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