Andrew's git / gitweb.git / grep.c
?
grep.con commit rebase -i: fix SIGSEGV when 'merge <branch>' fails (bc9238b)
   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
  21static void color_set(char *dst, const char *color_bytes)
  22{
  23        xsnprintf(dst, COLOR_MAXLEN, "%s", color_bytes);
  24}
  25
  26/*
  27 * Initialize the grep_defaults template with hardcoded defaults.
  28 * We could let the compiler do this, but without C99 initializers
  29 * the code gets unwieldy and unreadable, so...
  30 */
  31void init_grep_defaults(void)
  32{
  33        struct grep_opt *opt = &grep_defaults;
  34        static int run_once;
  35
  36        if (run_once)
  37                return;
  38        run_once++;
  39
  40        memset(opt, 0, sizeof(*opt));
  41        opt->relative = 1;
  42        opt->pathname = 1;
  43        opt->max_depth = -1;
  44        opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  45        color_set(opt->color_context, "");
  46        color_set(opt->color_filename, "");
  47        color_set(opt->color_function, "");
  48        color_set(opt->color_lineno, "");
  49        color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
  50        color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
  51        color_set(opt->color_selected, "");
  52        color_set(opt->color_sep, GIT_COLOR_CYAN);
  53        opt->color = -1;
  54        opt->output = std_output;
  55}
  56
  57static int parse_pattern_type_arg(const char *opt, const char *arg)
  58{
  59        if (!strcmp(arg, "default"))
  60                return GREP_PATTERN_TYPE_UNSPECIFIED;
  61        else if (!strcmp(arg, "basic"))
  62                return GREP_PATTERN_TYPE_BRE;
  63        else if (!strcmp(arg, "extended"))
  64                return GREP_PATTERN_TYPE_ERE;
  65        else if (!strcmp(arg, "fixed"))
  66                return GREP_PATTERN_TYPE_FIXED;
  67        else if (!strcmp(arg, "perl"))
  68                return GREP_PATTERN_TYPE_PCRE;
  69        die("bad %s argument: %s", opt, arg);
  70}
  71
  72/*
  73 * Read the configuration file once and store it in
  74 * the grep_defaults template.
  75 */
  76int grep_config(const char *var, const char *value, void *cb)
  77{
  78        struct grep_opt *opt = &grep_defaults;
  79        char *color = NULL;
  80
  81        if (userdiff_config(var, value) < 0)
  82                return -1;
  83
  84        if (!strcmp(var, "grep.extendedregexp")) {
  85                opt->extended_regexp_option = git_config_bool(var, value);
  86                return 0;
  87        }
  88
  89        if (!strcmp(var, "grep.patterntype")) {
  90                opt->pattern_type_option = parse_pattern_type_arg(var, value);
  91                return 0;
  92        }
  93
  94        if (!strcmp(var, "grep.linenumber")) {
  95                opt->linenum = git_config_bool(var, value);
  96                return 0;
  97        }
  98
  99        if (!strcmp(var, "grep.fullname")) {
 100                opt->relative = !git_config_bool(var, value);
 101                return 0;
 102        }
 103
 104        if (!strcmp(var, "color.grep"))
 105                opt->color = git_config_colorbool(var, value);
 106        else if (!strcmp(var, "color.grep.context"))
 107                color = opt->color_context;
 108        else if (!strcmp(var, "color.grep.filename"))
 109                color = opt->color_filename;
 110        else if (!strcmp(var, "color.grep.function"))
 111                color = opt->color_function;
 112        else if (!strcmp(var, "color.grep.linenumber"))
 113                color = opt->color_lineno;
 114        else if (!strcmp(var, "color.grep.matchcontext"))
 115                color = opt->color_match_context;
 116        else if (!strcmp(var, "color.grep.matchselected"))
 117                color = opt->color_match_selected;
 118        else if (!strcmp(var, "color.grep.selected"))
 119                color = opt->color_selected;
 120        else if (!strcmp(var, "color.grep.separator"))
 121                color = opt->color_sep;
 122        else if (!strcmp(var, "color.grep.match")) {
 123                int rc = 0;
 124                if (!value)
 125                        return config_error_nonbool(var);
 126                rc |= color_parse(value, opt->color_match_context);
 127                rc |= color_parse(value, opt->color_match_selected);
 128                return rc;
 129        }
 130
 131        if (color) {
 132                if (!value)
 133                        return config_error_nonbool(var);
 134                return color_parse(value, color);
 135        }
 136        return 0;
 137}
 138
 139/*
 140 * Initialize one instance of grep_opt and copy the
 141 * default values from the template we read the configuration
 142 * information in an earlier call to git_config(grep_config).
 143 */
 144void grep_init(struct grep_opt *opt, const char *prefix)
 145{
 146        struct grep_opt *def = &grep_defaults;
 147
 148        memset(opt, 0, sizeof(*opt));
 149        opt->prefix = prefix;
 150        opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 151        opt->pattern_tail = &opt->pattern_list;
 152        opt->header_tail = &opt->header_list;
 153
 154        opt->color = def->color;
 155        opt->extended_regexp_option = def->extended_regexp_option;
 156        opt->pattern_type_option = def->pattern_type_option;
 157        opt->linenum = def->linenum;
 158        opt->max_depth = def->max_depth;
 159        opt->pathname = def->pathname;
 160        opt->relative = def->relative;
 161        opt->output = def->output;
 162
 163        color_set(opt->color_context, def->color_context);
 164        color_set(opt->color_filename, def->color_filename);
 165        color_set(opt->color_function, def->color_function);
 166        color_set(opt->color_lineno, def->color_lineno);
 167        color_set(opt->color_match_context, def->color_match_context);
 168        color_set(opt->color_match_selected, def->color_match_selected);
 169        color_set(opt->color_selected, def->color_selected);
 170        color_set(opt->color_sep, def->color_sep);
 171}
 172
 173static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 174{
 175        /*
 176         * When committing to the pattern type by setting the relevant
 177         * fields in grep_opt it's generally not necessary to zero out
 178         * the fields we're not choosing, since they won't have been
 179         * set by anything. The extended_regexp_option field is the
 180         * only exception to this.
 181         *
 182         * This is because in the process of parsing grep.patternType
 183         * & grep.extendedRegexp we set opt->pattern_type_option and
 184         * opt->extended_regexp_option, respectively. We then
 185         * internally use opt->extended_regexp_option to see if we're
 186         * compiling an ERE. It must be unset if that's not actually
 187         * the case.
 188         */
 189        if (pattern_type != GREP_PATTERN_TYPE_ERE &&
 190            opt->extended_regexp_option)
 191                opt->extended_regexp_option = 0;
 192
 193        switch (pattern_type) {
 194        case GREP_PATTERN_TYPE_UNSPECIFIED:
 195                /* fall through */
 196
 197        case GREP_PATTERN_TYPE_BRE:
 198                break;
 199
 200        case GREP_PATTERN_TYPE_ERE:
 201                opt->extended_regexp_option = 1;
 202                break;
 203
 204        case GREP_PATTERN_TYPE_FIXED:
 205                opt->fixed = 1;
 206                break;
 207
 208        case GREP_PATTERN_TYPE_PCRE:
 209#ifdef USE_LIBPCRE2
 210                opt->pcre2 = 1;
 211#else
 212                /*
 213                 * It's important that pcre1 always be assigned to
 214                 * even when there's no USE_LIBPCRE* defined. We still
 215                 * call the PCRE stub function, it just dies with
 216                 * "cannot use Perl-compatible regexes[...]".
 217                 */
 218                opt->pcre1 = 1;
 219#endif
 220                break;
 221        }
 222}
 223
 224void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 225{
 226        if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 227                grep_set_pattern_type_option(pattern_type, opt);
 228        else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 229                grep_set_pattern_type_option(opt->pattern_type_option, opt);
 230        else if (opt->extended_regexp_option)
 231                /*
 232                 * This branch *must* happen after setting from the
 233                 * opt->pattern_type_option above, we don't want
 234                 * grep.extendedRegexp to override grep.patternType!
