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->pcre1 = 0;
 182                 break;
 183
 184         case GREP_PATTERN_TYPE_ERE:
 185                 opt->fixed = 0;
 186                 opt->pcre1 = 0;
 187                 opt->regflags |= REG_EXTENDED;
 188                 break;
 189
 190         case GREP_PATTERN_TYPE_FIXED:
 191                 opt->fixed = 1;
 192                 opt->pcre1 = 0;
 193                 break;
 194
 195         case GREP_PATTERN_TYPE_PCRE:
 196                 opt->fixed = 0;
 197                 opt->pcre1 = 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 static int is_fixed(const char *s, size_t len)
 325 {
 326         size_t i;
 327
 328         for (i = 0; i < len; i++) {
 329                 if (is_regex_special(s[i]))
 330                         return 0;
 331         }
 332
 333         return 1;
 334 }
 335
 336 static int has_null(const char *s, size_t len)
 337 {
 338         /*
 339          * regcomp cannot accept patterns with NULs so when using it
 340          * we consider any pattern containing a NUL fixed.
 341          */
 342         if (memchr(s, 0, len))
 343                 return 1;
 344
 345         return 0;
 346 }
 347
 348 #ifdef USE_LIBPCRE1
 349 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 350 {
 351         const char *error;
 352         int erroffset;
 353         int options = PCRE_MULTILINE;
 354
 355         if (opt->ignore_case) {
 356                 if (has_non_ascii(p->pattern))
 357                         p->pcre1_tables = pcre_maketables();
 358                 options |= PCRE_CASELESS;
 359         }
 360         if (is_utf8_locale() && has_non_ascii(p->pattern))
 361                 options |= PCRE_UTF8;
 362
 363         p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 364                                       p->pcre1_tables);
 365         if (!p->pcre1_regexp)
 366                 compile_regexp_failed(p, error);
 367
 368         p->pcre1_extra_info = pcre_study(p->pcre1_regexp, PCRE_STUDY_JIT_COMPILE, &error);
 369         if (!p->pcre1_extra_info && error)
 370                 die("%s", error);
 371
 372 #ifdef PCRE_CONFIG_JIT
 373         pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
 374         if (p->pcre1_jit_on == 1) {
 375                 p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
 376                 if (!p->pcre1_jit_stack)
 377                         die("Couldn't allocate PCRE JIT stack");
 378                 pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
 379         } else if (p->pcre1_jit_on != 0) {
 380                 die("BUG: The pcre1_jit_on variable should be 0 or 1, not %d",
 381                     p->pcre1_jit_on);
 382         }
 383 #endif
 384 }
 385
 386 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 387                 regmatch_t *match, int eflags)
 388 {
 389         int ovector[30], ret, flags = 0;
 390
 391         if (eflags & REG_NOTBOL)
 392                 flags |= PCRE_NOTBOL;
 393
 394 #ifdef PCRE_CONFIG_JIT
 395         if (p->pcre1_jit_on) {
 396                 ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 397                                     eol - line, 0, flags, ovector,
 398                                     ARRAY_SIZE(ovector), p->pcre1_jit_stack);
 399         } else
 400 #endif
 401         {
 402                 ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 403                                 eol - line, 0, flags, ovector,
 404                                 ARRAY_SIZE(ovector));
 405         }
 406
 407         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 408                 die("pcre_exec failed with error code %d", ret);
 409         if (ret > 0) {
 410                 ret = 0;
 411                 match->rm_so = ovector[0];
 412                 match->rm_eo = ovector[1];
 413         }
 414
 415         return ret;
 416 }
 417
 418 static void free_pcre1_regexp(struct grep_pat *p)
 419 {
 420         pcre_free(p->pcre1_regexp);
 421 #ifdef PCRE_CONFIG_JIT
 422         if (p->pcre1_jit_on) {
 423                 pcre_free_study(p->pcre1_extra_info);
 424                 pcre_jit_stack_free(p->pcre1_jit_stack);
 425         } else
 426 #endif
 427         {
 428                 pcre_free(p->pcre1_extra_info);
 429         }
 430         pcre_free((void *)p->pcre1_tables);
 431 }
 432 #else /* !USE_LIBPCRE1 */
 433 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 434 {
 435         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 436 }
 437
 438 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 439                 regmatch_t *match, int eflags)
 440 {
 441         return 1;
 442 }
 443
 444 static void free_pcre1_regexp(struct grep_pat *p)
 445 {
 446 }
 447 #endif /* !USE_LIBPCRE1 */
 448
 449 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 450 {
 451         struct strbuf sb = STRBUF_INIT;
 452         int err;
 453         int regflags = opt->regflags;
 454
 455         basic_regex_quote_buf(&sb, p->pattern);
 456         if (opt->ignore_case)
 457                 regflags |= REG_ICASE;
 458         err = regcomp(&p->regexp, sb.buf, regflags);
 459         if (opt->debug)
 460                 fprintf(stderr, "fixed %s\n", sb.buf);
 461         strbuf_release(&sb);
 462         if (err) {
 463                 char errbuf[1024];
 464                 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 465                 regfree(&p->regexp);
 466                 compile_regexp_failed(p, errbuf);
 467         }
 468 }
 469
 470 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 471 {
 472         int icase, ascii_only;
 473         int err;
 474
 475         p->word_regexp = opt->word_regexp;
 476         p->ignore_case = opt->ignore_case;
 477         icase          = opt->regflags & REG_ICASE || p->ignore_case;
 478         ascii_only     = !has_non_ascii(p->pattern);
 479
 480         /*
 481          * Even when -F (fixed) asks us to do a non-regexp search, we
 482          * may not be able to correctly case-fold when -i
 483          * (ignore-case) is asked (in which case, we'll synthesize a
 484          * regexp to match the pattern that matches regexp special
 485          * characters literally, while ignoring case differences).  On
 486          * the other hand, even without -F, if the pattern does not
 487          * have any regexp special characters and there is no need for
 488          * case-folding search, we can internally turn it into a
 489          * simple string match using kws.  p->fixed tells us if we
 490          * want to use kws.
