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