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 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
  16 {
  17         fwrite(buf, size, 1, stdout);
  18 }
  19
  20 /*
  21  * Initialize the grep_defaults template with hardcoded defaults.
  22  * We could let the compiler do this, but without C99 initializers
  23  * the code gets unwieldy and unreadable, so...
  24  */
  25 void init_grep_defaults(void)
  26 {
  27         struct grep_opt *opt = &grep_defaults;
  28         static int run_once;
  29
  30         if (run_once)
  31                 return;
  32         run_once++;
  33
  34         memset(opt, 0, sizeof(*opt));
  35         opt->relative = 1;
  36         opt->pathname = 1;
  37         opt->regflags = REG_NEWLINE;
  38         opt->max_depth = -1;
  39         opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  40         opt->extended_regexp_option = 0;
  41         color_set(opt->color_context, "");
  42         color_set(opt->color_filename, "");
  43         color_set(opt->color_function, "");
  44         color_set(opt->color_lineno, "");
  45         color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
  46         color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
  47         color_set(opt->color_selected, "");
  48         color_set(opt->color_sep, GIT_COLOR_CYAN);
  49         opt->color = -1;
  50         opt->output = std_output;
  51 }
  52
  53 static int parse_pattern_type_arg(const char *opt, const char *arg)
  54 {
  55         if (!strcmp(arg, "default"))
  56                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  57         else if (!strcmp(arg, "basic"))
  58                 return GREP_PATTERN_TYPE_BRE;
  59         else if (!strcmp(arg, "extended"))
  60                 return GREP_PATTERN_TYPE_ERE;
  61         else if (!strcmp(arg, "fixed"))
  62                 return GREP_PATTERN_TYPE_FIXED;
  63         else if (!strcmp(arg, "perl"))
  64                 return GREP_PATTERN_TYPE_PCRE;
  65         die("bad %s argument: %s", opt, arg);
  66 }
  67
  68 /*
  69  * Read the configuration file once and store it in
  70  * the grep_defaults template.
  71  */
  72 int grep_config(const char *var, const char *value, void *cb)
  73 {
  74         struct grep_opt *opt = &grep_defaults;
  75         char *color = NULL;
  76
  77         if (userdiff_config(var, value) < 0)
  78                 return -1;
  79
  80         if (!strcmp(var, "grep.extendedregexp")) {
  81                 if (git_config_bool(var, value))
  82                         opt->extended_regexp_option = 1;
  83                 else
  84                         opt->extended_regexp_option = 0;
  85                 return 0;
  86         }
  87
  88         if (!strcmp(var, "grep.patterntype")) {
  89                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
  90                 return 0;
  91         }
  92
  93         if (!strcmp(var, "grep.linenumber")) {
  94                 opt->linenum = git_config_bool(var, value);
  95                 return 0;
  96         }
  97
  98         if (!strcmp(var, "grep.fullname")) {
  99                 opt->relative = !git_config_bool(var, value);
 100                 return 0;
 101         }
 102
 103         if (!strcmp(var, "color.grep"))
 104                 opt->color = git_config_colorbool(var, value);
 105         else if (!strcmp(var, "color.grep.context"))
 106                 color = opt->color_context;
 107         else if (!strcmp(var, "color.grep.filename"))
 108                 color = opt->color_filename;
 109         else if (!strcmp(var, "color.grep.function"))
 110                 color = opt->color_function;
 111         else if (!strcmp(var, "color.grep.linenumber"))
 112                 color = opt->color_lineno;
 113         else if (!strcmp(var, "color.grep.matchcontext"))
 114                 color = opt->color_match_context;
 115         else if (!strcmp(var, "color.grep.matchselected"))
 116                 color = opt->color_match_selected;
 117         else if (!strcmp(var, "color.grep.selected"))
 118                 color = opt->color_selected;
 119         else if (!strcmp(var, "color.grep.separator"))
 120                 color = opt->color_sep;
 121         else if (!strcmp(var, "color.grep.match")) {
 122                 int rc = 0;
 123                 if (!value)
 124                         return config_error_nonbool(var);
 125                 rc |= color_parse(value, opt->color_match_context);
 126                 rc |= color_parse(value, opt->color_match_selected);
 127                 return rc;
 128         }
 129
 130         if (color) {
 131                 if (!value)
 132                         return config_error_nonbool(var);
 133                 return color_parse(value, color);
 134         }
 135         return 0;
 136 }
 137
 138 /*
 139  * Initialize one instance of grep_opt and copy the
 140  * default values from the template we read the configuration
 141  * information in an earlier call to git_config(grep_config).
