| 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 |
|
|---|
| 29 | static void map(char *pvector,
|
|---|
| 30 | unsigned char *string1, unsigned int string1_len,
|
|---|
| 31 | unsigned char *string2, unsigned int string2_len)
|
|---|
| 32 | {
|
|---|
| 33 | char last = '0';
|
|---|
| 34 | unsigned int i, j;
|
|---|
| 35 |
|
|---|
| 36 | for (j = 0, i = 0; i < string1_len; i++) {
|
|---|
| 37 | if (string2_len <= j)
|
|---|
| 38 | pvector[string1[i]] = last;
|
|---|
| 39 | else
|
|---|
| 40 | pvector[string1[i]] = last = string2[j++];
|
|---|
| 41 | }
|
|---|
| 42 | }
|
|---|
| 43 |
|
|---|
| 44 | /* supported constructs:
|
|---|
| 45 | * Ranges, e.g., [0-9] ==> 0123456789
|
|---|
| 46 | * Escapes, e.g., \a ==> Control-G
|
|---|
| 47 | * Character classes, e.g. [:upper:] ==> A ... Z
|
|---|
| 48 | */
|
|---|
| 49 | static unsigned int expand(const char *arg, char *buffer)
|
|---|
| 50 | {
|
|---|
| 51 | char *buffer_start = buffer;
|
|---|
| 52 | unsigned i; /* XXX: FIXME: use unsigned char? */
|
|---|
| 53 | unsigned char ac;
|
|---|
| 54 | #define CLO ":]\0"
|
|---|
| 55 | static const char classes[] ALIGN1 =
|
|---|
| 56 | "alpha"CLO "alnum"CLO "digit"CLO "lower"CLO "upper"CLO "space"CLO
|
|---|
| 57 | "blank"CLO "punct"CLO "cntrl"CLO;
|
|---|
| 58 | #define CLASS_invalid 0 /* we increment the retval */
|
|---|
| 59 | #define CLASS_alpha 1
|
|---|
| 60 | #define CLASS_alnum 2
|
|---|
| 61 | #define CLASS_digit 3
|
|---|
| 62 | #define CLASS_lower 4
|
|---|
| 63 | #define CLASS_upper 5
|
|---|
| 64 | #define CLASS_space 6
|
|---|
| 65 | #define CLASS_blank 7
|
|---|
| 66 | #define CLASS_punct 8
|
|---|
| 67 | #define CLASS_cntrl 9
|
|---|
| 68 | //#define CLASS_xdigit 10
|
|---|
| 69 | //#define CLASS_graph 11
|
|---|
| 70 | //#define CLASS_print 12
|
|---|
| 71 | while (*arg) {
|
|---|
| 72 | if (*arg == '\\') {
|
|---|
| 73 | arg++;
|
|---|
| 74 | *buffer++ = bb_process_escape_sequence(&arg);
|
|---|
| 75 | } else if (*(arg+1) == '-') {
|
|---|
| 76 | ac = *(arg+2);
|
|---|
| 77 | if (ac == 0) {
|
|---|
| 78 | *buffer++ = *arg++;
|
|---|
| 79 | continue;
|
|---|
| 80 | }
|
|---|
| 81 | i = *arg;
|
|---|
| 82 | while (i <= ac)
|
|---|
| 83 | *buffer++ = i++;
|
|---|
| 84 | arg += 3; /* Skip the assumed a-z */
|
|---|
| 85 | } else if (*arg == '[') {
|
|---|
| 86 | arg++;
|
|---|
| 87 | i = *arg++;
|
|---|
| 88 | if (ENABLE_FEATURE_TR_CLASSES && i == ':') {
|
|---|
| 89 | smalluint j;
|
|---|
| 90 | { /* not really pretty.. */
|
|---|
| 91 | char *tmp = xstrndup(arg, 7); // warning: xdigit needs 8, not 7
|
|---|
| 92 | j = index_in_strings(classes, tmp) + 1;
|
|---|
| 93 | free(tmp);
|
|---|
| 94 | }
|
|---|
| 95 | if (j == CLASS_alnum || j == CLASS_digit) {
|
|---|
| 96 | for (i = '0'; i <= '9'; i++)
|
|---|
