Andrew's git / gitweb.git / builtin/grep.c
?
builtin / grep.con commit grep: enable recurse-submodules to work on <tree> objects (74ed437)
   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        if (num_threads)
  77                pthread_mutex_lock(&grep_mutex);
  78}
  79
  80static inline void grep_unlock(void)
  81{
  82        if (num_threads)
  83                pthread_mutex_unlock(&grep_mutex);
  84}
  85
  86/* Signalled when a new work_item is added to todo. */
  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        }
 293
 294        return st;
 295}
 296
 297static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
 298{
 299        void *data;
 300
 301        grep_read_lock();
 302        data = read_sha1_file(sha1, type, size);
 303        grep_read_unlock();
 304        return data;
 305}
 306
 307static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
 308                     const char *filename, int tree_name_len,
 309                     const char *path)
 310{
 311        struct strbuf pathbuf = STRBUF_INIT;
 312
 313        if (opt->relative && opt->prefix_length) {
 314                quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
 315                strbuf_insert(&pathbuf, 0, filename, tree_name_len);
 316        } else if (super_prefix) {
 317                strbuf_add(&pathbuf, filename, tree_name_len);
 318                strbuf_addstr(&pathbuf, super_prefix);
 319                strbuf_addstr(&pathbuf, filename + tree_name_len);
 320        } else {
 321                strbuf_addstr(&pathbuf, filename);
 322        }
 323
 324#ifndef NO_PTHREADS
 325        if (num_threads) {
 326                add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, path, sha1);
 327                strbuf_release(&pathbuf);
 328                return 0;
 329        } else
 330#endif
 331        {
 332                struct grep_source gs;
 333                int hit;
 334
 335                grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, path, sha1);
 336                strbuf_release(&pathbuf);
 337                hit = grep_source(opt, &gs);
 338
 339                grep_source_clear(&gs);
 340                return hit;
 341        }
 342}
 343
 344static int grep_file(struct grep_opt *opt, const char *filename)
 345{
 346        struct strbuf buf = STRBUF_INIT;
 347
 348        if (opt->relative && opt->prefix_length) {
 349                quote_path_relative(filename, opt->prefix, &buf);
 350        } else {
 351                if (super_prefix)
 352                        strbuf_addstr(&buf, super_prefix);
 353                strbuf_addstr(&buf, filename);
 354        }
 355
 356#ifndef NO_PTHREADS
 357        if (num_threads) {
 358                add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
 359                strbuf_release(&buf);
 360                return 0;
 361        } else
 362#endif
 363        {
 364                struct grep_source gs;
 365                int hit;
 366
 367                grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
 368                strbuf_release(&buf);
 369                hit = grep_source(opt, &gs);
 370
 371                grep_source_clear(&gs);
 372                return hit;
 373        }
 374}
 375
 376static void append_path(struct grep_opt *opt, const void *data, size_t len)
 377{
 378        struct string_list *path_list = opt->output_priv;
 379
 380        if (len == 1 && *(const char *)data == '\0')
 381                return;
 382        string_list_append(path_list, xstrndup(data, len));
 383}
 384
 385static void run_pager(struct grep_opt *opt, const char *prefix)
 386{
 387        struct string_list *path_list = opt->output_priv;
 388        struct child_process child = CHILD_PROCESS_INIT;
 389        int i, status;
 390
 391        for (i = 0; i < path_list->nr; i++)
 392                argv_array_push(&child.args, path_list->items[i].string);
 393        child.dir = prefix;
 394        child.use_shell = 1;
 395
 396        status = run_command(&child);
 397        if (status)
 398                exit(status);
 399}
 400
 401static void compile_submodule_options(const struct grep_opt *opt,
 402                                      const struct pathspec *pathspec,
 403                                      int cached, int untracked,
 404                                      int opt_exclude, int use_index,
 405                                      int pattern_type_arg)
 406{
 407        struct grep_pat *pattern;
 408        int i;
 409
 410        if (recurse_submodules)
 411                argv_array_push(&submodule_options, "--recurse-submodules");
 412
 413        if (cached)
 414                argv_array_push(&submodule_options, "--cached");
