coreutils/tr.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * Mini tr implementation for busybox
   4  *
   5  * Copyright (c) Michiel Huisjes
   6  *
   7  * This version of tr is adapted from Minix tr and was modified
   8  * by Erik Andersen <andersen@codepoet.org> to be used in busybox.
   9  *
  10  * This program is free software; you can redistribute it and/or modify
  11  * it under the terms of the GNU General Public License as published by
  12  * the Free Software Foundation; either version 2 of the License, or
  13  * (at your option) any later version.
  14  *
  15  * This program is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  18  * General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU General Public License
  21  * along with this program; if not, write to the Free Software
  22  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  23  *
  24  * Original copyright notice is retained at the end of this file.
  25  */
  26
  27 #include <stdio.h>
  28 #include <string.h>
  29 #include <stdlib.h>
  30 #include <unistd.h>
  31 #include <ctype.h>
  32 #include <sys/types.h>
  33 #include "busybox.h"
  34
  35 #define ASCII 0377
  36
  37 /* some "globals" shared across this file */
  38 static char com_fl, del_fl, sq_fl;
  39 static short in_index, out_index;
  40 /* these last are pointers to static buffers declared in tr_main */
  41 static unsigned char *poutput, *pinput;
  42 static unsigned char *pvector;
  43 static char *pinvec, *poutvec;
  44
  45
  46 static void convert(void)
  47 {
  48         short read_chars = 0;
  49         short c, coded;
  50         short last = -1;
  51
  52         for (;;) {
  53                 if (in_index == read_chars) {
  54                         if ((read_chars = read(0, (char *) pinput, BUFSIZ)) <= 0) {
  55                                 if (write(1, (char *) poutput, out_index) != out_index)
  56                                         bb_error_msg(bb_msg_write_error);
  57                                 exit(0);
  58                         }
  59                         in_index = 0;
  60                 }
  61                 c = pinput[in_index++];
  62                 coded = pvector[c];
  63                 if (del_fl && pinvec[c])
  64                         continue;
  65                 if (sq_fl && last == coded && (pinvec[c] || poutvec[coded]))
  66                         continue;
  67                 poutput[out_index++] = last = coded;
  68                 if (out_index == BUFSIZ) {
  69                         if (write(1, (char *) poutput, out_index) != out_index)
  70                                 bb_error_msg_and_die(bb_msg_write_error);
  71                         out_index = 0;
  72                 }
  73         }
  74
  75         /* NOTREACHED */
  76 }
  77
  78 static void map(register unsigned char *string1, unsigned int string1_len,
  79                 register unsigned char *string2, unsigned int string2_len)
  80 {
  81         unsigned char last = '0';
  82         unsigned int i, j;
  83
  84         for (j = 0, i = 0; i < string1_len; i++) {
  85                 if (string2_len <= j)
  86                         pvector[string1[i]] = last;
  87                 else
  88                         pvector[string1[i]] = last = string2[j++];
  89         }
  90 }
  91
  92 /* supported constructs:
  93  *   Ranges,  e.g.,  [0-9]  ==>  0123456789
  94  *   Escapes, e.g.,  \a     ==>  Control-G
  95  *       Character classes, e.g. [:upper:] ==> A ... Z
  96  */
  97 static unsigned int expand(const char *arg, register unsigned char *buffer)
  98 {
  99         unsigned char *buffer_start = buffer;
 100         int i, ac;
 101
 102         while (*arg) {
 103                 if (*arg == '\\') {
 104                         arg++;
 105                         *buffer++ = bb_process_escape_sequence(&arg);
 106                 } else if (*(arg+1) == '-') {
 107                         ac = *(arg+2);
 108                         if(ac == 0) {
 109                                 *buffer++ = *arg++;
 110                                 continue;
 111                         }
 112                         i = *arg;
 113                         while (i <= ac)
 114                                 *buffer++ = i++;
 115                         arg += 3; /* Skip the assumed a-z */
 116                 } else if (*arg == '[') {
 117                         arg++;
 118                         if (ENABLE_FEATURE_TR_CLASSES && *arg++ == ':') {
 119                                 if (strncmp(arg, "alpha", 5) == 0) {
 120                                         for (i = 'A'; i <= 'Z'; i++)
 121                                                 *buffer++ = i;
 122                                         for (i = 'a'; i <= 'z'; i++)
 123                                                 *buffer++ = i;
 124                                 }
 125                                 else if (strncmp(arg, "alnum", 5) == 0) {
 126                                         for (i = 'A'; i <= 'Z'; i++)
 127                                                 *buffer++ = i;
 128                                         for (i = 'a'; i <= 'z'; i++)
 129                                                 *buffer++ = i;
 130                                         for (i = '0'; i <= '9'; i++)
 131                                                 *buffer++ = i;
 132                                 }
 133                                 else if (strncmp(arg, "digit", 5) == 0)
 134                                         for (i = '0'; i <= '9'; i++)
 135                                                 *buffer++ = i;
 136                                 else if (strncmp(arg, "lower", 5) == 0)
 137                                         for (i = 'a'; i <= 'z'; i++)
 138                                                 *buffer++ = i;
 139                                 else if (strncmp(arg, "upper", 5) == 0)
 140                                         for (i = 'A'; i <= 'Z'; i++)
 141                                                 *buffer++ = i;
 142                                 else if (strncmp(arg, "space", 5) == 0)
 143                                         strcat(buffer, " \f\n\r\t\v");
 144                                 else if (strncmp(arg, "blank", 5) == 0)
 145                                         strcat(buffer, " \t");
 146                                 /* gcc gives a warning if braces aren't used here */
 147                                 else if (strncmp(arg, "punct", 5) == 0) {
 148                                         for (i = 0; i <= ASCII; i++)
