5 #include "xdiff-interface.h"
11 static int grep_source_load(struct grep_source
*gs
);
12 static int grep_source_is_binary(struct grep_source
*gs
);
14 static struct grep_opt grep_defaults
;
16 static void std_output(struct grep_opt
*opt
, const void *buf
, size_t size
)
18 fwrite(buf
, size
, 1, stdout
);
22 * Initialize the grep_defaults template with hardcoded defaults.
23 * We could let the compiler do this, but without C99 initializers
24 * the code gets unwieldy and unreadable, so...
26 void init_grep_defaults(void)
28 struct grep_opt
*opt
= &grep_defaults
;
35 memset(opt
, 0, sizeof(*opt
));
38 opt
->regflags
= REG_NEWLINE
;
40 opt
->pattern_type_option
= GREP_PATTERN_TYPE_UNSPECIFIED
;
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 opt
->extended_regexp_option
= git_config_bool(var
, value
);
85 if (!strcmp(var
, "grep.patterntype")) {
86 opt
->pattern_type_option
= parse_pattern_type_arg(var
, value
);
90 if (!strcmp(var
, "grep.linenumber")) {
91 opt
->linenum
= git_config_bool(var
, value
);
95 if (!strcmp(var
, "grep.fullname")) {
96 opt
->relative
= !git_config_bool(var
, value
);
100 if (!strcmp(var
, "color.grep"))
101 opt
->color
= git_config_colorbool(var
, value
);
102 else if (!strcmp(var
, "color.grep.context"))
103 color
= opt
->color_context
;
104 else if (!strcmp(var
, "color.grep.filename"))
105 color
= opt
->color_filename
;
106 else if (!strcmp(var
, "color.grep.function"))
107 color
= opt
->color_function
;
108 else if (!strcmp(var
, "color.grep.linenumber"))
109 color
= opt
->color_lineno
;
110 else if (!strcmp(var
, "color.grep.matchcontext"))
111 color
= opt
->color_match_context
;
112 else if (!strcmp(var
, "color.grep.matchselected"))
113 color
= opt
->color_match_selected
;
114 else if (!strcmp(var
, "color.grep.selected"))
115 color
= opt
->color_selected
;
116 else if (!strcmp(var
, "color.grep.separator"))
117 color
= opt
->color_sep
;
118 else if (!strcmp(var
, "color.grep.match")) {
121 return config_error_nonbool(var
);
122 rc
|= color_parse(value
, opt
->color_match_context
);
123 rc
|= color_parse(value
, opt
->color_match_selected
);
129 return config_error_nonbool(var
);
130 return color_parse(value
, color
);
136 * Initialize one instance of grep_opt and copy the
137 * default values from the template we read the configuration
138 * information in an earlier call to git_config(grep_config).
140 void grep_init(struct grep_opt
*opt
, const char *prefix
)
142 struct grep_opt
*def
= &grep_defaults
;
144 memset(opt
, 0, sizeof(*opt
));
145 opt
->prefix
= prefix
;
146 opt
->prefix_length
= (prefix
&& *prefix
) ? strlen(prefix
) : 0;
147 opt
->pattern_tail
= &opt
->pattern_list
;
148 opt
->header_tail
= &opt
->header_list
;
150 opt
->color
= def
->color
;
151 opt
->extended_regexp_option
= def
->extended_regexp_option
;
152 opt
->pattern_type_option
= def
->pattern_type_option
;
153 opt
->linenum
= def
->linenum
;
154 opt
->max_depth
= def
->max_depth
;
155 opt
->pathname
= def
->pathname
;
156 opt
->regflags
= def
->regflags
;
157 opt
->relative
= def
->relative
;
158 opt
->output
= def
->output
;
160 color_set(opt
->color_context
, def
->color_context
);
161 color_set(opt
->color_filename
, def
->color_filename
);
162 color_set(opt
->color_function
, def
->color_function
);
163 color_set(opt
->color_lineno
, def
->color_lineno
);
164 color_set(opt
->color_match_context
, def
->color_match_context
);
165 color_set(opt
->color_match_selected
, def
->color_match_selected
);
166 color_set(opt
->color_selected
, def
->color_selected
);
167 color_set(opt
->color_sep
, def
->color_sep
);
170 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type
, struct grep_opt
*opt
)
172 switch (pattern_type
) {
173 case GREP_PATTERN_TYPE_UNSPECIFIED
:
176 case GREP_PATTERN_TYPE_BRE
:
182 case GREP_PATTERN_TYPE_ERE
:
186 opt
->regflags
|= REG_EXTENDED
;
189 case GREP_PATTERN_TYPE_FIXED
:
195 case GREP_PATTERN_TYPE_PCRE
:
202 * It's important that pcre1 always be assigned to
203 * even when there's no USE_LIBPCRE* defined. We still
204 * call the PCRE stub function, it just dies with
205 * "cannot use Perl-compatible regexes[...]".
214 void grep_commit_pattern_type(enum grep_pattern_type pattern_type
, struct grep_opt
*opt
)
216 if (pattern_type
!= GREP_PATTERN_TYPE_UNSPECIFIED
)
217 grep_set_pattern_type_option(pattern_type
, opt
);
218 else if (opt
->pattern_type_option
!= GREP_PATTERN_TYPE_UNSPECIFIED
)
219 grep_set_pattern_type_option(opt
->pattern_type_option
, opt
);
220 else if (opt
->extended_regexp_option
)
221 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE
, opt
);
224 static struct grep_pat
*create_grep_pat(const char *pat
, size_t patlen
,
225 const char *origin
, int no
,
226 enum grep_pat_token t
,
227 enum grep_header_field field
)
229 struct grep_pat
*p
= xcalloc(1, sizeof(*p
));
230 p
->pattern
= xmemdupz(pat
, patlen
);
231 p
->patternlen
= patlen
;
239 static void do_append_grep_pat(struct grep_pat
***tail
, struct grep_pat
*p
)
246 case GREP_PATTERN
: /* atom */
247 case GREP_PATTERN_HEAD
:
248 case GREP_PATTERN_BODY
:
250 struct grep_pat
*new_pat
;
252 char *cp
= p
->pattern
+ p
->patternlen
, *nl
= NULL
;
253 while (++len
<= p
->patternlen
) {
254 if (*(--cp
) == '\n') {
261 new_pat
= create_grep_pat(nl
+ 1, len
- 1, p
->origin
,
262 p
->no
, p
->token
, p
->field
);
263 new_pat
->next
= p
->next
;
265 *tail
= &new_pat
->next
;
268 p
->patternlen
-= len
;
276 void append_header_grep_pattern(struct grep_opt
*opt
,
277 enum grep_header_field field
, const char *pat
)
279 struct grep_pat
*p
= create_grep_pat(pat
, strlen(pat
), "header", 0,
280 GREP_PATTERN_HEAD
, field
);
281 if (field
== GREP_HEADER_REFLOG
)
282 opt
->use_reflog_filter
= 1;
283 do_append_grep_pat(&opt
->header_tail
, p
);
286 void append_grep_pattern(struct grep_opt
*opt
, const char *pat
,
287 const char *origin
, int no
, enum grep_pat_token t
)
289 append_grep_pat(opt
, pat
, strlen(pat
), origin
, no
, t
);
292 void append_grep_pat(struct grep_opt
*opt
, const char *pat
, size_t patlen
,
293 const char *origin
, int no
, enum grep_pat_token t
)
295 struct grep_pat
*p
= create_grep_pat(pat
, patlen
, origin
, no
, t
, 0);
296 do_append_grep_pat(&opt
->pattern_tail
, p
);
299 struct grep_opt
*grep_opt_dup(const struct grep_opt
*opt
)
301 struct grep_pat
*pat
;
302 struct grep_opt
*ret
= xmalloc(sizeof(struct grep_opt
));
305 ret
->pattern_list
= NULL
;
306 ret
->pattern_tail
= &ret
->pattern_list
;
308 for(pat
= opt
->pattern_list
; pat
!= NULL
; pat
= pat
->next
)
310 if(pat
->token
== GREP_PATTERN_HEAD
)
311 append_header_grep_pattern(ret
, pat
->field
,
314 append_grep_pat(ret
, pat
->pattern
, pat
->patternlen
,
315 pat
->origin
, pat
->no
, pat
->token
);
321 static NORETURN
void compile_regexp_failed(const struct grep_pat
*p
,
327 xsnprintf(where
, sizeof(where
), "In '%s' at %d, ", p
->origin
, p
->no
);
329 xsnprintf(where
, sizeof(where
), "%s, ", p
->origin
);
333 die("%s'%s': %s", where
, p
->pattern
, error
);
336 static int is_fixed(const char *s
, size_t len
)
340 for (i
= 0; i
< len
; i
++) {
341 if (is_regex_special(s
[i
]))
348 static int has_null(const char *s
, size_t len
)
351 * regcomp cannot accept patterns with NULs so when using it
352 * we consider any pattern containing a NUL fixed.
