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 "pathspec.h"
  21 #include "submodule.h"
  22 #include "submodule-config.h"
  23
  24 static char const * const grep_usage[] = {
  25         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  26         NULL
  27 };
  28
  29 static const char *super_prefix;
  30 static int recurse_submodules;
  31 static struct argv_array submodule_options = ARGV_ARRAY_INIT;
  32 static const char *parent_basename;
  33
  34 static int grep_submodule_launch(struct grep_opt *opt,
  35                                  const struct grep_source *gs);
  36
  37 #define GREP_NUM_THREADS_DEFAULT 8
  38 static int num_threads;
  39
  40 #ifndef NO_PTHREADS
  41 static pthread_t *threads;
  42
  43 /* We use one producer thread and THREADS consumer
  44  * threads. The producer adds struct work_items to 'todo' and the
  45  * consumers pick work items from the same array.
  46  */
  47 struct work_item {
  48         struct grep_source source;
  49         char done;
  50         struct strbuf out;
  51 };
  52
  53 /* In the range [todo_done, todo_start) in 'todo' we have work_items
  54  * that have been or are processed by a consumer thread. We haven't
  55  * written the result for these to stdout yet.
  56  *
  57  * The work_items in [todo_start, todo_end) are waiting to be picked
  58  * up by a consumer thread.
  59  *
  60  * The ranges are modulo TODO_SIZE.
  61  */
  62 #define TODO_SIZE 128
  63 static struct work_item todo[TODO_SIZE];
  64 static int todo_start;
  65 static int todo_end;
  66 static int todo_done;
  67
  68 /* Has all work items been added? */
  69 static int all_work_added;
  70
  71 /* This lock protects all the variables above. */
  72 static pthread_mutex_t grep_mutex;
  73
  74 static inline void grep_lock(void)
  75 {
  76         if (num_threads)
  77                 pthread_mutex_lock(&grep_mutex);
  78 }
  79
  80 static inline void grep_unlock(void)
  81 {
  82         if (num_threads)
  83                 pthread_mutex_unlock(&grep_mutex);
  84 }
  85
  86 /* Signalled when a new work_item is added to todo. */
  87 static pthread_cond_t cond_add;
  88
  89 /* Signalled when the result from one work_item is written to
  90  * stdout.
  91  */
  92 static pthread_cond_t cond_write;
  93
  94 /* Signalled when we are finished with everything. */
  95 static pthread_cond_t cond_result;
  96
  97 static int skip_first_line;
  98
  99 static void add_work(struct grep_opt *opt, enum grep_source_type type,
 100                      const char *name, const char *path, const void *id)
 101 {
 102         grep_lock();
 103
 104         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
 105                 pthread_cond_wait(&cond_write, &grep_mutex);
 106         }
 107
 108         grep_source_init(&todo[todo_end].source, type, name, path, id);
 109         if (opt->binary != GREP_BINARY_TEXT)
 110                 grep_source_load_driver(&todo[todo_end].source);
 111         todo[todo_end].done = 0;
 112         strbuf_reset(&todo[todo_end].out);
 113         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
 114
 115         pthread_cond_signal(&cond_add);
 116         grep_unlock();
 117 }
 118
 119 static struct work_item *get_work(void)
 120 {
 121         struct work_item *ret;
 122
 123         grep_lock();
 124         while (todo_start == todo_end && !all_work_added) {
 125                 pthread_cond_wait(&cond_add, &grep_mutex);
 126         }
 127
 128         if (todo_start == todo_end && all_work_added) {
 129                 ret = NULL;
 130         } else {
 131                 ret = &todo[todo_start];
 132                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
 133         }
 134         grep_unlock();
 135         return ret;
 136 }
 137
 138 static void work_done(struct work_item *w)
 139 {
 140         int old_done;
 141
 142         grep_lock();
 143         w->done = 1;
 144         old_done = todo_done;
 145         for(; todo[todo_done].done && todo_done != todo_start;
 146             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
 147                 w = &todo[todo_done];
 148                 if (w->out.len) {
 149                         const char *p = w->out.buf;
 150                         size_t len = w->out.len;
 151
 152                         /* Skip the leading hunk mark of the first file. */
 153                         if (skip_first_line) {
 154                                 while (len) {
 155                                         len--;
 156                                         if (*p++ == '\n')
 157                                                 break;
 158                                 }
 159                                 skip_first_line = 0;
 160                         }
 161
 162                         write_or_die(1, p, len);
 163                 }
 164                 grep_source_clear(&w->source);
 165         }
 166
 167         if (old_done != todo_done)
 168                 pthread_cond_signal(&cond_write);
 169
 170         if (all_work_added && todo_done == todo_end)
 171                 pthread_cond_signal(&cond_result);
 172
 173         grep_unlock();
 174 }
 175
 176 static void *run(void *arg)
 177 {
 178         int hit = 0;
 179         struct grep_opt *opt = arg;
 180
 181         while (1) {
 182                 struct work_item *w = get_work();
 183                 if (!w)
 184                         break;
 185
 186                 opt->output_priv = w;
 187                 if (w->source.type == GREP_SOURCE_SUBMODULE)
 188                         hit |= grep_submodule_launch(opt, &w->source);
 189                 else
 190                         hit |= grep_source(opt, &w->source);
 191                 grep_source_clear_data(&w->source);
 192                 work_done(w);
 193         }
 194         free_grep_patterns(arg);
 195         free(arg);
 196
 197         return (void*) (intptr_t) hit;
 198 }
 199
 200 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
