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 "thread-utils.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 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  34                        const char *name);
  35 static void *load_file(const char *filename, size_t *sz);
  36
  37 enum work_type {WORK_SHA1, WORK_FILE};
  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         enum work_type type;
  45         char *name;
  46
  47         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  48          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  49          * terminated filename.
  50          */
  51         void *identifier;
  52         char done;
  53         struct strbuf out;
  54 };
  55
  56 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  57  * that have been or are processed by a consumer thread. We haven't
  58  * written the result for these to stdout yet.
  59  *
  60  * The work_items in [todo_start, todo_end) are waiting to be picked
  61  * up by a consumer thread.
  62  *
  63  * The ranges are modulo TODO_SIZE.
  64  */
  65 #define TODO_SIZE 128
  66 static struct work_item todo[TODO_SIZE];
  67 static int todo_start;
  68 static int todo_end;
  69 static int todo_done;
  70
  71 /* Has all work items been added? */
  72 static int all_work_added;
  73
  74 /* This lock protects all the variables above. */
  75 static pthread_mutex_t grep_mutex;
  76
  77 static inline void grep_lock(void)
  78 {
  79         if (use_threads)
  80                 pthread_mutex_lock(&grep_mutex);
  81 }
  82
  83 static inline void grep_unlock(void)
  84 {
  85         if (use_threads)
  86                 pthread_mutex_unlock(&grep_mutex);
  87 }
  88
  89 /* Used to serialize calls to read_sha1_file. */
  90 static pthread_mutex_t read_sha1_mutex;
  91
  92 static inline void read_sha1_lock(void)
  93 {
  94         if (use_threads)
  95                 pthread_mutex_lock(&read_sha1_mutex);
  96 }
  97
  98 static inline void read_sha1_unlock(void)
  99 {
 100         if (use_threads)
 101                 pthread_mutex_unlock(&read_sha1_mutex);
 102 }
 103
 104 /* Signalled when a new work_item is added to todo. */
 105 static pthread_cond_t cond_add;
 106
 107 /* Signalled when the result from one work_item is written to
 108  * stdout.
 109  */
 110 static pthread_cond_t cond_write;
 111
 112 /* Signalled when we are finished with everything. */
 113 static pthread_cond_t cond_result;
 114
 115 static int skip_first_line;
 116
 117 static void add_work(enum work_type type, char *name, void *id)
 118 {
 119         grep_lock();
 120
 121         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 122                 pthread_cond_wait(&cond_write, &grep_mutex);
 123         }
 124
 125         todo[todo_end].type = type;
 126         todo[todo_end].name = name;
 127         todo[todo_end].identifier = id;
 128         todo[todo_end].done = 0;
 129         strbuf_reset(&todo[todo_end].out);
 130         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 131
 132         pthread_cond_signal(&cond_add);
 133         grep_unlock();
 134 }
 135
 136 static struct work_item *get_work(void)
 137 {
 138         struct work_item *ret;
 139
 140         grep_lock();
 141         while (todo_start == todo_end && !all_work_added) {
 142                 pthread_cond_wait(&cond_add, &grep_mutex);
 143         }
 144
 145         if (todo_start == todo_end && all_work_added) {
 146                 ret = NULL;
 147         } else {
 148                 ret = &todo[todo_start];
 149                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 150         }
 151         grep_unlock();
 152         return ret;
 153 }
 154
 155 static void grep_sha1_async(struct grep_opt *opt, char *name,
 156                             const unsigned char *sha1)
 157 {
 158         unsigned char *s;
 159         s = xmalloc(20);
 160         memcpy(s, sha1, 20);
 161         add_work(WORK_SHA1, name, s);
 162 }
 163
 164 static void grep_file_async(struct grep_opt *opt, char *name,
 165                             const char *filename)
 166 {
 167         add_work(WORK_FILE, name, xstrdup(filename));
 168 }
 169
 170 static void work_done(struct work_item *w)
 171 {
 172         int old_done;
