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
  21 static char const * const grep_usage[] = {
  22         "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
  23         NULL
  24 };
  25
  26 static int use_threads = 1;
  27
  28 #ifndef NO_PTHREADS
  29 #define THREADS 8
  30 static pthread_t threads[THREADS];
  31
  32 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  33                        const char *name);
  34 static void *load_file(const char *filename, size_t *sz);
  35
  36 enum work_type {WORK_SHA1, WORK_FILE};
  37
  38 /* We use one producer thread and THREADS consumer
  39  * threads. The producer adds struct work_items to 'todo' and the
  40  * consumers pick work items from the same array.
  41  */
  42 struct work_item {
  43         enum work_type type;
  44         char *name;
  45
  46         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  47          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  48          * terminated filename.
  49          */
  50         void *identifier;
  51         char done;
  52         struct strbuf out;
  53 };
  54
  55 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  56  * that have been or are processed by a consumer thread. We haven't
  57  * written the result for these to stdout yet.
  58  *
  59  * The work_items in [todo_start, todo_end) are waiting to be picked
  60  * up by a consumer thread.
  61  *
  62  * The ranges are modulo TODO_SIZE.
  63  */
  64 #define TODO_SIZE 128
  65 static struct work_item todo[TODO_SIZE];
  66 static int todo_start;
  67 static int todo_end;
  68 static int todo_done;
  69
  70 /* Has all work items been added? */
  71 static int all_work_added;
  72
  73 /* This lock protects all the variables above. */
  74 static pthread_mutex_t grep_mutex;
  75
  76 static inline void grep_lock(void)
  77 {
  78         if (use_threads)
  79                 pthread_mutex_lock(&grep_mutex);
  80 }
  81
  82 static inline void grep_unlock(void)
  83 {
  84         if (use_threads)
  85                 pthread_mutex_unlock(&grep_mutex);
  86 }
  87
  88 /* Used to serialize calls to read_sha1_file. */
  89 static pthread_mutex_t read_sha1_mutex;
  90
  91 static inline void read_sha1_lock(void)
  92 {
  93         if (use_threads)
  94                 pthread_mutex_lock(&read_sha1_mutex);
  95 }
  96
  97 static inline void read_sha1_unlock(void)
  98 {
  99         if (use_threads)
 100                 pthread_mutex_unlock(&read_sha1_mutex);
 101 }
 102
 103 /* Signalled when a new work_item is added to todo. */
 104 static pthread_cond_t cond_add;
 105
 106 /* Signalled when the result from one work_item is written to
 107  * stdout.
 108  */
 109 static pthread_cond_t cond_write;
 110
 111 /* Signalled when we are finished with everything. */
 112 static pthread_cond_t cond_result;
 113
 114 static int skip_first_line;
 115
 116 static void add_work(enum work_type type, char *name, void *id)
 117 {
 118         grep_lock();
 119
 120         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 121                 pthread_cond_wait(&cond_write, &grep_mutex);
 122         }
 123
 124         todo[todo_end].type = type;
 125         todo[todo_end].name = name;
 126         todo[todo_end].identifier = id;
 127         todo[todo_end].done = 0;
 128         strbuf_reset(&todo[todo_end].out);
 129         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 130
 131         pthread_cond_signal(&cond_add);
 132         grep_unlock();
 133 }
 134
 135 static struct work_item *get_work(void)
 136 {
 137         struct work_item *ret;
 138
 139         grep_lock();
 140         while (todo_start == todo_end && !all_work_added) {
 141                 pthread_cond_wait(&cond_add, &grep_mutex);
 142         }
 143
 144         if (todo_start == todo_end && all_work_added) {
 145                 ret = NULL;
 146         } else {
 147                 ret = &todo[todo_start];
 148                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 149         }
 150         grep_unlock();
 151         return ret;
 152 }
 153
 154 static void grep_sha1_async(struct grep_opt *opt, char *name,
 155                             const unsigned char *sha1)
 156 {
 157         unsigned char *s;
 158         s = xmalloc(20);
 159         memcpy(s, sha1, 20);
 160         add_work(WORK_SHA1, name, s);
 161 }
 162
 163 static void grep_file_async(struct grep_opt *opt, char *name,
 164                             const char *filename)
 165 {
 166         add_work(WORK_FILE, name, xstrdup(filename));
 167 }
 168
 169 static void work_done(struct work_item *w)
 170 {
 171         int old_done;
 172
 173         grep_lock();
 174         w->done = 1;
 175         old_done = todo_done;
 176         for(; todo[todo_done].done && todo_done != todo_start;
 177             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 178                 w = &todo[todo_done];
 179                 if (w->out.len) {
 180                         const char *p = w->out.buf;
 181                         size_t len = w->out.len;
 182
 183                         /* Skip the leading hunk mark of the first file. */
 184                         if (skip_first_line) {
 185                                 while (len) {
 186                                         len--;
 187                                         if (*p++ == '\n')
 188                                                 break;
