Sum all numbers of a file, separated by "\n"

sum.cpp

#include <fcntl.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <sys/mman.h>
#include <sys/stat.h>

off_t total_size = 0;

char* mmap_file(const char* name) {
  int fd = open(name, O_RDONLY);
  struct stat st;
  fstat(fd, &st);
  total_size = st.st_size;
  return (char*)mmap(0, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
}

// https://github.com/KholdStare/qnd-integer-parsing-experiments/blob/master/parsing.hh#L93
inline uint64_t parse_8_chars(const char* string) noexcept {
  uint64_t chunk = 0;
  memcpy(&chunk, string, sizeof(chunk));

  // 1-byte mask trick (works on 4 pairs of single digits)
  uint64_t lower_digits = (chunk & 0x0f000f000f000f00) >> 8;
  uint64_t upper_digits = (chunk & 0x000f000f000f000f) * 10;
  chunk = lower_digits + upper_digits;

  // 2-byte mask trick (works on 2 pairs of two digits)
  lower_digits = (chunk & 0x00ff000000ff0000) >> 16;
  upper_digits = (chunk & 0x000000ff000000ff) * 100;
  chunk = lower_digits + upper_digits;

  // 4-byte mask trick (works on pair of four digits)
  lower_digits = (chunk & 0x0000ffff00000000) >> 32;
  upper_digits = (chunk & 0x000000000000ffff) * 10000;
  chunk = lower_digits + upper_digits;

  return chunk;
}

int main() {
  char* data = mmap_file("file");

  uint64_t sum = 0;
  uint64_t i = 0;

  while (data[i]) {
    uint64_t value = 0;
    int g = 0;
    // https://graphics.stanford.edu/~seander/bithacks.html#ZeroInWord
    uint64_t first8 = *(uint64_t*)(data + i) ^ 0x0A0A0A0A0A0A0A0AULL;
    if (!((first8 - 0x0101010101010101ULL) & (~first8) & 0x8080808080808080ULL)) {
      value = parse_8_chars(data + i);
      i += 8;
      g = 8;
    }
    while (data[i] != '\n') {
      value = value * 10 + data[i] - '0';
      i++;
    }
    i++;
    sum += value;
  }

  printf("%lld\n", sum);
}