busybox/coreutils/tr.c
James Byrne 6937487be7 libbb: reduce the overhead of single parameter bb_error_msg() calls
Back in 2007, commit 0c97c9d437 ("'simple' error message functions by
Loic Grenie") introduced bb_simple_perror_msg() to allow for a lower
overhead call to bb_perror_msg() when only a string was being printed
with no parameters. This saves space for some CPU architectures because
it avoids the overhead of a call to a variadic function. However there
has never been a simple version of bb_error_msg(), and since 2007 many
new calls to bb_perror_msg() have been added that only take a single
parameter and so could have been using bb_simple_perror_message().

This changeset introduces 'simple' versions of bb_info_msg(),
bb_error_msg(), bb_error_msg_and_die(), bb_herror_msg() and
bb_herror_msg_and_die(), and replaces all calls that only take a
single parameter, or use something like ("%s", arg), with calls to the
corresponding 'simple' version.

Since it is likely that single parameter calls to the variadic functions
may be accidentally reintroduced in the future a new debugging config
option WARN_SIMPLE_MSG has been introduced. This uses some macro magic
which will cause any such calls to generate a warning, but this is
turned off by default to avoid use of the unpleasant macros in normal
circumstances.

This is a large changeset due to the number of calls that have been
replaced. The only files that contain changes other than simple
substitution of function calls are libbb.h, libbb/herror_msg.c,
libbb/verror_msg.c and libbb/xfuncs_printf.c. In miscutils/devfsd.c,
networking/udhcp/common.h and util-linux/mdev.c additonal macros have
been added for logging so that single parameter and multiple parameter
logging variants exist.

The amount of space saved varies considerably by architecture, and was
found to be as follows (for 'defconfig' using GCC 7.4):

Arm:     -92 bytes
MIPS:    -52 bytes
PPC:   -1836 bytes
x86_64: -938 bytes

Note that for the MIPS architecture only an exception had to be made
disabling the 'simple' calls for 'udhcp' (in networking/udhcp/common.h)
because it made these files larger on MIPS.

Signed-off-by: James Byrne <james.byrne@origamienergy.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2019-07-02 11:35:03 +02:00

