4 #include "xdiff-interface.h"
10 static int grep_source_load(struct grep_source *gs);
11 static int grep_source_is_binary(struct grep_source *gs);
13 static struct grep_opt grep_defaults;
15 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
17 fwrite(buf, size, 1, stdout);
21 * Initialize the grep_defaults template with hardcoded defaults.
22 * We could let the compiler do this, but without C99 initializers
23 * the code gets unwieldy and unreadable, so...
25 void init_grep_defaults(void)
27 struct grep_opt *opt = &grep_defaults;
34 memset(opt, 0, sizeof(*opt));
37 opt->regflags = REG_NEWLINE;
39 opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
40 opt->extended_regexp_option = 0;
41 color_set(opt->color_context, "");
42 color_set(opt->color_filename, "");
43 color_set(opt->color_function, "");
44 color_set(opt->color_lineno, "");
45 color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
46 color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
47 color_set(opt->color_selected, "");
48 color_set(opt->color_sep, GIT_COLOR_CYAN);
50 opt->output = std_output;
53 static int parse_pattern_type_arg(const char *opt, const char *arg)
55 if (!strcmp(arg, "default"))
56 return GREP_PATTERN_TYPE_UNSPECIFIED;
57 else if (!strcmp(arg, "basic"))
58 return GREP_PATTERN_TYPE_BRE;
59 else if (!strcmp(arg, "extended"))
60 return GREP_PATTERN_TYPE_ERE;
61 else if (!strcmp(arg, "fixed"))
62 return GREP_PATTERN_TYPE_FIXED;
63 else if (!strcmp(arg, "perl"))
64 return GREP_PATTERN_TYPE_PCRE;
65 die("bad %s argument: %s", opt, arg);
69 * Read the configuration file once and store it in
70 * the grep_defaults template.
72 int grep_config(const char *var, const char *value, void *cb)
74 struct grep_opt *opt = &grep_defaults;
77 if (userdiff_config(var, value) < 0)
80 if (!strcmp(var, "grep.extendedregexp")) {
81 if (git_config_bool(var, value))
82 opt->extended_regexp_option = 1;
84 opt->extended_regexp_option = 0;
88 if (!strcmp(var, "grep.patterntype")) {
89 opt->pattern_type_option = parse_pattern_type_arg(var, value);
93 if (!strcmp(var, "grep.linenumber")) {
94 opt->linenum = git_config_bool(var, value);
98 if (!strcmp(var, "grep.fullname")) {
99 opt->relative = !git_config_bool(var, value);
103 if (!strcmp(var, "color.grep"))
104 opt->color = git_config_colorbool(var, value);
105 else if (!strcmp(var, "color.grep.context"))
106 color = opt->color_context;
107 else if (!strcmp(var, "color.grep.filename"))
108 color = opt->color_filename;
109 else if (!strcmp(var, "color.grep.function"))
110 color = opt->color_function;
111 else if (!strcmp(var, "color.grep.linenumber"))
112 color = opt->color_lineno;
113 else if (!strcmp(var, "color.grep.matchcontext"))
114 color = opt->color_match_context;
115 else if (!strcmp(var, "color.grep.matchselected"))
116 color = opt->color_match_selected;
117 else if (!strcmp(var, "color.grep.selected"))
118 color = opt->color_selected;
119 else if (!strcmp(var, "color.grep.separator"))
120 color = opt->color_sep;
121 else if (!strcmp(var, "color.grep.match")) {
124 return config_error_nonbool(var);
125 rc |= color_parse(value, opt->color_match_context);
126 rc |= color_parse(value, opt->color_match_selected);
132 return config_error_nonbool(var);
133 return color_parse(value, color);
139 * Initialize one instance of grep_opt and copy the
140 * default values from the template we read the configuration
141 * information in an earlier call to git_config(grep_config).
143 void grep_init(struct grep_opt *opt, const char *prefix)
145 struct grep_opt *def = &grep_defaults;
147 memset(opt, 0, sizeof(*opt));
148 opt->prefix = prefix;
149 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
150 opt->pattern_tail = &opt->pattern_list;
151 opt->header_tail = &opt->header_list;
153 opt->color = def->color;
154 opt->extended_regexp_option = def->extended_regexp_option;
155 opt->pattern_type_option = def->pattern_type_option;
156 opt->linenum = def->linenum;
157 opt->max_depth = def->max_depth;
158 opt->pathname = def->pathname;
159 opt->regflags = def->regflags;
160 opt->relative = def->relative;
161 opt->output = def->output;
163 color_set(opt->color_context, def->color_context);
164 color_set(opt->color_filename, def->color_filename);
165 color_set(opt->color_function, def->color_function);
166 color_set(opt->color_lineno, def->color_lineno);
167 color_set(opt->color_match_context, def->color_match_context);
168 color_set(opt->color_match_selected, def->color_match_selected);
169 color_set(opt->color_selected, def->color_selected);
170 color_set(opt->color_sep, def->color_sep);
173 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
175 switch (pattern_type) {
176 case GREP_PATTERN_TYPE_UNSPECIFIED:
179 case GREP_PATTERN_TYPE_BRE:
185 case GREP_PATTERN_TYPE_ERE:
189 opt->regflags |= REG_EXTENDED;
192 case GREP_PATTERN_TYPE_FIXED:
198 case GREP_PATTERN_TYPE_PCRE:
205 * It's important that pcre1 always be assigned to
206 * even when there's no USE_LIBPCRE* defined. We still
207 * call the PCRE stub function, it just dies with
208 * "cannot use Perl-compatible regexes[...]".
