git.oblomov.eu Git - git/blob - builtin/grep.c

   1 /*
   2  * Builtin "git grep"
   3  *
   4  * Copyright (c) 2006 Junio C Hamano
   5  */
   6 #include "cache.h"
   7 #include "repository.h"
   8 #include "config.h"
   9 #include "blob.h"
  10 #include "tree.h"
  11 #include "commit.h"
  12 #include "tag.h"
  13 #include "tree-walk.h"
  14 #include "builtin.h"
  15 #include "parse-options.h"
  16 #include "string-list.h"
  17 #include "run-command.h"
  18 #include "userdiff.h"
  19 #include "grep.h"
  20 #include "quote.h"
  21 #include "dir.h"
  22 #include "pathspec.h"
  23 #include "submodule.h"
  24 #include "submodule-config.h"
  25 #include "object-store.h"
  26
  27 static char const * const grep_usage[] = {
  28         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  29         NULL
  30 };
  31
  32 static int recurse_submodules;
  33
  34 #define GREP_NUM_THREADS_DEFAULT 8
  35 static int num_threads;
  36
  37 #ifndef NO_PTHREADS
  38 static pthread_t *threads;
  39
  40 /* We use one producer thread and THREADS consumer
  41  * threads. The producer adds struct work_items to 'todo' and the
  42  * consumers pick work items from the same array.
  43  */
  44 struct work_item {
  45         struct grep_source source;
  46         char done;
  47         struct strbuf out;
  48 };
  49
  50 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  51  * that have been or are processed by a consumer thread. We haven't
  52  * written the result for these to stdout yet.
  53  *
  54  * The work_items in [todo_start, todo_end) are waiting to be picked
  55  * up by a consumer thread.
  56  *
  57  * The ranges are modulo TODO_SIZE.
  58  */
  59 #define TODO_SIZE 128
  60 static struct work_item todo[TODO_SIZE];
  61 static int todo_start;
  62 static int todo_end;
  63 static int todo_done;
  64
  65 /* Has all work items been added? */
  66 static int all_work_added;
  67
  68 /* This lock protects all the variables above. */
  69 static pthread_mutex_t grep_mutex;
  70
  71 static inline void grep_lock(void)
  72 {
  73         assert(num_threads);
  74         pthread_mutex_lock(&grep_mutex);
  75 }
  76
  77 static inline void grep_unlock(void)
  78 {
  79         assert(num_threads);
  80         pthread_mutex_unlock(&grep_mutex);
  81 }
  82
  83 /* Signalled when a new work_item is added to todo. */
  84 static pthread_cond_t cond_add;
  85
  86 /* Signalled when the result from one work_item is written to
  87  * stdout.
  88  */
  89 static pthread_cond_t cond_write;
  90
  91 /* Signalled when we are finished with everything. */
  92 static pthread_cond_t cond_result;
  93
  94 static int skip_first_line;
  95
  96 static void add_work(struct grep_opt *opt, const struct grep_source *gs)
  97 {
  98         grep_lock();
  99
 100         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 101                 pthread_cond_wait(&cond_write, &grep_mutex);
 102         }
 103
 104         todo[todo_end].source = *gs;
 105         if (opt->binary != GREP_BINARY_TEXT)
 106                 grep_source_load_driver(&todo[todo_end].source);
 107         todo[todo_end].done = 0;
 108         strbuf_reset(&todo[todo_end].out);
 109         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 110
 111         pthread_cond_signal(&cond_add);
 112         grep_unlock();
 113 }
 114
 115 static struct work_item *get_work(void)
 116 {
 117         struct work_item *ret;
 118
 119         grep_lock();
 120         while (todo_start == todo_end && !all_work_added) {
 121                 pthread_cond_wait(&cond_add, &grep_mutex);
 122         }
 123
 124         if (todo_start == todo_end && all_work_added) {
 125                 ret = NULL;
 126         } else {
 127                 ret = &todo[todo_start];
 128                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 129         }
 130         grep_unlock();
 131         return ret;
 132 }
 133
 134 static void work_done(struct work_item *w)
 135 {
 136         int old_done;
 137
 138         grep_lock();
 139         w->done = 1;
 140         old_done = todo_done;
 141         for(; todo[todo_done].done && todo_done != todo_start;
 142             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 143                 w = &todo[todo_done];
 144                 if (w->out.len) {
 145                         const char *p = w->out.buf;
 146                         size_t len = w->out.len;
 147
 148                         /* Skip the leading hunk mark of the first file. */
 149                         if (skip_first_line) {
 150                                 while (len) {
 151                                         len--;
 152                                         if (*p++ == '\n')
 153                                                 break;
 154                                 }
 155                                 skip_first_line = 0;
