Andrew's git / gitweb.git / builtin-grep.c
?
builtin-grep.con commit grep --no-index: allow use of "git grep" outside a git repository (3081623)
   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 "userdiff.h"
  15#include "grep.h"
  16#include "quote.h"
  17#include "dir.h"
  18
  19static char const * const grep_usage[] = {
  20        "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
  21        NULL
  22};
  23
  24static int grep_config(const char *var, const char *value, void *cb)
  25{
  26        struct grep_opt *opt = cb;
  27
  28        switch (userdiff_config(var, value)) {
  29        case 0: break;
  30        case -1: return -1;
  31        default: return 0;
  32        }
  33
  34        if (!strcmp(var, "color.grep")) {
  35                opt->color = git_config_colorbool(var, value, -1);
  36                return 0;
  37        }
  38        if (!strcmp(var, "color.grep.match")) {
  39                if (!value)
  40                        return config_error_nonbool(var);
  41                color_parse(value, var, opt->color_match);
  42                return 0;
  43        }
  44        return git_color_default_config(var, value, cb);
  45}
  46
  47/*
  48 * Return non-zero if max_depth is negative or path has no more then max_depth
  49 * slashes.
  50 */
  51static int accept_subdir(const char *path, int max_depth)
  52{
  53        if (max_depth < 0)
  54                return 1;
  55
  56        while ((path = strchr(path, '/')) != NULL) {
  57                max_depth--;
  58                if (max_depth < 0)
  59                        return 0;
  60                path++;
  61        }
  62        return 1;
  63}
  64
  65/*
  66 * Return non-zero if name is a subdirectory of match and is not too deep.
  67 */
  68static int is_subdir(const char *name, int namelen,
  69                const char *match, int matchlen, int max_depth)
  70{
  71        if (matchlen > namelen || strncmp(name, match, matchlen))
  72                return 0;
  73
  74        if (name[matchlen] == '\0') /* exact match */
  75                return 1;
  76
  77        if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
  78                return accept_subdir(name + matchlen + 1, max_depth);
  79
  80        return 0;
  81}
  82
  83/*
  84 * git grep pathspecs are somewhat different from diff-tree pathspecs;
  85 * pathname wildcards are allowed.
  86 */
  87static int pathspec_matches(const char **paths, const char *name, int max_depth)
  88{
  89        int namelen, i;
  90        if (!paths || !*paths)
  91                return accept_subdir(name, max_depth);
  92        namelen = strlen(name);
  93        for (i = 0; paths[i]; i++) {
  94                const char *match = paths[i];
  95                int matchlen = strlen(match);
  96                const char *cp, *meta;
  97
  98                if (is_subdir(name, namelen, match, matchlen, max_depth))
  99                        return 1;
 100                if (!fnmatch(match, name, 0))
 101                        return 1;
 102                if (name[namelen-1] != '/')
 103                        continue;
 104
 105                /* We are being asked if the directory ("name") is worth
 106                 * descending into.
 107                 *
 108                 * Find the longest leading directory name that does
 109                 * not have metacharacter in the pathspec; the name
 110                 * we are looking at must overlap with that directory.
 111                 */
 112                for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
 113                        char ch = *cp;
 114                        if (ch == '*' || ch == '[' || ch == '?') {
 115                                meta = cp;
 116                                break;
 117                        }
 118                }
 119                if (!meta)
 120                        meta = cp; /* fully literal */
 121
 122                if (namelen <= meta - match) {
 123                        /* Looking at "Documentation/" and
 124                         * the pattern says "Documentation/howto/", or
 125                         * "Documentation/diff*.txt".  The name we
 126                         * have should match prefix.
 127                         */
 128                        if (!memcmp(match, name, namelen))
 129                                return 1;
 130                        continue;
 131                }
 132
 133                if (meta - match < namelen) {
 134                        /* Looking at "Documentation/howto/" and
 135                         * the pattern says "Documentation/h*";
 136                         * match up to "Do.../h"; this avoids descending
 137                         * into "Documentation/technical/".
