Andrew's git / gitweb.git / grep.c
?
grep.con commit Merge branch 'maint' (8146f19)
   1#include "cache.h"
   2#include "grep.h"
   3#include "xdiff-interface.h"
   4
   5void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat)
   6{
   7        struct grep_pat *p = xcalloc(1, sizeof(*p));
   8        p->pattern = pat;
   9        p->origin = "header";
  10        p->no = 0;
  11        p->token = GREP_PATTERN_HEAD;
  12        p->field = field;
  13        *opt->pattern_tail = p;
  14        opt->pattern_tail = &p->next;
  15        p->next = NULL;
  16}
  17
  18void append_grep_pattern(struct grep_opt *opt, const char *pat,
  19                         const char *origin, int no, enum grep_pat_token t)
  20{
  21        struct grep_pat *p = xcalloc(1, sizeof(*p));
  22        p->pattern = pat;
  23        p->origin = origin;
  24        p->no = no;
  25        p->token = t;
  26        *opt->pattern_tail = p;
  27        opt->pattern_tail = &p->next;
  28        p->next = NULL;
  29}
  30
  31static int is_fixed(const char *s)
  32{
  33        while (*s && !is_regex_special(*s))
  34                s++;
  35        return !*s;
  36}
  37
  38static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
  39{
  40        int err;
  41
  42        p->word_regexp = opt->word_regexp;
  43
  44        if (opt->fixed || is_fixed(p->pattern))
  45                p->fixed = 1;
  46        if (opt->regflags & REG_ICASE)
  47                p->fixed = 0;
  48        if (p->fixed)
  49                return;
  50
  51        err = regcomp(&p->regexp, p->pattern, opt->regflags);
  52        if (err) {
  53                char errbuf[1024];
  54                char where[1024];
  55                if (p->no)
  56                        sprintf(where, "In '%s' at %d, ",
  57                                p->origin, p->no);
  58                else if (p->origin)
  59                        sprintf(where, "%s, ", p->origin);
  60                else
  61                        where[0] = 0;
  62                regerror(err, &p->regexp, errbuf, 1024);
  63                regfree(&p->regexp);
  64                die("%s'%s': %s", where, p->pattern, errbuf);
  65        }
  66}
  67
  68static struct grep_expr *compile_pattern_or(struct grep_pat **);
  69static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
  70{
  71        struct grep_pat *p;
  72        struct grep_expr *x;
  73
  74        p = *list;
  75        if (!p)
  76                return NULL;
  77        switch (p->token) {
  78        case GREP_PATTERN: /* atom */
  79        case GREP_PATTERN_HEAD:
  80        case GREP_PATTERN_BODY:
  81                x = xcalloc(1, sizeof (struct grep_expr));
  82                x->node = GREP_NODE_ATOM;
  83                x->u.atom = p;
  84                *list = p->next;
  85                return x;
  86        case GREP_OPEN_PAREN:
  87                *list = p->next;
  88                x = compile_pattern_or(list);
  89                if (!*list || (*list)->token != GREP_CLOSE_PAREN)
  90                        die("unmatched parenthesis");
  91                *list = (*list)->next;
  92                return x;
  93        default:
  94                return NULL;
  95        }
  96}
  97
  98static struct grep_expr *compile_pattern_not(struct grep_pat **list)
  99{
 100        struct grep_pat *p;
 101        struct grep_expr *x;
 102
 103        p = *list;
 104        if (!p)
 105                return NULL;
 106        switch (p->token) {
 107        case GREP_NOT:
 108                if (!p->next)
 109                        die("--not not followed by pattern expression");
 110                *list = p->next;
 111                x = xcalloc(1, sizeof (struct grep_expr));
 112                x->node = GREP_NODE_NOT;
 113                x->u.unary = compile_pattern_not(list);
 114                if (!x->u.unary)
 115                        die("--not followed by non pattern expression");
 116                return x;
 117        default:
 118                return compile_pattern_atom(list);
 119        }
 120}
 121
 122static struct grep_expr *compile_pattern_and(struct grep_pat **list)
 123{
 124        struct grep_pat *p;
 125        struct grep_expr *x, *y, *z;
 126
 127        x = compile_pattern_not(list);
 128        p = *list;
 129        if (p && p->token == GREP_AND) {
 130                if (!p->next)
 131                        die("--and not followed by pattern expression");
 132                *list = p->next;
 133                y = compile_pattern_and(list);
 134                if (!y)
 135                        die("--and not followed by pattern expression");
 136                z = xcalloc(1, sizeof (struct grep_expr));
 137                z->node = GREP_NODE_AND;
 138                z->u.binary.left = x;
 139                z->u.binary.right = y;
 140                return z;
 141        }
 142        return x;
 143}
 144
 145static struct grep_expr *compile_pattern_or(struct grep_pat **list)
 146{
 147        struct grep_pat *p;
