grep.c

   1 #include "cache.h"
   2 #include "grep.h"
   3 #include "userdiff.h"
   4 #include "xdiff-interface.h"
   5
   6 static int grep_source_load(struct grep_source *gs);
   7 static int grep_source_is_binary(struct grep_source *gs);
   8
   9
  10 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
  11                                         const char *origin, int no,
  12                                         enum grep_pat_token t,
  13                                         enum grep_header_field field)
  14 {
  15         struct grep_pat *p = xcalloc(1, sizeof(*p));
  16         p->pattern = xmemdupz(pat, patlen);
  17         p->patternlen = patlen;
  18         p->origin = origin;
  19         p->no = no;
  20         p->token = t;
  21         p->field = field;
  22         return p;
  23 }
  24
  25 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
  26 {
  27         **tail = p;
  28         *tail = &p->next;
  29         p->next = NULL;
  30
  31         switch (p->token) {
  32         case GREP_PATTERN: /* atom */
  33         case GREP_PATTERN_HEAD:
  34         case GREP_PATTERN_BODY:
  35                 for (;;) {
  36                         struct grep_pat *new_pat;
  37                         size_t len = 0;
  38                         char *cp = p->pattern + p->patternlen, *nl = NULL;
  39                         while (++len <= p->patternlen) {
  40                                 if (*(--cp) == '\n') {
  41                                         nl = cp;
  42                                         break;
  43                                 }
  44                         }
  45                         if (!nl)
  46                                 break;
  47                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
  48                                                   p->no, p->token, p->field);
  49                         new_pat->next = p->next;
  50                         if (!p->next)
  51                                 *tail = &new_pat->next;
  52                         p->next = new_pat;
  53                         *nl = '\0';
  54                         p->patternlen -= len;
  55                 }
  56                 break;
  57         default:
  58                 break;
  59         }
  60 }
  61
  62 void append_header_grep_pattern(struct grep_opt *opt,
  63                                 enum grep_header_field field, const char *pat)
  64 {
  65         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
  66                                              GREP_PATTERN_HEAD, field);
  67         do_append_grep_pat(&opt->header_tail, p);
  68 }
  69
  70 void append_grep_pattern(struct grep_opt *opt, const char *pat,
  71                          const char *origin, int no, enum grep_pat_token t)
  72 {
  73         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
  74 }
  75
  76 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
  77                      const char *origin, int no, enum grep_pat_token t)
  78 {
  79         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
  80         do_append_grep_pat(&opt->pattern_tail, p);
  81 }
  82
  83 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
  84 {
  85         struct grep_pat *pat;
  86         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
  87         *ret = *opt;
  88
  89         ret->pattern_list = NULL;
  90         ret->pattern_tail = &ret->pattern_list;
  91
  92         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
  93         {
  94                 if(pat->token == GREP_PATTERN_HEAD)
  95                         append_header_grep_pattern(ret, pat->field,
  96                                                    pat->pattern);
  97                 else
  98                         append_grep_pat(ret, pat->pattern, pat->patternlen,
  99                                         pat->origin, pat->no, pat->token);
 100         }
 101
 102         return ret;
 103 }
 104
 105 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 106                 const char *error)
 107 {
 108         char where[1024];
 109
 110         if (p->no)
 111                 sprintf(where, "In '%s' at %d, ", p->origin, p->no);
 112         else if (p->origin)
 113                 sprintf(where, "%s, ", p->origin);
 114         else
 115                 where[0] = 0;
 116
 117         die("%s'%s': %s", where, p->pattern, error);
 118 }
 119
 120 #ifdef USE_LIBPCRE
 121 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 122 {
 123         const char *error;
 124         int erroffset;
 125         int options = PCRE_MULTILINE;
 126
 127         if (opt->ignore_case)
 128                 options |= PCRE_CASELESS;
 129
 130         p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 131                         NULL);
 132         if (!p->pcre_regexp)
 133                 compile_regexp_failed(p, error);
 134
 135         p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 136         if (!p->pcre_extra_info && error)
 137                 die("%s", error);
 138 }
 139
 140 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 141                 regmatch_t *match, int eflags)
 142 {
 143         int ovector[30], ret, flags = 0;
 144
 145         if (eflags & REG_NOTBOL)
 146                 flags |= PCRE_NOTBOL;
 147
 148         ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 149                         0, flags, ovector, ARRAY_SIZE(ovector));
 150         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 151                 die("pcre_exec failed with error code %d", ret);
 152         if (ret > 0) {
 153                 ret = 0;
 154                 match->rm_so = ovector[0];
 155                 match->rm_eo = ovector[1];
 156         }
 157
 158         return ret;
 159 }
 160
 161 static void free_pcre_regexp(struct grep_pat *p)
 162 {
 163         pcre_free(p->pcre_regexp);
 164         pcre_free(p->pcre_extra_info);
 165 }
 166 #else /* !USE_LIBPCRE */
 167 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 168 {
 169         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 170 }
 171
 172 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 173                 regmatch_t *match, int eflags)
 174 {
 175         return 1;
 176 }
 177
 178 static void free_pcre_regexp(struct grep_pat *p)
 179 {
 180 }
 181 #endif /* !USE_LIBPCRE */
 182
 183 static int is_fixed(const char *s, size_t len)
 184 {
 185         size_t i;
 186
 187         /* regcomp cannot accept patterns with NULs so we
 188          * consider any pattern containing a NUL fixed.
