grep: add test script for binary file handling
[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
a4d7d2c6
JH
6void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat)
7{
8 struct grep_pat *p = xcalloc(1, sizeof(*p));
9 p->pattern = pat;
10 p->origin = "header";
11 p->no = 0;
12 p->token = GREP_PATTERN_HEAD;
13 p->field = field;
80235ba7
JH
14 *opt->header_tail = p;
15 opt->header_tail = &p->next;
a4d7d2c6
JH
16 p->next = NULL;
17}
18
83b5d2f5
JH
19void append_grep_pattern(struct grep_opt *opt, const char *pat,
20 const char *origin, int no, enum grep_pat_token t)
21{
22 struct grep_pat *p = xcalloc(1, sizeof(*p));
23 p->pattern = pat;
24 p->origin = origin;
25 p->no = no;
26 p->token = t;
27 *opt->pattern_tail = p;
28 opt->pattern_tail = &p->next;
29 p->next = NULL;
30}
31
5b594f45
FK
32struct grep_opt *grep_opt_dup(const struct grep_opt *opt)
33{
34 struct grep_pat *pat;
35 struct grep_opt *ret = xmalloc(sizeof(struct grep_opt));
36 *ret = *opt;
37
38 ret->pattern_list = NULL;
39 ret->pattern_tail = &ret->pattern_list;
40
41 for(pat = opt->pattern_list; pat != NULL; pat = pat->next)
42 {
43 if(pat->token == GREP_PATTERN_HEAD)
44 append_header_grep_pattern(ret, pat->field,
45 pat->pattern);
46 else
47 append_grep_pattern(ret, pat->pattern, pat->origin,
48 pat->no, pat->token);
49 }
50
51 return ret;
52}
53
83b5d2f5
JH
54static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
55{
c822255c
RS
56 int err;
57
d7eb527d 58 p->word_regexp = opt->word_regexp;
5183bf67 59 p->ignore_case = opt->ignore_case;
79286102 60 p->fixed = opt->fixed;
d7eb527d 61
c822255c
RS
62 if (p->fixed)
63 return;
64
65 err = regcomp(&p->regexp, p->pattern, opt->regflags);
83b5d2f5
JH
66 if (err) {
67 char errbuf[1024];
68 char where[1024];
69 if (p->no)
70 sprintf(where, "In '%s' at %d, ",
71 p->origin, p->no);
72 else if (p->origin)
73 sprintf(where, "%s, ", p->origin);
74 else
75 where[0] = 0;
76 regerror(err, &p->regexp, errbuf, 1024);
77 regfree(&p->regexp);
78 die("%s'%s': %s", where, p->pattern, errbuf);
79 }
80}
81
0ab7befa 82static struct grep_expr *compile_pattern_or(struct grep_pat **);
83b5d2f5
JH
83static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
84{
85 struct grep_pat *p;
86 struct grep_expr *x;
87
88 p = *list;
c922b01f
LT
89 if (!p)
90 return NULL;
83b5d2f5
JH
91 switch (p->token) {
92 case GREP_PATTERN: /* atom */
480c1ca6
JH
93 case GREP_PATTERN_HEAD:
94 case GREP_PATTERN_BODY:
83b5d2f5
JH
95 x = xcalloc(1, sizeof (struct grep_expr));
96 x->node = GREP_NODE_ATOM;
97 x->u.atom = p;
98 *list = p->next;
99 return x;
100 case GREP_OPEN_PAREN:
101 *list = p->next;
0ab7befa 102 x = compile_pattern_or(list);
83b5d2f5
JH
103 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
104 die("unmatched parenthesis");
105 *list = (*list)->next;
106 return x;
107 default:
108 return NULL;
109 }
110}
111
112static struct grep_expr *compile_pattern_not(struct grep_pat **list)
