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 struct grep_opt grep_defaults;
  18
  19 static const char *color_grep_slots[] = {
  20         [GREP_COLOR_CONTEXT]        = "context",
  21         [GREP_COLOR_FILENAME]       = "filename",
  22         [GREP_COLOR_FUNCTION]       = "function",
  23         [GREP_COLOR_LINENO]         = "lineNumber",
  24         [GREP_COLOR_COLUMNNO]       = "column",
  25         [GREP_COLOR_MATCH_CONTEXT]  = "matchContext",
  26         [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
  27         [GREP_COLOR_SELECTED]       = "selected",
  28         [GREP_COLOR_SEP]            = "separator",
  29 };
  30
  31 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
  32 {
  33         fwrite(buf, size, 1, stdout);
  34 }
  35
  36 static void color_set(char *dst, const char *color_bytes)
  37 {
  38         xsnprintf(dst, COLOR_MAXLEN, "%s", color_bytes);
  39 }
  40
  41 /*
  42  * Initialize the grep_defaults template with hardcoded defaults.
  43  * We could let the compiler do this, but without C99 initializers
  44  * the code gets unwieldy and unreadable, so...
  45  */
  46 void init_grep_defaults(struct repository *repo)
  47 {
  48         struct grep_opt *opt = &grep_defaults;
  49         static int run_once;
  50
  51         if (run_once)
  52                 return;
  53         run_once++;
  54
  55         memset(opt, 0, sizeof(*opt));
  56         opt->repo = repo;
  57         opt->relative = 1;
  58         opt->pathname = 1;
  59         opt->max_depth = -1;
  60         opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  61         color_set(opt->colors[GREP_COLOR_CONTEXT], "");
  62         color_set(opt->colors[GREP_COLOR_FILENAME], "");
  63         color_set(opt->colors[GREP_COLOR_FUNCTION], "");
  64         color_set(opt->colors[GREP_COLOR_LINENO], "");
  65         color_set(opt->colors[GREP_COLOR_COLUMNNO], "");
  66         color_set(opt->colors[GREP_COLOR_MATCH_CONTEXT], GIT_COLOR_BOLD_RED);
  67         color_set(opt->colors[GREP_COLOR_MATCH_SELECTED], GIT_COLOR_BOLD_RED);
  68         color_set(opt->colors[GREP_COLOR_SELECTED], "");
  69         color_set(opt->colors[GREP_COLOR_SEP], GIT_COLOR_CYAN);
  70         opt->only_matching = 0;
  71         opt->color = -1;
  72         opt->output = std_output;
  73 }
  74
  75 static int parse_pattern_type_arg(const char *opt, const char *arg)
  76 {
  77         if (!strcmp(arg, "default"))
  78                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  79         else if (!strcmp(arg, "basic"))
  80                 return GREP_PATTERN_TYPE_BRE;
  81         else if (!strcmp(arg, "extended"))
  82                 return GREP_PATTERN_TYPE_ERE;
  83         else if (!strcmp(arg, "fixed"))
  84                 return GREP_PATTERN_TYPE_FIXED;
  85         else if (!strcmp(arg, "perl"))
  86                 return GREP_PATTERN_TYPE_PCRE;
  87         die("bad %s argument: %s", opt, arg);
  88 }
  89
  90 define_list_config_array_extra(color_grep_slots, {"match"});
  91
  92 /*
  93  * Read the configuration file once and store it in
  94  * the grep_defaults template.
  95  */
  96 int grep_config(const char *var, const char *value, void *cb)
  97 {
  98         struct grep_opt *opt = &grep_defaults;
  99         const char *slot;
 100
 101         if (userdiff_config(var, value) < 0)
 102                 return -1;
 103
 104         if (!strcmp(var, "grep.extendedregexp")) {
 105                 opt->extended_regexp_option = git_config_bool(var, value);
 106                 return 0;
 107         }
 108
 109         if (!strcmp(var, "grep.patterntype")) {
 110                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
 111                 return 0;
 112         }
 113
 114         if (!strcmp(var, "grep.linenumber")) {
 115                 opt->linenum = git_config_bool(var, value);
 116                 return 0;
 117         }
 118         if (!strcmp(var, "grep.column")) {
 119                 opt->columnnum = git_config_bool(var, value);
 120                 return 0;
 121         }
 122
 123         if (!strcmp(var, "grep.fullname")) {
 124                 opt->relative = !git_config_bool(var, value);
 125                 return 0;
 126         }
 127
 128         if (!strcmp(var, "color.grep"))
 129                 opt->color = git_config_colorbool(var, value);
 130         if (!strcmp(var, "color.grep.match")) {
 131                 if (grep_config("color.grep.matchcontext", value, cb) < 0)
 132                         return -1;
 133                 if (grep_config("color.grep.matchselected", value, cb) < 0)
 134                         return -1;
 135         } else if (skip_prefix(var, "color.grep.", &slot)) {
 136                 int i = LOOKUP_CONFIG(color_grep_slots, slot);
 137                 char *color;
 138
 139                 if (i < 0)
 140                         return -1;
 141                 color = opt->colors[i];
 142                 if (!value)
 143                         return config_error_nonbool(var);
 144                 return color_parse(value, color);
 145         }
 146         return 0;
 147 }
 148
 149 /*
 150  * Initialize one instance of grep_opt and copy the
 151  * default values from the template we read the configuration
 152  * information in an earlier call to git_config(grep_config).
 153  */
 154 void grep_init(struct grep_opt *opt, struct repository *repo, const char *prefix)
 155 {
 156         struct grep_opt *def = &grep_defaults;
 157         int i;
 158
 159         memset(opt, 0, sizeof(*opt));
 160         opt->repo = repo;
 161         opt->prefix = prefix;
 162         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 163         opt->pattern_tail = &opt->pattern_list;
 164         opt->header_tail = &opt->header_list;
 165
 166         opt->only_matching = def->only_matching;
 167         opt->color = def->color;
 168         opt->extended_regexp_option = def->extended_regexp_option;
 169         opt->pattern_type_option = def->pattern_type_option;
 170         opt->linenum = def->linenum;
 171         opt->columnnum = def->columnnum;
 172         opt->max_depth = def->max_depth;
 173         opt->pathname = def->pathname;
 174         opt->relative = def->relative;
 175         opt->output = def->output;
 176
 177         for (i = 0; i < NR_GREP_COLORS; i++)
 178                 color_set(opt->colors[i], def->colors[i]);
 179 }
 180
 181 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 182 {
 183         /*
 184          * When committing to the pattern type by setting the relevant
 185          * fields in grep_opt it's generally not necessary to zero out
 186          * the fields we're not choosing, since they won't have been
 187          * set by anything. The extended_regexp_option field is the
 188          * only exception to this.
 189          *
 190          * This is because in the process of parsing grep.patternType
 191          * & grep.extendedRegexp we set opt->pattern_type_option and
 192          * opt->extended_regexp_option, respectively. We then
 193          * internally use opt->extended_regexp_option to see if we're
 194          * compiling an ERE. It must be unset if that's not actually
 195          * the case.
 196          */
 197         if (pattern_type != GREP_PATTERN_TYPE_ERE &&
 198             opt->extended_regexp_option)
 199                 opt->extended_regexp_option = 0;
 200
 201         switch (pattern_type) {
 202         case GREP_PATTERN_TYPE_UNSPECIFIED:
 203                 /* fall through */
 204
 205         case GREP_PATTERN_TYPE_BRE:
 206                 break;
 207
 208         case GREP_PATTERN_TYPE_ERE:
 209                 opt->extended_regexp_option = 1;
 210                 break;
 211
 212         case GREP_PATTERN_TYPE_FIXED:
 213                 opt->fixed = 1;
 214                 break;
 215
 216         case GREP_PATTERN_TYPE_PCRE:
 217 #ifdef USE_LIBPCRE2
 218                 opt->pcre2 = 1;
 219 #else
 220                 /*
 221                  * It's important that pcre1 always be assigned to
 222                  * even when there's no USE_LIBPCRE* defined. We still
 223                  * call the PCRE stub function, it just dies with
 224                  * "cannot use Perl-compatible regexes[...]".
 225                  */
 226                 opt->pcre1 = 1;
 227 #endif
 228                 break;
 229         }
 230 }
 231
 232 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 233 {
 234         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 235                 grep_set_pattern_type_option(pattern_type, opt);
 236         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 237                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 238         else if (opt->extended_regexp_option)
 239                 /*
 240                  * This branch *must* happen after setting from the
 241                  * opt->pattern_type_option above, we don't want
 242                  * grep.extendedRegexp to override grep.patternType!
