Problem with ctime

Hello all.
I am having a problem with the ctime() function defined in the header included by #include <ctime>.

The function is supposed to return a string in the format:

Www Mmm dd hh:mm:ss yyyy

Where Www is the weekday, Mmm the month in letters, dd the day of the month, hh:mm:ss the time, and yyyy the year.

There should only be one space between each thing. I just encountered a situation where the returned string had two spaces after the month, which breaks my properlyFormatDateAndTime() function's algorithm which depends on the above format. The string returned by the ctime function just a couple of minutes ago, looks like this:

Code:

Wed Mar 2 10:59:10 2011

As you can see, there are two spaces after the month, instead of just one. I don't know why that is, but just thought I'd go ahead and post about it before the behavior went away (seeing as it doesn't do that all the time, as I just ran my program which has the properlyFormatDateAndTime() function the other day, and it worked fine, and as expected, then).

EDIT: Ok, this is even weirder. The above is what I thought was happening, but actually this is what the ctime function is returning:

Code:

' Wed Mar 2 11:12:33 2011

I was reporting the output of a C++ string before that had the content of the C-style string returned by ctime assigned to it (because I assumed the content would be the same), but then I decided to check the content of the actual C-style string, and the above is what it is returning now, and what the C++ string which gets the content of the C-style string returned by ctime contains is this:

Code:

Wed Mar 2 11:12:33 2011

As you can see, the C++ string is missing the beginning hyphen and space that the original string contained for some reason.

It's not two spaces after the month, it's the single digit day printed with a leading space.

What code are you using?

Quote:

Originally Posted by Salem

It's not two spaces after the month, it's the single digit day printed with a leading space.

What code are you using?

Ahh..ok. But why does it put a leading space before the day?

Here's my code (please ignore the debugging statements, most of which are commented out):

Code:

const char* properlyFormatDateAndTime(char* c_time) { const char* properly_formatted_c_time; string c_time_str = c_time; cout<< "c_time is: ' " << c_time <<endl; cin.get(); cout<< "c_time_str is: " << c_time_str <<endl; cin.get(); string day_of_the_week = c_time_str.substr(0, c_time_str.find(' ', 0)); //cout<< "day_of_the_week is: " << day_of_the_week <<endl; //cin.get(); c_time_str = c_time_str.substr(c_time_str.find(' ', 0) + 1); //cout<< "c_time_str is: " << c_time_str <<endl; //cin.get(); string month = c_time_str.substr(0, c_time_str.find(' ', 0)); //cout<< "month is: " << month <<endl; //cin.get(); c_time_str = c_time_str.substr(c_time_str.find(' ', 0) + 1); //cout<< "c_time_str is: " << c_time_str <<endl; //cin.get(); string date = c_time_str.substr(0, c_time_str.find(' ', 0)); //cout<< "date is: " << date <<endl; //cin.get(); c_time_str = c_time_str.substr(c_time_str.find(' ', 0) + 1); //cout<< "c_time_str is: " << c_time_str <<endl; //cin.get(); string current_time = c_time_str.substr(0, c_time_str.find(' ', 0) + 1); //cout<< "current_time is: " << current_time <<endl; //cin.get(); string hour = current_time.substr(0, current_time.find(':', 0)); if (hour.at(0) == '0') { hour.erase(0, 1); } stringstream hour_ss; hour_ss<< hour; int hour_int; hour_ss>> hour_int; //cout<< "hour_int is: " << hour_int <<endl; //cin.get(); string ampm = "AM"; if (hour_int > 12) { ampm = "PM"; if (hour_int == 13) hour_int = 1; else if (hour_int == 14) hour_int = 2; else if (hour_int == 15) hour_int = 3; else if (hour_int == 16) hour_int = 4; else if (hour_int == 17) hour_int = 5; else if (hour_int == 18) hour_int = 6; else if (hour_int == 19) hour_int = 7; else if (hour_int == 20) hour_int = 8; else if (hour_int == 21) hour_int = 9; else if (hour_int == 22) hour_int = 10; else if (hour_int == 23) hour_int = 11; else if (hour_int == 24) { hour_int = 12; ampm = "AM"; } } //cout<< "hour_int is: " << hour_int <<endl; //cin.get(); hour_ss.str(""); hour_ss.clear(); hour_ss<< hour_int; hour = hour_ss.str(); //cout<< "hour is: " << hour <<endl; //cin.get(); current_time = current_time.substr(current_time.find(':', 0) + 1); string minutes = current_time.substr(0, current_time.find(':', 0)); //cout<< "minutes is: " << minutes <<endl; //cin.get(); current_time = current_time.substr(current_time.find(':', 0) + 1); string seconds = current_time.substr(0, current_time.find(' ', 0)); //cout<< "seconds is: " << seconds <<endl; //cin.get(); string complete_time = hour + ":" + minutes + ":" + seconds + " " + ampm; //cout<< "complete_time is: " << complete_time <<endl; //cin.get(); c_time_str = c_time_str.substr(c_time_str.find(' ', 0) + 1); //cout<< "c_time_str is: " << c_time_str <<endl; //cin.get(); string year = c_time_str.substr(0, c_time_str.find('\n', 0)); /* if (year.at(year.size() - 1) == '\n') { cout<< "year is followed by a new line." <<endl; cin.get(); } */ //cout<< "year is: " << year <<endl; //cin.get(); string properly_formatted_c_time_str = complete_time + " - " + day_of_the_week + " " + month + " " + date + ", " + year; //cout<< "properly_formatted_c_time_str is: " << properly_formatted_c_time_str <<endl; //cin.get(); properly_formatted_c_time = properly_formatted_c_time_str.c_str(); //cout<< "properly_formatted_c_time is: " << properly_formatted_c_time <<endl; //cin.get(); return properly_formatted_c_time; }

