grep.c

   1 #include "git-compat-util.h"
   2 #include "config.h"
   3 #include "gettext.h"
   4 #include "grep.h"
   5 #include "hex.h"
   6 #include "object-store-ll.h"
   7 #include "pretty.h"
   8 #include "userdiff.h"
   9 #include "xdiff-interface.h"
  10 #include "diff.h"
  11 #include "diffcore.h"
  12 #include "commit.h"
  13 #include "quote.h"
  14 #include "help.h"
  15
  16 static int grep_source_load(struct grep_source *gs);
  17 static int grep_source_is_binary(struct grep_source *gs,
  18                                  struct index_state *istate);
  19
  20 static void std_output(struct grep_opt *opt UNUSED, const void *buf, size_t size)
  21 {
  22         fwrite(buf, size, 1, stdout);
  23 }
  24
  25 static const char *color_grep_slots[] = {
  26         [GREP_COLOR_CONTEXT]        = "context",
  27         [GREP_COLOR_FILENAME]       = "filename",
  28         [GREP_COLOR_FUNCTION]       = "function",
  29         [GREP_COLOR_LINENO]         = "lineNumber",
  30         [GREP_COLOR_COLUMNNO]       = "column",
  31         [GREP_COLOR_MATCH_CONTEXT]  = "matchContext",
  32         [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
  33         [GREP_COLOR_SELECTED]       = "selected",
  34         [GREP_COLOR_SEP]            = "separator",
  35 };
  36
  37 static int parse_pattern_type_arg(const char *opt, const char *arg)
  38 {
  39         if (!strcmp(arg, "default"))
  40                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  41         else if (!strcmp(arg, "basic"))
  42                 return GREP_PATTERN_TYPE_BRE;
  43         else if (!strcmp(arg, "extended"))
  44                 return GREP_PATTERN_TYPE_ERE;
  45         else if (!strcmp(arg, "fixed"))
  46                 return GREP_PATTERN_TYPE_FIXED;
  47         else if (!strcmp(arg, "perl"))
  48                 return GREP_PATTERN_TYPE_PCRE;
  49         die("bad %s argument: %s", opt, arg);
  50 }
  51
  52 define_list_config_array_extra(color_grep_slots, {"match"});
  53
  54 /*
  55  * Read the configuration file once and store it in
  56  * the grep_defaults template.
  57  */
  58 int grep_config(const char *var, const char *value,
  59                 const struct config_context *ctx, void *cb)
  60 {
  61         struct grep_opt *opt = cb;
  62         const char *slot;
  63
  64         if (userdiff_config(var, value) < 0)
  65                 return -1;
  66
  67         if (!strcmp(var, "grep.extendedregexp")) {
  68                 opt->extended_regexp_option = git_config_bool(var, value);
  69                 return 0;
  70         }
  71
  72         if (!strcmp(var, "grep.patterntype")) {
  73                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
  74                 return 0;
  75         }
  76
  77         if (!strcmp(var, "grep.linenumber")) {
  78                 opt->linenum = git_config_bool(var, value);
  79                 return 0;
  80         }
  81         if (!strcmp(var, "grep.column")) {
  82                 opt->columnnum = git_config_bool(var, value);
  83                 return 0;
  84         }
  85
  86         if (!strcmp(var, "grep.fullname")) {
  87                 opt->relative = !git_config_bool(var, value);
  88                 return 0;
  89         }
  90
  91         if (!strcmp(var, "color.grep"))
  92                 opt->color = git_config_colorbool(var, value);
  93         if (!strcmp(var, "color.grep.match")) {
  94                 if (grep_config("color.grep.matchcontext", value, ctx, cb) < 0)
  95                         return -1;
  96                 if (grep_config("color.grep.matchselected", value, ctx, cb) < 0)
  97                         return -1;
  98         } else if (skip_prefix(var, "color.grep.", &slot)) {
  99                 int i = LOOKUP_CONFIG(color_grep_slots, slot);
 100                 char *color;
 101
 102                 if (i < 0)
 103                         return -1;
 104                 color = opt->colors[i];
 105                 if (!value)
 106                         return config_error_nonbool(var);
 107                 return color_parse(value, color);
 108         }
 109         return 0;
 110 }
 111
 112 void grep_init(struct grep_opt *opt, struct repository *repo)
 113 {
 114         struct grep_opt blank = GREP_OPT_INIT;
 115         memcpy(opt, &blank, sizeof(*opt));
 116
 117         opt->repo = repo;
 118         opt->pattern_tail = &opt->pattern_list;
 119         opt->header_tail = &opt->header_list;
 120 }
 121
 122 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 123                                         const char *origin, int no,
 124                                         enum grep_pat_token t,
 125                                         enum grep_header_field field)
 126 {
 127         struct grep_pat *p = xcalloc(1, sizeof(*p));
 128         p->pattern = xmemdupz(pat, patlen);
 129         p->patternlen = patlen;
 130         p->origin = origin;
 131         p->no = no;
 132         p->token = t;
 133         p->field = field;
 134         return p;
 135 }
 136
 137 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 138 {
 139         **tail = p;
 140         *tail = &p->next;
 141         p->next = NULL;
 142
 143         switch (p->token) {
 144         case GREP_PATTERN: /* atom */
 145         case GREP_PATTERN_HEAD:
 146         case GREP_PATTERN_BODY:
 147                 for (;;) {
 148                         struct grep_pat *new_pat;
 149                         size_t len = 0;
 150                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 151                         while (++len <= p->patternlen) {
 152                                 if (*(--cp) == '\n') {
 153                                         nl = cp;
 154                                         break;
 155                                 }
 156                         }
 157                         if (!nl)
 158                                 break;
 159                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 160                                                   p->no, p->token, p->field);
 161                         new_pat->next = p->next;
 162                         if (!p->next)
 163                                 *tail = &new_pat->next;
 164                         p->next = new_pat;
 165                         *nl = '\0';
 166                         p->patternlen -= len;
 167                 }
 168                 break;
 169         default:
 170                 break;
 171         }
 172 }
 173
 174 void append_header_grep_pattern(struct grep_opt *opt,
 175                                 enum grep_header_field field, const char *pat)
 176 {
 177         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 178                                              GREP_PATTERN_HEAD, field);
 179         if (field == GREP_HEADER_REFLOG)
 180                 opt->use_reflog_filter = 1;
 181         do_append_grep_pat(&opt->header_tail, p);
 182 }
 183
 184 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 185                          const char *origin, int no, enum grep_pat_token t)
 186 {
 187         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 188 }
 189
 190 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 191                      const char *origin, int no, enum grep_pat_token t)
 192 {
 193         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 194         do_append_grep_pat(&opt->pattern_tail, p);
 195 }
 196
 197 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 198 {
 199         struct grep_pat *pat;
 200         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 201         *ret = *opt;
 202
 203         ret->pattern_list = NULL;
 204         ret->pattern_tail = &ret->pattern_list;
 205
 206         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 207         {
 208                 if(pat->token == GREP_PATTERN_HEAD)
 209                         append_header_grep_pattern(ret, pat->field,
 210                                                    pat->pattern);
 211                 else
 212                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 213                                         pat->origin, pat->no, pat->token);
 214         }
 215
 216         return ret;
 217 }
 218
 219 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 220                 const char *error)
 221 {
 222         char where[1024];
 223
 224         if (p->no)
 225                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 226         else if (p->origin)
 227                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 228         else
 229                 where[0] = 0;
 230
 231         die("%s'%s': %s", where, p->pattern, error);
 232 }
 233
 234 static int is_fixed(const char *s, size_t len)
 235 {
 236         size_t i;
 237
 238         for (i = 0; i < len; i++) {
 239                 if (is_regex_special(s[i]))
 240                         return 0;
 241         }
 242
 243         return 1;
 244 }
 245
 246 #ifdef USE_LIBPCRE2
 247 #define GREP_PCRE2_DEBUG_MALLOC 0
 248
 249 static void *pcre2_malloc(PCRE2_SIZE size, MAYBE_UNUSED void *memory_data)
 250 {
 251         void *pointer = malloc(size);
 252 #if GREP_PCRE2_DEBUG_MALLOC
 253         static int count = 1;
 254         fprintf(stderr, "PCRE2:%p -> #%02d: alloc(%lu)\n", pointer, count++, size);
 255 #endif
 256         return pointer;
 257 }
 258
 259 static void pcre2_free(void *pointer, MAYBE_UNUSED void *memory_data)
 260 {
 261 #if GREP_PCRE2_DEBUG_MALLOC
 262         static int count = 1;
 263         if (pointer)
 264                 fprintf(stderr, "PCRE2:%p -> #%02d: free()\n", pointer, count++);
 265 #endif
 266         free(pointer);
 267 }
 268
 269 static int pcre2_jit_functional(void)
 270 {
 271         static int jit_working = -1;
 272         pcre2_code *code;
 273         size_t off;
 274         int err;
 275
 276         if (jit_working != -1)
 277                 return jit_working;
 278
 279         /*
 280          * Try to JIT compile a simple pattern to probe if the JIT is
 281          * working in general. It might fail for systems where creating
 282          * memory mappings for runtime code generation is restricted.