113{
114 struct grep_pat *p;
115 struct grep_expr *x;
116
117 p = *list;
c922b01f
LT
118 if (!p)
119 return NULL;
83b5d2f5
JH
120 switch (p->token) {
121 case GREP_NOT:
122 if (!p->next)
123 die("--not not followed by pattern expression");
124 *list = p->next;
125 x = xcalloc(1, sizeof (struct grep_expr));
126 x->node = GREP_NODE_NOT;
127 x->u.unary = compile_pattern_not(list);
128 if (!x->u.unary)
129 die("--not followed by non pattern expression");
130 return x;
131 default:
132 return compile_pattern_atom(list);
133 }
134}
135
136static struct grep_expr *compile_pattern_and(struct grep_pat **list)
137{
138 struct grep_pat *p;
139 struct grep_expr *x, *y, *z;
140
141 x = compile_pattern_not(list);
142 p = *list;
143 if (p && p->token == GREP_AND) {
144 if (!p->next)
145 die("--and not followed by pattern expression");
146 *list = p->next;
147 y = compile_pattern_and(list);
148 if (!y)
149 die("--and not followed by pattern expression");
150 z = xcalloc(1, sizeof (struct grep_expr));
151 z->node = GREP_NODE_AND;
152 z->u.binary.left = x;
153 z->u.binary.right = y;
154 return z;
155 }
156 return x;
157}
158
159static struct grep_expr *compile_pattern_or(struct grep_pat **list)
160{
161 struct grep_pat *p;
162 struct grep_expr *x, *y, *z;
163
164 x = compile_pattern_and(list);
165 p = *list;
166 if (x && p && p->token != GREP_CLOSE_PAREN) {
167 y = compile_pattern_or(list);
168 if (!y)
169 die("not a pattern expression %s", p->pattern);
170 z = xcalloc(1, sizeof (struct grep_expr));
171 z->node = GREP_NODE_OR;
172 z->u.binary.left = x;
173 z->u.binary.right = y;
174 return z;
175 }
176 return x;
177}
178
179static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
180{
181 return compile_pattern_or(list);
182}
183
184void compile_grep_patterns(struct grep_opt *opt)
185{
186 struct grep_pat *p;
80235ba7 187 struct grep_expr *header_expr = NULL;
83b5d2f5 188
80235ba7
JH
189 if (opt->header_list) {
190 p = opt->header_list;
191 header_expr = compile_pattern_expr(&p);
192 if (p)
193 die("incomplete pattern expression: %s", p->pattern);
194 for (p = opt->header_list; p; p = p->next) {
195 switch (p->token) {
196 case GREP_PATTERN: /* atom */
197 case GREP_PATTERN_HEAD:
198 case GREP_PATTERN_BODY:
199 compile_regexp(p, opt);
200 break;
201 default:
202 opt->extended = 1;
203 break;
204 }
205 }
206 }
0ab7befa 207
83b5d2f5 208 for (p = opt->pattern_list; p; p = p->next) {
480c1ca6
JH
209 switch (p->token) {
210 case GREP_PATTERN: /* atom */
211 case GREP_PATTERN_HEAD:
212 case GREP_PATTERN_BODY:
c822255c 213 compile_regexp(p, opt);
480c1ca6
JH
214 break;
215 default:
83b5d2f5 216 opt->extended = 1;
480c1ca6
JH
217 break;
218 }
83b5d2f5
JH
219 }
220
80235ba7
JH
221 if (opt->all_match || header_expr)
222 opt->extended = 1;
223 else if (!opt->extended)
83b5d2f5
JH
224 return;
225
226 /* Then bundle them up in an expression.
227 * A classic recursive descent parser would do.
