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 Reutner-Fischer
   8  *
   9  * Licensed under GPLv2 or later, see file LICENSE in this source tree.
  10  */
  11 //config:config CUT
  12 //config:       bool "cut (6.7 kb)"
  13 //config:       default y
  14 //config:       help
  15 //config:       cut is used to print selected parts of lines from
  16 //config:       each file to stdout.
  17 //config:
  18 //config:config FEATURE_CUT_REGEX
  19 //config:       bool "cut -F"
  20 //config:       default y
  21 //config:       depends on CUT
  22 //config:       help
  23 //config:       Allow regex based delimiters.
  24
  25 //applet:IF_CUT(APPLET_NOEXEC(cut, cut, BB_DIR_USR_BIN, BB_SUID_DROP, cut))
  26
  27 //kbuild:lib-$(CONFIG_CUT) += cut.o
  28
  29 //usage:#define cut_trivial_usage
  30 //usage:       "[OPTIONS] [FILE]..."
  31 //usage:#define cut_full_usage "\n\n"
  32 //usage:       "Print selected fields from FILEs to stdout\n"
  33 //usage:     "\n        -b LIST Output only bytes from LIST"
  34 //usage:     "\n        -c LIST Output only characters from LIST"
  35 //usage:     "\n        -d SEP  Field delimiter for input (default -f TAB, -F run of whitespace)"
  36 //usage:     "\n        -O SEP  Field delimeter for output (default = -d for -f, one space for -F)"
  37 //usage:     "\n        -D      Don't sort/collate sections or match -fF lines without delimeter"
  38 //usage:     "\n        -f LIST Print only these fields (-d is single char)"
  39 //usage:     IF_FEATURE_CUT_REGEX(
  40 //usage:     "\n        -F LIST Print only these fields (-d is regex)"
  41 //usage:     )
  42 //usage:     "\n        -s      Output only lines containing delimiter"
  43 //usage:     "\n        -n      Ignored"
  44 //(manpage:-n   with -b: don't split multibyte characters)
  45 //usage:
  46 //usage:#define cut_example_usage
  47 //usage:       "$ echo \"Hello world\" | cut -f 1 -d ' '\n"
  48 //usage:       "Hello\n"
  49 //usage:       "$ echo \"Hello world\" | cut -f 2 -d ' '\n"
  50 //usage:       "world\n"
  51
  52 #include "libbb.h"
  53
  54 #if ENABLE_FEATURE_CUT_REGEX
  55 #include "xregex.h"
  56 #else
  57 #define regex_t int
  58 typedef struct { int rm_eo, rm_so; } regmatch_t;
  59 #define xregcomp(x, ...) *(x) = 0
  60 #define regexec(...)     0
  61 #endif
  62
  63 /* This is a NOEXEC applet. Be very careful! */
  64
  65
  66 /* option vars */
  67 #define OPT_STR "b:c:f:d:O:sD"IF_FEATURE_CUT_REGEX("F:")"n"
  68 #define CUT_OPT_BYTE_FLGS     (1 << 0)
  69 #define CUT_OPT_CHAR_FLGS     (1 << 1)
  70 #define CUT_OPT_FIELDS_FLGS   (1 << 2)
  71 #define CUT_OPT_DELIM_FLGS    (1 << 3)
  72 #define CUT_OPT_ODELIM_FLGS   (1 << 4)
  73 #define CUT_OPT_SUPPRESS_FLGS (1 << 5)
  74 #define CUT_OPT_NOSORT_FLGS   (1 << 6)
  75 #define CUT_OPT_REGEX_FLGS    ((1 << 7) * ENABLE_FEATURE_CUT_REGEX)
  76
  77 struct cut_list {
  78         int startpos;
  79         int endpos;
  80 };
  81
  82 static int cmpfunc(const void *a, const void *b)
  83 {
  84         return (((struct cut_list *) a)->startpos -
  85                         ((struct cut_list *) b)->startpos);
  86 }
  87
  88 static void cut_file(FILE *file, const char *delim, const char *odelim,
  89                 const struct cut_list *cut_lists, unsigned nlists)
  90 {
  91         char *line;
  92         unsigned linenum = 0;   /* keep these zero-based to be consistent */
  93         regex_t reg;
  94         int spos, shoe = option_mask32 & CUT_OPT_REGEX_FLGS;
  95
  96         if (shoe) xregcomp(&reg, delim, REG_EXTENDED);
  97
  98         /* go through every line in the file */
  99         while ((line = xmalloc_fgetline(file)) != NULL) {
 100
 101                 /* set up a list so we can keep track of what's been printed */
 102                 int linelen = strlen(line);
 103                 char *printed = xzalloc(linelen + 1);
 104                 char *orig_line = line;
 105                 unsigned cl_pos = 0;
 106
 107                 /* cut based on chars/bytes XXX: only works when sizeof(char) == byte */
 108                 if (option_mask32 & (CUT_OPT_CHAR_FLGS | CUT_OPT_BYTE_FLGS)) {
 109                         /* print the chars specified in each cut list */
 110                         for (; cl_pos < nlists; cl_pos++) {
 111                                 for (spos = cut_lists[cl_pos].startpos; spos < linelen;) {
 112                                         if (!printed[spos]) {
 113                                                 printed[spos] = 'X';
 114                                                 putchar(line[spos]);
 115                                         }
 116                                         if (++spos > cut_lists[cl_pos].endpos) {
 117                                                 break;
 118                                         }
 119                                 }
 120                         }
 121                 } else if (*delim == '\n') {    /* cut by lines */
 122                         spos = cut_lists[cl_pos].startpos;
 123
 124                         /* get out if we have no more lists to process or if the lines
 125                          * are lower than what we're interested in */
 126                         if (((int)linenum < spos) || (cl_pos >= nlists))
 127                                 goto next_line;
 128
