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  * 4. 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
  30 #ifndef lint
  31 static const char copyright[] =
  32 "@(#) Copyright (c) 1980, 1987, 1991, 1993\n\
  33         The Regents of the University of California.  All rights reserved.\n";
  34 #endif /* not lint */
  35
  36 #if 0
  37 #ifndef lint
  38 static char sccsid[] = "@(#)wc.c        8.1 (Berkeley) 6/6/93";
  39 #endif /* not lint */
  40 #endif
  41
  42 #include <sys/cdefs.h>
  43 __FBSDID("$FreeBSD$");
  44
  45 #include <sys/param.h>
  46 #include <sys/stat.h>
  47
  48 #include <ctype.h>
  49 #include <err.h>
  50 #include <errno.h>
  51 #include <fcntl.h>
  52 #include <locale.h>
  53 #include <stdint.h>
  54 #include <stdio.h>
  55 #include <stdlib.h>
  56 #include <string.h>
  57 #include <unistd.h>
  58 #include <wchar.h>
  59 #include <wctype.h>
  60
  61 static uintmax_t tlinect, twordct, tcharct, tlongline;
  62 static int doline, doword, dochar, domulti, dolongline;
  63 static volatile sig_atomic_t siginfo;
  64
  65 static void     show_cnt(const char *file, uintmax_t linect, uintmax_t wordct,
  66                     uintmax_t charct, uintmax_t llct);
  67 static int      cnt(const char *);
  68 static void     usage(void);
  69
  70 static void
  71 siginfo_handler(int sig __unused)
  72 {
  73
  74         siginfo = 1;
  75 }
  76
  77 static void
  78 reset_siginfo(void)
  79 {
  80
  81         signal(SIGINFO, SIG_DFL);
  82         siginfo = 0;
  83 }
  84
  85 int
  86 main(int argc, char *argv[])
  87 {
  88         int ch, errors, total;
  89
  90         (void) setlocale(LC_CTYPE, "");
  91
  92         while ((ch = getopt(argc, argv, "clmwL")) != -1)
  93                 switch((char)ch) {
  94                 case 'l':
  95                         doline = 1;
  96                         break;
  97                 case 'w':
  98                         doword = 1;
  99                         break;
 100                 case 'c':
 101                         dochar = 1;
 102                         domulti = 0;
 103                         break;
 104                 case 'L':
 105                         dolongline = 1;
 106                         break;
 107                 case 'm':
 108                         domulti = 1;
 109                         dochar = 0;
 110                         break;
 111                 case '?':
 112                 default:
 113                         usage();
 114                 }
 115         argv += optind;
 116         argc -= optind;
 117
 118         (void)signal(SIGINFO, siginfo_handler);
 119
 120         /* Wc's flags are on by default. */
 121         if (doline + doword + dochar + domulti + dolongline == 0)
 122                 doline = doword = dochar = 1;
 123
 124         errors = 0;
 125         total = 0;
 126         if (!*argv) {
 127                 if (cnt((char *)NULL) != 0)
 128                         ++errors;
 129         } else {
 130                 do {
 131                         if (cnt(*argv) != 0)
 132                                 ++errors;
 133                         ++total;
 134                 } while(*++argv);
 135         }
 136
 137         if (total > 1)
 138                 show_cnt("total", tlinect, twordct, tcharct, tlongline);
 139         exit(errors == 0 ? 0 : 1);
 140 }
 141
 142 static void
 143 show_cnt(const char *file, uintmax_t linect, uintmax_t wordct,
 144     uintmax_t charct, uintmax_t llct)
 145 {
 146         FILE *out;
 147
 148         if (!siginfo)
 149                 out = stdout;
 150         else {
 151                 out = stderr;
 152                 siginfo = 0;
 153         }
 154
 155         if (doline)
 156                 (void)fprintf(out, " %7ju", linect);
 157         if (doword)
 158                 (void)fprintf(out, " %7ju", wordct);
 159         if (dochar || domulti)
 160                 (void)fprintf(out, " %7ju", charct);
 161         if (dolongline)
 162                 (void)fprintf(out, " %7ju", llct);
 163         if (file != NULL)
 164                 (void)fprintf(out, " %s\n", file);
 165         else
 166                 (void)fprintf(out, "\n");
 167 }
 168
 169 static int
 170 cnt(const char *file)
 171 {
 172         struct stat sb;
 173         uintmax_t linect, wordct, charct, llct, tmpll;
 174         int fd, len, warned;
 175         size_t clen;
 176         short gotsp;
 177         u_char *p;
 178         u_char buf[MAXBSIZE];
 179         wchar_t wch;
 180         mbstate_t mbs;
 181
 182         linect = wordct = charct = llct = tmpll = 0;
 183         if (file == NULL)
 184                 fd = STDIN_FILENO;
 185         else {
 186                 if ((fd = open(file, O_RDONLY, 0)) < 0) {
 187                         warn("%s: open", file);
 188                         return (1);
 189                 }
 190                 if (doword || (domulti && MB_CUR_MAX != 1))
 191                         goto word;
 192                 /*
 193                  * Line counting is split out because it's a lot faster to get
 194                  * lines than to get words, since the word count requires some
 195                  * logic.
