grep.c

   1 #include "cache.h"
   2 #include "grep.h"
   3 #include "userdiff.h"
   4 #include "xdiff-interface.h"
   5
   6 static int grep_source_load(struct grep_source *gs);
   7 static int grep_source_is_binary(struct grep_source *gs);
   8
   9 static struct grep_opt grep_defaults;
  10
  11 /*
  12  * Initialize the grep_defaults template with hardcoded defaults.
  13  * We could let the compiler do this, but without C99 initializers
  14  * the code gets unwieldy and unreadable, so...
  15  */
  16 void init_grep_defaults(void)
  17 {
  18         struct grep_opt *opt = &grep_defaults;
  19         static int run_once;
  20
  21         if (run_once)
  22                 return;
  23         run_once++;
  24
  25         memset(opt, 0, sizeof(*opt));
  26         opt->relative = 1;
  27         opt->pathname = 1;
  28         opt->regflags = REG_NEWLINE;
  29         opt->max_depth = -1;
  30         opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  31         opt->extended_regexp_option = 0;
  32         strcpy(opt->color_context, "");
  33         strcpy(opt->color_filename, "");
  34         strcpy(opt->color_function, "");
  35         strcpy(opt->color_lineno, "");
  36         strcpy(opt->color_match, GIT_COLOR_BOLD_RED);
  37         strcpy(opt->color_selected, "");
  38         strcpy(opt->color_sep, GIT_COLOR_CYAN);
  39         opt->color = -1;
  40 }
  41
  42 static int parse_pattern_type_arg(const char *opt, const char *arg)
  43 {
  44         if (!strcmp(arg, "default"))
  45                 return GREP_PATTERN_TYPE_UNSPECIFIED;
  46         else if (!strcmp(arg, "basic"))
  47                 return GREP_PATTERN_TYPE_BRE;
  48         else if (!strcmp(arg, "extended"))
  49                 return GREP_PATTERN_TYPE_ERE;
  50         else if (!strcmp(arg, "fixed"))
  51                 return GREP_PATTERN_TYPE_FIXED;
  52         else if (!strcmp(arg, "perl"))
  53                 return GREP_PATTERN_TYPE_PCRE;
  54         die("bad %s argument: %s", opt, arg);
  55 }
  56
  57 /*
  58  * Read the configuration file once and store it in
  59  * the grep_defaults template.
  60  */
  61 int grep_config(const char *var, const char *value, void *cb)
  62 {
  63         struct grep_opt *opt = &grep_defaults;
  64         char *color = NULL;
  65
  66         if (userdiff_config(var, value) < 0)
  67                 return -1;
  68
  69         if (!strcmp(var, "grep.extendedregexp")) {
  70                 if (git_config_bool(var, value))
  71                         opt->extended_regexp_option = 1;
  72                 else
  73                         opt->extended_regexp_option = 0;
  74                 return 0;
  75         }
  76
  77         if (!strcmp(var, "grep.patterntype")) {
  78                 opt->pattern_type_option = parse_pattern_type_arg(var, value);
  79                 return 0;
  80         }
  81
  82         if (!strcmp(var, "grep.linenumber")) {
  83                 opt->linenum = git_config_bool(var, value);
  84                 return 0;
  85         }
  86
  87         if (!strcmp(var, "color.grep"))
  88                 opt->color = git_config_colorbool(var, value);
  89         else if (!strcmp(var, "color.grep.context"))
  90                 color = opt->color_context;
  91         else if (!strcmp(var, "color.grep.filename"))
  92                 color = opt->color_filename;
  93         else if (!strcmp(var, "color.grep.function"))
  94                 color = opt->color_function;
  95         else if (!strcmp(var, "color.grep.linenumber"))
  96                 color = opt->color_lineno;
  97         else if (!strcmp(var, "color.grep.match"))
  98                 color = opt->color_match;
  99         else if (!strcmp(var, "color.grep.selected"))
 100                 color = opt->color_selected;
 101         else if (!strcmp(var, "color.grep.separator"))
 102                 color = opt->color_sep;
 103
 104         if (color) {
 105                 if (!value)
 106                         return config_error_nonbool(var);
 107                 color_parse(value, var, color);
 108         }
 109         return 0;
 110 }
 111
 112 /*
 113  * Initialize one instance of grep_opt and copy the
 114  * default values from the template we read the configuration
 115  * information in an earlier call to git_config(grep_config).
 116  */
 117 void grep_init(struct grep_opt *opt, const char *prefix)
 118 {
 119         struct grep_opt *def = &grep_defaults;
 120
 121         memset(opt, 0, sizeof(*opt));
 122         opt->prefix = prefix;
 123         opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 124         opt->pattern_tail = &opt->pattern_list;
 125         opt->header_tail = &opt->header_list;
 126
 127         opt->color = def->color;
 128         opt->extended_regexp_option = def->extended_regexp_option;
 129         opt->pattern_type_option = def->pattern_type_option;
 130         opt->linenum = def->linenum;
 131         opt->max_depth = def->max_depth;
 132         opt->pathname = def->pathname;
 133         opt->regflags = def->regflags;
 134         opt->relative = def->relative;
 135
 136         strcpy(opt->color_context, def->color_context);
 137         strcpy(opt->color_filename, def->color_filename);
 138         strcpy(opt->color_function, def->color_function);
 139         strcpy(opt->color_lineno, def->color_lineno);
 140         strcpy(opt->color_match, def->color_match);
 141         strcpy(opt->color_selected, def->color_selected);
 142         strcpy(opt->color_sep, def->color_sep);
 143 }
 144
 145 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 146 {
 147         if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 148                 grep_set_pattern_type_option(pattern_type, opt);
 149         else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 150                 grep_set_pattern_type_option(opt->pattern_type_option, opt);
 151         else if (opt->extended_regexp_option)
 152                 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 153 }
 154
 155 void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 156 {
 157         switch (pattern_type) {
 158         case GREP_PATTERN_TYPE_UNSPECIFIED:
 159                 /* fall through */
 160
 161         case GREP_PATTERN_TYPE_BRE:
 162                 opt->fixed = 0;
 163                 opt->pcre = 0;
 164                 opt->regflags &= ~REG_EXTENDED;
 165                 break;
 166
 167         case GREP_PATTERN_TYPE_ERE:
 168                 opt->fixed = 0;
 169                 opt->pcre = 0;
 170                 opt->regflags |= REG_EXTENDED;
 171                 break;
 172
 173         case GREP_PATTERN_TYPE_FIXED:
 174                 opt->fixed = 1;
 175                 opt->pcre = 0;
