grep.c

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "gettext.h"
   4 #include "grep.h"
   5 #include "hex.h"
   6 #include "object-store.h"
   7 #include "userdiff.h"
   8 #include "xdiff-interface.h"
   9 #include "diff.h"
  10 #include "diffcore.h"
  11 #include "commit.h"
  12 #include "quote.h"
  13 #include "help.h"
  14
  15 static int grep_source_load(struct grep_source *gs);
  16 static int grep_source_is_binary(struct grep_source *gs,
  17                                  struct index_state *istate);
  18
  19 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
  20 {
  21         fwrite(buf, size, 1, stdout);
  22 }
  23
  24 static const char *color_grep_slots[] = {
  25         [GREP_COLOR_CONTEXT]        = "context",
  26         [GREP_COLOR_FILENAME]       = "filename",
  27         [GREP_COLOR_FUNCTION]       = "function",
  28         [GREP_COLOR_LINENO]         = "lineNumber",
  29         [GREP_COLOR_COLUMNNO]       = "column",
  30         [GREP_COLOR_MATCH_CONTEXT]  = "matchContext",
  31         [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
  32         [GREP_COLOR_SELECTED]       = "selected",
  33         [GREP_COLOR_SEP]            = "separator",
  34 };
  35
  36 static int parse_pattern_type_arg(const char *opt, const char *arg)
  37 {
  38         if (!strcmp(arg, "default"))
  39                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  40         else if (!strcmp(arg, "basic"))
  41                 return GREP_PATTERN_TYPE_BRE;
  42         else if (!strcmp(arg, "extended"))
  43                 return GREP_PATTERN_TYPE_ERE;
  44         else if (!strcmp(arg, "fixed"))
  45                 return GREP_PATTERN_TYPE_FIXED;
  46         else if (!strcmp(arg, "perl"))
  47                 return GREP_PATTERN_TYPE_PCRE;
  48         die("bad %s argument: %s", opt, arg);
  49 }
  50
  51 define_list_config_array_extra(color_grep_slots, {"match"});
  52
  53 /*
  54  * Read the configuration file once and store it in
  55  * the grep_defaults template.
  56  */
  57 int grep_config(const char *var, const char *value, void *cb)
  58 {
  59         struct grep_opt *opt = cb;
  60         const char *slot;
  61
  62         if (userdiff_config(var, value) < 0)
  63                 return -1;
  64
  65         if (!strcmp(var, "grep.extendedregexp")) {
  66                 opt->extended_regexp_option = git_config_bool(var, value);
  67                 return 0;
  68         }
  69
  70         if (!strcmp(var, "grep.patterntype")) {
  71                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
  72                 return 0;
  73         }
  74
  75         if (!strcmp(var, "grep.linenumber")) {
  76                 opt->linenum = git_config_bool(var, value);
  77                 return 0;
  78         }
  79         if (!strcmp(var, "grep.column")) {
  80                 opt->columnnum = git_config_bool(var, value);
  81                 return 0;
  82         }
  83
  84         if (!strcmp(var, "grep.fullname")) {
  85                 opt->relative = !git_config_bool(var, value);
  86                 return 0;
  87         }
  88
  89         if (!strcmp(var, "color.grep"))
  90                 opt->color = git_config_colorbool(var, value);
  91         if (!strcmp(var, "color.grep.match")) {
  92                 if (grep_config("color.grep.matchcontext", value, cb) < 0)
  93                         return -1;
  94                 if (grep_config("color.grep.matchselected", value, cb) < 0)
  95                         return -1;
  96         } else if (skip_prefix(var, "color.grep.", &slot)) {
  97                 int i = LOOKUP_CONFIG(color_grep_slots, slot);
  98                 char *color;
  99
 100                 if (i < 0)
 101                         return -1;
 102                 color = opt->colors[i];
 103                 if (!value)
 104                         return config_error_nonbool(var);
 105                 return color_parse(value, color);
 106         }
 107         return 0;
 108 }
 109
 110 void grep_init(struct grep_opt *opt, struct repository *repo)
 111 {
 112         struct grep_opt blank = GREP_OPT_INIT;
 113         memcpy(opt, &blank, sizeof(*opt));
 114
 115         opt->repo = repo;
 116         opt->pattern_tail = &opt->pattern_list;
 117         opt->header_tail = &opt->header_list;
 118 }
 119
 120 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 121                                         const char *origin, int no,
 122                                         enum grep_pat_token t,
 123                                         enum grep_header_field field)
 124 {
 125         struct grep_pat *p = xcalloc(1, sizeof(*p));
 126         p->pattern = xmemdupz(pat, patlen);
 127         p->patternlen = patlen;
 128         p->origin = origin;
 129         p->no = no;
 130         p->token = t;
 131         p->field = field;
 132         return p;
 133 }
 134
 135 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 136 {
 137         **tail = p;
 138         *tail = &p->next;
 139         p->next = NULL;
 140
 141         switch (p->token) {
 142         case GREP_PATTERN: /* atom */
 143         case GREP_PATTERN_HEAD:
 144         case GREP_PATTERN_BODY:
 145                 for (;;) {
 146                         struct grep_pat *new_pat;
 147                         size_t len = 0;
 148                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 149                         while (++len <= p->patternlen) {
 150                                 if (*(--cp) == '\n') {
 151                                         nl = cp;
 152                                         break;
 153                                 }
 154                         }
 155                         if (!nl)
 156                                 break;
 157                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 158                                                   p->no, p->token, p->field);
 159                         new_pat->next = p->next;
 160                         if (!p->next)
 161                                 *tail = &new_pat->next;
 162                         p->next = new_pat;
 163                         *nl = '\0';
 164                         p->patternlen -= len;
 165                 }
 166                 break;
 167         default:
 168                 break;
 169         }
 170 }
 171
 172 void append_header_grep_pattern(struct grep_opt *opt,
 173                                 enum grep_header_field field, const char *pat)
 174 {
 175         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 176                                              GREP_PATTERN_HEAD, field);
 177         if (field == GREP_HEADER_REFLOG)
 178                 opt->use_reflog_filter = 1;
 179         do_append_grep_pat(&opt->header_tail, p);
 180 }
 181
 182 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 183                          const char *origin, int no, enum grep_pat_token t)
 184 {
 185         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 186 }
 187
 188 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 189                      const char *origin, int no, enum grep_pat_token t)
 190 {
 191         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 192         do_append_grep_pat(&opt->pattern_tail, p);
 193 }
 194
 195 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 196 {
 197         struct grep_pat *pat;
 198         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 199         *ret = *opt;
 200
 201         ret->pattern_list = NULL;
 202         ret->pattern_tail = &ret->pattern_list;
 203
 204         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 205         {
 206                 if(pat->token == GREP_PATTERN_HEAD)
 207                         append_header_grep_pattern(ret, pat->field,
 208                                                    pat->pattern);
 209                 else
 210                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 211                                         pat->origin, pat->no, pat->token);
 212         }
 213
 214         return ret;
 215 }
 216
 217 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 218                 const char *error)
 219 {
 220         char where[1024];
 221
 222         if (p->no)
 223                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 224         else if (p->origin)
 225                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 226         else
 227                 where[0] = 0;
 228
 229         die("%s'%s': %s", where, p->pattern, error);
 230 }
 231
 232 static int is_fixed(const char *s, size_t len)
 233 {
 234         size_t i;
 235
 236         for (i = 0; i < len; i++) {
 237                 if (is_regex_special(s[i]))
 238                         return 0;
 239         }
 240
 241         return 1;
 242 }
 243
 244 #ifdef USE_LIBPCRE2
 245 #define GREP_PCRE2_DEBUG_MALLOC 0
 246
 247 static void *pcre2_malloc(PCRE2_SIZE size, MAYBE_UNUSED void *memory_data)
 248 {
 249         void *pointer = malloc(size);
 250 #if GREP_PCRE2_DEBUG_MALLOC
 251         static int count = 1;
 252         fprintf(stderr, "PCRE2:%p -> #%02d: alloc(%lu)\n", pointer, count++, size);
 253 #endif
 254         return pointer;
 255 }
 256
 257 static void pcre2_free(void *pointer, MAYBE_UNUSED void *memory_data)
 258 {
 259 #if GREP_PCRE2_DEBUG_MALLOC
 260         static int count = 1;
 261         if (pointer)
 262                 fprintf(stderr, "PCRE2:%p -> #%02d: free()\n", pointer, count++);
 263 #endif
 264         free(pointer);
 265 }
 266
 267 static int pcre2_jit_functional(void)
 268 {
 269         static int jit_working = -1;
 270         pcre2_code *code;
 271         size_t off;
 272         int err;
 273
 274         if (jit_working != -1)
 275                 return jit_working;
 276
 277         /*
 278          * Try to JIT compile a simple pattern to probe if the JIT is
 279          * working in general. It might fail for systems where creating
 280          * memory mappings for runtime code generation is restricted.
