Merge git://ozlabs.org/~paulus/gitk
[git/git.git] / grep.c
CommitLineData
83b5d2f5 1#include "cache.h"
83b5d2f5 2#include "grep.h"
60ecac98 3#include "userdiff.h"
6bfce93e 4#include "xdiff-interface.h"
335ec3bf
JK
5#include "diff.h"
6#include "diffcore.h"
83b5d2f5 7
07a7d656
JH
8static int grep_source_load(struct grep_source *gs);
9static int grep_source_is_binary(struct grep_source *gs);
10
7687a054
JH
11static struct grep_opt grep_defaults;
12
13/*
14 * Initialize the grep_defaults template with hardcoded defaults.
15 * We could let the compiler do this, but without C99 initializers
16 * the code gets unwieldy and unreadable, so...
17 */
18void init_grep_defaults(void)
19{
20 struct grep_opt *opt = &grep_defaults;
918d4e1c
JH
21 static int run_once;
22
23 if (run_once)
24 return;
25 run_once++;
7687a054
JH
26
27 memset(opt, 0, sizeof(*opt));
28 opt->relative = 1;
29 opt->pathname = 1;
30 opt->regflags = REG_NEWLINE;
31 opt->max_depth = -1;
32 opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
33 opt->extended_regexp_option = 0;
34 strcpy(opt->color_context, "");
35 strcpy(opt->color_filename, "");
36 strcpy(opt->color_function, "");
37 strcpy(opt->color_lineno, "");
38 strcpy(opt->color_match, GIT_COLOR_BOLD_RED);
39 strcpy(opt->color_selected, "");
40 strcpy(opt->color_sep, GIT_COLOR_CYAN);
41 opt->color = -1;
42}
43
44static int parse_pattern_type_arg(const char *opt, const char *arg)
45{
46 if (!strcmp(arg, "default"))
47 return GREP_PATTERN_TYPE_UNSPECIFIED;
48 else if (!strcmp(arg, "basic"))
49 return GREP_PATTERN_TYPE_BRE;
50 else if (!strcmp(arg, "extended"))
51 return GREP_PATTERN_TYPE_ERE;
52 else if (!strcmp(arg, "fixed"))
53 return GREP_PATTERN_TYPE_FIXED;
54 else if (!strcmp(arg, "perl"))
55 return GREP_PATTERN_TYPE_PCRE;
56 die("bad %s argument: %s", opt, arg);
57}
58
59/*
60 * Read the configuration file once and store it in
61 * the grep_defaults template.
62 */
63int grep_config(const char *var, const char *value, void *cb)
64{
65 struct grep_opt *opt = &grep_defaults;
66 char *color = NULL;
67
68 if (userdiff_config(var, value) < 0)
69 return -1;
70
71 if (!strcmp(var, "grep.extendedregexp")) {
72 if (git_config_bool(var, value))
73 opt->extended_regexp_option = 1;
74 else
75 opt->extended_regexp_option = 0;
76 return 0;
77 }
78
79 if (!strcmp(var, "grep.patterntype")) {
80 opt->pattern_type_option = parse_pattern_type_arg(var, value);
81 return 0;
82 }
83
84 if (!strcmp(var, "grep.linenumber")) {
85 opt->linenum = git_config_bool(var, value);
86 return 0;
87 }
88
6453f7b3
AS
89 if (!strcmp(var, "grep.fullname")) {
90 opt->relative = !git_config_bool(var, value);
91 return 0;
92 }
93
7687a054
JH
94 if (!strcmp(var, "color.grep"))
95 opt->color = git_config_colorbool(var, value);
96 else if (!strcmp(var, "color.grep.context"))
97 color = opt->color_context;
98 else if (!strcmp(var, "color.grep.filename"))
99 color = opt->color_filename;
100 else if (!strcmp(var, "color.grep.function"))
101 color = opt->color_function;
102 else if (!strcmp(var, "color.grep.linenumber"))
103 color = opt->color_lineno;
104 else if (!strcmp(var, "color.grep.match"))
105 color = opt->color_match;
106 else if (!strcmp(var, "color.grep.selected"))
107 color = opt->color_selected;
108 else if (!strcmp(var, "color.grep.separator"))
109 color = opt->color_sep;
110
111 if (color) {
112 if (!value)
113 return config_error_nonbool(var);
f6c5a296 114 return color_parse(value, color);
7687a054
JH
115 }
116 return 0;
117}
118
119/*
120 * Initialize one instance of grep_opt and copy the
121 * default values from the template we read the configuration
122 * information in an earlier call to git_config(grep_config).
123 */
124void grep_init(struct grep_opt *opt, const char *prefix)
125{
126 struct grep_opt *def = &grep_defaults;
127
128 memset(opt, 0, sizeof(*opt));
129 opt->prefix = prefix;
130 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
131 opt->pattern_tail = &opt->pattern_list;
132 opt->header_tail = &opt->header_list;
133
134 opt->color = def->color;
135 opt->extended_regexp_option = def->extended_regexp_option;
136 opt->pattern_type_option = def->pattern_type_option;
137 opt->linenum = def->linenum;
138 opt->max_depth = def->max_depth;
139 opt->pathname = def->pathname;
140 opt->regflags = def->regflags;
141 opt->relative = def->relative;
142
143 strcpy(opt->color_context, def->color_context);
144 strcpy(opt->color_filename, def->color_filename);
145 strcpy(opt->color_function, def->color_function);
146 strcpy(opt->color_lineno, def->color_lineno);
147 strcpy(opt->color_match, def->color_match);
148 strcpy(opt->color_selected, def->color_selected);
149 strcpy(opt->color_sep, def->color_sep);
150}
07a7d656 151
c5c31d33
JH
152void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
153{
154 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
155 grep_set_pattern_type_option(pattern_type, opt);
156 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
157 grep_set_pattern_type_option(opt->pattern_type_option, opt);
158 else if (opt->extended_regexp_option)
159 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
160}
161
162void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
163{
164 switch (pattern_type) {
165 case GREP_PATTERN_TYPE_UNSPECIFIED:
166 /* fall through */
167
168 case GREP_PATTERN_TYPE_BRE:
169 opt->fixed = 0;
170 opt->pcre = 0;
171 opt->regflags &= ~REG_EXTENDED;
172 break;
173
174 case GREP_PATTERN_TYPE_ERE:
175 opt->fixed = 0;
176 opt->pcre = 0;
177 opt->regflags |= REG_EXTENDED;
178 break;
179
180 case GREP_PATTERN_TYPE_FIXED:
181 opt->fixed = 1;
182 opt->pcre = 0;
183 opt->regflags &= ~REG_EXTENDED;
184 break;
185
186 case GREP_PATTERN_TYPE_PCRE:
187 opt->fixed = 0;
188 opt->pcre = 1;
189 opt->regflags &= ~REG_EXTENDED;
190 break;
191 }
192}
193
fc456751
RS
194static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
195 const char *origin, int no,
196 enum grep_pat_token t,
197 enum grep_header_field field)
a4d7d2c6
JH
198{
199 struct grep_pat *p = xcalloc(1, sizeof(*p));
526a858a 200 p->pattern = xmemdupz(pat, patlen);
fc456751
RS
201 p->patternlen = patlen;
202 p->origin = origin;
203 p->no = no;
204 p->token = t;
a4d7d2c6 205 p->field = field;
fc456751
RS
206 return p;
207}
208
2b3873ff
RS
209static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
210{
211 **tail = p;
212 *tail = &p->next;
a4d7d2c6 213 p->next = NULL;
526a858a
RS
