grep: move pattern-type bits support to top-level grep.[ch]
[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;
19
20 memset(opt, 0, sizeof(*opt));
21 opt->relative = 1;
22 opt->pathname = 1;
23 opt->regflags = REG_NEWLINE;
24 opt->max_depth = -1;
25 opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
26 opt->extended_regexp_option = 0;
27 strcpy(opt->color_context, "");
28 strcpy(opt->color_filename, "");
29 strcpy(opt->color_function, "");
30 strcpy(opt->color_lineno, "");
31 strcpy(opt->color_match, GIT_COLOR_BOLD_RED);
32 strcpy(opt->color_selected, "");
33 strcpy(opt->color_sep, GIT_COLOR_CYAN);
34 opt->color = -1;
35}
36
37static int parse_pattern_type_arg(const char *opt, const char *arg)
38{
39 if (!strcmp(arg, "default"))
40 return GREP_PATTERN_TYPE_UNSPECIFIED;
41 else if (!strcmp(arg, "basic"))
42 return GREP_PATTERN_TYPE_BRE;
43 else if (!strcmp(arg, "extended"))
44 return GREP_PATTERN_TYPE_ERE;
45 else if (!strcmp(arg, "fixed"))
46 return GREP_PATTERN_TYPE_FIXED;
47 else if (!strcmp(arg, "perl"))
48 return GREP_PATTERN_TYPE_PCRE;
49 die("bad %s argument: %s", opt, arg);
50}
51
52/*
53 * Read the configuration file once and store it in
54 * the grep_defaults template.
55 */
56int grep_config(const char *var, const char *value, void *cb)
57{
58 struct grep_opt *opt = &grep_defaults;
59 char *color = NULL;
60
61 if (userdiff_config(var, value) < 0)
62 return -1;
63
64 if (!strcmp(var, "grep.extendedregexp")) {
65 if (git_config_bool(var, value))
66 opt->extended_regexp_option = 1;
67 else
68 opt->extended_regexp_option = 0;
69 return 0;
70 }
71
72 if (!strcmp(var, "grep.patterntype")) {
73 opt->pattern_type_option = parse_pattern_type_arg(var, value);
74 return 0;
75 }
76
77 if (!strcmp(var, "grep.linenumber")) {
78 opt->linenum = git_config_bool(var, value);
79 return 0;
80 }
81
82 if (!strcmp(var, "color.grep"))
83 opt->color = git_config_colorbool(var, value);
84 else if (!strcmp(var, "color.grep.context"))
85 color = opt->color_context;
86 else if (!strcmp(var, "color.grep.filename"))
87 color = opt->color_filename;
88 else if (!strcmp(var, "color.grep.function"))
89 color = opt->color_function;
90 else if (!strcmp(var, "color.grep.linenumber"))
91 color = opt->color_lineno;
92 else if (!strcmp(var, "color.grep.match"))
93 color = opt->color_match;
94 else if (!strcmp(var, "color.grep.selected"))
95 color = opt->color_selected;
96 else if (!strcmp(var, "color.grep.separator"))
97 color = opt->color_sep;
98
99 if (color) {
100 if (!value)
101 return config_error_nonbool(var);
102 color_parse(value, var, color);
103 }
104 return 0;
105}
106
107/*
108 * Initialize one instance of grep_opt and copy the
109 * default values from the template we read the configuration
110 * information in an earlier call to git_config(grep_config).
111 */
112void grep_init(struct grep_opt *opt, const char *prefix)
113{
114 struct grep_opt *def = &grep_defaults;
115
116 memset(opt, 0, sizeof(*opt));
117 opt->prefix = prefix;
118 opt->prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
119 opt->pattern_tail = &opt->pattern_list;
120 opt->header_tail = &opt->header_list;
121
122 opt->color = def->color;
123 opt->extended_regexp_option = def->extended_regexp_option;
124 opt->pattern_type_option = def->pattern_type_option;
125 opt->linenum = def->linenum;
126 opt->max_depth = def->max_depth;
127 opt->pathname = def->pathname;
128 opt->regflags = def->regflags;
129 opt->relative = def->relative;
130
131 strcpy(opt->color_context, def->color_context);
132 strcpy(opt->color_filename, def->color_filename);
133 strcpy(opt->color_function, def->color_function);
134 strcpy(opt->color_lineno, def->color_lineno);
135 strcpy(opt->color_match, def->color_match);
136 strcpy(opt->color_selected, def->color_selected);
137 strcpy(opt->color_sep, def->color_sep);
138}
07a7d656 139
c5c31d33
JH
140void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_opt *opt)
141{
142 if (pattern_type != GREP_PATTERN_TYPE_UNSPECIFIED)
143 grep_set_pattern_type_option(pattern_type, opt);
144 else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
145 grep_set_pattern_type_option(opt->pattern_type_option, opt);
146 else if (opt->extended_regexp_option)
147 grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
148}
149
150void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
151{
152 switch (pattern_type) {
153 case GREP_PATTERN_TYPE_UNSPECIFIED:
154 /* fall through */
155
156 case GREP_PATTERN_TYPE_BRE:
157 opt->fixed = 0;
158 opt->pcre = 0;
159 opt->regflags &= ~REG_EXTENDED;
160 break;
161
162 case GREP_PATTERN_TYPE_ERE:
163 opt->fixed = 0;
164 opt->pcre = 0;
165 opt->regflags |= REG_EXTENDED;
166 break;
167
168 case GREP_PATTERN_TYPE_FIXED:
169 opt->fixed = 1;
170 opt->pcre = 0;
171 opt->regflags &= ~REG_EXTENDED;
172 break;
173
174 case GREP_PATTERN_TYPE_PCRE:
175 opt->fixed = 0;
176 opt->pcre = 1;
177 opt->regflags &= ~REG_EXTENDED;
178 break;
179 }
180}
181
fc456751
RS
182static struct grep_pat *create_grep_pat(const char *pat, size_t patlen,
183 const char *origin, int no,
184 enum grep_pat_token t,
185 enum grep_header_field field)
a4d7d2c6
JH
186{
187 struct grep_pat *p = xcalloc(1, sizeof(*p));
526a858a 188 p->pattern = xmemdupz(pat, patlen);
fc456751
RS
189 p->patternlen = patlen;
190 p->origin = origin;
191 p->no = no;
192 p->token = t;
a4d7d2c6 193 p->field = field;
fc456751
RS
194 return p;
195}
196
2b3873ff
RS
197static void do_append_grep_pat(struct grep_pat ***tail, struct grep_pat *p)
198{
199 **tail = p;
200 *tail = &p->next;
a4d7d2c6 201 p->next = NULL;
526a858a
RS
202
203 switch (p->token) {
204 case GREP_PATTERN: /* atom */
205 case GREP_PATTERN_HEAD:
206 case GREP_PATTERN_BODY:
207 for (;;) {
208 struct grep_pat *new_pat;
209 size_t len = 0;
210 char *cp = p->pattern + p->patternlen, *nl = NULL;
211 while (++len <= p->patternlen) {
212 if (*(--cp) == '\n') {
213 nl = cp;
214 break;
215 }
216 }
217 if (!nl)
218 break;
219 new_pat = create_grep_pat(nl + 1, len - 1, p->origin,
220 p->no, p->token, p->field);
221 new_pat->next = p->next;
222 if (!p->next)
223 *tail = &new_pat->next;
224 p->next = new_pat;
225 *nl = '\0';
226 p->patternlen -= len;
227 }
228 break;
229 default:
230 break;