 189          */
 190         if (memchr(s, 0, len))
 191                 return 1;
 192
 193         for (i = 0; i < len; i++) {
 194                 if (is_regex_special(s[i]))
 195                         return 0;
 196         }
 197
 198         return 1;
 199 }
 200
 201 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 202 {
 203         int err;
 204
 205         p->word_regexp = opt->word_regexp;
 206         p->ignore_case = opt->ignore_case;
 207
 208         if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 209                 p->fixed = 1;
 210         else
 211                 p->fixed = 0;
 212
 213         if (p->fixed) {
 214                 if (opt->regflags & REG_ICASE || p->ignore_case)
 215                         p->kws = kwsalloc(tolower_trans_tbl);
 216                 else
 217                         p->kws = kwsalloc(NULL);
 218                 kwsincr(p->kws, p->pattern, p->patternlen);
 219                 kwsprep(p->kws);
 220                 return;
 221         }
 222
 223         if (opt->pcre) {
 224                 compile_pcre_regexp(p, opt);
 225                 return;
 226         }
 227
 228         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 229         if (err) {
 230                 char errbuf[1024];
 231                 regerror(err, &p->regexp, errbuf, 1024);
 232                 regfree(&p->regexp);
 233                 compile_regexp_failed(p, errbuf);
 234         }
 235 }
 236
 237 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 238 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 239 {
 240         struct grep_pat *p;
 241         struct grep_expr *x;
 242
 243         p = *list;
 244         if (!p)
 245                 return NULL;
 246         switch (p->token) {
 247         case GREP_PATTERN: /* atom */
 248         case GREP_PATTERN_HEAD:
 249         case GREP_PATTERN_BODY:
 250                 x = xcalloc(1, sizeof (struct grep_expr));
 251                 x->node = GREP_NODE_ATOM;
 252                 x->u.atom = p;
 253                 *list = p->next;
 254                 return x;
 255         case GREP_OPEN_PAREN:
 256                 *list = p->next;
 257                 x = compile_pattern_or(list);
 258                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 259                         die("unmatched parenthesis");
 260                 *list = (*list)->next;
 261                 return x;
 262         default:
 263                 return NULL;
 264         }
 265 }
 266
 267 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 268 {
 269         struct grep_pat *p;
 270         struct grep_expr *x;
 271
 272         p = *list;
 273         if (!p)
 274                 return NULL;
 275         switch (p->token) {
 276         case GREP_NOT:
 277                 if (!p->next)
 278                         die("--not not followed by pattern expression");
 279                 *list = p->next;
 280                 x = xcalloc(1, sizeof (struct grep_expr));
 281                 x->node = GREP_NODE_NOT;
 282                 x->u.unary = compile_pattern_not(list);
 283                 if (!x->u.unary)
 284                         die("--not followed by non pattern expression");
 285                 return x;
 286         default:
 287                 return compile_pattern_atom(list);
 288         }
 289 }
 290
 291 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 292 {
 293         struct grep_pat *p;
 294         struct grep_expr *x, *y, *z;
 295
 296         x = compile_pattern_not(list);
 297         p = *list;
 298         if (p && p->token == GREP_AND) {
 299                 if (!p->next)
 300                         die("--and not followed by pattern expression");
 301                 *list = p->next;
 302                 y = compile_pattern_and(list);
 303                 if (!y)
 304                         die("--and not followed by pattern expression");
 305                 z = xcalloc(1, sizeof (struct grep_expr));
 306                 z->node = GREP_NODE_AND;
 307                 z->u.binary.left = x;
 308                 z->u.binary.right = y;
 309                 return z;
 310         }
 311         return x;
 312 }
 313
 314 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 315 {
 316         struct grep_pat *p;
 317         struct grep_expr *x, *y, *z;
 318
 319         x = compile_pattern_and(list);
 320         p = *list;
 321         if (x && p && p->token != GREP_CLOSE_PAREN) {
 322                 y = compile_pattern_or(list);
 323                 if (!y)
 324                         die("not a pattern expression %s", p->pattern);
 325                 z = xcalloc(1, sizeof (struct grep_expr));
 326                 z->node = GREP_NODE_OR;
 327                 z->u.binary.left = x;
 328                 z->u.binary.right = y;
 329                 return z;
 330         }
 331         return x;
 332 }
 333
 334 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 335 {
 336         return compile_pattern_or(list);
 337 }
 338
 339 static void indent(int in)
 340 {
 341         while (in-- > 0)
 342                 fputc(' ', stderr);
 343 }
 344
 345 static void dump_grep_pat(struct grep_pat *p)
 346 {
 347         switch (p->token) {
 348         case GREP_AND: fprintf(stderr, "*and*"); break;
 349         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 350         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 351         case GREP_NOT: fprintf(stderr, "*not*"); break;
 352         case GREP_OR: fprintf(stderr, "*or*"); break;
 353
 354         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 355         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 356         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 357         }
 358
 359         switch (p->token) {
 360         default: break;
 361         case GREP_PATTERN_HEAD:
 362                 fprintf(stderr, "<head %d>", p->field); break;
 363         case GREP_PATTERN_BODY:
 364                 fprintf(stderr, "<body>"); break;
 365         }
 366         switch (p->token) {
 367         default: break;
 368         case GREP_PATTERN_HEAD:
 369         case GREP_PATTERN_BODY:
 370         case GREP_PATTERN:
 371                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 372                 break;
 373         }
 374         fputc('\n', stderr);
 375 }
 376
 377 static void dump_grep_expression_1(struct grep_expr *x, int in)
 378 {
 379         indent(in);
 380         switch (x->node) {
 381         case GREP_NODE_TRUE:
 382                 fprintf(stderr, "true\n");
 383                 break;
 384         case GREP_NODE_ATOM:
 385                 dump_grep_pat(x->u.atom);
 386                 break;
 387         case GREP_NODE_NOT:
 388                 fprintf(stderr, "(not\n");
