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