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 "parse-options.h"
  14 #include "string-list.h"
  15 #include "run-command.h"
  16 #include "userdiff.h"
  17 #include "grep.h"
  18 #include "quote.h"
  19 #include "dir.h"
  20 #include "pathspec.h"
  21 #include "submodule.h"
  22
  23 static char const * const grep_usage[] = {
  24         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  25         NULL
  26 };
  27
  28 static const char *super_prefix;
  29 static int recurse_submodules;
  30 static struct argv_array submodule_options = ARGV_ARRAY_INIT;
  31
  32 static int grep_submodule_launch(struct grep_opt *opt,
  33                                  const struct grep_source *gs);
  34
  35 #define GREP_NUM_THREADS_DEFAULT 8
  36 static int num_threads;
  37
  38 #ifndef NO_PTHREADS
  39 static pthread_t *threads;
  40
  41 /* We use one producer thread and THREADS consumer
  42  * threads. The producer adds struct work_items to 'todo' and the
  43  * consumers pick work items from the same array.
  44  */
  45 struct work_item {
  46         struct grep_source source;
  47         char done;
  48         struct strbuf out;
  49 };
  50
  51 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  52  * that have been or are processed by a consumer thread. We haven't
  53  * written the result for these to stdout yet.
  54  *
  55  * The work_items in [todo_start, todo_end) are waiting to be picked
  56  * up by a consumer thread.
  57  *
  58  * The ranges are modulo TODO_SIZE.
  59  */
  60 #define TODO_SIZE 128
  61 static struct work_item todo[TODO_SIZE];
  62 static int todo_start;
  63 static int todo_end;
  64 static int todo_done;
  65
  66 /* Has all work items been added? */
  67 static int all_work_added;
  68
  69 /* This lock protects all the variables above. */
  70 static pthread_mutex_t grep_mutex;
  71
  72 static inline void grep_lock(void)
  73 {
  74         if (num_threads)
  75                 pthread_mutex_lock(&grep_mutex);
  76 }
  77
  78 static inline void grep_unlock(void)
  79 {
  80         if (num_threads)
  81                 pthread_mutex_unlock(&grep_mutex);
  82 }
  83
  84 /* Signalled when a new work_item is added to todo. */
  85 static pthread_cond_t cond_add;
  86
  87 /* Signalled when the result from one work_item is written to
  88  * stdout.
  89  */
  90 static pthread_cond_t cond_write;
  91
  92 /* Signalled when we are finished with everything. */
  93 static pthread_cond_t cond_result;
  94
  95 static int skip_first_line;
  96
  97 static void add_work(struct grep_opt *opt, enum grep_source_type type,
  98                      const char *name, const char *path, const void *id)
  99 {
 100         grep_lock();
 101
 102         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 103                 pthread_cond_wait(&cond_write, &grep_mutex);
 104         }
 105
 106         grep_source_init(&todo[todo_end].source, type, name, path, id);
 107         if (opt->binary != GREP_BINARY_TEXT)
 108                 grep_source_load_driver(&todo[todo_end].source);
 109         todo[todo_end].done = 0;
 110         strbuf_reset(&todo[todo_end].out);
 111         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 112
 113         pthread_cond_signal(&cond_add);
 114         grep_unlock();
 115 }
 116
 117 static struct work_item *get_work(void)
 118 {
 119         struct work_item *ret;
 120
 121         grep_lock();
 122         while (todo_start == todo_end && !all_work_added) {
 123                 pthread_cond_wait(&cond_add, &grep_mutex);
 124         }
 125
 126         if (todo_start == todo_end && all_work_added) {
 127                 ret = NULL;
 128         } else {
 129                 ret = &todo[todo_start];
 130                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 131         }
 132         grep_unlock();
 133         return ret;
 134 }
 135
 136 static void work_done(struct work_item *w)
 137 {
 138         int old_done;
 139
 140         grep_lock();
 141         w->done = 1;
 142         old_done = todo_done;
 143         for(; todo[todo_done].done && todo_done != todo_start;
 144             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 145                 w = &todo[todo_done];
 146                 if (w->out.len) {
 147                         const char *p = w->out.buf;
 148                         size_t len = w->out.len;
 149
 150                         /* Skip the leading hunk mark of the first file. */
 151                         if (skip_first_line) {
 152                                 while (len) {
 153                                         len--;
 154                                         if (*p++ == '\n')
 155                                                 break;
 156                                 }
 157                                 skip_first_line = 0;
 158                         }
 159
 160                         write_or_die(1, p, len);
 161                 }
 162                 grep_source_clear(&w->source);
 163         }
 164
 165         if (old_done != todo_done)
 166                 pthread_cond_signal(&cond_write);
 167
 168         if (all_work_added && todo_done == todo_end)
 169                 pthread_cond_signal(&cond_result);
 170
 171         grep_unlock();
 172 }
 173
 174 static void *run(void *arg)
 175 {
 176         int hit = 0;
 177         struct grep_opt *opt = arg;
 178
 179         while (1) {
 180                 struct work_item *w = get_work();
 181                 if (!w)
 182                         break;
 183
 184                 opt->output_priv = w;
 185                 if (w->source.type == GREP_SOURCE_SUBMODULE)
 186                         hit |= grep_submodule_launch(opt, &w->source);
 187                 else
 188                         hit |= grep_source(opt, &w->source);
 189                 grep_source_clear_data(&w->source);
