4 #include "xdiff-interface.h"
6 static int grep_source_load(struct grep_source
*gs
);
7 static int grep_source_is_binary(struct grep_source
*gs
);
9 static struct grep_opt grep_defaults
;
12 * Initialize the grep_defaults template with hardcoded defaults.
13 * We could let the compiler do this, but without C99 initializers
14 * the code gets unwieldy and unreadable, so...
16 void init_grep_defaults(void)
18 struct grep_opt
*opt
= &grep_defaults
;
20 memset(opt
, 0, sizeof(*opt
));
23 opt
->regflags
= REG_NEWLINE
;
25 opt
->pattern_type_option
= GREP_PATTERN_TYPE_UNSPECIFIED
;
26 opt
->extended_regexp_option
= 0;
27 strcpy(opt
->color_context
, "");
28 strcpy(opt
->color_filename
, "");
29 strcpy(opt
->color_function
, "");
30 strcpy(opt
->color_lineno
, "");
31 strcpy(opt
->color_match
, GIT_COLOR_BOLD_RED
);
32 strcpy(opt
->color_selected
, "");
33 strcpy(opt
->color_sep
, GIT_COLOR_CYAN
);
37 static int parse_pattern_type_arg(const char *opt
, const char *arg
)
39 if (!strcmp(arg
, "default"))
40 return GREP_PATTERN_TYPE_UNSPECIFIED
;
41 else if (!strcmp(arg
, "basic"))
42 return GREP_PATTERN_TYPE_BRE
;
43 else if (!strcmp(arg
, "extended"))
44 return GREP_PATTERN_TYPE_ERE
;
45 else if (!strcmp(arg
, "fixed"))
46 return GREP_PATTERN_TYPE_FIXED
;
47 else if (!strcmp(arg
, "perl"))
48 return GREP_PATTERN_TYPE_PCRE
;
49 die("bad %s argument: %s", opt
, arg
);
53 * Read the configuration file once and store it in
54 * the grep_defaults template.
56 int grep_config(const char *var
, const char *value
, void *cb
)
58 struct grep_opt
*opt
= &grep_defaults
;
61 if (userdiff_config(var
, value
) < 0)
64 if (!strcmp(var
, "grep.extendedregexp")) {
65 if (git_config_bool(var
, value
))
66 opt
->extended_regexp_option
= 1;
68 opt
->extended_regexp_option
= 0;
72 if (!strcmp(var
, "grep.patterntype")) {
73 opt
->pattern_type_option
= parse_pattern_type_arg(var
, value
);
77 if (!strcmp(var
, "grep.linenumber")) {
78 opt
->linenum
= git_config_bool(var
, value
);
82 if (!strcmp(var
, "color.grep"))
83 opt
->color
= git_config_colorbool(var
, value
);
84 else if (!strcmp(var
, "color.grep.context"))
85 color
= opt
->color_context
;
86 else if (!strcmp(var
, "color.grep.filename"))
87 color
= opt
->color_filename
;
88 else if (!strcmp(var
, "color.grep.function"))
89 color
= opt
->color_function
;
90 else if (!strcmp(var
, "color.grep.linenumber"))
91 color
= opt
->color_lineno
;
92 else if (!strcmp(var
, "color.grep.match"))
93 color
= opt
->color_match
;
94 else if (!strcmp(var
, "color.grep.selected"))
95 color
= opt
->color_selected
;
96 else if (!strcmp(var
, "color.grep.separator"))
97 color
= opt
->color_sep
;
101 return config_error_nonbool(var
);
102 color_parse(value
, var
, color
);
108 * Initialize one instance of grep_opt and copy the
109 * default values from the template we read the configuration
110 * information in an earlier call to git_config(grep_config).
112 void grep_init(struct grep_opt
*opt
, const char *prefix
)
114 struct grep_opt
*def
= &grep_defaults
;
116 memset(opt
, 0, sizeof(*opt
));
117 opt
->prefix
= prefix
;
118 opt
->prefix_length
= (prefix
&& *prefix
) ? strlen(prefix
) : 0;
119 opt
->pattern_tail
= &opt
->pattern_list
;
120 opt
->header_tail
= &opt
->header_list
;
122 opt
->color
= def
->color
;
123 opt
->extended_regexp_option
= def
->extended_regexp_option
;
124 opt
->pattern_type_option
= def
->pattern_type_option
;
125 opt
->linenum
= def
->linenum
;
126 opt
->max_depth
= def
->max_depth
;
127 opt
->pathname
= def
->pathname
;
128 opt
->regflags
= def
->regflags
;
129 opt
->relative
= def
->relative
;
131 strcpy(opt
->color_context
, def
->color_context
);
132 strcpy(opt
->color_filename
, def
->color_filename
);
133 strcpy(opt
->color_function
, def
->color_function
);
134 strcpy(opt
->color_lineno
, def
->color_lineno
);
135 strcpy(opt
->color_match
, def
->color_match
);
136 strcpy(opt
->color_selected
, def
->color_selected
);
137 strcpy(opt