 281          */
 282         code = pcre2_compile((PCRE2_SPTR)".", 1, 0, &err, &off, NULL);
 283         if (!code)
 284                 return 0;
 285
 286         jit_working = pcre2_jit_compile(code, PCRE2_JIT_COMPLETE) == 0;
 287         pcre2_code_free(code);
 288
 289         return jit_working;
 290 }
 291
 292 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 293 {
 294         int error;
 295         PCRE2_UCHAR errbuf[256];
 296         PCRE2_SIZE erroffset;
 297         int options = PCRE2_MULTILINE;
 298         int jitret;
 299         int patinforet;
 300         size_t jitsizearg;
 301         int literal = !opt->ignore_case && (p->fixed || p->is_fixed);
 302
 303         /*
 304          * Call pcre2_general_context_create() before calling any
 305          * other pcre2_*(). It sets up our malloc()/free() functions
 306          * with which everything else is allocated.
 307          */
 308         p->pcre2_general_context = pcre2_general_context_create(
 309                 pcre2_malloc, pcre2_free, NULL);
 310         if (!p->pcre2_general_context)
 311                 die("Couldn't allocate PCRE2 general context");
 312
 313         if (opt->ignore_case) {
 314                 if (!opt->ignore_locale && has_non_ascii(p->pattern)) {
 315                         p->pcre2_tables = pcre2_maketables(p->pcre2_general_context);
 316                         p->pcre2_compile_context = pcre2_compile_context_create(p->pcre2_general_context);
 317                         pcre2_set_character_tables(p->pcre2_compile_context,
 318                                                         p->pcre2_tables);
 319                 }
 320                 options |= PCRE2_CASELESS;
 321         }
 322         if (!opt->ignore_locale && is_utf8_locale() && !literal)
 323                 options |= (PCRE2_UTF | PCRE2_UCP | PCRE2_MATCH_INVALID_UTF);
 324
 325 #ifndef GIT_PCRE2_VERSION_10_36_OR_HIGHER
 326         /* Work around https://bugs.exim.org/show_bug.cgi?id=2642 fixed in 10.36 */
 327         if (PCRE2_MATCH_INVALID_UTF && options & (PCRE2_UTF | PCRE2_CASELESS))
 328                 options |= PCRE2_NO_START_OPTIMIZE;
 329 #endif
 330
 331         p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
 332                                          p->patternlen, options, &error, &erroffset,
 333                                          p->pcre2_compile_context);
 334
 335         if (p->pcre2_pattern) {
 336                 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, p->pcre2_general_context);
 337                 if (!p->pcre2_match_data)
 338                         die("Couldn't allocate PCRE2 match data");
 339         } else {
 340                 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
 341                 compile_regexp_failed(p, (const char *)&errbuf);
 342         }
 343
 344         pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
 345         if (p->pcre2_jit_on) {
 346                 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
 347                 if (jitret == PCRE2_ERROR_NOMEMORY && !pcre2_jit_functional()) {
 348                         /*
 349                          * Even though pcre2_config(PCRE2_CONFIG_JIT, ...)
 350                          * indicated JIT support, the library might still
 351                          * fail to generate JIT code for various reasons,
 352                          * e.g. when SELinux's 'deny_execmem' or PaX's
 353                          * MPROTECT prevent creating W|X memory mappings.
 354                          *
 355                          * Instead of faling hard, fall back to interpreter
 356                          * mode, just as if the pattern was prefixed with
 357                          * '(*NO_JIT)'.
 358                          */
 359                         p->pcre2_jit_on = 0;
 360                         return;
 361                 } else if (jitret) {
 362                         int need_clip = p->patternlen > 64;
 363                         int clip_len = need_clip ? 64 : p->patternlen;
 364                         die("Couldn't JIT the PCRE2 pattern '%.*s'%s, got '%d'%s",
 365                             clip_len, p->pattern, need_clip ? "..." : "", jitret,
 366                             pcre2_jit_functional()
 367                             ? "\nPerhaps prefix (*NO_JIT) to your pattern?"
 368                             : "");
 369                 }
 370
 371                 /*
 372                  * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
 373                  * tells us whether the library itself supports JIT,
 374                  * but to see whether we're going to be actually using
 375                  * JIT we need to extract PCRE2_INFO_JITSIZE from the
 376                  * pattern *after* we do pcre2_jit_compile() above.
 377                  *
 378                  * This is because if the pattern contains the
 379                  * (*NO_JIT) verb (see pcre2syntax(3))
 380                  * pcre2_jit_compile() will exit early with 0. If we
 381                  * then proceed to call pcre2_jit_match() further down
 382                  * the line instead of pcre2_match() we'll either
 383                  * segfault (pre PCRE 10.31) or run into a fatal error
 384                  * (post PCRE2 10.31)
 385                  */
 386                 patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
 387                 if (patinforet)
 388                         BUG("pcre2_pattern_info() failed: %d", patinforet);
 389                 if (jitsizearg == 0) {
 390                         p->pcre2_jit_on = 0;
 391                         return;
 392                 }
 393         }
 394 }
 395
 396 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 397                 regmatch_t *match, int eflags)
 398 {
 399         int ret, flags = 0;
 400         PCRE2_SIZE *ovector;
 401         PCRE2_UCHAR errbuf[256];
 402
 403         if (eflags & REG_NOTBOL)
 404                 flags |= PCRE2_NOTBOL;
 405
 406         if (p->pcre2_jit_on)
 407                 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 408                                       eol - line, 0, flags, p->pcre2_match_data,
 409                                       NULL);
 410         else
 411                 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 412                                   eol - line, 0, flags, p->pcre2_match_data,
 413                                   NULL);
 414
 415         if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 416                 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 417                 die("%s failed with error code %d: %s",
 418                     (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 419                     errbuf);
 420         }
 421         if (ret > 0) {
 422                 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 423                 ret = 0;
 424                 match->rm_so = (int)ovector[0];
 425                 match->rm_eo = (int)ovector[1];
 426         }
 427
 428         return ret;
 429 }
 430
 431 static void free_pcre2_pattern(struct grep_pat *p)
 432 {
 433         pcre2_compile_context_free(p->pcre2_compile_context);
 434         pcre2_code_free(p->pcre2_pattern);
 435         pcre2_match_data_free(p->pcre2_match_data);
 436 #ifdef GIT_PCRE2_VERSION_10_34_OR_HIGHER
 437         pcre2_maketables_free(p->pcre2_general_context, p->pcre2_tables);
 438 #else
 439         free((void *)p->pcre2_tables);
 440 #endif
 441         pcre2_general_context_free(p->pcre2_general_context);
 442 }
 443 #else /* !USE_LIBPCRE2 */
 444 static void compile_pcre2_pattern(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 pcre2match(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_pcre2_pattern(struct grep_pat *p)
 456 {
 457 }
 458
 459 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 460 {
 461         struct strbuf sb = STRBUF_INIT;
 462         int err;
 463         int regflags = 0;
 464
 465         basic_regex_quote_buf(&sb, p->pattern);
 466         if (opt->ignore_case)
 467                 regflags |= REG_ICASE;
 468         err = regcomp(&p->regexp, sb.buf, regflags);
 469         strbuf_release(&sb);
 470         if (err) {
 471                 char errbuf[1024];
 472                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 473                 compile_regexp_failed(p, errbuf);
 474         }
 475 }
 476 #endif /* !USE_LIBPCRE2 */
 477
 478 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 479 {
 480         int err;
 481         int regflags = REG_NEWLINE;
 482
 483         if (opt->pattern_type_option == GREP_PATTERN_TYPE_UNSPECIFIED)
 484                 opt->pattern_type_option = (opt->extended_regexp_option
 485                                             ? GREP_PATTERN_TYPE_ERE
 486                                             : GREP_PATTERN_TYPE_BRE);
 487
 488         p->word_regexp = opt->word_regexp;
 489         p->ignore_case = opt->ignore_case;
 490         p->fixed = opt->pattern_type_option == GREP_PATTERN_TYPE_FIXED;
 491
 492         if (opt->pattern_type_option != GREP_PATTERN_TYPE_PCRE &&
 493             memchr(p->pattern, 0, p->patternlen))
 494                 die(_("given pattern contains NULL byte (via -f <file>). This is only supported with -P under PCRE v2"));
 495
 496         p->is_fixed = is_fixed(p->pattern, p->patternlen);
 497 #ifdef USE_LIBPCRE2
 498        if (!p->fixed && !p->is_fixed) {
 499                const char *no_jit = "(*NO_JIT)";
 500                const int no_jit_len = strlen(no_jit);
 501                if (starts_with(p->pattern, no_jit) &&
 502                    is_fixed(p->pattern + no_jit_len,
 503                             p->patternlen - no_jit_len))
 504                        p->is_fixed = 1;
 505        }
 506 #endif
 507         if (p->fixed || p->is_fixed) {
 508 #ifdef USE_LIBPCRE2
 509                 if (p->is_fixed) {
 510                         compile_pcre2_pattern(p, opt);
 511                 } else {
 512                         /*
 513                          * E.g. t7811-grep-open.sh relies on the
 514                          * pattern being restored.
