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