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 #ifndef NO_PTHREADS
  22 #include <pthread.h>
  23 #include "thread-utils.h"
  24 #endif
  25
  26 static char const * const grep_usage[] = {
  27         "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
  28         NULL
  29 };
  30
  31 static int use_threads = 1;
  32
  33 #ifndef NO_PTHREADS
  34 #define THREADS 8
  35 static pthread_t threads[THREADS];
  36
  37 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  38                        const char *name);
  39 static void *load_file(const char *filename, size_t *sz);
  40
  41 enum work_type {WORK_SHA1, WORK_FILE};
  42
  43 /* We use one producer thread and THREADS consumer
  44  * threads. The producer adds struct work_items to 'todo' and the
  45  * consumers pick work items from the same array.
  46  */
  47 struct work_item
  48 {
  49         enum work_type type;
  50         char *name;
  51
  52         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  53          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  54          * terminated filename.
  55          */
  56         void *identifier;
  57         char done;
  58         struct strbuf out;
  59 };
  60
  61 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  62  * that have been or are processed by a consumer thread. We haven't
  63  * written the result for these to stdout yet.
  64  *
  65  * The work_items in [todo_start, todo_end) are waiting to be picked
  66  * up by a consumer thread.
  67  *
  68  * The ranges are modulo TODO_SIZE.
  69  */
  70 #define TODO_SIZE 128
  71 static struct work_item todo[TODO_SIZE];
  72 static int todo_start;
  73 static int todo_end;
  74 static int todo_done;
  75
  76 /* Has all work items been added? */
  77 static int all_work_added;
  78
  79 /* This lock protects all the variables above. */
  80 static pthread_mutex_t grep_mutex;
  81
  82 /* Used to serialize calls to read_sha1_file. */
  83 static pthread_mutex_t read_sha1_mutex;
  84
  85 #define grep_lock() pthread_mutex_lock(&grep_mutex)
  86 #define grep_unlock() pthread_mutex_unlock(&grep_mutex)
  87 #define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
  88 #define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
  89
  90 /* Signalled when a new work_item is added to todo. */
  91 static pthread_cond_t cond_add;
  92
  93 /* Signalled when the result from one work_item is written to
  94  * stdout.
  95  */
  96 static pthread_cond_t cond_write;
  97
  98 /* Signalled when we are finished with everything. */
  99 static pthread_cond_t cond_result;
 100
 101 static int print_hunk_marks_between_files;
 102 static int printed_something;
 103
 104 static void add_work(enum work_type type, char *name, void *id)
 105 {
 106         grep_lock();
 107
 108         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 109                 pthread_cond_wait(&cond_write, &grep_mutex);
 110         }
 111
 112         todo[todo_end].type = type;
 113         todo[todo_end].name = name;
 114         todo[todo_end].identifier = id;
 115         todo[todo_end].done = 0;
 116         strbuf_reset(&todo[todo_end].out);
 117         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 118
 119         pthread_cond_signal(&cond_add);
 120         grep_unlock();
 121 }
 122
 123 static struct work_item *get_work(void)
 124 {
 125         struct work_item *ret;
 126
 127         grep_lock();
 128         while (todo_start == todo_end && !all_work_added) {
 129                 pthread_cond_wait(&cond_add, &grep_mutex);
 130         }
 131
 132         if (todo_start == todo_end && all_work_added) {
 133                 ret = NULL;
 134         } else {
 135                 ret = &todo[todo_start];
 136                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 137         }
 138         grep_unlock();
 139         return ret;
 140 }
 141
 142 static void grep_sha1_async(struct grep_opt *opt, char *name,
 143                             const unsigned char *sha1)
 144 {
 145         unsigned char *s;
 146         s = xmalloc(20);
 147         memcpy(s, sha1, 20);
 148         add_work(WORK_SHA1, name, s);
 149 }
 150
 151 static void grep_file_async(struct grep_opt *opt, char *name,
 152                             const char *filename)
 153 {
 154         add_work(WORK_FILE, name, xstrdup(filename));
 155 }
 156
 157 static void work_done(struct work_item *w)
 158 {
 159         int old_done;
 160
 161         grep_lock();
 162         w->done = 1;
 163         old_done = todo_done;
 164         for(; todo[todo_done].done && todo_done != todo_start;
 165             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 166                 w = &todo[todo_done];
 167                 if (w->out.len) {
 168                         if (print_hunk_marks_between_files && printed_something)
 169                                 write_or_die(1, "--\n", 3);
 170                         write_or_die(1, w->out.buf, w->out.len);
 171                         printed_something = 1;
 172                 }
 173                 free(w->name);
 174                 free(w->identifier);
 175         }
 176
 177         if (old_done != todo_done)
 178                 pthread_cond_signal(&cond_write);
 179
 180         if (all_work_added && todo_done == todo_end)
 181                 pthread_cond_signal(&cond_result);
