grep.c

   1 #include "cache.h"
   2 #include "grep.h"
   3 #include "userdiff.h"
   4 #include "xdiff-interface.h"
   5
   6 static int grep_source_load(struct grep_source *gs);
   7 static int grep_source_is_binary(struct grep_source *gs);
   8
   9 static struct grep_opt grep_defaults;
  10
  11 /*
  12  * Initialize the grep_defaults template with hardcoded defaults.
  13  * We could let the compiler do this, but without C99 initializers
  14  * the code gets unwieldy and unreadable, so...
  15  */
  16 void init_grep_defaults(void)
  17 {
  18         struct grep_opt *opt = &grep_defaults;
  19
  20         memset(opt, 0, sizeof(*opt));
  21         opt->relative = 1;
  22         opt->pathname = 1;
  23         opt->regflags = REG_NEWLINE;
  24         opt->max_depth = -1;
  25         opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  26         opt->extended_regexp_option = 0;
  27         strcpy(opt->color_context, "");
  28         strcpy(opt->color_filename, "");
  29         strcpy(opt->color_function, "");
  30         strcpy(opt->color_lineno, "");
  31         strcpy(opt->color_match, GIT_COLOR_BOLD_RED);
  32         strcpy(opt->color_selected, "");
  33         strcpy(opt->color_sep, GIT_COLOR_CYAN);
  34         opt->color = -1;
  35 }
  36
  37 static int parse_pattern_type_arg(const char *opt, const char *arg)
  38 {
  39         if (!strcmp(arg, "default"))
  40                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  41         else if (!strcmp(arg, "basic"))
  42                 return GREP_PATTERN_TYPE_BRE;
  43         else if (!strcmp(arg, "extended"))
  44                 return GREP_PATTERN_TYPE_ERE;
  45         else if (!strcmp(arg, "fixed"))
  46                 return GREP_PATTERN_TYPE_FIXED;
  47         else if (!strcmp(arg, "perl"))
  48                 return GREP_PATTERN_TYPE_PCRE;
  49         die("bad %s argument: %s", opt, arg);
  50 }
  51
  52 /*
  53  * Read the configuration file once and store it in
  54  * the grep_defaults template.
  55  */
  56 int grep_config(const char *var, const char *value, void *cb)
  57 {
  58         struct grep_opt *opt = &grep_defaults;
  59         char *color = NULL;
  60
  61         if (userdiff_config(var, value) < 0)
  62                 return -1;
  63
  64         if (!strcmp(var, "grep.extendedregexp")) {
  65                 if (git_config_bool(var, value))
  66                         opt->extended_regexp_option = 1;
  67                 else
  68                         opt->extended_regexp_option = 0;
  69                 return 0;
  70         }
  71
  72         if (!strcmp(var, "grep.patterntype")) {
  73                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
  74                 return 0;
  75         }
  76
  77         if (!strcmp(var, "grep.linenumber")) {
  78                 opt->linenum = git_config_bool(var, value);
  79                 return 0;
  80         }
  81
  82         if (!strcmp(var, "color.grep"))
  83                 opt->color = git_config_colorbool(var, value);
  84         else if (!strcmp(var, "color.grep.context"))
  85                 color = opt->color_context;
  86         else if (!strcmp(var, "color.grep.filename"))
  87                 color = opt->color_filename;
  88         else if (!strcmp(var, "color.grep.function"))
  89                 color = opt->color_function;
  90         else if (!strcmp(var, "color.grep.linenumber"))
  91                 color = opt->color_lineno;
  92         else if (!strcmp(var, "color.grep.match"))
  93                 color = opt->color_match;
  94         else if (!strcmp(var, "color.grep.selected"))
  95                 color = opt->color_selected;
  96         else if (!strcmp(var, "color.grep.separator"))
  97                 color = opt->color_sep;
  98
  99         if (color) {
 100                 if (!value)
 101                         return config_error_nonbool(var);
 102                 color_parse(value, var, color);
 103         }
 104         return 0;
 105 }
 106
 107 /*
 108  * Initialize one instance of grep_opt and copy the
 109  * default values from the template we read the configuration
 110  * information in an earlier call to git_config(grep_config).
 111  */
 112 void grep_init(struct grep_opt *opt, const char *prefix)
 113 {
 114         struct grep_opt *def = &grep_defaults;
 115
 116         memset(opt, 0, sizeof(*opt));
 117         opt->prefix = prefix;
 118         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 119         opt->pattern_tail = &opt->pattern_list;
 120         opt->header_tail = &opt->header_list;
 121
 122         opt->color = def->color;
 123         opt->extended_regexp_option = def->extended_regexp_option;
 124         opt->pattern_type_option = def->pattern_type_option;
 125         opt->linenum = def->linenum;
 126         opt->max_depth = def->max_depth;
 127         opt->pathname = def->pathname;
 128         opt->regflags = def->regflags;
 129         opt->relative = def->relative;
 130
 131         strcpy(opt->color_context, def->color_context);
 132         strcpy(opt->color_filename, def->color_filename);
 133         strcpy(opt->color_function, def->color_function);
 134         strcpy(opt->color_lineno, def->color_lineno);
 135         strcpy(opt->color_match, def->color_match);
 136         strcpy(opt->color_selected, def->color_selected);
 137         strcpy(opt->color_sep, def->color_sep);
 138 }
 139
 140 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 141                                         const char *origin, int no,
 142                                         enum grep_pat_token t,
 143                                         enum grep_header_field field)
 144 {
 145         struct grep_pat *p = xcalloc(1, sizeof(*p));
 146         p->pattern = xmemdupz(pat, patlen);
 147         p->patternlen = patlen;
 148         p->origin = origin;
 149         p->no = no;
 150         p->token = t;
 151         p->field = field;
 152         return p;
 153 }
 154
 155 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 156 {
 157         **tail = p;
 158         *tail = &p->next;
 159         p->next = NULL;
 160
 161         switch (p->token) {
 162         case GREP_PATTERN: /* atom */
 163         case GREP_PATTERN_HEAD:
 164         case GREP_PATTERN_BODY:
 165                 for (;;) {
 166                         struct grep_pat *new_pat;
 167                         size_t len = 0;
 168                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 169                         while (++len <= p->patternlen) {
 170                                 if (*(--cp) == '\n') {
 171                                         nl = cp;
 172                                         break;
 173                                 }
 174                         }
 175                         if (!nl)
 176                                 break;
 177                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 178                                                   p->no, p->token, p->field);
 179                         new_pat->next = p->next;
 180                         if (!p->next)
 181                                 *tail = &new_pat->next;
 182                         p->next = new_pat;
 183                         *nl = '\0';
 184                         p->patternlen -= len;
 185                 }
 186                 break;
 187         default:
 188                 break;
 189         }
 190 }
 191
 192 void append_header_grep_pattern(struct grep_opt *opt,
 193                                 enum grep_header_field field, const char *pat)
 194 {
 195         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 196                                              GREP_PATTERN_HEAD, field);
 197         if (field == GREP_HEADER_REFLOG)
