Andrew's git / gitweb.git / grep.c
?
grep.con commit diffcore-pickaxe: support case insensitive match on non-ascii (b51a9c1)
   1#include "cache.h"
   2#include "grep.h"
   3#include "userdiff.h"
   4#include "xdiff-interface.h"
   5#include "diff.h"
   6#include "diffcore.h"
   7#include "commit.h"
   8#include "quote.h"
   9
  10static int grep_source_load(struct grep_source *gs);
  11static int grep_source_is_binary(struct grep_source *gs);
  12
  13static struct grep_opt grep_defaults;
  14
  15/*
  16 * Initialize the grep_defaults template with hardcoded defaults.
  17 * We could let the compiler do this, but without C99 initializers
  18 * the code gets unwieldy and unreadable, so...
  19 */
  20void init_grep_defaults(void)
  21{
  22        struct grep_opt *opt = &grep_defaults;
  23        static int run_once;
  24
  25        if (run_once)
  26                return;
  27        run_once++;
  28
  29        memset(opt, 0, sizeof(*opt));
  30        opt->relative = 1;
  31        opt->pathname = 1;
  32        opt->regflags = REG_NEWLINE;
  33        opt->max_depth = -1;
  34        opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  35        opt->extended_regexp_option = 0;
  36        color_set(opt->color_context, "");
  37        color_set(opt->color_filename, "");
  38        color_set(opt->color_function, "");
  39        color_set(opt->color_lineno, "");
  40        color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
  41        color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
  42        color_set(opt->color_selected, "");
  43        color_set(opt->color_sep, GIT_COLOR_CYAN);
  44        opt->color = -1;
  45}
  46
  47static int parse_pattern_type_arg(const char *opt, const char *arg)
  48{
  49        if (!strcmp(arg, "default"))
  50                return GREP_PATTERN_TYPE_UNSPECIFIED;
  51        else if (!strcmp(arg, "basic"))
  52                return GREP_PATTERN_TYPE_BRE;
  53        else if (!strcmp(arg, "extended"))
  54                return GREP_PATTERN_TYPE_ERE;
  55        else if (!strcmp(arg, "fixed"))
  56                return GREP_PATTERN_TYPE_FIXED;
  57        else if (!strcmp(arg, "perl"))
  58                return GREP_PATTERN_TYPE_PCRE;
  59        die("bad %s argument: %s", opt, arg);
  60}
  61
  62/*
  63 * Read the configuration file once and store it in
  64 * the grep_defaults template.
  65 */
  66int grep_config(const char *var, const char *value, void *cb)
  67{
  68        struct grep_opt *opt = &grep_defaults;
  69        char *color = NULL;
  70
  71        if (userdiff_config(var, value) < 0)
  72                return -1;
  73
  74        if (!strcmp(var, "grep.extendedregexp")) {
  75                if (git_config_bool(var, value))
  76                        opt->extended_regexp_option = 1;
  77                else
  78                        opt->extended_regexp_option = 0;
  79                return 0;
  80        }
  81
  82        if (!strcmp(var, "grep.patterntype")) {
  83                opt->pattern_type_option = parse_pattern_type_arg(var, value);
  84                return 0;
  85        }
  86
  87        if (!strcmp(var, "grep.linenumber")) {
  88                opt->linenum = git_config_bool(var, value);
  89                return 0;
  90        }
  91
  92        if (!strcmp(var, "grep.fullname")) {
  93                opt->relative = !git_config_bool(var, value);
  94                return 0;
  95        }
  96
  97        if (!strcmp(var, "color.grep"))
  98                opt->color = git_config_colorbool(var, value);
  99        else if (!strcmp(var, "color.grep.context"))
 100                color = opt->color_context;
 101        else if (!strcmp(var, "color.grep.filename"))
 102                color = opt->color_filename;
 103        else if (!strcmp(var, "color.grep.function"))
 104                color = opt->color_function;
 105        else if (!strcmp(var, "color.grep.linenumber"))
 106                color = opt->color_lineno;
 107        else if (!strcmp(var, "color.grep.matchcontext"))
 108                color = opt->color_match_context;
 109        else if (!strcmp(var, "color.grep.matchselected"))
 110                color = opt->color_match_selected;
 111        else if (!strcmp(var, "color.grep.selected"))
 112                color = opt->color_selected;
 113        else if (!strcmp(var, "color.grep.separator"))
 114                color = opt->color_sep;
 115        else if (!strcmp(var, "color.grep.match")) {
 116                int rc = 0;
 117                if (!value)
 118                        return config_error_nonbool(var);
 119                rc |= color_parse(value, opt->color_match_context);
 120                rc |= color_parse(value, opt->color_match_selected);
 121                return rc;
 122        }
 123
 124        if (color) {
 125                if (!value)
 126                        return config_error_nonbool(var);
 127                return color_parse(value, color);
 128        }
 129        return 0;
 130}
 131
 132/*
 133 * Initialize one instance of grep_opt and copy the
 134 * default values from the template we read the configuration
 135 * information in an earlier call to git_config(grep_config).
 136 */
 137void grep_init(struct grep_opt *opt, const char *prefix)
 138{
 139        struct grep_opt *def = &grep_defaults;
 140
 141        memset(opt, 0, sizeof(*opt));
 142        opt->prefix = prefix;
 143        opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 144        opt->pattern_tail = &opt->pattern_list;
 145        opt->header_tail = &opt->header_list;
 146
 147        opt->color = def->color;
 148        opt->extended_regexp_option = def->extended_regexp_option;
 149        opt->pattern_type_option = def->pattern_type_option;
 150        opt->linenum = def->linenum;
 151        opt->max_depth = def->max_depth;
 152        opt->pathname = def->pathname;
 153        opt->regflags = def->regflags;
 154        opt->relative = def->relative;
 155
 156        color_set(opt->color_context, def->color_context);
 157        color_set(opt->color_filename, def->color_filename);
 158        color_set(opt->color_function, def->color_function);
 159        color_set(opt->color_lineno, def->color_lineno);
 160        color_set(opt->color_match_context, def->color_match_context);
 161        color_set(opt->color_match_selected, def->color_match_selected);
 162        color_set(opt->color_selected, def->color_selected);
 163        color_set(opt->color_sep, def->color_sep);
 164}
 165
 166void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 167{
 168        if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 169                grep_set_pattern_type_option(pattern_type, opt);
 170        else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 171                grep_set_pattern_type_option(opt->pattern_type_option, opt);
 172        else if (opt->extended_regexp_option)
 173                grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 174}
 175
 176void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 177{
 178        switch (pattern_type) {
 179        case GREP_PATTERN_TYPE_UNSPECIFIED:
 180                /* fall through */
 181
 182        case GREP_PATTERN_TYPE_BRE:
 183                opt->fixed = 0;
 184                opt->pcre = 0;
 185                opt->regflags &= ~REG_EXTENDED;
 186                break;
 187
 188        case GREP_PATTERN_TYPE_ERE:
 189                opt->fixed = 0;
 190                opt->pcre = 0;
 191                opt->regflags |= REG_EXTENDED;
 192                break;
 193
 194        case GREP_PATTERN_TYPE_FIXED:
 195                opt->fixed = 1;
 196                opt->pcre = 0;
 197                opt->regflags &= ~REG_EXTENDED;
 198                break;
 199
 200        case GREP_PATTERN_TYPE_PCRE:
 201                opt->fixed = 0;
 202                opt->pcre = 1;
 203                opt->regflags &= ~REG_EXTENDED;
 204                break;
 205        }
 206}
 207
 208static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 209                                        const char *origin, int no,
 210                                        enum grep_pat_token t,
 211                                        enum grep_header_field field)
 212{
 213        struct grep_pat *p = xcalloc(1, sizeof(*p));
 214        p->pattern = xmemdupz(pat, patlen);
