usr.bin/diff/diff.c

   1 /*      $OpenBSD: diff.c,v 1.67 2019/06/28 13:35:00 deraadt 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 <errno.h>
  31 #include <getopt.h>
  32 #include <stdlib.h>
  33 #include <stdio.h>
  34 #include <string.h>
  35 #include <unistd.h>
  36 #include <limits.h>
  37
  38 #include "diff.h"
  39 #include "xmalloc.h"
  40
  41 int      lflag, Nflag, Pflag, rflag, sflag, Tflag, cflag, Wflag;
  42 int      diff_format, diff_context, status, ignore_file_case, suppress_common;
  43 int      tabsize = 8, width = 130;
  44 char    *start, *ifdefname, *diffargs, *label[2], *ignore_pats;
  45 char    *group_format = NULL;
  46 struct stat stb1, stb2;
  47 struct excludes *excludes_list;
  48 regex_t  ignore_re;
  49
  50 #define OPTIONS "0123456789aBbC:cdD:efHhI:iL:lnNPpqrS:sTtU:uwW:X:x:y"
  51 enum {
  52         OPT_TSIZE = CHAR_MAX + 1,
  53         OPT_STRIPCR,
  54         OPT_IGN_FN_CASE,
  55         OPT_NO_IGN_FN_CASE,
  56         OPT_NORMAL,
  57         OPT_HORIZON_LINES,
  58         OPT_CHANGED_GROUP_FORMAT,
  59         OPT_SUPPRESS_COMMON,
  60 };
  61
  62 static struct option longopts[] = {
  63         { "text",                       no_argument,            0,      'a' },
  64         { "ignore-space-change",        no_argument,            0,      'b' },
  65         { "context",                    optional_argument,      0,      'C' },
  66         { "ifdef",                      required_argument,      0,      'D' },
  67         { "minimal",                    no_argument,            0,      'd' },
  68         { "ed",                         no_argument,            0,      'e' },
  69         { "forward-ed",                 no_argument,            0,      'f' },
  70         { "speed-large-files",          no_argument,            NULL,   'H' },
  71         { "ignore-blank-lines",         no_argument,            0,      'B' },
  72         { "ignore-matching-lines",      required_argument,      0,      'I' },
  73         { "ignore-case",                no_argument,            0,      'i' },
  74         { "paginate",                   no_argument,            NULL,   'l' },
  75         { "label",                      required_argument,      0,      'L' },
  76         { "new-file",                   no_argument,            0,      'N' },
  77         { "rcs",                        no_argument,            0,      'n' },
  78         { "unidirectional-new-file",    no_argument,            0,      'P' },
  79         { "show-c-function",            no_argument,            0,      'p' },
  80         { "brief",                      no_argument,            0,      'q' },
  81         { "recursive",                  no_argument,            0,      'r' },
  82         { "report-identical-files",     no_argument,            0,      's' },
  83         { "starting-file",              required_argument,      0,      'S' },
  84         { "expand-tabs",                no_argument,            0,      't' },
  85         { "initial-tab",                no_argument,            0,      'T' },
  86         { "unified",                    optional_argument,      0,      'U' },
  87         { "ignore-all-space",           no_argument,            0,      'w' },
  88         { "width",                      required_argument,      0,      'W' },
  89         { "exclude",                    required_argument,      0,      'x' },
  90         { "exclude-from",               required_argument,      0,      'X' },
  91         { "side-by-side",               no_argument,            NULL,   'y' },
  92         { "ignore-file-name-case",      no_argument,            NULL,   OPT_IGN_FN_CASE },
  93         { "horizon-lines",              required_argument,      NULL,   OPT_HORIZON_LINES },
  94         { "no-ignore-file-name-case",   no_argument,            NULL,   OPT_NO_IGN_FN_CASE },
  95         { "normal",                     no_argument,            NULL,   OPT_NORMAL },
  96         { "strip-trailing-cr",          no_argument,            NULL,   OPT_STRIPCR },
  97         { "tabsize",                    required_argument,      NULL,   OPT_TSIZE },
  98         { "changed-group-format",       required_argument,      NULL,   OPT_CHANGED_GROUP_FORMAT},
  99         { "suppress-common-lines",      no_argument,            NULL,   OPT_SUPPRESS_COMMON },
 100         { NULL,                         0,                      0,      '\0'}
 101 };
 102
 103 void usage(void) __dead2;
 104 void conflicting_format(void) __dead2;
 105 void push_excludes(char *);
 106 void push_ignore_pats(char *);
 107 void read_excludes_file(char *file);
 108 void set_argstr(char **, char **);
 109
 110 int
 111 main(int argc, char **argv)
 112 {
 113         const char *errstr = NULL;
