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