grep.c

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "grep.h"
   4 #include "object-store.h"
   5 #include "userdiff.h"
   6 #include "xdiff-interface.h"
   7 #include "diff.h"
   8 #include "diffcore.h"
   9 #include "commit.h"
  10 #include "quote.h"
  11 #include "help.h"
  12
  13 static int grep_source_load(struct grep_source *gs);
  14 static int grep_source_is_binary(struct grep_source *gs,
  15                                  struct index_state *istate);
  16
  17 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
  18 {
  19         fwrite(buf, size, 1, stdout);
  20 }
  21
  22 static struct grep_opt grep_defaults = {
  23         .relative = 1,
  24         .pathname = 1,
  25         .max_depth = -1,
  26         .pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED,
  27         .colors = {
  28                 [GREP_COLOR_CONTEXT] = "",
  29                 [GREP_COLOR_FILENAME] = "",
  30                 [GREP_COLOR_FUNCTION] = "",
  31                 [GREP_COLOR_LINENO] = "",
  32                 [GREP_COLOR_COLUMNNO] = "",
  33                 [GREP_COLOR_MATCH_CONTEXT] = GIT_COLOR_BOLD_RED,
  34                 [GREP_COLOR_MATCH_SELECTED] = GIT_COLOR_BOLD_RED,
  35                 [GREP_COLOR_SELECTED] = "",
  36                 [GREP_COLOR_SEP] = GIT_COLOR_CYAN,
  37         },
  38         .only_matching = 0,
  39         .color = -1,
  40         .output = std_output,
  41 };
  42
  43 #ifdef USE_LIBPCRE2
  44 static pcre2_general_context *pcre2_global_context;
  45
  46 static void *pcre2_malloc(PCRE2_SIZE size, MAYBE_UNUSED void *memory_data)
  47 {
  48         return malloc(size);
  49 }
  50
  51 static void pcre2_free(void *pointer, MAYBE_UNUSED void *memory_data)
  52 {
  53         free(pointer);
  54 }
  55 #endif
  56
  57 static const char *color_grep_slots[] = {
  58         [GREP_COLOR_CONTEXT]        = "context",
  59         [GREP_COLOR_FILENAME]       = "filename",
  60         [GREP_COLOR_FUNCTION]       = "function",
  61         [GREP_COLOR_LINENO]         = "lineNumber",
  62         [GREP_COLOR_COLUMNNO]       = "column",
  63         [GREP_COLOR_MATCH_CONTEXT]  = "matchContext",
  64         [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
  65         [GREP_COLOR_SELECTED]       = "selected",
  66         [GREP_COLOR_SEP]            = "separator",
  67 };
  68
  69 static int parse_pattern_type_arg(const char *opt, const char *arg)
  70 {
  71         if (!strcmp(arg, "default"))
  72                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  73         else if (!strcmp(arg, "basic"))
  74                 return GREP_PATTERN_TYPE_BRE;
  75         else if (!strcmp(arg, "extended"))
  76                 return GREP_PATTERN_TYPE_ERE;
  77         else if (!strcmp(arg, "fixed"))
  78                 return GREP_PATTERN_TYPE_FIXED;
  79         else if (!strcmp(arg, "perl"))
  80                 return GREP_PATTERN_TYPE_PCRE;
  81         die("bad %s argument: %s", opt, arg);
  82 }
  83
  84 define_list_config_array_extra(color_grep_slots, {"match"});
  85
  86 /*
  87  * Read the configuration file once and store it in
  88  * the grep_defaults template.
  89  */
  90 int grep_config(const char *var, const char *value, void *cb)
  91 {
  92         struct grep_opt *opt = &grep_defaults;
  93         const char *slot;
  94
  95         if (userdiff_config(var, value) < 0)
  96                 return -1;
  97
  98         /*
  99          * The instance of grep_opt that we set up here is copied by
 100          * grep_init() to be used by each individual invocation.
 101          * When populating a new field of this structure here, be
 102          * sure to think about ownership -- e.g., you might need to
 103          * override the shallow copy in grep_init() with a deep copy.
 104          */
 105
 106         if (!strcmp(var, "grep.extendedregexp")) {
 107                 opt->extended_regexp_option = git_config_bool(var, value);
 108                 return 0;
 109         }
 110
 111         if (!strcmp(var, "grep.patterntype")) {
 112                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
 113                 return 0;
 114         }
 115
 116         if (!strcmp(var, "grep.linenumber")) {
 117                 opt->linenum = git_config_bool(var, value);
 118                 return 0;
 119         }
 120         if (!strcmp(var, "grep.column")) {
 121                 opt->columnnum = git_config_bool(var, value);
 122                 return 0;
 123         }
 124
 125         if (!strcmp(var, "grep.fullname")) {
 126                 opt->relative = !git_config_bool(var, value);
 127                 return 0;
 128         }
 129
 130         if (!strcmp(var, "color.grep"))
 131                 opt->color = git_config_colorbool(var, value);
 132         if (!strcmp(var, "color.grep.match")) {
 133                 if (grep_config("color.grep.matchcontext", value, cb) < 0)
 134                         return -1;
 135                 if (grep_config("color.grep.matchselected", value, cb) < 0)
 136                         return -1;
 137         } else if (skip_prefix(var, "color.grep.", &slot)) {
 138                 int i = LOOKUP_CONFIG(color_grep_slots, slot);
 139                 char *color;
 140
 141                 if (i < 0)
 142                         return -1;
 143                 color = opt->colors[i];
 144                 if (!value)
 145                         return config_error_nonbool(var);
 146                 return color_parse(value, color);
 147         }
 148         return 0;
 149 }
 150
 151 /*
 152  * Initialize one instance of grep_opt and copy the
 153  * default values from the template we read the configuration
 154  * information in an earlier call to git_config(grep_config).
 155  *
 156  * If using PCRE, make sure that the library is configured
 157  * to use the same allocator as Git (e.g. nedmalloc on Windows).
 158  *
 159  * Any allocated memory needs to be released in grep_destroy().
 160  */
 161 void grep_init(struct grep_opt *opt, struct repository *repo, const char *prefix)
 162 {
 163 #if defined(USE_LIBPCRE2)
 164         if (!pcre2_global_context)
 165                 pcre2_global_context = pcre2_general_context_create(
 166                                         pcre2_malloc, pcre2_free, NULL);
 167 #endif
 168
 169 #ifdef USE_LIBPCRE1
 170         pcre_malloc = malloc;
 171         pcre_free = free;
 172 #endif
 173
 174         *opt = grep_defaults;
 175
 176         opt->repo = repo;
 177         opt->prefix = prefix;
 178         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 179         opt->pattern_tail = &opt->pattern_list;
 180         opt->header_tail = &opt->header_list;
 181 }
 182
 183 void grep_destroy(void)
 184 {
 185 #ifdef USE_LIBPCRE2
 186         pcre2_general_context_free(pcre2_global_context);
 187 #endif
 188 }
 189
 190 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 191 {
 192         /*
 193          * When committing to the pattern type by setting the relevant
 194          * fields in grep_opt it's generally not necessary to zero out
 195          * the fields we're not choosing, since they won't have been
 196          * set by anything. The extended_regexp_option field is the
 197          * only exception to this.
 198          *
 199          * This is because in the process of parsing grep.patternType
 200          * & grep.extendedRegexp we set opt->pattern_type_option and
 201          * opt->extended_regexp_option, respectively. We then
 202          * internally use opt->extended_regexp_option to see if we're
 203          * compiling an ERE. It must be unset if that's not actually
 204          * the case.
 205          */
 206         if (pattern_type != GREP_PATTERN_TYPE_ERE &&
 207             opt->extended_regexp_option)
 208                 opt->extended_regexp_option = 0;
 209
 210         switch (pattern_type) {
 211         case GREP_PATTERN_TYPE_UNSPECIFIED:
 212                 /* fall through */
 213
 214         case GREP_PATTERN_TYPE_BRE:
 215                 break;
 216
 217         case GREP_PATTERN_TYPE_ERE:
 218                 opt->extended_regexp_option = 1;
 219                 break;
 220
 221         case GREP_PATTERN_TYPE_FIXED:
 222                 opt->fixed = 1;
 223                 break;
 224
 225         case GREP_PATTERN_TYPE_PCRE:
 226 #ifdef USE_LIBPCRE2
 227                 opt->pcre2 = 1;
 228 #else
 229                 /*
 230                  * It's important that pcre1 always be assigned to
 231                  * even when there's no USE_LIBPCRE* defined. We still
 232                  * call the PCRE stub function, it just dies with
 233                  * "cannot use Perl-compatible regexes[...]".
