Andrew's git / gitweb.git / grep.c
?
grep.con commit Sync with maint-2.10 (a274e0a)
   1#include "cache.h"
   2#include "grep.h"
   3#include "userdiff.h"
   4#include "xdiff-interface.h"
   5#include "diff.h"
   6#include "diffcore.h"
   7#include "commit.h"
   8#include "quote.h"
   9
  10static int grep_source_load(struct grep_source *gs);
  11static int grep_source_is_binary(struct grep_source *gs);
  12
  13static struct grep_opt grep_defaults;
  14
  15/*
  16 * Initialize the grep_defaults template with hardcoded defaults.
  17 * We could let the compiler do this, but without C99 initializers
  18 * the code gets unwieldy and unreadable, so...
  19 */
  20void init_grep_defaults(void)
  21{
  22        struct grep_opt *opt = &grep_defaults;
  23        static int run_once;
  24
  25        if (run_once)
  26                return;
  27        run_once++;
  28
  29        memset(opt, 0, sizeof(*opt));
  30        opt->relative = 1;
  31        opt->pathname = 1;
  32        opt->regflags = REG_NEWLINE;
  33        opt->max_depth = -1;
  34        opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
  35        opt->extended_regexp_option = 0;
  36        color_set(opt->color_context, "");
  37        color_set(opt->color_filename, "");
  38        color_set(opt->color_function, "");
  39        color_set(opt->color_lineno, "");
  40        color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
  41        color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
  42        color_set(opt->color_selected, "");
  43        color_set(opt->color_sep, GIT_COLOR_CYAN);
  44        opt->color = -1;
  45}
  46
  47static int parse_pattern_type_arg(const char *opt, const char *arg)
  48{
  49        if (!strcmp(arg, "default"))
  50                return GREP_PATTERN_TYPE_UNSPECIFIED;
  51        else if (!strcmp(arg, "basic"))
  52                return GREP_PATTERN_TYPE_BRE;
  53        else if (!strcmp(arg, "extended"))
  54                return GREP_PATTERN_TYPE_ERE;
  55        else if (!strcmp(arg, "fixed"))
  56                return GREP_PATTERN_TYPE_FIXED;
  57        else if (!strcmp(arg, "perl"))
  58                return GREP_PATTERN_TYPE_PCRE;
  59        die("bad %s argument: %s", opt, arg);
  60}
  61
  62/*
  63 * Read the configuration file once and store it in
  64 * the grep_defaults template.
  65 */
  66int grep_config(const char *var, const char *value, void *cb)
  67{
  68        struct grep_opt *opt = &grep_defaults;
  69        char *color = NULL;
  70
  71        if (userdiff_config(var, value) < 0)
  72                return -1;
  73
  74        if (!strcmp(var, "grep.extendedregexp")) {
  75                if (git_config_bool(var, value))
  76                        opt->extended_regexp_option = 1;
  77                else
  78                        opt->extended_regexp_option = 0;
  79                return 0;
  80        }
  81
  82        if (!strcmp(var, "grep.patterntype")) {
  83                opt->pattern_type_option = parse_pattern_type_arg(var, value);
  84                return 0;
  85        }
  86
  87        if (!strcmp(var, "grep.linenumber")) {
  88                opt->linenum = git_config_bool(var, value);
  89                return 0;
  90        }
  91
  92        if (!strcmp(var, "grep.fullname")) {
  93                opt->relative = !git_config_bool(var, value);
  94                return 0;
  95        }
  96
  97        if (!strcmp(var, "color.grep"))
  98                opt->color = git_config_colorbool(var, value);
  99        else if (!strcmp(var, "color.grep.context"))
 100                color = opt->color_context;
 101        else if (!strcmp(var, "color.grep.filename"))
 102                color = opt->color_filename;
 103        else if (!strcmp(var, "color.grep.function"))
 104                color = opt->color_function;
 105        else if (!strcmp(var, "color.grep.linenumber"))
 106                color = opt->color_lineno;
 107        else if (!strcmp(var, "color.grep.matchcontext"))
 108                color = opt->color_match_context;
 109        else if (!strcmp(var, "color.grep.matchselected"))
 110                color = opt->color_match_selected;
 111        else if (!strcmp(var, "color.grep.selected"))
 112                color = opt->color_selected;
 113        else if (!strcmp(var, "color.grep.separator"))
 114                color = opt->color_sep;
 115        else if (!strcmp(var, "color.grep.match")) {
 116                int rc = 0;
 117                if (!value)
 118                        return config_error_nonbool(var);
 119                rc |= color_parse(value, opt->color_match_context);
 120                rc |= color_parse(value, opt->color_match_selected);
 121                return rc;
 122        }
 123
 124        if (color) {
 125                if (!value)
 126                        return config_error_nonbool(var);
 127                return color_parse(value, color);
 128        }
 129        return 0;
 130}
 131
 132/*
 133 * Initialize one instance of grep_opt and copy the
 134 * default values from the template we read the configuration
 135 * information in an earlier call to git_config(grep_config).
 136 */
 137void grep_init(struct grep_opt *opt, const char *prefix)
 138{
 139        struct grep_opt *def = &grep_defaults;
 140
 141        memset(opt, 0, sizeof(*opt));
 142        opt->prefix = prefix;
 143        opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
 144        opt->pattern_tail = &opt->pattern_list;
 145        opt->header_tail = &opt->header_list;
 146
 147        opt->color = def->color;
 148        opt->extended_regexp_option = def->extended_regexp_option;
 149        opt->pattern_type_option = def->pattern_type_option;
 150        opt->linenum = def->linenum;
 151        opt->max_depth = def->max_depth;
 152        opt->pathname = def->pathname;
 153        opt->regflags = def->regflags;
 154        opt->relative = def->relative;
 155
 156        color_set(opt->color_context, def->color_context);
 157        color_set(opt->color_filename, def->color_filename);
 158        color_set(opt->color_function, def->color_function);
 159        color_set(opt->color_lineno, def->color_lineno);
 160        color_set(opt->color_match_context, def->color_match_context);
 161        color_set(opt->color_match_selected, def->color_match_selected);
 162        color_set(opt->color_selected, def->color_selected);
 163        color_set(opt->color_sep, def->color_sep);
 164}
 165
 166static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 167{
 168        switch (pattern_type) {
 169        case GREP_PATTERN_TYPE_UNSPECIFIED:
 170                /* fall through */
 171
 172        case GREP_PATTERN_TYPE_BRE:
 173                opt->fixed = 0;
 174                opt->pcre = 0;
 175                opt->regflags &= ~REG_EXTENDED;
 176                break;
 177
 178        case GREP_PATTERN_TYPE_ERE:
 179                opt->fixed = 0;
 180                opt->pcre = 0;
 181                opt->regflags |= REG_EXTENDED;
 182                break;
 183
 184        case GREP_PATTERN_TYPE_FIXED:
 185                opt->fixed = 1;
 186                opt->pcre = 0;
 187                opt->regflags &= ~REG_EXTENDED;
 188                break;
 189
 190        case GREP_PATTERN_TYPE_PCRE:
 191                opt->fixed = 0;
 192                opt->pcre = 1;
 193                opt->regflags &= ~REG_EXTENDED;
 194                break;
 195        }
 196}
 197
 198void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 199{
 200        if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
 201                grep_set_pattern_type_option(pattern_type, opt);
 202        else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
 203                grep_set_pattern_type_option(opt->pattern_type_option, opt);
 204        else if (opt->extended_regexp_option)
 205                grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 206}
 207
 208static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
 209                                        const char *origin, int no,
 210                                        enum grep_pat_token t,
 211                                        enum grep_header_field field)
 212{
 213        struct grep_pat *p = xcalloc(1, sizeof(*p));
 214        p->pattern = xmemdupz(pat, patlen);
 215        p->patternlen = patlen;