 201 {
 202         struct work_item *w = opt->output_priv;
 203         strbuf_add(&w->out, buf, size);
 204 }
 205
 206 static void start_threads(struct grep_opt *opt)
 207 {
 208         int i;
 209
 210         pthread_mutex_init(&grep_mutex, NULL);
 211         pthread_mutex_init(&grep_read_mutex, NULL);
 212         pthread_mutex_init(&grep_attr_mutex, NULL);
 213         pthread_cond_init(&cond_add, NULL);
 214         pthread_cond_init(&cond_write, NULL);
 215         pthread_cond_init(&cond_result, NULL);
 216         grep_use_locks = 1;
 217
 218         for (i = 0; i < ARRAY_SIZE(todo); i++) {
 219                 strbuf_init(&todo[i].out, 0);
 220         }
 221
 222         threads = xcalloc(num_threads, sizeof(*threads));
 223         for (i = 0; i < num_threads; i++) {
 224                 int err;
 225                 struct grep_opt *o = grep_opt_dup(opt);
 226                 o->output = strbuf_out;
 227                 o->debug = 0;
 228                 compile_grep_patterns(o);
 229                 err = pthread_create(&threads[i], NULL, run, o);
 230
 231                 if (err)
 232                         die(_("grep: failed to create thread: %s"),
 233                             strerror(err));
 234         }
 235 }
 236
 237 static int wait_all(void)
 238 {
 239         int hit = 0;
 240         int i;
 241
 242         grep_lock();
 243         all_work_added = 1;
 244
 245         /* Wait until all work is done. */
 246         while (todo_done != todo_end)
 247                 pthread_cond_wait(&cond_result, &grep_mutex);
 248
 249         /* Wake up all the consumer threads so they can see that there
 250          * is no more work to do.
 251          */
 252         pthread_cond_broadcast(&cond_add);
 253         grep_unlock();
 254
 255         for (i = 0; i < num_threads; i++) {
 256                 void *h;
 257                 pthread_join(threads[i], &h);
 258                 hit |= (int) (intptr_t) h;
 259         }
 260
 261         free(threads);
 262
 263         pthread_mutex_destroy(&grep_mutex);
 264         pthread_mutex_destroy(&grep_read_mutex);
 265         pthread_mutex_destroy(&grep_attr_mutex);
 266         pthread_cond_destroy(&cond_add);
 267         pthread_cond_destroy(&cond_write);
 268         pthread_cond_destroy(&cond_result);
 269         grep_use_locks = 0;
 270
 271         return hit;
 272 }
 273 #else /* !NO_PTHREADS */
 274
 275 static int wait_all(void)
 276 {
 277         return 0;
 278 }
 279 #endif
 280
 281 static int grep_cmd_config(const char *var, const char *value, void *cb)
 282 {
 283         int st = grep_config(var, value, cb);
 284         if (git_color_default_config(var, value, cb) < 0)
 285                 st = -1;
 286
 287         if (!strcmp(var, "grep.threads")) {
 288                 num_threads = git_config_int(var, value);
 289                 if (num_threads < 0)
 290                         die(_("invalid number of threads specified (%d) for %s"),
 291                             num_threads, var);
 292         }
 293
 294         return st;
 295 }
 296
 297 static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 298 {
 299         void *data;
 300
 301         grep_read_lock();
 302         data = read_sha1_file(sha1, type, size);
 303         grep_read_unlock();
 304         return data;
 305 }
 306
 307 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 308                      const char *filename, int tree_name_len,
 309                      const char *path)
 310 {
 311         struct strbuf pathbuf = STRBUF_INIT;
 312
 313         if (opt->relative && opt->prefix_length) {
 314                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
 315                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 316         } else if (super_prefix) {
 317                 strbuf_add(&pathbuf, filename, tree_name_len);
 318                 strbuf_addstr(&pathbuf, super_prefix);
 319                 strbuf_addstr(&pathbuf, filename + tree_name_len);
 320         } else {
 321                 strbuf_addstr(&pathbuf, filename);
 322         }
 323
 324 #ifndef NO_PTHREADS
 325         if (num_threads) {
 326                 add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, path, sha1);
 327                 strbuf_release(&pathbuf);
 328                 return 0;
 329         } else
 330 #endif
 331         {
 332                 struct grep_source gs;
 333                 int hit;
 334
 335                 grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, path, sha1);
 336                 strbuf_release(&pathbuf);
 337                 hit = grep_source(opt, &gs);
 338
 339                 grep_source_clear(&gs);
 340                 return hit;
 341         }
 342 }
 343
 344 static int grep_file(struct grep_opt *opt, const char *filename)
 345 {
 346         struct strbuf buf = STRBUF_INIT;
 347
 348         if (opt->relative && opt->prefix_length) {
 349                 quote_path_relative(filename, opt->prefix, &buf);
 350         } else {
 351                 if (super_prefix)
 352                         strbuf_addstr(&buf, super_prefix);
 353                 strbuf_addstr(&buf, filename);
 354         }
 355
 356 #ifndef NO_PTHREADS
 357         if (num_threads) {
 358                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
 359                 strbuf_release(&buf);
 360                 return 0;
 361         } else
 362 #endif
 363         {
 364                 struct grep_source gs;
 365                 int hit;
 366
 367                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 368                 strbuf_release(&buf);
 369                 hit = grep_source(opt, &gs);
 370
 371                 grep_source_clear(&gs);
 372                 return hit;
 373         }
 374 }
 375
 376 static void append_path(struct grep_opt *opt, const void *data, size_t len)
 377 {
 378         struct string_list *path_list = opt->output_priv;
 379
