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 static int num_threads;
  37
  38 static pthread_t *threads;
  39
  40 /* We use one producer thread and THREADS consumer
  41  * threads. The producer adds struct work_items to 'todo' and the
  42  * consumers pick work items from the same array.
  43  */
  44 struct work_item {
  45         struct grep_source source;
  46         char done;
  47         struct strbuf out;
  48 };
  49
  50 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  51  * that have been or are processed by a consumer thread. We haven't
  52  * written the result for these to stdout yet.
  53  *
  54  * The work_items in [todo_start, todo_end) are waiting to be picked
  55  * up by a consumer thread.
  56  *
  57  * The ranges are modulo TODO_SIZE.
  58  */
  59 #define TODO_SIZE 128
  60 static struct work_item todo[TODO_SIZE];
  61 static int todo_start;
  62 static int todo_end;
  63 static int todo_done;
  64
  65 /* Has all work items been added? */
  66 static int all_work_added;
  67
  68 /* This lock protects all the variables above. */
  69 static pthread_mutex_t grep_mutex;
  70
  71 static inline void grep_lock(void)
  72 {
  73         pthread_mutex_lock(&grep_mutex);
  74 }
  75
  76 static inline void grep_unlock(void)
  77 {
  78         pthread_mutex_unlock(&grep_mutex);
  79 }
  80
  81 /* Signalled when a new work_item is added to todo. */
  82 static pthread_cond_t cond_add;
  83
  84 /* Signalled when the result from one work_item is written to
  85  * stdout.
  86  */
  87 static pthread_cond_t cond_write;
  88
  89 /* Signalled when we are finished with everything. */
  90 static pthread_cond_t cond_result;
  91
  92 static int skip_first_line;
  93
  94 static void add_work(struct grep_opt *opt, struct grep_source *gs)
  95 {
  96         if (opt->binary != GREP_BINARY_TEXT)
  97                 grep_source_load_driver(gs, opt->repo->index);
  98
  99         grep_lock();
 100
 101         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 102                 pthread_cond_wait(&cond_write, &grep_mutex);
 103         }
 104
 105         todo[todo_end].source = *gs;
 106         todo[todo_end].done = 0;
 107         strbuf_reset(&todo[todo_end].out);
 108         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 109
 110         pthread_cond_signal(&cond_add);
 111         grep_unlock();
 112 }
 113
 114 static struct work_item *get_work(void)
 115 {
 116         struct work_item *ret;
 117
 118         grep_lock();
 119         while (todo_start == todo_end && !all_work_added) {
 120                 pthread_cond_wait(&cond_add, &grep_mutex);
 121         }
 122
 123         if (todo_start == todo_end && all_work_added) {
 124                 ret = NULL;
 125         } else {
 126                 ret = &todo[todo_start];
 127                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 128         }
 129         grep_unlock();
 130         return ret;
 131 }
 132
 133 static void work_done(struct work_item *w)
 134 {
 135         int old_done;
 136
 137         grep_lock();
 138         w->done = 1;
 139         old_done = todo_done;
 140         for(; todo[todo_done].done && todo_done != todo_start;
 141             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 142                 w = &todo[todo_done];
 143                 if (w->out.len) {
 144                         const char *p = w->out.buf;
 145                         size_t len = w->out.len;
 146
 147                         /* Skip the leading hunk mark of the first file. */
 148                         if (skip_first_line) {
 149                                 while (len) {
 150                                         len--;
 151                                         if (*p++ == '\n')
 152                                                 break;
 153                                 }
 154                                 skip_first_line = 0;
 155                         }
 156
 157                         write_or_die(1, p, len);
 158                 }
 159                 grep_source_clear(&w->source);
 160         }
 161
 162         if (old_done != todo_done)
 163                 pthread_cond_signal(&cond_write);
 164
 165         if (all_work_added && todo_done == todo_end)
 166                 pthread_cond_signal(&cond_result);
 167
 168         grep_unlock();
 169 }
 170
 171 static void *run(void *arg)
 172 {
 173         int hit = 0;
 174         struct grep_opt *opt = arg;
 175
 176         while (1) {
 177                 struct work_item *w = get_work();
 178                 if (!w)
 179                         break;
 180
 181                 opt->output_priv = w;
 182                 hit |= grep_source(opt, &w->source);
 183                 grep_source_clear_data(&w->source);
 184                 work_done(w);
 185         }
 186         free_grep_patterns(arg);
 187         free(arg);
 188
 189         return (void*) (intptr_t) hit;
 190 }
 191
 192 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 193 {
 194         struct work_item *w = opt->output_priv;
 195         strbuf_add(&w->out, buf, size);
 196 }
 197
 198 static void start_threads(struct grep_opt *opt)
 199 {
 200         int i;
 201
 202         pthread_mutex_init(&grep_mutex, NULL);
 203         pthread_mutex_init(&grep_attr_mutex, NULL);
 204         pthread_cond_init(&cond_add, NULL);
 205         pthread_cond_init(&cond_write, NULL);
 206         pthread_cond_init(&cond_result, NULL);
 207         grep_use_locks = 1;
 208         enable_obj_read_lock();
 209
 210         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 211                 strbuf_init(&todo[i].out, 0);
 212         }
 213
 214         threads = xcalloc(num_threads, sizeof(*threads));
 215         for (i = 0; i < num_threads; i++) {
 216                 int err;
 217                 struct grep_opt *o = grep_opt_dup(opt);
 218                 o->output = strbuf_out;
 219                 compile_grep_patterns(o);
 220                 err = pthread_create(&threads[i], NULL, run, o);
 221
 222                 if (err)
 223                         die(_("grep: failed to create thread: %s"),
 224                             strerror(err));
 225         }
 226 }
 227
 228 static int wait_all(void)
 229 {
 230         int hit = 0;
 231         int i;
 232
 233         if (!HAVE_THREADS)
 234                 BUG("Never call this function unless you have started threads");
 235
 236         grep_lock();
 237         all_work_added = 1;
 238
 239         /* Wait until all work is done. */
 240         while (todo_done != todo_end)
 241                 pthread_cond_wait(&cond_result, &grep_mutex);
 242
 243         /* Wake up all the consumer threads so they can see that there
 244          * is no more work to do.