 491          */
 492         if (opt->fixed ||
 493             has_null(p->pattern, p->patternlen) ||
 494             is_fixed(p->pattern, p->patternlen))
 495                 p->fixed = !icase || ascii_only;
 496         else
 497                 p->fixed = 0;
 498
 499         if (p->fixed) {
 500                 p->kws = kwsalloc(icase ? tolower_trans_tbl : NULL);
 501                 kwsincr(p->kws, p->pattern, p->patternlen);
 502                 kwsprep(p->kws);
 503                 return;
 504         } else if (opt->fixed) {
 505                 /*
 506                  * We come here when the pattern has the non-ascii
 507                  * characters we cannot case-fold, and asked to
 508                  * ignore-case.
 509                  */
 510                 compile_fixed_regexp(p, opt);
 511                 return;
 512         }
 513
 514         if (opt->pcre1) {
 515                 compile_pcre1_regexp(p, opt);
 516                 return;
 517         }
 518
 519         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 520         if (err) {
 521                 char errbuf[1024];
 522                 regerror(err, &p->regexp, errbuf, 1024);
 523                 regfree(&p->regexp);
 524                 compile_regexp_failed(p, errbuf);
 525         }
 526 }
 527
 528 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 529 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 530 {
 531         struct grep_pat *p;
 532         struct grep_expr *x;
 533
 534         p = *list;
 535         if (!p)
 536                 return NULL;
 537         switch (p->token) {
 538         case GREP_PATTERN: /* atom */
 539         case GREP_PATTERN_HEAD:
 540         case GREP_PATTERN_BODY:
 541                 x = xcalloc(1, sizeof (struct grep_expr));
 542                 x->node = GREP_NODE_ATOM;
 543                 x->u.atom = p;
 544                 *list = p->next;
 545                 return x;
 546         case GREP_OPEN_PAREN:
 547                 *list = p->next;
 548                 x = compile_pattern_or(list);
 549                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 550                         die("unmatched parenthesis");
 551                 *list = (*list)->next;
 552                 return x;
 553         default:
 554                 return NULL;
 555         }
 556 }
 557
 558 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 559 {
 560         struct grep_pat *p;
 561         struct grep_expr *x;
 562
 563         p = *list;
 564         if (!p)
 565                 return NULL;
 566         switch (p->token) {
 567         case GREP_NOT:
 568                 if (!p->next)
 569                         die("--not not followed by pattern expression");
 570                 *list = p->next;
 571                 x = xcalloc(1, sizeof (struct grep_expr));
 572                 x->node = GREP_NODE_NOT;
 573                 x->u.unary = compile_pattern_not(list);
 574                 if (!x->u.unary)
 575                         die("--not followed by non pattern expression");
 576                 return x;
 577         default:
 578                 return compile_pattern_atom(list);
 579         }
 580 }
 581
 582 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 583 {
 584         struct grep_pat *p;
 585         struct grep_expr *x, *y, *z;
 586
 587         x = compile_pattern_not(list);
 588         p = *list;
 589         if (p && p->token == GREP_AND) {
 590                 if (!p->next)
 591                         die("--and not followed by pattern expression");
 592                 *list = p->next;
 593                 y = compile_pattern_and(list);
 594                 if (!y)
 595                         die("--and not followed by pattern expression");
 596                 z = xcalloc(1, sizeof (struct grep_expr));
 597                 z->node = GREP_NODE_AND;
 598                 z->u.binary.left = x;
 599                 z->u.binary.right = y;
 600                 return z;
 601         }
 602         return x;
 603 }
 604
 605 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 606 {
 607         struct grep_pat *p;
 608         struct grep_expr *x, *y, *z;
 609
 610         x = compile_pattern_and(list);
 611         p = *list;
 612         if (x && p && p->token != GREP_CLOSE_PAREN) {
 613                 y = compile_pattern_or(list);
 614                 if (!y)
 615                         die("not a pattern expression %s", p->pattern);
 616                 z = xcalloc(1, sizeof (struct grep_expr));
 617                 z->node = GREP_NODE_OR;
 618                 z->u.binary.left = x;
 619                 z->u.binary.right = y;
 620                 return z;
 621         }
 622         return x;
 623 }
 624
 625 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 626 {
 627         return compile_pattern_or(list);
 628 }
 629
 630 static void indent(int in)
 631 {
 632         while (in-- > 0)
 633                 fputc(' ', stderr);
 634 }
 635
 636 static void dump_grep_pat(struct grep_pat *p)
 637 {
 638         switch (p->token) {
 639         case GREP_AND: fprintf(stderr, "*and*"); break;
 640         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 641         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 642         case GREP_NOT: fprintf(stderr, "*not*"); break;
 643         case GREP_OR: fprintf(stderr, "*or*"); break;
 644
 645         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 646         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 647         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 648         }
 649
 650         switch (p->token) {
 651         default: break;
 652         case GREP_PATTERN_HEAD:
 653                 fprintf(stderr, "<head %d>", p->field); break;
 654         case GREP_PATTERN_BODY:
 655                 fprintf(stderr, "<body>"); break;
 656         }
 657         switch (p->token) {
 658         default: break;
 659         case GREP_PATTERN_HEAD:
 660         case GREP_PATTERN_BODY:
 661         case GREP_PATTERN:
 662                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 663                 break;
 664         }
 665         fputc('\n', stderr);
 666 }
 667
 668 static void dump_grep_expression_1(struct grep_expr *x, int in)
 669 {
 670         indent(in);
 671         switch (x->node) {
 672         case GREP_NODE_TRUE:
 673                 fprintf(stderr, "true\n");
 674                 break;
 675         case GREP_NODE_ATOM:
 676                 dump_grep_pat(x->u.atom);
 677                 break;
 678         case GREP_NODE_NOT:
 679                 fprintf(stderr, "(not\n");