214
215 switch (p->token) {
216 case GREP_PATTERN: /* atom */
217 case GREP_PATTERN_HEAD:
218 case GREP_PATTERN_BODY:
219 for (;;) {
220 struct grep_pat *new_pat;
221 size_t len = 0;
222 char *cp = p->pattern + p->patternlen, *nl = NULL;
223 while (++len <= p->patternlen) {
224 if (*(--cp) == '\n') {
225 nl = cp;
226 break;
227 }
228 }
229 if (!nl)
230 break;
231 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
232 p->no, p->token, p->field);
233 new_pat->next = p->next;
234 if (!p->next)
235 *tail = &new_pat->next;
236 p->next = new_pat;
237 *nl = '\0';
238 p->patternlen -= len;
239 }
240 break;
241 default:
242 break;
243 }
2b3873ff
RS
244}
245
fc456751
RS
246void append_header_grep_pattern(struct grep_opt *opt,
247 enum grep_header_field field, const char *pat)
248{
249 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
250 GREP_PATTERN_HEAD, field);
baa6378f
JH
251 if (field == GREP_HEADER_REFLOG)
252 opt->use_reflog_filter = 1;
2b3873ff 253 do_append_grep_pat(&opt->header_tail, p);
a4d7d2c6
JH
254}
255
83b5d2f5
JH
256void append_grep_pattern(struct grep_opt *opt, const char *pat,
257 const char *origin, int no, enum grep_pat_token t)
ed40a095
RS
258{
259 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
260}
261
262void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
263 const char *origin, int no, enum grep_pat_token t)
83b5d2f5 264{
fc456751 265 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
2b3873ff 266 do_append_grep_pat(&opt->pattern_tail, p);
83b5d2f5
JH
267}
268
5b594f45
FK
269struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
270{
271 struct grep_pat *pat;
272 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
273 *ret = *opt;
274
275 ret->pattern_list = NULL;
276 ret->pattern_tail = &ret->pattern_list;
277
278 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
279 {
280 if(pat->token == GREP_PATTERN_HEAD)
281 append_header_grep_pattern(ret, pat->field,
282 pat->pattern);
283 else
ed40a095
RS
284 append_grep_pat(ret, pat->pattern, pat->patternlen,
285 pat->origin, pat->no, pat->token);
5b594f45
FK
286 }
287
288 return ret;
289}
290
a30c148a
MK
291static NORETURN void compile_regexp_failed(const struct grep_pat *p,
292 const char *error)
293{
294 char where[1024];
295
296 if (p->no)
297 sprintf(where, "In '%s' at %d, ", p->origin, p->no);
298 else if (p->origin)
299 sprintf(where, "%s, ", p->origin);
300 else
301 where[0] = 0;
302
303 die("%s'%s': %s", where, p->pattern, error);
304}
305
63e7e9d8
MK
306#ifdef USE_LIBPCRE
307static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
308{
309 const char *error;
310 int erroffset;
fba4f125 311 int options = PCRE_MULTILINE;
63e7e9d8
MK
312
313 if (opt->ignore_case)
314 options |= PCRE_CASELESS;
315
316 p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
317 NULL);
318 if (!p->pcre_regexp)
319 compile_regexp_failed(p, error);
320
321 p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
322 if (!p->pcre_extra_info && error)
323 die("%s", error);
324}
325
326static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
327 regmatch_t *match, int eflags)
328{
329 int ovector[30], ret, flags = 0;
330
331 if (eflags & REG_NOTBOL)
332 flags |= PCRE_NOTBOL;
333
334 ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
335 0, flags, ovector, ARRAY_SIZE(ovector));
336 if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
337 die("pcre_exec failed with error code %d", ret);
338 if (ret > 0) {
339 ret = 0;
340 match->rm_so = ovector[0];
341 match->rm_eo = ovector[1];
342 }
343
344 return ret;
345}
346
347static void free_pcre_regexp(struct grep_pat *p)
348{
349 pcre_free(p->pcre_regexp);
350 pcre_free(p->pcre_extra_info);
351}
352#else /* !USE_LIBPCRE */
353static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
354{
355 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
356}
357
358static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
359 regmatch_t *match, int eflags)
360{
361 return 1;
362}
363
364static void free_pcre_regexp(struct grep_pat *p)
365{
366}
367#endif /* !USE_LIBPCRE */
368
9eceddee
FK
369static int is_fixed(const char *s, size_t len)
370{
371 size_t i;
372
373 /* regcomp cannot accept patterns with NULs so we
374 * consider any pattern containing a NUL fixed.
375 */
376 if (memchr(s, 0, len))
377 return 1;
378
379 for (i = 0; i < len; i++) {
380 if (is_regex_special(s[i]))
381 return 0;
382 }
383
384 return 1;
385}
386
83b5d2f5
JH
387static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
388{
c822255c
RS
389 int err;
390
d7eb527d 391 p->word_regexp = opt->word_regexp;
5183bf67 392 p->ignore_case = opt->ignore_case;
d7eb527d 393
9eceddee
FK
394 if (opt->fixed || is_fixed(p->pattern, p->patternlen))
395 p->fixed = 1;
396 else
397 p->fixed = 0;
398
399 if (p->fixed) {
0f871cf5
JH
400 if (opt->regflags & REG_ICASE || p->ignore_case)
401 p->kws = kwsalloc(tolower_trans_tbl);
402 else
9eceddee 403 p->kws = kwsalloc(NULL);
9eceddee
FK
404 kwsincr(p->kws, p->pattern, p->patternlen);
405 kwsprep(p->kws);
c822255c 406 return;
9eceddee 407 }
c822255c 408
63e7e9d8
MK
409 if (opt->pcre) {
410 compile_pcre_regexp(p, opt);
411 return;
412 }
413
c822255c 414 err = regcomp(&p->regexp, p->pattern, opt->regflags);
83b5d2f5
JH
415 if (err) {
416 char errbuf[1024];
83b5d2f5
JH
417 regerror(err, &p->regexp, errbuf, 1024);
418 regfree(&p->regexp);
a30c148a 419 compile_regexp_failed(p, errbuf);
83b5d2f5
JH
420 }
421}
422
0ab7befa 423static struct grep_expr *compile_pattern_or(struct grep_pat **);
83b5d2f5
JH
424static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
425{
426 struct grep_pat *p;
427 struct grep_expr *x;
428
429 p = *list;
c922b01f
LT
430 if (!p)
431 return NULL;
83b5d2f5
JH
432 switch (p->token) {
433 case GREP_PATTERN: /* atom */
480c1ca6
JH
434 case GREP_PATTERN_HEAD:
435 case GREP_PATTERN_BODY:
83b5d2f5
JH
436 x = xcalloc(1, sizeof (struct grep_expr));
437 x->node = GREP_NODE_ATOM;
438 x->u.atom = p;
439 *list = p->next;
440 return x;
441 case GREP_OPEN_PAREN:
442 *list = p->next;
0ab7befa 443 x = compile_pattern_or(list);
83b5d2f5
JH
444 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
445 die("unmatched parenthesis");
446 *list = (*list)->next;
447 return x;
448 default:
449 return NULL;
450 }
451}
452
453static struct grep_expr *compile_pattern_not(struct grep_pat **list)
454{
455 struct grep_pat *p;
456 struct grep_expr *x;
457