 189                                 }
 190                                 skip_first_line = 0;
 191                         }
 192
 193                         write_or_die(1, p, len);
 194                 }
 195                 free(w->name);
 196                 free(w->identifier);
 197         }
 198
 199         if (old_done != todo_done)
 200                 pthread_cond_signal(&cond_write);
 201
 202         if (all_work_added && todo_done == todo_end)
 203                 pthread_cond_signal(&cond_result);
 204
 205         grep_unlock();
 206 }
 207
 208 static void *run(void *arg)
 209 {
 210         int hit = 0;
 211         struct grep_opt *opt = arg;
 212
 213         while (1) {
 214                 struct work_item *w = get_work();
 215                 if (!w)
 216                         break;
 217
 218                 opt->output_priv = w;
 219                 if (w->type == WORK_SHA1) {
 220                         unsigned long sz;
 221                         void* data = load_sha1(w->identifier, &sz, w->name);
 222
 223                         if (data) {
 224                                 hit |= grep_buffer(opt, w->name, data, sz);
 225                                 free(data);
 226                         }
 227                 } else if (w->type == WORK_FILE) {
 228                         size_t sz;
 229                         void* data = load_file(w->identifier, &sz);
 230                         if (data) {
 231                                 hit |= grep_buffer(opt, w->name, data, sz);
 232                                 free(data);
 233                         }
 234                 } else {
 235                         assert(0);
 236                 }
 237
 238                 work_done(w);
 239         }
 240         free_grep_patterns(arg);
 241         free(arg);
 242
 243         return (void*) (intptr_t) hit;
 244 }
 245
 246 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 247 {
 248         struct work_item *w = opt->output_priv;
 249         strbuf_add(&w->out, buf, size);
 250 }
 251
 252 static void start_threads(struct grep_opt *opt)
 253 {
 254         int i;
 255
 256         pthread_mutex_init(&grep_mutex, NULL);
 257         pthread_mutex_init(&read_sha1_mutex, NULL);
 258         pthread_mutex_init(&grep_attr_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_mutex_destroy(&grep_attr_mutex);
 307         pthread_cond_destroy(&cond_add);
 308         pthread_cond_destroy(&cond_write);
 309         pthread_cond_destroy(&cond_result);
 310
 311         return hit;
 312 }
 313 #else /* !NO_PTHREADS */
 314 #define read_sha1_lock()
 315 #define read_sha1_unlock()
 316
 317 static int wait_all(void)
 318 {
 319         return 0;
 320 }
 321 #endif
 322
 323 static int grep_config(const char *var, const char *value, void *cb)
 324 {
 325         struct grep_opt *opt = cb;
 326         char *color = NULL;
 327
 328         switch (userdiff_config(var, value)) {
 329         case 0: break;
 330         case -1: return -1;
 331         default: return 0;
 332         }
 333
 334         if (!strcmp(var, "grep.extendedregexp")) {
 335                 if (git_config_bool(var, value))
 336                         opt->regflags |= REG_EXTENDED;
 337                 else
 338                         opt->regflags &= ~REG_EXTENDED;
 339                 return 0;
 340         }
 341
 342         if (!strcmp(var, "grep.linenumber")) {
 343                 opt->linenum = git_config_bool(var, value);
 344                 return 0;
 345         }
 346
 347         if (!strcmp(var, "color.grep"))
 348                 opt->color = git_config_colorbool(var, value);
 349         else if (!strcmp(var, "color.grep.context"))
 350                 color = opt->color_context;
 351         else if (!strcmp(var, "color.grep.filename"))
 352                 color = opt->color_filename;
 353         else if (!strcmp(var, "color.grep.function"))
 354                 color = opt->color_function;
 355         else if (!strcmp(var, "color.grep.linenumber"))
 356                 color = opt->color_lineno;
 357         else if (!strcmp(var, "color.grep.match"))
 358                 color = opt->color_match;
 359         else if (!strcmp(var, "color.grep.selected"))
 360                 color = opt->color_selected;
 361         else if (!strcmp(var, "color.grep.separator"))
 362                 color = opt->color_sep;
 363         else
 364                 return git_color_default_config(var, value, cb);
 365         if (color) {
 366                 if (!value)
 367                         return config_error_nonbool(var);
 368                 color_parse(value, var, color);
 369         }
 370         return 0;
 371 }
 372
 373 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 374 {
 375         void *data;
 376
 377         read_sha1_lock();
 378         data = read_sha1_file(sha1, type, size);
 379         read_sha1_unlock();
 380         return data;
 381 }
 382
 383 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 384                        const char *name)
 385 {
 386         enum object_type type;
 387         void *data = lock_and_read_sha1_file(sha1, &type, size);
 388
 389         if (!data)
 390                 error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
 391
 392         return data;
 393 }
 394
