grep.c

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "grep.h"
   4 #include "object-store.h"
   5 #include "userdiff.h"
   6 #include "xdiff-interface.h"
   7 #include "diff.h"
   8 #include "diffcore.h"
   9 #include "commit.h"
  10 #include "quote.h"
  11 #include "help.h"
  12
  13 static int grep_source_load(struct grep_source *gs);
  14 static int grep_source_is_binary(struct grep_source *gs,
  15                                  struct index_state *istate);
  16
  17 static struct grep_opt grep_defaults;
  18
  19 static const char *color_grep_slots[] = {
  20         [GREP_COLOR_CONTEXT]        = "context",
  21         [GREP_COLOR_FILENAME]       = "filename",
  22         [GREP_COLOR_FUNCTION]       = "function",
  23         [GREP_COLOR_LINENO]         = "lineNumber",
  24         [GREP_COLOR_COLUMNNO]       = "column",
  25         [GREP_COLOR_MATCH_CONTEXT]  = "matchContext",
  26         [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
  27         [GREP_COLOR_SELECTED]       = "selected",
  28         [GREP_COLOR_SEP]            = "separator",
  29 };
  30
  31 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
  32 {
  33         fwrite(buf, size, 1, stdout);
  34 }
  35
  36 static void color_set(char *dst, const char *color_bytes)
  37 {
  38         xsnprintf(dst, COLOR_MAXLEN, "%s", color_bytes);
  39 }
  40
  41 /*
  42  * Initialize the grep_defaults template with hardcoded defaults.
  43  * We could let the compiler do this, but without C99 initializers
  44  * the code gets unwieldy and unreadable, so...
  45  */
  46 void init_grep_defaults(struct repository *repo)
  47 {
  48         struct grep_opt *opt = &grep_defaults;
  49         static int run_once;
  50
  51         if (run_once)
  52                 return;
  53         run_once++;
  54
  55         memset(opt, 0, sizeof(*opt));
  56         opt->repo = repo;
  57         opt->relative = 1;
  58         opt->pathname = 1;
  59         opt->max_depth = -1;
  60         opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  61         color_set(opt->colors[GREP_COLOR_CONTEXT], "");
  62         color_set(opt->colors[GREP_COLOR_FILENAME], "");
  63         color_set(opt->colors[GREP_COLOR_FUNCTION], "");
  64         color_set(opt->colors[GREP_COLOR_LINENO], "");
  65         color_set(opt->colors[GREP_COLOR_COLUMNNO], "");
  66         color_set(opt->colors[GREP_COLOR_MATCH_CONTEXT], GIT_COLOR_BOLD_RED);
  67         color_set(opt->colors[GREP_COLOR_MATCH_SELECTED], GIT_COLOR_BOLD_RED);
  68         color_set(opt->colors[GREP_COLOR_SELECTED], "");
  69         color_set(opt->colors[GREP_COLOR_SEP], GIT_COLOR_CYAN);
  70         opt->only_matching = 0;
  71         opt->color = -1;
  72         opt->output = std_output;
  73 }
  74
  75 static int parse_pattern_type_arg(const char *opt, const char *arg)
  76 {
  77         if (!strcmp(arg, "default"))
  78                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  79         else if (!strcmp(arg, "basic"))
  80                 return GREP_PATTERN_TYPE_BRE;
  81         else if (!strcmp(arg, "extended"))
  82                 return GREP_PATTERN_TYPE_ERE;
  83         else if (!strcmp(arg, "fixed"))
  84                 return GREP_PATTERN_TYPE_FIXED;
  85         else if (!strcmp(arg, "perl"))
  86                 return GREP_PATTERN_TYPE_PCRE;
  87         die("bad %s argument: %s", opt, arg);
  88 }
  89
  90 define_list_config_array_extra(color_grep_slots, {"match"});
  91
  92 /*
  93  * Read the configuration file once and store it in
  94  * the grep_defaults template.
  95  */
  96 int grep_config(const char *var, const char *value, void *cb)
  97 {
  98         struct grep_opt *opt = &grep_defaults;
  99         const char *slot;
 100
 101         if (userdiff_config(var, value) < 0)
 102                 return -1;
 103
 104         if (!strcmp(var, "grep.extendedregexp")) {
 105                 opt->extended_regexp_option = git_config_bool(var, value);
 106                 return 0;
 107         }
 108
 109         if (!strcmp(var, "grep.patterntype")) {
 110                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
 111                 return 0;
 112         }
 113
 114         if (!strcmp(var, "grep.linenumber")) {
 115                 opt->linenum = git_config_bool(var, value);
 116                 return 0;
 117         }
 118         if (!strcmp(var, "grep.column")) {
 119                 opt->columnnum = git_config_bool(var, value);
 120                 return 0;
 121         }
 122
 123         if (!strcmp(var, "grep.fullname")) {
 124                 opt->relative = !git_config_bool(var, value);
 125                 return 0;
 126         }
 127
 128         if (!strcmp(var, "color.grep"))
 129                 opt->color = git_config_colorbool(var, value);
 130         if (!strcmp(var, "color.grep.match")) {
 131                 if (grep_config("color.grep.matchcontext", value, cb) < 0)
 132                         return -1;
 133                 if (grep_config("color.grep.matchselected", value, cb) < 0)
 134                         return -1;
 135         } else if (skip_prefix(var, "color.grep.", &slot)) {
 136                 int i = LOOKUP_CONFIG(color_grep_slots, slot);
 137                 char *color;
 138
 139                 if (i < 0)
 140                         return -1;
 141                 color = opt->colors[i];
 142                 if (!value)
 143                         return config_error_nonbool(var);
 144                 return color_parse(value, color);
 145         }
 146         return 0;
 147 }
 148
 149 /*
 150  * Initialize one instance of grep_opt and copy the
 151  * default values from the template we read the configuration
 152  * information in an earlier call to git_config(grep_config).
 153  */
 154 void grep_init(struct grep_opt *opt, struct repository *repo, const char *prefix)
 155 {
 156         struct grep_opt *def = &grep_defaults;
 157         int i;
 158
 159         memset(opt, 0, sizeof(*opt));
 160         opt->repo = repo;
 161         opt->prefix = prefix;
 162         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 163         opt->pattern_tail = &opt->pattern_list;
 164         opt->header_tail = &opt->header_list;
 165
 166         opt->only_matching = def->only_matching;
 167         opt->color = def->color;
 168         opt->extended_regexp_option = def->extended_regexp_option;
 169         opt->pattern_type_option = def->pattern_type_option;
 170         opt->linenum = def->linenum;
 171         opt->columnnum = def->columnnum;
 172         opt->max_depth = def->max_depth;
 173         opt->pathname = def->pathname;
 174         opt->relative = def->relative;
 175         opt->output = def->output;
 176
 177         for (i = 0; i < NR_GREP_COLORS; i++)
 178                 color_set(opt->colors[i], def->colors[i]);
 179 }
 180
 181 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 182 {
 183         /*
 184          * When committing to the pattern type by setting the relevant
 185          * fields in grep_opt it's generally not necessary to zero out
 186          * the fields we're not choosing, since they won't have been
 187          * set by anything. The extended_regexp_option field is the
 188          * only exception to this.
 189          *
 190          * This is because in the process of parsing grep.patternType
 191          * & grep.extendedRegexp we set opt->pattern_type_option and
 192          * opt->extended_regexp_option, respectively. We then
 193          * internally use opt->extended_regexp_option to see if we're
 194          * compiling an ERE. It must be unset if that's not actually
 195          * the case.
 196          */
 197         if (pattern_type != GREP_PATTERN_TYPE_ERE &&
 198             opt->extended_regexp_option)
 199                 opt->extended_regexp_option = 0;
 200
 201         switch (pattern_type) {
 202         case GREP_PATTERN_TYPE_UNSPECIFIED:
 203                 /* fall through */
 204
 205         case GREP_PATTERN_TYPE_BRE:
 206                 break;
 207
 208         case GREP_PATTERN_TYPE_ERE:
 209                 opt->extended_regexp_option = 1;
 210                 break;
 211
 212         case GREP_PATTERN_TYPE_FIXED:
 213                 opt->fixed = 1;
 214                 break;
 215
 216         case GREP_PATTERN_TYPE_PCRE:
 217 #ifdef USE_LIBPCRE2
 218                 opt->pcre2 = 1;
 219 #else
 220                 /*
 221                  * It's important that pcre1 always be assigned to
 222                  * even when there's no USE_LIBPCRE* defined. We still
 223                  * call the PCRE stub function, it just dies with
 224                  * "cannot use Perl-compatible regexes[...]".
