Thread: Array of getchar / count

Code:

int to_array_index(int c) {
    if (c >= 97 && c <= 122) /* ascii a - b */
    {
        return c - 97;
    }
    else if (c >= 65 && c <= 90)  /* ascii A - B */
    {
        return c - 65
    }
    else {
        return -1; /* error, not letter */
    }
}

Yeuch...

Code:

int to_array_index(int c) {
    if (c >= 'a' && c <= 'z') /* ascii a - b */
    {
        return c - 'a';
    }
    else if (c >= 'A' && c <= 'Z')  /* ascii A - B */
    {
        return c - 'A'
    }
    else {
        return -1; /* error, not letter */
    }
}

is much easier to read, don't you think?

Yes, I know that 97 is 'a', but why not let the compiler work it out for you. The one extra character you have to type in 5 out of the 6 places isn't really going to matter in the whole scheme of things, is it?

--
Mats

Originally Posted by matsp

Code:

int to_array_index(int c) {
    if (c >= 97 && c <= 122) /* ascii a - b */
    {
        return c - 97;
    }
    else if (c >= 65 && c <= 90)  /* ascii A - B */
    {
        return c - 65
    }
    else {
        return -1; /* error, not letter */
    }
}

Yeuch...

Code:

int to_array_index(int c) {
    if (c >= 'a' && c <= 'z') /* ascii a - b */
    {
        return c - 'a';
    }
    else if (c >= 'A' && c <= 'Z')  /* ascii A - B */
    {
        return c - 'A'
    }
    else {
        return -1; /* error, not letter */
    }
}

is much easier to read, don't you think?

Yes, I know that 97 is 'a', but why not let the compiler work it out for you. The one extra character you have to type in 5 out of the 6 places isn't really going to matter in the whole scheme of things, is it?

--
Mats

You are actually right, it's more better now.

Nonetheless note that matsp's modification still does not make the code absolutely portable unlike cas' suggestions, though it makes it slightly more portable and definitely more readable. On the other hand, such absolute portability can generally be ignored nowadays since Unicode is a superset of ASCII.

Originally Posted by laserlight

Nonetheless note that matsp's modification still does not make the code absolutely portable unlike cas' suggestions, though it makes it slightly more portable and definitely more readable. On the other hand, such absolute portability can generally be ignored nowadays since Unicode is a superset of ASCII.

To be clear, my solution was not an attempt to make the code more portable, but to make it more READABLE. I should perhaps have mentioned that in my post - for portability, we'd have to accept that the characters MAY not come in any particular order, and that there may be gaps between two letters in the alphabet.

More so if we start looking at international languages, e.g. Scandinavian languages which have ä, å, ö - those are outside the range of 32-126 that ASCII printables are within. But if you wanted to count the number of occurrences of the letters in a Swedish text, they would be necessary. In that case, something like cas' code would be much better suited.

--
Mats

Portability was not mentioned in the OP, so I ommitted it. If Marth_01 has this problem, then the last thing they would be thinking about is portability or worrying whether it works w/ Unicode & ASCII.
Also how do cas's suggestions differenciate between ASCII and EBDIC or unicode? If they don't then you are being hypocritical.
How about just support Unicode? I was meant to replace ASCII as the standard character set so when (&if) it replaces ASCII no one will use the other charactersets & will end up w/ absolute portability.