C Handling Large Inputs

Question

I have seen many questions for this in C++ section. And the most of them suggested using uint64_t, unsigned long long and i have tried

• long long
• unsigned long long
• uint64_t

I am giving input of 10^18 as its max input. But the number being processed is returned as some garbage values.

I am giving it input as 111111111111111110, And at the line when I divide it by 10^(step-1)and multiply it by 10^(step-1) The value seems to be oscillating at

Right Now In 111111111111111103
Right Now In 111111111111111104 //But it should be 111111111111111100 and then -- to decreas one more

This approach works well with lesser values. But not working with greater inputs

live demo

#include<stdio.h>
#include<stdbool.h>
#include<stdlib.h>
#include<math.h>
#include<stdint.h>
#include<inttypes.h>
int step = 0;

bool isTidy(uint64_t n){
    step++;
    if(n < 10) return true;
    int lastNumber = n % 10;
    int secondLastNumber = ((n - lastNumber)/10) % 10; 
    return (lastNumber >= secondLastNumber) && isTidy((n - lastNumber)/10);
}

int main(){
    char* inputString;
    uint64_t testNumber;
    int numberOfTestcases;
    int iterator = 1;

    scanf("%d", &numberOfTestcases);
    while(numberOfTestcases --> 0){
        scanf("%"SCNu64,&testNumber);
        do{
            step = 0;
            if(isTidy(testNumber)){
                printf("Case #%d: %"PRIu64"\n",iterator++, testNumber);
                break;
            }
            printf("Right Now In %"PRIu64"\n",testNumber);
            testNumber /= pow(10,step - 1);
            testNumber *= pow(10,step - 1);

            printf("Right Now In %"PRIu64"\n",testNumber);
        }while(testNumber --> 0);
    }
    return 0;
}

edits: Tried to clear up the question. And fixed the scan input problem

Answer 1

You have the wrong conversion specifier in the second scanf(). You should have:

scanf("%"SCNu64,&testNumber);

Also, there is a stray comma in one of the printf() statements:

printf("Right Now In %"PRIu64"\n",testNumber);

It is good to enable compiler warnings to help catch such simple errors. I always compile with at least:

gcc -Wall -Wextra -Wpedantic

The posted code will compile without warnings if no warnings are enabled.

The additional problem with large inputs can be traced to the pow() function, which returns a double, not a uint64_t. One solution is to replace the calls to pow() with calls to another function that returns uint64_t. Here is an example, though this is not the most efficient implementation by a long stretch:

uint64_t mypow(uint64_t base, unsigned exp)
{
    uint64_t res = 1;

    while (exp--) {
        res *= base;
    }
    return res;
}

Answer 2

I don't understand why you are using PRIu64.

I've finished this problem from Google CodeJam and a simple scanf("%lld"), using long long, was sufficient.

---

Edits

1. Since my answer apparently wasn't good enough, here are some clarifications, you are getting garbage value while printing because of line 32:

   printf("Right Now In %"PRIu64,"\n",testNumber);
   

   there is a comma after PRIu64, remove that and all values will be printed correctly. Provided you don't get RTE for other reasons (such as an infinity loop).

2. Regarding the problem of the code (from Google CodeJam), it is not working with large numbers (and a infinity loop in your example) because pow() is a double and a lot of precision is lost with large numbers, if you use your own simple pow() function, you will get an AC.