grep.c

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