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

   1 /*
   2  * Builtin "git grep"
   3  *
   4  * Copyright (c) 2006 Junio C Hamano
   5  */
   6 #include "cache.h"
   7 #include "repository.h"
   8 #include "config.h"
   9 #include "blob.h"
  10 #include "tree.h"
  11 #include "commit.h"
  12 #include "tag.h"
  13 #include "tree-walk.h"
  14 #include "builtin.h"
  15 #include "parse-options.h"
  16 #include "string-list.h"
  17 #include "run-command.h"
  18 #include "userdiff.h"
  19 #include "grep.h"
  20 #include "quote.h"
  21 #include "dir.h"
  22 #include "pathspec.h"
  23 #include "submodule.h"
  24 #include "submodule-config.h"
  25 #include "object-store.h"
  26
  27 static char const * const grep_usage[] = {
  28         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  29         NULL
  30 };
  31
  32 static int recurse_submodules;
  33
  34 #define GREP_NUM_THREADS_DEFAULT 8
  35 static int num_threads;
  36
  37 #ifndef NO_PTHREADS
  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         assert(num_threads);
  74         pthread_mutex_lock(&grep_mutex);
  75 }
  76
  77 static inline void grep_unlock(void)
  78 {
  79         assert(num_threads);
  80         pthread_mutex_unlock(&grep_mutex);
  81 }
  82
  83 /* Signalled when a new work_item is added to todo. */
  84 static pthread_cond_t cond_add;
  85
  86 /* Signalled when the result from one work_item is written to
  87  * stdout.
  88  */
  89 static pthread_cond_t cond_write;
  90
  91 /* Signalled when we are finished with everything. */
  92 static pthread_cond_t cond_result;
  93
  94 static int skip_first_line;
  95
  96 static void add_work(struct grep_opt *opt, enum grep_source_type type,
  97                      const char *name, const char *path, const void *id)
  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         grep_source_init(&todo[todo_end].source, type, name, path, id);
 106         if (opt->binary != GREP_BINARY_TEXT)
 107                 grep_source_load_driver(&todo[todo_end].source);
 108         todo[todo_end].done = 0;
 109         strbuf_reset(&todo[todo_end].out);
 110         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 111
 112         pthread_cond_signal(&cond_add);
 113         grep_unlock();
 114 }
 115
 116 static struct work_item *get_work(void)
 117 {
 118         struct work_item *ret;
 119
 120         grep_lock();
 121         while (todo_start == todo_end && !all_work_added) {
 122                 pthread_cond_wait(&cond_add, &grep_mutex);
 123         }
 124
 125         if (todo_start == todo_end && all_work_added) {
 126                 ret = NULL;
 127         } else {
 128                 ret = &todo[todo_start];
 129                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 130         }
 131         grep_unlock();
 132         return ret;
 133 }
 134
 135 static void work_done(struct work_item *w)
 136 {
 137         int old_done;
 138
 139         grep_lock();
 140         w->done = 1;
 141         old_done = todo_done;
 142         for(; todo[todo_done].done && todo_done != todo_start;
 143             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 144                 w = &todo[todo_done];
 145                 if (w->out.len) {
 146                         const char *p = w->out.buf;
 147                         size_t len = w->out.len;
 148
 149                         /* Skip the leading hunk mark of the first file. */
 150                         if (skip_first_line) {
 151                                 while (len) {
 152                                         len--;
 153                                         if (*p++ == '\n')
 154                                                 break;
 155                                 }
 156                                 skip_first_line = 0;
 157                         }
 158
 159                         write_or_die(1, p, len);
 160                 }
 161                 grep_source_clear(&w->source);
 162         }
 163
 164         if (old_done != todo_done)
 165                 pthread_cond_signal(&cond_write);
 166
 167         if (all_work_added && todo_done == todo_end)
 168                 pthread_cond_signal(&cond_result);
 169
 170         grep_unlock();
 171 }
 172
 173 static void *run(void *arg)
 174 {
 175         int hit = 0;
 176         struct grep_opt *opt = arg;
 177
 178         while (1) {
 179                 struct work_item *w = get_work();
 180                 if (!w)
 181                         break;
 182
 183                 opt->output_priv = w;
 184                 hit |= grep_source(opt, &w->source);
 185                 grep_source_clear_data(&w->source);
 186                 work_done(w);
 187         }
 188         free_grep_patterns(arg);
 189         free(arg);
 190
 191         return (void*) (intptr_t) hit;
 192 }
 193
 194 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 195 {
 196         struct work_item *w = opt->output_priv;
 197         strbuf_add(&w->out, buf, size);
 198 }
 199
 200 static void start_threads(struct grep_opt *opt)
 201 {
 202         int i;
 203
 204         pthread_mutex_init(&grep_mutex, NULL);
 205         pthread_mutex_init(&grep_read_mutex, NULL);
 206         pthread_mutex_init(&grep_attr_mutex, NULL);
 207         pthread_cond_init(&cond_add, NULL);
 208         pthread_cond_init(&cond_write, NULL);
 209         pthread_cond_init(&cond_result, NULL);
 210         grep_use_locks = 1;
 211
