate time representation. 08:05:07 %x Locale’s appropriate date representation. 11/09/2014 %y Year without century as a zero-padded decimal number. 14 %Z Time zone name IST (I have no idea why) %A Weekday as locale’s full name. Sunday %b Month as locale’s abbreviated name. Nov %B Month as locale’s full name. November
See also strftime in Python that I used as the source of some of the explanations.