->color_sep
, def
->color_sep
);
140 static struct grep_pat
*create_grep_pat(const char *pat
, size_t patlen
,
141 const char *origin
, int no
,
142 enum grep_pat_token t
,
143 enum grep_header_field field
)
145 struct grep_pat
*p
= xcalloc(1, sizeof(*p
));
146 p
->pattern
= xmemdupz(pat
, patlen
);
147 p
->patternlen
= patlen
;
155 static void do_append_grep_pat(struct grep_pat
***tail
, struct grep_pat
*p
)
162 case GREP_PATTERN
: /* atom */
163 case GREP_PATTERN_HEAD
:
164 case GREP_PATTERN_BODY
:
166 struct grep_pat
*new_pat
;
168 char *cp
= p
->pattern
+ p
->patternlen
, *nl
= NULL
;
169 while (++len
<= p
->patternlen
) {
170 if (*(--cp
) == '\n') {
177 new_pat
= create_grep_pat(nl
+ 1, len
- 1, p
->origin
,
178 p
->no
, p
->token
, p
->field
);
179 new_pat
->next
= p
->next
;
181 *tail
= &new_pat
->next
;
184 p
->patternlen
-= len
;
192 void append_header_grep_pattern(struct grep_opt
*opt
,
193 enum grep_header_field field
, const char *pat
)
195 struct grep_pat
*p
= create_grep_pat(pat
, strlen(pat
), "header", 0,
196 GREP_PATTERN_HEAD
, field
);
197 if (field
== GREP_HEADER_REFLOG
)
198 opt
->use_reflog_filter
= 1;
199 do_append_grep_pat(&opt
->header_tail
, p
);
202 void append_grep_pattern(struct grep_opt
*opt
, const char *pat
,
203 const char *origin
, int no
, enum grep_pat_token t
)
205 append_grep_pat(opt
, pat
, strlen(pat
), origin
, no
, t
);
208 void append_grep_pat(struct grep_opt
*opt
, const char *pat
, size_t patlen
,
209 const char *origin
, int no
, enum grep_pat_token t
)
211 struct grep_pat
*p
= create_grep_pat(pat
, patlen
, origin
, no
, t
, 0);
212 do_append_grep_pat(&opt
->pattern_tail
, p
);
215 struct grep_opt
*grep_opt_dup(const struct grep_opt
*opt
)
217 struct grep_pat
*pat
;
218 struct grep_opt
*ret
= xmalloc(sizeof(struct grep_opt
));
221 ret
->pattern_list
= NULL
;
222 ret
->pattern_tail
= &ret
->pattern_list
;
224 for(pat
= opt
->pattern_list
; pat
!= NULL
; pat
= pat
->next
)
226 if(pat
->token
== GREP_PATTERN_HEAD
)
227 append_header_grep_pattern(ret
, pat
->field
,
230 append_grep_pat(ret
, pat
->pattern
, pat
->patternlen
,
231 pat
->origin
, pat
->no
, pat
->token
);
237 static NORETURN
void compile_regexp_failed(const struct grep_pat
*p
,
243 sprintf(where
, "In '%s' at %d, ", p
->origin
, p
->no
);
245 sprintf(where
, "%s, ", p
->origin
);
249 die("%s'%s': %s", where
, p
->pattern
, error
);
253 static void compile_pcre_regexp(struct grep_pat
*p
, const struct grep_opt
*opt
)
257 int options
= PCRE_MULTILINE
;
259 if (opt
->ignore_case
)
260 options
|= PCRE_CASELESS
;
262 p
->pcre_regexp
= pcre_compile(p
->pattern
, options
, &error
, &erroffset
,
265 compile_regexp_failed(p
, error
);
267 p
->pcre_extra_info
= pcre_study(p
->pcre_regexp
, 0, &error
);
268 if (!p
->pcre_extra_info
&& error
)
272 static int pcrematch(struct grep_pat
*p
, const char *line
, const char *eol
,
273 regmatch_t
*match
, int eflags
)
275 int ovector
[30], ret
, flags
= 0;
277 if (eflags
& REG_NOTBOL
)
278 flags
|= PCRE_NOTBOL
;
280 ret
= pcre_exec(p
->pcre_regexp
, p
->pcre_extra_info
, line
, eol
- line
,
281 0, flags
, ovector
, ARRAY_SIZE(ovector
));
282 if (ret
< 0 && ret
!= PCRE_ERROR_NOMATCH
)
283 die("pcre_exec failed with error code %d", ret
);
286 match
->rm_so
= ovector
[0];
287 match
->rm_eo
= ovector
[1];
293 static void free_pcre_regexp(struct grep_pat
*p
)
295 pcre_free(p
->pcre_regexp
);
296 pcre_free(p
->pcre_extra_info
);
298 #else /* !USE_LIBPCRE */
299 static void compile_pcre_regexp(struct grep_pat
*p
, const struct grep_opt
*opt
)
301 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
304 static int pcrematch(struct grep_pat
*p
, const char *line
, const char *eol
,
305 regmatch_t
*match
, int eflags
)
310 static void free_pcre_regexp(struct grep_pat
*p
)
313 #endif /* !USE_LIBPCRE */
315 static int is_fixed(const char *s
, size_t len
)
319 /* regcomp cannot accept patterns with NULs so we
320 * consider any pattern containing a NUL fixed.
