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
  12 static int grep_source_load(struct grep_source *gs);
  13 static int grep_source_is_binary(struct grep_source *gs);
  14
  15 static struct grep_opt grep_defaults;
  16
  17 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
  18 {
  19         fwrite(buf, size, 1, stdout);
  20 }
  21
  22 static void color_set(char *dst, const char *color_bytes)
  23 {
  24         xsnprintf(dst, COLOR_MAXLEN, "%s", color_bytes);
  25 }
  26
  27 /*
  28  * Initialize the grep_defaults template with hardcoded defaults.
  29  * We could let the compiler do this, but without C99 initializers
  30  * the code gets unwieldy and unreadable, so...
  31  */
  32 void init_grep_defaults(void)
  33 {
  34         struct grep_opt *opt = &grep_defaults;
  35         static int run_once;
  36
  37         if (run_once)
  38                 return;
  39         run_once++;
  40
  41         memset(opt, 0, sizeof(*opt));
  42         opt->relative = 1;
  43         opt->pathname = 1;
  44         opt->max_depth = -1;
  45         opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  46         color_set(opt->color_context, "");
  47         color_set(opt->color_filename, "");
  48         color_set(opt->color_function, "");
  49         color_set(opt->color_lineno, "");
  50         color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
  51         color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
  52         color_set(opt->color_selected, "");
  53         color_set(opt->color_sep, GIT_COLOR_CYAN);
  54         opt->color = -1;
  55         opt->output = std_output;
  56 }
  57
  58 static int parse_pattern_type_arg(const char *opt, const char *arg)
  59 {
  60         if (!strcmp(arg, "default"))
  61                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  62         else if (!strcmp(arg, "basic"))
  63                 return GREP_PATTERN_TYPE_BRE;
  64         else if (!strcmp(arg, "extended"))
  65                 return GREP_PATTERN_TYPE_ERE;
  66         else if (!strcmp(arg, "fixed"))
  67                 return GREP_PATTERN_TYPE_FIXED;
  68         else if (!strcmp(arg, "perl"))
  69                 return GREP_PATTERN_TYPE_PCRE;
  70         die("bad %s argument: %s", opt, arg);
  71 }
  72
  73 /*
  74  * Read the configuration file once and store it in
  75  * the grep_defaults template.
  76  */
  77 int grep_config(const char *var, const char *value, void *cb)
  78 {
  79         struct grep_opt *opt = &grep_defaults;
  80         char *color = NULL;
  81
  82         if (userdiff_config(var, value) < 0)
  83                 return -1;
  84
  85         if (!strcmp(var, "grep.extendedregexp")) {
  86                 opt->extended_regexp_option = git_config_bool(var, value);
  87                 return 0;
  88         }
  89
  90         if (!strcmp(var, "grep.patterntype")) {
  91                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
  92                 return 0;
  93         }
  94
  95         if (!strcmp(var, "grep.linenumber")) {
  96                 opt->linenum = git_config_bool(var, value);
  97                 return 0;
  98         }
  99
 100         if (!strcmp(var, "grep.fullname")) {
 101                 opt->relative = !git_config_bool(var, value);
 102                 return 0;
 103         }
 104
 105         if (!strcmp(var, "color.grep"))
 106                 opt->color = git_config_colorbool(var, value);
 107         else if (!strcmp(var, "color.grep.context"))
 108                 color = opt->color_context;
 109         else if (!strcmp(var, "color.grep.filename"))
 110                 color = opt->color_filename;
 111         else if (!strcmp(var, "color.grep.function"))
 112                 color = opt->color_function;
 113         else if (!strcmp(var, "color.grep.linenumber"))
 114                 color = opt->color_lineno;
 115         else if (!strcmp(var, "color.grep.matchcontext"))
 116                 color = opt->color_match_context;
 117         else if (!strcmp(var, "color.grep.matchselected"))
 118                 color = opt->color_match_selected;
 119         else if (!strcmp(var, "color.grep.selected"))
 120                 color = opt->color_selected;
 121         else if (!strcmp(var, "color.grep.separator"))
 122                 color = opt->color_sep;
 123         else if (!strcmp(var, "color.grep.match")) {
 124                 int rc = 0;
 125                 if (!value)
 126                         return config_error_nonbool(var);
 127                 rc |= color_parse(value, opt->color_match_context);
 128                 rc |= color_parse(value, opt->color_match_selected);
 129                 return rc;
 130         }
 131
 132         if (color) {
 133                 if (!value)
 134                         return config_error_nonbool(var);
 135                 return color_parse(value, color);
 136         }
 137         return 0;
 138 }
 139
 140 /*
 141  * Initialize one instance of grep_opt and copy the
 142  * default values from the template we read the configuration
 143  * information in an earlier call to git_config(grep_config).
 144  */
 145 void grep_init(struct grep_opt *opt, const char *prefix)
 146 {
 147         struct grep_opt *def = &grep_defaults;
 148
 149         memset(opt, 0, sizeof(*opt));
 150         opt->prefix = prefix;
 151         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 152         opt->pattern_tail = &opt->pattern_list;
 153         opt->header_tail = &opt->header_list;
 154
 155         opt->color = def->color;
 156         opt->extended_regexp_option = def->extended_regexp_option;
 157         opt->pattern_type_option = def->pattern_type_option;
 158         opt->linenum = def->linenum;
 159         opt->max_depth = def->max_depth;
 160         opt->pathname = def->pathname;
 161         opt->relative = def->relative;
 162         opt->output = def->output;
 163
 164         color_set(opt->color_context, def->color_context);
 165         color_set(opt->color_filename, def->color_filename);
 166         color_set(opt->color_function, def->color_function);
 167         color_set(opt->color_lineno, def->color_lineno);
 168         color_set(opt->color_match_context, def->color_match_context);
 169         color_set(opt->color_match_selected, def->color_match_selected);
 170         color_set(opt->color_selected, def->color_selected);
 171         color_set(opt->color_sep, def->color_sep);
 172 }
 173
 174 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 175 {
 176         /*
 177          * When committing to the pattern type by setting the relevant
 178          * fields in grep_opt it's generally not necessary to zero out
 179          * the fields we're not choosing, since they won't have been
 180          * set by anything. The extended_regexp_option field is the
 181          * only exception to this.
 182          *
 183          * This is because in the process of parsing grep.patternType
 184          * & grep.extendedRegexp we set opt->pattern_type_option and
 185          * opt->extended_regexp_option, respectively. We then
 186          * internally use opt->extended_regexp_option to see if we're
 187          * compiling an ERE. It must be unset if that's not actually
 188          * the case.
 189          */
 190         if (pattern_type != GREP_PATTERN_TYPE_ERE &&
 191             opt->extended_regexp_option)
 192                 opt->extended_regexp_option = 0;
 193
 194         switch (pattern_type) {
 195         case GREP_PATTERN_TYPE_UNSPECIFIED:
 196                 /* fall through */
 197
 198         case GREP_PATTERN_TYPE_BRE:
 199                 break;
 200
 201         case GREP_PATTERN_TYPE_ERE:
 202                 opt->extended_regexp_option = 1;
 203                 break;
 204
 205         case GREP_PATTERN_TYPE_FIXED:
 206                 opt->fixed = 1;
 207                 break;
 208
 209         case GREP_PATTERN_TYPE_PCRE:
 210 #ifdef USE_LIBPCRE2
 211                 opt->pcre2 = 1;
 212 #else
 213                 /*
 214                  * It's important that pcre1 always be assigned to
 215                  * even when there's no USE_LIBPCRE* defined. We still
 216                  * call the PCRE stub function, it just dies with
 217                  * "cannot use Perl-compatible regexes[...]".
 218                  */
 219                 opt->pcre1 = 1;
 220 #endif
 221                 break;
 222         }
 223 }
 224
 225 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 226 {
 227         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 228                 grep_set_pattern_type_option(pattern_type, opt);
 229         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 230                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 231         else if (opt->extended_regexp_option)
 232                 /*
 233                  * This branch *must* happen after setting from the
 234                  * opt->pattern_type_option above, we don't want
 235                  * grep.extendedRegexp to override grep.patternType!