 415        if (!use_index)
 416                argv_array_push(&submodule_options, "--no-index");
 417        if (untracked)
 418                argv_array_push(&submodule_options, "--untracked");
 419        if (opt_exclude > 0)
 420                argv_array_push(&submodule_options, "--exclude-standard");
 421
 422        if (opt->invert)
 423                argv_array_push(&submodule_options, "-v");
 424        if (opt->ignore_case)
 425                argv_array_push(&submodule_options, "-i");
 426        if (opt->word_regexp)
 427                argv_array_push(&submodule_options, "-w");
 428        switch (opt->binary) {
 429        case GREP_BINARY_NOMATCH:
 430                argv_array_push(&submodule_options, "-I");
 431                break;
 432        case GREP_BINARY_TEXT:
 433                argv_array_push(&submodule_options, "-a");
 434                break;
 435        default:
 436                break;
 437        }
 438        if (opt->allow_textconv)
 439                argv_array_push(&submodule_options, "--textconv");
 440        if (opt->max_depth != -1)
 441                argv_array_pushf(&submodule_options, "--max-depth=%d",
 442                                 opt->max_depth);
 443        if (opt->linenum)
 444                argv_array_push(&submodule_options, "-n");
 445        if (!opt->pathname)
 446                argv_array_push(&submodule_options, "-h");
 447        if (!opt->relative)
 448                argv_array_push(&submodule_options, "--full-name");
 449        if (opt->name_only)
 450                argv_array_push(&submodule_options, "-l");
 451        if (opt->unmatch_name_only)
 452                argv_array_push(&submodule_options, "-L");
 453        if (opt->null_following_name)
 454                argv_array_push(&submodule_options, "-z");
 455        if (opt->count)
 456                argv_array_push(&submodule_options, "-c");
 457        if (opt->file_break)
 458                argv_array_push(&submodule_options, "--break");
 459        if (opt->heading)
 460                argv_array_push(&submodule_options, "--heading");
 461        if (opt->pre_context)
 462                argv_array_pushf(&submodule_options, "--before-context=%d",
 463                                 opt->pre_context);
 464        if (opt->post_context)
 465                argv_array_pushf(&submodule_options, "--after-context=%d",
 466                                 opt->post_context);
 467        if (opt->funcname)
 468                argv_array_push(&submodule_options, "-p");
 469        if (opt->funcbody)
 470                argv_array_push(&submodule_options, "-W");
 471        if (opt->all_match)
 472                argv_array_push(&submodule_options, "--all-match");
 473        if (opt->debug)
 474                argv_array_push(&submodule_options, "--debug");
 475        if (opt->status_only)
 476                argv_array_push(&submodule_options, "-q");
 477
 478        switch (pattern_type_arg) {
 479        case GREP_PATTERN_TYPE_BRE:
 480                argv_array_push(&submodule_options, "-G");
 481                break;
 482        case GREP_PATTERN_TYPE_ERE:
 483                argv_array_push(&submodule_options, "-E");
 484                break;
 485        case GREP_PATTERN_TYPE_FIXED:
 486                argv_array_push(&submodule_options, "-F");
 487                break;
 488        case GREP_PATTERN_TYPE_PCRE:
 489                argv_array_push(&submodule_options, "-P");
 490                break;
 491        case GREP_PATTERN_TYPE_UNSPECIFIED:
 492                break;
 493        }
 494
 495        for (pattern = opt->pattern_list; pattern != NULL;
 496             pattern = pattern->next) {
 497                switch (pattern->token) {
 498                case GREP_PATTERN:
 499                        argv_array_pushf(&submodule_options, "-e%s",
 500                                         pattern->pattern);
 501                        break;
 502                case GREP_AND:
 503                case GREP_OPEN_PAREN:
 504                case GREP_CLOSE_PAREN:
 505                case GREP_NOT:
 506                case GREP_OR:
 507                        argv_array_push(&submodule_options, pattern->pattern);
 508                        break;
 509                /* BODY and HEAD are not used by git-grep */
 510                case GREP_PATTERN_BODY:
 511                case GREP_PATTERN_HEAD:
 512                        break;
 513                }
 514        }
 515
 516        /*
 517         * Limit number of threads for child process to use.
 518         * This is to prevent potential fork-bomb behavior of git-grep as each
 519         * submodule process has its own thread pool.
