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
 712         for (p = opt->pattern_list; p; p = p->next) {
 713                 switch (p->token) {
 714                 case GREP_PATTERN: /* atom */
 715                 case GREP_PATTERN_HEAD:
 716                 case GREP_PATTERN_BODY:
 717                         compile_regexp(p, opt);
 718                         break;
 719                 default:
 720                         opt->extended = 1;
 721                         break;
 722                 }
 723         }
 724
 725         if (opt->all_match || opt->no_body_match || header_expr)
 726                 opt->extended = 1;
 727         else if (!opt->extended)
 728                 return;
 729
 730         p = opt->pattern_list;
 731         if (p)
 732                 opt->pattern_expression = compile_pattern_expr(&p);
 733         if (p)
 734                 die("incomplete pattern expression: %s", p->pattern);
 735
 736         if (opt->no_body_match && opt->pattern_expression)
 737                 opt->pattern_expression = grep_not_expr(opt->pattern_expression);
 738
 739         if (!header_expr)
 740                 return;
 741
 742         if (!opt->pattern_expression)
 743                 opt->pattern_expression = header_expr;
 744         else if (opt->all_match)
 745                 opt->pattern_expression = grep_splice_or(header_expr,
 746                                                          opt->pattern_expression);
 747         else
 748                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 749                                                        header_expr);
 750         opt->all_match = 1;
 751 }
 752
 753 static void free_pattern_expr(struct grep_expr *x)
 754 {
 755         switch (x->node) {
 756         case GREP_NODE_TRUE:
 757         case GREP_NODE_ATOM:
 758                 break;
 759         case GREP_NODE_NOT:
 760                 free_pattern_expr(x->u.unary);
 761                 break;
 762         case GREP_NODE_AND:
 763         case GREP_NODE_OR:
 764                 free_pattern_expr(x->u.binary.left);
 765                 free_pattern_expr(x->u.binary.right);
 766                 break;
 767         }
 768         free(x);
 769 }
 770
 771 void free_grep_patterns(struct grep_opt *opt)
 772 {
 773         struct grep_pat *p, *n;
 774
 775         for (p = opt->pattern_list; p; p = n) {
 776                 n = p->next;
 777                 switch (p->token) {
 778                 case GREP_PATTERN: /* atom */
 779                 case GREP_PATTERN_HEAD:
 780                 case GREP_PATTERN_BODY:
 781                         if (p->pcre2_pattern)
 782                                 free_pcre2_pattern(p);
 783                         else
 784                                 regfree(&p->regexp);
 785                         free(p->pattern);
 786                         break;
 787                 default:
 788                         break;
 789                 }
 790                 free(p);
 791         }
 792
 793         if (!opt->extended)
 794                 return;
 795         free_pattern_expr(opt->pattern_expression);
 796 }
 797
 798 static const char *end_of_line(const char *cp, unsigned long *left)
 799 {
 800         unsigned long l = *left;
 801         while (l && *cp != '\n') {
 802                 l--;
 803                 cp++;
 804         }
 805         *left = l;
 806         return cp;
 807 }
 808
 809 static int word_char(char ch)
 810 {
 811         return isalnum(ch) || ch == '_';
 812 }
 813
 814 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 815                          const char *color)
 816 {
 817         if (want_color(opt->color) && color && color[0]) {
 818                 opt->output(opt, color, strlen(color));
 819                 opt->output(opt, data, size);
 820                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 821         } else
 822                 opt->output(opt, data, size);
 823 }
 824
 825 static void output_sep(struct grep_opt *opt, char sign)
 826 {
 827         if (opt->null_following_name)
 828                 opt->output(opt, "\0", 1);
 829         else
 830                 output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
 831 }
 832
 833 static void show_name(struct grep_opt *opt, const char *name)
 834 {
 835         output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
 836         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 837 }
 838
 839 static int patmatch(struct grep_pat *p,
 840                     const char *line, const char *eol,
 841                     regmatch_t *match, int eflags)
 842 {
 843         int hit;
 844
 845         if (p->pcre2_pattern)
 846                 hit = !pcre2match(p, line, eol, match, eflags);
 847         else
 848                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
 849                                    eflags);
 850
 851         return hit;
 852 }
 853
 854 static void strip_timestamp(const char *bol, const char **eol_p)
 855 {
 856         const char *eol = *eol_p;
 857
 858         while (bol < --eol) {
 859                 if (*eol != '>')
 860                         continue;
 861                 *eol_p = ++eol;
 862                 break;
 863         }
 864 }
 865
 866 static struct {
 867         const char *field;
 868         size_t len;
 869 } header_field[] = {
 870         { "author ", 7 },
 871         { "committer ", 10 },
 872         { "reflog ", 7 },
 873 };
 874
 875 static int headerless_match_one_pattern(struct grep_pat *p,
 876                                         const char *bol, const char *eol,
 877                                         enum grep_context ctx,
 878                                         regmatch_t *pmatch, int eflags)
 879 {
 880         int hit = 0;
 881         const char *start = bol;
 882
 883         if ((p->token != GREP_PATTERN) &&
 884             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 885                 return 0;
 886
 887  again:
 888         hit = patmatch(p, bol, eol, pmatch, eflags);
 889
 890         if (hit && p->word_regexp) {
 891                 if ((pmatch[0].rm_so < 0) ||
 892                     (eol - bol) < pmatch[0].rm_so ||
 893                     (pmatch[0].rm_eo < 0) ||
 894                     (eol - bol) < pmatch[0].rm_eo)
 895                         die("regexp returned nonsense");
 896
 897                 /* Match beginning must be either beginning of the
 898                  * line, or at word boundary (i.e. the last char must
 899                  * not be a word char).  Similarly, match end must be
 900                  * either end of the line, or at word boundary
 901                  * (i.e. the next char must not be a word char).