 395 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 396                      const char *filename, int tree_name_len)
 397 {
 398         struct strbuf pathbuf = STRBUF_INIT;
 399         char *name;
 400
 401         if (opt->relative && opt->prefix_length) {
 402                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 403                                     opt->prefix);
 404                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 405         } else {
 406                 strbuf_addstr(&pathbuf, filename);
 407         }
 408
 409         name = strbuf_detach(&pathbuf, NULL);
 410
 411 #ifndef NO_PTHREADS
 412         if (use_threads) {
 413                 grep_sha1_async(opt, name, sha1);
 414                 return 0;
 415         } else
 416 #endif
 417         {
 418                 int hit;
 419                 unsigned long sz;
 420                 void *data = load_sha1(sha1, &sz, name);
 421                 if (!data)
 422                         hit = 0;
 423                 else
 424                         hit = grep_buffer(opt, name, data, sz);
 425
 426                 free(data);
 427                 free(name);
 428                 return hit;
 429         }
 430 }
 431
 432 static void *load_file(const char *filename, size_t *sz)
 433 {
 434         struct stat st;
 435         char *data;
 436         int i;
 437
 438         if (lstat(filename, &st) < 0) {
 439         err_ret:
 440                 if (errno != ENOENT)
 441                         error(_("'%s': %s"), filename, strerror(errno));
 442                 return NULL;
 443         }
 444         if (!S_ISREG(st.st_mode))
 445                 return NULL;
 446         *sz = xsize_t(st.st_size);
 447         i = open(filename, O_RDONLY);
 448         if (i < 0)
 449                 goto err_ret;
 450         data = xmalloc(*sz + 1);
 451         if (st.st_size != read_in_full(i, data, *sz)) {
 452                 error(_("'%s': short read %s"), filename, strerror(errno));
 453                 close(i);
 454                 free(data);
 455                 return NULL;
 456         }
 457         close(i);
 458         data[*sz] = 0;
 459         return data;
 460 }
 461
 462 static int grep_file(struct grep_opt *opt, const char *filename)
 463 {
 464         struct strbuf buf = STRBUF_INIT;
 465         char *name;
 466
 467         if (opt->relative && opt->prefix_length)
 468                 quote_path_relative(filename, -1, &buf, opt->prefix);
 469         else
 470                 strbuf_addstr(&buf, filename);
 471         name = strbuf_detach(&buf, NULL);
 472
 473 #ifndef NO_PTHREADS
 474         if (use_threads) {
 475                 grep_file_async(opt, name, filename);
 476                 return 0;
 477         } else
 478 #endif
 479         {
 480                 int hit;
 481                 size_t sz;
 482                 void *data = load_file(filename, &sz);
 483                 if (!data)
 484                         hit = 0;
 485                 else
 486                         hit = grep_buffer(opt, name, data, sz);
 487
 488                 free(data);
 489                 free(name);
 490                 return hit;
 491         }
 492 }
 493
 494 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 495 {
 496         struct string_list *path_list = opt->output_priv;
 497
 498         if (len == 1 && *(const char *)data == '\0')
 499                 return;
 500         string_list_append(path_list, xstrndup(data, len));
 501 }
 502
 503 static void run_pager(struct grep_opt *opt, const char *prefix)
 504 {
 505         struct string_list *path_list = opt->output_priv;
 506         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 507         int i, status;
 508
 509         for (i = 0; i < path_list->nr; i++)
 510                 argv[i] = path_list->items[i].string;
 511         argv[path_list->nr] = NULL;
 512
 513         if (prefix && chdir(prefix))
 514                 die(_("Failed to chdir: %s"), prefix);
 515         status = run_command_v_opt(argv, RUN_USING_SHELL);
 516         if (status)
 517                 exit(status);
 518         free(argv);
 519 }
 520
 521 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 522 {
 523         int hit = 0;
 524         int nr;
 525         read_cache();
 526
 527         for (nr = 0; nr < active_nr; nr++) {
 528                 struct cache_entry *ce = active_cache[nr];
 529                 if (!S_ISREG(ce->ce_mode))
 530                         continue;
 531                 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 532                         continue;
 533                 /*
 534                  * If CE_VALID is on, we assume worktree file and its cache entry
 535                  * are identical, even if worktree file has been modified, so use
 536                  * cache version instead
 537                  */
 538                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 539                         if (ce_stage(ce))
 540                                 continue;
 541                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 542                 }
 543                 else
 544                         hit |= grep_file(opt, ce->name);
 545                 if (ce_stage(ce)) {
 546                         do {
 547                                 nr++;
 548                         } while (nr < active_nr &&
 549                                  !strcmp(ce->name, active_cache[nr]->name));