 236                  */
 237                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 238 }
 239
 240 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 241                                         const char *origin, int no,
 242                                         enum grep_pat_token t,
 243                                         enum grep_header_field field)
 244 {
 245         struct grep_pat *p = xcalloc(1, sizeof(*p));
 246         p->pattern = xmemdupz(pat, patlen);
 247         p->patternlen = patlen;
 248         p->origin = origin;
 249         p->no = no;
 250         p->token = t;
 251         p->field = field;
 252         return p;
 253 }
 254
 255 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 256 {
 257         **tail = p;
 258         *tail = &p->next;
 259         p->next = NULL;
 260
 261         switch (p->token) {
 262         case GREP_PATTERN: /* atom */
 263         case GREP_PATTERN_HEAD:
 264         case GREP_PATTERN_BODY:
 265                 for (;;) {
 266                         struct grep_pat *new_pat;
 267                         size_t len = 0;
 268                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 269                         while (++len <= p->patternlen) {
 270                                 if (*(--cp) == '\n') {
 271                                         nl = cp;
 272                                         break;
 273                                 }
 274                         }
 275                         if (!nl)
 276                                 break;
 277                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 278                                                   p->no, p->token, p->field);
 279                         new_pat->next = p->next;
 280                         if (!p->next)
 281                                 *tail = &new_pat->next;
 282                         p->next = new_pat;
 283                         *nl = '\0';
 284                         p->patternlen -= len;
 285                 }
 286                 break;
 287         default:
 288                 break;
 289         }
 290 }
 291
 292 void append_header_grep_pattern(struct grep_opt *opt,
 293                                 enum grep_header_field field, const char *pat)
 294 {
 295         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 296                                              GREP_PATTERN_HEAD, field);
 297         if (field == GREP_HEADER_REFLOG)
 298                 opt->use_reflog_filter = 1;
 299         do_append_grep_pat(&opt->header_tail, p);
 300 }
 301
 302 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 303                          const char *origin, int no, enum grep_pat_token t)
 304 {
 305         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 306 }
 307
 308 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 309                      const char *origin, int no, enum grep_pat_token t)
 310 {
 311         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 312         do_append_grep_pat(&opt->pattern_tail, p);
 313 }
 314
 315 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 316 {
 317         struct grep_pat *pat;
 318         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 319         *ret = *opt;
 320
 321         ret->pattern_list = NULL;
 322         ret->pattern_tail = &ret->pattern_list;
 323
 324         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 325         {
 326                 if(pat->token == GREP_PATTERN_HEAD)
 327                         append_header_grep_pattern(ret, pat->field,
 328                                                    pat->pattern);
 329                 else
 330                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 331                                         pat->origin, pat->no, pat->token);
 332         }
 333
 334         return ret;
 335 }
 336
 337 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 338                 const char *error)
 339 {
 340         char where[1024];
 341
 342         if (p->no)
 343                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 344         else if (p->origin)
 345                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 346         else
 347                 where[0] = 0;
 348
 349         die("%s'%s': %s", where, p->pattern, error);
 350 }
 351
 352 static int is_fixed(const char *s, size_t len)
 353 {
 354         size_t i;
 355
 356         for (i = 0; i < len; i++) {
 357                 if (is_regex_special(s[i]))
 358                         return 0;
 359         }
 360
 361         return 1;
 362 }
 363
 364 static int has_null(const char *s, size_t len)
 365 {
 366         /*
 367          * regcomp cannot accept patterns with NULs so when using it
 368          * we consider any pattern containing a NUL fixed.
 369          */
 370         if (memchr(s, 0, len))
 371                 return 1;
 372
 373         return 0;
 374 }
 375
 376 #ifdef USE_LIBPCRE1
 377 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 378 {
 379         const char *error;
 380         int erroffset;
 381         int options = PCRE_MULTILINE;
 382
 383         if (opt->ignore_case) {
 384                 if (has_non_ascii(p->pattern))
 385                         p->pcre1_tables = pcre_maketables();
 386                 options |= PCRE_CASELESS;
 387         }
 388         if (is_utf8_locale() && has_non_ascii(p->pattern))
 389                 options |= PCRE_UTF8;
 390
 391         p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 392                                       p->pcre1_tables);
 393         if (!p->pcre1_regexp)
 394                 compile_regexp_failed(p, error);
 395
 396         p->pcre1_extra_info = pcre_study(p->pcre1_regexp, GIT_PCRE_STUDY_JIT_COMPILE, &error);
 397         if (!p->pcre1_extra_info && error)
 398                 die("%s", error);
 399
 400 #ifdef GIT_PCRE1_USE_JIT
 401         pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
 402         if (p->pcre1_jit_on == 1) {
 403                 p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
 404                 if (!p->pcre1_jit_stack)
 405                         die("Couldn't allocate PCRE JIT stack");
 406                 pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
 407         } else if (p->pcre1_jit_on != 0) {
 408                 die("BUG: The pcre1_jit_on variable should be 0 or 1, not %d",
 409                     p->pcre1_jit_on);
 410         }
 411 #endif
 412 }
 413
 414 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 415                 regmatch_t *match, int eflags)
 416 {
 417         int ovector[30], ret, flags = 0;
 418
 419         if (eflags & REG_NOTBOL)
 420                 flags |= PCRE_NOTBOL;
 421
 422 #ifdef GIT_PCRE1_USE_JIT
 423         if (p->pcre1_jit_on) {
 424                 ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 425                                     eol - line, 0, flags, ovector,
 426                                     ARRAY_SIZE(ovector), p->pcre1_jit_stack);
 427         } else
 428 #endif
 429         {
 430                 ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 431                                 eol - line, 0, flags, ovector,
 432                                 ARRAY_SIZE(ovector));
 433         }
 434
 435         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 436                 die("pcre_exec failed with error code %d", ret);
 437         if (ret > 0) {
 438                 ret = 0;
 439                 match->rm_so = ovector[0];
 440                 match->rm_eo = ovector[1];
 441         }
 442
 443         return ret;
 444 }
 445
 446 static void free_pcre1_regexp(struct grep_pat *p)
 447 {
 448         pcre_free(p->pcre1_regexp);
 449 #ifdef GIT_PCRE1_USE_JIT
 450         if (p->pcre1_jit_on) {
 451                 pcre_free_study(p->pcre1_extra_info);
 452                 pcre_jit_stack_free(p->pcre1_jit_stack);
 453         } else
 454 #endif
 455         {
 456                 pcre_free(p->pcre1_extra_info);
 457         }
 458         pcre_free((void *)p->pcre1_tables);
 459 }
 460 #else /* !USE_LIBPCRE1 */
 461 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 462 {
 463         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 464 }
 465
 466 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 467                 regmatch_t *match, int eflags)
 468 {
 469         return 1;
 470 }
 471
 472 static void free_pcre1_regexp(struct grep_pat *p)
 473 {
 474 }
 475 #endif /* !USE_LIBPCRE1 */
 476
 477 #ifdef USE_LIBPCRE2
 478 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 479 {
 480         int error;
 481         PCRE2_UCHAR errbuf[256];
 482         PCRE2_SIZE erroffset;
 483         int options = PCRE2_MULTILINE;
 484         const uint8_t *character_tables = NULL;
 485         int jitret;
 486         int patinforet;
 487         size_t jitsizearg;
 488
 489         assert(opt->pcre2);
 490
 491         p->pcre2_compile_context = NULL;
 492
 493         if (opt->ignore_case) {
 494                 if (has_non_ascii(p->pattern)) {
 495                         character_tables = pcre2_maketables(NULL);
 496                         p->pcre2_compile_context = pcre2_compile_context_create(NULL);
 497                         pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
 498                 }
 499                 options |= PCRE2_CASELESS;
 500         }
 501         if (is_utf8_locale() && has_non_ascii(p->pattern))
 502                 options |= PCRE2_UTF;
 503
 504         p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
 505                                          p->patternlen, options, &error, &erroffset,
 506                                          p->pcre2_compile_context);
 507
 508         if (p->pcre2_pattern) {
 509                 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
 510                 if (!p->pcre2_match_data)
 511                         die("Couldn't allocate PCRE2 match data");
 512         } else {
 513                 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
 514                 compile_regexp_failed(p, (const char *)&errbuf);
 515         }
 516
 517         pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
 518         if (p->pcre2_jit_on == 1) {
 519                 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
 520                 if (jitret)
 521                         die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
 522
 523                 /*
 524                  * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
 525                  * tells us whether the library itself supports JIT,
 526                  * but to see whether we're going to be actually using
 527                  * JIT we need to extract PCRE2_INFO_JITSIZE from the
 528                  * pattern *after* we do pcre2_jit_compile() above.
