Andrew's git / gitweb.git / grep.c
?
grep.con commit am: release strbuf on error return in hg_patch_to_mail() (b36474f)
   1#include "cache.h"
   2#include "config.h"
   3#include "grep.h"
   4#include "userdiff.h"
   5#include "xdiff-interface.h"
   6#include "diff.h"
   7#include "diffcore.h"
   8#include "commit.h"
   9#include "quote.h"
  10
  11static int grep_source_load(struct grep_source *gs);
  12static int grep_source_is_binary(struct grep_source *gs);
  13
  14static struct grep_opt grep_defaults;
  15
  16static void std_output(struct grep_opt *opt, const void *buf, size_t size)
  17{
  18        fwrite(buf, size, 1, stdout);
  19}
  20
  21/*
  22 * Initialize the grep_defaults template with hardcoded defaults.
  23 * We could let the compiler do this, but without C99 initializers
  24 * the code gets unwieldy and unreadable, so...
  25 */
  26void init_grep_defaults(void)
  27{
  28        struct grep_opt *opt = &grep_defaults;
  29        static int run_once;
  30
  31        if (run_once)
  32                return;
  33        run_once++;
  34
  35        memset(opt, 0, sizeof(*opt));
  36        opt->relative = 1;
  37        opt->pathname = 1;
  38        opt->max_depth = -1;
  39        opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  40        color_set(opt->color_context, "");
  41        color_set(opt->color_filename, "");
  42        color_set(opt->color_function, "");
  43        color_set(opt->color_lineno, "");
  44        color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
  45        color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
  46        color_set(opt->color_selected, "");
  47        color_set(opt->color_sep, GIT_COLOR_CYAN);
  48        opt->color = -1;
  49        opt->output = std_output;
  50}
  51
  52static int parse_pattern_type_arg(const char *opt, const char *arg)
  53{
  54        if (!strcmp(arg, "default"))
  55                return GREP_PATTERN_TYPE_UNSPECIFIED;
  56        else if (!strcmp(arg, "basic"))
  57                return GREP_PATTERN_TYPE_BRE;
  58        else if (!strcmp(arg, "extended"))
  59                return GREP_PATTERN_TYPE_ERE;
  60        else if (!strcmp(arg, "fixed"))
  61                return GREP_PATTERN_TYPE_FIXED;
  62        else if (!strcmp(arg, "perl"))
  63                return GREP_PATTERN_TYPE_PCRE;
  64        die("bad %s argument: %s", opt, arg);
  65}
  66
  67/*
  68 * Read the configuration file once and store it in
  69 * the grep_defaults template.
  70 */
  71int grep_config(const char *var, const char *value, void *cb)
  72{
  73        struct grep_opt *opt = &grep_defaults;
  74        char *color = NULL;
  75
  76        if (userdiff_config(var, value) < 0)
  77                return -1;
  78
  79        if (!strcmp(var, "grep.extendedregexp")) {
  80                opt->extended_regexp_option = git_config_bool(var, value);
  81                return 0;
  82        }
  83
  84        if (!strcmp(var, "grep.patterntype")) {
  85                opt->pattern_type_option = parse_pattern_type_arg(var, value);
  86                return 0;
  87        }
  88
  89        if (!strcmp(var, "grep.linenumber")) {
  90                opt->linenum = git_config_bool(var, value);
  91                return 0;
  92        }
  93
  94        if (!strcmp(var, "grep.fullname")) {
  95                opt->relative = !git_config_bool(var, value);
  96                return 0;
  97        }
  98
  99        if (!strcmp(var, "color.grep"))
 100                opt->color = git_config_colorbool(var, value);
 101        else if (!strcmp(var, "color.grep.context"))
 102                color = opt->color_context;
 103        else if (!strcmp(var, "color.grep.filename"))
 104                color = opt->color_filename;
 105        else if (!strcmp(var, "color.grep.function"))
 106                color = opt->color_function;
 107        else if (!strcmp(var, "color.grep.linenumber"))
 108                color = opt->color_lineno;
 109        else if (!strcmp(var, "color.grep.matchcontext"))
 110                color = opt->color_match_context;
 111        else if (!strcmp(var, "color.grep.matchselected"))
 112                color = opt->color_match_selected;
 113        else if (!strcmp(var, "color.grep.selected"))
 114                color = opt->color_selected;
 115        else if (!strcmp(var, "color.grep.separator"))
 116                color = opt->color_sep;
 117        else if (!strcmp(var, "color.grep.match")) {
 118                int rc = 0;
 119                if (!value)
 120                        return config_error_nonbool(var);
 121                rc |= color_parse(value, opt->color_match_context);
 122                rc |= color_parse(value, opt->color_match_selected);
 123                return rc;
 124        }
 125
 126        if (color) {
 127                if (!value)
 128                        return config_error_nonbool(var);
 129                return color_parse(value, color);
 130        }
 131        return 0;
 132}
 133
 134/*
 135 * Initialize one instance of grep_opt and copy the
 136 * default values from the template we read the configuration
 137 * information in an earlier call to git_config(grep_config).
 138 */
 139void grep_init(struct grep_opt *opt, const char *prefix)
 140{
 141        struct grep_opt *def = &grep_defaults;
 142
 143        memset(opt, 0, sizeof(*opt));
 144        opt->prefix = prefix;
 145        opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 146        opt->pattern_tail = &opt->pattern_list;
 147        opt->header_tail = &opt->header_list;
 148
 149        opt->color = def->color;
 150        opt->extended_regexp_option = def->extended_regexp_option;
 151        opt->pattern_type_option = def->pattern_type_option;
 152        opt->linenum = def->linenum;
 153        opt->max_depth = def->max_depth;
 154        opt->pathname = def->pathname;
 155        opt->relative = def->relative;
 156        opt->output = def->output;
 157
 158        color_set(opt->color_context, def->color_context);
 159        color_set(opt->color_filename, def->color_filename);
 160        color_set(opt->color_function, def->color_function);
 161        color_set(opt->color_lineno, def->color_lineno);
 162        color_set(opt->color_match_context, def->color_match_context);
 163        color_set(opt->color_match_selected, def->color_match_selected);
 164        color_set(opt->color_selected, def->color_selected);
 165        color_set(opt->color_sep, def->color_sep);
 166}
 167
 168static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 169{
 170        /*
 171         * When committing to the pattern type by setting the relevant
 172         * fields in grep_opt it's generally not necessary to zero out
 173         * the fields we're not choosing, since they won't have been
 174         * set by anything. The extended_regexp_option field is the
 175         * only exception to this.
 176         *
 177         * This is because in the process of parsing grep.patternType
 178         * & grep.extendedRegexp we set opt->pattern_type_option and
 179         * opt->extended_regexp_option, respectively. We then
 180         * internally use opt->extended_regexp_option to see if we're
 181         * compiling an ERE. It must be unset if that's not actually
 182         * the case.
 183         */
 184        if (pattern_type != GREP_PATTERN_TYPE_ERE &&
 185            opt->extended_regexp_option)
 186                opt->extended_regexp_option = 0;
 187
 188        switch (pattern_type) {
 189        case GREP_PATTERN_TYPE_UNSPECIFIED:
 190                /* fall through */
 191
 192        case GREP_PATTERN_TYPE_BRE:
 193                break;
 194
 195        case GREP_PATTERN_TYPE_ERE:
 196                opt->extended_regexp_option = 1;
 197                break;
 198
 199        case GREP_PATTERN_TYPE_FIXED:
 200                opt->fixed = 1;
 201                break;
 202
 203        case GREP_PATTERN_TYPE_PCRE:
 204#ifdef USE_LIBPCRE2
 205                opt->pcre2 = 1;
 206#else
 207                /*
 208                 * It's important that pcre1 always be assigned to
 209                 * even when there's no USE_LIBPCRE* defined. We still
 210                 * call the PCRE stub function, it just dies with
 211                 * "cannot use Perl-compatible regexes[...]".
