grep.c

   1 #include "cache.h"
   2 #include "grep.h"
   3 #include "userdiff.h"
   4 #include "xdiff-interface.h"
   5 #include "diff.h"
   6 #include "diffcore.h"
   7
   8 static int grep_source_load(struct grep_source *gs);
   9 static int grep_source_is_binary(struct grep_source *gs);
  10
  11 static struct grep_opt grep_defaults;
  12
  13 /*
  14  * Initialize the grep_defaults template with hardcoded defaults.
  15  * We could let the compiler do this, but without C99 initializers
  16  * the code gets unwieldy and unreadable, so...
  17  */
  18 void init_grep_defaults(void)
  19 {
  20         struct grep_opt *opt = &grep_defaults;
  21         static int run_once;
  22
  23         if (run_once)
  24                 return;
  25         run_once++;
  26
  27         memset(opt, 0, sizeof(*opt));
  28         opt->relative = 1;
  29         opt->pathname = 1;
  30         opt->regflags = REG_NEWLINE;
  31         opt->max_depth = -1;
  32         opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  33         opt->extended_regexp_option = 0;
  34         color_set(opt->color_context, "");
  35         color_set(opt->color_filename, "");
  36         color_set(opt->color_function, "");
  37         color_set(opt->color_lineno, "");
  38         color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
  39         color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
  40         color_set(opt->color_selected, "");
  41         color_set(opt->color_sep, GIT_COLOR_CYAN);
  42         opt->color = -1;
  43 }
  44
  45 static int parse_pattern_type_arg(const char *opt, const char *arg)
  46 {
  47         if (!strcmp(arg, "default"))
  48                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  49         else if (!strcmp(arg, "basic"))
  50                 return GREP_PATTERN_TYPE_BRE;
  51         else if (!strcmp(arg, "extended"))
  52                 return GREP_PATTERN_TYPE_ERE;
  53         else if (!strcmp(arg, "fixed"))
  54                 return GREP_PATTERN_TYPE_FIXED;
  55         else if (!strcmp(arg, "perl"))
  56                 return GREP_PATTERN_TYPE_PCRE;
  57         die("bad %s argument: %s", opt, arg);
  58 }
  59
  60 /*
  61  * Read the configuration file once and store it in
  62  * the grep_defaults template.
  63  */
  64 int grep_config(const char *var, const char *value, void *cb)
  65 {
  66         struct grep_opt *opt = &grep_defaults;
  67         char *color = NULL;
  68
  69         if (userdiff_config(var, value) < 0)
  70                 return -1;
  71
  72         if (!strcmp(var, "grep.extendedregexp")) {
  73                 if (git_config_bool(var, value))
  74                         opt->extended_regexp_option = 1;
  75                 else
  76                         opt->extended_regexp_option = 0;
  77                 return 0;
  78         }
  79
  80         if (!strcmp(var, "grep.patterntype")) {
  81                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
  82                 return 0;
  83         }
  84
  85         if (!strcmp(var, "grep.linenumber")) {
  86                 opt->linenum = git_config_bool(var, value);
  87                 return 0;
  88         }
  89
  90         if (!strcmp(var, "grep.fullname")) {
  91                 opt->relative = !git_config_bool(var, value);
  92                 return 0;
  93         }
  94
  95         if (!strcmp(var, "color.grep"))
  96                 opt->color = git_config_colorbool(var, value);
  97         else if (!strcmp(var, "color.grep.context"))
  98                 color = opt->color_context;
  99         else if (!strcmp(var, "color.grep.filename"))
 100                 color = opt->color_filename;
 101         else if (!strcmp(var, "color.grep.function"))
 102                 color = opt->color_function;
 103         else if (!strcmp(var, "color.grep.linenumber"))
 104                 color = opt->color_lineno;
 105         else if (!strcmp(var, "color.grep.matchcontext"))
 106                 color = opt->color_match_context;
 107         else if (!strcmp(var, "color.grep.matchselected"))
 108                 color = opt->color_match_selected;
 109         else if (!strcmp(var, "color.grep.selected"))
 110                 color = opt->color_selected;
 111         else if (!strcmp(var, "color.grep.separator"))
 112                 color = opt->color_sep;
 113         else if (!strcmp(var, "color.grep.match")) {
 114                 int rc = 0;
 115                 if (!value)
 116                         return config_error_nonbool(var);
 117                 rc |= color_parse(value, opt->color_match_context);
 118                 rc |= color_parse(value, opt->color_match_selected);
 119                 return rc;
 120         }
 121
 122         if (color) {
 123                 if (!value)
 124                         return config_error_nonbool(var);
 125                 return color_parse(value, color);
 126         }
 127         return 0;
 128 }
 129
 130 /*
 131  * Initialize one instance of grep_opt and copy the
 132  * default values from the template we read the configuration
 133  * information in an earlier call to git_config(grep_config).
 134  */
 135 void grep_init(struct grep_opt *opt, const char *prefix)
 136 {
 137         struct grep_opt *def = &grep_defaults;
 138
 139         memset(opt, 0, sizeof(*opt));
 140         opt->prefix = prefix;
 141         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 142         opt->pattern_tail = &opt->pattern_list;
 143         opt->header_tail = &opt->header_list;
 144
 145         opt->color = def->color;
 146         opt->extended_regexp_option = def->extended_regexp_option;
 147         opt->pattern_type_option = def->pattern_type_option;
 148         opt->linenum = def->linenum;
 149         opt->max_depth = def->max_depth;
 150         opt->pathname = def->pathname;
 151         opt->regflags = def->regflags;
 152         opt->relative = def->relative;
 153
 154         color_set(opt->color_context, def->color_context);
 155         color_set(opt->color_filename, def->color_filename);
 156         color_set(opt->color_function, def->color_function);
 157         color_set(opt->color_lineno, def->color_lineno);
 158         color_set(opt->color_match_context, def->color_match_context);
 159         color_set(opt->color_match_selected, def->color_match_selected);
 160         color_set(opt->color_selected, def->color_selected);
 161         color_set(opt->color_sep, def->color_sep);
 162 }
 163
 164 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 165 {
 166         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 167                 grep_set_pattern_type_option(pattern_type, opt);
 168         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 169                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 170         else if (opt->extended_regexp_option)
 171                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 172 }
 173
 174 void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 175 {
 176         switch (pattern_type) {
 177         case GREP_PATTERN_TYPE_UNSPECIFIED:
 178                 /* fall through */
 179
 180         case GREP_PATTERN_TYPE_BRE:
 181                 opt->fixed = 0;
 182                 opt->pcre = 0;
 183                 opt->regflags &= ~REG_EXTENDED;
 184                 break;
 185
 186         case GREP_PATTERN_TYPE_ERE:
 187                 opt->fixed = 0;
 188                 opt->pcre = 0;
 189                 opt->regflags |= REG_EXTENDED;
 190                 break;
 191
 192         case GREP_PATTERN_TYPE_FIXED:
 193                 opt->fixed = 1;
 194                 opt->pcre = 0;
 195                 opt->regflags &= ~REG_EXTENDED;
 196                 break;
 197
 198         case GREP_PATTERN_TYPE_PCRE:
 199                 opt->fixed = 0;
 200                 opt->pcre = 1;
 201                 opt->regflags &= ~REG_EXTENDED;
 202                 break;
 203         }
 204 }
 205
 206 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 207                                         const char *origin, int no,
 208                                         enum grep_pat_token t,
 209                                         enum grep_header_field field)
 210 {
 211         struct grep_pat *p = xcalloc(1, sizeof(*p));
 212         p->pattern = xmemdupz(pat, patlen);
 213         p->patternlen = patlen;
 214         p->origin = origin;
 215         p->no = no;
 216         p->token = t;
 217         p->field = field;
 218         return p;
 219 }
 220
 221 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 222 {
 223         **tail = p;
 224         *tail = &p->next;
 225         p->next = NULL;
 226
 227         switch (p->token) {
 228         case GREP_PATTERN: /* atom */
 229         case GREP_PATTERN_HEAD:
 230         case GREP_PATTERN_BODY:
 231                 for (;;) {
 232                         struct grep_pat *new_pat;
 233                         size_t len = 0;
 234                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 235                         while (++len <= p->patternlen) {
 236                                 if (*(--cp) == '\n') {
 237                                         nl = cp;
 238                                         break;
 239                                 }
 240                         }
 241                         if (!nl)
 242                                 break;
 243                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 244                                                   p->no, p->token, p->field);
 245                         new_pat->next = p->next;
 246                         if (!p->next)
 247                                 *tail = &new_pat->next;
 248                         p->next = new_pat;
 249                         *nl = '\0';
 250                         p->patternlen -= len;
 251                 }
 252                 break;
 253         default:
 254                 break;
 255         }
 256 }
 257
 258 void append_header_grep_pattern(struct grep_opt *opt,
 259                                 enum grep_header_field field, const char *pat)
 260 {
 261         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 262                                              GREP_PATTERN_HEAD, field);
 263         if (field == GREP_HEADER_REFLOG)
 264                 opt->use_reflog_filter = 1;
 265         do_append_grep_pat(&opt->header_tail, p);
 266 }
 267
 268 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 269                          const char *origin, int no, enum grep_pat_token t)
 270 {
 271         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 272 }
 273
 274 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 275                      const char *origin, int no, enum grep_pat_token t)
 276 {
 277         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 278         do_append_grep_pat(&opt->pattern_tail, p);
 279 }
 280
 281 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 282 {
 283         struct grep_pat *pat;
 284         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 285         *ret = *opt;
 286
 287         ret->pattern_list = NULL;
 288         ret->pattern_tail = &ret->pattern_list;
 289
 290         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 291         {
 292                 if(pat->token == GREP_PATTERN_HEAD)
 293                         append_header_grep_pattern(ret, pat->field,
 294                                                    pat->pattern);
 295                 else
 296                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 297                                         pat->origin, pat->no, pat->token);
 298         }
 299
 300         return ret;
 301 }
 302
 303 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 304                 const char *error)
 305 {
 306         char where[1024];
 307
 308         if (p->no)
 309                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 310         else if (p->origin)
 311                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 312         else
 313                 where[0] = 0;
 314
 315         die("%s'%s': %s", where, p->pattern, error);
 316 }
 317
 318 #ifdef USE_LIBPCRE
 319 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 320 {
 321         const char *error;
 322         int erroffset;
 323         int options = PCRE_MULTILINE;
 324
 325         if (opt->ignore_case)
 326                 options |= PCRE_CASELESS;
 327
 328         p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 329                         NULL);
 330         if (!p->pcre_regexp)
 331                 compile_regexp_failed(p, error);
 332
 333         p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 334         if (!p->pcre_extra_info && error)
 335                 die("%s", error);
 336 }
 337
 338 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 339                 regmatch_t *match, int eflags)
 340 {
 341         int ovector[30], ret, flags = 0;
 342
 343         if (eflags & REG_NOTBOL)
 344                 flags |= PCRE_NOTBOL;
 345
 346         ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 347                         0, flags, ovector, ARRAY_SIZE(ovector));
 348         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 349                 die("pcre_exec failed with error code %d", ret);
 350         if (ret > 0) {
 351                 ret = 0;
 352                 match->rm_so = ovector[0];
 353                 match->rm_eo = ovector[1];
 354         }
 355
 356         return ret;
 357 }
 358
 359 static void free_pcre_regexp(struct grep_pat *p)
 360 {
 361         pcre_free(p->pcre_regexp);
 362         pcre_free(p->pcre_extra_info);
 363 }
 364 #else /* !USE_LIBPCRE */
 365 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 366 {
 367         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 368 }
 369
 370 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 371                 regmatch_t *match, int eflags)
 372 {
 373         return 1;
 374 }
 375
 376 static void free_pcre_regexp(struct grep_pat *p)
 377 {
 378 }
 379 #endif /* !USE_LIBPCRE */
 380
 381 static int is_fixed(const char *s, size_t len)
 382 {
 383         size_t i;
 384
 385         /* regcomp cannot accept patterns with NULs so we
 386          * consider any pattern containing a NUL fixed.
