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