 212         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 213                 strbuf_init(&todo[i].out, 0);
 214         }
 215
 216         threads = xcalloc(num_threads, sizeof(*threads));
 217         for (i = 0; i < num_threads; i++) {
 218                 int err;
 219                 struct grep_opt *o = grep_opt_dup(opt);
 220                 o->output = strbuf_out;
 221                 if (i)
 222                         o->debug = 0;
 223                 compile_grep_patterns(o);
 224                 err = pthread_create(&threads[i], NULL, run, o);
 225
 226                 if (err)
 227                         die(_("grep: failed to create thread: %s"),
 228                             strerror(err));
 229         }
 230 }
 231
 232 static int wait_all(void)
 233 {
 234         int hit = 0;
 235         int i;
 236
 237         grep_lock();
 238         all_work_added = 1;
 239
 240         /* Wait until all work is done. */
 241         while (todo_done != todo_end)
 242                 pthread_cond_wait(&cond_result, &grep_mutex);
 243
 244         /* Wake up all the consumer threads so they can see that there
 245          * is no more work to do.
 246          */
 247         pthread_cond_broadcast(&cond_add);
 248         grep_unlock();
 249
 250         for (i = 0; i < num_threads; i++) {
 251                 void *h;
 252                 pthread_join(threads[i], &h);
 253                 hit |= (int) (intptr_t) h;
 254         }
 255
 256         free(threads);
 257
 258         pthread_mutex_destroy(&grep_mutex);
 259         pthread_mutex_destroy(&grep_read_mutex);
 260         pthread_mutex_destroy(&grep_attr_mutex);
 261         pthread_cond_destroy(&cond_add);
 262         pthread_cond_destroy(&cond_write);
 263         pthread_cond_destroy(&cond_result);
 264         grep_use_locks = 0;
 265
 266         return hit;
 267 }
 268 #else /* !NO_PTHREADS */
 269
 270 static int wait_all(void)
 271 {
 272         return 0;
 273 }
 274 #endif
 275
 276 static int grep_cmd_config(const char *var, const char *value, void *cb)
 277 {
 278         int st = grep_config(var, value, cb);
 279         if (git_color_default_config(var, value, cb) < 0)
 280                 st = -1;
 281
 282         if (!strcmp(var, "grep.threads")) {
 283                 num_threads = git_config_int(var, value);
 284                 if (num_threads < 0)
 285                         die(_("invalid number of threads specified (%d) for %s"),
 286                             num_threads, var);
 287 #ifdef NO_PTHREADS
 288                 else if (num_threads && num_threads != 1) {
 289                         /*
 290                          * TRANSLATORS: %s is the configuration
 291                          * variable for tweaking threads, currently
 292                          * grep.threads
 293                          */
 294                         warning(_("no threads support, ignoring %s"), var);
 295                         num_threads = 0;
 296                 }
 297 #endif
 298         }
 299
 300         if (!strcmp(var, "submodule.recurse"))
 301                 recurse_submodules = git_config_bool(var, value);
 302
 303         return st;
 304 }
 305
 306 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
 307 {
 308         void *data;
 309
 310         grep_read_lock();
 311         data = read_sha1_file(oid->hash, type, size);
 312         grep_read_unlock();
 313         return data;
 314 }
 315
 316 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
 317                      const char *filename, int tree_name_len,
 318                      const char *path)
 319 {
 320         struct strbuf pathbuf = STRBUF_INIT;
 321
 322         if (opt->relative && opt->prefix_length) {
 323                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
 324                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 325         } else {
 326                 strbuf_addstr(&pathbuf, filename);
 327         }
 328
 329 #ifndef NO_PTHREADS
 330         if (num_threads) {
 331                 add_work(opt, GREP_SOURCE_OID, pathbuf.buf, path, oid);
 332                 strbuf_release(&pathbuf);
 333                 return 0;
 334         } else
 335 #endif
 336         {
 337                 struct grep_source gs;
 338                 int hit;
 339
 340                 grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
 341                 strbuf_release(&pathbuf);
 342                 hit = grep_source(opt, &gs);
 343
 344                 grep_source_clear(&gs);
 345                 return hit;
 346         }
 347 }
 348
 349 static int grep_file(struct grep_opt *opt, const char *filename)
 350 {
 351         struct strbuf buf = STRBUF_INIT;
 352
 353         if (opt->relative && opt->prefix_length)
 354                 quote_path_relative(filename, opt->prefix, &buf);
 355         else
 356                 strbuf_addstr(&buf, filename);
 357
 358 #ifndef NO_PTHREADS
 359         if (num_threads) {
 360                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
 361                 strbuf_release(&buf);
 362                 return 0;
 363         } else
 364 #endif
 365         {
 366                 struct grep_source gs;
 367                 int hit;
 368
 369                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 370                 strbuf_release(&buf);
 371                 hit = grep_source(opt, &gs);
 372
 373                 grep_source_clear(&gs);