 235                 */
 236                grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 237}
 238
 239static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 240                                        const char *origin, int no,
 241                                        enum grep_pat_token t,
 242                                        enum grep_header_field field)
 243{
 244        struct grep_pat *p = xcalloc(1, sizeof(*p));
 245        p->pattern = xmemdupz(pat, patlen);
 246        p->patternlen = patlen;
 247        p->origin = origin;
 248        p->no = no;
 249        p->token = t;
 250        p->field = field;
 251        return p;
 252}
 253
 254static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 255{
 256        **tail = p;
 257        *tail = &p->next;
 258        p->next = NULL;
 259
 260        switch (p->token) {
 261        case GREP_PATTERN: /* atom */
 262        case GREP_PATTERN_HEAD:
 263        case GREP_PATTERN_BODY:
 264                for (;;) {
 265                        struct grep_pat *new_pat;
 266                        size_t len = 0;
 267                        char *cp = p->pattern + p->patternlen, *nl = NULL;
 268                        while (++len <= p->patternlen) {
 269                                if (*(--cp) == '\n') {
 270                                        nl = cp;
 271                                        break;
 272                                }
 273                        }
 274                        if (!nl)
 275                                break;
 276                        new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 277                                                  p->no, p->token, p->field);
 278                        new_pat->next = p->next;
 279                        if (!p->next)
 280                                *tail = &new_pat->next;
 281                        p->next = new_pat;
 282                        *nl = '\0';
 283                        p->patternlen -= len;
 284                }
 285                break;
 286        default:
 287                break;
 288        }
 289}
 290
 291void append_header_grep_pattern(struct grep_opt *opt,
 292                                enum grep_header_field field, const char *pat)
 293{
 294        struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 295                                             GREP_PATTERN_HEAD, field);
 296        if (field == GREP_HEADER_REFLOG)
 297                opt->use_reflog_filter = 1;
 298        do_append_grep_pat(&opt->header_tail, p);
 299}
 300
 301void append_grep_pattern(struct grep_opt *opt, const char *pat,
 302                         const char *origin, int no, enum grep_pat_token t)
 303{
 304        append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 305}
 306
 307void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 308                     const char *origin, int no, enum grep_pat_token t)
 309{
 310        struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 311        do_append_grep_pat(&opt->pattern_tail, p);
 312}
 313
 314struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 315{
 316        struct grep_pat *pat;
 317        struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 318        *ret = *opt;
 319
 320        ret->pattern_list = NULL;
 321        ret->pattern_tail = &ret->pattern_list;
 322
 323        for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 324        {
 325                if(pat->token == GREP_PATTERN_HEAD)
 326                        append_header_grep_pattern(ret, pat->field,
 327                                                   pat->pattern);
 328                else
 329                        append_grep_pat(ret, pat->pattern, pat->patternlen,
 330                                        pat->origin, pat->no, pat->token);
 331        }
 332
 333        return ret;
 334}
 335
 336static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 337                const char *error)
 338{
 339        char where[1024];
 340
 341        if (p->no)
 342                xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 343        else if (p->origin)
 344                xsnprintf(where, sizeof(where), "%s, ", p->origin);
 345        else
 346                where[0] = 0;
 347
 348        die("%s'%s': %s", where, p->pattern, error);
 349}
 350
 351static int is_fixed(const char *s, size_t len)
 352{
 353        size_t i;
 354
 355        for (i = 0; i < len; i++) {
 356                if (is_regex_special(s[i]))
 357                        return 0;
 358        }
 359
 360        return 1;
 361}
 362
 363static int has_null(const char *s, size_t len)
 364{
 365        /*
 366         * regcomp cannot accept patterns with NULs so when using it
 367         * we consider any pattern containing a NUL fixed.
 368         */
 369        if (memchr(s, 0, len))
 370                return 1;
 371
 372        return 0;
 373}
 374
 375#ifdef USE_LIBPCRE1
 376static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 377{
 378        const char *error;
 379        int erroffset;
 380        int options = PCRE_MULTILINE;
 381
 382        if (opt->ignore_case) {
 383                if (has_non_ascii(p->pattern))
 384                        p->pcre1_tables = pcre_maketables();
 385                options |= PCRE_CASELESS;
 386        }
 387        if (is_utf8_locale() && has_non_ascii(p->pattern))
 388                options |= PCRE_UTF8;
 389
 390        p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 391                                      p->pcre1_tables);
 392        if (!p->pcre1_regexp)
 393                compile_regexp_failed(p, error);
 394
 395        p->pcre1_extra_info = pcre_study(p->pcre1_regexp, GIT_PCRE_STUDY_JIT_COMPILE, &error);
 396        if (!p->pcre1_extra_info && error)
 397                die("%s", error);
 398
 399#ifdef GIT_PCRE1_USE_JIT
 400        pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
 401        if (p->pcre1_jit_on == 1) {
 402                p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
 403                if (!p->pcre1_jit_stack)
 404                        die("Couldn't allocate PCRE JIT stack");
 405                pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
 406        } else if (p->pcre1_jit_on != 0) {
 407                BUG("The pcre1_jit_on variable should be 0 or 1, not %d",
 408                    p->pcre1_jit_on);
 409        }
 410#endif
 411}
 412
 413static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 414                regmatch_t *match, int eflags)
 415{
 416        int ovector[30], ret, flags = 0;
 417
 418        if (eflags & REG_NOTBOL)
 419                flags |= PCRE_NOTBOL;
 420
 421#ifdef GIT_PCRE1_USE_JIT
 422        if (p->pcre1_jit_on) {
 423                ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 424                                    eol - line, 0, flags, ovector,
 425                                    ARRAY_SIZE(ovector), p->pcre1_jit_stack);
 426        } else
 427#endif
 428        {
 429                ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
 430                                eol - line, 0, flags, ovector,
 431                                ARRAY_SIZE(ovector));
 432        }
 433
 434        if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 435                die("pcre_exec failed with error code %d", ret);
 436        if (ret > 0) {
 437                ret = 0;
 438                match->rm_so = ovector[0];
 439                match->rm_eo = ovector[1];
 440        }
 441
 442        return ret;
 443}
 444
 445static void free_pcre1_regexp(struct grep_pat *p)
 446{
 447        pcre_free(p->pcre1_regexp);
 448#ifdef GIT_PCRE1_USE_JIT
 449        if (p->pcre1_jit_on) {
 450                pcre_free_study(p->pcre1_extra_info);
 451                pcre_jit_stack_free(p->pcre1_jit_stack);
 452        } else
 453#endif
 454        {
 455                pcre_free(p->pcre1_extra_info);
 456        }
 457        pcre_free((void *)p->pcre1_tables);
 458}
 459#else /* !USE_LIBPCRE1 */
 460static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 461{
 462        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 463}
 464
 465static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
 466                regmatch_t *match, int eflags)
 467{
 468        return 1;
 469}
 470
 471static void free_pcre1_regexp(struct grep_pat *p)
 472{
 473}
 474#endif /* !USE_LIBPCRE1 */
 475
 476#ifdef USE_LIBPCRE2
 477static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 478{
 479        int error;
 480        PCRE2_UCHAR errbuf[256];
 481        PCRE2_SIZE erroffset;
 482        int options = PCRE2_MULTILINE;
 483        const uint8_t *character_tables = NULL;
 484        int jitret;
 485        int patinforet;
 486        size_t jitsizearg;
 487
 488        assert(opt->pcre2);
 489
 490        p->pcre2_compile_context = NULL;
 491
 492        if (opt->ignore_case) {
 493                if (has_non_ascii(p->pattern)) {
 494                        character_tables = pcre2_maketables(NULL);
 495                        p->pcre2_compile_context = pcre2_compile_context_create(NULL);
 496                        pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
 497                }
 498                options |= PCRE2_CASELESS;
 499        }
 500        if (is_utf8_locale() && has_non_ascii(p->pattern))
 501                options |= PCRE2_UTF;
 502
 503        p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
 504                                         p->patternlen, options, &error, &erroffset,
 505                                         p->pcre2_compile_context);
 506
 507        if (p->pcre2_pattern) {
 508                p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
 509                if (!p->pcre2_match_data)
 510                        die("Couldn't allocate PCRE2 match data");
 511        } else {
 512                pcre2_get_error_message(error, errbuf, sizeof(errbuf));
 513                compile_regexp_failed(p, (const char *)&errbuf);
 514        }
 515
 516        pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
 517        if (p->pcre2_jit_on == 1) {
 518                jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
 519                if (jitret)
 520                        die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
 521
 522                /*
 523                 * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
 524                 * tells us whether the library itself supports JIT,
 525                 * but to see whether we're going to be actually using
 526                 * JIT we need to extract PCRE2_INFO_JITSIZE from the
 527                 * pattern *after* we do pcre2_jit_compile() above.