| 97 | *buffer++ = i;
|
|---|
| 98 | }
|
|---|
| 99 | if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) {
|
|---|
| 100 | for (i = 'A'; i <= 'Z'; i++)
|
|---|
| 101 | *buffer++ = i;
|
|---|
| 102 | }
|
|---|
| 103 | if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) {
|
|---|
| 104 | for (i = 'a'; i <= 'z'; i++)
|
|---|
| 105 | *buffer++ = i;
|
|---|
| 106 | }
|
|---|
| 107 | if (j == CLASS_space || j == CLASS_blank) {
|
|---|
| 108 | *buffer++ = '\t';
|
|---|
| 109 | if (j == CLASS_space) {
|
|---|
| 110 | *buffer++ = '\n';
|
|---|
| 111 | *buffer++ = '\v';
|
|---|
| 112 | *buffer++ = '\f';
|
|---|
| 113 | *buffer++ = '\r';
|
|---|
| 114 | }
|
|---|
| 115 | *buffer++ = ' ';
|
|---|
| 116 | }
|
|---|
| 117 | if (j == CLASS_punct || j == CLASS_cntrl) {
|
|---|
| 118 | for (i = 0; i <= ASCII; i++)
|
|---|
| 119 | if ((j == CLASS_punct &&
|
|---|
| 120 | isprint(i) && (!isalnum(i)) && (!isspace(i))) ||
|
|---|
| 121 | (j == CLASS_cntrl && iscntrl(i)))
|
|---|
| 122 | *buffer++ = i;
|
|---|
| 123 | }
|
|---|
| 124 | if (j == CLASS_invalid) {
|
|---|
| 125 | *buffer++ = '[';
|
|---|
| 126 | *buffer++ = ':';
|
|---|
| 127 | continue;
|
|---|
| 128 | }
|
|---|
| 129 | break;
|
|---|
| 130 | }
|
|---|
| 131 | if (ENABLE_FEATURE_TR_EQUIV && i == '=') {
|
|---|
| 132 | *buffer++ = *arg;
|
|---|
| 133 | arg += 3; /* Skip the closing =] */
|
|---|
| 134 | continue;
|
|---|
| 135 | }
|
|---|
| 136 | if (*arg++ != '-') {
|
|---|
| 137 | *buffer++ = '[';
|
|---|
| 138 | arg -= 2;
|
|---|
| 139 | continue;
|
|---|
| 140 | }
|
|---|
| 141 | ac = *arg++;
|
|---|
| 142 | while (i <= ac)
|
|---|
| 143 | *buffer++ = i++;
|
|---|
| 144 | arg++; /* Skip the assumed ']' */
|
|---|
| 145 | } else
|
|---|
| 146 | *buffer++ = *arg++;
|
|---|
| 147 | }
|
|---|
| 148 | return (buffer - buffer_start);
|
|---|
| 149 | }
|
|---|
| 150 |
|
|---|
| 151 | static int complement(char *buffer, int buffer_len)
|
|---|
| 152 | {
|
|---|
| 153 | int i, j, ix;
|
|---|
| 154 | char conv[ASCII + 2];
|
|---|
| 155 |
|
|---|
| 156 | ix = 0;
|
|---|
| 157 | for (i = 0; i <= ASCII; i++) {
|
|---|
| 158 | for (j = 0; j < buffer_len; j++)
|
|---|
| 159 | if (buffer[j] == i)
|
|---|
| 160 | break;
|
|---|
| 161 | if (j == buffer_len)
|
|---|
| 162 | conv[ix++] = i & ASCII;
|
|---|
| 163 | }
|
|---|
| 164 | memcpy(buffer, conv, ix);
|
|---|
| 165 | return ix;
|
|---|
| 166 | }
|
|---|
| 167 |
|
|---|
| 168 | int tr_main(int argc, char **argv);
|
|---|
| 169 | int tr_main(int argc, char **argv)
|
|---|
| 170 | {
|
|---|
| 171 | unsigned char *ptr;
|
|---|
| 172 | int output_length = 0, input_length;
|
|---|
| 173 | int idx = 1;
|
|---|
| 174 | int i;
|
|---|
| 175 | smalluint flags = 0;
|
|---|
| 176 | size_t read_chars = 0, in_index = 0, out_index = 0, c, coded, last = -1;
