builtin/grep.c

   1 /*
   2  * Builtin "git grep"
   3  *
   4  * Copyright (c) 2006 Junio C Hamano
   5  */
   6 #include "cache.h"
   7 #include "blob.h"
   8 #include "tree.h"
   9 #include "commit.h"
  10 #include "tag.h"
  11 #include "tree-walk.h"
  12 #include "builtin.h"
  13 #include "parse-options.h"
  14 #include "string-list.h"
  15 #include "run-command.h"
  16 #include "userdiff.h"
  17 #include "grep.h"
  18 #include "quote.h"
  19 #include "dir.h"
  20
  21 static char const * const grep_usage[] = {
  22         "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
  23         NULL
  24 };
  25
  26 static int use_threads = 1;
  27
  28 #ifndef NO_PTHREADS
  29 #define THREADS 8
  30 static pthread_t threads[THREADS];
  31
  32 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
  33                        const char *name);
  34 static void *load_file(const char *filename, size_t *sz);
  35
  36 enum work_type {WORK_SHA1, WORK_FILE};
  37
  38 /* We use one producer thread and THREADS consumer
  39  * threads. The producer adds struct work_items to 'todo' and the
  40  * consumers pick work items from the same array.
  41  */
  42 struct work_item {
  43         enum work_type type;
  44         char *name;
  45
  46         /* if type == WORK_SHA1, then 'identifier' is a SHA1,
  47          * otherwise type == WORK_FILE, and 'identifier' is a NUL
  48          * terminated filename.
  49          */
  50         void *identifier;
  51         char done;
  52         struct strbuf out;
  53 };
  54
  55 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  56  * that have been or are processed by a consumer thread. We haven't
  57  * written the result for these to stdout yet.
  58  *
  59  * The work_items in [todo_start, todo_end) are waiting to be picked
  60  * up by a consumer thread.
  61  *
  62  * The ranges are modulo TODO_SIZE.
  63  */
  64 #define TODO_SIZE 128
  65 static struct work_item todo[TODO_SIZE];
  66 static int todo_start;
  67 static int todo_end;
  68 static int todo_done;
  69
  70 /* Has all work items been added? */
  71 static int all_work_added;
  72
  73 /* This lock protects all the variables above. */
  74 static pthread_mutex_t grep_mutex;
  75
  76 static inline void grep_lock(void)
  77 {
  78         if (use_threads)
  79                 pthread_mutex_lock(&grep_mutex);
  80 }
  81
  82 static inline void grep_unlock(void)
  83 {
  84         if (use_threads)
  85                 pthread_mutex_unlock(&grep_mutex);
  86 }
  87
  88 /* Used to serialize calls to read_sha1_file. */
  89 static pthread_mutex_t read_sha1_mutex;
  90
  91 static inline void read_sha1_lock(void)
  92 {
  93         if (use_threads)
  94                 pthread_mutex_lock(&read_sha1_mutex);
  95 }
  96
  97 static inline void read_sha1_unlock(void)
  98 {
  99         if (use_threads)
 100                 pthread_mutex_unlock(&read_sha1_mutex);
 101 }
 102
 103 /* Signalled when a new work_item is added to todo. */
 104 static pthread_cond_t cond_add;
 105
 106 /* Signalled when the result from one work_item is written to
 107  * stdout.