 389                 dump_grep_expression_1(x->u.unary, in+1);
 390                 indent(in);
 391                 fprintf(stderr, ")\n");
 392                 break;
 393         case GREP_NODE_AND:
 394                 fprintf(stderr, "(and\n");
 395                 dump_grep_expression_1(x->u.binary.left, in+1);
 396                 dump_grep_expression_1(x->u.binary.right, in+1);
 397                 indent(in);
 398                 fprintf(stderr, ")\n");
 399                 break;
 400         case GREP_NODE_OR:
 401                 fprintf(stderr, "(or\n");
 402                 dump_grep_expression_1(x->u.binary.left, in+1);
 403                 dump_grep_expression_1(x->u.binary.right, in+1);
 404                 indent(in);
 405                 fprintf(stderr, ")\n");
 406                 break;
 407         }
 408 }
 409
 410 static void dump_grep_expression(struct grep_opt *opt)
 411 {
 412         struct grep_expr *x = opt->pattern_expression;
 413
 414         if (opt->all_match)
 415                 fprintf(stderr, "[all-match]\n");
 416         dump_grep_expression_1(x, 0);
 417         fflush(NULL);
 418 }
 419
 420 static struct grep_expr *grep_true_expr(void)
 421 {
 422         struct grep_expr *z = xcalloc(1, sizeof(*z));
 423         z->node = GREP_NODE_TRUE;
 424         return z;
 425 }
 426
 427 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 428 {
 429         struct grep_expr *z = xcalloc(1, sizeof(*z));
 430         z->node = GREP_NODE_OR;
 431         z->u.binary.left = left;
 432         z->u.binary.right = right;
 433         return z;
 434 }
 435
 436 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 437 {
 438         struct grep_pat *p;
 439         struct grep_expr *header_expr;
 440         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 441         enum grep_header_field fld;
 442
 443         if (!opt->header_list)
 444                 return NULL;
 445
 446         for (p = opt->header_list; p; p = p->next) {
 447                 if (p->token != GREP_PATTERN_HEAD)
 448                         die("bug: a non-header pattern in grep header list.");
 449                 if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
 450                         die("bug: unknown header field %d", p->field);
 451                 compile_regexp(p, opt);
 452         }
 453
 454         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 455                 header_group[fld] = NULL;
 456
 457         for (p = opt->header_list; p; p = p->next) {
 458                 struct grep_expr *h;
 459                 struct grep_pat *pp = p;
 460
 461                 h = compile_pattern_atom(&pp);
 462                 if (!h || pp != p->next)
 463                         die("bug: malformed header expr");
 464                 if (!header_group[p->field]) {
 465                         header_group[p->field] = h;
 466                         continue;
 467                 }
 468                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 469         }
 470
 471         header_expr = NULL;
 472
 473         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 474                 if (!header_group[fld])
 475                         continue;
 476                 if (!header_expr)
 477                         header_expr = grep_true_expr();
 478                 header_expr = grep_or_expr(header_group[fld], header_expr);
 479         }
 480         return header_expr;
 481 }
 482
 483 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 484 {
 485         struct grep_expr *z = x;
 486
 487         while (x) {
 488                 assert(x->node == GREP_NODE_OR);
 489                 if (x->u.binary.right &&
 490                     x->u.binary.right->node == GREP_NODE_TRUE) {
 491                         x->u.binary.right = y;
 492                         break;
 493                 }
 494                 x = x->u.binary.right;
 495         }
 496         return z;
 497 }
 498
 499 static void compile_grep_patterns_real(struct grep_opt *opt)
 500 {
 501         struct grep_pat *p;
 502         struct grep_expr *header_expr = prep_header_patterns(opt);
 503
 504         for (p = opt->pattern_list; p; p = p->next) {
 505                 switch (p->token) {
 506                 case GREP_PATTERN: /* atom */
 507                 case GREP_PATTERN_HEAD:
 508                 case GREP_PATTERN_BODY:
 509                         compile_regexp(p, opt);
 510                         break;
 511                 default:
 512                         opt->extended = 1;
 513                         break;
 514                 }
 515         }
 516
 517         if (opt->all_match || header_expr)
 518                 opt->extended = 1;
 519         else if (!opt->extended && !opt->debug)
 520                 return;
 521
 522         p = opt->pattern_list;
 523         if (p)
 524                 opt->pattern_expression = compile_pattern_expr(&p);
 525         if (p)
 526                 die("incomplete pattern expression: %s", p->pattern);
 527
 528         if (!header_expr)
 529                 return;
 530
 531         if (!opt->pattern_expression)
 532                 opt->pattern_expression = header_expr;
 533         else if (opt->all_match)
 534                 opt->pattern_expression = grep_splice_or(header_expr,
 535                                                          opt->pattern_expression);
 536         else
 537                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 538                                                        header_expr);
 539         opt->all_match = 1;
 540 }
 541
 542 void compile_grep_patterns(struct grep_opt *opt)
 543 {
 544         compile_grep_patterns_real(opt);
 545         if (opt->debug)
 546                 dump_grep_expression(opt);
 547 }
 548
 549 static void free_pattern_expr(struct grep_expr *x)
 550 {
 551         switch (x->node) {
 552         case GREP_NODE_TRUE:
 553         case GREP_NODE_ATOM:
 554                 break;
 555         case GREP_NODE_NOT:
 556                 free_pattern_expr(x->u.unary);
 557                 break;
 558         case GREP_NODE_AND:
 559         case GREP_NODE_OR:
 560                 free_pattern_expr(x->u.binary.left);
 561                 free_pattern_expr(x->u.binary.right);
 562                 break;
 563         }
 564         free(x);
 565 }
 566
 567 void free_grep_patterns(struct grep_opt *opt)
 568 {
 569         struct grep_pat *p, *n;
 570
 571         for (p = opt->pattern_list; p; p = n) {
 572                 n = p->next;
 573                 switch (p->token) {
 574                 case GREP_PATTERN: /* atom */
 575                 case GREP_PATTERN_HEAD:
 576                 case GREP_PATTERN_BODY:
 577                         if (p->kws)