 234                  */
 235                 opt->pcre1 = 1;
 236 #endif
 237                 break;
 238         }
 239 }
 240
 241 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 242 {
 243         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 244                 grep_set_pattern_type_option(pattern_type, opt);
 245         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 246                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 247         else if (opt->extended_regexp_option)
 248                 /*
 249                  * This branch *must* happen after setting from the
 250                  * opt->pattern_type_option above, we don't want
 251                  * grep.extendedRegexp to override grep.patternType!
 252                  */
 253                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 254 }
 255
 256 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 257                                         const char *origin, int no,
 258                                         enum grep_pat_token t,
 259                                         enum grep_header_field field)
 260 {
 261         struct grep_pat *p = xcalloc(1, sizeof(*p));
 262         p->pattern = xmemdupz(pat, patlen);
 263         p->patternlen = patlen;
 264         p->origin = origin;
 265         p->no = no;
 266         p->token = t;
 267         p->field = field;
 268         return p;
 269 }
 270
 271 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 272 {
 273         **tail = p;
 274         *tail = &p->next;
 275         p->next = NULL;
 276
 277         switch (p->token) {
 278         case GREP_PATTERN: /* atom */
 279         case GREP_PATTERN_HEAD:
 280         case GREP_PATTERN_BODY:
 281                 for (;;) {
 282                         struct grep_pat *new_pat;
 283                         size_t len = 0;
 284                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 285                         while (++len <= p->patternlen) {
 286                                 if (*(--cp) == '\n') {
 287                                         nl = cp;
 288                                         break;
 289                                 }
 290                         }
 291                         if (!nl)
 292                                 break;
 293                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 294                                                   p->no, p->token, p->field);
 295                         new_pat->next = p->next;
 296                         if (!p->next)
 297                                 *tail = &new_pat->next;
 298                         p->next = new_pat;
 299                         *nl = '\0';
 300                         p->patternlen -= len;
 301                 }
 302                 break;
 303         default:
 304                 break;
 305         }
 306 }
 307
 308 void append_header_grep_pattern(struct grep_opt *opt,
 309                                 enum grep_header_field field, const char *pat)
 310 {
 311         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 312                                              GREP_PATTERN_HEAD, field);
 313         if (field == GREP_HEADER_REFLOG)
 314                 opt->use_reflog_filter = 1;
 315         do_append_grep_pat(&opt->header_tail, p);
 316 }
 317
 318 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 319                          const char *origin, int no, enum grep_pat_token t)
 320 {
 321         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 322 }
 323
 324 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 325                      const char *origin, int no, enum grep_pat_token t)
 326 {
 327         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 328         do_append_grep_pat(&opt->pattern_tail, p);
 329 }
 330
 331 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 332 {
 333         struct grep_pat *pat;
 334         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 335         *ret = *opt;
 336
 337         ret->pattern_list = NULL;
 338         ret->pattern_tail = &ret->pattern_list;
 339
 340         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 341         {
 342                 if(pat->token == GREP_PATTERN_HEAD)
 343                         append_header_grep_pattern(ret, pat->field,
 344                                                    pat->pattern);
 345                 else
 346                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 347                                         pat->origin, pat->no, pat->token);
 348         }
 349
 350         return ret;
 351 }
 352
 353 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 354                 const char *error)
 355 {
 356         char where[1024];
 357
 358         if (p->no)
 359                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 360         else if (p->origin)
 361                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 362         else
 363                 where[0] = 0;
 364
 365         die("%s'%s': %s", where, p->pattern, error);
 366 }
 367
 368 static int is_fixed(const char *s, size_t len)
 369 {
 370         size_t i;
 371
 372         for (i = 0; i < len; i++) {
 373                 if (is_regex_special(s[i]))
 374                         return 0;
 375         }
 376
 377         return 1;
 378 }
 379
 380 #ifdef USE_LIBPCRE1
 381 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 382 {
 383         const char *error;
 384         int erroffset;
 385         int options = PCRE_MULTILINE;
 386         int study_options = 0;
 387
 388         if (opt->ignore_case) {
 389                 if (!opt->ignore_locale && has_non_ascii(p->pattern))
 390                         p->pcre1_tables = pcre_maketables();
 391                 options |= PCRE_CASELESS;
 392         }
 393         if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern))
 394                 options |= PCRE_UTF8;
 395
 396         p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 397                                       p->pcre1_tables);
 398         if (!p->pcre1_regexp)
 399                 compile_regexp_failed(p, error);
 400
 401 #if defined(PCRE_CONFIG_JIT) && !defined(NO_LIBPCRE1_JIT)
 402         pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
 403
 404         if (p->pcre1_jit_on)
 405                 study_options = PCRE_STUDY_JIT_COMPILE;
 406 #endif
 407
 408         p->pcre1_extra_info = pcre_study(p->pcre1_regexp, study_options, &error);
 409         if (!p->pcre1_extra_info && error)
 410                 die("%s", error);
 411 }
 412
 413 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 414                 regmatch_t *match, int eflags)
 415 {
 416         int ovector[30], ret, flags = PCRE_NO_UTF8_CHECK;
 417
 418         if (eflags & REG_NOTBOL)
 419                 flags |= PCRE_NOTBOL;
 420
 421         ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 422                         eol - line, 0, flags, ovector,
 423                         ARRAY_SIZE(ovector));
 424
 425         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 426                 die("pcre_exec failed with error code %d", ret);
 427         if (ret > 0) {
 428                 ret = 0;
 429                 match->rm_so = ovector[0];
 430                 match->rm_eo = ovector[1];
 431         }
 432
 433         return ret;
 434 }
 435
 436 static void free_pcre1_regexp(struct grep_pat *p)
 437 {
 438         pcre_free(p->pcre1_regexp);
 439 #ifdef PCRE_CONFIG_JIT
 440         if (p->pcre1_jit_on)
 441                 pcre_free_study(p->pcre1_extra_info);
 442         else
 443 #endif
 444                 pcre_free(p->pcre1_extra_info);
 445         pcre_free((void *)p->pcre1_tables);
 446 }
 447 #else /* !USE_LIBPCRE1 */
 448 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 449 {
 450         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 451 }
 452
 453 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 454                 regmatch_t *match, int eflags)
 455 {
 456         return 1;
 457 }
 458
 459 static void free_pcre1_regexp(struct grep_pat *p)
 460 {
 461 }
 462 #endif /* !USE_LIBPCRE1 */
 463
 464 #ifdef USE_LIBPCRE2
 465 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 466 {
 467         int error;
 468         PCRE2_UCHAR errbuf[256];
 469         PCRE2_SIZE erroffset;
 470         int options = PCRE2_MULTILINE;
 471         int jitret;
 472         int patinforet;
 473         size_t jitsizearg;
 474
 475         assert(opt->pcre2);
 476
 477         p->pcre2_compile_context = NULL;
 478
 479         /* pcre2_global_context is initialized in append_grep_pattern */
 480         if (opt->ignore_case) {
 481                 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
 482                         if (!pcre2_global_context)
 483                                 BUG("pcre2_global_context uninitialized");
 484                         p->pcre2_tables = pcre2_maketables(pcre2_global_context);
 485                         p->pcre2_compile_context = pcre2_compile_context_create(NULL);
 486                         pcre2_set_character_tables(p->pcre2_compile_context,
 487                                                         p->pcre2_tables);
 488                 }
 489                 options |= PCRE2_CASELESS;
 490         }
 491         if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern) &&
 492             !(!opt->ignore_case && (p->fixed || p->is_fixed)))
 493                 options |= PCRE2_UTF;
 494
 495         p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
 496                                          p->patternlen, options, &error, &erroffset,
 497                                          p->pcre2_compile_context);
 498
 499         if (p->pcre2_pattern) {
 500                 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
 501                 if (!p->pcre2_match_data)
 502                         die("Couldn't allocate PCRE2 match data");
 503         } else {
 504                 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
 505                 compile_regexp_failed(p, (const char *)&errbuf);
 506         }
 507
 508         pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
 509         if (p->pcre2_jit_on) {
 510                 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
 511                 if (jitret)
 512                         die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
 513
 514                 /*
 515                  * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
 516                  * tells us whether the library itself supports JIT,
 517                  * but to see whether we're going to be actually using
 518                  * JIT we need to extract PCRE2_INFO_JITSIZE from the
 519                  * pattern *after* we do pcre2_jit_compile() above.
