builtin/grep.c

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