grep.c

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