 142  */
 143 void grep_init(struct grep_opt *opt, const char *prefix)
 144 {
 145         struct grep_opt *def = &grep_defaults;
 146
 147         memset(opt, 0, sizeof(*opt));
 148         opt->prefix = prefix;
 149         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 150         opt->pattern_tail = &opt->pattern_list;
 151         opt->header_tail = &opt->header_list;
 152
 153         opt->color = def->color;
 154         opt->extended_regexp_option = def->extended_regexp_option;
 155         opt->pattern_type_option = def->pattern_type_option;
 156         opt->linenum = def->linenum;
 157         opt->max_depth = def->max_depth;
 158         opt->pathname = def->pathname;
 159         opt->regflags = def->regflags;
 160         opt->relative = def->relative;
 161         opt->output = def->output;
 162
 163         color_set(opt->color_context, def->color_context);
 164         color_set(opt->color_filename, def->color_filename);
 165         color_set(opt->color_function, def->color_function);
 166         color_set(opt->color_lineno, def->color_lineno);
 167         color_set(opt->color_match_context, def->color_match_context);
 168         color_set(opt->color_match_selected, def->color_match_selected);
 169         color_set(opt->color_selected, def->color_selected);
 170         color_set(opt->color_sep, def->color_sep);
 171 }
 172
 173 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 174 {
 175         switch (pattern_type) {
 176         case GREP_PATTERN_TYPE_UNSPECIFIED:
 177                 /* fall through */
 178
 179         case GREP_PATTERN_TYPE_BRE:
 180                 opt->fixed = 0;
 181                 opt->pcre = 0;
 182                 opt->regflags &= ~REG_EXTENDED;
 183                 break;
 184
 185         case GREP_PATTERN_TYPE_ERE:
 186                 opt->fixed = 0;
 187                 opt->pcre = 0;
 188                 opt->regflags |= REG_EXTENDED;
 189                 break;
 190
 191         case GREP_PATTERN_TYPE_FIXED:
 192                 opt->fixed = 1;
 193                 opt->pcre = 0;
 194                 opt->regflags &= ~REG_EXTENDED;
 195                 break;
 196
 197         case GREP_PATTERN_TYPE_PCRE:
 198                 opt->fixed = 0;
 199                 opt->pcre = 1;
 200                 opt->regflags &= ~REG_EXTENDED;
 201                 break;
 202         }
 203 }
 204
 205 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 206 {
 207         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 208                 grep_set_pattern_type_option(pattern_type, opt);
 209         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 210                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 211         else if (opt->extended_regexp_option)
 212                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 213 }
 214
 215 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 216                                         const char *origin, int no,
 217                                         enum grep_pat_token t,
 218                                         enum grep_header_field field)
 219 {
 220         struct grep_pat *p = xcalloc(1, sizeof(*p));
 221         p->pattern = xmemdupz(pat, patlen);
 222         p->patternlen = patlen;
 223         p->origin = origin;
 224         p->no = no;
 225         p->token = t;
 226         p->field = field;
 227         return p;
 228 }
 229
 230 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 231 {
 232         **tail = p;
 233         *tail = &p->next;
 234         p->next = NULL;
 235
 236         switch (p->token) {
 237         case GREP_PATTERN: /* atom */
 238         case GREP_PATTERN_HEAD:
 239         case GREP_PATTERN_BODY:
 240                 for (;;) {
 241                         struct grep_pat *new_pat;
 242                         size_t len = 0;
 243                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 244                         while (++len <= p->patternlen) {
 245                                 if (*(--cp) == '\n') {
 246                                         nl = cp;
 247                                         break;
 248                                 }
 249                         }
 250                         if (!nl)
 251                                 break;
 252                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 253                                                   p->no, p->token, p->field);
 254                         new_pat->next = p->next;
 255                         if (!p->next)
 256                                 *tail = &new_pat->next;
 257                         p->next = new_pat;
 258                         *nl = '\0';
 259                         p->patternlen -= len;
 260                 }
 261                 break;
 262         default:
 263                 break;
 264         }
 265 }
 266
 267 void append_header_grep_pattern(struct grep_opt *opt,
 268                                 enum grep_header_field field, const char *pat)
 269 {
 270         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 271                                              GREP_PATTERN_HEAD, field);
 272         if (field == GREP_HEADER_REFLOG)
 273                 opt->use_reflog_filter = 1;
 274         do_append_grep_pat(&opt->header_tail, p);
 275 }
 276
 277 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 278                          const char *origin, int no, enum grep_pat_token t)
 279 {
 280         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 281 }
 282
 283 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 284                      const char *origin, int no, enum grep_pat_token t)
 285 {
 286         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 287         do_append_grep_pat(&opt->pattern_tail, p);
 288 }
 289
 290 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 291 {
 292         struct grep_pat *pat;
 293         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 294         *ret = *opt;
 295
 296         ret->pattern_list = NULL;
 297         ret->pattern_tail = &ret->pattern_list;
 298
 299         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 300         {
 301                 if(pat->token == GREP_PATTERN_HEAD)
 302                         append_header_grep_pattern(ret, pat->field,
 303                                                    pat->pattern);
 304                 else
 305                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 306                                         pat->origin, pat->no, pat->token);
 307         }
 308
 309         return ret;
 310 }
 311
 312 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 313                 const char *error)
 314 {
 315         char where[1024];
 316
 317         if (p->no)
 318                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 319         else if (p->origin)
 320                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 321         else
 322                 where[0] = 0;
 323
 324         die("%s'%s': %s", where, p->pattern, error);
 325 }
 326
 327 #ifdef USE_LIBPCRE
 328 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 329 {
 330         const char *error;
 331         int erroffset;
 332         int options = PCRE_MULTILINE;
 333
 334         if (opt->ignore_case) {
 335                 if (has_non_ascii(p->pattern))
 336                         p->pcre_tables = pcre_maketables();
 337                 options |= PCRE_CASELESS;
 338         }
 339         if (is_utf8_locale() && has_non_ascii(p->pattern))
 340                 options |= PCRE_UTF8;
 341
 342         p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 343                                       p->pcre_tables);
 344         if (!p->pcre_regexp)
 345                 compile_regexp_failed(p, error);
 346
 347         p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 348         if (!p->pcre_extra_info && error)
 349                 die("%s", error);
 350 }
 351
 352 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 353                 regmatch_t *match, int eflags)
 354 {
 355         int ovector[30], ret, flags = 0;
 356
 357         if (eflags & REG_NOTBOL)
 358                 flags |= PCRE_NOTBOL;
 359
 360         ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 361                         0, flags, ovector, ARRAY_SIZE(ovector));
 362         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 363                 die("pcre_exec failed with error code %d", ret);
 364         if (ret > 0) {
 365                 ret = 0;
 366                 match->rm_so = ovector[0];
 367                 match->rm_eo = ovector[1];
 368         }
 369
 370         return ret;
 371 }
 372
 373 static void free_pcre_regexp(struct grep_pat *p)
 374 {
 375         pcre_free(p->pcre_regexp);
 376         pcre_free(p->pcre_extra_info);
 377         pcre_free((void *)p->pcre_tables);
 378 }
 379 #else /* !USE_LIBPCRE */
 380 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 381 {
 382         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 383 }
 384
 385 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 386                 regmatch_t *match, int eflags)
 387 {
 388         return 1;
 389 }
 390
 391 static void free_pcre_regexp(struct grep_pat *p)
 392 {
 393 }
 394 #endif /* !USE_LIBPCRE */
 395
 396 static int is_fixed(const char *s, size_t len)
 397 {
 398         size_t i;
 399
 400         /* regcomp cannot accept patterns with NULs so we
 401          * consider any pattern containing a NUL fixed.