 190                 work_done(w);
 191         }
 192         free_grep_patterns(arg);
 193         free(arg);
 194
 195         return (void*) (intptr_t) hit;
 196 }
 197
 198 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 199 {
 200         struct work_item *w = opt->output_priv;
 201         strbuf_add(&w->out, buf, size);
 202 }
 203
 204 static void start_threads(struct grep_opt *opt)
 205 {
 206         int i;
 207
 208         pthread_mutex_init(&grep_mutex, NULL);
 209         pthread_mutex_init(&grep_read_mutex, NULL);
 210         pthread_mutex_init(&grep_attr_mutex, NULL);
 211         pthread_cond_init(&cond_add, NULL);
 212         pthread_cond_init(&cond_write, NULL);
 213         pthread_cond_init(&cond_result, NULL);
 214         grep_use_locks = 1;
 215
 216         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 217                 strbuf_init(&todo[i].out, 0);
 218         }
 219
 220         threads = xcalloc(num_threads, sizeof(*threads));
 221         for (i = 0; i < num_threads; i++) {
 222                 int err;
 223                 struct grep_opt *o = grep_opt_dup(opt);
 224                 o->output = strbuf_out;
 225                 o->debug = 0;
 226                 compile_grep_patterns(o);
 227                 err = pthread_create(&threads[i], NULL, run, o);
 228
 229                 if (err)
 230                         die(_("grep: failed to create thread: %s"),
 231                             strerror(err));
 232         }
 233 }
 234
 235 static int wait_all(void)
 236 {
 237         int hit = 0;
 238         int i;
 239
 240         grep_lock();
 241         all_work_added = 1;
 242
 243         /* Wait until all work is done. */
 244         while (todo_done != todo_end)
 245                 pthread_cond_wait(&cond_result, &grep_mutex);
 246
 247         /* Wake up all the consumer threads so they can see that there
 248          * is no more work to do.
 249          */
 250         pthread_cond_broadcast(&cond_add);
 251         grep_unlock();
 252
 253         for (i = 0; i < num_threads; i++) {
 254                 void *h;
 255                 pthread_join(threads[i], &h);
 256                 hit |= (int) (intptr_t) h;
 257         }
 258
 259         free(threads);
 260
 261         pthread_mutex_destroy(&grep_mutex);
 262         pthread_mutex_destroy(&grep_read_mutex);
 263         pthread_mutex_destroy(&grep_attr_mutex);
 264         pthread_cond_destroy(&cond_add);
 265         pthread_cond_destroy(&cond_write);
 266         pthread_cond_destroy(&cond_result);
 267         grep_use_locks = 0;
 268
 269         return hit;
 270 }
 271 #else /* !NO_PTHREADS */
 272
 273 static int wait_all(void)
 274 {
 275         return 0;
 276 }
 277 #endif
 278
 279 static int grep_cmd_config(const char *var, const char *value, void *cb)
 280 {
 281         int st = grep_config(var, value, cb);
 282         if (git_color_default_config(var, value, cb) < 0)
 283                 st = -1;
 284
 285         if (!strcmp(var, "grep.threads")) {
 286                 num_threads = git_config_int(var, value);
 287                 if (num_threads < 0)
 288                         die(_("invalid number of threads specified (%d) for %s"),
 289                             num_threads, var);
 290         }
 291
 292         return st;
 293 }
 294
 295 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 296 {
 297         void *data;
 298
 299         grep_read_lock();
 300         data = read_sha1_file(sha1, type, size);
 301         grep_read_unlock();
 302         return data;
 303 }
 304
 305 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 306                      const char *filename, int tree_name_len,
 307                      const char *path)
 308 {
 309         struct strbuf pathbuf = STRBUF_INIT;
 310
 311         if (opt->relative && opt->prefix_length) {
 312                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
 313                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 314         } else if (super_prefix) {
 315                 strbuf_add(&pathbuf, filename, tree_name_len);
 316                 strbuf_addstr(&pathbuf, super_prefix);
 317                 strbuf_addstr(&pathbuf, filename + tree_name_len);
 318         } else {
 319                 strbuf_addstr(&pathbuf, filename);
 320         }
 321
 322 #ifndef NO_PTHREADS
 323         if (num_threads) {
 324                 add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, path, sha1);
 325                 strbuf_release(&pathbuf);
 326                 return 0;
 327         } else
 328 #endif
 329         {
 330                 struct grep_source gs;
 331                 int hit;
 332
 333                 grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, path, sha1);
 334                 strbuf_release(&pathbuf);
 335                 hit = grep_source(opt, &gs);
 336
 337                 grep_source_clear(&gs);
 338                 return hit;
 339         }
 340 }
 341
 342 static int grep_file(struct grep_opt *opt, const char *filename)
 343 {
 344         struct strbuf buf = STRBUF_INIT;
 345
 346         if (opt->relative && opt->prefix_length) {
 347                 quote_path_relative(filename, opt->prefix, &buf);
 348         } else {
 349                 if (super_prefix)
 350                         strbuf_addstr(&buf, super_prefix);
 351                 strbuf_addstr(&buf, filename);
 352         }
 353
 354 #ifndef NO_PTHREADS
 355         if (num_threads) {
 356                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
 357                 strbuf_release(&buf);
 358                 return 0;
 359         } else
 360 #endif
 361         {
 362                 struct grep_source gs;
 363                 int hit;
 364
 365                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 366                 strbuf_release(&buf);