 374                 return hit;
 375         }
 376 }
 377
 378 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 379 {
 380         struct string_list *path_list = opt->output_priv;
 381
 382         if (len == 1 && *(const char *)data == '\0')
 383                 return;
 384         string_list_append(path_list, xstrndup(data, len));
 385 }
 386
 387 static void run_pager(struct grep_opt *opt, const char *prefix)
 388 {
 389         struct string_list *path_list = opt->output_priv;
 390         struct child_process child = CHILD_PROCESS_INIT;
 391         int i, status;
 392
 393         for (i = 0; i < path_list->nr; i++)
 394                 argv_array_push(&child.args, path_list->items[i].string);
 395         child.dir = prefix;
 396         child.use_shell = 1;
 397
 398         status = run_command(&child);
 399         if (status)
 400                 exit(status);
 401 }
 402
 403 static int grep_cache(struct grep_opt *opt, struct repository *repo,
 404                       const struct pathspec *pathspec, int cached);
 405 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 406                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 407                      int check_attr, struct repository *repo);
 408
 409 static int grep_submodule(struct grep_opt *opt, struct repository *superproject,
 410                           const struct pathspec *pathspec,
 411                           const struct object_id *oid,
 412                           const char *filename, const char *path)
 413 {
 414         struct repository submodule;
 415         int hit;
 416
 417         if (!is_submodule_active(superproject, path))
 418                 return 0;
 419
 420         if (repo_submodule_init(&submodule, superproject, path))
 421                 return 0;
 422
 423         repo_read_gitmodules(&submodule);
 424
 425         /*
 426          * NEEDSWORK: This adds the submodule's object directory to the list of
 427          * alternates for the single in-memory object store.  This has some bad
 428          * consequences for memory (processed objects will never be freed) and
 429          * performance (this increases the number of pack files git has to pay
 430          * attention to, to the sum of the number of pack files in all the
 431          * repositories processed so far).  This can be removed once the object
 432          * store is no longer global and instead is a member of the repository
 433          * object.
 434          */
 435         grep_read_lock();
 436         add_to_alternates_memory(submodule.objects->objectdir);
 437         grep_read_unlock();
 438
 439         if (oid) {
 440                 struct object *object;
 441                 struct tree_desc tree;
 442                 void *data;
 443                 unsigned long size;
 444                 struct strbuf base = STRBUF_INIT;
 445
 446                 object = parse_object_or_die(oid, oid_to_hex(oid));
 447
 448                 grep_read_lock();
 449                 data = read_object_with_reference(object->oid.hash, tree_type,
 450                                                   &size, NULL);
 451                 grep_read_unlock();
 452
 453                 if (!data)
 454                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
 455
 456                 strbuf_addstr(&base, filename);
 457                 strbuf_addch(&base, '/');
 458
 459                 init_tree_desc(&tree, data, size);
 460                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 461                                 object->type == OBJ_COMMIT, &submodule);
 462                 strbuf_release(&base);
 463                 free(data);
 464         } else {
 465                 hit = grep_cache(opt, &submodule, pathspec, 1);
 466         }
 467
 468         repo_clear(&submodule);
 469         return hit;
 470 }
 471
 472 static int grep_cache(struct grep_opt *opt, struct repository *repo,
 473                       const struct pathspec *pathspec, int cached)
 474 {
 475         int hit = 0;
 476         int nr;
 477         struct strbuf name = STRBUF_INIT;
 478         int name_base_len = 0;
 479         if (repo->submodule_prefix) {
 480                 name_base_len = strlen(repo->submodule_prefix);
 481                 strbuf_addstr(&name, repo->submodule_prefix);
 482         }
 483
 484         repo_read_index(repo);
 485
 486         for (nr = 0; nr < repo->index->cache_nr; nr++) {
 487                 const struct cache_entry *ce = repo->index->cache[nr];
 488                 strbuf_setlen(&name, name_base_len);
 489                 strbuf_addstr(&name, ce->name);
 490
 491                 if (S_ISREG(ce->ce_mode) &&
 492                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
 493                                    S_ISDIR(ce->ce_mode) ||
 494                                    S_ISGITLINK(ce->ce_mode))) {
 495                         /*
 496                          * If CE_VALID is on, we assume worktree file and its
 497                          * cache entry are identical, even if worktree file has
 498                          * been modified, so use cache version instead
 499                          */
 500                         if (cached || (ce->ce_flags & CE_VALID) ||
 501                             ce_skip_worktree(ce)) {
 502                                 if (ce_stage(ce) || ce_intent_to_add(ce))
