grep.c

   1 #include "cache.h"
   2 #include "config.h"
   3 #include "grep.h"
   4 #include "userdiff.h"
   5 #include "xdiff-interface.h"
   6 #include "diff.h"
   7 #include "diffcore.h"
   8 #include "commit.h"
   9 #include "quote.h"
  10
  11 static int grep_source_load(struct grep_source *gs);
  12 static int grep_source_is_binary(struct grep_source *gs);
  13
  14 static struct grep_opt grep_defaults;
  15
  16 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
  17 {
  18         fwrite(buf, size, 1, stdout);
  19 }
  20
  21 /*
  22  * Initialize the grep_defaults template with hardcoded defaults.
  23  * We could let the compiler do this, but without C99 initializers
  24  * the code gets unwieldy and unreadable, so...
  25  */
  26 void init_grep_defaults(void)
  27 {
  28         struct grep_opt *opt = &grep_defaults;
  29         static int run_once;
  30
  31         if (run_once)
  32                 return;
  33         run_once++;
  34
  35         memset(opt, 0, sizeof(*opt));
  36         opt->relative = 1;
  37         opt->pathname = 1;
  38         opt->regflags = REG_NEWLINE;
  39         opt->max_depth = -1;
  40         opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  41         opt->extended_regexp_option = 0;
  42         color_set(opt->color_context, "");
  43         color_set(opt->color_filename, "");
  44         color_set(opt->color_function, "");
  45         color_set(opt->color_lineno, "");
  46         color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
  47         color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
  48         color_set(opt->color_selected, "");
  49         color_set(opt->color_sep, GIT_COLOR_CYAN);
  50         opt->color = -1;
  51         opt->output = std_output;
  52 }
  53
  54 static int parse_pattern_type_arg(const char *opt, const char *arg)
  55 {
  56         if (!strcmp(arg, "default"))
  57                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  58         else if (!strcmp(arg, "basic"))
  59                 return GREP_PATTERN_TYPE_BRE;
  60         else if (!strcmp(arg, "extended"))
  61                 return GREP_PATTERN_TYPE_ERE;
  62         else if (!strcmp(arg, "fixed"))
  63                 return GREP_PATTERN_TYPE_FIXED;
  64         else if (!strcmp(arg, "perl"))
  65                 return GREP_PATTERN_TYPE_PCRE;
  66         die("bad %s argument: %s", opt, arg);
  67 }
  68
  69 /*
  70  * Read the configuration file once and store it in
  71  * the grep_defaults template.
  72  */
  73 int grep_config(const char *var, const char *value, void *cb)
  74 {
  75         struct grep_opt *opt = &grep_defaults;
  76         char *color = NULL;
  77
  78         if (userdiff_config(var, value) < 0)
  79                 return -1;
  80
  81         if (!strcmp(var, "grep.extendedregexp")) {
  82                 if (git_config_bool(var, value))
  83                         opt->extended_regexp_option = 1;
  84                 else
  85                         opt->extended_regexp_option = 0;
  86                 return 0;
  87         }
  88
  89         if (!strcmp(var, "grep.patterntype")) {
  90                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
  91                 return 0;
  92         }
  93
  94         if (!strcmp(var, "grep.linenumber")) {
  95                 opt->linenum = git_config_bool(var, value);
  96                 return 0;
  97         }
  98
  99         if (!strcmp(var, "grep.fullname")) {
 100                 opt->relative = !git_config_bool(var, value);
 101                 return 0;
 102         }
 103
 104         if (!strcmp(var, "color.grep"))
 105                 opt->color = git_config_colorbool(var, value);
 106         else if (!strcmp(var, "color.grep.context"))
 107                 color = opt->color_context;
 108         else if (!strcmp(var, "color.grep.filename"))
 109                 color = opt->color_filename;
 110         else if (!strcmp(var, "color.grep.function"))
 111                 color = opt->color_function;
 112         else if (!strcmp(var, "color.grep.linenumber"))
 113                 color = opt->color_lineno;
 114         else if (!strcmp(var, "color.grep.matchcontext"))
 115                 color = opt->color_match_context;
 116         else if (!strcmp(var, "color.grep.matchselected"))
 117                 color = opt->color_match_selected;
 118         else if (!strcmp(var, "color.grep.selected"))
 119                 color = opt->color_selected;
 120         else if (!strcmp(var, "color.grep.separator"))
 121                 color = opt->color_sep;
 122         else if (!strcmp(var, "color.grep.match")) {
 123                 int rc = 0;
 124                 if (!value)
 125                         return config_error_nonbool(var);
 126                 rc |= color_parse(value, opt->color_match_context);
 127                 rc |= color_parse(value, opt->color_match_selected);
 128                 return rc;
 129         }
 130
 131         if (color) {
 132                 if (!value)
 133                         return config_error_nonbool(var);
 134                 return color_parse(value, color);
 135         }
 136         return 0;
 137 }
 138
 139 /*
 140  * Initialize one instance of grep_opt and copy the
 141  * default values from the template we read the configuration
 142  * information in an earlier call to git_config(grep_config).
 143  */
 144 void grep_init(struct grep_opt *opt, const char *prefix)
 145 {
 146         struct grep_opt *def = &grep_defaults;
 147
 148         memset(opt, 0, sizeof(*opt));
 149         opt->prefix = prefix;
 150         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 151         opt->pattern_tail = &opt->pattern_list;
 152         opt->header_tail = &opt->header_list;
 153
 154         opt->color = def->color;
 155         opt->extended_regexp_option = def->extended_regexp_option;
 156         opt->pattern_type_option = def->pattern_type_option;
 157         opt->linenum = def->linenum;
 158         opt->max_depth = def->max_depth;
 159         opt->pathname = def->pathname;
 160         opt->regflags = def->regflags;
 161         opt->relative = def->relative;
 162         opt->output = def->output;
 163
 164         color_set(opt->color_context, def->color_context);
 165         color_set(opt->color_filename, def->color_filename);
 166         color_set(opt->color_function, def->color_function);
 167         color_set(opt->color_lineno, def->color_lineno);
 168         color_set(opt->color_match_context, def->color_match_context);
 169         color_set(opt->color_match_selected, def->color_match_selected);
 170         color_set(opt->color_selected, def->color_selected);
 171         color_set(opt->color_sep, def->color_sep);
 172 }
 173
 174 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 175 {
 176         switch (pattern_type) {
 177         case GREP_PATTERN_TYPE_UNSPECIFIED:
 178                 /* fall through */
 179
 180         case GREP_PATTERN_TYPE_BRE:
 181                 opt->fixed = 0;
 182                 opt->pcre = 0;
 183                 opt->regflags &= ~REG_EXTENDED;
 184                 break;
 185
 186         case GREP_PATTERN_TYPE_ERE:
 187                 opt->fixed = 0;
 188                 opt->pcre = 0;
 189                 opt->regflags |= REG_EXTENDED;
 190                 break;
 191
 192         case GREP_PATTERN_TYPE_FIXED:
 193                 opt->fixed = 1;
 194                 opt->pcre = 0;
 195                 opt->regflags &= ~REG_EXTENDED;
 196                 break;
 197
 198         case GREP_PATTERN_TYPE_PCRE:
 199                 opt->fixed = 0;
 200                 opt->pcre = 1;
 201                 opt->regflags &= ~REG_EXTENDED;
 202                 break;
 203         }
 204 }
 205
 206 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 207 {
 208         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 209                 grep_set_pattern_type_option(pattern_type, opt);
 210         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 211                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 212         else if (opt->extended_regexp_option)
 213                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 214 }
 215
 216 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 217                                         const char *origin, int no,
 218                                         enum grep_pat_token t,
 219                                         enum grep_header_field field)
 220 {