354 if (memchr(s
, 0, len
))
361 static void compile_pcre1_regexp(struct grep_pat
*p
, const struct grep_opt
*opt
)
365 int options
= PCRE_MULTILINE
;
367 if (opt
->ignore_case
) {
368 if (has_non_ascii(p
->pattern
))
369 p
->pcre1_tables
= pcre_maketables();
370 options
|= PCRE_CASELESS
;
372 if (is_utf8_locale() && has_non_ascii(p
->pattern
))
373 options
|= PCRE_UTF8
;
375 p
->pcre1_regexp
= pcre_compile(p
->pattern
, options
, &error
, &erroffset
,
377 if (!p
->pcre1_regexp
)
378 compile_regexp_failed(p
, error
);
380 p
->pcre1_extra_info
= pcre_study(p
->pcre1_regexp
, PCRE_STUDY_JIT_COMPILE
, &error
);
381 if (!p
->pcre1_extra_info
&& error
)
384 #ifdef GIT_PCRE1_USE_JIT
385 pcre_config(PCRE_CONFIG_JIT
, &p
->pcre1_jit_on
);
386 if (p
->pcre1_jit_on
== 1) {
387 p
->pcre1_jit_stack
= pcre_jit_stack_alloc(1, 1024 * 1024);
388 if (!p
->pcre1_jit_stack
)
389 die("Couldn't allocate PCRE JIT stack");
390 pcre_assign_jit_stack(p
->pcre1_extra_info
, NULL
, p
->pcre1_jit_stack
);
391 } else if (p
->pcre1_jit_on
!= 0) {
392 die("BUG: The pcre1_jit_on variable should be 0 or 1, not %d",
398 static int pcre1match(struct grep_pat
*p
, const char *line
, const char *eol
,
399 regmatch_t
*match
, int eflags
)
401 int ovector
[30], ret
, flags
= 0;
403 if (eflags
& REG_NOTBOL
)
404 flags
|= PCRE_NOTBOL
;
406 #ifdef GIT_PCRE1_USE_JIT
407 if (p
->pcre1_jit_on
) {
408 ret
= pcre_jit_exec(p
->pcre1_regexp
, p
->pcre1_extra_info
, line
,
409 eol
- line
, 0, flags
, ovector
,
410 ARRAY_SIZE(ovector
), p
->pcre1_jit_stack
);
414 ret
= pcre_exec(p
->pcre1_regexp
, p
->pcre1_extra_info
, line
,
415 eol
- line
, 0, flags
, ovector
,
416 ARRAY_SIZE(ovector
));
419 if (ret
< 0 && ret
!= PCRE_ERROR_NOMATCH
)
420 die("pcre_exec failed with error code %d", ret
);
423 match
->rm_so
= ovector
[0];
424 match
->rm_eo
= ovector
[1];
430 static void free_pcre1_regexp(struct grep_pat
*p
)
432 pcre_free(p
->pcre1_regexp
);
433 #ifdef GIT_PCRE1_USE_JIT
434 if (p
->pcre1_jit_on
) {
435 pcre_free_study(p
->pcre1_extra_info
);
436 pcre_jit_stack_free(p
->pcre1_jit_stack
);
440 pcre_free(p
->pcre1_extra_info
);
442 pcre_free((void *)p
->pcre1_tables
);
444 #else /* !USE_LIBPCRE1 */
445 static void compile_pcre1_regexp(struct grep_pat
*p
, const struct grep_opt
*opt
)
447 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
450 static int pcre1match(struct grep_pat
*p
, const char *line
, const char *eol
,
451 regmatch_t
*match
, int eflags
)
456 static void free_pcre1_regexp(struct grep_pat
*p
)
459 #endif /* !USE_LIBPCRE1 */
462 static void compile_pcre2_pattern(struct grep_pat
*p
, const struct grep_opt
*opt
)
465 PCRE2_UCHAR errbuf
[256];
466 PCRE2_SIZE erroffset
;
467 int options
= PCRE2_MULTILINE
;
468 const uint8_t *character_tables
= NULL
;
473 p
->pcre2_compile_context
= NULL
;
475 if (opt
->ignore_case
) {
476 if (has_non_ascii(p
->pattern
)) {
477 character_tables
= pcre2_maketables(NULL
);
478 p
->pcre2_compile_context
= pcre2_compile_context_create(NULL
);
479 pcre2_set_character_tables(p
->pcre2_compile_context
, character_tables
);
481 options
|= PCRE2_CASELESS
;
483 if (is_utf8_locale() && has_non_ascii(p
->pattern
))
484 options
|= PCRE2_UTF
;
486 p
->pcre2_pattern
= pcre2_compile((PCRE2_SPTR
)p
->pattern
,
487 p
->patternlen
, options
, &error
, &erroffset
,
488 p
->pcre2_compile_context
);
490 if (p
->pcre2_pattern
) {
491 p
->pcre2_match_data
= pcre2_match_data_create_from_pattern(p
->pcre2_pattern
, NULL
);
492 if (!p
->pcre2_match_data
)
493 die("Couldn't allocate PCRE2 match data");
495 pcre2_get_error_message(error
, errbuf
, sizeof(errbuf
));
496 compile_regexp_failed(p
, (const char *)&errbuf
);
499 pcre2_config(PCRE2_CONFIG_JIT
, &p
->pcre2_jit_on
);
500 if (p
->pcre2_jit_on
== 1) {
501 jitret
= pcre2_jit_compile(p
->pcre2_pattern
, PCRE2_JIT_COMPLETE
);
503 die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p
->pattern
, jitret
);
504 p
->pcre2_jit_stack
= pcre2_jit_stack_create(1, 1024 * 1024, NULL
);
505 if (!p
->pcre2_jit_stack
)
506 die("Couldn't allocate PCRE2 JIT stack");
507 p
->pcre2_match_context
= pcre2_match_context_create(NULL
);
508 if (!p
->pcre2_match_context
)
509 die("Couldn't allocate PCRE2 match context");
510 pcre2_jit_stack_assign(p
->pcre2_match_context
, NULL
, p
->pcre2_jit_stack
);
511 } else if (p
->pcre2_jit_on
!= 0) {
512 die("BUG: The pcre2_jit_on variable should be 0 or 1, not %d",
517 static int pcre2match(struct grep_pat
*p
, const char *line
, const char *eol
,
518 regmatch_t
*match
, int eflags
)
522 PCRE2_UCHAR errbuf
[256];
524 if (eflags
& REG_NOTBOL
)
525 flags
|= PCRE2_NOTBOL
;