322 if (memchr(s
, 0, len
))
325 for (i
= 0; i
< len
; i
++) {
326 if (is_regex_special(s
[i
]))
333 static void compile_regexp(struct grep_pat
*p
, struct grep_opt
*opt
)
337 p
->word_regexp
= opt
->word_regexp
;
338 p
->ignore_case
= opt
->ignore_case
;
340 if (opt
->fixed
|| is_fixed(p
->pattern
, p
->patternlen
))
346 if (opt
->regflags
& REG_ICASE
|| p
->ignore_case
)
347 p
->kws
= kwsalloc(tolower_trans_tbl
);
349 p
->kws
= kwsalloc(NULL
);
350 kwsincr(p
->kws
, p
->pattern
, p
->patternlen
);
356 compile_pcre_regexp(p
, opt
);
360 err
= regcomp(&p
->regexp
, p
->pattern
, opt
->regflags
);
363 regerror(err
, &p
->regexp
, errbuf
, 1024);
365 compile_regexp_failed(p
, errbuf
);
369 static struct grep_expr
*compile_pattern_or(struct grep_pat
**);
370 static struct grep_expr
*compile_pattern_atom(struct grep_pat
**list
)
379 case GREP_PATTERN
: /* atom */
380 case GREP_PATTERN_HEAD
:
381 case GREP_PATTERN_BODY
:
382 x
= xcalloc(1, sizeof (struct grep_expr
));
383 x
->node
= GREP_NODE_ATOM
;
387 case GREP_OPEN_PAREN
:
389 x
= compile_pattern_or(list
);
390 if (!*list
|| (*list
)->token
!= GREP_CLOSE_PAREN
)
391 die("unmatched parenthesis");
392 *list
= (*list
)->next
;
399 static struct grep_expr
*compile_pattern_not(struct grep_pat
**list
)
410 die("--not not followed by pattern expression");
412 x
= xcalloc(1, sizeof (struct grep_expr
));
413 x
->node
= GREP_NODE_NOT
;
414 x
->u
.unary
= compile_pattern_not(list
);
416 die("--not followed by non pattern expression");
419 return compile_pattern_atom(list
);
423 static struct grep_expr
*compile_pattern_and(struct grep_pat
**list
)
426 struct grep_expr
*x
, *y
, *z
;
428 x
= compile_pattern_not(list
);
430 if (p
&& p
->token
== GREP_AND
) {
432 die("--and not followed by pattern expression");
434 y
= compile_pattern_and(list
);
436 die("--and not followed by pattern expression");
437 z
= xcalloc(1, sizeof (struct grep_expr
));
438 z
->node
= GREP_NODE_AND
;
439 z
->u
.binary
.left
= x
;
440 z
->u
.binary
.right
= y
;
446 static struct grep_expr
*compile_pattern_or(struct grep_pat
**list
)
449 struct grep_expr
*x
, *y
, *z
;
451 x
= compile_pattern_and(list
);
453 if (x
&& p
&& p
->token
!= GREP_CLOSE_PAREN
) {
454 y
= compile_pattern_or(list
);
456 die("not a pattern expression %s", p
->pattern
);
457 z
= xcalloc(1, sizeof (struct grep_expr
));
458 z
->node
= GREP_NODE_OR
;
459 z
->u
.binary
.left
= x
;
460 z
->u
.binary
.right
= y
;
466 static struct grep_expr
*compile_pattern_expr(struct grep_pat
**list
)
468 return compile_pattern_or(list
);
471 static void indent(int in
)
477 static void dump_grep_pat(struct grep_pat
*p
)
480 case GREP_AND
: fprintf(stderr
, "*and*"); break;
481 case GREP_OPEN_PAREN
: fprintf(stderr
, "*(*"); break;
482 case GREP_CLOSE_PAREN
: fprintf(stderr
, "*)*"); break;
483 case GREP_NOT
: fprintf(stderr
, "*not*"); break;
484 case GREP_OR
: fprintf(stderr
, "*or*"); break;
486 case GREP_PATTERN
: fprintf(stderr
, "pattern"); break;
487 case GREP_PATTERN_HEAD
: fprintf(stderr
, "pattern_head"); break;
488 case GREP_PATTERN_BODY
: fprintf(stderr
, "pattern_body"); break;
493 case GREP_PATTERN_HEAD
:
494 fprintf(stderr
, "<head %d>", p
->field
); break;
495 case GREP_PATTERN_BODY
:
496 fprintf(stderr
, "<body>"); break;
500 case GREP_PATTERN_HEAD
:
501 case GREP_PATTERN_BODY
:
503 fprintf(stderr
, "%.*s", (int)p
->patternlen
, p
->pattern
);
509 static void dump_grep_expression_1(struct grep_expr
*x
, int in
)
514 fprintf(stderr
, "true\n");
517 dump_grep_pat(x
->u
.atom
);
520 fprintf(stderr
, "(not\n");
521 dump_grep_expression_1(x
->u
.unary
, in
+1);
523 fprintf(stderr
, ")\n");
526 fprintf(stderr
, "(and\n");
527 dump_grep_expression_1(x
->u
.binary
.left
, in
+1);
528 dump_grep_expression_1(x
->u
.binary
.right
, in
+1);
530 fprintf(stderr
, ")\n");
533 fprintf(stderr
, "(or\n");
534 dump_grep_expression_1(x
->u
.binary
.left
, in
+1);
535 dump_grep_expression_1(x
->u
.binary
.right
, in
+1);
537 fprintf(stderr
, ")\n");
542 static void dump_grep_expression(struct grep_opt
*opt
)
544 struct grep_expr
*x
= opt
->pattern_expression
;
547 fprintf(stderr
, "[all-match]\n");
548 dump_grep_expression_1(x
, 0);
552 static struct grep_expr
*grep_true_expr(void)
554 struct grep_expr
*z
= xcalloc(1, sizeof(*z
));
555 z
->node
= GREP_NODE_TRUE
