Andrew's git / gitweb.git / builtin/grep.c
?
builtin / grep.con commit sha1_file: refactor has_sha1_file_with_flags (e83e71c)
   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
  24static char const * const grep_usage[] = {
  25        N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
  26        NULL
  27};
  28
  29static const char *super_prefix;
  30static int recurse_submodules;
  31static struct argv_array submodule_options = ARGV_ARRAY_INIT;
  32static const char *parent_basename;
  33
  34static int grep_submodule_launch(struct grep_opt *opt,
  35                                 const struct grep_source *gs);
  36
  37#define GREP_NUM_THREADS_DEFAULT 8
  38static int num_threads;
  39
  40#ifndef NO_PTHREADS
  41static 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 */
  47struct 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
  63static struct work_item todo[TODO_SIZE];
  64static int todo_start;
  65static int todo_end;
  66static int todo_done;
  67
  68/* Has all work items been added? */
  69static int all_work_added;
  70
  71/* This lock protects all the variables above. */
  72static pthread_mutex_t grep_mutex;
  73
  74static inline void grep_lock(void)
  75{
  76        assert(num_threads);
  77        pthread_mutex_lock(&grep_mutex);
  78}
  79
  80static inline void grep_unlock(void)
  81{
  82        assert(num_threads);
  83        pthread_mutex_unlock(&grep_mutex);
  84}
  85
  86/* Signalled when a new work_item is added to todo. */
  87static pthread_cond_t cond_add;
  88
  89/* Signalled when the result from one work_item is written to
  90 * stdout.
  91 */
  92static pthread_cond_t cond_write;
  93
  94/* Signalled when we are finished with everything. */
  95static pthread_cond_t cond_result;
  96
  97static int skip_first_line;
  98
  99static 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
 119static 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
 138static 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
 176static 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
 200static 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
 206static 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
 237static 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
 275static int wait_all(void)
 276{
 277        return 0;
 278}
 279#endif
 280
 281static 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#ifdef NO_PTHREADS
 293                else if (num_threads && num_threads != 1) {
 294                        /*
 295                         * TRANSLATORS: %s is the configuration
 296                         * variable for tweaking threads, currently
 297                         * grep.threads
 298                         */
 299                        warning(_("no threads support, ignoring %s"), var);
 300                        num_threads = 0;
 301                }
 302#endif
 303        }
 304
 305        if (!strcmp(var, "submodule.recurse"))
 306                recurse_submodules = git_config_bool(var, value);
 307
 308        return st;
 309}
 310
 311static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
 312{
 313        void *data;
 314
 315        grep_read_lock();
 316        data = read_sha1_file(oid->hash, type, size);
 317        grep_read_unlock();
 318        return data;
 319}
 320
 321static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
 322                     const char *filename, int tree_name_len,
 323                     const char *path)
 324{
 325        struct strbuf pathbuf = STRBUF_INIT;
 326
 327        if (super_prefix) {
 328                strbuf_add(&pathbuf, filename, tree_name_len);
 329                strbuf_addstr(&pathbuf, super_prefix);
 330                strbuf_addstr(&pathbuf, filename + tree_name_len);
 331        } else {
 332                strbuf_addstr(&pathbuf, filename);
 333        }
 334
 335        if (opt->relative && opt->prefix_length) {
 336                char *name = strbuf_detach(&pathbuf, NULL);
 337                quote_path_relative(name + tree_name_len, opt->prefix, &pathbuf);
 338                strbuf_insert(&pathbuf, 0, name, tree_name_len);
 339                free(name);
 340        }
 341
 342#ifndef NO_PTHREADS
 343        if (num_threads) {
 344                add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
 345                strbuf_release(&pathbuf);
 346                return 0;
 347        } else
 348#endif
 349        {
 350                struct grep_source gs;
 351                int hit;
 352
 353                grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
 354                strbuf_release(&pathbuf);
 355                hit = grep_source(opt, &gs);
 356
 357                grep_source_clear(&gs);
 358                return hit;
 359        }
 360}
 361
 362static int grep_file(struct grep_opt *opt, const char *filename)
 363{
 364        struct strbuf buf = STRBUF_INIT;
 365
 366        if (super_prefix)
 367                strbuf_addstr(&buf, super_prefix);
 368        strbuf_addstr(&buf, filename);
 369
 370        if (opt->relative && opt->prefix_length) {
 371                char *name = strbuf_detach(&buf, NULL);
 372                quote_path_relative(name, opt->prefix, &buf);
 373                free(name);
 374        }
 375
 376#ifndef NO_PTHREADS
 377        if (num_threads) {
 378                add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
 379                strbuf_release(&buf);
 380                return 0;
 381        } else
 382#endif
 383        {
 384                struct grep_source gs;
 385                int hit;
 386
 387                grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 388                strbuf_release(&buf);
 389                hit = grep_source(opt, &gs);
 390
 391                grep_source_clear(&gs);
 392                return hit;
 393        }
 394}
 395
 396static void append_path(struct grep_opt *opt, const void *data, size_t len)
 397{
 398        struct string_list *path_list = opt->output_priv;
