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                 break;
 183
 184         case GREP_PATTERN_TYPE_ERE:
 185                 opt->fixed = 0;
 186                 opt->pcre = 0;
 187                 opt->regflags |= REG_EXTENDED;
 188                 break;
 189
 190         case GREP_PATTERN_TYPE_FIXED:
 191                 opt->fixed = 1;
 192                 opt->pcre = 0;
 193                 break;
 194
 195         case GREP_PATTERN_TYPE_PCRE:
 196                 opt->fixed = 0;
 197                 opt->pcre = 1;
 198                 break;
 199         }
 200 }
 201
 202 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 203 {
 204         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 205                 grep_set_pattern_type_option(pattern_type, opt);
 206         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 207                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 208         else if (opt->extended_regexp_option)
 209                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 210 }
 211
 212 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 213                                         const char *origin, int no,
 214                                         enum grep_pat_token t,
 215                                         enum grep_header_field field)
 216 {
 217         struct grep_pat *p = xcalloc(1, sizeof(*p));
 218         p->pattern = xmemdupz(pat, patlen);
 219         p->patternlen = patlen;
 220         p->origin = origin;
 221         p->no = no;
 222         p->token = t;
 223         p->field = field;
 224         return p;
 225 }
 226
 227 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 228 {
 229         **tail = p;
 230         *tail = &p->next;
 231         p->next = NULL;
 232
 233         switch (p->token) {
 234         case GREP_PATTERN: /* atom */
 235         case GREP_PATTERN_HEAD:
 236         case GREP_PATTERN_BODY:
 237                 for (;;) {
 238                         struct grep_pat *new_pat;
 239                         size_t len = 0;
 240                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 241                         while (++len <= p->patternlen) {
 242                                 if (*(--cp) == '\n') {
 243                                         nl = cp;
 244                                         break;
 245                                 }
 246                         }
 247                         if (!nl)
 248                                 break;
 249                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 250                                                   p->no, p->token, p->field);
 251                         new_pat->next = p->next;
 252                         if (!p->next)
 253                                 *tail = &new_pat->next;
 254                         p->next = new_pat;
 255                         *nl = '\0';
 256                         p->patternlen -= len;
 257                 }
 258                 break;
 259         default:
 260                 break;
 261         }
 262 }
 263
 264 void append_header_grep_pattern(struct grep_opt *opt,
 265                                 enum grep_header_field field, const char *pat)
 266 {
 267         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 268                                              GREP_PATTERN_HEAD, field);
 269         if (field == GREP_HEADER_REFLOG)
 270                 opt->use_reflog_filter = 1;
 271         do_append_grep_pat(&opt->header_tail, p);
 272 }
 273
 274 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 275                          const char *origin, int no, enum grep_pat_token t)
 276 {
 277         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 278 }
 279
 280 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 281                      const char *origin, int no, enum grep_pat_token t)
 282 {
 283         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 284         do_append_grep_pat(&opt->pattern_tail, p);
 285 }
 286
 287 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 288 {
 289         struct grep_pat *pat;
 290         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 291         *ret = *opt;
 292
 293         ret->pattern_list = NULL;
 294         ret->pattern_tail = &ret->pattern_list;
 295
 296         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 297         {
 298                 if(pat->token == GREP_PATTERN_HEAD)
 299                         append_header_grep_pattern(ret, pat->field,
 300                                                    pat->pattern);
 301                 else
 302                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 303                                         pat->origin, pat->no, pat->token);
 304         }
 305
 306         return ret;
 307 }
 308
 309 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 310                 const char *error)
 311 {
 312         char where[1024];
 313
 314         if (p->no)
 315                 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 316         else if (p->origin)
 317                 xsnprintf(where, sizeof(where), "%s, ", p->origin);
 318         else
 319                 where[0] = 0;
 320
 321         die("%s'%s': %s", where, p->pattern, error);
 322 }
 323
 324 #ifdef USE_LIBPCRE
 325 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 326 {
 327         const char *error;
 328         int erroffset;
 329         int options = PCRE_MULTILINE;
 330
 331         if (opt->ignore_case) {
 332                 if (has_non_ascii(p->pattern))
 333                         p->pcre_tables = pcre_maketables();
 334                 options |= PCRE_CASELESS;
 335         }
 336         if (is_utf8_locale() && has_non_ascii(p->pattern))
 337                 options |= PCRE_UTF8;
 338
 339         p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 340                                       p->pcre_tables);
 341         if (!p->pcre_regexp)
 342                 compile_regexp_failed(p, error);
 343
 344         p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 345         if (!p->pcre_extra_info && error)
 346                 die("%s", error);
 347 }
 348
 349 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 350                 regmatch_t *match, int eflags)
 351 {
 352         int ovector[30], ret, flags = 0;
 353
 354         if (eflags & REG_NOTBOL)
 355                 flags |= PCRE_NOTBOL;
 356
 357         ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 358                         0, flags, ovector, ARRAY_SIZE(ovector));
 359         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 360                 die("pcre_exec failed with error code %d", ret);
 361         if (ret > 0) {
 362                 ret = 0;
 363                 match->rm_so = ovector[0];
 364                 match->rm_eo = ovector[1];
 365         }
 366
 367         return ret;
 368 }
 369
 370 static void free_pcre_regexp(struct grep_pat *p)
 371 {
 372         pcre_free(p->pcre_regexp);
 373         pcre_free(p->pcre_extra_info);
 374         pcre_free((void *)p->pcre_tables);
 375 }
 376 #else /* !USE_LIBPCRE */
 377 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 378 {
 379         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 380 }
 381
 382 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 383                 regmatch_t *match, int eflags)
 384 {
 385         return 1;
 386 }
 387
 388 static void free_pcre_regexp(struct grep_pat *p)
 389 {
 390 }
 391 #endif /* !USE_LIBPCRE */
 392
 393 static int is_fixed(const char *s, size_t len)
 394 {
 395         size_t i;
 396
 397         /* regcomp cannot accept patterns with NULs so we
 398          * consider any pattern containing a NUL fixed.
