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("$FreeBSD$");
  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 <stdint.h>
  58 #include <stdio.h>
  59 #include <stdlib.h>
  60 #include <string.h>
  61 #include <unistd.h>
  62 #include <wchar.h>
  63 #include <wctype.h>
  64
  65 uintmax_t tlinect, twordct, tcharct, tlongline;
  66 int doline, doword, dochar, domulti, dolongline;
  67
  68 static int      cnt(const char *);
  69 static void     usage(void);
  70
  71 int
  72 main(int argc, char *argv[])
  73 {
  74         int ch, errors, total;
  75
  76         (void) setlocale(LC_CTYPE, "");
  77
  78         while ((ch = getopt(argc, argv, "clmwL")) != -1)
  79                 switch((char)ch) {
  80                 case 'l':
  81                         doline = 1;
  82                         break;
  83                 case 'w':
  84                         doword = 1;
  85                         break;
  86                 case 'c':
  87                         dochar = 1;
  88                         domulti = 0;
  89                         break;
  90                 case 'L':
  91                         dolongline = 1;
  92                         break;
  93                 case 'm':
  94                         domulti = 1;
  95                         dochar = 0;
  96                         break;
  97                 case '?':
  98                 default:
  99                         usage();
 100                 }
 101         argv += optind;
 102         argc -= optind;
 103
 104         /* Wc's flags are on by default. */
 105         if (doline + doword + dochar + domulti + dolongline == 0)
 106                 doline = doword = dochar = 1;
 107
 108         errors = 0;
 109         total = 0;
 110         if (!*argv) {
 111                 if (cnt((char *)NULL) != 0)
 112                         ++errors;
 113                 else
 114                         (void)printf("\n");
 115         }
 116         else do {
 117                 if (cnt(*argv) != 0)
 118                         ++errors;
 119                 else
 120                         (void)printf(" %s\n", *argv);
 121                 ++total;
 122         } while(*++argv);
 123
 124         if (total > 1) {
 125                 if (doline)
 126                         (void)printf(" %7ju", tlinect);
 127                 if (doword)
 128                         (void)printf(" %7ju", twordct);
 129                 if (dochar || domulti)
 130                         (void)printf(" %7ju", tcharct);
 131                 if (dolongline)
 132                         (void)printf(" %7ju", tlongline);
 133                 (void)printf(" total\n");
 134         }
 135         exit(errors == 0 ? 0 : 1);
 136 }
 137
 138 static int
 139 cnt(const char *file)
 140 {
 141         struct stat sb;
 142         uintmax_t linect, wordct, charct, llct, tmpll;
 143         int fd, len, warned;
 144         size_t clen;
 145         short gotsp;
 146         u_char *p;
 147         u_char buf[MAXBSIZE];
 148         wchar_t wch;
 149         mbstate_t mbs;
 150
 151         linect = wordct = charct = llct = tmpll = 0;
 152         if (file == NULL) {
 153                 file = "stdin";
 154                 fd = STDIN_FILENO;
 155         } else {
 156                 if ((fd = open(file, O_RDONLY, 0)) < 0) {
 157                         warn("%s: open", file);
 158                         return (1);
 159                 }
 160                 if (doword || (domulti && MB_CUR_MAX != 1))
 161                         goto word;
 162                 /*
 163                  * Line counting is split out because it's a lot faster to get
 164                  * lines than to get words, since the word count requires some
 165                  * logic.