 680                 dump_grep_expression_1(x->u.unary, in+1);
 681                 indent(in);
 682                 fprintf(stderr, ")\n");
 683                 break;
 684         case GREP_NODE_AND:
 685                 fprintf(stderr, "(and\n");
 686                 dump_grep_expression_1(x->u.binary.left, in+1);
 687                 dump_grep_expression_1(x->u.binary.right, in+1);
 688                 indent(in);
 689                 fprintf(stderr, ")\n");
 690                 break;
 691         case GREP_NODE_OR:
 692                 fprintf(stderr, "(or\n");
 693                 dump_grep_expression_1(x->u.binary.left, in+1);
 694                 dump_grep_expression_1(x->u.binary.right, in+1);
 695                 indent(in);
 696                 fprintf(stderr, ")\n");
 697                 break;
 698         }
 699 }
 700
 701 static void dump_grep_expression(struct grep_opt *opt)
 702 {
 703         struct grep_expr *x = opt->pattern_expression;
 704
 705         if (opt->all_match)
 706                 fprintf(stderr, "[all-match]\n");
 707         dump_grep_expression_1(x, 0);
 708         fflush(NULL);
 709 }
 710
 711 static struct grep_expr *grep_true_expr(void)
 712 {
 713         struct grep_expr *z = xcalloc(1, sizeof(*z));
 714         z->node = GREP_NODE_TRUE;
 715         return z;
 716 }
 717
 718 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 719 {
 720         struct grep_expr *z = xcalloc(1, sizeof(*z));
 721         z->node = GREP_NODE_OR;
 722         z->u.binary.left = left;
 723         z->u.binary.right = right;
 724         return z;
 725 }
 726
 727 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 728 {
 729         struct grep_pat *p;
 730         struct grep_expr *header_expr;
 731         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 732         enum grep_header_field fld;
 733
 734         if (!opt->header_list)
 735                 return NULL;
 736
 737         for (p = opt->header_list; p; p = p->next) {
 738                 if (p->token != GREP_PATTERN_HEAD)
 739                         die("BUG: a non-header pattern in grep header list.");
 740                 if (p->field < GREP_HEADER_FIELD_MIN ||
 741                     GREP_HEADER_FIELD_MAX <= p->field)
 742                         die("BUG: unknown header field %d", p->field);
 743                 compile_regexp(p, opt);
 744         }
 745
 746         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 747                 header_group[fld] = NULL;
 748
 749         for (p = opt->header_list; p; p = p->next) {
 750                 struct grep_expr *h;
 751                 struct grep_pat *pp = p;
 752
 753                 h = compile_pattern_atom(&pp);
 754                 if (!h || pp != p->next)
 755                         die("BUG: malformed header expr");
 756                 if (!header_group[p->field]) {
 757                         header_group[p->field] = h;
 758                         continue;
 759                 }
 760                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 761         }
 762
 763         header_expr = NULL;
 764
 765         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 766                 if (!header_group[fld])
 767                         continue;
 768                 if (!header_expr)
 769                         header_expr = grep_true_expr();
 770                 header_expr = grep_or_expr(header_group[fld], header_expr);
 771         }
 772         return header_expr;
 773 }
 774
 775 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 776 {
 777         struct grep_expr *z = x;
 778
 779         while (x) {
 780                 assert(x->node == GREP_NODE_OR);
 781                 if (x->u.binary.right &&
 782                     x->u.binary.right->node == GREP_NODE_TRUE) {
 783                         x->u.binary.right = y;
 784                         break;
 785                 }
 786                 x = x->u.binary.right;
 787         }
 788         return z;
 789 }
 790
 791 static void compile_grep_patterns_real(struct grep_opt *opt)
 792 {
 793         struct grep_pat *p;
 794         struct grep_expr *header_expr = prep_header_patterns(opt);
 795
 796         for (p = opt->pattern_list; p; p = p->next) {
 797                 switch (p->token) {
 798                 case GREP_PATTERN: /* atom */
 799                 case GREP_PATTERN_HEAD:
 800                 case GREP_PATTERN_BODY:
 801                         compile_regexp(p, opt);
 802                         break;
 803                 default:
 804                         opt->extended = 1;
 805                         break;
 806                 }
 807         }
 808
 809         if (opt->all_match || header_expr)
 810                 opt->extended = 1;
 811         else if (!opt->extended && !opt->debug)
 812                 return;
 813
 814         p = opt->pattern_list;
 815         if (p)
 816                 opt->pattern_expression = compile_pattern_expr(&p);
 817         if (p)
 818                 die("incomplete pattern expression: %s", p->pattern);
 819
 820         if (!header_expr)
 821                 return;
 822
 823         if (!opt->pattern_expression)
 824                 opt->pattern_expression = header_expr;
 825         else if (opt->all_match)
 826                 opt->pattern_expression = grep_splice_or(header_expr,
 827                                                          opt->pattern_expression);
 828         else
 829                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 830                                                        header_expr);
 831         opt->all_match = 1;
 832 }
 833
 834 void compile_grep_patterns(struct grep_opt *opt)
 835 {
 836         compile_grep_patterns_real(opt);
 837         if (opt->debug)
 838                 dump_grep_expression(opt);
 839 }
 840
 841 static void free_pattern_expr(struct grep_expr *x)
 842 {
 843         switch (x->node) {
 844         case GREP_NODE_TRUE:
 845         case GREP_NODE_ATOM:
 846                 break;
 847         case GREP_NODE_NOT:
 848                 free_pattern_expr(x->u.unary);
 849                 break;
 850         case GREP_NODE_AND:
 851         case GREP_NODE_OR:
 852                 free_pattern_expr(x->u.binary.left);
 853                 free_pattern_expr(x->u.binary.right);
 854                 break;
 855         }
 856         free(x);
 857 }
 858
 859 void free_grep_patterns(struct grep_opt *opt)
 860 {
 861         struct grep_pat *p, *n;
 862
 863         for (p = opt->pattern_list; p; p = n) {
 864                 n = p->next;
 865                 switch (p->token) {
 866                 case GREP_PATTERN: /* atom */
