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

tr: fix overflow in expand and complement, fix stop after [:class:],

Browse Source 
fix order of [:xdigit:], make tr require one param at least.

function                                             old     new   delta
expand                                              1743    1849    +106
tr_main                                              463     518     +55
complement                                            72      75      +3
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 3/0 up/down: 164/0)             Total: 164 bytes
This commit is contained in:
Denis Vlasenko 2009-03-02 12:05:57 +00:00
parent 1bfcc8b08e
commit 491d6e3b56
1 changed files with 131 additions and 104 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

235
coreutils/tr.c
View File

@ -16,24 +16,31 @@
* 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.
*/ */
/* http://www.opengroup.org/onlinepubs/009695399/utilities/tr.html /* http://www.opengroup.org/onlinepubs/009695399/utilities/tr.html
* TODO: xdigit, graph, print * TODO: graph, print
*/ */
#include "libbb.h" #include "libbb.h"
#define ASCII 0377 enum {
ASCII = 256,
/* string buffer needs to be at least as big as the whole "alphabet".
* BUFSIZ == ASCII is ok, but we will realloc in expand
* even for smallest patterns, let's avoid that by using *2:
*/
TR_BUFSIZ = (BUFSIZ > ASCII*2) ? BUFSIZ : ASCII*2,
};
static void map(char *pvector, static void map(char *pvector,
unsigned char *string1, unsigned string1_len, char *string1, unsigned string1_len,
unsigned char *string2, unsigned string2_len) char *string2, unsigned string2_len)
{ {
char last = '0'; char last = '0';
unsigned 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)
pvector[string1[i]] = last; pvector[(unsigned char)(string1[i])] = last;
else else
pvector[string1[i]] = last = string2[j++]; pvector[(unsigned char)(string1[i])] = last = string2[j++];
} }
} }
@ -43,139 +50,155 @@ static void map(char *pvector,
* 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 expand(const char *arg, char *buffer) static unsigned expand(const char *arg, char **buffer_p)
{ {
char *buffer_start = buffer; char *buffer = *buffer_p;
unsigned pos = 0;
unsigned size = TR_BUFSIZ;
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 */
unsigned char ac; unsigned char ac;
while (*arg) { while (*arg) {
if (pos + ASCII > size) {
size += ASCII;
*buffer_p = buffer = xrealloc(buffer, size);
}
if (*arg == '\\') { if (*arg == '\\') {
arg++; arg++;
*buffer++ = bb_process_escape_sequence(&arg); buffer[pos++] = bb_process_escape_sequence(&arg);
continue; continue;
} }
if (arg[1] == '-') { /* "0-9..." */ if (arg[1] == '-') { /* "0-9..." */
ac = arg[2]; ac = arg[2];
if (ac == '\0') { /* "0-": copy verbatim */ if (ac == '\0') { /* "0-": copy verbatim */
*buffer++ = *arg++; /* copy '0' */ buffer[pos++] = *arg++; /* copy '0' */
continue; /* next iter will copy '-' and stop */ continue; /* next iter will copy '-' and stop */
} }
i = *arg; i = (unsigned char) *arg;
while (i <= ac) /* ok: i is unsigned _int_ */ while (i <= ac) /* ok: i is unsigned _int_ */
*buffer++ = i++; buffer[pos++] = i++;
arg += 3; /* skip 0-9 */ arg += 3; /* skip 0-9 */
continue; continue;
} }
if (*arg == '[') { /* "[xyz..." */ if ((ENABLE_FEATURE_TR_CLASSES || ENABLE_FEATURE_TR_EQUIV)
&& *arg == '['
) {
arg++; arg++;
i = *arg++; i = (unsigned char) *arg++;
/* "[xyz...", i=x, arg points to y */ /* "[xyz...". i=x, arg points to y */
if (ENABLE_FEATURE_TR_CLASSES && i == ':') { if (ENABLE_FEATURE_TR_CLASSES && i == ':') { /* [:class:] */
#define CLO ":]\0" #define CLO ":]\0"
