Andrew's git / gitweb.git / builtin/grep.c
?
builtin / grep.con commit ls-files: convert show_files to take an index (f587c8d)
   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        return st;
 306}
 307
 308static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
 309{
 310        void *data;
 311
 312        grep_read_lock();
 313        data = read_sha1_file(oid->hash, type, size);
 314        grep_read_unlock();
 315        return data;
 316}
 317
 318static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
 319                     const char *filename, int tree_name_len,
 320                     const char *path)
 321{
 322        struct strbuf pathbuf = STRBUF_INIT;
 323
 324        if (super_prefix) {
 325                strbuf_add(&pathbuf, filename, tree_name_len);
 326                strbuf_addstr(&pathbuf, super_prefix);
 327                strbuf_addstr(&pathbuf, filename + tree_name_len);
 328        } else {
 329                strbuf_addstr(&pathbuf, filename);
 330        }
 331
 332        if (opt->relative && opt->prefix_length) {
 333                char *name = strbuf_detach(&pathbuf, NULL);
 334                quote_path_relative(name + tree_name_len, opt->prefix, &pathbuf);
 335                strbuf_insert(&pathbuf, 0, name, tree_name_len);
 336                free(name);
 337        }
 338
 339#ifndef NO_PTHREADS
 340        if (num_threads) {
 341                add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
 342                strbuf_release(&pathbuf);
 343                return 0;
 344        } else
 345#endif
 346        {
 347                struct grep_source gs;
 348                int hit;
 349
 350                grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
 351                strbuf_release(&pathbuf);
 352                hit = grep_source(opt, &gs);
 353
 354                grep_source_clear(&gs);
 355                return hit;
 356        }
 357}
 358
 359static int grep_file(struct grep_opt *opt, const char *filename)
 360{
 361        struct strbuf buf = STRBUF_INIT;
 362
 363        if (super_prefix)
 364                strbuf_addstr(&buf, super_prefix);
 365        strbuf_addstr(&buf, filename);
 366
 367        if (opt->relative && opt->prefix_length) {
 368                char *name = strbuf_detach(&buf, NULL);
 369                quote_path_relative(name, opt->prefix, &buf);
 370                free(name);
 371        }
 372
 373#ifndef NO_PTHREADS
 374        if (num_threads) {
 375                add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
 376                strbuf_release(&buf);
 377                return 0;
 378        } else
 379#endif
 380        {
 381                struct grep_source gs;
 382                int hit;
 383
 384                grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 385                strbuf_release(&buf);
 386                hit = grep_source(opt, &gs);
 387
 388                grep_source_clear(&gs);
 389                return hit;
 390        }
 391}
 392
 393static void append_path(struct grep_opt *opt, const void *data, size_t len)
 394{
 395        struct string_list *path_list = opt->output_priv;
 396
 397        if (len == 1 && *(const char *)data == '\0')
 398                return;
 399        string_list_append(path_list, xstrndup(data, len));
 400}
 401
 402static void run_pager(struct grep_opt *opt, const char *prefix)
 403{
 404        struct string_list *path_list = opt->output_priv;
 405        struct child_process child = CHILD_PROCESS_INIT;
 406        int i, status;
 407
 408        for (i = 0; i < path_list->nr; i++)
 409                argv_array_push(&child.args, path_list->items[i].string);
 410        child.dir = prefix;
 411        child.use_shell = 1;
 412
 413        status = run_command(&child);
 414        if (status)
 415                exit(status);
 416}
 417
 418static void compile_submodule_options(const struct grep_opt *opt,
 419                                      const char **argv,
 420                                      int cached, int untracked,
 421                                      int opt_exclude, int use_index,
 422                                      int pattern_type_arg)
 423{
 424        struct grep_pat *pattern;
 425
 426        if (recurse_submodules)
 427                argv_array_push(&submodule_options, "--recurse-submodules");
 428
 429        if (cached)
 430                argv_array_push(&submodule_options, "--cached");
 431        if (!use_index)
 432                argv_array_push(&submodule_options, "--no-index");
 433        if (untracked)
 434                argv_array_push(&submodule_options, "--untracked");