361 lines
9.5 KiB
C

/* vi: set sw=4 ts=4: */
/*
* Mini tr implementation for busybox
*
* Copyright (c) 1987,1997, Prentice Hall All rights reserved.
*
* The name of Prentice Hall may not be used to endorse or promote
* products derived from this software without specific prior
* written permission.
*
* Copyright (c) Michiel Huisjes
*
* This version of tr is adapted from Minix tr and was modified
* by Erik Andersen <andersen@codepoet.org> to be used in busybox.
*
* Licensed under GPLv2 or later, see file LICENSE in this source tree.
*/
/* http://www.opengroup.org/onlinepubs/009695399/utilities/tr.html
* TODO: graph, print
*/
//config:config TR
//config: bool "tr (5.1 kb)"
//config: default y
//config: help
//config: tr is used to squeeze, and/or delete characters from standard
//config: input, writing to standard output.
//config:
//config:config FEATURE_TR_CLASSES
//config: bool "Enable character classes (such as [:upper:])"
//config: default y
//config: depends on TR
//config: help
//config: Enable character classes, enabling commands such as:
//config: tr [:upper:] [:lower:] to convert input into lowercase.
//config:
//config:config FEATURE_TR_EQUIV
//config: bool "Enable equivalence classes"
//config: default y
//config: depends on TR
//config: help
//config: Enable equivalence classes, which essentially add the enclosed
//config: character to the current set. For instance, tr [=a=] xyz would
//config: replace all instances of 'a' with 'xyz'. This option is mainly
//config: useful for cases when no other way of expressing a character
//config: is possible.
//applet:IF_TR(APPLET(tr, BB_DIR_USR_BIN, BB_SUID_DROP))
//kbuild:lib-$(CONFIG_TR) += tr.o
//usage:#define tr_trivial_usage
//usage: "[-cds] STRING1 [STRING2]"
//usage:#define tr_full_usage "\n\n"
//usage: "Translate, squeeze, or delete characters from stdin, writing to stdout\n"
//usage: "\n -c Take complement of STRING1"
//usage: "\n -d Delete input characters coded STRING1"
//usage: "\n -s Squeeze multiple output characters of STRING2 into one character"
//usage:
//usage:#define tr_example_usage
//usage: "$ echo \"gdkkn vnqkc\" | tr [a-y] [b-z]\n"
//usage: "hello world\n"
#include "libbb.h"
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,
char *string1, unsigned string1_len,
char *string2, unsigned string2_len)
{
char last = '0';
unsigned i, j;
for (j = 0, i = 0; i < string1_len; i++) {
if (string2_len <= j)
pvector[(unsigned char)(string1[i])] = last;
else
pvector[(unsigned char)(string1[i])] = last = string2[j++];
}
}
/* supported constructs:
* Ranges, e.g., 0-9 ==> 0123456789
* Escapes, e.g., \a ==> Control-G
* Character classes, e.g. [:upper:] ==> A...Z
* Equiv classess, e.g. [=A=] ==> A (hmmmmmmm?)
* not supported:
* [x*N] - repeat char x N times
* [x*] - repeat char x until it fills STRING2:
* # echo qwe123 | /usr/bin/tr 123456789 '[d]'
* qwe[d]
* # echo qwe123 | /usr/bin/tr 123456789 '[d*]'
* qweddd
*/
static unsigned expand(char *arg, char **buffer_p)
{
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 char ac;
while (*arg) {
if (pos + ASCII > size) {
size += ASCII;
*buffer_p = buffer = xrealloc(buffer, size);
}
if (*arg == '\\') {
const char *z;
arg++;
z = arg;
ac = bb_process_escape_sequence(&z);
arg = (char *)z;
arg--;
*arg = ac;
/*
* fall through, there may be a range.
* If not, current char will be treated anyway.
*/
}
if (arg[1] == '-') { /* "0-9..." */
ac = arg[2];
if (ac == '\0') { /* "0-": copy verbatim */
buffer[pos++] = *arg++; /* copy '0' */
continue; /* next iter will copy '-' and stop */
}
i = (unsigned char) *arg;
arg += 3; /* skip 0-9 or 0-\ */
if (ac == '\\') {
const char *z;
z = arg;
ac = bb_process_escape_sequence(&z);
arg = (char *)z;
}
while (i <= ac) /* ok: i is unsigned _int_ */
buffer[pos++] = i++;
continue;
}
if ((ENABLE_FEATURE_TR_CLASSES || ENABLE_FEATURE_TR_EQUIV)
&& *arg == '['
) {
arg++;
i = (unsigned char) *arg++;
/* "[xyz...". i=x, arg points to y */
if (ENABLE_FEATURE_TR_CLASSES && i == ':') { /* [:class:] */
#define CLO ":]\0"
static const char classes[] ALIGN1 =
"alpha"CLO "alnum"CLO "digit"CLO