 399          */
 400         if (memchr(s, 0, len))
 401                 return 1;
 402
 403         for (i = 0; i < len; i++) {
 404                 if (is_regex_special(s[i]))
 405                         return 0;
 406         }
 407
 408         return 1;
 409 }
 410
 411 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 412 {
 413         struct strbuf sb = STRBUF_INIT;
 414         int err;
 415         int regflags = opt->regflags;
 416
 417         basic_regex_quote_buf(&sb, p->pattern);
 418         if (opt->ignore_case)
 419                 regflags |= REG_ICASE;
 420         err = regcomp(&p->regexp, sb.buf, regflags);
 421         if (opt->debug)
 422                 fprintf(stderr, "fixed %s\n", sb.buf);
 423         strbuf_release(&sb);
 424         if (err) {
 425                 char errbuf[1024];
 426                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 427                 regfree(&p->regexp);
 428                 compile_regexp_failed(p, errbuf);
 429         }
 430 }
 431
 432 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 433 {
 434         int icase, ascii_only;
 435         int err;
 436
 437         p->word_regexp = opt->word_regexp;
 438         p->ignore_case = opt->ignore_case;
 439         icase          = opt->regflags & REG_ICASE || p->ignore_case;
 440         ascii_only     = !has_non_ascii(p->pattern);
 441
 442         /*
 443          * Even when -F (fixed) asks us to do a non-regexp search, we
 444          * may not be able to correctly case-fold when -i
 445          * (ignore-case) is asked (in which case, we'll synthesize a
 446          * regexp to match the pattern that matches regexp special
 447          * characters literally, while ignoring case differences).  On
 448          * the other hand, even without -F, if the pattern does not
 449          * have any regexp special characters and there is no need for
 450          * case-folding search, we can internally turn it into a
 451          * simple string match using kws.  p->fixed tells us if we
 452          * want to use kws.
 453          */
 454         if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 455                 p->fixed = !icase || ascii_only;
 456         else
 457                 p->fixed = 0;
 458
 459         if (p->fixed) {
 460                 p->kws = kwsalloc(icase ? tolower_trans_tbl : NULL);
 461                 kwsincr(p->kws, p->pattern, p->patternlen);
 462                 kwsprep(p->kws);
 463                 return;
 464         } else if (opt->fixed) {
 465                 /*
 466                  * We come here when the pattern has the non-ascii
 467                  * characters we cannot case-fold, and asked to
 468                  * ignore-case.
 469                  */
 470                 compile_fixed_regexp(p, opt);
 471                 return;
 472         }
 473
 474         if (opt->pcre) {
 475                 compile_pcre_regexp(p, opt);
 476                 return;
 477         }
 478
 479         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 480         if (err) {
 481                 char errbuf[1024];
 482                 regerror(err, &p->regexp, errbuf, 1024);
 483                 regfree(&p->regexp);
 484                 compile_regexp_failed(p, errbuf);
 485         }
 486 }
 487
 488 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 489 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 490 {
 491         struct grep_pat *p;
 492         struct grep_expr *x;
 493
 494         p = *list;
 495         if (!p)
 496                 return NULL;
 497         switch (p->token) {
 498         case GREP_PATTERN: /* atom */
 499         case GREP_PATTERN_HEAD:
 500         case GREP_PATTERN_BODY:
 501                 x = xcalloc(1, sizeof (struct grep_expr));
 502                 x->node = GREP_NODE_ATOM;
 503                 x->u.atom = p;
 504                 *list = p->next;
 505                 return x;
 506         case GREP_OPEN_PAREN:
 507                 *list = p->next;
 508                 x = compile_pattern_or(list);
 509                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 510                         die("unmatched parenthesis");
 511                 *list = (*list)->next;
 512                 return x;
 513         default:
 514                 return NULL;
 515         }
 516 }
 517
 518 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 519 {
 520         struct grep_pat *p;
 521         struct grep_expr *x;
 522
 523         p = *list;
 524         if (!p)
 525                 return NULL;
 526         switch (p->token) {
 527         case GREP_NOT:
 528                 if (!p->next)
 529                         die("--not not followed by pattern expression");
 530                 *list = p->next;
 531                 x = xcalloc(1, sizeof (struct grep_expr));
 532                 x->node = GREP_NODE_NOT;
 533                 x->u.unary = compile_pattern_not(list);
 534                 if (!x->u.unary)
 535                         die("--not followed by non pattern expression");
 536                 return x;
 537         default:
 538                 return compile_pattern_atom(list);
 539         }
 540 }
 541
 542 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 543 {
 544         struct grep_pat *p;
 545         struct grep_expr *x, *y, *z;
 546
 547         x = compile_pattern_not(list);
 548         p = *list;
 549         if (p && p->token == GREP_AND) {
 550                 if (!p->next)
 551                         die("--and not followed by pattern expression");
 552                 *list = p->next;
 553                 y = compile_pattern_and(list);
 554                 if (!y)
 555                         die("--and not followed by pattern expression");
 556                 z = xcalloc(1, sizeof (struct grep_expr));
 557                 z->node = GREP_NODE_AND;
 558                 z->u.binary.left = x;
 559                 z->u.binary.right = y;
 560                 return z;
 561         }
 562         return x;
 563 }
 564
 565 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 566 {
 567         struct grep_pat *p;
 568         struct grep_expr *x, *y, *z;
 569
 570         x = compile_pattern_and(list);
 571         p = *list;
 572         if (x && p && p->token != GREP_CLOSE_PAREN) {
 573                 y = compile_pattern_or(list);
 574                 if (!y)
 575                         die("not a pattern expression %s", p->pattern);
 576                 z = xcalloc(1, sizeof (struct grep_expr));
 577                 z->node = GREP_NODE_OR;
 578                 z->u.binary.left = x;
 579                 z->u.binary.right = y;
 580                 return z;
 581         }
 582         return x;
 583 }
 584
 585 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 586 {
 587         return compile_pattern_or(list);
 588 }
 589
 590 static void indent(int in)
 591 {
 592         while (in-- > 0)
 593                 fputc(' ', stderr);
 594 }
 595
 596 static void dump_grep_pat(struct grep_pat *p)
 597 {
 598         switch (p->token) {
 599         case GREP_AND: fprintf(stderr, "*and*"); break;
 600         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 601         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 602         case GREP_NOT: fprintf(stderr, "*not*"); break;