 283          */
 284         code = pcre2_compile((PCRE2_SPTR)".", 1, 0, &err, &off, NULL);
 285         if (!code)
 286                 return 0;
 287
 288         jit_working = pcre2_jit_compile(code, PCRE2_JIT_COMPLETE) == 0;
 289         pcre2_code_free(code);
 290
 291         return jit_working;
 292 }
 293
 294 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 295 {
 296         int error;
 297         PCRE2_UCHAR errbuf[256];
 298         PCRE2_SIZE erroffset;
 299         int options = PCRE2_MULTILINE;
 300         int jitret;
 301         int patinforet;
 302         size_t jitsizearg;
 303         int literal = !opt->ignore_case && (p->fixed || p->is_fixed);
 304
 305         /*
 306          * Call pcre2_general_context_create() before calling any
 307          * other pcre2_*(). It sets up our malloc()/free() functions
 308          * with which everything else is allocated.
 309          */
 310         p->pcre2_general_context = pcre2_general_context_create(
 311                 pcre2_malloc, pcre2_free, NULL);
 312         if (!p->pcre2_general_context)
 313                 die("Couldn't allocate PCRE2 general context");
 314
 315         if (opt->ignore_case) {
 316                 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
 317                         p->pcre2_tables = pcre2_maketables(p->pcre2_general_context);
 318                         p->pcre2_compile_context = pcre2_compile_context_create(p->pcre2_general_context);
 319                         pcre2_set_character_tables(p->pcre2_compile_context,
 320                                                         p->pcre2_tables);
 321                 }
 322                 options |= PCRE2_CASELESS;
 323         }
 324         if (!opt->ignore_locale && is_utf8_locale() && !literal)
 325                 options |= (PCRE2_UTF | PCRE2_UCP | PCRE2_MATCH_INVALID_UTF);
 326
 327 #ifndef GIT_PCRE2_VERSION_10_35_OR_HIGHER
 328         /*
 329          * Work around a JIT bug related to invalid Unicode character handling
 330          * fixed in 10.35:
 331          * https://github.com/PCRE2Project/pcre2/commit/c21bd977547d
 332          */
 333         options &= ~PCRE2_UCP;
 334 #endif
 335
 336 #ifndef GIT_PCRE2_VERSION_10_36_OR_HIGHER
 337         /* Work around https://bugs.exim.org/show_bug.cgi?id=2642 fixed in 10.36 */
 338         if (PCRE2_MATCH_INVALID_UTF && options & (PCRE2_UTF | PCRE2_CASELESS))
 339                 options |= PCRE2_NO_START_OPTIMIZE;
 340 #endif
 341
 342         p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
 343                                          p->patternlen, options, &error, &erroffset,
 344                                          p->pcre2_compile_context);
 345
 346         if (p->pcre2_pattern) {
 347                 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, p->pcre2_general_context);
 348                 if (!p->pcre2_match_data)
 349                         die("Couldn't allocate PCRE2 match data");
 350         } else {
 351                 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
 352                 compile_regexp_failed(p, (const char *)&errbuf);
 353         }
 354
 355         pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
 356         if (p->pcre2_jit_on) {
 357                 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
 358                 if (jitret == PCRE2_ERROR_NOMEMORY && !pcre2_jit_functional()) {
 359                         /*
 360                          * Even though pcre2_config(PCRE2_CONFIG_JIT, ...)
 361                          * indicated JIT support, the library might still
 362                          * fail to generate JIT code for various reasons,
 363                          * e.g. when SELinux's 'deny_execmem' or PaX's
 364                          * MPROTECT prevent creating W|X memory mappings.
 365                          *
 366                          * Instead of faling hard, fall back to interpreter
 367                          * mode, just as if the pattern was prefixed with
 368                          * '(*NO_JIT)'.
 369                          */
 370                         p->pcre2_jit_on = 0;
 371                         return;
 372                 } else if (jitret) {
 373                         int need_clip = p->patternlen > 64;
 374                         int clip_len = need_clip ? 64 : p->patternlen;
 375                         die("Couldn't JIT the PCRE2 pattern '%.*s'%s, got '%d'%s",
 376                             clip_len, p->pattern, need_clip ? "..." : "", jitret,
 377                             pcre2_jit_functional()
 378                             ? "\nPerhaps prefix (*NO_JIT) to your pattern?"
 379                             : "");
 380                 }
 381
 382                 /*
 383                  * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
 384                  * tells us whether the library itself supports JIT,
 385                  * but to see whether we're going to be actually using
 386                  * JIT we need to extract PCRE2_INFO_JITSIZE from the
 387                  * pattern *after* we do pcre2_jit_compile() above.
 388                  *
 389                  * This is because if the pattern contains the
 390                  * (*NO_JIT) verb (see pcre2syntax(3))
 391                  * pcre2_jit_compile() will exit early with 0. If we
 392                  * then proceed to call pcre2_jit_match() further down
 393                  * the line instead of pcre2_match() we'll either
 394                  * segfault (pre PCRE 10.31) or run into a fatal error
 395                  * (post PCRE2 10.31)
 396                  */
 397                 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
 398                 if (patinforet)
 399                         BUG("pcre2_pattern_info() failed: %d", patinforet);
 400                 if (jitsizearg == 0) {
 401                         p->pcre2_jit_on = 0;
 402                         return;
 403                 }
 404         }
 405 }
 406
 407 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 408                 regmatch_t *match, int eflags)
 409 {
 410         int ret, flags = 0;
 411         PCRE2_SIZE *ovector;
 412         PCRE2_UCHAR errbuf[256];
 413
 414         if (eflags & REG_NOTBOL)
 415                 flags |= PCRE2_NOTBOL;
 416
 417         if (p->pcre2_jit_on)
 418                 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 419                                       eol - line, 0, flags, p->pcre2_match_data,
 420                                       NULL);
 421         else
 422                 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 423                                   eol - line, 0, flags, p->pcre2_match_data,
 424                                   NULL);
 425
 426         if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 427                 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 428                 die("%s failed with error code %d: %s",
 429                     (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 430                     errbuf);
 431         }
 432         if (ret > 0) {
 433                 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 434                 ret = 0;
 435                 match->rm_so = (int)ovector[0];
 436                 match->rm_eo = (int)ovector[1];
 437         }
 438
 439         return ret;
 440 }
 441
 442 static void free_pcre2_pattern(struct grep_pat *p)
 443 {
 444         pcre2_compile_context_free(p->pcre2_compile_context);
 445         pcre2_code_free(p->pcre2_pattern);
 446         pcre2_match_data_free(p->pcre2_match_data);
 447 #ifdef GIT_PCRE2_VERSION_10_34_OR_HIGHER
 448         pcre2_maketables_free(p->pcre2_general_context, p->pcre2_tables);
 449 #else
 450         free((void *)p->pcre2_tables);
 451 #endif
 452         pcre2_general_context_free(p->pcre2_general_context);
 453 }
 454 #else /* !USE_LIBPCRE2 */
 455 static void compile_pcre2_pattern(struct grep_pat *p UNUSED,
 456                                   const struct grep_opt *opt UNUSED)
 457 {
 458         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 459 }
 460
 461 static int pcre2match(struct grep_pat *p UNUSED, const char *line UNUSED,
 462                       const char *eol UNUSED, regmatch_t *match UNUSED,
 463                       int eflags UNUSED)
 464 {
 465         return 1;
 466 }
 467
 468 static void free_pcre2_pattern(struct grep_pat *p UNUSED)
 469 {
 470 }
 471
 472 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 473 {
 474         struct strbuf sb = STRBUF_INIT;
 475         int err;
 476         int regflags = 0;
 477
 478         basic_regex_quote_buf(&sb, p->pattern);
 479         if (opt->ignore_case)
 480                 regflags |= REG_ICASE;
 481         err = regcomp(&p->regexp, sb.buf, regflags);
 482         strbuf_release(&sb);
 483         if (err) {
 484                 char errbuf[1024];
 485                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 486                 compile_regexp_failed(p, errbuf);
 487         }
 488 }
 489 #endif /* !USE_LIBPCRE2 */
 490
 491 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 492 {
 493         int err;
 494         int regflags = REG_NEWLINE;
 495
 496         if (opt->pattern_type_option == GREP_PATTERN_TYPE_UNSPECIFIED)
 497                 opt->pattern_type_option = (opt->extended_regexp_option
 498                                             ? GREP_PATTERN_TYPE_ERE
 499                                             : GREP_PATTERN_TYPE_BRE);
 500
 501         p->word_regexp = opt->word_regexp;
 502         p->ignore_case = opt->ignore_case;
 503         p->fixed = opt->pattern_type_option == GREP_PATTERN_TYPE_FIXED;
 504
 505         if (opt->pattern_type_option != GREP_PATTERN_TYPE_PCRE &&
 506             memchr(p->pattern, 0, p->patternlen))
 507                 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
 508
 509         p->is_fixed = is_fixed(p->pattern, p->patternlen);
 510 #ifdef USE_LIBPCRE2
 511        if (!p->fixed && !p->is_fixed) {
 512                const char *no_jit = "(*NO_JIT)";
 513                const int no_jit_len = strlen(no_jit);
 514                if (starts_with(p->pattern, no_jit) &&
 515                    is_fixed(p->pattern + no_jit_len,
 516                             p->patternlen - no_jit_len))
 517                        p->is_fixed = 1;
 518        }
 519 #endif
 520         if (p->fixed || p->is_fixed) {
 521 #ifdef USE_LIBPCRE2
 522                 if (p->is_fixed) {
 523                         compile_pcre2_pattern(p, opt);
 524                 } else {
 525                         /*
 526                          * E.g. t7811-grep-open.sh relies on the
 527                          * pattern being restored.