 435        if (opt_exclude > 0)
 436                argv_array_push(&submodule_options, "--exclude-standard");
 437
 438        if (opt->invert)
 439                argv_array_push(&submodule_options, "-v");
 440        if (opt->ignore_case)
 441                argv_array_push(&submodule_options, "-i");
 442        if (opt->word_regexp)
 443                argv_array_push(&submodule_options, "-w");
 444        switch (opt->binary) {
 445        case GREP_BINARY_NOMATCH:
 446                argv_array_push(&submodule_options, "-I");
 447                break;
 448        case GREP_BINARY_TEXT:
 449                argv_array_push(&submodule_options, "-a");
 450                break;
 451        default:
 452                break;
 453        }
 454        if (opt->allow_textconv)
 455                argv_array_push(&submodule_options, "--textconv");
 456        if (opt->max_depth != -1)
 457                argv_array_pushf(&submodule_options, "--max-depth=%d",
 458                                 opt->max_depth);
 459        if (opt->linenum)
 460                argv_array_push(&submodule_options, "-n");
 461        if (!opt->pathname)
 462                argv_array_push(&submodule_options, "-h");
 463        if (!opt->relative)
 464                argv_array_push(&submodule_options, "--full-name");
 465        if (opt->name_only)
 466                argv_array_push(&submodule_options, "-l");
 467        if (opt->unmatch_name_only)
 468                argv_array_push(&submodule_options, "-L");
 469        if (opt->null_following_name)
 470                argv_array_push(&submodule_options, "-z");
 471        if (opt->count)
 472                argv_array_push(&submodule_options, "-c");
 473        if (opt->file_break)
 474                argv_array_push(&submodule_options, "--break");
 475        if (opt->heading)
 476                argv_array_push(&submodule_options, "--heading");
 477        if (opt->pre_context)
 478                argv_array_pushf(&submodule_options, "--before-context=%d",
 479                                 opt->pre_context);
 480        if (opt->post_context)
 481                argv_array_pushf(&submodule_options, "--after-context=%d",
 482                                 opt->post_context);
 483        if (opt->funcname)
 484                argv_array_push(&submodule_options, "-p");
 485        if (opt->funcbody)
 486                argv_array_push(&submodule_options, "-W");
 487        if (opt->all_match)
 488                argv_array_push(&submodule_options, "--all-match");
 489        if (opt->debug)
 490                argv_array_push(&submodule_options, "--debug");
 491        if (opt->status_only)
 492                argv_array_push(&submodule_options, "-q");
 493
 494        switch (pattern_type_arg) {
 495        case GREP_PATTERN_TYPE_BRE:
 496                argv_array_push(&submodule_options, "-G");
 497                break;
 498        case GREP_PATTERN_TYPE_ERE:
 499                argv_array_push(&submodule_options, "-E");
 500                break;
 501        case GREP_PATTERN_TYPE_FIXED:
 502                argv_array_push(&submodule_options, "-F");
 503                break;
 504        case GREP_PATTERN_TYPE_PCRE:
 505                argv_array_push(&submodule_options, "-P");
 506                break;
 507        case GREP_PATTERN_TYPE_UNSPECIFIED:
 508                break;
 509        default:
 510                die("BUG: Added a new grep pattern type without updating switch statement");
 511        }
 512
 513        for (pattern = opt->pattern_list; pattern != NULL;
 514             pattern = pattern->next) {
 515                switch (pattern->token) {
 516                case GREP_PATTERN:
 517                        argv_array_pushf(&submodule_options, "-e%s",
 518                                         pattern->pattern);
 519                        break;
 520                case GREP_AND:
 521                case GREP_OPEN_PAREN:
 522                case GREP_CLOSE_PAREN:
 523                case GREP_NOT:
 524                case GREP_OR:
 525                        argv_array_push(&submodule_options, pattern->pattern);
 526                        break;
 527                /* BODY and HEAD are not used by git-grep */
 528                case GREP_PATTERN_BODY:
 529                case GREP_PATTERN_HEAD:
 530                        break;
 531                }
 532        }
 533
 534        /*
 535         * Limit number of threads for child process to use.
 536         * This is to prevent potential fork-bomb behavior of git-grep as each
 537         * submodule process has its own thread pool.
