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