grep.c

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "grep.h"
   4 #include "userdiff.h"
   5 #include "xdiff-interface.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "commit.h"
   9 #include "quote.h"
  10
  11 static int grep_source_load(struct grep_source *gs);
  12 static int grep_source_is_binary(struct grep_source *gs);
  13
  14 static struct grep_opt grep_defaults;
  15
  16 static const char *color_grep_slots[] = {
  17         [GREP_COLOR_CONTEXT]        = "context",
  18         [GREP_COLOR_FILENAME]       = "filename",
  19         [GREP_COLOR_FUNCTION]       = "function",
  20         [GREP_COLOR_LINENO]         = "lineNumber",
  21         [GREP_COLOR_MATCH_CONTEXT]  = "matchContext",
  22         [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
  23         [GREP_COLOR_SELECTED]       = "selected",
  24         [GREP_COLOR_SEP]            = "separator",
  25 };
  26
  27 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
  28 {
  29         fwrite(buf, size, 1, stdout);
  30 }
  31
  32 static void color_set(char *dst, const char *color_bytes)
  33 {
  34         xsnprintf(dst, COLOR_MAXLEN, "%s", color_bytes);
  35 }
  36
  37 /*
  38  * Initialize the grep_defaults template with hardcoded defaults.
  39  * We could let the compiler do this, but without C99 initializers
  40  * the code gets unwieldy and unreadable, so...
  41  */
  42 void init_grep_defaults(void)
  43 {
  44         struct grep_opt *opt = &grep_defaults;
  45         static int run_once;
  46
  47         if (run_once)
  48                 return;
  49         run_once++;
  50
  51         memset(opt, 0, sizeof(*opt));
  52         opt->relative = 1;
  53         opt->pathname = 1;
  54         opt->max_depth = -1;
  55         opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  56         color_set(opt->colors[GREP_COLOR_CONTEXT], "");
  57         color_set(opt->colors[GREP_COLOR_FILENAME], "");
  58         color_set(opt->colors[GREP_COLOR_FUNCTION], "");
  59         color_set(opt->colors[GREP_COLOR_LINENO], "");
  60         color_set(opt->colors[GREP_COLOR_MATCH_CONTEXT], GIT_COLOR_BOLD_RED);
  61         color_set(opt->colors[GREP_COLOR_MATCH_SELECTED], GIT_COLOR_BOLD_RED);
  62         color_set(opt->colors[GREP_COLOR_SELECTED], "");
  63         color_set(opt->colors[GREP_COLOR_SEP], GIT_COLOR_CYAN);
  64         opt->color = -1;
  65         opt->output = std_output;
  66 }
  67
  68 static int parse_pattern_type_arg(const char *opt, const char *arg)
  69 {
  70         if (!strcmp(arg, "default"))
  71                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  72         else if (!strcmp(arg, "basic"))
  73                 return GREP_PATTERN_TYPE_BRE;
  74         else if (!strcmp(arg, "extended"))
  75                 return GREP_PATTERN_TYPE_ERE;
  76         else if (!strcmp(arg, "fixed"))
  77                 return GREP_PATTERN_TYPE_FIXED;
  78         else if (!strcmp(arg, "perl"))
  79                 return GREP_PATTERN_TYPE_PCRE;
  80         die("bad %s argument: %s", opt, arg);
  81 }
  82
  83 /*
  84  * Read the configuration file once and store it in
  85  * the grep_defaults template.
  86  */
  87 int grep_config(const char *var, const char *value, void *cb)
  88 {
  89         struct grep_opt *opt = &grep_defaults;
  90         const char *slot;
  91
  92         if (userdiff_config(var, value) < 0)
  93                 return -1;
  94
  95         if (!strcmp(var, "grep.extendedregexp")) {
  96                 opt->extended_regexp_option = git_config_bool(var, value);
  97                 return 0;
  98         }
  99
 100         if (!strcmp(var, "grep.patterntype")) {
 101                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
 102                 return 0;
 103         }
 104
 105         if (!strcmp(var, "grep.linenumber")) {
 106                 opt->linenum = git_config_bool(var, value);
 107                 return 0;
 108         }
 109
 110         if (!strcmp(var, "grep.fullname")) {
 111                 opt->relative = !git_config_bool(var, value);
 112                 return 0;
 113         }
 114
 115         if (!strcmp(var, "color.grep"))
 116                 opt->color = git_config_colorbool(var, value);
 117         if (!strcmp(var, "color.grep.match")) {
 118                 if (grep_config("color.grep.matchcontext", value, cb) < 0)
 119                         return -1;
 120                 if (grep_config("color.grep.matchselected", value, cb) < 0)
 121                         return -1;
 122         } else if (skip_prefix(var, "color.grep.", &slot)) {
 123                 int i = LOOKUP_CONFIG(color_grep_slots, slot);
 124                 char *color;
 125
 126                 if (i < 0)
 127                         return -1;
 128                 color = opt->colors[i];
 129                 if (!value)
 130                         return config_error_nonbool(var);
 131                 return color_parse(value, color);
 132         }
 133         return 0;
 134 }
 135
 136 /*
 137  * Initialize one instance of grep_opt and copy the
 138  * default values from the template we read the configuration
 139  * information in an earlier call to git_config(grep_config).
 140  */
 141 void grep_init(struct grep_opt *opt, const char *prefix)
 142 {
 143         struct grep_opt *def = &grep_defaults;
 144         int i;
 145
 146         memset(opt, 0, sizeof(*opt));
 147         opt->prefix = prefix;
 148         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 149         opt->pattern_tail = &opt->pattern_list;
 150         opt->header_tail = &opt->header_list;
 151
 152         opt->color = def->color;
 153         opt->extended_regexp_option = def->extended_regexp_option;
 154         opt->pattern_type_option = def->pattern_type_option;
 155         opt->linenum = def->linenum;
 156         opt->max_depth = def->max_depth;
 157         opt->pathname = def->pathname;
 158         opt->relative = def->relative;
 159         opt->output = def->output;
 160
 161         for (i = 0; i < NR_GREP_COLORS; i++)
 162                 color_set(opt->colors[i], def->colors[i]);
 163 }
 164
 165 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 166 {
 167         /*
 168          * When committing to the pattern type by setting the relevant
 169          * fields in grep_opt it's generally not necessary to zero out
 170          * the fields we're not choosing, since they won't have been
 171          * set by anything. The extended_regexp_option field is the
 172          * only exception to this.
 173          *
 174          * This is because in the process of parsing grep.patternType
 175          * & grep.extendedRegexp we set opt->pattern_type_option and
 176          * opt->extended_regexp_option, respectively. We then
 177          * internally use opt->extended_regexp_option to see if we're
 178          * compiling an ERE. It must be unset if that's not actually
 179          * the case.
 180          */
 181         if (pattern_type != GREP_PATTERN_TYPE_ERE &&
 182             opt->extended_regexp_option)
 183                 opt->extended_regexp_option = 0;
 184
 185         switch (pattern_type) {
 186         case GREP_PATTERN_TYPE_UNSPECIFIED:
 187                 /* fall through */
 188
 189         case GREP_PATTERN_TYPE_BRE:
 190                 break;
 191
 192         case GREP_PATTERN_TYPE_ERE:
 193                 opt->extended_regexp_option = 1;
 194                 break;
 195
 196         case GREP_PATTERN_TYPE_FIXED:
 197                 opt->fixed = 1;
 198                 break;
 199
 200         case GREP_PATTERN_TYPE_PCRE:
 201 #ifdef USE_LIBPCRE2
 202                 opt->pcre2 = 1;
 203 #else
 204                 /*
 205                  * It's important that pcre1 always be assigned to
 206                  * even when there's no USE_LIBPCRE* defined. We still
 207                  * call the PCRE stub function, it just dies with
 208                  * "cannot use Perl-compatible regexes[...]".
 209                  */
 210                 opt->pcre1 = 1;
 211 #endif
 212                 break;
 213         }
 214 }
 215
 216 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 217 {
 218         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 219                 grep_set_pattern_type_option(pattern_type, opt);
 220         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 221                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 222         else if (opt->extended_regexp_option)
 223                 /*
 224                  * This branch *must* happen after setting from the
 225                  * opt->pattern_type_option above, we don't want
 226                  * grep.extendedRegexp to override grep.patternType!