 138                         */
 139                        if (!memcmp(match, name, meta - match))
 140                                return 1;
 141                        continue;
 142                }
 143        }
 144        return 0;
 145}
 146
 147static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
 148{
 149        unsigned long size;
 150        char *data;
 151        enum object_type type;
 152        int hit;
 153        struct strbuf pathbuf = STRBUF_INIT;
 154
 155        data = read_sha1_file(sha1, &type, &size);
 156        if (!data) {
 157                error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 158                return 0;
 159        }
 160        if (opt->relative && opt->prefix_length) {
 161                quote_path_relative(name + tree_name_len, -1, &pathbuf, opt->prefix);
 162                strbuf_insert(&pathbuf, 0, name, tree_name_len);
 163                name = pathbuf.buf;
 164        }
 165        hit = grep_buffer(opt, name, data, size);
 166        strbuf_release(&pathbuf);
 167        free(data);
 168        return hit;
 169}
 170
 171static int grep_file(struct grep_opt *opt, const char *filename)
 172{
 173        struct stat st;
 174        int i;
 175        char *data;
 176        size_t sz;
 177        struct strbuf buf = STRBUF_INIT;
 178
 179        if (lstat(filename, &st) < 0) {
 180        err_ret:
 181                if (errno != ENOENT)
 182                        error("'%s': %s", filename, strerror(errno));
 183                return 0;
 184        }
 185        if (!st.st_size)
 186                return 0; /* empty file -- no grep hit */
 187        if (!S_ISREG(st.st_mode))
 188                return 0;
 189        sz = xsize_t(st.st_size);
 190        i = open(filename, O_RDONLY);
 191        if (i < 0)
 192                goto err_ret;
 193        data = xmalloc(sz + 1);
 194        if (st.st_size != read_in_full(i, data, sz)) {
 195                error("'%s': short read %s", filename, strerror(errno));
 196                close(i);
 197                free(data);
 198                return 0;
 199        }
 200        close(i);
 201        if (opt->relative && opt->prefix_length)
 202                filename = quote_path_relative(filename, -1, &buf, opt->prefix);
 203        i = grep_buffer(opt, filename, data, sz);
 204        strbuf_release(&buf);
 205        free(data);
 206        return i;
 207}
 208
 209static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 210{
 211        int hit = 0;
 212        int nr;
 213        read_cache();
 214
 215        for (nr = 0; nr < active_nr; nr++) {
 216                struct cache_entry *ce = active_cache[nr];
 217                if (!S_ISREG(ce->ce_mode))
 218                        continue;
 219                if (!pathspec_matches(paths, ce->name, opt->max_depth))
 220                        continue;
 221                /*
 222                 * If CE_VALID is on, we assume worktree file and its cache entry
 223                 * are identical, even if worktree file has been modified, so use
 224                 * cache version instead
 225                 */
 226                if (cached || (ce->ce_flags & CE_VALID)) {
 227                        if (ce_stage(ce))
 228                                continue;
 229                        hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
 230                }
 231                else
 232                        hit |= grep_file(opt, ce->name);
 233                if (ce_stage(ce)) {
 234                        do {
 235                                nr++;
 236                        } while (nr < active_nr &&
 237                                 !strcmp(ce->name, active_cache[nr]->name));
 238                        nr--; /* compensate for loop control */
 239                }
 240        }
 241        free_grep_patterns(opt);
 242        return hit;
 243}
 244
 245static int grep_tree(struct grep_opt *opt, const char **paths,
 246                     struct tree_desc *tree,
 247                     const char *tree_name, const char *base)
 248{
 249        int len;
 250        int hit = 0;
 251        struct name_entry entry;
 252        char *down;
 253        int tn_len = strlen(tree_name);
 254        struct strbuf pathbuf;
 255
 256        strbuf_init(&pathbuf, PATH_MAX + tn_len);
 257
 258        if (tn_len) {
 259                strbuf_add(&pathbuf, tree_name, tn_len);
 260                strbuf_addch(&pathbuf, ':');
 261                tn_len = pathbuf.len;
 262        }
 263        strbuf_addstr(&pathbuf, base);
 264        len = pathbuf.len;
 265
 266        while (tree_entry(tree, &entry)) {
 267                int te_len = tree_entry_len(entry.path, entry.sha1);
 268                pathbuf.len = len;
 269                strbuf_add(&pathbuf, entry.path, te_len);
 270
 271                if (S_ISDIR(entry.mode))
 272                        /* Match "abc/" against pathspec to
 273                         * decide if we want to descend into "abc"
 274                         * directory.