 550                         nr--; /* compensate for loop control */
 551                 }
 552                 if (hit && opt->status_only)
 553                         break;
 554         }
 555         return hit;
 556 }
 557
 558 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 559                      struct tree_desc *tree, struct strbuf *base, int tn_len)
 560 {
 561         int hit = 0;
 562         enum interesting match = entry_not_interesting;
 563         struct name_entry entry;
 564         int old_baselen = base->len;
 565
 566         while (tree_entry(tree, &entry)) {
 567                 int te_len = tree_entry_len(&entry);
 568
 569                 if (match != all_entries_interesting) {
 570                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 571                         if (match == all_entries_not_interesting)
 572                                 break;
 573                         if (match == entry_not_interesting)
 574                                 continue;
 575                 }
 576
 577                 strbuf_add(base, entry.path, te_len);
 578
 579                 if (S_ISREG(entry.mode)) {
 580                         hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 581                 }
 582                 else if (S_ISDIR(entry.mode)) {
 583                         enum object_type type;
 584                         struct tree_desc sub;
 585                         void *data;
 586                         unsigned long size;
 587
 588                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 589                         if (!data)
 590                                 die(_("unable to read tree (%s)"),
 591                                     sha1_to_hex(entry.sha1));
 592
 593                         strbuf_addch(base, '/');
 594                         init_tree_desc(&sub, data, size);
 595                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 596                         free(data);
 597                 }
 598                 strbuf_setlen(base, old_baselen);
 599
 600                 if (hit && opt->status_only)
 601                         break;
 602         }
 603         return hit;
 604 }
 605
 606 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 607                        struct object *obj, const char *name)
 608 {
 609         if (obj->type == OBJ_BLOB)
 610                 return grep_sha1(opt, obj->sha1, name, 0);
 611         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 612                 struct tree_desc tree;
 613                 void *data;
 614                 unsigned long size;
 615                 struct strbuf base;
 616                 int hit, len;
 617
 618                 read_sha1_lock();
 619                 data = read_object_with_reference(obj->sha1, tree_type,
 620                                                   &size, NULL);
 621                 read_sha1_unlock();
 622
 623                 if (!data)
 624                         die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 625
 626                 len = name ? strlen(name) : 0;
 627                 strbuf_init(&base, PATH_MAX + len + 1);
 628                 if (len) {
 629                         strbuf_add(&base, name, len);
 630                         strbuf_addch(&base, ':');
 631                 }
 632                 init_tree_desc(&tree, data, size);
 633                 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 634                 strbuf_release(&base);
 635                 free(data);
 636                 return hit;
 637         }
 638         die(_("unable to grep from object of type %s"), typename(obj->type));
 639 }
 640
 641 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 642                         const struct object_array *list)
 643 {
 644         unsigned int i;
 645         int hit = 0;
 646         const unsigned int nr = list->nr;
 647
 648         for (i = 0; i < nr; i++) {
 649                 struct object *real_obj;
 650                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 651                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 652                         hit = 1;
 653                         if (opt->status_only)
 654                                 break;
 655                 }
 656         }
 657         return hit;
 658 }
 659
 660 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 661                           int exc_std)
 662 {
 663         struct dir_struct dir;
 664         int i, hit = 0;
 665
 666         memset(&dir, 0, sizeof(dir));
 667         if (exc_std)
 668                 setup_standard_excludes(&dir);
 669
 670         fill_directory(&dir, pathspec->raw);
 671         for (i = 0; i < dir.nr; i++) {
 672                 const char *name = dir.entries[i]->name;
 673                 int namelen = strlen(name);
 674                 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 675                         continue;
 676                 hit |= grep_file(opt, dir.entries[i]->name);
 677                 if (hit && opt->status_only)
 678                         break;
 679         }
 680         return hit;
 681 }
 682
 683 static int context_callback(const struct option *opt, const char *arg,
 684                             int unset)
 685 {
 686         struct grep_opt *grep_opt = opt->value;