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; "xdigit"CLO;
#define CLASS_invalid 0 /* we increment the retval */ enum {
#define CLASS_alpha 1 CLASS_invalid = 0, /* we increment the retval */
#define CLASS_alnum 2 CLASS_alpha = 1,
#define CLASS_digit 3 CLASS_alnum = 2,
#define CLASS_lower 4 CLASS_digit = 3,
#define CLASS_upper 5 CLASS_lower = 4,
#define CLASS_space 6 CLASS_upper = 5,
#define CLASS_blank 7 CLASS_space = 6,
#define CLASS_punct 8 CLASS_blank = 7,
#define CLASS_cntrl 9 CLASS_punct = 8,
#define CLASS_xdigit 10 CLASS_cntrl = 9,
//#define CLASS_graph 11 CLASS_xdigit = 10,
//#define CLASS_print 12 //CLASS_graph = 11,
//CLASS_print = 12,
};
smalluint j; smalluint j;
{ char *tmp;
/* xdigit needs 8, not 7 */
char *tmp = xstrndup(arg, 7 + (arg[0]=='x')); /* xdigit needs 8, not 7 */
j = index_in_strings(classes, tmp) + 1; i = 7 + (arg[0] == 'x');
free(tmp); tmp = xstrndup(arg, i);
} j = index_in_strings(classes, tmp) + 1;
free(tmp);
if (j == CLASS_invalid)
goto skip_bracket;
arg += i;
if (j == CLASS_alnum || j == CLASS_digit || j == CLASS_xdigit) { 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[pos++] = i;
} }
if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) { if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) {
for (i = 'A'; i <= 'Z'; i++) for (i = 'A'; i <= 'Z'; i++)
*buffer++ = i; buffer[pos++] = i;
} }
if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) { if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) {
for (i = 'a'; i <= 'z'; i++) for (i = 'a'; i <= 'z'; i++)
*buffer++ = i; buffer[pos++] = i;
} }
if (j == CLASS_space || j == CLASS_blank) { if (j == CLASS_space || j == CLASS_blank) {
*buffer++ = '\t'; buffer[pos++] = '\t';
if (j == CLASS_space) { if (j == CLASS_space) {
*buffer++ = '\n'; buffer[pos++] = '\n';
*buffer++ = '\v'; buffer[pos++] = '\v';
*buffer++ = '\f'; buffer[pos++] = '\f';
*buffer++ = '\r'; buffer[pos++] = '\r';
} }
*buffer++ = ' '; buffer[pos++] = ' ';
} }
if (j == CLASS_punct || j == CLASS_cntrl) { if (j == CLASS_punct || j == CLASS_cntrl) {
for (i = '\0'; i <= ASCII; i++) for (i = '\0'; i < ASCII; i++) {
if ((j == CLASS_punct && isprint(i) && !isalnum(i) && !isspace(i)) if ((j == CLASS_punct && isprint(i) && !isalnum(i) && !isspace(i))
|| (j == CLASS_cntrl && iscntrl(i))) || (j == CLASS_cntrl && iscntrl(i))
*buffer++ = i; ) {
buffer[pos++] = i;
}
}
} }
if (j == CLASS_xdigit) { if (j == CLASS_xdigit) {
for (i = 'A'; i <= 'F'; i++) { for (i = 'A'; i <= 'F'; i++) {
*buffer++ = i; buffer[pos + 6] = i | 0x20;
*buffer++ = i | 0x20; buffer[pos++] = i;
} }
pos += 6;
} }
if (j == CLASS_invalid) { continue;
*buffer++ = '[';
*buffer++ = ':';
continue;
}
break;
} }
/* "[xyz...", i=x, arg points to y */ /* "[xyz...", i=x, arg points to y */
if (ENABLE_FEATURE_TR_EQUIV && i == '=') { /* [=CHAR=] */ if (ENABLE_FEATURE_TR_EQUIV && i == '=') { /* [=CHAR=] */
*buffer++ = *arg; /* copy CHAR */ buffer[pos++] = *arg; /* copy CHAR */
if (!*arg || arg[1] != '=' || arg[2] != ']') if (!arg[0] || arg[1] != '=' || arg[2] != ']')
bb_show_usage(); bb_show_usage();
arg += 3; /* skip CHAR=] */ arg += 3; /* skip CHAR=] */
continue; continue;
} }
/* The rest of [xyz... cases is treated as normal /* The rest of "[xyz..." cases is treated as normal
* string, '[' has no special meaning here: * string, "[" has no special meaning here:
* tr "[a-z]" "[A-Z]" can be written as tr "a-z" "A-Z", * 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 * also try tr "[a-z]" "_A-Z+" and you'll see that
* [] is not special here. * [] is not special here.
*/ */
*buffer++ = '['; skip_bracket:
arg--; /* points to x */ arg -= 2; /* points to "[" in "[xyz..." */
continue;
} }