 156                         }
 157
 158                         write_or_die(1, p, len);
 159                 }
 160                 grep_source_clear(&w->source);
 161         }
 162
 163         if (old_done != todo_done)
 164                 pthread_cond_signal(&cond_write);
 165
 166         if (all_work_added && todo_done == todo_end)
 167                 pthread_cond_signal(&cond_result);
 168
 169         grep_unlock();
 170 }
 171
 172 static void *run(void *arg)
 173 {
 174         int hit = 0;
 175         struct grep_opt *opt = arg;
 176
 177         while (1) {
 178                 struct work_item *w = get_work();
 179                 if (!w)
 180                         break;
 181
 182                 opt->output_priv = w;
 183                 hit |= grep_source(opt, &w->source);
 184                 grep_source_clear_data(&w->source);
 185                 work_done(w);
 186         }
 187         free_grep_patterns(arg);
 188         free(arg);
 189
 190         return (void*) (intptr_t) hit;
 191 }
 192
 193 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 194 {
 195         struct work_item *w = opt->output_priv;
 196         strbuf_add(&w->out, buf, size);
 197 }
 198
 199 static void start_threads(struct grep_opt *opt)
 200 {
 201         int i;
 202
 203         pthread_mutex_init(&grep_mutex, NULL);
 204         pthread_mutex_init(&grep_read_mutex, NULL);
 205         pthread_mutex_init(&grep_attr_mutex, NULL);
 206         pthread_cond_init(&cond_add, NULL);
 207         pthread_cond_init(&cond_write, NULL);
 208         pthread_cond_init(&cond_result, NULL);
 209         grep_use_locks = 1;
 210
 211         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 212                 strbuf_init(&todo[i].out, 0);
 213         }
 214
 215         threads = xcalloc(num_threads, sizeof(*threads));
 216         for (i = 0; i < num_threads; i++) {
 217                 int err;
 218                 struct grep_opt *o = grep_opt_dup(opt);
 219                 o->output = strbuf_out;
 220                 if (i)
 221                         o->debug = 0;
 222                 compile_grep_patterns(o);
 223                 err = pthread_create(&threads[i], NULL, run, o);
 224
 225                 if (err)
 226                         die(_("grep: failed to create thread: %s"),
 227                             strerror(err));
 228         }
 229 }
 230
 231 static int wait_all(void)
 232 {
 233         int hit = 0;
 234         int i;
 235
 236         grep_lock();
 237         all_work_added = 1;
 238
 239         /* Wait until all work is done. */
 240         while (todo_done != todo_end)
 241                 pthread_cond_wait(&cond_result, &grep_mutex);
 242
 243         /* Wake up all the consumer threads so they can see that there
 244          * is no more work to do.
 245          */
 246         pthread_cond_broadcast(&cond_add);
 247         grep_unlock();
 248
 249         for (i = 0; i < num_threads; i++) {
 250                 void *h;
 251                 pthread_join(threads[i], &h);
 252                 hit |= (int) (intptr_t) h;
 253         }
 254
 255         free(threads);
 256
 257         pthread_mutex_destroy(&grep_mutex);
 258         pthread_mutex_destroy(&grep_read_mutex);
 259         pthread_mutex_destroy(&grep_attr_mutex);
 260         pthread_cond_destroy(&cond_add);
 261         pthread_cond_destroy(&cond_write);
 262         pthread_cond_destroy(&cond_result);
 263         grep_use_locks = 0;
 264
 265         return hit;
 266 }
 267 #else /* !NO_PTHREADS */
 268
 269 static int wait_all(void)
 270 {
 271         return 0;
 272 }
 273 #endif
 274
 275 static int grep_cmd_config(const char *var, const char *value, void *cb)
 276 {
 277         int st = grep_config(var, value, cb);
 278         if (git_color_default_config(var, value, cb) < 0)
 279                 st = -1;
 280
 281         if (!strcmp(var, "grep.threads")) {
 282                 num_threads = git_config_int(var, value);
 283                 if (num_threads < 0)
 284                         die(_("invalid number of threads specified (%d) for %s"),
 285                             num_threads, var);
 286 #ifdef NO_PTHREADS
 287                 else if (num_threads && num_threads != 1) {
 288                         /*
 289                          * TRANSLATORS: %s is the configuration
 290                          * variable for tweaking threads, currently
 291                          * grep.threads
 292                          */
 293                         warning(_("no threads support, ignoring %s"), var);
 294                         num_threads = 0;
 295                 }
 296 #endif
 297         }
 298
 299         if (!strcmp(var, "submodule.recurse"))
 300                 recurse_submodules = git_config_bool(var, value);
 301
 302         return st;
 303 }
 304
 305 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