 243                  */
 244                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 245 }
 246
 247 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 248                                         const char *origin, int no,
 249                                         enum grep_pat_token t,
 250                                         enum grep_header_field field)
 251 {
 252         struct grep_pat *p = xcalloc(1, sizeof(*p));
 253         p->pattern = xmemdupz(pat, patlen);
 254         p->patternlen = patlen;
 255         p->origin = origin;
 256         p->no = no;
 257         p->token = t;
 258         p->field = field;
 259         return p;
 260 }
 261
 262 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 263 {
 264         **tail = p;
 265         *tail = &p->next;
 266         p->next = NULL;
 267
 268         switch (p->token) {
 269         case GREP_PATTERN: /* atom */
 270         case GREP_PATTERN_HEAD:
 271         case GREP_PATTERN_BODY:
 272                 for (;;) {
 273                         struct grep_pat *new_pat;
 274                         size_t len = 0;
 275                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 276                         while (++len <= p->patternlen) {
 277                                 if (*(--cp) == '\n') {
 278                                         nl = cp;
 279                                         break;
 280                                 }
 281                         }
 282                         if (!nl)
 283                                 break;
 284                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 285                                                   p->no, p->token, p->field);
 286                         new_pat->next = p->next;
 287                         if (!p->next)
 288                                 *tail = &new_pat->next;
 289                         p->next = new_pat;
 290                         *nl = '\0';
 291                         p->patternlen -= len;
 292                 }
 293                 break;
 294         default:
 295                 break;
 296         }
 297 }
 298
 299 void append_header_grep_pattern(struct grep_opt *opt,
 300                                 enum grep_header_field field, const char *pat)
 301 {
 302         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 303                                              GREP_PATTERN_HEAD, field);
 304         if (field == GREP_HEADER_REFLOG)
 305                 opt->use_reflog_filter = 1;
 306         do_append_grep_pat(&opt->header_tail, p);
 307 }
 308
 309 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 310                          const char *origin, int no, enum grep_pat_token t)
 311 {
 312         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 313 }
 314
 315 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 316                      const char *origin, int no, enum grep_pat_token t)
 317 {
 318         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 319         do_append_grep_pat(&opt->pattern_tail, p);
 320 }
 321
 322 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 323 {
 324         struct grep_pat *pat;
 325         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 326         *ret = *opt;
 327
 328         ret->pattern_list = NULL;
 329         ret->pattern_tail = &ret->pattern_list;
 330
 331         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 332         {
 333                 if(pat->token == GREP_PATTERN_HEAD)
 334                         append_header_grep_pattern(ret, pat->field,
 335                                                    pat->pattern);
 336                 else
 337                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 338                                         pat->origin, pat->no, pat->token);
 339         }
 340
 341         return ret;
 342 }
 343
 344 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 345                 const char *error)
 346 {
 347         char where[1024];
 348
 349         if (p->no)
 350                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 351         else if (p->origin)
 352                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 353         else
 354                 where[0] = 0;
 355
 356         die("%s'%s': %s", where, p->pattern, error);
 357 }
 358
 359 static int is_fixed(const char *s, size_t len)
 360 {
 361         size_t i;
 362
 363         for (i = 0; i < len; i++) {
 364                 if (is_regex_special(s[i]))
 365                         return 0;
 366         }
 367
 368         return 1;
 369 }
 370
 371 #ifdef USE_LIBPCRE1
 372 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 373 {
 374         const char *error;
 375         int erroffset;
 376         int options = PCRE_MULTILINE;
 377
 378         if (opt->ignore_case) {
 379                 if (!opt->ignore_locale && has_non_ascii(p->pattern))
 380                         p->pcre1_tables = pcre_maketables();
 381                 options |= PCRE_CASELESS;
 382         }
 383         if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern))
 384                 options |= PCRE_UTF8;
 385
 386         p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 387                                       p->pcre1_tables);
 388         if (!p->pcre1_regexp)
 389                 compile_regexp_failed(p, error);
 390
 391         p->pcre1_extra_info = pcre_study(p->pcre1_regexp, GIT_PCRE_STUDY_JIT_COMPILE, &error);
 392         if (!p->pcre1_extra_info && error)
 393                 die("%s", error);
 394
 395 #ifdef GIT_PCRE1_USE_JIT
 396         pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
 397 #endif
 398 }
 399
 400 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 401                 regmatch_t *match, int eflags)
 402 {
 403         int ovector[30], ret, flags = 0;
 404
 405         if (eflags & REG_NOTBOL)
 406                 flags |= PCRE_NOTBOL;
 407
 408         ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 409                         eol - line, 0, flags, ovector,
 410                         ARRAY_SIZE(ovector));
 411
 412         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 413                 die("pcre_exec failed with error code %d", ret);
 414         if (ret > 0) {
 415                 ret = 0;
 416                 match->rm_so = ovector[0];
 417                 match->rm_eo = ovector[1];
 418         }
 419
 420         return ret;
 421 }
 422
 423 static void free_pcre1_regexp(struct grep_pat *p)
 424 {
 425         pcre_free(p->pcre1_regexp);
 426 #ifdef GIT_PCRE1_USE_JIT
 427         if (p->pcre1_jit_on)
 428                 pcre_free_study(p->pcre1_extra_info);
 429         else
 430 #endif
 431                 pcre_free(p->pcre1_extra_info);
 432         pcre_free((void *)p->pcre1_tables);
 433 }
 434 #else /* !USE_LIBPCRE1 */
 435 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 436 {
 437         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 438 }
 439
 440 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 441                 regmatch_t *match, int eflags)
 442 {
 443         return 1;
 444 }
 445
 446 static void free_pcre1_regexp(struct grep_pat *p)
 447 {
 448 }
 449 #endif /* !USE_LIBPCRE1 */
 450
 451 #ifdef USE_LIBPCRE2
 452 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 453 {
 454         int error;
 455         PCRE2_UCHAR errbuf[256];
 456         PCRE2_SIZE erroffset;
 457         int options = PCRE2_MULTILINE;
 458         const uint8_t *character_tables = NULL;
 459         int jitret;
 460         int patinforet;
 461         size_t jitsizearg;
 462
 463         assert(opt->pcre2);
 464
 465         p->pcre2_compile_context = NULL;
 466
 467         if (opt->ignore_case) {
 468                 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
 469                         character_tables = pcre2_maketables(NULL);
 470                         p->pcre2_compile_context = pcre2_compile_context_create(NULL);
 471                         pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
 472                 }
 473                 options |= PCRE2_CASELESS;
 474         }
 475         if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern))
 476                 options |= PCRE2_UTF;
 477
 478         p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
 479                                          p->patternlen, options, &error, &erroffset,
 480                                          p->pcre2_compile_context);
 481
 482         if (p->pcre2_pattern) {
 483                 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
 484                 if (!p->pcre2_match_data)
 485                         die("Couldn't allocate PCRE2 match data");
 486         } else {
 487                 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
 488                 compile_regexp_failed(p, (const char *)&errbuf);
 489         }
 490
 491         pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
 492         if (p->pcre2_jit_on) {
 493                 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
 494                 if (jitret)
 495                         die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
 496
 497                 /*
 498                  * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
 499                  * tells us whether the library itself supports JIT,
 500                  * but to see whether we're going to be actually using
 501                  * JIT we need to extract PCRE2_INFO_JITSIZE from the
 502                  * pattern *after* we do pcre2_jit_compile() above.