 503                                         continue;
 504                                 hit |= grep_oid(opt, &ce->oid, name.buf,
 505                                                  0, name.buf);
 506                         } else {
 507                                 hit |= grep_file(opt, name.buf);
 508                         }
 509                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 510                            submodule_path_match(pathspec, name.buf, NULL)) {
 511                         hit |= grep_submodule(opt, repo, pathspec, NULL, ce->name, ce->name);
 512                 } else {
 513                         continue;
 514                 }
 515
 516                 if (ce_stage(ce)) {
 517                         do {
 518                                 nr++;
 519                         } while (nr < repo->index->cache_nr &&
 520                                  !strcmp(ce->name, repo->index->cache[nr]->name));
 521                         nr--; /* compensate for loop control */
 522                 }
 523                 if (hit && opt->status_only)
 524                         break;
 525         }
 526
 527         strbuf_release(&name);
 528         return hit;
 529 }
 530
 531 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 532                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 533                      int check_attr, struct repository *repo)
 534 {
 535         int hit = 0;
 536         enum interesting match = entry_not_interesting;
 537         struct name_entry entry;
 538         int old_baselen = base->len;
 539         struct strbuf name = STRBUF_INIT;
 540         int name_base_len = 0;
 541         if (repo->submodule_prefix) {
 542                 strbuf_addstr(&name, repo->submodule_prefix);
 543                 name_base_len = name.len;
 544         }
 545
 546         while (tree_entry(tree, &entry)) {
 547                 int te_len = tree_entry_len(&entry);
 548
 549                 if (match != all_entries_interesting) {
 550                         strbuf_addstr(&name, base->buf + tn_len);
 551                         match = tree_entry_interesting(&entry, &name,
 552                                                        0, pathspec);
 553                         strbuf_setlen(&name, name_base_len);
 554
 555                         if (match == all_entries_not_interesting)
 556                                 break;
 557                         if (match == entry_not_interesting)
 558                                 continue;
 559                 }
 560
 561                 strbuf_add(base, entry.path, te_len);
 562
 563                 if (S_ISREG(entry.mode)) {
 564                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
 565                                          check_attr ? base->buf + tn_len : NULL);
 566                 } else if (S_ISDIR(entry.mode)) {
 567                         enum object_type type;
 568                         struct tree_desc sub;
 569                         void *data;
 570                         unsigned long size;
 571
 572                         data = lock_and_read_oid_file(entry.oid, &type, &size);
 573                         if (!data)
 574                                 die(_("unable to read tree (%s)"),
 575                                     oid_to_hex(entry.oid));
 576
 577                         strbuf_addch(base, '/');
 578                         init_tree_desc(&sub, data, size);
 579                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 580                                          check_attr, repo);
 581                         free(data);
 582                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 583                         hit |= grep_submodule(opt, repo, pathspec, entry.oid,
 584                                               base->buf, base->buf + tn_len);
 585                 }
 586
 587                 strbuf_setlen(base, old_baselen);
 588
 589                 if (hit && opt->status_only)
 590                         break;
 591         }
 592
 593         strbuf_release(&name);
 594         return hit;
 595 }
 596
 597 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 598                        struct object *obj, const char *name, const char *path,
 599                        struct repository *repo)
 600 {
 601         if (obj->type == OBJ_BLOB)
 602                 return grep_oid(opt, &obj->oid, name, 0, path);
 603         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 604                 struct tree_desc tree;
 605                 void *data;
 606                 unsigned long size;
 607                 struct strbuf base;
 608                 int hit, len;
 609
 610                 grep_read_lock();
 611                 data = read_object_with_reference(obj->oid.hash, tree_type,
 612                                                   &size, NULL);
 613                 grep_read_unlock();
 614
 615                 if (!data)
 616                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 617
 618                 len = name ? strlen(name) : 0;
 619                 strbuf_init(&base, PATH_MAX + len + 1);
 620                 if (len) {
 621                         strbuf_add(&base, name, len);
 622                         strbuf_addch(&base, ':');
 623                 }
 624                 init_tree_desc(&tree, data, size);
 625                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 626                                 obj->type == OBJ_COMMIT, repo);
