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 #include <sys/wait.h>
  16
  17 /*
  18  * git grep pathspecs are somewhat different from diff-tree pathspecs;
  19  * pathname wildcards are allowed.
  20  */
  21 static int pathspec_matches(const char **paths, const char *name)
  22 {
  23         int namelen, i;
  24         if (!paths || !*paths)
  25                 return 1;
  26         namelen = strlen(name);
  27         for (i = 0; paths[i]; i++) {
  28                 const char *match = paths[i];
  29                 int matchlen = strlen(match);
  30                 const char *cp, *meta;
  31
  32                 if (!matchlen ||
  33                     ((matchlen <= namelen) &&
  34                      !strncmp(name, match, matchlen) &&
  35                      (match[matchlen-1] == '/' ||
  36                       name[matchlen] == '\0' || name[matchlen] == '/')))
  37                         return 1;
  38                 if (!fnmatch(match, name, 0))
  39                         return 1;
  40                 if (name[namelen-1] != '/')
  41                         continue;
  42
  43                 /* We are being asked if the directory ("name") is worth
  44                  * descending into.
  45                  *
  46                  * Find the longest leading directory name that does
  47                  * not have metacharacter in the pathspec; the name
  48                  * we are looking at must overlap with that directory.
  49                  */
  50                 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
  51                         char ch = *cp;
  52                         if (ch == '*' || ch == '[' || ch == '?') {
  53                                 meta = cp;
  54                                 break;
  55                         }
  56                 }
  57                 if (!meta)
  58                         meta = cp; /* fully literal */
  59
  60                 if (namelen <= meta - match) {
  61                         /* Looking at "Documentation/" and
  62                          * the pattern says "Documentation/howto/", or
  63                          * "Documentation/diff*.txt".  The name we
  64                          * have should match prefix.
  65                          */
  66                         if (!memcmp(match, name, namelen))
  67                                 return 1;
  68                         continue;
  69                 }
  70
  71                 if (meta - match < namelen) {
  72                         /* Looking at "Documentation/howto/" and
  73                          * the pattern says "Documentation/h*";
  74                          * match up to "Do.../h"; this avoids descending
  75                          * into "Documentation/technical/".
  76                          */
  77                         if (!memcmp(match, name, meta - match))
  78                                 return 1;
  79                         continue;
  80                 }
  81         }
  82         return 0;
  83 }
  84
  85 enum grep_pat_token {
  86         GREP_PATTERN,
  87         GREP_AND,
  88         GREP_OPEN_PAREN,
  89         GREP_CLOSE_PAREN,
  90         GREP_NOT,
  91         GREP_OR,
  92 };
  93
  94 struct grep_pat {
  95         struct grep_pat *next;
  96         const char *origin;
  97         int no;
  98         enum grep_pat_token token;
  99         const char *pattern;
 100         regex_t regexp;
 101 };
 102
 103 enum grep_expr_node {
 104         GREP_NODE_ATOM,
 105         GREP_NODE_NOT,
 106         GREP_NODE_AND,
 107         GREP_NODE_OR,
 108 };
 109
 110 struct grep_expr {
 111         enum grep_expr_node node;
 112         union {
 113                 struct grep_pat *atom;
 114                 struct grep_expr *unary;
 115                 struct {
 116                         struct grep_expr *left;
 117                         struct grep_expr *right;
 118                 } binary;
 119         } u;
 120 };
 121
 122 struct grep_opt {
 123         struct grep_pat *pattern_list;
 124         struct grep_pat **pattern_tail;
 125         struct grep_expr *pattern_expression;
 126         regex_t regexp;
 127         unsigned linenum:1;
 128         unsigned invert:1;
 129         unsigned name_only:1;
 130         unsigned unmatch_name_only:1;
 131         unsigned count:1;
 132         unsigned word_regexp:1;
 133         unsigned fixed:1;
 134 #define GREP_BINARY_DEFAULT     0
 135 #define GREP_BINARY_NOMATCH     1
 136 #define GREP_BINARY_TEXT        2
 137         unsigned binary:2;
 138         unsigned extended:1;
 139         int regflags;
 140         unsigned pre_context;
 141         unsigned post_context;
 142 };
 143
 144 static void add_pattern(struct grep_opt *opt, const char *pat,
 145                         const char *origin, int no, enum grep_pat_token t)
 146 {
 147         struct grep_pat *p = xcalloc(1, sizeof(*p));
 148         p->pattern = pat;
 149         p->origin = origin;
 150         p->no = no;
 151         p->token = t;
 152         *opt->pattern_tail = p;
 153         opt->pattern_tail = &p->next;
 154         p->next = NULL;
 155 }
 156
 157 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 158 {
 159         int err = regcomp(&p->regexp, p->pattern, opt->regflags);
 160         if (err) {
 161                 char errbuf[1024];
 162                 char where[1024];
 163                 if (p->no)
 164                         sprintf(where, "In '%s' at %d, ",
 165                                 p->origin, p->no);
 166                 else if (p->origin)
 167                         sprintf(where, "%s, ", p->origin);
 168                 else
 169                         where[0] = 0;
 170                 regerror(err, &p->regexp, errbuf, 1024);