217 void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
219 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
220 grep_set_pattern_type_option(pattern_type, opt);
221 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
222 grep_set_pattern_type_option(opt->pattern_type_option, opt);
223 else if (opt->extended_regexp_option)
224 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
227 static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
228 const char *origin, int no,
229 enum grep_pat_token t,
230 enum grep_header_field field)
232 struct grep_pat *p = xcalloc(1, sizeof(*p));
233 p->pattern = xmemdupz(pat, patlen);
234 p->patternlen = patlen;
242 static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
249 case GREP_PATTERN: /* atom */
250 case GREP_PATTERN_HEAD:
251 case GREP_PATTERN_BODY:
253 struct grep_pat *new_pat;
255 char *cp = p->pattern + p->patternlen, *nl = NULL;
256 while (++len <= p->patternlen) {
257 if (*(--cp) == '\n') {
264 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
265 p->no, p->token, p->field);
266 new_pat->next = p->next;
268 *tail = &new_pat->next;
271 p->patternlen -= len;
279 void append_header_grep_pattern(struct grep_opt *opt,
280 enum grep_header_field field, const char *pat)
282 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
283 GREP_PATTERN_HEAD, field);
284 if (field == GREP_HEADER_REFLOG)
285 opt->use_reflog_filter = 1;
286 do_append_grep_pat(&opt->header_tail, p);
289 void append_grep_pattern(struct grep_opt *opt, const char *pat,
290 const char *origin, int no, enum grep_pat_token t)
292 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
295 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
296 const char *origin, int no, enum grep_pat_token t)
298 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
299 do_append_grep_pat(&opt->pattern_tail, p);
302 struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
304 struct grep_pat *pat;
305 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
308 ret->pattern_list = NULL;
309 ret->pattern_tail = &ret->pattern_list;
311 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
313 if(pat->token == GREP_PATTERN_HEAD)
314 append_header_grep_pattern(ret, pat->field,
317 append_grep_pat(ret, pat->pattern, pat->patternlen,
318 pat->origin, pat->no, pat->token);
324 static NORETURN void compile_regexp_failed(const struct grep_pat *p,
330 xsnprintf(where, sizeof(where), "In '%s' at %d, ", p->origin, p->no);
332 xsnprintf(where, sizeof(where), "%s, ", p->origin);
336 die("%s'%s': %s", where, p->pattern, error);
339 static int is_fixed(const char *s, size_t len)
343 for (i = 0; i < len; i++) {
344 if (is_regex_special(s[i]))
351 static int has_null(const char *s, size_t len)
354 * regcomp cannot accept patterns with NULs so when using it
355 * we consider any pattern containing a NUL fixed.
357 if (memchr(s, 0, len))
364 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
368 int options = PCRE_MULTILINE;
370 if (opt->ignore_case) {
371 if (has_non_ascii(p->pattern))
372 p->pcre1_tables = pcre_maketables();
373 options |= PCRE_CASELESS;
375 if (is_utf8_locale() && has_non_ascii(p->pattern))
376 options |= PCRE_UTF8;
378 p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
380 if (!p->pcre1_regexp)
381 compile_regexp_failed(p, error);
383 p->pcre1_extra_info = pcre_study(p->pcre1_regexp, PCRE_STUDY_JIT_COMPILE, &error);
384 if (!p->pcre1_extra_info && error)
387 #ifdef GIT_PCRE1_USE_JIT
388 pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
389 if (p->pcre1_jit_on == 1) {
390 p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
391 if (!p->pcre1_jit_stack)
392 die("Couldn't allocate PCRE JIT stack");
393 pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
394 } else if (p->pcre1_jit_on != 0) {
395 die("BUG: The pcre1_jit_on variable should be 0 or 1, not %d",
401 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
402 regmatch_t *match, int eflags)
404 int ovector[30], ret, flags = 0;
406 if (eflags & REG_NOTBOL)
407 flags |= PCRE_NOTBOL;
409 #ifdef GIT_PCRE1_USE_JIT
410 if (p->pcre1_jit_on) {
411 ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
412 eol - line, 0, flags, ovector,
413 ARRAY_SIZE(ovector), p->pcre1_jit_stack);
417 ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
418 eol - line, 0, flags, ovector,
419 ARRAY_SIZE(ovector));
422 if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
423 die("pcre_exec failed with error code %d", ret);
426 match->rm_so = ovector[0];
427 match->rm_eo = ovector[1];
433 static void free_pcre1_regexp(struct grep_pat *p)
435 pcre_free(p->pcre1_regexp);
436 #ifdef GIT_PCRE1_USE_JIT
437 if (p->pcre1_jit_on) {
438 pcre_free_study(p->pcre1_extra_info);
439 pcre_jit_stack_free(p->pcre1_jit_stack);
443 pcre_free(p->pcre1_extra_info);
445 pcre_free((void *)p->pcre1_tables);
447 #else /* !USE_LIBPCRE1 */
448 static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
450 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
