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 #include "packfile.h"
  28
  29 static char const * const grep_usage[] = {
  30         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  31         NULL
  32 };
  33
  34 static int recurse_submodules;
  35
  36 #define GREP_NUM_THREADS_DEFAULT 8
  37 static int num_threads;
  38
  39 static pthread_t *threads;
  40
  41 /* We use one producer thread and THREADS consumer
  42  * threads. The producer adds struct work_items to 'todo' and the
  43  * consumers pick work items from the same array.
  44  */
  45 struct work_item {
  46         struct grep_source source;
  47         char done;
  48         struct strbuf out;
  49 };
  50
  51 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  52  * that have been or are processed by a consumer thread. We haven't
  53  * written the result for these to stdout yet.
  54  *
  55  * The work_items in [todo_start, todo_end) are waiting to be picked
  56  * up by a consumer thread.
  57  *
  58  * The ranges are modulo TODO_SIZE.
  59  */
  60 #define TODO_SIZE 128
  61 static struct work_item todo[TODO_SIZE];
  62 static int todo_start;
  63 static int todo_end;
  64 static int todo_done;
  65
  66 /* Has all work items been added? */
  67 static int all_work_added;
  68
  69 /* This lock protects all the variables above. */
  70 static pthread_mutex_t grep_mutex;
  71
  72 static inline void grep_lock(void)
  73 {
  74         pthread_mutex_lock(&grep_mutex);
  75 }
  76
  77 static inline void grep_unlock(void)
  78 {
  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_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         enable_obj_read_lock();
 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                 BUG("Never call this function unless you have started threads");
 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_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         disable_obj_read_lock();
 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 > 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 = 1;
 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 int grep_oid(struct grep_opt *opt, const struct object_id *oid,
 300                      const char *filename, int tree_name_len,
 301                      const char *path)
 302 {
 303         struct strbuf pathbuf = STRBUF_INIT;
 304         struct grep_source gs;
 305
 306         if (opt->relative && opt->prefix_length) {
 307                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
 308                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 309         } else {
 310                 strbuf_addstr(&pathbuf, filename);
 311         }
 312
 313         grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
 314         strbuf_release(&pathbuf);
 315
 316         if (num_threads > 1) {
 317                 /*
 318                  * add_work() copies gs and thus assumes ownership of
 319                  * its fields, so do not call grep_source_clear()
 320                  */
 321                 add_work(opt, &gs);
 322                 return 0;
 323         } else {
 324                 int hit;
 325
 326                 hit = grep_source(opt, &gs);
 327
 328                 grep_source_clear(&gs);
 329                 return hit;
 330         }
 331 }
 332
 333 static int grep_file(struct grep_opt *opt, const char *filename)
 334 {
 335         struct strbuf buf = STRBUF_INIT;
 336         struct grep_source gs;
 337
 338         if (opt->relative && opt->prefix_length)
 339                 quote_path_relative(filename, opt->prefix, &buf);
 340         else
 341                 strbuf_addstr(&buf, filename);
 342
 343         grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 344         strbuf_release(&buf);
 345
 346         if (num_threads > 1) {
 347                 /*
 348                  * add_work() copies gs and thus assumes ownership of
 349                  * its fields, so do not call grep_source_clear()
 350                  */
 351                 add_work(opt, &gs);
 352                 return 0;
 353         } else {
 354                 int hit;
 355
 356                 hit = grep_source(opt, &gs);
 357
 358                 grep_source_clear(&gs);
 359                 return hit;
 360         }
 361 }
 362
 363 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 364 {
 365         struct string_list *path_list = opt->output_priv;
 366
 367         if (len == 1 && *(const char *)data == '\0')
 368                 return;
 369         string_list_append(path_list, xstrndup(data, len));
 370 }
 371
 372 static void run_pager(struct grep_opt *opt, const char *prefix)
 373 {
 374         struct string_list *path_list = opt->output_priv;
 375         struct child_process child = CHILD_PROCESS_INIT;
 376         int i, status;
 377
 378         for (i = 0; i < path_list->nr; i++)
 379                 argv_array_push(&child.args, path_list->items[i].string);
 380         child.dir = prefix;
 381         child.use_shell = 1;
 382
 383         status = run_command(&child);
 384         if (status)
 385                 exit(status);
 386 }
 387
 388 static int grep_cache(struct grep_opt *opt,
