[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>
|
---|
[2725] | 7 | * debloated by Bernhard Reutner-Fischer
|
---|
[821] | 8 | *
|
---|
[2725] | 9 | * Licensed under GPLv2 or later, see file LICENSE in this source tree.
|
---|
[821] | 10 | */
|
---|
| 11 |
|
---|
[3232] | 12 | //usage:#define cut_trivial_usage
|
---|
| 13 | //usage: "[OPTIONS] [FILE]..."
|
---|
| 14 | //usage:#define cut_full_usage "\n\n"
|
---|
| 15 | //usage: "Print selected fields from each input FILE to stdout\n"
|
---|
| 16 | //usage: "\n -b LIST Output only bytes from LIST"
|
---|
| 17 | //usage: "\n -c LIST Output only characters from LIST"
|
---|
| 18 | //usage: "\n -d CHAR Use CHAR instead of tab as the field delimiter"
|
---|
| 19 | //usage: "\n -s Output only the lines containing delimiter"
|
---|
| 20 | //usage: "\n -f N Print only these fields"
|
---|
| 21 | //usage: "\n -n Ignored"
|
---|
| 22 | //usage:
|
---|
| 23 | //usage:#define cut_example_usage
|
---|
| 24 | //usage: "$ echo \"Hello world\" | cut -f 1 -d ' '\n"
|
---|
| 25 | //usage: "Hello\n"
|
---|
| 26 | //usage: "$ echo \"Hello world\" | cut -f 2 -d ' '\n"
|
---|
| 27 | //usage: "world\n"
|
---|
| 28 |
|
---|
[1765] | 29 | #include "libbb.h"
|
---|
[821] | 30 |
|
---|
[1765] | 31 | /* This is a NOEXEC applet. Be very careful! */
|
---|
[821] | 32 |
|
---|
[1765] | 33 |
|
---|
[821] | 34 | /* option vars */
|
---|
[1765] | 35 | static const char optstring[] ALIGN1 = "b:c:f:d:sn";
|
---|
[2725] | 36 | #define CUT_OPT_BYTE_FLGS (1 << 0)
|
---|
| 37 | #define CUT_OPT_CHAR_FLGS (1 << 1)
|
---|
| 38 | #define CUT_OPT_FIELDS_FLGS (1 << 2)
|
---|
| 39 | #define CUT_OPT_DELIM_FLGS (1 << 3)
|
---|
| 40 | #define CUT_OPT_SUPPRESS_FLGS (1 << 4)
|
---|
[821] | 41 |
|
---|
| 42 | struct cut_list {
|
---|
| 43 | int startpos;
|
---|
| 44 | int endpos;
|
---|
| 45 | };
|
---|
| 46 |
|
---|
| 47 | enum {
|
---|
| 48 | BOL = 0,
|
---|
| 49 | EOL = INT_MAX,
|
---|
| 50 | NON_RANGE = -1
|
---|
| 51 | };
|
---|
| 52 |
|
---|
| 53 | static int cmpfunc(const void *a, const void *b)
|
---|
| 54 | {
|
---|
[1765] | 55 | return (((struct cut_list *) a)->startpos -
|
---|
| 56 | ((struct cut_list *) b)->startpos);
|
---|
[821] | 57 | }
|
---|
| 58 |
|
---|
[2725] | 59 | static void cut_file(FILE *file, char delim, const struct cut_list *cut_lists, unsigned nlists)
|
---|
[821] | 60 | {
|
---|
[2725] | 61 | char *line;
|
---|
| 62 | unsigned linenum = 0; /* keep these zero-based to be consistent */
|
---|
[821] | 63 |
|
---|
[1765] | 64 | /* go through every line in the file */
|
---|
[2725] | 65 | while ((line = xmalloc_fgetline(file)) != NULL) {
|
---|
[821] | 66 |
|
---|
[1765] | 67 | /* set up a list so we can keep track of what's been printed */
|
---|
[2725] | 68 | int linelen = strlen(line);
|
---|
| 69 | char *printed = xzalloc(linelen + 1);
|
---|
| 70 | char *orig_line = line;
|
---|
| 71 | unsigned cl_pos = 0;
|
---|
[1765] | 72 | int spos;
|
---|
[821] | 73 |
|
---|
[1765] | 74 | /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
|
---|
| 75 | if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
|
---|
| 76 | /* print the chars specified in each cut list */
|
---|
| 77 | for (; cl_pos < nlists; cl_pos++) {
|
---|