528 ret
= pcre2_jit_match(p
->pcre2_pattern
, (unsigned char *)line
,
529 eol
- line
, 0, flags
, p
->pcre2_match_data
,
532 ret
= pcre2_match(p
->pcre2_pattern
, (unsigned char *)line
,
533 eol
- line
, 0, flags
, p
->pcre2_match_data
,
536 if (ret
< 0 && ret
!= PCRE2_ERROR_NOMATCH
) {
537 pcre2_get_error_message(ret
, errbuf
, sizeof(errbuf
));
538 die("%s failed with error code %d: %s",
539 (p
->pcre2_jit_on
? "pcre2_jit_match" : "pcre2_match"), ret
,
543 ovector
= pcre2_get_ovector_pointer(p
->pcre2_match_data
);
545 match
->rm_so
= (int)ovector
[0];
546 match
->rm_eo
= (int)ovector
[1];
552 static void free_pcre2_pattern(struct grep_pat
*p
)
554 pcre2_compile_context_free(p
->pcre2_compile_context
);
555 pcre2_code_free(p
->pcre2_pattern
);
556 pcre2_match_data_free(p
->pcre2_match_data
);
557 pcre2_jit_stack_free(p
->pcre2_jit_stack
);
558 pcre2_match_context_free(p
->pcre2_match_context
);
560 #else /* !USE_LIBPCRE2 */
561 static void compile_pcre2_pattern(struct grep_pat
*p
, const struct grep_opt
*opt
)
564 * Unreachable until USE_LIBPCRE2 becomes synonymous with
565 * USE_LIBPCRE. See the sibling comment in
566 * grep_set_pattern_type_option().
568 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
571 static int pcre2match(struct grep_pat
*p
, const char *line
, const char *eol
,
572 regmatch_t
*match
, int eflags
)
577 static void free_pcre2_pattern(struct grep_pat
*p
)
580 #endif /* !USE_LIBPCRE2 */
582 static void compile_fixed_regexp(struct grep_pat
*p
, struct grep_opt
*opt
)
584 struct strbuf sb
= STRBUF_INIT
;
586 int regflags
= opt
->regflags
;
588 basic_regex_quote_buf(&sb
, p
->pattern
);
589 if (opt
->ignore_case
)
590 regflags
|= REG_ICASE
;
591 err
= regcomp(&p
->regexp
, sb
.buf
, regflags
);
593 fprintf(stderr
, "fixed %s\n", sb
.buf
);
597 regerror(err
, &p
->regexp
, errbuf
, sizeof(errbuf
));
599 compile_regexp_failed(p
, errbuf
);
603 static void compile_regexp(struct grep_pat
*p
, struct grep_opt
*opt
)
605 int icase
, ascii_only
;
608 p
->word_regexp
= opt
->word_regexp
;
609 p
->ignore_case
= opt
->ignore_case
;
610 icase
= opt
->regflags
& REG_ICASE
|| p
->ignore_case
;
611 ascii_only
= !has_non_ascii(p
->pattern
);
614 * Even when -F (fixed) asks us to do a non-regexp search, we
615 * may not be able to correctly case-fold when -i
616 * (ignore-case) is asked (in which case, we'll synthesize a
617 * regexp to match the pattern that matches regexp special
618 * characters literally, while ignoring case differences). On
619 * the other hand, even without -F, if the pattern does not
620 * have any regexp special characters and there is no need for
621 * case-folding search, we can internally turn it into a
622 * simple string match using kws. p->fixed tells us if we
626 has_null(p
->pattern
, p
->patternlen
) ||
627 is_fixed(p
->pattern
, p
->patternlen
))
628 p
->fixed
= !icase
|| ascii_only
;
633 p
->kws
= kwsalloc(icase
? tolower_trans_tbl
: NULL
);
634 kwsincr(p
->kws
, p
->pattern
, p
->patternlen
);
637 } else if (opt
->fixed
) {
639 * We come here when the pattern has the non-ascii
640 * characters we cannot case-fold, and asked to
643 compile_fixed_regexp(p
, opt
);
648 compile_pcre2_pattern(p
, opt
);
653 compile_pcre1_regexp(p
, opt
);
657 err
= regcomp(&p
->regexp
, p
->pattern
, opt
->regflags
);
660 regerror(err
, &p
->regexp
, errbuf
, 1024);
662 compile_regexp_failed(p
, errbuf
);
666 static struct grep_expr
*compile_pattern_or(struct grep_pat
**);
667 static struct grep_expr
*compile_pattern_atom(struct grep_pat
**list
)
676 case GREP_PATTERN
: /* atom */
677 case GREP_PATTERN_HEAD
:
678 case GREP_PATTERN_BODY
:
679 x
= xcalloc(1, sizeof (struct grep_expr
));
680 x
->node
= GREP_NODE_ATOM
;
684 case GREP_OPEN_PAREN
:
686 x
= compile_pattern_or(list
);
687 if (!*list
|| (*list
)->token
!= GREP_CLOSE_PAREN
)
688 die("unmatched parenthesis");
689 *list
= (*list
)->next
;
696 static struct grep_expr
*compile_pattern_not(struct grep_pat
**list
)
707 die("--not not followed by pattern expression");
709 x
= xcalloc(1, sizeof (struct grep_expr
));
710 x
->node
= GREP_NODE_NOT
;
711 x
->u
.unary
= compile_pattern_not(list
);
713 die("--not followed by non pattern expression");
716 return compile_pattern_atom(list
);
720 static struct grep_expr
*compile_pattern_and(struct grep_pat
**list
)
723 struct grep_expr
*x
, *y
, *z
;
725 x
= compile_pattern_not(list
);
727 if (p
&& p
->token
== GREP_AND
) {
729 die("--and not followed by pattern expression");
731 y
= compile_pattern_and(list
);
733 die("--and not followed by pattern expression");
734 z
= xcalloc(1, sizeof (struct grep_expr