 225                  */
 226                 opt->pcre1 = 1;
 227 #endif
 228                 break;
 229         }
 230 }
 231
 232 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 233 {
 234         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 235                 grep_set_pattern_type_option(pattern_type, opt);
 236         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 237                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 238         else if (opt->extended_regexp_option)
 239                 /*
 240                  * This branch *must* happen after setting from the
 241                  * opt->pattern_type_option above, we don't want
 242                  * grep.extendedRegexp to override grep.patternType!
 243                  */
 244                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 245 }
 246
 247 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 248                                         const char *origin, int no,
 249                                         enum grep_pat_token t,
 250                                         enum grep_header_field field)
 251 {
 252         struct grep_pat *p = xcalloc(1, sizeof(*p));
 253         p->pattern = xmemdupz(pat, patlen);
 254         p->patternlen = patlen;
 255         p->origin = origin;
 256         p->no = no;
 257         p->token = t;
 258         p->field = field;
 259         return p;
 260 }
 261
 262 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 263 {
 264         **tail = p;
 265         *tail = &p->next;
 266         p->next = NULL;
 267
 268         switch (p->token) {
 269         case GREP_PATTERN: /* atom */
 270         case GREP_PATTERN_HEAD:
 271         case GREP_PATTERN_BODY:
 272                 for (;;) {
 273                         struct grep_pat *new_pat;
 274                         size_t len = 0;
 275                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 276                         while (++len <= p->patternlen) {
 277                                 if (*(--cp) == '\n') {
 278                                         nl = cp;
 279                                         break;
 280                                 }
 281                         }
 282                         if (!nl)
 283                                 break;
 284                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 285                                                   p->no, p->token, p->field);
 286                         new_pat->next = p->next;
 287                         if (!p->next)
 288                                 *tail = &new_pat->next;
 289                         p->next = new_pat;
 290                         *nl = '\0';
 291                         p->patternlen -= len;
 292                 }
 293                 break;
 294         default:
 295                 break;
 296         }
 297 }
 298
 299 void append_header_grep_pattern(struct grep_opt *opt,
 300                                 enum grep_header_field field, const char *pat)
 301 {
 302         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 303                                              GREP_PATTERN_HEAD, field);
 304         if (field == GREP_HEADER_REFLOG)
 305                 opt->use_reflog_filter = 1;
 306         do_append_grep_pat(&opt->header_tail, p);
 307 }
 308
 309 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 310                          const char *origin, int no, enum grep_pat_token t)
 311 {
 312         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 313 }
 314
 315 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 316                      const char *origin, int no, enum grep_pat_token t)
 317 {
 318         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 319         do_append_grep_pat(&opt->pattern_tail, p);
 320 }
 321
 322 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 323 {
 324         struct grep_pat *pat;
 325         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 326         *ret = *opt;
 327
 328         ret->pattern_list = NULL;
 329         ret->pattern_tail = &ret->pattern_list;
 330
 331         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 332         {
 333                 if(pat->token == GREP_PATTERN_HEAD)
 334                         append_header_grep_pattern(ret, pat->field,
 335                                                    pat->pattern);
 336                 else
 337                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 338                                         pat->origin, pat->no, pat->token);
 339         }
 340
 341         return ret;
 342 }
 343
 344 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 345                 const char *error)
 346 {
 347         char where[1024];
 348
 349         if (p->no)
 350                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 351         else if (p->origin)
 352                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 353         else
 354                 where[0] = 0;
 355
 356         die("%s'%s': %s", where, p->pattern, error);
 357 }
 358
 359 #ifdef USE_LIBPCRE1
 360 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 361 {
 362         const char *error;
 363         int erroffset;
 364         int options = PCRE_MULTILINE;
 365
 366         if (opt->ignore_case) {
 367                 if (!opt->ignore_locale && has_non_ascii(p->pattern))
 368                         p->pcre1_tables = pcre_maketables();
 369                 options |= PCRE_CASELESS;
 370         }
 371         if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern))
 372                 options |= PCRE_UTF8;
 373
 374         p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 375                                       p->pcre1_tables);
 376         if (!p->pcre1_regexp)
 377                 compile_regexp_failed(p, error);
 378
 379         p->pcre1_extra_info = pcre_study(p->pcre1_regexp, GIT_PCRE_STUDY_JIT_COMPILE, &error);
 380         if (!p->pcre1_extra_info && error)
 381                 die("%s", error);
 382
 383 #ifdef GIT_PCRE1_USE_JIT
 384         pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
 385         if (p->pcre1_jit_on == 1) {
 386                 p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
 387                 if (!p->pcre1_jit_stack)
 388                         die("Couldn't allocate PCRE JIT stack");
 389                 pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
 390         } else if (p->pcre1_jit_on != 0) {
 391                 BUG("The pcre1_jit_on variable should be 0 or 1, not %d",
 392                     p->pcre1_jit_on);
 393         }
 394 #endif
 395 }
 396
 397 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 398                 regmatch_t *match, int eflags)
 399 {
 400         int ovector[30], ret, flags = 0;
 401
 402         if (eflags & REG_NOTBOL)
 403                 flags |= PCRE_NOTBOL;
 404
 405 #ifdef GIT_PCRE1_USE_JIT
 406         if (p->pcre1_jit_on) {
 407                 ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 408                                     eol - line, 0, flags, ovector,
 409                                     ARRAY_SIZE(ovector), p->pcre1_jit_stack);
 410         } else
 411 #endif
 412         {
 413                 ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 414                                 eol - line, 0, flags, ovector,
 415                                 ARRAY_SIZE(ovector));
 416         }
 417
 418         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 419                 die("pcre_exec failed with error code %d", ret);
 420         if (ret > 0) {
 421                 ret = 0;
 422                 match->rm_so = ovector[0];
 423                 match->rm_eo = ovector[1];
 424         }
 425
 426         return ret;
 427 }
 428
 429 static void free_pcre1_regexp(struct grep_pat *p)
 430 {
 431         pcre_free(p->pcre1_regexp);
 432 #ifdef GIT_PCRE1_USE_JIT
 433         if (p->pcre1_jit_on) {
 434                 pcre_free_study(p->pcre1_extra_info);
 435                 pcre_jit_stack_free(p->pcre1_jit_stack);
 436         } else
 437 #endif
 438         {
 439                 pcre_free(p->pcre1_extra_info);
 440         }
 441         pcre_free((void *)p->pcre1_tables);
 442 }
 443 #else /* !USE_LIBPCRE1 */
 444 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 445 {
 446         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 447 }
 448
 449 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 450                 regmatch_t *match, int eflags)
 451 {
 452         return 1;
 453 }
 454
 455 static void free_pcre1_regexp(struct grep_pat *p)
 456 {
 457 }
 458 #endif /* !USE_LIBPCRE1 */
 459
 460 #ifdef USE_LIBPCRE2
 461 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 462 {
 463         int error;
 464         PCRE2_UCHAR errbuf[256];
 465         PCRE2_SIZE erroffset;
 466         int options = PCRE2_MULTILINE;
 467         const uint8_t *character_tables = NULL;
 468         int jitret;
 469         int patinforet;
 470         size_t jitsizearg;
 471
 472         assert(opt->pcre2);
 473
 474         p->pcre2_compile_context = NULL;
 475
 476         if (opt->ignore_case) {
 477                 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
 478                         character_tables = pcre2_maketables(NULL);
 479                         p->pcre2_compile_context = pcre2_compile_context_create(NULL);
 480                         pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
 481                 }
 482                 options |= PCRE2_CASELESS;
 483         }
 484         if (!opt->ignore_locale && is_utf8_locale() && has_non_ascii(p->pattern))
 485                 options |= PCRE2_UTF;
 486
 487         p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
 488                                          p->patternlen, options, &error, &erroffset,
 489                                          p->pcre2_compile_context);
 490
 491         if (p->pcre2_pattern) {
 492                 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
 493                 if (!p->pcre2_match_data)
 494                         die("Couldn't allocate PCRE2 match data");
 495         } else {
 496                 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
 497                 compile_regexp_failed(p, (const char *)&errbuf);
 498         }
 499
 500         pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
 501         if (p->pcre2_jit_on == 1) {
 502                 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
 503                 if (jitret)
 504                         die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
 505
 506                 /*
 507                  * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
 508                  * tells us whether the library itself supports JIT,
 509                  * but to see whether we're going to be actually using
 510                  * JIT we need to extract PCRE2_INFO_JITSIZE from the
 511                  * pattern *after* we do pcre2_jit_compile() above.