 529                  *
 530                  * This is because if the pattern contains the
 531                  * (*NO_JIT) verb (see pcre2syntax(3))
 532                  * pcre2_jit_compile() will exit early with 0. If we
 533                  * then proceed to call pcre2_jit_match() further down
 534                  * the line instead of pcre2_match() we'll either
 535                  * segfault (pre PCRE 10.31) or run into a fatal error
 536                  * (post PCRE2 10.31)
 537                  */
 538                 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
 539                 if (patinforet)
 540                         BUG("pcre2_pattern_info() failed: %d", patinforet);
 541                 if (jitsizearg == 0) {
 542                         p->pcre2_jit_on = 0;
 543                         return;
 544                 }
 545
 546                 p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
 547                 if (!p->pcre2_jit_stack)
 548                         die("Couldn't allocate PCRE2 JIT stack");
 549                 p->pcre2_match_context = pcre2_match_context_create(NULL);
 550                 if (!p->pcre2_match_context)
 551                         die("Couldn't allocate PCRE2 match context");
 552                 pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
 553         } else if (p->pcre2_jit_on != 0) {
 554                 die("BUG: The pcre2_jit_on variable should be 0 or 1, not %d",
 555                     p->pcre1_jit_on);
 556         }
 557 }
 558
 559 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 560                 regmatch_t *match, int eflags)
 561 {
 562         int ret, flags = 0;
 563         PCRE2_SIZE *ovector;
 564         PCRE2_UCHAR errbuf[256];
 565
 566         if (eflags & REG_NOTBOL)
 567                 flags |= PCRE2_NOTBOL;
 568
 569         if (p->pcre2_jit_on)
 570                 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 571                                       eol - line, 0, flags, p->pcre2_match_data,
 572                                       NULL);
 573         else
 574                 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 575                                   eol - line, 0, flags, p->pcre2_match_data,
 576                                   NULL);
 577
 578         if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 579                 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 580                 die("%s failed with error code %d: %s",
 581                     (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 582                     errbuf);
 583         }
 584         if (ret > 0) {
 585                 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 586                 ret = 0;
 587                 match->rm_so = (int)ovector[0];
 588                 match->rm_eo = (int)ovector[1];
 589         }
 590
 591         return ret;
 592 }
 593
 594 static void free_pcre2_pattern(struct grep_pat *p)
 595 {
 596         pcre2_compile_context_free(p->pcre2_compile_context);
 597         pcre2_code_free(p->pcre2_pattern);
 598         pcre2_match_data_free(p->pcre2_match_data);
 599         pcre2_jit_stack_free(p->pcre2_jit_stack);
 600         pcre2_match_context_free(p->pcre2_match_context);
 601 }
 602 #else /* !USE_LIBPCRE2 */
 603 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 604 {
 605         /*
 606          * Unreachable until USE_LIBPCRE2 becomes synonymous with
 607          * USE_LIBPCRE. See the sibling comment in
 608          * grep_set_pattern_type_option().
 609          */
 610         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 611 }
 612
 613 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 614                 regmatch_t *match, int eflags)
 615 {
 616         return 1;
 617 }
 618
 619 static void free_pcre2_pattern(struct grep_pat *p)
 620 {
 621 }
 622 #endif /* !USE_LIBPCRE2 */
 623
 624 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 625 {
 626         struct strbuf sb = STRBUF_INIT;
 627         int err;
 628         int regflags = 0;
 629
 630         basic_regex_quote_buf(&sb, p->pattern);
 631         if (opt->ignore_case)
 632                 regflags |= REG_ICASE;
 633         err = regcomp(&p->regexp, sb.buf, regflags);
 634         if (opt->debug)
 635                 fprintf(stderr, "fixed %s\n", sb.buf);
 636         strbuf_release(&sb);
 637         if (err) {
 638                 char errbuf[1024];
 639                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 640                 regfree(&p->regexp);
 641                 compile_regexp_failed(p, errbuf);
 642         }
 643 }
 644
 645 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 646 {
 647         int ascii_only;
 648         int err;
 649         int regflags = REG_NEWLINE;
 650
 651         p->word_regexp = opt->word_regexp;
 652         p->ignore_case = opt->ignore_case;
 653         ascii_only     = !has_non_ascii(p->pattern);
 654
 655         /*
 656          * Even when -F (fixed) asks us to do a non-regexp search, we
 657          * may not be able to correctly case-fold when -i
 658          * (ignore-case) is asked (in which case, we'll synthesize a
 659          * regexp to match the pattern that matches regexp special
 660          * characters literally, while ignoring case differences).  On
 661          * the other hand, even without -F, if the pattern does not
 662          * have any regexp special characters and there is no need for
 663          * case-folding search, we can internally turn it into a
 664          * simple string match using kws.  p->fixed tells us if we
 665          * want to use kws.
 666          */
 667         if (opt->fixed ||
 668             has_null(p->pattern, p->patternlen) ||
 669             is_fixed(p->pattern, p->patternlen))
 670                 p->fixed = !p->ignore_case || ascii_only;
 671
 672         if (p->fixed) {
 673                 p->kws = kwsalloc(p->ignore_case ? tolower_trans_tbl : NULL);
 674                 kwsincr(p->kws, p->pattern, p->patternlen);
 675                 kwsprep(p->kws);
 676                 return;
 677         } else if (opt->fixed) {
 678                 /*
 679                  * We come here when the pattern has the non-ascii
 680                  * characters we cannot case-fold, and asked to
 681                  * ignore-case.
 682                  */
 683                 compile_fixed_regexp(p, opt);
 684                 return;
 685         }
 686
 687         if (opt->pcre2) {
 688                 compile_pcre2_pattern(p, opt);
 689                 return;
 690         }
 691
 692         if (opt->pcre1) {
 693                 compile_pcre1_regexp(p, opt);
 694                 return;
 695         }
 696
 697         if (p->ignore_case)
 698                 regflags |= REG_ICASE;
 699         if (opt->extended_regexp_option)
 700                 regflags |= REG_EXTENDED;
 701         err = regcomp(&p->regexp, p->pattern, regflags);
 702         if (err) {
 703                 char errbuf[1024];
 704                 regerror(err, &p->regexp, errbuf, 1024);
 705                 regfree(&p->regexp);
 706                 compile_regexp_failed(p, errbuf);
 707         }
 708 }
 709
 710 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 711 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 712 {
 713         struct grep_pat *p;
 714         struct grep_expr *x;
 715
 716         p = *list;
 717         if (!p)
 718                 return NULL;
 719         switch (p->token) {
 720         case GREP_PATTERN: /* atom */
 721         case GREP_PATTERN_HEAD:
 722         case GREP_PATTERN_BODY:
 723                 x = xcalloc(1, sizeof (struct grep_expr));
 724                 x->node = GREP_NODE_ATOM;
 725                 x->u.atom = p;
 726                 *list = p->next;
 727                 return x;
 728         case GREP_OPEN_PAREN:
 729                 *list = p->next;
 730                 x = compile_pattern_or(list);
 731                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 732                         die("unmatched parenthesis");
 733                 *list = (*list)->next;
 734                 return x;
 735         default:
 736                 return NULL;
 737         }
 738 }
 739
 740 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 741 {
 742         struct grep_pat *p;
 743         struct grep_expr *x;
 744
 745         p = *list;
 746         if (!p)
 747                 return NULL;
 748         switch (p->token) {
 749         case GREP_NOT:
 750                 if (!p->next)
 751                         die("--not not followed by pattern expression");
 752                 *list = p->next;
