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

   1 /*
   2  * Builtin "git grep"
   3  *
   4  * Copyright (c) 2006 Junio C Hamano
   5  */
   6 #define USE_THE_INDEX_COMPATIBILITY_MACROS
   7 #include "cache.h"
   8 #include "repository.h"
   9 #include "config.h"
  10 #include "blob.h"
  11 #include "tree.h"
  12 #include "commit.h"
  13 #include "tag.h"
  14 #include "tree-walk.h"
  15 #include "builtin.h"
  16 #include "parse-options.h"
  17 #include "string-list.h"
  18 #include "run-command.h"
  19 #include "userdiff.h"
  20 #include "grep.h"
  21 #include "quote.h"
  22 #include "dir.h"
  23 #include "pathspec.h"
  24 #include "submodule.h"
  25 #include "submodule-config.h"
  26 #include "object-store.h"
  27
  28 static char const * const grep_usage[] = {
  29         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  30         NULL
  31 };
  32
  33 static int recurse_submodules;
  34
  35 #define GREP_NUM_THREADS_DEFAULT 8
  36 static int num_threads;
  37
  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         pthread_mutex_lock(&grep_mutex);
  74 }
  75
  76 static inline void grep_unlock(void)
  77 {
  78         pthread_mutex_unlock(&grep_mutex);
  79 }
  80
  81 /* Signalled when a new work_item is added to todo. */
  82 static pthread_cond_t cond_add;
  83
  84 /* Signalled when the result from one work_item is written to
  85  * stdout.
  86  */
  87 static pthread_cond_t cond_write;
  88
  89 /* Signalled when we are finished with everything. */
  90 static pthread_cond_t cond_result;
  91
  92 static int skip_first_line;
  93
  94 static void add_work(struct grep_opt *opt, const struct grep_source *gs)
  95 {
  96         grep_lock();
  97
  98         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
  99                 pthread_cond_wait(&cond_write, &grep_mutex);
 100         }
 101
 102         todo[todo_end].source = *gs;
 103         if (opt->binary != GREP_BINARY_TEXT)
 104                 grep_source_load_driver(&todo[todo_end].source,
 105                                         opt->repo->index);
 106         todo[todo_end].done = 0;
 107         strbuf_reset(&todo[todo_end].out);
 108         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 109
 110         pthread_cond_signal(&cond_add);
 111         grep_unlock();
 112 }
 113
 114 static struct work_item *get_work(void)
 115 {
 116         struct work_item *ret;
 117
 118         grep_lock();
 119         while (todo_start == todo_end && !all_work_added) {
 120                 pthread_cond_wait(&cond_add, &grep_mutex);
 121         }
 122
 123         if (todo_start == todo_end && all_work_added) {
 124                 ret = NULL;
 125         } else {
 126                 ret = &todo[todo_start];
 127                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 128         }
 129         grep_unlock();
 130         return ret;
 131 }
 132
 133 static void work_done(struct work_item *w)
 134 {
 135         int old_done;
 136
 137         grep_lock();
 138         w->done = 1;
 139         old_done = todo_done;
 140         for(; todo[todo_done].done && todo_done != todo_start;
 141             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 142                 w = &todo[todo_done];
 143                 if (w->out.len) {
 144                         const char *p = w->out.buf;
 145                         size_t len = w->out.len;
 146
 147                         /* Skip the leading hunk mark of the first file. */
 148                         if (skip_first_line) {
 149                                 while (len) {
 150                                         len--;
 151                                         if (*p++ == '\n')
 152                                                 break;
 153                                 }
 154                                 skip_first_line = 0;
 155                         }
 156
 157                         write_or_die(1, p, len);
 158                 }
 159                 grep_source_clear(&w->source);
 160         }
 161
 162         if (old_done != todo_done)
 163                 pthread_cond_signal(&cond_write);
 164
 165         if (all_work_added && todo_done == todo_end)
 166                 pthread_cond_signal(&cond_result);
 167
 168         grep_unlock();
 169 }
 170
 171 static void *run(void *arg)
 172 {
 173         int hit = 0;
 174         struct grep_opt *opt = arg;
 175
 176         while (1) {
 177                 struct work_item *w = get_work();
 178                 if (!w)
 179                         break;
 180
 181                 opt->output_priv = w;
 182                 hit |= grep_source(opt, &w->source);
 183                 grep_source_clear_data(&w->source);
 184                 work_done(w);
 185         }
 186         free_grep_patterns(arg);
 187         free(arg);
 188
 189         return (void*) (intptr_t) hit;
 190 }
 191
 192 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 193 {
 194         struct work_item *w = opt->output_priv;
 195         strbuf_add(&w->out, buf, size);
 196 }
 197
 198 static void start_threads(struct grep_opt *opt)
 199 {
 200         int i;
 201
 202         pthread_mutex_init(&grep_mutex, NULL);
 203         pthread_mutex_init(&grep_read_mutex, NULL);
 204         pthread_mutex_init(&grep_attr_mutex, NULL);
 205         pthread_cond_init(&cond_add, NULL);
 206         pthread_cond_init(&cond_write, NULL);
 207         pthread_cond_init(&cond_result, NULL);
 208         grep_use_locks = 1;
 209
