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
22 //kbuild:lib-$(CONFIG_TR) += tr.o
28 //config: tr is used to squeeze, and/or delete characters from standard
29 //config: input, writing to standard output.
31 //config:config FEATURE_TR_CLASSES
32 //config: bool "Enable character classes (such as [:upper:])"
34 //config: depends on TR
36 //config: Enable character classes, enabling commands such as:
37 //config: tr [:upper:] [:lower:] to convert input into lowercase.
39 //config:config FEATURE_TR_EQUIV
40 //config: bool "Enable equivalence classes"
42 //config: depends on TR
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.
54 /* string buffer needs to be at least as big as the whole "alphabet".
55 * BUFSIZ == ASCII is ok, but we will realloc in expand
56 * even for smallest patterns, let's avoid that by using *2:
58 TR_BUFSIZ = (BUFSIZ > ASCII*2) ? BUFSIZ : ASCII*2,
61 static void map(char *pvector,
62 char *string1, unsigned string1_len,
63 char *string2, unsigned string2_len)
68 for (j = 0, i = 0; i < string1_len; i++) {
70 pvector[(unsigned char)(string1[i])] = last;
72 pvector[(unsigned char)(string1[i])] = last = string2[j++];
76 /* supported constructs:
77 * Ranges, e.g., 0-9 ==> 0123456789
78 * Escapes, e.g., \a ==> Control-G
79 * Character classes, e.g. [:upper:] ==> A...Z
80 * Equiv classess, e.g. [=A=] ==> A (hmmmmmmm?)
82 * \ooo-\ooo - octal ranges
83 * [x*N] - repeat char x N times
84 * [x*] - repeat char x until it fills STRING2:
85 * # echo qwe123 | /usr/bin/tr 123456789 '[d]'
87 * # echo qwe123 | /usr/bin/tr 123456789 '[d*]'
90 static unsigned expand(const char *arg, char **buffer_p)
92 char *buffer = *buffer_p;
94 unsigned size = TR_BUFSIZ;
95 unsigned i; /* can't be unsigned char: must be able to hold 256 */
99 if (pos + ASCII > size) {
101 *buffer_p = buffer = xrealloc(buffer, size);
105 buffer[pos++] = bb_process_escape_sequence(&arg);
108 if (arg[1] == '-') { /* "0-9..." */
110 if (ac == '\0') { /* "0-": copy verbatim */
111 buffer[pos++] = *arg++; /* copy '0' */
112 continue; /* next iter will copy '-' and stop */
114 i = (unsigned char) *arg;
115 while (i <= ac) /* ok: i is unsigned _int_ */
117 arg += 3; /* skip 0-9 */
120 if ((ENABLE_FEATURE_TR_CLASSES || ENABLE_FEATURE_TR_EQUIV)
124 i = (unsigned char) *arg++;
125 /* "[xyz...". i=x, arg points to y */
126 if (ENABLE_FEATURE_TR_CLASSES && i == ':') { /* [:class:] */
128 static const char classes[] ALIGN1 =
129 "alpha"CLO "alnum"CLO "digit"CLO
130 "lower"CLO "upper"CLO "space"CLO
131 "blank"CLO "punct"CLO "cntrl"CLO
134 CLASS_invalid = 0, /* we increment the retval */
151 /* xdigit needs 8, not 7 */
152 i = 7 + (arg[0] == 'x');
153 tmp = xstrndup(arg, i);
154 j = index_in_strings(classes, tmp) + 1;
157 if (j == CLASS_invalid)
161 if (j == CLASS_alnum || j == CLASS_digit || j == CLASS_xdigit) {
162 for (i = '0'; i <= '9'; i++)
165 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) {
166 for (i = 'A'; i <= 'Z'; i++)
169 if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) {
170 for (i = 'a'; i <= 'z'; i++)
173 if (j == CLASS_space || j == CLASS_blank) {
174 buffer[pos++] = '\t';