));
735 z
->node
= GREP_NODE_AND
;
736 z
->u
.binary
.left
= x
;
737 z
->u
.binary
.right
= y
;
743 static struct grep_expr
*compile_pattern_or(struct grep_pat
**list
)
746 struct grep_expr
*x
, *y
, *z
;
748 x
= compile_pattern_and(list
);
750 if (x
&& p
&& p
->token
!= GREP_CLOSE_PAREN
) {
751 y
= compile_pattern_or(list
);
753 die("not a pattern expression %s", p
->pattern
);
754 z
= xcalloc(1, sizeof (struct grep_expr
));
755 z
->node
= GREP_NODE_OR
;
756 z
->u
.binary
.left
= x
;
757 z
->u
.binary
.right
= y
;
763 static struct grep_expr
*compile_pattern_expr(struct grep_pat
**list
)
765 return compile_pattern_or(list
);
768 static void indent(int in
)
774 static void dump_grep_pat(struct grep_pat
*p
)
777 case GREP_AND
: fprintf(stderr
, "*and*"); break;
778 case GREP_OPEN_PAREN
: fprintf(stderr
, "*(*"); break;
779 case GREP_CLOSE_PAREN
: fprintf(stderr
, "*)*"); break;
780 case GREP_NOT
: fprintf(stderr
, "*not*"); break;
781 case GREP_OR
: fprintf(stderr
, "*or*"); break;
783 case GREP_PATTERN
: fprintf(stderr
, "pattern"); break;
784 case GREP_PATTERN_HEAD
: fprintf(stderr
, "pattern_head"); break;
785 case GREP_PATTERN_BODY
: fprintf(stderr
, "pattern_body"); break;
790 case GREP_PATTERN_HEAD
:
791 fprintf(stderr
, "<head %d>", p
->field
); break;
792 case GREP_PATTERN_BODY
:
793 fprintf(stderr
, "<body>"); break;
797 case GREP_PATTERN_HEAD
:
798 case GREP_PATTERN_BODY
:
800 fprintf(stderr
, "%.*s", (int)p
->patternlen
, p
->pattern
);
806 static void dump_grep_expression_1(struct grep_expr
*x
, int in
)
811 fprintf(stderr
, "true\n");
814 dump_grep_pat(x
->u
.atom
);
817 fprintf(stderr
, "(not\n");
818 dump_grep_expression_1(x
->u
.unary
, in
+1);
820 fprintf(stderr
, ")\n");
823 fprintf(stderr
, "(and\n");
824 dump_grep_expression_1(x
->u
.binary
.left
, in
+1);
825 dump_grep_expression_1(x
->u
.binary
.right
, in
+1);
827 fprintf(stderr
, ")\n");
830 fprintf(stderr
, "(or\n");
831 dump_grep_expression_1(x
->u
.binary
.left
, in
+1);
832 dump_grep_expression_1(x
->u
.binary
.right
, in
+1);
834 fprintf(stderr
, ")\n");
839 static void dump_grep_expression(struct grep_opt
*opt
)
841 struct grep_expr
*x
= opt
->pattern_expression
;
844 fprintf(stderr
, "[all-match]\n");
845 dump_grep_expression_1(x
, 0);
849 static struct grep_expr
*grep_true_expr(void)
851 struct grep_expr
*z
= xcalloc(1, sizeof(*z
));
852 z
->node
= GREP_NODE_TRUE
;
856 static struct grep_expr
*grep_or_expr(struct grep_expr
*left
, struct grep_expr
*right
)
858 struct grep_expr
*z
= xcalloc(1, sizeof(*z
));
859 z
->node
= GREP_NODE_OR
;
860 z
->u
.binary
.left
= left
;
861 z
->u
.binary
.right
= right
;
865 static struct grep_expr
*prep_header_patterns(struct grep_opt
*opt
)
868 struct grep_expr
*header_expr
;
869 struct grep_expr
*(header_group
[GREP_HEADER_FIELD_MAX
]);
870 enum grep_header_field fld
;
872 if (!opt
->header_list
)
875 for (p
= opt
->header_list
; p
; p
= p
->next
) {
876 if (p
->token
!= GREP_PATTERN_HEAD
)
877 die("BUG: a non-header pattern in grep header list.");
878 if (p
->field
< GREP_HEADER_FIELD_MIN
||
879 GREP_HEADER_FIELD_MAX
<= p
->field
)
880 die("BUG: unknown header field %d", p
->field
);
881 compile_regexp(p
, opt
);
884 for (fld
= 0; fld
< GREP_HEADER_FIELD_MAX
; fld
++)
885 header_group
[fld
] = NULL
;
887 for (p
= opt
->header_list
; p
; p
= p
->next
) {
889 struct grep_pat
*pp
= p
;
891 h
= compile_pattern_atom(&pp
);
892 if (!h
|| pp
!= p
->next
)
893 die("BUG: malformed header expr");
894 if (!header_group
[p
->field
]) {
895 header_group
[p
->field
] = h
;
898 header_group
[p
->field
] = grep_or_expr(h
, header_group
[p
->field
]);
903 for (fld
= 0; fld
< GREP_HEADER_FIELD_MAX
; fld
++) {
904 if (!header_group
[fld
])
907 header_expr
= grep_true_expr();
908 header_expr
= grep_or_expr(header_group
[fld
], header_expr
);
913 static struct grep_expr
*grep_splice_or(struct grep_expr
*x
, struct grep_expr
*y
)
915 struct grep_expr
*z
= x
;
918 assert(x
->node
== GREP_NODE_OR
);
919 if (x
->u
.binary
.right
&&
920 x
->u
.binary
.right
->node
== GREP_NODE_TRUE
) {
921 x
->u
.binary
.right
= y
;
924 x
= x
->u
.binary
.right
;
929 static void compile_grep_patterns_real(struct grep_opt
*opt
)
932 struct grep_expr
*header_expr
= prep_header_patterns(opt
);
934 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
936 case GREP_PATTERN
: /* atom */
937 case GREP_PATTERN_HEAD
:
938 case GREP_PATTERN_BODY
:
939 compile_regexp(p
, opt
);
947 if (opt
->all_match
|| header_expr
)
949 else if (!opt
->extended
&& !opt
->debug
)
952 p
= opt
->pattern_list
;
954 opt
->pattern_expression
= compile_pattern_expr(&p