 387          */
 388         if (memchr(s, 0, len))
 389                 return 1;
 390
 391         for (i = 0; i < len; i++) {
 392                 if (is_regex_special(s[i]))
 393                         return 0;
 394         }
 395
 396         return 1;
 397 }
 398
 399 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 400 {
 401         int err;
 402
 403         p->word_regexp = opt->word_regexp;
 404         p->ignore_case = opt->ignore_case;
 405
 406         if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 407                 p->fixed = 1;
 408         else
 409                 p->fixed = 0;
 410
 411         if (p->fixed) {
 412                 if (opt->regflags & REG_ICASE || p->ignore_case)
 413                         p->kws = kwsalloc(tolower_trans_tbl);
 414                 else
 415                         p->kws = kwsalloc(NULL);
 416                 kwsincr(p->kws, p->pattern, p->patternlen);
 417                 kwsprep(p->kws);
 418                 return;
 419         }
 420
 421         if (opt->pcre) {
 422                 compile_pcre_regexp(p, opt);
 423                 return;
 424         }
 425
 426         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 427         if (err) {
 428                 char errbuf[1024];
 429                 regerror(err, &p->regexp, errbuf, 1024);
 430                 regfree(&p->regexp);
 431                 compile_regexp_failed(p, errbuf);
 432         }
 433 }
 434
 435 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 436 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 437 {
 438         struct grep_pat *p;
 439         struct grep_expr *x;
 440
 441         p = *list;
 442         if (!p)
 443                 return NULL;
 444         switch (p->token) {
 445         case GREP_PATTERN: /* atom */
 446         case GREP_PATTERN_HEAD:
 447         case GREP_PATTERN_BODY:
 448                 x = xcalloc(1, sizeof (struct grep_expr));
 449                 x->node = GREP_NODE_ATOM;
 450                 x->u.atom = p;
 451                 *list = p->next;
 452                 return x;
 453         case GREP_OPEN_PAREN:
 454                 *list = p->next;
 455                 x = compile_pattern_or(list);
 456                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 457                         die("unmatched parenthesis");
 458                 *list = (*list)->next;
 459                 return x;
 460         default:
 461                 return NULL;
 462         }
 463 }
 464
 465 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 466 {
 467         struct grep_pat *p;
 468         struct grep_expr *x;
 469
 470         p = *list;
 471         if (!p)
 472                 return NULL;
 473         switch (p->token) {
 474         case GREP_NOT:
 475                 if (!p->next)
 476                         die("--not not followed by pattern expression");
 477                 *list = p->next;
 478                 x = xcalloc(1, sizeof (struct grep_expr));
 479                 x->node = GREP_NODE_NOT;
 480                 x->u.unary = compile_pattern_not(list);
 481                 if (!x->u.unary)
 482                         die("--not followed by non pattern expression");
 483                 return x;
 484         default:
 485                 return compile_pattern_atom(list);
 486         }
 487 }
 488
 489 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 490 {
 491         struct grep_pat *p;
 492         struct grep_expr *x, *y, *z;
 493
 494         x = compile_pattern_not(list);
 495         p = *list;
 496         if (p && p->token == GREP_AND) {
 497                 if (!p->next)
 498                         die("--and not followed by pattern expression");
 499                 *list = p->next;
 500                 y = compile_pattern_and(list);
 501                 if (!y)
 502                         die("--and not followed by pattern expression");
 503                 z = xcalloc(1, sizeof (struct grep_expr));
 504                 z->node = GREP_NODE_AND;
 505                 z->u.binary.left = x;
 506                 z->u.binary.right = y;
 507                 return z;
 508         }
 509         return x;
 510 }
 511
 512 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 513 {
 514         struct grep_pat *p;
 515         struct grep_expr *x, *y, *z;
 516
 517         x = compile_pattern_and(list);
 518         p = *list;
 519         if (x && p && p->token != GREP_CLOSE_PAREN) {
 520                 y = compile_pattern_or(list);
 521                 if (!y)
 522                         die("not a pattern expression %s", p->pattern);
 523                 z = xcalloc(1, sizeof (struct grep_expr));
 524                 z->node = GREP_NODE_OR;
 525                 z->u.binary.left = x;
 526                 z->u.binary.right = y;
 527                 return z;
 528         }
 529         return x;
 530 }
 531
 532 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 533 {
 534         return compile_pattern_or(list);
 535 }
 536
 537 static void indent(int in)
 538 {
 539         while (in-- > 0)
 540                 fputc(' ', stderr);
 541 }
 542
 543 static void dump_grep_pat(struct grep_pat *p)
 544 {
 545         switch (p->token) {
 546         case GREP_AND: fprintf(stderr, "*and*"); break;
 547         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 548         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 549         case GREP_NOT: fprintf(stderr, "*not*"); break;
 550         case GREP_OR: fprintf(stderr, "*or*"); break;
 551
 552         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 553         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 554         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 555         }
 556
 557         switch (p->token) {
 558         default: break;
 559         case GREP_PATTERN_HEAD:
 560                 fprintf(stderr, "<head %d>", p->field); break;
 561         case GREP_PATTERN_BODY:
 562                 fprintf(stderr, "<body>"); break;
 563         }
 564         switch (p->token) {
 565         default: break;
 566         case GREP_PATTERN_HEAD:
 567         case GREP_PATTERN_BODY:
 568         case GREP_PATTERN:
 569                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 570                 break;
 571         }
 572         fputc('\n', stderr);
 573 }
 574
 575 static void dump_grep_expression_1(struct grep_expr *x, int in)
 576 {
 577         indent(in);
 578         switch (x->node) {
 579         case GREP_NODE_TRUE:
 580                 fprintf(stderr, "true\n");
 581                 break;
 582         case GREP_NODE_ATOM:
 583                 dump_grep_pat(x->u.atom);
 584                 break;
 585         case GREP_NODE_NOT:
 586                 fprintf(stderr, "(not\n");
 587                 dump_grep_expression_1(x->u.unary, in+1);
 588                 indent(in);
 589                 fprintf(stderr, ")\n");
 590                 break;
 591         case GREP_NODE_AND:
 592                 fprintf(stderr, "(and\n");
 593                 dump_grep_expression_1(x->u.binary.left, in+1);
 594                 dump_grep_expression_1(x->u.binary.right, in+1);
 595                 indent(in);
 596                 fprintf(stderr, ")\n");
 597                 break;
 598         case GREP_NODE_OR:
 599                 fprintf(stderr, "(or\n");
 600                 dump_grep_expression_1(x->u.binary.left, in+1);
 601                 dump_grep_expression_1(x->u.binary.right, in+1);
 602                 indent(in);
 603                 fprintf(stderr, ")\n");
 604                 break;
 605         }
 606 }
 607
 608 static void dump_grep_expression(struct grep_opt *opt)
 609 {
 610         struct grep_expr *x = opt->pattern_expression;
 611
 612         if (opt->all_match)
 613                 fprintf(stderr, "[all-match]\n");
 614         dump_grep_expression_1(x, 0);
 615         fflush(NULL);
 616 }
 617
 618 static struct grep_expr *grep_true_expr(void)
 619 {
 620         struct grep_expr *z = xcalloc(1, sizeof(*z));
 621         z->node = GREP_NODE_TRUE;
 622         return z;
 623 }
 624
 625 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 626 {
 627         struct grep_expr *z = xcalloc(1, sizeof(*z));
 628         z->node = GREP_NODE_OR;
 629         z->u.binary.left = left;
 630         z->u.binary.right = right;
 631         return z;
 632 }
 633
 634 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 635 {
 636         struct grep_pat *p;
 637         struct grep_expr *header_expr;
 638         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 639         enum grep_header_field fld;
 640
 641         if (!opt->header_list)
 642                 return NULL;
 643
 644         for (p = opt->header_list; p; p = p->next) {
 645                 if (p->token != GREP_PATTERN_HEAD)
 646                         die("bug: a non-header pattern in grep header list.");
 647                 if (p->field < GREP_HEADER_FIELD_MIN ||
 648                     GREP_HEADER_FIELD_MAX <= p->field)
 649                         die("bug: unknown header field %d", p->field);
 650                 compile_regexp(p, opt);
 651         }
 652
 653         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 654                 header_group[fld] = NULL;
 655
 656         for (p = opt->header_list; p; p = p->next) {
 657                 struct grep_expr *h;
 658                 struct grep_pat *pp = p;
 659
 660                 h = compile_pattern_atom(&pp);
 661                 if (!h || pp != p->next)