;
559 static struct grep_expr
*grep_or_expr(struct grep_expr
*left
, struct grep_expr
*right
)
561 struct grep_expr
*z
= xcalloc(1, sizeof(*z
));
562 z
->node
= GREP_NODE_OR
;
563 z
->u
.binary
.left
= left
;
564 z
->u
.binary
.right
= right
;
568 static struct grep_expr
*prep_header_patterns(struct grep_opt
*opt
)
571 struct grep_expr
*header_expr
;
572 struct grep_expr
*(header_group
[GREP_HEADER_FIELD_MAX
]);
573 enum grep_header_field fld
;
575 if (!opt
->header_list
)
578 for (p
= opt
->header_list
; p
; p
= p
->next
) {
579 if (p
->token
!= GREP_PATTERN_HEAD
)
580 die("bug: a non-header pattern in grep header list.");
581 if (p
->field
< 0 || GREP_HEADER_FIELD_MAX
<= p
->field
)
582 die("bug: unknown header field %d", p
->field
);
583 compile_regexp(p
, opt
);
586 for (fld
= 0; fld
< GREP_HEADER_FIELD_MAX
; fld
++)
587 header_group
[fld
] = NULL
;
589 for (p
= opt
->header_list
; p
; p
= p
->next
) {
591 struct grep_pat
*pp
= p
;
593 h
= compile_pattern_atom(&pp
);
594 if (!h
|| pp
!= p
->next
)
595 die("bug: malformed header expr");
596 if (!header_group
[p
->field
]) {
597 header_group
[p
->field
] = h
;
600 header_group
[p
->field
] = grep_or_expr(h
, header_group
[p
->field
]);
605 for (fld
= 0; fld
< GREP_HEADER_FIELD_MAX
; fld
++) {
606 if (!header_group
[fld
])
609 header_expr
= grep_true_expr();
610 header_expr
= grep_or_expr(header_group
[fld
], header_expr
);
615 static struct grep_expr
*grep_splice_or(struct grep_expr
*x
, struct grep_expr
*y
)
617 struct grep_expr
*z
= x
;
620 assert(x
->node
== GREP_NODE_OR
);
621 if (x
->u
.binary
.right
&&
622 x
->u
.binary
.right
->node
== GREP_NODE_TRUE
) {
623 x
->u
.binary
.right
= y
;
626 x
= x
->u
.binary
.right
;
631 static void compile_grep_patterns_real(struct grep_opt
*opt
)
634 struct grep_expr
*header_expr
= prep_header_patterns(opt
);
636 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
638 case GREP_PATTERN
: /* atom */
639 case GREP_PATTERN_HEAD
:
640 case GREP_PATTERN_BODY
:
641 compile_regexp(p
, opt
);
649 if (opt
->all_match
|| header_expr
)
651 else if (!opt
->extended
&& !opt
->debug
)
654 p
= opt
->pattern_list
;
656 opt
->pattern_expression
= compile_pattern_expr(&p
);
658 die("incomplete pattern expression: %s", p
->pattern
);
663 if (!opt
->pattern_expression
)
664 opt
->pattern_expression
= header_expr
;
665 else if (opt
->all_match
)
666 opt
->pattern_expression
= grep_splice_or(header_expr
,
667 opt
->pattern_expression
);
669 opt
->pattern_expression
= grep_or_expr(opt
->pattern_expression
,
674 void compile_grep_patterns(struct grep_opt
*opt
)
676 compile_grep_patterns_real(opt
);
678 dump_grep_expression(opt
);
681 static void free_pattern_expr(struct grep_expr
*x
)
688 free_pattern_expr(x
->u
.unary
);
692 free_pattern_expr(x
->u
.binary
.left
);
693 free_pattern_expr(x
->u
.binary
.right
);
699 void free_grep_patterns(struct grep_opt
*opt
)
701 struct grep_pat
*p
, *n
;
703 for (p
= opt
->pattern_list
; p
; p
= n
) {
706 case GREP_PATTERN
: /* atom */
707 case GREP_PATTERN_HEAD
:
708 case GREP_PATTERN_BODY
:
711 else if (p
->pcre_regexp
)
725 free_pattern_expr(opt
->pattern_expression
);
728 static char *end_of_line(char *cp
, unsigned long *left
)
730 unsigned long l
= *left
;
731 while (l
&& *cp
!= '\n') {
739 static int word_char(char ch
)
741 return isalnum(ch
) || ch
== '_';
744 static void output_color(struct grep_opt
*opt
, const void *data
, size_t size
,
747 if (want_color(opt
->color
) && color
&& color
[0]) {
748 opt
->output(opt
, color
, strlen(color
));
749 opt
->output(opt
, data
, size
);
750 opt
->output(opt
, GIT_COLOR_RESET
, strlen(GIT_COLOR_RESET
));
752 opt
->output(opt
, data
, size
);
755 static void output_sep(struct grep_opt
*opt
, char sign
)
757 if (opt
->null_following_name
)
758 opt
->output(opt
, "\0", 1);
760 output_color(opt
, &sign
, 1, opt
->color_sep
);
763 static void show_name(struct grep_opt
*opt
, const char *name
)
765 output_color(opt
, name
, strlen(name
), opt
->color_filename
);
766 opt
->output(opt
, opt
->null_following_name
? "\0" : "\n", 1);
769 static int fixmatch(struct grep_pat
*p
, char *line
, char *eol
,
772 struct kwsmatch kwsm
;
773 size_t offset
= kwsexec(p
->kws
, line
, eol
- line
, &kwsm
);
775 match
->rm_so
= match
->rm_eo
= -1;
778 match
->rm_so
= offset