 198                 opt->use_reflog_filter = 1;
 199         do_append_grep_pat(&opt->header_tail, p);
 200 }
 201
 202 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 203                          const char *origin, int no, enum grep_pat_token t)
 204 {
 205         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 206 }
 207
 208 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 209                      const char *origin, int no, enum grep_pat_token t)
 210 {
 211         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 212         do_append_grep_pat(&opt->pattern_tail, p);
 213 }
 214
 215 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 216 {
 217         struct grep_pat *pat;
 218         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 219         *ret = *opt;
 220
 221         ret->pattern_list = NULL;
 222         ret->pattern_tail = &ret->pattern_list;
 223
 224         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 225         {
 226                 if(pat->token == GREP_PATTERN_HEAD)
 227                         append_header_grep_pattern(ret, pat->field,
 228                                                    pat->pattern);
 229                 else
 230                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 231                                         pat->origin, pat->no, pat->token);
 232         }
 233
 234         return ret;
 235 }
 236
 237 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 238                 const char *error)
 239 {
 240         char where[1024];
 241
 242         if (p->no)
 243                 sprintf(where, "In '%s' at %d, ", p->origin, p->no);
 244         else if (p->origin)
 245                 sprintf(where, "%s, ", p->origin);
 246         else
 247                 where[0] = 0;
 248
 249         die("%s'%s': %s", where, p->pattern, error);
 250 }
 251
 252 #ifdef USE_LIBPCRE
 253 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 254 {
 255         const char *error;
 256         int erroffset;
 257         int options = PCRE_MULTILINE;
 258
 259         if (opt->ignore_case)
 260                 options |= PCRE_CASELESS;
 261
 262         p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 263                         NULL);
 264         if (!p->pcre_regexp)
 265                 compile_regexp_failed(p, error);
 266
 267         p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 268         if (!p->pcre_extra_info && error)
 269                 die("%s", error);
 270 }
 271
 272 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 273                 regmatch_t *match, int eflags)
 274 {
 275         int ovector[30], ret, flags = 0;
 276
 277         if (eflags & REG_NOTBOL)
 278                 flags |= PCRE_NOTBOL;
 279
 280         ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 281                         0, flags, ovector, ARRAY_SIZE(ovector));
 282         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 283                 die("pcre_exec failed with error code %d", ret);
 284         if (ret > 0) {
 285                 ret = 0;
 286                 match->rm_so = ovector[0];
 287                 match->rm_eo = ovector[1];
 288         }
 289
 290         return ret;
 291 }
 292
 293 static void free_pcre_regexp(struct grep_pat *p)
 294 {
 295         pcre_free(p->pcre_regexp);
 296         pcre_free(p->pcre_extra_info);
 297 }
 298 #else /* !USE_LIBPCRE */
 299 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 300 {
 301         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 302 }
 303
 304 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 305                 regmatch_t *match, int eflags)
 306 {
 307         return 1;
 308 }
 309
 310 static void free_pcre_regexp(struct grep_pat *p)
 311 {
 312 }
 313 #endif /* !USE_LIBPCRE */
 314
 315 static int is_fixed(const char *s, size_t len)
 316 {
 317         size_t i;
 318
 319         /* regcomp cannot accept patterns with NULs so we
 320          * consider any pattern containing a NUL fixed.
 321          */
 322         if (memchr(s, 0, len))
 323                 return 1;
 324
 325         for (i = 0; i < len; i++) {
 326                 if (is_regex_special(s[i]))
 327                         return 0;
 328         }
 329
 330         return 1;
 331 }
 332
 333 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 334 {
 335         int err;
 336
 337         p->word_regexp = opt->word_regexp;
 338         p->ignore_case = opt->ignore_case;
 339
 340         if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 341                 p->fixed = 1;
 342         else
 343                 p->fixed = 0;
 344
 345         if (p->fixed) {
 346                 if (opt->regflags & REG_ICASE || p->ignore_case)
 347                         p->kws = kwsalloc(tolower_trans_tbl);
 348                 else
 349                         p->kws = kwsalloc(NULL);
 350                 kwsincr(p->kws, p->pattern, p->patternlen);
 351                 kwsprep(p->kws);
 352                 return;
 353         }
 354
 355         if (opt->pcre) {
 356                 compile_pcre_regexp(p, opt);
 357                 return;
 358         }
 359
 360         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 361         if (err) {
 362                 char errbuf[1024];
 363                 regerror(err, &p->regexp, errbuf, 1024);
 364                 regfree(&p->regexp);
 365                 compile_regexp_failed(p, errbuf);
 366         }
 367 }
 368
 369 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 370 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 371 {
 372         struct grep_pat *p;
 373         struct grep_expr *x;
 374
 375         p = *list;
 376         if (!p)
 377                 return NULL;
 378         switch (p->token) {
 379         case GREP_PATTERN: /* atom */
 380         case GREP_PATTERN_HEAD:
 381         case GREP_PATTERN_BODY:
 382                 x = xcalloc(1, sizeof (struct grep_expr));
 383                 x->node = GREP_NODE_ATOM;
 384                 x->u.atom = p;
 385                 *list = p->next;
 386                 return x;
 387         case GREP_OPEN_PAREN:
 388                 *list = p->next;
 389                 x = compile_pattern_or(list);
 390                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 391                         die("unmatched parenthesis");
 392                 *list = (*list)->next;
 393                 return x;
 394         default:
 395                 return NULL;
 396         }
 397 }
 398
 399 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 400 {
 401         struct grep_pat *p;
 402         struct grep_expr *x;
 403
 404         p = *list;
 405         if (!p)
 406                 return NULL;
 407         switch (p->token) {
 408         case GREP_NOT:
 409                 if (!p->next)
 410                         die("--not not followed by pattern expression");
 411                 *list = p->next;
 412                 x = xcalloc(1, sizeof (struct grep_expr));
 413                 x->node = GREP_NODE_NOT;
 414                 x->u.unary = compile_pattern_not(list);
 415                 if (!x->u.unary)
 416                         die("--not followed by non pattern expression");
 417                 return x;
 418         default:
 419                 return compile_pattern_atom(list);
 420         }
 421 }
 422
 423 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 424 {
 425         struct grep_pat *p;
 426         struct grep_expr *x, *y, *z;
 427
 428         x = compile_pattern_not(list);
 429         p = *list;
 430         if (p && p->token == GREP_AND) {
 431                 if (!p->next)
 432                         die("--and not followed by pattern expression");
 433                 *list = p->next;
 434                 y = compile_pattern_and(list);