 227                  */
 228                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 229 }
 230
 231 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 232                                         const char *origin, int no,
 233                                         enum grep_pat_token t,
 234                                         enum grep_header_field field)
 235 {
 236         struct grep_pat *p = xcalloc(1, sizeof(*p));
 237         p->pattern = xmemdupz(pat, patlen);
 238         p->patternlen = patlen;
 239         p->origin = origin;
 240         p->no = no;
 241         p->token = t;
 242         p->field = field;
 243         return p;
 244 }
 245
 246 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 247 {
 248         **tail = p;
 249         *tail = &p->next;
 250         p->next = NULL;
 251
 252         switch (p->token) {
 253         case GREP_PATTERN: /* atom */
 254         case GREP_PATTERN_HEAD:
 255         case GREP_PATTERN_BODY:
 256                 for (;;) {
 257                         struct grep_pat *new_pat;
 258                         size_t len = 0;
 259                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 260                         while (++len <= p->patternlen) {
 261                                 if (*(--cp) == '\n') {
 262                                         nl = cp;
 263                                         break;
 264                                 }
 265                         }
 266                         if (!nl)
 267                                 break;
 268                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 269                                                   p->no, p->token, p->field);
 270                         new_pat->next = p->next;
 271                         if (!p->next)
 272                                 *tail = &new_pat->next;
 273                         p->next = new_pat;
 274                         *nl = '\0';
 275                         p->patternlen -= len;
 276                 }
 277                 break;
 278         default:
 279                 break;
 280         }
 281 }
 282
 283 void append_header_grep_pattern(struct grep_opt *opt,
 284                                 enum grep_header_field field, const char *pat)
 285 {
 286         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 287                                              GREP_PATTERN_HEAD, field);
 288         if (field == GREP_HEADER_REFLOG)
 289                 opt->use_reflog_filter = 1;
 290         do_append_grep_pat(&opt->header_tail, p);
 291 }
 292
 293 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 294                          const char *origin, int no, enum grep_pat_token t)
 295 {
 296         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 297 }
 298
 299 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 300                      const char *origin, int no, enum grep_pat_token t)
 301 {
 302         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 303         do_append_grep_pat(&opt->pattern_tail, p);
 304 }
 305
 306 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 307 {
 308         struct grep_pat *pat;
 309         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 310         *ret = *opt;
 311
 312         ret->pattern_list = NULL;
 313         ret->pattern_tail = &ret->pattern_list;
 314
 315         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 316         {
 317                 if(pat->token == GREP_PATTERN_HEAD)
 318                         append_header_grep_pattern(ret, pat->field,
 319                                                    pat->pattern);
 320                 else
 321                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 322                                         pat->origin, pat->no, pat->token);
 323         }
 324
 325         return ret;
 326 }
 327
 328 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 329                 const char *error)
 330 {
 331         char where[1024];
 332
 333         if (p->no)
 334                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 335         else if (p->origin)
 336                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 337         else
 338                 where[0] = 0;
 339
 340         die("%s'%s': %s", where, p->pattern, error);
 341 }
 342
 343 static int is_fixed(const char *s, size_t len)
 344 {
 345         size_t i;
 346
 347         for (i = 0; i < len; i++) {
 348                 if (is_regex_special(s[i]))
 349                         return 0;
 350         }
 351
 352         return 1;
 353 }
 354
 355 static int has_null(const char *s, size_t len)
 356 {
 357         /*
 358          * regcomp cannot accept patterns with NULs so when using it
 359          * we consider any pattern containing a NUL fixed.
 360          */
 361         if (memchr(s, 0, len))
 362                 return 1;
 363
 364         return 0;
 365 }
 366
 367 #ifdef USE_LIBPCRE1
 368 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 369 {
 370         const char *error;
 371         int erroffset;
 372         int options = PCRE_MULTILINE;
 373
 374         if (opt->ignore_case) {
 375                 if (has_non_ascii(p->pattern))
 376                         p->pcre1_tables = pcre_maketables();
 377                 options |= PCRE_CASELESS;
 378         }
 379         if (is_utf8_locale() && has_non_ascii(p->pattern))
 380                 options |= PCRE_UTF8;
 381
 382         p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 383                                       p->pcre1_tables);
 384         if (!p->pcre1_regexp)
 385                 compile_regexp_failed(p, error);
 386
 387         p->pcre1_extra_info = pcre_study(p->pcre1_regexp, GIT_PCRE_STUDY_JIT_COMPILE, &error);
 388         if (!p->pcre1_extra_info && error)
 389                 die("%s", error);
 390
 391 #ifdef GIT_PCRE1_USE_JIT
 392         pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
 393         if (p->pcre1_jit_on == 1) {
 394                 p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
 395                 if (!p->pcre1_jit_stack)
 396                         die("Couldn't allocate PCRE JIT stack");
 397                 pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
 398         } else if (p->pcre1_jit_on != 0) {
 399                 die("BUG: The pcre1_jit_on variable should be 0 or 1, not %d",
 400                     p->pcre1_jit_on);
 401         }
 402 #endif
 403 }
 404
 405 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 406                 regmatch_t *match, int eflags)
 407 {
 408         int ovector[30], ret, flags = 0;
 409
 410         if (eflags & REG_NOTBOL)
 411                 flags |= PCRE_NOTBOL;
 412
 413 #ifdef GIT_PCRE1_USE_JIT
 414         if (p->pcre1_jit_on) {
 415                 ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 416                                     eol - line, 0, flags, ovector,
 417                                     ARRAY_SIZE(ovector), p->pcre1_jit_stack);
 418         } else
 419 #endif
 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
 426         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 427                 die("pcre_exec failed with error code %d", ret);
 428         if (ret > 0) {
 429                 ret = 0;
 430                 match->rm_so = ovector[0];
 431                 match->rm_eo = ovector[1];
 432         }
 433
 434         return ret;
 435 }
 436
 437 static void free_pcre1_regexp(struct grep_pat *p)
 438 {
 439         pcre_free(p->pcre1_regexp);
 440 #ifdef GIT_PCRE1_USE_JIT
 441         if (p->pcre1_jit_on) {
 442                 pcre_free_study(p->pcre1_extra_info);
 443                 pcre_jit_stack_free(p->pcre1_jit_stack);
 444         } else
 445 #endif
 446         {
 447                 pcre_free(p->pcre1_extra_info);
 448         }
 449         pcre_free((void *)p->pcre1_tables);
 450 }
 451 #else /* !USE_LIBPCRE1 */
 452 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 453 {
 454         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 455 }
 456
 457 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 458                 regmatch_t *match, int eflags)
 459 {
 460         return 1;
 461 }
 462
 463 static void free_pcre1_regexp(struct grep_pat *p)
 464 {
 465 }
 466 #endif /* !USE_LIBPCRE1 */
 467
 468 #ifdef USE_LIBPCRE2
 469 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 470 {
 471         int error;
 472         PCRE2_UCHAR errbuf[256];
 473         PCRE2_SIZE erroffset;
 474         int options = PCRE2_MULTILINE;
 475         const uint8_t *character_tables = NULL;
 476         int jitret;
 477         int patinforet;
 478         size_t jitsizearg;
 479
 480         assert(opt->pcre2);
 481
 482         p->pcre2_compile_context = NULL;
 483
 484         if (opt->ignore_case) {
 485                 if (has_non_ascii(p->pattern)) {
 486                         character_tables = pcre2_maketables(NULL);
 487                         p->pcre2_compile_context = pcre2_compile_context_create(NULL);
 488                         pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
 489                 }
 490                 options |= PCRE2_CASELESS;
 491         }
 492         if (is_utf8_locale() && has_non_ascii(p->pattern))
 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 == 1) {
 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                 p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
 538                 if (!p->pcre2_jit_stack)
 539                         die("Couldn't allocate PCRE2 JIT stack");
 540                 p->pcre2_match_context = pcre2_match_context_create(NULL);
 541                 if (!p->pcre2_match_context)
 542                         die("Couldn't allocate PCRE2 match context");
 543                 pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
 544         } else if (p->pcre2_jit_on != 0) {
 545                 die("BUG: The pcre2_jit_on variable should be 0 or 1, not %d",
 546                     p->pcre1_jit_on);
 547         }
 548 }
 549
 550 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 551                 regmatch_t *match, int eflags)
 552 {
 553         int ret, flags = 0;
 554         PCRE2_SIZE *ovector;
 555         PCRE2_UCHAR errbuf[256];
 556
 557         if (eflags & REG_NOTBOL)
 558                 flags |= PCRE2_NOTBOL;
 559
 560         if (p->pcre2_jit_on)
 561                 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 562                                       eol - line, 0, flags, p->pcre2_match_data,
 563                                       NULL);
 564         else
 565                 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 566                                   eol - line, 0, flags, p->pcre2_match_data,
 567                                   NULL);
 568
 569         if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 570                 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 571                 die("%s failed with error code %d: %s",
 572                     (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 573                     errbuf);
 574         }
 575         if (ret > 0) {
 576                 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 577                 ret = 0;
 578                 match->rm_so = (int)ovector[0];
 579                 match->rm_eo = (int)ovector[1];
 580         }
 581
 582         return ret;
 583 }
 584
 585 static void free_pcre2_pattern(struct grep_pat *p)
 586 {
 587         pcre2_compile_context_free(p->pcre2_compile_context);
 588         pcre2_code_free(p->pcre2_pattern);
 589         pcre2_match_data_free(p->pcre2_match_data);
 590         pcre2_jit_stack_free(p->pcre2_jit_stack);
 591         pcre2_match_context_free(p->pcre2_match_context);
 592 }
 593 #else /* !USE_LIBPCRE2 */
 594 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 595 {
 596         /*
 597          * Unreachable until USE_LIBPCRE2 becomes synonymous with
 598          * USE_LIBPCRE. See the sibling comment in
 599          * grep_set_pattern_type_option().