 380         if (len == 1 && *(const char *)data == '\0')
 381                 return;
 382         string_list_append(path_list, xstrndup(data, len));
 383 }
 384
 385 static void run_pager(struct grep_opt *opt, const char *prefix)
 386 {
 387         struct string_list *path_list = opt->output_priv;
 388         struct child_process child = CHILD_PROCESS_INIT;
 389         int i, status;
 390
 391         for (i = 0; i < path_list->nr; i++)
 392                 argv_array_push(&child.args, path_list->items[i].string);
 393         child.dir = prefix;
 394         child.use_shell = 1;
 395
 396         status = run_command(&child);
 397         if (status)
 398                 exit(status);
 399 }
 400
 401 static void compile_submodule_options(const struct grep_opt *opt,
 402                                       const struct pathspec *pathspec,
 403                                       int cached, int untracked,
 404                                       int opt_exclude, int use_index,
 405                                       int pattern_type_arg)
 406 {
 407         struct grep_pat *pattern;
 408         int i;
 409
 410         if (recurse_submodules)
 411                 argv_array_push(&submodule_options, "--recurse-submodules");
 412
 413         if (cached)
 414                 argv_array_push(&submodule_options, "--cached");
 415         if (!use_index)
 416                 argv_array_push(&submodule_options, "--no-index");
 417         if (untracked)
 418                 argv_array_push(&submodule_options, "--untracked");
 419         if (opt_exclude > 0)
 420                 argv_array_push(&submodule_options, "--exclude-standard");
 421
 422         if (opt->invert)
 423                 argv_array_push(&submodule_options, "-v");
 424         if (opt->ignore_case)
 425                 argv_array_push(&submodule_options, "-i");
 426         if (opt->word_regexp)
 427                 argv_array_push(&submodule_options, "-w");
 428         switch (opt->binary) {
 429         case GREP_BINARY_NOMATCH:
 430                 argv_array_push(&submodule_options, "-I");
 431                 break;
 432         case GREP_BINARY_TEXT:
 433                 argv_array_push(&submodule_options, "-a");
 434                 break;
 435         default:
 436                 break;
 437         }
 438         if (opt->allow_textconv)
 439                 argv_array_push(&submodule_options, "--textconv");
 440         if (opt->max_depth != -1)
 441                 argv_array_pushf(&submodule_options, "--max-depth=%d",
 442                                  opt->max_depth);
 443         if (opt->linenum)
 444                 argv_array_push(&submodule_options, "-n");
 445         if (!opt->pathname)
 446                 argv_array_push(&submodule_options, "-h");
 447         if (!opt->relative)
 448                 argv_array_push(&submodule_options, "--full-name");
 449         if (opt->name_only)
 450                 argv_array_push(&submodule_options, "-l");
 451         if (opt->unmatch_name_only)
 452                 argv_array_push(&submodule_options, "-L");
 453         if (opt->null_following_name)
 454                 argv_array_push(&submodule_options, "-z");
 455         if (opt->count)
 456                 argv_array_push(&submodule_options, "-c");
 457         if (opt->file_break)
 458                 argv_array_push(&submodule_options, "--break");
 459         if (opt->heading)
 460                 argv_array_push(&submodule_options, "--heading");
 461         if (opt->pre_context)
 462                 argv_array_pushf(&submodule_options, "--before-context=%d",
 463                                  opt->pre_context);
 464         if (opt->post_context)
 465                 argv_array_pushf(&submodule_options, "--after-context=%d",
 466                                  opt->post_context);
 467         if (opt->funcname)
 468                 argv_array_push(&submodule_options, "-p");
 469         if (opt->funcbody)
 470                 argv_array_push(&submodule_options, "-W");
 471         if (opt->all_match)
 472                 argv_array_push(&submodule_options, "--all-match");
 473         if (opt->debug)
 474                 argv_array_push(&submodule_options, "--debug");
 475         if (opt->status_only)
 476                 argv_array_push(&submodule_options, "-q");
 477
 478         switch (pattern_type_arg) {
 479         case GREP_PATTERN_TYPE_BRE:
 480                 argv_array_push(&submodule_options, "-G");
 481                 break;
 482         case GREP_PATTERN_TYPE_ERE:
 483                 argv_array_push(&submodule_options, "-E");
 484                 break;
 485         case GREP_PATTERN_TYPE_FIXED:
 486                 argv_array_push(&submodule_options, "-F");
 487                 break;
 488         case GREP_PATTERN_TYPE_PCRE:
 489                 argv_array_push(&submodule_options, "-P");
 490                 break;
 491         case GREP_PATTERN_TYPE_UNSPECIFIED:
 492                 break;
 493         }
 494
 495         for (pattern = opt->pattern_list; pattern != NULL;
 496              pattern = pattern->next) {
 497                 switch (pattern->token) {
 498                 case GREP_PATTERN:
 499                         argv_array_pushf(&submodule_options, "-e%s",
 500                                          pattern->pattern);
 501                         break;
 502                 case GREP_AND:
 503                 case GREP_OPEN_PAREN:
 504                 case GREP_CLOSE_PAREN:
 505                 case GREP_NOT:
 506                 case GREP_OR:
 507                         argv_array_push(&submodule_options, pattern->pattern);
 508                         break;
 509                 /* BODY and HEAD are not used by git-grep */
 510                 case GREP_PATTERN_BODY:
 511                 case GREP_PATTERN_HEAD:
 512                         break;
 513                 }
 514         }
 515
 516         /*
 517          * Limit number of threads for child process to use.