 389                       const struct pathspec *pathspec, int cached);
 390 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 391                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 392                      int check_attr);
 393
 394 static int grep_submodule(struct grep_opt *opt,
 395                           const struct pathspec *pathspec,
 396                           const struct object_id *oid,
 397                           const char *filename, const char *path, int cached)
 398 {
 399         struct repository subrepo;
 400         struct repository *superproject = opt->repo;
 401         const struct submodule *sub;
 402         struct grep_opt subopt;
 403         int hit;
 404
 405         sub = submodule_from_path(superproject, &null_oid, path);
 406
 407         if (!is_submodule_active(superproject, path))
 408                 return 0;
 409
 410         if (repo_submodule_init(&subrepo, superproject, sub))
 411                 return 0;
 412
 413         /*
 414          * NEEDSWORK: repo_read_gitmodules() might call
 415          * add_to_alternates_memory() via config_from_gitmodules(). This
 416          * operation causes a race condition with concurrent object readings
 417          * performed by the worker threads. That's why we need obj_read_lock()
 418          * here. It should be removed once it's no longer necessary to add the
 419          * subrepo's odbs to the in-memory alternates list.
 420          */
 421         obj_read_lock();
 422         repo_read_gitmodules(&subrepo, 0);
 423
 424         /*
 425          * NEEDSWORK: This adds the submodule's object directory to the list of
 426          * alternates for the single in-memory object store.  This has some bad
 427          * consequences for memory (processed objects will never be freed) and
 428          * performance (this increases the number of pack files git has to pay
 429          * attention to, to the sum of the number of pack files in all the
 430          * repositories processed so far).  This can be removed once the object
 431          * store is no longer global and instead is a member of the repository
 432          * object.
 433          */
 434         add_to_alternates_memory(subrepo.objects->odb->path);
 435         obj_read_unlock();
 436
 437         memcpy(&subopt, opt, sizeof(subopt));
 438         subopt.repo = &subrepo;
 439
 440         if (oid) {
 441                 struct object *object;
 442                 struct tree_desc tree;
 443                 void *data;
 444                 unsigned long size;
 445                 struct strbuf base = STRBUF_INIT;
 446
 447                 obj_read_lock();
 448                 object = parse_object_or_die(oid, oid_to_hex(oid));
 449                 obj_read_unlock();
 450                 data = read_object_with_reference(&subrepo,
 451                                                   &object->oid, tree_type,
 452                                                   &size, NULL);
 453                 if (!data)
 454                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
 455
 456                 strbuf_addstr(&base, filename);
 457                 strbuf_addch(&base, '/');
 458
 459                 init_tree_desc(&tree, data, size);
 460                 hit = grep_tree(&subopt, pathspec, &tree, &base, base.len,
 461                                 object->type == OBJ_COMMIT);
 462                 strbuf_release(&base);
 463                 free(data);
 464         } else {
 465                 hit = grep_cache(&subopt, pathspec, cached);
 466         }
 467
 468         repo_clear(&subrepo);
 469         return hit;
 470 }
 471
 472 static int grep_cache(struct grep_opt *opt,
 473                       const struct pathspec *pathspec, int cached)
 474 {
 475         struct repository *repo = opt->repo;
 476         int hit = 0;
 477         int nr;
 478         struct strbuf name = STRBUF_INIT;
 479         int name_base_len = 0;
 480         if (repo->submodule_prefix) {
 481                 name_base_len = strlen(repo->submodule_prefix);
 482                 strbuf_addstr(&name, repo->submodule_prefix);
 483         }
 484
 485         if (repo_read_index(repo) < 0)
 486                 die(_("index file corrupt"));
 487
 488         for (nr = 0; nr < repo->index->cache_nr; nr++) {
 489                 const struct cache_entry *ce = repo->index->cache[nr];
 490                 strbuf_setlen(&name, name_base_len);
 491                 strbuf_addstr(&name, ce->name);
 492
 493                 if (S_ISREG(ce->ce_mode) &&
 494                     match_pathspec(repo->index, pathspec, name.buf, name.len, 0, NULL,
 495                                    S_ISDIR(ce->ce_mode) ||
 496                                    S_ISGITLINK(ce->ce_mode))) {
 497                         /*
 498                          * If CE_VALID is on, we assume worktree file and its
 499                          * cache entry are identical, even if worktree file has
 500                          * been modified, so use cache version instead
 501                          */
 502                         if (cached || (ce->ce_flags & CE_VALID) ||