| 78 | spos = cut_lists[cl_pos].startpos;
|
---|
[2725] | 79 | while (spos < linelen) {
|
---|
[1765] | 80 | if (!printed[spos]) {
|
---|
| 81 | printed[spos] = 'X';
|
---|
| 82 | putchar(line[spos]);
|
---|
| 83 | }
|
---|
| 84 | spos++;
|
---|
| 85 | if (spos > cut_lists[cl_pos].endpos
|
---|
[2725] | 86 | /* NON_RANGE is -1, so if below is true,
|
---|
| 87 | * the above was true too (spos is >= 0) */
|
---|
| 88 | /* || cut_lists[cl_pos].endpos == NON_RANGE */
|
---|
| 89 | ) {
|
---|
[1765] | 90 | break;
|
---|
[2725] | 91 | }
|
---|
[1765] | 92 | }
|
---|
| 93 | }
|
---|
| 94 | } else if (delim == '\n') { /* cut by lines */
|
---|
| 95 | spos = cut_lists[cl_pos].startpos;
|
---|
[821] | 96 |
|
---|
[1765] | 97 | /* get out if we have no more lists to process or if the lines
|
---|
| 98 | * are lower than what we're interested in */
|
---|
[2725] | 99 | if (((int)linenum < spos) || (cl_pos >= nlists))
|
---|
[1765] | 100 | goto next_line;
|
---|
[821] | 101 |
|
---|
[1765] | 102 | /* if the line we're looking for is lower than the one we were
|
---|
| 103 | * passed, it means we displayed it already, so move on */
|
---|
[2725] | 104 | while (spos < (int)linenum) {
|
---|
[1765] | 105 | spos++;
|
---|
| 106 | /* go to the next list if we're at the end of this one */
|
---|
| 107 | if (spos > cut_lists[cl_pos].endpos
|
---|
[2725] | 108 | || cut_lists[cl_pos].endpos == NON_RANGE
|
---|
| 109 | ) {
|
---|
[1765] | 110 | cl_pos++;
|
---|
| 111 | /* get out if there's no more lists to process */
|
---|
| 112 | if (cl_pos >= nlists)
|
---|
| 113 | goto next_line;
|
---|
| 114 | spos = cut_lists[cl_pos].startpos;
|
---|
| 115 | /* get out if the current line is lower than the one
|
---|
| 116 | * we just became interested in */
|
---|
[2725] | 117 | if ((int)linenum < spos)
|
---|
[1765] | 118 | goto next_line;
|
---|
| 119 | }
|
---|
| 120 | }
|
---|
[821] | 121 |
|
---|
[1765] | 122 | /* If we made it here, it means we've found the line we're
|
---|
| 123 | * looking for, so print it */
|
---|
| 124 | puts(line);
|
---|
| 125 | goto next_line;
|
---|
| 126 | } else { /* cut by fields */
|
---|
| 127 | int ndelim = -1; /* zero-based / one-based problem */
|
---|
| 128 | int nfields_printed = 0;
|
---|
| 129 | char *field = NULL;
|
---|
[2725] | 130 | char delimiter[2];
|
---|
[821] | 131 |
|
---|
[2725] | 132 | delimiter[0] = delim;
|
---|
| 133 | delimiter[1] = 0;
|
---|
| 134 |
|
---|
[1765] | 135 | /* does this line contain any delimiters? */
|
---|
| 136 | if (strchr(line, delim) == NULL) {
|
---|
| 137 | if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
|
---|
| 138 | puts(line);
|
---|
| 139 | goto next_line;
|
---|
| 140 | }
|
---|
[821] | 141 |
|
---|
[1765] | 142 | /* process each list on this line, for as long as we've got
|
---|
| 143 | * a line to process */
|
---|
| 144 | for (; cl_pos < nlists && line; cl_pos++) {
|
---|
| 145 | spos = cut_lists[cl_pos].startpos;
|
---|
| 146 | do {
|
---|
| 147 | /* find the field we're looking for */
|
---|
| 148 | while (line && ndelim < spos) {
|
---|
| 149 | field = strsep(&line, delimiter);
|
---|
| 150 | ndelim++;
|
---|
| 151 | }
|
---|
[821] | 152 |
|
---|
[1765] | 153 | /* we found it, and it hasn't been printed yet */