 212                 */
 213                opt->pcre1 = 1;
 214#endif
 215                break;
 216        }
 217}
 218
 219void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 220{
 221        if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 222                grep_set_pattern_type_option(pattern_type, opt);
 223        else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 224                grep_set_pattern_type_option(opt->pattern_type_option, opt);
 225        else if (opt->extended_regexp_option)
 226                /*
 227                 * This branch *must* happen after setting from the
 228                 * opt->pattern_type_option above, we don't want
 229                 * grep.extendedRegexp to override grep.patternType!
 230                 */
 231                grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 232}
 233
 234static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 235                                        const char *origin, int no,
 236                                        enum grep_pat_token t,
 237                                        enum grep_header_field field)
 238{
 239        struct grep_pat *p = xcalloc(1, sizeof(*p));
 240        p->pattern = xmemdupz(pat, patlen);
 241        p->patternlen = patlen;
 242        p->origin = origin;
 243        p->no = no;
 244        p->token = t;
 245        p->field = field;
 246        return p;
 247}
 248
 249static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 250{
 251        **tail = p;
 252        *tail = &p->next;
 253        p->next = NULL;
 254
 255        switch (p->token) {
 256        case GREP_PATTERN: /* atom */
 257        case GREP_PATTERN_HEAD:
 258        case GREP_PATTERN_BODY:
 259                for (;;) {
 260                        struct grep_pat *new_pat;
 261                        size_t len = 0;
 262                        char *cp = p->pattern + p->patternlen, *nl = NULL;
 263                        while (++len <= p->patternlen) {
 264                                if (*(--cp) == '\n') {
 265                                        nl = cp;
 266                                        break;
 267                                }
 268                        }
 269                        if (!nl)
 270                                break;
 271                        new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 272                                                  p->no, p->token, p->field);
 273                        new_pat->next = p->next;
 274                        if (!p->next)
 275                                *tail = &new_pat->next;
 276                        p->next = new_pat;
 277                        *nl = '\0';
 278                        p->patternlen -= len;
 279                }
 280                break;
 281        default:
 282                break;
 283        }
 284}
 285
 286void append_header_grep_pattern(struct grep_opt *opt,
 287                                enum grep_header_field field, const char *pat)
 288{
 289        struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 290                                             GREP_PATTERN_HEAD, field);
 291        if (field == GREP_HEADER_REFLOG)
 292                opt->use_reflog_filter = 1;
 293        do_append_grep_pat(&opt->header_tail, p);
 294}
 295
 296void append_grep_pattern(struct grep_opt *opt, const char *pat,
 297                         const char *origin, int no, enum grep_pat_token t)
 298{
 299        append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 300}
 301
 302void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 303                     const char *origin, int no, enum grep_pat_token t)
 304{
 305        struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 306        do_append_grep_pat(&opt->pattern_tail, p);
 307}
 308
 309struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 310{
 311        struct grep_pat *pat;
 312        struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 313        *ret = *opt;
 314
 315        ret->pattern_list = NULL;
 316        ret->pattern_tail = &ret->pattern_list;
 317
 318        for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 319        {
 320                if(pat->token == GREP_PATTERN_HEAD)
 321                        append_header_grep_pattern(ret, pat->field,
 322                                                   pat->pattern);
 323                else
 324                        append_grep_pat(ret, pat->pattern, pat->patternlen,
 325                                        pat->origin, pat->no, pat->token);
 326        }
 327
 328        return ret;
 329}
 330
 331static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 332                const char *error)
 333{
 334        char where[1024];
 335
 336        if (p->no)
 337                xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 338        else if (p->origin)
 339                xsnprintf(where, sizeof(where), "%s, ", p->origin);
 340        else
 341                where[0] = 0;
 342
 343        die("%s'%s': %s", where, p->pattern, error);
 344}
 345
 346static int is_fixed(const char *s, size_t len)
 347{
 348        size_t i;
 349
 350        for (i = 0; i < len; i++) {
 351                if (is_regex_special(s[i]))
 352                        return 0;
 353        }
 354
 355        return 1;
 356}
 357
 358static int has_null(const char *s, size_t len)
 359{
 360        /*
 361         * regcomp cannot accept patterns with NULs so when using it
 362         * we consider any pattern containing a NUL fixed.
 363         */
 364        if (memchr(s, 0, len))
 365                return 1;
 366
 367        return 0;
 368}
 369
 370#ifdef USE_LIBPCRE1
 371static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 372{
 373        const char *error;
 374        int erroffset;
 375        int options = PCRE_MULTILINE;
 376
 377        if (opt->ignore_case) {
 378                if (has_non_ascii(p->pattern))
 379                        p->pcre1_tables = pcre_maketables();
 380                options |= PCRE_CASELESS;
 381        }
 382        if (is_utf8_locale() && has_non_ascii(p->pattern))
 383                options |= PCRE_UTF8;
 384
 385        p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 386                                      p->pcre1_tables);
 387        if (!p->pcre1_regexp)
 388                compile_regexp_failed(p, error);
 389
 390        p->pcre1_extra_info = pcre_study(p->pcre1_regexp, PCRE_STUDY_JIT_COMPILE, &error);
 391        if (!p->pcre1_extra_info && error)
 392                die("%s", error);
 393
 394#ifdef GIT_PCRE1_USE_JIT
 395        pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
 396        if (p->pcre1_jit_on == 1) {
 397                p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
 398                if (!p->pcre1_jit_stack)
 399                        die("Couldn't allocate PCRE JIT stack");
 400                pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
 401        } else if (p->pcre1_jit_on != 0) {
 402                die("BUG: The pcre1_jit_on variable should be 0 or 1, not %d",
 403                    p->pcre1_jit_on);
 404        }
 405#endif
 406}
 407
 408static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 409                regmatch_t *match, int eflags)
 410{
 411        int ovector[30], ret, flags = 0;
 412
 413        if (eflags & REG_NOTBOL)
 414                flags |= PCRE_NOTBOL;
 415
 416#ifdef GIT_PCRE1_USE_JIT
 417        if (p->pcre1_jit_on) {
 418                ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 419                                    eol - line, 0, flags, ovector,
 420                                    ARRAY_SIZE(ovector), p->pcre1_jit_stack);
 421        } else
 422#endif
 423        {
 424                ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 425                                eol - line, 0, flags, ovector,
 426                                ARRAY_SIZE(ovector));
 427        }
 428
 429        if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 430                die("pcre_exec failed with error code %d", ret);
 431        if (ret > 0) {
 432                ret = 0;
 433                match->rm_so = ovector[0];
 434                match->rm_eo = ovector[1];
 435        }
 436
 437        return ret;
 438}
 439
 440static void free_pcre1_regexp(struct grep_pat *p)
 441{
 442        pcre_free(p->pcre1_regexp);
 443#ifdef GIT_PCRE1_USE_JIT
 444        if (p->pcre1_jit_on) {
 445                pcre_free_study(p->pcre1_extra_info);
 446                pcre_jit_stack_free(p->pcre1_jit_stack);
 447        } else
 448#endif
 449        {
 450                pcre_free(p->pcre1_extra_info);
 451        }
 452        pcre_free((void *)p->pcre1_tables);
 453}
 454#else /* !USE_LIBPCRE1 */
 455static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 456{
 457        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 458}
 459
