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

   1 /*
   2  * Builtin "git grep"
   3  *
   4  * Copyright (c) 2006 Junio C Hamano
   5  */
   6 #include "cache.h"
   7 #include "repository.h"
   8 #include "config.h"
   9 #include "blob.h"
  10 #include "tree.h"
  11 #include "commit.h"
  12 #include "tag.h"
  13 #include "tree-walk.h"
  14 #include "builtin.h"
  15 #include "parse-options.h"
  16 #include "string-list.h"
  17 #include "run-command.h"
  18 #include "userdiff.h"
  19 #include "grep.h"
  20 #include "quote.h"
  21 #include "dir.h"
  22 #include "pathspec.h"
  23 #include "submodule.h"
  24 #include "submodule-config.h"
  25
  26 static char const * const grep_usage[] = {
  27         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  28         NULL
  29 };
  30
  31 static int recurse_submodules;
  32
  33 #define GREP_NUM_THREADS_DEFAULT 8
  34 static int num_threads;
  35
  36 #ifndef NO_PTHREADS
  37 static pthread_t *threads;
  38
  39 /* We use one producer thread and THREADS consumer
  40  * threads. The producer adds struct work_items to 'todo' and the
  41  * consumers pick work items from the same array.
  42  */
  43 struct work_item {
  44         struct grep_source source;
  45         char done;
  46         struct strbuf out;
  47 };
  48
  49 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  50  * that have been or are processed by a consumer thread. We haven't
  51  * written the result for these to stdout yet.
  52  *
  53  * The work_items in [todo_start, todo_end) are waiting to be picked
  54  * up by a consumer thread.
  55  *
  56  * The ranges are modulo TODO_SIZE.
  57  */
  58 #define TODO_SIZE 128
  59 static struct work_item todo[TODO_SIZE];
  60 static int todo_start;
  61 static int todo_end;
  62 static int todo_done;
  63
  64 /* Has all work items been added? */
  65 static int all_work_added;
  66
  67 /* This lock protects all the variables above. */
  68 static pthread_mutex_t grep_mutex;
  69
  70 static inline void grep_lock(void)
  71 {
  72         assert(num_threads);
  73         pthread_mutex_lock(&grep_mutex);
  74 }
  75
  76 static inline void grep_unlock(void)
  77 {
  78         assert(num_threads);
  79         pthread_mutex_unlock(&grep_mutex);
  80 }
  81
  82 /* Signalled when a new work_item is added to todo. */
  83 static pthread_cond_t cond_add;
  84
  85 /* Signalled when the result from one work_item is written to
  86  * stdout.
  87  */
  88 static pthread_cond_t cond_write;
  89
  90 /* Signalled when we are finished with everything. */
  91 static pthread_cond_t cond_result;
  92
  93 static int skip_first_line;
  94
  95 static void add_work(struct grep_opt *opt, enum grep_source_type type,
  96                      const char *name, const char *path, const void *id)
  97 {
  98         grep_lock();
  99
 100         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 101                 pthread_cond_wait(&cond_write, &grep_mutex);
 102         }
 103
 104         grep_source_init(&todo[todo_end].source, type, name, path, id);
 105         if (opt->binary != GREP_BINARY_TEXT)
 106                 grep_source_load_driver(&todo[todo_end].source);
 107         todo[todo_end].done = 0;
 108         strbuf_reset(&todo[todo_end].out);
 109         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 110
 111         pthread_cond_signal(&cond_add);
 112         grep_unlock();
 113 }
 114
 115 static struct work_item *get_work(void)
 116 {
 117         struct work_item *ret;
 118
 119         grep_lock();
 120         while (todo_start == todo_end && !all_work_added) {
 121                 pthread_cond_wait(&cond_add, &grep_mutex);
 122         }
 123
 124         if (todo_start == todo_end && all_work_added) {
 125                 ret = NULL;
 126         } else {
 127                 ret = &todo[todo_start];
 128                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 129         }
 130         grep_unlock();
 131         return ret;
 132 }
 133
 134 static void work_done(struct work_item *w)
 135 {
 136         int old_done;
 137
 138         grep_lock();
 139         w->done = 1;
 140         old_done = todo_done;
 141         for(; todo[todo_done].done && todo_done != todo_start;
 142             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 143                 w = &todo[todo_done];
 144                 if (w->out.len) {
 145                         const char *p = w->out.buf;
 146                         size_t len = w->out.len;
 147
 148                         /* Skip the leading hunk mark of the first file. */
 149                         if (skip_first_line) {
 150                                 while (len) {
 151                                         len--;
 152                                         if (*p++ == '\n')
 153                                                 break;
 154                                 }
 155                                 skip_first_line = 0;
 156                         }
 157
 158                         write_or_die(1, p, len);
 159                 }
 160                 grep_source_clear(&w->source);
 161         }
 162
 163         if (old_done != todo_done)
 164                 pthread_cond_signal(&cond_write);
 165
 166         if (all_work_added && todo_done == todo_end)
 167                 pthread_cond_signal(&cond_result);
 168
 169         grep_unlock();
 170 }
 171
 172 static void *run(void *arg)