 245          */
 246         pthread_cond_broadcast(&cond_add);
 247         grep_unlock();
 248
 249         for (i = 0; i < num_threads; i++) {
 250                 void *h;
 251                 pthread_join(threads[i], &h);
 252                 hit |= (int) (intptr_t) h;
 253         }
 254
 255         free(threads);
 256
 257         pthread_mutex_destroy(&grep_mutex);
 258         pthread_mutex_destroy(&grep_attr_mutex);
 259         pthread_cond_destroy(&cond_add);
 260         pthread_cond_destroy(&cond_write);
 261         pthread_cond_destroy(&cond_result);
 262         grep_use_locks = 0;
 263         disable_obj_read_lock();
 264
 265         return hit;
 266 }
 267
 268 static int grep_cmd_config(const char *var, const char *value, void *cb)
 269 {
 270         int st = grep_config(var, value, cb);
 271         if (git_color_default_config(var, value, cb) < 0)
 272                 st = -1;
 273
 274         if (!strcmp(var, "grep.threads")) {
 275                 num_threads = git_config_int(var, value);
 276                 if (num_threads < 0)
 277                         die(_("invalid number of threads specified (%d) for %s"),
 278                             num_threads, var);
 279                 else if (!HAVE_THREADS && num_threads > 1) {
 280                         /*
 281                          * TRANSLATORS: %s is the configuration
 282                          * variable for tweaking threads, currently
 283                          * grep.threads
 284                          */
 285                         warning(_("no threads support, ignoring %s"), var);
 286                         num_threads = 1;
 287                 }
 288         }
 289
 290         if (!strcmp(var, "submodule.recurse"))
 291                 recurse_submodules = git_config_bool(var, value);
 292
 293         return st;
 294 }
 295
 296 static void grep_source_name(struct grep_opt *opt, const char *filename,
 297                              int tree_name_len, struct strbuf *out)
 298 {
 299         strbuf_reset(out);
 300
 301         if (opt->null_following_name) {
 302                 if (opt->relative && opt->prefix_length) {
 303                         struct strbuf rel_buf = STRBUF_INIT;
 304                         const char *rel_name =
 305                                 relative_path(filename + tree_name_len,
 306                                               opt->prefix, &rel_buf);
 307
 308                         if (tree_name_len)
 309                                 strbuf_add(out, filename, tree_name_len);
 310
 311                         strbuf_addstr(out, rel_name);
 312                         strbuf_release(&rel_buf);
 313                 } else {
 314                         strbuf_addstr(out, filename);
 315                 }
 316                 return;
 317         }
 318
 319         if (opt->relative && opt->prefix_length)
 320                 quote_path(filename + tree_name_len, opt->prefix, out, 0);
 321         else
 322                 quote_c_style(filename + tree_name_len, out, NULL, 0);
 323
 324         if (tree_name_len)
 325                 strbuf_insert(out, 0, filename, tree_name_len);
 326 }
 327
 328 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
 329                      const char *filename, int tree_name_len,
 330                      const char *path)
 331 {
 332         struct strbuf pathbuf = STRBUF_INIT;
 333         struct grep_source gs;
 334
 335         grep_source_name(opt, filename, tree_name_len, &pathbuf);
 336         grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
 337         strbuf_release(&pathbuf);
 338
 339         if (num_threads > 1) {
 340                 /*
 341                  * add_work() copies gs and thus assumes ownership of
 342                  * its fields, so do not call grep_source_clear()
 343                  */
 344                 add_work(opt, &gs);
 345                 return 0;
 346         } else {
 347                 int hit;
 348
 349                 hit = grep_source(opt, &gs);
 350
 351                 grep_source_clear(&gs);
 352                 return hit;
 353         }
 354 }
 355
 356 static int grep_file(struct grep_opt *opt, const char *filename)
 357 {
 358         struct strbuf buf = STRBUF_INIT;
 359         struct grep_source gs;
 360
 361         grep_source_name(opt, filename, 0, &buf);
 362         grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 363         strbuf_release(&buf);
 364
 365         if (num_threads > 1) {
 366                 /*
 367                  * add_work() copies gs and thus assumes ownership of
 368                  * its fields, so do not call grep_source_clear()
 369                  */
 370                 add_work(opt, &gs);
 371                 return 0;
 372         } else {
 373                 int hit;
 374
 375                 hit = grep_source(opt, &gs);
 376
 377                 grep_source_clear(&gs);
 378                 return hit;
 379         }
 380 }
 381
 382 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 383 {
 384         struct string_list *path_list = opt->output_priv;
 385
 386         if (len == 1 && *(const char *)data == '\0')
 387                 return;
 388         string_list_append(path_list, xstrndup(data, len));
 389 }
 390
 391 static void run_pager(struct grep_opt *opt, const char *prefix)
 392 {
 393         struct string_list *path_list = opt->output_priv;
 394         struct child_process child = CHILD_PROCESS_INIT;
 395         int i, status;
 396
 397         for (i = 0; i < path_list->nr; i++)
 398                 strvec_push(&child.args, path_list->items[i].string);
 399         child.dir = prefix;
 400         child.use_shell = 1;
 401
 402         status = run_command(&child);
 403         if (status)
 404                 exit(status);