 528                          */
 529                         char *old_pattern = p->pattern;
 530                         size_t old_patternlen = p->patternlen;
 531                         struct strbuf sb = STRBUF_INIT;
 532
 533                         /*
 534                          * There is the PCRE2_LITERAL flag, but it's
 535                          * only in PCRE v2 10.30 and later. Needing to
 536                          * ifdef our way around that and dealing with
 537                          * it + PCRE2_MULTILINE being an error is more
 538                          * complex than just quoting this ourselves.
 539                         */
 540                         strbuf_add(&sb, "\\Q", 2);
 541                         strbuf_add(&sb, p->pattern, p->patternlen);
 542                         strbuf_add(&sb, "\\E", 2);
 543
 544                         p->pattern = sb.buf;
 545                         p->patternlen = sb.len;
 546                         compile_pcre2_pattern(p, opt);
 547                         p->pattern = old_pattern;
 548                         p->patternlen = old_patternlen;
 549                         strbuf_release(&sb);
 550                 }
 551 #else /* !USE_LIBPCRE2 */
 552                 compile_fixed_regexp(p, opt);
 553 #endif /* !USE_LIBPCRE2 */
 554                 return;
 555         }
 556
 557         if (opt->pattern_type_option == GREP_PATTERN_TYPE_PCRE) {
 558                 compile_pcre2_pattern(p, opt);
 559                 return;
 560         }
 561
 562         if (p->ignore_case)
 563                 regflags |= REG_ICASE;
 564         if (opt->pattern_type_option == GREP_PATTERN_TYPE_ERE)
 565                 regflags |= REG_EXTENDED;
 566         err = regcomp(&p->regexp, p->pattern, regflags);
 567         if (err) {
 568                 char errbuf[1024];
 569                 regerror(err, &p->regexp, errbuf, 1024);
 570                 compile_regexp_failed(p, errbuf);
 571         }
 572 }
 573
 574 static struct grep_expr *grep_not_expr(struct grep_expr *expr)
 575 {
 576         struct grep_expr *z = xcalloc(1, sizeof(*z));
 577         z->node = GREP_NODE_NOT;
 578         z->u.unary = expr;
 579         return z;
 580 }
 581
 582 static struct grep_expr *grep_binexp(enum grep_expr_node kind,
 583                                      struct grep_expr *left,
 584                                      struct grep_expr *right)
 585 {
 586         struct grep_expr *z = xcalloc(1, sizeof(*z));
 587         z->node = kind;
 588         z->u.binary.left = left;
 589         z->u.binary.right = right;
 590         return z;
 591 }
 592
 593 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 594 {
 595         return grep_binexp(GREP_NODE_OR, left, right);
 596 }
 597
 598 static struct grep_expr *grep_and_expr(struct grep_expr *left, struct grep_expr *right)
 599 {
 600         return grep_binexp(GREP_NODE_AND, left, right);
 601 }
 602
 603 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 604 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 605 {
 606         struct grep_pat *p;
 607         struct grep_expr *x;
 608
 609         p = *list;
 610         if (!p)
 611                 return NULL;
 612         switch (p->token) {
 613         case GREP_PATTERN: /* atom */
 614         case GREP_PATTERN_HEAD:
 615         case GREP_PATTERN_BODY:
 616                 CALLOC_ARRAY(x, 1);
 617                 x->node = GREP_NODE_ATOM;
 618                 x->u.atom = p;
 619                 *list = p->next;
 620                 return x;
 621         case GREP_OPEN_PAREN:
 622                 *list = p->next;
 623                 x = compile_pattern_or(list);
 624                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 625                         die("unmatched parenthesis");
 626                 *list = (*list)->next;
 627                 return x;
 628         default:
 629                 return NULL;
 630         }
 631 }
 632
 633 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 634 {
 635         struct grep_pat *p;
 636         struct grep_expr *x;
 637
 638         p = *list;
 639         if (!p)
 640                 return NULL;
 641         switch (p->token) {
 642         case GREP_NOT:
 643                 if (!p->next)
 644                         die("--not not followed by pattern expression");
 645                 *list = p->next;
 646                 x = compile_pattern_not(list);
 647                 if (!x)
 648                         die("--not followed by non pattern expression");
 649                 return grep_not_expr(x);
 650         default:
 651                 return compile_pattern_atom(list);
 652         }
 653 }
 654
 655 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 656 {
 657         struct grep_pat *p;
 658         struct grep_expr *x, *y;
 659
 660         x = compile_pattern_not(list);
 661         p = *list;
 662         if (p && p->token == GREP_AND) {
 663                 if (!x)
 664                         die("--and not preceded by pattern expression");
 665                 if (!p->next)
 666                         die("--and not followed by pattern expression");
 667                 *list = p->next;
 668                 y = compile_pattern_and(list);
 669                 if (!y)
 670                         die("--and not followed by pattern expression");
 671                 return grep_and_expr(x, y);
 672         }
 673         return x;
 674 }
 675
 676 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 677 {
 678         struct grep_pat *p;
 679         struct grep_expr *x, *y;
 680
 681         x = compile_pattern_and(list);
 682         p = *list;
 683         if (x && p && p->token != GREP_CLOSE_PAREN) {
 684                 y = compile_pattern_or(list);
 685                 if (!y)
 686                         die("not a pattern expression %s", p->pattern);
 687                 return grep_or_expr(x, y);
 688         }
 689         return x;
 690 }
 691
 692 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 693 {
 694         return compile_pattern_or(list);
 695 }
 696
 697 static struct grep_expr *grep_true_expr(void)
 698 {
 699         struct grep_expr *z = xcalloc(1, sizeof(*z));
 700         z->node = GREP_NODE_TRUE;
 701         return z;
 702 }
 703
 704 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 705 {
 706         struct grep_pat *p;
 707         struct grep_expr *header_expr;
 708         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 709         enum grep_header_field fld;
 710
 711         if (!opt->header_list)
 712                 return NULL;
 713
 714         for (p = opt->header_list; p; p = p->next) {
 715                 if (p->token != GREP_PATTERN_HEAD)
 716                         BUG("a non-header pattern in grep header list.");
 717                 if (p->field < GREP_HEADER_FIELD_MIN ||
 718                     GREP_HEADER_FIELD_MAX <= p->field)
 719                         BUG("unknown header field %d", p->field);
 720                 compile_regexp(p, opt);
 721         }
 722
 723         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 724                 header_group[fld] = NULL;
 725
 726         for (p = opt->header_list; p; p = p->next) {
 727                 struct grep_expr *h;
 728                 struct grep_pat *pp = p;
 729
 730                 h = compile_pattern_atom(&pp);
 731                 if (!h || pp != p->next)
 732                         BUG("malformed header expr");
 733                 if (!header_group[p->field]) {
 734                         header_group[p->field] = h;
 735                         continue;
 736                 }
 737                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 738         }
 739
 740         header_expr = NULL;
 741
 742         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 743                 if (!header_group[fld])
 744                         continue;
 745                 if (!header_expr)
 746                         header_expr = grep_true_expr();
 747                 header_expr = grep_or_expr(header_group[fld], header_expr);
 748         }
 749         return header_expr;
 750 }
 751
 752 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 753 {
 754         struct grep_expr *z = x;
 755
 756         while (x) {
 757                 assert(x->node == GREP_NODE_OR);
