builtin/grep.c

   1 /*
   2  * Builtin "git grep"
   3  *
   4  * Copyright (c) 2006 Junio C Hamano
   5  */
   6 #include "cache.h"
   7 #include "blob.h"
   8 #include "tree.h"
   9 #include "commit.h"
  10 #include "tag.h"
  11 #include "tree-walk.h"
  12 #include "builtin.h"
  13 #include "parse-options.h"
  14 #include "string-list.h"
  15 #include "run-command.h"
  16 #include "userdiff.h"
  17 #include "grep.h"
  18 #include "quote.h"
  19 #include "dir.h"
  20 #include "attr.h"
  21
  22 static char const * const grep_usage[] = {
  23         "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
  24         NULL
  25 };
  26
  27 static int use_threads = 1;
  28
  29 #ifndef NO_PTHREADS
  30 #define THREADS 8
  31 static pthread_t threads[THREADS];
  32
  33 /* We use one producer thread and THREADS consumer
  34  * threads. The producer adds struct work_items to 'todo' and the
  35  * consumers pick work items from the same array.
  36  */
  37 struct work_item {
  38         struct grep_source source;
  39         char done;
  40         struct strbuf out;
  41 };
  42
  43 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  44  * that have been or are processed by a consumer thread. We haven't
  45  * written the result for these to stdout yet.
  46  *
  47  * The work_items in [todo_start, todo_end) are waiting to be picked
  48  * up by a consumer thread.
  49  *
  50  * The ranges are modulo TODO_SIZE.
  51  */
  52 #define TODO_SIZE 128
  53 static struct work_item todo[TODO_SIZE];
  54 static int todo_start;
  55 static int todo_end;
  56 static int todo_done;
  57
  58 /* Has all work items been added? */
  59 static int all_work_added;
  60
  61 /* This lock protects all the variables above. */
  62 static pthread_mutex_t grep_mutex;
  63
  64 static inline void grep_lock(void)
  65 {
  66         if (use_threads)
  67                 pthread_mutex_lock(&grep_mutex);
  68 }
  69
  70 static inline void grep_unlock(void)
  71 {
  72         if (use_threads)
  73                 pthread_mutex_unlock(&grep_mutex);
  74 }
  75
  76 /* Signalled when a new work_item is added to todo. */
  77 static pthread_cond_t cond_add;
  78
  79 /* Signalled when the result from one work_item is written to
  80  * stdout.
  81  */
  82 static pthread_cond_t cond_write;
  83
  84 /* Signalled when we are finished with everything. */
  85 static pthread_cond_t cond_result;
  86
  87 static int skip_first_line;
  88
  89 static void add_work(struct grep_opt *opt, enum grep_source_type type,
  90                      const char *name, const void *id)
  91 {
  92         grep_lock();
  93
  94         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
  95                 pthread_cond_wait(&cond_write, &grep_mutex);
  96         }
  97
  98         grep_source_init(&todo[todo_end].source, type, name, id);
  99         if (opt->binary != GREP_BINARY_TEXT)
 100                 grep_source_load_driver(&todo[todo_end].source);
 101         todo[todo_end].done = 0;
 102         strbuf_reset(&todo[todo_end].out);
 103         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 104
 105         pthread_cond_signal(&cond_add);
 106         grep_unlock();
 107 }
 108
 109 static struct work_item *get_work(void)
 110 {
 111         struct work_item *ret;
 112
 113         grep_lock();
 114         while (todo_start == todo_end && !all_work_added) {
 115                 pthread_cond_wait(&cond_add, &grep_mutex);
 116         }
 117
 118         if (todo_start == todo_end && all_work_added) {
 119                 ret = NULL;
 120         } else {
 121                 ret = &todo[todo_start];
 122                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 123         }
 124         grep_unlock();
 125         return ret;
 126 }
 127
 128 static void work_done(struct work_item *w)
 129 {
 130         int old_done;
 131
 132         grep_lock();
 133         w->done = 1;
 134         old_done = todo_done;
 135         for(; todo[todo_done].done && todo_done != todo_start;
 136             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 137                 w = &todo[todo_done];