 171                 regfree(&p->regexp);
 172                 die("%s'%s': %s", where, p->pattern, errbuf);
 173         }
 174 }
 175
 176 #if DEBUG
 177 static inline void indent(int in)
 178 {
 179         int i;
 180         for (i = 0; i < in; i++) putchar(' ');
 181 }
 182
 183 static void dump_pattern_exp(struct grep_expr *x, int in)
 184 {
 185         switch (x->node) {
 186         case GREP_NODE_ATOM:
 187                 indent(in);
 188                 puts(x->u.atom->pattern);
 189                 break;
 190         case GREP_NODE_NOT:
 191                 indent(in);
 192                 puts("--not");
 193                 dump_pattern_exp(x->u.unary, in+1);
 194                 break;
 195         case GREP_NODE_AND:
 196                 dump_pattern_exp(x->u.binary.left, in+1);
 197                 indent(in);
 198                 puts("--and");
 199                 dump_pattern_exp(x->u.binary.right, in+1);
 200                 break;
 201         case GREP_NODE_OR:
 202                 dump_pattern_exp(x->u.binary.left, in+1);
 203                 indent(in);
 204                 puts("--or");
 205                 dump_pattern_exp(x->u.binary.right, in+1);
 206                 break;
 207         }
 208 }
 209
 210 static void looking_at(const char *msg, struct grep_pat **list)
 211 {
 212         struct grep_pat *p = *list;
 213         fprintf(stderr, "%s: looking at ", msg);
 214         if (!p)
 215                 fprintf(stderr, "empty\n");
 216         else
 217                 fprintf(stderr, "<%s>\n", p->pattern);
 218 }
 219 #else
 220 #define looking_at(a,b) do {} while(0)
 221 #endif
 222
 223 static struct grep_expr *compile_pattern_expr(struct grep_pat **);
 224 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 225 {
 226         struct grep_pat *p;
 227         struct grep_expr *x;
 228
 229         looking_at("atom", list);
 230
 231         p = *list;
 232         switch (p->token) {
 233         case GREP_PATTERN: /* atom */
 234                 x = xcalloc(1, sizeof (struct grep_expr));
 235                 x->node = GREP_NODE_ATOM;
 236                 x->u.atom = p;
 237                 *list = p->next;
 238                 return x;
 239         case GREP_OPEN_PAREN:
 240                 *list = p->next;
 241                 x = compile_pattern_expr(list);
 242                 if (!x)
 243                         return NULL;
 244                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 245                         die("unmatched parenthesis");
 246                 *list = (*list)->next;
 247                 return x;
 248         default:
 249                 return NULL;
 250         }
 251 }
 252
 253 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 254 {
 255         struct grep_pat *p;
 256         struct grep_expr *x;
 257
 258         looking_at("not", list);
 259
 260         p = *list;
 261         switch (p->token) {
 262         case GREP_NOT:
 263                 if (!p->next)
 264                         die("--not not followed by pattern expression");
 265                 *list = p->next;
 266                 x = xcalloc(1, sizeof (struct grep_expr));
 267                 x->node = GREP_NODE_NOT;
 268                 x->u.unary = compile_pattern_not(list);
 269                 if (!x->u.unary)
 270                         die("--not followed by non pattern expression");
 271                 return x;
 272         default:
 273                 return compile_pattern_atom(list);
 274         }
 275 }
 276
 277 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 278 {
 279         struct grep_pat *p;
 280         struct grep_expr *x, *y, *z;
 281
 282         looking_at("and", list);
 283
 284         x = compile_pattern_not(list);
 285         p = *list;
 286         if (p && p->token == GREP_AND) {
 287                 if (!p->next)
 288                         die("--and not followed by pattern expression");
 289                 *list = p->next;
 290                 y = compile_pattern_and(list);
 291                 if (!y)
 292                         die("--and not followed by pattern expression");
 293                 z = xcalloc(1, sizeof (struct grep_expr));
 294                 z->node = GREP_NODE_AND;
 295                 z->u.binary.left = x;
 296                 z->u.binary.right = y;
 297                 return z;
 298         }
 299         return x;
 300 }
 301
 302 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 303 {
 304         struct grep_pat *p;
 305         struct grep_expr *x, *y, *z;
 306
 307         looking_at("or", list);
 308
 309         x = compile_pattern_and(list);
 310         p = *list;
 311         if (x && p && p->token != GREP_CLOSE_PAREN) {
 312                 y = compile_pattern_or(list);
 313                 if (!y)
 314                         die("not a pattern expression %s", p->pattern);
 315                 z = xcalloc(1, sizeof (struct grep_expr));
 316                 z->node = GREP_NODE_OR;
 317                 z->u.binary.left = x;
 318                 z->u.binary.right = y;
 319                 return z;
 320         }
 321         return x;
 322 }
 323
 324 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 325 {
 326         looking_at("expr", list);
 327
 328         return compile_pattern_or(list);
 329 }
 330
 331 static void compile_patterns(struct grep_opt *opt)
 332 {
 333         struct grep_pat *p;
 334
 335         /* First compile regexps */
 336         for (p = opt->pattern_list; p; p = p->next) {
 337                 if (p->token == GREP_PATTERN)
 338                         compile_regexp(p, opt);
 339                 else
 340                         opt->extended = 1;
 341         }
 342
 343         if (!opt->extended)
 344                 return;
 345
 346         /* Then bundle them up in an expression.