 176                 opt->regflags &= ~REG_EXTENDED;
 177                 break;
 178
 179         case GREP_PATTERN_TYPE_PCRE:
 180                 opt->fixed = 0;
 181                 opt->pcre = 1;
 182                 opt->regflags &= ~REG_EXTENDED;
 183                 break;
 184         }
 185 }
 186
 187 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 188                                         const char *origin, int no,
 189                                         enum grep_pat_token t,
 190                                         enum grep_header_field field)
 191 {
 192         struct grep_pat *p = xcalloc(1, sizeof(*p));
 193         p->pattern = xmemdupz(pat, patlen);
 194         p->patternlen = patlen;
 195         p->origin = origin;
 196         p->no = no;
 197         p->token = t;
 198         p->field = field;
 199         return p;
 200 }
 201
 202 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 203 {
 204         **tail = p;
 205         *tail = &p->next;
 206         p->next = NULL;
 207
 208         switch (p->token) {
 209         case GREP_PATTERN: /* atom */
 210         case GREP_PATTERN_HEAD:
 211         case GREP_PATTERN_BODY:
 212                 for (;;) {
 213                         struct grep_pat *new_pat;
 214                         size_t len = 0;
 215                         char *cp = p->pattern + p->patternlen, *nl = NULL;
 216                         while (++len <= p->patternlen) {
 217                                 if (*(--cp) == '\n') {
 218                                         nl = cp;
 219                                         break;
 220                                 }
 221                         }
 222                         if (!nl)
 223                                 break;
 224                         new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 225                                                   p->no, p->token, p->field);
 226                         new_pat->next = p->next;
 227                         if (!p->next)
 228                                 *tail = &new_pat->next;
 229                         p->next = new_pat;
 230                         *nl = '\0';
 231                         p->patternlen -= len;
 232                 }
 233                 break;
 234         default:
 235                 break;
 236         }
 237 }
 238
 239 void append_header_grep_pattern(struct grep_opt *opt,
 240                                 enum grep_header_field field, const char *pat)
 241 {
 242         struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 243                                              GREP_PATTERN_HEAD, field);
 244         if (field == GREP_HEADER_REFLOG)
 245                 opt->use_reflog_filter = 1;
 246         do_append_grep_pat(&opt->header_tail, p);
 247 }
 248
 249 void append_grep_pattern(struct grep_opt *opt, const char *pat,
 250                          const char *origin, int no, enum grep_pat_token t)
 251 {
 252         append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 253 }
 254
 255 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 256                      const char *origin, int no, enum grep_pat_token t)
 257 {
 258         struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 259         do_append_grep_pat(&opt->pattern_tail, p);
 260 }
 261
 262 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 263 {
 264         struct grep_pat *pat;
 265         struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 266         *ret = *opt;
 267
 268         ret->pattern_list = NULL;
 269         ret->pattern_tail = &ret->pattern_list;
 270
 271         for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 272         {
 273                 if(pat->token == GREP_PATTERN_HEAD)
 274                         append_header_grep_pattern(ret, pat->field,
 275                                                    pat->pattern);
 276                 else
 277                         append_grep_pat(ret, pat->pattern, pat->patternlen,
 278                                         pat->origin, pat->no, pat->token);
 279         }
 280
 281         return ret;
 282 }
 283
 284 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 285                 const char *error)
 286 {
 287         char where[1024];
 288
 289         if (p->no)
 290                 sprintf(where, "In '%s' at %d, ", p->origin, p->no);
 291         else if (p->origin)
 292                 sprintf(where, "%s, ", p->origin);
 293         else
 294                 where[0] = 0;
 295
 296         die("%s'%s': %s", where, p->pattern, error);
 297 }
 298
 299 #ifdef USE_LIBPCRE
 300 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 301 {
 302         const char *error;
 303         int erroffset;
 304         int options = PCRE_MULTILINE;
 305
 306         if (opt->ignore_case)
 307                 options |= PCRE_CASELESS;
 308
 309         p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 310                         NULL);
 311         if (!p->pcre_regexp)
 312                 compile_regexp_failed(p, error);
 313
 314         p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 315         if (!p->pcre_extra_info && error)
 316                 die("%s", error);
 317 }
 318
 319 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 320                 regmatch_t *match, int eflags)
 321 {
 322         int ovector[30], ret, flags = 0;
 323
 324         if (eflags & REG_NOTBOL)
 325                 flags |= PCRE_NOTBOL;
 326
 327         ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 328                         0, flags, ovector, ARRAY_SIZE(ovector));
 329         if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 330                 die("pcre_exec failed with error code %d", ret);
 331         if (ret > 0) {
 332                 ret = 0;
 333                 match->rm_so = ovector[0];
 334                 match->rm_eo = ovector[1];
 335         }
 336
 337         return ret;
 338 }
 339
 340 static void free_pcre_regexp(struct grep_pat *p)
 341 {
 342         pcre_free(p->pcre_regexp);
 343         pcre_free(p->pcre_extra_info);
 344 }
 345 #else /* !USE_LIBPCRE */
 346 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 347 {
 348         die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 349 }
 350
 351 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 352                 regmatch_t *match, int eflags)
 353 {
 354         return 1;
 355 }
 356
 357 static void free_pcre_regexp(struct grep_pat *p)
 358 {
 359 }
 360 #endif /* !USE_LIBPCRE */
 361
 362 static int is_fixed(const char *s, size_t len)
 363 {
 364         size_t i;
 365
 366         /* regcomp cannot accept patterns with NULs so we
 367          * consider any pattern containing a NUL fixed.
