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 #include "attr.h"
  22
  23 static char const * const grep_usage[] = {
  24         "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
  25         NULL
  26 };
  27
  28 static int use_threads = 1;
  29
  30 #ifndef NO_PTHREADS
  31 #define THREADS 8
  32 static pthread_t threads[THREADS];
  33
  34 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  35                        const char *name);
  36 static void *load_file(const char *filename, size_t *sz);
  37
  38 enum work_type {WORK_SHA1, WORK_FILE};
  39
  40 /* We use one producer thread and THREADS consumer
  41  * threads. The producer adds struct work_items to 'todo' and the
  42  * consumers pick work items from the same array.
  43  */
  44 struct work_item {
  45         enum work_type type;
  46         char *name;
  47
  48         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  49          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  50          * terminated filename.
  51          */
  52         void *identifier;
  53         char done;
  54         struct strbuf out;
  55 };
  56
  57 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  58  * that have been or are processed by a consumer thread. We haven't
  59  * written the result for these to stdout yet.
  60  *
  61  * The work_items in [todo_start, todo_end) are waiting to be picked
  62  * up by a consumer thread.
  63  *
  64  * The ranges are modulo TODO_SIZE.
  65  */
  66 #define TODO_SIZE 128
  67 static struct work_item todo[TODO_SIZE];
  68 static int todo_start;
  69 static int todo_end;
  70 static int todo_done;
  71
  72 /* Has all work items been added? */
  73 static int all_work_added;
  74
  75 /* This lock protects all the variables above. */
  76 static pthread_mutex_t grep_mutex;
  77
  78 /* Used to serialize calls to read_sha1_file. */
  79 static pthread_mutex_t read_sha1_mutex;
  80
  81 #define grep_lock() pthread_mutex_lock(&grep_mutex)
  82 #define grep_unlock() pthread_mutex_unlock(&grep_mutex)
  83 #define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
  84 #define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
  85
  86 /* Signalled when a new work_item is added to todo. */
  87 static pthread_cond_t cond_add;
  88
  89 /* Signalled when the result from one work_item is written to
  90  * stdout.
  91  */
  92 static pthread_cond_t cond_write;
  93
  94 /* Signalled when we are finished with everything. */
  95 static pthread_cond_t cond_result;
  96
  97 static int print_hunk_marks_between_files;
  98 static int printed_something;
  99
 100 static void add_work(enum work_type type, char *name, void *id)
 101 {
 102         grep_lock();
 103
 104         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 105                 pthread_cond_wait(&cond_write, &grep_mutex);
 106         }
 107
 108         todo[todo_end].type = type;
 109         todo[todo_end].name = name;
 110         todo[todo_end].identifier = id;
 111         todo[todo_end].done = 0;
 112         strbuf_reset(&todo[todo_end].out);
 113         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 114
 115         pthread_cond_signal(&cond_add);
 116         grep_unlock();
 117 }
 118
 119 static struct work_item *get_work(void)
 120 {
 121         struct work_item *ret;
 122
 123         grep_lock();
 124         while (todo_start == todo_end && !all_work_added) {
 125                 pthread_cond_wait(&cond_add, &grep_mutex);
 126         }
 127
 128         if (todo_start == todo_end && all_work_added) {
 129                 ret = NULL;
 130         } else {
 131                 ret = &todo[todo_start];
 132                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 133         }
 134         grep_unlock();
 135         return ret;
 136 }
 137
 138 static void grep_sha1_async(struct grep_opt *opt, char *name,
 139                             const unsigned char *sha1)
 140 {
 141         unsigned char *s;
 142         s = xmalloc(20);
 143         memcpy(s, sha1, 20);
 144         add_work(WORK_SHA1, name, s);
 145 }
 146
 147 static void grep_file_async(struct grep_opt *opt, char *name,
 148                             const char *filename)
 149 {
 150         add_work(WORK_FILE, name, xstrdup(filename));
 151 }
 152
 153 static void work_done(struct work_item *w)
 154 {
 155         int old_done;
 156
 157         grep_lock();
 158         w->done = 1;
 159         old_done = todo_done;