 753                 x = xcalloc(1, sizeof (struct grep_expr));
 754                 x->node = GREP_NODE_NOT;
 755                 x->u.unary = compile_pattern_not(list);
 756                 if (!x->u.unary)
 757                         die("--not followed by non pattern expression");
 758                 return x;
 759         default:
 760                 return compile_pattern_atom(list);
 761         }
 762 }
 763
 764 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 765 {
 766         struct grep_pat *p;
 767         struct grep_expr *x, *y, *z;
 768
 769         x = compile_pattern_not(list);
 770         p = *list;
 771         if (p && p->token == GREP_AND) {
 772                 if (!p->next)
 773                         die("--and not followed by pattern expression");
 774                 *list = p->next;
 775                 y = compile_pattern_and(list);
 776                 if (!y)
 777                         die("--and not followed by pattern expression");
 778                 z = xcalloc(1, sizeof (struct grep_expr));
 779                 z->node = GREP_NODE_AND;
 780                 z->u.binary.left = x;
 781                 z->u.binary.right = y;
 782                 return z;
 783         }
 784         return x;
 785 }
 786
 787 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 788 {
 789         struct grep_pat *p;
 790         struct grep_expr *x, *y, *z;
 791
 792         x = compile_pattern_and(list);
 793         p = *list;
 794         if (x && p && p->token != GREP_CLOSE_PAREN) {
 795                 y = compile_pattern_or(list);
 796                 if (!y)
 797                         die("not a pattern expression %s", p->pattern);
 798                 z = xcalloc(1, sizeof (struct grep_expr));
 799                 z->node = GREP_NODE_OR;
 800                 z->u.binary.left = x;
 801                 z->u.binary.right = y;
 802                 return z;
 803         }
 804         return x;
 805 }
 806
 807 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 808 {
 809         return compile_pattern_or(list);
 810 }
 811
 812 static void indent(int in)
 813 {
 814         while (in-- > 0)
 815                 fputc(' ', stderr);
 816 }
 817
 818 static void dump_grep_pat(struct grep_pat *p)
 819 {
 820         switch (p->token) {
 821         case GREP_AND: fprintf(stderr, "*and*"); break;
 822         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 823         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 824         case GREP_NOT: fprintf(stderr, "*not*"); break;
 825         case GREP_OR: fprintf(stderr, "*or*"); break;
 826
 827         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 828         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 829         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 830         }
 831
 832         switch (p->token) {
 833         default: break;
 834         case GREP_PATTERN_HEAD:
 835                 fprintf(stderr, "<head %d>", p->field); break;
 836         case GREP_PATTERN_BODY:
 837                 fprintf(stderr, "<body>"); break;
 838         }
 839         switch (p->token) {
 840         default: break;
 841         case GREP_PATTERN_HEAD:
 842         case GREP_PATTERN_BODY:
 843         case GREP_PATTERN:
 844                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 845                 break;
 846         }
 847         fputc('\n', stderr);
 848 }
 849
 850 static void dump_grep_expression_1(struct grep_expr *x, int in)
 851 {
 852         indent(in);
 853         switch (x->node) {
 854         case GREP_NODE_TRUE:
 855                 fprintf(stderr, "true\n");
 856                 break;
 857         case GREP_NODE_ATOM:
 858                 dump_grep_pat(x->u.atom);
 859                 break;
 860         case GREP_NODE_NOT:
 861                 fprintf(stderr, "(not\n");
 862                 dump_grep_expression_1(x->u.unary, in+1);
 863                 indent(in);
 864                 fprintf(stderr, ")\n");
 865                 break;
 866         case GREP_NODE_AND:
 867                 fprintf(stderr, "(and\n");
 868                 dump_grep_expression_1(x->u.binary.left, in+1);
 869                 dump_grep_expression_1(x->u.binary.right, in+1);
 870                 indent(in);
 871                 fprintf(stderr, ")\n");
 872                 break;
 873         case GREP_NODE_OR:
 874                 fprintf(stderr, "(or\n");
 875                 dump_grep_expression_1(x->u.binary.left, in+1);
 876                 dump_grep_expression_1(x->u.binary.right, in+1);
 877                 indent(in);
 878                 fprintf(stderr, ")\n");
 879                 break;
 880         }
 881 }
 882
 883 static void dump_grep_expression(struct grep_opt *opt)
 884 {
 885         struct grep_expr *x = opt->pattern_expression;
 886
 887         if (opt->all_match)
 888                 fprintf(stderr, "[all-match]\n");
 889         dump_grep_expression_1(x, 0);
 890         fflush(NULL);
 891 }
 892
 893 static struct grep_expr *grep_true_expr(void)
 894 {
 895         struct grep_expr *z = xcalloc(1, sizeof(*z));
 896         z->node = GREP_NODE_TRUE;
 897         return z;
 898 }
 899
 900 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 901 {
 902         struct grep_expr *z = xcalloc(1, sizeof(*z));
 903         z->node = GREP_NODE_OR;
 904         z->u.binary.left = left;
 905         z->u.binary.right = right;
 906         return z;
 907 }
 908
 909 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 910 {
 911         struct grep_pat *p;
 912         struct grep_expr *header_expr;
 913         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 914         enum grep_header_field fld;
 915
 916         if (!opt->header_list)
 917                 return NULL;
 918
 919         for (p = opt->header_list; p; p = p->next) {
 920                 if (p->token != GREP_PATTERN_HEAD)
 921                         die("BUG: a non-header pattern in grep header list.");
 922                 if (p->field < GREP_HEADER_FIELD_MIN ||
 923                     GREP_HEADER_FIELD_MAX <= p->field)
 924                         die("BUG: unknown header field %d", p->field);
 925                 compile_regexp(p, opt);
 926         }
 927
 928         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 929                 header_group[fld] = NULL;
 930
 931         for (p = opt->header_list; p; p = p->next) {
 932                 struct grep_expr *h;
 933                 struct grep_pat *pp = p;
 934
 935                 h = compile_pattern_atom(&pp);
 936                 if (!h || pp != p->next)
 937                         die("BUG: malformed header expr");
 938                 if (!header_group[p->field]) {
 939                         header_group[p->field] = h;
 940                         continue;
 941                 }
 942                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 943         }
 944
 945         header_expr = NULL;
 946
 947         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 948                 if (!header_group[fld])
 949                         continue;
 950                 if (!header_expr)
 951                         header_expr = grep_true_expr();
 952                 header_expr = grep_or_expr(header_group[fld], header_expr);
 953         }
 954         return header_expr;
 955 }
 956
 957 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 958 {
 959         struct grep_expr *z = x;
 960
 961         while (x) {
 962                 assert(x->node == GREP_NODE_OR);
 963                 if (x->u.binary.right &&
 964                     x->u.binary.right->node == GREP_NODE_TRUE) {
 965                         x->u.binary.right = y;
 966                         break;
 967                 }
 968                 x = x->u.binary.right;
 969         }
 970         return z;
 971 }
 972
 973 static void compile_grep_patterns_real(struct grep_opt *opt)
 974 {
 975         struct grep_pat *p;
 976         struct grep_expr *header_expr = prep_header_patterns(opt);
 977
 978         for (p = opt->pattern_list; p; p = p->next) {
 979                 switch (p->token) {
 980                 case GREP_PATTERN: /* atom */
 981                 case GREP_PATTERN_HEAD:
 982                 case GREP_PATTERN_BODY:
 983                         compile_regexp(p, opt);
 984                         break;
 985                 default:
 986                         opt->extended = 1;
 987                         break;
 988                 }
 989         }
 990
 991         if (opt->all_match || header_expr)
 992                 opt->extended = 1;
 993         else if (!opt->extended && !opt->debug)
 994                 return;
 995
 996         p = opt->pattern_list;