 182
 183         grep_unlock();
 184 }
 185
 186 static void *run(void *arg)
 187 {
 188         int hit = 0;
 189         struct grep_opt *opt = arg;
 190
 191         while (1) {
 192                 struct work_item *w = get_work();
 193                 if (!w)
 194                         break;
 195
 196                 opt->output_priv = w;
 197                 if (w->type == WORK_SHA1) {
 198                         unsigned long sz;
 199                         void* data = load_sha1(w->identifier, &sz, w->name);
 200
 201                         if (data) {
 202                                 hit |= grep_buffer(opt, w->name, data, sz);
 203                                 free(data);
 204                         }
 205                 } else if (w->type == WORK_FILE) {
 206                         size_t sz;
 207                         void* data = load_file(w->identifier, &sz);
 208                         if (data) {
 209                                 hit |= grep_buffer(opt, w->name, data, sz);
 210                                 free(data);
 211                         }
 212                 } else {
 213                         assert(0);
 214                 }
 215
 216                 work_done(w);
 217         }
 218         free_grep_patterns(arg);
 219         free(arg);
 220
 221         return (void*) (intptr_t) hit;
 222 }
 223
 224 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 225 {
 226         struct work_item *w = opt->output_priv;
 227         strbuf_add(&w->out, buf, size);
 228 }
 229
 230 static void start_threads(struct grep_opt *opt)
 231 {
 232         int i;
 233
 234         pthread_mutex_init(&grep_mutex, NULL);
 235         pthread_mutex_init(&read_sha1_mutex, NULL);
 236         pthread_cond_init(&cond_add, NULL);
 237         pthread_cond_init(&cond_write, NULL);
 238         pthread_cond_init(&cond_result, NULL);
 239
 240         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 241                 strbuf_init(&todo[i].out, 0);
 242         }
 243
 244         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 245                 int err;
 246                 struct grep_opt *o = grep_opt_dup(opt);
 247                 o->output = strbuf_out;
 248                 compile_grep_patterns(o);
 249                 err = pthread_create(&threads[i], NULL, run, o);
 250
 251                 if (err)
 252                         die("grep: failed to create thread: %s",
 253                             strerror(err));
 254         }
 255 }
 256
 257 static int wait_all(void)
 258 {
 259         int hit = 0;
 260         int i;
 261
 262         grep_lock();
 263         all_work_added = 1;
 264
 265         /* Wait until all work is done. */
 266         while (todo_done != todo_end)
 267                 pthread_cond_wait(&cond_result, &grep_mutex);
 268
 269         /* Wake up all the consumer threads so they can see that there
 270          * is no more work to do.
 271          */
 272         pthread_cond_broadcast(&cond_add);
 273         grep_unlock();
 274
 275         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 276                 void *h;
 277                 pthread_join(threads[i], &h);
 278                 hit |= (int) (intptr_t) h;
 279         }
 280
 281         pthread_mutex_destroy(&grep_mutex);
 282         pthread_mutex_destroy(&read_sha1_mutex);
 283         pthread_cond_destroy(&cond_add);
 284         pthread_cond_destroy(&cond_write);
 285         pthread_cond_destroy(&cond_result);
 286
 287         return hit;
 288 }
 289 #else /* !NO_PTHREADS */
 290 #define read_sha1_lock()
 291 #define read_sha1_unlock()
 292
 293 static int wait_all(void)
 294 {
 295         return 0;
 296 }
 297 #endif
 298
 299 static int grep_config(const char *var, const char *value, void *cb)
 300 {
 301         struct grep_opt *opt = cb;
 302         char *color = NULL;
 303
 304         switch (userdiff_config(var, value)) {
 305         case 0: break;
 306         case -1: return -1;
 307         default: return 0;
 308         }
 309
 310         if (!strcmp(var, "color.grep"))
 311                 opt->color = git_config_colorbool(var, value, -1);
 312         else if (!strcmp(var, "color.grep.context"))
 313                 color = opt->color_context;
 314         else if (!strcmp(var, "color.grep.filename"))
 315                 color = opt->color_filename;
 316         else if (!strcmp(var, "color.grep.function"))
 317                 color = opt->color_function;
 318         else if (!strcmp(var, "color.grep.linenumber"))
 319                 color = opt->color_lineno;
 320         else if (!strcmp(var, "color.grep.match"))
 321                 color = opt->color_match;
 322         else if (!strcmp(var, "color.grep.selected"))
 323                 color = opt->color_selected;
 324         else if (!strcmp(var, "color.grep.separator"))
 325                 color = opt->color_sep;
 326         else
 327                 return git_color_default_config(var, value, cb);
 328         if (color) {
 329                 if (!value)
 330                         return config_error_nonbool(var);
 331                 color_parse(value, var, color);
 332         }
 333         return 0;
 334 }
 335
 336 /*
 337  * Return non-zero if max_depth is negative or path has no more then max_depth
 338  * slashes.
