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 "attr.h"
  21
  22 static char const * const grep_usage[] = {
  23         "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
  24         NULL
  25 };
  26
  27 static int use_threads = 1;
  28
  29 #ifndef NO_PTHREADS
  30 #define THREADS 8
  31 static pthread_t threads[THREADS];
  32
  33 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  34                        const char *name);
  35 static void *load_file(const char *filename, size_t *sz);
  36
  37 enum work_type {WORK_SHA1, WORK_FILE};
  38
  39 /* We use one producer thread and THREADS consumer
  40  * threads. The producer adds struct work_items to 'todo' and the
  41  * consumers pick work items from the same array.
  42  */
  43 struct work_item {
  44         enum work_type type;
  45         char *name;
  46
  47         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  48          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  49          * terminated filename.
  50          */
  51         void *identifier;
  52         char done;
  53         struct strbuf out;
  54 };
  55
  56 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  57  * that have been or are processed by a consumer thread. We haven't
  58  * written the result for these to stdout yet.
  59  *
  60  * The work_items in [todo_start, todo_end) are waiting to be picked
  61  * up by a consumer thread.
  62  *
  63  * The ranges are modulo TODO_SIZE.
  64  */
  65 #define TODO_SIZE 128
  66 static struct work_item todo[TODO_SIZE];
  67 static int todo_start;
  68 static int todo_end;
  69 static int todo_done;
  70
  71 /* Has all work items been added? */
  72 static int all_work_added;
  73
  74 /* This lock protects all the variables above. */
  75 static pthread_mutex_t grep_mutex;
  76
  77 static inline void grep_lock(void)
  78 {
  79         if (use_threads)
  80                 pthread_mutex_lock(&grep_mutex);
  81 }
  82
  83 static inline void grep_unlock(void)
  84 {
  85         if (use_threads)
  86                 pthread_mutex_unlock(&grep_mutex);
  87 }
  88
  89 /* Used to serialize calls to read_sha1_file. */
  90 static pthread_mutex_t read_sha1_mutex;
  91
  92 static inline void read_sha1_lock(void)
  93 {
  94         if (use_threads)
  95                 pthread_mutex_lock(&read_sha1_mutex);
  96 }
  97
  98 static inline void read_sha1_unlock(void)
  99 {
 100         if (use_threads)
 101                 pthread_mutex_unlock(&read_sha1_mutex);
 102 }
 103
 104 /* Signalled when a new work_item is added to todo. */
 105 static pthread_cond_t cond_add;
 106
 107 /* Signalled when the result from one work_item is written to
 108  * stdout.
 109  */
 110 static pthread_cond_t cond_write;
 111
 112 /* Signalled when we are finished with everything. */
 113 static pthread_cond_t cond_result;
 114
 115 static int skip_first_line;
 116
 117 static void add_work(enum work_type type, char *name, void *id)
 118 {
 119         grep_lock();
 120
 121         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 122                 pthread_cond_wait(&cond_write, &grep_mutex);
 123         }
 124
 125         todo[todo_end].type = type;
 126         todo[todo_end].name = name;
 127         todo[todo_end].identifier = id;
 128         todo[todo_end].done = 0;
 129         strbuf_reset(&todo[todo_end].out);
 130         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 131
 132         pthread_cond_signal(&cond_add);
 133         grep_unlock();
 134 }
 135
 136 static struct work_item *get_work(void)
 137 {
 138         struct work_item *ret;
 139
 140         grep_lock();
 141         while (todo_start == todo_end && !all_work_added) {
 142                 pthread_cond_wait(&cond_add, &grep_mutex);
 143         }
 144
 145         if (todo_start == todo_end && all_work_added) {
 146                 ret = NULL;
 147         } else {
 148                 ret = &todo[todo_start];
 149                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 150         }
 151         grep_unlock();
 152         return ret;
 153 }
 154
 155 static void grep_sha1_async(struct grep_opt *opt, char *name,
 156                             const unsigned char *sha1)
 157 {
 158         unsigned char *s;
 159         s = xmalloc(20);
 160         memcpy(s, sha1, 20);
 161         add_work(WORK_SHA1, name, s);
 162 }
 163
 164 static void grep_file_async(struct grep_opt *opt, char *name,
 165                             const char *filename)
 166 {
 167         add_work(WORK_FILE, name, xstrdup(filename));
 168 }
 169
 170 static void work_done(struct work_item *w)
 171 {
 172         int old_done;
 173
 174         grep_lock();
 175         w->done = 1;
 176         old_done = todo_done;
 177         for(; todo[todo_done].done && todo_done != todo_start;
