15 |
* |
* |
16 |
* Licensed under GPLv2 or later, see file LICENSE in this tarball for details. |
* 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 |
#include "busybox.h" |
* TODO: xdigit, graph, print |
20 |
|
*/ |
21 |
// Even with -funsigned-char, gcc still complains about char as an array index. |
#include "libbb.h" |
|
|
|
|
#define GCC4_IS_STUPID int |
|
22 |
|
|
23 |
#define ASCII 0377 |
#define ASCII 0377 |
24 |
|
|
25 |
/* some "globals" shared across this file */ |
static void map(char *pvector, |
26 |
static char com_fl, del_fl, sq_fl; |
unsigned char *string1, unsigned int string1_len, |
27 |
/* these last are pointers to static buffers declared in tr_main */ |
unsigned char *string2, unsigned int string2_len) |
|
static char *poutput, *pvector, *pinvec, *poutvec; |
|
|
|
|
|
static void convert(void) |
|
|
{ |
|
|
int read_chars = 0, in_index = 0, out_index = 0, c, coded, last = -1; |
|
|
|
|
|
for (;;) { |
|
|
// If we're out of input, flush output and read more input. |
|
|
|
|
|
if (in_index == read_chars) { |
|
|
if (out_index) { |
|
|
if (write(1, (char *) poutput, out_index) != out_index) |
|
|
bb_error_msg_and_die(bb_msg_write_error); |
|
|
out_index = 0; |
|
|
} |
|
|
|
|
|
if ((read_chars = read(0, bb_common_bufsiz1, BUFSIZ)) <= 0) { |
|
|
if (write(1, (char *) poutput, out_index) != out_index) |
|
|
bb_error_msg(bb_msg_write_error); |
|
|
exit(0); |
|
|
} |
|
|
in_index = 0; |
|
|
} |
|
|
c = bb_common_bufsiz1[in_index++]; |
|
|
coded = pvector[c]; |
|
|
if (del_fl && pinvec[c]) |
|
|
continue; |
|
|
if (sq_fl && last == coded && (pinvec[c] || poutvec[coded])) |
|
|
continue; |
|
|
poutput[out_index++] = last = coded; |
|
|
} |
|
|
|
|
|
/* NOTREACHED */ |
|
|
} |
|
|
|
|
|
static void map(char *string1, unsigned int string1_len, |
|
|
char *string2, unsigned int string2_len) |
|
28 |
{ |
{ |
29 |
char last = '0'; |
char last = '0'; |
30 |
unsigned int i, j; |
unsigned int i, j; |
31 |
|
|
32 |
for (j = 0, i = 0; i < string1_len; i++) { |
for (j = 0, i = 0; i < string1_len; i++) { |
33 |
if (string2_len <= j) |
if (string2_len <= j) |
34 |
pvector[(GCC4_IS_STUPID)string1[i]] = last; |
pvector[string1[i]] = last; |
35 |
else |
else |
36 |
pvector[(GCC4_IS_STUPID)string1[i]] = last = string2[j++]; |
pvector[string1[i]] = last = string2[j++]; |
37 |
} |
} |
38 |
} |
} |
39 |
|
|
40 |
/* supported constructs: |
/* supported constructs: |
41 |
* Ranges, e.g., [0-9] ==> 0123456789 |
* Ranges, e.g., 0-9 ==> 0123456789 |
42 |
* Escapes, e.g., \a ==> Control-G |
* Ranges, e.g., [0-9] ==> 0123456789 |
43 |
* Character classes, e.g. [:upper:] ==> A ... Z |
* Escapes, e.g., \a ==> Control-G |
44 |
|
* Character classes, e.g. [:upper:] ==> A...Z |
45 |
|
* Equiv classess, e.g. [=A=] ==> A (hmmmmmmm?) |
46 |
*/ |
*/ |
47 |
static unsigned int expand(const char *arg, char *buffer) |
static unsigned int expand(const char *arg, char *buffer) |
48 |
{ |
{ |
49 |
char *buffer_start = buffer; |
char *buffer_start = buffer; |
50 |
int i, ac; |
unsigned i; /* can't be unsigned char: must be able to hold 256 */ |
51 |
|
unsigned char ac; |
52 |
|
|
53 |
while (*arg) { |
while (*arg) { |
54 |
if (*arg == '\\') { |
if (*arg == '\\') { |
55 |
arg++; |
arg++; |
56 |
*buffer++ = bb_process_escape_sequence(&arg); |