 210         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 211                 strbuf_init(&todo[i].out, 0);
 212         }
 213
 214         threads = xcalloc(num_threads, sizeof(*threads));
 215         for (i = 0; i < num_threads; i++) {
 216                 int err;
 217                 struct grep_opt *o = grep_opt_dup(opt);
 218                 o->output = strbuf_out;
 219                 if (i)
 220                         o->debug = 0;
 221                 compile_grep_patterns(o);
 222                 err = pthread_create(&threads[i], NULL, run, o);
 223
 224                 if (err)
 225                         die(_("grep: failed to create thread: %s"),
 226                             strerror(err));
 227         }
 228 }
 229
 230 static int wait_all(void)
 231 {
 232         int hit = 0;
 233         int i;
 234
 235         if (!HAVE_THREADS)
 236                 BUG("Never call this function unless you have started threads");
 237
 238         grep_lock();
 239         all_work_added = 1;
 240
 241         /* Wait until all work is done. */
 242         while (todo_done != todo_end)
 243                 pthread_cond_wait(&cond_result, &grep_mutex);
 244
 245         /* Wake up all the consumer threads so they can see that there
 246          * is no more work to do.
 247          */
 248         pthread_cond_broadcast(&cond_add);
 249         grep_unlock();
 250
 251         for (i = 0; i < num_threads; i++) {
 252                 void *h;
 253                 pthread_join(threads[i], &h);
 254                 hit |= (int) (intptr_t) h;
 255         }
 256
 257         free(threads);
 258
 259         pthread_mutex_destroy(&grep_mutex);
 260         pthread_mutex_destroy(&grep_read_mutex);
 261         pthread_mutex_destroy(&grep_attr_mutex);
 262         pthread_cond_destroy(&cond_add);
 263         pthread_cond_destroy(&cond_write);
 264         pthread_cond_destroy(&cond_result);
 265         grep_use_locks = 0;
 266
 267         return hit;
 268 }
 269
 270 static int grep_cmd_config(const char *var, const char *value, void *cb)
 271 {
 272         int st = grep_config(var, value, cb);
 273         if (git_color_default_config(var, value, cb) < 0)
 274                 st = -1;
 275
 276         if (!strcmp(var, "grep.threads")) {
 277                 num_threads = git_config_int(var, value);
 278                 if (num_threads < 0)
 279                         die(_("invalid number of threads specified (%d) for %s"),
 280                             num_threads, var);
 281                 else if (!HAVE_THREADS && num_threads > 1) {
 282                         /*
 283                          * TRANSLATORS: %s is the configuration
 284                          * variable for tweaking threads, currently
 285                          * grep.threads
 286                          */
 287                         warning(_("no threads support, ignoring %s"), var);
 288                         num_threads = 1;
 289                 }
 290         }
 291
 292         if (!strcmp(var, "submodule.recurse"))
 293                 recurse_submodules = git_config_bool(var, value);
 294
 295         return st;
 296 }
 297
 298 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
 299 {
 300         void *data;
 301
 302         grep_read_lock();
 303         data = read_object_file(oid, type, size);
 304         grep_read_unlock();
 305         return data;
 306 }
 307
 308 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
 309                      const char *filename, int tree_name_len,
 310                      const char *path)
 311 {
 312         struct strbuf pathbuf = STRBUF_INIT;
 313         struct grep_source gs;
 314
 315         if (opt->relative && opt->prefix_length) {
 316                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
 317                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 318         } else {
 319                 strbuf_addstr(&pathbuf, filename);
 320         }
 321
 322         grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
 323         strbuf_release(&pathbuf);
 324
 325         if (num_threads > 1) {
 326                 /*
 327                  * add_work() copies gs and thus assumes ownership of
 328                  * its fields, so do not call grep_source_clear()
 329                  */
 330                 add_work(opt, &gs);
 331                 return 0;
 332         } else {
 333                 int hit;
 334
 335                 hit = grep_source(opt, &gs);
 336
 337                 grep_source_clear(&gs);
 338                 return hit;
 339         }
 340 }
 341
 342 static int grep_file(struct grep_opt *opt, const char *filename)
 343 {
 344         struct strbuf buf = STRBUF_INIT;
 345         struct grep_source gs;
 346
 347         if (opt->relative && opt->prefix_length)
 348                 quote_path_relative(filename, opt->prefix, &buf);
 349         else
 350                 strbuf_addstr(&buf, filename);
 351
 352         grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 353         strbuf_release(&buf);
 354
 355         if (num_threads > 1) {
 356                 /*
 357                  * add_work() copies gs and thus assumes ownership of
 358                  * its fields, so do not call grep_source_clear()
 359                  */
 360                 add_work(opt, &gs);
 361                 return 0;
 362         } else {
 363                 int hit;
 364
 365                 hit = grep_source(opt, &gs);
 366
 367                 grep_source_clear(&gs);
