[1765] | 1 | /* vi: set sw=4 ts=4: */
|
---|
[821] | 2 | /*
|
---|
| 3 | * cut.c - minimalist version of cut
|
---|
| 4 | *
|
---|
| 5 | * Copyright (C) 1999,2000,2001 by Lineo, inc.
|
---|
| 6 | * Written by Mark Whitley <markw@codepoet.org>
|
---|
[1765] | 7 | * debloated by Bernhard Fischer
|
---|
[821] | 8 | *
|
---|
[1765] | 9 | * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
|
---|
[821] | 10 | */
|
---|
| 11 |
|
---|
[1765] | 12 | #include "libbb.h"
|
---|
[821] | 13 |
|
---|
[1765] | 14 | /* This is a NOEXEC applet. Be very careful! */
|
---|
[821] | 15 |
|
---|
[1765] | 16 |
|
---|
[821] | 17 | /* option vars */
|
---|
[1765] | 18 | static const char optstring[] ALIGN1 = "b:c:f:d:sn";
|
---|
| 19 | #define CUT_OPT_BYTE_FLGS (1<<0)
|
---|
| 20 | #define CUT_OPT_CHAR_FLGS (1<<1)
|
---|
| 21 | #define CUT_OPT_FIELDS_FLGS (1<<2)
|
---|
| 22 | #define CUT_OPT_DELIM_FLGS (1<<3)
|
---|
| 23 | #define CUT_OPT_SUPPRESS_FLGS (1<<4)
|
---|
[821] | 24 |
|
---|
[1765] | 25 | static char delim = '\t'; /* delimiter, default is tab */
|
---|
| 26 |
|
---|
[821] | 27 | struct cut_list {
|
---|
| 28 | int startpos;
|
---|
| 29 | int endpos;
|
---|
| 30 | };
|
---|
| 31 |
|
---|
| 32 | enum {
|
---|
| 33 | BOL = 0,
|
---|
| 34 | EOL = INT_MAX,
|
---|
| 35 | NON_RANGE = -1
|
---|
| 36 | };
|
---|
| 37 |
|
---|
[1765] | 38 | /* growable array holding a series of lists */
|
---|
| 39 | static struct cut_list *cut_lists;
|
---|
| 40 | static unsigned int nlists; /* number of elements in above list */
|
---|
[821] | 41 |
|
---|
| 42 |
|
---|
| 43 | static int cmpfunc(const void *a, const void *b)
|
---|
| 44 | {
|
---|
[1765] | 45 | return (((struct cut_list *) a)->startpos -
|
---|
| 46 | ((struct cut_list *) b)->startpos);
|
---|
[821] | 47 |
|
---|
| 48 | }
|
---|
| 49 |
|
---|
[1765] | 50 | static void cut_file(FILE * file)
|
---|
[821] | 51 | {
|
---|
[1765] | 52 | char *line = NULL;
|
---|
| 53 | unsigned int linenum = 0; /* keep these zero-based to be consistent */
|
---|
[821] | 54 |
|
---|
[1765] | 55 | /* go through every line in the file */
|
---|
| 56 | while ((line = xmalloc_getline(file)) != NULL) {
|
---|
[821] | 57 |
|
---|
[1765] | 58 | /* set up a list so we can keep track of what's been printed */
|
---|
| 59 | char * printed = xzalloc(strlen(line) * sizeof(char));
|
---|
| 60 | char * orig_line = line;
|
---|
| 61 | unsigned int cl_pos = 0;
|
---|
| 62 | int spos;
|
---|
[821] | 63 |
|
---|
[1765] | 64 | /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
|
---|
| 65 | if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
|
---|
| 66 | /* print the chars specified in each cut list */
|
---|
| 67 | for (; cl_pos < nlists; cl_pos++) {
|
---|
| 68 | spos = cut_lists[cl_pos].startpos;
|
---|
| 69 | while (spos < strlen(line)) {
|
---|
