t7008: demonstrate behavior of grep with textconv
[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"
83b5d2f5 5
07a7d656
JH
6static int grep_source_load(struct grep_source *gs);
7static int grep_source_is_binary(struct grep_source *gs);
8
7687a054
JH
9static struct grep_opt grep_defaults;
10
11/*
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...
15 */
16void init_grep_defaults(void)
17{
18 struct grep_opt *opt = &grep_defaults;
918d4e1c
JH
19 static int run_once;
20
21 if (run_once)
22 return;
23 run_once++;
7687a054
JH
24
25 memset(opt, 0, sizeof(*opt));
26 opt->relative = 1;
27 opt->pathname = 1;
28 opt->regflags = REG_NEWLINE;
29 opt->max_depth = -1;
30 opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
31 opt->extended_regexp_option = 0;
32 strcpy(opt->color_context, "");
33 strcpy(opt->color_filename, "");
34 strcpy(opt->color_function, "");
35 strcpy(opt->color_lineno, "");
36 strcpy(opt->color_match, GIT_COLOR_BOLD_RED);
37 strcpy(opt->color_selected, "");
38 strcpy(opt->color_sep, GIT_COLOR_CYAN);
39 opt->color = -1;
40}
41
42static int parse_pattern_type_arg(const char *opt, const char *arg)
43{
44 if (!strcmp(arg, "default"))
45 return GREP_PATTERN_TYPE_UNSPECIFIED;
46 else if (!strcmp(arg, "basic"))
47 return GREP_PATTERN_TYPE_BRE;
48 else if (!strcmp(arg, "extended"))
49 return GREP_PATTERN_TYPE_ERE;
50 else if (!strcmp(arg, "fixed"))
51 return GREP_PATTERN_TYPE_FIXED;
52 else if (!strcmp(arg, "perl"))
53 return GREP_PATTERN_TYPE_PCRE;
54 die("bad %s argument: %s", opt, arg);
55}
56
57/*
58 * Read the configuration file once and store it in
59 * the grep_defaults template.
60 */
61int grep_config(const char *var, const char *value, void *cb)
62{
63 struct grep_opt *opt = &grep_defaults;
64 char *color = NULL;
65
66 if (userdiff_config(var, value) < 0)
67 return -1;
68
69 if (!strcmp(var, "grep.extendedregexp")) {
70 if (git_config_bool(var, value))
71 opt->extended_regexp_option = 1;
72 else
73 opt->extended_regexp_option = 0;
74 return 0;
75 }
76
77 if (!strcmp(var, "grep.patterntype")) {
78 opt->pattern_type_option = parse_pattern_type_arg(var, value);
79 return 0;
80 }
81
82 if (!strcmp(var, "grep.linenumber")) {
83 opt->linenum = git_config_bool(var, value);
84 return 0;
85 }
86
87 if (!strcmp(var, "color.grep"))
88 opt->color = git_config_colorbool(var, value);
89 else if (!strcmp(var, "color.grep.context"))
90 color = opt->color_context;
91 else if (!strcmp(var, "color.grep.filename"))
92 color = opt->color_filename;
93 else if (!strcmp(var, "color.grep.function"))
94 color = opt->color_function;
95 else if (!strcmp(var, "color.grep.linenumber"))
96 color = opt->color_lineno;
97 else if (!strcmp(var, "color.grep.match"))
98 color = opt->color_match;
99 else if (!strcmp(var, "color.grep.selected"))
100 color = opt->color_selected;
101 else if (!strcmp(var, "color.grep.separator"))
102 color = opt->color_sep;
103
104 if (color) {
105 if (!value)
106 return config_error_nonbool(var);
107 color_parse(value, var, color);
108 }
109 return 0;
110}
111
112/*
113 * Initialize one instance of grep_opt and copy the
114 * default values from the template we read the configuration
115 * information in an earlier call to git_config(grep_config).
116 */
117void grep_init(struct grep_opt *opt, const char *prefix)
118{
119 struct grep_opt *def = &grep_defaults;
120
121 memset(opt, 0, sizeof(*opt));
122 opt->prefix = prefix;
123 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
124 opt->pattern_tail = &opt->pattern_list;
125 opt->header_tail = &opt->header_list;
126
127 opt->color = def->color;
128 opt->extended_regexp_option = def->extended_regexp_option;
129 opt->pattern_type_option = def->pattern_type_option;
130 opt->linenum = def->linenum;
131 opt->max_depth = def->max_depth;
132 opt->pathname = def->pathname;
133 opt->regflags = def->regflags;
134 opt->relative = def->relative;
135
136 strcpy(opt->color_context, def->color_context);
137 strcpy(opt->color_filename, def->color_filename);
138 strcpy(opt->color_function, def->color_function);
139 strcpy(opt->color_lineno, def->color_lineno);
140 strcpy(opt->color_match, def->color_match);
141 strcpy(opt->color_selected, def->color_selected);
142 strcpy(opt->color_sep, def->color_sep);
143}
07a7d656 144
c5c31d33
JH
145void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
146{
147 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
148 grep_set_pattern_type_option(pattern_type, opt);
149 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
150 grep_set_pattern_type_option(opt->pattern_type_option, opt);
151 else if (opt->extended_regexp_option)
152 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
153}
154
155void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
156{
157 switch (pattern_type) {
158 case GREP_PATTERN_TYPE_UNSPECIFIED:
159 /* fall through */
160
161 case GREP_PATTERN_TYPE_BRE:
162 opt->fixed = 0;
163 opt->pcre = 0;
164 opt->regflags &= ~REG_EXTENDED;
165 break;
166
167 case GREP_PATTERN_TYPE_ERE:
168 opt->fixed = 0;
169 opt->pcre = 0;
170 opt->regflags |= REG_EXTENDED;
171 break;
172
173 case GREP_PATTERN_TYPE_FIXED:
174 opt->fixed = 1;
175 opt->pcre = 0;
176 opt->regflags &= ~REG_EXTENDED;
177 break;
178
179 case GREP_PATTERN_TYPE_PCRE:
180 opt->fixed = 0;
181 opt->pcre = 1;
182 opt->regflags &= ~REG_EXTENDED;
183 break;
184 }
185}
186
fc456751
RS
187static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
188 const char *origin, int no,
189 enum grep_pat_token t,
190 enum grep_header_field field)
a4d7d2c6
JH
191{
192 struct grep_pat *p = xcalloc(1, sizeof(*p));
526a858a 193 p->pattern = xmemdupz(pat, patlen);
fc456751
RS
194 p->patternlen = patlen;
195 p->origin = origin;
196 p->no = no;
197 p->token = t;
a4d7d2c6 198 p->field = field;
fc456751
RS
199 return p;
200}
201
2b3873ff
RS
202static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
203{
204 **tail = p;
205 *tail = &p->next;
a4d7d2c6 206 p->next = NULL;
526a858a
RS
207
208 switch (p->token) {
209 case GREP_PATTERN: /* atom */
210 case GREP_PATTERN_HEAD:
211 case GREP_PATTERN_BODY:
212 for (;;) {
213 struct grep_pat *new_pat;
214 size_t len = 0;
215 char *cp = p->pattern + p->patternlen, *nl = NULL;
216 while (++len <= p->patternlen) {
217 if (*(--cp) == '\n') {
