How to get the date and time for an arbitrary time since epoch?

[jmakkay]

I'm currently working on a code that requires that I read a file which title is on a ddMMyyyyHHmm format (e.g., 060420181720.txt.) It uses the
ctime library to get the current time and then convert it to strings.

This is the relevant snippet of my code.

	string dd, MM, yyyy, HH, mm, datetime;
	
	time_t now = time(0);
	tm *ltm = localtime(&now);
	
	a = ltm -> tm_mday;
	b = 1 + ltm -> tm_mon;
	c = 1900 + ltm -> tm_year;
	d = ltm -> tm_hour;
	e = ltm -> tm_min;

	dd = to_string(a);
	MM = to_string(b);
	yyyy = to_string(c);
	HH = to_string(d);
	mm = to_string(e);
	
	datetime = dd + MM + yyyy + HH + mm;

(Yes, I understand I could just write dd = to_string(ltm -> tm_mday) and so on but bear with me.)

There'll be times when the latest file will be unavailable, and I'll have to scan one from my files instead (say, suddenly the system goes off and 060420181720.txt never arrives on my end so I have to make do with 060420181710.txt or 060420181620.txt or 060420171720.txt). I could just reduce e, d and c by preset values I'm afraid I'd have to add in too many if statements and litter the code (say, if 010120190000.txt isn't available I'd have to reduce e by 10, then detect that e is negative, then make e = e + 60 and d = d - 1, then detect that d is negative, then make d = d + 24 and c = c - 1 and so on). I've been thinking instead of getting the time since epoch, reducing it by a preset number of seconds (say, time_since_epoch = time_since_epoch - 600) and then converting the new time_since_epoch variable into the datetime string. Would it be possible to implement this just using the ctime library (or any other default one for that matter)?

[helios]

Do you just need to read the latest file?

By the way, this is unrelated to your question, but date formats other than year-month-day-hour-minute-second are terrible because they can't be sorted chronologically by common string comparison functions. Meaning if you're browsing the directory all the files will be in an unusable order.

[jmakkay]

Do you just need to read the latest file?

This would actually do the trick in some occasions, but not always. Say, if all files from the last 12 hours were unavailable I'd get more usable data from one 24 hours ago than from one from 12.

By the way, this is unrelated to your question, but date formats other than year-month-day-hour-minute-second are terrible because they can't be sorted chronologically by common string comparison functions. Meaning if you're browsing the directory all the files will be in an unusable order.

Didn't even think of that. Thanks for the advice.

[jmakkay]

I've found what I wanted. In retrospect, it was quite a silly question.

I just created a new time_t variable, which I can set it to any value I want. Then I just call localtime(&new_variable).

Solved.

[helios]

Say, if all files from the last 12 hours were unavailable I'd get more usable data from one 24 hours ago than from one from 12.

But how could you encode this logic? Okay, so if there's no data for 12 hours you read data from 24 hours ago.
* What if there's been no data for three days?
* What if 24 hours ago there was a 2-hour period where there was no data?
* What if all the files got nuked except for a single one from 12 hours ago?

Are you maybe trying to automate too much? Maybe you need to limit the scope of your problem.