 758                 if (x->u.binary.right &&
 759                     x->u.binary.right->node == GREP_NODE_TRUE) {
 760                         x->u.binary.right = y;
 761                         break;
 762                 }
 763                 x = x->u.binary.right;
 764         }
 765         return z;
 766 }
 767
 768 void compile_grep_patterns(struct grep_opt *opt)
 769 {
 770         struct grep_pat *p;
 771         struct grep_expr *header_expr = prep_header_patterns(opt);
 772         int extended = 0;
 773
 774         for (p = opt->pattern_list; p; p = p->next) {
 775                 switch (p->token) {
 776                 case GREP_PATTERN: /* atom */
 777                 case GREP_PATTERN_HEAD:
 778                 case GREP_PATTERN_BODY:
 779                         compile_regexp(p, opt);
 780                         break;
 781                 default:
 782                         extended = 1;
 783                         break;
 784                 }
 785         }
 786
 787         if (opt->all_match || opt->no_body_match || header_expr)
 788                 extended = 1;
 789         else if (!extended)
 790                 return;
 791
 792         p = opt->pattern_list;
 793         if (p)
 794                 opt->pattern_expression = compile_pattern_expr(&p);
 795         if (p)
 796                 die("incomplete pattern expression: %s", p->pattern);
 797
 798         if (opt->no_body_match && opt->pattern_expression)
 799                 opt->pattern_expression = grep_not_expr(opt->pattern_expression);
 800
 801         if (!header_expr)
 802                 return;
 803
 804         if (!opt->pattern_expression)
 805                 opt->pattern_expression = header_expr;
 806         else if (opt->all_match)
 807                 opt->pattern_expression = grep_splice_or(header_expr,
 808                                                          opt->pattern_expression);
 809         else
 810                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 811                                                        header_expr);
 812         opt->all_match = 1;
 813 }
 814
 815 static void free_pattern_expr(struct grep_expr *x)
 816 {
 817         switch (x->node) {
 818         case GREP_NODE_TRUE:
 819         case GREP_NODE_ATOM:
 820                 break;
 821         case GREP_NODE_NOT:
 822                 free_pattern_expr(x->u.unary);
 823                 break;
 824         case GREP_NODE_AND:
 825         case GREP_NODE_OR:
 826                 free_pattern_expr(x->u.binary.left);
 827                 free_pattern_expr(x->u.binary.right);
 828                 break;
 829         }
 830         free(x);
 831 }
 832
 833 static void free_grep_pat(struct grep_pat *pattern)
 834 {
 835         struct grep_pat *p, *n;
 836
 837         for (p = pattern; p; p = n) {
 838                 n = p->next;
 839                 switch (p->token) {
 840                 case GREP_PATTERN: /* atom */
 841                 case GREP_PATTERN_HEAD:
 842                 case GREP_PATTERN_BODY:
 843                         if (p->pcre2_pattern)
 844                                 free_pcre2_pattern(p);
 845                         else
 846                                 regfree(&p->regexp);
 847                         free(p->pattern);
 848                         break;
 849                 default:
 850                         break;
 851                 }
 852                 free(p);
 853         }
 854 }
 855
 856 void free_grep_patterns(struct grep_opt *opt)
 857 {
 858         free_grep_pat(opt->pattern_list);
 859         free_grep_pat(opt->header_list);
 860
 861         if (opt->pattern_expression)
 862                 free_pattern_expr(opt->pattern_expression);
 863 }
 864
 865 static const char *end_of_line(const char *cp, unsigned long *left)
 866 {
 867         unsigned long l = *left;
 868         while (l && *cp != '\n') {
 869                 l--;
 870                 cp++;
 871         }
 872         *left = l;
 873         return cp;
 874 }
 875
 876 static int word_char(char ch)
 877 {
 878         return isalnum(ch) || ch == '_';
 879 }
 880
 881 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 882                          const char *color)
 883 {
 884         if (want_color(opt->color) && color && color[0]) {
 885                 opt->output(opt, color, strlen(color));
 886                 opt->output(opt, data, size);
 887                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 888         } else
 889                 opt->output(opt, data, size);
 890 }
 891
 892 static void output_sep(struct grep_opt *opt, char sign)
 893 {
 894         if (opt->null_following_name)
 895                 opt->output(opt, "\0", 1);
 896         else
 897                 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
 898 }
 899
 900 static void show_name(struct grep_opt *opt, const char *name)
 901 {
 902         output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
 903         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 904 }
 905
 906 static int patmatch(struct grep_pat *p,
 907                     const char *line, const char *eol,
 908                     regmatch_t *match, int eflags)
 909 {
 910         int hit;
 911
 912         if (p->pcre2_pattern)
 913                 hit = !pcre2match(p, line, eol, match, eflags);
 914         else
 915                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
 916                                    eflags);
 917
 918         return hit;
 919 }
 920
 921 static void strip_timestamp(const char *bol, const char **eol_p)
 922 {
 923         const char *eol = *eol_p;
 924
 925         while (bol < --eol) {
 926                 if (*eol != '>')
 927                         continue;
 928                 *eol_p = ++eol;
 929                 break;
 930         }
 931 }
 932
 933 static struct {
 934         const char *field;
 935         size_t len;
 936 } header_field[] = {
 937         { "author ", 7 },
 938         { "committer ", 10 },
 939         { "reflog ", 7 },
 940 };
 941
 942 static int headerless_match_one_pattern(struct grep_pat *p,
 943                                         const char *bol, const char *eol,
 944                                         enum grep_context ctx,
 945                                         regmatch_t *pmatch, int eflags)
 946 {
 947         int hit = 0;
 948         const char *start = bol;
 949
 950         if ((p->token != GREP_PATTERN) &&
 951             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 952                 return 0;
 953
 954  again:
 955         hit = patmatch(p, bol, eol, pmatch, eflags);
 956
 957         if (hit && p->word_regexp) {
 958                 if ((pmatch[0].rm_so < 0) ||
 959                     (eol - bol) < pmatch[0].rm_so ||
 960                     (pmatch[0].rm_eo < 0) ||
 961                     (eol - bol) < pmatch[0].rm_eo)
 962                         die("regexp returned nonsense");
 963
 964                 /* Match beginning must be either beginning of the
 965                  * line, or at word boundary (i.e. the last char must
 966                  * not be a word char).  Similarly, match end must be
 967                  * either end of the line, or at word boundary
 968                  * (i.e. the next char must not be a word char).
 969                  */
 970                 if ( ((pmatch[0].rm_so == 0) ||
 971                       !word_char(bol[pmatch[0].rm_so-1])) &&
 972                      ((pmatch[0].rm_eo == (eol-bol)) ||
 973                       !word_char(bol[pmatch[0].rm_eo])) )
 974                         ;
 975                 else
 976                         hit = 0;
 977
 978                 /* Words consist of at least one character. */
 979                 if (pmatch->rm_so == pmatch->rm_eo)
 980                         hit = 0;
 981
 982                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 983                         /* There could be more than one match on the
 984                          * line, and the first match might not be
 985                          * strict word match.  But later ones could be!
 986                          * Forward to the next possible start, i.e. the
 987                          * next position following a non-word char.