 997         if (p)
 998                 opt->pattern_expression = compile_pattern_expr(&p);
 999         if (p)
1000                 die("incomplete pattern expression: %s", p->pattern);
1001
1002         if (!header_expr)
1003                 return;
1004
1005         if (!opt->pattern_expression)
1006                 opt->pattern_expression = header_expr;
1007         else if (opt->all_match)
1008                 opt->pattern_expression = grep_splice_or(header_expr,
1009                                                          opt->pattern_expression);
1010         else
1011                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
1012                                                        header_expr);
1013         opt->all_match = 1;
1014 }
1015
1016 void compile_grep_patterns(struct grep_opt *opt)
1017 {
1018         compile_grep_patterns_real(opt);
1019         if (opt->debug)
1020                 dump_grep_expression(opt);
1021 }
1022
1023 static void free_pattern_expr(struct grep_expr *x)
1024 {
1025         switch (x->node) {
1026         case GREP_NODE_TRUE:
1027         case GREP_NODE_ATOM:
1028                 break;
1029         case GREP_NODE_NOT:
1030                 free_pattern_expr(x->u.unary);
1031                 break;
1032         case GREP_NODE_AND:
1033         case GREP_NODE_OR:
1034                 free_pattern_expr(x->u.binary.left);
1035                 free_pattern_expr(x->u.binary.right);
1036                 break;
1037         }
1038         free(x);
1039 }
1040
1041 void free_grep_patterns(struct grep_opt *opt)
1042 {
1043         struct grep_pat *p, *n;
1044
1045         for (p = opt->pattern_list; p; p = n) {
1046                 n = p->next;
1047                 switch (p->token) {
1048                 case GREP_PATTERN: /* atom */
1049                 case GREP_PATTERN_HEAD:
1050                 case GREP_PATTERN_BODY:
1051                         if (p->kws)
1052                                 kwsfree(p->kws);
1053                         else if (p->pcre1_regexp)
1054                                 free_pcre1_regexp(p);
1055                         else if (p->pcre2_pattern)
1056                                 free_pcre2_pattern(p);
1057                         else
1058                                 regfree(&p->regexp);
1059                         free(p->pattern);
1060                         break;
1061                 default:
1062                         break;
1063                 }
1064                 free(p);
1065         }
1066
1067         if (!opt->extended)
1068                 return;
1069         free_pattern_expr(opt->pattern_expression);
1070 }
1071
1072 static char *end_of_line(char *cp, unsigned long *left)
1073 {
1074         unsigned long l = *left;
1075         while (l && *cp != '\n') {
1076                 l--;
1077                 cp++;
1078         }
1079         *left = l;
1080         return cp;
1081 }
1082
1083 static int word_char(char ch)
1084 {
1085         return isalnum(ch) || ch == '_';
1086 }
1087
1088 static void output_color(struct grep_opt *opt, const void *data, size_t size,
1089                          const char *color)
1090 {
1091         if (want_color(opt->color) && color && color[0]) {
1092                 opt->output(opt, color, strlen(color));
1093                 opt->output(opt, data, size);
1094                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1095         } else
1096                 opt->output(opt, data, size);
1097 }
1098
1099 static void output_sep(struct grep_opt *opt, char sign)
1100 {
1101         if (opt->null_following_name)
1102                 opt->output(opt, "\0", 1);
1103         else
1104                 output_color(opt, &sign, 1, opt->color_sep);
1105 }
1106
1107 static void show_name(struct grep_opt *opt, const char *name)
1108 {
1109         output_color(opt, name, strlen(name), opt->color_filename);
1110         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1111 }
1112
1113 static int fixmatch(struct grep_pat *p, char *line, char *eol,
1114                     regmatch_t *match)
1115 {
1116         struct kwsmatch kwsm;
1117         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
1118         if (offset == -1) {
1119                 match->rm_so = match->rm_eo = -1;
1120                 return REG_NOMATCH;
1121         } else {
1122                 match->rm_so = offset;
1123                 match->rm_eo = match->rm_so + kwsm.size[0];
1124                 return 0;
1125         }
1126 }
1127
1128 static int patmatch(struct grep_pat *p, char *line, char *eol,
1129                     regmatch_t *match, int eflags)
1130 {
1131         int hit;
1132
1133         if (p->fixed)
1134                 hit = !fixmatch(p, line, eol, match);
1135         else if (p->pcre1_regexp)
1136                 hit = !pcre1match(p, line, eol, match, eflags);
1137         else if (p->pcre2_pattern)
1138                 hit = !pcre2match(p, line, eol, match, eflags);
1139         else
1140                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1141                                    eflags);
1142
1143         return hit;
1144 }
1145
1146 static int strip_timestamp(char *bol, char **eol_p)
1147 {
1148         char *eol = *eol_p;
1149         int ch;
1150
1151         while (bol < --eol) {
1152                 if (*eol != '>')
1153                         continue;
1154                 *eol_p = ++eol;
1155                 ch = *eol;
1156                 *eol = '\0';
1157                 return ch;
1158         }
1159         return 0;
1160 }
1161
1162 static struct {
1163         const char *field;
1164         size_t len;
1165 } header_field[] = {
1166         { "author ", 7 },
1167         { "committer ", 10 },
1168         { "reflog ", 7 },
1169 };
1170
1171 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1172                              enum grep_context ctx,
1173                              regmatch_t *pmatch, int eflags)
1174 {
1175         int hit = 0;
1176         int saved_ch = 0;
1177         const char *start = bol;
1178
1179         if ((p->token != GREP_PATTERN) &&
1180             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1181                 return 0;
1182
1183         if (p->token == GREP_PATTERN_HEAD) {
1184                 const char *field;
1185                 size_t len;
1186                 assert(p->field < ARRAY_SIZE(header_field));
1187                 field = header_field[p->field].field;
1188                 len = header_field[p->field].len;
1189                 if (strncmp(bol, field, len))
1190                         return 0;
1191                 bol += len;
1192                 switch (p->field) {
1193                 case GREP_HEADER_AUTHOR:
1194                 case GREP_HEADER_COMMITTER:
1195                         saved_ch = strip_timestamp(bol, &eol);
1196                         break;
1197                 default:
1198                         break;
1199                 }
1200         }
1201
1202  again:
1203         hit = patmatch(p, bol, eol, pmatch, eflags);
1204
1205         if (hit && p->word_regexp) {
1206                 if ((pmatch[0].rm_so < 0) ||
1207                     (eol - bol) < pmatch[0].rm_so ||
1208                     (pmatch[0].rm_eo < 0) ||
1209                     (eol - bol) < pmatch[0].rm_eo)
1210                         die("regexp returned nonsense");
1211
1212                 /* Match beginning must be either beginning of the
1213                  * line, or at word boundary (i.e. the last char must
1214                  * not be a word char).  Similarly, match end must be
1215                  * either end of the line, or at word boundary
1216                  * (i.e. the next char must not be a word char).
1217                  */
1218                 if ( ((pmatch[0].rm_so == 0) ||
1219                       !word_char(bol[pmatch[0].rm_so-1])) &&
1220                      ((pmatch[0].rm_eo == (eol-bol)) ||
1221                       !word_char(bol[pmatch[0].rm_eo])) )
1222                         ;
1223                 else
1224                         hit = 0;
1225
1226                 /* Words consist of at least one character. */
1227                 if (pmatch->rm_so == pmatch->rm_eo)
1228                         hit = 0;
1229
1230                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1231                         /* There could be more than one match on the
1232                          * line, and the first match might not be
1233                          * strict word match.  But later ones could be!
1234                          * Forward to the next possible start, i.e. the
1235                          * next position following a non-word char.