 520                  *
 521                  * This is because if the pattern contains the
 522                  * (*NO_JIT) verb (see pcre2syntax(3))
 523                  * pcre2_jit_compile() will exit early with 0. If we
 524                  * then proceed to call pcre2_jit_match() further down
 525                  * the line instead of pcre2_match() we'll either
 526                  * segfault (pre PCRE 10.31) or run into a fatal error
 527                  * (post PCRE2 10.31)
 528                  */
 529                 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
 530                 if (patinforet)
 531                         BUG("pcre2_pattern_info() failed: %d", patinforet);
 532                 if (jitsizearg == 0) {
 533                         p->pcre2_jit_on = 0;
 534                         return;
 535                 }
 536         }
 537 }
 538
 539 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 540                 regmatch_t *match, int eflags)
 541 {
 542         int ret, flags = 0;
 543         PCRE2_SIZE *ovector;
 544         PCRE2_UCHAR errbuf[256];
 545
 546         if (eflags & REG_NOTBOL)
 547                 flags |= PCRE2_NOTBOL;
 548
 549         if (p->pcre2_jit_on)
 550                 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 551                                       eol - line, 0, flags, p->pcre2_match_data,
 552                                       NULL);
 553         else
 554                 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 555                                   eol - line, 0, flags, p->pcre2_match_data,
 556                                   NULL);
 557
 558         if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 559                 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 560                 die("%s failed with error code %d: %s",
 561                     (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 562                     errbuf);
 563         }
 564         if (ret > 0) {
 565                 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 566                 ret = 0;
 567                 match->rm_so = (int)ovector[0];
 568                 match->rm_eo = (int)ovector[1];
 569         }
 570
 571         return ret;
 572 }
 573
 574 static void free_pcre2_pattern(struct grep_pat *p)
 575 {
 576         pcre2_compile_context_free(p->pcre2_compile_context);
 577         pcre2_code_free(p->pcre2_pattern);
 578         pcre2_match_data_free(p->pcre2_match_data);
 579         free((void *)p->pcre2_tables);
 580 }
 581 #else /* !USE_LIBPCRE2 */
 582 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 583 {
 584         /*
 585          * Unreachable until USE_LIBPCRE2 becomes synonymous with
 586          * USE_LIBPCRE. See the sibling comment in
 587          * grep_set_pattern_type_option().
 588          */
 589         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 590 }
 591
 592 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 593                 regmatch_t *match, int eflags)
 594 {
 595         return 1;
 596 }
 597
 598 static void free_pcre2_pattern(struct grep_pat *p)
 599 {
 600 }
 601
 602 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 603 {
 604         struct strbuf sb = STRBUF_INIT;
 605         int err;
 606         int regflags = 0;
 607
 608         basic_regex_quote_buf(&sb, p->pattern);
 609         if (opt->ignore_case)
 610                 regflags |= REG_ICASE;
 611         err = regcomp(&p->regexp, sb.buf, regflags);
 612         strbuf_release(&sb);
 613         if (err) {
 614                 char errbuf[1024];
 615                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 616                 compile_regexp_failed(p, errbuf);
 617         }
 618 }
 619 #endif /* !USE_LIBPCRE2 */
 620
 621 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 622 {
 623         int err;
 624         int regflags = REG_NEWLINE;
 625
 626         p->word_regexp = opt->word_regexp;
 627         p->ignore_case = opt->ignore_case;
 628         p->fixed = opt->fixed;
 629
 630         if (memchr(p->pattern, 0, p->patternlen) && !opt->pcre2)
 631                 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
 632
 633         p->is_fixed = is_fixed(p->pattern, p->patternlen);
 634 #ifdef USE_LIBPCRE2
 635        if (!p->fixed && !p->is_fixed) {
 636                const char *no_jit = "(*NO_JIT)";
 637                const int no_jit_len = strlen(no_jit);
 638                if (starts_with(p->pattern, no_jit) &&
 639                    is_fixed(p->pattern + no_jit_len,
 640                             p->patternlen - no_jit_len))
 641                        p->is_fixed = 1;
 642        }
 643 #endif
 644         if (p->fixed || p->is_fixed) {
 645 #ifdef USE_LIBPCRE2
 646                 opt->pcre2 = 1;
 647                 if (p->is_fixed) {
 648                         compile_pcre2_pattern(p, opt);
 649                 } else {
 650                         /*
 651                          * E.g. t7811-grep-open.sh relies on the
 652                          * pattern being restored.
 653                          */
 654                         char *old_pattern = p->pattern;
 655                         size_t old_patternlen = p->patternlen;
 656                         struct strbuf sb = STRBUF_INIT;
 657
 658                         /*
 659                          * There is the PCRE2_LITERAL flag, but it's
 660                          * only in PCRE v2 10.30 and later. Needing to
 661                          * ifdef our way around that and dealing with
 662                          * it + PCRE2_MULTILINE being an error is more
 663                          * complex than just quoting this ourselves.
 664                         */
 665                         strbuf_add(&sb, "\\Q", 2);
 666                         strbuf_add(&sb, p->pattern, p->patternlen);
 667                         strbuf_add(&sb, "\\E", 2);
 668
 669                         p->pattern = sb.buf;
 670                         p->patternlen = sb.len;
 671                         compile_pcre2_pattern(p, opt);
 672                         p->pattern = old_pattern;
 673                         p->patternlen = old_patternlen;
 674                         strbuf_release(&sb);
 675                 }
 676 #else /* !USE_LIBPCRE2 */
 677                 compile_fixed_regexp(p, opt);
 678 #endif /* !USE_LIBPCRE2 */
 679                 return;
 680         }
 681
 682         if (opt->pcre2) {
 683                 compile_pcre2_pattern(p, opt);
 684                 return;
 685         }
 686
 687         if (opt->pcre1) {
 688                 compile_pcre1_regexp(p, opt);
 689                 return;
 690         }
 691
 692         if (p->ignore_case)
 693                 regflags |= REG_ICASE;
 694         if (opt->extended_regexp_option)
 695                 regflags |= REG_EXTENDED;
 696         err = regcomp(&p->regexp, p->pattern, regflags);
 697         if (err) {
 698                 char errbuf[1024];
 699                 regerror(err, &p->regexp, errbuf, 1024);
 700                 compile_regexp_failed(p, errbuf);
 701         }
 702 }
 703
 704 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 705 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 706 {
 707         struct grep_pat *p;
 708         struct grep_expr *x;
 709
 710         p = *list;
 711         if (!p)
 712                 return NULL;
 713         switch (p->token) {
 714         case GREP_PATTERN: /* atom */
 715         case GREP_PATTERN_HEAD:
 716         case GREP_PATTERN_BODY:
 717                 x = xcalloc(1, sizeof (struct grep_expr));
 718                 x->node = GREP_NODE_ATOM;
 719                 x->u.atom = p;
 720                 *list = p->next;
 721                 return x;
 722         case GREP_OPEN_PAREN:
 723                 *list = p->next;
 724                 x = compile_pattern_or(list);
 725                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 726                         die("unmatched parenthesis");
 727                 *list = (*list)->next;
 728                 return x;
 729         default:
 730                 return NULL;
 731         }
 732 }
 733
 734 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 735 {
 736         struct grep_pat *p;
 737         struct grep_expr *x;
 738
 739         p = *list;
 740         if (!p)
 741                 return NULL;
 742         switch (p->token) {
 743         case GREP_NOT:
 744                 if (!p->next)
 745                         die("--not not followed by pattern expression");
 746                 *list = p->next;
 747                 x = xcalloc(1, sizeof (struct grep_expr));
 748                 x->node = GREP_NODE_NOT;
 749                 x->u.unary = compile_pattern_not(list);
 750                 if (!x->u.unary)
 751                         die("--not followed by non pattern expression");
 752                 return x;
 753         default:
 754                 return compile_pattern_atom(list);
 755         }
 756 }
 757
 758 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 759 {
 760         struct grep_pat *p;
 761         struct grep_expr *x, *y, *z;
 762
 763         x = compile_pattern_not(list);
 764         p = *list;
 765         if (p && p->token == GREP_AND) {
 766                 if (!p->next)
 767                         die("--and not followed by pattern expression");
 768                 *list = p->next;
 769                 y = compile_pattern_and(list);
 770                 if (!y)
 771                         die("--and not followed by pattern expression");
 772                 z = xcalloc(1, sizeof (struct grep_expr));
 773                 z->node = GREP_NODE_AND;
 774                 z->u.binary.left = x;
 775                 z->u.binary.right = y;
 776                 return z;
 777         }
 778         return x;
 779 }
 780
 781 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 782 {
 783         struct grep_pat *p;
 784         struct grep_expr *x, *y, *z;
 785
 786         x = compile_pattern_and(list);
 787         p = *list;
 788         if (x && p && p->token != GREP_CLOSE_PAREN) {
 789                 y = compile_pattern_or(list);
 790                 if (!y)
 791                         die("not a pattern expression %s", p->pattern);
 792                 z = xcalloc(1, sizeof (struct grep_expr));
 793                 z->node = GREP_NODE_OR;
 794                 z->u.binary.left = x;
 795                 z->u.binary.right = y;
 796                 return z;
 797         }
 798         return x;
 799 }
 800