 512                  *
 513                  * This is because if the pattern contains the
 514                  * (*NO_JIT) verb (see pcre2syntax(3))
 515                  * pcre2_jit_compile() will exit early with 0. If we
 516                  * then proceed to call pcre2_jit_match() further down
 517                  * the line instead of pcre2_match() we'll either
 518                  * segfault (pre PCRE 10.31) or run into a fatal error
 519                  * (post PCRE2 10.31)
 520                  */
 521                 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
 522                 if (patinforet)
 523                         BUG("pcre2_pattern_info() failed: %d", patinforet);
 524                 if (jitsizearg == 0) {
 525                         p->pcre2_jit_on = 0;
 526                         return;
 527                 }
 528
 529                 p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
 530                 if (!p->pcre2_jit_stack)
 531                         die("Couldn't allocate PCRE2 JIT stack");
 532                 p->pcre2_match_context = pcre2_match_context_create(NULL);
 533                 if (!p->pcre2_match_context)
 534                         die("Couldn't allocate PCRE2 match context");
 535                 pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
 536         } else if (p->pcre2_jit_on != 0) {
 537                 BUG("The pcre2_jit_on variable should be 0 or 1, not %d",
 538                     p->pcre1_jit_on);
 539         }
 540 }
 541
 542 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 543                 regmatch_t *match, int eflags)
 544 {
 545         int ret, flags = 0;
 546         PCRE2_SIZE *ovector;
 547         PCRE2_UCHAR errbuf[256];
 548
 549         if (eflags & REG_NOTBOL)
 550                 flags |= PCRE2_NOTBOL;
 551
 552         if (p->pcre2_jit_on)
 553                 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 554                                       eol - line, 0, flags, p->pcre2_match_data,
 555                                       NULL);
 556         else
 557                 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 558                                   eol - line, 0, flags, p->pcre2_match_data,
 559                                   NULL);
 560
 561         if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 562                 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 563                 die("%s failed with error code %d: %s",
 564                     (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 565                     errbuf);
 566         }
 567         if (ret > 0) {
 568                 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 569                 ret = 0;
 570                 match->rm_so = (int)ovector[0];
 571                 match->rm_eo = (int)ovector[1];
 572         }
 573
 574         return ret;
 575 }
 576
 577 static void free_pcre2_pattern(struct grep_pat *p)
 578 {
 579         pcre2_compile_context_free(p->pcre2_compile_context);
 580         pcre2_code_free(p->pcre2_pattern);
 581         pcre2_match_data_free(p->pcre2_match_data);
 582         pcre2_jit_stack_free(p->pcre2_jit_stack);
 583         pcre2_match_context_free(p->pcre2_match_context);
 584 }
 585 #else /* !USE_LIBPCRE2 */
 586 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 587 {
 588         /*
 589          * Unreachable until USE_LIBPCRE2 becomes synonymous with
 590          * USE_LIBPCRE. See the sibling comment in
 591          * grep_set_pattern_type_option().
 592          */
 593         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 594 }
 595
 596 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 597                 regmatch_t *match, int eflags)
 598 {
 599         return 1;
 600 }
 601
 602 static void free_pcre2_pattern(struct grep_pat *p)
 603 {
 604 }
 605 #endif /* !USE_LIBPCRE2 */
 606
 607 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 608 {
 609         struct strbuf sb = STRBUF_INIT;
 610         int err;
 611         int regflags = 0;
 612
 613         basic_regex_quote_buf(&sb, p->pattern);
 614         if (opt->ignore_case)
 615                 regflags |= REG_ICASE;
 616         err = regcomp(&p->regexp, sb.buf, regflags);
 617         if (opt->debug)
 618                 fprintf(stderr, "fixed %s\n", sb.buf);
 619         strbuf_release(&sb);
 620         if (err) {
 621                 char errbuf[1024];
 622                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 623                 compile_regexp_failed(p, errbuf);
 624         }
 625 }
 626
 627 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 628 {
 629         int err;
 630         int regflags = REG_NEWLINE;
 631
 632         p->word_regexp = opt->word_regexp;
 633         p->ignore_case = opt->ignore_case;
 634         p->fixed = opt->fixed;
 635
 636         if (memchr(p->pattern, 0, p->patternlen) && !opt->pcre2)
 637                 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
 638
 639         if (opt->fixed) {
 640                 compile_fixed_regexp(p, opt);
 641                 return;
 642         }
 643
 644         if (opt->pcre2) {
 645                 compile_pcre2_pattern(p, opt);
 646                 return;
 647         }
 648
 649         if (opt->pcre1) {
 650                 compile_pcre1_regexp(p, opt);
 651                 return;
 652         }
 653
 654         if (p->ignore_case)
 655                 regflags |= REG_ICASE;
 656         if (opt->extended_regexp_option)
 657                 regflags |= REG_EXTENDED;
 658         err = regcomp(&p->regexp, p->pattern, regflags);
 659         if (err) {
 660                 char errbuf[1024];
 661                 regerror(err, &p->regexp, errbuf, 1024);
 662                 compile_regexp_failed(p, errbuf);
 663         }
 664 }
 665
 666 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 667 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 668 {
 669         struct grep_pat *p;
 670         struct grep_expr *x;
 671
 672         p = *list;
 673         if (!p)
 674                 return NULL;
 675         switch (p->token) {
 676         case GREP_PATTERN: /* atom */
 677         case GREP_PATTERN_HEAD:
 678         case GREP_PATTERN_BODY:
 679                 x = xcalloc(1, sizeof (struct grep_expr));
 680                 x->node = GREP_NODE_ATOM;
 681                 x->u.atom = p;
 682                 *list = p->next;
 683                 return x;
 684         case GREP_OPEN_PAREN:
 685                 *list = p->next;
 686                 x = compile_pattern_or(list);
 687                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 688                         die("unmatched parenthesis");
 689                 *list = (*list)->next;
 690                 return x;
 691         default:
 692                 return NULL;
 693         }
 694 }
 695
 696 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 697 {
 698         struct grep_pat *p;
 699         struct grep_expr *x;
 700
 701         p = *list;
 702         if (!p)
 703                 return NULL;
 704         switch (p->token) {
 705         case GREP_NOT:
 706                 if (!p->next)
 707                         die("--not not followed by pattern expression");
 708                 *list = p->next;
 709                 x = xcalloc(1, sizeof (struct grep_expr));
 710                 x->node = GREP_NODE_NOT;
 711                 x->u.unary = compile_pattern_not(list);
 712                 if (!x->u.unary)
 713                         die("--not followed by non pattern expression");
 714                 return x;
 715         default:
 716                 return compile_pattern_atom(list);
 717         }
 718 }
 719
 720 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 721 {
 722         struct grep_pat *p;
 723         struct grep_expr *x, *y, *z;
 724
 725         x = compile_pattern_not(list);
 726         p = *list;
 727         if (p && p->token == GREP_AND) {
 728                 if (!p->next)
 729                         die("--and not followed by pattern expression");
 730                 *list = p->next;
 731                 y = compile_pattern_and(list);
 732                 if (!y)
 733                         die("--and not followed by pattern expression");
 734                 z = xcalloc(1, sizeof (struct grep_expr));
 735                 z->node = GREP_NODE_AND;
 736                 z->u.binary.left = x;
 737                 z->u.binary.right = y;
 738                 return z;
 739         }
 740         return x;
 741 }
 742
 743 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 744 {
 745         struct grep_pat *p;
 746         struct grep_expr *x, *y, *z;
 747
 748         x = compile_pattern_and(list);
 749         p = *list;
 750         if (x && p && p->token != GREP_CLOSE_PAREN) {
 751                 y = compile_pattern_or(list);
 752                 if (!y)
 753                         die("not a pattern expression %s", p->pattern);
 754                 z = xcalloc(1, sizeof (struct grep_expr));
 755                 z->node = GREP_NODE_OR;
 756                 z->u.binary.left = x;
 757                 z->u.binary.right = y;
 758                 return z;
 759         }
 760         return x;
 761 }
 762
 763 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 764 {
 765         return compile_pattern_or(list);
 766 }
 767
 768 static void indent(int in)
 769 {
 770         while (in-- > 0)
 771                 fputc(' ', stderr);
 772 }
 773
 774 static void dump_grep_pat(struct grep_pat *p)
 775 {
 776         switch (p->token) {
 777         case GREP_AND: fprintf(stderr, "*and*"); break;