 520         */
 521        argv_array_pushf(&submodule_options, "--threads=%d",
 522                         (num_threads + 1) / 2);
 523
 524        /* Add Pathspecs */
 525        argv_array_push(&submodule_options, "--");
 526        for (i = 0; i < pathspec->nr; i++)
 527                argv_array_push(&submodule_options,
 528                                pathspec->items[i].original);
 529}
 530
 531/*
 532 * Launch child process to grep contents of a submodule
 533 */
 534static int grep_submodule_launch(struct grep_opt *opt,
 535                                 const struct grep_source *gs)
 536{
 537        struct child_process cp = CHILD_PROCESS_INIT;
 538        int status, i;
 539        const char *end_of_base;
 540        const char *name;
 541        struct work_item *w = opt->output_priv;
 542
 543        end_of_base = strchr(gs->name, ':');
 544        if (gs->identifier && end_of_base)
 545                name = end_of_base + 1;
 546        else
 547                name = gs->name;
 548
 549        prepare_submodule_repo_env(&cp.env_array);
 550
 551        /* Add super prefix */
 552        argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
 553                         super_prefix ? super_prefix : "",
 554                         name);
 555        argv_array_push(&cp.args, "grep");
 556
 557        /*
 558         * Add basename of parent project
 559         * When performing grep on a tree object the filename is prefixed
 560         * with the object's name: 'tree-name:filename'.  In order to
 561         * provide uniformity of output we want to pass the name of the
 562         * parent project's object name to the submodule so the submodule can
 563         * prefix its output with the parent's name and not its own SHA1.
 564         */
 565        if (gs->identifier && end_of_base)
 566                argv_array_pushf(&cp.args, "--parent-basename=%.*s",
 567                                 (int) (end_of_base - gs->name),
 568                                 gs->name);
 569
 570        /* Add options */
 571        for (i = 0; i < submodule_options.argc; i++) {
 572                /*
 573                 * If there is a tree identifier for the submodule, add the
 574                 * rev after adding the submodule options but before the
 575                 * pathspecs.  To do this we listen for the '--' and insert the
 576                 * sha1 before pushing the '--' onto the child process argv
 577                 * array.
 578                 */
 579                if (gs->identifier &&
 580                    !strcmp("--", submodule_options.argv[i])) {
 581                        argv_array_push(&cp.args, sha1_to_hex(gs->identifier));
 582                }
 583
 584                argv_array_push(&cp.args, submodule_options.argv[i]);
 585        }
 586
 587        cp.git_cmd = 1;
 588        cp.dir = gs->path;
 589
 590        /*
 591         * Capture output to output buffer and check the return code from the
 592         * child process.  A '0' indicates a hit, a '1' indicates no hit and
 593         * anything else is an error.
 594         */
 595        status = capture_command(&cp, &w->out, 0);
 596        if (status && (status != 1)) {
 597                /* flush the buffer */
 598                write_or_die(1, w->out.buf, w->out.len);
 599                die("process for submodule '%s' failed with exit code: %d",
 600                    gs->name, status);
 601        }
 602
 603        /* invert the return code to make a hit equal to 1 */
 604        return !status;
 605}
 606
 607/*
 608 * Prep grep structures for a submodule grep
 609 * sha1: the sha1 of the submodule or NULL if using the working tree
 610 * filename: name of the submodule including tree name of parent
 611 * path: location of the submodule
 612 */
 613static int grep_submodule(struct grep_opt *opt, const unsigned char *sha1,
 614                          const char *filename, const char *path)
 615{
 616        if (!is_submodule_initialized(path))
 617                return 0;
 618        if (!is_submodule_populated(path))
 619                return 0;
 620
 621#ifndef NO_PTHREADS
 622        if (num_threads) {
 623                add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, sha1);
 624                return 0;
 625        } else
 626#endif
 627        {
 628                struct work_item w;
 629                int hit;
 630
 631                grep_source_init(&w.source, GREP_SOURCE_SUBMODULE,
 632                                 filename, path, sha1);
 633                strbuf_init(&w.out, 0);
 634                opt->output_priv = &w;
 635                hit = grep_submodule_launch(opt, &w.source);
 636
 637                write_or_die(1, w.out.buf, w.out.len);
 638
 639                grep_source_clear(&w.source);
 640                strbuf_release(&w.out);
 641                return hit;
 642        }
 643}
 644
 645static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
 646                      int cached)
 647{
 648        int hit = 0;
 649        int nr;
 650        struct strbuf name = STRBUF_INIT;
 651        int name_base_len = 0;
 652        if (super_prefix) {
 653                name_base_len = strlen(super_prefix);
 654                strbuf_addstr(&name, super_prefix);
 655        }
 656
 657        read_cache();
 658
 659        for (nr = 0; nr < active_nr; nr++) {
 660                const struct cache_entry *ce = active_cache[nr];