 515                          */
 516                         char *old_pattern = p->pattern;
 517                         size_t old_patternlen = p->patternlen;
 518                         struct strbuf sb = STRBUF_INIT;
 519
 520                         /*
 521                          * There is the PCRE2_LITERAL flag, but it's
 522                          * only in PCRE v2 10.30 and later. Needing to
 523                          * ifdef our way around that and dealing with
 524                          * it + PCRE2_MULTILINE being an error is more
 525                          * complex than just quoting this ourselves.
 526                         */
 527                         strbuf_add(&sb, "\\Q", 2);
 528                         strbuf_add(&sb, p->pattern, p->patternlen);
 529                         strbuf_add(&sb, "\\E", 2);
 530
 531                         p->pattern = sb.buf;
 532                         p->patternlen = sb.len;
 533                         compile_pcre2_pattern(p, opt);
 534                         p->pattern = old_pattern;
 535                         p->patternlen = old_patternlen;
 536                         strbuf_release(&sb);
 537                 }
 538 #else /* !USE_LIBPCRE2 */
 539                 compile_fixed_regexp(p, opt);
 540 #endif /* !USE_LIBPCRE2 */
 541                 return;
 542         }
 543
 544         if (opt->pattern_type_option == GREP_PATTERN_TYPE_PCRE) {
 545                 compile_pcre2_pattern(p, opt);
 546                 return;
 547         }
 548
 549         if (p->ignore_case)
 550                 regflags |= REG_ICASE;
 551         if (opt->pattern_type_option == GREP_PATTERN_TYPE_ERE)
 552                 regflags |= REG_EXTENDED;
 553         err = regcomp(&p->regexp, p->pattern, regflags);
 554         if (err) {
 555                 char errbuf[1024];
 556                 regerror(err, &p->regexp, errbuf, 1024);
 557                 compile_regexp_failed(p, errbuf);
 558         }
 559 }
 560
 561 static struct grep_expr *grep_not_expr(struct grep_expr *expr)
 562 {
 563         struct grep_expr *z = xcalloc(1, sizeof(*z));
 564         z->node = GREP_NODE_NOT;
 565         z->u.unary = expr;
 566         return z;
 567 }
 568
 569 static struct grep_expr *grep_binexp(enum grep_expr_node kind,
 570                                      struct grep_expr *left,
 571                                      struct grep_expr *right)
 572 {
 573         struct grep_expr *z = xcalloc(1, sizeof(*z));
 574         z->node = kind;
 575         z->u.binary.left = left;
 576         z->u.binary.right = right;
 577         return z;
 578 }
 579
 580 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 581 {
 582         return grep_binexp(GREP_NODE_OR, left, right);
 583 }
 584
 585 static struct grep_expr *grep_and_expr(struct grep_expr *left, struct grep_expr *right)
 586 {
 587         return grep_binexp(GREP_NODE_AND, left, right);
 588 }
 589
 590 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 591 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 592 {
 593         struct grep_pat *p;
 594         struct grep_expr *x;
 595
 596         p = *list;
 597         if (!p)
 598                 return NULL;
 599         switch (p->token) {
 600         case GREP_PATTERN: /* atom */
 601         case GREP_PATTERN_HEAD:
 602         case GREP_PATTERN_BODY:
 603                 CALLOC_ARRAY(x, 1);
 604                 x->node = GREP_NODE_ATOM;
 605                 x->u.atom = p;
 606                 *list = p->next;
 607                 return x;
 608         case GREP_OPEN_PAREN:
 609                 *list = p->next;
 610                 x = compile_pattern_or(list);
 611                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 612                         die("unmatched parenthesis");
 613                 *list = (*list)->next;
 614                 return x;
 615         default:
 616                 return NULL;
 617         }
 618 }
 619
 620 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 621 {
 622         struct grep_pat *p;
 623         struct grep_expr *x;
 624
 625         p = *list;
 626         if (!p)
 627                 return NULL;
 628         switch (p->token) {
 629         case GREP_NOT:
 630                 if (!p->next)
 631                         die("--not not followed by pattern expression");
 632                 *list = p->next;
 633                 x = compile_pattern_not(list);
 634                 if (!x)
 635                         die("--not followed by non pattern expression");
 636                 return grep_not_expr(x);
 637         default:
 638                 return compile_pattern_atom(list);
 639         }
 640 }
 641
 642 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 643 {
 644         struct grep_pat *p;
 645         struct grep_expr *x, *y;
 646
 647         x = compile_pattern_not(list);
 648         p = *list;
 649         if (p && p->token == GREP_AND) {
 650                 if (!x)
 651                         die("--and not preceded by pattern expression");
 652                 if (!p->next)
 653                         die("--and not followed by pattern expression");
 654                 *list = p->next;
 655                 y = compile_pattern_and(list);
 656                 if (!y)
 657                         die("--and not followed by pattern expression");
 658                 return grep_and_expr(x, y);
 659         }
 660         return x;
 661 }
 662
 663 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 664 {
 665         struct grep_pat *p;
 666         struct grep_expr *x, *y;
 667
 668         x = compile_pattern_and(list);
 669         p = *list;
 670         if (x && p && p->token != GREP_CLOSE_PAREN) {
 671                 y = compile_pattern_or(list);
 672                 if (!y)
 673                         die("not a pattern expression %s", p->pattern);
 674                 return grep_or_expr(x, y);
 675         }
 676         return x;
 677 }
 678
 679 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 680 {
 681         return compile_pattern_or(list);
 682 }
 683
 684 static struct grep_expr *grep_true_expr(void)
 685 {
 686         struct grep_expr *z = xcalloc(1, sizeof(*z));
 687         z->node = GREP_NODE_TRUE;
 688         return z;
 689 }
 690
 691 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 692 {
 693         struct grep_pat *p;
 694         struct grep_expr *header_expr;
 695         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 696         enum grep_header_field fld;
 697
 698         if (!opt->header_list)
 699                 return NULL;
 700
 701         for (p = opt->header_list; p; p = p->next) {
 702                 if (p->token != GREP_PATTERN_HEAD)
 703                         BUG("a non-header pattern in grep header list.");
 704                 if (p->field < GREP_HEADER_FIELD_MIN ||
 705                     GREP_HEADER_FIELD_MAX <= p->field)
 706                         BUG("unknown header field %d", p->field);
 707                 compile_regexp(p, opt);
 708         }
 709
 710         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 711                 header_group[fld] = NULL;
 712
 713         for (p = opt->header_list; p; p = p->next) {
 714                 struct grep_expr *h;
 715                 struct grep_pat *pp = p;
 716
 717                 h = compile_pattern_atom(&pp);
 718                 if (!h || pp != p->next)
 719                         BUG("malformed header expr");
 720                 if (!header_group[p->field]) {
 721                         header_group[p->field] = h;
 722                         continue;
 723                 }
 724                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 725         }
 726
 727         header_expr = NULL;
 728
 729         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 730                 if (!header_group[fld])
 731                         continue;
 732                 if (!header_expr)
 733                         header_expr = grep_true_expr();
 734                 header_expr = grep_or_expr(header_group[fld], header_expr);
 735         }
 736         return header_expr;
 737 }
 738
 739 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 740 {
 741         struct grep_expr *z = x;
 742
 743         while (x) {
 744                 assert(x->node == GREP_NODE_OR);
