Andrew's git / gitweb.git / grep.c
?
grep.con commit pickaxe: allow -i to search in patch case-insensitively (accccde)
   1#include "cache.h"
   2#include "grep.h"
   3#include "userdiff.h"
   4#include "xdiff-interface.h"
   5
   6void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat)
   7{
   8        struct grep_pat *p = xcalloc(1, sizeof(*p));
   9        p->pattern = pat;
  10        p->patternlen = strlen(pat);
  11        p->origin = "header";
  12        p->no = 0;
  13        p->token = GREP_PATTERN_HEAD;
  14        p->field = field;
  15        *opt->header_tail = p;
  16        opt->header_tail = &p->next;
  17        p->next = NULL;
  18}
  19
  20void append_grep_pattern(struct grep_opt *opt, const char *pat,
  21                         const char *origin, int no, enum grep_pat_token t)
  22{
  23        append_grep_pat(opt, pat, strlen(pat), origin, no, t);
  24}
  25
  26void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
  27                     const char *origin, int no, enum grep_pat_token t)
  28{
  29        struct grep_pat *p = xcalloc(1, sizeof(*p));
  30        p->pattern = pat;
  31        p->patternlen = patlen;
  32        p->origin = origin;
  33        p->no = no;
  34        p->token = t;
  35        *opt->pattern_tail = p;
  36        opt->pattern_tail = &p->next;
  37        p->next = NULL;
  38}
  39
  40struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
  41{
  42        struct grep_pat *pat;
  43        struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
  44        *ret = *opt;
  45
  46        ret->pattern_list = NULL;
  47        ret->pattern_tail = &ret->pattern_list;
  48
  49        for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
  50        {
  51                if(pat->token == GREP_PATTERN_HEAD)
  52                        append_header_grep_pattern(ret, pat->field,
  53                                                   pat->pattern);
  54                else
  55                        append_grep_pat(ret, pat->pattern, pat->patternlen,
  56                                        pat->origin, pat->no, pat->token);
  57        }
  58
  59        return ret;
  60}
  61
  62static NORETURN void compile_regexp_failed(const struct grep_pat *p,
  63                const char *error)
  64{
  65        char where[1024];
  66
  67        if (p->no)
  68                sprintf(where, "In '%s' at %d, ", p->origin, p->no);
  69        else if (p->origin)
  70                sprintf(where, "%s, ", p->origin);
  71        else
  72                where[0] = 0;
  73
  74        die("%s'%s': %s", where, p->pattern, error);
  75}
  76
  77#ifdef USE_LIBPCRE
  78static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
  79{
  80        const char *error;
  81        int erroffset;
  82        int options = 0;
  83
  84        if (opt->ignore_case)
  85                options |= PCRE_CASELESS;
  86
  87        p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
  88                        NULL);
  89        if (!p->pcre_regexp)
  90                compile_regexp_failed(p, error);
  91
  92        p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
  93        if (!p->pcre_extra_info && error)
  94                die("%s", error);
  95}
  96
  97static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
  98                regmatch_t *match, int eflags)
  99{
 100        int ovector[30], ret, flags = 0;
 101
 102        if (eflags & REG_NOTBOL)
 103                flags |= PCRE_NOTBOL;
 104
 105        ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 106                        0, flags, ovector, ARRAY_SIZE(ovector));
 107        if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 108                die("pcre_exec failed with error code %d", ret);
 109        if (ret > 0) {
 110                ret = 0;
 111                match->rm_so = ovector[0];
 112                match->rm_eo = ovector[1];
 113        }
 114
 115        return ret;
 116}
 117
 118static void free_pcre_regexp(struct grep_pat *p)
 119{
 120        pcre_free(p->pcre_regexp);
 121        pcre_free(p->pcre_extra_info);
 122}
 123#else /* !USE_LIBPCRE */
 124static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 125{
 126        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 127}
 128
 129static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 130                regmatch_t *match, int eflags)
 131{
 132        return 1;
 133}
 134
 135static void free_pcre_regexp(struct grep_pat *p)
 136{
 137}
 138#endif /* !USE_LIBPCRE */
 139
 140static int is_fixed(const char *s, size_t len)
 141{
 142        size_t i;
 143
 144        /* regcomp cannot accept patterns with NULs so we
 145         * consider any pattern containing a NUL fixed.