 661                strbuf_setlen(&name, name_base_len);
 662                strbuf_addstr(&name, ce->name);
 663
 664                if (S_ISREG(ce->ce_mode) &&
 665                    match_pathspec(pathspec, name.buf, name.len, 0, NULL,
 666                                   S_ISDIR(ce->ce_mode) ||
 667                                   S_ISGITLINK(ce->ce_mode))) {
 668                        /*
 669                         * If CE_VALID is on, we assume worktree file and its
 670                         * cache entry are identical, even if worktree file has
 671                         * been modified, so use cache version instead
 672                         */
 673                        if (cached || (ce->ce_flags & CE_VALID) ||
 674                            ce_skip_worktree(ce)) {
 675                                if (ce_stage(ce) || ce_intent_to_add(ce))
 676                                        continue;
 677                                hit |= grep_sha1(opt, ce->oid.hash, ce->name,
 678                                                 0, ce->name);
 679                        } else {
 680                                hit |= grep_file(opt, ce->name);
 681                        }
 682                } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
 683                           submodule_path_match(pathspec, name.buf, NULL)) {
 684                        hit |= grep_submodule(opt, NULL, ce->name, ce->name);
 685                } else {
 686                        continue;
 687                }
 688
 689                if (ce_stage(ce)) {
 690                        do {
 691                                nr++;
 692                        } while (nr < active_nr &&
 693                                 !strcmp(ce->name, active_cache[nr]->name));
 694                        nr--; /* compensate for loop control */
 695                }
 696                if (hit && opt->status_only)
 697                        break;
 698        }
 699
 700        strbuf_release(&name);
 701        return hit;
 702}
 703
 704static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
 705                     struct tree_desc *tree, struct strbuf *base, int tn_len,
 706                     int check_attr)
 707{
 708        int hit = 0;
 709        enum interesting match = entry_not_interesting;
 710        struct name_entry entry;
 711        int old_baselen = base->len;
 712        struct strbuf name = STRBUF_INIT;
 713        int name_base_len = 0;
 714        if (super_prefix) {
 715                strbuf_addstr(&name, super_prefix);
 716                name_base_len = name.len;
 717        }
 718
 719        while (tree_entry(tree, &entry)) {
 720                int te_len = tree_entry_len(&entry);
 721
 722                if (match != all_entries_interesting) {
 723                        strbuf_addstr(&name, base->buf + tn_len);
 724                        match = tree_entry_interesting(&entry, &name,
 725                                                       0, pathspec);
 726                        strbuf_setlen(&name, name_base_len);
 727
 728                        if (match == all_entries_not_interesting)
 729                                break;
 730                        if (match == entry_not_interesting)
 731                                continue;
 732                }
 733
 734                strbuf_add(base, entry.path, te_len);
 735
 736                if (S_ISREG(entry.mode)) {
 737                        hit |= grep_sha1(opt, entry.oid->hash, base->buf, tn_len,
 738                                         check_attr ? base->buf + tn_len : NULL);
 739                } else if (S_ISDIR(entry.mode)) {
 740                        enum object_type type;
 741                        struct tree_desc sub;
 742                        void *data;
 743                        unsigned long size;
 744
 745                        data = lock_and_read_sha1_file(entry.oid->hash, &type, &size);
 746                        if (!data)
 747                                die(_("unable to read tree (%s)"),
 748                                    oid_to_hex(entry.oid));
 749
 750                        strbuf_addch(base, '/');
 751                        init_tree_desc(&sub, data, size);
 752                        hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
 753                                         check_attr);
 754                        free(data);
 755                } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
 756                        hit |= grep_submodule(opt, entry.oid->hash, base->buf,
 757                                              base->buf + tn_len);
 758                }
 759
 760                strbuf_setlen(base, old_baselen);
 761
 762                if (hit && opt->status_only)
 763                        break;
 764        }
 765
 766        strbuf_release(&name);
 767        return hit;
 768}
 769
 770static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
 771                       struct object *obj, const char *name, const char *path)
 772{
 773        if (obj->type == OBJ_BLOB)
 774                return grep_sha1(opt, obj->oid.hash, name, 0, path);
 775        if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 776                struct tree_desc tree;
 777                void *data;
 778                unsigned long size;
 779                struct strbuf base;
