busybox/coreutils/cut.c

320 lines
8.7 KiB
C
Raw Normal View History

/* vi: set sw=4 ts=4: */
/*
* cut.c - minimalist version of cut
*
* Copyright (C) 1999,2000,2001 by Lineo, inc.
* Written by Mark Whitley <markw@codepoet.org>
* debloated by Bernhard Reutner-Fischer
*
* Licensed under GPLv2 or later, see file LICENSE in this source tree.
*/
//config:config CUT
//config: bool "cut (5.8 kb)"
//config: default y
//config: help
//config: cut is used to print selected parts of lines from
//config: each file to stdout.
//applet:IF_CUT(APPLET_NOEXEC(cut, cut, BB_DIR_USR_BIN, BB_SUID_DROP, cut))
//kbuild:lib-$(CONFIG_CUT) += cut.o
//usage:#define cut_trivial_usage
//usage: "[OPTIONS] [FILE]..."
//usage:#define cut_full_usage "\n\n"
//usage: "Print selected fields from each input FILE to stdout\n"
//usage: "\n -b LIST Output only bytes from LIST"
//usage: "\n -c LIST Output only characters from LIST"
//usage: "\n -d CHAR Use CHAR instead of tab as the field delimiter"
//usage: "\n -s Output only the lines containing delimiter"
//usage: "\n -f N Print only these fields"
//usage: "\n -n Ignored"
//usage:
//usage:#define cut_example_usage
//usage: "$ echo \"Hello world\" | cut -f 1 -d ' '\n"
//usage: "Hello\n"
//usage: "$ echo \"Hello world\" | cut -f 2 -d ' '\n"
//usage: "world\n"
#include "libbb.h"
/* This is a NOEXEC applet. Be very careful! */
/* option vars */
getopt32: remove opt_complementary function old new delta vgetopt32 1318 1392 +74 runsvdir_main 703 713 +10 bb_make_directory 423 425 +2 collect_cpu 546 545 -1 opt_chars 3 - -3 opt_complementary 4 - -4 tftpd_main 567 562 -5 ntp_init 476 471 -5 zcip_main 1266 1256 -10 xxd_main 428 418 -10 whois_main 140 130 -10 who_main 463 453 -10 which_main 212 202 -10 wget_main 2535 2525 -10 watchdog_main 291 281 -10 watch_main 222 212 -10 vlock_main 399 389 -10 uuencode_main 332 322 -10 uudecode_main 316 306 -10 unlink_main 45 35 -10 udhcpd_main 1482 1472 -10 udhcpc_main 2762 2752 -10 tune2fs_main 290 280 -10 tunctl_main 366 356 -10 truncate_main 218 208 -10 tr_main 518 508 -10 time_main 1134 1124 -10 tftp_main 286 276 -10 telnetd_main 1873 1863 -10 tcpudpsvd_main 1785 1775 -10 taskset_main 521 511 -10 tar_main 1009 999 -10 tail_main 1644 1634 -10 syslogd_main 1967 1957 -10 switch_root_main 368 358 -10 svlogd_main 1454 1444 -10 sv 1296 1286 -10 stat_main 104 94 -10 start_stop_daemon_main 1028 1018 -10 split_main 542 532 -10 sort_main 796 786 -10 slattach_main 624 614 -10 shuf_main 504 494 -10 setsid_main 96 86 -10 setserial_main 1132 1122 -10 setfont_main 388 378 -10 setconsole_main 78 68 -10 sendmail_main 1209 1199 -10 sed_main 677 667 -10 script_main 1077 1067 -10 run_parts_main 325 315 -10 rtcwake_main 454 444 -10 rm_main 175 165 -10 reformime_main 119 109 -10 readlink_main 123 113 -10 rdate_main 246 236 -10 pwdx_main 189 179 -10 pstree_main 317 307 -10 pscan_main 663 653 -10 popmaildir_main 818 808 -10 pmap_main 80 70 -10 nc_main 1042 1032 -10 mv_main 558 548 -10 mountpoint_main 477 467 -10 mount_main 1264 1254 -10 modprobe_main 768 758 -10 modinfo_main 333 323 -10 mktemp_main 200 190 -10 mkswap_main 324 314 -10 mkfs_vfat_main 1489 1479 -10 microcom_main 