 902                  */
 903                 if ( ((pmatch[0].rm_so == 0) ||
 904                       !word_char(bol[pmatch[0].rm_so-1])) &&
 905                      ((pmatch[0].rm_eo == (eol-bol)) ||
 906                       !word_char(bol[pmatch[0].rm_eo])) )
 907                         ;
 908                 else
 909                         hit = 0;
 910
 911                 /* Words consist of at least one character. */
 912                 if (pmatch->rm_so == pmatch->rm_eo)
 913                         hit = 0;
 914
 915                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 916                         /* There could be more than one match on the
 917                          * line, and the first match might not be
 918                          * strict word match.  But later ones could be!
 919                          * Forward to the next possible start, i.e. the
 920                          * next position following a non-word char.
 921                          */
 922                         bol = pmatch[0].rm_so + bol + 1;
 923                         while (word_char(bol[-1]) && bol < eol)
 924                                 bol++;
 925                         eflags |= REG_NOTBOL;
 926                         if (bol < eol)
 927                                 goto again;
 928                 }
 929         }
 930         if (hit) {
 931                 pmatch[0].rm_so += bol - start;
 932                 pmatch[0].rm_eo += bol - start;
 933         }
 934         return hit;
 935 }
 936
 937 static int match_one_pattern(struct grep_pat *p,
 938                              const char *bol, const char *eol,
 939                              enum grep_context ctx, regmatch_t *pmatch,
 940                              int eflags)
 941 {
 942         const char *field;
 943         size_t len;
 944
 945         if (p->token == GREP_PATTERN_HEAD) {
 946                 assert(p->field < ARRAY_SIZE(header_field));
 947                 field = header_field[p->field].field;
 948                 len = header_field[p->field].len;
 949                 if (strncmp(bol, field, len))
 950                         return 0;
 951                 bol += len;
 952
 953                 switch (p->field) {
 954                 case GREP_HEADER_AUTHOR:
 955                 case GREP_HEADER_COMMITTER:
 956                         strip_timestamp(bol, &eol);
 957                         break;
 958                 default:
 959                         break;
 960                 }
 961         }
 962
 963         return headerless_match_one_pattern(p, bol, eol, ctx, pmatch, eflags);
 964 }
 965
 966
 967 static int match_expr_eval(struct grep_opt *opt, struct grep_expr *x,
 968                            const char *bol, const char *eol,
 969                            enum grep_context ctx, ssize_t *col,
 970                            ssize_t *icol, int collect_hits)
 971 {
 972         int h = 0;
 973
 974         if (!x)
 975                 die("Not a valid grep expression");
 976         switch (x->node) {
 977         case GREP_NODE_TRUE:
 978                 h = 1;
 979                 break;
 980         case GREP_NODE_ATOM:
 981                 {
 982                         regmatch_t tmp;
 983                         h = match_one_pattern(x->u.atom, bol, eol, ctx,
 984                                               &tmp, 0);
 985                         if (h && (*col < 0 || tmp.rm_so < *col))
 986                                 *col = tmp.rm_so;
 987                 }
 988                 if (x->u.atom->token == GREP_PATTERN_BODY)
 989                         opt->body_hit |= h;
 990                 break;
 991         case GREP_NODE_NOT:
 992                 /*
 993                  * Upon visiting a GREP_NODE_NOT, col and icol become swapped.
 994                  */
 995                 h = !match_expr_eval(opt, x->u.unary, bol, eol, ctx, icol, col,
 996                                      0);
 997                 break;
 998         case GREP_NODE_AND:
 999                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1000                                     icol, 0);
1001                 if (h || opt->columnnum) {
1002                         /*
1003                          * Don't short-circuit AND when given --column, since a
1004                          * NOT earlier in the tree may turn this into an OR. In
1005                          * this case, see the below comment.
1006                          */
1007                         h &= match_expr_eval(opt, x->u.binary.right, bol, eol,
1008                                              ctx, col, icol, 0);
1009                 }
1010                 break;
1011         case GREP_NODE_OR:
1012                 if (!(collect_hits || opt->columnnum)) {
1013                         /*
1014                          * Don't short-circuit OR when given --column (or
1015                          * collecting hits) to ensure we don't skip a later
1016                          * child that would produce an earlier match.
