grep.c

   1 #include "cache.h"
   2 #include "grep.h"
   3 #include "userdiff.h"
   4 #include "xdiff-interface.h"
   5
   6 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
   7                                         const char *origin, int no,
   8                                         enum grep_pat_token t,
   9                                         enum grep_header_field field)
  10 {
  11         struct grep_pat *p = xcalloc(1, sizeof(*p));
  12         p->pattern = xmemdupz(pat, patlen);
  13         p->patternlen = patlen;
  14         p->origin = origin;
  15         p->no = no;
  16         p->token = t;
  17         p->field = field;
  18         return p;
  19 }
  20
  21 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
  22 {
  23         **tail = p;
  24         *tail = &p->next;
  25         p->next = NULL;
  26
  27         switch (p->token) {
  28         case GREP_PATTERN: /* atom */
  29         case GREP_PATTERN_HEAD:
  30         case GREP_PATTERN_BODY:
  31                 for (;;) {
  32                         struct grep_pat *new_pat;
  33                         size_t len = 0;
  34                         char *cp = p->pattern + p->patternlen, *nl = NULL;
  35                         while (++len <= p->patternlen) {
  36                                 if (*(--cp) == '\n') {
  37                                         nl = cp;
  38                                         break;
  39                                 }
  40                         }
  41                         if (!nl)
  42                                 break;
  43                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
  44                                                   p->no, p->token, p->field);
  45                         new_pat->next = p->next;
  46                         if (!p->next)
  47                                 *tail = &new_pat->next;
  48                         p->next = new_pat;
  49                         *nl = '\0';
  50                         p->patternlen -= len;
  51                 }
  52                 break;
  53         default:
  54                 break;
  55         }
  56 }
  57
  58 void append_header_grep_pattern(struct grep_opt *opt,
  59                                 enum grep_header_field field, const char *pat)
  60 {
  61         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
  62                                              GREP_PATTERN_HEAD, field);
  63         do_append_grep_pat(&opt->header_tail, p);
  64 }
  65
  66 void append_grep_pattern(struct grep_opt *opt, const char *pat,
  67                          const char *origin, int no, enum grep_pat_token t)
  68 {
  69         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
  70 }
  71
  72 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
  73                      const char *origin, int no, enum grep_pat_token t)
  74 {
  75         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
  76         do_append_grep_pat(&opt->pattern_tail, p);
  77 }
  78
  79 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
  80 {
  81         struct grep_pat *pat;
  82         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
  83         *ret = *opt;
  84
  85         ret->pattern_list = NULL;
  86         ret->pattern_tail = &ret->pattern_list;
  87
  88         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
  89         {
  90                 if(pat->token == GREP_PATTERN_HEAD)
  91                         append_header_grep_pattern(ret, pat->field,
  92                                                    pat->pattern);
  93                 else
  94                         append_grep_pat(ret, pat->pattern, pat->patternlen,
  95                                         pat->origin, pat->no, pat->token);
  96         }
  97
  98         return ret;
  99 }
 100
 101 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 102                 const char *error)
 103 {
 104         char where[1024];
 105
 106         if (p->no)
 107                 sprintf(where, "In '%s' at %d, ", p->origin, p->no);
 108         else if (p->origin)
 109                 sprintf(where, "%s, ", p->origin);
 110         else
 111                 where[0] = 0;
 112
 113         die("%s'%s': %s", where, p->pattern, error);
 114 }
 115
 116 #ifdef USE_LIBPCRE
 117 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 118 {
 119         const char *error;
 120         int erroffset;
 121         int options = 0;
 122
 123         if (opt->ignore_case)
 124                 options |= PCRE_CASELESS;
 125
 126         p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 127                         NULL);
 128         if (!p->pcre_regexp)
 129                 compile_regexp_failed(p, error);
 130
 131         p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 132         if (!p->pcre_extra_info && error)
 133                 die("%s", error);
 134 }
 135
 136 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 137                 regmatch_t *match, int eflags)
 138 {
 139         int ovector[30], ret, flags = 0;
 140
 141         if (eflags & REG_NOTBOL)
 142                 flags |= PCRE_NOTBOL;
 143
 144         ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 145                         0, flags, ovector, ARRAY_SIZE(ovector));