 215        p->patternlen = patlen;
 216        p->origin = origin;
 217        p->no = no;
 218        p->token = t;
 219        p->field = field;
 220        return p;
 221}
 222
 223static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 224{
 225        **tail = p;
 226        *tail = &p->next;
 227        p->next = NULL;
 228
 229        switch (p->token) {
 230        case GREP_PATTERN: /* atom */
 231        case GREP_PATTERN_HEAD:
 232        case GREP_PATTERN_BODY:
 233                for (;;) {
 234                        struct grep_pat *new_pat;
 235                        size_t len = 0;
 236                        char *cp = p->pattern + p->patternlen, *nl = NULL;
 237                        while (++len <= p->patternlen) {
 238                                if (*(--cp) == '\n') {
 239                                        nl = cp;
 240                                        break;
 241                                }
 242                        }
 243                        if (!nl)
 244                                break;
 245                        new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 246                                                  p->no, p->token, p->field);
 247                        new_pat->next = p->next;
 248                        if (!p->next)
 249                                *tail = &new_pat->next;
 250                        p->next = new_pat;
 251                        *nl = '\0';
 252                        p->patternlen -= len;
 253                }
 254                break;
 255        default:
 256                break;
 257        }
 258}
 259
 260void append_header_grep_pattern(struct grep_opt *opt,
 261                                enum grep_header_field field, const char *pat)
 262{
 263        struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 264                                             GREP_PATTERN_HEAD, field);
 265        if (field == GREP_HEADER_REFLOG)
 266                opt->use_reflog_filter = 1;
 267        do_append_grep_pat(&opt->header_tail, p);
 268}
 269
 270void append_grep_pattern(struct grep_opt *opt, const char *pat,
 271                         const char *origin, int no, enum grep_pat_token t)
 272{
 273        append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 274}
 275
 276void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 277                     const char *origin, int no, enum grep_pat_token t)
 278{
 279        struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 280        do_append_grep_pat(&opt->pattern_tail, p);
 281}
 282
 283struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 284{
 285        struct grep_pat *pat;
 286        struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 287        *ret = *opt;
 288
 289        ret->pattern_list = NULL;
 290        ret->pattern_tail = &ret->pattern_list;
 291
 292        for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 293        {
 294                if(pat->token == GREP_PATTERN_HEAD)
 295                        append_header_grep_pattern(ret, pat->field,
 296                                                   pat->pattern);
 297                else
 298                        append_grep_pat(ret, pat->pattern, pat->patternlen,
 299                                        pat->origin, pat->no, pat->token);
 300        }
 301
 302        return ret;
 303}
 304
 305static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 306                const char *error)
 307{
 308        char where[1024];
 309
 310        if (p->no)
 311                xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 312        else if (p->origin)
 313                xsnprintf(where, sizeof(where), "%s, ", p->origin);
 314        else
 315                where[0] = 0;
 316
 317        die("%s'%s': %s", where, p->pattern, error);
 318}
 319
 320#ifdef USE_LIBPCRE
 321static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 322{
 323        const char *error;
 324        int erroffset;
 325        int options = PCRE_MULTILINE;
 326
 327        if (opt->ignore_case) {
 328                if (has_non_ascii(p->pattern))
 329                        p->pcre_tables = pcre_maketables();
 330                options |= PCRE_CASELESS;
 331        }
 332        if (is_utf8_locale() && has_non_ascii(p->pattern))
 333                options |= PCRE_UTF8;
 334
 335        p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 336                                      p->pcre_tables);
 337        if (!p->pcre_regexp)
 338                compile_regexp_failed(p, error);
 339
 340        p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 341        if (!p->pcre_extra_info && error)
 342                die("%s", error);
 343}
 344
 345static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 346                regmatch_t *match, int eflags)
 347{
 348        int ovector[30], ret, flags = 0;
 349
 350        if (eflags & REG_NOTBOL)
 351                flags |= PCRE_NOTBOL;
 352
 353        ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 354                        0, flags, ovector, ARRAY_SIZE(ovector));
 355        if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 356                die("pcre_exec failed with error code %d", ret);
 357        if (ret > 0) {
 358                ret = 0;
 359                match->rm_so = ovector[0];
 360                match->rm_eo = ovector[1];
 361        }
 362
 363        return ret;
 364}
 365
 366static void free_pcre_regexp(struct grep_pat *p)
 367{
 368        pcre_free(p->pcre_regexp);
 369        pcre_free(p->pcre_extra_info);
 370        pcre_free((void *)p->pcre_tables);
 371}
 372#else /* !USE_LIBPCRE */
 373static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 374{
 375        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 376}
 377
 378static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 379                regmatch_t *match, int eflags)
 380{
 381        return 1;
 382}
 383
 384static void free_pcre_regexp(struct grep_pat *p)
 385{
 386}
 387#endif /* !USE_LIBPCRE */
 388
 389static int is_fixed(const char *s, size_t len)
 390{
 391        size_t i;
 392
 393        /* regcomp cannot accept patterns with NULs so we
 394         * consider any pattern containing a NUL fixed.
 395         */
 396        if (memchr(s, 0, len))
 397                return 1;
 398
 399        for (i = 0; i < len; i++) {
 400                if (is_regex_special(s[i]))
 401                        return 0;
 402        }
 403
 404        return 1;
 405}
 406
 407static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 408{
 409        struct strbuf sb = STRBUF_INIT;
 410        int err;
 411        int regflags;
 412
 413        basic_regex_quote_buf(&sb, p->pattern);
 414        regflags = opt->regflags & ~REG_EXTENDED;
 415        if (opt->ignore_case)
 416                regflags |= REG_ICASE;
 417        err = regcomp(&p->regexp, sb.buf, regflags);
 418        if (opt->debug)
 419                fprintf(stderr, "fixed %s\n", sb.buf);
 420        strbuf_release(&sb);
 421        if (err) {
 422                char errbuf[1024];
 423                regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 424                regfree(&p->regexp);
 425                compile_regexp_failed(p, errbuf);
 426        }
 427}
 428
 429static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 430{
 431        int icase, ascii_only;
 432        int err;
 433
 434        p->word_regexp = opt->word_regexp;
 435        p->ignore_case = opt->ignore_case;
 436        icase          = opt->regflags & REG_ICASE || p->ignore_case;
 437        ascii_only     = !has_non_ascii(p->pattern);
 438
 439        /*
 440         * Even when -F (fixed) asks us to do a non-regexp search, we
 441         * may not be able to correctly case-fold when -i
 442         * (ignore-case) is asked (in which case, we'll synthesize a
 443         * regexp to match the pattern that matches regexp special
 444         * characters literally, while ignoring case differences).  On
 445         * the other hand, even without -F, if the pattern does not
 446         * have any regexp special characters and there is no need for
 447         * case-folding search, we can internally turn it into a
 448         * simple string match using kws.  p->fixed tells us if we
 449         * want to use kws.