 528                 *
 529                 * This is because if the pattern contains the
 530                 * (*NO_JIT) verb (see pcre2syntax(3))
 531                 * pcre2_jit_compile() will exit early with 0. If we
 532                 * then proceed to call pcre2_jit_match() further down
 533                 * the line instead of pcre2_match() we'll either
 534                 * segfault (pre PCRE 10.31) or run into a fatal error
 535                 * (post PCRE2 10.31)
 536                 */
 537                patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
 538                if (patinforet)
 539                        BUG("pcre2_pattern_info() failed: %d", patinforet);
 540                if (jitsizearg == 0) {
 541                        p->pcre2_jit_on = 0;
 542                        return;
 543                }
 544
 545                p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
 546                if (!p->pcre2_jit_stack)
 547                        die("Couldn't allocate PCRE2 JIT stack");
 548                p->pcre2_match_context = pcre2_match_context_create(NULL);
 549                if (!p->pcre2_match_context)
 550                        die("Couldn't allocate PCRE2 match context");
 551                pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
 552        } else if (p->pcre2_jit_on != 0) {
 553                BUG("The pcre2_jit_on variable should be 0 or 1, not %d",
 554                    p->pcre1_jit_on);
 555        }
 556}
 557
 558static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 559                regmatch_t *match, int eflags)
 560{
 561        int ret, flags = 0;
 562        PCRE2_SIZE *ovector;
 563        PCRE2_UCHAR errbuf[256];
 564
 565        if (eflags & REG_NOTBOL)
 566                flags |= PCRE2_NOTBOL;
 567
 568        if (p->pcre2_jit_on)
 569                ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
 570                                      eol - line, 0, flags, p->pcre2_match_data,
 571                                      NULL);
 572        else
 573                ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
 574                                  eol - line, 0, flags, p->pcre2_match_data,
 575                                  NULL);
 576
 577        if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
 578                pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
 579                die("%s failed with error code %d: %s",
 580                    (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
 581                    errbuf);
 582        }
 583        if (ret > 0) {
 584                ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
 585                ret = 0;
 586                match->rm_so = (int)ovector[0];
 587                match->rm_eo = (int)ovector[1];
 588        }
 589
 590        return ret;
 591}
 592
 593static void free_pcre2_pattern(struct grep_pat *p)
 594{
 595        pcre2_compile_context_free(p->pcre2_compile_context);
 596        pcre2_code_free(p->pcre2_pattern);
 597        pcre2_match_data_free(p->pcre2_match_data);
 598        pcre2_jit_stack_free(p->pcre2_jit_stack);
 599        pcre2_match_context_free(p->pcre2_match_context);
 600}
 601#else /* !USE_LIBPCRE2 */
 602static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 603{
 604        /*
 605         * Unreachable until USE_LIBPCRE2 becomes synonymous with
 606         * USE_LIBPCRE. See the sibling comment in
 607         * grep_set_pattern_type_option().
 608         */
 609        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 610}
 611
 612static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
 613                regmatch_t *match, int eflags)
 614{
 615        return 1;
 616}
 617
 618static void free_pcre2_pattern(struct grep_pat *p)
 619{
 620}
 621#endif /* !USE_LIBPCRE2 */
 622
 623static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 624{
 625        struct strbuf sb = STRBUF_INIT;
 626        int err;
 627        int regflags = 0;
 628
 629        basic_regex_quote_buf(&sb, p->pattern);
 630        if (opt->ignore_case)
 631                regflags |= REG_ICASE;
 632        err = regcomp(&p->regexp, sb.buf, regflags);
 633        if (opt->debug)
 634                fprintf(stderr, "fixed %s\n", sb.buf);
 635        strbuf_release(&sb);
 636        if (err) {
 637                char errbuf[1024];
 638                regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 639                compile_regexp_failed(p, errbuf);
 640        }
 641}
 642
 643static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 644{
 645        int ascii_only;
 646        int err;
 647        int regflags = REG_NEWLINE;
 648
 649        p->word_regexp = opt->word_regexp;
 650        p->ignore_case = opt->ignore_case;
 651        ascii_only     = !has_non_ascii(p->pattern);
 652
 653        /*
 654         * Even when -F (fixed) asks us to do a non-regexp search, we
 655         * may not be able to correctly case-fold when -i
 656         * (ignore-case) is asked (in which case, we'll synthesize a
 657         * regexp to match the pattern that matches regexp special
 658         * characters literally, while ignoring case differences).  On
 659         * the other hand, even without -F, if the pattern does not
 660         * have any regexp special characters and there is no need for
 661         * case-folding search, we can internally turn it into a
 662         * simple string match using kws.  p->fixed tells us if we
 663         * want to use kws.
 664         */
 665        if (opt->fixed ||
 666            has_null(p->pattern, p->patternlen) ||
 667            is_fixed(p->pattern, p->patternlen))
 668                p->fixed = !p->ignore_case || ascii_only;
 669
 670        if (p->fixed) {
 671                p->kws = kwsalloc(p->ignore_case ? tolower_trans_tbl : NULL);
 672                kwsincr(p->kws, p->pattern, p->patternlen);
 673                kwsprep(p->kws);
 674                return;
 675        } else if (opt->fixed) {
 676                /*
 677                 * We come here when the pattern has the non-ascii
 678                 * characters we cannot case-fold, and asked to
 679                 * ignore-case.
 680                 */
 681                compile_fixed_regexp(p, opt);
 682                return;
 683        }
 684
 685        if (opt->pcre2) {
 686                compile_pcre2_pattern(p, opt);
 687                return;
 688        }
 689
 690        if (opt->pcre1) {
 691                compile_pcre1_regexp(p, opt);
 692                return;
 693        }
 694
 695        if (p->ignore_case)
 696                regflags |= REG_ICASE;
 697        if (opt->extended_regexp_option)
 698                regflags |= REG_EXTENDED;
 699        err = regcomp(&p->regexp, p->pattern, regflags);
 700        if (err) {
 701                char errbuf[1024];
 702                regerror(err, &p->regexp, errbuf, 1024);
 703                compile_regexp_failed(p, errbuf);
 704        }
 705}
 706
 707static struct grep_expr *compile_pattern_or(struct grep_pat **);
 708static struct grep_expr *compile_pattern_atom(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_PATTERN: /* atom */
 718        case GREP_PATTERN_HEAD:
 719        case GREP_PATTERN_BODY:
 720                x = xcalloc(1, sizeof (struct grep_expr));
 721                x->node = GREP_NODE_ATOM;
 722                x->u.atom = p;
 723                *list = p->next;
 724                return x;
 725        case GREP_OPEN_PAREN:
 726                *list = p->next;
 727                x = compile_pattern_or(list);
 728                if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 729                        die("unmatched parenthesis");
 730                *list = (*list)->next;
 731                return x;
 732        default:
 733                return NULL;
 734        }
 735}
 736
 737static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 738{
 739        struct grep_pat *p;
 740        struct grep_expr *x;
 741
 742        p = *list;
 743        if (!p)
 744                return NULL;
 745        switch (p->token) {
 746        case GREP_NOT:
 747                if (!p->next)
 748                        die("--not not followed by pattern expression");
 749                *list = p->next;
 750                x = xcalloc(1, sizeof (struct grep_expr));
 751                x->node = GREP_NODE_NOT;
 752                x->u.unary = compile_pattern_not(list);