218 nl = cp;
219 break;
220 }
221 }
222 if (!nl)
223 break;
224 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
225 p->no, p->token, p->field);
226 new_pat->next = p->next;
227 if (!p->next)
228 *tail = &new_pat->next;
229 p->next = new_pat;
230 *nl = '\0';
231 p->patternlen -= len;
232 }
233 break;
234 default:
235 break;
236 }
2b3873ff
RS
237}
238
fc456751
RS
239void append_header_grep_pattern(struct grep_opt *opt,
240 enum grep_header_field field, const char *pat)
241{
242 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
243 GREP_PATTERN_HEAD, field);
baa6378f
JH
244 if (field == GREP_HEADER_REFLOG)
245 opt->use_reflog_filter = 1;
2b3873ff 246 do_append_grep_pat(&opt->header_tail, p);
a4d7d2c6
JH
247}
248
83b5d2f5
JH
249void append_grep_pattern(struct grep_opt *opt, const char *pat,
250 const char *origin, int no, enum grep_pat_token t)
ed40a095
RS
251{
252 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
253}
254
255void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
256 const char *origin, int no, enum grep_pat_token t)
83b5d2f5 257{
fc456751 258 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
2b3873ff 259 do_append_grep_pat(&opt->pattern_tail, p);
83b5d2f5
JH
260}
261
5b594f45
FK
262struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
263{
264 struct grep_pat *pat;
265 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
266 *ret = *opt;
267
268 ret->pattern_list = NULL;
269 ret->pattern_tail = &ret->pattern_list;
270
271 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
272 {
273 if(pat->token == GREP_PATTERN_HEAD)
274 append_header_grep_pattern(ret, pat->field,
275 pat->pattern);
276 else
ed40a095
RS
277 append_grep_pat(ret, pat->pattern, pat->patternlen,
278 pat->origin, pat->no, pat->token);
5b594f45
FK
279 }
280
281 return ret;
282}
283
a30c148a
MK
284static NORETURN void compile_regexp_failed(const struct grep_pat *p,
285 const char *error)
286{
287 char where[1024];
288
289 if (p->no)
290 sprintf(where, "In '%s' at %d, ", p->origin, p->no);
291 else if (p->origin)
292 sprintf(where, "%s, ", p->origin);
293 else
294 where[0] = 0;
295
296 die("%s'%s': %s", where, p->pattern, error);
297}
298
63e7e9d8
MK
299#ifdef USE_LIBPCRE
300static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
301{
302 const char *error;
303 int erroffset;
fba4f125 304 int options = PCRE_MULTILINE;
63e7e9d8
MK
305
306 if (opt->ignore_case)
307 options |= PCRE_CASELESS;
308
309 p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
310 NULL);
311 if (!p->pcre_regexp)
312 compile_regexp_failed(p, error);
313
314 p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
315 if (!p->pcre_extra_info && error)
316 die("%s", error);
317}
318
319static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
320 regmatch_t *match, int eflags)
321{
322 int ovector[30], ret, flags = 0;
323
324 if (eflags & REG_NOTBOL)
325 flags |= PCRE_NOTBOL;
326
327 ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
328 0, flags, ovector, ARRAY_SIZE(ovector));
329 if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
330 die("pcre_exec failed with error code %d", ret);
331 if (ret > 0) {
332 ret = 0;
333 match->rm_so = ovector[0];
334 match->rm_eo = ovector[1];
335 }
336
337 return ret;
338}
339
340static void free_pcre_regexp(struct grep_pat *p)
341{
342 pcre_free(p->pcre_regexp);
343 pcre_free(p->pcre_extra_info);
344}
345#else /* !USE_LIBPCRE */
346static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
347{
348 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
349}
350
351static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
352 regmatch_t *match, int eflags)
353{
354 return 1;
355}
356
357static void free_pcre_regexp(struct grep_pat *p)
358{
359}
360#endif /* !USE_LIBPCRE */
361
9eceddee
FK
362static int is_fixed(const char *s, size_t len)
363{
364 size_t i;
365
366 /* regcomp cannot accept patterns with NULs so we
367 * consider any pattern containing a NUL fixed.
368 */
369 if (memchr(s, 0, len))
370 return 1;
371
372 for (i = 0; i < len; i++) {
373 if (is_regex_special(s[i]))
374 return 0;
375 }
376
377 return 1;
378}
379
83b5d2f5
JH
380static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
381{
c822255c
RS
382 int err;
383
d7eb527d 384 p->word_regexp = opt->word_regexp;
5183bf67 385 p->ignore_case = opt->ignore_case;
d7eb527d 386
9eceddee
FK
387 if (opt->fixed || is_fixed(p->pattern, p->patternlen))
388 p->fixed = 1;
389 else
390 p->fixed = 0;
391
392 if (p->fixed) {
0f871cf5
JH
393 if (opt->regflags & REG_ICASE || p->ignore_case)
394 p->kws = kwsalloc(tolower_trans_tbl);
395 else
9eceddee 396 p->kws = kwsalloc(NULL);
9eceddee
FK
397 kwsincr(p->kws, p->pattern, p->patternlen);
398 kwsprep(p->kws);
c822255c 399 return;
9eceddee 400 }
c822255c 401
63e7e9d8
MK
402 if (opt->pcre) {
403 compile_pcre_regexp(p, opt);
404 return;
405 }
406
c822255c 407 err = regcomp(&p->regexp, p->pattern, opt->regflags);
83b5d2f5
JH
408 if (err) {
409 char errbuf[1024];
83b5d2f5
JH
410 regerror(err, &p->regexp, errbuf, 1024);
411 regfree(&p->regexp);
a30c148a 412 compile_regexp_failed(p, errbuf);
83b5d2f5
JH
413 }
414}
415
0ab7befa 416static struct grep_expr *compile_pattern_or(struct grep_pat **);
83b5d2f5
JH
417static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
418{
419 struct grep_pat *p;
420 struct grep_expr *x;
421
422 p = *list;
c922b01f
LT
423 if (!p)
424 return NULL;
83b5d2f5
JH
425 switch (p->token) {
426 case GREP_PATTERN: /* atom */
480c1ca6
JH
427 case GREP_PATTERN_HEAD:
428 case GREP_PATTERN_BODY:
83b5d2f5
JH
429 x = xcalloc(1, sizeof (struct grep_expr));
430 x->node = GREP_NODE_ATOM;
431 x->u.atom = p;
432 *list = p->next;
433 return x;
434 case GREP_OPEN_PAREN:
435 *list = p->next;
0ab7befa 436 x = compile_pattern_or(list);
83b5d2f5
JH
437 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
438 die("unmatched parenthesis");
439 *list = (*list)->next;
440 return x;
441 default:
442 return NULL;
443 }
444}
445
446static struct grep_expr *compile_pattern_not(struct grep_pat **list)
447{
448 struct grep_pat *p;
449 struct grep_expr *x;
450
451 p = *list;
c922b01f
LT
452 if (!p)
453 return NULL;
83b5d2f5
JH
454 switch (p->token) {
455 case GREP_NOT:
456 if (!p->next)
457 die("--not not followed by pattern expression");