 138                 if (w->out.len) {
 139                         const char *p = w->out.buf;
 140                         size_t len = w->out.len;
 141
 142                         /* Skip the leading hunk mark of the first file. */
 143                         if (skip_first_line) {
 144                                 while (len) {
 145                                         len--;
 146                                         if (*p++ == '\n')
 147                                                 break;
 148                                 }
 149                                 skip_first_line = 0;
 150                         }
 151
 152                         write_or_die(1, p, len);
 153                 }
 154                 grep_source_clear(&w->source);
 155         }
 156
 157         if (old_done != todo_done)
 158                 pthread_cond_signal(&cond_write);
 159
 160         if (all_work_added && todo_done == todo_end)
 161                 pthread_cond_signal(&cond_result);
 162
 163         grep_unlock();
 164 }
 165
 166 static int skip_binary(struct grep_opt *opt, const char *filename)
 167 {
 168         if ((opt->binary & GREP_BINARY_NOMATCH)) {
 169                 static struct git_attr *attr_text;
 170                 struct git_attr_check check;
 171
 172                 if (!attr_text)
 173                         attr_text = git_attr("text");
 174                 memset(&check, 0, sizeof(check));
 175                 check.attr = attr_text;
 176                 return !git_check_attr(filename, 1, &check) &&
 177                                 ATTR_FALSE(check.value);
 178         }
 179         return 0;
 180 }
 181
 182 static void *run(void *arg)
 183 {
 184         int hit = 0;
 185         struct grep_opt *opt = arg;
 186
 187         while (1) {
 188                 struct work_item *w = get_work();
 189                 if (!w)
 190                         break;
 191
 192                 if (skip_binary(opt, (const char *)w->source.identifier))
 193                         continue;
 194
 195                 opt->output_priv = w;
 196                 hit |= grep_source(opt, &w->source);
 197                 grep_source_clear_data(&w->source);
 198                 work_done(w);
 199         }
 200         free_grep_patterns(arg);
 201         free(arg);
 202
 203         return (void*) (intptr_t) hit;
 204 }
 205
 206 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 207 {
 208         struct work_item *w = opt->output_priv;
 209         strbuf_add(&w->out, buf, size);
 210 }
 211
 212 static void start_threads(struct grep_opt *opt)
 213 {
 214         int i;
 215
 216         pthread_mutex_init(&grep_mutex, NULL);
 217         pthread_mutex_init(&grep_read_mutex, NULL);
 218         pthread_mutex_init(&grep_attr_mutex, NULL);
 219         pthread_cond_init(&cond_add, NULL);
 220         pthread_cond_init(&cond_write, NULL);
 221         pthread_cond_init(&cond_result, NULL);
 222         grep_use_locks = 1;
 223
 224         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 225                 strbuf_init(&todo[i].out, 0);
 226         }
 227
 228         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 229                 int err;
 230                 struct grep_opt *o = grep_opt_dup(opt);
 231                 o->output = strbuf_out;
 232                 compile_grep_patterns(o);
 233                 err = pthread_create(&threads[i], NULL, run, o);
 234
 235                 if (err)
 236                         die(_("grep: failed to create thread: %s"),
 237                             strerror(err));
 238         }
 239 }
 240
 241 static int wait_all(void)
 242 {
 243         int hit = 0;
 244         int i;
 245
 246         grep_lock();
 247         all_work_added = 1;
 248
 249         /* Wait until all work is done. */
 250         while (todo_done != todo_end)
 251                 pthread_cond_wait(&cond_result, &grep_mutex);
 252
 253         /* Wake up all the consumer threads so they can see that there
 254          * is no more work to do.