 578                                 kwsfree(p->kws);
 579                         else if (p->pcre_regexp)
 580                                 free_pcre_regexp(p);
 581                         else
 582                                 regfree(&p->regexp);
 583                         free(p->pattern);
 584                         break;
 585                 default:
 586                         break;
 587                 }
 588                 free(p);
 589         }
 590
 591         if (!opt->extended)
 592                 return;
 593         free_pattern_expr(opt->pattern_expression);
 594 }
 595
 596 static char *end_of_line(char *cp, unsigned long *left)
 597 {
 598         unsigned long l = *left;
 599         while (l && *cp != '\n') {
 600                 l--;
 601                 cp++;
 602         }
 603         *left = l;
 604         return cp;
 605 }
 606
 607 static int word_char(char ch)
 608 {
 609         return isalnum(ch) || ch == '_';
 610 }
 611
 612 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 613                          const char *color)
 614 {
 615         if (want_color(opt->color) && color && color[0]) {
 616                 opt->output(opt, color, strlen(color));
 617                 opt->output(opt, data, size);
 618                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 619         } else
 620                 opt->output(opt, data, size);
 621 }
 622
 623 static void output_sep(struct grep_opt *opt, char sign)
 624 {
 625         if (opt->null_following_name)
 626                 opt->output(opt, "\0", 1);
 627         else
 628                 output_color(opt, &sign, 1, opt->color_sep);
 629 }
 630
 631 static void show_name(struct grep_opt *opt, const char *name)
 632 {
 633         output_color(opt, name, strlen(name), opt->color_filename);
 634         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 635 }
 636
 637 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 638                     regmatch_t *match)
 639 {
 640         struct kwsmatch kwsm;
 641         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 642         if (offset == -1) {
 643                 match->rm_so = match->rm_eo = -1;
 644                 return REG_NOMATCH;
 645         } else {
 646                 match->rm_so = offset;
 647                 match->rm_eo = match->rm_so + kwsm.size[0];
 648                 return 0;
 649         }
 650 }
 651
 652 static int regmatch(const regex_t *preg, char *line, char *eol,
 653                     regmatch_t *match, int eflags)
 654 {
 655 #ifdef REG_STARTEND
 656         match->rm_so = 0;
 657         match->rm_eo = eol - line;
 658         eflags |= REG_STARTEND;
 659 #endif
 660         return regexec(preg, line, 1, match, eflags);
 661 }
 662
 663 static int patmatch(struct grep_pat *p, char *line, char *eol,
 664                     regmatch_t *match, int eflags)
 665 {
 666         int hit;
 667
 668         if (p->fixed)
 669                 hit = !fixmatch(p, line, eol, match);
 670         else if (p->pcre_regexp)
 671                 hit = !pcrematch(p, line, eol, match, eflags);
 672         else
 673                 hit = !regmatch(&p->regexp, line, eol, match, eflags);
 674
 675         return hit;
 676 }
 677
 678 static int strip_timestamp(char *bol, char **eol_p)
 679 {
 680         char *eol = *eol_p;
 681         int ch;
 682
 683         while (bol < --eol) {
 684                 if (*eol != '>')
 685                         continue;
 686                 *eol_p = ++eol;
 687                 ch = *eol;
 688                 *eol = '\0';
 689                 return ch;
 690         }
 691         return 0;
 692 }
 693
 694 static struct {
 695         const char *field;
 696         size_t len;
 697 } header_field[] = {
 698         { "author ", 7 },
 699         { "committer ", 10 },
 700 };
 701
 702 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 703                              enum grep_context ctx,
 704                              regmatch_t *pmatch, int eflags)
 705 {
 706         int hit = 0;
 707         int saved_ch = 0;
 708         const char *start = bol;
 709
 710         if ((p->token != GREP_PATTERN) &&
 711             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 712                 return 0;
 713
 714         if (p->token == GREP_PATTERN_HEAD) {
 715                 const char *field;
 716                 size_t len;
 717                 assert(p->field < ARRAY_SIZE(header_field));
 718                 field = header_field[p->field].field;
 719                 len = header_field[p->field].len;
 720                 if (strncmp(bol, field, len))
 721                         return 0;
 722                 bol += len;
 723                 switch (p->field) {
 724                 case GREP_HEADER_AUTHOR:
 725                 case GREP_HEADER_COMMITTER:
 726                         saved_ch = strip_timestamp(bol, &eol);
 727                         break;
 728                 default:
 729                         break;
 730                 }
 731         }
 732
 733  again:
 734         hit = patmatch(p, bol, eol, pmatch, eflags);
 735
 736         if (hit && p->word_regexp) {
 737                 if ((pmatch[0].rm_so < 0) ||
 738                     (eol - bol) < pmatch[0].rm_so ||
 739                     (pmatch[0].rm_eo < 0) ||
 740                     (eol - bol) < pmatch[0].rm_eo)
 741                         die("regexp returned nonsense");
 742
 743                 /* Match beginning must be either beginning of the
 744                  * line, or at word boundary (i.e. the last char must
 745                  * not be a word char).  Similarly, match end must be
 746                  * either end of the line, or at word boundary
 747                  * (i.e. the next char must not be a word char).
 748                  */
 749                 if ( ((pmatch[0].rm_so == 0) ||
 750                       !word_char(bol[pmatch[0].rm_so-1])) &&
 751                      ((pmatch[0].rm_eo == (eol-bol)) ||
 752                       !word_char(bol[pmatch[0].rm_eo])) )
 753                         ;
 754                 else
 755                         hit = 0;
 756
 757                 /* Words consist of at least one character. */
 758                 if (pmatch->rm_so == pmatch->rm_eo)
 759                         hit = 0;
 760
 761                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 762                         /* There could be more than one match on the
 763                          * line, and the first match might not be
 764                          * strict word match.  But later ones could be!
 765                          * Forward to the next possible start, i.e. the
 766                          * next position following a non-word char.