 114         char *ep, **oargv;
 115         long  l;
 116         int   ch, dflags, lastch, gotstdin, prevoptind, newarg;
 117
 118         oargv = argv;
 119         gotstdin = 0;
 120         dflags = 0;
 121         lastch = '\0';
 122         prevoptind = 1;
 123         newarg = 1;
 124         diff_context = 3;
 125         diff_format = D_UNSET;
 126 #define FORMAT_MISMATCHED(type) \
 127         (diff_format != D_UNSET && diff_format != (type))
 128         while ((ch = getopt_long(argc, argv, OPTIONS, longopts, NULL)) != -1) {
 129                 switch (ch) {
 130                 case '0': case '1': case '2': case '3': case '4':
 131                 case '5': case '6': case '7': case '8': case '9':
 132                         if (newarg)
 133                                 usage();        /* disallow -[0-9]+ */
 134                         else if (lastch == 'c' || lastch == 'u')
 135                                 diff_context = 0;
 136                         else if (!isdigit(lastch) || diff_context > INT_MAX / 10)
 137                                 usage();
 138                         diff_context = (diff_context * 10) + (ch - '0');
 139                         break;
 140                 case 'a':
 141                         dflags |= D_FORCEASCII;
 142                         break;
 143                 case 'b':
 144                         dflags |= D_FOLDBLANKS;
 145                         break;
 146                 case 'C':
 147                 case 'c':
 148                         if (FORMAT_MISMATCHED(D_CONTEXT))
 149                                 conflicting_format();
 150                         cflag = 1;
 151                         diff_format = D_CONTEXT;
 152                         if (optarg != NULL) {
 153                                 l = strtol(optarg, &ep, 10);
 154                                 if (*ep != '\0' || l < 0 || l >= INT_MAX)
 155                                         usage();
 156                                 diff_context = (int)l;
 157                         }
 158                         break;
 159                 case 'd':
 160                         dflags |= D_MINIMAL;
 161                         break;
 162                 case 'D':
 163                         if (FORMAT_MISMATCHED(D_IFDEF))
 164                                 conflicting_format();
 165                         diff_format = D_IFDEF;
 166                         ifdefname = optarg;
 167                         break;
 168                 case 'e':
 169                         if (FORMAT_MISMATCHED(D_EDIT))
 170                                 conflicting_format();
 171                         diff_format = D_EDIT;
 172                         break;
 173                 case 'f':
 174                         if (FORMAT_MISMATCHED(D_REVERSE))
 175                                 conflicting_format();
 176                         diff_format = D_REVERSE;
 177                         break;
 178                 case 'H':
 179                         /* ignore but needed for compatibility with GNU diff */
 180                         break;
 181                 case 'h':
 182                         /* silently ignore for backwards compatibility */
 183                         break;
 184                 case 'B':
 185                         dflags |= D_SKIPBLANKLINES;
 186                         break;
 187                 case 'I':
 188                         push_ignore_pats(optarg);
 189                         break;
 190                 case 'i':
 191                         dflags |= D_IGNORECASE;
 192                         break;
 193                 case 'L':
 194                         if (label[0] == NULL)
 195                                 label[0] = optarg;
 196                         else if (label[1] == NULL)
 197                                 label[1] = optarg;
 198                         else
 199                                 usage();
 200                         break;
 201                 case 'l':
 202                         lflag = 1;
 203                         break;
 204                 case 'N':
 205                         Nflag = 1;
 206                         break;
 207                 case 'n':
 208                         if (FORMAT_MISMATCHED(D_NREVERSE))
 209                                 conflicting_format();
 210                         diff_format = D_NREVERSE;
 211                         break;
 212                 case 'p':
 213                         dflags |= D_PROTOTYPE;
 214                         break;
 215                 case 'P':
 216                         Pflag = 1;
 217                         break;
 218                 case 'r':
 219                         rflag = 1;
