Andrew's git / gitweb.git / grep.c
?
grep.con commit test-regex: expose full regcomp() to the command line (d8acfe1)
   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
   8static int grep_source_load(struct grep_source *gs);
   9static int grep_source_is_binary(struct grep_source *gs);
  10
  11static struct grep_opt grep_defaults;
  12
  13/*
  14 * Initialize the grep_defaults template with hardcoded defaults.
  15 * We could let the compiler do this, but without C99 initializers
  16 * the code gets unwieldy and unreadable, so...
  17 */
  18void init_grep_defaults(void)
  19{
  20        struct grep_opt *opt = &grep_defaults;
  21        static int run_once;
  22
  23        if (run_once)
  24                return;
  25        run_once++;
  26
  27        memset(opt, 0, sizeof(*opt));
  28        opt->relative = 1;
  29        opt->pathname = 1;
  30        opt->regflags = REG_NEWLINE;
  31        opt->max_depth = -1;
  32        opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  33        opt->extended_regexp_option = 0;
  34        color_set(opt->color_context, "");
  35        color_set(opt->color_filename, "");
  36        color_set(opt->color_function, "");
  37        color_set(opt->color_lineno, "");
  38        color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
  39        color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
  40        color_set(opt->color_selected, "");
  41        color_set(opt->color_sep, GIT_COLOR_CYAN);
  42        opt->color = -1;
  43}
  44
  45static int parse_pattern_type_arg(const char *opt, const char *arg)
  46{
  47        if (!strcmp(arg, "default"))
  48                return GREP_PATTERN_TYPE_UNSPECIFIED;
  49        else if (!strcmp(arg, "basic"))
  50                return GREP_PATTERN_TYPE_BRE;
  51        else if (!strcmp(arg, "extended"))
  52                return GREP_PATTERN_TYPE_ERE;
  53        else if (!strcmp(arg, "fixed"))
  54                return GREP_PATTERN_TYPE_FIXED;
  55        else if (!strcmp(arg, "perl"))
  56                return GREP_PATTERN_TYPE_PCRE;
  57        die("bad %s argument: %s", opt, arg);
  58}
  59
  60/*
  61 * Read the configuration file once and store it in
  62 * the grep_defaults template.
  63 */
  64int grep_config(const char *var, const char *value, void *cb)
  65{
  66        struct grep_opt *opt = &grep_defaults;
  67        char *color = NULL;
  68
  69        if (userdiff_config(var, value) < 0)
  70                return -1;
  71
  72        if (!strcmp(var, "grep.extendedregexp")) {
  73                if (git_config_bool(var, value))
  74                        opt->extended_regexp_option = 1;
  75                else
  76                        opt->extended_regexp_option = 0;
  77                return 0;
  78        }
  79
  80        if (!strcmp(var, "grep.patterntype")) {
  81                opt->pattern_type_option = parse_pattern_type_arg(var, value);
  82                return 0;
  83        }
  84
  85        if (!strcmp(var, "grep.linenumber")) {
  86                opt->linenum = git_config_bool(var, value);
  87                return 0;
  88        }
  89
  90        if (!strcmp(var, "grep.fullname")) {
  91                opt->relative = !git_config_bool(var, value);
  92                return 0;
  93        }
  94
  95        if (!strcmp(var, "color.grep"))
  96                opt->color = git_config_colorbool(var, value);
  97        else if (!strcmp(var, "color.grep.context"))
  98                color = opt->color_context;
  99        else if (!strcmp(var, "color.grep.filename"))
 100                color = opt->color_filename;
 101        else if (!strcmp(var, "color.grep.function"))
 102                color = opt->color_function;
 103        else if (!strcmp(var, "color.grep.linenumber"))
 104                color = opt->color_lineno;
 105        else if (!strcmp(var, "color.grep.matchcontext"))
 106                color = opt->color_match_context;
 107        else if (!strcmp(var, "color.grep.matchselected"))
 108                color = opt->color_match_selected;
 109        else if (!strcmp(var, "color.grep.selected"))
 110                color = opt->color_selected;
 111        else if (!strcmp(var, "color.grep.separator"))
 112                color = opt->color_sep;
 113        else if (!strcmp(var, "color.grep.match")) {
 114                int rc = 0;
 115                if (!value)
 116                        return config_error_nonbool(var);
 117                rc |= color_parse(value, opt->color_match_context);
 118                rc |= color_parse(value, opt->color_match_selected);
 119                return rc;
 120        }
 121
 122        if (color) {
 123                if (!value)
 124                        return config_error_nonbool(var);
 125                return color_parse(value, color);
 126        }
 127        return 0;
 128}
 129
 130/*
 131 * Initialize one instance of grep_opt and copy the
 132 * default values from the template we read the configuration
 133 * information in an earlier call to git_config(grep_config).
 134 */
 135void grep_init(struct grep_opt *opt, const char *prefix)
 136{
 137        struct grep_opt *def = &grep_defaults;
 138
 139        memset(opt, 0, sizeof(*opt));
 140        opt->prefix = prefix;
 141        opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 142        opt->pattern_tail = &opt->pattern_list;
 143        opt->header_tail = &opt->header_list;
 144
 145        opt->color = def->color;
 146        opt->extended_regexp_option = def->extended_regexp_option;
 147        opt->pattern_type_option = def->pattern_type_option;
 148        opt->linenum = def->linenum;
 149        opt->max_depth = def->max_depth;
 150        opt->pathname = def->pathname;
 151        opt->regflags = def->regflags;
 152        opt->relative = def->relative;
 153
 154        color_set(opt->color_context, def->color_context);
 155        color_set(opt->color_filename, def->color_filename);
 156        color_set(opt->color_function, def->color_function);
 157        color_set(opt->color_lineno, def->color_lineno);
 158        color_set(opt->color_match_context, def->color_match_context);
 159        color_set(opt->color_match_selected, def->color_match_selected);
 160        color_set(opt->color_selected, def->color_selected);
 161        color_set(opt->color_sep, def->color_sep);
 162}
 163
 164void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 165{
 166        if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 167                grep_set_pattern_type_option(pattern_type, opt);
 168        else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 169                grep_set_pattern_type_option(opt->pattern_type_option, opt);
 170        else if (opt->extended_regexp_option)
 171                grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 172}
 173
 174void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 175{
 176        switch (pattern_type) {
 177        case GREP_PATTERN_TYPE_UNSPECIFIED:
 178                /* fall through */
 179
 180        case GREP_PATTERN_TYPE_BRE:
 181                opt->fixed = 0;