 402          */
 403         if (memchr(s, 0, len))
 404                 return 1;
 405
 406         for (i = 0; i < len; i++) {
 407                 if (is_regex_special(s[i]))
 408                         return 0;
 409         }
 410
 411         return 1;
 412 }
 413
 414 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 415 {
 416         struct strbuf sb = STRBUF_INIT;
 417         int err;
 418         int regflags;
 419
 420         basic_regex_quote_buf(&sb, p->pattern);
 421         regflags = opt->regflags & ~REG_EXTENDED;
 422         if (opt->ignore_case)
 423                 regflags |= REG_ICASE;
 424         err = regcomp(&p->regexp, sb.buf, regflags);
 425         if (opt->debug)
 426                 fprintf(stderr, "fixed %s\n", sb.buf);
 427         strbuf_release(&sb);
 428         if (err) {
 429                 char errbuf[1024];
 430                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 431                 regfree(&p->regexp);
 432                 compile_regexp_failed(p, errbuf);
 433         }
 434 }
 435
 436 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 437 {
 438         int icase, ascii_only;
 439         int err;
 440
 441         p->word_regexp = opt->word_regexp;
 442         p->ignore_case = opt->ignore_case;
 443         icase          = opt->regflags & REG_ICASE || p->ignore_case;
 444         ascii_only     = !has_non_ascii(p->pattern);
 445
 446         /*
 447          * Even when -F (fixed) asks us to do a non-regexp search, we
 448          * may not be able to correctly case-fold when -i
 449          * (ignore-case) is asked (in which case, we'll synthesize a
 450          * regexp to match the pattern that matches regexp special
 451          * characters literally, while ignoring case differences).  On
 452          * the other hand, even without -F, if the pattern does not
 453          * have any regexp special characters and there is no need for
 454          * case-folding search, we can internally turn it into a
 455          * simple string match using kws.  p->fixed tells us if we
 456          * want to use kws.
 457          */
 458         if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 459                 p->fixed = !icase || ascii_only;
 460         else
 461                 p->fixed = 0;
 462
 463         if (p->fixed) {
 464                 p->kws = kwsalloc(icase ? tolower_trans_tbl : NULL);
 465                 kwsincr(p->kws, p->pattern, p->patternlen);
 466                 kwsprep(p->kws);
 467                 return;
 468         } else if (opt->fixed) {
 469                 /*
 470                  * We come here when the pattern has the non-ascii
 471                  * characters we cannot case-fold, and asked to
 472                  * ignore-case.
 473                  */
 474                 compile_fixed_regexp(p, opt);
 475                 return;
 476         }
 477
 478         if (opt->pcre) {
 479                 compile_pcre_regexp(p, opt);
 480                 return;
 481         }
 482
 483         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 484         if (err) {
 485                 char errbuf[1024];
 486                 regerror(err, &p->regexp, errbuf, 1024);
 487                 regfree(&p->regexp);
 488                 compile_regexp_failed(p, errbuf);
 489         }
 490 }
 491
 492 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 493 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 494 {
 495         struct grep_pat *p;
 496         struct grep_expr *x;
 497
 498         p = *list;
 499         if (!p)
 500                 return NULL;
 501         switch (p->token) {
 502         case GREP_PATTERN: /* atom */
 503         case GREP_PATTERN_HEAD:
 504         case GREP_PATTERN_BODY:
 505                 x = xcalloc(1, sizeof (struct grep_expr));
 506                 x->node = GREP_NODE_ATOM;
 507                 x->u.atom = p;
 508                 *list = p->next;
 509                 return x;
 510         case GREP_OPEN_PAREN:
 511                 *list = p->next;
 512                 x = compile_pattern_or(list);
 513                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 514                         die("unmatched parenthesis");
 515                 *list = (*list)->next;
 516                 return x;
 517         default:
 518                 return NULL;
 519         }
 520 }
 521
 522 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 523 {
 524         struct grep_pat *p;
 525         struct grep_expr *x;
 526
 527         p = *list;
 528         if (!p)
 529                 return NULL;
 530         switch (p->token) {
 531         case GREP_NOT:
 532                 if (!p->next)
 533                         die("--not not followed by pattern expression");
 534                 *list = p->next;
 535                 x = xcalloc(1, sizeof (struct grep_expr));
 536                 x->node = GREP_NODE_NOT;
 537                 x->u.unary = compile_pattern_not(list);
 538                 if (!x->u.unary)
 539                         die("--not followed by non pattern expression");
 540                 return x;
 541         default:
 542                 return compile_pattern_atom(list);
 543         }
 544 }
 545
 546 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 547 {
 548         struct grep_pat *p;
 549         struct grep_expr *x, *y, *z;
 550
 551         x = compile_pattern_not(list);
 552         p = *list;
 553         if (p && p->token == GREP_AND) {
 554                 if (!p->next)
 555                         die("--and not followed by pattern expression");
 556                 *list = p->next;
 557                 y = compile_pattern_and(list);
 558                 if (!y)
 559                         die("--and not followed by pattern expression");
 560                 z = xcalloc(1, sizeof (struct grep_expr));
 561                 z->node = GREP_NODE_AND;
 562                 z->u.binary.left = x;
 563                 z->u.binary.right = y;
 564                 return z;
 565         }
 566         return x;
 567 }
 568
 569 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 570 {
 571         struct grep_pat *p;
 572         struct grep_expr *x, *y, *z;
 573
 574         x = compile_pattern_and(list);
 575         p = *list;
 576         if (x && p && p->token != GREP_CLOSE_PAREN) {