458 *list = p->next;
459 x = xcalloc(1, sizeof (struct grep_expr));
460 x->node = GREP_NODE_NOT;
461 x->u.unary = compile_pattern_not(list);
462 if (!x->u.unary)
463 die("--not followed by non pattern expression");
464 return x;
465 default:
466 return compile_pattern_atom(list);
467 }
468}
469
470static struct grep_expr *compile_pattern_and(struct grep_pat **list)
471{
472 struct grep_pat *p;
473 struct grep_expr *x, *y, *z;
474
475 x = compile_pattern_not(list);
476 p = *list;
477 if (p && p->token == GREP_AND) {
478 if (!p->next)
479 die("--and not followed by pattern expression");
480 *list = p->next;
481 y = compile_pattern_and(list);
482 if (!y)
483 die("--and not followed by pattern expression");
484 z = xcalloc(1, sizeof (struct grep_expr));
485 z->node = GREP_NODE_AND;
486 z->u.binary.left = x;
487 z->u.binary.right = y;
488 return z;
489 }
490 return x;
491}
492
493static struct grep_expr *compile_pattern_or(struct grep_pat **list)
494{
495 struct grep_pat *p;
496 struct grep_expr *x, *y, *z;
497
498 x = compile_pattern_and(list);
499 p = *list;
500 if (x && p && p->token != GREP_CLOSE_PAREN) {
501 y = compile_pattern_or(list);
502 if (!y)
503 die("not a pattern expression %s", p->pattern);
504 z = xcalloc(1, sizeof (struct grep_expr));
505 z->node = GREP_NODE_OR;
506 z->u.binary.left = x;
507 z->u.binary.right = y;
508 return z;
509 }
510 return x;
511}
512
513static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
514{
515 return compile_pattern_or(list);
516}
517
17bf35a3
JH
518static void indent(int in)
519{
520 while (in-- > 0)
521 fputc(' ', stderr);
522}
523
524static void dump_grep_pat(struct grep_pat *p)
525{
526 switch (p->token) {
527 case GREP_AND: fprintf(stderr, "*and*"); break;
528 case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
529 case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
530 case GREP_NOT: fprintf(stderr, "*not*"); break;
531 case GREP_OR: fprintf(stderr, "*or*"); break;
532
533 case GREP_PATTERN: fprintf(stderr, "pattern"); break;
534 case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
535 case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
536 }
537
538 switch (p->token) {
539 default: break;
540 case GREP_PATTERN_HEAD:
541 fprintf(stderr, "<head %d>", p->field); break;
542 case GREP_PATTERN_BODY:
543 fprintf(stderr, "<body>"); break;
544 }
545 switch (p->token) {
546 default: break;
547 case GREP_PATTERN_HEAD:
548 case GREP_PATTERN_BODY:
549 case GREP_PATTERN:
550 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
551 break;
552 }
553 fputc('\n', stderr);
554}
555
556static void dump_grep_expression_1(struct grep_expr *x, int in)
557{
558 indent(in);
559 switch (x->node) {
560 case GREP_NODE_TRUE:
561 fprintf(stderr, "true\n");
562 break;
563 case GREP_NODE_ATOM:
564 dump_grep_pat(x->u.atom);
565 break;
566 case GREP_NODE_NOT:
567 fprintf(stderr, "(not\n");
568 dump_grep_expression_1(x->u.unary, in+1);
569 indent(in);
570 fprintf(stderr, ")\n");
571 break;
572 case GREP_NODE_AND:
573 fprintf(stderr, "(and\n");
574 dump_grep_expression_1(x->u.binary.left, in+1);
575 dump_grep_expression_1(x->u.binary.right, in+1);
576 indent(in);
577 fprintf(stderr, ")\n");
578 break;
579 case GREP_NODE_OR:
580 fprintf(stderr, "(or\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 }
587}
588
07a7d656 589static void dump_grep_expression(struct grep_opt *opt)
17bf35a3
JH
590{
591 struct grep_expr *x = opt->pattern_expression;
592
593 if (opt->all_match)
594 fprintf(stderr, "[all-match]\n");
595 dump_grep_expression_1(x, 0);
596 fflush(NULL);
597}
598
5aaeb733
JH
599static struct grep_expr *grep_true_expr(void)
600{
601 struct grep_expr *z = xcalloc(1, sizeof(*z));
602 z->node = GREP_NODE_TRUE;
603 return z;
604}
605
606static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
607{
608 struct grep_expr *z = xcalloc(1, sizeof(*z));
609 z->node = GREP_NODE_OR;
610 z->u.binary.left = left;
611 z->u.binary.right = right;
612 return z;
613}
614
95ce9ce2 615static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
83b5d2f5
JH
616{
617 struct grep_pat *p;
95ce9ce2 618 struct grep_expr *header_expr;
5aaeb733
JH
619 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
620 enum grep_header_field fld;
83b5d2f5 621
95ce9ce2
JH
622 if (!opt->header_list)
623 return NULL;
2385f246 624
95ce9ce2
JH
625 for (p = opt->header_list; p; p = p->next) {
626 if (p->token != GREP_PATTERN_HEAD)
627 die("bug: a non-header pattern in grep header list.");
3ce3ffb8
AP
628 if (p->field < GREP_HEADER_FIELD_MIN ||
629 GREP_HEADER_FIELD_MAX <= p->field)
95ce9ce2
JH
630 die("bug: unknown header field %d", p->field);
631 compile_regexp(p, opt);
80235ba7 632 }
5aaeb733
JH
633
634 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
635 header_group[fld] = NULL;
636
637 for (p = opt->header_list; p; p = p->next) {
638 struct grep_expr *h;
639 struct grep_pat *pp = p;
640
641 h = compile_pattern_atom(&pp);
642 if (!h || pp != p->next)
643 die("bug: malformed header expr");
644 if (!header_group[p->field]) {
645 header_group[p->field] = h;
646 continue;
647 }
648 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
649 }
650
651 header_expr = NULL;
652
653 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
654 if (!header_group[fld])
655 continue;
656 if (!header_expr)
657 header_expr = grep_true_expr();
658 header_expr = grep_or_expr(header_group[fld], header_expr);
659 }
95ce9ce2
JH
660 return header_expr;
661}
662
13e4fc7e
JH
663static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
664{
665 struct grep_expr *z = x;
666
667 while (x) {
668 assert(x->node == GREP_NODE_OR);
669 if (x->u.binary.right &&
670 x->u.binary.right->node == GREP_NODE_TRUE) {
671 x->u.binary.right = y;
672 break;
673 }
674 x = x->u.binary.right;
675 }
676 return z;
677}
678
17bf35a3 679static void compile_grep_patterns_real(struct grep_opt *opt)
95ce9ce2
JH
680{
681 struct grep_pat *p;
682 struct grep_expr *header_expr = prep_header_patterns(opt);
0ab7befa 683
83b5d2f5 684 for (p = opt->pattern_list; p; p = p->next) {
480c1ca6
JH
685 switch (p->token) {
686 case GREP_PATTERN: /* atom */
687 case GREP_PATTERN_HEAD:
688 case GREP_PATTERN_BODY:
c822255c 689 compile_regexp(p, opt);
480c1ca6
JH
690 break;
691 default:
83b5d2f5 692 opt->extended = 1;
480c1ca6
JH
693 break;
694 }
83b5d2f5
JH
695 }
696
80235ba7
JH
697 if (opt->all_match || header_expr)
698 opt->extended = 1;
17bf35a3 699 else if (!opt->extended && !opt->debug)
83b5d2f5
JH
700 return;
701
83b5d2f5 702 p = opt->pattern_list;
ba150a3f
MB
703 if (p)
704 opt->pattern_expression = compile_pattern_expr(&p);
83b5d2f5
JH
705 if (p)
706 die("incomplete pattern expression: %s", p->pattern);
80235ba7
JH
707
708 if (!header_expr)
709 return;
710
5aaeb733 711 if (!opt->pattern_expression)
80235ba7 712 opt->pattern_expression = header_expr;
13e4fc7e
JH
713 else if (opt->all_match)
714 opt->pattern_expression = grep_splice_or(header_expr,
715 opt->pattern_expression);
5aaeb733
JH
716 else
717 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
718 header_expr);
80235ba7 719 opt->all_match = 1;
83b5d2f5
JH
720}
721
17bf35a3
JH
722void compile_grep_patterns(struct grep_opt *opt)
723{
724 compile_grep_patterns_real(opt);
725 if (opt->debug)
726 dump_grep_expression(opt);
727}
728
b48fb5b6
JH
729static void free_pattern_expr(struct grep_expr *x)
730{
731 switch (x->node) {
5aaeb733 732 case GREP_NODE_TRUE:
b48fb5b6
JH
733 case GREP_NODE_ATOM:
734 break;
735 case GREP_NODE_NOT:
736 free_pattern_expr(x->u.unary);
737 break;
738 case GREP_NODE_AND:
739 case GREP_NODE_OR:
740 free_pattern_expr(x->u.binary.left);
741 free_pattern_expr(x->u.binary.right);
742 break;
743 }
744 free(x);
745}
746
747void free_grep_patterns(struct grep_opt *opt)
748{
749 struct grep_pat *p, *n;
750
751 for (p = opt->pattern_list; p; p = n) {
752 n = p->next;
753 switch (p->token) {
754 case GREP_PATTERN: /* atom */
755 case GREP_PATTERN_HEAD:
756 case GREP_PATTERN_BODY:
9eceddee
FK
757 if (p->kws)
758 kwsfree(p->kws);
759 else if (p->pcre_regexp)
63e7e9d8
MK
760 free_pcre_regexp(p);
761 else
762 regfree(&p->regexp);
526a858a 763 free(p->pattern);
b48fb5b6
JH
764 break;
765 default:
766 break;
767 }
768 free(p);
769 }
770
771 if (!opt->extended)
772 return;
773 free_pattern_expr(opt->pattern_expression);
774}
775
83b5d2f5
JH
776static char *end_of_line(char *cp, unsigned long *left)
777{
778 unsigned long l = *left;
779 while (l && *cp != '\n') {
780 l--;
781 cp++;
782 }
783 *left = l;
784 return cp;
785}
786
787static int word_char(char ch)
788{
789 return isalnum(ch) || ch == '_';
790}
791
55f638bd
ML
792static void output_color(struct grep_opt *opt, const void *data, size_t size,
793 const char *color)
794{
daa0c3d9 795 if (want_color(opt->color) && color && color[0]) {
55f638bd
ML
796 opt->output(opt, color, strlen(color));
797 opt->output(opt, data, size);
798 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
799 } else
800 opt->output(opt, data, size);
801}
802
803static void output_sep(struct grep_opt *opt, char sign)
804{
805 if (opt->null_following_name)
806 opt->output(opt, "\0", 1);
807 else
808 output_color(opt, &sign, 1, opt->color_sep);
809}
810
83caecca
RZ
811static void show_name(struct grep_opt *opt, const char *name)
812{
55f638bd 813 output_color(opt, name, strlen(name), opt->color_filename);
5b594f45 814 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
83caecca
RZ
815}
816
ed40a095
RS
817static int fixmatch(struct grep_pat *p, char *line, char *eol,
818 regmatch_t *match)
83b5d2f5 819{
9eceddee
FK
820 struct kwsmatch kwsm;
821 size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
822 if (offset == -1) {
83b5d2f5
JH
823 match->rm_so = match->rm_eo = -1;
824 return REG_NOMATCH;
9eceddee
FK
825 } else {
826 match->rm_so = offset;
827 match->rm_eo = match->rm_so + kwsm.size[0];
83b5d2f5
JH
828 return 0;
829 }
830}
831
f96e5673
RS
832static int regmatch(const regex_t *preg, char *line, char *eol,
833 regmatch_t *match, int eflags)
834{
835#ifdef REG_STARTEND
836 match->rm_so = 0;
837 match->rm_eo = eol - line;
838 eflags |= REG_STARTEND;
839#endif
840 return regexec(preg, line, 1, match, eflags);
841}
842
97e77784
MK
843static int patmatch(struct grep_pat *p, char *line, char *eol,
844 regmatch_t *match, int eflags)
845{
846 int hit;
847
848 if (p->fixed)
849 hit = !fixmatch(p, line, eol, match);
63e7e9d8
MK
850 else if (p->pcre_regexp)
851 hit = !pcrematch(p, line, eol, match, eflags);
97e77784
MK
852 else
853 hit = !regmatch(&p->regexp, line, eol, match, eflags);
854
855 return hit;
856}
857
a4d7d2c6
JH
858static int strip_timestamp(char *bol, char **eol_p)
859{
860 char *eol = *eol_p;
861 int ch;
862
863 while (bol < --eol) {
864 if (*eol != '>')
865 continue;
866 *eol_p = ++eol;
867 ch = *eol;
868 *eol = '\0';
869 return ch;
870 }
871 return 0;
872}
873
874static struct {
875 const char *field;
876 size_t len;
877} header_field[] = {
878 { "author ", 7 },
879 { "committer ", 10 },
72fd13f7 880 { "reflog ", 7 },
a4d7d2c6
JH
881};
882
d7eb527d 883static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
79212772
RS
884 enum grep_context ctx,
885 regmatch_t *pmatch, int eflags)
83b5d2f5
JH
886{
887 int hit = 0;
a4d7d2c6 888 int saved_ch = 0;
e701fadb 889 const char *start = bol;
83b5d2f5 890
480c1ca6
JH
891 if ((p->token != GREP_PATTERN) &&
892 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
893 return 0;
894
a4d7d2c6
JH
895 if (p->token == GREP_PATTERN_HEAD) {
896 const char *field;
897 size_t len;
898 assert(p->field < ARRAY_SIZE(header_field));
899 field = header_field[p->field].field;
900 len = header_field[p->field].len;
901 if (strncmp(bol, field, len))
902 return 0;
903 bol += len;
ad4813b3
NTND
904 switch (p->field) {
905 case GREP_HEADER_AUTHOR:
906 case GREP_HEADER_COMMITTER:
907 saved_ch = strip_timestamp(bol, &eol);
908 break;
909 default:
910 break;
911 }
a4d7d2c6
JH
912 }
913
83b5d2f5 914 again:
97e77784 915 hit = patmatch(p, bol, eol, pmatch, eflags);
83b5d2f5 916
d7eb527d 917 if (hit && p->word_regexp) {
83b5d2f5 918 if ((pmatch[0].rm_so < 0) ||
84201eae 919 (eol - bol) < pmatch[0].rm_so ||
83b5d2f5
JH
920 (pmatch[0].rm_eo < 0) ||
921 (eol - bol) < pmatch[0].rm_eo)
922 die("regexp returned nonsense");
923
924 /* Match beginning must be either beginning of the
925 * line, or at word boundary (i.e. the last char must
926 * not be a word char). Similarly, match end must be
927 * either end of the line, or at word boundary
928 * (i.e. the next char must not be a word char).