 173
 174         grep_lock();
 175         w->done = 1;
 176         old_done = todo_done;
 177         for(; todo[todo_done].done && todo_done != todo_start;
 178             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 179                 w = &todo[todo_done];
 180                 if (w->out.len) {
 181                         const char *p = w->out.buf;
 182                         size_t len = w->out.len;
 183
 184                         /* Skip the leading hunk mark of the first file. */
 185                         if (skip_first_line) {
 186                                 while (len) {
 187                                         len--;
 188                                         if (*p++ == '\n')
 189                                                 break;
 190                                 }
 191                                 skip_first_line = 0;
 192                         }
 193
 194                         write_or_die(1, p, len);
 195                 }
 196                 free(w->name);
 197                 free(w->identifier);
 198         }
 199
 200         if (old_done != todo_done)
 201                 pthread_cond_signal(&cond_write);
 202
 203         if (all_work_added && todo_done == todo_end)
 204                 pthread_cond_signal(&cond_result);
 205
 206         grep_unlock();
 207 }
 208
 209 static void *run(void *arg)
 210 {
 211         int hit = 0;
 212         struct grep_opt *opt = arg;
 213
 214         while (1) {
 215                 struct work_item *w = get_work();
 216                 if (!w)
 217                         break;
 218
 219                 opt->output_priv = w;
 220                 if (w->type == WORK_SHA1) {
 221                         unsigned long sz;
 222                         void* data = load_sha1(w->identifier, &sz, w->name);
 223
 224                         if (data) {
 225                                 hit |= grep_buffer(opt, w->name, data, sz);
 226                                 free(data);
 227                         }
 228                 } else if (w->type == WORK_FILE) {
 229                         size_t sz;
 230                         void* data = load_file(w->identifier, &sz);
 231                         if (data) {
 232                                 hit |= grep_buffer(opt, w->name, data, sz);
 233                                 free(data);
 234                         }
 235                 } else {
 236                         assert(0);
 237                 }
 238
 239                 work_done(w);
 240         }
 241         free_grep_patterns(arg);
 242         free(arg);
 243
 244         return (void*) (intptr_t) hit;
 245 }
 246
 247 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 248 {
 249         struct work_item *w = opt->output_priv;
 250         strbuf_add(&w->out, buf, size);
 251 }
 252
 253 static void start_threads(struct grep_opt *opt)
 254 {
 255         int i;
 256
 257         pthread_mutex_init(&grep_mutex, NULL);
 258         pthread_mutex_init(&read_sha1_mutex, NULL);
 259         pthread_cond_init(&cond_add, NULL);
 260         pthread_cond_init(&cond_write, NULL);
 261         pthread_cond_init(&cond_result, NULL);
 262
 263         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 264                 strbuf_init(&todo[i].out, 0);
 265         }
 266
 267         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 268                 int err;
 269                 struct grep_opt *o = grep_opt_dup(opt);
 270                 o->output = strbuf_out;
 271                 compile_grep_patterns(o);
 272                 err = pthread_create(&threads[i], NULL, run, o);
 273
 274                 if (err)
 275                         die(_("grep: failed to create thread: %s"),
 276                             strerror(err));
 277         }
 278 }
 279
 280 static int wait_all(void)
 281 {
 282         int hit = 0;
 283         int i;
 284
 285         grep_lock();
 286         all_work_added = 1;
 287
 288         /* Wait until all work is done. */
 289         while (todo_done != todo_end)
 290                 pthread_cond_wait(&cond_result, &grep_mutex);
 291
 292         /* Wake up all the consumer threads so they can see that there
 293          * is no more work to do.
