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