929 */
fb62eb7f 930 if ( ((pmatch[0].rm_so == 0) ||
83b5d2f5
JH
931 !word_char(bol[pmatch[0].rm_so-1])) &&
932 ((pmatch[0].rm_eo == (eol-bol)) ||
933 !word_char(bol[pmatch[0].rm_eo])) )
934 ;
935 else
936 hit = 0;
937
84201eae
RS
938 /* Words consist of at least one character. */
939 if (pmatch->rm_so == pmatch->rm_eo)
940 hit = 0;
941
83b5d2f5
JH
942 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
943 /* There could be more than one match on the
944 * line, and the first match might not be
945 * strict word match. But later ones could be!
fb62eb7f
RS
946 * Forward to the next possible start, i.e. the
947 * next position following a non-word char.
83b5d2f5
JH
948 */
949 bol = pmatch[0].rm_so + bol + 1;
fb62eb7f
RS
950 while (word_char(bol[-1]) && bol < eol)
951 bol++;
dbb6a4ad 952 eflags |= REG_NOTBOL;
fb62eb7f
RS
953 if (bol < eol)
954 goto again;
83b5d2f5
JH
955 }
956 }
a4d7d2c6
JH
957 if (p->token == GREP_PATTERN_HEAD && saved_ch)
958 *eol = saved_ch;
e701fadb
RS
959 if (hit) {
960 pmatch[0].rm_so += bol - start;
961 pmatch[0].rm_eo += bol - start;
962 }
83b5d2f5
JH
963 return hit;
964}
965
d7eb527d
RS
966static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
967 enum grep_context ctx, int collect_hits)
83b5d2f5 968{
0ab7befa 969 int h = 0;
79212772 970 regmatch_t match;
0ab7befa 971
c922b01f
LT
972 if (!x)
973 die("Not a valid grep expression");
83b5d2f5 974 switch (x->node) {
5aaeb733
JH
975 case GREP_NODE_TRUE:
976 h = 1;
977 break;
83b5d2f5 978 case GREP_NODE_ATOM:
79212772 979 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
83b5d2f5
JH
980 break;
981 case GREP_NODE_NOT:
d7eb527d 982 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
0ab7befa 983 break;
83b5d2f5 984 case GREP_NODE_AND:
d7eb527d 985 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
252d560d 986 return 0;
d7eb527d 987 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
0ab7befa 988 break;
83b5d2f5 989 case GREP_NODE_OR:
0ab7befa 990 if (!collect_hits)
d7eb527d 991 return (match_expr_eval(x->u.binary.left,
0ab7befa 992 bol, eol, ctx, 0) ||
d7eb527d 993 match_expr_eval(x->u.binary.right,
0ab7befa 994 bol, eol, ctx, 0));
d7eb527d 995 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
0ab7befa 996 x->u.binary.left->hit |= h;
d7eb527d 997 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
0ab7befa
JH
998 break;
999 default:
d7530708 1000 die("Unexpected node type (internal error) %d", x->node);
83b5d2f5 1001 }
0ab7befa
JH
1002 if (collect_hits)
1003 x->hit |= h;
1004 return h;
83b5d2f5
JH
1005}
1006
480c1ca6 1007static int match_expr(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1008 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1009{
1010 struct grep_expr *x = opt->pattern_expression;
d7eb527d 1011 return match_expr_eval(x, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1012}
1013
480c1ca6 1014static int match_line(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1015 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1016{
1017 struct grep_pat *p;
79212772
RS
1018 regmatch_t match;
1019
83b5d2f5 1020 if (opt->extended)
0ab7befa
JH
1021 return match_expr(opt, bol, eol, ctx, collect_hits);
1022
1023 /* we do not call with collect_hits without being extended */
83b5d2f5 1024 for (p = opt->pattern_list; p; p = p->next) {
79212772 1025 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
83b5d2f5
JH
1026 return 1;
1027 }
1028 return 0;
1029}
1030
7e8f59d5
RS
1031static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1032 enum grep_context ctx,
1033 regmatch_t *pmatch, int eflags)
1034{
1035 regmatch_t match;
1036
1037 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1038 return 0;
1039 if (match.rm_so < 0 || match.rm_eo < 0)
1040 return 0;
1041 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1042 if (match.rm_so > pmatch->rm_so)
1043 return 1;
1044 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1045 return 1;
1046 }
1047 pmatch->rm_so = match.rm_so;
1048 pmatch->rm_eo = match.rm_eo;
1049 return 1;
1050}
1051
1052static int next_match(struct grep_opt *opt, char *bol, char *eol,
1053 enum grep_context ctx, regmatch_t *pmatch, int eflags)
1054{
1055 struct grep_pat *p;
1056 int hit = 0;
1057
1058 pmatch->rm_so = pmatch->rm_eo = -1;
1059 if (bol < eol) {
1060 for (p = opt->pattern_list; p; p = p->next) {
1061 switch (p->token) {
1062 case GREP_PATTERN: /* atom */
1063 case GREP_PATTERN_HEAD:
1064 case GREP_PATTERN_BODY:
1065 hit |= match_next_pattern(p, bol, eol, ctx,
1066 pmatch, eflags);
1067 break;
1068 default:
1069 break;
1070 }
1071 }
1072 }
1073 return hit;
1074}
1075
1076static void show_line(struct grep_opt *opt, char *bol, char *eol,
1077 const char *name, unsigned lno, char sign)
1078{
1079 int rest = eol - bol;
00588bb5 1080 char *line_color = NULL;
7e8f59d5 1081
a8f0e764
RS
1082 if (opt->file_break && opt->last_shown == 0) {
1083 if (opt->show_hunk_mark)
1084 opt->output(opt, "\n", 1);
ba8ea749 1085 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
046802d0 1086 if (opt->last_shown == 0) {
55f638bd
ML
1087 if (opt->show_hunk_mark) {
1088 output_color(opt, "--", 2, opt->color_sep);
1089 opt->output(opt, "\n", 1);
07b838f0 1090 }
55f638bd
ML
1091 } else if (lno > opt->last_shown + 1) {
1092 output_color(opt, "--", 2, opt->color_sep);
1093 opt->output(opt, "\n", 1);
1094 }
5dd06d38 1095 }
1d84f72e
RS
1096 if (opt->heading && opt->last_shown == 0) {
1097 output_color(opt, name, strlen(name), opt->color_filename);
1098 opt->output(opt, "\n", 1);
1099 }
5dd06d38
RS
1100 opt->last_shown = lno;
1101
1d84f72e 1102 if (!opt->heading && opt->pathname) {
55f638bd
ML
1103 output_color(opt, name, strlen(name), opt->color_filename);
1104 output_sep(opt, sign);
5b594f45
FK
1105 }
1106 if (opt->linenum) {
1107 char buf[32];
1108 snprintf(buf, sizeof(buf), "%d", lno);
55f638bd
ML
1109 output_color(opt, buf, strlen(buf), opt->color_lineno);
1110 output_sep(opt, sign);
5b594f45 1111 }
7e8f59d5
RS
1112 if (opt->color) {
1113 regmatch_t match;
1114 enum grep_context ctx = GREP_CONTEXT_BODY;
1115 int ch = *eol;
1116 int eflags = 0;
1117
00588bb5
ML
1118 if (sign == ':')
1119 line_color = opt->color_selected;
1120 else if (sign == '-')
1121 line_color = opt->color_context;
1122 else if (sign == '=')
1123 line_color = opt->color_function;
7e8f59d5
RS
1124 *eol = '\0';
1125 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1f5b9cc4
RS
1126 if (match.rm_so == match.rm_eo)
1127 break;
5b594f45 1128
00588bb5 1129 output_color(opt, bol, match.rm_so, line_color);
55f638bd
ML
1130 output_color(opt, bol + match.rm_so,
1131 match.rm_eo - match.rm_so,
1132 opt->color_match);
7e8f59d5
RS
1133 bol += match.rm_eo;
1134 rest -= match.rm_eo;
1135 eflags = REG_NOTBOL;
1136 }
1137 *eol = ch;
1138 }
00588bb5 1139 output_color(opt, bol, rest, line_color);
5b594f45 1140 opt->output(opt, "\n", 1);
7e8f59d5
RS
1141}
1142
0579f91d 1143#ifndef NO_PTHREADS
78db6ea9
JK
1144int grep_use_locks;
1145
0579f91d
TR
1146/*
1147 * This lock protects access to the gitattributes machinery, which is
1148 * not thread-safe.