 347          * A classic recursive descent parser would do.
 348          */
 349         p = opt->pattern_list;
 350         opt->pattern_expression = compile_pattern_expr(&p);
 351 #if DEBUG
 352         dump_pattern_exp(opt->pattern_expression, 0);
 353 #endif
 354         if (p)
 355                 die("incomplete pattern expression: %s", p->pattern);
 356 }
 357
 358 static char *end_of_line(char *cp, unsigned long *left)
 359 {
 360         unsigned long l = *left;
 361         while (l && *cp != '\n') {
 362                 l--;
 363                 cp++;
 364         }
 365         *left = l;
 366         return cp;
 367 }
 368
 369 static int word_char(char ch)
 370 {
 371         return isalnum(ch) || ch == '_';
 372 }
 373
 374 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
 375                       const char *name, unsigned lno, char sign)
 376 {
 377         printf("%s%c", name, sign);
 378         if (opt->linenum)
 379                 printf("%d%c", lno, sign);
 380         printf("%.*s\n", (int)(eol-bol), bol);
 381 }
 382
 383 /*
 384  * NEEDSWORK: share code with diff.c
 385  */
 386 #define FIRST_FEW_BYTES 8000
 387 static int buffer_is_binary(const char *ptr, unsigned long size)
 388 {
 389         if (FIRST_FEW_BYTES < size)
 390                 size = FIRST_FEW_BYTES;
 391         if (memchr(ptr, 0, size))
 392                 return 1;
 393         return 0;
 394 }
 395
 396 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 397 {
 398         char *hit = strstr(line, pattern);
 399         if (!hit) {
 400                 match->rm_so = match->rm_eo = -1;
 401                 return REG_NOMATCH;
 402         }
 403         else {
 404                 match->rm_so = hit - line;
 405                 match->rm_eo = match->rm_so + strlen(pattern);
 406                 return 0;
 407         }
 408 }
 409
 410 static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol, char *eol)
 411 {
 412         int hit = 0;
 413         regmatch_t pmatch[10];
 414
 415         if (!opt->fixed) {
 416                 regex_t *exp = &p->regexp;
 417                 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
 418                                pmatch, 0);
 419         }
 420         else {
 421                 hit = !fixmatch(p->pattern, bol, pmatch);
 422         }
 423
 424         if (hit && opt->word_regexp) {
 425                 /* Match beginning must be either
 426                  * beginning of the line, or at word
 427                  * boundary (i.e. the last char must
 428                  * not be alnum or underscore).
 429                  */
 430                 if ((pmatch[0].rm_so < 0) ||
 431                     (eol - bol) <= pmatch[0].rm_so ||
 432                     (pmatch[0].rm_eo < 0) ||
 433                     (eol - bol) < pmatch[0].rm_eo)
 434                         die("regexp returned nonsense");
 435                 if (pmatch[0].rm_so != 0 &&
 436                     word_char(bol[pmatch[0].rm_so-1]))
 437                         hit = 0;
 438                 if (pmatch[0].rm_eo != (eol-bol) &&
 439                     word_char(bol[pmatch[0].rm_eo]))
 440                         hit = 0;
 441         }
 442         return hit;
 443 }
 444
 445 static int match_expr_eval(struct grep_opt *opt,
 446                            struct grep_expr *x,
 447                            char *bol, char *eol)
 448 {
 449         switch (x->node) {
 450         case GREP_NODE_ATOM:
 451                 return match_one_pattern(opt, x->u.atom, bol, eol);
 452                 break;
 453         case GREP_NODE_NOT:
 454                 return !match_expr_eval(opt, x->u.unary, bol, eol);
 455         case GREP_NODE_AND:
 456                 return (match_expr_eval(opt, x->u.binary.left, bol, eol) &&
 457                         match_expr_eval(opt, x->u.binary.right, bol, eol));
 458         case GREP_NODE_OR:
 459                 return (match_expr_eval(opt, x->u.binary.left, bol, eol) ||
 460                         match_expr_eval(opt, x->u.binary.right, bol, eol));
 461         }
 462         die("Unexpected node type (internal error) %d\n", x->node);
 463 }
 464
 465 static int match_expr(struct grep_opt *opt, char *bol, char *eol)
 466 {
 467         struct grep_expr *x = opt->pattern_expression;
 468         return match_expr_eval(opt, x, bol, eol);
 469 }
 470
 471 static int match_line(struct grep_opt *opt, char *bol, char *eol)
 472 {
 473         struct grep_pat *p;
 474         if (opt->extended)
 475                 return match_expr(opt, bol, eol);
 476         for (p = opt->pattern_list; p; p = p->next) {
 477                 if (match_one_pattern(opt, p, bol, eol))
 478                         return 1;
 479         }
 480         return 0;
 481 }
 482
 483 static int grep_buffer(struct grep_opt *opt, const char *name,
 484                        char *buf, unsigned long size)
 485 {
 486         char *bol = buf;
 487         unsigned long left = size;
 488         unsigned lno = 1;
 489         struct pre_context_line {
 490                 char *bol;
 491                 char *eol;
 492         } *prev = NULL, *pcl;
 493         unsigned last_hit = 0;
 494         unsigned last_shown = 0;
 495         int binary_match_only = 0;
