usr.bin/diff/diff.c

   1 /*      $OpenBSD: diff.c,v 1.65 2015/12/29 19:04:46 gsoares Exp $       */
   2
   3 /*
   4  * Copyright (c) 2003 Todd C. Miller <Todd.Miller@courtesan.com>
   5  *
   6  * Permission to use, copy, modify, and distribute this software for any
   7  * purpose with or without fee is hereby granted, provided that the above
   8  * copyright notice and this permission notice appear in all copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  17  *
  18  * Sponsored in part by the Defense Advanced Research Projects
  19  * Agency (DARPA) and Air Force Research Laboratory, Air Force
  20  * Materiel Command, USAF, under agreement number F39502-99-1-0512.
  21  */
  22
  23 #include <sys/cdefs.h>
  24 __FBSDID("$FreeBSD$");
  25
  26 #include <sys/stat.h>
  27
  28 #include <ctype.h>
  29 #include <err.h>
  30 #include <getopt.h>
  31 #include <stdlib.h>
  32 #include <stdio.h>
  33 #include <string.h>
  34 #include <unistd.h>
  35 #include <limits.h>
  36
  37 #include "diff.h"
  38 #include "xmalloc.h"
  39
  40 int      lflag, Nflag, Pflag, rflag, sflag, Tflag, cflag;
  41 int      diff_format, diff_context, status, ignore_file_case;
  42 int      tabsize = 8;
  43 char    *start, *ifdefname, *diffargs, *label[2], *ignore_pats;
  44 char    *group_format = NULL;
  45 struct stat stb1, stb2;
  46 struct excludes *excludes_list;
  47 regex_t  ignore_re;
  48
  49 #define OPTIONS "0123456789aBbC:cdD:efhI:iL:lnNPpqrS:sTtU:uwX:x:"
  50 enum {
  51         OPT_TSIZE = CHAR_MAX + 1,
  52         OPT_STRIPCR,
  53         OPT_IGN_FN_CASE,
  54         OPT_NO_IGN_FN_CASE,
  55         OPT_NORMAL,
  56         OPT_HORIZON_LINES,
  57         OPT_SPEED_LARGE_FILES,
  58         OPT_CHANGED_GROUP_FORMAT,
  59 };
  60
  61 static struct option longopts[] = {
  62         { "text",                       no_argument,            0,      'a' },
  63         { "ignore-space-change",        no_argument,            0,      'b' },
  64         { "context",                    optional_argument,      0,      'C' },
  65         { "ifdef",                      required_argument,      0,      'D' },
  66         { "minimal",                    no_argument,            0,      'd' },
  67         { "ed",                         no_argument,            0,      'e' },
  68         { "forward-ed",                 no_argument,            0,      'f' },
  69         { "ignore-matching-lines",      required_argument,      0,      'I' },
  70         { "ignore-case",                no_argument,            0,      'i' },
  71         { "paginate",                   no_argument,            NULL,   'l' },
  72         { "label",                      required_argument,      0,      'L' },
  73         { "new-file",                   no_argument,            0,      'N' },
  74         { "rcs",                        no_argument,            0,      'n' },
  75         { "unidirectional-new-file",    no_argument,            0,      'P' },
  76         { "show-c-function",            no_argument,            0,      'p' },
  77         { "brief",                      no_argument,            0,      'q' },
  78         { "recursive",                  no_argument,            0,      'r' },
  79         { "report-identical-files",     no_argument,            0,      's' },
  80         { "starting-file",              required_argument,      0,      'S' },
  81         { "expand-tabs",                no_argument,            0,      't' },
  82         { "initial-tab",                no_argument,            0,      'T' },
  83         { "unified",                    optional_argument,      0,      'U' },
  84         { "ignore-all-space",           no_argument,            0,      'w' },
  85         { "exclude",                    required_argument,      0,      'x' },
  86         { "exclude-from",               required_argument,      0,      'X' },
  87         { "ignore-file-name-case",      no_argument,            NULL,   OPT_IGN_FN_CASE },
  88         { "horizon-lines",              required_argument,      NULL,   OPT_HORIZON_LINES },
  89         { "no-ignore-file-name-case",   no_argument,            NULL,   OPT_NO_IGN_FN_CASE },
  90         { "normal",                     no_argument,            NULL,   OPT_NORMAL },
  91         { "speed-large-files",          no_argument,            NULL,   OPT_SPEED_LARGE_FILES},
  92         { "strip-trailing-cr",          no_argument,            NULL,   OPT_STRIPCR },
  93         { "tabsize",                    optional_argument,      NULL,   OPT_TSIZE },
  94         { "changed-group-format",       required_argument,      NULL,   OPT_CHANGED_GROUP_FORMAT},
  95         { NULL,                         0,                      0,      '\0'}
  96 };
  97
  98 void usage(void) __dead2;
  99 void push_excludes(char *);
 100 void push_ignore_pats(char *);
 101 void read_excludes_file(char *file);
 102 void set_argstr(char **, char **);
 103
 104 int
 105 main(int argc, char **argv)
 106 {
 107         const char *errstr = NULL;
 108         char *ep, **oargv;