1236                          */
1237                         bol = pmatch[0].rm_so + bol + 1;
1238                         while (word_char(bol[-1]) && bol < eol)
1239                                 bol++;
1240                         eflags |= REG_NOTBOL;
1241                         if (bol < eol)
1242                                 goto again;
1243                 }
1244         }
1245         if (p->token == GREP_PATTERN_HEAD && saved_ch)
1246                 *eol = saved_ch;
1247         if (hit) {
1248                 pmatch[0].rm_so += bol - start;
1249                 pmatch[0].rm_eo += bol - start;
1250         }
1251         return hit;
1252 }
1253
1254 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1255                            enum grep_context ctx, int collect_hits)
1256 {
1257         int h = 0;
1258         regmatch_t match;
1259
1260         if (!x)
1261                 die("Not a valid grep expression");
1262         switch (x->node) {
1263         case GREP_NODE_TRUE:
1264                 h = 1;
1265                 break;
1266         case GREP_NODE_ATOM:
1267                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1268                 break;
1269         case GREP_NODE_NOT:
1270                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1271                 break;
1272         case GREP_NODE_AND:
1273                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1274                         return 0;
1275                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1276                 break;
1277         case GREP_NODE_OR:
1278                 if (!collect_hits)
1279                         return (match_expr_eval(x->u.binary.left,
1280                                                 bol, eol, ctx, 0) ||
1281                                 match_expr_eval(x->u.binary.right,
1282                                                 bol, eol, ctx, 0));
1283                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1284                 x->u.binary.left->hit |= h;
1285                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1286                 break;
1287         default:
1288                 die("Unexpected node type (internal error) %d", x->node);
1289         }
1290         if (collect_hits)
1291                 x->hit |= h;
1292         return h;
1293 }
1294
1295 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1296                       enum grep_context ctx, int collect_hits)
1297 {
1298         struct grep_expr *x = opt->pattern_expression;
1299         return match_expr_eval(x, bol, eol, ctx, collect_hits);
1300 }
1301
1302 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1303                       enum grep_context ctx, int collect_hits)
1304 {
1305         struct grep_pat *p;
1306         regmatch_t match;
1307
1308         if (opt->extended)
1309                 return match_expr(opt, bol, eol, ctx, collect_hits);
1310
1311         /* we do not call with collect_hits without being extended */
1312         for (p = opt->pattern_list; p; p = p->next) {
1313                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1314                         return 1;
1315         }
1316         return 0;
1317 }
1318
1319 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1320                               enum grep_context ctx,
1321                               regmatch_t *pmatch, int eflags)
1322 {
1323         regmatch_t match;
1324
1325         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1326                 return 0;
1327         if (match.rm_so < 0 || match.rm_eo < 0)
1328                 return 0;
1329         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1330                 if (match.rm_so > pmatch->rm_so)
1331                         return 1;
1332                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1333                         return 1;
1334         }
1335         pmatch->rm_so = match.rm_so;
1336         pmatch->rm_eo = match.rm_eo;
1337         return 1;
1338 }
1339
1340 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1341                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1342 {
1343         struct grep_pat *p;
1344         int hit = 0;
1345
1346         pmatch->rm_so = pmatch->rm_eo = -1;
1347         if (bol < eol) {
1348                 for (p = opt->pattern_list; p; p = p->next) {
1349                         switch (p->token) {
1350                         case GREP_PATTERN: /* atom */
1351                         case GREP_PATTERN_HEAD:
1352                         case GREP_PATTERN_BODY:
1353                                 hit |= match_next_pattern(p, bol, eol, ctx,
1354                                                           pmatch, eflags);
1355                                 break;
1356                         default:
1357                                 break;
1358                         }
1359                 }
1360         }
1361         return hit;
1362 }
1363
1364 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1365                       const char *name, unsigned lno, char sign)
1366 {
1367         int rest = eol - bol;
1368         const char *match_color, *line_color = NULL;
1369
1370         if (opt->file_break && opt->last_shown == 0) {
1371                 if (opt->show_hunk_mark)
1372                         opt->output(opt, "\n", 1);
1373         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1374                 if (opt->last_shown == 0) {
1375                         if (opt->show_hunk_mark) {
1376                                 output_color(opt, "--", 2, opt->color_sep);
1377                                 opt->output(opt, "\n", 1);
1378                         }
1379                 } else if (lno > opt->last_shown + 1) {
1380                         output_color(opt, "--", 2, opt->color_sep);
1381                         opt->output(opt, "\n", 1);
1382                 }
1383         }
1384         if (opt->heading && opt->last_shown == 0) {
1385                 output_color(opt, name, strlen(name), opt->color_filename);
1386                 opt->output(opt, "\n", 1);
1387         }
1388         opt->last_shown = lno;
1389
1390         if (!opt->heading && opt->pathname) {
1391                 output_color(opt, name, strlen(name), opt->color_filename);
1392                 output_sep(opt, sign);
1393         }
1394         if (opt->linenum) {
1395                 char buf[32];
1396                 xsnprintf(buf, sizeof(buf), "%d", lno);
1397                 output_color(opt, buf, strlen(buf), opt->color_lineno);
1398                 output_sep(opt, sign);
1399         }
1400         if (opt->color) {
1401                 regmatch_t match;
1402                 enum grep_context ctx = GREP_CONTEXT_BODY;
1403                 int ch = *eol;
1404                 int eflags = 0;
1405
1406                 if (sign == ':')
1407                         match_color = opt->color_match_selected;
1408                 else
1409                         match_color = opt->color_match_context;
1410                 if (sign == ':')
1411                         line_color = opt->color_selected;
1412                 else if (sign == '-')
1413                         line_color = opt->color_context;
1414                 else if (sign == '=')
1415                         line_color = opt->color_function;
1416                 *eol = '\0';
1417                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1418                         if (match.rm_so == match.rm_eo)
1419                                 break;
1420
1421                         output_color(opt, bol, match.rm_so, line_color);
1422                         output_color(opt, bol + match.rm_so,
1423                                      match.rm_eo - match.rm_so, match_color);
1424                         bol += match.rm_eo;
1425                         rest -= match.rm_eo;
1426                         eflags = REG_NOTBOL;
1427                 }
1428                 *eol = ch;
1429         }
1430         output_color(opt, bol, rest, line_color);
1431         opt->output(opt, "\n", 1);
1432 }
1433
1434 #ifndef NO_PTHREADS
1435 int grep_use_locks;
1436
1437 /*
1438  * This lock protects access to the gitattributes machinery, which is
1439  * not thread-safe.
1440  */
1441 pthread_mutex_t grep_attr_mutex;
1442
1443 static inline void grep_attr_lock(void)
1444 {
1445         if (grep_use_locks)
1446                 pthread_mutex_lock(&grep_attr_mutex);
1447 }
1448
1449 static inline void grep_attr_unlock(void)
1450 {
1451         if (grep_use_locks)
1452                 pthread_mutex_unlock(&grep_attr_mutex);
1453 }
1454
1455 /*
1456  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1457  */
1458 pthread_mutex_t grep_read_mutex;
1459
1460 #else
1461 #define grep_attr_lock()
1462 #define grep_attr_unlock()
1463 #endif
1464
1465 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1466 {
1467         xdemitconf_t *xecfg = opt->priv;
1468         if (xecfg && !xecfg->find_func) {
1469                 grep_source_load_driver(gs);
1470                 if (gs->driver->funcname.pattern) {
1471                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1472                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1473                 } else {
1474                         xecfg = opt->priv = NULL;
1475                 }
1476         }
1477
1478         if (xecfg) {
1479                 char buf[1];
1480                 return xecfg->find_func(bol, eol - bol, buf, 1,
1481                                         xecfg->find_func_priv) >= 0;
1482         }
1483
1484         if (bol == eol)
1485                 return 0;
1486         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1487                 return 1;
1488         return 0;
1489 }
1490
1491 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1492                                char *bol, unsigned lno)
1493 {
1494         while (bol > gs->buf) {
1495                 char *eol = --bol;
1496
1497                 while (bol > gs->buf && bol[-1] != '\n')
1498                         bol--;
1499                 lno--;
1500
1501                 if (lno <= opt->last_shown)
1502                         break;
1503
1504                 if (match_funcname(opt, gs, bol, eol)) {
1505                         show_line(opt, bol, eol, gs->name, lno, '=');
1506                         break;
1507                 }
1508         }
1509 }
1510
1511 static int is_empty_line(const char *bol, const char *eol);
1512
1513 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1514                              char *bol, char *end, unsigned lno)
1515 {
1516         unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1517         int funcname_needed = !!opt->funcname, comment_needed = 0;
1518
1519         if (opt->pre_context < lno)
1520                 from = lno - opt->pre_context;
1521         if (from <= opt->last_shown)
1522                 from = opt->last_shown + 1;
1523         orig_from = from;
1524         if (opt->funcbody) {
1525                 if (match_funcname(opt, gs, bol, end))
1526                         comment_needed = 1;
1527                 else
1528                         funcname_needed = 1;
1529                 from = opt->last_shown + 1;
1530         }
1531
1532         /* Rewind. */
1533         while (bol > gs->buf && cur > from) {
1534                 char *next_bol = bol;
1535                 char *eol = --bol;
1536
1537                 while (bol > gs->buf && bol[-1] != '\n')
1538                         bol--;
1539                 cur--;
1540                 if (comment_needed && (is_empty_line(bol, eol) ||
1541                                        match_funcname(opt, gs, bol, eol))) {
1542                         comment_needed = 0;
1543                         from = orig_from;
1544                         if (cur < from) {
1545                                 cur++;
1546                                 bol = next_bol;
1547                                 break;
1548                         }
1549                 }
1550                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1551                         funcname_lno = cur;
1552                         funcname_needed = 0;
1553                         if (opt->funcbody)
1554                                 comment_needed = 1;
1555                         else
1556                                 from = orig_from;
1557                 }
1558         }
1559
1560         /* We need to look even further back to find a function signature. */
1561         if (opt->funcname && funcname_needed)
1562                 show_funcname_line(opt, gs, bol, cur);
1563
1564         /* Back forward. */
1565         while (cur < lno) {
1566                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1567
1568                 while (*eol != '\n')
1569                         eol++;
1570                 show_line(opt, bol, eol, gs->name, cur, sign);
1571                 bol = eol + 1;
1572                 cur++;
1573         }
1574 }
1575
1576 static int should_lookahead(struct grep_opt *opt)
1577 {
1578         struct grep_pat *p;
1579
1580         if (opt->extended)
1581                 return 0; /* punt for too complex stuff */
1582         if (opt->invert)
1583                 return 0;
1584         for (p = opt->pattern_list; p; p = p->next) {
1585                 if (p->token != GREP_PATTERN)
1586                         return 0; /* punt for "header only" and stuff */
1587         }
1588         return 1;
1589 }
1590
1591 static int look_ahead(struct grep_opt *opt,
1592                       unsigned long *left_p,
1593                       unsigned *lno_p,
1594                       char **bol_p)
1595 {
1596         unsigned lno = *lno_p;
1597         char *bol = *bol_p;
1598         struct grep_pat *p;
1599         char *sp, *last_bol;
1600         regoff_t earliest = -1;
1601
1602         for (p = opt->pattern_list; p; p = p->next) {
1603                 int hit;
1604                 regmatch_t m;
1605
1606                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1607                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1608                         continue;
1609                 if (earliest < 0 || m.rm_so < earliest)
1610                         earliest = m.rm_so;
1611         }
1612
1613         if (earliest < 0) {
1614                 *bol_p = bol + *left_p;
1615                 *left_p = 0;
1616                 return 1;
1617         }
1618         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1619                 ; /* find the beginning of the line */
1620         last_bol = sp;
1621
1622         for (sp = bol; sp < last_bol; sp++) {
1623                 if (*sp == '\n')
1624                         lno++;
1625         }
1626         *left_p -= last_bol - bol;
1627         *bol_p = last_bol;
1628         *lno_p = lno;
1629         return 0;
1630 }
1631
1632 static int fill_textconv_grep(struct userdiff_driver *driver,
1633                               struct grep_source *gs)
1634 {
1635         struct diff_filespec *df;
1636         char *buf;
1637         size_t size;
1638
1639         if (!driver || !driver->textconv)
1640                 return grep_source_load(gs);
1641
1642         /*
1643          * The textconv interface is intimately tied to diff_filespecs, so we
1644          * have to pretend to be one. If we could unify the grep_source
1645          * and diff_filespec structs, this mess could just go away.