 306 {
 307         void *data;
 308
 309         grep_read_lock();
 310         data = read_object_file(oid, type, size);
 311         grep_read_unlock();
 312         return data;
 313 }
 314
 315 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
 316                      const char *filename, int tree_name_len,
 317                      const char *path)
 318 {
 319         struct strbuf pathbuf = STRBUF_INIT;
 320         struct grep_source gs;
 321
 322         if (opt->relative && opt->prefix_length) {
 323                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
 324                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 325         } else {
 326                 strbuf_addstr(&pathbuf, filename);
 327         }
 328
 329         grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
 330         strbuf_release(&pathbuf);
 331
 332 #ifndef NO_PTHREADS
 333         if (num_threads) {
 334                 /*
 335                  * add_work() copies gs and thus assumes ownership of
 336                  * its fields, so do not call grep_source_clear()
 337                  */
 338                 add_work(opt, &gs);
 339                 return 0;
 340         } else
 341 #endif
 342         {
 343                 int hit;
 344
 345                 hit = grep_source(opt, &gs);
 346
 347                 grep_source_clear(&gs);
 348                 return hit;
 349         }
 350 }
 351
 352 static int grep_file(struct grep_opt *opt, const char *filename)
 353 {
 354         struct strbuf buf = STRBUF_INIT;
 355         struct grep_source gs;
 356
 357         if (opt->relative && opt->prefix_length)
 358                 quote_path_relative(filename, opt->prefix, &buf);
 359         else
 360                 strbuf_addstr(&buf, filename);
 361
 362         grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 363         strbuf_release(&buf);
 364
 365 #ifndef NO_PTHREADS
 366         if (num_threads) {
 367                 /*
 368                  * add_work() copies gs and thus assumes ownership of
 369                  * its fields, so do not call grep_source_clear()
 370                  */
 371                 add_work(opt, &gs);
 372                 return 0;
 373         } else
 374 #endif
 375         {
 376                 int hit;
 377
 378                 hit = grep_source(opt, &gs);
 379
 380                 grep_source_clear(&gs);
 381                 return hit;
 382         }
 383 }
 384
 385 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 386 {
 387         struct string_list *path_list = opt->output_priv;
 388
 389         if (len == 1 && *(const char *)data == '\0')
 390                 return;
 391         string_list_append(path_list, xstrndup(data, len));
 392 }
 393
 394 static void run_pager(struct grep_opt *opt, const char *prefix)
 395 {
 396         struct string_list *path_list = opt->output_priv;
 397         struct child_process child = CHILD_PROCESS_INIT;
 398         int i, status;
 399
 400         for (i = 0; i < path_list->nr; i++)
 401                 argv_array_push(&child.args, path_list->items[i].string);
 402         child.dir = prefix;
 403         child.use_shell = 1;
 404
 405         status = run_command(&child);
 406         if (status)
 407                 exit(status);
 408 }
 409
 410 static int grep_cache(struct grep_opt *opt, struct repository *repo,
 411                       const struct pathspec *pathspec, int cached);
 412 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 413                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 414                      int check_attr, struct repository *repo);
 415
 416 static int grep_submodule(struct grep_opt *opt, struct repository *superproject,
 417                           const struct pathspec *pathspec,
 418                           const struct object_id *oid,
 419                           const char *filename, const char *path)
 420 {
 421         struct repository submodule;
 422         int hit;
 423
 424         if (!is_submodule_active(superproject, path))
 425                 return 0;
 426
 427         if (repo_submodule_init(&submodule, superproject, path))
 428                 return 0;
 429
 430         repo_read_gitmodules(&submodule);
 431
 432         /*
 433          * NEEDSWORK: This adds the submodule's object directory to the list of
 434          * alternates for the single in-memory object store.  This has some bad
 435          * consequences for memory (processed objects will never be freed) and
 436          * performance (this increases the number of pack files git has to pay
 437          * attention to, to the sum of the number of pack files in all the
 438          * repositories processed so far).  This can be removed once the object
 439          * store is no longer global and instead is a member of the repository
 440          * object.
