1 /* tr.c - translate or delete characters
2 *
3 * Copyright 2014 Sandeep Sharma <[email protected]>
4 *
5 * See http://pubs.opengroup.org/onlinepubs/9699919799/utilities/tr.html
6 * TODO: -a (ascii)
7
8 USE_TR(NEWTOY(tr, "^<1>2Ccstd[+cC]", TOYFLAG_USR|TOYFLAG_BIN))
9
10 config TR
11 bool "tr"
12 default n
13 help
14 usage: tr [-cdst] SET1 [SET2]
15
16 Translate, squeeze, or delete characters from stdin, writing to stdout
17
18 -c/-C Take complement of SET1
19 -d Delete input characters coded SET1
20 -s Squeeze multiple output characters of SET2 into one character
21 -t Truncate SET1 to length of SET2
22 */
23
24 #define FOR_tr
25 #include "toys.h"
26
27 GLOBALS(
28 short *map;
29 int len1, len2;
30 )
31
32 enum {
33 class_alpha, class_alnum, class_digit,
34 class_lower,class_upper,class_space,class_blank,
35 class_punct,class_cntrl,class_xdigit,class_invalid
36 };
37
map_translation(char * set1,char * set2)38 static void map_translation(char *set1 , char *set2)
39 {
40 int i = TT.len1, k = 0;
41
42 if (FLAG(d))
43 for (; i; i--, k++) TT.map[set1[k]] = set1[k]|0x100; //set delete bit
44
45 if (FLAG(s)) {
46 for (i = TT.len1, k = 0; i; i--, k++)
47 TT.map[set1[k]] = TT.map[set1[k]]|0x200;
48 for (i = TT.len2, k = 0; i; i--, k++)
49 TT.map[set2[k]] = TT.map[set2[k]]|0x200;
50 }
51 i = k = 0;
52 while (!FLAG(d) && set2 && TT.len1--) { //ignore set2 if -d present
53 TT.map[set1[i]] = ((TT.map[set1[i]] & 0xFF00) | set2[k]);
54 if (set2[k + 1]) k++;
55 i++;
56 }
57 }
58
handle_escape_char(char ** esc_val)59 static int handle_escape_char(char **esc_val) //taken from printf
60 {
61 char *ptr = *esc_val;
62 int esc_length = 0;
63 unsigned base = 0, num = 0, result = 0, count = 0;
64
65 if (*ptr == 'x') {
66 ptr++;
67 esc_length++;
68 base = 16;
69 } else if (isdigit(*ptr)) base = 8;
70
71 while (esc_length < 3 && base) {
72 num = tolower(*ptr) - '0';
73 if (num > 10) num += ('0' - 'a' + 10);
74 if (num >= base) {
75 if (base == 16) {
76 esc_length--;
77 if (!esc_length) {// Invalid hex value eg. /xvd, print as it is /xvd
78 result = '\\';
79 ptr--;
80 }
81 }
82 break;
83 }
84 esc_length++;
85 result = (char)(count = (count * base) + num);
86 ptr++;
87 }
88 if (base) ptr--;
89 else if (!(result = unescape(*ptr))) {
90 result = '\\';
91 ptr--;
92 }
93 *esc_val = ptr;
94 return result;
95 }
96
find_class(char * class_name)97 static int find_class(char *class_name)
98 {
99 int i;
100 static char *class[] = {
101 "[:alpha:]","[:alnum:]","[:digit:]", "[:lower:]","[:upper:]","[:space:]",
102 "[:blank:]","[:punct:]","[:cntrl:]", "[:xdigit:]"
103 };
104
105 for (i = 0; i != class_invalid; i++)
106 if (!memcmp(class_name, class[i], 9+(*class_name == 'x'))) break;
107
108 return i;
109 }
110
expand_set(char * arg,int * len,size_t until)111 static char *expand_set(char *arg, int *len, size_t until)
112 {
113 int i = 0, j, k, size = 256;
114 char *set = xzalloc(size), *orig = arg;
115
116 while (*arg) {
117 if (arg-orig >= until) break;
118 if (i >= size) {
119 size += 256;
120 set = xrealloc(set, size);
121 }
122 if (*arg == '\\') {
123 arg++;