 435                 if (!y)
 436                         die("--and not followed by pattern expression");
 437                 z = xcalloc(1, sizeof (struct grep_expr));
 438                 z->node = GREP_NODE_AND;
 439                 z->u.binary.left = x;
 440                 z->u.binary.right = y;
 441                 return z;
 442         }
 443         return x;
 444 }
 445
 446 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 447 {
 448         struct grep_pat *p;
 449         struct grep_expr *x, *y, *z;
 450
 451         x = compile_pattern_and(list);
 452         p = *list;
 453         if (x && p && p->token != GREP_CLOSE_PAREN) {
 454                 y = compile_pattern_or(list);
 455                 if (!y)
 456                         die("not a pattern expression %s", p->pattern);
 457                 z = xcalloc(1, sizeof (struct grep_expr));
 458                 z->node = GREP_NODE_OR;
 459                 z->u.binary.left = x;
 460                 z->u.binary.right = y;
 461                 return z;
 462         }
 463         return x;
 464 }
 465
 466 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 467 {
 468         return compile_pattern_or(list);
 469 }
 470
 471 static void indent(int in)
 472 {
 473         while (in-- > 0)
 474                 fputc(' ', stderr);
 475 }
 476
 477 static void dump_grep_pat(struct grep_pat *p)
 478 {
 479         switch (p->token) {
 480         case GREP_AND: fprintf(stderr, "*and*"); break;
 481         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 482         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 483         case GREP_NOT: fprintf(stderr, "*not*"); break;
 484         case GREP_OR: fprintf(stderr, "*or*"); break;
 485
 486         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 487         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 488         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 489         }
 490
 491         switch (p->token) {
 492         default: break;
 493         case GREP_PATTERN_HEAD:
 494                 fprintf(stderr, "<head %d>", p->field); break;
 495         case GREP_PATTERN_BODY:
 496                 fprintf(stderr, "<body>"); break;
 497         }
 498         switch (p->token) {
 499         default: break;
 500         case GREP_PATTERN_HEAD:
 501         case GREP_PATTERN_BODY:
 502         case GREP_PATTERN:
 503                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 504                 break;
 505         }
 506         fputc('\n', stderr);
 507 }
 508
 509 static void dump_grep_expression_1(struct grep_expr *x, int in)
 510 {
 511         indent(in);
 512         switch (x->node) {
 513         case GREP_NODE_TRUE:
 514                 fprintf(stderr, "true\n");
 515                 break;
 516         case GREP_NODE_ATOM:
 517                 dump_grep_pat(x->u.atom);
 518                 break;
 519         case GREP_NODE_NOT:
 520                 fprintf(stderr, "(not\n");
 521                 dump_grep_expression_1(x->u.unary, in+1);
 522                 indent(in);
 523                 fprintf(stderr, ")\n");
 524                 break;
 525         case GREP_NODE_AND:
 526                 fprintf(stderr, "(and\n");
 527                 dump_grep_expression_1(x->u.binary.left, in+1);
 528                 dump_grep_expression_1(x->u.binary.right, in+1);
 529                 indent(in);
 530                 fprintf(stderr, ")\n");
 531                 break;
 532         case GREP_NODE_OR:
 533                 fprintf(stderr, "(or\n");
 534                 dump_grep_expression_1(x->u.binary.left, in+1);
 535                 dump_grep_expression_1(x->u.binary.right, in+1);
 536                 indent(in);
 537                 fprintf(stderr, ")\n");
 538                 break;
 539         }
 540 }
 541
 542 static void dump_grep_expression(struct grep_opt *opt)
 543 {
 544         struct grep_expr *x = opt->pattern_expression;
 545
 546         if (opt->all_match)
 547                 fprintf(stderr, "[all-match]\n");
 548         dump_grep_expression_1(x, 0);
 549         fflush(NULL);
 550 }
 551
 552 static struct grep_expr *grep_true_expr(void)
 553 {
 554         struct grep_expr *z = xcalloc(1, sizeof(*z));
 555         z->node = GREP_NODE_TRUE;
 556         return z;
 557 }
 558
 559 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 560 {
 561         struct grep_expr *z = xcalloc(1, sizeof(*z));
 562         z->node = GREP_NODE_OR;
 563         z->u.binary.left = left;
 564         z->u.binary.right = right;
 565         return z;
 566 }
 567
 568 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 569 {
 570         struct grep_pat *p;
 571         struct grep_expr *header_expr;
 572         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 573         enum grep_header_field fld;
 574
 575         if (!opt->header_list)
 576                 return NULL;
 577
 578         for (p = opt->header_list; p; p = p->next) {
 579                 if (p->token != GREP_PATTERN_HEAD)
 580                         die("bug: a non-header pattern in grep header list.");
 581                 if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
 582                         die("bug: unknown header field %d", p->field);
 583                 compile_regexp(p, opt);
 584         }
 585
 586         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 587                 header_group[fld] = NULL;
 588
 589         for (p = opt->header_list; p; p = p->next) {
 590                 struct grep_expr *h;
 591                 struct grep_pat *pp = p;
 592
 593                 h = compile_pattern_atom(&pp);
 594                 if (!h || pp != p->next)
 595                         die("bug: malformed header expr");
 596                 if (!header_group[p->field]) {
 597                         header_group[p->field] = h;
 598                         continue;
 599                 }
 600                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 601         }
 602
 603         header_expr = NULL;
 604
 605         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 606                 if (!header_group[fld])
 607                         continue;
 608                 if (!header_expr)
 609                         header_expr = grep_true_expr();
 610                 header_expr = grep_or_expr(header_group[fld], header_expr);
 611         }
 612         return header_expr;
 613 }
 614
 615 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 616 {
 617         struct grep_expr *z = x;
 618
 619         while (x) {
 620                 assert(x->node == GREP_NODE_OR);
 621                 if (x->u.binary.right &&
 622                     x->u.binary.right->node == GREP_NODE_TRUE) {
 623                         x->u.binary.right = y;
 624                         break;
 625                 }
 626                 x = x->u.binary.right;
 627         }
 628         return z;
 629 }
 630
 631 static void compile_grep_patterns_real(struct grep_opt *opt)
 632 {
 633         struct grep_pat *p;
 634         struct grep_expr *header_expr = prep_header_patterns(opt);
 635
 636         for (p = opt->pattern_list; p; p = p->next) {
 637                 switch (p->token) {
 638                 case GREP_PATTERN: /* atom */
 639                 case GREP_PATTERN_HEAD:
 640                 case GREP_PATTERN_BODY:
 641                         compile_regexp(p, opt);
 642                         break;
 643                 default:
 644                         opt->extended = 1;
 645                         break;
 646                 }
 647         }
 648
 649         if (opt->all_match || header_expr)
 650                 opt->extended = 1;
 651         else if (!opt->extended && !opt->debug)
 652                 return;
 653