 753                if (!x->u.unary)
 754                        die("--not followed by non pattern expression");
 755                return x;
 756        default:
 757                return compile_pattern_atom(list);
 758        }
 759}
 760
 761static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 762{
 763        struct grep_pat *p;
 764        struct grep_expr *x, *y, *z;
 765
 766        x = compile_pattern_not(list);
 767        p = *list;
 768        if (p && p->token == GREP_AND) {
 769                if (!p->next)
 770                        die("--and not followed by pattern expression");
 771                *list = p->next;
 772                y = compile_pattern_and(list);
 773                if (!y)
 774                        die("--and not followed by pattern expression");
 775                z = xcalloc(1, sizeof (struct grep_expr));
 776                z->node = GREP_NODE_AND;
 777                z->u.binary.left = x;
 778                z->u.binary.right = y;
 779                return z;
 780        }
 781        return x;
 782}
 783
 784static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 785{
 786        struct grep_pat *p;
 787        struct grep_expr *x, *y, *z;
 788
 789        x = compile_pattern_and(list);
 790        p = *list;
 791        if (x && p && p->token != GREP_CLOSE_PAREN) {
 792                y = compile_pattern_or(list);
 793                if (!y)
 794                        die("not a pattern expression %s", p->pattern);
 795                z = xcalloc(1, sizeof (struct grep_expr));
 796                z->node = GREP_NODE_OR;
 797                z->u.binary.left = x;
 798                z->u.binary.right = y;
 799                return z;
 800        }
 801        return x;
 802}
 803
 804static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 805{
 806        return compile_pattern_or(list);
 807}
 808
 809static void indent(int in)
 810{
 811        while (in-- > 0)
 812                fputc(' ', stderr);
 813}
 814
 815static void dump_grep_pat(struct grep_pat *p)
 816{
 817        switch (p->token) {
 818        case GREP_AND: fprintf(stderr, "*and*"); break;
 819        case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 820        case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 821        case GREP_NOT: fprintf(stderr, "*not*"); break;
 822        case GREP_OR: fprintf(stderr, "*or*"); break;
 823
 824        case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 825        case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 826        case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 827        }
 828
 829        switch (p->token) {
 830        default: break;
 831        case GREP_PATTERN_HEAD:
 832                fprintf(stderr, "<head %d>", p->field); break;
 833        case GREP_PATTERN_BODY:
 834                fprintf(stderr, "<body>"); break;
 835        }
 836        switch (p->token) {
 837        default: break;
 838        case GREP_PATTERN_HEAD:
 839        case GREP_PATTERN_BODY:
 840        case GREP_PATTERN:
 841                fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 842                break;
 843        }
 844        fputc('\n', stderr);
 845}
 846
 847static void dump_grep_expression_1(struct grep_expr *x, int in)
 848{
 849        indent(in);
 850        switch (x->node) {
 851        case GREP_NODE_TRUE:
 852                fprintf(stderr, "true\n");
 853                break;
 854        case GREP_NODE_ATOM:
 855                dump_grep_pat(x->u.atom);
 856                break;
 857        case GREP_NODE_NOT:
 858                fprintf(stderr, "(not\n");
 859                dump_grep_expression_1(x->u.unary, in+1);
 860                indent(in);
 861                fprintf(stderr, ")\n");
 862                break;
 863        case GREP_NODE_AND:
 864                fprintf(stderr, "(and\n");
 865                dump_grep_expression_1(x->u.binary.left, in+1);
 866                dump_grep_expression_1(x->u.binary.right, in+1);
 867                indent(in);
 868                fprintf(stderr, ")\n");
 869                break;
 870        case GREP_NODE_OR:
 871                fprintf(stderr, "(or\n");
 872                dump_grep_expression_1(x->u.binary.left, in+1);
 873                dump_grep_expression_1(x->u.binary.right, in+1);
 874                indent(in);
 875                fprintf(stderr, ")\n");
 876                break;
 877        }
 878}
 879
 880static void dump_grep_expression(struct grep_opt *opt)
 881{
 882        struct grep_expr *x = opt->pattern_expression;
 883
 884        if (opt->all_match)
 885                fprintf(stderr, "[all-match]\n");
 886        dump_grep_expression_1(x, 0);
 887        fflush(NULL);
 888}
 889
 890static struct grep_expr *grep_true_expr(void)
 891{
 892        struct grep_expr *z = xcalloc(1, sizeof(*z));
 893        z->node = GREP_NODE_TRUE;
 894        return z;
 895}
 896
 897static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 898{
 899        struct grep_expr *z = xcalloc(1, sizeof(*z));
 900        z->node = GREP_NODE_OR;
 901        z->u.binary.left = left;
 902        z->u.binary.right = right;
 903        return z;
 904}
 905
 906static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 907{
 908        struct grep_pat *p;
 909        struct grep_expr *header_expr;
 910        struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 911        enum grep_header_field fld;
 912
 913        if (!opt->header_list)
 914                return NULL;
 915
 916        for (p = opt->header_list; p; p = p->next) {
 917                if (p->token != GREP_PATTERN_HEAD)
 918                        BUG("a non-header pattern in grep header list.");
 919                if (p->field < GREP_HEADER_FIELD_MIN ||
 920                    GREP_HEADER_FIELD_MAX <= p->field)
 921                        BUG("unknown header field %d", p->field);
 922                compile_regexp(p, opt);
 923        }
 924
 925        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 926                header_group[fld] = NULL;
 927
 928        for (p = opt->header_list; p; p = p->next) {
 929                struct grep_expr *h;
 930                struct grep_pat *pp = p;
 931
 932                h = compile_pattern_atom(&pp);
 933                if (!h || pp != p->next)
 934                        BUG("malformed header expr");
 935                if (!header_group[p->field]) {
 936                        header_group[p->field] = h;
 937                        continue;
 938                }
 939                header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 940        }
 941
 942        header_expr = NULL;
 943
 944        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 945                if (!header_group[fld])
 946                        continue;
 947                if (!header_expr)
 948                        header_expr = grep_true_expr();
 949                header_expr = grep_or_expr(header_group[fld], header_expr);
 950        }
 951        return header_expr;
 952}
 953
 954static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 955{
 956        struct grep_expr *z = x;
 957
 958        while (x) {
 959                assert(x->node == GREP_NODE_OR);
 960                if (x->u.binary.right &&
 961                    x->u.binary.right->node == GREP_NODE_TRUE) {
 962                        x->u.binary.right = y;
 963                        break;
 964                }
 965                x = x->u.binary.right;
 966        }
 967        return z;
 968}
 969
 970static void compile_grep_patterns_real(struct grep_opt *opt)
 971{
 972        struct grep_pat *p;
 973        struct grep_expr *header_expr = prep_header_patterns(opt);
 974
 975        for (p = opt->pattern_list; p; p = p->next) {
 976                switch (p->token) {
 977                case GREP_PATTERN: /* atom */
 978                case GREP_PATTERN_HEAD:
 979                case GREP_PATTERN_BODY:
 980                        compile_regexp(p, opt);
 981                        break;
 982                default:
 983                        opt->extended = 1;
 984                        break;
 985                }
 986        }
 987
 988        if (opt->all_match || header_expr)
 989                opt->extended = 1;
 990        else if (!opt->extended && !opt->debug)
 991                return;
 992
 993        p = opt->pattern_list;
 994        if (p)