 216        p->origin = origin;
 217        p->no = no;
 218        p->token = t;
 219        p->field = field;
 220        return p;
 221}
 222
 223static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
 224{
 225        **tail = p;
 226        *tail = &p->next;
 227        p->next = NULL;
 228
 229        switch (p->token) {
 230        case GREP_PATTERN: /* atom */
 231        case GREP_PATTERN_HEAD:
 232        case GREP_PATTERN_BODY:
 233                for (;;) {
 234                        struct grep_pat *new_pat;
 235                        size_t len = 0;
 236                        char *cp = p->pattern + p->patternlen, *nl = NULL;
 237                        while (++len <= p->patternlen) {
 238                                if (*(--cp) == '\n') {
 239                                        nl = cp;
 240                                        break;
 241                                }
 242                        }
 243                        if (!nl)
 244                                break;
 245                        new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
 246                                                  p->no, p->token, p->field);
 247                        new_pat->next = p->next;
 248                        if (!p->next)
 249                                *tail = &new_pat->next;
 250                        p->next = new_pat;
 251                        *nl = '\0';
 252                        p->patternlen -= len;
 253                }
 254                break;
 255        default:
 256                break;
 257        }
 258}
 259
 260void append_header_grep_pattern(struct grep_opt *opt,
 261                                enum grep_header_field field, const char *pat)
 262{
 263        struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
 264                                             GREP_PATTERN_HEAD, field);
 265        if (field == GREP_HEADER_REFLOG)
 266                opt->use_reflog_filter = 1;
 267        do_append_grep_pat(&opt->header_tail, p);
 268}
 269
 270void append_grep_pattern(struct grep_opt *opt, const char *pat,
 271                         const char *origin, int no, enum grep_pat_token t)
 272{
 273        append_grep_pat(opt, pat, strlen(pat), origin, no, t);
 274}
 275
 276void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
 277                     const char *origin, int no, enum grep_pat_token t)
 278{
 279        struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
 280        do_append_grep_pat(&opt->pattern_tail, p);
 281}
 282
 283struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
 284{
 285        struct grep_pat *pat;
 286        struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
 287        *ret = *opt;
 288
 289        ret->pattern_list = NULL;
 290        ret->pattern_tail = &ret->pattern_list;
 291
 292        for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
 293        {
 294                if(pat->token == GREP_PATTERN_HEAD)
 295                        append_header_grep_pattern(ret, pat->field,
 296                                                   pat->pattern);
 297                else
 298                        append_grep_pat(ret, pat->pattern, pat->patternlen,
 299                                        pat->origin, pat->no, pat->token);
 300        }
 301
 302        return ret;
 303}
 304
 305static NORETURN void compile_regexp_failed(const struct grep_pat *p,
 306                const char *error)
 307{
 308        char where[1024];
 309
 310        if (p->no)
 311                xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
 312        else if (p->origin)
 313                xsnprintf(where, sizeof(where), "%s, ", p->origin);
 314        else
 315                where[0] = 0;
 316
 317        die("%s'%s': %s", where, p->pattern, error);
 318}
 319
 320#ifdef USE_LIBPCRE
 321static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 322{
 323        const char *error;
 324        int erroffset;
 325        int options = PCRE_MULTILINE;
 326
 327        if (opt->ignore_case) {
 328                if (has_non_ascii(p->pattern))
 329                        p->pcre_tables = pcre_maketables();
 330                options |= PCRE_CASELESS;
 331        }
 332        if (is_utf8_locale() && has_non_ascii(p->pattern))
 333                options |= PCRE_UTF8;
 334
 335        p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
 336                                      p->pcre_tables);
 337        if (!p->pcre_regexp)
 338                compile_regexp_failed(p, error);
 339
 340        p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
 341        if (!p->pcre_extra_info && error)
 342                die("%s", error);
 343}
 344
 345static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 346                regmatch_t *match, int eflags)
 347{
 348        int ovector[30], ret, flags = 0;
 349
 350        if (eflags & REG_NOTBOL)
 351                flags |= PCRE_NOTBOL;
 352
 353        ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
 354                        0, flags, ovector, ARRAY_SIZE(ovector));
 355        if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
 356                die("pcre_exec failed with error code %d", ret);
 357        if (ret > 0) {
 358                ret = 0;
 359                match->rm_so = ovector[0];
 360                match->rm_eo = ovector[1];
 361        }
 362
 363        return ret;
 364}
 365
 366static void free_pcre_regexp(struct grep_pat *p)
 367{
 368        pcre_free(p->pcre_regexp);
 369        pcre_free(p->pcre_extra_info);
 370        pcre_free((void *)p->pcre_tables);
 371}
 372#else /* !USE_LIBPCRE */
 373static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 374{
 375        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 376}
 377
 378static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
 379                regmatch_t *match, int eflags)
 380{
 381        return 1;
 382}
 383
 384static void free_pcre_regexp(struct grep_pat *p)
 385{
 386}
 387#endif /* !USE_LIBPCRE */
 388
 389static int is_fixed(const char *s, size_t len)
 390{
 391        size_t i;
 392
 393        /* regcomp cannot accept patterns with NULs so we
 394         * consider any pattern containing a NUL fixed.
 395         */
 396        if (memchr(s, 0, len))
 397                return 1;
 398
 399        for (i = 0; i < len; i++) {
 400                if (is_regex_special(s[i]))
 401                        return 0;
 402        }
 403
 404        return 1;
 405}
 406
 407static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 408{
 409        struct strbuf sb = STRBUF_INIT;
 410        int err;
 411        int regflags;
 412
 413        basic_regex_quote_buf(&sb, p->pattern);
 414        regflags = opt->regflags & ~REG_EXTENDED;
 415        if (opt->ignore_case)
 416                regflags |= REG_ICASE;
 417        err = regcomp(&p->regexp, sb.buf, regflags);
 418        if (opt->debug)
 419                fprintf(stderr, "fixed %s\n", sb.buf);
 420        strbuf_release(&sb);
 421        if (err) {
 422                char errbuf[1024];
 423                regerror(err, &p->regexp, errbuf, sizeof(errbuf));
 424                regfree(&p->regexp);
 425                compile_regexp_failed(p, errbuf);
 426        }
 427}
 428
 429static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 430{
 431        int icase, ascii_only;
 432        int err;
 433
 434        p->word_regexp = opt->word_regexp;
 435        p->ignore_case = opt->ignore_case;
 436        icase          = opt->regflags & REG_ICASE || p->ignore_case;
 437        ascii_only     = !has_non_ascii(p->pattern);
 438
 439        /*
 440         * Even when -F (fixed) asks us to do a non-regexp search, we
 441         * may not be able to correctly case-fold when -i
 442         * (ignore-case) is asked (in which case, we'll synthesize a
 443         * regexp to match the pattern that matches regexp special
 444         * characters literally, while ignoring case differences).  On
 445         * the other hand, even without -F, if the pattern does not
 446         * have any regexp special characters and there is no need for
 447         * case-folding search, we can internally turn it into a
 448         * simple string match using kws.  p->fixed tells us if we
 449         * want to use kws.
