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 "userdiff.h"
  15 #include "grep.h"
  16 #include "quote.h"
  17 #include "dir.h"
  18
  19 #ifndef NO_PTHREADS
  20 #include "thread-utils.h"
  21 #include <pthread.h>
  22 #endif
  23
  24 static char const * const grep_usage[] = {
  25         "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
  26         NULL
  27 };
  28
  29 static int use_threads = 1;
  30
  31 #ifndef NO_PTHREADS
  32 #define THREADS 8
  33 static pthread_t threads[THREADS];
  34
  35 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  36                        const char *name);
  37 static void *load_file(const char *filename, size_t *sz);
  38
  39 enum work_type {WORK_SHA1, WORK_FILE};
  40
  41 /* We use one producer thread and THREADS consumer
  42  * threads. The producer adds struct work_items to 'todo' and the
  43  * consumers pick work items from the same array.
  44  */
  45 struct work_item
  46 {
  47         enum work_type type;
  48         char *name;
  49
  50         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  51          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  52          * terminated filename.
  53          */
  54         void *identifier;
  55         char done;
  56         struct strbuf out;
  57 };
  58
  59 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  60  * that have been or are processed by a consumer thread. We haven't
  61  * written the result for these to stdout yet.
  62  *
  63  * The work_items in [todo_start, todo_end) are waiting to be picked
  64  * up by a consumer thread.
  65  *
  66  * The ranges are modulo TODO_SIZE.
  67  */
  68 #define TODO_SIZE 128
  69 static struct work_item todo[TODO_SIZE];
  70 static int todo_start;
  71 static int todo_end;
  72 static int todo_done;
  73
  74 /* Has all work items been added? */
  75 static int all_work_added;
  76
  77 /* This lock protects all the variables above. */
  78 static pthread_mutex_t grep_mutex;
  79
  80 /* Used to serialize calls to read_sha1_file. */
  81 static pthread_mutex_t read_sha1_mutex;
  82
  83 #define grep_lock() pthread_mutex_lock(&grep_mutex)
  84 #define grep_unlock() pthread_mutex_unlock(&grep_mutex)
  85 #define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
  86 #define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
  87
  88 /* Signalled when a new work_item is added to todo. */
  89 static pthread_cond_t cond_add;
  90
  91 /* Signalled when the result from one work_item is written to
  92  * stdout.
  93  */
  94 static pthread_cond_t cond_write;
  95
  96 /* Signalled when we are finished with everything. */
  97 static pthread_cond_t cond_result;
  98
  99 static void add_work(enum work_type type, char *name, void *id)
 100 {
 101         grep_lock();
 102
 103         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 104                 pthread_cond_wait(&cond_write, &grep_mutex);
 105         }
 106
 107         todo[todo_end].type = type;
 108         todo[todo_end].name = name;
 109         todo[todo_end].identifier = id;
 110         todo[todo_end].done = 0;
 111         strbuf_reset(&todo[todo_end].out);
 112         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 113
 114         pthread_cond_signal(&cond_add);
 115         grep_unlock();
 116 }
 117
 118 static struct work_item *get_work(void)
 119 {
 120         struct work_item *ret;
 121
 122         grep_lock();
 123         while (todo_start == todo_end && !all_work_added) {
 124                 pthread_cond_wait(&cond_add, &grep_mutex);
 125         }
 126
 127         if (todo_start == todo_end && all_work_added) {
 128                 ret = NULL;
 129         } else {
 130                 ret = &todo[todo_start];
 131                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 132         }
 133         grep_unlock();
 134         return ret;
 135 }
 136
 137 static void grep_sha1_async(struct grep_opt *opt, char *name,
 138                             const unsigned char *sha1)
 139 {