 441          */
 442         grep_read_lock();
 443         add_to_alternates_memory(submodule.objects->objectdir);
 444         grep_read_unlock();
 445
 446         if (oid) {
 447                 struct object *object;
 448                 struct tree_desc tree;
 449                 void *data;
 450                 unsigned long size;
 451                 struct strbuf base = STRBUF_INIT;
 452
 453                 object = parse_object_or_die(oid, oid_to_hex(oid));
 454
 455                 grep_read_lock();
 456                 data = read_object_with_reference(&object->oid, tree_type,
 457                                                   &size, NULL);
 458                 grep_read_unlock();
 459
 460                 if (!data)
 461                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
 462
 463                 strbuf_addstr(&base, filename);
 464                 strbuf_addch(&base, '/');
 465
 466                 init_tree_desc(&tree, data, size);
 467                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 468                                 object->type == OBJ_COMMIT, &submodule);
 469                 strbuf_release(&base);
 470                 free(data);
 471         } else {
 472                 hit = grep_cache(opt, &submodule, pathspec, 1);
 473         }
 474
 475         repo_clear(&submodule);
 476         return hit;
 477 }
 478
 479 static int grep_cache(struct grep_opt *opt, struct repository *repo,
 480                       const struct pathspec *pathspec, int cached)
 481 {
 482         int hit = 0;
 483         int nr;
 484         struct strbuf name = STRBUF_INIT;
 485         int name_base_len = 0;
 486         if (repo->submodule_prefix) {
 487                 name_base_len = strlen(repo->submodule_prefix);
 488                 strbuf_addstr(&name, repo->submodule_prefix);
 489         }
 490
 491         repo_read_index(repo);
 492
 493         for (nr = 0; nr < repo->index->cache_nr; nr++) {
 494                 const struct cache_entry *ce = repo->index->cache[nr];
 495                 strbuf_setlen(&name, name_base_len);
 496                 strbuf_addstr(&name, ce->name);
 497
 498                 if (S_ISREG(ce->ce_mode) &&
 499                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
 500                                    S_ISDIR(ce->ce_mode) ||
 501                                    S_ISGITLINK(ce->ce_mode))) {
 502                         /*
 503                          * If CE_VALID is on, we assume worktree file and its
 504                          * cache entry are identical, even if worktree file has
 505                          * been modified, so use cache version instead
 506                          */
 507                         if (cached || (ce->ce_flags & CE_VALID) ||
 508                             ce_skip_worktree(ce)) {
 509                                 if (ce_stage(ce) || ce_intent_to_add(ce))
 510                                         continue;
 511                                 hit |= grep_oid(opt, &ce->oid, name.buf,
 512                                                  0, name.buf);
 513                         } else {
 514                                 hit |= grep_file(opt, name.buf);
 515                         }
 516                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 517                            submodule_path_match(pathspec, name.buf, NULL)) {
 518                         hit |= grep_submodule(opt, repo, pathspec, NULL, ce->name, ce->name);
 519                 } else {
 520                         continue;
 521                 }
 522
 523                 if (ce_stage(ce)) {
 524                         do {
 525                                 nr++;
 526                         } while (nr < repo->index->cache_nr &&
 527                                  !strcmp(ce->name, repo->index->cache[nr]->name));
 528                         nr--; /* compensate for loop control */
 529                 }
 530                 if (hit && opt->status_only)
 531                         break;
 532         }
 533
 534         strbuf_release(&name);
 535         return hit;
 536 }
 537
 538 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 539                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 540                      int check_attr, struct repository *repo)
 541 {
 542         int hit = 0;
 543         enum interesting match = entry_not_interesting;
 544         struct name_entry entry;
 545         int old_baselen = base->len;
 546         struct strbuf name = STRBUF_INIT;
 547         int name_base_len = 0;
 548         if (repo->submodule_prefix) {
 549                 strbuf_addstr(&name, repo->submodule_prefix);
 550                 name_base_len = name.len;
 551         }
 552
 553         while (tree_entry(tree, &entry)) {
 554                 int te_len = tree_entry_len(&entry);
 555
 556                 if (match != all_entries_interesting) {
 557                         strbuf_addstr(&name, base->buf + tn_len);
 558                         match = tree_entry_interesting(&entry, &name,
 559                                                        0, pathspec);
 560                         strbuf_setlen(&name, name_base_len);
 561
 562                         if (match == all_entries_not_interesting)
 563                                 break;
 564                         if (match == entry_not_interesting)
 565                                 continue;
 566                 }
 567
 568                 strbuf_add(base, entry.path, te_len);
 569
 570                 if (S_ISREG(entry.mode)) {
 571                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
 572                                          check_attr ? base->buf + tn_len : NULL);
 573                 } else if (S_ISDIR(entry.mode)) {
 574                         enum object_type type;
 575                         struct tree_desc sub;
 576                         void *data;
 577                         unsigned long size;
 578
 579                         data = lock_and_read_oid_file(entry.oid, &type, &size);
 580                         if (!data)
 581                                 die(_("unable to read tree (%s)"),