 460static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 461                regmatch_t *match, int eflags)
 462{
 463        return 1;
 464}
 465
 466static void free_pcre1_regexp(struct grep_pat *p)
 467{
 468}
 469#endif /* !USE_LIBPCRE1 */
 470
 471#ifdef USE_LIBPCRE2
 472static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 473{
 474        int error;
 475        PCRE2_UCHAR errbuf[256];
 476        PCRE2_SIZE erroffset;
 477        int options = PCRE2_MULTILINE;
 478        const uint8_t *character_tables = NULL;
 479        int jitret;
 480
 481        assert(opt->pcre2);
 482
 483        p->pcre2_compile_context = NULL;
 484
 485        if (opt->ignore_case) {
 486                if (has_non_ascii(p->pattern)) {
 487                        character_tables = pcre2_maketables(NULL);
 488                        p->pcre2_compile_context = pcre2_compile_context_create(NULL);
 489                        pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
 490                }
 491                options |= PCRE2_CASELESS;
 492        }
 493        if (is_utf8_locale() && has_non_ascii(p->pattern))
 494                options |= PCRE2_UTF;
 495
 496        p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
 497                                         p->patternlen, options, &error, &erroffset,
 498                                         p->pcre2_compile_context);
 499
 500        if (p->pcre2_pattern) {
 501                p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
 502                if (!p->pcre2_match_data)
 503                        die("Couldn't allocate PCRE2 match data");
 504        } else {
 505                pcre2_get_error_message(error, errbuf, sizeof(errbuf));
 506                compile_regexp_failed(p, (const char *)&errbuf);
 507        }
 508
 509        pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
 510        if (p->pcre2_jit_on == 1) {
 511                jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
 512                if (jitret)
 513                        die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
 514                p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
 515                if (!p->pcre2_jit_stack)
 516                        die("Couldn't allocate PCRE2 JIT stack");
 517                p->pcre2_match_context = pcre2_match_context_create(NULL);
 518                if (!p->pcre2_match_context)
 519                        die("Couldn't allocate PCRE2 match context");
 520                pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
 521        } else if (p->pcre2_jit_on != 0) {
 522                die("BUG: The pcre2_jit_on variable should be 0 or 1, not %d",
 523                    p->pcre1_jit_on);
 524        }
 525}
 526
 527static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 528                regmatch_t *match, int eflags)
 529{
 530        int ret, flags = 0;
 531        PCRE2_SIZE *ovector;
 532        PCRE2_UCHAR errbuf[256];
 533
 534        if (eflags & REG_NOTBOL)
 535                flags |= PCRE2_NOTBOL;
 536
 537        if (p->pcre2_jit_on)
 538                ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 539                                      eol - line, 0, flags, p->pcre2_match_data,
 540                                      NULL);
 541        else
 542                ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 543                                  eol - line, 0, flags, p->pcre2_match_data,
 544                                  NULL);
 545
 546        if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 547                pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 548                die("%s failed with error code %d: %s",
 549                    (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 550                    errbuf);
 551        }
 552        if (ret > 0) {
 553                ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 554                ret = 0;
 555                match->rm_so = (int)ovector[0];
 556                match->rm_eo = (int)ovector[1];
 557        }
 558
 559        return ret;
 560}
 561
 562static void free_pcre2_pattern(struct grep_pat *p)
 563{
 564        pcre2_compile_context_free(p->pcre2_compile_context);
 565        pcre2_code_free(p->pcre2_pattern);
 566        pcre2_match_data_free(p->pcre2_match_data);
 567        pcre2_jit_stack_free(p->pcre2_jit_stack);
 568        pcre2_match_context_free(p->pcre2_match_context);
 569}
 570#else /* !USE_LIBPCRE2 */
 571static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 572{
 573        /*
 574         * Unreachable until USE_LIBPCRE2 becomes synonymous with
 575         * USE_LIBPCRE. See the sibling comment in
 576         * grep_set_pattern_type_option().
 577         */
 578        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 579}
 580
 581static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 582                regmatch_t *match, int eflags)
 583{
 584        return 1;
 585}
 586
 587static void free_pcre2_pattern(struct grep_pat *p)
 588{
 589}
 590#endif /* !USE_LIBPCRE2 */
 591
 592static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 593{
 594        struct strbuf sb = STRBUF_INIT;
 595        int err;
 596        int regflags = 0;
 597
 598        basic_regex_quote_buf(&sb, p->pattern);
 599        if (opt->ignore_case)
 600                regflags |= REG_ICASE;
 601        err = regcomp(&p->regexp, sb.buf, regflags);
 602        if (opt->debug)
 603                fprintf(stderr, "fixed %s\n", sb.buf);
 604        strbuf_release(&sb);
 605        if (err) {
 606                char errbuf[1024];
 607                regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 608                regfree(&p->regexp);
 609                compile_regexp_failed(p, errbuf);
 610        }
 611}
 612
 613static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 614{
 615        int ascii_only;
 616        int err;
 617        int regflags = REG_NEWLINE;
 618
 619        p->word_regexp = opt->word_regexp;
 620        p->ignore_case = opt->ignore_case;
 621        ascii_only     = !has_non_ascii(p->pattern);
 622
 623        /*
 624         * Even when -F (fixed) asks us to do a non-regexp search, we
 625         * may not be able to correctly case-fold when -i
 626         * (ignore-case) is asked (in which case, we'll synthesize a
 627         * regexp to match the pattern that matches regexp special
 628         * characters literally, while ignoring case differences).  On
 629         * the other hand, even without -F, if the pattern does not
 630         * have any regexp special characters and there is no need for
 631         * case-folding search, we can internally turn it into a
 632         * simple string match using kws.  p->fixed tells us if we
 633         * want to use kws.
 634         */
 635        if (opt->fixed ||
 636            has_null(p->pattern, p->patternlen) ||
 637            is_fixed(p->pattern, p->patternlen))
 638                p->fixed = !p->ignore_case || ascii_only;
 639
 640        if (p->fixed) {
 641                p->kws = kwsalloc(p->ignore_case ? tolower_trans_tbl : NULL);
 642                kwsincr(p->kws, p->pattern, p->patternlen);
 643                kwsprep(p->kws);
 644                return;
 645        } else if (opt->fixed) {
 646                /*
 647                 * We come here when the pattern has the non-ascii
 648                 * characters we cannot case-fold, and asked to
 649                 * ignore-case.
 650                 */
 651                compile_fixed_regexp(p, opt);
 652                return;
 653        }
 654
 655        if (opt->pcre2) {
 656                compile_pcre2_pattern(p, opt);
 657                return;
 658        }
 659
 660        if (opt->pcre1) {
 661                compile_pcre1_regexp(p, opt);
 662                return;
 663        }
 664
 665        if (p->ignore_case)
 666                regflags |= REG_ICASE;
 667        if (opt->extended_regexp_option)
 668                regflags |= REG_EXTENDED;
 669        err = regcomp(&p->regexp, p->pattern, regflags);
 670        if (err) {
 671                char errbuf[1024];
 672                regerror(err, &p->regexp, errbuf, 1024);
 673                regfree(&p->regexp);
 674                compile_regexp_failed(p, errbuf);
 675        }
 676}
 677
 678static struct grep_expr *compile_pattern_or(struct grep_pat **);
 679static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 680{
 681        struct grep_pat *p;
 682        struct grep_expr *x;
 683
 684        p = *list;
 685        if (!p)
 686                return NULL;
 687        switch (p->token) {
 688        case GREP_PATTERN: /* atom */
 689        case GREP_PATTERN_HEAD:
 690        case GREP_PATTERN_BODY:
 691                x = xcalloc(1, sizeof (struct grep_expr));
 692                x->node = GREP_NODE_ATOM;
 693                x->u.atom = p;
 694                *list = p->next;
