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
  34 #ifndef lint
  35 static const char copyright[] =
  36 "@(#) Copyright (c) 1980, 1987, 1991, 1993\n\
  37         The Regents of the University of California.  All rights reserved.\n";
  38 #endif /* not lint */
  39
  40 #if 0
  41 #ifndef lint
  42 static char sccsid[] = "@(#)wc.c        8.1 (Berkeley) 6/6/93";
  43 #endif /* not lint */
  44 #endif
  45
  46 #include <sys/cdefs.h>
  47 __FBSDID("$FreeBSD: src/usr.bin/wc/wc.c,v 1.11.2.1 2002/08/25 02:47:04 tjr Exp $");
  48
  49 #include <sys/param.h>
  50 #include <sys/stat.h>
  51
  52 #include <ctype.h>
  53 #include <err.h>
  54 #include <errno.h>
  55 #include <fcntl.h>
  56 #include <locale.h>
  57 #include <stdio.h>
  58 #include <stdlib.h>
  59 #include <string.h>
  60 #include <unistd.h>
  61
  62 u_quad_t tlinect, twordct, tcharct;
  63 int doline, doword, dochar, domulti;
  64
  65 static int      cnt(const char *);
  66 static void     usage(void);
  67
  68 int
  69 main(argc, argv)
  70         int argc;
  71         char *argv[];
  72 {
  73         int ch, errors, total;
  74
  75         (void) setlocale(LC_CTYPE, "");
  76
  77         while ((ch = getopt(argc, argv, "clmw")) != -1)
  78                 switch((char)ch) {
  79                 case 'l':
  80                         doline = 1;
  81                         break;
  82                 case 'w':
  83                         doword = 1;
  84                         break;
  85                 case 'c':
  86                         dochar = 1;
  87                         domulti = 0;
  88                         break;
  89                 case 'm':
  90                         domulti = 1;
  91                         dochar = 0;
  92                         break;
  93                 case '?':
  94                 default:
  95                         usage();
  96                 }
  97         argv += optind;
  98         argc -= optind;
  99
 100         /* Wc's flags are on by default. */
 101         if (doline + doword + dochar + domulti == 0)
 102                 doline = doword = dochar = 1;
 103
 104         errors = 0;
 105         total = 0;
 106         if (!*argv) {
 107                 if (cnt((char *)NULL) != 0)
 108                         ++errors;
 109                 else
 110                         (void)printf("\n");
 111         }
 112         else do {
 113                 if (cnt(*argv) != 0)
 114                         ++errors;
 115                 else
 116                         (void)printf(" %s\n", *argv);
 117                 ++total;
 118         } while(*++argv);
 119
 120         if (total > 1) {
 121                 if (doline)
 122                         (void)printf(" %7qu", tlinect);
 123                 if (doword)
 124                         (void)printf(" %7qu", twordct);
 125                 if (dochar || domulti)
 126                         (void)printf(" %7qu", tcharct);
 127                 (void)printf(" total\n");
 128         }
 129         exit(errors == 0 ? 0 : 1);
 130 }
 131
 132 static int
 133 cnt(file)
 134         const char *file;
 135 {
 136         struct stat sb;
 137         u_quad_t linect, wordct, charct;
 138         ssize_t nread;
 139         int clen, fd, len, warned;
 140         short gotsp;
 141         u_char *p;
 142         u_char buf[MAXBSIZE];
 143         wchar_t wch;
 144
 145         linect = wordct = charct = 0;
 146         if (file == NULL) {
 147                 file = "stdin";
 148                 fd = STDIN_FILENO;
 149         } else {
 150                 if ((fd = open(file, O_RDONLY, 0)) < 0) {
 151                         warn("%s: open", file);
 152                         return (1);
 153                 }
 154                 if (doword || (domulti && MB_CUR_MAX != 1))
 155                         goto word;
 156                 /*
 157                  * Line counting is split out because it's a lot faster to get
 158                  * lines than to get words, since the word count requires some
 159                  * logic.
 160                  */
 161                 if (doline) {
 162                         while ((len = read(fd, buf, MAXBSIZE))) {
 163                                 if (len == -1) {
 164                                         warn("%s: read", file);
 165                                         (void)close(fd);
 166                                         return (1);
 167                                 }
 168                                 charct += len;
 169                                 for (p = buf; len--; ++p)
 170                                         if (*p == '\n')
 171                                                 ++linect;
 172                         }
 173                         tlinect += linect;
 174                         (void)printf(" %7qu", linect);
 175                         if (dochar) {
 176                                 tcharct += charct;
 177                                 (void)printf(" %7qu", charct);
 178                         }
 179                         (void)close(fd);
 180                         return (0);
 181                 }
 182                 /*
 183                  * If all we need is the number of characters and it's a
 184                  * regular file, just stat the puppy.
 185                  */
 186                 if (dochar || domulti) {
 187                         if (fstat(fd, &sb)) {
 188                                 warn("%s: fstat", file);
 189                                 (void)close(fd);
 190                                 return (1);
 191                         }
 192                         if (S_ISREG(sb.st_mode)) {
 193                                 (void)printf(" %7lld", (long long)sb.st_size);
 194                                 tcharct += sb.st_size;
 195                                 (void)close(fd);
 196                                 return (0);
 197                         }
 198                 }
 199         }
 200
 201         /* Do it the hard way... */
 202 word:   gotsp = 1;
 203         len = 0;
 204         warned = 0;
 205         while ((nread = read(fd, buf + len, MAXBSIZE - len)) != 0) {
 206                 if (nread == -1) {
 207                         warn("%s: read", file);
 208                         (void)close(fd);
 209                         return (1);
 210                 }
 211                 len += nread;
 212                 p = buf;
 213                 while (len > 0) {
 214                         if (!domulti || MB_CUR_MAX == 1) {
 215                                 clen = 1;
 216                                 wch = (unsigned char)*p;
 217                         } else if ((clen = mbtowc(&wch, p, len)) <= 0) {
 218                                 if (len > MB_CUR_MAX) {
 219                                         clen = 1;
 220                                         wch = (unsigned char)*p;
 221                                         if (!warned) {
 222                                                 errno = EILSEQ;
 223                                                 warn("%s", file);
 224                                                 warned = 1;
 225                                         }
 226                                 } else {
 227                                         memmove(buf, p, len);
 228                                         break;
 229                                 }
 230                         }
 231                         charct++;
 232                         len -= clen;
 233                         p += clen;
 234                         if (wch == L'\n')
 235                                 ++linect;
 236                         if (isspace(wch))
 237                                 gotsp = 1;
 238                         else if (gotsp) {
 239                                 gotsp = 0;
 240                                 ++wordct;
 241                         }
 242                 }
 243         }
 244         if (doline) {
 245                 tlinect += linect;
 246                 (void)printf(" %7qu", linect);
 247         }
 248         if (doword) {
 249                 twordct += wordct;
 250                 (void)printf(" %7qu", wordct);
 251         }
 252         if (dochar || domulti) {
 253                 tcharct += charct;
 254                 (void)printf(" %7qu", charct);
 255         }
 256         (void)close(fd);
 257         return (0);
 258 }
 259
 260 static void
 261 usage()
 262 {
 263         (void)fprintf(stderr, "usage: wc [-clmw] [file ...]\n");
 264         exit(1);
 265 }