 109         long  l;
 110         int   ch, dflags, lastch, gotstdin, prevoptind, newarg;
 111
 112         oargv = argv;
 113         gotstdin = 0;
 114         dflags = 0;
 115         lastch = '\0';
 116         prevoptind = 1;
 117         newarg = 1;
 118         diff_context = 3;
 119         diff_format = 0;
 120         while ((ch = getopt_long(argc, argv, OPTIONS, longopts, NULL)) != -1) {
 121                 switch (ch) {
 122                 case '0': case '1': case '2': case '3': case '4':
 123                 case '5': case '6': case '7': case '8': case '9':
 124                         if (newarg)
 125                                 usage();        /* disallow -[0-9]+ */
 126                         else if (lastch == 'c' || lastch == 'u')
 127                                 diff_context = 0;
 128                         else if (!isdigit(lastch) || diff_context > INT_MAX / 10)
 129                                 usage();
 130                         diff_context = (diff_context * 10) + (ch - '0');
 131                         break;
 132                 case 'a':
 133                         dflags |= D_FORCEASCII;
 134                         break;
 135                 case 'b':
 136                         dflags |= D_FOLDBLANKS;
 137                         break;
 138                 case 'C':
 139                 case 'c':
 140                         cflag = 1;
 141                         diff_format = D_CONTEXT;
 142                         if (optarg != NULL) {
 143                                 l = strtol(optarg, &ep, 10);
 144                                 if (*ep != '\0' || l < 0 || l >= INT_MAX)
 145                                         usage();
 146                                 diff_context = (int)l;
 147                         }
 148                         break;
 149                 case 'd':
 150                         dflags |= D_MINIMAL;
 151                         break;
 152                 case 'D':
 153                         diff_format = D_IFDEF;
 154                         ifdefname = optarg;
 155                         break;
 156                 case 'e':
 157                         diff_format = D_EDIT;
 158                         break;
 159                 case 'f':
 160                         diff_format = D_REVERSE;
 161                         break;
 162                 case 'h':
 163                         /* silently ignore for backwards compatibility */
 164                         break;
 165                 case 'I':
 166                         push_ignore_pats(optarg);
 167                         break;
 168                 case 'i':
 169                         dflags |= D_IGNORECASE;
 170                         break;
 171                 case 'L':
 172                         if (label[0] == NULL)
 173                                 label[0] = optarg;
 174                         else if (label[1] == NULL)
 175                                 label[1] = optarg;
 176                         else
 177                                 usage();
 178                         break;
 179                 case 'l':
 180                         lflag = 1;
 181                         break;
 182                 case 'N':
 183                         Nflag = 1;
 184                         break;
 185                 case 'n':
 186                         diff_format = D_NREVERSE;
 187                         break;
 188                 case 'p':
 189                         if (diff_format == 0)
 190                                 diff_format = D_CONTEXT;
 191                         dflags |= D_PROTOTYPE;
 192                         break;
 193                 case 'P':
 194                         Pflag = 1;
 195                         break;
 196                 case 'r':
 197                         rflag = 1;
 198                         break;
 199                 case 'q':
 200                         diff_format = D_BRIEF;
 201                         break;
 202                 case 'S':
 203                         start = optarg;
 204                         break;
 205                 case 's':
 206                         sflag = 1;
 207                         break;
 208                 case 'T':
 209                         Tflag = 1;
 210                         break;
 211                 case 't':
 212                         dflags |= D_EXPANDTABS;
 213                         break;
 214                 case 'U':
 215                 case 'u':
 216                         diff_format = D_UNIFIED;
 217                         if (optarg != NULL) {
 218                                 l = strtol(optarg, &ep, 10);
 219                                 if (*ep != '\0' || l < 0 || l >= INT_MAX)
 220                                         usage();
 221                                 diff_context = (int)l;
 222                         }
 223                         break;
 224                 case 'w':
 225                         dflags |= D_IGNOREBLANKS;
 226                         break;