 294          */
 295         pthread_cond_broadcast(&cond_add);
 296         grep_unlock();
 297
 298         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 299                 void *h;
 300                 pthread_join(threads[i], &h);
 301                 hit |= (int) (intptr_t) h;
 302         }
 303
 304         pthread_mutex_destroy(&grep_mutex);
 305         pthread_mutex_destroy(&read_sha1_mutex);
 306         pthread_cond_destroy(&cond_add);
 307         pthread_cond_destroy(&cond_write);
 308         pthread_cond_destroy(&cond_result);
 309
 310         return hit;
 311 }
 312 #else /* !NO_PTHREADS */
 313 #define read_sha1_lock()
 314 #define read_sha1_unlock()
 315
 316 static int wait_all(void)
 317 {
 318         return 0;
 319 }
 320 #endif
 321
 322 static int grep_config(const char *var, const char *value, void *cb)
 323 {
 324         struct grep_opt *opt = cb;
 325         char *color = NULL;
 326
 327         switch (userdiff_config(var, value)) {
 328         case 0: break;
 329         case -1: return -1;
 330         default: return 0;
 331         }
 332
 333         if (!strcmp(var, "grep.extendedregexp")) {
 334                 if (git_config_bool(var, value))
 335                         opt->regflags |= REG_EXTENDED;
 336                 else
 337                         opt->regflags &= ~REG_EXTENDED;
 338                 return 0;
 339         }
 340
 341         if (!strcmp(var, "grep.linenumber")) {
 342                 opt->linenum = git_config_bool(var, value);
 343                 return 0;
 344         }
 345
 346         if (!strcmp(var, "color.grep"))
 347                 opt->color = git_config_colorbool(var, value);
 348         else if (!strcmp(var, "color.grep.context"))
 349                 color = opt->color_context;
 350         else if (!strcmp(var, "color.grep.filename"))
 351                 color = opt->color_filename;
 352         else if (!strcmp(var, "color.grep.function"))
 353                 color = opt->color_function;
 354         else if (!strcmp(var, "color.grep.linenumber"))
 355                 color = opt->color_lineno;
 356         else if (!strcmp(var, "color.grep.match"))
 357                 color = opt->color_match;
 358         else if (!strcmp(var, "color.grep.selected"))
 359                 color = opt->color_selected;
 360         else if (!strcmp(var, "color.grep.separator"))
 361                 color = opt->color_sep;
 362         else
 363                 return git_color_default_config(var, value, cb);
 364         if (color) {
 365                 if (!value)
 366                         return config_error_nonbool(var);
 367                 color_parse(value, var, color);
 368         }
 369         return 0;
 370 }
 371
 372 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 373 {
 374         void *data;
 375
 376         if (use_threads) {
 377                 read_sha1_lock();
 378                 data = read_sha1_file(sha1, type, size);
 379                 read_sha1_unlock();
 380         } else {
 381                 data = read_sha1_file(sha1, type, size);
 382         }
 383         return data;
 384 }
 385
 386 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 387                        const char *name)
 388 {
 389         enum object_type type;
 390         void *data = lock_and_read_sha1_file(sha1, &type, size);
 391
 392         if (!data)
 393                 error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
 394
 395         return data;
 396 }
 397
 398 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 399                      const char *filename, int tree_name_len)
 400 {
 401         struct strbuf pathbuf = STRBUF_INIT;
 402         char *name;
 403
 404         if (opt->relative && opt->prefix_length) {
 405                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 406                                     opt->prefix);
 407                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 408         } else {
 409                 strbuf_addstr(&pathbuf, filename);
 410         }
 411
 412         name = strbuf_detach(&pathbuf, NULL);
 413
 414 #ifndef NO_PTHREADS
 415         if (use_threads) {
 416                 grep_sha1_async(opt, name, sha1);
 417                 return 0;
 418         } else
 419 #endif
 420         {
 421                 int hit;
 422                 unsigned long sz;
 423                 void *data = load_sha1(sha1, &sz, name);
 424                 if (!data)
 425                         hit = 0;
 426                 else
 427                         hit = grep_buffer(opt, name, data, sz);
 428
 429                 free(data);
 430                 free(name);
 431                 return hit;
 432         }
 433 }
 434
 435 static void *load_file(const char *filename, size_t *sz)
 436 {
 437         struct stat st;
 438         char *data;
 439         int i;
 440
 441         if (lstat(filename, &st) < 0) {
 442         err_ret:
 443                 if (errno != ENOENT)
 444                         error(_("'%s': %s"), filename, strerror(errno));
 445                 return NULL;
 446         }
 447         if (!S_ISREG(st.st_mode))
 448                 return NULL;
 449         *sz = xsize_t(st.st_size);
 450         i = open(filename, O_RDONLY);
 451         if (i < 0)
 452                 goto err_ret;