 687         int value;
 688         const char *endp;
 689
 690         if (unset) {
 691                 grep_opt->pre_context = grep_opt->post_context = 0;
 692                 return 0;
 693         }
 694         value = strtol(arg, (char **)&endp, 10);
 695         if (*endp) {
 696                 return error(_("switch `%c' expects a numerical value"),
 697                              opt->short_name);
 698         }
 699         grep_opt->pre_context = grep_opt->post_context = value;
 700         return 0;
 701 }
 702
 703 static int file_callback(const struct option *opt, const char *arg, int unset)
 704 {
 705         struct grep_opt *grep_opt = opt->value;
 706         int from_stdin = !strcmp(arg, "-");
 707         FILE *patterns;
 708         int lno = 0;
 709         struct strbuf sb = STRBUF_INIT;
 710
 711         patterns = from_stdin ? stdin : fopen(arg, "r");
 712         if (!patterns)
 713                 die_errno(_("cannot open '%s'"), arg);
 714         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 715                 char *s;
 716                 size_t len;
 717
 718                 /* ignore empty line like grep does */
 719                 if (sb.len == 0)
 720                         continue;
 721
 722                 s = strbuf_detach(&sb, &len);
 723                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 724         }
 725         if (!from_stdin)
 726                 fclose(patterns);
 727         strbuf_release(&sb);
 728         return 0;
 729 }
 730
 731 static int not_callback(const struct option *opt, const char *arg, int unset)
 732 {
 733         struct grep_opt *grep_opt = opt->value;
 734         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 735         return 0;
 736 }
 737
 738 static int and_callback(const struct option *opt, const char *arg, int unset)
 739 {
 740         struct grep_opt *grep_opt = opt->value;
 741         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 742         return 0;
 743 }
 744
 745 static int open_callback(const struct option *opt, const char *arg, int unset)
 746 {
 747         struct grep_opt *grep_opt = opt->value;
 748         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 749         return 0;
 750 }
 751
 752 static int close_callback(const struct option *opt, const char *arg, int unset)
 753 {
 754         struct grep_opt *grep_opt = opt->value;
 755         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 756         return 0;
 757 }
 758
 759 static int pattern_callback(const struct option *opt, const char *arg,
 760                             int unset)
 761 {
 762         struct grep_opt *grep_opt = opt->value;
 763         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 764         return 0;
 765 }
 766
 767 static int help_callback(const struct option *opt, const char *arg, int unset)
 768 {
 769         return -1;
 770 }
 771
 772 int cmd_grep(int argc, const char **argv, const char *prefix)
 773 {
 774         int hit = 0;
 775         int cached = 0, untracked = 0, opt_exclude = -1;
 776         int seen_dashdash = 0;
 777         int external_grep_allowed__ignored;
 778         const char *show_in_pager = NULL, *default_pager = "dummy";
 779         struct grep_opt opt;
 780         struct object_array list = OBJECT_ARRAY_INIT;
 781         const char **paths = NULL;
 782         struct pathspec pathspec;
 783         struct string_list path_list = STRING_LIST_INIT_NODUP;
 784         int i;
 785         int dummy;
 786         int use_index = 1;
 787         enum {
 788                 pattern_type_unspecified = 0,
 789                 pattern_type_bre,
 790                 pattern_type_ere,
 791                 pattern_type_fixed,
 792                 pattern_type_pcre,
 793         };
 794         int pattern_type = pattern_type_unspecified;
 795
 796         struct option options[] = {
 797                 OPT_BOOLEAN(0, "cached", &cached,
 798                         "search in index instead of in the work tree"),
 799                 { OPTION_BOOLEAN, 0, "index", &use_index, NULL,
 800                         "finds in contents not managed by git",
 801                         PARSE_OPT_NOARG | PARSE_OPT_NEGHELP },
 802                 OPT_BOOLEAN(0, "untracked", &untracked,
 803                         "search in both tracked and untracked files"),
 804                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 805                             "search also in ignored files", 1),
 806                 OPT_GROUP(""),
 807                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 808                         "show non-matching lines"),
 809                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 810                         "case insensitive matching"),
 811                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 812                         "match patterns only at word boundaries"),
 813                 OPT_SET_INT('a', "text", &opt.binary,
 814                         "process binary files as text", GREP_BINARY_TEXT),
 815                 OPT_SET_INT('I', NULL, &opt.binary,
 816                         "don't match patterns in binary files",
 817                         GREP_BINARY_NOMATCH),
 818                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 819                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 820                         NULL, 1 },