 654         p = opt->pattern_list;
 655         if (p)
 656                 opt->pattern_expression = compile_pattern_expr(&p);
 657         if (p)
 658                 die("incomplete pattern expression: %s", p->pattern);
 659
 660         if (!header_expr)
 661                 return;
 662
 663         if (!opt->pattern_expression)
 664                 opt->pattern_expression = header_expr;
 665         else if (opt->all_match)
 666                 opt->pattern_expression = grep_splice_or(header_expr,
 667                                                          opt->pattern_expression);
 668         else
 669                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 670                                                        header_expr);
 671         opt->all_match = 1;
 672 }
 673
 674 void compile_grep_patterns(struct grep_opt *opt)
 675 {
 676         compile_grep_patterns_real(opt);
 677         if (opt->debug)
 678                 dump_grep_expression(opt);
 679 }
 680
 681 static void free_pattern_expr(struct grep_expr *x)
 682 {
 683         switch (x->node) {
 684         case GREP_NODE_TRUE:
 685         case GREP_NODE_ATOM:
 686                 break;
 687         case GREP_NODE_NOT:
 688                 free_pattern_expr(x->u.unary);
 689                 break;
 690         case GREP_NODE_AND:
 691         case GREP_NODE_OR:
 692                 free_pattern_expr(x->u.binary.left);
 693                 free_pattern_expr(x->u.binary.right);
 694                 break;
 695         }
 696         free(x);
 697 }
 698
 699 void free_grep_patterns(struct grep_opt *opt)
 700 {
 701         struct grep_pat *p, *n;
 702
 703         for (p = opt->pattern_list; p; p = n) {
 704                 n = p->next;
 705                 switch (p->token) {
 706                 case GREP_PATTERN: /* atom */
 707                 case GREP_PATTERN_HEAD:
 708                 case GREP_PATTERN_BODY:
 709                         if (p->kws)
 710                                 kwsfree(p->kws);
 711                         else if (p->pcre_regexp)
 712                                 free_pcre_regexp(p);
 713                         else
 714                                 regfree(&p->regexp);
 715                         free(p->pattern);
 716                         break;
 717                 default:
 718                         break;
 719                 }
 720                 free(p);
 721         }
 722
 723         if (!opt->extended)
 724                 return;
 725         free_pattern_expr(opt->pattern_expression);
 726 }
 727
 728 static char *end_of_line(char *cp, unsigned long *left)
 729 {
 730         unsigned long l = *left;
 731         while (l && *cp != '\n') {
 732                 l--;
 733                 cp++;
 734         }
 735         *left = l;
 736         return cp;
 737 }
 738
 739 static int word_char(char ch)
 740 {
 741         return isalnum(ch) || ch == '_';
 742 }
 743
 744 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 745                          const char *color)
 746 {
 747         if (want_color(opt->color) && color && color[0]) {
 748                 opt->output(opt, color, strlen(color));
 749                 opt->output(opt, data, size);
 750                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 751         } else
 752                 opt->output(opt, data, size);
 753 }
 754
 755 static void output_sep(struct grep_opt *opt, char sign)
 756 {
 757         if (opt->null_following_name)
 758                 opt->output(opt, "\0", 1);
 759         else
 760                 output_color(opt, &sign, 1, opt->color_sep);
 761 }
 762
 763 static void show_name(struct grep_opt *opt, const char *name)
 764 {
 765         output_color(opt, name, strlen(name), opt->color_filename);
 766         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 767 }
 768
 769 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 770                     regmatch_t *match)
 771 {
 772         struct kwsmatch kwsm;
 773         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 774         if (offset == -1) {
 775                 match->rm_so = match->rm_eo = -1;
 776                 return REG_NOMATCH;
 777         } else {
 778                 match->rm_so = offset;
 779                 match->rm_eo = match->rm_so + kwsm.size[0];
 780                 return 0;
 781         }
 782 }
 783
 784 static int regmatch(const regex_t *preg, char *line, char *eol,
 785                     regmatch_t *match, int eflags)
 786 {
 787 #ifdef REG_STARTEND
 788         match->rm_so = 0;
 789         match->rm_eo = eol - line;
 790         eflags |= REG_STARTEND;
 791 #endif
 792         return regexec(preg, line, 1, match, eflags);
 793 }
 794
 795 static int patmatch(struct grep_pat *p, char *line, char *eol,
 796                     regmatch_t *match, int eflags)
 797 {
 798         int hit;
 799
 800         if (p->fixed)
 801                 hit = !fixmatch(p, line, eol, match);
 802         else if (p->pcre_regexp)
 803                 hit = !pcrematch(p, line, eol, match, eflags);
 804         else
 805                 hit = !regmatch(&p->regexp, line, eol, match, eflags);
 806
 807         return hit;
 808 }
 809
 810 static int strip_timestamp(char *bol, char **eol_p)
 811 {
 812         char *eol = *eol_p;
 813         int ch;
 814
 815         while (bol < --eol) {
 816                 if (*eol != '>')
 817                         continue;
 818                 *eol_p = ++eol;
 819                 ch = *eol;
 820                 *eol = '\0';
 821                 return ch;
 822         }
 823         return 0;
 824 }
 825
 826 static struct {
 827         const char *field;
 828         size_t len;
 829 } header_field[] = {
 830         { "author ", 7 },
 831         { "committer ", 10 },
 832         { "reflog ", 7 },
 833 };
 834
 835 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 836                              enum grep_context ctx,
 837                              regmatch_t *pmatch, int eflags)
 838 {
 839         int hit = 0;
 840         int saved_ch = 0;
 841         const char *start = bol;
 842
 843         if ((p->token != GREP_PATTERN) &&
 844             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 845                 return 0;
 846
 847         if (p->token == GREP_PATTERN_HEAD) {
 848                 const char *field;
 849                 size_t len;
 850                 assert(p->field < ARRAY_SIZE(header_field));
 851                 field = header_field[p->field].field;
 852                 len = header_field[p->field].len;
 853                 if (strncmp(bol, field, len))
 854                         return 0;
 855                 bol += len;
 856                 switch (p->field) {
 857                 case GREP_HEADER_AUTHOR:
 858                 case GREP_HEADER_COMMITTER:
 859                         saved_ch = strip_timestamp(bol, &eol);
 860                         break;
 861                 default:
 862                         break;
 863                 }
 864         }
 865
 866  again:
 867         hit = patmatch(p, bol, eol, pmatch, eflags);
 868
 869         if (hit && p->word_regexp) {
 870                 if ((pmatch[0].rm_so < 0) ||
 871                     (eol - bol) < pmatch[0].rm_so ||
 872                     (pmatch[0].rm_eo < 0) ||
 873                     (eol - bol) < pmatch[0].rm_eo)
 874                         die("regexp returned nonsense");
 875
 876                 /* Match beginning must be either beginning of the
 877                  * line, or at word boundary (i.e. the last char must
 878                  * not be a word char).  Similarly, match end must be
 879                  * either end of the line, or at word boundary
 880                  * (i.e. the next char must not be a word char).