 662                         die("bug: malformed header expr");
 663                 if (!header_group[p->field]) {
 664                         header_group[p->field] = h;
 665                         continue;
 666                 }
 667                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 668         }
 669
 670         header_expr = NULL;
 671
 672         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 673                 if (!header_group[fld])
 674                         continue;
 675                 if (!header_expr)
 676                         header_expr = grep_true_expr();
 677                 header_expr = grep_or_expr(header_group[fld], header_expr);
 678         }
 679         return header_expr;
 680 }
 681
 682 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 683 {
 684         struct grep_expr *z = x;
 685
 686         while (x) {
 687                 assert(x->node == GREP_NODE_OR);
 688                 if (x->u.binary.right &&
 689                     x->u.binary.right->node == GREP_NODE_TRUE) {
 690                         x->u.binary.right = y;
 691                         break;
 692                 }
 693                 x = x->u.binary.right;
 694         }
 695         return z;
 696 }
 697
 698 static void compile_grep_patterns_real(struct grep_opt *opt)
 699 {
 700         struct grep_pat *p;
 701         struct grep_expr *header_expr = prep_header_patterns(opt);
 702
 703         for (p = opt->pattern_list; p; p = p->next) {
 704                 switch (p->token) {
 705                 case GREP_PATTERN: /* atom */
 706                 case GREP_PATTERN_HEAD:
 707                 case GREP_PATTERN_BODY:
 708                         compile_regexp(p, opt);
 709                         break;
 710                 default:
 711                         opt->extended = 1;
 712                         break;
 713                 }
 714         }
 715
 716         if (opt->all_match || header_expr)
 717                 opt->extended = 1;
 718         else if (!opt->extended && !opt->debug)
 719                 return;
 720
 721         p = opt->pattern_list;
 722         if (p)
 723                 opt->pattern_expression = compile_pattern_expr(&p);
 724         if (p)
 725                 die("incomplete pattern expression: %s", p->pattern);
 726
 727         if (!header_expr)
 728                 return;
 729
 730         if (!opt->pattern_expression)
 731                 opt->pattern_expression = header_expr;
 732         else if (opt->all_match)
 733                 opt->pattern_expression = grep_splice_or(header_expr,
 734                                                          opt->pattern_expression);
 735         else
 736                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 737                                                        header_expr);
 738         opt->all_match = 1;
 739 }
 740
 741 void compile_grep_patterns(struct grep_opt *opt)
 742 {
 743         compile_grep_patterns_real(opt);
 744         if (opt->debug)
 745                 dump_grep_expression(opt);
 746 }
 747
 748 static void free_pattern_expr(struct grep_expr *x)
 749 {
 750         switch (x->node) {
 751         case GREP_NODE_TRUE:
 752         case GREP_NODE_ATOM:
 753                 break;
 754         case GREP_NODE_NOT:
 755                 free_pattern_expr(x->u.unary);
 756                 break;
 757         case GREP_NODE_AND:
 758         case GREP_NODE_OR:
 759                 free_pattern_expr(x->u.binary.left);
 760                 free_pattern_expr(x->u.binary.right);
 761                 break;
 762         }
 763         free(x);
 764 }
 765
 766 void free_grep_patterns(struct grep_opt *opt)
 767 {
 768         struct grep_pat *p, *n;
 769
 770         for (p = opt->pattern_list; p; p = n) {
 771                 n = p->next;
 772                 switch (p->token) {
 773                 case GREP_PATTERN: /* atom */
 774                 case GREP_PATTERN_HEAD:
 775                 case GREP_PATTERN_BODY:
 776                         if (p->kws)
 777                                 kwsfree(p->kws);
 778                         else if (p->pcre_regexp)
 779                                 free_pcre_regexp(p);
 780                         else
 781                                 regfree(&p->regexp);
 782                         free(p->pattern);
 783                         break;
 784                 default:
 785                         break;
 786                 }
 787                 free(p);
 788         }
 789
 790         if (!opt->extended)
 791                 return;
 792         free_pattern_expr(opt->pattern_expression);
 793 }
 794
 795 static char *end_of_line(char *cp, unsigned long *left)
 796 {
 797         unsigned long l = *left;
 798         while (l && *cp != '\n') {
 799                 l--;
 800                 cp++;
 801         }
 802         *left = l;
 803         return cp;
 804 }
 805
 806 static int word_char(char ch)
 807 {
 808         return isalnum(ch) || ch == '_';
 809 }
 810
 811 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 812                          const char *color)
 813 {
 814         if (want_color(opt->color) && color && color[0]) {
 815                 opt->output(opt, color, strlen(color));
 816                 opt->output(opt, data, size);
 817                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 818         } else
 819                 opt->output(opt, data, size);
 820 }
 821
 822 static void output_sep(struct grep_opt *opt, char sign)
 823 {
 824         if (opt->null_following_name)
 825                 opt->output(opt, "\0", 1);
 826         else
 827                 output_color(opt, &sign, 1, opt->color_sep);
 828 }
 829
 830 static void show_name(struct grep_opt *opt, const char *name)
 831 {
 832         output_color(opt, name, strlen(name), opt->color_filename);
 833         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 834 }
 835
 836 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 837                     regmatch_t *match)
 838 {
 839         struct kwsmatch kwsm;
 840         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 841         if (offset == -1) {
 842                 match->rm_so = match->rm_eo = -1;
 843                 return REG_NOMATCH;
 844         } else {
 845                 match->rm_so = offset;
 846                 match->rm_eo = match->rm_so + kwsm.size[0];
 847                 return 0;
 848         }
 849 }
 850
 851 static int patmatch(struct grep_pat *p, char *line, char *eol,
 852                     regmatch_t *match, int eflags)
 853 {
 854         int hit;
 855
 856         if (p->fixed)
 857                 hit = !fixmatch(p, line, eol, match);
 858         else if (p->pcre_regexp)
 859                 hit = !pcrematch(p, line, eol, match, eflags);
 860         else
 861                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
 862                                    eflags);
 863
 864         return hit;
 865 }
 866
 867 static int strip_timestamp(char *bol, char **eol_p)
 868 {
 869         char *eol = *eol_p;
 870         int ch;
 871
 872         while (bol < --eol) {
 873                 if (*eol != '>')
 874                         continue;
 875                 *eol_p = ++eol;
 876                 ch = *eol;
 877                 *eol = '\0';
 878                 return ch;
 879         }
 880         return 0;
 881 }
 882
 883 static struct {
 884         const char *field;
 885         size_t len;
 886 } header_field[] = {
 887         { "author ", 7 },
 888         { "committer ", 10 },
 889         { "reflog ", 7 },
 890 };
 891
 892 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 893                              enum grep_context ctx,
 894                              regmatch_t *pmatch, int eflags)
 895 {
 896         int hit = 0;
 897         int saved_ch = 0;
 898         const char *start = bol;
 899
 900         if ((p->token != GREP_PATTERN) &&
 901             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 902                 return 0;
 903
 904         if (p->token == GREP_PATTERN_HEAD) {
 905                 const char *field;
 906                 size_t len;
 907                 assert(p->field < ARRAY_SIZE(header_field));
 908                 field = header_field[p->field].field;
 909                 len = header_field[p->field].len;
 910                 if (strncmp(bol, field, len))
 911                         return 0;
 912                 bol += len;
 913                 switch (p->field) {
 914                 case GREP_HEADER_AUTHOR:
 915                 case GREP_HEADER_COMMITTER:
 916                         saved_ch = strip_timestamp(bol, &eol);
 917                         break;
 918                 default:
 919                         break;
 920                 }
 921         }
 922
 923  again:
 924         hit = patmatch(p, bol, eol, pmatch, eflags);
 925
 926         if (hit && p->word_regexp) {
 927                 if ((pmatch[0].rm_so < 0) ||
 928                     (eol - bol) < pmatch[0].rm_so ||
 929                     (pmatch[0].rm_eo < 0) ||
 930                     (eol - bol) < pmatch[0].rm_eo)
 931                         die("regexp returned nonsense");
 932
 933                 /* Match beginning must be either beginning of the
 934                  * line, or at word boundary (i.e. the last char must
 935                  * not be a word char).  Similarly, match end must be
 936                  * either end of the line, or at word boundary
 937                  * (i.e. the next char must not be a word char).