 140         unsigned char *s;
 141         s = xmalloc(20);
 142         memcpy(s, sha1, 20);
 143         add_work(WORK_SHA1, name, s);
 144 }
 145
 146 static void grep_file_async(struct grep_opt *opt, char *name,
 147                             const char *filename)
 148 {
 149         add_work(WORK_FILE, name, xstrdup(filename));
 150 }
 151
 152 static void work_done(struct work_item *w)
 153 {
 154         int old_done;
 155
 156         grep_lock();
 157         w->done = 1;
 158         old_done = todo_done;
 159         for(; todo[todo_done].done && todo_done != todo_start;
 160             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 161                 w = &todo[todo_done];
 162                 write_or_die(1, w->out.buf, w->out.len);
 163                 free(w->name);
 164                 free(w->identifier);
 165         }
 166
 167         if (old_done != todo_done)
 168                 pthread_cond_signal(&cond_write);
 169
 170         if (all_work_added && todo_done == todo_end)
 171                 pthread_cond_signal(&cond_result);
 172
 173         grep_unlock();
 174 }
 175
 176 static void *run(void *arg)
 177 {
 178         int hit = 0;
 179         struct grep_opt *opt = arg;
 180
 181         while (1) {
 182                 struct work_item *w = get_work();
 183                 if (!w)
 184                         break;
 185
 186                 opt->output_priv = w;
 187                 if (w->type == WORK_SHA1) {
 188                         unsigned long sz;
 189                         void* data = load_sha1(w->identifier, &sz, w->name);
 190
 191                         if (data) {
 192                                 hit |= grep_buffer(opt, w->name, data, sz);
 193                                 free(data);
 194                         }
 195                 } else if (w->type == WORK_FILE) {
 196                         size_t sz;
 197                         void* data = load_file(w->identifier, &sz);
 198                         if (data) {
 199                                 hit |= grep_buffer(opt, w->name, data, sz);
 200                                 free(data);
 201                         }
 202                 } else {
 203                         assert(0);
 204                 }
 205
 206                 work_done(w);
 207         }
 208         free_grep_patterns(arg);
 209         free(arg);
 210
 211         return (void*) (intptr_t) hit;
 212 }
 213
 214 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 215 {
 216         struct work_item *w = opt->output_priv;
 217         strbuf_add(&w->out, buf, size);
 218 }
 219
 220 static void start_threads(struct grep_opt *opt)
 221 {
 222         int i;
 223
 224         pthread_mutex_init(&grep_mutex, NULL);
 225         pthread_mutex_init(&read_sha1_mutex, NULL);
 226         pthread_cond_init(&cond_add, NULL);
 227         pthread_cond_init(&cond_write, NULL);
 228         pthread_cond_init(&cond_result, NULL);
 229
 230         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 231                 strbuf_init(&todo[i].out, 0);
 232         }
 233
 234         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 235                 int err;
 236                 struct grep_opt *o = grep_opt_dup(opt);
 237                 o->output = strbuf_out;
 238                 compile_grep_patterns(o);
 239                 err = pthread_create(&threads[i], NULL, run, o);
 240
 241                 if (err)
 242                         die("grep: failed to create thread: %s",
 243                             strerror(err));
 244         }
 245 }
 246
 247 static int wait_all(void)
 248 {
 249         int hit = 0;
 250         int i;
 251
 252         grep_lock();
 253         all_work_added = 1;
 254
 255         /* Wait until all work is done. */
 256         while (todo_done != todo_end)
 257                 pthread_cond_wait(&cond_result, &grep_mutex);
 258
 259         /* Wake up all the consumer threads so they can see that there
 260          * is no more work to do.