 367                 hit = grep_source(opt, &gs);
 368
 369                 grep_source_clear(&gs);
 370                 return hit;
 371         }
 372 }
 373
 374 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 375 {
 376         struct string_list *path_list = opt->output_priv;
 377
 378         if (len == 1 && *(const char *)data == '\0')
 379                 return;
 380         string_list_append(path_list, xstrndup(data, len));
 381 }
 382
 383 static void run_pager(struct grep_opt *opt, const char *prefix)
 384 {
 385         struct string_list *path_list = opt->output_priv;
 386         struct child_process child = CHILD_PROCESS_INIT;
 387         int i, status;
 388
 389         for (i = 0; i < path_list->nr; i++)
 390                 argv_array_push(&child.args, path_list->items[i].string);
 391         child.dir = prefix;
 392         child.use_shell = 1;
 393
 394         status = run_command(&child);
 395         if (status)
 396                 exit(status);
 397 }
 398
 399 static void compile_submodule_options(const struct grep_opt *opt,
 400                                       const struct pathspec *pathspec,
 401                                       int cached, int untracked,
 402                                       int opt_exclude, int use_index,
 403                                       int pattern_type_arg)
 404 {
 405         struct grep_pat *pattern;
 406         int i;
 407
 408         if (recurse_submodules)
 409                 argv_array_push(&submodule_options, "--recurse-submodules");
 410
 411         if (cached)
 412                 argv_array_push(&submodule_options, "--cached");
 413         if (!use_index)
 414                 argv_array_push(&submodule_options, "--no-index");
 415         if (untracked)
 416                 argv_array_push(&submodule_options, "--untracked");
 417         if (opt_exclude > 0)
 418                 argv_array_push(&submodule_options, "--exclude-standard");
 419
 420         if (opt->invert)
 421                 argv_array_push(&submodule_options, "-v");
 422         if (opt->ignore_case)
 423                 argv_array_push(&submodule_options, "-i");
 424         if (opt->word_regexp)
 425                 argv_array_push(&submodule_options, "-w");
 426         switch (opt->binary) {
 427         case GREP_BINARY_NOMATCH:
 428                 argv_array_push(&submodule_options, "-I");
 429                 break;
 430         case GREP_BINARY_TEXT:
 431                 argv_array_push(&submodule_options, "-a");
 432                 break;
 433         default:
 434                 break;
 435         }
 436         if (opt->allow_textconv)
 437                 argv_array_push(&submodule_options, "--textconv");
 438         if (opt->max_depth != -1)
 439                 argv_array_pushf(&submodule_options, "--max-depth=%d",
 440                                  opt->max_depth);
 441         if (opt->linenum)
 442                 argv_array_push(&submodule_options, "-n");
 443         if (!opt->pathname)
 444                 argv_array_push(&submodule_options, "-h");
 445         if (!opt->relative)
 446                 argv_array_push(&submodule_options, "--full-name");
 447         if (opt->name_only)
 448                 argv_array_push(&submodule_options, "-l");
 449         if (opt->unmatch_name_only)
 450                 argv_array_push(&submodule_options, "-L");
 451         if (opt->null_following_name)
 452                 argv_array_push(&submodule_options, "-z");
 453         if (opt->count)
 454                 argv_array_push(&submodule_options, "-c");
 455         if (opt->file_break)
 456                 argv_array_push(&submodule_options, "--break");
 457         if (opt->heading)
 458                 argv_array_push(&submodule_options, "--heading");
 459         if (opt->pre_context)
 460                 argv_array_pushf(&submodule_options, "--before-context=%d",
 461                                  opt->pre_context);
 462         if (opt->post_context)
 463                 argv_array_pushf(&submodule_options, "--after-context=%d",
 464                                  opt->post_context);
 465         if (opt->funcname)
 466                 argv_array_push(&submodule_options, "-p");
 467         if (opt->funcbody)
 468                 argv_array_push(&submodule_options, "-W");
 469         if (opt->all_match)
 470                 argv_array_push(&submodule_options, "--all-match");
 471         if (opt->debug)
 472                 argv_array_push(&submodule_options, "--debug");
 473         if (opt->status_only)
 474                 argv_array_push(&submodule_options, "-q");
 475
 476         switch (pattern_type_arg) {
 477         case GREP_PATTERN_TYPE_BRE:
 478                 argv_array_push(&submodule_options, "-G");
 479                 break;
 480         case GREP_PATTERN_TYPE_ERE:
 481                 argv_array_push(&submodule_options, "-E");
 482                 break;
 483         case GREP_PATTERN_TYPE_FIXED:
 484                 argv_array_push(&submodule_options, "-F");
 485                 break;
 486         case GREP_PATTERN_TYPE_PCRE:
 487                 argv_array_push(&submodule_options, "-P");
 488                 break;
 489         case GREP_PATTERN_TYPE_UNSPECIFIED:
 490                 break;
 491         }
 492
 493         for (pattern = opt->pattern_list; pattern != NULL;
 494              pattern = pattern->next) {
 495                 switch (pattern->token) {
 496                 case GREP_PATTERN:
 497                         argv_array_pushf(&submodule_options, "-e%s",
 498                                          pattern->pattern);
 499                         break;
 500                 case GREP_AND:
 501                 case GREP_OPEN_PAREN:
 502                 case GREP_CLOSE_PAREN:
 503                 case GREP_NOT:
 504                 case GREP_OR:
 505                         argv_array_push(&submodule_options, pattern->pattern);
 506                         break;
 507                 /* BODY and HEAD are not used by git-grep */
 508                 case GREP_PATTERN_BODY:
 509                 case GREP_PATTERN_HEAD:
 510                         break;
 511                 }
 512         }
 513
 514         /*
 515          * Limit number of threads for child process to use.
 516          * This is to prevent potential fork-bomb behavior of git-grep as each
 517          * submodule process has its own thread pool.
 518          */
 519         argv_array_pushf(&submodule_options, "--threads=%d",
 520                          (num_threads + 1) / 2);
 521
 522         /* Add Pathspecs */
 523         argv_array_push(&submodule_options, "--");
 524         for (i = 0; i < pathspec->nr; i++)
 525                 argv_array_push(&submodule_options,
 526                                 pathspec->items[i].original);
 527 }
 528
 529 /*
 530  * Launch child process to grep contents of a submodule
 531  */
 532 static int grep_submodule_launch(struct grep_opt *opt,
 533                                  const struct grep_source *gs)
 534 {
 535         struct child_process cp = CHILD_PROCESS_INIT;
 536         int status, i;
 537         struct work_item *w = opt->output_priv;
 538
 539         prepare_submodule_repo_env(&cp.env_array);
 540
 541         /* Add super prefix */
 542         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
 543                          super_prefix ? super_prefix : "",
 544                          gs->name);
 545         argv_array_push(&cp.args, "grep");
 546
 547         /* Add options */
 548         for (i = 0; i < submodule_options.argc; i++)
 549                 argv_array_push(&cp.args, submodule_options.argv[i]);
 550
 551         cp.git_cmd = 1;
 552         cp.dir = gs->path;
 553
 554         /*
 555          * Capture output to output buffer and check the return code from the
 556          * child process.  A '0' indicates a hit, a '1' indicates no hit and
 557          * anything else is an error.
 558          */
 559         status = capture_command(&cp, &w->out, 0);
 560         if (status && (status != 1)) {
 561                 /* flush the buffer */
 562                 write_or_die(1, w->out.buf, w->out.len);
 563                 die("process for submodule '%s' failed with exit code: %d",
 564                     gs->name, status);
 565         }
 566
 567         /* invert the return code to make a hit equal to 1 */
 568         return !status;
 569 }
 570
 571 /*
 572  * Prep grep structures for a submodule grep
 573  * sha1: the sha1 of the submodule or NULL if using the working tree
 574  * filename: name of the submodule including tree name of parent
 575  * path: location of the submodule
 576  */
 577 static int grep_submodule(struct grep_opt *opt, const unsigned char *sha1,
 578                           const char *filename, const char *path)
 579 {
 580         if (!is_submodule_initialized(path))
 581                 return 0;
 582         if (!is_submodule_populated(path))
 583                 return 0;
 584
 585 #ifndef NO_PTHREADS
 586         if (num_threads) {
 587                 add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, sha1);
 588                 return 0;
 589         } else
 590 #endif
 591         {
 592                 struct work_item w;
 593                 int hit;
 594
 595                 grep_source_init(&w.source, GREP_SOURCE_SUBMODULE,
 596                                  filename, path, sha1);
 597                 strbuf_init(&w.out, 0);
 598                 opt->output_priv = &w;
 599                 hit = grep_submodule_launch(opt, &w.source);
 600
 601                 write_or_die(1, w.out.buf, w.out.len);
 602
 603                 grep_source_clear(&w.source);
 604                 strbuf_release(&w.out);
 605                 return hit;
 606         }
 607 }
 608
 609 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
 610                       int cached)
 611 {
 612         int hit = 0;
 613         int nr;
 614         struct strbuf name = STRBUF_INIT;
 615         int name_base_len = 0;
 616         if (super_prefix) {
 617                 name_base_len = strlen(super_prefix);
 618                 strbuf_addstr(&name, super_prefix);
 619         }
 620
 621         read_cache();
 622
 623         for (nr = 0; nr < active_nr; nr++) {
 624                 const struct cache_entry *ce = active_cache[nr];
 625                 strbuf_setlen(&name, name_base_len);
 626                 strbuf_addstr(&name, ce->name);
 627
 628                 if (S_ISREG(ce->ce_mode) &&
 629                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
 630                                    S_ISDIR(ce->ce_mode) ||
 631                                    S_ISGITLINK(ce->ce_mode))) {
 632                         /*
 633                          * If CE_VALID is on, we assume worktree file and its
 634                          * cache entry are identical, even if worktree file has
 635                          * been modified, so use cache version instead
 636                          */
 637                         if (cached || (ce->ce_flags & CE_VALID) ||
 638                             ce_skip_worktree(ce)) {
 639                                 if (ce_stage(ce) || ce_intent_to_add(ce))
 640                                         continue;
 641                                 hit |= grep_sha1(opt, ce->oid.hash, ce->name,
 642                                                  0, ce->name);
 643                         } else {
 644                                 hit |= grep_file(opt, ce->name);
 645                         }
 646                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 647                            submodule_path_match(pathspec, name.buf, NULL)) {