 173 {
 174         int hit = 0;
 175         struct grep_opt *opt = arg;
 176
 177         while (1) {
 178                 struct work_item *w = get_work();
 179                 if (!w)
 180                         break;
 181
 182                 opt->output_priv = w;
 183                 hit |= grep_source(opt, &w->source);
 184                 grep_source_clear_data(&w->source);
 185                 work_done(w);
 186         }
 187         free_grep_patterns(arg);
 188         free(arg);
 189
 190         return (void*) (intptr_t) hit;
 191 }
 192
 193 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 194 {
 195         struct work_item *w = opt->output_priv;
 196         strbuf_add(&w->out, buf, size);
 197 }
 198
 199 static void start_threads(struct grep_opt *opt)
 200 {
 201         int i;
 202
 203         pthread_mutex_init(&grep_mutex, NULL);
 204         pthread_mutex_init(&grep_read_mutex, NULL);
 205         pthread_mutex_init(&grep_attr_mutex, NULL);
 206         pthread_cond_init(&cond_add, NULL);
 207         pthread_cond_init(&cond_write, NULL);
 208         pthread_cond_init(&cond_result, NULL);
 209         grep_use_locks = 1;
 210
 211         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 212                 strbuf_init(&todo[i].out, 0);
 213         }
 214
 215         threads = xcalloc(num_threads, sizeof(*threads));
 216         for (i = 0; i < num_threads; i++) {
 217                 int err;
 218                 struct grep_opt *o = grep_opt_dup(opt);
 219                 o->output = strbuf_out;
 220                 if (i)
 221                         o->debug = 0;
 222                 compile_grep_patterns(o);
 223                 err = pthread_create(&threads[i], NULL, run, o);
 224
 225                 if (err)
 226                         die(_("grep: failed to create thread: %s"),
 227                             strerror(err));
 228         }
 229 }
 230
 231 static int wait_all(void)
 232 {
 233         int hit = 0;
 234         int i;
 235
 236         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_read_mutex);
 259         pthread_mutex_destroy(&grep_attr_mutex);
 260         pthread_cond_destroy(&cond_add);
 261         pthread_cond_destroy(&cond_write);
 262         pthread_cond_destroy(&cond_result);
 263         grep_use_locks = 0;
 264
 265         return hit;
 266 }
 267 #else /* !NO_PTHREADS */
 268
 269 static int wait_all(void)
 270 {
 271         return 0;
 272 }
 273 #endif
 274
 275 static int grep_cmd_config(const char *var, const char *value, void *cb)
 276 {
 277         int st = grep_config(var, value, cb);
 278         if (git_color_default_config(var, value, cb) < 0)
 279                 st = -1;
 280
 281         if (!strcmp(var, "grep.threads")) {
 282                 num_threads = git_config_int(var, value);
 283                 if (num_threads < 0)
 284                         die(_("invalid number of threads specified (%d) for %s"),
 285                             num_threads, var);
 286 #ifdef NO_PTHREADS
 287                 else if (num_threads && num_threads != 1) {
 288                         /*
 289                          * TRANSLATORS: %s is the configuration
 290                          * variable for tweaking threads, currently
 291                          * grep.threads
 292                          */
 293                         warning(_("no threads support, ignoring %s"), var);
 294                         num_threads = 0;
 295                 }
 296 #endif
 297         }
 298
 299         if (!strcmp(var, "submodule.recurse"))
 300                 recurse_submodules = git_config_bool(var, value);
 301
 302         return st;
 303 }
 304
 305 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
 306 {
 307         void *data;
 308
 309         grep_read_lock();
 310         data = read_sha1_file(oid->hash, type, size);
 311         grep_read_unlock();
 312         return data;
 313 }
 314
 315 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
 316                      const char *filename, int tree_name_len,
 317                      const char *path)
 318 {
 319         struct strbuf pathbuf = STRBUF_INIT;
 320
 321         if (opt->relative && opt->prefix_length) {
 322                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
 323                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 324         } else {
 325                 strbuf_addstr(&pathbuf, filename);
 326         }
 327
 328 #ifndef NO_PTHREADS
 329         if (num_threads) {
 330                 add_work(opt, GREP_SOURCE_OID, pathbuf.buf, path, oid);
 331                 strbuf_release(&pathbuf);
 332                 return 0;
 333         } else
 334 #endif
 335         {
 336                 struct grep_source gs;
 337                 int hit;
 338
 339                 grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
 340                 strbuf_release(&pathbuf);
 341                 hit = grep_source(opt, &gs);
 342
 343                 grep_source_clear(&gs);
 344                 return hit;
 345         }
 346 }
 347
 348 static int grep_file(struct grep_opt *opt, const char *filename)
 349 {
 350         struct strbuf buf = STRBUF_INIT;
 351
 352         if (opt->relative && opt->prefix_length)
 353                 quote_path_relative(filename, opt->prefix, &buf);
 354         else
 355                 strbuf_addstr(&buf, filename);
 356
 357 #ifndef NO_PTHREADS