458 p = *list;
c922b01f
LT
459 if (!p)
460 return NULL;
83b5d2f5
JH
461 switch (p->token) {
462 case GREP_NOT:
463 if (!p->next)
464 die("--not not followed by pattern expression");
465 *list = p->next;
466 x = xcalloc(1, sizeof (struct grep_expr));
467 x->node = GREP_NODE_NOT;
468 x->u.unary = compile_pattern_not(list);
469 if (!x->u.unary)
470 die("--not followed by non pattern expression");
471 return x;
472 default:
473 return compile_pattern_atom(list);
474 }
475}
476
477static struct grep_expr *compile_pattern_and(struct grep_pat **list)
478{
479 struct grep_pat *p;
480 struct grep_expr *x, *y, *z;
481
482 x = compile_pattern_not(list);
483 p = *list;
484 if (p && p->token == GREP_AND) {
485 if (!p->next)
486 die("--and not followed by pattern expression");
487 *list = p->next;
488 y = compile_pattern_and(list);
489 if (!y)
490 die("--and not followed by pattern expression");
491 z = xcalloc(1, sizeof (struct grep_expr));
492 z->node = GREP_NODE_AND;
493 z->u.binary.left = x;
494 z->u.binary.right = y;
495 return z;
496 }
497 return x;
498}
499
500static struct grep_expr *compile_pattern_or(struct grep_pat **list)
501{
502 struct grep_pat *p;
503 struct grep_expr *x, *y, *z;
504
505 x = compile_pattern_and(list);
506 p = *list;
507 if (x && p && p->token != GREP_CLOSE_PAREN) {
508 y = compile_pattern_or(list);
509 if (!y)
510 die("not a pattern expression %s", p->pattern);
511 z = xcalloc(1, sizeof (struct grep_expr));
512 z->node = GREP_NODE_OR;
513 z->u.binary.left = x;
514 z->u.binary.right = y;
515 return z;
516 }
517 return x;
518}
519
520static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
521{
522 return compile_pattern_or(list);
523}
524
17bf35a3
JH
525static void indent(int in)
526{
527 while (in-- > 0)
528 fputc(' ', stderr);
529}
530
531static void dump_grep_pat(struct grep_pat *p)
532{
533 switch (p->token) {
534 case GREP_AND: fprintf(stderr, "*and*"); break;
535 case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
536 case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
537 case GREP_NOT: fprintf(stderr, "*not*"); break;
538 case GREP_OR: fprintf(stderr, "*or*"); break;
539
540 case GREP_PATTERN: fprintf(stderr, "pattern"); break;
541 case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
542 case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
543 }
544
545 switch (p->token) {
546 default: break;
547 case GREP_PATTERN_HEAD:
548 fprintf(stderr, "<head %d>", p->field); break;
549 case GREP_PATTERN_BODY:
550 fprintf(stderr, "<body>"); break;
551 }
552 switch (p->token) {
553 default: break;
554 case GREP_PATTERN_HEAD:
555 case GREP_PATTERN_BODY:
556 case GREP_PATTERN:
557 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
558 break;
559 }
560 fputc('\n', stderr);
561}
562
563static void dump_grep_expression_1(struct grep_expr *x, int in)
564{
565 indent(in);
566 switch (x->node) {
567 case GREP_NODE_TRUE:
568 fprintf(stderr, "true\n");
569 break;
570 case GREP_NODE_ATOM:
571 dump_grep_pat(x->u.atom);
572 break;
573 case GREP_NODE_NOT:
574 fprintf(stderr, "(not\n");
575 dump_grep_expression_1(x->u.unary, in+1);
576 indent(in);
577 fprintf(stderr, ")\n");
578 break;
579 case GREP_NODE_AND:
580 fprintf(stderr, "(and\n");
581 dump_grep_expression_1(x->u.binary.left, in+1);
582 dump_grep_expression_1(x->u.binary.right, in+1);
583 indent(in);
584 fprintf(stderr, ")\n");
585 break;
586 case GREP_NODE_OR:
587 fprintf(stderr, "(or\n");
588 dump_grep_expression_1(x->u.binary.left, in+1);
589 dump_grep_expression_1(x->u.binary.right, in+1);
590 indent(in);
591 fprintf(stderr, ")\n");
592 break;
593 }
594}
595
07a7d656 596static void dump_grep_expression(struct grep_opt *opt)
17bf35a3
JH
597{
598 struct grep_expr *x = opt->pattern_expression;
599
600 if (opt->all_match)
601 fprintf(stderr, "[all-match]\n");
602 dump_grep_expression_1(x, 0);
603 fflush(NULL);
604}
605
5aaeb733
JH
606static struct grep_expr *grep_true_expr(void)
607{
608 struct grep_expr *z = xcalloc(1, sizeof(*z));
609 z->node = GREP_NODE_TRUE;
610 return z;
611}
612
613static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
614{
615 struct grep_expr *z = xcalloc(1, sizeof(*z));
616 z->node = GREP_NODE_OR;
617 z->u.binary.left = left;
618 z->u.binary.right = right;
619 return z;
620}
621
95ce9ce2 622static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
83b5d2f5
JH
623{
624 struct grep_pat *p;
95ce9ce2 625 struct grep_expr *header_expr;
5aaeb733
JH
626 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
627 enum grep_header_field fld;
83b5d2f5 628
95ce9ce2
JH
629 if (!opt->header_list)
630 return NULL;
2385f246 631
95ce9ce2
JH
632 for (p = opt->header_list; p; p = p->next) {
633 if (p->token != GREP_PATTERN_HEAD)
634 die("bug: a non-header pattern in grep header list.");
3ce3ffb8
AP
635 if (p->field < GREP_HEADER_FIELD_MIN ||
636 GREP_HEADER_FIELD_MAX <= p->field)
95ce9ce2
JH
637 die("bug: unknown header field %d", p->field);
638 compile_regexp(p, opt);
80235ba7 639 }
5aaeb733
JH
640
641 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
642 header_group[fld] = NULL;
643
644 for (p = opt->header_list; p; p = p->next) {
645 struct grep_expr *h;
646 struct grep_pat *pp = p;
647
648 h = compile_pattern_atom(&pp);
649 if (!h || pp != p->next)
650 die("bug: malformed header expr");
651 if (!header_group[p->field]) {
652 header_group[p->field] = h;
653 continue;
654 }
655 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
656 }
657
658 header_expr = NULL;
659
660 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
661 if (!header_group[fld])
662 continue;
663 if (!header_expr)
664 header_expr = grep_true_expr();
665 header_expr = grep_or_expr(header_group[fld], header_expr);
666 }
95ce9ce2
JH
667 return header_expr;
668}
669
13e4fc7e
JH
670static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
671{
672 struct grep_expr *z = x;
673
674 while (x) {
675 assert(x->node == GREP_NODE_OR);
676 if (x->u.binary.right &&
677 x->u.binary.right->node == GREP_NODE_TRUE) {
678 x->u.binary.right = y;
679 break;
680 }
681 x = x->u.binary.right;
682 }
683 return z;
684}
685
17bf35a3 686static void compile_grep_patterns_real(struct grep_opt *opt)
95ce9ce2
JH
687{
688 struct grep_pat *p;
689 struct grep_expr *header_expr = prep_header_patterns(opt);
0ab7befa 690
83b5d2f5 691 for (p = opt->pattern_list; p; p = p->next) {
480c1ca6
JH
692 switch (p->token) {
693 case GREP_PATTERN: /* atom */
