builtin-grep.c

   1 /*
   2  * Builtin "git grep"
   3  *
   4  * Copyright (c) 2006 Junio C Hamano
   5  */
   6 #include "cache.h"
   7 #include "blob.h"
   8 #include "tree.h"
   9 #include "commit.h"
  10 #include "tag.h"
  11 #include "tree-walk.h"
  12 #include "builtin.h"
  13 #include <regex.h>
  14 #include "grep.h"
  15 #include <fnmatch.h>
  16 #include <sys/wait.h>
  17
  18 /*
  19  * git grep pathspecs are somewhat different from diff-tree pathspecs;
  20  * pathname wildcards are allowed.
  21  */
  22 static int pathspec_matches(const char **paths, const char *name)
  23 {
  24         int namelen, i;
  25         if (!paths || !*paths)
  26                 return 1;
  27         namelen = strlen(name);
  28         for (i = 0; paths[i]; i++) {
  29                 const char *match = paths[i];
  30                 int matchlen = strlen(match);
  31                 const char *cp, *meta;
  32
  33                 if (!matchlen ||
  34                     ((matchlen <= namelen) &&
  35                      !strncmp(name, match, matchlen) &&
  36                      (match[matchlen-1] == '/' ||
  37                       name[matchlen] == '\0' || name[matchlen] == '/')))
  38                         return 1;
  39                 if (!fnmatch(match, name, 0))
  40                         return 1;
  41                 if (name[namelen-1] != '/')
  42                         continue;
  43
  44                 /* We are being asked if the directory ("name") is worth
  45                  * descending into.
  46                  *
  47                  * Find the longest leading directory name that does
  48                  * not have metacharacter in the pathspec; the name
  49                  * we are looking at must overlap with that directory.
  50                  */
  51                 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
  52                         char ch = *cp;
  53                         if (ch == '*' || ch == '[' || ch == '?') {
  54                                 meta = cp;
  55                                 break;
  56                         }
  57                 }
  58                 if (!meta)
  59                         meta = cp; /* fully literal */
  60
  61                 if (namelen <= meta - match) {
  62                         /* Looking at "Documentation/" and
  63                          * the pattern says "Documentation/howto/", or
  64                          * "Documentation/diff*.txt".  The name we
  65                          * have should match prefix.
  66                          */
  67                         if (!memcmp(match, name, namelen))
  68                                 return 1;
  69                         continue;
  70                 }
  71
  72                 if (meta - match < namelen) {
  73                         /* Looking at "Documentation/howto/" and
  74                          * the pattern says "Documentation/h*";
  75                          * match up to "Do.../h"; this avoids descending
  76                          * into "Documentation/technical/".
  77                          */
  78                         if (!memcmp(match, name, meta - match))
  79                                 return 1;
  80                         continue;
  81                 }
  82         }
  83         return 0;
  84 }
  85
  86 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
  87 {
  88         unsigned long size;
  89         char *data;
  90         char type[20];
  91         char *to_free = NULL;
  92         int hit;
  93
  94         data = read_sha1_file(sha1, type, &size);
  95         if (!data) {
  96                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
  97                 return 0;
  98         }
  99         if (opt->relative && opt->prefix_length) {
 100                 static char name_buf[PATH_MAX];
 101                 char *cp;
 102                 int name_len = strlen(name) - opt->prefix_length + 1;
 103
 104                 if (!tree_name_len)
 105                         name += opt->prefix_length;
 106                 else {
 107                         if (ARRAY_SIZE(name_buf) <= name_len)
 108                                 cp = to_free = xmalloc(name_len);
 109                         else
 110                                 cp = name_buf;
 111                         memcpy(cp, name, tree_name_len);
 112                         strcpy(cp + tree_name_len,
 113                                name + tree_name_len + opt->prefix_length);
 114                         name = cp;
 115                 }
 116         }
 117         hit = grep_buffer(opt, name, data, size);
 118         free(data);
 119         free(to_free);
 120         return hit;
 121 }
 122
 123 static int grep_file(struct grep_opt *opt, const char *filename)
 124 {
 125         struct stat st;
 126         int i;
 127         char *data;
 128         if (lstat(filename, &st) < 0) {
 129         err_ret:
 130                 if (errno != ENOENT)
 131                         error("'%s': %s", filename, strerror(errno));