 339  */
 340 static int accept_subdir(const char *path, int max_depth)
 341 {
 342         if (max_depth < 0)
 343                 return 1;
 344
 345         while ((path = strchr(path, '/')) != NULL) {
 346                 max_depth--;
 347                 if (max_depth < 0)
 348                         return 0;
 349                 path++;
 350         }
 351         return 1;
 352 }
 353
 354 /*
 355  * Return non-zero if name is a subdirectory of match and is not too deep.
 356  */
 357 static int is_subdir(const char *name, int namelen,
 358                 const char *match, int matchlen, int max_depth)
 359 {
 360         if (matchlen > namelen || strncmp(name, match, matchlen))
 361                 return 0;
 362
 363         if (name[matchlen] == '\0') /* exact match */
 364                 return 1;
 365
 366         if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
 367                 return accept_subdir(name + matchlen + 1, max_depth);
 368
 369         return 0;
 370 }
 371
 372 /*
 373  * git grep pathspecs are somewhat different from diff-tree pathspecs;
 374  * pathname wildcards are allowed.
 375  */
 376 static int pathspec_matches(const char **paths, const char *name, int max_depth)
 377 {
 378         int namelen, i;
 379         if (!paths || !*paths)
 380                 return accept_subdir(name, max_depth);
 381         namelen = strlen(name);
 382         for (i = 0; paths[i]; i++) {
 383                 const char *match = paths[i];
 384                 int matchlen = strlen(match);
 385                 const char *cp, *meta;
 386
 387                 if (is_subdir(name, namelen, match, matchlen, max_depth))
 388                         return 1;
 389                 if (!fnmatch(match, name, 0))
 390                         return 1;
 391                 if (name[namelen-1] != '/')
 392                         continue;
 393
 394                 /* We are being asked if the directory ("name") is worth
 395                  * descending into.
 396                  *
 397                  * Find the longest leading directory name that does
 398                  * not have metacharacter in the pathspec; the name
 399                  * we are looking at must overlap with that directory.
 400                  */
 401                 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
 402                         char ch = *cp;
 403                         if (ch == '*' || ch == '[' || ch == '?') {
 404                                 meta = cp;
 405                                 break;
 406                         }
 407                 }
 408                 if (!meta)
 409                         meta = cp; /* fully literal */
 410
 411                 if (namelen <= meta - match) {
 412                         /* Looking at "Documentation/" and
 413                          * the pattern says "Documentation/howto/", or
 414                          * "Documentation/diff*.txt".  The name we
 415                          * have should match prefix.
 416                          */
 417                         if (!memcmp(match, name, namelen))
 418                                 return 1;
 419                         continue;
 420                 }
 421
 422                 if (meta - match < namelen) {
 423                         /* Looking at "Documentation/howto/" and
 424                          * the pattern says "Documentation/h*";
 425                          * match up to "Do.../h"; this avoids descending
 426                          * into "Documentation/technical/".