 399
 400        if (len == 1 && *(const char *)data == '\0')
 401                return;
 402        string_list_append(path_list, xstrndup(data, len));
 403}
 404
 405static void run_pager(struct grep_opt *opt, const char *prefix)
 406{
 407        struct string_list *path_list = opt->output_priv;
 408        struct child_process child = CHILD_PROCESS_INIT;
 409        int i, status;
 410
 411        for (i = 0; i < path_list->nr; i++)
 412                argv_array_push(&child.args, path_list->items[i].string);
 413        child.dir = prefix;
 414        child.use_shell = 1;
 415
 416        status = run_command(&child);
 417        if (status)
 418                exit(status);
 419}
 420
 421static void compile_submodule_options(const struct grep_opt *opt,
 422                                      const char **argv,
 423                                      int cached, int untracked,
 424                                      int opt_exclude, int use_index,
 425                                      int pattern_type_arg)
 426{
 427        struct grep_pat *pattern;
 428
 429        if (recurse_submodules)
 430                argv_array_push(&submodule_options, "--recurse-submodules");
 431
 432        if (cached)
 433                argv_array_push(&submodule_options, "--cached");
 434        if (!use_index)
 435                argv_array_push(&submodule_options, "--no-index");
 436        if (untracked)
 437                argv_array_push(&submodule_options, "--untracked");
 438        if (opt_exclude > 0)
 439                argv_array_push(&submodule_options, "--exclude-standard");
 440
 441        if (opt->invert)
 442                argv_array_push(&submodule_options, "-v");
 443        if (opt->ignore_case)
 444                argv_array_push(&submodule_options, "-i");
 445        if (opt->word_regexp)
 446                argv_array_push(&submodule_options, "-w");
 447        switch (opt->binary) {
 448        case GREP_BINARY_NOMATCH:
 449                argv_array_push(&submodule_options, "-I");
 450                break;
 451        case GREP_BINARY_TEXT:
 452                argv_array_push(&submodule_options, "-a");
 453                break;
 454        default:
 455                break;
 456        }
 457        if (opt->allow_textconv)
 458                argv_array_push(&submodule_options, "--textconv");
 459        if (opt->max_depth != -1)
 460                argv_array_pushf(&submodule_options, "--max-depth=%d",
 461                                 opt->max_depth);
 462        if (opt->linenum)
 463                argv_array_push(&submodule_options, "-n");
 464        if (!opt->pathname)
 465                argv_array_push(&submodule_options, "-h");
 466        if (!opt->relative)
 467                argv_array_push(&submodule_options, "--full-name");
 468        if (opt->name_only)
 469                argv_array_push(&submodule_options, "-l");
 470        if (opt->unmatch_name_only)
 471                argv_array_push(&submodule_options, "-L");
 472        if (opt->null_following_name)
 473                argv_array_push(&submodule_options, "-z");
 474        if (opt->count)
 475                argv_array_push(&submodule_options, "-c");
 476        if (opt->file_break)
 477                argv_array_push(&submodule_options, "--break");
 478        if (opt->heading)
 479                argv_array_push(&submodule_options, "--heading");
 480        if (opt->pre_context)
 481                argv_array_pushf(&submodule_options, "--before-context=%d",
 482                                 opt->pre_context);
 483        if (opt->post_context)
 484                argv_array_pushf(&submodule_options, "--after-context=%d",
 485                                 opt->post_context);
 486        if (opt->funcname)
 487                argv_array_push(&submodule_options, "-p");
 488        if (opt->funcbody)
 489                argv_array_push(&submodule_options, "-W");
 490        if (opt->all_match)
 491                argv_array_push(&submodule_options, "--all-match");
 492        if (opt->debug)
 493                argv_array_push(&submodule_options, "--debug");
 494        if (opt->status_only)
 495                argv_array_push(&submodule_options, "-q");
 496
 497        switch (pattern_type_arg) {
 498        case GREP_PATTERN_TYPE_BRE:
 499                argv_array_push(&submodule_options, "-G");
 500                break;
 501        case GREP_PATTERN_TYPE_ERE:
 502                argv_array_push(&submodule_options, "-E");
 503                break;
 504        case GREP_PATTERN_TYPE_FIXED:
 505                argv_array_push(&submodule_options, "-F");
 506                break;
 507        case GREP_PATTERN_TYPE_PCRE:
 508                argv_array_push(&submodule_options, "-P");
 509                break;
 510        case GREP_PATTERN_TYPE_UNSPECIFIED:
 511                break;
 512        default:
 513                die("BUG: Added a new grep pattern type without updating switch statement");
 514        }
 515
 516        for (pattern = opt->pattern_list; pattern != NULL;
 517             pattern = pattern->next) {
 518                switch (pattern->token) {
 519                case GREP_PATTERN:
 520                        argv_array_pushf(&submodule_options, "-e%s",
 521                                         pattern->pattern);
 522                        break;
 523                case GREP_AND:
 524                case GREP_OPEN_PAREN:
 525                case GREP_CLOSE_PAREN:
 526                case GREP_NOT:
 527                case GREP_OR:
 528                        argv_array_push(&submodule_options, pattern->pattern);
 529                        break;
 530                /* BODY and HEAD are not used by git-grep */
 531                case GREP_PATTERN_BODY:
 532                case GREP_PATTERN_HEAD:
 533                        break;
 534                }
 535        }
 536
 537        /*
 538         * Limit number of threads for child process to use.