 695                return x;
 696        case GREP_OPEN_PAREN:
 697                *list = p->next;
 698                x = compile_pattern_or(list);
 699                if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 700                        die("unmatched parenthesis");
 701                *list = (*list)->next;
 702                return x;
 703        default:
 704                return NULL;
 705        }
 706}
 707
 708static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 709{
 710        struct grep_pat *p;
 711        struct grep_expr *x;
 712
 713        p = *list;
 714        if (!p)
 715                return NULL;
 716        switch (p->token) {
 717        case GREP_NOT:
 718                if (!p->next)
 719                        die("--not not followed by pattern expression");
 720                *list = p->next;
 721                x = xcalloc(1, sizeof (struct grep_expr));
 722                x->node = GREP_NODE_NOT;
 723                x->u.unary = compile_pattern_not(list);
 724                if (!x->u.unary)
 725                        die("--not followed by non pattern expression");
 726                return x;
 727        default:
 728                return compile_pattern_atom(list);
 729        }
 730}
 731
 732static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 733{
 734        struct grep_pat *p;
 735        struct grep_expr *x, *y, *z;
 736
 737        x = compile_pattern_not(list);
 738        p = *list;
 739        if (p && p->token == GREP_AND) {
 740                if (!p->next)
 741                        die("--and not followed by pattern expression");
 742                *list = p->next;
 743                y = compile_pattern_and(list);
 744                if (!y)
 745                        die("--and not followed by pattern expression");
 746                z = xcalloc(1, sizeof (struct grep_expr));
 747                z->node = GREP_NODE_AND;
 748                z->u.binary.left = x;
 749                z->u.binary.right = y;
 750                return z;
 751        }
 752        return x;
 753}
 754
 755static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 756{
 757        struct grep_pat *p;
 758        struct grep_expr *x, *y, *z;
 759
 760        x = compile_pattern_and(list);
 761        p = *list;
 762        if (x && p && p->token != GREP_CLOSE_PAREN) {
 763                y = compile_pattern_or(list);
 764                if (!y)
 765                        die("not a pattern expression %s", p->pattern);
 766                z = xcalloc(1, sizeof (struct grep_expr));
 767                z->node = GREP_NODE_OR;
 768                z->u.binary.left = x;
 769                z->u.binary.right = y;
 770                return z;
 771        }
 772        return x;
 773}
 774
 775static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 776{
 777        return compile_pattern_or(list);
 778}
 779
 780static void indent(int in)
 781{
 782        while (in-- > 0)
 783                fputc(' ', stderr);
 784}
 785
 786static void dump_grep_pat(struct grep_pat *p)
 787{
 788        switch (p->token) {
 789        case GREP_AND: fprintf(stderr, "*and*"); break;
 790        case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 791        case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 792        case GREP_NOT: fprintf(stderr, "*not*"); break;
 793        case GREP_OR: fprintf(stderr, "*or*"); break;
 794
 795        case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 796        case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 797        case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 798        }
 799
 800        switch (p->token) {
 801        default: break;
 802        case GREP_PATTERN_HEAD:
 803                fprintf(stderr, "<head %d>", p->field); break;
 804        case GREP_PATTERN_BODY:
 805                fprintf(stderr, "<body>"); break;
 806        }
 807        switch (p->token) {
 808        default: break;
 809        case GREP_PATTERN_HEAD:
 810        case GREP_PATTERN_BODY:
 811        case GREP_PATTERN:
 812                fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 813                break;
 814        }
 815        fputc('\n', stderr);
 816}
 817
 818static void dump_grep_expression_1(struct grep_expr *x, int in)
 819{
 820        indent(in);
 821        switch (x->node) {
 822        case GREP_NODE_TRUE:
 823                fprintf(stderr, "true\n");
 824                break;
 825        case GREP_NODE_ATOM:
 826                dump_grep_pat(x->u.atom);
 827                break;
 828        case GREP_NODE_NOT:
 829                fprintf(stderr, "(not\n");
 830                dump_grep_expression_1(x->u.unary, in+1);
 831                indent(in);
 832                fprintf(stderr, ")\n");
 833                break;
 834        case GREP_NODE_AND:
 835                fprintf(stderr, "(and\n");
 836                dump_grep_expression_1(x->u.binary.left, in+1);
 837                dump_grep_expression_1(x->u.binary.right, in+1);
 838                indent(in);
 839                fprintf(stderr, ")\n");
 840                break;
 841        case GREP_NODE_OR:
 842                fprintf(stderr, "(or\n");
 843                dump_grep_expression_1(x->u.binary.left, in+1);
 844                dump_grep_expression_1(x->u.binary.right, in+1);
 845                indent(in);
 846                fprintf(stderr, ")\n");
 847                break;
 848        }
 849}
 850
 851static void dump_grep_expression(struct grep_opt *opt)
 852{
 853        struct grep_expr *x = opt->pattern_expression;
 854
 855        if (opt->all_match)
 856                fprintf(stderr, "[all-match]\n");
 857        dump_grep_expression_1(x, 0);
 858        fflush(NULL);
 859}
 860
 861static struct grep_expr *grep_true_expr(void)
 862{
 863        struct grep_expr *z = xcalloc(1, sizeof(*z));
 864        z->node = GREP_NODE_TRUE;
 865        return z;
 866}
 867
 868static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 869{
 870        struct grep_expr *z = xcalloc(1, sizeof(*z));
 871        z->node = GREP_NODE_OR;
 872        z->u.binary.left = left;
 873        z->u.binary.right = right;
 874        return z;
 875}
 876
 877static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 878{
 879        struct grep_pat *p;
 880        struct grep_expr *header_expr;
 881        struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 882        enum grep_header_field fld;
 883
 884        if (!opt->header_list)
 885                return NULL;
 886
 887        for (p = opt->header_list; p; p = p->next) {
 888                if (p->token != GREP_PATTERN_HEAD)
 889                        die("BUG: a non-header pattern in grep header list.");
 890                if (p->field < GREP_HEADER_FIELD_MIN ||
 891                    GREP_HEADER_FIELD_MAX <= p->field)
 892                        die("BUG: unknown header field %d", p->field);
 893                compile_regexp(p, opt);
 894        }
 895
 896        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 897                header_group[fld] = NULL;
 898
 899        for (p = opt->header_list; p; p = p->next) {
 900                struct grep_expr *h;
 901                struct grep_pat *pp = p;
 902
 903                h = compile_pattern_atom(&pp);
 904                if (!h || pp != p->next)
 905                        die("BUG: malformed header expr");
 906                if (!header_group[p->field]) {
 907                        header_group[p->field] = h;
 908                        continue;
 909                }
 910                header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 911        }
 912
 913        header_expr = NULL;
 914
 915        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 916                if (!header_group[fld])
 917                        continue;
 918                if (!header_expr)
 919                        header_expr = grep_true_expr();
 920                header_expr = grep_or_expr(header_group[fld], header_expr);
 921        }
 922        return header_expr;
 923}
 924
 925static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 926{
 927        struct grep_expr *z = x;
 928
 929        while (x) {
 930                assert(x->node == GREP_NODE_OR);
 931                if (x->u.binary.right &&
 932                    x->u.binary.right->node == GREP_NODE_TRUE) {
 933                        x->u.binary.right = y;
 934                        break;
 935                }
 936                x = x->u.binary.right;
 937        }
 938        return z;
 939}
 940
 941static void compile_grep_patterns_real(struct grep_opt *opt)
 942{
 943        struct grep_pat *p;
 944        struct grep_expr *header_expr = prep_header_patterns(opt);
 945
 946        for (p = opt->pattern_list; p; p = p->next) {
 947                switch (p->token) {
 948                case GREP_PATTERN: /* atom */
 949                case GREP_PATTERN_HEAD:
 950                case GREP_PATTERN_BODY:
 951                        compile_regexp(p, opt);
 952                        break;
 953                default:
 954                        opt->extended = 1;
 955                        break;
 956                }
 957        }
 958
 959        if (opt->all_match || header_expr)
 960                opt->extended = 1;
 961        else if (!opt->extended && !opt->debug)
 962                return;
 963
 964        p = opt->pattern_list;
 965        if (p)
 966                opt->pattern_expression = compile_pattern_expr(&p);
 967        if (p)
 968                die("incomplete pattern expression: %s", p->pattern);
 969
 970        if (!header_expr)
 971                return;
 972
 973        if (!opt->pattern_expression)
 974                opt->pattern_expression = header_expr;
 975        else if (opt->all_match)
 976                opt->pattern_expression = grep_splice_or(header_expr,
 977                                                         opt->pattern_expression);
 978        else
 979                opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 980                                                       header_expr);
 981        opt->all_match = 1;
 982}
 983
 984void compile_grep_patterns(struct grep_opt *opt)
 985{
 986        compile_grep_patterns_real(opt);
 987        if (opt->debug)
 988                dump_grep_expression(opt);
 989}
 990
 991static void free_pattern_expr(struct grep_expr *x)
 992{
 993        switch (x->node) {
 994        case GREP_NODE_TRUE:
 995        case GREP_NODE_ATOM:
 996                break;
 997        case GREP_NODE_NOT:
 998                free_pattern_expr(x->u.unary);
 999                break;
1000        case GREP_NODE_AND:
1001        case GREP_NODE_OR:
1002                free_pattern_expr(x->u.binary.left);
1003                free_pattern_expr(x->u.binary.right);
1004                break;
1005        }
1006        free(x);
1007}
1008
1009void free_grep_patterns(struct grep_opt *opt)
1010{
1011        struct grep_pat *p, *n;
1012
1013        for (p = opt->pattern_list; p; p = n) {
1014                n = p->next;
1015                switch (p->token) {
1016                case GREP_PATTERN: /* atom */
1017                case GREP_PATTERN_HEAD:
1018                case GREP_PATTERN_BODY:
1019                        if (p->kws)
1020                                kwsfree(p->kws);
1021                        else if (p->pcre1_regexp)
1022                                free_pcre1_regexp(p);
1023                        else if (p->pcre2_pattern)
1024                                free_pcre2_pattern(p);
1025                        else
1026                                regfree(&p->regexp);
1027                        free(p->pattern);
1028                        break;
1029                default:
1030                        break;
1031                }
1032                free(p);
1033        }
1034
1035        if (!opt->extended)
1036                return;
1037        free_pattern_expr(opt->pattern_expression);
1038}
1039
1040static char *end_of_line(char *cp, unsigned long *left)
1041{
1042        unsigned long l = *left;
1043        while (l && *cp != '\n') {
1044                l--;
1045                cp++;
1046        }
1047        *left = l;
1048        return cp;
1049}
1050
1051static int word_char(char ch)
1052{
1053        return isalnum(ch) || ch == '_';
1054}
1055
1056static void output_color(struct grep_opt *opt, const void *data, size_t size,
1057                         const char *color)
1058{
1059        if (want_color(opt->color) && color && color[0]) {
1060                opt->output(opt, color, strlen(color));
1061                opt->output(opt, data, size);
1062                opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1063        } else
1064                opt->output(opt, data, size);
1065}
1066
1067static void output_sep(struct grep_opt *opt, char sign)
1068{
1069        if (opt->null_following_name)
1070                opt->output(opt, "\0", 1);
1071        else
1072                output_color(opt, &sign, 1, opt->color_sep);
1073}
1074
1075static void show_name(struct grep_opt *opt, const char *name)
1076{
1077        output_color(opt, name, strlen(name), opt->color_filename);
1078        opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1079}
1080
1081static int fixmatch(struct grep_pat *p, char *line, char *eol,
1082                    regmatch_t *match)
1083{
1084        struct kwsmatch kwsm;
1085        size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
1086        if (offset == -1) {
1087                match->rm_so = match->rm_eo = -1;
1088                return REG_NOMATCH;
1089        } else {
1090                match->rm_so = offset;
1091                match->rm_eo = match->rm_so + kwsm.size[0];
1092                return 0;
1093        }
1094}
1095
1096static int patmatch(struct grep_pat *p, char *line, char *eol,
1097                    regmatch_t *match, int eflags)
1098{
1099        int hit;
1100
1101        if (p->fixed)
1102                hit = !fixmatch(p, line, eol, match);
1103        else if (p->pcre1_regexp)
1104                hit = !pcre1match(p, line, eol, match, eflags);
1105        else if (p->pcre2_pattern)
1106                hit = !pcre2match(p, line, eol, match, eflags);
1107        else
1108                hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1109                                   eflags);
1110
1111        return hit;
1112}
1113
1114static int strip_timestamp(char *bol, char **eol_p)
1115{
1116        char *eol = *eol_p;
1117        int ch;
1118
1119        while (bol < --eol) {
1120                if (*eol != '>')
1121                        continue;
1122                *eol_p = ++eol;
1123                ch = *eol;
1124                *eol = '\0';
1125                return ch;
1126        }
1127        return 0;
1128}
1129
1130static struct {
1131        const char *field;
1132        size_t len;
1133} header_field[] = {
1134        { "author ", 7 },
1135        { "committer ", 10 },
1136        { "reflog ", 7 },
1137};
1138
1139static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1140                             enum grep_context ctx,
1141                             regmatch_t *pmatch, int eflags)
1142{
1143        int hit = 0;
1144        int saved_ch = 0;
1145        const char *start = bol;
1146
1147        if ((p->token != GREP_PATTERN) &&
1148            ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1149                return 0;
1150
1151        if (p->token == GREP_PATTERN_HEAD) {
1152                const char *field;
1153                size_t len;
1154                assert(p->field < ARRAY_SIZE(header_field));
1155                field = header_field[p->field].field;
1156                len = header_field[p->field].len;
1157                if (strncmp(bol, field, len))
1158                        return 0;
1159                bol += len;
1160                switch (p->field) {
1161                case GREP_HEADER_AUTHOR:
1162                case GREP_HEADER_COMMITTER:
1163                        saved_ch = strip_timestamp(bol, &eol);
1164                        break;
1165                default:
1166                        break;
1167                }
1168        }
1169
1170 again:
1171        hit = patmatch(p, bol, eol, pmatch, eflags);
1172
1173        if (hit && p->word_regexp) {
1174                if ((pmatch[0].rm_so < 0) ||
1175                    (eol - bol) < pmatch[0].rm_so ||
1176                    (pmatch[0].rm_eo < 0) ||
1177                    (eol - bol) < pmatch[0].rm_eo)
1178                        die("regexp returned nonsense");
1179
1180                /* Match beginning must be either beginning of the
1181                 * line, or at word boundary (i.e. the last char must
1182                 * not be a word char).  Similarly, match end must be
1183                 * either end of the line, or at word boundary
1184                 * (i.e. the next char must not be a word char).
1185                 */
1186                if ( ((pmatch[0].rm_so == 0) ||
1187                      !word_char(bol[pmatch[0].rm_so-1])) &&
1188                     ((pmatch[0].rm_eo == (eol-bol)) ||
1189                      !word_char(bol[pmatch[0].rm_eo])) )
1190                        ;
1191                else
1192                        hit = 0;
1193
1194                /* Words consist of at least one character. */
1195                if (pmatch->rm_so == pmatch->rm_eo)
1196                        hit = 0;
1197
1198                if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1199                        /* There could be more than one match on the
1200                         * line, and the first match might not be
1201                         * strict word match.  But later ones could be!
1202                         * Forward to the next possible start, i.e. the
1203                         * next position following a non-word char.