694 case GREP_PATTERN_HEAD:
695 case GREP_PATTERN_BODY:
c822255c 696 compile_regexp(p, opt);
480c1ca6
JH
697 break;
698 default:
83b5d2f5 699 opt->extended = 1;
480c1ca6
JH
700 break;
701 }
83b5d2f5
JH
702 }
703
80235ba7
JH
704 if (opt->all_match || header_expr)
705 opt->extended = 1;
17bf35a3 706 else if (!opt->extended && !opt->debug)
83b5d2f5
JH
707 return;
708
83b5d2f5 709 p = opt->pattern_list;
ba150a3f
MB
710 if (p)
711 opt->pattern_expression = compile_pattern_expr(&p);
83b5d2f5
JH
712 if (p)
713 die("incomplete pattern expression: %s", p->pattern);
80235ba7
JH
714
715 if (!header_expr)
716 return;
717
5aaeb733 718 if (!opt->pattern_expression)
80235ba7 719 opt->pattern_expression = header_expr;
13e4fc7e
JH
720 else if (opt->all_match)
721 opt->pattern_expression = grep_splice_or(header_expr,
722 opt->pattern_expression);
5aaeb733
JH
723 else
724 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
725 header_expr);
80235ba7 726 opt->all_match = 1;
83b5d2f5
JH
727}
728
17bf35a3
JH
729void compile_grep_patterns(struct grep_opt *opt)
730{
731 compile_grep_patterns_real(opt);
732 if (opt->debug)
733 dump_grep_expression(opt);
734}
735
b48fb5b6
JH
736static void free_pattern_expr(struct grep_expr *x)
737{
738 switch (x->node) {
5aaeb733 739 case GREP_NODE_TRUE:
b48fb5b6
JH
740 case GREP_NODE_ATOM:
741 break;
742 case GREP_NODE_NOT:
743 free_pattern_expr(x->u.unary);
744 break;
745 case GREP_NODE_AND:
746 case GREP_NODE_OR:
747 free_pattern_expr(x->u.binary.left);
748 free_pattern_expr(x->u.binary.right);
749 break;
750 }
751 free(x);
752}
753
754void free_grep_patterns(struct grep_opt *opt)
755{
756 struct grep_pat *p, *n;
757
758 for (p = opt->pattern_list; p; p = n) {
759 n = p->next;
760 switch (p->token) {
761 case GREP_PATTERN: /* atom */
762 case GREP_PATTERN_HEAD:
763 case GREP_PATTERN_BODY:
9eceddee
FK
764 if (p->kws)
765 kwsfree(p->kws);
766 else if (p->pcre_regexp)
63e7e9d8
MK
767 free_pcre_regexp(p);
768 else
769 regfree(&p->regexp);
526a858a 770 free(p->pattern);
b48fb5b6
JH
771 break;
772 default:
773 break;
774 }
775 free(p);
776 }
777
778 if (!opt->extended)
779 return;
780 free_pattern_expr(opt->pattern_expression);
781}
782
83b5d2f5
JH
783static char *end_of_line(char *cp, unsigned long *left)
784{
785 unsigned long l = *left;
786 while (l && *cp != '\n') {
787 l--;
788 cp++;
789 }
790 *left = l;
791 return cp;
792}
793
794static int word_char(char ch)
795{
796 return isalnum(ch) || ch == '_';
797}
798
55f638bd
ML
799static void output_color(struct grep_opt *opt, const void *data, size_t size,
800 const char *color)
801{
daa0c3d9 802 if (want_color(opt->color) && color && color[0]) {
55f638bd
ML
803 opt->output(opt, color, strlen(color));
804 opt->output(opt, data, size);
805 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
806 } else
807 opt->output(opt, data, size);
808}
809
810static void output_sep(struct grep_opt *opt, char sign)
811{
812 if (opt->null_following_name)
813 opt->output(opt, "\0", 1);
814 else
815 output_color(opt, &sign, 1, opt->color_sep);
816}
817
83caecca
RZ
818static void show_name(struct grep_opt *opt, const char *name)
819{
55f638bd 820 output_color(opt, name, strlen(name), opt->color_filename);
5b594f45 821 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
83caecca
RZ
822}
823
ed40a095
RS
824static int fixmatch(struct grep_pat *p, char *line, char *eol,
825 regmatch_t *match)
83b5d2f5 826{
9eceddee
FK
827 struct kwsmatch kwsm;
828 size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
829 if (offset == -1) {
83b5d2f5
JH
830 match->rm_so = match->rm_eo = -1;
831 return REG_NOMATCH;
9eceddee
FK
832 } else {
833 match->rm_so = offset;
834 match->rm_eo = match->rm_so + kwsm.size[0];
83b5d2f5
JH
835 return 0;
836 }
837}
838
f96e5673
RS
839static int regmatch(const regex_t *preg, char *line, char *eol,
840 regmatch_t *match, int eflags)
841{
842#ifdef REG_STARTEND
843 match->rm_so = 0;
844 match->rm_eo = eol - line;
845 eflags |= REG_STARTEND;
846#endif
847 return regexec(preg, line, 1, match, eflags);
848}
849
97e77784
MK
850static int patmatch(struct grep_pat *p, char *line, char *eol,
851 regmatch_t *match, int eflags)
852{
853 int hit;
854
855 if (p->fixed)
856 hit = !fixmatch(p, line, eol, match);
63e7e9d8
MK
857 else if (p->pcre_regexp)
858 hit = !pcrematch(p, line, eol, match, eflags);
97e77784
MK
859 else
860 hit = !regmatch(&p->regexp, line, eol, match, eflags);
861
862 return hit;
863}
864
a4d7d2c6
JH
865static int strip_timestamp(char *bol, char **eol_p)
866{
867 char *eol = *eol_p;
868 int ch;
869
870 while (bol < --eol) {
871 if (*eol != '>')
872 continue;
873 *eol_p = ++eol;
874 ch = *eol;
875 *eol = '\0';
876 return ch;
877 }
878 return 0;
879}
880
881static struct {
882 const char *field;
883 size_t len;
884} header_field[] = {
885 { "author ", 7 },
886 { "committer ", 10 },
72fd13f7 887 { "reflog ", 7 },
a4d7d2c6
JH
888};
889
d7eb527d 890static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
79212772
RS
891 enum grep_context ctx,
892 regmatch_t *pmatch, int eflags)
83b5d2f5
JH
893{
894 int hit = 0;
a4d7d2c6 895 int saved_ch = 0;
e701fadb 896 const char *start = bol;
83b5d2f5 897
480c1ca6
JH
898 if ((p->token != GREP_PATTERN) &&
899 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
900 return 0;
901
a4d7d2c6
JH
902 if (p->token == GREP_PATTERN_HEAD) {
903 const char *field;
904 size_t len;
905 assert(p->field < ARRAY_SIZE(header_field));
906 field = header_field[p->field].field;
907 len = header_field[p->field].len;
908 if (strncmp(bol, field, len))
909 return 0;
910 bol += len;
ad4813b3
NTND
911 switch (p->field) {
912 case GREP_HEADER_AUTHOR:
913 case GREP_HEADER_COMMITTER:
914 saved_ch = strip_timestamp(bol, &eol);
915 break;
916 default:
917 break;
918 }
a4d7d2c6
JH
919 }
920
83b5d2f5 921 again:
97e77784 922 hit = patmatch(p, bol, eol, pmatch, eflags);
83b5d2f5 923
d7eb527d 924 if (hit && p->word_regexp) {
83b5d2f5 925 if ((pmatch[0].rm_so < 0) ||
84201eae 926 (eol - bol) < pmatch[0].rm_so ||
83b5d2f5
JH
927 (pmatch[0].rm_eo < 0) ||
928 (eol - bol) < pmatch[0].rm_eo)
929 die("regexp returned nonsense");
930
931 /* Match beginning must be either beginning of the
932 * line, or at word boundary (i.e. the last char must
933 * not be a word char). Similarly, match end must be
934 * either end of the line, or at word boundary
935 * (i.e. the next char must not be a word char).