 450         */
 451        if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 452                p->fixed = !icase || ascii_only;
 453        else
 454                p->fixed = 0;
 455
 456        if (p->fixed) {
 457                if (opt->regflags & REG_ICASE || p->ignore_case)
 458                        p->kws = kwsalloc(tolower_trans_tbl);
 459                else
 460                        p->kws = kwsalloc(NULL);
 461                kwsincr(p->kws, p->pattern, p->patternlen);
 462                kwsprep(p->kws);
 463                return;
 464        } else if (opt->fixed) {
 465                /*
 466                 * We come here when the pattern has the non-ascii
 467                 * characters we cannot case-fold, and asked to
 468                 * ignore-case.
 469                 */
 470                compile_fixed_regexp(p, opt);
 471                return;
 472        }
 473
 474        if (opt->pcre) {
 475                compile_pcre_regexp(p, opt);
 476                return;
 477        }
 478
 479        err = regcomp(&p->regexp, p->pattern, opt->regflags);
 480        if (err) {
 481                char errbuf[1024];
 482                regerror(err, &p->regexp, errbuf, 1024);
 483                regfree(&p->regexp);
 484                compile_regexp_failed(p, errbuf);
 485        }
 486}
 487
 488static struct grep_expr *compile_pattern_or(struct grep_pat **);
 489static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 490{
 491        struct grep_pat *p;
 492        struct grep_expr *x;
 493
 494        p = *list;
 495        if (!p)
 496                return NULL;
 497        switch (p->token) {
 498        case GREP_PATTERN: /* atom */
 499        case GREP_PATTERN_HEAD:
 500        case GREP_PATTERN_BODY:
 501                x = xcalloc(1, sizeof (struct grep_expr));
 502                x->node = GREP_NODE_ATOM;
 503                x->u.atom = p;
 504                *list = p->next;
 505                return x;
 506        case GREP_OPEN_PAREN:
 507                *list = p->next;
 508                x = compile_pattern_or(list);
 509                if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 510                        die("unmatched parenthesis");
 511                *list = (*list)->next;
 512                return x;
 513        default:
 514                return NULL;
 515        }
 516}
 517
 518static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 519{
 520        struct grep_pat *p;
 521        struct grep_expr *x;
 522
 523        p = *list;
 524        if (!p)
 525                return NULL;
 526        switch (p->token) {
 527        case GREP_NOT:
 528                if (!p->next)
 529                        die("--not not followed by pattern expression");
 530                *list = p->next;
 531                x = xcalloc(1, sizeof (struct grep_expr));
 532                x->node = GREP_NODE_NOT;
 533                x->u.unary = compile_pattern_not(list);
 534                if (!x->u.unary)
 535                        die("--not followed by non pattern expression");
 536                return x;
 537        default:
 538                return compile_pattern_atom(list);
 539        }
 540}
 541
 542static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 543{
 544        struct grep_pat *p;
 545        struct grep_expr *x, *y, *z;
 546
 547        x = compile_pattern_not(list);
 548        p = *list;
 549        if (p && p->token == GREP_AND) {
 550                if (!p->next)
 551                        die("--and not followed by pattern expression");
 552                *list = p->next;
 553                y = compile_pattern_and(list);
 554                if (!y)
 555                        die("--and not followed by pattern expression");
 556                z = xcalloc(1, sizeof (struct grep_expr));
 557                z->node = GREP_NODE_AND;
 558                z->u.binary.left = x;
 559                z->u.binary.right = y;
 560                return z;
 561        }
 562        return x;
 563}
 564
 565static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 566{
 567        struct grep_pat *p;
 568        struct grep_expr *x, *y, *z;
 569
 570        x = compile_pattern_and(list);
 571        p = *list;
 572        if (x && p && p->token != GREP_CLOSE_PAREN) {
 573                y = compile_pattern_or(list);
 574                if (!y)
 575                        die("not a pattern expression %s", p->pattern);
 576                z = xcalloc(1, sizeof (struct grep_expr));
 577                z->node = GREP_NODE_OR;
 578                z->u.binary.left = x;
 579                z->u.binary.right = y;
 580                return z;
 581        }
 582        return x;
 583}
 584
 585static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 586{
 587        return compile_pattern_or(list);
 588}
 589
 590static void indent(int in)
 591{
 592        while (in-- > 0)
 593                fputc(' ', stderr);
 594}
 595
 596static void dump_grep_pat(struct grep_pat *p)
 597{
 598        switch (p->token) {
 599        case GREP_AND: fprintf(stderr, "*and*"); break;
 600        case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 601        case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 602        case GREP_NOT: fprintf(stderr, "*not*"); break;
 603        case GREP_OR: fprintf(stderr, "*or*"); break;
 604
 605        case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 606        case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 607        case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 608        }
 609
 610        switch (p->token) {
 611        default: break;
 612        case GREP_PATTERN_HEAD:
 613                fprintf(stderr, "<head %d>", p->field); break;
 614        case GREP_PATTERN_BODY:
 615                fprintf(stderr, "<body>"); break;
 616        }
 617        switch (p->token) {
 618        default: break;
 619        case GREP_PATTERN_HEAD:
 620        case GREP_PATTERN_BODY:
 621        case GREP_PATTERN:
 622                fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 623                break;
 624        }
 625        fputc('\n', stderr);
 626}
 627
 628static void dump_grep_expression_1(struct grep_expr *x, int in)
 629{
 630        indent(in);
 631        switch (x->node) {
 632        case GREP_NODE_TRUE:
 633                fprintf(stderr, "true\n");
 634                break;
 635        case GREP_NODE_ATOM:
 636                dump_grep_pat(x->u.atom);
 637                break;
 638        case GREP_NODE_NOT:
 639                fprintf(stderr, "(not\n");
 640                dump_grep_expression_1(x->u.unary, in+1);
 641                indent(in);
 642                fprintf(stderr, ")\n");
 643                break;
 644        case GREP_NODE_AND:
 645                fprintf(stderr, "(and\n");
 646                dump_grep_expression_1(x->u.binary.left, in+1);
 647                dump_grep_expression_1(x->u.binary.right, in+1);
 648                indent(in);
 649                fprintf(stderr, ")\n");
 650                break;
 651        case GREP_NODE_OR:
 652                fprintf(stderr, "(or\n");
 653                dump_grep_expression_1(x->u.binary.left, in+1);
 654                dump_grep_expression_1(x->u.binary.right, in+1);
 655                indent(in);
 656                fprintf(stderr, ")\n");
 657                break;
 658        }
 659}
 660
 661static void dump_grep_expression(struct grep_opt *opt)
 662{
 663        struct grep_expr *x = opt->pattern_expression;
 664
 665        if (opt->all_match)
 666                fprintf(stderr, "[all-match]\n");
 667        dump_grep_expression_1(x, 0);
 668        fflush(NULL);
 669}
 670
 671static struct grep_expr *grep_true_expr(void)
 672{
 673        struct grep_expr *z = xcalloc(1, sizeof(*z));
 674        z->node = GREP_NODE_TRUE;
 675        return z;
 676}
 677
 678static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 679{
 680        struct grep_expr *z = xcalloc(1, sizeof(*z));
 681        z->node = GREP_NODE_OR;
 682        z->u.binary.left = left;
 683        z->u.binary.right = right;
 684        return z;
 685}
 686
 687static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 688{
 689        struct grep_pat *p;
 690        struct grep_expr *header_expr;
 691        struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 692        enum grep_header_field fld;
 693
 694        if (!opt->header_list)
 695                return NULL;
 696
 697        for (p = opt->header_list; p; p = p->next) {
 698                if (p->token != GREP_PATTERN_HEAD)