231 }
2b3873ff
RS
232}
233
fc456751
RS
234void append_header_grep_pattern(struct grep_opt *opt,
235 enum grep_header_field field, const char *pat)
236{
237 struct grep_pat *p = create_grep_pat(pat, strlen(pat), "header", 0,
238 GREP_PATTERN_HEAD, field);
baa6378f
JH
239 if (field == GREP_HEADER_REFLOG)
240 opt->use_reflog_filter = 1;
2b3873ff 241 do_append_grep_pat(&opt->header_tail, p);
a4d7d2c6
JH
242}
243
83b5d2f5
JH
244void append_grep_pattern(struct grep_opt *opt, const char *pat,
245 const char *origin, int no, enum grep_pat_token t)
ed40a095
RS
246{
247 append_grep_pat(opt, pat, strlen(pat), origin, no, t);
248}
249
250void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen,
251 const char *origin, int no, enum grep_pat_token t)
83b5d2f5 252{
fc456751 253 struct grep_pat *p = create_grep_pat(pat, patlen, origin, no, t, 0);
2b3873ff 254 do_append_grep_pat(&opt->pattern_tail, p);
83b5d2f5
JH
255}
256
5b594f45
FK
257struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
258{
259 struct grep_pat *pat;
260 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
261 *ret = *opt;
262
263 ret->pattern_list = NULL;
264 ret->pattern_tail = &ret->pattern_list;
265
266 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
267 {
268 if(pat->token == GREP_PATTERN_HEAD)
269 append_header_grep_pattern(ret, pat->field,
270 pat->pattern);
271 else
ed40a095
RS
272 append_grep_pat(ret, pat->pattern, pat->patternlen,
273 pat->origin, pat->no, pat->token);
5b594f45
FK
274 }
275
276 return ret;
277}
278
a30c148a
MK
279static NORETURN void compile_regexp_failed(const struct grep_pat *p,
280 const char *error)
281{
282 char where[1024];
283
284 if (p->no)
285 sprintf(where, "In '%s' at %d, ", p->origin, p->no);
286 else if (p->origin)
287 sprintf(where, "%s, ", p->origin);
288 else
289 where[0] = 0;
290
291 die("%s'%s': %s", where, p->pattern, error);
292}
293
63e7e9d8
MK
294#ifdef USE_LIBPCRE
295static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
296{
297 const char *error;
298 int erroffset;
fba4f125 299 int options = PCRE_MULTILINE;
63e7e9d8
MK
300
301 if (opt->ignore_case)
302 options |= PCRE_CASELESS;
303
304 p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
305 NULL);
306 if (!p->pcre_regexp)
307 compile_regexp_failed(p, error);
308
309 p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
310 if (!p->pcre_extra_info && error)
311 die("%s", error);
312}
313
314static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
315 regmatch_t *match, int eflags)
316{
317 int ovector[30], ret, flags = 0;
318
319 if (eflags & REG_NOTBOL)
320 flags |= PCRE_NOTBOL;
321
322 ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
323 0, flags, ovector, ARRAY_SIZE(ovector));
324 if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
325 die("pcre_exec failed with error code %d", ret);
326 if (ret > 0) {
327 ret = 0;
328 match->rm_so = ovector[0];
329 match->rm_eo = ovector[1];
330 }
331
332 return ret;
333}
334
335static void free_pcre_regexp(struct grep_pat *p)
336{
337 pcre_free(p->pcre_regexp);
338 pcre_free(p->pcre_extra_info);
339}
340#else /* !USE_LIBPCRE */
341static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
342{
343 die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
344}
345
346static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
347 regmatch_t *match, int eflags)
348{
349 return 1;
350}
351
352static void free_pcre_regexp(struct grep_pat *p)
353{
354}
355#endif /* !USE_LIBPCRE */
356
9eceddee
FK
357static int is_fixed(const char *s, size_t len)
358{
359 size_t i;
360
361 /* regcomp cannot accept patterns with NULs so we
362 * consider any pattern containing a NUL fixed.
363 */
364 if (memchr(s, 0, len))
365 return 1;
366
367 for (i = 0; i < len; i++) {
368 if (is_regex_special(s[i]))
369 return 0;
370 }
371
372 return 1;
373}
374
83b5d2f5
JH
375static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
376{
c822255c
RS
377 int err;
378
d7eb527d 379 p->word_regexp = opt->word_regexp;
5183bf67 380 p->ignore_case = opt->ignore_case;
d7eb527d 381
9eceddee
FK
382 if (opt->fixed || is_fixed(p->pattern, p->patternlen))
383 p->fixed = 1;
384 else
385 p->fixed = 0;
386
387 if (p->fixed) {
0f871cf5
JH
388 if (opt->regflags & REG_ICASE || p->ignore_case)
389 p->kws = kwsalloc(tolower_trans_tbl);
390 else
9eceddee 391 p->kws = kwsalloc(NULL);
9eceddee
FK
392 kwsincr(p->kws, p->pattern, p->patternlen);
393 kwsprep(p->kws);
c822255c 394 return;
9eceddee 395 }
c822255c 396
63e7e9d8
MK
397 if (opt->pcre) {
398 compile_pcre_regexp(p, opt);
399 return;
400 }
401
c822255c 402 err = regcomp(&p->regexp, p->pattern, opt->regflags);
83b5d2f5
JH
403 if (err) {
404 char errbuf[1024];
83b5d2f5
JH
405 regerror(err, &p->regexp, errbuf, 1024);
406 regfree(&p->regexp);
a30c148a 407 compile_regexp_failed(p, errbuf);
83b5d2f5
JH
408 }
409}
410
0ab7befa 411static struct grep_expr *compile_pattern_or(struct grep_pat **);
83b5d2f5
JH
412static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
413{
414 struct grep_pat *p;
415 struct grep_expr *x;
416
417 p = *list;
c922b01f
LT
418 if (!p)
419 return NULL;
83b5d2f5
JH
420 switch (p->token) {
421 case GREP_PATTERN: /* atom */
480c1ca6
JH
422 case GREP_PATTERN_HEAD:
423 case GREP_PATTERN_BODY:
83b5d2f5
JH
424 x = xcalloc(1, sizeof (struct grep_expr));
425 x->node = GREP_NODE_ATOM;
426 x->u.atom = p;
427 *list = p->next;
428 return x;
429 case GREP_OPEN_PAREN:
430 *list = p->next;
0ab7befa 431 x = compile_pattern_or(list);
83b5d2f5
JH
432 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
433 die("unmatched parenthesis");
434 *list = (*list)->next;
435 return x;
436 default:
437 return NULL;
438 }
439}
440
441static struct grep_expr *compile_pattern_not(struct grep_pat **list)
442{
443 struct grep_pat *p;
444 struct grep_expr *x;
445
446 p = *list;
c922b01f
LT
447 if (!p)
448 return NULL;
83b5d2f5
JH
449 switch (p->token) {
450 case GREP_NOT:
451 if (!p->next)
452 die("--not not followed by pattern expression");
453 *list = p->next;
454 x = xcalloc(1, sizeof (struct grep_expr));
455 x->node = GREP_NODE_NOT;
456 x->u.unary = compile_pattern_not(list);
457 if (!x->u.unary)
458 die("--not followed by non pattern expression");
459 return x;