715 705 -10 md5_sha1_sum_main 521 511 -10 man_main 867 857 -10 makedevs_main 1052 1042 -10 ls_main 563 553 -10 losetup_main 432 422 -10 loadfont_main 89 79 -10 ln_main 524 514 -10 link_main 75 65 -10 ipcalc_main 544 534 -10 iostat_main 2397 2387 -10 install_main 768 758 -10 id_main 480 470 -10 i2cset_main 1239 1229 -10 i2cget_main 380 370 -10 i2cdump_main 1482 1472 -10 i2cdetect_main 682 672 -10 hwclock_main 406 396 -10 httpd_main 741 731 -10 grep_main 837 827 -10 getty_main 1559 1549 -10 fuser_main 297 287 -10 ftpgetput_main 345 335 -10 ftpd_main 2232 2222 -10 fstrim_main 251 241 -10 fsfreeze_main 77 67 -10 fsck_minix_main 2921 2911 -10 flock_main 314 304 -10 flashcp_main 740 730 -10 flash_eraseall_main 833 823 -10 fdformat_main 532 522 -10 expand_main 680 670 -10 eject_main 335 325 -10 dumpleases_main 630 620 -10 du_main 314 304 -10 dos2unix_main 441 431 -10 diff_main 1350 1340 -10 df_main 1064 1054 -10 date_main 1095 1085 -10 cut_main 961 951 -10 cryptpw_main 228 218 -10 crontab_main 575 565 -10 crond_main 1149 1139 -10 cp_main 370 360 -10 common_traceroute_main 3834 3824 -10 common_ping_main 1767 1757 -10 comm_main 239 229 -10 cmp_main 655 645 -10 chrt_main 379 369 -10 chpst_main 704 694 -10 chpasswd_main 308 298 -10 chown_main 171 161 -10 chmod_main 158 148 -10 cat_main 428 418 -10 bzip2_main 120 110 -10 blkdiscard_main 264 254 -10 base64_main 221 211 -10 arping_main 1665 1655 -10 ar_main 556 546 -10 adjtimex_main 406 396 -10 adduser_main 882 872 -10 addgroup_main 411 401 -10 acpid_main 1198 1188 -10 optstring 11 - -11 opt_string 18 - -18 OPT_STR 25 - -25 ubi_tools_main 1288 1258 -30 ls_options 31 - -31 ------------------------------------------------------------------------------ (add/remove: 0/6 grow/shrink: 3/129 up/down: 86/-1383) Total: -1297 bytes text data bss dec hex filename 915428 485 6876 922789 e14a5 busybox_old 914629 485 6872 921986 e1182 busybox_unstripped Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2017-08-09 01:25:02 +05:30
#define OPT_STR "b:c:f:d:sn"
#define CUT_OPT_BYTE_FLGS (1 << 0)
#define CUT_OPT_CHAR_FLGS (1 << 1)
#define CUT_OPT_FIELDS_FLGS (1 << 2)
#define CUT_OPT_DELIM_FLGS (1 << 3)
#define CUT_OPT_SUPPRESS_FLGS (1 << 4)
struct cut_list {
int startpos;
int endpos;
};
enum {
BOL = 0,
EOL = INT_MAX,
NON_RANGE = -1
};
static int cmpfunc(const void *a, const void *b)
{
return (((struct cut_list *) a)->startpos -
((struct cut_list *) b)->startpos);
}
static void cut_file(FILE *file, char delim, const struct cut_list *cut_lists, unsigned nlists)
{
char *line;
unsigned linenum = 0; /* keep these zero-based to be consistent */
/* go through every line in the file */
while ((line = xmalloc_fgetline(file)) != NULL) {
/* set up a list so we can keep track of what's been printed */
int linelen = strlen(line);
2008-08-16 02:50:23 +05:30
char *printed = xzalloc(linelen + 1);
char *orig_line = line;
unsigned cl_pos = 0;
int spos;
/* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
2006-10-13 04:12:33 +05:30
if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
/* print the chars specified in each cut list */
for (; cl_pos < nlists; cl_pos++) {
spos = cut_lists[cl_pos].startpos;