 221         struct grep_pat *p = xcalloc(1, sizeof(*p));
 222         p->pattern = xmemdupz(pat, patlen);
 223         p->patternlen = patlen;
 224         p->origin = origin;
 225         p->no = no;
 226         p->token = t;
 227         p->field = field;
 228         return p;
 229 }
 230
 231 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 232 {
 233         **tail = p;
 234         *tail = &p->next;
 235         p->next = NULL;
 236
 237         switch (p->token) {
 238         case GREP_PATTERN: /* atom */
 239         case GREP_PATTERN_HEAD:
 240         case GREP_PATTERN_BODY:
 241                 for (;;) {
 242                         struct grep_pat *new_pat;
 243                         size_t len = 0;
 244                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 245                         while (++len <= p->patternlen) {
 246                                 if (*(--cp) == '\n') {
 247                                         nl = cp;
 248                                         break;
 249                                 }
 250                         }
 251                         if (!nl)
 252                                 break;
 253                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 254                                                   p->no, p->token, p->field);
 255                         new_pat->next = p->next;
 256                         if (!p->next)
 257                                 *tail = &new_pat->next;
 258                         p->next = new_pat;
 259                         *nl = '\0';
 260                         p->patternlen -= len;
 261                 }
 262                 break;
 263         default:
 264                 break;
 265         }
 266 }
 267
 268 void append_header_grep_pattern(struct grep_opt *opt,
 269                                 enum grep_header_field field, const char *pat)
 270 {
 271         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 272                                              GREP_PATTERN_HEAD, field);
 273         if (field == GREP_HEADER_REFLOG)
 274                 opt->use_reflog_filter = 1;
 275         do_append_grep_pat(&opt->header_tail, p);
 276 }
 277
 278 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 279                          const char *origin, int no, enum grep_pat_token t)
 280 {
 281         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 282 }
 283
 284 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 285                      const char *origin, int no, enum grep_pat_token t)
 286 {
 287         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 288         do_append_grep_pat(&opt->pattern_tail, p);
 289 }
 290
 291 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 292 {
 293         struct grep_pat *pat;
 294         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 295         *ret = *opt;
 296
 297         ret->pattern_list = NULL;
 298         ret->pattern_tail = &ret->pattern_list;
 299
 300         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 301         {
 302                 if(pat->token == GREP_PATTERN_HEAD)
 303                         append_header_grep_pattern(ret, pat->field,
 304                                                    pat->pattern);
 305                 else
 306                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 307                                         pat->origin, pat->no, pat->token);
 308         }
 309
 310         return ret;
 311 }
 312
 313 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 314                 const char *error)
 315 {
 316         char where[1024];
 317
 318         if (p->no)
 319                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 320         else if (p->origin)
 321                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 322         else
 323                 where[0] = 0;
 324
 325         die("%s'%s': %s", where, p->pattern, error);
 326 }
 327
 328 #ifdef USE_LIBPCRE
 329 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 330 {
 331         const char *error;
 332         int erroffset;
 333         int options = PCRE_MULTILINE;
 334
 335         if (opt->ignore_case) {
 336                 if (has_non_ascii(p->pattern))
 337                         p->pcre_tables = pcre_maketables();
 338                 options |= PCRE_CASELESS;
 339         }
 340         if (is_utf8_locale() && has_non_ascii(p->pattern))
 341                 options |= PCRE_UTF8;
 342
 343         p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 344                                       p->pcre_tables);
 345         if (!p->pcre_regexp)
 346                 compile_regexp_failed(p, error);
 347
 348         p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 349         if (!p->pcre_extra_info && error)
 350                 die("%s", error);
 351 }
 352
 353 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 354                 regmatch_t *match, int eflags)
 355 {
 356         int ovector[30], ret, flags = 0;
 357
 358         if (eflags & REG_NOTBOL)
 359                 flags |= PCRE_NOTBOL;
 360
 361         ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 362                         0, flags, ovector, ARRAY_SIZE(ovector));
 363         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 364                 die("pcre_exec failed with error code %d", ret);
 365         if (ret > 0) {
 366                 ret = 0;
 367                 match->rm_so = ovector[0];
 368                 match->rm_eo = ovector[1];
 369         }
 370
 371         return ret;
 372 }
 373
 374 static void free_pcre_regexp(struct grep_pat *p)
 375 {
 376         pcre_free(p->pcre_regexp);
 377         pcre_free(p->pcre_extra_info);
 378         pcre_free((void *)p->pcre_tables);
 379 }
 380 #else /* !USE_LIBPCRE */
 381 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 382 {
 383         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 384 }
 385
 386 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 387                 regmatch_t *match, int eflags)
 388 {
 389         return 1;
 390 }
 391
 392 static void free_pcre_regexp(struct grep_pat *p)
 393 {
 394 }
 395 #endif /* !USE_LIBPCRE */
 396
 397 static int is_fixed(const char *s, size_t len)
 398 {
 399         size_t i;
 400
 401         /* regcomp cannot accept patterns with NULs so we
 402          * consider any pattern containing a NUL fixed.
 403          */
 404         if (memchr(s, 0, len))
 405                 return 1;
 406
 407         for (i = 0; i < len; i++) {
 408                 if (is_regex_special(s[i]))
 409                         return 0;
 410         }
 411
 412         return 1;
 413 }
 414
 415 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 416 {
 417         struct strbuf sb = STRBUF_INIT;
 418         int err;
 419         int regflags;
 420
 421         basic_regex_quote_buf(&sb, p->pattern);
 422         regflags = opt->regflags & ~REG_EXTENDED;
 423         if (opt->ignore_case)
 424                 regflags |= REG_ICASE;
 425         err = regcomp(&p->regexp, sb.buf, regflags);
 426         if (opt->debug)
 427                 fprintf(stderr, "fixed %s\n", sb.buf);
 428         strbuf_release(&sb);
 429         if (err) {
 430                 char errbuf[1024];
 431                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 432                 regfree(&p->regexp);
 433                 compile_regexp_failed(p, errbuf);
 434         }
 435 }
 436
 437 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 438 {
 439         int icase, ascii_only;
 440         int err;
 441
 442         p->word_regexp = opt->word_regexp;
 443         p->ignore_case = opt->ignore_case;
 444         icase          = opt->regflags & REG_ICASE || p->ignore_case;
 445         ascii_only     = !has_non_ascii(p->pattern);
 446
 447         /*
 448          * Even when -F (fixed) asks us to do a non-regexp search, we
 449          * may not be able to correctly case-fold when -i
 450          * (ignore-case) is asked (in which case, we'll synthesize a
 451          * regexp to match the pattern that matches regexp special
 452          * characters literally, while ignoring case differences).  On
 453          * the other hand, even without -F, if the pattern does not
 454          * have any regexp special characters and there is no need for
 455          * case-folding search, we can internally turn it into a
 456          * simple string match using kws.  p->fixed tells us if we
 457          * want to use kws.
