Contents of /tags/mkinitrd-6_2_3/busybox/coreutils/tr.c
Parent Directory | Revision Log
Revision 1054 -
(show annotations)
(download)
Mon May 31 20:54:44 2010 UTC (14 years, 3 months ago) by niro
File MIME type: text/plain
File size: 7864 byte(s)
Mon May 31 20:54:44 2010 UTC (14 years, 3 months ago) by niro
File MIME type: text/plain
File size: 7864 byte(s)
tagged 'mkinitrd-6_2_3'
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: graph, print |
20 | */ |
21 | #include "libbb.h" |
22 | |
23 | enum { |
24 | ASCII = 256, |
25 | /* string buffer needs to be at least as big as the whole "alphabet". |
26 | * BUFSIZ == ASCII is ok, but we will realloc in expand |
27 | * even for smallest patterns, let's avoid that by using *2: |
28 | */ |
29 | TR_BUFSIZ = (BUFSIZ > ASCII*2) ? BUFSIZ : ASCII*2, |
30 | }; |
31 | |
32 | static void map(char *pvector, |
33 | char *string1, unsigned string1_len, |
34 | char *string2, unsigned string2_len) |
35 | { |
36 | char last = '0'; |
37 | unsigned i, j; |
38 | |
39 | for (j = 0, i = 0; i < string1_len; i++) { |
40 | if (string2_len <= j) |
41 | pvector[(unsigned char)(string1[i])] = last; |
42 | else |
43 | pvector[(unsigned char)(string1[i])] = last = string2[j++]; |
44 | } |
45 | } |
46 | |
47 | /* supported constructs: |
48 | * Ranges, e.g., 0-9 ==> 0123456789 |
49 | * Escapes, e.g., \a ==> Control-G |
50 | * Character classes, e.g. [:upper:] ==> A...Z |
51 | * Equiv classess, e.g. [=A=] ==> A (hmmmmmmm?) |
52 | * not supported: |
53 | * \ooo-\ooo - octal ranges |
54 | * [x*N] - repeat char x N times |
55 | * [x*] - repeat char x until it fills STRING2: |
56 | * # echo qwe123 | /usr/bin/tr 123456789 '[d]' |
57 | * qwe[d] |
58 | * # echo qwe123 | /usr/bin/tr 123456789 '[d*]' |
59 | * qweddd |
60 | */ |
61 | static unsigned expand(const char *arg, char **buffer_p) |
62 | { |
63 | char *buffer = *buffer_p; |
64 | unsigned pos = 0; |
65 | unsigned size = TR_BUFSIZ; |
66 | unsigned i; /* can't be unsigned char: must be able to hold 256 */ |
67 | unsigned char ac; |
68 | |
69 | while (*arg) { |
70 | if (pos + ASCII > size) { |
71 | size += ASCII; |
72 | *buffer_p = buffer = xrealloc(buffer, size); |
73 | } |
74 | if (*arg == '\\') { |
75 | arg++; |
76 | buffer[pos++] = bb_process_escape_sequence(&arg); |
77 | continue; |
78 | } |
79 | if (arg[1] == '-') { /* "0-9..." */ |
80 | ac = arg[2]; |
81 | if (ac == '\0') { /* "0-": copy verbatim */ |
82 | buffer[pos++] = *arg++; /* copy '0' */ |
83 | continue; /* next iter will copy '-' and stop */ |
84 | } |
85 | i = (unsigned char) *arg; |
86 | while (i <= ac) /* ok: i is unsigned _int_ */ |
87 | buffer[pos++] = i++; |
88 | arg += 3; /* skip 0-9 */ |
89 | continue; |
90 | } |
91 | if ((ENABLE_FEATURE_TR_CLASSES || ENABLE_FEATURE_TR_EQUIV) |
92 | && *arg == '[' |
93 | ) { |
94 | arg++; |
95 | i = (unsigned char) *arg++; |
96 | /* "[xyz...". i=x, arg points to y */ |
97 | if (ENABLE_FEATURE_TR_CLASSES && i == ':') { /* [:class:] */ |
98 | #define CLO ":]\0" |
99 | static const char classes[] ALIGN1 = |
