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         free_grep_patterns(opt);
 329         return hit;
 330 }
 331
 332 static int grep_tree(struct grep_opt *opt, const char **paths,
 333                      struct tree_desc *tree,
 334                      const char *tree_name, const char *base)
 335 {
 336         int len;
 337         int hit = 0;
 338         struct name_entry entry;
 339         char *down;
 340         int tn_len = strlen(tree_name);
 341         char *path_buf = xmalloc(PATH_MAX + tn_len + 100);
 342
 343         if (tn_len) {
 344                 tn_len = sprintf(path_buf, "%s:", tree_name);
 345                 down = path_buf + tn_len;
 346                 strcat(down, base);
 347         }
 348         else {
 349                 down = path_buf;
 350                 strcpy(down, base);
 351         }
 352         len = strlen(path_buf);
 353
 354         while (tree_entry(tree, &entry)) {
 355                 strcpy(path_buf + len, entry.path);
 356
 357                 if (S_ISDIR(entry.mode))
 358                         /* Match "abc/" against pathspec to
 359                          * decide if we want to descend into "abc"
 360                          * directory.
 361                          */
 362                         strcpy(path_buf + len + entry.pathlen, "/");
 363
 364                 if (!pathspec_matches(paths, down))
 365                         ;
 366                 else if (S_ISREG(entry.mode))
 367                         hit |= grep_sha1(opt, entry.sha1, path_buf, tn_len);
 368                 else if (S_ISDIR(entry.mode)) {
 369                         char type[20];
 370                         struct tree_desc sub;
 371                         void *data;
 372                         data = read_sha1_file(entry.sha1, type, &sub.size);
 373                         if (!data)
 374                                 die("unable to read tree (%s)",
 375                                     sha1_to_hex(entry.sha1));
 376                         sub.buf = data;
 377                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 378                         free(data);
 379                 }
 380         }
 381         return hit;
 382 }
 383
 384 static int grep_object(struct grep_opt *opt, const char **paths,
 385                        struct object *obj, const char *name)
 386 {
 387         if (obj->type == OBJ_BLOB)
 388                 return grep_sha1(opt, obj->sha1, name, 0);
 389         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 390                 struct tree_desc tree;
 391                 void *data;
 392                 int hit;
 393                 data = read_object_with_reference(obj->sha1, tree_type,
 394                                                   &tree.size, NULL);
 395                 if (!data)
 396                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 397                 tree.buf = data;
 398                 hit = grep_tree(opt, paths, &tree, name, "");
 399                 free(data);
 400                 return hit;
 401         }
 402         die("unable to grep from object of type %s", typename(obj->type));
 403 }
 404
 405 static const char builtin_grep_usage[] =
 406 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 407
 408 static const char emsg_invalid_context_len[] =
 409 "%s: invalid context length argument";
 410 static const char emsg_missing_context_len[] =
 411 "missing context length argument";
 412 static const char emsg_missing_argument[] =
 413 "option requires an argument -%s";
 414
 415 int cmd_grep(int argc, const char **argv, const char *prefix)
 416 {
 417         int hit = 0;
 418         int cached = 0;
 419         int seen_dashdash = 0;
 420         struct grep_opt opt;
 421         struct object_array list = { 0, 0, NULL };
 422         const char **paths = NULL;
 423         int i;
 424
 425         memset(&opt, 0, sizeof(opt));
 426         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 427         opt.relative = 1;
 428         opt.pathname = 1;
 429         opt.pattern_tail = &opt.pattern_list;
 430         opt.regflags = REG_NEWLINE;
 431
 432         /*
 433          * If there is no -- then the paths must exist in the working
 434          * tree.  If there is no explicit pattern specified with -e or
 435          * -f, we take the first unrecognized non option to be the
 436          * pattern, but then what follows it must be zero or more
 437          * valid refs up to the -- (if exists), and then existing
 438          * paths.  If there is an explicit pattern, then the first
 439          * unrecognized non option is the beginning of the refs list
 440          * that continues up to the -- (if exists), and then paths.