 518          * This is to prevent potential fork-bomb behavior of git-grep as each
 519          * submodule process has its own thread pool.
 520          */
 521         argv_array_pushf(&submodule_options, "--threads=%d",
 522                          (num_threads + 1) / 2);
 523
 524         /* Add Pathspecs */
 525         argv_array_push(&submodule_options, "--");
 526         for (i = 0; i < pathspec->nr; i++)
 527                 argv_array_push(&submodule_options,
 528                                 pathspec->items[i].original);
 529 }
 530
 531 /*
 532  * Launch child process to grep contents of a submodule
 533  */
 534 static int grep_submodule_launch(struct grep_opt *opt,
 535                                  const struct grep_source *gs)
 536 {
 537         struct child_process cp = CHILD_PROCESS_INIT;
 538         int status, i;
 539         const char *end_of_base;
 540         const char *name;
 541         struct work_item *w = opt->output_priv;
 542
 543         end_of_base = strchr(gs->name, ':');
 544         if (gs->identifier && end_of_base)
 545                 name = end_of_base + 1;
 546         else
 547                 name = gs->name;
 548
 549         prepare_submodule_repo_env(&cp.env_array);
 550         argv_array_push(&cp.env_array, GIT_DIR_ENVIRONMENT);
 551
 552         /* Add super prefix */
 553         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
 554                          super_prefix ? super_prefix : "",
 555                          name);
 556         argv_array_push(&cp.args, "grep");
 557
 558         /*
 559          * Add basename of parent project
 560          * When performing grep on a tree object the filename is prefixed
 561          * with the object's name: 'tree-name:filename'.  In order to
 562          * provide uniformity of output we want to pass the name of the
 563          * parent project's object name to the submodule so the submodule can
 564          * prefix its output with the parent's name and not its own SHA1.
 565          */
 566         if (gs->identifier && end_of_base)
 567                 argv_array_pushf(&cp.args, "--parent-basename=%.*s",
 568                                  (int) (end_of_base - gs->name),
 569                                  gs->name);
 570
 571         /* Add options */
 572         for (i = 0; i < submodule_options.argc; i++) {
 573                 /*
 574                  * If there is a tree identifier for the submodule, add the
 575                  * rev after adding the submodule options but before the
 576                  * pathspecs.  To do this we listen for the '--' and insert the
 577                  * sha1 before pushing the '--' onto the child process argv
 578                  * array.
 579                  */
 580                 if (gs->identifier &&
 581                     !strcmp("--", submodule_options.argv[i])) {
 582                         argv_array_push(&cp.args, sha1_to_hex(gs->identifier));
 583                 }
 584
 585                 argv_array_push(&cp.args, submodule_options.argv[i]);
 586         }
 587
 588         cp.git_cmd = 1;
 589         cp.dir = gs->path;
 590
 591         /*
 592          * Capture output to output buffer and check the return code from the
 593          * child process.  A '0' indicates a hit, a '1' indicates no hit and
 594          * anything else is an error.
 595          */
 596         status = capture_command(&cp, &w->out, 0);
 597         if (status && (status != 1)) {
 598                 /* flush the buffer */
 599                 write_or_die(1, w->out.buf, w->out.len);
 600                 die("process for submodule '%s' failed with exit code: %d",
 601                     gs->name, status);
 602         }
 603
 604         /* invert the return code to make a hit equal to 1 */
 605         return !status;
 606 }
 607
 608 /*
 609  * Prep grep structures for a submodule grep
 610  * sha1: the sha1 of the submodule or NULL if using the working tree
 611  * filename: name of the submodule including tree name of parent
 612  * path: location of the submodule
 613  */
 614 static int grep_submodule(struct grep_opt *opt, const unsigned char *sha1,
 615                           const char *filename, const char *path)
 616 {
 617         if (!is_submodule_initialized(path))
 618                 return 0;
 619         if (!is_submodule_populated(path)) {
 620                 /*
 621                  * If searching history, check for the presense of the
 622                  * submodule's gitdir before skipping the submodule.