 427                          */
 428                         if (!memcmp(match, name, meta - match))
 429                                 return 1;
 430                         continue;
 431                 }
 432         }
 433         return 0;
 434 }
 435
 436 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 437 {
 438         void *data;
 439
 440         if (use_threads) {
 441                 read_sha1_lock();
 442                 data = read_sha1_file(sha1, type, size);
 443                 read_sha1_unlock();
 444         } else {
 445                 data = read_sha1_file(sha1, type, size);
 446         }
 447         return data;
 448 }
 449
 450 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 451                        const char *name)
 452 {
 453         enum object_type type;
 454         void *data = lock_and_read_sha1_file(sha1, &type, size);
 455
 456         if (!data)
 457                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 458
 459         return data;
 460 }
 461
 462 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 463                      const char *filename, int tree_name_len)
 464 {
 465         struct strbuf pathbuf = STRBUF_INIT;
 466         char *name;
 467
 468         if (opt->relative && opt->prefix_length) {
 469                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 470                                     opt->prefix);
 471                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 472         } else {
 473                 strbuf_addstr(&pathbuf, filename);
 474         }
 475
 476         name = strbuf_detach(&pathbuf, NULL);
 477
 478 #ifndef NO_PTHREADS
 479         if (use_threads) {
 480                 grep_sha1_async(opt, name, sha1);
 481                 return 0;
 482         } else
 483 #endif
 484         {
 485                 int hit;
 486                 unsigned long sz;
 487                 void *data = load_sha1(sha1, &sz, name);
 488                 if (!data)
 489                         hit = 0;
 490                 else
 491                         hit = grep_buffer(opt, name, data, sz);
 492
 493                 free(data);
 494                 free(name);
 495                 return hit;
 496         }
 497 }
 498
 499 static void *load_file(const char *filename, size_t *sz)
 500 {
 501         struct stat st;
 502         char *data;
 503         int i;
 504
 505         if (lstat(filename, &st) < 0) {
 506         err_ret:
 507                 if (errno != ENOENT)
 508                         error("'%s': %s", filename, strerror(errno));
 509                 return 0;
 510         }
 511         if (!S_ISREG(st.st_mode))
 512                 return 0;
 513         *sz = xsize_t(st.st_size);
 514         i = open(filename, O_RDONLY);
 515         if (i < 0)
 516                 goto err_ret;
 517         data = xmalloc(*sz + 1);
 518         if (st.st_size != read_in_full(i, data, *sz)) {
 519                 error("'%s': short read %s", filename, strerror(errno));
 520                 close(i);
 521                 free(data);
 522                 return 0;
 523         }
 524         close(i);
 525         data[*sz] = 0;
 526         return data;
 527 }
 528
 529 static int grep_file(struct grep_opt *opt, const char *filename)
 530 {
 531         struct strbuf buf = STRBUF_INIT;
 532         char *name;
 533
 534         if (opt->relative && opt->prefix_length)
 535                 quote_path_relative(filename, -1, &buf, opt->prefix);
 536         else
 537                 strbuf_addstr(&buf, filename);
 538         name = strbuf_detach(&buf, NULL);
 539
 540 #ifndef NO_PTHREADS
 541         if (use_threads) {
 542                 grep_file_async(opt, name, filename);
 543                 return 0;
 544         } else
 545 #endif
 546         {
 547                 int hit;
 548                 size_t sz;
 549                 void *data = load_file(filename, &sz);
 550                 if (!data)
 551                         hit = 0;
 552                 else
 553                         hit = grep_buffer(opt, name, data, sz);
 554
 555                 free(data);
 556                 free(name);
 557                 return hit;
 558         }
 559 }
 560
 561 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 562 {
 563         struct string_list *path_list = opt->output_priv;
 564
 565         if (len == 1 && *(const char *)data == '\0')
 566                 return;
 567         string_list_append(path_list, xstrndup(data, len));
 568 }
 569
 570 static void run_pager(struct grep_opt *opt, const char *prefix)
 571 {
 572         struct string_list *path_list = opt->output_priv;
 573         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 574         int i, status;
 575
 576         for (i = 0; i < path_list->nr; i++)
 577                 argv[i] = path_list->items[i].string;
 578         argv[path_list->nr] = NULL;
 579
 580         if (prefix && chdir(prefix))
 581                 die("Failed to chdir: %s", prefix);
 582         status = run_command_v_opt(argv, RUN_USING_SHELL);
 583         if (status)
 584                 exit(status);
 585         free(argv);
 586 }
 587
 588 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 589 {
 590         int hit = 0;
 591         int nr;
 592         read_cache();
 593
 594         for (nr = 0; nr < active_nr; nr++) {
 595                 struct cache_entry *ce = active_cache[nr];
 596                 if (!S_ISREG(ce->ce_mode))
 597                         continue;
 598                 if (!pathspec_matches(paths, ce->name, opt->max_depth))
 599                         continue;