 458          */
 459         if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 460                 p->fixed = !icase || ascii_only;
 461         else
 462                 p->fixed = 0;
 463
 464         if (p->fixed) {
 465                 p->kws = kwsalloc(icase ? tolower_trans_tbl : NULL);
 466                 kwsincr(p->kws, p->pattern, p->patternlen);
 467                 kwsprep(p->kws);
 468                 return;
 469         } else if (opt->fixed) {
 470                 /*
 471                  * We come here when the pattern has the non-ascii
 472                  * characters we cannot case-fold, and asked to
 473                  * ignore-case.
 474                  */
 475                 compile_fixed_regexp(p, opt);
 476                 return;
 477         }
 478
 479         if (opt->pcre) {
 480                 compile_pcre_regexp(p, opt);
 481                 return;
 482         }
 483
 484         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 485         if (err) {
 486                 char errbuf[1024];
 487                 regerror(err, &p->regexp, errbuf, 1024);
 488                 regfree(&p->regexp);
 489                 compile_regexp_failed(p, errbuf);
 490         }
 491 }
 492
 493 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 494 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 495 {
 496         struct grep_pat *p;
 497         struct grep_expr *x;
 498
 499         p = *list;
 500         if (!p)
 501                 return NULL;
 502         switch (p->token) {
 503         case GREP_PATTERN: /* atom */
 504         case GREP_PATTERN_HEAD:
 505         case GREP_PATTERN_BODY:
 506                 x = xcalloc(1, sizeof (struct grep_expr));
 507                 x->node = GREP_NODE_ATOM;
 508                 x->u.atom = p;
 509                 *list = p->next;
 510                 return x;
 511         case GREP_OPEN_PAREN:
 512                 *list = p->next;
 513                 x = compile_pattern_or(list);
 514                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 515                         die("unmatched parenthesis");
 516                 *list = (*list)->next;
 517                 return x;
 518         default:
 519                 return NULL;
 520         }
 521 }
 522
 523 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 524 {
 525         struct grep_pat *p;
 526         struct grep_expr *x;
 527
 528         p = *list;
 529         if (!p)
 530                 return NULL;
 531         switch (p->token) {
 532         case GREP_NOT:
 533                 if (!p->next)
 534                         die("--not not followed by pattern expression");
 535                 *list = p->next;
 536                 x = xcalloc(1, sizeof (struct grep_expr));
 537                 x->node = GREP_NODE_NOT;
 538                 x->u.unary = compile_pattern_not(list);
 539                 if (!x->u.unary)
 540                         die("--not followed by non pattern expression");
 541                 return x;
 542         default:
 543                 return compile_pattern_atom(list);
 544         }
 545 }
 546
 547 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 548 {
 549         struct grep_pat *p;
 550         struct grep_expr *x, *y, *z;
 551
 552         x = compile_pattern_not(list);
 553         p = *list;
 554         if (p && p->token == GREP_AND) {
 555                 if (!p->next)
 556                         die("--and not followed by pattern expression");
 557                 *list = p->next;
 558                 y = compile_pattern_and(list);
 559                 if (!y)
 560                         die("--and not followed by pattern expression");
 561                 z = xcalloc(1, sizeof (struct grep_expr));
 562                 z->node = GREP_NODE_AND;
 563                 z->u.binary.left = x;
 564                 z->u.binary.right = y;
 565                 return z;
 566         }
 567         return x;
 568 }
 569
 570 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 571 {
 572         struct grep_pat *p;
 573         struct grep_expr *x, *y, *z;
 574
 575         x = compile_pattern_and(list);
 576         p = *list;
 577         if (x && p && p->token != GREP_CLOSE_PAREN) {
 578                 y = compile_pattern_or(list);
 579                 if (!y)
 580                         die("not a pattern expression %s", p->pattern);
 581                 z = xcalloc(1, sizeof (struct grep_expr));
 582                 z->node = GREP_NODE_OR;
 583                 z->u.binary.left = x;
 584                 z->u.binary.right = y;
 585                 return z;
 586         }
 587         return x;
 588 }
 589
 590 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 591 {
 592         return compile_pattern_or(list);
 593 }
 594
 595 static void indent(int in)
 596 {
 597         while (in-- > 0)
 598                 fputc(' ', stderr);
 599 }
 600
 601 static void dump_grep_pat(struct grep_pat *p)
 602 {
 603         switch (p->token) {
 604         case GREP_AND: fprintf(stderr, "*and*"); break;
 605         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 606         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 607         case GREP_NOT: fprintf(stderr, "*not*"); break;
 608         case GREP_OR: fprintf(stderr, "*or*"); break;
 609
 610         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 611         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 612         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 613         }
 614
 615         switch (p->token) {
 616         default: break;
 617         case GREP_PATTERN_HEAD:
 618                 fprintf(stderr, "<head %d>", p->field); break;
 619         case GREP_PATTERN_BODY:
 620                 fprintf(stderr, "<body>"); break;
 621         }
 622         switch (p->token) {
 623         default: break;
 624         case GREP_PATTERN_HEAD:
 625         case GREP_PATTERN_BODY:
 626         case GREP_PATTERN:
 627                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 628                 break;
 629         }
 630         fputc('\n', stderr);
 631 }
 632
 633 static void dump_grep_expression_1(struct grep_expr *x, int in)
 634 {
 635         indent(in);
 636         switch (x->node) {
 637         case GREP_NODE_TRUE:
 638                 fprintf(stderr, "true\n");
 639                 break;
 640         case GREP_NODE_ATOM:
 641                 dump_grep_pat(x->u.atom);
 642                 break;
 643         case GREP_NODE_NOT:
 644                 fprintf(stderr, "(not\n");
 645                 dump_grep_expression_1(x->u.unary, in+1);
 646                 indent(in);
 647                 fprintf(stderr, ")\n");
 648                 break;
 649         case GREP_NODE_AND:
 650                 fprintf(stderr, "(and\n");
 651                 dump_grep_expression_1(x->u.binary.left, in+1);
 652                 dump_grep_expression_1(x->u.binary.right, in+1);
 653                 indent(in);
 654                 fprintf(stderr, ")\n");
 655                 break;
 656         case GREP_NODE_OR:
 657                 fprintf(stderr, "(or\n");
 658                 dump_grep_expression_1(x->u.binary.left, in+1);
 659                 dump_grep_expression_1(x->u.binary.right, in+1);
 660                 indent(in);
 661                 fprintf(stderr, ")\n");
 662                 break;
 663         }
 664 }
 665
 666 static void dump_grep_expression(struct grep_opt *opt)
 667 {
 668         struct grep_expr *x = opt->pattern_expression;
 669
 670         if (opt->all_match)
 671                 fprintf(stderr, "[all-match]\n");
 672         dump_grep_expression_1(x, 0);
 673         fflush(NULL);
 674 }
 675
 676 static struct grep_expr *grep_true_expr(void)
 677 {
 678         struct grep_expr *z = xcalloc(1, sizeof(*z));
 679         z->node = GREP_NODE_TRUE;
 680         return z;
 681 }
 682
 683 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 684 {
 685         struct grep_expr *z = xcalloc(1, sizeof(*z));
 686         z->node = GREP_NODE_OR;
 687         z->u.binary.left = left;
 688         z->u.binary.right = right;
 689         return z;
 690 }
 691
 692 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 693 {
 694         struct grep_pat *p;
 695         struct grep_expr *header_expr;
 696         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 697         enum grep_header_field fld;
 698
 699         if (!opt->header_list)
 700                 return NULL;
 701
 702         for (p = opt->header_list; p; p = p->next) {
 703                 if (p->token != GREP_PATTERN_HEAD)
 704                         die("BUG: a non-header pattern in grep header list.");
 705                 if (p->field < GREP_HEADER_FIELD_MIN ||
 706                     GREP_HEADER_FIELD_MAX <= p->field)
 707                         die("BUG: unknown header field %d", p->field);
 708                 compile_regexp(p, opt);
 709         }
 710
 711         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 712                 header_group[fld] = NULL;
 713
 714         for (p = opt->header_list; p; p = p->next) {
 715                 struct grep_expr *h;
 716                 struct grep_pat *pp = p;
 717
 718                 h = compile_pattern_atom(&pp);
 719                 if (!h || pp != p->next)
 720                         die("BUG: malformed header expr");
 721                 if (!header_group[p->field]) {
 722                         header_group[p->field] = h;
