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 int
  78 main(int argc, char *argv[])
  79 {
  80         int ch, errors, total;
  81
  82         (void) setlocale(LC_CTYPE, "");
  83
  84         while ((ch = getopt(argc, argv, "clmwL")) != -1)
  85                 switch((char)ch) {
  86                 case 'l':
  87                         doline = 1;
  88                         break;
  89                 case 'w':
  90                         doword = 1;
  91                         break;
  92                 case 'c':
  93                         dochar = 1;
  94                         domulti = 0;
  95                         break;
  96                 case 'L':
  97                         dolongline = 1;
  98                         break;
  99                 case 'm':
 100                         domulti = 1;
 101                         dochar = 0;
 102                         break;
 103                 case '?':
 104                 default:
 105                         usage();
 106                 }
 107         argv += optind;
 108         argc -= optind;
 109
 110         (void)signal(SIGINFO, siginfo_handler);
 111
 112         /* Wc's flags are on by default. */
 113         if (doline + doword + dochar + domulti + dolongline == 0)
 114                 doline = doword = dochar = 1;
 115
 116         errors = 0;
 117         total = 0;
 118         if (!*argv) {
 119                 if (cnt((char *)NULL) != 0)
 120                         ++errors;
 121         } else {
 122                 do {
 123                         if (cnt(*argv) != 0)
 124                                 ++errors;
 125                         ++total;
 126                 } while(*++argv);
 127         }
 128
 129         if (total > 1)
 130                 show_cnt("total", tlinect, twordct, tcharct, tlongline);
 131         exit(errors == 0 ? 0 : 1);
 132 }
 133
 134 static void
 135 show_cnt(const char *file, uintmax_t linect, uintmax_t wordct,
 136     uintmax_t charct, uintmax_t llct)
 137 {
 138         FILE *out;
 139
 140         if (!siginfo)
 141                 out = stdout;
 142         else {
 143                 out = stderr;
 144                 siginfo = 0;
 145         }
 146
 147         if (doline)
 148                 (void)fprintf(out, " %7ju", linect);
 149         if (doword)
 150                 (void)fprintf(out, " %7ju", wordct);
 151         if (dochar || domulti)
 152                 (void)fprintf(out, " %7ju", charct);
 153         if (dolongline)
 154                 (void)fprintf(out, " %7ju", llct);
 155         if (file != NULL)
 156                 (void)fprintf(out, " %s\n", file);
 157         else
 158                 (void)fprintf(out, "\n");
 159 }
 160
 161 static int
 162 cnt(const char *file)
 163 {
 164         struct stat sb;
 165         uintmax_t linect, wordct, charct, llct, tmpll;
 166         int fd, len, warned;
 167         size_t clen;
 168         short gotsp;
 169         u_char *p;
 170         u_char buf[MAXBSIZE];
 171         wchar_t wch;
 172         mbstate_t mbs;
 173
 174         linect = wordct = charct = llct = tmpll = 0;
 175         if (file == NULL)
 176                 fd = STDIN_FILENO;
 177         else {
 178                 if ((fd = open(file, O_RDONLY, 0)) < 0) {
 179                         warn("%s: open", file);
 180                         return (1);
 181                 }
 182                 if (doword || (domulti && MB_CUR_MAX != 1))
 183                         goto word;
 184                 /*
 185                  * Line counting is split out because it's a lot faster to get
 186                  * lines than to get words, since the word count requires some
 187                  * logic.