 227                 case 'X':
 228                         read_excludes_file(optarg);
 229                         break;
 230                 case 'x':
 231                         push_excludes(optarg);
 232                         break;
 233                 case OPT_CHANGED_GROUP_FORMAT:
 234                         diff_format = D_GFORMAT;
 235                         group_format = optarg;
 236                         break;
 237                 case OPT_HORIZON_LINES:
 238                         break; /* XXX TODO for compatibility with GNU diff3 */
 239                 case OPT_IGN_FN_CASE:
 240                         ignore_file_case = 1;
 241                         break;
 242                 case OPT_NO_IGN_FN_CASE:
 243                         ignore_file_case = 0;
 244                         break;
 245                 case OPT_NORMAL:
 246                         diff_format = D_NORMAL;
 247                         break;
 248                 case OPT_TSIZE:
 249                         tabsize = (int) strtonum(optarg, 1, INT_MAX, &errstr);
 250                         if (errstr) {
 251                                 warnx("Invalid argument for tabsize");
 252                                 usage();
 253                         }
 254                         break;
 255                 case OPT_SPEED_LARGE_FILES:
 256                         break; /* ignore but needed for compatibility with GNU diff */
 257                 case OPT_STRIPCR:
 258                         dflags |= D_STRIPCR;
 259                         break;
 260                 default:
 261                         usage();
 262                         break;
 263                 }
 264                 lastch = ch;
 265                 newarg = optind != prevoptind;
 266                 prevoptind = optind;
 267         }
 268         argc -= optind;
 269         argv += optind;
 270
 271 #ifdef __OpenBSD__
 272         if (pledge("stdio rpath tmppath", NULL) == -1)
 273                 err(2, "pledge");
 274 #endif
 275
 276         /*
 277          * Do sanity checks, fill in stb1 and stb2 and call the appropriate
 278          * driver routine.  Both drivers use the contents of stb1 and stb2.
 279          */
 280         if (argc != 2)
 281                 usage();
 282         if (ignore_pats != NULL) {
 283                 char buf[BUFSIZ];
 284                 int error;
 285
 286                 if ((error = regcomp(&ignore_re, ignore_pats,
 287                                      REG_NEWLINE | REG_EXTENDED)) != 0) {
 288                         regerror(error, &ignore_re, buf, sizeof(buf));
 289                         if (*ignore_pats != '\0')
 290                                 errx(2, "%s: %s", ignore_pats, buf);
 291                         else
 292                                 errx(2, "%s", buf);
 293                 }
 294         }
 295         if (strcmp(argv[0], "-") == 0) {
 296                 fstat(STDIN_FILENO, &stb1);
 297                 gotstdin = 1;
 298         } else if (stat(argv[0], &stb1) != 0)
 299                 err(2, "%s", argv[0]);
 300         if (strcmp(argv[1], "-") == 0) {
 301                 fstat(STDIN_FILENO, &stb2);
 302                 gotstdin = 1;
 303         } else if (stat(argv[1], &stb2) != 0)
 304                 err(2, "%s", argv[1]);
 305         if (gotstdin && (S_ISDIR(stb1.st_mode) || S_ISDIR(stb2.st_mode)))
 306                 errx(2, "can't compare - to a directory");
 307         set_argstr(oargv, argv);
 308         if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) {
 309                 if (diff_format == D_IFDEF)
 310                         errx(2, "-D option not supported with directories");
 311                 diffdir(argv[0], argv[1], dflags);
 312         } else {
 313                 if (S_ISDIR(stb1.st_mode)) {
 314                         argv[0] = splice(argv[0], argv[1]);
 315                         if (stat(argv[0], &stb1) < 0)
 316                                 err(2, "%s", argv[0]);
 317                 }
 318                 if (S_ISDIR(stb2.st_mode)) {
 319                         argv[1] = splice(argv[1], argv[0]);
 320                         if (stat(argv[1], &stb2) < 0)
 321                                 err(2, "%s", argv[1]);
 322                 }
 323                 print_status(diffreg(argv[0], argv[1], dflags, 1), argv[0],
 324                     argv[1], "");
 325         }
 326         exit(status);
 327 }
 328
 329 void
 330 set_argstr(char **av, char **ave)
 331 {
 332         size_t argsize;
 333         char **ap;
 334
 335         argsize = 4 + *ave - *av + 1;
 336         diffargs = xmalloc(argsize);
 337         strlcpy(diffargs, "diff", argsize);
 338         for (ap = av + 1; ap < ave; ap++) {
 339                 if (strcmp(*ap, "--") != 0) {
 340                         strlcat(diffargs, " ", argsize);
 341                         strlcat(diffargs, *ap, argsize);
 342                 }
 343         }
 344 }
 345
 346 /*
 347  * Read in an excludes file and push each line.