 577                 y = compile_pattern_or(list);
 578                 if (!y)
 579                         die("not a pattern expression %s", p->pattern);
 580                 z = xcalloc(1, sizeof (struct grep_expr));
 581                 z->node = GREP_NODE_OR;
 582                 z->u.binary.left = x;
 583                 z->u.binary.right = y;
 584                 return z;
 585         }
 586         return x;
 587 }
 588
 589 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 590 {
 591         return compile_pattern_or(list);
 592 }
 593
 594 static void indent(int in)
 595 {
 596         while (in-- > 0)
 597                 fputc(' ', stderr);
 598 }
 599
 600 static void dump_grep_pat(struct grep_pat *p)
 601 {
 602         switch (p->token) {
 603         case GREP_AND: fprintf(stderr, "*and*"); break;
 604         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 605         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 606         case GREP_NOT: fprintf(stderr, "*not*"); break;
 607         case GREP_OR: fprintf(stderr, "*or*"); break;
 608
 609         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 610         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 611         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 612         }
 613
 614         switch (p->token) {
 615         default: break;
 616         case GREP_PATTERN_HEAD:
 617                 fprintf(stderr, "<head %d>", p->field); break;
 618         case GREP_PATTERN_BODY:
 619                 fprintf(stderr, "<body>"); break;
 620         }
 621         switch (p->token) {
 622         default: break;
 623         case GREP_PATTERN_HEAD:
 624         case GREP_PATTERN_BODY:
 625         case GREP_PATTERN:
 626                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 627                 break;
 628         }
 629         fputc('\n', stderr);
 630 }
 631
 632 static void dump_grep_expression_1(struct grep_expr *x, int in)
 633 {
 634         indent(in);
 635         switch (x->node) {
 636         case GREP_NODE_TRUE:
 637                 fprintf(stderr, "true\n");
 638                 break;
 639         case GREP_NODE_ATOM:
 640                 dump_grep_pat(x->u.atom);
 641                 break;
 642         case GREP_NODE_NOT:
 643                 fprintf(stderr, "(not\n");
 644                 dump_grep_expression_1(x->u.unary, in+1);
 645                 indent(in);
 646                 fprintf(stderr, ")\n");
 647                 break;
 648         case GREP_NODE_AND:
 649                 fprintf(stderr, "(and\n");
 650                 dump_grep_expression_1(x->u.binary.left, in+1);
 651                 dump_grep_expression_1(x->u.binary.right, in+1);
 652                 indent(in);
 653                 fprintf(stderr, ")\n");
 654                 break;
 655         case GREP_NODE_OR:
 656                 fprintf(stderr, "(or\n");
 657                 dump_grep_expression_1(x->u.binary.left, in+1);
 658                 dump_grep_expression_1(x->u.binary.right, in+1);
 659                 indent(in);
 660                 fprintf(stderr, ")\n");
 661                 break;
 662         }
 663 }
 664
 665 static void dump_grep_expression(struct grep_opt *opt)
 666 {
 667         struct grep_expr *x = opt->pattern_expression;
 668
 669         if (opt->all_match)
 670                 fprintf(stderr, "[all-match]\n");
 671         dump_grep_expression_1(x, 0);
 672         fflush(NULL);
 673 }
 674
 675 static struct grep_expr *grep_true_expr(void)
 676 {
 677         struct grep_expr *z = xcalloc(1, sizeof(*z));
 678         z->node = GREP_NODE_TRUE;
 679         return z;
 680 }
 681
 682 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 683 {
 684         struct grep_expr *z = xcalloc(1, sizeof(*z));
 685         z->node = GREP_NODE_OR;
 686         z->u.binary.left = left;
 687         z->u.binary.right = right;
 688         return z;
 689 }
 690
 691 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 692 {
 693         struct grep_pat *p;
 694         struct grep_expr *header_expr;
 695         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 696         enum grep_header_field fld;
 697
 698         if (!opt->header_list)
 699                 return NULL;
 700
 701         for (p = opt->header_list; p; p = p->next) {
 702                 if (p->token != GREP_PATTERN_HEAD)
 703                         die("BUG: a non-header pattern in grep header list.");
 704                 if (p->field < GREP_HEADER_FIELD_MIN ||
 705                     GREP_HEADER_FIELD_MAX <= p->field)
 706                         die("BUG: unknown header field %d", p->field);
 707                 compile_regexp(p, opt);
 708         }
 709
 710         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 711                 header_group[fld] = NULL;
 712
 713         for (p = opt->header_list; p; p = p->next) {
 714                 struct grep_expr *h;
 715                 struct grep_pat *pp = p;
 716
 717                 h = compile_pattern_atom(&pp);
 718                 if (!h || pp != p->next)
 719                         die("BUG: malformed header expr");
 720                 if (!header_group[p->field]) {
 721                         header_group[p->field] = h;
 722                         continue;
 723                 }
 724                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 725         }
 726
 727         header_expr = NULL;
 728
 729         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 730                 if (!header_group[fld])
 731                         continue;
 732                 if (!header_expr)
 733                         header_expr = grep_true_expr();
 734                 header_expr = grep_or_expr(header_group[fld], header_expr);
 735         }
 736         return header_expr;
 737 }
 738
 739 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 740 {
 741         struct grep_expr *z = x;
 742
 743         while (x) {
 744                 assert(x->node == GREP_NODE_OR);
 745                 if (x->u.binary.right &&
 746                     x->u.binary.right->node == GREP_NODE_TRUE) {
 747                         x->u.binary.right = y;
 748                         break;
 749                 }
 750                 x = x->u.binary.right;
 751         }
 752         return z;
 753 }
 754
 755 static void compile_grep_patterns_real(struct grep_opt *opt)