 196                  */
 197                 if (doline) {
 198                         while ((len = read(fd, buf, MAXBSIZE))) {
 199                                 if (len == -1) {
 200                                         warn("%s: read", file);
 201                                         (void)close(fd);
 202                                         return (1);
 203                                 }
 204                                 if (siginfo) {
 205                                         show_cnt(file, linect, wordct, charct,
 206                                             llct);
 207                                 }
 208                                 charct += len;
 209                                 for (p = buf; len--; ++p)
 210                                         if (*p == '\n') {
 211                                                 if (tmpll > llct)
 212                                                         llct = tmpll;
 213                                                 tmpll = 0;
 214                                                 ++linect;
 215                                         } else
 216                                                 tmpll++;
 217                         }
 218                         reset_siginfo();
 219                         tlinect += linect;
 220                         if (dochar)
 221                                 tcharct += charct;
 222                         if (dolongline) {
 223                                 if (llct > tlongline)
 224                                         tlongline = llct;
 225                         }
 226                         show_cnt(file, linect, wordct, charct, llct);
 227                         (void)close(fd);
 228                         return (0);
 229                 }
 230                 /*
 231                  * If all we need is the number of characters and it's a
 232                  * regular file, just stat the puppy.
 233                  */
 234                 if (dochar || domulti) {
 235                         if (fstat(fd, &sb)) {
 236                                 warn("%s: fstat", file);
 237                                 (void)close(fd);
 238                                 return (1);
 239                         }
 240                         if (S_ISREG(sb.st_mode)) {
 241                                 reset_siginfo();
 242                                 charct = sb.st_size;
 243                                 show_cnt(file, linect, wordct, charct, llct);
 244                                 tcharct += charct;
 245                                 (void)close(fd);
 246                                 return (0);
 247                         }
 248                 }
 249         }
 250
 251         /* Do it the hard way... */
 252 word:   gotsp = 1;
 253         warned = 0;
 254         memset(&mbs, 0, sizeof(mbs));
 255         while ((len = read(fd, buf, MAXBSIZE)) != 0) {
 256                 if (len == -1) {
 257                         warn("%s: read", file != NULL ? file : "stdin");
 258                         (void)close(fd);
 259                         return (1);
 260                 }
 261                 p = buf;
 262                 while (len > 0) {
 263                         if (siginfo)
 264                                 show_cnt(file, linect, wordct, charct, llct);
 265                         if (!domulti || MB_CUR_MAX == 1) {
 266                                 clen = 1;
 267                                 wch = (unsigned char)*p;
 268                         } else if ((clen = mbrtowc(&wch, p, len, &mbs)) ==
 269                             (size_t)-1) {
 270                                 if (!warned) {
 271                                         errno = EILSEQ;
 272                                         warn("%s",
 273                                             file != NULL ? file : "stdin");
 274                                         warned = 1;
 275                                 }
 276                                 memset(&mbs, 0, sizeof(mbs));
 277                                 clen = 1;
 278                                 wch = (unsigned char)*p;
 279                         } else if (clen == (size_t)-2)
 280                                 break;
 281                         else if (clen == 0)
 282                                 clen = 1;
 283                         charct++;
 284                         if (wch != L'\n')
 285                                 tmpll++;
 286                         len -= clen;
 287                         p += clen;
 288                         if (wch == L'\n') {
 289                                 if (tmpll > llct)
 290                                         llct = tmpll;
 291                                 tmpll = 0;
 292                                 ++linect;
 293                         }
 294                         if (iswspace(wch))
 295                                 gotsp = 1;
 296                         else if (gotsp) {
 297                                 gotsp = 0;
 298                                 ++wordct;
 299                         }
 300                 }
 301         }
 302         reset_siginfo();
 303         if (domulti && MB_CUR_MAX > 1)
 304                 if (mbrtowc(NULL, NULL, 0, &mbs) == (size_t)-1 && !warned)
 305                         warn("%s", file != NULL ? file : "stdin");
 306         if (doline)
 307                 tlinect += linect;
 308         if (doword)
 309                 twordct += wordct;
 310         if (dochar || domulti)
 311                 tcharct += charct;
 312         if (dolongline) {
 313                 if (llct > tlongline)
 314                         tlongline = llct;
 315         }
 316         show_cnt(file, linect, wordct, charct, llct);
 317         (void)close(fd);
 318         return (0);
 319 }
 320
 321 static void
 322 usage(void)
 323 {
 324         (void)fprintf(stderr, "usage: wc [-Lclmw] [file ...]\n");
 325         exit(1);
 326 }