);
956 die("incomplete pattern expression: %s", p
->pattern
);
961 if (!opt
->pattern_expression
)
962 opt
->pattern_expression
= header_expr
;
963 else if (opt
->all_match
)
964 opt
->pattern_expression
= grep_splice_or(header_expr
,
965 opt
->pattern_expression
);
967 opt
->pattern_expression
= grep_or_expr(opt
->pattern_expression
,
972 void compile_grep_patterns(struct grep_opt
*opt
)
974 compile_grep_patterns_real(opt
);
976 dump_grep_expression(opt
);
979 static void free_pattern_expr(struct grep_expr
*x
)
986 free_pattern_expr(x
->u
.unary
);
990 free_pattern_expr(x
->u
.binary
.left
);
991 free_pattern_expr(x
->u
.binary
.right
);
997 void free_grep_patterns(struct grep_opt
*opt
)
999 struct grep_pat
*p
, *n
;
1001 for (p
= opt
->pattern_list
; p
; p
= n
) {
1004 case GREP_PATTERN
: /* atom */
1005 case GREP_PATTERN_HEAD
:
1006 case GREP_PATTERN_BODY
:
1009 else if (p
->pcre1_regexp
)
1010 free_pcre1_regexp(p
);
1011 else if (p
->pcre2_pattern
)
1012 free_pcre2_pattern(p
);
1014 regfree(&p
->regexp
);
1025 free_pattern_expr(opt
->pattern_expression
);
1028 static char *end_of_line(char *cp
, unsigned long *left
)
1030 unsigned long l
= *left
;
1031 while (l
&& *cp
!= '\n') {
1039 static int word_char(char ch
)
1041 return isalnum(ch
) || ch
== '_';
1044 static void output_color(struct grep_opt
*opt
, const void *data
, size_t size
,
1047 if (want_color(opt
->color
) && color
&& color
[0]) {
1048 opt
->output(opt
, color
, strlen(color
));
1049 opt
->output(opt
, data
, size
);
1050 opt
->output(opt
, GIT_COLOR_RESET
, strlen(GIT_COLOR_RESET
));
1052 opt
->output(opt
, data
, size
);
1055 static void output_sep(struct grep_opt
*opt
, char sign
)
1057 if (opt
->null_following_name
)
1058 opt
->output(opt
, "\0", 1);
1060 output_color(opt
, &sign
, 1, opt
->color_sep
);
1063 static void show_name(struct grep_opt
*opt
, const char *name
)
1065 output_color(opt
, name
, strlen(name
), opt
->color_filename
);
1066 opt
->output(opt
, opt
->null_following_name
? "\0" : "\n", 1);
1069 static int fixmatch(struct grep_pat
*p
, char *line
, char *eol
,
1072 struct kwsmatch kwsm
;
1073 size_t offset
= kwsexec(p
->kws
, line
, eol
- line
, &kwsm
);
1075 match
->rm_so
= match
->rm_eo
= -1;
1078 match
->rm_so
= offset
;
1079 match
->rm_eo
= match
->rm_so
+ kwsm
.size
[0];
1084 static int patmatch(struct grep_pat
*p
, char *line
, char *eol
,
1085 regmatch_t
*match
, int eflags
)
1090 hit
= !fixmatch(p
, line
, eol
, match
);
1091 else if (p
->pcre1_regexp
)
1092 hit
= !pcre1match(p
, line
, eol
, match
, eflags
);
1093 else if (p
->pcre2_pattern
)
1094 hit
= !pcre2match(p
, line
, eol
, match
, eflags
);
1096 hit
= !regexec_buf(&p
->regexp
, line
, eol
- line
, 1, match
,
1102 static int strip_timestamp(char *bol
, char **eol_p
)
1107 while (bol
< --eol
) {
1121 } header_field
[] = {
1123 { "committer ", 10 },
1127 static int match_one_pattern(struct grep_pat
*p
, char *bol
, char *eol
,
1128 enum grep_context ctx
,
1129 regmatch_t
*pmatch
, int eflags
)
1133 const char *start
= bol
;
1135 if ((p
->token
!= GREP_PATTERN
) &&
1136 ((p
->token
== GREP_PATTERN_HEAD
) != (ctx
== GREP_CONTEXT_HEAD
)))
1139 if (p
->token
== GREP_PATTERN_HEAD
) {
1142 assert(p
->field
< ARRAY_SIZE(header_field
));
1143 field
= header_field
[p
->field
].field
;
1144 len
= header_field
[p
->field
].len
;
1145 if (strncmp(bol
, field
, len
))
1149 case GREP_HEADER_AUTHOR
:
1150 case GREP_HEADER_COMMITTER
:
1151 saved_ch
= strip_timestamp(bol
, &eol
);
1159 hit
= patmatch(p
, bol
, eol
, pmatch
, eflags
);
1161 if (hit
&& p
->word_regexp
) {
1162 if ((pmatch
[0].rm_so
< 0) ||
1163 (eol
- bol
) < pmatch
[0].rm_so
||
1164 (pmatch
[0].rm_eo
< 0) ||
1165 (eol
- bol
) < pmatch
[0].rm_eo
)
1166 die("regexp returned nonsense");
1168 /* Match beginning must be either beginning of the
1169 * line, or at word boundary (i.e. the last char must
1170 * not be a word char). Similarly, match end must be
1171 * either end of the line, or at word boundary
1172 * (i.e. the next char must not be a word char).
1174 if ( ((pmatch
[0].rm_so
== 0) ||
1175 !word_char(bol
[pmatch
[0].rm_so
-1])) &&
1176 ((pmatch
[0].rm_eo
== (eol
-bol
)) ||
1177 !word_char(bol
[pmatch
[0].rm_eo
])) )
1182 /* Words consist of at least one character. */
1183 if (pmatch
->rm_so
== pmatch
->rm_eo
)
1186 if (!hit
&& pmatch
[0].rm_so
+ bol
+ 1 < eol
) {
1187 /* There could be more than one match on the
1188 * line, and the first match might not be
1189 * strict word match. But later ones could be!
1190 * Forward to the next possible start, i.e. the
1191 * next position following a non-word char.
