emo/busybox
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

tr: support [:xdigit:], fix handling of ranges and [x]'s.

Browse Source 
add testsuite entry for each of 3 bugs fixed.

function                                             old     new   delta      
static.classes                                        73      82      +9      
expand                                              1738    1743      +5      
complement                                            74      72      -2      
tr_main                                              472     463      -9      
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 2/2 up/down: 14/-11)              Total: 3 bytes
This commit is contained in:
Denis Vlasenko 2009-03-01 04:50:18 +00:00
parent 31773b71e8
commit 1bfcc8b08e
2 changed files with 69 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

81
coreutils/tr.c
View File

@ -23,11 +23,11 @@
#define ASCII 0377 #define ASCII 0377
static void map(char *pvector, static void map(char *pvector,
unsigned char *string1, unsigned int string1_len, unsigned char *string1, unsigned string1_len,
unsigned char *string2, unsigned int string2_len) unsigned char *string2, unsigned string2_len)
{ {
char last = '0'; char last = '0';
unsigned int i, j; unsigned i, j;
for (j = 0, i = 0; i < string1_len; i++) { for (j = 0, i = 0; i < string1_len; i++) {
if (string2_len <= j) if (string2_len <= j)
@ -39,12 +39,11 @@ static void map(char *pvector,
/* supported constructs: /* supported constructs:
* Ranges, e.g., 0-9 ==> 0123456789 * Ranges, e.g., 0-9 ==> 0123456789
* Ranges, e.g., [0-9] ==> 0123456789
* Escapes, e.g., \a ==> Control-G * Escapes, e.g., \a ==> Control-G
* Character classes, e.g. [:upper:] ==> A...Z * Character classes, e.g. [:upper:] ==> A...Z
* Equiv classess, e.g. [=A=] ==> A (hmmmmmmm?) * Equiv classess, e.g. [=A=] ==> A (hmmmmmmm?)
*/ */
static unsigned int expand(const char *arg, char *buffer) static unsigned expand(const char *arg, char *buffer)
{ {
char *buffer_start = buffer; char *buffer_start = buffer;
unsigned i; /* can't be unsigned char: must be able to hold 256 */ unsigned i; /* can't be unsigned char: must be able to hold 256 */
@ -77,7 +76,8 @@ static unsigned int expand(const char *arg, char *buffer)
static const char classes[] ALIGN1 = static const char classes[] ALIGN1 =
"alpha"CLO "alnum"CLO "digit"CLO "alpha"CLO "alnum"CLO "digit"CLO
"lower"CLO "upper"CLO "space"CLO "lower"CLO "upper"CLO "space"CLO
"blank"CLO "punct"CLO "cntrl"CLO; "blank"CLO "punct"CLO "cntrl"CLO
"xdigit"CLO;
#define CLASS_invalid 0 /* we increment the retval */ #define CLASS_invalid 0 /* we increment the retval */
#define CLASS_alpha 1 #define CLASS_alpha 1
#define CLASS_alnum 2 #define CLASS_alnum 2
@ -88,16 +88,17 @@ static unsigned int expand(const char *arg, char *buffer)
#define CLASS_blank 7 #define CLASS_blank 7
#define CLASS_punct 8 #define CLASS_punct 8
#define CLASS_cntrl 9 #define CLASS_cntrl 9
//#define CLASS_xdigit 10 #define CLASS_xdigit 10
//#define CLASS_graph 11 //#define CLASS_graph 11
//#define CLASS_print 12 //#define CLASS_print 12
smalluint j; smalluint j;
{ /* not really pretty.. */ {
char *tmp = xstrndup(arg, 7); // warning: xdigit would need 8, not 7 /* xdigit needs 8, not 7 */
char *tmp = xstrndup(arg, 7 + (arg[0]=='x'));
j = index_in_strings(classes, tmp) + 1; j = index_in_strings(classes, tmp) + 1;
free(tmp); free(tmp);
} }
if (j == CLASS_alnum || j == CLASS_digit) { if (j == CLASS_alnum || j == CLASS_digit || j == CLASS_xdigit) {
for (i = '0'; i <= '9'; i++) for (i = '0'; i <= '9'; i++)
*buffer++ = i; *buffer++ = i;
} }
@ -125,6 +126,12 @@ static unsigned int expand(const char *arg, char *buffer)
|| (j == CLASS_cntrl && iscntrl(i))) || (j == CLASS_cntrl && iscntrl(i)))
*buffer++ = i; *buffer++ = i;
} }
if (j == CLASS_xdigit) {
for (i = 'A'; i <= 'F'; i++) {
*buffer++ = i;
*buffer++ = i | 0x20;
}
}
if (j == CLASS_invalid) { if (j == CLASS_invalid) {
*buffer++ = '['; *buffer++ = '[';
*buffer++ = ':'; *buffer++ = ':';
@ -140,19 +147,14 @@ static unsigned int expand(const char *arg, char *buffer)
arg += 3; /* skip CHAR=] */ arg += 3; /* skip CHAR=] */
continue; continue;
} }
if (i == '\0' || *arg != '-') { /* not [x-...] - copy verbatim */ /* The rest of [xyz... cases is treated as normal
