Andrew's git - gitweb.git/blob

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