 220                         break;
 221                 case 'q':
 222                         if (FORMAT_MISMATCHED(D_BRIEF))
 223                                 conflicting_format();
 224                         diff_format = D_BRIEF;
 225                         break;
 226                 case 'S':
 227                         start = optarg;
 228                         break;
 229                 case 's':
 230                         sflag = 1;
 231                         break;
 232                 case 'T':
 233                         Tflag = 1;
 234                         break;
 235                 case 't':
 236                         dflags |= D_EXPANDTABS;
 237                         break;
 238                 case 'U':
 239                 case 'u':
 240                         if (FORMAT_MISMATCHED(D_UNIFIED))
 241                                 conflicting_format();
 242                         diff_format = D_UNIFIED;
 243                         if (optarg != NULL) {
 244                                 l = strtol(optarg, &ep, 10);
 245                                 if (*ep != '\0' || l < 0 || l >= INT_MAX)
 246                                         usage();
 247                                 diff_context = (int)l;
 248                         }
 249                         break;
 250                 case 'w':
 251                         dflags |= D_IGNOREBLANKS;
 252                         break;
 253                 case 'W':
 254                         Wflag = 1;
 255                         width = (int) strtonum(optarg, 1, INT_MAX, &errstr);
 256                         if (errstr) {
 257                                 warnx("Invalid argument for width");
 258                                 usage();
 259                         }
 260                         break;
 261                 case 'X':
 262                         read_excludes_file(optarg);
 263                         break;
 264                 case 'x':
 265                         push_excludes(optarg);
 266                         break;
 267                 case 'y':
 268                         if (FORMAT_MISMATCHED(D_SIDEBYSIDE))
 269                                 conflicting_format();
 270                         diff_format = D_SIDEBYSIDE;
 271                         break;
 272                 case OPT_CHANGED_GROUP_FORMAT:
 273                         if (FORMAT_MISMATCHED(D_GFORMAT))
 274                                 conflicting_format();
 275                         diff_format = D_GFORMAT;
 276                         group_format = optarg;
 277                         break;
 278                 case OPT_HORIZON_LINES:
 279                         break; /* XXX TODO for compatibility with GNU diff3 */
 280                 case OPT_IGN_FN_CASE:
 281                         ignore_file_case = 1;
 282                         break;
 283                 case OPT_NO_IGN_FN_CASE:
 284                         ignore_file_case = 0;
 285                         break;
 286                 case OPT_NORMAL:
 287                         if (FORMAT_MISMATCHED(D_NORMAL))
 288                                 conflicting_format();
 289                         diff_format = D_NORMAL;
 290                         break;
 291                 case OPT_TSIZE:
 292                         tabsize = (int) strtonum(optarg, 1, INT_MAX, &errstr);
 293                         if (errstr) {
 294                                 warnx("Invalid argument for tabsize");
 295                                 usage();
 296                         }
 297                         break;
 298                 case OPT_STRIPCR:
 299                         dflags |= D_STRIPCR;
 300                         break;
 301                 case OPT_SUPPRESS_COMMON:
 302                         suppress_common = 1;
 303                         break;
 304                 default:
 305                         usage();
 306                         break;
 307                 }
 308                 lastch = ch;
 309                 newarg = optind != prevoptind;
 310                 prevoptind = optind;
 311         }
 312         if (diff_format == D_UNSET && (dflags & D_PROTOTYPE) != 0)
 313                 diff_format = D_CONTEXT;
 314         if (diff_format == D_UNSET)
 315                 diff_format = D_NORMAL;
 316         argc -= optind;
 317         argv += optind;
 318
 319 #ifdef __OpenBSD__
 320         if (pledge("stdio rpath tmppath", NULL) == -1)
 321                 err(2, "pledge");
 322 #endif
 323
 324         /*
 325          * Do sanity checks, fill in stb1 and stb2 and call the appropriate
 326          * driver routine.  Both drivers use the contents of stb1 and stb2.