 938                  */
 939                 if ( ((pmatch[0].rm_so == 0) ||
 940                       !word_char(bol[pmatch[0].rm_so-1])) &&
 941                      ((pmatch[0].rm_eo == (eol-bol)) ||
 942                       !word_char(bol[pmatch[0].rm_eo])) )
 943                         ;
 944                 else
 945                         hit = 0;
 946
 947                 /* Words consist of at least one character. */
 948                 if (pmatch->rm_so == pmatch->rm_eo)
 949                         hit = 0;
 950
 951                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 952                         /* There could be more than one match on the
 953                          * line, and the first match might not be
 954                          * strict word match.  But later ones could be!
 955                          * Forward to the next possible start, i.e. the
 956                          * next position following a non-word char.
 957                          */
 958                         bol = pmatch[0].rm_so + bol + 1;
 959                         while (word_char(bol[-1]) && bol < eol)
 960                                 bol++;
 961                         eflags |= REG_NOTBOL;
 962                         if (bol < eol)
 963                                 goto again;
 964                 }
 965         }
 966         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 967                 *eol = saved_ch;
 968         if (hit) {
 969                 pmatch[0].rm_so += bol - start;
 970                 pmatch[0].rm_eo += bol - start;
 971         }
 972         return hit;
 973 }
 974
 975 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 976                            enum grep_context ctx, int collect_hits)
 977 {
 978         int h = 0;
 979         regmatch_t match;
 980
 981         if (!x)
 982                 die("Not a valid grep expression");
 983         switch (x->node) {
 984         case GREP_NODE_TRUE:
 985                 h = 1;
 986                 break;
 987         case GREP_NODE_ATOM:
 988                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 989                 break;
 990         case GREP_NODE_NOT:
 991                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 992                 break;
 993         case GREP_NODE_AND:
 994                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 995                         return 0;
 996                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 997                 break;
 998         case GREP_NODE_OR:
 999                 if (!collect_hits)
1000                         return (match_expr_eval(x->u.binary.left,
1001                                                 bol, eol, ctx, 0) ||
1002                                 match_expr_eval(x->u.binary.right,
1003                                                 bol, eol, ctx, 0));
1004                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1005                 x->u.binary.left->hit |= h;
1006                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1007                 break;
1008         default:
1009                 die("Unexpected node type (internal error) %d", x->node);
1010         }
1011         if (collect_hits)
1012                 x->hit |= h;
1013         return h;
1014 }
1015
1016 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1017                       enum grep_context ctx, int collect_hits)
1018 {
1019         struct grep_expr *x = opt->pattern_expression;
1020         return match_expr_eval(x, bol, eol, ctx, collect_hits);
1021 }
1022
1023 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1024                       enum grep_context ctx, int collect_hits)
1025 {
1026         struct grep_pat *p;
1027         regmatch_t match;
1028
1029         if (opt->extended)
1030                 return match_expr(opt, bol, eol, ctx, collect_hits);
1031
1032         /* we do not call with collect_hits without being extended */
1033         for (p = opt->pattern_list; p; p = p->next) {
1034                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1035                         return 1;
1036         }
1037         return 0;
1038 }
1039
1040 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1041                               enum grep_context ctx,
1042                               regmatch_t *pmatch, int eflags)
1043 {
1044         regmatch_t match;
1045
1046         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1047                 return 0;
1048         if (match.rm_so < 0 || match.rm_eo < 0)
1049                 return 0;
1050         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1051                 if (match.rm_so > pmatch->rm_so)
1052                         return 1;
1053                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1054                         return 1;
1055         }
1056         pmatch->rm_so = match.rm_so;
1057         pmatch->rm_eo = match.rm_eo;
1058         return 1;
1059 }
1060
1061 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1062                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1063 {
1064         struct grep_pat *p;
1065         int hit = 0;
1066
1067         pmatch->rm_so = pmatch->rm_eo = -1;
1068         if (bol < eol) {
1069                 for (p = opt->pattern_list; p; p = p->next) {
1070                         switch (p->token) {
1071                         case GREP_PATTERN: /* atom */
1072                         case GREP_PATTERN_HEAD:
1073                         case GREP_PATTERN_BODY:
1074                                 hit |= match_next_pattern(p, bol, eol, ctx,
1075                                                           pmatch, eflags);
1076                                 break;
1077                         default:
1078                                 break;
1079                         }
1080                 }
1081         }
1082         return hit;
1083 }
1084
1085 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1086                       const char *name, unsigned lno, char sign)
1087 {
1088         int rest = eol - bol;
1089         const char *match_color, *line_color = NULL;
1090
1091         if (opt->file_break && opt->last_shown == 0) {
1092                 if (opt->show_hunk_mark)
1093                         opt->output(opt, "\n", 1);
1094         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1095                 if (opt->last_shown == 0) {
1096                         if (opt->show_hunk_mark) {
1097                                 output_color(opt, "--", 2, opt->color_sep);
1098                                 opt->output(opt, "\n", 1);
1099                         }
1100                 } else if (lno > opt->last_shown + 1) {
1101                         output_color(opt, "--", 2, opt->color_sep);
1102                         opt->output(opt, "\n", 1);
1103                 }
1104         }
1105         if (opt->heading && opt->last_shown == 0) {
1106                 output_color(opt, name, strlen(name), opt->color_filename);
1107                 opt->output(opt, "\n", 1);
1108         }
1109         opt->last_shown = lno;
1110
1111         if (!opt->heading && opt->pathname) {
1112                 output_color(opt, name, strlen(name), opt->color_filename);
1113                 output_sep(opt, sign);
1114         }
1115         if (opt->linenum) {
1116                 char buf[32];
1117                 snprintf(buf, sizeof(buf), "%d", lno);
1118                 output_color(opt, buf, strlen(buf), opt->color_lineno);
1119                 output_sep(opt, sign);
1120         }
1121         if (opt->color) {
1122                 regmatch_t match;
1123                 enum grep_context ctx = GREP_CONTEXT_BODY;
1124                 int ch = *eol;
1125                 int eflags = 0;
1126
1127                 if (sign == ':')
1128                         match_color = opt->color_match_selected;
1129                 else
1130                         match_color = opt->color_match_context;
1131                 if (sign == ':')
1132                         line_color = opt->color_selected;
1133                 else if (sign == '-')
1134                         line_color = opt->color_context;
1135                 else if (sign == '=')
1136                         line_color = opt->color_function;
1137                 *eol = '\0';
1138                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1139                         if (match.rm_so == match.rm_eo)
1140                                 break;
1141
1142                         output_color(opt, bol, match.rm_so, line_color);
1143                         output_color(opt, bol + match.rm_so,
1144                                      match.rm_eo - match.rm_so, match_color);
1145                         bol += match.rm_eo;
1146                         rest -= match.rm_eo;
1147                         eflags = REG_NOTBOL;
1148                 }
1149                 *eol = ch;
1150         }
1151         output_color(opt, bol, rest, line_color);
1152         opt->output(opt, "\n", 1);
1153 }
1154
1155 #ifndef NO_PTHREADS
1156 int grep_use_locks;
1157
1158 /*
1159  * This lock protects access to the gitattributes machinery, which is
1160  * not thread-safe.