 182                opt->pcre = 0;
 183                opt->regflags &= ~REG_EXTENDED;
 184                break;
 185
 186        case GREP_PATTERN_TYPE_ERE:
 187                opt->fixed = 0;
 188                opt->pcre = 0;
 189                opt->regflags |= REG_EXTENDED;
 190                break;
 191
 192        case GREP_PATTERN_TYPE_FIXED:
 193                opt->fixed = 1;
 194                opt->pcre = 0;
 195                opt->regflags &= ~REG_EXTENDED;
 196                break;
 197
 198        case GREP_PATTERN_TYPE_PCRE:
 199                opt->fixed = 0;
 200                opt->pcre = 1;
 201                opt->regflags &= ~REG_EXTENDED;
 202                break;
 203        }
 204}
 205
 206static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 207                                        const char *origin, int no,
 208                                        enum grep_pat_token t,
 209                                        enum grep_header_field field)
 210{
 211        struct grep_pat *p = xcalloc(1, sizeof(*p));
 212        p->pattern = xmemdupz(pat, patlen);
 213        p->patternlen = patlen;
 214        p->origin = origin;
 215        p->no = no;
 216        p->token = t;
 217        p->field = field;
 218        return p;
 219}
 220
 221static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 222{
 223        **tail = p;
 224        *tail = &p->next;
 225        p->next = NULL;
 226
 227        switch (p->token) {
 228        case GREP_PATTERN: /* atom */
 229        case GREP_PATTERN_HEAD:
 230        case GREP_PATTERN_BODY:
 231                for (;;) {
 232                        struct grep_pat *new_pat;
 233                        size_t len = 0;
 234                        char *cp = p->pattern + p->patternlen, *nl = NULL;
 235                        while (++len <= p->patternlen) {
 236                                if (*(--cp) == '\n') {
 237                                        nl = cp;
 238                                        break;
 239                                }
 240                        }
 241                        if (!nl)
 242                                break;
 243                        new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 244                                                  p->no, p->token, p->field);
 245                        new_pat->next = p->next;
 246                        if (!p->next)
 247                                *tail = &new_pat->next;
 248                        p->next = new_pat;
 249                        *nl = '\0';
 250                        p->patternlen -= len;
 251                }
 252                break;
 253        default:
 254                break;
 255        }
 256}
 257
 258void append_header_grep_pattern(struct grep_opt *opt,
 259                                enum grep_header_field field, const char *pat)
 260{
 261        struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 262                                             GREP_PATTERN_HEAD, field);
 263        if (field == GREP_HEADER_REFLOG)
 264                opt->use_reflog_filter = 1;
 265        do_append_grep_pat(&opt->header_tail, p);
 266}
 267
 268void append_grep_pattern(struct grep_opt *opt, const char *pat,
 269                         const char *origin, int no, enum grep_pat_token t)
 270{
 271        append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 272}
 273
 274void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 275                     const char *origin, int no, enum grep_pat_token t)
 276{
 277        struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 278        do_append_grep_pat(&opt->pattern_tail, p);
 279}
 280
 281struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 282{
 283        struct grep_pat *pat;
 284        struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 285        *ret = *opt;
 286
 287        ret->pattern_list = NULL;
 288        ret->pattern_tail = &ret->pattern_list;
 289
 290        for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 291        {
 292                if(pat->token == GREP_PATTERN_HEAD)
 293                        append_header_grep_pattern(ret, pat->field,
 294                                                   pat->pattern);
 295                else
 296                        append_grep_pat(ret, pat->pattern, pat->patternlen,
 297                                        pat->origin, pat->no, pat->token);
 298        }
 299
 300        return ret;
 301}
 302
 303static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 304                const char *error)
 305{
 306        char where[1024];
 307
 308        if (p->no)
 309                xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 310        else if (p->origin)
 311                xsnprintf(where, sizeof(where), "%s, ", p->origin);
 312        else
 313                where[0] = 0;
 314
 315        die("%s'%s': %s", where, p->pattern, error);
 316}
 317
 318#ifdef USE_LIBPCRE
 319static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 320{
 321        const char *error;
 322        int erroffset;
 323        int options = PCRE_MULTILINE;
 324
 325        if (opt->ignore_case)
 326                options |= PCRE_CASELESS;
 327
 328        p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 329                        NULL);
 330        if (!p->pcre_regexp)
 331                compile_regexp_failed(p, error);
 332
 333        p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 334        if (!p->pcre_extra_info && error)
 335                die("%s", error);
 336}
 337
 338static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 339                regmatch_t *match, int eflags)
 340{
 341        int ovector[30], ret, flags = 0;
 342
 343        if (eflags & REG_NOTBOL)
 344                flags |= PCRE_NOTBOL;
 345
 346        ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 347                        0, flags, ovector, ARRAY_SIZE(ovector));
 348        if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 349                die("pcre_exec failed with error code %d", ret);
 350        if (ret > 0) {
 351                ret = 0;
 352                match->rm_so = ovector[0];
 353                match->rm_eo = ovector[1];
 354        }
 355
 356        return ret;
 357}
 358
 359static void free_pcre_regexp(struct grep_pat *p)
 360{
 361        pcre_free(p->pcre_regexp);
 362        pcre_free(p->pcre_extra_info);
 363}
 364#else /* !USE_LIBPCRE */
 365static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 366{
 367        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 368}
 369
 370static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 371                regmatch_t *match, int eflags)
 372{
 373        return 1;
 374}
 375
 376static void free_pcre_regexp(struct grep_pat *p)
 377{
 378}
 379#endif /* !USE_LIBPCRE */
 380
 381static int is_fixed(const char *s, size_t len)
 382{
 383        size_t i;
 384
 385        /* regcomp cannot accept patterns with NULs so we
 386         * consider any pattern containing a NUL fixed.