 261          */
 262         pthread_cond_broadcast(&cond_add);
 263         grep_unlock();
 264
 265         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 266                 void *h;
 267                 pthread_join(threads[i], &h);
 268                 hit |= (int) (intptr_t) h;
 269         }
 270
 271         pthread_mutex_destroy(&grep_mutex);
 272         pthread_mutex_destroy(&read_sha1_mutex);
 273         pthread_cond_destroy(&cond_add);
 274         pthread_cond_destroy(&cond_write);
 275         pthread_cond_destroy(&cond_result);
 276
 277         return hit;
 278 }
 279 #else /* !NO_PTHREADS */
 280 #define read_sha1_lock()
 281 #define read_sha1_unlock()
 282
 283 static int wait_all(void)
 284 {
 285         return 0;
 286 }
 287 #endif
 288
 289 static int grep_config(const char *var, const char *value, void *cb)
 290 {
 291         struct grep_opt *opt = cb;
 292         char *color = NULL;
 293
 294         switch (userdiff_config(var, value)) {
 295         case 0: break;
 296         case -1: return -1;
 297         default: return 0;
 298         }
 299
 300         if (!strcmp(var, "color.grep"))
 301                 opt->color = git_config_colorbool(var, value, -1);
 302         else if (!strcmp(var, "color.grep.context"))
 303                 color = opt->color_context;
 304         else if (!strcmp(var, "color.grep.filename"))
 305                 color = opt->color_filename;
 306         else if (!strcmp(var, "color.grep.function"))
 307                 color = opt->color_function;
 308         else if (!strcmp(var, "color.grep.linenumber"))
 309                 color = opt->color_lineno;
 310         else if (!strcmp(var, "color.grep.match"))
 311                 color = opt->color_match;
 312         else if (!strcmp(var, "color.grep.selected"))
 313                 color = opt->color_selected;
 314         else if (!strcmp(var, "color.grep.separator"))
 315                 color = opt->color_sep;
 316         else
 317                 return git_color_default_config(var, value, cb);
 318         if (color) {
 319                 if (!value)
 320                         return config_error_nonbool(var);
 321                 color_parse(value, var, color);
 322         }
 323         return 0;
 324 }
 325
 326 /*
 327  * Return non-zero if max_depth is negative or path has no more then max_depth
 328  * slashes.
 329  */
 330 static int accept_subdir(const char *path, int max_depth)
 331 {
 332         if (max_depth < 0)
 333                 return 1;
 334
 335         while ((path = strchr(path, '/')) != NULL) {
 336                 max_depth--;
 337                 if (max_depth < 0)
 338                         return 0;
 339                 path++;
 340         }
 341         return 1;
 342 }
 343
 344 /*
 345  * Return non-zero if name is a subdirectory of match and is not too deep.
 346  */
 347 static int is_subdir(const char *name, int namelen,
 348                 const char *match, int matchlen, int max_depth)
 349 {
 350         if (matchlen > namelen || strncmp(name, match, matchlen))
 351                 return 0;
 352
 353         if (name[matchlen] == '\0') /* exact match */
 354                 return 1;
 355
 356         if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
 357                 return accept_subdir(name + matchlen + 1, max_depth);
 358
 359         return 0;
 360 }
 361
 362 /*
 363  * git grep pathspecs are somewhat different from diff-tree pathspecs;
 364  * pathname wildcards are allowed.
 365  */
 366 static int pathspec_matches(const char **paths, const char *name, int max_depth)
 367 {
 368         int namelen, i;
 369         if (!paths || !*paths)
 370                 return accept_subdir(name, max_depth);
 371         namelen = strlen(name);
 372         for (i = 0; paths[i]; i++) {
 373                 const char *match = paths[i];
 374                 int matchlen = strlen(match);
 375                 const char *cp, *meta;
 376
 377                 if (is_subdir(name, namelen, match, matchlen, max_depth))
 378                         return 1;
 379                 if (!fnmatch(match, name, 0))
 380                         return 1;
 381                 if (name[namelen-1] != '/')
 382                         continue;
 383
 384                 /* We are being asked if the directory ("name") is worth
 385                  * descending into.
 386                  *
 387                  * Find the longest leading directory name that does
 388                  * not have metacharacter in the pathspec; the name
 389                  * we are looking at must overlap with that directory.
 390                  */
 391                 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
 392                         char ch = *cp;
 393                         if (ch == '*' || ch == '[' || ch == '?') {
 394                                 meta = cp;
 395                                 break;
 396                         }
 397                 }
 398                 if (!meta)
 399                         meta = cp; /* fully literal */
 400
 401                 if (namelen <= meta - match) {
 402                         /* Looking at "Documentation/" and
 403                          * the pattern says "Documentation/howto/", or
 404                          * "Documentation/diff*.txt".  The name we
 405                          * have should match prefix.