 358         if (num_threads) {
 359                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
 360                 strbuf_release(&buf);
 361                 return 0;
 362         } else
 363 #endif
 364         {
 365                 struct grep_source gs;
 366                 int hit;
 367
 368                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 369                 strbuf_release(&buf);
 370                 hit = grep_source(opt, &gs);
 371
 372                 grep_source_clear(&gs);
 373                 return hit;
 374         }
 375 }
 376
 377 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 378 {
 379         struct string_list *path_list = opt->output_priv;
 380
 381         if (len == 1 && *(const char *)data == '\0')
 382                 return;
 383         string_list_append(path_list, xstrndup(data, len));
 384 }
 385
 386 static void run_pager(struct grep_opt *opt, const char *prefix)
 387 {
 388         struct string_list *path_list = opt->output_priv;
 389         struct child_process child = CHILD_PROCESS_INIT;
 390         int i, status;
 391
 392         for (i = 0; i < path_list->nr; i++)
 393                 argv_array_push(&child.args, path_list->items[i].string);
 394         child.dir = prefix;
 395         child.use_shell = 1;
 396
 397         status = run_command(&child);
 398         if (status)
 399                 exit(status);
 400 }
 401
 402 static int grep_cache(struct grep_opt *opt, struct repository *repo,
 403                       const struct pathspec *pathspec, int cached);
 404 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 405                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 406                      int check_attr, struct repository *repo);
 407
 408 static int grep_submodule(struct grep_opt *opt, struct repository *superproject,
 409                           const struct pathspec *pathspec,
 410                           const struct object_id *oid,
 411                           const char *filename, const char *path)
 412 {
 413         struct repository submodule;
 414         int hit;
 415
 416         if (!is_submodule_active(superproject, path))
 417                 return 0;
 418
 419         if (repo_submodule_init(&submodule, superproject, path))
 420                 return 0;
 421
 422         repo_read_gitmodules(&submodule);
 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         grep_read_lock();
 435         add_to_alternates_memory(submodule.objectdir);
 436         grep_read_unlock();
 437
 438         if (oid) {
 439                 struct object *object;
 440                 struct tree_desc tree;
 441                 void *data;
 442                 unsigned long size;
 443                 struct strbuf base = STRBUF_INIT;
 444
 445                 object = parse_object_or_die(oid, oid_to_hex(oid));
 446
 447                 grep_read_lock();
 448                 data = read_object_with_reference(object->oid.hash, tree_type,
 449                                                   &size, NULL);
 450                 grep_read_unlock();
 451
 452                 if (!data)
 453                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
 454
 455                 strbuf_addstr(&base, filename);
 456                 strbuf_addch(&base, '/');
 457
 458                 init_tree_desc(&tree, data, size);
 459                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 460                                 object->type == OBJ_COMMIT, &submodule);
 461                 strbuf_release(&base);
 462                 free(data);
 463         } else {
 464                 hit = grep_cache(opt, &submodule, pathspec, 1);
 465         }
 466
 467         repo_clear(&submodule);
 468         return hit;
 469 }
 470
 471 static int grep_cache(struct grep_opt *opt, struct repository *repo,
 472                       const struct pathspec *pathspec, int cached)
 473 {
 474         int hit = 0;
 475         int nr;
 476         struct strbuf name = STRBUF_INIT;
 477         int name_base_len = 0;
 478         if (repo->submodule_prefix) {
 479                 name_base_len = strlen(repo->submodule_prefix);
 480                 strbuf_addstr(&name, repo->submodule_prefix);
 481         }
 482
 483         repo_read_index(repo);
 484
 485         for (nr = 0; nr < repo->index->cache_nr; nr++) {
 486                 const struct cache_entry *ce = repo->index->cache[nr];
 487                 strbuf_setlen(&name, name_base_len);
 488                 strbuf_addstr(&name, ce->name);
 489
 490                 if (S_ISREG(ce->ce_mode) &&
 491                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
 492                                    S_ISDIR(ce->ce_mode) ||
 493                                    S_ISGITLINK(ce->ce_mode))) {
 494                         /*
 495                          * If CE_VALID is on, we assume worktree file and its
 496                          * cache entry are identical, even if worktree file has
 497                          * been modified, so use cache version instead
 498                          */
 499                         if (cached || (ce->ce_flags & CE_VALID) ||
 500                             ce_skip_worktree(ce)) {
 501                                 if (ce_stage(ce) || ce_intent_to_add(ce))
 502                                         continue;
 503                                 hit |= grep_oid(opt, &ce->oid, name.buf,