while (spos < linelen) {
if (!printed[spos]) {
printed[spos] = 'X';
putchar(line[spos]);
}
spos++;
if (spos > cut_lists[cl_pos].endpos
/* NON_RANGE is -1, so if below is true,
* the above was true too (spos is >= 0) */
/* || cut_lists[cl_pos].endpos == NON_RANGE */
) {
break;
}
}
}
} else if (delim == '\n') { /* cut by lines */
spos = cut_lists[cl_pos].startpos;
/* get out if we have no more lists to process or if the lines
* are lower than what we're interested in */
if (((int)linenum < spos) || (cl_pos >= nlists))
goto next_line;
/* if the line we're looking for is lower than the one we were
* passed, it means we displayed it already, so move on */
while (spos < (int)linenum) {
spos++;
/* go to the next list if we're at the end of this one */
if (spos > cut_lists[cl_pos].endpos
|| cut_lists[cl_pos].endpos == NON_RANGE
) {
cl_pos++;
/* get out if there's no more lists to process */
if (cl_pos >= nlists)
goto next_line;
spos = cut_lists[cl_pos].startpos;
/* get out if the current line is lower than the one
* we just became interested in */
if ((int)linenum < spos)
goto next_line;
}
}
/* If we made it here, it means we've found the line we're
* looking for, so print it */
puts(line);
goto next_line;
} else { /* cut by fields */
int ndelim = -1; /* zero-based / one-based problem */
int nfields_printed = 0;
char *field = NULL;
char delimiter[2];
delimiter[0] = delim;
delimiter[1] = 0;
/* does this line contain any delimiters? */
if (strchr(line, delim) == NULL) {
2006-10-13 04:12:33 +05:30
if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
puts(line);
goto next_line;
}
/* process each list on this line, for as long as we've got
* a line to process */
for (; cl_pos < nlists && line; cl_pos++) {
spos = cut_lists[cl_pos].startpos;
do {
/* find the field we're looking for */
while (line && ndelim < spos) {
field = strsep(&line, delimiter);
ndelim++;
}
/* we found it, and it hasn't been printed yet */
if (field && ndelim == spos && !printed[ndelim]) {
/* if this isn't our first time through, we need to
* print the delimiter after the last field that was
* printed */
if (nfields_printed > 0)
putchar(delim);
fputs(field, stdout);
printed[ndelim] = 'X';
nfields_printed++; /* shouldn't overflow.. */
}
spos++;
/* keep going as long as we have a line to work with,
* this is a list, and we're not at the end of that
* list */
} while (spos <= cut_lists[cl_pos].endpos && line
&& cut_lists[cl_pos].endpos != NON_RANGE);
}
}
/* if we printed anything at all, we need to finish it with a
* newline cuz we were handed a chomped line */
putchar('\n');
2006-10-13 04:12:33 +05:30
next_line:
linenum++;
free(printed);
free(orig_line);
}
}
int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
2008-07-05 14:48:54 +05:30
int cut_main(int argc UNUSED_PARAM, char **argv)
{
/* growable array holding a series of lists */
struct cut_list *cut_lists = NULL;
unsigned nlists = 0; /* number of elements in above list */
char delim = '\t'; /* delimiter, default is tab */
char *sopt, *ltok;
unsigned opt;