 453         data = xmalloc(*sz + 1);
 454         if (st.st_size != read_in_full(i, data, *sz)) {
 455                 error(_("'%s': short read %s"), filename, strerror(errno));
 456                 close(i);
 457                 free(data);
 458                 return NULL;
 459         }
 460         close(i);
 461         data[*sz] = 0;
 462         return data;
 463 }
 464
 465 static int grep_file(struct grep_opt *opt, const char *filename)
 466 {
 467         struct strbuf buf = STRBUF_INIT;
 468         char *name;
 469
 470         if (opt->relative && opt->prefix_length)
 471                 quote_path_relative(filename, -1, &buf, opt->prefix);
 472         else
 473                 strbuf_addstr(&buf, filename);
 474         name = strbuf_detach(&buf, NULL);
 475
 476 #ifndef NO_PTHREADS
 477         if (use_threads) {
 478                 grep_file_async(opt, name, filename);
 479                 return 0;
 480         } else
 481 #endif
 482         {
 483                 int hit;
 484                 size_t sz;
 485                 void *data = load_file(filename, &sz);
 486                 if (!data)
 487                         hit = 0;
 488                 else
 489                         hit = grep_buffer(opt, name, data, sz);
 490
 491                 free(data);
 492                 free(name);
 493                 return hit;
 494         }
 495 }
 496
 497 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 498 {
 499         struct string_list *path_list = opt->output_priv;
 500
 501         if (len == 1 && *(const char *)data == '\0')
 502                 return;
 503         string_list_append(path_list, xstrndup(data, len));
 504 }
 505
 506 static void run_pager(struct grep_opt *opt, const char *prefix)
 507 {
 508         struct string_list *path_list = opt->output_priv;
 509         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 510         int i, status;
 511
 512         for (i = 0; i < path_list->nr; i++)
 513                 argv[i] = path_list->items[i].string;
 514         argv[path_list->nr] = NULL;
 515
 516         if (prefix && chdir(prefix))
 517                 die(_("Failed to chdir: %s"), prefix);
 518         status = run_command_v_opt(argv, RUN_USING_SHELL);
 519         if (status)
 520                 exit(status);
 521         free(argv);
 522 }
 523
 524 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 525 {
 526         int hit = 0;
 527         int nr;
 528         read_cache();
 529
 530         for (nr = 0; nr < active_nr; nr++) {
 531                 struct cache_entry *ce = active_cache[nr];
 532                 if (!S_ISREG(ce->ce_mode))
 533                         continue;
 534                 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 535                         continue;
 536                 /*
 537                  * If CE_VALID is on, we assume worktree file and its cache entry
 538                  * are identical, even if worktree file has been modified, so use
 539                  * cache version instead
 540                  */
 541                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 542                         if (ce_stage(ce))
 543                                 continue;
 544                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 545                 }
 546                 else
 547                         hit |= grep_file(opt, ce->name);
 548                 if (ce_stage(ce)) {
 549                         do {
 550                                 nr++;
 551                         } while (nr < active_nr &&
 552                                  !strcmp(ce->name, active_cache[nr]->name));
 553                         nr--; /* compensate for loop control */
 554                 }
 555                 if (hit && opt->status_only)
 556                         break;
 557         }
 558         return hit;
 559 }
 560
 561 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 562                      struct tree_desc *tree, struct strbuf *base, int tn_len)
 563 {
 564         int hit = 0, match = 0;
 565         struct name_entry entry;
 566         int old_baselen = base->len;
 567
 568         while (tree_entry(tree, &entry)) {
 569                 int te_len = tree_entry_len(entry.path, entry.sha1);
 570
 571                 if (match != 2) {
 572                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 573                         if (match < 0)
 574                                 break;
 575                         if (match == 0)
 576                                 continue;
 577                 }
 578
 579                 strbuf_add(base, entry.path, te_len);
 580
 581                 if (S_ISREG(entry.mode)) {
 582                         hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 583                 }
 584                 else if (S_ISDIR(entry.mode)) {
 585                         enum object_type type;
 586                         struct tree_desc sub;
 587                         void *data;
 588                         unsigned long size;
 589
 590                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 591                         if (!data)
 592                                 die(_("unable to read tree (%s)"),
 593                                     sha1_to_hex(entry.sha1));