 387         */
 388        if (memchr(s, 0, len))
 389                return 1;
 390
 391        for (i = 0; i < len; i++) {
 392                if (is_regex_special(s[i]))
 393                        return 0;
 394        }
 395
 396        return 1;
 397}
 398
 399static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 400{
 401        int err;
 402
 403        p->word_regexp = opt->word_regexp;
 404        p->ignore_case = opt->ignore_case;
 405
 406        if (opt->fixed)
 407                p->fixed = 1;
 408        else if (is_fixed(p->pattern, p->patternlen))
 409                p->fixed = 1;
 410        else
 411                p->fixed = 0;
 412
 413        if (p->fixed) {
 414                if (opt->regflags & REG_ICASE || p->ignore_case)
 415                        p->kws = kwsalloc(tolower_trans_tbl);
 416                else
 417                        p->kws = kwsalloc(NULL);
 418                kwsincr(p->kws, p->pattern, p->patternlen);
 419                kwsprep(p->kws);
 420                return;
 421        }
 422
 423        if (opt->pcre) {
 424                compile_pcre_regexp(p, opt);
 425                return;
 426        }
 427
 428        err = regcomp(&p->regexp, p->pattern, opt->regflags);
 429        if (err) {
 430                char errbuf[1024];
 431                regerror(err, &p->regexp, errbuf, 1024);
 432                regfree(&p->regexp);
 433                compile_regexp_failed(p, errbuf);
 434        }
 435}
 436
 437static struct grep_expr *compile_pattern_or(struct grep_pat **);
 438static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 439{
 440        struct grep_pat *p;
 441        struct grep_expr *x;
 442
 443        p = *list;
 444        if (!p)
 445                return NULL;
 446        switch (p->token) {
 447        case GREP_PATTERN: /* atom */
 448        case GREP_PATTERN_HEAD:
 449        case GREP_PATTERN_BODY:
 450                x = xcalloc(1, sizeof (struct grep_expr));
 451                x->node = GREP_NODE_ATOM;
 452                x->u.atom = p;
 453                *list = p->next;
 454                return x;
 455        case GREP_OPEN_PAREN:
 456                *list = p->next;
 457                x = compile_pattern_or(list);
 458                if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 459                        die("unmatched parenthesis");
 460                *list = (*list)->next;
 461                return x;
 462        default:
 463                return NULL;
 464        }
 465}
 466
 467static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 468{
 469        struct grep_pat *p;
 470        struct grep_expr *x;
 471
 472        p = *list;
 473        if (!p)
 474                return NULL;
 475        switch (p->token) {
 476        case GREP_NOT:
 477                if (!p->next)
 478                        die("--not not followed by pattern expression");
 479                *list = p->next;
 480                x = xcalloc(1, sizeof (struct grep_expr));
 481                x->node = GREP_NODE_NOT;
 482                x->u.unary = compile_pattern_not(list);
 483                if (!x->u.unary)
 484                        die("--not followed by non pattern expression");
 485                return x;
 486        default:
 487                return compile_pattern_atom(list);
 488        }
 489}
 490
 491static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 492{
 493        struct grep_pat *p;
 494        struct grep_expr *x, *y, *z;
 495
 496        x = compile_pattern_not(list);
 497        p = *list;
 498        if (p && p->token == GREP_AND) {
 499                if (!p->next)
 500                        die("--and not followed by pattern expression");
 501                *list = p->next;
 502                y = compile_pattern_and(list);
 503                if (!y)
 504                        die("--and not followed by pattern expression");
 505                z = xcalloc(1, sizeof (struct grep_expr));
 506                z->node = GREP_NODE_AND;
 507                z->u.binary.left = x;
 508                z->u.binary.right = y;
 509                return z;
 510        }
 511        return x;
 512}
 513
 514static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 515{
 516        struct grep_pat *p;
 517        struct grep_expr *x, *y, *z;
 518
 519        x = compile_pattern_and(list);
 520        p = *list;
 521        if (x && p && p->token != GREP_CLOSE_PAREN) {
 522                y = compile_pattern_or(list);
 523                if (!y)
 524                        die("not a pattern expression %s", p->pattern);
 525                z = xcalloc(1, sizeof (struct grep_expr));
 526                z->node = GREP_NODE_OR;
 527                z->u.binary.left = x;
 528                z->u.binary.right = y;
 529                return z;
 530        }
 531        return x;
 532}
 533
 534static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 535{
 536        return compile_pattern_or(list);
 537}
 538
 539static void indent(int in)
 540{
 541        while (in-- > 0)
 542                fputc(' ', stderr);
 543}
 544
 545static void dump_grep_pat(struct grep_pat *p)
 546{
 547        switch (p->token) {
 548        case GREP_AND: fprintf(stderr, "*and*"); break;
 549        case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 550        case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 551        case GREP_NOT: fprintf(stderr, "*not*"); break;
 552        case GREP_OR: fprintf(stderr, "*or*"); break;
 553
 554        case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 555        case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 556        case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 557        }
 558
 559        switch (p->token) {
 560        default: break;
 561        case GREP_PATTERN_HEAD:
 562                fprintf(stderr, "<head %d>", p->field); break;
 563        case GREP_PATTERN_BODY:
 564                fprintf(stderr, "<body>"); break;
 565        }
 566        switch (p->token) {
 567        default: break;
 568        case GREP_PATTERN_HEAD:
 569        case GREP_PATTERN_BODY:
 570        case GREP_PATTERN:
 571                fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 572                break;
 573        }
 574        fputc('\n', stderr);
 575}
 576
 577static void dump_grep_expression_1(struct grep_expr *x, int in)
 578{
 579        indent(in);
 580        switch (x->node) {
 581        case GREP_NODE_TRUE:
 582                fprintf(stderr, "true\n");
 583                break;
 584        case GREP_NODE_ATOM:
 585                dump_grep_pat(x->u.atom);
 586                break;
 587        case GREP_NODE_NOT:
 588                fprintf(stderr, "(not\n");
 589                dump_grep_expression_1(x->u.unary, in+1);
 590                indent(in);
 591                fprintf(stderr, ")\n");
 592                break;
 593        case GREP_NODE_AND:
 594                fprintf(stderr, "(and\n");
 595                dump_grep_expression_1(x->u.binary.left, in+1);
 596                dump_grep_expression_1(x->u.binary.right, in+1);
 597                indent(in);
 598                fprintf(stderr, ")\n");
 599                break;
 600        case GREP_NODE_OR:
 601                fprintf(stderr, "(or\n");
 602                dump_grep_expression_1(x->u.binary.left, in+1);
 603                dump_grep_expression_1(x->u.binary.right, in+1);
 604                indent(in);
 605                fprintf(stderr, ")\n");
 606                break;
 607        }
 608}
 609
 610static void dump_grep_expression(struct grep_opt *opt)
 611{
 612        struct grep_expr *x = opt->pattern_expression;
 613
 614        if (opt->all_match)
 615                fprintf(stderr, "[all-match]\n");
 616        dump_grep_expression_1(x, 0);
 617        fflush(NULL);
 618}
 619
 620static struct grep_expr *grep_true_expr(void)
 621{
 622        struct grep_expr *z = xcalloc(1, sizeof(*z));
 623        z->node = GREP_NODE_TRUE;
 624        return z;
 625}
 626
 627static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 628{
 629        struct grep_expr *z = xcalloc(1, sizeof(*z));
 630        z->node = GREP_NODE_OR;
 631        z->u.binary.left = left;
 632        z->u.binary.right = right;
 633        return z;
 634}
 635
 636static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 637{
 638        struct grep_pat *p;
 639        struct grep_expr *header_expr;
 640        struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 641        enum grep_header_field fld;
 642
 643        if (!opt->header_list)
 644                return NULL;
 645
 646        for (p = opt->header_list; p; p = p->next) {
 647                if (p->token != GREP_PATTERN_HEAD)
 648                        die("bug: a non-header pattern in grep header list.");
 649                if (p->field < GREP_HEADER_FIELD_MIN ||
 650                    GREP_HEADER_FIELD_MAX <= p->field)
 651                        die("bug: unknown header field %d", p->field);
 652                compile_regexp(p, opt);
 653        }
 654
 655        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 656                header_group[fld] = NULL;
 657
 658        for (p = opt->header_list; p; p = p->next) {
 659                struct grep_expr *h;
 660                struct grep_pat *pp = p;
 661
 662                h = compile_pattern_atom(&pp);
 663                if (!h || pp != p->next)
 664                        die("bug: malformed header expr");
 665                if (!header_group[p->field]) {
