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