 648                         hit |= grep_submodule(opt, NULL, ce->name, ce->name);
 649                 } else {
 650                         continue;
 651                 }
 652
 653                 if (ce_stage(ce)) {
 654                         do {
 655                                 nr++;
 656                         } while (nr < active_nr &&
 657                                  !strcmp(ce->name, active_cache[nr]->name));
 658                         nr--; /* compensate for loop control */
 659                 }
 660                 if (hit && opt->status_only)
 661                         break;
 662         }
 663
 664         strbuf_release(&name);
 665         return hit;
 666 }
 667
 668 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 669                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 670                      int check_attr)
 671 {
 672         int hit = 0;
 673         enum interesting match = entry_not_interesting;
 674         struct name_entry entry;
 675         int old_baselen = base->len;
 676
 677         while (tree_entry(tree, &entry)) {
 678                 int te_len = tree_entry_len(&entry);
 679
 680                 if (match != all_entries_interesting) {
 681                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 682                         if (match == all_entries_not_interesting)
 683                                 break;
 684                         if (match == entry_not_interesting)
 685                                 continue;
 686                 }
 687
 688                 strbuf_add(base, entry.path, te_len);
 689
 690                 if (S_ISREG(entry.mode)) {
 691                         hit |= grep_sha1(opt, entry.oid->hash, base->buf, tn_len,
 692                                          check_attr ? base->buf + tn_len : NULL);
 693                 }
 694                 else if (S_ISDIR(entry.mode)) {
 695                         enum object_type type;
 696                         struct tree_desc sub;
 697                         void *data;
 698                         unsigned long size;
 699
 700                         data = lock_and_read_sha1_file(entry.oid->hash, &type, &size);
 701                         if (!data)
 702                                 die(_("unable to read tree (%s)"),
 703                                     oid_to_hex(entry.oid));
 704
 705                         strbuf_addch(base, '/');
 706                         init_tree_desc(&sub, data, size);
 707                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 708                                          check_attr);
 709                         free(data);
 710                 }
 711                 strbuf_setlen(base, old_baselen);
 712
 713                 if (hit && opt->status_only)
 714                         break;
 715         }
 716         return hit;
 717 }
 718
 719 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 720                        struct object *obj, const char *name, const char *path)
 721 {
 722         if (obj->type == OBJ_BLOB)
 723                 return grep_sha1(opt, obj->oid.hash, name, 0, path);
 724         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 725                 struct tree_desc tree;
 726                 void *data;
 727                 unsigned long size;
 728                 struct strbuf base;
 729                 int hit, len;
 730
 731                 grep_read_lock();
 732                 data = read_object_with_reference(obj->oid.hash, tree_type,
 733                                                   &size, NULL);
 734                 grep_read_unlock();
 735
 736                 if (!data)
 737                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 738
 739                 len = name ? strlen(name) : 0;
 740                 strbuf_init(&base, PATH_MAX + len + 1);
 741                 if (len) {
 742                         strbuf_add(&base, name, len);
 743                         strbuf_addch(&base, ':');
 744                 }
 745                 init_tree_desc(&tree, data, size);
 746                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 747                                 obj->type == OBJ_COMMIT);
 748                 strbuf_release(&base);
 749                 free(data);
 750                 return hit;
 751         }
 752         die(_("unable to grep from object of type %s"), typename(obj->type));
 753 }
 754
 755 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 756                         const struct object_array *list)
 757 {
 758         unsigned int i;
 759         int hit = 0;
 760         const unsigned int nr = list->nr;
 761
 762         for (i = 0; i < nr; i++) {
 763                 struct object *real_obj;
 764                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 765                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
 766                         hit = 1;
 767                         if (opt->status_only)
 768                                 break;
 769                 }
 770         }
 771         return hit;
 772 }
 773
 774 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 775                           int exc_std, int use_index)
 776 {
 777         struct dir_struct dir;
 778         int i, hit = 0;
 779
 780         memset(&dir, 0, sizeof(dir));
 781         if (!use_index)
 782                 dir.flags |= DIR_NO_GITLINKS;
 783         if (exc_std)
 784                 setup_standard_excludes(&dir);
 785
 786         fill_directory(&dir, pathspec);
 787         for (i = 0; i < dir.nr; i++) {
 788                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
 789                         continue;
 790                 hit |= grep_file(opt, dir.entries[i]->name);
 791                 if (hit && opt->status_only)
 792                         break;