 406                          */
 407                         if (!memcmp(match, name, namelen))
 408                                 return 1;
 409                         continue;
 410                 }
 411
 412                 if (meta - match < namelen) {
 413                         /* Looking at "Documentation/howto/" and
 414                          * the pattern says "Documentation/h*";
 415                          * match up to "Do.../h"; this avoids descending
 416                          * into "Documentation/technical/".
 417                          */
 418                         if (!memcmp(match, name, meta - match))
 419                                 return 1;
 420                         continue;
 421                 }
 422         }
 423         return 0;
 424 }
 425
 426 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 427 {
 428         void *data;
 429
 430         if (use_threads) {
 431                 read_sha1_lock();
 432                 data = read_sha1_file(sha1, type, size);
 433                 read_sha1_unlock();
 434         } else {
 435                 data = read_sha1_file(sha1, type, size);
 436         }
 437         return data;
 438 }
 439
 440 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 441                        const char *name)
 442 {
 443         enum object_type type;
 444         void *data = lock_and_read_sha1_file(sha1, &type, size);
 445
 446         if (!data)
 447                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 448
 449         return data;
 450 }
 451
 452 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 453                      const char *filename, int tree_name_len)
 454 {
 455         struct strbuf pathbuf = STRBUF_INIT;
 456         char *name;
 457
 458         if (opt->relative && opt->prefix_length) {
 459                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 460                                     opt->prefix);
 461                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 462         } else {
 463                 strbuf_addstr(&pathbuf, filename);
 464         }
 465
 466         name = strbuf_detach(&pathbuf, NULL);
 467
 468 #ifndef NO_PTHREADS
 469         if (use_threads) {
 470                 grep_sha1_async(opt, name, sha1);
 471                 return 0;
 472         } else
 473 #endif
 474         {
 475                 int hit;
 476                 unsigned long sz;
 477                 void *data = load_sha1(sha1, &sz, name);
 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 *load_file(const char *filename, size_t *sz)
 490 {
 491         struct stat st;
 492         char *data;
 493         int i;
 494
 495         if (lstat(filename, &st) < 0) {
 496         err_ret:
 497                 if (errno != ENOENT)
 498                         error("'%s': %s", filename, strerror(errno));
 499                 return 0;
 500         }
 501         if (!S_ISREG(st.st_mode))
 502                 return 0;
 503         *sz = xsize_t(st.st_size);
 504         i = open(filename, O_RDONLY);
 505         if (i < 0)
 506                 goto err_ret;
 507         data = xmalloc(*sz + 1);
 508         if (st.st_size != read_in_full(i, data, *sz)) {
 509                 error("'%s': short read %s", filename, strerror(errno));
 510                 close(i);
 511                 free(data);
 512                 return 0;
 513         }
 514         close(i);
 515         data[*sz] = 0;
 516         return data;
 517 }
 518
 519 static int grep_file(struct grep_opt *opt, const char *filename)
 520 {
 521         struct strbuf buf = STRBUF_INIT;
 522         char *name;
 523
 524         if (opt->relative && opt->prefix_length)
 525                 quote_path_relative(filename, -1, &buf, opt->prefix);
 526         else
 527                 strbuf_addstr(&buf, filename);
 528         name = strbuf_detach(&buf, NULL);
 529
 530 #ifndef NO_PTHREADS
 531         if (use_threads) {
 532                 grep_file_async(opt, name, filename);
 533                 return 0;
 534         } else
 535 #endif
 536         {
 537                 int hit;
 538                 size_t sz;
 539                 void *data = load_file(filename, &sz);
 540                 if (!data)
 541                         hit = 0;
 542                 else
 543                         hit = grep_buffer(opt, name, data, sz);
 544
 545                 free(data);
 546                 free(name);
 547                 return hit;
 548         }
 549 }
 550
 551 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 552 {
 553         int hit = 0;
 554         int nr;
 555         read_cache();
 556
 557         for (nr = 0; nr < active_nr; nr++) {
 558                 struct cache_entry *ce = active_cache[nr];
 559                 if (!S_ISREG(ce->ce_mode))
 560                         continue;
 561                 if (!pathspec_matches(paths, ce->name, opt->max_depth))
 562                         continue;
 563                 /*
 564                  * If CE_VALID is on, we assume worktree file and its cache entry
 565                  * are identical, even if worktree file has been modified, so use