 767                          */
 768                         bol = pmatch[0].rm_so + bol + 1;
 769                         while (word_char(bol[-1]) && bol < eol)
 770                                 bol++;
 771                         eflags |= REG_NOTBOL;
 772                         if (bol < eol)
 773                                 goto again;
 774                 }
 775         }
 776         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 777                 *eol = saved_ch;
 778         if (hit) {
 779                 pmatch[0].rm_so += bol - start;
 780                 pmatch[0].rm_eo += bol - start;
 781         }
 782         return hit;
 783 }
 784
 785 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 786                            enum grep_context ctx, int collect_hits)
 787 {
 788         int h = 0;
 789         regmatch_t match;
 790
 791         if (!x)
 792                 die("Not a valid grep expression");
 793         switch (x->node) {
 794         case GREP_NODE_TRUE:
 795                 h = 1;
 796                 break;
 797         case GREP_NODE_ATOM:
 798                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 799                 break;
 800         case GREP_NODE_NOT:
 801                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 802                 break;
 803         case GREP_NODE_AND:
 804                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 805                         return 0;
 806                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 807                 break;
 808         case GREP_NODE_OR:
 809                 if (!collect_hits)
 810                         return (match_expr_eval(x->u.binary.left,
 811                                                 bol, eol, ctx, 0) ||
 812                                 match_expr_eval(x->u.binary.right,
 813                                                 bol, eol, ctx, 0));
 814                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 815                 x->u.binary.left->hit |= h;
 816                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 817                 break;
 818         default:
 819                 die("Unexpected node type (internal error) %d", x->node);
 820         }
 821         if (collect_hits)
 822                 x->hit |= h;
 823         return h;
 824 }
 825
 826 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 827                       enum grep_context ctx, int collect_hits)
 828 {
 829         struct grep_expr *x = opt->pattern_expression;
 830         return match_expr_eval(x, bol, eol, ctx, collect_hits);
 831 }
 832
 833 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 834                       enum grep_context ctx, int collect_hits)
 835 {
 836         struct grep_pat *p;
 837         regmatch_t match;
 838
 839         if (opt->extended)
 840                 return match_expr(opt, bol, eol, ctx, collect_hits);
 841
 842         /* we do not call with collect_hits without being extended */
 843         for (p = opt->pattern_list; p; p = p->next) {
 844                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 845                         return 1;
 846         }
 847         return 0;
 848 }
 849
 850 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 851                               enum grep_context ctx,
 852                               regmatch_t *pmatch, int eflags)
 853 {
 854         regmatch_t match;
 855
 856         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 857                 return 0;
 858         if (match.rm_so < 0 || match.rm_eo < 0)
 859                 return 0;
 860         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 861                 if (match.rm_so > pmatch->rm_so)
 862                         return 1;
 863                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 864                         return 1;
 865         }
 866         pmatch->rm_so = match.rm_so;
 867         pmatch->rm_eo = match.rm_eo;
 868         return 1;
 869 }
 870
 871 static int next_match(struct grep_opt *opt, char *bol, char *eol,
 872                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
 873 {
 874         struct grep_pat *p;
 875         int hit = 0;
 876
 877         pmatch->rm_so = pmatch->rm_eo = -1;
 878         if (bol < eol) {
 879                 for (p = opt->pattern_list; p; p = p->next) {
 880                         switch (p->token) {
 881                         case GREP_PATTERN: /* atom */
 882                         case GREP_PATTERN_HEAD:
 883                         case GREP_PATTERN_BODY:
 884                                 hit |= match_next_pattern(p, bol, eol, ctx,
 885                                                           pmatch, eflags);
 886                                 break;
 887                         default:
 888                                 break;
 889                         }
 890                 }
 891         }
 892         return hit;
 893 }
 894
 895 static void show_line(struct grep_opt *opt, char *bol, char *eol,
 896                       const char *name, unsigned lno, char sign)
 897 {
 898         int rest = eol - bol;
 899         char *line_color = NULL;
 900
 901         if (opt->file_break && opt->last_shown == 0) {
 902                 if (opt->show_hunk_mark)
 903                         opt->output(opt, "\n", 1);
 904         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
 905                 if (opt->last_shown == 0) {
 906                         if (opt->show_hunk_mark) {
 907                                 output_color(opt, "--", 2, opt->color_sep);
 908                                 opt->output(opt, "\n", 1);
 909                         }
 910                 } else if (lno > opt->last_shown + 1) {
 911                         output_color(opt, "--", 2, opt->color_sep);
 912                         opt->output(opt, "\n", 1);
 913                 }
 914         }
 915         if (opt->heading && opt->last_shown == 0) {
 916                 output_color(opt, name, strlen(name), opt->color_filename);
 917                 opt->output(opt, "\n", 1);
 918         }
 919         opt->last_shown = lno;
 920
 921         if (!opt->heading && opt->pathname) {
 922                 output_color(opt, name, strlen(name), opt->color_filename);
 923                 output_sep(opt, sign);
 924         }
 925         if (opt->linenum) {
 926                 char buf[32];
 927                 snprintf(buf, sizeof(buf), "%d", lno);
 928                 output_color(opt, buf, strlen(buf), opt->color_lineno);
 929                 output_sep(opt, sign);
 930         }
 931         if (opt->color) {
 932                 regmatch_t match;
 933                 enum grep_context ctx = GREP_CONTEXT_BODY;
 934                 int ch = *eol;
 935                 int eflags = 0;
 936
 937                 if (sign == ':')
 938                         line_color = opt->color_selected;
 939                 else if (sign == '-')
 940                         line_color = opt->color_context;
 941                 else if (sign == '=')
 942                         line_color = opt->color_function;
 943                 *eol = '\0';
 944                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
 945                         if (match.rm_so == match.rm_eo)
 946                                 break;
 947
 948                         output_color(opt, bol, match.rm_so, line_color);
 949                         output_color(opt, bol + match.rm_so,
 950                                      match.rm_eo - match.rm_so,
 951                                      opt->color_match);
 952                         bol += match.rm_eo;
 953                         rest -= match.rm_eo;
 954                         eflags = REG_NOTBOL;
 955                 }
 956                 *eol = ch;
 957         }
 958         output_color(opt, bol, rest, line_color);
 959         opt->output(opt, "\n", 1);
 960 }
 961
 962 #ifndef NO_PTHREADS
 963 int grep_use_locks;
 964
 965 /*
 966  * This lock protects access to the gitattributes machinery, which is
 967  * not thread-safe.