 504                                                  0, name.buf);
 505                         } else {
 506                                 hit |= grep_file(opt, name.buf);
 507                         }
 508                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 509                            submodule_path_match(pathspec, name.buf, NULL)) {
 510                         hit |= grep_submodule(opt, repo, pathspec, NULL, ce->name, ce->name);
 511                 } else {
 512                         continue;
 513                 }
 514
 515                 if (ce_stage(ce)) {
 516                         do {
 517                                 nr++;
 518                         } while (nr < repo->index->cache_nr &&
 519                                  !strcmp(ce->name, repo->index->cache[nr]->name));
 520                         nr--; /* compensate for loop control */
 521                 }
 522                 if (hit && opt->status_only)
 523                         break;
 524         }
 525
 526         strbuf_release(&name);
 527         return hit;
 528 }
 529
 530 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 531                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 532                      int check_attr, struct repository *repo)
 533 {
 534         int hit = 0;
 535         enum interesting match = entry_not_interesting;
 536         struct name_entry entry;
 537         int old_baselen = base->len;
 538         struct strbuf name = STRBUF_INIT;
 539         int name_base_len = 0;
 540         if (repo->submodule_prefix) {
 541                 strbuf_addstr(&name, repo->submodule_prefix);
 542                 name_base_len = name.len;
 543         }
 544
 545         while (tree_entry(tree, &entry)) {
 546                 int te_len = tree_entry_len(&entry);
 547
 548                 if (match != all_entries_interesting) {
 549                         strbuf_addstr(&name, base->buf + tn_len);
 550                         match = tree_entry_interesting(&entry, &name,
 551                                                        0, pathspec);
 552                         strbuf_setlen(&name, name_base_len);
 553
 554                         if (match == all_entries_not_interesting)
 555                                 break;
 556                         if (match == entry_not_interesting)
 557                                 continue;
 558                 }
 559
 560                 strbuf_add(base, entry.path, te_len);
 561
 562                 if (S_ISREG(entry.mode)) {
 563                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
 564                                          check_attr ? base->buf + tn_len : NULL);
 565                 } else if (S_ISDIR(entry.mode)) {
 566                         enum object_type type;
 567                         struct tree_desc sub;
 568                         void *data;
 569                         unsigned long size;
 570
 571                         data = lock_and_read_oid_file(entry.oid, &type, &size);
 572                         if (!data)
 573                                 die(_("unable to read tree (%s)"),
 574                                     oid_to_hex(entry.oid));
 575
 576                         strbuf_addch(base, '/');
 577                         init_tree_desc(&sub, data, size);
 578                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 579                                          check_attr, repo);
 580                         free(data);
 581                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 582                         hit |= grep_submodule(opt, repo, pathspec, entry.oid,
 583                                               base->buf, base->buf + tn_len);
 584                 }
 585
 586                 strbuf_setlen(base, old_baselen);
 587
 588                 if (hit && opt->status_only)
 589                         break;
 590         }
 591
 592         strbuf_release(&name);
 593         return hit;
 594 }
 595
 596 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 597                        struct object *obj, const char *name, const char *path,
 598                        struct repository *repo)
 599 {
 600         if (obj->type == OBJ_BLOB)
 601                 return grep_oid(opt, &obj->oid, name, 0, path);
 602         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 603                 struct tree_desc tree;
 604                 void *data;
 605                 unsigned long size;
 606                 struct strbuf base;
 607                 int hit, len;
 608
 609                 grep_read_lock();
 610                 data = read_object_with_reference(obj->oid.hash, tree_type,
 611                                                   &size, NULL);
 612                 grep_read_unlock();
 613
 614                 if (!data)
 615                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 616
 617                 len = name ? strlen(name) : 0;
 618                 strbuf_init(&base, PATH_MAX + len + 1);
 619                 if (len) {
 620                         strbuf_add(&base, name, len);
 621                         strbuf_addch(&base, ':');
 622                 }
 623                 init_tree_desc(&tree, data, size);
 624                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 625                                 obj->type == OBJ_COMMIT, repo);
