source: MondoRescue/branches/3.3/mindi-busybox/coreutils/tr.c@ 3621

Last change on this file since 3621 was 3621, checked in by Bruno Cornec, 7 years ago

New 3?3 banch for incorporation of latest busybox 1.25. Changing minor version to handle potential incompatibilities.
File size: 9.5 KB
Line 
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 source tree.
17 */
18/* http://www.opengroup.org/onlinepubs/009695399/utilities/tr.html
19 * TODO: graph, print
20 */
21
22//kbuild:lib-$(CONFIG_TR) += tr.o
23
24//config:config TR
25//config: bool "tr"
26//config: default y
27//config: help
28//config: tr is used to squeeze, and/or delete characters from standard
29//config: input, writing to standard output.
30//config:
31//config:config FEATURE_TR_CLASSES
32//config: bool "Enable character classes (such as [:upper:])"
33//config: default y
34//config: depends on TR
35//config: help
36//config: Enable character classes, enabling commands such as:
37//config: tr [:upper:] [:lower:] to convert input into lowercase.
38//config:
39//config:config FEATURE_TR_EQUIV
40//config: bool "Enable equivalence classes"
41//config: default y
42//config: depends on TR
43//config: help
44//config: Enable equivalence classes, which essentially add the enclosed
45//config: character to the current set. For instance, tr [=a=] xyz would
46//config: replace all instances of 'a' with 'xyz'. This option is mainly
47//config: useful for cases when no other way of expressing a character
48//config: is possible.
49
50//usage:#define tr_trivial_usage
51//usage: "[-cds] STRING1 [STRING2]"
52//usage:#define tr_full_usage "\n\n"
53//usage: "Translate, squeeze, or delete characters from stdin, writing to stdout\n"
54//usage: "\n -c Take complement of STRING1"
55//usage: "\n -d Delete input characters coded STRING1"
56//usage: "\n -s Squeeze multiple output characters of STRING2 into one character"
57//usage:
58//usage:#define tr_example_usage
59//usage: "$ echo \"gdkkn vnqkc\" | tr [a-y] [b-z]\n"
60//usage: "hello world\n"
61
62#include "libbb.h"
63
64enum {
65 ASCII = 256,
66 /* string buffer needs to be at least as big as the whole "alphabet".
67 * BUFSIZ == ASCII is ok, but we will realloc in expand
68 * even for smallest patterns, let's avoid that by using *2:
69 */
70 TR_BUFSIZ = (BUFSIZ > ASCII*2) ? BUFSIZ : ASCII*2,
71};
72
73static void map(char *pvector,
74 char *string1, unsigned string1_len,
75 char *string2, unsigned string2_len)
76{
77 char last = '0';
78 unsigned i, j;
79
80 for (j = 0, i = 0; i < string1_len; i++) {
81 if (string2_len <= j)
82 pvector[(unsigned char)(string1[i])] = last;
83 else
84 pvector[(unsigned char)(string1[i])] = last = string2[j++];
85 }
86}
87
88/* supported constructs:
89 * Ranges, e.g., 0-9 ==> 0123456789
90 * Escapes, e.g., \a ==> Control-G
91 * Character classes, e.g. [:upper:] ==> A...Z
92 * Equiv classess, e.g. [=A=] ==> A (hmmmmmmm?)
93 * not supported:
94 * [x*N] - repeat char x N times
95 * [x*] - repeat char x until it fills STRING2:
96 * # echo qwe123 | /usr/bin/tr 123456789 '[d]'
97 * qwe[d]
98 * # echo qwe123 | /usr/bin/tr 123456789 '[d*]'
99 * qweddd
100 */
101static unsigned expand(char *arg, char **buffer_p)
102{
103 char *buffer = *buffer_p;
104 unsigned pos = 0;
105 unsigned size = TR_BUFSIZ;
106 unsigned i; /* can't be unsigned char: must be able to hold 256 */
107 unsigned char ac;
108
109 while (*arg) {
110 if (pos + ASCII > size) {
111 size += ASCII;
112 *buffer_p = buffer = xrealloc(buffer, size);
113 }
114 if (*arg == '\\') {
115 const char *z;
116 arg++;
117 z = arg;
118 ac = bb_process_escape_sequence(&z);
119 arg = (char *)z;
120 arg--;
121 *arg = ac;
122 /*
123 * fall through, there may be a range.