 566                  * cache version instead
 567                  */
 568                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 569                         if (ce_stage(ce))
 570                                 continue;
 571                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 572                 }
 573                 else
 574                         hit |= grep_file(opt, ce->name);
 575                 if (ce_stage(ce)) {
 576                         do {
 577                                 nr++;
 578                         } while (nr < active_nr &&
 579                                  !strcmp(ce->name, active_cache[nr]->name));
 580                         nr--; /* compensate for loop control */
 581                 }
 582                 if (hit && opt->status_only)
 583                         break;
 584         }
 585         free_grep_patterns(opt);
 586         return hit;
 587 }
 588
 589 static int grep_tree(struct grep_opt *opt, const char **paths,
 590                      struct tree_desc *tree,
 591                      const char *tree_name, const char *base)
 592 {
 593         int len;
 594         int hit = 0;
 595         struct name_entry entry;
 596         char *down;
 597         int tn_len = strlen(tree_name);
 598         struct strbuf pathbuf;
 599
 600         strbuf_init(&pathbuf, PATH_MAX + tn_len);
 601
 602         if (tn_len) {
 603                 strbuf_add(&pathbuf, tree_name, tn_len);
 604                 strbuf_addch(&pathbuf, ':');
 605                 tn_len = pathbuf.len;
 606         }
 607         strbuf_addstr(&pathbuf, base);
 608         len = pathbuf.len;
 609
 610         while (tree_entry(tree, &entry)) {
 611                 int te_len = tree_entry_len(entry.path, entry.sha1);
 612                 pathbuf.len = len;
 613                 strbuf_add(&pathbuf, entry.path, te_len);
 614
 615                 if (S_ISDIR(entry.mode))
 616                         /* Match "abc/" against pathspec to
 617                          * decide if we want to descend into "abc"
 618                          * directory.
 619                          */
 620                         strbuf_addch(&pathbuf, '/');
 621
 622                 down = pathbuf.buf + tn_len;
 623                 if (!pathspec_matches(paths, down, opt->max_depth))
 624                         ;
 625                 else if (S_ISREG(entry.mode))
 626                         hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
 627                 else if (S_ISDIR(entry.mode)) {
 628                         enum object_type type;
 629                         struct tree_desc sub;
 630                         void *data;
 631                         unsigned long size;
 632
 633                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 634                         if (!data)
 635                                 die("unable to read tree (%s)",
 636                                     sha1_to_hex(entry.sha1));
 637                         init_tree_desc(&sub, data, size);
 638                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 639                         free(data);
 640                 }
 641                 if (hit && opt->status_only)
 642                         break;
 643         }
 644         strbuf_release(&pathbuf);
 645         return hit;
 646 }
 647
 648 static int grep_object(struct grep_opt *opt, const char **paths,
 649                        struct object *obj, const char *name)
 650 {
 651         if (obj->type == OBJ_BLOB)
 652                 return grep_sha1(opt, obj->sha1, name, 0);
 653         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 654                 struct tree_desc tree;
 655                 void *data;
 656                 unsigned long size;
 657                 int hit;
 658                 data = read_object_with_reference(obj->sha1, tree_type,
 659                                                   &size, NULL);
 660                 if (!data)
 661                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 662                 init_tree_desc(&tree, data, size);
 663                 hit = grep_tree(opt, paths, &tree, name, "");
 664                 free(data);
 665                 return hit;
 666         }
 667         die("unable to grep from object of type %s", typename(obj->type));
 668 }
 669
 670 static int grep_directory(struct grep_opt *opt, const char **paths)
 671 {
 672         struct dir_struct dir;
 673         int i, hit = 0;
 674
 675         memset(&dir, 0, sizeof(dir));
 676         setup_standard_excludes(&dir);
 677
 678         fill_directory(&dir, paths);
 679         for (i = 0; i < dir.nr; i++) {
 680                 hit |= grep_file(opt, dir.entries[i]->name);
 681                 if (hit && opt->status_only)
 682                         break;
 683         }
 684         free_grep_patterns(opt);
 685         return hit;
 686 }
 687
 688 static int context_callback(const struct option *opt, const char *arg,
 689                             int unset)
 690 {