 756 {
 757         struct grep_pat *p;
 758         struct grep_expr *header_expr = prep_header_patterns(opt);
 759
 760         for (p = opt->pattern_list; p; p = p->next) {
 761                 switch (p->token) {
 762                 case GREP_PATTERN: /* atom */
 763                 case GREP_PATTERN_HEAD:
 764                 case GREP_PATTERN_BODY:
 765                         compile_regexp(p, opt);
 766                         break;
 767                 default:
 768                         opt->extended = 1;
 769                         break;
 770                 }
 771         }
 772
 773         if (opt->all_match || header_expr)
 774                 opt->extended = 1;
 775         else if (!opt->extended && !opt->debug)
 776                 return;
 777
 778         p = opt->pattern_list;
 779         if (p)
 780                 opt->pattern_expression = compile_pattern_expr(&p);
 781         if (p)
 782                 die("incomplete pattern expression: %s", p->pattern);
 783
 784         if (!header_expr)
 785                 return;
 786
 787         if (!opt->pattern_expression)
 788                 opt->pattern_expression = header_expr;
 789         else if (opt->all_match)
 790                 opt->pattern_expression = grep_splice_or(header_expr,
 791                                                          opt->pattern_expression);
 792         else
 793                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 794                                                        header_expr);
 795         opt->all_match = 1;
 796 }
 797
 798 void compile_grep_patterns(struct grep_opt *opt)
 799 {
 800         compile_grep_patterns_real(opt);
 801         if (opt->debug)
 802                 dump_grep_expression(opt);
 803 }
 804
 805 static void free_pattern_expr(struct grep_expr *x)
 806 {
 807         switch (x->node) {
 808         case GREP_NODE_TRUE:
 809         case GREP_NODE_ATOM:
 810                 break;
 811         case GREP_NODE_NOT:
 812                 free_pattern_expr(x->u.unary);
 813                 break;
 814         case GREP_NODE_AND:
 815         case GREP_NODE_OR:
 816                 free_pattern_expr(x->u.binary.left);
 817                 free_pattern_expr(x->u.binary.right);
 818                 break;
 819         }
 820         free(x);
 821 }
 822
 823 void free_grep_patterns(struct grep_opt *opt)
 824 {
 825         struct grep_pat *p, *n;
 826
 827         for (p = opt->pattern_list; p; p = n) {
 828                 n = p->next;
 829                 switch (p->token) {
 830                 case GREP_PATTERN: /* atom */
 831                 case GREP_PATTERN_HEAD:
 832                 case GREP_PATTERN_BODY:
 833                         if (p->kws)
 834                                 kwsfree(p->kws);
 835                         else if (p->pcre_regexp)
 836                                 free_pcre_regexp(p);
 837                         else
 838                                 regfree(&p->regexp);
 839                         free(p->pattern);
 840                         break;
 841                 default:
 842                         break;
 843                 }
 844                 free(p);
 845         }
 846
 847         if (!opt->extended)
 848                 return;
 849         free_pattern_expr(opt->pattern_expression);
 850 }
 851
 852 static char *end_of_line(char *cp, unsigned long *left)
 853 {
 854         unsigned long l = *left;
 855         while (l && *cp != '\n') {
 856                 l--;
 857                 cp++;
 858         }
 859         *left = l;
 860         return cp;
 861 }
 862
 863 static int word_char(char ch)
 864 {
 865         return isalnum(ch) || ch == '_';
 866 }
 867
 868 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 869                          const char *color)
 870 {
 871         if (want_color(opt->color) && color && color[0]) {
 872                 opt->output(opt, color, strlen(color));
 873                 opt->output(opt, data, size);
 874                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 875         } else
 876                 opt->output(opt, data, size);
 877 }
 878
 879 static void output_sep(struct grep_opt *opt, char sign)
 880 {
 881         if (opt->null_following_name)
 882                 opt->output(opt, "\0", 1);
 883         else
 884                 output_color(opt, &sign, 1, opt->color_sep);
 885 }
 886
 887 static void show_name(struct grep_opt *opt, const char *name)
 888 {
 889         output_color(opt, name, strlen(name), opt->color_filename);
 890         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 891 }
 892
 893 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 894                     regmatch_t *match)
 895 {
 896         struct kwsmatch kwsm;
 897         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 898         if (offset == -1) {
 899                 match->rm_so = match->rm_eo = -1;
 900                 return REG_NOMATCH;
 901         } else {
 902                 match->rm_so = offset;
 903                 match->rm_eo = match->rm_so + kwsm.size[0];
 904                 return 0;
 905         }
 906 }
 907
 908 static int patmatch(struct grep_pat *p, char *line, char *eol,
 909                     regmatch_t *match, int eflags)
 910 {
 911         int hit;
 912
 913         if (p->fixed)
 914                 hit = !fixmatch(p, line, eol, match);
 915         else if (p->pcre_regexp)
 916                 hit = !pcrematch(p, line, eol, match, eflags);
 917         else
 918                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
 919                                    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                 xsnprintf(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 int fill_textconv_grep(struct userdiff_driver *driver,
1390                               struct grep_source *gs)
1391 {
1392         struct diff_filespec *df;
1393         char *buf;
1394         size_t size;
1395
1396         if (!driver || !driver->textconv)
1397                 return grep_source_load(gs);
1398
1399         /*
1400          * The textconv interface is intimately tied to diff_filespecs, so we
1401          * have to pretend to be one. If we could unify the grep_source
1402          * and diff_filespec structs, this mess could just go away.