*buffer++ = *arg++; buffer[pos++] = *arg++;
} }
return (buffer - buffer_start); return pos;
} }
/* NB: buffer is guaranteed to be at least TR_BUFSIZE
* (which is >= ASCII) big.
*/
static int complement(char *buffer, int buffer_len) static int complement(char *buffer, int buffer_len)
{ {
int ch, j, len; int len;
char conv[ASCII + 2]; char conv[ASCII];
unsigned char ch;
len = 0; len = 0;
for (ch = '\0'; ch <= ASCII; ch++) { ch = '\0';
for (j = 0; j < buffer_len; j++) while (1) {
if (buffer[j] == ch) if (memchr(buffer, ch, buffer_len) == NULL)
goto next_ch; conv[len++] = ch;
/* Didn't find it */ if (++ch == '\0')
conv[len++] = (char) ch; break;
next_ch:
continue;
} }
memcpy(buffer, conv, len); memcpy(buffer, conv, len);
return len; return len;
@ -190,54 +213,56 @@ int tr_main(int argc UNUSED_PARAM, char **argv)
size_t in_index, out_index; 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); char *str1 = xmalloc(TR_BUFSIZ);
char *vector = xzalloc((ASCII+1) * 3); char *str2 = xmalloc(TR_BUFSIZ);
char *invec = vector + (ASCII+1); int str2_length;
char *outvec = vector + (ASCII+1) * 2; int str1_length;
char *vector = xzalloc(ASCII * 3);
char *invec = vector + ASCII;
char *outvec = vector + ASCII * 2;
#define TR_OPT_complement (1 << 0) #define TR_OPT_complement (1 << 0)
#define TR_OPT_delete (1 << 1) #define TR_OPT_delete (1 << 1)
#define TR_OPT_squeeze_reps (1 << 2) #define TR_OPT_squeeze_reps (1 << 2)
flags = getopt32(argv, "+cds"); /* '+': stop at first non-option */ for (i = 0; i < ASCII; i++) {
argv += optind;
for (i = 0; i <= ASCII; i++) {
vector[i] = i; vector[i] = i;
/*invec[i] = outvec[i] = FALSE; - done by xzalloc */ /*invec[i] = outvec[i] = FALSE; - done by xzalloc */
} }
#define tr_buf bb_common_bufsiz1 opt_complementary = "-1";
if (*argv != NULL) { flags = getopt32(argv, "+cds"); /* '+': stop at first non-option */
int output_length = 0; argv += optind;
int input_length;
input_length = expand(*argv++, tr_buf); str1_length = expand(*argv++, &str1);
if (flags & TR_OPT_complement) str2_length = 0;
input_length = complement(tr_buf, input_length); if (flags & TR_OPT_complement)
if (*argv) { str1_length = complement(str1, str1_length);
if (argv[0][0] == '\0') if (*argv) {
bb_error_msg_and_die("STRING2 cannot be empty"); if (argv[0][0] == '\0')
output_length = expand(*argv, (char *)output); bb_error_msg_and_die("STRING2 cannot be empty");
map(vector, (unsigned char *)tr_buf, input_length, output, output_length); str2_length = expand(*argv, &str2);
} map(vector, str1, str1_length,
for (i = 0; i < input_length; i++) str2, str2_length);
invec[(unsigned char)tr_buf[i]] = TRUE;
for (i = 0; i < output_length; i++)
outvec[output[i]] = TRUE;
} }
for (i = 0; i < str1_length; i++)
invec[(unsigned char)(str1[i])] = TRUE;
for (i = 0; i < str2_length; i++)
outvec[(unsigned char)(str2[i])] = TRUE;
goto start_from; goto start_from;
/* In this loop, str1 space is reused as input buffer,
* str2 - as output one. */
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, str2, out_index);
start_from: start_from:
out_index = 0; out_index = 0;
} }
read_chars = safe_read(STDIN_FILENO, tr_buf, BUFSIZ); read_chars = safe_read(STDIN_FILENO, str1, TR_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);
@ -245,14 +270,16 @@ int tr_main(int argc UNUSED_PARAM, char **argv)
} }
in_index = 0; in_index = 0;
} }
c = tr_buf[in_index++]; c = str1[in_index++];
if ((flags & TR_OPT_delete) && invec[c]) if ((flags & TR_OPT_delete) && invec[c])
continue; continue;
coded = vector[c]; 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; }
str2[out_index++] = last = coded;
} }
return EXIT_SUCCESS; return EXIT_SUCCESS;