|
|---|
| 177 | RESERVE_CONFIG_UBUFFER(output, BUFSIZ);
|
|---|
| 178 | RESERVE_CONFIG_BUFFER(vector, ASCII+1);
|
|---|
| 179 | RESERVE_CONFIG_BUFFER(invec, ASCII+1);
|
|---|
| 180 | RESERVE_CONFIG_BUFFER(outvec, ASCII+1);
|
|---|
| 181 |
|
|---|
| 182 | if (argc > 1 && argv[idx][0] == '-') {
|
|---|
| 183 | for (ptr = (unsigned char *) &argv[idx][1]; *ptr; ptr++) {
|
|---|
| 184 | if (*ptr == 'c')
|
|---|
| 185 | flags |= TR_OPT_complement;
|
|---|
| 186 | else if (*ptr == 'd')
|
|---|
| 187 | flags |= TR_OPT_delete;
|
|---|
| 188 | else if (*ptr == 's')
|
|---|
| 189 | flags |= TR_OPT_squeeze_reps;
|
|---|
| 190 | else
|
|---|
| 191 | bb_show_usage();
|
|---|
| 192 | }
|
|---|
| 193 | idx++;
|
|---|
| 194 | }
|
|---|
| 195 | for (i = 0; i <= ASCII; i++) {
|
|---|
| 196 | vector[i] = i;
|
|---|
| 197 | invec[i] = outvec[i] = FALSE;
|
|---|
| 198 | }
|
|---|
| 199 |
|
|---|
| 200 | #define tr_buf bb_common_bufsiz1
|
|---|
| 201 | if (argv[idx] != NULL) {
|
|---|
| 202 | input_length = expand(argv[idx++], tr_buf);
|
|---|
| 203 | if (flags & TR_OPT_complement)
|
|---|
| 204 | input_length = complement(tr_buf, input_length);
|
|---|
| 205 | if (argv[idx] != NULL) {
|
|---|
| 206 | if (*argv[idx] == '\0')
|
|---|
| 207 | bb_error_msg_and_die("STRING2 cannot be empty");
|
|---|
| 208 | output_length = expand(argv[idx], output);
|
|---|
| 209 | map(vector, tr_buf, input_length, output, output_length);
|
|---|
| 210 | }
|
|---|
| 211 | for (i = 0; i < input_length; i++)
|
|---|
| 212 | invec[(unsigned char)tr_buf[i]] = TRUE;
|
|---|
| 213 | for (i = 0; i < output_length; i++)
|
|---|
| 214 | outvec[output[i]] = TRUE;
|
|---|
| 215 | }
|
|---|
| 216 |
|
|---|
| 217 | for (;;) {
|
|---|
| 218 | /* If we're out of input, flush output and read more input. */
|
|---|
| 219 | if (in_index == read_chars) {
|
|---|
| 220 | if (out_index) {
|
|---|
| 221 | xwrite(STDOUT_FILENO, (char *)output, out_index);
|
|---|
| 222 | out_index = 0;
|
|---|
| 223 | }
|
|---|
| 224 | read_chars = read(STDIN_FILENO, tr_buf, BUFSIZ);
|
|---|
| 225 | if (read_chars <= 0) {
|
|---|
| 226 | if (write(STDOUT_FILENO, (char *)output, out_index) != out_index)
|
|---|
| 227 | bb_perror_msg(bb_msg_write_error);
|
|---|
| 228 | exit(EXIT_SUCCESS);
|
|---|
| 229 | }
|
|---|
| 230 | in_index = 0;
|
|---|
| 231 | }
|
|---|
| 232 | c = tr_buf[in_index++];
|
|---|
| 233 | coded = vector[c];
|
|---|
| 234 | if ((flags & TR_OPT_delete) && invec[c])
|
|---|
| 235 | continue;
|
|---|
| 236 | if ((flags & TR_OPT_squeeze_reps) && last == coded &&
|
|---|
| 237 | (invec[c] || outvec[coded]))
|
|---|
| 238 | continue;
|
|---|
| 239 | output[out_index++] = last = coded;
|
|---|
| 240 | }
|
|---|
| 241 | /* NOTREACHED */
|
|---|
| 242 | return EXIT_SUCCESS;
|
|---|
| 243 | }
|
|---|