 626                 strbuf_release(&base);
 627                 free(data);
 628                 return hit;
 629         }
 630         die(_("unable to grep from object of type %s"), type_name(obj->type));
 631 }
 632
 633 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 634                         struct repository *repo,
 635                         const struct object_array *list)
 636 {
 637         unsigned int i;
 638         int hit = 0;
 639         const unsigned int nr = list->nr;
 640
 641         for (i = 0; i < nr; i++) {
 642                 struct object *real_obj;
 643                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 644
 645                 /* load the gitmodules file for this rev */
 646                 if (recurse_submodules) {
 647                         submodule_free();
 648                         gitmodules_config_oid(&real_obj->oid);
 649                 }
 650                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path,
 651                                 repo)) {
 652                         hit = 1;
 653                         if (opt->status_only)
 654                                 break;
 655                 }
 656         }
 657         return hit;
 658 }
 659
 660 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 661                           int exc_std, int use_index)
 662 {
 663         struct dir_struct dir;
 664         int i, hit = 0;
 665
 666         memset(&dir, 0, sizeof(dir));
 667         if (!use_index)
 668                 dir.flags |= DIR_NO_GITLINKS;
 669         if (exc_std)
 670                 setup_standard_excludes(&dir);
 671
 672         fill_directory(&dir, &the_index, pathspec);
 673         for (i = 0; i < dir.nr; i++) {
 674                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
 675                         continue;
 676                 hit |= grep_file(opt, dir.entries[i]->name);
 677                 if (hit && opt->status_only)
 678                         break;
 679         }
 680         return hit;
 681 }
 682
 683 static int context_callback(const struct option *opt, const char *arg,
 684                             int unset)
 685 {
 686         struct grep_opt *grep_opt = opt->value;
 687         int value;
 688         const char *endp;
 689
 690         if (unset) {
 691                 grep_opt->pre_context = grep_opt->post_context = 0;
 692                 return 0;
 693         }
 694         value = strtol(arg, (char **)&endp, 10);
 695         if (*endp) {
 696                 return error(_("switch `%c' expects a numerical value"),
 697                              opt->short_name);
 698         }
 699         grep_opt->pre_context = grep_opt->post_context = value;
 700         return 0;
 701 }
 702
 703 static int file_callback(const struct option *opt, const char *arg, int unset)
 704 {
 705         struct grep_opt *grep_opt = opt->value;
 706         int from_stdin = !strcmp(arg, "-");
 707         FILE *patterns;
 708         int lno = 0;
 709         struct strbuf sb = STRBUF_INIT;
 710
 711         patterns = from_stdin ? stdin : fopen(arg, "r");
 712         if (!patterns)
 713                 die_errno(_("cannot open '%s'"), arg);
 714         while (strbuf_getline(&sb, patterns) == 0) {
 715                 /* ignore empty line like grep does */
 716                 if (sb.len == 0)
 717                         continue;
 718
 719                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 720                                 GREP_PATTERN);
 721         }
 722         if (!from_stdin)
 723                 fclose(patterns);
 724         strbuf_release(&sb);
 725         return 0;
 726 }
 727
 728 static int not_callback(const struct option *opt, const char *arg, int unset)
 729 {
 730         struct grep_opt *grep_opt = opt->value;
 731         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 732         return 0;
 733 }
 734
 735 static int and_callback(const struct option *opt, const char *arg, int unset)
 736 {
 737         struct grep_opt *grep_opt = opt->value;
 738         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 739         return 0;
 740 }
 741
 742 static int open_callback(const struct option *opt, const char *arg, int unset)
 743 {
 744         struct grep_opt *grep_opt = opt->value;
 745         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 746         return 0;
 747 }
 748
 749 static int close_callback(const struct option *opt, const char *arg, int unset)
 750 {
 751         struct grep_opt *grep_opt = opt->value;
 752         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 753         return 0;
 754 }
 755
 756 static int pattern_callback(const struct option *opt, const char *arg,
 757                             int unset)
 758 {
 759         struct grep_opt *grep_opt = opt->value;
 760         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 761         return 0;
 762 }
 763
 764 int cmd_grep(int argc, const char **argv, const char *prefix)
 765 {
 766         int hit = 0;
 767         int cached = 0, untracked = 0, opt_exclude = -1;
 768         int seen_dashdash = 0;
 769         int external_grep_allowed__ignored;
 770         const char *show_in_pager = NULL, *default_pager = "dummy";
 771         struct grep_opt opt;
 772         struct object_array list = OBJECT_ARRAY_INIT;
 773         struct pathspec pathspec;
 774         struct string_list path_list = STRING_LIST_INIT_NODUP;