1403          */
1404         df = alloc_filespec(gs->path);
1405         switch (gs->type) {
1406         case GREP_SOURCE_SHA1:
1407                 fill_filespec(df, gs->identifier, 1, 0100644);
1408                 break;
1409         case GREP_SOURCE_FILE:
1410                 fill_filespec(df, null_sha1, 0, 0100644);
1411                 break;
1412         default:
1413                 die("BUG: attempt to textconv something without a path?");
1414         }
1415
1416         /*
1417          * fill_textconv is not remotely thread-safe; it may load objects
1418          * behind the scenes, and it modifies the global diff tempfile
1419          * structure.
1420          */
1421         grep_read_lock();
1422         size = fill_textconv(driver, df, &buf);
1423         grep_read_unlock();
1424         free_filespec(df);
1425
1426         /*
1427          * The normal fill_textconv usage by the diff machinery would just keep
1428          * the textconv'd buf separate from the diff_filespec. But much of the
1429          * grep code passes around a grep_source and assumes that its "buf"
1430          * pointer is the beginning of the thing we are searching. So let's
1431          * install our textconv'd version into the grep_source, taking care not
1432          * to leak any existing buffer.
1433          */
1434         grep_source_clear_data(gs);
1435         gs->buf = buf;
1436         gs->size = size;
1437
1438         return 0;
1439 }
1440
1441 static int is_empty_line(const char *bol, const char *eol)
1442 {
1443         while (bol < eol && isspace(*bol))
1444                 bol++;
1445         return bol == eol;
1446 }
1447
1448 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1449 {
1450         char *bol;
1451         char *peek_bol = NULL;
1452         unsigned long left;
1453         unsigned lno = 1;
1454         unsigned last_hit = 0;
1455         int binary_match_only = 0;
1456         unsigned count = 0;
1457         int try_lookahead = 0;
1458         int show_function = 0;
1459         struct userdiff_driver *textconv = NULL;
1460         enum grep_context ctx = GREP_CONTEXT_HEAD;
1461         xdemitconf_t xecfg;
1462
1463         if (!opt->output)
1464                 opt->output = std_output;
1465
1466         if (opt->pre_context || opt->post_context || opt->file_break ||
1467             opt->funcbody) {
1468                 /* Show hunk marks, except for the first file. */
1469                 if (opt->last_shown)
1470                         opt->show_hunk_mark = 1;
1471                 /*
1472                  * If we're using threads then we can't easily identify
1473                  * the first file.  Always put hunk marks in that case
1474                  * and skip the very first one later in work_done().
1475                  */
1476                 if (opt->output != std_output)
1477                         opt->show_hunk_mark = 1;
1478         }
1479         opt->last_shown = 0;
1480
1481         if (opt->allow_textconv) {
1482                 grep_source_load_driver(gs);
1483                 /*
1484                  * We might set up the shared textconv cache data here, which
1485                  * is not thread-safe.
1486                  */
1487                 grep_attr_lock();
1488                 textconv = userdiff_get_textconv(gs->driver);
1489                 grep_attr_unlock();
1490         }
1491
1492         /*
1493          * We know the result of a textconv is text, so we only have to care
1494          * about binary handling if we are not using it.
1495          */
1496         if (!textconv) {
1497                 switch (opt->binary) {
1498                 case GREP_BINARY_DEFAULT:
1499                         if (grep_source_is_binary(gs))
1500                                 binary_match_only = 1;
1501                         break;
1502                 case GREP_BINARY_NOMATCH:
1503                         if (grep_source_is_binary(gs))
1504                                 return 0; /* Assume unmatch */
1505                         break;
1506                 case GREP_BINARY_TEXT:
1507                         break;
1508                 default:
1509                         die("BUG: unknown binary handling mode");
1510                 }
1511         }
1512
1513         memset(&xecfg, 0, sizeof(xecfg));
1514         opt->priv = &xecfg;
1515
1516         try_lookahead = should_lookahead(opt);
1517
1518         if (fill_textconv_grep(textconv, gs) < 0)
1519                 return 0;
1520
1521         bol = gs->buf;
1522         left = gs->size;
1523         while (left) {
1524                 char *eol, ch;
1525                 int hit;
1526
1527                 /*
1528                  * look_ahead() skips quickly to the line that possibly
1529                  * has the next hit; don't call it if we need to do
1530                  * something more than just skipping the current line
1531                  * in response to an unmatch for the current line.  E.g.