 132                 return 0;
 133         }
 134         if (!st.st_size)
 135                 return 0; /* empty file -- no grep hit */
 136         if (!S_ISREG(st.st_mode))
 137                 return 0;
 138         i = open(filename, O_RDONLY);
 139         if (i < 0)
 140                 goto err_ret;
 141         data = xmalloc(st.st_size + 1);
 142         if (st.st_size != xread(i, data, st.st_size)) {
 143                 error("'%s': short read %s", filename, strerror(errno));
 144                 close(i);
 145                 free(data);
 146                 return 0;
 147         }
 148         close(i);
 149         if (opt->relative && opt->prefix_length)
 150                 filename += opt->prefix_length;
 151         i = grep_buffer(opt, filename, data, st.st_size);
 152         free(data);
 153         return i;
 154 }
 155
 156 static int exec_grep(int argc, const char **argv)
 157 {
 158         pid_t pid;
 159         int status;
 160
 161         argv[argc] = NULL;
 162         pid = fork();
 163         if (pid < 0)
 164                 return pid;
 165         if (!pid) {
 166                 execvp("grep", (char **) argv);
 167                 exit(255);
 168         }
 169         while (waitpid(pid, &status, 0) < 0) {
 170                 if (errno == EINTR)
 171                         continue;
 172                 return -1;
 173         }
 174         if (WIFEXITED(status)) {
 175                 if (!WEXITSTATUS(status))
 176                         return 1;
 177                 return 0;
 178         }
 179         return -1;
 180 }
 181
 182 #define MAXARGS 1000
 183 #define ARGBUF 4096
 184 #define push_arg(a) do { \
 185         if (nr < MAXARGS) argv[nr++] = (a); \
 186         else die("maximum number of args exceeded"); \
 187         } while (0)
 188
 189 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
 190 {
 191         int i, nr, argc, hit, len, status;
 192         const char *argv[MAXARGS+1];
 193         char randarg[ARGBUF];
 194         char *argptr = randarg;
 195         struct grep_pat *p;
 196
 197         if (opt->extended || (opt->relative && opt->prefix_length))
 198                 return -1;
 199         len = nr = 0;
 200         push_arg("grep");
 201         if (opt->fixed)
 202                 push_arg("-F");
 203         if (opt->linenum)
 204                 push_arg("-n");
 205         if (!opt->pathname)
 206                 push_arg("-h");
 207         if (opt->regflags & REG_EXTENDED)
 208                 push_arg("-E");
 209         if (opt->regflags & REG_ICASE)
 210                 push_arg("-i");
 211         if (opt->word_regexp)
 212                 push_arg("-w");
 213         if (opt->name_only)
 214                 push_arg("-l");
 215         if (opt->unmatch_name_only)
 216                 push_arg("-L");
 217         if (opt->count)
 218                 push_arg("-c");
 219         if (opt->post_context || opt->pre_context) {
 220                 if (opt->post_context != opt->pre_context) {
 221                         if (opt->pre_context) {
 222                                 push_arg("-B");
 223                                 len += snprintf(argptr, sizeof(randarg)-len,
 224                                                 "%u", opt->pre_context);
 225                                 if (sizeof(randarg) <= len)
 226                                         die("maximum length of args exceeded");
 227                                 push_arg(argptr);
 228                                 argptr += len;
 229                         }
 230                         if (opt->post_context) {
 231                                 push_arg("-A");
 232                                 len += snprintf(argptr, sizeof(randarg)-len,
 233                                                 "%u", opt->post_context);
 234                                 if (sizeof(randarg) <= len)
 235                                         die("maximum length of args exceeded");
 236                                 push_arg(argptr);
 237                                 argptr += len;
 238                         }
 239                 }
 240                 else {
 241                         push_arg("-C");
 242                         len += snprintf(argptr, sizeof(randarg)-len,
 243                                         "%u", opt->post_context);
 244                         if (sizeof(randarg) <= len)
 245                                 die("maximum length of args exceeded");
 246                         push_arg(argptr);
 247                         argptr += len;
 248                 }
 249         }
 250         for (p = opt->pattern_list; p; p = p->next) {
 251                 push_arg("-e");
 252                 push_arg(p->pattern);
 253         }
 254
 255         /*
 256          * To make sure we get the header printed out when we want it,
 257          * add /dev/null to the paths to grep.  This is unnecessary
 258          * (and wrong) with "-l" or "-L", which always print out the
 259          * name anyway.