 780                int hit, len;
 781
 782                grep_read_lock();
 783                data = read_object_with_reference(obj->oid.hash, tree_type,
 784                                                  &size, NULL);
 785                grep_read_unlock();
 786
 787                if (!data)
 788                        die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
 789
 790                /* Use parent's name as base when recursing submodules */
 791                if (recurse_submodules && parent_basename)
 792                        name = parent_basename;
 793
 794                len = name ? strlen(name) : 0;
 795                strbuf_init(&base, PATH_MAX + len + 1);
 796                if (len) {
 797                        strbuf_add(&base, name, len);
 798                        strbuf_addch(&base, ':');
 799                }
 800                init_tree_desc(&tree, data, size);
 801                hit = grep_tree(opt, pathspec, &tree, &base, base.len,
 802                                obj->type == OBJ_COMMIT);
 803                strbuf_release(&base);
 804                free(data);
 805                return hit;
 806        }
 807        die(_("unable to grep from object of type %s"), typename(obj->type));
 808}
 809
 810static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
 811                        const struct object_array *list)
 812{
 813        unsigned int i;
 814        int hit = 0;
 815        const unsigned int nr = list->nr;
 816
 817        for (i = 0; i < nr; i++) {
 818                struct object *real_obj;
 819                real_obj = deref_tag(list->objects[i].item, NULL, 0);
 820
 821                /* load the gitmodules file for this rev */
 822                if (recurse_submodules) {
 823                        submodule_free();
 824                        gitmodules_config_sha1(real_obj->oid.hash);
 825                }
 826                if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
 827                        hit = 1;
 828                        if (opt->status_only)
 829                                break;
 830                }
 831        }
 832        return hit;
 833}
 834
 835static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
 836                          int exc_std, int use_index)
 837{
 838        struct dir_struct dir;
 839        int i, hit = 0;
 840
 841        memset(&dir, 0, sizeof(dir));
 842        if (!use_index)
 843                dir.flags |= DIR_NO_GITLINKS;
 844        if (exc_std)
 845                setup_standard_excludes(&dir);
 846
 847        fill_directory(&dir, pathspec);
 848        for (i = 0; i < dir.nr; i++) {
 849                if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
 850                        continue;
 851                hit |= grep_file(opt, dir.entries[i]->name);
 852                if (hit && opt->status_only)
 853                        break;
 854        }
 855        return hit;
 856}
 857
 858static int context_callback(const struct option *opt, const char *arg,
 859                            int unset)
 860{
 861        struct grep_opt *grep_opt = opt->value;
 862        int value;
 863        const char *endp;
 864
 865        if (unset) {
 866                grep_opt->pre_context = grep_opt->post_context = 0;
 867                return 0;
 868        }
 869        value = strtol(arg, (char **)&endp, 10);
 870        if (*endp) {
 871                return error(_("switch `%c' expects a numerical value"),
 872                             opt->short_name);
 873        }
 874        grep_opt->pre_context = grep_opt->post_context = value;
 875        return 0;
 876}
 877
 878static int file_callback(const struct option *opt, const char *arg, int unset)
 879{
 880        struct grep_opt *grep_opt = opt->value;
 881        int from_stdin = !strcmp(arg, "-");
 882        FILE *patterns;
 883        int lno = 0;
 884        struct strbuf sb = STRBUF_INIT;
 885
 886        patterns = from_stdin ? stdin : fopen(arg, "r");
 887        if (!patterns)
 888                die_errno(_("cannot open '%s'"), arg);
 889        while (strbuf_getline(&sb, patterns) == 0) {
 890                /* ignore empty line like grep does */
 891                if (sb.len == 0)
 892                        continue;
 893
 894                append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
 895                                GREP_PATTERN);
 896        }
 897        if (!from_stdin)
 898                fclose(patterns);
 899        strbuf_release(&sb);
 900        return 0;
 901}
 902
 903static int not_callback(const struct option *opt, const char *arg, int unset)
 904{
 905        struct grep_opt *grep_opt = opt->value;
 906        append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 907        return 0;
 908}
 909
 910static int and_callback(const struct option *opt, const char *arg, int unset)
 911{
 912        struct grep_opt *grep_opt = opt->value;
 913        append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 914        return 0;
 915}
 916
 917static int open_callback(const struct option *opt, const char *arg, int unset)
 918{
 919        struct grep_opt *grep_opt = opt->value;
 920        append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 921        return 0;
 922}
 923
 924static int close_callback(const struct option *opt, const char *arg, int unset)
 925{
 926        struct grep_opt *grep_opt = opt->value;