 441          */
 442
 443         while (1 < argc) {
 444                 const char *arg = argv[1];
 445                 argc--; argv++;
 446                 if (!strcmp("--cached", arg)) {
 447                         cached = 1;
 448                         continue;
 449                 }
 450                 if (!strcmp("-a", arg) ||
 451                     !strcmp("--text", arg)) {
 452                         opt.binary = GREP_BINARY_TEXT;
 453                         continue;
 454                 }
 455                 if (!strcmp("-i", arg) ||
 456                     !strcmp("--ignore-case", arg)) {
 457                         opt.regflags |= REG_ICASE;
 458                         continue;
 459                 }
 460                 if (!strcmp("-I", arg)) {
 461                         opt.binary = GREP_BINARY_NOMATCH;
 462                         continue;
 463                 }
 464                 if (!strcmp("-v", arg) ||
 465                     !strcmp("--invert-match", arg)) {
 466                         opt.invert = 1;
 467                         continue;
 468                 }
 469                 if (!strcmp("-E", arg) ||
 470                     !strcmp("--extended-regexp", arg)) {
 471                         opt.regflags |= REG_EXTENDED;
 472                         continue;
 473                 }
 474                 if (!strcmp("-F", arg) ||
 475                     !strcmp("--fixed-strings", arg)) {
 476                         opt.fixed = 1;
 477                         continue;
 478                 }
 479                 if (!strcmp("-G", arg) ||
 480                     !strcmp("--basic-regexp", arg)) {
 481                         opt.regflags &= ~REG_EXTENDED;
 482                         continue;
 483                 }
 484                 if (!strcmp("-n", arg)) {
 485                         opt.linenum = 1;
 486                         continue;
 487                 }
 488                 if (!strcmp("-h", arg)) {
 489                         opt.pathname = 0;
 490                         continue;
 491                 }
 492                 if (!strcmp("-H", arg)) {
 493                         opt.pathname = 1;
 494                         continue;
 495                 }
 496                 if (!strcmp("-l", arg) ||
 497                     !strcmp("--files-with-matches", arg)) {
 498                         opt.name_only = 1;
 499                         continue;
 500                 }
 501                 if (!strcmp("-L", arg) ||
 502                     !strcmp("--files-without-match", arg)) {
 503                         opt.unmatch_name_only = 1;
 504                         continue;
 505                 }
 506                 if (!strcmp("-c", arg) ||
 507                     !strcmp("--count", arg)) {
 508                         opt.count = 1;
 509                         continue;
 510                 }
 511                 if (!strcmp("-w", arg) ||
 512                     !strcmp("--word-regexp", arg)) {
 513                         opt.word_regexp = 1;
 514                         continue;
 515                 }
 516                 if (!strncmp("-A", arg, 2) ||
 517                     !strncmp("-B", arg, 2) ||
 518                     !strncmp("-C", arg, 2) ||
 519                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
 520                         unsigned num;
 521                         const char *scan;
 522                         switch (arg[1]) {
 523                         case 'A': case 'B': case 'C':
 524                                 if (!arg[2]) {
 525                                         if (argc <= 1)
 526                                                 die(emsg_missing_context_len);
 527                                         scan = *++argv;
 528                                         argc--;
 529                                 }
 530                                 else
 531                                         scan = arg + 2;
 532                                 break;
 533                         default:
 534                                 scan = arg + 1;
 535                                 break;
 536                         }
 537                         if (sscanf(scan, "%u", &num) != 1)
 538                                 die(emsg_invalid_context_len, scan);
 539                         switch (arg[1]) {
 540                         case 'A':
 541                                 opt.post_context = num;
 542                                 break;
 543                         default:
 544                         case 'C':
 545                                 opt.post_context = num;
 546                         case 'B':
 547                                 opt.pre_context = num;
 548                                 break;
 549                         }
 550                         continue;
 551                 }
 552                 if (!strcmp("-f", arg)) {
 553                         FILE *patterns;
 554                         int lno = 0;
 555                         char buf[1024];
 556                         if (argc <= 1)
 557                                 die(emsg_missing_argument, arg);
 558                         patterns = fopen(argv[1], "r");
 559                         if (!patterns)
 560                                 die("'%s': %s", argv[1], strerror(errno));
 561                         while (fgets(buf, sizeof(buf), patterns)) {
 562                                 int len = strlen(buf);
 563                                 if (buf[len-1] == '\n')
 564                                         buf[len-1] = 0;
 565                                 /* ignore empty line like grep does */
 566                                 if (!buf[0])
 567                                         continue;