 450         */
 451        if (opt->fixed || is_fixed(p->pattern, p->patternlen))
 452                p->fixed = !icase || ascii_only;
 453        else
 454                p->fixed = 0;
 455
 456        if (p->fixed) {
 457                p->kws = kwsalloc(icase ? tolower_trans_tbl : NULL);
 458                kwsincr(p->kws, p->pattern, p->patternlen);
 459                kwsprep(p->kws);
 460                return;
 461        } else if (opt->fixed) {
 462                /*
 463                 * We come here when the pattern has the non-ascii
 464                 * characters we cannot case-fold, and asked to
 465                 * ignore-case.
 466                 */
 467                compile_fixed_regexp(p, opt);
 468                return;
 469        }
 470
 471        if (opt->pcre) {
 472                compile_pcre_regexp(p, opt);
 473                return;
 474        }
 475
 476        err = regcomp(&p->regexp, p->pattern, opt->regflags);
 477        if (err) {
 478                char errbuf[1024];
 479                regerror(err, &p->regexp, errbuf, 1024);
 480                regfree(&p->regexp);
 481                compile_regexp_failed(p, errbuf);
 482        }
 483}
 484
 485static struct grep_expr *compile_pattern_or(struct grep_pat **);
 486static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
 487{
 488        struct grep_pat *p;
 489        struct grep_expr *x;
 490
 491        p = *list;
 492        if (!p)
 493                return NULL;
 494        switch (p->token) {
 495        case GREP_PATTERN: /* atom */
 496        case GREP_PATTERN_HEAD:
 497        case GREP_PATTERN_BODY:
 498                x = xcalloc(1, sizeof (struct grep_expr));
 499                x->node = GREP_NODE_ATOM;
 500                x->u.atom = p;
 501                *list = p->next;
 502                return x;
 503        case GREP_OPEN_PAREN:
 504                *list = p->next;
 505                x = compile_pattern_or(list);
 506                if (!*list || (*list)->token != GREP_CLOSE_PAREN)
 507                        die("unmatched parenthesis");
 508                *list = (*list)->next;
 509                return x;
 510        default:
 511                return NULL;
 512        }
 513}
 514
 515static struct grep_expr *compile_pattern_not(struct grep_pat **list)
 516{
 517        struct grep_pat *p;
 518        struct grep_expr *x;
 519
 520        p = *list;
 521        if (!p)
 522                return NULL;
 523        switch (p->token) {
 524        case GREP_NOT:
 525                if (!p->next)
 526                        die("--not not followed by pattern expression");
 527                *list = p->next;
 528                x = xcalloc(1, sizeof (struct grep_expr));
 529                x->node = GREP_NODE_NOT;
 530                x->u.unary = compile_pattern_not(list);
 531                if (!x->u.unary)
 532                        die("--not followed by non pattern expression");
 533                return x;
 534        default:
 535                return compile_pattern_atom(list);
 536        }
 537}
 538
 539static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 540{
 541        struct grep_pat *p;
 542        struct grep_expr *x, *y, *z;
 543
 544        x = compile_pattern_not(list);
 545        p = *list;
 546        if (p && p->token == GREP_AND) {
 547                if (!p->next)
 548                        die("--and not followed by pattern expression");
 549                *list = p->next;
 550                y = compile_pattern_and(list);
 551                if (!y)
 552                        die("--and not followed by pattern expression");
 553                z = xcalloc(1, sizeof (struct grep_expr));
 554                z->node = GREP_NODE_AND;
 555                z->u.binary.left = x;
 556                z->u.binary.right = y;
 557                return z;
 558        }
 559        return x;
 560}
 561
 562static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 563{
 564        struct grep_pat *p;
 565        struct grep_expr *x, *y, *z;
 566
 567        x = compile_pattern_and(list);
 568        p = *list;
 569        if (x && p && p->token != GREP_CLOSE_PAREN) {
 570                y = compile_pattern_or(list);
 571                if (!y)
 572                        die("not a pattern expression %s", p->pattern);
 573                z = xcalloc(1, sizeof (struct grep_expr));
 574                z->node = GREP_NODE_OR;
 575                z->u.binary.left = x;
 576                z->u.binary.right = y;
 577                return z;
 578        }
 579        return x;
 580}
 581
 582static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 583{
 584        return compile_pattern_or(list);
 585}
 586
 587static void indent(int in)
 588{
 589        while (in-- > 0)
 590                fputc(' ', stderr);
 591}
 592
 593static void dump_grep_pat(struct grep_pat *p)
 594{
 595        switch (p->token) {
 596        case GREP_AND: fprintf(stderr, "*and*"); break;
 597        case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
 598        case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
 599        case GREP_NOT: fprintf(stderr, "*not*"); break;
 600        case GREP_OR: fprintf(stderr, "*or*"); break;
 601
 602        case GREP_PATTERN: fprintf(stderr, "pattern"); break;
 603        case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
 604        case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
 605        }
 606
 607        switch (p->token) {
 608        default: break;
 609        case GREP_PATTERN_HEAD:
 610                fprintf(stderr, "<head %d>", p->field); break;
 611        case GREP_PATTERN_BODY:
 612                fprintf(stderr, "<body>"); break;
 613        }
 614        switch (p->token) {
 615        default: break;
 616        case GREP_PATTERN_HEAD:
 617        case GREP_PATTERN_BODY:
 618        case GREP_PATTERN:
 619                fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
 620                break;
 621        }
 622        fputc('\n', stderr);
 623}
 624
 625static void dump_grep_expression_1(struct grep_expr *x, int in)
 626{
 627        indent(in);
 628        switch (x->node) {
 629        case GREP_NODE_TRUE:
 630                fprintf(stderr, "true\n");
 631                break;
 632        case GREP_NODE_ATOM:
 633                dump_grep_pat(x->u.atom);
 634                break;
 635        case GREP_NODE_NOT:
 636                fprintf(stderr, "(not\n");
 637                dump_grep_expression_1(x->u.unary, in+1);
 638                indent(in);
 639                fprintf(stderr, ")\n");
 640                break;
 641        case GREP_NODE_AND:
 642                fprintf(stderr, "(and\n");
 643                dump_grep_expression_1(x->u.binary.left, in+1);
 644                dump_grep_expression_1(x->u.binary.right, in+1);
 645                indent(in);
 646                fprintf(stderr, ")\n");
 647                break;
 648        case GREP_NODE_OR:
 649                fprintf(stderr, "(or\n");
 650                dump_grep_expression_1(x->u.binary.left, in+1);
 651                dump_grep_expression_1(x->u.binary.right, in+1);
 652                indent(in);
 653                fprintf(stderr, ")\n");
 654                break;
 655        }
 656}
 657
 658static void dump_grep_expression(struct grep_opt *opt)
 659{
 660        struct grep_expr *x = opt->pattern_expression;
 661
 662        if (opt->all_match)
 663                fprintf(stderr, "[all-match]\n");
 664        dump_grep_expression_1(x, 0);
 665        fflush(NULL);
 666}
 667
 668static struct grep_expr *grep_true_expr(void)
 669{
 670        struct grep_expr *z = xcalloc(1, sizeof(*z));
 671        z->node = GREP_NODE_TRUE;
 672        return z;
 673}
 674
 675static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
 676{
 677        struct grep_expr *z = xcalloc(1, sizeof(*z));
 678        z->node = GREP_NODE_OR;
 679        z->u.binary.left = left;
 680        z->u.binary.right = right;
 681        return z;
 682}
 683
 684static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 685{
 686        struct grep_pat *p;
 687        struct grep_expr *header_expr;
 688        struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
 689        enum grep_header_field fld;
 690
 691        if (!opt->header_list)
 692                return NULL;
 693
 694        for (p = opt->header_list; p; p = p->next) {
 695                if (p->token != GREP_PATTERN_HEAD)
 696                        die("BUG: a non-header pattern in grep header list.");
 697                if (p->field < GREP_HEADER_FIELD_MIN ||
 698                    GREP_HEADER_FIELD_MAX <= p->field)
 699                        die("BUG: unknown header field %d", p->field);
 700                compile_regexp(p, opt);
 701        }
 702
 703        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
 704                header_group[fld] = NULL;
 705
 706        for (p = opt->header_list; p; p = p->next) {
 707                struct grep_expr *h;
 708                struct grep_pat *pp = p;
 709
 710                h = compile_pattern_atom(&pp);
 711                if (!h || pp != p->next)
 712                        die("BUG: malformed header expr");
 713                if (!header_group[p->field]) {