 582                                     oid_to_hex(entry.oid));
 583
 584                         strbuf_addch(base, '/');
 585                         init_tree_desc(&sub, data, size);
 586                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 587                                          check_attr, repo);
 588                         free(data);
 589                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 590                         hit |= grep_submodule(opt, repo, pathspec, entry.oid,
 591                                               base->buf, base->buf + tn_len);
 592                 }
 593
 594                 strbuf_setlen(base, old_baselen);
 595
 596                 if (hit && opt->status_only)
 597                         break;
 598         }
 599
 600         strbuf_release(&name);
 601         return hit;
 602 }
 603
 604 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 605                        struct object *obj, const char *name, const char *path)
 606 {
 607         if (obj->type == OBJ_BLOB)
 608                 return grep_oid(opt, &obj->oid, name, 0, path);
 609         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 610                 struct tree_desc tree;
 611                 void *data;
 612                 unsigned long size;
 613                 struct strbuf base;
 614                 int hit, len;
 615
 616                 grep_read_lock();
 617                 data = read_object_with_reference(&obj->oid, tree_type,
 618                                                   &size, NULL);
 619                 grep_read_unlock();
 620
 621                 if (!data)
 622                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 623
 624                 len = name ? strlen(name) : 0;
 625                 strbuf_init(&base, PATH_MAX + len + 1);
 626                 if (len) {
 627                         strbuf_add(&base, name, len);
 628                         strbuf_addch(&base, ':');
 629                 }
 630                 init_tree_desc(&tree, data, size);
 631                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 632                                 obj->type == OBJ_COMMIT, the_repository);
 633                 strbuf_release(&base);
 634                 free(data);
 635                 return hit;
 636         }
 637         die(_("unable to grep from object of type %s"), type_name(obj->type));
 638 }
 639
 640 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 641                         const struct object_array *list)
 642 {
 643         unsigned int i;
 644         int hit = 0;
 645         const unsigned int nr = list->nr;
 646
 647         for (i = 0; i < nr; i++) {
 648                 struct object *real_obj;
 649                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 650
 651                 /* load the gitmodules file for this rev */
 652                 if (recurse_submodules) {
 653                         submodule_free(the_repository);
 654                         gitmodules_config_oid(&real_obj->oid);
 655                 }
 656                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
 657                                 list->objects[i].path)) {
 658                         hit = 1;
 659                         if (opt->status_only)
 660                                 break;
 661                 }
 662         }
 663         return hit;
 664 }
 665
 666 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 667                           int exc_std, int use_index)
 668 {
 669         struct dir_struct dir;
 670         int i, hit = 0;
 671
 672         memset(&dir, 0, sizeof(dir));
 673         if (!use_index)
 674                 dir.flags |= DIR_NO_GITLINKS;
 675         if (exc_std)
 676                 setup_standard_excludes(&dir);
 677
 678         fill_directory(&dir, &the_index, pathspec);
 679         for (i = 0; i < dir.nr; i++) {
 680                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
 681                         continue;
 682                 hit |= grep_file(opt, dir.entries[i]->name);
 683                 if (hit && opt->status_only)
 684                         break;
 685         }
 686         return hit;
 687 }
 688
 689 static int context_callback(const struct option *opt, const char *arg,
 690                             int unset)
 691 {
 692         struct grep_opt *grep_opt = opt->value;
 693         int value;
 694         const char *endp;
 695
 696         if (unset) {
 697                 grep_opt->pre_context = grep_opt->post_context = 0;
 698                 return 0;
 699         }
 700         value = strtol(arg, (char **)&endp, 10);
 701         if (*endp) {
 702                 return error(_("switch `%c' expects a numerical value"),
 703                              opt->short_name);
 704         }
 705         grep_opt->pre_context = grep_opt->post_context = value;
 706         return 0;
 707 }
 708
 709 static int file_callback(const struct option *opt, const char *arg, int unset)
 710 {
 711         struct grep_opt *grep_opt = opt->value;
 712         int from_stdin = !strcmp(arg, "-");
 713         FILE *patterns;
 714         int lno = 0;
 715         struct strbuf sb = STRBUF_INIT;
 716
 717         patterns = from_stdin ? stdin : fopen(arg, "r");
 718         if (!patterns)
 719                 die_errno(_("cannot open '%s'"), arg);
 720         while (strbuf_getline(&sb, patterns) == 0) {
 721                 /* ignore empty line like grep does */
 722                 if (sb.len == 0)
 723                         continue;
 724
 725                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 726                                 GREP_PATTERN);
 727         }
 728         if (!from_stdin)
 729                 fclose(patterns);
 730         strbuf_release(&sb);
 731         return 0;
 732 }
 733
 734 static int not_callback(const struct option *opt, const char *arg, int unset)
 735 {
 736         struct grep_opt *grep_opt = opt->value;
 737         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 738         return 0;
 739 }
 740
 741 static int and_callback(const struct option *opt, const char *arg, int unset)
 742 {