 867                 case GREP_PATTERN_HEAD:
 868                 case GREP_PATTERN_BODY:
 869                         if (p->kws)
 870                                 kwsfree(p->kws);
 871                         else if (p->pcre1_regexp)
 872                                 free_pcre1_regexp(p);
 873                         else
 874                                 regfree(&p->regexp);
 875                         free(p->pattern);
 876                         break;
 877                 default:
 878                         break;
 879                 }
 880                 free(p);
 881         }
 882
 883         if (!opt->extended)
 884                 return;
 885         free_pattern_expr(opt->pattern_expression);
 886 }
 887
 888 static char *end_of_line(char *cp, unsigned long *left)
 889 {
 890         unsigned long l = *left;
 891         while (l && *cp != '\n') {
 892                 l--;
 893                 cp++;
 894         }
 895         *left = l;
 896         return cp;
 897 }
 898
 899 static int word_char(char ch)
 900 {
 901         return isalnum(ch) || ch == '_';
 902 }
 903
 904 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 905                          const char *color)
 906 {
 907         if (want_color(opt->color) && color && color[0]) {
 908                 opt->output(opt, color, strlen(color));
 909                 opt->output(opt, data, size);
 910                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 911         } else
 912                 opt->output(opt, data, size);
 913 }
 914
 915 static void output_sep(struct grep_opt *opt, char sign)
 916 {
 917         if (opt->null_following_name)
 918                 opt->output(opt, "\0", 1);
 919         else
 920                 output_color(opt, &sign, 1, opt->color_sep);
 921 }
 922
 923 static void show_name(struct grep_opt *opt, const char *name)
 924 {
 925         output_color(opt, name, strlen(name), opt->color_filename);
 926         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 927 }
 928
 929 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 930                     regmatch_t *match)
 931 {
 932         struct kwsmatch kwsm;
 933         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 934         if (offset == -1) {
 935                 match->rm_so = match->rm_eo = -1;
 936                 return REG_NOMATCH;
 937         } else {
 938                 match->rm_so = offset;
 939                 match->rm_eo = match->rm_so + kwsm.size[0];
 940                 return 0;
 941         }
 942 }
 943
 944 static int patmatch(struct grep_pat *p, char *line, char *eol,
 945                     regmatch_t *match, int eflags)
 946 {
 947         int hit;
 948
 949         if (p->fixed)
 950                 hit = !fixmatch(p, line, eol, match);
 951         else if (p->pcre1_regexp)
 952                 hit = !pcre1match(p, line, eol, match, eflags);
 953         else
 954                 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
 955                                    eflags);
 956
 957         return hit;
 958 }
 959
 960 static int strip_timestamp(char *bol, char **eol_p)
 961 {
 962         char *eol = *eol_p;
 963         int ch;
 964
 965         while (bol < --eol) {
 966                 if (*eol != '>')
 967                         continue;
 968                 *eol_p = ++eol;
 969                 ch = *eol;
 970                 *eol = '\0';
 971                 return ch;
 972         }
 973         return 0;
 974 }
 975
 976 static struct {
 977         const char *field;
 978         size_t len;
 979 } header_field[] = {
 980         { "author ", 7 },
 981         { "committer ", 10 },
 982         { "reflog ", 7 },
 983 };
 984
 985 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 986                              enum grep_context ctx,
 987                              regmatch_t *pmatch, int eflags)
 988 {
 989         int hit = 0;
 990         int saved_ch = 0;
 991         const char *start = bol;
 992
 993         if ((p->token != GREP_PATTERN) &&
 994             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 995                 return 0;
 996
 997         if (p->token == GREP_PATTERN_HEAD) {
 998                 const char *field;
 999                 size_t len;
1000                 assert(p->field < ARRAY_SIZE(header_field));
1001                 field = header_field[p->field].field;
1002                 len = header_field[p->field].len;
1003                 if (strncmp(bol, field, len))
1004                         return 0;
1005                 bol += len;
1006                 switch (p->field) {
1007                 case GREP_HEADER_AUTHOR:
1008                 case GREP_HEADER_COMMITTER:
1009                         saved_ch = strip_timestamp(bol, &eol);
1010                         break;
1011                 default:
1012                         break;
1013                 }
1014         }
1015
1016  again:
1017         hit = patmatch(p, bol, eol, pmatch, eflags);
1018
1019         if (hit && p->word_regexp) {
1020                 if ((pmatch[0].rm_so < 0) ||
1021                     (eol - bol) < pmatch[0].rm_so ||
1022                     (pmatch[0].rm_eo < 0) ||
1023                     (eol - bol) < pmatch[0].rm_eo)
1024                         die("regexp returned nonsense");
1025
1026                 /* Match beginning must be either beginning of the
1027                  * line, or at word boundary (i.e. the last char must
1028                  * not be a word char).  Similarly, match end must be
1029                  * either end of the line, or at word boundary
1030                  * (i.e. the next char must not be a word char).
1031                  */
1032                 if ( ((pmatch[0].rm_so == 0) ||
1033                       !word_char(bol[pmatch[0].rm_so-1])) &&
1034                      ((pmatch[0].rm_eo == (eol-bol)) ||
1035                       !word_char(bol[pmatch[0].rm_eo])) )
1036                         ;
1037                 else
1038                         hit = 0;
1039
1040                 /* Words consist of at least one character. */
1041                 if (pmatch->rm_so == pmatch->rm_eo)
1042                         hit = 0;
1043
1044                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1045                         /* There could be more than one match on the
1046                          * line, and the first match might not be
1047                          * strict word match.  But later ones could be!
1048                          * Forward to the next possible start, i.e. the
1049                          * next position following a non-word char.