 714                        header_group[p->field] = h;
 715                        continue;
 716                }
 717                header_group[p->field] = grep_or_expr(h, header_group[p->field]);
 718        }
 719
 720        header_expr = NULL;
 721
 722        for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
 723                if (!header_group[fld])
 724                        continue;
 725                if (!header_expr)
 726                        header_expr = grep_true_expr();
 727                header_expr = grep_or_expr(header_group[fld], header_expr);
 728        }
 729        return header_expr;
 730}
 731
 732static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
 733{
 734        struct grep_expr *z = x;
 735
 736        while (x) {
 737                assert(x->node == GREP_NODE_OR);
 738                if (x->u.binary.right &&
 739                    x->u.binary.right->node == GREP_NODE_TRUE) {
 740                        x->u.binary.right = y;
 741                        break;
 742                }
 743                x = x->u.binary.right;
 744        }
 745        return z;
 746}
 747
 748static void compile_grep_patterns_real(struct grep_opt *opt)
 749{
 750        struct grep_pat *p;
 751        struct grep_expr *header_expr = prep_header_patterns(opt);
 752
 753        for (p = opt->pattern_list; p; p = p->next) {
 754                switch (p->token) {
 755                case GREP_PATTERN: /* atom */
 756                case GREP_PATTERN_HEAD:
 757                case GREP_PATTERN_BODY:
 758                        compile_regexp(p, opt);
 759                        break;
 760                default:
 761                        opt->extended = 1;
 762                        break;
 763                }
 764        }
 765
 766        if (opt->all_match || header_expr)
 767                opt->extended = 1;
 768        else if (!opt->extended && !opt->debug)
 769                return;
 770
 771        p = opt->pattern_list;
 772        if (p)
 773                opt->pattern_expression = compile_pattern_expr(&p);
 774        if (p)
 775                die("incomplete pattern expression: %s", p->pattern);
 776
 777        if (!header_expr)
 778                return;
 779
 780        if (!opt->pattern_expression)
 781                opt->pattern_expression = header_expr;
 782        else if (opt->all_match)
 783                opt->pattern_expression = grep_splice_or(header_expr,
 784                                                         opt->pattern_expression);
 785        else
 786                opt->pattern_expression = grep_or_expr(opt->pattern_expression,
 787                                                       header_expr);
 788        opt->all_match = 1;
 789}
 790
 791void compile_grep_patterns(struct grep_opt *opt)
 792{
 793        compile_grep_patterns_real(opt);
 794        if (opt->debug)
 795                dump_grep_expression(opt);
 796}
 797
 798static void free_pattern_expr(struct grep_expr *x)
 799{
 800        switch (x->node) {
 801        case GREP_NODE_TRUE:
 802        case GREP_NODE_ATOM:
 803                break;
 804        case GREP_NODE_NOT:
 805                free_pattern_expr(x->u.unary);
 806                break;
 807        case GREP_NODE_AND:
 808        case GREP_NODE_OR:
 809                free_pattern_expr(x->u.binary.left);
 810                free_pattern_expr(x->u.binary.right);
 811                break;
 812        }
 813        free(x);
 814}
 815
 816void free_grep_patterns(struct grep_opt *opt)
 817{
 818        struct grep_pat *p, *n;
 819
 820        for (p = opt->pattern_list; p; p = n) {
 821                n = p->next;
 822                switch (p->token) {
 823                case GREP_PATTERN: /* atom */
 824                case GREP_PATTERN_HEAD:
 825                case GREP_PATTERN_BODY:
 826                        if (p->kws)
 827                                kwsfree(p->kws);
 828                        else if (p->pcre_regexp)
 829                                free_pcre_regexp(p);
 830                        else
 831                                regfree(&p->regexp);
 832                        free(p->pattern);
 833                        break;
 834                default:
 835                        break;
 836                }
 837                free(p);
 838        }
 839
 840        if (!opt->extended)
 841                return;
 842        free_pattern_expr(opt->pattern_expression);
 843}
 844
 845static char *end_of_line(char *cp, unsigned long *left)
 846{
 847        unsigned long l = *left;
 848        while (l && *cp != '\n') {
 849                l--;
 850                cp++;
 851        }
 852        *left = l;
 853        return cp;
 854}
 855
 856static int word_char(char ch)
 857{
 858        return isalnum(ch) || ch == '_';
 859}
 860
 861static void output_color(struct grep_opt *opt, const void *data, size_t size,
 862                         const char *color)
 863{
 864        if (want_color(opt->color) && color && color[0]) {
 865                opt->output(opt, color, strlen(color));
 866                opt->output(opt, data, size);
 867                opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
 868        } else
 869                opt->output(opt, data, size);
 870}
 871
 872static void output_sep(struct grep_opt *opt, char sign)
 873{
 874        if (opt->null_following_name)
 875                opt->output(opt, "\0", 1);
 876        else
 877                output_color(opt, &sign, 1, opt->color_sep);
 878}
 879
 880static void show_name(struct grep_opt *opt, const char *name)
 881{
 882        output_color(opt, name, strlen(name), opt->color_filename);
 883        opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 884}
 885
 886static int fixmatch(struct grep_pat *p, char *line, char *eol,
 887                    regmatch_t *match)
 888{
 889        struct kwsmatch kwsm;
 890        size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
 891        if (offset == -1) {
 892                match->rm_so = match->rm_eo = -1;
 893                return REG_NOMATCH;
 894        } else {
 895                match->rm_so = offset;
 896                match->rm_eo = match->rm_so + kwsm.size[0];
 897                return 0;
 898        }
 899}
 900
 901static int patmatch(struct grep_pat *p, char *line, char *eol,
 902                    regmatch_t *match, int eflags)
 903{
 904        int hit;
 905
 906        if (p->fixed)
 907                hit = !fixmatch(p, line, eol, match);
 908        else if (p->pcre_regexp)
 909                hit = !pcrematch(p, line, eol, match, eflags);
 910        else
 911                hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
 912                                   eflags);
 913
 914        return hit;
 915}
 916
 917static int strip_timestamp(char *bol, char **eol_p)
 918{
 919        char *eol = *eol_p;
 920        int ch;
 921
 922        while (bol < --eol) {
 923                if (*eol != '>')
 924                        continue;
 925                *eol_p = ++eol;
 926                ch = *eol;
 927                *eol = '\0';
 928                return ch;
 929        }
 930        return 0;
 931}
 932
 933static struct {
 934        const char *field;
 935        size_t len;
 936} header_field[] = {
 937        { "author ", 7 },
 938        { "committer ", 10 },
 939        { "reflog ", 7 },
 940};
 941
 942static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 943                             enum grep_context ctx,
 944                             regmatch_t *pmatch, int eflags)
 945{
 946        int hit = 0;
 947        int saved_ch = 0;
 948        const char *start = bol;
 949
 950        if ((p->token != GREP_PATTERN) &&
 951            ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 952                return 0;
 953
 954        if (p->token == GREP_PATTERN_HEAD) {
 955                const char *field;
 956                size_t len;
 957                assert(p->field < ARRAY_SIZE(header_field));
 958                field = header_field[p->field].field;
 959                len = header_field[p->field].len;
 960                if (strncmp(bol, field, len))
 961                        return 0;
 962                bol += len;
 963                switch (p->field) {
 964                case GREP_HEADER_AUTHOR:
 965                case GREP_HEADER_COMMITTER:
 966                        saved_ch = strip_timestamp(bol, &eol);
 967                        break;
 968                default:
 969                        break;
 970                }
 971        }
 972
 973 again:
 974        hit = patmatch(p, bol, eol, pmatch, eflags);
 975
 976        if (hit && p->word_regexp) {
 977                if ((pmatch[0].rm_so < 0) ||
 978                    (eol - bol) < pmatch[0].rm_so ||
 979                    (pmatch[0].rm_eo < 0) ||
 980                    (eol - bol) < pmatch[0].rm_eo)
 981                        die("regexp returned nonsense");
 982
 983                /* Match beginning must be either beginning of the
 984                 * line, or at word boundary (i.e. the last char must
 985                 * not be a word char).  Similarly, match end must be
 986                 * either end of the line, or at word boundary
 987                 * (i.e. the next char must not be a word char).