 496         const char *hunk_mark = "";
 497         unsigned count = 0;
 498
 499         if (buffer_is_binary(buf, size)) {
 500                 switch (opt->binary) {
 501                 case GREP_BINARY_DEFAULT:
 502                         binary_match_only = 1;
 503                         break;
 504                 case GREP_BINARY_NOMATCH:
 505                         return 0; /* Assume unmatch */
 506                         break;
 507                 default:
 508                         break;
 509                 }
 510         }
 511
 512         if (opt->pre_context)
 513                 prev = xcalloc(opt->pre_context, sizeof(*prev));
 514         if (opt->pre_context || opt->post_context)
 515                 hunk_mark = "--\n";
 516
 517         while (left) {
 518                 char *eol, ch;
 519                 int hit = 0;
 520
 521                 eol = end_of_line(bol, &left);
 522                 ch = *eol;
 523                 *eol = 0;
 524
 525                 hit = match_line(opt, bol, eol);
 526
 527                 /* "grep -v -e foo -e bla" should list lines
 528                  * that do not have either, so inversion should
 529                  * be done outside.
 530                  */
 531                 if (opt->invert)
 532                         hit = !hit;
 533                 if (opt->unmatch_name_only) {
 534                         if (hit)
 535                                 return 0;
 536                         goto next_line;
 537                 }
 538                 if (hit) {
 539                         count++;
 540                         if (binary_match_only) {
 541                                 printf("Binary file %s matches\n", name);
 542                                 return 1;
 543                         }
 544                         if (opt->name_only) {
 545                                 printf("%s\n", name);
 546                                 return 1;
 547                         }
 548                         /* Hit at this line.  If we haven't shown the
 549                          * pre-context lines, we would need to show them.
 550                          * When asked to do "count", this still show
 551                          * the context which is nonsense, but the user
 552                          * deserves to get that ;-).
 553                          */
 554                         if (opt->pre_context) {
 555                                 unsigned from;
 556                                 if (opt->pre_context < lno)
 557                                         from = lno - opt->pre_context;
 558                                 else
 559                                         from = 1;
 560                                 if (from <= last_shown)
 561                                         from = last_shown + 1;
 562                                 if (last_shown && from != last_shown + 1)
 563                                         printf(hunk_mark);
 564                                 while (from < lno) {
 565                                         pcl = &prev[lno-from-1];
 566                                         show_line(opt, pcl->bol, pcl->eol,
 567                                                   name, from, '-');
 568                                         from++;
 569                                 }
 570                                 last_shown = lno-1;
 571                         }
 572                         if (last_shown && lno != last_shown + 1)
 573                                 printf(hunk_mark);
 574                         if (!opt->count)
 575                                 show_line(opt, bol, eol, name, lno, ':');
 576                         last_shown = last_hit = lno;
 577                 }
 578                 else if (last_hit &&
 579                          lno <= last_hit + opt->post_context) {
 580                         /* If the last hit is within the post context,
 581                          * we need to show this line.
 582                          */
 583                         if (last_shown && lno != last_shown + 1)
 584                                 printf(hunk_mark);
 585                         show_line(opt, bol, eol, name, lno, '-');
 586                         last_shown = lno;
 587                 }
 588                 if (opt->pre_context) {
 589                         memmove(prev+1, prev,
 590                                 (opt->pre_context-1) * sizeof(*prev));
 591                         prev->bol = bol;
 592                         prev->eol = eol;
 593                 }
 594
 595         next_line:
 596                 *eol = ch;
 597                 bol = eol + 1;
 598                 if (!left)
 599                         break;
 600                 left--;
 601                 lno++;
 602         }
 603
 604         if (opt->unmatch_name_only) {
 605                 /* We did not see any hit, so we want to show this */
 606                 printf("%s\n", name);
 607                 return 1;
 608         }
 609
 610         /* NEEDSWORK:
 611          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 612          * which feels mostly useless but sometimes useful.  Maybe
 613          * make it another option?  For now suppress them.
