Andrew's git / gitweb.git / grep.c
?
grep.con commit Merge branch 'ks/submodule-doc-updates' into maint (60736db)
   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, GIT_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        int patinforet;
 481        size_t jitsizearg;
 482
 483        assert(opt->pcre2);
 484
 485        p->pcre2_compile_context = NULL;
 486
 487        if (opt->ignore_case) {
 488                if (has_non_ascii(p->pattern)) {
 489                        character_tables = pcre2_maketables(NULL);
 490                        p->pcre2_compile_context = pcre2_compile_context_create(NULL);
 491                        pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
 492                }
 493                options |= PCRE2_CASELESS;
 494        }
 495        if (is_utf8_locale() && has_non_ascii(p->pattern))
 496                options |= PCRE2_UTF;
 497
 498        p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
 499                                         p->patternlen, options, &error, &erroffset,
 500                                         p->pcre2_compile_context);
 501
 502        if (p->pcre2_pattern) {
 503                p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
 504                if (!p->pcre2_match_data)
 505                        die("Couldn't allocate PCRE2 match data");
 506        } else {
 507                pcre2_get_error_message(error, errbuf, sizeof(errbuf));
 508                compile_regexp_failed(p, (const char *)&errbuf);
 509        }
 510
 511        pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
 512        if (p->pcre2_jit_on == 1) {
 513                jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
 514                if (jitret)
 515                        die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
 516
 517                /*
 518                 * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
 519                 * tells us whether the library itself supports JIT,
 520                 * but to see whether we're going to be actually using
 521                 * JIT we need to extract PCRE2_INFO_JITSIZE from the
 522                 * pattern *after* we do pcre2_jit_compile() above.
 523                 *
 524                 * This is because if the pattern contains the
 525                 * (*NO_JIT) verb (see pcre2syntax(3))
 526                 * pcre2_jit_compile() will exit early with 0. If we
 527                 * then proceed to call pcre2_jit_match() further down
 528                 * the line instead of pcre2_match() we'll either
 529                 * segfault (pre PCRE 10.31) or run into a fatal error
 530                 * (post PCRE2 10.31)
 531                 */
 532                patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
 533                if (patinforet)
 534                        BUG("pcre2_pattern_info() failed: %d", patinforet);
 535                if (jitsizearg == 0) {
 536                        p->pcre2_jit_on = 0;
 537                        return;
 538                }
 539
 540                p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
 541                if (!p->pcre2_jit_stack)
 542                        die("Couldn't allocate PCRE2 JIT stack");
 543                p->pcre2_match_context = pcre2_match_context_create(NULL);
 544                if (!p->pcre2_match_context)
 545                        die("Couldn't allocate PCRE2 match context");
 546                pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
 547        } else if (p->pcre2_jit_on != 0) {
 548                die("BUG: The pcre2_jit_on variable should be 0 or 1, not %d",
 549                    p->pcre1_jit_on);
 550        }
 551}
 552
 553static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 554                regmatch_t *match, int eflags)
 555{
 556        int ret, flags = 0;
 557        PCRE2_SIZE *ovector;
 558        PCRE2_UCHAR errbuf[256];
 559
 560        if (eflags & REG_NOTBOL)
 561                flags |= PCRE2_NOTBOL;
 562
 563        if (p->pcre2_jit_on)
 564                ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 565                                      eol - line, 0, flags, p->pcre2_match_data,
 566                                      NULL);
 567        else
 568                ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 569                                  eol - line, 0, flags, p->pcre2_match_data,
 570                                  NULL);
 571
 572        if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 573                pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 574                die("%s failed with error code %d: %s",
 575                    (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 576                    errbuf);
 577        }
 578        if (ret > 0) {
 579                ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 580                ret = 0;
 581                match->rm_so = (int)ovector[0];
 582                match->rm_eo = (int)ovector[1];
 583        }
 584
 585        return ret;
 586}
 587
 588static void free_pcre2_pattern(struct grep_pat *p)
 589{
 590        pcre2_compile_context_free(p->pcre2_compile_context);
 591        pcre2_code_free(p->pcre2_pattern);
 592        pcre2_match_data_free(p->pcre2_match_data);
 593        pcre2_jit_stack_free(p->pcre2_jit_stack);
 594        pcre2_match_context_free(p->pcre2_match_context);
 595}
 596#else /* !USE_LIBPCRE2 */
 597static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 598{
 599        /*
 600         * Unreachable until USE_LIBPCRE2 becomes synonymous with
 601         * USE_LIBPCRE. See the sibling comment in
 602         * grep_set_pattern_type_option().
 603         */
 604        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 605}
 606
 607static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 608                regmatch_t *match, int eflags)
 609{
 610        return 1;
 611}
 612
 613static void free_pcre2_pattern(struct grep_pat *p)
 614{
 615}
 616#endif /* !USE_LIBPCRE2 */
 617
 618static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 619{
 620        struct strbuf sb = STRBUF_INIT;
 621        int err;
 622        int regflags = 0;
 623
 624        basic_regex_quote_buf(&sb, p->pattern);
 625        if (opt->ignore_case)
 626                regflags |= REG_ICASE;
 627        err = regcomp(&p->regexp, sb.buf, regflags);
 628        if (opt->debug)
 629                fprintf(stderr, "fixed %s\n", sb.buf);
 630        strbuf_release(&sb);
 631        if (err) {
 632                char errbuf[1024];
 633                regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 634                regfree(&p->regexp);
 635                compile_regexp_failed(p, errbuf);
 636        }
 637}
 638
 639static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 640{
 641        int ascii_only;
 642        int err;
 643        int regflags = REG_NEWLINE;
 644
 645        p->word_regexp = opt->word_regexp;
 646        p->ignore_case = opt->ignore_case;
 647        ascii_only     = !has_non_ascii(p->pattern);
 648
 649        /*
 650         * Even when -F (fixed) asks us to do a non-regexp search, we
 651         * may not be able to correctly case-fold when -i
 652         * (ignore-case) is asked (in which case, we'll synthesize a
 653         * regexp to match the pattern that matches regexp special
 654         * characters literally, while ignoring case differences).  On
 655         * the other hand, even without -F, if the pattern does not
 656         * have any regexp special characters and there is no need for
 657         * case-folding search, we can internally turn it into a
 658         * simple string match using kws.  p->fixed tells us if we
 659         * want to use kws.