1017                          */
1018                         return (match_expr_eval(opt, x->u.binary.left, bol, eol,
1019                                                 ctx, col, icol, 0) ||
1020                                 match_expr_eval(opt, x->u.binary.right, bol,
1021                                                 eol, ctx, col, icol, 0));
1022                 }
1023                 h = match_expr_eval(opt, x->u.binary.left, bol, eol, ctx, col,
1024                                     icol, 0);
1025                 if (collect_hits)
1026                         x->u.binary.left->hit |= h;
1027                 h |= match_expr_eval(opt, x->u.binary.right, bol, eol, ctx, col,
1028                                      icol, collect_hits);
1029                 break;
1030         default:
1031                 die("Unexpected node type (internal error) %d", x->node);
1032         }
1033         if (collect_hits)
1034                 x->hit |= h;
1035         return h;
1036 }
1037
1038 static int match_expr(struct grep_opt *opt,
1039                       const char *bol, const char *eol,
1040                       enum grep_context ctx, ssize_t *col,
1041                       ssize_t *icol, int collect_hits)
1042 {
1043         struct grep_expr *x = opt->pattern_expression;
1044         return match_expr_eval(opt, x, bol, eol, ctx, col, icol, collect_hits);
1045 }
1046
1047 static int match_line(struct grep_opt *opt,
1048                       const char *bol, const char *eol,
1049                       ssize_t *col, ssize_t *icol,
1050                       enum grep_context ctx, int collect_hits)
1051 {
1052         struct grep_pat *p;
1053         int hit = 0;
1054
1055         if (opt->extended)
1056                 return match_expr(opt, bol, eol, ctx, col, icol,
1057                                   collect_hits);
1058
1059         /* we do not call with collect_hits without being extended */
1060         for (p = opt->pattern_list; p; p = p->next) {
1061                 regmatch_t tmp;
1062                 if (match_one_pattern(p, bol, eol, ctx, &tmp, 0)) {
1063                         hit |= 1;
1064                         if (!opt->columnnum) {
1065                                 /*
1066                                  * Without --column, any single match on a line
1067                                  * is enough to know that it needs to be
1068                                  * printed. With --column, scan _all_ patterns
1069                                  * to find the earliest.
1070                                  */
1071                                 break;
1072                         }
1073                         if (*col < 0 || tmp.rm_so < *col)
1074                                 *col = tmp.rm_so;
1075                 }
1076         }
1077         return hit;
1078 }
1079
1080 static int match_next_pattern(struct grep_pat *p,
1081                               const char *bol, const char *eol,
1082                               enum grep_context ctx,
1083                               regmatch_t *pmatch, int eflags)
1084 {
1085         regmatch_t match;
1086
1087         if (!headerless_match_one_pattern(p, bol, eol, ctx, &match, eflags))
1088                 return 0;
1089         if (match.rm_so < 0 || match.rm_eo < 0)
1090                 return 0;
1091         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1092                 if (match.rm_so > pmatch->rm_so)
1093                         return 1;
1094                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1095                         return 1;
1096         }
1097         pmatch->rm_so = match.rm_so;
1098         pmatch->rm_eo = match.rm_eo;
1099         return 1;
1100 }
1101
1102 int grep_next_match(struct grep_opt *opt,
1103                     const char *bol, const char *eol,
1104                     enum grep_context ctx, regmatch_t *pmatch,
1105                     enum grep_header_field field, int eflags)
1106 {
1107         struct grep_pat *p;
1108         int hit = 0;
1109
1110         pmatch->rm_so = pmatch->rm_eo = -1;
1111         if (bol < eol) {
1112                 for (p = ((ctx == GREP_CONTEXT_HEAD)
1113                            ? opt->header_list : opt->pattern_list);
1114                           p; p = p->next) {
1115                         switch (p->token) {
1116                         case GREP_PATTERN_HEAD:
1117                                 if ((field != GREP_HEADER_FIELD_MAX) &&
1118                                     (p->field != field))
1119                                         continue;
1120                                 /* fall thru */
1121                         case GREP_PATTERN: /* atom */
1122                         case GREP_PATTERN_BODY:
1123                                 hit |= match_next_pattern(p, bol, eol, ctx,
1124                                                           pmatch, eflags);
1125                                 break;
1126                         default:
1127                                 break;
1128                         }
1129                 }
1130         }
1131         return hit;
1132 }
1133
1134 static void show_line_header(struct grep_opt *opt, const char *name,
1135                              unsigned lno, ssize_t cno, char sign)
1136 {
1137         if (opt->heading && opt->last_shown == 0) {
1138                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1139                 opt->output(opt, "\n", 1);
1140         }
1141         opt->last_shown = lno;
1142
1143         if (!opt->heading && opt->pathname) {
1144                 output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
1145                 output_sep(opt, sign);
1146         }
1147         if (opt->linenum) {
1148                 char buf[32];
1149                 xsnprintf(buf, sizeof(buf), "%d", lno);
1150                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
1151                 output_sep(opt, sign);
1152         }
1153         /*
1154          * Treat 'cno' as the 1-indexed offset from the start of a non-context
1155          * line to its first match. Otherwise, 'cno' is 0 indicating that we are
1156          * being called with a context line.