 600          */
 601         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 602 }
 603
 604 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 605                 regmatch_t *match, int eflags)
 606 {
 607         return 1;
 608 }
 609
 610 static void free_pcre2_pattern(struct grep_pat *p)
 611 {
 612 }
 613 #endif /* !USE_LIBPCRE2 */
 614
 615 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 616 {
 617         struct strbuf sb = STRBUF_INIT;
 618         int err;
 619         int regflags = 0;
 620
 621         basic_regex_quote_buf(&sb, p->pattern);
 622         if (opt->ignore_case)
 623                 regflags |= REG_ICASE;
 624         err = regcomp(&p->regexp, sb.buf, regflags);
 625         if (opt->debug)
 626                 fprintf(stderr, "fixed %s\n", sb.buf);
 627         strbuf_release(&sb);
 628         if (err) {
 629                 char errbuf[1024];
 630                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 631                 regfree(&p->regexp);
 632                 compile_regexp_failed(p, errbuf);
 633         }
 634 }
 635
 636 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 637 {
 638         int ascii_only;
 639         int err;
 640         int regflags = REG_NEWLINE;
 641
 642         p->word_regexp = opt->word_regexp;
 643         p->ignore_case = opt->ignore_case;
 644         ascii_only     = !has_non_ascii(p->pattern);
 645
 646         /*
 647          * Even when -F (fixed) asks us to do a non-regexp search, we
 648          * may not be able to correctly case-fold when -i
 649          * (ignore-case) is asked (in which case, we'll synthesize a
 650          * regexp to match the pattern that matches regexp special
 651          * characters literally, while ignoring case differences).  On
 652          * the other hand, even without -F, if the pattern does not
 653          * have any regexp special characters and there is no need for
 654          * case-folding search, we can internally turn it into a
 655          * simple string match using kws.  p->fixed tells us if we
 656          * want to use kws.
 657          */
 658         if (opt->fixed ||
 659             has_null(p->pattern, p->patternlen) ||
 660             is_fixed(p->pattern, p->patternlen))
 661                 p->fixed = !p->ignore_case || ascii_only;
 662
 663         if (p->fixed) {
 664                 p->kws = kwsalloc(p->ignore_case ? tolower_trans_tbl : NULL);
 665                 kwsincr(p->kws, p->pattern, p->patternlen);
 666                 kwsprep(p->kws);
 667                 return;
 668         } else if (opt->fixed) {
 669                 /*
 670                  * We come here when the pattern has the non-ascii
 671                  * characters we cannot case-fold, and asked to
 672                  * ignore-case.
 673                  */
 674                 compile_fixed_regexp(p, opt);
 675                 return;
 676         }
 677
 678         if (opt->pcre2) {
 679                 compile_pcre2_pattern(p, opt);
 680                 return;
 681         }
 682
 683         if (opt->pcre1) {
 684                 compile_pcre1_regexp(p, opt);
 685                 return;
 686         }
 687
 688         if (p->ignore_case)
 689                 regflags |= REG_ICASE;
 690         if (opt->extended_regexp_option)
 691                 regflags |= REG_EXTENDED;
 692         err = regcomp(&p->regexp, p->pattern, regflags);
 693         if (err) {
 694                 char errbuf[1024];
 695                 regerror(err, &p->regexp, errbuf, 1024);
 696                 regfree(&p->regexp);
 697                 compile_regexp_failed(p, errbuf);
 698         }
 699 }
 700
 701 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 702 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 703 {
 704         struct grep_pat *p;
 705         struct grep_expr *x;
 706
 707         p = *list;
 708         if (!p)
 709                 return NULL;
 710         switch (p->token) {
 711         case GREP_PATTERN: /* atom */
 712         case GREP_PATTERN_HEAD:
 713         case GREP_PATTERN_BODY:
 714                 x = xcalloc(1, sizeof (struct grep_expr));
 715                 x->node = GREP_NODE_ATOM;
 716                 x->u.atom = p;
 717                 *list = p->next;
 718                 return x;
 719         case GREP_OPEN_PAREN:
 720                 *list = p->next;
 721                 x = compile_pattern_or(list);
 722                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 723                         die("unmatched parenthesis");
 724                 *list = (*list)->next;
 725                 return x;
 726         default:
 727                 return NULL;
 728         }
 729 }
 730
 731 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 732 {
 733         struct grep_pat *p;
 734         struct grep_expr *x;
 735
 736         p = *list;
 737         if (!p)
 738                 return NULL;
 739         switch (p->token) {
 740         case GREP_NOT:
 741                 if (!p->next)
 742                         die("--not not followed by pattern expression");
 743                 *list = p->next;
 744                 x = xcalloc(1, sizeof (struct grep_expr));
 745                 x->node = GREP_NODE_NOT;
 746                 x->u.unary = compile_pattern_not(list);
 747                 if (!x->u.unary)
 748                         die("--not followed by non pattern expression");
 749                 return x;
 750         default:
 751                 return compile_pattern_atom(list);
 752         }
 753 }
 754
 755 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 756 {
 757         struct grep_pat *p;
 758         struct grep_expr *x, *y, *z;
 759
 760         x = compile_pattern_not(list);
 761         p = *list;
 762         if (p && p->token == GREP_AND) {
 763                 if (!p->next)
 764                         die("--and not followed by pattern expression");
 765                 *list = p->next;
 766                 y = compile_pattern_and(list);
 767                 if (!y)
 768                         die("--and not followed by pattern expression");
 769                 z = xcalloc(1, sizeof (struct grep_expr));
 770                 z->node = GREP_NODE_AND;
 771                 z->u.binary.left = x;
 772                 z->u.binary.right = y;
 773                 return z;
 774         }
 775         return x;
 776 }
 777
 778 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 779 {
 780         struct grep_pat *p;
 781         struct grep_expr *x, *y, *z;
 782
 783         x = compile_pattern_and(list);
 784         p = *list;
 785         if (x && p && p->token != GREP_CLOSE_PAREN) {
 786                 y = compile_pattern_or(list);
 787                 if (!y)
 788                         die("not a pattern expression %s", p->pattern);
 789                 z = xcalloc(1, sizeof (struct grep_expr));
 790                 z->node = GREP_NODE_OR;
 791                 z->u.binary.left = x;
 792                 z->u.binary.right = y;
 793                 return z;
 794         }
 795         return x;
 796 }
 797
 798 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 799 {
 800         return compile_pattern_or(list);
 801 }
 802
 803 static void indent(int in)
 804 {
 805         while (in-- > 0)
 806                 fputc(' ', stderr);
 807 }
 808
 809 static void dump_grep_pat(struct grep_pat *p)
 810 {
 811         switch (p->token) {
 812         case GREP_AND: fprintf(stderr, "*and*"); break;
 813         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 814         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 815         case GREP_NOT: fprintf(stderr, "*not*"); break;
 816         case GREP_OR: fprintf(stderr, "*or*"); break;
 817
 818         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 819         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 820         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 821         }
 822
 823         switch (p->token) {
 824         default: break;
 825         case GREP_PATTERN_HEAD:
 826                 fprintf(stderr, "<head %d>", p->field); break;
 827         case GREP_PATTERN_BODY:
 828                 fprintf(stderr, "<body>"); break;
 829         }
 830         switch (p->token) {
 831         default: break;
 832         case GREP_PATTERN_HEAD:
 833         case GREP_PATTERN_BODY:
 834         case GREP_PATTERN:
 835                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 836                 break;
 837         }
 838         fputc('\n', stderr);
 839 }
 840
 841 static void dump_grep_expression_1(struct grep_expr *x, int in)
 842 {
 843         indent(in);
 844         switch (x->node) {
 845         case GREP_NODE_TRUE:
 846                 fprintf(stderr, "true\n");
 847                 break;
 848         case GREP_NODE_ATOM:
 849                 dump_grep_pat(x->u.atom);
 850                 break;
 851         case GREP_NODE_NOT:
 852                 fprintf(stderr, "(not\n");
 853                 dump_grep_expression_1(x->u.unary, in+1);
 854                 indent(in);
 855                 fprintf(stderr, ")\n");
 856                 break;
 857         case GREP_NODE_AND:
 858                 fprintf(stderr, "(and\n");
 859                 dump_grep_expression_1(x->u.binary.left, in+1);
 860                 dump_grep_expression_1(x->u.binary.right, in+1);
 861                 indent(in);
 862                 fprintf(stderr, ")\n");
 863                 break;
 864         case GREP_NODE_OR:
 865                 fprintf(stderr, "(or\n");