 178             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 179                 w = &todo[todo_done];
 180                 if (w->out.len) {
 181                         const char *p = w->out.buf;
 182                         size_t len = w->out.len;
 183
 184                         /* Skip the leading hunk mark of the first file. */
 185                         if (skip_first_line) {
 186                                 while (len) {
 187                                         len--;
 188                                         if (*p++ == '\n')
 189                                                 break;
 190                                 }
 191                                 skip_first_line = 0;
 192                         }
 193
 194                         write_or_die(1, p, len);
 195                 }
 196                 free(w->name);
 197                 free(w->identifier);
 198         }
 199
 200         if (old_done != todo_done)
 201                 pthread_cond_signal(&cond_write);
 202
 203         if (all_work_added && todo_done == todo_end)
 204                 pthread_cond_signal(&cond_result);
 205
 206         grep_unlock();
 207 }
 208
 209 static int skip_binary(struct grep_opt *opt, const char *filename)
 210 {
 211         if ((opt->binary & GREP_BINARY_NOMATCH)) {
 212                 static struct git_attr *attr_text;
 213                 struct git_attr_check check;
 214
 215                 if (!attr_text)
 216                         attr_text = git_attr("text");
 217                 memset(&check, 0, sizeof(check));
 218                 check.attr = attr_text;
 219                 return !git_check_attr(filename, 1, &check) &&
 220                                 ATTR_FALSE(check.value);
 221         }
 222         return 0;
 223 }
 224
 225 static void *run(void *arg)
 226 {
 227         int hit = 0;
 228         struct grep_opt *opt = arg;
 229
 230         while (1) {
 231                 struct work_item *w = get_work();
 232                 if (!w)
 233                         break;
 234
 235                 if (skip_binary(opt, (const char *)w->identifier))
 236                         continue;
 237
 238                 opt->output_priv = w;
 239                 if (w->type == WORK_SHA1) {
 240                         unsigned long sz;
 241                         void* data = load_sha1(w->identifier, &sz, w->name);
 242
 243                         if (data) {
 244                                 hit |= grep_buffer(opt, w->name, data, sz);
 245                                 free(data);
 246                         }
 247                 } else if (w->type == WORK_FILE) {
 248                         size_t sz;
 249                         void* data = load_file(w->identifier, &sz);
 250                         if (data) {
 251                                 hit |= grep_buffer(opt, w->name, data, sz);
 252                                 free(data);
 253                         }
 254                 } else {
 255                         assert(0);
 256                 }
 257
 258                 work_done(w);
 259         }
 260         free_grep_patterns(arg);
 261         free(arg);
 262
 263         return (void*) (intptr_t) hit;
 264 }
 265
 266 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 267 {
 268         struct work_item *w = opt->output_priv;
 269         strbuf_add(&w->out, buf, size);
 270 }
 271
 272 static void start_threads(struct grep_opt *opt)
 273 {
 274         int i;
 275
 276         pthread_mutex_init(&grep_mutex, NULL);
 277         pthread_mutex_init(&read_sha1_mutex, NULL);
 278         pthread_mutex_init(&grep_attr_mutex, NULL);
 279         pthread_cond_init(&cond_add, NULL);
 280         pthread_cond_init(&cond_write, NULL);
 281         pthread_cond_init(&cond_result, NULL);
 282
 283         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 284                 strbuf_init(&todo[i].out, 0);
 285         }
 286
 287         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 288                 int err;
 289                 struct grep_opt *o = grep_opt_dup(opt);
 290                 o->output = strbuf_out;
 291                 compile_grep_patterns(o);
 292                 err = pthread_create(&threads[i], NULL, run, o);
 293
 294                 if (err)
 295                         die(_("grep: failed to create thread: %s"),
 296                             strerror(err));
 297         }
 298 }
 299
 300 static int wait_all(void)
 301 {
 302         int hit = 0;
 303         int i;
 304
 305         grep_lock();
 306         all_work_added = 1;
 307
 308         /* Wait until all work is done. */
 309         while (todo_done != todo_end)
 310                 pthread_cond_wait(&cond_result, &grep_mutex);
 311
 312         /* Wake up all the consumer threads so they can see that there
 313          * is no more work to do.
