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 // Even with -funsigned-char, gcc still complains about char as an array index.
  22
  23 #define GCC4_IS_STUPID int
  24
  25 #define ASCII 0377
  26
  27 /* some "globals" shared across this file */
  28 static char com_fl, del_fl, sq_fl;
  29 /* these last are pointers to static buffers declared in tr_main */
  30 static char *poutput, *pvector, *pinvec, *poutvec;
  31
  32 static void convert(void)
  33 {
  34         int read_chars = 0, in_index = 0, out_index = 0, c, coded, last = -1;
  35
  36         for (;;) {
  37                 // If we're out of input, flush output and read more input.
  38
  39                 if (in_index == read_chars) {
  40                         if (out_index) {
  41                                 if (write(1, (char *) poutput, out_index) != out_index)
  42                                         bb_error_msg_and_die(bb_msg_write_error);
  43                                 out_index = 0;
  44                         }
  45
  46                         if ((read_chars = read(0, bb_common_bufsiz1, BUFSIZ)) <= 0) {
  47                                 if (write(1, (char *) poutput, out_index) != out_index)
  48                                         bb_error_msg(bb_msg_write_error);
  49                                 exit(0);
  50                         }
  51                         in_index = 0;
  52                 }
  53                 c = bb_common_bufsiz1[in_index++];
  54                 coded = pvector[c];
  55                 if (del_fl && pinvec[c])
  56                         continue;
  57                 if (sq_fl && last == coded && (pinvec[c] || poutvec[coded]))
  58                         continue;
  59                 poutput[out_index++] = last = coded;
  60         }
  61
  62         /* NOTREACHED */
  63 }
  64
  65 static void map(char *string1, unsigned int string1_len,
  66                 char *string2, unsigned int string2_len)
  67 {
  68         char last = '0';
  69         unsigned int i, j;
  70
  71         for (j = 0, i = 0; i < string1_len; i++) {
  72                 if (string2_len <= j)
  73                         pvector[(GCC4_IS_STUPID)string1[i]] = last;
  74                 else
  75                         pvector[(GCC4_IS_STUPID)string1[i]] = last = string2[j++];
  76         }
  77 }
  78
  79 /* supported constructs:
  80  *   Ranges,  e.g.,  [0-9]  ==>  0123456789
  81  *   Escapes, e.g.,  \a     ==>  Control-G
  82  *       Character classes, e.g. [:upper:] ==> A ... Z
  83  */
  84 static unsigned int expand(const char *arg, char *buffer)
  85 {
  86         char *buffer_start = buffer;
  87         int i, ac;
  88
  89         while (*arg) {
  90                 if (*arg == '\\') {
  91                         arg++;
  92                         *buffer++ = bb_process_escape_sequence(&arg);
  93                 } else if (*(arg+1) == '-') {
  94                         ac = *(arg+2);
  95                         if(ac == 0) {
  96                                 *buffer++ = *arg++;
  97                                 continue;
  98                         }
  99                         i = *arg;
 100                         while (i <= ac)
 101                                 *buffer++ = i++;
 102                         arg += 3; /* Skip the assumed a-z */
 103                 } else if (*arg == '[') {
 104                         arg++;
 105                         i = *arg++;
 106                         if (ENABLE_FEATURE_TR_CLASSES && i == ':') {
 107                                 if (strncmp(arg, "alpha", 5) == 0) {
 108                                         for (i = 'A'; i <= 'Z'; i++)
 109                                                 *buffer++ = i;
 110                                         for (i = 'a'; i <= 'z'; i++)
 111                                                 *buffer++ = i;
 112                                 }
 113                                 else if (strncmp(arg, "alnum", 5) == 0) {
 114                                         for (i = '0'; i <= '9'; i++)
 115                                                 *buffer++ = i;
 116                                         for (i = 'A'; i <= 'Z'; i++)
 117                                                 *buffer++ = i;
 118                                         for (i = 'a'; i <= 'z'; i++)
 119                                                 *buffer++ = i;
 120                                 }
 121                                 else if (strncmp(arg, "digit", 5) == 0)
 122                                         for (i = '0'; i <= '9'; i++)
 123                                                 *buffer++ = i;
 124                                 else if (strncmp(arg, "lower", 5) == 0)
 125                                         for (i = 'a'; i <= 'z'; i++)
 126                                                 *buffer++ = i;
 127                                 else if (strncmp(arg, "upper", 5) == 0)
 128                                         for (i = 'A'; i <= 'Z'; i++)
 129                                                 *buffer++ = i;
 130                                 else if (strncmp(arg, "space", 5) == 0) {
 131                                     const char s[] = "\t\n\v\f\r ";
 132                                         strcat((char*)buffer, s);