| 70 | if (!printed[spos]) {
|
---|
| 71 | printed[spos] = 'X';
|
---|
| 72 | putchar(line[spos]);
|
---|
| 73 | }
|
---|
| 74 | spos++;
|
---|
| 75 | if (spos > cut_lists[cl_pos].endpos
|
---|
| 76 | || cut_lists[cl_pos].endpos == NON_RANGE)
|
---|
| 77 | break;
|
---|
| 78 | }
|
---|
| 79 | }
|
---|
| 80 | } else if (delim == '\n') { /* cut by lines */
|
---|
| 81 | spos = cut_lists[cl_pos].startpos;
|
---|
[821] | 82 |
|
---|
[1765] | 83 | /* get out if we have no more lists to process or if the lines
|
---|
| 84 | * are lower than what we're interested in */
|
---|
| 85 | if (linenum < spos || cl_pos >= nlists)
|
---|
| 86 | goto next_line;
|
---|
[821] | 87 |
|
---|
[1765] | 88 | /* if the line we're looking for is lower than the one we were
|
---|
| 89 | * passed, it means we displayed it already, so move on */
|
---|
| 90 | while (spos < linenum) {
|
---|
| 91 | spos++;
|
---|
| 92 | /* go to the next list if we're at the end of this one */
|
---|
| 93 | if (spos > cut_lists[cl_pos].endpos
|
---|
| 94 | || cut_lists[cl_pos].endpos == NON_RANGE) {
|
---|
| 95 | cl_pos++;
|
---|
| 96 | /* get out if there's no more lists to process */
|
---|
| 97 | if (cl_pos >= nlists)
|
---|
| 98 | goto next_line;
|
---|
| 99 | spos = cut_lists[cl_pos].startpos;
|
---|
| 100 | /* get out if the current line is lower than the one
|
---|
| 101 | * we just became interested in */
|
---|
| 102 | if (linenum < spos)
|
---|
| 103 | goto next_line;
|
---|
| 104 | }
|
---|
| 105 | }
|
---|
[821] | 106 |
|
---|
[1765] | 107 | /* If we made it here, it means we've found the line we're
|
---|
| 108 | * looking for, so print it */
|
---|
| 109 | puts(line);
|
---|
| 110 | goto next_line;
|
---|
| 111 | } else { /* cut by fields */
|
---|
| 112 | int ndelim = -1; /* zero-based / one-based problem */
|
---|
| 113 | int nfields_printed = 0;
|
---|
| 114 | char *field = NULL;
|
---|
| 115 | const char delimiter[2] = { delim, 0 };
|
---|
[821] | 116 |
|
---|
[1765] | 117 | /* does this line contain any delimiters? */
|
---|
| 118 | if (strchr(line, delim) == NULL) {
|
---|
| 119 | if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
|
---|
| 120 | puts(line);
|
---|
| 121 | goto next_line;
|
---|
| 122 | }
|
---|
[821] | 123 |
|
---|
[1765] | 124 | /* process each list on this line, for as long as we've got
|
---|
| 125 | * a line to process */
|
---|
| 126 | for (; cl_pos < nlists && line; cl_pos++) {
|
---|
| 127 | spos = cut_lists[cl_pos].startpos;
|
---|
| 128 | do {
|
---|
| 129 | /* find the field we're looking for */
|
---|
| 130 | while (line && ndelim < spos) {
|
---|
| 131 | field = strsep(&line, delimiter);
|
---|
| 132 | ndelim++;
|
---|
| 133 | }
|
---|
[821] | 134 |
|
---|
[1765] | 135 | /* we found it, and it hasn't been printed yet */
|
---|