 927        append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 928        return 0;
 929}
 930
 931static int pattern_callback(const struct option *opt, const char *arg,
 932                            int unset)
 933{
 934        struct grep_opt *grep_opt = opt->value;
 935        append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 936        return 0;
 937}
 938
 939int cmd_grep(int argc, const char **argv, const char *prefix)
 940{
 941        int hit = 0;
 942        int cached = 0, untracked = 0, opt_exclude = -1;
 943        int seen_dashdash = 0;
 944        int external_grep_allowed__ignored;
 945        const char *show_in_pager = NULL, *default_pager = "dummy";
 946        struct grep_opt opt;
 947        struct object_array list = OBJECT_ARRAY_INIT;
 948        struct pathspec pathspec;
 949        struct string_list path_list = STRING_LIST_INIT_NODUP;
 950        int i;
 951        int dummy;
 952        int use_index = 1;
 953        int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
 954
 955        struct option options[] = {
 956                OPT_BOOL(0, "cached", &cached,
 957                        N_("search in index instead of in the work tree")),
 958                OPT_NEGBIT(0, "no-index", &use_index,
 959                         N_("find in contents not managed by git"), 1),
 960                OPT_BOOL(0, "untracked", &untracked,
 961                        N_("search in both tracked and untracked files")),
 962                OPT_SET_INT(0, "exclude-standard", &opt_exclude,
 963                            N_("ignore files specified via '.gitignore'"), 1),
 964                OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
 965                         N_("recursivley search in each submodule")),
 966                OPT_STRING(0, "parent-basename", &parent_basename,
 967                           N_("basename"),
 968                           N_("prepend parent project's basename to output")),
 969                OPT_GROUP(""),
 970                OPT_BOOL('v', "invert-match", &opt.invert,
 971                        N_("show non-matching lines")),
 972                OPT_BOOL('i', "ignore-case", &opt.ignore_case,
 973                        N_("case insensitive matching")),
 974                OPT_BOOL('w', "word-regexp", &opt.word_regexp,
 975                        N_("match patterns only at word boundaries")),
 976                OPT_SET_INT('a', "text", &opt.binary,
 977                        N_("process binary files as text"), GREP_BINARY_TEXT),
 978                OPT_SET_INT('I', NULL, &opt.binary,
 979                        N_("don't match patterns in binary files"),
 980                        GREP_BINARY_NOMATCH),
 981                OPT_BOOL(0, "textconv", &opt.allow_textconv,
 982                         N_("process binary files with textconv filters")),
 983                { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
 984                        N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
 985                        NULL, 1 },
 986                OPT_GROUP(""),
 987                OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
 988                            N_("use extended POSIX regular expressions"),
 989                            GREP_PATTERN_TYPE_ERE),
 990                OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
 991                            N_("use basic POSIX regular expressions (default)"),
 992                            GREP_PATTERN_TYPE_BRE),
 993                OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
 994                            N_("interpret patterns as fixed strings"),
 995                            GREP_PATTERN_TYPE_FIXED),
 996                OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
 997                            N_("use Perl-compatible regular expressions"),
 998                            GREP_PATTERN_TYPE_PCRE),
 999                OPT_GROUP(""),
1000                OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1001                OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1002                OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1003                OPT_NEGBIT(0, "full-name", &opt.relative,
1004                        N_("show filenames relative to top directory"), 1),
1005                OPT_BOOL('l', "files-with-matches", &opt.name_only,
1006                        N_("show only filenames instead of matching lines")),
1007                OPT_BOOL(0, "name-only", &opt.name_only,
1008                        N_("synonym for --files-with-matches")),
1009                OPT_BOOL('L', "files-without-match",
1010                        &opt.unmatch_name_only,
1011                        N_("show only the names of files without match")),
1012                OPT_BOOL('z', "null", &opt.null_following_name,
1013                        N_("print NUL after filenames")),
1014                OPT_BOOL('c', "count", &opt.count,
1015                        N_("show the number of matches instead of matching lines")),
1016                OPT__COLOR(&opt.color, N_("highlight matches")),
1017                OPT_BOOL(0, "break", &opt.file_break,
1018                        N_("print empty line between matches from different files")),
1019                OPT_BOOL(0, "heading", &opt.heading,
1020                        N_("show filename only once above matches from same file")),
