strtol() question

[movl0x1]

hello

#include <stdio.h>
#include <stdlib.h>
#include <errno.h>

int main(void)
{
     char *str = "ffffffff";
     int c;

     c = strtol(str, 0, 16);

     if (errno != 0)
         printf("Strtol overflowed!!!\n");

     printf("&#37;d\n", c);

     return 0;
}

How do I make strtol() read ffffffff as -1 and not UNSIGNED_LONGMAX??

I'm basically making a conversion program that converts between
different bases, e.g decimal to hex, hex to decimal, decimal to binary, etc.

The user can enter ---> cvt -h ffffffff -d and it prints -1
If the user enters --> cvt -uh ffffffff -d (u stands for unsigned) it prints whatever
UNSIGNED LONG MAX is.

The above overflows because it's reading ffffffff as UNSIGNED_LONGMAX.

Similarly with:

         char *srt = "11111111";
   
         c = strtol(str, 0, 2);

strtol reads 11111111 as 255. Is there a way to make it read 11111111 as -1?
I know there are various ways without using strtol. Im just wondering if strtol
has some built in mechanism for reading 11111111 or ffffffff as negative instead
of positive.

thanks

[brewbuck]

Sounds like your strtol() function is broken. Even if it did overflow, it should be returning LONG_MAX, not UNSIGNED_LONGMAX.

EDIT: On second thought, strtol() isn't broken (although it should be returning LONG_MAX not UNSIGNED_LONGMAX). You're trying to use it in a way it can't be used. It isn't meant to convert bit patterns into integers, it is a number converter. Just because FFFFFFFF happens to represent -1 doesn't mean strtol() will do that. Instead, it sees that FFFFFFFF is larger than 2^31-1, the maximum range of a signed integer, and returns an overflow. This is as it should be.

Anyway, what makes you think that FFFFFFFF is equal to -1? That's an assumption of 2's complement, an assumption you shouldn't make.

[itsme86]

If you want to use an unsigned conversion, use strtoul() instead.

[movl0x1]

No, strtol is returning LONG_MAX, but interpreting ffffffff as UNSIGNED LONG MAX
causing it to overflow.

thanks

[brewbuck]

No, strtol is returning LONG_MAX, but interpreting ffffffff as UNSIGNED LONG MAX
causing it to overflow.

thanks

It's not interpreting it as anything other than the number it is, which is 2^32-1. This is larger than the largest signed integer value, so you get an overflow.

[Dave_Sinkula]

Im just wondering if strtol
has some built in mechanism for reading 11111111 or ffffffff as negative instead
of positive.

A minus sign?

     char *str = "-7fffffff";

[brewbuck]

If what you want to do is convert 2's complement values to decimal, strtol() will not do that. Just to be crystal clear. The only way strtol() will return a negative value is if the number has a minus sign in front of it.

[movl0x1]

thanks guys


If I put a minus in front of 11111111 (-11111111) strtol spits out -255 though, not -1.

But if the user enters 11111111 and thinks of it as unsigned, he wants the
program (when converting to decimal) to spit out 255, and if he enters
11111111 and thinks of it as signed, he wants the program to spit out -1
when converting to decimal.

What I'm saying is that 11111111 is converted to 255 with strtol() when it
reads that character string and converts it to base 2.

I was wondering if strtol() can convert it to -1 instead.
Probably not. Guess I'm gonna have to create some type of wrapper.

[Dave_Sinkula]

If I put a minus in front of 11111111 (-11111111) strtol spits out -255 though, not -1.

Sure, that's what it is.

But if the user enters 11111111 and thinks of it as unsigned, he wants the
program (when converting to decimal) to spit out 255, and if he enters
11111111 and thinks of it as signed, he wants the program to spit out -1
when converting to decimal.

What I'm saying is that 11111111 is converted to 255 with strtol() when it
reads that character string and converts it to base 2.

I was wondering if strtol() can convert it to -1 instead.
Probably not. Guess I'm gonna have to create some type of wrapper.

I'd guess so. It sounds like you first want to create an integral image given the text representation in a certain base, and then display (interpret) it according to a specified signedness.

For that I'd say use unsigned conversion to an integral type first, and then a cast for the interpretation. And remember that implementations vary, so this will not be portable as I have described.

[movl0x1]

Oh yeah. Right, thanks. I forgot that 255 in an int (well, on my machine) would be

00000000 00000000 00000000 11111111. I think that's how strtol would return it in
the int it assigns it to.

So read that string "11111111" in with strtoul(), and it put's it in the unsigned int as

00000000 00000000 00000000 11111111 (255 unsigned)

typecast it to

11111111 11111111 11111111 11111111 (-1 unsigned)


I hope I got it! Thanks to all for your help.


I did a little example and it did what I wanted

#include <stdio.h>
#include <stdlib.h>

int main(void)
{
           char *str = "ffffffff";
           unsigned = c;
           
           /* user wants 0xffffffff to be interpreted as signed (-1) */
           
           /* get string and convert it to all 1 bits inside unsigned int c */
           c = strtoul(str, 0, 16);

           /* use printf  (I believe printf calls itoa() ) to build ascii string out of the int */
           /* printf() interprets c (all 1 bits) as unsigned, therefore printing -1.
           printf("&#37;d\n", c);

           return 0;
}

[vart]

note that the solution is not portable - entered value should be the same size as int, if int is 64bit - your trick will not work...

Why user does not want to enter -1? It is really shorter.

[movl0x1]

this is gonna be a program that converts numbers.

If you enter ---> cvt -h ffffffff -d

cvt = (convert - name of my prog) -h string is interpreted as signed hexadecimal
-d string is converted from ffffffff to its decimal representation.

Now, if you enter ----> cvt -uh ffffffff -d (-uh meaning unsigned hex)

It spits out the whatever ULONG_MAX is as a decimal number.

If I enter ---> cvt -b 11111111 -d it spits out -1

cvt -d 10 -h spits out ---> 0a

10 (decimal string) converted to hex is 0a

It's for GNU/Linux
Just a little conversion prog I can use to convert numbers back an forth from hex to binary, binary to hex, etc.

[vart]

11111111 in binary is not a LONG_MAX...

And LONG_MAX is different on different compilers...

[movl0x1]

If I have a 32-bit register or variable this is LONG and ULONG_MAX

11111111 11111111 11111111 11111111

ULONG_MAX under my GNU/LINUX 32 bit. Yes, I know it's different on
different architectures.

thanks