box/coreutils/cut.c

   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 Fischer
   8  *
   9  * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
  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[] = "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 static char delim = '\t';       /* delimiter, default is tab */
  26
  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
  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 */
  41
  42
  43 static int cmpfunc(const void *a, const void *b)
  44 {
  45         return (((struct cut_list *) a)->startpos -
  46                         ((struct cut_list *) b)->startpos);
  47
  48 }
  49
  50 static void cut_file(FILE * file)
  51 {
  52         char *line = NULL;
  53         unsigned int linenum = 0;       /* keep these zero-based to be consistent */
  54
  55         /* go through every line in the file */
  56         while ((line = xmalloc_getline(file)) != NULL) {
  57
  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;
  63
  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;
  82
  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;
  87
  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                         }
 106
 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 };
 116
 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                         }
 123
 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                                         }
 134
 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                                         }
 146
 147                                         spos++;
 148
 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);
 154                         }
 155                 }
 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);
 163         }
 164 }
 165
 166 static const char _op_on_field[] = " only when operating on fields";
 167
 168 int cut_main(int argc, char **argv);
 169 int cut_main(int argc, char **argv)
 170 {
 171         char *sopt, *ltok;
 172
 173         opt_complementary = "b--bcf:c--bcf:f--bcf";
 174         getopt32(argc, argv, optstring, &sopt, &sopt, &sopt, &ltok);
 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");
 179         if (option_mask32 & BB_GETOPT_ERROR)
 180                 bb_error_msg_and_die("only one type of list may be specified");
 181
 182         if (option_mask32 & CUT_OPT_DELIM_FLGS) {
 183                 if (strlen(ltok) > 1) {
 184                         bb_error_msg_and_die("the delimiter must be a single character");
 185                 }
 186                 delim = ltok[0];
 187         }
 188
 189         /*  non-field (char or byte) cutting has some special handling */
 190         if (!(option_mask32 & CUT_OPT_FIELDS_FLGS)) {
 191                 if (option_mask32 & CUT_OPT_SUPPRESS_FLGS) {
 192                         bb_error_msg_and_die
 193                                 ("suppressing non-delimited lines makes sense%s",
 194                                  _op_on_field);
 195                 }
 196                 if (delim != '\t') {
 197                         bb_error_msg_and_die
 198                                 ("a delimiter may be specified%s", _op_on_field);
 199                 }
 200         }
 201
 202         /*
 203          * parse list and put values into startpos and endpos.
 204          * valid list formats: N, N-, N-M, -M
 205          * more than one list can be separated by commas
 206          */
 207         {
 208                 char *ntok;
 209                 int s = 0, e = 0;
 210
 211                 /* take apart the lists, one by one (they are separated with commas */
 212                 while ((ltok = strsep(&sopt, ",")) != NULL) {
 213
 214                         /* it's actually legal to pass an empty list */
 215                         if (strlen(ltok) == 0)
 216                                 continue;
 217
 218                         /* get the start pos */
 219                         ntok = strsep(&ltok, "-");
 220                         if (ntok == NULL) {
 221                                 bb_error_msg
 222                                         ("internal error: ntok is null for start pos!?\n");
 223                         } else if (strlen(ntok) == 0) {
 224                                 s = BOL;
 225                         } else {
 226                                 s = xatoi_u(ntok);
 227                                 /* account for the fact that arrays are zero based, while
 228                                  * the user expects the first char on the line to be char #1 */
 229                                 if (s != 0)
 230                                         s--;
 231                         }
 232
 233                         /* get the end pos */
 234                         ntok = strsep(&ltok, "-");
 235                         if (ntok == NULL) {
 236                                 e = NON_RANGE;
 237                         } else if (strlen(ntok) == 0) {
 238                                 e = EOL;
 239                         } else {
 240                                 e = xatoi_u(ntok);
 241                                 /* if the user specified and end position of 0, that means "til the
 242                                  * 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                         /* if there's something left to tokenize, the user passed
 251                          * an invalid list */
 252                         if (ltok)
 253                                 bb_error_msg_and_die("invalid byte or field list");
 254
 255                         /* add the new list */
 256                         cut_lists = xrealloc(cut_lists, sizeof(struct cut_list) * (++nlists));
 257                         cut_lists[nlists-1].startpos = s;
 258                         cut_lists[nlists-1].endpos = e;
 259                 }
 260
 261                 /* make sure we got some cut positions out of all that */
 262                 if (nlists == 0)
 263                         bb_error_msg_and_die("missing list of positions");
 264
 265                 /* now that the lists are parsed, we need to sort them to make life
 266                  * easier on us when it comes time to print the chars / fields / lines
 267                  */
 268                 qsort(cut_lists, nlists, sizeof(struct cut_list), cmpfunc);
 269         }
 270
 271         /* argv[0..argc-1] should be names of file to process. If no
 272          * files were specified or '-' was specified, take input from stdin.
 273          * Otherwise, we process all the files specified. */
 274         if (argv[0] == NULL || LONE_DASH(argv[0])) {
 275                 cut_file(stdin);
 276         } else {
 277                 FILE *file;
 278
 279                 do {
 280                         file = fopen_or_warn(argv[0], "r");
 281                         if (file) {
 282                                 cut_file(file);
 283                                 fclose(file);
 284                         }
 285                 } while (*++argv);
 286         }
 287         if (ENABLE_FEATURE_CLEAN_UP)
 288                 free(cut_lists);
 289         return EXIT_SUCCESS;
 290 }