box/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 /* http://www.opengroup.org/onlinepubs/009695399/utilities/tr.html
  19  * TODO: xdigit, graph, print
  20  */
  21 #include "libbb.h"
  22
  23 #define ASCII 0377
  24
  25 #define TR_OPT_complement       (1<<0)
  26 #define TR_OPT_delete           (1<<1)
  27 #define TR_OPT_squeeze_reps     (1<<2)
  28 /* some "globals" shared across this file */
  29 /* these last are pointers to static buffers declared in tr_main */
  30 static char *poutput, *pvector, *pinvec, *poutvec;
  31
  32 static void ATTRIBUTE_NORETURN convert(const smalluint flags)
  33 {
  34         size_t 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                 if (in_index == read_chars) {
  39                         if (out_index) {
  40                                 xwrite(STDOUT_FILENO, (char *)poutput, out_index);
  41                                 out_index = 0;
  42                         }
  43                         if ((read_chars = read(STDIN_FILENO, bb_common_bufsiz1, BUFSIZ)) <= 0) {
  44                                 if (write(STDOUT_FILENO, (char *)poutput, out_index) != out_index)
  45                                         bb_perror_msg(bb_msg_write_error);
  46                                 exit(EXIT_SUCCESS);
  47                         }
  48                         in_index = 0;
  49                 }
  50                 c = bb_common_bufsiz1[in_index++];
  51                 coded = pvector[c];
  52                 if ((flags & TR_OPT_delete) && pinvec[c])
  53                         continue;
  54                 if ((flags & TR_OPT_squeeze_reps) && last == coded &&
  55                         (pinvec[c] || poutvec[coded]))
  56                         continue;
  57                 poutput[out_index++] = last = coded;
  58         }
  59         /* NOTREACHED */
  60 }
  61
  62 static void map(unsigned char *string1, unsigned int string1_len,
  63                 unsigned char *string2, unsigned int string2_len)
  64 {
  65         char last = '0';
  66         unsigned int i, j;
  67
  68         for (j = 0, i = 0; i < string1_len; i++) {
  69                 if (string2_len <= j)
  70                         pvector[string1[i]] = last;
  71                 else
  72                         pvector[string1[i]] = last = string2[j++];
  73         }
  74 }
  75
  76 /* supported constructs:
  77  *   Ranges,  e.g.,  [0-9]  ==>  0123456789
  78  *   Escapes, e.g.,  \a     ==>  Control-G
  79  *       Character classes, e.g. [:upper:] ==> A ... Z
  80  */
  81 static unsigned int expand(const char *arg, char *buffer)
  82 {
  83         char *buffer_start = buffer;
  84         unsigned i; /* XXX: FIXME: use unsigned char? */
  85         unsigned char ac;
  86 #define CLO ":]"
  87         static const char * const classes[] = {
  88                 "alpha"CLO, "alnum"CLO, "digit"CLO, "lower"CLO, "upper"CLO, "space"CLO,
  89                 "blank"CLO, "punct"CLO, "cntrl"CLO, NULL
  90         };
  91 #define CLASS_invalid 0 /* we increment the retval */
  92 #define CLASS_alpha 1
  93 #define CLASS_alnum 2
  94 #define CLASS_digit 3
  95 #define CLASS_lower 4
  96 #define CLASS_upper 5
  97 #define CLASS_space 6
  98 #define CLASS_blank 7
  99 #define CLASS_punct 8
 100 #define CLASS_cntrl 9
 101 //#define CLASS_xdigit 10
 102 //#define CLASS_graph 11
 103 //#define CLASS_print 12
 104         while (*arg) {
 105                 if (*arg == '\\') {
 106                         arg++;
 107                         *buffer++ = bb_process_escape_sequence(&arg);
 108                 } else if (*(arg+1) == '-') {
 109                         ac = *(arg+2);
 110                         if (ac == 0) {
 111                                 *buffer++ = *arg++;
 112                                 continue;
 113                         }
 114                         i = *arg;
 115                         while (i <= ac)
 116                                 *buffer++ = i++;
 117                         arg += 3; /* Skip the assumed a-z */
 118                 } else if (*arg == '[') {
 119                         arg++;
 120                         i = *arg++;
 121                         if (ENABLE_FEATURE_TR_CLASSES && i == ':') {
 122                                 smalluint j;
 123                                 { /* not really pretty.. */
 124                                 char *tmp = xstrndup(arg, 7); // warning: xdigit needs 8, not 7
 125                                 j = index_in_str_array(classes, tmp) + 1;
 126                                 free(tmp);
 127                                 }
 128                                 if (j == CLASS_alnum || j == CLASS_digit) {
 129                                         for (i = '0'; i <= '9'; i++)
 130                                                 *buffer++ = i;
 131                                 }
 132                                 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) {
 133                                         for (i = 'A'; i <= 'Z'; i++)
 134                                                 *buffer++ = i;
 135                                 }
 136                                 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) {
 137                                         for (i = 'a'; i <= 'z'; i++)
 138                                                 *buffer++ = i;