 503                             ce_skip_worktree(ce)) {
 504                                 if (ce_stage(ce) || ce_intent_to_add(ce))
 505                                         continue;
 506                                 hit |= grep_oid(opt, &ce->oid, name.buf,
 507                                                  0, name.buf);
 508                         } else {
 509                                 hit |= grep_file(opt, name.buf);
 510                         }
 511                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 512                            submodule_path_match(repo->index, pathspec, name.buf, NULL)) {
 513                         hit |= grep_submodule(opt, pathspec, NULL, ce->name,
 514                                               ce->name, cached);
 515                 } else {
 516                         continue;
 517                 }
 518
 519                 if (ce_stage(ce)) {
 520                         do {
 521                                 nr++;
 522                         } while (nr < repo->index->cache_nr &&
 523                                  !strcmp(ce->name, repo->index->cache[nr]->name));
 524                         nr--; /* compensate for loop control */
 525                 }
 526                 if (hit && opt->status_only)
 527                         break;
 528         }
 529
 530         strbuf_release(&name);
 531         return hit;
 532 }
 533
 534 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 535                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 536                      int check_attr)
 537 {
 538         struct repository *repo = opt->repo;
 539         int hit = 0;
 540         enum interesting match = entry_not_interesting;
 541         struct name_entry entry;
 542         int old_baselen = base->len;
 543         struct strbuf name = STRBUF_INIT;
 544         int name_base_len = 0;
 545         if (repo->submodule_prefix) {
 546                 strbuf_addstr(&name, repo->submodule_prefix);
 547                 name_base_len = name.len;
 548         }
 549
 550         while (tree_entry(tree, &entry)) {
 551                 int te_len = tree_entry_len(&entry);
 552
 553                 if (match != all_entries_interesting) {
 554                         strbuf_addstr(&name, base->buf + tn_len);
 555                         match = tree_entry_interesting(repo->index,
 556                                                        &entry, &name,
 557                                                        0, pathspec);
 558                         strbuf_setlen(&name, name_base_len);
 559
 560                         if (match == all_entries_not_interesting)
 561                                 break;
 562                         if (match == entry_not_interesting)
 563                                 continue;
 564                 }
 565
 566                 strbuf_add(base, entry.path, te_len);
 567
 568                 if (S_ISREG(entry.mode)) {
 569                         hit |= grep_oid(opt, &entry.oid, base->buf, tn_len,
 570                                          check_attr ? base->buf + tn_len : NULL);
 571                 } else if (S_ISDIR(entry.mode)) {
 572                         enum object_type type;
 573                         struct tree_desc sub;
 574                         void *data;
 575                         unsigned long size;
 576
 577                         data = read_object_file(&entry.oid, &type, &size);
 578                         if (!data)
 579                                 die(_("unable to read tree (%s)"),
 580                                     oid_to_hex(&entry.oid));
 581
 582                         strbuf_addch(base, '/');
 583                         init_tree_desc(&sub, data, size);
 584                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 585                                          check_attr);
 586                         free(data);
 587                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 588                         hit |= grep_submodule(opt, pathspec, &entry.oid,
 589                                               base->buf, base->buf + tn_len,
 590                                               1); /* ignored */
 591                 }
 592
 593                 strbuf_setlen(base, old_baselen);
 594
 595                 if (hit && opt->status_only)
 596                         break;
 597         }
 598
 599         strbuf_release(&name);
 600         return hit;
 601 }
 602
 603 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 604                        struct object *obj, const char *name, const char *path)
 605 {
 606         if (obj->type == OBJ_BLOB)
 607                 return grep_oid(opt, &obj->oid, name, 0, path);
 608         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 609                 struct tree_desc tree;
 610                 void *data;
 611                 unsigned long size;
 612                 struct strbuf base;
 613                 int hit, len;
 614
 615                 data = read_object_with_reference(opt->repo,
 616                                                   &obj->oid, tree_type,