 699                        die("bug: a non-header pattern in grep header list.");
 700                if (p->field < GREP_HEADER_FIELD_MIN ||
 701                    GREP_HEADER_FIELD_MAX <= p->field)
 702                        die("bug: unknown header field %d", p->field);
 703                compile_regexp(p, opt);
 704        }
 705
 706        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 707                header_group[fld] = NULL;
 708
 709        for (p = opt->header_list; p; p = p->next) {
 710                struct grep_expr *h;
 711                struct grep_pat *pp = p;
 712
 713                h = compile_pattern_atom(&pp);
 714                if (!h || pp != p->next)
 715                        die("bug: malformed header expr");
 716                if (!header_group[p->field]) {
 717                        header_group[p->field] = h;
 718                        continue;
 719                }
 720                header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 721        }
 722
 723        header_expr = NULL;
 724
 725        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 726                if (!header_group[fld])
 727                        continue;
 728                if (!header_expr)
 729                        header_expr = grep_true_expr();
 730                header_expr = grep_or_expr(header_group[fld], header_expr);
 731        }
 732        return header_expr;
 733}
 734
 735static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 736{
 737        struct grep_expr *z = x;
 738
 739        while (x) {
 740                assert(x->node == GREP_NODE_OR);
 741                if (x->u.binary.right &&
 742                    x->u.binary.right->node == GREP_NODE_TRUE) {
 743                        x->u.binary.right = y;
 744                        break;
 745                }
 746                x = x->u.binary.right;
 747        }
 748        return z;
 749}
 750
 751static void compile_grep_patterns_real(struct grep_opt *opt)
 752{
 753        struct grep_pat *p;
 754        struct grep_expr *header_expr = prep_header_patterns(opt);
 755
 756        for (p = opt->pattern_list; p; p = p->next) {
 757                switch (p->token) {
 758                case GREP_PATTERN: /* atom */
 759                case GREP_PATTERN_HEAD:
 760                case GREP_PATTERN_BODY:
 761                        compile_regexp(p, opt);
 762                        break;
 763                default:
 764                        opt->extended = 1;
 765                        break;
 766                }
 767        }
 768
 769        if (opt->all_match || header_expr)
 770                opt->extended = 1;
 771        else if (!opt->extended && !opt->debug)
 772                return;
 773
 774        p = opt->pattern_list;
 775        if (p)
 776                opt->pattern_expression = compile_pattern_expr(&p);
 777        if (p)
 778                die("incomplete pattern expression: %s", p->pattern);
 779
 780        if (!header_expr)
 781                return;
 782
 783        if (!opt->pattern_expression)
 784                opt->pattern_expression = header_expr;
 785        else if (opt->all_match)
 786                opt->pattern_expression = grep_splice_or(header_expr,
 787                                                         opt->pattern_expression);
 788        else
 789                opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 790                                                       header_expr);
 791        opt->all_match = 1;
 792}
 793
 794void compile_grep_patterns(struct grep_opt *opt)
 795{
 796        compile_grep_patterns_real(opt);
 797        if (opt->debug)
 798                dump_grep_expression(opt);
 799}
 800
 801static void free_pattern_expr(struct grep_expr *x)
 802{
 803        switch (x->node) {
 804        case GREP_NODE_TRUE:
 805        case GREP_NODE_ATOM:
 806                break;
 807        case GREP_NODE_NOT:
 808                free_pattern_expr(x->u.unary);
 809                break;
 810        case GREP_NODE_AND:
 811        case GREP_NODE_OR:
 812                free_pattern_expr(x->u.binary.left);
 813                free_pattern_expr(x->u.binary.right);
 814                break;
 815        }
 816        free(x);
 817}
 818
 819void free_grep_patterns(struct grep_opt *opt)
 820{
 821        struct grep_pat *p, *n;
 822
 823        for (p = opt->pattern_list; p; p = n) {
 824                n = p->next;
 825                switch (p->token) {
 826                case GREP_PATTERN: /* atom */
 827                case GREP_PATTERN_HEAD:
 828                case GREP_PATTERN_BODY:
 829                        if (p->kws)
 830                                kwsfree(p->kws);
 831                        else if (p->pcre_regexp)
 832                                free_pcre_regexp(p);
 833                        else
 834                                regfree(&p->regexp);
 835                        free(p->pattern);
 836                        break;
 837                default:
 838                        break;
 839                }
 840                free(p);
 841        }
 842
 843        if (!opt->extended)
 844                return;
 845        free_pattern_expr(opt->pattern_expression);
 846}
 847
 848static char *end_of_line(char *cp, unsigned long *left)
 849{
 850        unsigned long l = *left;
 851        while (l && *cp != '\n') {
 852                l--;
 853                cp++;
 854        }
 855        *left = l;
 856        return cp;
 857}
 858
 859static int word_char(char ch)
 860{
 861        return isalnum(ch) || ch == '_';
 862}
 863
 864static void output_color(struct grep_opt *opt, const void *data, size_t size,
 865                         const char *color)
 866{
 867        if (want_color(opt->color) && color && color[0]) {
 868                opt->output(opt, color, strlen(color));
 869                opt->output(opt, data, size);
 870                opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 871        } else
 872                opt->output(opt, data, size);
 873}
 874
 875static void output_sep(struct grep_opt *opt, char sign)
 876{
 877        if (opt->null_following_name)
 878                opt->output(opt, "\0", 1);
 879        else
 880                output_color(opt, &sign, 1, opt->color_sep);
 881}
 882
 883static void show_name(struct grep_opt *opt, const char *name)
 884{
 885        output_color(opt, name, strlen(name), opt->color_filename);
 886        opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 887}
 888
 889static int fixmatch(struct grep_pat *p, char *line, char *eol,
 890                    regmatch_t *match)
 891{
 892        struct kwsmatch kwsm;
 893        size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 894        if (offset == -1) {
 895                match->rm_so = match->rm_eo = -1;
 896                return REG_NOMATCH;
 897        } else {
 898                match->rm_so = offset;
 899                match->rm_eo = match->rm_so + kwsm.size[0];
 900                return 0;
 901        }
 902}
 903
 904static int regmatch(const regex_t *preg, char *line, char *eol,
 905                    regmatch_t *match, int eflags)
 906{
 907#ifdef REG_STARTEND
 908        match->rm_so = 0;
 909        match->rm_eo = eol - line;
 910        eflags |= REG_STARTEND;
 911#endif
 912        return regexec(preg, line, 1, match, eflags);
 913}
 914
 915static int patmatch(struct grep_pat *p, char *line, char *eol,
 916                    regmatch_t *match, int eflags)
 917{
 918        int hit;
 919
 920        if (p->fixed)
 921                hit = !fixmatch(p, line, eol, match);
 922        else if (p->pcre_regexp)
 923                hit = !pcrematch(p, line, eol, match, eflags);
 924        else
 925                hit = !regmatch(&p->regexp, line, eol, match, eflags);
 926
 927        return hit;
 928}
 929
 930static int strip_timestamp(char *bol, char **eol_p)
 931{
 932        char *eol = *eol_p;
 933        int ch;
 934
 935        while (bol < --eol) {
 936                if (*eol != '>')
 937                        continue;
 938                *eol_p = ++eol;
 939                ch = *eol;
 940                *eol = '\0';
 941                return ch;
 942        }
 943        return 0;
 944}
 945
 946static struct {
 947        const char *field;
 948        size_t len;
 949} header_field[] = {
 950        { "author ", 7 },
 951        { "committer ", 10 },
 952        { "reflog ", 7 },
 953};
 954
 955static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 956                             enum grep_context ctx,