 146         */
 147        if (memchr(s, 0, len))
 148                return 1;
 149
 150        for (i = 0; i < len; i++) {
 151                if (is_regex_special(s[i]))
 152                        return 0;
 153        }
 154
 155        return 1;
 156}
 157
 158static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 159{
 160        int err;
 161
 162        p->word_regexp = opt->word_regexp;
 163        p->ignore_case = opt->ignore_case;
 164
 165        if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 166                p->fixed = 1;
 167        else
 168                p->fixed = 0;
 169
 170        if (p->fixed) {
 171                if (opt->regflags & REG_ICASE || p->ignore_case)
 172                        p->kws = kwsalloc(tolower_trans_tbl);
 173                else
 174                        p->kws = kwsalloc(NULL);
 175                kwsincr(p->kws, p->pattern, p->patternlen);
 176                kwsprep(p->kws);
 177                return;
 178        }
 179
 180        if (opt->pcre) {
 181                compile_pcre_regexp(p, opt);
 182                return;
 183        }
 184
 185        err = regcomp(&p->regexp, p->pattern, opt->regflags);
 186        if (err) {
 187                char errbuf[1024];
 188                regerror(err, &p->regexp, errbuf, 1024);
 189                regfree(&p->regexp);
 190                compile_regexp_failed(p, errbuf);
 191        }
 192}
 193
 194static struct grep_expr *compile_pattern_or(struct grep_pat **);
 195static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 196{
 197        struct grep_pat *p;
 198        struct grep_expr *x;
 199
 200        p = *list;
 201        if (!p)
 202                return NULL;
 203        switch (p->token) {
 204        case GREP_PATTERN: /* atom */
 205        case GREP_PATTERN_HEAD:
 206        case GREP_PATTERN_BODY:
 207                x = xcalloc(1, sizeof (struct grep_expr));
 208                x->node = GREP_NODE_ATOM;
 209                x->u.atom = p;
 210                *list = p->next;
 211                return x;
 212        case GREP_OPEN_PAREN:
 213                *list = p->next;
 214                x = compile_pattern_or(list);
 215                if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 216                        die("unmatched parenthesis");
 217                *list = (*list)->next;
 218                return x;
 219        default:
 220                return NULL;
 221        }
 222}
 223
 224static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 225{
 226        struct grep_pat *p;
 227        struct grep_expr *x;
 228
 229        p = *list;
 230        if (!p)
 231                return NULL;
 232        switch (p->token) {
 233        case GREP_NOT:
 234                if (!p->next)
 235                        die("--not not followed by pattern expression");
 236                *list = p->next;
 237                x = xcalloc(1, sizeof (struct grep_expr));
 238                x->node = GREP_NODE_NOT;
 239                x->u.unary = compile_pattern_not(list);
 240                if (!x->u.unary)
 241                        die("--not followed by non pattern expression");
 242                return x;
 243        default:
 244                return compile_pattern_atom(list);
 245        }
 246}
 247
 248static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 249{
 250        struct grep_pat *p;
 251        struct grep_expr *x, *y, *z;
 252
 253        x = compile_pattern_not(list);
 254        p = *list;
 255        if (p && p->token == GREP_AND) {
 256                if (!p->next)
 257                        die("--and not followed by pattern expression");
 258                *list = p->next;
 259                y = compile_pattern_and(list);
 260                if (!y)
 261                        die("--and not followed by pattern expression");
 262                z = xcalloc(1, sizeof (struct grep_expr));
 263                z->node = GREP_NODE_AND;
 264                z->u.binary.left = x;
 265                z->u.binary.right = y;
 266                return z;
 267        }
 268        return x;
 269}
 270
 271static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 272{
 273        struct grep_pat *p;
 274        struct grep_expr *x, *y, *z;
 275
 276        x = compile_pattern_and(list);
 277        p = *list;
 278        if (x && p && p->token != GREP_CLOSE_PAREN) {
 279                y = compile_pattern_or(list);
 280                if (!y)
 281                        die("not a pattern expression %s", p->pattern);
 282                z = xcalloc(1, sizeof (struct grep_expr));
 283                z->node = GREP_NODE_OR;
 284                z->u.binary.left = x;
 285                z->u.binary.right = y;
 286                return z;
 287        }
 288        return x;
 289}
 290
 291static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 292{
 293        return compile_pattern_or(list);
 294}
 295
 296static struct grep_expr *grep_true_expr(void)
 297{
 298        struct grep_expr *z = xcalloc(1, sizeof(*z));
 299        z->node = GREP_NODE_TRUE;
 300        return z;
 301}
 302
 303static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 304{
 305        struct grep_expr *z = xcalloc(1, sizeof(*z));
 306        z->node = GREP_NODE_OR;
 307        z->u.binary.left = left;
