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