1157          */
1158         if (opt->columnnum && cno) {
1159                 char buf[32];
1160                 xsnprintf(buf, sizeof(buf), "%"PRIuMAX, (uintmax_t)cno);
1161                 output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_COLUMNNO]);
1162                 output_sep(opt, sign);
1163         }
1164 }
1165
1166 static void show_line(struct grep_opt *opt,
1167                       const char *bol, const char *eol,
1168                       const char *name, unsigned lno, ssize_t cno, char sign)
1169 {
1170         int rest = eol - bol;
1171         const char *match_color = NULL;
1172         const char *line_color = NULL;
1173
1174         if (opt->file_break && opt->last_shown == 0) {
1175                 if (opt->show_hunk_mark)
1176                         opt->output(opt, "\n", 1);
1177         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1178                 if (opt->last_shown == 0) {
1179                         if (opt->show_hunk_mark) {
1180                                 output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1181                                 opt->output(opt, "\n", 1);
1182                         }
1183                 } else if (lno > opt->last_shown + 1) {
1184                         output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
1185                         opt->output(opt, "\n", 1);
1186                 }
1187         }
1188         if (!opt->only_matching) {
1189                 /*
1190                  * In case the line we're being called with contains more than
1191                  * one match, leave printing each header to the loop below.
1192                  */
1193                 show_line_header(opt, name, lno, cno, sign);
1194         }
1195         if (opt->color || opt->only_matching) {
1196                 regmatch_t match;
1197                 enum grep_context ctx = GREP_CONTEXT_BODY;
1198                 int eflags = 0;
1199
1200                 if (opt->color) {
1201                         if (sign == ':')
1202                                 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
1203                         else
1204                                 match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
1205                         if (sign == ':')
1206                                 line_color = opt->colors[GREP_COLOR_SELECTED];
1207                         else if (sign == '-')
1208                                 line_color = opt->colors[GREP_COLOR_CONTEXT];
1209                         else if (sign == '=')
1210                                 line_color = opt->colors[GREP_COLOR_FUNCTION];
1211                 }
1212                 while (grep_next_match(opt, bol, eol, ctx, &match,
1213                                        GREP_HEADER_FIELD_MAX, eflags)) {
1214                         if (match.rm_so == match.rm_eo)
1215                                 break;
1216
1217                         if (opt->only_matching)
1218                                 show_line_header(opt, name, lno, cno, sign);
1219                         else
1220                                 output_color(opt, bol, match.rm_so, line_color);
1221                         output_color(opt, bol + match.rm_so,
1222                                      match.rm_eo - match.rm_so, match_color);
1223                         if (opt->only_matching)
1224                                 opt->output(opt, "\n", 1);
1225                         bol += match.rm_eo;
1226                         cno += match.rm_eo;
1227                         rest -= match.rm_eo;
1228                         eflags = REG_NOTBOL;
1229                 }
1230         }
1231         if (!opt->only_matching) {
1232                 output_color(opt, bol, rest, line_color);
1233                 opt->output(opt, "\n", 1);
1234         }
1235 }
1236
1237 int grep_use_locks;
1238
1239 /*
1240  * This lock protects access to the gitattributes machinery, which is
1241  * not thread-safe.
1242  */
1243 pthread_mutex_t grep_attr_mutex;
1244
1245 static inline void grep_attr_lock(void)
1246 {
1247         if (grep_use_locks)
1248                 pthread_mutex_lock(&grep_attr_mutex);
1249 }
1250
1251 static inline void grep_attr_unlock(void)
1252 {
1253         if (grep_use_locks)
1254                 pthread_mutex_unlock(&grep_attr_mutex);
1255 }
1256
1257 static int match_funcname(struct grep_opt *opt, struct grep_source *gs,
1258                           const char *bol, const char *eol)
1259 {
1260         xdemitconf_t *xecfg = opt->priv;
1261         if (xecfg && !xecfg->find_func) {
1262                 grep_source_load_driver(gs, opt->repo->index);
1263                 if (gs->driver->funcname.pattern) {
1264                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1265                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1266                 } else {
1267                         xecfg = opt->priv = NULL;
1268                 }
1269         }
1270
1271         if (xecfg) {
1272                 char buf[1];
1273                 return xecfg->find_func(bol, eol - bol, buf, 1,
1274                                         xecfg->find_func_priv) >= 0;
1275         }
1276
1277         if (bol == eol)
1278                 return 0;
1279         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1280                 return 1;
1281         return 0;
1282 }
1283
1284 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1285                                const char *bol, unsigned lno)
1286 {
1287         while (bol > gs->buf) {
1288                 const char *eol = --bol;
1289
1290                 while (bol > gs->buf && bol[-1] != '\n')
1291                         bol--;
1292                 lno--;
1293
1294                 if (lno <= opt->last_shown)
1295                         break;
1296
1297                 if (match_funcname(opt, gs, bol, eol)) {
1298                         show_line(opt, bol, eol, gs->name, lno, 0, '=');
1299                         break;
1300                 }
1301         }
1302 }
1303
1304 static int is_empty_line(const char *bol, const char *eol);
1305
1306 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1307                              const char *bol, const char *end, unsigned lno)
1308 {
1309         unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1310         int funcname_needed = !!opt->funcname, comment_needed = 0;
1311
1312         if (opt->pre_context < lno)
1313                 from = lno - opt->pre_context;
