4 #include "xdiff-interface.h"
6 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
7 const char *origin, int no,
9 enum grep_header_field field)
11 struct grep_pat *p = xcalloc(1, sizeof(*p));
12 p->pattern = xmemdupz(pat, patlen);
13 p->patternlen = patlen;
21 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
28 case GREP_PATTERN: /* atom */
29 case GREP_PATTERN_HEAD:
30 case GREP_PATTERN_BODY:
32 struct grep_pat *new_pat;
34 char *cp = p->pattern + p->patternlen, *nl = NULL;
35 while (++len <= p->patternlen) {
36 if (*(--cp) == '\n') {
43 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
44 p->no, p->token, p->field);
45 new_pat->next = p->next;
47 *tail = &new_pat->next;
58 void append_header_grep_pattern(struct grep_opt *opt,
59 enum grep_header_field field, const char *pat)
61 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
62 GREP_PATTERN_HEAD, field);
63 do_append_grep_pat(&opt->header_tail, p);
66 void append_grep_pattern(struct grep_opt *opt, const char *pat,
67 const char *origin, int no, enum grep_pat_token t)
69 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
72 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
73 const char *origin, int no, enum grep_pat_token t)
75 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
76 do_append_grep_pat(&opt->pattern_tail, p);
79 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
82 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
85 ret->pattern_list = NULL;
86 ret->pattern_tail = &ret->pattern_list;
88 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
90 if(pat->token == GREP_PATTERN_HEAD)
91 append_header_grep_pattern(ret, pat->field,
94 append_grep_pat(ret, pat->pattern, pat->patternlen,
95 pat->origin, pat->no, pat->token);
101 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
107 sprintf(where, "In '%s' at %d, ", p->origin, p->no);
109 sprintf(where, "%s, ", p->origin);
113 die("%s'%s': %s", where, p->pattern, error);
117 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
121 int options = PCRE_MULTILINE;
123 if (opt->ignore_case)
124 options |= PCRE_CASELESS;
126 p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
129 compile_regexp_failed(p, error);
131 p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
132 if (!p->pcre_extra_info && error)
136 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
137 regmatch_t *match, int eflags)
139 int ovector[30], ret, flags = 0;
141 if (eflags & REG_NOTBOL)
142 flags |= PCRE_NOTBOL;
144 ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
145 0, flags, ovector, ARRAY_SIZE(ovector));
146 if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
147 die("pcre_exec failed with error code %d", ret);
150 match->rm_so = ovector[0];
151 match->rm_eo = ovector[1];
157 static void free_pcre_regexp(struct grep_pat *p)
159 pcre_free(p->pcre_regexp);
160 pcre_free(p->pcre_extra_info);
162 #else /* !USE_LIBPCRE */
163 static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
165 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
168 static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
169 regmatch_t *match, int eflags)
174 static void free_pcre_regexp(struct grep_pat *p)
177 #endif /* !USE_LIBPCRE */
179 static int is_fixed(const char *s, size_t len)
183 /* regcomp cannot accept patterns with NULs so we
184 * consider any pattern containing a NUL fixed.