 368          */
 369         if (memchr(s, 0, len))
 370                 return 1;
 371
 372         for (i = 0; i < len; i++) {
 373                 if (is_regex_special(s[i]))
 374                         return 0;
 375         }
 376
 377         return 1;
 378 }
 379
 380 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 381 {
 382         int err;
 383
 384         p->word_regexp = opt->word_regexp;
 385         p->ignore_case = opt->ignore_case;
 386
 387         if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 388                 p->fixed = 1;
 389         else
 390                 p->fixed = 0;
 391
 392         if (p->fixed) {
 393                 if (opt->regflags & REG_ICASE || p->ignore_case)
 394                         p->kws = kwsalloc(tolower_trans_tbl);
 395                 else
 396                         p->kws = kwsalloc(NULL);
 397                 kwsincr(p->kws, p->pattern, p->patternlen);
 398                 kwsprep(p->kws);
 399                 return;
 400         }
 401
 402         if (opt->pcre) {
 403                 compile_pcre_regexp(p, opt);
 404                 return;
 405         }
 406
 407         err = regcomp(&p->regexp, p->pattern, opt->regflags);
 408         if (err) {
 409                 char errbuf[1024];
 410                 regerror(err, &p->regexp, errbuf, 1024);
 411                 regfree(&p->regexp);
 412                 compile_regexp_failed(p, errbuf);
 413         }
 414 }
 415
 416 static struct grep_expr *compile_pattern_or(struct grep_pat **);
 417 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 418 {
 419         struct grep_pat *p;
 420         struct grep_expr *x;
 421
 422         p = *list;
 423         if (!p)
 424                 return NULL;
 425         switch (p->token) {
 426         case GREP_PATTERN: /* atom */
 427         case GREP_PATTERN_HEAD:
 428         case GREP_PATTERN_BODY:
 429                 x = xcalloc(1, sizeof (struct grep_expr));
 430                 x->node = GREP_NODE_ATOM;
 431                 x->u.atom = p;
 432                 *list = p->next;
 433                 return x;
 434         case GREP_OPEN_PAREN:
 435                 *list = p->next;
 436                 x = compile_pattern_or(list);
 437                 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 438                         die("unmatched parenthesis");
 439                 *list = (*list)->next;
 440                 return x;
 441         default:
 442                 return NULL;
 443         }
 444 }
 445
 446 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 447 {
 448         struct grep_pat *p;
 449         struct grep_expr *x;
 450
 451         p = *list;
 452         if (!p)
 453                 return NULL;
 454         switch (p->token) {
 455         case GREP_NOT:
 456                 if (!p->next)
 457                         die("--not not followed by pattern expression");
 458                 *list = p->next;
 459                 x = xcalloc(1, sizeof (struct grep_expr));
 460                 x->node = GREP_NODE_NOT;
 461                 x->u.unary = compile_pattern_not(list);
 462                 if (!x->u.unary)
 463                         die("--not followed by non pattern expression");
 464                 return x;
 465         default:
 466                 return compile_pattern_atom(list);
 467         }
 468 }
 469
 470 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 471 {
 472         struct grep_pat *p;
 473         struct grep_expr *x, *y, *z;
 474
 475         x = compile_pattern_not(list);
 476         p = *list;
 477         if (p && p->token == GREP_AND) {
 478                 if (!p->next)
 479                         die("--and not followed by pattern expression");
 480                 *list = p->next;
 481                 y = compile_pattern_and(list);
 482                 if (!y)
 483                         die("--and not followed by pattern expression");
 484                 z = xcalloc(1, sizeof (struct grep_expr));
 485                 z->node = GREP_NODE_AND;
 486                 z->u.binary.left = x;
 487                 z->u.binary.right = y;
 488                 return z;
 489         }
 490         return x;
 491 }
 492
 493 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 494 {
 495         struct grep_pat *p;
 496         struct grep_expr *x, *y, *z;
 497
 498         x = compile_pattern_and(list);
 499         p = *list;
 500         if (x && p && p->token != GREP_CLOSE_PAREN) {
 501                 y = compile_pattern_or(list);
 502                 if (!y)
 503                         die("not a pattern expression %s", p->pattern);
 504                 z = xcalloc(1, sizeof (struct grep_expr));
 505                 z->node = GREP_NODE_OR;
 506                 z->u.binary.left = x;
 507                 z->u.binary.right = y;
 508                 return z;
 509         }
 510         return x;
 511 }
 512
 513 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 514 {
 515         return compile_pattern_or(list);
 516 }
 517
 518 static void indent(int in)
 519 {
 520         while (in-- > 0)
 521                 fputc(' ', stderr);
 522 }
 523
 524 static void dump_grep_pat(struct grep_pat *p)
 525 {
 526         switch (p->token) {
 527         case GREP_AND: fprintf(stderr, "*and*"); break;
 528         case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 529         case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 530         case GREP_NOT: fprintf(stderr, "*not*"); break;
 531         case GREP_OR: fprintf(stderr, "*or*"); break;
 532
 533         case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 534         case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 535         case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 536         }
 537
 538         switch (p->token) {
 539         default: break;
 540         case GREP_PATTERN_HEAD:
 541                 fprintf(stderr, "<head %d>", p->field); break;
 542         case GREP_PATTERN_BODY:
 543                 fprintf(stderr, "<body>"); break;
 544         }
 545         switch (p->token) {
 546         default: break;
 547         case GREP_PATTERN_HEAD:
 548         case GREP_PATTERN_BODY:
 549         case GREP_PATTERN:
 550                 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 551                 break;
 552         }
 553         fputc('\n', stderr);
 554 }
 555
 556 static void dump_grep_expression_1(struct grep_expr *x, int in)
 557 {
 558         indent(in);
 559         switch (x->node) {
 560         case GREP_NODE_TRUE:
 561                 fprintf(stderr, "true\n");
 562                 break;
 563         case GREP_NODE_ATOM:
 564                 dump_grep_pat(x->u.atom);
 565                 break;
 566         case GREP_NODE_NOT:
 567                 fprintf(stderr, "(not\n");
 568                 dump_grep_expression_1(x->u.unary, in+1);
 569                 indent(in);
 570                 fprintf(stderr, ")\n");
 571                 break;
 572         case GREP_NODE_AND:
 573                 fprintf(stderr, "(and\n");
 574                 dump_grep_expression_1(x->u.binary.left, in+1);
 575                 dump_grep_expression_1(x->u.binary.right, in+1);
 576                 indent(in);
 577                 fprintf(stderr, ")\n");
 578                 break;
 579         case GREP_NODE_OR:
 580                 fprintf(stderr, "(or\n");
 581                 dump_grep_expression_1(x->u.binary.left, in+1);
 582                 dump_grep_expression_1(x->u.binary.right, in+1);
 583                 indent(in);
 584                 fprintf(stderr, ")\n");
 585                 break;
 586         }
 587 }
 588
 589 static void dump_grep_expression(struct grep_opt *opt)
 590 {
 591         struct grep_expr *x = opt->pattern_expression;
 592
 593         if (opt->all_match)
 594                 fprintf(stderr, "[all-match]\n");
 595         dump_grep_expression_1(x, 0);
 596         fflush(NULL);
 597 }
 598
 599 static struct grep_expr *grep_true_expr(void)
 600 {
 601         struct grep_expr *z = xcalloc(1, sizeof(*z));
 602         z->node = GREP_NODE_TRUE;
 603         return z;
 604 }
 605
 606 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 607 {
 608         struct grep_expr *z = xcalloc(1, sizeof(*z));