 957                             regmatch_t *pmatch, int eflags)
 958{
 959        int hit = 0;
 960        int saved_ch = 0;
 961        const char *start = bol;
 962
 963        if ((p->token != GREP_PATTERN) &&
 964            ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 965                return 0;
 966
 967        if (p->token == GREP_PATTERN_HEAD) {
 968                const char *field;
 969                size_t len;
 970                assert(p->field < ARRAY_SIZE(header_field));
 971                field = header_field[p->field].field;
 972                len = header_field[p->field].len;
 973                if (strncmp(bol, field, len))
 974                        return 0;
 975                bol += len;
 976                switch (p->field) {
 977                case GREP_HEADER_AUTHOR:
 978                case GREP_HEADER_COMMITTER:
 979                        saved_ch = strip_timestamp(bol, &eol);
 980                        break;
 981                default:
 982                        break;
 983                }
 984        }
 985
 986 again:
 987        hit = patmatch(p, bol, eol, pmatch, eflags);
 988
 989        if (hit && p->word_regexp) {
 990                if ((pmatch[0].rm_so < 0) ||
 991                    (eol - bol) < pmatch[0].rm_so ||
 992                    (pmatch[0].rm_eo < 0) ||
 993                    (eol - bol) < pmatch[0].rm_eo)
 994                        die("regexp returned nonsense");
 995
 996                /* Match beginning must be either beginning of the
 997                 * line, or at word boundary (i.e. the last char must
 998                 * not be a word char).  Similarly, match end must be
 999                 * either end of the line, or at word boundary
1000                 * (i.e. the next char must not be a word char).
1001                 */
1002                if ( ((pmatch[0].rm_so == 0) ||
1003                      !word_char(bol[pmatch[0].rm_so-1])) &&
1004                     ((pmatch[0].rm_eo == (eol-bol)) ||
1005                      !word_char(bol[pmatch[0].rm_eo])) )
1006                        ;
1007                else
1008                        hit = 0;
1009
1010                /* Words consist of at least one character. */
1011                if (pmatch->rm_so == pmatch->rm_eo)
1012                        hit = 0;
1013
1014                if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1015                        /* There could be more than one match on the
1016                         * line, and the first match might not be
1017                         * strict word match.  But later ones could be!
1018                         * Forward to the next possible start, i.e. the
1019                         * next position following a non-word char.
1020                         */
1021                        bol = pmatch[0].rm_so + bol + 1;
1022                        while (word_char(bol[-1]) && bol < eol)
1023                                bol++;
1024                        eflags |= REG_NOTBOL;
1025                        if (bol < eol)
1026                                goto again;
1027                }
1028        }
1029        if (p->token == GREP_PATTERN_HEAD && saved_ch)
1030                *eol = saved_ch;
1031        if (hit) {
1032                pmatch[0].rm_so += bol - start;
1033                pmatch[0].rm_eo += bol - start;
1034        }
1035        return hit;
1036}
1037
1038static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1039                           enum grep_context ctx, int collect_hits)
1040{
1041        int h = 0;
1042        regmatch_t match;
1043
1044        if (!x)
1045                die("Not a valid grep expression");
1046        switch (x->node) {
1047        case GREP_NODE_TRUE:
1048                h = 1;
1049                break;
1050        case GREP_NODE_ATOM:
1051                h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1052                break;
1053        case GREP_NODE_NOT:
1054                h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1055                break;
1056        case GREP_NODE_AND:
1057                if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1058                        return 0;
1059                h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1060                break;
1061        case GREP_NODE_OR:
1062                if (!collect_hits)
1063                        return (match_expr_eval(x->u.binary.left,
1064                                                bol, eol, ctx, 0) ||
1065                                match_expr_eval(x->u.binary.right,
1066                                                bol, eol, ctx, 0));
1067                h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1068                x->u.binary.left->hit |= h;
1069                h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1070                break;
1071        default:
1072                die("Unexpected node type (internal error) %d", x->node);
1073        }
1074        if (collect_hits)
1075                x->hit |= h;
1076        return h;
1077}
1078
1079static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1080                      enum grep_context ctx, int collect_hits)
1081{
1082        struct grep_expr *x = opt->pattern_expression;
1083        return match_expr_eval(x, bol, eol, ctx, collect_hits);
1084}
1085
1086static int match_line(struct grep_opt *opt, char *bol, char *eol,
1087                      enum grep_context ctx, int collect_hits)
1088{
1089        struct grep_pat *p;
1090        regmatch_t match;
1091
1092        if (opt->extended)
1093                return match_expr(opt, bol, eol, ctx, collect_hits);
1094
1095        /* we do not call with collect_hits without being extended */
1096        for (p = opt->pattern_list; p; p = p->next) {
1097                if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1098                        return 1;
1099        }
1100        return 0;
1101}
1102
1103static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1104                              enum grep_context ctx,
1105                              regmatch_t *pmatch, int eflags)
1106{
1107        regmatch_t match;
1108
1109        if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1110                return 0;
1111        if (match.rm_so < 0 || match.rm_eo < 0)
1112                return 0;
1113        if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1114                if (match.rm_so > pmatch->rm_so)
1115                        return 1;
1116                if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1117                        return 1;
1118        }
1119        pmatch->rm_so = match.rm_so;
1120        pmatch->rm_eo = match.rm_eo;
1121        return 1;
1122}
1123
1124static int next_match(struct grep_opt *opt, char *bol, char *eol,
1125                      enum grep_context ctx, regmatch_t *pmatch, int eflags)
1126{
1127        struct grep_pat *p;
1128        int hit = 0;
1129
1130        pmatch->rm_so = pmatch->rm_eo = -1;
1131        if (bol < eol) {
1132                for (p = opt->pattern_list; p; p = p->next) {
1133                        switch (p->token) {
1134                        case GREP_PATTERN: /* atom */
1135                        case GREP_PATTERN_HEAD:
1136                        case GREP_PATTERN_BODY:
1137                                hit |= match_next_pattern(p, bol, eol, ctx,
1138                                                          pmatch, eflags);
1139                                break;
1140                        default:
1141                                break;
1142                        }
1143                }
1144        }
1145        return hit;
1146}
1147
1148static void show_line(struct grep_opt *opt, char *bol, char *eol,
1149                      const char *name, unsigned lno, char sign)
1150{
1151        int rest = eol - bol;
1152        const char *match_color, *line_color = NULL;
1153
1154        if (opt->file_break && opt->last_shown == 0) {
1155                if (opt->show_hunk_mark)
1156                        opt->output(opt, "\n", 1);
1157        } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1158                if (opt->last_shown == 0) {
1159                        if (opt->show_hunk_mark) {
1160                                output_color(opt, "--", 2, opt->color_sep);
1161                                opt->output(opt, "\n", 1);
1162                        }
1163                } else if (lno > opt->last_shown + 1) {
1164                        output_color(opt, "--", 2, opt->color_sep);
1165                        opt->output(opt, "\n", 1);