 778         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 779         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 780         case GREP_NOT: fprintf(stderr, "*not*"); break;
 781         case GREP_OR: fprintf(stderr, "*or*"); break;
 782
 783         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 784         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 785         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 786         }
 787
 788         switch (p->token) {
 789         default: break;
 790         case GREP_PATTERN_HEAD:
 791                 fprintf(stderr, "<head %d>", p->field); break;
 792         case GREP_PATTERN_BODY:
 793                 fprintf(stderr, "<body>"); break;
 794         }
 795         switch (p->token) {
 796         default: break;
 797         case GREP_PATTERN_HEAD:
 798         case GREP_PATTERN_BODY:
 799         case GREP_PATTERN:
 800                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 801                 break;
 802         }
 803         fputc('\n', stderr);
 804 }
 805
 806 static void dump_grep_expression_1(struct grep_expr *x, int in)
 807 {
 808         indent(in);
 809         switch (x->node) {
 810         case GREP_NODE_TRUE:
 811                 fprintf(stderr, "true\n");
 812                 break;
 813         case GREP_NODE_ATOM:
 814                 dump_grep_pat(x->u.atom);
 815                 break;
 816         case GREP_NODE_NOT:
 817                 fprintf(stderr, "(not\n");
 818                 dump_grep_expression_1(x->u.unary, in+1);
 819                 indent(in);
 820                 fprintf(stderr, ")\n");
 821                 break;
 822         case GREP_NODE_AND:
 823                 fprintf(stderr, "(and\n");
 824                 dump_grep_expression_1(x->u.binary.left, in+1);
 825                 dump_grep_expression_1(x->u.binary.right, in+1);
 826                 indent(in);
 827                 fprintf(stderr, ")\n");
 828                 break;
 829         case GREP_NODE_OR:
 830                 fprintf(stderr, "(or\n");
 831                 dump_grep_expression_1(x->u.binary.left, in+1);
 832                 dump_grep_expression_1(x->u.binary.right, in+1);
 833                 indent(in);
 834                 fprintf(stderr, ")\n");
 835                 break;
 836         }
 837 }
 838
 839 static void dump_grep_expression(struct grep_opt *opt)
 840 {
 841         struct grep_expr *x = opt->pattern_expression;
 842
 843         if (opt->all_match)
 844                 fprintf(stderr, "[all-match]\n");
 845         dump_grep_expression_1(x, 0);
 846         fflush(NULL);
 847 }
 848
 849 static struct grep_expr *grep_true_expr(void)
 850 {
 851         struct grep_expr *z = xcalloc(1, sizeof(*z));
 852         z->node = GREP_NODE_TRUE;
 853         return z;
 854 }
 855
 856 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 857 {
 858         struct grep_expr *z = xcalloc(1, sizeof(*z));
 859         z->node = GREP_NODE_OR;
 860         z->u.binary.left = left;
 861         z->u.binary.right = right;
 862         return z;
 863 }
 864
 865 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 866 {
 867         struct grep_pat *p;
 868         struct grep_expr *header_expr;
 869         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 870         enum grep_header_field fld;
 871
 872         if (!opt->header_list)
 873                 return NULL;
 874
 875         for (p = opt->header_list; p; p = p->next) {
 876                 if (p->token != GREP_PATTERN_HEAD)
 877                         BUG("a non-header pattern in grep header list.");
 878                 if (p->field < GREP_HEADER_FIELD_MIN ||
 879                     GREP_HEADER_FIELD_MAX <= p->field)
 880                         BUG("unknown header field %d", p->field);
 881                 compile_regexp(p, opt);
 882         }
 883
 884         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 885                 header_group[fld] = NULL;
 886
 887         for (p = opt->header_list; p; p = p->next) {
 888                 struct grep_expr *h;
 889                 struct grep_pat *pp = p;
 890
 891                 h = compile_pattern_atom(&pp);
 892                 if (!h || pp != p->next)
 893                         BUG("malformed header expr");
 894                 if (!header_group[p->field]) {
 895                         header_group[p->field] = h;
 896                         continue;
 897                 }
 898                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 899         }
 900
 901         header_expr = NULL;
 902
 903         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 904                 if (!header_group[fld])
 905                         continue;
 906                 if (!header_expr)
 907                         header_expr = grep_true_expr();
 908                 header_expr = grep_or_expr(header_group[fld], header_expr);
 909         }
 910         return header_expr;
 911 }
 912
 913 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 914 {
 915         struct grep_expr *z = x;
 916
 917         while (x) {
 918                 assert(x->node == GREP_NODE_OR);
 919                 if (x->u.binary.right &&
 920                     x->u.binary.right->node == GREP_NODE_TRUE) {
 921                         x->u.binary.right = y;
 922                         break;
 923                 }
 924                 x = x->u.binary.right;
 925         }
 926         return z;
 927 }
 928
 929 static void compile_grep_patterns_real(struct grep_opt *opt)
 930 {
 931         struct grep_pat *p;
 932         struct grep_expr *header_expr = prep_header_patterns(opt);
 933
 934         for (p = opt->pattern_list; p; p = p->next) {
 935                 switch (p->token) {
 936                 case GREP_PATTERN: /* atom */
 937                 case GREP_PATTERN_HEAD:
 938                 case GREP_PATTERN_BODY:
 939                         compile_regexp(p, opt);
 940                         break;
 941                 default:
 942                         opt->extended = 1;
 943                         break;
 944                 }
 945         }
 946
 947         if (opt->all_match || header_expr)
 948                 opt->extended = 1;
 949         else if (!opt->extended && !opt->debug)
 950                 return;
 951
 952         p = opt->pattern_list;
 953         if (p)
 954                 opt->pattern_expression = compile_pattern_expr(&p);
 955         if (p)
 956                 die("incomplete pattern expression: %s", p->pattern);
 957
 958         if (!header_expr)
 959                 return;
 960
 961         if (!opt->pattern_expression)
 962                 opt->pattern_expression = header_expr;
 963         else if (opt->all_match)
 964                 opt->pattern_expression = grep_splice_or(header_expr,
 965                                                          opt->pattern_expression);
 966         else
 967                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 968                                                        header_expr);
 969         opt->all_match = 1;
 970 }
 971
 972 void compile_grep_patterns(struct grep_opt *opt)
 973 {
 974         compile_grep_patterns_real(opt);
 975         if (opt->debug)
 976                 dump_grep_expression(opt);
 977 }
 978
 979 static void free_pattern_expr(struct grep_expr *x)
 980 {
 981         switch (x->node) {
 982         case GREP_NODE_TRUE:
 983         case GREP_NODE_ATOM:
 984                 break;
 985         case GREP_NODE_NOT:
 986                 free_pattern_expr(x->u.unary);
 987                 break;
 988         case GREP_NODE_AND:
 989         case GREP_NODE_OR:
 990                 free_pattern_expr(x->u.binary.left);
 991                 free_pattern_expr(x->u.binary.right);
 992                 break;
 993         }
 994         free(x);
 995 }
 996
 997 void free_grep_patterns(struct grep_opt *opt)
 998 {
 999         struct grep_pat *p, *n;
1000
1001         for (p = opt->pattern_list; p; p = n) {
1002                 n = p->next;
1003                 switch (p->token) {
1004                 case GREP_PATTERN: /* atom */
1005                 case GREP_PATTERN_HEAD:
1006                 case GREP_PATTERN_BODY:
1007                         if (p->pcre1_regexp)
1008                                 free_pcre1_regexp(p);
1009                         else if (p->pcre2_pattern)
1010                                 free_pcre2_pattern(p);
1011                         else
1012                                 regfree(&p->regexp);
1013                         free(p->pattern);
1014                         break;
1015                 default:
1016                         break;
1017                 }
1018                 free(p);
1019         }
1020
1021         if (!opt->extended)
1022                 return;
1023         free_pattern_expr(opt->pattern_expression);
1024 }
1025
1026 static char *end_of_line(char *cp, unsigned long *left)
1027 {
1028         unsigned long l = *left;
1029         while (l && *cp != '\n') {
1030                 l--;
1031                 cp++;
1032         }
1033         *left = l;
1034         return cp;
1035 }
1036
1037 static int word_char(char ch)
1038 {
1039         return isalnum(ch) || ch == '_';
1040 }
1041
1042 static void output_color(struct grep_opt *opt, const void *data, size_t size,
1043                          const char *color)
1044 {
1045         if (want_color(opt->color) && color && color[0]) {
1046                 opt->output(opt, color, strlen(color));
1047                 opt->output(opt, data, size);
1048                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1049         } else
1050                 opt->output(opt, data, size);
1051 }
1052
1053 static void output_sep(struct grep_opt *opt, char sign)
1054 {
1055         if (opt->null_following_name)