 609         z->node = GREP_NODE_OR;
 610         z->u.binary.left = left;
 611         z->u.binary.right = right;
 612         return z;
 613 }
 614
 615 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 616 {
 617         struct grep_pat *p;
 618         struct grep_expr *header_expr;
 619         struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 620         enum grep_header_field fld;
 621
 622         if (!opt->header_list)
 623                 return NULL;
 624
 625         for (p = opt->header_list; p; p = p->next) {
 626                 if (p->token != GREP_PATTERN_HEAD)
 627                         die("bug: a non-header pattern in grep header list.");
 628                 if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
 629                         die("bug: unknown header field %d", p->field);
 630                 compile_regexp(p, opt);
 631         }
 632
 633         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 634                 header_group[fld] = NULL;
 635
 636         for (p = opt->header_list; p; p = p->next) {
 637                 struct grep_expr *h;
 638                 struct grep_pat *pp = p;
 639
 640                 h = compile_pattern_atom(&pp);
 641                 if (!h || pp != p->next)
 642                         die("bug: malformed header expr");
 643                 if (!header_group[p->field]) {
 644                         header_group[p->field] = h;
 645                         continue;
 646                 }
 647                 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 648         }
 649
 650         header_expr = NULL;
 651
 652         for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 653                 if (!header_group[fld])
 654                         continue;
 655                 if (!header_expr)
 656                         header_expr = grep_true_expr();
 657                 header_expr = grep_or_expr(header_group[fld], header_expr);
 658         }
 659         return header_expr;
 660 }
 661
 662 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 663 {
 664         struct grep_expr *z = x;
 665
 666         while (x) {
 667                 assert(x->node == GREP_NODE_OR);
 668                 if (x->u.binary.right &&
 669                     x->u.binary.right->node == GREP_NODE_TRUE) {
 670                         x->u.binary.right = y;
 671                         break;
 672                 }
 673                 x = x->u.binary.right;
 674         }
 675         return z;
 676 }
 677
 678 static void compile_grep_patterns_real(struct grep_opt *opt)
 679 {
 680         struct grep_pat *p;
 681         struct grep_expr *header_expr = prep_header_patterns(opt);
 682
 683         for (p = opt->pattern_list; p; p = p->next) {
 684                 switch (p->token) {
 685                 case GREP_PATTERN: /* atom */
 686                 case GREP_PATTERN_HEAD:
 687                 case GREP_PATTERN_BODY:
 688                         compile_regexp(p, opt);
 689                         break;
 690                 default:
 691                         opt->extended = 1;
 692                         break;
 693                 }
 694         }
 695
 696         if (opt->all_match || header_expr)
 697                 opt->extended = 1;
 698         else if (!opt->extended && !opt->debug)
 699                 return;
 700
 701         p = opt->pattern_list;
 702         if (p)
 703                 opt->pattern_expression = compile_pattern_expr(&p);
 704         if (p)
 705                 die("incomplete pattern expression: %s", p->pattern);
 706
 707         if (!header_expr)
 708                 return;
 709
 710         if (!opt->pattern_expression)
 711                 opt->pattern_expression = header_expr;
 712         else if (opt->all_match)
 713                 opt->pattern_expression = grep_splice_or(header_expr,
 714                                                          opt->pattern_expression);
 715         else
 716                 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 717                                                        header_expr);
 718         opt->all_match = 1;
 719 }
 720
 721 void compile_grep_patterns(struct grep_opt *opt)
 722 {
 723         compile_grep_patterns_real(opt);
 724         if (opt->debug)
 725                 dump_grep_expression(opt);
 726 }
 727
 728 static void free_pattern_expr(struct grep_expr *x)
 729 {
 730         switch (x->node) {
 731         case GREP_NODE_TRUE:
 732         case GREP_NODE_ATOM:
 733                 break;
 734         case GREP_NODE_NOT:
 735                 free_pattern_expr(x->u.unary);
 736                 break;
 737         case GREP_NODE_AND:
 738         case GREP_NODE_OR:
 739                 free_pattern_expr(x->u.binary.left);
 740                 free_pattern_expr(x->u.binary.right);
 741                 break;
 742         }
 743         free(x);
 744 }
 745
 746 void free_grep_patterns(struct grep_opt *opt)
 747 {
 748         struct grep_pat *p, *n;
 749
 750         for (p = opt->pattern_list; p; p = n) {
 751                 n = p->next;
 752                 switch (p->token) {
 753                 case GREP_PATTERN: /* atom */
 754                 case GREP_PATTERN_HEAD:
 755                 case GREP_PATTERN_BODY:
 756                         if (p->kws)
 757                                 kwsfree(p->kws);
 758                         else if (p->pcre_regexp)
 759                                 free_pcre_regexp(p);
 760                         else
 761                                 regfree(&p->regexp);
 762                         free(p->pattern);
 763                         break;
 764                 default:
 765                         break;
 766                 }
 767                 free(p);
 768         }
 769
 770         if (!opt->extended)
 771                 return;
 772         free_pattern_expr(opt->pattern_expression);
 773 }
 774
 775 static char *end_of_line(char *cp, unsigned long *left)
 776 {
 777         unsigned long l = *left;
 778         while (l && *cp != '\n') {
 779                 l--;
 780                 cp++;
 781         }
 782         *left = l;
 783         return cp;
 784 }
 785
 786 static int word_char(char ch)
 787 {
 788         return isalnum(ch) || ch == '_';
 789 }
 790
 791 static void output_color(struct grep_opt *opt, const void *data, size_t size,
 792                          const char *color)
 793 {
 794         if (want_color(opt->color) && color && color[0]) {
 795                 opt->output(opt, color, strlen(color));
 796                 opt->output(opt, data, size);
 797                 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 798         } else
 799                 opt->output(opt, data, size);
 800 }
 801
 802 static void output_sep(struct grep_opt *opt, char sign)
 803 {
 804         if (opt->null_following_name)
 805                 opt->output(opt, "\0", 1);
 806         else
 807                 output_color(opt, &sign, 1, opt->color_sep);
 808 }
 809
 810 static void show_name(struct grep_opt *opt, const char *name)
 811 {
 812         output_color(opt, name, strlen(name), opt->color_filename);
 813         opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 814 }
 815
 816 static int fixmatch(struct grep_pat *p, char *line, char *eol,
 817                     regmatch_t *match)
 818 {
 819         struct kwsmatch kwsm;
 820         size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 821         if (offset == -1) {
 822                 match->rm_so = match->rm_eo = -1;
 823                 return REG_NOMATCH;
 824         } else {
 825                 match->rm_so = offset;
 826                 match->rm_eo = match->rm_so + kwsm.size[0];
 827                 return 0;
 828         }
 829 }
 830
 831 static int regmatch(const regex_t *preg, char *line, char *eol,
 832                     regmatch_t *match, int eflags)
 833 {
 834 #ifdef REG_STARTEND
 835         match->rm_so = 0;
 836         match->rm_eo = eol - line;
 837         eflags |= REG_STARTEND;
 838 #endif
 839         return regexec(preg, line, 1, match, eflags);
 840 }
 841
 842 static int patmatch(struct grep_pat *p, char *line, char *eol,
 843                     regmatch_t *match, int eflags)
 844 {
 845         int hit;