 793         }
 794         return hit;
 795 }
 796
 797 static int context_callback(const struct option *opt, const char *arg,
 798                             int unset)
 799 {
 800         struct grep_opt *grep_opt = opt->value;
 801         int value;
 802         const char *endp;
 803
 804         if (unset) {
 805                 grep_opt->pre_context = grep_opt->post_context = 0;
 806                 return 0;
 807         }
 808         value = strtol(arg, (char **)&endp, 10);
 809         if (*endp) {
 810                 return error(_("switch `%c' expects a numerical value"),
 811                              opt->short_name);
 812         }
 813         grep_opt->pre_context = grep_opt->post_context = value;
 814         return 0;
 815 }
 816
 817 static int file_callback(const struct option *opt, const char *arg, int unset)
 818 {
 819         struct grep_opt *grep_opt = opt->value;
 820         int from_stdin = !strcmp(arg, "-");
 821         FILE *patterns;
 822         int lno = 0;
 823         struct strbuf sb = STRBUF_INIT;
 824
 825         patterns = from_stdin ? stdin : fopen(arg, "r");
 826         if (!patterns)
 827                 die_errno(_("cannot open '%s'"), arg);
 828         while (strbuf_getline(&sb, patterns) == 0) {
 829                 /* ignore empty line like grep does */
 830                 if (sb.len == 0)
 831                         continue;
 832
 833                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 834                                 GREP_PATTERN);
 835         }
 836         if (!from_stdin)
 837                 fclose(patterns);
 838         strbuf_release(&sb);
 839         return 0;
 840 }
 841
 842 static int not_callback(const struct option *opt, const char *arg, int unset)
 843 {
 844         struct grep_opt *grep_opt = opt->value;
 845         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 846         return 0;
 847 }
 848
 849 static int and_callback(const struct option *opt, const char *arg, int unset)
 850 {
 851         struct grep_opt *grep_opt = opt->value;
 852         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 853         return 0;
 854 }
 855
 856 static int open_callback(const struct option *opt, const char *arg, int unset)
 857 {
 858         struct grep_opt *grep_opt = opt->value;
 859         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 860         return 0;
 861 }
 862
 863 static int close_callback(const struct option *opt, const char *arg, int unset)
 864 {
 865         struct grep_opt *grep_opt = opt->value;
 866         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 867         return 0;
 868 }
 869
 870 static int pattern_callback(const struct option *opt, const char *arg,
 871                             int unset)
 872 {
 873         struct grep_opt *grep_opt = opt->value;
 874         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 875         return 0;
 876 }
 877
 878 int cmd_grep(int argc, const char **argv, const char *prefix)
 879 {
 880         int hit = 0;
 881         int cached = 0, untracked = 0, opt_exclude = -1;
 882         int seen_dashdash = 0;
 883         int external_grep_allowed__ignored;
 884         const char *show_in_pager = NULL, *default_pager = "dummy";
 885         struct grep_opt opt;
 886         struct object_array list = OBJECT_ARRAY_INIT;
 887         struct pathspec pathspec;
 888         struct string_list path_list = STRING_LIST_INIT_NODUP;
 889         int i;
 890         int dummy;
 891         int use_index = 1;
 892         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 893
 894         struct option options[] = {
 895                 OPT_BOOL(0, "cached", &cached,
 896                         N_("search in index instead of in the work tree")),
 897                 OPT_NEGBIT(0, "no-index", &use_index,
 898                          N_("find in contents not managed by git"), 1),
 899                 OPT_BOOL(0, "untracked", &untracked,
 900                         N_("search in both tracked and untracked files")),
 901                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 902                             N_("ignore files specified via '.gitignore'"), 1),
 903                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 904                          N_("recursivley search in each submodule")),
 905                 OPT_GROUP(""),
 906                 OPT_BOOL('v', "invert-match", &opt.invert,
 907                         N_("show non-matching lines")),
 908                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 909                         N_("case insensitive matching")),
 910                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 911                         N_("match patterns only at word boundaries")),
 912                 OPT_SET_INT('a', "text", &opt.binary,
 913                         N_("process binary files as text"), GREP_BINARY_TEXT),
 914                 OPT_SET_INT('I', NULL, &opt.binary,
 915                         N_("don't match patterns in binary files"),
 916                         GREP_BINARY_NOMATCH),
 917                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
 918                          N_("process binary files with textconv filters")),
 919                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
 920                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
 921                         NULL, 1 },
 922                 OPT_GROUP(""),
 923                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
 924                             N_("use extended POSIX regular expressions"),
 925                             GREP_PATTERN_TYPE_ERE),
 926                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
 927                             N_("use basic POSIX regular expressions (default)"),