 723                         continue;
 724                 }
 725                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 726         }
 727
 728         header_expr = NULL;
 729
 730         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 731                 if (!header_group[fld])
 732                         continue;
 733                 if (!header_expr)
 734                         header_expr = grep_true_expr();
 735                 header_expr = grep_or_expr(header_group[fld], header_expr);
 736         }
 737         return header_expr;
 738 }
 739
 740 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 741 {
 742         struct grep_expr *z = x;
 743
 744         while (x) {
 745                 assert(x->node == GREP_NODE_OR);
 746                 if (x->u.binary.right &&
 747                     x->u.binary.right->node == GREP_NODE_TRUE) {
 748                         x->u.binary.right = y;
 749                         break;
 750                 }
 751                 x = x->u.binary.right;
 752         }
 753         return z;
 754 }
 755
 756 static void compile_grep_patterns_real(struct grep_opt *opt)
 757 {
 758         struct grep_pat *p;
 759         struct grep_expr *header_expr = prep_header_patterns(opt);
 760
 761         for (p = opt->pattern_list; p; p = p->next) {
 762                 switch (p->token) {
 763                 case GREP_PATTERN: /* atom */
 764                 case GREP_PATTERN_HEAD:
 765                 case GREP_PATTERN_BODY:
 766                         compile_regexp(p, opt);
 767                         break;
 768                 default:
 769                         opt->extended = 1;
 770                         break;
 771                 }
 772         }
 773
 774         if (opt->all_match || header_expr)
 775                 opt->extended = 1;
 776         else if (!opt->extended && !opt->debug)
 777                 return;
 778
 779         p = opt->pattern_list;
 780         if (p)
 781                 opt->pattern_expression = compile_pattern_expr(&p);
 782         if (p)
 783                 die("incomplete pattern expression: %s", p->pattern);
 784
 785         if (!header_expr)
 786                 return;
 787
 788         if (!opt->pattern_expression)
 789                 opt->pattern_expression = header_expr;
 790         else if (opt->all_match)
 791                 opt->pattern_expression = grep_splice_or(header_expr,
 792                                                          opt->pattern_expression);
 793         else
 794                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 795                                                        header_expr);
 796         opt->all_match = 1;
 797 }
 798
 799 void compile_grep_patterns(struct grep_opt *opt)
 800 {
 801         compile_grep_patterns_real(opt);
 802         if (opt->debug)
 803                 dump_grep_expression(opt);
 804 }
 805
 806 static void free_pattern_expr(struct grep_expr *x)
 807 {
 808         switch (x->node) {
 809         case GREP_NODE_TRUE:
 810         case GREP_NODE_ATOM:
 811                 break;
 812         case GREP_NODE_NOT:
 813                 free_pattern_expr(x->u.unary);
 814                 break;
 815         case GREP_NODE_AND:
 816         case GREP_NODE_OR:
 817                 free_pattern_expr(x->u.binary.left);
 818                 free_pattern_expr(x->u.binary.right);
 819                 break;
 820         }
 821         free(x);
 822 }
 823
 824 void free_grep_patterns(struct grep_opt *opt)
 825 {
 826         struct grep_pat *p, *n;
 827
 828         for (p = opt->pattern_list; p; p = n) {
 829                 n = p->next;
 830                 switch (p->token) {
 831                 case GREP_PATTERN: /* atom */
 832                 case GREP_PATTERN_HEAD:
 833                 case GREP_PATTERN_BODY:
 834                         if (p->kws)
 835                                 kwsfree(p->kws);
 836                         else if (p->pcre_regexp)
 837                                 free_pcre_regexp(p);
 838                         else
 839                                 regfree(&p->regexp);
 840                         free(p->pattern);
 841                         break;
 842                 default:
 843                         break;
 844                 }
 845                 free(p);
 846         }
 847
 848         if (!opt->extended)
 849                 return;
 850         free_pattern_expr(opt->pattern_expression);
 851 }
 852
 853 static char *end_of_line(char *cp, unsigned long *left)
 854 {
 855         unsigned long l = *left;
 856         while (l && *cp != '\n') {
 857                 l--;
 858                 cp++;
 859         }
 860         *left = l;
 861         return cp;
 862 }
 863
 864 static int word_char(char ch)
 865 {
 866         return isalnum(ch) || ch == '_';
 867 }
 868
 869 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 870                          const char *color)
 871 {
 872         if (want_color(opt->color) && color && color[0]) {
 873                 opt->output(opt, color, strlen(color));
 874                 opt->output(opt, data, size);
 875                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 876         } else
 877                 opt->output(opt, data, size);
 878 }
 879
 880 static void output_sep(struct grep_opt *opt, char sign)
 881 {
 882         if (opt->null_following_name)
 883                 opt->output(opt, "\0", 1);
 884         else
 885                 output_color(opt, &sign, 1, opt->color_sep);
 886 }
 887
 888 static void show_name(struct grep_opt *opt, const char *name)
 889 {
 890         output_color(opt, name, strlen(name), opt->color_filename);
 891         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 892 }
 893
 894 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 895                     regmatch_t *match)
 896 {
 897         struct kwsmatch kwsm;
 898         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 899         if (offset == -1) {
 900                 match->rm_so = match->rm_eo = -1;
 901                 return REG_NOMATCH;
 902         } else {
 903                 match->rm_so = offset;
 904                 match->rm_eo = match->rm_so + kwsm.size[0];
 905                 return 0;
 906         }
 907 }
 908
 909 static int patmatch(struct grep_pat *p, char *line, char *eol,
 910                     regmatch_t *match, int eflags)
 911 {
 912         int hit;
 913
 914         if (p->fixed)
 915                 hit = !fixmatch(p, line, eol, match);
 916         else if (p->pcre_regexp)
 917                 hit = !pcrematch(p, line, eol, match, eflags);
 918         else
 919                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
 920                                    eflags);
 921
 922         return hit;
 923 }
 924
 925 static int strip_timestamp(char *bol, char **eol_p)
 926 {
 927         char *eol = *eol_p;
 928         int ch;
 929
 930         while (bol < --eol) {
 931                 if (*eol != '>')
 932                         continue;
 933                 *eol_p = ++eol;
 934                 ch = *eol;
 935                 *eol = '\0';
 936                 return ch;
 937         }
 938         return 0;
 939 }
 940
 941 static struct {
 942         const char *field;
 943         size_t len;
 944 } header_field[] = {
 945         { "author ", 7 },
 946         { "committer ", 10 },
 947         { "reflog ", 7 },
 948 };
 949
 950 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 951                              enum grep_context ctx,
 952                              regmatch_t *pmatch, int eflags)
 953 {
 954         int hit = 0;
 955         int saved_ch = 0;
 956         const char *start = bol;
 957
 958         if ((p->token != GREP_PATTERN) &&
 959             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 960                 return 0;
 961
 962         if (p->token == GREP_PATTERN_HEAD) {
 963                 const char *field;
 964                 size_t len;
 965                 assert(p->field < ARRAY_SIZE(header_field));
 966                 field = header_field[p->field].field;
 967                 len = header_field[p->field].len;
 968                 if (strncmp(bol, field, len))
 969                         return 0;
 970                 bol += len;
 971                 switch (p->field) {
 972                 case GREP_HEADER_AUTHOR:
 973                 case GREP_HEADER_COMMITTER:
 974                         saved_ch = strip_timestamp(bol, &eol);
 975                         break;
 976                 default:
 977                         break;
 978                 }
 979         }
 980
 981  again:
 982         hit = patmatch(p, bol, eol, pmatch, eflags);
 983
 984         if (hit && p->word_regexp) {
 985                 if ((pmatch[0].rm_so < 0) ||
 986                     (eol - bol) < pmatch[0].rm_so ||
 987                     (pmatch[0].rm_eo < 0) ||
 988                     (eol - bol) < pmatch[0].rm_eo)
 989                         die("regexp returned nonsense");
 990
 991                 /* Match beginning must be either beginning of the
 992                  * line, or at word boundary (i.e. the last char must
 993                  * not be a word char).  Similarly, match end must be
 994                  * either end of the line, or at word boundary
 995                  * (i.e. the next char must not be a word char).