|
---|
| 154 | if (field && ndelim == spos && !printed[ndelim]) {
|
---|
| 155 | /* if this isn't our first time through, we need to
|
---|
| 156 | * print the delimiter after the last field that was
|
---|
| 157 | * printed */
|
---|
| 158 | if (nfields_printed > 0)
|
---|
| 159 | putchar(delim);
|
---|
| 160 | fputs(field, stdout);
|
---|
| 161 | printed[ndelim] = 'X';
|
---|
| 162 | nfields_printed++; /* shouldn't overflow.. */
|
---|
| 163 | }
|
---|
[821] | 164 |
|
---|
[1765] | 165 | spos++;
|
---|
[821] | 166 |
|
---|
[1765] | 167 | /* keep going as long as we have a line to work with,
|
---|
| 168 | * this is a list, and we're not at the end of that
|
---|
| 169 | * list */
|
---|
| 170 | } while (spos <= cut_lists[cl_pos].endpos && line
|
---|
[2725] | 171 | && cut_lists[cl_pos].endpos != NON_RANGE);
|
---|
[821] | 172 | }
|
---|
| 173 | }
|
---|
[1765] | 174 | /* if we printed anything at all, we need to finish it with a
|
---|
| 175 | * newline cuz we were handed a chomped line */
|
---|
| 176 | putchar('\n');
|
---|
| 177 | next_line:
|
---|
| 178 | linenum++;
|
---|
| 179 | free(printed);
|
---|
| 180 | free(orig_line);
|
---|
[821] | 181 | }
|
---|
| 182 | }
|
---|
| 183 |
|
---|
[2725] | 184 | int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
|
---|
| 185 | int cut_main(int argc UNUSED_PARAM, char **argv)
|
---|
[821] | 186 | {
|
---|
[2725] | 187 | /* growable array holding a series of lists */
|
---|
| 188 | struct cut_list *cut_lists = NULL;
|
---|
| 189 | unsigned nlists = 0; /* number of elements in above list */
|
---|
| 190 | char delim = '\t'; /* delimiter, default is tab */
|
---|
[1765] | 191 | char *sopt, *ltok;
|
---|
[2725] | 192 | unsigned opt;
|
---|
[821] | 193 |
|
---|
[1765] | 194 | opt_complementary = "b--bcf:c--bcf:f--bcf";
|
---|
[2725] | 195 | opt = getopt32(argv, optstring, &sopt, &sopt, &sopt, <ok);
|
---|
[1765] | 196 | // argc -= optind;
|
---|
| 197 | argv += optind;
|
---|
[2725] | 198 | if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
|
---|
[1765] | 199 | bb_error_msg_and_die("expected a list of bytes, characters, or fields");
|
---|
[821] | 200 |
|
---|
[2725] | 201 | if (opt & CUT_OPT_DELIM_FLGS) {
|
---|
| 202 | if (ltok[0] && ltok[1]) { /* more than 1 char? */
|
---|
[1765] | 203 | bb_error_msg_and_die("the delimiter must be a single character");
|
---|
| 204 | }
|
---|
| 205 | delim = ltok[0];
|
---|
[821] | 206 | }
|
---|
| 207 |
|
---|
[1765] | 208 | /* non-field (char or byte) cutting has some special handling */
|
---|
[2725] | 209 | if (!(opt & CUT_OPT_FIELDS_FLGS)) {
|
---|
| 210 | static const char _op_on_field[] ALIGN1 = " only when operating on fields";
|
---|
| 211 |
|
---|
| 212 | if (opt & CUT_OPT_SUPPRESS_FLGS) {
|
---|
[1765] | 213 | bb_error_msg_and_die
|
---|
| 214 | ("suppressing non-delimited lines makes sense%s",
|
---|
[3232] | 215 | _op_on_field);
|
---|
[821] | 216 | }
|
---|
[1765] | 217 | if (delim != '\t') {
|
---|
| 218 | bb_error_msg_and_die
|
---|
| 219 | ("a delimiter may be specified%s", _op_on_field);
|
---|
| 220 | }
|
---|
[821] | 221 | }
|
---|
| 222 |
|
---|
[1765] | 223 | /*
|
---|
| 224 | * parse list and put values into startpos and endpos.