 260          *
 261          * GNU grep has "-H", but this is portable.
 262          */
 263         if (!opt->name_only && !opt->unmatch_name_only)
 264                 push_arg("/dev/null");
 265
 266         hit = 0;
 267         argc = nr;
 268         for (i = 0; i < active_nr; i++) {
 269                 struct cache_entry *ce = active_cache[i];
 270                 char *name;
 271                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 272                         continue;
 273                 if (!pathspec_matches(paths, ce->name))
 274                         continue;
 275                 name = ce->name;
 276                 if (name[0] == '-') {
 277                         int len = ce_namelen(ce);
 278                         name = xmalloc(len + 3);
 279                         memcpy(name, "./", 2);
 280                         memcpy(name + 2, ce->name, len + 1);
 281                 }
 282                 argv[argc++] = name;
 283                 if (argc < MAXARGS)
 284                         continue;
 285                 status = exec_grep(argc, argv);
 286                 if (0 < status)
 287                         hit = 1;
 288                 argc = nr;
 289         }
 290         if (argc > nr) {
 291                 status = exec_grep(argc, argv);
 292                 if (0 < status)
 293                         hit = 1;
 294         }
 295         return hit;
 296 }
 297
 298 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 299 {
 300         int hit = 0;
 301         int nr;
 302         read_cache();
 303
 304 #ifdef __unix__
 305         /*
 306          * Use the external "grep" command for the case where
 307          * we grep through the checked-out files. It tends to
 308          * be a lot more optimized
 309          */
 310         if (!cached) {
 311                 hit = external_grep(opt, paths, cached);
 312                 if (hit >= 0)
 313                         return hit;
 314         }
 315 #endif
 316
 317         for (nr = 0; nr < active_nr; nr++) {
 318                 struct cache_entry *ce = active_cache[nr];
 319                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 320                         continue;
 321                 if (!pathspec_matches(paths, ce->name))
 322                         continue;
 323                 if (cached)
 324                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 325                 else
 326                         hit |= grep_file(opt, ce->name);
 327         }
 328         return hit;
 329 }
 330
 331 static int grep_tree(struct grep_opt *opt, const char **paths,
 332                      struct tree_desc *tree,
 333                      const char *tree_name, const char *base)
 334 {
 335         int len;
 336         int hit = 0;
 337         struct name_entry entry;
 338         char *down;
 339         int tn_len = strlen(tree_name);
 340         char *path_buf = xmalloc(PATH_MAX + tn_len + 100);
 341
 342         if (tn_len) {
 343                 tn_len = sprintf(path_buf, "%s:", tree_name);
 344                 down = path_buf + tn_len;
 345                 strcat(down, base);
 346         }
 347         else {
 348                 down = path_buf;
 349                 strcpy(down, base);
 350         }
 351         len = strlen(path_buf);
 352
 353         while (tree_entry(tree, &entry)) {
 354                 strcpy(path_buf + len, entry.path);
 355
 356                 if (S_ISDIR(entry.mode))
 357                         /* Match "abc/" against pathspec to
 358                          * decide if we want to descend into "abc"
 359                          * directory.