 775         int i;
 776         int dummy;
 777         int use_index = 1;
 778         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 779         int allow_revs;
 780
 781         struct option options[] = {
 782                 OPT_BOOL(0, "cached", &cached,
 783                         N_("search in index instead of in the work tree")),
 784                 OPT_NEGBIT(0, "no-index", &use_index,
 785                          N_("find in contents not managed by git"), 1),
 786                 OPT_BOOL(0, "untracked", &untracked,
 787                         N_("search in both tracked and untracked files")),
 788                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 789                             N_("ignore files specified via '.gitignore'"), 1),
 790                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 791                          N_("recursively search in each submodule")),
 792                 OPT_GROUP(""),
 793                 OPT_BOOL('v', "invert-match", &opt.invert,
 794                         N_("show non-matching lines")),
 795                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 796                         N_("case insensitive matching")),
 797                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 798                         N_("match patterns only at word boundaries")),
 799                 OPT_SET_INT('a', "text", &opt.binary,
 800                         N_("process binary files as text"), GREP_BINARY_TEXT),
 801                 OPT_SET_INT('I', NULL, &opt.binary,
 802                         N_("don't match patterns in binary files"),
 803                         GREP_BINARY_NOMATCH),
 804                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
 805                          N_("process binary files with textconv filters")),
 806                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
 807                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
 808                         NULL, 1 },
 809                 OPT_GROUP(""),
 810                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
 811                             N_("use extended POSIX regular expressions"),
 812                             GREP_PATTERN_TYPE_ERE),
 813                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
 814                             N_("use basic POSIX regular expressions (default)"),
 815                             GREP_PATTERN_TYPE_BRE),
 816                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
 817                             N_("interpret patterns as fixed strings"),
 818                             GREP_PATTERN_TYPE_FIXED),
 819                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
 820                             N_("use Perl-compatible regular expressions"),
 821                             GREP_PATTERN_TYPE_PCRE),
 822                 OPT_GROUP(""),
 823                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
 824                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
 825                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
 826                 OPT_NEGBIT(0, "full-name", &opt.relative,
 827                         N_("show filenames relative to top directory"), 1),
 828                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
 829                         N_("show only filenames instead of matching lines")),
 830                 OPT_BOOL(0, "name-only", &opt.name_only,
 831                         N_("synonym for --files-with-matches")),
 832                 OPT_BOOL('L', "files-without-match",
 833                         &opt.unmatch_name_only,
 834                         N_("show only the names of files without match")),
 835                 OPT_BOOL('z', "null", &opt.null_following_name,
 836                         N_("print NUL after filenames")),
 837                 OPT_BOOL('c', "count", &opt.count,
 838                         N_("show the number of matches instead of matching lines")),
 839                 OPT__COLOR(&opt.color, N_("highlight matches")),
 840                 OPT_BOOL(0, "break", &opt.file_break,
 841                         N_("print empty line between matches from different files")),
 842                 OPT_BOOL(0, "heading", &opt.heading,
 843                         N_("show filename only once above matches from same file")),
 844                 OPT_GROUP(""),
 845                 OPT_CALLBACK('C', "context", &opt, N_("n"),
 846                         N_("show <n> context lines before and after matches"),
 847                         context_callback),
 848                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 849                         N_("show <n> context lines before matches")),
 850                 OPT_INTEGER('A', "after-context", &opt.post_context,
 851                         N_("show <n> context lines after matches")),
 852                 OPT_INTEGER(0, "threads", &num_threads,
 853                         N_("use <n> worker threads")),
 854                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
 855                         context_callback),
 856                 OPT_BOOL('p', "show-function", &opt.funcname,
 857                         N_("show a line with the function name before matches")),
 858                 OPT_BOOL('W', "function-context", &opt.funcbody,