 988                 */
 989                if ( ((pmatch[0].rm_so == 0) ||
 990                      !word_char(bol[pmatch[0].rm_so-1])) &&
 991                     ((pmatch[0].rm_eo == (eol-bol)) ||
 992                      !word_char(bol[pmatch[0].rm_eo])) )
 993                        ;
 994                else
 995                        hit = 0;
 996
 997                /* Words consist of at least one character. */
 998                if (pmatch->rm_so == pmatch->rm_eo)
 999                        hit = 0;
1000
1001                if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1002                        /* There could be more than one match on the
1003                         * line, and the first match might not be
1004                         * strict word match.  But later ones could be!
1005                         * Forward to the next possible start, i.e. the
1006                         * next position following a non-word char.
1007                         */
1008                        bol = pmatch[0].rm_so + bol + 1;
1009                        while (word_char(bol[-1]) && bol < eol)
1010                                bol++;
1011                        eflags |= REG_NOTBOL;
1012                        if (bol < eol)
1013                                goto again;
1014                }
1015        }
1016        if (p->token == GREP_PATTERN_HEAD && saved_ch)
1017                *eol = saved_ch;
1018        if (hit) {
1019                pmatch[0].rm_so += bol - start;
1020                pmatch[0].rm_eo += bol - start;
1021        }
1022        return hit;
1023}
1024
1025static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1026                           enum grep_context ctx, int collect_hits)
1027{
1028        int h = 0;
1029        regmatch_t match;
1030
1031        if (!x)
1032                die("Not a valid grep expression");
1033        switch (x->node) {
1034        case GREP_NODE_TRUE:
1035                h = 1;
1036                break;
1037        case GREP_NODE_ATOM:
1038                h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1039                break;
1040        case GREP_NODE_NOT:
1041                h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1042                break;
1043        case GREP_NODE_AND:
1044                if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1045                        return 0;
1046                h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1047                break;
1048        case GREP_NODE_OR:
1049                if (!collect_hits)
1050                        return (match_expr_eval(x->u.binary.left,
1051                                                bol, eol, ctx, 0) ||
1052                                match_expr_eval(x->u.binary.right,
1053                                                bol, eol, ctx, 0));
1054                h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1055                x->u.binary.left->hit |= h;
1056                h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1057                break;
1058        default:
1059                die("Unexpected node type (internal error) %d", x->node);
1060        }
1061        if (collect_hits)
1062                x->hit |= h;
1063        return h;
1064}
1065
1066static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1067                      enum grep_context ctx, int collect_hits)
1068{
1069        struct grep_expr *x = opt->pattern_expression;
1070        return match_expr_eval(x, bol, eol, ctx, collect_hits);
1071}
1072
1073static int match_line(struct grep_opt *opt, char *bol, char *eol,
1074                      enum grep_context ctx, int collect_hits)
1075{
1076        struct grep_pat *p;
1077        regmatch_t match;
1078
1079        if (opt->extended)
1080                return match_expr(opt, bol, eol, ctx, collect_hits);
1081
1082        /* we do not call with collect_hits without being extended */
1083        for (p = opt->pattern_list; p; p = p->next) {
1084                if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1085                        return 1;
1086        }
1087        return 0;
1088}
1089
1090static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1091                              enum grep_context ctx,
1092                              regmatch_t *pmatch, int eflags)
1093{
1094        regmatch_t match;
1095
1096        if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1097                return 0;
1098        if (match.rm_so < 0 || match.rm_eo < 0)
1099                return 0;
1100        if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1101                if (match.rm_so > pmatch->rm_so)
1102                        return 1;
1103                if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1104                        return 1;
1105        }
1106        pmatch->rm_so = match.rm_so;
1107        pmatch->rm_eo = match.rm_eo;
1108        return 1;
1109}
1110
1111static int next_match(struct grep_opt *opt, char *bol, char *eol,
1112                      enum grep_context ctx, regmatch_t *pmatch, int eflags)
1113{
1114        struct grep_pat *p;
1115        int hit = 0;
1116
1117        pmatch->rm_so = pmatch->rm_eo = -1;
1118        if (bol < eol) {
1119                for (p = opt->pattern_list; p; p = p->next) {
1120                        switch (p->token) {
1121                        case GREP_PATTERN: /* atom */
1122                        case GREP_PATTERN_HEAD:
1123                        case GREP_PATTERN_BODY:
1124                                hit |= match_next_pattern(p, bol, eol, ctx,
1125                                                          pmatch, eflags);
1126                                break;
1127                        default:
1128                                break;
1129                        }
1130                }
1131        }
1132        return hit;
1133}
1134
1135static void show_line(struct grep_opt *opt, char *bol, char *eol,
1136                      const char *name, unsigned lno, char sign)
1137{
1138        int rest = eol - bol;
1139        const char *match_color, *line_color = NULL;
1140
1141        if (opt->file_break && opt->last_shown == 0) {
1142                if (opt->show_hunk_mark)
1143                        opt->output(opt, "\n", 1);
1144        } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1145                if (opt->last_shown == 0) {
1146                        if (opt->show_hunk_mark) {
1147                                output_color(opt, "--", 2, opt->color_sep);
1148                                opt->output(opt, "\n", 1);
1149                        }
1150                } else if (lno > opt->last_shown + 1) {
1151                        output_color(opt, "--", 2, opt->color_sep);
1152                        opt->output(opt, "\n", 1);
1153                }
1154        }
1155        if (opt->heading && opt->last_shown == 0) {
1156                output_color(opt, name, strlen(name), opt->color_filename);
1157                opt->output(opt, "\n", 1);
1158        }
1159        opt->last_shown = lno;
1160
1161        if (!opt->heading && opt->pathname) {
1162                output_color(opt, name, strlen(name), opt->color_filename);
1163                output_sep(opt, sign);
1164        }
1165        if (opt->linenum) {
1166                char buf[32];
1167                snprintf(buf, sizeof(buf), "%d", lno);
1168                output_color(opt, buf, strlen(buf), opt->color_lineno);
1169                output_sep(opt, sign);
1170        }
1171        if (opt->color) {
1172                regmatch_t match;
1173                enum grep_context ctx = GREP_CONTEXT_BODY;
1174                int ch = *eol;
1175                int eflags = 0;
1176
1177                if (sign == ':')
1178                        match_color = opt->color_match_selected;
1179                else
1180                        match_color = opt->color_match_context;
1181                if (sign == ':')
1182                        line_color = opt->color_selected;
1183                else if (sign == '-')
1184                        line_color = opt->color_context;
1185                else if (sign == '=')
1186                        line_color = opt->color_function;
1187                *eol = '\0';
1188                while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1189                        if (match.rm_so == match.rm_eo)
1190                                break;
1191
1192                        output_color(opt, bol, match.rm_so, line_color);
1193                        output_color(opt, bol + match.rm_so,
1194                                     match.rm_eo - match.rm_so, match_color);
1195                        bol += match.rm_eo;
1196                        rest -= match.rm_eo;
1197                        eflags = REG_NOTBOL;
1198                }
1199                *eol = ch;
1200        }
1201        output_color(opt, bol, rest, line_color);
1202        opt->output(opt, "\n", 1);
1203}
1204
1205#ifndef NO_PTHREADS
1206int grep_use_locks;
1207
1208/*
1209 * This lock protects access to the gitattributes machinery, which is
1210 * not thread-safe.