 160         for(; todo[todo_done].done && todo_done != todo_start;
 161             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 162                 w = &todo[todo_done];
 163                 if (w->out.len) {
 164                         if (print_hunk_marks_between_files && printed_something)
 165                                 write_or_die(1, "--\n", 3);
 166                         write_or_die(1, w->out.buf, w->out.len);
 167                         printed_something = 1;
 168                 }
 169                 free(w->name);
 170                 free(w->identifier);
 171         }
 172
 173         if (old_done != todo_done)
 174                 pthread_cond_signal(&cond_write);
 175
 176         if (all_work_added && todo_done == todo_end)
 177                 pthread_cond_signal(&cond_result);
 178
 179         grep_unlock();
 180 }
 181
 182 static int skip_binary(struct grep_opt *opt, const char *filename)
 183 {
 184         if ((opt->binary & GREP_BINARY_NOMATCH)) {
 185                 static struct git_attr *attr_text;
 186                 struct git_attr_check check;
 187
 188                 if (!attr_text)
 189                         attr_text = git_attr("text");
 190                 memset(&check, 0, sizeof(check));
 191                 check.attr = attr_text;
 192                 return !git_checkattr(filename, 1, &check) &&
 193                                 ATTR_FALSE(check.value);
 194         }
 195         return 0;
 196 }
 197
 198 static void *run(void *arg)
 199 {
 200         int hit = 0;
 201         struct grep_opt *opt = arg;
 202
 203         while (1) {
 204                 struct work_item *w = get_work();
 205                 if (!w)
 206                         break;
 207
 208                 if (skip_binary(opt, (const char *)w->identifier))
 209                         continue;
 210
 211                 opt->output_priv = w;
 212                 if (w->type == WORK_SHA1) {
 213                         unsigned long sz;
 214                         void* data = load_sha1(w->identifier, &sz, w->name);
 215
 216                         if (data) {
 217                                 hit |= grep_buffer(opt, w->name, data, sz);
 218                                 free(data);
 219                         }
 220                 } else if (w->type == WORK_FILE) {
 221                         size_t sz;
 222                         void* data = load_file(w->identifier, &sz);
 223                         if (data) {
 224                                 hit |= grep_buffer(opt, w->name, data, sz);
 225                                 free(data);
 226                         }
 227                 } else {
 228                         assert(0);
 229                 }
 230
 231                 work_done(w);
 232         }
 233         free_grep_patterns(arg);
 234         free(arg);
 235
 236         return (void*) (intptr_t) hit;
 237 }
 238
 239 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 240 {
 241         struct work_item *w = opt->output_priv;
 242         strbuf_add(&w->out, buf, size);
 243 }
 244
 245 static void start_threads(struct grep_opt *opt)
 246 {
 247         int i;
 248
 249         pthread_mutex_init(&grep_mutex, NULL);
 250         pthread_mutex_init(&read_sha1_mutex, NULL);
 251         pthread_cond_init(&cond_add, NULL);
 252         pthread_cond_init(&cond_write, NULL);
 253         pthread_cond_init(&cond_result, NULL);
 254
 255         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 256                 strbuf_init(&todo[i].out, 0);
 257         }
 258
 259         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 260                 int err;
 261                 struct grep_opt *o = grep_opt_dup(opt);
 262                 o->output = strbuf_out;
 263                 compile_grep_patterns(o);
 264                 err = pthread_create(&threads[i], NULL, run, o);
 265
 266                 if (err)
 267                         die(_("grep: failed to create thread: %s"),
 268                             strerror(err));
 269         }
 270 }
 271
 272 static int wait_all(void)
 273 {
 274         int hit = 0;
 275         int i;
 276
 277         grep_lock();
 278         all_work_added = 1;
 279
 280         /* Wait until all work is done. */
 281         while (todo_done != todo_end)
 282                 pthread_cond_wait(&cond_result, &grep_mutex);
 283
 284         /* Wake up all the consumer threads so they can see that there
 285          * is no more work to do.