 255          */
 256         pthread_cond_broadcast(&cond_add);
 257         grep_unlock();
 258
 259         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 260                 void *h;
 261                 pthread_join(threads[i], &h);
 262                 hit |= (int) (intptr_t) h;
 263         }
 264
 265         pthread_mutex_destroy(&grep_mutex);
 266         pthread_mutex_destroy(&grep_read_mutex);
 267         pthread_mutex_destroy(&grep_attr_mutex);
 268         pthread_cond_destroy(&cond_add);
 269         pthread_cond_destroy(&cond_write);
 270         pthread_cond_destroy(&cond_result);
 271         grep_use_locks = 0;
 272
 273         return hit;
 274 }
 275 #else /* !NO_PTHREADS */
 276
 277 static int wait_all(void)
 278 {
 279         return 0;
 280 }
 281 #endif
 282
 283 static int grep_config(const char *var, const char *value, void *cb)
 284 {
 285         struct grep_opt *opt = cb;
 286         char *color = NULL;
 287
 288         if (userdiff_config(var, value) < 0)
 289                 return -1;
 290
 291         if (!strcmp(var, "grep.extendedregexp")) {
 292                 if (git_config_bool(var, value))
 293                         opt->regflags |= REG_EXTENDED;
 294                 else
 295                         opt->regflags &= ~REG_EXTENDED;
 296                 return 0;
 297         }
 298
 299         if (!strcmp(var, "grep.linenumber")) {
 300                 opt->linenum = git_config_bool(var, value);
 301                 return 0;
 302         }
 303
 304         if (!strcmp(var, "color.grep"))
 305                 opt->color = git_config_colorbool(var, value);
 306         else if (!strcmp(var, "color.grep.context"))
 307                 color = opt->color_context;
 308         else if (!strcmp(var, "color.grep.filename"))
 309                 color = opt->color_filename;
 310         else if (!strcmp(var, "color.grep.function"))
 311                 color = opt->color_function;
 312         else if (!strcmp(var, "color.grep.linenumber"))
 313                 color = opt->color_lineno;
 314         else if (!strcmp(var, "color.grep.match"))
 315                 color = opt->color_match;
 316         else if (!strcmp(var, "color.grep.selected"))
 317                 color = opt->color_selected;
 318         else if (!strcmp(var, "color.grep.separator"))
 319                 color = opt->color_sep;
 320         else
 321                 return git_color_default_config(var, value, cb);
 322         if (color) {
 323                 if (!value)
 324                         return config_error_nonbool(var);
 325                 color_parse(value, var, color);
 326         }
 327         return 0;
 328 }
 329
 330 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 331 {
 332         void *data;
 333
 334         grep_read_lock();
 335         data = read_sha1_file(sha1, type, size);
 336         grep_read_unlock();
 337         return data;
 338 }
 339
 340 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 341                      const char *filename, int tree_name_len)
 342 {
 343         struct strbuf pathbuf = STRBUF_INIT;
 344
 345         if (opt->relative && opt->prefix_length) {
 346                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 347                                     opt->prefix);
 348                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 349         } else {
 350                 strbuf_addstr(&pathbuf, filename);
 351         }
 352
 353 #ifndef NO_PTHREADS
 354         if (use_threads) {
 355                 add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, sha1);
 356                 strbuf_release(&pathbuf);
 357                 return 0;
 358         } else
 359 #endif
 360         {
 361                 struct grep_source gs;
 362                 int hit;
 363
 364                 grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, sha1);
 365                 strbuf_release(&pathbuf);
 366                 hit = grep_source(opt, &gs);
 367
 368                 grep_source_clear(&gs);
 369                 return hit;
 370         }
 371 }
 372
 373 static int grep_file(struct grep_opt *opt, const char *filename)
 374 {
 375         struct strbuf buf = STRBUF_INIT;
 376
 377         if (opt->relative && opt->prefix_length)
 378                 quote_path_relative(filename, -1, &buf, opt->prefix);
 379         else
 380                 strbuf_addstr(&buf, filename);
 381
 382 #ifndef NO_PTHREADS
 383         if (use_threads) {
 384                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename);
 385                 strbuf_release(&buf);
 386                 return 0;
 387         } else
 388 #endif
 389         {
 390                 struct grep_source gs;
 391                 int hit;
 392
 393                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename);
 394                 strbuf_release(&buf);
 395                 hit = grep_source(opt, &gs);
 396
 397                 grep_source_clear(&gs);
 398                 return hit;
 399         }
 400 }
 401
 402 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 403 {
 404         struct string_list *path_list = opt->output_priv;
 405
 406         if (len == 1 && *(const char *)data == '\0')
 407                 return;
 408         string_list_append(path_list, xstrndup(data, len));
 409 }
 410
 411 static void run_pager(struct grep_opt *opt, const char *prefix)
 412 {
 413         struct string_list *path_list = opt->output_priv;
 414         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 415         int i, status;
 416
 417         for (i = 0; i < path_list->nr; i++)
 418                 argv[i] = path_list->items[i].string;
 419         argv[path_list->nr] = NULL;
 420
 421         if (prefix && chdir(prefix))
 422                 die(_("Failed to chdir: %s"), prefix);