 859                         N_("show the surrounding function")),
 860                 OPT_GROUP(""),
 861                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
 862                         N_("read patterns from file"), file_callback),
 863                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
 864                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
 865                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 866                   N_("combine patterns specified with -e"),
 867                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 868                 OPT_BOOL(0, "or", &dummy, ""),
 869                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 870                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 871                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 872                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 873                   open_callback },
 874                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 875                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 876                   close_callback },
 877                 OPT__QUIET(&opt.status_only,
 878                            N_("indicate hit with exit status without output")),
 879                 OPT_BOOL(0, "all-match", &opt.all_match,
 880                         N_("show only matches from files that match all patterns")),
 881                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
 882                   N_("show parse tree for grep expression"),
 883                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
 884                 OPT_GROUP(""),
 885                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 886                         N_("pager"), N_("show matching files in the pager"),
 887                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 888                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
 889                          N_("allow calling of grep(1) (ignored by this build)")),
 890                 OPT_END()
 891         };
 892
 893         init_grep_defaults();
 894         git_config(grep_cmd_config, NULL);
 895         grep_init(&opt, prefix);
 896
 897         /*
 898          * If there is no -- then the paths must exist in the working
 899          * tree.  If there is no explicit pattern specified with -e or
 900          * -f, we take the first unrecognized non option to be the
 901          * pattern, but then what follows it must be zero or more
 902          * valid refs up to the -- (if exists), and then existing
 903          * paths.  If there is an explicit pattern, then the first
 904          * unrecognized non option is the beginning of the refs list
 905          * that continues up to the -- (if exists), and then paths.
 906          */
 907         argc = parse_options(argc, argv, prefix, options, grep_usage,
 908                              PARSE_OPT_KEEP_DASHDASH |
 909                              PARSE_OPT_STOP_AT_NON_OPTION);
 910         grep_commit_pattern_type(pattern_type_arg, &opt);
 911
 912         if (use_index && !startup_info->have_repository) {
 913                 int fallback = 0;
 914                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
 915                 if (fallback)
 916                         use_index = 0;
 917                 else
 918                         /* die the same way as if we did it at the beginning */
 919                         setup_git_directory();
 920         }
 921
 922         /*
 923          * skip a -- separator; we know it cannot be
 924          * separating revisions from pathnames if
 925          * we haven't even had any patterns yet
 926          */
 927         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 928                 argv++;
 929                 argc--;
 930         }
 931
 932         /* First unrecognized non-option token */
 933         if (argc > 0 && !opt.pattern_list) {
 934                 append_grep_pattern(&opt, argv[0], "command line", 0,
 935                                     GREP_PATTERN);
 936                 argv++;
 937                 argc--;
 938         }
 939
 940         if (show_in_pager == default_pager)
 941                 show_in_pager = git_pager(1);
 942         if (show_in_pager) {
 943                 opt.color = 0;
 944                 opt.name_only = 1;
 945                 opt.null_following_name = 1;
 946                 opt.output_priv = &path_list;
 947                 opt.output = append_path;
 948                 string_list_append(&path_list, show_in_pager);
 949         }
 950
 951         if (!opt.pattern_list)
 952                 die(_("no pattern given."));
 953
 954         /*
 955          * We have to find "--" in a separate pass, because its presence
 956          * influences how we will parse arguments that come before it.
 957          */
 958         for (i = 0; i < argc; i++) {
 959                 if (!strcmp(argv[i], "--")) {
 960                         seen_dashdash = 1;
 961                         break;
 962                 }
 963         }
 964
 965         /*
 966          * Resolve any rev arguments. If we have a dashdash, then everything up
 967          * to it must resolve as a rev. If not, then we stop at the first
 968          * non-rev and assume everything else is a path.