1532                  * inside a post-context window, we will show the current
1533                  * line as a context around the previous hit when it
1534                  * doesn't hit.
1535                  */
1536                 if (try_lookahead
1537                     && !(last_hit
1538                          && (show_function ||
1539                              lno <= last_hit + opt->post_context))
1540                     && look_ahead(opt, &left, &lno, &bol))
1541                         break;
1542                 eol = end_of_line(bol, &left);
1543                 ch = *eol;
1544                 *eol = 0;
1545
1546                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1547                         ctx = GREP_CONTEXT_BODY;
1548
1549                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1550                 *eol = ch;
1551
1552                 if (collect_hits)
1553                         goto next_line;
1554
1555                 /* "grep -v -e foo -e bla" should list lines
1556                  * that do not have either, so inversion should
1557                  * be done outside.
1558                  */
1559                 if (opt->invert)
1560                         hit = !hit;
1561                 if (opt->unmatch_name_only) {
1562                         if (hit)
1563                                 return 0;
1564                         goto next_line;
1565                 }
1566                 if (hit) {
1567                         count++;
1568                         if (opt->status_only)
1569                                 return 1;
1570                         if (opt->name_only) {
1571                                 show_name(opt, gs->name);
1572                                 return 1;
1573                         }
1574                         if (opt->count)
1575                                 goto next_line;
1576                         if (binary_match_only) {
1577                                 opt->output(opt, "Binary file ", 12);
1578                                 output_color(opt, gs->name, strlen(gs->name),
1579                                              opt->color_filename);
1580                                 opt->output(opt, " matches\n", 9);
1581                                 return 1;
1582                         }
1583                         /* Hit at this line.  If we haven't shown the
1584                          * pre-context lines, we would need to show them.
1585                          */
1586                         if (opt->pre_context || opt->funcbody)
1587                                 show_pre_context(opt, gs, bol, eol, lno);
1588                         else if (opt->funcname)
1589                                 show_funcname_line(opt, gs, bol, lno);
1590                         show_line(opt, bol, eol, gs->name, lno, ':');
1591                         last_hit = lno;
1592                         if (opt->funcbody)
1593                                 show_function = 1;
1594                         goto next_line;
1595                 }
1596                 if (show_function && (!peek_bol || peek_bol < bol)) {
1597                         unsigned long peek_left = left;
1598                         char *peek_eol = eol;
1599
1600                         /*
1601                          * Trailing empty lines are not interesting.
1602                          * Peek past them to see if they belong to the
1603                          * body of the current function.
1604                          */
1605                         peek_bol = bol;
1606                         while (is_empty_line(peek_bol, peek_eol)) {
1607                                 peek_bol = peek_eol + 1;
1608                                 peek_eol = end_of_line(peek_bol, &peek_left);
1609                         }
1610
1611                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1612                                 show_function = 0;
1613                 }
1614                 if (show_function ||
1615                     (last_hit && lno <= last_hit + opt->post_context)) {
1616                         /* If the last hit is within the post context,
1617                          * we need to show this line.
1618                          */
1619                         show_line(opt, bol, eol, gs->name, lno, '-');
1620                 }
1621
1622         next_line:
1623                 bol = eol + 1;
1624                 if (!left)
1625                         break;
1626                 left--;
1627                 lno++;
1628         }
1629
1630         if (collect_hits)
1631                 return 0;
1632
1633         if (opt->status_only)
1634                 return 0;
1635         if (opt->unmatch_name_only) {
1636                 /* We did not see any hit, so we want to show this */
1637                 show_name(opt, gs->name);
1638                 return 1;
1639         }
1640
1641         xdiff_clear_find_func(&xecfg);
1642         opt->priv = NULL;
1643
1644         /* NEEDSWORK:
1645          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1646          * which feels mostly useless but sometimes useful.  Maybe
1647          * make it another option?  For now suppress them.
1648          */
1649         if (opt->count && count) {
1650                 char buf[32];
1651                 if (opt->pathname) {
1652                         output_color(opt, gs->name, strlen(gs->name),
1653                                      opt->color_filename);
1654                         output_sep(opt, ':');
1655                 }
1656                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1657                 opt->output(opt, buf, strlen(buf));
1658                 return 1;
1659         }
1660         return !!last_hit;
1661 }
1662
1663 static void clr_hit_marker(struct grep_expr *x)
1664 {
1665         /* All-hit markers are meaningful only at the very top level
1666          * OR node.
1667          */
1668         while (1) {
1669                 x->hit = 0;
1670                 if (x->node != GREP_NODE_OR)
1671                         return;
1672                 x->u.binary.left->hit = 0;
1673                 x = x->u.binary.right;
1674         }
1675 }
1676
1677 static int chk_hit_marker(struct grep_expr *x)
1678 {
1679         /* Top level nodes have hit markers.  See if they all are hits */
1680         while (1) {
1681                 if (x->node != GREP_NODE_OR)
1682                         return x->hit;
1683                 if (!x->u.binary.left->hit)
1684                         return 0;
1685                 x = x->u.binary.right;
1686         }
1687 }
1688
1689 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1690 {
1691         /*
1692          * we do not have to do the two-pass grep when we do not check
1693          * buffer-wide "all-match".
