builtin-grep.c

   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 <regex.h>
  14 #include <fnmatch.h>
  15
  16 /*
  17  * git grep pathspecs are somewhat different from diff-tree pathspecs;
  18  * pathname wildcards are allowed.
  19  */
  20 static int pathspec_matches(const char **paths, const char *name)
  21 {
  22         int namelen, i;
  23         if (!paths || !*paths)
  24                 return 1;
  25         namelen = strlen(name);
  26         for (i = 0; paths[i]; i++) {
  27                 const char *match = paths[i];
  28                 int matchlen = strlen(match);
  29                 const char *cp, *meta;
  30
  31                 if ((matchlen <= namelen) &&
  32                     !strncmp(name, match, matchlen) &&
  33                     (match[matchlen-1] == '/' ||
  34                      name[matchlen] == '\0' || name[matchlen] == '/'))
  35                         return 1;
  36                 if (!fnmatch(match, name, 0))
  37                         return 1;
  38                 if (name[namelen-1] != '/')
  39                         continue;
  40
  41                 /* We are being asked if the directory ("name") is worth
  42                  * descending into.
  43                  *
  44                  * Find the longest leading directory name that does
  45                  * not have metacharacter in the pathspec; the name
  46                  * we are looking at must overlap with that directory.
  47                  */
  48                 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
  49                         char ch = *cp;
  50                         if (ch == '*' || ch == '[' || ch == '?') {
  51                                 meta = cp;
  52                                 break;
  53                         }
  54                 }
  55                 if (!meta)
  56                         meta = cp; /* fully literal */
  57
  58                 if (namelen <= meta - match) {
  59                         /* Looking at "Documentation/" and
  60                          * the pattern says "Documentation/howto/", or
  61                          * "Documentation/diff*.txt".  The name we
  62                          * have should match prefix.
  63                          */
  64                         if (!memcmp(match, name, namelen))
  65                                 return 1;
  66                         continue;
  67                 }
  68
  69                 if (meta - match < namelen) {
  70                         /* Looking at "Documentation/howto/" and
  71                          * the pattern says "Documentation/h*";
  72                          * match up to "Do.../h"; this avoids descending
  73                          * into "Documentation/technical/".
  74                          */
  75                         if (!memcmp(match, name, meta - match))
  76                                 return 1;
  77                         continue;
  78                 }
  79         }
  80         return 0;
  81 }
  82
  83 struct grep_pat {
  84         struct grep_pat *next;
  85         const char *pattern;
  86         regex_t regexp;
  87 };
  88
  89 struct grep_opt {
  90         struct grep_pat *pattern_list;
  91         struct grep_pat **pattern_tail;
  92         regex_t regexp;
  93         unsigned linenum:1;
  94         unsigned invert:1;
  95         unsigned name_only:1;
  96         unsigned unmatch_name_only:1;
  97         unsigned count:1;
  98         unsigned word_regexp:1;
  99 #define GREP_BINARY_DEFAULT     0
 100 #define GREP_BINARY_NOMATCH     1
 101 #define GREP_BINARY_TEXT        2
 102         unsigned binary:2;
 103         int regflags;
 104         unsigned pre_context;
 105         unsigned post_context;
 106 };
 107
 108 static void add_pattern(struct grep_opt *opt, const char *pat)
 109 {
 110         struct grep_pat *p = xcalloc(1, sizeof(*p));
 111         p->pattern = pat;
 112         *opt->pattern_tail = p;
 113         opt->pattern_tail = &p->next;
 114         p->next = NULL;
 115 }
 116
 117 static void compile_patterns(struct grep_opt *opt)
 118 {
 119         struct grep_pat *p;
 120         for (p = opt->pattern_list; p; p = p->next) {
 121                 int err = regcomp(&p->regexp, p->pattern, opt->regflags);
 122                 if (err) {
 123                         char errbuf[1024];
 124                         regerror(err, &p->regexp, errbuf, 1024);
 125                         regfree(&p->regexp);
 126                         die("'%s': %s", p->pattern, errbuf);
 127                 }
 128         }
 129 }
 130
 131 static char *end_of_line(char *cp, unsigned long *left)
 132 {
 133         unsigned long l = *left;
 134         while (l && *cp != '\n') {
 135                 l--;
 136                 cp++;
 137         }
 138         *left = l;
 139         return cp;
 140 }
 141
 142 static int word_char(char ch)
 143 {
 144         return isalnum(ch) || ch == '_';
 145 }
 146
 147 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 148                       const char *name, unsigned lno, char sign)