 286          */
 287         pthread_cond_broadcast(&cond_add);
 288         grep_unlock();
 289
 290         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 291                 void *h;
 292                 pthread_join(threads[i], &h);
 293                 hit |= (int) (intptr_t) h;
 294         }
 295
 296         pthread_mutex_destroy(&grep_mutex);
 297         pthread_mutex_destroy(&read_sha1_mutex);
 298         pthread_cond_destroy(&cond_add);
 299         pthread_cond_destroy(&cond_write);
 300         pthread_cond_destroy(&cond_result);
 301
 302         return hit;
 303 }
 304 #else /* !NO_PTHREADS */
 305 #define read_sha1_lock()
 306 #define read_sha1_unlock()
 307
 308 static int wait_all(void)
 309 {
 310         return 0;
 311 }
 312 #endif
 313
 314 static int grep_config(const char *var, const char *value, void *cb)
 315 {
 316         struct grep_opt *opt = cb;
 317         char *color = NULL;
 318
 319         switch (userdiff_config(var, value)) {
 320         case 0: break;
 321         case -1: return -1;
 322         default: return 0;
 323         }
 324
 325         if (!strcmp(var, "grep.extendedregexp")) {
 326                 if (git_config_bool(var, value))
 327                         opt->regflags |= REG_EXTENDED;
 328                 else
 329                         opt->regflags &= ~REG_EXTENDED;
 330                 return 0;
 331         }
 332
 333         if (!strcmp(var, "grep.linenumber")) {
 334                 opt->linenum = git_config_bool(var, value);
 335                 return 0;
 336         }
 337
 338         if (!strcmp(var, "color.grep"))
 339                 opt->color = git_config_colorbool(var, value, -1);
 340         else if (!strcmp(var, "color.grep.context"))
 341                 color = opt->color_context;
 342         else if (!strcmp(var, "color.grep.filename"))
 343                 color = opt->color_filename;
 344         else if (!strcmp(var, "color.grep.function"))
 345                 color = opt->color_function;
 346         else if (!strcmp(var, "color.grep.linenumber"))
 347                 color = opt->color_lineno;
 348         else if (!strcmp(var, "color.grep.match"))
 349                 color = opt->color_match;
 350         else if (!strcmp(var, "color.grep.selected"))
 351                 color = opt->color_selected;
 352         else if (!strcmp(var, "color.grep.separator"))
 353                 color = opt->color_sep;
 354         else
 355                 return git_color_default_config(var, value, cb);
 356         if (color) {
 357                 if (!value)
 358                         return config_error_nonbool(var);
 359                 color_parse(value, var, color);
 360         }
 361         return 0;
 362 }
 363
 364 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 365 {
 366         void *data;
 367
 368         if (use_threads) {
 369                 read_sha1_lock();
 370                 data = read_sha1_file(sha1, type, size);
 371                 read_sha1_unlock();
 372         } else {
 373                 data = read_sha1_file(sha1, type, size);
 374         }
 375         return data;
 376 }
 377
 378 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 379                        const char *name)
 380 {
 381         enum object_type type;
 382         void *data = lock_and_read_sha1_file(sha1, &type, size);
 383
 384         if (!data)
 385                 error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
 386
 387         return data;
 388 }
 389
 390 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 391                      const char *filename, int tree_name_len)
 392 {
 393         struct strbuf pathbuf = STRBUF_INIT;
 394         char *name;
 395
 396         if (opt->relative && opt->prefix_length) {
 397                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 398                                     opt->prefix);
 399                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 400         } else {
 401                 strbuf_addstr(&pathbuf, filename);
 402         }
 403
 404         name = strbuf_detach(&pathbuf, NULL);
 405
 406 #ifndef NO_PTHREADS
 407         if (use_threads) {
 408                 grep_sha1_async(opt, name, sha1);
 409                 return 0;
 410         } else
 411 #endif
 412         {
 413                 int hit;
 414                 unsigned long sz;
 415                 void *data = load_sha1(sha1, &sz, name);
 416                 if (!data)
 417                         hit = 0;
 418                 else
 419                         hit = grep_buffer(opt, name, data, sz);
 420
 421                 free(data);
 422                 free(name);
 423                 return hit;
 424         }
 425 }
 426
 427 static void *load_file(const char *filename, size_t *sz)
 428 {
 429         struct stat st;
 430         char *data;
 431         int i;
 432
 433         if (lstat(filename, &st) < 0) {
 434         err_ret:
 435                 if (errno != ENOENT)
 436                         error(_("'%s': %s"), filename, strerror(errno));
 437                 return NULL;
 438         }
 439         if (!S_ISREG(st.st_mode))
 440                 return NULL;
 441         *sz = xsize_t(st.st_size);
 442         i = open(filename, O_RDONLY);
 443         if (i < 0)
 444                 goto err_ret;
 445         data = xmalloc(*sz + 1);