 146         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 147                 die("pcre_exec failed with error code %d", ret);
 148         if (ret > 0) {
 149                 ret = 0;
 150                 match->rm_so = ovector[0];
 151                 match->rm_eo = ovector[1];
 152         }
 153
 154         return ret;
 155 }
 156
 157 static void free_pcre_regexp(struct grep_pat *p)
 158 {
 159         pcre_free(p->pcre_regexp);
 160         pcre_free(p->pcre_extra_info);
 161 }
 162 #else /* !USE_LIBPCRE */
 163 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 164 {
 165         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 166 }
 167
 168 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 169                 regmatch_t *match, int eflags)
 170 {
 171         return 1;
 172 }
 173
 174 static void free_pcre_regexp(struct grep_pat *p)
 175 {
 176 }
 177 #endif /* !USE_LIBPCRE */
 178
 179 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 180 {
 181         int err;
 182
 183         p->word_regexp = opt->word_regexp;
 184         p->ignore_case = opt->ignore_case;
 185         p->fixed = opt->fixed;
 186
 187         if (p->fixed)
 188                 return;
 189
 190         if (opt->pcre) {
 191                 compile_pcre_regexp(p, opt);
 192                 return;
 193         }
 194
 195         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 196         if (err) {
 197                 char errbuf[1024];
 198                 regerror(err, &p->regexp, errbuf, 1024);
 199                 regfree(&p->regexp);
 200                 compile_regexp_failed(p, errbuf);
 201         }
 202 }
 203
 204 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 205 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 206 {
 207         struct grep_pat *p;
 208         struct grep_expr *x;
 209
 210         p = *list;
 211         if (!p)
 212                 return NULL;
 213         switch (p->token) {
 214         case GREP_PATTERN: /* atom */
 215         case GREP_PATTERN_HEAD:
 216         case GREP_PATTERN_BODY:
 217                 x = xcalloc(1, sizeof (struct grep_expr));
 218                 x->node = GREP_NODE_ATOM;
 219                 x->u.atom = p;
 220                 *list = p->next;
 221                 return x;
 222         case GREP_OPEN_PAREN:
 223                 *list = p->next;
 224                 x = compile_pattern_or(list);
 225                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 226                         die("unmatched parenthesis");
 227                 *list = (*list)->next;
 228                 return x;
 229         default:
 230                 return NULL;
 231         }
 232 }
 233
 234 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 235 {
 236         struct grep_pat *p;
 237         struct grep_expr *x;
 238
 239         p = *list;
 240         if (!p)
 241                 return NULL;
 242         switch (p->token) {
 243         case GREP_NOT:
 244                 if (!p->next)
 245                         die("--not not followed by pattern expression");
 246                 *list = p->next;
 247                 x = xcalloc(1, sizeof (struct grep_expr));
 248                 x->node = GREP_NODE_NOT;
 249                 x->u.unary = compile_pattern_not(list);
 250                 if (!x->u.unary)
 251                         die("--not followed by non pattern expression");
 252                 return x;
 253         default:
 254                 return compile_pattern_atom(list);
 255         }
 256 }
 257
 258 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 259 {
 260         struct grep_pat *p;
 261         struct grep_expr *x, *y, *z;
 262
 263         x = compile_pattern_not(list);
 264         p = *list;
 265         if (p && p->token == GREP_AND) {
 266                 if (!p->next)
 267                         die("--and not followed by pattern expression");
 268                 *list = p->next;
 269                 y = compile_pattern_and(list);
 270                 if (!y)
 271                         die("--and not followed by pattern expression");
 272                 z = xcalloc(1, sizeof (struct grep_expr));
 273                 z->node = GREP_NODE_AND;
 274                 z->u.binary.left = x;
 275                 z->u.binary.right = y;
 276                 return z;
 277         }
 278         return x;
 279 }
 280
 281 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 282 {
 283         struct grep_pat *p;
 284         struct grep_expr *x, *y, *z;
 285
 286         x = compile_pattern_and(list);
 287         p = *list;
 288         if (x && p && p->token != GREP_CLOSE_PAREN) {
 289                 y = compile_pattern_or(list);
 290                 if (!y)
 291                         die("not a pattern expression %s", p->pattern);
 292                 z = xcalloc(1, sizeof (struct grep_expr));
 293                 z->node = GREP_NODE_OR;
 294                 z->u.binary.left = x;
 295                 z->u.binary.right = y;
 296                 return z;
 297         }
 298         return x;
 299 }
 300
 301 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 302 {
 303         return compile_pattern_or(list);
 304 }
 305