 503                  *
 504                  * This is because if the pattern contains the
 505                  * (*NO_JIT) verb (see pcre2syntax(3))
 506                  * pcre2_jit_compile() will exit early with 0. If we
 507                  * then proceed to call pcre2_jit_match() further down
 508                  * the line instead of pcre2_match() we'll either
 509                  * segfault (pre PCRE 10.31) or run into a fatal error
 510                  * (post PCRE2 10.31)
 511                  */
 512                 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
 513                 if (patinforet)
 514                         BUG("pcre2_pattern_info() failed: %d", patinforet);
 515                 if (jitsizearg == 0) {
 516                         p->pcre2_jit_on = 0;
 517                         return;
 518                 }
 519         }
 520 }
 521
 522 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 523                 regmatch_t *match, int eflags)
 524 {
 525         int ret, flags = 0;
 526         PCRE2_SIZE *ovector;
 527         PCRE2_UCHAR errbuf[256];
 528
 529         if (eflags & REG_NOTBOL)
 530                 flags |= PCRE2_NOTBOL;
 531
 532         if (p->pcre2_jit_on)
 533                 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 534                                       eol - line, 0, flags, p->pcre2_match_data,
 535                                       NULL);
 536         else
 537                 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 538                                   eol - line, 0, flags, p->pcre2_match_data,
 539                                   NULL);
 540
 541         if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 542                 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 543                 die("%s failed with error code %d: %s",
 544                     (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 545                     errbuf);
 546         }
 547         if (ret > 0) {
 548                 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 549                 ret = 0;
 550                 match->rm_so = (int)ovector[0];
 551                 match->rm_eo = (int)ovector[1];
 552         }
 553
 554         return ret;
 555 }
 556
 557 static void free_pcre2_pattern(struct grep_pat *p)
 558 {
 559         pcre2_compile_context_free(p->pcre2_compile_context);
 560         pcre2_code_free(p->pcre2_pattern);
 561         pcre2_match_data_free(p->pcre2_match_data);
 562 }
 563 #else /* !USE_LIBPCRE2 */
 564 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 565 {
 566         /*
 567          * Unreachable until USE_LIBPCRE2 becomes synonymous with
 568          * USE_LIBPCRE. See the sibling comment in
 569          * grep_set_pattern_type_option().
 570          */
 571         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 572 }
 573
 574 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 575                 regmatch_t *match, int eflags)
 576 {
 577         return 1;
 578 }
 579
 580 static void free_pcre2_pattern(struct grep_pat *p)
 581 {
 582 }
 583
 584 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 585 {
 586         struct strbuf sb = STRBUF_INIT;
 587         int err;
 588         int regflags = 0;
 589
 590         basic_regex_quote_buf(&sb, p->pattern);
 591         if (opt->ignore_case)
 592                 regflags |= REG_ICASE;
 593         err = regcomp(&p->regexp, sb.buf, regflags);
 594         if (opt->debug)
 595                 fprintf(stderr, "fixed %s\n", sb.buf);
 596         strbuf_release(&sb);
 597         if (err) {
 598                 char errbuf[1024];
 599                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 600                 compile_regexp_failed(p, errbuf);
 601         }
 602 }
 603 #endif /* !USE_LIBPCRE2 */
 604
 605 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 606 {
 607         int err;
 608         int regflags = REG_NEWLINE;
 609
 610         p->word_regexp = opt->word_regexp;
 611         p->ignore_case = opt->ignore_case;
 612         p->fixed = opt->fixed;
 613
 614         if (memchr(p->pattern, 0, p->patternlen) && !opt->pcre2)
 615                 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
 616
 617         p->is_fixed = is_fixed(p->pattern, p->patternlen);
 618         if (p->fixed || p->is_fixed) {
 619 #ifdef USE_LIBPCRE2
 620                 opt->pcre2 = 1;
 621                 if (p->is_fixed) {
 622                         compile_pcre2_pattern(p, opt);
 623                 } else {
 624                         /*
 625                          * E.g. t7811-grep-open.sh relies on the
 626                          * pattern being restored.
 627                          */
 628                         char *old_pattern = p->pattern;
 629                         size_t old_patternlen = p->patternlen;
 630                         struct strbuf sb = STRBUF_INIT;
 631
 632                         /*
 633                          * There is the PCRE2_LITERAL flag, but it's
 634                          * only in PCRE v2 10.30 and later. Needing to
 635                          * ifdef our way around that and dealing with
 636                          * it + PCRE2_MULTILINE being an error is more
 637                          * complex than just quoting this ourselves.
 638                         */
 639                         strbuf_add(&sb, "\\Q", 2);
 640                         strbuf_add(&sb, p->pattern, p->patternlen);
 641                         strbuf_add(&sb, "\\E", 2);
 642
 643                         p->pattern = sb.buf;
 644                         p->patternlen = sb.len;
 645                         compile_pcre2_pattern(p, opt);
 646                         p->pattern = old_pattern;
 647                         p->patternlen = old_patternlen;
 648                         strbuf_release(&sb);
 649                 }
 650 #else /* !USE_LIBPCRE2 */
 651                 compile_fixed_regexp(p, opt);
 652 #endif /* !USE_LIBPCRE2 */
 653                 return;
 654         }
 655
 656         if (opt->pcre2) {
 657                 compile_pcre2_pattern(p, opt);
 658                 return;
 659         }
 660
 661         if (opt->pcre1) {
 662                 compile_pcre1_regexp(p, opt);
 663                 return;
 664         }
 665
 666         if (p->ignore_case)
 667                 regflags |= REG_ICASE;
 668         if (opt->extended_regexp_option)
 669                 regflags |= REG_EXTENDED;
 670         err = regcomp(&p->regexp, p->pattern, regflags);
 671         if (err) {
 672                 char errbuf[1024];
 673                 regerror(err, &p->regexp, errbuf, 1024);
 674                 compile_regexp_failed(p, errbuf);
 675         }
 676 }
 677
 678 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 679 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 680 {
 681         struct grep_pat *p;
 682         struct grep_expr *x;
 683
 684         p = *list;
 685         if (!p)
 686                 return NULL;
 687         switch (p->token) {
 688         case GREP_PATTERN: /* atom */
 689         case GREP_PATTERN_HEAD:
 690         case GREP_PATTERN_BODY:
 691                 x = xcalloc(1, sizeof (struct grep_expr));
 692                 x->node = GREP_NODE_ATOM;
 693                 x->u.atom = p;
 694                 *list = p->next;
 695                 return x;
 696         case GREP_OPEN_PAREN:
 697                 *list = p->next;
 698                 x = compile_pattern_or(list);
 699                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 700                         die("unmatched parenthesis");
 701                 *list = (*list)->next;
 702                 return x;
 703         default:
 704                 return NULL;
 705         }
 706 }
 707
 708 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 709 {
 710         struct grep_pat *p;
 711         struct grep_expr *x;
 712
 713         p = *list;
 714         if (!p)
 715                 return NULL;
 716         switch (p->token) {
 717         case GREP_NOT:
 718                 if (!p->next)
 719                         die("--not not followed by pattern expression");
 720                 *list = p->next;
 721                 x = xcalloc(1, sizeof (struct grep_expr));
 722                 x->node = GREP_NODE_NOT;
 723                 x->u.unary = compile_pattern_not(list);
 724                 if (!x->u.unary)
 725                         die("--not followed by non pattern expression");
 726                 return x;
 727         default:
 728                 return compile_pattern_atom(list);
 729         }
 730 }
 731
 732 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 733 {
 734         struct grep_pat *p;
 735         struct grep_expr *x, *y, *z;
 736
 737         x = compile_pattern_not(list);
 738         p = *list;
 739         if (p && p->token == GREP_AND) {
 740                 if (!p->next)
 741                         die("--and not followed by pattern expression");
 742                 *list = p->next;
 743                 y = compile_pattern_and(list);
 744                 if (!y)
 745                         die("--and not followed by pattern expression");
 746                 z = xcalloc(1, sizeof (struct grep_expr));
 747                 z->node = GREP_NODE_AND;
 748                 z->u.binary.left = x;
 749                 z->u.binary.right = y;
 750                 return z;
 751         }
 752         return x;
 753 }
 754
 755 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 756 {
 757         struct grep_pat *p;
 758         struct grep_expr *x, *y, *z;
 759
 760         x = compile_pattern_and(list);
 761         p = *list;
 762         if (x && p && p->token != GREP_CLOSE_PAREN) {
 763                 y = compile_pattern_or(list);
 764                 if (!y)
 765                         die("not a pattern expression %s", p->pattern);
 766                 z = xcalloc(1, sizeof (struct grep_expr));
 767                 z->node = GREP_NODE_OR;
 768                 z->u.binary.left = x;