 660         */
 661        if (opt->fixed ||
 662            has_null(p->pattern, p->patternlen) ||
 663            is_fixed(p->pattern, p->patternlen))
 664                p->fixed = !p->ignore_case || ascii_only;
 665
 666        if (p->fixed) {
 667                p->kws = kwsalloc(p->ignore_case ? tolower_trans_tbl : NULL);
 668                kwsincr(p->kws, p->pattern, p->patternlen);
 669                kwsprep(p->kws);
 670                return;
 671        } else if (opt->fixed) {
 672                /*
 673                 * We come here when the pattern has the non-ascii
 674                 * characters we cannot case-fold, and asked to
 675                 * ignore-case.
 676                 */
 677                compile_fixed_regexp(p, opt);
 678                return;
 679        }
 680
 681        if (opt->pcre2) {
 682                compile_pcre2_pattern(p, opt);
 683                return;
 684        }
 685
 686        if (opt->pcre1) {
 687                compile_pcre1_regexp(p, opt);
 688                return;
 689        }
 690
 691        if (p->ignore_case)
 692                regflags |= REG_ICASE;
 693        if (opt->extended_regexp_option)
 694                regflags |= REG_EXTENDED;
 695        err = regcomp(&p->regexp, p->pattern, regflags);
 696        if (err) {
 697                char errbuf[1024];
 698                regerror(err, &p->regexp, errbuf, 1024);
 699                regfree(&p->regexp);
 700                compile_regexp_failed(p, errbuf);
 701        }
 702}
 703
 704static struct grep_expr *compile_pattern_or(struct grep_pat **);
 705static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 706{
 707        struct grep_pat *p;
 708        struct grep_expr *x;
 709
 710        p = *list;
 711        if (!p)
 712                return NULL;
 713        switch (p->token) {
 714        case GREP_PATTERN: /* atom */
 715        case GREP_PATTERN_HEAD:
 716        case GREP_PATTERN_BODY:
 717                x = xcalloc(1, sizeof (struct grep_expr));
 718                x->node = GREP_NODE_ATOM;
 719                x->u.atom = p;
 720                *list = p->next;
 721                return x;
 722        case GREP_OPEN_PAREN:
 723                *list = p->next;
 724                x = compile_pattern_or(list);
 725                if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 726                        die("unmatched parenthesis");
 727                *list = (*list)->next;
 728                return x;
 729        default:
 730                return NULL;
 731        }
 732}
 733
 734static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 735{
 736        struct grep_pat *p;
 737        struct grep_expr *x;
 738
 739        p = *list;
 740        if (!p)
 741                return NULL;
 742        switch (p->token) {
 743        case GREP_NOT:
 744                if (!p->next)
 745                        die("--not not followed by pattern expression");
 746                *list = p->next;
 747                x = xcalloc(1, sizeof (struct grep_expr));
 748                x->node = GREP_NODE_NOT;
 749                x->u.unary = compile_pattern_not(list);
 750                if (!x->u.unary)
 751                        die("--not followed by non pattern expression");
 752                return x;
 753        default:
 754                return compile_pattern_atom(list);
 755        }
 756}
 757
 758static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 759{
 760        struct grep_pat *p;
 761        struct grep_expr *x, *y, *z;
 762
 763        x = compile_pattern_not(list);
 764        p = *list;
 765        if (p && p->token == GREP_AND) {
 766                if (!p->next)
 767                        die("--and not followed by pattern expression");
 768                *list = p->next;
 769                y = compile_pattern_and(list);
 770                if (!y)
 771                        die("--and not followed by pattern expression");
 772                z = xcalloc(1, sizeof (struct grep_expr));
 773                z->node = GREP_NODE_AND;
 774                z->u.binary.left = x;
 775                z->u.binary.right = y;
 776                return z;
 777        }
 778        return x;
 779}
 780
 781static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 782{
 783        struct grep_pat *p;
 784        struct grep_expr *x, *y, *z;
 785
 786        x = compile_pattern_and(list);
 787        p = *list;
 788        if (x && p && p->token != GREP_CLOSE_PAREN) {
 789                y = compile_pattern_or(list);
 790                if (!y)
 791                        die("not a pattern expression %s", p->pattern);
 792                z = xcalloc(1, sizeof (struct grep_expr));
 793                z->node = GREP_NODE_OR;
 794                z->u.binary.left = x;
 795                z->u.binary.right = y;
 796                return z;
 797        }
 798        return x;
 799}
 800
 801static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 802{
 803        return compile_pattern_or(list);
 804}
 805
 806static void indent(int in)
 807{
 808        while (in-- > 0)
 809                fputc(' ', stderr);
 810}
 811
 812static void dump_grep_pat(struct grep_pat *p)
 813{
 814        switch (p->token) {
 815        case GREP_AND: fprintf(stderr, "*and*"); break;
 816        case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 817        case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 818        case GREP_NOT: fprintf(stderr, "*not*"); break;
 819        case GREP_OR: fprintf(stderr, "*or*"); break;
 820
 821        case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 822        case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 823        case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 824        }
 825
 826        switch (p->token) {
 827        default: break;
 828        case GREP_PATTERN_HEAD:
 829                fprintf(stderr, "<head %d>", p->field); break;
 830        case GREP_PATTERN_BODY:
 831                fprintf(stderr, "<body>"); break;
 832        }
 833        switch (p->token) {
 834        default: break;
 835        case GREP_PATTERN_HEAD:
 836        case GREP_PATTERN_BODY:
 837        case GREP_PATTERN:
 838                fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 839                break;
 840        }
 841        fputc('\n', stderr);
 842}
 843
 844static void dump_grep_expression_1(struct grep_expr *x, int in)
 845{
 846        indent(in);
 847        switch (x->node) {
 848        case GREP_NODE_TRUE:
 849                fprintf(stderr, "true\n");
 850                break;
 851        case GREP_NODE_ATOM:
 852                dump_grep_pat(x->u.atom);
 853                break;