 423         status = run_command_v_opt(argv, RUN_USING_SHELL);
 424         if (status)
 425                 exit(status);
 426         free(argv);
 427 }
 428
 429 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 430 {
 431         int hit = 0;
 432         int nr;
 433         read_cache();
 434
 435         for (nr = 0; nr < active_nr; nr++) {
 436                 struct cache_entry *ce = active_cache[nr];
 437                 if (!S_ISREG(ce->ce_mode))
 438                         continue;
 439                 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 440                         continue;
 441                 if (skip_binary(opt, ce->name))
 442                         continue;
 443
 444                 /*
 445                  * If CE_VALID is on, we assume worktree file and its cache entry
 446                  * are identical, even if worktree file has been modified, so use
 447                  * cache version instead
 448                  */
 449                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 450                         if (ce_stage(ce))
 451                                 continue;
 452                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 453                 }
 454                 else
 455                         hit |= grep_file(opt, ce->name);
 456                 if (ce_stage(ce)) {
 457                         do {
 458                                 nr++;
 459                         } while (nr < active_nr &&
 460                                  !strcmp(ce->name, active_cache[nr]->name));
 461                         nr--; /* compensate for loop control */
 462                 }
 463                 if (hit && opt->status_only)
 464                         break;
 465         }
 466         return hit;
 467 }
 468
 469 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 470                      struct tree_desc *tree, struct strbuf *base, int tn_len)
 471 {
 472         int hit = 0;
 473         enum interesting match = entry_not_interesting;
 474         struct name_entry entry;
 475         int old_baselen = base->len;
 476
 477         while (tree_entry(tree, &entry)) {
 478                 int te_len = tree_entry_len(&entry);
 479
 480                 if (match != all_entries_interesting) {
 481                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 482                         if (match == all_entries_not_interesting)
 483                                 break;
 484                         if (match == entry_not_interesting)
 485                                 continue;
 486                 }
 487
 488                 strbuf_add(base, entry.path, te_len);
 489
 490                 if (S_ISREG(entry.mode)) {
 491                         hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 492                 }
 493                 else if (S_ISDIR(entry.mode)) {
 494                         enum object_type type;
 495                         struct tree_desc sub;
 496                         void *data;
 497                         unsigned long size;
 498
 499                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 500                         if (!data)
 501                                 die(_("unable to read tree (%s)"),
 502                                     sha1_to_hex(entry.sha1));
 503
 504                         strbuf_addch(base, '/');
 505                         init_tree_desc(&sub, data, size);
 506                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 507                         free(data);
 508                 }
 509                 strbuf_setlen(base, old_baselen);
 510
 511                 if (hit && opt->status_only)
 512                         break;
 513         }
 514         return hit;
 515 }
 516
 517 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 518                        struct object *obj, const char *name)
 519 {
 520         if (obj->type == OBJ_BLOB)
 521                 return grep_sha1(opt, obj->sha1, name, 0);
 522         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 523                 struct tree_desc tree;
 524                 void *data;
 525                 unsigned long size;
 526                 struct strbuf base;
 527                 int hit, len;
 528
 529                 grep_read_lock();
 530                 data = read_object_with_reference(obj->sha1, tree_type,
 531                                                   &size, NULL);
 532                 grep_read_unlock();
 533
 534                 if (!data)
 535                         die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 536
 537                 len = name ? strlen(name) : 0;
 538                 strbuf_init(&base, PATH_MAX + len + 1);
 539                 if (len) {
 540                         strbuf_add(&base, name, len);
 541                         strbuf_addch(&base, ':');
 542                 }
 543                 init_tree_desc(&tree, data, size);
 544                 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 545                 strbuf_release(&base);
 546                 free(data);
 547                 return hit;
 548         }
 549         die(_("unable to grep from object of type %s"), typename(obj->type));
 550 }
 551
 552 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 553                         const struct object_array *list)
 554 {
 555         unsigned int i;
 556         int hit = 0;
 557         const unsigned int nr = list->nr;
 558
 559         for (i = 0; i < nr; i++) {
 560                 struct object *real_obj;
 561                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 562                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 563                         hit = 1;
 564                         if (opt->status_only)
 565                                 break;
 566                 }