 623                  */
 624                 if (sha1) {
 625                         const struct submodule *sub =
 626                                         submodule_from_path(null_sha1, path);
 627                         if (sub)
 628                                 path = git_path("modules/%s", sub->name);
 629
 630                         if (!(is_directory(path) && is_git_directory(path)))
 631                                 return 0;
 632                 } else {
 633                         return 0;
 634                 }
 635         }
 636
 637 #ifndef NO_PTHREADS
 638         if (num_threads) {
 639                 add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, sha1);
 640                 return 0;
 641         } else
 642 #endif
 643         {
 644                 struct work_item w;
 645                 int hit;
 646
 647                 grep_source_init(&w.source, GREP_SOURCE_SUBMODULE,
 648                                  filename, path, sha1);
 649                 strbuf_init(&w.out, 0);
 650                 opt->output_priv = &w;
 651                 hit = grep_submodule_launch(opt, &w.source);
 652
 653                 write_or_die(1, w.out.buf, w.out.len);
 654
 655                 grep_source_clear(&w.source);
 656                 strbuf_release(&w.out);
 657                 return hit;
 658         }
 659 }
 660
 661 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
 662                       int cached)
 663 {
 664         int hit = 0;
 665         int nr;
 666         struct strbuf name = STRBUF_INIT;
 667         int name_base_len = 0;
 668         if (super_prefix) {
 669                 name_base_len = strlen(super_prefix);
 670                 strbuf_addstr(&name, super_prefix);
 671         }
 672
 673         read_cache();
 674
 675         for (nr = 0; nr < active_nr; nr++) {
 676                 const struct cache_entry *ce = active_cache[nr];
 677                 strbuf_setlen(&name, name_base_len);
 678                 strbuf_addstr(&name, ce->name);
 679
 680                 if (S_ISREG(ce->ce_mode) &&
 681                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
 682                                    S_ISDIR(ce->ce_mode) ||
 683                                    S_ISGITLINK(ce->ce_mode))) {
 684                         /*
 685                          * If CE_VALID is on, we assume worktree file and its
 686                          * cache entry are identical, even if worktree file has
 687                          * been modified, so use cache version instead
 688                          */
 689                         if (cached || (ce->ce_flags & CE_VALID) ||
 690                             ce_skip_worktree(ce)) {
 691                                 if (ce_stage(ce) || ce_intent_to_add(ce))
 692                                         continue;
 693                                 hit |= grep_sha1(opt, ce->oid.hash, ce->name,
 694                                                  0, ce->name);
 695                         } else {
 696                                 hit |= grep_file(opt, ce->name);
 697                         }
 698                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 699                            submodule_path_match(pathspec, name.buf, NULL)) {
 700                         hit |= grep_submodule(opt, NULL, ce->name, ce->name);
 701                 } else {
 702                         continue;
 703                 }
 704
 705                 if (ce_stage(ce)) {
 706                         do {
 707                                 nr++;
 708                         } while (nr < active_nr &&
 709                                  !strcmp(ce->name, active_cache[nr]->name));
 710                         nr--; /* compensate for loop control */
 711                 }
 712                 if (hit && opt->status_only)
 713                         break;
 714         }
 715
 716         strbuf_release(&name);
 717         return hit;
 718 }
 719
 720 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 721                      struct tree_desc *tree, struct strbuf *base, int tn_len,
 722                      int check_attr)
 723 {
 724         int hit = 0;
 725         enum interesting match = entry_not_interesting;
 726         struct name_entry entry;
 727         int old_baselen = base->len;
 728         struct strbuf name = STRBUF_INIT;
 729         int name_base_len = 0;
 730         if (super_prefix) {
 731                 strbuf_addstr(&name, super_prefix);
 732                 name_base_len = name.len;
 733         }
 734
 735         while (tree_entry(tree, &entry)) {
 736                 int te_len = tree_entry_len(&entry);
 737
 738                 if (match != all_entries_interesting) {
 739                         strbuf_addstr(&name, base->buf + tn_len);
 740                         match = tree_entry_interesting(&entry, &name,
 741                                                        0, pathspec);
 742                         strbuf_setlen(&name, name_base_len);
 743
 744                         if (match == all_entries_not_interesting)
 745                                 break;
 746                         if (match == entry_not_interesting)
 747                                 continue;
 748                 }
 749
 750                 strbuf_add(base, entry.path, te_len);
 751
 752                 if (S_ISREG(entry.mode)) {
 753                         hit |= grep_sha1(opt, entry.oid->hash, base->buf, tn_len,
 754                                          check_attr ? base->buf + tn_len : NULL);
 755                 } else if (S_ISDIR(entry.mode)) {
 756                         enum object_type type;
 757                         struct tree_desc sub;
 758                         void *data;
 759                         unsigned long size;
 760
 761                         data = lock_and_read_sha1_file(entry.oid->hash, &type, &size);
 762                         if (!data)
 763                                 die(_("unable to read tree (%s)"),
 764                                     oid_to_hex(entry.oid));
 765
 766                         strbuf_addch(base, '/');
 767                         init_tree_desc(&sub, data, size);
 768                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 769                                          check_attr);
 770                         free(data);
 771                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 772                         hit |= grep_submodule(opt, entry.oid->hash, base->buf,
 773                                               base->buf + tn_len);
 774                 }
 775
 776                 strbuf_setlen(base, old_baselen);
 777
 778                 if (hit && opt->status_only)
 779                         break;
 780         }
 781
 782         strbuf_release(&name);
 783         return hit;
 784 }
 785
 786 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 787                        struct object *obj, const char *name, const char *path)
 788 {
 789         if (obj->type == OBJ_BLOB)
 790                 return grep_sha1(opt, obj->oid.hash, name, 0, path);
 791         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 792                 struct tree_desc tree;
 793                 void *data;
 794                 unsigned long size;
 795                 struct strbuf base;
 796                 int hit, len;
 797
 798                 grep_read_lock();
 799                 data = read_object_with_reference(obj->oid.hash, tree_type,
 800                                                   &size, NULL);
 801                 grep_read_unlock();
 802
 803                 if (!data)
 804                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 805
 806                 /* Use parent's name as base when recursing submodules */
 807                 if (recurse_submodules && parent_basename)
 808                         name = parent_basename;
 809
 810                 len = name ? strlen(name) : 0;
 811                 strbuf_init(&base, PATH_MAX + len + 1);
 812                 if (len) {
 813                         strbuf_add(&base, name, len);