186 if (memchr(s, 0, len))
189 for (i = 0; i < len; i++) {
190 if (is_regex_special(s[i]))
197 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
201 p->word_regexp = opt->word_regexp;
202 p->ignore_case = opt->ignore_case;
204 if (opt->fixed || is_fixed(p->pattern, p->patternlen))
210 if (opt->regflags & REG_ICASE || p->ignore_case)
211 p->kws = kwsalloc(tolower_trans_tbl);
213 p->kws = kwsalloc(NULL);
214 kwsincr(p->kws, p->pattern, p->patternlen);
220 compile_pcre_regexp(p, opt);
224 err = regcomp(&p->regexp, p->pattern, opt->regflags);
227 regerror(err, &p->regexp, errbuf, 1024);
229 compile_regexp_failed(p, errbuf);
233 static struct grep_expr *compile_pattern_or(struct grep_pat **);
234 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
243 case GREP_PATTERN: /* atom */
244 case GREP_PATTERN_HEAD:
245 case GREP_PATTERN_BODY:
246 x = xcalloc(1, sizeof (struct grep_expr));
247 x->node = GREP_NODE_ATOM;
251 case GREP_OPEN_PAREN:
253 x = compile_pattern_or(list);
254 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
255 die("unmatched parenthesis");
256 *list = (*list)->next;
263 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
274 die("--not not followed by pattern expression");
276 x = xcalloc(1, sizeof (struct grep_expr));
277 x->node = GREP_NODE_NOT;
278 x->u.unary = compile_pattern_not(list);
280 die("--not followed by non pattern expression");
283 return compile_pattern_atom(list);
287 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
290 struct grep_expr *x, *y, *z;
292 x = compile_pattern_not(list);
294 if (p && p->token == GREP_AND) {
296 die("--and not followed by pattern expression");
298 y = compile_pattern_and(list);
300 die("--and not followed by pattern expression");
301 z = xcalloc(1, sizeof (struct grep_expr));
302 z->node = GREP_NODE_AND;
303 z->u.binary.left = x;
304 z->u.binary.right = y;
310 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
313 struct grep_expr *x, *y, *z;
315 x = compile_pattern_and(list);
317 if (x && p && p->token != GREP_CLOSE_PAREN) {
318 y = compile_pattern_or(list);
320 die("not a pattern expression %s", p->pattern);
321 z = xcalloc(1, sizeof (struct grep_expr));
322 z->node = GREP_NODE_OR;
323 z->u.binary.left = x;
324 z->u.binary.right = y;
330 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
332 return compile_pattern_or(list);
335 static void indent(int in)
341 static void dump_grep_pat(struct grep_pat *p)
344 case GREP_AND: fprintf(stderr, "*and*"); break;
345 case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
346 case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
347 case GREP_NOT: fprintf(stderr, "*not*"); break;
348 case GREP_OR: fprintf(stderr, "*or*"); break;
350 case GREP_PATTERN: fprintf(stderr, "pattern"); break;
351 case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
352 case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
357 case GREP_PATTERN_HEAD:
358 fprintf(stderr, "<head %d>", p->field); break;
359 case GREP_PATTERN_BODY:
360 fprintf(stderr, "<body>"); break;
364 case GREP_PATTERN_HEAD:
365 case GREP_PATTERN_BODY:
367 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
373 static void dump_grep_expression_1(struct grep_expr *x, int in)
378 fprintf(stderr, "true\n");
381 dump_grep_pat(x->u.atom);
384 fprintf(stderr, "(not\n");
385 dump_grep_expression_1(x->u.unary, in+1);
387 fprintf(stderr, ")\n");
390 fprintf(stderr, "(and\n");
391 dump_grep_expression_1(x->u.binary.left, in+1);
392 dump_grep_expression_1(x->u.binary.right, in+1);
394 fprintf(stderr, ")\n");
397 fprintf(stderr, "(or\n");
398 dump_grep_expression_1(x->u.binary.left, in+1);
399 dump_grep_expression_1(x->u.binary.right, in+1);
401 fprintf(stderr, ")\n");
406 void dump_grep_expression(struct grep_opt *opt)
408 struct grep_expr *x = opt->pattern_expression;
411 fprintf(stderr, "[all-match]\n");
412 dump_grep_expression_1(x, 0);
416 static struct grep_expr *grep_true_expr(void)
418 struct grep_expr *z = xcalloc(1, sizeof(*z));
419 z->node = GREP_NODE_TRUE;
423 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
425 struct grep_expr *z = xcalloc(1, sizeof(*z));
426 z->node = GREP_NODE_OR;
427 z->u.binary.left = left;
428 z->u.binary.right = right;
432 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
435 struct grep_expr *header_expr;
436 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
437 enum grep_header_field fld;
439 if (!opt->header_list)
442 for (p = opt->header_list; p; p = p->next) {
443 if (p->token != GREP_PATTERN_HEAD)
444 die("bug: a non-header pattern in grep header list.");
445 if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