 614          */
 615         if (opt->count && count)
 616                 printf("%s:%u\n", name, count);
 617         return !!last_hit;
 618 }
 619
 620 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name)
 621 {
 622         unsigned long size;
 623         char *data;
 624         char type[20];
 625         int hit;
 626         data = read_sha1_file(sha1, type, &size);
 627         if (!data) {
 628                 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
 629                 return 0;
 630         }
 631         hit = grep_buffer(opt, name, data, size);
 632         free(data);
 633         return hit;
 634 }
 635
 636 static int grep_file(struct grep_opt *opt, const char *filename)
 637 {
 638         struct stat st;
 639         int i;
 640         char *data;
 641         if (lstat(filename, &st) < 0) {
 642         err_ret:
 643                 if (errno != ENOENT)
 644                         error("'%s': %s", filename, strerror(errno));
 645                 return 0;
 646         }
 647         if (!st.st_size)
 648                 return 0; /* empty file -- no grep hit */
 649         if (!S_ISREG(st.st_mode))
 650                 return 0;
 651         i = open(filename, O_RDONLY);
 652         if (i < 0)
 653                 goto err_ret;
 654         data = xmalloc(st.st_size + 1);
 655         if (st.st_size != xread(i, data, st.st_size)) {
 656                 error("'%s': short read %s", filename, strerror(errno));
 657                 close(i);
 658                 free(data);
 659                 return 0;
 660         }
 661         close(i);
 662         i = grep_buffer(opt, filename, data, st.st_size);
 663         free(data);
 664         return i;
 665 }
 666
 667 static int exec_grep(int argc, const char **argv)
 668 {
 669         pid_t pid;
 670         int status;
 671
 672         argv[argc] = NULL;
 673         pid = fork();
 674         if (pid < 0)
 675                 return pid;
 676         if (!pid) {
 677                 execvp("grep", (char **) argv);
 678                 exit(255);
 679         }
 680         while (waitpid(pid, &status, 0) < 0) {
 681                 if (errno == EINTR)
 682                         continue;
 683                 return -1;
 684         }
 685         if (WIFEXITED(status)) {
 686                 if (!WEXITSTATUS(status))
 687                         return 1;
 688                 return 0;
 689         }
 690         return -1;
 691 }
 692
 693 #define MAXARGS 1000
 694 #define ARGBUF 4096
 695 #define push_arg(a) do { \
 696         if (nr < MAXARGS) argv[nr++] = (a); \
 697         else die("maximum number of args exceeded"); \
 698         } while (0)
 699
 700 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
 701 {
 702         int i, nr, argc, hit, len, status;
 703         const char *argv[MAXARGS+1];
 704         char randarg[ARGBUF];
 705         char *argptr = randarg;
 706         struct grep_pat *p;
 707
 708         if (opt->extended)
 709                 return -1;
 710         len = nr = 0;
 711         push_arg("grep");
 712         if (opt->fixed)
 713                 push_arg("-F");
 714         if (opt->linenum)
 715                 push_arg("-n");
 716         if (opt->regflags & REG_EXTENDED)
 717                 push_arg("-E");
 718         if (opt->regflags & REG_ICASE)
 719                 push_arg("-i");
 720         if (opt->word_regexp)
 721                 push_arg("-w");
 722         if (opt->name_only)
 723                 push_arg("-l");
 724         if (opt->unmatch_name_only)
 725                 push_arg("-L");
 726         if (opt->count)
 727                 push_arg("-c");
 728         if (opt->post_context || opt->pre_context) {
 729                 if (opt->post_context != opt->pre_context) {
 730                         if (opt->pre_context) {
 731                                 push_arg("-B");
 732                                 len += snprintf(argptr, sizeof(randarg)-len,
 733                                                 "%u", opt->pre_context);
 734                                 if (sizeof(randarg) <= len)
 735                                         die("maximum length of args exceeded");
 736                                 push_arg(argptr);
 737                                 argptr += len;
 738                         }
 739                         if (opt->post_context) {
 740                                 push_arg("-A");
 741                                 len += snprintf(argptr, sizeof(randarg)-len,
 742                                                 "%u", opt->post_context);
 743                                 if (sizeof(randarg) <= len)
 744                                         die("maximum length of args exceeded");
 745                                 push_arg(argptr);
 746                                 argptr += len;
 747                         }
 748                 }
 749                 else {
 750                         push_arg("-C");
 751                         len += snprintf(argptr, sizeof(randarg)-len,
 752                                         "%u", opt->post_context);
 753                         if (sizeof(randarg) <= len)
 754                                 die("maximum length of args exceeded");
 755                         push_arg(argptr);
 756                         argptr += len;
 757                 }
 758         }
 759         for (p = opt->pattern_list; p; p = p->next) {
 760                 push_arg("-e");
 761                 push_arg(p->pattern);
 762         }
 763
 764         /*
 765          * To make sure we get the header printed out when we want it,
 766          * add /dev/null to the paths to grep.  This is unnecessary
 767          * (and wrong) with "-l" or "-L", which always print out the
 768          * name anyway.