460 default:
461 return compile_pattern_atom(list);
462 }
463}
464
465static struct grep_expr *compile_pattern_and(struct grep_pat **list)
466{
467 struct grep_pat *p;
468 struct grep_expr *x, *y, *z;
469
470 x = compile_pattern_not(list);
471 p = *list;
472 if (p && p->token == GREP_AND) {
473 if (!p->next)
474 die("--and not followed by pattern expression");
475 *list = p->next;
476 y = compile_pattern_and(list);
477 if (!y)
478 die("--and not followed by pattern expression");
479 z = xcalloc(1, sizeof (struct grep_expr));
480 z->node = GREP_NODE_AND;
481 z->u.binary.left = x;
482 z->u.binary.right = y;
483 return z;
484 }
485 return x;
486}
487
488static struct grep_expr *compile_pattern_or(struct grep_pat **list)
489{
490 struct grep_pat *p;
491 struct grep_expr *x, *y, *z;
492
493 x = compile_pattern_and(list);
494 p = *list;
495 if (x && p && p->token != GREP_CLOSE_PAREN) {
496 y = compile_pattern_or(list);
497 if (!y)
498 die("not a pattern expression %s", p->pattern);
499 z = xcalloc(1, sizeof (struct grep_expr));
500 z->node = GREP_NODE_OR;
501 z->u.binary.left = x;
502 z->u.binary.right = y;
503 return z;
504 }
505 return x;
506}
507
508static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
509{
510 return compile_pattern_or(list);
511}
512
17bf35a3
JH
513static void indent(int in)
514{
515 while (in-- > 0)
516 fputc(' ', stderr);
517}
518
519static void dump_grep_pat(struct grep_pat *p)
520{
521 switch (p->token) {
522 case GREP_AND: fprintf(stderr, "*and*"); break;
523 case GREP_OPEN_PAREN: fprintf(stderr, "*(*"); break;
524 case GREP_CLOSE_PAREN: fprintf(stderr, "*)*"); break;
525 case GREP_NOT: fprintf(stderr, "*not*"); break;
526 case GREP_OR: fprintf(stderr, "*or*"); break;
527
528 case GREP_PATTERN: fprintf(stderr, "pattern"); break;
529 case GREP_PATTERN_HEAD: fprintf(stderr, "pattern_head"); break;
530 case GREP_PATTERN_BODY: fprintf(stderr, "pattern_body"); break;
531 }
532
533 switch (p->token) {
534 default: break;
535 case GREP_PATTERN_HEAD:
536 fprintf(stderr, "<head %d>", p->field); break;
537 case GREP_PATTERN_BODY:
538 fprintf(stderr, "<body>"); break;
539 }
540 switch (p->token) {
541 default: break;
542 case GREP_PATTERN_HEAD:
543 case GREP_PATTERN_BODY:
544 case GREP_PATTERN:
545 fprintf(stderr, "%.*s", (int)p->patternlen, p->pattern);
546 break;
547 }
548 fputc('\n', stderr);
549}
550
551static void dump_grep_expression_1(struct grep_expr *x, int in)
552{
553 indent(in);
554 switch (x->node) {
555 case GREP_NODE_TRUE:
556 fprintf(stderr, "true\n");
557 break;
558 case GREP_NODE_ATOM:
559 dump_grep_pat(x->u.atom);
560 break;
561 case GREP_NODE_NOT:
562 fprintf(stderr, "(not\n");
563 dump_grep_expression_1(x->u.unary, in+1);
564 indent(in);
565 fprintf(stderr, ")\n");
566 break;
567 case GREP_NODE_AND:
568 fprintf(stderr, "(and\n");
569 dump_grep_expression_1(x->u.binary.left, in+1);
570 dump_grep_expression_1(x->u.binary.right, in+1);
571 indent(in);
572 fprintf(stderr, ")\n");
573 break;
574 case GREP_NODE_OR:
575 fprintf(stderr, "(or\n");
576 dump_grep_expression_1(x->u.binary.left, in+1);
577 dump_grep_expression_1(x->u.binary.right, in+1);
578 indent(in);
579 fprintf(stderr, ")\n");
580 break;
581 }
582}
583
07a7d656 584static void dump_grep_expression(struct grep_opt *opt)
17bf35a3
JH
585{
586 struct grep_expr *x = opt->pattern_expression;
587
588 if (opt->all_match)
589 fprintf(stderr, "[all-match]\n");
590 dump_grep_expression_1(x, 0);
591 fflush(NULL);
592}
593
5aaeb733
JH
594static struct grep_expr *grep_true_expr(void)
595{
596 struct grep_expr *z = xcalloc(1, sizeof(*z));
597 z->node = GREP_NODE_TRUE;
598 return z;
599}
600
601static struct grep_expr *grep_or_expr(struct grep_expr *left, struct grep_expr *right)
602{
603 struct grep_expr *z = xcalloc(1, sizeof(*z));
604 z->node = GREP_NODE_OR;
605 z->u.binary.left = left;
606 z->u.binary.right = right;
607 return z;
608}
609
95ce9ce2 610static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
83b5d2f5
JH
611{
612 struct grep_pat *p;
95ce9ce2 613 struct grep_expr *header_expr;
5aaeb733
JH
614 struct grep_expr *(header_group[GREP_HEADER_FIELD_MAX]);
615 enum grep_header_field fld;
83b5d2f5 616
95ce9ce2
JH
617 if (!opt->header_list)
618 return NULL;
2385f246 619
95ce9ce2
JH
620 for (p = opt->header_list; p; p = p->next) {
621 if (p->token != GREP_PATTERN_HEAD)
622 die("bug: a non-header pattern in grep header list.");
623 if (p->field < 0 || GREP_HEADER_FIELD_MAX <= p->field)
624 die("bug: unknown header field %d", p->field);
625 compile_regexp(p, opt);
80235ba7 626 }
5aaeb733
JH
627
628 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++)
629 header_group[fld] = NULL;
630
631 for (p = opt->header_list; p; p = p->next) {
632 struct grep_expr *h;
633 struct grep_pat *pp = p;
634
635 h = compile_pattern_atom(&pp);
636 if (!h || pp != p->next)
637 die("bug: malformed header expr");
638 if (!header_group[p->field]) {
639 header_group[p->field] = h;
640 continue;
641 }
642 header_group[p->field] = grep_or_expr(h, header_group[p->field]);
643 }
644
645 header_expr = NULL;
646
647 for (fld = 0; fld < GREP_HEADER_FIELD_MAX; fld++) {
648 if (!header_group[fld])
649 continue;
650 if (!header_expr)
651 header_expr = grep_true_expr();
652 header_expr = grep_or_expr(header_group[fld], header_expr);
653 }
95ce9ce2
JH
654 return header_expr;
655}
656
13e4fc7e
JH
657static struct grep_expr *grep_splice_or(struct grep_expr *x, struct grep_expr *y)
658{
659 struct grep_expr *z = x;
660
661 while (x) {
662 assert(x->node == GREP_NODE_OR);
663 if (x->u.binary.right &&
664 x->u.binary.right->node == GREP_NODE_TRUE) {
665 x->u.binary.right = y;
666 break;
667 }
668 x = x->u.binary.right;
669 }
670 return z;
671}
672
17bf35a3 673static void compile_grep_patterns_real(struct grep_opt *opt)
95ce9ce2
JH
674{
675 struct grep_pat *p;
676 struct grep_expr *header_expr = prep_header_patterns(opt);
0ab7befa 677
83b5d2f5 678 for (p = opt->pattern_list; p; p = p->next) {
480c1ca6
JH
679 switch (p->token) {
680 case GREP_PATTERN: /* atom */
681 case GREP_PATTERN_HEAD:
682 case GREP_PATTERN_BODY:
c822255c 683 compile_regexp(p, opt);
480c1ca6
JH
684 break;
685 default:
83b5d2f5 686 opt->extended = 1;
480c1ca6
JH
687 break;