 108  */
 109 static pthread_cond_t cond_write;
 110
 111 /* Signalled when we are finished with everything. */
 112 static pthread_cond_t cond_result;
 113
 114 static int skip_first_line;
 115
 116 static void add_work(enum work_type type, char *name, void *id)
 117 {
 118         grep_lock();
 119
 120         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 121                 pthread_cond_wait(&cond_write, &grep_mutex);
 122         }
 123
 124         todo[todo_end].type = type;
 125         todo[todo_end].name = name;
 126         todo[todo_end].identifier = id;
 127         todo[todo_end].done = 0;
 128         strbuf_reset(&todo[todo_end].out);
 129         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 130
 131         pthread_cond_signal(&cond_add);
 132         grep_unlock();
 133 }
 134
 135 static struct work_item *get_work(void)
 136 {
 137         struct work_item *ret;
 138
 139         grep_lock();
 140         while (todo_start == todo_end && !all_work_added) {
 141                 pthread_cond_wait(&cond_add, &grep_mutex);
 142         }
 143
 144         if (todo_start == todo_end && all_work_added) {
 145                 ret = NULL;
 146         } else {
 147                 ret = &todo[todo_start];
 148                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 149         }
 150         grep_unlock();
 151         return ret;
 152 }
 153
 154 static void grep_sha1_async(struct grep_opt *opt, char *name,
 155                             const unsigned char *sha1)
 156 {
 157         unsigned char *s;
 158         s = xmalloc(20);
 159         memcpy(s, sha1, 20);
 160         add_work(WORK_SHA1, name, s);
 161 }
 162
 163 static void grep_file_async(struct grep_opt *opt, char *name,
 164                             const char *filename)
 165 {
 166         add_work(WORK_FILE, name, xstrdup(filename));
 167 }
 168
 169 static void work_done(struct work_item *w)
 170 {
 171         int old_done;
 172
 173         grep_lock();
 174         w->done = 1;
 175         old_done = todo_done;
 176         for(; todo[todo_done].done && todo_done != todo_start;
 177             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 178                 w = &todo[todo_done];
 179                 if (w->out.len) {
 180                         const char *p = w->out.buf;
 181                         size_t len = w->out.len;
 182
 183                         /* Skip the leading hunk mark of the first file. */
 184                         if (skip_first_line) {
 185                                 while (len) {
 186                                         len--;
 187                                         if (*p++ == '\n')
 188                                                 break;
 189                                 }
 190                                 skip_first_line = 0;
 191                         }
 192
 193                         write_or_die(1, p, len);
 194                 }
 195                 free(w->name);
 196                 free(w->identifier);
 197         }
 198
 199         if (old_done != todo_done)
 200                 pthread_cond_signal(&cond_write);
 201
 202         if (all_work_added && todo_done == todo_end)
 203                 pthread_cond_signal(&cond_result);
 204
 205         grep_unlock();
 206 }
 207
 208 static void *run(void *arg)
 209 {
 210         int hit = 0;
 211         struct grep_opt *opt = arg;
 212
 213         while (1) {
 214                 struct work_item *w = get_work();
 215                 if (!w)
 216                         break;
 217
 218                 opt->output_priv = w;
 219                 if (w->type == WORK_SHA1) {
 220                         unsigned long sz;
 221                         void* data = load_sha1(w->identifier, &sz, w->name);
 222
 223                         if (data) {
 224                                 hit |= grep_buffer(opt, w->name, data, sz);
 225                                 free(data);
 226                         }
 227                 } else if (w->type == WORK_FILE) {
 228                         size_t sz;
 229                         void* data = load_file(w->identifier, &sz);
 230                         if (data) {
 231                                 hit |= grep_buffer(opt, w->name, data, sz);
 232                                 free(data);
 233                         }
 234                 } else {
 235                         assert(0);
 236                 }
 237
 238                 work_done(w);
 239         }
 240         free_grep_patterns(arg);
 241         free(arg);
 242
 243         return (void*) (intptr_t) hit;
 244 }
 245
 246 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 247 {
 248         struct work_item *w = opt->output_priv;
 249         strbuf_add(&w->out, buf, size);
 250 }
 251
 252 static void start_threads(struct grep_opt *opt)
 253 {
 254         int i;
 255
 256         pthread_mutex_init(&grep_mutex, NULL);
 257         pthread_mutex_init(&read_sha1_mutex, NULL);
 258         pthread_mutex_init(&grep_attr_mutex, NULL);
 259         pthread_cond_init(&cond_add, NULL);
 260         pthread_cond_init(&cond_write, NULL);
 261         pthread_cond_init(&cond_result, NULL);
 262
 263         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 264                 strbuf_init(&todo[i].out, 0);
 265         }
 266
 267         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 268                 int err;
 269                 struct grep_opt *o = grep_opt_dup(opt);
 270                 o->output = strbuf_out;
 271                 compile_grep_patterns(o);
 272                 err = pthread_create(&threads[i], NULL, run, o);
 273
 274                 if (err)
 275                         die(_("grep: failed to create thread: %s"),
 276                             strerror(err));
 277         }
 278 }
 279
 280 static int wait_all(void)
 281 {
 282         int hit = 0;
 283         int i;
 284
 285         grep_lock();
 286         all_work_added = 1;
 287
 288         /* Wait until all work is done. */
 289         while (todo_done != todo_end)
 290                 pthread_cond_wait(&cond_result, &grep_mutex);
 291
 292         /* Wake up all the consumer threads so they can see that there
 293          * is no more work to do.