 308        z->u.binary.right = right;
 309        return z;
 310}
 311
 312static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 313{
 314        struct grep_pat *p;
 315        struct grep_expr *header_expr;
 316        struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 317        enum grep_header_field fld;
 318
 319        if (!opt->header_list)
 320                return NULL;
 321        p = opt->header_list;
 322        for (p = opt->header_list; p; p = p->next) {
 323                if (p->token != GREP_PATTERN_HEAD)
 324                        die("bug: a non-header pattern in grep header list.");
 325                if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
 326                        die("bug: unknown header field %d", p->field);
 327                compile_regexp(p, opt);
 328        }
 329
 330        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 331                header_group[fld] = NULL;
 332
 333        for (p = opt->header_list; p; p = p->next) {
 334                struct grep_expr *h;
 335                struct grep_pat *pp = p;
 336
 337                h = compile_pattern_atom(&pp);
 338                if (!h || pp != p->next)
 339                        die("bug: malformed header expr");
 340                if (!header_group[p->field]) {
 341                        header_group[p->field] = h;
 342                        continue;
 343                }
 344                header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 345        }
 346
 347        header_expr = NULL;
 348
 349        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 350                if (!header_group[fld])
 351                        continue;
 352                if (!header_expr)
 353                        header_expr = grep_true_expr();
 354                header_expr = grep_or_expr(header_group[fld], header_expr);
 355        }
 356        return header_expr;
 357}
 358
 359void compile_grep_patterns(struct grep_opt *opt)
 360{
 361        struct grep_pat *p;
 362        struct grep_expr *header_expr = prep_header_patterns(opt);
 363
 364        for (p = opt->pattern_list; p; p = p->next) {
 365                switch (p->token) {
 366                case GREP_PATTERN: /* atom */
 367                case GREP_PATTERN_HEAD:
 368                case GREP_PATTERN_BODY:
 369                        compile_regexp(p, opt);
 370                        break;
 371                default:
 372                        opt->extended = 1;
 373                        break;
 374                }
 375        }
 376
 377        if (opt->all_match || header_expr)
 378                opt->extended = 1;
 379        else if (!opt->extended)
 380                return;
 381
 382        p = opt->pattern_list;
 383        if (p)
 384                opt->pattern_expression = compile_pattern_expr(&p);
 385        if (p)
 386                die("incomplete pattern expression: %s", p->pattern);
 387
 388        if (!header_expr)
 389                return;
 390
 391        if (!opt->pattern_expression)
 392                opt->pattern_expression = header_expr;
 393        else
 394                opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 395                                                       header_expr);
 396        opt->all_match = 1;
 397}
 398
 399static void free_pattern_expr(struct grep_expr *x)
 400{
 401        switch (x->node) {
 402        case GREP_NODE_TRUE:
 403        case GREP_NODE_ATOM:
 404                break;
 405        case GREP_NODE_NOT:
 406                free_pattern_expr(x->u.unary);
 407                break;
 408        case GREP_NODE_AND:
 409        case GREP_NODE_OR:
 410                free_pattern_expr(x->u.binary.left);
 411                free_pattern_expr(x->u.binary.right);
 412                break;
 413        }
 414        free(x);
 415}
 416
 417void free_grep_patterns(struct grep_opt *opt)
 418{
 419        struct grep_pat *p, *n;
 420
 421        for (p = opt->pattern_list; p; p = n) {
 422                n = p->next;
 423                switch (p->token) {
 424                case GREP_PATTERN: /* atom */
 425                case GREP_PATTERN_HEAD:
 426                case GREP_PATTERN_BODY:
 427                        if (p->kws)
 428                                kwsfree(p->kws);
 429                        else if (p->pcre_regexp)
 430                                free_pcre_regexp(p);
 431                        else
 432                                regfree(&p->regexp);
 433                        break;
 434                default:
 435                        break;
 436                }
 437                free(p);
 438        }
 439
 440        if (!opt->extended)
 441                return;
 442        free_pattern_expr(opt->pattern_expression);
 443}
 444
 445static char *end_of_line(char *cp, unsigned long *left)
 446{
 447        unsigned long l = *left;
 448        while (l && *cp != '\n') {
 449                l--;
 450                cp++;
 451        }
 452        *left = l;
 453        return cp;
 454}
 455
 456static int word_char(char ch)
 457{
 458        return isalnum(ch) || ch == '_';
 459}
 460
 461static void output_color(struct grep_opt *opt, const void *data, size_t size,
 462                         const char *color)
 463{