 567         }
 568         return hit;
 569 }
 570
 571 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 572                           int exc_std)
 573 {
 574         struct dir_struct dir;
 575         int i, hit = 0;
 576
 577         memset(&dir, 0, sizeof(dir));
 578         if (exc_std)
 579                 setup_standard_excludes(&dir);
 580
 581         fill_directory(&dir, pathspec->raw);
 582         for (i = 0; i < dir.nr; i++) {
 583                 const char *name = dir.entries[i]->name;
 584                 int namelen = strlen(name);
 585                 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 586                         continue;
 587                 hit |= grep_file(opt, dir.entries[i]->name);
 588                 if (hit && opt->status_only)
 589                         break;
 590         }
 591         return hit;
 592 }
 593
 594 static int context_callback(const struct option *opt, const char *arg,
 595                             int unset)
 596 {
 597         struct grep_opt *grep_opt = opt->value;
 598         int value;
 599         const char *endp;
 600
 601         if (unset) {
 602                 grep_opt->pre_context = grep_opt->post_context = 0;
 603                 return 0;
 604         }
 605         value = strtol(arg, (char **)&endp, 10);
 606         if (*endp) {
 607                 return error(_("switch `%c' expects a numerical value"),
 608                              opt->short_name);
 609         }
 610         grep_opt->pre_context = grep_opt->post_context = value;
 611         return 0;
 612 }
 613
 614 static int file_callback(const struct option *opt, const char *arg, int unset)
 615 {
 616         struct grep_opt *grep_opt = opt->value;
 617         int from_stdin = !strcmp(arg, "-");
 618         FILE *patterns;
 619         int lno = 0;
 620         struct strbuf sb = STRBUF_INIT;
 621
 622         patterns = from_stdin ? stdin : fopen(arg, "r");
 623         if (!patterns)
 624                 die_errno(_("cannot open '%s'"), arg);
 625         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 626                 /* ignore empty line like grep does */
 627                 if (sb.len == 0)
 628                         continue;
 629
 630                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 631                                 GREP_PATTERN);
 632         }
 633         if (!from_stdin)
 634                 fclose(patterns);
 635         strbuf_release(&sb);
 636         return 0;
 637 }
 638
 639 static int not_callback(const struct option *opt, const char *arg, int unset)
 640 {
 641         struct grep_opt *grep_opt = opt->value;
 642         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 643         return 0;
 644 }
 645
 646 static int and_callback(const struct option *opt, const char *arg, int unset)
 647 {
 648         struct grep_opt *grep_opt = opt->value;
 649         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 650         return 0;
 651 }
 652
 653 static int open_callback(const struct option *opt, const char *arg, int unset)
 654 {
 655         struct grep_opt *grep_opt = opt->value;
 656         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 657         return 0;
 658 }
 659
 660 static int close_callback(const struct option *opt, const char *arg, int unset)
 661 {
 662         struct grep_opt *grep_opt = opt->value;
 663         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 664         return 0;
 665 }
 666
 667 static int pattern_callback(const struct option *opt, const char *arg,
 668                             int unset)
 669 {
 670         struct grep_opt *grep_opt = opt->value;
 671         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 672         return 0;
 673 }
 674
 675 static int help_callback(const struct option *opt, const char *arg, int unset)
 676 {
 677         return -1;
 678 }
 679
 680 int cmd_grep(int argc, const char **argv, const char *prefix)
 681 {
 682         int hit = 0;
 683         int cached = 0, untracked = 0, opt_exclude = -1;
 684         int seen_dashdash = 0;
 685         int external_grep_allowed__ignored;
 686         const char *show_in_pager = NULL, *default_pager = "dummy";
 687         struct grep_opt opt;
 688         struct object_array list = OBJECT_ARRAY_INIT;
 689         const char **paths = NULL;
 690         struct pathspec pathspec;
 691         struct string_list path_list = STRING_LIST_INIT_NODUP;
 692         int i;
 693         int dummy;
 694         int use_index = 1;
 695         enum {
 696                 pattern_type_unspecified = 0,
 697                 pattern_type_bre,
 698                 pattern_type_ere,
 699                 pattern_type_fixed,
 700                 pattern_type_pcre,
 701         };
 702         int pattern_type = pattern_type_unspecified;
 703
 704         struct option options[] = {
 705                 OPT_BOOLEAN(0, "cached", &cached,
 706                         "search in index instead of in the work tree"),
 707                 OPT_NEGBIT(0, "no-index", &use_index,
 708                          "finds in contents not managed by git", 1),
 709                 OPT_BOOLEAN(0, "untracked", &untracked,
 710                         "search in both tracked and untracked files"),
 711                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 712                             "search also in ignored files", 1),
 713                 OPT_GROUP(""),
 714                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 715                         "show non-matching lines"),
 716                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 717                         "case insensitive matching"),
 718                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 719                         "match patterns only at word boundaries"),