 801 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 802 {
 803         return compile_pattern_or(list);
 804 }
 805
 806 static struct grep_expr *grep_true_expr(void)
 807 {
 808         struct grep_expr *z = xcalloc(1, sizeof(*z));
 809         z->node = GREP_NODE_TRUE;
 810         return z;
 811 }
 812
 813 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 814 {
 815         struct grep_expr *z = xcalloc(1, sizeof(*z));
 816         z->node = GREP_NODE_OR;
 817         z->u.binary.left = left;
 818         z->u.binary.right = right;
 819         return z;
 820 }
 821
 822 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 823 {
 824         struct grep_pat *p;
 825         struct grep_expr *header_expr;
 826         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 827         enum grep_header_field fld;
 828
 829         if (!opt->header_list)
 830                 return NULL;
 831
 832         for (p = opt->header_list; p; p = p->next) {
 833                 if (p->token != GREP_PATTERN_HEAD)
 834                         BUG("a non-header pattern in grep header list.");
 835                 if (p->field < GREP_HEADER_FIELD_MIN ||
 836                     GREP_HEADER_FIELD_MAX <= p->field)
 837                         BUG("unknown header field %d", p->field);
 838                 compile_regexp(p, opt);
 839         }
 840
 841         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 842                 header_group[fld] = NULL;
 843
 844         for (p = opt->header_list; p; p = p->next) {
 845                 struct grep_expr *h;
 846                 struct grep_pat *pp = p;
 847
 848                 h = compile_pattern_atom(&pp);
 849                 if (!h || pp != p->next)
 850                         BUG("malformed header expr");
 851                 if (!header_group[p->field]) {
 852                         header_group[p->field] = h;
 853                         continue;
 854                 }
 855                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 856         }
 857
 858         header_expr = NULL;
 859
 860         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 861                 if (!header_group[fld])
 862                         continue;
 863                 if (!header_expr)
 864                         header_expr = grep_true_expr();
 865                 header_expr = grep_or_expr(header_group[fld], header_expr);
 866         }
 867         return header_expr;
 868 }
 869
 870 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 871 {
 872         struct grep_expr *z = x;
 873
 874         while (x) {
 875                 assert(x->node == GREP_NODE_OR);
 876                 if (x->u.binary.right &&
 877                     x->u.binary.right->node == GREP_NODE_TRUE) {
 878                         x->u.binary.right = y;
 879                         break;
 880                 }
 881                 x = x->u.binary.right;
 882         }
 883         return z;
 884 }
 885
 886 void compile_grep_patterns(struct grep_opt *opt)
 887 {
 888         struct grep_pat *p;
 889         struct grep_expr *header_expr = prep_header_patterns(opt);
 890
 891         for (p = opt->pattern_list; p; p = p->next) {
 892                 switch (p->token) {
 893                 case GREP_PATTERN: /* atom */
 894                 case GREP_PATTERN_HEAD:
 895                 case GREP_PATTERN_BODY:
 896                         compile_regexp(p, opt);
 897                         break;
 898                 default:
 899                         opt->extended = 1;
 900                         break;
 901                 }
 902         }
 903
 904         if (opt->all_match || header_expr)
 905                 opt->extended = 1;
 906         else if (!opt->extended)
 907                 return;
 908
 909         p = opt->pattern_list;
 910         if (p)
 911                 opt->pattern_expression = compile_pattern_expr(&p);
 912         if (p)
 913                 die("incomplete pattern expression: %s", p->pattern);
 914
 915         if (!header_expr)
 916                 return;
 917
 918         if (!opt->pattern_expression)
 919                 opt->pattern_expression = header_expr;
 920         else if (opt->all_match)
 921                 opt->pattern_expression = grep_splice_or(header_expr,
 922                                                          opt->pattern_expression);
 923         else
 924                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 925                                                        header_expr);
 926         opt->all_match = 1;
 927 }
 928
 929 static void free_pattern_expr(struct grep_expr *x)
 930 {
 931         switch (x->node) {
 932         case GREP_NODE_TRUE:
 933         case GREP_NODE_ATOM:
 934                 break;
 935         case GREP_NODE_NOT:
 936                 free_pattern_expr(x->u.unary);
 937                 break;
 938         case GREP_NODE_AND:
 939         case GREP_NODE_OR:
 940                 free_pattern_expr(x->u.binary.left);
 941                 free_pattern_expr(x->u.binary.right);
 942                 break;
 943         }
 944         free(x);
 945 }
 946
 947 void free_grep_patterns(struct grep_opt *opt)
 948 {
 949         struct grep_pat *p, *n;
 950
 951         for (p = opt->pattern_list; p; p = n) {
 952                 n = p->next;
 953                 switch (p->token) {
 954                 case GREP_PATTERN: /* atom */
 955                 case GREP_PATTERN_HEAD:
 956                 case GREP_PATTERN_BODY:
 957                         if (p->pcre1_regexp)
 958                                 free_pcre1_regexp(p);
 959                         else if (p->pcre2_pattern)
 960                                 free_pcre2_pattern(p);
 961                         else
 962                                 regfree(&p->regexp);
 963                         free(p->pattern);
 964                         break;
 965                 default:
 966                         break;
 967                 }
 968                 free(p);
 969         }
 970
 971         if (!opt->extended)
 972                 return;
 973         free_pattern_expr(opt->pattern_expression);
 974 }
 975
 976 static char *end_of_line(char *cp, unsigned long *left)
 977 {
 978         unsigned long l = *left;
 979         while (l && *cp != '\n') {
 980                 l--;
 981                 cp++;
 982         }
 983         *left = l;
 984         return cp;
 985 }
 986
 987 static int word_char(char ch)
 988 {
 989         return isalnum(ch) || ch == '_';
 990 }
 991
 992 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 993                          const char *color)
 994 {
 995         if (want_color(opt->color) && color && color[0]) {
 996                 opt->output(opt, color, strlen(color));
 997                 opt->output(opt, data, size);
 998                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 999         } else
1000                 opt->output(opt, data, size);
1001 }
1002
1003 static void output_sep(struct grep_opt *opt, char sign)
1004 {
1005         if (opt->null_following_name)
1006                 opt->output(opt, "\0", 1);
1007         else
1008                 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
1009 }
1010
1011 static void show_name(struct grep_opt *opt, const char *name)
1012 {
1013         output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1014         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1015 }
1016
1017 static int patmatch(struct grep_pat *p, char *line, char *eol,
1018                     regmatch_t *match, int eflags)
1019 {
1020         int hit;
1021
1022         if (p->pcre1_regexp)
1023                 hit = !pcre1match(p, line, eol, match, eflags);
1024         else if (p->pcre2_pattern)
1025                 hit = !pcre2match(p, line, eol, match, eflags);
1026         else
1027                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1028                                    eflags);
1029
1030         return hit;
1031 }
1032
1033 static int strip_timestamp(char *bol, char **eol_p)
1034 {
1035         char *eol = *eol_p;
1036         int ch;
1037
1038         while (bol < --eol) {
1039                 if (*eol != '>')
1040                         continue;
1041                 *eol_p = ++eol;
1042                 ch = *eol;
1043                 *eol = '\0';
1044                 return ch;
1045         }
1046         return 0;
1047 }
1048
1049 static struct {
1050         const char *field;
1051         size_t len;
1052 } header_field[] = {
1053         { "author ", 7 },
1054         { "committer ", 10 },
1055         { "reflog ", 7 },
1056 };
1057
1058 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1059                              enum grep_context ctx,
1060                              regmatch_t *pmatch, int eflags)
1061 {
1062         int hit = 0;
1063         int saved_ch = 0;
1064         const char *start = bol;
1065
1066         if ((p->token != GREP_PATTERN) &&
1067             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1068                 return 0;
1069
1070         if (p->token == GREP_PATTERN_HEAD) {
1071                 const char *field;
1072                 size_t len;
1073                 assert(p->field < ARRAY_SIZE(header_field));
1074                 field = header_field[p->field].field;
1075                 len = header_field[p->field].len;
1076                 if (strncmp(bol, field, len))
1077                         return 0;
1078                 bol += len;
1079                 switch (p->field) {
1080                 case GREP_HEADER_AUTHOR:
1081                 case GREP_HEADER_COMMITTER:
1082                         saved_ch = strip_timestamp(bol, &eol);
1083                         break;
1084                 default:
1085                         break;
1086                 }
1087         }
1088
1089  again:
1090         hit = patmatch(p, bol, eol, pmatch, eflags);
1091
1092         if (hit && p->word_regexp) {
1093                 if ((pmatch[0].rm_so < 0) ||
1094                     (eol - bol) < pmatch[0].rm_so ||
1095                     (pmatch[0].rm_eo < 0) ||
1096                     (eol - bol) < pmatch[0].rm_eo)
1097                         die("regexp returned nonsense");
1098
1099                 /* Match beginning must be either beginning of the
1100                  * line, or at word boundary (i.e. the last char must
1101                  * not be a word char).  Similarly, match end must be
1102                  * either end of the line, or at word boundary
1103                  * (i.e. the next char must not be a word char).