 745                 if (x->u.binary.right &&
 746                     x->u.binary.right->node == GREP_NODE_TRUE) {
 747                         x->u.binary.right = y;
 748                         break;
 749                 }
 750                 x = x->u.binary.right;
 751         }
 752         return z;
 753 }
 754
 755 void compile_grep_patterns(struct grep_opt *opt)
 756 {
 757         struct grep_pat *p;
 758         struct grep_expr *header_expr = prep_header_patterns(opt);
 759         int extended = 0;
 760
 761         for (p = opt->pattern_list; p; p = p->next) {
 762                 switch (p->token) {
 763                 case GREP_PATTERN: /* atom */
 764                 case GREP_PATTERN_HEAD:
 765                 case GREP_PATTERN_BODY:
 766                         compile_regexp(p, opt);
 767                         break;
 768                 default:
 769                         extended = 1;
 770                         break;
 771                 }
 772         }
 773
 774         if (opt->all_match || opt->no_body_match || header_expr)
 775                 extended = 1;
 776         else if (!extended)
 777                 return;
 778
 779         p = opt->pattern_list;
 780         if (p)
 781                 opt->pattern_expression = compile_pattern_expr(&p);
 782         if (p)
 783                 die("incomplete pattern expression: %s", p->pattern);
 784
 785         if (opt->no_body_match && opt->pattern_expression)
 786                 opt->pattern_expression = grep_not_expr(opt->pattern_expression);
 787
 788         if (!header_expr)
 789                 return;
 790
 791         if (!opt->pattern_expression)
 792                 opt->pattern_expression = header_expr;
 793         else if (opt->all_match)
 794                 opt->pattern_expression = grep_splice_or(header_expr,
 795                                                          opt->pattern_expression);
 796         else
 797                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 798                                                        header_expr);
 799         opt->all_match = 1;
 800 }
 801
 802 static void free_pattern_expr(struct grep_expr *x)
 803 {
 804         switch (x->node) {
 805         case GREP_NODE_TRUE:
 806         case GREP_NODE_ATOM:
 807                 break;
 808         case GREP_NODE_NOT:
 809                 free_pattern_expr(x->u.unary);
 810                 break;
 811         case GREP_NODE_AND:
 812         case GREP_NODE_OR:
 813                 free_pattern_expr(x->u.binary.left);
 814                 free_pattern_expr(x->u.binary.right);
 815                 break;
 816         }
 817         free(x);
 818 }
 819
 820 static void free_grep_pat(struct grep_pat *pattern)
 821 {
 822         struct grep_pat *p, *n;
 823
 824         for (p = pattern; p; p = n) {
 825                 n = p->next;
 826                 switch (p->token) {
 827                 case GREP_PATTERN: /* atom */
 828                 case GREP_PATTERN_HEAD:
 829                 case GREP_PATTERN_BODY:
 830                         if (p->pcre2_pattern)
 831                                 free_pcre2_pattern(p);
 832                         else
 833                                 regfree(&p->regexp);
 834                         free(p->pattern);
 835                         break;
 836                 default:
 837                         break;
 838                 }
 839                 free(p);
 840         }
 841 }
 842
 843 void free_grep_patterns(struct grep_opt *opt)
 844 {
 845         free_grep_pat(opt->pattern_list);
 846         free_grep_pat(opt->header_list);
 847
 848         if (opt->pattern_expression)
 849                 free_pattern_expr(opt->pattern_expression);
 850 }
 851
 852 static const char *end_of_line(const char *cp, unsigned long *left)
 853 {
 854         unsigned long l = *left;
 855         while (l && *cp != '\n') {
 856                 l--;
 857                 cp++;
 858         }
 859         *left = l;
 860         return cp;
 861 }
 862
 863 static int word_char(char ch)
 864 {
 865         return isalnum(ch) || ch == '_';
 866 }
 867
 868 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 869                          const char *color)
 870 {
 871         if (want_color(opt->color) && color && color[0]) {
 872                 opt->output(opt, color, strlen(color));
 873                 opt->output(opt, data, size);
 874                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 875         } else
 876                 opt->output(opt, data, size);
 877 }
 878
 879 static void output_sep(struct grep_opt *opt, char sign)
 880 {
 881         if (opt->null_following_name)
 882                 opt->output(opt, "\0", 1);
 883         else
 884                 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
 885 }
 886
 887 static void show_name(struct grep_opt *opt, const char *name)
 888 {
 889         output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
 890         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 891 }
 892
 893 static int patmatch(struct grep_pat *p,
 894                     const char *line, const char *eol,
 895                     regmatch_t *match, int eflags)
 896 {
 897         int hit;
 898
 899         if (p->pcre2_pattern)
 900                 hit = !pcre2match(p, line, eol, match, eflags);
 901         else
 902                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
 903                                    eflags);
 904
 905         return hit;
 906 }
 907
 908 static void strip_timestamp(const char *bol, const char **eol_p)
 909 {
 910         const char *eol = *eol_p;
 911
 912         while (bol < --eol) {
 913                 if (*eol != '>')
 914                         continue;
 915                 *eol_p = ++eol;
 916                 break;
 917         }
 918 }
 919
 920 static struct {
 921         const char *field;
 922         size_t len;
 923 } header_field[] = {
 924         { "author ", 7 },
 925         { "committer ", 10 },
 926         { "reflog ", 7 },
 927 };
 928
 929 static int headerless_match_one_pattern(struct grep_pat *p,
 930                                         const char *bol, const char *eol,
 931                                         enum grep_context ctx,
 932                                         regmatch_t *pmatch, int eflags)
 933 {
 934         int hit = 0;
 935         const char *start = bol;
 936
 937         if ((p->token != GREP_PATTERN) &&
 938             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 939                 return 0;
 940
 941  again:
 942         hit = patmatch(p, bol, eol, pmatch, eflags);
 943
 944         if (hit && p->word_regexp) {
 945                 if ((pmatch[0].rm_so < 0) ||
 946                     (eol - bol) < pmatch[0].rm_so ||
 947                     (pmatch[0].rm_eo < 0) ||
 948                     (eol - bol) < pmatch[0].rm_eo)
 949                         die("regexp returned nonsense");
 950
 951                 /* Match beginning must be either beginning of the
 952                  * line, or at word boundary (i.e. the last char must
 953                  * not be a word char).  Similarly, match end must be
 954                  * either end of the line, or at word boundary
 955                  * (i.e. the next char must not be a word char).
 956                  */
 957                 if ( ((pmatch[0].rm_so == 0) ||
 958                       !word_char(bol[pmatch[0].rm_so-1])) &&
 959                      ((pmatch[0].rm_eo == (eol-bol)) ||
 960                       !word_char(bol[pmatch[0].rm_eo])) )
 961                         ;
 962                 else
 963                         hit = 0;
 964
 965                 /* Words consist of at least one character. */
 966                 if (pmatch->rm_so == pmatch->rm_eo)
 967                         hit = 0;
 968
 969                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 970                         /* There could be more than one match on the
 971                          * line, and the first match might not be
 972                          * strict word match.  But later ones could be!
 973                          * Forward to the next possible start, i.e. the
 974                          * next position following a non-word char.