1166                }
1167        }
1168        if (opt->heading && opt->last_shown == 0) {
1169                output_color(opt, name, strlen(name), opt->color_filename);
1170                opt->output(opt, "\n", 1);
1171        }
1172        opt->last_shown = lno;
1173
1174        if (!opt->heading && opt->pathname) {
1175                output_color(opt, name, strlen(name), opt->color_filename);
1176                output_sep(opt, sign);
1177        }
1178        if (opt->linenum) {
1179                char buf[32];
1180                snprintf(buf, sizeof(buf), "%d", lno);
1181                output_color(opt, buf, strlen(buf), opt->color_lineno);
1182                output_sep(opt, sign);
1183        }
1184        if (opt->color) {
1185                regmatch_t match;
1186                enum grep_context ctx = GREP_CONTEXT_BODY;
1187                int ch = *eol;
1188                int eflags = 0;
1189
1190                if (sign == ':')
1191                        match_color = opt->color_match_selected;
1192                else
1193                        match_color = opt->color_match_context;
1194                if (sign == ':')
1195                        line_color = opt->color_selected;
1196                else if (sign == '-')
1197                        line_color = opt->color_context;
1198                else if (sign == '=')
1199                        line_color = opt->color_function;
1200                *eol = '\0';
1201                while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1202                        if (match.rm_so == match.rm_eo)
1203                                break;
1204
1205                        output_color(opt, bol, match.rm_so, line_color);
1206                        output_color(opt, bol + match.rm_so,
1207                                     match.rm_eo - match.rm_so, match_color);
1208                        bol += match.rm_eo;
1209                        rest -= match.rm_eo;
1210                        eflags = REG_NOTBOL;
1211                }
1212                *eol = ch;
1213        }
1214        output_color(opt, bol, rest, line_color);
1215        opt->output(opt, "\n", 1);
1216}
1217
1218#ifndef NO_PTHREADS
1219int grep_use_locks;
1220
1221/*
1222 * This lock protects access to the gitattributes machinery, which is
1223 * not thread-safe.
1224 */
1225pthread_mutex_t grep_attr_mutex;
1226
1227static inline void grep_attr_lock(void)
1228{
1229        if (grep_use_locks)
1230                pthread_mutex_lock(&grep_attr_mutex);
1231}
1232
1233static inline void grep_attr_unlock(void)
1234{
1235        if (grep_use_locks)
1236                pthread_mutex_unlock(&grep_attr_mutex);
1237}
1238
1239/*
1240 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1241 */
1242pthread_mutex_t grep_read_mutex;
1243
1244#else
1245#define grep_attr_lock()
1246#define grep_attr_unlock()
1247#endif
1248
1249static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1250{
1251        xdemitconf_t *xecfg = opt->priv;
1252        if (xecfg && !xecfg->find_func) {
1253                grep_source_load_driver(gs);
1254                if (gs->driver->funcname.pattern) {
1255                        const struct userdiff_funcname *pe = &gs->driver->funcname;
1256                        xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1257                } else {
1258                        xecfg = opt->priv = NULL;
1259                }
1260        }
1261
1262        if (xecfg) {
1263                char buf[1];
1264                return xecfg->find_func(bol, eol - bol, buf, 1,
1265                                        xecfg->find_func_priv) >= 0;
1266        }
1267
1268        if (bol == eol)
1269                return 0;
1270        if (isalpha(*bol) || *bol == '_' || *bol == '$')
1271                return 1;
1272        return 0;
1273}
1274
1275static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1276                               char *bol, unsigned lno)
1277{
1278        while (bol > gs->buf) {
1279                char *eol = --bol;
1280
1281                while (bol > gs->buf && bol[-1] != '\n')
1282                        bol--;
1283                lno--;
1284
1285                if (lno <= opt->last_shown)
1286                        break;
1287
1288                if (match_funcname(opt, gs, bol, eol)) {
1289                        show_line(opt, bol, eol, gs->name, lno, '=');
1290                        break;
1291                }
1292        }
1293}
1294
1295static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1296                             char *bol, char *end, unsigned lno)
1297{
1298        unsigned cur = lno, from = 1, funcname_lno = 0;
1299        int funcname_needed = !!opt->funcname;
1300
1301        if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1302                funcname_needed = 2;
1303
1304        if (opt->pre_context < lno)
1305                from = lno - opt->pre_context;
1306        if (from <= opt->last_shown)
1307                from = opt->last_shown + 1;
1308
1309        /* Rewind. */
1310        while (bol > gs->buf &&
1311               cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1312                char *eol = --bol;
1313
1314                while (bol > gs->buf && bol[-1] != '\n')
1315                        bol--;
1316                cur--;
1317                if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1318                        funcname_lno = cur;
1319                        funcname_needed = 0;
1320                }
1321        }
1322
1323        /* We need to look even further back to find a function signature. */
1324        if (opt->funcname && funcname_needed)
1325                show_funcname_line(opt, gs, bol, cur);
1326
1327        /* Back forward. */
1328        while (cur < lno) {
1329                char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1330
1331                while (*eol != '\n')
1332                        eol++;
1333                show_line(opt, bol, eol, gs->name, cur, sign);
1334                bol = eol + 1;
1335                cur++;
1336        }
1337}
1338
1339static int should_lookahead(struct grep_opt *opt)
1340{
1341        struct grep_pat *p;
1342
1343        if (opt->extended)
1344                return 0; /* punt for too complex stuff */
1345        if (opt->invert)
1346                return 0;
1347        for (p = opt->pattern_list; p; p = p->next) {
1348                if (p->token != GREP_PATTERN)
1349                        return 0; /* punt for "header only" and stuff */
1350        }
1351        return 1;
1352}
1353
1354static int look_ahead(struct grep_opt *opt,
1355                      unsigned long *left_p,
1356                      unsigned *lno_p,
1357                      char **bol_p)
1358{
1359        unsigned lno = *lno_p;
1360        char *bol = *bol_p;
1361        struct grep_pat *p;
1362        char *sp, *last_bol;
1363        regoff_t earliest = -1;
1364
1365        for (p = opt->pattern_list; p; p = p->next) {
1366                int hit;
1367                regmatch_t m;
1368
1369                hit = patmatch(p, bol, bol + *left_p, &m, 0);
1370                if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1371                        continue;
1372                if (earliest < 0 || m.rm_so < earliest)
1373                        earliest = m.rm_so;
1374        }
1375
1376        if (earliest < 0) {
1377                *bol_p = bol + *left_p;
1378                *left_p = 0;
1379                return 1;
1380        }
1381        for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1382                ; /* find the beginning of the line */
1383        last_bol = sp;
1384
1385        for (sp = bol; sp < last_bol; sp++) {
1386                if (*sp == '\n')
1387                        lno++;
1388        }
1389        *left_p -= last_bol - bol;
1390        *bol_p = last_bol;
1391        *lno_p = lno;
1392        return 0;
1393}
1394
1395static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1396{
1397        fwrite(buf, size, 1, stdout);
1398}
1399
1400static int fill_textconv_grep(struct userdiff_driver *driver,
1401                              struct grep_source *gs)
1402{
1403        struct diff_filespec *df;
1404        char *buf;
1405        size_t size;
1406
1407        if (!driver || !driver->textconv)
1408                return grep_source_load(gs);
1409
1410        /*
1411         * The textconv interface is intimately tied to diff_filespecs, so we
1412         * have to pretend to be one. If we could unify the grep_source
1413         * and diff_filespec structs, this mess could just go away.