 166                  */
 167                 if (doline) {
 168                         while ((len = read(fd, buf, MAXBSIZE))) {
 169                                 if (len == -1) {
 170                                         warn("%s: read", file);
 171                                         (void)close(fd);
 172                                         return (1);
 173                                 }
 174                                 charct += len;
 175                                 for (p = buf; len--; ++p)
 176                                         if (*p == '\n') {
 177                                                 if (tmpll > llct)
 178                                                         llct = tmpll;
 179                                                 tmpll = 0;
 180                                                 ++linect;
 181                                         } else
 182                                                 tmpll++;
 183                         }
 184                         tlinect += linect;
 185                         (void)printf(" %7ju", linect);
 186                         if (dochar) {
 187                                 tcharct += charct;
 188                                 (void)printf(" %7ju", charct);
 189                         }
 190                         if (dolongline) {
 191                                 if (llct > tlongline)
 192                                         tlongline = llct;
 193                                 (void)printf(" %7ju", tlongline);
 194                         }
 195                         (void)close(fd);
 196                         return (0);
 197                 }
 198                 /*
 199                  * If all we need is the number of characters and it's a
 200                  * regular file, just stat the puppy.
 201                  */
 202                 if (dochar || domulti) {
 203                         if (fstat(fd, &sb)) {
 204                                 warn("%s: fstat", file);
 205                                 (void)close(fd);
 206                                 return (1);
 207                         }
 208                         if (S_ISREG(sb.st_mode)) {
 209                                 (void)printf(" %7lld", (long long)sb.st_size);
 210                                 tcharct += sb.st_size;
 211                                 (void)close(fd);
 212                                 return (0);
 213                         }
 214                 }
 215         }
 216
 217         /* Do it the hard way... */
 218 word:   gotsp = 1;
 219         warned = 0;
 220         memset(&mbs, 0, sizeof(mbs));
 221         while ((len = read(fd, buf, MAXBSIZE)) != 0) {
 222                 if (len == -1) {
 223                         warn("%s: read", file);
 224                         (void)close(fd);
 225                         return (1);
 226                 }
 227                 p = buf;
 228                 while (len > 0) {
 229                         if (!domulti || MB_CUR_MAX == 1) {
 230                                 clen = 1;
 231                                 wch = (unsigned char)*p;
 232                         } else if ((clen = mbrtowc(&wch, p, len, &mbs)) ==
 233                             (size_t)-1) {
 234                                 if (!warned) {
 235                                         errno = EILSEQ;
 236                                         warn("%s", file);
 237                                         warned = 1;
 238                                 }
 239                                 memset(&mbs, 0, sizeof(mbs));
 240                                 clen = 1;
 241                                 wch = (unsigned char)*p;
 242                         } else if (clen == (size_t)-2)
 243                                 break;
 244                         else if (clen == 0)
 245                                 clen = 1;
 246                         charct++;
 247                         if (wch != L'\n')
 248                                 tmpll++;
 249                         len -= clen;
 250                         p += clen;
 251                         if (wch == L'\n') {
 252                                 if (tmpll > llct)
 253                                         llct = tmpll;
 254                                 tmpll = 0;
 255                                 ++linect;
 256                         }
 257                         if (iswspace(wch))
 258                                 gotsp = 1;
 259                         else if (gotsp) {
 260                                 gotsp = 0;
 261                                 ++wordct;
 262                         }
 263                 }
 264         }
 265         if (domulti && MB_CUR_MAX > 1)
 266                 if (mbrtowc(NULL, NULL, 0, &mbs) == (size_t)-1 && !warned)
 267                         warn("%s", file);
 268         if (doline) {
 269                 tlinect += linect;
 270                 (void)printf(" %7ju", linect);
 271         }
 272         if (doword) {
 273                 twordct += wordct;
 274                 (void)printf(" %7ju", wordct);
 275         }
 276         if (dochar || domulti) {
 277                 tcharct += charct;
 278                 (void)printf(" %7ju", charct);
 279         }
 280         if (dolongline) {
 281                 if (llct > tlongline)
 282                         tlongline = llct;
 283                 (void)printf(" %7ju", llct);
 284         }
 285         (void)close(fd);
 286         return (0);
 287 }
 288
 289 static void
 290 usage()
 291 {
 292         (void)fprintf(stderr, "usage: wc [-Lclmw] [file ...]\n");
 293         exit(1);
 294 }