446 die("bug: unknown header field %d", p->field);
447 compile_regexp(p, opt);
450 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
451 header_group[fld] = NULL;
453 for (p = opt->header_list; p; p = p->next) {
455 struct grep_pat *pp = p;
457 h = compile_pattern_atom(&pp);
458 if (!h || pp != p->next)
459 die("bug: malformed header expr");
460 if (!header_group[p->field]) {
461 header_group[p->field] = h;
464 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
469 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
470 if (!header_group[fld])
473 header_expr = grep_true_expr();
474 header_expr = grep_or_expr(header_group[fld], header_expr);
479 static void compile_grep_patterns_real(struct grep_opt *opt)
482 struct grep_expr *header_expr = prep_header_patterns(opt);
484 for (p = opt->pattern_list; p; p = p->next) {
486 case GREP_PATTERN: /* atom */
487 case GREP_PATTERN_HEAD:
488 case GREP_PATTERN_BODY:
489 compile_regexp(p, opt);
497 if (opt->all_match || header_expr)
499 else if (!opt->extended && !opt->debug)
502 p = opt->pattern_list;
504 opt->pattern_expression = compile_pattern_expr(&p);
506 die("incomplete pattern expression: %s", p->pattern);
511 if (!opt->pattern_expression)
512 opt->pattern_expression = header_expr;
514 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
519 void compile_grep_patterns(struct grep_opt *opt)
521 compile_grep_patterns_real(opt);
523 dump_grep_expression(opt);
526 static void free_pattern_expr(struct grep_expr *x)
533 free_pattern_expr(x->u.unary);
537 free_pattern_expr(x->u.binary.left);
538 free_pattern_expr(x->u.binary.right);
544 void free_grep_patterns(struct grep_opt *opt)
546 struct grep_pat *p, *n;
548 for (p = opt->pattern_list; p; p = n) {
551 case GREP_PATTERN: /* atom */
552 case GREP_PATTERN_HEAD:
553 case GREP_PATTERN_BODY:
556 else if (p->pcre_regexp)
570 free_pattern_expr(opt->pattern_expression);
573 static char *end_of_line(char *cp, unsigned long *left)
575 unsigned long l = *left;
576 while (l && *cp != '\n') {
584 static int word_char(char ch)
586 return isalnum(ch) || ch == '_';
589 static void output_color(struct grep_opt *opt, const void *data, size_t size,
592 if (want_color(opt->color) && color && color[0]) {
593 opt->output(opt, color, strlen(color));
594 opt->output(opt, data, size);
595 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
597 opt->output(opt, data, size);
600 static void output_sep(struct grep_opt *opt, char sign)
602 if (opt->null_following_name)
603 opt->output(opt, "\0", 1);
605 output_color(opt, &sign, 1, opt->color_sep);
608 static void show_name(struct grep_opt *opt, const char *name)
610 output_color(opt, name, strlen(name), opt->color_filename);
611 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
614 static int fixmatch(struct grep_pat *p, char *line, char *eol,
617 struct kwsmatch kwsm;
618 size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
620 match->rm_so = match->rm_eo = -1;
623 match->rm_so = offset;
624 match->rm_eo = match->rm_so + kwsm.size[0];
629 static int regmatch(const regex_t *preg, char *line, char *eol,
630 regmatch_t *match, int eflags)
634 match->rm_eo = eol - line;
635 eflags |= REG_STARTEND;
637 return regexec(preg, line, 1, match, eflags);
640 static int patmatch(struct grep_pat *p, char *line, char *eol,
641 regmatch_t *match, int eflags)
646 hit = !fixmatch(p, line, eol, match);
647 else if (p->pcre_regexp)
648 hit = !pcrematch(p, line, eol, match, eflags);
650 hit = !regmatch(&p->regexp, line, eol, match, eflags);
655 static int strip_timestamp(char *bol, char **eol_p)
660 while (bol < --eol) {
676 { "committer ", 10 },
679 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
680 enum grep_context ctx,
681 regmatch_t *pmatch, int eflags)
685 const char *start = bol;
687 if ((p->token != GREP_PATTERN) &&
688 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
691 if (p->token == GREP_PATTERN_HEAD) {
694 assert(p->field < ARRAY_SIZE(header_field));
695 field = header_field[p->field].field;
696 len = header_field[p->field].len;
697 if (strncmp(bol, field, len))
700 saved_ch = strip_timestamp(bol, &eol);
704 hit = patmatch(p, bol, eol, pmatch, eflags);
706 if (hit && p->word_regexp) {
707 if ((pmatch[0].rm_so < 0) ||
708 (eol - bol) < pmatch[0].rm_so ||
709 (pmatch[0].rm_eo < 0) ||
710 (eol - bol) < pmatch[0].rm_eo)
711 die("regexp returned nonsense");
713 /* Match beginning must be either beginning of the
714 * line, or at word boundary (i.e. the last char must
715 * not be a word char). Similarly, match end must be
716 * either end of the line, or at word boundary
717 * (i.e. the next char must not be a word char).