 866                 dump_grep_expression_1(x->u.binary.left, in+1);
 867                 dump_grep_expression_1(x->u.binary.right, in+1);
 868                 indent(in);
 869                 fprintf(stderr, ")\n");
 870                 break;
 871         }
 872 }
 873
 874 static void dump_grep_expression(struct grep_opt *opt)
 875 {
 876         struct grep_expr *x = opt->pattern_expression;
 877
 878         if (opt->all_match)
 879                 fprintf(stderr, "[all-match]\n");
 880         dump_grep_expression_1(x, 0);
 881         fflush(NULL);
 882 }
 883
 884 static struct grep_expr *grep_true_expr(void)
 885 {
 886         struct grep_expr *z = xcalloc(1, sizeof(*z));
 887         z->node = GREP_NODE_TRUE;
 888         return z;
 889 }
 890
 891 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 892 {
 893         struct grep_expr *z = xcalloc(1, sizeof(*z));
 894         z->node = GREP_NODE_OR;
 895         z->u.binary.left = left;
 896         z->u.binary.right = right;
 897         return z;
 898 }
 899
 900 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 901 {
 902         struct grep_pat *p;
 903         struct grep_expr *header_expr;
 904         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 905         enum grep_header_field fld;
 906
 907         if (!opt->header_list)
 908                 return NULL;
 909
 910         for (p = opt->header_list; p; p = p->next) {
 911                 if (p->token != GREP_PATTERN_HEAD)
 912                         die("BUG: a non-header pattern in grep header list.");
 913                 if (p->field < GREP_HEADER_FIELD_MIN ||
 914                     GREP_HEADER_FIELD_MAX <= p->field)
 915                         die("BUG: unknown header field %d", p->field);
 916                 compile_regexp(p, opt);
 917         }
 918
 919         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 920                 header_group[fld] = NULL;
 921
 922         for (p = opt->header_list; p; p = p->next) {
 923                 struct grep_expr *h;
 924                 struct grep_pat *pp = p;
 925
 926                 h = compile_pattern_atom(&pp);
 927                 if (!h || pp != p->next)
 928                         die("BUG: malformed header expr");
 929                 if (!header_group[p->field]) {
 930                         header_group[p->field] = h;
 931                         continue;
 932                 }
 933                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 934         }
 935
 936         header_expr = NULL;
 937
 938         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 939                 if (!header_group[fld])
 940                         continue;
 941                 if (!header_expr)
 942                         header_expr = grep_true_expr();
 943                 header_expr = grep_or_expr(header_group[fld], header_expr);
 944         }
 945         return header_expr;
 946 }
 947
 948 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 949 {
 950         struct grep_expr *z = x;
 951
 952         while (x) {
 953                 assert(x->node == GREP_NODE_OR);
 954                 if (x->u.binary.right &&
 955                     x->u.binary.right->node == GREP_NODE_TRUE) {
 956                         x->u.binary.right = y;
 957                         break;
 958                 }
 959                 x = x->u.binary.right;
 960         }
 961         return z;
 962 }
 963
 964 static void compile_grep_patterns_real(struct grep_opt *opt)
 965 {
 966         struct grep_pat *p;
 967         struct grep_expr *header_expr = prep_header_patterns(opt);
 968
 969         for (p = opt->pattern_list; p; p = p->next) {
 970                 switch (p->token) {
 971                 case GREP_PATTERN: /* atom */
 972                 case GREP_PATTERN_HEAD:
 973                 case GREP_PATTERN_BODY:
 974                         compile_regexp(p, opt);
 975                         break;
 976                 default:
 977                         opt->extended = 1;
 978                         break;
 979                 }
 980         }
 981
 982         if (opt->all_match || header_expr)
 983                 opt->extended = 1;
 984         else if (!opt->extended && !opt->debug)
 985                 return;
 986
 987         p = opt->pattern_list;
 988         if (p)
 989                 opt->pattern_expression = compile_pattern_expr(&p);
 990         if (p)
 991                 die("incomplete pattern expression: %s", p->pattern);
 992
 993         if (!header_expr)
 994                 return;
 995
 996         if (!opt->pattern_expression)
 997                 opt->pattern_expression = header_expr;
 998         else if (opt->all_match)
 999                 opt->pattern_expression = grep_splice_or(header_expr,
1000                                                          opt->pattern_expression);
1001         else
1002                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
1003                                                        header_expr);
1004         opt->all_match = 1;
1005 }
1006
1007 void compile_grep_patterns(struct grep_opt *opt)
1008 {
1009         compile_grep_patterns_real(opt);
1010         if (opt->debug)
1011                 dump_grep_expression(opt);
1012 }
1013
1014 static void free_pattern_expr(struct grep_expr *x)
1015 {
1016         switch (x->node) {
1017         case GREP_NODE_TRUE:
1018         case GREP_NODE_ATOM:
1019                 break;
1020         case GREP_NODE_NOT:
1021                 free_pattern_expr(x->u.unary);
1022                 break;
1023         case GREP_NODE_AND:
1024         case GREP_NODE_OR:
1025                 free_pattern_expr(x->u.binary.left);
1026                 free_pattern_expr(x->u.binary.right);
1027                 break;
1028         }
1029         free(x);
1030 }
1031
1032 void free_grep_patterns(struct grep_opt *opt)
1033 {
1034         struct grep_pat *p, *n;
1035
1036         for (p = opt->pattern_list; p; p = n) {
1037                 n = p->next;
1038                 switch (p->token) {
1039                 case GREP_PATTERN: /* atom */
1040                 case GREP_PATTERN_HEAD:
1041                 case GREP_PATTERN_BODY:
1042                         if (p->kws)
1043                                 kwsfree(p->kws);
1044                         else if (p->pcre1_regexp)
1045                                 free_pcre1_regexp(p);
1046                         else if (p->pcre2_pattern)
1047                                 free_pcre2_pattern(p);
1048                         else
1049                                 regfree(&p->regexp);
1050                         free(p->pattern);
1051                         break;
1052                 default:
1053                         break;
1054                 }
1055                 free(p);
1056         }
1057
1058         if (!opt->extended)
1059                 return;
1060         free_pattern_expr(opt->pattern_expression);
1061 }
1062
1063 static char *end_of_line(char *cp, unsigned long *left)
1064 {
1065         unsigned long l = *left;
1066         while (l && *cp != '\n') {
1067                 l--;
1068                 cp++;
1069         }
1070         *left = l;
1071         return cp;
1072 }
1073
1074 static int word_char(char ch)
1075 {
1076         return isalnum(ch) || ch == '_';
1077 }
1078
1079 static void output_color(struct grep_opt *opt, const void *data, size_t size,
1080                          const char *color)
1081 {
1082         if (want_color(opt->color) && color && color[0]) {
1083                 opt->output(opt, color, strlen(color));
1084                 opt->output(opt, data, size);
1085                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1086         } else
1087                 opt->output(opt, data, size);
1088 }
1089
1090 static void output_sep(struct grep_opt *opt, char sign)
1091 {
1092         if (opt->null_following_name)
1093                 opt->output(opt, "\0", 1);
1094         else
1095                 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
1096 }
1097
1098 static void show_name(struct grep_opt *opt, const char *name)
1099 {
1100         output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1101         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1102 }
1103
1104 static int fixmatch(struct grep_pat *p, char *line, char *eol,
1105                     regmatch_t *match)
1106 {
1107         struct kwsmatch kwsm;
1108         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
1109         if (offset == -1) {
1110                 match->rm_so = match->rm_eo = -1;
1111                 return REG_NOMATCH;
1112         } else {
1113                 match->rm_so = offset;
1114                 match->rm_eo = match->rm_so + kwsm.size[0];
1115                 return 0;
1116         }
1117 }
1118
1119 static int patmatch(struct grep_pat *p, char *line, char *eol,
1120                     regmatch_t *match, int eflags)
1121 {
1122         int hit;
1123
1124         if (p->fixed)
1125                 hit = !fixmatch(p, line, eol, match);
1126         else if (p->pcre1_regexp)
1127                 hit = !pcre1match(p, line, eol, match, eflags);
1128         else if (p->pcre2_pattern)
1129                 hit = !pcre2match(p, line, eol, match, eflags);
1130         else
1131                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1132                                    eflags);
1133
1134         return hit;
1135 }
1136
1137 static int strip_timestamp(char *bol, char **eol_p)
1138 {
