coreutils/tr.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * Mini tr implementation for busybox
   4  *
   5  ** Copyright (c) 1987,1997, Prentice Hall   All rights reserved.
   6  *
   7  * The name of Prentice Hall may not be used to endorse or promote
   8  * products derived from this software without specific prior
   9  * written permission.
  10  *
  11  * Copyright (c) Michiel Huisjes
  12  *
  13  * This version of tr is adapted from Minix tr and was modified
  14  * by Erik Andersen <andersen@codepoet.org> to be used in busybox.
  15  *
  16  * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
  17  */
  18
  19 #include "busybox.h"
  20
  21 #define ASCII 0377
  22
  23 /* some "globals" shared across this file */
  24 static char com_fl, del_fl, sq_fl;
  25 /* these last are pointers to static buffers declared in tr_main */
  26 static unsigned char *poutput;
  27 static unsigned char *pvector;
  28 static unsigned char *pinvec, *poutvec;
  29
  30 static void convert(void)
  31 {
  32         int read_chars = 0, in_index = 0, out_index = 0, c, coded, last = -1;
  33
  34         for (;;) {
  35                 // If we're out of input, flush output and read more input.
  36
  37                 if (in_index == read_chars) {
  38                         if (out_index) {
  39                                 if (write(1, (char *) poutput, out_index) != out_index)
  40                                         bb_error_msg_and_die(bb_msg_write_error);
  41                                 out_index = 0;
  42                         }
  43
  44                         if ((read_chars = read(0, bb_common_bufsiz1, BUFSIZ)) <= 0) {
  45                                 if (write(1, (char *) poutput, out_index) != out_index)
  46                                         bb_error_msg(bb_msg_write_error);
  47                                 exit(0);
  48                         }
  49                         in_index = 0;
  50                 }
  51                 c = bb_common_bufsiz1[in_index++];
  52                 coded = pvector[c];
  53                 if (del_fl && pinvec[c])
  54                         continue;
  55                 if (sq_fl && last == coded && (pinvec[c] || poutvec[coded]))
  56                         continue;
  57                 poutput[out_index++] = last = coded;
  58         }
  59
  60         /* NOTREACHED */
  61 }
  62
  63 static void map(register unsigned char *string1, unsigned int string1_len,
  64                 register unsigned char *string2, unsigned int string2_len)
  65 {
  66         unsigned char last = '0';
  67         unsigned int i, j;
  68
  69         for (j = 0, i = 0; i < string1_len; i++) {
  70                 if (string2_len <= j)
  71                         pvector[string1[i]] = last;
  72                 else
  73                         pvector[string1[i]] = last = string2[j++];
  74         }
  75 }
  76
  77 /* supported constructs:
  78  *   Ranges,  e.g.,  [0-9]  ==>  0123456789
  79  *   Escapes, e.g.,  \a     ==>  Control-G
  80  *       Character classes, e.g. [:upper:] ==> A ... Z
  81  */
  82 static unsigned int expand(const char *arg, register unsigned char *buffer)
  83 {
  84         unsigned char *buffer_start = buffer;
  85         int i, ac;
  86
  87         while (*arg) {
  88                 if (*arg == '\\') {
  89                         arg++;
  90                         *buffer++ = bb_process_escape_sequence(&arg);
  91                 } else if (*(arg+1) == '-') {
  92                         ac = *(arg+2);
  93                         if(ac == 0) {
  94                                 *buffer++ = *arg++;
  95                                 continue;
  96                         }
  97                         i = *arg;
  98                         while (i <= ac)
  99                                 *buffer++ = i++;
 100                         arg += 3; /* Skip the assumed a-z */
 101                 } else if (*arg == '[') {
 102                         arg++;
 103                         i = *arg++;
 104                         if (ENABLE_FEATURE_TR_CLASSES && i == ':') {
 105                                 if (strncmp(arg, "alpha", 5) == 0) {
 106                                         for (i = 'A'; i <= 'Z'; i++)
 107                                                 *buffer++ = i;
 108                                         for (i = 'a'; i <= 'z'; i++)
 109                                                 *buffer++ = i;
 110                                 }
 111                                 else if (strncmp(arg, "alnum", 5) == 0) {
 112                                         for (i = '0'; i <= '9'; i++)
 113                                                 *buffer++ = i;
 114                                         for (i = 'A'; i <= 'Z'; i++)
 115                                                 *buffer++ = i;
 116                                         for (i = 'a'; i <= 'z'; i++)
 117                                                 *buffer++ = i;
 118                                 }
 119                                 else if (strncmp(arg, "digit", 5) == 0)
 120                                         for (i = '0'; i <= '9'; i++)
 121                                                 *buffer++ = i;
 122                                 else if (strncmp(arg, "lower", 5) == 0)
 123                                         for (i = 'a'; i <= 'z'; i++)
 124                                                 *buffer++ = i;
 125                                 else if (strncmp(arg, "upper", 5) == 0)
 126                                         for (i = 'A'; i <= 'Z'; i++)
 127                                                 *buffer++ = i;