 294          */
 295         pthread_cond_broadcast(&cond_add);
 296         grep_unlock();
 297
 298         for (i = 0; i < ARRAY_SIZE(threads); i++) {
 299                 void *h;
 300                 pthread_join(threads[i], &h);
 301                 hit |= (int) (intptr_t) h;
 302         }
 303
 304         pthread_mutex_destroy(&grep_mutex);
 305         pthread_mutex_destroy(&read_sha1_mutex);
 306         pthread_mutex_destroy(&grep_attr_mutex);
 307         pthread_cond_destroy(&cond_add);
 308         pthread_cond_destroy(&cond_write);
 309         pthread_cond_destroy(&cond_result);
 310
 311         return hit;
 312 }
 313 #else /* !NO_PTHREADS */
 314 #define read_sha1_lock()
 315 #define read_sha1_unlock()
 316
 317 static int wait_all(void)
 318 {
 319         return 0;
 320 }
 321 #endif
 322
 323 static int grep_config(const char *var, const char *value, void *cb)
 324 {
 325         struct grep_opt *opt = cb;
 326         char *color = NULL;
 327
 328         if (userdiff_config(var, value) < 0)
 329                 return -1;
 330
 331         if (!strcmp(var, "grep.extendedregexp")) {
 332                 if (git_config_bool(var, value))
 333                         opt->regflags |= REG_EXTENDED;
 334                 else
 335                         opt->regflags &= ~REG_EXTENDED;
 336                 return 0;
 337         }
 338
 339         if (!strcmp(var, "grep.linenumber")) {
 340                 opt->linenum = git_config_bool(var, value);
 341                 return 0;
 342         }
 343
 344         if (!strcmp(var, "color.grep"))
 345                 opt->color = git_config_colorbool(var, value);
 346         else if (!strcmp(var, "color.grep.context"))
 347                 color = opt->color_context;
 348         else if (!strcmp(var, "color.grep.filename"))
 349                 color = opt->color_filename;
 350         else if (!strcmp(var, "color.grep.function"))
 351                 color = opt->color_function;
 352         else if (!strcmp(var, "color.grep.linenumber"))
 353                 color = opt->color_lineno;
 354         else if (!strcmp(var, "color.grep.match"))
 355                 color = opt->color_match;
 356         else if (!strcmp(var, "color.grep.selected"))
 357                 color = opt->color_selected;
 358         else if (!strcmp(var, "color.grep.separator"))
 359                 color = opt->color_sep;
 360         else
 361                 return git_color_default_config(var, value, cb);
 362         if (color) {
 363                 if (!value)
 364                         return config_error_nonbool(var);
 365                 color_parse(value, var, color);
 366         }
 367         return 0;
 368 }
 369
 370 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 371 {
 372         void *data;
 373
 374         read_sha1_lock();
 375         data = read_sha1_file(sha1, type, size);
 376         read_sha1_unlock();
 377         return data;
 378 }
 379
 380 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
 381                        const char *name)
 382 {
 383         enum object_type type;
 384         void *data = lock_and_read_sha1_file(sha1, &type, size);
 385
 386         if (!data)
 387                 error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
 388
 389         return data;
 390 }
 391
 392 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 393                      const char *filename, int tree_name_len)
 394 {
 395         struct strbuf pathbuf = STRBUF_INIT;
 396         char *name;
 397
 398         if (opt->relative && opt->prefix_length) {
 399                 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
 400                                     opt->prefix);
 401                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 402         } else {
 403                 strbuf_addstr(&pathbuf, filename);
 404         }
 405
 406         name = strbuf_detach(&pathbuf, NULL);
 407
 408 #ifndef NO_PTHREADS
 409         if (use_threads) {
 410                 grep_sha1_async(opt, name, sha1);
 411                 return 0;
 412         } else
 413 #endif
 414         {
 415                 int hit;
 416                 unsigned long sz;
 417                 void *data = load_sha1(sha1, &sz, name);
 418                 if (!data)
 419                         hit = 0;
 420                 else
 421                         hit = grep_buffer(opt, name, data, sz);
 422
 423                 free(data);
 424                 free(name);
 425                 return hit;
 426         }
 427 }
 428
 429 static void *load_file(const char *filename, size_t *sz)
 430 {
 431         struct stat st;
 432         char *data;
 433         int i;
 434
 435         if (lstat(filename, &st) < 0) {
 436         err_ret:
 437                 if (errno != ENOENT)
 438                         error(_("'%s': %s"), filename, strerror(errno));
 439                 return NULL;
 440         }
 441         if (!S_ISREG(st.st_mode))
 442                 return NULL;
 443         *sz = xsize_t(st.st_size);
 444         i = open(filename, O_RDONLY);
 445         if (i < 0)
 446                 goto err_ret;
 447         data = xmalloc(*sz + 1);
 448         if (st.st_size != read_in_full(i, data, *sz)) {
 449                 error(_("'%s': short read %s"), filename, strerror(errno));