 691         struct grep_opt *grep_opt = opt->value;
 692         int value;
 693         const char *endp;
 694
 695         if (unset) {
 696                 grep_opt->pre_context = grep_opt->post_context = 0;
 697                 return 0;
 698         }
 699         value = strtol(arg, (char **)&endp, 10);
 700         if (*endp) {
 701                 return error("switch `%c' expects a numerical value",
 702                              opt->short_name);
 703         }
 704         grep_opt->pre_context = grep_opt->post_context = value;
 705         return 0;
 706 }
 707
 708 static int file_callback(const struct option *opt, const char *arg, int unset)
 709 {
 710         struct grep_opt *grep_opt = opt->value;
 711         FILE *patterns;
 712         int lno = 0;
 713         struct strbuf sb = STRBUF_INIT;
 714
 715         patterns = fopen(arg, "r");
 716         if (!patterns)
 717                 die_errno("cannot open '%s'", arg);
 718         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 719                 /* ignore empty line like grep does */
 720                 if (sb.len == 0)
 721                         continue;
 722                 append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg,
 723                                     ++lno, GREP_PATTERN);
 724         }
 725         fclose(patterns);
 726         strbuf_release(&sb);
 727         return 0;
 728 }
 729
 730 static int not_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, "--not", "command line", 0, GREP_NOT);
 734         return 0;
 735 }
 736
 737 static int and_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, "--and", "command line", 0, GREP_AND);
 741         return 0;
 742 }
 743
 744 static int open_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_OPEN_PAREN);
 748         return 0;
 749 }
 750
 751 static int close_callback(const struct option *opt, const char *arg, int unset)
 752 {
 753         struct grep_opt *grep_opt = opt->value;
 754         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 755         return 0;
 756 }
 757
 758 static int pattern_callback(const struct option *opt, const char *arg,
 759                             int unset)
 760 {
 761         struct grep_opt *grep_opt = opt->value;
 762         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 763         return 0;
 764 }
 765
 766 static int help_callback(const struct option *opt, const char *arg, int unset)
 767 {
 768         return -1;
 769 }
 770
 771 int cmd_grep(int argc, const char **argv, const char *prefix)
 772 {
 773         int hit = 0;
 774         int cached = 0;
 775         int seen_dashdash = 0;
 776         int external_grep_allowed__ignored;
 777         struct grep_opt opt;
 778         struct object_array list = { 0, 0, NULL };
 779         const char **paths = NULL;
 780         int i;
 781         int dummy;
 782         int nongit = 0, use_index = 1;
 783         struct option options[] = {
 784                 OPT_BOOLEAN(0, "cached", &cached,
 785                         "search in index instead of in the work tree"),
 786                 OPT_BOOLEAN(0, "index", &use_index,
 787                         "--no-index finds in contents not managed by git"),
 788                 OPT_GROUP(""),
 789                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 790                         "show non-matching lines"),
 791                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 792                         "case insensitive matching"),
 793                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 794                         "match patterns only at word boundaries"),
 795                 OPT_SET_INT('a', "text", &opt.binary,
 796                         "process binary files as text", GREP_BINARY_TEXT),
 797                 OPT_SET_INT('I', NULL, &opt.binary,
 798                         "don't match patterns in binary files",
 799                         GREP_BINARY_NOMATCH),
 800                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 801                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 802                         NULL, 1 },
 803                 OPT_GROUP(""),
 804                 OPT_BIT('E', "extended-regexp", &opt.regflags,
 805                         "use extended POSIX regular expressions", REG_EXTENDED),
 806                 OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
 807                         "use basic POSIX regular expressions (default)",
 808                         REG_EXTENDED),
 809                 OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
 810                         "interpret patterns as fixed strings"),
 811                 OPT_GROUP(""),
 812                 OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
 813                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 814                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 815                 OPT_NEGBIT(0, "full-name", &opt.relative,
 816                         "show filenames relative to top directory", 1),