* string, '[' has no special meaning here:
* tr "[a-z]" "[A-Z]" can be written as tr "a-z" "A-Z",
* also try tr "[a-z]" "_A-Z+" and you'll see that
* [] is not special here.
*/
*buffer++ = '['; *buffer++ = '[';
arg--; /* points to x */ arg--; /* points to x */
continue; /* copy all, including eventual ']' */
}
/* [x-z] */
arg++; /* skip - */
if (arg[0] == '\0' || arg[1] != ']')
bb_show_usage();
ac = *arg++;
while (i <= ac)
*buffer++ = i++;
arg++; /* skip ] */
continue; continue;
} }
*buffer++ = *arg++; *buffer++ = *arg++;
@ -162,29 +164,30 @@ static unsigned int expand(const char *arg, char *buffer)
static int complement(char *buffer, int buffer_len) static int complement(char *buffer, int buffer_len)
{ {
int i, j, ix; int ch, j, len;
char conv[ASCII + 2]; char conv[ASCII + 2];
ix = 0; len = 0;
for (i = '\0'; i <= ASCII; i++) { for (ch = '\0'; ch <= ASCII; ch++) {
for (j = 0; j < buffer_len; j++) for (j = 0; j < buffer_len; j++)
if (buffer[j] == i) if (buffer[j] == ch)
break; goto next_ch;
if (j == buffer_len) /* Didn't find it */
conv[ix++] = i & ASCII; conv[len++] = (char) ch;
next_ch:
continue;
} }
memcpy(buffer, conv, ix); memcpy(buffer, conv, len);
return ix; return len;
} }
int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int tr_main(int argc UNUSED_PARAM, char **argv) int tr_main(int argc UNUSED_PARAM, char **argv)
{ {
int output_length = 0, input_length;
int i; int i;
smalluint flags; smalluint flags;
ssize_t read_chars = 0; ssize_t read_chars;
size_t in_index = 0, out_index = 0; size_t in_index, out_index;
unsigned last = UCHAR_MAX + 1; /* not equal to any char */ unsigned last = UCHAR_MAX + 1; /* not equal to any char */
unsigned char coded, c; unsigned char coded, c;
unsigned char *output = xmalloc(BUFSIZ); unsigned char *output = xmalloc(BUFSIZ);
@ -206,6 +209,9 @@ int tr_main(int argc UNUSED_PARAM, char **argv)
#define tr_buf bb_common_bufsiz1 #define tr_buf bb_common_bufsiz1
if (*argv != NULL) { if (*argv != NULL) {
int output_length = 0;
int input_length;
input_length = expand(*argv++, tr_buf); input_length = expand(*argv++, tr_buf);
if (flags & TR_OPT_complement) if (flags & TR_OPT_complement)
input_length = complement(tr_buf, input_length); input_length = complement(tr_buf, input_length);
@ -221,30 +227,33 @@ int tr_main(int argc UNUSED_PARAM, char **argv)
outvec[output[i]] = TRUE; outvec[output[i]] = TRUE;
} }
goto start_from;
for (;;) { for (;;) {
/* If we're out of input, flush output and read more input. */ /* If we're out of input, flush output and read more input. */
if ((ssize_t)in_index == read_chars) { if ((ssize_t)in_index == read_chars) {
if (out_index) { if (out_index) {
xwrite(STDOUT_FILENO, (char *)output, out_index); xwrite(STDOUT_FILENO, (char *)output, out_index);
start_from:
out_index = 0; out_index = 0;
} }
read_chars = safe_read(STDIN_FILENO, tr_buf, BUFSIZ); read_chars = safe_read(STDIN_FILENO, tr_buf, BUFSIZ);
if (read_chars <= 0) { if (read_chars <= 0) {
if (read_chars < 0) if (read_chars < 0)
bb_perror_msg_and_die(bb_msg_read_error); bb_perror_msg_and_die(bb_msg_read_error);
exit(EXIT_SUCCESS); break;
} }
in_index = 0; in_index = 0;
} }
c = tr_buf[in_index++]; c = tr_buf[in_index++];
coded = vector[c];
if ((flags & TR_OPT_delete) && invec[c]) if ((flags & TR_OPT_delete) && invec[c])
continue; continue;
coded = vector[c];
if ((flags & TR_OPT_squeeze_reps) && last == coded if ((flags & TR_OPT_squeeze_reps) && last == coded
&& (invec[c] || outvec[coded])) && (invec[c] || outvec[coded]))
continue; continue;
output[out_index++] = last = coded; output[out_index++] = last = coded;
} }
/* NOTREACHED */
return EXIT_SUCCESS; return EXIT_SUCCESS;
} }

22
testsuite/tr.tests Normal file
View File

@ -0,0 +1,22 @@
#!/bin/sh
# Copyright 2009 by Denys Vlasenko <vda.linux@googlemail.com>
# Licensed under GPL v2, see file LICENSE for details.
. testing.sh
# testing "description" "arguments" "result" "infile" "stdin"
testing "tr does not treat [] in [a-z] as special" \
"tr '[q-z]' '_Q-Z+'" \
"_QWe+" "" "[qwe]"
testing "tr understands 0-9A-F" \
"tr -cd '[0-9A-F]'" \
"19AF" "" "19AFH\n"
testing "tr understands [:xdigit:]" \
"tr -cd '[:xdigit:]'" \
"19AF" "" "19AFH\n"
exit $FAILCOUNT