453 static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
454 regmatch_t *match, int eflags)
459 static void free_pcre1_regexp(struct grep_pat *p)
462 #endif /* !USE_LIBPCRE1 */
465 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
468 PCRE2_UCHAR errbuf[256];
469 PCRE2_SIZE erroffset;
470 int options = PCRE2_MULTILINE;
471 const uint8_t *character_tables = NULL;
476 p->pcre2_compile_context = NULL;
478 if (opt->ignore_case) {
479 if (has_non_ascii(p->pattern)) {
480 character_tables = pcre2_maketables(NULL);
481 p->pcre2_compile_context = pcre2_compile_context_create(NULL);
482 pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
484 options |= PCRE2_CASELESS;
486 if (is_utf8_locale() && has_non_ascii(p->pattern))
487 options |= PCRE2_UTF;
489 p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
490 p->patternlen, options, &error, &erroffset,
491 p->pcre2_compile_context);
493 if (p->pcre2_pattern) {
494 p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
495 if (!p->pcre2_match_data)
496 die("Couldn't allocate PCRE2 match data");
498 pcre2_get_error_message(error, errbuf, sizeof(errbuf));
499 compile_regexp_failed(p, (const char *)&errbuf);
502 pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
503 if (p->pcre2_jit_on == 1) {
504 jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
506 die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
507 p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
508 if (!p->pcre2_jit_stack)
509 die("Couldn't allocate PCRE2 JIT stack");
510 p->pcre2_match_context = pcre2_match_context_create(NULL);
511 if (!p->pcre2_jit_stack)
512 die("Couldn't allocate PCRE2 match context");
513 pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
514 } else if (p->pcre2_jit_on != 0) {
515 die("BUG: The pcre2_jit_on variable should be 0 or 1, not %d",
520 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
521 regmatch_t *match, int eflags)
525 PCRE2_UCHAR errbuf[256];
527 if (eflags & REG_NOTBOL)
528 flags |= PCRE2_NOTBOL;
531 ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
532 eol - line, 0, flags, p->pcre2_match_data,
535 ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
536 eol - line, 0, flags, p->pcre2_match_data,
539 if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
540 pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
541 die("%s failed with error code %d: %s",
542 (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
546 ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
548 match->rm_so = (int)ovector[0];
549 match->rm_eo = (int)ovector[1];
555 static void free_pcre2_pattern(struct grep_pat *p)
557 pcre2_compile_context_free(p->pcre2_compile_context);
558 pcre2_code_free(p->pcre2_pattern);
559 pcre2_match_data_free(p->pcre2_match_data);
560 pcre2_jit_stack_free(p->pcre2_jit_stack);
561 pcre2_match_context_free(p->pcre2_match_context);
563 #else /* !USE_LIBPCRE2 */
564 static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
567 * Unreachable until USE_LIBPCRE2 becomes synonymous with
568 * USE_LIBPCRE. See the sibling comment in
569 * grep_set_pattern_type_option().
571 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
574 static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
575 regmatch_t *match, int eflags)
580 static void free_pcre2_pattern(struct grep_pat *p)
583 #endif /* !USE_LIBPCRE2 */
585 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
587 struct strbuf sb = STRBUF_INIT;
589 int regflags = opt->regflags;
591 basic_regex_quote_buf(&sb, p->pattern);
592 if (opt->ignore_case)
593 regflags |= REG_ICASE;
594 err = regcomp(&p->regexp, sb.buf, regflags);
596 fprintf(stderr, "fixed %s\n", sb.buf);
600 regerror(err, &p->regexp, errbuf, sizeof(errbuf));
602 compile_regexp_failed(p, errbuf);
606 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
608 int icase, ascii_only;
611 p->word_regexp = opt->word_regexp;
612 p->ignore_case = opt->ignore_case;
613 icase = opt->regflags & REG_ICASE || p->ignore_case;
614 ascii_only = !has_non_ascii(p->pattern);
617 * Even when -F (fixed) asks us to do a non-regexp search, we
618 * may not be able to correctly case-fold when -i
619 * (ignore-case) is asked (in which case, we'll synthesize a
620 * regexp to match the pattern that matches regexp special
621 * characters literally, while ignoring case differences). On
622 * the other hand, even without -F, if the pattern does not
623 * have any regexp special characters and there is no need for
624 * case-folding search, we can internally turn it into a
625 * simple string match using kws. p->fixed tells us if we
629 has_null(p->pattern, p->patternlen) ||
630 is_fixed(p->pattern, p->patternlen))
631 p->fixed = !icase || ascii_only;
636 p->kws = kwsalloc(icase ? tolower_trans_tbl : NULL);
637 kwsincr(p->kws, p->pattern, p->patternlen);
640 } else if (opt->fixed) {
642 * We come here when the pattern has the non-ascii
643 * characters we cannot case-fold, and asked to
646 compile_fixed_regexp(p, opt);
