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;
  42 static unsigned char *pvector;
  43 static unsigned char *pinvec, *poutvec;
  44
  45 #define input bb_common_bufsiz1
  46
  47 static void convert(void)
  48 {
  49         short read_chars = 0;
  50         short c, coded;
  51         short last = -1;
  52
  53         for (;;) {
  54                 if (in_index == read_chars) {
  55                         if ((read_chars = read(0, input, BUFSIZ)) <= 0) {
  56                                 if (write(1, (char *) poutput, out_index) != out_index)
  57                                         bb_error_msg(bb_msg_write_error);
  58                                 exit(0);
  59                         }
  60                         in_index = 0;
  61                 }
  62                 c = input[in_index++];
  63                 coded = pvector[c];
  64                 if (del_fl && pinvec[c])
  65                         continue;
  66                 if (sq_fl && last == coded && (pinvec[c] || poutvec[coded]))
  67                         continue;
  68                 poutput[out_index++] = last = coded;
  69                 if (out_index == BUFSIZ) {
  70                         if (write(1, (char *) poutput, out_index) != out_index)
  71                                 bb_error_msg_and_die(bb_msg_write_error);
  72                         out_index = 0;
  73                 }
  74         }
  75
  76         /* NOTREACHED */
  77 }
  78
  79 static void map(register unsigned char *string1, unsigned int string1_len,
  80                 register unsigned char *string2, unsigned int string2_len)
  81 {
  82         unsigned char last = '0';
  83         unsigned int i, j;
  84
  85         for (j = 0, i = 0; i < string1_len; i++) {
  86                 if (string2_len <= j)
  87                         pvector[string1[i]] = last;
  88                 else
  89                         pvector[string1[i]] = last = string2[j++];
  90         }
  91 }
  92
  93 /* supported constructs:
  94  *   Ranges,  e.g.,  [0-9]  ==>  0123456789
  95  *   Escapes, e.g.,  \a     ==>  Control-G
  96  *       Character classes, e.g. [:upper:] ==> A ... Z
  97  */
  98 static unsigned int expand(const char *arg, register unsigned char *buffer)
  99 {
 100         unsigned char *buffer_start = buffer;
 101         int i, ac;
 102
 103         while (*arg) {
 104                 if (*arg == '\\') {
 105                         arg++;
 106                         *buffer++ = bb_process_escape_sequence(&arg);
 107                 } else if (*(arg+1) == '-') {
 108                         ac = *(arg+2);
 109                         if(ac == 0) {
 110                                 *buffer++ = *arg++;
 111                                 continue;
 112                         }
 113                         i = *arg;
 114                         while (i <= ac)
 115                                 *buffer++ = i++;
 116                         arg += 3; /* Skip the assumed a-z */
 117                 } else if (*arg == '[') {
 118                         arg++;
 119                         i = *arg++;
 120                         if (ENABLE_FEATURE_TR_CLASSES && i == ':') {
 121                                 if (strncmp(arg, "alpha", 5) == 0) {
 122                                         for (i = 'A'; i <= 'Z'; i++)
 123                                                 *buffer++ = i;
 124                                         for (i = 'a'; i <= 'z'; i++)
 125                                                 *buffer++ = i;
 126                                 }
 127                                 else if (strncmp(arg, "alnum", 5) == 0) {
 128                                         for (i = '0'; i <= '9'; i++)
 129                                                 *buffer++ = i;
 130                                         for (i = 'A'; i <= 'Z'; i++)
 131                                                 *buffer++ = i;
 132                                         for (i = 'a'; i <= 'z'; i++)
 133                                                 *buffer++ = i;
 134                                 }
 135                                 else if (strncmp(arg, "digit", 5) == 0)
 136                                         for (i = '0'; i <= '9'; i++)
 137                                                 *buffer++ = i;
 138                                 else if (strncmp(arg, "lower", 5) == 0)
 139                                         for (i = 'a'; i <= 'z'; i++)
 140                                                 *buffer++ = i;
 141                                 else if (strncmp(arg, "upper", 5) == 0)
 142                                         for (i = 'A'; i <= 'Z'; i++)
 143                                                 *buffer++ = i;
 144                                 else if (strncmp(arg, "space", 5) == 0) {
 145                                     const char s[] = "\t\n\v\f\r ";
 146                                         strcat((char*)buffer, s);
 147                                         buffer += sizeof(s) - 1;
 148                                 }
 149                                 else if (strncmp(arg, "blank", 5) == 0) {
 150                                         *buffer++ = '\t';
 151                                         *buffer++ = ' ';
 152                                 }
 153                                 /* gcc gives a warning if braces aren't used here */
 154                                 else if (strncmp(arg, "punct", 5) == 0) {
 155                                         for (i = 0; i <= ASCII; i++)
 156                                                 if (isprint(i) && (!isalnum(i)) && (!isspace(i)))
 157                                                         *buffer++ = i;
 158                                 }
 159                                 else if (strncmp(arg, "cntrl", 5) == 0) {
 160                                         for (i = 0; i <= ASCII; i++)
 161                                                 if (iscntrl(i))
 162                                                         *buffer++ = i;