 603         case GREP_OR: fprintf(stderr, "*or*"); break;
 604
 605         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 606         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 607         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 608         }
 609
 610         switch (p->token) {
 611         default: break;
 612         case GREP_PATTERN_HEAD:
 613                 fprintf(stderr, "<head %d>", p->field); break;
 614         case GREP_PATTERN_BODY:
 615                 fprintf(stderr, "<body>"); break;
 616         }
 617         switch (p->token) {
 618         default: break;
 619         case GREP_PATTERN_HEAD:
 620         case GREP_PATTERN_BODY:
 621         case GREP_PATTERN:
 622                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 623                 break;
 624         }
 625         fputc('\n', stderr);
 626 }
 627
 628 static void dump_grep_expression_1(struct grep_expr *x, int in)
 629 {
 630         indent(in);
 631         switch (x->node) {
 632         case GREP_NODE_TRUE:
 633                 fprintf(stderr, "true\n");
 634                 break;
 635         case GREP_NODE_ATOM:
 636                 dump_grep_pat(x->u.atom);
 637                 break;
 638         case GREP_NODE_NOT:
 639                 fprintf(stderr, "(not\n");
 640                 dump_grep_expression_1(x->u.unary, in+1);
 641                 indent(in);
 642                 fprintf(stderr, ")\n");
 643                 break;
 644         case GREP_NODE_AND:
 645                 fprintf(stderr, "(and\n");
 646                 dump_grep_expression_1(x->u.binary.left, in+1);
 647                 dump_grep_expression_1(x->u.binary.right, in+1);
 648                 indent(in);
 649                 fprintf(stderr, ")\n");
 650                 break;
 651         case GREP_NODE_OR:
 652                 fprintf(stderr, "(or\n");
 653                 dump_grep_expression_1(x->u.binary.left, in+1);
 654                 dump_grep_expression_1(x->u.binary.right, in+1);
 655                 indent(in);
 656                 fprintf(stderr, ")\n");
 657                 break;
 658         }
 659 }
 660
 661 static void dump_grep_expression(struct grep_opt *opt)
 662 {
 663         struct grep_expr *x = opt->pattern_expression;
 664
 665         if (opt->all_match)
 666                 fprintf(stderr, "[all-match]\n");
 667         dump_grep_expression_1(x, 0);
 668         fflush(NULL);
 669 }
 670
 671 static struct grep_expr *grep_true_expr(void)
 672 {
 673         struct grep_expr *z = xcalloc(1, sizeof(*z));
 674         z->node = GREP_NODE_TRUE;
 675         return z;
 676 }
 677
 678 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 679 {
 680         struct grep_expr *z = xcalloc(1, sizeof(*z));
 681         z->node = GREP_NODE_OR;
 682         z->u.binary.left = left;
 683         z->u.binary.right = right;
 684         return z;
 685 }
 686
 687 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 688 {
 689         struct grep_pat *p;
 690         struct grep_expr *header_expr;
 691         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 692         enum grep_header_field fld;
 693
 694         if (!opt->header_list)
 695                 return NULL;
 696
 697         for (p = opt->header_list; p; p = p->next) {
 698                 if (p->token != GREP_PATTERN_HEAD)
 699                         die("BUG: a non-header pattern in grep header list.");
 700                 if (p->field < GREP_HEADER_FIELD_MIN ||
 701                     GREP_HEADER_FIELD_MAX <= p->field)
 702                         die("BUG: unknown header field %d", p->field);
 703                 compile_regexp(p, opt);
 704         }
 705
 706         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 707                 header_group[fld] = NULL;
 708
 709         for (p = opt->header_list; p; p = p->next) {
 710                 struct grep_expr *h;
 711                 struct grep_pat *pp = p;
 712
 713                 h = compile_pattern_atom(&pp);
 714                 if (!h || pp != p->next)
 715                         die("BUG: malformed header expr");
 716                 if (!header_group[p->field]) {
 717                         header_group[p->field] = h;
 718                         continue;
 719                 }
 720                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 721         }
 722
 723         header_expr = NULL;
 724
 725         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 726                 if (!header_group[fld])
 727                         continue;
 728                 if (!header_expr)
 729                         header_expr = grep_true_expr();
 730                 header_expr = grep_or_expr(header_group[fld], header_expr);
 731         }
 732         return header_expr;
 733 }
 734
 735 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 736 {
 737         struct grep_expr *z = x;
 738
 739         while (x) {
 740                 assert(x->node == GREP_NODE_OR);
 741                 if (x->u.binary.right &&
 742                     x->u.binary.right->node == GREP_NODE_TRUE) {
 743                         x->u.binary.right = y;
 744                         break;
 745                 }
 746                 x = x->u.binary.right;
 747         }
 748         return z;
 749 }
 750
 751 static void compile_grep_patterns_real(struct grep_opt *opt)
 752 {
 753         struct grep_pat *p;
 754         struct grep_expr *header_expr = prep_header_patterns(opt);
 755
 756         for (p = opt->pattern_list; p; p = p->next) {
 757                 switch (p->token) {
 758                 case GREP_PATTERN: /* atom */
 759                 case GREP_PATTERN_HEAD:
 760                 case GREP_PATTERN_BODY:
 761                         compile_regexp(p, opt);
 762                         break;
 763                 default:
 764                         opt->extended = 1;
 765                         break;
 766                 }
 767         }
 768
 769         if (opt->all_match || header_expr)
 770                 opt->extended = 1;
 771         else if (!opt->extended && !opt->debug)
 772                 return;
 773
 774         p = opt->pattern_list;
 775         if (p)
 776                 opt->pattern_expression = compile_pattern_expr(&p);
 777         if (p)
 778                 die("incomplete pattern expression: %s", p->pattern);
 779
 780         if (!header_expr)
 781                 return;
 782
 783         if (!opt->pattern_expression)
 784                 opt->pattern_expression = header_expr;
 785         else if (opt->all_match)
 786                 opt->pattern_expression = grep_splice_or(header_expr,
 787                                                          opt->pattern_expression);
 788         else
 789                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 790                                                        header_expr);
 791         opt->all_match = 1;
 792 }
 793
 794 void compile_grep_patterns(struct grep_opt *opt)
 795 {
 796         compile_grep_patterns_real(opt);
 797         if (opt->debug)
 798                 dump_grep_expression(opt);
 799 }
 800
 801 static void free_pattern_expr(struct grep_expr *x)
 802 {