 814                         strbuf_addch(&base, ':');
 815                 }
 816                 init_tree_desc(&tree, data, size);
 817                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 818                                 obj->type == OBJ_COMMIT);
 819                 strbuf_release(&base);
 820                 free(data);
 821                 return hit;
 822         }
 823         die(_("unable to grep from object of type %s"), typename(obj->type));
 824 }
 825
 826 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 827                         const struct object_array *list)
 828 {
 829         unsigned int i;
 830         int hit = 0;
 831         const unsigned int nr = list->nr;
 832
 833         for (i = 0; i < nr; i++) {
 834                 struct object *real_obj;
 835                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
 836
 837                 /* load the gitmodules file for this rev */
 838                 if (recurse_submodules) {
 839                         submodule_free();
 840                         gitmodules_config_sha1(real_obj->oid.hash);
 841                 }
 842                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
 843                         hit = 1;
 844                         if (opt->status_only)
 845                                 break;
 846                 }
 847         }
 848         return hit;
 849 }
 850
 851 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 852                           int exc_std, int use_index)
 853 {
 854         struct dir_struct dir;
 855         int i, hit = 0;
 856
 857         memset(&dir, 0, sizeof(dir));
 858         if (!use_index)
 859                 dir.flags |= DIR_NO_GITLINKS;
 860         if (exc_std)
 861                 setup_standard_excludes(&dir);
 862
 863         fill_directory(&dir, pathspec);
 864         for (i = 0; i < dir.nr; i++) {
 865                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
 866                         continue;
 867                 hit |= grep_file(opt, dir.entries[i]->name);
 868                 if (hit && opt->status_only)
 869                         break;
 870         }
 871         return hit;
 872 }
 873
 874 static int context_callback(const struct option *opt, const char *arg,
 875                             int unset)
 876 {
 877         struct grep_opt *grep_opt = opt->value;
 878         int value;
 879         const char *endp;
 880
 881         if (unset) {
 882                 grep_opt->pre_context = grep_opt->post_context = 0;
 883                 return 0;
 884         }
 885         value = strtol(arg, (char **)&endp, 10);
 886         if (*endp) {
 887                 return error(_("switch `%c' expects a numerical value"),
 888                              opt->short_name);
 889         }
 890         grep_opt->pre_context = grep_opt->post_context = value;
 891         return 0;
 892 }
 893
 894 static int file_callback(const struct option *opt, const char *arg, int unset)
 895 {
 896         struct grep_opt *grep_opt = opt->value;
 897         int from_stdin = !strcmp(arg, "-");
 898         FILE *patterns;
 899         int lno = 0;
 900         struct strbuf sb = STRBUF_INIT;
 901
 902         patterns = from_stdin ? stdin : fopen(arg, "r");
 903         if (!patterns)
 904                 die_errno(_("cannot open '%s'"), arg);
 905         while (strbuf_getline(&sb, patterns) == 0) {
 906                 /* ignore empty line like grep does */
 907                 if (sb.len == 0)
 908                         continue;
 909
 910                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 911                                 GREP_PATTERN);
 912         }
 913         if (!from_stdin)
 914                 fclose(patterns);
 915         strbuf_release(&sb);
 916         return 0;
 917 }
 918
 919 static int not_callback(const struct option *opt, const char *arg, int unset)
 920 {
 921         struct grep_opt *grep_opt = opt->value;
 922         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 923         return 0;
 924 }
 925
 926 static int and_callback(const struct option *opt, const char *arg, int unset)
 927 {
 928         struct grep_opt *grep_opt = opt->value;
 929         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 930         return 0;
 931 }
 932
 933 static int open_callback(const struct option *opt, const char *arg, int unset)
 934 {
 935         struct grep_opt *grep_opt = opt->value;
 936         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 937         return 0;
 938 }
 939
 940 static int close_callback(const struct option *opt, const char *arg, int unset)
 941 {
 942         struct grep_opt *grep_opt = opt->value;
 943         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 944         return 0;
 945 }
 946
 947 static int pattern_callback(const struct option *opt, const char *arg,
 948                             int unset)
 949 {
 950         struct grep_opt *grep_opt = opt->value;
 951         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 952         return 0;
 953 }
 954
 955 int cmd_grep(int argc, const char **argv, const char *prefix)
 956 {
 957         int hit = 0;
 958         int cached = 0, untracked = 0, opt_exclude = -1;
 959         int seen_dashdash = 0;
 960         int external_grep_allowed__ignored;
 961         const char *show_in_pager = NULL, *default_pager = "dummy";
 962         struct grep_opt opt;
 963         struct object_array list = OBJECT_ARRAY_INIT;
 964         struct pathspec pathspec;
 965         struct string_list path_list = STRING_LIST_INIT_NODUP;
 966         int i;
 967         int dummy;
 968         int use_index = 1;
 969         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 970         int allow_revs;
 971
 972         struct option options[] = {
 973                 OPT_BOOL(0, "cached", &cached,
 974                         N_("search in index instead of in the work tree")),
 975                 OPT_NEGBIT(0, "no-index", &use_index,
 976                          N_("find in contents not managed by git"), 1),
 977                 OPT_BOOL(0, "untracked", &untracked,
 978                         N_("search in both tracked and untracked files")),
 979                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 980                             N_("ignore files specified via '.gitignore'"), 1),
 981                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 982                          N_("recursivley search in each submodule")),
 983                 OPT_STRING(0, "parent-basename", &parent_basename,
 984                            N_("basename"),
 985                            N_("prepend parent project's basename to output")),
 986                 OPT_GROUP(""),
 987                 OPT_BOOL('v', "invert-match", &opt.invert,
 988                         N_("show non-matching lines")),
 989                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 990                         N_("case insensitive matching")),
 991                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 992                         N_("match patterns only at word boundaries")),
 993                 OPT_SET_INT('a', "text", &opt.binary,
 994                         N_("process binary files as text"), GREP_BINARY_TEXT),
 995                 OPT_SET_INT('I', NULL, &opt.binary,
 996                         N_("don't match patterns in binary files"),
 997                         GREP_BINARY_NOMATCH),
 998                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
 999                          N_("process binary files with textconv filters")),
1000                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
1001                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
1002                         NULL, 1 },
1003                 OPT_GROUP(""),
1004                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
1005                             N_("use extended POSIX regular expressions"),
1006                             GREP_PATTERN_TYPE_ERE),
1007                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