 928                             GREP_PATTERN_TYPE_BRE),
 929                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
 930                             N_("interpret patterns as fixed strings"),
 931                             GREP_PATTERN_TYPE_FIXED),
 932                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
 933                             N_("use Perl-compatible regular expressions"),
 934                             GREP_PATTERN_TYPE_PCRE),
 935                 OPT_GROUP(""),
 936                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
 937                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
 938                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
 939                 OPT_NEGBIT(0, "full-name", &opt.relative,
 940                         N_("show filenames relative to top directory"), 1),
 941                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
 942                         N_("show only filenames instead of matching lines")),
 943                 OPT_BOOL(0, "name-only", &opt.name_only,
 944                         N_("synonym for --files-with-matches")),
 945                 OPT_BOOL('L', "files-without-match",
 946                         &opt.unmatch_name_only,
 947                         N_("show only the names of files without match")),
 948                 OPT_BOOL('z', "null", &opt.null_following_name,
 949                         N_("print NUL after filenames")),
 950                 OPT_BOOL('c', "count", &opt.count,
 951                         N_("show the number of matches instead of matching lines")),
 952                 OPT__COLOR(&opt.color, N_("highlight matches")),
 953                 OPT_BOOL(0, "break", &opt.file_break,
 954                         N_("print empty line between matches from different files")),
 955                 OPT_BOOL(0, "heading", &opt.heading,
 956                         N_("show filename only once above matches from same file")),
 957                 OPT_GROUP(""),
 958                 OPT_CALLBACK('C', "context", &opt, N_("n"),
 959                         N_("show <n> context lines before and after matches"),
 960                         context_callback),
 961                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 962                         N_("show <n> context lines before matches")),
 963                 OPT_INTEGER('A', "after-context", &opt.post_context,
 964                         N_("show <n> context lines after matches")),
 965                 OPT_INTEGER(0, "threads", &num_threads,
 966                         N_("use <n> worker threads")),
 967                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
 968                         context_callback),
 969                 OPT_BOOL('p', "show-function", &opt.funcname,
 970                         N_("show a line with the function name before matches")),
 971                 OPT_BOOL('W', "function-context", &opt.funcbody,
 972                         N_("show the surrounding function")),
 973                 OPT_GROUP(""),
 974                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
 975                         N_("read patterns from file"), file_callback),
 976                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
 977                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
 978                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 979                   N_("combine patterns specified with -e"),
 980                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 981                 OPT_BOOL(0, "or", &dummy, ""),
 982                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 983                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 984                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 985                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 986                   open_callback },
 987                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 988                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 989                   close_callback },
 990                 OPT__QUIET(&opt.status_only,
 991                            N_("indicate hit with exit status without output")),
 992                 OPT_BOOL(0, "all-match", &opt.all_match,
 993                         N_("show only matches from files that match all patterns")),
 994                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
 995                   N_("show parse tree for grep expression"),
 996                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
 997                 OPT_GROUP(""),
 998                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 999                         N_("pager"), N_("show matching files in the pager"),
1000                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1001                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1002                          N_("allow calling of grep(1) (ignored by this build)")),
1003                 OPT_END()
1004         };
1005
1006         init_grep_defaults();
1007         git_config(grep_cmd_config, NULL);
1008         grep_init(&opt, prefix);
1009         super_prefix = get_super_prefix();
1010
1011         /*
1012          * If there is no -- then the paths must exist in the working
1013          * tree.  If there is no explicit pattern specified with -e or
1014          * -f, we take the first unrecognized non option to be the
1015          * pattern, but then what follows it must be zero or more
1016          * valid refs up to the -- (if exists), and then existing
1017          * paths.  If there is an explicit pattern, then the first
1018          * unrecognized non option is the beginning of the refs list
1019          * that continues up to the -- (if exists), and then paths.