 149 {
 150         printf("%s%c", name, sign);
 151         if (opt->linenum)
 152                 printf("%d%c", lno, sign);
 153         printf("%.*s\n", (int)(eol-bol), bol);
 154 }
 155
 156 /*
 157  * NEEDSWORK: share code with diff.c
 158  */
 159 #define FIRST_FEW_BYTES 8000
 160 static int buffer_is_binary(const char *ptr, unsigned long size)
 161 {
 162         if (FIRST_FEW_BYTES < size)
 163                 size = FIRST_FEW_BYTES;
 164         if (memchr(ptr, 0, size))
 165                 return 1;
 166         return 0;
 167 }
 168
 169 static int grep_buffer(struct grep_opt *opt, const char *name,
 170                        char *buf, unsigned long size)
 171 {
 172         char *bol = buf;
 173         unsigned long left = size;
 174         unsigned lno = 1;
 175         struct pre_context_line {
 176                 char *bol;
 177                 char *eol;
 178         } *prev = NULL, *pcl;
 179         unsigned last_hit = 0;
 180         unsigned last_shown = 0;
 181         int binary_match_only = 0;
 182         const char *hunk_mark = "";
 183         unsigned count = 0;
 184
 185         if (buffer_is_binary(buf, size)) {
 186                 switch (opt->binary) {
 187                 case GREP_BINARY_DEFAULT:
 188                         binary_match_only = 1;
 189                         break;
 190                 case GREP_BINARY_NOMATCH:
 191                         return 0; /* Assume unmatch */
 192                         break;
 193                 default:
 194                         break;
 195                 }
 196         }
 197
 198         if (opt->pre_context)
 199                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 200         if (opt->pre_context || opt->post_context)
 201                 hunk_mark = "--\n";
 202
 203         while (left) {
 204                 regmatch_t pmatch[10];
 205                 char *eol, ch;
 206                 int hit = 0;
 207                 struct grep_pat *p;
 208
 209                 eol = end_of_line(bol, &left);
 210                 ch = *eol;
 211                 *eol = 0;
 212
 213                 for (p = opt->pattern_list; p; p = p->next) {
 214                         regex_t *exp = &p->regexp;
 215                         hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 216                                        pmatch, 0);
 217
 218                         if (hit && opt->word_regexp) {
 219                                 /* Match beginning must be either
 220                                  * beginning of the line, or at word
 221                                  * boundary (i.e. the last char must
 222                                  * not be alnum or underscore).
 223                                  */
 224                                 if ((pmatch[0].rm_so < 0) ||
 225                                     (eol - bol) <= pmatch[0].rm_so ||
 226                                     (pmatch[0].rm_eo < 0) ||
 227                                     (eol - bol) < pmatch[0].rm_eo)
 228                                         die("regexp returned nonsense");
 229                                 if (pmatch[0].rm_so != 0 &&
 230                                     word_char(bol[pmatch[0].rm_so-1]))
 231                                         continue; /* not a word boundary */
 232                                 if ((eol-bol) < pmatch[0].rm_eo &&
 233                                     word_char(bol[pmatch[0].rm_eo]))
 234                                         continue; /* not a word boundary */
 235                         }
 236                         if (hit)
 237                                 break;
 238                 }
 239                 /* "grep -v -e foo -e bla" should list lines
 240                  * that do not have either, so inversion should
 241                  * be done outside.
 242                  */
 243                 if (opt->invert)
 244                         hit = !hit;
 245                 if (opt->unmatch_name_only) {
 246                         if (hit)
 247                                 return 0;
 248                         goto next_line;
 249                 }
 250                 if (hit) {
 251                         count++;
 252                         if (binary_match_only) {
 253                                 printf("Binary file %s matches\n", name);
 254                                 return 1;
 255                         }
 256                         if (opt->name_only) {
 257                                 printf("%s\n", name);
 258                                 return 1;
 259                         }
 260                         /* Hit at this line.  If we haven't shown the
 261                          * pre-context lines, we would need to show them.
 262                          * When asked to do "count", this still show
 263                          * the context which is nonsense, but the user
 264                          * deserves to get that ;-).