651 compile_pcre2_pattern(p, opt);
656 compile_pcre1_regexp(p, opt);
660 err = regcomp(&p->regexp, p->pattern, opt->regflags);
663 regerror(err, &p->regexp, errbuf, 1024);
665 compile_regexp_failed(p, errbuf);
669 static struct grep_expr *compile_pattern_or(struct grep_pat **);
670 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
679 case GREP_PATTERN: /* atom */
680 case GREP_PATTERN_HEAD:
681 case GREP_PATTERN_BODY:
682 x = xcalloc(1, sizeof (struct grep_expr));
683 x->node = GREP_NODE_ATOM;
687 case GREP_OPEN_PAREN:
689 x = compile_pattern_or(list);
690 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
691 die("unmatched parenthesis");
692 *list = (*list)->next;
699 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
710 die("--not not followed by pattern expression");
712 x = xcalloc(1, sizeof (struct grep_expr));
713 x->node = GREP_NODE_NOT;
714 x->u.unary = compile_pattern_not(list);
716 die("--not followed by non pattern expression");
719 return compile_pattern_atom(list);
723 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
726 struct grep_expr *x, *y, *z;
728 x = compile_pattern_not(list);
730 if (p && p->token == GREP_AND) {
732 die("--and not followed by pattern expression");
734 y = compile_pattern_and(list);
736 die("--and not followed by pattern expression");
737 z = xcalloc(1, sizeof (struct grep_expr));
738 z->node = GREP_NODE_AND;
739 z->u.binary.left = x;
740 z->u.binary.right = y;
746 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
749 struct grep_expr *x, *y, *z;
751 x = compile_pattern_and(list);
753 if (x && p && p->token != GREP_CLOSE_PAREN) {
754 y = compile_pattern_or(list);
756 die("not a pattern expression %s", p->pattern);
757 z = xcalloc(1, sizeof (struct grep_expr));
758 z->node = GREP_NODE_OR;
759 z->u.binary.left = x;
760 z->u.binary.right = y;
766 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
768 return compile_pattern_or(list);
771 static void indent(int in)
777 static void dump_grep_pat(struct grep_pat *p)
780 case GREP_AND: fprintf(stderr, "*and*"); break;
781 case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
782 case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
783 case GREP_NOT: fprintf(stderr, "*not*"); break;
784 case GREP_OR: fprintf(stderr, "*or*"); break;
786 case GREP_PATTERN: fprintf(stderr, "pattern"); break;
787 case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
788 case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
793 case GREP_PATTERN_HEAD:
794 fprintf(stderr, "<head %d>", p->field); break;
795 case GREP_PATTERN_BODY:
796 fprintf(stderr, "<body>"); break;
800 case GREP_PATTERN_HEAD:
801 case GREP_PATTERN_BODY:
803 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
809 static void dump_grep_expression_1(struct grep_expr *x, int in)
814 fprintf(stderr, "true\n");
817 dump_grep_pat(x->u.atom);
820 fprintf(stderr, "(not\n");
821 dump_grep_expression_1(x->u.unary, in+1);
823 fprintf(stderr, ")\n");
826 fprintf(stderr, "(and\n");
827 dump_grep_expression_1(x->u.binary.left, in+1);
828 dump_grep_expression_1(x->u.binary.right, in+1);
830 fprintf(stderr, ")\n");
833 fprintf(stderr, "(or\n");
834 dump_grep_expression_1(x->u.binary.left, in+1);
835 dump_grep_expression_1(x->u.binary.right, in+1);
837 fprintf(stderr, ")\n");
842 static void dump_grep_expression(struct grep_opt *opt)
844 struct grep_expr *x = opt->pattern_expression;
847 fprintf(stderr, "[all-match]\n");
848 dump_grep_expression_1(x, 0);
852 static struct grep_expr *grep_true_expr(void)
854 struct grep_expr *z = xcalloc(1, sizeof(*z));
855 z->node = GREP_NODE_TRUE;
859 static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
861 struct grep_expr *z = xcalloc(1, sizeof(*z));
862 z->node = GREP_NODE_OR;
863 z->u.binary.left = left;
864 z->u.binary.right = right;
868 static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
871 struct grep_expr *header_expr;
872 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
873 enum grep_header_field fld;
875 if (!opt->header_list)
878 for (p = opt->header_list; p; p = p->next) {
879 if (p->token != GREP_PATTERN_HEAD)
880 die("BUG: a non-header pattern in grep header list.");
881 if (p->field < GREP_HEADER_FIELD_MIN ||
882 GREP_HEADER_FIELD_MAX <= p->field)
883 die("BUG: unknown header field %d", p->field);
884 compile_regexp(p, opt);
887 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
888 header_group[fld] = NULL;
890 for (p = opt->header_list; p; p = p->next) {
892 struct grep_pat *pp = p;
894 h = compile_pattern_atom(&pp);
895 if (!h || pp != p->next)
896 die("BUG: malformed header expr");
897 if (!header_group[p->field]) {
898 header_group[p->field] = h;
901 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
906 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
907 if (!header_group[fld])
910 header_expr = grep_true_expr();
911 header_expr = grep_or_expr(header_group[fld], header_expr);
916 static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
918 struct grep_expr *z = x;