 446         if (st.st_size != read_in_full(i, data, *sz)) {
 447                 error(_("'%s': short read %s"), filename, strerror(errno));
 448                 close(i);
 449                 free(data);
 450                 return NULL;
 451         }
 452         close(i);
 453         data[*sz] = 0;
 454         return data;
 455 }
 456
 457 static int grep_file(struct grep_opt *opt, const char *filename)
 458 {
 459         struct strbuf buf = STRBUF_INIT;
 460         char *name;
 461
 462         if (opt->relative && opt->prefix_length)
 463                 quote_path_relative(filename, -1, &buf, opt->prefix);
 464         else
 465                 strbuf_addstr(&buf, filename);
 466         name = strbuf_detach(&buf, NULL);
 467
 468 #ifndef NO_PTHREADS
 469         if (use_threads) {
 470                 grep_file_async(opt, name, filename);
 471                 return 0;
 472         } else
 473 #endif
 474         {
 475                 int hit;
 476                 size_t sz;
 477                 void *data = load_file(filename, &sz);
 478                 if (!data)
 479                         hit = 0;
 480                 else
 481                         hit = grep_buffer(opt, name, data, sz);
 482
 483                 free(data);
 484                 free(name);
 485                 return hit;
 486         }
 487 }
 488
 489 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 490 {
 491         struct string_list *path_list = opt->output_priv;
 492
 493         if (len == 1 && *(const char *)data == '\0')
 494                 return;
 495         string_list_append(path_list, xstrndup(data, len));
 496 }
 497
 498 static void run_pager(struct grep_opt *opt, const char *prefix)
 499 {
 500         struct string_list *path_list = opt->output_priv;
 501         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 502         int i, status;
 503
 504         for (i = 0; i < path_list->nr; i++)
 505                 argv[i] = path_list->items[i].string;
 506         argv[path_list->nr] = NULL;
 507
 508         if (prefix && chdir(prefix))
 509                 die(_("Failed to chdir: %s"), prefix);
 510         status = run_command_v_opt(argv, RUN_USING_SHELL);
 511         if (status)
 512                 exit(status);
 513         free(argv);
 514 }
 515
 516 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 517 {
 518         int hit = 0;
 519         int nr;
 520         read_cache();
 521
 522         for (nr = 0; nr < active_nr; nr++) {
 523                 struct cache_entry *ce = active_cache[nr];
 524                 if (!S_ISREG(ce->ce_mode))
 525                         continue;
 526                 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 527                         continue;
 528                 if (skip_binary(opt, ce->name))
 529                         continue;
 530
 531                 /*
 532                  * If CE_VALID is on, we assume worktree file and its cache entry
 533                  * are identical, even if worktree file has been modified, so use
 534                  * cache version instead
 535                  */
 536                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 537                         if (ce_stage(ce))
 538                                 continue;
 539                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 540                 }
 541                 else
 542                         hit |= grep_file(opt, ce->name);
 543                 if (ce_stage(ce)) {
 544                         do {
 545                                 nr++;
 546                         } while (nr < active_nr &&
 547                                  !strcmp(ce->name, active_cache[nr]->name));
 548                         nr--; /* compensate for loop control */
 549                 }
 550                 if (hit && opt->status_only)
 551                         break;
 552         }
 553         return hit;
 554 }
 555
 556 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 557                      struct tree_desc *tree, struct strbuf *base, int tn_len)
 558 {
 559         int hit = 0, match = 0;
 560         struct name_entry entry;
 561         int old_baselen = base->len;
 562
 563         while (tree_entry(tree, &entry)) {
 564                 int te_len = tree_entry_len(entry.path, entry.sha1);
 565
 566                 if (match != 2) {
 567                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 568                         if (match < 0)
 569                                 break;
 570                         if (match == 0)
 571                                 continue;
 572                 }
 573
 574                 strbuf_add(base, entry.path, te_len);
 575
 576                 if (S_ISREG(entry.mode)) {
 577                         hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 578                 }
 579                 else if (S_ISDIR(entry.mode)) {
 580                         enum object_type type;
 581                         struct tree_desc sub;
 582                         void *data;
 583                         unsigned long size;
 584
 585                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 586                         if (!data)
 587                                 die(_("unable to read tree (%s)"),
 588                                     sha1_to_hex(entry.sha1));