1104                  */
1105                 if ( ((pmatch[0].rm_so == 0) ||
1106                       !word_char(bol[pmatch[0].rm_so-1])) &&
1107                      ((pmatch[0].rm_eo == (eol-bol)) ||
1108                       !word_char(bol[pmatch[0].rm_eo])) )
1109                         ;
1110                 else
1111                         hit = 0;
1112
1113                 /* Words consist of at least one character. */
1114                 if (pmatch->rm_so == pmatch->rm_eo)
1115                         hit = 0;
1116
1117                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1118                         /* There could be more than one match on the
1119                          * line, and the first match might not be
1120                          * strict word match.  But later ones could be!
1121                          * Forward to the next possible start, i.e. the
1122                          * next position following a non-word char.
1123                          */
1124                         bol = pmatch[0].rm_so + bol + 1;
1125                         while (word_char(bol[-1]) && bol < eol)
1126                                 bol++;
1127                         eflags |= REG_NOTBOL;
1128                         if (bol < eol)
1129                                 goto again;
1130                 }
1131         }
1132         if (p->token == GREP_PATTERN_HEAD && saved_ch)
1133                 *eol = saved_ch;
1134         if (hit) {
1135                 pmatch[0].rm_so += bol - start;
1136                 pmatch[0].rm_eo += bol - start;
1137         }
1138         return hit;
1139 }
1140
1141 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x, char *bol,
1142                            char *eol, enum grep_context ctx, ssize_t *col,
1143                            ssize_t *icol, int collect_hits)
1144 {
1145         int h = 0;
1146
1147         if (!x)
1148                 die("Not a valid grep expression");
1149         switch (x->node) {
1150         case GREP_NODE_TRUE:
1151                 h = 1;
1152                 break;
1153         case GREP_NODE_ATOM:
1154                 {
1155                         regmatch_t tmp;
1156                         h = match_one_pattern(x->u.atom, bol, eol, ctx,
1157                                               &tmp, 0);
1158                         if (h && (*col < 0 || tmp.rm_so < *col))
1159                                 *col = tmp.rm_so;
1160                 }
1161                 break;
1162         case GREP_NODE_NOT:
1163                 /*
1164                  * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1165                  */
1166                 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1167                                      0);
1168                 break;
1169         case GREP_NODE_AND:
1170                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1171                                     icol, 0);
1172                 if (h || opt->columnnum) {
1173                         /*
1174                          * Don't short-circuit AND when given --column, since a
1175                          * NOT earlier in the tree may turn this into an OR. In
1176                          * this case, see the below comment.
1177                          */
1178                         h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1179                                              ctx, col, icol, 0);
1180                 }
1181                 break;
1182         case GREP_NODE_OR:
1183                 if (!(collect_hits || opt->columnnum)) {
1184                         /*
1185                          * Don't short-circuit OR when given --column (or
1186                          * collecting hits) to ensure we don't skip a later
1187                          * child that would produce an earlier match.
1188                          */
1189                         return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1190                                                 ctx, col, icol, 0) ||
1191                                 match_expr_eval(opt, x->u.binary.right, bol,
1192                                                 eol, ctx, col, icol, 0));
1193                 }
1194                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1195                                     icol, 0);
1196                 if (collect_hits)
1197                         x->u.binary.left->hit |= h;
1198                 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1199                                      icol, collect_hits);
1200                 break;
1201         default:
1202                 die("Unexpected node type (internal error) %d", x->node);
1203         }
1204         if (collect_hits)
1205                 x->hit |= h;
1206         return h;
1207 }
1208
1209 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1210                       enum grep_context ctx, ssize_t *col,
1211                       ssize_t *icol, int collect_hits)
1212 {
1213         struct grep_expr *x = opt->pattern_expression;
1214         return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1215 }
1216
1217 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1218                       ssize_t *col, ssize_t *icol,
1219                       enum grep_context ctx, int collect_hits)
1220 {
1221         struct grep_pat *p;
1222         int hit = 0;
1223
1224         if (opt->extended)
1225                 return match_expr(opt, bol, eol, ctx, col, icol,
1226                                   collect_hits);
1227
1228         /* we do not call with collect_hits without being extended */
1229         for (p = opt->pattern_list; p; p = p->next) {
1230                 regmatch_t tmp;
1231                 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1232                         hit |= 1;
1233                         if (!opt->columnnum) {
1234                                 /*
1235                                  * Without --column, any single match on a line
1236                                  * is enough to know that it needs to be
1237                                  * printed. With --column, scan _all_ patterns
1238                                  * to find the earliest.
1239                                  */
1240                                 break;
1241                         }
1242                         if (*col < 0 || tmp.rm_so < *col)
1243                                 *col = tmp.rm_so;
1244                 }
1245         }
1246         return hit;
1247 }
1248
1249 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1250                               enum grep_context ctx,
1251                               regmatch_t *pmatch, int eflags)
1252 {
1253         regmatch_t match;
1254
1255         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1256                 return 0;
1257         if (match.rm_so < 0 || match.rm_eo < 0)
1258                 return 0;
1259         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1260                 if (match.rm_so > pmatch->rm_so)
1261                         return 1;
1262                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1263                         return 1;
1264         }
1265         pmatch->rm_so = match.rm_so;
1266         pmatch->rm_eo = match.rm_eo;
1267         return 1;
1268 }
1269
1270 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1271                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1272 {
1273         struct grep_pat *p;
1274         int hit = 0;
1275
1276         pmatch->rm_so = pmatch->rm_eo = -1;
1277         if (bol < eol) {
1278                 for (p = opt->pattern_list; p; p = p->next) {
1279                         switch (p->token) {
1280                         case GREP_PATTERN: /* atom */
1281                         case GREP_PATTERN_HEAD:
1282                         case GREP_PATTERN_BODY:
1283                                 hit |= match_next_pattern(p, bol, eol, ctx,
1284                                                           pmatch, eflags);
1285                                 break;
1286                         default:
1287                                 break;
1288                         }
1289                 }
1290         }
1291         return hit;
1292 }
1293
1294 static void show_line_header(struct grep_opt *opt, const char *name,
1295                              unsigned lno, ssize_t cno, char sign)
1296 {
1297         if (opt->heading && opt->last_shown == 0) {
1298                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1299                 opt->output(opt, "\n", 1);
1300         }
1301         opt->last_shown = lno;
1302
1303         if (!opt->heading && opt->pathname) {
1304                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1305                 output_sep(opt, sign);
1306         }
1307         if (opt->linenum) {
1308                 char buf[32];
1309                 xsnprintf(buf, sizeof(buf), "%d", lno);
1310                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1311                 output_sep(opt, sign);
1312         }
1313         /*
1314          * Treat 'cno' as the 1-indexed offset from the start of a non-context
1315          * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1316          * being called with a context line.