 129                         /* if the line we're looking for is lower than the one we were
 130                          * passed, it means we displayed it already, so move on */
 131                         while (spos < (int)linenum) {
 132                                 spos++;
 133                                 /* go to the next list if we're at the end of this one */
 134                                 if (spos > cut_lists[cl_pos].endpos) {
 135                                         cl_pos++;
 136                                         /* get out if there's no more lists to process */
 137                                         if (cl_pos >= nlists)
 138                                                 goto next_line;
 139                                         spos = cut_lists[cl_pos].startpos;
 140                                         /* get out if the current line is lower than the one
 141                                          * we just became interested in */
 142                                         if ((int)linenum < spos)
 143                                                 goto next_line;
 144                                 }
 145                         }
 146
 147                         /* If we made it here, it means we've found the line we're
 148                          * looking for, so print it */
 149                         puts(line);
 150                         goto next_line;
 151                 } else {                /* cut by fields */
 152                         unsigned uu = 0, start = 0, end = 0, out = 0;
 153                         int dcount = 0;
 154
 155                         /* Loop through bytes, finding next delimiter */
 156                         for (;;) {
 157                                 /* End of current range? */
 158                                 if (end == linelen || dcount > cut_lists[cl_pos].endpos) {
 159                                         if (++cl_pos >= nlists) break;
 160                                         if (option_mask32 & CUT_OPT_NOSORT_FLGS)
 161                                                 start = dcount = uu = 0;
 162                                         end = 0;
 163                                 }
 164                                 /* End of current line? */
 165                                 if (uu == linelen) {
 166                                         /* If we've seen no delimiters, check -s */
 167                                         if (!cl_pos && !dcount && !shoe) {
 168                                                 if (option_mask32 & CUT_OPT_SUPPRESS_FLGS)
 169                                                         goto next_line;
 170                                         } else if (dcount < cut_lists[cl_pos].startpos)
 171                                                 start = linelen;
 172                                         end = linelen;
 173                                 } else {
 174                                         /* Find next delimiter */
 175                                         if (shoe) {
 176                                                 regmatch_t rr = {-1, -1};
 177
 178                                                 if (!regexec(&reg, line+uu, 1, &rr, REG_NOTBOL|REG_NOTEOL)) {
 179                                                         end = uu + rr.rm_so;
 180                                                         uu += rr.rm_eo;
 181                                                 } else {
 182                                                         uu = linelen;
 183                                                         continue;
 184                                                 }
 185                                         } else if (line[end = uu++] != *delim)
 186                                                 continue;
 187
 188                                         /* Got delimiter. Loop if not yet within range. */
 189                                         if (dcount++ < cut_lists[cl_pos].startpos) {
 190                                                 start = uu;
 191                                                 continue;
 192                                         }
 193                                 }
 194                                 if (end != start || !shoe)
 195                                         printf("%s%.*s", out++ ? odelim : "", end-start, line + start);
 196                                 start = uu;
 197                                 if (!dcount)
 198                                         break;
 199                         }
 200                 }
 201                 /* if we printed anything, finish with newline */
 202                 putchar('\n');
 203  next_line:
 204                 linenum++;
 205                 free(printed);
 206                 free(orig_line);
 207         }
 208 }
 209
 210 int cut_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 211 int cut_main(int argc UNUSED_PARAM, char **argv)
 212 {
 213         /* growable array holding a series of lists */
 214         struct cut_list *cut_lists = NULL;
 215         unsigned nlists = 0;    /* number of elements in above list */
 216         char *sopt, *ltok;
 217         const char *delim = NULL;
 218         const char *odelim = NULL;
 219         unsigned opt;
 220
 221 #define ARG "bcf"IF_FEATURE_CUT_REGEX("F")
 222         opt = getopt32(argv, "^"