1314         if (from <= opt->last_shown)
1315                 from = opt->last_shown + 1;
1316         orig_from = from;
1317         if (opt->funcbody) {
1318                 if (match_funcname(opt, gs, bol, end))
1319                         comment_needed = 1;
1320                 else
1321                         funcname_needed = 1;
1322                 from = opt->last_shown + 1;
1323         }
1324
1325         /* Rewind. */
1326         while (bol > gs->buf && cur > from) {
1327                 const char *next_bol = bol;
1328                 const char *eol = --bol;
1329
1330                 while (bol > gs->buf && bol[-1] != '\n')
1331                         bol--;
1332                 cur--;
1333                 if (comment_needed && (is_empty_line(bol, eol) ||
1334                                        match_funcname(opt, gs, bol, eol))) {
1335                         comment_needed = 0;
1336                         from = orig_from;
1337                         if (cur < from) {
1338                                 cur++;
1339                                 bol = next_bol;
1340                                 break;
1341                         }
1342                 }
1343                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1344                         funcname_lno = cur;
1345                         funcname_needed = 0;
1346                         if (opt->funcbody)
1347                                 comment_needed = 1;
1348                         else
1349                                 from = orig_from;
1350                 }
1351         }
1352
1353         /* We need to look even further back to find a function signature. */
1354         if (opt->funcname && funcname_needed)
1355                 show_funcname_line(opt, gs, bol, cur);
1356
1357         /* Back forward. */
1358         while (cur < lno) {
1359                 const char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1360
1361                 while (*eol != '\n')
1362                         eol++;
1363                 show_line(opt, bol, eol, gs->name, cur, 0, sign);
1364                 bol = eol + 1;
1365                 cur++;
1366         }
1367 }
1368
1369 static int should_lookahead(struct grep_opt *opt)
1370 {
1371         struct grep_pat *p;
1372
1373         if (opt->extended)
1374                 return 0; /* punt for too complex stuff */
1375         if (opt->invert)
1376                 return 0;
1377         for (p = opt->pattern_list; p; p = p->next) {
1378                 if (p->token != GREP_PATTERN)
1379                         return 0; /* punt for "header only" and stuff */
1380         }
1381         return 1;
1382 }
1383
1384 static int look_ahead(struct grep_opt *opt,
1385                       unsigned long *left_p,
1386                       unsigned *lno_p,
1387                       const char **bol_p)
1388 {
1389         unsigned lno = *lno_p;
1390         const char *bol = *bol_p;
1391         struct grep_pat *p;
1392         const char *sp, *last_bol;
1393         regoff_t earliest = -1;
1394
1395         for (p = opt->pattern_list; p; p = p->next) {
1396                 int hit;
1397                 regmatch_t m;
1398
1399                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1400                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1401                         continue;
1402                 if (earliest < 0 || m.rm_so < earliest)
1403                         earliest = m.rm_so;
1404         }
1405
1406         if (earliest < 0) {
1407                 *bol_p = bol + *left_p;
1408                 *left_p = 0;
1409                 return 1;
1410         }
1411         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1412                 ; /* find the beginning of the line */
1413         last_bol = sp;
1414
1415         for (sp = bol; sp < last_bol; sp++) {
1416                 if (*sp == '\n')
1417                         lno++;
1418         }
1419         *left_p -= last_bol - bol;
1420         *bol_p = last_bol;
1421         *lno_p = lno;
1422         return 0;
1423 }
1424
1425 static int fill_textconv_grep(struct repository *r,
1426                               struct userdiff_driver *driver,
1427                               struct grep_source *gs)
1428 {
1429         struct diff_filespec *df;
1430         char *buf;
1431         size_t size;
1432
1433         if (!driver || !driver->textconv)
1434                 return grep_source_load(gs);
1435
1436         /*
1437          * The textconv interface is intimately tied to diff_filespecs, so we
1438          * have to pretend to be one. If we could unify the grep_source
1439          * and diff_filespec structs, this mess could just go away.
1440          */
1441         df = alloc_filespec(gs->path);
1442         switch (gs->type) {
1443         case GREP_SOURCE_OID:
1444                 fill_filespec(df, gs->identifier, 1, 0100644);
1445                 break;
1446         case GREP_SOURCE_FILE:
1447                 fill_filespec(df, null_oid(), 0, 0100644);
1448                 break;
1449         default:
1450                 BUG("attempt to textconv something without a path?");
1451         }
1452
1453         /*
1454          * fill_textconv is not remotely thread-safe; it modifies the global
1455          * diff tempfile structure, writes to the_repo's odb and might
1456          * internally call thread-unsafe functions such as the
1457          * prepare_packed_git() lazy-initializator. Because of the last two, we
1458          * must ensure mutual exclusion between this call and the object reading
1459          * API, thus we use obj_read_lock() here.
1460          *
1461          * TODO: allowing text conversion to run in parallel with object
1462          * reading operations might increase performance in the multithreaded
1463          * non-worktreee git-grep with --textconv.
1464          */
1465         obj_read_lock();
1466         size = fill_textconv(r, driver, df, &buf);
1467         obj_read_unlock();
1468         free_filespec(df);
1469
1470         /*
1471          * The normal fill_textconv usage by the diff machinery would just keep
1472          * the textconv'd buf separate from the diff_filespec. But much of the
1473          * grep code passes around a grep_source and assumes that its "buf"
1474          * pointer is the beginning of the thing we are searching. So let's
1475          * install our textconv'd version into the grep_source, taking care not
1476          * to leak any existing buffer.