 306 static struct grep_expr *grep_true_expr(void)
 307 {
 308         struct grep_expr *z = xcalloc(1, sizeof(*z));
 309         z->node = GREP_NODE_TRUE;
 310         return z;
 311 }
 312
 313 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 314 {
 315         struct grep_expr *z = xcalloc(1, sizeof(*z));
 316         z->node = GREP_NODE_OR;
 317         z->u.binary.left = left;
 318         z->u.binary.right = right;
 319         return z;
 320 }
 321
 322 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 323 {
 324         struct grep_pat *p;
 325         struct grep_expr *header_expr;
 326         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 327         enum grep_header_field fld;
 328
 329         if (!opt->header_list)
 330                 return NULL;
 331         p = opt->header_list;
 332         for (p = opt->header_list; p; p = p->next) {
 333                 if (p->token != GREP_PATTERN_HEAD)
 334                         die("bug: a non-header pattern in grep header list.");
 335                 if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
 336                         die("bug: unknown header field %d", p->field);
 337                 compile_regexp(p, opt);
 338         }
 339
 340         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 341                 header_group[fld] = NULL;
 342
 343         for (p = opt->header_list; p; p = p->next) {
 344                 struct grep_expr *h;
 345                 struct grep_pat *pp = p;
 346
 347                 h = compile_pattern_atom(&pp);
 348                 if (!h || pp != p->next)
 349                         die("bug: malformed header expr");
 350                 if (!header_group[p->field]) {
 351                         header_group[p->field] = h;
 352                         continue;
 353                 }
 354                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 355         }
 356
 357         header_expr = NULL;
 358
 359         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 360                 if (!header_group[fld])
 361                         continue;
 362                 if (!header_expr)
 363                         header_expr = grep_true_expr();
 364                 header_expr = grep_or_expr(header_group[fld], header_expr);
 365         }
 366         return header_expr;
 367 }
 368
 369 void compile_grep_patterns(struct grep_opt *opt)
 370 {
 371         struct grep_pat *p;
 372         struct grep_expr *header_expr = prep_header_patterns(opt);
 373
 374         for (p = opt->pattern_list; p; p = p->next) {
 375                 switch (p->token) {
 376                 case GREP_PATTERN: /* atom */
 377                 case GREP_PATTERN_HEAD:
 378                 case GREP_PATTERN_BODY:
 379                         compile_regexp(p, opt);
 380                         break;
 381                 default:
 382                         opt->extended = 1;
 383                         break;
 384                 }
 385         }
 386
 387         if (opt->all_match || header_expr)
 388                 opt->extended = 1;
 389         else if (!opt->extended)
 390                 return;
 391
 392         p = opt->pattern_list;
 393         if (p)
 394                 opt->pattern_expression = compile_pattern_expr(&p);
 395         if (p)
 396                 die("incomplete pattern expression: %s", p->pattern);
 397
 398         if (!header_expr)
 399                 return;
 400
 401         if (!opt->pattern_expression)
 402                 opt->pattern_expression = header_expr;
 403         else
 404                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 405                                                        header_expr);
 406         opt->all_match = 1;
 407 }
 408
 409 static void free_pattern_expr(struct grep_expr *x)
 410 {
 411         switch (x->node) {
 412         case GREP_NODE_TRUE:
 413         case GREP_NODE_ATOM:
 414                 break;
 415         case GREP_NODE_NOT:
 416                 free_pattern_expr(x->u.unary);
 417                 break;
 418         case GREP_NODE_AND:
 419         case GREP_NODE_OR:
 420                 free_pattern_expr(x->u.binary.left);
 421                 free_pattern_expr(x->u.binary.right);
 422                 break;
 423         }
 424         free(x);
 425 }
 426
 427 void free_grep_patterns(struct grep_opt *opt)
 428 {
 429         struct grep_pat *p, *n;
 430
 431         for (p = opt->pattern_list; p; p = n) {
 432                 n = p->next;
 433                 switch (p->token) {
 434                 case GREP_PATTERN: /* atom */
 435                 case GREP_PATTERN_HEAD:
 436                 case GREP_PATTERN_BODY:
 437                         if (p->pcre_regexp)
 438                                 free_pcre_regexp(p);
 439                         else
 440                                 regfree(&p->regexp);
 441                         free(p->pattern);
 442                         break;
 443                 default:
 444                         break;
 445                 }
 446                 free(p);
 447         }
 448
 449         if (!opt->extended)
 450                 return;
 451         free_pattern_expr(opt->pattern_expression);
 452 }
 453
 454 static char *end_of_line(char *cp, unsigned long *left)