 854        case GREP_NODE_NOT:
 855                fprintf(stderr, "(not\n");
 856                dump_grep_expression_1(x->u.unary, in+1);
 857                indent(in);
 858                fprintf(stderr, ")\n");
 859                break;
 860        case GREP_NODE_AND:
 861                fprintf(stderr, "(and\n");
 862                dump_grep_expression_1(x->u.binary.left, in+1);
 863                dump_grep_expression_1(x->u.binary.right, in+1);
 864                indent(in);
 865                fprintf(stderr, ")\n");
 866                break;
 867        case GREP_NODE_OR:
 868                fprintf(stderr, "(or\n");
 869                dump_grep_expression_1(x->u.binary.left, in+1);
 870                dump_grep_expression_1(x->u.binary.right, in+1);
 871                indent(in);
 872                fprintf(stderr, ")\n");
 873                break;
 874        }
 875}
 876
 877static void dump_grep_expression(struct grep_opt *opt)
 878{
 879        struct grep_expr *x = opt->pattern_expression;
 880
 881        if (opt->all_match)
 882                fprintf(stderr, "[all-match]\n");
 883        dump_grep_expression_1(x, 0);
 884        fflush(NULL);
 885}
 886
 887static struct grep_expr *grep_true_expr(void)
 888{
 889        struct grep_expr *z = xcalloc(1, sizeof(*z));
 890        z->node = GREP_NODE_TRUE;
 891        return z;
 892}
 893
 894static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 895{
 896        struct grep_expr *z = xcalloc(1, sizeof(*z));
 897        z->node = GREP_NODE_OR;
 898        z->u.binary.left = left;
 899        z->u.binary.right = right;
 900        return z;
 901}
 902
 903static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 904{
 905        struct grep_pat *p;
 906        struct grep_expr *header_expr;
 907        struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 908        enum grep_header_field fld;
 909
 910        if (!opt->header_list)
 911                return NULL;
 912
 913        for (p = opt->header_list; p; p = p->next) {
 914                if (p->token != GREP_PATTERN_HEAD)
 915                        die("BUG: a non-header pattern in grep header list.");
 916                if (p->field < GREP_HEADER_FIELD_MIN ||
 917                    GREP_HEADER_FIELD_MAX <= p->field)
 918                        die("BUG: unknown header field %d", p->field);
 919                compile_regexp(p, opt);
 920        }
 921
 922        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 923                header_group[fld] = NULL;
 924
 925        for (p = opt->header_list; p; p = p->next) {
 926                struct grep_expr *h;
 927                struct grep_pat *pp = p;
 928
 929                h = compile_pattern_atom(&pp);
 930                if (!h || pp != p->next)
 931                        die("BUG: malformed header expr");
 932                if (!header_group[p->field]) {
 933                        header_group[p->field] = h;
 934                        continue;
 935                }
 936                header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 937        }
 938
 939        header_expr = NULL;
 940
 941        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 942                if (!header_group[fld])
 943                        continue;
 944                if (!header_expr)
 945                        header_expr = grep_true_expr();
 946                header_expr = grep_or_expr(header_group[fld], header_expr);
 947        }
 948        return header_expr;
 949}
 950
 951static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 952{
 953        struct grep_expr *z = x;
 954
 955        while (x) {
 956                assert(x->node == GREP_NODE_OR);
 957                if (x->u.binary.right &&
 958                    x->u.binary.right->node == GREP_NODE_TRUE) {
 959                        x->u.binary.right = y;
 960                        break;
 961                }
 962                x = x->u.binary.right;
 963        }
 964        return z;
 965}
 966
 967static void compile_grep_patterns_real(struct grep_opt *opt)
 968{
 969        struct grep_pat *p;
 970        struct grep_expr *header_expr = prep_header_patterns(opt);
 971
 972        for (p = opt->pattern_list; p; p = p->next) {
 973                switch (p->token) {
 974                case GREP_PATTERN: /* atom */
 975                case GREP_PATTERN_HEAD:
 976                case GREP_PATTERN_BODY:
 977                        compile_regexp(p, opt);
 978                        break;
 979                default:
 980                        opt->extended = 1;
 981                        break;
 982                }
 983        }
 984
 985        if (opt->all_match || header_expr)
 986                opt->extended = 1;
 987        else if (!opt->extended && !opt->debug)
 988                return;
 989
 990        p = opt->pattern_list;
 991        if (p)
 992                opt->pattern_expression = compile_pattern_expr(&p);
 993        if (p)
 994                die("incomplete pattern expression: %s", p->pattern);
 995
 996        if (!header_expr)
 997                return;
 998
 999        if (!opt->pattern_expression)
1000                opt->pattern_expression = header_expr;
1001        else if (opt->all_match)
1002                opt->pattern_expression = grep_splice_or(header_expr,
1003                                                         opt->pattern_expression);
1004        else
1005                opt->pattern_expression = grep_or_expr(opt->pattern_expression,
1006                                                       header_expr);
1007        opt->all_match = 1;
1008}
1009
1010void compile_grep_patterns(struct grep_opt *opt)
1011{
1012        compile_grep_patterns_real(opt);
1013        if (opt->debug)
1014                dump_grep_expression(opt);
1015}
1016
1017static void free_pattern_expr(struct grep_expr *x)
1018{
1019        switch (x->node) {
1020        case GREP_NODE_TRUE:
1021        case GREP_NODE_ATOM:
1022                break;
1023        case GREP_NODE_NOT:
1024                free_pattern_expr(x->u.unary);
1025                break;
1026        case GREP_NODE_AND:
1027        case GREP_NODE_OR:
1028                free_pattern_expr(x->u.binary.left);
1029                free_pattern_expr(x->u.binary.right);
1030                break;
1031        }
1032        free(x);
1033}
1034
1035void free_grep_patterns(struct grep_opt *opt)
1036{
1037        struct grep_pat *p, *n;
1038
1039        for (p = opt->pattern_list; p; p = n) {
1040                n = p->next;
1041                switch (p->token) {
1042                case GREP_PATTERN: /* atom */
1043                case GREP_PATTERN_HEAD:
1044                case GREP_PATTERN_BODY:
1045                        if (p->kws)
1046                                kwsfree(p->kws);
1047                        else if (p->pcre1_regexp)
1048                                free_pcre1_regexp(p);
1049                        else if (p->pcre2_pattern)
1050                                free_pcre2_pattern(p);
1051                        else
1052                                regfree(&p->regexp);
1053                        free(p->pattern);
1054                        break;
1055                default:
1056                        break;
1057                }
1058                free(p);
1059        }
1060
1061        if (!opt->extended)
1062                return;
1063        free_pattern_expr(opt->pattern_expression);
1064}
1065
1066static char *end_of_line(char *cp, unsigned long *left)
1067{
1068        unsigned long l = *left;
1069        while (l && *cp != '\n') {
1070                l--;
1071                cp++;
1072        }
1073        *left = l;
1074        return cp;
1075}
1076
1077static int word_char(char ch)
1078{
1079        return isalnum(ch) || ch == '_';
1080}
1081
1082static void output_color(struct grep_opt *opt, const void *data, size_t size,
1083                         const char *color)
1084{
1085        if (want_color(opt->color) && color && color[0]) {
1086                opt->output(opt, color, strlen(color));
1087                opt->output(opt, data, size);
1088                opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1089        } else
1090                opt->output(opt, data, size);
1091}
1092
1093static void output_sep(struct grep_opt *opt, char sign)
1094{
1095        if (opt->null_following_name)
1096                opt->output(opt, "\0", 1);
1097        else
1098                output_color(opt, &sign, 1, opt->color_sep);
1099}
1100
1101static void show_name(struct grep_opt *opt, const char *name)
1102{
1103        output_color(opt, name, strlen(name), opt->color_filename);
1104        opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1105}
1106
1107static int fixmatch(struct grep_pat *p, char *line, char *eol,
1108                    regmatch_t *match)
1109{
1110        struct kwsmatch kwsm;
1111        size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
1112        if (offset == -1) {
1113                match->rm_so = match->rm_eo = -1;
1114                return REG_NOMATCH;
1115        } else {
1116                match->rm_so = offset;
1117                match->rm_eo = match->rm_so + kwsm.size[0];
1118                return 0;
1119        }
1120}
1121
1122static int patmatch(struct grep_pat *p, char *line, char *eol,
1123                    regmatch_t *match, int eflags)
1124{
1125        int hit;
1126
1127        if (p->fixed)
1128                hit = !fixmatch(p, line, eol, match);
1129        else if (p->pcre1_regexp)
1130                hit = !pcre1match(p, line, eol, match, eflags);
1131        else if (p->pcre2_pattern)
1132                hit = !pcre2match(p, line, eol, match, eflags);
1133        else
1134                hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1135                                   eflags);
1136
1137        return hit;
1138}
1139
1140static int strip_timestamp(char *bol, char **eol_p)
1141{
1142        char *eol = *eol_p;
1143        int ch;
1144
1145        while (bol < --eol) {
1146                if (*eol != '>')
1147                        continue;
1148                *eol_p = ++eol;
1149                ch = *eol;
1150                *eol = '\0';
1151                return ch;
1152        }
1153        return 0;
1154}
1155
1156static struct {
1157        const char *field;
1158        size_t len;
1159} header_field[] = {
1160        { "author ", 7 },
1161        { "committer ", 10 },
1162        { "reflog ", 7 },
1163};
1164
1165static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1166                             enum grep_context ctx,
1167                             regmatch_t *pmatch, int eflags)
1168{
1169        int hit = 0;
1170        int saved_ch = 0;
1171        const char *start = bol;
1172
1173        if ((p->token != GREP_PATTERN) &&
1174            ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1175                return 0;
1176
1177        if (p->token == GREP_PATTERN_HEAD) {
1178                const char *field;
1179                size_t len;
1180                assert(p->field < ARRAY_SIZE(header_field));
1181                field = header_field[p->field].field;
1182                len = header_field[p->field].len;
1183                if (strncmp(bol, field, len))
1184                        return 0;
1185                bol += len;
1186                switch (p->field) {
1187                case GREP_HEADER_AUTHOR:
1188                case GREP_HEADER_COMMITTER:
1189                        saved_ch = strip_timestamp(bol, &eol);
1190                        break;
1191                default:
1192                        break;
1193                }
1194        }
1195
1196 again:
1197        hit = patmatch(p, bol, eol, pmatch, eflags);
1198
1199        if (hit && p->word_regexp) {
1200                if ((pmatch[0].rm_so < 0) ||
1201                    (eol - bol) < pmatch[0].rm_so ||
1202                    (pmatch[0].rm_eo < 0) ||
1203                    (eol - bol) < pmatch[0].rm_eo)
1204                        die("regexp returned nonsense");
1205
1206                /* Match beginning must be either beginning of the
1207                 * line, or at word boundary (i.e. the last char must
1208                 * not be a word char).  Similarly, match end must be
1209                 * either end of the line, or at word boundary
1210                 * (i.e. the next char must not be a word char).
1211                 */
1212                if ( ((pmatch[0].rm_so == 0) ||
1213                      !word_char(bol[pmatch[0].rm_so-1])) &&
1214                     ((pmatch[0].rm_eo == (eol-bol)) ||
1215                      !word_char(bol[pmatch[0].rm_eo])) )
1216                        ;
1217                else
1218                        hit = 0;
1219
1220                /* Words consist of at least one character. */
1221                if (pmatch->rm_so == pmatch->rm_eo)
1222                        hit = 0;
1223
1224                if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1225                        /* There could be more than one match on the
1226                         * line, and the first match might not be
1227                         * strict word match.  But later ones could be!
1228                         * Forward to the next possible start, i.e. the
1229                         * next position following a non-word char.
