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 static void map(char *pvector,
  26                 unsigned char *string1, unsigned string1_len,
  27                 unsigned char *string2, unsigned string2_len)
  28 {
  29         char last = '0';
  30         unsigned i, j;
  31
  32         for (j = 0, i = 0; i < string1_len; i++) {
  33                 if (string2_len <= j)
  34                         pvector[string1[i]] = last;
  35                 else
  36                         pvector[string1[i]] = last = string2[j++];
  37         }
  38 }
  39
  40 /* supported constructs:
  41  *   Ranges,  e.g.,  0-9   ==>  0123456789
  42  *   Escapes, e.g.,  \a    ==>  Control-G
  43  *   Character classes, e.g. [:upper:] ==> A...Z
  44  *   Equiv classess, e.g. [=A=] ==> A   (hmmmmmmm?)
  45  */
  46 static unsigned expand(const char *arg, char *buffer)
  47 {
  48         char *buffer_start = buffer;
  49         unsigned i; /* can't be unsigned char: must be able to hold 256 */
  50         unsigned char ac;
  51
  52         while (*arg) {
  53                 if (*arg == '\\') {
  54                         arg++;
  55                         *buffer++ = bb_process_escape_sequence(&arg);
  56                         continue;
  57                 }
  58                 if (arg[1] == '-') { /* "0-9..." */
  59                         ac = arg[2];
  60                         if (ac == '\0') { /* "0-": copy verbatim */
  61                                 *buffer++ = *arg++; /* copy '0' */
  62                                 continue; /* next iter will copy '-' and stop */
  63                         }
  64                         i = *arg;
  65                         while (i <= ac) /* ok: i is unsigned _int_ */
  66                                 *buffer++ = i++;
  67                         arg += 3; /* skip 0-9 */
  68                         continue;
  69                 }
  70                 if (*arg == '[') { /* "[xyz..." */
  71                         arg++;
  72                         i = *arg++;
  73                         /* "[xyz...", i=x, arg points to y */
  74                         if (ENABLE_FEATURE_TR_CLASSES && i == ':') {
  75 #define CLO ":]\0"
  76                                 static const char classes[] ALIGN1 =
  77                                         "alpha"CLO "alnum"CLO "digit"CLO
  78                                         "lower"CLO "upper"CLO "space"CLO
  79                                         "blank"CLO "punct"CLO "cntrl"CLO
  80                                         "xdigit"CLO;
  81 #define CLASS_invalid 0 /* we increment the retval */
  82 #define CLASS_alpha 1
  83 #define CLASS_alnum 2
  84 #define CLASS_digit 3
  85 #define CLASS_lower 4
  86 #define CLASS_upper 5
  87 #define CLASS_space 6
  88 #define CLASS_blank 7
  89 #define CLASS_punct 8
  90 #define CLASS_cntrl 9
  91 #define CLASS_xdigit 10
  92 //#define CLASS_graph 11
  93 //#define CLASS_print 12
  94                                 smalluint j;
  95                                 {
  96                                         /* xdigit needs 8, not 7 */
  97                                         char *tmp = xstrndup(arg, 7 + (arg[0]=='x'));
  98                                         j = index_in_strings(classes, tmp) + 1;
  99                                         free(tmp);
 100                                 }
 101                                 if (j == CLASS_alnum || j == CLASS_digit || j == CLASS_xdigit) {
 102                                         for (i = '0'; i <= '9'; i++)
 103                                                 *buffer++ = i;
 104                                 }
 105                                 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) {
 106                                         for (i = 'A'; i <= 'Z'; i++)
 107                                                 *buffer++ = i;
 108                                 }
 109                                 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) {
 110                                         for (i = 'a'; i <= 'z'; i++)
 111                                                 *buffer++ = i;
 112                                 }
 113                                 if (j == CLASS_space || j == CLASS_blank) {
 114                                         *buffer++ = '\t';
 115                                         if (j == CLASS_space) {
 116                                                 *buffer++ = '\n';
 117                                                 *buffer++ = '\v';
 118                                                 *buffer++ = '\f';
 119                                                 *buffer++ = '\r';
 120                                         }
 121                                         *buffer++ = ' ';
 122                                 }
 123                                 if (j == CLASS_punct || j == CLASS_cntrl) {
 124                                         for (i = '\0'; i <= ASCII; i++)
 125                                                 if ((j == CLASS_punct && isprint(i) && !isalnum(i) && !isspace(i))
 126                                                  || (j == CLASS_cntrl && iscntrl(i)))
 127                                                         *buffer++ = i;
 128                                 }
 129                                 if (j == CLASS_xdigit) {
 130                                         for (i = 'A'; i <= 'F'; i++) {
 131                                                 *buffer++ = i;
 132                                                 *buffer++ = i | 0x20;
 133                                         }
 134                                 }
 135                                 if (j == CLASS_invalid) {
 136                                         *buffer++ = '[';
 137                                         *buffer++ = ':';
 138                                         continue;
 139                                 }
 140                                 break;
 141                         }
 142                         /* "[xyz...", i=x, arg points to y */
 143                         if (ENABLE_FEATURE_TR_EQUIV && i == '=') { /* [=CHAR=] */
 144                                 *buffer++ = *arg; /* copy CHAR */
 145                                 if (!*arg || arg[1] != '=' || arg[2] != ']')
 146                                         bb_show_usage();
 147                                 arg += 3;       /* skip CHAR=] */
 148                                 continue;
 149                         }
 150                         /* The rest of [xyz... cases is treated as normal
 151                          * string, '[' has no special meaning here:
 152                          * tr "[a-z]" "[A-Z]" can be written as tr "a-z" "A-Z",
 153                          * also try tr "[a-z]" "_A-Z+" and you'll see that
 154                          * [] is not special here.