1149 */
1150pthread_mutex_t grep_attr_mutex;
1151
78db6ea9 1152static inline void grep_attr_lock(void)
0579f91d 1153{
78db6ea9 1154 if (grep_use_locks)
0579f91d
TR
1155 pthread_mutex_lock(&grep_attr_mutex);
1156}
1157
78db6ea9 1158static inline void grep_attr_unlock(void)
0579f91d 1159{
78db6ea9 1160 if (grep_use_locks)
0579f91d
TR
1161 pthread_mutex_unlock(&grep_attr_mutex);
1162}
b3aeb285
JK
1163
1164/*
1165 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1166 */
1167pthread_mutex_t grep_read_mutex;
1168
0579f91d 1169#else
78db6ea9
JK
1170#define grep_attr_lock()
1171#define grep_attr_unlock()
0579f91d
TR
1172#endif
1173
e1327023 1174static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
2944e4e6 1175{
60ecac98 1176 xdemitconf_t *xecfg = opt->priv;
0579f91d 1177 if (xecfg && !xecfg->find_func) {
94ad9d9e
JK
1178 grep_source_load_driver(gs);
1179 if (gs->driver->funcname.pattern) {
1180 const struct userdiff_funcname *pe = &gs->driver->funcname;
0579f91d
TR
1181 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1182 } else {
1183 xecfg = opt->priv = NULL;
1184 }
1185 }
1186
1187 if (xecfg) {
60ecac98
RS
1188 char buf[1];
1189 return xecfg->find_func(bol, eol - bol, buf, 1,
1190 xecfg->find_func_priv) >= 0;
1191 }
1192
2944e4e6
RS
1193 if (bol == eol)
1194 return 0;
1195 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1196 return 1;
1197 return 0;
1198}
1199
e1327023
JK
1200static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1201 char *bol, unsigned lno)
2944e4e6 1202{
e1327023 1203 while (bol > gs->buf) {
2944e4e6
RS
1204 char *eol = --bol;
1205
e1327023 1206 while (bol > gs->buf && bol[-1] != '\n')
2944e4e6
RS
1207 bol--;
1208 lno--;
1209
1210 if (lno <= opt->last_shown)
1211 break;
1212
e1327023
JK
1213 if (match_funcname(opt, gs, bol, eol)) {
1214 show_line(opt, bol, eol, gs->name, lno, '=');
2944e4e6
RS
1215 break;
1216 }
1217 }
1218}
1219
e1327023 1220static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
ba8ea749 1221 char *bol, char *end, unsigned lno)
49de3216 1222{
2944e4e6 1223 unsigned cur = lno, from = 1, funcname_lno = 0;
ba8ea749
RS
1224 int funcname_needed = !!opt->funcname;
1225
e1327023 1226 if (opt->funcbody && !match_funcname(opt, gs, bol, end))
ba8ea749 1227 funcname_needed = 2;
49de3216
RS
1228
1229 if (opt->pre_context < lno)
1230 from = lno - opt->pre_context;
1231 if (from <= opt->last_shown)
1232 from = opt->last_shown + 1;
1233
1234 /* Rewind. */
e1327023 1235 while (bol > gs->buf &&
ba8ea749 1236 cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
2944e4e6
RS
1237 char *eol = --bol;
1238
e1327023 1239 while (bol > gs->buf && bol[-1] != '\n')
49de3216
RS
1240 bol--;
1241 cur--;
e1327023 1242 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
2944e4e6
RS
1243 funcname_lno = cur;
1244 funcname_needed = 0;
1245 }
49de3216
RS
1246 }
1247
2944e4e6
RS
1248 /* We need to look even further back to find a function signature. */
1249 if (opt->funcname && funcname_needed)
e1327023 1250 show_funcname_line(opt, gs, bol, cur);
2944e4e6 1251
49de3216
RS
1252 /* Back forward. */
1253 while (cur < lno) {
2944e4e6 1254 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
49de3216
RS
1255
1256 while (*eol != '\n')
1257 eol++;
e1327023 1258 show_line(opt, bol, eol, gs->name, cur, sign);
49de3216
RS
1259 bol = eol + 1;
1260 cur++;
1261 }
1262}
1263
a26345b6
JH
1264static int should_lookahead(struct grep_opt *opt)
1265{
1266 struct grep_pat *p;
1267
1268 if (opt->extended)
1269 return 0; /* punt for too complex stuff */
1270 if (opt->invert)
1271 return 0;
1272 for (p = opt->pattern_list; p; p = p->next) {
1273 if (p->token != GREP_PATTERN)
1274 return 0; /* punt for "header only" and stuff */
1275 }
1276 return 1;
1277}
1278
1279static int look_ahead(struct grep_opt *opt,
1280 unsigned long *left_p,
1281 unsigned *lno_p,
1282 char **bol_p)
1283{
1284 unsigned lno = *lno_p;
1285 char *bol = *bol_p;
1286 struct grep_pat *p;
1287 char *sp, *last_bol;
1288 regoff_t earliest = -1;
1289
1290 for (p = opt->pattern_list; p; p = p->next) {
1291 int hit;
1292 regmatch_t m;
1293
97e77784 1294 hit = patmatch(p, bol, bol + *left_p, &m, 0);
a26345b6
JH
1295 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1296 continue;
1297 if (earliest < 0 || m.rm_so < earliest)
1298 earliest = m.rm_so;
1299 }
1300
1301 if (earliest < 0) {
1302 *bol_p = bol + *left_p;
1303 *left_p = 0;
1304 return 1;
1305 }
1306 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1307 ; /* find the beginning of the line */
1308 last_bol = sp;
1309
1310 for (sp = bol; sp < last_bol; sp++) {
1311 if (*sp == '\n')
1312 lno++;
1313 }
1314 *left_p -= last_bol - bol;
1315 *bol_p = last_bol;
1316 *lno_p = lno;
1317 return 0;
1318}
1319
5b594f45
FK
1320static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1321{
1322 fwrite(buf, size, 1, stdout);
1323}
1324
e1327023 1325static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
83b5d2f5 1326{
e1327023
JK
1327 char *bol;
1328 unsigned long left;
83b5d2f5 1329 unsigned lno = 1;
83b5d2f5 1330 unsigned last_hit = 0;
83b5d2f5 1331 int binary_match_only = 0;
83b5d2f5 1332 unsigned count = 0;
a26345b6 1333 int try_lookahead = 0;
ba8ea749 1334 int show_function = 0;
480c1ca6 1335 enum grep_context ctx = GREP_CONTEXT_HEAD;
60ecac98 1336 xdemitconf_t xecfg;
83b5d2f5 1337
5b594f45
FK
1338 if (!opt->output)
1339 opt->output = std_output;
1340
ba8ea749
RS
1341 if (opt->pre_context || opt->post_context || opt->file_break ||
1342 opt->funcbody) {
08303c36
RS
1343 /* Show hunk marks, except for the first file. */
1344 if (opt->last_shown)
1345 opt->show_hunk_mark = 1;
1346 /*
1347 * If we're using threads then we can't easily identify
1348 * the first file. Always put hunk marks in that case
1349 * and skip the very first one later in work_done().