 769                 z->u.binary.right = y;
 770                 return z;
 771         }
 772         return x;
 773 }
 774
 775 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 776 {
 777         return compile_pattern_or(list);
 778 }
 779
 780 static void indent(int in)
 781 {
 782         while (in-- > 0)
 783                 fputc(' ', stderr);
 784 }
 785
 786 static void dump_grep_pat(struct grep_pat *p)
 787 {
 788         switch (p->token) {
 789         case GREP_AND: fprintf(stderr, "*and*"); break;
 790         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 791         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 792         case GREP_NOT: fprintf(stderr, "*not*"); break;
 793         case GREP_OR: fprintf(stderr, "*or*"); break;
 794
 795         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 796         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 797         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 798         }
 799
 800         switch (p->token) {
 801         default: break;
 802         case GREP_PATTERN_HEAD:
 803                 fprintf(stderr, "<head %d>", p->field); break;
 804         case GREP_PATTERN_BODY:
 805                 fprintf(stderr, "<body>"); break;
 806         }
 807         switch (p->token) {
 808         default: break;
 809         case GREP_PATTERN_HEAD:
 810         case GREP_PATTERN_BODY:
 811         case GREP_PATTERN:
 812                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 813                 break;
 814         }
 815         fputc('\n', stderr);
 816 }
 817
 818 static void dump_grep_expression_1(struct grep_expr *x, int in)
 819 {
 820         indent(in);
 821         switch (x->node) {
 822         case GREP_NODE_TRUE:
 823                 fprintf(stderr, "true\n");
 824                 break;
 825         case GREP_NODE_ATOM:
 826                 dump_grep_pat(x->u.atom);
 827                 break;
 828         case GREP_NODE_NOT:
 829                 fprintf(stderr, "(not\n");
 830                 dump_grep_expression_1(x->u.unary, in+1);
 831                 indent(in);
 832                 fprintf(stderr, ")\n");
 833                 break;
 834         case GREP_NODE_AND:
 835                 fprintf(stderr, "(and\n");
 836                 dump_grep_expression_1(x->u.binary.left, in+1);
 837                 dump_grep_expression_1(x->u.binary.right, in+1);
 838                 indent(in);
 839                 fprintf(stderr, ")\n");
 840                 break;
 841         case GREP_NODE_OR:
 842                 fprintf(stderr, "(or\n");
 843                 dump_grep_expression_1(x->u.binary.left, in+1);
 844                 dump_grep_expression_1(x->u.binary.right, in+1);
 845                 indent(in);
 846                 fprintf(stderr, ")\n");
 847                 break;
 848         }
 849 }
 850
 851 static void dump_grep_expression(struct grep_opt *opt)
 852 {
 853         struct grep_expr *x = opt->pattern_expression;
 854
 855         if (opt->all_match)
 856                 fprintf(stderr, "[all-match]\n");
 857         dump_grep_expression_1(x, 0);
 858         fflush(NULL);
 859 }
 860
 861 static struct grep_expr *grep_true_expr(void)
 862 {
 863         struct grep_expr *z = xcalloc(1, sizeof(*z));
 864         z->node = GREP_NODE_TRUE;
 865         return z;
 866 }
 867
 868 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 869 {
 870         struct grep_expr *z = xcalloc(1, sizeof(*z));
 871         z->node = GREP_NODE_OR;
 872         z->u.binary.left = left;
 873         z->u.binary.right = right;
 874         return z;
 875 }
 876
 877 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 878 {
 879         struct grep_pat *p;
 880         struct grep_expr *header_expr;
 881         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 882         enum grep_header_field fld;
 883
 884         if (!opt->header_list)
 885                 return NULL;
 886
 887         for (p = opt->header_list; p; p = p->next) {
 888                 if (p->token != GREP_PATTERN_HEAD)
 889                         BUG("a non-header pattern in grep header list.");
 890                 if (p->field < GREP_HEADER_FIELD_MIN ||
 891                     GREP_HEADER_FIELD_MAX <= p->field)
 892                         BUG("unknown header field %d", p->field);
 893                 compile_regexp(p, opt);
 894         }
 895
 896         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 897                 header_group[fld] = NULL;
 898
 899         for (p = opt->header_list; p; p = p->next) {
 900                 struct grep_expr *h;
 901                 struct grep_pat *pp = p;
 902
 903                 h = compile_pattern_atom(&pp);
 904                 if (!h || pp != p->next)
 905                         BUG("malformed header expr");
 906                 if (!header_group[p->field]) {
 907                         header_group[p->field] = h;
 908                         continue;
 909                 }
 910                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 911         }
 912
 913         header_expr = NULL;
 914
 915         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 916                 if (!header_group[fld])
 917                         continue;
 918                 if (!header_expr)
 919                         header_expr = grep_true_expr();
 920                 header_expr = grep_or_expr(header_group[fld], header_expr);
 921         }
 922         return header_expr;
 923 }
 924
 925 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 926 {
 927         struct grep_expr *z = x;
 928
 929         while (x) {
 930                 assert(x->node == GREP_NODE_OR);
 931                 if (x->u.binary.right &&
 932                     x->u.binary.right->node == GREP_NODE_TRUE) {
 933                         x->u.binary.right = y;
 934                         break;
 935                 }
 936                 x = x->u.binary.right;
 937         }
 938         return z;
 939 }
 940
 941 static void compile_grep_patterns_real(struct grep_opt *opt)
 942 {
 943         struct grep_pat *p;
 944         struct grep_expr *header_expr = prep_header_patterns(opt);
 945
 946         for (p = opt->pattern_list; p; p = p->next) {
 947                 switch (p->token) {
 948                 case GREP_PATTERN: /* atom */
 949                 case GREP_PATTERN_HEAD:
 950                 case GREP_PATTERN_BODY:
 951                         compile_regexp(p, opt);
 952                         break;
 953                 default:
 954                         opt->extended = 1;
 955                         break;
 956                 }
 957         }
 958
 959         if (opt->all_match || header_expr)
 960                 opt->extended = 1;
 961         else if (!opt->extended && !opt->debug)
 962                 return;
 963
 964         p = opt->pattern_list;
 965         if (p)
 966                 opt->pattern_expression = compile_pattern_expr(&p);
 967         if (p)
 968                 die("incomplete pattern expression: %s", p->pattern);
 969
 970         if (!header_expr)
 971                 return;
 972
 973         if (!opt->pattern_expression)
 974                 opt->pattern_expression = header_expr;
 975         else if (opt->all_match)
 976                 opt->pattern_expression = grep_splice_or(header_expr,
 977                                                          opt->pattern_expression);
 978         else
 979                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 980                                                        header_expr);
 981         opt->all_match = 1;
 982 }
 983
 984 void compile_grep_patterns(struct grep_opt *opt)
 985 {
 986         compile_grep_patterns_real(opt);
 987         if (opt->debug)
 988                 dump_grep_expression(opt);
 989 }
 990
 991 static void free_pattern_expr(struct grep_expr *x)
 992 {
 993         switch (x->node) {
 994         case GREP_NODE_TRUE:
 995         case GREP_NODE_ATOM:
 996                 break;
 997         case GREP_NODE_NOT:
 998                 free_pattern_expr(x->u.unary);
 999                 break;
1000         case GREP_NODE_AND:
1001         case GREP_NODE_OR:
1002                 free_pattern_expr(x->u.binary.left);
1003                 free_pattern_expr(x->u.binary.right);
1004                 break;
1005         }
1006         free(x);
1007 }
1008
1009 void free_grep_patterns(struct grep_opt *opt)
1010 {
1011         struct grep_pat *p, *n;
1012
1013         for (p = opt->pattern_list; p; p = n) {
1014                 n = p->next;
1015                 switch (p->token) {
1016                 case GREP_PATTERN: /* atom */
1017                 case GREP_PATTERN_HEAD:
1018                 case GREP_PATTERN_BODY:
1019                         if (p->pcre1_regexp)
1020                                 free_pcre1_regexp(p);
1021                         else if (p->pcre2_pattern)
1022                                 free_pcre2_pattern(p);
1023                         else
1024                                 regfree(&p->regexp);
1025                         free(p->pattern);
1026                         break;
1027                 default:
1028                         break;
1029                 }
1030                 free(p);
1031         }
1032
1033         if (!opt->extended)
1034                 return;
1035         free_pattern_expr(opt->pattern_expression);
1036 }
1037
1038 static char *end_of_line(char *cp, unsigned long *left)
1039 {
1040         unsigned long l = *left;
1041         while (l && *cp != '\n') {
1042                 l--;
1043                 cp++;
1044         }
1045         *left = l;
1046         return cp;
1047 }
1048
1049 static int word_char(char ch)
1050 {
1051         return isalnum(ch) || ch == '_';
1052 }
1053
1054 static void output_color(struct grep_opt *opt, const void *data, size_t size,
1055                          const char *color)
1056 {
1057         if (want_color(opt->color) && color && color[0]) {
1058                 opt->output(opt, color, strlen(color));