1414         */
1415        df = alloc_filespec(gs->path);
1416        switch (gs->type) {
1417        case GREP_SOURCE_SHA1:
1418                fill_filespec(df, gs->identifier, 1, 0100644);
1419                break;
1420        case GREP_SOURCE_FILE:
1421                fill_filespec(df, null_sha1, 0, 0100644);
1422                break;
1423        default:
1424                die("BUG: attempt to textconv something without a path?");
1425        }
1426
1427        /*
1428         * fill_textconv is not remotely thread-safe; it may load objects
1429         * behind the scenes, and it modifies the global diff tempfile
1430         * structure.
1431         */
1432        grep_read_lock();
1433        size = fill_textconv(driver, df, &buf);
1434        grep_read_unlock();
1435        free_filespec(df);
1436
1437        /*
1438         * The normal fill_textconv usage by the diff machinery would just keep
1439         * the textconv'd buf separate from the diff_filespec. But much of the
1440         * grep code passes around a grep_source and assumes that its "buf"
1441         * pointer is the beginning of the thing we are searching. So let's
1442         * install our textconv'd version into the grep_source, taking care not
1443         * to leak any existing buffer.
1444         */
1445        grep_source_clear_data(gs);
1446        gs->buf = buf;
1447        gs->size = size;
1448
1449        return 0;
1450}
1451
1452static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1453{
1454        char *bol;
1455        unsigned long left;
1456        unsigned lno = 1;
1457        unsigned last_hit = 0;
1458        int binary_match_only = 0;
1459        unsigned count = 0;
1460        int try_lookahead = 0;
1461        int show_function = 0;
1462        struct userdiff_driver *textconv = NULL;
1463        enum grep_context ctx = GREP_CONTEXT_HEAD;
1464        xdemitconf_t xecfg;
1465
1466        if (!opt->output)
1467                opt->output = std_output;
1468
1469        if (opt->pre_context || opt->post_context || opt->file_break ||
1470            opt->funcbody) {
1471                /* Show hunk marks, except for the first file. */
1472                if (opt->last_shown)
1473                        opt->show_hunk_mark = 1;
1474                /*
1475                 * If we're using threads then we can't easily identify
1476                 * the first file.  Always put hunk marks in that case
1477                 * and skip the very first one later in work_done().
1478                 */
1479                if (opt->output != std_output)
1480                        opt->show_hunk_mark = 1;
1481        }
1482        opt->last_shown = 0;
1483
1484        if (opt->allow_textconv) {
1485                grep_source_load_driver(gs);
1486                /*
1487                 * We might set up the shared textconv cache data here, which
1488                 * is not thread-safe.
1489                 */
1490                grep_attr_lock();
1491                textconv = userdiff_get_textconv(gs->driver);
1492                grep_attr_unlock();
1493        }
1494
1495        /*
1496         * We know the result of a textconv is text, so we only have to care
1497         * about binary handling if we are not using it.
1498         */
1499        if (!textconv) {
1500                switch (opt->binary) {
1501                case GREP_BINARY_DEFAULT:
1502                        if (grep_source_is_binary(gs))
1503                                binary_match_only = 1;
1504                        break;
1505                case GREP_BINARY_NOMATCH:
1506                        if (grep_source_is_binary(gs))
1507                                return 0; /* Assume unmatch */
1508                        break;
1509                case GREP_BINARY_TEXT:
1510                        break;
1511                default:
1512                        die("bug: unknown binary handling mode");
1513                }
1514        }
1515
1516        memset(&xecfg, 0, sizeof(xecfg));
1517        opt->priv = &xecfg;
1518
1519        try_lookahead = should_lookahead(opt);
1520
1521        if (fill_textconv_grep(textconv, gs) < 0)
1522                return 0;
1523
1524        bol = gs->buf;
1525        left = gs->size;
1526        while (left) {
1527                char *eol, ch;
1528                int hit;
1529
1530                /*
1531                 * look_ahead() skips quickly to the line that possibly
1532                 * has the next hit; don't call it if we need to do
1533                 * something more than just skipping the current line
1534                 * in response to an unmatch for the current line.  E.g.
1535                 * inside a post-context window, we will show the current
1536                 * line as a context around the previous hit when it
1537                 * doesn't hit.
1538                 */
1539                if (try_lookahead
1540                    && !(last_hit
1541                         && (show_function ||
1542                             lno <= last_hit + opt->post_context))
1543                    && look_ahead(opt, &left, &lno, &bol))
1544                        break;
1545                eol = end_of_line(bol, &left);
1546                ch = *eol;
1547                *eol = 0;
1548
1549                if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1550                        ctx = GREP_CONTEXT_BODY;
1551
1552                hit = match_line(opt, bol, eol, ctx, collect_hits);
1553                *eol = ch;
1554
1555                if (collect_hits)
1556                        goto next_line;
1557
1558                /* "grep -v -e foo -e bla" should list lines
1559                 * that do not have either, so inversion should
1560                 * be done outside.
1561                 */
1562                if (opt->invert)
1563                        hit = !hit;
1564                if (opt->unmatch_name_only) {
1565                        if (hit)
1566                                return 0;
1567                        goto next_line;
1568                }
1569                if (hit) {
1570                        count++;
1571                        if (opt->status_only)
1572                                return 1;
1573                        if (opt->name_only) {
1574                                show_name(opt, gs->name);
1575                                return 1;
1576                        }
1577                        if (opt->count)
1578                                goto next_line;
1579                        if (binary_match_only) {
1580                                opt->output(opt, "Binary file ", 12);
1581                                output_color(opt, gs->name, strlen(gs->name),
1582                                             opt->color_filename);
1583                                opt->output(opt, " matches\n", 9);
1584                                return 1;
1585                        }
1586                        /* Hit at this line.  If we haven't shown the
1587                         * pre-context lines, we would need to show them.
1588                         */
1589                        if (opt->pre_context || opt->funcbody)
1590                                show_pre_context(opt, gs, bol, eol, lno);
1591                        else if (opt->funcname)
1592                                show_funcname_line(opt, gs, bol, lno);
1593                        show_line(opt, bol, eol, gs->name, lno, ':');
1594                        last_hit = lno;
1595                        if (opt->funcbody)
1596                                show_function = 1;
1597                        goto next_line;
1598                }
1599                if (show_function && match_funcname(opt, gs, bol, eol))
1600                        show_function = 0;
1601                if (show_function ||
1602                    (last_hit && lno <= last_hit + opt->post_context)) {
1603                        /* If the last hit is within the post context,
1604                         * we need to show this line.