1161  */
1162 pthread_mutex_t grep_attr_mutex;
1163
1164 static inline void grep_attr_lock(void)
1165 {
1166         if (grep_use_locks)
1167                 pthread_mutex_lock(&grep_attr_mutex);
1168 }
1169
1170 static inline void grep_attr_unlock(void)
1171 {
1172         if (grep_use_locks)
1173                 pthread_mutex_unlock(&grep_attr_mutex);
1174 }
1175
1176 /*
1177  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1178  */
1179 pthread_mutex_t grep_read_mutex;
1180
1181 #else
1182 #define grep_attr_lock()
1183 #define grep_attr_unlock()
1184 #endif
1185
1186 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1187 {
1188         xdemitconf_t *xecfg = opt->priv;
1189         if (xecfg && !xecfg->find_func) {
1190                 grep_source_load_driver(gs);
1191                 if (gs->driver->funcname.pattern) {
1192                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1193                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1194                 } else {
1195                         xecfg = opt->priv = NULL;
1196                 }
1197         }
1198
1199         if (xecfg) {
1200                 char buf[1];
1201                 return xecfg->find_func(bol, eol - bol, buf, 1,
1202                                         xecfg->find_func_priv) >= 0;
1203         }
1204
1205         if (bol == eol)
1206                 return 0;
1207         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1208                 return 1;
1209         return 0;
1210 }
1211
1212 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1213                                char *bol, unsigned lno)
1214 {
1215         while (bol > gs->buf) {
1216                 char *eol = --bol;
1217
1218                 while (bol > gs->buf && bol[-1] != '\n')
1219                         bol--;
1220                 lno--;
1221
1222                 if (lno <= opt->last_shown)
1223                         break;
1224
1225                 if (match_funcname(opt, gs, bol, eol)) {
1226                         show_line(opt, bol, eol, gs->name, lno, '=');
1227                         break;
1228                 }
1229         }
1230 }
1231
1232 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1233                              char *bol, char *end, unsigned lno)
1234 {
1235         unsigned cur = lno, from = 1, funcname_lno = 0;
1236         int funcname_needed = !!opt->funcname;
1237
1238         if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1239                 funcname_needed = 2;
1240
1241         if (opt->pre_context < lno)
1242                 from = lno - opt->pre_context;
1243         if (from <= opt->last_shown)
1244                 from = opt->last_shown + 1;
1245
1246         /* Rewind. */
1247         while (bol > gs->buf &&
1248                cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1249                 char *eol = --bol;
1250
1251                 while (bol > gs->buf && bol[-1] != '\n')
1252                         bol--;
1253                 cur--;
1254                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1255                         funcname_lno = cur;
1256                         funcname_needed = 0;
1257                 }
1258         }
1259
1260         /* We need to look even further back to find a function signature. */
1261         if (opt->funcname && funcname_needed)
1262                 show_funcname_line(opt, gs, bol, cur);
1263
1264         /* Back forward. */
1265         while (cur < lno) {
1266                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1267
1268                 while (*eol != '\n')
1269                         eol++;
1270                 show_line(opt, bol, eol, gs->name, cur, sign);
1271                 bol = eol + 1;
1272                 cur++;
1273         }
1274 }
1275
1276 static int should_lookahead(struct grep_opt *opt)
1277 {
1278         struct grep_pat *p;
1279
1280         if (opt->extended)
1281                 return 0; /* punt for too complex stuff */
1282         if (opt->invert)
1283                 return 0;
1284         for (p = opt->pattern_list; p; p = p->next) {
1285                 if (p->token != GREP_PATTERN)
1286                         return 0; /* punt for "header only" and stuff */
1287         }
1288         return 1;
1289 }
1290
1291 static int look_ahead(struct grep_opt *opt,
1292                       unsigned long *left_p,
1293                       unsigned *lno_p,
1294                       char **bol_p)
1295 {
1296         unsigned lno = *lno_p;
1297         char *bol = *bol_p;
1298         struct grep_pat *p;
1299         char *sp, *last_bol;
1300         regoff_t earliest = -1;
1301
1302         for (p = opt->pattern_list; p; p = p->next) {
1303                 int hit;
1304                 regmatch_t m;
1305
1306                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1307                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1308                         continue;
1309                 if (earliest < 0 || m.rm_so < earliest)
1310                         earliest = m.rm_so;
1311         }
1312
1313         if (earliest < 0) {
1314                 *bol_p = bol + *left_p;
1315                 *left_p = 0;
1316                 return 1;
1317         }
1318         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1319                 ; /* find the beginning of the line */
1320         last_bol = sp;
1321
1322         for (sp = bol; sp < last_bol; sp++) {
1323                 if (*sp == '\n')
1324                         lno++;
1325         }
1326         *left_p -= last_bol - bol;
1327         *bol_p = last_bol;
1328         *lno_p = lno;
1329         return 0;
1330 }
1331
1332 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1333 {
1334         fwrite(buf, size, 1, stdout);
1335 }
1336
1337 static int fill_textconv_grep(struct userdiff_driver *driver,
1338                               struct grep_source *gs)
1339 {
1340         struct diff_filespec *df;
1341         char *buf;
1342         size_t size;
1343
1344         if (!driver || !driver->textconv)
1345                 return grep_source_load(gs);
1346
1347         /*
1348          * The textconv interface is intimately tied to diff_filespecs, so we
1349          * have to pretend to be one. If we could unify the grep_source
1350          * and diff_filespec structs, this mess could just go away.