1139         char *eol = *eol_p;
1140         int ch;
1141
1142         while (bol < --eol) {
1143                 if (*eol != '>')
1144                         continue;
1145                 *eol_p = ++eol;
1146                 ch = *eol;
1147                 *eol = '\0';
1148                 return ch;
1149         }
1150         return 0;
1151 }
1152
1153 static struct {
1154         const char *field;
1155         size_t len;
1156 } header_field[] = {
1157         { "author ", 7 },
1158         { "committer ", 10 },
1159         { "reflog ", 7 },
1160 };
1161
1162 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1163                              enum grep_context ctx,
1164                              regmatch_t *pmatch, int eflags)
1165 {
1166         int hit = 0;
1167         int saved_ch = 0;
1168         const char *start = bol;
1169
1170         if ((p->token != GREP_PATTERN) &&
1171             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1172                 return 0;
1173
1174         if (p->token == GREP_PATTERN_HEAD) {
1175                 const char *field;
1176                 size_t len;
1177                 assert(p->field < ARRAY_SIZE(header_field));
1178                 field = header_field[p->field].field;
1179                 len = header_field[p->field].len;
1180                 if (strncmp(bol, field, len))
1181                         return 0;
1182                 bol += len;
1183                 switch (p->field) {
1184                 case GREP_HEADER_AUTHOR:
1185                 case GREP_HEADER_COMMITTER:
1186                         saved_ch = strip_timestamp(bol, &eol);
1187                         break;
1188                 default:
1189                         break;
1190                 }
1191         }
1192
1193  again:
1194         hit = patmatch(p, bol, eol, pmatch, eflags);
1195
1196         if (hit && p->word_regexp) {
1197                 if ((pmatch[0].rm_so < 0) ||
1198                     (eol - bol) < pmatch[0].rm_so ||
1199                     (pmatch[0].rm_eo < 0) ||
1200                     (eol - bol) < pmatch[0].rm_eo)
1201                         die("regexp returned nonsense");
1202
1203                 /* Match beginning must be either beginning of the
1204                  * line, or at word boundary (i.e. the last char must
1205                  * not be a word char).  Similarly, match end must be
1206                  * either end of the line, or at word boundary
1207                  * (i.e. the next char must not be a word char).
1208                  */
1209                 if ( ((pmatch[0].rm_so == 0) ||
1210                       !word_char(bol[pmatch[0].rm_so-1])) &&
1211                      ((pmatch[0].rm_eo == (eol-bol)) ||
1212                       !word_char(bol[pmatch[0].rm_eo])) )
1213                         ;
1214                 else
1215                         hit = 0;
1216
1217                 /* Words consist of at least one character. */
1218                 if (pmatch->rm_so == pmatch->rm_eo)
1219                         hit = 0;
1220
1221                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1222                         /* There could be more than one match on the
1223                          * line, and the first match might not be
1224                          * strict word match.  But later ones could be!
1225                          * Forward to the next possible start, i.e. the
1226                          * next position following a non-word char.
1227                          */
1228                         bol = pmatch[0].rm_so + bol + 1;
1229                         while (word_char(bol[-1]) && bol < eol)
1230                                 bol++;
1231                         eflags |= REG_NOTBOL;
1232                         if (bol < eol)
1233                                 goto again;
1234                 }
1235         }
1236         if (p->token == GREP_PATTERN_HEAD && saved_ch)
1237                 *eol = saved_ch;
1238         if (hit) {
1239                 pmatch[0].rm_so += bol - start;
1240                 pmatch[0].rm_eo += bol - start;
1241         }
1242         return hit;
1243 }
1244
1245 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1246                            enum grep_context ctx, int collect_hits)
1247 {
1248         int h = 0;
1249         regmatch_t match;
1250
1251         if (!x)
1252                 die("Not a valid grep expression");
1253         switch (x->node) {
1254         case GREP_NODE_TRUE:
1255                 h = 1;
1256                 break;
1257         case GREP_NODE_ATOM:
1258                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1259                 break;
1260         case GREP_NODE_NOT:
1261                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1262                 break;
1263         case GREP_NODE_AND:
1264                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1265                         return 0;
1266                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1267                 break;
1268         case GREP_NODE_OR:
1269                 if (!collect_hits)
1270                         return (match_expr_eval(x->u.binary.left,
1271                                                 bol, eol, ctx, 0) ||
1272                                 match_expr_eval(x->u.binary.right,
1273                                                 bol, eol, ctx, 0));
1274                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1275                 x->u.binary.left->hit |= h;
1276                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1277                 break;
1278         default:
1279                 die("Unexpected node type (internal error) %d", x->node);
1280         }
1281         if (collect_hits)
1282                 x->hit |= h;
1283         return h;
1284 }
1285
1286 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1287                       enum grep_context ctx, int collect_hits)
1288 {
1289         struct grep_expr *x = opt->pattern_expression;
1290         return match_expr_eval(x, bol, eol, ctx, collect_hits);
1291 }
1292
1293 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1294                       enum grep_context ctx, int collect_hits)
1295 {
1296         struct grep_pat *p;
1297         regmatch_t match;
1298
1299         if (opt->extended)
1300                 return match_expr(opt, bol, eol, ctx, collect_hits);
1301
1302         /* we do not call with collect_hits without being extended */
1303         for (p = opt->pattern_list; p; p = p->next) {
1304                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1305                         return 1;
1306         }
1307         return 0;
1308 }
1309
1310 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1311                               enum grep_context ctx,
1312                               regmatch_t *pmatch, int eflags)
1313 {
1314         regmatch_t match;
1315
1316         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1317                 return 0;
1318         if (match.rm_so < 0 || match.rm_eo < 0)
1319                 return 0;
1320         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1321                 if (match.rm_so > pmatch->rm_so)
1322                         return 1;
1323                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1324                         return 1;
1325         }
1326         pmatch->rm_so = match.rm_so;
1327         pmatch->rm_eo = match.rm_eo;
1328         return 1;
1329 }
1330
1331 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1332                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1333 {
1334         struct grep_pat *p;
1335         int hit = 0;
1336
1337         pmatch->rm_so = pmatch->rm_eo = -1;
1338         if (bol < eol) {
1339                 for (p = opt->pattern_list; p; p = p->next) {
1340                         switch (p->token) {
1341                         case GREP_PATTERN: /* atom */
1342                         case GREP_PATTERN_HEAD:
1343                         case GREP_PATTERN_BODY:
1344                                 hit |= match_next_pattern(p, bol, eol, ctx,
1345                                                           pmatch, eflags);
1346                                 break;
1347                         default:
1348                                 break;
1349                         }
1350                 }
1351         }
1352         return hit;
1353 }
1354
1355 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1356                       const char *name, unsigned lno, char sign)
1357 {
1358         int rest = eol - bol;
1359         const char *match_color, *line_color = NULL;
1360
1361         if (opt->file_break && opt->last_shown == 0) {
1362                 if (opt->show_hunk_mark)
1363                         opt->output(opt, "\n", 1);
1364         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1365                 if (opt->last_shown == 0) {
1366                         if (opt->show_hunk_mark) {
1367                                 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1368                                 opt->output(opt, "\n", 1);
1369                         }
1370                 } else if (lno > opt->last_shown + 1) {
1371                         output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1372                         opt->output(opt, "\n", 1);
1373                 }
1374         }
1375         if (opt->heading && opt->last_shown == 0) {