1050                          */
1051                         bol = pmatch[0].rm_so + bol + 1;
1052                         while (word_char(bol[-1]) && bol < eol)
1053                                 bol++;
1054                         eflags |= REG_NOTBOL;
1055                         if (bol < eol)
1056                                 goto again;
1057                 }
1058         }
1059         if (p->token == GREP_PATTERN_HEAD && saved_ch)
1060                 *eol = saved_ch;
1061         if (hit) {
1062                 pmatch[0].rm_so += bol - start;
1063                 pmatch[0].rm_eo += bol - start;
1064         }
1065         return hit;
1066 }
1067
1068 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1069                            enum grep_context ctx, int collect_hits)
1070 {
1071         int h = 0;
1072         regmatch_t match;
1073
1074         if (!x)
1075                 die("Not a valid grep expression");
1076         switch (x->node) {
1077         case GREP_NODE_TRUE:
1078                 h = 1;
1079                 break;
1080         case GREP_NODE_ATOM:
1081                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1082                 break;
1083         case GREP_NODE_NOT:
1084                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1085                 break;
1086         case GREP_NODE_AND:
1087                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1088                         return 0;
1089                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1090                 break;
1091         case GREP_NODE_OR:
1092                 if (!collect_hits)
1093                         return (match_expr_eval(x->u.binary.left,
1094                                                 bol, eol, ctx, 0) ||
1095                                 match_expr_eval(x->u.binary.right,
1096                                                 bol, eol, ctx, 0));
1097                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1098                 x->u.binary.left->hit |= h;
1099                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1100                 break;
1101         default:
1102                 die("Unexpected node type (internal error) %d", x->node);
1103         }
1104         if (collect_hits)
1105                 x->hit |= h;
1106         return h;
1107 }
1108
1109 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1110                       enum grep_context ctx, int collect_hits)
1111 {
1112         struct grep_expr *x = opt->pattern_expression;
1113         return match_expr_eval(x, bol, eol, ctx, collect_hits);
1114 }
1115
1116 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1117                       enum grep_context ctx, int collect_hits)
1118 {
1119         struct grep_pat *p;
1120         regmatch_t match;
1121
1122         if (opt->extended)
1123                 return match_expr(opt, bol, eol, ctx, collect_hits);
1124
1125         /* we do not call with collect_hits without being extended */
1126         for (p = opt->pattern_list; p; p = p->next) {
1127                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1128                         return 1;
1129         }
1130         return 0;
1131 }
1132
1133 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1134                               enum grep_context ctx,
1135                               regmatch_t *pmatch, int eflags)
1136 {
1137         regmatch_t match;
1138
1139         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1140                 return 0;
1141         if (match.rm_so < 0 || match.rm_eo < 0)
1142                 return 0;
1143         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1144                 if (match.rm_so > pmatch->rm_so)
1145                         return 1;
1146                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1147                         return 1;
1148         }
1149         pmatch->rm_so = match.rm_so;
1150         pmatch->rm_eo = match.rm_eo;
1151         return 1;
1152 }
1153
1154 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1155                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1156 {
1157         struct grep_pat *p;
1158         int hit = 0;
1159
1160         pmatch->rm_so = pmatch->rm_eo = -1;
1161         if (bol < eol) {
1162                 for (p = opt->pattern_list; p; p = p->next) {
1163                         switch (p->token) {
1164                         case GREP_PATTERN: /* atom */
1165                         case GREP_PATTERN_HEAD:
1166                         case GREP_PATTERN_BODY:
1167                                 hit |= match_next_pattern(p, bol, eol, ctx,
1168                                                           pmatch, eflags);
1169                                 break;
1170                         default:
1171                                 break;
1172                         }
1173                 }
1174         }
1175         return hit;
1176 }
1177
1178 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1179                       const char *name, unsigned lno, char sign)
1180 {
1181         int rest = eol - bol;
1182         const char *match_color, *line_color = NULL;
1183
1184         if (opt->file_break && opt->last_shown == 0) {
1185                 if (opt->show_hunk_mark)
1186                         opt->output(opt, "\n", 1);
1187         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1188                 if (opt->last_shown == 0) {
1189                         if (opt->show_hunk_mark) {
1190                                 output_color(opt, "--", 2, opt->color_sep);
1191                                 opt->output(opt, "\n", 1);
1192                         }
1193                 } else if (lno > opt->last_shown + 1) {
1194                         output_color(opt, "--", 2, opt->color_sep);
1195                         opt->output(opt, "\n", 1);
1196                 }
1197         }
1198         if (opt->heading && opt->last_shown == 0) {
1199                 output_color(opt, name, strlen(name), opt->color_filename);
1200                 opt->output(opt, "\n", 1);
1201         }
1202         opt->last_shown = lno;
1203
1204         if (!opt->heading && opt->pathname) {
1205                 output_color(opt, name, strlen(name), opt->color_filename);
1206                 output_sep(opt, sign);
1207         }
1208         if (opt->linenum) {
1209                 char buf[32];
1210                 xsnprintf(buf, sizeof(buf), "%d", lno);
1211                 output_color(opt, buf, strlen(buf), opt->color_lineno);
1212                 output_sep(opt, sign);
1213         }
1214         if (opt->color) {
1215                 regmatch_t match;
1216                 enum grep_context ctx = GREP_CONTEXT_BODY;
1217                 int ch = *eol;
1218                 int eflags = 0;
1219
1220                 if (sign == ':')
1221                         match_color = opt->color_match_selected;
1222                 else
1223                         match_color = opt->color_match_context;
1224                 if (sign == ':')
1225                         line_color = opt->color_selected;
1226                 else if (sign == '-')
1227                         line_color = opt->color_context;
1228                 else if (sign == '=')
1229                         line_color = opt->color_function;
1230                 *eol = '\0';
1231                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1232                         if (match.rm_so == match.rm_eo)
1233                                 break;
1234
1235                         output_color(opt, bol, match.rm_so, line_color);
1236                         output_color(opt, bol + match.rm_so,
1237                                      match.rm_eo - match.rm_so, match_color);
1238                         bol += match.rm_eo;
1239                         rest -= match.rm_eo;
1240                         eflags = REG_NOTBOL;
1241                 }
1242                 *eol = ch;
1243         }
1244         output_color(opt, bol, rest, line_color);
1245         opt->output(opt, "\n", 1);
1246 }
1247
1248 #ifndef NO_PTHREADS
1249 int grep_use_locks;
1250
1251 /*
1252  * This lock protects access to the gitattributes machinery, which is
1253  * not thread-safe.