 539         * This is to prevent potential fork-bomb behavior of git-grep as each
 540         * submodule process has its own thread pool.
 541         */
 542        argv_array_pushf(&submodule_options, "--threads=%d",
 543                         (num_threads + 1) / 2);
 544
 545        /* Add Pathspecs */
 546        argv_array_push(&submodule_options, "--");
 547        for (; *argv; argv++)
 548                argv_array_push(&submodule_options, *argv);
 549}
 550
 551/*
 552 * Launch child process to grep contents of a submodule
 553 */
 554static int grep_submodule_launch(struct grep_opt *opt,
 555                                 const struct grep_source *gs)
 556{
 557        struct child_process cp = CHILD_PROCESS_INIT;
 558        int status, i;
 559        const char *end_of_base;
 560        const char *name;
 561        struct strbuf child_output = STRBUF_INIT;
 562
 563        end_of_base = strchr(gs->name, ':');
 564        if (gs->identifier && end_of_base)
 565                name = end_of_base + 1;
 566        else
 567                name = gs->name;
 568
 569        prepare_submodule_repo_env(&cp.env_array);
 570        argv_array_push(&cp.env_array, GIT_DIR_ENVIRONMENT);
 571
 572        if (opt->relative && opt->prefix_length)
 573                argv_array_pushf(&cp.env_array, "%s=%s",
 574                                 GIT_TOPLEVEL_PREFIX_ENVIRONMENT,
 575                                 opt->prefix);
 576
 577        /* Add super prefix */
 578        argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
 579                         super_prefix ? super_prefix : "",
 580                         name);
 581        argv_array_push(&cp.args, "grep");
 582
 583        /*
 584         * Add basename of parent project
 585         * When performing grep on a tree object the filename is prefixed
 586         * with the object's name: 'tree-name:filename'.  In order to
 587         * provide uniformity of output we want to pass the name of the
 588         * parent project's object name to the submodule so the submodule can
 589         * prefix its output with the parent's name and not its own SHA1.
 590         */
 591        if (gs->identifier && end_of_base)
 592                argv_array_pushf(&cp.args, "--parent-basename=%.*s",
 593                                 (int) (end_of_base - gs->name),
 594                                 gs->name);
 595
 596        /* Add options */
 597        for (i = 0; i < submodule_options.argc; i++) {
 598                /*
 599                 * If there is a tree identifier for the submodule, add the
 600                 * rev after adding the submodule options but before the
 601                 * pathspecs.  To do this we listen for the '--' and insert the
 602                 * sha1 before pushing the '--' onto the child process argv
 603                 * array.
 604                 */
 605                if (gs->identifier &&
 606                    !strcmp("--", submodule_options.argv[i])) {
 607                        argv_array_push(&cp.args, sha1_to_hex(gs->identifier));
 608                }
 609
 610                argv_array_push(&cp.args, submodule_options.argv[i]);
 611        }
 612
 613        cp.git_cmd = 1;
 614        cp.dir = gs->path;
 615
 616        /*
 617         * Capture output to output buffer and check the return code from the
 618         * child process.  A '0' indicates a hit, a '1' indicates no hit and
 619         * anything else is an error.
 620         */
 621        status = capture_command(&cp, &child_output, 0);
 622        if (status && (status != 1)) {
 623                /* flush the buffer */
 624                write_or_die(1, child_output.buf, child_output.len);
 625                die("process for submodule '%s' failed with exit code: %d",
 626                    gs->name, status);
 627        }
 628
 629        opt->output(opt, child_output.buf, child_output.len);
 630        strbuf_release(&child_output);
 631        /* invert the return code to make a hit equal to 1 */
 632        return !status;
 633}
 634
 635/*
 636 * Prep grep structures for a submodule grep
 637 * sha1: the sha1 of the submodule or NULL if using the working tree
 638 * filename: name of the submodule including tree name of parent
 639 * path: location of the submodule
 640 */
 641static int grep_submodule(struct grep_opt *opt, const unsigned char *sha1,
 642                          const char *filename, const char *path)
 643{
 644        if (!is_submodule_initialized(path))
 645                return 0;
 646        if (!is_submodule_populated_gently(path, NULL)) {
 647                /*
 648                 * If searching history, check for the presense of the
 649                 * submodule's gitdir before skipping the submodule.