 275                         */
 276                        strbuf_addch(&pathbuf, '/');
 277
 278                down = pathbuf.buf + tn_len;
 279                if (!pathspec_matches(paths, down, opt->max_depth))
 280                        ;
 281                else if (S_ISREG(entry.mode))
 282                        hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
 283                else if (S_ISDIR(entry.mode)) {
 284                        enum object_type type;
 285                        struct tree_desc sub;
 286                        void *data;
 287                        unsigned long size;
 288
 289                        data = read_sha1_file(entry.sha1, &type, &size);
 290                        if (!data)
 291                                die("unable to read tree (%s)",
 292                                    sha1_to_hex(entry.sha1));
 293                        init_tree_desc(&sub, data, size);
 294                        hit |= grep_tree(opt, paths, &sub, tree_name, down);
 295                        free(data);
 296                }
 297        }
 298        strbuf_release(&pathbuf);
 299        return hit;
 300}
 301
 302static int grep_object(struct grep_opt *opt, const char **paths,
 303                       struct object *obj, const char *name)
 304{
 305        if (obj->type == OBJ_BLOB)
 306                return grep_sha1(opt, obj->sha1, name, 0);
 307        if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
 308                struct tree_desc tree;
 309                void *data;
 310                unsigned long size;
 311                int hit;
 312                data = read_object_with_reference(obj->sha1, tree_type,
 313                                                  &size, NULL);
 314                if (!data)
 315                        die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 316                init_tree_desc(&tree, data, size);
 317                hit = grep_tree(opt, paths, &tree, name, "");
 318                free(data);
 319                return hit;
 320        }
 321        die("unable to grep from object of type %s", typename(obj->type));
 322}
 323
 324static int grep_directory(struct grep_opt *opt, const char **paths)
 325{
 326        struct dir_struct dir;
 327        int i, hit = 0;
 328
 329        memset(&dir, 0, sizeof(dir));
 330        setup_standard_excludes(&dir);
 331
 332        fill_directory(&dir, paths);
 333        for (i = 0; i < dir.nr; i++)
 334                hit |= grep_file(opt, dir.entries[i]->name);
 335        free_grep_patterns(opt);
 336        return hit;
 337}
 338
 339static int context_callback(const struct option *opt, const char *arg,
 340                            int unset)
 341{
 342        struct grep_opt *grep_opt = opt->value;
 343        int value;
 344        const char *endp;
 345
 346        if (unset) {
 347                grep_opt->pre_context = grep_opt->post_context = 0;
 348                return 0;
 349        }
 350        value = strtol(arg, (char **)&endp, 10);
 351        if (*endp) {
 352                return error("switch `%c' expects a numerical value",
 353                             opt->short_name);
 354        }
 355        grep_opt->pre_context = grep_opt->post_context = value;
 356        return 0;
 357}
 358
 359static int file_callback(const struct option *opt, const char *arg, int unset)
 360{
 361        struct grep_opt *grep_opt = opt->value;
 362        FILE *patterns;
 363        int lno = 0;
 364        struct strbuf sb = STRBUF_INIT;
 365
 366        patterns = fopen(arg, "r");
 367        if (!patterns)
 368                die_errno("cannot open '%s'", arg);
 369        while (strbuf_getline(&sb, patterns, '\n') == 0) {
 370                /* ignore empty line like grep does */
 371                if (sb.len == 0)
 372                        continue;
 373                append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg,
 374                                    ++lno, GREP_PATTERN);
 375        }
 376        fclose(patterns);
 377        strbuf_release(&sb);
 378        return 0;
 379}
 380
 381static int not_callback(const struct option *opt, const char *arg, int unset)
 382{
 383        struct grep_opt *grep_opt = opt->value;
 384        append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
 385        return 0;
 386}
 387
 388static int and_callback(const struct option *opt, const char *arg, int unset)
 389{
 390        struct grep_opt *grep_opt = opt->value;
 391        append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
 392        return 0;
 393}
 394
 395static int open_callback(const struct option *opt, const char *arg, int unset)
 396{
 397        struct grep_opt *grep_opt = opt->value;
 398        append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
 399        return 0;
 400}
 401
 402static int close_callback(const struct option *opt, const char *arg, int unset)
 403{
 404        struct grep_opt *grep_opt = opt->value;
 405        append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
 406        return 0;
 407}
 408
 409static int pattern_callback(const struct option *opt, const char *arg,
 410                            int unset)
 411{
 412        struct grep_opt *grep_opt = opt->value;
 413        append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
 414        return 0;
 415}
 416
 417static int help_callback(const struct option *opt, const char *arg, int unset)
 418{
 419        return -1;
 420}
 421
 422int cmd_grep(int argc, const char **argv, const char *prefix)
 423{
 424        int hit = 0;
 425        int cached = 0;
 426        int seen_dashdash = 0;
 427        int external_grep_allowed__ignored;
 428        struct grep_opt opt;