1056                 opt->output(opt, "\0", 1);
1057         else
1058                 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
1059 }
1060
1061 static void show_name(struct grep_opt *opt, const char *name)
1062 {
1063         output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1064         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1065 }
1066
1067 static int patmatch(struct grep_pat *p, char *line, char *eol,
1068                     regmatch_t *match, int eflags)
1069 {
1070         int hit;
1071
1072         if (p->pcre1_regexp)
1073                 hit = !pcre1match(p, line, eol, match, eflags);
1074         else if (p->pcre2_pattern)
1075                 hit = !pcre2match(p, line, eol, match, eflags);
1076         else
1077                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1078                                    eflags);
1079
1080         return hit;
1081 }
1082
1083 static int strip_timestamp(char *bol, char **eol_p)
1084 {
1085         char *eol = *eol_p;
1086         int ch;
1087
1088         while (bol < --eol) {
1089                 if (*eol != '>')
1090                         continue;
1091                 *eol_p = ++eol;
1092                 ch = *eol;
1093                 *eol = '\0';
1094                 return ch;
1095         }
1096         return 0;
1097 }
1098
1099 static struct {
1100         const char *field;
1101         size_t len;
1102 } header_field[] = {
1103         { "author ", 7 },
1104         { "committer ", 10 },
1105         { "reflog ", 7 },
1106 };
1107
1108 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1109                              enum grep_context ctx,
1110                              regmatch_t *pmatch, int eflags)
1111 {
1112         int hit = 0;
1113         int saved_ch = 0;
1114         const char *start = bol;
1115
1116         if ((p->token != GREP_PATTERN) &&
1117             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1118                 return 0;
1119
1120         if (p->token == GREP_PATTERN_HEAD) {
1121                 const char *field;
1122                 size_t len;
1123                 assert(p->field < ARRAY_SIZE(header_field));
1124                 field = header_field[p->field].field;
1125                 len = header_field[p->field].len;
1126                 if (strncmp(bol, field, len))
1127                         return 0;
1128                 bol += len;
1129                 switch (p->field) {
1130                 case GREP_HEADER_AUTHOR:
1131                 case GREP_HEADER_COMMITTER:
1132                         saved_ch = strip_timestamp(bol, &eol);
1133                         break;
1134                 default:
1135                         break;
1136                 }
1137         }
1138
1139  again:
1140         hit = patmatch(p, bol, eol, pmatch, eflags);
1141
1142         if (hit && p->word_regexp) {
1143                 if ((pmatch[0].rm_so < 0) ||
1144                     (eol - bol) < pmatch[0].rm_so ||
1145                     (pmatch[0].rm_eo < 0) ||
1146                     (eol - bol) < pmatch[0].rm_eo)
1147                         die("regexp returned nonsense");
1148
1149                 /* Match beginning must be either beginning of the
1150                  * line, or at word boundary (i.e. the last char must
1151                  * not be a word char).  Similarly, match end must be
1152                  * either end of the line, or at word boundary
1153                  * (i.e. the next char must not be a word char).
1154                  */
1155                 if ( ((pmatch[0].rm_so == 0) ||
1156                       !word_char(bol[pmatch[0].rm_so-1])) &&
1157                      ((pmatch[0].rm_eo == (eol-bol)) ||
1158                       !word_char(bol[pmatch[0].rm_eo])) )
1159                         ;
1160                 else
1161                         hit = 0;
1162
1163                 /* Words consist of at least one character. */
1164                 if (pmatch->rm_so == pmatch->rm_eo)
1165                         hit = 0;
1166
1167                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1168                         /* There could be more than one match on the
1169                          * line, and the first match might not be
1170                          * strict word match.  But later ones could be!
1171                          * Forward to the next possible start, i.e. the
1172                          * next position following a non-word char.
1173                          */
1174                         bol = pmatch[0].rm_so + bol + 1;
1175                         while (word_char(bol[-1]) && bol < eol)
1176                                 bol++;
1177                         eflags |= REG_NOTBOL;
1178                         if (bol < eol)
1179                                 goto again;
1180                 }
1181         }
1182         if (p->token == GREP_PATTERN_HEAD && saved_ch)
1183                 *eol = saved_ch;
1184         if (hit) {
1185                 pmatch[0].rm_so += bol - start;
1186                 pmatch[0].rm_eo += bol - start;
1187         }
1188         return hit;
1189 }
1190
1191 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x, char *bol,
1192                            char *eol, enum grep_context ctx, ssize_t *col,
1193                            ssize_t *icol, int collect_hits)
1194 {
1195         int h = 0;
1196
1197         if (!x)
1198                 die("Not a valid grep expression");
1199         switch (x->node) {
1200         case GREP_NODE_TRUE:
1201                 h = 1;
1202                 break;
1203         case GREP_NODE_ATOM:
1204                 {
1205                         regmatch_t tmp;
1206                         h = match_one_pattern(x->u.atom, bol, eol, ctx,
1207                                               &tmp, 0);
1208                         if (h && (*col < 0 || tmp.rm_so < *col))
1209                                 *col = tmp.rm_so;
1210                 }
1211                 break;
1212         case GREP_NODE_NOT:
1213                 /*
1214                  * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1215                  */
1216                 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1217                                      0);
1218                 break;
1219         case GREP_NODE_AND:
1220                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1221                                     icol, 0);
1222                 if (h || opt->columnnum) {
1223                         /*
1224                          * Don't short-circuit AND when given --column, since a
1225                          * NOT earlier in the tree may turn this into an OR. In
1226                          * this case, see the below comment.
1227                          */
1228                         h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1229                                              ctx, col, icol, 0);
1230                 }
1231                 break;
1232         case GREP_NODE_OR:
1233                 if (!(collect_hits || opt->columnnum)) {
1234                         /*
1235                          * Don't short-circuit OR when given --column (or
1236                          * collecting hits) to ensure we don't skip a later
1237                          * child that would produce an earlier match.
1238                          */
1239                         return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1240                                                 ctx, col, icol, 0) ||
1241                                 match_expr_eval(opt, x->u.binary.right, bol,
1242                                                 eol, ctx, col, icol, 0));
1243                 }
1244                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1245                                     icol, 0);
1246                 if (collect_hits)
1247                         x->u.binary.left->hit |= h;
1248                 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1249                                      icol, collect_hits);
1250                 break;
1251         default:
1252                 die("Unexpected node type (internal error) %d", x->node);
1253         }
1254         if (collect_hits)
1255                 x->hit |= h;
1256         return h;
1257 }
1258
1259 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1260                       enum grep_context ctx, ssize_t *col,
1261                       ssize_t *icol, int collect_hits)
1262 {
1263         struct grep_expr *x = opt->pattern_expression;
1264         return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1265 }
1266
1267 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1268                       ssize_t *col, ssize_t *icol,
1269                       enum grep_context ctx, int collect_hits)
1270 {
1271         struct grep_pat *p;
1272         int hit = 0;
1273
1274         if (opt->extended)
1275                 return match_expr(opt, bol, eol, ctx, col, icol,
1276                                   collect_hits);
1277
1278         /* we do not call with collect_hits without being extended */
1279         for (p = opt->pattern_list; p; p = p->next) {
1280                 regmatch_t tmp;
1281                 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1282                         hit |= 1;
1283                         if (!opt->columnnum) {
1284                                 /*
1285                                  * Without --column, any single match on a line
1286                                  * is enough to know that it needs to be
1287                                  * printed. With --column, scan _all_ patterns
1288                                  * to find the earliest.