 405 }
 406
 407 static int grep_cache(struct grep_opt *opt,
 408                       const struct pathspec *pathspec, int cached);
 409 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 410                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 411                      int check_attr);
 412
 413 static int grep_submodule(struct grep_opt *opt,
 414                           const struct pathspec *pathspec,
 415                           const struct object_id *oid,
 416                           const char *filename, const char *path, int cached)
 417 {
 418         struct repository subrepo;
 419         struct repository *superproject = opt->repo;
 420         const struct submodule *sub;
 421         struct grep_opt subopt;
 422         int hit;
 423
 424         sub = submodule_from_path(superproject, &null_oid, path);
 425
 426         if (!is_submodule_active(superproject, path))
 427                 return 0;
 428
 429         if (repo_submodule_init(&subrepo, superproject, sub))
 430                 return 0;
 431
 432         /*
 433          * NEEDSWORK: repo_read_gitmodules() might call
 434          * add_to_alternates_memory() via config_from_gitmodules(). This
 435          * operation causes a race condition with concurrent object readings
 436          * performed by the worker threads. That's why we need obj_read_lock()
 437          * here. It should be removed once it's no longer necessary to add the
 438          * subrepo's odbs to the in-memory alternates list.
 439          */
 440         obj_read_lock();
 441         repo_read_gitmodules(&subrepo, 0);
 442
 443         /*
 444          * NEEDSWORK: This adds the submodule's object directory to the list of
 445          * alternates for the single in-memory object store.  This has some bad
 446          * consequences for memory (processed objects will never be freed) and
 447          * performance (this increases the number of pack files git has to pay
 448          * attention to, to the sum of the number of pack files in all the
 449          * repositories processed so far).  This can be removed once the object
 450          * store is no longer global and instead is a member of the repository
 451          * object.
 452          */
 453         add_to_alternates_memory(subrepo.objects->odb->path);
 454         obj_read_unlock();
 455
 456         memcpy(&subopt, opt, sizeof(subopt));
 457         subopt.repo = &subrepo;
 458
 459         if (oid) {
 460                 struct object *object;
 461                 struct tree_desc tree;
 462                 void *data;
 463                 unsigned long size;
 464                 struct strbuf base = STRBUF_INIT;
 465
 466                 obj_read_lock();
 467                 object = parse_object_or_die(oid, NULL);
 468                 obj_read_unlock();
 469                 data = read_object_with_reference(&subrepo,
 470                                                   &object->oid, tree_type,
 471                                                   &size, NULL);
 472                 if (!data)
 473                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
 474
 475                 strbuf_addstr(&base, filename);
 476                 strbuf_addch(&base, '/');
 477
 478                 init_tree_desc(&tree, data, size);
 479                 hit = grep_tree(&subopt, pathspec, &tree, &base, base.len,
 480                                 object->type == OBJ_COMMIT);
 481                 strbuf_release(&base);
 482                 free(data);
 483         } else {
 484                 hit = grep_cache(&subopt, pathspec, cached);
 485         }
 486
 487         repo_clear(&subrepo);
 488         return hit;
 489 }
 490
 491 static int grep_cache(struct grep_opt *opt,
 492                       const struct pathspec *pathspec, int cached)
 493 {
 494         struct repository *repo = opt->repo;
 495         int hit = 0;
 496         int nr;
 497         struct strbuf name = STRBUF_INIT;
 498         int name_base_len = 0;
 499         if (repo->submodule_prefix) {
 500                 name_base_len = strlen(repo->submodule_prefix);
 501                 strbuf_addstr(&name, repo->submodule_prefix);
 502         }
 503
 504         if (repo_read_index(repo) < 0)
 505                 die(_("index file corrupt"));
 506
 507         for (nr = 0; nr < repo->index->cache_nr; nr++) {
 508                 const struct cache_entry *ce = repo->index->cache[nr];
 509                 strbuf_setlen(&name, name_base_len);
 510                 strbuf_addstr(&name, ce->name);
 511
 512                 if (S_ISREG(ce->ce_mode) &&
 513                     match_pathspec(repo->index, pathspec, name.buf, name.len, 0, NULL,
 514                                    S_ISDIR(ce->ce_mode) ||
 515                                    S_ISGITLINK(ce->ce_mode))) {
 516                         /*
 517                          * If CE_VALID is on, we assume worktree file and its
 518                          * cache entry are identical, even if worktree file has
 519                          * been modified, so use cache version instead
 520                          */
 521                         if (cached || (ce->ce_flags & CE_VALID) ||
 522                             ce_skip_worktree(ce)) {
 523                                 if (ce_stage(ce) || ce_intent_to_add(ce))
 524                                         continue;
 525                                 hit |= grep_oid(opt, &ce->oid, name.buf,
 526                                                  0, name.buf);
 527                         } else {
 528                                 hit |= grep_file(opt, name.buf);
 529                         }