 314          */
 315         pthread_cond_broadcast(&cond_add);
 316         grep_unlock();
 317
 318         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 319                 void *h;
 320                 pthread_join(threads[i], &h);
 321                 hit |= (int) (intptr_t) h;
 322         }
 323
 324         pthread_mutex_destroy(&grep_mutex);
 325         pthread_mutex_destroy(&read_sha1_mutex);
 326         pthread_mutex_destroy(&grep_attr_mutex);
 327         pthread_cond_destroy(&cond_add);
 328         pthread_cond_destroy(&cond_write);
 329         pthread_cond_destroy(&cond_result);
 330
 331         return hit;
 332 }
 333 #else /* !NO_PTHREADS */
 334 #define read_sha1_lock()
 335 #define read_sha1_unlock()
 336
 337 static int wait_all(void)
 338 {
 339         return 0;
 340 }
 341 #endif
 342
 343 static int grep_config(const char *var, const char *value, void *cb)
 344 {
 345         struct grep_opt *opt = cb;
 346         char *color = NULL;
 347
 348         switch (userdiff_config(var, value)) {
 349         case 0: break;
 350         case -1: return -1;
 351         default: return 0;
 352         }
 353
 354         if (!strcmp(var, "grep.extendedregexp")) {
 355                 if (git_config_bool(var, value))
 356                         opt->regflags |= REG_EXTENDED;
 357                 else
 358                         opt->regflags &= ~REG_EXTENDED;
 359                 return 0;
 360         }
 361
 362         if (!strcmp(var, "grep.linenumber")) {
 363                 opt->linenum = git_config_bool(var, value);
 364                 return 0;
 365         }
 366
 367         if (!strcmp(var, "color.grep"))
 368                 opt->color = git_config_colorbool(var, value);
 369         else if (!strcmp(var, "color.grep.context"))
 370                 color = opt->color_context;
 371         else if (!strcmp(var, "color.grep.filename"))
 372                 color = opt->color_filename;
 373         else if (!strcmp(var, "color.grep.function"))
 374                 color = opt->color_function;
 375         else if (!strcmp(var, "color.grep.linenumber"))
 376                 color = opt->color_lineno;
 377         else if (!strcmp(var, "color.grep.match"))
 378                 color = opt->color_match;
 379         else if (!strcmp(var, "color.grep.selected"))
 380                 color = opt->color_selected;
 381         else if (!strcmp(var, "color.grep.separator"))
 382                 color = opt->color_sep;
 383         else
 384                 return git_color_default_config(var, value, cb);
 385         if (color) {
 386                 if (!value)
 387                         return config_error_nonbool(var);
 388                 color_parse(value, var, color);
 389         }
 390         return 0;
 391 }
 392
 393 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 394 {
 395         void *data;
 396
 397         read_sha1_lock();
 398         data = read_sha1_file(sha1, type, size);
 399         read_sha1_unlock();
 400         return data;
 401 }
 402
 403 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 404                        const char *name)
 405 {
 406         enum object_type type;
 407         void *data = lock_and_read_sha1_file(sha1, &type, size);
 408
 409         if (!data)
 410                 error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
 411
 412         return data;
 413 }
 414
 415 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 416                      const char *filename, int tree_name_len)
 417 {
 418         struct strbuf pathbuf = STRBUF_INIT;
 419         char *name;
 420
 421         if (opt->relative && opt->prefix_length) {
 422                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 423                                     opt->prefix);
 424                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 425         } else {
 426                 strbuf_addstr(&pathbuf, filename);
 427         }
 428
 429         name = strbuf_detach(&pathbuf, NULL);
 430
 431 #ifndef NO_PTHREADS
 432         if (use_threads) {
 433                 grep_sha1_async(opt, name, sha1);
 434                 return 0;
 435         } else
 436 #endif
 437         {
 438                 int hit;
 439                 unsigned long sz;
 440                 void *data = load_sha1(sha1, &sz, name);
 441                 if (!data)
 442                         hit = 0;
 443                 else
 444                         hit = grep_buffer(opt, name, data, sz);
 445
 446                 free(data);
 447                 free(name);
 448                 return hit;
 449         }
 450 }
 451
 452 static void *load_file(const char *filename, size_t *sz)
 453 {
 454         struct stat st;
 455         char *data;
 456         int i;
 457
 458         if (lstat(filename, &st) < 0) {
 459         err_ret:
 460                 if (errno != ENOENT)
 461                         error(_("'%s': %s"), filename, strerror(errno));
 462                 return NULL;
 463         }
 464         if (!S_ISREG(st.st_mode))
 465                 return NULL;
 466         *sz = xsize_t(st.st_size);
 467         i = open(filename, O_RDONLY);
 468         if (i < 0)
 469                 goto err_ret;
 470         data = xmalloc(*sz + 1);
 471         if (st.st_size != read_in_full(i, data, *sz)) {
 472                 error(_("'%s': short read %s"), filename, strerror(errno));
 473                 close(i);
 474                 free(data);
 475                 return NULL;
 476         }
 477         close(i);