1254  */
1255 pthread_mutex_t grep_attr_mutex;
1256
1257 static inline void grep_attr_lock(void)
1258 {
1259         if (grep_use_locks)
1260                 pthread_mutex_lock(&grep_attr_mutex);
1261 }
1262
1263 static inline void grep_attr_unlock(void)
1264 {
1265         if (grep_use_locks)
1266                 pthread_mutex_unlock(&grep_attr_mutex);
1267 }
1268
1269 /*
1270  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1271  */
1272 pthread_mutex_t grep_read_mutex;
1273
1274 #else
1275 #define grep_attr_lock()
1276 #define grep_attr_unlock()
1277 #endif
1278
1279 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1280 {
1281         xdemitconf_t *xecfg = opt->priv;
1282         if (xecfg && !xecfg->find_func) {
1283                 grep_source_load_driver(gs);
1284                 if (gs->driver->funcname.pattern) {
1285                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1286                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1287                 } else {
1288                         xecfg = opt->priv = NULL;
1289                 }
1290         }
1291
1292         if (xecfg) {
1293                 char buf[1];
1294                 return xecfg->find_func(bol, eol - bol, buf, 1,
1295                                         xecfg->find_func_priv) >= 0;
1296         }
1297
1298         if (bol == eol)
1299                 return 0;
1300         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1301                 return 1;
1302         return 0;
1303 }
1304
1305 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1306                                char *bol, unsigned lno)
1307 {
1308         while (bol > gs->buf) {
1309                 char *eol = --bol;
1310
1311                 while (bol > gs->buf && bol[-1] != '\n')
1312                         bol--;
1313                 lno--;
1314
1315                 if (lno <= opt->last_shown)
1316                         break;
1317
1318                 if (match_funcname(opt, gs, bol, eol)) {
1319                         show_line(opt, bol, eol, gs->name, lno, '=');
1320                         break;
1321                 }
1322         }
1323 }
1324
1325 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1326                              char *bol, char *end, unsigned lno)
1327 {
1328         unsigned cur = lno, from = 1, funcname_lno = 0;
1329         int funcname_needed = !!opt->funcname;
1330
1331         if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1332                 funcname_needed = 2;
1333
1334         if (opt->pre_context < lno)
1335                 from = lno - opt->pre_context;
1336         if (from <= opt->last_shown)
1337                 from = opt->last_shown + 1;
1338
1339         /* Rewind. */
1340         while (bol > gs->buf &&
1341                cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1342                 char *eol = --bol;
1343
1344                 while (bol > gs->buf && bol[-1] != '\n')
1345                         bol--;
1346                 cur--;
1347                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1348                         funcname_lno = cur;
1349                         funcname_needed = 0;
1350                 }
1351         }
1352
1353         /* We need to look even further back to find a function signature. */
1354         if (opt->funcname && funcname_needed)
1355                 show_funcname_line(opt, gs, bol, cur);
1356
1357         /* Back forward. */
1358         while (cur < lno) {
1359                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1360
1361                 while (*eol != '\n')
1362                         eol++;
1363                 show_line(opt, bol, eol, gs->name, cur, sign);
1364                 bol = eol + 1;
1365                 cur++;
1366         }
1367 }
1368
1369 static int should_lookahead(struct grep_opt *opt)
1370 {
1371         struct grep_pat *p;
1372
1373         if (opt->extended)
1374                 return 0; /* punt for too complex stuff */
1375         if (opt->invert)
1376                 return 0;
1377         for (p = opt->pattern_list; p; p = p->next) {
1378                 if (p->token != GREP_PATTERN)
1379                         return 0; /* punt for "header only" and stuff */
1380         }
1381         return 1;
1382 }
1383
1384 static int look_ahead(struct grep_opt *opt,
1385                       unsigned long *left_p,
1386                       unsigned *lno_p,
1387                       char **bol_p)
1388 {
1389         unsigned lno = *lno_p;
1390         char *bol = *bol_p;
1391         struct grep_pat *p;
1392         char *sp, *last_bol;
1393         regoff_t earliest = -1;
1394
1395         for (p = opt->pattern_list; p; p = p->next) {
1396                 int hit;
1397                 regmatch_t m;
1398
1399                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1400                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1401                         continue;
1402                 if (earliest < 0 || m.rm_so < earliest)
1403                         earliest = m.rm_so;
1404         }
1405
1406         if (earliest < 0) {
1407                 *bol_p = bol + *left_p;
1408                 *left_p = 0;
1409                 return 1;
1410         }
1411         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1412                 ; /* find the beginning of the line */
1413         last_bol = sp;
1414
1415         for (sp = bol; sp < last_bol; sp++) {
1416                 if (*sp == '\n')
1417                         lno++;
1418         }
1419         *left_p -= last_bol - bol;
1420         *bol_p = last_bol;
1421         *lno_p = lno;
1422         return 0;
1423 }
1424
1425 static int fill_textconv_grep(struct userdiff_driver *driver,
1426                               struct grep_source *gs)
1427 {
1428         struct diff_filespec *df;
1429         char *buf;
1430         size_t size;
1431
1432         if (!driver || !driver->textconv)
1433                 return grep_source_load(gs);
1434
1435         /*
1436          * The textconv interface is intimately tied to diff_filespecs, so we
1437          * have to pretend to be one. If we could unify the grep_source
1438          * and diff_filespec structs, this mess could just go away.
1439          */
1440         df = alloc_filespec(gs->path);
1441         switch (gs->type) {
1442         case GREP_SOURCE_SHA1:
1443                 fill_filespec(df, gs->identifier, 1, 0100644);
1444                 break;
1445         case GREP_SOURCE_FILE:
1446                 fill_filespec(df, null_sha1, 0, 0100644);
1447                 break;
1448         default:
1449                 die("BUG: attempt to textconv something without a path?");
1450         }
1451
1452         /*
1453          * fill_textconv is not remotely thread-safe; it may load objects
1454          * behind the scenes, and it modifies the global diff tempfile
1455          * structure.
1456          */
1457         grep_read_lock();
1458         size = fill_textconv(driver, df, &buf);
1459         grep_read_unlock();
1460         free_filespec(df);
1461
1462         /*
1463          * The normal fill_textconv usage by the diff machinery would just keep
1464          * the textconv'd buf separate from the diff_filespec. But much of the
1465          * grep code passes around a grep_source and assumes that its "buf"
1466          * pointer is the beginning of the thing we are searching. So let's
1467          * install our textconv'd version into the grep_source, taking care not
1468          * to leak any existing buffer.
1469          */
1470         grep_source_clear_data(gs);
1471         gs->buf = buf;
1472         gs->size = size;
1473
1474         return 0;
1475 }
1476
1477 static int is_empty_line(const char *bol, const char *eol)
1478 {
1479         while (bol < eol && isspace(*bol))
1480                 bol++;
1481         return bol == eol;
1482 }
1483
1484 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1485 {
1486         char *bol;
1487         char *peek_bol = NULL;
1488         unsigned long left;
1489         unsigned lno = 1;
1490         unsigned last_hit = 0;
1491         int binary_match_only = 0;
1492         unsigned count = 0;
1493         int try_lookahead = 0;
1494         int show_function = 0;
1495         struct userdiff_driver *textconv = NULL;
1496         enum grep_context ctx = GREP_CONTEXT_HEAD;
1497         xdemitconf_t xecfg;
1498
1499         if (!opt->output)
1500                 opt->output = std_output;
1501
1502         if (opt->pre_context || opt->post_context || opt->file_break ||
1503             opt->funcbody) {
1504                 /* Show hunk marks, except for the first file. */
1505                 if (opt->last_shown)
1506                         opt->show_hunk_mark = 1;
1507                 /*
1508                  * If we're using threads then we can't easily identify
1509                  * the first file.  Always put hunk marks in that case
1510                  * and skip the very first one later in work_done().