 666                        header_group[p->field] = h;
 667                        continue;
 668                }
 669                header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 670        }
 671
 672        header_expr = NULL;
 673
 674        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 675                if (!header_group[fld])
 676                        continue;
 677                if (!header_expr)
 678                        header_expr = grep_true_expr();
 679                header_expr = grep_or_expr(header_group[fld], header_expr);
 680        }
 681        return header_expr;
 682}
 683
 684static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 685{
 686        struct grep_expr *z = x;
 687
 688        while (x) {
 689                assert(x->node == GREP_NODE_OR);
 690                if (x->u.binary.right &&
 691                    x->u.binary.right->node == GREP_NODE_TRUE) {
 692                        x->u.binary.right = y;
 693                        break;
 694                }
 695                x = x->u.binary.right;
 696        }
 697        return z;
 698}
 699
 700static void compile_grep_patterns_real(struct grep_opt *opt)
 701{
 702        struct grep_pat *p;
 703        struct grep_expr *header_expr = prep_header_patterns(opt);
 704
 705        for (p = opt->pattern_list; p; p = p->next) {
 706                switch (p->token) {
 707                case GREP_PATTERN: /* atom */
 708                case GREP_PATTERN_HEAD:
 709                case GREP_PATTERN_BODY:
 710                        compile_regexp(p, opt);
 711                        break;
 712                default:
 713                        opt->extended = 1;
 714                        break;
 715                }
 716        }
 717
 718        if (opt->all_match || header_expr)
 719                opt->extended = 1;
 720        else if (!opt->extended && !opt->debug)
 721                return;
 722
 723        p = opt->pattern_list;
 724        if (p)
 725                opt->pattern_expression = compile_pattern_expr(&p);
 726        if (p)
 727                die("incomplete pattern expression: %s", p->pattern);
 728
 729        if (!header_expr)
 730                return;
 731
 732        if (!opt->pattern_expression)
 733                opt->pattern_expression = header_expr;
 734        else if (opt->all_match)
 735                opt->pattern_expression = grep_splice_or(header_expr,
 736                                                         opt->pattern_expression);
 737        else
 738                opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 739                                                       header_expr);
 740        opt->all_match = 1;
 741}
 742
 743void compile_grep_patterns(struct grep_opt *opt)
 744{
 745        compile_grep_patterns_real(opt);
 746        if (opt->debug)
 747                dump_grep_expression(opt);
 748}
 749
 750static void free_pattern_expr(struct grep_expr *x)
 751{
 752        switch (x->node) {
 753        case GREP_NODE_TRUE:
 754        case GREP_NODE_ATOM:
 755                break;
 756        case GREP_NODE_NOT:
 757                free_pattern_expr(x->u.unary);
 758                break;
 759        case GREP_NODE_AND:
 760        case GREP_NODE_OR:
 761                free_pattern_expr(x->u.binary.left);
 762                free_pattern_expr(x->u.binary.right);
 763                break;
 764        }
 765        free(x);
 766}
 767
 768void free_grep_patterns(struct grep_opt *opt)
 769{
 770        struct grep_pat *p, *n;
 771
 772        for (p = opt->pattern_list; p; p = n) {
 773                n = p->next;
 774                switch (p->token) {
 775                case GREP_PATTERN: /* atom */
 776                case GREP_PATTERN_HEAD:
 777                case GREP_PATTERN_BODY:
 778                        if (p->kws)
 779                                kwsfree(p->kws);
 780                        else if (p->pcre_regexp)
 781                                free_pcre_regexp(p);
 782                        else
 783                                regfree(&p->regexp);
 784                        free(p->pattern);
 785                        break;
 786                default:
 787                        break;
 788                }
 789                free(p);
 790        }
 791
 792        if (!opt->extended)
 793                return;
 794        free_pattern_expr(opt->pattern_expression);
 795}
 796
 797static char *end_of_line(char *cp, unsigned long *left)
 798{
 799        unsigned long l = *left;
 800        while (l && *cp != '\n') {
 801                l--;
 802                cp++;
 803        }
 804        *left = l;
 805        return cp;
 806}
 807
 808static int word_char(char ch)
 809{
 810        return isalnum(ch) || ch == '_';
 811}
 812
 813static void output_color(struct grep_opt *opt, const void *data, size_t size,
 814                         const char *color)
 815{
 816        if (want_color(opt->color) && color && color[0]) {
 817                opt->output(opt, color, strlen(color));
 818                opt->output(opt, data, size);
 819                opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 820        } else
 821                opt->output(opt, data, size);
 822}
 823
 824static void output_sep(struct grep_opt *opt, char sign)
 825{
 826        if (opt->null_following_name)
 827                opt->output(opt, "\0", 1);
 828        else
 829                output_color(opt, &sign, 1, opt->color_sep);
 830}
 831
 832static void show_name(struct grep_opt *opt, const char *name)
 833{
 834        output_color(opt, name, strlen(name), opt->color_filename);
 835        opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 836}
 837
 838static int fixmatch(struct grep_pat *p, char *line, char *eol,
 839                    regmatch_t *match)
 840{
 841        struct kwsmatch kwsm;
 842        size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 843        if (offset == -1) {
 844                match->rm_so = match->rm_eo = -1;
 845                return REG_NOMATCH;
 846        } else {
 847                match->rm_so = offset;
 848                match->rm_eo = match->rm_so + kwsm.size[0];
 849                return 0;
 850        }
 851}
 852
 853static int regmatch(const regex_t *preg, char *line, char *eol,
 854                    regmatch_t *match, int eflags)
 855{
 856#ifdef REG_STARTEND
 857        match->rm_so = 0;
 858        match->rm_eo = eol - line;
 859        eflags |= REG_STARTEND;
 860#endif
 861        return regexec(preg, line, 1, match, eflags);
 862}
 863
 864static int patmatch(struct grep_pat *p, char *line, char *eol,
 865                    regmatch_t *match, int eflags)
 866{
 867        int hit;
 868
 869        if (p->fixed)
 870                hit = !fixmatch(p, line, eol, match);
 871        else if (p->pcre_regexp)
 872                hit = !pcrematch(p, line, eol, match, eflags);
 873        else
 874                hit = !regmatch(&p->regexp, line, eol, match, eflags);
 875
 876        return hit;
 877}
 878
 879static int strip_timestamp(char *bol, char **eol_p)
 880{
 881        char *eol = *eol_p;
 882        int ch;
 883
 884        while (bol < --eol) {
 885                if (*eol != '>')
 886                        continue;
 887                *eol_p = ++eol;
 888                ch = *eol;
 889                *eol = '\0';
 890                return ch;
 891        }
 892        return 0;
 893}
 894
 895static struct {
 896        const char *field;
 897        size_t len;
 898} header_field[] = {
 899        { "author ", 7 },
 900        { "committer ", 10 },
 901        { "reflog ", 7 },
 902};
 903
 904static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 905                             enum grep_context ctx,
 906                             regmatch_t *pmatch, int eflags)
 907{
 908        int hit = 0;
 909        int saved_ch = 0;
 910        const char *start = bol;
 911
 912        if ((p->token != GREP_PATTERN) &&
 913            ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 914                return 0;
 915
 916        if (p->token == GREP_PATTERN_HEAD) {
 917                const char *field;
 918                size_t len;
 919                assert(p->field < ARRAY_SIZE(header_field));
 920                field = header_field[p->field].field;
 921                len = header_field[p->field].len;
 922                if (strncmp(bol, field, len))
 923                        return 0;
 924                bol += len;
 925                switch (p->field) {
 926                case GREP_HEADER_AUTHOR:
 927                case GREP_HEADER_COMMITTER:
 928                        saved_ch = strip_timestamp(bol, &eol);
 929                        break;
 930                default:
 931                        break;
 932                }
 933        }
 934
 935 again:
 936        hit = patmatch(p, bol, eol, pmatch, eflags);
 937
 938        if (hit && p->word_regexp) {
 939                if ((pmatch[0].rm_so < 0) ||
 940                    (eol - bol) < pmatch[0].rm_so ||
 941                    (pmatch[0].rm_eo < 0) ||
 942                    (eol - bol) < pmatch[0].rm_eo)
 943                        die("regexp returned nonsense");
 944
 945                /* Match beginning must be either beginning of the
 946                 * line, or at word boundary (i.e. the last char must
 947                 * not be a word char).  Similarly, match end must be
 948                 * either end of the line, or at word boundary
 949                 * (i.e. the next char must not be a word char).