 538         */
 539        argv_array_pushf(&submodule_options, "--threads=%d",
 540                         (num_threads + 1) / 2);
 541
 542        /* Add Pathspecs */
 543        argv_array_push(&submodule_options, "--");
 544        for (; *argv; argv++)
 545                argv_array_push(&submodule_options, *argv);
 546}
 547
 548/*
 549 * Launch child process to grep contents of a submodule
 550 */
 551static int grep_submodule_launch(struct grep_opt *opt,
 552                                 const struct grep_source *gs)
 553{
 554        struct child_process cp = CHILD_PROCESS_INIT;
 555        int status, i;
 556        const char *end_of_base;
 557        const char *name;
 558        struct strbuf child_output = STRBUF_INIT;
 559
 560        end_of_base = strchr(gs->name, ':');
 561        if (gs->identifier && end_of_base)
 562                name = end_of_base + 1;
 563        else
 564                name = gs->name;
 565
 566        prepare_submodule_repo_env(&cp.env_array);
 567        argv_array_push(&cp.env_array, GIT_DIR_ENVIRONMENT);
 568
 569        if (opt->relative && opt->prefix_length)
 570                argv_array_pushf(&cp.env_array, "%s=%s",
 571                                 GIT_TOPLEVEL_PREFIX_ENVIRONMENT,
 572                                 opt->prefix);
 573
 574        /* Add super prefix */
 575        argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
 576                         super_prefix ? super_prefix : "",
 577                         name);
 578        argv_array_push(&cp.args, "grep");
 579
 580        /*
 581         * Add basename of parent project
 582         * When performing grep on a tree object the filename is prefixed
 583         * with the object's name: 'tree-name:filename'.  In order to
 584         * provide uniformity of output we want to pass the name of the
 585         * parent project's object name to the submodule so the submodule can
 586         * prefix its output with the parent's name and not its own SHA1.
 587         */
 588        if (gs->identifier && end_of_base)
 589                argv_array_pushf(&cp.args, "--parent-basename=%.*s",
 590                                 (int) (end_of_base - gs->name),
 591                                 gs->name);
 592
 593        /* Add options */
 594        for (i = 0; i < submodule_options.argc; i++) {
 595                /*
 596                 * If there is a tree identifier for the submodule, add the
 597                 * rev after adding the submodule options but before the
 598                 * pathspecs.  To do this we listen for the '--' and insert the
 599                 * sha1 before pushing the '--' onto the child process argv
 600                 * array.
 601                 */
 602                if (gs->identifier &&
 603                    !strcmp("--", submodule_options.argv[i])) {
 604                        argv_array_push(&cp.args, sha1_to_hex(gs->identifier));
 605                }
 606
 607                argv_array_push(&cp.args, submodule_options.argv[i]);
 608        }
 609
 610        cp.git_cmd = 1;
 611        cp.dir = gs->path;
 612
 613        /*
 614         * Capture output to output buffer and check the return code from the
 615         * child process.  A '0' indicates a hit, a '1' indicates no hit and
 616         * anything else is an error.
 617         */
 618        status = capture_command(&cp, &child_output, 0);
 619        if (status && (status != 1)) {
 620                /* flush the buffer */
 621                write_or_die(1, child_output.buf, child_output.len);
 622                die("process for submodule '%s' failed with exit code: %d",
 623                    gs->name, status);
 624        }
 625
 626        opt->output(opt, child_output.buf, child_output.len);
 627        strbuf_release(&child_output);
 628        /* invert the return code to make a hit equal to 1 */
 629        return !status;
 630}
 631
 632/*
 633 * Prep grep structures for a submodule grep
 634 * sha1: the sha1 of the submodule or NULL if using the working tree
 635 * filename: name of the submodule including tree name of parent
 636 * path: location of the submodule
 637 */
 638static int grep_submodule(struct grep_opt *opt, const unsigned char *sha1,
 639                          const char *filename, const char *path)
 640{
 641        if (!is_submodule_initialized(path))
 642                return 0;
 643        if (!is_submodule_populated_gently(path, NULL)) {
 644                /*
 645                 * If searching history, check for the presense of the
 646                 * submodule's gitdir before skipping the submodule.