 988                          */
 989                         bol = pmatch[0].rm_so + bol + 1;
 990                         while (word_char(bol[-1]) && bol < eol)
 991                                 bol++;
 992                         eflags |= REG_NOTBOL;
 993                         if (bol < eol)
 994                                 goto again;
 995                 }
 996         }
 997         if (hit) {
 998                 pmatch[0].rm_so += bol - start;
 999                 pmatch[0].rm_eo += bol - start;
1000         }
1001         return hit;
1002 }
1003
1004 static int match_one_pattern(struct grep_pat *p,
1005                              const char *bol, const char *eol,
1006                              enum grep_context ctx, regmatch_t *pmatch,
1007                              int eflags)
1008 {
1009         const char *field;
1010         size_t len;
1011
1012         if (p->token == GREP_PATTERN_HEAD) {
1013                 assert(p->field < ARRAY_SIZE(header_field));
1014                 field = header_field[p->field].field;
1015                 len = header_field[p->field].len;
1016                 if (strncmp(bol, field, len))
1017                         return 0;
1018                 bol += len;
1019
1020                 switch (p->field) {
1021                 case GREP_HEADER_AUTHOR:
1022                 case GREP_HEADER_COMMITTER:
1023                         strip_timestamp(bol, &eol);
1024                         break;
1025                 default:
1026                         break;
1027                 }
1028         }
1029
1030         return headerless_match_one_pattern(p, bol, eol, ctx, pmatch, eflags);
1031 }
1032
1033
1034 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x,
1035                            const char *bol, const char *eol,
1036                            enum grep_context ctx, ssize_t *col,
1037                            ssize_t *icol, int collect_hits)
1038 {
1039         int h = 0;
1040
1041         switch (x->node) {
1042         case GREP_NODE_TRUE:
1043                 h = 1;
1044                 break;
1045         case GREP_NODE_ATOM:
1046                 {
1047                         regmatch_t tmp;
1048                         h = match_one_pattern(x->u.atom, bol, eol, ctx,
1049                                               &tmp, 0);
1050                         if (h && (*col < 0 || tmp.rm_so < *col))
1051                                 *col = tmp.rm_so;
1052                 }
1053                 if (x->u.atom->token == GREP_PATTERN_BODY)
1054                         opt->body_hit |= h;
1055                 break;
1056         case GREP_NODE_NOT:
1057                 /*
1058                  * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1059                  */
1060                 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1061                                      0);
1062                 break;
1063         case GREP_NODE_AND:
1064                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1065                                     icol, 0);
1066                 if (h || opt->columnnum) {
1067                         /*
1068                          * Don't short-circuit AND when given --column, since a
1069                          * NOT earlier in the tree may turn this into an OR. In
1070                          * this case, see the below comment.
1071                          */
1072                         h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1073                                              ctx, col, icol, 0);
1074                 }
1075                 break;
1076         case GREP_NODE_OR:
1077                 if (!(collect_hits || opt->columnnum)) {
1078                         /*
1079                          * Don't short-circuit OR when given --column (or
1080                          * collecting hits) to ensure we don't skip a later
1081                          * child that would produce an earlier match.
1082                          */
1083                         return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1084                                                 ctx, col, icol, 0) ||
1085                                 match_expr_eval(opt, x->u.binary.right, bol,
1086                                                 eol, ctx, col, icol, 0));
1087                 }
1088                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1089                                     icol, 0);
1090                 if (collect_hits)
1091                         x->u.binary.left->hit |= h;
1092                 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1093                                      icol, collect_hits);
1094                 break;
1095         default:
1096                 die("Unexpected node type (internal error) %d", x->node);
1097         }
1098         if (collect_hits)
1099                 x->hit |= h;
1100         return h;
1101 }
1102
1103 static int match_expr(struct grep_opt *opt,
1104                       const char *bol, const char *eol,
1105                       enum grep_context ctx, ssize_t *col,
1106                       ssize_t *icol, int collect_hits)
1107 {
1108         struct grep_expr *x = opt->pattern_expression;
1109         return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1110 }
1111
1112 static int match_line(struct grep_opt *opt,
1113                       const char *bol, const char *eol,
1114                       ssize_t *col, ssize_t *icol,
1115                       enum grep_context ctx, int collect_hits)
1116 {
1117         struct grep_pat *p;
1118         int hit = 0;
1119
1120         if (opt->pattern_expression)
1121                 return match_expr(opt, bol, eol, ctx, col, icol,
1122                                   collect_hits);
1123
1124         /* we do not call with collect_hits without being extended */
1125         for (p = opt->pattern_list; p; p = p->next) {
1126                 regmatch_t tmp;
1127                 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1128                         hit |= 1;
1129                         if (!opt->columnnum) {
1130                                 /*
1131                                  * Without --column, any single match on a line
1132                                  * is enough to know that it needs to be
1133                                  * printed. With --column, scan _all_ patterns
1134                                  * to find the earliest.
1135                                  */
1136                                 break;
1137                         }
1138                         if (*col < 0 || tmp.rm_so < *col)
1139                                 *col = tmp.rm_so;
1140                 }
1141         }
1142         return hit;
1143 }
1144
1145 static int match_next_pattern(struct grep_pat *p,
1146                               const char *bol, const char *eol,
1147                               enum grep_context ctx,
1148                               regmatch_t *pmatch, int eflags)
1149 {
1150         regmatch_t match;
1151
1152         if (!headerless_match_one_pattern(p, bol, eol, ctx, &match, eflags))
1153                 return 0;
1154         if (match.rm_so < 0 || match.rm_eo < 0)
1155                 return 0;
1156         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1157                 if (match.rm_so > pmatch->rm_so)
1158                         return 1;
1159                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1160                         return 1;
1161         }
1162         pmatch->rm_so = match.rm_so;
1163         pmatch->rm_eo = match.rm_eo;
1164         return 1;
1165 }
1166
1167 int grep_next_match(struct grep_opt *opt,
1168                     const char *bol, const char *eol,
1169                     enum grep_context ctx, regmatch_t *pmatch,
1170                     enum grep_header_field field, int eflags)
1171 {
1172         struct grep_pat *p;
1173         int hit = 0;
1174
1175         pmatch->rm_so = pmatch->rm_eo = -1;
1176         if (bol < eol) {
1177                 for (p = ((ctx == GREP_CONTEXT_HEAD)
1178                            ? opt->header_list : opt->pattern_list);
1179                           p; p = p->next) {
1180                         switch (p->token) {
1181                         case GREP_PATTERN_HEAD:
1182                                 if ((field != GREP_HEADER_FIELD_MAX) &&
1183                                     (p->field != field))
1184                                         continue;
1185                                 /* fall thru */
1186                         case GREP_PATTERN: /* atom */
1187                         case GREP_PATTERN_BODY:
1188                                 hit |= match_next_pattern(p, bol, eol, ctx,
1189                                                           pmatch, eflags);
1190                                 break;
1191                         default:
1192                                 break;
1193                         }
1194                 }
1195         }
1196         return hit;
1197 }
1198
1199 static void show_line_header(struct grep_opt *opt, const char *name,
1200                              unsigned lno, ssize_t cno, char sign)
1201 {
1202         if (opt->heading && opt->last_shown == 0) {
1203                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1204                 opt->output(opt, "\n", 1);
1205         }
1206         opt->last_shown = lno;
1207
1208         if (!opt->heading && opt->pathname) {
1209                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1210                 output_sep(opt, sign);
1211         }
1212         if (opt->linenum) {
1213                 char buf[32];
1214                 xsnprintf(buf, sizeof(buf), "%d", lno);
1215                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1216                 output_sep(opt, sign);
1217         }
1218         /*
1219          * Treat 'cno' as the 1-indexed offset from the start of a non-context
1220          * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1221          * being called with a context line.
1222          */
1223         if (opt->columnnum && cno) {
1224                 char buf[32];
1225                 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1226                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1227                 output_sep(opt, sign);
1228         }
1229 }
1230
1231 static void show_line(struct grep_opt *opt,
1232                       const char *bol, const char *eol,
1233                       const char *name, unsigned lno, ssize_t cno, char sign)
1234 {
1235         int rest = eol - bol;
1236         const char *match_color = NULL;
1237         const char *line_color = NULL;
1238
1239         if (opt->file_break && opt->last_shown == 0) {
1240                 if (opt->show_hunk_mark)
1241                         opt->output(opt, "\n", 1);
1242         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1243                 if (opt->last_shown == 0) {
1244                         if (opt->show_hunk_mark) {
1245                                 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1246                                 opt->output(opt, "\n", 1);
1247                         }
1248                 } else if (lno > opt->last_shown + 1) {
1249                         output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1250                         opt->output(opt, "\n", 1);
1251                 }
1252         }
1253         if (!opt->only_matching) {
1254                 /*
1255                  * In case the line we're being called with contains more than
1256                  * one match, leave printing each header to the loop below.