1193 bol
= pmatch
[0].rm_so
+ bol
+ 1;
1194 while (word_char(bol
[-1]) && bol
< eol
)
1196 eflags
|= REG_NOTBOL
;
1201 if (p
->token
== GREP_PATTERN_HEAD
&& saved_ch
)
1204 pmatch
[0].rm_so
+= bol
- start
;
1205 pmatch
[0].rm_eo
+= bol
- start
;
1210 static int match_expr_eval(struct grep_expr
*x
, char *bol
, char *eol
,
1211 enum grep_context ctx
, int collect_hits
)
1217 die("Not a valid grep expression");
1219 case GREP_NODE_TRUE
:
1222 case GREP_NODE_ATOM
:
1223 h
= match_one_pattern(x
->u
.atom
, bol
, eol
, ctx
, &match
, 0);
1226 h
= !match_expr_eval(x
->u
.unary
, bol
, eol
, ctx
, 0);
1229 if (!match_expr_eval(x
->u
.binary
.left
, bol
, eol
, ctx
, 0))
1231 h
= match_expr_eval(x
->u
.binary
.right
, bol
, eol
, ctx
, 0);
1235 return (match_expr_eval(x
->u
.binary
.left
,
1236 bol
, eol
, ctx
, 0) ||
1237 match_expr_eval(x
->u
.binary
.right
,
1239 h
= match_expr_eval(x
->u
.binary
.left
, bol
, eol
, ctx
, 0);
1240 x
->u
.binary
.left
->hit
|= h
;
1241 h
|= match_expr_eval(x
->u
.binary
.right
, bol
, eol
, ctx
, 1);
1244 die("Unexpected node type (internal error) %d", x
->node
);
1251 static int match_expr(struct grep_opt
*opt
, char *bol
, char *eol
,
1252 enum grep_context ctx
, int collect_hits
)
1254 struct grep_expr
*x
= opt
->pattern_expression
;
1255 return match_expr_eval(x
, bol
, eol
, ctx
, collect_hits
);
1258 static int match_line(struct grep_opt
*opt
, char *bol
, char *eol
,
1259 enum grep_context ctx
, int collect_hits
)
1265 return match_expr(opt
, bol
, eol
, ctx
, collect_hits
);
1267 /* we do not call with collect_hits without being extended */
1268 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
1269 if (match_one_pattern(p
, bol
, eol
, ctx
, &match
, 0))
1275 static int match_next_pattern(struct grep_pat
*p
, char *bol
, char *eol
,
1276 enum grep_context ctx
,
1277 regmatch_t
*pmatch
, int eflags
)
1281 if (!match_one_pattern(p
, bol
, eol
, ctx
, &match
, eflags
))
1283 if (match
.rm_so
< 0 || match
.rm_eo
< 0)
1285 if (pmatch
->rm_so
>= 0 && pmatch
->rm_eo
>= 0) {
1286 if (match
.rm_so
> pmatch
->rm_so
)
1288 if (match
.rm_so
== pmatch
->rm_so
&& match
.rm_eo
< pmatch
->rm_eo
)
1291 pmatch
->rm_so
= match
.rm_so
;
1292 pmatch
->rm_eo
= match
.rm_eo
;
1296 static int next_match(struct grep_opt
*opt
, char *bol
, char *eol
,
1297 enum grep_context ctx
, regmatch_t
*pmatch
, int eflags
)
1302 pmatch
->rm_so
= pmatch
->rm_eo
= -1;
1304 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
1306 case GREP_PATTERN
: /* atom */
1307 case GREP_PATTERN_HEAD
:
1308 case GREP_PATTERN_BODY
:
1309 hit
|= match_next_pattern(p
, bol
, eol
, ctx
,
1320 static void show_line(struct grep_opt
*opt
, char *bol
, char *eol
,
1321 const char *name
, unsigned lno
, char sign
)
1323 int rest
= eol
- bol
;
1324 const char *match_color
, *line_color
= NULL
;
1326 if (opt
->file_break
&& opt
->last_shown
== 0) {
1327 if (opt
->show_hunk_mark
)
1328 opt
->output(opt
, "\n", 1);
1329 } else if (opt
->pre_context
|| opt
->post_context
|| opt
->funcbody
) {
1330 if (opt
->last_shown
== 0) {
1331 if (opt
->show_hunk_mark
) {
1332 output_color(opt
, "--", 2, opt
->color_sep
);
1333 opt
->output(opt
, "\n", 1);
1335 } else if (lno
> opt
->last_shown
+ 1) {
1336 output_color(opt
, "--", 2, opt
->color_sep
);
1337 opt
->output(opt
, "\n", 1);
1340 if (opt
->heading
&& opt
->last_shown
== 0) {
1341 output_color(opt
, name
, strlen(name
), opt
->color_filename
);
1342 opt
->output(opt
, "\n", 1);
1344 opt
->last_shown
= lno
;
1346 if (!opt
->heading
&& opt
->pathname
) {
1347 output_color(opt
, name
, strlen(name
), opt
->color_filename
);
1348 output_sep(opt
, sign
);
1352 xsnprintf(buf
, sizeof(buf
), "%d", lno
);
1353 output_color(opt
, buf
, strlen(buf
), opt
->color_lineno
);
1354 output_sep(opt
, sign
);
1358 enum grep_context ctx
= GREP_CONTEXT_BODY
;
1363 match_color
= opt
->color_match_selected
;
1365 match_color
= opt
->color_match_context
;
1367 line_color
= opt
->color_selected
;
1368 else if (sign
== '-')
1369 line_color
= opt
->color_context
;
1370 else if (sign
== '=')
1371 line_color
= opt
->color_function
;
1373 while (next_match(opt
, bol
, eol
, ctx
, &match
, eflags
)) {
1374 if (match
.rm_so
== match
.rm_eo
)
1377 output_color(opt
, bol
, match
.rm_so
, line_color
);
1378 output_color(opt
, bol
+ match
.rm_so
,
1379 match
.rm_eo
- match
.rm_so
, match_color
);
1381 rest
-= match
.rm_eo
;
1382 eflags
= REG_NOTBOL
;
1386 output_color(opt
, bol
, rest
, line_color
);
1387 opt
->output(opt
, "\n", 1);
1394 * This lock protects access to the gitattributes machinery, which is
1397 pthread_mutex_t grep_attr_mutex
;
1399 static inline void grep_attr_lock(void)
1402 pthread_mutex_lock(&grep_attr_mutex
);
1405 static inline void grep_attr_unlock(void)
1408 pthread_mutex_unlock(&grep_attr_mutex
);
1412 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1414 pthread_mutex_t grep_read_mutex
;
1417 #define grep_attr_lock()
1418 #define grep_attr_unlock()
1421 static int match_funcname(struct grep_opt
*opt
, struct grep_source
*gs
, char *bol
, char *eol
)
1423 xdemitconf_t
*xecfg
= opt
->priv
;
1424 if (xecfg
&& !xecfg
->find_func
) {
1425 grep_source_load_driver(gs
);
1426 if (gs
->driver
->funcname
.pattern
) {
1427 const struct userdiff_funcname
*pe
= &gs
->driver
->funcname
;
1428 xdiff_set_find_func(xecfg
, pe
->pattern
, pe
->cflags
);
1430 xecfg
= opt
->priv
= NULL
;
1436 return xecfg
->find_func(bol
, eol
- bol
, buf
, 1,
1437 xecfg
->find_func_priv
) >= 0;
1442 if (isalpha(*bol
) || *bol
== '_' || *bol
== '$')
1447 static void show_funcname_line(struct grep_opt
*opt
, struct grep_source
*gs
,
1448 char *bol
, unsigned lno
)
1450 while (bol
> gs
->buf
) {
1453 while (bol
> gs
->buf
&& bol
[-1] != '\n')
1457 if (lno
<= opt
->last_shown
)
1460 if (match_funcname(opt
, gs
, bol
, eol
)) {
1461 show_line(opt
, bol
, eol
, gs
->name
, lno
, '=');
1467 static void show_pre_context(struct grep_opt
*opt
, struct grep_source
*gs
,
1468 char *bol
, char *end