|
---|
| 225 | * valid list formats: N, N-, N-M, -M
|
---|
| 226 | * more than one list can be separated by commas
|
---|
| 227 | */
|
---|
| 228 | {
|
---|
| 229 | char *ntok;
|
---|
| 230 | int s = 0, e = 0;
|
---|
[821] | 231 |
|
---|
[2725] | 232 | /* take apart the lists, one by one (they are separated with commas) */
|
---|
[1765] | 233 | while ((ltok = strsep(&sopt, ",")) != NULL) {
|
---|
[821] | 234 |
|
---|
[1765] | 235 | /* it's actually legal to pass an empty list */
|
---|
[2725] | 236 | if (!ltok[0])
|
---|
[1765] | 237 | continue;
|
---|
[821] | 238 |
|
---|
[1765] | 239 | /* get the start pos */
|
---|
| 240 | ntok = strsep(<ok, "-");
|
---|
[2725] | 241 | if (!ntok[0]) {
|
---|
[1765] | 242 | s = BOL;
|
---|
| 243 | } else {
|
---|
[2725] | 244 | s = xatoi_positive(ntok);
|
---|
[1765] | 245 | /* account for the fact that arrays are zero based, while
|
---|
| 246 | * the user expects the first char on the line to be char #1 */
|
---|
| 247 | if (s != 0)
|
---|
| 248 | s--;
|
---|
| 249 | }
|
---|
[821] | 250 |
|
---|
[1765] | 251 | /* get the end pos */
|
---|
[2725] | 252 | if (ltok == NULL) {
|
---|
[1765] | 253 | e = NON_RANGE;
|
---|
[2725] | 254 | } else if (!ltok[0]) {
|
---|
[1765] | 255 | e = EOL;
|
---|
| 256 | } else {
|
---|
[2725] | 257 | e = xatoi_positive(ltok);
|
---|
| 258 | /* if the user specified and end position of 0,
|
---|
| 259 | * that means "til the end of the line" */
|
---|
[1765] | 260 | if (e == 0)
|
---|
| 261 | e = EOL;
|
---|
| 262 | e--; /* again, arrays are zero based, lines are 1 based */
|
---|
| 263 | if (e == s)
|
---|
| 264 | e = NON_RANGE;
|
---|
| 265 | }
|
---|
[821] | 266 |
|
---|
[1765] | 267 | /* add the new list */
|
---|
[2725] | 268 | cut_lists = xrealloc_vector(cut_lists, 4, nlists);
|
---|
| 269 | /* NB: startpos is always >= 0,
|
---|
| 270 | * while endpos may be = NON_RANGE (-1) */
|
---|
| 271 | cut_lists[nlists].startpos = s;
|
---|
| 272 | cut_lists[nlists].endpos = e;
|
---|
| 273 | nlists++;
|
---|
[821] | 274 | }
|
---|
| 275 |
|
---|
[1765] | 276 | /* make sure we got some cut positions out of all that */
|
---|
| 277 | if (nlists == 0)
|
---|
| 278 | bb_error_msg_and_die("missing list of positions");
|
---|
[821] | 279 |
|
---|
[1765] | 280 | /* now that the lists are parsed, we need to sort them to make life
|
---|
| 281 | * easier on us when it comes time to print the chars / fields / lines
|
---|
| 282 | */
|
---|
[2725] | 283 | qsort(cut_lists, nlists, sizeof(cut_lists[0]), cmpfunc);
|
---|
[821] | 284 | }
|
---|
| 285 |
|
---|
[2725] | 286 | {
|
---|
| 287 | int retval = EXIT_SUCCESS;
|
---|
[1765] | 288 |
|
---|
[2725] | 289 | if (!*argv)
|
---|
| 290 | *--argv = (char *)"-";
|
---|
| 291 |
|
---|
[1765] | 292 | do {
|
---|
[2725] | 293 | FILE *file = fopen_or_warn_stdin(*argv);
|
---|
| 294 | if (!file) {
|
---|
| 295 | retval = EXIT_FAILURE;
|
---|
| 296 | continue;
|
---|
[821] | 297 | }
|
---|
[2725] | 298 | cut_file(file, delim, cut_lists, nlists);
|
---|
| 299 | fclose_if_not_stdin(file);
|
---|
[1765] | 300 | } while (*++argv);
|
---|
[2725] | 301 |
|
---|
| 302 | if (ENABLE_FEATURE_CLEAN_UP)
|
---|
| 303 | free(cut_lists);
|
---|
| 304 | fflush_stdout_and_exit(retval);
|
---|
[821] | 305 | }
|
---|
| 306 | }
|
---|