 568                                 append_grep_pattern(&opt, xstrdup(buf),
 569                                                     argv[1], ++lno,
 570                                                     GREP_PATTERN);
 571                         }
 572                         fclose(patterns);
 573                         argv++;
 574                         argc--;
 575                         continue;
 576                 }
 577                 if (!strcmp("--not", arg)) {
 578                         append_grep_pattern(&opt, arg, "command line", 0,
 579                                             GREP_NOT);
 580                         continue;
 581                 }
 582                 if (!strcmp("--and", arg)) {
 583                         append_grep_pattern(&opt, arg, "command line", 0,
 584                                             GREP_AND);
 585                         continue;
 586                 }
 587                 if (!strcmp("--or", arg))
 588                         continue; /* no-op */
 589                 if (!strcmp("(", arg)) {
 590                         append_grep_pattern(&opt, arg, "command line", 0,
 591                                             GREP_OPEN_PAREN);
 592                         continue;
 593                 }
 594                 if (!strcmp(")", arg)) {
 595                         append_grep_pattern(&opt, arg, "command line", 0,
 596                                             GREP_CLOSE_PAREN);
 597                         continue;
 598                 }
 599                 if (!strcmp("-e", arg)) {
 600                         if (1 < argc) {
 601                                 append_grep_pattern(&opt, argv[1],
 602                                                     "-e option", 0,
 603                                                     GREP_PATTERN);
 604                                 argv++;
 605                                 argc--;
 606                                 continue;
 607                         }
 608                         die(emsg_missing_argument, arg);
 609                 }
 610                 if (!strcmp("--full-name", arg)) {
 611                         opt.relative = 0;
 612                         continue;
 613                 }
 614                 if (!strcmp("--", arg)) {
 615                         /* later processing wants to have this at argv[1] */
 616                         argv--;
 617                         argc++;
 618                         break;
 619                 }
 620                 if (*arg == '-')
 621                         usage(builtin_grep_usage);
 622
 623                 /* First unrecognized non-option token */
 624                 if (!opt.pattern_list) {
 625                         append_grep_pattern(&opt, arg, "command line", 0,
 626                                             GREP_PATTERN);
 627                         break;
 628                 }
 629                 else {
 630                         /* We are looking at the first path or rev;
 631                          * it is found at argv[1] after leaving the
 632                          * loop.
 633                          */
 634                         argc++; argv--;
 635                         break;
 636                 }
 637         }
 638
 639         if (!opt.pattern_list)
 640                 die("no pattern given.");
 641         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
 642                 die("cannot mix --fixed-strings and regexp");
 643         compile_grep_patterns(&opt);
 644
 645         /* Check revs and then paths */
 646         for (i = 1; i < argc; i++) {
 647                 const char *arg = argv[i];
 648                 unsigned char sha1[20];
 649                 /* Is it a rev? */
 650                 if (!get_sha1(arg, sha1)) {
 651                         struct object *object = parse_object(sha1);
 652                         if (!object)
 653                                 die("bad object %s", arg);
 654                         add_object_array(object, arg, &list);
 655                         continue;
 656                 }
 657                 if (!strcmp(arg, "--")) {
 658                         i++;
 659                         seen_dashdash = 1;
 660                 }
 661                 break;
 662         }
 663
 664         /* The rest are paths */
 665         if (!seen_dashdash) {
 666                 int j;
 667                 for (j = i; j < argc; j++)
 668                         verify_filename(prefix, argv[j]);
 669         }
 670
 671         if (i < argc) {
 672                 paths = get_pathspec(prefix, argv + i);
 673                 if (opt.prefix_length && opt.relative) {
 674                         /* Make sure we do not get outside of paths */
 675                         for (i = 0; paths[i]; i++)
 676                                 if (strncmp(prefix, paths[i], opt.prefix_length))
 677                                         die("git-grep: cannot generate relative filenames containing '..'");
 678                 }
 679         }
 680         else if (prefix) {
 681                 paths = xcalloc(2, sizeof(const char *));
 682                 paths[0] = prefix;
 683                 paths[1] = NULL;
 684         }
 685
 686         if (!list.nr)
 687                 return !grep_cache(&opt, paths, cached);
 688
 689         if (cached)
 690                 die("both --cached and trees are given.");
 691
 692         for (i = 0; i < list.nr; i++) {
 693                 struct object *real_obj;
 694                 real_obj = deref_tag(list.objects[i].item, NULL, 0);
 695                 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
 696                         hit = 1;
 697         }
 698         free_grep_patterns(&opt);
 699         return !hit;
 700 }