 600                 /*
 601                  * If CE_VALID is on, we assume worktree file and its cache entry
 602                  * are identical, even if worktree file has been modified, so use
 603                  * cache version instead
 604                  */
 605                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 606                         if (ce_stage(ce))
 607                                 continue;
 608                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 609                 }
 610                 else
 611                         hit |= grep_file(opt, ce->name);
 612                 if (ce_stage(ce)) {
 613                         do {
 614                                 nr++;
 615                         } while (nr < active_nr &&
 616                                  !strcmp(ce->name, active_cache[nr]->name));
 617                         nr--; /* compensate for loop control */
 618                 }
 619                 if (hit && opt->status_only)
 620                         break;
 621         }
 622         return hit;
 623 }
 624
 625 static int grep_tree(struct grep_opt *opt, const char **paths,
 626                      struct tree_desc *tree,
 627                      const char *tree_name, const char *base)
 628 {
 629         int len;
 630         int hit = 0;
 631         struct name_entry entry;
 632         char *down;
 633         int tn_len = strlen(tree_name);
 634         struct strbuf pathbuf;
 635
 636         strbuf_init(&pathbuf, PATH_MAX + tn_len);
 637
 638         if (tn_len) {
 639                 strbuf_add(&pathbuf, tree_name, tn_len);
 640                 strbuf_addch(&pathbuf, ':');
 641                 tn_len = pathbuf.len;
 642         }
 643         strbuf_addstr(&pathbuf, base);
 644         len = pathbuf.len;
 645
 646         while (tree_entry(tree, &entry)) {
 647                 int te_len = tree_entry_len(entry.path, entry.sha1);
 648                 pathbuf.len = len;
 649                 strbuf_add(&pathbuf, entry.path, te_len);
 650
 651                 if (S_ISDIR(entry.mode))
 652                         /* Match "abc/" against pathspec to
 653                          * decide if we want to descend into "abc"
 654                          * directory.
 655                          */
 656                         strbuf_addch(&pathbuf, '/');
 657
 658                 down = pathbuf.buf + tn_len;
 659                 if (!pathspec_matches(paths, down, opt->max_depth))
 660                         ;
 661                 else if (S_ISREG(entry.mode))
 662                         hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
 663                 else if (S_ISDIR(entry.mode)) {
 664                         enum object_type type;
 665                         struct tree_desc sub;
 666                         void *data;
 667                         unsigned long size;
 668
 669                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 670                         if (!data)
 671                                 die("unable to read tree (%s)",
 672                                     sha1_to_hex(entry.sha1));
 673                         init_tree_desc(&sub, data, size);
 674                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 675                         free(data);
 676                 }
 677                 if (hit && opt->status_only)
 678                         break;
 679         }
 680         strbuf_release(&pathbuf);
 681         return hit;
 682 }
 683
 684 static int grep_object(struct grep_opt *opt, const char **paths,
 685                        struct object *obj, const char *name)
 686 {
 687         if (obj->type == OBJ_BLOB)
 688                 return grep_sha1(opt, obj->sha1, name, 0);
 689         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 690                 struct tree_desc tree;
 691                 void *data;
 692                 unsigned long size;
 693                 int hit;
 694                 data = read_object_with_reference(obj->sha1, tree_type,
 695                                                   &size, NULL);
 696                 if (!data)
 697                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 698                 init_tree_desc(&tree, data, size);
 699                 hit = grep_tree(opt, paths, &tree, name, "");
 700                 free(data);
 701                 return hit;
 702         }
 703         die("unable to grep from object of type %s", typename(obj->type));
 704 }
 705
 706 static int grep_objects(struct grep_opt *opt, const char **paths,
 707                         const struct object_array *list)
 708 {
 709         unsigned int i;
 710         int hit = 0;
 711         const unsigned int nr = list->nr;
 712
 713         for (i = 0; i < nr; i++) {
 714                 struct object *real_obj;
 715                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 716                 if (grep_object(opt, paths, real_obj, list->objects[i].name)) {
 717                         hit = 1;
 718                         if (opt->status_only)
 719                                 break;
 720                 }
 721         }
 722         return hit;
 723 }
 724
 725 static int grep_directory(struct grep_opt *opt, const char **paths)
 726 {
 727         struct dir_struct dir;
 728         int i, hit = 0;
 729
 730         memset(&dir, 0, sizeof(dir));
 731         setup_standard_excludes(&dir);
 732
 733         fill_directory(&dir, paths);
 734         for (i = 0; i < dir.nr; i++) {
 735                 hit |= grep_file(opt, dir.entries[i]->name);