 821                 OPT_GROUP(""),
 822                 OPT_SET_INT('E', "extended-regexp", &pattern_type,
 823                             "use extended POSIX regular expressions",
 824                             pattern_type_ere),
 825                 OPT_SET_INT('G', "basic-regexp", &pattern_type,
 826                             "use basic POSIX regular expressions (default)",
 827                             pattern_type_bre),
 828                 OPT_SET_INT('F', "fixed-strings", &pattern_type,
 829                             "interpret patterns as fixed strings",
 830                             pattern_type_fixed),
 831                 OPT_SET_INT('P', "perl-regexp", &pattern_type,
 832                             "use Perl-compatible regular expressions",
 833                             pattern_type_pcre),
 834                 OPT_GROUP(""),
 835                 OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
 836                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 837                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 838                 OPT_NEGBIT(0, "full-name", &opt.relative,
 839                         "show filenames relative to top directory", 1),
 840                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 841                         "show only filenames instead of matching lines"),
 842                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 843                         "synonym for --files-with-matches"),
 844                 OPT_BOOLEAN('L', "files-without-match",
 845                         &opt.unmatch_name_only,
 846                         "show only the names of files without match"),
 847                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 848                         "print NUL after filenames"),
 849                 OPT_BOOLEAN('c', "count", &opt.count,
 850                         "show the number of matches instead of matching lines"),
 851                 OPT__COLOR(&opt.color, "highlight matches"),
 852                 OPT_BOOLEAN(0, "break", &opt.file_break,
 853                         "print empty line between matches from different files"),
 854                 OPT_BOOLEAN(0, "heading", &opt.heading,
 855                         "show filename only once above matches from same file"),
 856                 OPT_GROUP(""),
 857                 OPT_CALLBACK('C', "context", &opt, "n",
 858                         "show <n> context lines before and after matches",
 859                         context_callback),
 860                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 861                         "show <n> context lines before matches"),
 862                 OPT_INTEGER('A', "after-context", &opt.post_context,
 863                         "show <n> context lines after matches"),
 864                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 865                         context_callback),
 866                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 867                         "show a line with the function name before matches"),
 868                 OPT_BOOLEAN('W', "function-context", &opt.funcbody,
 869                         "show the surrounding function"),
 870                 OPT_GROUP(""),
 871                 OPT_CALLBACK('f', NULL, &opt, "file",
 872                         "read patterns from file", file_callback),
 873                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 874                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 875                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 876                   "combine patterns specified with -e",
 877                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 878                 OPT_BOOLEAN(0, "or", &dummy, ""),
 879                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 880                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 881                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 882                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 883                   open_callback },
 884                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 885                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 886                   close_callback },
 887                 OPT__QUIET(&opt.status_only,
 888                            "indicate hit with exit status without output"),
 889                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 890                         "show only matches from files that match all patterns"),
 891                 OPT_GROUP(""),
 892                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 893                         "pager", "show matching files in the pager",
 894                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 895                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 896                             "allow calling of grep(1) (ignored by this build)"),
 897                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 898                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 899                 OPT_END()
 900         };
 901
 902         /*
 903          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 904          * to show usage information and exit.