 223                         OPT_STR  // = "b:c:f:d:O:sD"IF_FEATURE_CUT_REGEX("F:")"n"
 224                         "\0" "b--"ARG":c--"ARG":f--"ARG IF_FEATURE_CUT_REGEX("F--"ARG),
 225                         &sopt, &sopt, &sopt, &delim, &odelim IF_FEATURE_CUT_REGEX(, &sopt)
 226         );
 227         if (!delim || !*delim)
 228                 delim = (opt & CUT_OPT_REGEX_FLGS) ? "[[:space:]]+" : "\t";
 229         if (!odelim) odelim = (opt & CUT_OPT_REGEX_FLGS) ? " " : delim;
 230
 231 //      argc -= optind;
 232         argv += optind;
 233         if (!(opt & (CUT_OPT_BYTE_FLGS | CUT_OPT_CHAR_FLGS | CUT_OPT_FIELDS_FLGS | CUT_OPT_REGEX_FLGS)))
 234                 bb_simple_error_msg_and_die("expected a list of bytes, characters, or fields");
 235
 236         /*  non-field (char or byte) cutting has some special handling */
 237         if (!(opt & (CUT_OPT_FIELDS_FLGS|CUT_OPT_REGEX_FLGS))) {
 238                 static const char _op_on_field[] ALIGN1 = " only when operating on fields";
 239
 240                 if (opt & CUT_OPT_SUPPRESS_FLGS) {
 241                         bb_error_msg_and_die
 242                                 ("suppressing non-delimited lines makes sense%s", _op_on_field);
 243                 }
 244                 if (opt & CUT_OPT_DELIM_FLGS) {
 245                         bb_error_msg_and_die
 246                                 ("a delimiter may be specified%s", _op_on_field);
 247                 }
 248         }
 249
 250         /*
 251          * parse list and put values into startpos and endpos.
 252          * valid list formats: N, N-, N-M, -M
 253          * more than one list can be separated by commas
 254          */
 255         {
 256                 char *ntok;
 257                 int s = 0, e = 0;
 258
 259                 /* take apart the lists, one by one (they are separated with commas) */
 260                 while ((ltok = strsep(&sopt, ",")) != NULL) {
 261
 262                         /* it's actually legal to pass an empty list */
 263                         if (!ltok[0])
 264                                 continue;
 265
 266                         /* get the start pos */
 267                         ntok = strsep(&ltok, "-");
 268                         if (!ntok[0]) {
 269                                 s = 0;
 270                         } else {
 271                                 s = xatoi_positive(ntok);
 272                                 /* account for the fact that arrays are zero based, while
 273                                  * the user expects the first char on the line to be char #1 */
 274                                 if (s != 0)
 275                                         s--;
 276                         }
 277
 278                         /* get the end pos */
 279                         if (ltok == NULL) {
 280                                 e = s;
 281                         } else if (!ltok[0]) {
 282                                 e = INT_MAX;
 283                         } else {
 284                                 e = xatoi_positive(ltok);
 285                                 /* if the user specified and end position of 0,
 286                                  * that means "til the end of the line" */
 287                                 if (!*ltok)
 288                                         e = INT_MAX;
 289                                 else if (e < s)
 290                                         bb_error_msg_and_die("%d<%d", e, s);
 291                                 e--;    /* again, arrays are zero based, lines are 1 based */
 292                         }
 293
 294                         /* add the new list */
 295                         cut_lists = xrealloc_vector(cut_lists, 4, nlists);
 296                         /* NB: startpos is always >= 0 */
 297                         cut_lists[nlists].startpos = s;
 298                         cut_lists[nlists].endpos = e;
 299                         nlists++;
 300                 }
 301
 302                 /* make sure we got some cut positions out of all that */
 303                 if (nlists == 0)
 304                         bb_simple_error_msg_and_die("missing list of positions");
 305
 306                 /* now that the lists are parsed, we need to sort them to make life
 307                  * easier on us when it comes time to print the chars / fields / lines
 308                  */
 309                 if (!(opt & CUT_OPT_NOSORT_FLGS))
 310                         qsort(cut_lists, nlists, sizeof(cut_lists[0]), cmpfunc);
 311         }
 312
 313         {
 314                 exitcode_t retval = EXIT_SUCCESS;
 315
 316                 if (!*argv)
 317                         *--argv = (char *)"-";
 318
 319                 do {
 320                         FILE *file = fopen_or_warn_stdin(*argv);
 321                         if (!file) {
 322                                 retval = EXIT_FAILURE;
 323                                 continue;
 324                         }
 325                         cut_file(file, delim, odelim, cut_lists, nlists);
 326                         fclose_if_not_stdin(file);
 327                 } while (*++argv);
 328
 329                 if (ENABLE_FEATURE_CLEAN_UP)
 330                         free(cut_lists);
 331                 fflush_stdout_and_exit(retval);
 332         }
 333 }