 803         switch (x->node) {
 804         case GREP_NODE_TRUE:
 805         case GREP_NODE_ATOM:
 806                 break;
 807         case GREP_NODE_NOT:
 808                 free_pattern_expr(x->u.unary);
 809                 break;
 810         case GREP_NODE_AND:
 811         case GREP_NODE_OR:
 812                 free_pattern_expr(x->u.binary.left);
 813                 free_pattern_expr(x->u.binary.right);
 814                 break;
 815         }
 816         free(x);
 817 }
 818
 819 void free_grep_patterns(struct grep_opt *opt)
 820 {
 821         struct grep_pat *p, *n;
 822
 823         for (p = opt->pattern_list; p; p = n) {
 824                 n = p->next;
 825                 switch (p->token) {
 826                 case GREP_PATTERN: /* atom */
 827                 case GREP_PATTERN_HEAD:
 828                 case GREP_PATTERN_BODY:
 829                         if (p->kws)
 830                                 kwsfree(p->kws);
 831                         else if (p->pcre_regexp)
 832                                 free_pcre_regexp(p);
 833                         else
 834                                 regfree(&p->regexp);
 835                         free(p->pattern);
 836                         break;
 837                 default:
 838                         break;
 839                 }
 840                 free(p);
 841         }
 842
 843         if (!opt->extended)
 844                 return;
 845         free_pattern_expr(opt->pattern_expression);
 846 }
 847
 848 static char *end_of_line(char *cp, unsigned long *left)
 849 {
 850         unsigned long l = *left;
 851         while (l && *cp != '\n') {
 852                 l--;
 853                 cp++;
 854         }
 855         *left = l;
 856         return cp;
 857 }
 858
 859 static int word_char(char ch)
 860 {
 861         return isalnum(ch) || ch == '_';
 862 }
 863
 864 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 865                          const char *color)
 866 {
 867         if (want_color(opt->color) && color && color[0]) {
 868                 opt->output(opt, color, strlen(color));
 869                 opt->output(opt, data, size);
 870                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 871         } else
 872                 opt->output(opt, data, size);
 873 }
 874
 875 static void output_sep(struct grep_opt *opt, char sign)
 876 {
 877         if (opt->null_following_name)
 878                 opt->output(opt, "\0", 1);
 879         else
 880                 output_color(opt, &sign, 1, opt->color_sep);
 881 }
 882
 883 static void show_name(struct grep_opt *opt, const char *name)
 884 {
 885         output_color(opt, name, strlen(name), opt->color_filename);
 886         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 887 }
 888
 889 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 890                     regmatch_t *match)
 891 {
 892         struct kwsmatch kwsm;
 893         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 894         if (offset == -1) {
 895                 match->rm_so = match->rm_eo = -1;
 896                 return REG_NOMATCH;
 897         } else {
 898                 match->rm_so = offset;
 899                 match->rm_eo = match->rm_so + kwsm.size[0];
 900                 return 0;
 901         }
 902 }
 903
 904 static int patmatch(struct grep_pat *p, char *line, char *eol,
 905                     regmatch_t *match, int eflags)
 906 {
 907         int hit;
 908
 909         if (p->fixed)
 910                 hit = !fixmatch(p, line, eol, match);
 911         else if (p->pcre_regexp)
 912                 hit = !pcrematch(p, line, eol, match, eflags);
 913         else
 914                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
 915                                    eflags);
 916
 917         return hit;
 918 }
 919
 920 static int strip_timestamp(char *bol, char **eol_p)
 921 {
 922         char *eol = *eol_p;
 923         int ch;
 924
 925         while (bol < --eol) {
 926                 if (*eol != '>')
 927                         continue;
 928                 *eol_p = ++eol;
 929                 ch = *eol;
 930                 *eol = '\0';
 931                 return ch;
 932         }
 933         return 0;
 934 }
 935
 936 static struct {
 937         const char *field;
 938         size_t len;
 939 } header_field[] = {
 940         { "author ", 7 },
 941         { "committer ", 10 },
 942         { "reflog ", 7 },
 943 };
 944
 945 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 946                              enum grep_context ctx,
 947                              regmatch_t *pmatch, int eflags)
 948 {
 949         int hit = 0;
 950         int saved_ch = 0;
 951         const char *start = bol;
 952
 953         if ((p->token != GREP_PATTERN) &&
 954             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 955                 return 0;
 956
 957         if (p->token == GREP_PATTERN_HEAD) {
 958                 const char *field;
 959                 size_t len;
 960                 assert(p->field < ARRAY_SIZE(header_field));
 961                 field = header_field[p->field].field;
 962                 len = header_field[p->field].len;
 963                 if (strncmp(bol, field, len))
 964                         return 0;
 965                 bol += len;
 966                 switch (p->field) {
 967                 case GREP_HEADER_AUTHOR:
 968                 case GREP_HEADER_COMMITTER:
 969                         saved_ch = strip_timestamp(bol, &eol);
 970                         break;
 971                 default:
 972                         break;
 973                 }
 974         }
 975
 976  again:
 977         hit = patmatch(p, bol, eol, pmatch, eflags);
 978
 979         if (hit && p->word_regexp) {
 980                 if ((pmatch[0].rm_so < 0) ||
 981                     (eol - bol) < pmatch[0].rm_so ||
 982                     (pmatch[0].rm_eo < 0) ||
 983                     (eol - bol) < pmatch[0].rm_eo)
 984                         die("regexp returned nonsense");
 985
 986                 /* Match beginning must be either beginning of the
 987                  * line, or at word boundary (i.e. the last char must
 988                  * not be a word char).  Similarly, match end must be
 989                  * either end of the line, or at word boundary
 990                  * (i.e. the next char must not be a word char).
 991                  */
 992                 if ( ((pmatch[0].rm_so == 0) ||
 993                       !word_char(bol[pmatch[0].rm_so-1])) &&
 994                      ((pmatch[0].rm_eo == (eol-bol)) ||
 995                       !word_char(bol[pmatch[0].rm_eo])) )
 996                         ;
 997                 else
 998                         hit = 0;
 999
1000                 /* Words consist of at least one character. */
1001                 if (pmatch->rm_so == pmatch->rm_eo)
1002                         hit = 0;
1003
1004                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1005                         /* There could be more than one match on the
1006                          * line, and the first match might not be
1007                          * strict word match.  But later ones could be!