100 | "alpha"CLO "alnum"CLO "digit"CLO |
101 | "lower"CLO "upper"CLO "space"CLO |
102 | "blank"CLO "punct"CLO "cntrl"CLO |
103 | "xdigit"CLO; |
104 | enum { |
105 | CLASS_invalid = 0, /* we increment the retval */ |
106 | CLASS_alpha = 1, |
107 | CLASS_alnum = 2, |
108 | CLASS_digit = 3, |
109 | CLASS_lower = 4, |
110 | CLASS_upper = 5, |
111 | CLASS_space = 6, |
112 | CLASS_blank = 7, |
113 | CLASS_punct = 8, |
114 | CLASS_cntrl = 9, |
115 | CLASS_xdigit = 10, |
116 | //CLASS_graph = 11, |
117 | //CLASS_print = 12, |
118 | }; |
119 | smalluint j; |
120 | char *tmp; |
121 | |
122 | /* xdigit needs 8, not 7 */ |
123 | i = 7 + (arg[0] == 'x'); |
124 | tmp = xstrndup(arg, i); |
125 | j = index_in_strings(classes, tmp) + 1; |
126 | free(tmp); |
127 | |
128 | if (j == CLASS_invalid) |
129 | goto skip_bracket; |
130 | |
131 | arg += i; |
132 | if (j == CLASS_alnum || j == CLASS_digit || j == CLASS_xdigit) { |
133 | for (i = '0'; i <= '9'; i++) |
134 | buffer[pos++] = i; |
135 | } |
136 | if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) { |
137 | for (i = 'A'; i <= 'Z'; i++) |
138 | buffer[pos++] = i; |
139 | } |
140 | if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) { |
141 | for (i = 'a'; i <= 'z'; i++) |
142 | buffer[pos++] = i; |
143 | } |
144 | if (j == CLASS_space || j == CLASS_blank) { |
145 | buffer[pos++] = '\t'; |
146 | if (j == CLASS_space) { |
147 | buffer[pos++] = '\n'; |
148 | buffer[pos++] = '\v'; |
149 | buffer[pos++] = '\f'; |
150 | buffer[pos++] = '\r'; |
151 | } |
152 | buffer[pos++] = ' '; |
153 | } |
154 | if (j == CLASS_punct || j == CLASS_cntrl) { |
155 | for (i = '\0'; i < ASCII; i++) { |
156 | if ((j == CLASS_punct && isprint_asciionly(i) && !isalnum(i) && !isspace(i)) |
157 | || (j == CLASS_cntrl && iscntrl(i)) |
158 | ) { |
159 | buffer[pos++] = i; |
160 | } |
161 | } |
162 | } |
163 | if (j == CLASS_xdigit) { |
164 | for (i = 'A'; i <= 'F'; i++) { |
165 | buffer[pos + 6] = i | 0x20; |
166 | buffer[pos++] = i; |
167 | } |
168 | pos += 6; |
169 | } |
170 | continue; |
171 | } |
172 | /* "[xyz...", i=x, arg points to y */ |
173 | if (ENABLE_FEATURE_TR_EQUIV && i == '=') { /* [=CHAR=] */ |
174 | buffer[pos++] = *arg; /* copy CHAR */ |
175 | if (!arg[0] || arg[1] != '=' || arg[2] != ']') |
176 | bb_show_usage(); |
177 | arg += 3; /* skip CHAR=] */ |
178 | continue; |
179 | } |
180 | /* The rest of "[xyz..." cases is treated as normal |
181 | * string, "[" has no special meaning here: |
182 | * tr "[a-z]" "[A-Z]" can be written as tr "a-z" "A-Z", |
183 | * also try tr "[a-z]" "_A-Z+" and you'll see that |
184 | * [] is not special here. |
185 | */ |
186 | skip_bracket: |
187 | arg -= 2; /* points to "[" in "[xyz..." */ |
188 | } |
189 | buffer[pos++] = *arg++; |
190 | } |
191 | return pos; |
192 | } |
193 | |
194 | /* NB: buffer is guaranteed to be at least TR_BUFSIZE |
195 | * (which is >= ASCII) big. |
196 | */ |
197 | static int complement(char *buffer, int buffer_len) |
198 | { |
199 | int len; |
200 | char conv[ASCII]; |
201 | unsigned char ch; |