1211 */
1212pthread_mutex_t grep_attr_mutex;
1213
1214static inline void grep_attr_lock(void)
1215{
1216        if (grep_use_locks)
1217                pthread_mutex_lock(&grep_attr_mutex);
1218}
1219
1220static inline void grep_attr_unlock(void)
1221{
1222        if (grep_use_locks)
1223                pthread_mutex_unlock(&grep_attr_mutex);
1224}
1225
1226/*
1227 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1228 */
1229pthread_mutex_t grep_read_mutex;
1230
1231#else
1232#define grep_attr_lock()
1233#define grep_attr_unlock()
1234#endif
1235
1236static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1237{
1238        xdemitconf_t *xecfg = opt->priv;
1239        if (xecfg && !xecfg->find_func) {
1240                grep_source_load_driver(gs);
1241                if (gs->driver->funcname.pattern) {
1242                        const struct userdiff_funcname *pe = &gs->driver->funcname;
1243                        xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1244                } else {
1245                        xecfg = opt->priv = NULL;
1246                }
1247        }
1248
1249        if (xecfg) {
1250                char buf[1];
1251                return xecfg->find_func(bol, eol - bol, buf, 1,
1252                                        xecfg->find_func_priv) >= 0;
1253        }
1254
1255        if (bol == eol)
1256                return 0;
1257        if (isalpha(*bol) || *bol == '_' || *bol == '$')
1258                return 1;
1259        return 0;
1260}
1261
1262static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1263                               char *bol, unsigned lno)
1264{
1265        while (bol > gs->buf) {
1266                char *eol = --bol;
1267
1268                while (bol > gs->buf && bol[-1] != '\n')
1269                        bol--;
1270                lno--;
1271
1272                if (lno <= opt->last_shown)
1273                        break;
1274
1275                if (match_funcname(opt, gs, bol, eol)) {
1276                        show_line(opt, bol, eol, gs->name, lno, '=');
1277                        break;
1278                }
1279        }
1280}
1281
1282static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1283                             char *bol, char *end, unsigned lno)
1284{
1285        unsigned cur = lno, from = 1, funcname_lno = 0;
1286        int funcname_needed = !!opt->funcname;
1287
1288        if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1289                funcname_needed = 2;
1290
1291        if (opt->pre_context < lno)
1292                from = lno - opt->pre_context;
1293        if (from <= opt->last_shown)
1294                from = opt->last_shown + 1;
1295
1296        /* Rewind. */
1297        while (bol > gs->buf &&
1298               cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1299                char *eol = --bol;
1300
1301                while (bol > gs->buf && bol[-1] != '\n')
1302                        bol--;
1303                cur--;
1304                if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1305                        funcname_lno = cur;
1306                        funcname_needed = 0;
1307                }
1308        }
1309
1310        /* We need to look even further back to find a function signature. */
1311        if (opt->funcname && funcname_needed)
1312                show_funcname_line(opt, gs, bol, cur);
1313
1314        /* Back forward. */
1315        while (cur < lno) {
1316                char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1317
1318                while (*eol != '\n')
1319                        eol++;
1320                show_line(opt, bol, eol, gs->name, cur, sign);
1321                bol = eol + 1;
1322                cur++;
1323        }
1324}
1325
1326static int should_lookahead(struct grep_opt *opt)
1327{
1328        struct grep_pat *p;
1329
1330        if (opt->extended)
1331                return 0; /* punt for too complex stuff */
1332        if (opt->invert)
1333                return 0;
1334        for (p = opt->pattern_list; p; p = p->next) {
1335                if (p->token != GREP_PATTERN)
1336                        return 0; /* punt for "header only" and stuff */
1337        }
1338        return 1;
1339}
1340
1341static int look_ahead(struct grep_opt *opt,
1342                      unsigned long *left_p,
1343                      unsigned *lno_p,
1344                      char **bol_p)
1345{
1346        unsigned lno = *lno_p;
1347        char *bol = *bol_p;
1348        struct grep_pat *p;
1349        char *sp, *last_bol;
1350        regoff_t earliest = -1;
1351
1352        for (p = opt->pattern_list; p; p = p->next) {
1353                int hit;
1354                regmatch_t m;
1355
1356                hit = patmatch(p, bol, bol + *left_p, &m, 0);
1357                if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1358                        continue;
1359                if (earliest < 0 || m.rm_so < earliest)
1360                        earliest = m.rm_so;
1361        }
1362
1363        if (earliest < 0) {
1364                *bol_p = bol + *left_p;
1365                *left_p = 0;
1366                return 1;
1367        }
1368        for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1369                ; /* find the beginning of the line */
1370        last_bol = sp;
1371
1372        for (sp = bol; sp < last_bol; sp++) {
1373                if (*sp == '\n')
1374                        lno++;
1375        }
1376        *left_p -= last_bol - bol;
1377        *bol_p = last_bol;
1378        *lno_p = lno;
1379        return 0;
1380}
1381
1382static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1383{
1384        fwrite(buf, size, 1, stdout);
1385}
1386
1387static int fill_textconv_grep(struct userdiff_driver *driver,
1388                              struct grep_source *gs)
1389{
1390        struct diff_filespec *df;
1391        char *buf;
1392        size_t size;
1393
1394        if (!driver || !driver->textconv)
1395                return grep_source_load(gs);
1396
1397        /*
1398         * The textconv interface is intimately tied to diff_filespecs, so we
1399         * have to pretend to be one. If we could unify the grep_source
1400         * and diff_filespec structs, this mess could just go away.