 155                          */
 156                         *buffer++ = '[';
 157                         arg--; /* points to x */
 158                         continue;
 159                 }
 160                 *buffer++ = *arg++;
 161         }
 162         return (buffer - buffer_start);
 163 }
 164
 165 static int complement(char *buffer, int buffer_len)
 166 {
 167         int ch, j, len;
 168         char conv[ASCII + 2];
 169
 170         len = 0;
 171         for (ch = '\0'; ch <= ASCII; ch++) {
 172                 for (j = 0; j < buffer_len; j++)
 173                         if (buffer[j] == ch)
 174                                 goto next_ch;
 175                 /* Didn't find it */
 176                 conv[len++] = (char) ch;
 177  next_ch:
 178                 continue;
 179         }
 180         memcpy(buffer, conv, len);
 181         return len;
 182 }
 183
 184 int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 185 int tr_main(int argc UNUSED_PARAM, char **argv)
 186 {
 187         int i;
 188         smalluint flags;
 189         ssize_t read_chars;
 190         size_t in_index, out_index;
 191         unsigned last = UCHAR_MAX + 1; /* not equal to any char */
 192         unsigned char coded, c;
 193         unsigned char *output = xmalloc(BUFSIZ);
 194         char *vector = xzalloc((ASCII+1) * 3);
 195         char *invec  = vector + (ASCII+1);
 196         char *outvec = vector + (ASCII+1) * 2;
 197
 198 #define TR_OPT_complement       (1 << 0)
 199 #define TR_OPT_delete           (1 << 1)
 200 #define TR_OPT_squeeze_reps     (1 << 2)
 201
 202         flags = getopt32(argv, "+cds"); /* '+': stop at first non-option */
 203         argv += optind;
 204
 205         for (i = 0; i <= ASCII; i++) {
 206                 vector[i] = i;
 207                 /*invec[i] = outvec[i] = FALSE; - done by xzalloc */
 208         }
 209
 210 #define tr_buf bb_common_bufsiz1
 211         if (*argv != NULL) {
 212                 int output_length = 0;
 213                 int input_length;
 214
 215                 input_length = expand(*argv++, tr_buf);
 216                 if (flags & TR_OPT_complement)
 217                         input_length = complement(tr_buf, input_length);
 218                 if (*argv) {
 219                         if (argv[0][0] == '\0')
 220                                 bb_error_msg_and_die("STRING2 cannot be empty");
 221                         output_length = expand(*argv, (char *)output);
 222                         map(vector, (unsigned char *)tr_buf, input_length, output, output_length);
 223                 }
 224                 for (i = 0; i < input_length; i++)
 225                         invec[(unsigned char)tr_buf[i]] = TRUE;
 226                 for (i = 0; i < output_length; i++)
 227                         outvec[output[i]] = TRUE;
 228         }
 229
 230         goto start_from;
 231
 232         for (;;) {
 233                 /* If we're out of input, flush output and read more input. */
 234                 if ((ssize_t)in_index == read_chars) {
 235                         if (out_index) {
 236                                 xwrite(STDOUT_FILENO, (char *)output, out_index);
 237  start_from:
 238                                 out_index = 0;
 239                         }
 240                         read_chars = safe_read(STDIN_FILENO, tr_buf, BUFSIZ);
 241                         if (read_chars <= 0) {
 242                                 if (read_chars < 0)
 243                                         bb_perror_msg_and_die(bb_msg_read_error);
 244                                 break;
 245                         }
 246                         in_index = 0;
 247                 }
 248                 c = tr_buf[in_index++];
 249                 if ((flags & TR_OPT_delete) && invec[c])
 250                         continue;
 251                 coded = vector[c];
 252                 if ((flags & TR_OPT_squeeze_reps) && last == coded
 253                  && (invec[c] || outvec[coded]))
 254                         continue;
 255                 output[out_index++] = last = coded;
 256         }
 257
 258         return EXIT_SUCCESS;
 259 }