[3320] | 1 | /* vi: set sw=4 ts=4: */
|
---|
| 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>
|
---|
| 7 | * debloated by Bernhard Reutner-Fischer
|
---|
| 8 | *
|
---|
| 9 | * Licensed under GPLv2 or later, see file LICENSE in this source tree.
|
---|
| 10 | */
|
---|
| 11 |
|
---|
| 12 | #include "libbb.h"
|
---|
| 13 |
|
---|
| 14 | /* This is a NOEXEC applet. Be very careful! */
|
---|
| 15 |
|
---|
| 16 |
|
---|
| 17 | /* option vars */
|
---|
| 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)
|
---|
| 24 |
|
---|
| 25 | struct cut_list {
|
---|
| 26 | int startpos;
|
---|
| 27 | int endpos;
|
---|
| 28 | };
|
---|
| 29 |
|
---|
| 30 | enum {
|
---|
| 31 | BOL = 0,
|
---|
| 32 | EOL = INT_MAX,
|
---|
| 33 | NON_RANGE = -1
|
---|
| 34 | };
|
---|
| 35 |
|
---|
| 36 | static int cmpfunc(const void *a, const void *b)
|
---|
| 37 | {
|
---|
| 38 | return (((struct cut_list *) a)->startpos -
|
---|
| 39 | ((struct cut_list *) b)->startpos);
|
---|
| 40 | }
|
---|
| 41 |
|
---|
| 42 | static void cut_file(FILE *file, char delim, const struct cut_list *cut_lists, unsigned nlists)
|
---|
| 43 | {
|
---|
| 44 | char *line;
|
---|
| 45 | unsigned linenum = 0; /* keep these zero-based to be consistent */
|
---|
| 46 |
|
---|
| 47 | /* go through every line in the file */
|
---|
| 48 | while ((line = xmalloc_fgetline(file)) != NULL) {
|
---|
| 49 |
|
---|
| 50 | /* set up a list so we can keep track of what's been printed */
|
---|
| 51 | int linelen = strlen(line);
|
---|
| 52 | char *printed = xzalloc(linelen + 1);
|
---|
| 53 | char *orig_line = line;
|
---|
| 54 | unsigned cl_pos = 0;
|
---|
| 55 | int spos;
|
---|
| 56 |
|
---|
| 57 | /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
|
---|
| 58 | if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
|
---|
| 59 | /* print the chars specified in each cut list */
|
---|
| 60 | for (; cl_pos < nlists; cl_pos++) {
|
---|
| 61 | spos = cut_lists[cl_pos].startpos;
|
---|
| 62 | while (spos < linelen) {
|
---|
| 63 | if (!printed[spos]) {
|
---|
| 64 | printed[spos] = 'X';
|
---|
| 65 | putchar(line[spos]);
|
---|
| 66 | }
|
---|
| 67 | spos++;
|
---|
| 68 | if (spos > cut_lists[cl_pos].endpos
|
---|
| 69 | /* NON_RANGE is -1, so if below is true,
|
---|
| 70 | * the above was true too (spos is >= 0) */
|
---|
| 71 | /* || cut_lists[cl_pos].endpos == NON_RANGE */
|
---|
| 72 | ) {
|
---|
| 73 | break;
|
---|
| 74 | }
|
---|
| 75 | }
|
---|
| 76 | }
|
---|
| 77 | } else if (delim == '\n') { /* cut by lines */
|
---|
| 78 | spos = cut_lists[cl_pos].startpos;
|
---|
| 79 |
|
---|
| 80 | /* get out if we have no more lists to process or if the lines
|
---|
| 81 | * are lower than what we're interested in */
|
---|
| 82 | if (((int)linenum < spos) || (cl_pos >= nlists))
|
---|
| 83 | goto next_line;
|
---|
| 84 |
|
---|
| 85 | /* if the line we're looking for is lower than the one we were
|
---|
| 86 | * passed, it means we displayed it already, so move on */
|
---|
| 87 | while (spos < (int)linenum) {
|
---|
| 88 | spos++;
|
---|
| 89 | /* go to the next list if we're at the end of this one */
|
---|
| 90 | if (spos > cut_lists[cl_pos].endpos
|
---|
| 91 | || cut_lists[cl_pos].endpos == NON_RANGE
|
---|
| 92 | ) {
|
---|
| 93 | cl_pos++;
|
---|
| 94 | /* get out if there's no more lists to process */
|
---|
| 95 | if (cl_pos >= nlists)
|
---|
| 96 | goto next_line;
|
---|
| 97 | spos = cut_lists[cl_pos].startpos;
|
---|
| 98 | /* get out if the current line is lower than the one
|
---|
| 99 | * we just became interested in */
|
---|
| 100 | if ((int)linenum < spos)
|
---|
| 101 | goto next_line;
|
---|
| 102 | }
|
---|
| 103 | }
|
---|
| 104 |
|
---|
| 105 | /* If we made it here, it means we've found the line we're
|
---|
| 106 | * looking for, so print it */
|
---|
| 107 | puts(line);
|
---|
| 108 | goto next_line;
|
---|
| 109 | } else { /* cut by fields */
|
---|
| 110 | int ndelim = -1; /* zero-based / one-based problem */
|
---|