 589
 590                         strbuf_addch(base, '/');
 591                         init_tree_desc(&sub, data, size);
 592                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 593                         free(data);
 594                 }
 595                 strbuf_setlen(base, old_baselen);
 596
 597                 if (hit && opt->status_only)
 598                         break;
 599         }
 600         return hit;
 601 }
 602
 603 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 604                        struct object *obj, const char *name)
 605 {
 606         if (obj->type == OBJ_BLOB)
 607                 return grep_sha1(opt, obj->sha1, name, 0);
 608         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 609                 struct tree_desc tree;
 610                 void *data;
 611                 unsigned long size;
 612                 struct strbuf base;
 613                 int hit, len;
 614
 615                 data = read_object_with_reference(obj->sha1, tree_type,
 616                                                   &size, NULL);
 617                 if (!data)
 618                         die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 619
 620                 len = name ? strlen(name) : 0;
 621                 strbuf_init(&base, PATH_MAX + len + 1);
 622                 if (len) {
 623                         strbuf_add(&base, name, len);
 624                         strbuf_addch(&base, ':');
 625                 }
 626                 init_tree_desc(&tree, data, size);
 627                 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 628                 strbuf_release(&base);
 629                 free(data);
 630                 return hit;
 631         }
 632         die(_("unable to grep from object of type %s"), typename(obj->type));
 633 }
 634
 635 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 636                         const struct object_array *list)
 637 {
 638         unsigned int i;
 639         int hit = 0;
 640         const unsigned int nr = list->nr;
 641
 642         for (i = 0; i < nr; i++) {
 643                 struct object *real_obj;
 644                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 645                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 646                         hit = 1;
 647                         if (opt->status_only)
 648                                 break;
 649                 }
 650         }
 651         return hit;
 652 }
 653
 654 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec)
 655 {
 656         struct dir_struct dir;
 657         int i, hit = 0;
 658
 659         memset(&dir, 0, sizeof(dir));
 660         setup_standard_excludes(&dir);
 661
 662         fill_directory(&dir, pathspec->raw);
 663         for (i = 0; i < dir.nr; i++) {
 664                 const char *name = dir.entries[i]->name;
 665                 int namelen = strlen(name);
 666                 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 667                         continue;
 668                 hit |= grep_file(opt, dir.entries[i]->name);
 669                 if (hit && opt->status_only)
 670                         break;
 671         }
 672         return hit;
 673 }
 674
 675 static int context_callback(const struct option *opt, const char *arg,
 676                             int unset)
 677 {
 678         struct grep_opt *grep_opt = opt->value;
 679         int value;
 680         const char *endp;
 681
 682         if (unset) {
 683                 grep_opt->pre_context = grep_opt->post_context = 0;
 684                 return 0;
 685         }
 686         value = strtol(arg, (char **)&endp, 10);
 687         if (*endp) {
 688                 return error(_("switch `%c' expects a numerical value"),
 689                              opt->short_name);
 690         }
 691         grep_opt->pre_context = grep_opt->post_context = value;
 692         return 0;
 693 }
 694
 695 static int file_callback(const struct option *opt, const char *arg, int unset)
 696 {
 697         struct grep_opt *grep_opt = opt->value;
 698         int from_stdin = !strcmp(arg, "-");
 699         FILE *patterns;
 700         int lno = 0;
 701         struct strbuf sb = STRBUF_INIT;
 702
 703         patterns = from_stdin ? stdin : fopen(arg, "r");
 704         if (!patterns)
 705                 die_errno(_("cannot open '%s'"), arg);
 706         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 707                 char *s;
 708                 size_t len;
 709
 710                 /* ignore empty line like grep does */
 711                 if (sb.len == 0)
 712                         continue;
 713
 714                 s = strbuf_detach(&sb, &len);
 715                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 716         }
 717         if (!from_stdin)
 718                 fclose(patterns);
 719         strbuf_release(&sb);
 720         return 0;
 721 }
 722
 723 static int not_callback(const struct option *opt, const char *arg, int unset)
 724 {
 725         struct grep_opt *grep_opt = opt->value;
 726         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 727         return 0;
 728 }
 729
 730 static int and_callback(const struct option *opt, const char *arg, int unset)
 731 {
 732         struct grep_opt *grep_opt = opt->value;