Why not use strftime, and get the format you exactly want?
man page strftime section 3

I think it's very consistent with what happens when you print a single digit in a field two characters wide. sprintf wouldn't pad zeros on a number unless you told it to a la "%02d". I can't say ctime() or asctime() are implemented in such a way, I never looked. Anyway, using strftime() instead should solve your format issues. See strftime - C++ Reference.

Quote:

Originally Posted by Salem

Why not use strftime, and get the format you exactly want?
man page strftime section 3

Quote:

Originally Posted by whiteflags

I think it's very consistent with what happens when you print a single digit in a field two characters wide. sprintf wouldn't pad zeros on a number unless you told it to a la "%02d". I can't say ctime() or asctime() are implemented in such a way, I never looked. Anyway, using strftime() instead should solve your format issues. See strftime - C++ Reference.

Thanks guys.
My properlyFormatDateAndTime() function definition now looks like this:

Code:

const char* properlyFormatDateAndTime(const struct tm* timeptr) { char properly_formatted_date_and_time[80]; strftime(properly_formatted_date_and_time, 80, "%I:%M:%S - %A %B %d, %Y", timeptr); cout<< "Properly formatted date and time is: " << properly_formatted_date_and_time <<endl; cin.get(); return properly_formatted_date_and_time; }

Which formats it in roughly the way I want it to be, only I can't figure out how to get rid of the 0 in front of the hour if its < 10 and the 0 in front of day of the month if its < 10, when using strftime. Any advice on that?

Thanks.

EDIT: Nevermind...just realized what I need to do.

EDIT again: Hmm...maybe not. I was able to remove the one for the hour, but I can't figure out how to remove the one for the date (or I mean I probably could, but I don't feel like working on this issue any longer right now :p).