 617                                                   &size, NULL);
 618                 if (!data)
 619                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 620
 621                 len = name ? strlen(name) : 0;
 622                 strbuf_init(&base, PATH_MAX + len + 1);
 623                 if (len) {
 624                         strbuf_add(&base, name, len);
 625                         strbuf_addch(&base, ':');
 626                 }
 627                 init_tree_desc(&tree, data, size);
 628                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 629                                 obj->type == OBJ_COMMIT);
 630                 strbuf_release(&base);
 631                 free(data);
 632                 return hit;
 633         }
 634         die(_("unable to grep from object of type %s"), type_name(obj->type));
 635 }
 636
 637 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 638                         const struct object_array *list)
 639 {
 640         unsigned int i;
 641         int hit = 0;
 642         const unsigned int nr = list->nr;
 643
 644         for (i = 0; i < nr; i++) {
 645                 struct object *real_obj;
 646
 647                 obj_read_lock();
 648                 real_obj = deref_tag(opt->repo, list->objects[i].item,
 649                                      NULL, 0);
 650                 obj_read_unlock();
 651
 652                 /* load the gitmodules file for this rev */
 653                 if (recurse_submodules) {
 654                         submodule_free(opt->repo);
 655                         obj_read_lock();
 656                         gitmodules_config_oid(&real_obj->oid);
 657                         obj_read_unlock();
 658                 }
 659                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
 660                                 list->objects[i].path)) {
 661                         hit = 1;
 662                         if (opt->status_only)
 663                                 break;
 664                 }
 665         }
 666         return hit;
 667 }
 668
 669 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 670                           int exc_std, int use_index)
 671 {
 672         struct dir_struct dir;
 673         int i, hit = 0;
 674
 675         memset(&dir, 0, sizeof(dir));
 676         if (!use_index)
 677                 dir.flags |= DIR_NO_GITLINKS;
 678         if (exc_std)
 679                 setup_standard_excludes(&dir);
 680
 681         fill_directory(&dir, opt->repo->index, pathspec);
 682         for (i = 0; i < dir.nr; i++) {
 683                 if (!dir_path_match(opt->repo->index, dir.entries[i], pathspec, 0, NULL))
 684                         continue;
 685                 hit |= grep_file(opt, dir.entries[i]->name);
 686                 if (hit && opt->status_only)
 687                         break;
 688         }
 689         return hit;
 690 }
 691
 692 static int context_callback(const struct option *opt, const char *arg,
 693                             int unset)
 694 {
 695         struct grep_opt *grep_opt = opt->value;
 696         int value;
 697         const char *endp;
 698
 699         if (unset) {
 700                 grep_opt->pre_context = grep_opt->post_context = 0;
 701                 return 0;
 702         }
 703         value = strtol(arg, (char **)&endp, 10);
 704         if (*endp) {
 705                 return error(_("switch `%c' expects a numerical value"),
 706                              opt->short_name);
 707         }
 708         grep_opt->pre_context = grep_opt->post_context = value;
 709         return 0;
 710 }
 711
 712 static int file_callback(const struct option *opt, const char *arg, int unset)
 713 {
 714         struct grep_opt *grep_opt = opt->value;
 715         int from_stdin;
 716         FILE *patterns;
 717         int lno = 0;
 718         struct strbuf sb = STRBUF_INIT;
 719
 720         BUG_ON_OPT_NEG(unset);
 721
 722         from_stdin = !strcmp(arg, "-");
 723         patterns = from_stdin ? stdin : fopen(arg, "r");
 724         if (!patterns)
 725                 die_errno(_("cannot open '%s'"), arg);
 726         while (strbuf_getline(&sb, patterns) == 0) {
 727                 /* ignore empty line like grep does */
 728                 if (sb.len == 0)
 729                         continue;
 730
 731                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 732                                 GREP_PATTERN);
 733         }
 734         if (!from_stdin)
 735                 fclose(patterns);
 736         strbuf_release(&sb);
 737         return 0;
 738 }
 739
 740 static int not_callback(const struct option *opt, const char *arg, int unset)
 741 {
 742         struct grep_opt *grep_opt = opt->value;
 743         BUG_ON_OPT_NEG(unset);
 744         BUG_ON_OPT_ARG(arg);
 745         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 746         return 0;
 747 }
 748
 749 static int and_callback(const struct option *opt, const char *arg, int unset)
 750 {
 751         struct grep_opt *grep_opt = opt->value;
 752         BUG_ON_OPT_NEG(unset);
 753         BUG_ON_OPT_ARG(arg);