 650                 */
 651                if (sha1) {
 652                        const struct submodule *sub =
 653                                        submodule_from_path(null_sha1, path);
 654                        if (sub)
 655                                path = git_path("modules/%s", sub->name);
 656
 657                        if (!(is_directory(path) && is_git_directory(path)))
 658                                return 0;
 659                } else {
 660                        return 0;
 661                }
 662        }
 663
 664#ifndef NO_PTHREADS
 665        if (num_threads) {
 666                add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, sha1);
 667                return 0;
 668        } else
 669#endif
 670        {
 671                struct grep_source gs;
 672                int hit;
 673
 674                grep_source_init(&gs, GREP_SOURCE_SUBMODULE,
 675                                 filename, path, sha1);
 676                hit = grep_submodule_launch(opt, &gs);
 677
 678                grep_source_clear(&gs);
 679                return hit;
 680        }
 681}
 682
 683static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
 684                      int cached)
 685{
 686        int hit = 0;
 687        int nr;
 688        struct strbuf name = STRBUF_INIT;
 689        int name_base_len = 0;
 690        if (super_prefix) {
 691                name_base_len = strlen(super_prefix);
 692                strbuf_addstr(&name, super_prefix);
 693        }
 694
 695        read_cache();
 696
 697        for (nr = 0; nr < active_nr; nr++) {
 698                const struct cache_entry *ce = active_cache[nr];
 699                strbuf_setlen(&name, name_base_len);
 700                strbuf_addstr(&name, ce->name);
 701
 702                if (S_ISREG(ce->ce_mode) &&
 703                    match_pathspec(pathspec, name.buf, name.len, 0, NULL,
 704                                   S_ISDIR(ce->ce_mode) ||
 705                                   S_ISGITLINK(ce->ce_mode))) {
 706                        /*
 707                         * If CE_VALID is on, we assume worktree file and its
 708                         * cache entry are identical, even if worktree file has
 709                         * been modified, so use cache version instead
 710                         */
 711                        if (cached || (ce->ce_flags & CE_VALID) ||
 712                            ce_skip_worktree(ce)) {
 713                                if (ce_stage(ce) || ce_intent_to_add(ce))
 714                                        continue;
 715                                hit |= grep_oid(opt, &ce->oid, ce->name,
 716                                                 0, ce->name);
 717                        } else {
 718                                hit |= grep_file(opt, ce->name);
 719                        }
 720                } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 721                           submodule_path_match(pathspec, name.buf, NULL)) {
 722                        hit |= grep_submodule(opt, NULL, ce->name, ce->name);
 723                } else {
 724                        continue;
 725                }
 726
 727                if (ce_stage(ce)) {
 728                        do {
 729                                nr++;
 730                        } while (nr < active_nr &&
 731                                 !strcmp(ce->name, active_cache[nr]->name));
 732                        nr--; /* compensate for loop control */
 733                }
 734                if (hit && opt->status_only)
 735                        break;
 736        }
 737
 738        strbuf_release(&name);
 739        return hit;
 740}
 741
 742static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 743                     struct tree_desc *tree, struct strbuf *base, int tn_len,
 744                     int check_attr)
 745{
 746        int hit = 0;
 747        enum interesting match = entry_not_interesting;
 748        struct name_entry entry;
 749        int old_baselen = base->len;
 750        struct strbuf name = STRBUF_INIT;
 751        int name_base_len = 0;
 752        if (super_prefix) {
 753                strbuf_addstr(&name, super_prefix);
 754                name_base_len = name.len;
 755        }
 756
 757        while (tree_entry(tree, &entry)) {
 758                int te_len = tree_entry_len(&entry);
 759
 760                if (match != all_entries_interesting) {
 761                        strbuf_addstr(&name, base->buf + tn_len);
 762                        match = tree_entry_interesting(&entry, &name,
 763                                                       0, pathspec);
 764                        strbuf_setlen(&name, name_base_len);
 765
 766                        if (match == all_entries_not_interesting)
 767                                break;
 768                        if (match == entry_not_interesting)
 769                                continue;
 770                }
 771
 772                strbuf_add(base, entry.path, te_len);
 773
 774                if (S_ISREG(entry.mode)) {
 775                        hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
 776                                         check_attr ? base->buf + tn_len : NULL);
 777                } else if (S_ISDIR(entry.mode)) {
 778                        enum object_type type;
 779                        struct tree_desc sub;
 780                        void *data;
 781                        unsigned long size;
 782
 783                        data = lock_and_read_oid_file(entry.oid, &type, &size);
 784                        if (!data)
 785                                die(_("unable to read tree (%s)"),
 786                                    oid_to_hex(entry.oid));
 787
 788                        strbuf_addch(base, '/');
 789                        init_tree_desc(&sub, data, size);
 790                        hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 791                                         check_attr);