 627                 strbuf_release(&base);
 628                 free(data);
 629                 return hit;
 630         }
 631         die(_("unable to grep from object of type %s"), typename(obj->type));
 632 }
 633
 634 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 635                         struct repository *repo,
 636                         const struct object_array *list)
 637 {
 638         unsigned int i;
 639         int hit = 0;
 640         const unsigned int nr = list->nr;
 641
 642         for (i = 0; i < nr; i++) {
 643                 struct object *real_obj;
 644                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 645
 646                 /* load the gitmodules file for this rev */
 647                 if (recurse_submodules) {
 648                         submodule_free();
 649                         gitmodules_config_oid(&real_obj->oid);
 650                 }
 651                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path,
 652                                 repo)) {
 653                         hit = 1;
 654                         if (opt->status_only)
 655                                 break;
 656                 }
 657         }
 658         return hit;
 659 }
 660
 661 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 662                           int exc_std, int use_index)
 663 {
 664         struct dir_struct dir;
 665         int i, hit = 0;
 666
 667         memset(&dir, 0, sizeof(dir));
 668         if (!use_index)
 669                 dir.flags |= DIR_NO_GITLINKS;
 670         if (exc_std)
 671                 setup_standard_excludes(&dir);
 672
 673         fill_directory(&dir, &the_index, pathspec);
 674         for (i = 0; i < dir.nr; i++) {
 675                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
 676                         continue;
 677                 hit |= grep_file(opt, dir.entries[i]->name);
 678                 if (hit && opt->status_only)
 679                         break;
 680         }
 681         return hit;
 682 }
 683
 684 static int context_callback(const struct option *opt, const char *arg,
 685                             int unset)
 686 {
 687         struct grep_opt *grep_opt = opt->value;
 688         int value;
 689         const char *endp;
 690
 691         if (unset) {
 692                 grep_opt->pre_context = grep_opt->post_context = 0;
 693                 return 0;
 694         }
 695         value = strtol(arg, (char **)&endp, 10);
 696         if (*endp) {
 697                 return error(_("switch `%c' expects a numerical value"),
 698                              opt->short_name);
 699         }
 700         grep_opt->pre_context = grep_opt->post_context = value;
 701         return 0;
 702 }
 703
 704 static int file_callback(const struct option *opt, const char *arg, int unset)
 705 {
 706         struct grep_opt *grep_opt = opt->value;
 707         int from_stdin = !strcmp(arg, "-");
 708         FILE *patterns;
 709         int lno = 0;
 710         struct strbuf sb = STRBUF_INIT;
 711
 712         patterns = from_stdin ? stdin : fopen(arg, "r");
 713         if (!patterns)
 714                 die_errno(_("cannot open '%s'"), arg);
 715         while (strbuf_getline(&sb, patterns) == 0) {
 716                 /* ignore empty line like grep does */
 717                 if (sb.len == 0)
 718                         continue;
 719
 720                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 721                                 GREP_PATTERN);
 722         }
 723         if (!from_stdin)
 724                 fclose(patterns);
 725         strbuf_release(&sb);
 726         return 0;
 727 }
 728
 729 static int not_callback(const struct option *opt, const char *arg, int unset)
 730 {
 731         struct grep_opt *grep_opt = opt->value;
 732         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 733         return 0;
 734 }
 735
 736 static int and_callback(const struct option *opt, const char *arg, int unset)
 737 {
 738         struct grep_opt *grep_opt = opt->value;
 739         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 740         return 0;
 741 }
 742
 743 static int open_callback(const struct option *opt, const char *arg, int unset)
 744 {
 745         struct grep_opt *grep_opt = opt->value;
 746         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 747         return 0;
 748 }
 749
 750 static int close_callback(const struct option *opt, const char *arg, int unset)
 751 {
 752         struct grep_opt *grep_opt = opt->value;
 753         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 754         return 0;
 755 }
 756
 757 static int pattern_callback(const struct option *opt, const char *arg,
 758                             int unset)
 759 {
 760         struct grep_opt *grep_opt = opt->value;
 761         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 762         return 0;
 763 }
 764
 765 int cmd_grep(int argc, const char **argv, const char *prefix)
 766 {
 767         int hit = 0;
 768         int cached = 0, untracked = 0, opt_exclude = -1;
 769         int seen_dashdash = 0;
 770         int external_grep_allowed__ignored;
 771         const char *show_in_pager = NULL, *default_pager = "dummy";
 772         struct grep_opt opt;
 773         struct object_array list = OBJECT_ARRAY_INIT;
 774         struct pathspec pathspec;