;
779 match
->rm_eo
= match
->rm_so
+ kwsm
.size
[0];
784 static int regmatch(const regex_t
*preg
, char *line
, char *eol
,
785 regmatch_t
*match
, int eflags
)
789 match
->rm_eo
= eol
- line
;
790 eflags
|= REG_STARTEND
;
792 return regexec(preg
, line
, 1, match
, eflags
);
795 static int patmatch(struct grep_pat
*p
, char *line
, char *eol
,
796 regmatch_t
*match
, int eflags
)
801 hit
= !fixmatch(p
, line
, eol
, match
);
802 else if (p
->pcre_regexp
)
803 hit
= !pcrematch(p
, line
, eol
, match
, eflags
);
805 hit
= !regmatch(&p
->regexp
, line
, eol
, match
, eflags
);
810 static int strip_timestamp(char *bol
, char **eol_p
)
815 while (bol
< --eol
) {
831 { "committer ", 10 },
835 static int match_one_pattern(struct grep_pat
*p
, char *bol
, char *eol
,
836 enum grep_context ctx
,
837 regmatch_t
*pmatch
, int eflags
)
841 const char *start
= bol
;
843 if ((p
->token
!= GREP_PATTERN
) &&
844 ((p
->token
== GREP_PATTERN_HEAD
) != (ctx
== GREP_CONTEXT_HEAD
)))
847 if (p
->token
== GREP_PATTERN_HEAD
) {
850 assert(p
->field
< ARRAY_SIZE(header_field
));
851 field
= header_field
[p
->field
].field
;
852 len
= header_field
[p
->field
].len
;
853 if (strncmp(bol
, field
, len
))
857 case GREP_HEADER_AUTHOR
:
858 case GREP_HEADER_COMMITTER
:
859 saved_ch
= strip_timestamp(bol
, &eol
);
867 hit
= patmatch(p
, bol
, eol
, pmatch
, eflags
);
869 if (hit
&& p
->word_regexp
) {
870 if ((pmatch
[0].rm_so
< 0) ||
871 (eol
- bol
) < pmatch
[0].rm_so
||
872 (pmatch
[0].rm_eo
< 0) ||
873 (eol
- bol
) < pmatch
[0].rm_eo
)
874 die("regexp returned nonsense");
876 /* Match beginning must be either beginning of the
877 * line, or at word boundary (i.e. the last char must
878 * not be a word char). Similarly, match end must be
879 * either end of the line, or at word boundary
880 * (i.e. the next char must not be a word char).
882 if ( ((pmatch
[0].rm_so
== 0) ||
883 !word_char(bol
[pmatch
[0].rm_so
-1])) &&
884 ((pmatch
[0].rm_eo
== (eol
-bol
)) ||
885 !word_char(bol
[pmatch
[0].rm_eo
])) )
890 /* Words consist of at least one character. */
891 if (pmatch
->rm_so
== pmatch
->rm_eo
)
894 if (!hit
&& pmatch
[0].rm_so
+ bol
+ 1 < eol
) {
895 /* There could be more than one match on the
896 * line, and the first match might not be
897 * strict word match. But later ones could be!
898 * Forward to the next possible start, i.e. the
899 * next position following a non-word char.
901 bol
= pmatch
[0].rm_so
+ bol
+ 1;
902 while (word_char(bol
[-1]) && bol
< eol
)
904 eflags
|= REG_NOTBOL
;
909 if (p
->token
== GREP_PATTERN_HEAD
&& saved_ch
)
912 pmatch
[0].rm_so
+= bol
- start
;
913 pmatch
[0].rm_eo
+= bol
- start
;
918 static int match_expr_eval(struct grep_expr
*x
, char *bol
, char *eol
,
919 enum grep_context ctx
, int collect_hits
)
925 die("Not a valid grep expression");
931 h
= match_one_pattern(x
->u
.atom
, bol
, eol
, ctx
, &match
, 0);
934 h
= !match_expr_eval(x
->u
.unary
, bol
, eol
, ctx
, 0);
937 if (!match_expr_eval(x
->u
.binary
.left
, bol
, eol
, ctx
, 0))
939 h
= match_expr_eval(x
->u
.binary
.right
, bol
, eol
, ctx
, 0);
943 return (match_expr_eval(x
->u
.binary
.left
,
945 match_expr_eval(x
->u
.binary
.right
,
947 h
= match_expr_eval(x
->u
.binary
.left
, bol
, eol
, ctx
, 0);
948 x
->u
.binary
.left
->hit
|= h
;
949 h
|= match_expr_eval(x
->u
.binary
.right
, bol
, eol
, ctx
, 1);
952 die("Unexpected node type (internal error) %d", x
->node
);
959 static int match_expr(struct grep_opt
*opt
, char *bol
, char *eol
,
960 enum grep_context ctx
, int collect_hits
)
962 struct grep_expr
*x
= opt
->pattern_expression
;
963 return match_expr_eval(x
, bol
, eol
, ctx
, collect_hits
);
966 static int match_line(struct grep_opt
*opt
, char *bol
, char *eol
,
967 enum grep_context ctx
, int collect_hits
)
973 return match_expr(opt
, bol
, eol
, ctx
, collect_hits
);
975 /* we do not call with collect_hits without being extended */
976 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
977 if (match_one_pattern(p
, bol
, eol
, ctx
, &match
, 0))
983 static int match_next_pattern(struct grep_pat
*p
, char *bol
, char *eol
,
984 enum grep_context ctx
,
985 regmatch_t
*pmatch
, int eflags
)
989 if (!match_one_pattern(p
, bol
, eol
, ctx
, &match
, eflags
))
991 if (match
.rm_so
< 0 || match
.rm_eo
< 0)
993 if (pmatch