 464        if (want_color(opt->color) && color && color[0]) {
 465                opt->output(opt, color, strlen(color));
 466                opt->output(opt, data, size);
 467                opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 468        } else
 469                opt->output(opt, data, size);
 470}
 471
 472static void output_sep(struct grep_opt *opt, char sign)
 473{
 474        if (opt->null_following_name)
 475                opt->output(opt, "\0", 1);
 476        else
 477                output_color(opt, &sign, 1, opt->color_sep);
 478}
 479
 480static void show_name(struct grep_opt *opt, const char *name)
 481{
 482        output_color(opt, name, strlen(name), opt->color_filename);
 483        opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 484}
 485
 486static int fixmatch(struct grep_pat *p, char *line, char *eol,
 487                    regmatch_t *match)
 488{
 489        struct kwsmatch kwsm;
 490        size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 491        if (offset == -1) {
 492                match->rm_so = match->rm_eo = -1;
 493                return REG_NOMATCH;
 494        } else {
 495                match->rm_so = offset;
 496                match->rm_eo = match->rm_so + kwsm.size[0];
 497                return 0;
 498        }
 499}
 500
 501static int regmatch(const regex_t *preg, char *line, char *eol,
 502                    regmatch_t *match, int eflags)
 503{
 504#ifdef REG_STARTEND
 505        match->rm_so = 0;
 506        match->rm_eo = eol - line;
 507        eflags |= REG_STARTEND;
 508#endif
 509        return regexec(preg, line, 1, match, eflags);
 510}
 511
 512static int patmatch(struct grep_pat *p, char *line, char *eol,
 513                    regmatch_t *match, int eflags)
 514{
 515        int hit;
 516
 517        if (p->fixed)
 518                hit = !fixmatch(p, line, eol, match);
 519        else if (p->pcre_regexp)
 520                hit = !pcrematch(p, line, eol, match, eflags);
 521        else
 522                hit = !regmatch(&p->regexp, line, eol, match, eflags);
 523
 524        return hit;
 525}
 526
 527static int strip_timestamp(char *bol, char **eol_p)
 528{
 529        char *eol = *eol_p;
 530        int ch;
 531
 532        while (bol < --eol) {
 533                if (*eol != '>')
 534                        continue;
 535                *eol_p = ++eol;
 536                ch = *eol;
 537                *eol = '\0';
 538                return ch;
 539        }
 540        return 0;
 541}
 542
 543static struct {
 544        const char *field;
 545        size_t len;
 546} header_field[] = {
 547        { "author ", 7 },
 548        { "committer ", 10 },
 549};
 550
 551static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 552                             enum grep_context ctx,
 553                             regmatch_t *pmatch, int eflags)
 554{
 555        int hit = 0;
 556        int saved_ch = 0;
 557        const char *start = bol;
 558
 559        if ((p->token != GREP_PATTERN) &&
 560            ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 561                return 0;
 562
 563        if (p->token == GREP_PATTERN_HEAD) {
 564                const char *field;
 565                size_t len;
 566                assert(p->field < ARRAY_SIZE(header_field));
 567                field = header_field[p->field].field;
 568                len = header_field[p->field].len;
 569                if (strncmp(bol, field, len))
 570                        return 0;
 571                bol += len;
 572                saved_ch = strip_timestamp(bol, &eol);
 573        }
 574
 575 again:
 576        hit = patmatch(p, bol, eol, pmatch, eflags);
 577
 578        if (hit && p->word_regexp) {
 579                if ((pmatch[0].rm_so < 0) ||
 580                    (eol - bol) < pmatch[0].rm_so ||
 581                    (pmatch[0].rm_eo < 0) ||
 582                    (eol - bol) < pmatch[0].rm_eo)
 583                        die("regexp returned nonsense");
 584
 585                /* Match beginning must be either beginning of the
 586                 * line, or at word boundary (i.e. the last char must
 587                 * not be a word char).  Similarly, match end must be
 588                 * either end of the line, or at word boundary
 589                 * (i.e. the next char must not be a word char).
 590                 */
 591                if ( ((pmatch[0].rm_so == 0) ||
 592                      !word_char(bol[pmatch[0].rm_so-1])) &&
 593                     ((pmatch[0].rm_eo == (eol-bol)) ||
 594                      !word_char(bol[pmatch[0].rm_eo])) )
 595                        ;
 596                else
 597                        hit = 0;
 598
 599                /* Words consist of at least one character. */
 600                if (pmatch->rm_so == pmatch->rm_eo)
 601                        hit = 0;
 602
 603                if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 604                        /* There could be more than one match on the
 605                         * line, and the first match might not be
 606                         * strict word match.  But later ones could be!
 607                         * Forward to the next possible start, i.e. the
 608                         * next position following a non-word char.