1289                                  */
1290                                 break;
1291                         }
1292                         if (*col < 0 || tmp.rm_so < *col)
1293                                 *col = tmp.rm_so;
1294                 }
1295         }
1296         return hit;
1297 }
1298
1299 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1300                               enum grep_context ctx,
1301                               regmatch_t *pmatch, int eflags)
1302 {
1303         regmatch_t match;
1304
1305         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1306                 return 0;
1307         if (match.rm_so < 0 || match.rm_eo < 0)
1308                 return 0;
1309         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1310                 if (match.rm_so > pmatch->rm_so)
1311                         return 1;
1312                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1313                         return 1;
1314         }
1315         pmatch->rm_so = match.rm_so;
1316         pmatch->rm_eo = match.rm_eo;
1317         return 1;
1318 }
1319
1320 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1321                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1322 {
1323         struct grep_pat *p;
1324         int hit = 0;
1325
1326         pmatch->rm_so = pmatch->rm_eo = -1;
1327         if (bol < eol) {
1328                 for (p = opt->pattern_list; p; p = p->next) {
1329                         switch (p->token) {
1330                         case GREP_PATTERN: /* atom */
1331                         case GREP_PATTERN_HEAD:
1332                         case GREP_PATTERN_BODY:
1333                                 hit |= match_next_pattern(p, bol, eol, ctx,
1334                                                           pmatch, eflags);
1335                                 break;
1336                         default:
1337                                 break;
1338                         }
1339                 }
1340         }
1341         return hit;
1342 }
1343
1344 static void show_line_header(struct grep_opt *opt, const char *name,
1345                              unsigned lno, ssize_t cno, char sign)
1346 {
1347         if (opt->heading && opt->last_shown == 0) {
1348                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1349                 opt->output(opt, "\n", 1);
1350         }
1351         opt->last_shown = lno;
1352
1353         if (!opt->heading && opt->pathname) {
1354                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1355                 output_sep(opt, sign);
1356         }
1357         if (opt->linenum) {
1358                 char buf[32];
1359                 xsnprintf(buf, sizeof(buf), "%d", lno);
1360                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1361                 output_sep(opt, sign);
1362         }
1363         /*
1364          * Treat 'cno' as the 1-indexed offset from the start of a non-context
1365          * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1366          * being called with a context line.
1367          */
1368         if (opt->columnnum && cno) {
1369                 char buf[32];
1370                 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1371                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1372                 output_sep(opt, sign);
1373         }
1374 }
1375
1376 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1377                       const char *name, unsigned lno, ssize_t cno, char sign)
1378 {
1379         int rest = eol - bol;
1380         const char *match_color = NULL;
1381         const char *line_color = NULL;
1382
1383         if (opt->file_break && opt->last_shown == 0) {
1384                 if (opt->show_hunk_mark)
1385                         opt->output(opt, "\n", 1);
1386         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1387                 if (opt->last_shown == 0) {
1388                         if (opt->show_hunk_mark) {
1389                                 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1390                                 opt->output(opt, "\n", 1);
1391                         }
1392                 } else if (lno > opt->last_shown + 1) {
1393                         output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1394                         opt->output(opt, "\n", 1);
1395                 }
1396         }
1397         if (!opt->only_matching) {
1398                 /*
1399                  * In case the line we're being called with contains more than
1400                  * one match, leave printing each header to the loop below.
1401                  */
1402                 show_line_header(opt, name, lno, cno, sign);
1403         }
1404         if (opt->color || opt->only_matching) {
1405                 regmatch_t match;
1406                 enum grep_context ctx = GREP_CONTEXT_BODY;
1407                 int ch = *eol;
1408                 int eflags = 0;
1409
1410                 if (opt->color) {
1411                         if (sign == ':')
1412                                 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1413                         else
1414                                 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1415                         if (sign == ':')
1416                                 line_color = opt->colors[GREP_COLOR_SELECTED];
1417                         else if (sign == '-')
1418                                 line_color = opt->colors[GREP_COLOR_CONTEXT];
1419                         else if (sign == '=')
1420                                 line_color = opt->colors[GREP_COLOR_FUNCTION];
1421                 }
1422                 *eol = '\0';
1423                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1424                         if (match.rm_so == match.rm_eo)
1425                                 break;
1426
1427                         if (opt->only_matching)
1428                                 show_line_header(opt, name, lno, cno, sign);
1429                         else
1430                                 output_color(opt, bol, match.rm_so, line_color);
1431                         output_color(opt, bol + match.rm_so,
1432                                      match.rm_eo - match.rm_so, match_color);
1433                         if (opt->only_matching)
1434                                 opt->output(opt, "\n", 1);
1435                         bol += match.rm_eo;
1436                         cno += match.rm_eo;
1437                         rest -= match.rm_eo;
1438                         eflags = REG_NOTBOL;
1439                 }
1440                 *eol = ch;
1441         }
1442         if (!opt->only_matching) {
1443                 output_color(opt, bol, rest, line_color);
1444                 opt->output(opt, "\n", 1);
1445         }
1446 }
1447
1448 int grep_use_locks;
1449
1450 /*
1451  * This lock protects access to the gitattributes machinery, which is
1452  * not thread-safe.
1453  */
1454 pthread_mutex_t grep_attr_mutex;
1455
1456 static inline void grep_attr_lock(void)
1457 {
1458         if (grep_use_locks)
1459                 pthread_mutex_lock(&grep_attr_mutex);
1460 }
1461
1462 static inline void grep_attr_unlock(void)
1463 {
1464         if (grep_use_locks)
1465                 pthread_mutex_unlock(&grep_attr_mutex);
1466 }
1467
1468 /*
1469  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1470  */
1471 pthread_mutex_t grep_read_mutex;
1472
1473 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1474 {
1475         xdemitconf_t *xecfg = opt->priv;
1476         if (xecfg && !xecfg->find_func) {
1477                 grep_source_load_driver(gs, opt->repo->index);
1478                 if (gs->driver->funcname.pattern) {
1479                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1480                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1481                 } else {
1482                         xecfg = opt->priv = NULL;
1483                 }
1484         }
1485
1486         if (xecfg) {
1487                 char buf[1];
1488                 return xecfg->find_func(bol, eol - bol, buf, 1,
1489                                         xecfg->find_func_priv) >= 0;
1490         }
1491
1492         if (bol == eol)
1493                 return 0;
1494         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1495                 return 1;
1496         return 0;
1497 }
1498
1499 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1500                                char *bol, unsigned lno)
1501 {
1502         while (bol > gs->buf) {
1503                 char *eol = --bol;
1504
1505                 while (bol > gs->buf && bol[-1] != '\n')
1506                         bol--;
1507                 lno--;
1508
1509                 if (lno <= opt->last_shown)
1510                         break;
1511
1512                 if (match_funcname(opt, gs, bol, eol)) {
1513                         show_line(opt, bol, eol, gs->name, lno, 0, '=');
1514                         break;
1515                 }
1516         }
1517 }
1518
1519 static int is_empty_line(const char *bol, const char *eol);
1520
1521 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1522                              char *bol, char *end, unsigned lno)
1523 {
1524         unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1525         int funcname_needed = !!opt->funcname, comment_needed = 0;
1526
1527         if (opt->pre_context < lno)
1528                 from = lno - opt->pre_context;
1529         if (from <= opt->last_shown)
1530                 from = opt->last_shown + 1;
1531         orig_from = from;
1532         if (opt->funcbody) {
1533                 if (match_funcname(opt, gs, bol, end))
1534                         comment_needed = 1;
1535                 else
1536                         funcname_needed = 1;
1537                 from = opt->last_shown + 1;
1538         }
1539
1540         /* Rewind. */
1541         while (bol > gs->buf && cur > from) {
1542                 char *next_bol = bol;
1543                 char *eol = --bol;
1544
1545                 while (bol > gs->buf && bol[-1] != '\n')
1546                         bol--;
1547                 cur--;
1548                 if (comment_needed && (is_empty_line(bol, eol) ||
1549                                        match_funcname(opt, gs, bol, eol))) {
1550                         comment_needed = 0;
1551                         from = orig_from;
1552                         if (cur < from) {
1553                                 cur++;
1554                                 bol = next_bol;
1555                                 break;
1556                         }
1557                 }
1558                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1559                         funcname_lno = cur;
1560                         funcname_needed = 0;
1561                         if (opt->funcbody)
1562                                 comment_needed = 1;
1563                         else
1564                                 from = orig_from;
1565                 }
1566         }
1567
1568         /* We need to look even further back to find a function signature. */
1569         if (opt->funcname && funcname_needed)
1570                 show_funcname_line(opt, gs, bol, cur);
1571
1572         /* Back forward. */
1573         while (cur < lno) {
1574                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1575
1576                 while (*eol != '\n')
1577                         eol++;
1578                 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1579                 bol = eol + 1;
1580                 cur++;
1581         }
1582 }
1583
1584 static int should_lookahead(struct grep_opt *opt)
1585 {
1586         struct grep_pat *p;
1587
1588         if (opt->extended)
1589                 return 0; /* punt for too complex stuff */
1590         if (opt->invert)
1591                 return 0;
1592         for (p = opt->pattern_list; p; p = p->next) {
1593                 if (p->token != GREP_PATTERN)
1594                         return 0; /* punt for "header only" and stuff */
1595         }
1596         return 1;
1597 }
1598
1599 static int look_ahead(struct grep_opt *opt,
1600                       unsigned long *left_p,
1601                       unsigned *lno_p,
1602                       char **bol_p)
1603 {
1604         unsigned lno = *lno_p;
1605         char *bol = *bol_p;
1606         struct grep_pat *p;
1607         char *sp, *last_bol;
1608         regoff_t earliest = -1;
1609
1610         for (p = opt->pattern_list; p; p = p->next) {
1611                 int hit;
1612                 regmatch_t m;
1613
1614                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1615                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1616                         continue;
1617                 if (earliest < 0 || m.rm_so < earliest)
1618                         earliest = m.rm_so;
1619         }
1620
1621         if (earliest < 0) {
1622                 *bol_p = bol + *left_p;
1623                 *left_p = 0;
1624                 return 1;
1625         }
1626         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1627                 ; /* find the beginning of the line */
1628         last_bol = sp;
1629
1630         for (sp = bol; sp < last_bol; sp++) {
1631                 if (*sp == '\n')
1632                         lno++;
1633         }
1634         *left_p -= last_bol - bol;
1635         *bol_p = last_bol;
1636         *lno_p = lno;
1637         return 0;
1638 }
1639
1640 static int fill_textconv_grep(struct repository *r,
1641                               struct userdiff_driver *driver,
1642                               struct grep_source *gs)
1643 {
1644         struct diff_filespec *df;
1645         char *buf;
1646         size_t size;
1647
1648         if (!driver || !driver->textconv)
1649                 return grep_source_load(gs);
1650
1651         /*
1652          * The textconv interface is intimately tied to diff_filespecs, so we
1653          * have to pretend to be one. If we could unify the grep_source
1654          * and diff_filespec structs, this mess could just go away.