 769          *
 770          * GNU grep has "-H", but this is portable.
 771          */
 772         if (!opt->name_only && !opt->unmatch_name_only)
 773                 push_arg("/dev/null");
 774
 775         hit = 0;
 776         argc = nr;
 777         for (i = 0; i < active_nr; i++) {
 778                 struct cache_entry *ce = active_cache[i];
 779                 char *name;
 780                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 781                         continue;
 782                 if (!pathspec_matches(paths, ce->name))
 783                         continue;
 784                 name = ce->name;
 785                 if (name[0] == '-') {
 786                         int len = ce_namelen(ce);
 787                         name = xmalloc(len + 3);
 788                         memcpy(name, "./", 2);
 789                         memcpy(name + 2, ce->name, len + 1);
 790                 }
 791                 argv[argc++] = name;
 792                 if (argc < MAXARGS)
 793                         continue;
 794                 status = exec_grep(argc, argv);
 795                 if (0 < status)
 796                         hit = 1;
 797                 argc = nr;
 798         }
 799         if (argc > nr) {
 800                 status = exec_grep(argc, argv);
 801                 if (0 < status)
 802                         hit = 1;
 803         }
 804         return hit;
 805 }
 806
 807 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 808 {
 809         int hit = 0;
 810         int nr;
 811         read_cache();
 812
 813 #ifdef __unix__
 814         /*
 815          * Use the external "grep" command for the case where
 816          * we grep through the checked-out files. It tends to
 817          * be a lot more optimized
 818          */
 819         if (!cached) {
 820                 hit = external_grep(opt, paths, cached);
 821                 if (hit >= 0)
 822                         return hit;
 823         }
 824 #endif
 825
 826         for (nr = 0; nr < active_nr; nr++) {
 827                 struct cache_entry *ce = active_cache[nr];
 828                 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
 829                         continue;
 830                 if (!pathspec_matches(paths, ce->name))
 831                         continue;
 832                 if (cached)
 833                         hit |= grep_sha1(opt, ce->sha1, ce->name);
 834                 else
 835                         hit |= grep_file(opt, ce->name);
 836         }
 837         return hit;
 838 }
 839
 840 static int grep_tree(struct grep_opt *opt, const char **paths,
 841                      struct tree_desc *tree,
 842                      const char *tree_name, const char *base)
 843 {
 844         int len;
 845         int hit = 0;
 846         struct name_entry entry;
 847         char *down;
 848         char *path_buf = xmalloc(PATH_MAX + strlen(tree_name) + 100);
 849
 850         if (tree_name[0]) {
 851                 int offset = sprintf(path_buf, "%s:", tree_name);
 852                 down = path_buf + offset;
 853                 strcat(down, base);
 854         }
 855         else {
 856                 down = path_buf;
 857                 strcpy(down, base);
 858         }
 859         len = strlen(path_buf);
 860
 861         while (tree_entry(tree, &entry)) {
 862                 strcpy(path_buf + len, entry.path);
 863
 864                 if (S_ISDIR(entry.mode))
 865                         /* Match "abc/" against pathspec to
 866                          * decide if we want to descend into "abc"
 867                          * directory.
 868                          */
 869                         strcpy(path_buf + len + entry.pathlen, "/");
 870
 871                 if (!pathspec_matches(paths, down))
 872                         ;
 873                 else if (S_ISREG(entry.mode))
 874                         hit |= grep_sha1(opt, entry.sha1, path_buf);
 875                 else if (S_ISDIR(entry.mode)) {
 876                         char type[20];
 877                         struct tree_desc sub;
 878                         void *data;
 879                         data = read_sha1_file(entry.sha1, type, &sub.size);
 880                         if (!data)
 881                                 die("unable to read tree (%s)",
 882                                     sha1_to_hex(entry.sha1));
 883                         sub.buf = data;
 884                         hit |= grep_tree(opt, paths, &sub, tree_name, down);
 885                         free(data);
 886                 }
 887         }
 888         return hit;
 889 }
 890
 891 static int grep_object(struct grep_opt *opt, const char **paths,
 892                        struct object *obj, const char *name)
 893 {
 894         if (obj->type == TYPE_BLOB)
 895                 return grep_sha1(opt, obj->sha1, name);
 896         if (obj->type == TYPE_COMMIT || obj->type == TYPE_TREE) {
 897                 struct tree_desc tree;
 898                 void *data;
 899                 int hit;
 900                 data = read_object_with_reference(obj->sha1, tree_type,
 901                                                   &tree.size, NULL);
 902                 if (!data)
 903                         die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
 904                 tree.buf = data;
 905                 hit = grep_tree(opt, paths, &tree, name, "");
 906                 free(data);
 907                 return hit;
 908         }
 909         die("unable to grep from object of type %s", typename(obj->type));
 910 }
 911
 912 static const char builtin_grep_usage[] =
 913 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
 914
 915 static const char emsg_invalid_context_len[] =
 916 "%s: invalid context length argument";
 917 static const char emsg_missing_context_len[] =
 918 "missing context length argument";
 919 static const char emsg_missing_argument[] =
 920 "option requires an argument -%s";
 921
 922 int cmd_grep(int argc, const char **argv, char **envp)
 923 {
 924         int hit = 0;
 925         int cached = 0;
 926         int seen_dashdash = 0;
 927         struct grep_opt opt;
 928         struct object_array list = { 0, 0, NULL };
 929         const char *prefix = setup_git_directory();
 930         const char **paths = NULL;
 931         int i;
 932
 933         memset(&opt, 0, sizeof(opt));
 934         opt.pattern_tail = &opt.pattern_list;
 935         opt.regflags = REG_NEWLINE;
 936
 937         /*
 938          * If there is no -- then the paths must exist in the working
 939          * tree.  If there is no explicit pattern specified with -e or
 940          * -f, we take the first unrecognized non option to be the
 941          * pattern, but then what follows it must be zero or more
 942          * valid refs up to the -- (if exists), and then existing
 943          * paths.  If there is an explicit pattern, then the first
 944          * unrecognized non option is the beginning of the refs list
 945          * that continues up to the -- (if exists), and then paths.
