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