124 * If not, current char will be treated anyway.
125 */
126 }
127 if (arg[1] == '-') { /* "0-9..." */
128 ac = arg[2];
129 if (ac == '\0') { /* "0-": copy verbatim */
130 buffer[pos++] = *arg++; /* copy '0' */
131 continue; /* next iter will copy '-' and stop */
132 }
133 i = (unsigned char) *arg;
134 arg += 3; /* skip 0-9 or 0-\ */
135 if (ac == '\\') {
136 const char *z;
137 z = arg;
138 ac = bb_process_escape_sequence(&z);
139 arg = (char *)z;
140 }
141 while (i <= ac) /* ok: i is unsigned _int_ */
142 buffer[pos++] = i++;
143 continue;
144 }
145 if ((ENABLE_FEATURE_TR_CLASSES || ENABLE_FEATURE_TR_EQUIV)
146 && *arg == '['
147 ) {
148 arg++;
149 i = (unsigned char) *arg++;
150 /* "[xyz...". i=x, arg points to y */
151 if (ENABLE_FEATURE_TR_CLASSES && i == ':') { /* [:class:] */
152#define CLO ":]\0"
153 static const char classes[] ALIGN1 =
154 "alpha"CLO "alnum"CLO "digit"CLO
155 "lower"CLO "upper"CLO "space"CLO
156 "blank"CLO "punct"CLO "cntrl"CLO
157 "xdigit"CLO;
158 enum {
159 CLASS_invalid = 0, /* we increment the retval */
160 CLASS_alpha = 1,
161 CLASS_alnum = 2,
162 CLASS_digit = 3,
163 CLASS_lower = 4,
164 CLASS_upper = 5,
165 CLASS_space = 6,
166 CLASS_blank = 7,
167 CLASS_punct = 8,
168 CLASS_cntrl = 9,
169 CLASS_xdigit = 10,
170 //CLASS_graph = 11,
171 //CLASS_print = 12,
172 };
173 smalluint j;
174 char *tmp;
175
176 /* xdigit needs 8, not 7 */
177 i = 7 + (arg[0] == 'x');
178 tmp = xstrndup(arg, i);
179 j = index_in_strings(classes, tmp) + 1;
180 free(tmp);
181
182 if (j == CLASS_invalid)
183 goto skip_bracket;
184
185 arg += i;
186 if (j == CLASS_alnum || j == CLASS_digit || j == CLASS_xdigit) {
187 for (i = '0'; i <= '9'; i++)
188 buffer[pos++] = i;
189 }
190 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) {
191 for (i = 'A'; i <= 'Z'; i++)
192 buffer[pos++] = i;
193 }
194 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) {
195 for (i = 'a'; i <= 'z'; i++)
196 buffer[pos++] = i;
197 }
198 if (j == CLASS_space || j == CLASS_blank) {
199 buffer[pos++] = '\t';
200 if (j == CLASS_space) {
201 buffer[pos++] = '\n';
202 buffer[pos++] = '\v';
203 buffer[pos++] = '\f';
204 buffer[pos++] = '\r';
205 }
206 buffer[pos++] = ' ';
207 }
208 if (j == CLASS_punct || j == CLASS_cntrl) {
209 for (i = '\0'; i < ASCII; i++) {
210 if ((j == CLASS_punct && isprint_asciionly(i) && !isalnum(i) && !isspace(i))
211 || (j == CLASS_cntrl && iscntrl(i))
212 ) {
213 buffer[pos++] = i;
214 }
215 }
216 }
217 if (j == CLASS_xdigit) {
218 for (i = 'A'; i <= 'F'; i++) {
219 buffer[pos + 6] = i | 0x20;
220 buffer[pos++] = i;
221 }
222 pos += 6;
223 }
224 continue;
225 }
226 /* "[xyz...", i=x, arg points to y */
227 if (ENABLE_FEATURE_TR_EQUIV && i == '=') { /* [=CHAR=] */
228 buffer[pos++] = *arg; /* copy CHAR */
229 if (!arg[0] || arg[1] != '=' || arg[2] != ']')
230 bb_show_usage();
231 arg += 3; /* skip CHAR=] */
232 continue;
233 }
234 /* The rest of "[xyz..." cases is treated as normal
235 * string, "[" has no special meaning here:
236 * tr "[a-z]" "[A-Z]" can be written as tr "a-z" "A-Z",
237 * also try tr "[a-z]" "_A-Z+" and you'll see that
238 * [] is not special here.