1008                             N_("use basic POSIX regular expressions (default)"),
1009                             GREP_PATTERN_TYPE_BRE),
1010                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
1011                             N_("interpret patterns as fixed strings"),
1012                             GREP_PATTERN_TYPE_FIXED),
1013                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
1014                             N_("use Perl-compatible regular expressions"),
1015                             GREP_PATTERN_TYPE_PCRE),
1016                 OPT_GROUP(""),
1017                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1018                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1019                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1020                 OPT_NEGBIT(0, "full-name", &opt.relative,
1021                         N_("show filenames relative to top directory"), 1),
1022                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
1023                         N_("show only filenames instead of matching lines")),
1024                 OPT_BOOL(0, "name-only", &opt.name_only,
1025                         N_("synonym for --files-with-matches")),
1026                 OPT_BOOL('L', "files-without-match",
1027                         &opt.unmatch_name_only,
1028                         N_("show only the names of files without match")),
1029                 OPT_BOOL('z', "null", &opt.null_following_name,
1030                         N_("print NUL after filenames")),
1031                 OPT_BOOL('c', "count", &opt.count,
1032                         N_("show the number of matches instead of matching lines")),
1033                 OPT__COLOR(&opt.color, N_("highlight matches")),
1034                 OPT_BOOL(0, "break", &opt.file_break,
1035                         N_("print empty line between matches from different files")),
1036                 OPT_BOOL(0, "heading", &opt.heading,
1037                         N_("show filename only once above matches from same file")),
1038                 OPT_GROUP(""),
1039                 OPT_CALLBACK('C', "context", &opt, N_("n"),
1040                         N_("show <n> context lines before and after matches"),
1041                         context_callback),
1042                 OPT_INTEGER('B', "before-context", &opt.pre_context,
1043                         N_("show <n> context lines before matches")),
1044                 OPT_INTEGER('A', "after-context", &opt.post_context,
1045                         N_("show <n> context lines after matches")),
1046                 OPT_INTEGER(0, "threads", &num_threads,
1047                         N_("use <n> worker threads")),
1048                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1049                         context_callback),
1050                 OPT_BOOL('p', "show-function", &opt.funcname,
1051                         N_("show a line with the function name before matches")),
1052                 OPT_BOOL('W', "function-context", &opt.funcbody,
1053                         N_("show the surrounding function")),
1054                 OPT_GROUP(""),
1055                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
1056                         N_("read patterns from file"), file_callback),
1057                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1058                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1059                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
1060                   N_("combine patterns specified with -e"),
1061                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1062                 OPT_BOOL(0, "or", &dummy, ""),
1063                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1064                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1065                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1066                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1067                   open_callback },
1068                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1069                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1070                   close_callback },
1071                 OPT__QUIET(&opt.status_only,
1072                            N_("indicate hit with exit status without output")),
1073                 OPT_BOOL(0, "all-match", &opt.all_match,
1074                         N_("show only matches from files that match all patterns")),
1075                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1076                   N_("show parse tree for grep expression"),
1077                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1078                 OPT_GROUP(""),
1079                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1080                         N_("pager"), N_("show matching files in the pager"),
1081                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1082                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1083                          N_("allow calling of grep(1) (ignored by this build)")),
1084                 OPT_END()
1085         };
1086
1087         init_grep_defaults();
1088         git_config(grep_cmd_config, NULL);
1089         grep_init(&opt, prefix);
1090         super_prefix = get_super_prefix();
1091
1092         /*
1093          * If there is no -- then the paths must exist in the working
1094          * tree.  If there is no explicit pattern specified with -e or
1095          * -f, we take the first unrecognized non option to be the
1096          * pattern, but then what follows it must be zero or more
1097          * valid refs up to the -- (if exists), and then existing
1098          * paths.  If there is an explicit pattern, then the first
1099          * unrecognized non option is the beginning of the refs list
1100          * that continues up to the -- (if exists), and then paths.
1101          */
1102         argc = parse_options(argc, argv, prefix, options, grep_usage,
1103                              PARSE_OPT_KEEP_DASHDASH |
1104                              PARSE_OPT_STOP_AT_NON_OPTION);
1105         grep_commit_pattern_type(pattern_type_arg, &opt);
1106
1107         if (use_index && !startup_info->have_repository) {
1108                 int fallback = 0;
1109                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
1110                 if (fallback)
1111                         use_index = 0;
1112                 else
1113                         /* die the same way as if we did it at the beginning */
1114                         setup_git_directory();
1115         }
1116
1117         /*
1118          * skip a -- separator; we know it cannot be
1119          * separating revisions from pathnames if
1120          * we haven't even had any patterns yet
1121          */
1122         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1123                 argv++;
1124                 argc--;
1125         }
1126
1127         /* First unrecognized non-option token */
1128         if (argc > 0 && !opt.pattern_list) {
1129                 append_grep_pattern(&opt, argv[0], "command line", 0,
1130                                     GREP_PATTERN);
1131                 argv++;
1132                 argc--;
1133         }
1134
1135         if (show_in_pager == default_pager)
1136                 show_in_pager = git_pager(1);
1137         if (show_in_pager) {
1138                 opt.color = 0;
1139                 opt.name_only = 1;
1140                 opt.null_following_name = 1;
1141                 opt.output_priv = &path_list;
1142                 opt.output = append_path;
1143                 string_list_append(&path_list, show_in_pager);
1144         }
1145
1146         if (!opt.pattern_list)
1147                 die(_("no pattern given."));
1148         if (!opt.fixed && opt.ignore_case)
1149                 opt.regflags |= REG_ICASE;
1150
1151         compile_grep_patterns(&opt);
1152
1153         /*
1154          * We have to find "--" in a separate pass, because its presence
1155          * influences how we will parse arguments that come before it.