1021                OPT_GROUP(""),
1022                OPT_CALLBACK('C', "context", &opt, N_("n"),
1023                        N_("show <n> context lines before and after matches"),
1024                        context_callback),
1025                OPT_INTEGER('B', "before-context", &opt.pre_context,
1026                        N_("show <n> context lines before matches")),
1027                OPT_INTEGER('A', "after-context", &opt.post_context,
1028                        N_("show <n> context lines after matches")),
1029                OPT_INTEGER(0, "threads", &num_threads,
1030                        N_("use <n> worker threads")),
1031                OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1032                        context_callback),
1033                OPT_BOOL('p', "show-function", &opt.funcname,
1034                        N_("show a line with the function name before matches")),
1035                OPT_BOOL('W', "function-context", &opt.funcbody,
1036                        N_("show the surrounding function")),
1037                OPT_GROUP(""),
1038                OPT_CALLBACK('f', NULL, &opt, N_("file"),
1039                        N_("read patterns from file"), file_callback),
1040                { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1041                        N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1042                { OPTION_CALLBACK, 0, "and", &opt, NULL,
1043                  N_("combine patterns specified with -e"),
1044                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1045                OPT_BOOL(0, "or", &dummy, ""),
1046                { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1047                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1048                { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1049                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1050                  open_callback },
1051                { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1052                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1053                  close_callback },
1054                OPT__QUIET(&opt.status_only,
1055                           N_("indicate hit with exit status without output")),
1056                OPT_BOOL(0, "all-match", &opt.all_match,
1057                        N_("show only matches from files that match all patterns")),
1058                { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1059                  N_("show parse tree for grep expression"),
1060                  PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1061                OPT_GROUP(""),
1062                { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1063                        N_("pager"), N_("show matching files in the pager"),
1064                        PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1065                OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1066                         N_("allow calling of grep(1) (ignored by this build)")),
1067                OPT_END()
1068        };
1069
1070        init_grep_defaults();
1071        git_config(grep_cmd_config, NULL);
1072        grep_init(&opt, prefix);
1073        super_prefix = get_super_prefix();
1074
1075        /*
1076         * If there is no -- then the paths must exist in the working
1077         * tree.  If there is no explicit pattern specified with -e or
1078         * -f, we take the first unrecognized non option to be the
1079         * pattern, but then what follows it must be zero or more
1080         * valid refs up to the -- (if exists), and then existing
1081         * paths.  If there is an explicit pattern, then the first
1082         * unrecognized non option is the beginning of the refs list
1083         * that continues up to the -- (if exists), and then paths.
1084         */
1085        argc = parse_options(argc, argv, prefix, options, grep_usage,
1086                             PARSE_OPT_KEEP_DASHDASH |
1087                             PARSE_OPT_STOP_AT_NON_OPTION);
1088        grep_commit_pattern_type(pattern_type_arg, &opt);
1089
1090        if (use_index && !startup_info->have_repository) {
1091                int fallback = 0;
1092                git_config_get_bool("grep.fallbacktonoindex", &fallback);
1093                if (fallback)
1094                        use_index = 0;
1095                else
1096                        /* die the same way as if we did it at the beginning */
1097                        setup_git_directory();
1098        }
1099
1100        /*
1101         * skip a -- separator; we know it cannot be
1102         * separating revisions from pathnames if
1103         * we haven't even had any patterns yet
1104         */
1105        if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1106                argv++;
1107                argc--;
1108        }
1109
1110        /* First unrecognized non-option token */
1111        if (argc > 0 && !opt.pattern_list) {
1112                append_grep_pattern(&opt, argv[0], "command line", 0,
1113                                    GREP_PATTERN);
1114                argv++;
1115                argc--;
1116        }
1117
1118        if (show_in_pager == default_pager)
1119                show_in_pager = git_pager(1);
1120        if (show_in_pager) {
1121                opt.color = 0;
1122                opt.name_only = 1;