1230                         */
1231                        bol = pmatch[0].rm_so + bol + 1;
1232                        while (word_char(bol[-1]) && bol < eol)
1233                                bol++;
1234                        eflags |= REG_NOTBOL;
1235                        if (bol < eol)
1236                                goto again;
1237                }
1238        }
1239        if (p->token == GREP_PATTERN_HEAD && saved_ch)
1240                *eol = saved_ch;
1241        if (hit) {
1242                pmatch[0].rm_so += bol - start;
1243                pmatch[0].rm_eo += bol - start;
1244        }
1245        return hit;
1246}
1247
1248static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1249                           enum grep_context ctx, int collect_hits)
1250{
1251        int h = 0;
1252        regmatch_t match;
1253
1254        if (!x)
1255                die("Not a valid grep expression");
1256        switch (x->node) {
1257        case GREP_NODE_TRUE:
1258                h = 1;
1259                break;
1260        case GREP_NODE_ATOM:
1261                h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1262                break;
1263        case GREP_NODE_NOT:
1264                h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1265                break;
1266        case GREP_NODE_AND:
1267                if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1268                        return 0;
1269                h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1270                break;
1271        case GREP_NODE_OR:
1272                if (!collect_hits)
1273                        return (match_expr_eval(x->u.binary.left,
1274                                                bol, eol, ctx, 0) ||
1275                                match_expr_eval(x->u.binary.right,
1276                                                bol, eol, ctx, 0));
1277                h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1278                x->u.binary.left->hit |= h;
1279                h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1280                break;
1281        default:
1282                die("Unexpected node type (internal error) %d", x->node);
1283        }
1284        if (collect_hits)
1285                x->hit |= h;
1286        return h;
1287}
1288
1289static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1290                      enum grep_context ctx, int collect_hits)
1291{
1292        struct grep_expr *x = opt->pattern_expression;
1293        return match_expr_eval(x, bol, eol, ctx, collect_hits);
1294}
1295
1296static int match_line(struct grep_opt *opt, char *bol, char *eol,
1297                      enum grep_context ctx, int collect_hits)
1298{
1299        struct grep_pat *p;
1300        regmatch_t match;
1301
1302        if (opt->extended)
1303                return match_expr(opt, bol, eol, ctx, collect_hits);
1304
1305        /* we do not call with collect_hits without being extended */
1306        for (p = opt->pattern_list; p; p = p->next) {
1307                if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1308                        return 1;
1309        }
1310        return 0;
1311}
1312
1313static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1314                              enum grep_context ctx,
1315                              regmatch_t *pmatch, int eflags)
1316{
1317        regmatch_t match;
1318
1319        if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1320                return 0;
1321        if (match.rm_so < 0 || match.rm_eo < 0)
1322                return 0;
1323        if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1324                if (match.rm_so > pmatch->rm_so)
1325                        return 1;
1326                if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1327                        return 1;
1328        }
1329        pmatch->rm_so = match.rm_so;
1330        pmatch->rm_eo = match.rm_eo;
1331        return 1;
1332}
1333
1334static int next_match(struct grep_opt *opt, char *bol, char *eol,
1335                      enum grep_context ctx, regmatch_t *pmatch, int eflags)
1336{
1337        struct grep_pat *p;
1338        int hit = 0;
1339
1340        pmatch->rm_so = pmatch->rm_eo = -1;
1341        if (bol < eol) {
1342                for (p = opt->pattern_list; p; p = p->next) {
1343                        switch (p->token) {
1344                        case GREP_PATTERN: /* atom */
1345                        case GREP_PATTERN_HEAD:
1346                        case GREP_PATTERN_BODY:
1347                                hit |= match_next_pattern(p, bol, eol, ctx,
1348                                                          pmatch, eflags);
1349                                break;
1350                        default:
1351                                break;
1352                        }
1353                }
1354        }
1355        return hit;
1356}
1357
1358static void show_line(struct grep_opt *opt, char *bol, char *eol,
1359                      const char *name, unsigned lno, char sign)
1360{
1361        int rest = eol - bol;
1362        const char *match_color, *line_color = NULL;
1363
1364        if (opt->file_break && opt->last_shown == 0) {
1365                if (opt->show_hunk_mark)
1366                        opt->output(opt, "\n", 1);
1367        } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1368                if (opt->last_shown == 0) {
1369                        if (opt->show_hunk_mark) {
1370                                output_color(opt, "--", 2, opt->color_sep);
1371                                opt->output(opt, "\n", 1);
1372                        }
1373                } else if (lno > opt->last_shown + 1) {
1374                        output_color(opt, "--", 2, opt->color_sep);
1375                        opt->output(opt, "\n", 1);
1376                }
1377        }
1378        if (opt->heading && opt->last_shown == 0) {
1379                output_color(opt, name, strlen(name), opt->color_filename);
1380                opt->output(opt, "\n", 1);
1381        }
1382        opt->last_shown = lno;
1383
1384        if (!opt->heading && opt->pathname) {
1385                output_color(opt, name, strlen(name), opt->color_filename);
1386                output_sep(opt, sign);
1387        }
1388        if (opt->linenum) {
1389                char buf[32];
1390                xsnprintf(buf, sizeof(buf), "%d", lno);
1391                output_color(opt, buf, strlen(buf), opt->color_lineno);
1392                output_sep(opt, sign);
1393        }
1394        if (opt->color) {
1395                regmatch_t match;
1396                enum grep_context ctx = GREP_CONTEXT_BODY;
1397                int ch = *eol;
1398                int eflags = 0;
1399
1400                if (sign == ':')
1401                        match_color = opt->color_match_selected;
1402                else
1403                        match_color = opt->color_match_context;
1404                if (sign == ':')
1405                        line_color = opt->color_selected;
1406                else if (sign == '-')
1407                        line_color = opt->color_context;
1408                else if (sign == '=')
1409                        line_color = opt->color_function;
1410                *eol = '\0';
1411                while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1412                        if (match.rm_so == match.rm_eo)
1413                                break;
1414
1415                        output_color(opt, bol, match.rm_so, line_color);
1416                        output_color(opt, bol + match.rm_so,
1417                                     match.rm_eo - match.rm_so, match_color);
1418                        bol += match.rm_eo;
1419                        rest -= match.rm_eo;
1420                        eflags = REG_NOTBOL;
1421                }
1422                *eol = ch;
1423        }
1424        output_color(opt, bol, rest, line_color);
1425        opt->output(opt, "\n", 1);
1426}
1427
1428#ifndef NO_PTHREADS
1429int grep_use_locks;
1430
1431/*
1432 * This lock protects access to the gitattributes machinery, which is
1433 * not thread-safe.