 530                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 531                            submodule_path_match(repo->index, pathspec, name.buf, NULL)) {
 532                         hit |= grep_submodule(opt, pathspec, NULL, ce->name,
 533                                               ce->name, cached);
 534                 } else {
 535                         continue;
 536                 }
 537
 538                 if (ce_stage(ce)) {
 539                         do {
 540                                 nr++;
 541                         } while (nr < repo->index->cache_nr &&
 542                                  !strcmp(ce->name, repo->index->cache[nr]->name));
 543                         nr--; /* compensate for loop control */
 544                 }
 545                 if (hit && opt->status_only)
 546                         break;
 547         }
 548
 549         strbuf_release(&name);
 550         return hit;
 551 }
 552
 553 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 554                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 555                      int check_attr)
 556 {
 557         struct repository *repo = opt->repo;
 558         int hit = 0;
 559         enum interesting match = entry_not_interesting;
 560         struct name_entry entry;
 561         int old_baselen = base->len;
 562         struct strbuf name = STRBUF_INIT;
 563         int name_base_len = 0;
 564         if (repo->submodule_prefix) {
 565                 strbuf_addstr(&name, repo->submodule_prefix);
 566                 name_base_len = name.len;
 567         }
 568
 569         while (tree_entry(tree, &entry)) {
 570                 int te_len = tree_entry_len(&entry);
 571
 572                 if (match != all_entries_interesting) {
 573                         strbuf_addstr(&name, base->buf + tn_len);
 574                         match = tree_entry_interesting(repo->index,
 575                                                        &entry, &name,
 576                                                        0, pathspec);
 577                         strbuf_setlen(&name, name_base_len);
 578
 579                         if (match == all_entries_not_interesting)
 580                                 break;
 581                         if (match == entry_not_interesting)
 582                                 continue;
 583                 }
 584
 585                 strbuf_add(base, entry.path, te_len);
 586
 587                 if (S_ISREG(entry.mode)) {
 588                         hit |= grep_oid(opt, &entry.oid, base->buf, tn_len,
 589                                          check_attr ? base->buf + tn_len : NULL);
 590                 } else if (S_ISDIR(entry.mode)) {
 591                         enum object_type type;
 592                         struct tree_desc sub;
 593                         void *data;
 594                         unsigned long size;
 595
 596                         data = read_object_file(&entry.oid, &type, &size);
 597                         if (!data)
 598                                 die(_("unable to read tree (%s)"),
 599                                     oid_to_hex(&entry.oid));
 600
 601                         strbuf_addch(base, '/');
 602                         init_tree_desc(&sub, data, size);
 603                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 604                                          check_attr);
 605                         free(data);
 606                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 607                         hit |= grep_submodule(opt, pathspec, &entry.oid,
 608                                               base->buf, base->buf + tn_len,
 609                                               1); /* ignored */
 610                 }
 611
 612                 strbuf_setlen(base, old_baselen);
 613
 614                 if (hit && opt->status_only)
 615                         break;
 616         }
 617
 618         strbuf_release(&name);
 619         return hit;
 620 }
 621
 622 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 623                        struct object *obj, const char *name, const char *path)
 624 {
 625         if (obj->type == OBJ_BLOB)
 626                 return grep_oid(opt, &obj->oid, name, 0, path);
 627         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 628                 struct tree_desc tree;
 629                 void *data;
 630                 unsigned long size;
 631                 struct strbuf base;
 632                 int hit, len;
 633
 634                 data = read_object_with_reference(opt->repo,
 635                                                   &obj->oid, tree_type,
 636                                                   &size, NULL);
 637                 if (!data)
 638                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 639
 640                 len = name ? strlen(name) : 0;
 641                 strbuf_init(&base, PATH_MAX + len + 1);
 642                 if (len) {
 643                         strbuf_add(&base, name, len);
 644                         strbuf_addch(&base, ':');
 645                 }
 646                 init_tree_desc(&tree, data, size);
 647                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 648                                 obj->type == OBJ_COMMIT);
 649                 strbuf_release(&base);
 650                 free(data);
 651                 return hit;
 652         }
 653         die(_("unable to grep from object of type %s"), type_name(obj->type));
 654 }
 655
 656 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 657                         const struct object_array *list)
 658 {
 659         unsigned int i;
 660         int hit = 0;
 661         const unsigned int nr = list->nr;
 662
 663         for (i = 0; i < nr; i++) {
 664                 struct object *real_obj;
