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 1.2 2003/06/17 04:29:33 dillon 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(argc, argv)
  60         int argc;
  61         char *argv[];
  62 {
  63         int ch, errors, total;
  64
  65         (void) setlocale(LC_CTYPE, "");
  66
  67         while ((ch = getopt(argc, argv, "clmw")) != -1)
  68                 switch((char)ch) {
  69                 case 'l':
  70                         doline = 1;
  71                         break;
  72                 case 'w':
  73                         doword = 1;
  74                         break;
  75                 case 'c':
  76                         dochar = 1;
  77                         domulti = 0;
  78                         break;
  79                 case 'm':
  80                         domulti = 1;
  81                         dochar = 0;
  82                         break;
  83                 case '?':
  84                 default:
  85                         usage();
  86                 }
  87         argv += optind;
  88         argc -= optind;
  89
  90         /* Wc's flags are on by default. */
  91         if (doline + doword + dochar + domulti == 0)
  92                 doline = doword = dochar = 1;
  93
  94         errors = 0;
  95         total = 0;
  96         if (!*argv) {
  97                 if (cnt((char *)NULL) != 0)
  98                         ++errors;
  99                 else
 100                         (void)printf("\n");
 101         }
 102         else do {
 103                 if (cnt(*argv) != 0)
 104                         ++errors;
 105                 else
 106                         (void)printf(" %s\n", *argv);
 107                 ++total;
 108         } while(*++argv);
 109
 110         if (total > 1) {
 111                 if (doline)
 112                         (void)printf(" %7qu", tlinect);
 113                 if (doword)
 114                         (void)printf(" %7qu", twordct);
 115                 if (dochar || domulti)
 116                         (void)printf(" %7qu", tcharct);
 117                 (void)printf(" total\n");
 118         }
 119         exit(errors == 0 ? 0 : 1);
 120 }
 121
 122 static int
 123 cnt(file)
 124         const char *file;
 125 {
 126         struct stat sb;
 127         u_quad_t linect, wordct, charct;
 128         ssize_t nread;
 129         int clen, fd, len, warned;
 130         short gotsp;
 131         u_char *p;
 132         u_char buf[MAXBSIZE];
 133         wchar_t wch;
 134
 135         linect = wordct = charct = 0;
 136         if (file == NULL) {
 137                 file = "stdin";
 138                 fd = STDIN_FILENO;
 139         } else {
 140                 if ((fd = open(file, O_RDONLY, 0)) < 0) {
 141                         warn("%s: open", file);
 142                         return (1);
 143                 }
 144                 if (doword || (domulti && MB_CUR_MAX != 1))
 145                         goto word;
 146                 /*
 147                  * Line counting is split out because it's a lot faster to get
 148                  * lines than to get words, since the word count requires some
 149                  * logic.
 150                  */
 151                 if (doline) {
 152                         while ((len = read(fd, buf, MAXBSIZE))) {
 153                                 if (len == -1) {
 154                                         warn("%s: read", file);
 155                                         (void)close(fd);
 156                                         return (1);
 157                                 }
 158                                 charct += len;
 159                                 for (p = buf; len--; ++p)
 160                                         if (*p == '\n')
 161                                                 ++linect;
 162                         }
 163                         tlinect += linect;
 164                         (void)printf(" %7qu", linect);
 165                         if (dochar) {
 166                                 tcharct += charct;
 167                                 (void)printf(" %7qu", charct);
 168                         }
 169                         (void)close(fd);
 170                         return (0);
 171                 }
 172                 /*
 173                  * If all we need is the number of characters and it's a
 174                  * regular file, just stat the puppy.
 175                  */
 176                 if (dochar || domulti) {
 177                         if (fstat(fd, &sb)) {
 178                                 warn("%s: fstat", file);
 179                                 (void)close(fd);
 180                                 return (1);
 181                         }
 182                         if (S_ISREG(sb.st_mode)) {
 183                                 (void)printf(" %7lld", (long long)sb.st_size);
 184                                 tcharct += sb.st_size;
 185                                 (void)close(fd);
 186                                 return (0);
 187                         }
 188                 }
 189         }
 190
 191         /* Do it the hard way... */
 192 word:   gotsp = 1;
 193         len = 0;
 194         warned = 0;
 195         while ((nread = read(fd, buf + len, MAXBSIZE - len)) != 0) {
 196                 if (nread == -1) {
 197                         warn("%s: read", file);
 198                         (void)close(fd);
 199                         return (1);
 200                 }
 201                 len += nread;
 202                 p = buf;
 203                 while (len > 0) {
 204                         if (!domulti || MB_CUR_MAX == 1) {
 205                                 clen = 1;
 206                                 wch = (unsigned char)*p;
 207                         } else if ((clen = mbtowc(&wch, p, len)) <= 0) {
 208                                 if (len > MB_CUR_MAX) {
 209                                         clen = 1;
 210                                         wch = (unsigned char)*p;
 211                                         if (!warned) {
 212                                                 errno = EILSEQ;
 213                                                 warn("%s", file);
 214                                                 warned = 1;
 215                                         }
 216                                 } else {
 217                                         memmove(buf, p, len);
 218                                         break;
 219                                 }
 220                         }
 221                         charct++;
 222                         len -= clen;
 223                         p += clen;
 224                         if (wch == L'\n')
 225                                 ++linect;
 226                         if (isspace(wch))
 227                                 gotsp = 1;
 228                         else if (gotsp) {
 229                                 gotsp = 0;
 230                                 ++wordct;
 231                         }
 232                 }
 233         }
 234         if (doline) {
 235                 tlinect += linect;
 236                 (void)printf(" %7qu", linect);
 237         }
 238         if (doword) {
 239                 twordct += wordct;
 240                 (void)printf(" %7qu", wordct);
 241         }
 242         if (dochar || domulti) {
 243                 tcharct += charct;
 244                 (void)printf(" %7qu", charct);
 245         }
 246         (void)close(fd);
 247         return (0);
 248 }
 249
 250 static void
 251 usage()
 252 {
 253         (void)fprintf(stderr, "usage: wc [-clmw] [file ...]\n");
 254         exit(1);
 255 }