 594
 595                         strbuf_addch(base, '/');
 596                         init_tree_desc(&sub, data, size);
 597                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 598                         free(data);
 599                 }
 600                 strbuf_setlen(base, old_baselen);
 601
 602                 if (hit && opt->status_only)
 603                         break;
 604         }
 605         return hit;
 606 }
 607
 608 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 609                        struct object *obj, const char *name)
 610 {
 611         if (obj->type == OBJ_BLOB)
 612                 return grep_sha1(opt, obj->sha1, name, 0);
 613         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 614                 struct tree_desc tree;
 615                 void *data;
 616                 unsigned long size;
 617                 struct strbuf base;
 618                 int hit, len;
 619
 620                 read_sha1_lock();
 621                 data = read_object_with_reference(obj->sha1, tree_type,
 622                                                   &size, NULL);
 623                 read_sha1_unlock();
 624
 625                 if (!data)
 626                         die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 627
 628                 len = name ? strlen(name) : 0;
 629                 strbuf_init(&base, PATH_MAX + len + 1);
 630                 if (len) {
 631                         strbuf_add(&base, name, len);
 632                         strbuf_addch(&base, ':');
 633                 }
 634                 init_tree_desc(&tree, data, size);
 635                 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 636                 strbuf_release(&base);
 637                 free(data);
 638                 return hit;
 639         }
 640         die(_("unable to grep from object of type %s"), typename(obj->type));
 641 }
 642
 643 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 644                         const struct object_array *list)
 645 {
 646         unsigned int i;
 647         int hit = 0;
 648         const unsigned int nr = list->nr;
 649
 650         for (i = 0; i < nr; i++) {
 651                 struct object *real_obj;
 652                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 653                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 654                         hit = 1;
 655                         if (opt->status_only)
 656                                 break;
 657                 }
 658         }
 659         return hit;
 660 }
 661
 662 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 663                           int exc_std)
 664 {
 665         struct dir_struct dir;
 666         int i, hit = 0;
 667
 668         memset(&dir, 0, sizeof(dir));
 669         if (exc_std)
 670                 setup_standard_excludes(&dir);
 671
 672         fill_directory(&dir, pathspec->raw);
 673         for (i = 0; i < dir.nr; i++) {
 674                 const char *name = dir.entries[i]->name;
 675                 int namelen = strlen(name);
 676                 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 677                         continue;
 678                 hit |= grep_file(opt, dir.entries[i]->name);
 679                 if (hit && opt->status_only)
 680                         break;
 681         }
 682         return hit;
 683 }
 684
 685 static int context_callback(const struct option *opt, const char *arg,
 686                             int unset)
 687 {
 688         struct grep_opt *grep_opt = opt->value;
 689         int value;
 690         const char *endp;
 691
 692         if (unset) {
 693                 grep_opt->pre_context = grep_opt->post_context = 0;
 694                 return 0;
 695         }
 696         value = strtol(arg, (char **)&endp, 10);
 697         if (*endp) {
 698                 return error(_("switch `%c' expects a numerical value"),
 699                              opt->short_name);
 700         }
 701         grep_opt->pre_context = grep_opt->post_context = value;
 702         return 0;
 703 }
 704
 705 static int file_callback(const struct option *opt, const char *arg, int unset)
 706 {
 707         struct grep_opt *grep_opt = opt->value;
 708         int from_stdin = !strcmp(arg, "-");
 709         FILE *patterns;
 710         int lno = 0;
 711         struct strbuf sb = STRBUF_INIT;
 712
 713         patterns = from_stdin ? stdin : fopen(arg, "r");
 714         if (!patterns)
 715                 die_errno(_("cannot open '%s'"), arg);
 716         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 717                 char *s;
 718                 size_t len;
 719
 720                 /* ignore empty line like grep does */
 721                 if (sb.len == 0)
 722                         continue;
 723
 724                 s = strbuf_detach(&sb, &len);
 725                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 726         }
 727         if (!from_stdin)
 728                 fclose(patterns);
 729         strbuf_release(&sb);
 730         return 0;
 731 }
 732
 733 static int not_callback(const struct option *opt, const char *arg, int unset)
 734 {
 735         struct grep_opt *grep_opt = opt->value;