 792                        free(data);
 793                } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 794                        hit |= grep_submodule(opt, entry.oid->hash, base->buf,
 795                                              base->buf + tn_len);
 796                }
 797
 798                strbuf_setlen(base, old_baselen);
 799
 800                if (hit && opt->status_only)
 801                        break;
 802        }
 803
 804        strbuf_release(&name);
 805        return hit;
 806}
 807
 808static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 809                       struct object *obj, const char *name, const char *path)
 810{
 811        if (obj->type == OBJ_BLOB)
 812                return grep_oid(opt, &obj->oid, name, 0, path);
 813        if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 814                struct tree_desc tree;
 815                void *data;
 816                unsigned long size;
 817                struct strbuf base;
 818                int hit, len;
 819
 820                grep_read_lock();
 821                data = read_object_with_reference(obj->oid.hash, tree_type,
 822                                                  &size, NULL);
 823                grep_read_unlock();
 824
 825                if (!data)
 826                        die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 827
 828                /* Use parent's name as base when recursing submodules */
 829                if (recurse_submodules && parent_basename)
 830                        name = parent_basename;
 831
 832                len = name ? strlen(name) : 0;
 833                strbuf_init(&base, PATH_MAX + len + 1);
 834                if (len) {
 835                        strbuf_add(&base, name, len);
 836                        strbuf_addch(&base, ':');
 837                }
 838                init_tree_desc(&tree, data, size);
 839                hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 840                                obj->type == OBJ_COMMIT);
 841                strbuf_release(&base);
 842                free(data);
 843                return hit;
 844        }
 845        die(_("unable to grep from object of type %s"), typename(obj->type));
 846}
 847
 848static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 849                        const struct object_array *list)
 850{
 851        unsigned int i;
 852        int hit = 0;
 853        const unsigned int nr = list->nr;
 854
 855        for (i = 0; i < nr; i++) {
 856                struct object *real_obj;
 857                real_obj = deref_tag(list->objects[i].item, NULL, 0);
 858
 859                /* load the gitmodules file for this rev */
 860                if (recurse_submodules) {
 861                        submodule_free();
 862                        gitmodules_config_sha1(real_obj->oid.hash);
 863                }
 864                if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
 865                        hit = 1;
 866                        if (opt->status_only)
 867                                break;
 868                }
 869        }
 870        return hit;
 871}
 872
 873static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 874                          int exc_std, int use_index)
 875{
 876        struct dir_struct dir;
 877        int i, hit = 0;
 878
 879        memset(&dir, 0, sizeof(dir));
 880        if (!use_index)
 881                dir.flags |= DIR_NO_GITLINKS;
 882        if (exc_std)
 883                setup_standard_excludes(&dir);
 884
 885        fill_directory(&dir, &the_index, pathspec);
 886        for (i = 0; i < dir.nr; i++) {
 887                if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
 888                        continue;
 889                hit |= grep_file(opt, dir.entries[i]->name);
 890                if (hit && opt->status_only)
 891                        break;
 892        }
 893        return hit;
 894}
 895
 896static int context_callback(const struct option *opt, const char *arg,
 897                            int unset)
 898{
 899        struct grep_opt *grep_opt = opt->value;
 900        int value;
 901        const char *endp;
 902
 903        if (unset) {
 904                grep_opt->pre_context = grep_opt->post_context = 0;
 905                return 0;
 906        }
 907        value = strtol(arg, (char **)&endp, 10);
 908        if (*endp) {
 909                return error(_("switch `%c' expects a numerical value"),
 910                             opt->short_name);
 911        }
 912        grep_opt->pre_context = grep_opt->post_context = value;
 913        return 0;
 914}
 915
 916static int file_callback(const struct option *opt, const char *arg, int unset)
 917{
 918        struct grep_opt *grep_opt = opt->value;
 919        int from_stdin = !strcmp(arg, "-");
 920        FILE *patterns;
 921        int lno = 0;
 922        struct strbuf sb = STRBUF_INIT;
 923
 924        patterns = from_stdin ? stdin : fopen(arg, "r");
 925        if (!patterns)
 926                die_errno(_("cannot open '%s'"), arg);
 927        while (strbuf_getline(&sb, patterns) == 0) {
 928                /* ignore empty line like grep does */
 929                if (sb.len == 0)
 930                        continue;
 931
 932                append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 933                                GREP_PATTERN);
 934        }
 935        if (!from_stdin)
 936                fclose(patterns);
 937        strbuf_release(&sb);
 938        return 0;
 939}
 940
 941static int not_callback(const struct option *opt, const char *arg, int unset)
 942{
 943        struct grep_opt *grep_opt = opt->value;
 944        append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 945        return 0;
 946}
 947
 948static int and_callback(const struct option *opt, const char *arg, int unset)
 949{
 950        struct grep_opt *grep_opt = opt->value;