 881                  */
 882                 if ( ((pmatch[0].rm_so == 0) ||
 883                       !word_char(bol[pmatch[0].rm_so-1])) &&
 884                      ((pmatch[0].rm_eo == (eol-bol)) ||
 885                       !word_char(bol[pmatch[0].rm_eo])) )
 886                         ;
 887                 else
 888                         hit = 0;
 889
 890                 /* Words consist of at least one character. */
 891                 if (pmatch->rm_so == pmatch->rm_eo)
 892                         hit = 0;
 893
 894                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 895                         /* There could be more than one match on the
 896                          * line, and the first match might not be
 897                          * strict word match.  But later ones could be!
 898                          * Forward to the next possible start, i.e. the
 899                          * next position following a non-word char.
 900                          */
 901                         bol = pmatch[0].rm_so + bol + 1;
 902                         while (word_char(bol[-1]) && bol < eol)
 903                                 bol++;
 904                         eflags |= REG_NOTBOL;
 905                         if (bol < eol)
 906                                 goto again;
 907                 }
 908         }
 909         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 910                 *eol = saved_ch;
 911         if (hit) {
 912                 pmatch[0].rm_so += bol - start;
 913                 pmatch[0].rm_eo += bol - start;
 914         }
 915         return hit;
 916 }
 917
 918 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 919                            enum grep_context ctx, int collect_hits)
 920 {
 921         int h = 0;
 922         regmatch_t match;
 923
 924         if (!x)
 925                 die("Not a valid grep expression");
 926         switch (x->node) {
 927         case GREP_NODE_TRUE:
 928                 h = 1;
 929                 break;
 930         case GREP_NODE_ATOM:
 931                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 932                 break;
 933         case GREP_NODE_NOT:
 934                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 935                 break;
 936         case GREP_NODE_AND:
 937                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 938                         return 0;
 939                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 940                 break;
 941         case GREP_NODE_OR:
 942                 if (!collect_hits)
 943                         return (match_expr_eval(x->u.binary.left,
 944                                                 bol, eol, ctx, 0) ||
 945                                 match_expr_eval(x->u.binary.right,
 946                                                 bol, eol, ctx, 0));
 947                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 948                 x->u.binary.left->hit |= h;
 949                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 950                 break;
 951         default:
 952                 die("Unexpected node type (internal error) %d", x->node);
 953         }
 954         if (collect_hits)
 955                 x->hit |= h;
 956         return h;
 957 }
 958
 959 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 960                       enum grep_context ctx, int collect_hits)
 961 {
 962         struct grep_expr *x = opt->pattern_expression;
 963         return match_expr_eval(x, bol, eol, ctx, collect_hits);
 964 }
 965
 966 static int match_line(struct grep_opt *opt, char *bol, char *eol,
 967                       enum grep_context ctx, int collect_hits)
 968 {
 969         struct grep_pat *p;
 970         regmatch_t match;
 971
 972         if (opt->extended)
 973                 return match_expr(opt, bol, eol, ctx, collect_hits);
 974
 975         /* we do not call with collect_hits without being extended */
 976         for (p = opt->pattern_list; p; p = p->next) {
 977                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 978                         return 1;
 979         }
 980         return 0;
 981 }
 982
 983 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 984                               enum grep_context ctx,
 985                               regmatch_t *pmatch, int eflags)
 986 {
 987         regmatch_t match;
 988
 989         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 990                 return 0;
 991         if (match.rm_so < 0 || match.rm_eo < 0)
 992                 return 0;
 993         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 994                 if (match.rm_so > pmatch->rm_so)
 995                         return 1;
 996                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 997                         return 1;
 998         }
 999         pmatch->rm_so = match.rm_so;
1000         pmatch->rm_eo = match.rm_eo;
1001         return 1;
1002 }
1003
1004 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1005                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1006 {
1007         struct grep_pat *p;
1008         int hit = 0;
1009
1010         pmatch->rm_so = pmatch->rm_eo = -1;
1011         if (bol < eol) {
1012                 for (p = opt->pattern_list; p; p = p->next) {
1013                         switch (p->token) {
1014                         case GREP_PATTERN: /* atom */
1015                         case GREP_PATTERN_HEAD:
1016                         case GREP_PATTERN_BODY:
1017                                 hit |= match_next_pattern(p, bol, eol, ctx,
1018                                                           pmatch, eflags);
1019                                 break;
1020                         default:
1021                                 break;
1022                         }
1023                 }
1024         }
1025         return hit;
1026 }
1027
1028 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1029                       const char *name, unsigned lno, char sign)
1030 {
1031         int rest = eol - bol;
1032         char *line_color = NULL;
1033
1034         if (opt->file_break && opt->last_shown == 0) {
1035                 if (opt->show_hunk_mark)
1036                         opt->output(opt, "\n", 1);
1037         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1038                 if (opt->last_shown == 0) {
1039                         if (opt->show_hunk_mark) {
1040                                 output_color(opt, "--", 2, opt->color_sep);
1041                                 opt->output(opt, "\n", 1);
1042                         }
1043                 } else if (lno > opt->last_shown + 1) {
1044                         output_color(opt, "--", 2, opt->color_sep);
1045                         opt->output(opt, "\n", 1);
1046                 }
1047         }
1048         if (opt->heading && opt->last_shown == 0) {
1049                 output_color(opt, name, strlen(name), opt->color_filename);
1050                 opt->output(opt, "\n", 1);
1051         }
1052         opt->last_shown = lno;
1053
1054         if (!opt->heading && opt->pathname) {
1055                 output_color(opt, name, strlen(name), opt->color_filename);