 950                 */
 951                if ( ((pmatch[0].rm_so == 0) ||
 952                      !word_char(bol[pmatch[0].rm_so-1])) &&
 953                     ((pmatch[0].rm_eo == (eol-bol)) ||
 954                      !word_char(bol[pmatch[0].rm_eo])) )
 955                        ;
 956                else
 957                        hit = 0;
 958
 959                /* Words consist of at least one character. */
 960                if (pmatch->rm_so == pmatch->rm_eo)
 961                        hit = 0;
 962
 963                if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 964                        /* There could be more than one match on the
 965                         * line, and the first match might not be
 966                         * strict word match.  But later ones could be!
 967                         * Forward to the next possible start, i.e. the
 968                         * next position following a non-word char.
 969                         */
 970                        bol = pmatch[0].rm_so + bol + 1;
 971                        while (word_char(bol[-1]) && bol < eol)
 972                                bol++;
 973                        eflags |= REG_NOTBOL;
 974                        if (bol < eol)
 975                                goto again;
 976                }
 977        }
 978        if (p->token == GREP_PATTERN_HEAD && saved_ch)
 979                *eol = saved_ch;
 980        if (hit) {
 981                pmatch[0].rm_so += bol - start;
 982                pmatch[0].rm_eo += bol - start;
 983        }
 984        return hit;
 985}
 986
 987static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 988                           enum grep_context ctx, int collect_hits)
 989{
 990        int h = 0;
 991        regmatch_t match;
 992
 993        if (!x)
 994                die("Not a valid grep expression");
 995        switch (x->node) {
 996        case GREP_NODE_TRUE:
 997                h = 1;
 998                break;
 999        case GREP_NODE_ATOM:
1000                h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1001                break;
1002        case GREP_NODE_NOT:
1003                h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1004                break;
1005        case GREP_NODE_AND:
1006                if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1007                        return 0;
1008                h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1009                break;
1010        case GREP_NODE_OR:
1011                if (!collect_hits)
1012                        return (match_expr_eval(x->u.binary.left,
1013                                                bol, eol, ctx, 0) ||
1014                                match_expr_eval(x->u.binary.right,
1015                                                bol, eol, ctx, 0));
1016                h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1017                x->u.binary.left->hit |= h;
1018                h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1019                break;
1020        default:
1021                die("Unexpected node type (internal error) %d", x->node);
1022        }
1023        if (collect_hits)
1024                x->hit |= h;
1025        return h;
1026}
1027
1028static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1029                      enum grep_context ctx, int collect_hits)
1030{
1031        struct grep_expr *x = opt->pattern_expression;
1032        return match_expr_eval(x, bol, eol, ctx, collect_hits);
1033}
1034
1035static int match_line(struct grep_opt *opt, char *bol, char *eol,
1036                      enum grep_context ctx, int collect_hits)
1037{
1038        struct grep_pat *p;
1039        regmatch_t match;
1040
1041        if (opt->extended)
1042                return match_expr(opt, bol, eol, ctx, collect_hits);
1043
1044        /* we do not call with collect_hits without being extended */
1045        for (p = opt->pattern_list; p; p = p->next) {
1046                if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1047                        return 1;
1048        }
1049        return 0;
1050}
1051
1052static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1053                              enum grep_context ctx,
1054                              regmatch_t *pmatch, int eflags)
1055{
1056        regmatch_t match;
1057
1058        if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1059                return 0;
1060        if (match.rm_so < 0 || match.rm_eo < 0)
1061                return 0;
1062        if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1063                if (match.rm_so > pmatch->rm_so)
1064                        return 1;
1065                if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1066                        return 1;
1067        }
1068        pmatch->rm_so = match.rm_so;
1069        pmatch->rm_eo = match.rm_eo;
1070        return 1;
1071}
1072
1073static int next_match(struct grep_opt *opt, char *bol, char *eol,
1074                      enum grep_context ctx, regmatch_t *pmatch, int eflags)
1075{
1076        struct grep_pat *p;
1077        int hit = 0;
1078
1079        pmatch->rm_so = pmatch->rm_eo = -1;
1080        if (bol < eol) {
1081                for (p = opt->pattern_list; p; p = p->next) {
1082                        switch (p->token) {
1083                        case GREP_PATTERN: /* atom */
1084                        case GREP_PATTERN_HEAD:
1085                        case GREP_PATTERN_BODY:
1086                                hit |= match_next_pattern(p, bol, eol, ctx,
1087                                                          pmatch, eflags);
1088                                break;
1089                        default:
1090                                break;
1091                        }
1092                }
1093        }
1094        return hit;
1095}
1096
1097static void show_line(struct grep_opt *opt, char *bol, char *eol,
1098                      const char *name, unsigned lno, char sign)
1099{
1100        int rest = eol - bol;
1101        const char *match_color, *line_color = NULL;
1102
1103        if (opt->file_break && opt->last_shown == 0) {
1104                if (opt->show_hunk_mark)
1105                        opt->output(opt, "\n", 1);
1106        } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1107                if (opt->last_shown == 0) {
1108                        if (opt->show_hunk_mark) {
1109                                output_color(opt, "--", 2, opt->color_sep);
1110                                opt->output(opt, "\n", 1);
1111                        }
1112                } else if (lno > opt->last_shown + 1) {
1113                        output_color(opt, "--", 2, opt->color_sep);
1114                        opt->output(opt, "\n", 1);
1115                }
1116        }
1117        if (opt->heading && opt->last_shown == 0) {
1118                output_color(opt, name, strlen(name), opt->color_filename);
1119                opt->output(opt, "\n", 1);
1120        }
1121        opt->last_shown = lno;
1122
1123        if (!opt->heading && opt->pathname) {
1124                output_color(opt, name, strlen(name), opt->color_filename);
1125                output_sep(opt, sign);
1126        }
1127        if (opt->linenum) {
1128                char buf[32];
1129                snprintf(buf, sizeof(buf), "%d", lno);
1130                output_color(opt, buf, strlen(buf), opt->color_lineno);
1131                output_sep(opt, sign);
1132        }
1133        if (opt->color) {
1134                regmatch_t match;
1135                enum grep_context ctx = GREP_CONTEXT_BODY;
1136                int ch = *eol;
1137                int eflags = 0;
1138
1139                if (sign == ':')
1140                        match_color = opt->color_match_selected;
1141                else
1142                        match_color = opt->color_match_context;
1143                if (sign == ':')
1144                        line_color = opt->color_selected;
1145                else if (sign == '-')
1146                        line_color = opt->color_context;
1147                else if (sign == '=')
1148                        line_color = opt->color_function;
1149                *eol = '\0';
1150                while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1151                        if (match.rm_so == match.rm_eo)
1152                                break;
1153
1154                        output_color(opt, bol, match.rm_so, line_color);
1155                        output_color(opt, bol + match.rm_so,
1156                                     match.rm_eo - match.rm_so, match_color);
1157                        bol += match.rm_eo;
1158                        rest -= match.rm_eo;
1159                        eflags = REG_NOTBOL;
1160                }
1161                *eol = ch;
1162        }
1163        output_color(opt, bol, rest, line_color);
1164        opt->output(opt, "\n", 1);
1165}
1166
1167#ifndef NO_PTHREADS
1168int grep_use_locks;
1169
1170/*
1171 * This lock protects access to the gitattributes machinery, which is
1172 * not thread-safe.