| 111 | int nfields_printed = 0;
|
---|
| 112 | char *field = NULL;
|
---|
| 113 | char delimiter[2];
|
---|
| 114 |
|
---|
| 115 | delimiter[0] = delim;
|
---|
| 116 | delimiter[1] = 0;
|
---|
| 117 |
|
---|
| 118 | /* does this line contain any delimiters? */
|
---|
| 119 | if (strchr(line, delim) == NULL) {
|
---|
| 120 | if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS))
|
---|
| 121 | puts(line);
|
---|
| 122 | goto next_line;
|
---|
| 123 | }
|
---|
| 124 |
|
---|
| 125 | /* process each list on this line, for as long as we've got
|
---|
| 126 | * a line to process */
|
---|
| 127 | for (; cl_pos < nlists && line; cl_pos++) {
|
---|
| 128 | spos = cut_lists[cl_pos].startpos;
|
---|
| 129 | do {
|
---|
| 130 | /* find the field we're looking for */
|
---|
| 131 | while (line && ndelim < spos) {
|
---|
| 132 | field = strsep(&line, delimiter);
|
---|
| 133 | ndelim++;
|
---|
| 134 | }
|
---|
| 135 |
|
---|
| 136 | /* we found it, and it hasn't been printed yet */
|
---|
| 137 | if (field && ndelim == spos && !printed[ndelim]) {
|
---|
| 138 | /* if this isn't our first time through, we need to
|
---|
| 139 | * print the delimiter after the last field that was
|
---|
| 140 | * printed */
|
---|
| 141 | if (nfields_printed > 0)
|
---|
| 142 | putchar(delim);
|
---|
| 143 | fputs(field, stdout);
|
---|
| 144 | printed[ndelim] = 'X';
|
---|
| 145 | nfields_printed++; /* shouldn't overflow.. */
|
---|
| 146 | }
|
---|
| 147 |
|
---|
| 148 | spos++;
|
---|
| 149 |
|
---|
| 150 | /* keep going as long as we have a line to work with,
|
---|
| 151 | * this is a list, and we're not at the end of that
|
---|
| 152 | * list */
|
---|
| 153 | } while (spos <= cut_lists[cl_pos].endpos && line
|
---|
| 154 | && cut_lists[cl_pos].endpos != NON_RANGE);
|
---|
| 155 | }
|
---|
| 156 | }
|
---|
| 157 | /* if we printed anything at all, we need to finish it with a
|
---|
| 158 | * newline cuz we were handed a chomped line */
|
---|
| 159 | putchar('\n');
|
---|
| 160 | next_line:
|
---|
| 161 | linenum++;
|
---|
| 162 | free(printed);
|
---|
| 163 | free(orig_line);
|
---|
| 164 | }
|
---|
| 165 | }
|
---|
| 166 |
|
---|
| 167 | int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
|
---|
| 168 | int cut_main(int argc UNUSED_PARAM, char **argv)
|
---|
| 169 | {
|
---|
| 170 | /* growable array holding a series of lists */
|
---|
| 171 | struct cut_list *cut_lists = NULL;
|
---|
| 172 | unsigned nlists = 0; /* number of elements in above list */
|
---|
| 173 | char delim = '\t'; /* delimiter, default is tab */
|
---|
| 174 | char *sopt, *ltok;
|
---|
| 175 | unsigned opt;
|
---|
| 176 |
|
---|
| 177 | opt_complementary = "b--bcf:c--bcf:f--bcf";
|
---|
| 178 | opt = getopt32(argv, optstring, &sopt, &sopt, &sopt, <ok);
|
---|
| 179 | // argc -= optind;
|
---|
| 180 | argv += optind;
|
---|
| 181 | if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS)))
|
---|
| 182 | bb_error_msg_and_die("expected a list of bytes, characters, or fields");
|
---|
| 183 |
|
---|
| 184 | if (opt & CUT_OPT_DELIM_FLGS) {
|
---|
| 185 | if (ltok[0] && ltok[1]) { /* more than 1 char? */
|
---|
| 186 | bb_error_msg_and_die("the delimiter must be a single character");
|
---|
| 187 | }
|
---|
| 188 | delim = ltok[0];
|
---|
| 189 | }
|
---|
| 190 |
|
---|
| 191 | /* non-field (char or byte) cutting has some special handling */
|
---|
| 192 | if (!(opt & CUT_OPT_FIELDS_FLGS)) {
|
---|
| 193 | static const char _op_on_field[] ALIGN1 = " only when operating on fields";
|
---|
| 194 |
|
---|
| 195 | if (opt & CUT_OPT_SUPPRESS_FLGS) {
|
---|
| 196 | bb_error_msg_and_die
|
---|
| 197 | ("suppressing non-delimited lines makes sense%s",
|
---|
| 198 | _op_on_field);
|
---|
| 199 | }
|
---|
| 200 | if (delim != '\t') {
|
---|
| 201 | bb_error_msg_and_die
|
---|
| 202 | ("a delimiter may be specified%s", _op_on_field);
|
---|
| 203 | }
|
---|
| 204 | }
|
---|
| 205 |
|
---|
| 206 | /*
|
---|
| 207 | * parse list and put values into startpos and endpos.