 968  */
 969 pthread_mutex_t grep_attr_mutex;
 970
 971 static inline void grep_attr_lock(void)
 972 {
 973         if (grep_use_locks)
 974                 pthread_mutex_lock(&grep_attr_mutex);
 975 }
 976
 977 static inline void grep_attr_unlock(void)
 978 {
 979         if (grep_use_locks)
 980                 pthread_mutex_unlock(&grep_attr_mutex);
 981 }
 982
 983 /*
 984  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
 985  */
 986 pthread_mutex_t grep_read_mutex;
 987
 988 #else
 989 #define grep_attr_lock()
 990 #define grep_attr_unlock()
 991 #endif
 992
 993 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
 994 {
 995         xdemitconf_t *xecfg = opt->priv;
 996         if (xecfg && !xecfg->find_func) {
 997                 grep_source_load_driver(gs);
 998                 if (gs->driver->funcname.pattern) {
 999                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1000                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1001                 } else {
1002                         xecfg = opt->priv = NULL;
1003                 }
1004         }
1005
1006         if (xecfg) {
1007                 char buf[1];
1008                 return xecfg->find_func(bol, eol - bol, buf, 1,
1009                                         xecfg->find_func_priv) >= 0;
1010         }
1011
1012         if (bol == eol)
1013                 return 0;
1014         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1015                 return 1;
1016         return 0;
1017 }
1018
1019 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1020                                char *bol, unsigned lno)
1021 {
1022         while (bol > gs->buf) {
1023                 char *eol = --bol;
1024
1025                 while (bol > gs->buf && bol[-1] != '\n')
1026                         bol--;
1027                 lno--;
1028
1029                 if (lno <= opt->last_shown)
1030                         break;
1031
1032                 if (match_funcname(opt, gs, bol, eol)) {
1033                         show_line(opt, bol, eol, gs->name, lno, '=');
1034                         break;
1035                 }
1036         }
1037 }
1038
1039 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1040                              char *bol, char *end, unsigned lno)
1041 {
1042         unsigned cur = lno, from = 1, funcname_lno = 0;
1043         int funcname_needed = !!opt->funcname;
1044
1045         if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1046                 funcname_needed = 2;
1047
1048         if (opt->pre_context < lno)
1049                 from = lno - opt->pre_context;
1050         if (from <= opt->last_shown)
1051                 from = opt->last_shown + 1;
1052
1053         /* Rewind. */
1054         while (bol > gs->buf &&
1055                cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1056                 char *eol = --bol;
1057
1058                 while (bol > gs->buf && bol[-1] != '\n')
1059                         bol--;
1060                 cur--;
1061                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1062                         funcname_lno = cur;
1063                         funcname_needed = 0;
1064                 }
1065         }
1066
1067         /* We need to look even further back to find a function signature. */
1068         if (opt->funcname && funcname_needed)
1069                 show_funcname_line(opt, gs, bol, cur);
1070
1071         /* Back forward. */
1072         while (cur < lno) {
1073                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1074
1075                 while (*eol != '\n')
1076                         eol++;
1077                 show_line(opt, bol, eol, gs->name, cur, sign);
1078                 bol = eol + 1;
1079                 cur++;
1080         }
1081 }
1082
1083 static int should_lookahead(struct grep_opt *opt)
1084 {
1085         struct grep_pat *p;
1086
1087         if (opt->extended)
1088                 return 0; /* punt for too complex stuff */
1089         if (opt->invert)
1090                 return 0;
1091         for (p = opt->pattern_list; p; p = p->next) {
1092                 if (p->token != GREP_PATTERN)
1093                         return 0; /* punt for "header only" and stuff */
1094         }
1095         return 1;
1096 }
1097
1098 static int look_ahead(struct grep_opt *opt,
1099                       unsigned long *left_p,
1100                       unsigned *lno_p,
1101                       char **bol_p)
1102 {
1103         unsigned lno = *lno_p;
1104         char *bol = *bol_p;
1105         struct grep_pat *p;
1106         char *sp, *last_bol;
1107         regoff_t earliest = -1;
1108
1109         for (p = opt->pattern_list; p; p = p->next) {
1110                 int hit;
1111                 regmatch_t m;
1112
1113                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1114                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1115                         continue;
1116                 if (earliest < 0 || m.rm_so < earliest)
1117                         earliest = m.rm_so;
1118         }
1119
1120         if (earliest < 0) {
1121                 *bol_p = bol + *left_p;
1122                 *left_p = 0;
1123                 return 1;
1124         }
1125         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1126                 ; /* find the beginning of the line */
1127         last_bol = sp;
1128
1129         for (sp = bol; sp < last_bol; sp++) {
1130                 if (*sp == '\n')
1131                         lno++;
1132         }
1133         *left_p -= last_bol - bol;
1134         *bol_p = last_bol;
1135         *lno_p = lno;
1136         return 0;
1137 }
1138
1139 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1140 {
1141         fwrite(buf, size, 1, stdout);
1142 }
1143
1144 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1145 {
1146         char *bol;
1147         unsigned long left;
1148         unsigned lno = 1;
1149         unsigned last_hit = 0;
1150         int binary_match_only = 0;
1151         unsigned count = 0;
1152         int try_lookahead = 0;
1153         int show_function = 0;
1154         enum grep_context ctx = GREP_CONTEXT_HEAD;
1155         xdemitconf_t xecfg;
1156
1157         if (!opt->output)
1158                 opt->output = std_output;
1159
1160         if (opt->pre_context || opt->post_context || opt->file_break ||
1161             opt->funcbody) {
1162                 /* Show hunk marks, except for the first file. */
1163                 if (opt->last_shown)
1164                         opt->show_hunk_mark = 1;
1165                 /*
1166                  * If we're using threads then we can't easily identify
1167                  * the first file.  Always put hunk marks in that case
1168                  * and skip the very first one later in work_done().