 665
 666                 obj_read_lock();
 667                 real_obj = deref_tag(opt->repo, list->objects[i].item,
 668                                      NULL, 0);
 669                 obj_read_unlock();
 670
 671                 if (!real_obj) {
 672                         char hex[GIT_MAX_HEXSZ + 1];
 673                         const char *name = list->objects[i].name;
 674
 675                         if (!name) {
 676                                 oid_to_hex_r(hex, &list->objects[i].item->oid);
 677                                 name = hex;
 678                         }
 679                         die(_("invalid object '%s' given."), name);
 680                 }
 681
 682                 /* load the gitmodules file for this rev */
 683                 if (recurse_submodules) {
 684                         submodule_free(opt->repo);
 685                         obj_read_lock();
 686                         gitmodules_config_oid(&real_obj->oid);
 687                         obj_read_unlock();
 688                 }
 689                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
 690                                 list->objects[i].path)) {
 691                         hit = 1;
 692                         if (opt->status_only)
 693                                 break;
 694                 }
 695         }
 696         return hit;
 697 }
 698
 699 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 700                           int exc_std, int use_index)
 701 {
 702         struct dir_struct dir;
 703         int i, hit = 0;
 704
 705         dir_init(&dir);
 706         if (!use_index)
 707                 dir.flags |= DIR_NO_GITLINKS;
 708         if (exc_std)
 709                 setup_standard_excludes(&dir);
 710
 711         fill_directory(&dir, opt->repo->index, pathspec);
 712         for (i = 0; i < dir.nr; i++) {
 713                 hit |= grep_file(opt, dir.entries[i]->name);
 714                 if (hit && opt->status_only)
 715                         break;
 716         }
 717         dir_clear(&dir);
 718         return hit;
 719 }
 720
 721 static int context_callback(const struct option *opt, const char *arg,
 722                             int unset)
 723 {
 724         struct grep_opt *grep_opt = opt->value;
 725         int value;
 726         const char *endp;
 727
 728         if (unset) {
 729                 grep_opt->pre_context = grep_opt->post_context = 0;
 730                 return 0;
 731         }
 732         value = strtol(arg, (char **)&endp, 10);
 733         if (*endp) {
 734                 return error(_("switch `%c' expects a numerical value"),
 735                              opt->short_name);
 736         }
 737         grep_opt->pre_context = grep_opt->post_context = value;
 738         return 0;
 739 }
 740
 741 static int file_callback(const struct option *opt, const char *arg, int unset)
 742 {
 743         struct grep_opt *grep_opt = opt->value;
 744         int from_stdin;
 745         FILE *patterns;
 746         int lno = 0;
 747         struct strbuf sb = STRBUF_INIT;
 748
 749         BUG_ON_OPT_NEG(unset);
 750
 751         from_stdin = !strcmp(arg, "-");
 752         patterns = from_stdin ? stdin : fopen(arg, "r");
 753         if (!patterns)
 754                 die_errno(_("cannot open '%s'"), arg);
 755         while (strbuf_getline(&sb, patterns) == 0) {
 756                 /* ignore empty line like grep does */
 757                 if (sb.len == 0)
 758                         continue;
 759
 760                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 761                                 GREP_PATTERN);
 762         }
 763         if (!from_stdin)
 764                 fclose(patterns);
 765         strbuf_release(&sb);
 766         return 0;
 767 }
 768
 769 static int not_callback(const struct option *opt, const char *arg, int unset)
 770 {
 771         struct grep_opt *grep_opt = opt->value;
 772         BUG_ON_OPT_NEG(unset);
 773         BUG_ON_OPT_ARG(arg);
 774         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 775         return 0;
 776 }
 777
 778 static int and_callback(const struct option *opt, const char *arg, int unset)
 779 {
 780         struct grep_opt *grep_opt = opt->value;
 781         BUG_ON_OPT_NEG(unset);
 782         BUG_ON_OPT_ARG(arg);
 783         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 784         return 0;
 785 }
 786
 787 static int open_callback(const struct option *opt, const char *arg, int unset)
 788 {
 789         struct grep_opt *grep_opt = opt->value;
 790         BUG_ON_OPT_NEG(unset);
 791         BUG_ON_OPT_ARG(arg);
 792         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 793         return 0;
 794 }
 795
 796 static int close_callback(const struct option *opt, const char *arg, int unset)
 797 {
 798         struct grep_opt *grep_opt = opt->value;
 799         BUG_ON_OPT_NEG(unset);
 800         BUG_ON_OPT_ARG(arg);
 801         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 802         return 0;
 803 }
 804
 805 static int pattern_callback(const struct option *opt, const char *arg,
 806                             int unset)
 807 {
 808         struct grep_opt *grep_opt = opt->value;
 809         BUG_ON_OPT_NEG(unset);
 810         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 811         return 0;
 812 }
 813
 814 int cmd_grep(int argc, const char **argv, const char *prefix)
 815 {
 816         int hit = 0;
 817         int cached = 0, untracked = 0, opt_exclude = -1;
 818         int seen_dashdash = 0;
 819         int external_grep_allowed__ignored;
 820         const char *show_in_pager = NULL, *default_pager = "dummy";
 821         struct grep_opt opt;
 822         struct object_array list = OBJECT_ARRAY_INIT;