 139                                 }
 140                                 if (j == CLASS_space || j == CLASS_blank) {
 141                                         *buffer++ = '\t';
 142                                         if (j == CLASS_space) {
 143                                                 *buffer++ = '\n';
 144                                                 *buffer++ = '\v';
 145                                                 *buffer++ = '\f';
 146                                                 *buffer++ = '\r';
 147                                         }
 148                                         *buffer++ = ' ';
 149                                 }
 150                                 if (j == CLASS_punct || j == CLASS_cntrl) {
 151                                         for (i = 0; i <= ASCII; i++)
 152                                                 if ((j == CLASS_punct &&
 153                                                          isprint(i) && (!isalnum(i)) && (!isspace(i))) ||
 154                                                         (j == CLASS_cntrl && iscntrl(i)))
 155                                                         *buffer++ = i;
 156                                 }
 157                                 if (j == CLASS_invalid) {
 158                                         *buffer++ = '[';
 159                                         *buffer++ = ':';
 160                                         continue;
 161                                 }
 162                                 break;
 163                         }
 164                         if (ENABLE_FEATURE_TR_EQUIV && i == '=') {
 165                                 *buffer++ = *arg;
 166                                 arg += 3;       /* Skip the closing =] */
 167                                 continue;
 168                         }
 169                         if (*arg++ != '-') {
 170                                 *buffer++ = '[';
 171                                 arg -= 2;
 172                                 continue;
 173                         }
 174                         ac = *arg++;
 175                         while (i <= ac)
 176                                 *buffer++ = i++;
 177                         arg++;  /* Skip the assumed ']' */
 178                 } else
 179                         *buffer++ = *arg++;
 180         }
 181         return (buffer - buffer_start);
 182 }
 183
 184 static int complement(char *buffer, int buffer_len)
 185 {
 186         short i, j, ix;
 187         char conv[ASCII + 2];
 188
 189         ix = 0;
 190         for (i = 0; i <= ASCII; i++) {
 191                 for (j = 0; j < buffer_len; j++)
 192                         if (buffer[j] == i)
 193                                 break;
 194                 if (j == buffer_len)
 195                         conv[ix++] = i & ASCII;
 196         }
 197         memcpy(buffer, conv, ix);
 198         return ix;
 199 }
 200
 201 int tr_main(int argc, char **argv);
 202 int tr_main(int argc, char **argv)
 203 {
 204         unsigned char *ptr;
 205         int output_length = 0, input_length;
 206         int idx = 1;
 207         int i;
 208         smalluint flags = 0;
 209         RESERVE_CONFIG_UBUFFER(output, BUFSIZ);
 210         RESERVE_CONFIG_BUFFER(vector, ASCII+1);
 211         RESERVE_CONFIG_BUFFER(invec,  ASCII+1);
 212         RESERVE_CONFIG_BUFFER(outvec, ASCII+1);
 213
 214         /* ... but make them available globally */
 215         poutput = output;
 216         pvector = vector;
 217         pinvec  = invec;
 218         poutvec = outvec;
 219
 220         if (argc > 1 && argv[idx][0] == '-') {
 221                 for (ptr = (unsigned char *) &argv[idx][1]; *ptr; ptr++) {
 222                         if (*ptr == 'c')
 223                                 flags |= TR_OPT_complement;
 224                         else if (*ptr == 'd')
 225                                 flags |= TR_OPT_delete;
 226                         else if (*ptr == 's')
 227                                 flags |= TR_OPT_squeeze_reps;
 228                         else
 229                                 bb_show_usage();
 230                 }
 231                 idx++;
 232         }
 233         for (i = 0; i <= ASCII; i++) {
 234                 vector[i] = i;
 235                 invec[i] = outvec[i] = FALSE;
 236         }
 237
 238         if (argv[idx] != NULL) {
 239                 input_length = expand(argv[idx++], bb_common_bufsiz1);
 240                 if (flags & TR_OPT_complement)
 241                         input_length = complement(bb_common_bufsiz1, input_length);
 242                 if (argv[idx] != NULL) {
 243                         if (*argv[idx] == '\0')
 244                                 bb_error_msg_and_die("STRING2 cannot be empty");
 245                         output_length = expand(argv[idx], output);
 246                         map(bb_common_bufsiz1, input_length, output, output_length);
 247                 }
 248                 for (i = 0; i < input_length; i++)
 249                         invec[(unsigned char)bb_common_bufsiz1[i]] = TRUE;
 250                 for (i = 0; i < output_length; i++)
 251                         outvec[output[i]] = TRUE;
 252         }
 253         convert(flags);
 254         return EXIT_SUCCESS;
 255 }