 609                         */
 610                        bol = pmatch[0].rm_so + bol + 1;
 611                        while (word_char(bol[-1]) && bol < eol)
 612                                bol++;
 613                        eflags |= REG_NOTBOL;
 614                        if (bol < eol)
 615                                goto again;
 616                }
 617        }
 618        if (p->token == GREP_PATTERN_HEAD && saved_ch)
 619                *eol = saved_ch;
 620        if (hit) {
 621                pmatch[0].rm_so += bol - start;
 622                pmatch[0].rm_eo += bol - start;
 623        }
 624        return hit;
 625}
 626
 627static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 628                           enum grep_context ctx, int collect_hits)
 629{
 630        int h = 0;
 631        regmatch_t match;
 632
 633        if (!x)
 634                die("Not a valid grep expression");
 635        switch (x->node) {
 636        case GREP_NODE_TRUE:
 637                h = 1;
 638                break;
 639        case GREP_NODE_ATOM:
 640                h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 641                break;
 642        case GREP_NODE_NOT:
 643                h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 644                break;
 645        case GREP_NODE_AND:
 646                if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 647                        return 0;
 648                h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 649                break;
 650        case GREP_NODE_OR:
 651                if (!collect_hits)
 652                        return (match_expr_eval(x->u.binary.left,
 653                                                bol, eol, ctx, 0) ||
 654                                match_expr_eval(x->u.binary.right,
 655                                                bol, eol, ctx, 0));
 656                h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 657                x->u.binary.left->hit |= h;
 658                h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 659                break;
 660        default:
 661                die("Unexpected node type (internal error) %d", x->node);
 662        }
 663        if (collect_hits)
 664                x->hit |= h;
 665        return h;
 666}
 667
 668static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 669                      enum grep_context ctx, int collect_hits)
 670{
 671        struct grep_expr *x = opt->pattern_expression;
 672        return match_expr_eval(x, bol, eol, ctx, collect_hits);
 673}
 674
 675static int match_line(struct grep_opt *opt, char *bol, char *eol,
 676                      enum grep_context ctx, int collect_hits)
 677{
 678        struct grep_pat *p;
 679        regmatch_t match;
 680
 681        if (opt->extended)
 682                return match_expr(opt, bol, eol, ctx, collect_hits);
 683
 684        /* we do not call with collect_hits without being extended */
 685        for (p = opt->pattern_list; p; p = p->next) {
 686                if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 687                        return 1;
 688        }
 689        return 0;
 690}
 691
 692static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 693                              enum grep_context ctx,
 694                              regmatch_t *pmatch, int eflags)
 695{
 696        regmatch_t match;
 697
 698        if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 699                return 0;
 700        if (match.rm_so < 0 || match.rm_eo < 0)
 701                return 0;
 702        if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 703                if (match.rm_so > pmatch->rm_so)
 704                        return 1;
 705                if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 706                        return 1;
 707        }
 708        pmatch->rm_so = match.rm_so;
 709        pmatch->rm_eo = match.rm_eo;
 710        return 1;
 711}
 712
 713static int next_match(struct grep_opt *opt, char *bol, char *eol,
 714                      enum grep_context ctx, regmatch_t *pmatch, int eflags)
 715{
 716        struct grep_pat *p;
 717        int hit = 0;
 718
 719        pmatch->rm_so = pmatch->rm_eo = -1;
 720        if (bol < eol) {
 721                for (p = opt->pattern_list; p; p = p->next) {
 722                        switch (p->token) {
 723                        case GREP_PATTERN: /* atom */
 724                        case GREP_PATTERN_HEAD:
 725                        case GREP_PATTERN_BODY:
 726                                hit |= match_next_pattern(p, bol, eol, ctx,
 727                                                          pmatch, eflags);
 728                                break;
 729                        default:
 730                                break;
 731                        }
 732                }
 733        }
 734        return hit;
 735}
 736
 737static void show_line(struct grep_opt *opt, char *bol, char *eol,
 738                      const char *name, unsigned lno, char sign)
 739{
 740        int rest = eol - bol;
 741        char *line_color = NULL;
 742
 743        if (opt->file_break && opt->last_shown == 0) {
 744                if (opt->show_hunk_mark)
 745                        opt->output(opt, "\n", 1);
 746        } else if (opt->pre_context || opt->post_context || opt->funcbody) {