 775         struct string_list path_list = STRING_LIST_INIT_NODUP;
 776         int i;
 777         int dummy;
 778         int use_index = 1;
 779         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 780         int allow_revs;
 781
 782         struct option options[] = {
 783                 OPT_BOOL(0, "cached", &cached,
 784                         N_("search in index instead of in the work tree")),
 785                 OPT_NEGBIT(0, "no-index", &use_index,
 786                          N_("find in contents not managed by git"), 1),
 787                 OPT_BOOL(0, "untracked", &untracked,
 788                         N_("search in both tracked and untracked files")),
 789                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 790                             N_("ignore files specified via '.gitignore'"), 1),
 791                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 792                          N_("recursively search in each submodule")),
 793                 OPT_GROUP(""),
 794                 OPT_BOOL('v', "invert-match", &opt.invert,
 795                         N_("show non-matching lines")),
 796                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 797                         N_("case insensitive matching")),
 798                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 799                         N_("match patterns only at word boundaries")),
 800                 OPT_SET_INT('a', "text", &opt.binary,
 801                         N_("process binary files as text"), GREP_BINARY_TEXT),
 802                 OPT_SET_INT('I', NULL, &opt.binary,
 803                         N_("don't match patterns in binary files"),
 804                         GREP_BINARY_NOMATCH),
 805                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
 806                          N_("process binary files with textconv filters")),
 807                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
 808                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
 809                         NULL, 1 },
 810                 OPT_GROUP(""),
 811                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
 812                             N_("use extended POSIX regular expressions"),
 813                             GREP_PATTERN_TYPE_ERE),
 814                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
 815                             N_("use basic POSIX regular expressions (default)"),
 816                             GREP_PATTERN_TYPE_BRE),
 817                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
 818                             N_("interpret patterns as fixed strings"),
 819                             GREP_PATTERN_TYPE_FIXED),
 820                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
 821                             N_("use Perl-compatible regular expressions"),
 822                             GREP_PATTERN_TYPE_PCRE),
 823                 OPT_GROUP(""),
 824                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
 825                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
 826                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
 827                 OPT_NEGBIT(0, "full-name", &opt.relative,
 828                         N_("show filenames relative to top directory"), 1),
 829                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
 830                         N_("show only filenames instead of matching lines")),
 831                 OPT_BOOL(0, "name-only", &opt.name_only,
 832                         N_("synonym for --files-with-matches")),
 833                 OPT_BOOL('L', "files-without-match",
 834                         &opt.unmatch_name_only,
 835                         N_("show only the names of files without match")),
 836                 OPT_BOOL('z', "null", &opt.null_following_name,
 837                         N_("print NUL after filenames")),
 838                 OPT_BOOL('c', "count", &opt.count,
 839                         N_("show the number of matches instead of matching lines")),
 840                 OPT__COLOR(&opt.color, N_("highlight matches")),
 841                 OPT_BOOL(0, "break", &opt.file_break,
 842                         N_("print empty line between matches from different files")),
 843                 OPT_BOOL(0, "heading", &opt.heading,
 844                         N_("show filename only once above matches from same file")),
 845                 OPT_GROUP(""),
 846                 OPT_CALLBACK('C', "context", &opt, N_("n"),
 847                         N_("show <n> context lines before and after matches"),
 848                         context_callback),
 849                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 850                         N_("show <n> context lines before matches")),
 851                 OPT_INTEGER('A', "after-context", &opt.post_context,
 852                         N_("show <n> context lines after matches")),
 853                 OPT_INTEGER(0, "threads", &num_threads,
 854                         N_("use <n> worker threads")),
 855                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
 856                         context_callback),
 857                 OPT_BOOL('p', "show-function", &opt.funcname,
 858                         N_("show a line with the function name before matches")),
 859                 OPT_BOOL('W', "function-context", &opt.funcbody,