 368                 return hit;
 369         }
 370 }
 371
 372 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 373 {
 374         struct string_list *path_list = opt->output_priv;
 375
 376         if (len == 1 && *(const char *)data == '\0')
 377                 return;
 378         string_list_append(path_list, xstrndup(data, len));
 379 }
 380
 381 static void run_pager(struct grep_opt *opt, const char *prefix)
 382 {
 383         struct string_list *path_list = opt->output_priv;
 384         struct child_process child = CHILD_PROCESS_INIT;
 385         int i, status;
 386
 387         for (i = 0; i < path_list->nr; i++)
 388                 argv_array_push(&child.args, path_list->items[i].string);
 389         child.dir = prefix;
 390         child.use_shell = 1;
 391
 392         status = run_command(&child);
 393         if (status)
 394                 exit(status);
 395 }
 396
 397 static int grep_cache(struct grep_opt *opt,
 398                       const struct pathspec *pathspec, int cached);
 399 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 400                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 401                      int check_attr);
 402
 403 static int grep_submodule(struct grep_opt *opt,
 404                           const struct pathspec *pathspec,
 405                           const struct object_id *oid,
 406                           const char *filename, const char *path)
 407 {
 408         struct repository *superproject = opt->repo;
 409         struct repository submodule;
 410         struct grep_opt subopt;
 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(&submodule, superproject, path)) {
 427                 grep_read_unlock();
 428                 return 0;
 429         }
 430
 431         repo_read_gitmodules(&submodule);
 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(submodule.objects->odb->path);
 444         grep_read_unlock();
 445
 446         memcpy(&subopt, opt, sizeof(subopt));
 447         subopt.repo = &submodule;
 448
 449         if (oid) {
 450                 struct object *object;
 451                 struct tree_desc tree;
 452                 void *data;
 453                 unsigned long size;
 454                 struct strbuf base = STRBUF_INIT;
 455
 456                 object = parse_object_or_die(oid, oid_to_hex(oid));
 457
 458                 grep_read_lock();
 459                 data = read_object_with_reference(&object->oid, tree_type,
 460                                                   &size, NULL);
 461                 grep_read_unlock();
 462
 463                 if (!data)
 464                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
 465
 466                 strbuf_addstr(&base, filename);
 467                 strbuf_addch(&base, '/');
 468
 469                 init_tree_desc(&tree, data, size);
 470                 hit = grep_tree(&subopt, pathspec, &tree, &base, base.len,
 471                                 object->type == OBJ_COMMIT);
 472                 strbuf_release(&base);
 473                 free(data);
 474         } else {
 475                 hit = grep_cache(&subopt, pathspec, 1);
 476         }
 477
 478         repo_clear(&submodule);
 479         return hit;
 480 }
 481
 482 static int grep_cache(struct grep_opt *opt,
 483                       const struct pathspec *pathspec, int cached)
 484 {
 485         struct repository *repo = opt->repo;
 486         int hit = 0;
 487         int nr;
 488         struct strbuf name = STRBUF_INIT;
 489         int name_base_len = 0;
 490         if (repo->submodule_prefix) {
 491                 name_base_len = strlen(repo->submodule_prefix);
 492                 strbuf_addstr(&name, repo->submodule_prefix);
 493         }
 494
 495         if (repo_read_index(repo) < 0)
 496                 die(_("index file corrupt"));
 497
 498         for (nr = 0; nr < repo->index->cache_nr; nr++) {
 499                 const struct cache_entry *ce = repo->index->cache[nr];
 500                 strbuf_setlen(&name, name_base_len);
 501                 strbuf_addstr(&name, ce->name);
 502
 503                 if (S_ISREG(ce->ce_mode) &&
 504                     match_pathspec(repo->index, pathspec, name.buf, name.len, 0, NULL,
 505                                    S_ISDIR(ce->ce_mode) ||
 506                                    S_ISGITLINK(ce->ce_mode))) {
 507                         /*
 508                          * If CE_VALID is on, we assume worktree file and its
 509                          * cache entry are identical, even if worktree file has
 510                          * been modified, so use cache version instead
 511                          */
 512                         if (cached || (ce->ce_flags & CE_VALID) ||
 513                             ce_skip_worktree(ce)) {
 514                                 if (ce_stage(ce) || ce_intent_to_add(ce))
 515                                         continue;
 516                                 hit |= grep_oid(opt, &ce->oid, name.buf,
 517                                                  0, name.buf);
 518                         } else {
 519                                 hit |= grep_file(opt, name.buf);
 520                         }
 521                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 522                            submodule_path_match(repo->index, pathspec, name.buf, NULL)) {
 523                         hit |= grep_submodule(opt, pathspec, NULL, ce->name, ce->name);
 524                 } else {