 360                          */
 361                         strcpy(path_buf + len + entry.pathlen, "/");
 362
 363                 if (!pathspec_matches(paths, down))
 364                         ;
 365                 else if (S_ISREG(entry.mode))
 366                         hit |= grep_sha1(opt, entry.sha1, path_buf, tn_len);
 367                 else if (S_ISDIR(entry.mode)) {
 368                         char type[20];
 369                         struct tree_desc sub;
 370                         void *data;
 371                         data = read_sha1_file(entry.sha1, type, &sub.size);
 372                         if (!data)
 373                                 die("unable to read tree (%s)",
 374                                     sha1_to_hex(entry.sha1));
 375                         sub.buf = data;
 376                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 377                         free(data);
 378                 }
 379         }
 380         return hit;
 381 }
 382
 383 static int grep_object(struct grep_opt *opt, const char **paths,
 384                        struct object *obj, const char *name)
 385 {
 386         if (obj->type == OBJ_BLOB)
 387                 return grep_sha1(opt, obj->sha1, name, 0);
 388         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 389                 struct tree_desc tree;
 390                 void *data;
 391                 int hit;
 392                 data = read_object_with_reference(obj->sha1, tree_type,
 393                                                   &tree.size, NULL);
 394                 if (!data)
 395                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 396                 tree.buf = data;
 397                 hit = grep_tree(opt, paths, &tree, name, "");
 398                 free(data);
 399                 return hit;
 400         }
 401         die("unable to grep from object of type %s", typename(obj->type));
 402 }
 403
 404 static const char builtin_grep_usage[] =
 405 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 406
 407 static const char emsg_invalid_context_len[] =
 408 "%s: invalid context length argument";
 409 static const char emsg_missing_context_len[] =
 410 "missing context length argument";
 411 static const char emsg_missing_argument[] =
 412 "option requires an argument -%s";
 413
 414 int cmd_grep(int argc, const char **argv, const char *prefix)
 415 {
 416         int hit = 0;
 417         int cached = 0;
 418         int seen_dashdash = 0;
 419         struct grep_opt opt;
 420         struct object_array list = { 0, 0, NULL };
 421         const char **paths = NULL;
 422         int i;
 423
 424         memset(&opt, 0, sizeof(opt));
 425         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 426         opt.relative = 1;
 427         opt.pathname = 1;
 428         opt.pattern_tail = &opt.pattern_list;
 429         opt.regflags = REG_NEWLINE;
 430
 431         /*
 432          * If there is no -- then the paths must exist in the working
 433          * tree.  If there is no explicit pattern specified with -e or
 434          * -f, we take the first unrecognized non option to be the
 435          * pattern, but then what follows it must be zero or more
 436          * valid refs up to the -- (if exists), and then existing
 437          * paths.  If there is an explicit pattern, then the first
 438          * unrecognized non option is the beginning of the refs list
 439          * that continues up to the -- (if exists), and then paths.