 975                          */
 976                         bol = pmatch[0].rm_so + bol + 1;
 977                         while (word_char(bol[-1]) && bol < eol)
 978                                 bol++;
 979                         eflags |= REG_NOTBOL;
 980                         if (bol < eol)
 981                                 goto again;
 982                 }
 983         }
 984         if (hit) {
 985                 pmatch[0].rm_so += bol - start;
 986                 pmatch[0].rm_eo += bol - start;
 987         }
 988         return hit;
 989 }
 990
 991 static int match_one_pattern(struct grep_pat *p,
 992                              const char *bol, const char *eol,
 993                              enum grep_context ctx, regmatch_t *pmatch,
 994                              int eflags)
 995 {
 996         const char *field;
 997         size_t len;
 998
 999         if (p->token == GREP_PATTERN_HEAD) {
1000                 assert(p->field < ARRAY_SIZE(header_field));
1001                 field = header_field[p->field].field;
1002                 len = header_field[p->field].len;
1003                 if (strncmp(bol, field, len))
1004                         return 0;
1005                 bol += len;
1006
1007                 switch (p->field) {
1008                 case GREP_HEADER_AUTHOR:
1009                 case GREP_HEADER_COMMITTER:
1010                         strip_timestamp(bol, &eol);
1011                         break;
1012                 default:
1013                         break;
1014                 }
1015         }
1016
1017         return headerless_match_one_pattern(p, bol, eol, ctx, pmatch, eflags);
1018 }
1019
1020
1021 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x,
1022                            const char *bol, const char *eol,
1023                            enum grep_context ctx, ssize_t *col,
1024                            ssize_t *icol, int collect_hits)
1025 {
1026         int h = 0;
1027
1028         switch (x->node) {
1029         case GREP_NODE_TRUE:
1030                 h = 1;
1031                 break;
1032         case GREP_NODE_ATOM:
1033                 {
1034                         regmatch_t tmp;
1035                         h = match_one_pattern(x->u.atom, bol, eol, ctx,
1036                                               &tmp, 0);
1037                         if (h && (*col < 0 || tmp.rm_so < *col))
1038                                 *col = tmp.rm_so;
1039                 }
1040                 if (x->u.atom->token == GREP_PATTERN_BODY)
1041                         opt->body_hit |= h;
1042                 break;
1043         case GREP_NODE_NOT:
1044                 /*
1045                  * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
1046                  */
1047                 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
1048                                      0);
1049                 break;
1050         case GREP_NODE_AND:
1051                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1052                                     icol, 0);
1053                 if (h || opt->columnnum) {
1054                         /*
1055                          * Don't short-circuit AND when given --column, since a
1056                          * NOT earlier in the tree may turn this into an OR. In
1057                          * this case, see the below comment.
1058                          */
1059                         h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1060                                              ctx, col, icol, 0);
1061                 }
1062                 break;
1063         case GREP_NODE_OR:
1064                 if (!(collect_hits || opt->columnnum)) {
1065                         /*
1066                          * Don't short-circuit OR when given --column (or
1067                          * collecting hits) to ensure we don't skip a later
1068                          * child that would produce an earlier match.
1069                          */
1070                         return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1071                                                 ctx, col, icol, 0) ||
1072                                 match_expr_eval(opt, x->u.binary.right, bol,
1073                                                 eol, ctx, col, icol, 0));
1074                 }
1075                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1076                                     icol, 0);
1077                 if (collect_hits)
1078                         x->u.binary.left->hit |= h;
1079                 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1080                                      icol, collect_hits);
1081                 break;
1082         default:
1083                 die("Unexpected node type (internal error) %d", x->node);
1084         }
1085         if (collect_hits)
1086                 x->hit |= h;
1087         return h;
1088 }
1089
1090 static int match_expr(struct grep_opt *opt,
1091                       const char *bol, const char *eol,
1092                       enum grep_context ctx, ssize_t *col,
1093                       ssize_t *icol, int collect_hits)
1094 {
1095         struct grep_expr *x = opt->pattern_expression;
1096         return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1097 }
1098
1099 static int match_line(struct grep_opt *opt,
1100                       const char *bol, const char *eol,
1101                       ssize_t *col, ssize_t *icol,
1102                       enum grep_context ctx, int collect_hits)
1103 {
1104         struct grep_pat *p;
1105         int hit = 0;
1106
1107         if (opt->pattern_expression)
1108                 return match_expr(opt, bol, eol, ctx, col, icol,
1109                                   collect_hits);
1110
1111         /* we do not call with collect_hits without being extended */
1112         for (p = opt->pattern_list; p; p = p->next) {
1113                 regmatch_t tmp;
1114                 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1115                         hit |= 1;
1116                         if (!opt->columnnum) {
1117                                 /*
1118                                  * Without --column, any single match on a line
1119                                  * is enough to know that it needs to be
1120                                  * printed. With --column, scan _all_ patterns
1121                                  * to find the earliest.
1122                                  */
1123                                 break;
1124                         }
1125                         if (*col < 0 || tmp.rm_so < *col)
1126                                 *col = tmp.rm_so;
1127                 }
1128         }
1129         return hit;
1130 }
1131
1132 static int match_next_pattern(struct grep_pat *p,
1133                               const char *bol, const char *eol,
1134                               enum grep_context ctx,
1135                               regmatch_t *pmatch, int eflags)
1136 {
1137         regmatch_t match;
1138
1139         if (!headerless_match_one_pattern(p, bol, eol, ctx, &match, eflags))
1140                 return 0;
1141         if (match.rm_so < 0 || match.rm_eo < 0)
1142                 return 0;
1143         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1144                 if (match.rm_so > pmatch->rm_so)
1145                         return 1;
1146                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1147                         return 1;
1148         }
1149         pmatch->rm_so = match.rm_so;
1150         pmatch->rm_eo = match.rm_eo;
1151         return 1;
1152 }
1153
1154 int grep_next_match(struct grep_opt *opt,
1155                     const char *bol, const char *eol,
1156                     enum grep_context ctx, regmatch_t *pmatch,
1157                     enum grep_header_field field, int eflags)
1158 {
1159         struct grep_pat *p;
1160         int hit = 0;
1161
1162         pmatch->rm_so = pmatch->rm_eo = -1;
1163         if (bol < eol) {
1164                 for (p = ((ctx == GREP_CONTEXT_HEAD)
1165                            ? opt->header_list : opt->pattern_list);
1166                           p; p = p->next) {
1167                         switch (p->token) {
1168                         case GREP_PATTERN_HEAD:
1169                                 if ((field != GREP_HEADER_FIELD_MAX) &&
1170                                     (p->field != field))
1171                                         continue;
1172                                 /* fall thru */
1173                         case GREP_PATTERN: /* atom */
1174                         case GREP_PATTERN_BODY:
1175                                 hit |= match_next_pattern(p, bol, eol, ctx,
1176                                                           pmatch, eflags);
1177                                 break;
1178                         default:
1179                                 break;
1180                         }
1181                 }
1182         }
1183         return hit;
1184 }
1185
1186 static void show_line_header(struct grep_opt *opt, const char *name,
1187                              unsigned lno, ssize_t cno, char sign)
1188 {
1189         if (opt->heading && opt->last_shown == 0) {
1190                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1191                 opt->output(opt, "\n", 1);
1192         }
1193         opt->last_shown = lno;
1194
1195         if (!opt->heading && opt->pathname) {
1196                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1197                 output_sep(opt, sign);
1198         }
1199         if (opt->linenum) {
1200                 char buf[32];
1201                 xsnprintf(buf, sizeof(buf), "%d", lno);
1202                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1203                 output_sep(opt, sign);
1204         }
1205         /*
1206          * Treat 'cno' as the 1-indexed offset from the start of a non-context
1207          * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1208          * being called with a context line.
1209          */
1210         if (opt->columnnum && cno) {
1211                 char buf[32];
1212                 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1213                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1214                 output_sep(opt, sign);
1215         }
1216 }
1217
1218 static void show_line(struct grep_opt *opt,
1219                       const char *bol, const char *eol,
1220                       const char *name, unsigned lno, ssize_t cno, char sign)
1221 {
1222         int rest = eol - bol;
1223         const char *match_color = NULL;
1224         const char *line_color = NULL;
1225
1226         if (opt->file_break && opt->last_shown == 0) {
1227                 if (opt->show_hunk_mark)
1228                         opt->output(opt, "\n", 1);
1229         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1230                 if (opt->last_shown == 0) {
1231                         if (opt->show_hunk_mark) {
1232                                 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1233                                 opt->output(opt, "\n", 1);
1234                         }
1235                 } else if (lno > opt->last_shown + 1) {
1236                         output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1237                         opt->output(opt, "\n", 1);
1238                 }
1239         }
1240         if (!opt->only_matching) {
1241                 /*
1242                  * In case the line we're being called with contains more than
1243                  * one match, leave printing each header to the loop below.