936 */
fb62eb7f 937 if ( ((pmatch[0].rm_so == 0) ||
83b5d2f5
JH
938 !word_char(bol[pmatch[0].rm_so-1])) &&
939 ((pmatch[0].rm_eo == (eol-bol)) ||
940 !word_char(bol[pmatch[0].rm_eo])) )
941 ;
942 else
943 hit = 0;
944
84201eae
RS
945 /* Words consist of at least one character. */
946 if (pmatch->rm_so == pmatch->rm_eo)
947 hit = 0;
948
83b5d2f5
JH
949 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
950 /* There could be more than one match on the
951 * line, and the first match might not be
952 * strict word match. But later ones could be!
fb62eb7f
RS
953 * Forward to the next possible start, i.e. the
954 * next position following a non-word char.
83b5d2f5
JH
955 */
956 bol = pmatch[0].rm_so + bol + 1;
fb62eb7f
RS
957 while (word_char(bol[-1]) && bol < eol)
958 bol++;
dbb6a4ad 959 eflags |= REG_NOTBOL;
fb62eb7f
RS
960 if (bol < eol)
961 goto again;
83b5d2f5
JH
962 }
963 }
a4d7d2c6
JH
964 if (p->token == GREP_PATTERN_HEAD && saved_ch)
965 *eol = saved_ch;
e701fadb
RS
966 if (hit) {
967 pmatch[0].rm_so += bol - start;
968 pmatch[0].rm_eo += bol - start;
969 }
83b5d2f5
JH
970 return hit;
971}
972
d7eb527d
RS
973static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
974 enum grep_context ctx, int collect_hits)
83b5d2f5 975{
0ab7befa 976 int h = 0;
79212772 977 regmatch_t match;
0ab7befa 978
c922b01f
LT
979 if (!x)
980 die("Not a valid grep expression");
83b5d2f5 981 switch (x->node) {
5aaeb733
JH
982 case GREP_NODE_TRUE:
983 h = 1;
984 break;
83b5d2f5 985 case GREP_NODE_ATOM:
79212772 986 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
83b5d2f5
JH
987 break;
988 case GREP_NODE_NOT:
d7eb527d 989 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
0ab7befa 990 break;
83b5d2f5 991 case GREP_NODE_AND:
d7eb527d 992 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
252d560d 993 return 0;
d7eb527d 994 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
0ab7befa 995 break;
83b5d2f5 996 case GREP_NODE_OR:
0ab7befa 997 if (!collect_hits)
d7eb527d 998 return (match_expr_eval(x->u.binary.left,
0ab7befa 999 bol, eol, ctx, 0) ||
d7eb527d 1000 match_expr_eval(x->u.binary.right,
0ab7befa 1001 bol, eol, ctx, 0));
d7eb527d 1002 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
0ab7befa 1003 x->u.binary.left->hit |= h;
d7eb527d 1004 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
0ab7befa
JH
1005 break;
1006 default:
d7530708 1007 die("Unexpected node type (internal error) %d", x->node);
83b5d2f5 1008 }
0ab7befa
JH
1009 if (collect_hits)
1010 x->hit |= h;
1011 return h;
83b5d2f5
JH
1012}
1013
480c1ca6 1014static int match_expr(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1015 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1016{
1017 struct grep_expr *x = opt->pattern_expression;
d7eb527d 1018 return match_expr_eval(x, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1019}
1020
480c1ca6 1021static int match_line(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1022 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1023{
1024 struct grep_pat *p;
79212772
RS
1025 regmatch_t match;
1026
83b5d2f5 1027 if (opt->extended)
0ab7befa
JH
1028 return match_expr(opt, bol, eol, ctx, collect_hits);
1029
1030 /* we do not call with collect_hits without being extended */
83b5d2f5 1031 for (p = opt->pattern_list; p; p = p->next) {
79212772 1032 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
83b5d2f5
JH
1033 return 1;
1034 }
1035 return 0;
1036}
1037
7e8f59d5
RS
1038static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1039 enum grep_context ctx,
1040 regmatch_t *pmatch, int eflags)
1041{
1042 regmatch_t match;
1043
1044 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1045 return 0;
1046 if (match.rm_so < 0 || match.rm_eo < 0)
1047 return 0;
1048 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1049 if (match.rm_so > pmatch->rm_so)
1050 return 1;
1051 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1052 return 1;
1053 }
1054 pmatch->rm_so = match.rm_so;
1055 pmatch->rm_eo = match.rm_eo;
1056 return 1;
1057}
1058
1059static int next_match(struct grep_opt *opt, char *bol, char *eol,
1060 enum grep_context ctx, regmatch_t *pmatch, int eflags)
1061{
1062 struct grep_pat *p;
1063 int hit = 0;
1064
1065 pmatch->rm_so = pmatch->rm_eo = -1;
1066 if (bol < eol) {
1067 for (p = opt->pattern_list; p; p = p->next) {
1068 switch (p->token) {
1069 case GREP_PATTERN: /* atom */
1070 case GREP_PATTERN_HEAD:
1071 case GREP_PATTERN_BODY:
1072 hit |= match_next_pattern(p, bol, eol, ctx,
1073 pmatch, eflags);
1074 break;
1075 default:
1076 break;
1077 }
1078 }
1079 }
1080 return hit;
1081}
1082
1083static void show_line(struct grep_opt *opt, char *bol, char *eol,
1084 const char *name, unsigned lno, char sign)
1085{
1086 int rest = eol - bol;
00588bb5 1087 char *line_color = NULL;
7e8f59d5 1088
a8f0e764
RS
1089 if (opt->file_break && opt->last_shown == 0) {
1090 if (opt->show_hunk_mark)
1091 opt->output(opt, "\n", 1);
ba8ea749 1092 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
046802d0 1093 if (opt->last_shown == 0) {
55f638bd
ML
1094 if (opt->show_hunk_mark) {
1095 output_color(opt, "--", 2, opt->color_sep);
1096 opt->output(opt, "\n", 1);
07b838f0 1097 }
55f638bd
ML
1098 } else if (lno > opt->last_shown + 1) {
1099 output_color(opt, "--", 2, opt->color_sep);
1100 opt->output(opt, "\n", 1);
1101 }
5dd06d38 1102 }
1d84f72e
RS
1103 if (opt->heading && opt->last_shown == 0) {
1104 output_color(opt, name, strlen(name), opt->color_filename);
1105 opt->output(opt, "\n", 1);
1106 }
5dd06d38
RS
1107 opt->last_shown = lno;
1108
1d84f72e 1109 if (!opt->heading && opt->pathname) {
55f638bd
ML
1110 output_color(opt, name, strlen(name), opt->color_filename);
1111 output_sep(opt, sign);
5b594f45
FK
1112 }
1113 if (opt->linenum) {
1114 char buf[32];
1115 snprintf(buf, sizeof(buf), "%d", lno);
55f638bd
ML
1116 output_color(opt, buf, strlen(buf), opt->color_lineno);
1117 output_sep(opt, sign);
5b594f45 1118 }
7e8f59d5
RS
1119 if (opt->color) {
1120 regmatch_t match;
1121 enum grep_context ctx = GREP_CONTEXT_BODY;
1122 int ch = *eol;
1123 int eflags = 0;
1124
00588bb5
ML
1125 if (sign == ':')
1126 line_color = opt->color_selected;
1127 else if (sign == '-')
1128 line_color = opt->color_context;
1129 else if (sign == '=')
1130 line_color = opt->color_function;
7e8f59d5
RS
1131 *eol = '\0';
1132 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1f5b9cc4
RS
1133 if (match.rm_so == match.rm_eo)
1134 break;
5b594f45 1135
00588bb5 1136 output_color(opt, bol, match.rm_so, line_color);
55f638bd
ML
1137 output_color(opt, bol + match.rm_so,
1138 match.rm_eo - match.rm_so,
1139 opt->color_match);
7e8f59d5
RS
1140 bol += match.rm_eo;
1141 rest -= match.rm_eo;
1142 eflags = REG_NOTBOL;
1143 }
1144 *eol = ch;
1145 }
00588bb5 1146 output_color(opt, bol, rest, line_color);
5b594f45 1147 opt->output(opt, "\n", 1);
7e8f59d5
RS
1148}
1149
0579f91d 1150#ifndef NO_PTHREADS
78db6ea9
JK
1151int grep_use_locks;
1152
0579f91d
TR
1153/*
1154 * This lock protects access to the gitattributes machinery, which is
1155 * not thread-safe.