719 if ( ((pmatch[0].rm_so == 0) ||
720 !word_char(bol[pmatch[0].rm_so-1])) &&
721 ((pmatch[0].rm_eo == (eol-bol)) ||
722 !word_char(bol[pmatch[0].rm_eo])) )
727 /* Words consist of at least one character. */
728 if (pmatch->rm_so == pmatch->rm_eo)
731 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
732 /* There could be more than one match on the
733 * line, and the first match might not be
734 * strict word match. But later ones could be!
735 * Forward to the next possible start, i.e. the
736 * next position following a non-word char.
738 bol = pmatch[0].rm_so + bol + 1;
739 while (word_char(bol[-1]) && bol < eol)
741 eflags |= REG_NOTBOL;
746 if (p->token == GREP_PATTERN_HEAD && saved_ch)
749 pmatch[0].rm_so += bol - start;
750 pmatch[0].rm_eo += bol - start;
755 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
756 enum grep_context ctx, int collect_hits)
762 die("Not a valid grep expression");
768 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
771 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
774 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
776 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
780 return (match_expr_eval(x->u.binary.left,
782 match_expr_eval(x->u.binary.right,
784 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
785 x->u.binary.left->hit |= h;
786 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
789 die("Unexpected node type (internal error) %d", x->node);
796 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
797 enum grep_context ctx, int collect_hits)
799 struct grep_expr *x = opt->pattern_expression;
800 return match_expr_eval(x, bol, eol, ctx, collect_hits);
803 static int match_line(struct grep_opt *opt, char *bol, char *eol,
804 enum grep_context ctx, int collect_hits)
810 return match_expr(opt, bol, eol, ctx, collect_hits);
812 /* we do not call with collect_hits without being extended */
813 for (p = opt->pattern_list; p; p = p->next) {
814 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
820 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
821 enum grep_context ctx,
822 regmatch_t *pmatch, int eflags)
826 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
828 if (match.rm_so < 0 || match.rm_eo < 0)
830 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
831 if (match.rm_so > pmatch->rm_so)
833 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
836 pmatch->rm_so = match.rm_so;
837 pmatch->rm_eo = match.rm_eo;
841 static int next_match(struct grep_opt *opt, char *bol, char *eol,
842 enum grep_context ctx, regmatch_t *pmatch, int eflags)
847 pmatch->rm_so = pmatch->rm_eo = -1;
849 for (p = opt->pattern_list; p; p = p->next) {
851 case GREP_PATTERN: /* atom */
852 case GREP_PATTERN_HEAD:
853 case GREP_PATTERN_BODY:
854 hit |= match_next_pattern(p, bol, eol, ctx,
865 static void show_line(struct grep_opt *opt, char *bol, char *eol,
866 const char *name, unsigned lno, char sign)
868 int rest = eol - bol;
869 char *line_color = NULL;
871 if (opt->file_break && opt->last_shown == 0) {
872 if (opt->show_hunk_mark)
873 opt->output(opt, "\n", 1);
874 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
875 if (opt->last_shown == 0) {
876 if (opt->show_hunk_mark) {
877 output_color(opt, "--", 2, opt->color_sep);
878 opt->output(opt, "\n", 1);
880 } else if (lno > opt->last_shown + 1) {
881 output_color(opt, "--", 2, opt->color_sep);
882 opt->output(opt, "\n", 1);
885 if (opt->heading && opt->last_shown == 0) {
886 output_color(opt, name, strlen(name), opt->color_filename);
887 opt->output(opt, "\n", 1);
889 opt->last_shown = lno;
891 if (!opt->heading && opt->pathname) {
892 output_color(opt, name, strlen(name), opt->color_filename);
893 output_sep(opt, sign);
897 snprintf(buf, sizeof(buf), "%d", lno);
898 output_color(opt, buf, strlen(buf), opt->color_lineno);
899 output_sep(opt, sign);
903 enum grep_context ctx = GREP_CONTEXT_BODY;
908 line_color = opt->color_selected;
909 else if (sign == '-')
910 line_color = opt->color_context;
911 else if (sign == '=')