 647                 */
 648                if (sha1) {
 649                        const struct submodule *sub =
 650                                        submodule_from_path(null_sha1, path);
 651                        if (sub)
 652                                path = git_path("modules/%s", sub->name);
 653
 654                        if (!(is_directory(path) && is_git_directory(path)))
 655                                return 0;
 656                } else {
 657                        return 0;
 658                }
 659        }
 660
 661#ifndef NO_PTHREADS
 662        if (num_threads) {
 663                add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, sha1);
 664                return 0;
 665        } else
 666#endif
 667        {
 668                struct grep_source gs;
 669                int hit;
 670
 671                grep_source_init(&gs, GREP_SOURCE_SUBMODULE,
 672                                 filename, path, sha1);
 673                hit = grep_submodule_launch(opt, &gs);
 674
 675                grep_source_clear(&gs);
 676                return hit;
 677        }
 678}
 679
 680static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
 681                      int cached)
 682{
 683        int hit = 0;
 684        int nr;
 685        struct strbuf name = STRBUF_INIT;
 686        int name_base_len = 0;
 687        if (super_prefix) {
 688                name_base_len = strlen(super_prefix);
 689                strbuf_addstr(&name, super_prefix);
 690        }
 691
 692        read_cache();
 693
 694        for (nr = 0; nr < active_nr; nr++) {
 695                const struct cache_entry *ce = active_cache[nr];
 696                strbuf_setlen(&name, name_base_len);
 697                strbuf_addstr(&name, ce->name);
 698
 699                if (S_ISREG(ce->ce_mode) &&
 700                    match_pathspec(pathspec, name.buf, name.len, 0, NULL,
 701                                   S_ISDIR(ce->ce_mode) ||
 702                                   S_ISGITLINK(ce->ce_mode))) {
 703                        /*
 704                         * If CE_VALID is on, we assume worktree file and its
 705                         * cache entry are identical, even if worktree file has
 706                         * been modified, so use cache version instead
 707                         */
 708                        if (cached || (ce->ce_flags & CE_VALID) ||
 709                            ce_skip_worktree(ce)) {
 710                                if (ce_stage(ce) || ce_intent_to_add(ce))
 711                                        continue;
 712                                hit |= grep_oid(opt, &ce->oid, ce->name,
 713                                                 0, ce->name);
 714                        } else {
 715                                hit |= grep_file(opt, ce->name);
 716                        }
 717                } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 718                           submodule_path_match(pathspec, name.buf, NULL)) {
 719                        hit |= grep_submodule(opt, NULL, ce->name, ce->name);
 720                } else {
 721                        continue;
 722                }
 723
 724                if (ce_stage(ce)) {
 725                        do {
 726                                nr++;
 727                        } while (nr < active_nr &&
 728                                 !strcmp(ce->name, active_cache[nr]->name));
 729                        nr--; /* compensate for loop control */
 730                }
 731                if (hit && opt->status_only)
 732                        break;
 733        }
 734
 735        strbuf_release(&name);
 736        return hit;
 737}
 738
 739static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 740                     struct tree_desc *tree, struct strbuf *base, int tn_len,
 741                     int check_attr)
 742{
 743        int hit = 0;
 744        enum interesting match = entry_not_interesting;
 745        struct name_entry entry;
 746        int old_baselen = base->len;
 747        struct strbuf name = STRBUF_INIT;
 748        int name_base_len = 0;
 749        if (super_prefix) {
 750                strbuf_addstr(&name, super_prefix);
 751                name_base_len = name.len;
 752        }
 753
 754        while (tree_entry(tree, &entry)) {
 755                int te_len = tree_entry_len(&entry);
 756
 757                if (match != all_entries_interesting) {
 758                        strbuf_addstr(&name, base->buf + tn_len);
 759                        match = tree_entry_interesting(&entry, &name,
 760                                                       0, pathspec);
 761                        strbuf_setlen(&name, name_base_len);
 762
 763                        if (match == all_entries_not_interesting)
 764                                break;
 765                        if (match == entry_not_interesting)
 766                                continue;
 767                }
 768
 769                strbuf_add(base, entry.path, te_len);
 770
 771                if (S_ISREG(entry.mode)) {
 772                        hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
 773                                         check_attr ? base->buf + tn_len : NULL);
 774                } else if (S_ISDIR(entry.mode)) {
 775                        enum object_type type;
 776                        struct tree_desc sub;
 777                        void *data;
 778                        unsigned long size;
 779
 780                        data = lock_and_read_oid_file(entry.oid, &type, &size);
 781                        if (!data)
 782                                die(_("unable to read tree (%s)"),
 783                                    oid_to_hex(entry.oid));
 784
 785                        strbuf_addch(base, '/');
 786                        init_tree_desc(&sub, data, size);
 787                        hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 788                                         check_attr);
 789                        free(data);