1434 */
1435pthread_mutex_t grep_attr_mutex;
1436
1437static inline void grep_attr_lock(void)
1438{
1439        if (grep_use_locks)
1440                pthread_mutex_lock(&grep_attr_mutex);
1441}
1442
1443static inline void grep_attr_unlock(void)
1444{
1445        if (grep_use_locks)
1446                pthread_mutex_unlock(&grep_attr_mutex);
1447}
1448
1449/*
1450 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1451 */
1452pthread_mutex_t grep_read_mutex;
1453
1454#else
1455#define grep_attr_lock()
1456#define grep_attr_unlock()
1457#endif
1458
1459static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1460{
1461        xdemitconf_t *xecfg = opt->priv;
1462        if (xecfg && !xecfg->find_func) {
1463                grep_source_load_driver(gs);
1464                if (gs->driver->funcname.pattern) {
1465                        const struct userdiff_funcname *pe = &gs->driver->funcname;
1466                        xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1467                } else {
1468                        xecfg = opt->priv = NULL;
1469                }
1470        }
1471
1472        if (xecfg) {
1473                char buf[1];
1474                return xecfg->find_func(bol, eol - bol, buf, 1,
1475                                        xecfg->find_func_priv) >= 0;
1476        }
1477
1478        if (bol == eol)
1479                return 0;
1480        if (isalpha(*bol) || *bol == '_' || *bol == '$')
1481                return 1;
1482        return 0;
1483}
1484
1485static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1486                               char *bol, unsigned lno)
1487{
1488        while (bol > gs->buf) {
1489                char *eol = --bol;
1490
1491                while (bol > gs->buf && bol[-1] != '\n')
1492                        bol--;
1493                lno--;
1494
1495                if (lno <= opt->last_shown)
1496                        break;
1497
1498                if (match_funcname(opt, gs, bol, eol)) {
1499                        show_line(opt, bol, eol, gs->name, lno, '=');
1500                        break;
1501                }
1502        }
1503}
1504
1505static int is_empty_line(const char *bol, const char *eol);
1506
1507static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1508                             char *bol, char *end, unsigned lno)
1509{
1510        unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1511        int funcname_needed = !!opt->funcname, comment_needed = 0;
1512
1513        if (opt->pre_context < lno)
1514                from = lno - opt->pre_context;
1515        if (from <= opt->last_shown)
1516                from = opt->last_shown + 1;
1517        orig_from = from;
1518        if (opt->funcbody) {
1519                if (match_funcname(opt, gs, bol, end))
1520                        comment_needed = 1;
1521                else
1522                        funcname_needed = 1;
1523                from = opt->last_shown + 1;
1524        }
1525
1526        /* Rewind. */
1527        while (bol > gs->buf && cur > from) {
1528                char *next_bol = bol;
1529                char *eol = --bol;
1530
1531                while (bol > gs->buf && bol[-1] != '\n')
1532                        bol--;
1533                cur--;
1534                if (comment_needed && (is_empty_line(bol, eol) ||
1535                                       match_funcname(opt, gs, bol, eol))) {
1536                        comment_needed = 0;
1537                        from = orig_from;
1538                        if (cur < from) {
1539                                cur++;
1540                                bol = next_bol;
1541                                break;
1542                        }
1543                }
1544                if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1545                        funcname_lno = cur;
1546                        funcname_needed = 0;
1547                        if (opt->funcbody)
1548                                comment_needed = 1;
1549                        else
1550                                from = orig_from;
1551                }
1552        }
1553
1554        /* We need to look even further back to find a function signature. */
1555        if (opt->funcname && funcname_needed)
1556                show_funcname_line(opt, gs, bol, cur);
1557
1558        /* Back forward. */
1559        while (cur < lno) {
1560                char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1561
1562                while (*eol != '\n')
1563                        eol++;
1564                show_line(opt, bol, eol, gs->name, cur, sign);
1565                bol = eol + 1;
1566                cur++;
1567        }
1568}
1569
1570static int should_lookahead(struct grep_opt *opt)
1571{
1572        struct grep_pat *p;
1573
1574        if (opt->extended)
1575                return 0; /* punt for too complex stuff */
1576        if (opt->invert)
1577                return 0;
1578        for (p = opt->pattern_list; p; p = p->next) {
1579                if (p->token != GREP_PATTERN)
1580                        return 0; /* punt for "header only" and stuff */
1581        }
1582        return 1;
1583}
1584
1585static int look_ahead(struct grep_opt *opt,
1586                      unsigned long *left_p,
1587                      unsigned *lno_p,
1588                      char **bol_p)
1589{
1590        unsigned lno = *lno_p;
1591        char *bol = *bol_p;
1592        struct grep_pat *p;
1593        char *sp, *last_bol;
1594        regoff_t earliest = -1;
1595
1596        for (p = opt->pattern_list; p; p = p->next) {
1597                int hit;
1598                regmatch_t m;
1599
1600                hit = patmatch(p, bol, bol + *left_p, &m, 0);
1601                if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1602                        continue;
1603                if (earliest < 0 || m.rm_so < earliest)
1604                        earliest = m.rm_so;
1605        }
1606
1607        if (earliest < 0) {
1608                *bol_p = bol + *left_p;
1609                *left_p = 0;
1610                return 1;
1611        }
1612        for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1613                ; /* find the beginning of the line */
1614        last_bol = sp;
1615
1616        for (sp = bol; sp < last_bol; sp++) {
1617                if (*sp == '\n')
1618                        lno++;
1619        }
1620        *left_p -= last_bol - bol;
1621        *bol_p = last_bol;
1622        *lno_p = lno;
1623        return 0;
1624}
1625
1626static int fill_textconv_grep(struct userdiff_driver *driver,
1627                              struct grep_source *gs)
1628{
1629        struct diff_filespec *df;
1630        char *buf;
1631        size_t size;
1632
1633        if (!driver || !driver->textconv)
1634                return grep_source_load(gs);
1635
1636        /*
1637         * The textconv interface is intimately tied to diff_filespecs, so we
1638         * have to pretend to be one. If we could unify the grep_source
1639         * and diff_filespec structs, this mess could just go away.
1640         */
1641        df = alloc_filespec(gs->path);
1642        switch (gs->type) {
1643        case GREP_SOURCE_OID:
1644                fill_filespec(df, gs->identifier, 1, 0100644);
1645                break;
1646        case GREP_SOURCE_FILE:
1647                fill_filespec(df, &null_oid, 0, 0100644);
1648                break;
1649        default:
1650                die("BUG: attempt to textconv something without a path?");
1651        }
1652
1653        /*
1654         * fill_textconv is not remotely thread-safe; it may load objects
1655         * behind the scenes, and it modifies the global diff tempfile
1656         * structure.
1657         */
1658        grep_read_lock();
1659        size = fill_textconv(driver, df, &buf);
1660        grep_read_unlock();
1661        free_filespec(df);
1662
1663        /*
1664         * The normal fill_textconv usage by the diff machinery would just keep
1665         * the textconv'd buf separate from the diff_filespec. But much of the
1666         * grep code passes around a grep_source and assumes that its "buf"
1667         * pointer is the beginning of the thing we are searching. So let's
1668         * install our textconv'd version into the grep_source, taking care not
1669         * to leak any existing buffer.