1694          */
1695         if (!opt->all_match)
1696                 return grep_source_1(opt, gs, 0);
1697
1698         /* Otherwise the toplevel "or" terms hit a bit differently.
1699          * We first clear hit markers from them.
1700          */
1701         clr_hit_marker(opt->pattern_expression);
1702         grep_source_1(opt, gs, 1);
1703
1704         if (!chk_hit_marker(opt->pattern_expression))
1705                 return 0;
1706
1707         return grep_source_1(opt, gs, 0);
1708 }
1709
1710 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1711 {
1712         struct grep_source gs;
1713         int r;
1714
1715         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1716         gs.buf = buf;
1717         gs.size = size;
1718
1719         r = grep_source(opt, &gs);
1720
1721         grep_source_clear(&gs);
1722         return r;
1723 }
1724
1725 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1726                       const char *name, const char *path,
1727                       const void *identifier)
1728 {
1729         gs->type = type;
1730         gs->name = xstrdup_or_null(name);
1731         gs->path = xstrdup_or_null(path);
1732         gs->buf = NULL;
1733         gs->size = 0;
1734         gs->driver = NULL;
1735
1736         switch (type) {
1737         case GREP_SOURCE_FILE:
1738                 gs->identifier = xstrdup(identifier);
1739                 break;
1740         case GREP_SOURCE_SUBMODULE:
1741                 if (!identifier) {
1742                         gs->identifier = NULL;
1743                         break;
1744                 }
1745                 /*
1746                  * FALL THROUGH
1747                  * If the identifier is non-NULL (in the submodule case) it
1748                  * will be a SHA1 that needs to be copied.
1749                  */
1750         case GREP_SOURCE_SHA1:
1751                 gs->identifier = xmalloc(20);
1752                 hashcpy(gs->identifier, identifier);
1753                 break;
1754         case GREP_SOURCE_BUF:
1755                 gs->identifier = NULL;
1756                 break;
1757         }
1758 }
1759
1760 void grep_source_clear(struct grep_source *gs)
1761 {
1762         free(gs->name);
1763         gs->name = NULL;
1764         free(gs->path);
1765         gs->path = NULL;
1766         free(gs->identifier);
1767         gs->identifier = NULL;
1768         grep_source_clear_data(gs);
1769 }
1770
1771 void grep_source_clear_data(struct grep_source *gs)
1772 {
1773         switch (gs->type) {
1774         case GREP_SOURCE_FILE:
1775         case GREP_SOURCE_SHA1:
1776         case GREP_SOURCE_SUBMODULE:
1777                 free(gs->buf);
1778                 gs->buf = NULL;
1779                 gs->size = 0;
1780                 break;
1781         case GREP_SOURCE_BUF:
1782                 /* leave user-provided buf intact */
1783                 break;
1784         }
1785 }
1786
1787 static int grep_source_load_sha1(struct grep_source *gs)
1788 {
1789         enum object_type type;
1790
1791         grep_read_lock();
1792         gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1793         grep_read_unlock();
1794
1795         if (!gs->buf)
1796                 return error(_("'%s': unable to read %s"),
1797                              gs->name,
1798                              sha1_to_hex(gs->identifier));
1799         return 0;
1800 }
1801
1802 static int grep_source_load_file(struct grep_source *gs)
1803 {
1804         const char *filename = gs->identifier;
1805         struct stat st;
1806         char *data;
1807         size_t size;
1808         int i;
1809
1810         if (lstat(filename, &st) < 0) {
1811         err_ret:
1812                 if (errno != ENOENT)
1813                         error_errno(_("failed to stat '%s'"), filename);
1814                 return -1;
1815         }
1816         if (!S_ISREG(st.st_mode))
1817                 return -1;
1818         size = xsize_t(st.st_size);
1819         i = open(filename, O_RDONLY);
1820         if (i < 0)
1821                 goto err_ret;
1822         data = xmallocz(size);
1823         if (st.st_size != read_in_full(i, data, size)) {
1824                 error_errno(_("'%s': short read"), filename);
1825                 close(i);
1826                 free(data);
1827                 return -1;
1828         }
1829         close(i);
1830
1831         gs->buf = data;
1832         gs->size = size;
1833         return 0;
1834 }
1835
1836 static int grep_source_load(struct grep_source *gs)
1837 {
1838         if (gs->buf)
1839                 return 0;
1840
1841         switch (gs->type) {
1842         case GREP_SOURCE_FILE:
1843                 return grep_source_load_file(gs);
1844         case GREP_SOURCE_SHA1:
1845                 return grep_source_load_sha1(gs);
1846         case GREP_SOURCE_BUF:
1847                 return gs->buf ? 0 : -1;
1848         case GREP_SOURCE_SUBMODULE:
1849                 break;
1850         }
1851         die("BUG: invalid grep_source type to load");
1852 }
1853
1854 void grep_source_load_driver(struct grep_source *gs)
1855 {
1856         if (gs->driver)
1857                 return;
1858
1859         grep_attr_lock();
1860         if (gs->path)
1861                 gs->driver = userdiff_find_by_path(gs->path);
1862         if (!gs->driver)
1863                 gs->driver = userdiff_find_by_name("default");
1864         grep_attr_unlock();
1865 }
1866
1867 static int grep_source_is_binary(struct grep_source *gs)
1868 {
1869         grep_source_load_driver(gs);
1870         if (gs->driver->binary != -1)
1871                 return gs->driver->binary;
1872
1873         if (!grep_source_load(gs))
1874                 return buffer_is_binary(gs->buf, gs->size);
1875
1876         return 0;
1877 }