 790                } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 791                        hit |= grep_submodule(opt, entry.oid->hash, base->buf,
 792                                              base->buf + tn_len);
 793                }
 794
 795                strbuf_setlen(base, old_baselen);
 796
 797                if (hit && opt->status_only)
 798                        break;
 799        }
 800
 801        strbuf_release(&name);
 802        return hit;
 803}
 804
 805static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 806                       struct object *obj, const char *name, const char *path)
 807{
 808        if (obj->type == OBJ_BLOB)
 809                return grep_oid(opt, &obj->oid, name, 0, path);
 810        if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 811                struct tree_desc tree;
 812                void *data;
 813                unsigned long size;
 814                struct strbuf base;
 815                int hit, len;
 816
 817                grep_read_lock();
 818                data = read_object_with_reference(obj->oid.hash, tree_type,
 819                                                  &size, NULL);
 820                grep_read_unlock();
 821
 822                if (!data)
 823                        die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 824
 825                /* Use parent's name as base when recursing submodules */
 826                if (recurse_submodules && parent_basename)
 827                        name = parent_basename;
 828
 829                len = name ? strlen(name) : 0;
 830                strbuf_init(&base, PATH_MAX + len + 1);
 831                if (len) {
 832                        strbuf_add(&base, name, len);
 833                        strbuf_addch(&base, ':');
 834                }
 835                init_tree_desc(&tree, data, size);
 836                hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 837                                obj->type == OBJ_COMMIT);
 838                strbuf_release(&base);
 839                free(data);
 840                return hit;
 841        }
 842        die(_("unable to grep from object of type %s"), typename(obj->type));
 843}
 844
 845static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 846                        const struct object_array *list)
 847{
 848        unsigned int i;
 849        int hit = 0;
 850        const unsigned int nr = list->nr;
 851
 852        for (i = 0; i < nr; i++) {
 853                struct object *real_obj;
 854                real_obj = deref_tag(list->objects[i].item, NULL, 0);
 855
 856                /* load the gitmodules file for this rev */
 857                if (recurse_submodules) {
 858                        submodule_free();
 859                        gitmodules_config_sha1(real_obj->oid.hash);
 860                }
 861                if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
 862                        hit = 1;
 863                        if (opt->status_only)
 864                                break;
 865                }
 866        }
 867        return hit;
 868}
 869
 870static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 871                          int exc_std, int use_index)
 872{
 873        struct dir_struct dir;
 874        int i, hit = 0;
 875
 876        memset(&dir, 0, sizeof(dir));
 877        if (!use_index)
 878                dir.flags |= DIR_NO_GITLINKS;
 879        if (exc_std)
 880                setup_standard_excludes(&dir);
 881
 882        fill_directory(&dir, &the_index, pathspec);
 883        for (i = 0; i < dir.nr; i++) {
 884                if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
 885                        continue;
 886                hit |= grep_file(opt, dir.entries[i]->name);
 887                if (hit && opt->status_only)
 888                        break;
 889        }
 890        return hit;
 891}
 892
 893static int context_callback(const struct option *opt, const char *arg,
 894                            int unset)
 895{
 896        struct grep_opt *grep_opt = opt->value;
 897        int value;
 898        const char *endp;
 899
 900        if (unset) {
 901                grep_opt->pre_context = grep_opt->post_context = 0;
 902                return 0;
 903        }
 904        value = strtol(arg, (char **)&endp, 10);
 905        if (*endp) {
 906                return error(_("switch `%c' expects a numerical value"),
 907                             opt->short_name);
 908        }
 909        grep_opt->pre_context = grep_opt->post_context = value;
 910        return 0;
 911}
 912
 913static int file_callback(const struct option *opt, const char *arg, int unset)
 914{
 915        struct grep_opt *grep_opt = opt->value;
 916        int from_stdin = !strcmp(arg, "-");
 917        FILE *patterns;
 918        int lno = 0;
 919        struct strbuf sb = STRBUF_INIT;
 920
 921        patterns = from_stdin ? stdin : fopen(arg, "r");
 922        if (!patterns)
 923                die_errno(_("cannot open '%s'"), arg);
 924        while (strbuf_getline(&sb, patterns) == 0) {
 925                /* ignore empty line like grep does */
 926                if (sb.len == 0)
 927                        continue;
 928
 929                append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 930                                GREP_PATTERN);
 931        }
 932        if (!from_stdin)
 933                fclose(patterns);
 934        strbuf_release(&sb);
 935        return 0;
 936}
 937
 938static int not_callback(const struct option *opt, const char *arg, int unset)
 939{
 940        struct grep_opt *grep_opt = opt->value;
 941        append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 942        return 0;
 943}
 944
 945static int and_callback(const struct option *opt, const char *arg, int unset)
 946{
 947        struct grep_opt *grep_opt = opt->value;
 948        append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 949        return 0;
 950}
 951
 952static int open_callback(const struct option *opt, const char *arg, int unset)
 953{
 954        struct grep_opt *grep_opt = opt->value;