1020          */
1021         argc = parse_options(argc, argv, prefix, options, grep_usage,
1022                              PARSE_OPT_KEEP_DASHDASH |
1023                              PARSE_OPT_STOP_AT_NON_OPTION);
1024         grep_commit_pattern_type(pattern_type_arg, &opt);
1025
1026         if (use_index && !startup_info->have_repository) {
1027                 int fallback = 0;
1028                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
1029                 if (fallback)
1030                         use_index = 0;
1031                 else
1032                         /* die the same way as if we did it at the beginning */
1033                         setup_git_directory();
1034         }
1035
1036         /*
1037          * skip a -- separator; we know it cannot be
1038          * separating revisions from pathnames if
1039          * we haven't even had any patterns yet
1040          */
1041         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1042                 argv++;
1043                 argc--;
1044         }
1045
1046         /* First unrecognized non-option token */
1047         if (argc > 0 && !opt.pattern_list) {
1048                 append_grep_pattern(&opt, argv[0], "command line", 0,
1049                                     GREP_PATTERN);
1050                 argv++;
1051                 argc--;
1052         }
1053
1054         if (show_in_pager == default_pager)
1055                 show_in_pager = git_pager(1);
1056         if (show_in_pager) {
1057                 opt.color = 0;
1058                 opt.name_only = 1;
1059                 opt.null_following_name = 1;
1060                 opt.output_priv = &path_list;
1061                 opt.output = append_path;
1062                 string_list_append(&path_list, show_in_pager);
1063         }
1064
1065         if (!opt.pattern_list)
1066                 die(_("no pattern given."));
1067         if (!opt.fixed && opt.ignore_case)
1068                 opt.regflags |= REG_ICASE;
1069
1070         compile_grep_patterns(&opt);
1071
1072         /* Check revs and then paths */
1073         for (i = 0; i < argc; i++) {
1074                 const char *arg = argv[i];
1075                 unsigned char sha1[20];
1076                 struct object_context oc;
1077                 /* Is it a rev? */
1078                 if (!get_sha1_with_context(arg, 0, sha1, &oc)) {
1079                         struct object *object = parse_object_or_die(sha1, arg);
1080                         if (!seen_dashdash)
1081                                 verify_non_filename(prefix, arg);
1082                         add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1083                         continue;
1084                 }
1085                 if (!strcmp(arg, "--")) {
1086                         i++;
1087                         seen_dashdash = 1;
1088                 }
1089                 break;
1090         }
1091
1092 #ifndef NO_PTHREADS
1093         if (list.nr || cached || show_in_pager)
1094                 num_threads = 0;
1095         else if (num_threads == 0)
1096                 num_threads = GREP_NUM_THREADS_DEFAULT;
1097         else if (num_threads < 0)
1098                 die(_("invalid number of threads specified (%d)"), num_threads);
1099 #else
1100         num_threads = 0;
1101 #endif
1102
1103 #ifndef NO_PTHREADS
1104         if (num_threads) {
1105                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1106                     && (opt.pre_context || opt.post_context ||
1107                         opt.file_break || opt.funcbody))
1108                         skip_first_line = 1;
1109                 start_threads(&opt);
1110         }
1111 #endif
1112
1113         /* The rest are paths */
1114         if (!seen_dashdash) {
1115                 int j;
1116                 for (j = i; j < argc; j++)
1117                         verify_filename(prefix, argv[j], j == i);
1118         }
1119
1120         parse_pathspec(&pathspec, 0,
1121                        PATHSPEC_PREFER_CWD |
1122                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1123                        prefix, argv + i);
1124         pathspec.max_depth = opt.max_depth;
1125         pathspec.recursive = 1;
1126
1127         if (recurse_submodules) {
1128                 gitmodules_config();
1129                 compile_submodule_options(&opt, &pathspec, cached, untracked,
1130                                           opt_exclude, use_index,
1131                                           pattern_type_arg);
1132         }
1133
1134         if (show_in_pager && (cached || list.nr))
1135                 die(_("--open-files-in-pager only works on the worktree"));
1136
1137         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1138                 const char *pager = path_list.items[0].string;
1139                 int len = strlen(pager);
1140
1141                 if (len > 4 && is_dir_sep(pager[len - 5]))
1142                         pager += len - 4;
1143
1144                 if (opt.ignore_case && !strcmp("less", pager))
1145                         string_list_append(&path_list, "-I");
1146
1147                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1148                         struct strbuf buf = STRBUF_INIT;
1149                         strbuf_addf(&buf, "+/%s%s",
1150                                         strcmp("less", pager) ? "" : "*",
1151                                         opt.pattern_list->pattern);
1152                         string_list_append(&path_list, buf.buf);
1153                         strbuf_detach(&buf, NULL);
1154                 }
1155         }
1156
1157         if (recurse_submodules && (!use_index || untracked || list.nr))
1158                 die(_("option not supported with --recurse-submodules."));
1159
1160         if (!show_in_pager && !opt.status_only)
1161                 setup_pager();
1162
1163         if (!use_index && (untracked || cached))
1164                 die(_("--cached or --untracked cannot be used with --no-index."));
1165
1166         if (!use_index || untracked) {
1167                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1168                 if (list.nr)
1169                         die(_("--no-index or --untracked cannot be used with revs."));
1170                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1171         } else if (0 <= opt_exclude) {
1172                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1173         } else if (!list.nr) {
1174                 if (!cached)
1175                         setup_work_tree();
1176
1177                 hit = grep_cache(&opt, &pathspec, cached);
1178         } else {
1179                 if (cached)
1180                         die(_("both --cached and trees are given."));
1181                 hit = grep_objects(&opt, &pathspec, &list);
1182         }
1183
1184         if (num_threads)
1185                 hit |= wait_all();
1186         if (hit && show_in_pager)
1187                 run_pager(&opt, prefix);
1188         free_grep_patterns(&opt);
1189         return !hit;
1190 }