 823         struct pathspec pathspec;
 824         struct string_list path_list = STRING_LIST_INIT_NODUP;
 825         int i;
 826         int dummy;
 827         int use_index = 1;
 828         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 829         int allow_revs;
 830
 831         struct option options[] = {
 832                 OPT_BOOL(0, "cached", &cached,
 833                         N_("search in index instead of in the work tree")),
 834                 OPT_NEGBIT(0, "no-index", &use_index,
 835                          N_("find in contents not managed by git"), 1),
 836                 OPT_BOOL(0, "untracked", &untracked,
 837                         N_("search in both tracked and untracked files")),
 838                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 839                             N_("ignore files specified via '.gitignore'"), 1),
 840                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 841                          N_("recursively search in each submodule")),
 842                 OPT_GROUP(""),
 843                 OPT_BOOL('v', "invert-match", &opt.invert,
 844                         N_("show non-matching lines")),
 845                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 846                         N_("case insensitive matching")),
 847                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 848                         N_("match patterns only at word boundaries")),
 849                 OPT_SET_INT('a', "text", &opt.binary,
 850                         N_("process binary files as text"), GREP_BINARY_TEXT),
 851                 OPT_SET_INT('I', NULL, &opt.binary,
 852                         N_("don't match patterns in binary files"),
 853                         GREP_BINARY_NOMATCH),
 854                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
 855                          N_("process binary files with textconv filters")),
 856                 OPT_SET_INT('r', "recursive", &opt.max_depth,
 857                             N_("search in subdirectories (default)"), -1),
 858                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
 859                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
 860                         NULL, 1 },
 861                 OPT_GROUP(""),
 862                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
 863                             N_("use extended POSIX regular expressions"),
 864                             GREP_PATTERN_TYPE_ERE),
 865                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
 866                             N_("use basic POSIX regular expressions (default)"),
 867                             GREP_PATTERN_TYPE_BRE),
 868                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
 869                             N_("interpret patterns as fixed strings"),
 870                             GREP_PATTERN_TYPE_FIXED),
 871                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
 872                             N_("use Perl-compatible regular expressions"),
 873                             GREP_PATTERN_TYPE_PCRE),
 874                 OPT_GROUP(""),
 875                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
 876                 OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")),
 877                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
 878                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
 879                 OPT_NEGBIT(0, "full-name", &opt.relative,
 880                         N_("show filenames relative to top directory"), 1),
 881                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
 882                         N_("show only filenames instead of matching lines")),
 883                 OPT_BOOL(0, "name-only", &opt.name_only,
 884                         N_("synonym for --files-with-matches")),
 885                 OPT_BOOL('L', "files-without-match",
 886                         &opt.unmatch_name_only,
 887                         N_("show only the names of files without match")),
 888                 OPT_BOOL_F('z', "null", &opt.null_following_name,
 889                            N_("print NUL after filenames"),
 890                            PARSE_OPT_NOCOMPLETE),
 891                 OPT_BOOL('o', "only-matching", &opt.only_matching,
 892                         N_("show only matching parts of a line")),
 893                 OPT_BOOL('c', "count", &opt.count,
 894                         N_("show the number of matches instead of matching lines")),
 895                 OPT__COLOR(&opt.color, N_("highlight matches")),
 896                 OPT_BOOL(0, "break", &opt.file_break,
 897                         N_("print empty line between matches from different files")),
 898                 OPT_BOOL(0, "heading", &opt.heading,
 899                         N_("show filename only once above matches from same file")),
 900                 OPT_GROUP(""),
 901                 OPT_CALLBACK('C', "context", &opt, N_("n"),
 902                         N_("show <n> context lines before and after matches"),
 903                         context_callback),
 904                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 905                         N_("show <n> context lines before matches")),
 906                 OPT_INTEGER('A', "after-context", &opt.post_context,
 907                         N_("show <n> context lines after matches")),
 908                 OPT_INTEGER(0, "threads", &num_threads,
 909                         N_("use <n> worker threads")),
 910                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
 911                         context_callback),
 912                 OPT_BOOL('p', "show-function", &opt.funcname,