 450                 close(i);
 451                 free(data);
 452                 return NULL;
 453         }
 454         close(i);
 455         data[*sz] = 0;
 456         return data;
 457 }
 458
 459 static int grep_file(struct grep_opt *opt, const char *filename)
 460 {
 461         struct strbuf buf = STRBUF_INIT;
 462         char *name;
 463
 464         if (opt->relative && opt->prefix_length)
 465                 quote_path_relative(filename, -1, &buf, opt->prefix);
 466         else
 467                 strbuf_addstr(&buf, filename);
 468         name = strbuf_detach(&buf, NULL);
 469
 470 #ifndef NO_PTHREADS
 471         if (use_threads) {
 472                 grep_file_async(opt, name, filename);
 473                 return 0;
 474         } else
 475 #endif
 476         {
 477                 int hit;
 478                 size_t sz;
 479                 void *data = load_file(filename, &sz);
 480                 if (!data)
 481                         hit = 0;
 482                 else
 483                         hit = grep_buffer(opt, name, data, sz);
 484
 485                 free(data);
 486                 free(name);
 487                 return hit;
 488         }
 489 }
 490
 491 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 492 {
 493         struct string_list *path_list = opt->output_priv;
 494
 495         if (len == 1 && *(const char *)data == '\0')
 496                 return;
 497         string_list_append(path_list, xstrndup(data, len));
 498 }
 499
 500 static void run_pager(struct grep_opt *opt, const char *prefix)
 501 {
 502         struct string_list *path_list = opt->output_priv;
 503         const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
 504         int i, status;
 505
 506         for (i = 0; i < path_list->nr; i++)
 507                 argv[i] = path_list->items[i].string;
 508         argv[path_list->nr] = NULL;
 509
 510         if (prefix && chdir(prefix))
 511                 die(_("Failed to chdir: %s"), prefix);
 512         status = run_command_v_opt(argv, RUN_USING_SHELL);
 513         if (status)
 514                 exit(status);
 515         free(argv);
 516 }
 517
 518 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
 519 {
 520         int hit = 0;
 521         int nr;
 522         read_cache();
 523
 524         for (nr = 0; nr < active_nr; nr++) {
 525                 struct cache_entry *ce = active_cache[nr];
 526                 if (!S_ISREG(ce->ce_mode))
 527                         continue;
 528                 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
 529                         continue;
 530                 /*
 531                  * If CE_VALID is on, we assume worktree file and its cache entry
 532                  * are identical, even if worktree file has been modified, so use
 533                  * cache version instead
 534                  */
 535                 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
 536                         if (ce_stage(ce))
 537                                 continue;
 538                         hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 539                 }
 540                 else
 541                         hit |= grep_file(opt, ce->name);
 542                 if (ce_stage(ce)) {
 543                         do {
 544                                 nr++;
 545                         } while (nr < active_nr &&
 546                                  !strcmp(ce->name, active_cache[nr]->name));
 547                         nr--; /* compensate for loop control */
 548                 }
 549                 if (hit && opt->status_only)
 550                         break;
 551         }
 552         return hit;
 553 }
 554
 555 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 556                      struct tree_desc *tree, struct strbuf *base, int tn_len)
 557 {
 558         int hit = 0;
 559         enum interesting match = entry_not_interesting;
 560         struct name_entry entry;
 561         int old_baselen = base->len;
 562
 563         while (tree_entry(tree, &entry)) {
 564                 int te_len = tree_entry_len(&entry);
 565
 566                 if (match != all_entries_interesting) {
 567                         match = tree_entry_interesting(&entry, base, tn_len, pathspec);
 568                         if (match == all_entries_not_interesting)
 569                                 break;
 570                         if (match == entry_not_interesting)
 571                                 continue;
 572                 }
 573
 574                 strbuf_add(base, entry.path, te_len);
 575
 576                 if (S_ISREG(entry.mode)) {
 577                         hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
 578                 }
 579                 else if (S_ISDIR(entry.mode)) {
 580                         enum object_type type;
 581                         struct tree_desc sub;
 582                         void *data;
 583                         unsigned long size;
 584
 585                         data = lock_and_read_sha1_file(entry.sha1, &type, &size);
 586                         if (!data)
 587                                 die(_("unable to read tree (%s)"),
 588                                     sha1_to_hex(entry.sha1));
 589