1204                         */
1205                        bol = pmatch[0].rm_so + bol + 1;
1206                        while (word_char(bol[-1]) && bol < eol)
1207                                bol++;
1208                        eflags |= REG_NOTBOL;
1209                        if (bol < eol)
1210                                goto again;
1211                }
1212        }
1213        if (p->token == GREP_PATTERN_HEAD && saved_ch)
1214                *eol = saved_ch;
1215        if (hit) {
1216                pmatch[0].rm_so += bol - start;
1217                pmatch[0].rm_eo += bol - start;
1218        }
1219        return hit;
1220}
1221
1222static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1223                           enum grep_context ctx, int collect_hits)
1224{
1225        int h = 0;
1226        regmatch_t match;
1227
1228        if (!x)
1229                die("Not a valid grep expression");
1230        switch (x->node) {
1231        case GREP_NODE_TRUE:
1232                h = 1;
1233                break;
1234        case GREP_NODE_ATOM:
1235                h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1236                break;
1237        case GREP_NODE_NOT:
1238                h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1239                break;
1240        case GREP_NODE_AND:
1241                if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1242                        return 0;
1243                h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1244                break;
1245        case GREP_NODE_OR:
1246                if (!collect_hits)
1247                        return (match_expr_eval(x->u.binary.left,
1248                                                bol, eol, ctx, 0) ||
1249                                match_expr_eval(x->u.binary.right,
1250                                                bol, eol, ctx, 0));
1251                h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1252                x->u.binary.left->hit |= h;
1253                h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1254                break;
1255        default:
1256                die("Unexpected node type (internal error) %d", x->node);
1257        }
1258        if (collect_hits)
1259                x->hit |= h;
1260        return h;
1261}
1262
1263static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1264                      enum grep_context ctx, int collect_hits)
1265{
1266        struct grep_expr *x = opt->pattern_expression;
1267        return match_expr_eval(x, bol, eol, ctx, collect_hits);
1268}
1269
1270static int match_line(struct grep_opt *opt, char *bol, char *eol,
1271                      enum grep_context ctx, int collect_hits)
1272{
1273        struct grep_pat *p;
1274        regmatch_t match;
1275
1276        if (opt->extended)
1277                return match_expr(opt, bol, eol, ctx, collect_hits);
1278
1279        /* we do not call with collect_hits without being extended */
1280        for (p = opt->pattern_list; p; p = p->next) {
1281                if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1282                        return 1;
1283        }
1284        return 0;
1285}
1286
1287static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1288                              enum grep_context ctx,
1289                              regmatch_t *pmatch, int eflags)
1290{
1291        regmatch_t match;
1292
1293        if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1294                return 0;
1295        if (match.rm_so < 0 || match.rm_eo < 0)
1296                return 0;
1297        if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1298                if (match.rm_so > pmatch->rm_so)
1299                        return 1;
1300                if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1301                        return 1;
1302        }
1303        pmatch->rm_so = match.rm_so;
1304        pmatch->rm_eo = match.rm_eo;
1305        return 1;
1306}
1307
1308static int next_match(struct grep_opt *opt, char *bol, char *eol,
1309                      enum grep_context ctx, regmatch_t *pmatch, int eflags)
1310{
1311        struct grep_pat *p;
1312        int hit = 0;
1313
1314        pmatch->rm_so = pmatch->rm_eo = -1;
1315        if (bol < eol) {
1316                for (p = opt->pattern_list; p; p = p->next) {
1317                        switch (p->token) {
1318                        case GREP_PATTERN: /* atom */
1319                        case GREP_PATTERN_HEAD:
1320                        case GREP_PATTERN_BODY:
1321                                hit |= match_next_pattern(p, bol, eol, ctx,
1322                                                          pmatch, eflags);
1323                                break;
1324                        default:
1325                                break;
1326                        }
1327                }
1328        }
1329        return hit;
1330}
1331
1332static void show_line(struct grep_opt *opt, char *bol, char *eol,
1333                      const char *name, unsigned lno, char sign)
1334{
1335        int rest = eol - bol;
1336        const char *match_color, *line_color = NULL;
1337
1338        if (opt->file_break && opt->last_shown == 0) {
1339                if (opt->show_hunk_mark)
1340                        opt->output(opt, "\n", 1);
1341        } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1342                if (opt->last_shown == 0) {
1343                        if (opt->show_hunk_mark) {
1344                                output_color(opt, "--", 2, opt->color_sep);
1345                                opt->output(opt, "\n", 1);
1346                        }
1347                } else if (lno > opt->last_shown + 1) {
1348                        output_color(opt, "--", 2, opt->color_sep);
1349                        opt->output(opt, "\n", 1);
1350                }
1351        }
1352        if (opt->heading && opt->last_shown == 0) {
1353                output_color(opt, name, strlen(name), opt->color_filename);
1354                opt->output(opt, "\n", 1);
1355        }
1356        opt->last_shown = lno;
1357
1358        if (!opt->heading && opt->pathname) {
1359                output_color(opt, name, strlen(name), opt->color_filename);
1360                output_sep(opt, sign);
1361        }
1362        if (opt->linenum) {
1363                char buf[32];
1364                xsnprintf(buf, sizeof(buf), "%d", lno);
1365                output_color(opt, buf, strlen(buf), opt->color_lineno);
1366                output_sep(opt, sign);
1367        }
1368        if (opt->color) {
1369                regmatch_t match;
1370                enum grep_context ctx = GREP_CONTEXT_BODY;
1371                int ch = *eol;
1372                int eflags = 0;
1373
1374                if (sign == ':')
1375                        match_color = opt->color_match_selected;
1376                else
1377                        match_color = opt->color_match_context;
1378                if (sign == ':')
1379                        line_color = opt->color_selected;
1380                else if (sign == '-')
1381                        line_color = opt->color_context;
1382                else if (sign == '=')
1383                        line_color = opt->color_function;
1384                *eol = '\0';
1385                while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1386                        if (match.rm_so == match.rm_eo)
1387                                break;
1388
1389                        output_color(opt, bol, match.rm_so, line_color);
1390                        output_color(opt, bol + match.rm_so,
1391                                     match.rm_eo - match.rm_so, match_color);
1392                        bol += match.rm_eo;
1393                        rest -= match.rm_eo;
1394                        eflags = REG_NOTBOL;
1395                }
1396                *eol = ch;
1397        }
1398        output_color(opt, bol, rest, line_color);
1399        opt->output(opt, "\n", 1);
1400}
1401
1402#ifndef NO_PTHREADS
1403int grep_use_locks;
1404
1405/*
1406 * This lock protects access to the gitattributes machinery, which is
1407 * not thread-safe.
1408 */
1409pthread_mutex_t grep_attr_mutex;
1410
1411static inline void grep_attr_lock(void)
1412{
1413        if (grep_use_locks)
1414                pthread_mutex_lock(&grep_attr_mutex);
1415}
1416
1417static inline void grep_attr_unlock(void)
1418{
1419        if (grep_use_locks)
1420                pthread_mutex_unlock(&grep_attr_mutex);
1421}
1422
1423/*
1424 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1425 */
1426pthread_mutex_t grep_read_mutex;
1427
1428#else
1429#define grep_attr_lock()
1430#define grep_attr_unlock()
1431#endif
1432
1433static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1434{
1435        xdemitconf_t *xecfg = opt->priv;
1436        if (xecfg && !xecfg->find_func) {
1437                grep_source_load_driver(gs);
1438                if (gs->driver->funcname.pattern) {
1439                        const struct userdiff_funcname *pe = &gs->driver->funcname;
1440                        xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1441                } else {
1442                        xecfg = opt->priv = NULL;
1443                }
1444        }
1445
1446        if (xecfg) {
1447                char buf[1];
1448                return xecfg->find_func(bol, eol - bol, buf, 1,
1449                                        xecfg->find_func_priv) >= 0;
1450        }
1451
1452        if (bol == eol)
1453                return 0;
1454        if (isalpha(*bol) || *bol == '_' || *bol == '$')
1455                return 1;
1456        return 0;
1457}
1458
1459static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1460                               char *bol, unsigned lno)
1461{
1462        while (bol > gs->buf) {
1463                char *eol = --bol;
1464
1465                while (bol > gs->buf && bol[-1] != '\n')
1466                        bol--;
1467                lno--;
1468
1469                if (lno <= opt->last_shown)
1470                        break;
1471
1472                if (match_funcname(opt, gs, bol, eol)) {
1473                        show_line(opt, bol, eol, gs->name, lno, '=');
1474                        break;
1475                }
1476        }
1477}
1478
1479static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1480                             char *bol, char *end, unsigned lno)
1481{
1482        unsigned cur = lno, from = 1, funcname_lno = 0;
1483        int funcname_needed = !!opt->funcname;
1484
1485        if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1486                funcname_needed = 2;
1487
1488        if (opt->pre_context < lno)
1489                from = lno - opt->pre_context;
1490        if (from <= opt->last_shown)
1491                from = opt->last_shown + 1;
1492
1493        /* Rewind. */
1494        while (bol > gs->buf &&
1495               cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1496                char *eol = --bol;
1497
1498                while (bol > gs->buf && bol[-1] != '\n')
1499                        bol--;
1500                cur--;
1501                if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1502                        funcname_lno = cur;
1503                        funcname_needed = 0;
1504                }
1505        }
1506
1507        /* We need to look even further back to find a function signature. */
1508        if (opt->funcname && funcname_needed)
1509                show_funcname_line(opt, gs, bol, cur);
1510
1511        /* Back forward. */
1512        while (cur < lno) {
1513                char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1514
1515                while (*eol != '\n')
1516                        eol++;
1517                show_line(opt, bol, eol, gs->name, cur, sign);
1518                bol = eol + 1;
1519                cur++;
1520        }
1521}
1522
1523static int should_lookahead(struct grep_opt *opt)
1524{
1525        struct grep_pat *p;
1526
1527        if (opt->extended)
1528                return 0; /* punt for too complex stuff */
1529        if (opt->invert)
1530                return 0;
1531        for (p = opt->pattern_list; p; p = p->next) {
1532                if (p->token != GREP_PATTERN)
1533                        return 0; /* punt for "header only" and stuff */
1534        }
1535        return 1;
1536}
1537
1538static int look_ahead(struct grep_opt *opt,
1539                      unsigned long *left_p,
1540                      unsigned *lno_p,
1541                      char **bol_p)
1542{
1543        unsigned lno = *lno_p;
1544        char *bol = *bol_p;
1545        struct grep_pat *p;
1546        char *sp, *last_bol;
1547        regoff_t earliest = -1;
1548
1549        for (p = opt->pattern_list; p; p = p->next) {
1550                int hit;
1551                regmatch_t m;
1552
1553                hit = patmatch(p, bol, bol + *left_p, &m, 0);
1554                if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1555                        continue;
1556                if (earliest < 0 || m.rm_so < earliest)
1557                        earliest = m.rm_so;
1558        }
1559
1560        if (earliest < 0) {
1561                *bol_p = bol + *left_p;
1562                *left_p = 0;
1563                return 1;
1564        }
1565        for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1566                ; /* find the beginning of the line */
1567        last_bol = sp;
1568
1569        for (sp = bol; sp < last_bol; sp++) {
1570                if (*sp == '\n')
1571                        lno++;
1572        }
1573        *left_p -= last_bol - bol;
1574        *bol_p = last_bol;
1575        *lno_p = lno;
1576        return 0;
1577}
1578
1579static int fill_textconv_grep(struct userdiff_driver *driver,
1580                              struct grep_source *gs)
1581{
1582        struct diff_filespec *df;
1583        char *buf;
1584        size_t size;
1585
1586        if (!driver || !driver->textconv)
1587                return grep_source_load(gs);
1588
1589        /*
1590         * The textconv interface is intimately tied to diff_filespecs, so we
1591         * have to pretend to be one. If we could unify the grep_source
1592         * and diff_filespec structs, this mess could just go away.