 913                         N_("show a line with the function name before matches")),
 914                 OPT_BOOL('W', "function-context", &opt.funcbody,
 915                         N_("show the surrounding function")),
 916                 OPT_GROUP(""),
 917                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
 918                         N_("read patterns from file"), file_callback),
 919                 OPT_CALLBACK_F('e', NULL, &opt, N_("pattern"),
 920                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback),
 921                 OPT_CALLBACK_F(0, "and", &opt, NULL,
 922                         N_("combine patterns specified with -e"),
 923                         PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback),
 924                 OPT_BOOL(0, "or", &dummy, ""),
 925                 OPT_CALLBACK_F(0, "not", &opt, NULL, "",
 926                         PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback),
 927                 OPT_CALLBACK_F('(', NULL, &opt, NULL, "",
 928                         PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 929                         open_callback),
 930                 OPT_CALLBACK_F(')', NULL, &opt, NULL, "",
 931                         PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 932                         close_callback),
 933                 OPT__QUIET(&opt.status_only,
 934                            N_("indicate hit with exit status without output")),
 935                 OPT_BOOL(0, "all-match", &opt.all_match,
 936                         N_("show only matches from files that match all patterns")),
 937                 OPT_GROUP(""),
 938                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 939                         N_("pager"), N_("show matching files in the pager"),
 940                         PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
 941                         NULL, (intptr_t)default_pager },
 942                 OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
 943                            N_("allow calling of grep(1) (ignored by this build)"),
 944                            PARSE_OPT_NOCOMPLETE),
 945                 OPT_END()
 946         };
 947
 948         git_config(grep_cmd_config, NULL);
 949         grep_init(&opt, the_repository, prefix);
 950
 951         /*
 952          * If there is no -- then the paths must exist in the working
 953          * tree.  If there is no explicit pattern specified with -e or
 954          * -f, we take the first unrecognized non option to be the
 955          * pattern, but then what follows it must be zero or more
 956          * valid refs up to the -- (if exists), and then existing
 957          * paths.  If there is an explicit pattern, then the first
 958          * unrecognized non option is the beginning of the refs list
 959          * that continues up to the -- (if exists), and then paths.
 960          */
 961         argc = parse_options(argc, argv, prefix, options, grep_usage,
 962                              PARSE_OPT_KEEP_DASHDASH |
 963                              PARSE_OPT_STOP_AT_NON_OPTION);
 964         grep_commit_pattern_type(pattern_type_arg, &opt);
 965
 966         if (use_index && !startup_info->have_repository) {
 967                 int fallback = 0;
 968                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
 969                 if (fallback)
 970                         use_index = 0;
 971                 else
 972                         /* die the same way as if we did it at the beginning */
 973                         setup_git_directory();
 974         }
 975         /* Ignore --recurse-submodules if --no-index is given or implied */
 976         if (!use_index)
 977                 recurse_submodules = 0;
 978
 979         /*
 980          * skip a -- separator; we know it cannot be
 981          * separating revisions from pathnames if
 982          * we haven't even had any patterns yet
 983          */
 984         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 985                 argv++;
 986                 argc--;
 987         }
 988
 989         /* First unrecognized non-option token */
 990         if (argc > 0 && !opt.pattern_list) {
 991                 append_grep_pattern(&opt, argv[0], "command line", 0,
 992                                     GREP_PATTERN);
 993                 argv++;
 994                 argc--;
 995         }
 996
 997         if (show_in_pager == default_pager)
 998                 show_in_pager = git_pager(1);
 999         if (show_in_pager) {
1000                 opt.color = 0;
1001                 opt.name_only = 1;
1002                 opt.null_following_name = 1;
1003                 opt.output_priv = &path_list;
1004                 opt.output = append_path;
1005                 string_list_append(&path_list, show_in_pager);
1006         }
1007
1008         if (!opt.pattern_list)
1009                 die(_("no pattern given"));
1010
1011         /* --only-matching has no effect with --invert. */
1012         if (opt.invert)
1013                 opt.only_matching = 0;
1014
1015         /*
1016          * We have to find "--" in a separate pass, because its presence
1017          * influences how we will parse arguments that come before it.
1018          */
1019         for (i = 0; i < argc; i++) {
1020                 if (!strcmp(argv[i], "--")) {
1021                         seen_dashdash = 1;
1022                         break;
1023                 }
1024         }
1025
1026         /*
1027          * Resolve any rev arguments. If we have a dashdash, then everything up
1028          * to it must resolve as a rev. If not, then we stop at the first
1029          * non-rev and assume everything else is a path.