1156 */
1157pthread_mutex_t grep_attr_mutex;
1158
78db6ea9 1159static inline void grep_attr_lock(void)
0579f91d 1160{
78db6ea9 1161 if (grep_use_locks)
0579f91d
TR
1162 pthread_mutex_lock(&grep_attr_mutex);
1163}
1164
78db6ea9 1165static inline void grep_attr_unlock(void)
0579f91d 1166{
78db6ea9 1167 if (grep_use_locks)
0579f91d
TR
1168 pthread_mutex_unlock(&grep_attr_mutex);
1169}
b3aeb285
JK
1170
1171/*
1172 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1173 */
1174pthread_mutex_t grep_read_mutex;
1175
0579f91d 1176#else
78db6ea9
JK
1177#define grep_attr_lock()
1178#define grep_attr_unlock()
0579f91d
TR
1179#endif
1180
e1327023 1181static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
2944e4e6 1182{
60ecac98 1183 xdemitconf_t *xecfg = opt->priv;
0579f91d 1184 if (xecfg && !xecfg->find_func) {
94ad9d9e
JK
1185 grep_source_load_driver(gs);
1186 if (gs->driver->funcname.pattern) {
1187 const struct userdiff_funcname *pe = &gs->driver->funcname;
0579f91d
TR
1188 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1189 } else {
1190 xecfg = opt->priv = NULL;
1191 }
1192 }
1193
1194 if (xecfg) {
60ecac98
RS
1195 char buf[1];
1196 return xecfg->find_func(bol, eol - bol, buf, 1,
1197 xecfg->find_func_priv) >= 0;
1198 }
1199
2944e4e6
RS
1200 if (bol == eol)
1201 return 0;
1202 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1203 return 1;
1204 return 0;
1205}
1206
e1327023
JK
1207static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1208 char *bol, unsigned lno)
2944e4e6 1209{
e1327023 1210 while (bol > gs->buf) {
2944e4e6
RS
1211 char *eol = --bol;
1212
e1327023 1213 while (bol > gs->buf && bol[-1] != '\n')
2944e4e6
RS
1214 bol--;
1215 lno--;
1216
1217 if (lno <= opt->last_shown)
1218 break;
1219
e1327023
JK
1220 if (match_funcname(opt, gs, bol, eol)) {
1221 show_line(opt, bol, eol, gs->name, lno, '=');
2944e4e6
RS
1222 break;
1223 }
1224 }
1225}
1226
e1327023 1227static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
ba8ea749 1228 char *bol, char *end, unsigned lno)
49de3216 1229{
2944e4e6 1230 unsigned cur = lno, from = 1, funcname_lno = 0;
ba8ea749
RS
1231 int funcname_needed = !!opt->funcname;
1232
e1327023 1233 if (opt->funcbody && !match_funcname(opt, gs, bol, end))
ba8ea749 1234 funcname_needed = 2;
49de3216
RS
1235
1236 if (opt->pre_context < lno)
1237 from = lno - opt->pre_context;
1238 if (from <= opt->last_shown)
1239 from = opt->last_shown + 1;
1240
1241 /* Rewind. */
e1327023 1242 while (bol > gs->buf &&
ba8ea749 1243 cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
2944e4e6
RS
1244 char *eol = --bol;
1245
e1327023 1246 while (bol > gs->buf && bol[-1] != '\n')
49de3216
RS
1247 bol--;
1248 cur--;
e1327023 1249 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
2944e4e6
RS
1250 funcname_lno = cur;
1251 funcname_needed = 0;
1252 }
49de3216
RS
1253 }
1254
2944e4e6
RS
1255 /* We need to look even further back to find a function signature. */
1256 if (opt->funcname && funcname_needed)
e1327023 1257 show_funcname_line(opt, gs, bol, cur);
2944e4e6 1258
49de3216
RS
1259 /* Back forward. */
1260 while (cur < lno) {
2944e4e6 1261 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
49de3216
RS
1262
1263 while (*eol != '\n')
1264 eol++;
e1327023 1265 show_line(opt, bol, eol, gs->name, cur, sign);
49de3216
RS
1266 bol = eol + 1;
1267 cur++;
1268 }
1269}
1270
a26345b6
JH
1271static int should_lookahead(struct grep_opt *opt)
1272{
1273 struct grep_pat *p;
1274
1275 if (opt->extended)
1276 return 0; /* punt for too complex stuff */
1277 if (opt->invert)
1278 return 0;
1279 for (p = opt->pattern_list; p; p = p->next) {
1280 if (p->token != GREP_PATTERN)
1281 return 0; /* punt for "header only" and stuff */
1282 }
1283 return 1;
1284}
1285
1286static int look_ahead(struct grep_opt *opt,
1287 unsigned long *left_p,
1288 unsigned *lno_p,
1289 char **bol_p)
1290{
1291 unsigned lno = *lno_p;
1292 char *bol = *bol_p;
1293 struct grep_pat *p;
1294 char *sp, *last_bol;
1295 regoff_t earliest = -1;
1296
1297 for (p = opt->pattern_list; p; p = p->next) {
1298 int hit;
1299 regmatch_t m;
1300
97e77784 1301 hit = patmatch(p, bol, bol + *left_p, &m, 0);
a26345b6
JH
1302 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1303 continue;
1304 if (earliest < 0 || m.rm_so < earliest)
1305 earliest = m.rm_so;
1306 }
1307
1308 if (earliest < 0) {
1309 *bol_p = bol + *left_p;
1310 *left_p = 0;
1311 return 1;
1312 }
1313 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1314 ; /* find the beginning of the line */
1315 last_bol = sp;
1316
1317 for (sp = bol; sp < last_bol; sp++) {
1318 if (*sp == '\n')
1319 lno++;
1320 }
1321 *left_p -= last_bol - bol;
1322 *bol_p = last_bol;
1323 *lno_p = lno;
1324 return 0;
1325}
1326
5b594f45
FK
1327static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1328{
1329 fwrite(buf, size, 1, stdout);
1330}
1331
335ec3bf
JK
1332static int fill_textconv_grep(struct userdiff_driver *driver,
1333 struct grep_source *gs)
1334{
1335 struct diff_filespec *df;
1336 char *buf;
1337 size_t size;
1338
1339 if (!driver || !driver->textconv)
1340 return grep_source_load(gs);
1341
1342 /*
1343 * The textconv interface is intimately tied to diff_filespecs, so we
1344 * have to pretend to be one. If we could unify the grep_source
1345 * and diff_filespec structs, this mess could just go away.
1346 */
1347 df = alloc_filespec(gs->path);
1348 switch (gs->type) {
1349 case GREP_SOURCE_SHA1:
1350 fill_filespec(df, gs->identifier, 1, 0100644);
1351 break;
1352 case GREP_SOURCE_FILE:
1353 fill_filespec(df, null_sha1, 0, 0100644);
1354 break;
1355 default:
1356 die("BUG: attempt to textconv something without a path?");
1357 }
1358
1359 /*
1360 * fill_textconv is not remotely thread-safe; it may load objects
1361 * behind the scenes, and it modifies the global diff tempfile
1362 * structure.