1593         */
1594        df = alloc_filespec(gs->path);
1595        switch (gs->type) {
1596        case GREP_SOURCE_OID:
1597                fill_filespec(df, gs->identifier, 1, 0100644);
1598                break;
1599        case GREP_SOURCE_FILE:
1600                fill_filespec(df, &null_oid, 0, 0100644);
1601                break;
1602        default:
1603                die("BUG: attempt to textconv something without a path?");
1604        }
1605
1606        /*
1607         * fill_textconv is not remotely thread-safe; it may load objects
1608         * behind the scenes, and it modifies the global diff tempfile
1609         * structure.
1610         */
1611        grep_read_lock();
1612        size = fill_textconv(driver, df, &buf);
1613        grep_read_unlock();
1614        free_filespec(df);
1615
1616        /*
1617         * The normal fill_textconv usage by the diff machinery would just keep
1618         * the textconv'd buf separate from the diff_filespec. But much of the
1619         * grep code passes around a grep_source and assumes that its "buf"
1620         * pointer is the beginning of the thing we are searching. So let's
1621         * install our textconv'd version into the grep_source, taking care not
1622         * to leak any existing buffer.
1623         */
1624        grep_source_clear_data(gs);
1625        gs->buf = buf;
1626        gs->size = size;
1627
1628        return 0;
1629}
1630
1631static int is_empty_line(const char *bol, const char *eol)
1632{
1633        while (bol < eol && isspace(*bol))
1634                bol++;
1635        return bol == eol;
1636}
1637
1638static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1639{
1640        char *bol;
1641        char *peek_bol = NULL;
1642        unsigned long left;
1643        unsigned lno = 1;
1644        unsigned last_hit = 0;
1645        int binary_match_only = 0;
1646        unsigned count = 0;
1647        int try_lookahead = 0;
1648        int show_function = 0;
1649        struct userdiff_driver *textconv = NULL;
1650        enum grep_context ctx = GREP_CONTEXT_HEAD;
1651        xdemitconf_t xecfg;
1652
1653        if (!opt->output)
1654                opt->output = std_output;
1655
1656        if (opt->pre_context || opt->post_context || opt->file_break ||
1657            opt->funcbody) {
1658                /* Show hunk marks, except for the first file. */
1659                if (opt->last_shown)
1660                        opt->show_hunk_mark = 1;
1661                /*
1662                 * If we're using threads then we can't easily identify
1663                 * the first file.  Always put hunk marks in that case
1664                 * and skip the very first one later in work_done().
1665                 */
1666                if (opt->output != std_output)
1667                        opt->show_hunk_mark = 1;
1668        }
1669        opt->last_shown = 0;
1670
1671        if (opt->allow_textconv) {
1672                grep_source_load_driver(gs);
1673                /*
1674                 * We might set up the shared textconv cache data here, which
1675                 * is not thread-safe.
1676                 */
1677                grep_attr_lock();
1678                textconv = userdiff_get_textconv(gs->driver);
1679                grep_attr_unlock();
1680        }
1681
1682        /*
1683         * We know the result of a textconv is text, so we only have to care
1684         * about binary handling if we are not using it.
1685         */
1686        if (!textconv) {
1687                switch (opt->binary) {
1688                case GREP_BINARY_DEFAULT:
1689                        if (grep_source_is_binary(gs))
1690                                binary_match_only = 1;
1691                        break;
1692                case GREP_BINARY_NOMATCH:
1693                        if (grep_source_is_binary(gs))
1694                                return 0; /* Assume unmatch */
1695                        break;
1696                case GREP_BINARY_TEXT:
1697                        break;
1698                default:
1699                        die("BUG: unknown binary handling mode");
1700                }
1701        }
1702
1703        memset(&xecfg, 0, sizeof(xecfg));
1704        opt->priv = &xecfg;
1705
1706        try_lookahead = should_lookahead(opt);
1707
1708        if (fill_textconv_grep(textconv, gs) < 0)
1709                return 0;
1710
1711        bol = gs->buf;
1712        left = gs->size;
1713        while (left) {
1714                char *eol, ch;
1715                int hit;
1716
1717                /*
1718                 * look_ahead() skips quickly to the line that possibly
1719                 * has the next hit; don't call it if we need to do
1720                 * something more than just skipping the current line
1721                 * in response to an unmatch for the current line.  E.g.
1722                 * inside a post-context window, we will show the current
1723                 * line as a context around the previous hit when it
1724                 * doesn't hit.
1725                 */
1726                if (try_lookahead
1727                    && !(last_hit
1728                         && (show_function ||
1729                             lno <= last_hit + opt->post_context))
1730                    && look_ahead(opt, &left, &lno, &bol))
1731                        break;
1732                eol = end_of_line(bol, &left);
1733                ch = *eol;
1734                *eol = 0;
1735
1736                if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1737                        ctx = GREP_CONTEXT_BODY;
1738
1739                hit = match_line(opt, bol, eol, ctx, collect_hits);
1740                *eol = ch;
1741
1742                if (collect_hits)
1743                        goto next_line;
1744
1745                /* "grep -v -e foo -e bla" should list lines
1746                 * that do not have either, so inversion should
1747                 * be done outside.
1748                 */
1749                if (opt->invert)
1750                        hit = !hit;
1751                if (opt->unmatch_name_only) {
1752                        if (hit)
1753                                return 0;
1754                        goto next_line;
1755                }
1756                if (hit) {
1757                        count++;
1758                        if (opt->status_only)
1759                                return 1;
1760                        if (opt->name_only) {
1761                                show_name(opt, gs->name);
1762                                return 1;
1763                        }
1764                        if (opt->count)
1765                                goto next_line;
1766                        if (binary_match_only) {
1767                                opt->output(opt, "Binary file ", 12);
1768                                output_color(opt, gs->name, strlen(gs->name),
1769                                             opt->color_filename);
1770                                opt->output(opt, " matches\n", 9);
1771                                return 1;
1772                        }
1773                        /* Hit at this line.  If we haven't shown the
1774                         * pre-context lines, we would need to show them.