 720                 OPT_SET_INT('a', "text", &opt.binary,
 721                         "process binary files as text", GREP_BINARY_TEXT),
 722                 OPT_SET_INT('I', NULL, &opt.binary,
 723                         "don't match patterns in binary files",
 724                         GREP_BINARY_NOMATCH),
 725                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 726                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 727                         NULL, 1 },
 728                 OPT_GROUP(""),
 729                 OPT_SET_INT('E', "extended-regexp", &pattern_type,
 730                             "use extended POSIX regular expressions",
 731                             pattern_type_ere),
 732                 OPT_SET_INT('G', "basic-regexp", &pattern_type,
 733                             "use basic POSIX regular expressions (default)",
 734                             pattern_type_bre),
 735                 OPT_SET_INT('F', "fixed-strings", &pattern_type,
 736                             "interpret patterns as fixed strings",
 737                             pattern_type_fixed),
 738                 OPT_SET_INT('P', "perl-regexp", &pattern_type,
 739                             "use Perl-compatible regular expressions",
 740                             pattern_type_pcre),
 741                 OPT_GROUP(""),
 742                 OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
 743                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 744                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 745                 OPT_NEGBIT(0, "full-name", &opt.relative,
 746                         "show filenames relative to top directory", 1),
 747                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 748                         "show only filenames instead of matching lines"),
 749                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 750                         "synonym for --files-with-matches"),
 751                 OPT_BOOLEAN('L', "files-without-match",
 752                         &opt.unmatch_name_only,
 753                         "show only the names of files without match"),
 754                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 755                         "print NUL after filenames"),
 756                 OPT_BOOLEAN('c', "count", &opt.count,
 757                         "show the number of matches instead of matching lines"),
 758                 OPT__COLOR(&opt.color, "highlight matches"),
 759                 OPT_BOOLEAN(0, "break", &opt.file_break,
 760                         "print empty line between matches from different files"),
 761                 OPT_BOOLEAN(0, "heading", &opt.heading,
 762                         "show filename only once above matches from same file"),
 763                 OPT_GROUP(""),
 764                 OPT_CALLBACK('C', "context", &opt, "n",
 765                         "show <n> context lines before and after matches",
 766                         context_callback),
 767                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 768                         "show <n> context lines before matches"),
 769                 OPT_INTEGER('A', "after-context", &opt.post_context,
 770                         "show <n> context lines after matches"),
 771                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 772                         context_callback),
 773                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 774                         "show a line with the function name before matches"),
 775                 OPT_BOOLEAN('W', "function-context", &opt.funcbody,
 776                         "show the surrounding function"),
 777                 OPT_GROUP(""),
 778                 OPT_CALLBACK('f', NULL, &opt, "file",
 779                         "read patterns from file", file_callback),
 780                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 781                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 782                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 783                   "combine patterns specified with -e",
 784                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 785                 OPT_BOOLEAN(0, "or", &dummy, ""),
 786                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 787                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 788                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 789                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 790                   open_callback },
 791                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 792                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 793                   close_callback },
 794                 OPT__QUIET(&opt.status_only,
 795                            "indicate hit with exit status without output"),
 796                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 797                         "show only matches from files that match all patterns"),
 798                 OPT_GROUP(""),
 799                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 800                         "pager", "show matching files in the pager",
 801                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 802                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 803                             "allow calling of grep(1) (ignored by this build)"),
 804                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 805                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 806                 OPT_END()
 807         };
 808
 809         /*
 810          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 811          * to show usage information and exit.