1401         */
1402        df = alloc_filespec(gs->path);
1403        switch (gs->type) {
1404        case GREP_SOURCE_SHA1:
1405                fill_filespec(df, gs->identifier, 1, 0100644);
1406                break;
1407        case GREP_SOURCE_FILE:
1408                fill_filespec(df, null_sha1, 0, 0100644);
1409                break;
1410        default:
1411                die("BUG: attempt to textconv something without a path?");
1412        }
1413
1414        /*
1415         * fill_textconv is not remotely thread-safe; it may load objects
1416         * behind the scenes, and it modifies the global diff tempfile
1417         * structure.
1418         */
1419        grep_read_lock();
1420        size = fill_textconv(driver, df, &buf);
1421        grep_read_unlock();
1422        free_filespec(df);
1423
1424        /*
1425         * The normal fill_textconv usage by the diff machinery would just keep
1426         * the textconv'd buf separate from the diff_filespec. But much of the
1427         * grep code passes around a grep_source and assumes that its "buf"
1428         * pointer is the beginning of the thing we are searching. So let's
1429         * install our textconv'd version into the grep_source, taking care not
1430         * to leak any existing buffer.
1431         */
1432        grep_source_clear_data(gs);
1433        gs->buf = buf;
1434        gs->size = size;
1435
1436        return 0;
1437}
1438
1439static int is_empty_line(const char *bol, const char *eol)
1440{
1441        while (bol < eol && isspace(*bol))
1442                bol++;
1443        return bol == eol;
1444}
1445
1446static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1447{
1448        char *bol;
1449        char *peek_bol = NULL;
1450        unsigned long left;
1451        unsigned lno = 1;
1452        unsigned last_hit = 0;
1453        int binary_match_only = 0;
1454        unsigned count = 0;
1455        int try_lookahead = 0;
1456        int show_function = 0;
1457        struct userdiff_driver *textconv = NULL;
1458        enum grep_context ctx = GREP_CONTEXT_HEAD;
1459        xdemitconf_t xecfg;
1460
1461        if (!opt->output)
1462                opt->output = std_output;
1463
1464        if (opt->pre_context || opt->post_context || opt->file_break ||
1465            opt->funcbody) {
1466                /* Show hunk marks, except for the first file. */
1467                if (opt->last_shown)
1468                        opt->show_hunk_mark = 1;
1469                /*
1470                 * If we're using threads then we can't easily identify
1471                 * the first file.  Always put hunk marks in that case
1472                 * and skip the very first one later in work_done().
1473                 */
1474                if (opt->output != std_output)
1475                        opt->show_hunk_mark = 1;
1476        }
1477        opt->last_shown = 0;
1478
1479        if (opt->allow_textconv) {
1480                grep_source_load_driver(gs);
1481                /*
1482                 * We might set up the shared textconv cache data here, which
1483                 * is not thread-safe.
1484                 */
1485                grep_attr_lock();
1486                textconv = userdiff_get_textconv(gs->driver);
1487                grep_attr_unlock();
1488        }
1489
1490        /*
1491         * We know the result of a textconv is text, so we only have to care
1492         * about binary handling if we are not using it.
1493         */
1494        if (!textconv) {
1495                switch (opt->binary) {
1496                case GREP_BINARY_DEFAULT:
1497                        if (grep_source_is_binary(gs))
1498                                binary_match_only = 1;
1499                        break;
1500                case GREP_BINARY_NOMATCH:
1501                        if (grep_source_is_binary(gs))
1502                                return 0; /* Assume unmatch */
1503                        break;
1504                case GREP_BINARY_TEXT:
1505                        break;
1506                default:
1507                        die("BUG: unknown binary handling mode");
1508                }
1509        }
1510
1511        memset(&xecfg, 0, sizeof(xecfg));
1512        opt->priv = &xecfg;
1513
1514        try_lookahead = should_lookahead(opt);
1515
1516        if (fill_textconv_grep(textconv, gs) < 0)
1517                return 0;
1518
1519        bol = gs->buf;
1520        left = gs->size;
1521        while (left) {
1522                char *eol, ch;
1523                int hit;
1524
1525                /*
1526                 * look_ahead() skips quickly to the line that possibly
1527                 * has the next hit; don't call it if we need to do
1528                 * something more than just skipping the current line
1529                 * in response to an unmatch for the current line.  E.g.
1530                 * inside a post-context window, we will show the current
1531                 * line as a context around the previous hit when it
1532                 * doesn't hit.
1533                 */
1534                if (try_lookahead
1535                    && !(last_hit
1536                         && (show_function ||
1537                             lno <= last_hit + opt->post_context))
1538                    && look_ahead(opt, &left, &lno, &bol))
1539                        break;
1540                eol = end_of_line(bol, &left);
1541                ch = *eol;
1542                *eol = 0;
1543
1544                if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1545                        ctx = GREP_CONTEXT_BODY;
1546
1547                hit = match_line(opt, bol, eol, ctx, collect_hits);
1548                *eol = ch;
1549
1550                if (collect_hits)
1551                        goto next_line;
1552
1553                /* "grep -v -e foo -e bla" should list lines
1554                 * that do not have either, so inversion should
1555                 * be done outside.
1556                 */
1557                if (opt->invert)
1558                        hit = !hit;
1559                if (opt->unmatch_name_only) {
1560                        if (hit)
1561                                return 0;
1562                        goto next_line;
1563                }
1564                if (hit) {
1565                        count++;
1566                        if (opt->status_only)
1567                                return 1;
1568                        if (opt->name_only) {
1569                                show_name(opt, gs->name);
1570                                return 1;
1571                        }
1572                        if (opt->count)
1573                                goto next_line;
1574                        if (binary_match_only) {
1575                                opt->output(opt, "Binary file ", 12);
1576                                output_color(opt, gs->name, strlen(gs->name),
1577                                             opt->color_filename);
1578                                opt->output(opt, " matches\n", 9);
1579                                return 1;
1580                        }
1581                        /* Hit at this line.  If we haven't shown the
1582                         * pre-context lines, we would need to show them.
1583                         */
1584                        if (opt->pre_context || opt->funcbody)
1585                                show_pre_context(opt, gs, bol, eol, lno);
1586                        else if (opt->funcname)
1587                                show_funcname_line(opt, gs, bol, lno);
1588                        show_line(opt, bol, eol, gs->name, lno, ':');
1589                        last_hit = lno;
1590                        if (opt->funcbody)
1591                                show_function = 1;
1592                        goto next_line;
1593                }
1594                if (show_function && (!peek_bol || peek_bol < bol)) {
1595                        unsigned long peek_left = left;
1596                        char *peek_eol = eol;
1597
1598                        /*
1599                         * Trailing empty lines are not interesting.
1600                         * Peek past them to see if they belong to the
1601                         * body of the current function.
1602                         */
1603                        peek_bol = bol;
1604                        while (is_empty_line(peek_bol, peek_eol)) {
1605                                peek_bol = peek_eol + 1;
1606                                peek_eol = end_of_line(peek_bol, &peek_left);
1607                        }
1608
1609                        if (match_funcname(opt, gs, peek_bol, peek_eol))
1610                                show_function = 0;
1611                }
1612                if (show_function ||
1613                    (last_hit && lno <= last_hit + opt->post_context)) {
1614                        /* If the last hit is within the post context,
1615                         * we need to show this line.