getopt32: remove opt_complementary function old new delta vgetopt32 1318 1392 +74 runsvdir_main 703 713 +10 bb_make_directory 423 425 +2 collect_cpu 546 545 -1 opt_chars 3 - -3 opt_complementary 4 - -4 tftpd_main 567 562 -5 ntp_init 476 471 -5 zcip_main 1266 1256 -10 xxd_main 428 418 -10 whois_main 140 130 -10 who_main 463 453 -10 which_main 212 202 -10 wget_main 2535 2525 -10 watchdog_main 291 281 -10 watch_main 222 212 -10 vlock_main 399 389 -10 uuencode_main 332 322 -10 uudecode_main 316 306 -10 unlink_main 45 35 -10 udhcpd_main 1482 1472 -10 udhcpc_main 2762 2752 -10 tune2fs_main 290 280 -10 tunctl_main 366 356 -10 truncate_main 218 208 -10 tr_main 518 508 -10 time_main 1134 1124 -10 tftp_main 286 276 -10 telnetd_main 1873 1863 -10 tcpudpsvd_main 1785 1775 -10 taskset_main 521 511 -10 tar_main 1009 999 -10 tail_main 1644 1634 -10 syslogd_main 1967 1957 -10 switch_root_main 368 358 -10 svlogd_main 1454 1444 -10 sv 1296 1286 -10 stat_main 104 94 -10 start_stop_daemon_main 1028 1018 -10 split_main 542 532 -10 sort_main 796 786 -10 slattach_main 624 614 -10 shuf_main 504 494 -10 setsid_main 96 86 -10 setserial_main 1132 1122 -10 setfont_main 388 378 -10 setconsole_main 78 68 -10 sendmail_main 1209 1199 -10 sed_main 677 667 -10 script_main 1077 1067 -10 run_parts_main 325 315 -10 rtcwake_main 454 444 -10 rm_main 175 165 -10 reformime_main 119 109 -10 readlink_main 123 113 -10 rdate_main 246 236 -10 pwdx_main 189 179 -10 pstree_main 317 307 -10 pscan_main 663 653 -10 popmaildir_main 818 808 -10 pmap_main 80 70 -10 nc_main 1042 1032 -10 mv_main 558 548 -10 mountpoint_main 477 467 -10 mount_main 1264 1254 -10 modprobe_main 768 758 -10 modinfo_main 333 323 -10 mktemp_main 200 190 -10 mkswap_main 324 314 -10 mkfs_vfat_main 1489 1479 -10 microcom_main 715 705 -10 md5_sha1_sum_main 521 511 -10 man_main 867 857 -10 makedevs_main 1052 1042 -10 ls_main 563 553 -10 losetup_main 432 422 -10 loadfont_main 89 79 -10 ln_main 524 514 -10 link_main 75 65 -10 ipcalc_main 544 534 -10 iostat_main 2397 2387 -10 install_main 768 758 -10 id_main 480 470 -10 i2cset_main 1239 1229 -10 i2cget_main 380 370 -10 i2cdump_main 1482 1472 -10 i2cdetect_main 682 672 -10 hwclock_main 406 396 -10 httpd_main 741 731 -10 grep_main 837 827 -10 getty_main 1559 1549 -10 fuser_main 297 287 -10 ftpgetput_main 345 335 -10 ftpd_main 2232 2222 -10 fstrim_main 251 241 -10 fsfreeze_main 77 67 -10 fsck_minix_main 2921 2911 -10 flock_main 314 304 -10 flashcp_main 740 730 -10 flash_eraseall_main 833 823 -10 fdformat_main 532 522 -10 expand_main 680 670 -10 eject_main 335 325 -10 dumpleases_main 630 620 -10 du_main 314 304 -10 dos2unix_main 441 431 -10 diff_main 1350 1340 -10 df_main 1064 1054 -10 date_main 1095 1085 -10 cut_main 961 951 -10 cryptpw_main 228 218 -10 crontab_main 575 565 -10 crond_main 1149 1139 -10 cp_main 370 360 -10 common_traceroute_main 3834 3824 -10 common_ping_main 1767 1757 -10 comm_main 239 229 -10 cmp_main 655 645 -10 chrt_main 379 369 -10 chpst_main 704 694 -10 chpasswd_main 308 298 -10 chown_main 171 161 -10 chmod_main 158 148 -10 cat_main 428 418 -10 bzip2_main 120 110 -10 blkdiscard_main 264 254 -10 base64_main 221 211 -10 arping_main 1665 1655 -10 ar_main 556 546 -10 adjtimex_main 406 396 -10 adduser_main 882 872 -10 addgroup_main 411 