1605                         */
1606                        show_line(opt, bol, eol, gs->name, lno, '-');
1607                }
1608
1609        next_line:
1610                bol = eol + 1;
1611                if (!left)
1612                        break;
1613                left--;
1614                lno++;
1615        }
1616
1617        if (collect_hits)
1618                return 0;
1619
1620        if (opt->status_only)
1621                return 0;
1622        if (opt->unmatch_name_only) {
1623                /* We did not see any hit, so we want to show this */
1624                show_name(opt, gs->name);
1625                return 1;
1626        }
1627
1628        xdiff_clear_find_func(&xecfg);
1629        opt->priv = NULL;
1630
1631        /* NEEDSWORK:
1632         * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1633         * which feels mostly useless but sometimes useful.  Maybe
1634         * make it another option?  For now suppress them.
1635         */
1636        if (opt->count && count) {
1637                char buf[32];
1638                if (opt->pathname) {
1639                        output_color(opt, gs->name, strlen(gs->name),
1640                                     opt->color_filename);
1641                        output_sep(opt, ':');
1642                }
1643                snprintf(buf, sizeof(buf), "%u\n", count);
1644                opt->output(opt, buf, strlen(buf));
1645                return 1;
1646        }
1647        return !!last_hit;
1648}
1649
1650static void clr_hit_marker(struct grep_expr *x)
1651{
1652        /* All-hit markers are meaningful only at the very top level
1653         * OR node.
1654         */
1655        while (1) {
1656                x->hit = 0;
1657                if (x->node != GREP_NODE_OR)
1658                        return;
1659                x->u.binary.left->hit = 0;
1660                x = x->u.binary.right;
1661        }
1662}
1663
1664static int chk_hit_marker(struct grep_expr *x)
1665{
1666        /* Top level nodes have hit markers.  See if they all are hits */
1667        while (1) {
1668                if (x->node != GREP_NODE_OR)
1669                        return x->hit;
1670                if (!x->u.binary.left->hit)
1671                        return 0;
1672                x = x->u.binary.right;
1673        }
1674}
1675
1676int grep_source(struct grep_opt *opt, struct grep_source *gs)
1677{
1678        /*
1679         * we do not have to do the two-pass grep when we do not check
1680         * buffer-wide "all-match".
1681         */
1682        if (!opt->all_match)
1683                return grep_source_1(opt, gs, 0);
1684
1685        /* Otherwise the toplevel "or" terms hit a bit differently.
1686         * We first clear hit markers from them.
1687         */
1688        clr_hit_marker(opt->pattern_expression);
1689        grep_source_1(opt, gs, 1);
1690
1691        if (!chk_hit_marker(opt->pattern_expression))
1692                return 0;
1693
1694        return grep_source_1(opt, gs, 0);
1695}
1696
1697int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1698{
1699        struct grep_source gs;
1700        int r;
1701
1702        grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1703        gs.buf = buf;
1704        gs.size = size;
1705
1706        r = grep_source(opt, &gs);
1707
1708        grep_source_clear(&gs);
1709        return r;
1710}
1711
1712void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1713                      const char *name, const char *path,
1714                      const void *identifier)
1715{
1716        gs->type = type;
1717        gs->name = xstrdup_or_null(name);
1718        gs->path = xstrdup_or_null(path);
1719        gs->buf = NULL;
1720        gs->size = 0;
1721        gs->driver = NULL;
1722
1723        switch (type) {
1724        case GREP_SOURCE_FILE:
1725                gs->identifier = xstrdup(identifier);
1726                break;
1727        case GREP_SOURCE_SHA1:
1728                gs->identifier = xmalloc(20);
1729                hashcpy(gs->identifier, identifier);
1730                break;
1731        case GREP_SOURCE_BUF:
1732                gs->identifier = NULL;
1733        }
1734}
1735
1736void grep_source_clear(struct grep_source *gs)
1737{
1738        free(gs->name);
1739        gs->name = NULL;
1740        free(gs->path);
1741        gs->path = NULL;
1742        free(gs->identifier);
1743        gs->identifier = NULL;
1744        grep_source_clear_data(gs);
1745}
1746
1747void grep_source_clear_data(struct grep_source *gs)
1748{
1749        switch (gs->type) {
1750        case GREP_SOURCE_FILE:
1751        case GREP_SOURCE_SHA1:
1752                free(gs->buf);
1753                gs->buf = NULL;
1754                gs->size = 0;
1755                break;
1756        case GREP_SOURCE_BUF:
1757                /* leave user-provided buf intact */
1758                break;
1759        }
1760}
1761
1762static int grep_source_load_sha1(struct grep_source *gs)
1763{
1764        enum object_type type;
1765
1766        grep_read_lock();
1767        gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1768        grep_read_unlock();
1769
1770        if (!gs->buf)
1771                return error(_("'%s': unable to read %s"),
1772                             gs->name,
1773                             sha1_to_hex(gs->identifier));
1774        return 0;
1775}
1776
1777static int grep_source_load_file(struct grep_source *gs)
1778{
1779        const char *filename = gs->identifier;
1780        struct stat st;
1781        char *data;
1782        size_t size;
1783        int i;
1784
1785        if (lstat(filename, &st) < 0) {
1786        err_ret:
1787                if (errno != ENOENT)
1788                        error(_("'%s': %s"), filename, strerror(errno));
1789                return -1;
1790        }
1791        if (!S_ISREG(st.st_mode))
1792                return -1;
1793        size = xsize_t(st.st_size);
1794        i = open(filename, O_RDONLY);
1795        if (i < 0)
1796                goto err_ret;
1797        data = xmalloc(size + 1);
1798        if (st.st_size != read_in_full(i, data, size)) {
1799                error(_("'%s': short read %s"), filename, strerror(errno));
1800                close(i);
1801                free(data);
1802                return -1;
1803        }
1804        close(i);
1805        data[size] = 0;
1806
1807        gs->buf = data;
1808        gs->size = size;
1809        return 0;
1810}
1811
1812static int grep_source_load(struct grep_source *gs)
1813{
1814        if (gs->buf)
1815                return 0;
1816
1817        switch (gs->type) {
1818        case GREP_SOURCE_FILE:
1819                return grep_source_load_file(gs);
1820        case GREP_SOURCE_SHA1:
1821                return grep_source_load_sha1(gs);
1822        case GREP_SOURCE_BUF:
1823                return gs->buf ? 0 : -1;
1824        }
1825        die("BUG: invalid grep_source type");
1826}
1827
1828void grep_source_load_driver(struct grep_source *gs)
1829{
1830        if (gs->driver)
1831                return;
1832
1833        grep_attr_lock();
1834        if (gs->path)
1835                gs->driver = userdiff_find_by_path(gs->path);
1836        if (!gs->driver)
1837                gs->driver = userdiff_find_by_name("default");
1838        grep_attr_unlock();
1839}
1840
1841static int grep_source_is_binary(struct grep_source *gs)
1842{
1843        grep_source_load_driver(gs);
1844        if (gs->driver->binary != -1)
1845                return gs->driver->binary;
1846
1847        if (!grep_source_load(gs))
1848                return buffer_is_binary(gs->buf, gs->size);
1849
1850        return 0;
1851}