 946          */
 947
 948         while (1 < argc) {
 949                 const char *arg = argv[1];
 950                 argc--; argv++;
 951                 if (!strcmp("--cached", arg)) {
 952                         cached = 1;
 953                         continue;
 954                 }
 955                 if (!strcmp("-a", arg) ||
 956                     !strcmp("--text", arg)) {
 957                         opt.binary = GREP_BINARY_TEXT;
 958                         continue;
 959                 }
 960                 if (!strcmp("-i", arg) ||
 961                     !strcmp("--ignore-case", arg)) {
 962                         opt.regflags |= REG_ICASE;
 963                         continue;
 964                 }
 965                 if (!strcmp("-I", arg)) {
 966                         opt.binary = GREP_BINARY_NOMATCH;
 967                         continue;
 968                 }
 969                 if (!strcmp("-v", arg) ||
 970                     !strcmp("--invert-match", arg)) {
 971                         opt.invert = 1;
 972                         continue;
 973                 }
 974                 if (!strcmp("-E", arg) ||
 975                     !strcmp("--extended-regexp", arg)) {
 976                         opt.regflags |= REG_EXTENDED;
 977                         continue;
 978                 }
 979                 if (!strcmp("-F", arg) ||
 980                     !strcmp("--fixed-strings", arg)) {
 981                         opt.fixed = 1;
 982                         continue;
 983                 }
 984                 if (!strcmp("-G", arg) ||
 985                     !strcmp("--basic-regexp", arg)) {
 986                         opt.regflags &= ~REG_EXTENDED;
 987                         continue;
 988                 }
 989                 if (!strcmp("-n", arg)) {
 990                         opt.linenum = 1;
 991                         continue;
 992                 }
 993                 if (!strcmp("-H", arg)) {
 994                         /* We always show the pathname, so this
 995                          * is a noop.
 996                          */
 997                         continue;
 998                 }
 999                 if (!strcmp("-l", arg) ||
1000                     !strcmp("--files-with-matches", arg)) {
1001                         opt.name_only = 1;
1002                         continue;
1003                 }
1004                 if (!strcmp("-L", arg) ||
1005                     !strcmp("--files-without-match", arg)) {
1006                         opt.unmatch_name_only = 1;
1007                         continue;
1008                 }
1009                 if (!strcmp("-c", arg) ||
1010                     !strcmp("--count", arg)) {
1011                         opt.count = 1;
1012                         continue;
1013                 }
1014                 if (!strcmp("-w", arg) ||
1015                     !strcmp("--word-regexp", arg)) {
1016                         opt.word_regexp = 1;
1017                         continue;
1018                 }
1019                 if (!strncmp("-A", arg, 2) ||
1020                     !strncmp("-B", arg, 2) ||
1021                     !strncmp("-C", arg, 2) ||
1022                     (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
1023                         unsigned num;
1024                         const char *scan;
1025                         switch (arg[1]) {
1026                         case 'A': case 'B': case 'C':
1027                                 if (!arg[2]) {
1028                                         if (argc <= 1)
1029                                                 die(emsg_missing_context_len);
1030                                         scan = *++argv;
1031                                         argc--;
1032                                 }
1033                                 else
1034                                         scan = arg + 2;
1035                                 break;
1036                         default:
1037                                 scan = arg + 1;
1038                                 break;
1039                         }
1040                         if (sscanf(scan, "%u", &num) != 1)
1041                                 die(emsg_invalid_context_len, scan);
1042                         switch (arg[1]) {
1043                         case 'A':
1044                                 opt.post_context = num;
1045                                 break;
1046                         default:
1047                         case 'C':
1048                                 opt.post_context = num;
1049                         case 'B':
1050                                 opt.pre_context = num;
1051                                 break;
1052                         }
1053                         continue;
1054                 }
1055                 if (!strcmp("-f", arg)) {
1056                         FILE *patterns;
1057                         int lno = 0;
1058                         char buf[1024];
1059                         if (argc <= 1)
1060                                 die(emsg_missing_argument, arg);
1061                         patterns = fopen(argv[1], "r");
1062                         if (!patterns)
1063                                 die("'%s': %s", argv[1], strerror(errno));