 743         struct grep_opt *grep_opt = opt->value;
 744         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 745         return 0;
 746 }
 747
 748 static int open_callback(const struct option *opt, const char *arg, int unset)
 749 {
 750         struct grep_opt *grep_opt = opt->value;
 751         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 752         return 0;
 753 }
 754
 755 static int close_callback(const struct option *opt, const char *arg, int unset)
 756 {
 757         struct grep_opt *grep_opt = opt->value;
 758         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 759         return 0;
 760 }
 761
 762 static int pattern_callback(const struct option *opt, const char *arg,
 763                             int unset)
 764 {
 765         struct grep_opt *grep_opt = opt->value;
 766         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 767         return 0;
 768 }
 769
 770 int cmd_grep(int argc, const char **argv, const char *prefix)
 771 {
 772         int hit = 0;
 773         int cached = 0, untracked = 0, opt_exclude = -1;
 774         int seen_dashdash = 0;
 775         int external_grep_allowed__ignored;
 776         const char *show_in_pager = NULL, *default_pager = "dummy";
 777         struct grep_opt opt;
 778         struct object_array list = OBJECT_ARRAY_INIT;
 779         struct pathspec pathspec;
 780         struct string_list path_list = STRING_LIST_INIT_NODUP;
 781         int i;
 782         int dummy;
 783         int use_index = 1;
 784         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 785         int allow_revs;
 786
 787         struct option options[] = {
 788                 OPT_BOOL(0, "cached", &cached,
 789                         N_("search in index instead of in the work tree")),
 790                 OPT_NEGBIT(0, "no-index", &use_index,
 791                          N_("find in contents not managed by git"), 1),
 792                 OPT_BOOL(0, "untracked", &untracked,
 793                         N_("search in both tracked and untracked files")),
 794                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 795                             N_("ignore files specified via '.gitignore'"), 1),
 796                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 797                          N_("recursively search in each submodule")),
 798                 OPT_GROUP(""),
 799                 OPT_BOOL('v', "invert-match", &opt.invert,
 800                         N_("show non-matching lines")),
 801                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 802                         N_("case insensitive matching")),
 803                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 804                         N_("match patterns only at word boundaries")),
 805                 OPT_SET_INT('a', "text", &opt.binary,
 806                         N_("process binary files as text"), GREP_BINARY_TEXT),
 807                 OPT_SET_INT('I', NULL, &opt.binary,
 808                         N_("don't match patterns in binary files"),
 809                         GREP_BINARY_NOMATCH),
 810                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
 811                          N_("process binary files with textconv filters")),
 812                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
 813                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
 814                         NULL, 1 },
 815                 OPT_GROUP(""),
 816                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
 817                             N_("use extended POSIX regular expressions"),
 818                             GREP_PATTERN_TYPE_ERE),
 819                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
 820                             N_("use basic POSIX regular expressions (default)"),
 821                             GREP_PATTERN_TYPE_BRE),
 822                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
 823                             N_("interpret patterns as fixed strings"),
 824                             GREP_PATTERN_TYPE_FIXED),
 825                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
 826                             N_("use Perl-compatible regular expressions"),
 827                             GREP_PATTERN_TYPE_PCRE),
 828                 OPT_GROUP(""),
 829                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
 830                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
 831                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
 832                 OPT_NEGBIT(0, "full-name", &opt.relative,
 833                         N_("show filenames relative to top directory"), 1),
 834                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
 835                         N_("show only filenames instead of matching lines")),
 836                 OPT_BOOL(0, "name-only", &opt.name_only,
 837                         N_("synonym for --files-with-matches")),
 838                 OPT_BOOL('L', "files-without-match",
 839                         &opt.unmatch_name_only,
 840                         N_("show only the names of files without match")),
 841                 OPT_BOOL_F('z', "null", &opt.null_following_name,
 842                            N_("print NUL after filenames"),
 843                            PARSE_OPT_NOCOMPLETE),
 844                 OPT_BOOL('c', "count", &opt.count,
 845                         N_("show the number of matches instead of matching lines")),
 846                 OPT__COLOR(&opt.color, N_("highlight matches")),
 847                 OPT_BOOL(0, "break", &opt.file_break,
 848                         N_("print empty line between matches from different files")),
 849                 OPT_BOOL(0, "heading", &opt.heading,
 850                         N_("show filename only once above matches from same file")),
 851                 OPT_GROUP(""),
 852                 OPT_CALLBACK('C', "context", &opt, N_("n"),
 853                         N_("show <n> context lines before and after matches"),
 854                         context_callback),
 855                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 856                         N_("show <n> context lines before matches")),