 996                  */
 997                 if ( ((pmatch[0].rm_so == 0) ||
 998                       !word_char(bol[pmatch[0].rm_so-1])) &&
 999                      ((pmatch[0].rm_eo == (eol-bol)) ||
1000                       !word_char(bol[pmatch[0].rm_eo])) )
1001                         ;
1002                 else
1003                         hit = 0;
1004
1005                 /* Words consist of at least one character. */
1006                 if (pmatch->rm_so == pmatch->rm_eo)
1007                         hit = 0;
1008
1009                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1010                         /* There could be more than one match on the
1011                          * line, and the first match might not be
1012                          * strict word match.  But later ones could be!
1013                          * Forward to the next possible start, i.e. the
1014                          * next position following a non-word char.
1015                          */
1016                         bol = pmatch[0].rm_so + bol + 1;
1017                         while (word_char(bol[-1]) && bol < eol)
1018                                 bol++;
1019                         eflags |= REG_NOTBOL;
1020                         if (bol < eol)
1021                                 goto again;
1022                 }
1023         }
1024         if (p->token == GREP_PATTERN_HEAD && saved_ch)
1025                 *eol = saved_ch;
1026         if (hit) {
1027                 pmatch[0].rm_so += bol - start;
1028                 pmatch[0].rm_eo += bol - start;
1029         }
1030         return hit;
1031 }
1032
1033 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1034                            enum grep_context ctx, int collect_hits)
1035 {
1036         int h = 0;
1037         regmatch_t match;
1038
1039         if (!x)
1040                 die("Not a valid grep expression");
1041         switch (x->node) {
1042         case GREP_NODE_TRUE:
1043                 h = 1;
1044                 break;
1045         case GREP_NODE_ATOM:
1046                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1047                 break;
1048         case GREP_NODE_NOT:
1049                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1050                 break;
1051         case GREP_NODE_AND:
1052                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1053                         return 0;
1054                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1055                 break;
1056         case GREP_NODE_OR:
1057                 if (!collect_hits)
1058                         return (match_expr_eval(x->u.binary.left,
1059                                                 bol, eol, ctx, 0) ||
1060                                 match_expr_eval(x->u.binary.right,
1061                                                 bol, eol, ctx, 0));
1062                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1063                 x->u.binary.left->hit |= h;
1064                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1065                 break;
1066         default:
1067                 die("Unexpected node type (internal error) %d", x->node);
1068         }
1069         if (collect_hits)
1070                 x->hit |= h;
1071         return h;
1072 }
1073
1074 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1075                       enum grep_context ctx, int collect_hits)
1076 {
1077         struct grep_expr *x = opt->pattern_expression;
1078         return match_expr_eval(x, bol, eol, ctx, collect_hits);
1079 }
1080
1081 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1082                       enum grep_context ctx, int collect_hits)
1083 {
1084         struct grep_pat *p;
1085         regmatch_t match;
1086
1087         if (opt->extended)
1088                 return match_expr(opt, bol, eol, ctx, collect_hits);
1089
1090         /* we do not call with collect_hits without being extended */
1091         for (p = opt->pattern_list; p; p = p->next) {
1092                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1093                         return 1;
1094         }
1095         return 0;
1096 }
1097
1098 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1099                               enum grep_context ctx,
1100                               regmatch_t *pmatch, int eflags)
1101 {
1102         regmatch_t match;
1103
1104         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1105                 return 0;
1106         if (match.rm_so < 0 || match.rm_eo < 0)
1107                 return 0;
1108         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1109                 if (match.rm_so > pmatch->rm_so)
1110                         return 1;
1111                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1112                         return 1;
1113         }
1114         pmatch->rm_so = match.rm_so;
1115         pmatch->rm_eo = match.rm_eo;
1116         return 1;
1117 }
1118
1119 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1120                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1121 {
1122         struct grep_pat *p;
1123         int hit = 0;
1124
1125         pmatch->rm_so = pmatch->rm_eo = -1;
1126         if (bol < eol) {
1127                 for (p = opt->pattern_list; p; p = p->next) {
1128                         switch (p->token) {
1129                         case GREP_PATTERN: /* atom */
1130                         case GREP_PATTERN_HEAD:
1131                         case GREP_PATTERN_BODY:
1132                                 hit |= match_next_pattern(p, bol, eol, ctx,
1133                                                           pmatch, eflags);
1134                                 break;
1135                         default:
1136                                 break;
1137                         }
1138                 }
1139         }
1140         return hit;
1141 }
1142
1143 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1144                       const char *name, unsigned lno, char sign)
1145 {
1146         int rest = eol - bol;
1147         const char *match_color, *line_color = NULL;
1148
1149         if (opt->file_break && opt->last_shown == 0) {
1150                 if (opt->show_hunk_mark)
1151                         opt->output(opt, "\n", 1);
1152         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1153                 if (opt->last_shown == 0) {
1154                         if (opt->show_hunk_mark) {
1155                                 output_color(opt, "--", 2, opt->color_sep);
1156                                 opt->output(opt, "\n", 1);
1157                         }
1158                 } else if (lno > opt->last_shown + 1) {
1159                         output_color(opt, "--", 2, opt->color_sep);
1160                         opt->output(opt, "\n", 1);
1161                 }
1162         }
1163         if (opt->heading && opt->last_shown == 0) {
1164                 output_color(opt, name, strlen(name), opt->color_filename);
1165                 opt->output(opt, "\n", 1);
1166         }
1167         opt->last_shown = lno;
1168
1169         if (!opt->heading && opt->pathname) {
1170                 output_color(opt, name, strlen(name), opt->color_filename);
1171                 output_sep(opt, sign);
1172         }
1173         if (opt->linenum) {
1174                 char buf[32];
1175                 xsnprintf(buf, sizeof(buf), "%d", lno);
1176                 output_color(opt, buf, strlen(buf), opt->color_lineno);
1177                 output_sep(opt, sign);
1178         }
1179         if (opt->color) {
1180                 regmatch_t match;
1181                 enum grep_context ctx = GREP_CONTEXT_BODY;
1182                 int ch = *eol;
1183                 int eflags = 0;
1184
1185                 if (sign == ':')
1186                         match_color = opt->color_match_selected;
1187                 else
1188                         match_color = opt->color_match_context;
1189                 if (sign == ':')
1190                         line_color = opt->color_selected;
1191                 else if (sign == '-')
1192                         line_color = opt->color_context;
1193                 else if (sign == '=')
1194                         line_color = opt->color_function;
1195                 *eol = '\0';
1196                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1197                         if (match.rm_so == match.rm_eo)
1198                                 break;
1199
1200                         output_color(opt, bol, match.rm_so, line_color);
1201                         output_color(opt, bol + match.rm_so,
1202                                      match.rm_eo - match.rm_so, match_color);
1203                         bol += match.rm_eo;
1204                         rest -= match.rm_eo;
1205                         eflags = REG_NOTBOL;
1206                 }
1207                 *eol = ch;
1208         }
1209         output_color(opt, bol, rest, line_color);
1210         opt->output(opt, "\n", 1);
1211 }
1212
1213 #ifndef NO_PTHREADS
1214 int grep_use_locks;
1215
1216 /*
1217  * This lock protects access to the gitattributes machinery, which is
1218  * not thread-safe.