1123                opt.null_following_name = 1;
1124                opt.output_priv = &path_list;
1125                opt.output = append_path;
1126                string_list_append(&path_list, show_in_pager);
1127        }
1128
1129        if (!opt.pattern_list)
1130                die(_("no pattern given."));
1131        if (!opt.fixed && opt.ignore_case)
1132                opt.regflags |= REG_ICASE;
1133
1134        compile_grep_patterns(&opt);
1135
1136        /* Check revs and then paths */
1137        for (i = 0; i < argc; i++) {
1138                const char *arg = argv[i];
1139                unsigned char sha1[20];
1140                struct object_context oc;
1141                /* Is it a rev? */
1142                if (!get_sha1_with_context(arg, 0, sha1, &oc)) {
1143                        struct object *object = parse_object_or_die(sha1, arg);
1144                        if (!seen_dashdash)
1145                                verify_non_filename(prefix, arg);
1146                        add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1147                        continue;
1148                }
1149                if (!strcmp(arg, "--")) {
1150                        i++;
1151                        seen_dashdash = 1;
1152                }
1153                break;
1154        }
1155
1156#ifndef NO_PTHREADS
1157        if (list.nr || cached || show_in_pager)
1158                num_threads = 0;
1159        else if (num_threads == 0)
1160                num_threads = GREP_NUM_THREADS_DEFAULT;
1161        else if (num_threads < 0)
1162                die(_("invalid number of threads specified (%d)"), num_threads);
1163#else
1164        num_threads = 0;
1165#endif
1166
1167#ifndef NO_PTHREADS
1168        if (num_threads) {
1169                if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1170                    && (opt.pre_context || opt.post_context ||
1171                        opt.file_break || opt.funcbody))
1172                        skip_first_line = 1;
1173                start_threads(&opt);
1174        }
1175#endif
1176
1177        /* The rest are paths */
1178        if (!seen_dashdash) {
1179                int j;
1180                for (j = i; j < argc; j++)
1181                        verify_filename(prefix, argv[j], j == i);
1182        }
1183
1184        parse_pathspec(&pathspec, 0,
1185                       PATHSPEC_PREFER_CWD |
1186                       (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1187                       prefix, argv + i);
1188        pathspec.max_depth = opt.max_depth;
1189        pathspec.recursive = 1;
1190
1191        if (recurse_submodules) {
1192                gitmodules_config();
1193                compile_submodule_options(&opt, &pathspec, cached, untracked,
1194                                          opt_exclude, use_index,
1195                                          pattern_type_arg);
1196        }
1197
1198        if (show_in_pager && (cached || list.nr))
1199                die(_("--open-files-in-pager only works on the worktree"));
1200
1201        if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1202                const char *pager = path_list.items[0].string;
1203                int len = strlen(pager);
1204
1205                if (len > 4 && is_dir_sep(pager[len - 5]))
1206                        pager += len - 4;
1207
1208                if (opt.ignore_case && !strcmp("less", pager))
1209                        string_list_append(&path_list, "-I");
1210
1211                if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1212                        struct strbuf buf = STRBUF_INIT;
1213                        strbuf_addf(&buf, "+/%s%s",
1214                                        strcmp("less", pager) ? "" : "*",
1215                                        opt.pattern_list->pattern);
1216                        string_list_append(&path_list, buf.buf);
1217                        strbuf_detach(&buf, NULL);
1218                }
1219        }
1220
1221        if (recurse_submodules && (!use_index || untracked))
1222                die(_("option not supported with --recurse-submodules."));
1223
1224        if (!show_in_pager && !opt.status_only)
1225                setup_pager();
1226
1227        if (!use_index && (untracked || cached))
1228                die(_("--cached or --untracked cannot be used with --no-index."));
1229
1230        if (!use_index || untracked) {
1231                int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1232                if (list.nr)
1233                        die(_("--no-index or --untracked cannot be used with revs."));
1234                hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1235        } else if (0 <= opt_exclude) {
1236                die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1237        } else if (!list.nr) {
1238                if (!cached)
1239                        setup_work_tree();
1240
1241                hit = grep_cache(&opt, &pathspec, cached);
1242        } else {
1243                if (cached)
1244                        die(_("both --cached and trees are given."));
1245                hit = grep_objects(&opt, &pathspec, &list);
1246        }
1247
1248        if (num_threads)
1249                hit |= wait_all();
1250        if (hit && show_in_pager)
1251                run_pager(&opt, prefix);
1252        free_grep_patterns(&opt);
1253        return !hit;
1254}