 754         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 755         return 0;
 756 }
 757
 758 static int open_callback(const struct option *opt, const char *arg, int unset)
 759 {
 760         struct grep_opt *grep_opt = opt->value;
 761         BUG_ON_OPT_NEG(unset);
 762         BUG_ON_OPT_ARG(arg);
 763         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 764         return 0;
 765 }
 766
 767 static int close_callback(const struct option *opt, const char *arg, int unset)
 768 {
 769         struct grep_opt *grep_opt = opt->value;
 770         BUG_ON_OPT_NEG(unset);
 771         BUG_ON_OPT_ARG(arg);
 772         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 773         return 0;
 774 }
 775
 776 static int pattern_callback(const struct option *opt, const char *arg,
 777                             int unset)
 778 {
 779         struct grep_opt *grep_opt = opt->value;
 780         BUG_ON_OPT_NEG(unset);
 781         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 782         return 0;
 783 }
 784
 785 int cmd_grep(int argc, const char **argv, const char *prefix)
 786 {
 787         int hit = 0;
 788         int cached = 0, untracked = 0, opt_exclude = -1;
 789         int seen_dashdash = 0;
 790         int external_grep_allowed__ignored;
 791         const char *show_in_pager = NULL, *default_pager = "dummy";
 792         struct grep_opt opt;
 793         struct object_array list = OBJECT_ARRAY_INIT;
 794         struct pathspec pathspec;
 795         struct string_list path_list = STRING_LIST_INIT_NODUP;
 796         int i;
 797         int dummy;
 798         int use_index = 1;
 799         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 800         int allow_revs;
 801
 802         struct option options[] = {
 803                 OPT_BOOL(0, "cached", &cached,
 804                         N_("search in index instead of in the work tree")),
 805                 OPT_NEGBIT(0, "no-index", &use_index,
 806                          N_("find in contents not managed by git"), 1),
 807                 OPT_BOOL(0, "untracked", &untracked,
 808                         N_("search in both tracked and untracked files")),
 809                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 810                             N_("ignore files specified via '.gitignore'"), 1),
 811                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 812                          N_("recursively search in each submodule")),
 813                 OPT_GROUP(""),
 814                 OPT_BOOL('v', "invert-match", &opt.invert,
 815                         N_("show non-matching lines")),
 816                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 817                         N_("case insensitive matching")),
 818                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 819                         N_("match patterns only at word boundaries")),
 820                 OPT_SET_INT('a', "text", &opt.binary,
 821                         N_("process binary files as text"), GREP_BINARY_TEXT),
 822                 OPT_SET_INT('I', NULL, &opt.binary,
 823                         N_("don't match patterns in binary files"),
 824                         GREP_BINARY_NOMATCH),
 825                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
 826                          N_("process binary files with textconv filters")),
 827                 OPT_SET_INT('r', "recursive", &opt.max_depth,
 828                             N_("search in subdirectories (default)"), -1),
 829                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
 830                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
 831                         NULL, 1 },
 832                 OPT_GROUP(""),
 833                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
 834                             N_("use extended POSIX regular expressions"),
 835                             GREP_PATTERN_TYPE_ERE),
 836                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
 837                             N_("use basic POSIX regular expressions (default)"),
 838                             GREP_PATTERN_TYPE_BRE),
 839                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
 840                             N_("interpret patterns as fixed strings"),
 841                             GREP_PATTERN_TYPE_FIXED),
 842                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
 843                             N_("use Perl-compatible regular expressions"),
 844                             GREP_PATTERN_TYPE_PCRE),
 845                 OPT_GROUP(""),
 846                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
 847                 OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")),
 848                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
 849                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
 850                 OPT_NEGBIT(0, "full-name", &opt.relative,
 851                         N_("show filenames relative to top directory"), 1),
 852                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
 853                         N_("show only filenames instead of matching lines")),