1350 */
1351 if (opt->output != std_output)
1352 opt->show_hunk_mark = 1;
1353 }
431d6e7b
RS
1354 opt->last_shown = 0;
1355
64fcec78
RS
1356 switch (opt->binary) {
1357 case GREP_BINARY_DEFAULT:
41b59bfc 1358 if (grep_source_is_binary(gs))
83b5d2f5 1359 binary_match_only = 1;
64fcec78
RS
1360 break;
1361 case GREP_BINARY_NOMATCH:
41b59bfc 1362 if (grep_source_is_binary(gs))
83b5d2f5 1363 return 0; /* Assume unmatch */
64fcec78
RS
1364 break;
1365 case GREP_BINARY_TEXT:
1366 break;
1367 default:
1368 die("bug: unknown binary handling mode");
83b5d2f5
JH
1369 }
1370
60ecac98 1371 memset(&xecfg, 0, sizeof(xecfg));
0579f91d
TR
1372 opt->priv = &xecfg;
1373
a26345b6 1374 try_lookahead = should_lookahead(opt);
60ecac98 1375
08265798
JK
1376 if (grep_source_load(gs) < 0)
1377 return 0;
1378
e1327023
JK
1379 bol = gs->buf;
1380 left = gs->size;
83b5d2f5
JH
1381 while (left) {
1382 char *eol, ch;
0ab7befa 1383 int hit;
83b5d2f5 1384
a26345b6 1385 /*
8997da38 1386 * look_ahead() skips quickly to the line that possibly
a26345b6
JH
1387 * has the next hit; don't call it if we need to do
1388 * something more than just skipping the current line
1389 * in response to an unmatch for the current line. E.g.
1390 * inside a post-context window, we will show the current
1391 * line as a context around the previous hit when it
1392 * doesn't hit.
1393 */
1394 if (try_lookahead
1395 && !(last_hit
ba8ea749
RS
1396 && (show_function ||
1397 lno <= last_hit + opt->post_context))
a26345b6
JH
1398 && look_ahead(opt, &left, &lno, &bol))
1399 break;
83b5d2f5
JH
1400 eol = end_of_line(bol, &left);
1401 ch = *eol;
1402 *eol = 0;
1403
480c1ca6
JH
1404 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1405 ctx = GREP_CONTEXT_BODY;
1406
0ab7befa 1407 hit = match_line(opt, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1408 *eol = ch;
1409
0ab7befa
JH
1410 if (collect_hits)
1411 goto next_line;
1412
83b5d2f5
JH
1413 /* "grep -v -e foo -e bla" should list lines
1414 * that do not have either, so inversion should
1415 * be done outside.
1416 */
1417 if (opt->invert)
1418 hit = !hit;
1419 if (opt->unmatch_name_only) {
1420 if (hit)
1421 return 0;
1422 goto next_line;
1423 }
1424 if (hit) {
1425 count++;
1426 if (opt->status_only)
1427 return 1;
321ffcc0 1428 if (opt->name_only) {
e1327023 1429 show_name(opt, gs->name);
321ffcc0
RS
1430 return 1;
1431 }
c30c10cf
RS
1432 if (opt->count)
1433 goto next_line;
83b5d2f5 1434 if (binary_match_only) {
5b594f45 1435 opt->output(opt, "Binary file ", 12);
e1327023 1436 output_color(opt, gs->name, strlen(gs->name),
55f638bd 1437 opt->color_filename);
5b594f45 1438 opt->output(opt, " matches\n", 9);
83b5d2f5
JH
1439 return 1;
1440 }
83b5d2f5
JH
1441 /* Hit at this line. If we haven't shown the
1442 * pre-context lines, we would need to show them.
83b5d2f5 1443 */
ba8ea749 1444 if (opt->pre_context || opt->funcbody)
e1327023 1445 show_pre_context(opt, gs, bol, eol, lno);
2944e4e6 1446 else if (opt->funcname)
e1327023
JK
1447 show_funcname_line(opt, gs, bol, lno);
1448 show_line(opt, bol, eol, gs->name, lno, ':');
5dd06d38 1449 last_hit = lno;
ba8ea749
RS
1450 if (opt->funcbody)
1451 show_function = 1;
1452 goto next_line;
83b5d2f5 1453 }
e1327023 1454 if (show_function && match_funcname(opt, gs, bol, eol))
ba8ea749
RS
1455 show_function = 0;
1456 if (show_function ||
1457 (last_hit && lno <= last_hit + opt->post_context)) {
83b5d2f5
JH
1458 /* If the last hit is within the post context,
1459 * we need to show this line.
1460 */
e1327023 1461 show_line(opt, bol, eol, gs->name, lno, '-');
83b5d2f5 1462 }
83b5d2f5
JH
1463
1464 next_line:
1465 bol = eol + 1;
1466 if (!left)
1467 break;
1468 left--;
1469 lno++;
1470 }
1471
0ab7befa
JH
1472 if (collect_hits)
1473 return 0;
b48fb5b6 1474
83b5d2f5
JH
1475 if (opt->status_only)
1476 return 0;
1477 if (opt->unmatch_name_only) {
1478 /* We did not see any hit, so we want to show this */
e1327023 1479 show_name(opt, gs->name);
83b5d2f5
JH
1480 return 1;
1481 }
1482
60ecac98
RS
1483 xdiff_clear_find_func(&xecfg);
1484 opt->priv = NULL;
1485
83b5d2f5
JH
1486 /* NEEDSWORK:
1487 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1488 * which feels mostly useless but sometimes useful. Maybe
1489 * make it another option? For now suppress them.