1244                  */
1245                 show_line_header(opt, name, lno, cno, sign);
1246         }
1247         if (opt->color || opt->only_matching) {
1248                 regmatch_t match;
1249                 enum grep_context ctx = GREP_CONTEXT_BODY;
1250                 int eflags = 0;
1251
1252                 if (opt->color) {
1253                         if (sign == ':')
1254                                 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1255                         else
1256                                 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1257                         if (sign == ':')
1258                                 line_color = opt->colors[GREP_COLOR_SELECTED];
1259                         else if (sign == '-')
1260                                 line_color = opt->colors[GREP_COLOR_CONTEXT];
1261                         else if (sign == '=')
1262                                 line_color = opt->colors[GREP_COLOR_FUNCTION];
1263                 }
1264                 while (grep_next_match(opt, bol, eol, ctx, &match,
1265                                        GREP_HEADER_FIELD_MAX, eflags)) {
1266                         if (match.rm_so == match.rm_eo)
1267                                 break;
1268
1269                         if (opt->only_matching)
1270                                 show_line_header(opt, name, lno, cno, sign);
1271                         else
1272                                 output_color(opt, bol, match.rm_so, line_color);
1273                         output_color(opt, bol + match.rm_so,
1274                                      match.rm_eo - match.rm_so, match_color);
1275                         if (opt->only_matching)
1276                                 opt->output(opt, "\n", 1);
1277                         bol += match.rm_eo;
1278                         cno += match.rm_eo;
1279                         rest -= match.rm_eo;
1280                         eflags = REG_NOTBOL;
1281                 }
1282         }
1283         if (!opt->only_matching) {
1284                 output_color(opt, bol, rest, line_color);
1285                 opt->output(opt, "\n", 1);
1286         }
1287 }
1288
1289 int grep_use_locks;
1290
1291 /*
1292  * This lock protects access to the gitattributes machinery, which is
1293  * not thread-safe.
1294  */
1295 pthread_mutex_t grep_attr_mutex;
1296
1297 static inline void grep_attr_lock(void)
1298 {
1299         if (grep_use_locks)
1300                 pthread_mutex_lock(&grep_attr_mutex);
1301 }
1302
1303 static inline void grep_attr_unlock(void)
1304 {
1305         if (grep_use_locks)
1306                 pthread_mutex_unlock(&grep_attr_mutex);
1307 }
1308
1309 static int match_funcname(struct grep_opt *opt, struct grep_source *gs,
1310                           const char *bol, const char *eol)
1311 {
1312         xdemitconf_t *xecfg = opt->priv;
1313         if (xecfg && !xecfg->find_func) {
1314                 grep_source_load_driver(gs, opt->repo->index);
1315                 if (gs->driver->funcname.pattern) {
1316                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1317                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1318                 } else {
1319                         xecfg = opt->priv = NULL;
1320                 }
1321         }
1322
1323         if (xecfg) {
1324                 char buf[1];
1325                 return xecfg->find_func(bol, eol - bol, buf, 1,
1326                                         xecfg->find_func_priv) >= 0;
1327         }
1328
1329         if (bol == eol)
1330                 return 0;
1331         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1332                 return 1;
1333         return 0;
1334 }
1335
1336 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1337                                const char *bol, unsigned lno)
1338 {
1339         while (bol > gs->buf) {
1340                 const char *eol = --bol;
1341
1342                 while (bol > gs->buf && bol[-1] != '\n')
1343                         bol--;
1344                 lno--;
1345
1346                 if (lno <= opt->last_shown)
1347                         break;
1348
1349                 if (match_funcname(opt, gs, bol, eol)) {
1350                         show_line(opt, bol, eol, gs->name, lno, 0, '=');
1351                         break;
1352                 }
1353         }
1354 }
1355
1356 static int is_empty_line(const char *bol, const char *eol);
1357
1358 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1359                              const char *bol, const char *end, unsigned lno)
1360 {
1361         unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1362         int funcname_needed = !!opt->funcname, comment_needed = 0;
1363
1364         if (opt->pre_context < lno)
1365                 from = lno - opt->pre_context;
1366         if (from <= opt->last_shown)
1367                 from = opt->last_shown + 1;
1368         orig_from = from;
1369         if (opt->funcbody) {
1370                 if (match_funcname(opt, gs, bol, end))
1371                         comment_needed = 1;
1372                 else
1373                         funcname_needed = 1;
1374                 from = opt->last_shown + 1;
1375         }
1376
1377         /* Rewind. */
1378         while (bol > gs->buf && cur > from) {
1379                 const char *next_bol = bol;
1380                 const char *eol = --bol;
1381
1382                 while (bol > gs->buf && bol[-1] != '\n')
1383                         bol--;
1384                 cur--;
1385                 if (comment_needed && (is_empty_line(bol, eol) ||
1386                                        match_funcname(opt, gs, bol, eol))) {
1387                         comment_needed = 0;
1388                         from = orig_from;
1389                         if (cur < from) {
1390                                 cur++;
1391                                 bol = next_bol;
1392                                 break;
1393                         }
1394                 }
1395                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1396                         funcname_lno = cur;
1397                         funcname_needed = 0;
1398                         if (opt->funcbody)
1399                                 comment_needed = 1;
1400                         else
1401                                 from = orig_from;
1402                 }
1403         }
1404
1405         /* We need to look even further back to find a function signature. */
1406         if (opt->funcname && funcname_needed)
1407                 show_funcname_line(opt, gs, bol, cur);
1408
1409         /* Back forward. */
1410         while (cur < lno) {
1411                 const char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1412
1413                 while (*eol != '\n')
1414                         eol++;
1415                 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1416                 bol = eol + 1;
1417                 cur++;
1418         }
1419 }
1420
1421 static int should_lookahead(struct grep_opt *opt)
1422 {
1423         struct grep_pat *p;
1424
1425         if (opt->pattern_expression)
1426                 return 0; /* punt for too complex stuff */
1427         if (opt->invert)
1428                 return 0;
1429         for (p = opt->pattern_list; p; p = p->next) {
1430                 if (p->token != GREP_PATTERN)
1431                         return 0; /* punt for "header only" and stuff */
1432         }
1433         return 1;
1434 }
1435
1436 static int look_ahead(struct grep_opt *opt,
1437                       unsigned long *left_p,
1438                       unsigned *lno_p,
1439                       const char **bol_p)
1440 {
1441         unsigned lno = *lno_p;
1442         const char *bol = *bol_p;
1443         struct grep_pat *p;
1444         const char *sp, *last_bol;
1445         regoff_t earliest = -1;
1446
1447         for (p = opt->pattern_list; p; p = p->next) {
1448                 int hit;
1449                 regmatch_t m;
1450
1451                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1452                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1453                         continue;
1454                 if (earliest < 0 || m.rm_so < earliest)
1455                         earliest = m.rm_so;
1456         }
1457
1458         if (earliest < 0) {
1459                 *bol_p = bol + *left_p;
1460                 *left_p = 0;
1461                 return 1;
1462         }
1463         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1464                 ; /* find the beginning of the line */
1465         last_bol = sp;
1466
1467         for (sp = bol; sp < last_bol; sp++) {
1468                 if (*sp == '\n')
1469                         lno++;
1470         }
1471         *left_p -= last_bol - bol;
1472         *bol_p = last_bol;
1473         *lno_p = lno;
1474         return 0;
1475 }
1476
1477 static int fill_textconv_grep(struct repository *r,
1478                               struct userdiff_driver *driver,
1479                               struct grep_source *gs)
1480 {
1481         struct diff_filespec *df;
1482         char *buf;
1483         size_t size;
1484
1485         if (!driver || !driver->textconv)
1486                 return grep_source_load(gs);
1487
1488         /*
1489          * The textconv interface is intimately tied to diff_filespecs, so we
1490          * have to pretend to be one. If we could unify the grep_source
1491          * and diff_filespec structs, this mess could just go away.