1670         */
1671        grep_source_clear_data(gs);
1672        gs->buf = buf;
1673        gs->size = size;
1674
1675        return 0;
1676}
1677
1678static int is_empty_line(const char *bol, const char *eol)
1679{
1680        while (bol < eol && isspace(*bol))
1681                bol++;
1682        return bol == eol;
1683}
1684
1685static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1686{
1687        char *bol;
1688        char *peek_bol = NULL;
1689        unsigned long left;
1690        unsigned lno = 1;
1691        unsigned last_hit = 0;
1692        int binary_match_only = 0;
1693        unsigned count = 0;
1694        int try_lookahead = 0;
1695        int show_function = 0;
1696        struct userdiff_driver *textconv = NULL;
1697        enum grep_context ctx = GREP_CONTEXT_HEAD;
1698        xdemitconf_t xecfg;
1699
1700        if (!opt->output)
1701                opt->output = std_output;
1702
1703        if (opt->pre_context || opt->post_context || opt->file_break ||
1704            opt->funcbody) {
1705                /* Show hunk marks, except for the first file. */
1706                if (opt->last_shown)
1707                        opt->show_hunk_mark = 1;
1708                /*
1709                 * If we're using threads then we can't easily identify
1710                 * the first file.  Always put hunk marks in that case
1711                 * and skip the very first one later in work_done().
1712                 */
1713                if (opt->output != std_output)
1714                        opt->show_hunk_mark = 1;
1715        }
1716        opt->last_shown = 0;
1717
1718        if (opt->allow_textconv) {
1719                grep_source_load_driver(gs);
1720                /*
1721                 * We might set up the shared textconv cache data here, which
1722                 * is not thread-safe.
1723                 */
1724                grep_attr_lock();
1725                textconv = userdiff_get_textconv(gs->driver);
1726                grep_attr_unlock();
1727        }
1728
1729        /*
1730         * We know the result of a textconv is text, so we only have to care
1731         * about binary handling if we are not using it.
1732         */
1733        if (!textconv) {
1734                switch (opt->binary) {
1735                case GREP_BINARY_DEFAULT:
1736                        if (grep_source_is_binary(gs))
1737                                binary_match_only = 1;
1738                        break;
1739                case GREP_BINARY_NOMATCH:
1740                        if (grep_source_is_binary(gs))
1741                                return 0; /* Assume unmatch */
1742                        break;
1743                case GREP_BINARY_TEXT:
1744                        break;
1745                default:
1746                        die("BUG: unknown binary handling mode");
1747                }
1748        }
1749
1750        memset(&xecfg, 0, sizeof(xecfg));
1751        opt->priv = &xecfg;
1752
1753        try_lookahead = should_lookahead(opt);
1754
1755        if (fill_textconv_grep(textconv, gs) < 0)
1756                return 0;
1757
1758        bol = gs->buf;
1759        left = gs->size;
1760        while (left) {
1761                char *eol, ch;
1762                int hit;
1763
1764                /*
1765                 * look_ahead() skips quickly to the line that possibly
1766                 * has the next hit; don't call it if we need to do
1767                 * something more than just skipping the current line
1768                 * in response to an unmatch for the current line.  E.g.
1769                 * inside a post-context window, we will show the current
1770                 * line as a context around the previous hit when it
1771                 * doesn't hit.
1772                 */
1773                if (try_lookahead
1774                    && !(last_hit
1775                         && (show_function ||
1776                             lno <= last_hit + opt->post_context))
1777                    && look_ahead(opt, &left, &lno, &bol))
1778                        break;
1779                eol = end_of_line(bol, &left);
1780                ch = *eol;
1781                *eol = 0;
1782
1783                if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1784                        ctx = GREP_CONTEXT_BODY;
1785
1786                hit = match_line(opt, bol, eol, ctx, collect_hits);
1787                *eol = ch;
1788
1789                if (collect_hits)
1790                        goto next_line;
1791
1792                /* "grep -v -e foo -e bla" should list lines
1793                 * that do not have either, so inversion should
1794                 * be done outside.
1795                 */
1796                if (opt->invert)
1797                        hit = !hit;
1798                if (opt->unmatch_name_only) {
1799                        if (hit)
1800                                return 0;
1801                        goto next_line;
1802                }
1803                if (hit) {
1804                        count++;
1805                        if (opt->status_only)
1806                                return 1;
1807                        if (opt->name_only) {
1808                                show_name(opt, gs->name);
1809                                return 1;
1810                        }
1811                        if (opt->count)
1812                                goto next_line;
1813                        if (binary_match_only) {
1814                                opt->output(opt, "Binary file ", 12);
1815                                output_color(opt, gs->name, strlen(gs->name),
1816                                             opt->color_filename);
1817                                opt->output(opt, " matches\n", 9);
1818                                return 1;
1819                        }
1820                        /* Hit at this line.  If we haven't shown the
1821                         * pre-context lines, we would need to show them.
1822                         */
1823                        if (opt->pre_context || opt->funcbody)
1824                                show_pre_context(opt, gs, bol, eol, lno);
1825                        else if (opt->funcname)
1826                                show_funcname_line(opt, gs, bol, lno);
1827                        show_line(opt, bol, eol, gs->name, lno, ':');
1828                        last_hit = lno;
1829                        if (opt->funcbody)
1830                                show_function = 1;
1831                        goto next_line;
1832                }
1833                if (show_function && (!peek_bol || peek_bol < bol)) {
1834                        unsigned long peek_left = left;
1835                        char *peek_eol = eol;
1836
1837                        /*
1838                         * Trailing empty lines are not interesting.
1839                         * Peek past them to see if they belong to the
1840                         * body of the current function.
1841                         */
1842                        peek_bol = bol;
1843                        while (is_empty_line(peek_bol, peek_eol)) {
1844                                peek_bol = peek_eol + 1;
1845                                peek_eol = end_of_line(peek_bol, &peek_left);
1846                        }
1847
1848                        if (match_funcname(opt, gs, peek_bol, peek_eol))
1849                                show_function = 0;
1850                }
1851                if (show_function ||
1852                    (last_hit && lno <= last_hit + opt->post_context)) {
1853                        /* If the last hit is within the post context,
1854                         * we need to show this line.
1855                         */
1856                        show_line(opt, bol, eol, gs->name, lno, '-');
1857                }
1858
1859        next_line:
1860                bol = eol + 1;
1861                if (!left)
1862                        break;
1863                left--;
1864                lno++;
1865        }
1866
1867        if (collect_hits)
1868                return 0;
1869
1870        if (opt->status_only)
1871                return opt->unmatch_name_only;
1872        if (opt->unmatch_name_only) {
1873                /* We did not see any hit, so we want to show this */
1874                show_name(opt, gs->name);
1875                return 1;
1876        }
1877
1878        xdiff_clear_find_func(&xecfg);
1879        opt->priv = NULL;
1880
1881        /* NEEDSWORK:
1882         * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1883         * which feels mostly useless but sometimes useful.  Maybe
1884         * make it another option?  For now suppress them.