 265                          */
 266                         if (opt->pre_context) {
 267                                 unsigned from;
 268                                 if (opt->pre_context < lno)
 269                                         from = lno - opt->pre_context;
 270                                 else
 271                                         from = 1;
 272                                 if (from <= last_shown)
 273                                         from = last_shown + 1;
 274                                 if (last_shown && from != last_shown + 1)
 275                                         printf(hunk_mark);
 276                                 while (from < lno) {
 277                                         pcl = &prev[lno-from-1];
 278                                         show_line(opt, pcl->bol, pcl->eol,
 279                                                   name, from, '-');
 280                                         from++;
 281                                 }
 282                                 last_shown = lno-1;
 283                         }
 284                         if (last_shown && lno != last_shown + 1)
 285                                 printf(hunk_mark);
 286                         if (!opt->count)
 287                                 show_line(opt, bol, eol, name, lno, ':');
 288                         last_shown = last_hit = lno;
 289                 }
 290                 else if (last_hit &&
 291                          lno <= last_hit + opt->post_context) {
 292                         /* If the last hit is within the post context,
 293                          * we need to show this line.
 294                          */
 295                         if (last_shown && lno != last_shown + 1)
 296                                 printf(hunk_mark);
 297                         show_line(opt, bol, eol, name, lno, '-');
 298                         last_shown = lno;
 299                 }
 300                 if (opt->pre_context) {
 301                         memmove(prev+1, prev,
 302                                 (opt->pre_context-1) * sizeof(*prev));
 303                         prev->bol = bol;
 304                         prev->eol = eol;
 305                 }
 306
 307         next_line:
 308                 *eol = ch;
 309                 bol = eol + 1;
 310                 if (!left)
 311                         break;
 312                 left--;
 313                 lno++;
 314         }
 315
 316         if (opt->unmatch_name_only) {
 317                 /* We did not see any hit, so we want to show this */
 318                 printf("%s\n", name);
 319                 return 1;
 320         }
 321
 322         /* NEEDSWORK:
 323          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 324          * which feels mostly useless but sometimes useful.  Maybe
 325          * make it another option?  For now suppress them.
 326          */
 327         if (opt->count && count)
 328                 printf("%s:%u\n", name, count);
 329         return !!last_hit;
 330 }
 331
 332 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name)
 333 {
 334         unsigned long size;
 335         char *data;
 336         char type[20];
 337         int hit;
 338         data = read_sha1_file(sha1, type, &size);
 339         if (!data) {
 340                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 341                 return 0;
 342         }
 343         hit = grep_buffer(opt, name, data, size);
 344         free(data);
 345         return hit;
 346 }
 347
 348 static int grep_file(struct grep_opt *opt, const char *filename)
 349 {
 350         struct stat st;
 351         int i;
 352         char *data;
 353         if (lstat(filename, &st) < 0) {
 354         err_ret:
 355                 if (errno != ENOENT)
 356                         error("'%s': %s", filename, strerror(errno));
 357                 return 0;
 358         }
 359         if (!st.st_size)
 360                 return 0; /* empty file -- no grep hit */
 361         if (!S_ISREG(st.st_mode))
 362                 return 0;
 363         i = open(filename, O_RDONLY);
 364         if (i < 0)
 365                 goto err_ret;
 366         data = xmalloc(st.st_size + 1);
 367         if (st.st_size != xread(i, data, st.st_size)) {
 368                 error("'%s': short read %s", filename, strerror(errno));
 369                 close(i);
 370                 free(data);
 371                 return 0;
 372         }
 373         close(i);
 374         i = grep_buffer(opt, filename, data, st.st_size);
 375         free(data);
 376         return i;
 377 }
 378
 379 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 380 {
 381         int hit = 0;
 382         int nr;
 383         read_cache();
 384
 385         for (nr = 0; nr < active_nr; nr++) {
 386                 struct cache_entry *ce = active_cache[nr];
 387                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 388                         continue;
 389                 if (!pathspec_matches(paths, ce->name))
 390                         continue;
 391                 if (cached)
 392                         hit |= grep_sha1(opt, ce->sha1, ce->name);
 393                 else
 394                         hit |= grep_file(opt, ce->name);
 395         }
 396         return hit;
 397 }
 398
 399 static int grep_tree(struct grep_opt *opt, const char **paths,
 400                      struct tree_desc *tree,
 401                      const char *tree_name, const char *base)
 402 {
 403         unsigned mode;
 404         int len;
 405         int hit = 0;
 406         const char *path;
 407         const unsigned char *sha1;
 408         char *down;
 409         char *path_buf = xmalloc(PATH_MAX + strlen(tree_name) + 100);
 410
 411         if (tree_name[0]) {
 412                 int offset = sprintf(path_buf, "%s:", tree_name);
 413                 down = path_buf + offset;
 414                 strcat(down, base);
 415         }
 416         else {
 417                 down = path_buf;
 418                 strcpy(down, base);
 419         }
 420         len = strlen(path_buf);
 421
 422         while (tree->size) {
 423                 int pathlen;
 424                 sha1 = tree_entry_extract(tree, &path, &mode);
 425                 pathlen = strlen(path);
 426                 strcpy(path_buf + len, path);
 427
 428                 if (S_ISDIR(mode))
 429                         /* Match "abc/" against pathspec to
 430                          * decide if we want to descend into "abc"
 431                          * directory.