 148        struct grep_expr *x, *y, *z;
 149
 150        x = compile_pattern_and(list);
 151        p = *list;
 152        if (x && p && p->token != GREP_CLOSE_PAREN) {
 153                y = compile_pattern_or(list);
 154                if (!y)
 155                        die("not a pattern expression %s", p->pattern);
 156                z = xcalloc(1, sizeof (struct grep_expr));
 157                z->node = GREP_NODE_OR;
 158                z->u.binary.left = x;
 159                z->u.binary.right = y;
 160                return z;
 161        }
 162        return x;
 163}
 164
 165static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
 166{
 167        return compile_pattern_or(list);
 168}
 169
 170void compile_grep_patterns(struct grep_opt *opt)
 171{
 172        struct grep_pat *p;
 173
 174        if (opt->all_match)
 175                opt->extended = 1;
 176
 177        for (p = opt->pattern_list; p; p = p->next) {
 178                switch (p->token) {
 179                case GREP_PATTERN: /* atom */
 180                case GREP_PATTERN_HEAD:
 181                case GREP_PATTERN_BODY:
 182                        compile_regexp(p, opt);
 183                        break;
 184                default:
 185                        opt->extended = 1;
 186                        break;
 187                }
 188        }
 189
 190        if (!opt->extended)
 191                return;
 192
 193        /* Then bundle them up in an expression.
 194         * A classic recursive descent parser would do.
 195         */
 196        p = opt->pattern_list;
 197        if (p)
 198                opt->pattern_expression = compile_pattern_expr(&p);
 199        if (p)
 200                die("incomplete pattern expression: %s", p->pattern);
 201}
 202
 203static void free_pattern_expr(struct grep_expr *x)
 204{
 205        switch (x->node) {
 206        case GREP_NODE_ATOM:
 207                break;
 208        case GREP_NODE_NOT:
 209                free_pattern_expr(x->u.unary);
 210                break;
 211        case GREP_NODE_AND:
 212        case GREP_NODE_OR:
 213                free_pattern_expr(x->u.binary.left);
 214                free_pattern_expr(x->u.binary.right);
 215                break;
 216        }
 217        free(x);
 218}
 219
 220void free_grep_patterns(struct grep_opt *opt)
 221{
 222        struct grep_pat *p, *n;
 223
 224        for (p = opt->pattern_list; p; p = n) {
 225                n = p->next;
 226                switch (p->token) {
 227                case GREP_PATTERN: /* atom */
 228                case GREP_PATTERN_HEAD:
 229                case GREP_PATTERN_BODY:
 230                        regfree(&p->regexp);
 231                        break;
 232                default:
 233                        break;
 234                }
 235                free(p);
 236        }
 237
 238        if (!opt->extended)
 239                return;
 240        free_pattern_expr(opt->pattern_expression);
 241}
 242
 243static char *end_of_line(char *cp, unsigned long *left)
 244{
 245        unsigned long l = *left;
 246        while (l && *cp != '\n') {
 247                l--;
 248                cp++;
 249        }
 250        *left = l;
 251        return cp;
 252}
 253
 254static int word_char(char ch)
 255{
 256        return isalnum(ch) || ch == '_';
 257}
 258
 259static void show_name(struct grep_opt *opt, const char *name)
 260{
 261        printf("%s%c", name, opt->null_following_name ? '\0' : '\n');
 262}
 263
 264static int fixmatch(const char *pattern, char *line, regmatch_t *match)
 265{
 266        char *hit = strstr(line, pattern);
 267        if (!hit) {
 268                match->rm_so = match->rm_eo = -1;
 269                return REG_NOMATCH;
 270        }
 271        else {
 272                match->rm_so = hit - line;
 273                match->rm_eo = match->rm_so + strlen(pattern);
 274                return 0;
 275        }
 276}
 277
 278static int strip_timestamp(char *bol, char **eol_p)
 279{
 280        char *eol = *eol_p;
 281        int ch;
 282
 283        while (bol < --eol) {
 284                if (*eol != '>')
 285                        continue;
 286                *eol_p = ++eol;
 287                ch = *eol;
 288                *eol = '\0';
 289                return ch;
 290        }
 291        return 0;
 292}
 293
 294static struct {
 295        const char *field;
 296        size_t len;
 297} header_field[] = {
 298        { "author ", 7 },
 299        { "committer ", 10 },
 300};
 301
 302static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
 303                             enum grep_context ctx,
 304                             regmatch_t *pmatch, int eflags)
 305{
 306        int hit = 0;
 307        int saved_ch = 0;
 308
 309        if ((p->token != GREP_PATTERN) &&
 310            ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
 311                return 0;
 312
 313        if (p->token == GREP_PATTERN_HEAD) {
 314                const char *field;