 955        append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 956        return 0;
 957}
 958
 959static int close_callback(const struct option *opt, const char *arg, int unset)
 960{
 961        struct grep_opt *grep_opt = opt->value;
 962        append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 963        return 0;
 964}
 965
 966static int pattern_callback(const struct option *opt, const char *arg,
 967                            int unset)
 968{
 969        struct grep_opt *grep_opt = opt->value;
 970        append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 971        return 0;
 972}
 973
 974int cmd_grep(int argc, const char **argv, const char *prefix)
 975{
 976        int hit = 0;
 977        int cached = 0, untracked = 0, opt_exclude = -1;
 978        int seen_dashdash = 0;
 979        int external_grep_allowed__ignored;
 980        const char *show_in_pager = NULL, *default_pager = "dummy";
 981        struct grep_opt opt;
 982        struct object_array list = OBJECT_ARRAY_INIT;
 983        struct pathspec pathspec;
 984        struct string_list path_list = STRING_LIST_INIT_NODUP;
 985        int i;
 986        int dummy;
 987        int use_index = 1;
 988        int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 989        int allow_revs;
 990
 991        struct option options[] = {
 992                OPT_BOOL(0, "cached", &cached,
 993                        N_("search in index instead of in the work tree")),
 994                OPT_NEGBIT(0, "no-index", &use_index,
 995                         N_("find in contents not managed by git"), 1),
 996                OPT_BOOL(0, "untracked", &untracked,
 997                        N_("search in both tracked and untracked files")),
 998                OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 999                            N_("ignore files specified via '.gitignore'"), 1),
1000                OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
1001                         N_("recursively search in each submodule")),
1002                OPT_STRING(0, "parent-basename", &parent_basename,
1003                           N_("basename"),
1004                           N_("prepend parent project's basename to output")),
1005                OPT_GROUP(""),
1006                OPT_BOOL('v', "invert-match", &opt.invert,
1007                        N_("show non-matching lines")),
1008                OPT_BOOL('i', "ignore-case", &opt.ignore_case,
1009                        N_("case insensitive matching")),
1010                OPT_BOOL('w', "word-regexp", &opt.word_regexp,
1011                        N_("match patterns only at word boundaries")),
1012                OPT_SET_INT('a', "text", &opt.binary,
1013                        N_("process binary files as text"), GREP_BINARY_TEXT),
1014                OPT_SET_INT('I', NULL, &opt.binary,
1015                        N_("don't match patterns in binary files"),
1016                        GREP_BINARY_NOMATCH),
1017                OPT_BOOL(0, "textconv", &opt.allow_textconv,
1018                         N_("process binary files with textconv filters")),
1019                { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
1020                        N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
1021                        NULL, 1 },
1022                OPT_GROUP(""),
1023                OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
1024                            N_("use extended POSIX regular expressions"),
1025                            GREP_PATTERN_TYPE_ERE),
1026                OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
1027                            N_("use basic POSIX regular expressions (default)"),
1028                            GREP_PATTERN_TYPE_BRE),
1029                OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
1030                            N_("interpret patterns as fixed strings"),
1031                            GREP_PATTERN_TYPE_FIXED),
1032                OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
1033                            N_("use Perl-compatible regular expressions"),
1034                            GREP_PATTERN_TYPE_PCRE),
1035                OPT_GROUP(""),
1036                OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1037                OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1038                OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1039                OPT_NEGBIT(0, "full-name", &opt.relative,
1040                        N_("show filenames relative to top directory"), 1),
1041                OPT_BOOL('l', "files-with-matches", &opt.name_only,
1042                        N_("show only filenames instead of matching lines")),
1043                OPT_BOOL(0, "name-only", &opt.name_only,
1044                        N_("synonym for --files-with-matches")),
1045                OPT_BOOL('L', "files-without-match",
1046                        &opt.unmatch_name_only,
1047                        N_("show only the names of files without match")),
1048                OPT_BOOL('z', "null", &opt.null_following_name,
1049                        N_("print NUL after filenames")),
1050                OPT_BOOL('c', "count", &opt.count,
1051                        N_("show the number of matches instead of matching lines")),
1052                OPT__COLOR(&opt.color, N_("highlight matches")),
1053                OPT_BOOL(0, "break", &opt.file_break,
1054                        N_("print empty line between matches from different files")),
1055                OPT_BOOL(0, "heading", &opt.heading,
1056                        N_("show filename only once above matches from same file")),
1057                OPT_GROUP(""),
1058                OPT_CALLBACK('C', "context", &opt, N_("n"),
1059                        N_("show <n> context lines before and after matches"),
1060                        context_callback),
1061                OPT_INTEGER('B', "before-context", &opt.pre_context,
1062                        N_("show <n> context lines before matches")),