688 }
83b5d2f5
JH
689 }
690
80235ba7
JH
691 if (opt->all_match || header_expr)
692 opt->extended = 1;
17bf35a3 693 else if (!opt->extended && !opt->debug)
83b5d2f5
JH
694 return;
695
83b5d2f5 696 p = opt->pattern_list;
ba150a3f
MB
697 if (p)
698 opt->pattern_expression = compile_pattern_expr(&p);
83b5d2f5
JH
699 if (p)
700 die("incomplete pattern expression: %s", p->pattern);
80235ba7
JH
701
702 if (!header_expr)
703 return;
704
5aaeb733 705 if (!opt->pattern_expression)
80235ba7 706 opt->pattern_expression = header_expr;
13e4fc7e
JH
707 else if (opt->all_match)
708 opt->pattern_expression = grep_splice_or(header_expr,
709 opt->pattern_expression);
5aaeb733
JH
710 else
711 opt->pattern_expression = grep_or_expr(opt->pattern_expression,
712 header_expr);
80235ba7 713 opt->all_match = 1;
83b5d2f5
JH
714}
715
17bf35a3
JH
716void compile_grep_patterns(struct grep_opt *opt)
717{
718 compile_grep_patterns_real(opt);
719 if (opt->debug)
720 dump_grep_expression(opt);
721}
722
b48fb5b6
JH
723static void free_pattern_expr(struct grep_expr *x)
724{
725 switch (x->node) {
5aaeb733 726 case GREP_NODE_TRUE:
b48fb5b6
JH
727 case GREP_NODE_ATOM:
728 break;
729 case GREP_NODE_NOT:
730 free_pattern_expr(x->u.unary);
731 break;
732 case GREP_NODE_AND:
733 case GREP_NODE_OR:
734 free_pattern_expr(x->u.binary.left);
735 free_pattern_expr(x->u.binary.right);
736 break;
737 }
738 free(x);
739}
740
741void free_grep_patterns(struct grep_opt *opt)
742{
743 struct grep_pat *p, *n;
744
745 for (p = opt->pattern_list; p; p = n) {
746 n = p->next;
747 switch (p->token) {
748 case GREP_PATTERN: /* atom */
749 case GREP_PATTERN_HEAD:
750 case GREP_PATTERN_BODY:
9eceddee
FK
751 if (p->kws)
752 kwsfree(p->kws);
753 else if (p->pcre_regexp)
63e7e9d8
MK
754 free_pcre_regexp(p);
755 else
756 regfree(&p->regexp);
526a858a 757 free(p->pattern);
b48fb5b6
JH
758 break;
759 default:
760 break;
761 }
762 free(p);
763 }
764
765 if (!opt->extended)
766 return;
767 free_pattern_expr(opt->pattern_expression);
768}
769
83b5d2f5
JH
770static char *end_of_line(char *cp, unsigned long *left)
771{
772 unsigned long l = *left;
773 while (l && *cp != '\n') {
774 l--;
775 cp++;
776 }
777 *left = l;
778 return cp;
779}
780
781static int word_char(char ch)
782{
783 return isalnum(ch) || ch == '_';
784}
785
55f638bd
ML
786static void output_color(struct grep_opt *opt, const void *data, size_t size,
787 const char *color)
788{
daa0c3d9 789 if (want_color(opt->color) && color && color[0]) {
55f638bd
ML
790 opt->output(opt, color, strlen(color));
791 opt->output(opt, data, size);
792 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
793 } else
794 opt->output(opt, data, size);
795}
796
797static void output_sep(struct grep_opt *opt, char sign)
798{
799 if (opt->null_following_name)
800 opt->output(opt, "\0", 1);
801 else
802 output_color(opt, &sign, 1, opt->color_sep);
803}
804
83caecca
RZ
805static void show_name(struct grep_opt *opt, const char *name)
806{
55f638bd 807 output_color(opt, name, strlen(name), opt->color_filename);
5b594f45 808 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
83caecca
RZ
809}
810
ed40a095
RS
811static int fixmatch(struct grep_pat *p, char *line, char *eol,
812 regmatch_t *match)
83b5d2f5 813{
9eceddee
FK
814 struct kwsmatch kwsm;
815 size_t offset = kwsexec(p->kws, line, eol - line, &kwsm);
816 if (offset == -1) {
83b5d2f5
JH
817 match->rm_so = match->rm_eo = -1;
818 return REG_NOMATCH;
9eceddee
FK
819 } else {
820 match->rm_so = offset;
821 match->rm_eo = match->rm_so + kwsm.size[0];
83b5d2f5
JH
822 return 0;
823 }
824}
825
f96e5673
RS
826static int regmatch(const regex_t *preg, char *line, char *eol,
827 regmatch_t *match, int eflags)
828{
829#ifdef REG_STARTEND
830 match->rm_so = 0;
831 match->rm_eo = eol - line;
832 eflags |= REG_STARTEND;
833#endif
834 return regexec(preg, line, 1, match, eflags);
835}
836
97e77784
MK
837static int patmatch(struct grep_pat *p, char *line, char *eol,
838 regmatch_t *match, int eflags)
839{
840 int hit;
841
842 if (p->fixed)
843 hit = !fixmatch(p, line, eol, match);
63e7e9d8
MK
844 else if (p->pcre_regexp)
845 hit = !pcrematch(p, line, eol, match, eflags);
97e77784
MK
846 else
847 hit = !regmatch(&p->regexp, line, eol, match, eflags);
848
849 return hit;
850}
851
a4d7d2c6
JH
852static int strip_timestamp(char *bol, char **eol_p)
853{
854 char *eol = *eol_p;
855 int ch;
856
857 while (bol < --eol) {
858 if (*eol != '>')
859 continue;
860 *eol_p = ++eol;
861 ch = *eol;
862 *eol = '\0';
863 return ch;
864 }
865 return 0;
866}
867
868static struct {
869 const char *field;
870 size_t len;
871} header_field[] = {
872 { "author ", 7 },
873 { "committer ", 10 },
72fd13f7 874 { "reflog ", 7 },
a4d7d2c6
JH
875};
876
d7eb527d 877static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
79212772
RS
878 enum grep_context ctx,
879 regmatch_t *pmatch, int eflags)
83b5d2f5
JH
880{
881 int hit = 0;
a4d7d2c6 882 int saved_ch = 0;
e701fadb 883 const char *start = bol;
83b5d2f5 884
480c1ca6
JH
885 if ((p->token != GREP_PATTERN) &&
886 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
887 return 0;
888
a4d7d2c6
JH
889 if (p->token == GREP_PATTERN_HEAD) {
890 const char *field;
891 size_t len;
892 assert(p->field < ARRAY_SIZE(header_field));
893 field = header_field[p->field].field;
894 len = header_field[p->field].len;
895 if (strncmp(bol, field, len))
896 return 0;
897 bol += len;
ad4813b3
NTND
898 switch (p->field) {
899 case GREP_HEADER_AUTHOR:
900 case GREP_HEADER_COMMITTER:
901 saved_ch = strip_timestamp(bol, &eol);
902 break;
903 default:
904 break;
905 }
a4d7d2c6
JH
906 }
907
83b5d2f5 908 again:
97e77784 909 hit = patmatch(p, bol, eol, pmatch, eflags);
83b5d2f5 910
d7eb527d 911 if (hit && p->word_regexp) {
83b5d2f5 912 if ((pmatch[0].rm_so < 0) ||
84201eae 913 (eol - bol) < pmatch[0].rm_so ||
83b5d2f5
JH
914 (pmatch[0].rm_eo < 0) ||
915 (eol - bol) < pmatch[0].rm_eo)
916 die("regexp returned nonsense");
917
918 /* Match beginning must be either beginning of the
919 * line, or at word boundary (i.e. the last char must
920 * not be a word char). Similarly, match end must be
921 * either end of the line, or at word boundary
922 * (i.e. the next char must not be a word char).