921 assert(x->node == GREP_NODE_OR);
922 if (x->u.binary.right &&
923 x->u.binary.right->node == GREP_NODE_TRUE) {
924 x->u.binary.right = y;
927 x = x->u.binary.right;
932 static void compile_grep_patterns_real(struct grep_opt *opt)
935 struct grep_expr *header_expr = prep_header_patterns(opt);
937 for (p = opt->pattern_list; p; p = p->next) {
939 case GREP_PATTERN: /* atom */
940 case GREP_PATTERN_HEAD:
941 case GREP_PATTERN_BODY:
942 compile_regexp(p, opt);
950 if (opt->all_match || header_expr)
952 else if (!opt->extended && !opt->debug)
955 p = opt->pattern_list;
957 opt->pattern_expression = compile_pattern_expr(&p);
959 die("incomplete pattern expression: %s", p->pattern);
964 if (!opt->pattern_expression)
965 opt->pattern_expression = header_expr;
966 else if (opt->all_match)
967 opt->pattern_expression = grep_splice_or(header_expr,
968 opt->pattern_expression);
970 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
975 void compile_grep_patterns(struct grep_opt *opt)
977 compile_grep_patterns_real(opt);
979 dump_grep_expression(opt);
982 static void free_pattern_expr(struct grep_expr *x)
989 free_pattern_expr(x->u.unary);
993 free_pattern_expr(x->u.binary.left);
994 free_pattern_expr(x->u.binary.right);
1000 void free_grep_patterns(struct grep_opt *opt)
1002 struct grep_pat *p, *n;
1004 for (p = opt->pattern_list; p; p = n) {
1007 case GREP_PATTERN: /* atom */
1008 case GREP_PATTERN_HEAD:
1009 case GREP_PATTERN_BODY:
1012 else if (p->pcre1_regexp)
1013 free_pcre1_regexp(p);
1014 else if (p->pcre2_pattern)
1015 free_pcre2_pattern(p);
1017 regfree(&p->regexp);
1028 free_pattern_expr(opt->pattern_expression);
1031 static char *end_of_line(char *cp, unsigned long *left)
1033 unsigned long l = *left;
1034 while (l && *cp != '\n') {
1042 static int word_char(char ch)
1044 return isalnum(ch) || ch == '_';
1047 static void output_color(struct grep_opt *opt, const void *data, size_t size,
1050 if (want_color(opt->color) && color && color[0]) {
1051 opt->output(opt, color, strlen(color));
1052 opt->output(opt, data, size);
1053 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
1055 opt->output(opt, data, size);
1058 static void output_sep(struct grep_opt *opt, char sign)
1060 if (opt->null_following_name)
1061 opt->output(opt, "\0", 1);
1063 output_color(opt, &sign, 1, opt->color_sep);
1066 static void show_name(struct grep_opt *opt, const char *name)
1068 output_color(opt, name, strlen(name), opt->color_filename);
1069 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
1072 static int fixmatch(struct grep_pat *p, char *line, char *eol,
1075 struct kwsmatch kwsm;
1076 size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
1078 match->rm_so = match->rm_eo = -1;
1081 match->rm_so = offset;
1082 match->rm_eo = match->rm_so + kwsm.size[0];
1087 static int patmatch(struct grep_pat *p, char *line, char *eol,
1088 regmatch_t *match, int eflags)
1093 hit = !fixmatch(p, line, eol, match);
1094 else if (p->pcre1_regexp)
1095 hit = !pcre1match(p, line, eol, match, eflags);
1096 else if (p->pcre2_pattern)
1097 hit = !pcre2match(p, line, eol, match, eflags);
1099 hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
1105 static int strip_timestamp(char *bol, char **eol_p)
1110 while (bol < --eol) {
1124 } header_field[] = {
1126 { "committer ", 10 },
1130 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
1131 enum grep_context ctx,
1132 regmatch_t *pmatch, int eflags)
1136 const char *start = bol;
1138 if ((p->token != GREP_PATTERN) &&
1139 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
1142 if (p->token == GREP_PATTERN_HEAD) {
1145 assert(p->field < ARRAY_SIZE(header_field));
1146 field = header_field[p->field].field;
1147 len = header_field[p->field].len;
1148 if (strncmp(bol, field, len))
1152 case GREP_HEADER_AUTHOR:
1153 case GREP_HEADER_COMMITTER:
1154 saved_ch = strip_timestamp(bol, &eol);
1162 hit = patmatch(p, bol, eol, pmatch, eflags);
1164 if (hit && p->word_regexp) {
1165 if ((pmatch[0].rm_so < 0) ||
1166 (eol - bol) < pmatch[0].rm_so ||
1167 (pmatch[0].rm_eo < 0) ||
1168 (eol - bol) < pmatch[0].rm_eo)
1169 die("regexp returned nonsense");
1171 /* Match beginning must be either beginning of the
1172 * line, or at word boundary (i.e. the last char must
1173 * not be a word char). Similarly, match end must be
1174 * either end of the line, or at word boundary
1175 * (i.e. the next char must not be a word char).
1177 if ( ((pmatch[0].rm_so == 0) ||
1178 !word_char(bol[pmatch[0].rm_so-1])) &&
1179 ((pmatch[0].rm_eo == (eol-bol)) ||
1180 !word_char(bol[pmatch[0].rm_eo])) )
1185 /* Words consist of at least one character. */
1186 if (pmatch->rm_so == pmatch->rm_eo)
1189 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
1190 /* There could be more than one match on the
1191 * line, and the first match might not be
1192 * strict word match. But later ones could be!
1193 * Forward to the next possible start, i.e. the
1194 * next position following a non-word char.
