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