Question

I have written a program that works well in C that converts non-readable ASCII to their character values. I would appreciate if a C master? would show me a better way of doing it that I have currently done, mainly this section:

if (isascii(ch)) {
    switch (ch) {
        case 0:
            printControl("NUL");
            break;
        case 1:
            printControl("SOH");
            break;

        .. etc (32 in total)

        case default:
            putchar(ch);
            break;
    }
}

Is it normal to make a switch that big? Or should I be using some other method (input from an ascii table?)

Answer 1

Too many years ago when assembly languages for 8-bit micros were how I spent my time, I would have written something like

printf("%3.3s", 
       ("NULSOHSTXETXEOTENQACKBELBS HT LF VT FF CR SO SI "
        "DLEDC1DC2DC3DC4NAKSYNETBCANEM SUBESCFS GS RS US ")[3*ch]);

but not because its particularly better. And the multiply by three is annoying because 8-bit micros don’t multiply so it would have required both a shift and an add, as well as a spare register.

A much more C-like result would be to use a table with four bytes per control, with the NUL bytes included. That allows each entry to be referred to as a string constant, but saves the extra storage for 32 pointers.

const char *charname(int ch) {
    if (ch >= 0 && ch <= 0x20)
        return ("NUL\0"  "SOH\0"  "STX\0"  "ETX\0"  /* 00..03 */
                "EOT\0"  "ENQ\0"  "ACK\0"  "BEL\0"  /* 04..07 */
                "BS\0\0" "HT\0\0" "LF\0\0" "VT\0\0" /* 08..0B */
                "FF\0\0" "CR\0\0" "SO\0\0" "SI\0\0" /* 0C..0F */
                "DLE\0"  "DC1\0"  "DC2\0"  "DC3\0"  /* 10..13 */
                "DC4\0"  "NAK\0"  "SYN\0"  "ETB\0"  /* 14..17 */
                "CAN\0"  "EM\0\0" "SUB\0"  "ESC\0"  /* 18..1B */
                "FS\0\0" "GS\0\0" "RS\0\0" "US\0\0" /* 1C..1F */
                "SP\0\0") + (ch<<2);                /* 20 */
    if (ch == 0x7f)
        return "DEL";
    if (ch == EOF)
        return "EOF";
    return NULL;
}

I’ve tried to format the main table so its organization is clear. The function returns NULL for characters that name themselves, or are not 7-bit ASCII. Otherwise, it returns a pointer to a NUL-terminated ASCII string containing the conventional abbreviation of that control character, or “EOF” for the non-character EOF returned by C standard IO routines on end of file.

Note the effort taken to pad each character name slot to exactly four bytes. This is a case where building this table with a scripting language or a separate program would be a good idea. In that case, the simple answer is to build a 129-entry table (or 257-entry) containing the names of all 7-bit ASCII (or 8-bit extended in your preferred code page) characters with an extra slot for EOF.

See the sources to the functions declared in <ctype.h> for a sample of handling the extra space for EOF.