1008                          * Forward to the next possible start, i.e. the
1009                          * next position following a non-word char.
1010                          */
1011                         bol = pmatch[0].rm_so + bol + 1;
1012                         while (word_char(bol[-1]) && bol < eol)
1013                                 bol++;
1014                         eflags |= REG_NOTBOL;
1015                         if (bol < eol)
1016                                 goto again;
1017                 }
1018         }
1019         if (p->token == GREP_PATTERN_HEAD && saved_ch)
1020                 *eol = saved_ch;
1021         if (hit) {
1022                 pmatch[0].rm_so += bol - start;
1023                 pmatch[0].rm_eo += bol - start;
1024         }
1025         return hit;
1026 }
1027
1028 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1029                            enum grep_context ctx, int collect_hits)
1030 {
1031         int h = 0;
1032         regmatch_t match;
1033
1034         if (!x)
1035                 die("Not a valid grep expression");
1036         switch (x->node) {
1037         case GREP_NODE_TRUE:
1038                 h = 1;
1039                 break;
1040         case GREP_NODE_ATOM:
1041                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1042                 break;
1043         case GREP_NODE_NOT:
1044                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1045                 break;
1046         case GREP_NODE_AND:
1047                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1048                         return 0;
1049                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1050                 break;
1051         case GREP_NODE_OR:
1052                 if (!collect_hits)
1053                         return (match_expr_eval(x->u.binary.left,
1054                                                 bol, eol, ctx, 0) ||
1055                                 match_expr_eval(x->u.binary.right,
1056                                                 bol, eol, ctx, 0));
1057                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1058                 x->u.binary.left->hit |= h;
1059                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1060                 break;
1061         default:
1062                 die("Unexpected node type (internal error) %d", x->node);
1063         }
1064         if (collect_hits)
1065                 x->hit |= h;
1066         return h;
1067 }
1068
1069 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1070                       enum grep_context ctx, int collect_hits)
1071 {
1072         struct grep_expr *x = opt->pattern_expression;
1073         return match_expr_eval(x, bol, eol, ctx, collect_hits);
1074 }
1075
1076 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1077                       enum grep_context ctx, int collect_hits)
1078 {
1079         struct grep_pat *p;
1080         regmatch_t match;
1081
1082         if (opt->extended)
1083                 return match_expr(opt, bol, eol, ctx, collect_hits);
1084
1085         /* we do not call with collect_hits without being extended */
1086         for (p = opt->pattern_list; p; p = p->next) {
1087                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1088                         return 1;
1089         }
1090         return 0;
1091 }
1092
1093 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1094                               enum grep_context ctx,
1095                               regmatch_t *pmatch, int eflags)
1096 {
1097         regmatch_t match;
1098
1099         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1100                 return 0;
1101         if (match.rm_so < 0 || match.rm_eo < 0)
1102                 return 0;
1103         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1104                 if (match.rm_so > pmatch->rm_so)
1105                         return 1;
1106                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1107                         return 1;
1108         }
1109         pmatch->rm_so = match.rm_so;
1110         pmatch->rm_eo = match.rm_eo;
1111         return 1;
1112 }
1113
1114 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1115                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1116 {
1117         struct grep_pat *p;
1118         int hit = 0;
1119
1120         pmatch->rm_so = pmatch->rm_eo = -1;
1121         if (bol < eol) {
1122                 for (p = opt->pattern_list; p; p = p->next) {
1123                         switch (p->token) {
1124                         case GREP_PATTERN: /* atom */
1125                         case GREP_PATTERN_HEAD:
1126                         case GREP_PATTERN_BODY:
1127                                 hit |= match_next_pattern(p, bol, eol, ctx,
1128                                                           pmatch, eflags);
1129                                 break;
1130                         default:
1131                                 break;
1132                         }
1133                 }
1134         }
1135         return hit;
1136 }
1137
1138 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1139                       const char *name, unsigned lno, char sign)
1140 {
1141         int rest = eol - bol;
1142         const char *match_color, *line_color = NULL;
1143
1144         if (opt->file_break && opt->last_shown == 0) {
1145                 if (opt->show_hunk_mark)
1146                         opt->output(opt, "\n", 1);
1147         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1148                 if (opt->last_shown == 0) {
1149                         if (opt->show_hunk_mark) {
1150                                 output_color(opt, "--", 2, opt->color_sep);
1151                                 opt->output(opt, "\n", 1);
1152                         }
1153                 } else if (lno > opt->last_shown + 1) {
1154                         output_color(opt, "--", 2, opt->color_sep);
1155                         opt->output(opt, "\n", 1);
1156                 }
1157         }
1158         if (opt->heading && opt->last_shown == 0) {
1159                 output_color(opt, name, strlen(name), opt->color_filename);
1160                 opt->output(opt, "\n", 1);
1161         }
1162         opt->last_shown = lno;
1163
1164         if (!opt->heading && opt->pathname) {
1165                 output_color(opt, name, strlen(name), opt->color_filename);
1166                 output_sep(opt, sign);
1167         }
1168         if (opt->linenum) {
1169                 char buf[32];
1170                 xsnprintf(buf, sizeof(buf), "%d", lno);
1171                 output_color(opt, buf, strlen(buf), opt->color_lineno);
1172                 output_sep(opt, sign);
1173         }
1174         if (opt->color) {
1175                 regmatch_t match;
1176                 enum grep_context ctx = GREP_CONTEXT_BODY;
1177                 int ch = *eol;
1178                 int eflags = 0;
1179
1180                 if (sign == ':')
1181                         match_color = opt->color_match_selected;
1182                 else
1183                         match_color = opt->color_match_context;
1184                 if (sign == ':')
1185                         line_color = opt->color_selected;
1186                 else if (sign == '-')
1187                         line_color = opt->color_context;
1188                 else if (sign == '=')
1189                         line_color = opt->color_function;
1190                 *eol = '\0';
1191                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1192                         if (match.rm_so == match.rm_eo)
1193                                 break;
1194
1195                         output_color(opt, bol, match.rm_so, line_color);
1196                         output_color(opt, bol + match.rm_so,
1197                                      match.rm_eo - match.rm_so, match_color);
1198                         bol += match.rm_eo;
1199                         rest -= match.rm_eo;
1200                         eflags = REG_NOTBOL;
1201                 }
1202                 *eol = ch;
1203         }
1204         output_color(opt, bol, rest, line_color);
1205         opt->output(opt, "\n", 1);
1206 }
1207
1208 #ifndef NO_PTHREADS
1209 int grep_use_locks;
1210
1211 /*
1212  * This lock protects access to the gitattributes machinery, which is
1213  * not thread-safe.