 478         data[*sz] = 0;
 479         return data;
 480 }
 481
 482 static int grep_file(struct grep_opt *opt, const char *filename)
 483 {
 484         struct strbuf buf = STRBUF_INIT;
 485         char *name;
 486
 487         if (opt->relative && opt->prefix_length)
 488                 quote_path_relative(filename, -1, &buf, opt->prefix);
 489         else
 490                 strbuf_addstr(&buf, filename);
 491         name = strbuf_detach(&buf, NULL);
 492
 493 #ifndef NO_PTHREADS
 494         if (use_threads) {
 495                 grep_file_async(opt, name, filename);
 496                 return 0;
 497         } else
 498 #endif
 499         {
 500                 int hit;
 501                 size_t sz;
 502                 void *data = load_file(filename, &sz);
 503                 if (!data)
 504                         hit = 0;
 505                 else
 506                         hit = grep_buffer(opt, name, data, sz);
 507
 508                 free(data);
 509                 free(name);
 510                 return hit;
 511         }
 512 }
 513
 514 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 515 {
 516         struct string_list *path_list = opt->output_priv;
 517
 518         if (len == 1 && *(const char *)data == '\0')
 519                 return;
 520         string_list_append(path_list, xstrndup(data, len));
 521 }
 522
 523 static void run_pager(struct grep_opt *opt, const char *prefix)
 524 {
 525         struct string_list *path_list = opt->output_priv;
 526         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 527         int i, status;
 528
 529         for (i = 0; i < path_list->nr; i++)
 530                 argv[i] = path_list->items[i].string;
 531         argv[path_list->nr] = NULL;
 532
 533         if (prefix && chdir(prefix))
 534                 die(_("Failed to chdir: %s"), prefix);
 535         status = run_command_v_opt(argv, RUN_USING_SHELL);
 536         if (status)
 537                 exit(status);
 538         free(argv);
 539 }
 540
 541 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 542 {
 543         int hit = 0;
 544         int nr;
 545         read_cache();
 546
 547         for (nr = 0; nr < active_nr; nr++) {
 548                 struct cache_entry *ce = active_cache[nr];
 549                 if (!S_ISREG(ce->ce_mode))
 550                         continue;
 551                 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 552                         continue;
 553                 if (skip_binary(opt, ce->name))
 554                         continue;
 555
 556                 /*
 557                  * If CE_VALID is on, we assume worktree file and its cache entry
 558                  * are identical, even if worktree file has been modified, so use
 559                  * cache version instead
 560                  */
 561                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 562                         if (ce_stage(ce))
 563                                 continue;
 564                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 565                 }
 566                 else
 567                         hit |= grep_file(opt, ce->name);
 568                 if (ce_stage(ce)) {
 569                         do {
 570                                 nr++;
 571                         } while (nr < active_nr &&
 572                                  !strcmp(ce->name, active_cache[nr]->name));
 573                         nr--; /* compensate for loop control */
 574                 }
 575                 if (hit && opt->status_only)
 576                         break;
 577         }
 578         return hit;
 579 }
 580
 581 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 582                      struct tree_desc *tree, struct strbuf *base, int tn_len)
 583 {
 584         int hit = 0;
 585         enum interesting match = entry_not_interesting;
 586         struct name_entry entry;
 587         int old_baselen = base->len;
 588
 589         while (tree_entry(tree, &entry)) {
 590                 int te_len = tree_entry_len(&entry);
 591
 592                 if (match != all_entries_interesting) {
 593                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 594                         if (match == all_entries_not_interesting)
 595                                 break;
 596                         if (match == entry_not_interesting)
 597                                 continue;
 598                 }
 599
 600                 strbuf_add(base, entry.path, te_len);
 601
 602                 if (S_ISREG(entry.mode)) {
 603                         hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 604                 }
 605                 else if (S_ISDIR(entry.mode)) {
 606                         enum object_type type;
 607                         struct tree_desc sub;
 608                         void *data;
 609                         unsigned long size;
 610
 611                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 612                         if (!data)
 613                                 die(_("unable to read tree (%s)"),
 614                                     sha1_to_hex(entry.sha1));
 615
 616                         strbuf_addch(base, '/');
 617                         init_tree_desc(&sub, data, size);
 618                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 619                         free(data);
 620                 }
 621                 strbuf_setlen(base, old_baselen);
 622
 623                 if (hit && opt->status_only)