923 */
fb62eb7f 924 if ( ((pmatch[0].rm_so == 0) ||
83b5d2f5
JH
925 !word_char(bol[pmatch[0].rm_so-1])) &&
926 ((pmatch[0].rm_eo == (eol-bol)) ||
927 !word_char(bol[pmatch[0].rm_eo])) )
928 ;
929 else
930 hit = 0;
931
84201eae
RS
932 /* Words consist of at least one character. */
933 if (pmatch->rm_so == pmatch->rm_eo)
934 hit = 0;
935
83b5d2f5
JH
936 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
937 /* There could be more than one match on the
938 * line, and the first match might not be
939 * strict word match. But later ones could be!
fb62eb7f
RS
940 * Forward to the next possible start, i.e. the
941 * next position following a non-word char.
83b5d2f5
JH
942 */
943 bol = pmatch[0].rm_so + bol + 1;
fb62eb7f
RS
944 while (word_char(bol[-1]) && bol < eol)
945 bol++;
dbb6a4ad 946 eflags |= REG_NOTBOL;
fb62eb7f
RS
947 if (bol < eol)
948 goto again;
83b5d2f5
JH
949 }
950 }
a4d7d2c6
JH
951 if (p->token == GREP_PATTERN_HEAD && saved_ch)
952 *eol = saved_ch;
e701fadb
RS
953 if (hit) {
954 pmatch[0].rm_so += bol - start;
955 pmatch[0].rm_eo += bol - start;
956 }
83b5d2f5
JH
957 return hit;
958}
959
d7eb527d
RS
960static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
961 enum grep_context ctx, int collect_hits)
83b5d2f5 962{
0ab7befa 963 int h = 0;
79212772 964 regmatch_t match;
0ab7befa 965
c922b01f
LT
966 if (!x)
967 die("Not a valid grep expression");
83b5d2f5 968 switch (x->node) {
5aaeb733
JH
969 case GREP_NODE_TRUE:
970 h = 1;
971 break;
83b5d2f5 972 case GREP_NODE_ATOM:
79212772 973 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
83b5d2f5
JH
974 break;
975 case GREP_NODE_NOT:
d7eb527d 976 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
0ab7befa 977 break;
83b5d2f5 978 case GREP_NODE_AND:
d7eb527d 979 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
252d560d 980 return 0;
d7eb527d 981 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
0ab7befa 982 break;
83b5d2f5 983 case GREP_NODE_OR:
0ab7befa 984 if (!collect_hits)
d7eb527d 985 return (match_expr_eval(x->u.binary.left,
0ab7befa 986 bol, eol, ctx, 0) ||
d7eb527d 987 match_expr_eval(x->u.binary.right,
0ab7befa 988 bol, eol, ctx, 0));
d7eb527d 989 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
0ab7befa 990 x->u.binary.left->hit |= h;
d7eb527d 991 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
0ab7befa
JH
992 break;
993 default:
d7530708 994 die("Unexpected node type (internal error) %d", x->node);
83b5d2f5 995 }
0ab7befa
JH
996 if (collect_hits)
997 x->hit |= h;
998 return h;
83b5d2f5
JH
999}
1000
480c1ca6 1001static int match_expr(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1002 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1003{
1004 struct grep_expr *x = opt->pattern_expression;
d7eb527d 1005 return match_expr_eval(x, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1006}
1007
480c1ca6 1008static int match_line(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 1009 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
1010{
1011 struct grep_pat *p;
79212772
RS
1012 regmatch_t match;
1013
83b5d2f5 1014 if (opt->extended)
0ab7befa
JH
1015 return match_expr(opt, bol, eol, ctx, collect_hits);
1016
1017 /* we do not call with collect_hits without being extended */
83b5d2f5 1018 for (p = opt->pattern_list; p; p = p->next) {
79212772 1019 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
83b5d2f5
JH
1020 return 1;
1021 }
1022 return 0;
1023}
1024
7e8f59d5
RS
1025static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
1026 enum grep_context ctx,
1027 regmatch_t *pmatch, int eflags)
1028{
1029 regmatch_t match;
1030
1031 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
1032 return 0;
1033 if (match.rm_so < 0 || match.rm_eo < 0)
1034 return 0;
1035 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
1036 if (match.rm_so > pmatch->rm_so)
1037 return 1;
1038 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
1039 return 1;
1040 }
1041 pmatch->rm_so = match.rm_so;
1042 pmatch->rm_eo = match.rm_eo;
1043 return 1;
1044}
1045
1046static int next_match(struct grep_opt *opt, char *bol, char *eol,
1047 enum grep_context ctx, regmatch_t *pmatch, int eflags)
1048{
1049 struct grep_pat *p;
1050 int hit = 0;
1051
1052 pmatch->rm_so = pmatch->rm_eo = -1;
1053 if (bol < eol) {
1054 for (p = opt->pattern_list; p; p = p->next) {
1055 switch (p->token) {
1056 case GREP_PATTERN: /* atom */
1057 case GREP_PATTERN_HEAD:
1058 case GREP_PATTERN_BODY:
1059 hit |= match_next_pattern(p, bol, eol, ctx,
1060 pmatch, eflags);
1061 break;
1062 default:
1063 break;
1064 }
1065 }
1066 }
1067 return hit;
1068}
1069
1070static void show_line(struct grep_opt *opt, char *bol, char *eol,
1071 const char *name, unsigned lno, char sign)
1072{
1073 int rest = eol - bol;
00588bb5 1074 char *line_color = NULL;
7e8f59d5 1075
a8f0e764
RS
1076 if (opt->file_break && opt->last_shown == 0) {
1077 if (opt->show_hunk_mark)
1078 opt->output(opt, "\n", 1);
ba8ea749 1079 } else if (opt->pre_context || opt->post_context || opt->funcbody) {
046802d0 1080 if (opt->last_shown == 0) {
55f638bd
ML
1081 if (opt->show_hunk_mark) {
1082 output_color(opt, "--", 2, opt->color_sep);
1083 opt->output(opt, "\n", 1);
07b838f0 1084 }
55f638bd
ML
1085 } else if (lno > opt->last_shown + 1) {
1086 output_color(opt, "--", 2, opt->color_sep);
1087 opt->output(opt, "\n", 1);
1088 }
5dd06d38 1089 }
1d84f72e
RS
1090 if (opt->heading && opt->last_shown == 0) {
1091 output_color(opt, name, strlen(name), opt->color_filename);
1092 opt->output(opt, "\n", 1);
1093 }
5dd06d38
RS
1094 opt->last_shown = lno;
1095
1d84f72e 1096 if (!opt->heading && opt->pathname) {
55f638bd
ML
1097 output_color(opt, name, strlen(name), opt->color_filename);
1098 output_sep(opt, sign);
5b594f45
FK
1099 }
1100 if (opt->linenum) {
1101 char buf[32];
1102 snprintf(buf, sizeof(buf), "%d", lno);
55f638bd
ML
1103 output_color(opt, buf, strlen(buf), opt->color_lineno);
1104 output_sep(opt, sign);
5b594f45 1105 }
7e8f59d5
RS
1106 if (opt->color) {
1107 regmatch_t match;
1108 enum grep_context ctx = GREP_CONTEXT_BODY;
1109 int ch = *eol;
1110 int eflags = 0;
1111
00588bb5
ML
1112 if (sign == ':')
1113 line_color = opt->color_selected;
1114 else if (sign == '-')
1115 line_color = opt->color_context;
1116 else if (sign == '=')
1117 line_color = opt->color_function;
7e8f59d5
RS
1118 *eol = '\0';
1119 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1f5b9cc4
RS
1120 if (match.rm_so == match.rm_eo)
1121 break;
5b594f45 1122
00588bb5 1123 output_color(opt, bol, match.rm_so, line_color);
55f638bd
ML
1124 output_color(opt, bol + match.rm_so,
1125 match.rm_eo - match.rm_so,
1126 opt->color_match);
7e8f59d5
RS
1127 bol += match.rm_eo;
1128 rest -= match.rm_eo;
1129 eflags = REG_NOTBOL;
1130 }
1131 *eol = ch;
1132 }
00588bb5 1133 output_color(opt, bol, rest, line_color);
5b594f45 1134 opt->output(opt, "\n", 1);
7e8f59d5
RS
1135}
1136
0579f91d 1137#ifndef NO_PTHREADS
78db6ea9
JK
1138int grep_use_locks;
1139
0579f91d
TR
1140/*
1141 * This lock protects access to the gitattributes machinery, which is
1142 * not thread-safe.