1376                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1377                 opt->output(opt, "\n", 1);
1378         }
1379         opt->last_shown = lno;
1380
1381         if (!opt->heading && opt->pathname) {
1382                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1383                 output_sep(opt, sign);
1384         }
1385         if (opt->linenum) {
1386                 char buf[32];
1387                 xsnprintf(buf, sizeof(buf), "%d", lno);
1388                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1389                 output_sep(opt, sign);
1390         }
1391         if (opt->color) {
1392                 regmatch_t match;
1393                 enum grep_context ctx = GREP_CONTEXT_BODY;
1394                 int ch = *eol;
1395                 int eflags = 0;
1396
1397                 if (sign == ':')
1398                         match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1399                 else
1400                         match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1401                 if (sign == ':')
1402                         line_color = opt->colors[GREP_COLOR_SELECTED];
1403                 else if (sign == '-')
1404                         line_color = opt->colors[GREP_COLOR_CONTEXT];
1405                 else if (sign == '=')
1406                         line_color = opt->colors[GREP_COLOR_FUNCTION];
1407                 *eol = '\0';
1408                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1409                         if (match.rm_so == match.rm_eo)
1410                                 break;
1411
1412                         output_color(opt, bol, match.rm_so, line_color);
1413                         output_color(opt, bol + match.rm_so,
1414                                      match.rm_eo - match.rm_so, match_color);
1415                         bol += match.rm_eo;
1416                         rest -= match.rm_eo;
1417                         eflags = REG_NOTBOL;
1418                 }
1419                 *eol = ch;
1420         }
1421         output_color(opt, bol, rest, line_color);
1422         opt->output(opt, "\n", 1);
1423 }
1424
1425 #ifndef NO_PTHREADS
1426 int grep_use_locks;
1427
1428 /*
1429  * This lock protects access to the gitattributes machinery, which is
1430  * not thread-safe.
1431  */
1432 pthread_mutex_t grep_attr_mutex;
1433
1434 static inline void grep_attr_lock(void)
1435 {
1436         if (grep_use_locks)
1437                 pthread_mutex_lock(&grep_attr_mutex);
1438 }
1439
1440 static inline void grep_attr_unlock(void)
1441 {
1442         if (grep_use_locks)
1443                 pthread_mutex_unlock(&grep_attr_mutex);
1444 }
1445
1446 /*
1447  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1448  */
1449 pthread_mutex_t grep_read_mutex;
1450
1451 #else
1452 #define grep_attr_lock()
1453 #define grep_attr_unlock()
1454 #endif
1455
1456 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1457 {
1458         xdemitconf_t *xecfg = opt->priv;
1459         if (xecfg && !xecfg->find_func) {
1460                 grep_source_load_driver(gs);
1461                 if (gs->driver->funcname.pattern) {
1462                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1463                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1464                 } else {
1465                         xecfg = opt->priv = NULL;
1466                 }
1467         }
1468
1469         if (xecfg) {
1470                 char buf[1];
1471                 return xecfg->find_func(bol, eol - bol, buf, 1,
1472                                         xecfg->find_func_priv) >= 0;
1473         }
1474
1475         if (bol == eol)
1476                 return 0;
1477         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1478                 return 1;
1479         return 0;
1480 }
1481
1482 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1483                                char *bol, unsigned lno)
1484 {
1485         while (bol > gs->buf) {
1486                 char *eol = --bol;
1487
1488                 while (bol > gs->buf && bol[-1] != '\n')
1489                         bol--;
1490                 lno--;
1491
1492                 if (lno <= opt->last_shown)
1493                         break;
1494
1495                 if (match_funcname(opt, gs, bol, eol)) {
1496                         show_line(opt, bol, eol, gs->name, lno, '=');
1497                         break;
1498                 }
1499         }
1500 }
1501
1502 static int is_empty_line(const char *bol, const char *eol);
1503
1504 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1505                              char *bol, char *end, unsigned lno)
1506 {
1507         unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1508         int funcname_needed = !!opt->funcname, comment_needed = 0;
1509
1510         if (opt->pre_context < lno)
1511                 from = lno - opt->pre_context;
1512         if (from <= opt->last_shown)
1513                 from = opt->last_shown + 1;
1514         orig_from = from;
1515         if (opt->funcbody) {
1516                 if (match_funcname(opt, gs, bol, end))
1517                         comment_needed = 1;
1518                 else
1519                         funcname_needed = 1;
1520                 from = opt->last_shown + 1;
1521         }
1522
1523         /* Rewind. */
1524         while (bol > gs->buf && cur > from) {
1525                 char *next_bol = bol;
1526                 char *eol = --bol;
1527
1528                 while (bol > gs->buf && bol[-1] != '\n')
1529                         bol--;
1530                 cur--;
1531                 if (comment_needed && (is_empty_line(bol, eol) ||
1532                                        match_funcname(opt, gs, bol, eol))) {
1533                         comment_needed = 0;
1534                         from = orig_from;
1535                         if (cur < from) {
1536                                 cur++;
1537                                 bol = next_bol;
1538                                 break;
1539                         }
1540                 }
1541                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1542                         funcname_lno = cur;
1543                         funcname_needed = 0;
1544                         if (opt->funcbody)
1545                                 comment_needed = 1;
1546                         else
1547                                 from = orig_from;
1548                 }
1549         }
1550
1551         /* We need to look even further back to find a function signature. */
1552         if (opt->funcname && funcname_needed)
1553                 show_funcname_line(opt, gs, bol, cur);
1554
1555         /* Back forward. */
1556         while (cur < lno) {
1557                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1558
1559                 while (*eol != '\n')
1560                         eol++;
1561                 show_line(opt, bol, eol, gs->name, cur, sign);
1562                 bol = eol + 1;
1563                 cur++;
1564         }
1565 }
1566
1567 static int should_lookahead(struct grep_opt *opt)
1568 {
1569         struct grep_pat *p;
1570
1571         if (opt->extended)
1572                 return 0; /* punt for too complex stuff */
1573         if (opt->invert)
1574                 return 0;
1575         for (p = opt->pattern_list; p; p = p->next) {
1576                 if (p->token != GREP_PATTERN)
1577                         return 0; /* punt for "header only" and stuff */
1578         }
1579         return 1;
1580 }
1581
1582 static int look_ahead(struct grep_opt *opt,
1583                       unsigned long *left_p,
1584                       unsigned *lno_p,
1585                       char **bol_p)
1586 {
1587         unsigned lno = *lno_p;
1588         char *bol = *bol_p;
1589         struct grep_pat *p;
1590         char *sp, *last_bol;
1591         regoff_t earliest = -1;
1592
1593         for (p = opt->pattern_list; p; p = p->next) {
1594                 int hit;
1595                 regmatch_t m;
1596
1597                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1598                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1599                         continue;
1600                 if (earliest < 0 || m.rm_so < earliest)
1601                         earliest = m.rm_so;
1602         }
1603
1604         if (earliest < 0) {
1605                 *bol_p = bol + *left_p;
1606                 *left_p = 0;
1607                 return 1;
1608         }
1609         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1610                 ; /* find the beginning of the line */
1611         last_bol = sp;
1612
1613         for (sp = bol; sp < last_bol; sp++) {
1614                 if (*sp == '\n')
1615                         lno++;
1616         }
1617         *left_p -= last_bol - bol;
1618         *bol_p = last_bol;
1619         *lno_p = lno;
1620         return 0;
1621 }
1622
1623 static int fill_textconv_grep(struct userdiff_driver *driver,
1624                               struct grep_source *gs)
1625 {
1626         struct diff_filespec *df;
1627         char *buf;
1628         size_t size;
1629
1630         if (!driver || !driver->textconv)
1631                 return grep_source_load(gs);
1632
1633         /*
1634          * The textconv interface is intimately tied to diff_filespecs, so we
1635          * have to pretend to be one. If we could unify the grep_source
1636          * and diff_filespec structs, this mess could just go away.