 525                         continue;
 526                 }
 527
 528                 if (ce_stage(ce)) {
 529                         do {
 530                                 nr++;
 531                         } while (nr < repo->index->cache_nr &&
 532                                  !strcmp(ce->name, repo->index->cache[nr]->name));
 533                         nr--; /* compensate for loop control */
 534                 }
 535                 if (hit && opt->status_only)
 536                         break;
 537         }
 538
 539         strbuf_release(&name);
 540         return hit;
 541 }
 542
 543 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 544                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 545                      int check_attr)
 546 {
 547         struct repository *repo = opt->repo;
 548         int hit = 0;
 549         enum interesting match = entry_not_interesting;
 550         struct name_entry entry;
 551         int old_baselen = base->len;
 552         struct strbuf name = STRBUF_INIT;
 553         int name_base_len = 0;
 554         if (repo->submodule_prefix) {
 555                 strbuf_addstr(&name, repo->submodule_prefix);
 556                 name_base_len = name.len;
 557         }
 558
 559         while (tree_entry(tree, &entry)) {
 560                 int te_len = tree_entry_len(&entry);
 561
 562                 if (match != all_entries_interesting) {
 563                         strbuf_addstr(&name, base->buf + tn_len);
 564                         match = tree_entry_interesting(&entry, &name,
 565                                                        0, pathspec);
 566                         strbuf_setlen(&name, name_base_len);
 567
 568                         if (match == all_entries_not_interesting)
 569                                 break;
 570                         if (match == entry_not_interesting)
 571                                 continue;
 572                 }
 573
 574                 strbuf_add(base, entry.path, te_len);
 575
 576                 if (S_ISREG(entry.mode)) {
 577                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
 578                                          check_attr ? base->buf + tn_len : NULL);
 579                 } else if (S_ISDIR(entry.mode)) {
 580                         enum object_type type;
 581                         struct tree_desc sub;
 582                         void *data;
 583                         unsigned long size;
 584
 585                         data = lock_and_read_oid_file(entry.oid, &type, &size);
 586                         if (!data)
 587                                 die(_("unable to read tree (%s)"),
 588                                     oid_to_hex(entry.oid));
 589
 590                         strbuf_addch(base, '/');
 591                         init_tree_desc(&sub, data, size);
 592                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 593                                          check_attr);
 594                         free(data);
 595                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 596                         hit |= grep_submodule(opt, pathspec, entry.oid,
 597                                               base->buf, base->buf + tn_len);
 598                 }
 599
 600                 strbuf_setlen(base, old_baselen);
 601
 602                 if (hit && opt->status_only)
 603                         break;
 604         }
 605
 606         strbuf_release(&name);
 607         return hit;
 608 }
 609
 610 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 611                        struct object *obj, const char *name, const char *path)
 612 {
 613         if (obj->type == OBJ_BLOB)
 614                 return grep_oid(opt, &obj->oid, name, 0, path);
 615         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 616                 struct tree_desc tree;
 617                 void *data;
 618                 unsigned long size;
 619                 struct strbuf base;
 620                 int hit, len;
 621
 622                 grep_read_lock();
 623                 data = read_object_with_reference(&obj->oid, tree_type,
 624                                                   &size, NULL);
 625                 grep_read_unlock();
 626
 627                 if (!data)
 628                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 629
 630                 len = name ? strlen(name) : 0;
 631                 strbuf_init(&base, PATH_MAX + len + 1);
 632                 if (len) {
 633                         strbuf_add(&base, name, len);
 634                         strbuf_addch(&base, ':');
 635                 }
 636                 init_tree_desc(&tree, data, size);
 637                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 638                                 obj->type == OBJ_COMMIT);
 639                 strbuf_release(&base);
 640                 free(data);
 641                 return hit;
 642         }
 643         die(_("unable to grep from object of type %s"), type_name(obj->type));
 644 }
 645
 646 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 647                         const struct object_array *list)
 648 {
 649         unsigned int i;
 650         int hit = 0;
 651         const unsigned int nr = list->nr;
 652
 653         for (i = 0; i < nr; i++) {
 654                 struct object *real_obj;
 655                 real_obj = deref_tag(opt->repo, list->objects[i].item,
 656                                      NULL, 0);