 315                size_t len;
 316                assert(p->field < ARRAY_SIZE(header_field));
 317                field = header_field[p->field].field;
 318                len = header_field[p->field].len;
 319                if (strncmp(bol, field, len))
 320                        return 0;
 321                bol += len;
 322                saved_ch = strip_timestamp(bol, &eol);
 323        }
 324
 325 again:
 326        if (p->fixed)
 327                hit = !fixmatch(p->pattern, bol, pmatch);
 328        else
 329                hit = !regexec(&p->regexp, bol, 1, pmatch, eflags);
 330
 331        if (hit && p->word_regexp) {
 332                if ((pmatch[0].rm_so < 0) ||
 333                    (eol - bol) <= pmatch[0].rm_so ||
 334                    (pmatch[0].rm_eo < 0) ||
 335                    (eol - bol) < pmatch[0].rm_eo)
 336                        die("regexp returned nonsense");
 337
 338                /* Match beginning must be either beginning of the
 339                 * line, or at word boundary (i.e. the last char must
 340                 * not be a word char).  Similarly, match end must be
 341                 * either end of the line, or at word boundary
 342                 * (i.e. the next char must not be a word char).
 343                 */
 344                if ( ((pmatch[0].rm_so == 0) ||
 345                      !word_char(bol[pmatch[0].rm_so-1])) &&
 346                     ((pmatch[0].rm_eo == (eol-bol)) ||
 347                      !word_char(bol[pmatch[0].rm_eo])) )
 348                        ;
 349                else
 350                        hit = 0;
 351
 352                if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
 353                        /* There could be more than one match on the
 354                         * line, and the first match might not be
 355                         * strict word match.  But later ones could be!
 356                         * Forward to the next possible start, i.e. the
 357                         * next position following a non-word char.
 358                         */
 359                        bol = pmatch[0].rm_so + bol + 1;
 360                        while (word_char(bol[-1]) && bol < eol)
 361                                bol++;
 362                        if (bol < eol)
 363                                goto again;
 364                }
 365        }
 366        if (p->token == GREP_PATTERN_HEAD && saved_ch)
 367                *eol = saved_ch;
 368        return hit;
 369}
 370
 371static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
 372                           enum grep_context ctx, int collect_hits)
 373{
 374        int h = 0;
 375        regmatch_t match;
 376
 377        if (!x)
 378                die("Not a valid grep expression");
 379        switch (x->node) {
 380        case GREP_NODE_ATOM:
 381                h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
 382                break;
 383        case GREP_NODE_NOT:
 384                h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
 385                break;
 386        case GREP_NODE_AND:
 387                if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
 388                        return 0;
 389                h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
 390                break;
 391        case GREP_NODE_OR:
 392                if (!collect_hits)
 393                        return (match_expr_eval(x->u.binary.left,
 394                                                bol, eol, ctx, 0) ||
 395                                match_expr_eval(x->u.binary.right,
 396                                                bol, eol, ctx, 0));
 397                h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
 398                x->u.binary.left->hit |= h;
 399                h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
 400                break;
 401        default:
 402                die("Unexpected node type (internal error) %d", x->node);
 403        }
 404        if (collect_hits)
 405                x->hit |= h;
 406        return h;
 407}
 408
 409static int match_expr(struct grep_opt *opt, char *bol, char *eol,
 410                      enum grep_context ctx, int collect_hits)
 411{
 412        struct grep_expr *x = opt->pattern_expression;
 413        return match_expr_eval(x, bol, eol, ctx, collect_hits);
 414}
 415
 416static int match_line(struct grep_opt *opt, char *bol, char *eol,
 417                      enum grep_context ctx, int collect_hits)
 418{
 419        struct grep_pat *p;
 420        regmatch_t match;
 421
 422        if (opt->extended)
 423                return match_expr(opt, bol, eol, ctx, collect_hits);
 424
 425        /* we do not call with collect_hits without being extended */
 426        for (p = opt->pattern_list; p; p = p->next) {
 427                if (match_one_pattern(p, bol, eol, ctx, &match, 0))
 428                        return 1;
 429        }
 430        return 0;
 431}
 432