1214  */
1215 pthread_mutex_t grep_attr_mutex;
1216
1217 static inline void grep_attr_lock(void)
1218 {
1219         if (grep_use_locks)
1220                 pthread_mutex_lock(&grep_attr_mutex);
1221 }
1222
1223 static inline void grep_attr_unlock(void)
1224 {
1225         if (grep_use_locks)
1226                 pthread_mutex_unlock(&grep_attr_mutex);
1227 }
1228
1229 /*
1230  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1231  */
1232 pthread_mutex_t grep_read_mutex;
1233
1234 #else
1235 #define grep_attr_lock()
1236 #define grep_attr_unlock()
1237 #endif
1238
1239 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1240 {
1241         xdemitconf_t *xecfg = opt->priv;
1242         if (xecfg && !xecfg->find_func) {
1243                 grep_source_load_driver(gs);
1244                 if (gs->driver->funcname.pattern) {
1245                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1246                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1247                 } else {
1248                         xecfg = opt->priv = NULL;
1249                 }
1250         }
1251
1252         if (xecfg) {
1253                 char buf[1];
1254                 return xecfg->find_func(bol, eol - bol, buf, 1,
1255                                         xecfg->find_func_priv) >= 0;
1256         }
1257
1258         if (bol == eol)
1259                 return 0;
1260         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1261                 return 1;
1262         return 0;
1263 }
1264
1265 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1266                                char *bol, unsigned lno)
1267 {
1268         while (bol > gs->buf) {
1269                 char *eol = --bol;
1270
1271                 while (bol > gs->buf && bol[-1] != '\n')
1272                         bol--;
1273                 lno--;
1274
1275                 if (lno <= opt->last_shown)
1276                         break;
1277
1278                 if (match_funcname(opt, gs, bol, eol)) {
1279                         show_line(opt, bol, eol, gs->name, lno, '=');
1280                         break;
1281                 }
1282         }
1283 }
1284
1285 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1286                              char *bol, char *end, unsigned lno)
1287 {
1288         unsigned cur = lno, from = 1, funcname_lno = 0;
1289         int funcname_needed = !!opt->funcname;
1290
1291         if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1292                 funcname_needed = 2;
1293
1294         if (opt->pre_context < lno)
1295                 from = lno - opt->pre_context;
1296         if (from <= opt->last_shown)
1297                 from = opt->last_shown + 1;
1298
1299         /* Rewind. */
1300         while (bol > gs->buf &&
1301                cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1302                 char *eol = --bol;
1303
1304                 while (bol > gs->buf && bol[-1] != '\n')
1305                         bol--;
1306                 cur--;
1307                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1308                         funcname_lno = cur;
1309                         funcname_needed = 0;
1310                 }
1311         }
1312
1313         /* We need to look even further back to find a function signature. */
1314         if (opt->funcname && funcname_needed)
1315                 show_funcname_line(opt, gs, bol, cur);
1316
1317         /* Back forward. */
1318         while (cur < lno) {
1319                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1320
1321                 while (*eol != '\n')
1322                         eol++;
1323                 show_line(opt, bol, eol, gs->name, cur, sign);
1324                 bol = eol + 1;
1325                 cur++;
1326         }
1327 }
1328
1329 static int should_lookahead(struct grep_opt *opt)
1330 {
1331         struct grep_pat *p;
1332
1333         if (opt->extended)
1334                 return 0; /* punt for too complex stuff */
1335         if (opt->invert)
1336                 return 0;
1337         for (p = opt->pattern_list; p; p = p->next) {
1338                 if (p->token != GREP_PATTERN)
1339                         return 0; /* punt for "header only" and stuff */
1340         }
1341         return 1;
1342 }
1343
1344 static int look_ahead(struct grep_opt *opt,
1345                       unsigned long *left_p,
1346                       unsigned *lno_p,
1347                       char **bol_p)
1348 {
1349         unsigned lno = *lno_p;
1350         char *bol = *bol_p;
1351         struct grep_pat *p;
1352         char *sp, *last_bol;
1353         regoff_t earliest = -1;
1354
1355         for (p = opt->pattern_list; p; p = p->next) {
1356                 int hit;
1357                 regmatch_t m;
1358
1359                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1360                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1361                         continue;
1362                 if (earliest < 0 || m.rm_so < earliest)
1363                         earliest = m.rm_so;
1364         }
1365
1366         if (earliest < 0) {
1367                 *bol_p = bol + *left_p;
1368                 *left_p = 0;
1369                 return 1;
1370         }
1371         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1372                 ; /* find the beginning of the line */
1373         last_bol = sp;
1374
1375         for (sp = bol; sp < last_bol; sp++) {
1376                 if (*sp == '\n')
1377                         lno++;
1378         }
1379         *left_p -= last_bol - bol;
1380         *bol_p = last_bol;
1381         *lno_p = lno;
1382         return 0;
1383 }
1384
1385 static int fill_textconv_grep(struct userdiff_driver *driver,
1386                               struct grep_source *gs)
1387 {
1388         struct diff_filespec *df;
1389         char *buf;
1390         size_t size;
1391
1392         if (!driver || !driver->textconv)
1393                 return grep_source_load(gs);
1394
1395         /*
1396          * The textconv interface is intimately tied to diff_filespecs, so we
1397          * have to pretend to be one. If we could unify the grep_source
1398          * and diff_filespec structs, this mess could just go away.
1399          */
1400         df = alloc_filespec(gs->path);
1401         switch (gs->type) {
1402         case GREP_SOURCE_SHA1:
1403                 fill_filespec(df, gs->identifier, 1, 0100644);
1404                 break;
1405         case GREP_SOURCE_FILE:
1406                 fill_filespec(df, null_sha1, 0, 0100644);
1407                 break;
1408         default:
1409                 die("BUG: attempt to textconv something without a path?");
1410         }
1411
1412         /*
1413          * fill_textconv is not remotely thread-safe; it may load objects
1414          * behind the scenes, and it modifies the global diff tempfile
1415          * structure.