 657
 658                 /* load the gitmodules file for this rev */
 659                 if (recurse_submodules) {
 660                         submodule_free(opt->repo);
 661                         gitmodules_config_oid(&real_obj->oid);
 662                 }
 663                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
 664                                 list->objects[i].path)) {
 665                         hit = 1;
 666                         if (opt->status_only)
 667                                 break;
 668                 }
 669         }
 670         return hit;
 671 }
 672
 673 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 674                           int exc_std, int use_index)
 675 {
 676         struct dir_struct dir;
 677         int i, hit = 0;
 678
 679         memset(&dir, 0, sizeof(dir));
 680         if (!use_index)
 681                 dir.flags |= DIR_NO_GITLINKS;
 682         if (exc_std)
 683                 setup_standard_excludes(&dir);
 684
 685         fill_directory(&dir, opt->repo->index, pathspec);
 686         for (i = 0; i < dir.nr; i++) {
 687                 if (!dir_path_match(opt->repo->index, dir.entries[i], pathspec, 0, NULL))
 688                         continue;
 689                 hit |= grep_file(opt, dir.entries[i]->name);
 690                 if (hit && opt->status_only)
 691                         break;
 692         }
 693         return hit;
 694 }
 695
 696 static int context_callback(const struct option *opt, const char *arg,
 697                             int unset)
 698 {
 699         struct grep_opt *grep_opt = opt->value;
 700         int value;
 701         const char *endp;
 702
 703         if (unset) {
 704                 grep_opt->pre_context = grep_opt->post_context = 0;
 705                 return 0;
 706         }
 707         value = strtol(arg, (char **)&endp, 10);
 708         if (*endp) {
 709                 return error(_("switch `%c' expects a numerical value"),
 710                              opt->short_name);
 711         }
 712         grep_opt->pre_context = grep_opt->post_context = value;
 713         return 0;
 714 }
 715
 716 static int file_callback(const struct option *opt, const char *arg, int unset)
 717 {
 718         struct grep_opt *grep_opt = opt->value;
 719         int from_stdin;
 720         FILE *patterns;
 721         int lno = 0;
 722         struct strbuf sb = STRBUF_INIT;
 723
 724         BUG_ON_OPT_NEG(unset);
 725
 726         from_stdin = !strcmp(arg, "-");
 727         patterns = from_stdin ? stdin : fopen(arg, "r");
 728         if (!patterns)
 729                 die_errno(_("cannot open '%s'"), arg);
 730         while (strbuf_getline(&sb, patterns) == 0) {
 731                 /* ignore empty line like grep does */
 732                 if (sb.len == 0)
 733                         continue;
 734
 735                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 736                                 GREP_PATTERN);
 737         }
 738         if (!from_stdin)
 739                 fclose(patterns);
 740         strbuf_release(&sb);
 741         return 0;
 742 }
 743
 744 static int not_callback(const struct option *opt, const char *arg, int unset)
 745 {
 746         struct grep_opt *grep_opt = opt->value;
 747         BUG_ON_OPT_NEG(unset);
 748         BUG_ON_OPT_ARG(arg);
 749         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 750         return 0;
 751 }
 752
 753 static int and_callback(const struct option *opt, const char *arg, int unset)
 754 {
 755         struct grep_opt *grep_opt = opt->value;
 756         BUG_ON_OPT_NEG(unset);
 757         BUG_ON_OPT_ARG(arg);
 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         BUG_ON_OPT_NEG(unset);
 766         BUG_ON_OPT_ARG(arg);
 767         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 768         return 0;
 769 }
 770
 771 static int close_callback(const struct option *opt, const char *arg, int unset)
 772 {
 773         struct grep_opt *grep_opt = opt->value;
 774         BUG_ON_OPT_NEG(unset);
 775         BUG_ON_OPT_ARG(arg);
 776         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 777         return 0;
 778 }
 779
 780 static int pattern_callback(const struct option *opt, const char *arg,
 781                             int unset)
 782 {
 783         struct grep_opt *grep_opt = opt->value;
 784         BUG_ON_OPT_NEG(unset);
 785         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 786         return 0;
 787 }
 788
 789 int cmd_grep(int argc, const char **argv, const char *prefix)
 790 {
 791         int hit = 0;
 792         int cached = 0, untracked = 0, opt_exclude = -1;
 793         int seen_dashdash = 0;
 794         int external_grep_allowed__ignored;
 795         const char *show_in_pager = NULL, *default_pager = "dummy";
 796         struct grep_opt opt;
 797         struct object_array list = OBJECT_ARRAY_INIT;
 798         struct pathspec pathspec;
 799         struct string_list path_list = STRING_LIST_INIT_NODUP;
 800         int i;
 801         int dummy;
 802         int use_index = 1;
 803         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 804         int allow_revs;
 805
 806         struct option options[] = {
 807                 OPT_BOOL(0, "cached", &cached,
 808                         N_("search in index instead of in the work tree")),
 809                 OPT_NEGBIT(0, "no-index", &use_index,
 810                          N_("find in contents not managed by git"), 1),
