coreutils/wc.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * wc implementation for busybox
   4  *
   5  * Copyright (C) 2003  Manuel Novoa III  <mjn3@codepoet.org>
   6  *
   7  * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
   8  */
   9
  10 /* BB_AUDIT SUSv3 _NOT_ compliant -- option -m is not currently supported. */
  11 /* http://www.opengroup.org/onlinepubs/007904975/utilities/wc.html */
  12
  13 /* Mar 16, 2003      Manuel Novoa III   (mjn3@codepoet.org)
  14  *
  15  * Rewritten to fix a number of problems and do some size optimizations.
  16  * Problems in the previous busybox implementation (besides bloat) included:
  17  *  1) broken 'wc -c' optimization (read note below)
  18  *  2) broken handling of '-' args
  19  *  3) no checking of ferror on EOF returns
  20  *  4) isprint() wasn't considered when word counting.
  21  *
  22  * TODO:
  23  *
  24  * When locale support is enabled, count multibyte chars in the '-m' case.
  25  *
  26  * NOTES:
  27  *
  28  * The previous busybox wc attempted an optimization using stat for the
  29  * case of counting chars only.  I omitted that because it was broken.
  30  * It didn't take into account the possibility of input coming from a
  31  * pipe, or input from a file with file pointer not at the beginning.
  32  *
  33  * To implement such a speed optimization correctly, not only do you
  34  * need the size, but also the file position.  Note also that the
  35  * file position may be past the end of file.  Consider the example
  36  * (adapted from example in gnu wc.c)
  37  *
  38  *      echo hello > /tmp/testfile &&
  39  *      (dd ibs=1k skip=1 count=0 &> /dev/null; wc -c) < /tmp/testfile
  40  *
  41  * for which 'wc -c' should output '0'.
  42  */
  43
  44 #include "libbb.h"
  45
  46 #if !ENABLE_LOCALE_SUPPORT
  47 # undef isprint
  48 # undef isspace
  49 # define isprint(c) ((unsigned)((c) - 0x20) <= (0x7e - 0x20))
  50 # define isspace(c) ((c) == ' ')
  51 #endif
  52
  53 #if ENABLE_FEATURE_WC_LARGE
  54 # define COUNT_T unsigned long long
  55 # define COUNT_FMT "llu"
  56 #else
  57 # define COUNT_T unsigned
  58 # define COUNT_FMT "u"
  59 #endif
  60
  61 enum {
  62         WC_LINES        = 0,
  63         WC_WORDS        = 1,
  64         WC_CHARS        = 2,
  65         WC_LENGTH       = 3
  66 };
  67
  68 int wc_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  69 int wc_main(int argc UNUSED_PARAM, char **argv)
  70 {
  71         const char *arg;
  72         const char *start_fmt = " %9"COUNT_FMT + 1;
  73         const char *fname_fmt = " %s\n";
  74         COUNT_T *pcounts;
  75         COUNT_T counts[4];
  76         COUNT_T totals[4];
  77         int num_files;
  78         smallint status = EXIT_SUCCESS;
  79         unsigned print_type;
  80
  81         print_type = getopt32(argv, "lwcL");
  82
  83         if (print_type == 0) {
  84                 print_type = (1 << WC_LINES) | (1 << WC_WORDS) | (1 << WC_CHARS);
  85         }
  86
  87         argv += optind;
  88         if (!argv[0]) {
  89                 *--argv = (char *) bb_msg_standard_input;
  90                 fname_fmt = "\n";
  91                 if (!((print_type-1) & print_type)) /* exactly one option? */
  92                         start_fmt = "%"COUNT_FMT;
  93         }
  94
  95         memset(totals, 0, sizeof(totals));
  96
  97         pcounts = counts;
  98
  99         num_files = 0;
 100         while ((arg = *argv++) != 0) {
 101                 FILE *fp;
 102                 const char *s;
 103                 unsigned u;
 104                 unsigned linepos;
 105                 smallint in_word;
 106
 107                 ++num_files;
 108                 fp = fopen_or_warn_stdin(arg);
 109                 if (!fp) {
 110                         status = EXIT_FAILURE;
 111                         continue;
 112                 }
 113
 114                 memset(counts, 0, sizeof(counts));
 115                 linepos = 0;
 116                 in_word = 0;
 117
 118                 do {
 119                         int c;
 120                         /* Our -w doesn't match GNU wc exactly... oh well */
 121
 122                         ++counts[WC_CHARS];
 123                         c = getc(fp);
 124                         if (c == EOF) {
 125                                 if (ferror(fp)) {
 126                                         bb_simple_perror_msg(arg);
 127                                         status = EXIT_FAILURE;
 128                                 }
 129                                 --counts[WC_CHARS];
 130                                 goto DO_EOF;            /* Treat an EOF as '\r'. */
 131                         }
 132                         if (isprint_asciionly(c)) {
 133                                 ++linepos;
 134                                 if (!isspace(c)) {
 135                                         in_word = 1;
 136                                         continue;
 137                                 }
 138                         } else if ((unsigned)(c - 9) <= 4) {
 139                                 /* \t  9
 140                                  * \n 10
 141                                  * \v 11
 142                                  * \f 12
 143                                  * \r 13
 144                                  */
 145                                 if (c == '\t') {
 146                                         linepos = (linepos | 7) + 1;
 147                                 } else {                        /* '\n', '\r', '\f', or '\v' */
 148  DO_EOF:
 149                                         if (linepos > counts[WC_LENGTH]) {
 150                                                 counts[WC_LENGTH] = linepos;
 151                                         }
 152                                         if (c == '\n') {
 153                                                 ++counts[WC_LINES];
 154                                         }
 155                                         if (c != '\v') {
 156                                                 linepos = 0;
 157                                         }
 158                                 }
 159                         } else {
 160                                 continue;
 161                         }
 162
 163                         counts[WC_WORDS] += in_word;
 164                         in_word = 0;
 165                         if (c == EOF) {
 166                                 break;
 167                         }
 168                 } while (1);
 169
 170                 if (totals[WC_LENGTH] < counts[WC_LENGTH]) {
 171                         totals[WC_LENGTH] = counts[WC_LENGTH];
 172                 }
 173                 totals[WC_LENGTH] -= counts[WC_LENGTH];
 174
 175                 fclose_if_not_stdin(fp);
 176
 177  OUTPUT:
 178                 /* coreutils wc tries hard to print pretty columns
 179                  * (saves results for all files, find max col len etc...)
 180                  * we won't try that hard, it will bloat us too much */
 181                 s = start_fmt;
 182                 u = 0;
 183                 do {
 184                         if (print_type & (1 << u)) {
 185                                 printf(s, pcounts[u]);
 186                                 s = " %9"COUNT_FMT; /* Ok... restore the leading space. */
 187                         }
 188                         totals[u] += pcounts[u];
 189                 } while (++u < 4);
 190                 printf(fname_fmt, arg);
 191         }
 192
 193         /* If more than one file was processed, we want the totals.  To save some
 194          * space, we set the pcounts ptr to the totals array.  This has the side
 195          * effect of trashing the totals array after outputting it, but that's
 196          * irrelavent since we no longer need it. */
 197         if (num_files > 1) {
 198                 num_files = 0;                          /* Make sure we don't get here again. */
 199                 arg = "total";
 200                 pcounts = totals;
 201                 --argv;
 202                 goto OUTPUT;
 203         }
 204
 205         fflush_stdout_and_exit(status);
 206 }