 857                 OPT_INTEGER('A', "after-context", &opt.post_context,
 858                         N_("show <n> context lines after matches")),
 859                 OPT_INTEGER(0, "threads", &num_threads,
 860                         N_("use <n> worker threads")),
 861                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
 862                         context_callback),
 863                 OPT_BOOL('p', "show-function", &opt.funcname,
 864                         N_("show a line with the function name before matches")),
 865                 OPT_BOOL('W', "function-context", &opt.funcbody,
 866                         N_("show the surrounding function")),
 867                 OPT_GROUP(""),
 868                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
 869                         N_("read patterns from file"), file_callback),
 870                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
 871                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
 872                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 873                   N_("combine patterns specified with -e"),
 874                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 875                 OPT_BOOL(0, "or", &dummy, ""),
 876                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 877                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 878                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 879                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 880                   open_callback },
 881                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 882                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 883                   close_callback },
 884                 OPT__QUIET(&opt.status_only,
 885                            N_("indicate hit with exit status without output")),
 886                 OPT_BOOL(0, "all-match", &opt.all_match,
 887                         N_("show only matches from files that match all patterns")),
 888                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
 889                   N_("show parse tree for grep expression"),
 890                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
 891                 OPT_GROUP(""),
 892                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 893                         N_("pager"), N_("show matching files in the pager"),
 894                         PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
 895                         NULL, (intptr_t)default_pager },
 896                 OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
 897                            N_("allow calling of grep(1) (ignored by this build)"),
 898                            PARSE_OPT_NOCOMPLETE),
 899                 OPT_END()
 900         };
 901
 902         init_grep_defaults();
 903         git_config(grep_cmd_config, NULL);
 904         grep_init(&opt, prefix);
 905
 906         /*
 907          * If there is no -- then the paths must exist in the working
 908          * tree.  If there is no explicit pattern specified with -e or
 909          * -f, we take the first unrecognized non option to be the
 910          * pattern, but then what follows it must be zero or more
 911          * valid refs up to the -- (if exists), and then existing
 912          * paths.  If there is an explicit pattern, then the first
 913          * unrecognized non option is the beginning of the refs list
 914          * that continues up to the -- (if exists), and then paths.
 915          */
 916         argc = parse_options(argc, argv, prefix, options, grep_usage,
 917                              PARSE_OPT_KEEP_DASHDASH |
 918                              PARSE_OPT_STOP_AT_NON_OPTION);
 919         grep_commit_pattern_type(pattern_type_arg, &opt);
 920
 921         if (use_index && !startup_info->have_repository) {
 922                 int fallback = 0;
 923                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
 924                 if (fallback)
 925                         use_index = 0;
 926                 else
 927                         /* die the same way as if we did it at the beginning */
 928                         setup_git_directory();
 929         }
 930
 931         /*
 932          * skip a -- separator; we know it cannot be
 933          * separating revisions from pathnames if
 934          * we haven't even had any patterns yet
 935          */
 936         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 937                 argv++;
 938                 argc--;
 939         }
 940
 941         /* First unrecognized non-option token */
 942         if (argc > 0 && !opt.pattern_list) {
 943                 append_grep_pattern(&opt, argv[0], "command line", 0,
 944                                     GREP_PATTERN);
 945                 argv++;
 946                 argc--;
 947         }
 948
 949         if (show_in_pager == default_pager)
 950                 show_in_pager = git_pager(1);
 951         if (show_in_pager) {
 952                 opt.color = 0;
 953                 opt.name_only = 1;
 954                 opt.null_following_name = 1;
 955                 opt.output_priv = &path_list;
 956                 opt.output = append_path;
 957                 string_list_append(&path_list, show_in_pager);
 958         }
 959
 960         if (!opt.pattern_list)
 961                 die(_("no pattern given."));
 962
 963         /*
 964          * We have to find "--" in a separate pass, because its presence
 965          * influences how we will parse arguments that come before it.
 966          */
 967         for (i = 0; i < argc; i++) {
 968                 if (!strcmp(argv[i], "--")) {
 969                         seen_dashdash = 1;
 970                         break;
 971                 }
 972         }
 973
 974         /*
 975          * Resolve any rev arguments. If we have a dashdash, then everything up
 976          * to it must resolve as a rev. If not, then we stop at the first
 977          * non-rev and assume everything else is a path.