1511                  */
1512                 if (opt->output != std_output)
1513                         opt->show_hunk_mark = 1;
1514         }
1515         opt->last_shown = 0;
1516
1517         if (opt->allow_textconv) {
1518                 grep_source_load_driver(gs);
1519                 /*
1520                  * We might set up the shared textconv cache data here, which
1521                  * is not thread-safe.
1522                  */
1523                 grep_attr_lock();
1524                 textconv = userdiff_get_textconv(gs->driver);
1525                 grep_attr_unlock();
1526         }
1527
1528         /*
1529          * We know the result of a textconv is text, so we only have to care
1530          * about binary handling if we are not using it.
1531          */
1532         if (!textconv) {
1533                 switch (opt->binary) {
1534                 case GREP_BINARY_DEFAULT:
1535                         if (grep_source_is_binary(gs))
1536                                 binary_match_only = 1;
1537                         break;
1538                 case GREP_BINARY_NOMATCH:
1539                         if (grep_source_is_binary(gs))
1540                                 return 0; /* Assume unmatch */
1541                         break;
1542                 case GREP_BINARY_TEXT:
1543                         break;
1544                 default:
1545                         die("BUG: unknown binary handling mode");
1546                 }
1547         }
1548
1549         memset(&xecfg, 0, sizeof(xecfg));
1550         opt->priv = &xecfg;
1551
1552         try_lookahead = should_lookahead(opt);
1553
1554         if (fill_textconv_grep(textconv, gs) < 0)
1555                 return 0;
1556
1557         bol = gs->buf;
1558         left = gs->size;
1559         while (left) {
1560                 char *eol, ch;
1561                 int hit;
1562
1563                 /*
1564                  * look_ahead() skips quickly to the line that possibly
1565                  * has the next hit; don't call it if we need to do
1566                  * something more than just skipping the current line
1567                  * in response to an unmatch for the current line.  E.g.
1568                  * inside a post-context window, we will show the current
1569                  * line as a context around the previous hit when it
1570                  * doesn't hit.
1571                  */
1572                 if (try_lookahead
1573                     && !(last_hit
1574                          && (show_function ||
1575                              lno <= last_hit + opt->post_context))
1576                     && look_ahead(opt, &left, &lno, &bol))
1577                         break;
1578                 eol = end_of_line(bol, &left);
1579                 ch = *eol;
1580                 *eol = 0;
1581
1582                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1583                         ctx = GREP_CONTEXT_BODY;
1584
1585                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1586                 *eol = ch;
1587
1588                 if (collect_hits)
1589                         goto next_line;
1590
1591                 /* "grep -v -e foo -e bla" should list lines
1592                  * that do not have either, so inversion should
1593                  * be done outside.
1594                  */
1595                 if (opt->invert)
1596                         hit = !hit;
1597                 if (opt->unmatch_name_only) {
1598                         if (hit)
1599                                 return 0;
1600                         goto next_line;
1601                 }
1602                 if (hit) {
1603                         count++;
1604                         if (opt->status_only)
1605                                 return 1;
1606                         if (opt->name_only) {
1607                                 show_name(opt, gs->name);
1608                                 return 1;
1609                         }
1610                         if (opt->count)
1611                                 goto next_line;
1612                         if (binary_match_only) {
1613                                 opt->output(opt, "Binary file ", 12);
1614                                 output_color(opt, gs->name, strlen(gs->name),
1615                                              opt->color_filename);
1616                                 opt->output(opt, " matches\n", 9);
1617                                 return 1;
1618                         }
1619                         /* Hit at this line.  If we haven't shown the
1620                          * pre-context lines, we would need to show them.
1621                          */
1622                         if (opt->pre_context || opt->funcbody)
1623                                 show_pre_context(opt, gs, bol, eol, lno);
1624                         else if (opt->funcname)
1625                                 show_funcname_line(opt, gs, bol, lno);
1626                         show_line(opt, bol, eol, gs->name, lno, ':');
1627                         last_hit = lno;
1628                         if (opt->funcbody)
1629                                 show_function = 1;
1630                         goto next_line;
1631                 }
1632                 if (show_function && (!peek_bol || peek_bol < bol)) {
1633                         unsigned long peek_left = left;
1634                         char *peek_eol = eol;
1635
1636                         /*
1637                          * Trailing empty lines are not interesting.
1638                          * Peek past them to see if they belong to the
1639                          * body of the current function.
1640                          */
1641                         peek_bol = bol;
1642                         while (is_empty_line(peek_bol, peek_eol)) {
1643                                 peek_bol = peek_eol + 1;
1644                                 peek_eol = end_of_line(peek_bol, &peek_left);
1645                         }
1646
1647                         if (match_funcname(opt, gs, peek_bol, peek_eol))
1648                                 show_function = 0;
1649                 }
1650                 if (show_function ||
1651                     (last_hit && lno <= last_hit + opt->post_context)) {
1652                         /* If the last hit is within the post context,
1653                          * we need to show this line.
1654                          */
1655                         show_line(opt, bol, eol, gs->name, lno, '-');
1656                 }
1657
1658         next_line:
1659                 bol = eol + 1;
1660                 if (!left)
1661                         break;
1662                 left--;
1663                 lno++;
1664         }
1665
1666         if (collect_hits)
1667                 return 0;
1668
1669         if (opt->status_only)
1670                 return 0;
1671         if (opt->unmatch_name_only) {
1672                 /* We did not see any hit, so we want to show this */
1673                 show_name(opt, gs->name);
1674                 return 1;
1675         }
1676
1677         xdiff_clear_find_func(&xecfg);
1678         opt->priv = NULL;
1679
1680         /* NEEDSWORK:
1681          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1682          * which feels mostly useless but sometimes useful.  Maybe
1683          * make it another option?  For now suppress them.
1684          */
1685         if (opt->count && count) {
1686                 char buf[32];
1687                 if (opt->pathname) {
1688                         output_color(opt, gs->name, strlen(gs->name),
1689                                      opt->color_filename);
1690                         output_sep(opt, ':');
1691                 }
1692                 xsnprintf(buf, sizeof(buf), "%u\n", count);
1693                 opt->output(opt, buf, strlen(buf));
1694                 return 1;
1695         }
1696         return !!last_hit;
1697 }
1698
1699 static void clr_hit_marker(struct grep_expr *x)
1700 {
1701         /* All-hit markers are meaningful only at the very top level
1702          * OR node.