 455 {
 456         unsigned long l = *left;
 457         while (l && *cp != '\n') {
 458                 l--;
 459                 cp++;
 460         }
 461         *left = l;
 462         return cp;
 463 }
 464
 465 static int word_char(char ch)
 466 {
 467         return isalnum(ch) || ch == '_';
 468 }
 469
 470 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 471                          const char *color)
 472 {
 473         if (opt->color && color && color[0]) {
 474                 opt->output(opt, color, strlen(color));
 475                 opt->output(opt, data, size);
 476                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 477         } else
 478                 opt->output(opt, data, size);
 479 }
 480
 481 static void output_sep(struct grep_opt *opt, char sign)
 482 {
 483         if (opt->null_following_name)
 484                 opt->output(opt, "\0", 1);
 485         else
 486                 output_color(opt, &sign, 1, opt->color_sep);
 487 }
 488
 489 static void show_name(struct grep_opt *opt, const char *name)
 490 {
 491         output_color(opt, name, strlen(name), opt->color_filename);
 492         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 493 }
 494
 495 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 496                     regmatch_t *match)
 497 {
 498         char *hit;
 499
 500         if (p->ignore_case) {
 501                 char *s = line;
 502                 do {
 503                         hit = strcasestr(s, p->pattern);
 504                         if (hit)
 505                                 break;
 506                         s += strlen(s) + 1;
 507                 } while (s < eol);
 508         } else
 509                 hit = memmem(line, eol - line, p->pattern, p->patternlen);
 510
 511         if (!hit) {
 512                 match->rm_so = match->rm_eo = -1;
 513                 return REG_NOMATCH;
 514         }
 515         else {
 516                 match->rm_so = hit - line;
 517                 match->rm_eo = match->rm_so + p->patternlen;
 518                 return 0;
 519         }
 520 }
 521
 522 static int regmatch(const regex_t *preg, char *line, char *eol,
 523                     regmatch_t *match, int eflags)
 524 {
 525 #ifdef REG_STARTEND
 526         match->rm_so = 0;
 527         match->rm_eo = eol - line;
 528         eflags |= REG_STARTEND;
 529 #endif
 530         return regexec(preg, line, 1, match, eflags);
 531 }
 532
 533 static int patmatch(struct grep_pat *p, char *line, char *eol,
 534                     regmatch_t *match, int eflags)
 535 {
 536         int hit;
 537
 538         if (p->fixed)
 539                 hit = !fixmatch(p, line, eol, match);
 540         else if (p->pcre_regexp)
 541                 hit = !pcrematch(p, line, eol, match, eflags);
 542         else
 543                 hit = !regmatch(&p->regexp, line, eol, match, eflags);
 544
 545         return hit;
 546 }
 547
 548 static int strip_timestamp(char *bol, char **eol_p)
 549 {
 550         char *eol = *eol_p;
 551         int ch;
 552
 553         while (bol < --eol) {
 554                 if (*eol != '>')
 555                         continue;
 556                 *eol_p = ++eol;
 557                 ch = *eol;
 558                 *eol = '\0';
 559                 return ch;
 560         }
 561         return 0;
 562 }
 563
 564 static struct {
 565         const char *field;
 566         size_t len;
 567 } header_field[] = {
 568         { "author ", 7 },
 569         { "committer ", 10 },
 570 };
 571
 572 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 573                              enum grep_context ctx,
 574                              regmatch_t *pmatch, int eflags)
 575 {
 576         int hit = 0;
 577         int saved_ch = 0;
 578         const char *start = bol;
 579
 580         if ((p->token != GREP_PATTERN) &&
 581             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 582                 return 0;
 583
 584         if (p->token == GREP_PATTERN_HEAD) {
 585                 const char *field;
 586                 size_t len;
 587                 assert(p->field < ARRAY_SIZE(header_field));
 588                 field = header_field[p->field].field;
 589                 len = header_field[p->field].len;
 590                 if (strncmp(bol, field, len))
 591                         return 0;
 592                 bol += len;
 593                 saved_ch = strip_timestamp(bol, &eol);
 594         }
 595
 596  again:
 597         hit = patmatch(p, bol, eol, pmatch, eflags);
 598
 599         if (hit && p->word_regexp) {
 600                 if ((pmatch[0].rm_so < 0) ||
 601                     (eol - bol) < pmatch[0].rm_so ||
 602                     (pmatch[0].rm_eo < 0) ||
 603                     (eol - bol) < pmatch[0].rm_eo)
 604                         die("regexp returned nonsense");
 605
 606                 /* Match beginning must be either beginning of the
 607                  * line, or at word boundary (i.e. the last char must
 608                  * not be a word char).  Similarly, match end must be
 609                  * either end of the line, or at word boundary
 610                  * (i.e. the next char must not be a word char).