 736                 if (hit && opt->status_only)
 737                         break;
 738         }
 739         return hit;
 740 }
 741
 742 static int context_callback(const struct option *opt, const char *arg,
 743                             int unset)
 744 {
 745         struct grep_opt *grep_opt = opt->value;
 746         int value;
 747         const char *endp;
 748
 749         if (unset) {
 750                 grep_opt->pre_context = grep_opt->post_context = 0;
 751                 return 0;
 752         }
 753         value = strtol(arg, (char **)&endp, 10);
 754         if (*endp) {
 755                 return error("switch `%c' expects a numerical value",
 756                              opt->short_name);
 757         }
 758         grep_opt->pre_context = grep_opt->post_context = value;
 759         return 0;
 760 }
 761
 762 static int file_callback(const struct option *opt, const char *arg, int unset)
 763 {
 764         struct grep_opt *grep_opt = opt->value;
 765         FILE *patterns;
 766         int lno = 0;
 767         struct strbuf sb = STRBUF_INIT;
 768
 769         patterns = fopen(arg, "r");
 770         if (!patterns)
 771                 die_errno("cannot open '%s'", arg);
 772         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 773                 char *s;
 774                 size_t len;
 775
 776                 /* ignore empty line like grep does */
 777                 if (sb.len == 0)
 778                         continue;
 779
 780                 s = strbuf_detach(&sb, &len);
 781                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 782         }
 783         fclose(patterns);
 784         strbuf_release(&sb);
 785         return 0;
 786 }
 787
 788 static int not_callback(const struct option *opt, const char *arg, int unset)
 789 {
 790         struct grep_opt *grep_opt = opt->value;
 791         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 792         return 0;
 793 }
 794
 795 static int and_callback(const struct option *opt, const char *arg, int unset)
 796 {
 797         struct grep_opt *grep_opt = opt->value;
 798         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 799         return 0;
 800 }
 801
 802 static int open_callback(const struct option *opt, const char *arg, int unset)
 803 {
 804         struct grep_opt *grep_opt = opt->value;
 805         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 806         return 0;
 807 }
 808
 809 static int close_callback(const struct option *opt, const char *arg, int unset)
 810 {
 811         struct grep_opt *grep_opt = opt->value;
 812         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 813         return 0;
 814 }
 815
 816 static int pattern_callback(const struct option *opt, const char *arg,
 817                             int unset)
 818 {
 819         struct grep_opt *grep_opt = opt->value;
 820         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 821         return 0;
 822 }
 823
 824 static int help_callback(const struct option *opt, const char *arg, int unset)
 825 {
 826         return -1;
 827 }
 828
 829 int cmd_grep(int argc, const char **argv, const char *prefix)
 830 {
 831         int hit = 0;
 832         int cached = 0;
 833         int seen_dashdash = 0;
 834         int external_grep_allowed__ignored;
 835         const char *show_in_pager = NULL, *default_pager = "dummy";
 836         struct grep_opt opt;
 837         struct object_array list = OBJECT_ARRAY_INIT;
 838         const char **paths = NULL;
 839         struct string_list path_list = STRING_LIST_INIT_NODUP;
 840         int i;
 841         int dummy;
 842         int use_index = 1;
 843         struct option options[] = {
 844                 OPT_BOOLEAN(0, "cached", &cached,
 845                         "search in index instead of in the work tree"),
 846                 OPT_BOOLEAN(0, "index", &use_index,
 847                         "--no-index finds in contents not managed by git"),
 848                 OPT_GROUP(""),
 849                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 850                         "show non-matching lines"),
 851                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 852                         "case insensitive matching"),
 853                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 854                         "match patterns only at word boundaries"),
 855                 OPT_SET_INT('a', "text", &opt.binary,
 856                         "process binary files as text", GREP_BINARY_TEXT),
 857                 OPT_SET_INT('I', NULL, &opt.binary,
 858                         "don't match patterns in binary files",
 859                         GREP_BINARY_NOMATCH),
 860                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 861                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 862                         NULL, 1 },
 863                 OPT_GROUP(""),
 864                 OPT_BIT('E', "extended-regexp", &opt.regflags,
 865                         "use extended POSIX regular expressions", REG_EXTENDED),
 866                 OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
 867                         "use basic POSIX regular expressions (default)",
 868                         REG_EXTENDED),
 869                 OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
 870                         "interpret patterns as fixed strings"),
 871                 OPT_GROUP(""),
 872                 OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