 747                if (opt->last_shown == 0) {
 748                        if (opt->show_hunk_mark) {
 749                                output_color(opt, "--", 2, opt->color_sep);
 750                                opt->output(opt, "\n", 1);
 751                        }
 752                } else if (lno > opt->last_shown + 1) {
 753                        output_color(opt, "--", 2, opt->color_sep);
 754                        opt->output(opt, "\n", 1);
 755                }
 756        }
 757        if (opt->heading && opt->last_shown == 0) {
 758                output_color(opt, name, strlen(name), opt->color_filename);
 759                opt->output(opt, "\n", 1);
 760        }
 761        opt->last_shown = lno;
 762
 763        if (!opt->heading && opt->pathname) {
 764                output_color(opt, name, strlen(name), opt->color_filename);
 765                output_sep(opt, sign);
 766        }
 767        if (opt->linenum) {
 768                char buf[32];
 769                snprintf(buf, sizeof(buf), "%d", lno);
 770                output_color(opt, buf, strlen(buf), opt->color_lineno);
 771                output_sep(opt, sign);
 772        }
 773        if (opt->color) {
 774                regmatch_t match;
 775                enum grep_context ctx = GREP_CONTEXT_BODY;
 776                int ch = *eol;
 777                int eflags = 0;
 778
 779                if (sign == ':')
 780                        line_color = opt->color_selected;
 781                else if (sign == '-')
 782                        line_color = opt->color_context;
 783                else if (sign == '=')
 784                        line_color = opt->color_function;
 785                *eol = '\0';
 786                while (next_match(opt, bol, eol, ctx, &match, eflags)) {
 787                        if (match.rm_so == match.rm_eo)
 788                                break;
 789
 790                        output_color(opt, bol, match.rm_so, line_color);
 791                        output_color(opt, bol + match.rm_so,
 792                                     match.rm_eo - match.rm_so,
 793                                     opt->color_match);
 794                        bol += match.rm_eo;
 795                        rest -= match.rm_eo;
 796                        eflags = REG_NOTBOL;
 797                }
 798                *eol = ch;
 799        }
 800        output_color(opt, bol, rest, line_color);
 801        opt->output(opt, "\n", 1);
 802}
 803
 804static int match_funcname(struct grep_opt *opt, char *bol, char *eol)
 805{
 806        xdemitconf_t *xecfg = opt->priv;
 807        if (xecfg && xecfg->find_func) {
 808                char buf[1];
 809                return xecfg->find_func(bol, eol - bol, buf, 1,
 810                                        xecfg->find_func_priv) >= 0;
 811        }
 812
 813        if (bol == eol)
 814                return 0;
 815        if (isalpha(*bol) || *bol == '_' || *bol == '$')
 816                return 1;
 817        return 0;
 818}
 819
 820static void show_funcname_line(struct grep_opt *opt, const char *name,
 821                               char *buf, char *bol, unsigned lno)
 822{
 823        while (bol > buf) {
 824                char *eol = --bol;
 825
 826                while (bol > buf && bol[-1] != '\n')
 827                        bol--;
 828                lno--;
 829
 830                if (lno <= opt->last_shown)
 831                        break;
 832
 833                if (match_funcname(opt, bol, eol)) {
 834                        show_line(opt, bol, eol, name, lno, '=');
 835                        break;
 836                }
 837        }
 838}
 839
 840static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
 841                             char *bol, char *end, unsigned lno)
 842{
 843        unsigned cur = lno, from = 1, funcname_lno = 0;
 844        int funcname_needed = !!opt->funcname;
 845
 846        if (opt->funcbody && !match_funcname(opt, bol, end))
 847                funcname_needed = 2;
 848
 849        if (opt->pre_context < lno)
 850                from = lno - opt->pre_context;
 851        if (from <= opt->last_shown)
 852                from = opt->last_shown + 1;
 853
 854        /* Rewind. */
 855        while (bol > buf &&
 856               cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
 857                char *eol = --bol;
 858
 859                while (bol > buf && bol[-1] != '\n')
 860                        bol--;
 861                cur--;
 862                if (funcname_needed && match_funcname(opt, bol, eol)) {
 863                        funcname_lno = cur;
 864                        funcname_needed = 0;
 865                }
 866        }
 867
 868        /* We need to look even further back to find a function signature. */
 869        if (opt->funcname && funcname_needed)
 870                show_funcname_line(opt, name, buf, bol, cur);
 871
 872        /* Back forward. */
 873        while (cur < lno) {
 874                char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
 875
 876                while (*eol != '\n')
 877                        eol++;
 878                show_line(opt, bol, eol, name, cur, sign);
 879                bol = eol + 1;
 880                cur++;
 881        }
 882}
 883
 884static int should_lookahead(struct grep_opt *opt)
 885{