1655          */
1656         df = alloc_filespec(gs->path);
1657         switch (gs->type) {
1658         case GREP_SOURCE_OID:
1659                 fill_filespec(df, gs->identifier, 1, 0100644);
1660                 break;
1661         case GREP_SOURCE_FILE:
1662                 fill_filespec(df, &null_oid, 0, 0100644);
1663                 break;
1664         default:
1665                 BUG("attempt to textconv something without a path?");
1666         }
1667
1668         /*
1669          * fill_textconv is not remotely thread-safe; it may load objects
1670          * behind the scenes, and it modifies the global diff tempfile
1671          * structure.
1672          */
1673         grep_read_lock();
1674         size = fill_textconv(r, driver, df, &buf);
1675         grep_read_unlock();
1676         free_filespec(df);
1677
1678         /*
1679          * The normal fill_textconv usage by the diff machinery would just keep
1680          * the textconv'd buf separate from the diff_filespec. But much of the
1681          * grep code passes around a grep_source and assumes that its "buf"
1682          * pointer is the beginning of the thing we are searching. So let's
1683          * install our textconv'd version into the grep_source, taking care not
1684          * to leak any existing buffer.
1685          */
1686         grep_source_clear_data(gs);
1687         gs->buf = buf;
1688         gs->size = size;
1689
1690         return 0;
1691 }
1692
1693 static int is_empty_line(const char *bol, const char *eol)
1694 {
1695         while (bol < eol && isspace(*bol))
1696                 bol++;
1697         return bol == eol;
1698 }
1699
1700 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1701 {
1702         char *bol;
1703         char *peek_bol = NULL;
1704         unsigned long left;
1705         unsigned lno = 1;
1706         unsigned last_hit = 0;
1707         int binary_match_only = 0;
1708         unsigned count = 0;
1709         int try_lookahead = 0;
1710         int show_function = 0;
1711         struct userdiff_driver *textconv = NULL;
1712         enum grep_context ctx = GREP_CONTEXT_HEAD;
1713         xdemitconf_t xecfg;
1714
1715         if (!opt->status_only && gs->name == NULL)
1716                 BUG("grep call which could print a name requires "
1717                     "grep_source.name be non-NULL");
1718
1719         if (!opt->output)
1720                 opt->output = std_output;
1721
1722         if (opt->pre_context || opt->post_context || opt->file_break ||
1723             opt->funcbody) {
1724                 /* Show hunk marks, except for the first file. */
1725                 if (opt->last_shown)
1726                         opt->show_hunk_mark = 1;
1727                 /*
1728                  * If we're using threads then we can't easily identify
1729                  * the first file.  Always put hunk marks in that case
1730                  * and skip the very first one later in work_done().
1731                  */
1732                 if (opt->output != std_output)
1733                         opt->show_hunk_mark = 1;
1734         }
1735         opt->last_shown = 0;
1736
1737         if (opt->allow_textconv) {
1738                 grep_source_load_driver(gs, opt->repo->index);
1739                 /*
1740                  * We might set up the shared textconv cache data here, which
1741                  * is not thread-safe.
1742                  */
1743                 grep_attr_lock();
1744                 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1745                 grep_attr_unlock();
1746         }
1747
1748         /*
1749          * We know the result of a textconv is text, so we only have to care
1750          * about binary handling if we are not using it.
1751          */
1752         if (!textconv) {
1753                 switch (opt->binary) {
1754                 case GREP_BINARY_DEFAULT:
1755                         if (grep_source_is_binary(gs, opt->repo->index))
1756                                 binary_match_only = 1;
1757                         break;
1758                 case GREP_BINARY_NOMATCH:
1759                         if (grep_source_is_binary(gs, opt->repo->index))
1760                                 return 0; /* Assume unmatch */
1761                         break;
1762                 case GREP_BINARY_TEXT:
1763                         break;
1764                 default:
1765                         BUG("unknown binary handling mode");
1766                 }
1767         }
1768
1769         memset(&xecfg, 0, sizeof(xecfg));
1770         opt->priv = &xecfg;
1771
1772         try_lookahead = should_lookahead(opt);
1773
1774         if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1775                 return 0;
1776
1777         bol = gs->buf;
1778         left = gs->size;
1779         while (left) {
1780                 char *eol, ch;
1781                 int hit;
1782                 ssize_t cno;
1783                 ssize_t col = -1, icol = -1;
1784
1785                 /*
1786                  * look_ahead() skips quickly to the line that possibly
1787                  * has the next hit; don't call it if we need to do
1788                  * something more than just skipping the current line
1789                  * in response to an unmatch for the current line.  E.g.
1790                  * inside a post-context window, we will show the current
1791                  * line as a context around the previous hit when it
1792                  * doesn't hit.
1793                  */
1794                 if (try_lookahead
1795                     && !(last_hit
1796                          && (show_function ||
1797                              lno <= last_hit + opt->post_context))
1798                     && look_ahead(opt, &left, &lno, &bol))
1799                         break;
1800                 eol = end_of_line(bol, &left);
1801                 ch = *eol;
1802                 *eol = 0;
1803
1804                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1805                         ctx = GREP_CONTEXT_BODY;
1806
1807                 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1808                 *eol = ch;
1809
1810                 if (collect_hits)
1811                         goto next_line;
1812
1813                 /* "grep -v -e foo -e bla" should list lines
1814                  * that do not have either, so inversion should
1815                  * be done outside.
1816                  */
1817                 if (opt->invert)
1818                         hit = !hit;
1819                 if (opt->unmatch_name_only) {
1820                         if (hit)
1821                                 return 0;
1822                         goto next_line;
1823                 }
1824                 if (hit) {
1825                         count++;
1826                         if (opt->status_only)
1827                                 return 1;
1828                         if (opt->name_only) {
1829                                 show_name(opt, gs->name);
1830                                 return 1;
1831                         }
1832                         if (opt->count)
1833                                 goto next_line;
1834                         if (binary_match_only) {
1835                                 opt->output(opt, "Binary file ", 12);
1836                                 output_color(opt, gs->name, strlen(gs->name),
1837                                              opt->colors[GREP_COLOR_FILENAME]);
1838                                 opt->output(opt, " matches\n", 9);
1839                                 return 1;
1840                         }
1841                         /* Hit at this line.  If we haven't shown the
1842                          * pre-context lines, we would need to show them.
1843                          */
1844                         if (opt->pre_context || opt->funcbody)
1845                                 show_pre_context(opt, gs, bol, eol, lno);
1846                         else if (opt->funcname)
1847                                 show_funcname_line(opt, gs, bol, lno);
1848                         cno = opt->invert ? icol : col;
1849                         if (cno < 0) {
1850                                 /*
1851                                  * A negative cno indicates that there was no
1852                                  * match on the line. We are thus inverted and
1853                                  * being asked to show all lines that _don't_
1854                                  * match a given expression. Therefore, set cno
1855                                  * to 0 to suggest the whole line matches.
