usr.bin/wc/wc.c

   1 /*
   2  * Copyright (c) 1980, 1987, 1991, 1993
   3  *      The Regents of the University of California.  All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  * 3. Neither the name of the University nor the names of its contributors
  14  *    may be used to endorse or promote products derived from this software
  15  *    without specific prior written permission.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  27  * SUCH DAMAGE.
  28  *
  29  * @(#) Copyright (c) 1980, 1987, 1991, 1993 The Regents of the University of California.  All rights reserved.
  30  * @(#)wc.c     8.1 (Berkeley) 6/6/93
  31  * $FreeBSD: src/usr.bin/wc/wc.c,v 1.11.2.1 2002/08/25 02:47:04 tjr Exp $
  32  * $DragonFly: src/usr.bin/wc/wc.c,v 1.5 2005/02/05 16:07:08 liamfoy Exp $
  33  */
  34
  35 #include <sys/param.h>
  36 #include <sys/stat.h>
  37
  38 #include <ctype.h>
  39 #include <err.h>
  40 #include <errno.h>
  41 #include <fcntl.h>
  42 #include <locale.h>
  43 #include <stdio.h>
  44 #include <stdlib.h>
  45 #include <string.h>
  46 #include <unistd.h>
  47
  48 uintmax_t tlinect, twordct, tcharct;
  49 int doline, doword, dochar, domulti;
  50
  51 static int      cnt(const char *);
  52 static void     usage(void);
  53
  54 int
  55 main(int argc, char **argv)
  56 {
  57         int ch, errors, total;
  58
  59         setlocale(LC_CTYPE, "");
  60
  61         while ((ch = getopt(argc, argv, "clmw")) != -1) {
  62                 switch (ch) {
  63                 case 'l':
  64                         doline = 1;
  65                         break;
  66                 case 'w':
  67                         doword = 1;
  68                         break;
  69                 case 'c':
  70                         dochar = 1;
  71                         domulti = 0;
  72                         break;
  73                 case 'm':
  74                         domulti = 1;
  75                         dochar = 0;
  76                         break;
  77                 case '?':
  78                 default:
  79                         usage();
  80                 }
  81         }
  82         argv += optind;
  83         argc -= optind;
  84
  85         /* Wc's flags are on by default. */
  86         if (doline + doword + dochar + domulti == 0)
  87                 doline = doword = dochar = 1;
  88
  89         errors = 0;
  90         total = 0;
  91         if (*argv == NULL) {
  92                 if (cnt(NULL) != 0)
  93                         ++errors;
  94                 else
  95                         printf("\n");
  96         }
  97         else do {
  98                 if (cnt(*argv) != 0)
  99                         ++errors;
 100                 else
 101                         printf(" %s\n", *argv);
 102                 ++total;
 103         } while(*++argv);
 104
 105         if (total > 1) {
 106                 if (doline)
 107                         printf(" %7ju", tlinect);
 108                 if (doword)
 109                         printf(" %7ju", twordct);
 110                 if (dochar || domulti)
 111                         printf(" %7ju", tcharct);
 112                 printf(" total\n");
 113         }
 114         exit(errors == 0 ? 0 : 1);
 115 }
 116
 117 static int
 118 cnt(const char *file)
 119 {
 120         struct stat sb;
 121         u_quad_t linect, wordct, charct;
 122         ssize_t nread;
 123         int clen, fd, len, warned;
 124         short gotsp;
 125         u_char *p;
 126         u_char buf[MAXBSIZE];
 127         wchar_t wch;
 128
 129         linect = wordct = charct = 0;
 130         if (file == NULL) {
 131                 file = "stdin";
 132                 fd = STDIN_FILENO;
 133         } else {
 134                 if ((fd = open(file, O_RDONLY)) < 0) {
 135                         warn("%s: open", file);
 136                         return (1);
 137                 }
 138                 if (doword || (domulti && MB_CUR_MAX != 1))
 139                         goto word;
 140                 /*
 141                  * Line counting is split out because it's a lot faster to get
 142                  * lines than to get words, since the word count requires some
 143                  * logic.
 144                  */
 145                 if (doline) {
 146                         while ((len = read(fd, buf, MAXBSIZE))) {
 147                                 if (len == -1) {
 148                                         warn("%s: read", file);
 149                                         close(fd);
 150                                         return (1);
 151                                 }
 152                                 charct += len;
 153                                 for (p = buf; len--; ++p) {
 154                                         if (*p == '\n')
 155                                                 ++linect;
 156                                 }
 157                         }
 158                         tlinect += linect;
 159                         printf(" %7ju", linect);
 160                         if (dochar) {
 161                                 tcharct += charct;
 162                                 printf(" %7ju", charct);
 163                         }
 164                         close(fd);
 165                         return (0);
 166                 }
 167                 /*
 168                  * If all we need is the number of characters and it's a
 169                  * regular file, just stat the puppy.
 170                  */
 171                 if (dochar || domulti) {
 172                         if (fstat(fd, &sb)) {
 173                                 warn("%s: fstat", file);
 174                                 close(fd);
 175                                 return (1);
 176                         }
 177                         if (S_ISREG(sb.st_mode)) {
 178                                 printf(" %7lld", (long long)sb.st_size);
 179                                 tcharct += sb.st_size;
 180                                 close(fd);
 181                                 return (0);
 182                         }
 183                 }
 184         }
 185
 186         /* Do it the hard way... */
 187 word:   gotsp = 1;
 188         len = 0;
 189         warned = 0;
 190         while ((nread = read(fd, buf + len, MAXBSIZE - len)) != 0) {
 191                 if (nread == -1) {
 192                         warn("%s: read", file);
 193                         close(fd);
 194                         return (1);
 195                 }
 196                 len += nread;
 197                 p = buf;
 198                 while (len > 0) {
 199                         if (!domulti || MB_CUR_MAX == 1) {
 200                                 clen = 1;
 201                                 wch = (unsigned char)*p;
 202                         } else if ((clen = mbtowc(&wch, p, len)) <= 0) {
 203                                 if (len > MB_CUR_MAX) {
 204                                         clen = 1;
 205                                         wch = (unsigned char)*p;
 206                                         if (!warned) {
 207                                                 errno = EILSEQ;
 208                                                 warn("%s", file);
 209                                                 warned = 1;
 210                                         }
 211                                 } else {
 212                                         memmove(buf, p, len);
 213                                         break;
 214                                 }
 215                         }
 216                         charct++;
 217                         len -= clen;
 218                         p += clen;
 219                         if (wch == L'\n')
 220                                 ++linect;
 221                         if (isspace(wch))
 222                                 gotsp = 1;
 223                         else if (gotsp) {
 224                                 gotsp = 0;
 225                                 ++wordct;
 226                         }
 227                 }
 228         }
 229         if (doline) {
 230                 tlinect += linect;
 231                 printf(" %7ju", linect);
 232         }
 233         if (doword) {
 234                 twordct += wordct;
 235                 printf(" %7ju", wordct);
 236         }
 237         if (dochar || domulti) {
 238                 tcharct += charct;
 239                 printf(" %7ju", charct);
 240         }
 241         close(fd);
 242         return (0);
 243 }
 244
 245 static void
 246 usage(void)
 247 {
 248         fprintf(stderr, "usage: wc [-clmw] [file ...]\n");
 249         exit(1);
 250 }