1169                  */
1170                 if (opt->output != std_output)
1171                         opt->show_hunk_mark = 1;
1172         }
1173         opt->last_shown = 0;
1174
1175         switch (opt->binary) {
1176         case GREP_BINARY_DEFAULT:
1177                 if (grep_source_is_binary(gs))
1178                         binary_match_only = 1;
1179                 break;
1180         case GREP_BINARY_NOMATCH:
1181                 if (grep_source_is_binary(gs))
1182                         return 0; /* Assume unmatch */
1183                 break;
1184         case GREP_BINARY_TEXT:
1185                 break;
1186         default:
1187                 die("bug: unknown binary handling mode");
1188         }
1189
1190         memset(&xecfg, 0, sizeof(xecfg));
1191         opt->priv = &xecfg;
1192
1193         try_lookahead = should_lookahead(opt);
1194
1195         if (grep_source_load(gs) < 0)
1196                 return 0;
1197
1198         bol = gs->buf;
1199         left = gs->size;
1200         while (left) {
1201                 char *eol, ch;
1202                 int hit;
1203
1204                 /*
1205                  * look_ahead() skips quickly to the line that possibly
1206                  * has the next hit; don't call it if we need to do
1207                  * something more than just skipping the current line
1208                  * in response to an unmatch for the current line.  E.g.
1209                  * inside a post-context window, we will show the current
1210                  * line as a context around the previous hit when it
1211                  * doesn't hit.
1212                  */
1213                 if (try_lookahead
1214                     && !(last_hit
1215                          && (show_function ||
1216                              lno <= last_hit + opt->post_context))
1217                     && look_ahead(opt, &left, &lno, &bol))
1218                         break;
1219                 eol = end_of_line(bol, &left);
1220                 ch = *eol;
1221                 *eol = 0;
1222
1223                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1224                         ctx = GREP_CONTEXT_BODY;
1225
1226                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1227                 *eol = ch;
1228
1229                 if (collect_hits)
1230                         goto next_line;
1231
1232                 /* "grep -v -e foo -e bla" should list lines
1233                  * that do not have either, so inversion should
1234                  * be done outside.
1235                  */
1236                 if (opt->invert)
1237                         hit = !hit;
1238                 if (opt->unmatch_name_only) {
1239                         if (hit)
1240                                 return 0;
1241                         goto next_line;
1242                 }
1243                 if (hit) {
1244                         count++;
1245                         if (opt->status_only)
1246                                 return 1;
1247                         if (opt->name_only) {
1248                                 show_name(opt, gs->name);
1249                                 return 1;
1250                         }
1251                         if (opt->count)
1252                                 goto next_line;
1253                         if (binary_match_only) {
1254                                 opt->output(opt, "Binary file ", 12);
1255                                 output_color(opt, gs->name, strlen(gs->name),
1256                                              opt->color_filename);
1257                                 opt->output(opt, " matches\n", 9);
1258                                 return 1;
1259                         }
1260                         /* Hit at this line.  If we haven't shown the
1261                          * pre-context lines, we would need to show them.
1262                          */
1263                         if (opt->pre_context || opt->funcbody)
1264                                 show_pre_context(opt, gs, bol, eol, lno);
1265                         else if (opt->funcname)
1266                                 show_funcname_line(opt, gs, bol, lno);
1267                         show_line(opt, bol, eol, gs->name, lno, ':');
1268                         last_hit = lno;
1269                         if (opt->funcbody)
1270                                 show_function = 1;
1271                         goto next_line;
1272                 }
1273                 if (show_function && match_funcname(opt, gs, bol, eol))
1274                         show_function = 0;
1275                 if (show_function ||
1276                     (last_hit && lno <= last_hit + opt->post_context)) {
1277                         /* If the last hit is within the post context,
1278                          * we need to show this line.
1279                          */
1280                         show_line(opt, bol, eol, gs->name, lno, '-');
1281                 }
1282
1283         next_line:
1284                 bol = eol + 1;
1285                 if (!left)
1286                         break;
1287                 left--;
1288                 lno++;
1289         }
1290
1291         if (collect_hits)
1292                 return 0;
1293
1294         if (opt->status_only)
1295                 return 0;
1296         if (opt->unmatch_name_only) {
1297                 /* We did not see any hit, so we want to show this */
1298                 show_name(opt, gs->name);
1299                 return 1;
1300         }
1301
1302         xdiff_clear_find_func(&xecfg);
1303         opt->priv = NULL;
1304
1305         /* NEEDSWORK:
1306          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1307          * which feels mostly useless but sometimes useful.  Maybe
1308          * make it another option?  For now suppress them.
1309          */
1310         if (opt->count && count) {
1311                 char buf[32];
1312                 output_color(opt, gs->name, strlen(gs->name), opt->color_filename);
1313                 output_sep(opt, ':');
1314                 snprintf(buf, sizeof(buf), "%u\n", count);
1315                 opt->output(opt, buf, strlen(buf));
1316                 return 1;
1317         }
1318         return !!last_hit;
1319 }
1320
1321 static void clr_hit_marker(struct grep_expr *x)
1322 {
1323         /* All-hit markers are meaningful only at the very top level
1324          * OR node.