 860                         N_("show the surrounding function")),
 861                 OPT_GROUP(""),
 862                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
 863                         N_("read patterns from file"), file_callback),
 864                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
 865                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
 866                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 867                   N_("combine patterns specified with -e"),
 868                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 869                 OPT_BOOL(0, "or", &dummy, ""),
 870                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 871                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 872                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 873                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 874                   open_callback },
 875                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 876                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 877                   close_callback },
 878                 OPT__QUIET(&opt.status_only,
 879                            N_("indicate hit with exit status without output")),
 880                 OPT_BOOL(0, "all-match", &opt.all_match,
 881                         N_("show only matches from files that match all patterns")),
 882                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
 883                   N_("show parse tree for grep expression"),
 884                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
 885                 OPT_GROUP(""),
 886                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 887                         N_("pager"), N_("show matching files in the pager"),
 888                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 889                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
 890                          N_("allow calling of grep(1) (ignored by this build)")),
 891                 OPT_END()
 892         };
 893
 894         init_grep_defaults();
 895         git_config(grep_cmd_config, NULL);
 896         grep_init(&opt, prefix);
 897
 898         /*
 899          * If there is no -- then the paths must exist in the working
 900          * tree.  If there is no explicit pattern specified with -e or
 901          * -f, we take the first unrecognized non option to be the
 902          * pattern, but then what follows it must be zero or more
 903          * valid refs up to the -- (if exists), and then existing
 904          * paths.  If there is an explicit pattern, then the first
 905          * unrecognized non option is the beginning of the refs list
 906          * that continues up to the -- (if exists), and then paths.
 907          */
 908         argc = parse_options(argc, argv, prefix, options, grep_usage,
 909                              PARSE_OPT_KEEP_DASHDASH |
 910                              PARSE_OPT_STOP_AT_NON_OPTION);
 911         grep_commit_pattern_type(pattern_type_arg, &opt);
 912
 913         if (use_index && !startup_info->have_repository) {
 914                 int fallback = 0;
 915                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
 916                 if (fallback)
 917                         use_index = 0;
 918                 else
 919                         /* die the same way as if we did it at the beginning */
 920                         setup_git_directory();
 921         }
 922
 923         /*
 924          * skip a -- separator; we know it cannot be
 925          * separating revisions from pathnames if
 926          * we haven't even had any patterns yet
 927          */
 928         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 929                 argv++;
 930                 argc--;
 931         }
 932
 933         /* First unrecognized non-option token */
 934         if (argc > 0 && !opt.pattern_list) {
 935                 append_grep_pattern(&opt, argv[0], "command line", 0,
 936                                     GREP_PATTERN);
 937                 argv++;
 938                 argc--;
 939         }
 940
 941         if (show_in_pager == default_pager)
 942                 show_in_pager = git_pager(1);
 943         if (show_in_pager) {
 944                 opt.color = 0;
 945                 opt.name_only = 1;
 946                 opt.null_following_name = 1;
 947                 opt.output_priv = &path_list;
 948                 opt.output = append_path;
 949                 string_list_append(&path_list, show_in_pager);
 950         }
 951
 952         if (!opt.pattern_list)
 953                 die(_("no pattern given."));
 954
 955         /*
 956          * We have to find "--" in a separate pass, because its presence
 957          * influences how we will parse arguments that come before it.
 958          */
 959         for (i = 0; i < argc; i++) {
 960                 if (!strcmp(argv[i], "--")) {
 961                         seen_dashdash = 1;
 962                         break;
 963                 }
 964         }
 965
 966         /*
 967          * Resolve any rev arguments. If we have a dashdash, then everything up
 968          * to it must resolve as a rev. If not, then we stop at the first
 969          * non-rev and assume everything else is a path.