 590                         strbuf_addch(base, '/');
 591                         init_tree_desc(&sub, data, size);
 592                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
 593                         free(data);
 594                 }
 595                 strbuf_setlen(base, old_baselen);
 596
 597                 if (hit && opt->status_only)
 598                         break;
 599         }
 600         return hit;
 601 }
 602
 603 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 604                        struct object *obj, const char *name)
 605 {
 606         if (obj->type == OBJ_BLOB)
 607                 return grep_sha1(opt, obj->sha1, name, 0);
 608         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 609                 struct tree_desc tree;
 610                 void *data;
 611                 unsigned long size;
 612                 struct strbuf base;
 613                 int hit, len;
 614
 615                 read_sha1_lock();
 616                 data = read_object_with_reference(obj->sha1, tree_type,
 617                                                   &size, NULL);
 618                 read_sha1_unlock();
 619
 620                 if (!data)
 621                         die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
 622
 623                 len = name ? strlen(name) : 0;
 624                 strbuf_init(&base, PATH_MAX + len + 1);
 625                 if (len) {
 626                         strbuf_add(&base, name, len);
 627                         strbuf_addch(&base, ':');
 628                 }
 629                 init_tree_desc(&tree, data, size);
 630                 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
 631                 strbuf_release(&base);
 632                 free(data);
 633                 return hit;
 634         }
 635         die(_("unable to grep from object of type %s"), typename(obj->type));
 636 }
 637
 638 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 639                         const struct object_array *list)
 640 {
 641         unsigned int i;
 642         int hit = 0;
 643         const unsigned int nr = list->nr;
 644
 645         for (i = 0; i < nr; i++) {
 646                 struct object *real_obj;
 647                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 648                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
 649                         hit = 1;
 650                         if (opt->status_only)
 651                                 break;
 652                 }
 653         }
 654         return hit;
 655 }
 656
 657 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 658                           int exc_std)
 659 {
 660         struct dir_struct dir;
 661         int i, hit = 0;
 662
 663         memset(&dir, 0, sizeof(dir));
 664         if (exc_std)
 665                 setup_standard_excludes(&dir);
 666
 667         fill_directory(&dir, pathspec->raw);
 668         for (i = 0; i < dir.nr; i++) {
 669                 const char *name = dir.entries[i]->name;
 670                 int namelen = strlen(name);
 671                 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
 672                         continue;
 673                 hit |= grep_file(opt, dir.entries[i]->name);
 674                 if (hit && opt->status_only)
 675                         break;
 676         }
 677         return hit;
 678 }
 679
 680 static int context_callback(const struct option *opt, const char *arg,
 681                             int unset)
 682 {
 683         struct grep_opt *grep_opt = opt->value;
 684         int value;
 685         const char *endp;
 686
 687         if (unset) {
 688                 grep_opt->pre_context = grep_opt->post_context = 0;
 689                 return 0;
 690         }
 691         value = strtol(arg, (char **)&endp, 10);
 692         if (*endp) {
 693                 return error(_("switch `%c' expects a numerical value"),
 694                              opt->short_name);
 695         }
 696         grep_opt->pre_context = grep_opt->post_context = value;
 697         return 0;
 698 }
 699
 700 static int file_callback(const struct option *opt, const char *arg, int unset)
 701 {
 702         struct grep_opt *grep_opt = opt->value;
 703         int from_stdin = !strcmp(arg, "-");
 704         FILE *patterns;
 705         int lno = 0;
 706         struct strbuf sb = STRBUF_INIT;
 707
 708         patterns = from_stdin ? stdin : fopen(arg, "r");
 709         if (!patterns)
 710                 die_errno(_("cannot open '%s'"), arg);
 711         while (strbuf_getline(&sb, patterns, '\n') == 0) {
 712                 char *s;
 713                 size_t len;
 714
 715                 /* ignore empty line like grep does */
 716                 if (sb.len == 0)
 717                         continue;
 718
 719                 s = strbuf_detach(&sb, &len);
 720                 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
 721         }
 722         if (!from_stdin)
 723                 fclose(patterns);
 724         strbuf_release(&sb);
 725         return 0;
 726 }
 727
 728 static int not_callback(const struct option *opt, const char *arg, int unset)
 729 {
 730         struct grep_opt *grep_opt = opt->value;
 731         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 732         return 0;
