C使用系统调用将.txt转换为二进制文件

Question

I'm writing a program that replaces every character where position % step = 0 As command line arguments I'm giving 1. file 2. character and 3. step . I can only use system calls. Here is my main function:

int main(int argc, char **argv){

   assert(argc ==  4);

   int fdInput = open(argv[1], O_WRONLY);
   if(fdInput == -1)
      fatalError("Error opening input file.\n");

   char c[1];
   c[0] = argv[2][0];
   unsigned step = atoi(argv[3]);

   int fileSize;
   if((fileSize = lseek(fdInput,0,SEEK_END)) < 0)
      fatalError("Lseek error: Determining file size\n");

   if(lseek(fdInput,0,SEEK_SET) == -1)
      fatalError("Lseek error: Returning to the beginning\n");

   int i;
   for(i = 0; i*step < fileSize; i++)

      if(step - 1 > 0){
         if(lseek(fdInput, i*step - 1, SEEK_SET) == -1)
            fatalError("Lseek error: Within loop\n");

         if(write(fdInput, c, 1) != 1)
            fatalError("Writing error\n");

      }
      else {

         if(write(fdInput, c, 1) != 1)
            fatalError("Writing error.\n");
      }
      close(fdInput);
      return 0;
   }

Example:

input.txt: 123456789

./output input.txt x 3 would return 12x45x78x

Problem: For some reason, when I compile and execute for the first time, everything works fine ! BUT: when I execute it for the second time, it won't work. When I try to cat/less input.txt it tells me that file is binary.

• echo "123456789" > input.txt -> creates .txt file
• ./output input.txt x 3 -> 12x45x78x
• ./output input.txt x 3 -> won't work (program is finished), but:
• less input.txt -> input.txt" may be a binary file. See it anyway?

How can file be binary? It should be plain text file. What am I doing wrong here? Am I doing something wrong with open ?

Answer 1

Your first seek offset is wrong:

i*step - 1

with i == 0 this yields -1 , considering only the pure mathematical calculation and leaving implicit type casts out of the game. This is then converted to an unsigned ( off_t is signed and a negative value passed should lead to EINVAL , thus it has to be the implicit conversion) which is going to be extremely large ( UINT_MAX ). The result is an extremely large (but sparse) file.

Oh, and:

assert(argc ==  4);

assert is meant to check invariants, not for handling incorrect user input.

Answer 2

You should initialize i to 1 instead of 0 ie

for(i = 1; i*step < fileSize; i++)

Otherwise, as @Daniel Jour said, your first seek offset will be wrong (i*step - 1 == -1, considering only the pure mathematical calculation and leaving implicit type casts out of the game).

Also, you should add for-loop 's opening and closing braces for better legibility.