 854                 OPT_BOOL(0, "name-only", &opt.name_only,
 855                         N_("synonym for --files-with-matches")),
 856                 OPT_BOOL('L', "files-without-match",
 857                         &opt.unmatch_name_only,
 858                         N_("show only the names of files without match")),
 859                 OPT_BOOL_F('z', "null", &opt.null_following_name,
 860                            N_("print NUL after filenames"),
 861                            PARSE_OPT_NOCOMPLETE),
 862                 OPT_BOOL('o', "only-matching", &opt.only_matching,
 863                         N_("show only matching parts of a line")),
 864                 OPT_BOOL('c', "count", &opt.count,
 865                         N_("show the number of matches instead of matching lines")),
 866                 OPT__COLOR(&opt.color, N_("highlight matches")),
 867                 OPT_BOOL(0, "break", &opt.file_break,
 868                         N_("print empty line between matches from different files")),
 869                 OPT_BOOL(0, "heading", &opt.heading,
 870                         N_("show filename only once above matches from same file")),
 871                 OPT_GROUP(""),
 872                 OPT_CALLBACK('C', "context", &opt, N_("n"),
 873                         N_("show <n> context lines before and after matches"),
 874                         context_callback),
 875                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 876                         N_("show <n> context lines before matches")),
 877                 OPT_INTEGER('A', "after-context", &opt.post_context,
 878                         N_("show <n> context lines after matches")),
 879                 OPT_INTEGER(0, "threads", &num_threads,
 880                         N_("use <n> worker threads")),
 881                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
 882                         context_callback),
 883                 OPT_BOOL('p', "show-function", &opt.funcname,
 884                         N_("show a line with the function name before matches")),
 885                 OPT_BOOL('W', "function-context", &opt.funcbody,
 886                         N_("show the surrounding function")),
 887                 OPT_GROUP(""),
 888                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
 889                         N_("read patterns from file"), file_callback),
 890                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
 891                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
 892                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 893                   N_("combine patterns specified with -e"),
 894                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 895                 OPT_BOOL(0, "or", &dummy, ""),
 896                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 897                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 898                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 899                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 900                   open_callback },
 901                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 902                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 903                   close_callback },
 904                 OPT__QUIET(&opt.status_only,
 905                            N_("indicate hit with exit status without output")),
 906                 OPT_BOOL(0, "all-match", &opt.all_match,
 907                         N_("show only matches from files that match all patterns")),
 908                 OPT_SET_INT_F(0, "debug", &opt.debug,
 909                               N_("show parse tree for grep expression"),
 910                               1, PARSE_OPT_HIDDEN),
 911                 OPT_GROUP(""),
 912                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 913                         N_("pager"), N_("show matching files in the pager"),
 914                         PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
 915                         NULL, (intptr_t)default_pager },
 916                 OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
 917                            N_("allow calling of grep(1) (ignored by this build)"),
 918                            PARSE_OPT_NOCOMPLETE),
 919                 OPT_END()
 920         };
 921
 922         init_grep_defaults(the_repository);
 923         git_config(grep_cmd_config, NULL);
 924         grep_init(&opt, the_repository, prefix);
 925
 926         /*
 927          * If there is no -- then the paths must exist in the working
 928          * tree.  If there is no explicit pattern specified with -e or
 929          * -f, we take the first unrecognized non option to be the
 930          * pattern, but then what follows it must be zero or more
 931          * valid refs up to the -- (if exists), and then existing
 932          * paths.  If there is an explicit pattern, then the first
 933          * unrecognized non option is the beginning of the refs list
 934          * that continues up to the -- (if exists), and then paths.
