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 | #include "libbb.h"
|
---|
51 |
|
---|
52 | enum {
|
---|
53 | ASCII = 256,
|
---|
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:
|
---|
57 | */
|
---|
58 | TR_BUFSIZ = (BUFSIZ > ASCII*2) ? BUFSIZ : ASCII*2,
|
---|
59 | };
|
---|
60 |
|
---|
61 | static void map(char *pvector,
|
---|
62 | char *string1, unsigned string1_len,
|
---|
63 | char *string2, unsigned string2_len)
|
---|
64 | {
|
---|
65 | char last = '0';
|
---|
66 | unsigned i, j;
|
---|
67 |
|
---|
68 | for (j = 0, i = 0; i < string1_len; i++) {
|
---|
69 | if (string2_len <= j)
|
---|
70 | pvector[(unsigned char)(string1[i])] = last;
|
---|
71 | else
|
---|
72 | pvector[(unsigned char)(string1[i])] = last = string2[j++];
|
---|
73 | }
|
---|
74 | }
|
---|
75 |
|
---|
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?)
|
---|
81 | * not supported:
|
---|
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]'
|
---|
86 | * qwe[d]
|
---|
87 | * # echo qwe123 | /usr/bin/tr 123456789 '[d*]'
|
---|
88 | * qweddd
|
---|
89 | */
|
---|
90 | static unsigned expand(const char *arg, char **buffer_p)
|
---|
91 | {
|
---|
92 | char *buffer = *buffer_p;
|
---|
93 | unsigned pos = 0;
|
---|
94 | unsigned size = TR_BUFSIZ;
|
---|
95 | unsigned i; /* can't be unsigned char: must be able to hold 256 */
|
---|
96 | unsigned char ac;
|
---|
97 |
|
---|
98 | while (*arg) {
|
---|
99 | if (pos + ASCII > size) {
|
---|
100 | size += ASCII;
|
---|
101 | *buffer_p = buffer = xrealloc(buffer, size);
|
---|
102 | }
|
---|
103 | if (*arg == '\\') {
|
---|
104 | arg++;
|
---|
105 | buffer[pos++] = bb_process_escape_sequence(&arg);
|
---|
106 | continue;
|
---|
107 | }
|
---|
108 | if (arg[1] == '-') { /* "0-9..." */
|
---|
109 | ac = arg[2];
|
---|
110 | if (ac == '\0') { /* "0-": copy verbatim */
|
---|
111 | buffer[pos++] = *arg++; /* copy '0' */
|
---|
112 | continue; /* next iter will copy '-' and stop */
|
---|
113 | }
|
---|
114 | i = (unsigned char) *arg;
|
---|
115 | while (i <= ac) /* ok: i is unsigned _int_ */
|
---|
116 | buffer[pos++] = i++;
|
---|
117 | arg += 3; /* skip 0-9 */
|
---|
118 | continue;
|
---|
119 | }
|
---|
120 | if ((ENABLE_FEATURE_TR_CLASSES || ENABLE_FEATURE_TR_EQUIV)
|
---|
121 | && *arg == '['
|
---|
122 | ) {
|
---|
123 | arg++;
|
---|
124 | i = (unsigned char) *arg++;
|
---|
125 | /* "[xyz...". i=x, arg points to y */
|
---|
126 | if (ENABLE_FEATURE_TR_CLASSES && i == ':') { /* [:class:] */
|
---|
127 | #define CLO ":]\0"
|
---|
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
|
---|
132 | "xdigit"CLO;
|
---|
133 | enum {
|
---|
134 | CLASS_invalid = 0, /* we increment the retval */
|
---|
135 | CLASS_alpha = 1,
|
---|
136 | CLASS_alnum = 2,
|
---|
137 | CLASS_digit = 3,
|
---|
138 | CLASS_lower = 4,
|
---|
139 | CLASS_upper = 5,
|
---|
140 | CLASS_space = 6,
|
---|
141 | CLASS_blank = 7,
|
---|
142 | CLASS_punct = 8,
|
---|
143 | CLASS_cntrl = 9,
|
---|
144 | CLASS_xdigit = 10,
|
---|
145 | //CLASS_graph = 11,
|
---|
146 | //CLASS_print = 12,
|
---|
147 | };
|
---|
148 | smalluint j;
|
---|
149 | char *tmp;
|
---|
150 |
|
---|
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;
|
---|
155 | free(tmp);
|
---|
156 |
|
---|
157 | if (j == CLASS_invalid)
|
---|
158 | goto skip_bracket;
|
---|
159 |
|
---|
160 | arg += i;
|
---|
161 | if (j == CLASS_alnum || j == CLASS_digit || j == CLASS_xdigit) {
|
---|
162 | for (i = '0'; i <= '9'; i++)
|
---|
163 | buffer[pos++] = i;
|
---|
164 | }
|
---|
165 | if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) {
|
---|
166 | for (i = 'A'; i <= 'Z'; i++)
|
---|
167 | buffer[pos++] = i;
|
---|
168 | }
|
---|
169 | if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) {
|
---|
170 | for (i = 'a'; i <= 'z'; i++)
|
---|
171 | buffer[pos++] = i;
|
---|
172 | }
|
---|
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';
|
---|
180 | }
|
---|
181 | buffer[pos++] = ' ';
|
---|
182 | }
|
---|
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))
|
---|
187 | ) {
|
---|
188 | buffer[pos++] = i;
|
---|
189 | }
|
---|
190 | }
|
---|
191 | }
|
---|
192 | if (j == CLASS_xdigit) {
|
---|
193 | for (i = 'A'; i <= 'F'; i++) {
|
---|
194 | buffer[pos + 6] = i | 0x20;
|
---|
195 | buffer[pos++] = i;
|
---|
196 | }
|
---|
197 | pos += 6;
|
---|
198 | }
|
---|
199 | continue;
|
---|
200 | }
|
---|
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] != ']')
|
---|
205 | bb_show_usage();
|
---|
206 | arg += 3; /* skip CHAR=] */
|
---|
207 | continue;
|
---|
208 | }
|
---|
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.