1063                OPT_INTEGER('A', "after-context", &opt.post_context,
1064                        N_("show <n> context lines after matches")),
1065                OPT_INTEGER(0, "threads", &num_threads,
1066                        N_("use <n> worker threads")),
1067                OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1068                        context_callback),
1069                OPT_BOOL('p', "show-function", &opt.funcname,
1070                        N_("show a line with the function name before matches")),
1071                OPT_BOOL('W', "function-context", &opt.funcbody,
1072                        N_("show the surrounding function")),
1073                OPT_GROUP(""),
1074                OPT_CALLBACK('f', NULL, &opt, N_("file"),
1075                        N_("read patterns from file"), file_callback),
1076                { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1077                        N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1078                { OPTION_CALLBACK, 0, "and", &opt, NULL,
1079                  N_("combine patterns specified with -e"),
1080                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1081                OPT_BOOL(0, "or", &dummy, ""),
1082                { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1083                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1084                { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1085                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1086                  open_callback },
1087                { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1088                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1089                  close_callback },
1090                OPT__QUIET(&opt.status_only,
1091                           N_("indicate hit with exit status without output")),
1092                OPT_BOOL(0, "all-match", &opt.all_match,
1093                        N_("show only matches from files that match all patterns")),
1094                { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1095                  N_("show parse tree for grep expression"),
1096                  PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1097                OPT_GROUP(""),
1098                { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1099                        N_("pager"), N_("show matching files in the pager"),
1100                        PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1101                OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1102                         N_("allow calling of grep(1) (ignored by this build)")),
1103                OPT_END()
1104        };
1105
1106        init_grep_defaults();
1107        git_config(grep_cmd_config, NULL);
1108        grep_init(&opt, prefix);
1109        super_prefix = get_super_prefix();
1110
1111        /*
1112         * If there is no -- then the paths must exist in the working
1113         * tree.  If there is no explicit pattern specified with -e or
1114         * -f, we take the first unrecognized non option to be the
1115         * pattern, but then what follows it must be zero or more
1116         * valid refs up to the -- (if exists), and then existing
1117         * paths.  If there is an explicit pattern, then the first
1118         * unrecognized non option is the beginning of the refs list
1119         * that continues up to the -- (if exists), and then paths.
1120         */
1121        argc = parse_options(argc, argv, prefix, options, grep_usage,
1122                             PARSE_OPT_KEEP_DASHDASH |
1123                             PARSE_OPT_STOP_AT_NON_OPTION);
1124        grep_commit_pattern_type(pattern_type_arg, &opt);
1125
1126        if (use_index && !startup_info->have_repository) {
1127                int fallback = 0;
1128                git_config_get_bool("grep.fallbacktonoindex", &fallback);
1129                if (fallback)
1130                        use_index = 0;
1131                else
1132                        /* die the same way as if we did it at the beginning */
1133                        setup_git_directory();
1134        }
1135
1136        /*
1137         * skip a -- separator; we know it cannot be
1138         * separating revisions from pathnames if
1139         * we haven't even had any patterns yet
1140         */
1141        if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1142                argv++;
1143                argc--;
1144        }
1145
1146        /* First unrecognized non-option token */
1147        if (argc > 0 && !opt.pattern_list) {
1148                append_grep_pattern(&opt, argv[0], "command line", 0,
1149                                    GREP_PATTERN);
1150                argv++;
1151                argc--;
1152        }
1153
1154        if (show_in_pager == default_pager)
1155                show_in_pager = git_pager(1);
1156        if (show_in_pager) {
1157                opt.color = 0;
1158                opt.name_only = 1;
1159                opt.null_following_name = 1;
1160                opt.output_priv = &path_list;
1161                opt.output = append_path;
1162                string_list_append(&path_list, show_in_pager);
1163        }
1164
1165        if (!opt.pattern_list)
1166                die(_("no pattern given."));
1167        if (!opt.fixed && opt.ignore_case)
1168                opt.regflags |= REG_ICASE;
1169
1170        compile_grep_patterns(&opt);
1171
1172        /*
1173         * We have to find "--" in a separate pass, because its presence
1174         * influences how we will parse arguments that come before it.
1175         */
1176        for (i = 0; i < argc; i++) {
1177                if (!strcmp(argv[i], "--")) {
1178                        seen_dashdash = 1;
1179                        break;
1180                }
1181        }
1182
1183        /*
1184         * Resolve any rev arguments. If we have a dashdash, then everything up
1185         * to it must resolve as a rev. If not, then we stop at the first
1186         * non-rev and assume everything else is a path.
