C/C++ Users Group Library 1996 July/C-C++ Users Group Library July 1996.iso/vol_400/422_02/dosutil/tr.c

C/C++ Source or Header | 1994-03-20 | 4KB | 226 lines

/* tr - like unix tr

** THIS PROGRAM MUST BE COMPILED WITH THE EXTENDED PREPROCESSOR
** IE: cc TR -p

 by jim mckim

 this program may be used in several ways.  By example:

     tr A-Z a-z <input >output
    will convert upper case to lower case

    tr -c !-~ ? <input >output
    will convert unprintable characters to question marks

    tr -cs !-~ \40 <input >output
    will convert unprintable characters to spaces and squeeze out extra spaces

    tr -d \177-\377 <input
    will remove all characters whose eighth bit is set

 comments about this can reach me via BIX (jmckim)
 or telephone [phone])
 or USPS
 (Jim McKim
 30224 Wolf Rd.
 Bay Village, OH 44140)

 (i gladly accept bug reports; bug fixes will make me ecstatic)

Compile command: cc tr -fop
*/

#include <stdio.h>
#include <file.h>

#define TTSIZE 257
#define isodigit(c) ((c)>='0' && (c)<='7')

TellHow()
{
fputs("usage: tr [-flags] string1 string2 <input >output\n", stderr);
fputs("where flags are:\n",stderr);
fputs("c  compliment string1 (e.g. all NOT in string1)\n",stderr);
fputs("d  delete if in string1\n",stderr);
fputs("s  squeeze out duplicates if in string2\n",stderr);
fputs("r  raw (no cr/lf conversion) (MSDOS only)\n",stderr);
exit(-1);
}

Interpret(input,output)
char *input, *output;
{
    int startc, endc;
    int i;
    int range_flag;  /* true when '-' has been seen */
    int got_next;  /* true when char after '-' has been seen */

    *output = range_flag = got_next = 0;
    while (*input)
    {
        if (*input == '\\')
        {
            ++input;
            if (isodigit(*input))
            {
                /* got an octal constant */
                i = *input - '0';
                if (isodigit(*(input+1)))
                {
                    i = i*8 + *++input - '0';
                    if (isodigit(*(input+1)))
                        i = i*8 + *++input - '0';
                }
                *output++ = i;
                if (range_flag)
                    got_next = 1;
            }
            else
            {
                /* got a \c type character */
                *output++ = *input;
                if (range_flag)
                    got_next = 1;
            }
        }
        else if (*input == '-')
        {
            /* got a range */
            range_flag = 1;
        }
        else
        {
            /* got a normal character */
            *output++ = *input;
            if (range_flag)
                got_next = 1;
        }

        ++input;

        if (range_flag && got_next)
        {
            startc = *(output-2);
            endc = *(output-1);
            output -= 2;
            for (i=startc; i<=endc; ++i)
                *output++ = i;
            range_flag = got_next = 0;
        }
    } /* end while */
    *output = '\0';
}

main(argc,argv)
int argc;
char *argv[];
{
    char incode[TTSIZE], outcode[TTSIZE], trtbl[TTSIZE];
    char inset[TTSIZE];
    char inch, outch, old_outch;
    int i, j, iarg;
    char c_flag, d_flag, s_flag;
    char *iap;

    iarg = 1;
    c_flag = d_flag = s_flag = 0;

    /* get the switches */
    while ((iarg <= argc) && (*argv[iarg] == '-'))
    {
        iap = argv[iarg++] + 1;  /* point past '-' */
        while (*iap)
            switch (*iap++)
            {
            case 'c':    /* Compliment string */
                c_flag = 1;
                break;
            case 'd':    /* Delete */
                d_flag = 1;
                break;
            case 's':    /* Squeeze out duplicates */
                s_flag = 1;
                break;
            case 'r':    /* Raw I/O mode */
                *stdin |= F_BINARY;
                *stdout |= F_BINARY;
                break;
            default:
                TellHow();
            }
    }

    if (iarg+1 >= argc)
    {
        TellHow();
    }

    for (i=0; i<TTSIZE; ++i)
    {
        trtbl[i] = i;
        inset[i] = '\0';
        incode[i] = '\0';
        outcode[i] = '\0';
    }

    Interpret(argv[iarg++],incode);
    Interpret(argv[iarg],outcode);

    i = 0;
    j = 0;
    while (incode[i])
    {
        inset[incode[i]] = 1;
        if (c_flag == 0)
        {
            trtbl[incode[i]] = outcode[j];
            /* if next is null, dup last */
            if (outcode[j+1])
                ++j;
        }
        ++i;
    }

    if (c_flag == 1)
    {
        j = 0;
        for (i=0; i<TTSIZE; ++i)
        {
            if (inset[i] == 0)
            {
                inset[i] = 1;
                trtbl[i] = outcode[j];
                if (outcode[j+1])
                    ++j;
            }
            else
            {
                inset[i] = 0;
                trtbl[i] = i;
            }
        }
    }

/* To speed up processsing, switch stdio streams to buffered */
    stdin = setbuf(stdin, 1024);
    stdout = setbuf(stdout, 1024);

    outch = '\0';
    while ((inch=getc(stdin)) != EOF)
    {
        old_outch = outch;
        outch = trtbl[inch];
        if (d_flag)
            if (inset[inch])
                continue;
        if (s_flag)
            if (inset[inch])
                if (old_outch == outch)
                    continue;
        putc(outch, stdout);
    }

    fflush(stdout);        /* Insure output buffer is written */
}