 817                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 818                         "show only filenames instead of matching lines"),
 819                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 820                         "synonym for --files-with-matches"),
 821                 OPT_BOOLEAN('L', "files-without-match",
 822                         &opt.unmatch_name_only,
 823                         "show only the names of files without match"),
 824                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 825                         "print NUL after filenames"),
 826                 OPT_BOOLEAN('c', "count", &opt.count,
 827                         "show the number of matches instead of matching lines"),
 828                 OPT__COLOR(&opt.color, "highlight matches"),
 829                 OPT_GROUP(""),
 830                 OPT_CALLBACK('C', NULL, &opt, "n",
 831                         "show <n> context lines before and after matches",
 832                         context_callback),
 833                 OPT_INTEGER('B', NULL, &opt.pre_context,
 834                         "show <n> context lines before matches"),
 835                 OPT_INTEGER('A', NULL, &opt.post_context,
 836                         "show <n> context lines after matches"),
 837                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 838                         context_callback),
 839                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 840                         "show a line with the function name before matches"),
 841                 OPT_GROUP(""),
 842                 OPT_CALLBACK('f', NULL, &opt, "file",
 843                         "read patterns from file", file_callback),
 844                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 845                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 846                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 847                   "combine patterns specified with -e",
 848                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 849                 OPT_BOOLEAN(0, "or", &dummy, ""),
 850                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 851                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 852                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 853                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 854                   open_callback },
 855                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 856                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 857                   close_callback },
 858                 OPT_BOOLEAN('q', "quiet", &opt.status_only,
 859                             "indicate hit with exit status without output"),
 860                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 861                         "show only matches from files that match all patterns"),
 862                 OPT_GROUP(""),
 863                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 864                             "allow calling of grep(1) (ignored by this build)"),
 865                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 866                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 867                 OPT_END()
 868         };
 869
 870         prefix = setup_git_directory_gently(&nongit);
 871
 872         /*
 873          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 874          * to show usage information and exit.
 875          */
 876         if (argc == 2 && !strcmp(argv[1], "-h"))
 877                 usage_with_options(grep_usage, options);
 878
 879         memset(&opt, 0, sizeof(opt));
 880         opt.prefix = prefix;
 881         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 882         opt.relative = 1;
 883         opt.pathname = 1;
 884         opt.pattern_tail = &opt.pattern_list;
 885         opt.header_tail = &opt.header_list;
 886         opt.regflags = REG_NEWLINE;
 887         opt.max_depth = -1;
 888
 889         strcpy(opt.color_context, "");
 890         strcpy(opt.color_filename, "");
 891         strcpy(opt.color_function, "");
 892         strcpy(opt.color_lineno, "");
 893         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 894         strcpy(opt.color_selected, "");
 895         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 896         opt.color = -1;
 897         git_config(grep_config, &opt);
 898         if (opt.color == -1)
 899                 opt.color = git_use_color_default;
 900
 901         /*
 902          * If there is no -- then the paths must exist in the working
 903          * tree.  If there is no explicit pattern specified with -e or
 904          * -f, we take the first unrecognized non option to be the
 905          * pattern, but then what follows it must be zero or more
 906          * valid refs up to the -- (if exists), and then existing
 907          * paths.  If there is an explicit pattern, then the first
 908          * unrecognized non option is the beginning of the refs list
 909          * that continues up to the -- (if exists), and then paths.