->rm_so
>= 0 && pmatch
->rm_eo
>= 0) {
994 if (match
.rm_so
> pmatch
->rm_so
)
996 if (match
.rm_so
== pmatch
->rm_so
&& match
.rm_eo
< pmatch
->rm_eo
)
999 pmatch
->rm_so
= match
.rm_so
;
1000 pmatch
->rm_eo
= match
.rm_eo
;
1004 static int next_match(struct grep_opt
*opt
, char *bol
, char *eol
,
1005 enum grep_context ctx
, regmatch_t
*pmatch
, int eflags
)
1010 pmatch
->rm_so
= pmatch
->rm_eo
= -1;
1012 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
1014 case GREP_PATTERN
: /* atom */
1015 case GREP_PATTERN_HEAD
:
1016 case GREP_PATTERN_BODY
:
1017 hit
|= match_next_pattern(p
, bol
, eol
, ctx
,
1028 static void show_line(struct grep_opt
*opt
, char *bol
, char *eol
,
1029 const char *name
, unsigned lno
, char sign
)
1031 int rest
= eol
- bol
;
1032 char *line_color
= NULL
;
1034 if (opt
->file_break
&& opt
->last_shown
== 0) {
1035 if (opt
->show_hunk_mark
)
1036 opt
->output(opt
, "\n", 1);
1037 } else if (opt
->pre_context
|| opt
->post_context
|| opt
->funcbody
) {
1038 if (opt
->last_shown
== 0) {
1039 if (opt
->show_hunk_mark
) {
1040 output_color(opt
, "--", 2, opt
->color_sep
);
1041 opt
->output(opt
, "\n", 1);
1043 } else if (lno
> opt
->last_shown
+ 1) {
1044 output_color(opt
, "--", 2, opt
->color_sep
);
1045 opt
->output(opt
, "\n", 1);
1048 if (opt
->heading
&& opt
->last_shown
== 0) {
1049 output_color(opt
, name
, strlen(name
), opt
->color_filename
);
1050 opt
->output(opt
, "\n", 1);
1052 opt
->last_shown
= lno
;
1054 if (!opt
->heading
&& opt
->pathname
) {
1055 output_color(opt
, name
, strlen(name
), opt
->color_filename
);
1056 output_sep(opt
, sign
);
1060 snprintf(buf
, sizeof(buf
), "%d", lno
);
1061 output_color(opt
, buf
, strlen(buf
), opt
->color_lineno
);
1062 output_sep(opt
, sign
);
1066 enum grep_context ctx
= GREP_CONTEXT_BODY
;
1071 line_color
= opt
->color_selected
;
1072 else if (sign
== '-')
1073 line_color
= opt
->color_context
;
1074 else if (sign
== '=')
1075 line_color
= opt
->color_function
;
1077 while (next_match(opt
, bol
, eol
, ctx
, &match
, eflags
)) {
1078 if (match
.rm_so
== match
.rm_eo
)
1081 output_color(opt
, bol
, match
.rm_so
, line_color
);
1082 output_color(opt
, bol
+ match
.rm_so
,
1083 match
.rm_eo
- match
.rm_so
,
1086 rest
-= match
.rm_eo
;
1087 eflags
= REG_NOTBOL
;
1091 output_color(opt
, bol
, rest
, line_color
);
1092 opt
->output(opt
, "\n", 1);
1099 * This lock protects access to the gitattributes machinery, which is
1102 pthread_mutex_t grep_attr_mutex
;
1104 static inline void grep_attr_lock(void)
1107 pthread_mutex_lock(&grep_attr_mutex
);
1110 static inline void grep_attr_unlock(void)
1113 pthread_mutex_unlock(&grep_attr_mutex
);
1117 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1119 pthread_mutex_t grep_read_mutex
;
1122 #define grep_attr_lock()
1123 #define grep_attr_unlock()
1126 static int match_funcname(struct grep_opt
*opt
, struct grep_source
*gs
, char *bol
, char *eol
)
1128 xdemitconf_t
*xecfg
= opt
->priv
;
1129 if (xecfg
&& !xecfg
->find_func
) {
1130 grep_source_load_driver(gs
);
1131 if (gs
->driver
->funcname
.pattern
) {
1132 const struct userdiff_funcname
*pe
= &gs
->driver
->funcname
;
1133 xdiff_set_find_func(xecfg
, pe
->pattern
, pe
->cflags
);
1135 xecfg
= opt
->priv
= NULL
;
1141 return xecfg
->find_func(bol
, eol
- bol
, buf
, 1,
1142 xecfg
->find_func_priv
) >= 0;
1147 if (isalpha(*bol
) || *bol
== '_' || *bol
== '$')
1152 static void show_funcname_line(struct grep_opt
*opt
, struct grep_source
*gs
,
1153 char *bol
, unsigned lno
)
1155 while (bol
> gs
->buf
) {
1158 while (bol
> gs
->buf
&& bol
[-1] != '\n')
1162 if (lno
<= opt
->last_shown
)
1165 if (match_funcname(opt
, gs
, bol
, eol
)) {
1166 show_line(opt
, bol
, eol
, gs
->name
, lno
, '=');
1172 static void show_pre_context(struct grep_opt
*opt
, struct grep_source
*gs
,
1173 char *bol
, char *end
, unsigned lno
)
1175 unsigned cur
= lno
, from
= 1, funcname_lno
= 0;
1176 int funcname_needed
= !!opt
->funcname
;
1178 if (opt
->funcbody
&& !match_funcname(opt
, gs
, bol
, end
))
1179 funcname_needed
= 2;
1181 if (opt
->pre_context
< lno
)
1182 from
= lno
- opt
->pre_context
;
1183 if (from
<= opt
->last_shown
)
1184 from
= opt
->last_shown
+ 1;
1187 while (bol
> gs
->buf
&&
1188 cur
> (funcname_needed
== 2 ? opt
->last_shown
+ 1 : from