 846
 847         if (p->fixed)
 848                 hit = !fixmatch(p, line, eol, match);
 849         else if (p->pcre_regexp)
 850                 hit = !pcrematch(p, line, eol, match, eflags);
 851         else
 852                 hit = !regmatch(&p->regexp, line, eol, match, eflags);
 853
 854         return hit;
 855 }
 856
 857 static int strip_timestamp(char *bol, char **eol_p)
 858 {
 859         char *eol = *eol_p;
 860         int ch;
 861
 862         while (bol < --eol) {
 863                 if (*eol != '>')
 864                         continue;
 865                 *eol_p = ++eol;
 866                 ch = *eol;
 867                 *eol = '\0';
 868                 return ch;
 869         }
 870         return 0;
 871 }
 872
 873 static struct {
 874         const char *field;
 875         size_t len;
 876 } header_field[] = {
 877         { "author ", 7 },
 878         { "committer ", 10 },
 879         { "reflog ", 7 },
 880 };
 881
 882 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 883                              enum grep_context ctx,
 884                              regmatch_t *pmatch, int eflags)
 885 {
 886         int hit = 0;
 887         int saved_ch = 0;
 888         const char *start = bol;
 889
 890         if ((p->token != GREP_PATTERN) &&
 891             ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 892                 return 0;
 893
 894         if (p->token == GREP_PATTERN_HEAD) {
 895                 const char *field;
 896                 size_t len;
 897                 assert(p->field < ARRAY_SIZE(header_field));
 898                 field = header_field[p->field].field;
 899                 len = header_field[p->field].len;
 900                 if (strncmp(bol, field, len))
 901                         return 0;
 902                 bol += len;
 903                 switch (p->field) {
 904                 case GREP_HEADER_AUTHOR:
 905                 case GREP_HEADER_COMMITTER:
 906                         saved_ch = strip_timestamp(bol, &eol);
 907                         break;
 908                 default:
 909                         break;
 910                 }
 911         }
 912
 913  again:
 914         hit = patmatch(p, bol, eol, pmatch, eflags);
 915
 916         if (hit && p->word_regexp) {
 917                 if ((pmatch[0].rm_so < 0) ||
 918                     (eol - bol) < pmatch[0].rm_so ||
 919                     (pmatch[0].rm_eo < 0) ||
 920                     (eol - bol) < pmatch[0].rm_eo)
 921                         die("regexp returned nonsense");
 922
 923                 /* Match beginning must be either beginning of the
 924                  * line, or at word boundary (i.e. the last char must
 925                  * not be a word char).  Similarly, match end must be
 926                  * either end of the line, or at word boundary
 927                  * (i.e. the next char must not be a word char).
 928                  */
 929                 if ( ((pmatch[0].rm_so == 0) ||
 930                       !word_char(bol[pmatch[0].rm_so-1])) &&
 931                      ((pmatch[0].rm_eo == (eol-bol)) ||
 932                       !word_char(bol[pmatch[0].rm_eo])) )
 933                         ;
 934                 else
 935                         hit = 0;
 936
 937                 /* Words consist of at least one character. */
 938                 if (pmatch->rm_so == pmatch->rm_eo)
 939                         hit = 0;
 940
 941                 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 942                         /* There could be more than one match on the
 943                          * line, and the first match might not be
 944                          * strict word match.  But later ones could be!
 945                          * Forward to the next possible start, i.e. the
 946                          * next position following a non-word char.
 947                          */
 948                         bol = pmatch[0].rm_so + bol + 1;
 949                         while (word_char(bol[-1]) && bol < eol)
 950                                 bol++;
 951                         eflags |= REG_NOTBOL;
 952                         if (bol < eol)
 953                                 goto again;
 954                 }
 955         }
 956         if (p->token == GREP_PATTERN_HEAD && saved_ch)
 957                 *eol = saved_ch;
 958         if (hit) {
 959                 pmatch[0].rm_so += bol - start;
 960                 pmatch[0].rm_eo += bol - start;
 961         }
 962         return hit;
 963 }
 964
 965 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 966                            enum grep_context ctx, int collect_hits)
 967 {
 968         int h = 0;
 969         regmatch_t match;
 970
 971         if (!x)
 972                 die("Not a valid grep expression");
 973         switch (x->node) {
 974         case GREP_NODE_TRUE:
 975                 h = 1;
 976                 break;
 977         case GREP_NODE_ATOM:
 978                 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 979                 break;
 980         case GREP_NODE_NOT:
 981                 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 982                 break;
 983         case GREP_NODE_AND:
 984                 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 985                         return 0;
 986                 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 987                 break;
 988         case GREP_NODE_OR:
 989                 if (!collect_hits)
 990                         return (match_expr_eval(x->u.binary.left,
 991                                                 bol, eol, ctx, 0) ||
 992                                 match_expr_eval(x->u.binary.right,
 993                                                 bol, eol, ctx, 0));
 994                 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 995                 x->u.binary.left->hit |= h;
 996                 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 997                 break;
 998         default:
 999                 die("Unexpected node type (internal error) %d", x->node);
1000         }
1001         if (collect_hits)
1002                 x->hit |= h;
1003         return h;
1004 }
1005
1006 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1007                       enum grep_context ctx, int collect_hits)
1008 {
1009         struct grep_expr *x = opt->pattern_expression;
1010         return match_expr_eval(x, bol, eol, ctx, collect_hits);
1011 }
1012
1013 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1014                       enum grep_context ctx, int collect_hits)
1015 {
1016         struct grep_pat *p;
1017         regmatch_t match;
1018
1019         if (opt->extended)
1020                 return match_expr(opt, bol, eol, ctx, collect_hits);
1021
1022         /* we do not call with collect_hits without being extended */
1023         for (p = opt->pattern_list; p; p = p->next) {
1024                 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1025                         return 1;
1026         }
1027         return 0;
1028 }
1029
1030 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1031                               enum grep_context ctx,
1032                               regmatch_t *pmatch, int eflags)
1033 {
1034         regmatch_t match;
1035
1036         if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1037                 return 0;
1038         if (match.rm_so < 0 || match.rm_eo < 0)
1039                 return 0;
1040         if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1041                 if (match.rm_so > pmatch->rm_so)
1042                         return 1;
1043                 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1044                         return 1;
1045         }
1046         pmatch->rm_so = match.rm_so;
1047         pmatch->rm_eo = match.rm_eo;
1048         return 1;
1049 }
1050
1051 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1052                       enum grep_context ctx, regmatch_t *pmatch, int eflags)
1053 {
1054         struct grep_pat *p;
1055         int hit = 0;
1056
1057         pmatch->rm_so = pmatch->rm_eo = -1;
1058         if (bol < eol) {