1317          */
1318         if (opt->columnnum && cno) {
1319                 char buf[32];
1320                 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1321                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1322                 output_sep(opt, sign);
1323         }
1324 }
1325
1326 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1327                       const char *name, unsigned lno, ssize_t cno, char sign)
1328 {
1329         int rest = eol - bol;
1330         const char *match_color = NULL;
1331         const char *line_color = NULL;
1332
1333         if (opt->file_break && opt->last_shown == 0) {
1334                 if (opt->show_hunk_mark)
1335                         opt->output(opt, "\n", 1);
1336         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1337                 if (opt->last_shown == 0) {
1338                         if (opt->show_hunk_mark) {
1339                                 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1340                                 opt->output(opt, "\n", 1);
1341                         }
1342                 } else if (lno > opt->last_shown + 1) {
1343                         output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1344                         opt->output(opt, "\n", 1);
1345                 }
1346         }
1347         if (!opt->only_matching) {
1348                 /*
1349                  * In case the line we're being called with contains more than
1350                  * one match, leave printing each header to the loop below.
1351                  */
1352                 show_line_header(opt, name, lno, cno, sign);
1353         }
1354         if (opt->color || opt->only_matching) {
1355                 regmatch_t match;
1356                 enum grep_context ctx = GREP_CONTEXT_BODY;
1357                 int ch = *eol;
1358                 int eflags = 0;
1359
1360                 if (opt->color) {
1361                         if (sign == ':')
1362                                 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1363                         else
1364                                 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1365                         if (sign == ':')
1366                                 line_color = opt->colors[GREP_COLOR_SELECTED];
1367                         else if (sign == '-')
1368                                 line_color = opt->colors[GREP_COLOR_CONTEXT];
1369                         else if (sign == '=')
1370                                 line_color = opt->colors[GREP_COLOR_FUNCTION];
1371                 }
1372                 *eol = '\0';
1373                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1374                         if (match.rm_so == match.rm_eo)
1375                                 break;
1376
1377                         if (opt->only_matching)
1378                                 show_line_header(opt, name, lno, cno, sign);
1379                         else
1380                                 output_color(opt, bol, match.rm_so, line_color);
1381                         output_color(opt, bol + match.rm_so,
1382                                      match.rm_eo - match.rm_so, match_color);
1383                         if (opt->only_matching)
1384                                 opt->output(opt, "\n", 1);
1385                         bol += match.rm_eo;
1386                         cno += match.rm_eo;
1387                         rest -= match.rm_eo;
1388                         eflags = REG_NOTBOL;
1389                 }
1390                 *eol = ch;
1391         }
1392         if (!opt->only_matching) {
1393                 output_color(opt, bol, rest, line_color);
1394                 opt->output(opt, "\n", 1);
1395         }
1396 }
1397
1398 int grep_use_locks;
1399
1400 /*
1401  * This lock protects access to the gitattributes machinery, which is
1402  * not thread-safe.
1403  */
1404 pthread_mutex_t grep_attr_mutex;
1405
1406 static inline void grep_attr_lock(void)
1407 {
1408         if (grep_use_locks)
1409                 pthread_mutex_lock(&grep_attr_mutex);
1410 }
1411
1412 static inline void grep_attr_unlock(void)
1413 {
1414         if (grep_use_locks)
1415                 pthread_mutex_unlock(&grep_attr_mutex);
1416 }
1417
1418 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1419 {
1420         xdemitconf_t *xecfg = opt->priv;
1421         if (xecfg && !xecfg->find_func) {
1422                 grep_source_load_driver(gs, opt->repo->index);
1423                 if (gs->driver->funcname.pattern) {
1424                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1425                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1426                 } else {
1427                         xecfg = opt->priv = NULL;
1428                 }
1429         }
1430
1431         if (xecfg) {
1432                 char buf[1];
1433                 return xecfg->find_func(bol, eol - bol, buf, 1,
1434                                         xecfg->find_func_priv) >= 0;
1435         }
1436
1437         if (bol == eol)
1438                 return 0;
1439         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1440                 return 1;
1441         return 0;
1442 }
1443
1444 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1445                                char *bol, unsigned lno)
1446 {
1447         while (bol > gs->buf) {
1448                 char *eol = --bol;
1449
1450                 while (bol > gs->buf && bol[-1] != '\n')
1451                         bol--;
1452                 lno--;
1453
1454                 if (lno <= opt->last_shown)
1455                         break;
1456
1457                 if (match_funcname(opt, gs, bol, eol)) {
1458                         show_line(opt, bol, eol, gs->name, lno, 0, '=');
1459                         break;
1460                 }
1461         }
1462 }
1463
1464 static int is_empty_line(const char *bol, const char *eol);
1465
1466 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1467                              char *bol, char *end, unsigned lno)
1468 {
1469         unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1470         int funcname_needed = !!opt->funcname, comment_needed = 0;
1471
1472         if (opt->pre_context < lno)
1473                 from = lno - opt->pre_context;
1474         if (from <= opt->last_shown)
1475                 from = opt->last_shown + 1;
1476         orig_from = from;
1477         if (opt->funcbody) {
1478                 if (match_funcname(opt, gs, bol, end))
1479                         comment_needed = 1;
1480                 else
1481                         funcname_needed = 1;
1482                 from = opt->last_shown + 1;
1483         }
1484
1485         /* Rewind. */
1486         while (bol > gs->buf && cur > from) {
1487                 char *next_bol = bol;
1488                 char *eol = --bol;
1489
1490                 while (bol > gs->buf && bol[-1] != '\n')
1491                         bol--;
1492                 cur--;
1493                 if (comment_needed && (is_empty_line(bol, eol) ||
1494                                        match_funcname(opt, gs, bol, eol))) {
1495                         comment_needed = 0;
1496                         from = orig_from;
1497                         if (cur < from) {
1498                                 cur++;
1499                                 bol = next_bol;
1500                                 break;
1501                         }
1502                 }
1503                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1504                         funcname_lno = cur;
1505                         funcname_needed = 0;
1506                         if (opt->funcbody)
1507                                 comment_needed = 1;
1508                         else
1509                                 from = orig_from;
1510                 }
1511         }
1512
1513         /* We need to look even further back to find a function signature. */
1514         if (opt->funcname && funcname_needed)
1515                 show_funcname_line(opt, gs, bol, cur);
1516
1517         /* Back forward. */
1518         while (cur < lno) {
1519                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1520
1521                 while (*eol != '\n')
1522                         eol++;
1523                 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1524                 bol = eol + 1;
1525                 cur++;
1526         }
1527 }
1528
1529 static int should_lookahead(struct grep_opt *opt)
1530 {
1531         struct grep_pat *p;
1532
1533         if (opt->extended)
1534                 return 0; /* punt for too complex stuff */
1535         if (opt->invert)
1536                 return 0;
1537         for (p = opt->pattern_list; p; p = p->next) {
1538                 if (p->token != GREP_PATTERN)
1539                         return 0; /* punt for "header only" and stuff */
1540         }
1541         return 1;
1542 }
1543
1544 static int look_ahead(struct grep_opt *opt,
1545                       unsigned long *left_p,
1546                       unsigned *lno_p,
1547                       char **bol_p)
1548 {
1549         unsigned lno = *lno_p;
1550         char *bol = *bol_p;
1551         struct grep_pat *p;
1552         char *sp, *last_bol;
1553         regoff_t earliest = -1;
1554
1555         for (p = opt->pattern_list; p; p = p->next) {
1556                 int hit;
1557                 regmatch_t m;
1558
1559                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1560                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1561                         continue;
1562                 if (earliest < 0 || m.rm_so < earliest)
1563                         earliest = m.rm_so;
1564         }
1565
1566         if (earliest < 0) {
1567                 *bol_p = bol + *left_p;
1568                 *left_p = 0;
1569                 return 1;
1570         }
1571         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1572                 ; /* find the beginning of the line */
1573         last_bol = sp;
1574
1575         for (sp = bol; sp < last_bol; sp++) {
1576                 if (*sp == '\n')
1577                         lno++;
1578         }
1579         *left_p -= last_bol - bol;
1580         *bol_p = last_bol;
1581         *lno_p = lno;
1582         return 0;
1583 }
1584
1585 static int fill_textconv_grep(struct repository *r,
1586                               struct userdiff_driver *driver,
1587                               struct grep_source *gs)
1588 {
1589         struct diff_filespec *df;
1590         char *buf;
1591         size_t size;
1592
1593         if (!driver || !driver->textconv)
1594                 return grep_source_load(gs);
1595
1596         /*
1597          * The textconv interface is intimately tied to diff_filespecs, so we
1598          * have to pretend to be one. If we could unify the grep_source
1599          * and diff_filespec structs, this mess could just go away.
