Difficulties with an example 1.9 of The C Programming Language

Question

I'm am working my way through the exercises of the first chapter of The C Programming Language and while I understand most of what is said and shown, there is one example that I don't understand.

In 1.9, there is a function shown to return the length of a line while setting a char array, passed as an argument, to the contents.

int get_line(char s[], int lim)
{
    int c, i, l;

    for (i = 0, l = 0; (c = getchar()) != EOF && c != '\n'; ++i) {
        if (i < lim - 1)
            s[l++] = c;
    }
    if (c == '\n')
        if (l < lim - 1)
            s[l++] = c;
    s[l] = '\0';

    return l;
}

The thing I do not understand, is why we need this: if (c == '\n') {...}. Could this not be combined in the for-loop? Where we explicitly check that c is not equal to '\n'? I'm having trouble wrapping my head around why this needs to be an external condition.

Any light shed would be helpful! Thanks!

Answer 1

The for loop is exited if either c equals EOF or c equals '\n'. Therefore, immediately after the for loop, if you want to know which value c has, you must test.

Answer 2

is why we need this: if (c == '\n') {...}.

get_line() is structurally:

get_line() {
  initialize

  while get, A not true and B not true
    perform X

  if B
    perform X
  
  finalize

The loop quits under 2 conditions. With one of those (c == '\n'), we still want to perform X somewhere as that is part of the function goal.

---

​Could this not be combined in the for-loop?

It could be combined, yet then we have 2 locations that exit the loop.

Typical coding guidelines promote a single location to quit the loop. If we set aside that goal, then:

get_line() {
  initialize

  while get, A not true
    perform X
    if B quit the loop
  
  finalize

As below with the same number of conditions checks, yet 2 loop exit points.

int get_line(char s[], int lim) {
    int c, i, l;

    for (i = 0, l = 0; (c = getchar()) != EOF; ++i) {
        if (i < lim - 1)
            s[l++] = c;
        if (c == '\n')
            break;
    }

    s[l] = '\0';
    return l;
}

We could contort the code to get the 2 checks back on the same line and not have that pesky after the loop if (c == '\n'). Stylistically this may be harder to follow.

int get_line(char s[], int lim) {
    int c, i, l;

    for (i = 0, l = 0, c = 0; c != '\n' && (c = getchar()) != EOF; ++i) {
        if (i < lim - 1)
            s[l++] = c;
    }
    s[l] = '\0';

    return l;
}

Lastly, code could use improvements:

• No need for i and l index counters. One is enough.

• Array sizing and index best uses size_t type. Warning: size_t is some unsigned type.

• Using a leading size parameter allows for better static code analysis and self-documenting code: the lim relates to s[].

• Avoid math on input parameters to not incur overflow. We have more range control on local objects.

• Careful when lim is at an extreme or zero.

• Rather than assign after declaration, where practical, initialize. E.g. int i = 0;

---

get_line() {
  initialize

  while B not true, get, A not true
    perform X
  
  finalize

or

#include <stdio.h>
#include <stdlib.h>
      
size_t get_line(size_t size, char s[size]) {
  int ch = 0;
  size_t i = 0;

  while (ch != '\n' && (ch = getchar()) != EOF) {
    if (i + 1 < size)
      s[i++] = (char) ch;
  }

  // size might have been pathologically 0, so no room for \0
  if (i < size) {
    s[i] = '\0';
  }
  return i;
}

Answer 3

If you want to put it in the loop, you have to do something like that:

int get_line(char s[], int lim)
{
    int c, i, l;

    for (i = 0, l = 0; (c = getchar()) != EOF; ++i) {
        if ((i < lim - 1) && (c != '\n'))
            s[l++] = c;
        else if (c == '\n') {
            if (l < lim - 1)
                s[l++] = c;
             break;
         }
    }

    s[l] = '\0';

    return l;
}

So as you see, wrapping the condition inside the loop, led to more conditions checks and a break statatement.