 995                opt->pattern_expression = compile_pattern_expr(&p);
 996        if (p)
 997                die("incomplete pattern expression: %s", p->pattern);
 998
 999        if (!header_expr)
1000                return;
1001
1002        if (!opt->pattern_expression)
1003                opt->pattern_expression = header_expr;
1004        else if (opt->all_match)
1005                opt->pattern_expression = grep_splice_or(header_expr,
1006                                                         opt->pattern_expression);
1007        else
1008                opt->pattern_expression = grep_or_expr(opt->pattern_expression,
1009                                                       header_expr);
1010        opt->all_match = 1;
1011}
1012
1013void compile_grep_patterns(struct grep_opt *opt)
1014{
1015        compile_grep_patterns_real(opt);
1016        if (opt->debug)
1017                dump_grep_expression(opt);
1018}
1019
1020static void free_pattern_expr(struct grep_expr *x)
1021{
1022        switch (x->node) {
1023        case GREP_NODE_TRUE:
1024        case GREP_NODE_ATOM:
1025                break;
1026        case GREP_NODE_NOT:
1027                free_pattern_expr(x->u.unary);
1028                break;
1029        case GREP_NODE_AND:
1030        case GREP_NODE_OR:
1031                free_pattern_expr(x->u.binary.left);
1032                free_pattern_expr(x->u.binary.right);
1033                break;
1034        }
1035        free(x);
1036}
1037
1038void free_grep_patterns(struct grep_opt *opt)
1039{
1040        struct grep_pat *p, *n;
1041
1042        for (p = opt->pattern_list; p; p = n) {
1043                n = p->next;
1044                switch (p->token) {
1045                case GREP_PATTERN: /* atom */
1046                case GREP_PATTERN_HEAD:
1047                case GREP_PATTERN_BODY:
1048                        if (p->kws)
1049                                kwsfree(p->kws);
1050                        else if (p->pcre1_regexp)
1051                                free_pcre1_regexp(p);
1052                        else if (p->pcre2_pattern)
1053                                free_pcre2_pattern(p);
1054                        else
1055                                regfree(&p->regexp);
1056                        free(p->pattern);
1057                        break;
1058                default:
1059                        break;
1060                }
1061                free(p);
1062        }
1063
1064        if (!opt->extended)
1065                return;
1066        free_pattern_expr(opt->pattern_expression);
1067}
1068
1069static char *end_of_line(char *cp, unsigned long *left)
1070{
1071        unsigned long l = *left;
1072        while (l && *cp != '\n') {
1073                l--;
1074                cp++;
1075        }
1076        *left = l;
1077        return cp;
1078}
1079
1080static int word_char(char ch)
1081{
1082        return isalnum(ch) || ch == '_';
1083}
1084
1085static void output_color(struct grep_opt *opt, const void *data, size_t size,
1086                         const char *color)
1087{
1088        if (want_color(opt->color) && color && color[0]) {
1089                opt->output(opt, color, strlen(color));
1090                opt->output(opt, data, size);
1091                opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1092        } else
1093                opt->output(opt, data, size);
1094}
1095
1096static void output_sep(struct grep_opt *opt, char sign)
1097{
1098        if (opt->null_following_name)
1099                opt->output(opt, "\0", 1);
1100        else
1101                output_color(opt, &sign, 1, opt->color_sep);
1102}
1103
1104static void show_name(struct grep_opt *opt, const char *name)
1105{
1106        output_color(opt, name, strlen(name), opt->color_filename);
1107        opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1108}
1109
1110static int fixmatch(struct grep_pat *p, char *line, char *eol,
1111                    regmatch_t *match)
1112{
1113        struct kwsmatch kwsm;
1114        size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
1115        if (offset == -1) {
1116                match->rm_so = match->rm_eo = -1;
1117                return REG_NOMATCH;
1118        } else {
1119                match->rm_so = offset;
1120                match->rm_eo = match->rm_so + kwsm.size[0];
1121                return 0;
1122        }
1123}
1124
1125static int patmatch(struct grep_pat *p, char *line, char *eol,
1126                    regmatch_t *match, int eflags)
1127{
1128        int hit;
1129
1130        if (p->fixed)
1131                hit = !fixmatch(p, line, eol, match);
1132        else if (p->pcre1_regexp)
1133                hit = !pcre1match(p, line, eol, match, eflags);
1134        else if (p->pcre2_pattern)
1135                hit = !pcre2match(p, line, eol, match, eflags);
1136        else
1137                hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1138                                   eflags);
1139
1140        return hit;
1141}
1142
1143static int strip_timestamp(char *bol, char **eol_p)
1144{
1145        char *eol = *eol_p;
1146        int ch;
1147
1148        while (bol < --eol) {
1149                if (*eol != '>')
1150                        continue;
1151                *eol_p = ++eol;
1152                ch = *eol;
1153                *eol = '\0';
1154                return ch;
1155        }
1156        return 0;
1157}
1158
1159static struct {
1160        const char *field;
1161        size_t len;
1162} header_field[] = {
1163        { "author ", 7 },
1164        { "committer ", 10 },
1165        { "reflog ", 7 },
1166};
1167
1168static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1169                             enum grep_context ctx,
1170                             regmatch_t *pmatch, int eflags)
1171{
1172        int hit = 0;
1173        int saved_ch = 0;
1174        const char *start = bol;
1175
1176        if ((p->token != GREP_PATTERN) &&
1177            ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1178                return 0;
1179
1180        if (p->token == GREP_PATTERN_HEAD) {
1181                const char *field;
1182                size_t len;
1183                assert(p->field < ARRAY_SIZE(header_field));
1184                field = header_field[p->field].field;
1185                len = header_field[p->field].len;
1186                if (strncmp(bol, field, len))
1187                        return 0;
1188                bol += len;
1189                switch (p->field) {
1190                case GREP_HEADER_AUTHOR:
1191                case GREP_HEADER_COMMITTER:
1192                        saved_ch = strip_timestamp(bol, &eol);
1193                        break;
1194                default:
1195                        break;
1196                }
1197        }
1198
1199 again:
1200        hit = patmatch(p, bol, eol, pmatch, eflags);
1201
1202        if (hit && p->word_regexp) {
1203                if ((pmatch[0].rm_so < 0) ||
1204                    (eol - bol) < pmatch[0].rm_so ||
1205                    (pmatch[0].rm_eo < 0) ||
1206                    (eol - bol) < pmatch[0].rm_eo)
1207                        die("regexp returned nonsense");
1208
1209                /* Match beginning must be either beginning of the
1210                 * line, or at word boundary (i.e. the last char must
1211                 * not be a word char).  Similarly, match end must be
1212                 * either end of the line, or at word boundary
1213                 * (i.e. the next char must not be a word char).
1214                 */
1215                if ( ((pmatch[0].rm_so == 0) ||
1216                      !word_char(bol[pmatch[0].rm_so-1])) &&
1217                     ((pmatch[0].rm_eo == (eol-bol)) ||
1218                      !word_char(bol[pmatch[0].rm_eo])) )
1219                        ;
1220                else
1221                        hit = 0;
1222
1223                /* Words consist of at least one character. */
1224                if (pmatch->rm_so == pmatch->rm_eo)
1225                        hit = 0;
1226
1227                if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1228                        /* There could be more than one match on the
1229                         * line, and the first match might not be
1230                         * strict word match.  But later ones could be!
1231                         * Forward to the next possible start, i.e. the
1232                         * next position following a non-word char.