 624                         break;
 625         }
 626         return hit;
 627 }
 628
 629 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 630                        struct object *obj, const char *name)
 631 {
 632         if (obj->type == OBJ_BLOB)
 633                 return grep_sha1(opt, obj->sha1, name, 0);
 634         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 635                 struct tree_desc tree;
 636                 void *data;
 637                 unsigned long size;
 638                 struct strbuf base;
 639                 int hit, len;
 640
 641                 read_sha1_lock();
 642                 data = read_object_with_reference(obj->sha1, tree_type,
 643                                                   &size, NULL);
 644                 read_sha1_unlock();
 645
 646                 if (!data)
 647                         die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 648
 649                 len = name ? strlen(name) : 0;
 650                 strbuf_init(&base, PATH_MAX + len + 1);
 651                 if (len) {
 652                         strbuf_add(&base, name, len);
 653                         strbuf_addch(&base, ':');
 654                 }
 655                 init_tree_desc(&tree, data, size);
 656                 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 657                 strbuf_release(&base);
 658                 free(data);
 659                 return hit;
 660         }
 661         die(_("unable to grep from object of type %s"), typename(obj->type));
 662 }
 663
 664 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 665                         const struct object_array *list)
 666 {
 667         unsigned int i;
 668         int hit = 0;
 669         const unsigned int nr = list->nr;
 670
 671         for (i = 0; i < nr; i++) {
 672                 struct object *real_obj;
 673                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 674                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 675                         hit = 1;
 676                         if (opt->status_only)
 677                                 break;
 678                 }
 679         }
 680         return hit;
 681 }
 682
 683 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 684                           int exc_std)
 685 {
 686         struct dir_struct dir;
 687         int i, hit = 0;
 688
 689         memset(&dir, 0, sizeof(dir));
 690         if (exc_std)
 691                 setup_standard_excludes(&dir);
 692
 693         fill_directory(&dir, pathspec->raw);
 694         for (i = 0; i < dir.nr; i++) {
 695                 const char *name = dir.entries[i]->name;
 696                 int namelen = strlen(name);
 697                 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 698                         continue;
 699                 hit |= grep_file(opt, dir.entries[i]->name);
 700                 if (hit && opt->status_only)
 701                         break;
 702         }
 703         return hit;
 704 }
 705
 706 static int context_callback(const struct option *opt, const char *arg,
 707                             int unset)
 708 {
 709         struct grep_opt *grep_opt = opt->value;
 710         int value;
 711         const char *endp;
 712
 713         if (unset) {
 714                 grep_opt->pre_context = grep_opt->post_context = 0;
 715                 return 0;
 716         }
 717         value = strtol(arg, (char **)&endp, 10);
 718         if (*endp) {
 719                 return error(_("switch `%c' expects a numerical value"),
 720                              opt->short_name);
 721         }
 722         grep_opt->pre_context = grep_opt->post_context = value;
 723         return 0;
 724 }
 725
 726 static int file_callback(const struct option *opt, const char *arg, int unset)
 727 {
 728         struct grep_opt *grep_opt = opt->value;
 729         int from_stdin = !strcmp(arg, "-");
 730         FILE *patterns;
 731         int lno = 0;
 732         struct strbuf sb = STRBUF_INIT;
 733
 734         patterns = from_stdin ? stdin : fopen(arg, "r");
 735         if (!patterns)
 736                 die_errno(_("cannot open '%s'"), arg);
 737         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 738                 char *s;
 739                 size_t len;
 740
 741                 /* ignore empty line like grep does */
 742                 if (sb.len == 0)
 743                         continue;
 744
 745                 s = strbuf_detach(&sb, &len);
 746                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 747         }
 748         if (!from_stdin)
 749                 fclose(patterns);
 750         strbuf_release(&sb);
 751         return 0;
 752 }
 753
 754 static int not_callback(const struct option *opt, const char *arg, int unset)
 755 {
 756         struct grep_opt *grep_opt = opt->value;
 757         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 758         return 0;
 759 }
 760
 761 static int and_callback(const struct option *opt, const char *arg, int unset)
 762 {
 763         struct grep_opt *grep_opt = opt->value;
 764         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 765         return 0;
 766 }
 767
 768 static int open_callback(const struct option *opt, const char *arg, int unset)
 769 {
 770         struct grep_opt *grep_opt = opt->value;
 771         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 772         return 0;
 773 }
 774
 775 static int close_callback(const struct option *opt, const char *arg, int unset)