1056                 output_sep(opt, sign);
1057         }
1058         if (opt->linenum) {
1059                 char buf[32];
1060                 snprintf(buf, sizeof(buf), "%d", lno);
1061                 output_color(opt, buf, strlen(buf), opt->color_lineno);
1062                 output_sep(opt, sign);
1063         }
1064         if (opt->color) {
1065                 regmatch_t match;
1066                 enum grep_context ctx = GREP_CONTEXT_BODY;
1067                 int ch = *eol;
1068                 int eflags = 0;
1069
1070                 if (sign == ':')
1071                         line_color = opt->color_selected;
1072                 else if (sign == '-')
1073                         line_color = opt->color_context;
1074                 else if (sign == '=')
1075                         line_color = opt->color_function;
1076                 *eol = '\0';
1077                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1078                         if (match.rm_so == match.rm_eo)
1079                                 break;
1080
1081                         output_color(opt, bol, match.rm_so, line_color);
1082                         output_color(opt, bol + match.rm_so,
1083                                      match.rm_eo - match.rm_so,
1084                                      opt->color_match);
1085                         bol += match.rm_eo;
1086                         rest -= match.rm_eo;
1087                         eflags = REG_NOTBOL;
1088                 }
1089                 *eol = ch;
1090         }
1091         output_color(opt, bol, rest, line_color);
1092         opt->output(opt, "\n", 1);
1093 }
1094
1095 #ifndef NO_PTHREADS
1096 int grep_use_locks;
1097
1098 /*
1099  * This lock protects access to the gitattributes machinery, which is
1100  * not thread-safe.
1101  */
1102 pthread_mutex_t grep_attr_mutex;
1103
1104 static inline void grep_attr_lock(void)
1105 {
1106         if (grep_use_locks)
1107                 pthread_mutex_lock(&grep_attr_mutex);
1108 }
1109
1110 static inline void grep_attr_unlock(void)
1111 {
1112         if (grep_use_locks)
1113                 pthread_mutex_unlock(&grep_attr_mutex);
1114 }
1115
1116 /*
1117  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1118  */
1119 pthread_mutex_t grep_read_mutex;
1120
1121 #else
1122 #define grep_attr_lock()
1123 #define grep_attr_unlock()
1124 #endif
1125
1126 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1127 {
1128         xdemitconf_t *xecfg = opt->priv;
1129         if (xecfg && !xecfg->find_func) {
1130                 grep_source_load_driver(gs);
1131                 if (gs->driver->funcname.pattern) {
1132                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1133                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1134                 } else {
1135                         xecfg = opt->priv = NULL;
1136                 }
1137         }
1138
1139         if (xecfg) {
1140                 char buf[1];
1141                 return xecfg->find_func(bol, eol - bol, buf, 1,
1142                                         xecfg->find_func_priv) >= 0;
1143         }
1144
1145         if (bol == eol)
1146                 return 0;
1147         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1148                 return 1;
1149         return 0;
1150 }
1151
1152 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1153                                char *bol, unsigned lno)
1154 {
1155         while (bol > gs->buf) {
1156                 char *eol = --bol;
1157
1158                 while (bol > gs->buf && bol[-1] != '\n')
1159                         bol--;
1160                 lno--;
1161
1162                 if (lno <= opt->last_shown)
1163                         break;
1164
1165                 if (match_funcname(opt, gs, bol, eol)) {
1166                         show_line(opt, bol, eol, gs->name, lno, '=');
1167                         break;
1168                 }
1169         }
1170 }
1171
1172 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1173                              char *bol, char *end, unsigned lno)
1174 {
1175         unsigned cur = lno, from = 1, funcname_lno = 0;
1176         int funcname_needed = !!opt->funcname;
1177
1178         if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1179                 funcname_needed = 2;
1180
1181         if (opt->pre_context < lno)
1182                 from = lno - opt->pre_context;
1183         if (from <= opt->last_shown)
1184                 from = opt->last_shown + 1;
1185
1186         /* Rewind. */
1187         while (bol > gs->buf &&
1188                cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1189                 char *eol = --bol;
1190
1191                 while (bol > gs->buf && bol[-1] != '\n')
1192                         bol--;
1193                 cur--;
1194                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1195                         funcname_lno = cur;
1196                         funcname_needed = 0;
1197                 }
1198         }
1199
1200         /* We need to look even further back to find a function signature. */
1201         if (opt->funcname && funcname_needed)
1202                 show_funcname_line(opt, gs, bol, cur);
1203
1204         /* Back forward. */
1205         while (cur < lno) {
1206                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1207
1208                 while (*eol != '\n')
1209                         eol++;
1210                 show_line(opt, bol, eol, gs->name, cur, sign);
1211                 bol = eol + 1;
1212                 cur++;
1213         }
1214 }
1215
1216 static int should_lookahead(struct grep_opt *opt)
1217 {
1218         struct grep_pat *p;
1219
1220         if (opt->extended)
1221                 return 0; /* punt for too complex stuff */
1222         if (opt->invert)
1223                 return 0;
1224         for (p = opt->pattern_list; p; p = p->next) {
1225                 if (p->token != GREP_PATTERN)
1226                         return 0; /* punt for "header only" and stuff */
1227         }
1228         return 1;
1229 }
1230
1231 static int look_ahead(struct grep_opt *opt,
1232                       unsigned long *left_p,
1233                       unsigned *lno_p,
1234                       char **bol_p)
1235 {
1236         unsigned lno = *lno_p;
1237         char *bol = *bol_p;
1238         struct grep_pat *p;
1239         char *sp, *last_bol;
1240         regoff_t earliest = -1;
1241
1242         for (p = opt->pattern_list; p; p = p->next) {
1243                 int hit;
1244                 regmatch_t m;
1245
1246                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1247                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1248                         continue;
1249                 if (earliest < 0 || m.rm_so < earliest)
1250                         earliest = m.rm_so;
1251         }
1252
1253         if (earliest < 0) {
1254                 *bol_p = bol + *left_p;
1255                 *left_p = 0;
1256                 return 1;
1257         }
1258         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1259                 ; /* find the beginning of the line */
1260         last_bol = sp;
1261