1416          */
1417         grep_read_lock();
1418         size = fill_textconv(driver, df, &buf);
1419         grep_read_unlock();
1420         free_filespec(df);
1421
1422         /*
1423          * The normal fill_textconv usage by the diff machinery would just keep
1424          * the textconv'd buf separate from the diff_filespec. But much of the
1425          * grep code passes around a grep_source and assumes that its "buf"
1426          * pointer is the beginning of the thing we are searching. So let's
1427          * install our textconv'd version into the grep_source, taking care not
1428          * to leak any existing buffer.
1429          */
1430         grep_source_clear_data(gs);
1431         gs->buf = buf;
1432         gs->size = size;
1433
1434         return 0;
1435 }
1436
1437 static int is_empty_line(const char *bol, const char *eol)
1438 {
1439         while (bol < eol && isspace(*bol))
1440                 bol++;
1441         return bol == eol;
1442 }
1443
1444 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1445 {
1446         char *bol;
1447         char *peek_bol = NULL;
1448         unsigned long left;
1449         unsigned lno = 1;
1450         unsigned last_hit = 0;
1451         int binary_match_only = 0;
1452         unsigned count = 0;
1453         int try_lookahead = 0;
1454         int show_function = 0;
1455         struct userdiff_driver *textconv = NULL;
1456         enum grep_context ctx = GREP_CONTEXT_HEAD;
1457         xdemitconf_t xecfg;
1458
1459         if (!opt->output)
1460                 opt->output = std_output;
1461
1462         if (opt->pre_context || opt->post_context || opt->file_break ||
1463             opt->funcbody) {
1464                 /* Show hunk marks, except for the first file. */
1465                 if (opt->last_shown)
1466                         opt->show_hunk_mark = 1;
1467                 /*
1468                  * If we're using threads then we can't easily identify
1469                  * the first file.  Always put hunk marks in that case
1470                  * and skip the very first one later in work_done().
1471                  */
1472                 if (opt->output != std_output)
1473                         opt->show_hunk_mark = 1;
1474         }
1475         opt->last_shown = 0;
1476
1477         if (opt->allow_textconv) {
1478                 grep_source_load_driver(gs);
1479                 /*
1480                  * We might set up the shared textconv cache data here, which
1481                  * is not thread-safe.
1482                  */
1483                 grep_attr_lock();
1484                 textconv = userdiff_get_textconv(gs->driver);
1485                 grep_attr_unlock();
1486         }
1487
1488         /*
1489          * We know the result of a textconv is text, so we only have to care
1490          * about binary handling if we are not using it.
1491          */
1492         if (!textconv) {
1493                 switch (opt->binary) {
1494                 case GREP_BINARY_DEFAULT:
1495                         if (grep_source_is_binary(gs))
1496                                 binary_match_only = 1;
1497                         break;
1498                 case GREP_BINARY_NOMATCH:
1499                         if (grep_source_is_binary(gs))
1500                                 return 0; /* Assume unmatch */
1501                         break;
1502                 case GREP_BINARY_TEXT:
1503                         break;
1504                 default:
1505                         die("BUG: unknown binary handling mode");
1506                 }
1507         }
1508
1509         memset(&xecfg, 0, sizeof(xecfg));
1510         opt->priv = &xecfg;
1511
1512         try_lookahead = should_lookahead(opt);
1513
1514         if (fill_textconv_grep(textconv, gs) < 0)
1515                 return 0;
1516
1517         bol = gs->buf;
1518         left = gs->size;
1519         while (left) {
1520                 char *eol, ch;
1521                 int hit;
1522
1523                 /*
1524                  * look_ahead() skips quickly to the line that possibly
1525                  * has the next hit; don't call it if we need to do
1526                  * something more than just skipping the current line
1527                  * in response to an unmatch for the current line.  E.g.
1528                  * inside a post-context window, we will show the current
1529                  * line as a context around the previous hit when it
1530                  * doesn't hit.
1531                  */
1532                 if (try_lookahead
1533                     && !(last_hit
1534                          && (show_function ||
1535                              lno <= last_hit + opt->post_context))
1536                     && look_ahead(opt, &left, &lno, &bol))
1537                         break;
1538                 eol = end_of_line(bol, &left);
1539                 ch = *eol;
1540                 *eol = 0;
1541
1542                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1543                         ctx = GREP_CONTEXT_BODY;
1544
1545                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1546                 *eol = ch;
1547
1548                 if (collect_hits)
1549                         goto next_line;
1550
1551                 /* "grep -v -e foo -e bla" should list lines
1552                  * that do not have either, so inversion should
1553                  * be done outside.
1554                  */
1555                 if (opt->invert)
1556                         hit = !hit;
1557                 if (opt->unmatch_name_only) {
1558                         if (hit)
1559                                 return 0;
1560                         goto next_line;
1561                 }
1562                 if (hit) {
1563                         count++;
1564                         if (opt->status_only)
1565                                 return 1;
1566                         if (opt->name_only) {
1567                                 show_name(opt, gs->name);
1568                                 return 1;
1569                         }
1570                         if (opt->count)
1571                                 goto next_line;
1572                         if (binary_match_only) {
1573                                 opt->output(opt, "Binary file ", 12);
1574                                 output_color(opt, gs->name, strlen(gs->name),
1575                                              opt->color_filename);
1576                                 opt->output(opt, " matches\n", 9);
1577                                 return 1;
1578                         }
1579                         /* Hit at this line.  If we haven't shown the
1580                          * pre-context lines, we would need to show them.
1581                          */
1582                         if (opt->pre_context || opt->funcbody)
1583                                 show_pre_context(opt, gs, bol, eol, lno);
1584                         else if (opt->funcname)
1585                                 show_funcname_line(opt, gs, bol, lno);
1586                         show_line(opt, bol, eol, gs->name, lno, ':');
1587                         last_hit = lno;
1588                         if (opt->funcbody)
1589                                 show_function = 1;
1590                         goto next_line;
1591                 }
1592                 if (show_function && (!peek_bol || peek_bol < bol)) {
1593                         unsigned long peek_left = left;
1594                         char *peek_eol = eol;
1595
1596                         /*
1597                          * Trailing empty lines are not interesting.