|
---|
| 208 | * valid list formats: N, N-, N-M, -M
|
---|
| 209 | * more than one list can be separated by commas
|
---|
| 210 | */
|
---|
| 211 | {
|
---|
| 212 | char *ntok;
|
---|
| 213 | int s = 0, e = 0;
|
---|
| 214 |
|
---|
| 215 | /* take apart the lists, one by one (they are separated with commas) */
|
---|
| 216 | while ((ltok = strsep(&sopt, ",")) != NULL) {
|
---|
| 217 |
|
---|
| 218 | /* it's actually legal to pass an empty list */
|
---|
| 219 | if (!ltok[0])
|
---|
| 220 | continue;
|
---|
| 221 |
|
---|
| 222 | /* get the start pos */
|
---|
| 223 | ntok = strsep(<ok, "-");
|
---|
| 224 | if (!ntok[0]) {
|
---|
| 225 | s = BOL;
|
---|
| 226 | } else {
|
---|
| 227 | s = xatoi_positive(ntok);
|
---|
| 228 | /* account for the fact that arrays are zero based, while
|
---|
| 229 | * the user expects the first char on the line to be char #1 */
|
---|
| 230 | if (s != 0)
|
---|
| 231 | s--;
|
---|
| 232 | }
|
---|
| 233 |
|
---|
| 234 | /* get the end pos */
|
---|
| 235 | if (ltok == NULL) {
|
---|
| 236 | e = NON_RANGE;
|
---|
| 237 | } else if (!ltok[0]) {
|
---|
| 238 | e = EOL;
|
---|
| 239 | } else {
|
---|
| 240 | e = xatoi_positive(ltok);
|
---|
| 241 | /* if the user specified and end position of 0,
|
---|
| 242 | * that means "til the end of the line" */
|
---|
| 243 | if (e == 0)
|
---|
| 244 | e = EOL;
|
---|
| 245 | e--; /* again, arrays are zero based, lines are 1 based */
|
---|
| 246 | if (e == s)
|
---|
| 247 | e = NON_RANGE;
|
---|
| 248 | }
|
---|
| 249 |
|
---|
| 250 | /* add the new list */
|
---|
| 251 | cut_lists = xrealloc_vector(cut_lists, 4, nlists);
|
---|
| 252 | /* NB: startpos is always >= 0,
|
---|
| 253 | * while endpos may be = NON_RANGE (-1) */
|
---|
| 254 | cut_lists[nlists].startpos = s;
|
---|
| 255 | cut_lists[nlists].endpos = e;
|
---|
| 256 | nlists++;
|
---|
| 257 | }
|
---|
| 258 |
|
---|
| 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");
|
---|
| 262 |
|
---|
| 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(cut_lists[0]), cmpfunc);
|
---|
| 267 | }
|
---|
| 268 |
|
---|
| 269 | {
|
---|
| 270 | int retval = EXIT_SUCCESS;
|
---|
| 271 |
|
---|
| 272 | if (!*argv)
|
---|
| 273 | *--argv = (char *)"-";
|
---|
| 274 |
|
---|
| 275 | do {
|
---|
| 276 | FILE *file = fopen_or_warn_stdin(*argv);
|
---|
| 277 | if (!file) {
|
---|
| 278 | retval = EXIT_FAILURE;
|
---|
| 279 | continue;
|
---|
| 280 | }
|
---|
| 281 | cut_file(file, delim, cut_lists, nlists);
|
---|
| 282 | fclose_if_not_stdin(file);
|
---|
| 283 | } while (*++argv);
|
---|
| 284 |
|
---|
| 285 | if (ENABLE_FEATURE_CLEAN_UP)
|
---|
| 286 | free(cut_lists);
|
---|
| 287 | fflush_stdout_and_exit(retval);
|
---|
| 288 | }
|
---|
| 289 | }
|
---|