1775                         */
1776                        if (opt->pre_context || opt->funcbody)
1777                                show_pre_context(opt, gs, bol, eol, lno);
1778                        else if (opt->funcname)
1779                                show_funcname_line(opt, gs, bol, lno);
1780                        show_line(opt, bol, eol, gs->name, lno, ':');
1781                        last_hit = lno;
1782                        if (opt->funcbody)
1783                                show_function = 1;
1784                        goto next_line;
1785                }
1786                if (show_function && (!peek_bol || peek_bol < bol)) {
1787                        unsigned long peek_left = left;
1788                        char *peek_eol = eol;
1789
1790                        /*
1791                         * Trailing empty lines are not interesting.
1792                         * Peek past them to see if they belong to the
1793                         * body of the current function.
1794                         */
1795                        peek_bol = bol;
1796                        while (is_empty_line(peek_bol, peek_eol)) {
1797                                peek_bol = peek_eol + 1;
1798                                peek_eol = end_of_line(peek_bol, &peek_left);
1799                        }
1800
1801                        if (match_funcname(opt, gs, peek_bol, peek_eol))
1802                                show_function = 0;
1803                }
1804                if (show_function ||
1805                    (last_hit && lno <= last_hit + opt->post_context)) {
1806                        /* If the last hit is within the post context,
1807                         * we need to show this line.
1808                         */
1809                        show_line(opt, bol, eol, gs->name, lno, '-');
1810                }
1811
1812        next_line:
1813                bol = eol + 1;
1814                if (!left)
1815                        break;
1816                left--;
1817                lno++;
1818        }
1819
1820        if (collect_hits)
1821                return 0;
1822
1823        if (opt->status_only)
1824                return opt->unmatch_name_only;
1825        if (opt->unmatch_name_only) {
1826                /* We did not see any hit, so we want to show this */
1827                show_name(opt, gs->name);
1828                return 1;
1829        }
1830
1831        xdiff_clear_find_func(&xecfg);
1832        opt->priv = NULL;
1833
1834        /* NEEDSWORK:
1835         * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1836         * which feels mostly useless but sometimes useful.  Maybe
1837         * make it another option?  For now suppress them.
1838         */
1839        if (opt->count && count) {
1840                char buf[32];
1841                if (opt->pathname) {
1842                        output_color(opt, gs->name, strlen(gs->name),
1843                                     opt->color_filename);
1844                        output_sep(opt, ':');
1845                }
1846                xsnprintf(buf, sizeof(buf), "%u\n", count);
1847                opt->output(opt, buf, strlen(buf));
1848                return 1;
1849        }
1850        return !!last_hit;
1851}
1852
1853static void clr_hit_marker(struct grep_expr *x)
1854{
1855        /* All-hit markers are meaningful only at the very top level
1856         * OR node.
1857         */
1858        while (1) {
1859                x->hit = 0;
1860                if (x->node != GREP_NODE_OR)
1861                        return;
1862                x->u.binary.left->hit = 0;
1863                x = x->u.binary.right;
1864        }
1865}
1866
1867static int chk_hit_marker(struct grep_expr *x)
1868{
1869        /* Top level nodes have hit markers.  See if they all are hits */
1870        while (1) {
1871                if (x->node != GREP_NODE_OR)
1872                        return x->hit;
1873                if (!x->u.binary.left->hit)
1874                        return 0;
1875                x = x->u.binary.right;
1876        }
1877}
1878
1879int grep_source(struct grep_opt *opt, struct grep_source *gs)
1880{
1881        /*
1882         * we do not have to do the two-pass grep when we do not check
1883         * buffer-wide "all-match".
1884         */
1885        if (!opt->all_match)
1886                return grep_source_1(opt, gs, 0);
1887
1888        /* Otherwise the toplevel "or" terms hit a bit differently.
1889         * We first clear hit markers from them.
1890         */
1891        clr_hit_marker(opt->pattern_expression);
1892        grep_source_1(opt, gs, 1);
1893
1894        if (!chk_hit_marker(opt->pattern_expression))
1895                return 0;
1896
1897        return grep_source_1(opt, gs, 0);
1898}
1899
1900int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1901{
1902        struct grep_source gs;
1903        int r;
1904
1905        grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1906        gs.buf = buf;
1907        gs.size = size;
1908
1909        r = grep_source(opt, &gs);
1910
1911        grep_source_clear(&gs);
1912        return r;
1913}
1914
1915void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1916                      const char *name, const char *path,
1917                      const void *identifier)
1918{
1919        gs->type = type;
1920        gs->name = xstrdup_or_null(name);
1921        gs->path = xstrdup_or_null(path);
1922        gs->buf = NULL;
1923        gs->size = 0;
1924        gs->driver = NULL;
1925
1926        switch (type) {
1927        case GREP_SOURCE_FILE:
1928                gs->identifier = xstrdup(identifier);
1929                break;
1930        case GREP_SOURCE_OID:
1931                gs->identifier = oiddup(identifier);
1932                break;
1933        case GREP_SOURCE_BUF:
1934                gs->identifier = NULL;
1935                break;
1936        }
1937}
1938
1939void grep_source_clear(struct grep_source *gs)
1940{
1941        FREE_AND_NULL(gs->name);
1942        FREE_AND_NULL(gs->path);
1943        FREE_AND_NULL(gs->identifier);
1944        grep_source_clear_data(gs);
1945}
1946
1947void grep_source_clear_data(struct grep_source *gs)
1948{
1949        switch (gs->type) {
1950        case GREP_SOURCE_FILE:
1951        case GREP_SOURCE_OID:
1952                FREE_AND_NULL(gs->buf);
1953                gs->size = 0;
1954                break;
1955        case GREP_SOURCE_BUF:
1956                /* leave user-provided buf intact */
1957                break;
1958        }
1959}
1960
1961static int grep_source_load_oid(struct grep_source *gs)
1962{
1963        enum object_type type;
1964
1965        grep_read_lock();
1966        gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1967        grep_read_unlock();
1968
1969        if (!gs->buf)
1970                return error(_("'%s': unable to read %s"),
1971                             gs->name,
1972                             oid_to_hex(gs->identifier));
1973        return 0;
1974}
1975
1976static int grep_source_load_file(struct grep_source *gs)
1977{
1978        const char *filename = gs->identifier;
1979        struct stat st;
1980        char *data;
1981        size_t size;
1982        int i;
1983
1984        if (lstat(filename, &st) < 0) {
1985        err_ret:
1986                if (errno != ENOENT)
1987                        error_errno(_("failed to stat '%s'"), filename);
1988                return -1;
1989        }
1990        if (!S_ISREG(st.st_mode))
1991                return -1;
1992        size = xsize_t(st.st_size);
1993        i = open(filename, O_RDONLY);
1994        if (i < 0)
1995                goto err_ret;
1996        data = xmallocz(size);
1997        if (st.st_size != read_in_full(i, data, size)) {
1998                error_errno(_("'%s': short read"), filename);
1999                close(i);
2000                free(data);
2001                return -1;
2002        }
2003        close(i);
2004
2005        gs->buf = data;
2006        gs->size = size;
2007        return 0;
2008}
2009
2010static int grep_source_load(struct grep_source *gs)
2011{
2012        if (gs->buf)
2013                return 0;
2014
2015        switch (gs->type) {
2016        case GREP_SOURCE_FILE:
2017                return grep_source_load_file(gs);
2018        case GREP_SOURCE_OID:
2019                return grep_source_load_oid(gs);
2020        case GREP_SOURCE_BUF:
2021                return gs->buf ? 0 : -1;
2022        }
2023        die("BUG: invalid grep_source type to load");
2024}
2025
2026void grep_source_load_driver(struct grep_source *gs)
2027{
2028        if (gs->driver)
2029                return;
2030
2031        grep_attr_lock();
2032        if (gs->path)
2033                gs->driver = userdiff_find_by_path(gs->path);
2034        if (!gs->driver)
2035                gs->driver = userdiff_find_by_name("default");
2036        grep_attr_unlock();
2037}
2038
2039static int grep_source_is_binary(struct grep_source *gs)
2040{
2041        grep_source_load_driver(gs);
2042        if (gs->driver->binary != -1)
2043                return gs->driver->binary;
2044
2045        if (!grep_source_load(gs))
2046                return buffer_is_binary(gs->buf, gs->size);
2047
2048        return 0;
2049}