1492          */
1493         df = alloc_filespec(gs->path);
1494         switch (gs->type) {
1495         case GREP_SOURCE_OID:
1496                 fill_filespec(df, gs->identifier, 1, 0100644);
1497                 break;
1498         case GREP_SOURCE_FILE:
1499                 fill_filespec(df, null_oid(), 0, 0100644);
1500                 break;
1501         default:
1502                 BUG("attempt to textconv something without a path?");
1503         }
1504
1505         /*
1506          * fill_textconv is not remotely thread-safe; it modifies the global
1507          * diff tempfile structure, writes to the_repo's odb and might
1508          * internally call thread-unsafe functions such as the
1509          * prepare_packed_git() lazy-initializator. Because of the last two, we
1510          * must ensure mutual exclusion between this call and the object reading
1511          * API, thus we use obj_read_lock() here.
1512          *
1513          * TODO: allowing text conversion to run in parallel with object
1514          * reading operations might increase performance in the multithreaded
1515          * non-worktreee git-grep with --textconv.
1516          */
1517         obj_read_lock();
1518         size = fill_textconv(r, driver, df, &buf);
1519         obj_read_unlock();
1520         free_filespec(df);
1521
1522         /*
1523          * The normal fill_textconv usage by the diff machinery would just keep
1524          * the textconv'd buf separate from the diff_filespec. But much of the
1525          * grep code passes around a grep_source and assumes that its "buf"
1526          * pointer is the beginning of the thing we are searching. So let's
1527          * install our textconv'd version into the grep_source, taking care not
1528          * to leak any existing buffer.
1529          */
1530         grep_source_clear_data(gs);
1531         gs->buf = buf;
1532         gs->size = size;
1533
1534         return 0;
1535 }
1536
1537 static int is_empty_line(const char *bol, const char *eol)
1538 {
1539         while (bol < eol && isspace(*bol))
1540                 bol++;
1541         return bol == eol;
1542 }
1543
1544 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1545 {
1546         const char *bol;
1547         const char *peek_bol = NULL;
1548         unsigned long left;
1549         unsigned lno = 1;
1550         unsigned last_hit = 0;
1551         int binary_match_only = 0;
1552         unsigned count = 0;
1553         int try_lookahead = 0;
1554         int show_function = 0;
1555         struct userdiff_driver *textconv = NULL;
1556         enum grep_context ctx = GREP_CONTEXT_HEAD;
1557         xdemitconf_t xecfg;
1558
1559         if (!opt->status_only && gs->name == NULL)
1560                 BUG("grep call which could print a name requires "
1561                     "grep_source.name be non-NULL");
1562
1563         if (!opt->output)
1564                 opt->output = std_output;
1565
1566         if (opt->pre_context || opt->post_context || opt->file_break ||
1567             opt->funcbody) {
1568                 /* Show hunk marks, except for the first file. */
1569                 if (opt->last_shown)
1570                         opt->show_hunk_mark = 1;
1571                 /*
1572                  * If we're using threads then we can't easily identify
1573                  * the first file.  Always put hunk marks in that case
1574                  * and skip the very first one later in work_done().
1575                  */
1576                 if (opt->output != std_output)
1577                         opt->show_hunk_mark = 1;
1578         }
1579         opt->last_shown = 0;
1580
1581         if (opt->allow_textconv) {
1582                 grep_source_load_driver(gs, opt->repo->index);
1583                 /*
1584                  * We might set up the shared textconv cache data here, which
1585                  * is not thread-safe. Also, get_oid_with_context() and
1586                  * parse_object() might be internally called. As they are not
1587                  * currently thread-safe and might be racy with object reading,
1588                  * obj_read_lock() must be called.
1589                  */
1590                 grep_attr_lock();
1591                 obj_read_lock();
1592                 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1593                 obj_read_unlock();
1594                 grep_attr_unlock();
1595         }
1596
1597         /*
1598          * We know the result of a textconv is text, so we only have to care
1599          * about binary handling if we are not using it.
1600          */
1601         if (!textconv) {
1602                 switch (opt->binary) {
1603                 case GREP_BINARY_DEFAULT:
1604                         if (grep_source_is_binary(gs, opt->repo->index))
1605                                 binary_match_only = 1;
1606                         break;
1607                 case GREP_BINARY_NOMATCH:
1608                         if (grep_source_is_binary(gs, opt->repo->index))
1609                                 return 0; /* Assume unmatch */
1610                         break;
1611                 case GREP_BINARY_TEXT:
1612                         break;
1613                 default:
1614                         BUG("unknown binary handling mode");
1615                 }
1616         }
1617
1618         memset(&xecfg, 0, sizeof(xecfg));
1619         opt->priv = &xecfg;
1620
1621         try_lookahead = should_lookahead(opt);
1622
1623         if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1624                 return 0;
1625
1626         bol = gs->buf;
1627         left = gs->size;
1628         while (left) {
1629                 const char *eol;
1630                 int hit;
1631                 ssize_t cno;
1632                 ssize_t col = -1, icol = -1;
1633
1634                 /*
1635                  * look_ahead() skips quickly to the line that possibly
1636                  * has the next hit; don't call it if we need to do
1637                  * something more than just skipping the current line
1638                  * in response to an unmatch for the current line.  E.g.
1639                  * inside a post-context window, we will show the current
1640                  * line as a context around the previous hit when it
1641                  * doesn't hit.
1642                  */
1643                 if (try_lookahead
1644                     && !(last_hit
1645                          && (show_function ||
1646                              lno <= last_hit + opt->post_context))
1647                     && look_ahead(opt, &left, &lno, &bol))
1648                         break;
1649                 eol = end_of_line(bol, &left);
1650
1651                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1652                         ctx = GREP_CONTEXT_BODY;
1653
1654                 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1655
1656                 if (collect_hits)
1657                         goto next_line;
1658
1659                 /* "grep -v -e foo -e bla" should list lines
1660                  * that do not have either, so inversion should
1661                  * be done outside.
1662                  */
1663                 if (opt->invert)
1664                         hit = !hit;
1665                 if (opt->unmatch_name_only) {
1666                         if (hit)
1667                                 return 0;
1668                         goto next_line;
1669                 }
1670                 if (hit && (opt->max_count < 0 || count < opt->max_count)) {
1671                         count++;
1672                         if (opt->status_only)
1673                                 return 1;
1674                         if (opt->name_only) {
1675                                 show_name(opt, gs->name);
1676                                 return 1;
1677                         }
1678                         if (opt->count)
1679                                 goto next_line;
1680                         if (binary_match_only) {
1681                                 opt->output(opt, "Binary file ", 12);
1682                                 output_color(opt, gs->name, strlen(gs->name),
1683                                              opt->colors[GREP_COLOR_FILENAME]);
1684                                 opt->output(opt, " matches\n", 9);
1685                                 return 1;
1686                         }
1687                         /* Hit at this line.  If we haven't shown the
1688                          * pre-context lines, we would need to show them.
1689                          */
1690                         if (opt->pre_context || opt->funcbody)
1691                                 show_pre_context(opt, gs, bol, eol, lno);
1692                         else if (opt->funcname)
1693                                 show_funcname_line(opt, gs, bol, lno);
1694                         cno = opt->invert ? icol : col;
1695                         if (cno < 0) {
1696                                 /*
1697                                  * A negative cno indicates that there was no
1698                                  * match on the line. We are thus inverted and
1699                                  * being asked to show all lines that _don't_
1700                                  * match a given expression. Therefore, set cno
1701                                  * to 0 to suggest the whole line matches.
1702                                  */
1703                                 cno = 0;
1704                         }
1705                         show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1706                         last_hit = lno;
1707                         if (opt->funcbody)
1708                                 show_function = 1;
1709                         goto next_line;
1710                 }
1711                 if (show_function && (!peek_bol || peek_bol < bol)) {
1712                         unsigned long peek_left = left;
1713                         const char *peek_eol = eol;
1714
1715                         /*
1716                          * Trailing empty lines are not interesting.
1717                          * Peek past them to see if they belong to the
1718                          * body of the current function.