*buffer++ = bb_process_escape_sequence(&arg); |
57 |
} else if (*(arg+1) == '-') { |
continue; |
58 |
ac = *(arg+2); |
} |
59 |
if(ac == 0) { |
if (arg[1] == '-') { /* "0-9..." */ |
60 |
*buffer++ = *arg++; |
ac = arg[2]; |
61 |
continue; |
if (ac == '\0') { /* "0-": copy verbatim */ |
62 |
|
*buffer++ = *arg++; /* copy '0' */ |
63 |
|
continue; /* next iter will copy '-' and stop */ |
64 |
} |
} |
65 |
i = *arg; |
i = *arg; |
66 |
while (i <= ac) |
while (i <= ac) /* ok: i is unsigned _int_ */ |
67 |
*buffer++ = i++; |
*buffer++ = i++; |
68 |
arg += 3; /* Skip the assumed a-z */ |
arg += 3; /* skip 0-9 */ |
69 |
} else if (*arg == '[') { |
continue; |
70 |
|
} |
71 |
|
if (*arg == '[') { /* "[xyz..." */ |
72 |
arg++; |
arg++; |
73 |
i = *arg++; |
i = *arg++; |
74 |
|
/* "[xyz...", i=x, arg points to y */ |
75 |
if (ENABLE_FEATURE_TR_CLASSES && i == ':') { |
if (ENABLE_FEATURE_TR_CLASSES && i == ':') { |
76 |
if (strncmp(arg, "alpha", 5) == 0) { |
#define CLO ":]\0" |
77 |
for (i = 'A'; i <= 'Z'; i++) |
static const char classes[] ALIGN1 = |
78 |
*buffer++ = i; |
"alpha"CLO "alnum"CLO "digit"CLO |
79 |
for (i = 'a'; i <= 'z'; i++) |
"lower"CLO "upper"CLO "space"CLO |
80 |
*buffer++ = i; |
"blank"CLO "punct"CLO "cntrl"CLO; |
81 |
|
#define CLASS_invalid 0 /* we increment the retval */ |
82 |
|
#define CLASS_alpha 1 |
83 |
|
#define CLASS_alnum 2 |
84 |
|
#define CLASS_digit 3 |
85 |
|
#define CLASS_lower 4 |
86 |
|
#define CLASS_upper 5 |
87 |
|
#define CLASS_space 6 |
88 |
|
#define CLASS_blank 7 |
89 |
|
#define CLASS_punct 8 |
90 |
|
#define CLASS_cntrl 9 |
91 |
|
//#define CLASS_xdigit 10 |
92 |
|
//#define CLASS_graph 11 |
93 |
|
//#define CLASS_print 12 |
94 |
|
smalluint j; |
95 |
|
{ /* not really pretty.. */ |
96 |
|
char *tmp = xstrndup(arg, 7); // warning: xdigit would need 8, not 7 |
97 |
|
j = index_in_strings(classes, tmp) + 1; |
98 |
|
free(tmp); |
99 |
} |
} |
100 |
else if (strncmp(arg, "alnum", 5) == 0) { |
if (j == CLASS_alnum || j == CLASS_digit) { |
101 |
for (i = '0'; i <= '9'; i++) |
for (i = '0'; i <= '9'; i++) |
102 |
*buffer++ = i; |
*buffer++ = i; |
103 |
|
} |
104 |
|
if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) { |
105 |
for (i = 'A'; i <= 'Z'; i++) |
for (i = 'A'; i <= 'Z'; i++) |
106 |
*buffer++ = i; |
*buffer++ = i; |
|
for (i = 'a'; i <= 'z'; i++) |
|
|
*buffer++ = i; |
|
107 |
} |
} |
108 |
else if (strncmp(arg, "digit", 5) == 0) |
if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) { |
|
for (i = '0'; i <= '9'; i++) |
|
|
*buffer++ = i; |
|
|
else if (strncmp(arg, "lower", 5) == 0) |
|
109 |
for (i = 'a'; i <= 'z'; i++) |
for (i = 'a'; i <= 'z'; i++) |
110 |
*buffer++ = i; |
*buffer++ = i; |
|
else if (strncmp(arg, "upper", 5) == 0) |
|
|
for (i = 'A'; i <= 'Z'; i++) |
|
|
*buffer++ = i; |
|
|
else if (strncmp(arg, "space", 5) == 0) { |
|
|
const char s[] = "\t\n\v\f\r "; |
|
|
strcat((char*)buffer, s); |
|
|
buffer += sizeof(s) - 1; |
|
111 |
} |
} |
112 |
else if (strncmp(arg, "blank", 5) == 0) { |
if (j == CLASS_space || j == CLASS_blank) { |
113 |
*buffer++ = '\t'; |
*buffer++ = '\t'; |
114 |
|
if (j == CLASS_space) { |
115 |
|
*buffer++ = '\n'; |
116 |
|
*buffer++ = '\v'; |
117 |
|
*buffer++ = '\f'; |
118 |
|
*buffer++ = '\r'; |
119 |
|
} |
120 |