 736         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 737         return 0;
 738 }
 739
 740 static int and_callback(const struct option *opt, const char *arg, int unset)
 741 {
 742         struct grep_opt *grep_opt = opt->value;
 743         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 744         return 0;
 745 }
 746
 747 static int open_callback(const struct option *opt, const char *arg, int unset)
 748 {
 749         struct grep_opt *grep_opt = opt->value;
 750         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 751         return 0;
 752 }
 753
 754 static int close_callback(const struct option *opt, const char *arg, int unset)
 755 {
 756         struct grep_opt *grep_opt = opt->value;
 757         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 758         return 0;
 759 }
 760
 761 static int pattern_callback(const struct option *opt, const char *arg,
 762                             int unset)
 763 {
 764         struct grep_opt *grep_opt = opt->value;
 765         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 766         return 0;
 767 }
 768
 769 static int help_callback(const struct option *opt, const char *arg, int unset)
 770 {
 771         return -1;
 772 }
 773
 774 int cmd_grep(int argc, const char **argv, const char *prefix)
 775 {
 776         int hit = 0;
 777         int cached = 0, untracked = 0, opt_exclude = -1;
 778         int seen_dashdash = 0;
 779         int external_grep_allowed__ignored;
 780         const char *show_in_pager = NULL, *default_pager = "dummy";
 781         struct grep_opt opt;
 782         struct object_array list = OBJECT_ARRAY_INIT;
 783         const char **paths = NULL;
 784         struct pathspec pathspec;
 785         struct string_list path_list = STRING_LIST_INIT_NODUP;
 786         int i;
 787         int dummy;
 788         int use_index = 1;
 789         enum {
 790                 pattern_type_unspecified = 0,
 791                 pattern_type_bre,
 792                 pattern_type_ere,
 793                 pattern_type_fixed,
 794                 pattern_type_pcre,
 795         };
 796         int pattern_type = pattern_type_unspecified;
 797
 798         struct option options[] = {
 799                 OPT_BOOLEAN(0, "cached", &cached,
 800                         "search in index instead of in the work tree"),
 801                 { OPTION_BOOLEAN, 0, "index", &use_index, NULL,
 802                         "finds in contents not managed by git",
 803                         PARSE_OPT_NOARG | PARSE_OPT_NEGHELP },
 804                 OPT_BOOLEAN(0, "untracked", &untracked,
 805                         "search in both tracked and untracked files"),
 806                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 807                             "search also in ignored files", 1),
 808                 OPT_GROUP(""),
 809                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 810                         "show non-matching lines"),
 811                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 812                         "case insensitive matching"),
 813                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 814                         "match patterns only at word boundaries"),
 815                 OPT_SET_INT('a', "text", &opt.binary,
 816                         "process binary files as text", GREP_BINARY_TEXT),
 817                 OPT_SET_INT('I', NULL, &opt.binary,
 818                         "don't match patterns in binary files",
 819                         GREP_BINARY_NOMATCH),
 820                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 821                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 822                         NULL, 1 },
 823                 OPT_GROUP(""),
 824                 OPT_SET_INT('E', "extended-regexp", &pattern_type,
 825                             "use extended POSIX regular expressions",
 826                             pattern_type_ere),
 827                 OPT_SET_INT('G', "basic-regexp", &pattern_type,
 828                             "use basic POSIX regular expressions (default)",
 829                             pattern_type_bre),
 830                 OPT_SET_INT('F', "fixed-strings", &pattern_type,
 831                             "interpret patterns as fixed strings",
 832                             pattern_type_fixed),
 833                 OPT_SET_INT('P', "perl-regexp", &pattern_type,
 834                             "use Perl-compatible regular expressions",
 835                             pattern_type_pcre),
 836                 OPT_GROUP(""),
 837                 OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
 838                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 839                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 840                 OPT_NEGBIT(0, "full-name", &opt.relative,
 841                         "show filenames relative to top directory", 1),
 842                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 843                         "show only filenames instead of matching lines"),
 844                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 845                         "synonym for --files-with-matches"),
 846                 OPT_BOOLEAN('L', "files-without-match",