1173 */
1174pthread_mutex_t grep_attr_mutex;
1175
1176static inline void grep_attr_lock(void)
1177{
1178        if (grep_use_locks)
1179                pthread_mutex_lock(&grep_attr_mutex);
1180}
1181
1182static inline void grep_attr_unlock(void)
1183{
1184        if (grep_use_locks)
1185                pthread_mutex_unlock(&grep_attr_mutex);
1186}
1187
1188/*
1189 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1190 */
1191pthread_mutex_t grep_read_mutex;
1192
1193#else
1194#define grep_attr_lock()
1195#define grep_attr_unlock()
1196#endif
1197
1198static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1199{
1200        xdemitconf_t *xecfg = opt->priv;
1201        if (xecfg && !xecfg->find_func) {
1202                grep_source_load_driver(gs);
1203                if (gs->driver->funcname.pattern) {
1204                        const struct userdiff_funcname *pe = &gs->driver->funcname;
1205                        xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1206                } else {
1207                        xecfg = opt->priv = NULL;
1208                }
1209        }
1210
1211        if (xecfg) {
1212                char buf[1];
1213                return xecfg->find_func(bol, eol - bol, buf, 1,
1214                                        xecfg->find_func_priv) >= 0;
1215        }
1216
1217        if (bol == eol)
1218                return 0;
1219        if (isalpha(*bol) || *bol == '_' || *bol == '$')
1220                return 1;
1221        return 0;
1222}
1223
1224static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1225                               char *bol, unsigned lno)
1226{
1227        while (bol > gs->buf) {
1228                char *eol = --bol;
1229
1230                while (bol > gs->buf && bol[-1] != '\n')
1231                        bol--;
1232                lno--;
1233
1234                if (lno <= opt->last_shown)
1235                        break;
1236
1237                if (match_funcname(opt, gs, bol, eol)) {
1238                        show_line(opt, bol, eol, gs->name, lno, '=');
1239                        break;
1240                }
1241        }
1242}
1243
1244static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1245                             char *bol, char *end, unsigned lno)
1246{
1247        unsigned cur = lno, from = 1, funcname_lno = 0;
1248        int funcname_needed = !!opt->funcname;
1249
1250        if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1251                funcname_needed = 2;
1252
1253        if (opt->pre_context < lno)
1254                from = lno - opt->pre_context;
1255        if (from <= opt->last_shown)
1256                from = opt->last_shown + 1;
1257
1258        /* Rewind. */
1259        while (bol > gs->buf &&
1260               cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1261                char *eol = --bol;
1262
1263                while (bol > gs->buf && bol[-1] != '\n')
1264                        bol--;
1265                cur--;
1266                if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1267                        funcname_lno = cur;
1268                        funcname_needed = 0;
1269                }
1270        }
1271
1272        /* We need to look even further back to find a function signature. */
1273        if (opt->funcname && funcname_needed)
1274                show_funcname_line(opt, gs, bol, cur);
1275
1276        /* Back forward. */
1277        while (cur < lno) {
1278                char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1279
1280                while (*eol != '\n')
1281                        eol++;
1282                show_line(opt, bol, eol, gs->name, cur, sign);
1283                bol = eol + 1;
1284                cur++;
1285        }
1286}
1287
1288static int should_lookahead(struct grep_opt *opt)
1289{
1290        struct grep_pat *p;
1291
1292        if (opt->extended)
1293                return 0; /* punt for too complex stuff */
1294        if (opt->invert)
1295                return 0;
1296        for (p = opt->pattern_list; p; p = p->next) {
1297                if (p->token != GREP_PATTERN)
1298                        return 0; /* punt for "header only" and stuff */
1299        }
1300        return 1;
1301}
1302
1303static int look_ahead(struct grep_opt *opt,
1304                      unsigned long *left_p,
1305                      unsigned *lno_p,
1306                      char **bol_p)
1307{
1308        unsigned lno = *lno_p;
1309        char *bol = *bol_p;
1310        struct grep_pat *p;
1311        char *sp, *last_bol;
1312        regoff_t earliest = -1;
1313
1314        for (p = opt->pattern_list; p; p = p->next) {
1315                int hit;
1316                regmatch_t m;
1317
1318                hit = patmatch(p, bol, bol + *left_p, &m, 0);
1319                if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1320                        continue;
1321                if (earliest < 0 || m.rm_so < earliest)
1322                        earliest = m.rm_so;
1323        }
1324
1325        if (earliest < 0) {
1326                *bol_p = bol + *left_p;
1327                *left_p = 0;
1328                return 1;
1329        }
1330        for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1331                ; /* find the beginning of the line */
1332        last_bol = sp;
1333
1334        for (sp = bol; sp < last_bol; sp++) {
1335                if (*sp == '\n')
1336                        lno++;
1337        }
1338        *left_p -= last_bol - bol;
1339        *bol_p = last_bol;
1340        *lno_p = lno;
1341        return 0;
1342}
1343
1344static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1345{
1346        fwrite(buf, size, 1, stdout);
1347}
1348
1349static int fill_textconv_grep(struct userdiff_driver *driver,
1350                              struct grep_source *gs)
1351{
1352        struct diff_filespec *df;
1353        char *buf;
1354        size_t size;
1355
1356        if (!driver || !driver->textconv)
1357                return grep_source_load(gs);
1358
1359        /*
1360         * The textconv interface is intimately tied to diff_filespecs, so we
1361         * have to pretend to be one. If we could unify the grep_source
1362         * and diff_filespec structs, this mess could just go away.
