从C中的文件逐字符读取

Question

I wrote the following program :

int main(){
    char str[500],c;
    FILE *f1=fopen("input.txt","r");
    FILE *f2=fopen("output.txt","w");


    while(c=fgetc(f1)!=EOF)
    fputc(toupper(c),f2);


    fclose(f1);
}

I was not getting the desired result though. I rewrote the code using a do while loop.

int main(){
  char str[500];
  FILE *f1=fopen("input.txt","r");
  FILE *f2=fopen("output.txt","w");
  char c;

  do
  {
    fputc(toupper(c),f2);
    c=fgetc(f1);
  }while(c!=EOF);
}

I figured out that the reason the first code fails is because in the while loop while(c=fgetc(f1)!=EOF) , we cannot guarantee that the left part of != is evaluated first and hence the results are not proper. Is this correct the explanation?

Answer 1

Yes you are correct; in your first code your while loop is written wrongly:

while(c=fgetc(f1)!=EOF)

Should be:

while((c=fgetc(f1))!=EOF)
//    ^           ^   added parenthesis

Because the precedence of operator != is greater than = operator in conditional expression c=fgetc(f1)!=EOF , the first returned the result of comparing the value from fgetc() with EOF (either 0 or 1) and assigned that to c . (That means simply c=fgetc(f1)!=EOF expression is equivalent to c=(fgetc(f1)!=EOF) and this is not what you need.)

You need () to overwrite precedence as I suggested.

But you have second thing to improve that is c variable must be an int ( not char ) in order to hold an EOF-value.

A very good Read: Definition of EOF and how to use it effectively

Answer 2

I will add a little why c should be int , not char . Suppose you write

#include <stdio.h>
#include <ctype.h>
#include <locale.h>

int main(){
    setlocale(LC_ALL,"");
    FILE *f1=fopen("input.txt","r");
    FILE *f2=fopen("output.txt","w");
    char c;

    while(EOF != (c=fgetc(f1))){
        if(isalpha(c)) c = toupper(c);
        fputc(c,f2);
    }
return 0;
}

And your input.txt is

some text
Некоторый текст
Ъ - on this letter program will stop

in KOI8-R symbol Ъ have code 255 == -1 (when you use char ). That's why in case of using char instead of int will give your output.txt with only that text:

SOME TEXT
НЕКОТОРЫЙ ТЕКСТ

As for non-working code with parentheses: c=fgetc(f1)!=EOF could be denote by compiler as c = (fgetc(f1)!=EOF) , that's why it's better always to add parentheses.

I recommend you to use flags -Wall -Werror when compiling your applications. In that case "forgetting" of parentheses would give you an error:

11.c: In function 'main':
11.c:9:2: error: suggest parentheses around assignment used as truth value [-Werror=parentheses]
cc1: all warnings being treated as errors