"Robert P. J. Day" | 63fc1a9 | 2006-07-02 19:47:05 +0000 | [diff] [blame] | 1 | /* vi: set sw=4 ts=4: */ |
Erik Andersen | 7ab9c7e | 2000-05-12 19:41:47 +0000 | [diff] [blame] | 2 | /* |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 3 | * cut.c - minimalist version of cut |
Erik Andersen | 7ab9c7e | 2000-05-12 19:41:47 +0000 | [diff] [blame] | 4 | * |
Eric Andersen | 8ec10a9 | 2001-01-27 09:33:39 +0000 | [diff] [blame] | 5 | * Copyright (C) 1999,2000,2001 by Lineo, inc. |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 6 | * Written by Mark Whitley <markw@codepoet.org> |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 7 | * debloated by Bernhard Fischer |
Erik Andersen | 7ab9c7e | 2000-05-12 19:41:47 +0000 | [diff] [blame] | 8 | * |
"Robert P. J. Day" | 801ab14 | 2006-07-12 07:56:04 +0000 | [diff] [blame] | 9 | * Licensed under GPLv2 or later, see file LICENSE in this tarball for details. |
Erik Andersen | 7ab9c7e | 2000-05-12 19:41:47 +0000 | [diff] [blame] | 10 | */ |
| 11 | |
Denis Vlasenko | b6adbf1 | 2007-05-26 19:00:18 +0000 | [diff] [blame] | 12 | #include "libbb.h" |
Erik Andersen | 7ab9c7e | 2000-05-12 19:41:47 +0000 | [diff] [blame] | 13 | |
Denis Vlasenko | 99912ca | 2007-04-10 15:43:37 +0000 | [diff] [blame] | 14 | /* This is a NOEXEC applet. Be very careful! */ |
| 15 | |
| 16 | |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 17 | /* option vars */ |
Denis Vlasenko | 6ca409e | 2007-08-12 20:58:27 +0000 | [diff] [blame] | 18 | static const char optstring[] ALIGN1 = "b:c:f:d:sn"; |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 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) |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 24 | |
| 25 | struct cut_list { |
| 26 | int startpos; |
| 27 | int endpos; |
| 28 | }; |
| 29 | |
Rob Landley | bc68cd1 | 2006-03-10 19:22:06 +0000 | [diff] [blame] | 30 | enum { |
| 31 | BOL = 0, |
| 32 | EOL = INT_MAX, |
| 33 | NON_RANGE = -1 |
| 34 | }; |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 35 | |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 36 | /* growable array holding a series of lists */ |
| 37 | static struct cut_list *cut_lists; |
| 38 | static unsigned int nlists; /* number of elements in above list */ |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 39 | |
| 40 | |
| 41 | static int cmpfunc(const void *a, const void *b) |
| 42 | { |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 43 | return (((struct cut_list *) a)->startpos - |
| 44 | ((struct cut_list *) b)->startpos); |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 45 | |
| 46 | } |
| 47 | |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 48 | static void cut_file(FILE *file, char delim) |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 49 | { |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 50 | char *line = NULL; |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 51 | unsigned int linenum = 0; /* keep these zero-based to be consistent */ |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 52 | |
| 53 | /* go through every line in the file */ |
Denis Vlasenko | 8ee649a | 2008-03-26 20:04:27 +0000 | [diff] [blame] | 54 | while ((line = xmalloc_fgetline(file)) != NULL) { |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 55 | |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 56 | /* set up a list so we can keep track of what's been printed */ |
| 57 | char * printed = xzalloc(strlen(line) * sizeof(char)); |
| 58 | char * orig_line = line; |
| 59 | unsigned int cl_pos = 0; |
| 60 | int spos; |
| 61 | |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 62 | /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */ |
Denis Vlasenko | 372686b | 2006-10-12 22:42:33 +0000 | [diff] [blame] | 63 | if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) { |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 64 | /* print the chars specified in each cut list */ |
| 65 | for (; cl_pos < nlists; cl_pos++) { |
| 66 | spos = cut_lists[cl_pos].startpos; |
| 67 | while (spos < strlen(line)) { |
| 68 | if (!printed[spos]) { |
| 69 | printed[spos] = 'X'; |
| 70 | putchar(line[spos]); |
| 71 | } |
| 72 | spos++; |
| 73 | if (spos > cut_lists[cl_pos].endpos |
| 74 | || cut_lists[cl_pos].endpos == NON_RANGE) |
| 75 | break; |
| 76 | } |
| 77 | } |
| 78 | } else if (delim == '\n') { /* cut by lines */ |
| 79 | spos = cut_lists[cl_pos].startpos; |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 80 | |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 81 | /* get out if we have no more lists to process or if the lines |
| 82 | * are lower than what we're interested in */ |
| 83 | if (linenum < spos || cl_pos >= nlists) |
| 84 | goto next_line; |
| 85 | |
| 86 | /* if the line we're looking for is lower than the one we were |
| 87 | * passed, it means we displayed it already, so move on */ |
| 88 | while (spos < linenum) { |