1703          */
1704         while (1) {
1705                 x->hit = 0;
1706                 if (x->node != GREP_NODE_OR)
1707                         return;
1708                 x->u.binary.left->hit = 0;
1709                 x = x->u.binary.right;
1710         }
1711 }
1712
1713 static int chk_hit_marker(struct grep_expr *x)
1714 {
1715         /* Top level nodes have hit markers.  See if they all are hits */
1716         while (1) {
1717                 if (x->node != GREP_NODE_OR)
1718                         return x->hit;
1719                 if (!x->u.binary.left->hit)
1720                         return 0;
1721                 x = x->u.binary.right;
1722         }
1723 }
1724
1725 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1726 {
1727         /*
1728          * we do not have to do the two-pass grep when we do not check
1729          * buffer-wide "all-match".
1730          */
1731         if (!opt->all_match)
1732                 return grep_source_1(opt, gs, 0);
1733
1734         /* Otherwise the toplevel "or" terms hit a bit differently.
1735          * We first clear hit markers from them.
1736          */
1737         clr_hit_marker(opt->pattern_expression);
1738         grep_source_1(opt, gs, 1);
1739
1740         if (!chk_hit_marker(opt->pattern_expression))
1741                 return 0;
1742
1743         return grep_source_1(opt, gs, 0);
1744 }
1745
1746 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1747 {
1748         struct grep_source gs;
1749         int r;
1750
1751         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1752         gs.buf = buf;
1753         gs.size = size;
1754
1755         r = grep_source(opt, &gs);
1756
1757         grep_source_clear(&gs);
1758         return r;
1759 }
1760
1761 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1762                       const char *name, const char *path,
1763                       const void *identifier)
1764 {
1765         gs->type = type;
1766         gs->name = xstrdup_or_null(name);
1767         gs->path = xstrdup_or_null(path);
1768         gs->buf = NULL;
1769         gs->size = 0;
1770         gs->driver = NULL;
1771
1772         switch (type) {
1773         case GREP_SOURCE_FILE:
1774                 gs->identifier = xstrdup(identifier);
1775                 break;
1776         case GREP_SOURCE_SUBMODULE:
1777                 if (!identifier) {
1778                         gs->identifier = NULL;
1779                         break;
1780                 }
1781                 /*
1782                  * FALL THROUGH
1783                  * If the identifier is non-NULL (in the submodule case) it
1784                  * will be a SHA1 that needs to be copied.
1785                  */
1786         case GREP_SOURCE_SHA1:
1787                 gs->identifier = xmalloc(20);
1788                 hashcpy(gs->identifier, identifier);
1789                 break;
1790         case GREP_SOURCE_BUF:
1791                 gs->identifier = NULL;
1792                 break;
1793         }
1794 }
1795
1796 void grep_source_clear(struct grep_source *gs)
1797 {
1798         free(gs->name);
1799         gs->name = NULL;
1800         free(gs->path);
1801         gs->path = NULL;
1802         free(gs->identifier);
1803         gs->identifier = NULL;
1804         grep_source_clear_data(gs);
1805 }
1806
1807 void grep_source_clear_data(struct grep_source *gs)
1808 {
1809         switch (gs->type) {
1810         case GREP_SOURCE_FILE:
1811         case GREP_SOURCE_SHA1:
1812         case GREP_SOURCE_SUBMODULE:
1813                 free(gs->buf);
1814                 gs->buf = NULL;
1815                 gs->size = 0;
1816                 break;
1817         case GREP_SOURCE_BUF:
1818                 /* leave user-provided buf intact */
1819                 break;
1820         }
1821 }
1822
1823 static int grep_source_load_sha1(struct grep_source *gs)
1824 {
1825         enum object_type type;
1826
1827         grep_read_lock();
1828         gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1829         grep_read_unlock();
1830
1831         if (!gs->buf)
1832                 return error(_("'%s': unable to read %s"),
1833                              gs->name,
1834                              sha1_to_hex(gs->identifier));
1835         return 0;
1836 }
1837
1838 static int grep_source_load_file(struct grep_source *gs)
1839 {
1840         const char *filename = gs->identifier;
1841         struct stat st;
1842         char *data;
1843         size_t size;
1844         int i;
1845
1846         if (lstat(filename, &st) < 0) {
1847         err_ret:
1848                 if (errno != ENOENT)
1849                         error_errno(_("failed to stat '%s'"), filename);
1850                 return -1;
1851         }
1852         if (!S_ISREG(st.st_mode))
1853                 return -1;
1854         size = xsize_t(st.st_size);
1855         i = open(filename, O_RDONLY);
1856         if (i < 0)
1857                 goto err_ret;
1858         data = xmallocz(size);
1859         if (st.st_size != read_in_full(i, data, size)) {
1860                 error_errno(_("'%s': short read"), filename);
1861                 close(i);
1862                 free(data);
1863                 return -1;
1864         }
1865         close(i);
1866
1867         gs->buf = data;
1868         gs->size = size;
1869         return 0;
1870 }
1871
1872 static int grep_source_load(struct grep_source *gs)
1873 {
1874         if (gs->buf)
1875                 return 0;
1876
1877         switch (gs->type) {
1878         case GREP_SOURCE_FILE:
1879                 return grep_source_load_file(gs);
1880         case GREP_SOURCE_SHA1:
1881                 return grep_source_load_sha1(gs);
1882         case GREP_SOURCE_BUF:
1883                 return gs->buf ? 0 : -1;
1884         case GREP_SOURCE_SUBMODULE:
1885                 break;
1886         }
1887         die("BUG: invalid grep_source type to load");
1888 }
1889
1890 void grep_source_load_driver(struct grep_source *gs)
1891 {
1892         if (gs->driver)
1893                 return;
1894
1895         grep_attr_lock();
1896         if (gs->path)
1897                 gs->driver = userdiff_find_by_path(gs->path);
1898         if (!gs->driver)
1899                 gs->driver = userdiff_find_by_name("default");
1900         grep_attr_unlock();
1901 }
1902
1903 static int grep_source_is_binary(struct grep_source *gs)
1904 {
1905         grep_source_load_driver(gs);
1906         if (gs->driver->binary != -1)
1907                 return gs->driver->binary;
1908
1909         if (!grep_source_load(gs))
1910                 return buffer_is_binary(gs->buf, gs->size);
1911
1912         return 0;
1913 }