912 line_color = opt->color_function;
914 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
915 if (match.rm_so == match.rm_eo)
918 output_color(opt, bol, match.rm_so, line_color);
919 output_color(opt, bol + match.rm_so,
920 match.rm_eo - match.rm_so,
928 output_color(opt, bol, rest, line_color);
929 opt->output(opt, "\n", 1);
936 * This lock protects access to the gitattributes machinery, which is
939 pthread_mutex_t grep_attr_mutex;
941 static inline void grep_attr_lock(void)
944 pthread_mutex_lock(&grep_attr_mutex);
947 static inline void grep_attr_unlock(void)
950 pthread_mutex_unlock(&grep_attr_mutex);
954 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
956 pthread_mutex_t grep_read_mutex;
959 #define grep_attr_lock()
960 #define grep_attr_unlock()
963 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
965 xdemitconf_t *xecfg = opt->priv;
966 if (xecfg && !xecfg->find_func) {
967 grep_source_load_driver(gs);
968 if (gs->driver->funcname.pattern) {
969 const struct userdiff_funcname *pe = &gs->driver->funcname;
970 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
972 xecfg = opt->priv = NULL;
978 return xecfg->find_func(bol, eol - bol, buf, 1,
979 xecfg->find_func_priv) >= 0;
984 if (isalpha(*bol) || *bol == '_' || *bol == '$')
989 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
990 char *bol, unsigned lno)
992 while (bol > gs->buf) {
995 while (bol > gs->buf && bol[-1] != '\n')
999 if (lno <= opt->last_shown)
1002 if (match_funcname(opt, gs, bol, eol)) {
1003 show_line(opt, bol, eol, gs->name, lno, '=');
1009 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1010 char *bol, char *end, unsigned lno)
1012 unsigned cur = lno, from = 1, funcname_lno = 0;
1013 int funcname_needed = !!opt->funcname;
1015 if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1016 funcname_needed = 2;
1018 if (opt->pre_context < lno)
1019 from = lno - opt->pre_context;
1020 if (from <= opt->last_shown)
1021 from = opt->last_shown + 1;
1024 while (bol > gs->buf &&
1025 cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1028 while (bol > gs->buf && bol[-1] != '\n')
1031 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1033 funcname_needed = 0;
1037 /* We need to look even further back to find a function signature. */
1038 if (opt->funcname && funcname_needed)
1039 show_funcname_line(opt, gs, bol, cur);
1043 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1045 while (*eol != '\n')
1047 show_line(opt, bol, eol, gs->name, cur, sign);
1053 static int should_lookahead(struct grep_opt *opt)
1058 return 0; /* punt for too complex stuff */
1061 for (p = opt->pattern_list; p; p = p->next) {
1062 if (p->token != GREP_PATTERN)
1063 return 0; /* punt for "header only" and stuff */
1068 static int look_ahead(struct grep_opt *opt,
1069 unsigned long *left_p,
1073 unsigned lno = *lno_p;
1076 char *sp, *last_bol;
1077 regoff_t earliest = -1;
1079 for (p = opt->pattern_list; p; p = p->next) {
1083 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1084 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1086 if (earliest < 0 || m.rm_so < earliest)
1091 *bol_p = bol + *left_p;
1095 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1096 ; /* find the beginning of the line */
1099 for (sp = bol; sp < last_bol; sp++) {
1103 *left_p -= last_bol - bol;
1109 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1111 fwrite(buf, size, 1, stdout);
1114 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1119 unsigned last_hit = 0;
1120 int binary_match_only = 0;
1122 int try_lookahead = 0;
1123 int show_function = 0;
1124 enum grep_context ctx = GREP_CONTEXT_HEAD;
1128 opt->output = std_output;
1130 if (opt->pre_context || opt->post_context || opt->file_break ||
1132 /* Show hunk marks, except for the first file. */
1133 if (opt->last_shown)
1134 opt->show_hunk_mark = 1;
1136 * If we're using threads then we can't easily identify
1137 * the first file. Always put hunk marks in that case
1138 * and skip the very first one later in work_done().