1059                 opt->output(opt, data, size);
1060                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1061         } else
1062                 opt->output(opt, data, size);
1063 }
1064
1065 static void output_sep(struct grep_opt *opt, char sign)
1066 {
1067         if (opt->null_following_name)
1068                 opt->output(opt, "\0", 1);
1069         else
1070                 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
1071 }
1072
1073 static void show_name(struct grep_opt *opt, const char *name)
1074 {
1075         output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1076         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1077 }
1078
1079 static int patmatch(struct grep_pat *p, char *line, char *eol,
1080                     regmatch_t *match, int eflags)
1081 {
1082         int hit;
1083
1084         if (p->pcre1_regexp)
1085                 hit = !pcre1match(p, line, eol, match, eflags);
1086         else if (p->pcre2_pattern)
1087                 hit = !pcre2match(p, line, eol, match, eflags);
1088         else
1089                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1090                                    eflags);
1091
1092         return hit;
1093 }
1094
1095 static int strip_timestamp(char *bol, char **eol_p)
1096 {
1097         char *eol = *eol_p;
1098         int ch;
1099
1100         while (bol < --eol) {
1101                 if (*eol != '>')
1102                         continue;
1103                 *eol_p = ++eol;
1104                 ch = *eol;
1105                 *eol = '\0';
1106                 return ch;
1107         }
1108         return 0;
1109 }
1110
1111 static struct {
1112         const char *field;
1113         size_t len;
1114 } header_field[] = {
1115         { "author ", 7 },
1116         { "committer ", 10 },
1117         { "reflog ", 7 },
1118 };
1119
1120 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1121                              enum grep_context ctx,
1122                              regmatch_t *pmatch, int eflags)
1123 {
1124         int hit = 0;
1125         int saved_ch = 0;
1126         const char *start = bol;
1127
1128         if ((p->token != GREP_PATTERN) &&
1129             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1130                 return 0;
1131
1132         if (p->token == GREP_PATTERN_HEAD) {
1133                 const char *field;
1134                 size_t len;
1135                 assert(p->field < ARRAY_SIZE(header_field));
1136                 field = header_field[p->field].field;
1137                 len = header_field[p->field].len;
1138                 if (strncmp(bol, field, len))
1139                         return 0;
1140                 bol += len;
1141                 switch (p->field) {
1142                 case GREP_HEADER_AUTHOR:
1143                 case GREP_HEADER_COMMITTER:
1144                         saved_ch = strip_timestamp(bol, &eol);
1145                         break;
1146                 default:
1147                         break;
1148                 }
1149         }
1150
1151  again:
1152         hit = patmatch(p, bol, eol, pmatch, eflags);
1153
1154         if (hit && p->word_regexp) {
1155                 if ((pmatch[0].rm_so < 0) ||
1156                     (eol - bol) < pmatch[0].rm_so ||
1157                     (pmatch[0].rm_eo < 0) ||
1158                     (eol - bol) < pmatch[0].rm_eo)
1159                         die("regexp returned nonsense");
1160
1161                 /* Match beginning must be either beginning of the
1162                  * line, or at word boundary (i.e. the last char must
1163                  * not be a word char).  Similarly, match end must be
1164                  * either end of the line, or at word boundary
1165                  * (i.e. the next char must not be a word char).
1166                  */
1167                 if ( ((pmatch[0].rm_so == 0) ||
1168                       !word_char(bol[pmatch[0].rm_so-1])) &&
1169                      ((pmatch[0].rm_eo == (eol-bol)) ||
1170                       !word_char(bol[pmatch[0].rm_eo])) )
1171                         ;
1172                 else
1173                         hit = 0;
1174
1175                 /* Words consist of at least one character. */
1176                 if (pmatch->rm_so == pmatch->rm_eo)
1177                         hit = 0;
1178
1179                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1180                         /* There could be more than one match on the
1181                          * line, and the first match might not be
1182                          * strict word match.  But later ones could be!
1183                          * Forward to the next possible start, i.e. the
1184                          * next position following a non-word char.
1185                          */
1186                         bol = pmatch[0].rm_so + bol + 1;
1187                         while (word_char(bol[-1]) && bol < eol)
1188                                 bol++;
1189                         eflags |= REG_NOTBOL;
1190                         if (bol < eol)
1191                                 goto again;
1192                 }
1193         }
1194         if (p->token == GREP_PATTERN_HEAD && saved_ch)
1195                 *eol = saved_ch;
1196         if (hit) {
1197                 pmatch[0].rm_so += bol - start;
1198                 pmatch[0].rm_eo += bol - start;
1199         }
1200         return hit;
1201 }
1202
1203 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x, char *bol,
1204                            char *eol, enum grep_context ctx, ssize_t *col,
1205                            ssize_t *icol, int collect_hits)
1206 {
1207         int h = 0;
1208
1209         if (!x)
1210                 die("Not a valid grep expression");
1211         switch (x->node) {
1212         case GREP_NODE_TRUE:
1213                 h = 1;
1214                 break;
1215         case GREP_NODE_ATOM:
1216                 {
1217                         regmatch_t tmp;
1218                         h = match_one_pattern(x->u.atom, bol, eol, ctx,
1219                                               &tmp, 0);
1220                         if (h && (*col < 0 || tmp.rm_so < *col))
1221                                 *col = tmp.rm_so;
1222                 }
1223                 break;
1224         case GREP_NODE_NOT:
1225                 /*
1226                  * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1227                  */
1228                 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1229                                      0);
1230                 break;
1231         case GREP_NODE_AND:
1232                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1233                                     icol, 0);
1234                 if (h || opt->columnnum) {
1235                         /*
1236                          * Don't short-circuit AND when given --column, since a
1237                          * NOT earlier in the tree may turn this into an OR. In
1238                          * this case, see the below comment.
1239                          */
1240                         h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1241                                              ctx, col, icol, 0);
1242                 }
1243                 break;
1244         case GREP_NODE_OR:
1245                 if (!(collect_hits || opt->columnnum)) {
1246                         /*
1247                          * Don't short-circuit OR when given --column (or
1248                          * collecting hits) to ensure we don't skip a later
1249                          * child that would produce an earlier match.
1250                          */
1251                         return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1252                                                 ctx, col, icol, 0) ||
1253                                 match_expr_eval(opt, x->u.binary.right, bol,
1254                                                 eol, ctx, col, icol, 0));
1255                 }
1256                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1257                                     icol, 0);
1258                 if (collect_hits)
1259                         x->u.binary.left->hit |= h;
1260                 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1261                                      icol, collect_hits);
1262                 break;
1263         default:
1264                 die("Unexpected node type (internal error) %d", x->node);
1265         }
1266         if (collect_hits)
1267                 x->hit |= h;
1268         return h;
1269 }
1270
1271 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1272                       enum grep_context ctx, ssize_t *col,
1273                       ssize_t *icol, int collect_hits)
1274 {
1275         struct grep_expr *x = opt->pattern_expression;
1276         return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1277 }
1278
1279 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1280                       ssize_t *col, ssize_t *icol,
1281                       enum grep_context ctx, int collect_hits)
1282 {
1283         struct grep_pat *p;
1284         int hit = 0;
1285
1286         if (opt->extended)
1287                 return match_expr(opt, bol, eol, ctx, col, icol,
1288                                   collect_hits);
1289
1290         /* we do not call with collect_hits without being extended */
1291         for (p = opt->pattern_list; p; p = p->next) {
1292                 regmatch_t tmp;
1293                 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1294                         hit |= 1;
1295                         if (!opt->columnnum) {
1296                                 /*
1297                                  * Without --column, any single match on a line
1298                                  * is enough to know that it needs to be
1299                                  * printed. With --column, scan _all_ patterns
1300                                  * to find the earliest.