 970          */
 971         allow_revs = use_index && !untracked;
 972         for (i = 0; i < argc; i++) {
 973                 const char *arg = argv[i];
 974                 struct object_id oid;
 975                 struct object_context oc;
 976                 struct object *object;
 977
 978                 if (!strcmp(arg, "--")) {
 979                         i++;
 980                         break;
 981                 }
 982
 983                 if (!allow_revs) {
 984                         if (seen_dashdash)
 985                                 die(_("--no-index or --untracked cannot be used with revs"));
 986                         break;
 987                 }
 988
 989                 if (get_oid_with_context(arg, GET_OID_RECORD_PATH,
 990                                          &oid, &oc)) {
 991                         if (seen_dashdash)
 992                                 die(_("unable to resolve revision: %s"), arg);
 993                         break;
 994                 }
 995
 996                 object = parse_object_or_die(&oid, arg);
 997                 if (!seen_dashdash)
 998                         verify_non_filename(prefix, arg);
 999                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1000                 free(oc.path);
1001         }
1002
1003         /*
1004          * Anything left over is presumed to be a path. But in the non-dashdash
1005          * "do what I mean" case, we verify and complain when that isn't true.
1006          */
1007         if (!seen_dashdash) {
1008                 int j;
1009                 for (j = i; j < argc; j++)
1010                         verify_filename(prefix, argv[j], j == i && allow_revs);
1011         }
1012
1013         parse_pathspec(&pathspec, 0,
1014                        PATHSPEC_PREFER_CWD |
1015                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1016                        prefix, argv + i);
1017         pathspec.max_depth = opt.max_depth;
1018         pathspec.recursive = 1;
1019         pathspec.recurse_submodules = !!recurse_submodules;
1020
1021 #ifndef NO_PTHREADS
1022         if (list.nr || cached || show_in_pager)
1023                 num_threads = 0;
1024         else if (num_threads == 0)
1025                 num_threads = GREP_NUM_THREADS_DEFAULT;
1026         else if (num_threads < 0)
1027                 die(_("invalid number of threads specified (%d)"), num_threads);
1028         if (num_threads == 1)
1029                 num_threads = 0;
1030 #else
1031         if (num_threads)
1032                 warning(_("no threads support, ignoring --threads"));
1033         num_threads = 0;
1034 #endif
1035
1036         if (!num_threads)
1037                 /*
1038                  * The compiled patterns on the main path are only
1039                  * used when not using threading. Otherwise
1040                  * start_threads() below calls compile_grep_patterns()
1041                  * for each thread.
1042                  */
1043                 compile_grep_patterns(&opt);
1044
1045 #ifndef NO_PTHREADS
1046         if (num_threads) {
1047                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1048                     && (opt.pre_context || opt.post_context ||
1049                         opt.file_break || opt.funcbody))
1050                         skip_first_line = 1;
1051                 start_threads(&opt);
1052         }
1053 #endif
1054
1055         if (show_in_pager && (cached || list.nr))
1056                 die(_("--open-files-in-pager only works on the worktree"));
1057
1058         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1059                 const char *pager = path_list.items[0].string;
1060                 int len = strlen(pager);
1061
1062                 if (len > 4 && is_dir_sep(pager[len - 5]))
1063                         pager += len - 4;
1064
1065                 if (opt.ignore_case && !strcmp("less", pager))
1066                         string_list_append(&path_list, "-I");
1067
1068                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1069                         struct strbuf buf = STRBUF_INIT;
1070                         strbuf_addf(&buf, "+/%s%s",
1071                                         strcmp("less", pager) ? "" : "*",
1072                                         opt.pattern_list->pattern);
1073                         string_list_append(&path_list, buf.buf);
1074                         strbuf_detach(&buf, NULL);
1075                 }
1076         }
1077
1078         if (recurse_submodules && (!use_index || untracked))
1079                 die(_("option not supported with --recurse-submodules."));
1080
1081         if (!show_in_pager && !opt.status_only)
1082                 setup_pager();
1083
1084         if (!use_index && (untracked || cached))
1085                 die(_("--cached or --untracked cannot be used with --no-index."));
1086
1087         if (!use_index || untracked) {
1088                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1089                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1090         } else if (0 <= opt_exclude) {
1091                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1092         } else if (!list.nr) {
1093                 if (!cached)
1094                         setup_work_tree();
1095
1096                 hit = grep_cache(&opt, the_repository, &pathspec, cached);
1097         } else {
1098                 if (cached)
1099                         die(_("both --cached and trees are given."));
1100
1101                 hit = grep_objects(&opt, &pathspec, the_repository, &list);
1102         }
1103
1104         if (num_threads)
1105                 hit |= wait_all();
1106         if (hit && show_in_pager)
1107                 run_pager(&opt, prefix);
1108         clear_pathspec(&pathspec);
1109         free_grep_patterns(&opt);
1110         return !hit;
1111 }