*buffer++ = ' '; |
*buffer++ = ' '; |
121 |
} |
} |
122 |
/* gcc gives a warning if braces aren't used here */ |
if (j == CLASS_punct || j == CLASS_cntrl) { |
123 |
else if (strncmp(arg, "punct", 5) == 0) { |
for (i = '\0'; i <= ASCII; i++) |
124 |
for (i = 0; i <= ASCII; i++) |
if ((j == CLASS_punct && isprint(i) && !isalnum(i) && !isspace(i)) |
125 |
if (isprint(i) && (!isalnum(i)) && (!isspace(i))) |
|| (j == CLASS_cntrl && iscntrl(i))) |
126 |
*buffer++ = i; |
*buffer++ = i; |
127 |
} |
} |
128 |
else if (strncmp(arg, "cntrl", 5) == 0) { |
if (j == CLASS_invalid) { |
|
for (i = 0; i <= ASCII; i++) |
|
|
if (iscntrl(i)) |
|
|
*buffer++ = i; |
|
|
} |
|
|
else { |
|
129 |
*buffer++ = '['; |
*buffer++ = '['; |
130 |
*buffer++ = ':'; |
*buffer++ = ':'; |
131 |
continue; |
continue; |
132 |
} |
} |
133 |
break; |
break; |
134 |
} |
} |
135 |
if (ENABLE_FEATURE_TR_EQUIV && i == '=') { |
/* "[xyz...", i=x, arg points to y */ |
136 |
*buffer++ = *arg; |
if (ENABLE_FEATURE_TR_EQUIV && i == '=') { /* [=CHAR=] */ |
137 |
/* skip the closing =] */ |
*buffer++ = *arg; /* copy CHAR */ |
138 |
arg += 3; |
if (!*arg || arg[1] != '=' || arg[2] != ']') |
139 |
|
bb_show_usage(); |
140 |
|
arg += 3; /* skip CHAR=] */ |
141 |
continue; |
continue; |
142 |
} |
} |
143 |
if (*arg++ != '-') { |
if (i == '\0' || *arg != '-') { /* not [x-...] - copy verbatim */ |
144 |
*buffer++ = '['; |
*buffer++ = '['; |
145 |
arg -= 2; |
arg--; /* points to x */ |
146 |
continue; |
continue; /* copy all, including eventual ']' */ |
147 |
} |
} |
148 |
|
/* [x-z] */ |
149 |
|
arg++; /* skip - */ |
150 |
|
if (arg[0] == '\0' || arg[1] != ']') |
151 |
|
bb_show_usage(); |
152 |
ac = *arg++; |
ac = *arg++; |
153 |
while (i <= ac) |
while (i <= ac) |
154 |
*buffer++ = i++; |
*buffer++ = i++; |
155 |
arg++; /* Skip the assumed ']' */ |
arg++; /* skip ] */ |
156 |
} else |
continue; |
157 |
*buffer++ = *arg++; |
} |
158 |
|
*buffer++ = *arg++; |
159 |
} |
} |
|
|
|
160 |
return (buffer - buffer_start); |
return (buffer - buffer_start); |
161 |
} |
} |
162 |
|
|
163 |
static int complement(char *buffer, int buffer_len) |
static int complement(char *buffer, int buffer_len) |
164 |
{ |
{ |
165 |
short i, j, ix; |
int i, j, ix; |
166 |
char conv[ASCII + 2]; |
char conv[ASCII + 2]; |
167 |
|
|
168 |
ix = 0; |
ix = 0; |
169 |
for (i = 0; i <= ASCII; i++) { |
for (i = '\0'; i <= ASCII; i++) { |
170 |
for (j = 0; j < buffer_len; j++) |
for (j = 0; j < buffer_len; j++) |
171 |
if (buffer[j] == i) |
if (buffer[j] == i) |
172 |
break; |
break; |
177 |
return ix; |
return ix; |
178 |
} |
} |
179 |
|
|
180 |
int tr_main(int argc, char **argv) |
int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
181 |
|
int tr_main(int argc UNUSED_PARAM, char **argv) |
182 |
{ |
{ |
183 |
unsigned char *ptr; |
int output_length = 0, input_length; |
|
int output_length=0, input_length; |
|
|
int idx = 1; |
|
184 |
int i; |
int i; |
185 |
RESERVE_CONFIG_BUFFER(output, BUFSIZ); |
smalluint flags; |
186 |
RESERVE_CONFIG_BUFFER(vector, ASCII+1); |
ssize_t read_chars = 0; |
187 |
RESERVE_CONFIG_BUFFER(invec, ASCII+1); |
size_t in_index = 0, out_index = 0; |
188 |
RESERVE_CONFIG_BUFFER(outvec, ASCII+1); |
unsigned last = UCHAR_MAX + 1; /* not equal to any char */ |
189 |
|
unsigned char coded, c; |