1325          */
1326         while (1) {
1327                 x->hit = 0;
1328                 if (x->node != GREP_NODE_OR)
1329                         return;
1330                 x->u.binary.left->hit = 0;
1331                 x = x->u.binary.right;
1332         }
1333 }
1334
1335 static int chk_hit_marker(struct grep_expr *x)
1336 {
1337         /* Top level nodes have hit markers.  See if they all are hits */
1338         while (1) {
1339                 if (x->node != GREP_NODE_OR)
1340                         return x->hit;
1341                 if (!x->u.binary.left->hit)
1342                         return 0;
1343                 x = x->u.binary.right;
1344         }
1345 }
1346
1347 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1348 {
1349         /*
1350          * we do not have to do the two-pass grep when we do not check
1351          * buffer-wide "all-match".
1352          */
1353         if (!opt->all_match)
1354                 return grep_source_1(opt, gs, 0);
1355
1356         /* Otherwise the toplevel "or" terms hit a bit differently.
1357          * We first clear hit markers from them.
1358          */
1359         clr_hit_marker(opt->pattern_expression);
1360         grep_source_1(opt, gs, 1);
1361
1362         if (!chk_hit_marker(opt->pattern_expression))
1363                 return 0;
1364
1365         return grep_source_1(opt, gs, 0);
1366 }
1367
1368 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1369 {
1370         struct grep_source gs;
1371         int r;
1372
1373         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL);
1374         gs.buf = buf;
1375         gs.size = size;
1376
1377         r = grep_source(opt, &gs);
1378
1379         grep_source_clear(&gs);
1380         return r;
1381 }
1382
1383 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1384                       const char *name, const void *identifier)
1385 {
1386         gs->type = type;
1387         gs->name = name ? xstrdup(name) : NULL;
1388         gs->buf = NULL;
1389         gs->size = 0;
1390         gs->driver = NULL;
1391
1392         switch (type) {
1393         case GREP_SOURCE_FILE:
1394                 gs->identifier = xstrdup(identifier);
1395                 break;
1396         case GREP_SOURCE_SHA1:
1397                 gs->identifier = xmalloc(20);
1398                 memcpy(gs->identifier, identifier, 20);
1399                 break;
1400         case GREP_SOURCE_BUF:
1401                 gs->identifier = NULL;
1402         }
1403 }
1404
1405 void grep_source_clear(struct grep_source *gs)
1406 {
1407         free(gs->name);
1408         gs->name = NULL;
1409         free(gs->identifier);
1410         gs->identifier = NULL;
1411         grep_source_clear_data(gs);
1412 }
1413
1414 void grep_source_clear_data(struct grep_source *gs)
1415 {
1416         switch (gs->type) {
1417         case GREP_SOURCE_FILE:
1418         case GREP_SOURCE_SHA1:
1419                 free(gs->buf);
1420                 gs->buf = NULL;
1421                 gs->size = 0;
1422                 break;
1423         case GREP_SOURCE_BUF:
1424                 /* leave user-provided buf intact */
1425                 break;
1426         }
1427 }
1428
1429 static int grep_source_load_sha1(struct grep_source *gs)
1430 {
1431         enum object_type type;
1432
1433         grep_read_lock();
1434         gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1435         grep_read_unlock();
1436
1437         if (!gs->buf)
1438                 return error(_("'%s': unable to read %s"),
1439                              gs->name,
1440                              sha1_to_hex(gs->identifier));
1441         return 0;
1442 }
1443
1444 static int grep_source_load_file(struct grep_source *gs)
1445 {
1446         const char *filename = gs->identifier;
1447         struct stat st;
1448         char *data;
1449         size_t size;
1450         int i;
1451
1452         if (lstat(filename, &st) < 0) {
1453         err_ret:
1454                 if (errno != ENOENT)
1455                         error(_("'%s': %s"), filename, strerror(errno));
1456                 return -1;
1457         }
1458         if (!S_ISREG(st.st_mode))
1459                 return -1;
1460         size = xsize_t(st.st_size);
1461         i = open(filename, O_RDONLY);
1462         if (i < 0)
1463                 goto err_ret;
1464         data = xmalloc(size + 1);
1465         if (st.st_size != read_in_full(i, data, size)) {
1466                 error(_("'%s': short read %s"), filename, strerror(errno));
1467                 close(i);
1468                 free(data);
1469                 return -1;
1470         }
1471         close(i);
1472         data[size] = 0;
1473
1474         gs->buf = data;
1475         gs->size = size;
1476         return 0;
1477 }
1478
1479 static int grep_source_load(struct grep_source *gs)
1480 {
1481         if (gs->buf)
1482                 return 0;
1483
1484         switch (gs->type) {
1485         case GREP_SOURCE_FILE:
1486                 return grep_source_load_file(gs);
1487         case GREP_SOURCE_SHA1:
1488                 return grep_source_load_sha1(gs);
1489         case GREP_SOURCE_BUF:
1490                 return gs->buf ? 0 : -1;
1491         }
1492         die("BUG: invalid grep_source type");
1493 }
1494
1495 void grep_source_load_driver(struct grep_source *gs)
1496 {
1497         if (gs->driver)
1498                 return;
1499
1500         grep_attr_lock();
1501         gs->driver = userdiff_find_by_path(gs->name);
1502         if (!gs->driver)
1503                 gs->driver = userdiff_find_by_name("default");
1504         grep_attr_unlock();
1505 }
1506
1507 static int grep_source_is_binary(struct grep_source *gs)
1508 {
1509         grep_source_load_driver(gs);
1510         if (gs->driver->binary != -1)
1511                 return gs->driver->binary;
1512
1513         if (!grep_source_load(gs))
1514                 return buffer_is_binary(gs->buf, gs->size);
1515
1516         return 0;
1517 }