1219  */
1220 pthread_mutex_t grep_attr_mutex;
1221
1222 static inline void grep_attr_lock(void)
1223 {
1224         if (grep_use_locks)
1225                 pthread_mutex_lock(&grep_attr_mutex);
1226 }
1227
1228 static inline void grep_attr_unlock(void)
1229 {
1230         if (grep_use_locks)
1231                 pthread_mutex_unlock(&grep_attr_mutex);
1232 }
1233
1234 /*
1235  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1236  */
1237 pthread_mutex_t grep_read_mutex;
1238
1239 #else
1240 #define grep_attr_lock()
1241 #define grep_attr_unlock()
1242 #endif
1243
1244 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1245 {
1246         xdemitconf_t *xecfg = opt->priv;
1247         if (xecfg && !xecfg->find_func) {
1248                 grep_source_load_driver(gs);
1249                 if (gs->driver->funcname.pattern) {
1250                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1251                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1252                 } else {
1253                         xecfg = opt->priv = NULL;
1254                 }
1255         }
1256
1257         if (xecfg) {
1258                 char buf[1];
1259                 return xecfg->find_func(bol, eol - bol, buf, 1,
1260                                         xecfg->find_func_priv) >= 0;
1261         }
1262
1263         if (bol == eol)
1264                 return 0;
1265         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1266                 return 1;
1267         return 0;
1268 }
1269
1270 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1271                                char *bol, unsigned lno)
1272 {
1273         while (bol > gs->buf) {
1274                 char *eol = --bol;
1275
1276                 while (bol > gs->buf && bol[-1] != '\n')
1277                         bol--;
1278                 lno--;
1279
1280                 if (lno <= opt->last_shown)
1281                         break;
1282
1283                 if (match_funcname(opt, gs, bol, eol)) {
1284                         show_line(opt, bol, eol, gs->name, lno, '=');
1285                         break;
1286                 }
1287         }
1288 }
1289
1290 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1291                              char *bol, char *end, unsigned lno)
1292 {
1293         unsigned cur = lno, from = 1, funcname_lno = 0;
1294         int funcname_needed = !!opt->funcname;
1295
1296         if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1297                 funcname_needed = 2;
1298
1299         if (opt->pre_context < lno)
1300                 from = lno - opt->pre_context;
1301         if (from <= opt->last_shown)
1302                 from = opt->last_shown + 1;
1303
1304         /* Rewind. */
1305         while (bol > gs->buf &&
1306                cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1307                 char *eol = --bol;
1308
1309                 while (bol > gs->buf && bol[-1] != '\n')
1310                         bol--;
1311                 cur--;
1312                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1313                         funcname_lno = cur;
1314                         funcname_needed = 0;
1315                 }
1316         }
1317
1318         /* We need to look even further back to find a function signature. */
1319         if (opt->funcname && funcname_needed)
1320                 show_funcname_line(opt, gs, bol, cur);
1321
1322         /* Back forward. */
1323         while (cur < lno) {
1324                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1325
1326                 while (*eol != '\n')
1327                         eol++;
1328                 show_line(opt, bol, eol, gs->name, cur, sign);
1329                 bol = eol + 1;
1330                 cur++;
1331         }
1332 }
1333
1334 static int should_lookahead(struct grep_opt *opt)
1335 {
1336         struct grep_pat *p;
1337
1338         if (opt->extended)
1339                 return 0; /* punt for too complex stuff */
1340         if (opt->invert)
1341                 return 0;
1342         for (p = opt->pattern_list; p; p = p->next) {
1343                 if (p->token != GREP_PATTERN)
1344                         return 0; /* punt for "header only" and stuff */
1345         }
1346         return 1;
1347 }
1348
1349 static int look_ahead(struct grep_opt *opt,
1350                       unsigned long *left_p,
1351                       unsigned *lno_p,
1352                       char **bol_p)
1353 {
1354         unsigned lno = *lno_p;
1355         char *bol = *bol_p;
1356         struct grep_pat *p;
1357         char *sp, *last_bol;
1358         regoff_t earliest = -1;
1359
1360         for (p = opt->pattern_list; p; p = p->next) {
1361                 int hit;
1362                 regmatch_t m;
1363
1364                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1365                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1366                         continue;
1367                 if (earliest < 0 || m.rm_so < earliest)
1368                         earliest = m.rm_so;
1369         }
1370
1371         if (earliest < 0) {
1372                 *bol_p = bol + *left_p;
1373                 *left_p = 0;
1374                 return 1;
1375         }
1376         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1377                 ; /* find the beginning of the line */
1378         last_bol = sp;
1379
1380         for (sp = bol; sp < last_bol; sp++) {
1381                 if (*sp == '\n')
1382                         lno++;
1383         }
1384         *left_p -= last_bol - bol;
1385         *bol_p = last_bol;
1386         *lno_p = lno;
1387         return 0;
1388 }
1389
1390 static int fill_textconv_grep(struct userdiff_driver *driver,
1391                               struct grep_source *gs)
1392 {
1393         struct diff_filespec *df;
1394         char *buf;
1395         size_t size;
1396
1397         if (!driver || !driver->textconv)
1398                 return grep_source_load(gs);
1399
1400         /*
1401          * The textconv interface is intimately tied to diff_filespecs, so we
1402          * have to pretend to be one. If we could unify the grep_source
1403          * and diff_filespec structs, this mess could just go away.
1404          */
1405         df = alloc_filespec(gs->path);
1406         switch (gs->type) {
1407         case GREP_SOURCE_SHA1:
1408                 fill_filespec(df, gs->identifier, 1, 0100644);
1409                 break;
1410         case GREP_SOURCE_FILE:
1411                 fill_filespec(df, null_sha1, 0, 0100644);
1412                 break;
1413         default:
1414                 die("BUG: attempt to textconv something without a path?");
1415         }
1416
1417         /*
1418          * fill_textconv is not remotely thread-safe; it may load objects
1419          * behind the scenes, and it modifies the global diff tempfile
1420          * structure.