1140 if (opt->output != std_output)
1141 opt->show_hunk_mark = 1;
1143 opt->last_shown = 0;
1145 switch (opt->binary) {
1146 case GREP_BINARY_DEFAULT:
1147 if (grep_source_is_binary(gs))
1148 binary_match_only = 1;
1150 case GREP_BINARY_NOMATCH:
1151 if (grep_source_is_binary(gs))
1152 return 0; /* Assume unmatch */
1154 case GREP_BINARY_TEXT:
1157 die("bug: unknown binary handling mode");
1160 memset(&xecfg, 0, sizeof(xecfg));
1163 try_lookahead = should_lookahead(opt);
1165 if (grep_source_load(gs) < 0)
1175 * look_ahead() skips quickly to the line that possibly
1176 * has the next hit; don't call it if we need to do
1177 * something more than just skipping the current line
1178 * in response to an unmatch for the current line. E.g.
1179 * inside a post-context window, we will show the current
1180 * line as a context around the previous hit when it
1185 && (show_function ||
1186 lno <= last_hit + opt->post_context))
1187 && look_ahead(opt, &left, &lno, &bol))
1189 eol = end_of_line(bol, &left);
1193 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1194 ctx = GREP_CONTEXT_BODY;
1196 hit = match_line(opt, bol, eol, ctx, collect_hits);
1202 /* "grep -v -e foo -e bla" should list lines
1203 * that do not have either, so inversion should
1208 if (opt->unmatch_name_only) {
1215 if (opt->status_only)
1217 if (opt->name_only) {
1218 show_name(opt, gs->name);
1223 if (binary_match_only) {
1224 opt->output(opt, "Binary file ", 12);
1225 output_color(opt, gs->name, strlen(gs->name),
1226 opt->color_filename);
1227 opt->output(opt, " matches\n", 9);
1230 /* Hit at this line. If we haven't shown the
1231 * pre-context lines, we would need to show them.
1233 if (opt->pre_context || opt->funcbody)
1234 show_pre_context(opt, gs, bol, eol, lno);
1235 else if (opt->funcname)
1236 show_funcname_line(opt, gs, bol, lno);
1237 show_line(opt, bol, eol, gs->name, lno, ':');
1243 if (show_function && match_funcname(opt, gs, bol, eol))
1245 if (show_function ||
1246 (last_hit && lno <= last_hit + opt->post_context)) {
1247 /* If the last hit is within the post context,
1248 * we need to show this line.
1250 show_line(opt, bol, eol, gs->name, lno, '-');
1264 if (opt->status_only)
1266 if (opt->unmatch_name_only) {
1267 /* We did not see any hit, so we want to show this */
1268 show_name(opt, gs->name);
1272 xdiff_clear_find_func(&xecfg);
1276 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1277 * which feels mostly useless but sometimes useful. Maybe
1278 * make it another option? For now suppress them.