 327          */
 328         if (argc != 2)
 329                 usage();
 330         if (ignore_pats != NULL) {
 331                 char buf[BUFSIZ];
 332                 int error;
 333
 334                 if ((error = regcomp(&ignore_re, ignore_pats,
 335                                      REG_NEWLINE | REG_EXTENDED)) != 0) {
 336                         regerror(error, &ignore_re, buf, sizeof(buf));
 337                         if (*ignore_pats != '\0')
 338                                 errx(2, "%s: %s", ignore_pats, buf);
 339                         else
 340                                 errx(2, "%s", buf);
 341                 }
 342         }
 343         if (strcmp(argv[0], "-") == 0) {
 344                 fstat(STDIN_FILENO, &stb1);
 345                 gotstdin = 1;
 346         } else if (stat(argv[0], &stb1) != 0) {
 347                 if (!Nflag || errno != ENOENT)
 348                         err(2, "%s", argv[0]);
 349                 dflags |= D_EMPTY1;
 350                 memset(&stb1, 0, sizeof(struct stat));
 351         }
 352
 353         if (strcmp(argv[1], "-") == 0) {
 354                 fstat(STDIN_FILENO, &stb2);
 355                 gotstdin = 1;
 356         } else if (stat(argv[1], &stb2) != 0) {
 357                 if (!Nflag || errno != ENOENT)
 358                         err(2, "%s", argv[1]);
 359                 dflags |= D_EMPTY2;
 360                 memset(&stb2, 0, sizeof(stb2));
 361                 stb2.st_mode = stb1.st_mode;
 362         }
 363
 364         if (dflags & D_EMPTY1 && dflags & D_EMPTY2){
 365                 warn("%s", argv[0]);
 366                 warn("%s", argv[1]);
 367                 exit(2);
 368         }
 369
 370         if (stb1.st_mode == 0)
 371                 stb1.st_mode = stb2.st_mode;
 372
 373         if (gotstdin && (S_ISDIR(stb1.st_mode) || S_ISDIR(stb2.st_mode)))
 374                 errx(2, "can't compare - to a directory");
 375         set_argstr(oargv, argv);
 376         if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) {
 377                 if (diff_format == D_IFDEF)
 378                         errx(2, "-D option not supported with directories");
 379                 diffdir(argv[0], argv[1], dflags);
 380         } else {
 381                 if (S_ISDIR(stb1.st_mode)) {
 382                         argv[0] = splice(argv[0], argv[1]);
 383                         if (stat(argv[0], &stb1) == -1)
 384                                 err(2, "%s", argv[0]);
 385                 }
 386                 if (S_ISDIR(stb2.st_mode)) {
 387                         argv[1] = splice(argv[1], argv[0]);
 388                         if (stat(argv[1], &stb2) == -1)
 389                                 err(2, "%s", argv[1]);
 390                 }
 391                 print_status(diffreg(argv[0], argv[1], dflags, 1), argv[0],
 392                     argv[1], "");
 393         }
 394         exit(status);
 395 }
 396
 397 void
 398 set_argstr(char **av, char **ave)
 399 {
 400         size_t argsize;
 401         char **ap;
 402
 403         argsize = 4 + *ave - *av + 1;
 404         diffargs = xmalloc(argsize);
 405         strlcpy(diffargs, "diff", argsize);
 406         for (ap = av + 1; ap < ave; ap++) {
 407                 if (strcmp(*ap, "--") != 0) {
 408                         strlcat(diffargs, " ", argsize);
 409                         strlcat(diffargs, *ap, argsize);
 410                 }
 411         }
 412 }
 413
 414 /*
 415  * Read in an excludes file and push each line.
 416  */
 417 void
 418 read_excludes_file(char *file)
 419 {
 420         FILE *fp;
 421         char *buf, *pattern;
 422         size_t len;
 423
 424         if (strcmp(file, "-") == 0)
 425                 fp = stdin;
 426         else if ((fp = fopen(file, "r")) == NULL)
 427                 err(2, "%s", file);
 428         while ((buf = fgetln(fp, &len)) != NULL) {
 429                 if (buf[len - 1] == '\n')
 430                         len--;
 431                 if ((pattern = strndup(buf, len)) == NULL)
 432                         err(2, "xstrndup");
 433                 push_excludes(pattern);
 434         }
 435         if (strcmp(file, "-") != 0)
 436                 fclose(fp);
 437 }
 438
 439 /*
 440  * Push a pattern onto the excludes list.