1421          */
1422         grep_read_lock();
1423         size = fill_textconv(driver, df, &buf);
1424         grep_read_unlock();
1425         free_filespec(df);
1426
1427         /*
1428          * The normal fill_textconv usage by the diff machinery would just keep
1429          * the textconv'd buf separate from the diff_filespec. But much of the
1430          * grep code passes around a grep_source and assumes that its "buf"
1431          * pointer is the beginning of the thing we are searching. So let's
1432          * install our textconv'd version into the grep_source, taking care not
1433          * to leak any existing buffer.
1434          */
1435         grep_source_clear_data(gs);
1436         gs->buf = buf;
1437         gs->size = size;
1438
1439         return 0;
1440 }
1441
1442 static int is_empty_line(const char *bol, const char *eol)
1443 {
1444         while (bol < eol && isspace(*bol))
1445                 bol++;
1446         return bol == eol;
1447 }
1448
1449 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1450 {
1451         char *bol;
1452         char *peek_bol = NULL;
1453         unsigned long left;
1454         unsigned lno = 1;
1455         unsigned last_hit = 0;
1456         int binary_match_only = 0;
1457         unsigned count = 0;
1458         int try_lookahead = 0;
1459         int show_function = 0;
1460         struct userdiff_driver *textconv = NULL;
1461         enum grep_context ctx = GREP_CONTEXT_HEAD;
1462         xdemitconf_t xecfg;
1463
1464         if (!opt->output)
1465                 opt->output = std_output;
1466
1467         if (opt->pre_context || opt->post_context || opt->file_break ||
1468             opt->funcbody) {
1469                 /* Show hunk marks, except for the first file. */
1470                 if (opt->last_shown)
1471                         opt->show_hunk_mark = 1;
1472                 /*
1473                  * If we're using threads then we can't easily identify
1474                  * the first file.  Always put hunk marks in that case
1475                  * and skip the very first one later in work_done().
1476                  */
1477                 if (opt->output != std_output)
1478                         opt->show_hunk_mark = 1;
1479         }
1480         opt->last_shown = 0;
1481
1482         if (opt->allow_textconv) {
1483                 grep_source_load_driver(gs);
1484                 /*
1485                  * We might set up the shared textconv cache data here, which
1486                  * is not thread-safe.
1487                  */
1488                 grep_attr_lock();
1489                 textconv = userdiff_get_textconv(gs->driver);
1490                 grep_attr_unlock();
1491         }
1492
1493         /*
1494          * We know the result of a textconv is text, so we only have to care
1495          * about binary handling if we are not using it.
1496          */
1497         if (!textconv) {
1498                 switch (opt->binary) {
1499                 case GREP_BINARY_DEFAULT:
1500                         if (grep_source_is_binary(gs))
1501                                 binary_match_only = 1;
1502                         break;
1503                 case GREP_BINARY_NOMATCH:
1504                         if (grep_source_is_binary(gs))
1505                                 return 0; /* Assume unmatch */
1506                         break;
1507                 case GREP_BINARY_TEXT:
1508                         break;
1509                 default:
1510                         die("BUG: unknown binary handling mode");
1511                 }
1512         }
1513
1514         memset(&xecfg, 0, sizeof(xecfg));
1515         opt->priv = &xecfg;
1516
1517         try_lookahead = should_lookahead(opt);
1518
1519         if (fill_textconv_grep(textconv, gs) < 0)
1520                 return 0;
1521
1522         bol = gs->buf;
1523         left = gs->size;
1524         while (left) {
1525                 char *eol, ch;
1526                 int hit;
1527
1528                 /*
1529                  * look_ahead() skips quickly to the line that possibly
1530                  * has the next hit; don't call it if we need to do
1531                  * something more than just skipping the current line
1532                  * in response to an unmatch for the current line.  E.g.
1533                  * inside a post-context window, we will show the current
1534                  * line as a context around the previous hit when it
1535                  * doesn't hit.
1536                  */
1537                 if (try_lookahead
1538                     && !(last_hit
1539                          && (show_function ||
1540                              lno <= last_hit + opt->post_context))
1541                     && look_ahead(opt, &left, &lno, &bol))
1542                         break;
1543                 eol = end_of_line(bol, &left);
1544                 ch = *eol;
1545                 *eol = 0;
1546
1547                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1548                         ctx = GREP_CONTEXT_BODY;
1549
1550                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1551                 *eol = ch;
1552
1553                 if (collect_hits)
1554                         goto next_line;
1555
1556                 /* "grep -v -e foo -e bla" should list lines
1557                  * that do not have either, so inversion should
1558                  * be done outside.
1559                  */
1560                 if (opt->invert)
1561                         hit = !hit;
1562                 if (opt->unmatch_name_only) {
1563                         if (hit)
1564                                 return 0;
1565                         goto next_line;
1566                 }
1567                 if (hit) {
1568                         count++;
1569                         if (opt->status_only)
1570                                 return 1;
1571                         if (opt->name_only) {
1572                                 show_name(opt, gs->name);
1573                                 return 1;
1574                         }
1575                         if (opt->count)
1576                                 goto next_line;
1577                         if (binary_match_only) {
1578                                 opt->output(opt, "Binary file ", 12);
1579                                 output_color(opt, gs->name, strlen(gs->name),
1580                                              opt->color_filename);
1581                                 opt->output(opt, " matches\n", 9);
1582                                 return 1;
1583                         }
1584                         /* Hit at this line.  If we haven't shown the
1585                          * pre-context lines, we would need to show them.
1586                          */
1587                         if (opt->pre_context || opt->funcbody)
1588                                 show_pre_context(opt, gs, bol, eol, lno);
1589                         else if (opt->funcname)
1590                                 show_funcname_line(opt, gs, bol, lno);
1591                         show_line(opt, bol, eol, gs->name, lno, ':');
1592                         last_hit = lno;
1593                         if (opt->funcbody)
1594                                 show_function = 1;
1595                         goto next_line;
1596                 }
1597                 if (show_function && (!peek_bol || peek_bol < bol)) {
1598                         unsigned long peek_left = left;
1599                         char *peek_eol = eol;
1600
1601                         /*
1602                          * Trailing empty lines are not interesting.
1603                          * Peek past them to see if they belong to the
1604                          * body of the current function.
1605                          */
1606                         peek_bol = bol;
1607                         while (is_empty_line(peek_bol, peek_eol)) {
1608                                 peek_bol = peek_eol + 1;
1609                                 peek_eol = end_of_line(peek_bol, &peek_left);
1610                         }
1611
1612                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1613                                 show_function = 0;
1614                 }
1615                 if (show_function ||
1616                     (last_hit && lno <= last_hit + opt->post_context)) {
1617                         /* If the last hit is within the post context,
1618                          * we need to show this line.