1363 */
1364 grep_read_lock();
1365 size = fill_textconv(driver, df, &buf);
1366 grep_read_unlock();
1367 free_filespec(df);
1368
1369 /*
1370 * The normal fill_textconv usage by the diff machinery would just keep
1371 * the textconv'd buf separate from the diff_filespec. But much of the
1372 * grep code passes around a grep_source and assumes that its "buf"
1373 * pointer is the beginning of the thing we are searching. So let's
1374 * install our textconv'd version into the grep_source, taking care not
1375 * to leak any existing buffer.
1376 */
1377 grep_source_clear_data(gs);
1378 gs->buf = buf;
1379 gs->size = size;
1380
1381 return 0;
1382}
1383
e1327023 1384static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
83b5d2f5 1385{
e1327023
JK
1386 char *bol;
1387 unsigned long left;
83b5d2f5 1388 unsigned lno = 1;
83b5d2f5 1389 unsigned last_hit = 0;
83b5d2f5 1390 int binary_match_only = 0;
83b5d2f5 1391 unsigned count = 0;
a26345b6 1392 int try_lookahead = 0;
ba8ea749 1393 int show_function = 0;
335ec3bf 1394 struct userdiff_driver *textconv = NULL;
480c1ca6 1395 enum grep_context ctx = GREP_CONTEXT_HEAD;
60ecac98 1396 xdemitconf_t xecfg;
83b5d2f5 1397
5b594f45
FK
1398 if (!opt->output)
1399 opt->output = std_output;
1400
ba8ea749
RS
1401 if (opt->pre_context || opt->post_context || opt->file_break ||
1402 opt->funcbody) {
08303c36
RS
1403 /* Show hunk marks, except for the first file. */
1404 if (opt->last_shown)
1405 opt->show_hunk_mark = 1;
1406 /*
1407 * If we're using threads then we can't easily identify
1408 * the first file. Always put hunk marks in that case
1409 * and skip the very first one later in work_done().
1410 */
1411 if (opt->output != std_output)
1412 opt->show_hunk_mark = 1;
1413 }
431d6e7b
RS
1414 opt->last_shown = 0;
1415
335ec3bf
JK
1416 if (opt->allow_textconv) {
1417 grep_source_load_driver(gs);
1418 /*
1419 * We might set up the shared textconv cache data here, which
1420 * is not thread-safe.
1421 */
1422 grep_attr_lock();
1423 textconv = userdiff_get_textconv(gs->driver);
1424 grep_attr_unlock();
1425 }
1426
1427 /*
1428 * We know the result of a textconv is text, so we only have to care
1429 * about binary handling if we are not using it.
1430 */
1431 if (!textconv) {
1432 switch (opt->binary) {
1433 case GREP_BINARY_DEFAULT:
1434 if (grep_source_is_binary(gs))
1435 binary_match_only = 1;
1436 break;
1437 case GREP_BINARY_NOMATCH:
1438 if (grep_source_is_binary(gs))
1439 return 0; /* Assume unmatch */
1440 break;
1441 case GREP_BINARY_TEXT:
1442 break;
1443 default:
1444 die("bug: unknown binary handling mode");
1445 }
83b5d2f5
JH
1446 }
1447
60ecac98 1448 memset(&xecfg, 0, sizeof(xecfg));
0579f91d
TR
1449 opt->priv = &xecfg;
1450
a26345b6 1451 try_lookahead = should_lookahead(opt);
60ecac98 1452
335ec3bf 1453 if (fill_textconv_grep(textconv, gs) < 0)
08265798
JK
1454 return 0;
1455
e1327023
JK
1456 bol = gs->buf;
1457 left = gs->size;
83b5d2f5
JH
1458 while (left) {
1459 char *eol, ch;
0ab7befa 1460 int hit;
83b5d2f5 1461
a26345b6 1462 /*
8997da38 1463 * look_ahead() skips quickly to the line that possibly
a26345b6
JH
1464 * has the next hit; don't call it if we need to do
1465 * something more than just skipping the current line
1466 * in response to an unmatch for the current line. E.g.
1467 * inside a post-context window, we will show the current
1468 * line as a context around the previous hit when it
1469 * doesn't hit.
1470 */
1471 if (try_lookahead
1472 && !(last_hit
ba8ea749
RS
1473 && (show_function ||
1474 lno <= last_hit + opt->post_context))
a26345b6
JH
1475 && look_ahead(opt, &left, &lno, &bol))
1476 break;
83b5d2f5
JH
1477 eol = end_of_line(bol, &left);
1478 ch = *eol;
1479 *eol = 0;
1480
480c1ca6
JH
1481 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1482 ctx = GREP_CONTEXT_BODY;
1483
0ab7befa 1484 hit = match_line(opt, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1485 *eol = ch;
1486
0ab7befa
JH
1487 if (collect_hits)
1488 goto next_line;
1489
83b5d2f5
JH
1490 /* "grep -v -e foo -e bla" should list lines
1491 * that do not have either, so inversion should
1492 * be done outside.
1493 */
1494 if (opt->invert)
1495 hit = !hit;
1496 if (opt->unmatch_name_only) {
1497 if (hit)
1498 return 0;
1499 goto next_line;
1500 }
1501 if (hit) {
1502 count++;
1503 if (opt->status_only)
1504 return 1;
321ffcc0 1505 if (opt->name_only) {
e1327023 1506 show_name(opt, gs->name);
321ffcc0
RS
1507 return 1;
1508 }
c30c10cf
RS
1509 if (opt->count)
1510 goto next_line;
83b5d2f5 1511 if (binary_match_only) {
5b594f45 1512 opt->output(opt, "Binary file ", 12);
e1327023 1513 output_color(opt, gs->name, strlen(gs->name),
55f638bd 1514 opt->color_filename);
5b594f45 1515 opt->output(opt, " matches\n", 9);
83b5d2f5
JH
1516 return 1;
1517 }
83b5d2f5
JH
1518 /* Hit at this line. If we haven't shown the
1519 * pre-context lines, we would need to show them.
83b5d2f5 1520 */
ba8ea749 1521 if (opt->pre_context || opt->funcbody)
e1327023 1522 show_pre_context(opt, gs, bol, eol, lno);
2944e4e6 1523 else if (opt->funcname)
e1327023
JK
1524 show_funcname_line(opt, gs, bol, lno);
1525 show_line(opt, bol, eol, gs->name, lno, ':');
5dd06d38 1526 last_hit = lno;
ba8ea749
RS
1527 if (opt->funcbody)
1528 show_function = 1;
1529 goto next_line;
83b5d2f5 1530 }
e1327023 1531 if (show_function && match_funcname(opt, gs, bol, eol))
ba8ea749
RS
1532 show_function = 0;
1533 if (show_function ||
1534 (last_hit && lno <= last_hit + opt->post_context)) {
83b5d2f5
JH
1535 /* If the last hit is within the post context,
1536 * we need to show this line.
1537 */
e1327023 1538 show_line(opt, bol, eol, gs->name, lno, '-');
83b5d2f5 1539 }
83b5d2f5
JH
1540
1541 next_line:
1542 bol = eol + 1;
1543 if (!left)
1544 break;
1545 left--;
1546 lno++;
1547 }
1548
0ab7befa
JH
1549 if (collect_hits)
1550 return 0;
b48fb5b6 1551
83b5d2f5
JH
1552 if (opt->status_only)
1553 return 0;
1554 if (opt->unmatch_name_only) {
1555 /* We did not see any hit, so we want to show this */
e1327023 1556 show_name(opt, gs->name);
83b5d2f5
JH
1557 return 1;
1558 }
1559
60ecac98
RS
1560 xdiff_clear_find_func(&xecfg);
1561 opt->priv = NULL;
1562
83b5d2f5
JH
1563 /* NEEDSWORK:
1564 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1565 * which feels mostly useless but sometimes useful. Maybe
1566 * make it another option? For now suppress them.