"lower"CLO "upper"CLO "space"CLO
"blank"CLO "punct"CLO "cntrl"CLO
"xdigit"CLO;
enum {
CLASS_invalid = 0, /* we increment the retval */
CLASS_alpha = 1,
CLASS_alnum = 2,
CLASS_digit = 3,
CLASS_lower = 4,
CLASS_upper = 5,
CLASS_space = 6,
CLASS_blank = 7,
CLASS_punct = 8,
CLASS_cntrl = 9,
CLASS_xdigit = 10,
//CLASS_graph = 11,
//CLASS_print = 12,
};
smalluint j;
char *tmp;
/* xdigit needs 8, not 7 */
i = 7 + (arg[0] == 'x');
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) {
for (i = '0'; i <= '9'; i++)
buffer[pos++] = i;
}
if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_upper) {
for (i = 'A'; i <= 'Z'; i++)
buffer[pos++] = i;
}
if (j == CLASS_alpha || j == CLASS_alnum || j == CLASS_lower) {
for (i = 'a'; i <= 'z'; i++)
buffer[pos++] = i;
}
if (j == CLASS_space || j == CLASS_blank) {
buffer[pos++] = '\t';
if (j == CLASS_space) {
buffer[pos++] = '\n';
buffer[pos++] = '\v';
buffer[pos++] = '\f';
buffer[pos++] = '\r';
}
buffer[pos++] = ' ';
}
if (j == CLASS_punct || j == CLASS_cntrl) {
for (i = '\0'; i < ASCII; i++) {
if ((j == CLASS_punct && isprint_asciionly(i) && !isalnum(i) && !isspace(i))
|| (j == CLASS_cntrl && iscntrl(i))
) {
buffer[pos++] = i;
}
}
}
if (j == CLASS_xdigit) {
for (i = 'A'; i <= 'F'; i++) {
buffer[pos + 6] = i | 0x20;
buffer[pos++] = i;
}
pos += 6;
}
continue;
}
/* "[xyz...", i=x, arg points to y */
if (ENABLE_FEATURE_TR_EQUIV && i == '=') { /* [=CHAR=] */
buffer[pos++] = *arg; /* copy CHAR */
if (!arg[0] || arg[1] != '=' || arg[2] != ']')
bb_show_usage();
arg += 3; /* skip CHAR=] */
continue;
}
/* 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.
*/
skip_bracket:
arg -= 2; /* points to "[" in "[xyz..." */
}
buffer[pos++] = *arg++;
}
return pos;
}
/* NB: buffer is guaranteed to be at least TR_BUFSIZE
* (which is >= ASCII) big.
*/
static int complement(char *buffer, int buffer_len)
{
int len;
char conv[ASCII];
unsigned char ch;
len = 0;
ch = '\0';
while (1) {
if (memchr(buffer, ch, buffer_len) == NULL)
conv[len++] = ch;
if (++ch == '\0')
break;
}
memcpy(buffer, conv, len);
return len;
}
int tr_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int tr_main(int argc UNUSED_PARAM, char **argv)
{
int i;
smalluint opts;
ssize_t read_chars;
size_t in_index, out_index;
unsigned last = UCHAR_MAX + 1; /* not equal to any char */
unsigned char coded, c;
char *str1 = xmalloc(TR_BUFSIZ);
char *str2 = xmalloc(TR_BUFSIZ);
int str2_length;
int str1_length;
char *vector = xzalloc(ASCII * 3);
char *invec = vector + ASCII;
char *outvec = vector + ASCII * 2;
#define TR_OPT_complement (3 << 0)
#define TR_OPT_delete (1 << 2)
#define TR_OPT_squeeze_reps (1 << 3)
for (i = 0; i < ASCII; i++) {
vector[i] = i;
/*invec[i] = outvec[i] = FALSE; - done by xzalloc */
}
/* -C/-c difference is that -C complements "characters",
* and -c complements "values" (binary bytes I guess).
* In POSIX locale, these are the same.
*/
/* '+': stop at first non-option */
opts = getopt32(argv, "^+" "Ccds" "\0" "-1");
argv += optind;
str1_length = expand(*argv++, &str1);
str2_length = 0;
if (opts & TR_OPT_complement)
str1_length = complement(str1, str1_length);
if (*argv) {
if (argv[0][0] == '\0')
bb_simple_error_msg_and_die("STRING2 cannot be empty");
str2_length = expand(*argv, &str2);
map(vector, str1, str1_length,
str2, str2_length);
}
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;
/* In this loop, str1 space is reused as input buffer,
* str2 - as output one. */
for (;;) {
/* If we're out of input, flush output and read more input. */
if ((ssize_t)in_index == read_chars) {
if (out_index) {
xwrite(STDOUT_FILENO, str2, out_index);
start_from:
out_index = 0;
}
read_chars = safe_read(STDIN_FILENO, str1, TR_BUFSIZ);
if (read_chars <= 0) {
if (read_chars < 0)
bb_simple_perror_msg_and_die(bb_msg_read_error);
break;
}
in_index = 0;
}
c = str1[in_index++];
if ((opts & TR_OPT_delete) && invec[c])
continue;
coded = vector[c];
if ((opts & TR_OPT_squeeze_reps) && last == coded
&& (invec[c] || outvec[coded])
) {
continue;
}
str2[out_index++] = last = coded;
}
if (ENABLE_FEATURE_CLEAN_UP) {
free(vector);
free(str2);
free(str1);
}
return EXIT_SUCCESS;
}