1856                                  */
1857                                 cno = 0;
1858                         }
1859                         show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1860                         last_hit = lno;
1861                         if (opt->funcbody)
1862                                 show_function = 1;
1863                         goto next_line;
1864                 }
1865                 if (show_function && (!peek_bol || peek_bol < bol)) {
1866                         unsigned long peek_left = left;
1867                         char *peek_eol = eol;
1868
1869                         /*
1870                          * Trailing empty lines are not interesting.
1871                          * Peek past them to see if they belong to the
1872                          * body of the current function.
1873                          */
1874                         peek_bol = bol;
1875                         while (is_empty_line(peek_bol, peek_eol)) {
1876                                 peek_bol = peek_eol + 1;
1877                                 peek_eol = end_of_line(peek_bol, &peek_left);
1878                         }
1879
1880                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1881                                 show_function = 0;
1882                 }
1883                 if (show_function ||
1884                     (last_hit && lno <= last_hit + opt->post_context)) {
1885                         /* If the last hit is within the post context,
1886                          * we need to show this line.
1887                          */
1888                         show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1889                 }
1890
1891         next_line:
1892                 bol = eol + 1;
1893                 if (!left)
1894                         break;
1895                 left--;
1896                 lno++;
1897         }
1898
1899         if (collect_hits)
1900                 return 0;
1901
1902         if (opt->status_only)
1903                 return opt->unmatch_name_only;
1904         if (opt->unmatch_name_only) {
1905                 /* We did not see any hit, so we want to show this */
1906                 show_name(opt, gs->name);
1907                 return 1;
1908         }
1909
1910         xdiff_clear_find_func(&xecfg);
1911         opt->priv = NULL;
1912
1913         /* NEEDSWORK:
1914          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1915          * which feels mostly useless but sometimes useful.  Maybe
1916          * make it another option?  For now suppress them.
1917          */
1918         if (opt->count && count) {
1919                 char buf[32];
1920                 if (opt->pathname) {
1921                         output_color(opt, gs->name, strlen(gs->name),
1922                                      opt->colors[GREP_COLOR_FILENAME]);
1923                         output_sep(opt, ':');
1924                 }
1925                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1926                 opt->output(opt, buf, strlen(buf));
1927                 return 1;
1928         }
1929         return !!last_hit;
1930 }
1931
1932 static void clr_hit_marker(struct grep_expr *x)
1933 {
1934         /* All-hit markers are meaningful only at the very top level
1935          * OR node.
1936          */
1937         while (1) {
1938                 x->hit = 0;
1939                 if (x->node != GREP_NODE_OR)
1940                         return;
1941                 x->u.binary.left->hit = 0;
1942                 x = x->u.binary.right;
1943         }
1944 }
1945
1946 static int chk_hit_marker(struct grep_expr *x)
1947 {
1948         /* Top level nodes have hit markers.  See if they all are hits */
1949         while (1) {
1950                 if (x->node != GREP_NODE_OR)
1951                         return x->hit;
1952                 if (!x->u.binary.left->hit)
1953                         return 0;
1954                 x = x->u.binary.right;
1955         }
1956 }
1957
1958 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1959 {
1960         /*
1961          * we do not have to do the two-pass grep when we do not check
1962          * buffer-wide "all-match".
1963          */
1964         if (!opt->all_match)
1965                 return grep_source_1(opt, gs, 0);
1966
1967         /* Otherwise the toplevel "or" terms hit a bit differently.
1968          * We first clear hit markers from them.
1969          */
1970         clr_hit_marker(opt->pattern_expression);
1971         grep_source_1(opt, gs, 1);
1972
1973         if (!chk_hit_marker(opt->pattern_expression))
1974                 return 0;
1975
1976         return grep_source_1(opt, gs, 0);
1977 }
1978
1979 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1980 {
1981         struct grep_source gs;
1982         int r;
1983
1984         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1985         gs.buf = buf;
1986         gs.size = size;
1987
1988         r = grep_source(opt, &gs);
1989
1990         grep_source_clear(&gs);
1991         return r;
1992 }
1993
1994 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1995                       const char *name, const char *path,
1996                       const void *identifier)
1997 {
1998         gs->type = type;
1999         gs->name = xstrdup_or_null(name);
2000         gs->path = xstrdup_or_null(path);
2001         gs->buf = NULL;
2002         gs->size = 0;
2003         gs->driver = NULL;
2004
2005         switch (type) {
2006         case GREP_SOURCE_FILE:
2007                 gs->identifier = xstrdup(identifier);
2008                 break;
2009         case GREP_SOURCE_OID:
2010                 gs->identifier = oiddup(identifier);
2011                 break;
2012         case GREP_SOURCE_BUF:
2013                 gs->identifier = NULL;
2014                 break;
2015         }
2016 }
2017
2018 void grep_source_clear(struct grep_source *gs)
2019 {
2020         FREE_AND_NULL(gs->name);
2021         FREE_AND_NULL(gs->path);
2022         FREE_AND_NULL(gs->identifier);
2023         grep_source_clear_data(gs);
2024 }
2025
2026 void grep_source_clear_data(struct grep_source *gs)
2027 {
2028         switch (gs->type) {
2029         case GREP_SOURCE_FILE:
2030         case GREP_SOURCE_OID:
2031                 FREE_AND_NULL(gs->buf);
2032                 gs->size = 0;
2033                 break;
2034         case GREP_SOURCE_BUF:
2035                 /* leave user-provided buf intact */
2036                 break;
2037         }
2038 }
2039
2040 static int grep_source_load_oid(struct grep_source *gs)
2041 {
2042         enum object_type type;
2043
2044         grep_read_lock();
2045         gs->buf = read_object_file(gs->identifier, &type, &gs->size);
2046         grep_read_unlock();
2047
2048         if (!gs->buf)
2049                 return error(_("'%s': unable to read %s"),
2050                              gs->name,
2051                              oid_to_hex(gs->identifier));
2052         return 0;
2053 }
2054
2055 static int grep_source_load_file(struct grep_source *gs)
2056 {
2057         const char *filename = gs->identifier;
2058         struct stat st;
2059         char *data;
2060         size_t size;
2061         int i;
2062
2063         if (lstat(filename, &st) < 0) {
2064         err_ret:
2065                 if (errno != ENOENT)
2066                         error_errno(_("failed to stat '%s'"), filename);
2067                 return -1;
2068         }
2069         if (!S_ISREG(st.st_mode))
2070                 return -1;
2071         size = xsize_t(st.st_size);
2072         i = open(filename, O_RDONLY);
2073         if (i < 0)
2074                 goto err_ret;
2075         data = xmallocz(size);
2076         if (st.st_size != read_in_full(i, data, size)) {
2077                 error_errno(_("'%s': short read"), filename);
2078                 close(i);
2079                 free(data);
2080                 return -1;
2081         }
2082         close(i);
2083
2084         gs->buf = data;
2085         gs->size = size;
2086         return 0;
2087 }
2088
2089 static int grep_source_load(struct grep_source *gs)
2090 {
2091         if (gs->buf)
2092                 return 0;
2093
2094         switch (gs->type) {
2095         case GREP_SOURCE_FILE:
2096                 return grep_source_load_file(gs);
2097         case GREP_SOURCE_OID:
2098                 return grep_source_load_oid(gs);
2099         case GREP_SOURCE_BUF:
2100                 return gs->buf ? 0 : -1;
2101         }
2102         BUG("invalid grep_source type to load");
2103 }
2104
2105 void grep_source_load_driver(struct grep_source *gs,
2106                              struct index_state *istate)
2107 {
2108         if (gs->driver)
2109                 return;
2110
2111         grep_attr_lock();
2112         if (gs->path)
2113                 gs->driver = userdiff_find_by_path(istate, gs->path);
2114         if (!gs->driver)
2115                 gs->driver = userdiff_find_by_name("default");
2116         grep_attr_unlock();
2117 }
2118
2119 static int grep_source_is_binary(struct grep_source *gs,
2120                                  struct index_state *istate)
2121 {
2122         grep_source_load_driver(gs, istate);
2123         if (gs->driver->binary != -1)
2124                 return gs->driver->binary;
2125
2126         if (!grep_source_load(gs))
2127                 return buffer_is_binary(gs->buf, gs->size);
2128
2129         return 0;
2130 }