1262         for (sp = bol; sp < last_bol; sp++) {
1263                 if (*sp == '\n')
1264                         lno++;
1265         }
1266         *left_p -= last_bol - bol;
1267         *bol_p = last_bol;
1268         *lno_p = lno;
1269         return 0;
1270 }
1271
1272 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1273 {
1274         fwrite(buf, size, 1, stdout);
1275 }
1276
1277 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1278 {
1279         char *bol;
1280         unsigned long left;
1281         unsigned lno = 1;
1282         unsigned last_hit = 0;
1283         int binary_match_only = 0;
1284         unsigned count = 0;
1285         int try_lookahead = 0;
1286         int show_function = 0;
1287         enum grep_context ctx = GREP_CONTEXT_HEAD;
1288         xdemitconf_t xecfg;
1289
1290         if (!opt->output)
1291                 opt->output = std_output;
1292
1293         if (opt->pre_context || opt->post_context || opt->file_break ||
1294             opt->funcbody) {
1295                 /* Show hunk marks, except for the first file. */
1296                 if (opt->last_shown)
1297                         opt->show_hunk_mark = 1;
1298                 /*
1299                  * If we're using threads then we can't easily identify
1300                  * the first file.  Always put hunk marks in that case
1301                  * and skip the very first one later in work_done().
1302                  */
1303                 if (opt->output != std_output)
1304                         opt->show_hunk_mark = 1;
1305         }
1306         opt->last_shown = 0;
1307
1308         switch (opt->binary) {
1309         case GREP_BINARY_DEFAULT:
1310                 if (grep_source_is_binary(gs))
1311                         binary_match_only = 1;
1312                 break;
1313         case GREP_BINARY_NOMATCH:
1314                 if (grep_source_is_binary(gs))
1315                         return 0; /* Assume unmatch */
1316                 break;
1317         case GREP_BINARY_TEXT:
1318                 break;
1319         default:
1320                 die("bug: unknown binary handling mode");
1321         }
1322
1323         memset(&xecfg, 0, sizeof(xecfg));
1324         opt->priv = &xecfg;
1325
1326         try_lookahead = should_lookahead(opt);
1327
1328         if (grep_source_load(gs) < 0)
1329                 return 0;
1330
1331         bol = gs->buf;
1332         left = gs->size;
1333         while (left) {
1334                 char *eol, ch;
1335                 int hit;
1336
1337                 /*
1338                  * look_ahead() skips quickly to the line that possibly
1339                  * has the next hit; don't call it if we need to do
1340                  * something more than just skipping the current line
1341                  * in response to an unmatch for the current line.  E.g.
1342                  * inside a post-context window, we will show the current
1343                  * line as a context around the previous hit when it
1344                  * doesn't hit.
1345                  */
1346                 if (try_lookahead
1347                     && !(last_hit
1348                          && (show_function ||
1349                              lno <= last_hit + opt->post_context))
1350                     && look_ahead(opt, &left, &lno, &bol))
1351                         break;
1352                 eol = end_of_line(bol, &left);
1353                 ch = *eol;
1354                 *eol = 0;
1355
1356                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1357                         ctx = GREP_CONTEXT_BODY;
1358
1359                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1360                 *eol = ch;
1361
1362                 if (collect_hits)
1363                         goto next_line;
1364
1365                 /* "grep -v -e foo -e bla" should list lines
1366                  * that do not have either, so inversion should
1367                  * be done outside.
1368                  */
1369                 if (opt->invert)
1370                         hit = !hit;
1371                 if (opt->unmatch_name_only) {
1372                         if (hit)
1373                                 return 0;
1374                         goto next_line;
1375                 }
1376                 if (hit) {
1377                         count++;
1378                         if (opt->status_only)
1379                                 return 1;
1380                         if (opt->name_only) {
1381                                 show_name(opt, gs->name);
1382                                 return 1;
1383                         }
1384                         if (opt->count)
1385                                 goto next_line;
1386                         if (binary_match_only) {
1387                                 opt->output(opt, "Binary file ", 12);
1388                                 output_color(opt, gs->name, strlen(gs->name),
1389                                              opt->color_filename);
1390                                 opt->output(opt, " matches\n", 9);
1391                                 return 1;
1392                         }
1393                         /* Hit at this line.  If we haven't shown the
1394                          * pre-context lines, we would need to show them.
1395                          */
1396                         if (opt->pre_context || opt->funcbody)
1397                                 show_pre_context(opt, gs, bol, eol, lno);
1398                         else if (opt->funcname)
1399                                 show_funcname_line(opt, gs, bol, lno);
1400                         show_line(opt, bol, eol, gs->name, lno, ':');
1401                         last_hit = lno;
1402                         if (opt->funcbody)
1403                                 show_function = 1;
1404                         goto next_line;
1405                 }
1406                 if (show_function && match_funcname(opt, gs, bol, eol))
1407                         show_function = 0;
1408                 if (show_function ||
1409                     (last_hit && lno <= last_hit + opt->post_context)) {
1410                         /* If the last hit is within the post context,
1411                          * we need to show this line.
1412                          */
1413                         show_line(opt, bol, eol, gs->name, lno, '-');
1414                 }
1415
1416         next_line:
1417                 bol = eol + 1;
1418                 if (!left)
1419                         break;
1420                 left--;
1421                 lno++;
1422         }
1423
1424         if (collect_hits)
1425                 return 0;
1426
1427         if (opt->status_only)
1428                 return 0;
1429         if (opt->unmatch_name_only) {
1430                 /* We did not see any hit, so we want to show this */
1431                 show_name(opt, gs->name);
1432                 return 1;
1433         }
1434
1435         xdiff_clear_find_func(&xecfg);
1436         opt->priv = NULL;
1437
1438         /* NEEDSWORK:
1439          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1440          * which feels mostly useless but sometimes useful.  Maybe
1441          * make it another option?  For now suppress them.