 776 {
 777         struct grep_opt *grep_opt = opt->value;
 778         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 779         return 0;
 780 }
 781
 782 static int pattern_callback(const struct option *opt, const char *arg,
 783                             int unset)
 784 {
 785         struct grep_opt *grep_opt = opt->value;
 786         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 787         return 0;
 788 }
 789
 790 static int help_callback(const struct option *opt, const char *arg, int unset)
 791 {
 792         return -1;
 793 }
 794
 795 int cmd_grep(int argc, const char **argv, const char *prefix)
 796 {
 797         int hit = 0;
 798         int cached = 0, untracked = 0, opt_exclude = -1;
 799         int seen_dashdash = 0;
 800         int external_grep_allowed__ignored;
 801         const char *show_in_pager = NULL, *default_pager = "dummy";
 802         struct grep_opt opt;
 803         struct object_array list = OBJECT_ARRAY_INIT;
 804         const char **paths = NULL;
 805         struct pathspec pathspec;
 806         struct string_list path_list = STRING_LIST_INIT_NODUP;
 807         int i;
 808         int dummy;
 809         int use_index = 1;
 810         enum {
 811                 pattern_type_unspecified = 0,
 812                 pattern_type_bre,
 813                 pattern_type_ere,
 814                 pattern_type_fixed,
 815                 pattern_type_pcre,
 816         };
 817         int pattern_type = pattern_type_unspecified;
 818
 819         struct option options[] = {
 820                 OPT_BOOLEAN(0, "cached", &cached,
 821                         "search in index instead of in the work tree"),
 822                 { OPTION_BOOLEAN, 0, "index", &use_index, NULL,
 823                         "finds in contents not managed by git",
 824                         PARSE_OPT_NOARG | PARSE_OPT_NEGHELP },
 825                 OPT_BOOLEAN(0, "untracked", &untracked,
 826                         "search in both tracked and untracked files"),
 827                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 828                             "search also in ignored files", 1),
 829                 OPT_GROUP(""),
 830                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 831                         "show non-matching lines"),
 832                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 833                         "case insensitive matching"),
 834                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 835                         "match patterns only at word boundaries"),
 836                 OPT_SET_INT('a', "text", &opt.binary,
 837                         "process binary files as text", GREP_BINARY_TEXT),
 838                 OPT_SET_INT('I', NULL, &opt.binary,
 839                         "don't match patterns in binary files",
 840                         GREP_BINARY_NOMATCH),
 841                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 842                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 843                         NULL, 1 },
 844                 OPT_GROUP(""),
 845                 OPT_SET_INT('E', "extended-regexp", &pattern_type,
 846                             "use extended POSIX regular expressions",
 847                             pattern_type_ere),
 848                 OPT_SET_INT('G', "basic-regexp", &pattern_type,
 849                             "use basic POSIX regular expressions (default)",
 850                             pattern_type_bre),
 851                 OPT_SET_INT('F', "fixed-strings", &pattern_type,
 852                             "interpret patterns as fixed strings",
 853                             pattern_type_fixed),
 854                 OPT_SET_INT('P', "perl-regexp", &pattern_type,
 855                             "use Perl-compatible regular expressions",
 856                             pattern_type_pcre),
 857                 OPT_GROUP(""),
 858                 OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
 859                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 860                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 861                 OPT_NEGBIT(0, "full-name", &opt.relative,
 862                         "show filenames relative to top directory", 1),
 863                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 864                         "show only filenames instead of matching lines"),
 865                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 866                         "synonym for --files-with-matches"),
 867                 OPT_BOOLEAN('L', "files-without-match",
 868                         &opt.unmatch_name_only,
 869                         "show only the names of files without match"),
 870                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 871                         "print NUL after filenames"),
 872                 OPT_BOOLEAN('c', "count", &opt.count,
 873                         "show the number of matches instead of matching lines"),
 874                 OPT__COLOR(&opt.color, "highlight matches"),
 875                 OPT_BOOLEAN(0, "break", &opt.file_break,
 876                         "print empty line between matches from different files"),
 877                 OPT_BOOLEAN(0, "heading", &opt.heading,
 878                         "show filename only once above matches from same file"),
 879                 OPT_GROUP(""),
 880                 OPT_CALLBACK('C', "context", &opt, "n",
