Python Hex to Signed Integer conversion

Question

I am debugging a C program which accepts input through scanf() using %ld as the format.

I need to send an input such that it is equal to 0xfffffffffffffffa.

If I convert the above hex value to decimal using Python, I get: 18446744073709551610 as shown below:

>>> a = 0xfffffffffffffffa
>>> a
18446744073709551610L

However, when I debug the program, I can see that the value was read by the program as: 0x7fffffffffffffff

So, is there a way to generate a decimal such that scanf() reads it as 0xfffffffffffffffa instead of 0x7fffffffffffffff?

Did you look at: https://stackoverflow.com/questions/6727875/hex-string-to-signed-int-in-python-3-2#6727975 — Jens, Jan 04 '19 at 12:23
`%ld` stores the result as a 64-bit *signed* integer (`signed long long`) and `0xfffffffffffffffa` overflows. — Martijn Pieters, Jan 04 '19 at 12:25
Must your C code accept a *signed* integer? Is -1 a valid input? If not, don't use `d`, use `l`: `%llu`, to switch the acceptable range to positive integers up to `0xffffffffffffffff`. — Martijn Pieters, Jan 04 '19 at 12:37
@MartijnPieters `signed long`; and the conversion specifier is `u` for unsigned, and it must be matched with `unsigned long` or `unsigned long long` argument — Antti Haapala -- Слава Україні, Jan 04 '19 at 12:43
@AnttiHaapala: yeah, already corrected. C scanf manpage reading is not my strong point. :-P — Martijn Pieters, Jan 04 '19 at 12:46

Antti Haapala -- Слава Україні · Answer 1 · 2019-01-04T12:51:33.407

There are 2 conversion specifiers for scanf for reading decimal numbers. u reads an unsigned decimal integer - into a variable of unsigned integer type, and d reads a signed decimal integer (into a variable of signed integer type). When reading signed decimal integer, the maximum for a 64-bit 2's complement number would be 2 ** 63 - 1 i.e. 9223372036854775807. When a number exceeding this is matched, scanf would still match it, but set the corresponding argument to the maximum value.

For unsigned 64-bit, the maximum would be 2 ** 64 - 1. 18446744073709551610 would work for unsigned input, but you need the signed number for which 0xfffffffffffffffa is the 2's complement representation. It is actually

>>> 0xfffffffffffffffa - 2 ** 64
-6

because the negative numbers are converted to unsigned by repeatedly adding or subtracting one larger than the maximum value until the value fits within the range - we now invert the operation and get -6.

Note that it is not correct to scanf into an object using %ld and then printf its value using %lx as you seem to do, unless the variable is a signed int and printf invocation has a cast to the argument to unsigned - because %lx requires an unsigned long int! If your compiler does not warn about it, then do add a switch such as -Wall to enable extra diagnostics.

It should be mantioned that the format specifier must exactly match the type of the destination variable; trying to read into `signed long long` with `%llu` causes undefined behaviour — M.M, Jan 04 '19 at 12:45
better although it's already UB to scan with `%ld` if the object is declared as having unsigned type, even if you don't go on to print — M.M, Jan 04 '19 at 12:57
@M.M the last addition there... blah can't bother to be `-pedantic-errors-extra-whiney`:D — Antti Haapala -- Слава Україні, Jan 04 '19 at 12:58

Python Hex to Signed Integer conversion

1 Answers1