 188                  */
 189                 if (doline) {
 190                         while ((len = read(fd, buf, MAXBSIZE))) {
 191                                 if (len == -1) {
 192                                         warn("%s: read", file);
 193                                         (void)close(fd);
 194                                         return (1);
 195                                 }
 196                                 if (siginfo) {
 197                                         show_cnt(file, linect, wordct, charct,
 198                                             llct);
 199                                 }
 200                                 charct += len;
 201                                 for (p = buf; len--; ++p)
 202                                         if (*p == '\n') {
 203                                                 if (tmpll > llct)
 204                                                         llct = tmpll;
 205                                                 tmpll = 0;
 206                                                 ++linect;
 207                                         } else
 208                                                 tmpll++;
 209                         }
 210                         tlinect += linect;
 211                         if (dochar)
 212                                 tcharct += charct;
 213                         if (dolongline) {
 214                                 if (llct > tlongline)
 215                                         tlongline = llct;
 216                         }
 217                         show_cnt(file, linect, wordct, charct, llct);
 218                         (void)close(fd);
 219                         return (0);
 220                 }
 221                 /*
 222                  * If all we need is the number of characters and it's a
 223                  * regular file, just stat the puppy.
 224                  */
 225                 if (dochar || domulti) {
 226                         if (fstat(fd, &sb)) {
 227                                 warn("%s: fstat", file);
 228                                 (void)close(fd);
 229                                 return (1);
 230                         }
 231                         if (S_ISREG(sb.st_mode)) {
 232                                 charct = sb.st_size;
 233                                 show_cnt(file, linect, wordct, charct, llct);
 234                                 tcharct += charct;
 235                                 (void)close(fd);
 236                                 return (0);
 237                         }
 238                 }
 239         }
 240
 241         /* Do it the hard way... */
 242 word:   gotsp = 1;
 243         warned = 0;
 244         memset(&mbs, 0, sizeof(mbs));
 245         while ((len = read(fd, buf, MAXBSIZE)) != 0) {
 246                 if (len == -1) {
 247                         warn("%s: read", file != NULL ? file : "stdin");
 248                         (void)close(fd);
 249                         return (1);
 250                 }
 251                 p = buf;
 252                 while (len > 0) {
 253                         if (siginfo)
 254                                 show_cnt(file, linect, wordct, charct, llct);
 255                         if (!domulti || MB_CUR_MAX == 1) {
 256                                 clen = 1;
 257                                 wch = (unsigned char)*p;
 258                         } else if ((clen = mbrtowc(&wch, p, len, &mbs)) ==
 259                             (size_t)-1) {
 260                                 if (!warned) {
 261                                         errno = EILSEQ;
 262                                         warn("%s",
 263                                             file != NULL ? file : "stdin");
 264                                         warned = 1;
 265                                 }
 266                                 memset(&mbs, 0, sizeof(mbs));
 267                                 clen = 1;
 268                                 wch = (unsigned char)*p;
 269                         } else if (clen == (size_t)-2)
 270                                 break;
 271                         else if (clen == 0)
 272                                 clen = 1;
 273                         charct++;
 274                         if (wch != L'\n')
 275                                 tmpll++;
 276                         len -= clen;
 277                         p += clen;
 278                         if (wch == L'\n') {
 279                                 if (tmpll > llct)
 280                                         llct = tmpll;
 281                                 tmpll = 0;
 282                                 ++linect;
 283                         }
 284                         if (iswspace(wch))
 285                                 gotsp = 1;
 286                         else if (gotsp) {
 287                                 gotsp = 0;
 288                                 ++wordct;
 289                         }
 290                 }
 291         }
 292         if (domulti && MB_CUR_MAX > 1)
 293                 if (mbrtowc(NULL, NULL, 0, &mbs) == (size_t)-1 && !warned)
 294                         warn("%s", file != NULL ? file : "stdin");
 295         if (doline)
 296                 tlinect += linect;
 297         if (doword)
 298                 twordct += wordct;
 299         if (dochar || domulti)
 300                 tcharct += charct;
 301         if (dolongline) {
 302                 if (llct > tlongline)
 303                         tlongline = llct;
 304         }
 305         show_cnt(file, linect, wordct, charct, llct);
 306         (void)close(fd);
 307         return (0);
 308 }
 309
 310 static void
 311 usage(void)
 312 {
 313         (void)fprintf(stderr, "usage: wc [-Lclmw] [file ...]\n");
 314         exit(1);
 315 }