1646          */
1647         df = alloc_filespec(gs->path);
1648         switch (gs->type) {
1649         case GREP_SOURCE_OID:
1650                 fill_filespec(df, gs->identifier, 1, 0100644);
1651                 break;
1652         case GREP_SOURCE_FILE:
1653                 fill_filespec(df, &null_oid, 0, 0100644);
1654                 break;
1655         default:
1656                 die("BUG: attempt to textconv something without a path?");
1657         }
1658
1659         /*
1660          * fill_textconv is not remotely thread-safe; it may load objects
1661          * behind the scenes, and it modifies the global diff tempfile
1662          * structure.
1663          */
1664         grep_read_lock();
1665         size = fill_textconv(driver, df, &buf);
1666         grep_read_unlock();
1667         free_filespec(df);
1668
1669         /*
1670          * The normal fill_textconv usage by the diff machinery would just keep
1671          * the textconv'd buf separate from the diff_filespec. But much of the
1672          * grep code passes around a grep_source and assumes that its "buf"
1673          * pointer is the beginning of the thing we are searching. So let's
1674          * install our textconv'd version into the grep_source, taking care not
1675          * to leak any existing buffer.
1676          */
1677         grep_source_clear_data(gs);
1678         gs->buf = buf;
1679         gs->size = size;
1680
1681         return 0;
1682 }
1683
1684 static int is_empty_line(const char *bol, const char *eol)
1685 {
1686         while (bol < eol && isspace(*bol))
1687                 bol++;
1688         return bol == eol;
1689 }
1690
1691 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1692 {
1693         char *bol;
1694         char *peek_bol = NULL;
1695         unsigned long left;
1696         unsigned lno = 1;
1697         unsigned last_hit = 0;
1698         int binary_match_only = 0;
1699         unsigned count = 0;
1700         int try_lookahead = 0;
1701         int show_function = 0;
1702         struct userdiff_driver *textconv = NULL;
1703         enum grep_context ctx = GREP_CONTEXT_HEAD;
1704         xdemitconf_t xecfg;
1705
1706         if (!opt->output)
1707                 opt->output = std_output;
1708
1709         if (opt->pre_context || opt->post_context || opt->file_break ||
1710             opt->funcbody) {
1711                 /* Show hunk marks, except for the first file. */
1712                 if (opt->last_shown)
1713                         opt->show_hunk_mark = 1;
1714                 /*
1715                  * If we're using threads then we can't easily identify
1716                  * the first file.  Always put hunk marks in that case
1717                  * and skip the very first one later in work_done().
1718                  */
1719                 if (opt->output != std_output)
1720                         opt->show_hunk_mark = 1;
1721         }
1722         opt->last_shown = 0;
1723
1724         if (opt->allow_textconv) {
1725                 grep_source_load_driver(gs);
1726                 /*
1727                  * We might set up the shared textconv cache data here, which
1728                  * is not thread-safe.
1729                  */
1730                 grep_attr_lock();
1731                 textconv = userdiff_get_textconv(gs->driver);
1732                 grep_attr_unlock();
1733         }
1734
1735         /*
1736          * We know the result of a textconv is text, so we only have to care
1737          * about binary handling if we are not using it.
1738          */
1739         if (!textconv) {
1740                 switch (opt->binary) {
1741                 case GREP_BINARY_DEFAULT:
1742                         if (grep_source_is_binary(gs))
1743                                 binary_match_only = 1;
1744                         break;
1745                 case GREP_BINARY_NOMATCH:
1746                         if (grep_source_is_binary(gs))
1747                                 return 0; /* Assume unmatch */
1748                         break;
1749                 case GREP_BINARY_TEXT:
1750                         break;
1751                 default:
1752                         die("BUG: unknown binary handling mode");
1753                 }
1754         }
1755
1756         memset(&xecfg, 0, sizeof(xecfg));
1757         opt->priv = &xecfg;
1758
1759         try_lookahead = should_lookahead(opt);
1760
1761         if (fill_textconv_grep(textconv, gs) < 0)
1762                 return 0;
1763
1764         bol = gs->buf;
1765         left = gs->size;
1766         while (left) {
1767                 char *eol, ch;
1768                 int hit;
1769
1770                 /*
1771                  * look_ahead() skips quickly to the line that possibly
1772                  * has the next hit; don't call it if we need to do
1773                  * something more than just skipping the current line
1774                  * in response to an unmatch for the current line.  E.g.
1775                  * inside a post-context window, we will show the current
1776                  * line as a context around the previous hit when it
1777                  * doesn't hit.
1778                  */
1779                 if (try_lookahead
1780                     && !(last_hit
1781                          && (show_function ||
1782                              lno <= last_hit + opt->post_context))
1783                     && look_ahead(opt, &left, &lno, &bol))
1784                         break;
1785                 eol = end_of_line(bol, &left);
1786                 ch = *eol;
1787                 *eol = 0;
1788
1789                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1790                         ctx = GREP_CONTEXT_BODY;
1791
1792                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1793                 *eol = ch;
1794
1795                 if (collect_hits)
1796                         goto next_line;
1797
1798                 /* "grep -v -e foo -e bla" should list lines
1799                  * that do not have either, so inversion should
1800                  * be done outside.
1801                  */
1802                 if (opt->invert)
1803                         hit = !hit;
1804                 if (opt->unmatch_name_only) {
1805                         if (hit)
1806                                 return 0;
1807                         goto next_line;
1808                 }
1809                 if (hit) {
1810                         count++;
1811                         if (opt->status_only)
1812                                 return 1;
1813                         if (opt->name_only) {
1814                                 show_name(opt, gs->name);
1815                                 return 1;
1816                         }
1817                         if (opt->count)
1818                                 goto next_line;
1819                         if (binary_match_only) {
1820                                 opt->output(opt, "Binary file ", 12);
1821                                 output_color(opt, gs->name, strlen(gs->name),
1822                                              opt->color_filename);
1823                                 opt->output(opt, " matches\n", 9);
1824                                 return 1;
1825                         }
1826                         /* Hit at this line.  If we haven't shown the
1827                          * pre-context lines, we would need to show them.