 733         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 734         return 0;
 735 }
 736
 737 static int open_callback(const struct option *opt, const char *arg, int unset)
 738 {
 739         struct grep_opt *grep_opt = opt->value;
 740         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 741         return 0;
 742 }
 743
 744 static int close_callback(const struct option *opt, const char *arg, int unset)
 745 {
 746         struct grep_opt *grep_opt = opt->value;
 747         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 748         return 0;
 749 }
 750
 751 static int pattern_callback(const struct option *opt, const char *arg,
 752                             int unset)
 753 {
 754         struct grep_opt *grep_opt = opt->value;
 755         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 756         return 0;
 757 }
 758
 759 static int help_callback(const struct option *opt, const char *arg, int unset)
 760 {
 761         return -1;
 762 }
 763
 764 int cmd_grep(int argc, const char **argv, const char *prefix)
 765 {
 766         int hit = 0;
 767         int cached = 0;
 768         int seen_dashdash = 0;
 769         int external_grep_allowed__ignored;
 770         const char *show_in_pager = NULL, *default_pager = "dummy";
 771         struct grep_opt opt;
 772         struct object_array list = OBJECT_ARRAY_INIT;
 773         const char **paths = NULL;
 774         struct pathspec pathspec;
 775         struct string_list path_list = STRING_LIST_INIT_NODUP;
 776         int i;
 777         int dummy;
 778         int use_index = 1;
 779         enum {
 780                 pattern_type_unspecified = 0,
 781                 pattern_type_bre,
 782                 pattern_type_ere,
 783                 pattern_type_fixed,
 784                 pattern_type_pcre,
 785         };
 786         int pattern_type = pattern_type_unspecified;
 787
 788         struct option options[] = {
 789                 OPT_BOOLEAN(0, "cached", &cached,
 790                         "search in index instead of in the work tree"),
 791                 OPT_BOOLEAN(0, "index", &use_index,
 792                         "--no-index finds in contents not managed by git"),
 793                 OPT_GROUP(""),
 794                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 795                         "show non-matching lines"),
 796                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 797                         "case insensitive matching"),
 798                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 799                         "match patterns only at word boundaries"),
 800                 OPT_SET_INT('a', "text", &opt.binary,
 801                         "process binary files as text", GREP_BINARY_TEXT),
 802                 OPT_SET_INT('I', NULL, &opt.binary,
 803                         "don't match patterns in binary files",
 804                         GREP_BINARY_NOMATCH),
 805                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 806                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 807                         NULL, 1 },
 808                 OPT_GROUP(""),
 809                 OPT_SET_INT('E', "extended-regexp", &pattern_type,
 810                             "use extended POSIX regular expressions",
 811                             pattern_type_ere),
 812                 OPT_SET_INT('G', "basic-regexp", &pattern_type,
 813                             "use basic POSIX regular expressions (default)",
 814                             pattern_type_bre),
 815                 OPT_SET_INT('F', "fixed-strings", &pattern_type,
 816                             "interpret patterns as fixed strings",
 817                             pattern_type_fixed),
 818                 OPT_SET_INT('P', "perl-regexp", &pattern_type,
 819                             "use Perl-compatible regular expressions",
 820                             pattern_type_pcre),
 821                 OPT_GROUP(""),
 822                 OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
 823                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 824                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 825                 OPT_NEGBIT(0, "full-name", &opt.relative,
 826                         "show filenames relative to top directory", 1),
 827                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 828                         "show only filenames instead of matching lines"),
 829                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 830                         "synonym for --files-with-matches"),
 831                 OPT_BOOLEAN('L', "files-without-match",
 832                         &opt.unmatch_name_only,
 833                         "show only the names of files without match"),
 834                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 835                         "print NUL after filenames"),
 836                 OPT_BOOLEAN('c', "count", &opt.count,
 837                         "show the number of matches instead of matching lines"),
 838                 OPT__COLOR(&opt.color, "highlight matches"),
 839                 OPT_GROUP(""),
 840                 OPT_CALLBACK('C', NULL, &opt, "n",
 841                         "show <n> context lines before and after matches",