1351          */
1352         df = alloc_filespec(gs->path);
1353         switch (gs->type) {
1354         case GREP_SOURCE_SHA1:
1355                 fill_filespec(df, gs->identifier, 1, 0100644);
1356                 break;
1357         case GREP_SOURCE_FILE:
1358                 fill_filespec(df, null_sha1, 0, 0100644);
1359                 break;
1360         default:
1361                 die("BUG: attempt to textconv something without a path?");
1362         }
1363
1364         /*
1365          * fill_textconv is not remotely thread-safe; it may load objects
1366          * behind the scenes, and it modifies the global diff tempfile
1367          * structure.
1368          */
1369         grep_read_lock();
1370         size = fill_textconv(driver, df, &buf);
1371         grep_read_unlock();
1372         free_filespec(df);
1373
1374         /*
1375          * The normal fill_textconv usage by the diff machinery would just keep
1376          * the textconv'd buf separate from the diff_filespec. But much of the
1377          * grep code passes around a grep_source and assumes that its "buf"
1378          * pointer is the beginning of the thing we are searching. So let's
1379          * install our textconv'd version into the grep_source, taking care not
1380          * to leak any existing buffer.
1381          */
1382         grep_source_clear_data(gs);
1383         gs->buf = buf;
1384         gs->size = size;
1385
1386         return 0;
1387 }
1388
1389 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1390 {
1391         char *bol;
1392         unsigned long left;
1393         unsigned lno = 1;
1394         unsigned last_hit = 0;
1395         int binary_match_only = 0;
1396         unsigned count = 0;
1397         int try_lookahead = 0;
1398         int show_function = 0;
1399         struct userdiff_driver *textconv = NULL;
1400         enum grep_context ctx = GREP_CONTEXT_HEAD;
1401         xdemitconf_t xecfg;
1402
1403         if (!opt->output)
1404                 opt->output = std_output;
1405
1406         if (opt->pre_context || opt->post_context || opt->file_break ||
1407             opt->funcbody) {
1408                 /* Show hunk marks, except for the first file. */
1409                 if (opt->last_shown)
1410                         opt->show_hunk_mark = 1;
1411                 /*
1412                  * If we're using threads then we can't easily identify
1413                  * the first file.  Always put hunk marks in that case
1414                  * and skip the very first one later in work_done().
1415                  */
1416                 if (opt->output != std_output)
1417                         opt->show_hunk_mark = 1;
1418         }
1419         opt->last_shown = 0;
1420
1421         if (opt->allow_textconv) {
1422                 grep_source_load_driver(gs);
1423                 /*
1424                  * We might set up the shared textconv cache data here, which
1425                  * is not thread-safe.
1426                  */
1427                 grep_attr_lock();
1428                 textconv = userdiff_get_textconv(gs->driver);
1429                 grep_attr_unlock();
1430         }
1431
1432         /*
1433          * We know the result of a textconv is text, so we only have to care
1434          * about binary handling if we are not using it.
1435          */
1436         if (!textconv) {
1437                 switch (opt->binary) {
1438                 case GREP_BINARY_DEFAULT:
1439                         if (grep_source_is_binary(gs))
1440                                 binary_match_only = 1;
1441                         break;
1442                 case GREP_BINARY_NOMATCH:
1443                         if (grep_source_is_binary(gs))
1444                                 return 0; /* Assume unmatch */
1445                         break;
1446                 case GREP_BINARY_TEXT:
1447                         break;
1448                 default:
1449                         die("bug: unknown binary handling mode");
1450                 }
1451         }
1452
1453         memset(&xecfg, 0, sizeof(xecfg));
1454         opt->priv = &xecfg;
1455
1456         try_lookahead = should_lookahead(opt);
1457
1458         if (fill_textconv_grep(textconv, gs) < 0)
1459                 return 0;
1460
1461         bol = gs->buf;
1462         left = gs->size;
1463         while (left) {
1464                 char *eol, ch;
1465                 int hit;
1466
1467                 /*
1468                  * look_ahead() skips quickly to the line that possibly
1469                  * has the next hit; don't call it if we need to do
1470                  * something more than just skipping the current line
1471                  * in response to an unmatch for the current line.  E.g.
1472                  * inside a post-context window, we will show the current
1473                  * line as a context around the previous hit when it
1474                  * doesn't hit.
1475                  */
1476                 if (try_lookahead
1477                     && !(last_hit
1478                          && (show_function ||
1479                              lno <= last_hit + opt->post_context))
1480                     && look_ahead(opt, &left, &lno, &bol))
1481                         break;
1482                 eol = end_of_line(bol, &left);
1483                 ch = *eol;
1484                 *eol = 0;
1485
1486                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1487                         ctx = GREP_CONTEXT_BODY;
1488
1489                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1490                 *eol = ch;
1491
1492                 if (collect_hits)
1493                         goto next_line;
1494
1495                 /* "grep -v -e foo -e bla" should list lines
1496                  * that do not have either, so inversion should
1497                  * be done outside.
1498                  */
1499                 if (opt->invert)
1500                         hit = !hit;
1501                 if (opt->unmatch_name_only) {
1502                         if (hit)
1503                                 return 0;
1504                         goto next_line;
1505                 }
1506                 if (hit) {
1507                         count++;
1508                         if (opt->status_only)
1509                                 return 1;
1510                         if (opt->name_only) {
1511                                 show_name(opt, gs->name);
1512                                 return 1;
1513                         }
1514                         if (opt->count)
1515                                 goto next_line;
1516                         if (binary_match_only) {
1517                                 opt->output(opt, "Binary file ", 12);
1518                                 output_color(opt, gs->name, strlen(gs->name),
1519                                              opt->color_filename);
1520                                 opt->output(opt, " matches\n", 9);
1521                                 return 1;
1522                         }
1523                         /* Hit at this line.  If we haven't shown the
1524                          * pre-context lines, we would need to show them.
1525                          */
1526                         if (opt->pre_context || opt->funcbody)
1527                                 show_pre_context(opt, gs, bol, eol, lno);
1528                         else if (opt->funcname)
1529                                 show_funcname_line(opt, gs, bol, lno);
1530                         show_line(opt, bol, eol, gs->name, lno, ':');
1531                         last_hit = lno;
1532                         if (opt->funcbody)
1533                                 show_function = 1;
1534                         goto next_line;
1535                 }
1536                 if (show_function && match_funcname(opt, gs, bol, eol))
1537                         show_function = 0;
1538                 if (show_function ||
1539                     (last_hit && lno <= last_hit + opt->post_context)) {
1540                         /* If the last hit is within the post context,
1541                          * we need to show this line.