 611                  */
 612                 if ( ((pmatch[0].rm_so == 0) ||
 613                       !word_char(bol[pmatch[0].rm_so-1])) &&
 614                      ((pmatch[0].rm_eo == (eol-bol)) ||
 615                       !word_char(bol[pmatch[0].rm_eo])) )
 616                         ;
 617                 else
 618                         hit = 0;
 619
 620                 /* Words consist of at least one character. */
 621                 if (pmatch->rm_so == pmatch->rm_eo)
 622                         hit = 0;
 623
 624                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 625                         /* There could be more than one match on the
 626                          * line, and the first match might not be
 627                          * strict word match.  But later ones could be!
 628                          * Forward to the next possible start, i.e. the
 629                          * next position following a non-word char.
 630                          */
 631                         bol = pmatch[0].rm_so + bol + 1;
 632                         while (word_char(bol[-1]) && bol < eol)
 633                                 bol++;
 634                         eflags |= REG_NOTBOL;
 635                         if (bol < eol)
 636                                 goto again;
 637                 }
 638         }
 639         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 640                 *eol = saved_ch;
 641         if (hit) {
 642                 pmatch[0].rm_so += bol - start;
 643                 pmatch[0].rm_eo += bol - start;
 644         }
 645         return hit;
 646 }
 647
 648 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 649                            enum grep_context ctx, int collect_hits)
 650 {
 651         int h = 0;
 652         regmatch_t match;
 653
 654         if (!x)
 655                 die("Not a valid grep expression");
 656         switch (x->node) {
 657         case GREP_NODE_TRUE:
 658                 h = 1;
 659                 break;
 660         case GREP_NODE_ATOM:
 661                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 662                 break;
 663         case GREP_NODE_NOT:
 664                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 665                 break;
 666         case GREP_NODE_AND:
 667                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 668                         return 0;
 669                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 670                 break;
 671         case GREP_NODE_OR:
 672                 if (!collect_hits)
 673                         return (match_expr_eval(x->u.binary.left,
 674                                                 bol, eol, ctx, 0) ||
 675                                 match_expr_eval(x->u.binary.right,
 676                                                 bol, eol, ctx, 0));
 677                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 678                 x->u.binary.left->hit |= h;
 679                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 680                 break;
 681         default:
 682                 die("Unexpected node type (internal error) %d", x->node);
 683         }
 684         if (collect_hits)
 685                 x->hit |= h;
 686         return h;
 687 }
 688
 689 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 690                       enum grep_context ctx, int collect_hits)
 691 {
 692         struct grep_expr *x = opt->pattern_expression;
 693         return match_expr_eval(x, bol, eol, ctx, collect_hits);
 694 }
 695
 696 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 697                       enum grep_context ctx, int collect_hits)
 698 {
 699         struct grep_pat *p;
 700         regmatch_t match;
 701
 702         if (opt->extended)
 703                 return match_expr(opt, bol, eol, ctx, collect_hits);
 704
 705         /* we do not call with collect_hits without being extended */
 706         for (p = opt->pattern_list; p; p = p->next) {
 707                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 708                         return 1;
 709         }
 710         return 0;
 711 }
 712
 713 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 714                               enum grep_context ctx,
 715                               regmatch_t *pmatch, int eflags)
 716 {
 717         regmatch_t match;
 718
 719         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 720                 return 0;
 721         if (match.rm_so < 0 || match.rm_eo < 0)
 722                 return 0;
 723         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 724                 if (match.rm_so > pmatch->rm_so)
 725                         return 1;
 726                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 727                         return 1;
 728         }
 729         pmatch->rm_so = match.rm_so;
 730         pmatch->rm_eo = match.rm_eo;
 731         return 1;
 732 }
 733
 734 static int next_match(struct grep_opt *opt, char *bol, char *eol,
 735                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
 736 {
 737         struct grep_pat *p;
 738         int hit = 0;
 739
 740         pmatch->rm_so = pmatch->rm_eo = -1;
 741         if (bol < eol) {
 742                 for (p = opt->pattern_list; p; p = p->next) {
 743                         switch (p->token) {