)) {
1191 while (bol
> gs
->buf
&& bol
[-1] != '\n')
1194 if (funcname_needed
&& match_funcname(opt
, gs
, bol
, eol
)) {
1196 funcname_needed
= 0;
1200 /* We need to look even further back to find a function signature. */
1201 if (opt
->funcname
&& funcname_needed
)
1202 show_funcname_line(opt
, gs
, bol
, cur
);
1206 char *eol
= bol
, sign
= (cur
== funcname_lno
) ? '=' : '-';
1208 while (*eol
!= '\n')
1210 show_line(opt
, bol
, eol
, gs
->name
, cur
, sign
);
1216 static int should_lookahead(struct grep_opt
*opt
)
1221 return 0; /* punt for too complex stuff */
1224 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
1225 if (p
->token
!= GREP_PATTERN
)
1226 return 0; /* punt for "header only" and stuff */
1231 static int look_ahead(struct grep_opt
*opt
,
1232 unsigned long *left_p
,
1236 unsigned lno
= *lno_p
;
1239 char *sp
, *last_bol
;
1240 regoff_t earliest
= -1;
1242 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
1246 hit
= patmatch(p
, bol
, bol
+ *left_p
, &m
, 0);
1247 if (!hit
|| m
.rm_so
< 0 || m
.rm_eo
< 0)
1249 if (earliest
< 0 || m
.rm_so
< earliest
)
1254 *bol_p
= bol
+ *left_p
;
1258 for (sp
= bol
+ earliest
; bol
< sp
&& sp
[-1] != '\n'; sp
--)
1259 ; /* find the beginning of the line */
1262 for (sp
= bol
; sp
< last_bol
; sp
++) {
1266 *left_p
-= last_bol
- bol
;
1272 static void std_output(struct grep_opt
*opt
, const void *buf
, size_t size
)
1274 fwrite(buf
, size
, 1, stdout
);
1277 static int grep_source_1(struct grep_opt
*opt
, struct grep_source
*gs
, int collect_hits
)
1282 unsigned last_hit
= 0;
1283 int binary_match_only
= 0;
1285 int try_lookahead
= 0;
1286 int show_function
= 0;
1287 enum grep_context ctx
= GREP_CONTEXT_HEAD
;
1291 opt
->output
= std_output
;
1293 if (opt
->pre_context
|| opt
->post_context
|| opt
->file_break
||
1295 /* Show hunk marks, except for the first file. */
1296 if (opt
->last_shown
)
1297 opt
->show_hunk_mark
= 1;
1299 * If we're using threads then we can't easily identify
1300 * the first file. Always put hunk marks in that case
1301 * and skip the very first one later in work_done().
1303 if (opt
->output
!= std_output
)
1304 opt
->show_hunk_mark
= 1;
1306 opt
->last_shown
= 0;
1308 switch (opt
->binary
) {
1309 case GREP_BINARY_DEFAULT
:
1310 if (grep_source_is_binary(gs
))
1311 binary_match_only
= 1;
1313 case GREP_BINARY_NOMATCH
:
1314 if (grep_source_is_binary(gs
))
1315 return 0; /* Assume unmatch */
1317 case GREP_BINARY_TEXT
:
1320 die("bug: unknown binary handling mode");
1323 memset(&xecfg
, 0, sizeof(xecfg
));
1326 try_lookahead
= should_lookahead(opt
);
1328 if (grep_source_load(gs
) < 0)
1338 * look_ahead() skips quickly to the line that possibly
1339 * has the next hit; don't call it if we need to do
1340 * something more than just skipping the current line
1341 * in response to an unmatch for the current line. E.g.
1342 * inside a post-context window, we will show the current
1343 * line as a context around the previous hit when it
1348 && (show_function
||
1349 lno
<= last_hit
+ opt
->post_context
))
1350 && look_ahead(opt
, &left
, &lno
, &bol
))
1352 eol
= end_of_line(bol
, &left
);
1356 if ((ctx
== GREP_CONTEXT_HEAD
) && (eol
== bol
))
1357 ctx
= GREP_CONTEXT_BODY
;
1359 hit
= match_line(opt
, bol
, eol
, ctx
, collect_hits
);
1365 /* "grep -v -e foo -e bla" should list lines
1366 * that do not have either, so inversion should
1371 if (opt
->unmatch_name_only
) {
1378 if (opt
->status_only
)
1380 if (opt
->name_only
) {
1381 show_name(opt
, gs
->name
);
1386 if (binary_match_only
) {
1387 opt
->output(opt
, "Binary file ", 12);
1388 output_color(opt
, gs
->name
, strlen(gs
->name
),
1389 opt
->color_filename
);
1390 opt
->output(opt
, " matches\n", 9);
1393 /* Hit at this line. If we haven't shown the
1394 * pre-context lines, we would need to show them.
1396 if (opt
->pre_context
|| opt
->funcbody
)
1397 show_pre_context(opt
, gs
, bol
, eol
, lno
);
1398 else if (opt
->funcname
)
1399 show_funcname_line(opt
, gs
, bol
, lno
);
1400 show_line(opt
, bol
, eol
, gs
->name
, lno
, ':');
1406 if (show_function
&& match_funcname(opt
, gs
, bol
, eol
))
1408 if (show_function
||
1409 (last_hit
&& lno
<= last_hit
+ opt
->post_context
)) {
1410 /* If the last hit is within the post context,
1411 * we need to show this line.