 910          */
 911         argc = parse_options(argc, argv, prefix, options, grep_usage,
 912                              PARSE_OPT_KEEP_DASHDASH |
 913                              PARSE_OPT_STOP_AT_NON_OPTION |
 914                              PARSE_OPT_NO_INTERNAL_HELP);
 915
 916         if (use_index && nongit)
 917                 /* die the same way as if we did it at the beginning */
 918                 setup_git_directory();
 919
 920         /*
 921          * skip a -- separator; we know it cannot be
 922          * separating revisions from pathnames if
 923          * we haven't even had any patterns yet
 924          */
 925         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 926                 argv++;
 927                 argc--;
 928         }
 929
 930         /* First unrecognized non-option token */
 931         if (argc > 0 && !opt.pattern_list) {
 932                 append_grep_pattern(&opt, argv[0], "command line", 0,
 933                                     GREP_PATTERN);
 934                 argv++;
 935                 argc--;
 936         }
 937
 938         if (!opt.pattern_list)
 939                 die("no pattern given.");
 940         if (!opt.fixed && opt.ignore_case)
 941                 opt.regflags |= REG_ICASE;
 942         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
 943                 die("cannot mix --fixed-strings and regexp");
 944
 945 #ifndef NO_PTHREADS
 946         if (online_cpus() == 1 || !grep_threads_ok(&opt))
 947                 use_threads = 0;
 948
 949         if (use_threads)
 950                 start_threads(&opt);
 951 #else
 952         use_threads = 0;
 953 #endif
 954
 955         compile_grep_patterns(&opt);
 956
 957         /* Check revs and then paths */
 958         for (i = 0; i < argc; i++) {
 959                 const char *arg = argv[i];
 960                 unsigned char sha1[20];
 961                 /* Is it a rev? */
 962                 if (!get_sha1(arg, sha1)) {
 963                         struct object *object = parse_object(sha1);
 964                         if (!object)
 965                                 die("bad object %s", arg);
 966                         add_object_array(object, arg, &list);
 967                         continue;
 968                 }
 969                 if (!strcmp(arg, "--")) {
 970                         i++;
 971                         seen_dashdash = 1;
 972                 }
 973                 break;
 974         }
 975
 976         /* The rest are paths */
 977         if (!seen_dashdash) {
 978                 int j;
 979                 for (j = i; j < argc; j++)
 980                         verify_filename(prefix, argv[j]);
 981         }
 982
 983         if (i < argc)
 984                 paths = get_pathspec(prefix, argv + i);
 985         else if (prefix) {
 986                 paths = xcalloc(2, sizeof(const char *));
 987                 paths[0] = prefix;
 988                 paths[1] = NULL;
 989         }
 990
 991         if (!use_index) {
 992                 int hit;
 993                 if (cached)
 994                         die("--cached cannot be used with --no-index.");
 995                 if (list.nr)
 996                         die("--no-index cannot be used with revs.");
 997                 hit = grep_directory(&opt, paths);
 998                 if (use_threads)
 999                         hit |= wait_all();
1000                 return !hit;
1001         }
1002
1003         if (!list.nr) {
1004                 int hit;
1005                 if (!cached)
1006                         setup_work_tree();
1007
1008                 hit = grep_cache(&opt, paths, cached);
1009                 if (use_threads)
1010                         hit |= wait_all();
1011                 return !hit;
1012         }
1013
1014         if (cached)
1015                 die("both --cached and trees are given.");
1016
1017         for (i = 0; i < list.nr; i++) {
1018                 struct object *real_obj;
1019                 real_obj = deref_tag(list.objects[i].item, NULL, 0);
1020                 if (grep_object(&opt, paths, real_obj, list.objects[i].name)) {
1021                         hit = 1;
1022                         if (opt.status_only)
1023                                 break;
1024                 }
1025         }
1026
1027         if (use_threads)
1028                 hit |= wait_all();
1029         free_grep_patterns(&opt);
1030         return !hit;
1031 }