 951        append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 952        return 0;
 953}
 954
 955static int open_callback(const struct option *opt, const char *arg, int unset)
 956{
 957        struct grep_opt *grep_opt = opt->value;
 958        append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 959        return 0;
 960}
 961
 962static int close_callback(const struct option *opt, const char *arg, int unset)
 963{
 964        struct grep_opt *grep_opt = opt->value;
 965        append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 966        return 0;
 967}
 968
 969static int pattern_callback(const struct option *opt, const char *arg,
 970                            int unset)
 971{
 972        struct grep_opt *grep_opt = opt->value;
 973        append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 974        return 0;
 975}
 976
 977int cmd_grep(int argc, const char **argv, const char *prefix)
 978{
 979        int hit = 0;
 980        int cached = 0, untracked = 0, opt_exclude = -1;
 981        int seen_dashdash = 0;
 982        int external_grep_allowed__ignored;
 983        const char *show_in_pager = NULL, *default_pager = "dummy";
 984        struct grep_opt opt;
 985        struct object_array list = OBJECT_ARRAY_INIT;
 986        struct pathspec pathspec;
 987        struct string_list path_list = STRING_LIST_INIT_NODUP;
 988        int i;
 989        int dummy;
 990        int use_index = 1;
 991        int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 992        int allow_revs;
 993
 994        struct option options[] = {
 995                OPT_BOOL(0, "cached", &cached,
 996                        N_("search in index instead of in the work tree")),
 997                OPT_NEGBIT(0, "no-index", &use_index,
 998                         N_("find in contents not managed by git"), 1),
 999                OPT_BOOL(0, "untracked", &untracked,
1000                        N_("search in both tracked and untracked files")),
1001                OPT_SET_INT(0, "exclude-standard", &opt_exclude,
1002                            N_("ignore files specified via '.gitignore'"), 1),
1003                OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
1004                         N_("recursively search in each submodule")),
1005                OPT_STRING(0, "parent-basename", &parent_basename,
1006                           N_("basename"),
1007                           N_("prepend parent project's basename to output")),
1008                OPT_GROUP(""),
1009                OPT_BOOL('v', "invert-match", &opt.invert,
1010                        N_("show non-matching lines")),
1011                OPT_BOOL('i', "ignore-case", &opt.ignore_case,
1012                        N_("case insensitive matching")),
1013                OPT_BOOL('w', "word-regexp", &opt.word_regexp,
1014                        N_("match patterns only at word boundaries")),
1015                OPT_SET_INT('a', "text", &opt.binary,
1016                        N_("process binary files as text"), GREP_BINARY_TEXT),
1017                OPT_SET_INT('I', NULL, &opt.binary,
1018                        N_("don't match patterns in binary files"),
1019                        GREP_BINARY_NOMATCH),
1020                OPT_BOOL(0, "textconv", &opt.allow_textconv,
1021                         N_("process binary files with textconv filters")),
1022                { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
1023                        N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
1024                        NULL, 1 },
1025                OPT_GROUP(""),
1026                OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
1027                            N_("use extended POSIX regular expressions"),
1028                            GREP_PATTERN_TYPE_ERE),
1029                OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
1030                            N_("use basic POSIX regular expressions (default)"),
1031                            GREP_PATTERN_TYPE_BRE),
1032                OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
1033                            N_("interpret patterns as fixed strings"),
1034                            GREP_PATTERN_TYPE_FIXED),
1035                OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
1036                            N_("use Perl-compatible regular expressions"),
1037                            GREP_PATTERN_TYPE_PCRE),
1038                OPT_GROUP(""),
1039                OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1040                OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1041                OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1042                OPT_NEGBIT(0, "full-name", &opt.relative,
1043                        N_("show filenames relative to top directory"), 1),
1044                OPT_BOOL('l', "files-with-matches", &opt.name_only,
1045                        N_("show only filenames instead of matching lines")),
1046                OPT_BOOL(0, "name-only", &opt.name_only,
1047                        N_("synonym for --files-with-matches")),
1048                OPT_BOOL('L', "files-without-match",
1049                        &opt.unmatch_name_only,
1050                        N_("show only the names of files without match")),
1051                OPT_BOOL('z', "null", &opt.null_following_name,
1052                        N_("print NUL after filenames")),
1053                OPT_BOOL('c', "count", &opt.count,
1054                        N_("show the number of matches instead of matching lines")),
1055                OPT__COLOR(&opt.color, N_("highlight matches")),
1056                OPT_BOOL(0, "break", &opt.file_break,
1057                        N_("print empty line between matches from different files")),
1058                OPT_BOOL(0, "heading", &opt.heading,
1059                        N_("show filename only once above matches from same file")),
1060                OPT_GROUP(""),