 812          */
 813         if (argc == 2 && !strcmp(argv[1], "-h"))
 814                 usage_with_options(grep_usage, options);
 815
 816         memset(&opt, 0, sizeof(opt));
 817         opt.prefix = prefix;
 818         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 819         opt.relative = 1;
 820         opt.pathname = 1;
 821         opt.pattern_tail = &opt.pattern_list;
 822         opt.header_tail = &opt.header_list;
 823         opt.regflags = REG_NEWLINE;
 824         opt.max_depth = -1;
 825
 826         strcpy(opt.color_context, "");
 827         strcpy(opt.color_filename, "");
 828         strcpy(opt.color_function, "");
 829         strcpy(opt.color_lineno, "");
 830         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 831         strcpy(opt.color_selected, "");
 832         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 833         opt.color = -1;
 834         git_config(grep_config, &opt);
 835
 836         /*
 837          * If there is no -- then the paths must exist in the working
 838          * tree.  If there is no explicit pattern specified with -e or
 839          * -f, we take the first unrecognized non option to be the
 840          * pattern, but then what follows it must be zero or more
 841          * valid refs up to the -- (if exists), and then existing
 842          * paths.  If there is an explicit pattern, then the first
 843          * unrecognized non option is the beginning of the refs list
 844          * that continues up to the -- (if exists), and then paths.
 845          */
 846         argc = parse_options(argc, argv, prefix, options, grep_usage,
 847                              PARSE_OPT_KEEP_DASHDASH |
 848                              PARSE_OPT_STOP_AT_NON_OPTION |
 849                              PARSE_OPT_NO_INTERNAL_HELP);
 850         switch (pattern_type) {
 851         case pattern_type_fixed:
 852                 opt.fixed = 1;
 853                 opt.pcre = 0;
 854                 break;
 855         case pattern_type_bre:
 856                 opt.fixed = 0;
 857                 opt.pcre = 0;
 858                 opt.regflags &= ~REG_EXTENDED;
 859                 break;
 860         case pattern_type_ere:
 861                 opt.fixed = 0;
 862                 opt.pcre = 0;
 863                 opt.regflags |= REG_EXTENDED;
 864                 break;
 865         case pattern_type_pcre:
 866                 opt.fixed = 0;
 867                 opt.pcre = 1;
 868                 break;
 869         default:
 870                 break; /* nothing */
 871         }
 872
 873         if (use_index && !startup_info->have_repository)
 874                 /* die the same way as if we did it at the beginning */
 875                 setup_git_directory();
 876
 877         /*
 878          * skip a -- separator; we know it cannot be
 879          * separating revisions from pathnames if
 880          * we haven't even had any patterns yet
 881          */
 882         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 883                 argv++;
 884                 argc--;
 885         }
 886
 887         /* First unrecognized non-option token */
 888         if (argc > 0 && !opt.pattern_list) {
 889                 append_grep_pattern(&opt, argv[0], "command line", 0,
 890                                     GREP_PATTERN);
 891                 argv++;
 892                 argc--;
 893         }
 894
 895         if (show_in_pager == default_pager)
 896                 show_in_pager = git_pager(1);
 897         if (show_in_pager) {
 898                 opt.color = 0;
 899                 opt.name_only = 1;
 900                 opt.null_following_name = 1;
 901                 opt.output_priv = &path_list;
 902                 opt.output = append_path;
 903                 string_list_append(&path_list, show_in_pager);
 904                 use_threads = 0;
 905         }
 906         if ((opt.binary & GREP_BINARY_NOMATCH))
 907                 use_threads = 0;
 908
 909         if (!opt.pattern_list)
 910                 die(_("no pattern given."));
 911         if (!opt.fixed && opt.ignore_case)
 912                 opt.regflags |= REG_ICASE;