 733 }
 734
 735 static int and_callback(const struct option *opt, const char *arg, int unset)
 736 {
 737         struct grep_opt *grep_opt = opt->value;
 738         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 739         return 0;
 740 }
 741
 742 static int open_callback(const struct option *opt, const char *arg, int unset)
 743 {
 744         struct grep_opt *grep_opt = opt->value;
 745         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 746         return 0;
 747 }
 748
 749 static int close_callback(const struct option *opt, const char *arg, int unset)
 750 {
 751         struct grep_opt *grep_opt = opt->value;
 752         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 753         return 0;
 754 }
 755
 756 static int pattern_callback(const struct option *opt, const char *arg,
 757                             int unset)
 758 {
 759         struct grep_opt *grep_opt = opt->value;
 760         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 761         return 0;
 762 }
 763
 764 static int help_callback(const struct option *opt, const char *arg, int unset)
 765 {
 766         return -1;
 767 }
 768
 769 int cmd_grep(int argc, const char **argv, const char *prefix)
 770 {
 771         int hit = 0;
 772         int cached = 0, untracked = 0, opt_exclude = -1;
 773         int seen_dashdash = 0;
 774         int external_grep_allowed__ignored;
 775         const char *show_in_pager = NULL, *default_pager = "dummy";
 776         struct grep_opt opt;
 777         struct object_array list = OBJECT_ARRAY_INIT;
 778         const char **paths = NULL;
 779         struct pathspec pathspec;
 780         struct string_list path_list = STRING_LIST_INIT_NODUP;
 781         int i;
 782         int dummy;
 783         int use_index = 1;
 784         enum {
 785                 pattern_type_unspecified = 0,
 786                 pattern_type_bre,
 787                 pattern_type_ere,
 788                 pattern_type_fixed,
 789                 pattern_type_pcre,
 790         };
 791         int pattern_type = pattern_type_unspecified;
 792
 793         struct option options[] = {
 794                 OPT_BOOLEAN(0, "cached", &cached,
 795                         "search in index instead of in the work tree"),
 796                 { OPTION_BOOLEAN, 0, "index", &use_index, NULL,
 797                         "finds in contents not managed by git",
 798                         PARSE_OPT_NOARG | PARSE_OPT_NEGHELP },
 799                 OPT_BOOLEAN(0, "untracked", &untracked,
 800                         "search in both tracked and untracked files"),
 801                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 802                             "search also in ignored files", 1),
 803                 OPT_GROUP(""),
 804                 OPT_BOOLEAN('v', "invert-match", &opt.invert,
 805                         "show non-matching lines"),
 806                 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 807                         "case insensitive matching"),
 808                 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 809                         "match patterns only at word boundaries"),
 810                 OPT_SET_INT('a', "text", &opt.binary,
 811                         "process binary files as text", GREP_BINARY_TEXT),
 812                 OPT_SET_INT('I', NULL, &opt.binary,
 813                         "don't match patterns in binary files",
 814                         GREP_BINARY_NOMATCH),
 815                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 816                         "descend at most <depth> levels", PARSE_OPT_NONEG,
 817                         NULL, 1 },
 818                 OPT_GROUP(""),
 819                 OPT_SET_INT('E', "extended-regexp", &pattern_type,
 820                             "use extended POSIX regular expressions",
 821                             pattern_type_ere),
 822                 OPT_SET_INT('G', "basic-regexp", &pattern_type,
 823                             "use basic POSIX regular expressions (default)",
 824                             pattern_type_bre),
 825                 OPT_SET_INT('F', "fixed-strings", &pattern_type,
 826                             "interpret patterns as fixed strings",
 827                             pattern_type_fixed),
 828                 OPT_SET_INT('P', "perl-regexp", &pattern_type,
 829                             "use Perl-compatible regular expressions",
 830                             pattern_type_pcre),
 831                 OPT_GROUP(""),
 832                 OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
 833                 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 834                 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 835                 OPT_NEGBIT(0, "full-name", &opt.relative,
 836                         "show filenames relative to top directory", 1),
 837                 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 838                         "show only filenames instead of matching lines"),
 839                 OPT_BOOLEAN(0, "name-only", &opt.name_only,
 840                         "synonym for --files-with-matches"),
 841                 OPT_BOOLEAN('L', "files-without-match",
 842                         &opt.unmatch_name_only,
