grep.c

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