I have a hex value FFFFFFCE that actually represents -50. When I simply use hex function to convert the above value to an integer, I get 4294967246. That means that integer overflow took place. How do I handle such cases, so that conversion will be correct?
Converting hex string to 32 bit signed integer
Collapse
X
-
Sorry. Let me try to clarify. I am trying to convert 8-character hex values to their 32-bit integer representation. The problem arises when a hex string starts with FFFF (ex. FFFFFFCE). I know that the real answer once the conversion is done is "-50" for "FFFFFFCE". You can check it by entering "-50" into any calculator that does conversions, and you will see that the hex value for it is "FFFFFFCE". But if you enter "FFFFFFCE" first and try to convert it to an integer, you will get "4294967246 ", which indicates overflow. My current solution is to subtract the maxint value (FFFFFFFF) from the given value (FFFFFFCE) and to subtract 1 from the result, which will yield the correct value of "-50". I don' t know if that is the best way and also if there are any other gotchas. I would imagine underflow will also present problems.Comment
-
Have you tried using the Data::Translate module from cpan? It looks like it does conversions like the one you are looking to do.
Regards,
JeffComment
-
Originally posted by whirlySorry. Let me try to clarify. I am trying to convert 8-character hex values to their 32-bit integer representation.
Fortunately, the rules are fairly simple. But let's start with unsigned first:
unsigned int
Dec: 0 to 4_294_967_295 (2^32-1)
Hex Representation: 0x00000000 to 0xFFFFFFFF
As you can see, straight forward enough. And the hex translation can be handled directly with the hex function and sprintf.
Signed integers are a little harder. The positive range is handled easily enough. However, the negative range is indicated by the 32nd bit, and the values are stored opposite of the positive range. This made arithmetic simpler back in the "old days" I suspect.
signed int
Dec: (-1 * 2 ^ 31) -2_147_483_648 to 2_147_483_647 (2 ^ 31 - 1)
hex positive: 0x00000000 to 0x7FFFFFFF
hex negative: 0xFFFFFFFF to 0x80000000
The easiest way to code your own translation function would be to rely on a little bit arithmetic:
[CODE=perl]
print hexstr_to_signe d32int('FFFFFFC E'), "\n"; # Outputs -50
print signed32int_to_ hexstr(-50), "\n"; # Outputs ffffffce
sub hexstr_to_signe d32int {
my ($hexstr) = @_;
die "Invalid hex string: $hexstr"
if $hexstr !~ /^[0-9A-Fa-f]{1,8}$/;
my $num = hex($hexstr);
return $num >> 31 ? $num - 2 ** 32 : $num;
}
sub signed32int_to_ hexstr {
my ($num) = @_;
die "Number outside of signed 32-bit range: $num"
if $num > 2_147_483_647 or $num < -2_147_483_648;
my $unsigned = $num < 0 ? 2 ** 32 + $num : $num;
return sprintf "%x", $unsigned;
}
[/CODE]
If you wanted capital hex strings, just change %x to %X
- MillerComment
-
Really great answer! Your thorough explanation helped me with a loosely related issue in creating a hashing function compatible with ones we use in Java and C. Perl had the binary result intact, but just needed a little adjustment to represent it as an unsigned integer value. Much appreciated.
Originally posted by millerThe problem you have is that there is no such thing as a strictly 32-bit integer in perl, so you must code the rules yourself or use a module as Jeff suggested.
...
[CODE=perl]
return $num >> 31 ? $num - 2 ** 32 : $num;
[/CODE]
- MillerComment
Comment