| 89 | spos++; |
| 90 | /* go to the next list if we're at the end of this one */ |
| 91 | if (spos > cut_lists[cl_pos].endpos |
| 92 | || cut_lists[cl_pos].endpos == NON_RANGE) { |
| 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 (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 | const char delimiter[2] = { delim, 0 }; |
| 114 | |
| 115 | /* does this line contain any delimiters? */ |
| 116 | if (strchr(line, delim) == NULL) { |
Denis Vlasenko | 372686b | 2006-10-12 22:42:33 +0000 | [diff] [blame] | 117 | if (!(option_mask32 & CUT_OPT_SUPPRESS_FLGS)) |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 118 | puts(line); |
| 119 | goto next_line; |
| 120 | } |
| 121 | |
| 122 | /* process each list on this line, for as long as we've got |
| 123 | * a line to process */ |
| 124 | for (; cl_pos < nlists && line; cl_pos++) { |
| 125 | spos = cut_lists[cl_pos].startpos; |
| 126 | do { |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 127 | /* find the field we're looking for */ |
| 128 | while (line && ndelim < spos) { |
| 129 | field = strsep(&line, delimiter); |
| 130 | ndelim++; |
| 131 | } |
| 132 | |
| 133 | /* we found it, and it hasn't been printed yet */ |
| 134 | if (field && ndelim == spos && !printed[ndelim]) { |
| 135 | /* if this isn't our first time through, we need to |
| 136 | * print the delimiter after the last field that was |
| 137 | * printed */ |
| 138 | if (nfields_printed > 0) |
| 139 | putchar(delim); |
| 140 | fputs(field, stdout); |
| 141 | printed[ndelim] = 'X'; |
| 142 | nfields_printed++; /* shouldn't overflow.. */ |
| 143 | } |
| 144 | |
| 145 | spos++; |
| 146 | |
| 147 | /* keep going as long as we have a line to work with, |
| 148 | * this is a list, and we're not at the end of that |
| 149 | * list */ |
| 150 | } while (spos <= cut_lists[cl_pos].endpos && line |
| 151 | && cut_lists[cl_pos].endpos != NON_RANGE); |
| 152 | } |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 153 | } |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 154 | /* if we printed anything at all, we need to finish it with a |
| 155 | * newline cuz we were handed a chomped line */ |
| 156 | putchar('\n'); |
Denis Vlasenko | 372686b | 2006-10-12 22:42:33 +0000 | [diff] [blame] | 157 | next_line: |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 158 | linenum++; |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 159 | free(printed); |
| 160 | free(orig_line); |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 161 | } |
| 162 | } |
| 163 | |
Denis Vlasenko | 9b49a5e | 2007-10-11 10:05:36 +0000 | [diff] [blame] | 164 | int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
Denis Vlasenko | 68404f1 | 2008-03-17 09:00:54 +0000 | [diff] [blame] | 165 | int cut_main(int argc ATTRIBUTE_UNUSED, char **argv) |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 166 | { |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 167 | char delim = '\t'; /* delimiter, default is tab */ |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 168 | char *sopt, *ltok; |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 169 | |
Denis Vlasenko | 67b23e6 | 2006-10-03 21:00:06 +0000 | [diff] [blame] | 170 | opt_complementary = "b--bcf:c--bcf:f--bcf"; |
Denis Vlasenko | fe7cd64 | 2007-08-18 15:32:12 +0000 | [diff] [blame] | 171 | getopt32(argv, optstring, &sopt, &sopt, &sopt, <ok); |
Denis Vlasenko | 9f73944 | 2006-12-16 23:49:13 +0000 | [diff] [blame] | 172 | // argc -= optind; |
| 173 | argv += optind; |
Denis Vlasenko | 372686b | 2006-10-12 22:42:33 +0000 | [diff] [blame] | 174 | if (!(option_mask32 & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS))) |
| 175 | bb_error_msg_and_die("expected a list of bytes, characters, or fields"); |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 176 | |
Denis Vlasenko | 372686b | 2006-10-12 22:42:33 +0000 | [diff] [blame] | 177 | if (option_mask32 & CUT_OPT_DELIM_FLGS) { |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 178 | if (ltok[0] && ltok[1]) { /* more than 1 char? */ |
Eric Andersen | c9e7024 | 2003-06-20 09:16:00 +0000 | [diff] [blame] | 179 | bb_error_msg_and_die("the delimiter must be a single character"); |
| 180 | } |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 181 | delim = ltok[0]; |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 182 | } |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 183 | |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 184 | /* non-field (char or byte) cutting has some special handling */ |
Denis Vlasenko | 372686b | 2006-10-12 22:42:33 +0000 | [diff] [blame] | 185 | if (!(option_mask32 & CUT_OPT_FIELDS_FLGS)) { |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 186 | static const char _op_on_field[] ALIGN1 = " only when operating on fields"; |
| 187 | |
Denis Vlasenko | 372686b | 2006-10-12 22:42:33 +0000 | [diff] [blame] | 188 | if (option_mask32 & CUT_OPT_SUPPRESS_FLGS) { |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 189 | bb_error_msg_and_die |