 811                 OPT_BOOL(0, "untracked", &untracked,
 812                         N_("search in both tracked and untracked files")),
 813                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 814                             N_("ignore files specified via '.gitignore'"), 1),
 815                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 816                          N_("recursively search in each submodule")),
 817                 OPT_GROUP(""),
 818                 OPT_BOOL('v', "invert-match", &opt.invert,
 819                         N_("show non-matching lines")),
 820                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 821                         N_("case insensitive matching")),
 822                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 823                         N_("match patterns only at word boundaries")),
 824                 OPT_SET_INT('a', "text", &opt.binary,
 825                         N_("process binary files as text"), GREP_BINARY_TEXT),
 826                 OPT_SET_INT('I', NULL, &opt.binary,
 827                         N_("don't match patterns in binary files"),
 828                         GREP_BINARY_NOMATCH),
 829                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
 830                          N_("process binary files with textconv filters")),
 831                 OPT_SET_INT('r', "recursive", &opt.max_depth,
 832                             N_("search in subdirectories (default)"), -1),
 833                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
 834                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
 835                         NULL, 1 },
 836                 OPT_GROUP(""),
 837                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
 838                             N_("use extended POSIX regular expressions"),
 839                             GREP_PATTERN_TYPE_ERE),
 840                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
 841                             N_("use basic POSIX regular expressions (default)"),
 842                             GREP_PATTERN_TYPE_BRE),
 843                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
 844                             N_("interpret patterns as fixed strings"),
 845                             GREP_PATTERN_TYPE_FIXED),
 846                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
 847                             N_("use Perl-compatible regular expressions"),
 848                             GREP_PATTERN_TYPE_PCRE),
 849                 OPT_GROUP(""),
 850                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
 851                 OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")),
 852                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
 853                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
 854                 OPT_NEGBIT(0, "full-name", &opt.relative,
 855                         N_("show filenames relative to top directory"), 1),
 856                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
 857                         N_("show only filenames instead of matching lines")),
 858                 OPT_BOOL(0, "name-only", &opt.name_only,
 859                         N_("synonym for --files-with-matches")),
 860                 OPT_BOOL('L', "files-without-match",
 861                         &opt.unmatch_name_only,
 862                         N_("show only the names of files without match")),
 863                 OPT_BOOL_F('z', "null", &opt.null_following_name,
 864                            N_("print NUL after filenames"),
 865                            PARSE_OPT_NOCOMPLETE),
 866                 OPT_BOOL('o', "only-matching", &opt.only_matching,
 867                         N_("show only matching parts of a line")),
 868                 OPT_BOOL('c', "count", &opt.count,
 869                         N_("show the number of matches instead of matching lines")),
 870                 OPT__COLOR(&opt.color, N_("highlight matches")),
 871                 OPT_BOOL(0, "break", &opt.file_break,
 872                         N_("print empty line between matches from different files")),
 873                 OPT_BOOL(0, "heading", &opt.heading,
 874                         N_("show filename only once above matches from same file")),
 875                 OPT_GROUP(""),
 876                 OPT_CALLBACK('C', "context", &opt, N_("n"),
 877                         N_("show <n> context lines before and after matches"),
 878                         context_callback),
 879                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 880                         N_("show <n> context lines before matches")),
 881                 OPT_INTEGER('A', "after-context", &opt.post_context,
 882                         N_("show <n> context lines after matches")),
 883                 OPT_INTEGER(0, "threads", &num_threads,
 884                         N_("use <n> worker threads")),
 885                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
 886                         context_callback),
 887                 OPT_BOOL('p', "show-function", &opt.funcname,
 888                         N_("show a line with the function name before matches")),
 889                 OPT_BOOL('W', "function-context", &opt.funcbody,
 890                         N_("show the surrounding function")),
 891                 OPT_GROUP(""),
 892                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