202 | |
203 | len = 0; |
204 | ch = '\0'; |
205 | while (1) { |
206 | if (memchr(buffer, ch, buffer_len) == NULL) |
207 | conv[len++] = ch; |
208 | if (++ch == '\0') |
209 | break; |
210 | } |
211 | memcpy(buffer, conv, len); |
212 | return len; |
213 | } |
214 | |
215 | int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
216 | int tr_main(int argc UNUSED_PARAM, char **argv) |
217 | { |
218 | int i; |
219 | smalluint opts; |
220 | ssize_t read_chars; |
221 | size_t in_index, out_index; |
222 | unsigned last = UCHAR_MAX + 1; /* not equal to any char */ |
223 | unsigned char coded, c; |
224 | char *str1 = xmalloc(TR_BUFSIZ); |
225 | char *str2 = xmalloc(TR_BUFSIZ); |
226 | int str2_length; |
227 | int str1_length; |
228 | char *vector = xzalloc(ASCII * 3); |
229 | char *invec = vector + ASCII; |
230 | char *outvec = vector + ASCII * 2; |
231 | |
232 | #define TR_OPT_complement (3 << 0) |
233 | #define TR_OPT_delete (1 << 2) |
234 | #define TR_OPT_squeeze_reps (1 << 3) |
235 | |
236 | for (i = 0; i < ASCII; i++) { |
237 | vector[i] = i; |
238 | /*invec[i] = outvec[i] = FALSE; - done by xzalloc */ |
239 | } |
240 | |
241 | /* -C/-c difference is that -C complements "characters", |
242 | * and -c complements "values" (binary bytes I guess). |
243 | * In POSIX locale, these are the same. |
244 | */ |
245 | |
246 | opt_complementary = "-1"; |
247 | opts = getopt32(argv, "+Ccds"); /* '+': stop at first non-option */ |
248 | argv += optind; |
249 | |
250 | str1_length = expand(*argv++, &str1); |
251 | str2_length = 0; |
252 | if (opts & TR_OPT_complement) |
253 | str1_length = complement(str1, str1_length); |
254 | if (*argv) { |
255 | if (argv[0][0] == '\0') |
256 | bb_error_msg_and_die("STRING2 cannot be empty"); |
257 | str2_length = expand(*argv, &str2); |
258 | map(vector, str1, str1_length, |
259 | str2, str2_length); |
260 | } |
261 | for (i = 0; i < str1_length; i++) |
262 | invec[(unsigned char)(str1[i])] = TRUE; |
263 | for (i = 0; i < str2_length; i++) |
264 | outvec[(unsigned char)(str2[i])] = TRUE; |
265 | |
266 | goto start_from; |
267 | |
268 | /* In this loop, str1 space is reused as input buffer, |
269 | * str2 - as output one. */ |
270 | for (;;) { |
271 | /* If we're out of input, flush output and read more input. */ |
272 | if ((ssize_t)in_index == read_chars) { |
273 | if (out_index) { |
274 | xwrite(STDOUT_FILENO, str2, out_index); |
275 | start_from: |
276 | out_index = 0; |
277 | } |
278 | read_chars = safe_read(STDIN_FILENO, str1, TR_BUFSIZ); |
279 | if (read_chars <= 0) { |
280 | if (read_chars < 0) |
281 | bb_perror_msg_and_die(bb_msg_read_error); |
282 | break; |
283 | } |
284 | in_index = 0; |
285 | } |
286 | c = str1[in_index++]; |
287 | if ((opts & TR_OPT_delete) && invec[c]) |
288 | continue; |
289 | coded = vector[c]; |
290 | if ((opts & TR_OPT_squeeze_reps) && last == coded |
291 | && (invec[c] || outvec[coded]) |
292 | ) { |
293 | continue; |
294 | } |
295 | str2[out_index++] = last = coded; |
296 | } |
297 | |
298 | return EXIT_SUCCESS; |
299 | } |