1567 */
5b594f45
FK
1568 if (opt->count && count) {
1569 char buf[32];
f76d947a
RS
1570 if (opt->pathname) {
1571 output_color(opt, gs->name, strlen(gs->name),
1572 opt->color_filename);
1573 output_sep(opt, ':');
1574 }
55f638bd 1575 snprintf(buf, sizeof(buf), "%u\n", count);
5b594f45 1576 opt->output(opt, buf, strlen(buf));
c30c10cf 1577 return 1;
5b594f45 1578 }
83b5d2f5
JH
1579 return !!last_hit;
1580}
1581
0ab7befa
JH
1582static void clr_hit_marker(struct grep_expr *x)
1583{
1584 /* All-hit markers are meaningful only at the very top level
1585 * OR node.
1586 */
1587 while (1) {
1588 x->hit = 0;
1589 if (x->node != GREP_NODE_OR)
1590 return;
1591 x->u.binary.left->hit = 0;
1592 x = x->u.binary.right;
1593 }
1594}
1595
1596static int chk_hit_marker(struct grep_expr *x)
1597{
1598 /* Top level nodes have hit markers. See if they all are hits */
1599 while (1) {
1600 if (x->node != GREP_NODE_OR)
1601 return x->hit;
1602 if (!x->u.binary.left->hit)
1603 return 0;
1604 x = x->u.binary.right;
1605 }
1606}
1607
e1327023 1608int grep_source(struct grep_opt *opt, struct grep_source *gs)
0ab7befa
JH
1609{
1610 /*
1611 * we do not have to do the two-pass grep when we do not check
1612 * buffer-wide "all-match".
1613 */
1614 if (!opt->all_match)
e1327023 1615 return grep_source_1(opt, gs, 0);
0ab7befa
JH
1616
1617 /* Otherwise the toplevel "or" terms hit a bit differently.
1618 * We first clear hit markers from them.
1619 */
1620 clr_hit_marker(opt->pattern_expression);
e1327023 1621 grep_source_1(opt, gs, 1);
0ab7befa
JH
1622
1623 if (!chk_hit_marker(opt->pattern_expression))
1624 return 0;
1625
e1327023
JK
1626 return grep_source_1(opt, gs, 0);
1627}
1628
c876d6da 1629int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
e1327023
JK
1630{
1631 struct grep_source gs;
1632 int r;
1633
55c61688 1634 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
e1327023
JK
1635 gs.buf = buf;
1636 gs.size = size;
1637
1638 r = grep_source(opt, &gs);
1639
1640 grep_source_clear(&gs);
1641 return r;
1642}
1643
1644void grep_source_init(struct grep_source *gs, enum grep_source_type type,
55c61688
NTND
1645 const char *name, const char *path,
1646 const void *identifier)
e1327023
JK
1647{
1648 gs->type = type;
1649 gs->name = name ? xstrdup(name) : NULL;
55c61688 1650 gs->path = path ? xstrdup(path) : NULL;
e1327023
JK
1651 gs->buf = NULL;
1652 gs->size = 0;
94ad9d9e 1653 gs->driver = NULL;
e1327023
JK
1654
1655 switch (type) {
1656 case GREP_SOURCE_FILE:
1657 gs->identifier = xstrdup(identifier);
1658 break;
1659 case GREP_SOURCE_SHA1:
1660 gs->identifier = xmalloc(20);
50546b15 1661 hashcpy(gs->identifier, identifier);
e1327023
JK
1662 break;
1663 case GREP_SOURCE_BUF:
1664 gs->identifier = NULL;
1665 }
1666}
1667
1668void grep_source_clear(struct grep_source *gs)
1669{
1670 free(gs->name);
1671 gs->name = NULL;
55c61688
NTND
1672 free(gs->path);
1673 gs->path = NULL;
e1327023
JK
1674 free(gs->identifier);
1675 gs->identifier = NULL;
1676 grep_source_clear_data(gs);
1677}
1678
1679void grep_source_clear_data(struct grep_source *gs)
1680{
1681 switch (gs->type) {
1682 case GREP_SOURCE_FILE:
1683 case GREP_SOURCE_SHA1:
1684 free(gs->buf);
1685 gs->buf = NULL;
1686 gs->size = 0;
1687 break;
1688 case GREP_SOURCE_BUF:
1689 /* leave user-provided buf intact */
1690 break;
1691 }
1692}
1693
1694static int grep_source_load_sha1(struct grep_source *gs)
1695{
1696 enum object_type type;
1697
1698 grep_read_lock();
1699 gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1700 grep_read_unlock();
1701
1702 if (!gs->buf)
1703 return error(_("'%s': unable to read %s"),
1704 gs->name,
1705 sha1_to_hex(gs->identifier));
1706 return 0;
1707}
1708
1709static int grep_source_load_file(struct grep_source *gs)
1710{
1711 const char *filename = gs->identifier;
1712 struct stat st;
1713 char *data;
1714 size_t size;
1715 int i;
1716
1717 if (lstat(filename, &st) < 0) {
1718 err_ret:
1719 if (errno != ENOENT)
1720 error(_("'%s': %s"), filename, strerror(errno));
1721 return -1;
1722 }
1723 if (!S_ISREG(st.st_mode))
1724 return -1;
1725 size = xsize_t(st.st_size);
1726 i = open(filename, O_RDONLY);
1727 if (i < 0)
1728 goto err_ret;
1729 data = xmalloc(size + 1);
1730 if (st.st_size != read_in_full(i, data, size)) {
1731 error(_("'%s': short read %s"), filename, strerror(errno));
1732 close(i);
1733 free(data);
1734 return -1;
1735 }
1736 close(i);
1737 data[size] = 0;
1738
1739 gs->buf = data;
1740 gs->size = size;
1741 return 0;
1742}
1743
3083301e 1744static int grep_source_load(struct grep_source *gs)
e1327023
JK
1745{
1746 if (gs->buf)
1747 return 0;
1748
1749 switch (gs->type) {
1750 case GREP_SOURCE_FILE:
1751 return grep_source_load_file(gs);
1752 case GREP_SOURCE_SHA1:
1753 return grep_source_load_sha1(gs);
1754 case GREP_SOURCE_BUF:
1755 return gs->buf ? 0 : -1;
1756 }
1757 die("BUG: invalid grep_source type");
0ab7befa 1758}
94ad9d9e
JK
1759
1760void grep_source_load_driver(struct grep_source *gs)
1761{
1762 if (gs->driver)
1763 return;
1764
1765 grep_attr_lock();
55c61688
NTND
1766 if (gs->path)
1767 gs->driver = userdiff_find_by_path(gs->path);
94ad9d9e
JK
1768 if (!gs->driver)
1769 gs->driver = userdiff_find_by_name("default");
1770 grep_attr_unlock();
1771}
41b59bfc 1772
3083301e 1773static int grep_source_is_binary(struct grep_source *gs)
41b59bfc
JK
1774{
1775 grep_source_load_driver(gs);
1776 if (gs->driver->binary != -1)
1777 return gs->driver->binary;
1778
1779 if (!grep_source_load(gs))
1780 return buffer_is_binary(gs->buf, gs->size);
1781
1782 return 0;
1783}