1616                         */
1617                        show_line(opt, bol, eol, gs->name, lno, '-');
1618                }
1619
1620        next_line:
1621                bol = eol + 1;
1622                if (!left)
1623                        break;
1624                left--;
1625                lno++;
1626        }
1627
1628        if (collect_hits)
1629                return 0;
1630
1631        if (opt->status_only)
1632                return 0;
1633        if (opt->unmatch_name_only) {
1634                /* We did not see any hit, so we want to show this */
1635                show_name(opt, gs->name);
1636                return 1;
1637        }
1638
1639        xdiff_clear_find_func(&xecfg);
1640        opt->priv = NULL;
1641
1642        /* NEEDSWORK:
1643         * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1644         * which feels mostly useless but sometimes useful.  Maybe
1645         * make it another option?  For now suppress them.
1646         */
1647        if (opt->count && count) {
1648                char buf[32];
1649                if (opt->pathname) {
1650                        output_color(opt, gs->name, strlen(gs->name),
1651                                     opt->color_filename);
1652                        output_sep(opt, ':');
1653                }
1654                snprintf(buf, sizeof(buf), "%u\n", count);
1655                opt->output(opt, buf, strlen(buf));
1656                return 1;
1657        }
1658        return !!last_hit;
1659}
1660
1661static void clr_hit_marker(struct grep_expr *x)
1662{
1663        /* All-hit markers are meaningful only at the very top level
1664         * OR node.
1665         */
1666        while (1) {
1667                x->hit = 0;
1668                if (x->node != GREP_NODE_OR)
1669                        return;
1670                x->u.binary.left->hit = 0;
1671                x = x->u.binary.right;
1672        }
1673}
1674
1675static int chk_hit_marker(struct grep_expr *x)
1676{
1677        /* Top level nodes have hit markers.  See if they all are hits */
1678        while (1) {
1679                if (x->node != GREP_NODE_OR)
1680                        return x->hit;
1681                if (!x->u.binary.left->hit)
1682                        return 0;
1683                x = x->u.binary.right;
1684        }
1685}
1686
1687int grep_source(struct grep_opt *opt, struct grep_source *gs)
1688{
1689        /*
1690         * we do not have to do the two-pass grep when we do not check
1691         * buffer-wide "all-match".
1692         */
1693        if (!opt->all_match)
1694                return grep_source_1(opt, gs, 0);
1695
1696        /* Otherwise the toplevel "or" terms hit a bit differently.
1697         * We first clear hit markers from them.
1698         */
1699        clr_hit_marker(opt->pattern_expression);
1700        grep_source_1(opt, gs, 1);
1701
1702        if (!chk_hit_marker(opt->pattern_expression))
1703                return 0;
1704
1705        return grep_source_1(opt, gs, 0);
1706}
1707
1708int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1709{
1710        struct grep_source gs;
1711        int r;
1712
1713        grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1714        gs.buf = buf;
1715        gs.size = size;
1716
1717        r = grep_source(opt, &gs);
1718
1719        grep_source_clear(&gs);
1720        return r;
1721}
1722
1723void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1724                      const char *name, const char *path,
1725                      const void *identifier)
1726{
1727        gs->type = type;
1728        gs->name = xstrdup_or_null(name);
1729        gs->path = xstrdup_or_null(path);
1730        gs->buf = NULL;
1731        gs->size = 0;
1732        gs->driver = NULL;
1733
1734        switch (type) {
1735        case GREP_SOURCE_FILE:
1736                gs->identifier = xstrdup(identifier);
1737                break;
1738        case GREP_SOURCE_SHA1:
1739                gs->identifier = xmalloc(20);
1740                hashcpy(gs->identifier, identifier);
1741                break;
1742        case GREP_SOURCE_BUF:
1743                gs->identifier = NULL;
1744        }
1745}
1746
1747void grep_source_clear(struct grep_source *gs)
1748{
1749        free(gs->name);
1750        gs->name = NULL;
1751        free(gs->path);
1752        gs->path = NULL;
1753        free(gs->identifier);
1754        gs->identifier = NULL;
1755        grep_source_clear_data(gs);
1756}
1757
1758void grep_source_clear_data(struct grep_source *gs)
1759{
1760        switch (gs->type) {
1761        case GREP_SOURCE_FILE:
1762        case GREP_SOURCE_SHA1:
1763                free(gs->buf);
1764                gs->buf = NULL;
1765                gs->size = 0;
1766                break;
1767        case GREP_SOURCE_BUF:
1768                /* leave user-provided buf intact */
1769                break;
1770        }
1771}
1772
1773static int grep_source_load_sha1(struct grep_source *gs)
1774{
1775        enum object_type type;
1776
1777        grep_read_lock();
1778        gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1779        grep_read_unlock();
1780
1781        if (!gs->buf)
1782                return error(_("'%s': unable to read %s"),
1783                             gs->name,
1784                             sha1_to_hex(gs->identifier));
1785        return 0;
1786}
1787
1788static int grep_source_load_file(struct grep_source *gs)
1789{
1790        const char *filename = gs->identifier;
1791        struct stat st;
1792        char *data;
1793        size_t size;
1794        int i;
1795
1796        if (lstat(filename, &st) < 0) {
1797        err_ret:
1798                if (errno != ENOENT)
1799                        error_errno(_("failed to stat '%s'"), filename);
1800                return -1;
1801        }
1802        if (!S_ISREG(st.st_mode))
1803                return -1;
1804        size = xsize_t(st.st_size);
1805        i = open(filename, O_RDONLY);
1806        if (i < 0)
1807                goto err_ret;
1808        data = xmallocz(size);
1809        if (st.st_size != read_in_full(i, data, size)) {
1810                error_errno(_("'%s': short read"), filename);
1811                close(i);
1812                free(data);
1813                return -1;
1814        }
1815        close(i);
1816
1817        gs->buf = data;
1818        gs->size = size;
1819        return 0;
1820}
1821
1822static int grep_source_load(struct grep_source *gs)
1823{
1824        if (gs->buf)
1825                return 0;
1826
1827        switch (gs->type) {
1828        case GREP_SOURCE_FILE:
1829                return grep_source_load_file(gs);
1830        case GREP_SOURCE_SHA1:
1831                return grep_source_load_sha1(gs);
1832        case GREP_SOURCE_BUF:
1833                return gs->buf ? 0 : -1;
1834        }
1835        die("BUG: invalid grep_source type");
1836}
1837
1838void grep_source_load_driver(struct grep_source *gs)
1839{
1840        if (gs->driver)
1841                return;
1842
1843        grep_attr_lock();
1844        if (gs->path)
1845                gs->driver = userdiff_find_by_path(gs->path);
1846        if (!gs->driver)
1847                gs->driver = userdiff_find_by_name("default");
1848        grep_attr_unlock();
1849}
1850
1851static int grep_source_is_binary(struct grep_source *gs)
1852{
1853        grep_source_load_driver(gs);
1854        if (gs->driver->binary != -1)
1855                return gs->driver->binary;
1856
1857        if (!grep_source_load(gs))
1858                return buffer_is_binary(gs->buf, gs->size);
1859
1860        return 0;
1861}