1719                          */
1720                         peek_bol = bol;
1721                         while (is_empty_line(peek_bol, peek_eol)) {
1722                                 peek_bol = peek_eol + 1;
1723                                 peek_eol = end_of_line(peek_bol, &peek_left);
1724                         }
1725
1726                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1727                                 show_function = 0;
1728                 }
1729                 if (show_function ||
1730                     (last_hit && lno <= last_hit + opt->post_context)) {
1731                         /* If the last hit is within the post context,
1732                          * we need to show this line.
1733                          */
1734                         show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1735                 }
1736
1737         next_line:
1738                 bol = eol + 1;
1739                 if (!left)
1740                         break;
1741                 left--;
1742                 lno++;
1743         }
1744
1745         if (collect_hits)
1746                 return 0;
1747
1748         if (opt->status_only)
1749                 return opt->unmatch_name_only;
1750         if (opt->unmatch_name_only) {
1751                 /* We did not see any hit, so we want to show this */
1752                 show_name(opt, gs->name);
1753                 return 1;
1754         }
1755
1756         xdiff_clear_find_func(&xecfg);
1757         opt->priv = NULL;
1758
1759         /* NEEDSWORK:
1760          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1761          * which feels mostly useless but sometimes useful.  Maybe
1762          * make it another option?  For now suppress them.
1763          */
1764         if (opt->count && count) {
1765                 char buf[32];
1766                 if (opt->pathname) {
1767                         output_color(opt, gs->name, strlen(gs->name),
1768                                      opt->colors[GREP_COLOR_FILENAME]);
1769                         output_sep(opt, ':');
1770                 }
1771                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1772                 opt->output(opt, buf, strlen(buf));
1773                 return 1;
1774         }
1775         return !!last_hit;
1776 }
1777
1778 static void clr_hit_marker(struct grep_expr *x)
1779 {
1780         /* All-hit markers are meaningful only at the very top level
1781          * OR node.
1782          */
1783         while (1) {
1784                 x->hit = 0;
1785                 if (x->node != GREP_NODE_OR)
1786                         return;
1787                 x->u.binary.left->hit = 0;
1788                 x = x->u.binary.right;
1789         }
1790 }
1791
1792 static int chk_hit_marker(struct grep_expr *x)
1793 {
1794         /* Top level nodes have hit markers.  See if they all are hits */
1795         while (1) {
1796                 if (x->node != GREP_NODE_OR)
1797                         return x->hit;
1798                 if (!x->u.binary.left->hit)
1799                         return 0;
1800                 x = x->u.binary.right;
1801         }
1802 }
1803
1804 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1805 {
1806         /*
1807          * we do not have to do the two-pass grep when we do not check
1808          * buffer-wide "all-match".
1809          */
1810         if (!opt->all_match && !opt->no_body_match)
1811                 return grep_source_1(opt, gs, 0);
1812
1813         /* Otherwise the toplevel "or" terms hit a bit differently.
1814          * We first clear hit markers from them.
1815          */
1816         clr_hit_marker(opt->pattern_expression);
1817         opt->body_hit = 0;
1818         grep_source_1(opt, gs, 1);
1819
1820         if (opt->all_match && !chk_hit_marker(opt->pattern_expression))
1821                 return 0;
1822         if (opt->no_body_match && opt->body_hit)
1823                 return 0;
1824
1825         return grep_source_1(opt, gs, 0);
1826 }
1827
1828 static void grep_source_init_buf(struct grep_source *gs,
1829                                  const char *buf,
1830                                  unsigned long size)
1831 {
1832         gs->type = GREP_SOURCE_BUF;
1833         gs->name = NULL;
1834         gs->path = NULL;
1835         gs->buf = buf;
1836         gs->size = size;
1837         gs->driver = NULL;
1838         gs->identifier = NULL;
1839 }
1840
1841 int grep_buffer(struct grep_opt *opt, const char *buf, unsigned long size)
1842 {
1843         struct grep_source gs;
1844         int r;
1845
1846         grep_source_init_buf(&gs, buf, size);
1847
1848         r = grep_source(opt, &gs);
1849
1850         grep_source_clear(&gs);
1851         return r;
1852 }
1853
1854 void grep_source_init_file(struct grep_source *gs, const char *name,
1855                            const char *path)
1856 {
1857         gs->type = GREP_SOURCE_FILE;
1858         gs->name = xstrdup_or_null(name);
1859         gs->path = xstrdup_or_null(path);
1860         gs->buf = NULL;
1861         gs->size = 0;
1862         gs->driver = NULL;
1863         gs->identifier = xstrdup(path);
1864 }
1865
1866 void grep_source_init_oid(struct grep_source *gs, const char *name,
1867                           const char *path, const struct object_id *oid,
1868                           struct repository *repo)
1869 {
1870         gs->type = GREP_SOURCE_OID;
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 = oiddup(oid);
1877         gs->repo = repo;
1878 }
1879
1880 void grep_source_clear(struct grep_source *gs)
1881 {
1882         FREE_AND_NULL(gs->name);
1883         FREE_AND_NULL(gs->path);
1884         FREE_AND_NULL(gs->identifier);
1885         grep_source_clear_data(gs);
1886 }
1887
1888 void grep_source_clear_data(struct grep_source *gs)
1889 {
1890         switch (gs->type) {
1891         case GREP_SOURCE_FILE:
1892         case GREP_SOURCE_OID:
1893                 /* these types own the buffer */
1894                 free((char *)gs->buf);
1895                 gs->buf = NULL;
1896                 gs->size = 0;
1897                 break;
1898         case GREP_SOURCE_BUF:
1899                 /* leave user-provided buf intact */
1900                 break;
1901         }
1902 }
1903
1904 static int grep_source_load_oid(struct grep_source *gs)
1905 {
1906         enum object_type type;
1907
1908         gs->buf = repo_read_object_file(gs->repo, gs->identifier, &type,
1909                                         &gs->size);
1910         if (!gs->buf)
1911                 return error(_("'%s': unable to read %s"),
1912                              gs->name,
1913                              oid_to_hex(gs->identifier));
1914         return 0;
1915 }
1916
1917 static int grep_source_load_file(struct grep_source *gs)
1918 {
1919         const char *filename = gs->identifier;
1920         struct stat st;
1921         char *data;
1922         size_t size;
1923         int i;
1924
1925         if (lstat(filename, &st) < 0) {
1926         err_ret:
1927                 if (errno != ENOENT)
1928                         error_errno(_("failed to stat '%s'"), filename);
1929                 return -1;
1930         }
1931         if (!S_ISREG(st.st_mode))
1932                 return -1;
1933         size = xsize_t(st.st_size);
1934         i = open(filename, O_RDONLY);
1935         if (i < 0)
1936                 goto err_ret;
1937         data = xmallocz(size);
1938         if (st.st_size != read_in_full(i, data, size)) {
1939                 error_errno(_("'%s': short read"), filename);
1940                 close(i);
1941                 free(data);
1942                 return -1;
1943         }
1944         close(i);
1945
1946         gs->buf = data;
1947         gs->size = size;
1948         return 0;
1949 }
1950
1951 static int grep_source_load(struct grep_source *gs)
1952 {
1953         if (gs->buf)
1954                 return 0;
1955
1956         switch (gs->type) {
1957         case GREP_SOURCE_FILE:
1958                 return grep_source_load_file(gs);
1959         case GREP_SOURCE_OID:
1960                 return grep_source_load_oid(gs);
1961         case GREP_SOURCE_BUF:
1962                 return gs->buf ? 0 : -1;
1963         }
1964         BUG("invalid grep_source type to load");
1965 }
1966
1967 void grep_source_load_driver(struct grep_source *gs,
1968                              struct index_state *istate)
1969 {
1970         if (gs->driver)
1971                 return;
1972
1973         grep_attr_lock();
1974         if (gs->path)
1975                 gs->driver = userdiff_find_by_path(istate, gs->path);
1976         if (!gs->driver)
1977                 gs->driver = userdiff_find_by_name("default");
1978         grep_attr_unlock();
1979 }
1980
1981 static int grep_source_is_binary(struct grep_source *gs,
1982                                  struct index_state *istate)
1983 {
1984         grep_source_load_driver(gs, istate);
1985         if (gs->driver->binary != -1)
1986                 return gs->driver->binary;
1987
1988         if (!grep_source_load(gs))
1989                 return buffer_is_binary(gs->buf, gs->size);
1990
1991         return 0;
1992 }