1477          */
1478         grep_source_clear_data(gs);
1479         gs->buf = buf;
1480         gs->size = size;
1481
1482         return 0;
1483 }
1484
1485 static int is_empty_line(const char *bol, const char *eol)
1486 {
1487         while (bol < eol && isspace(*bol))
1488                 bol++;
1489         return bol == eol;
1490 }
1491
1492 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1493 {
1494         const char *bol;
1495         const char *peek_bol = NULL;
1496         unsigned long left;
1497         unsigned lno = 1;
1498         unsigned last_hit = 0;
1499         int binary_match_only = 0;
1500         unsigned count = 0;
1501         int try_lookahead = 0;
1502         int show_function = 0;
1503         struct userdiff_driver *textconv = NULL;
1504         enum grep_context ctx = GREP_CONTEXT_HEAD;
1505         xdemitconf_t xecfg;
1506
1507         if (!opt->status_only && gs->name == NULL)
1508                 BUG("grep call which could print a name requires "
1509                     "grep_source.name be non-NULL");
1510
1511         if (!opt->output)
1512                 opt->output = std_output;
1513
1514         if (opt->pre_context || opt->post_context || opt->file_break ||
1515             opt->funcbody) {
1516                 /* Show hunk marks, except for the first file. */
1517                 if (opt->last_shown)
1518                         opt->show_hunk_mark = 1;
1519                 /*
1520                  * If we're using threads then we can't easily identify
1521                  * the first file.  Always put hunk marks in that case
1522                  * and skip the very first one later in work_done().
1523                  */
1524                 if (opt->output != std_output)
1525                         opt->show_hunk_mark = 1;
1526         }
1527         opt->last_shown = 0;
1528
1529         if (opt->allow_textconv) {
1530                 grep_source_load_driver(gs, opt->repo->index);
1531                 /*
1532                  * We might set up the shared textconv cache data here, which
1533                  * is not thread-safe. Also, get_oid_with_context() and
1534                  * parse_object() might be internally called. As they are not
1535                  * currently thread-safe and might be racy with object reading,
1536                  * obj_read_lock() must be called.
1537                  */
1538                 grep_attr_lock();
1539                 obj_read_lock();
1540                 textconv = userdiff_get_textconv(opt->repo, gs->driver);
1541                 obj_read_unlock();
1542                 grep_attr_unlock();
1543         }
1544
1545         /*
1546          * We know the result of a textconv is text, so we only have to care
1547          * about binary handling if we are not using it.
1548          */
1549         if (!textconv) {
1550                 switch (opt->binary) {
1551                 case GREP_BINARY_DEFAULT:
1552                         if (grep_source_is_binary(gs, opt->repo->index))
1553                                 binary_match_only = 1;
1554                         break;
1555                 case GREP_BINARY_NOMATCH:
1556                         if (grep_source_is_binary(gs, opt->repo->index))
1557                                 return 0; /* Assume unmatch */
1558                         break;
1559                 case GREP_BINARY_TEXT:
1560                         break;
1561                 default:
1562                         BUG("unknown binary handling mode");
1563                 }
1564         }
1565
1566         memset(&xecfg, 0, sizeof(xecfg));
1567         opt->priv = &xecfg;
1568
1569         try_lookahead = should_lookahead(opt);
1570
1571         if (fill_textconv_grep(opt->repo, textconv, gs) < 0)
1572                 return 0;
1573
1574         bol = gs->buf;
1575         left = gs->size;
1576         while (left) {
1577                 const char *eol;
1578                 int hit;
1579                 ssize_t cno;
1580                 ssize_t col = -1, icol = -1;
1581
1582                 /*
1583                  * look_ahead() skips quickly to the line that possibly
1584                  * has the next hit; don't call it if we need to do
1585                  * something more than just skipping the current line
1586                  * in response to an unmatch for the current line.  E.g.
1587                  * inside a post-context window, we will show the current
1588                  * line as a context around the previous hit when it
1589                  * doesn't hit.
1590                  */
1591                 if (try_lookahead
1592                     && !(last_hit
1593                          && (show_function ||
1594                              lno <= last_hit + opt->post_context))
1595                     && look_ahead(opt, &left, &lno, &bol))
1596                         break;
1597                 eol = end_of_line(bol, &left);
1598
1599                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1600                         ctx = GREP_CONTEXT_BODY;
1601
1602                 hit = match_line(opt, bol, eol, &col, &icol, ctx, collect_hits);
1603
1604                 if (collect_hits)
1605                         goto next_line;
1606
1607                 /* "grep -v -e foo -e bla" should list lines
1608                  * that do not have either, so inversion should
1609                  * be done outside.
1610                  */
1611                 if (opt->invert)
1612                         hit = !hit;
1613                 if (opt->unmatch_name_only) {
1614                         if (hit)
1615                                 return 0;
1616                         goto next_line;
1617                 }
1618                 if (hit) {
1619                         count++;
1620                         if (opt->status_only)
1621                                 return 1;
1622                         if (opt->name_only) {
1623                                 show_name(opt, gs->name);
1624                                 return 1;
1625                         }
1626                         if (opt->count)
1627                                 goto next_line;
1628                         if (binary_match_only) {
1629                                 opt->output(opt, "Binary file ", 12);
1630                                 output_color(opt, gs->name, strlen(gs->name),
1631                                              opt->colors[GREP_COLOR_FILENAME]);
1632                                 opt->output(opt, " matches\n", 9);
1633                                 return 1;
1634                         }
1635                         /* Hit at this line.  If we haven't shown the
1636                          * pre-context lines, we would need to show them.