1637          */
1638         df = alloc_filespec(gs->path);
1639         switch (gs->type) {
1640         case GREP_SOURCE_OID:
1641                 fill_filespec(df, gs->identifier, 1, 0100644);
1642                 break;
1643         case GREP_SOURCE_FILE:
1644                 fill_filespec(df, &null_oid, 0, 0100644);
1645                 break;
1646         default:
1647                 die("BUG: attempt to textconv something without a path?");
1648         }
1649
1650         /*
1651          * fill_textconv is not remotely thread-safe; it may load objects
1652          * behind the scenes, and it modifies the global diff tempfile
1653          * structure.
1654          */
1655         grep_read_lock();
1656         size = fill_textconv(driver, df, &buf);
1657         grep_read_unlock();
1658         free_filespec(df);
1659
1660         /*
1661          * The normal fill_textconv usage by the diff machinery would just keep
1662          * the textconv'd buf separate from the diff_filespec. But much of the
1663          * grep code passes around a grep_source and assumes that its "buf"
1664          * pointer is the beginning of the thing we are searching. So let's
1665          * install our textconv'd version into the grep_source, taking care not
1666          * to leak any existing buffer.
1667          */
1668         grep_source_clear_data(gs);
1669         gs->buf = buf;
1670         gs->size = size;
1671
1672         return 0;
1673 }
1674
1675 static int is_empty_line(const char *bol, const char *eol)
1676 {
1677         while (bol < eol && isspace(*bol))
1678                 bol++;
1679         return bol == eol;
1680 }
1681
1682 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1683 {
1684         char *bol;
1685         char *peek_bol = NULL;
1686         unsigned long left;
1687         unsigned lno = 1;
1688         unsigned last_hit = 0;
1689         int binary_match_only = 0;
1690         unsigned count = 0;
1691         int try_lookahead = 0;
1692         int show_function = 0;
1693         struct userdiff_driver *textconv = NULL;
1694         enum grep_context ctx = GREP_CONTEXT_HEAD;
1695         xdemitconf_t xecfg;
1696
1697         if (!opt->output)
1698                 opt->output = std_output;
1699
1700         if (opt->pre_context || opt->post_context || opt->file_break ||
1701             opt->funcbody) {
1702                 /* Show hunk marks, except for the first file. */
1703                 if (opt->last_shown)
1704                         opt->show_hunk_mark = 1;
1705                 /*
1706                  * If we're using threads then we can't easily identify
1707                  * the first file.  Always put hunk marks in that case
1708                  * and skip the very first one later in work_done().
1709                  */
1710                 if (opt->output != std_output)
1711                         opt->show_hunk_mark = 1;
1712         }
1713         opt->last_shown = 0;
1714
1715         if (opt->allow_textconv) {
1716                 grep_source_load_driver(gs);
1717                 /*
1718                  * We might set up the shared textconv cache data here, which
1719                  * is not thread-safe.
1720                  */
1721                 grep_attr_lock();
1722                 textconv = userdiff_get_textconv(gs->driver);
1723                 grep_attr_unlock();
1724         }
1725
1726         /*
1727          * We know the result of a textconv is text, so we only have to care
1728          * about binary handling if we are not using it.
1729          */
1730         if (!textconv) {
1731                 switch (opt->binary) {
1732                 case GREP_BINARY_DEFAULT:
1733                         if (grep_source_is_binary(gs))
1734                                 binary_match_only = 1;
1735                         break;
1736                 case GREP_BINARY_NOMATCH:
1737                         if (grep_source_is_binary(gs))
1738                                 return 0; /* Assume unmatch */
1739                         break;
1740                 case GREP_BINARY_TEXT:
1741                         break;
1742                 default:
1743                         die("BUG: unknown binary handling mode");
1744                 }
1745         }
1746
1747         memset(&xecfg, 0, sizeof(xecfg));
1748         opt->priv = &xecfg;
1749
1750         try_lookahead = should_lookahead(opt);
1751
1752         if (fill_textconv_grep(textconv, gs) < 0)
1753                 return 0;
1754
1755         bol = gs->buf;
1756         left = gs->size;
1757         while (left) {
1758                 char *eol, ch;
1759                 int hit;
1760
1761                 /*
1762                  * look_ahead() skips quickly to the line that possibly
1763                  * has the next hit; don't call it if we need to do
1764                  * something more than just skipping the current line
1765                  * in response to an unmatch for the current line.  E.g.
1766                  * inside a post-context window, we will show the current
1767                  * line as a context around the previous hit when it
1768                  * doesn't hit.
1769                  */
1770                 if (try_lookahead
1771                     && !(last_hit
1772                          && (show_function ||
1773                              lno <= last_hit + opt->post_context))
1774                     && look_ahead(opt, &left, &lno, &bol))
1775                         break;
1776                 eol = end_of_line(bol, &left);
1777                 ch = *eol;
1778                 *eol = 0;
1779
1780                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1781                         ctx = GREP_CONTEXT_BODY;
1782
1783                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1784                 *eol = ch;
1785
1786                 if (collect_hits)
1787                         goto next_line;
1788
1789                 /* "grep -v -e foo -e bla" should list lines
1790                  * that do not have either, so inversion should
1791                  * be done outside.
1792                  */
1793                 if (opt->invert)
1794                         hit = !hit;
1795                 if (opt->unmatch_name_only) {
1796                         if (hit)
1797                                 return 0;
1798                         goto next_line;
1799                 }
1800                 if (hit) {
1801                         count++;
1802                         if (opt->status_only)
1803                                 return 1;
1804                         if (opt->name_only) {
1805                                 show_name(opt, gs->name);
1806                                 return 1;
1807                         }
1808                         if (opt->count)
1809                                 goto next_line;
1810                         if (binary_match_only) {
1811                                 opt->output(opt, "Binary file ", 12);
1812                                 output_color(opt, gs->name, strlen(gs->name),
1813                                              opt->colors[GREP_COLOR_FILENAME]);
1814                                 opt->output(opt, " matches\n", 9);
1815                                 return 1;
1816                         }
1817                         /* Hit at this line.  If we haven't shown the
1818                          * pre-context lines, we would need to show them.
1819                          */
1820                         if (opt->pre_context || opt->funcbody)
1821                                 show_pre_context(opt, gs, bol, eol, lno);
1822                         else if (opt->funcname)
1823                                 show_funcname_line(opt, gs, bol, lno);
1824                         show_line(opt, bol, eol, gs->name, lno, ':');
1825                         last_hit = lno;
1826                         if (opt->funcbody)
1827                                 show_function = 1;
1828                         goto next_line;
1829                 }
1830                 if (show_function && (!peek_bol || peek_bol < bol)) {
1831                         unsigned long peek_left = left;
1832                         char *peek_eol = eol;
1833
1834                         /*
1835                          * Trailing empty lines are not interesting.