 873                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 874                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 875                 OPT_NEGBIT(0, "full-name", &opt.relative,
 876                         "show filenames relative to top directory", 1),
 877                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 878                         "show only filenames instead of matching lines"),
 879                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 880                         "synonym for --files-with-matches"),
 881                 OPT_BOOLEAN('L', "files-without-match",
 882                         &opt.unmatch_name_only,
 883                         "show only the names of files without match"),
 884                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 885                         "print NUL after filenames"),
 886                 OPT_BOOLEAN('c', "count", &opt.count,
 887                         "show the number of matches instead of matching lines"),
 888                 OPT__COLOR(&opt.color, "highlight matches"),
 889                 OPT_GROUP(""),
 890                 OPT_CALLBACK('C', NULL, &opt, "n",
 891                         "show <n> context lines before and after matches",
 892                         context_callback),
 893                 OPT_INTEGER('B', NULL, &opt.pre_context,
 894                         "show <n> context lines before matches"),
 895                 OPT_INTEGER('A', NULL, &opt.post_context,
 896                         "show <n> context lines after matches"),
 897                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 898                         context_callback),
 899                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 900                         "show a line with the function name before matches"),
 901                 OPT_GROUP(""),
 902                 OPT_CALLBACK('f', NULL, &opt, "file",
 903                         "read patterns from file", file_callback),
 904                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 905                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 906                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 907                   "combine patterns specified with -e",
 908                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 909                 OPT_BOOLEAN(0, "or", &dummy, ""),
 910                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 911                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 912                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 913                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 914                   open_callback },
 915                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 916                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 917                   close_callback },
 918                 OPT_BOOLEAN('q', "quiet", &opt.status_only,
 919                             "indicate hit with exit status without output"),
 920                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 921                         "show only matches from files that match all patterns"),
 922                 OPT_GROUP(""),
 923                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 924                         "pager", "show matching files in the pager",
 925                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 926                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 927                             "allow calling of grep(1) (ignored by this build)"),
 928                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 929                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 930                 OPT_END()
 931         };
 932
 933         /*
 934          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 935          * to show usage information and exit.
 936          */
 937         if (argc == 2 && !strcmp(argv[1], "-h"))
 938                 usage_with_options(grep_usage, options);
 939
 940         memset(&opt, 0, sizeof(opt));
 941         opt.prefix = prefix;
 942         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 943         opt.relative = 1;
 944         opt.pathname = 1;
 945         opt.pattern_tail = &opt.pattern_list;
 946         opt.header_tail = &opt.header_list;
 947         opt.regflags = REG_NEWLINE;
 948         opt.max_depth = -1;
 949
 950         strcpy(opt.color_context, "");
 951         strcpy(opt.color_filename, "");
 952         strcpy(opt.color_function, "");
 953         strcpy(opt.color_lineno, "");
 954         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 955         strcpy(opt.color_selected, "");
 956         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 957         opt.color = -1;
 958         git_config(grep_config, &opt);
 959         if (opt.color == -1)
 960                 opt.color = git_use_color_default;
 961
 962         /*
 963          * If there is no -- then the paths must exist in the working
 964          * tree.  If there is no explicit pattern specified with -e or
 965          * -f, we take the first unrecognized non option to be the
 966          * pattern, but then what follows it must be zero or more
 967          * valid refs up to the -- (if exists), and then existing
 968          * paths.  If there is an explicit pattern, then the first
 969          * unrecognized non option is the beginning of the refs list
 970          * that continues up to the -- (if exists), and then paths.
