grep.c

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