| 190 | ("suppressing non-delimited lines makes sense%s", |
| 191 | _op_on_field); |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 192 | } |
Eric Andersen | 8876fb2 | 2003-06-20 09:01:58 +0000 | [diff] [blame] | 193 | if (delim != '\t') { |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 194 | bb_error_msg_and_die |
| 195 | ("a delimiter may be specified%s", _op_on_field); |
Mark Whitley | b696763 | 2001-05-18 23:04:51 +0000 | [diff] [blame] | 196 | } |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 197 | } |
| 198 | |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 199 | /* |
| 200 | * parse list and put values into startpos and endpos. |
| 201 | * valid list formats: N, N-, N-M, -M |
| 202 | * more than one list can be separated by commas |
| 203 | */ |
| 204 | { |
| 205 | char *ntok; |
| 206 | int s = 0, e = 0; |
| 207 | |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 208 | /* take apart the lists, one by one (they are separated with commas) */ |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 209 | while ((ltok = strsep(&sopt, ",")) != NULL) { |
| 210 | |
| 211 | /* it's actually legal to pass an empty list */ |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 212 | if (!ltok[0]) |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 213 | continue; |
| 214 | |
| 215 | /* get the start pos */ |
| 216 | ntok = strsep(<ok, "-"); |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 217 | if (!ntok[0]) { |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 218 | s = BOL; |
| 219 | } else { |
Denis Vlasenko | 1385899 | 2006-10-08 12:49:22 +0000 | [diff] [blame] | 220 | s = xatoi_u(ntok); |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 221 | /* account for the fact that arrays are zero based, while |
| 222 | * the user expects the first char on the line to be char #1 */ |
| 223 | if (s != 0) |
| 224 | s--; |
| 225 | } |
| 226 | |
| 227 | /* get the end pos */ |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 228 | if (ltok == NULL) { |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 229 | e = NON_RANGE; |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 230 | } else if (!ltok[0]) { |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 231 | e = EOL; |
| 232 | } else { |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 233 | e = xatoi_u(ltok); |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 234 | /* if the user specified and end position of 0, that means "til the |
| 235 | * end of the line */ |
| 236 | if (e == 0) |
| 237 | e = EOL; |
| 238 | e--; /* again, arrays are zero based, lines are 1 based */ |
| 239 | if (e == s) |
| 240 | e = NON_RANGE; |
| 241 | } |
| 242 | |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 243 | /* add the new list */ |
Denis Vlasenko | 372686b | 2006-10-12 22:42:33 +0000 | [diff] [blame] | 244 | cut_lists = xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists)); |
| 245 | cut_lists[nlists-1].startpos = s; |
| 246 | cut_lists[nlists-1].endpos = e; |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 247 | } |
| 248 | |
| 249 | /* make sure we got some cut positions out of all that */ |
| 250 | if (nlists == 0) |
| 251 | bb_error_msg_and_die("missing list of positions"); |
| 252 | |
| 253 | /* now that the lists are parsed, we need to sort them to make life |
| 254 | * easier on us when it comes time to print the chars / fields / lines |
| 255 | */ |
| 256 | qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc); |
| 257 | } |
| 258 | |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 259 | { |
| 260 | int retval = EXIT_SUCCESS; |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 261 | |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 262 | if (!*argv) |
| 263 | *--argv = (char *)"-"; |
Bernhard Reutner-Fischer | 73561cc | 2006-08-28 23:31:54 +0000 | [diff] [blame] | 264 | |
Denis Vlasenko | 9f73944 | 2006-12-16 23:49:13 +0000 | [diff] [blame] | 265 | do { |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 266 | FILE *file = fopen_or_warn_stdin(*argv); |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 267 | if (!file) { |
| 268 | retval = EXIT_FAILURE; |
| 269 | continue; |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 270 | } |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 271 | cut_file(file, delim); |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 272 | fclose_if_not_stdin(file); |
Denis Vlasenko | 9f73944 | 2006-12-16 23:49:13 +0000 | [diff] [blame] | 273 | } while (*++argv); |
Denis Vlasenko | 3831c91 | 2007-11-23 07:26:15 +0000 | [diff] [blame] | 274 | |
| 275 | if (ENABLE_FEATURE_CLEAN_UP) |
| 276 | free(cut_lists); |
| 277 | fflush_stdout_and_exit(retval); |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 278 | } |
Mark Whitley | 807f0fd | 2000-08-02 18:30:11 +0000 | [diff] [blame] | 279 | } |