1363         */
1364        df = alloc_filespec(gs->path);
1365        switch (gs->type) {
1366        case GREP_SOURCE_SHA1:
1367                fill_filespec(df, gs->identifier, 1, 0100644);
1368                break;
1369        case GREP_SOURCE_FILE:
1370                fill_filespec(df, null_sha1, 0, 0100644);
1371                break;
1372        default:
1373                die("BUG: attempt to textconv something without a path?");
1374        }
1375
1376        /*
1377         * fill_textconv is not remotely thread-safe; it may load objects
1378         * behind the scenes, and it modifies the global diff tempfile
1379         * structure.
1380         */
1381        grep_read_lock();
1382        size = fill_textconv(driver, df, &buf);
1383        grep_read_unlock();
1384        free_filespec(df);
1385
1386        /*
1387         * The normal fill_textconv usage by the diff machinery would just keep
1388         * the textconv'd buf separate from the diff_filespec. But much of the
1389         * grep code passes around a grep_source and assumes that its "buf"
1390         * pointer is the beginning of the thing we are searching. So let's
1391         * install our textconv'd version into the grep_source, taking care not
1392         * to leak any existing buffer.
1393         */
1394        grep_source_clear_data(gs);
1395        gs->buf = buf;
1396        gs->size = size;
1397
1398        return 0;
1399}
1400
1401static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1402{
1403        char *bol;
1404        unsigned long left;
1405        unsigned lno = 1;
1406        unsigned last_hit = 0;
1407        int binary_match_only = 0;
1408        unsigned count = 0;
1409        int try_lookahead = 0;
1410        int show_function = 0;
1411        struct userdiff_driver *textconv = NULL;
1412        enum grep_context ctx = GREP_CONTEXT_HEAD;
1413        xdemitconf_t xecfg;
1414
1415        if (!opt->output)
1416                opt->output = std_output;
1417
1418        if (opt->pre_context || opt->post_context || opt->file_break ||
1419            opt->funcbody) {
1420                /* Show hunk marks, except for the first file. */
1421                if (opt->last_shown)
1422                        opt->show_hunk_mark = 1;
1423                /*
1424                 * If we're using threads then we can't easily identify
1425                 * the first file.  Always put hunk marks in that case
1426                 * and skip the very first one later in work_done().
1427                 */
1428                if (opt->output != std_output)
1429                        opt->show_hunk_mark = 1;
1430        }
1431        opt->last_shown = 0;
1432
1433        if (opt->allow_textconv) {
1434                grep_source_load_driver(gs);
1435                /*
1436                 * We might set up the shared textconv cache data here, which
1437                 * is not thread-safe.
1438                 */
1439                grep_attr_lock();
1440                textconv = userdiff_get_textconv(gs->driver);
1441                grep_attr_unlock();
1442        }
1443
1444        /*
1445         * We know the result of a textconv is text, so we only have to care
1446         * about binary handling if we are not using it.
1447         */
1448        if (!textconv) {
1449                switch (opt->binary) {
1450                case GREP_BINARY_DEFAULT:
1451                        if (grep_source_is_binary(gs))
1452                                binary_match_only = 1;
1453                        break;
1454                case GREP_BINARY_NOMATCH:
1455                        if (grep_source_is_binary(gs))
1456                                return 0; /* Assume unmatch */
1457                        break;
1458                case GREP_BINARY_TEXT:
1459                        break;
1460                default:
1461                        die("bug: unknown binary handling mode");
1462                }
1463        }
1464
1465        memset(&xecfg, 0, sizeof(xecfg));
1466        opt->priv = &xecfg;
1467
1468        try_lookahead = should_lookahead(opt);
1469
1470        if (fill_textconv_grep(textconv, gs) < 0)
1471                return 0;
1472
1473        bol = gs->buf;
1474        left = gs->size;
1475        while (left) {
1476                char *eol, ch;
1477                int hit;
1478
1479                /*
1480                 * look_ahead() skips quickly to the line that possibly
1481                 * has the next hit; don't call it if we need to do
1482                 * something more than just skipping the current line
1483                 * in response to an unmatch for the current line.  E.g.
1484                 * inside a post-context window, we will show the current
1485                 * line as a context around the previous hit when it
1486                 * doesn't hit.
1487                 */
1488                if (try_lookahead
1489                    && !(last_hit
1490                         && (show_function ||
1491                             lno <= last_hit + opt->post_context))
1492                    && look_ahead(opt, &left, &lno, &bol))
1493                        break;
1494                eol = end_of_line(bol, &left);
1495                ch = *eol;
1496                *eol = 0;
1497
1498                if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1499                        ctx = GREP_CONTEXT_BODY;
1500
1501                hit = match_line(opt, bol, eol, ctx, collect_hits);
1502                *eol = ch;
1503
1504                if (collect_hits)
1505                        goto next_line;
1506
1507                /* "grep -v -e foo -e bla" should list lines
1508                 * that do not have either, so inversion should
1509                 * be done outside.
1510                 */
1511                if (opt->invert)
1512                        hit = !hit;
1513                if (opt->unmatch_name_only) {
1514                        if (hit)
1515                                return 0;
1516                        goto next_line;
1517                }
1518                if (hit) {
1519                        count++;
1520                        if (opt->status_only)
1521                                return 1;
1522                        if (opt->name_only) {
1523                                show_name(opt, gs->name);
1524                                return 1;
1525                        }
1526                        if (opt->count)
1527                                goto next_line;
1528                        if (binary_match_only) {
1529                                opt->output(opt, "Binary file ", 12);
1530                                output_color(opt, gs->name, strlen(gs->name),
1531                                             opt->color_filename);
1532                                opt->output(opt, " matches\n", 9);
1533                                return 1;
1534                        }
1535                        /* Hit at this line.  If we haven't shown the
1536                         * pre-context lines, we would need to show them.
1537                         */
1538                        if (opt->pre_context || opt->funcbody)
1539                                show_pre_context(opt, gs, bol, eol, lno);
1540                        else if (opt->funcname)
1541                                show_funcname_line(opt, gs, bol, lno);
1542                        show_line(opt, bol, eol, gs->name, lno, ':');
1543                        last_hit = lno;
1544                        if (opt->funcbody)
1545                                show_function = 1;
1546                        goto next_line;
1547                }
1548                if (show_function && match_funcname(opt, gs, bol, eol))
1549                        show_function = 0;
1550                if (show_function ||
1551                    (last_hit && lno <= last_hit + opt->post_context)) {
1552                        /* If the last hit is within the post context,
1553                         * we need to show this line.