 429        struct object_array list = { 0, 0, NULL };
 430        const char **paths = NULL;
 431        int i;
 432        int dummy;
 433        int nongit = 0, use_index = 1;
 434        struct option options[] = {
 435                OPT_BOOLEAN(0, "cached", &cached,
 436                        "search in index instead of in the work tree"),
 437                OPT_BOOLEAN(0, "index", &use_index,
 438                        "--no-index finds in contents not managed by git"),
 439                OPT_GROUP(""),
 440                OPT_BOOLEAN('v', "invert-match", &opt.invert,
 441                        "show non-matching lines"),
 442                OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
 443                        "case insensitive matching"),
 444                OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
 445                        "match patterns only at word boundaries"),
 446                OPT_SET_INT('a', "text", &opt.binary,
 447                        "process binary files as text", GREP_BINARY_TEXT),
 448                OPT_SET_INT('I', NULL, &opt.binary,
 449                        "don't match patterns in binary files",
 450                        GREP_BINARY_NOMATCH),
 451                { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
 452                        "descend at most <depth> levels", PARSE_OPT_NONEG,
 453                        NULL, 1 },
 454                OPT_GROUP(""),
 455                OPT_BIT('E', "extended-regexp", &opt.regflags,
 456                        "use extended POSIX regular expressions", REG_EXTENDED),
 457                OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
 458                        "use basic POSIX regular expressions (default)",
 459                        REG_EXTENDED),
 460                OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
 461                        "interpret patterns as fixed strings"),
 462                OPT_GROUP(""),
 463                OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
 464                OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
 465                OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
 466                OPT_NEGBIT(0, "full-name", &opt.relative,
 467                        "show filenames relative to top directory", 1),
 468                OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
 469                        "show only filenames instead of matching lines"),
 470                OPT_BOOLEAN(0, "name-only", &opt.name_only,
 471                        "synonym for --files-with-matches"),
 472                OPT_BOOLEAN('L', "files-without-match",
 473                        &opt.unmatch_name_only,
 474                        "show only the names of files without match"),
 475                OPT_BOOLEAN('z', "null", &opt.null_following_name,
 476                        "print NUL after filenames"),
 477                OPT_BOOLEAN('c', "count", &opt.count,
 478                        "show the number of matches instead of matching lines"),
 479                OPT_SET_INT(0, "color", &opt.color, "highlight matches", 1),
 480                OPT_GROUP(""),
 481                OPT_CALLBACK('C', NULL, &opt, "n",
 482                        "show <n> context lines before and after matches",
 483                        context_callback),
 484                OPT_INTEGER('B', NULL, &opt.pre_context,
 485                        "show <n> context lines before matches"),
 486                OPT_INTEGER('A', NULL, &opt.post_context,
 487                        "show <n> context lines after matches"),
 488                OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
 489                        context_callback),
 490                OPT_BOOLEAN('p', "show-function", &opt.funcname,
 491                        "show a line with the function name before matches"),
 492                OPT_GROUP(""),
 493                OPT_CALLBACK('f', NULL, &opt, "file",
 494                        "read patterns from file", file_callback),
 495                { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
 496                        "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
 497                { OPTION_CALLBACK, 0, "and", &opt, NULL,
 498                  "combine patterns specified with -e",
 499                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
 500                OPT_BOOLEAN(0, "or", &dummy, ""),
 501                { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
 502                  PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
 503                { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
 504                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 505                  open_callback },
 506                { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
 507                  PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
 508                  close_callback },
 509                OPT_BOOLEAN(0, "all-match", &opt.all_match,
 510                        "show only matches from files that match all patterns"),
 511                OPT_GROUP(""),
 512                OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
 513                            "allow calling of grep(1) (ignored by this build)"),
 514                { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
 515                  PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
 516                OPT_END()
 517        };
 518
 519        prefix = setup_git_directory_gently(&nongit);
 520
 521        /*
 522         * 'git grep -h', unlike 'git grep -h <pattern>', is a request
 523         * to show usage information and exit.