, unsigned lno
)
1470 unsigned cur
= lno
, from
= 1, funcname_lno
= 0;
1471 int funcname_needed
= !!opt
->funcname
;
1473 if (opt
->funcbody
&& !match_funcname(opt
, gs
, bol
, end
))
1474 funcname_needed
= 2;
1476 if (opt
->pre_context
< lno
)
1477 from
= lno
- opt
->pre_context
;
1478 if (from
<= opt
->last_shown
)
1479 from
= opt
->last_shown
+ 1;
1482 while (bol
> gs
->buf
&&
1483 cur
> (funcname_needed
== 2 ? opt
->last_shown
+ 1 : from
)) {
1486 while (bol
> gs
->buf
&& bol
[-1] != '\n')
1489 if (funcname_needed
&& match_funcname(opt
, gs
, bol
, eol
)) {
1491 funcname_needed
= 0;
1495 /* We need to look even further back to find a function signature. */
1496 if (opt
->funcname
&& funcname_needed
)
1497 show_funcname_line(opt
, gs
, bol
, cur
);
1501 char *eol
= bol
, sign
= (cur
== funcname_lno
) ? '=' : '-';
1503 while (*eol
!= '\n')
1505 show_line(opt
, bol
, eol
, gs
->name
, cur
, sign
);
1511 static int should_lookahead(struct grep_opt
*opt
)
1516 return 0; /* punt for too complex stuff */
1519 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
1520 if (p
->token
!= GREP_PATTERN
)
1521 return 0; /* punt for "header only" and stuff */
1526 static int look_ahead(struct grep_opt
*opt
,
1527 unsigned long *left_p
,
1531 unsigned lno
= *lno_p
;
1534 char *sp
, *last_bol
;
1535 regoff_t earliest
= -1;
1537 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
1541 hit
= patmatch(p
, bol
, bol
+ *left_p
, &m
, 0);
1542 if (!hit
|| m
.rm_so
< 0 || m
.rm_eo
< 0)
1544 if (earliest
< 0 || m
.rm_so
< earliest
)
1549 *bol_p
= bol
+ *left_p
;
1553 for (sp
= bol
+ earliest
; bol
< sp
&& sp
[-1] != '\n'; sp
--)
1554 ; /* find the beginning of the line */
1557 for (sp
= bol
; sp
< last_bol
; sp
++) {
1561 *left_p
-= last_bol
- bol
;
1567 static int fill_textconv_grep(struct userdiff_driver
*driver
,
1568 struct grep_source
*gs
)
1570 struct diff_filespec
*df
;
1574 if (!driver
|| !driver
->textconv
)
1575 return grep_source_load(gs
);
1578 * The textconv interface is intimately tied to diff_filespecs, so we
1579 * have to pretend to be one. If we could unify the grep_source
1580 * and diff_filespec structs, this mess could just go away.
1582 df
= alloc_filespec(gs
->path
);
1584 case GREP_SOURCE_OID
:
1585 fill_filespec(df
, gs
->identifier
, 1, 0100644);
1587 case GREP_SOURCE_FILE
:
1588 fill_filespec(df
, &null_oid
, 0, 0100644);
1591 die("BUG: attempt to textconv something without a path?");
1595 * fill_textconv is not remotely thread-safe; it may load objects
1596 * behind the scenes, and it modifies the global diff tempfile
1600 size
= fill_textconv(driver
, df
, &buf
);
1605 * The normal fill_textconv usage by the diff machinery would just keep
1606 * the textconv'd buf separate from the diff_filespec. But much of the
1607 * grep code passes around a grep_source and assumes that its "buf"
1608 * pointer is the beginning of the thing we are searching. So let's
1609 * install our textconv'd version into the grep_source, taking care not
1610 * to leak any existing buffer.
1612 grep_source_clear_data(gs
);
1619 static int is_empty_line(const char *bol
, const char *eol
)
1621 while (bol
< eol
&& isspace(*bol
))
1626 static int grep_source_1(struct grep_opt
*opt
, struct grep_source
*gs
, int collect_hits
)
1629 char *peek_bol
= NULL
;
1632 unsigned last_hit
= 0;
1633 int binary_match_only
= 0;
1635 int try_lookahead
= 0;
1636 int show_function
= 0;
1637 struct userdiff_driver
*textconv
= NULL
;
1638 enum grep_context ctx
= GREP_CONTEXT_HEAD
;
1642 opt
->output
= std_output
;
1644 if (opt
->pre_context
|| opt
->post_context
|| opt
->file_break
||
1646 /* Show hunk marks, except for the first file. */
1647 if (opt
->last_shown
)
1648 opt
->show_hunk_mark
= 1;
1650 * If we're using threads then we can't easily identify
1651 * the first file. Always put hunk marks in that case
1652 * and skip the very first one later in work_done().
1654 if (opt
->output
!= std_output
)
1655 opt
->show_hunk_mark
= 1;
1657 opt
->last_shown
= 0;
1659 if (opt
->allow_textconv
) {
1660 grep_source_load_driver(gs
);
1662 * We might set up the shared textconv cache data here, which
1663 * is not thread-safe.
1666 textconv
= userdiff_get_textconv(gs
->driver
);
1671 * We know the result of a textconv is text, so we only have to care
1672 * about binary handling if we are not using it.
1675 switch (opt
->binary
) {
1676 case GREP_BINARY_DEFAULT
:
1677 if (grep_source_is_binary(gs
))
1678 binary_match_only
= 1;
1680 case GREP_BINARY_NOMATCH
:
1681 if (grep_source_is_binary(gs
))
1682 return 0; /* Assume unmatch */
1684 case GREP_BINARY_TEXT
:
1687 die("BUG: unknown binary handling mode");
1691 memset(&xecfg
, 0, sizeof(xecfg
));
1694 try_lookahead
= should_lookahead(opt
);
1696 if (fill_textconv_grep(textconv
, gs
) < 0)
1706 * look_ahead() skips quickly to the line that possibly
1707 * has the next hit; don't call it if we need to do
1708 * something more than just skipping the current line
1709 * in response to an unmatch for the current line. E.g.
1710 * inside a post-context window, we will show the current
1711 * line as a context around the previous hit when it
1716 && (show_function
||
1717 lno
<= last_hit
+ opt
->post_context
))
1718 && look_ahead(opt
, &left
, &lno
, &bol
))
1720 eol
= end_of_line(bol
, &left
);
1724 if ((ctx
== GREP_CONTEXT_HEAD
) && (eol
== bol
))
1725 ctx
= GREP_CONTEXT_BODY
;
1727 hit
= match_line(opt
, bol
, eol
, ctx
, collect_hits
);
1733 /* "grep -v -e foo -e bla" should list lines
1734 * that do not have either, so inversion should
1739 if (opt
->unmatch_name_only
) {
1746 if (opt
->status_only
)
1748 if (opt
->name_only
) {
1749 show_name(opt
, gs
->name
);
1754 if (binary_match_only
) {
1755 opt
->output(opt
, "Binary file ", 12);
1756 output_color(opt
, gs
->name
, strlen(gs
->name
),
1757 opt
->color_filename
);
1758 opt
->output(opt
, " matches\n", 9);
1761 /* Hit at this line. If we haven't shown the
1762 * pre-context lines, we would need to show them.