401 -10 acpid_main 1198 1188 -10 optstring 11 - -11 opt_string 18 - -18 OPT_STR 25 - -25 ubi_tools_main 1288 1258 -30 ls_options 31 - -31 ------------------------------------------------------------------------------ (add/remove: 0/6 grow/shrink: 3/129 up/down: 86/-1383) Total: -1297 bytes text data bss dec hex filename 915428 485 6876 922789 e14a5 busybox_old 914629 485 6872 921986 e1182 busybox_unstripped Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2017-08-09 01:25:02 +05:30
opt = getopt32(argv, "^"
OPT_STR
"\0" "b--bcf:c--bcf:f--bcf",
&sopt, &sopt, &sopt, &ltok
);
// argc -= optind;
argv += optind;
if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
2006-10-13 04:12:33 +05:30
bb_error_msg_and_die("expected a list of bytes, characters, or fields");
if (opt & CUT_OPT_DELIM_FLGS) {
if (ltok[0] && ltok[1]) { /* more than 1 char? */
bb_error_msg_and_die("the delimiter must be a single character");
}
delim = ltok[0];
}
/* non-field (char or byte) cutting has some special handling */
if (!(opt & CUT_OPT_FIELDS_FLGS)) {
static const char _op_on_field[] ALIGN1 = " only when operating on fields";
if (opt & CUT_OPT_SUPPRESS_FLGS) {
bb_error_msg_and_die
("suppressing non-delimited lines makes sense%s",
_op_on_field);
}
if (delim != '\t') {
bb_error_msg_and_die
("a delimiter may be specified%s", _op_on_field);
}
}
/*
* parse list and put values into startpos and endpos.
* valid list formats: N, N-, N-M, -M
* more than one list can be separated by commas
*/
{
char *ntok;
int s = 0, e = 0;
/* take apart the lists, one by one (they are separated with commas) */
while ((ltok = strsep(&sopt, ",")) != NULL) {
/* it's actually legal to pass an empty list */
if (!ltok[0])
continue;
/* get the start pos */
ntok = strsep(&ltok, "-");
if (!ntok[0]) {
s = BOL;
} else {
s = xatoi_positive(ntok);
/* account for the fact that arrays are zero based, while
* the user expects the first char on the line to be char #1 */
if (s != 0)
s--;
}
/* get the end pos */
if (ltok == NULL) {
e = NON_RANGE;
} else if (!ltok[0]) {
e = EOL;
} else {
e = xatoi_positive(ltok);
/* if the user specified and end position of 0,
* that means "til the end of the line" */
if (e == 0)
e = EOL;
e--; /* again, arrays are zero based, lines are 1 based */
if (e == s)
e = NON_RANGE;
}
/* add the new list */
cut_lists = xrealloc_vector(cut_lists, 4, nlists);
/* NB: startpos is always >= 0,
* while endpos may be = NON_RANGE (-1) */
cut_lists[nlists].startpos = s;
cut_lists[nlists].endpos = e;
nlists++;
}
/* make sure we got some cut positions out of all that */
if (nlists == 0)
bb_error_msg_and_die("missing list of positions");
/* now that the lists are parsed, we need to sort them to make life
* easier on us when it comes time to print the chars / fields / lines
*/
qsort(cut_lists, nlists, sizeof(cut_lists[0]), cmpfunc);
}
{
int retval = EXIT_SUCCESS;
if (!*argv)
*--argv = (char *)"-";
do {
FILE *file = fopen_or_warn_stdin(*argv);
if (!file) {
retval = EXIT_FAILURE;
continue;
}
cut_file(file, delim, cut_lists, nlists);
fclose_if_not_stdin(file);
} while (*++argv);
if (ENABLE_FEATURE_CLEAN_UP)
free(cut_lists);
fflush_stdout_and_exit(retval);
}
}