 843                         "show only the names of files without match"),
 844                 OPT_BOOLEAN('z', "null", &opt.null_following_name,
 845                         "print NUL after filenames"),
 846                 OPT_BOOLEAN('c', "count", &opt.count,
 847                         "show the number of matches instead of matching lines"),
 848                 OPT__COLOR(&opt.color, "highlight matches"),
 849                 OPT_BOOLEAN(0, "break", &opt.file_break,
 850                         "print empty line between matches from different files"),
 851                 OPT_BOOLEAN(0, "heading", &opt.heading,
 852                         "show filename only once above matches from same file"),
 853                 OPT_GROUP(""),
 854                 OPT_CALLBACK('C', "context", &opt, "n",
 855                         "show <n> context lines before and after matches",
 856                         context_callback),
 857                 OPT_INTEGER('B', "before-context", &opt.pre_context,
 858                         "show <n> context lines before matches"),
 859                 OPT_INTEGER('A', "after-context", &opt.post_context,
 860                         "show <n> context lines after matches"),
 861                 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 862                         context_callback),
 863                 OPT_BOOLEAN('p', "show-function", &opt.funcname,
 864                         "show a line with the function name before matches"),
 865                 OPT_BOOLEAN('W', "function-context", &opt.funcbody,
 866                         "show the surrounding function"),
 867                 OPT_GROUP(""),
 868                 OPT_CALLBACK('f', NULL, &opt, "file",
 869                         "read patterns from file", file_callback),
 870                 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 871                         "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 872                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
 873                   "combine patterns specified with -e",
 874                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 875                 OPT_BOOLEAN(0, "or", &dummy, ""),
 876                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 877                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 878                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 879                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 880                   open_callback },
 881                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 882                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 883                   close_callback },
 884                 OPT__QUIET(&opt.status_only,
 885                            "indicate hit with exit status without output"),
 886                 OPT_BOOLEAN(0, "all-match", &opt.all_match,
 887                         "show only matches from files that match all patterns"),
 888                 OPT_GROUP(""),
 889                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
 890                         "pager", "show matching files in the pager",
 891                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
 892                 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 893                             "allow calling of grep(1) (ignored by this build)"),
 894                 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 895                   PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 896                 OPT_END()
 897         };
 898
 899         /*
 900          * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 901          * to show usage information and exit.
 902          */
 903         if (argc == 2 && !strcmp(argv[1], "-h"))
 904                 usage_with_options(grep_usage, options);
 905
 906         memset(&opt, 0, sizeof(opt));
 907         opt.prefix = prefix;
 908         opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 909         opt.relative = 1;
 910         opt.pathname = 1;
 911         opt.pattern_tail = &opt.pattern_list;
 912         opt.header_tail = &opt.header_list;
 913         opt.regflags = REG_NEWLINE;
 914         opt.max_depth = -1;
 915
 916         strcpy(opt.color_context, "");
 917         strcpy(opt.color_filename, "");
 918         strcpy(opt.color_function, "");
 919         strcpy(opt.color_lineno, "");
 920         strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
 921         strcpy(opt.color_selected, "");
 922         strcpy(opt.color_sep, GIT_COLOR_CYAN);
 923         opt.color = -1;
 924         git_config(grep_config, &opt);
 925
 926         /*
 927          * If there is no -- then the paths must exist in the working
 928          * tree.  If there is no explicit pattern specified with -e or
 929          * -f, we take the first unrecognized non option to be the
 930          * pattern, but then what follows it must be zero or more
 931          * valid refs up to the -- (if exists), and then existing
 932          * paths.  If there is an explicit pattern, then the first
 933          * unrecognized non option is the beginning of the refs list
 934          * that continues up to the -- (if exists), and then paths.