Here is my new code:

Code:

const char* properlyFormatDateAndTime(const struct tm* timeptr) { char properly_formatted_date_and_time_non_const[80]; strftime(properly_formatted_date_and_time_non_const, 80, "%I:%M:%S - %A %B %d, %Y", timeptr); string date_and_time_str = properly_formatted_date_and_time_non_const; string hour = date_and_time_str.substr(0, date_and_time_str.find(':', 0)); //cout<< "hour is: " << hour <<endl; //cin.get(); stringstream hour_ss; hour_ss<< hour; int hour_int; hour_ss>> hour_int; //cout<< "hour_int is: " << hour_int <<endl; //cin.get(); if (hour_int < 10) { string new_hour = hour_ss.str(); new_hour.erase(0, 1); //cout<< "new_hour.size() is: " << new_hour.size() <<endl; //cout<< "new_hour is: " << new_hour <<endl; date_and_time_str.replace(date_and_time_str.find(hour, 0), hour.size(), new_hour); } //cout<< "Date and time is: " << date_and_time_str <<endl; //cin.get(); const char* properly_formatted_date_and_time = date_and_time_str.c_str(); return properly_formatted_date_and_time; }

How about using Boost's Date/Time library?

Code:

const char* properlyFormatDateAndTime(const struct tm* timeptr) { char properly_formatted_date_and_time_non_const[80]; strftime(properly_formatted_date_and_time_non_const, 80, "%I:%M:%S - %A %B %d, %Y", timeptr); string date_and_time_str = properly_formatted_date_and_time_non_const; string hour = date_and_time_str.substr(0, date_and_time_str.find(':', 0)); //cout<< "hour is: " << hour <<endl; //cin.get(); stringstream hour_ss; hour_ss<< hour; int hour_int; hour_ss>> hour_int; //cout<< "hour_int is: " << hour_int <<endl; //cin.get(); if (hour_int < 10) { string new_hour = hour_ss.str(); new_hour.erase(0, 1); //cout<< "new_hour.size() is: " << new_hour.size() <<endl; //cout<< "new_hour is: " << new_hour <<endl; date_and_time_str.replace(date_and_time_str.find(hour, 0), hour.size(), new_hour); } //cout<< "Date and time is: " << date_and_time_str <<endl; //cin.get(); const char* properly_formatted_date_and_time = date_and_time_str.c_str(); return properly_formatted_date_and_time; }

You are returning a pointer to an internal area of memory managed by a local string object which is destructed after the function exits.

Quote:

Originally Posted by Elysia

How about using Boost's Date/Time library?

I'm trying to stay away from 3rd party libraries as much as possible, for right now.
I want to get the standard C++ library down first. :p

Quote:

Originally Posted by hk_mp5kpdw

Code:

const char* properlyFormatDateAndTime(const struct tm* timeptr) { char properly_formatted_date_and_time_non_const[80]; strftime(properly_formatted_date_and_time_non_const, 80, "%I:%M:%S - %A %B %d, %Y", timeptr); string date_and_time_str = properly_formatted_date_and_time_non_const; string hour = date_and_time_str.substr(0, date_and_time_str.find(':', 0)); //cout<< "hour is: " << hour <<endl; //cin.get(); stringstream hour_ss; hour_ss<< hour; int hour_int; hour_ss>> hour_int; //cout<< "hour_int is: " << hour_int <<endl; //cin.get(); if (hour_int < 10) { string new_hour = hour_ss.str(); new_hour.erase(0, 1); //cout<< "new_hour.size() is: " << new_hour.size() <<endl; //cout<< "new_hour is: " << new_hour <<endl; date_and_time_str.replace(date_and_time_str.find(hour, 0), hour.size(), new_hour); } //cout<< "Date and time is: " << date_and_time_str <<endl; //cin.get(); const char* properly_formatted_date_and_time = date_and_time_str.c_str(); return properly_formatted_date_and_time; }

You are returning a pointer to an internal area of memory managed by a local string object which is destructed after the function exits.

Maybe so, but it still works.