 842                         context_callback),
 843                 OPT_INTEGER('B', NULL, &opt.pre_context,
 844                         "show <n> context lines before matches"),
 845                 OPT_INTEGER('A', NULL, &opt.post_context,
 846                         "show <n> context lines after matches"),
 847                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 848                         context_callback),
 849                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 850                         "show a line with the function name before matches"),
 851                 OPT_GROUP(""),
 852                 OPT_CALLBACK('f', NULL, &opt, "file",
 853                         "read patterns from file", file_callback),
 854                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 855                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 856                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 857                   "combine patterns specified with -e",
 858                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 859                 OPT_BOOLEAN(0, "or", &dummy, ""),
 860                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 861                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 862                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 863                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 864                   open_callback },
 865                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 866                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 867                   close_callback },
 868                 OPT__QUIET(&opt.status_only,
 869                            "indicate hit with exit status without output"),
 870                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 871                         "show only matches from files that match all patterns"),
 872                 OPT_GROUP(""),
 873                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 874                         "pager", "show matching files in the pager",
 875                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 876                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 877                             "allow calling of grep(1) (ignored by this build)"),
 878                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 879                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 880                 OPT_END()
 881         };
 882
 883         /*
 884          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 885          * to show usage information and exit.
 886          */
 887         if (argc == 2 && !strcmp(argv[1], "-h"))
 888                 usage_with_options(grep_usage, options);
 889
 890         memset(&opt, 0, sizeof(opt));
 891         opt.prefix = prefix;
 892         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 893         opt.relative = 1;
 894         opt.pathname = 1;
 895         opt.pattern_tail = &opt.pattern_list;
 896         opt.header_tail = &opt.header_list;
 897         opt.regflags = REG_NEWLINE;
 898         opt.max_depth = -1;
 899
 900         strcpy(opt.color_context, "");
 901         strcpy(opt.color_filename, "");
 902         strcpy(opt.color_function, "");
 903         strcpy(opt.color_lineno, "");
 904         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 905         strcpy(opt.color_selected, "");
 906         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 907         opt.color = -1;
 908         git_config(grep_config, &opt);
 909         if (opt.color == -1)
 910                 opt.color = git_use_color_default;
 911
 912         /*
 913          * If there is no -- then the paths must exist in the working
 914          * tree.  If there is no explicit pattern specified with -e or
 915          * -f, we take the first unrecognized non option to be the
 916          * pattern, but then what follows it must be zero or more
 917          * valid refs up to the -- (if exists), and then existing
 918          * paths.  If there is an explicit pattern, then the first
 919          * unrecognized non option is the beginning of the refs list
 920          * that continues up to the -- (if exists), and then paths.
 921          */
 922         argc = parse_options(argc, argv, prefix, options, grep_usage,
 923                              PARSE_OPT_KEEP_DASHDASH |
 924                              PARSE_OPT_STOP_AT_NON_OPTION |
 925                              PARSE_OPT_NO_INTERNAL_HELP);
 926         switch (pattern_type) {
 927         case pattern_type_fixed:
 928                 opt.fixed = 1;
 929                 opt.pcre = 0;
 930                 break;
 931         case pattern_type_bre:
 932                 opt.fixed = 0;
 933                 opt.pcre = 0;
 934                 opt.regflags &= ~REG_EXTENDED;
 935                 break;
 936         case pattern_type_ere:
 937                 opt.fixed = 0;
 938                 opt.pcre = 0;
 939                 opt.regflags |= REG_EXTENDED;
 940                 break;
 941         case pattern_type_pcre:
 942                 opt.fixed = 0;
 943                 opt.pcre = 1;
 944                 break;
 945         default:
 946                 break; /* nothing */