 149                                                 if (isprint(i) && (!isalnum(i)) && (!isspace(i)))
 150                                                         *buffer++ = i;
 151                                 }
 152                                 else if (strncmp(arg, "cntrl", 5) == 0) {
 153                                         for (i = 0; i <= ASCII; i++)
 154                                                 if (iscntrl(i))
 155                                                         *buffer++ = i;
 156                                 }
 157                                 else {
 158                                         strcat(buffer, "[:");
 159                                         arg++;
 160                                         continue;
 161                                 }
 162                                 break;
 163                         }
 164                         if (ENABLE_FEATURE_TR_EQUIV && *arg++ == '=') {
 165                                 *buffer++ = *arg;
 166                                 /* skip the closing =] */
 167                                 arg += 3;
 168                                 continue;
 169                         }
 170                         if (*arg++ != '-') {
 171                                 *buffer++ = '[';
 172                                 arg -= 2;
 173                                 continue;
 174                         }
 175                         i = *arg++;
 176                         ac = *arg++;
 177                         while (i <= ac)
 178                                 *buffer++ = i++;
 179                         arg++;                          /* Skip the assumed ']' */
 180                 } else
 181                         *buffer++ = *arg++;
 182         }
 183
 184         return (buffer - buffer_start);
 185 }
 186
 187 static int complement(unsigned char *buffer, int buffer_len)
 188 {
 189         register short i, j, ix;
 190         char conv[ASCII + 2];
 191
 192         ix = 0;
 193         for (i = 0; i <= ASCII; i++) {
 194                 for (j = 0; j < buffer_len; j++)
 195                         if (buffer[j] == i)
 196                                 break;
 197                 if (j == buffer_len)
 198                         conv[ix++] = i & ASCII;
 199         }
 200         memcpy(buffer, conv, ix);
 201         return ix;
 202 }
 203
 204 extern int tr_main(int argc, char **argv)
 205 {
 206         register unsigned char *ptr;
 207         int output_length=0, input_length;
 208         int idx = 1;
 209         int i;
 210         RESERVE_CONFIG_BUFFER(output, BUFSIZ);
 211         RESERVE_CONFIG_BUFFER(input,  BUFSIZ);
 212         RESERVE_CONFIG_UBUFFER(vector, ASCII+1);
 213         RESERVE_CONFIG_BUFFER(invec,  ASCII+1);
 214         RESERVE_CONFIG_BUFFER(outvec, ASCII+1);
 215
 216         /* ... but make them available globally */
 217         poutput = output;
 218         pinput  = input;
 219         pvector = vector;
 220         pinvec  = invec;
 221         poutvec = outvec;
 222
 223         if (argc > 1 && argv[idx][0] == '-') {
 224                 for (ptr = (unsigned char *) &argv[idx][1]; *ptr; ptr++) {
 225                         switch (*ptr) {
 226                         case 'c':
 227                                 com_fl = TRUE;
 228                                 break;
 229                         case 'd':
 230                                 del_fl = TRUE;
 231                                 break;
 232                         case 's':
 233                                 sq_fl = TRUE;
 234                                 break;
 235                         default:
 236                                 bb_show_usage();
 237                         }
 238                 }
 239                 idx++;
 240         }
 241         for (i = 0; i <= ASCII; i++) {
 242                 vector[i] = i;
 243                 invec[i] = outvec[i] = FALSE;
 244         }
 245
 246         if (argv[idx] != NULL) {
 247                 input_length = expand(argv[idx++], input);
 248                 if (com_fl)
 249                         input_length = complement(input, input_length);
 250                 if (argv[idx] != NULL) {
 251                         if (*argv[idx] == '\0')
 252                                 bb_error_msg_and_die("STRING2 cannot be empty");
 253                         output_length = expand(argv[idx], output);
 254                         map(input, input_length, output, output_length);
 255                 }
 256                 for (i = 0; i < input_length; i++)
 257                         invec[(unsigned char)input[i]] = TRUE;
 258                 for (i = 0; i < output_length; i++)
 259                         outvec[(unsigned char)output[i]] = TRUE;
 260         }
 261         convert();
 262         return (0);
 263 }
 264
 265 /*
 266  * Copyright (c) 1987,1997, Prentice Hall
 267  * All rights reserved.
 268  *
 269  * Redistribution and use of the MINIX operating system in source and
 270  * binary forms, with or without modification, are permitted provided
 271  * that the following conditions are met:
 272  *
 273  * Redistributions of source code must retain the above copyright
 274  * notice, this list of conditions and the following disclaimer.
 275  *
 276  * Redistributions in binary form must reproduce the above
 277  * copyright notice, this list of conditions and the following
 278  * disclaimer in the documentation and/or other materials provided
 279  * with the distribution.
 280  *
 281  * Neither the name of Prentice Hall nor the names of the software
 282  * authors or contributors may be used to endorse or promote
 283  * products derived from this software without specific prior
 284  * written permission.
 285  *
 286  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS, AUTHORS, AND
 287  * CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES,
 288  * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
 289  * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
 290  * IN NO EVENT SHALL PRENTICE HALL OR ANY AUTHORS OR CONTRIBUTORS BE
 291  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 292  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 293  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
 294  * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
 295  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
 296  * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
 297  * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 298  *
 299  */
 300