1442          */
1443         if (opt->count && count) {
1444                 char buf[32];
1445                 output_color(opt, gs->name, strlen(gs->name), opt->color_filename);
1446                 output_sep(opt, ':');
1447                 snprintf(buf, sizeof(buf), "%u\n", count);
1448                 opt->output(opt, buf, strlen(buf));
1449                 return 1;
1450         }
1451         return !!last_hit;
1452 }
1453
1454 static void clr_hit_marker(struct grep_expr *x)
1455 {
1456         /* All-hit markers are meaningful only at the very top level
1457          * OR node.
1458          */
1459         while (1) {
1460                 x->hit = 0;
1461                 if (x->node != GREP_NODE_OR)
1462                         return;
1463                 x->u.binary.left->hit = 0;
1464                 x = x->u.binary.right;
1465         }
1466 }
1467
1468 static int chk_hit_marker(struct grep_expr *x)
1469 {
1470         /* Top level nodes have hit markers.  See if they all are hits */
1471         while (1) {
1472                 if (x->node != GREP_NODE_OR)
1473                         return x->hit;
1474                 if (!x->u.binary.left->hit)
1475                         return 0;
1476                 x = x->u.binary.right;
1477         }
1478 }
1479
1480 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1481 {
1482         /*
1483          * we do not have to do the two-pass grep when we do not check
1484          * buffer-wide "all-match".
1485          */
1486         if (!opt->all_match)
1487                 return grep_source_1(opt, gs, 0);
1488
1489         /* Otherwise the toplevel "or" terms hit a bit differently.
1490          * We first clear hit markers from them.
1491          */
1492         clr_hit_marker(opt->pattern_expression);
1493         grep_source_1(opt, gs, 1);
1494
1495         if (!chk_hit_marker(opt->pattern_expression))
1496                 return 0;
1497
1498         return grep_source_1(opt, gs, 0);
1499 }
1500
1501 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1502 {
1503         struct grep_source gs;
1504         int r;
1505
1506         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL);
1507         gs.buf = buf;
1508         gs.size = size;
1509
1510         r = grep_source(opt, &gs);
1511
1512         grep_source_clear(&gs);
1513         return r;
1514 }
1515
1516 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1517                       const char *name, const void *identifier)
1518 {
1519         gs->type = type;
1520         gs->name = name ? xstrdup(name) : NULL;
1521         gs->buf = NULL;
1522         gs->size = 0;
1523         gs->driver = NULL;
1524
1525         switch (type) {
1526         case GREP_SOURCE_FILE:
1527                 gs->identifier = xstrdup(identifier);
1528                 break;
1529         case GREP_SOURCE_SHA1:
1530                 gs->identifier = xmalloc(20);
1531                 memcpy(gs->identifier, identifier, 20);
1532                 break;
1533         case GREP_SOURCE_BUF:
1534                 gs->identifier = NULL;
1535         }
1536 }
1537
1538 void grep_source_clear(struct grep_source *gs)
1539 {
1540         free(gs->name);
1541         gs->name = NULL;
1542         free(gs->identifier);
1543         gs->identifier = NULL;
1544         grep_source_clear_data(gs);
1545 }
1546
1547 void grep_source_clear_data(struct grep_source *gs)
1548 {
1549         switch (gs->type) {
1550         case GREP_SOURCE_FILE:
1551         case GREP_SOURCE_SHA1:
1552                 free(gs->buf);
1553                 gs->buf = NULL;
1554                 gs->size = 0;
1555                 break;
1556         case GREP_SOURCE_BUF:
1557                 /* leave user-provided buf intact */
1558                 break;
1559         }
1560 }
1561
1562 static int grep_source_load_sha1(struct grep_source *gs)
1563 {
1564         enum object_type type;
1565
1566         grep_read_lock();
1567         gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1568         grep_read_unlock();
1569
1570         if (!gs->buf)
1571                 return error(_("'%s': unable to read %s"),
1572                              gs->name,
1573                              sha1_to_hex(gs->identifier));
1574         return 0;
1575 }
1576
1577 static int grep_source_load_file(struct grep_source *gs)
1578 {
1579         const char *filename = gs->identifier;
1580         struct stat st;
1581         char *data;
1582         size_t size;
1583         int i;
1584
1585         if (lstat(filename, &st) < 0) {
1586         err_ret:
1587                 if (errno != ENOENT)
1588                         error(_("'%s': %s"), filename, strerror(errno));
1589                 return -1;
1590         }
1591         if (!S_ISREG(st.st_mode))
1592                 return -1;
1593         size = xsize_t(st.st_size);
1594         i = open(filename, O_RDONLY);
1595         if (i < 0)
1596                 goto err_ret;
1597         data = xmalloc(size + 1);
1598         if (st.st_size != read_in_full(i, data, size)) {
1599                 error(_("'%s': short read %s"), filename, strerror(errno));
1600                 close(i);
1601                 free(data);
1602                 return -1;
1603         }
1604         close(i);
1605         data[size] = 0;
1606
1607         gs->buf = data;
1608         gs->size = size;
1609         return 0;
1610 }
1611
1612 static int grep_source_load(struct grep_source *gs)
1613 {
1614         if (gs->buf)
1615                 return 0;
1616
1617         switch (gs->type) {
1618         case GREP_SOURCE_FILE:
1619                 return grep_source_load_file(gs);
1620         case GREP_SOURCE_SHA1:
1621                 return grep_source_load_sha1(gs);
1622         case GREP_SOURCE_BUF:
1623                 return gs->buf ? 0 : -1;
1624         }
1625         die("BUG: invalid grep_source type");
1626 }
1627
1628 void grep_source_load_driver(struct grep_source *gs)
1629 {
1630         if (gs->driver)
1631                 return;
1632
1633         grep_attr_lock();
1634         gs->driver = userdiff_find_by_path(gs->name);
1635         if (!gs->driver)
1636                 gs->driver = userdiff_find_by_name("default");
1637         grep_attr_unlock();
1638 }
1639
1640 static int grep_source_is_binary(struct grep_source *gs)
1641 {
1642         grep_source_load_driver(gs);
1643         if (gs->driver->binary != -1)
1644                 return gs->driver->binary;
1645
1646         if (!grep_source_load(gs))
1647                 return buffer_is_binary(gs->buf, gs->size);
1648
1649         return 0;
1650 }