1030          */
1031         allow_revs = use_index && !untracked;
1032         for (i = 0; i < argc; i++) {
1033                 const char *arg = argv[i];
1034                 struct object_id oid;
1035                 struct object_context oc;
1036                 struct object *object;
1037
1038                 if (!strcmp(arg, "--")) {
1039                         i++;
1040                         break;
1041                 }
1042
1043                 if (!allow_revs) {
1044                         if (seen_dashdash)
1045                                 die(_("--no-index or --untracked cannot be used with revs"));
1046                         break;
1047                 }
1048
1049                 if (get_oid_with_context(the_repository, arg,
1050                                          GET_OID_RECORD_PATH,
1051                                          &oid, &oc)) {
1052                         if (seen_dashdash)
1053                                 die(_("unable to resolve revision: %s"), arg);
1054                         break;
1055                 }
1056
1057                 object = parse_object_or_die(&oid, arg);
1058                 if (!seen_dashdash)
1059                         verify_non_filename(prefix, arg);
1060                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1061                 free(oc.path);
1062         }
1063
1064         /*
1065          * Anything left over is presumed to be a path. But in the non-dashdash
1066          * "do what I mean" case, we verify and complain when that isn't true.
1067          */
1068         if (!seen_dashdash) {
1069                 int j;
1070                 for (j = i; j < argc; j++)
1071                         verify_filename(prefix, argv[j], j == i && allow_revs);
1072         }
1073
1074         parse_pathspec(&pathspec, 0,
1075                        PATHSPEC_PREFER_CWD |
1076                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1077                        prefix, argv + i);
1078         pathspec.max_depth = opt.max_depth;
1079         pathspec.recursive = 1;
1080         pathspec.recurse_submodules = !!recurse_submodules;
1081
1082         if (recurse_submodules && untracked)
1083                 die(_("--untracked not supported with --recurse-submodules"));
1084
1085         if (show_in_pager) {
1086                 if (num_threads > 1)
1087                         warning(_("invalid option combination, ignoring --threads"));
1088                 num_threads = 1;
1089         } else if (!HAVE_THREADS && num_threads > 1) {
1090                 warning(_("no threads support, ignoring --threads"));
1091                 num_threads = 1;
1092         } else if (num_threads < 0)
1093                 die(_("invalid number of threads specified (%d)"), num_threads);
1094         else if (num_threads == 0)
1095                 num_threads = HAVE_THREADS ? online_cpus() : 1;
1096
1097         if (num_threads > 1) {
1098                 if (!HAVE_THREADS)
1099                         BUG("Somebody got num_threads calculation wrong!");
1100                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1101                     && (opt.pre_context || opt.post_context ||
1102                         opt.file_break || opt.funcbody))
1103                         skip_first_line = 1;
1104
1105                 /*
1106                  * Pre-read gitmodules (if not read already) and force eager
1107                  * initialization of packed_git to prevent racy lazy
1108                  * reading/initialization once worker threads are started.
1109                  */
1110                 if (recurse_submodules)
1111                         repo_read_gitmodules(the_repository, 1);
1112                 if (startup_info->have_repository)
1113                         (void)get_packed_git(the_repository);
1114
1115                 start_threads(&opt);
1116         } else {
1117                 /*
1118                  * The compiled patterns on the main path are only
1119                  * used when not using threading. Otherwise
1120                  * start_threads() above calls compile_grep_patterns()
1121                  * for each thread.
1122                  */
1123                 compile_grep_patterns(&opt);
1124         }
1125
1126         if (show_in_pager && (cached || list.nr))
1127                 die(_("--open-files-in-pager only works on the worktree"));
1128
1129         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1130                 const char *pager = path_list.items[0].string;
1131                 int len = strlen(pager);
1132
1133                 if (len > 4 && is_dir_sep(pager[len - 5]))
1134                         pager += len - 4;
1135
1136                 if (opt.ignore_case && !strcmp("less", pager))
1137                         string_list_append(&path_list, "-I");
1138
1139                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1140                         struct strbuf buf = STRBUF_INIT;
1141                         strbuf_addf(&buf, "+/%s%s",
1142                                         strcmp("less", pager) ? "" : "*",
1143                                         opt.pattern_list->pattern);
1144                         string_list_append(&path_list,
1145                                            strbuf_detach(&buf, NULL));
1146                 }
1147         }
1148
1149         if (!show_in_pager && !opt.status_only)
1150                 setup_pager();
1151
1152         if (!use_index && (untracked || cached))
1153                 die(_("--cached or --untracked cannot be used with --no-index"));
1154
1155         if (!use_index || untracked) {
1156                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1157                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1158         } else if (0 <= opt_exclude) {
1159                 die(_("--[no-]exclude-standard cannot be used for tracked contents"));
1160         } else if (!list.nr) {
1161                 if (!cached)
1162                         setup_work_tree();
1163
1164                 hit = grep_cache(&opt, &pathspec, cached);
1165         } else {
1166                 if (cached)
1167                         die(_("both --cached and trees are given"));
1168
1169                 hit = grep_objects(&opt, &pathspec, &list);
1170         }
1171
1172         if (num_threads > 1)
1173                 hit |= wait_all();
1174         if (hit && show_in_pager)
1175                 run_pager(&opt, prefix);
1176         clear_pathspec(&pathspec);
1177         free_grep_patterns(&opt);
1178         grep_destroy();
1179         return !hit;
1180 }