1828                          */
1829                         if (opt->pre_context || opt->funcbody)
1830                                 show_pre_context(opt, gs, bol, eol, lno);
1831                         else if (opt->funcname)
1832                                 show_funcname_line(opt, gs, bol, lno);
1833                         show_line(opt, bol, eol, gs->name, lno, ':');
1834                         last_hit = lno;
1835                         if (opt->funcbody)
1836                                 show_function = 1;
1837                         goto next_line;
1838                 }
1839                 if (show_function && (!peek_bol || peek_bol < bol)) {
1840                         unsigned long peek_left = left;
1841                         char *peek_eol = eol;
1842
1843                         /*
1844                          * Trailing empty lines are not interesting.
1845                          * Peek past them to see if they belong to the
1846                          * body of the current function.
1847                          */
1848                         peek_bol = bol;
1849                         while (is_empty_line(peek_bol, peek_eol)) {
1850                                 peek_bol = peek_eol + 1;
1851                                 peek_eol = end_of_line(peek_bol, &peek_left);
1852                         }
1853
1854                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1855                                 show_function = 0;
1856                 }
1857                 if (show_function ||
1858                     (last_hit && lno <= last_hit + opt->post_context)) {
1859                         /* If the last hit is within the post context,
1860                          * we need to show this line.
1861                          */
1862                         show_line(opt, bol, eol, gs->name, lno, '-');
1863                 }
1864
1865         next_line:
1866                 bol = eol + 1;
1867                 if (!left)
1868                         break;
1869                 left--;
1870                 lno++;
1871         }
1872
1873         if (collect_hits)
1874                 return 0;
1875
1876         if (opt->status_only)
1877                 return opt->unmatch_name_only;
1878         if (opt->unmatch_name_only) {
1879                 /* We did not see any hit, so we want to show this */
1880                 show_name(opt, gs->name);
1881                 return 1;
1882         }
1883
1884         xdiff_clear_find_func(&xecfg);
1885         opt->priv = NULL;
1886
1887         /* NEEDSWORK:
1888          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1889          * which feels mostly useless but sometimes useful.  Maybe
1890          * make it another option?  For now suppress them.
1891          */
1892         if (opt->count && count) {
1893                 char buf[32];
1894                 if (opt->pathname) {
1895                         output_color(opt, gs->name, strlen(gs->name),
1896                                      opt->color_filename);
1897                         output_sep(opt, ':');
1898                 }
1899                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1900                 opt->output(opt, buf, strlen(buf));
1901                 return 1;
1902         }
1903         return !!last_hit;
1904 }
1905
1906 static void clr_hit_marker(struct grep_expr *x)
1907 {
1908         /* All-hit markers are meaningful only at the very top level
1909          * OR node.
1910          */
1911         while (1) {
1912                 x->hit = 0;
1913                 if (x->node != GREP_NODE_OR)
1914                         return;
1915                 x->u.binary.left->hit = 0;
1916                 x = x->u.binary.right;
1917         }
1918 }
1919
1920 static int chk_hit_marker(struct grep_expr *x)
1921 {
1922         /* Top level nodes have hit markers.  See if they all are hits */
1923         while (1) {
1924                 if (x->node != GREP_NODE_OR)
1925                         return x->hit;
1926                 if (!x->u.binary.left->hit)
1927                         return 0;
1928                 x = x->u.binary.right;
1929         }
1930 }
1931
1932 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1933 {
1934         /*
1935          * we do not have to do the two-pass grep when we do not check
1936          * buffer-wide "all-match".
1937          */
1938         if (!opt->all_match)
1939                 return grep_source_1(opt, gs, 0);
1940
1941         /* Otherwise the toplevel "or" terms hit a bit differently.
1942          * We first clear hit markers from them.
1943          */
1944         clr_hit_marker(opt->pattern_expression);
1945         grep_source_1(opt, gs, 1);
1946
1947         if (!chk_hit_marker(opt->pattern_expression))
1948                 return 0;
1949
1950         return grep_source_1(opt, gs, 0);
1951 }
1952
1953 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1954 {
1955         struct grep_source gs;
1956         int r;
1957
1958         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1959         gs.buf = buf;
1960         gs.size = size;
1961
1962         r = grep_source(opt, &gs);
1963
1964         grep_source_clear(&gs);
1965         return r;
1966 }
1967
1968 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1969                       const char *name, const char *path,
1970                       const void *identifier)
1971 {
1972         gs->type = type;
1973         gs->name = xstrdup_or_null(name);
1974         gs->path = xstrdup_or_null(path);
1975         gs->buf = NULL;
1976         gs->size = 0;
1977         gs->driver = NULL;
1978
1979         switch (type) {
1980         case GREP_SOURCE_FILE:
1981                 gs->identifier = xstrdup(identifier);
1982                 break;
1983         case GREP_SOURCE_OID:
1984                 gs->identifier = oiddup(identifier);
1985                 break;
1986         case GREP_SOURCE_BUF:
1987                 gs->identifier = NULL;
1988                 break;
1989         }
1990 }
1991
1992 void grep_source_clear(struct grep_source *gs)
1993 {
1994         FREE_AND_NULL(gs->name);
1995         FREE_AND_NULL(gs->path);
1996         FREE_AND_NULL(gs->identifier);
1997         grep_source_clear_data(gs);
1998 }
1999
2000 void grep_source_clear_data(struct grep_source *gs)
2001 {
2002         switch (gs->type) {
2003         case GREP_SOURCE_FILE:
2004         case GREP_SOURCE_OID:
2005                 FREE_AND_NULL(gs->buf);
2006                 gs->size = 0;
2007                 break;
2008         case GREP_SOURCE_BUF:
2009                 /* leave user-provided buf intact */
2010                 break;
2011         }
2012 }
2013
2014 static int grep_source_load_oid(struct grep_source *gs)
2015 {
2016         enum object_type type;
2017
2018         grep_read_lock();
2019         gs->buf = read_object_file(gs->identifier, &type, &gs->size);
2020         grep_read_unlock();
2021
2022         if (!gs->buf)
2023                 return error(_("'%s': unable to read %s"),
2024                              gs->name,
2025                              oid_to_hex(gs->identifier));
2026         return 0;
2027 }
2028
2029 static int grep_source_load_file(struct grep_source *gs)
2030 {
2031         const char *filename = gs->identifier;
2032         struct stat st;
2033         char *data;
2034         size_t size;
2035         int i;
2036
2037         if (lstat(filename, &st) < 0) {
2038         err_ret:
2039                 if (errno != ENOENT)
2040                         error_errno(_("failed to stat '%s'"), filename);
2041                 return -1;
2042         }
2043         if (!S_ISREG(st.st_mode))
2044                 return -1;
2045         size = xsize_t(st.st_size);
2046         i = open(filename, O_RDONLY);
2047         if (i < 0)
2048                 goto err_ret;
2049         data = xmallocz(size);
2050         if (st.st_size != read_in_full(i, data, size)) {
2051                 error_errno(_("'%s': short read"), filename);
2052                 close(i);
2053                 free(data);
2054                 return -1;
2055         }
2056         close(i);
2057
2058         gs->buf = data;
2059         gs->size = size;
2060         return 0;
2061 }
2062
2063 static int grep_source_load(struct grep_source *gs)
2064 {
2065         if (gs->buf)
2066                 return 0;
2067
2068         switch (gs->type) {
2069         case GREP_SOURCE_FILE:
2070                 return grep_source_load_file(gs);
2071         case GREP_SOURCE_OID:
2072                 return grep_source_load_oid(gs);
2073         case GREP_SOURCE_BUF:
2074                 return gs->buf ? 0 : -1;
2075         }
2076         die("BUG: invalid grep_source type to load");
2077 }
2078
2079 void grep_source_load_driver(struct grep_source *gs)
2080 {
2081         if (gs->driver)
2082                 return;
2083
2084         grep_attr_lock();
2085         if (gs->path)
2086                 gs->driver = userdiff_find_by_path(gs->path);
2087         if (!gs->driver)
2088                 gs->driver = userdiff_find_by_name("default");
2089         grep_attr_unlock();
2090 }
2091
2092 static int grep_source_is_binary(struct grep_source *gs)
2093 {
2094         grep_source_load_driver(gs);
2095         if (gs->driver->binary != -1)
2096                 return gs->driver->binary;
2097
2098         if (!grep_source_load(gs))
2099                 return buffer_is_binary(gs->buf, gs->size);
2100
2101         return 0;
2102 }