 524         */
 525        if (argc == 2 && !strcmp(argv[1], "-h"))
 526                usage_with_options(grep_usage, options);
 527
 528        memset(&opt, 0, sizeof(opt));
 529        opt.prefix = prefix;
 530        opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 531        opt.relative = 1;
 532        opt.pathname = 1;
 533        opt.pattern_tail = &opt.pattern_list;
 534        opt.regflags = REG_NEWLINE;
 535        opt.max_depth = -1;
 536
 537        strcpy(opt.color_match, GIT_COLOR_RED GIT_COLOR_BOLD);
 538        opt.color = -1;
 539        git_config(grep_config, &opt);
 540        if (opt.color == -1)
 541                opt.color = git_use_color_default;
 542
 543        /*
 544         * If there is no -- then the paths must exist in the working
 545         * tree.  If there is no explicit pattern specified with -e or
 546         * -f, we take the first unrecognized non option to be the
 547         * pattern, but then what follows it must be zero or more
 548         * valid refs up to the -- (if exists), and then existing
 549         * paths.  If there is an explicit pattern, then the first
 550         * unrecognized non option is the beginning of the refs list
 551         * that continues up to the -- (if exists), and then paths.
 552         */
 553        argc = parse_options(argc, argv, prefix, options, grep_usage,
 554                             PARSE_OPT_KEEP_DASHDASH |
 555                             PARSE_OPT_STOP_AT_NON_OPTION |
 556                             PARSE_OPT_NO_INTERNAL_HELP);
 557
 558        if (use_index && nongit)
 559                /* die the same way as if we did it at the beginning */
 560                setup_git_directory();
 561
 562        /* First unrecognized non-option token */
 563        if (argc > 0 && !opt.pattern_list) {
 564                append_grep_pattern(&opt, argv[0], "command line", 0,
 565                                    GREP_PATTERN);
 566                argv++;
 567                argc--;
 568        }
 569
 570        if (!opt.pattern_list)
 571                die("no pattern given.");
 572        if (!opt.fixed && opt.ignore_case)
 573                opt.regflags |= REG_ICASE;
 574        if ((opt.regflags != REG_NEWLINE) && opt.fixed)
 575                die("cannot mix --fixed-strings and regexp");
 576        compile_grep_patterns(&opt);
 577
 578        /* Check revs and then paths */
 579        for (i = 0; i < argc; i++) {
 580                const char *arg = argv[i];
 581                unsigned char sha1[20];
 582                /* Is it a rev? */
 583                if (!get_sha1(arg, sha1)) {
 584                        struct object *object = parse_object(sha1);
 585                        if (!object)
 586                                die("bad object %s", arg);
 587                        add_object_array(object, arg, &list);
 588                        continue;
 589                }
 590                if (!strcmp(arg, "--")) {
 591                        i++;
 592                        seen_dashdash = 1;
 593                }
 594                break;
 595        }
 596
 597        /* The rest are paths */
 598        if (!seen_dashdash) {
 599                int j;
 600                for (j = i; j < argc; j++)
 601                        verify_filename(prefix, argv[j]);
 602        }
 603
 604        if (i < argc)
 605                paths = get_pathspec(prefix, argv + i);
 606        else if (prefix) {
 607                paths = xcalloc(2, sizeof(const char *));
 608                paths[0] = prefix;
 609                paths[1] = NULL;
 610        }
 611
 612        if (!use_index) {
 613                if (cached)
 614                        die("--cached cannot be used with --no-index.");
 615                if (list.nr)
 616                        die("--no-index cannot be used with revs.");
 617                return !grep_directory(&opt, paths);
 618        }
 619
 620        if (!list.nr) {
 621                if (!cached)
 622                        setup_work_tree();
 623                return !grep_cache(&opt, paths, cached);
 624        }
 625
 626        if (cached)
 627                die("both --cached and trees are given.");
 628
 629        for (i = 0; i < list.nr; i++) {
 630                struct object *real_obj;
 631                real_obj = deref_tag(list.objects[i].item, NULL, 0);
 632                if (grep_object(&opt, paths, real_obj, list.objects[i].name))
 633                        hit = 1;
 634        }
 635        free_grep_patterns(&opt);
 636        return !hit;
 637}