 893                         N_("read patterns from file"), file_callback),
 894                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
 895                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
 896                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 897                   N_("combine patterns specified with -e"),
 898                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 899                 OPT_BOOL(0, "or", &dummy, ""),
 900                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 901                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 902                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 903                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 904                   open_callback },
 905                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 906                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 907                   close_callback },
 908                 OPT__QUIET(&opt.status_only,
 909                            N_("indicate hit with exit status without output")),
 910                 OPT_BOOL(0, "all-match", &opt.all_match,
 911                         N_("show only matches from files that match all patterns")),
 912                 OPT_SET_INT_F(0, "debug", &opt.debug,
 913                               N_("show parse tree for grep expression"),
 914                               1, PARSE_OPT_HIDDEN),
 915                 OPT_GROUP(""),
 916                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 917                         N_("pager"), N_("show matching files in the pager"),
 918                         PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
 919                         NULL, (intptr_t)default_pager },
 920                 OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
 921                            N_("allow calling of grep(1) (ignored by this build)"),
 922                            PARSE_OPT_NOCOMPLETE),
 923                 OPT_END()
 924         };
 925
 926         init_grep_defaults(the_repository);
 927         git_config(grep_cmd_config, NULL);
 928         grep_init(&opt, the_repository, prefix);
 929
 930         /*
 931          * If there is no -- then the paths must exist in the working
 932          * tree.  If there is no explicit pattern specified with -e or
 933          * -f, we take the first unrecognized non option to be the
 934          * pattern, but then what follows it must be zero or more
 935          * valid refs up to the -- (if exists), and then existing
 936          * paths.  If there is an explicit pattern, then the first
 937          * unrecognized non option is the beginning of the refs list
 938          * that continues up to the -- (if exists), and then paths.
 939          */
 940         argc = parse_options(argc, argv, prefix, options, grep_usage,
 941                              PARSE_OPT_KEEP_DASHDASH |
 942                              PARSE_OPT_STOP_AT_NON_OPTION);
 943         grep_commit_pattern_type(pattern_type_arg, &opt);
 944
 945         if (use_index && !startup_info->have_repository) {
 946                 int fallback = 0;
 947                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
 948                 if (fallback)
 949                         use_index = 0;
 950                 else
 951                         /* die the same way as if we did it at the beginning */
 952                         setup_git_directory();
 953         }
 954
 955         /*
 956          * skip a -- separator; we know it cannot be
 957          * separating revisions from pathnames if
 958          * we haven't even had any patterns yet
 959          */
 960         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 961                 argv++;
 962                 argc--;
 963         }
 964
 965         /* First unrecognized non-option token */
 966         if (argc > 0 && !opt.pattern_list) {
 967                 append_grep_pattern(&opt, argv[0], "command line", 0,
 968                                     GREP_PATTERN);
 969                 argv++;
 970                 argc--;
 971         }
 972
 973         if (show_in_pager == default_pager)
 974                 show_in_pager = git_pager(1);
 975         if (show_in_pager) {
 976                 opt.color = 0;
 977                 opt.name_only = 1;
 978                 opt.null_following_name = 1;
 979                 opt.output_priv = &path_list;
 980                 opt.output = append_path;
 981                 string_list_append(&path_list, show_in_pager);
 982         }
 983
 984         if (!opt.pattern_list)
 985                 die(_("no pattern given"));
 986
 987         /* --only-matching has no effect with --invert. */
 988         if (opt.invert)
 989                 opt.only_matching = 0;
 990
 991         /*
 992          * We have to find "--" in a separate pass, because its presence
 993          * influences how we will parse arguments that come before it.
 994          */
 995         for (i = 0; i < argc; i++) {
 996                 if (!strcmp(argv[i], "--")) {
 997                         seen_dashdash = 1;
 998                         break;
 999                 }
1000         }
1001
1002         /*
1003          * Resolve any rev arguments. If we have a dashdash, then everything up
1004          * to it must resolve as a rev. If not, then we stop at the first
1005          * non-rev and assume everything else is a path.