1257                  */
1258                 show_line_header(opt, name, lno, cno, sign);
1259         }
1260         if (opt->color || opt->only_matching) {
1261                 regmatch_t match;
1262                 enum grep_context ctx = GREP_CONTEXT_BODY;
1263                 int eflags = 0;
1264
1265                 if (opt->color) {
1266                         if (sign == ':')
1267                                 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1268                         else
1269                                 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1270                         if (sign == ':')
1271                                 line_color = opt->colors[GREP_COLOR_SELECTED];
1272                         else if (sign == '-')
1273                                 line_color = opt->colors[GREP_COLOR_CONTEXT];
1274                         else if (sign == '=')
1275                                 line_color = opt->colors[GREP_COLOR_FUNCTION];
1276                 }
1277                 while (grep_next_match(opt, bol, eol, ctx, &match,
1278                                        GREP_HEADER_FIELD_MAX, eflags)) {
1279                         if (match.rm_so == match.rm_eo)
1280                                 break;
1281
1282                         if (opt->only_matching)
1283                                 show_line_header(opt, name, lno, cno, sign);
1284                         else
1285                                 output_color(opt, bol, match.rm_so, line_color);
1286                         output_color(opt, bol + match.rm_so,
1287                                      match.rm_eo - match.rm_so, match_color);
1288                         if (opt->only_matching)
1289                                 opt->output(opt, "\n", 1);
1290                         bol += match.rm_eo;
1291                         cno += match.rm_eo;
1292                         rest -= match.rm_eo;
1293                         eflags = REG_NOTBOL;
1294                 }
1295         }
1296         if (!opt->only_matching) {
1297                 output_color(opt, bol, rest, line_color);
1298                 opt->output(opt, "\n", 1);
1299         }
1300 }
1301
1302 int grep_use_locks;
1303
1304 /*
1305  * This lock protects access to the gitattributes machinery, which is
1306  * not thread-safe.
1307  */
1308 pthread_mutex_t grep_attr_mutex;
1309
1310 static inline void grep_attr_lock(void)
1311 {
1312         if (grep_use_locks)
1313                 pthread_mutex_lock(&grep_attr_mutex);
1314 }
1315
1316 static inline void grep_attr_unlock(void)
1317 {
1318         if (grep_use_locks)
1319                 pthread_mutex_unlock(&grep_attr_mutex);
1320 }
1321
1322 static int match_funcname(struct grep_opt *opt, struct grep_source *gs,
1323                           const char *bol, const char *eol)
1324 {
1325         xdemitconf_t *xecfg = opt->priv;
1326         if (xecfg && !xecfg->find_func) {
1327                 grep_source_load_driver(gs, opt->repo->index);
1328                 if (gs->driver->funcname.pattern) {
1329                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1330                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1331                 } else {
1332                         xecfg = opt->priv = NULL;
1333                 }
1334         }
1335
1336         if (xecfg) {
1337                 char buf[1];
1338                 return xecfg->find_func(bol, eol - bol, buf, 1,
1339                                         xecfg->find_func_priv) >= 0;
1340         }
1341
1342         if (bol == eol)
1343                 return 0;
1344         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1345                 return 1;
1346         return 0;
1347 }
1348
1349 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1350                                const char *bol, unsigned lno)
1351 {
1352         while (bol > gs->buf) {
1353                 const char *eol = --bol;
1354
1355                 while (bol > gs->buf && bol[-1] != '\n')
1356                         bol--;
1357                 lno--;
1358
1359                 if (lno <= opt->last_shown)
1360                         break;
1361
1362                 if (match_funcname(opt, gs, bol, eol)) {
1363                         show_line(opt, bol, eol, gs->name, lno, 0, '=');
1364                         break;
1365                 }
1366         }
1367 }
1368
1369 static int is_empty_line(const char *bol, const char *eol);
1370
1371 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1372                              const char *bol, const char *end, unsigned lno)
1373 {
1374         unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1375         int funcname_needed = !!opt->funcname, comment_needed = 0;
1376
1377         if (opt->pre_context < lno)
1378                 from = lno - opt->pre_context;
1379         if (from <= opt->last_shown)
1380                 from = opt->last_shown + 1;
1381         orig_from = from;
1382         if (opt->funcbody) {
1383                 if (match_funcname(opt, gs, bol, end))
1384                         comment_needed = 1;
1385                 else
1386                         funcname_needed = 1;
1387                 from = opt->last_shown + 1;
1388         }
1389
1390         /* Rewind. */
1391         while (bol > gs->buf && cur > from) {
1392                 const char *next_bol = bol;
1393                 const char *eol = --bol;
1394
1395                 while (bol > gs->buf && bol[-1] != '\n')
1396                         bol--;
1397                 cur--;
1398                 if (comment_needed && (is_empty_line(bol, eol) ||
1399                                        match_funcname(opt, gs, bol, eol))) {
1400                         comment_needed = 0;
1401                         from = orig_from;
1402                         if (cur < from) {
1403                                 cur++;
1404                                 bol = next_bol;
1405                                 break;
1406                         }
1407                 }
1408                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1409                         funcname_lno = cur;
1410                         funcname_needed = 0;
1411                         if (opt->funcbody)
1412                                 comment_needed = 1;
1413                         else
1414                                 from = orig_from;
1415                 }
1416         }
1417
1418         /* We need to look even further back to find a function signature. */
1419         if (opt->funcname && funcname_needed)
1420                 show_funcname_line(opt, gs, bol, cur);
1421
1422         /* Back forward. */
1423         while (cur < lno) {
1424                 const char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1425
1426                 while (*eol != '\n')
1427                         eol++;
1428                 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1429                 bol = eol + 1;
1430                 cur++;
1431         }
1432 }
1433
1434 static int should_lookahead(struct grep_opt *opt)
1435 {
1436         struct grep_pat *p;
1437
1438         if (opt->pattern_expression)
1439                 return 0; /* punt for too complex stuff */
1440         if (opt->invert)
1441                 return 0;
1442         for (p = opt->pattern_list; p; p = p->next) {
1443                 if (p->token != GREP_PATTERN)
1444                         return 0; /* punt for "header only" and stuff */
1445         }
1446         return 1;
1447 }
1448
1449 static int look_ahead(struct grep_opt *opt,
1450                       unsigned long *left_p,
1451                       unsigned *lno_p,
1452                       const char **bol_p)
1453 {
1454         unsigned lno = *lno_p;
1455         const char *bol = *bol_p;
1456         struct grep_pat *p;
1457         const char *sp, *last_bol;
1458         regoff_t earliest = -1;
1459
1460         for (p = opt->pattern_list; p; p = p->next) {
1461                 int hit;
1462                 regmatch_t m;
1463
1464                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1465                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1466                         continue;
1467                 if (earliest < 0 || m.rm_so < earliest)
1468                         earliest = m.rm_so;
1469         }
1470
1471         if (earliest < 0) {
1472                 *bol_p = bol + *left_p;
1473                 *left_p = 0;
1474                 return 1;
1475         }
1476         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1477                 ; /* find the beginning of the line */
1478         last_bol = sp;
1479
1480         for (sp = bol; sp < last_bol; sp++) {
1481                 if (*sp == '\n')
1482                         lno++;
1483         }
1484         *left_p -= last_bol - bol;
1485         *bol_p = last_bol;
1486         *lno_p = lno;
1487         return 0;
1488 }
1489
1490 static int fill_textconv_grep(struct repository *r,
1491                               struct userdiff_driver *driver,
1492                               struct grep_source *gs)
1493 {
1494         struct diff_filespec *df;
1495         char *buf;
1496         size_t size;
1497
1498         if (!driver || !driver->textconv)
1499                 return grep_source_load(gs);
1500
1501         /*
1502          * The textconv interface is intimately tied to diff_filespecs, so we
1503          * have to pretend to be one. If we could unify the grep_source
1504          * and diff_filespec structs, this mess could just go away.