239 */
240 skip_bracket:
241 arg -= 2; /* points to "[" in "[xyz..." */
242 }
243 buffer[pos++] = *arg++;
244 }
245 return pos;
246}
247
248/* NB: buffer is guaranteed to be at least TR_BUFSIZE
249 * (which is >= ASCII) big.
250 */
251static int complement(char *buffer, int buffer_len)
252{
253 int len;
254 char conv[ASCII];
255 unsigned char ch;
256
257 len = 0;
258 ch = '\0';
259 while (1) {
260 if (memchr(buffer, ch, buffer_len) == NULL)
261 conv[len++] = ch;
262 if (++ch == '\0')
263 break;
264 }
265 memcpy(buffer, conv, len);
266 return len;
267}
268
269int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
270int tr_main(int argc UNUSED_PARAM, char **argv)
271{
272 int i;
273 smalluint opts;
274 ssize_t read_chars;
275 size_t in_index, out_index;
276 unsigned last = UCHAR_MAX + 1; /* not equal to any char */
277 unsigned char coded, c;
278 char *str1 = xmalloc(TR_BUFSIZ);
279 char *str2 = xmalloc(TR_BUFSIZ);
280 int str2_length;
281 int str1_length;
282 char *vector = xzalloc(ASCII * 3);
283 char *invec = vector + ASCII;
284 char *outvec = vector + ASCII * 2;
285
286#define TR_OPT_complement (3 << 0)
287#define TR_OPT_delete (1 << 2)
288#define TR_OPT_squeeze_reps (1 << 3)
289
290 for (i = 0; i < ASCII; i++) {
291 vector[i] = i;
292 /*invec[i] = outvec[i] = FALSE; - done by xzalloc */
293 }
294
295 /* -C/-c difference is that -C complements "characters",
296 * and -c complements "values" (binary bytes I guess).
297 * In POSIX locale, these are the same.
298 */
299
300 opt_complementary = "-1";
301 opts = getopt32(argv, "+Ccds"); /* '+': stop at first non-option */
302 argv += optind;
303
304 str1_length = expand(*argv++, &str1);
305 str2_length = 0;
306 if (opts & TR_OPT_complement)
307 str1_length = complement(str1, str1_length);
308 if (*argv) {
309 if (argv[0][0] == '\0')
310 bb_error_msg_and_die("STRING2 cannot be empty");
311 str2_length = expand(*argv, &str2);
312 map(vector, str1, str1_length,
313 str2, str2_length);
314 }
315 for (i = 0; i < str1_length; i++)
316 invec[(unsigned char)(str1[i])] = TRUE;
317 for (i = 0; i < str2_length; i++)
318 outvec[(unsigned char)(str2[i])] = TRUE;
319
320 goto start_from;
321
322 /* In this loop, str1 space is reused as input buffer,
323 * str2 - as output one. */
324 for (;;) {
325 /* If we're out of input, flush output and read more input. */
326 if ((ssize_t)in_index == read_chars) {
327 if (out_index) {
328 xwrite(STDOUT_FILENO, str2, out_index);
329 start_from:
330 out_index = 0;
331 }
332 read_chars = safe_read(STDIN_FILENO, str1, TR_BUFSIZ);
333 if (read_chars <= 0) {
334 if (read_chars < 0)
335 bb_perror_msg_and_die(bb_msg_read_error);
336 break;
337 }
338 in_index = 0;
339 }
340 c = str1[in_index++];
341 if ((opts & TR_OPT_delete) && invec[c])
342 continue;
343 coded = vector[c];
344 if ((opts & TR_OPT_squeeze_reps) && last == coded
345 && (invec[c] || outvec[coded])
346 ) {
347 continue;
348 }
349 str2[out_index++] = last = coded;
350 }
351
352 if (ENABLE_FEATURE_CLEAN_UP) {
353 free(vector);
354 free(str2);
355 free(str1);
356 }
357
358 return EXIT_SUCCESS;
359}
Note: See TracBrowser for help on using the repository browser.