 881                         "show <n> context lines before and after matches",
 882                         context_callback),
 883                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 884                         "show <n> context lines before matches"),
 885                 OPT_INTEGER('A', "after-context", &opt.post_context,
 886                         "show <n> context lines after matches"),
 887                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 888                         context_callback),
 889                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 890                         "show a line with the function name before matches"),
 891                 OPT_BOOLEAN('W', "function-context", &opt.funcbody,
 892                         "show the surrounding function"),
 893                 OPT_GROUP(""),
 894                 OPT_CALLBACK('f', NULL, &opt, "file",
 895                         "read patterns from file", file_callback),
 896                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 897                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 898                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 899                   "combine patterns specified with -e",
 900                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 901                 OPT_BOOLEAN(0, "or", &dummy, ""),
 902                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 903                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 904                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 905                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 906                   open_callback },
 907                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 908                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 909                   close_callback },
 910                 OPT__QUIET(&opt.status_only,
 911                            "indicate hit with exit status without output"),
 912                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 913                         "show only matches from files that match all patterns"),
 914                 OPT_GROUP(""),
 915                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 916                         "pager", "show matching files in the pager",
 917                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 918                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 919                             "allow calling of grep(1) (ignored by this build)"),
 920                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 921                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 922                 OPT_END()
 923         };
 924
 925         /*
 926          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 927          * to show usage information and exit.
 928          */
 929         if (argc == 2 && !strcmp(argv[1], "-h"))
 930                 usage_with_options(grep_usage, options);
 931
 932         memset(&opt, 0, sizeof(opt));
 933         opt.prefix = prefix;
 934         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 935         opt.relative = 1;
 936         opt.pathname = 1;
 937         opt.pattern_tail = &opt.pattern_list;
 938         opt.header_tail = &opt.header_list;
 939         opt.regflags = REG_NEWLINE;
 940         opt.max_depth = -1;
 941
 942         strcpy(opt.color_context, "");
 943         strcpy(opt.color_filename, "");
 944         strcpy(opt.color_function, "");
 945         strcpy(opt.color_lineno, "");
 946         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 947         strcpy(opt.color_selected, "");
 948         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 949         opt.color = -1;
 950         git_config(grep_config, &opt);
 951
 952         /*
 953          * If there is no -- then the paths must exist in the working
 954          * tree.  If there is no explicit pattern specified with -e or
 955          * -f, we take the first unrecognized non option to be the
 956          * pattern, but then what follows it must be zero or more
 957          * valid refs up to the -- (if exists), and then existing
 958          * paths.  If there is an explicit pattern, then the first
 959          * unrecognized non option is the beginning of the refs list
 960          * that continues up to the -- (if exists), and then paths.
 961          */
 962         argc = parse_options(argc, argv, prefix, options, grep_usage,
 963                              PARSE_OPT_KEEP_DASHDASH |
 964                              PARSE_OPT_STOP_AT_NON_OPTION |
 965                              PARSE_OPT_NO_INTERNAL_HELP);
 966         switch (pattern_type) {
 967         case pattern_type_fixed:
 968                 opt.fixed = 1;
 969                 opt.pcre = 0;
 970                 break;
 971         case pattern_type_bre:
 972                 opt.fixed = 0;
 973                 opt.pcre = 0;
 974                 opt.regflags &= ~REG_EXTENDED;
 975                 break;
 976         case pattern_type_ere:
 977                 opt.fixed = 0;
 978                 opt.pcre = 0;
 979                 opt.regflags |= REG_EXTENDED;
 980                 break;
 981         case pattern_type_pcre:
 982                 opt.fixed = 0;
 983                 opt.pcre = 1;
 984                 break;
 985         default:
 986                 break; /* nothing */
 987         }
 988
 989         if (use_index && !startup_info->have_repository)