1619                          */
1620                         show_line(opt, bol, eol, gs->name, lno, '-');
1621                 }
1622
1623         next_line:
1624                 bol = eol + 1;
1625                 if (!left)
1626                         break;
1627                 left--;
1628                 lno++;
1629         }
1630
1631         if (collect_hits)
1632                 return 0;
1633
1634         if (opt->status_only)
1635                 return 0;
1636         if (opt->unmatch_name_only) {
1637                 /* We did not see any hit, so we want to show this */
1638                 show_name(opt, gs->name);
1639                 return 1;
1640         }
1641
1642         xdiff_clear_find_func(&xecfg);
1643         opt->priv = NULL;
1644
1645         /* NEEDSWORK:
1646          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1647          * which feels mostly useless but sometimes useful.  Maybe
1648          * make it another option?  For now suppress them.
1649          */
1650         if (opt->count && count) {
1651                 char buf[32];
1652                 if (opt->pathname) {
1653                         output_color(opt, gs->name, strlen(gs->name),
1654                                      opt->color_filename);
1655                         output_sep(opt, ':');
1656                 }
1657                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1658                 opt->output(opt, buf, strlen(buf));
1659                 return 1;
1660         }
1661         return !!last_hit;
1662 }
1663
1664 static void clr_hit_marker(struct grep_expr *x)
1665 {
1666         /* All-hit markers are meaningful only at the very top level
1667          * OR node.
1668          */
1669         while (1) {
1670                 x->hit = 0;
1671                 if (x->node != GREP_NODE_OR)
1672                         return;
1673                 x->u.binary.left->hit = 0;
1674                 x = x->u.binary.right;
1675         }
1676 }
1677
1678 static int chk_hit_marker(struct grep_expr *x)
1679 {
1680         /* Top level nodes have hit markers.  See if they all are hits */
1681         while (1) {
1682                 if (x->node != GREP_NODE_OR)
1683                         return x->hit;
1684                 if (!x->u.binary.left->hit)
1685                         return 0;
1686                 x = x->u.binary.right;
1687         }
1688 }
1689
1690 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1691 {
1692         /*
1693          * we do not have to do the two-pass grep when we do not check
1694          * buffer-wide "all-match".
1695          */
1696         if (!opt->all_match)
1697                 return grep_source_1(opt, gs, 0);
1698
1699         /* Otherwise the toplevel "or" terms hit a bit differently.
1700          * We first clear hit markers from them.
1701          */
1702         clr_hit_marker(opt->pattern_expression);
1703         grep_source_1(opt, gs, 1);
1704
1705         if (!chk_hit_marker(opt->pattern_expression))
1706                 return 0;
1707
1708         return grep_source_1(opt, gs, 0);
1709 }
1710
1711 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1712 {
1713         struct grep_source gs;
1714         int r;
1715
1716         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1717         gs.buf = buf;
1718         gs.size = size;
1719
1720         r = grep_source(opt, &gs);
1721
1722         grep_source_clear(&gs);
1723         return r;
1724 }
1725
1726 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1727                       const char *name, const char *path,
1728                       const void *identifier)
1729 {
1730         gs->type = type;
1731         gs->name = xstrdup_or_null(name);
1732         gs->path = xstrdup_or_null(path);
1733         gs->buf = NULL;
1734         gs->size = 0;
1735         gs->driver = NULL;
1736
1737         switch (type) {
1738         case GREP_SOURCE_FILE:
1739                 gs->identifier = xstrdup(identifier);
1740                 break;
1741         case GREP_SOURCE_SUBMODULE:
1742                 if (!identifier) {
1743                         gs->identifier = NULL;
1744                         break;
1745                 }
1746                 /*
1747                  * FALL THROUGH
1748                  * If the identifier is non-NULL (in the submodule case) it
1749                  * will be a SHA1 that needs to be copied.
1750                  */
1751         case GREP_SOURCE_SHA1:
1752                 gs->identifier = xmalloc(20);
1753                 hashcpy(gs->identifier, identifier);
1754                 break;
1755         case GREP_SOURCE_BUF:
1756                 gs->identifier = NULL;
1757                 break;
1758         }
1759 }
1760
1761 void grep_source_clear(struct grep_source *gs)
1762 {
1763         free(gs->name);
1764         gs->name = NULL;
1765         free(gs->path);
1766         gs->path = NULL;
1767         free(gs->identifier);
1768         gs->identifier = NULL;
1769         grep_source_clear_data(gs);
1770 }
1771
1772 void grep_source_clear_data(struct grep_source *gs)
1773 {
1774         switch (gs->type) {
1775         case GREP_SOURCE_FILE:
1776         case GREP_SOURCE_SHA1:
1777         case GREP_SOURCE_SUBMODULE:
1778                 free(gs->buf);
1779                 gs->buf = NULL;
1780                 gs->size = 0;
1781                 break;
1782         case GREP_SOURCE_BUF:
1783                 /* leave user-provided buf intact */
1784                 break;
1785         }
1786 }
1787
1788 static int grep_source_load_sha1(struct grep_source *gs)
1789 {
1790         enum object_type type;
1791
1792         grep_read_lock();
1793         gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1794         grep_read_unlock();
1795
1796         if (!gs->buf)
1797                 return error(_("'%s': unable to read %s"),
1798                              gs->name,
1799                              sha1_to_hex(gs->identifier));
1800         return 0;
1801 }
1802
1803 static int grep_source_load_file(struct grep_source *gs)
1804 {
1805         const char *filename = gs->identifier;
1806         struct stat st;
1807         char *data;
1808         size_t size;
1809         int i;
1810
1811         if (lstat(filename, &st) < 0) {
1812         err_ret:
1813                 if (errno != ENOENT)
1814                         error_errno(_("failed to stat '%s'"), filename);
1815                 return -1;
1816         }
1817         if (!S_ISREG(st.st_mode))
1818                 return -1;
1819         size = xsize_t(st.st_size);
1820         i = open(filename, O_RDONLY);
1821         if (i < 0)
1822                 goto err_ret;
1823         data = xmallocz(size);
1824         if (st.st_size != read_in_full(i, data, size)) {
1825                 error_errno(_("'%s': short read"), filename);
1826                 close(i);
1827                 free(data);
1828                 return -1;
1829         }
1830         close(i);
1831
1832         gs->buf = data;
1833         gs->size = size;
1834         return 0;
1835 }
1836
1837 static int grep_source_load(struct grep_source *gs)
1838 {
1839         if (gs->buf)
1840                 return 0;
1841
1842         switch (gs->type) {
1843         case GREP_SOURCE_FILE:
1844                 return grep_source_load_file(gs);
1845         case GREP_SOURCE_SHA1:
1846                 return grep_source_load_sha1(gs);
1847         case GREP_SOURCE_BUF:
1848                 return gs->buf ? 0 : -1;
1849         case GREP_SOURCE_SUBMODULE:
1850                 break;
1851         }
1852         die("BUG: invalid grep_source type to load");
1853 }
1854
1855 void grep_source_load_driver(struct grep_source *gs)
1856 {
1857         if (gs->driver)
1858                 return;
1859
1860         grep_attr_lock();
1861         if (gs->path)
1862                 gs->driver = userdiff_find_by_path(gs->path);
1863         if (!gs->driver)
1864                 gs->driver = userdiff_find_by_name("default");
1865         grep_attr_unlock();
1866 }
1867
1868 static int grep_source_is_binary(struct grep_source *gs)
1869 {
1870         grep_source_load_driver(gs);
1871         if (gs->driver->binary != -1)
1872                 return gs->driver->binary;
1873
1874         if (!grep_source_load(gs))
1875                 return buffer_is_binary(gs->buf, gs->size);
1876
1877         return 0;
1878 }