 433static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
 434                              enum grep_context ctx,
 435                              regmatch_t *pmatch, int eflags)
 436{
 437        regmatch_t match;
 438
 439        if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
 440                return 0;
 441        if (match.rm_so < 0 || match.rm_eo < 0)
 442                return 0;
 443        if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
 444                if (match.rm_so > pmatch->rm_so)
 445                        return 1;
 446                if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
 447                        return 1;
 448        }
 449        pmatch->rm_so = match.rm_so;
 450        pmatch->rm_eo = match.rm_eo;
 451        return 1;
 452}
 453
 454static int next_match(struct grep_opt *opt, char *bol, char *eol,
 455                      enum grep_context ctx, regmatch_t *pmatch, int eflags)
 456{
 457        struct grep_pat *p;
 458        int hit = 0;
 459
 460        pmatch->rm_so = pmatch->rm_eo = -1;
 461        if (bol < eol) {
 462                for (p = opt->pattern_list; p; p = p->next) {
 463                        switch (p->token) {
 464                        case GREP_PATTERN: /* atom */
 465                        case GREP_PATTERN_HEAD:
 466                        case GREP_PATTERN_BODY:
 467                                hit |= match_next_pattern(p, bol, eol, ctx,
 468                                                          pmatch, eflags);
 469                                break;
 470                        default:
 471                                break;
 472                        }
 473                }
 474        }
 475        return hit;
 476}
 477
 478static void show_line(struct grep_opt *opt, char *bol, char *eol,
 479                      const char *name, unsigned lno, char sign)
 480{
 481        int rest = eol - bol;
 482
 483        if (opt->null_following_name)
 484                sign = '\0';
 485        if (opt->pathname)
 486                printf("%s%c", name, sign);
 487        if (opt->linenum)
 488                printf("%d%c", lno, sign);
 489        if (opt->color) {
 490                regmatch_t match;
 491                enum grep_context ctx = GREP_CONTEXT_BODY;
 492                int ch = *eol;
 493                int eflags = 0;
 494
 495                *eol = '\0';
 496                while (next_match(opt, bol, eol, ctx, &match, eflags)) {
 497                        printf("%.*s%s%.*s%s",
 498                               (int)match.rm_so, bol,
 499                               opt->color_match,
 500                               (int)(match.rm_eo - match.rm_so), bol + match.rm_so,
 501                               GIT_COLOR_RESET);
 502                        bol += match.rm_eo;
 503                        rest -= match.rm_eo;
 504                        eflags = REG_NOTBOL;
 505                }
 506                *eol = ch;
 507        }
 508        printf("%.*s\n", rest, bol);
 509}
 510
 511static int grep_buffer_1(struct grep_opt *opt, const char *name,
 512                         char *buf, unsigned long size, int collect_hits)
 513{
 514        char *bol = buf;
 515        unsigned long left = size;
 516        unsigned lno = 1;
 517        struct pre_context_line {
 518                char *bol;
 519                char *eol;
 520        } *prev = NULL, *pcl;
 521        unsigned last_hit = 0;
 522        unsigned last_shown = 0;
 523        int binary_match_only = 0;
 524        const char *hunk_mark = "";
 525        unsigned count = 0;
 526        enum grep_context ctx = GREP_CONTEXT_HEAD;
 527
 528        if (buffer_is_binary(buf, size)) {
 529                switch (opt->binary) {
 530                case GREP_BINARY_DEFAULT:
 531                        binary_match_only = 1;
 532                        break;
 533                case GREP_BINARY_NOMATCH:
 534                        return 0; /* Assume unmatch */
 535                        break;
 536                default:
 537                        break;
 538                }
 539        }
 540
 541        if (opt->pre_context)
 542                prev = xcalloc(opt->pre_context, sizeof(*prev));
 543        if (opt->pre_context || opt->post_context)
 544                hunk_mark = "--\n";
 545
 546        while (left) {
 547                char *eol, ch;
 548                int hit;
 549
 550                eol = end_of_line(bol, &left);
 551                ch = *eol;
 552                *eol = 0;
 553
 554                if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
 555                        ctx = GREP_CONTEXT_BODY;
 556
 557                hit = match_line(opt, bol, eol, ctx, collect_hits);
 558                *eol = ch;
 559
 560                if (collect_hits)
 561                        goto next_line;
 562
 563                /* "grep -v -e foo -e bla" should list lines
 564                 * that do not have either, so inversion should
 565                 * be done outside.
