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. 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  * @(#) Copyright (c) 1980, 1987, 1991, 1993 The Regents of the University of California.  All rights reserved.
  30  * @(#)wc.c     8.1 (Berkeley) 6/6/93
  31  * $FreeBSD: head/usr.bin/wc/wc.c 281617 2015-04-16 21:44:35Z bdrewery $
  32  */
  33
  34 #include <sys/param.h>
  35 #include <sys/stat.h>
  36
  37 #include <ctype.h>
  38 #include <err.h>
  39 #include <errno.h>
  40 #include <fcntl.h>
  41 #include <locale.h>
  42 #include <stdint.h>
  43 #include <stdio.h>
  44 #include <stdlib.h>
  45 #include <string.h>
  46 #include <unistd.h>
  47 #include <wchar.h>
  48 #include <wctype.h>
  49
  50 static uintmax_t tlinect, twordct, tcharct, tlongline;
  51 static int doline, doword, dochar, domulti, dolongline;
  52 static volatile sig_atomic_t siginfo;
  53
  54 static void     show_cnt(const char *file, uintmax_t linect, uintmax_t wordct,
  55                     uintmax_t charct, uintmax_t llct);
  56 static int      cnt(const char *);
  57 static void     usage(void);
  58
  59 static void
  60 siginfo_handler(int sig __unused)
  61 {
  62
  63         siginfo = 1;
  64 }
  65
  66 static void
  67 reset_siginfo(void)
  68 {
  69
  70         signal(SIGINFO, SIG_DFL);
  71         siginfo = 0;
  72 }
  73
  74 int
  75 main(int argc, char *argv[])
  76 {
  77         int ch, errors, total;
  78
  79         setlocale(LC_CTYPE, "");
  80
  81         while ((ch = getopt(argc, argv, "clmwL")) != -1) {
  82                 switch (ch) {
  83                 case 'l':
  84                         doline = 1;
  85                         break;
  86                 case 'w':
  87                         doword = 1;
  88                         break;
  89                 case 'c':
  90                         dochar = 1;
  91                         domulti = 0;
  92                         break;
  93                 case 'L':
  94                         dolongline = 1;
  95                         break;
  96                 case 'm':
  97                         domulti = 1;
  98                         dochar = 0;
  99                         break;
 100                 case '?':
 101                 default:
 102                         usage();
 103                 }
 104         }
 105         argv += optind;
 106         argc -= optind;
 107
 108         signal(SIGINFO, siginfo_handler);
 109
 110         /* Wc's flags are on by default. */
 111         if (doline + doword + dochar + domulti + dolongline == 0)
 112                 doline = doword = dochar = 1;
 113
 114         errors = 0;
 115         total = 0;
 116         if (*argv == NULL) {
 117                 if (cnt(NULL) != 0)
 118                         ++errors;
 119         } else {
 120                 do {
 121                         if (cnt(*argv) != 0)
 122                                 ++errors;
 123                         ++total;
 124                 } while(*++argv);
 125         }
 126
 127         if (total > 1)
 128                 show_cnt("total", tlinect, twordct, tcharct, tlongline);
 129         exit(errors == 0 ? 0 : 1);
 130 }
 131
 132 static void
 133 show_cnt(const char *file, uintmax_t linect, uintmax_t wordct,
 134     uintmax_t charct, uintmax_t llct)
 135 {
 136         FILE *out;
 137
 138         if (!siginfo)
 139                 out = stdout;
 140         else {
 141                 out = stderr;
 142                 siginfo = 0;
 143         }
 144
 145         if (doline)
 146                 fprintf(out, " %7ju", linect);
 147         if (doword)
 148                 fprintf(out, " %7ju", wordct);
 149         if (dochar || domulti)
 150                 fprintf(out, " %7ju", charct);
 151         if (dolongline)
 152                 fprintf(out, " %7ju", llct);
 153         if (file != NULL)
 154                 fprintf(out, " %s\n", file);
 155         else
 156                 fprintf(out, "\n");
 157 }
 158
 159 static int
 160 cnt(const char *file)
 161 {
 162         struct stat sb;
 163         uintmax_t linect, wordct, charct, llct, tmpll;
 164         int fd, len, warned;
 165         size_t clen;
 166         short gotsp;
 167         u_char *p;
 168         u_char buf[MAXBSIZE];
 169         wchar_t wch;
 170         mbstate_t mbs;
 171
 172         linect = wordct = charct = llct = tmpll = 0;
 173         if (file == NULL) {
 174                 fd = STDIN_FILENO;
 175         } else {
 176                 if ((fd = open(file, O_RDONLY)) < 0) {
 177                         warn("%s: open", file);
 178                         return (1);
 179                 }
 180                 if (doword || (domulti && MB_CUR_MAX != 1))
 181                         goto word;
 182                 /*
 183                  * Line counting is split out because it's a lot faster to get
 184                  * lines than to get words, since the word count requires some
 185                  * logic.