124 set[i++] = handle_escape_char(&arg);
125 arg++;
126 continue;
127 }
128 if (arg[1] == '-') {
129 if (!arg[2]) goto save;
130 j = *arg;
131 k = arg[2];
132 if (j > k) perror_exit("reverse colating order");
133 while (j <= k) set[i++] = j++;
134 arg += 3;
135 continue;
136 }
137 if (*arg == '[' && arg[1] == ':') {
138
139 if ((j = find_class(arg)) == class_invalid) goto save;
140
141 if ((j == class_alpha) || (j == class_upper) || (j == class_alnum))
142 for (k = 'A'; k <= 'Z'; k++) set[i++] = k;
143 if ((j == class_alpha) || (j == class_lower) || (j == class_alnum))
144 for (k = 'a'; k <= 'z'; k++) set[i++] = k;
145 if ((j == class_alnum) || (j == class_digit) || (j == class_xdigit))
146 for (k = '0'; k <= '9'; k++) set[i++] = k;
147 if (j == class_space || j == class_blank) {
148 set[i++] = '\t';
149 if (j == class_space) {
150 set[i++] = '\n';
151 set[i++] = '\f';
152 set[i++] = '\r';
153 set[i++] = '\v';
154 }
155 set[i++] = ' ';
156 }
157 if (j == class_punct)
158 for (k = 0; k <= 255; k++) if (ispunct(k)) set[i++] = k;
159 if (j == class_cntrl)
160 for (k = 0; k <= 255; k++) if (iscntrl(k)) set[i++] = k;
161 if (j == class_xdigit) {
162 for (k = 'A'; k <= 'F'; k++) {
163 set[i + 6] = k | 0x20;
164 set[i++] = k;
165 }
166 i += 6;
167 arg += 10;
168 continue;
169 }
170
171 arg += 9; //never here for class_xdigit.
172 continue;
173 }
174 if (*arg == '[' && arg[1] == '=') { //[=char=] only
175 arg += 2;
176 if (*arg) set[i++] = *arg;
177 if (!arg[1] || arg[1] != '=' || arg[2] != ']')
178 error_exit("bad equiv class");
179 continue;
180 }
181 save:
182 set[i++] = *arg++;
183 }
184 *len = i;
185 return set;
186 }
187
print_map(char * set1,char * set2)188 static void print_map(char *set1, char *set2)
189 {
190 int n, ch, src, dst, prev = -1;
191
192 while ((n = read(0, toybuf, sizeof(toybuf)))) {
193 if (!FLAG(d) && !FLAG(s))
194 for (dst = 0; dst < n; dst++) toybuf[dst] = TT.map[toybuf[dst]];
195 else for (src = dst = 0; src < n; src++) {
196 ch = TT.map[toybuf[src]];
197 if (FLAG(d) && (ch & 0x100)) continue;
198 if (FLAG(s) && ((ch & 0x200) && prev == ch)) continue;
199 toybuf[dst++] = prev = ch;
200 }
201 xwrite(1, toybuf, dst);
202 }
203 }
204
do_complement(char ** set)205 static void do_complement(char **set)
206 {
207 int i = 0, j = 0;
208 char *comp = xmalloc(256);
209
210 for (; i < 256; i++) {
211 if (memchr(*set, i, TT.len1)) continue;
212 else comp[j++] = (char)i;
213 }
214 free(*set);
215 TT.len1 = j;
216 *set = comp;
217 }
218
tr_main(void)219 void tr_main(void)
220 {
221 char *set1, *set2 = NULL;
222 int i = 0;
223
224 TT.map = xmalloc(256*sizeof(*TT.map));
225 for (; i < 256; i++) TT.map[i] = i; //init map
226
227 set1 = expand_set(*toys.optargs, &TT.len1,
228 (FLAG(t) && toys.optargs[1]) ? strlen(toys.optargs[1]) : -1);
229 if (FLAG(c)) do_complement(&set1);
230 if (toys.optargs[1]) {
231 if (!*toys.optargs[1]) error_exit("set2 can't be empty string");
232 set2 = expand_set(toys.optargs[1], &TT.len2, -1);
233 }
234 map_translation(set1, set2);
235
236 print_map(set1, set2);
237 free(set1);
238 free(set2);
239 }
240