1061                OPT_CALLBACK('C', "context", &opt, N_("n"),
1062                        N_("show <n> context lines before and after matches"),
1063                        context_callback),
1064                OPT_INTEGER('B', "before-context", &opt.pre_context,
1065                        N_("show <n> context lines before matches")),
1066                OPT_INTEGER('A', "after-context", &opt.post_context,
1067                        N_("show <n> context lines after matches")),
1068                OPT_INTEGER(0, "threads", &num_threads,
1069                        N_("use <n> worker threads")),
1070                OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1071                        context_callback),
1072                OPT_BOOL('p', "show-function", &opt.funcname,
1073                        N_("show a line with the function name before matches")),
1074                OPT_BOOL('W', "function-context", &opt.funcbody,
1075                        N_("show the surrounding function")),
1076                OPT_GROUP(""),
1077                OPT_CALLBACK('f', NULL, &opt, N_("file"),
1078                        N_("read patterns from file"), file_callback),
1079                { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1080                        N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1081                { OPTION_CALLBACK, 0, "and", &opt, NULL,
1082                  N_("combine patterns specified with -e"),
1083                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1084                OPT_BOOL(0, "or", &dummy, ""),
1085                { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1086                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1087                { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1088                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1089                  open_callback },
1090                { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1091                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1092                  close_callback },
1093                OPT__QUIET(&opt.status_only,
1094                           N_("indicate hit with exit status without output")),
1095                OPT_BOOL(0, "all-match", &opt.all_match,
1096                        N_("show only matches from files that match all patterns")),
1097                { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1098                  N_("show parse tree for grep expression"),
1099                  PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1100                OPT_GROUP(""),
1101                { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1102                        N_("pager"), N_("show matching files in the pager"),
1103                        PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1104                OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1105                         N_("allow calling of grep(1) (ignored by this build)")),
1106                OPT_END()
1107        };
1108
1109        init_grep_defaults();
1110        git_config(grep_cmd_config, NULL);
1111        grep_init(&opt, prefix);
1112        super_prefix = get_super_prefix();
1113
1114        /*
1115         * If there is no -- then the paths must exist in the working
1116         * tree.  If there is no explicit pattern specified with -e or
1117         * -f, we take the first unrecognized non option to be the
1118         * pattern, but then what follows it must be zero or more
1119         * valid refs up to the -- (if exists), and then existing
1120         * paths.  If there is an explicit pattern, then the first
1121         * unrecognized non option is the beginning of the refs list
1122         * that continues up to the -- (if exists), and then paths.
1123         */
1124        argc = parse_options(argc, argv, prefix, options, grep_usage,
1125                             PARSE_OPT_KEEP_DASHDASH |
1126                             PARSE_OPT_STOP_AT_NON_OPTION);
1127        grep_commit_pattern_type(pattern_type_arg, &opt);
1128
1129        if (use_index && !startup_info->have_repository) {
1130                int fallback = 0;
1131                git_config_get_bool("grep.fallbacktonoindex", &fallback);
1132                if (fallback)
1133                        use_index = 0;
1134                else
1135                        /* die the same way as if we did it at the beginning */
1136                        setup_git_directory();
1137        }
1138
1139        /*
1140         * skip a -- separator; we know it cannot be
1141         * separating revisions from pathnames if
1142         * we haven't even had any patterns yet
1143         */
1144        if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1145                argv++;
1146                argc--;
1147        }
1148
1149        /* First unrecognized non-option token */
1150        if (argc > 0 && !opt.pattern_list) {
1151                append_grep_pattern(&opt, argv[0], "command line", 0,
1152                                    GREP_PATTERN);
1153                argv++;
1154                argc--;
1155        }
1156
1157        if (show_in_pager == default_pager)
1158                show_in_pager = git_pager(1);
1159        if (show_in_pager) {
1160                opt.color = 0;
1161                opt.name_only = 1;
1162                opt.null_following_name = 1;
1163                opt.output_priv = &path_list;
1164                opt.output = append_path;
1165                string_list_append(&path_list, show_in_pager);
1166        }
1167
1168        if (!opt.pattern_list)
1169                die(_("no pattern given."));
1170        if (!opt.fixed && opt.ignore_case)
1171                opt.regflags |= REG_ICASE;
1172
1173        compile_grep_patterns(&opt);
1174
1175        /*
1176         * We have to find "--" in a separate pass, because its presence
1177         * influences how we will parse arguments that come before it.
