В результатах некоторых тестов выскакивают сообщения о том, что нужно было вернуть True. Функция должна возвращать строку, а не булевы значения. Также встречаются тесты, где возникает ошибка "p(i) should be in increasing order", хотя это невозможно в моём алгоритме.

May I suggest that the "data science" tag be removed from this kata? Some basic forumals should not qualify as such.

The correct way to account for floating point inaccuracies in assertions is not to round the result to the nearest integer but to assert for approximate equality with a tolerance.

There are problems with rounding to an integer value.
Functions such as: round, int, ceil, floor and decimal module. They do not give results.

.

.

.

.

int cur, idx, amount = 0;

idx is not initialized, meaning its value when the function is entered is arbitrary, and will sometimes cause out-of-bounds accesses, depending on what's on the stack at this point in time

your allocation of 5 bytes for cStr is also rather optimistic, what if your function is called with "0123456.789.10.11 ? (i dont know if such inputs are tested since the kata is underspecified, but in C you should always be careful)

Language?

.

.

.

.

.