 441  */
 442 void
 443 push_excludes(char *pattern)
 444 {
 445         struct excludes *entry;
 446
 447         entry = xmalloc(sizeof(*entry));
 448         entry->pattern = pattern;
 449         entry->next = excludes_list;
 450         excludes_list = entry;
 451 }
 452
 453 void
 454 push_ignore_pats(char *pattern)
 455 {
 456         size_t len;
 457
 458         if (ignore_pats == NULL)
 459                 ignore_pats = xstrdup(pattern);
 460         else {
 461                 /* old + "|" + new + NUL */
 462                 len = strlen(ignore_pats) + strlen(pattern) + 2;
 463                 ignore_pats = xreallocarray(ignore_pats, 1, len);
 464                 strlcat(ignore_pats, "|", len);
 465                 strlcat(ignore_pats, pattern, len);
 466         }
 467 }
 468
 469 void
 470 print_only(const char *path, size_t dirlen, const char *entry)
 471 {
 472         if (dirlen > 1)
 473                 dirlen--;
 474         printf("Only in %.*s: %s\n", (int)dirlen, path, entry);
 475 }
 476
 477 void
 478 print_status(int val, char *path1, char *path2, const char *entry)
 479 {
 480         if (label[0] != NULL) path1 = label[0];
 481         if (label[1] != NULL) path2 = label[1];
 482
 483         switch (val) {
 484         case D_BINARY:
 485                 printf("Binary files %s%s and %s%s differ\n",
 486                     path1, entry, path2, entry);
 487                 break;
 488         case D_DIFFER:
 489                 if (diff_format == D_BRIEF)
 490                         printf("Files %s%s and %s%s differ\n",
 491                             path1, entry, path2, entry);
 492                 break;
 493         case D_SAME:
 494                 if (sflag)
 495                         printf("Files %s%s and %s%s are identical\n",
 496                             path1, entry, path2, entry);
 497                 break;
 498         case D_MISMATCH1:
 499                 printf("File %s%s is a directory while file %s%s is a regular file\n",
 500                     path1, entry, path2, entry);
 501                 break;
 502         case D_MISMATCH2:
 503                 printf("File %s%s is a regular file while file %s%s is a directory\n",
 504                     path1, entry, path2, entry);
 505                 break;
 506         case D_SKIPPED1:
 507                 printf("File %s%s is not a regular file or directory and was skipped\n",
 508                     path1, entry);
 509                 break;
 510         case D_SKIPPED2:
 511                 printf("File %s%s is not a regular file or directory and was skipped\n",
 512                     path2, entry);
 513                 break;
 514         }
 515 }
 516
 517 void
 518 usage(void)
 519 {
 520         (void)fprintf(stderr,
 521             "usage: diff [-aBbdilpTtw] [-c | -e | -f | -n | -q | -u] [--ignore-case]\n"
 522             "            [--no-ignore-case] [--normal] [--strip-trailing-cr] [--tabsize]\n"
 523             "            [-I pattern] [-L label] file1 file2\n"
 524             "       diff [-aBbdilpTtw] [-I pattern] [-L label] [--ignore-case]\n"
 525             "            [--no-ignore-case] [--normal] [--strip-trailing-cr] [--tabsize]\n"
 526             "            -C number file1 file2\n"
 527             "       diff [-aBbdiltw] [-I pattern] [--ignore-case] [--no-ignore-case]\n"
 528             "            [--normal] [--strip-trailing-cr] [--tabsize] -D string file1 file2\n"
 529             "       diff [-aBbdilpTtw] [-I pattern] [-L label] [--ignore-case]\n"
 530             "            [--no-ignore-case] [--normal] [--tabsize] [--strip-trailing-cr]\n"
 531             "            -U number file1 file2\n"
 532             "       diff [-aBbdilNPprsTtw] [-c | -e | -f | -n | -q | -u] [--ignore-case]\n"
 533             "            [--no-ignore-case] [--normal] [--tabsize] [-I pattern] [-L label]\n"
 534             "            [-S name] [-X file] [-x pattern] dir1 dir2\n"
 535             "       diff [-aBbditwW] [--expand-tabs] [--ignore-all-blanks]\n"
 536             "            [--ignore-blank-lines] [--ignore-case] [--minimal]\n"
 537             "            [--no-ignore-file-name-case] [--strip-trailing-cr]\n"
 538             "            [--suppress-common-lines] [--tabsize] [--text] [--width]\n"
 539             "            -y | --side-by-side file1 file2\n");
 540
 541         exit(2);
 542 }
 543
 544 void
 545 conflicting_format(void)
 546 {
 547
 548         fprintf(stderr, "error: conflicting output format options.\n");
 549         usage();
 550 }