1413 show_line(opt
, bol
, eol
, gs
->name
, lno
, '-');
1427 if (opt
->status_only
)
1429 if (opt
->unmatch_name_only
) {
1430 /* We did not see any hit, so we want to show this */
1431 show_name(opt
, gs
->name
);
1435 xdiff_clear_find_func(&xecfg
);
1439 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1440 * which feels mostly useless but sometimes useful. Maybe
1441 * make it another option? For now suppress them.
1443 if (opt
->count
&& count
) {
1445 output_color(opt
, gs
->name
, strlen(gs
->name
), opt
->color_filename
);
1446 output_sep(opt
, ':');
1447 snprintf(buf
, sizeof(buf
), "%u\n", count
);
1448 opt
->output(opt
, buf
, strlen(buf
));
1454 static void clr_hit_marker(struct grep_expr
*x
)
1456 /* All-hit markers are meaningful only at the very top level
1461 if (x
->node
!= GREP_NODE_OR
)
1463 x
->u
.binary
.left
->hit
= 0;
1464 x
= x
->u
.binary
.right
;
1468 static int chk_hit_marker(struct grep_expr
*x
)
1470 /* Top level nodes have hit markers. See if they all are hits */
1472 if (x
->node
!= GREP_NODE_OR
)
1474 if (!x
->u
.binary
.left
->hit
)
1476 x
= x
->u
.binary
.right
;
1480 int grep_source(struct grep_opt
*opt
, struct grep_source
*gs
)
1483 * we do not have to do the two-pass grep when we do not check
1484 * buffer-wide "all-match".
1486 if (!opt
->all_match
)
1487 return grep_source_1(opt
, gs
, 0);
1489 /* Otherwise the toplevel "or" terms hit a bit differently.
1490 * We first clear hit markers from them.
1492 clr_hit_marker(opt
->pattern_expression
);
1493 grep_source_1(opt
, gs
, 1);
1495 if (!chk_hit_marker(opt
->pattern_expression
))
1498 return grep_source_1(opt
, gs
, 0);
1501 int grep_buffer(struct grep_opt
*opt
, char *buf
, unsigned long size
)
1503 struct grep_source gs
;
1506 grep_source_init(&gs
, GREP_SOURCE_BUF
, NULL
, NULL
);
1510 r
= grep_source(opt
, &gs
);
1512 grep_source_clear(&gs
);
1516 void grep_source_init(struct grep_source
*gs
, enum grep_source_type type
,
1517 const char *name
, const void *identifier
)
1520 gs
->name
= name
? xstrdup(name
) : NULL
;
1526 case GREP_SOURCE_FILE
:
1527 gs
->identifier
= xstrdup(identifier
);
1529 case GREP_SOURCE_SHA1
:
1530 gs
->identifier
= xmalloc(20);
1531 memcpy(gs
->identifier
, identifier
, 20);
1533 case GREP_SOURCE_BUF
:
1534 gs
->identifier
= NULL
;
1538 void grep_source_clear(struct grep_source
*gs
)
1542 free(gs
->identifier
);
1543 gs
->identifier
= NULL
;
1544 grep_source_clear_data(gs
);
1547 void grep_source_clear_data(struct grep_source
*gs
)
1550 case GREP_SOURCE_FILE
:
1551 case GREP_SOURCE_SHA1
:
1556 case GREP_SOURCE_BUF
:
1557 /* leave user-provided buf intact */
1562 static int grep_source_load_sha1(struct grep_source
*gs
)
1564 enum object_type type
;
1567 gs
->buf
= read_sha1_file(gs
->identifier
, &type
, &gs
->size
);
1571 return error(_("'%s': unable to read %s"),
1573 sha1_to_hex(gs
->identifier
));
1577 static int grep_source_load_file(struct grep_source
*gs
)
1579 const char *filename
= gs
->identifier
;
1585 if (lstat(filename
, &st
) < 0) {
1587 if (errno
!= ENOENT
)
1588 error(_("'%s': %s"), filename
, strerror(errno
));
1591 if (!S_ISREG(st
.st_mode
))
1593 size
= xsize_t(st
.st_size
);
1594 i
= open(filename
, O_RDONLY
);
1597 data
= xmalloc(size
+ 1);
1598 if (st
.st_size
!= read_in_full(i
, data
, size
)) {
1599 error(_("'%s': short read %s"), filename
, strerror(errno
));
1612 static int grep_source_load(struct grep_source
*gs
)
1618 case GREP_SOURCE_FILE
:
1619 return grep_source_load_file(gs
);
1620 case GREP_SOURCE_SHA1
:
1621 return grep_source_load_sha1(gs
);
1622 case GREP_SOURCE_BUF
:
1623 return gs
->buf
? 0 : -1;
1625 die("BUG: invalid grep_source type");
1628 void grep_source_load_driver(struct grep_source
*gs
)
1634 gs
->driver
= userdiff_find_by_path(gs
->name
);
1636 gs
->driver
= userdiff_find_by_name("default");
1640 static int grep_source_is_binary(struct grep_source
*gs
)
1642 grep_source_load_driver(gs
);
1643 if (gs
->driver
->binary
!= -1)
1644 return gs
->driver
->binary
;
1646 if (!grep_source_load(gs
))
1647 return buffer_is_binary(gs
->buf
, gs
->size
);