 886        struct grep_pat *p;
 887
 888        if (opt->extended)
 889                return 0; /* punt for too complex stuff */
 890        if (opt->invert)
 891                return 0;
 892        for (p = opt->pattern_list; p; p = p->next) {
 893                if (p->token != GREP_PATTERN)
 894                        return 0; /* punt for "header only" and stuff */
 895        }
 896        return 1;
 897}
 898
 899static int look_ahead(struct grep_opt *opt,
 900                      unsigned long *left_p,
 901                      unsigned *lno_p,
 902                      char **bol_p)
 903{
 904        unsigned lno = *lno_p;
 905        char *bol = *bol_p;
 906        struct grep_pat *p;
 907        char *sp, *last_bol;
 908        regoff_t earliest = -1;
 909
 910        for (p = opt->pattern_list; p; p = p->next) {
 911                int hit;
 912                regmatch_t m;
 913
 914                hit = patmatch(p, bol, bol + *left_p, &m, 0);
 915                if (!hit || m.rm_so < 0 || m.rm_eo < 0)
 916                        continue;
 917                if (earliest < 0 || m.rm_so < earliest)
 918                        earliest = m.rm_so;
 919        }
 920
 921        if (earliest < 0) {
 922                *bol_p = bol + *left_p;
 923                *left_p = 0;
 924                return 1;
 925        }
 926        for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
 927                ; /* find the beginning of the line */
 928        last_bol = sp;
 929
 930        for (sp = bol; sp < last_bol; sp++) {
 931                if (*sp == '\n')
 932                        lno++;
 933        }
 934        *left_p -= last_bol - bol;
 935        *bol_p = last_bol;
 936        *lno_p = lno;
 937        return 0;
 938}
 939
 940int grep_threads_ok(const struct grep_opt *opt)
 941{
 942        /* If this condition is true, then we may use the attribute
 943         * machinery in grep_buffer_1. The attribute code is not
 944         * thread safe, so we disable the use of threads.
 945         */
 946        if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
 947            !opt->name_only)
 948                return 0;
 949
 950        return 1;
 951}
 952
 953static void std_output(struct grep_opt *opt, const void *buf, size_t size)
 954{
 955        fwrite(buf, size, 1, stdout);
 956}
 957
 958static int grep_buffer_1(struct grep_opt *opt, const char *name,
 959                         char *buf, unsigned long size, int collect_hits)
 960{
 961        char *bol = buf;
 962        unsigned long left = size;
 963        unsigned lno = 1;
 964        unsigned last_hit = 0;
 965        int binary_match_only = 0;
 966        unsigned count = 0;
 967        int try_lookahead = 0;
 968        int show_function = 0;
 969        enum grep_context ctx = GREP_CONTEXT_HEAD;
 970        xdemitconf_t xecfg;
 971
 972        if (!opt->output)
 973                opt->output = std_output;
 974
 975        if (opt->pre_context || opt->post_context || opt->file_break ||
 976            opt->funcbody) {
 977                /* Show hunk marks, except for the first file. */
 978                if (opt->last_shown)
 979                        opt->show_hunk_mark = 1;
 980                /*
 981                 * If we're using threads then we can't easily identify
 982                 * the first file.  Always put hunk marks in that case
 983                 * and skip the very first one later in work_done().
 984                 */
 985                if (opt->output != std_output)
 986                        opt->show_hunk_mark = 1;
 987        }
 988        opt->last_shown = 0;
 989
 990        switch (opt->binary) {
 991        case GREP_BINARY_DEFAULT:
 992                if (buffer_is_binary(buf, size))
 993                        binary_match_only = 1;
 994                break;
 995        case GREP_BINARY_NOMATCH:
 996                if (buffer_is_binary(buf, size))
 997                        return 0; /* Assume unmatch */
 998                break;
 999        case GREP_BINARY_TEXT:
1000                break;
1001        default:
1002                die("bug: unknown binary handling mode");
1003        }
1004
1005        memset(&xecfg, 0, sizeof(xecfg));
1006        if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
1007            !opt->name_only && !binary_match_only && !collect_hits) {
1008                struct userdiff_driver *drv = userdiff_find_by_path(name);
1009                if (drv && drv->funcname.pattern) {
1010                        const struct userdiff_funcname *pe = &drv->funcname;
1011                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
1012                        opt->priv = &xecfg;
1013                }
1014        }
1015        try_lookahead = should_lookahead(opt);
1016
1017        while (left) {
1018                char *eol, ch;
1019                int hit;
1020
1021                /*
1022                 * look_ahead() skips quickly to the line that possibly
1023                 * has the next hit; don't call it if we need to do
1024                 * something more than just skipping the current line
1025                 * in response to an unmatch for the current line.  E.g.
1026                 * inside a post-context window, we will show the current
1027                 * line as a context around the previous hit when it
1028                 * doesn't hit.