 905          */
 906         if (argc == 2 && !strcmp(argv[1], "-h"))
 907                 usage_with_options(grep_usage, options);
 908
 909         memset(&opt, 0, sizeof(opt));
 910         opt.prefix = prefix;
 911         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 912         opt.relative = 1;
 913         opt.pathname = 1;
 914         opt.pattern_tail = &opt.pattern_list;
 915         opt.header_tail = &opt.header_list;
 916         opt.regflags = REG_NEWLINE;
 917         opt.max_depth = -1;
 918
 919         strcpy(opt.color_context, "");
 920         strcpy(opt.color_filename, "");
 921         strcpy(opt.color_function, "");
 922         strcpy(opt.color_lineno, "");
 923         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 924         strcpy(opt.color_selected, "");
 925         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 926         opt.color = -1;
 927         git_config(grep_config, &opt);
 928
 929         /*
 930          * If there is no -- then the paths must exist in the working
 931          * tree.  If there is no explicit pattern specified with -e or
 932          * -f, we take the first unrecognized non option to be the
 933          * pattern, but then what follows it must be zero or more
 934          * valid refs up to the -- (if exists), and then existing
 935          * paths.  If there is an explicit pattern, then the first
 936          * unrecognized non option is the beginning of the refs list
 937          * that continues up to the -- (if exists), and then paths.
 938          */
 939         argc = parse_options(argc, argv, prefix, options, grep_usage,
 940                              PARSE_OPT_KEEP_DASHDASH |
 941                              PARSE_OPT_STOP_AT_NON_OPTION |
 942                              PARSE_OPT_NO_INTERNAL_HELP);
 943         switch (pattern_type) {
 944         case pattern_type_fixed:
 945                 opt.fixed = 1;
 946                 opt.pcre = 0;
 947                 break;
 948         case pattern_type_bre:
 949                 opt.fixed = 0;
 950                 opt.pcre = 0;
 951                 opt.regflags &= ~REG_EXTENDED;
 952                 break;
 953         case pattern_type_ere:
 954                 opt.fixed = 0;
 955                 opt.pcre = 0;
 956                 opt.regflags |= REG_EXTENDED;
 957                 break;
 958         case pattern_type_pcre:
 959                 opt.fixed = 0;
 960                 opt.pcre = 1;
 961                 break;
 962         default:
 963                 break; /* nothing */
 964         }
 965
 966         if (use_index && !startup_info->have_repository)
 967                 /* die the same way as if we did it at the beginning */
 968                 setup_git_directory();
 969
 970         /*
 971          * skip a -- separator; we know it cannot be
 972          * separating revisions from pathnames if
 973          * we haven't even had any patterns yet
 974          */
 975         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 976                 argv++;
 977                 argc--;
 978         }
 979
 980         /* First unrecognized non-option token */
 981         if (argc > 0 && !opt.pattern_list) {
 982                 append_grep_pattern(&opt, argv[0], "command line", 0,
 983                                     GREP_PATTERN);
 984                 argv++;
 985                 argc--;
 986         }
 987
 988         if (show_in_pager == default_pager)
 989                 show_in_pager = git_pager(1);
 990         if (show_in_pager) {
 991                 opt.color = 0;
 992                 opt.name_only = 1;
 993                 opt.null_following_name = 1;
 994                 opt.output_priv = &path_list;
 995                 opt.output = append_path;
 996                 string_list_append(&path_list, show_in_pager);
 997                 use_threads = 0;
 998         }
 999
1000         if (!opt.pattern_list)
1001                 die(_("no pattern given."));
1002         if (!opt.fixed && opt.ignore_case)