1598                          * Peek past them to see if they belong to the
1599                          * body of the current function.
1600                          */
1601                         peek_bol = bol;
1602                         while (is_empty_line(peek_bol, peek_eol)) {
1603                                 peek_bol = peek_eol + 1;
1604                                 peek_eol = end_of_line(peek_bol, &peek_left);
1605                         }
1606
1607                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1608                                 show_function = 0;
1609                 }
1610                 if (show_function ||
1611                     (last_hit && lno <= last_hit + opt->post_context)) {
1612                         /* If the last hit is within the post context,
1613                          * we need to show this line.
1614                          */
1615                         show_line(opt, bol, eol, gs->name, lno, '-');
1616                 }
1617
1618         next_line:
1619                 bol = eol + 1;
1620                 if (!left)
1621                         break;
1622                 left--;
1623                 lno++;
1624         }
1625
1626         if (collect_hits)
1627                 return 0;
1628
1629         if (opt->status_only)
1630                 return 0;
1631         if (opt->unmatch_name_only) {
1632                 /* We did not see any hit, so we want to show this */
1633                 show_name(opt, gs->name);
1634                 return 1;
1635         }
1636
1637         xdiff_clear_find_func(&xecfg);
1638         opt->priv = NULL;
1639
1640         /* NEEDSWORK:
1641          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1642          * which feels mostly useless but sometimes useful.  Maybe
1643          * make it another option?  For now suppress them.
1644          */
1645         if (opt->count && count) {
1646                 char buf[32];
1647                 if (opt->pathname) {
1648                         output_color(opt, gs->name, strlen(gs->name),
1649                                      opt->color_filename);
1650                         output_sep(opt, ':');
1651                 }
1652                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1653                 opt->output(opt, buf, strlen(buf));
1654                 return 1;
1655         }
1656         return !!last_hit;
1657 }
1658
1659 static void clr_hit_marker(struct grep_expr *x)
1660 {
1661         /* All-hit markers are meaningful only at the very top level
1662          * OR node.
1663          */
1664         while (1) {
1665                 x->hit = 0;
1666                 if (x->node != GREP_NODE_OR)
1667                         return;
1668                 x->u.binary.left->hit = 0;
1669                 x = x->u.binary.right;
1670         }
1671 }
1672
1673 static int chk_hit_marker(struct grep_expr *x)
1674 {
1675         /* Top level nodes have hit markers.  See if they all are hits */
1676         while (1) {
1677                 if (x->node != GREP_NODE_OR)
1678                         return x->hit;
1679                 if (!x->u.binary.left->hit)
1680                         return 0;
1681                 x = x->u.binary.right;
1682         }
1683 }
1684
1685 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1686 {
1687         /*
1688          * we do not have to do the two-pass grep when we do not check
1689          * buffer-wide "all-match".
1690          */
1691         if (!opt->all_match)
1692                 return grep_source_1(opt, gs, 0);
1693
1694         /* Otherwise the toplevel "or" terms hit a bit differently.
1695          * We first clear hit markers from them.
1696          */
1697         clr_hit_marker(opt->pattern_expression);
1698         grep_source_1(opt, gs, 1);
1699
1700         if (!chk_hit_marker(opt->pattern_expression))
1701                 return 0;
1702
1703         return grep_source_1(opt, gs, 0);
1704 }
1705
1706 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1707 {
1708         struct grep_source gs;
1709         int r;
1710
1711         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1712         gs.buf = buf;
1713         gs.size = size;
1714
1715         r = grep_source(opt, &gs);
1716
1717         grep_source_clear(&gs);
1718         return r;
1719 }
1720
1721 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1722                       const char *name, const char *path,
1723                       const void *identifier)
1724 {
1725         gs->type = type;
1726         gs->name = xstrdup_or_null(name);
1727         gs->path = xstrdup_or_null(path);
1728         gs->buf = NULL;
1729         gs->size = 0;
1730         gs->driver = NULL;
1731
1732         switch (type) {
1733         case GREP_SOURCE_FILE:
1734                 gs->identifier = xstrdup(identifier);
1735                 break;
1736         case GREP_SOURCE_SUBMODULE:
1737                 if (!identifier) {
1738                         gs->identifier = NULL;
1739                         break;
1740                 }
1741                 /*
1742                  * FALL THROUGH
1743                  * If the identifier is non-NULL (in the submodule case) it
1744                  * will be a SHA1 that needs to be copied.
1745                  */
1746         case GREP_SOURCE_SHA1:
1747                 gs->identifier = xmalloc(20);
1748                 hashcpy(gs->identifier, identifier);
1749                 break;
1750         case GREP_SOURCE_BUF:
1751                 gs->identifier = NULL;
1752                 break;
1753         }
1754 }
1755
1756 void grep_source_clear(struct grep_source *gs)
1757 {
1758         free(gs->name);
1759         gs->name = NULL;
1760         free(gs->path);
1761         gs->path = NULL;
1762         free(gs->identifier);
1763         gs->identifier = NULL;
1764         grep_source_clear_data(gs);
1765 }
1766
1767 void grep_source_clear_data(struct grep_source *gs)
1768 {
1769         switch (gs->type) {
1770         case GREP_SOURCE_FILE:
1771         case GREP_SOURCE_SHA1:
1772         case GREP_SOURCE_SUBMODULE:
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         case GREP_SOURCE_SUBMODULE:
1845                 break;
1846         }
1847         die("BUG: invalid grep_source type to load");
1848 }
1849
1850 void grep_source_load_driver(struct grep_source *gs)
1851 {
1852         if (gs->driver)
1853                 return;
1854
1855         grep_attr_lock();
1856         if (gs->path)
1857                 gs->driver = userdiff_find_by_path(gs->path);
1858         if (!gs->driver)
1859                 gs->driver = userdiff_find_by_name("default");
1860         grep_attr_unlock();
1861 }
1862
1863 static int grep_source_is_binary(struct grep_source *gs)
1864 {
1865         grep_source_load_driver(gs);
1866         if (gs->driver->binary != -1)
1867                 return gs->driver->binary;
1868
1869         if (!grep_source_load(gs))
1870                 return buffer_is_binary(gs->buf, gs->size);
1871
1872         return 0;
1873 }