1064                         while (fgets(buf, sizeof(buf), patterns)) {
1065                                 int len = strlen(buf);
1066                                 if (buf[len-1] == '\n')
1067                                         buf[len-1] = 0;
1068                                 /* ignore empty line like grep does */
1069                                 if (!buf[0])
1070                                         continue;
1071                                 add_pattern(&opt, strdup(buf), argv[1], ++lno,
1072                                             GREP_PATTERN);
1073                         }
1074                         fclose(patterns);
1075                         argv++;
1076                         argc--;
1077                         continue;
1078                 }
1079                 if (!strcmp("--not", arg)) {
1080                         add_pattern(&opt, arg, "command line", 0, GREP_NOT);
1081                         continue;
1082                 }
1083                 if (!strcmp("--and", arg)) {
1084                         add_pattern(&opt, arg, "command line", 0, GREP_AND);
1085                         continue;
1086                 }
1087                 if (!strcmp("--or", arg))
1088                         continue; /* no-op */
1089                 if (!strcmp("(", arg)) {
1090                         add_pattern(&opt, arg, "command line", 0, GREP_OPEN_PAREN);
1091                         continue;
1092                 }
1093                 if (!strcmp(")", arg)) {
1094                         add_pattern(&opt, arg, "command line", 0, GREP_CLOSE_PAREN);
1095                         continue;
1096                 }
1097                 if (!strcmp("-e", arg)) {
1098                         if (1 < argc) {
1099                                 add_pattern(&opt, argv[1], "-e option", 0,
1100                                             GREP_PATTERN);
1101                                 argv++;
1102                                 argc--;
1103                                 continue;
1104                         }
1105                         die(emsg_missing_argument, arg);
1106                 }
1107                 if (!strcmp("--", arg)) {
1108                         /* later processing wants to have this at argv[1] */
1109                         argv--;
1110                         argc++;
1111                         break;
1112                 }
1113                 if (*arg == '-')
1114                         usage(builtin_grep_usage);
1115
1116                 /* First unrecognized non-option token */
1117                 if (!opt.pattern_list) {
1118                         add_pattern(&opt, arg, "command line", 0,
1119                                     GREP_PATTERN);
1120                         break;
1121                 }
1122                 else {
1123                         /* We are looking at the first path or rev;
1124                          * it is found at argv[1] after leaving the
1125                          * loop.
1126                          */
1127                         argc++; argv--;
1128                         break;
1129                 }
1130         }
1131
1132         if (!opt.pattern_list)
1133                 die("no pattern given.");
1134         if ((opt.regflags != REG_NEWLINE) && opt.fixed)
1135                 die("cannot mix --fixed-strings and regexp");
1136         if (!opt.fixed)
1137                 compile_patterns(&opt);
1138
1139         /* Check revs and then paths */
1140         for (i = 1; i < argc; i++) {
1141                 const char *arg = argv[i];
1142                 unsigned char sha1[20];
1143                 /* Is it a rev? */
1144                 if (!get_sha1(arg, sha1)) {
1145                         struct object *object = parse_object(sha1);
1146                         if (!object)
1147                                 die("bad object %s", arg);
1148                         add_object_array(object, arg, &list);
1149                         continue;
1150                 }
1151                 if (!strcmp(arg, "--")) {
1152                         i++;
1153                         seen_dashdash = 1;
1154                 }
1155                 break;
1156         }
1157
1158         /* The rest are paths */
1159         if (!seen_dashdash) {
1160                 int j;
1161                 for (j = i; j < argc; j++)
1162                         verify_filename(prefix, argv[j]);
1163         }
1164
1165         if (i < argc)
1166                 paths = get_pathspec(prefix, argv + i);
1167         else if (prefix) {
1168                 paths = xcalloc(2, sizeof(const char *));
1169                 paths[0] = prefix;
1170                 paths[1] = NULL;
1171         }
1172
1173         if (!list.nr)
1174                 return !grep_cache(&opt, paths, cached);
1175
1176         if (cached)
1177                 die("both --cached and trees are given.");
1178
1179         for (i = 0; i < list.nr; i++) {
1180                 struct object *real_obj;
1181                 real_obj = deref_tag(list.objects[i].item, NULL, 0);
1182                 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
1183                         hit = 1;
1184         }
1185         return !hit;
1186 }