175 if (j == CLASS_space) {
176 buffer[pos++] = '\n';
177 buffer[pos++] = '\v';
178 buffer[pos++] = '\f';
179 buffer[pos++] = '\r';
183 if (j == CLASS_punct || j == CLASS_cntrl) {
184 for (i = '\0'; i < ASCII; i++) {
185 if ((j == CLASS_punct && isprint_asciionly(i) && !isalnum(i) && !isspace(i))
186 || (j == CLASS_cntrl && iscntrl(i))
192 if (j == CLASS_xdigit) {
193 for (i = 'A'; i <= 'F'; i++) {
194 buffer[pos + 6] = i | 0x20;
201 /* "[xyz...", i=x, arg points to y */
202 if (ENABLE_FEATURE_TR_EQUIV && i == '=') { /* [=CHAR=] */
203 buffer[pos++] = *arg; /* copy CHAR */
204 if (!arg[0] || arg[1] != '=' || arg[2] != ']')
206 arg += 3; /* skip CHAR=] */
209 /* The rest of "[xyz..." cases is treated as normal
210 * string, "[" has no special meaning here:
211 * tr "[a-z]" "[A-Z]" can be written as tr "a-z" "A-Z",
212 * also try tr "[a-z]" "_A-Z+" and you'll see that
213 * [] is not special here.
216 arg -= 2; /* points to "[" in "[xyz..." */
218 buffer[pos++] = *arg++;
223 /* NB: buffer is guaranteed to be at least TR_BUFSIZE
224 * (which is >= ASCII) big.
226 static int complement(char *buffer, int buffer_len)
235 if (memchr(buffer, ch, buffer_len) == NULL)
240 memcpy(buffer, conv, len);
244 int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
245 int tr_main(int argc UNUSED_PARAM, char **argv)
250 size_t in_index, out_index;
251 unsigned last = UCHAR_MAX + 1; /* not equal to any char */
252 unsigned char coded, c;
253 char *str1 = xmalloc(TR_BUFSIZ);
254 char *str2 = xmalloc(TR_BUFSIZ);
257 char *vector = xzalloc(ASCII * 3);
258 char *invec = vector + ASCII;
259 char *outvec = vector + ASCII * 2;
261 #define TR_OPT_complement (3 << 0)
262 #define TR_OPT_delete (1 << 2)
263 #define TR_OPT_squeeze_reps (1 << 3)
265 for (i = 0; i < ASCII; i++) {
267 /*invec[i] = outvec[i] = FALSE; - done by xzalloc */
270 /* -C/-c difference is that -C complements "characters",
271 * and -c complements "values" (binary bytes I guess).
272 * In POSIX locale, these are the same.
275 opt_complementary = "-1";
276 opts = getopt32(argv, "+Ccds"); /* '+': stop at first non-option */
279 str1_length = expand(*argv++, &str1);
281 if (opts & TR_OPT_complement)
282 str1_length = complement(str1, str1_length);
284 if (argv[0][0] == '\0')
285 bb_error_msg_and_die("STRING2 cannot be empty");
286 str2_length = expand(*argv, &str2);
287 map(vector, str1, str1_length,
290 for (i = 0; i < str1_length; i++)
291 invec[(unsigned char)(str1[i])] = TRUE;
292 for (i = 0; i < str2_length; i++)
293 outvec[(unsigned char)(str2[i])] = TRUE;
297 /* In this loop, str1 space is reused as input buffer,
298 * str2 - as output one. */
300 /* If we're out of input, flush output and read more input. */
301 if ((ssize_t)in_index == read_chars) {
303 xwrite(STDOUT_FILENO, str2, out_index);
307 read_chars = safe_read(STDIN_FILENO, str1, TR_BUFSIZ);
308 if (read_chars <= 0) {
310 bb_perror_msg_and_die(bb_msg_read_error);
315 c = str1[in_index++];
316 if ((opts & TR_OPT_delete) && invec[c])
319 if ((opts & TR_OPT_squeeze_reps) && last == coded
320 && (invec[c] || outvec[coded])
324 str2[out_index++] = last = coded;