190 |
/* ... but make them available globally */ |
unsigned char *output = xmalloc(BUFSIZ); |
191 |
poutput = output; |
char *vector = xzalloc((ASCII+1) * 3); |
192 |
pvector = vector; |
char *invec = vector + (ASCII+1); |
193 |
pinvec = invec; |
char *outvec = vector + (ASCII+1) * 2; |
194 |
poutvec = outvec; |
|
195 |
|
#define TR_OPT_complement (1 << 0) |
196 |
if (argc > 1 && argv[idx][0] == '-') { |
#define TR_OPT_delete (1 << 1) |
197 |
for (ptr = (unsigned char *) &argv[idx][1]; *ptr; ptr++) { |
#define TR_OPT_squeeze_reps (1 << 2) |
198 |
switch (*ptr) { |
|
199 |
case 'c': |
flags = getopt32(argv, "+cds"); /* '+': stop at first non-option */ |
200 |
com_fl = TRUE; |
argv += optind; |
201 |
break; |
|
|
case 'd': |
|
|
del_fl = TRUE; |
|
|
break; |
|
|
case 's': |
|
|
sq_fl = TRUE; |
|
|
break; |
|
|
default: |
|
|
bb_show_usage(); |
|
|
} |
|
|
} |
|
|
idx++; |
|
|
} |
|
202 |
for (i = 0; i <= ASCII; i++) { |
for (i = 0; i <= ASCII; i++) { |
203 |
vector[i] = i; |
vector[i] = i; |
204 |
invec[i] = outvec[i] = FALSE; |
/*invec[i] = outvec[i] = FALSE; - done by xzalloc */ |
205 |
} |
} |
206 |
|
|
207 |
if (argv[idx] != NULL) { |
#define tr_buf bb_common_bufsiz1 |
208 |
input_length = expand(argv[idx++], bb_common_bufsiz1); |
if (*argv != NULL) { |
209 |
if (com_fl) |
input_length = expand(*argv++, tr_buf); |
210 |
input_length = complement(bb_common_bufsiz1, input_length); |
if (flags & TR_OPT_complement) |
211 |
if (argv[idx] != NULL) { |
input_length = complement(tr_buf, input_length); |
212 |
if (*argv[idx] == '\0') |
if (*argv) { |
213 |
|
if (argv[0][0] == '\0') |
214 |
bb_error_msg_and_die("STRING2 cannot be empty"); |
bb_error_msg_and_die("STRING2 cannot be empty"); |
215 |
output_length = expand(argv[idx], output); |
output_length = expand(*argv, (char *)output); |
216 |
map(bb_common_bufsiz1, input_length, output, output_length); |
map(vector, (unsigned char *)tr_buf, input_length, output, output_length); |
217 |
} |
} |
218 |
for (i = 0; i < input_length; i++) |
for (i = 0; i < input_length; i++) |
219 |
invec[(GCC4_IS_STUPID)bb_common_bufsiz1[i]] = TRUE; |
invec[(unsigned char)tr_buf[i]] = TRUE; |
220 |
for (i = 0; i < output_length; i++) |
for (i = 0; i < output_length; i++) |
221 |
outvec[(GCC4_IS_STUPID)output[i]] = TRUE; |
outvec[output[i]] = TRUE; |
222 |
} |
} |
223 |
convert(); |
|
224 |
return 0; |
for (;;) { |
225 |
|
/* If we're out of input, flush output and read more input. */ |
226 |
|
if ((ssize_t)in_index == read_chars) { |
227 |
|
if (out_index) { |
228 |
|
xwrite(STDOUT_FILENO, (char *)output, out_index); |
229 |
|
out_index = 0; |
230 |
|
} |
231 |
|
read_chars = safe_read(STDIN_FILENO, tr_buf, BUFSIZ); |
232 |
|
if (read_chars <= 0) { |
233 |
|
if (read_chars < 0) |
234 |
|
bb_perror_msg_and_die(bb_msg_read_error); |
235 |
|
exit(EXIT_SUCCESS); |
236 |
|
} |
237 |
|
in_index = 0; |
238 |
|
} |
239 |
|
c = tr_buf[in_index++]; |
240 |
|
coded = vector[c]; |
241 |
|
if ((flags & TR_OPT_delete) && invec[c]) |
242 |
|
continue; |
243 |
|
if ((flags & TR_OPT_squeeze_reps) && last == coded |
244 |
|
&& (invec[c] || outvec[coded])) |
245 |
|
continue; |
246 |
|
output[out_index++] = last = coded; |
247 |
|
} |
248 |
|
/* NOTREACHED */ |
249 |
|
return EXIT_SUCCESS; |
250 |
} |
} |