1029                 */
1030                if (try_lookahead
1031                    && !(last_hit
1032                         && (show_function ||
1033                             lno <= last_hit + opt->post_context))
1034                    && look_ahead(opt, &left, &lno, &bol))
1035                        break;
1036                eol = end_of_line(bol, &left);
1037                ch = *eol;
1038                *eol = 0;
1039
1040                if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1041                        ctx = GREP_CONTEXT_BODY;
1042
1043                hit = match_line(opt, bol, eol, ctx, collect_hits);
1044                *eol = ch;
1045
1046                if (collect_hits)
1047                        goto next_line;
1048
1049                /* "grep -v -e foo -e bla" should list lines
1050                 * that do not have either, so inversion should
1051                 * be done outside.
1052                 */
1053                if (opt->invert)
1054                        hit = !hit;
1055                if (opt->unmatch_name_only) {
1056                        if (hit)
1057                                return 0;
1058                        goto next_line;
1059                }
1060                if (hit) {
1061                        count++;
1062                        if (opt->status_only)
1063                                return 1;
1064                        if (opt->name_only) {
1065                                show_name(opt, name);
1066                                return 1;
1067                        }
1068                        if (opt->count)
1069                                goto next_line;
1070                        if (binary_match_only) {
1071                                opt->output(opt, "Binary file ", 12);
1072                                output_color(opt, name, strlen(name),
1073                                             opt->color_filename);
1074                                opt->output(opt, " matches\n", 9);
1075                                return 1;
1076                        }
1077                        /* Hit at this line.  If we haven't shown the
1078                         * pre-context lines, we would need to show them.
1079                         */
1080                        if (opt->pre_context || opt->funcbody)
1081                                show_pre_context(opt, name, buf, bol, eol, lno);
1082                        else if (opt->funcname)
1083                                show_funcname_line(opt, name, buf, bol, lno);
1084                        show_line(opt, bol, eol, name, lno, ':');
1085                        last_hit = lno;
1086                        if (opt->funcbody)
1087                                show_function = 1;
1088                        goto next_line;
1089                }
1090                if (show_function && match_funcname(opt, bol, eol))
1091                        show_function = 0;
1092                if (show_function ||
1093                    (last_hit && lno <= last_hit + opt->post_context)) {
1094                        /* If the last hit is within the post context,
1095                         * we need to show this line.
1096                         */
1097                        show_line(opt, bol, eol, name, lno, '-');
1098                }
1099
1100        next_line:
1101                bol = eol + 1;
1102                if (!left)
1103                        break;
1104                left--;
1105                lno++;
1106        }
1107
1108        if (collect_hits)
1109                return 0;
1110
1111        if (opt->status_only)
1112                return 0;
1113        if (opt->unmatch_name_only) {
1114                /* We did not see any hit, so we want to show this */
1115                show_name(opt, name);
1116                return 1;
1117        }
1118
1119        xdiff_clear_find_func(&xecfg);
1120        opt->priv = NULL;
1121
1122        /* NEEDSWORK:
1123         * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1124         * which feels mostly useless but sometimes useful.  Maybe
1125         * make it another option?  For now suppress them.
1126         */
1127        if (opt->count && count) {
1128                char buf[32];
1129                output_color(opt, name, strlen(name), opt->color_filename);
1130                output_sep(opt, ':');
1131                snprintf(buf, sizeof(buf), "%u\n", count);
1132                opt->output(opt, buf, strlen(buf));
1133                return 1;
1134        }
1135        return !!last_hit;
1136}
1137
1138static void clr_hit_marker(struct grep_expr *x)
1139{
1140        /* All-hit markers are meaningful only at the very top level
1141         * OR node.
1142         */
1143        while (1) {
1144                x->hit = 0;
1145                if (x->node != GREP_NODE_OR)
1146                        return;
1147                x->u.binary.left->hit = 0;
1148                x = x->u.binary.right;
1149        }
1150}
1151
1152static int chk_hit_marker(struct grep_expr *x)
1153{
1154        /* Top level nodes have hit markers.  See if they all are hits */
1155        while (1) {
1156                if (x->node != GREP_NODE_OR)
1157                        return x->hit;
1158                if (!x->u.binary.left->hit)
1159                        return 0;
1160                x = x->u.binary.right;
1161        }
1162}
1163
1164int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
1165{
1166        /*
1167         * we do not have to do the two-pass grep when we do not check
1168         * buffer-wide "all-match".
1169         */
1170        if (!opt->all_match)
1171                return grep_buffer_1(opt, name, buf, size, 0);
1172
1173        /* Otherwise the toplevel "or" terms hit a bit differently.
1174         * We first clear hit markers from them.
1175         */
1176        clr_hit_marker(opt->pattern_expression);
1177        grep_buffer_1(opt, name, buf, size, 1);
1178
1179        if (!chk_hit_marker(opt->pattern_expression))
1180                return 0;
1181
1182        return grep_buffer_1(opt, name, buf, size, 0);
1183}