1 /* vi: set sw=4 ts=4: */
3 * Mini tr implementation for busybox
5 ** Copyright (c) 1987,1997, Prentice Hall All rights reserved.
7 * The name of Prentice Hall may not be used to endorse or promote
8 * products derived from this software without specific prior
11 * Copyright (c) Michiel Huisjes
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.
16 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
18 /* http://www.opengroup.org/onlinepubs/009695399/utilities/tr.html
19 * TODO: xdigit, graph, print
25 #define TR_OPT_complement (1<<0)
26 #define TR_OPT_delete (1<<1)
27 #define TR_OPT_squeeze_reps (1<<2)
29 static void map(char *pvector,
30 unsigned char *string1, unsigned int string1_len,
31 unsigned char *string2, unsigned int string2_len)
36 for (j = 0, i = 0; i < string1_len; i++) {
38 pvector[string1[i]] = last;
40 pvector[string1[i]] = last = string2[j++];
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
49 static unsigned int expand(const char *arg, char *buffer)
51 char *buffer_start = buffer;
52 unsigned i; /* XXX: FIXME: use unsigned char? */
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 */
68 //#define CLASS_xdigit 10
69 //#define CLASS_graph 11
70 //#define CLASS_print 12
74 *buffer++ = bb_process_escape_sequence(&arg);
75 } else if (*(arg+1) == '-') {
84 arg += 3; /* Skip the assumed a-z */
85 } else if (*arg == '[') {
88 if (ENABLE_FEATURE_TR_CLASSES && i == ':') {
90 { /* not really pretty.. */
91 char *tmp = xstrndup(arg, 7); // warning: xdigit needs 8, not 7
92 j = index_in_strings(classes, tmp) + 1;
95 if (j == CLASS_alnum || j == CLASS_digit) {
96 for (i = '0'; i <= '9'; i++)
99 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) {
100 for (i = 'A'; i <= 'Z'; i++)
103 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) {
104 for (i = 'a'; i <= 'z'; i++)
107 if (j == CLASS_space || j == CLASS_blank) {
109 if (j == CLASS_space) {
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)))
124 if (j == CLASS_invalid) {
131 if (ENABLE_FEATURE_TR_EQUIV && i == '=') {
133 arg += 3; /* Skip the closing =] */
144 arg++; /* Skip the assumed ']' */
148 return (buffer - buffer_start);
151 static int complement(char *buffer, int buffer_len)
154 char conv[ASCII + 2];
157 for (i = 0; i <= ASCII; i++) {
158 for (j = 0; j < buffer_len; j++)
162 conv[ix++] = i & ASCII;
164 memcpy(buffer, conv, ix);
168 int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
169 int tr_main(int argc, char **argv)
172 int output_length = 0, input_length;
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);
182 if (argc > 1 && argv[idx][0] == '-') {
183 for (ptr = (unsigned char *) &argv[idx][1]; *ptr; ptr++) {
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;
195 for (i = 0; i <= ASCII; i++) {
197 invec[i] = outvec[i] = FALSE;
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);
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;
218 /* If we're out of input, flush output and read more input. */
219 if (in_index == read_chars) {
221 xwrite(STDOUT_FILENO, (char *)output, out_index);
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);
232 c = tr_buf[in_index++];
234 if ((flags & TR_OPT_delete) && invec[c])
236 if ((flags & TR_OPT_squeeze_reps) && last == coded &&
237 (invec[c] || outvec[coded]))
239 output[out_index++] = last = coded;