1542                          */
1543                         show_line(opt, bol, eol, gs->name, lno, '-');
1544                 }
1545
1546         next_line:
1547                 bol = eol + 1;
1548                 if (!left)
1549                         break;
1550                 left--;
1551                 lno++;
1552         }
1553
1554         if (collect_hits)
1555                 return 0;
1556
1557         if (opt->status_only)
1558                 return 0;
1559         if (opt->unmatch_name_only) {
1560                 /* We did not see any hit, so we want to show this */
1561                 show_name(opt, gs->name);
1562                 return 1;
1563         }
1564
1565         xdiff_clear_find_func(&xecfg);
1566         opt->priv = NULL;
1567
1568         /* NEEDSWORK:
1569          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1570          * which feels mostly useless but sometimes useful.  Maybe
1571          * make it another option?  For now suppress them.
1572          */
1573         if (opt->count && count) {
1574                 char buf[32];
1575                 if (opt->pathname) {
1576                         output_color(opt, gs->name, strlen(gs->name),
1577                                      opt->color_filename);
1578                         output_sep(opt, ':');
1579                 }
1580                 snprintf(buf, sizeof(buf), "%u\n", count);
1581                 opt->output(opt, buf, strlen(buf));
1582                 return 1;
1583         }
1584         return !!last_hit;
1585 }
1586
1587 static void clr_hit_marker(struct grep_expr *x)
1588 {
1589         /* All-hit markers are meaningful only at the very top level
1590          * OR node.
1591          */
1592         while (1) {
1593                 x->hit = 0;
1594                 if (x->node != GREP_NODE_OR)
1595                         return;
1596                 x->u.binary.left->hit = 0;
1597                 x = x->u.binary.right;
1598         }
1599 }
1600
1601 static int chk_hit_marker(struct grep_expr *x)
1602 {
1603         /* Top level nodes have hit markers.  See if they all are hits */
1604         while (1) {
1605                 if (x->node != GREP_NODE_OR)
1606                         return x->hit;
1607                 if (!x->u.binary.left->hit)
1608                         return 0;
1609                 x = x->u.binary.right;
1610         }
1611 }
1612
1613 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1614 {
1615         /*
1616          * we do not have to do the two-pass grep when we do not check
1617          * buffer-wide "all-match".
1618          */
1619         if (!opt->all_match)
1620                 return grep_source_1(opt, gs, 0);
1621
1622         /* Otherwise the toplevel "or" terms hit a bit differently.
1623          * We first clear hit markers from them.
1624          */
1625         clr_hit_marker(opt->pattern_expression);
1626         grep_source_1(opt, gs, 1);
1627
1628         if (!chk_hit_marker(opt->pattern_expression))
1629                 return 0;
1630
1631         return grep_source_1(opt, gs, 0);
1632 }
1633
1634 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1635 {
1636         struct grep_source gs;
1637         int r;
1638
1639         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1640         gs.buf = buf;
1641         gs.size = size;
1642
1643         r = grep_source(opt, &gs);
1644
1645         grep_source_clear(&gs);
1646         return r;
1647 }
1648
1649 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1650                       const char *name, const char *path,
1651                       const void *identifier)
1652 {
1653         gs->type = type;
1654         gs->name = xstrdup_or_null(name);
1655         gs->path = xstrdup_or_null(path);
1656         gs->buf = NULL;
1657         gs->size = 0;
1658         gs->driver = NULL;
1659
1660         switch (type) {
1661         case GREP_SOURCE_FILE:
1662                 gs->identifier = xstrdup(identifier);
1663                 break;
1664         case GREP_SOURCE_SHA1:
1665                 gs->identifier = xmalloc(20);
1666                 hashcpy(gs->identifier, identifier);
1667                 break;
1668         case GREP_SOURCE_BUF:
1669                 gs->identifier = NULL;
1670         }
1671 }
1672
1673 void grep_source_clear(struct grep_source *gs)
1674 {
1675         free(gs->name);
1676         gs->name = NULL;
1677         free(gs->path);
1678         gs->path = NULL;
1679         free(gs->identifier);
1680         gs->identifier = NULL;
1681         grep_source_clear_data(gs);
1682 }
1683
1684 void grep_source_clear_data(struct grep_source *gs)
1685 {
1686         switch (gs->type) {
1687         case GREP_SOURCE_FILE:
1688         case GREP_SOURCE_SHA1:
1689                 free(gs->buf);
1690                 gs->buf = NULL;
1691                 gs->size = 0;
1692                 break;
1693         case GREP_SOURCE_BUF:
1694                 /* leave user-provided buf intact */
1695                 break;
1696         }
1697 }
1698
1699 static int grep_source_load_sha1(struct grep_source *gs)
1700 {
1701         enum object_type type;
1702
1703         grep_read_lock();
1704         gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1705         grep_read_unlock();
1706
1707         if (!gs->buf)
1708                 return error(_("'%s': unable to read %s"),
1709                              gs->name,
1710                              sha1_to_hex(gs->identifier));
1711         return 0;
1712 }
1713
1714 static int grep_source_load_file(struct grep_source *gs)
1715 {
1716         const char *filename = gs->identifier;
1717         struct stat st;
1718         char *data;
1719         size_t size;
1720         int i;
1721
1722         if (lstat(filename, &st) < 0) {
1723         err_ret:
1724                 if (errno != ENOENT)
1725                         error(_("'%s': %s"), filename, strerror(errno));
1726                 return -1;
1727         }
1728         if (!S_ISREG(st.st_mode))
1729                 return -1;
1730         size = xsize_t(st.st_size);
1731         i = open(filename, O_RDONLY);
1732         if (i < 0)
1733                 goto err_ret;
1734         data = xmallocz(size);
1735         if (st.st_size != read_in_full(i, data, size)) {
1736                 error(_("'%s': short read %s"), filename, strerror(errno));
1737                 close(i);
1738                 free(data);
1739                 return -1;
1740         }
1741         close(i);
1742
1743         gs->buf = data;
1744         gs->size = size;
1745         return 0;
1746 }
1747
1748 static int grep_source_load(struct grep_source *gs)
1749 {
1750         if (gs->buf)
1751                 return 0;
1752
1753         switch (gs->type) {
1754         case GREP_SOURCE_FILE:
1755                 return grep_source_load_file(gs);
1756         case GREP_SOURCE_SHA1:
1757                 return grep_source_load_sha1(gs);
1758         case GREP_SOURCE_BUF:
1759                 return gs->buf ? 0 : -1;
1760         }
1761         die("BUG: invalid grep_source type");
1762 }
1763
1764 void grep_source_load_driver(struct grep_source *gs)
1765 {
1766         if (gs->driver)
1767                 return;
1768
1769         grep_attr_lock();
1770         if (gs->path)
1771                 gs->driver = userdiff_find_by_path(gs->path);
1772         if (!gs->driver)
1773                 gs->driver = userdiff_find_by_name("default");
1774         grep_attr_unlock();
1775 }
1776
1777 static int grep_source_is_binary(struct grep_source *gs)
1778 {
1779         grep_source_load_driver(gs);
1780         if (gs->driver->binary != -1)
1781                 return gs->driver->binary;
1782
1783         if (!grep_source_load(gs))
1784                 return buffer_is_binary(gs->buf, gs->size);
1785
1786         return 0;
1787 }