1490 */
5b594f45
FK
1491 if (opt->count && count) {
1492 char buf[32];
e1327023 1493 output_color(opt, gs->name, strlen(gs->name), opt->color_filename);
55f638bd
ML
1494 output_sep(opt, ':');
1495 snprintf(buf, sizeof(buf), "%u\n", count);
5b594f45 1496 opt->output(opt, buf, strlen(buf));
c30c10cf 1497 return 1;
5b594f45 1498 }
83b5d2f5
JH
1499 return !!last_hit;
1500}
1501
0ab7befa
JH
1502static void clr_hit_marker(struct grep_expr *x)
1503{
1504 /* All-hit markers are meaningful only at the very top level
1505 * OR node.
1506 */
1507 while (1) {
1508 x->hit = 0;
1509 if (x->node != GREP_NODE_OR)
1510 return;
1511 x->u.binary.left->hit = 0;
1512 x = x->u.binary.right;
1513 }
1514}
1515
1516static int chk_hit_marker(struct grep_expr *x)
1517{
1518 /* Top level nodes have hit markers. See if they all are hits */
1519 while (1) {
1520 if (x->node != GREP_NODE_OR)
1521 return x->hit;
1522 if (!x->u.binary.left->hit)
1523 return 0;
1524 x = x->u.binary.right;
1525 }
1526}
1527
e1327023 1528int grep_source(struct grep_opt *opt, struct grep_source *gs)
0ab7befa
JH
1529{
1530 /*
1531 * we do not have to do the two-pass grep when we do not check
1532 * buffer-wide "all-match".
1533 */
1534 if (!opt->all_match)
e1327023 1535 return grep_source_1(opt, gs, 0);
0ab7befa
JH
1536
1537 /* Otherwise the toplevel "or" terms hit a bit differently.
1538 * We first clear hit markers from them.
1539 */
1540 clr_hit_marker(opt->pattern_expression);
e1327023 1541 grep_source_1(opt, gs, 1);
0ab7befa
JH
1542
1543 if (!chk_hit_marker(opt->pattern_expression))
1544 return 0;
1545
e1327023
JK
1546 return grep_source_1(opt, gs, 0);
1547}
1548
c876d6da 1549int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
e1327023
JK
1550{
1551 struct grep_source gs;
1552 int r;
1553
55c61688 1554 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL, NULL);
e1327023
JK
1555 gs.buf = buf;
1556 gs.size = size;
1557
1558 r = grep_source(opt, &gs);
1559
1560 grep_source_clear(&gs);
1561 return r;
1562}
1563
1564void grep_source_init(struct grep_source *gs, enum grep_source_type type,
55c61688
NTND
1565 const char *name, const char *path,
1566 const void *identifier)
e1327023
JK
1567{
1568 gs->type = type;
1569 gs->name = name ? xstrdup(name) : NULL;
55c61688 1570 gs->path = path ? xstrdup(path) : NULL;
e1327023
JK
1571 gs->buf = NULL;
1572 gs->size = 0;
94ad9d9e 1573 gs->driver = NULL;
e1327023
JK
1574
1575 switch (type) {
1576 case GREP_SOURCE_FILE:
1577 gs->identifier = xstrdup(identifier);
1578 break;
1579 case GREP_SOURCE_SHA1:
1580 gs->identifier = xmalloc(20);
1581 memcpy(gs->identifier, identifier, 20);
1582 break;
1583 case GREP_SOURCE_BUF:
1584 gs->identifier = NULL;
1585 }
1586}
1587
1588void grep_source_clear(struct grep_source *gs)
1589{
1590 free(gs->name);
1591 gs->name = NULL;
55c61688
NTND
1592 free(gs->path);
1593 gs->path = NULL;
e1327023
JK
1594 free(gs->identifier);
1595 gs->identifier = NULL;
1596 grep_source_clear_data(gs);
1597}
1598
1599void grep_source_clear_data(struct grep_source *gs)
1600{
1601 switch (gs->type) {
1602 case GREP_SOURCE_FILE:
1603 case GREP_SOURCE_SHA1:
1604 free(gs->buf);
1605 gs->buf = NULL;
1606 gs->size = 0;
1607 break;
1608 case GREP_SOURCE_BUF:
1609 /* leave user-provided buf intact */
1610 break;
1611 }
1612}
1613
1614static int grep_source_load_sha1(struct grep_source *gs)
1615{
1616 enum object_type type;
1617
1618 grep_read_lock();
1619 gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1620 grep_read_unlock();
1621
1622 if (!gs->buf)
1623 return error(_("'%s': unable to read %s"),
1624 gs->name,
1625 sha1_to_hex(gs->identifier));
1626 return 0;
1627}
1628
1629static int grep_source_load_file(struct grep_source *gs)
1630{
1631 const char *filename = gs->identifier;
1632 struct stat st;
1633 char *data;
1634 size_t size;
1635 int i;
1636
1637 if (lstat(filename, &st) < 0) {
1638 err_ret:
1639 if (errno != ENOENT)
1640 error(_("'%s': %s"), filename, strerror(errno));
1641 return -1;
1642 }
1643 if (!S_ISREG(st.st_mode))
1644 return -1;
1645 size = xsize_t(st.st_size);
1646 i = open(filename, O_RDONLY);
1647 if (i < 0)
1648 goto err_ret;
1649 data = xmalloc(size + 1);
1650 if (st.st_size != read_in_full(i, data, size)) {
1651 error(_("'%s': short read %s"), filename, strerror(errno));
1652 close(i);
1653 free(data);
1654 return -1;
1655 }
1656 close(i);
1657 data[size] = 0;
1658
1659 gs->buf = data;
1660 gs->size = size;
1661 return 0;
1662}
1663
3083301e 1664static int grep_source_load(struct grep_source *gs)
e1327023
JK
1665{
1666 if (gs->buf)
1667 return 0;
1668
1669 switch (gs->type) {
1670 case GREP_SOURCE_FILE:
1671 return grep_source_load_file(gs);
1672 case GREP_SOURCE_SHA1:
1673 return grep_source_load_sha1(gs);
1674 case GREP_SOURCE_BUF:
1675 return gs->buf ? 0 : -1;
1676 }
1677 die("BUG: invalid grep_source type");
0ab7befa 1678}
94ad9d9e
JK
1679
1680void grep_source_load_driver(struct grep_source *gs)
1681{
1682 if (gs->driver)
1683 return;
1684
1685 grep_attr_lock();
55c61688
NTND
1686 if (gs->path)
1687 gs->driver = userdiff_find_by_path(gs->path);
94ad9d9e
JK
1688 if (!gs->driver)
1689 gs->driver = userdiff_find_by_name("default");
1690 grep_attr_unlock();
1691}
41b59bfc 1692
3083301e 1693static int grep_source_is_binary(struct grep_source *gs)
41b59bfc
JK
1694{
1695 grep_source_load_driver(gs);
1696 if (gs->driver->binary != -1)
1697 return gs->driver->binary;
1698
1699 if (!grep_source_load(gs))
1700 return buffer_is_binary(gs->buf, gs->size);
1701
1702 return 0;
1703}