 744                         case GREP_PATTERN: /* atom */
 745                         case GREP_PATTERN_HEAD:
 746                         case GREP_PATTERN_BODY:
 747                                 hit |= match_next_pattern(p, bol, eol, ctx,
 748                                                           pmatch, eflags);
 749                                 break;
 750                         default:
 751                                 break;
 752                         }
 753                 }
 754         }
 755         return hit;
 756 }
 757
 758 static void show_line(struct grep_opt *opt, char *bol, char *eol,
 759                       const char *name, unsigned lno, char sign)
 760 {
 761         int rest = eol - bol;
 762         char *line_color = NULL;
 763
 764         if (opt->pre_context || opt->post_context) {
 765                 if (opt->last_shown == 0) {
 766                         if (opt->show_hunk_mark) {
 767                                 output_color(opt, "--", 2, opt->color_sep);
 768                                 opt->output(opt, "\n", 1);
 769                         }
 770                 } else if (lno > opt->last_shown + 1) {
 771                         output_color(opt, "--", 2, opt->color_sep);
 772                         opt->output(opt, "\n", 1);
 773                 }
 774         }
 775         opt->last_shown = lno;
 776
 777         if (opt->pathname) {
 778                 output_color(opt, name, strlen(name), opt->color_filename);
 779                 output_sep(opt, sign);
 780         }
 781         if (opt->linenum) {
 782                 char buf[32];
 783                 snprintf(buf, sizeof(buf), "%d", lno);
 784                 output_color(opt, buf, strlen(buf), opt->color_lineno);
 785                 output_sep(opt, sign);
 786         }
 787         if (opt->color) {
 788                 regmatch_t match;
 789                 enum grep_context ctx = GREP_CONTEXT_BODY;
 790                 int ch = *eol;
 791                 int eflags = 0;
 792
 793                 if (sign == ':')
 794                         line_color = opt->color_selected;
 795                 else if (sign == '-')
 796                         line_color = opt->color_context;
 797                 else if (sign == '=')
 798                         line_color = opt->color_function;
 799                 *eol = '\0';
 800                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
 801                         if (match.rm_so == match.rm_eo)
 802                                 break;
 803
 804                         output_color(opt, bol, match.rm_so, line_color);
 805                         output_color(opt, bol + match.rm_so,
 806                                      match.rm_eo - match.rm_so,
 807                                      opt->color_match);
 808                         bol += match.rm_eo;
 809                         rest -= match.rm_eo;
 810                         eflags = REG_NOTBOL;
 811                 }
 812                 *eol = ch;
 813         }
 814         output_color(opt, bol, rest, line_color);
 815         opt->output(opt, "\n", 1);
 816 }
 817
 818 static int match_funcname(struct grep_opt *opt, char *bol, char *eol)
 819 {
 820         xdemitconf_t *xecfg = opt->priv;
 821         if (xecfg && xecfg->find_func) {
 822                 char buf[1];
 823                 return xecfg->find_func(bol, eol - bol, buf, 1,
 824                                         xecfg->find_func_priv) >= 0;
 825         }
 826
 827         if (bol == eol)
 828                 return 0;
 829         if (isalpha(*bol) || *bol == '_' || *bol == '$')
 830                 return 1;
 831         return 0;
 832 }
 833
 834 static void show_funcname_line(struct grep_opt *opt, const char *name,
 835                                char *buf, char *bol, unsigned lno)
 836 {
 837         while (bol > buf) {
 838                 char *eol = --bol;
 839
 840                 while (bol > buf && bol[-1] != '\n')
 841                         bol--;
 842                 lno--;
 843
 844                 if (lno <= opt->last_shown)
 845                         break;
 846
 847                 if (match_funcname(opt, bol, eol)) {
 848                         show_line(opt, bol, eol, name, lno, '=');
 849                         break;
 850                 }
 851         }
 852 }
 853
 854 static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
 855                              char *bol, unsigned lno)
 856 {
 857         unsigned cur = lno, from = 1, funcname_lno = 0;
 858         int funcname_needed = opt->funcname;
 859
 860         if (opt->pre_context < lno)
 861                 from = lno - opt->pre_context;
 862         if (from <= opt->last_shown)
 863                 from = opt->last_shown + 1;
 864
 865         /* Rewind. */
 866         while (bol > buf && cur > from) {
 867                 char *eol = --bol;
 868
 869                 while (bol > buf && bol[-1] != '\n')
 870                         bol--;
 871                 cur--;
 872                 if (funcname_needed && match_funcname(opt, bol, eol)) {
 873                         funcname_lno = cur;
 874                         funcname_needed = 0;
 875                 }
 876         }
 877
 878         /* We need to look even further back to find a function signature. */
 879         if (opt->funcname && funcname_needed)
 880                 show_funcname_line(opt, name, buf, bol, cur);