 348  */
 349 void
 350 read_excludes_file(char *file)
 351 {
 352         FILE *fp;
 353         char *buf, *pattern;
 354         size_t len;
 355
 356         if (strcmp(file, "-") == 0)
 357                 fp = stdin;
 358         else if ((fp = fopen(file, "r")) == NULL)
 359                 err(2, "%s", file);
 360         while ((buf = fgetln(fp, &len)) != NULL) {
 361                 if (buf[len - 1] == '\n')
 362                         len--;
 363                 if ((pattern = strndup(buf, len)) == NULL)
 364                         err(2, "xstrndup");
 365                 push_excludes(pattern);
 366         }
 367         if (strcmp(file, "-") != 0)
 368                 fclose(fp);
 369 }
 370
 371 /*
 372  * Push a pattern onto the excludes list.
 373  */
 374 void
 375 push_excludes(char *pattern)
 376 {
 377         struct excludes *entry;
 378
 379         entry = xmalloc(sizeof(*entry));
 380         entry->pattern = pattern;
 381         entry->next = excludes_list;
 382         excludes_list = entry;
 383 }
 384
 385 void
 386 push_ignore_pats(char *pattern)
 387 {
 388         size_t len;
 389
 390         if (ignore_pats == NULL)
 391                 ignore_pats = xstrdup(pattern);
 392         else {
 393                 /* old + "|" + new + NUL */
 394                 len = strlen(ignore_pats) + strlen(pattern) + 2;
 395                 ignore_pats = xreallocarray(ignore_pats, 1, len);
 396                 strlcat(ignore_pats, "|", len);
 397                 strlcat(ignore_pats, pattern, len);
 398         }
 399 }
 400
 401 void
 402 print_only(const char *path, size_t dirlen, const char *entry)
 403 {
 404         if (dirlen > 1)
 405                 dirlen--;
 406         printf("Only in %.*s: %s\n", (int)dirlen, path, entry);
 407 }
 408
 409 void
 410 print_status(int val, char *path1, char *path2, const char *entry)
 411 {
 412         switch (val) {
 413         case D_BINARY:
 414                 printf("Binary files %s%s and %s%s differ\n",
 415                     path1, entry, path2, entry);
 416                 break;
 417         case D_DIFFER:
 418                 if (diff_format == D_BRIEF)
 419                         printf("Files %s%s and %s%s differ\n",
 420                             path1, entry, path2, entry);
 421                 break;
 422         case D_SAME:
 423                 if (sflag)
 424                         printf("Files %s%s and %s%s are identical\n",
 425                             path1, entry, path2, entry);
 426                 break;
 427         case D_MISMATCH1:
 428                 printf("File %s%s is a directory while file %s%s is a regular file\n",
 429                     path1, entry, path2, entry);
 430                 break;
 431         case D_MISMATCH2:
 432                 printf("File %s%s is a regular file while file %s%s is a directory\n",
 433                     path1, entry, path2, entry);
 434                 break;
 435         case D_SKIPPED1:
 436                 printf("File %s%s is not a regular file or directory and was skipped\n",
 437                     path1, entry);
 438                 break;
 439         case D_SKIPPED2:
 440                 printf("File %s%s is not a regular file or directory and was skipped\n",
 441                     path2, entry);
 442                 break;
 443         }
 444 }
 445
 446 void
 447 usage(void)
 448 {
 449         (void)fprintf(stderr,
 450             "usage: diff [-abdilpTtw] [-c | -e | -f | -n | -q | -u] [--ignore-case]\n"
 451             "            [--no-ignore-case] [--normal] [--strip-trailing-cr] [--tabsize]\n"
 452             "            [-I pattern] [-L label] file1 file2\n"
 453             "       diff [-abdilpTtw] [-I pattern] [-L label] [--ignore-case]\n"
 454             "            [--no-ignore-case] [--normal] [--strip-trailing-cr] [--tabsize]\n"
 455             "            -C number file1 file2\n"
 456             "       diff [-abdiltw] [-I pattern] [--ignore-case] [--no-ignore-case]\n"
 457             "            [--normal] [--strip-trailing-cr] [--tabsize] -D string file1 file2\n"
 458             "       diff [-abdilpTtw] [-I pattern] [-L label] [--ignore-case]\n"
 459             "            [--no-ignore-case] [--normal] [--tabsize] [--strip-trailing-cr]\n"
 460             "            -U number file1 file2\n"
 461             "       diff [-abdilNPprsTtw] [-c | -e | -f | -n | -q | -u] [--ignore-case]\n"
 462             "            [--no-ignore-case] [--normal] [--tabsize] [-I pattern] [-L label]\n"
 463             "            [-S name] [-X file] [-x pattern] dir1 dir2\n");
 464
 465         exit(2);
 466 }