1280 if (opt->count && count) {
1282 output_color(opt, gs->name, strlen(gs->name), opt->color_filename);
1283 output_sep(opt, ':');
1284 snprintf(buf, sizeof(buf), "%u\n", count);
1285 opt->output(opt, buf, strlen(buf));
1291 static void clr_hit_marker(struct grep_expr *x)
1293 /* All-hit markers are meaningful only at the very top level
1298 if (x->node != GREP_NODE_OR)
1300 x->u.binary.left->hit = 0;
1301 x = x->u.binary.right;
1305 static int chk_hit_marker(struct grep_expr *x)
1307 /* Top level nodes have hit markers. See if they all are hits */
1309 if (x->node != GREP_NODE_OR)
1311 if (!x->u.binary.left->hit)
1313 x = x->u.binary.right;
1317 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1320 * we do not have to do the two-pass grep when we do not check
1321 * buffer-wide "all-match".
1323 if (!opt->all_match)
1324 return grep_source_1(opt, gs, 0);
1326 /* Otherwise the toplevel "or" terms hit a bit differently.
1327 * We first clear hit markers from them.
1329 clr_hit_marker(opt->pattern_expression);
1330 grep_source_1(opt, gs, 1);
1332 if (!chk_hit_marker(opt->pattern_expression))
1335 return grep_source_1(opt, gs, 0);
1338 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1340 struct grep_source gs;
1343 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL);
1347 r = grep_source(opt, &gs);
1349 grep_source_clear(&gs);
1353 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1354 const char *name, const void *identifier)
1357 gs->name = name ? xstrdup(name) : NULL;
1363 case GREP_SOURCE_FILE:
1364 gs->identifier = xstrdup(identifier);
1366 case GREP_SOURCE_SHA1:
1367 gs->identifier = xmalloc(20);
1368 memcpy(gs->identifier, identifier, 20);
1370 case GREP_SOURCE_BUF:
1371 gs->identifier = NULL;
1375 void grep_source_clear(struct grep_source *gs)
1379 free(gs->identifier);
1380 gs->identifier = NULL;
1381 grep_source_clear_data(gs);
1384 void grep_source_clear_data(struct grep_source *gs)
1387 case GREP_SOURCE_FILE:
1388 case GREP_SOURCE_SHA1:
1393 case GREP_SOURCE_BUF:
1394 /* leave user-provided buf intact */
1399 static int grep_source_load_sha1(struct grep_source *gs)
1401 enum object_type type;
1404 gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1408 return error(_("'%s': unable to read %s"),
1410 sha1_to_hex(gs->identifier));
1414 static int grep_source_load_file(struct grep_source *gs)
1416 const char *filename = gs->identifier;
1422 if (lstat(filename, &st) < 0) {
1424 if (errno != ENOENT)
1425 error(_("'%s': %s"), filename, strerror(errno));
1428 if (!S_ISREG(st.st_mode))
1430 size = xsize_t(st.st_size);
1431 i = open(filename, O_RDONLY);
1434 data = xmalloc(size + 1);
1435 if (st.st_size != read_in_full(i, data, size)) {
1436 error(_("'%s': short read %s"), filename, strerror(errno));
1449 int grep_source_load(struct grep_source *gs)
1455 case GREP_SOURCE_FILE:
1456 return grep_source_load_file(gs);
1457 case GREP_SOURCE_SHA1:
1458 return grep_source_load_sha1(gs);
1459 case GREP_SOURCE_BUF:
1460 return gs->buf ? 0 : -1;
1462 die("BUG: invalid grep_source type");
1465 void grep_source_load_driver(struct grep_source *gs)
1471 gs->driver = userdiff_find_by_path(gs->name);
1473 gs->driver = userdiff_find_by_name("default");
1477 int grep_source_is_binary(struct grep_source *gs)
1479 grep_source_load_driver(gs);
1480 if (gs->driver->binary != -1)
1481 return gs->driver->binary;
1483 if (!grep_source_load(gs))
1484 return buffer_is_binary(gs->buf, gs->size);