 163                                 }
 164                                 else {
 165                                         *buffer++ = '[';
 166                                         *buffer++ = ':';
 167                                         continue;
 168                                 }
 169                                 break;
 170                         }
 171                         if (ENABLE_FEATURE_TR_EQUIV && i == '=') {
 172                                 *buffer++ = *arg;
 173                                 /* skip the closing =] */
 174                                 arg += 3;
 175                                 continue;
 176                         }
 177                         if (*arg++ != '-') {
 178                                 *buffer++ = '[';
 179                                 arg -= 2;
 180                                 continue;
 181                         }
 182                         ac = *arg++;
 183                         while (i <= ac)
 184                                 *buffer++ = i++;
 185                         arg++;                          /* Skip the assumed ']' */
 186                 } else
 187                         *buffer++ = *arg++;
 188         }
 189
 190         return (buffer - buffer_start);
 191 }
 192
 193 static int complement(unsigned char *buffer, int buffer_len)
 194 {
 195         register short i, j, ix;
 196         char conv[ASCII + 2];
 197
 198         ix = 0;
 199         for (i = 0; i <= ASCII; i++) {
 200                 for (j = 0; j < buffer_len; j++)
 201                         if (buffer[j] == i)
 202                                 break;
 203                 if (j == buffer_len)
 204                         conv[ix++] = i & ASCII;
 205         }
 206         memcpy(buffer, conv, ix);
 207         return ix;
 208 }
 209
 210 int tr_main(int argc, char **argv)
 211 {
 212         register unsigned char *ptr;
 213         int output_length=0, input_length;
 214         int idx = 1;
 215         int i;
 216         RESERVE_CONFIG_BUFFER(output, BUFSIZ);
 217         RESERVE_CONFIG_BUFFER(vector, ASCII+1);
 218         RESERVE_CONFIG_BUFFER(invec,  ASCII+1);
 219         RESERVE_CONFIG_BUFFER(outvec, ASCII+1);
 220
 221         /* ... but make them available globally */
 222         poutput = (unsigned char*)output;
 223         pvector = (unsigned char*)vector;
 224         pinvec  = (unsigned char*)invec;
 225         poutvec = (unsigned char*)outvec;
 226
 227         if (argc > 1 && argv[idx][0] == '-') {
 228                 for (ptr = (unsigned char *) &argv[idx][1]; *ptr; ptr++) {
 229                         switch (*ptr) {
 230                         case 'c':
 231                                 com_fl = TRUE;
 232                                 break;
 233                         case 'd':
 234                                 del_fl = TRUE;
 235                                 break;
 236                         case 's':
 237                                 sq_fl = TRUE;
 238                                 break;
 239                         default:
 240                                 bb_show_usage();
 241                         }
 242                 }
 243                 idx++;
 244         }
 245         for (i = 0; i <= ASCII; i++) {
 246                 vector[i] = i;
 247                 invec[i] = outvec[i] = FALSE;
 248         }
 249
 250         if (argv[idx] != NULL) {
 251                 input_length = expand(argv[idx++], (unsigned char*)input);
 252                 if (com_fl)
 253                         input_length = complement((unsigned char*)input, input_length);
 254                 if (argv[idx] != NULL) {
 255                         if (*argv[idx] == '\0')
 256                                 bb_error_msg_and_die("STRING2 cannot be empty");
 257                         output_length = expand(argv[idx], (unsigned char*)output);
 258                         map((unsigned char*)input, input_length, (unsigned char*)output, output_length);
 259                 }
 260                 for (i = 0; i < input_length; i++)
 261                         invec[(unsigned char)input[i]] = TRUE;
 262                 for (i = 0; i < output_length; i++)
 263                         outvec[(unsigned char)output[i]] = TRUE;
 264         }
 265         convert();
 266         return (0);
 267 }
 268
 269 /*
 270  * Copyright (c) 1987,1997, Prentice Hall
 271  * All rights reserved.
 272  *
 273  * Redistribution and use of the MINIX operating system in source and
 274  * binary forms, with or without modification, are permitted provided
 275  * that the following conditions are met:
 276  *
 277  * Redistributions of source code must retain the above copyright
 278  * notice, this list of conditions and the following disclaimer.
 279  *
 280  * Redistributions in binary form must reproduce the above
 281  * copyright notice, this list of conditions and the following
 282  * disclaimer in the documentation and/or other materials provided
 283  * with the distribution.
 284  *
 285  * Neither the name of Prentice Hall nor the names of the software
 286  * authors or contributors may be used to endorse or promote
 287  * products derived from this software without specific prior
 288  * written permission.
 289  *
 290  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS, AUTHORS, AND
 291  * CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES,
 292  * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
 293  * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
 294  * IN NO EVENT SHALL PRENTICE HALL OR ANY AUTHORS OR CONTRIBUTORS BE
 295  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 296  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 297  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
 298  * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
 299  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
 300  * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
 301  * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 302  *
 303  */
 304