 847                         &opt.unmatch_name_only,
 848                         "show only the names of files without match"),
 849                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 850                         "print NUL after filenames"),
 851                 OPT_BOOLEAN('c', "count", &opt.count,
 852                         "show the number of matches instead of matching lines"),
 853                 OPT__COLOR(&opt.color, "highlight matches"),
 854                 OPT_BOOLEAN(0, "break", &opt.file_break,
 855                         "print empty line between matches from different files"),
 856                 OPT_BOOLEAN(0, "heading", &opt.heading,
 857                         "show filename only once above matches from same file"),
 858                 OPT_GROUP(""),
 859                 OPT_CALLBACK('C', "context", &opt, "n",
 860                         "show <n> context lines before and after matches",
 861                         context_callback),
 862                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 863                         "show <n> context lines before matches"),
 864                 OPT_INTEGER('A', "after-context", &opt.post_context,
 865                         "show <n> context lines after matches"),
 866                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 867                         context_callback),
 868                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 869                         "show a line with the function name before matches"),
 870                 OPT_BOOLEAN('W', "function-context", &opt.funcbody,
 871                         "show the surrounding function"),
 872                 OPT_GROUP(""),
 873                 OPT_CALLBACK('f', NULL, &opt, "file",
 874                         "read patterns from file", file_callback),
 875                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 876                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 877                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 878                   "combine patterns specified with -e",
 879                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 880                 OPT_BOOLEAN(0, "or", &dummy, ""),
 881                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 882                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 883                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 884                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 885                   open_callback },
 886                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 887                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 888                   close_callback },
 889                 OPT__QUIET(&opt.status_only,
 890                            "indicate hit with exit status without output"),
 891                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 892                         "show only matches from files that match all patterns"),
 893                 OPT_GROUP(""),
 894                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 895                         "pager", "show matching files in the pager",
 896                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 897                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 898                             "allow calling of grep(1) (ignored by this build)"),
 899                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 900                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 901                 OPT_END()
 902         };
 903
 904         /*
 905          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 906          * to show usage information and exit.
 907          */
 908         if (argc == 2 && !strcmp(argv[1], "-h"))
 909                 usage_with_options(grep_usage, options);
 910
 911         memset(&opt, 0, sizeof(opt));
 912         opt.prefix = prefix;
 913         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 914         opt.relative = 1;
 915         opt.pathname = 1;
 916         opt.pattern_tail = &opt.pattern_list;
 917         opt.header_tail = &opt.header_list;
 918         opt.regflags = REG_NEWLINE;
 919         opt.max_depth = -1;
 920
 921         strcpy(opt.color_context, "");
 922         strcpy(opt.color_filename, "");
 923         strcpy(opt.color_function, "");
 924         strcpy(opt.color_lineno, "");
 925         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 926         strcpy(opt.color_selected, "");
 927         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 928         opt.color = -1;
 929         git_config(grep_config, &opt);
 930
 931         /*
 932          * If there is no -- then the paths must exist in the working
 933          * tree.  If there is no explicit pattern specified with -e or
 934          * -f, we take the first unrecognized non option to be the
 935          * pattern, but then what follows it must be zero or more
 936          * valid refs up to the -- (if exists), and then existing
 937          * paths.  If there is an explicit pattern, then the first
 938          * unrecognized non option is the beginning of the refs list
 939          * that continues up to the -- (if exists), and then paths.