1196 bol = pmatch[0].rm_so + bol + 1;
1197 while (word_char(bol[-1]) && bol < eol)
1199 eflags |= REG_NOTBOL;
1204 if (p->token == GREP_PATTERN_HEAD && saved_ch)
1207 pmatch[0].rm_so += bol - start;
1208 pmatch[0].rm_eo += bol - start;
1213 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
1214 enum grep_context ctx, int collect_hits)
1220 die("Not a valid grep expression");
1222 case GREP_NODE_TRUE:
1225 case GREP_NODE_ATOM:
1226 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
1229 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
1232 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
1234 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
1238 return (match_expr_eval(x->u.binary.left,
1239 bol, eol, ctx, 0) ||
1240 match_expr_eval(x->u.binary.right,
1242 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
1243 x->u.binary.left->hit |= h;
1244 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
1247 die("Unexpected node type (internal error) %d", x->node);
1254 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
1255 enum grep_context ctx, int collect_hits)
1257 struct grep_expr *x = opt->pattern_expression;
1258 return match_expr_eval(x, bol, eol, ctx, collect_hits);
1261 static int match_line(struct grep_opt *opt, char *bol, char *eol,
1262 enum grep_context ctx, int collect_hits)
1268 return match_expr(opt, bol, eol, ctx, collect_hits);
1270 /* we do not call with collect_hits without being extended */
1271 for (p = opt->pattern_list; p; p = p->next) {
1272 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
1278 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1279 enum grep_context ctx,
1280 regmatch_t *pmatch, int eflags)
1284 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1286 if (match.rm_so < 0 || match.rm_eo < 0)
1288 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1289 if (match.rm_so > pmatch->rm_so)
1291 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1294 pmatch->rm_so = match.rm_so;
1295 pmatch->rm_eo = match.rm_eo;
1299 static int next_match(struct grep_opt *opt, char *bol, char *eol,
1300 enum grep_context ctx, regmatch_t *pmatch, int eflags)
1305 pmatch->rm_so = pmatch->rm_eo = -1;
1307 for (p = opt->pattern_list; p; p = p->next) {
1309 case GREP_PATTERN: /* atom */
1310 case GREP_PATTERN_HEAD:
1311 case GREP_PATTERN_BODY:
1312 hit |= match_next_pattern(p, bol, eol, ctx,
1323 static void show_line(struct grep_opt *opt, char *bol, char *eol,
1324 const char *name, unsigned lno, char sign)
1326 int rest = eol - bol;
1327 const char *match_color, *line_color = NULL;
1329 if (opt->file_break && opt->last_shown == 0) {
1330 if (opt->show_hunk_mark)
1331 opt->output(opt, "\n", 1);
1332 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
1333 if (opt->last_shown == 0) {
1334 if (opt->show_hunk_mark) {
1335 output_color(opt, "--", 2, opt->color_sep);
1336 opt->output(opt, "\n", 1);
1338 } else if (lno > opt->last_shown + 1) {
1339 output_color(opt, "--", 2, opt->color_sep);
1340 opt->output(opt, "\n", 1);
1343 if (opt->heading && opt->last_shown == 0) {
1344 output_color(opt, name, strlen(name), opt->color_filename);
1345 opt->output(opt, "\n", 1);
1347 opt->last_shown = lno;
1349 if (!opt->heading && opt->pathname) {
1350 output_color(opt, name, strlen(name), opt->color_filename);
1351 output_sep(opt, sign);
1355 xsnprintf(buf, sizeof(buf), "%d", lno);
1356 output_color(opt, buf, strlen(buf), opt->color_lineno);
1357 output_sep(opt, sign);
1361 enum grep_context ctx = GREP_CONTEXT_BODY;
1366 match_color = opt->color_match_selected;
1368 match_color = opt->color_match_context;
1370 line_color = opt->color_selected;
1371 else if (sign == '-')
1372 line_color = opt->color_context;
1373 else if (sign == '=')
1374 line_color = opt->color_function;
1376 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1377 if (match.rm_so == match.rm_eo)
1380 output_color(opt, bol, match.rm_so, line_color);
1381 output_color(opt, bol + match.rm_so,
1382 match.rm_eo - match.rm_so, match_color);
1384 rest -= match.rm_eo;
1385 eflags = REG_NOTBOL;
1389 output_color(opt, bol, rest, line_color);
1390 opt->output(opt, "\n", 1);
1397 * This lock protects access to the gitattributes machinery, which is
1400 pthread_mutex_t grep_attr_mutex;
1402 static inline void grep_attr_lock(void)
1405 pthread_mutex_lock(&grep_attr_mutex);
1408 static inline void grep_attr_unlock(void)
1411 pthread_mutex_unlock(&grep_attr_mutex);
1415 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1417 pthread_mutex_t grep_read_mutex;
1420 #define grep_attr_lock()
1421 #define grep_attr_unlock()
1424 static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
1426 xdemitconf_t *xecfg = opt->priv;
1427 if (xecfg && !xecfg->find_func) {
1428 grep_source_load_driver(gs);
1429 if (gs->driver->funcname.pattern) {
1430 const struct userdiff_funcname *pe = &gs->driver->funcname;
1431 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1433 xecfg = opt->priv = NULL;
1439 return xecfg->find_func(bol, eol - bol, buf, 1,
1440 xecfg->find_func_priv) >= 0;
1445 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1450 static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1451 char *bol, unsigned lno)
1453 while (bol > gs->buf) {
1456 while (bol > gs->buf && bol[-1] != '\n')
1460 if (lno <= opt->last_shown)
1463 if (match_funcname(opt, gs, bol, eol)) {
1464 show_line(opt, bol, eol, gs->name, lno, '=');
1470 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
1471 char *bol, char *end, unsigned lno)
1473 unsigned cur = lno, from = 1, funcname_lno = 0;
1474 int funcname_needed = !!opt->funcname;
1476 if (opt->funcbody && !match_funcname(opt, gs, bol, end))
1477 funcname_needed = 2;
1479 if (opt->pre_context < lno)
1480 from = lno - opt->pre_context;
1481 if (from <= opt->last_shown)
1482 from = opt->last_shown + 1;
1485 while (bol > gs->buf &&
1486 cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
1489 while (bol > gs->buf && bol[-1] != '\n')
1492 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
1494 funcname_needed = 0;
1498 /* We need to look even further back to find a function signature. */
1499 if (opt->funcname && funcname_needed)
1500 show_funcname_line(opt, gs, bol, cur);
1504 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
1506 while (*eol != '\n')
1508 show_line(opt, bol, eol, gs->name, cur, sign);
1514 static int should_lookahead(struct grep_opt *opt)
1519 return 0; /* punt for too complex stuff */
1522 for (p = opt->pattern_list; p; p = p->next) {
1523 if (p->token != GREP_PATTERN)
1524 return 0; /* punt for "header only" and stuff */
1529 static int look_ahead(struct grep_opt *opt,
1530 unsigned long *left_p,
1534 unsigned lno = *lno_p;
1537 char *sp, *last_bol;
1538 regoff_t earliest = -1;
1540 for (p = opt->pattern_list; p; p = p->next) {
1544 hit = patmatch(p, bol, bol + *left_p, &m, 0);
1545 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1547 if (earliest < 0 || m.rm_so < earliest)
1552 *bol_p = bol + *left_p;
1556 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1557 ; /* find the beginning of the line */
1560 for (sp = bol; sp < last_bol; sp++) {
1564 *left_p -= last_bol - bol;
1570 static int fill_textconv_grep(struct userdiff_driver *driver,
1571 struct grep_source *gs)
1573 struct diff_filespec *df;
1577 if (!driver || !driver->textconv)
1578 return grep_source_load(gs);
1581 * The textconv interface is intimately tied to diff_filespecs, so we
1582 * have to pretend to be one. If we could unify the grep_source
1583 * and diff_filespec structs, this mess could just go away.