 440          */
 441
 442         while (1 < argc) {
 443                 const char *arg = argv[1];
 444                 argc--; argv++;
 445                 if (!strcmp("--cached", arg)) {
 446                         cached = 1;
 447                         continue;
 448                 }
 449                 if (!strcmp("-a", arg) ||
 450                     !strcmp("--text", arg)) {
 451                         opt.binary = GREP_BINARY_TEXT;
 452                         continue;
 453                 }
 454                 if (!strcmp("-i", arg) ||
 455                     !strcmp("--ignore-case", arg)) {
 456                         opt.regflags |= REG_ICASE;
 457                         continue;
 458                 }
 459                 if (!strcmp("-I", arg)) {
 460                         opt.binary = GREP_BINARY_NOMATCH;
 461                         continue;
 462                 }
 463                 if (!strcmp("-v", arg) ||
 464                     !strcmp("--invert-match", arg)) {
 465                         opt.invert = 1;
 466                         continue;
 467                 }
 468                 if (!strcmp("-E", arg) ||
 469                     !strcmp("--extended-regexp", arg)) {
 470                         opt.regflags |= REG_EXTENDED;
 471                         continue;
 472                 }
 473                 if (!strcmp("-F", arg) ||
 474                     !strcmp("--fixed-strings", arg)) {
 475                         opt.fixed = 1;
 476                         continue;
 477                 }
 478                 if (!strcmp("-G", arg) ||
 479                     !strcmp("--basic-regexp", arg)) {
 480                         opt.regflags &= ~REG_EXTENDED;
 481                         continue;
 482                 }
 483                 if (!strcmp("-n", arg)) {
 484                         opt.linenum = 1;
 485                         continue;
 486                 }
 487                 if (!strcmp("-h", arg)) {
 488                         opt.pathname = 0;
 489                         continue;
 490                 }
 491                 if (!strcmp("-H", arg)) {
 492                         opt.pathname = 1;
 493                         continue;
 494                 }
 495                 if (!strcmp("-l", arg) ||
 496                     !strcmp("--files-with-matches", arg)) {
 497                         opt.name_only = 1;
 498                         continue;
 499                 }
 500                 if (!strcmp("-L", arg) ||
 501                     !strcmp("--files-without-match", arg)) {
 502                         opt.unmatch_name_only = 1;
 503                         continue;
 504                 }
 505                 if (!strcmp("-c", arg) ||
 506                     !strcmp("--count", arg)) {
 507                         opt.count = 1;
 508                         continue;
 509                 }
 510                 if (!strcmp("-w", arg) ||
 511                     !strcmp("--word-regexp", arg)) {
 512                         opt.word_regexp = 1;
 513                         continue;
 514                 }
 515                 if (!strncmp("-A", arg, 2) ||
 516                     !strncmp("-B", arg, 2) ||
 517                     !strncmp("-C", arg, 2) ||
 518                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
 519                         unsigned num;
 520                         const char *scan;
 521                         switch (arg[1]) {
 522                         case 'A': case 'B': case 'C':
 523                                 if (!arg[2]) {
 524                                         if (argc <= 1)
 525                                                 die(emsg_missing_context_len);
 526                                         scan = *++argv;
 527                                         argc--;
 528                                 }
 529                                 else
 530                                         scan = arg + 2;
 531                                 break;
 532                         default:
 533                                 scan = arg + 1;
 534                                 break;
 535                         }
 536                         if (sscanf(scan, "%u", &num) != 1)
 537                                 die(emsg_invalid_context_len, scan);
 538                         switch (arg[1]) {
 539                         case 'A':
 540                                 opt.post_context = num;
 541                                 break;
 542                         default:
 543                         case 'C':
 544                                 opt.post_context = num;
 545                         case 'B':
 546                                 opt.pre_context = num;
 547                                 break;
 548                         }
 549                         continue;
 550                 }
 551                 if (!strcmp("-f", arg)) {
 552                         FILE *patterns;
 553                         int lno = 0;
 554                         char buf[1024];
 555                         if (argc <= 1)
 556                                 die(emsg_missing_argument, arg);
 557                         patterns = fopen(argv[1], "r");
 558                         if (!patterns)
 559                                 die("'%s': %s", argv[1], strerror(errno));
 560                         while (fgets(buf, sizeof(buf), patterns)) {
 561                                 int len = strlen(buf);
 562                                 if (buf[len-1] == '\n')
 563                                         buf[len-1] = 0;
 564                                 /* ignore empty line like grep does */
 565                                 if (!buf[0])
 566                                         continue;