|
---|
214 | */
|
---|
215 | skip_bracket:
|
---|
216 | arg -= 2; /* points to "[" in "[xyz..." */
|
---|
217 | }
|
---|
218 | buffer[pos++] = *arg++;
|
---|
219 | }
|
---|
220 | return pos;
|
---|
221 | }
|
---|
222 |
|
---|
223 | /* NB: buffer is guaranteed to be at least TR_BUFSIZE
|
---|
224 | * (which is >= ASCII) big.
|
---|
225 | */
|
---|
226 | static int complement(char *buffer, int buffer_len)
|
---|
227 | {
|
---|
228 | int len;
|
---|
229 | char conv[ASCII];
|
---|
230 | unsigned char ch;
|
---|
231 |
|
---|
232 | len = 0;
|
---|
233 | ch = '\0';
|
---|
234 | while (1) {
|
---|
235 | if (memchr(buffer, ch, buffer_len) == NULL)
|
---|
236 | conv[len++] = ch;
|
---|
237 | if (++ch == '\0')
|
---|
238 | break;
|
---|
239 | }
|
---|
240 | memcpy(buffer, conv, len);
|
---|
241 | return len;
|
---|
242 | }
|
---|
243 |
|
---|
244 | int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
|
---|
245 | int tr_main(int argc UNUSED_PARAM, char **argv)
|
---|
246 | {
|
---|
247 | int i;
|
---|
248 | smalluint opts;
|
---|
249 | ssize_t read_chars;
|
---|
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);
|
---|
255 | int str2_length;
|
---|
256 | int str1_length;
|
---|
257 | char *vector = xzalloc(ASCII * 3);
|
---|
258 | char *invec = vector + ASCII;
|
---|
259 | char *outvec = vector + ASCII * 2;
|
---|
260 |
|
---|
261 | #define TR_OPT_complement (3 << 0)
|
---|
262 | #define TR_OPT_delete (1 << 2)
|
---|
263 | #define TR_OPT_squeeze_reps (1 << 3)
|
---|
264 |
|
---|
265 | for (i = 0; i < ASCII; i++) {
|
---|
266 | vector[i] = i;
|
---|
267 | /*invec[i] = outvec[i] = FALSE; - done by xzalloc */
|
---|
268 | }
|
---|
269 |
|
---|
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.
|
---|
273 | */
|
---|
274 |
|
---|
275 | opt_complementary = "-1";
|
---|
276 | opts = getopt32(argv, "+Ccds"); /* '+': stop at first non-option */
|
---|
277 | argv += optind;
|
---|
278 |
|
---|
279 | str1_length = expand(*argv++, &str1);
|
---|
280 | str2_length = 0;
|
---|
281 | if (opts & TR_OPT_complement)
|
---|
282 | str1_length = complement(str1, str1_length);
|
---|
283 | if (*argv) {
|
---|
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,
|
---|
288 | str2, str2_length);
|
---|
289 | }
|
---|
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;
|
---|
294 |
|
---|
295 | goto start_from;
|
---|
296 |
|
---|
297 | /* In this loop, str1 space is reused as input buffer,
|
---|
298 | * str2 - as output one. */
|
---|
299 | for (;;) {
|
---|
300 | /* If we're out of input, flush output and read more input. */
|
---|
301 | if ((ssize_t)in_index == read_chars) {
|
---|
302 | if (out_index) {
|
---|
303 | xwrite(STDOUT_FILENO, str2, out_index);
|
---|
304 | start_from:
|
---|
305 | out_index = 0;
|
---|
306 | }
|
---|
307 | read_chars = safe_read(STDIN_FILENO, str1, TR_BUFSIZ);
|
---|
308 | if (read_chars <= 0) {
|
---|
309 | if (read_chars < 0)
|
---|
310 | bb_perror_msg_and_die(bb_msg_read_error);
|
---|
311 | break;
|
---|
312 | }
|
---|
313 | in_index = 0;
|
---|
314 | }
|
---|
315 | c = str1[in_index++];
|
---|
316 | if ((opts & TR_OPT_delete) && invec[c])
|
---|
317 | continue;
|
---|
318 | coded = vector[c];
|
---|
319 | if ((opts & TR_OPT_squeeze_reps) && last == coded
|
---|
320 | && (invec[c] || outvec[coded])
|
---|
321 | ) {
|
---|
322 | continue;
|
---|
323 | }
|
---|
324 | str2[out_index++] = last = coded;
|
---|
325 | }
|
---|
326 |
|
---|
327 | return EXIT_SUCCESS;
|
---|
328 | }
|
---|