 978          */
 979         allow_revs = use_index && !untracked;
 980         for (i = 0; i < argc; i++) {
 981                 const char *arg = argv[i];
 982                 struct object_id oid;
 983                 struct object_context oc;
 984                 struct object *object;
 985
 986                 if (!strcmp(arg, "--")) {
 987                         i++;
 988                         break;
 989                 }
 990
 991                 if (!allow_revs) {
 992                         if (seen_dashdash)
 993                                 die(_("--no-index or --untracked cannot be used with revs"));
 994                         break;
 995                 }
 996
 997                 if (get_oid_with_context(arg, GET_OID_RECORD_PATH,
 998                                          &oid, &oc)) {
 999                         if (seen_dashdash)
1000                                 die(_("unable to resolve revision: %s"), arg);
1001                         break;
1002                 }
1003
1004                 object = parse_object_or_die(&oid, arg);
1005                 if (!seen_dashdash)
1006                         verify_non_filename(prefix, arg);
1007                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1008                 free(oc.path);
1009         }
1010
1011         /*
1012          * Anything left over is presumed to be a path. But in the non-dashdash
1013          * "do what I mean" case, we verify and complain when that isn't true.
1014          */
1015         if (!seen_dashdash) {
1016                 int j;
1017                 for (j = i; j < argc; j++)
1018                         verify_filename(prefix, argv[j], j == i && allow_revs);
1019         }
1020
1021         parse_pathspec(&pathspec, 0,
1022                        PATHSPEC_PREFER_CWD |
1023                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1024                        prefix, argv + i);
1025         pathspec.max_depth = opt.max_depth;
1026         pathspec.recursive = 1;
1027         pathspec.recurse_submodules = !!recurse_submodules;
1028
1029 #ifndef NO_PTHREADS
1030         if (list.nr || cached || show_in_pager)
1031                 num_threads = 0;
1032         else if (num_threads == 0)
1033                 num_threads = GREP_NUM_THREADS_DEFAULT;
1034         else if (num_threads < 0)
1035                 die(_("invalid number of threads specified (%d)"), num_threads);
1036         if (num_threads == 1)
1037                 num_threads = 0;
1038 #else
1039         if (num_threads)
1040                 warning(_("no threads support, ignoring --threads"));
1041         num_threads = 0;
1042 #endif
1043
1044         if (!num_threads)
1045                 /*
1046                  * The compiled patterns on the main path are only
1047                  * used when not using threading. Otherwise
1048                  * start_threads() below calls compile_grep_patterns()
1049                  * for each thread.
1050                  */
1051                 compile_grep_patterns(&opt);
1052
1053 #ifndef NO_PTHREADS
1054         if (num_threads) {
1055                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1056                     && (opt.pre_context || opt.post_context ||
1057                         opt.file_break || opt.funcbody))
1058                         skip_first_line = 1;
1059                 start_threads(&opt);
1060         }
1061 #endif
1062
1063         if (show_in_pager && (cached || list.nr))
1064                 die(_("--open-files-in-pager only works on the worktree"));
1065
1066         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1067                 const char *pager = path_list.items[0].string;
1068                 int len = strlen(pager);
1069
1070                 if (len > 4 && is_dir_sep(pager[len - 5]))
1071                         pager += len - 4;
1072
1073                 if (opt.ignore_case && !strcmp("less", pager))
1074                         string_list_append(&path_list, "-I");
1075
1076                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1077                         struct strbuf buf = STRBUF_INIT;
1078                         strbuf_addf(&buf, "+/%s%s",
1079                                         strcmp("less", pager) ? "" : "*",
1080                                         opt.pattern_list->pattern);
1081                         string_list_append(&path_list, buf.buf);
1082                         strbuf_detach(&buf, NULL);
1083                 }
1084         }
1085
1086         if (recurse_submodules && (!use_index || untracked))
1087                 die(_("option not supported with --recurse-submodules."));
1088
1089         if (!show_in_pager && !opt.status_only)
1090                 setup_pager();
1091
1092         if (!use_index && (untracked || cached))
1093                 die(_("--cached or --untracked cannot be used with --no-index."));
1094
1095         if (!use_index || untracked) {
1096                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1097                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1098         } else if (0 <= opt_exclude) {
1099                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1100         } else if (!list.nr) {
1101                 if (!cached)
1102                         setup_work_tree();
1103
1104                 hit = grep_cache(&opt, the_repository, &pathspec, cached);
1105         } else {
1106                 if (cached)
1107                         die(_("both --cached and trees are given."));
1108
1109                 hit = grep_objects(&opt, &pathspec, &list);
1110         }
1111
1112         if (num_threads)
1113                 hit |= wait_all();
1114         if (hit && show_in_pager)
1115                 run_pager(&opt, prefix);
1116         clear_pathspec(&pathspec);
1117         free_grep_patterns(&opt);
1118         return !hit;
1119 }