1006          */
1007         allow_revs = use_index && !untracked;
1008         for (i = 0; i < argc; i++) {
1009                 const char *arg = argv[i];
1010                 struct object_id oid;
1011                 struct object_context oc;
1012                 struct object *object;
1013
1014                 if (!strcmp(arg, "--")) {
1015                         i++;
1016                         break;
1017                 }
1018
1019                 if (!allow_revs) {
1020                         if (seen_dashdash)
1021                                 die(_("--no-index or --untracked cannot be used with revs"));
1022                         break;
1023                 }
1024
1025                 if (get_oid_with_context(the_repository, arg,
1026                                          GET_OID_RECORD_PATH,
1027                                          &oid, &oc)) {
1028                         if (seen_dashdash)
1029                                 die(_("unable to resolve revision: %s"), arg);
1030                         break;
1031                 }
1032
1033                 object = parse_object_or_die(&oid, arg);
1034                 if (!seen_dashdash)
1035                         verify_non_filename(prefix, arg);
1036                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1037                 free(oc.path);
1038         }
1039
1040         /*
1041          * Anything left over is presumed to be a path. But in the non-dashdash
1042          * "do what I mean" case, we verify and complain when that isn't true.
1043          */
1044         if (!seen_dashdash) {
1045                 int j;
1046                 for (j = i; j < argc; j++)
1047                         verify_filename(prefix, argv[j], j == i && allow_revs);
1048         }
1049
1050         parse_pathspec(&pathspec, 0,
1051                        PATHSPEC_PREFER_CWD |
1052                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1053                        prefix, argv + i);
1054         pathspec.max_depth = opt.max_depth;
1055         pathspec.recursive = 1;
1056         pathspec.recurse_submodules = !!recurse_submodules;
1057
1058         if (list.nr || cached || show_in_pager) {
1059                 if (num_threads > 1)
1060                         warning(_("invalid option combination, ignoring --threads"));
1061                 num_threads = 1;
1062         } else if (!HAVE_THREADS && num_threads > 1) {
1063                 warning(_("no threads support, ignoring --threads"));
1064                 num_threads = 1;
1065         } else if (num_threads < 0)
1066                 die(_("invalid number of threads specified (%d)"), num_threads);
1067         else if (num_threads == 0)
1068                 num_threads = HAVE_THREADS ? GREP_NUM_THREADS_DEFAULT : 1;
1069
1070         if (num_threads > 1) {
1071                 if (!HAVE_THREADS)
1072                         BUG("Somebody got num_threads calculation wrong!");
1073                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1074                     && (opt.pre_context || opt.post_context ||
1075                         opt.file_break || opt.funcbody))
1076                         skip_first_line = 1;
1077                 start_threads(&opt);
1078         } else {
1079                 /*
1080                  * The compiled patterns on the main path are only
1081                  * used when not using threading. Otherwise
1082                  * start_threads() above calls compile_grep_patterns()
1083                  * for each thread.
1084                  */
1085                 compile_grep_patterns(&opt);
1086         }
1087
1088         if (show_in_pager && (cached || list.nr))
1089                 die(_("--open-files-in-pager only works on the worktree"));
1090
1091         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1092                 const char *pager = path_list.items[0].string;
1093                 int len = strlen(pager);
1094
1095                 if (len > 4 && is_dir_sep(pager[len - 5]))
1096                         pager += len - 4;
1097
1098                 if (opt.ignore_case && !strcmp("less", pager))
1099                         string_list_append(&path_list, "-I");
1100
1101                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1102                         struct strbuf buf = STRBUF_INIT;
1103                         strbuf_addf(&buf, "+/%s%s",
1104                                         strcmp("less", pager) ? "" : "*",
1105                                         opt.pattern_list->pattern);
1106                         string_list_append(&path_list, buf.buf);
1107                         strbuf_detach(&buf, NULL);
1108                 }
1109         }
1110
1111         if (recurse_submodules && (!use_index || untracked))
1112                 die(_("option not supported with --recurse-submodules"));
1113
1114         if (!show_in_pager && !opt.status_only)
1115                 setup_pager();
1116
1117         if (!use_index && (untracked || cached))
1118                 die(_("--cached or --untracked cannot be used with --no-index"));
1119
1120         if (!use_index || untracked) {
1121                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1122                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1123         } else if (0 <= opt_exclude) {
1124                 die(_("--[no-]exclude-standard cannot be used for tracked contents"));
1125         } else if (!list.nr) {
1126                 if (!cached)
1127                         setup_work_tree();
1128
1129                 hit = grep_cache(&opt, &pathspec, cached);
1130         } else {
1131                 if (cached)
1132                         die(_("both --cached and trees are given"));
1133
1134                 hit = grep_objects(&opt, &pathspec, &list);
1135         }
1136
1137         if (num_threads > 1)
1138                 hit |= wait_all();
1139         if (hit && show_in_pager)
1140                 run_pager(&opt, prefix);
1141         clear_pathspec(&pathspec);
1142         free_grep_patterns(&opt);
1143         return !hit;
1144 }