1059                 for (p = opt->pattern_list; p; p = p->next) {
1060                         switch (p->token) {
1061                         case GREP_PATTERN: /* atom */
1062                         case GREP_PATTERN_HEAD:
1063                         case GREP_PATTERN_BODY:
1064                                 hit |= match_next_pattern(p, bol, eol, ctx,
1065                                                           pmatch, eflags);
1066                                 break;
1067                         default:
1068                                 break;
1069                         }
1070                 }
1071         }
1072         return hit;
1073 }
1074
1075 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1076                       const char *name, unsigned lno, char sign)
1077 {
1078         int rest = eol - bol;
1079         char *line_color = NULL;
1080
1081         if (opt->file_break && opt->last_shown == 0) {
1082                 if (opt->show_hunk_mark)
1083                         opt->output(opt, "\n", 1);
1084         } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1085                 if (opt->last_shown == 0) {
1086                         if (opt->show_hunk_mark) {
1087                                 output_color(opt, "--", 2, opt->color_sep);
1088                                 opt->output(opt, "\n", 1);
1089                         }
1090                 } else if (lno > opt->last_shown + 1) {
1091                         output_color(opt, "--", 2, opt->color_sep);
1092                         opt->output(opt, "\n", 1);
1093                 }
1094         }
1095         if (opt->heading && opt->last_shown == 0) {
1096                 output_color(opt, name, strlen(name), opt->color_filename);
1097                 opt->output(opt, "\n", 1);
1098         }
1099         opt->last_shown = lno;
1100
1101         if (!opt->heading && opt->pathname) {
1102                 output_color(opt, name, strlen(name), opt->color_filename);
1103                 output_sep(opt, sign);
1104         }
1105         if (opt->linenum) {
1106                 char buf[32];
1107                 snprintf(buf, sizeof(buf), "%d", lno);
1108                 output_color(opt, buf, strlen(buf), opt->color_lineno);
1109                 output_sep(opt, sign);
1110         }
1111         if (opt->color) {
1112                 regmatch_t match;
1113                 enum grep_context ctx = GREP_CONTEXT_BODY;
1114                 int ch = *eol;
1115                 int eflags = 0;
1116
1117                 if (sign == ':')
1118                         line_color = opt->color_selected;
1119                 else if (sign == '-')
1120                         line_color = opt->color_context;
1121                 else if (sign == '=')
1122                         line_color = opt->color_function;
1123                 *eol = '\0';
1124                 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1125                         if (match.rm_so == match.rm_eo)
1126                                 break;
1127
1128                         output_color(opt, bol, match.rm_so, line_color);
1129                         output_color(opt, bol + match.rm_so,
1130                                      match.rm_eo - match.rm_so,
1131                                      opt->color_match);
1132                         bol += match.rm_eo;
1133                         rest -= match.rm_eo;
1134                         eflags = REG_NOTBOL;
1135                 }
1136                 *eol = ch;
1137         }
1138         output_color(opt, bol, rest, line_color);
1139         opt->output(opt, "\n", 1);
1140 }
1141
1142 #ifndef NO_PTHREADS
1143 int grep_use_locks;
1144
1145 /*
1146  * This lock protects access to the gitattributes machinery, which is
1147  * not thread-safe.
1148  */
1149 pthread_mutex_t grep_attr_mutex;
1150
1151 static inline void grep_attr_lock(void)
1152 {
1153         if (grep_use_locks)
1154                 pthread_mutex_lock(&grep_attr_mutex);
1155 }
1156
1157 static inline void grep_attr_unlock(void)
1158 {
1159         if (grep_use_locks)
1160                 pthread_mutex_unlock(&grep_attr_mutex);
1161 }
1162
1163 /*
1164  * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1165  */
1166 pthread_mutex_t grep_read_mutex;
1167
1168 #else
1169 #define grep_attr_lock()
1170 #define grep_attr_unlock()
1171 #endif
1172
1173 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1174 {
1175         xdemitconf_t *xecfg = opt->priv;
1176         if (xecfg && !xecfg->find_func) {
1177                 grep_source_load_driver(gs);
1178                 if (gs->driver->funcname.pattern) {
1179                         const struct userdiff_funcname *pe = &gs->driver->funcname;
1180                         xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1181                 } else {
1182                         xecfg = opt->priv = NULL;
1183                 }
1184         }
1185
1186         if (xecfg) {
1187                 char buf[1];
1188                 return xecfg->find_func(bol, eol - bol, buf, 1,
1189                                         xecfg->find_func_priv) >= 0;
1190         }
1191
1192         if (bol == eol)
1193                 return 0;
1194         if (isalpha(*bol) || *bol == '_' || *bol == '$')
1195                 return 1;
1196         return 0;
1197 }
1198
1199 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1200                                char *bol, unsigned lno)
1201 {
1202         while (bol > gs->buf) {
1203                 char *eol = --bol;
1204
1205                 while (bol > gs->buf && bol[-1] != '\n')
1206                         bol--;
1207                 lno--;
1208
1209                 if (lno <= opt->last_shown)
1210                         break;
1211
1212                 if (match_funcname(opt, gs, bol, eol)) {
1213                         show_line(opt, bol, eol, gs->name, lno, '=');
1214                         break;
1215                 }
1216         }
1217 }
1218
1219 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1220                              char *bol, char *end, unsigned lno)
1221 {
1222         unsigned cur = lno, from = 1, funcname_lno = 0;
1223         int funcname_needed = !!opt->funcname;
1224
1225         if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1226                 funcname_needed = 2;
1227
1228         if (opt->pre_context < lno)
1229                 from = lno - opt->pre_context;
1230         if (from <= opt->last_shown)
1231                 from = opt->last_shown + 1;
1232
1233         /* Rewind. */
1234         while (bol > gs->buf &&
1235                cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1236                 char *eol = --bol;
1237
1238                 while (bol > gs->buf && bol[-1] != '\n')
1239                         bol--;
1240                 cur--;
1241                 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1242                         funcname_lno = cur;
1243                         funcname_needed = 0;
1244                 }
1245         }
1246
1247         /* We need to look even further back to find a function signature. */
1248         if (opt->funcname && funcname_needed)
1249                 show_funcname_line(opt, gs, bol, cur);
1250
1251         /* Back forward. */
1252         while (cur < lno) {
1253                 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1254
1255                 while (*eol != '\n')
1256                         eol++;
1257                 show_line(opt, bol, eol, gs->name, cur, sign);
1258                 bol = eol + 1;
1259                 cur++;
1260         }
1261 }
1262
1263 static int should_lookahead(struct grep_opt *opt)
1264 {
1265         struct grep_pat *p;
1266
1267         if (opt->extended)
1268                 return 0; /* punt for too complex stuff */
1269         if (opt->invert)
1270                 return 0;
1271         for (p = opt->pattern_list; p; p = p->next) {
1272                 if (p->token != GREP_PATTERN)
1273                         return 0; /* punt for "header only" and stuff */
1274         }
1275         return 1;
1276 }
1277
1278 static int look_ahead(struct grep_opt *opt,
1279                       unsigned long *left_p,
1280                       unsigned *lno_p,