1301                                  */
1302                                 break;
1303                         }
1304                         if (*col < 0 || tmp.rm_so < *col)
1305                                 *col = tmp.rm_so;
1306                 }
1307         }
1308         return hit;
1309 }
1310
1311 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1312                               enum grep_context ctx,
1313                               regmatch_t *pmatch, int eflags)
1314 {
1315         regmatch_t match;
1316
1317         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1318                 return 0;
1319         if (match.rm_so < 0 || match.rm_eo < 0)
1320                 return 0;
1321         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1322                 if (match.rm_so > pmatch->rm_so)
1323                         return 1;
1324                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1325                         return 1;
1326         }
1327         pmatch->rm_so = match.rm_so;
1328         pmatch->rm_eo = match.rm_eo;
1329         return 1;
1330 }
1331
1332 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1333                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1334 {
1335         struct grep_pat *p;
1336         int hit = 0;
1337
1338         pmatch->rm_so = pmatch->rm_eo = -1;
1339         if (bol < eol) {
1340                 for (p = opt->pattern_list; p; p = p->next) {
1341                         switch (p->token) {
1342                         case GREP_PATTERN: /* atom */
1343                         case GREP_PATTERN_HEAD:
1344                         case GREP_PATTERN_BODY:
1345                                 hit |= match_next_pattern(p, bol, eol, ctx,
1346                                                           pmatch, eflags);
1347                                 break;
1348                         default:
1349                                 break;
1350                         }
1351                 }
1352         }
1353         return hit;
1354 }
1355
1356 static void show_line_header(struct grep_opt *opt, const char *name,
1357                              unsigned lno, ssize_t cno, char sign)
1358 {
1359         if (opt->heading && opt->last_shown == 0) {
1360                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1361                 opt->output(opt, "\n", 1);
1362         }
1363         opt->last_shown = lno;
1364
1365         if (!opt->heading && opt->pathname) {
1366                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1367                 output_sep(opt, sign);
1368         }
1369         if (opt->linenum) {
1370                 char buf[32];
1371                 xsnprintf(buf, sizeof(buf), "%d", lno);
1372                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1373                 output_sep(opt, sign);
1374         }
1375         /*
1376          * Treat 'cno' as the 1-indexed offset from the start of a non-context
1377          * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1378          * being called with a context line.
1379          */
1380         if (opt->columnnum && cno) {
1381                 char buf[32];
1382                 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1383                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1384                 output_sep(opt, sign);
1385         }
1386 }
1387
1388 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1389                       const char *name, unsigned lno, ssize_t cno, char sign)
1390 {
1391         int rest = eol - bol;
1392         const char *match_color = NULL;
1393         const char *line_color = NULL;
1394
1395         if (opt->file_break && opt->last_shown == 0) {
1396                 if (opt->show_hunk_mark)
1397                         opt->output(opt, "\n", 1);
1398         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1399                 if (opt->last_shown == 0) {
1400                         if (opt->show_hunk_mark) {
1401                                 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1402                                 opt->output(opt, "\n", 1);
1403                         }
1404                 } else if (lno > opt->last_shown + 1) {
1405                         output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1406                         opt->output(opt, "\n", 1);
1407                 }
1408         }
1409         if (!opt->only_matching) {
1410                 /*
1411                  * In case the line we're being called with contains more than
1412                  * one match, leave printing each header to the loop below.
1413                  */
1414                 show_line_header(opt, name, lno, cno, sign);
1415         }
1416         if (opt->color || opt->only_matching) {
1417                 regmatch_t match;
1418                 enum grep_context ctx = GREP_CONTEXT_BODY;
1419                 int ch = *eol;
1420                 int eflags = 0;
1421
1422                 if (opt->color) {
1423                         if (sign == ':')
1424                                 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1425                         else
1426                                 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1427                         if (sign == ':')
1428                                 line_color = opt->colors[GREP_COLOR_SELECTED];
1429                         else if (sign == '-')
1430                                 line_color = opt->colors[GREP_COLOR_CONTEXT];
1431                         else if (sign == '=')
1432                                 line_color = opt->colors[GREP_COLOR_FUNCTION];
1433                 }
1434                 *eol = '\0';
1435                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1436                         if (match.rm_so == match.rm_eo)
1437                                 break;
1438
1439                         if (opt->only_matching)
1440                                 show_line_header(opt, name, lno, cno, sign);
1441                         else
1442                                 output_color(opt, bol, match.rm_so, line_color);
1443                         output_color(opt, bol + match.rm_so,
1444                                      match.rm_eo - match.rm_so, match_color);
1445                         if (opt->only_matching)
1446                                 opt->output(opt, "\n", 1);
1447                         bol += match.rm_eo;
1448                         cno += match.rm_eo;
1449                         rest -= match.rm_eo;
1450                         eflags = REG_NOTBOL;
1451                 }
1452                 *eol = ch;
1453         }
1454         if (!opt->only_matching) {
1455                 output_color(opt, bol, rest, line_color);
1456                 opt->output(opt, "\n", 1);
1457         }
1458 }
1459
1460 int grep_use_locks;
1461
1462 /*
1463  * This lock protects access to the gitattributes machinery, which is
1464  * not thread-safe.
1465  */
1466 pthread_mutex_t grep_attr_mutex;
1467
1468 static inline void grep_attr_lock(void)
1469 {
1470         if (grep_use_locks)
1471                 pthread_mutex_lock(&grep_attr_mutex);
1472 }
1473
1474 static inline void grep_attr_unlock(void)
1475 {
1476         if (grep_use_locks)
1477                 pthread_mutex_unlock(&grep_attr_mutex);
1478 }
1479
1480 /*
1481  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1482  */
1483 pthread_mutex_t grep_read_mutex;
1484
1485 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1486 {
1487         xdemitconf_t *xecfg = opt->priv;
1488         if (xecfg && !xecfg->find_func) {
1489                 grep_source_load_driver(gs, opt->repo->index);
1490                 if (gs->driver->funcname.pattern) {
1491                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1492                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1493                 } else {
1494                         xecfg = opt->priv = NULL;
1495                 }
1496         }
1497
1498         if (xecfg) {
1499                 char buf[1];
1500                 return xecfg->find_func(bol, eol - bol, buf, 1,
1501                                         xecfg->find_func_priv) >= 0;
1502         }
1503
1504         if (bol == eol)
1505                 return 0;
1506         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1507                 return 1;
1508         return 0;
1509 }
1510
1511 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1512                                char *bol, unsigned lno)
1513 {
1514         while (bol > gs->buf) {
1515                 char *eol = --bol;
1516
1517                 while (bol > gs->buf && bol[-1] != '\n')
1518                         bol--;
1519                 lno--;
1520
1521                 if (lno <= opt->last_shown)
1522                         break;
1523
1524                 if (match_funcname(opt, gs, bol, eol)) {
1525                         show_line(opt, bol, eol, gs->name, lno, 0, '=');
1526                         break;
1527                 }
1528         }
1529 }
1530
1531 static int is_empty_line(const char *bol, const char *eol);
1532
1533 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1534                              char *bol, char *end, unsigned lno)
1535 {
1536         unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1537         int funcname_needed = !!opt->funcname, comment_needed = 0;
1538
1539         if (opt->pre_context < lno)
1540                 from = lno - opt->pre_context;
1541         if (from <= opt->last_shown)
1542                 from = opt->last_shown + 1;
1543         orig_from = from;
1544         if (opt->funcbody) {
1545                 if (match_funcname(opt, gs, bol, end))
1546                         comment_needed = 1;
1547                 else
1548                         funcname_needed = 1;
1549                 from = opt->last_shown + 1;
1550         }
1551
1552         /* Rewind. */
1553         while (bol > gs->buf && cur > from) {
1554                 char *next_bol = bol;
1555                 char *eol = --bol;
1556
1557                 while (bol > gs->buf && bol[-1] != '\n')
1558                         bol--;
1559                 cur--;
1560                 if (comment_needed && (is_empty_line(bol, eol) ||
1561                                        match_funcname(opt, gs, bol, eol))) {
1562                         comment_needed = 0;
1563                         from = orig_from;
1564                         if (cur < from) {
1565                                 cur++;
1566                                 bol = next_bol;
1567                                 break;
1568                         }
1569                 }
1570                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1571                         funcname_lno = cur;
1572                         funcname_needed = 0;
1573                         if (opt->funcbody)
1574                                 comment_needed = 1;
1575                         else
1576                                 from = orig_from;
1577                 }
1578         }
1579
1580         /* We need to look even further back to find a function signature. */
1581         if (opt->funcname && funcname_needed)
1582                 show_funcname_line(opt, gs, bol, cur);
1583
1584         /* Back forward. */
1585         while (cur < lno) {
1586                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1587
1588                 while (*eol != '\n')
1589                         eol++;
1590                 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1591                 bol = eol + 1;
1592                 cur++;
1593         }
1594 }
1595
1596 static int should_lookahead(struct grep_opt *opt)
1597 {
1598         struct grep_pat *p;
1599
1600         if (opt->extended)
1601                 return 0; /* punt for too complex stuff */
1602         if (opt->invert)
1603                 return 0;
1604         for (p = opt->pattern_list; p; p = p->next) {
1605                 if (p->token != GREP_PATTERN)
1606                         return 0; /* punt for "header only" and stuff */
1607         }
1608         return 1;
1609 }
1610
1611 static int look_ahead(struct grep_opt *opt,
1612                       unsigned long *left_p,
1613                       unsigned *lno_p,
1614                       char **bol_p)
1615 {
1616         unsigned lno = *lno_p;
1617         char *bol = *bol_p;
1618         struct grep_pat *p;
1619         char *sp, *last_bol;
1620         regoff_t earliest = -1;
1621
1622         for (p = opt->pattern_list; p; p = p->next) {
1623                 int hit;
1624                 regmatch_t m;
1625
1626                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1627                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1628                         continue;
1629                 if (earliest < 0 || m.rm_so < earliest)
1630                         earliest = m.rm_so;
1631         }
1632
1633         if (earliest < 0) {
1634                 *bol_p = bol + *left_p;
1635                 *left_p = 0;
1636                 return 1;
1637         }
1638         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1639                 ; /* find the beginning of the line */
1640         last_bol = sp;
1641
1642         for (sp = bol; sp < last_bol; sp++) {
1643                 if (*sp == '\n')
1644                         lno++;
1645         }
1646         *left_p -= last_bol - bol;
1647         *bol_p = last_bol;
1648         *lno_p = lno;
1649         return 0;
1650 }
1651
1652 static int fill_textconv_grep(struct repository *r,
1653                               struct userdiff_driver *driver,
1654                               struct grep_source *gs)
1655 {
1656         struct diff_filespec *df;
1657         char *buf;
1658         size_t size;
1659
1660         if (!driver || !driver->textconv)
1661                 return grep_source_load(gs);
1662
1663         /*
1664          * The textconv interface is intimately tied to diff_filespecs, so we
1665          * have to pretend to be one. If we could unify the grep_source
1666          * and diff_filespec structs, this mess could just go away.