1143 */
1144pthread_mutex_t grep_attr_mutex;
1145
78db6ea9 1146static inline void grep_attr_lock(void)
0579f91d 1147{
78db6ea9 1148 if (grep_use_locks)
0579f91d
TR
1149 pthread_mutex_lock(&grep_attr_mutex);
1150}
1151
78db6ea9 1152static inline void grep_attr_unlock(void)
0579f91d 1153{
78db6ea9 1154 if (grep_use_locks)
0579f91d
TR
1155 pthread_mutex_unlock(&grep_attr_mutex);
1156}
b3aeb285
JK
1157
1158/*
1159 * Same as git_attr_mutex, but protecting the thread-unsafe object db access.
1160 */
1161pthread_mutex_t grep_read_mutex;
1162
0579f91d 1163#else
78db6ea9
JK
1164#define grep_attr_lock()
1165#define grep_attr_unlock()
0579f91d
TR
1166#endif
1167
e1327023 1168static int match_funcname(struct grep_opt *opt, struct grep_source *gs, char *bol, char *eol)
2944e4e6 1169{
60ecac98 1170 xdemitconf_t *xecfg = opt->priv;
0579f91d 1171 if (xecfg && !xecfg->find_func) {
94ad9d9e
JK
1172 grep_source_load_driver(gs);
1173 if (gs->driver->funcname.pattern) {
1174 const struct userdiff_funcname *pe = &gs->driver->funcname;
0579f91d
TR
1175 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
1176 } else {
1177 xecfg = opt->priv = NULL;
1178 }
1179 }
1180
1181 if (xecfg) {
60ecac98
RS
1182 char buf[1];
1183 return xecfg->find_func(bol, eol - bol, buf, 1,
1184 xecfg->find_func_priv) >= 0;
1185 }
1186
2944e4e6
RS
1187 if (bol == eol)
1188 return 0;
1189 if (isalpha(*bol) || *bol == '_' || *bol == '$')
1190 return 1;
1191 return 0;
1192}
1193
e1327023
JK
1194static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
1195 char *bol, unsigned lno)
2944e4e6 1196{
e1327023 1197 while (bol > gs->buf) {
2944e4e6
RS
1198 char *eol = --bol;
1199
e1327023 1200 while (bol > gs->buf && bol[-1] != '\n')
2944e4e6
RS
1201 bol--;
1202 lno--;
1203
1204 if (lno <= opt->last_shown)
1205 break;
1206
e1327023
JK
1207 if (match_funcname(opt, gs, bol, eol)) {
1208 show_line(opt, bol, eol, gs->name, lno, '=');
2944e4e6
RS
1209 break;
1210 }
1211 }
1212}
1213
e1327023 1214static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
ba8ea749 1215 char *bol, char *end, unsigned lno)
49de3216 1216{
2944e4e6 1217 unsigned cur = lno, from = 1, funcname_lno = 0;
ba8ea749
RS
1218 int funcname_needed = !!opt->funcname;
1219
e1327023 1220 if (opt->funcbody && !match_funcname(opt, gs, bol, end))
ba8ea749 1221 funcname_needed = 2;
49de3216
RS
1222
1223 if (opt->pre_context < lno)
1224 from = lno - opt->pre_context;
1225 if (from <= opt->last_shown)
1226 from = opt->last_shown + 1;
1227
1228 /* Rewind. */
e1327023 1229 while (bol > gs->buf &&
ba8ea749 1230 cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
2944e4e6
RS
1231 char *eol = --bol;
1232
e1327023 1233 while (bol > gs->buf && bol[-1] != '\n')
49de3216
RS
1234 bol--;
1235 cur--;
e1327023 1236 if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
2944e4e6
RS
1237 funcname_lno = cur;
1238 funcname_needed = 0;
1239 }
49de3216
RS
1240 }
1241
2944e4e6
RS
1242 /* We need to look even further back to find a function signature. */
1243 if (opt->funcname && funcname_needed)
e1327023 1244 show_funcname_line(opt, gs, bol, cur);
2944e4e6 1245
49de3216
RS
1246 /* Back forward. */
1247 while (cur < lno) {
2944e4e6 1248 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
49de3216
RS
1249
1250 while (*eol != '\n')
1251 eol++;
e1327023 1252 show_line(opt, bol, eol, gs->name, cur, sign);
49de3216
RS
1253 bol = eol + 1;
1254 cur++;
1255 }
1256}
1257
a26345b6
JH
1258static int should_lookahead(struct grep_opt *opt)
1259{
1260 struct grep_pat *p;
1261
1262 if (opt->extended)
1263 return 0; /* punt for too complex stuff */
1264 if (opt->invert)
1265 return 0;
1266 for (p = opt->pattern_list; p; p = p->next) {
1267 if (p->token != GREP_PATTERN)
1268 return 0; /* punt for "header only" and stuff */
1269 }
1270 return 1;
1271}
1272
1273static int look_ahead(struct grep_opt *opt,
1274 unsigned long *left_p,
1275 unsigned *lno_p,
1276 char **bol_p)
1277{
1278 unsigned lno = *lno_p;
1279 char *bol = *bol_p;
1280 struct grep_pat *p;
1281 char *sp, *last_bol;
1282 regoff_t earliest = -1;
1283
1284 for (p = opt->pattern_list; p; p = p->next) {
1285 int hit;
1286 regmatch_t m;
1287
97e77784 1288 hit = patmatch(p, bol, bol + *left_p, &m, 0);
a26345b6
JH
1289 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
1290 continue;
1291 if (earliest < 0 || m.rm_so < earliest)
1292 earliest = m.rm_so;
1293 }
1294
1295 if (earliest < 0) {
1296 *bol_p = bol + *left_p;
1297 *left_p = 0;
1298 return 1;
1299 }
1300 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
1301 ; /* find the beginning of the line */
1302 last_bol = sp;
1303
1304 for (sp = bol; sp < last_bol; sp++) {
1305 if (*sp == '\n')
1306 lno++;
1307 }
1308 *left_p -= last_bol - bol;
1309 *bol_p = last_bol;
1310 *lno_p = lno;
1311 return 0;
1312}
1313
5b594f45
FK
1314static void std_output(struct grep_opt *opt, const void *buf, size_t size)
1315{
1316 fwrite(buf, size, 1, stdout);
1317}
1318
e1327023 1319static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int collect_hits)
83b5d2f5 1320{
e1327023
JK
1321 char *bol;
1322 unsigned long left;