1281                       char **bol_p)
1282 {
1283         unsigned lno = *lno_p;
1284         char *bol = *bol_p;
1285         struct grep_pat *p;
1286         char *sp, *last_bol;
1287         regoff_t earliest = -1;
1288
1289         for (p = opt->pattern_list; p; p = p->next) {
1290                 int hit;
1291                 regmatch_t m;
1292
1293                 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1294                 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1295                         continue;
1296                 if (earliest < 0 || m.rm_so < earliest)
1297                         earliest = m.rm_so;
1298         }
1299
1300         if (earliest < 0) {
1301                 *bol_p = bol + *left_p;
1302                 *left_p = 0;
1303                 return 1;
1304         }
1305         for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1306                 ; /* find the beginning of the line */
1307         last_bol = sp;
1308
1309         for (sp = bol; sp < last_bol; sp++) {
1310                 if (*sp == '\n')
1311                         lno++;
1312         }
1313         *left_p -= last_bol - bol;
1314         *bol_p = last_bol;
1315         *lno_p = lno;
1316         return 0;
1317 }
1318
1319 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1320 {
1321         fwrite(buf, size, 1, stdout);
1322 }
1323
1324 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1325 {
1326         char *bol;
1327         unsigned long left;
1328         unsigned lno = 1;
1329         unsigned last_hit = 0;
1330         int binary_match_only = 0;
1331         unsigned count = 0;
1332         int try_lookahead = 0;
1333         int show_function = 0;
1334         enum grep_context ctx = GREP_CONTEXT_HEAD;
1335         xdemitconf_t xecfg;
1336
1337         if (!opt->output)
1338                 opt->output = std_output;
1339
1340         if (opt->pre_context || opt->post_context || opt->file_break ||
1341             opt->funcbody) {
1342                 /* Show hunk marks, except for the first file. */
1343                 if (opt->last_shown)
1344                         opt->show_hunk_mark = 1;
1345                 /*
1346                  * If we're using threads then we can't easily identify
1347                  * the first file.  Always put hunk marks in that case
1348                  * and skip the very first one later in work_done().
1349                  */
1350                 if (opt->output != std_output)
1351                         opt->show_hunk_mark = 1;
1352         }
1353         opt->last_shown = 0;
1354
1355         switch (opt->binary) {
1356         case GREP_BINARY_DEFAULT:
1357                 if (grep_source_is_binary(gs))
1358                         binary_match_only = 1;
1359                 break;
1360         case GREP_BINARY_NOMATCH:
1361                 if (grep_source_is_binary(gs))
1362                         return 0; /* Assume unmatch */
1363                 break;
1364         case GREP_BINARY_TEXT:
1365                 break;
1366         default:
1367                 die("bug: unknown binary handling mode");
1368         }
1369
1370         memset(&xecfg, 0, sizeof(xecfg));
1371         opt->priv = &xecfg;
1372
1373         try_lookahead = should_lookahead(opt);
1374
1375         if (grep_source_load(gs) < 0)
1376                 return 0;
1377
1378         bol = gs->buf;
1379         left = gs->size;
1380         while (left) {
1381                 char *eol, ch;
1382                 int hit;
1383
1384                 /*
1385                  * look_ahead() skips quickly to the line that possibly
1386                  * has the next hit; don't call it if we need to do
1387                  * something more than just skipping the current line
1388                  * in response to an unmatch for the current line.  E.g.
1389                  * inside a post-context window, we will show the current
1390                  * line as a context around the previous hit when it
1391                  * doesn't hit.
1392                  */
1393                 if (try_lookahead
1394                     && !(last_hit
1395                          && (show_function ||
1396                              lno <= last_hit + opt->post_context))
1397                     && look_ahead(opt, &left, &lno, &bol))
1398                         break;
1399                 eol = end_of_line(bol, &left);
1400                 ch = *eol;
1401                 *eol = 0;
1402
1403                 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1404                         ctx = GREP_CONTEXT_BODY;
1405
1406                 hit = match_line(opt, bol, eol, ctx, collect_hits);
1407                 *eol = ch;
1408
1409                 if (collect_hits)
1410                         goto next_line;
1411
1412                 /* "grep -v -e foo -e bla" should list lines
1413                  * that do not have either, so inversion should
1414                  * be done outside.
1415                  */
1416                 if (opt->invert)
1417                         hit = !hit;
1418                 if (opt->unmatch_name_only) {
1419                         if (hit)
1420                                 return 0;
1421                         goto next_line;
1422                 }
1423                 if (hit) {
1424                         count++;
1425                         if (opt->status_only)
1426                                 return 1;
1427                         if (opt->name_only) {
1428                                 show_name(opt, gs->name);
1429                                 return 1;
1430                         }
1431                         if (opt->count)
1432                                 goto next_line;
1433                         if (binary_match_only) {
1434                                 opt->output(opt, "Binary file ", 12);
1435                                 output_color(opt, gs->name, strlen(gs->name),
1436                                              opt->color_filename);
1437                                 opt->output(opt, " matches\n", 9);
1438                                 return 1;
1439                         }
1440                         /* Hit at this line.  If we haven't shown the
1441                          * pre-context lines, we would need to show them.
1442                          */
1443                         if (opt->pre_context || opt->funcbody)
1444                                 show_pre_context(opt, gs, bol, eol, lno);
1445                         else if (opt->funcname)
1446                                 show_funcname_line(opt, gs, bol, lno);
1447                         show_line(opt, bol, eol, gs->name, lno, ':');
1448                         last_hit = lno;
1449                         if (opt->funcbody)
1450                                 show_function = 1;
1451                         goto next_line;
1452                 }
1453                 if (show_function && match_funcname(opt, gs, bol, eol))
1454                         show_function = 0;
1455                 if (show_function ||
1456                     (last_hit && lno <= last_hit + opt->post_context)) {
1457                         /* If the last hit is within the post context,
1458                          * we need to show this line.
1459                          */
1460                         show_line(opt, bol, eol, gs->name, lno, '-');
1461                 }
1462
1463         next_line:
1464                 bol = eol + 1;
1465                 if (!left)
1466                         break;
1467                 left--;
1468                 lno++;
1469         }
1470
1471         if (collect_hits)
1472                 return 0;
1473
1474         if (opt->status_only)
1475                 return 0;
1476         if (opt->unmatch_name_only) {
1477                 /* We did not see any hit, so we want to show this */
1478                 show_name(opt, gs->name);
1479                 return 1;
1480         }
1481
1482         xdiff_clear_find_func(&xecfg);
1483         opt->priv = NULL;
1484
1485         /* NEEDSWORK:
1486          * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1487          * which feels mostly useless but sometimes useful.  Maybe
1488          * make it another option?  For now suppress them.