 913
 914         compile_grep_patterns(&opt);
 915
 916         /* Check revs and then paths */
 917         for (i = 0; i < argc; i++) {
 918                 const char *arg = argv[i];
 919                 unsigned char sha1[20];
 920                 /* Is it a rev? */
 921                 if (!get_sha1(arg, sha1)) {
 922                         struct object *object = parse_object(sha1);
 923                         if (!object)
 924                                 die(_("bad object %s"), arg);
 925                         add_object_array(object, arg, &list);
 926                         continue;
 927                 }
 928                 if (!strcmp(arg, "--")) {
 929                         i++;
 930                         seen_dashdash = 1;
 931                 }
 932                 break;
 933         }
 934
 935 #ifndef NO_PTHREADS
 936         if (list.nr || cached || online_cpus() == 1)
 937                 use_threads = 0;
 938 #else
 939         use_threads = 0;
 940 #endif
 941
 942 #ifndef NO_PTHREADS
 943         if (use_threads) {
 944                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
 945                     && (opt.pre_context || opt.post_context ||
 946                         opt.file_break || opt.funcbody))
 947                         skip_first_line = 1;
 948                 start_threads(&opt);
 949         }
 950 #endif
 951
 952         /* The rest are paths */
 953         if (!seen_dashdash) {
 954                 int j;
 955                 for (j = i; j < argc; j++)
 956                         verify_filename(prefix, argv[j]);
 957         }
 958
 959         paths = get_pathspec(prefix, argv + i);
 960         init_pathspec(&pathspec, paths);
 961         pathspec.max_depth = opt.max_depth;
 962         pathspec.recursive = 1;
 963
 964         if (show_in_pager && (cached || list.nr))
 965                 die(_("--open-files-in-pager only works on the worktree"));
 966
 967         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
 968                 const char *pager = path_list.items[0].string;
 969                 int len = strlen(pager);
 970
 971                 if (len > 4 && is_dir_sep(pager[len - 5]))
 972                         pager += len - 4;
 973
 974                 if (opt.ignore_case && !strcmp("less", pager))
 975                         string_list_append(&path_list, "-i");
 976
 977                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
 978                         struct strbuf buf = STRBUF_INIT;
 979                         strbuf_addf(&buf, "+/%s%s",
 980                                         strcmp("less", pager) ? "" : "*",
 981                                         opt.pattern_list->pattern);
 982                         string_list_append(&path_list, buf.buf);
 983                         strbuf_detach(&buf, NULL);
 984                 }
 985         }
 986
 987         if (!show_in_pager)
 988                 setup_pager();
 989
 990         if (!use_index && (untracked || cached))
 991                 die(_("--cached or --untracked cannot be used with --no-index."));
 992
 993         if (!use_index || untracked) {
 994                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
 995                 if (list.nr)
 996                         die(_("--no-index or --untracked cannot be used with revs."));
 997                 hit = grep_directory(&opt, &pathspec, use_exclude);
 998         } else if (0 <= opt_exclude) {
 999                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1000         } else if (!list.nr) {
1001                 if (!cached)
1002                         setup_work_tree();
1003
1004                 hit = grep_cache(&opt, &pathspec, cached);
1005         } else {
1006                 if (cached)
1007                         die(_("both --cached and trees are given."));
1008                 hit = grep_objects(&opt, &pathspec, &list);
1009         }
1010
1011         if (use_threads)
1012                 hit |= wait_all();
1013         if (hit && show_in_pager)
1014                 run_pager(&opt, prefix);
1015         free_grep_patterns(&opt);
1016         return !hit;
1017 }