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