1489          */
1490         if (opt->count && count) {
1491                 char buf[32];
1492                 output_color(opt, gs->name, strlen(gs->name), opt->color_filename);
1493                 output_sep(opt, ':');
1494                 snprintf(buf, sizeof(buf), "%u\n", count);
1495                 opt->output(opt, buf, strlen(buf));
1496                 return 1;
1497         }
1498         return !!last_hit;
1499 }
1500
1501 static void clr_hit_marker(struct grep_expr *x)
1502 {
1503         /* All-hit markers are meaningful only at the very top level
1504          * OR node.
1505          */
1506         while (1) {
1507                 x->hit = 0;
1508                 if (x->node != GREP_NODE_OR)
1509                         return;
1510                 x->u.binary.left->hit = 0;
1511                 x = x->u.binary.right;
1512         }
1513 }
1514
1515 static int chk_hit_marker(struct grep_expr *x)
1516 {
1517         /* Top level nodes have hit markers.  See if they all are hits */
1518         while (1) {
1519                 if (x->node != GREP_NODE_OR)
1520                         return x->hit;
1521                 if (!x->u.binary.left->hit)
1522                         return 0;
1523                 x = x->u.binary.right;
1524         }
1525 }
1526
1527 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1528 {
1529         /*
1530          * we do not have to do the two-pass grep when we do not check
1531          * buffer-wide "all-match".
1532          */
1533         if (!opt->all_match)
1534                 return grep_source_1(opt, gs, 0);
1535
1536         /* Otherwise the toplevel "or" terms hit a bit differently.
1537          * We first clear hit markers from them.
1538          */
1539         clr_hit_marker(opt->pattern_expression);
1540         grep_source_1(opt, gs, 1);
1541
1542         if (!chk_hit_marker(opt->pattern_expression))
1543                 return 0;
1544
1545         return grep_source_1(opt, gs, 0);
1546 }
1547
1548 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1549 {
1550         struct grep_source gs;
1551         int r;
1552
1553         grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1554         gs.buf = buf;
1555         gs.size = size;
1556
1557         r = grep_source(opt, &gs);
1558
1559         grep_source_clear(&gs);
1560         return r;
1561 }
1562
1563 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1564                       const char *name, const char *path,
1565                       const void *identifier)
1566 {
1567         gs->type = type;
1568         gs->name = name ? xstrdup(name) : NULL;
1569         gs->path = path ? xstrdup(path) : NULL;
1570         gs->buf = NULL;
1571         gs->size = 0;
1572         gs->driver = NULL;
1573
1574         switch (type) {
1575         case GREP_SOURCE_FILE:
1576                 gs->identifier = xstrdup(identifier);
1577                 break;
1578         case GREP_SOURCE_SHA1:
1579                 gs->identifier = xmalloc(20);
1580                 memcpy(gs->identifier, identifier, 20);
1581                 break;
1582         case GREP_SOURCE_BUF:
1583                 gs->identifier = NULL;
1584         }
1585 }
1586
1587 void grep_source_clear(struct grep_source *gs)
1588 {
1589         free(gs->name);
1590         gs->name = NULL;
1591         free(gs->path);
1592         gs->path = NULL;
1593         free(gs->identifier);
1594         gs->identifier = NULL;
1595         grep_source_clear_data(gs);
1596 }
1597
1598 void grep_source_clear_data(struct grep_source *gs)
1599 {
1600         switch (gs->type) {
1601         case GREP_SOURCE_FILE:
1602         case GREP_SOURCE_SHA1:
1603                 free(gs->buf);
1604                 gs->buf = NULL;
1605                 gs->size = 0;
1606                 break;
1607         case GREP_SOURCE_BUF:
1608                 /* leave user-provided buf intact */
1609                 break;
1610         }
1611 }
1612
1613 static int grep_source_load_sha1(struct grep_source *gs)
1614 {
1615         enum object_type type;
1616
1617         grep_read_lock();
1618         gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1619         grep_read_unlock();
1620
1621         if (!gs->buf)
1622                 return error(_("'%s': unable to read %s"),
1623                              gs->name,
1624                              sha1_to_hex(gs->identifier));
1625         return 0;
1626 }
1627
1628 static int grep_source_load_file(struct grep_source *gs)
1629 {
1630         const char *filename = gs->identifier;
1631         struct stat st;
1632         char *data;
1633         size_t size;
1634         int i;
1635
1636         if (lstat(filename, &st) < 0) {
1637         err_ret:
1638                 if (errno != ENOENT)
1639                         error(_("'%s': %s"), filename, strerror(errno));
1640                 return -1;
1641         }
1642         if (!S_ISREG(st.st_mode))
1643                 return -1;
1644         size = xsize_t(st.st_size);
1645         i = open(filename, O_RDONLY);
1646         if (i < 0)
1647                 goto err_ret;
1648         data = xmalloc(size + 1);
1649         if (st.st_size != read_in_full(i, data, size)) {
1650                 error(_("'%s': short read %s"), filename, strerror(errno));
1651                 close(i);
1652                 free(data);
1653                 return -1;
1654         }
1655         close(i);
1656         data[size] = 0;
1657
1658         gs->buf = data;
1659         gs->size = size;
1660         return 0;
1661 }
1662
1663 static int grep_source_load(struct grep_source *gs)
1664 {
1665         if (gs->buf)
1666                 return 0;
1667
1668         switch (gs->type) {
1669         case GREP_SOURCE_FILE:
1670                 return grep_source_load_file(gs);
1671         case GREP_SOURCE_SHA1:
1672                 return grep_source_load_sha1(gs);
1673         case GREP_SOURCE_BUF:
1674                 return gs->buf ? 0 : -1;
1675         }
1676         die("BUG: invalid grep_source type");
1677 }
1678
1679 void grep_source_load_driver(struct grep_source *gs)
1680 {
1681         if (gs->driver)
1682                 return;
1683
1684         grep_attr_lock();
1685         if (gs->path)
1686                 gs->driver = userdiff_find_by_path(gs->path);
1687         if (!gs->driver)
1688                 gs->driver = userdiff_find_by_name("default");
1689         grep_attr_unlock();
1690 }
1691
1692 static int grep_source_is_binary(struct grep_source *gs)
1693 {
1694         grep_source_load_driver(gs);
1695         if (gs->driver->binary != -1)
1696                 return gs->driver->binary;
1697
1698         if (!grep_source_load(gs))
1699                 return buffer_is_binary(gs->buf, gs->size);
1700
1701         return 0;
1702 }