 567                                 append_grep_pattern(&opt, xstrdup(buf),
 568                                                     argv[1], ++lno,
 569                                                     GREP_PATTERN);
 570                         }
 571                         fclose(patterns);
 572                         argv++;
 573                         argc--;
 574                         continue;
 575                 }
 576                 if (!strcmp("--not", arg)) {
 577                         append_grep_pattern(&opt, arg, "command line", 0,
 578                                             GREP_NOT);
 579                         continue;
 580                 }
 581                 if (!strcmp("--and", arg)) {
 582                         append_grep_pattern(&opt, arg, "command line", 0,
 583                                             GREP_AND);
 584                         continue;
 585                 }
 586                 if (!strcmp("--or", arg))
 587                         continue; /* no-op */
 588                 if (!strcmp("(", arg)) {
 589                         append_grep_pattern(&opt, arg, "command line", 0,
 590                                             GREP_OPEN_PAREN);
 591                         continue;
 592                 }
 593                 if (!strcmp(")", arg)) {
 594                         append_grep_pattern(&opt, arg, "command line", 0,
 595                                             GREP_CLOSE_PAREN);
 596                         continue;
 597                 }
 598                 if (!strcmp("-e", arg)) {
 599                         if (1 < argc) {
 600                                 append_grep_pattern(&opt, argv[1],
 601                                                     "-e option", 0,
 602                                                     GREP_PATTERN);
 603                                 argv++;
 604                                 argc--;
 605                                 continue;
 606                         }
 607                         die(emsg_missing_argument, arg);
 608                 }
 609                 if (!strcmp("--full-name", arg)) {
 610                         opt.relative = 0;
 611                         continue;
 612                 }
 613                 if (!strcmp("--", arg)) {
 614                         /* later processing wants to have this at argv[1] */
 615                         argv--;
 616                         argc++;
 617                         break;
 618                 }
 619                 if (*arg == '-')
 620                         usage(builtin_grep_usage);
 621
 622                 /* First unrecognized non-option token */
 623                 if (!opt.pattern_list) {
 624                         append_grep_pattern(&opt, arg, "command line", 0,
 625                                             GREP_PATTERN);
 626                         break;
 627                 }
 628                 else {
 629                         /* We are looking at the first path or rev;
 630                          * it is found at argv[1] after leaving the
 631                          * loop.
 632                          */
 633                         argc++; argv--;
 634                         break;
 635                 }
 636         }
 637
 638         if (!opt.pattern_list)
 639                 die("no pattern given.");
 640         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
 641                 die("cannot mix --fixed-strings and regexp");
 642         compile_grep_patterns(&opt);
 643
 644         /* Check revs and then paths */
 645         for (i = 1; i < argc; i++) {
 646                 const char *arg = argv[i];
 647                 unsigned char sha1[20];
 648                 /* Is it a rev? */
 649                 if (!get_sha1(arg, sha1)) {
 650                         struct object *object = parse_object(sha1);
 651                         if (!object)
 652                                 die("bad object %s", arg);
 653                         add_object_array(object, arg, &list);
 654                         continue;
 655                 }
 656                 if (!strcmp(arg, "--")) {
 657                         i++;
 658                         seen_dashdash = 1;
 659                 }
 660                 break;
 661         }
 662
 663         /* The rest are paths */
 664         if (!seen_dashdash) {
 665                 int j;
 666                 for (j = i; j < argc; j++)
 667                         verify_filename(prefix, argv[j]);
 668         }
 669
 670         if (i < argc) {
 671                 paths = get_pathspec(prefix, argv + i);
 672                 if (opt.prefix_length && opt.relative) {
 673                         /* Make sure we do not get outside of paths */
 674                         for (i = 0; paths[i]; i++)
 675                                 if (strncmp(prefix, paths[i], opt.prefix_length))
 676                                         die("git-grep: cannot generate relative filenames containing '..'");
 677                 }
 678         }
 679         else if (prefix) {
 680                 paths = xcalloc(2, sizeof(const char *));
 681                 paths[0] = prefix;
 682                 paths[1] = NULL;
 683         }
 684
 685         if (!list.nr)
 686                 return !grep_cache(&opt, paths, cached);
 687
 688         if (cached)
 689                 die("both --cached and trees are given.");
 690
 691         for (i = 0; i < list.nr; i++) {
 692                 struct object *real_obj;
 693                 real_obj = deref_tag(list.objects[i].item, NULL, 0);
 694                 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
 695                         hit = 1;
 696         }
 697         return !hit;
 698 }