 940          */
 941         argc = parse_options(argc, argv, prefix, options, grep_usage,
 942                              PARSE_OPT_KEEP_DASHDASH |
 943                              PARSE_OPT_STOP_AT_NON_OPTION |
 944                              PARSE_OPT_NO_INTERNAL_HELP);
 945         switch (pattern_type) {
 946         case pattern_type_fixed:
 947                 opt.fixed = 1;
 948                 opt.pcre = 0;
 949                 break;
 950         case pattern_type_bre:
 951                 opt.fixed = 0;
 952                 opt.pcre = 0;
 953                 opt.regflags &= ~REG_EXTENDED;
 954                 break;
 955         case pattern_type_ere:
 956                 opt.fixed = 0;
 957                 opt.pcre = 0;
 958                 opt.regflags |= REG_EXTENDED;
 959                 break;
 960         case pattern_type_pcre:
 961                 opt.fixed = 0;
 962                 opt.pcre = 1;
 963                 break;
 964         default:
 965                 break; /* nothing */
 966         }
 967
 968         if (use_index && !startup_info->have_repository)
 969                 /* die the same way as if we did it at the beginning */
 970                 setup_git_directory();
 971
 972         /*
 973          * skip a -- separator; we know it cannot be
 974          * separating revisions from pathnames if
 975          * we haven't even had any patterns yet
 976          */
 977         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 978                 argv++;
 979                 argc--;
 980         }
 981
 982         /* First unrecognized non-option token */
 983         if (argc > 0 && !opt.pattern_list) {
 984                 append_grep_pattern(&opt, argv[0], "command line", 0,
 985                                     GREP_PATTERN);
 986                 argv++;
 987                 argc--;
 988         }
 989
 990         if (show_in_pager == default_pager)
 991                 show_in_pager = git_pager(1);
 992         if (show_in_pager) {
 993                 opt.color = 0;
 994                 opt.name_only = 1;
 995                 opt.null_following_name = 1;
 996                 opt.output_priv = &path_list;
 997                 opt.output = append_path;
 998                 string_list_append(&path_list, show_in_pager);
 999                 use_threads = 0;
1000         }
1001
1002         if (!opt.pattern_list)
1003                 die(_("no pattern given."));
1004         if (!opt.fixed && opt.ignore_case)
1005                 opt.regflags |= REG_ICASE;
1006
1007 #ifndef NO_PTHREADS
1008         if (online_cpus() == 1 || !grep_threads_ok(&opt))
1009                 use_threads = 0;
1010
1011         if (use_threads) {
1012                 if (opt.pre_context || opt.post_context || opt.file_break ||
1013                     opt.funcbody)
1014                         skip_first_line = 1;
1015                 start_threads(&opt);
1016         }
1017 #else
1018         use_threads = 0;
1019 #endif
1020
1021         compile_grep_patterns(&opt);
1022
1023         /* Check revs and then paths */
1024         for (i = 0; i < argc; i++) {
1025                 const char *arg = argv[i];
1026                 unsigned char sha1[20];
1027                 /* Is it a rev? */
1028                 if (!get_sha1(arg, sha1)) {
1029                         struct object *object = parse_object(sha1);
1030                         if (!object)
1031                                 die(_("bad object %s"), arg);
1032                         add_object_array(object, arg, &list);
1033                         continue;
1034                 }
1035                 if (!strcmp(arg, "--")) {
1036                         i++;
1037                         seen_dashdash = 1;
1038                 }
1039                 break;
1040         }
1041
1042         /* The rest are paths */
1043         if (!seen_dashdash) {
1044                 int j;
1045                 for (j = i; j < argc; j++)
1046                         verify_filename(prefix, argv[j]);
1047         }
1048
1049         paths = get_pathspec(prefix, argv + i);
1050         init_pathspec(&pathspec, paths);
1051         pathspec.max_depth = opt.max_depth;
1052         pathspec.recursive = 1;
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 (!strcmp("less", pager) || !strcmp("vi", pager)) {
1065                         struct strbuf buf = STRBUF_INIT;
1066                         strbuf_addf(&buf, "+/%s%s",
1067                                         strcmp("less", pager) ? "" : "*",
1068                                         opt.pattern_list->pattern);
1069                         string_list_append(&path_list, buf.buf);
1070                         strbuf_detach(&buf, NULL);
1071                 }
1072         }
1073
1074         if (!show_in_pager)
1075                 setup_pager();
1076
1077         if (!use_index && (untracked || cached))
1078                 die(_("--cached or --untracked cannot be used with --no-index."));
1079
1080         if (!use_index || untracked) {
1081                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1082                 if (list.nr)
1083                         die(_("--no-index or --untracked cannot be used with revs."));
1084                 hit = grep_directory(&opt, &pathspec, use_exclude);
1085         } else if (0 <= opt_exclude) {
1086                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1087         } else if (!list.nr) {
1088                 if (!cached)
1089                         setup_work_tree();
1090
1091                 hit = grep_cache(&opt, &pathspec, cached);
1092         } else {
1093                 if (cached)
1094                         die(_("both --cached and trees are given."));
1095                 hit = grep_objects(&opt, &pathspec, &list);
1096         }
1097
1098         if (use_threads)
1099                 hit |= wait_all();
1100         if (hit && show_in_pager)
1101                 run_pager(&opt, prefix);
1102         free_grep_patterns(&opt);
1103         return !hit;
1104 }