1233                         */
1234                        bol = pmatch[0].rm_so + bol + 1;
1235                        while (word_char(bol[-1]) && bol < eol)
1236                                bol++;
1237                        eflags |= REG_NOTBOL;
1238                        if (bol < eol)
1239                                goto again;
1240                }
1241        }
1242        if (p->token == GREP_PATTERN_HEAD && saved_ch)
1243                *eol = saved_ch;
1244        if (hit) {
1245                pmatch[0].rm_so += bol - start;
1246                pmatch[0].rm_eo += bol - start;
1247        }
1248        return hit;
1249}
1250
1251static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1252                           enum grep_context ctx, int collect_hits)
1253{
1254        int h = 0;
1255        regmatch_t match;
1256
1257        if (!x)
1258                die("Not a valid grep expression");
1259        switch (x->node) {
1260        case GREP_NODE_TRUE:
1261                h = 1;
1262                break;
1263        case GREP_NODE_ATOM:
1264                h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1265                break;
1266        case GREP_NODE_NOT:
1267                h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1268                break;
1269        case GREP_NODE_AND:
1270                if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1271                        return 0;
1272                h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1273                break;
1274        case GREP_NODE_OR:
1275                if (!collect_hits)
1276                        return (match_expr_eval(x->u.binary.left,
1277                                                bol, eol, ctx, 0) ||
1278                                match_expr_eval(x->u.binary.right,
1279                                                bol, eol, ctx, 0));
1280                h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1281                x->u.binary.left->hit |= h;
1282                h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1283                break;
1284        default:
1285                die("Unexpected node type (internal error) %d", x->node);
1286        }
1287        if (collect_hits)
1288                x->hit |= h;
1289        return h;
1290}
1291
1292static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1293                      enum grep_context ctx, int collect_hits)
1294{
1295        struct grep_expr *x = opt->pattern_expression;
1296        return match_expr_eval(x, bol, eol, ctx, collect_hits);
1297}
1298
1299static int match_line(struct grep_opt *opt, char *bol, char *eol,
1300                      enum grep_context ctx, int collect_hits)
1301{
1302        struct grep_pat *p;
1303        regmatch_t match;
1304
1305        if (opt->extended)
1306                return match_expr(opt, bol, eol, ctx, collect_hits);
1307
1308        /* we do not call with collect_hits without being extended */
1309        for (p = opt->pattern_list; p; p = p->next) {
1310                if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1311                        return 1;
1312        }
1313        return 0;
1314}
1315
1316static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1317                              enum grep_context ctx,
1318                              regmatch_t *pmatch, int eflags)
1319{
1320        regmatch_t match;
1321
1322        if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1323                return 0;
1324        if (match.rm_so < 0 || match.rm_eo < 0)
1325                return 0;
1326        if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1327                if (match.rm_so > pmatch->rm_so)
1328                        return 1;
1329                if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1330                        return 1;
1331        }
1332        pmatch->rm_so = match.rm_so;
1333        pmatch->rm_eo = match.rm_eo;
1334        return 1;
1335}
1336
1337static int next_match(struct grep_opt *opt, char *bol, char *eol,
1338                      enum grep_context ctx, regmatch_t *pmatch, int eflags)
1339{
1340        struct grep_pat *p;
1341        int hit = 0;
1342
1343        pmatch->rm_so = pmatch->rm_eo = -1;
1344        if (bol < eol) {
1345                for (p = opt->pattern_list; p; p = p->next) {
1346                        switch (p->token) {
1347                        case GREP_PATTERN: /* atom */
1348                        case GREP_PATTERN_HEAD:
1349                        case GREP_PATTERN_BODY:
1350                                hit |= match_next_pattern(p, bol, eol, ctx,
1351                                                          pmatch, eflags);
1352                                break;
1353                        default:
1354                                break;
1355                        }
1356                }
1357        }
1358        return hit;
1359}
1360
1361static void show_line(struct grep_opt *opt, char *bol, char *eol,
1362                      const char *name, unsigned lno, char sign)
1363{
1364        int rest = eol - bol;
1365        const char *match_color, *line_color = NULL;
1366
1367        if (opt->file_break && opt->last_shown == 0) {
1368                if (opt->show_hunk_mark)
1369                        opt->output(opt, "\n", 1);
1370        } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1371                if (opt->last_shown == 0) {
1372                        if (opt->show_hunk_mark) {
1373                                output_color(opt, "--", 2, opt->color_sep);
1374                                opt->output(opt, "\n", 1);
1375                        }
1376                } else if (lno > opt->last_shown + 1) {
1377                        output_color(opt, "--", 2, opt->color_sep);
1378                        opt->output(opt, "\n", 1);
1379                }
1380        }
1381        if (opt->heading && opt->last_shown == 0) {
1382                output_color(opt, name, strlen(name), opt->color_filename);
1383                opt->output(opt, "\n", 1);
1384        }
1385        opt->last_shown = lno;
1386
1387        if (!opt->heading && opt->pathname) {
1388                output_color(opt, name, strlen(name), opt->color_filename);
1389                output_sep(opt, sign);
1390        }
1391        if (opt->linenum) {
1392                char buf[32];
1393                xsnprintf(buf, sizeof(buf), "%d", lno);
1394                output_color(opt, buf, strlen(buf), opt->color_lineno);
1395                output_sep(opt, sign);
1396        }
1397        if (opt->color) {
1398                regmatch_t match;
1399                enum grep_context ctx = GREP_CONTEXT_BODY;
1400                int ch = *eol;
1401                int eflags = 0;
1402
1403                if (sign == ':')
1404                        match_color = opt->color_match_selected;
1405                else
1406                        match_color = opt->color_match_context;
1407                if (sign == ':')
1408                        line_color = opt->color_selected;
1409                else if (sign == '-')
1410                        line_color = opt->color_context;
1411                else if (sign == '=')
1412                        line_color = opt->color_function;
1413                *eol = '\0';
1414                while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1415                        if (match.rm_so == match.rm_eo)
1416                                break;
1417
1418                        output_color(opt, bol, match.rm_so, line_color);
1419                        output_color(opt, bol + match.rm_so,
1420                                     match.rm_eo - match.rm_so, match_color);
1421                        bol += match.rm_eo;
1422                        rest -= match.rm_eo;
1423                        eflags = REG_NOTBOL;
1424                }
1425                *eol = ch;
1426        }
1427        output_color(opt, bol, rest, line_color);
1428        opt->output(opt, "\n", 1);
1429}
1430
1431#ifndef NO_PTHREADS
1432int grep_use_locks;
1433
1434/*
1435 * This lock protects access to the gitattributes machinery, which is
1436 * not thread-safe.
1437 */
1438pthread_mutex_t grep_attr_mutex;
1439
1440static inline void grep_attr_lock(void)
1441{
1442        if (grep_use_locks)
1443                pthread_mutex_lock(&grep_attr_mutex);
1444}
1445
1446static inline void grep_attr_unlock(void)
1447{
1448        if (grep_use_locks)
1449                pthread_mutex_unlock(&grep_attr_mutex);
1450}
1451
1452/*
1453 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1454 */
1455pthread_mutex_t grep_read_mutex;
1456
1457#else
1458#define grep_attr_lock()
1459#define grep_attr_unlock()
1460#endif
1461
1462static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1463{
1464        xdemitconf_t *xecfg = opt->priv;
1465        if (xecfg && !xecfg->find_func) {
1466                grep_source_load_driver(gs);
1467                if (gs->driver->funcname.pattern) {
1468                        const struct userdiff_funcname *pe = &gs->driver->funcname;
1469                        xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1470                } else {
1471                        xecfg = opt->priv = NULL;
1472                }
1473        }
1474
1475        if (xecfg) {
1476                char buf[1];
1477                return xecfg->find_func(bol, eol - bol, buf, 1,
1478                                        xecfg->find_func_priv) >= 0;
1479        }
1480
1481        if (bol == eol)
1482                return 0;
1483        if (isalpha(*bol) || *bol == '_' || *bol == '$')
1484                return 1;
1485        return 0;
1486}
1487
1488static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1489                               char *bol, unsigned lno)
1490{
1491        while (bol > gs->buf) {
1492                char *eol = --bol;
1493
1494                while (bol > gs->buf && bol[-1] != '\n')
1495                        bol--;
1496                lno--;
1497
1498                if (lno <= opt->last_shown)
1499                        break;
1500
1501                if (match_funcname(opt, gs, bol, eol)) {
1502                        show_line(opt, bol, eol, gs->name, lno, '=');
1503                        break;
1504                }
1505        }
1506}
1507
1508static int is_empty_line(const char *bol, const char *eol);
1509
1510static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1511                             char *bol, char *end, unsigned lno)
1512{
1513        unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
1514        int funcname_needed = !!opt->funcname, comment_needed = 0;
1515
1516        if (opt->pre_context < lno)
1517                from = lno - opt->pre_context;
1518        if (from <= opt->last_shown)
1519                from = opt->last_shown + 1;
1520        orig_from = from;
1521        if (opt->funcbody) {
1522                if (match_funcname(opt, gs, bol, end))
1523                        comment_needed = 1;
1524                else
1525                        funcname_needed = 1;
1526                from = opt->last_shown + 1;
1527        }
1528
1529        /* Rewind. */
1530        while (bol > gs->buf && cur > from) {
1531                char *next_bol = bol;
1532                char *eol = --bol;
1533
1534                while (bol > gs->buf && bol[-1] != '\n')
1535                        bol--;
1536                cur--;
1537                if (comment_needed && (is_empty_line(bol, eol) ||
1538                                       match_funcname(opt, gs, bol, eol))) {
1539                        comment_needed = 0;
1540                        from = orig_from;
1541                        if (cur < from) {
1542                                cur++;
1543                                bol = next_bol;
1544                                break;
1545                        }
1546                }
1547                if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1548                        funcname_lno = cur;
1549                        funcname_needed = 0;
1550                        if (opt->funcbody)
1551                                comment_needed = 1;
1552                        else
1553                                from = orig_from;
1554                }
1555        }
1556
1557        /* We need to look even further back to find a function signature. */
1558        if (opt->funcname && funcname_needed)
1559                show_funcname_line(opt, gs, bol, cur);
1560
1561        /* Back forward. */
1562        while (cur < lno) {
1563                char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1564
1565                while (*eol != '\n')
1566                        eol++;
1567                show_line(opt, bol, eol, gs->name, cur, sign);
1568                bol = eol + 1;
1569                cur++;
1570        }
1571}
1572
1573static int should_lookahead(struct grep_opt *opt)
1574{
1575        struct grep_pat *p;
1576
1577        if (opt->extended)
1578                return 0; /* punt for too complex stuff */
1579        if (opt->invert)
1580                return 0;
1581        for (p = opt->pattern_list; p; p = p->next) {
1582                if (p->token != GREP_PATTERN)
1583                        return 0; /* punt for "header only" and stuff */
1584        }
1585        return 1;
1586}
1587
1588static int look_ahead(struct grep_opt *opt,
1589                      unsigned long *left_p,
1590                      unsigned *lno_p,
1591                      char **bol_p)
1592{
1593        unsigned lno = *lno_p;
1594        char *bol = *bol_p;
1595        struct grep_pat *p;
1596        char *sp, *last_bol;
1597        regoff_t earliest = -1;
1598
1599        for (p = opt->pattern_list; p; p = p->next) {
1600                int hit;
1601                regmatch_t m;
1602
1603                hit = patmatch(p, bol, bol + *left_p, &m, 0);
1604                if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1605                        continue;
1606                if (earliest < 0 || m.rm_so < earliest)
1607                        earliest = m.rm_so;
1608        }
1609
1610        if (earliest < 0) {
1611                *bol_p = bol + *left_p;
1612                *left_p = 0;
1613                return 1;
1614        }
1615        for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1616                ; /* find the beginning of the line */
1617        last_bol = sp;
1618
1619        for (sp = bol; sp < last_bol; sp++) {
1620                if (*sp == '\n')
1621                        lno++;
1622        }
1623        *left_p -= last_bol - bol;
1624        *bol_p = last_bol;
1625        *lno_p = lno;
1626        return 0;
1627}
1628
1629static int fill_textconv_grep(struct userdiff_driver *driver,
1630                              struct grep_source *gs)
1631{
1632        struct diff_filespec *df;
1633        char *buf;
1634        size_t size;
1635
1636        if (!driver || !driver->textconv)
1637                return grep_source_load(gs);
1638
1639        /*
1640         * The textconv interface is intimately tied to diff_filespecs, so we
1641         * have to pretend to be one. If we could unify the grep_source
1642         * and diff_filespec structs, this mess could just go away.