 969          */
 970         allow_revs = use_index && !untracked;
 971         for (i = 0; i < argc; i++) {
 972                 const char *arg = argv[i];
 973                 struct object_id oid;
 974                 struct object_context oc;
 975                 struct object *object;
 976
 977                 if (!strcmp(arg, "--")) {
 978                         i++;
 979                         break;
 980                 }
 981
 982                 if (!allow_revs) {
 983                         if (seen_dashdash)
 984                                 die(_("--no-index or --untracked cannot be used with revs"));
 985                         break;
 986                 }
 987
 988                 if (get_oid_with_context(arg, GET_OID_RECORD_PATH,
 989                                          &oid, &oc)) {
 990                         if (seen_dashdash)
 991                                 die(_("unable to resolve revision: %s"), arg);
 992                         break;
 993                 }
 994
 995                 object = parse_object_or_die(&oid, arg);
 996                 if (!seen_dashdash)
 997                         verify_non_filename(prefix, arg);
 998                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
 999                 free(oc.path);
1000         }
1001
1002         /*
1003          * Anything left over is presumed to be a path. But in the non-dashdash
1004          * "do what I mean" case, we verify and complain when that isn't true.
1005          */
1006         if (!seen_dashdash) {
1007                 int j;
1008                 for (j = i; j < argc; j++)
1009                         verify_filename(prefix, argv[j], j == i && allow_revs);
1010         }
1011
1012         parse_pathspec(&pathspec, 0,
1013                        PATHSPEC_PREFER_CWD |
1014                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1015                        prefix, argv + i);
1016         pathspec.max_depth = opt.max_depth;
1017         pathspec.recursive = 1;
1018         pathspec.recurse_submodules = !!recurse_submodules;
1019
1020 #ifndef NO_PTHREADS
1021         if (list.nr || cached || show_in_pager)
1022                 num_threads = 0;
1023         else if (num_threads == 0)
1024                 num_threads = GREP_NUM_THREADS_DEFAULT;
1025         else if (num_threads < 0)
1026                 die(_("invalid number of threads specified (%d)"), num_threads);
1027         if (num_threads == 1)
1028                 num_threads = 0;
1029 #else
1030         if (num_threads)
1031                 warning(_("no threads support, ignoring --threads"));
1032         num_threads = 0;
1033 #endif
1034
1035         if (!num_threads)
1036                 /*
1037                  * The compiled patterns on the main path are only
1038                  * used when not using threading. Otherwise
1039                  * start_threads() below calls compile_grep_patterns()
1040                  * for each thread.
1041                  */
1042                 compile_grep_patterns(&opt);
1043
1044 #ifndef NO_PTHREADS
1045         if (num_threads) {
1046                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1047                     && (opt.pre_context || opt.post_context ||
1048                         opt.file_break || opt.funcbody))
1049                         skip_first_line = 1;
1050                 start_threads(&opt);
1051         }
1052 #endif
1053
1054         if (show_in_pager && (cached || list.nr))
1055                 die(_("--open-files-in-pager only works on the worktree"));
1056
1057         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1058                 const char *pager = path_list.items[0].string;
1059                 int len = strlen(pager);
1060
1061                 if (len > 4 && is_dir_sep(pager[len - 5]))
1062                         pager += len - 4;
1063
1064                 if (opt.ignore_case && !strcmp("less", pager))
1065                         string_list_append(&path_list, "-I");
1066
1067                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1068                         struct strbuf buf = STRBUF_INIT;
1069                         strbuf_addf(&buf, "+/%s%s",
1070                                         strcmp("less", pager) ? "" : "*",
1071                                         opt.pattern_list->pattern);
1072                         string_list_append(&path_list, buf.buf);
1073                         strbuf_detach(&buf, NULL);
1074                 }
1075         }
1076
1077         if (recurse_submodules && (!use_index || untracked))
1078                 die(_("option not supported with --recurse-submodules."));
1079
1080         if (!show_in_pager && !opt.status_only)
1081                 setup_pager();
1082
1083         if (!use_index && (untracked || cached))
1084                 die(_("--cached or --untracked cannot be used with --no-index."));
1085
1086         if (!use_index || untracked) {
1087                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1088                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1089         } else if (0 <= opt_exclude) {
1090                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1091         } else if (!list.nr) {
1092                 if (!cached)
1093                         setup_work_tree();
1094
1095                 hit = grep_cache(&opt, the_repository, &pathspec, cached);
1096         } else {
1097                 if (cached)
1098                         die(_("both --cached and trees are given."));
1099
1100                 hit = grep_objects(&opt, &pathspec, the_repository, &list);
1101         }
1102
1103         if (num_threads)
1104                 hit |= wait_all();
1105         if (hit && show_in_pager)
1106                 run_pager(&opt, prefix);
1107         clear_pathspec(&pathspec);
1108         free_grep_patterns(&opt);
1109         return !hit;
1110 }