1836                          * Peek past them to see if they belong to the
1837                          * body of the current function.
1838                          */
1839                         peek_bol = bol;
1840                         while (is_empty_line(peek_bol, peek_eol)) {
1841                                 peek_bol = peek_eol + 1;
1842                                 peek_eol = end_of_line(peek_bol, &peek_left);
1843                         }
1844
1845                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1846                                 show_function = 0;
1847                 }
1848                 if (show_function ||
1849                     (last_hit && lno <= last_hit + opt->post_context)) {
1850                         /* If the last hit is within the post context,
1851                          * we need to show this line.
1852                          */
1853                         show_line(opt, bol, eol, gs->name, lno, '-');
1854                 }
1855
1856         next_line:
1857                 bol = eol + 1;
1858                 if (!left)
1859                         break;
1860                 left--;
1861                 lno++;
1862         }
1863
1864         if (collect_hits)
1865                 return 0;
1866
1867         if (opt->status_only)
1868                 return opt->unmatch_name_only;
1869         if (opt->unmatch_name_only) {
1870                 /* We did not see any hit, so we want to show this */
1871                 show_name(opt, gs->name);
1872                 return 1;
1873         }
1874
1875         xdiff_clear_find_func(&xecfg);
1876         opt->priv = NULL;
1877
1878         /* NEEDSWORK:
1879          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1880          * which feels mostly useless but sometimes useful.  Maybe
1881          * make it another option?  For now suppress them.
1882          */
1883         if (opt->count && count) {
1884                 char buf[32];
1885                 if (opt->pathname) {
1886                         output_color(opt, gs->name, strlen(gs->name),
1887                                      opt->colors[GREP_COLOR_FILENAME]);
1888                         output_sep(opt, ':');
1889                 }
1890                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1891                 opt->output(opt, buf, strlen(buf));
1892                 return 1;
1893         }
1894         return !!last_hit;
1895 }
1896
1897 static void clr_hit_marker(struct grep_expr *x)
1898 {
1899         /* All-hit markers are meaningful only at the very top level
1900          * OR node.
1901          */
1902         while (1) {
1903                 x->hit = 0;
1904                 if (x->node != GREP_NODE_OR)
1905                         return;
1906                 x->u.binary.left->hit = 0;
1907                 x = x->u.binary.right;
1908         }
1909 }
1910
1911 static int chk_hit_marker(struct grep_expr *x)
1912 {
1913         /* Top level nodes have hit markers.  See if they all are hits */
1914         while (1) {
1915                 if (x->node != GREP_NODE_OR)
1916                         return x->hit;
1917                 if (!x->u.binary.left->hit)
1918                         return 0;
1919                 x = x->u.binary.right;
1920         }
1921 }
1922
1923 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1924 {
1925         /*
1926          * we do not have to do the two-pass grep when we do not check
1927          * buffer-wide "all-match".
1928          */
1929         if (!opt->all_match)
1930                 return grep_source_1(opt, gs, 0);
1931
1932         /* Otherwise the toplevel "or" terms hit a bit differently.
1933          * We first clear hit markers from them.
1934          */
1935         clr_hit_marker(opt->pattern_expression);
1936         grep_source_1(opt, gs, 1);
1937
1938         if (!chk_hit_marker(opt->pattern_expression))
1939                 return 0;
1940
1941         return grep_source_1(opt, gs, 0);
1942 }
1943
1944 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1945 {
1946         struct grep_source gs;
1947         int r;
1948
1949         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1950         gs.buf = buf;
1951         gs.size = size;
1952
1953         r = grep_source(opt, &gs);
1954
1955         grep_source_clear(&gs);
1956         return r;
1957 }
1958
1959 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1960                       const char *name, const char *path,
1961                       const void *identifier)
1962 {
1963         gs->type = type;
1964         gs->name = xstrdup_or_null(name);
1965         gs->path = xstrdup_or_null(path);
1966         gs->buf = NULL;
1967         gs->size = 0;
1968         gs->driver = NULL;
1969
1970         switch (type) {
1971         case GREP_SOURCE_FILE:
1972                 gs->identifier = xstrdup(identifier);
1973                 break;
1974         case GREP_SOURCE_OID:
1975                 gs->identifier = oiddup(identifier);
1976                 break;
1977         case GREP_SOURCE_BUF:
1978                 gs->identifier = NULL;
1979                 break;
1980         }
1981 }
1982
1983 void grep_source_clear(struct grep_source *gs)
1984 {
1985         FREE_AND_NULL(gs->name);
1986         FREE_AND_NULL(gs->path);
1987         FREE_AND_NULL(gs->identifier);
1988         grep_source_clear_data(gs);
1989 }
1990
1991 void grep_source_clear_data(struct grep_source *gs)
1992 {
1993         switch (gs->type) {
1994         case GREP_SOURCE_FILE:
1995         case GREP_SOURCE_OID:
1996                 FREE_AND_NULL(gs->buf);
1997                 gs->size = 0;
1998                 break;
1999         case GREP_SOURCE_BUF:
2000                 /* leave user-provided buf intact */
2001                 break;
2002         }
2003 }
2004
2005 static int grep_source_load_oid(struct grep_source *gs)
2006 {
2007         enum object_type type;
2008
2009         grep_read_lock();
2010         gs->buf = read_object_file(gs->identifier, &type, &gs->size);
2011         grep_read_unlock();
2012
2013         if (!gs->buf)
2014                 return error(_("'%s': unable to read %s"),
2015                              gs->name,
2016                              oid_to_hex(gs->identifier));
2017         return 0;
2018 }
2019
2020 static int grep_source_load_file(struct grep_source *gs)
2021 {
2022         const char *filename = gs->identifier;
2023         struct stat st;
2024         char *data;
2025         size_t size;
2026         int i;
2027
2028         if (lstat(filename, &st) < 0) {
2029         err_ret:
2030                 if (errno != ENOENT)
2031                         error_errno(_("failed to stat '%s'"), filename);
2032                 return -1;
2033         }
2034         if (!S_ISREG(st.st_mode))
2035                 return -1;
2036         size = xsize_t(st.st_size);
2037         i = open(filename, O_RDONLY);
2038         if (i < 0)
2039                 goto err_ret;
2040         data = xmallocz(size);
2041         if (st.st_size != read_in_full(i, data, size)) {
2042                 error_errno(_("'%s': short read"), filename);
2043                 close(i);
2044                 free(data);
2045                 return -1;
2046         }
2047         close(i);
2048
2049         gs->buf = data;
2050         gs->size = size;
2051         return 0;
2052 }
2053
2054 static int grep_source_load(struct grep_source *gs)
2055 {
2056         if (gs->buf)
2057                 return 0;
2058
2059         switch (gs->type) {
2060         case GREP_SOURCE_FILE:
2061                 return grep_source_load_file(gs);
2062         case GREP_SOURCE_OID:
2063                 return grep_source_load_oid(gs);
2064         case GREP_SOURCE_BUF:
2065                 return gs->buf ? 0 : -1;
2066         }
2067         die("BUG: invalid grep_source type to load");
2068 }
2069
2070 void grep_source_load_driver(struct grep_source *gs)
2071 {
2072         if (gs->driver)
2073                 return;
2074
2075         grep_attr_lock();
2076         if (gs->path)
2077                 gs->driver = userdiff_find_by_path(gs->path);
2078         if (!gs->driver)
2079                 gs->driver = userdiff_find_by_name("default");
2080         grep_attr_unlock();
2081 }
2082
2083 static int grep_source_is_binary(struct grep_source *gs)
2084 {
2085         grep_source_load_driver(gs);
2086         if (gs->driver->binary != -1)
2087                 return gs->driver->binary;
2088
2089         if (!grep_source_load(gs))
2090                 return buffer_is_binary(gs->buf, gs->size);
2091
2092         return 0;
2093 }