 947         }
 948
 949         if (use_index && !startup_info->have_repository)
 950                 /* die the same way as if we did it at the beginning */
 951                 setup_git_directory();
 952
 953         /*
 954          * skip a -- separator; we know it cannot be
 955          * separating revisions from pathnames if
 956          * we haven't even had any patterns yet
 957          */
 958         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 959                 argv++;
 960                 argc--;
 961         }
 962
 963         /* First unrecognized non-option token */
 964         if (argc > 0 && !opt.pattern_list) {
 965                 append_grep_pattern(&opt, argv[0], "command line", 0,
 966                                     GREP_PATTERN);
 967                 argv++;
 968                 argc--;
 969         }
 970
 971         if (show_in_pager == default_pager)
 972                 show_in_pager = git_pager(1);
 973         if (show_in_pager) {
 974                 opt.color = 0;
 975                 opt.name_only = 1;
 976                 opt.null_following_name = 1;
 977                 opt.output_priv = &path_list;
 978                 opt.output = append_path;
 979                 string_list_append(&path_list, show_in_pager);
 980                 use_threads = 0;
 981         }
 982         if ((opt.binary & GREP_BINARY_NOMATCH))
 983                 use_threads = 0;
 984
 985         if (!opt.pattern_list)
 986                 die(_("no pattern given."));
 987         if (!opt.fixed && opt.ignore_case)
 988                 opt.regflags |= REG_ICASE;
 989
 990 #ifndef NO_PTHREADS
 991         if (online_cpus() == 1 || !grep_threads_ok(&opt))
 992                 use_threads = 0;
 993
 994         if (use_threads) {
 995                 if (opt.pre_context || opt.post_context)
 996                         print_hunk_marks_between_files = 1;
 997                 start_threads(&opt);
 998         }
 999 #else
1000         use_threads = 0;
1001 #endif
1002
1003         compile_grep_patterns(&opt);
1004
1005         /* Check revs and then paths */
1006         for (i = 0; i < argc; i++) {
1007                 const char *arg = argv[i];
1008                 unsigned char sha1[20];
1009                 /* Is it a rev? */
1010                 if (!get_sha1(arg, sha1)) {
1011                         struct object *object = parse_object(sha1);
1012                         if (!object)
1013                                 die(_("bad object %s"), arg);
1014                         add_object_array(object, arg, &list);
1015                         continue;
1016                 }
1017                 if (!strcmp(arg, "--")) {
1018                         i++;
1019                         seen_dashdash = 1;
1020                 }
1021                 break;
1022         }
1023
1024         /* The rest are paths */
1025         if (!seen_dashdash) {
1026                 int j;
1027                 for (j = i; j < argc; j++)
1028                         verify_filename(prefix, argv[j]);
1029         }
1030
1031         paths = get_pathspec(prefix, argv + i);
1032         init_pathspec(&pathspec, paths);
1033         pathspec.max_depth = opt.max_depth;
1034         pathspec.recursive = 1;
1035
1036         if (show_in_pager && (cached || list.nr))
1037                 die(_("--open-files-in-pager only works on the worktree"));
1038
1039         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1040                 const char *pager = path_list.items[0].string;
1041                 int len = strlen(pager);
1042
1043                 if (len > 4 && is_dir_sep(pager[len - 5]))
1044                         pager += len - 4;
1045
1046                 if (opt.ignore_case && !strcmp("less", pager))
1047                         string_list_append(&path_list, "-i");
1048
1049                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1050                         struct strbuf buf = STRBUF_INIT;
1051                         strbuf_addf(&buf, "+/%s%s",
1052                                         strcmp("less", pager) ? "" : "*",
1053                                         opt.pattern_list->pattern);
1054                         string_list_append(&path_list, buf.buf);
1055                         strbuf_detach(&buf, NULL);
1056                 }
1057         }
1058
1059         if (!show_in_pager)
1060                 setup_pager();
1061
1062
1063         if (!use_index) {
1064                 if (cached)
1065                         die(_("--cached cannot be used with --no-index."));
1066                 if (list.nr)
1067                         die(_("--no-index cannot be used with revs."));
1068                 hit = grep_directory(&opt, &pathspec);
1069         } else if (!list.nr) {
1070                 if (!cached)
1071                         setup_work_tree();
1072
1073                 hit = grep_cache(&opt, &pathspec, cached);
1074         } else {
1075                 if (cached)
1076                         die(_("both --cached and trees are given."));
1077                 hit = grep_objects(&opt, &pathspec, &list);
1078         }
1079
1080         if (use_threads)
1081                 hit |= wait_all();
1082         if (hit && show_in_pager)
1083                 run_pager(&opt, prefix);
1084         free_grep_patterns(&opt);
1085         return !hit;
1086 }