1667          */
1668         df = alloc_filespec(gs->path);
1669         switch (gs->type) {
1670         case GREP_SOURCE_OID:
1671                 fill_filespec(df, gs->identifier, 1, 0100644);
1672                 break;
1673         case GREP_SOURCE_FILE:
1674                 fill_filespec(df, &null_oid, 0, 0100644);
1675                 break;
1676         default:
1677                 BUG("attempt to textconv something without a path?");
1678         }
1679
1680         /*
1681          * fill_textconv is not remotely thread-safe; it may load objects
1682          * behind the scenes, and it modifies the global diff tempfile
1683          * structure.
1684          */
1685         grep_read_lock();
1686         size = fill_textconv(r, driver, df, &buf);
1687         grep_read_unlock();
1688         free_filespec(df);
1689
1690         /*
1691          * The normal fill_textconv usage by the diff machinery would just keep
1692          * the textconv'd buf separate from the diff_filespec. But much of the
1693          * grep code passes around a grep_source and assumes that its "buf"
1694          * pointer is the beginning of the thing we are searching. So let's
1695          * install our textconv'd version into the grep_source, taking care not
1696          * to leak any existing buffer.
1697          */
1698         grep_source_clear_data(gs);
1699         gs->buf = buf;
1700         gs->size = size;
1701
1702         return 0;
1703 }
1704
1705 static int is_empty_line(const char *bol, const char *eol)
1706 {
1707         while (bol < eol && isspace(*bol))
1708                 bol++;
1709         return bol == eol;
1710 }
1711
1712 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1713 {
1714         char *bol;
1715         char *peek_bol = NULL;
1716         unsigned long left;
1717         unsigned lno = 1;
1718         unsigned last_hit = 0;
1719         int binary_match_only = 0;
1720         unsigned count = 0;
1721         int try_lookahead = 0;
1722         int show_function = 0;
1723         struct userdiff_driver *textconv = NULL;
1724         enum grep_context ctx = GREP_CONTEXT_HEAD;
1725         xdemitconf_t xecfg;
1726
1727         if (!opt->status_only && gs->name == NULL)
1728                 BUG("grep call which could print a name requires "
1729                     "grep_source.name be non-NULL");
1730
1731         if (!opt->output)
1732                 opt->output = std_output;
1733
1734         if (opt->pre_context || opt->post_context || opt->file_break ||
1735             opt->funcbody) {
1736                 /* Show hunk marks, except for the first file. */
1737                 if (opt->last_shown)
1738                         opt->show_hunk_mark = 1;
1739                 /*
1740                  * If we're using threads then we can't easily identify
1741                  * the first file.  Always put hunk marks in that case
1742                  * and skip the very first one later in work_done().
1743                  */
1744                 if (opt->output != std_output)
1745                         opt->show_hunk_mark = 1;
1746         }
1747         opt->last_shown = 0;
1748
1749         if (opt->allow_textconv) {
1750                 grep_source_load_driver(gs, opt->repo->index);
1751                 /*
1752                  * We might set up the shared textconv cache data here, which
1753                  * is not thread-safe.
1754                  */
1755                 grep_attr_lock();
1756                 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1757                 grep_attr_unlock();
1758         }
1759
1760         /*
1761          * We know the result of a textconv is text, so we only have to care
1762          * about binary handling if we are not using it.
1763          */
1764         if (!textconv) {
1765                 switch (opt->binary) {
1766                 case GREP_BINARY_DEFAULT:
1767                         if (grep_source_is_binary(gs, opt->repo->index))
1768                                 binary_match_only = 1;
1769                         break;
1770                 case GREP_BINARY_NOMATCH:
1771                         if (grep_source_is_binary(gs, opt->repo->index))
1772                                 return 0; /* Assume unmatch */
1773                         break;
1774                 case GREP_BINARY_TEXT:
1775                         break;
1776                 default:
1777                         BUG("unknown binary handling mode");
1778                 }
1779         }
1780
1781         memset(&xecfg, 0, sizeof(xecfg));
1782         opt->priv = &xecfg;
1783
1784         try_lookahead = should_lookahead(opt);
1785
1786         if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1787                 return 0;
1788
1789         bol = gs->buf;
1790         left = gs->size;
1791         while (left) {
1792                 char *eol, ch;
1793                 int hit;
1794                 ssize_t cno;
1795                 ssize_t col = -1, icol = -1;
1796
1797                 /*
1798                  * look_ahead() skips quickly to the line that possibly
1799                  * has the next hit; don't call it if we need to do
1800                  * something more than just skipping the current line
1801                  * in response to an unmatch for the current line.  E.g.
1802                  * inside a post-context window, we will show the current
1803                  * line as a context around the previous hit when it
1804                  * doesn't hit.
1805                  */
1806                 if (try_lookahead
1807                     && !(last_hit
1808                          && (show_function ||
1809                              lno <= last_hit + opt->post_context))
1810                     && look_ahead(opt, &left, &lno, &bol))
1811                         break;
1812                 eol = end_of_line(bol, &left);
1813                 ch = *eol;
1814                 *eol = 0;
1815
1816                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1817                         ctx = GREP_CONTEXT_BODY;
1818
1819                 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1820                 *eol = ch;
1821
1822                 if (collect_hits)
1823                         goto next_line;
1824
1825                 /* "grep -v -e foo -e bla" should list lines
1826                  * that do not have either, so inversion should
1827                  * be done outside.
1828                  */
1829                 if (opt->invert)
1830                         hit = !hit;
1831                 if (opt->unmatch_name_only) {
1832                         if (hit)
1833                                 return 0;
1834                         goto next_line;
1835                 }
1836                 if (hit) {
1837                         count++;
1838                         if (opt->status_only)
1839                                 return 1;
1840                         if (opt->name_only) {
1841                                 show_name(opt, gs->name);
1842                                 return 1;
1843                         }
1844                         if (opt->count)
1845                                 goto next_line;
1846                         if (binary_match_only) {
1847                                 opt->output(opt, "Binary file ", 12);
1848                                 output_color(opt, gs->name, strlen(gs->name),
1849                                              opt->colors[GREP_COLOR_FILENAME]);
1850                                 opt->output(opt, " matches\n", 9);
1851                                 return 1;
1852                         }
1853                         /* Hit at this line.  If we haven't shown the
1854                          * pre-context lines, we would need to show them.
1855                          */
1856                         if (opt->pre_context || opt->funcbody)
1857                                 show_pre_context(opt, gs, bol, eol, lno);
1858                         else if (opt->funcname)
1859                                 show_funcname_line(opt, gs, bol, lno);
1860                         cno = opt->invert ? icol : col;
1861                         if (cno < 0) {
1862                                 /*
1863                                  * A negative cno indicates that there was no
1864                                  * match on the line. We are thus inverted and
1865                                  * being asked to show all lines that _don't_
1866                                  * match a given expression. Therefore, set cno
1867                                  * to 0 to suggest the whole line matches.