1505          */
1506         df = alloc_filespec(gs->path);
1507         switch (gs->type) {
1508         case GREP_SOURCE_OID:
1509                 fill_filespec(df, gs->identifier, 1, 0100644);
1510                 break;
1511         case GREP_SOURCE_FILE:
1512                 fill_filespec(df, null_oid(), 0, 0100644);
1513                 break;
1514         default:
1515                 BUG("attempt to textconv something without a path?");
1516         }
1517
1518         /*
1519          * fill_textconv is not remotely thread-safe; it modifies the global
1520          * diff tempfile structure, writes to the_repo's odb and might
1521          * internally call thread-unsafe functions such as the
1522          * prepare_packed_git() lazy-initializator. Because of the last two, we
1523          * must ensure mutual exclusion between this call and the object reading
1524          * API, thus we use obj_read_lock() here.
1525          *
1526          * TODO: allowing text conversion to run in parallel with object
1527          * reading operations might increase performance in the multithreaded
1528          * non-worktreee git-grep with --textconv.
1529          */
1530         obj_read_lock();
1531         size = fill_textconv(r, driver, df, &buf);
1532         obj_read_unlock();
1533         free_filespec(df);
1534
1535         /*
1536          * The normal fill_textconv usage by the diff machinery would just keep
1537          * the textconv'd buf separate from the diff_filespec. But much of the
1538          * grep code passes around a grep_source and assumes that its "buf"
1539          * pointer is the beginning of the thing we are searching. So let's
1540          * install our textconv'd version into the grep_source, taking care not
1541          * to leak any existing buffer.
1542          */
1543         grep_source_clear_data(gs);
1544         gs->buf = buf;
1545         gs->size = size;
1546
1547         return 0;
1548 }
1549
1550 static int is_empty_line(const char *bol, const char *eol)
1551 {
1552         while (bol < eol && isspace(*bol))
1553                 bol++;
1554         return bol == eol;
1555 }
1556
1557 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1558 {
1559         const char *bol;
1560         const char *peek_bol = NULL;
1561         unsigned long left;
1562         unsigned lno = 1;
1563         unsigned last_hit = 0;
1564         int binary_match_only = 0;
1565         unsigned count = 0;
1566         int try_lookahead = 0;
1567         int show_function = 0;
1568         struct userdiff_driver *textconv = NULL;
1569         enum grep_context ctx = GREP_CONTEXT_HEAD;
1570         xdemitconf_t xecfg;
1571
1572         if (!opt->status_only && gs->name == NULL)
1573                 BUG("grep call which could print a name requires "
1574                     "grep_source.name be non-NULL");
1575
1576         if (!opt->output)
1577                 opt->output = std_output;
1578
1579         if (opt->pre_context || opt->post_context || opt->file_break ||
1580             opt->funcbody) {
1581                 /* Show hunk marks, except for the first file. */
1582                 if (opt->last_shown)
1583                         opt->show_hunk_mark = 1;
1584                 /*
1585                  * If we're using threads then we can't easily identify
1586                  * the first file.  Always put hunk marks in that case
1587                  * and skip the very first one later in work_done().
1588                  */
1589                 if (opt->output != std_output)
1590                         opt->show_hunk_mark = 1;
1591         }
1592         opt->last_shown = 0;
1593
1594         if (opt->allow_textconv) {
1595                 grep_source_load_driver(gs, opt->repo->index);
1596                 /*
1597                  * We might set up the shared textconv cache data here, which
1598                  * is not thread-safe. Also, get_oid_with_context() and
1599                  * parse_object() might be internally called. As they are not
1600                  * currently thread-safe and might be racy with object reading,
1601                  * obj_read_lock() must be called.
1602                  */
1603                 grep_attr_lock();
1604                 obj_read_lock();
1605                 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1606                 obj_read_unlock();
1607                 grep_attr_unlock();
1608         }
1609
1610         /*
1611          * We know the result of a textconv is text, so we only have to care
1612          * about binary handling if we are not using it.
1613          */
1614         if (!textconv) {
1615                 switch (opt->binary) {
1616                 case GREP_BINARY_DEFAULT:
1617                         if (grep_source_is_binary(gs, opt->repo->index))
1618                                 binary_match_only = 1;
1619                         break;
1620                 case GREP_BINARY_NOMATCH:
1621                         if (grep_source_is_binary(gs, opt->repo->index))
1622                                 return 0; /* Assume unmatch */
1623                         break;
1624                 case GREP_BINARY_TEXT:
1625                         break;
1626                 default:
1627                         BUG("unknown binary handling mode");
1628                 }
1629         }
1630
1631         memset(&xecfg, 0, sizeof(xecfg));
1632         opt->priv = &xecfg;
1633
1634         try_lookahead = should_lookahead(opt);
1635
1636         if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1637                 return 0;
1638
1639         bol = gs->buf;
1640         left = gs->size;
1641         while (left) {
1642                 const char *eol;
1643                 int hit;
1644                 ssize_t cno;
1645                 ssize_t col = -1, icol = -1;
1646
1647                 /*
1648                  * look_ahead() skips quickly to the line that possibly
1649                  * has the next hit; don't call it if we need to do
1650                  * something more than just skipping the current line
1651                  * in response to an unmatch for the current line.  E.g.
1652                  * inside a post-context window, we will show the current
1653                  * line as a context around the previous hit when it
1654                  * doesn't hit.
1655                  */
1656                 if (try_lookahead
1657                     && !(last_hit
1658                          && (show_function ||
1659                              lno <= last_hit + opt->post_context))
1660                     && look_ahead(opt, &left, &lno, &bol))
1661                         break;
1662                 eol = end_of_line(bol, &left);
1663
1664                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1665                         ctx = GREP_CONTEXT_BODY;
1666
1667                 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1668
1669                 if (collect_hits)
1670                         goto next_line;
1671
1672                 /* "grep -v -e foo -e bla" should list lines
1673                  * that do not have either, so inversion should
1674                  * be done outside.
1675                  */
1676                 if (opt->invert)
1677                         hit = !hit;
1678                 if (opt->unmatch_name_only) {
1679                         if (hit)
1680                                 return 0;
1681                         goto next_line;
1682                 }
1683                 if (hit && (opt->max_count < 0 || count < opt->max_count)) {
1684                         count++;
1685                         if (opt->status_only)
1686                                 return 1;
1687                         if (opt->name_only) {
1688                                 show_name(opt, gs->name);
1689                                 return 1;
1690                         }
1691                         if (opt->count)
1692                                 goto next_line;
1693                         if (binary_match_only) {
1694                                 opt->output(opt, "Binary file ", 12);
1695                                 output_color(opt, gs->name, strlen(gs->name),
1696                                              opt->colors[GREP_COLOR_FILENAME]);
1697                                 opt->output(opt, " matches\n", 9);
1698                                 return 1;
1699                         }
1700                         /* Hit at this line.  If we haven't shown the
1701                          * pre-context lines, we would need to show them.
1702                          */
1703                         if (opt->pre_context || opt->funcbody)
1704                                 show_pre_context(opt, gs, bol, eol, lno);
1705                         else if (opt->funcname)
1706                                 show_funcname_line(opt, gs, bol, lno);
1707                         cno = opt->invert ? icol : col;
1708                         if (cno < 0) {
1709                                 /*
1710                                  * A negative cno indicates that there was no
1711                                  * match on the line. We are thus inverted and
1712                                  * being asked to show all lines that _don't_
1713                                  * match a given expression. Therefore, set cno
1714                                  * to 0 to suggest the whole line matches.
1715                                  */
1716                                 cno = 0;
1717                         }
1718                         show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1719                         last_hit = lno;
1720                         if (opt->funcbody)
1721                                 show_function = 1;
1722                         goto next_line;
1723                 }
1724                 if (show_function && (!peek_bol || peek_bol < bol)) {
1725                         unsigned long peek_left = left;
1726                         const char *peek_eol = eol;
1727
1728                         /*
1729                          * Trailing empty lines are not interesting.
1730                          * Peek past them to see if they belong to the
1731                          * body of the current function.
1732                          */
1733                         peek_bol = bol;
1734                         while (is_empty_line(peek_bol, peek_eol)) {
1735                                 peek_bol = peek_eol + 1;
1736                                 peek_eol = end_of_line(peek_bol, &peek_left);
1737                         }
1738
1739                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1740                                 show_function = 0;
1741                 }
1742                 if (show_function ||
1743                     (last_hit && lno <= last_hit + opt->post_context)) {
1744                         /* If the last hit is within the post context,
1745                          * we need to show this line.