 133                                         buffer += sizeof(s) - 1;
 134                                 }
 135                                 else if (strncmp(arg, "blank", 5) == 0) {
 136                                         *buffer++ = '\t';
 137                                         *buffer++ = ' ';
 138                                 }
 139                                 /* gcc gives a warning if braces aren't used here */
 140                                 else if (strncmp(arg, "punct", 5) == 0) {
 141                                         for (i = 0; i <= ASCII; i++)
 142                                                 if (isprint(i) && (!isalnum(i)) && (!isspace(i)))
 143                                                         *buffer++ = i;
 144                                 }
 145                                 else if (strncmp(arg, "cntrl", 5) == 0) {
 146                                         for (i = 0; i <= ASCII; i++)
 147                                                 if (iscntrl(i))
 148                                                         *buffer++ = i;
 149                                 }
 150                                 else {
 151                                         *buffer++ = '[';
 152                                         *buffer++ = ':';
 153                                         continue;
 154                                 }
 155                                 break;
 156                         }
 157                         if (ENABLE_FEATURE_TR_EQUIV && i == '=') {
 158                                 *buffer++ = *arg;
 159                                 /* skip the closing =] */
 160                                 arg += 3;
 161                                 continue;
 162                         }
 163                         if (*arg++ != '-') {
 164                                 *buffer++ = '[';
 165                                 arg -= 2;
 166                                 continue;
 167                         }
 168                         ac = *arg++;
 169                         while (i <= ac)
 170                                 *buffer++ = i++;
 171                         arg++;                          /* Skip the assumed ']' */
 172                 } else
 173                         *buffer++ = *arg++;
 174         }
 175
 176         return (buffer - buffer_start);
 177 }
 178
 179 static int complement(char *buffer, int buffer_len)
 180 {
 181         short i, j, ix;
 182         char conv[ASCII + 2];
 183
 184         ix = 0;
 185         for (i = 0; i <= ASCII; i++) {
 186                 for (j = 0; j < buffer_len; j++)
 187                         if (buffer[j] == i)
 188                                 break;
 189                 if (j == buffer_len)
 190                         conv[ix++] = i & ASCII;
 191         }
 192         memcpy(buffer, conv, ix);
 193         return ix;
 194 }
 195
 196 int tr_main(int argc, char **argv)
 197 {
 198         unsigned char *ptr;
 199         int output_length=0, input_length;
 200         int idx = 1;
 201         int i;
 202         RESERVE_CONFIG_BUFFER(output, BUFSIZ);
 203         RESERVE_CONFIG_BUFFER(vector, ASCII+1);
 204         RESERVE_CONFIG_BUFFER(invec,  ASCII+1);
 205         RESERVE_CONFIG_BUFFER(outvec, ASCII+1);
 206
 207         /* ... but make them available globally */
 208         poutput = output;
 209         pvector = vector;
 210         pinvec  = invec;
 211         poutvec = outvec;
 212
 213         if (argc > 1 && argv[idx][0] == '-') {
 214                 for (ptr = (unsigned char *) &argv[idx][1]; *ptr; ptr++) {
 215                         switch (*ptr) {
 216                         case 'c':
 217                                 com_fl = TRUE;
 218                                 break;
 219                         case 'd':
 220                                 del_fl = TRUE;
 221                                 break;
 222                         case 's':
 223                                 sq_fl = TRUE;
 224                                 break;
 225                         default:
 226                                 bb_show_usage();
 227                         }
 228                 }
 229                 idx++;
 230         }
 231         for (i = 0; i <= ASCII; i++) {
 232                 vector[i] = i;
 233                 invec[i] = outvec[i] = FALSE;
 234         }
 235
 236         if (argv[idx] != NULL) {
 237                 input_length = expand(argv[idx++], bb_common_bufsiz1);
 238                 if (com_fl)
 239                         input_length = complement(bb_common_bufsiz1, input_length);
 240                 if (argv[idx] != NULL) {
 241                         if (*argv[idx] == '\0')
 242                                 bb_error_msg_and_die("STRING2 cannot be empty");
 243                         output_length = expand(argv[idx], output);
 244                         map(bb_common_bufsiz1, input_length, output, output_length);
 245                 }
 246                 for (i = 0; i < input_length; i++)
 247                         invec[(GCC4_IS_STUPID)bb_common_bufsiz1[i]] = TRUE;
 248                 for (i = 0; i < output_length; i++)
 249                         outvec[(GCC4_IS_STUPID)output[i]] = TRUE;
 250         }
 251         convert();
 252         return (0);
 253 }