1178         */
1179        for (i = 0; i < argc; i++) {
1180                if (!strcmp(argv[i], "--")) {
1181                        seen_dashdash = 1;
1182                        break;
1183                }
1184        }
1185
1186        /*
1187         * Resolve any rev arguments. If we have a dashdash, then everything up
1188         * to it must resolve as a rev. If not, then we stop at the first
1189         * non-rev and assume everything else is a path.
1190         */
1191        allow_revs = use_index && !untracked;
1192        for (i = 0; i < argc; i++) {
1193                const char *arg = argv[i];
1194                struct object_id oid;
1195                struct object_context oc;
1196                struct object *object;
1197
1198                if (!strcmp(arg, "--")) {
1199                        i++;
1200                        break;
1201                }
1202
1203                if (!allow_revs) {
1204                        if (seen_dashdash)
1205                                die(_("--no-index or --untracked cannot be used with revs"));
1206                        break;
1207                }
1208
1209                if (get_sha1_with_context(arg, GET_SHA1_RECORD_PATH,
1210                                          oid.hash, &oc)) {
1211                        if (seen_dashdash)
1212                                die(_("unable to resolve revision: %s"), arg);
1213                        break;
1214                }
1215
1216                object = parse_object_or_die(&oid, arg);
1217                if (!seen_dashdash)
1218                        verify_non_filename(prefix, arg);
1219                add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1220                free(oc.path);
1221        }
1222
1223        /*
1224         * Anything left over is presumed to be a path. But in the non-dashdash
1225         * "do what I mean" case, we verify and complain when that isn't true.
1226         */
1227        if (!seen_dashdash) {
1228                int j;
1229                for (j = i; j < argc; j++)
1230                        verify_filename(prefix, argv[j], j == i && allow_revs);
1231        }
1232
1233        parse_pathspec(&pathspec, 0,
1234                       PATHSPEC_PREFER_CWD |
1235                       (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1236                       prefix, argv + i);
1237        pathspec.max_depth = opt.max_depth;
1238        pathspec.recursive = 1;
1239
1240#ifndef NO_PTHREADS
1241        if (list.nr || cached || show_in_pager)
1242                num_threads = 0;
1243        else if (num_threads == 0)
1244                num_threads = GREP_NUM_THREADS_DEFAULT;
1245        else if (num_threads < 0)
1246                die(_("invalid number of threads specified (%d)"), num_threads);
1247#else
1248        if (num_threads)
1249                warning(_("no threads support, ignoring --threads"));
1250        num_threads = 0;
1251#endif
1252
1253#ifndef NO_PTHREADS
1254        if (num_threads) {
1255                if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1256                    && (opt.pre_context || opt.post_context ||
1257                        opt.file_break || opt.funcbody))
1258                        skip_first_line = 1;
1259                start_threads(&opt);
1260        }
1261#endif
1262
1263        if (recurse_submodules) {
1264                gitmodules_config();
1265                compile_submodule_options(&opt, argv + i, cached, untracked,
1266                                          opt_exclude, use_index,
1267                                          pattern_type_arg);
1268        }
1269
1270        if (show_in_pager && (cached || list.nr))
1271                die(_("--open-files-in-pager only works on the worktree"));
1272
1273        if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1274                const char *pager = path_list.items[0].string;
1275                int len = strlen(pager);
1276
1277                if (len > 4 && is_dir_sep(pager[len - 5]))
1278                        pager += len - 4;
1279
1280                if (opt.ignore_case && !strcmp("less", pager))
1281                        string_list_append(&path_list, "-I");
1282
1283                if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1284                        struct strbuf buf = STRBUF_INIT;
1285                        strbuf_addf(&buf, "+/%s%s",
1286                                        strcmp("less", pager) ? "" : "*",
1287                                        opt.pattern_list->pattern);
1288                        string_list_append(&path_list, buf.buf);
1289                        strbuf_detach(&buf, NULL);
1290                }
1291        }
1292
1293        if (recurse_submodules && (!use_index || untracked))
1294                die(_("option not supported with --recurse-submodules."));
1295
1296        if (!show_in_pager && !opt.status_only)
1297                setup_pager();
1298
1299        if (!use_index && (untracked || cached))
1300                die(_("--cached or --untracked cannot be used with --no-index."));
1301
1302        if (!use_index || untracked) {
1303                int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1304                hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1305        } else if (0 <= opt_exclude) {
1306                die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1307        } else if (!list.nr) {
1308                if (!cached)
1309                        setup_work_tree();
1310
1311                hit = grep_cache(&opt, &pathspec, cached);
1312        } else {
1313                if (cached)
1314                        die(_("both --cached and trees are given."));
1315                hit = grep_objects(&opt, &pathspec, &list);
1316        }
1317
1318        if (num_threads)
1319                hit |= wait_all();
1320        if (hit && show_in_pager)
1321                run_pager(&opt, prefix);
1322        clear_pathspec(&pathspec);
1323        free_grep_patterns(&opt);
1324        return !hit;
1325}