 881
 882         /* Back forward. */
 883         while (cur < lno) {
 884                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
 885
 886                 while (*eol != '\n')
 887                         eol++;
 888                 show_line(opt, bol, eol, name, cur, sign);
 889                 bol = eol + 1;
 890                 cur++;
 891         }
 892 }
 893
 894 static int should_lookahead(struct grep_opt *opt)
 895 {
 896         struct grep_pat *p;
 897
 898         if (opt->extended)
 899                 return 0; /* punt for too complex stuff */
 900         if (opt->invert)
 901                 return 0;
 902         for (p = opt->pattern_list; p; p = p->next) {
 903                 if (p->token != GREP_PATTERN)
 904                         return 0; /* punt for "header only" and stuff */
 905         }
 906         return 1;
 907 }
 908
 909 static int look_ahead(struct grep_opt *opt,
 910                       unsigned long *left_p,
 911                       unsigned *lno_p,
 912                       char **bol_p)
 913 {
 914         unsigned lno = *lno_p;
 915         char *bol = *bol_p;
 916         struct grep_pat *p;
 917         char *sp, *last_bol;
 918         regoff_t earliest = -1;
 919
 920         for (p = opt->pattern_list; p; p = p->next) {
 921                 int hit;
 922                 regmatch_t m;
 923
 924                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
 925                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
 926                         continue;
 927                 if (earliest < 0 || m.rm_so < earliest)
 928                         earliest = m.rm_so;
 929         }
 930
 931         if (earliest < 0) {
 932                 *bol_p = bol + *left_p;
 933                 *left_p = 0;
 934                 return 1;
 935         }
 936         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
 937                 ; /* find the beginning of the line */
 938         last_bol = sp;
 939
 940         for (sp = bol; sp < last_bol; sp++) {
 941                 if (*sp == '\n')
 942                         lno++;
 943         }
 944         *left_p -= last_bol - bol;
 945         *bol_p = last_bol;
 946         *lno_p = lno;
 947         return 0;
 948 }
 949
 950 int grep_threads_ok(const struct grep_opt *opt)
 951 {
 952         /* If this condition is true, then we may use the attribute
 953          * machinery in grep_buffer_1. The attribute code is not
 954          * thread safe, so we disable the use of threads.
 955          */
 956         if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
 957             !opt->name_only)
 958                 return 0;
 959
 960         return 1;
 961 }
 962
 963 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
 964 {
 965         fwrite(buf, size, 1, stdout);
 966 }
 967
 968 static int grep_buffer_1(struct grep_opt *opt, const char *name,
 969                          char *buf, unsigned long size, int collect_hits)
 970 {
 971         char *bol = buf;
 972         unsigned long left = size;
 973         unsigned lno = 1;
 974         unsigned last_hit = 0;
 975         int binary_match_only = 0;
 976         unsigned count = 0;
 977         int try_lookahead = 0;
 978         enum grep_context ctx = GREP_CONTEXT_HEAD;
 979         xdemitconf_t xecfg;
 980
 981         if (!opt->output)
 982                 opt->output = std_output;
 983
 984         if (opt->last_shown && (opt->pre_context || opt->post_context) &&
 985             opt->output == std_output)
 986                 opt->show_hunk_mark = 1;
 987         opt->last_shown = 0;
 988
 989         switch (opt->binary) {
 990         case GREP_BINARY_DEFAULT:
 991                 if (buffer_is_binary(buf, size))
 992                         binary_match_only = 1;
 993                 break;
 994         case GREP_BINARY_NOMATCH:
 995                 if (buffer_is_binary(buf, size))
 996                         return 0; /* Assume unmatch */
 997                 break;
 998         case GREP_BINARY_TEXT:
 999                 break;
1000         default:
1001                 die("bug: unknown binary handling mode");
1002         }
1003
1004         memset(&xecfg, 0, sizeof(xecfg));
1005         if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
1006             !opt->name_only && !binary_match_only && !collect_hits) {
1007                 struct userdiff_driver *drv = userdiff_find_by_path(name);
1008                 if (drv && drv->funcname.pattern) {
1009                         const struct userdiff_funcname *pe = &drv->funcname;
1010                         xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1011                         opt->priv = &xecfg;
1012                 }
1013         }
1014         try_lookahead = should_lookahead(opt);
1015
1016         while (left) {
1017                 char *eol, ch;
1018                 int hit;
1019
1020                 /*
1021                  * look_ahead() skips quickly to the line that possibly
1022                  * has the next hit; don't call it if we need to do
1023                  * something more than just skipping the current line
1024                  * in response to an unmatch for the current line.  E.g.
1025                  * inside a post-context window, we will show the current
1026                  * line as a context around the previous hit when it
1027                  * doesn't hit.