83b5d2f5 1323 unsigned lno = 1;
83b5d2f5 1324 unsigned last_hit = 0;
83b5d2f5 1325 int binary_match_only = 0;
83b5d2f5 1326 unsigned count = 0;
a26345b6 1327 int try_lookahead = 0;
ba8ea749 1328 int show_function = 0;
480c1ca6 1329 enum grep_context ctx = GREP_CONTEXT_HEAD;
60ecac98 1330 xdemitconf_t xecfg;
83b5d2f5 1331
5b594f45
FK
1332 if (!opt->output)
1333 opt->output = std_output;
1334
ba8ea749
RS
1335 if (opt->pre_context || opt->post_context || opt->file_break ||
1336 opt->funcbody) {
08303c36
RS
1337 /* Show hunk marks, except for the first file. */
1338 if (opt->last_shown)
1339 opt->show_hunk_mark = 1;
1340 /*
1341 * If we're using threads then we can't easily identify
1342 * the first file. Always put hunk marks in that case
1343 * and skip the very first one later in work_done().
1344 */
1345 if (opt->output != std_output)
1346 opt->show_hunk_mark = 1;
1347 }
431d6e7b
RS
1348 opt->last_shown = 0;
1349
64fcec78
RS
1350 switch (opt->binary) {
1351 case GREP_BINARY_DEFAULT:
41b59bfc 1352 if (grep_source_is_binary(gs))
83b5d2f5 1353 binary_match_only = 1;
64fcec78
RS
1354 break;
1355 case GREP_BINARY_NOMATCH:
41b59bfc 1356 if (grep_source_is_binary(gs))
83b5d2f5 1357 return 0; /* Assume unmatch */
64fcec78
RS
1358 break;
1359 case GREP_BINARY_TEXT:
1360 break;
1361 default:
1362 die("bug: unknown binary handling mode");
83b5d2f5
JH
1363 }
1364
60ecac98 1365 memset(&xecfg, 0, sizeof(xecfg));
0579f91d
TR
1366 opt->priv = &xecfg;
1367
a26345b6 1368 try_lookahead = should_lookahead(opt);
60ecac98 1369
08265798
JK
1370 if (grep_source_load(gs) < 0)
1371 return 0;
1372
e1327023
JK
1373 bol = gs->buf;
1374 left = gs->size;
83b5d2f5
JH
1375 while (left) {
1376 char *eol, ch;
0ab7befa 1377 int hit;
83b5d2f5 1378
a26345b6 1379 /*
8997da38 1380 * look_ahead() skips quickly to the line that possibly
a26345b6
JH
1381 * has the next hit; don't call it if we need to do
1382 * something more than just skipping the current line
1383 * in response to an unmatch for the current line. E.g.
1384 * inside a post-context window, we will show the current
1385 * line as a context around the previous hit when it
1386 * doesn't hit.
1387 */
1388 if (try_lookahead
1389 && !(last_hit
ba8ea749
RS
1390 && (show_function ||
1391 lno <= last_hit + opt->post_context))
a26345b6
JH
1392 && look_ahead(opt, &left, &lno, &bol))
1393 break;
83b5d2f5
JH
1394 eol = end_of_line(bol, &left);
1395 ch = *eol;
1396 *eol = 0;
1397
480c1ca6
JH
1398 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
1399 ctx = GREP_CONTEXT_BODY;
1400
0ab7befa 1401 hit = match_line(opt, bol, eol, ctx, collect_hits);
83b5d2f5
JH
1402 *eol = ch;
1403
0ab7befa
JH
1404 if (collect_hits)
1405 goto next_line;
1406
83b5d2f5
JH
1407 /* "grep -v -e foo -e bla" should list lines
1408 * that do not have either, so inversion should
1409 * be done outside.
1410 */
1411 if (opt->invert)
1412 hit = !hit;
1413 if (opt->unmatch_name_only) {
1414 if (hit)
1415 return 0;
1416 goto next_line;
1417 }
1418 if (hit) {
1419 count++;
1420 if (opt->status_only)
1421 return 1;
321ffcc0 1422 if (opt->name_only) {
e1327023 1423 show_name(opt, gs->name);
321ffcc0
RS
1424 return 1;
1425 }
c30c10cf
RS
1426 if (opt->count)
1427 goto next_line;
83b5d2f5 1428 if (binary_match_only) {
5b594f45 1429 opt->output(opt, "Binary file ", 12);
e1327023 1430 output_color(opt, gs->name, strlen(gs->name),
55f638bd 1431 opt->color_filename);
5b594f45 1432 opt->output(opt, " matches\n", 9);
83b5d2f5
JH
1433 return 1;
1434 }
83b5d2f5
JH
1435 /* Hit at this line. If we haven't shown the
1436 * pre-context lines, we would need to show them.
83b5d2f5 1437 */
ba8ea749 1438 if (opt->pre_context || opt->funcbody)
e1327023 1439 show_pre_context(opt, gs, bol, eol, lno);
2944e4e6 1440 else if (opt->funcname)
e1327023
JK
1441 show_funcname_line(opt, gs, bol, lno);
1442 show_line(opt, bol, eol, gs->name, lno, ':');
5dd06d38 1443 last_hit = lno;
ba8ea749
RS
1444 if (opt->funcbody)
1445 show_function = 1;
1446 goto next_line;
83b5d2f5 1447 }
e1327023 1448 if (show_function && match_funcname(opt, gs, bol, eol))
ba8ea749
RS
1449 show_function = 0;
1450 if (show_function ||
1451 (last_hit && lno <= last_hit + opt->post_context)) {
83b5d2f5
JH
1452 /* If the last hit is within the post context,
1453 * we need to show this line.
1454 */
e1327023 1455 show_line(opt, bol, eol, gs->name, lno, '-');
83b5d2f5 1456 }
83b5d2f5
JH
1457
1458 next_line:
1459 bol = eol + 1;
1460 if (!left)
1461 break;
1462 left--;
1463 lno++;
1464 }
1465
0ab7befa
JH
1466 if (collect_hits)
1467 return 0;
b48fb5b6 1468
83b5d2f5
JH
1469 if (opt->status_only)
1470 return 0;
1471 if (opt->unmatch_name_only) {
1472 /* We did not see any hit, so we want to show this */
e1327023 1473 show_name(opt, gs->name);
83b5d2f5
JH
1474 return 1;
1475 }
1476
60ecac98
RS
1477 xdiff_clear_find_func(&xecfg);
1478 opt->priv = NULL;
1479
83b5d2f5
JH
1480 /* NEEDSWORK:
1481 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
1482 * which feels mostly useless but sometimes useful. Maybe
1483 * make it another option? For now suppress them.