1600          */
1601         df = alloc_filespec(gs->path);
1602         switch (gs->type) {
1603         case GREP_SOURCE_OID:
1604                 fill_filespec(df, gs->identifier, 1, 0100644);
1605                 break;
1606         case GREP_SOURCE_FILE:
1607                 fill_filespec(df, &null_oid, 0, 0100644);
1608                 break;
1609         default:
1610                 BUG("attempt to textconv something without a path?");
1611         }
1612
1613         /*
1614          * fill_textconv is not remotely thread-safe; it modifies the global
1615          * diff tempfile structure, writes to the_repo's odb and might
1616          * internally call thread-unsafe functions such as the
1617          * prepare_packed_git() lazy-initializator. Because of the last two, we
1618          * must ensure mutual exclusion between this call and the object reading
1619          * API, thus we use obj_read_lock() here.
1620          *
1621          * TODO: allowing text conversion to run in parallel with object
1622          * reading operations might increase performance in the multithreaded
1623          * non-worktreee git-grep with --textconv.
1624          */
1625         obj_read_lock();
1626         size = fill_textconv(r, driver, df, &buf);
1627         obj_read_unlock();
1628         free_filespec(df);
1629
1630         /*
1631          * The normal fill_textconv usage by the diff machinery would just keep
1632          * the textconv'd buf separate from the diff_filespec. But much of the
1633          * grep code passes around a grep_source and assumes that its "buf"
1634          * pointer is the beginning of the thing we are searching. So let's
1635          * install our textconv'd version into the grep_source, taking care not
1636          * to leak any existing buffer.
1637          */
1638         grep_source_clear_data(gs);
1639         gs->buf = buf;
1640         gs->size = size;
1641
1642         return 0;
1643 }
1644
1645 static int is_empty_line(const char *bol, const char *eol)
1646 {
1647         while (bol < eol && isspace(*bol))
1648                 bol++;
1649         return bol == eol;
1650 }
1651
1652 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1653 {
1654         char *bol;
1655         char *peek_bol = NULL;
1656         unsigned long left;
1657         unsigned lno = 1;
1658         unsigned last_hit = 0;
1659         int binary_match_only = 0;
1660         unsigned count = 0;
1661         int try_lookahead = 0;
1662         int show_function = 0;
1663         struct userdiff_driver *textconv = NULL;
1664         enum grep_context ctx = GREP_CONTEXT_HEAD;
1665         xdemitconf_t xecfg;
1666
1667         if (!opt->status_only && gs->name == NULL)
1668                 BUG("grep call which could print a name requires "
1669                     "grep_source.name be non-NULL");
1670
1671         if (!opt->output)
1672                 opt->output = std_output;
1673
1674         if (opt->pre_context || opt->post_context || opt->file_break ||
1675             opt->funcbody) {
1676                 /* Show hunk marks, except for the first file. */
1677                 if (opt->last_shown)
1678                         opt->show_hunk_mark = 1;
1679                 /*
1680                  * If we're using threads then we can't easily identify
1681                  * the first file.  Always put hunk marks in that case
1682                  * and skip the very first one later in work_done().
1683                  */
1684                 if (opt->output != std_output)
1685                         opt->show_hunk_mark = 1;
1686         }
1687         opt->last_shown = 0;
1688
1689         if (opt->allow_textconv) {
1690                 grep_source_load_driver(gs, opt->repo->index);
1691                 /*
1692                  * We might set up the shared textconv cache data here, which
1693                  * is not thread-safe. Also, get_oid_with_context() and
1694                  * parse_object() might be internally called. As they are not
1695                  * currently thread-safe and might be racy with object reading,
1696                  * obj_read_lock() must be called.
1697                  */
1698                 grep_attr_lock();
1699                 obj_read_lock();
1700                 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1701                 obj_read_unlock();
1702                 grep_attr_unlock();
1703         }
1704
1705         /*
1706          * We know the result of a textconv is text, so we only have to care
1707          * about binary handling if we are not using it.
1708          */
1709         if (!textconv) {
1710                 switch (opt->binary) {
1711                 case GREP_BINARY_DEFAULT:
1712                         if (grep_source_is_binary(gs, opt->repo->index))
1713                                 binary_match_only = 1;
1714                         break;
1715                 case GREP_BINARY_NOMATCH:
1716                         if (grep_source_is_binary(gs, opt->repo->index))
1717                                 return 0; /* Assume unmatch */
1718                         break;
1719                 case GREP_BINARY_TEXT:
1720                         break;
1721                 default:
1722                         BUG("unknown binary handling mode");
1723                 }
1724         }
1725
1726         memset(&xecfg, 0, sizeof(xecfg));
1727         opt->priv = &xecfg;
1728
1729         try_lookahead = should_lookahead(opt);
1730
1731         if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1732                 return 0;
1733
1734         bol = gs->buf;
1735         left = gs->size;
1736         while (left) {
1737                 char *eol, ch;
1738                 int hit;
1739                 ssize_t cno;
1740                 ssize_t col = -1, icol = -1;
1741
1742                 /*
1743                  * look_ahead() skips quickly to the line that possibly
1744                  * has the next hit; don't call it if we need to do
1745                  * something more than just skipping the current line
1746                  * in response to an unmatch for the current line.  E.g.
1747                  * inside a post-context window, we will show the current
1748                  * line as a context around the previous hit when it
1749                  * doesn't hit.
1750                  */
1751                 if (try_lookahead
1752                     && !(last_hit
1753                          && (show_function ||
1754                              lno <= last_hit + opt->post_context))
1755                     && look_ahead(opt, &left, &lno, &bol))
1756                         break;
1757                 eol = end_of_line(bol, &left);
1758                 ch = *eol;
1759                 *eol = 0;
1760
1761                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1762                         ctx = GREP_CONTEXT_BODY;
1763
1764                 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1765                 *eol = ch;
1766
1767                 if (collect_hits)
1768                         goto next_line;
1769
1770                 /* "grep -v -e foo -e bla" should list lines
1771                  * that do not have either, so inversion should
1772                  * be done outside.
1773                  */
1774                 if (opt->invert)
1775                         hit = !hit;
1776                 if (opt->unmatch_name_only) {
1777                         if (hit)
1778                                 return 0;
1779                         goto next_line;
1780                 }
1781                 if (hit) {
1782                         count++;
1783                         if (opt->status_only)
1784                                 return 1;
1785                         if (opt->name_only) {
1786                                 show_name(opt, gs->name);
1787                                 return 1;
1788                         }
1789                         if (opt->count)
1790                                 goto next_line;
1791                         if (binary_match_only) {
1792                                 opt->output(opt, "Binary file ", 12);
1793                                 output_color(opt, gs->name, strlen(gs->name),
1794                                              opt->colors[GREP_COLOR_FILENAME]);
1795                                 opt->output(opt, " matches\n", 9);
1796                                 return 1;
1797                         }
1798                         /* Hit at this line.  If we haven't shown the
1799                          * pre-context lines, we would need to show them.
1800                          */
1801                         if (opt->pre_context || opt->funcbody)
1802                                 show_pre_context(opt, gs, bol, eol, lno);
1803                         else if (opt->funcname)
1804                                 show_funcname_line(opt, gs, bol, lno);
1805                         cno = opt->invert ? icol : col;
1806                         if (cno < 0) {
1807                                 /*
1808                                  * A negative cno indicates that there was no
1809                                  * match on the line. We are thus inverted and
1810                                  * being asked to show all lines that _don't_
1811                                  * match a given expression. Therefore, set cno
1812                                  * to 0 to suggest the whole line matches.