1764 if (opt
->pre_context
|| opt
->funcbody
)
1765 show_pre_context(opt
, gs
, bol
, eol
, lno
);
1766 else if (opt
->funcname
)
1767 show_funcname_line(opt
, gs
, bol
, lno
);
1768 show_line(opt
, bol
, eol
, gs
->name
, lno
, ':');
1774 if (show_function
&& (!peek_bol
|| peek_bol
< bol
)) {
1775 unsigned long peek_left
= left
;
1776 char *peek_eol
= eol
;
1779 * Trailing empty lines are not interesting.
1780 * Peek past them to see if they belong to the
1781 * body of the current function.
1784 while (is_empty_line(peek_bol
, peek_eol
)) {
1785 peek_bol
= peek_eol
+ 1;
1786 peek_eol
= end_of_line(peek_bol
, &peek_left
);
1789 if (match_funcname(opt
, gs
, peek_bol
, peek_eol
))
1792 if (show_function
||
1793 (last_hit
&& lno
<= last_hit
+ opt
->post_context
)) {
1794 /* If the last hit is within the post context,
1795 * we need to show this line.
1797 show_line(opt
, bol
, eol
, gs
->name
, lno
, '-');
1811 if (opt
->status_only
)
1813 if (opt
->unmatch_name_only
) {
1814 /* We did not see any hit, so we want to show this */
1815 show_name(opt
, gs
->name
);
1819 xdiff_clear_find_func(&xecfg
);
1823 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1824 * which feels mostly useless but sometimes useful. Maybe
1825 * make it another option? For now suppress them.
1827 if (opt
->count
&& count
) {
1829 if (opt
->pathname
) {
1830 output_color(opt
, gs
->name
, strlen(gs
->name
),
1831 opt
->color_filename
);
1832 output_sep(opt
, ':');
1834 xsnprintf(buf
, sizeof(buf
), "%u\n", count
);
1835 opt
->output(opt
, buf
, strlen(buf
));
1841 static void clr_hit_marker(struct grep_expr
*x
)
1843 /* All-hit markers are meaningful only at the very top level
1848 if (x
->node
!= GREP_NODE_OR
)
1850 x
->u
.binary
.left
->hit
= 0;
1851 x
= x
->u
.binary
.right
;
1855 static int chk_hit_marker(struct grep_expr
*x
)
1857 /* Top level nodes have hit markers. See if they all are hits */
1859 if (x
->node
!= GREP_NODE_OR
)
1861 if (!x
->u
.binary
.left
->hit
)
1863 x
= x
->u
.binary
.right
;
1867 int grep_source(struct grep_opt
*opt
, struct grep_source
*gs
)
1870 * we do not have to do the two-pass grep when we do not check
1871 * buffer-wide "all-match".
1873 if (!opt
->all_match
)
1874 return grep_source_1(opt
, gs
, 0);
1876 /* Otherwise the toplevel "or" terms hit a bit differently.
1877 * We first clear hit markers from them.
1879 clr_hit_marker(opt
->pattern_expression
);
1880 grep_source_1(opt
, gs
, 1);
1882 if (!chk_hit_marker(opt
->pattern_expression
))
1885 return grep_source_1(opt
, gs
, 0);
1888 int grep_buffer(struct grep_opt
*opt
, char *buf
, unsigned long size
)
1890 struct grep_source gs
;
1893 grep_source_init(&gs
, GREP_SOURCE_BUF
, NULL
, NULL
, NULL
);
1897 r
= grep_source(opt
, &gs
);
1899 grep_source_clear(&gs
);
1903 void grep_source_init(struct grep_source
*gs
, enum grep_source_type type
,
1904 const char *name
, const char *path
,
1905 const void *identifier
)
1908 gs
->name
= xstrdup_or_null(name
);
1909 gs
->path
= xstrdup_or_null(path
);
1915 case GREP_SOURCE_FILE
:
1916 gs
->identifier
= xstrdup(identifier
);
1918 case GREP_SOURCE_SUBMODULE
:
1920 gs
->identifier
= NULL
;
1925 * If the identifier is non-NULL (in the submodule case) it
1926 * will be a SHA1 that needs to be copied.
1928 case GREP_SOURCE_OID
:
1929 gs
->identifier
= oiddup(identifier
);
1931 case GREP_SOURCE_BUF
:
1932 gs
->identifier
= NULL
;
1937 void grep_source_clear(struct grep_source
*gs
)
1939 FREE_AND_NULL(gs
->name
);
1940 FREE_AND_NULL(gs
->path
);
1941 FREE_AND_NULL(gs
->identifier
);
1942 grep_source_clear_data(gs
);
1945 void grep_source_clear_data(struct grep_source
*gs
)
1948 case GREP_SOURCE_FILE
:
1949 case GREP_SOURCE_OID
:
1950 case GREP_SOURCE_SUBMODULE
:
1951 FREE_AND_NULL(gs
->buf
);
1954 case GREP_SOURCE_BUF
:
1955 /* leave user-provided buf intact */
1960 static int grep_source_load_oid(struct grep_source
*gs
)
1962 enum object_type type
;
1965 gs
->buf
= read_sha1_file(gs
->identifier
, &type
, &gs
->size
);
1969 return error(_("'%s': unable to read %s"),
1971 oid_to_hex(gs
->identifier
));
1975 static int grep_source_load_file(struct grep_source
*gs
)
1977 const char *filename
= gs
->identifier
;
1983 if (lstat(filename
, &st
) < 0) {
1985 if (errno
!= ENOENT
)
1986 error_errno(_("failed to stat '%s'"), filename
);
1989 if (!S_ISREG(st
.st_mode
))
1991 size
= xsize_t(st
.st_size
);
1992 i
= open(filename
, O_RDONLY
);
1995 data
= xmallocz(size
);
1996 if (st
.st_size
!= read_in_full(i
, data
, size
)) {
1997 error_errno(_("'%s': short read"), filename
);
2009 static int grep_source_load(struct grep_source
*gs
)
2015 case GREP_SOURCE_FILE
:
2016 return grep_source_load_file(gs
);
2017 case GREP_SOURCE_OID
:
2018 return grep_source_load_oid(gs
);
2019 case GREP_SOURCE_BUF
:
2020 return gs
->buf
? 0 : -1;
2021 case GREP_SOURCE_SUBMODULE
:
2024 die("BUG: invalid grep_source type to load");
2027 void grep_source_load_driver(struct grep_source
*gs
)
2034 gs
->driver
= userdiff_find_by_path(gs
->path
);
2036 gs
->driver
= userdiff_find_by_name("default");
2040 static int grep_source_is_binary(struct grep_source
*gs
)
2042 grep_source_load_driver(gs
);
2043 if (gs
->driver
->binary
!= -1)
2044 return gs
->driver
->binary
;
2046 if (!grep_source_load(gs
))
2047 return buffer_is_binary(gs
->buf
, gs
->size
);