1484 */
5b594f45
FK
1485 if (opt->count && count) {
1486 char buf[32];
e1327023 1487 output_color(opt, gs->name, strlen(gs->name), opt->color_filename);
55f638bd
ML
1488 output_sep(opt, ':');
1489 snprintf(buf, sizeof(buf), "%u\n", count);
5b594f45 1490 opt->output(opt, buf, strlen(buf));
c30c10cf 1491 return 1;
5b594f45 1492 }
83b5d2f5
JH
1493 return !!last_hit;
1494}
1495
0ab7befa
JH
1496static void clr_hit_marker(struct grep_expr *x)
1497{
1498 /* All-hit markers are meaningful only at the very top level
1499 * OR node.
1500 */
1501 while (1) {
1502 x->hit = 0;
1503 if (x->node != GREP_NODE_OR)
1504 return;
1505 x->u.binary.left->hit = 0;
1506 x = x->u.binary.right;
1507 }
1508}
1509
1510static int chk_hit_marker(struct grep_expr *x)
1511{
1512 /* Top level nodes have hit markers. See if they all are hits */
1513 while (1) {
1514 if (x->node != GREP_NODE_OR)
1515 return x->hit;
1516 if (!x->u.binary.left->hit)
1517 return 0;
1518 x = x->u.binary.right;
1519 }
1520}
1521
e1327023 1522int grep_source(struct grep_opt *opt, struct grep_source *gs)
0ab7befa
JH
1523{
1524 /*
1525 * we do not have to do the two-pass grep when we do not check
1526 * buffer-wide "all-match".
1527 */
1528 if (!opt->all_match)
e1327023 1529 return grep_source_1(opt, gs, 0);
0ab7befa
JH
1530
1531 /* Otherwise the toplevel "or" terms hit a bit differently.
1532 * We first clear hit markers from them.
1533 */
1534 clr_hit_marker(opt->pattern_expression);
e1327023 1535 grep_source_1(opt, gs, 1);
0ab7befa
JH
1536
1537 if (!chk_hit_marker(opt->pattern_expression))
1538 return 0;
1539
e1327023
JK
1540 return grep_source_1(opt, gs, 0);
1541}
1542
c876d6da 1543int grep_buffer(struct grep_opt *opt, char *buf, unsigned long size)
e1327023
JK
1544{
1545 struct grep_source gs;
1546 int r;
1547
c876d6da 1548 grep_source_init(&gs, GREP_SOURCE_BUF, NULL, NULL);
e1327023
JK
1549 gs.buf = buf;
1550 gs.size = size;
1551
1552 r = grep_source(opt, &gs);
1553
1554 grep_source_clear(&gs);
1555 return r;
1556}
1557
1558void grep_source_init(struct grep_source *gs, enum grep_source_type type,
1559 const char *name, const void *identifier)
1560{
1561 gs->type = type;
1562 gs->name = name ? xstrdup(name) : NULL;
1563 gs->buf = NULL;
1564 gs->size = 0;
94ad9d9e 1565 gs->driver = NULL;
e1327023
JK
1566
1567 switch (type) {
1568 case GREP_SOURCE_FILE:
1569 gs->identifier = xstrdup(identifier);
1570 break;
1571 case GREP_SOURCE_SHA1:
1572 gs->identifier = xmalloc(20);
1573 memcpy(gs->identifier, identifier, 20);
1574 break;
1575 case GREP_SOURCE_BUF:
1576 gs->identifier = NULL;
1577 }
1578}
1579
1580void grep_source_clear(struct grep_source *gs)
1581{
1582 free(gs->name);
1583 gs->name = NULL;
1584 free(gs->identifier);
1585 gs->identifier = NULL;
1586 grep_source_clear_data(gs);
1587}
1588
1589void grep_source_clear_data(struct grep_source *gs)
1590{
1591 switch (gs->type) {
1592 case GREP_SOURCE_FILE:
1593 case GREP_SOURCE_SHA1:
1594 free(gs->buf);
1595 gs->buf = NULL;
1596 gs->size = 0;
1597 break;
1598 case GREP_SOURCE_BUF:
1599 /* leave user-provided buf intact */
1600 break;
1601 }
1602}
1603
1604static int grep_source_load_sha1(struct grep_source *gs)
1605{
1606 enum object_type type;
1607
1608 grep_read_lock();
1609 gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
1610 grep_read_unlock();
1611
1612 if (!gs->buf)
1613 return error(_("'%s': unable to read %s"),
1614 gs->name,
1615 sha1_to_hex(gs->identifier));
1616 return 0;
1617}
1618
1619static int grep_source_load_file(struct grep_source *gs)
1620{
1621 const char *filename = gs->identifier;
1622 struct stat st;
1623 char *data;
1624 size_t size;
1625 int i;
1626
1627 if (lstat(filename, &st) < 0) {
1628 err_ret:
1629 if (errno != ENOENT)
1630 error(_("'%s': %s"), filename, strerror(errno));
1631 return -1;
1632 }
1633 if (!S_ISREG(st.st_mode))
1634 return -1;
1635 size = xsize_t(st.st_size);
1636 i = open(filename, O_RDONLY);
1637 if (i < 0)
1638 goto err_ret;
1639 data = xmalloc(size + 1);
1640 if (st.st_size != read_in_full(i, data, size)) {
1641 error(_("'%s': short read %s"), filename, strerror(errno));
1642 close(i);
1643 free(data);
1644 return -1;
1645 }
1646 close(i);
1647 data[size] = 0;
1648
1649 gs->buf = data;
1650 gs->size = size;
1651 return 0;
1652}
1653
3083301e 1654static int grep_source_load(struct grep_source *gs)
e1327023
JK
1655{
1656 if (gs->buf)
1657 return 0;
1658
1659 switch (gs->type) {
1660 case GREP_SOURCE_FILE:
1661 return grep_source_load_file(gs);
1662 case GREP_SOURCE_SHA1:
1663 return grep_source_load_sha1(gs);
1664 case GREP_SOURCE_BUF:
1665 return gs->buf ? 0 : -1;
1666 }
1667 die("BUG: invalid grep_source type");
0ab7befa 1668}
94ad9d9e
JK
1669
1670void grep_source_load_driver(struct grep_source *gs)
1671{
1672 if (gs->driver)
1673 return;
1674
1675 grep_attr_lock();
1676 gs->driver = userdiff_find_by_path(gs->name);
1677 if (!gs->driver)
1678 gs->driver = userdiff_find_by_name("default");
1679 grep_attr_unlock();
1680}
41b59bfc 1681
3083301e 1682static int grep_source_is_binary(struct grep_source *gs)
41b59bfc
JK
1683{
1684 grep_source_load_driver(gs);
1685 if (gs->driver->binary != -1)
1686 return gs->driver->binary;
1687
1688 if (!grep_source_load(gs))
1689 return buffer_is_binary(gs->buf, gs->size);
1690
1691 return 0;
1692}