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. All advertising materials mentioning features or use of this software
  14  *    must display the following acknowledgement:
  15  *      This product includes software developed by the University of
  16  *      California, Berkeley and its contributors.
  17  * 4. Neither the name of the University nor the names of its contributors
  18  *    may be used to endorse or promote products derived from this software
  19  *    without specific prior written permission.
  20  *
  21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  31  * SUCH DAMAGE.
  32  *
  33  * @(#) Copyright (c) 1980, 1987, 1991, 1993 The Regents of the University of California.  All rights reserved.
  34  * @(#)wc.c     8.1 (Berkeley) 6/6/93
  35  * $FreeBSD: src/usr.bin/wc/wc.c,v 1.11.2.1 2002/08/25 02:47:04 tjr Exp $
  36  * $DragonFly: src/usr.bin/wc/wc.c,v 1.3 2003/10/04 20:36:55 hmp Exp $
  37  */
  38
  39 #include <sys/param.h>
  40 #include <sys/stat.h>
  41
  42 #include <ctype.h>
  43 #include <err.h>
  44 #include <errno.h>
  45 #include <fcntl.h>
  46 #include <locale.h>
  47 #include <stdio.h>
  48 #include <stdlib.h>
  49 #include <string.h>
  50 #include <unistd.h>
  51
  52 u_quad_t tlinect, twordct, tcharct;
  53 int doline, doword, dochar, domulti;
  54
  55 static int      cnt(const char *);
  56 static void     usage(void);
  57
  58 int
  59 main(int argc, char **argv)
  60 {
  61         int ch, errors, total;
  62
  63         (void) setlocale(LC_CTYPE, "");
  64
  65         while ((ch = getopt(argc, argv, "clmw")) != -1)
  66                 switch((char)ch) {
  67                 case 'l':
  68                         doline = 1;
  69                         break;
  70                 case 'w':
  71                         doword = 1;
  72                         break;
  73                 case 'c':
  74                         dochar = 1;
  75                         domulti = 0;
  76                         break;
  77                 case 'm':
  78                         domulti = 1;
  79                         dochar = 0;
  80                         break;
  81                 case '?':
  82                 default:
  83                         usage();
  84                 }
  85         argv += optind;
  86         argc -= optind;
  87
  88         /* Wc's flags are on by default. */
  89         if (doline + doword + dochar + domulti == 0)
  90                 doline = doword = dochar = 1;
  91
  92         errors = 0;
  93         total = 0;
  94         if (!*argv) {
  95                 if (cnt((char *)NULL) != 0)
  96                         ++errors;
  97                 else
  98                         (void)printf("\n");
  99         }
 100         else do {
 101                 if (cnt(*argv) != 0)
 102                         ++errors;
 103                 else
 104                         (void)printf(" %s\n", *argv);
 105                 ++total;
 106         } while(*++argv);
 107
 108         if (total > 1) {
 109                 if (doline)
 110                         (void)printf(" %7qu", tlinect);
 111                 if (doword)
 112                         (void)printf(" %7qu", twordct);
 113                 if (dochar || domulti)
 114                         (void)printf(" %7qu", tcharct);
 115                 (void)printf(" total\n");
 116         }
 117         exit(errors == 0 ? 0 : 1);
 118 }
 119
 120 static int
 121 cnt(const char *file)
 122 {
 123         struct stat sb;
 124         u_quad_t linect, wordct, charct;
 125         ssize_t nread;
 126         int clen, fd, len, warned;
 127         short gotsp;
 128         u_char *p;
 129         u_char buf[MAXBSIZE];
 130         wchar_t wch;
 131
 132         linect = wordct = charct = 0;
 133         if (file == NULL) {
 134                 file = "stdin";
 135                 fd = STDIN_FILENO;
 136         } else {
 137                 if ((fd = open(file, O_RDONLY, 0)) < 0) {
 138                         warn("%s: open", file);
 139                         return (1);
 140                 }
 141                 if (doword || (domulti && MB_CUR_MAX != 1))
 142                         goto word;
 143                 /*
 144                  * Line counting is split out because it's a lot faster to get
 145                  * lines than to get words, since the word count requires some
 146                  * logic.
 147                  */
 148                 if (doline) {
 149                         while ((len = read(fd, buf, MAXBSIZE))) {
 150                                 if (len == -1) {
 151                                         warn("%s: read", file);
 152                                         (void)close(fd);
 153                                         return (1);
 154                                 }
 155                                 charct += len;
 156                                 for (p = buf; len--; ++p)
 157                                         if (*p == '\n')
 158                                                 ++linect;
 159                         }
 160                         tlinect += linect;
 161                         (void)printf(" %7qu", linect);
 162                         if (dochar) {
 163                                 tcharct += charct;
 164                                 (void)printf(" %7qu", charct);
 165                         }
 166                         (void)close(fd);
 167                         return (0);
 168                 }
 169                 /*
 170                  * If all we need is the number of characters and it's a
 171                  * regular file, just stat the puppy.
 172                  */
 173                 if (dochar || domulti) {
 174                         if (fstat(fd, &sb)) {
 175                                 warn("%s: fstat", file);
 176                                 (void)close(fd);
 177                                 return (1);
 178                         }
 179                         if (S_ISREG(sb.st_mode)) {
 180                                 (void)printf(" %7lld", (long long)sb.st_size);
 181                                 tcharct += sb.st_size;
 182                                 (void)close(fd);
 183                                 return (0);
 184                         }
 185                 }
 186         }
 187
 188         /* Do it the hard way... */
 189 word:   gotsp = 1;
 190         len = 0;
 191         warned = 0;
 192         while ((nread = read(fd, buf + len, MAXBSIZE - len)) != 0) {
 193                 if (nread == -1) {
 194                         warn("%s: read", file);
 195                         (void)close(fd);
 196                         return (1);
 197                 }
 198                 len += nread;
 199                 p = buf;
 200                 while (len > 0) {
 201                         if (!domulti || MB_CUR_MAX == 1) {
 202                                 clen = 1;
 203                                 wch = (unsigned char)*p;
 204                         } else if ((clen = mbtowc(&wch, p, len)) <= 0) {
 205                                 if (len > MB_CUR_MAX) {
 206                                         clen = 1;
 207                                         wch = (unsigned char)*p;
 208                                         if (!warned) {
 209                                                 errno = EILSEQ;
 210                                                 warn("%s", file);
 211                                                 warned = 1;
 212                                         }
 213                                 } else {
 214                                         memmove(buf, p, len);
 215                                         break;
 216                                 }
 217                         }
 218                         charct++;
 219                         len -= clen;
 220                         p += clen;
 221                         if (wch == L'\n')
 222                                 ++linect;
 223                         if (isspace(wch))
 224                                 gotsp = 1;
 225                         else if (gotsp) {
 226                                 gotsp = 0;
 227                                 ++wordct;
 228                         }
 229                 }
 230         }
 231         if (doline) {
 232                 tlinect += linect;
 233                 (void)printf(" %7qu", linect);
 234         }
 235         if (doword) {
 236                 twordct += wordct;
 237                 (void)printf(" %7qu", wordct);
 238         }
 239         if (dochar || domulti) {
 240                 tcharct += charct;
 241                 (void)printf(" %7qu", charct);
 242         }
 243         (void)close(fd);
 244         return (0);
 245 }
 246
 247 static void
 248 usage(void)
 249 {
 250         (void)fprintf(stderr, "usage: wc [-clmw] [file ...]\n");
 251         exit(1);
 252 }