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