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