1554                         */
1555                        show_line(opt, bol, eol, gs->name, lno, '-');
1556                }
1557
1558        next_line:
1559                bol = eol + 1;
1560                if (!left)
1561                        break;
1562                left--;
1563                lno++;
1564        }
1565
1566        if (collect_hits)
1567                return 0;
1568
1569        if (opt->status_only)
1570                return 0;
1571        if (opt->unmatch_name_only) {
1572                /* We did not see any hit, so we want to show this */
1573                show_name(opt, gs->name);
1574                return 1;
1575        }
1576
1577        xdiff_clear_find_func(&xecfg);
1578        opt->priv = NULL;
1579
1580        /* NEEDSWORK:
1581         * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1582         * which feels mostly useless but sometimes useful.  Maybe
1583         * make it another option?  For now suppress them.
1584         */
1585        if (opt->count && count) {
1586                char buf[32];
1587                if (opt->pathname) {
1588                        output_color(opt, gs->name, strlen(gs->name),
1589                                     opt->color_filename);
1590                        output_sep(opt, ':');
1591                }
1592                snprintf(buf, sizeof(buf), "%u\n", count);
1593                opt->output(opt, buf, strlen(buf));
1594                return 1;
1595        }
1596        return !!last_hit;
1597}
1598
1599static void clr_hit_marker(struct grep_expr *x)
1600{
1601        /* All-hit markers are meaningful only at the very top level
1602         * OR node.
1603         */
1604        while (1) {
1605                x->hit = 0;
1606                if (x->node != GREP_NODE_OR)
1607                        return;
1608                x->u.binary.left->hit = 0;
1609                x = x->u.binary.right;
1610        }
1611}
1612
1613static int chk_hit_marker(struct grep_expr *x)
1614{
1615        /* Top level nodes have hit markers.  See if they all are hits */
1616        while (1) {
1617                if (x->node != GREP_NODE_OR)
1618                        return x->hit;
1619                if (!x->u.binary.left->hit)
1620                        return 0;
1621                x = x->u.binary.right;
1622        }
1623}
1624
1625int grep_source(struct grep_opt *opt, struct grep_source *gs)
1626{
1627        /*
1628         * we do not have to do the two-pass grep when we do not check
1629         * buffer-wide "all-match".
1630         */
1631        if (!opt->all_match)
1632                return grep_source_1(opt, gs, 0);
1633
1634        /* Otherwise the toplevel "or" terms hit a bit differently.
1635         * We first clear hit markers from them.
1636         */
1637        clr_hit_marker(opt->pattern_expression);
1638        grep_source_1(opt, gs, 1);
1639
1640        if (!chk_hit_marker(opt->pattern_expression))
1641                return 0;
1642
1643        return grep_source_1(opt, gs, 0);
1644}
1645
1646int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1647{
1648        struct grep_source gs;
1649        int r;
1650
1651        grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1652        gs.buf = buf;
1653        gs.size = size;
1654
1655        r = grep_source(opt, &gs);
1656
1657        grep_source_clear(&gs);
1658        return r;
1659}
1660
1661void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1662                      const char *name, const char *path,
1663                      const void *identifier)
1664{
1665        gs->type = type;
1666        gs->name = xstrdup_or_null(name);
1667        gs->path = xstrdup_or_null(path);
1668        gs->buf = NULL;
1669        gs->size = 0;
1670        gs->driver = NULL;
1671
1672        switch (type) {
1673        case GREP_SOURCE_FILE:
1674                gs->identifier = xstrdup(identifier);
1675                break;
1676        case GREP_SOURCE_SHA1:
1677                gs->identifier = xmalloc(20);
1678                hashcpy(gs->identifier, identifier);
1679                break;
1680        case GREP_SOURCE_BUF:
1681                gs->identifier = NULL;
1682        }
1683}
1684
1685void grep_source_clear(struct grep_source *gs)
1686{
1687        free(gs->name);
1688        gs->name = NULL;
1689        free(gs->path);
1690        gs->path = NULL;
1691        free(gs->identifier);
1692        gs->identifier = NULL;
1693        grep_source_clear_data(gs);
1694}
1695
1696void grep_source_clear_data(struct grep_source *gs)
1697{
1698        switch (gs->type) {
1699        case GREP_SOURCE_FILE:
1700        case GREP_SOURCE_SHA1:
1701                free(gs->buf);
1702                gs->buf = NULL;
1703                gs->size = 0;
1704                break;
1705        case GREP_SOURCE_BUF:
1706                /* leave user-provided buf intact */
1707                break;
1708        }
1709}
1710
1711static int grep_source_load_sha1(struct grep_source *gs)
1712{
1713        enum object_type type;
1714
1715        grep_read_lock();
1716        gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1717        grep_read_unlock();
1718
1719        if (!gs->buf)
1720                return error(_("'%s': unable to read %s"),
1721                             gs->name,
1722                             sha1_to_hex(gs->identifier));
1723        return 0;
1724}
1725
1726static int grep_source_load_file(struct grep_source *gs)
1727{
1728        const char *filename = gs->identifier;
1729        struct stat st;
1730        char *data;
1731        size_t size;
1732        int i;
1733
1734        if (lstat(filename, &st) < 0) {
1735        err_ret:
1736                if (errno != ENOENT)
1737                        error(_("'%s': %s"), filename, strerror(errno));
1738                return -1;
1739        }
1740        if (!S_ISREG(st.st_mode))
1741                return -1;
1742        size = xsize_t(st.st_size);
1743        i = open(filename, O_RDONLY);
1744        if (i < 0)
1745                goto err_ret;
1746        data = xmalloc(size + 1);
1747        if (st.st_size != read_in_full(i, data, size)) {
1748                error(_("'%s': short read %s"), filename, strerror(errno));
1749                close(i);
1750                free(data);
1751                return -1;
1752        }
1753        close(i);
1754        data[size] = 0;
1755
1756        gs->buf = data;
1757        gs->size = size;
1758        return 0;
1759}
1760
1761static int grep_source_load(struct grep_source *gs)
1762{
1763        if (gs->buf)
1764                return 0;
1765
1766        switch (gs->type) {
1767        case GREP_SOURCE_FILE:
1768                return grep_source_load_file(gs);
1769        case GREP_SOURCE_SHA1:
1770                return grep_source_load_sha1(gs);
1771        case GREP_SOURCE_BUF:
1772                return gs->buf ? 0 : -1;
1773        }
1774        die("BUG: invalid grep_source type");
1775}
1776
1777void grep_source_load_driver(struct grep_source *gs)
1778{
1779        if (gs->driver)
1780                return;
1781
1782        grep_attr_lock();
1783        if (gs->path)
1784                gs->driver = userdiff_find_by_path(gs->path);
1785        if (!gs->driver)
1786                gs->driver = userdiff_find_by_name("default");
1787        grep_attr_unlock();
1788}
1789
1790static int grep_source_is_binary(struct grep_source *gs)
1791{
1792        grep_source_load_driver(gs);
1793        if (gs->driver->binary != -1)
1794                return gs->driver->binary;
1795
1796        if (!grep_source_load(gs))
1797                return buffer_is_binary(gs->buf, gs->size);
1798
1799        return 0;
1800}