 432                          */
 433                         strcpy(path_buf + len + pathlen, "/");
 434
 435                 if (!pathspec_matches(paths, down))
 436                         ;
 437                 else if (S_ISREG(mode))
 438                         hit |= grep_sha1(opt, sha1, path_buf);
 439                 else if (S_ISDIR(mode)) {
 440                         char type[20];
 441                         struct tree_desc sub;
 442                         void *data;
 443                         data = read_sha1_file(sha1, type, &sub.size);
 444                         if (!data)
 445                                 die("unable to read tree (%s)",
 446                                     sha1_to_hex(sha1));
 447                         sub.buf = data;
 448                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 449                         free(data);
 450                 }
 451                 update_tree_entry(tree);
 452         }
 453         return hit;
 454 }
 455
 456 static int grep_object(struct grep_opt *opt, const char **paths,
 457                        struct object *obj, const char *name)
 458 {
 459         if (!strcmp(obj->type, blob_type))
 460                 return grep_sha1(opt, obj->sha1, name);
 461         if (!strcmp(obj->type, commit_type) ||
 462             !strcmp(obj->type, tree_type)) {
 463                 struct tree_desc tree;
 464                 void *data;
 465                 int hit;
 466                 data = read_object_with_reference(obj->sha1, tree_type,
 467                                                   &tree.size, NULL);
 468                 if (!data)
 469                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 470                 tree.buf = data;
 471                 hit = grep_tree(opt, paths, &tree, name, "");
 472                 free(data);
 473                 return hit;
 474         }
 475         die("unable to grep from object of type %s", obj->type);
 476 }
 477
 478 static const char builtin_grep_usage[] =
 479 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 480
 481 int cmd_grep(int argc, const char **argv, char **envp)
 482 {
 483         int hit = 0;
 484         int no_more_flags = 0;
 485         int seen_noncommit = 0;
 486         int cached = 0;
 487         struct grep_opt opt;
 488         struct object_list *list, **tail, *object_list = NULL;
 489         const char *prefix = setup_git_directory();
 490         const char **paths = NULL;
 491
 492         memset(&opt, 0, sizeof(opt));
 493         opt.pattern_tail = &opt.pattern_list;
 494         opt.regflags = REG_NEWLINE;
 495
 496         /*
 497          * No point using rev_info, really.
 498          */
 499         while (1 < argc) {
 500                 const char *arg = argv[1];
 501                 argc--; argv++;
 502                 if (!strcmp("--cached", arg)) {
 503                         cached = 1;
 504                         continue;
 505                 }
 506                 if (!strcmp("-a", arg) ||
 507                     !strcmp("--text", arg)) {
 508                         opt.binary = GREP_BINARY_TEXT;
 509                         continue;
 510                 }
 511                 if (!strcmp("-i", arg) ||
 512                     !strcmp("--ignore-case", arg)) {
 513                         opt.regflags |= REG_ICASE;
 514                         continue;
 515                 }
 516                 if (!strcmp("-I", arg)) {
 517                         opt.binary = GREP_BINARY_NOMATCH;
 518                         continue;
 519                 }
 520                 if (!strcmp("-v", arg) ||
 521                     !strcmp("--invert-match", arg)) {
 522                         opt.invert = 1;
 523                         continue;
 524                 }
 525                 if (!strcmp("-E", arg) ||
 526                     !strcmp("--extended-regexp", arg)) {
 527                         opt.regflags |= REG_EXTENDED;
 528                         continue;
 529                 }
 530                 if (!strcmp("-G", arg) ||
 531                     !strcmp("--basic-regexp", arg)) {
 532                         opt.regflags &= ~REG_EXTENDED;
 533                         continue;
 534                 }
 535                 if (!strcmp("-n", arg)) {
 536                         opt.linenum = 1;
 537                         continue;
 538                 }
 539                 if (!strcmp("-H", arg)) {
 540                         /* We always show the pathname, so this
 541                          * is a noop.