| 136 | if (field && ndelim == spos && !printed[ndelim]) {
|
---|
| 137 | /* if this isn't our first time through, we need to
|
---|
| 138 | * print the delimiter after the last field that was
|
---|
| 139 | * printed */
|
---|
| 140 | if (nfields_printed > 0)
|
---|
| 141 | putchar(delim);
|
---|
| 142 | fputs(field, stdout);
|
---|
| 143 | printed[ndelim] = 'X';
|
---|
| 144 | nfields_printed++; /* shouldn't overflow.. */
|
---|
| 145 | }
|
---|
[821] | 146 |
|
---|
[1765] | 147 | spos++;
|
---|
[821] | 148 |
|
---|
[1765] | 149 | /* keep going as long as we have a line to work with,
|
---|
| 150 | * this is a list, and we're not at the end of that
|
---|
| 151 | * list */
|
---|
| 152 | } while (spos <= cut_lists[cl_pos].endpos && line
|
---|
| 153 | && cut_lists[cl_pos].endpos != NON_RANGE);
|
---|
[821] | 154 | }
|
---|
| 155 | }
|
---|
[1765] | 156 | /* if we printed anything at all, we need to finish it with a
|
---|
| 157 | * newline cuz we were handed a chomped line */
|
---|
| 158 | putchar('\n');
|
---|
| 159 | next_line:
|
---|
| 160 | linenum++;
|
---|
| 161 | free(printed);
|
---|
| 162 | free(orig_line);
|
---|
[821] | 163 | }
|
---|
| 164 | }
|
---|
| 165 |
|
---|
[1765] | 166 | static const char _op_on_field[] ALIGN1 = " only when operating on fields";
|
---|
[821] | 167 |
|
---|
[1765] | 168 | int cut_main(int argc, char **argv);
|
---|
| 169 | int cut_main(int argc, char **argv)
|
---|
[821] | 170 | {
|
---|
[1765] | 171 | char *sopt, *ltok;
|
---|
[821] | 172 |
|
---|
[1765] | 173 | opt_complementary = "b--bcf:c--bcf:f--bcf";
|
---|
| 174 | getopt32(argv, optstring, &sopt, &sopt, &sopt, <ok);
|
---|
| 175 | // argc -= optind;
|
---|
| 176 | argv += optind;
|
---|
| 177 | if (!(option_mask32 & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
|
---|
| 178 | bb_error_msg_and_die("expected a list of bytes, characters, or fields");
|
---|
[821] | 179 |
|
---|
[1765] | 180 | if (option_mask32 & CUT_OPT_DELIM_FLGS) {
|
---|
| 181 | if (strlen(ltok) > 1) {
|
---|
| 182 | bb_error_msg_and_die("the delimiter must be a single character");
|
---|
| 183 | }
|
---|
| 184 | delim = ltok[0];
|
---|
[821] | 185 | }
|
---|
| 186 |
|
---|
[1765] | 187 | /* non-field (char or byte) cutting has some special handling */
|
---|
| 188 | if (!(option_mask32 & CUT_OPT_FIELDS_FLGS)) {
|
---|
| 189 | if (option_mask32 & CUT_OPT_SUPPRESS_FLGS) {
|
---|
| 190 | bb_error_msg_and_die
|
---|
| 191 | ("suppressing non-delimited lines makes sense%s",
|
---|
| 192 | _op_on_field);
|
---|
[821] | 193 | }
|
---|
[1765] | 194 | if (delim != '\t') {
|
---|
| 195 | bb_error_msg_and_die
|
---|
| 196 | ("a delimiter may be specified%s", _op_on_field);
|
---|
| 197 | }
|
---|
[821] | 198 | }
|
---|
| 199 |
|
---|
[1765] | 200 | /*
|
---|
| 201 | * parse list and put values into startpos and endpos.