1003                 opt.regflags |= REG_ICASE;
1004
1005 #ifndef NO_PTHREADS
1006         if (online_cpus() == 1)
1007                 use_threads = 0;
1008 #else
1009         use_threads = 0;
1010 #endif
1011
1012         opt.use_threads = use_threads;
1013
1014 #ifndef NO_PTHREADS
1015         if (use_threads) {
1016                 if (opt.pre_context || opt.post_context || opt.file_break ||
1017                     opt.funcbody)
1018                         skip_first_line = 1;
1019                 start_threads(&opt);
1020         }
1021 #endif
1022
1023         compile_grep_patterns(&opt);
1024
1025         /* Check revs and then paths */
1026         for (i = 0; i < argc; i++) {
1027                 const char *arg = argv[i];
1028                 unsigned char sha1[20];
1029                 /* Is it a rev? */
1030                 if (!get_sha1(arg, sha1)) {
1031                         struct object *object = parse_object(sha1);
1032                         if (!object)
1033                                 die(_("bad object %s"), arg);
1034                         add_object_array(object, arg, &list);
1035                         continue;
1036                 }
1037                 if (!strcmp(arg, "--")) {
1038                         i++;
1039                         seen_dashdash = 1;
1040                 }
1041                 break;
1042         }
1043
1044         /* The rest are paths */
1045         if (!seen_dashdash) {
1046                 int j;
1047                 for (j = i; j < argc; j++)
1048                         verify_filename(prefix, argv[j]);
1049         }
1050
1051         paths = get_pathspec(prefix, argv + i);
1052         init_pathspec(&pathspec, paths);
1053         pathspec.max_depth = opt.max_depth;
1054         pathspec.recursive = 1;
1055
1056         if (show_in_pager && (cached || list.nr))
1057                 die(_("--open-files-in-pager only works on the worktree"));
1058
1059         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1060                 const char *pager = path_list.items[0].string;
1061                 int len = strlen(pager);
1062
1063                 if (len > 4 && is_dir_sep(pager[len - 5]))
1064                         pager += len - 4;
1065
1066                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1067                         struct strbuf buf = STRBUF_INIT;
1068                         strbuf_addf(&buf, "+/%s%s",
1069                                         strcmp("less", pager) ? "" : "*",
1070                                         opt.pattern_list->pattern);
1071                         string_list_append(&path_list, buf.buf);
1072                         strbuf_detach(&buf, NULL);
1073                 }
1074         }
1075
1076         if (!show_in_pager)
1077                 setup_pager();
1078
1079         if (!use_index && (untracked || cached))
1080                 die(_("--cached or --untracked cannot be used with --no-index."));
1081
1082         if (!use_index || untracked) {
1083                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1084                 if (list.nr)
1085                         die(_("--no-index or --untracked cannot be used with revs."));
1086                 hit = grep_directory(&opt, &pathspec, use_exclude);
1087         } else if (0 <= opt_exclude) {
1088                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1089         } else if (!list.nr) {
1090                 if (!cached)
1091                         setup_work_tree();
1092
1093                 hit = grep_cache(&opt, &pathspec, cached);
1094         } else {
1095                 if (cached)
1096                         die(_("both --cached and trees are given."));
1097                 hit = grep_objects(&opt, &pathspec, &list);
1098         }
1099
1100         if (use_threads)
1101                 hit |= wait_all();
1102         if (hit && show_in_pager)
1103                 run_pager(&opt, prefix);
1104         free_grep_patterns(&opt);
1105         return !hit;
1106 }