Quote:

Originally Posted by Programmer_P

I'm trying to stay away from 3rd party libraries as much as possible, for right now.
I want to get the standard C++ library down first. :p

Is learning the C standard library on your list too, especially when there is a C++ portable alternative available?

Quote:

Maybe so, but it still works.

Geez. Do you not realize that you are running on undefined behavior?
Just because it works, doesn't that it's right.

Quote:

Originally Posted by Elysia

Is learning the C standard library on your list too, especially when there is a C++ portable alternative available?

Not really, as I'm not really interested in C, because I understand C++ is a step up from C which basically includes all of C's features anyway (though I'm sure someone will correct me on the "all" part I said).

Quote:

Geez. Do you not realize that you are running on undefined behavior?

Haha, I knew that would get some responses.
How so? I mean, if it works, how is it "undefined behavior"?
Computers can only do what you tell it to do, so if my computer is outputting the time correctly from my program when I use the above function, then its behavior must obviously be defined somewhere.

Allow me to demonstrate why it's undefined behavior...

Quote:

Originally Posted by Elysia

Allow me to demonstrate why it's undefined behavior...

Please do. :D

Code:

#include <string> #include <iostream> const char* foo() { std::string s = "Hello World!\n"; std::cout << "foo: " << s; return s.c_str(); } int main() { const char* s = foo(); std::cout << "main: " << s; }

Output:
foo: Hello World!
main: Press any key to continue . . .

Not what you expected?

Quote:

Originally Posted by Programmer_P

Not really, as I'm not really interested in C, because I understand C++ is a step up from C which basically includes all of C's features anyway (though I'm sure someone will correct me on the "all" part I said).

Right. So why bother with C functions where there is a better C++ alternative available?

If you're only really interested in C++, then return a std::string rather than messing with char* pointers of various persuations.

Only use char arrays for the very minimum of using legacy functions. As soon as you have your char array result, copy it to a std::string and use that in the rest of the code.

Quote:

Originally Posted by Elysia

Right. So why bother with C functions where there is a better C++ alternative available?

But there isn't a C++ language alternative: there is a Boost alternative written in C++. There is a difference. C++'s solution for time is C's solution for time.

That's really a semantics argument. That was not my point.

Facts are not arguments.

Whatever. Still not my point.
Do I have to reword my reply?

Don't even ask. You do this every time someone has anything to say about random things you recommend, no matter how well intentioned you are or they are.

Your vagueness confounds me. Sorry, I have no idea what you are trying to say.

if you run a program once and it works in every path. it will always work until you change it.
therefore, behavior is only undefined until you test it. once you test it and it works it will do the same thing every time, until you change it.
however, if there is an external variable behavior can be undefined because the variable will affect the program, BUT then the behavior for the program is defined, but it relies on the variable.

plz argue this. i want to see what points you will use

AWESOME!

I am so freaking glad that you've told me.

I'm serious; I will never have to track down undefined behavior ever again. I'm so happy!

Soma ^_^

well, you still have to debug.

when you get "undefined behavior" you have something that looks like it would work but varies from system to system. this can still can cause problems. most of the point of my argument was that "undefined behavior" is a misnomer. if it does it correctly once it will work correctly until a variable is changed

O_o

You have no idea what you are talking about. I know from your posts that you are a newbie. There is nothing wrong with that, but please stop offering such statements as if you know something. It might be the case that another newbie comes along and believes you.

The phrases "undefined behavior" and "the behavior is undefined" appear in the C++ standard document to describe situations where the standard committee chose not to require a particular behavior. In practice, this means that a platform is allowed to do virtually anything in those situations. It may be that, for any given case of undefined behavior, the platform does define the behavior as an extension to the standard, but that need not be the case.

In the real world employing undefined behavior may result in seemingly random crashes. This is simply the nature of the beast.

Feeding invalid values, outside the range or domain, to the standard library functions is undefined behavior, and depending on the state of the system, outside the program environment, your program may or may not crash even if the program environment is exactly the same.

Soma