 935          */
 936         argc = parse_options(argc, argv, prefix, options, grep_usage,
 937                              PARSE_OPT_KEEP_DASHDASH |
 938                              PARSE_OPT_STOP_AT_NON_OPTION);
 939         grep_commit_pattern_type(pattern_type_arg, &opt);
 940
 941         if (use_index && !startup_info->have_repository) {
 942                 int fallback = 0;
 943                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
 944                 if (fallback)
 945                         use_index = 0;
 946                 else
 947                         /* die the same way as if we did it at the beginning */
 948                         setup_git_directory();
 949         }
 950
 951         /*
 952          * skip a -- separator; we know it cannot be
 953          * separating revisions from pathnames if
 954          * we haven't even had any patterns yet
 955          */
 956         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 957                 argv++;
 958                 argc--;
 959         }
 960
 961         /* First unrecognized non-option token */
 962         if (argc > 0 && !opt.pattern_list) {
 963                 append_grep_pattern(&opt, argv[0], "command line", 0,
 964                                     GREP_PATTERN);
 965                 argv++;
 966                 argc--;
 967         }
 968
 969         if (show_in_pager == default_pager)
 970                 show_in_pager = git_pager(1);
 971         if (show_in_pager) {
 972                 opt.color = 0;
 973                 opt.name_only = 1;
 974                 opt.null_following_name = 1;
 975                 opt.output_priv = &path_list;
 976                 opt.output = append_path;
 977                 string_list_append(&path_list, show_in_pager);
 978         }
 979
 980         if (!opt.pattern_list)
 981                 die(_("no pattern given"));
 982
 983         /* --only-matching has no effect with --invert. */
 984         if (opt.invert)
 985                 opt.only_matching = 0;
 986
 987         /*
 988          * We have to find "--" in a separate pass, because its presence
 989          * influences how we will parse arguments that come before it.
 990          */
 991         for (i = 0; i < argc; i++) {
 992                 if (!strcmp(argv[i], "--")) {
 993                         seen_dashdash = 1;
 994                         break;
 995                 }
 996         }
 997
 998         /*
 999          * Resolve any rev arguments. If we have a dashdash, then everything up
1000          * to it must resolve as a rev. If not, then we stop at the first
1001          * non-rev and assume everything else is a path.
1002          */
1003         allow_revs = use_index && !untracked;
1004         for (i = 0; i < argc; i++) {
1005                 const char *arg = argv[i];
1006                 struct object_id oid;
1007                 struct object_context oc;
1008                 struct object *object;
1009
1010                 if (!strcmp(arg, "--")) {
1011                         i++;
1012                         break;
1013                 }
1014
1015                 if (!allow_revs) {
1016                         if (seen_dashdash)
1017                                 die(_("--no-index or --untracked cannot be used with revs"));
1018                         break;
1019                 }
1020
1021                 if (get_oid_with_context(the_repository, arg,
1022                                          GET_OID_RECORD_PATH,
1023                                          &oid, &oc)) {
1024                         if (seen_dashdash)
1025                                 die(_("unable to resolve revision: %s"), arg);
1026                         break;
1027                 }
1028
1029                 object = parse_object_or_die(&oid, arg);
1030                 if (!seen_dashdash)
1031                         verify_non_filename(prefix, arg);
1032                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1033                 free(oc.path);
1034         }
1035
1036         /*
1037          * Anything left over is presumed to be a path. But in the non-dashdash
1038          * "do what I mean" case, we verify and complain when that isn't true.