1746                          */
1747                         show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1748                 }
1749
1750         next_line:
1751                 bol = eol + 1;
1752                 if (!left)
1753                         break;
1754                 left--;
1755                 lno++;
1756         }
1757
1758         if (collect_hits)
1759                 return 0;
1760
1761         if (opt->status_only)
1762                 return opt->unmatch_name_only;
1763         if (opt->unmatch_name_only) {
1764                 /* We did not see any hit, so we want to show this */
1765                 show_name(opt, gs->name);
1766                 return 1;
1767         }
1768
1769         xdiff_clear_find_func(&xecfg);
1770         opt->priv = NULL;
1771
1772         /* NEEDSWORK:
1773          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1774          * which feels mostly useless but sometimes useful.  Maybe
1775          * make it another option?  For now suppress them.
1776          */
1777         if (opt->count && count) {
1778                 char buf[32];
1779                 if (opt->pathname) {
1780                         output_color(opt, gs->name, strlen(gs->name),
1781                                      opt->colors[GREP_COLOR_FILENAME]);
1782                         output_sep(opt, ':');
1783                 }
1784                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1785                 opt->output(opt, buf, strlen(buf));
1786                 return 1;
1787         }
1788         return !!last_hit;
1789 }
1790
1791 static void clr_hit_marker(struct grep_expr *x)
1792 {
1793         /* All-hit markers are meaningful only at the very top level
1794          * OR node.
1795          */
1796         while (1) {
1797                 x->hit = 0;
1798                 if (x->node != GREP_NODE_OR)
1799                         return;
1800                 x->u.binary.left->hit = 0;
1801                 x = x->u.binary.right;
1802         }
1803 }
1804
1805 static int chk_hit_marker(struct grep_expr *x)
1806 {
1807         /* Top level nodes have hit markers.  See if they all are hits */
1808         while (1) {
1809                 if (x->node != GREP_NODE_OR)
1810                         return x->hit;
1811                 if (!x->u.binary.left->hit)
1812                         return 0;
1813                 x = x->u.binary.right;
1814         }
1815 }
1816
1817 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1818 {
1819         /*
1820          * we do not have to do the two-pass grep when we do not check
1821          * buffer-wide "all-match".
1822          */
1823         if (!opt->all_match && !opt->no_body_match)
1824                 return grep_source_1(opt, gs, 0);
1825
1826         /* Otherwise the toplevel "or" terms hit a bit differently.
1827          * We first clear hit markers from them.
1828          */
1829         clr_hit_marker(opt->pattern_expression);
1830         opt->body_hit = 0;
1831         grep_source_1(opt, gs, 1);
1832
1833         if (opt->all_match && !chk_hit_marker(opt->pattern_expression))
1834                 return 0;
1835         if (opt->no_body_match && opt->body_hit)
1836                 return 0;
1837
1838         return grep_source_1(opt, gs, 0);
1839 }
1840
1841 static void grep_source_init_buf(struct grep_source *gs,
1842                                  const char *buf,
1843                                  unsigned long size)
1844 {
1845         gs->type = GREP_SOURCE_BUF;
1846         gs->name = NULL;
1847         gs->path = NULL;
1848         gs->buf = buf;
1849         gs->size = size;
1850         gs->driver = NULL;
1851         gs->identifier = NULL;
1852 }
1853
1854 int grep_buffer(struct grep_opt *opt, const char *buf, unsigned long size)
1855 {
1856         struct grep_source gs;
1857         int r;
1858
1859         grep_source_init_buf(&gs, buf, size);
1860
1861         r = grep_source(opt, &gs);
1862
1863         grep_source_clear(&gs);
1864         return r;
1865 }
1866
1867 void grep_source_init_file(struct grep_source *gs, const char *name,
1868                            const char *path)
1869 {
1870         gs->type = GREP_SOURCE_FILE;
1871         gs->name = xstrdup_or_null(name);
1872         gs->path = xstrdup_or_null(path);
1873         gs->buf = NULL;
1874         gs->size = 0;
1875         gs->driver = NULL;
1876         gs->identifier = xstrdup(path);
1877 }
1878
1879 void grep_source_init_oid(struct grep_source *gs, const char *name,
1880                           const char *path, const struct object_id *oid,
1881                           struct repository *repo)
1882 {
1883         gs->type = GREP_SOURCE_OID;
1884         gs->name = xstrdup_or_null(name);
1885         gs->path = xstrdup_or_null(path);
1886         gs->buf = NULL;
1887         gs->size = 0;
1888         gs->driver = NULL;
1889         gs->identifier = oiddup(oid);
1890         gs->repo = repo;
1891 }
1892
1893 void grep_source_clear(struct grep_source *gs)
1894 {
1895         FREE_AND_NULL(gs->name);
1896         FREE_AND_NULL(gs->path);
1897         FREE_AND_NULL(gs->identifier);
1898         grep_source_clear_data(gs);
1899 }
1900
1901 void grep_source_clear_data(struct grep_source *gs)
1902 {
1903         switch (gs->type) {
1904         case GREP_SOURCE_FILE:
1905         case GREP_SOURCE_OID:
1906                 /* these types own the buffer */
1907                 free((char *)gs->buf);
1908                 gs->buf = NULL;
1909                 gs->size = 0;
1910                 break;
1911         case GREP_SOURCE_BUF:
1912                 /* leave user-provided buf intact */
1913                 break;
1914         }
1915 }
1916
1917 static int grep_source_load_oid(struct grep_source *gs)
1918 {
1919         enum object_type type;
1920
1921         gs->buf = repo_read_object_file(gs->repo, gs->identifier, &type,
1922                                         &gs->size);
1923         if (!gs->buf)
1924                 return error(_("'%s': unable to read %s"),
1925                              gs->name,
1926                              oid_to_hex(gs->identifier));
1927         return 0;
1928 }
1929
1930 static int grep_source_load_file(struct grep_source *gs)
1931 {
1932         const char *filename = gs->identifier;
1933         struct stat st;
1934         char *data;
1935         size_t size;
1936         int i;
1937
1938         if (lstat(filename, &st) < 0) {
1939         err_ret:
1940                 if (errno != ENOENT)
1941                         error_errno(_("failed to stat '%s'"), filename);
1942                 return -1;
1943         }
1944         if (!S_ISREG(st.st_mode))
1945                 return -1;
1946         size = xsize_t(st.st_size);
1947         i = open(filename, O_RDONLY);
1948         if (i < 0)
1949                 goto err_ret;
1950         data = xmallocz(size);
1951         if (st.st_size != read_in_full(i, data, size)) {
1952                 error_errno(_("'%s': short read"), filename);
1953                 close(i);
1954                 free(data);
1955                 return -1;
1956         }
1957         close(i);
1958
1959         gs->buf = data;
1960         gs->size = size;
1961         return 0;
1962 }
1963
1964 static int grep_source_load(struct grep_source *gs)
1965 {
1966         if (gs->buf)
1967                 return 0;
1968
1969         switch (gs->type) {
1970         case GREP_SOURCE_FILE:
1971                 return grep_source_load_file(gs);
1972         case GREP_SOURCE_OID:
1973                 return grep_source_load_oid(gs);
1974         case GREP_SOURCE_BUF:
1975                 return gs->buf ? 0 : -1;
1976         }
1977         BUG("invalid grep_source type to load");
1978 }
1979
1980 void grep_source_load_driver(struct grep_source *gs,
1981                              struct index_state *istate)
1982 {
1983         if (gs->driver)
1984                 return;
1985
1986         grep_attr_lock();
1987         if (gs->path)
1988                 gs->driver = userdiff_find_by_path(istate, gs->path);
1989         if (!gs->driver)
1990                 gs->driver = userdiff_find_by_name("default");
1991         grep_attr_unlock();
1992 }
1993
1994 static int grep_source_is_binary(struct grep_source *gs,
1995                                  struct index_state *istate)
1996 {
1997         grep_source_load_driver(gs, istate);
1998         if (gs->driver->binary != -1)
1999                 return gs->driver->binary;
2000
2001         if (!grep_source_load(gs))
2002                 return buffer_is_binary(gs->buf, gs->size);
2003
2004         return 0;
2005 }