1187         */
1188        allow_revs = use_index && !untracked;
1189        for (i = 0; i < argc; i++) {
1190                const char *arg = argv[i];
1191                struct object_id oid;
1192                struct object_context oc;
1193                struct object *object;
1194
1195                if (!strcmp(arg, "--")) {
1196                        i++;
1197                        break;
1198                }
1199
1200                if (!allow_revs) {
1201                        if (seen_dashdash)
1202                                die(_("--no-index or --untracked cannot be used with revs"));
1203                        break;
1204                }
1205
1206                if (get_sha1_with_context(arg, GET_SHA1_RECORD_PATH,
1207                                          oid.hash, &oc)) {
1208                        if (seen_dashdash)
1209                                die(_("unable to resolve revision: %s"), arg);
1210                        break;
1211                }
1212
1213                object = parse_object_or_die(&oid, arg);
1214                if (!seen_dashdash)
1215                        verify_non_filename(prefix, arg);
1216                add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1217                free(oc.path);
1218        }
1219
1220        /*
1221         * Anything left over is presumed to be a path. But in the non-dashdash
1222         * "do what I mean" case, we verify and complain when that isn't true.
1223         */
1224        if (!seen_dashdash) {
1225                int j;
1226                for (j = i; j < argc; j++)
1227                        verify_filename(prefix, argv[j], j == i && allow_revs);
1228        }
1229
1230        parse_pathspec(&pathspec, 0,
1231                       PATHSPEC_PREFER_CWD |
1232                       (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1233                       prefix, argv + i);
1234        pathspec.max_depth = opt.max_depth;
1235        pathspec.recursive = 1;
1236
1237#ifndef NO_PTHREADS
1238        if (list.nr || cached || show_in_pager)
1239                num_threads = 0;
1240        else if (num_threads == 0)
1241                num_threads = GREP_NUM_THREADS_DEFAULT;
1242        else if (num_threads < 0)
1243                die(_("invalid number of threads specified (%d)"), num_threads);
1244#else
1245        if (num_threads)
1246                warning(_("no threads support, ignoring --threads"));
1247        num_threads = 0;
1248#endif
1249
1250#ifndef NO_PTHREADS
1251        if (num_threads) {
1252                if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1253                    && (opt.pre_context || opt.post_context ||
1254                        opt.file_break || opt.funcbody))
1255                        skip_first_line = 1;
1256                start_threads(&opt);
1257        }
1258#endif
1259
1260        if (recurse_submodules) {
1261                gitmodules_config();
1262                compile_submodule_options(&opt, argv + i, cached, untracked,
1263                                          opt_exclude, use_index,
1264                                          pattern_type_arg);
1265        }
1266
1267        if (show_in_pager && (cached || list.nr))
1268                die(_("--open-files-in-pager only works on the worktree"));
1269
1270        if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1271                const char *pager = path_list.items[0].string;
1272                int len = strlen(pager);
1273
1274                if (len > 4 && is_dir_sep(pager[len - 5]))
1275                        pager += len - 4;
1276
1277                if (opt.ignore_case && !strcmp("less", pager))
1278                        string_list_append(&path_list, "-I");
1279
1280                if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1281                        struct strbuf buf = STRBUF_INIT;
1282                        strbuf_addf(&buf, "+/%s%s",
1283                                        strcmp("less", pager) ? "" : "*",
1284                                        opt.pattern_list->pattern);
1285                        string_list_append(&path_list, buf.buf);
1286                        strbuf_detach(&buf, NULL);
1287                }
1288        }
1289
1290        if (recurse_submodules && (!use_index || untracked))
1291                die(_("option not supported with --recurse-submodules."));
1292
1293        if (!show_in_pager && !opt.status_only)
1294                setup_pager();
1295
1296        if (!use_index && (untracked || cached))
1297                die(_("--cached or --untracked cannot be used with --no-index."));
1298
1299        if (!use_index || untracked) {
1300                int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1301                hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1302        } else if (0 <= opt_exclude) {
1303                die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1304        } else if (!list.nr) {
1305                if (!cached)
1306                        setup_work_tree();
1307
1308                hit = grep_cache(&opt, &pathspec, cached);
1309        } else {
1310                if (cached)
1311                        die(_("both --cached and trees are given."));
1312                hit = grep_objects(&opt, &pathspec, &list);
1313        }
1314
1315        if (num_threads)
1316                hit |= wait_all();
1317        if (hit && show_in_pager)
1318                run_pager(&opt, prefix);
1319        clear_pathspec(&pathspec);
1320        free_grep_patterns(&opt);
1321        return !hit;
1322}