1156          */
1157         for (i = 0; i < argc; i++) {
1158                 if (!strcmp(argv[i], "--")) {
1159                         seen_dashdash = 1;
1160                         break;
1161                 }
1162         }
1163
1164         /*
1165          * Resolve any rev arguments. If we have a dashdash, then everything up
1166          * to it must resolve as a rev. If not, then we stop at the first
1167          * non-rev and assume everything else is a path.
1168          */
1169         allow_revs = use_index && !untracked;
1170         for (i = 0; i < argc; i++) {
1171                 const char *arg = argv[i];
1172                 unsigned char sha1[20];
1173                 struct object_context oc;
1174                 struct object *object;
1175
1176                 if (!strcmp(arg, "--")) {
1177                         i++;
1178                         break;
1179                 }
1180
1181                 if (!allow_revs) {
1182                         if (seen_dashdash)
1183                                 die(_("--no-index or --untracked cannot be used with revs"));
1184                         break;
1185                 }
1186
1187                 if (get_sha1_with_context(arg, 0, sha1, &oc)) {
1188                         if (seen_dashdash)
1189                                 die(_("unable to resolve revision: %s"), arg);
1190                         break;
1191                 }
1192
1193                 object = parse_object_or_die(sha1, arg);
1194                 if (!seen_dashdash)
1195                         verify_non_filename(prefix, arg);
1196                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1197         }
1198
1199         /*
1200          * Anything left over is presumed to be a path. But in the non-dashdash
1201          * "do what I mean" case, we verify and complain when that isn't true.
1202          */
1203         if (!seen_dashdash) {
1204                 int j;
1205                 for (j = i; j < argc; j++)
1206                         verify_filename(prefix, argv[j], j == i && allow_revs);
1207         }
1208
1209         parse_pathspec(&pathspec, 0,
1210                        PATHSPEC_PREFER_CWD |
1211                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1212                        prefix, argv + i);
1213         pathspec.max_depth = opt.max_depth;
1214         pathspec.recursive = 1;
1215
1216 #ifndef NO_PTHREADS
1217         if (list.nr || cached || show_in_pager)
1218                 num_threads = 0;
1219         else if (num_threads == 0)
1220                 num_threads = GREP_NUM_THREADS_DEFAULT;
1221         else if (num_threads < 0)
1222                 die(_("invalid number of threads specified (%d)"), num_threads);
1223 #else
1224         num_threads = 0;
1225 #endif
1226
1227 #ifndef NO_PTHREADS
1228         if (num_threads) {
1229                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1230                     && (opt.pre_context || opt.post_context ||
1231                         opt.file_break || opt.funcbody))
1232                         skip_first_line = 1;
1233                 start_threads(&opt);
1234         }
1235 #endif
1236
1237         if (recurse_submodules) {
1238                 gitmodules_config();
1239                 compile_submodule_options(&opt, &pathspec, cached, untracked,
1240                                           opt_exclude, use_index,
1241                                           pattern_type_arg);
1242         }
1243
1244         if (show_in_pager && (cached || list.nr))
1245                 die(_("--open-files-in-pager only works on the worktree"));
1246
1247         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1248                 const char *pager = path_list.items[0].string;
1249                 int len = strlen(pager);
1250
1251                 if (len > 4 && is_dir_sep(pager[len - 5]))
1252                         pager += len - 4;
1253
1254                 if (opt.ignore_case && !strcmp("less", pager))
1255                         string_list_append(&path_list, "-I");
1256
1257                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1258                         struct strbuf buf = STRBUF_INIT;
1259                         strbuf_addf(&buf, "+/%s%s",
1260                                         strcmp("less", pager) ? "" : "*",
1261                                         opt.pattern_list->pattern);
1262                         string_list_append(&path_list, buf.buf);
1263                         strbuf_detach(&buf, NULL);
1264                 }
1265         }
1266
1267         if (recurse_submodules && (!use_index || untracked))
1268                 die(_("option not supported with --recurse-submodules."));
1269
1270         if (!show_in_pager && !opt.status_only)
1271                 setup_pager();
1272
1273         if (!use_index && (untracked || cached))
1274                 die(_("--cached or --untracked cannot be used with --no-index."));
1275
1276         if (!use_index || untracked) {
1277                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1278                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1279         } else if (0 <= opt_exclude) {
1280                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1281         } else if (!list.nr) {
1282                 if (!cached)
1283                         setup_work_tree();
1284
1285                 hit = grep_cache(&opt, &pathspec, cached);
1286         } else {
1287                 if (cached)
1288                         die(_("both --cached and trees are given."));
1289                 hit = grep_objects(&opt, &pathspec, &list);
1290         }
1291
1292         if (num_threads)
1293                 hit |= wait_all();
1294         if (hit && show_in_pager)
1295                 run_pager(&opt, prefix);
1296         free_grep_patterns(&opt);
1297         return !hit;
1298 }