 971          */
 972         argc = parse_options(argc, argv, prefix, options, grep_usage,
 973                              PARSE_OPT_KEEP_DASHDASH |
 974                              PARSE_OPT_STOP_AT_NON_OPTION |
 975                              PARSE_OPT_NO_INTERNAL_HELP);
 976
 977         if (use_index && !startup_info->have_repository)
 978                 /* die the same way as if we did it at the beginning */
 979                 setup_git_directory();
 980
 981         /*
 982          * skip a -- separator; we know it cannot be
 983          * separating revisions from pathnames if
 984          * we haven't even had any patterns yet
 985          */
 986         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 987                 argv++;
 988                 argc--;
 989         }
 990
 991         /* First unrecognized non-option token */
 992         if (argc > 0 && !opt.pattern_list) {
 993                 append_grep_pattern(&opt, argv[0], "command line", 0,
 994                                     GREP_PATTERN);
 995                 argv++;
 996                 argc--;
 997         }
 998
 999         if (show_in_pager == default_pager)
1000                 show_in_pager = git_pager(1);
1001         if (show_in_pager) {
1002                 opt.color = 0;
1003                 opt.name_only = 1;
1004                 opt.null_following_name = 1;
1005                 opt.output_priv = &path_list;
1006                 opt.output = append_path;
1007                 string_list_append(&path_list, show_in_pager);
1008                 use_threads = 0;
1009         }
1010
1011         if (!opt.pattern_list)
1012                 die("no pattern given.");
1013         if (!opt.fixed && opt.ignore_case)
1014                 opt.regflags |= REG_ICASE;
1015         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
1016                 die("cannot mix --fixed-strings and regexp");
1017
1018 #ifndef NO_PTHREADS
1019         if (online_cpus() == 1 || !grep_threads_ok(&opt))
1020                 use_threads = 0;
1021
1022         if (use_threads) {
1023                 if (opt.pre_context || opt.post_context)
1024                         print_hunk_marks_between_files = 1;
1025                 start_threads(&opt);
1026         }
1027 #else
1028         use_threads = 0;
1029 #endif
1030
1031         compile_grep_patterns(&opt);
1032
1033         /* Check revs and then paths */
1034         for (i = 0; i < argc; i++) {
1035                 const char *arg = argv[i];
1036                 unsigned char sha1[20];
1037                 /* Is it a rev? */
1038                 if (!get_sha1(arg, sha1)) {
1039                         struct object *object = parse_object(sha1);
1040                         if (!object)
1041                                 die("bad object %s", arg);
1042                         add_object_array(object, arg, &list);
1043                         continue;
1044                 }
1045                 if (!strcmp(arg, "--")) {
1046                         i++;
1047                         seen_dashdash = 1;
1048                 }
1049                 break;
1050         }
1051
1052         /* The rest are paths */
1053         if (!seen_dashdash) {
1054                 int j;
1055                 for (j = i; j < argc; j++)
1056                         verify_filename(prefix, argv[j]);
1057         }
1058
1059         if (i < argc)
1060                 paths = get_pathspec(prefix, argv + i);
1061         else if (prefix) {
1062                 paths = xcalloc(2, sizeof(const char *));
1063                 paths[0] = prefix;
1064                 paths[1] = NULL;
1065         }
1066
1067         if (show_in_pager && (cached || list.nr))
1068                 die("--open-files-in-pager only works on the worktree");
1069
1070         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1071                 const char *pager = path_list.items[0].string;
1072                 int len = strlen(pager);
1073
1074                 if (len > 4 && is_dir_sep(pager[len - 5]))
1075                         pager += len - 4;
1076
1077                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1078                         struct strbuf buf = STRBUF_INIT;
1079                         strbuf_addf(&buf, "+/%s%s",
1080                                         strcmp("less", pager) ? "" : "*",
1081                                         opt.pattern_list->pattern);
1082                         string_list_append(&path_list, buf.buf);
1083                         strbuf_detach(&buf, NULL);
1084                 }
1085         }
1086
1087         if (!show_in_pager)
1088                 setup_pager();
1089
1090
1091         if (!use_index) {
1092                 if (cached)
1093                         die("--cached cannot be used with --no-index.");
1094                 if (list.nr)
1095                         die("--no-index cannot be used with revs.");
1096                 hit = grep_directory(&opt, paths);
1097         } else if (!list.nr) {
1098                 if (!cached)
1099                         setup_work_tree();
1100
1101                 hit = grep_cache(&opt, paths, cached);
1102         } else {
1103                 if (cached)
1104                         die("both --cached and trees are given.");
1105                 hit = grep_objects(&opt, paths, &list);
1106         }
1107
1108         if (use_threads)
1109                 hit |= wait_all();
1110         if (hit && show_in_pager)
1111                 run_pager(&opt, prefix);
1112         free_grep_patterns(&opt);
1113         return !hit;
1114 }