1637                          */
1638                         if (opt->pre_context || opt->funcbody)
1639                                 show_pre_context(opt, gs, bol, eol, lno);
1640                         else if (opt->funcname)
1641                                 show_funcname_line(opt, gs, bol, lno);
1642                         cno = opt->invert ? icol : col;
1643                         if (cno < 0) {
1644                                 /*
1645                                  * A negative cno indicates that there was no
1646                                  * match on the line. We are thus inverted and
1647                                  * being asked to show all lines that _don't_
1648                                  * match a given expression. Therefore, set cno
1649                                  * to 0 to suggest the whole line matches.
1650                                  */
1651                                 cno = 0;
1652                         }
1653                         show_line(opt, bol, eol, gs->name, lno, cno + 1, ':');
1654                         last_hit = lno;
1655                         if (opt->funcbody)
1656                                 show_function = 1;
1657                         goto next_line;
1658                 }
1659                 if (show_function && (!peek_bol || peek_bol < bol)) {
1660                         unsigned long peek_left = left;
1661                         const char *peek_eol = eol;
1662
1663                         /*
1664                          * Trailing empty lines are not interesting.
1665                          * Peek past them to see if they belong to the
1666                          * body of the current function.
1667                          */
1668                         peek_bol = bol;
1669                         while (is_empty_line(peek_bol, peek_eol)) {
1670                                 peek_bol = peek_eol + 1;
1671                                 peek_eol = end_of_line(peek_bol, &peek_left);
1672                         }
1673
1674                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1675                                 show_function = 0;
1676                 }
1677                 if (show_function ||
1678                     (last_hit && lno <= last_hit + opt->post_context)) {
1679                         /* If the last hit is within the post context,
1680                          * we need to show this line.
1681                          */
1682                         show_line(opt, bol, eol, gs->name, lno, col + 1, '-');
1683                 }
1684
1685         next_line:
1686                 bol = eol + 1;
1687                 if (!left)
1688                         break;
1689                 left--;
1690                 lno++;
1691         }
1692
1693         if (collect_hits)
1694                 return 0;
1695
1696         if (opt->status_only)
1697                 return opt->unmatch_name_only;
1698         if (opt->unmatch_name_only) {
1699                 /* We did not see any hit, so we want to show this */
1700                 show_name(opt, gs->name);
1701                 return 1;
1702         }
1703
1704         xdiff_clear_find_func(&xecfg);
1705         opt->priv = NULL;
1706
1707         /* NEEDSWORK:
1708          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1709          * which feels mostly useless but sometimes useful.  Maybe
1710          * make it another option?  For now suppress them.
1711          */
1712         if (opt->count && count) {
1713                 char buf[32];
1714                 if (opt->pathname) {
1715                         output_color(opt, gs->name, strlen(gs->name),
1716                                      opt->colors[GREP_COLOR_FILENAME]);
1717                         output_sep(opt, ':');
1718                 }
1719                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1720                 opt->output(opt, buf, strlen(buf));
1721                 return 1;
1722         }
1723         return !!last_hit;
1724 }
1725
1726 static void clr_hit_marker(struct grep_expr *x)
1727 {
1728         /* All-hit markers are meaningful only at the very top level
1729          * OR node.
1730          */
1731         while (1) {
1732                 x->hit = 0;
1733                 if (x->node != GREP_NODE_OR)
1734                         return;
1735                 x->u.binary.left->hit = 0;
1736                 x = x->u.binary.right;
1737         }
1738 }
1739
1740 static int chk_hit_marker(struct grep_expr *x)
1741 {
1742         /* Top level nodes have hit markers.  See if they all are hits */
1743         while (1) {
1744                 if (x->node != GREP_NODE_OR)
1745                         return x->hit;
1746                 if (!x->u.binary.left->hit)
1747                         return 0;
1748                 x = x->u.binary.right;
1749         }
1750 }
1751
1752 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1753 {
1754         /*
1755          * we do not have to do the two-pass grep when we do not check
1756          * buffer-wide "all-match".
1757          */
1758         if (!opt->all_match && !opt->no_body_match)
1759                 return grep_source_1(opt, gs, 0);
1760
1761         /* Otherwise the toplevel "or" terms hit a bit differently.
1762          * We first clear hit markers from them.