 566                 */
 567                if (opt->invert)
 568                        hit = !hit;
 569                if (opt->unmatch_name_only) {
 570                        if (hit)
 571                                return 0;
 572                        goto next_line;
 573                }
 574                if (hit) {
 575                        count++;
 576                        if (opt->status_only)
 577                                return 1;
 578                        if (binary_match_only) {
 579                                printf("Binary file %s matches\n", name);
 580                                return 1;
 581                        }
 582                        if (opt->name_only) {
 583                                show_name(opt, name);
 584                                return 1;
 585                        }
 586                        /* Hit at this line.  If we haven't shown the
 587                         * pre-context lines, we would need to show them.
 588                         * When asked to do "count", this still show
 589                         * the context which is nonsense, but the user
 590                         * deserves to get that ;-).
 591                         */
 592                        if (opt->pre_context) {
 593                                unsigned from;
 594                                if (opt->pre_context < lno)
 595                                        from = lno - opt->pre_context;
 596                                else
 597                                        from = 1;
 598                                if (from <= last_shown)
 599                                        from = last_shown + 1;
 600                                if (last_shown && from != last_shown + 1)
 601                                        fputs(hunk_mark, stdout);
 602                                while (from < lno) {
 603                                        pcl = &prev[lno-from-1];
 604                                        show_line(opt, pcl->bol, pcl->eol,
 605                                                  name, from, '-');
 606                                        from++;
 607                                }
 608                                last_shown = lno-1;
 609                        }
 610                        if (last_shown && lno != last_shown + 1)
 611                                fputs(hunk_mark, stdout);
 612                        if (!opt->count)
 613                                show_line(opt, bol, eol, name, lno, ':');
 614                        last_shown = last_hit = lno;
 615                }
 616                else if (last_hit &&
 617                         lno <= last_hit + opt->post_context) {
 618                        /* If the last hit is within the post context,
 619                         * we need to show this line.
 620                         */
 621                        if (last_shown && lno != last_shown + 1)
 622                                fputs(hunk_mark, stdout);
 623                        show_line(opt, bol, eol, name, lno, '-');
 624                        last_shown = lno;
 625                }
 626                if (opt->pre_context) {
 627                        memmove(prev+1, prev,
 628                                (opt->pre_context-1) * sizeof(*prev));
 629                        prev->bol = bol;
 630                        prev->eol = eol;
 631                }
 632
 633        next_line:
 634                bol = eol + 1;
 635                if (!left)
 636                        break;
 637                left--;
 638                lno++;
 639        }
 640
 641        free(prev);
 642        if (collect_hits)
 643                return 0;
 644
 645        if (opt->status_only)
 646                return 0;
 647        if (opt->unmatch_name_only) {
 648                /* We did not see any hit, so we want to show this */
 649                show_name(opt, name);
 650                return 1;
 651        }
 652
 653        /* NEEDSWORK:
 654         * The real "grep -c foo *.c" gives many "bar.c:0" lines,
 655         * which feels mostly useless but sometimes useful.  Maybe
 656         * make it another option?  For now suppress them.
 657         */
 658        if (opt->count && count)
 659                printf("%s%c%u\n", name,
 660                       opt->null_following_name ? '\0' : ':', count);
 661        return !!last_hit;
 662}
 663
 664static void clr_hit_marker(struct grep_expr *x)
 665{
 666        /* All-hit markers are meaningful only at the very top level
 667         * OR node.
 668         */
 669        while (1) {
 670                x->hit = 0;
 671                if (x->node != GREP_NODE_OR)
 672                        return;
 673                x->u.binary.left->hit = 0;
 674                x = x->u.binary.right;
 675        }
 676}
 677
 678static int chk_hit_marker(struct grep_expr *x)
 679{
 680        /* Top level nodes have hit markers.  See if they all are hits */
 681        while (1) {
 682                if (x->node != GREP_NODE_OR)
 683                        return x->hit;
 684                if (!x->u.binary.left->hit)
 685                        return 0;
 686                x = x->u.binary.right;
 687        }
 688}
 689
 690int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
 691{
 692        /*
 693         * we do not have to do the two-pass grep when we do not check
 694         * buffer-wide "all-match".
 695         */
 696        if (!opt->all_match)
 697                return grep_buffer_1(opt, name, buf, size, 0);
 698
 699        /* Otherwise the toplevel "or" terms hit a bit differently.
 700         * We first clear hit markers from them.
 701         */
 702        clr_hit_marker(opt->pattern_expression);
 703        grep_buffer_1(opt, name, buf, size, 1);
 704
 705        if (!chk_hit_marker(opt->pattern_expression))
 706                return 0;
 707
 708        return grep_buffer_1(opt, name, buf, size, 0);
 709}