1039          */
1040         if (!seen_dashdash) {
1041                 int j;
1042                 for (j = i; j < argc; j++)
1043                         verify_filename(prefix, argv[j], j == i && allow_revs);
1044         }
1045
1046         parse_pathspec(&pathspec, 0,
1047                        PATHSPEC_PREFER_CWD |
1048                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1049                        prefix, argv + i);
1050         pathspec.max_depth = opt.max_depth;
1051         pathspec.recursive = 1;
1052         pathspec.recurse_submodules = !!recurse_submodules;
1053
1054         if (recurse_submodules && (!use_index || untracked))
1055                 die(_("option not supported with --recurse-submodules"));
1056
1057         if (list.nr || cached || show_in_pager) {
1058                 if (num_threads > 1)
1059                         warning(_("invalid option combination, ignoring --threads"));
1060                 num_threads = 1;
1061         } else if (!HAVE_THREADS && num_threads > 1) {
1062                 warning(_("no threads support, ignoring --threads"));
1063                 num_threads = 1;
1064         } else if (num_threads < 0)
1065                 die(_("invalid number of threads specified (%d)"), num_threads);
1066         else if (num_threads == 0)
1067                 num_threads = HAVE_THREADS ? GREP_NUM_THREADS_DEFAULT : 1;
1068
1069         if (num_threads > 1) {
1070                 if (!HAVE_THREADS)
1071                         BUG("Somebody got num_threads calculation wrong!");
1072                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1073                     && (opt.pre_context || opt.post_context ||
1074                         opt.file_break || opt.funcbody))
1075                         skip_first_line = 1;
1076
1077                 /*
1078                  * Pre-read gitmodules (if not read already) and force eager
1079                  * initialization of packed_git to prevent racy lazy
1080                  * reading/initialization once worker threads are started.
1081                  */
1082                 if (recurse_submodules)
1083                         repo_read_gitmodules(the_repository, 1);
1084                 if (startup_info->have_repository)
1085                         (void)get_packed_git(the_repository);
1086
1087                 start_threads(&opt);
1088         } else {
1089                 /*
1090                  * The compiled patterns on the main path are only
1091                  * used when not using threading. Otherwise
1092                  * start_threads() above calls compile_grep_patterns()
1093                  * for each thread.
1094                  */
1095                 compile_grep_patterns(&opt);
1096         }
1097
1098         if (show_in_pager && (cached || list.nr))
1099                 die(_("--open-files-in-pager only works on the worktree"));
1100
1101         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1102                 const char *pager = path_list.items[0].string;
1103                 int len = strlen(pager);
1104
1105                 if (len > 4 && is_dir_sep(pager[len - 5]))
1106                         pager += len - 4;
1107
1108                 if (opt.ignore_case && !strcmp("less", pager))
1109                         string_list_append(&path_list, "-I");
1110
1111                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1112                         struct strbuf buf = STRBUF_INIT;
1113                         strbuf_addf(&buf, "+/%s%s",
1114                                         strcmp("less", pager) ? "" : "*",
1115                                         opt.pattern_list->pattern);
1116                         string_list_append(&path_list,
1117                                            strbuf_detach(&buf, NULL));
1118                 }
1119         }
1120
1121         if (!show_in_pager && !opt.status_only)
1122                 setup_pager();
1123
1124         if (!use_index && (untracked || cached))
1125                 die(_("--cached or --untracked cannot be used with --no-index"));
1126
1127         if (!use_index || untracked) {
1128                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1129                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1130         } else if (0 <= opt_exclude) {
1131                 die(_("--[no-]exclude-standard cannot be used for tracked contents"));
1132         } else if (!list.nr) {
1133                 if (!cached)
1134                         setup_work_tree();
1135
1136                 hit = grep_cache(&opt, &pathspec, cached);
1137         } else {
1138                 if (cached)
1139                         die(_("both --cached and trees are given"));
1140
1141                 hit = grep_objects(&opt, &pathspec, &list);
1142         }
1143
1144         if (num_threads > 1)
1145                 hit |= wait_all();
1146         if (hit && show_in_pager)
1147                 run_pager(&opt, prefix);
1148         clear_pathspec(&pathspec);
1149         free_grep_patterns(&opt);
1150         grep_destroy();
1151         return !hit;
1152 }