1643         */
1644        df = alloc_filespec(gs->path);
1645        switch (gs->type) {
1646        case GREP_SOURCE_OID:
1647                fill_filespec(df, gs->identifier, 1, 0100644);
1648                break;
1649        case GREP_SOURCE_FILE:
1650                fill_filespec(df, &null_oid, 0, 0100644);
1651                break;
1652        default:
1653                BUG("attempt to textconv something without a path?");
1654        }
1655
1656        /*
1657         * fill_textconv is not remotely thread-safe; it may load objects
1658         * behind the scenes, and it modifies the global diff tempfile
1659         * structure.
1660         */
1661        grep_read_lock();
1662        size = fill_textconv(driver, df, &buf);
1663        grep_read_unlock();
1664        free_filespec(df);
1665
1666        /*
1667         * The normal fill_textconv usage by the diff machinery would just keep
1668         * the textconv'd buf separate from the diff_filespec. But much of the
1669         * grep code passes around a grep_source and assumes that its "buf"
1670         * pointer is the beginning of the thing we are searching. So let's
1671         * install our textconv'd version into the grep_source, taking care not
1672         * to leak any existing buffer.
1673         */
1674        grep_source_clear_data(gs);
1675        gs->buf = buf;
1676        gs->size = size;
1677
1678        return 0;
1679}
1680
1681static int is_empty_line(const char *bol, const char *eol)
1682{
1683        while (bol < eol && isspace(*bol))
1684                bol++;
1685        return bol == eol;
1686}
1687
1688static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1689{
1690        char *bol;
1691        char *peek_bol = NULL;
1692        unsigned long left;
1693        unsigned lno = 1;
1694        unsigned last_hit = 0;
1695        int binary_match_only = 0;
1696        unsigned count = 0;
1697        int try_lookahead = 0;
1698        int show_function = 0;
1699        struct userdiff_driver *textconv = NULL;
1700        enum grep_context ctx = GREP_CONTEXT_HEAD;
1701        xdemitconf_t xecfg;
1702
1703        if (!opt->output)
1704                opt->output = std_output;
1705
1706        if (opt->pre_context || opt->post_context || opt->file_break ||
1707            opt->funcbody) {
1708                /* Show hunk marks, except for the first file. */
1709                if (opt->last_shown)
1710                        opt->show_hunk_mark = 1;
1711                /*
1712                 * If we're using threads then we can't easily identify
1713                 * the first file.  Always put hunk marks in that case
1714                 * and skip the very first one later in work_done().
1715                 */
1716                if (opt->output != std_output)
1717                        opt->show_hunk_mark = 1;
1718        }
1719        opt->last_shown = 0;
1720
1721        if (opt->allow_textconv) {
1722                grep_source_load_driver(gs);
1723                /*
1724                 * We might set up the shared textconv cache data here, which
1725                 * is not thread-safe.
1726                 */
1727                grep_attr_lock();
1728                textconv = userdiff_get_textconv(gs->driver);
1729                grep_attr_unlock();
1730        }
1731
1732        /*
1733         * We know the result of a textconv is text, so we only have to care
1734         * about binary handling if we are not using it.
1735         */
1736        if (!textconv) {
1737                switch (opt->binary) {
1738                case GREP_BINARY_DEFAULT:
1739                        if (grep_source_is_binary(gs))
1740                                binary_match_only = 1;
1741                        break;
1742                case GREP_BINARY_NOMATCH:
1743                        if (grep_source_is_binary(gs))
1744                                return 0; /* Assume unmatch */
1745                        break;
1746                case GREP_BINARY_TEXT:
1747                        break;
1748                default:
1749                        BUG("unknown binary handling mode");
1750                }
1751        }
1752
1753        memset(&xecfg, 0, sizeof(xecfg));
1754        opt->priv = &xecfg;
1755
1756        try_lookahead = should_lookahead(opt);
1757
1758        if (fill_textconv_grep(textconv, gs) < 0)
1759                return 0;
1760
1761        bol = gs->buf;
1762        left = gs->size;
1763        while (left) {
1764                char *eol, ch;
1765                int hit;
1766
1767                /*
1768                 * look_ahead() skips quickly to the line that possibly
1769                 * has the next hit; don't call it if we need to do
1770                 * something more than just skipping the current line
1771                 * in response to an unmatch for the current line.  E.g.
1772                 * inside a post-context window, we will show the current
1773                 * line as a context around the previous hit when it
1774                 * doesn't hit.
1775                 */
1776                if (try_lookahead
1777                    && !(last_hit
1778                         && (show_function ||
1779                             lno <= last_hit + opt->post_context))
1780                    && look_ahead(opt, &left, &lno, &bol))
1781                        break;
1782                eol = end_of_line(bol, &left);
1783                ch = *eol;
1784                *eol = 0;
1785
1786                if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1787                        ctx = GREP_CONTEXT_BODY;
1788
1789                hit = match_line(opt, bol, eol, ctx, collect_hits);
1790                *eol = ch;
1791
1792                if (collect_hits)
1793                        goto next_line;
1794
1795                /* "grep -v -e foo -e bla" should list lines
1796                 * that do not have either, so inversion should
1797                 * be done outside.
1798                 */
1799                if (opt->invert)
1800                        hit = !hit;
1801                if (opt->unmatch_name_only) {
1802                        if (hit)
1803                                return 0;
1804                        goto next_line;
1805                }
1806                if (hit) {
1807                        count++;
1808                        if (opt->status_only)
1809                                return 1;
1810                        if (opt->name_only) {
1811                                show_name(opt, gs->name);
1812                                return 1;
1813                        }
1814                        if (opt->count)
1815                                goto next_line;
1816                        if (binary_match_only) {
1817                                opt->output(opt, "Binary file ", 12);
1818                                output_color(opt, gs->name, strlen(gs->name),
1819                                             opt->color_filename);
1820                                opt->output(opt, " matches\n", 9);
1821                                return 1;
1822                        }
1823                        /* Hit at this line.  If we haven't shown the
1824                         * pre-context lines, we would need to show them.