1585 df = alloc_filespec(gs->path);
1587 case GREP_SOURCE_OID:
1588 fill_filespec(df, gs->identifier, 1, 0100644);
1590 case GREP_SOURCE_FILE:
1591 fill_filespec(df, &null_oid, 0, 0100644);
1594 die("BUG: attempt to textconv something without a path?");
1598 * fill_textconv is not remotely thread-safe; it may load objects
1599 * behind the scenes, and it modifies the global diff tempfile
1603 size = fill_textconv(driver, df, &buf);
1608 * The normal fill_textconv usage by the diff machinery would just keep
1609 * the textconv'd buf separate from the diff_filespec. But much of the
1610 * grep code passes around a grep_source and assumes that its "buf"
1611 * pointer is the beginning of the thing we are searching. So let's
1612 * install our textconv'd version into the grep_source, taking care not
1613 * to leak any existing buffer.
1615 grep_source_clear_data(gs);
1622 static int is_empty_line(const char *bol, const char *eol)
1624 while (bol < eol && isspace(*bol))
1629 static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
1632 char *peek_bol = NULL;
1635 unsigned last_hit = 0;
1636 int binary_match_only = 0;
1638 int try_lookahead = 0;
1639 int show_function = 0;
1640 struct userdiff_driver *textconv = NULL;
1641 enum grep_context ctx = GREP_CONTEXT_HEAD;
1645 opt->output = std_output;
1647 if (opt->pre_context || opt->post_context || opt->file_break ||
1649 /* Show hunk marks, except for the first file. */
1650 if (opt->last_shown)
1651 opt->show_hunk_mark = 1;
1653 * If we're using threads then we can't easily identify
1654 * the first file. Always put hunk marks in that case
1655 * and skip the very first one later in work_done().
1657 if (opt->output != std_output)
1658 opt->show_hunk_mark = 1;
1660 opt->last_shown = 0;
1662 if (opt->allow_textconv) {
1663 grep_source_load_driver(gs);
1665 * We might set up the shared textconv cache data here, which
1666 * is not thread-safe.
1669 textconv = userdiff_get_textconv(gs->driver);
1674 * We know the result of a textconv is text, so we only have to care
1675 * about binary handling if we are not using it.
1678 switch (opt->binary) {
1679 case GREP_BINARY_DEFAULT:
1680 if (grep_source_is_binary(gs))
1681 binary_match_only = 1;
1683 case GREP_BINARY_NOMATCH:
1684 if (grep_source_is_binary(gs))
1685 return 0; /* Assume unmatch */
1687 case GREP_BINARY_TEXT:
1690 die("BUG: unknown binary handling mode");
1694 memset(&xecfg, 0, sizeof(xecfg));
1697 try_lookahead = should_lookahead(opt);
1699 if (fill_textconv_grep(textconv, gs) < 0)
1709 * look_ahead() skips quickly to the line that possibly
1710 * has the next hit; don't call it if we need to do
1711 * something more than just skipping the current line
1712 * in response to an unmatch for the current line. E.g.
1713 * inside a post-context window, we will show the current
1714 * line as a context around the previous hit when it
1719 && (show_function ||
1720 lno <= last_hit + opt->post_context))
1721 && look_ahead(opt, &left, &lno, &bol))
1723 eol = end_of_line(bol, &left);
1727 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1728 ctx = GREP_CONTEXT_BODY;
1730 hit = match_line(opt, bol, eol, ctx, collect_hits);
1736 /* "grep -v -e foo -e bla" should list lines
1737 * that do not have either, so inversion should
1742 if (opt->unmatch_name_only) {
1749 if (opt->status_only)
1751 if (opt->name_only) {
1752 show_name(opt, gs->name);
1757 if (binary_match_only) {
1758 opt->output(opt, "Binary file ", 12);
1759 output_color(opt, gs->name, strlen(gs->name),
1760 opt->color_filename);
1761 opt->output(opt, " matches\n", 9);
1764 /* Hit at this line. If we haven't shown the
1765 * pre-context lines, we would need to show them.
1767 if (opt->pre_context || opt->funcbody)
1768 show_pre_context(opt, gs, bol, eol, lno);
1769 else if (opt->funcname)
1770 show_funcname_line(opt, gs, bol, lno);
1771 show_line(opt, bol, eol, gs->name, lno, ':');
1777 if (show_function && (!peek_bol || peek_bol < bol)) {
1778 unsigned long peek_left = left;
1779 char *peek_eol = eol;
1782 * Trailing empty lines are not interesting.