 935          */
 936         argc = parse_options(argc, argv, prefix, options, grep_usage,
 937                              PARSE_OPT_KEEP_DASHDASH |
 938                              PARSE_OPT_STOP_AT_NON_OPTION |
 939                              PARSE_OPT_NO_INTERNAL_HELP);
 940         switch (pattern_type) {
 941         case pattern_type_fixed:
 942                 opt.fixed = 1;
 943                 opt.pcre = 0;
 944                 break;
 945         case pattern_type_bre:
 946                 opt.fixed = 0;
 947                 opt.pcre = 0;
 948                 opt.regflags &= ~REG_EXTENDED;
 949                 break;
 950         case pattern_type_ere:
 951                 opt.fixed = 0;
 952                 opt.pcre = 0;
 953                 opt.regflags |= REG_EXTENDED;
 954                 break;
 955         case pattern_type_pcre:
 956                 opt.fixed = 0;
 957                 opt.pcre = 1;
 958                 break;
 959         default:
 960                 break; /* nothing */
 961         }
 962
 963         if (use_index && !startup_info->have_repository)
 964                 /* die the same way as if we did it at the beginning */
 965                 setup_git_directory();
 966
 967         /*
 968          * skip a -- separator; we know it cannot be
 969          * separating revisions from pathnames if
 970          * we haven't even had any patterns yet
 971          */
 972         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
 973                 argv++;
 974                 argc--;
 975         }
 976
 977         /* First unrecognized non-option token */
 978         if (argc > 0 && !opt.pattern_list) {
 979                 append_grep_pattern(&opt, argv[0], "command line", 0,
 980                                     GREP_PATTERN);
 981                 argv++;
 982                 argc--;
 983         }
 984
 985         if (show_in_pager == default_pager)
 986                 show_in_pager = git_pager(1);
 987         if (show_in_pager) {
 988                 opt.color = 0;
 989                 opt.name_only = 1;
 990                 opt.null_following_name = 1;
 991                 opt.output_priv = &path_list;
 992                 opt.output = append_path;
 993                 string_list_append(&path_list, show_in_pager);
 994                 use_threads = 0;
 995         }
 996
 997         if (!opt.pattern_list)
 998                 die(_("no pattern given."));
 999         if (!opt.fixed && opt.ignore_case)
1000                 opt.regflags |= REG_ICASE;
1001
1002         compile_grep_patterns(&opt);
1003
1004         /* Check revs and then paths */
1005         for (i = 0; i < argc; i++) {
1006                 const char *arg = argv[i];
1007                 unsigned char sha1[20];
1008                 /* Is it a rev? */
1009                 if (!get_sha1(arg, sha1)) {
1010                         struct object *object = parse_object(sha1);
1011                         if (!object)
1012                                 die(_("bad object %s"), arg);
1013                         add_object_array(object, arg, &list);
1014                         continue;
1015                 }
1016                 if (!strcmp(arg, "--")) {
1017                         i++;
1018                         seen_dashdash = 1;
1019                 }
1020                 break;
1021         }
1022
1023 #ifndef NO_PTHREADS
1024         if (list.nr || cached || online_cpus() == 1)
1025                 use_threads = 0;
1026 #else
1027         use_threads = 0;
1028 #endif
1029
1030         opt.use_threads = use_threads;
1031
1032 #ifndef NO_PTHREADS
1033         if (use_threads) {
1034                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1035                     && (opt.pre_context || opt.post_context ||
1036                         opt.file_break || opt.funcbody))
1037                         skip_first_line = 1;
1038                 start_threads(&opt);
1039         }
1040 #endif
1041
1042         /* The rest are paths */
1043         if (!seen_dashdash) {
1044                 int j;
1045                 for (j = i; j < argc; j++)
1046                         verify_filename(prefix, argv[j]);
1047         }
1048
1049         paths = get_pathspec(prefix, argv + i);
1050         init_pathspec(&pathspec, paths);
1051         pathspec.max_depth = opt.max_depth;
1052         pathspec.recursive = 1;
1053
1054         if (show_in_pager && (cached || list.nr))
1055                 die(_("--open-files-in-pager only works on the worktree"));
1056
1057         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1058                 const char *pager = path_list.items[0].string;
1059                 int len = strlen(pager);
1060
1061                 if (len > 4 && is_dir_sep(pager[len - 5]))
1062                         pager += len - 4;
1063
1064                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1065                         struct strbuf buf = STRBUF_INIT;
1066                         strbuf_addf(&buf, "+/%s%s",
1067                                         strcmp("less", pager) ? "" : "*",
1068                                         opt.pattern_list->pattern);
1069                         string_list_append(&path_list, buf.buf);
1070                         strbuf_detach(&buf, NULL);
1071                 }
1072         }
1073
1074         if (!show_in_pager)
1075                 setup_pager();
1076
1077         if (!use_index && (untracked || cached))
1078                 die(_("--cached or --untracked cannot be used with --no-index."));
1079
1080         if (!use_index || untracked) {
1081                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1082                 if (list.nr)
1083                         die(_("--no-index or --untracked cannot be used with revs."));
1084                 hit = grep_directory(&opt, &pathspec, use_exclude);
1085         } else if (0 <= opt_exclude) {
1086                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1087         } else if (!list.nr) {
1088                 if (!cached)
1089                         setup_work_tree();
1090
1091                 hit = grep_cache(&opt, &pathspec, cached);
1092         } else {
1093                 if (cached)
1094                         die(_("both --cached and trees are given."));
1095                 hit = grep_objects(&opt, &pathspec, &list);
1096         }
1097
1098         if (use_threads)
1099                 hit |= wait_all();
1100         if (hit && show_in_pager)
1101                 run_pager(&opt, prefix);
1102         free_grep_patterns(&opt);
1103         return !hit;
1104 }