1868                                  */
1869                                 cno = 0;
1870                         }
1871                         show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1872                         last_hit = lno;
1873                         if (opt->funcbody)
1874                                 show_function = 1;
1875                         goto next_line;
1876                 }
1877                 if (show_function && (!peek_bol || peek_bol < bol)) {
1878                         unsigned long peek_left = left;
1879                         char *peek_eol = eol;
1880
1881                         /*
1882                          * Trailing empty lines are not interesting.
1883                          * Peek past them to see if they belong to the
1884                          * body of the current function.
1885                          */
1886                         peek_bol = bol;
1887                         while (is_empty_line(peek_bol, peek_eol)) {
1888                                 peek_bol = peek_eol + 1;
1889                                 peek_eol = end_of_line(peek_bol, &peek_left);
1890                         }
1891
1892                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1893                                 show_function = 0;
1894                 }
1895                 if (show_function ||
1896                     (last_hit && lno <= last_hit + opt->post_context)) {
1897                         /* If the last hit is within the post context,
1898                          * we need to show this line.
1899                          */
1900                         show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1901                 }
1902
1903         next_line:
1904                 bol = eol + 1;
1905                 if (!left)
1906                         break;
1907                 left--;
1908                 lno++;
1909         }
1910
1911         if (collect_hits)
1912                 return 0;
1913
1914         if (opt->status_only)
1915                 return opt->unmatch_name_only;
1916         if (opt->unmatch_name_only) {
1917                 /* We did not see any hit, so we want to show this */
1918                 show_name(opt, gs->name);
1919                 return 1;
1920         }
1921
1922         xdiff_clear_find_func(&xecfg);
1923         opt->priv = NULL;
1924
1925         /* NEEDSWORK:
1926          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1927          * which feels mostly useless but sometimes useful.  Maybe
1928          * make it another option?  For now suppress them.
1929          */
1930         if (opt->count && count) {
1931                 char buf[32];
1932                 if (opt->pathname) {
1933                         output_color(opt, gs->name, strlen(gs->name),
1934                                      opt->colors[GREP_COLOR_FILENAME]);
1935                         output_sep(opt, ':');
1936                 }
1937                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1938                 opt->output(opt, buf, strlen(buf));
1939                 return 1;
1940         }
1941         return !!last_hit;
1942 }
1943
1944 static void clr_hit_marker(struct grep_expr *x)
1945 {
1946         /* All-hit markers are meaningful only at the very top level
1947          * OR node.
1948          */
1949         while (1) {
1950                 x->hit = 0;
1951                 if (x->node != GREP_NODE_OR)
1952                         return;
1953                 x->u.binary.left->hit = 0;
1954                 x = x->u.binary.right;
1955         }
1956 }
1957
1958 static int chk_hit_marker(struct grep_expr *x)
1959 {
1960         /* Top level nodes have hit markers.  See if they all are hits */
1961         while (1) {
1962                 if (x->node != GREP_NODE_OR)
1963                         return x->hit;
1964                 if (!x->u.binary.left->hit)
1965                         return 0;
1966                 x = x->u.binary.right;
1967         }
1968 }
1969
1970 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1971 {
1972         /*
1973          * we do not have to do the two-pass grep when we do not check
1974          * buffer-wide "all-match".
1975          */
1976         if (!opt->all_match)
1977                 return grep_source_1(opt, gs, 0);
1978
1979         /* Otherwise the toplevel "or" terms hit a bit differently.
1980          * We first clear hit markers from them.
1981          */
1982         clr_hit_marker(opt->pattern_expression);
1983         grep_source_1(opt, gs, 1);
1984
1985         if (!chk_hit_marker(opt->pattern_expression))
1986                 return 0;
1987
1988         return grep_source_1(opt, gs, 0);
1989 }
1990
1991 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1992 {
1993         struct grep_source gs;
1994         int r;
1995
1996         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1997         gs.buf = buf;
1998         gs.size = size;
1999
2000         r = grep_source(opt, &gs);
2001
2002         grep_source_clear(&gs);
2003         return r;
2004 }
2005
2006 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
2007                       const char *name, const char *path,
2008                       const void *identifier)
2009 {
2010         gs->type = type;
2011         gs->name = xstrdup_or_null(name);
2012         gs->path = xstrdup_or_null(path);
2013         gs->buf = NULL;
2014         gs->size = 0;
2015         gs->driver = NULL;
2016
2017         switch (type) {
2018         case GREP_SOURCE_FILE:
2019                 gs->identifier = xstrdup(identifier);
2020                 break;
2021         case GREP_SOURCE_OID:
2022                 gs->identifier = oiddup(identifier);
2023                 break;
2024         case GREP_SOURCE_BUF:
2025                 gs->identifier = NULL;
2026                 break;
2027         }
2028 }
2029
2030 void grep_source_clear(struct grep_source *gs)
2031 {
2032         FREE_AND_NULL(gs->name);
2033         FREE_AND_NULL(gs->path);
2034         FREE_AND_NULL(gs->identifier);
2035         grep_source_clear_data(gs);
2036 }
2037
2038 void grep_source_clear_data(struct grep_source *gs)
2039 {
2040         switch (gs->type) {
2041         case GREP_SOURCE_FILE:
2042         case GREP_SOURCE_OID:
2043                 FREE_AND_NULL(gs->buf);
2044                 gs->size = 0;
2045                 break;
2046         case GREP_SOURCE_BUF:
2047                 /* leave user-provided buf intact */
2048                 break;
2049         }
2050 }
2051
2052 static int grep_source_load_oid(struct grep_source *gs)
2053 {
2054         enum object_type type;
2055
2056         grep_read_lock();
2057         gs->buf = read_object_file(gs->identifier, &type, &gs->size);
2058         grep_read_unlock();
2059
2060         if (!gs->buf)
2061                 return error(_("'%s': unable to read %s"),
2062                              gs->name,
2063                              oid_to_hex(gs->identifier));
2064         return 0;
2065 }
2066
2067 static int grep_source_load_file(struct grep_source *gs)
2068 {
2069         const char *filename = gs->identifier;
2070         struct stat st;
2071         char *data;
2072         size_t size;
2073         int i;
2074
2075         if (lstat(filename, &st) < 0) {
2076         err_ret:
2077                 if (errno != ENOENT)
2078                         error_errno(_("failed to stat '%s'"), filename);
2079                 return -1;
2080         }
2081         if (!S_ISREG(st.st_mode))
2082                 return -1;
2083         size = xsize_t(st.st_size);
2084         i = open(filename, O_RDONLY);
2085         if (i < 0)
2086                 goto err_ret;
2087         data = xmallocz(size);
2088         if (st.st_size != read_in_full(i, data, size)) {
2089                 error_errno(_("'%s': short read"), filename);
2090                 close(i);
2091                 free(data);
2092                 return -1;
2093         }
2094         close(i);
2095
2096         gs->buf = data;
2097         gs->size = size;
2098         return 0;
2099 }
2100
2101 static int grep_source_load(struct grep_source *gs)
2102 {
2103         if (gs->buf)
2104                 return 0;
2105
2106         switch (gs->type) {
2107         case GREP_SOURCE_FILE:
2108                 return grep_source_load_file(gs);
2109         case GREP_SOURCE_OID:
2110                 return grep_source_load_oid(gs);
2111         case GREP_SOURCE_BUF:
2112                 return gs->buf ? 0 : -1;
2113         }
2114         BUG("invalid grep_source type to load");
2115 }
2116
2117 void grep_source_load_driver(struct grep_source *gs,
2118                              struct index_state *istate)
2119 {
2120         if (gs->driver)
2121                 return;
2122
2123         grep_attr_lock();
2124         if (gs->path)
2125                 gs->driver = userdiff_find_by_path(istate, gs->path);
2126         if (!gs->driver)
2127                 gs->driver = userdiff_find_by_name("default");
2128         grep_attr_unlock();
2129 }
2130
2131 static int grep_source_is_binary(struct grep_source *gs,
2132                                  struct index_state *istate)
2133 {
2134         grep_source_load_driver(gs, istate);
2135         if (gs->driver->binary != -1)
2136                 return gs->driver->binary;
2137
2138         if (!grep_source_load(gs))
2139                 return buffer_is_binary(gs->buf, gs->size);
2140
2141         return 0;
2142 }