1825                         */
1826                        if (opt->pre_context || opt->funcbody)
1827                                show_pre_context(opt, gs, bol, eol, lno);
1828                        else if (opt->funcname)
1829                                show_funcname_line(opt, gs, bol, lno);
1830                        show_line(opt, bol, eol, gs->name, lno, ':');
1831                        last_hit = lno;
1832                        if (opt->funcbody)
1833                                show_function = 1;
1834                        goto next_line;
1835                }
1836                if (show_function && (!peek_bol || peek_bol < bol)) {
1837                        unsigned long peek_left = left;
1838                        char *peek_eol = eol;
1839
1840                        /*
1841                         * Trailing empty lines are not interesting.
1842                         * Peek past them to see if they belong to the
1843                         * body of the current function.
1844                         */
1845                        peek_bol = bol;
1846                        while (is_empty_line(peek_bol, peek_eol)) {
1847                                peek_bol = peek_eol + 1;
1848                                peek_eol = end_of_line(peek_bol, &peek_left);
1849                        }
1850
1851                        if (match_funcname(opt, gs, peek_bol, peek_eol))
1852                                show_function = 0;
1853                }
1854                if (show_function ||
1855                    (last_hit && lno <= last_hit + opt->post_context)) {
1856                        /* If the last hit is within the post context,
1857                         * we need to show this line.
1858                         */
1859                        show_line(opt, bol, eol, gs->name, lno, '-');
1860                }
1861
1862        next_line:
1863                bol = eol + 1;
1864                if (!left)
1865                        break;
1866                left--;
1867                lno++;
1868        }
1869
1870        if (collect_hits)
1871                return 0;
1872
1873        if (opt->status_only)
1874                return opt->unmatch_name_only;
1875        if (opt->unmatch_name_only) {
1876                /* We did not see any hit, so we want to show this */
1877                show_name(opt, gs->name);
1878                return 1;
1879        }
1880
1881        xdiff_clear_find_func(&xecfg);
1882        opt->priv = NULL;
1883
1884        /* NEEDSWORK:
1885         * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1886         * which feels mostly useless but sometimes useful.  Maybe
1887         * make it another option?  For now suppress them.
1888         */
1889        if (opt->count && count) {
1890                char buf[32];
1891                if (opt->pathname) {
1892                        output_color(opt, gs->name, strlen(gs->name),
1893                                     opt->color_filename);
1894                        output_sep(opt, ':');
1895                }
1896                xsnprintf(buf, sizeof(buf), "%u\n", count);
1897                opt->output(opt, buf, strlen(buf));
1898                return 1;
1899        }
1900        return !!last_hit;
1901}
1902
1903static void clr_hit_marker(struct grep_expr *x)
1904{
1905        /* All-hit markers are meaningful only at the very top level
1906         * OR node.
1907         */
1908        while (1) {
1909                x->hit = 0;
1910                if (x->node != GREP_NODE_OR)
1911                        return;
1912                x->u.binary.left->hit = 0;
1913                x = x->u.binary.right;
1914        }
1915}
1916
1917static int chk_hit_marker(struct grep_expr *x)
1918{
1919        /* Top level nodes have hit markers.  See if they all are hits */
1920        while (1) {
1921                if (x->node != GREP_NODE_OR)
1922                        return x->hit;
1923                if (!x->u.binary.left->hit)
1924                        return 0;
1925                x = x->u.binary.right;
1926        }
1927}
1928
1929int grep_source(struct grep_opt *opt, struct grep_source *gs)
1930{
1931        /*
1932         * we do not have to do the two-pass grep when we do not check
1933         * buffer-wide "all-match".
1934         */
1935        if (!opt->all_match)
1936                return grep_source_1(opt, gs, 0);
1937
1938        /* Otherwise the toplevel "or" terms hit a bit differently.
1939         * We first clear hit markers from them.
1940         */
1941        clr_hit_marker(opt->pattern_expression);
1942        grep_source_1(opt, gs, 1);
1943
1944        if (!chk_hit_marker(opt->pattern_expression))
1945                return 0;
1946
1947        return grep_source_1(opt, gs, 0);
1948}
1949
1950int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1951{
1952        struct grep_source gs;
1953        int r;
1954
1955        grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1956        gs.buf = buf;
1957        gs.size = size;
1958
1959        r = grep_source(opt, &gs);
1960
1961        grep_source_clear(&gs);
1962        return r;
1963}
1964
1965void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1966                      const char *name, const char *path,
1967                      const void *identifier)
1968{
1969        gs->type = type;
1970        gs->name = xstrdup_or_null(name);
1971        gs->path = xstrdup_or_null(path);
1972        gs->buf = NULL;
1973        gs->size = 0;
1974        gs->driver = NULL;
1975
1976        switch (type) {
1977        case GREP_SOURCE_FILE:
1978                gs->identifier = xstrdup(identifier);
1979                break;
1980        case GREP_SOURCE_OID:
1981                gs->identifier = oiddup(identifier);
1982                break;
1983        case GREP_SOURCE_BUF:
1984                gs->identifier = NULL;
1985                break;
1986        }
1987}
1988
1989void grep_source_clear(struct grep_source *gs)
1990{
1991        FREE_AND_NULL(gs->name);
1992        FREE_AND_NULL(gs->path);
1993        FREE_AND_NULL(gs->identifier);
1994        grep_source_clear_data(gs);
1995}
1996
1997void grep_source_clear_data(struct grep_source *gs)
1998{
1999        switch (gs->type) {
2000        case GREP_SOURCE_FILE:
2001        case GREP_SOURCE_OID:
2002                FREE_AND_NULL(gs->buf);
2003                gs->size = 0;
2004                break;
2005        case GREP_SOURCE_BUF:
2006                /* leave user-provided buf intact */
2007                break;
2008        }
2009}
2010
2011static int grep_source_load_oid(struct grep_source *gs)
2012{
2013        enum object_type type;
2014
2015        grep_read_lock();
2016        gs->buf = read_object_file(gs->identifier, &type, &gs->size);
2017        grep_read_unlock();
2018
2019        if (!gs->buf)
2020                return error(_("'%s': unable to read %s"),
2021                             gs->name,
2022                             oid_to_hex(gs->identifier));
2023        return 0;
2024}
2025
2026static int grep_source_load_file(struct grep_source *gs)
2027{
2028        const char *filename = gs->identifier;
2029        struct stat st;
2030        char *data;
2031        size_t size;
2032        int i;
2033
2034        if (lstat(filename, &st) < 0) {
2035        err_ret:
2036                if (errno != ENOENT)
2037                        error_errno(_("failed to stat '%s'"), filename);
2038                return -1;
2039        }
2040        if (!S_ISREG(st.st_mode))
2041                return -1;
2042        size = xsize_t(st.st_size);
2043        i = open(filename, O_RDONLY);
2044        if (i < 0)
2045                goto err_ret;
2046        data = xmallocz(size);
2047        if (st.st_size != read_in_full(i, data, size)) {
2048                error_errno(_("'%s': short read"), filename);
2049                close(i);
2050                free(data);
2051                return -1;
2052        }
2053        close(i);
2054
2055        gs->buf = data;
2056        gs->size = size;
2057        return 0;
2058}
2059
2060static int grep_source_load(struct grep_source *gs)
2061{
2062        if (gs->buf)
2063                return 0;
2064
2065        switch (gs->type) {
2066        case GREP_SOURCE_FILE:
2067                return grep_source_load_file(gs);
2068        case GREP_SOURCE_OID:
2069                return grep_source_load_oid(gs);
2070        case GREP_SOURCE_BUF:
2071                return gs->buf ? 0 : -1;
2072        }
2073        BUG("invalid grep_source type to load");
2074}
2075
2076void grep_source_load_driver(struct grep_source *gs)
2077{
2078        if (gs->driver)
2079                return;
2080
2081        grep_attr_lock();
2082        if (gs->path)
2083                gs->driver = userdiff_find_by_path(gs->path);
2084        if (!gs->driver)
2085                gs->driver = userdiff_find_by_name("default");
2086        grep_attr_unlock();
2087}
2088
2089static int grep_source_is_binary(struct grep_source *gs)
2090{
2091        grep_source_load_driver(gs);
2092        if (gs->driver->binary != -1)
2093                return gs->driver->binary;
2094
2095        if (!grep_source_load(gs))
2096                return buffer_is_binary(gs->buf, gs->size);
2097
2098        return 0;
2099}