1885         */
1886        if (opt->count && count) {
1887                char buf[32];
1888                if (opt->pathname) {
1889                        output_color(opt, gs->name, strlen(gs->name),
1890                                     opt->color_filename);
1891                        output_sep(opt, ':');
1892                }
1893                xsnprintf(buf, sizeof(buf), "%u\n", count);
1894                opt->output(opt, buf, strlen(buf));
1895                return 1;
1896        }
1897        return !!last_hit;
1898}
1899
1900static void clr_hit_marker(struct grep_expr *x)
1901{
1902        /* All-hit markers are meaningful only at the very top level
1903         * OR node.
1904         */
1905        while (1) {
1906                x->hit = 0;
1907                if (x->node != GREP_NODE_OR)
1908                        return;
1909                x->u.binary.left->hit = 0;
1910                x = x->u.binary.right;
1911        }
1912}
1913
1914static int chk_hit_marker(struct grep_expr *x)
1915{
1916        /* Top level nodes have hit markers.  See if they all are hits */
1917        while (1) {
1918                if (x->node != GREP_NODE_OR)
1919                        return x->hit;
1920                if (!x->u.binary.left->hit)
1921                        return 0;
1922                x = x->u.binary.right;
1923        }
1924}
1925
1926int grep_source(struct grep_opt *opt, struct grep_source *gs)
1927{
1928        /*
1929         * we do not have to do the two-pass grep when we do not check
1930         * buffer-wide "all-match".
1931         */
1932        if (!opt->all_match)
1933                return grep_source_1(opt, gs, 0);
1934
1935        /* Otherwise the toplevel "or" terms hit a bit differently.
1936         * We first clear hit markers from them.
1937         */
1938        clr_hit_marker(opt->pattern_expression);
1939        grep_source_1(opt, gs, 1);
1940
1941        if (!chk_hit_marker(opt->pattern_expression))
1942                return 0;
1943
1944        return grep_source_1(opt, gs, 0);
1945}
1946
1947int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1948{
1949        struct grep_source gs;
1950        int r;
1951
1952        grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1953        gs.buf = buf;
1954        gs.size = size;
1955
1956        r = grep_source(opt, &gs);
1957
1958        grep_source_clear(&gs);
1959        return r;
1960}
1961
1962void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1963                      const char *name, const char *path,
1964                      const void *identifier)
1965{
1966        gs->type = type;
1967        gs->name = xstrdup_or_null(name);
1968        gs->path = xstrdup_or_null(path);
1969        gs->buf = NULL;
1970        gs->size = 0;
1971        gs->driver = NULL;
1972
1973        switch (type) {
1974        case GREP_SOURCE_FILE:
1975                gs->identifier = xstrdup(identifier);
1976                break;
1977        case GREP_SOURCE_OID:
1978                gs->identifier = oiddup(identifier);
1979                break;
1980        case GREP_SOURCE_BUF:
1981                gs->identifier = NULL;
1982                break;
1983        }
1984}
1985
1986void grep_source_clear(struct grep_source *gs)
1987{
1988        FREE_AND_NULL(gs->name);
1989        FREE_AND_NULL(gs->path);
1990        FREE_AND_NULL(gs->identifier);
1991        grep_source_clear_data(gs);
1992}
1993
1994void grep_source_clear_data(struct grep_source *gs)
1995{
1996        switch (gs->type) {
1997        case GREP_SOURCE_FILE:
1998        case GREP_SOURCE_OID:
1999                FREE_AND_NULL(gs->buf);
2000                gs->size = 0;
2001                break;
2002        case GREP_SOURCE_BUF:
2003                /* leave user-provided buf intact */
2004                break;
2005        }
2006}
2007
2008static int grep_source_load_oid(struct grep_source *gs)
2009{
2010        enum object_type type;
2011
2012        grep_read_lock();
2013        gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
2014        grep_read_unlock();
2015
2016        if (!gs->buf)
2017                return error(_("'%s': unable to read %s"),
2018                             gs->name,
2019                             oid_to_hex(gs->identifier));
2020        return 0;
2021}
2022
2023static int grep_source_load_file(struct grep_source *gs)
2024{
2025        const char *filename = gs->identifier;
2026        struct stat st;
2027        char *data;
2028        size_t size;
2029        int i;
2030
2031        if (lstat(filename, &st) < 0) {
2032        err_ret:
2033                if (errno != ENOENT)
2034                        error_errno(_("failed to stat '%s'"), filename);
2035                return -1;
2036        }
2037        if (!S_ISREG(st.st_mode))
2038                return -1;
2039        size = xsize_t(st.st_size);
2040        i = open(filename, O_RDONLY);
2041        if (i < 0)
2042                goto err_ret;
2043        data = xmallocz(size);
2044        if (st.st_size != read_in_full(i, data, size)) {
2045                error_errno(_("'%s': short read"), filename);
2046                close(i);
2047                free(data);
2048                return -1;
2049        }
2050        close(i);
2051
2052        gs->buf = data;
2053        gs->size = size;
2054        return 0;
2055}
2056
2057static int grep_source_load(struct grep_source *gs)
2058{
2059        if (gs->buf)
2060                return 0;
2061
2062        switch (gs->type) {
2063        case GREP_SOURCE_FILE:
2064                return grep_source_load_file(gs);
2065        case GREP_SOURCE_OID:
2066                return grep_source_load_oid(gs);
2067        case GREP_SOURCE_BUF:
2068                return gs->buf ? 0 : -1;
2069        }
2070        die("BUG: invalid grep_source type to load");
2071}
2072
2073void grep_source_load_driver(struct grep_source *gs)
2074{
2075        if (gs->driver)
2076                return;
2077
2078        grep_attr_lock();
2079        if (gs->path)
2080                gs->driver = userdiff_find_by_path(gs->path);
2081        if (!gs->driver)
2082                gs->driver = userdiff_find_by_name("default");
2083        grep_attr_unlock();
2084}
2085
2086static int grep_source_is_binary(struct grep_source *gs)
2087{
2088        grep_source_load_driver(gs);
2089        if (gs->driver->binary != -1)
2090                return gs->driver->binary;
2091
2092        if (!grep_source_load(gs))
2093                return buffer_is_binary(gs->buf, gs->size);
2094
2095        return 0;
2096}