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