1783 * Peek past them to see if they belong to the
1784 * body of the current function.
1787 while (is_empty_line(peek_bol, peek_eol)) {
1788 peek_bol = peek_eol + 1;
1789 peek_eol = end_of_line(peek_bol, &peek_left);
1792 if (match_funcname(opt, gs, peek_bol, peek_eol))
1795 if (show_function ||
1796 (last_hit && lno <= last_hit + opt->post_context)) {
1797 /* If the last hit is within the post context,
1798 * we need to show this line.
1800 show_line(opt, bol, eol, gs->name, lno, '-');
1814 if (opt->status_only)
1816 if (opt->unmatch_name_only) {
1817 /* We did not see any hit, so we want to show this */
1818 show_name(opt, gs->name);
1822 xdiff_clear_find_func(&xecfg);
1826 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1827 * which feels mostly useless but sometimes useful. Maybe
1828 * make it another option? For now suppress them.
1830 if (opt->count && count) {
1832 if (opt->pathname) {
1833 output_color(opt, gs->name, strlen(gs->name),
1834 opt->color_filename);
1835 output_sep(opt, ':');
1837 xsnprintf(buf, sizeof(buf), "%u\n", count);
1838 opt->output(opt, buf, strlen(buf));
1844 static void clr_hit_marker(struct grep_expr *x)
1846 /* All-hit markers are meaningful only at the very top level
1851 if (x->node != GREP_NODE_OR)
1853 x->u.binary.left->hit = 0;
1854 x = x->u.binary.right;
1858 static int chk_hit_marker(struct grep_expr *x)
1860 /* Top level nodes have hit markers. See if they all are hits */
1862 if (x->node != GREP_NODE_OR)
1864 if (!x->u.binary.left->hit)
1866 x = x->u.binary.right;
1870 int grep_source(struct grep_opt *opt, struct grep_source *gs)
1873 * we do not have to do the two-pass grep when we do not check
1874 * buffer-wide "all-match".
1876 if (!opt->all_match)
1877 return grep_source_1(opt, gs, 0);
1879 /* Otherwise the toplevel "or" terms hit a bit differently.
1880 * We first clear hit markers from them.
1882 clr_hit_marker(opt->pattern_expression);
1883 grep_source_1(opt, gs, 1);
1885 if (!chk_hit_marker(opt->pattern_expression))
1888 return grep_source_1(opt, gs, 0);
1891 int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
1893 struct grep_source gs;
1896 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
1900 r = grep_source(opt, &gs);
1902 grep_source_clear(&gs);
1906 void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1907 const char *name, const char *path,
1908 const void *identifier)
1911 gs->name = xstrdup_or_null(name);
1912 gs->path = xstrdup_or_null(path);
1918 case GREP_SOURCE_FILE:
1919 gs->identifier = xstrdup(identifier);
1921 case GREP_SOURCE_SUBMODULE:
1923 gs->identifier = NULL;
1928 * If the identifier is non-NULL (in the submodule case) it
1929 * will be a SHA1 that needs to be copied.
1931 case GREP_SOURCE_OID:
1932 gs->identifier = oiddup(identifier);
1934 case GREP_SOURCE_BUF:
1935 gs->identifier = NULL;
1940 void grep_source_clear(struct grep_source *gs)
1946 free(gs->identifier);
1947 gs->identifier = NULL;
1948 grep_source_clear_data(gs);
1951 void grep_source_clear_data(struct grep_source *gs)
1954 case GREP_SOURCE_FILE:
1955 case GREP_SOURCE_OID:
1956 case GREP_SOURCE_SUBMODULE:
1961 case GREP_SOURCE_BUF:
1962 /* leave user-provided buf intact */
1967 static int grep_source_load_oid(struct grep_source *gs)
1969 enum object_type type;
1972 gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1976 return error(_("'%s': unable to read %s"),
1978 oid_to_hex(gs->identifier));
1982 static int grep_source_load_file(struct grep_source *gs)
1984 const char *filename = gs->identifier;
1990 if (lstat(filename, &st) < 0) {
1992 if (errno != ENOENT)
1993 error_errno(_("failed to stat '%s'"), filename);
1996 if (!S_ISREG(st.st_mode))
1998 size = xsize_t(st.st_size);
1999 i = open(filename, O_RDONLY);
2002 data = xmallocz(size);
2003 if (st.st_size != read_in_full(i, data, size)) {
2004 error_errno(_("'%s': short read"), filename);
2016 static int grep_source_load(struct grep_source *gs)
2022 case GREP_SOURCE_FILE:
2023 return grep_source_load_file(gs);
2024 case GREP_SOURCE_OID:
2025 return grep_source_load_oid(gs);
2026 case GREP_SOURCE_BUF:
2027 return gs->buf ? 0 : -1;
2028 case GREP_SOURCE_SUBMODULE:
2031 die("BUG: invalid grep_source type to load");
2034 void grep_source_load_driver(struct grep_source *gs)
2041 gs->driver = userdiff_find_by_path(gs->path);
2043 gs->driver = userdiff_find_by_name("default");
2047 static int grep_source_is_binary(struct grep_source *gs)
2049 grep_source_load_driver(gs);
2050 if (gs->driver->binary != -1)
2051 return gs->driver->binary;
2053 if (!grep_source_load(gs))
2054 return buffer_is_binary(gs->buf, gs->size);