1763          */
1764         clr_hit_marker(opt->pattern_expression);
1765         opt->body_hit = 0;
1766         grep_source_1(opt, gs, 1);
1767
1768         if (opt->all_match && !chk_hit_marker(opt->pattern_expression))
1769                 return 0;
1770         if (opt->no_body_match && opt->body_hit)
1771                 return 0;
1772
1773         return grep_source_1(opt, gs, 0);
1774 }
1775
1776 static void grep_source_init_buf(struct grep_source *gs,
1777                                  const char *buf,
1778                                  unsigned long size)
1779 {
1780         gs->type = GREP_SOURCE_BUF;
1781         gs->name = NULL;
1782         gs->path = NULL;
1783         gs->buf = buf;
1784         gs->size = size;
1785         gs->driver = NULL;
1786         gs->identifier = NULL;
1787 }
1788
1789 int grep_buffer(struct grep_opt *opt, const char *buf, unsigned long size)
1790 {
1791         struct grep_source gs;
1792         int r;
1793
1794         grep_source_init_buf(&gs, buf, size);
1795
1796         r = grep_source(opt, &gs);
1797
1798         grep_source_clear(&gs);
1799         return r;
1800 }
1801
1802 void grep_source_init_file(struct grep_source *gs, const char *name,
1803                            const char *path)
1804 {
1805         gs->type = GREP_SOURCE_FILE;
1806         gs->name = xstrdup_or_null(name);
1807         gs->path = xstrdup_or_null(path);
1808         gs->buf = NULL;
1809         gs->size = 0;
1810         gs->driver = NULL;
1811         gs->identifier = xstrdup(path);
1812 }
1813
1814 void grep_source_init_oid(struct grep_source *gs, const char *name,
1815                           const char *path, const struct object_id *oid,
1816                           struct repository *repo)
1817 {
1818         gs->type = GREP_SOURCE_OID;
1819         gs->name = xstrdup_or_null(name);
1820         gs->path = xstrdup_or_null(path);
1821         gs->buf = NULL;
1822         gs->size = 0;
1823         gs->driver = NULL;
1824         gs->identifier = oiddup(oid);
1825         gs->repo = repo;
1826 }
1827
1828 void grep_source_clear(struct grep_source *gs)
1829 {
1830         FREE_AND_NULL(gs->name);
1831         FREE_AND_NULL(gs->path);
1832         FREE_AND_NULL(gs->identifier);
1833         grep_source_clear_data(gs);
1834 }
1835
1836 void grep_source_clear_data(struct grep_source *gs)
1837 {
1838         switch (gs->type) {
1839         case GREP_SOURCE_FILE:
1840         case GREP_SOURCE_OID:
1841                 /* these types own the buffer */
1842                 free((char *)gs->buf);
1843                 gs->buf = NULL;
1844                 gs->size = 0;
1845                 break;
1846         case GREP_SOURCE_BUF:
1847                 /* leave user-provided buf intact */
1848                 break;
1849         }
1850 }
1851
1852 static int grep_source_load_oid(struct grep_source *gs)
1853 {
1854         enum object_type type;
1855
1856         gs->buf = repo_read_object_file(gs->repo, gs->identifier, &type,
1857                                         &gs->size);
1858         if (!gs->buf)
1859                 return error(_("'%s': unable to read %s"),
1860                              gs->name,
1861                              oid_to_hex(gs->identifier));
1862         return 0;
1863 }
1864
1865 static int grep_source_load_file(struct grep_source *gs)
1866 {
1867         const char *filename = gs->identifier;
1868         struct stat st;
1869         char *data;
1870         size_t size;
1871         int i;
1872
1873         if (lstat(filename, &st) < 0) {
1874         err_ret:
1875                 if (errno != ENOENT)
1876                         error_errno(_("failed to stat '%s'"), filename);
1877                 return -1;
1878         }
1879         if (!S_ISREG(st.st_mode))
1880                 return -1;
1881         size = xsize_t(st.st_size);
1882         i = open(filename, O_RDONLY);
1883         if (i < 0)
1884                 goto err_ret;
1885         data = xmallocz(size);
1886         if (st.st_size != read_in_full(i, data, size)) {
1887                 error_errno(_("'%s': short read"), filename);
1888                 close(i);
1889                 free(data);
1890                 return -1;
1891         }
1892         close(i);
1893
1894         gs->buf = data;
1895         gs->size = size;
1896         return 0;
1897 }
1898
1899 static int grep_source_load(struct grep_source *gs)
1900 {
1901         if (gs->buf)
1902                 return 0;
1903
1904         switch (gs->type) {
1905         case GREP_SOURCE_FILE:
1906                 return grep_source_load_file(gs);
1907         case GREP_SOURCE_OID:
1908                 return grep_source_load_oid(gs);
1909         case GREP_SOURCE_BUF:
1910                 return gs->buf ? 0 : -1;
1911         }
1912         BUG("invalid grep_source type to load");
1913 }
1914
1915 void grep_source_load_driver(struct grep_source *gs,
1916                              struct index_state *istate)
1917 {
1918         if (gs->driver)
1919                 return;
1920
1921         grep_attr_lock();
1922         if (gs->path)
1923                 gs->driver = userdiff_find_by_path(istate, gs->path);
1924         if (!gs->driver)
1925                 gs->driver = userdiff_find_by_name("default");
1926         grep_attr_unlock();
1927 }
1928
1929 static int grep_source_is_binary(struct grep_source *gs,
1930                                  struct index_state *istate)
1931 {
1932         grep_source_load_driver(gs, istate);
1933         if (gs->driver->binary != -1)
1934                 return gs->driver->binary;
1935
1936         if (!grep_source_load(gs))
1937                 return buffer_is_binary(gs->buf, gs->size);
1938
1939         return 0;
1940 }