 542                          */
 543                         continue;
 544                 }
 545                 if (!strcmp("-l", arg) ||
 546                     !strcmp("--files-with-matches", arg)) {
 547                         opt.name_only = 1;
 548                         continue;
 549                 }
 550                 if (!strcmp("-L", arg) ||
 551                     !strcmp("--files-without-match", arg)) {
 552                         opt.unmatch_name_only = 1;
 553                         continue;
 554                 }
 555                 if (!strcmp("-c", arg) ||
 556                     !strcmp("--count", arg)) {
 557                         opt.count = 1;
 558                         continue;
 559                 }
 560                 if (!strcmp("-w", arg) ||
 561                     !strcmp("--word-regexp", arg)) {
 562                         opt.word_regexp = 1;
 563                         continue;
 564                 }
 565                 if (!strncmp("-A", arg, 2) ||
 566                     !strncmp("-B", arg, 2) ||
 567                     !strncmp("-C", arg, 2) ||
 568                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
 569                         unsigned num;
 570                         const char *scan;
 571                         switch (arg[1]) {
 572                         case 'A': case 'B': case 'C':
 573                                 if (!arg[2]) {
 574                                         if (argc <= 1)
 575                                                 usage(builtin_grep_usage);
 576                                         scan = *++argv;
 577                                         argc--;
 578                                 }
 579                                 else
 580                                         scan = arg + 2;
 581                                 break;
 582                         default:
 583                                 scan = arg + 1;
 584                                 break;
 585                         }
 586                         if (sscanf(scan, "%u", &num) != 1)
 587                                 usage(builtin_grep_usage);
 588                         switch (arg[1]) {
 589                         case 'A':
 590                                 opt.post_context = num;
 591                                 break;
 592                         default:
 593                         case 'C':
 594                                 opt.post_context = num;
 595                         case 'B':
 596                                 opt.pre_context = num;
 597                                 break;
 598                         }
 599                         continue;
 600                 }
 601                 if (!strcmp("-e", arg)) {
 602                         if (1 < argc) {
 603                                 add_pattern(&opt, argv[1]);
 604                                 argv++;
 605                                 argc--;
 606                                 continue;
 607                         }
 608                         usage(builtin_grep_usage);
 609                 }
 610                 if (!strcmp("--", arg)) {
 611                         no_more_flags = 1;
 612                         continue;
 613                 }
 614                 /* Either unrecognized option or a single pattern */
 615                 if (!no_more_flags && *arg == '-')
 616                         usage(builtin_grep_usage);
 617                 if (!opt.pattern_list) {
 618                         add_pattern(&opt, arg);
 619                         break;
 620                 }
 621                 else {
 622                         /* We are looking at the first path or rev;
 623                          * it is found at argv[0] after leaving the
 624                          * loop.
 625                          */
 626                         argc++; argv--;
 627                         break;
 628                 }
 629         }
 630         if (!opt.pattern_list)
 631                 die("no pattern given.");
 632         compile_patterns(&opt);
 633         tail = &object_list;
 634         while (1 < argc) {
 635                 struct object *object;
 636                 struct object_list *elem;
 637                 const char *arg = argv[1];
 638                 unsigned char sha1[20];
 639                 if (get_sha1(arg, sha1) < 0)
 640                         break;
 641                 object = parse_object(sha1);
 642                 if (!object)
 643                         die("bad object %s", arg);
 644                 elem = object_list_insert(object, tail);
 645                 elem->name = arg;
 646                 tail = &elem->next;
 647                 argc--; argv++;
 648         }
 649         if (1 < argc)
 650                 paths = get_pathspec(prefix, argv + 1);
 651         else if (prefix) {
 652                 paths = xcalloc(2, sizeof(const char *));
 653                 paths[0] = prefix;
 654                 paths[1] = NULL;
 655         }
 656
 657         if (!object_list)
 658                 return !grep_cache(&opt, paths, cached);
 659         /*
 660          * Do not walk "grep -e foo master next pu -- Documentation/"
 661          * but do walk "grep -e foo master..next -- Documentation/".
 662          * Ranged request mixed with a blob or tree object, like
 663          * "grep -e foo v1.0.0:Documentation/ master..next"
 664          * so detect that and complain.
 665          */
 666         for (list = object_list; list; list = list->next) {
 667                 struct object *real_obj;
 668                 real_obj = deref_tag(list->item, NULL, 0);
 669                 if (strcmp(real_obj->type, commit_type))
 670                         seen_noncommit = 1;
 671         }
 672         if (cached)
 673                 die("both --cached and revisions given.");
 674
 675         for (list = object_list; list; list = list->next) {
 676                 struct object *real_obj;
 677                 real_obj = deref_tag(list->item, NULL, 0);
 678                 if (grep_object(&opt, paths, real_obj, list->name))
 679                         hit = 1;
 680         }
 681         return !hit;
 682 }