1813                                  */
1814                                 cno = 0;
1815                         }
1816                         show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1817                         last_hit = lno;
1818                         if (opt->funcbody)
1819                                 show_function = 1;
1820                         goto next_line;
1821                 }
1822                 if (show_function && (!peek_bol || peek_bol < bol)) {
1823                         unsigned long peek_left = left;
1824                         char *peek_eol = eol;
1825
1826                         /*
1827                          * Trailing empty lines are not interesting.
1828                          * Peek past them to see if they belong to the
1829                          * body of the current function.
1830                          */
1831                         peek_bol = bol;
1832                         while (is_empty_line(peek_bol, peek_eol)) {
1833                                 peek_bol = peek_eol + 1;
1834                                 peek_eol = end_of_line(peek_bol, &peek_left);
1835                         }
1836
1837                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1838                                 show_function = 0;
1839                 }
1840                 if (show_function ||
1841                     (last_hit && lno <= last_hit + opt->post_context)) {
1842                         /* If the last hit is within the post context,
1843                          * we need to show this line.
1844                          */
1845                         show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1846                 }
1847
1848         next_line:
1849                 bol = eol + 1;
1850                 if (!left)
1851                         break;
1852                 left--;
1853                 lno++;
1854         }
1855
1856         if (collect_hits)
1857                 return 0;
1858
1859         if (opt->status_only)
1860                 return opt->unmatch_name_only;
1861         if (opt->unmatch_name_only) {
1862                 /* We did not see any hit, so we want to show this */
1863                 show_name(opt, gs->name);
1864                 return 1;
1865         }
1866
1867         xdiff_clear_find_func(&xecfg);
1868         opt->priv = NULL;
1869
1870         /* NEEDSWORK:
1871          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1872          * which feels mostly useless but sometimes useful.  Maybe
1873          * make it another option?  For now suppress them.
1874          */
1875         if (opt->count && count) {
1876                 char buf[32];
1877                 if (opt->pathname) {
1878                         output_color(opt, gs->name, strlen(gs->name),
1879                                      opt->colors[GREP_COLOR_FILENAME]);
1880                         output_sep(opt, ':');
1881                 }
1882                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1883                 opt->output(opt, buf, strlen(buf));
1884                 return 1;
1885         }
1886         return !!last_hit;
1887 }
1888
1889 static void clr_hit_marker(struct grep_expr *x)
1890 {
1891         /* All-hit markers are meaningful only at the very top level
1892          * OR node.
1893          */
1894         while (1) {
1895                 x->hit = 0;
1896                 if (x->node != GREP_NODE_OR)
1897                         return;
1898                 x->u.binary.left->hit = 0;
1899                 x = x->u.binary.right;
1900         }
1901 }
1902
1903 static int chk_hit_marker(struct grep_expr *x)
1904 {
1905         /* Top level nodes have hit markers.  See if they all are hits */
1906         while (1) {
1907                 if (x->node != GREP_NODE_OR)
1908                         return x->hit;
1909                 if (!x->u.binary.left->hit)
1910                         return 0;
1911                 x = x->u.binary.right;
1912         }
1913 }
1914
1915 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1916 {
1917         /*
1918          * we do not have to do the two-pass grep when we do not check
1919          * buffer-wide "all-match".
1920          */
1921         if (!opt->all_match)
1922                 return grep_source_1(opt, gs, 0);
1923
1924         /* Otherwise the toplevel "or" terms hit a bit differently.
1925          * We first clear hit markers from them.
1926          */
1927         clr_hit_marker(opt->pattern_expression);
1928         grep_source_1(opt, gs, 1);
1929
1930         if (!chk_hit_marker(opt->pattern_expression))
1931                 return 0;
1932
1933         return grep_source_1(opt, gs, 0);
1934 }
1935
1936 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1937 {
1938         struct grep_source gs;
1939         int r;
1940
1941         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1942         gs.buf = buf;
1943         gs.size = size;
1944
1945         r = grep_source(opt, &gs);
1946
1947         grep_source_clear(&gs);
1948         return r;
1949 }
1950
1951 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1952                       const char *name, const char *path,
1953                       const void *identifier)
1954 {
1955         gs->type = type;
1956         gs->name = xstrdup_or_null(name);
1957         gs->path = xstrdup_or_null(path);
1958         gs->buf = NULL;
1959         gs->size = 0;
1960         gs->driver = NULL;
1961
1962         switch (type) {
1963         case GREP_SOURCE_FILE:
1964                 gs->identifier = xstrdup(identifier);
1965                 break;
1966         case GREP_SOURCE_OID:
1967                 gs->identifier = oiddup(identifier);
1968                 break;
1969         case GREP_SOURCE_BUF:
1970                 gs->identifier = NULL;
1971                 break;
1972         }
1973 }
1974
1975 void grep_source_clear(struct grep_source *gs)
1976 {
1977         FREE_AND_NULL(gs->name);
1978         FREE_AND_NULL(gs->path);
1979         FREE_AND_NULL(gs->identifier);
1980         grep_source_clear_data(gs);
1981 }
1982
1983 void grep_source_clear_data(struct grep_source *gs)
1984 {
1985         switch (gs->type) {
1986         case GREP_SOURCE_FILE:
1987         case GREP_SOURCE_OID:
1988                 FREE_AND_NULL(gs->buf);
1989                 gs->size = 0;
1990                 break;
1991         case GREP_SOURCE_BUF:
1992                 /* leave user-provided buf intact */
1993                 break;
1994         }
1995 }
1996
1997 static int grep_source_load_oid(struct grep_source *gs)
1998 {
1999         enum object_type type;
2000
2001         gs->buf = read_object_file(gs->identifier, &type, &gs->size);
2002         if (!gs->buf)
2003                 return error(_("'%s': unable to read %s"),
2004                              gs->name,
2005                              oid_to_hex(gs->identifier));
2006         return 0;
2007 }
2008
2009 static int grep_source_load_file(struct grep_source *gs)
2010 {
2011         const char *filename = gs->identifier;
2012         struct stat st;
2013         char *data;
2014         size_t size;
2015         int i;
2016
2017         if (lstat(filename, &st) < 0) {
2018         err_ret:
2019                 if (errno != ENOENT)
2020                         error_errno(_("failed to stat '%s'"), filename);
2021                 return -1;
2022         }
2023         if (!S_ISREG(st.st_mode))
2024                 return -1;
2025         size = xsize_t(st.st_size);
2026         i = open(filename, O_RDONLY);
2027         if (i < 0)
2028                 goto err_ret;
2029         data = xmallocz(size);
2030         if (st.st_size != read_in_full(i, data, size)) {
2031                 error_errno(_("'%s': short read"), filename);
2032                 close(i);
2033                 free(data);
2034                 return -1;
2035         }
2036         close(i);
2037
2038         gs->buf = data;
2039         gs->size = size;
2040         return 0;
2041 }
2042
2043 static int grep_source_load(struct grep_source *gs)
2044 {
2045         if (gs->buf)
2046                 return 0;
2047
2048         switch (gs->type) {
2049         case GREP_SOURCE_FILE:
2050                 return grep_source_load_file(gs);
2051         case GREP_SOURCE_OID:
2052                 return grep_source_load_oid(gs);
2053         case GREP_SOURCE_BUF:
2054                 return gs->buf ? 0 : -1;
2055         }
2056         BUG("invalid grep_source type to load");
2057 }
2058
2059 void grep_source_load_driver(struct grep_source *gs,
2060                              struct index_state *istate)
2061 {
2062         if (gs->driver)
2063                 return;
2064
2065         grep_attr_lock();
2066         if (gs->path)
2067                 gs->driver = userdiff_find_by_path(istate, gs->path);
2068         if (!gs->driver)
2069                 gs->driver = userdiff_find_by_name("default");
2070         grep_attr_unlock();
2071 }
2072
2073 static int grep_source_is_binary(struct grep_source *gs,
2074                                  struct index_state *istate)
2075 {
2076         grep_source_load_driver(gs, istate);
2077         if (gs->driver->binary != -1)
2078                 return gs->driver->binary;
2079
2080         if (!grep_source_load(gs))
2081                 return buffer_is_binary(gs->buf, gs->size);
2082
2083         return 0;
2084 }