 186                  */
 187                 if (doline) {
 188                         while ((len = read(fd, buf, MAXBSIZE))) {
 189                                 if (len == -1) {
 190                                         warn("%s: read", file);
 191                                         close(fd);
 192                                         return (1);
 193                                 }
 194                                 if (siginfo) {
 195                                         show_cnt(file, linect, wordct, charct,
 196                                             llct);
 197                                 }
 198                                 charct += len;
 199                                 for (p = buf; len--; ++p) {
 200                                         if (*p == '\n') {
 201                                                 if (tmpll > llct)
 202                                                         llct = tmpll;
 203                                                 tmpll = 0;
 204                                                 ++linect;
 205                                         } else
 206                                                 tmpll++;
 207                                 }
 208                         }
 209                         reset_siginfo();
 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                         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                                 close(fd);
 229                                 return (1);
 230                         }
 231                         if (S_ISREG(sb.st_mode)) {
 232                                 reset_siginfo();
 233                                 charct = sb.st_size;
 234                                 show_cnt(file, linect, wordct, charct, llct);
 235                                 tcharct += charct;
 236                                 close(fd);
 237                                 return (0);
 238                         }
 239                 }
 240         }
 241
 242         /* Do it the hard way... */
 243 word:   gotsp = 1;
 244         warned = 0;
 245         memset(&mbs, 0, sizeof(mbs));
 246         while ((len = read(fd, buf, MAXBSIZE)) != 0) {
 247                 if (len == -1) {
 248                         warn("%s: read", file != NULL ? file : "stdin");
 249                         close(fd);
 250                         return (1);
 251                 }
 252                 p = buf;
 253                 while (len > 0) {
 254                         if (siginfo)
 255                                 show_cnt(file, linect, wordct, charct, llct);
 256                         if (!domulti || MB_CUR_MAX == 1) {
 257                                 clen = 1;
 258                                 wch = (unsigned char)*p;
 259                         } else if ((clen = mbrtowc(&wch, p, len, &mbs)) ==
 260                             (size_t)-1) {
 261                                 if (!warned) {
 262                                         errno = EILSEQ;
 263                                         warn("%s",
 264                                             file != NULL ? file : "stdin");
 265                                         warned = 1;
 266                                 }
 267                                 memset(&mbs, 0, sizeof(mbs));
 268                                 clen = 1;
 269                                 wch = (unsigned char)*p;
 270                         } else if (clen == (size_t)-2)
 271                                 break;
 272                         else if (clen == 0)
 273                                 clen = 1;
 274                         charct++;
 275                         if (wch != L'\n')
 276                                 tmpll++;
 277                         len -= clen;
 278                         p += clen;
 279                         if (wch == L'\n') {
 280                                 if (tmpll > llct)
 281                                         llct = tmpll;
 282                                 tmpll = 0;
 283                                 ++linect;
 284                         }
 285                         if (iswspace(wch))
 286                                 gotsp = 1;
 287                         else if (gotsp) {
 288                                 gotsp = 0;
 289                                 ++wordct;
 290                         }
 291                 }
 292         }
 293         reset_siginfo();
 294         if (domulti && MB_CUR_MAX > 1)
 295                 if (mbrtowc(NULL, NULL, 0, &mbs) == (size_t)-1 && !warned)
 296                         warn("%s", file != NULL ? file : "stdin");
 297         if (doline)
 298                 tlinect += linect;
 299         if (doword)
 300                 twordct += wordct;
 301         if (dochar || domulti)
 302                 tcharct += charct;
 303         if (dolongline) {
 304                 if (llct > tlongline)
 305                         tlongline = llct;
 306         }
 307         show_cnt(file, linect, wordct, charct, llct);
 308         close(fd);
 309         return (0);
 310 }
 311
 312 static void
 313 usage(void)
 314 {
 315         fprintf(stderr, "usage: wc [-Lclmw] [file ...]\n");
 316         exit(1);
 317 }