 128                                 else if (strncmp(arg, "space", 5) == 0) {
 129                                     const char s[] = "\t\n\v\f\r ";
 130                                         strcat((char*)buffer, s);
 131                                         buffer += sizeof(s) - 1;
 132                                 }
 133                                 else if (strncmp(arg, "blank", 5) == 0) {
 134                                         *buffer++ = '\t';
 135                                         *buffer++ = ' ';
 136                                 }
 137                                 /* gcc gives a warning if braces aren't used here */
 138                                 else if (strncmp(arg, "punct", 5) == 0) {
 139                                         for (i = 0; i <= ASCII; i++)
 140                                                 if (isprint(i) && (!isalnum(i)) && (!isspace(i)))
 141                                                         *buffer++ = i;
 142                                 }
 143                                 else if (strncmp(arg, "cntrl", 5) == 0) {
 144                                         for (i = 0; i <= ASCII; i++)
 145                                                 if (iscntrl(i))
 146                                                         *buffer++ = i;
 147                                 }
 148                                 else {
 149                                         *buffer++ = '[';
 150                                         *buffer++ = ':';
 151                                         continue;
 152                                 }
 153                                 break;
 154                         }
 155                         if (ENABLE_FEATURE_TR_EQUIV && i == '=') {
 156                                 *buffer++ = *arg;
 157                                 /* skip the closing =] */
 158                                 arg += 3;
 159                                 continue;
 160                         }
 161                         if (*arg++ != '-') {
 162                                 *buffer++ = '[';
 163                                 arg -= 2;
 164                                 continue;
 165                         }
 166                         ac = *arg++;
 167                         while (i <= ac)
 168                                 *buffer++ = i++;
 169                         arg++;                          /* Skip the assumed ']' */
 170                 } else
 171                         *buffer++ = *arg++;
 172         }
 173
 174         return (buffer - buffer_start);
 175 }
 176
 177 static int complement(unsigned char *buffer, int buffer_len)
 178 {
 179         register short i, j, ix;
 180         char conv[ASCII + 2];
 181
 182         ix = 0;
 183         for (i = 0; i <= ASCII; i++) {
 184                 for (j = 0; j < buffer_len; j++)
 185                         if (buffer[j] == i)
 186                                 break;
 187                 if (j == buffer_len)
 188                         conv[ix++] = i & ASCII;
 189         }
 190         memcpy(buffer, conv, ix);
 191         return ix;
 192 }
 193
 194 int tr_main(int argc, char **argv)
 195 {
 196         register unsigned char *ptr;
 197         int output_length=0, input_length;
 198         int idx = 1;
 199         int i;
 200         RESERVE_CONFIG_BUFFER(output, BUFSIZ);
 201         RESERVE_CONFIG_BUFFER(vector, ASCII+1);
 202         RESERVE_CONFIG_BUFFER(invec,  ASCII+1);
 203         RESERVE_CONFIG_BUFFER(outvec, ASCII+1);
 204
 205         /* ... but make them available globally */
 206         poutput = (unsigned char*)output;
 207         pvector = (unsigned char*)vector;
 208         pinvec  = (unsigned char*)invec;
 209         poutvec = (unsigned char*)outvec;
 210
 211         if (argc > 1 && argv[idx][0] == '-') {
 212                 for (ptr = (unsigned char *) &argv[idx][1]; *ptr; ptr++) {
 213                         switch (*ptr) {
 214                         case 'c':
 215                                 com_fl = TRUE;
 216                                 break;
 217                         case 'd':
 218                                 del_fl = TRUE;
 219                                 break;
 220                         case 's':
 221                                 sq_fl = TRUE;
 222                                 break;
 223                         default:
 224                                 bb_show_usage();
 225                         }
 226                 }
 227                 idx++;
 228         }
 229         for (i = 0; i <= ASCII; i++) {
 230                 vector[i] = i;
 231                 invec[i] = outvec[i] = FALSE;
 232         }
 233
 234         if (argv[idx] != NULL) {
 235                 input_length = expand(argv[idx++], bb_common_bufsiz1);
 236                 if (com_fl)
 237                         input_length = complement(bb_common_bufsiz1, input_length);
 238                 if (argv[idx] != NULL) {
 239                         if (*argv[idx] == '\0')
 240                                 bb_error_msg_and_die("STRING2 cannot be empty");
 241                         output_length = expand(argv[idx], (unsigned char*)output);
 242                         map(bb_common_bufsiz1, input_length, (unsigned char*)output, output_length);
 243                 }
 244                 for (i = 0; i < input_length; i++)
 245                         invec[bb_common_bufsiz1[i]] = TRUE;
 246                 for (i = 0; i < output_length; i++)
 247                         outvec[(unsigned char)output[i]] = TRUE;
 248         }
 249         convert();
 250         return (0);
 251 }