 990                 /* die the same way as if we did it at the beginning */
 991                 setup_git_directory();
 992
 993         /*
 994          * skip a -- separator; we know it cannot be
 995          * separating revisions from pathnames if
 996          * we haven't even had any patterns yet
 997          */
 998         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 999                 argv++;
1000                 argc--;
1001         }
1002
1003         /* First unrecognized non-option token */
1004         if (argc > 0 && !opt.pattern_list) {
1005                 append_grep_pattern(&opt, argv[0], "command line", 0,
1006                                     GREP_PATTERN);
1007                 argv++;
1008                 argc--;
1009         }
1010
1011         if (show_in_pager == default_pager)
1012                 show_in_pager = git_pager(1);
1013         if (show_in_pager) {
1014                 opt.color = 0;
1015                 opt.name_only = 1;
1016                 opt.null_following_name = 1;
1017                 opt.output_priv = &path_list;
1018                 opt.output = append_path;
1019                 string_list_append(&path_list, show_in_pager);
1020                 use_threads = 0;
1021         }
1022         if ((opt.binary & GREP_BINARY_NOMATCH))
1023                 use_threads = 0;
1024
1025         if (!opt.pattern_list)
1026                 die(_("no pattern given."));
1027         if (!opt.fixed && opt.ignore_case)
1028                 opt.regflags |= REG_ICASE;
1029
1030         compile_grep_patterns(&opt);
1031
1032         /* Check revs and then paths */
1033         for (i = 0; i < argc; i++) {
1034                 const char *arg = argv[i];
1035                 unsigned char sha1[20];
1036                 /* Is it a rev? */
1037                 if (!get_sha1(arg, sha1)) {
1038                         struct object *object = parse_object(sha1);
1039                         if (!object)
1040                                 die(_("bad object %s"), arg);
1041                         add_object_array(object, arg, &list);
1042                         continue;
1043                 }
1044                 if (!strcmp(arg, "--")) {
1045                         i++;
1046                         seen_dashdash = 1;
1047                 }
1048                 break;
1049         }
1050
1051 #ifndef NO_PTHREADS
1052         if (list.nr || cached || online_cpus() == 1)
1053                 use_threads = 0;
1054 #else
1055         use_threads = 0;
1056 #endif
1057
1058         opt.use_threads = use_threads;
1059
1060 #ifndef NO_PTHREADS
1061         if (use_threads) {
1062                 if (opt.pre_context || opt.post_context || opt.file_break ||
1063                     opt.funcbody)
1064                         skip_first_line = 1;
1065                 start_threads(&opt);
1066         }
1067 #endif
1068
1069         /* The rest are paths */
1070         if (!seen_dashdash) {
1071                 int j;
1072                 for (j = i; j < argc; j++)
1073                         verify_filename(prefix, argv[j]);
1074         }
1075
1076         paths = get_pathspec(prefix, argv + i);
1077         init_pathspec(&pathspec, paths);
1078         pathspec.max_depth = opt.max_depth;
1079         pathspec.recursive = 1;
1080
1081         if (show_in_pager && (cached || list.nr))
1082                 die(_("--open-files-in-pager only works on the worktree"));
1083
1084         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1085                 const char *pager = path_list.items[0].string;
1086                 int len = strlen(pager);
1087
1088                 if (len > 4 && is_dir_sep(pager[len - 5]))
1089                         pager += len - 4;
1090
1091                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1092                         struct strbuf buf = STRBUF_INIT;
1093                         strbuf_addf(&buf, "+/%s%s",
1094                                         strcmp("less", pager) ? "" : "*",
1095                                         opt.pattern_list->pattern);
1096                         string_list_append(&path_list, buf.buf);
1097                         strbuf_detach(&buf, NULL);
1098                 }
1099         }
1100
1101         if (!show_in_pager)
1102                 setup_pager();
1103
1104         if (!use_index && (untracked || cached))
1105                 die(_("--cached or --untracked cannot be used with --no-index."));
1106
1107         if (!use_index || untracked) {
1108                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1109                 if (list.nr)
1110                         die(_("--no-index or --untracked cannot be used with revs."));
1111                 hit = grep_directory(&opt, &pathspec, use_exclude);
1112         } else if (0 <= opt_exclude) {
1113                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1114         } else if (!list.nr) {
1115                 if (!cached)
1116                         setup_work_tree();
1117
1118                 hit = grep_cache(&opt, &pathspec, cached);
1119         } else {
1120                 if (cached)
1121                         die(_("both --cached and trees are given."));
1122                 hit = grep_objects(&opt, &pathspec, &list);
1123         }
1124
1125         if (use_threads)
1126                 hit |= wait_all();
1127         if (hit && show_in_pager)
1128                 run_pager(&opt, prefix);
1129         free_grep_patterns(&opt);
1130         return !hit;
1131 }