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