|
---|
| 202 | * valid list formats: N, N-, N-M, -M
|
---|
| 203 | * more than one list can be separated by commas
|
---|
| 204 | */
|
---|
| 205 | {
|
---|
| 206 | char *ntok;
|
---|
| 207 | int s = 0, e = 0;
|
---|
[821] | 208 |
|
---|
[1765] | 209 | /* take apart the lists, one by one (they are separated with commas */
|
---|
| 210 | while ((ltok = strsep(&sopt, ",")) != NULL) {
|
---|
[821] | 211 |
|
---|
[1765] | 212 | /* it's actually legal to pass an empty list */
|
---|
| 213 | if (strlen(ltok) == 0)
|
---|
| 214 | continue;
|
---|
[821] | 215 |
|
---|
[1765] | 216 | /* get the start pos */
|
---|
| 217 | ntok = strsep(<ok, "-");
|
---|
| 218 | if (ntok == NULL) {
|
---|
| 219 | bb_error_msg
|
---|
| 220 | ("internal error: ntok is null for start pos!?\n");
|
---|
| 221 | } else if (strlen(ntok) == 0) {
|
---|
| 222 | s = BOL;
|
---|
| 223 | } else {
|
---|
| 224 | s = xatoi_u(ntok);
|
---|
| 225 | /* account for the fact that arrays are zero based, while
|
---|
| 226 | * the user expects the first char on the line to be char #1 */
|
---|
| 227 | if (s != 0)
|
---|
| 228 | s--;
|
---|
| 229 | }
|
---|
[821] | 230 |
|
---|
[1765] | 231 | /* get the end pos */
|
---|
| 232 | ntok = strsep(<ok, "-");
|
---|
| 233 | if (ntok == NULL) {
|
---|
| 234 | e = NON_RANGE;
|
---|
| 235 | } else if (strlen(ntok) == 0) {
|
---|
| 236 | e = EOL;
|
---|
| 237 | } else {
|
---|
| 238 | e = xatoi_u(ntok);
|
---|
| 239 | /* if the user specified and end position of 0, that means "til the
|
---|
| 240 | * end of the line */
|
---|
| 241 | if (e == 0)
|
---|
| 242 | e = EOL;
|
---|
| 243 | e--; /* again, arrays are zero based, lines are 1 based */
|
---|
| 244 | if (e == s)
|
---|
| 245 | e = NON_RANGE;
|
---|
| 246 | }
|
---|
[821] | 247 |
|
---|
[1765] | 248 | /* if there's something left to tokenize, the user passed
|
---|
| 249 | * an invalid list */
|
---|
| 250 | if (ltok)
|
---|
| 251 | bb_error_msg_and_die("invalid byte or field list");
|
---|
| 252 |
|
---|
| 253 | /* add the new list */
|
---|
| 254 | cut_lists = xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists));
|
---|
| 255 | cut_lists[nlists-1].startpos = s;
|
---|
| 256 | cut_lists[nlists-1].endpos = e;
|
---|
[821] | 257 | }
|
---|
| 258 |
|
---|
[1765] | 259 | /* make sure we got some cut positions out of all that */
|
---|
| 260 | if (nlists == 0)
|
---|
| 261 | bb_error_msg_and_die("missing list of positions");
|
---|
[821] | 262 |
|
---|
[1765] | 263 | /* now that the lists are parsed, we need to sort them to make life
|
---|
| 264 | * easier on us when it comes time to print the chars / fields / lines
|
---|
| 265 | */
|
---|
| 266 | qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc);
|
---|
[821] | 267 | }
|
---|
| 268 |
|
---|
[1765] | 269 | /* argv[0..argc-1] should be names of file to process. If no
|
---|
[821] | 270 | * files were specified or '-' was specified, take input from stdin.
|
---|
| 271 | * Otherwise, we process all the files specified. */
|
---|
[1765] | 272 | if (argv[0] == NULL || LONE_DASH(argv[0])) {
|
---|
[821] | 273 | cut_file(stdin);
|
---|
[1765] | 274 | } else {
|
---|
[821] | 275 | FILE *file;
|
---|
[1765] | 276 |
|
---|
| 277 | do {
|
---|
| 278 | file = fopen_or_warn(argv[0], "r");
|
---|
| 279 | if (file) {
|
---|
[821] | 280 | cut_file(file);
|
---|
| 281 | fclose(file);
|
---|
| 282 | }
|
---|
[1765] | 283 | } while (*++argv);
|
---|
[821] | 284 | }
|
---|
[1765] | 285 | if (ENABLE_FEATURE_CLEAN_UP)
|
---|
| 286 | free(cut_lists);
|
---|
[821] | 287 | return EXIT_SUCCESS;
|
---|
| 288 | }
|
---|