1028                  */
1029                 if (try_lookahead
1030                     && !(last_hit
1031                          && lno <= last_hit + opt->post_context)
1032                     && look_ahead(opt, &left, &lno, &bol))
1033                         break;
1034                 eol = end_of_line(bol, &left);
1035                 ch = *eol;
1036                 *eol = 0;
1037
1038                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1039                         ctx = GREP_CONTEXT_BODY;
1040
1041                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1042                 *eol = ch;
1043
1044                 if (collect_hits)
1045                         goto next_line;
1046
1047                 /* "grep -v -e foo -e bla" should list lines
1048                  * that do not have either, so inversion should
1049                  * be done outside.
1050                  */
1051                 if (opt->invert)
1052                         hit = !hit;
1053                 if (opt->unmatch_name_only) {
1054                         if (hit)
1055                                 return 0;
1056                         goto next_line;
1057                 }
1058                 if (hit) {
1059                         count++;
1060                         if (opt->status_only)
1061                                 return 1;
1062                         if (opt->name_only) {
1063                                 show_name(opt, name);
1064                                 return 1;
1065                         }
1066                         if (opt->count)
1067                                 goto next_line;
1068                         if (binary_match_only) {
1069                                 opt->output(opt, "Binary file ", 12);
1070                                 output_color(opt, name, strlen(name),
1071                                              opt->color_filename);
1072                                 opt->output(opt, " matches\n", 9);
1073                                 return 1;
1074                         }
1075                         /* Hit at this line.  If we haven't shown the
1076                          * pre-context lines, we would need to show them.
1077                          */
1078                         if (opt->pre_context)
1079                                 show_pre_context(opt, name, buf, bol, lno);
1080                         else if (opt->funcname)
1081                                 show_funcname_line(opt, name, buf, bol, lno);
1082                         show_line(opt, bol, eol, name, lno, ':');
1083                         last_hit = lno;
1084                 }
1085                 else if (last_hit &&
1086                          lno <= last_hit + opt->post_context) {
1087                         /* If the last hit is within the post context,
1088                          * we need to show this line.
1089                          */
1090                         show_line(opt, bol, eol, name, lno, '-');
1091                 }
1092
1093         next_line:
1094                 bol = eol + 1;
1095                 if (!left)
1096                         break;
1097                 left--;
1098                 lno++;
1099         }
1100
1101         if (collect_hits)
1102                 return 0;
1103
1104         if (opt->status_only)
1105                 return 0;
1106         if (opt->unmatch_name_only) {
1107                 /* We did not see any hit, so we want to show this */
1108                 show_name(opt, name);
1109                 return 1;
1110         }
1111
1112         xdiff_clear_find_func(&xecfg);
1113         opt->priv = NULL;
1114
1115         /* NEEDSWORK:
1116          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1117          * which feels mostly useless but sometimes useful.  Maybe
1118          * make it another option?  For now suppress them.
1119          */
1120         if (opt->count && count) {
1121                 char buf[32];
1122                 output_color(opt, name, strlen(name), opt->color_filename);
1123                 output_sep(opt, ':');
1124                 snprintf(buf, sizeof(buf), "%u\n", count);
1125                 opt->output(opt, buf, strlen(buf));
1126                 return 1;
1127         }
1128         return !!last_hit;
1129 }
1130
1131 static void clr_hit_marker(struct grep_expr *x)
1132 {
1133         /* All-hit markers are meaningful only at the very top level
1134          * OR node.
1135          */
1136         while (1) {
1137                 x->hit = 0;
1138                 if (x->node != GREP_NODE_OR)
1139                         return;
1140                 x->u.binary.left->hit = 0;
1141                 x = x->u.binary.right;
1142         }
1143 }
1144
1145 static int chk_hit_marker(struct grep_expr *x)
1146 {
1147         /* Top level nodes have hit markers.  See if they all are hits */
1148         while (1) {
1149                 if (x->node != GREP_NODE_OR)
1150                         return x->hit;
1151                 if (!x->u.binary.left->hit)
1152                         return 0;
1153                 x = x->u.binary.right;
1154         }
1155 }
1156
1157 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
1158 {
1159         /*
1160          * we do not have to do the two-pass grep when we do not check
1161          * buffer-wide "all-match".
1162          */
1163         if (!opt->all_match)
1164                 return grep_buffer_1(opt, name, buf, size, 0);
1165
1166         /* Otherwise the toplevel "or" terms hit a bit differently.
1167          * We first clear hit markers from them.
1168          */
1169         clr_hit_marker(opt->pattern_expression);
1170         grep_buffer_1(opt, name, buf, size, 1);
1171
1172         if (!chk_hit_marker(opt->pattern_expression))
1173                 return 0;
1174
1175         return grep_buffer_1(opt, name, buf, size, 0);
1176 }