228 */
229 p = opt->pattern_list;
ba150a3f
MB
230 if (p)
231 opt->pattern_expression = compile_pattern_expr(&p);
83b5d2f5
JH
232 if (p)
233 die("incomplete pattern expression: %s", p->pattern);
80235ba7
JH
234
235 if (!header_expr)
236 return;
237
238 if (opt->pattern_expression) {
239 struct grep_expr *z;
240 z = xcalloc(1, sizeof(*z));
241 z->node = GREP_NODE_OR;
242 z->u.binary.left = opt->pattern_expression;
243 z->u.binary.right = header_expr;
244 opt->pattern_expression = z;
245 } else {
246 opt->pattern_expression = header_expr;
247 }
248 opt->all_match = 1;
83b5d2f5
JH
249}
250
b48fb5b6
JH
251static void free_pattern_expr(struct grep_expr *x)
252{
253 switch (x->node) {
254 case GREP_NODE_ATOM:
255 break;
256 case GREP_NODE_NOT:
257 free_pattern_expr(x->u.unary);
258 break;
259 case GREP_NODE_AND:
260 case GREP_NODE_OR:
261 free_pattern_expr(x->u.binary.left);
262 free_pattern_expr(x->u.binary.right);
263 break;
264 }
265 free(x);
266}
267
268void free_grep_patterns(struct grep_opt *opt)
269{
270 struct grep_pat *p, *n;
271
272 for (p = opt->pattern_list; p; p = n) {
273 n = p->next;
274 switch (p->token) {
275 case GREP_PATTERN: /* atom */
276 case GREP_PATTERN_HEAD:
277 case GREP_PATTERN_BODY:
278 regfree(&p->regexp);
279 break;
280 default:
281 break;
282 }
283 free(p);
284 }
285
286 if (!opt->extended)
287 return;
288 free_pattern_expr(opt->pattern_expression);
289}
290
83b5d2f5
JH
291static char *end_of_line(char *cp, unsigned long *left)
292{
293 unsigned long l = *left;
294 while (l && *cp != '\n') {
295 l--;
296 cp++;
297 }
298 *left = l;
299 return cp;
300}
301
302static int word_char(char ch)
303{
304 return isalnum(ch) || ch == '_';
305}
306
55f638bd
ML
307static void output_color(struct grep_opt *opt, const void *data, size_t size,
308 const char *color)
309{
310 if (opt->color && color && color[0]) {
311 opt->output(opt, color, strlen(color));
312 opt->output(opt, data, size);
313 opt->output(opt, GIT_COLOR_RESET, strlen(GIT_COLOR_RESET));
314 } else
315 opt->output(opt, data, size);
316}
317
318static void output_sep(struct grep_opt *opt, char sign)
319{
320 if (opt->null_following_name)
321 opt->output(opt, "\0", 1);
322 else
323 output_color(opt, &sign, 1, opt->color_sep);
324}
325
83caecca
RZ
326static void show_name(struct grep_opt *opt, const char *name)
327{
55f638bd 328 output_color(opt, name, strlen(name), opt->color_filename);
5b594f45 329 opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
83caecca
RZ
330}
331
5183bf67
BC
332
333static int fixmatch(const char *pattern, char *line, int ignore_case, regmatch_t *match)
83b5d2f5 334{
5183bf67
BC
335 char *hit;
336 if (ignore_case)
337 hit = strcasestr(line, pattern);
338 else
339 hit = strstr(line, pattern);
340
83b5d2f5
JH
341 if (!hit) {
342 match->rm_so = match->rm_eo = -1;
343 return REG_NOMATCH;
344 }
345 else {
346 match->rm_so = hit - line;
347 match->rm_eo = match->rm_so + strlen(pattern);
348 return 0;
349 }
350}
351
a4d7d2c6
JH
352static int strip_timestamp(char *bol, char **eol_p)
353{
354 char *eol = *eol_p;
355 int ch;
356
357 while (bol < --eol) {
358 if (*eol != '>')
359 continue;
360 *eol_p = ++eol;
361 ch = *eol;
362 *eol = '\0';
363 return ch;
364 }
365 return 0;
366}
367
368static struct {
369 const char *field;
370 size_t len;
371} header_field[] = {
372 { "author ", 7 },
373 { "committer ", 10 },
374};
375
d7eb527d 376static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
79212772
RS
377 enum grep_context ctx,
378 regmatch_t *pmatch, int eflags)
83b5d2f5
JH
379{
380 int hit = 0;
a4d7d2c6 381 int saved_ch = 0;
e701fadb 382 const char *start = bol;
83b5d2f5 383
480c1ca6
JH
384 if ((p->token != GREP_PATTERN) &&
385 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
386 return 0;
387
a4d7d2c6
JH
388 if (p->token == GREP_PATTERN_HEAD) {
389 const char *field;
390 size_t len;
391 assert(p->field < ARRAY_SIZE(header_field));
392 field = header_field[p->field].field;
393 len = header_field[p->field].len;
394 if (strncmp(bol, field, len))
395 return 0;
396 bol += len;
397 saved_ch = strip_timestamp(bol, &eol);
398 }
399
83b5d2f5 400 again:
79212772 401 if (p->fixed)
5183bf67 402 hit = !fixmatch(p->pattern, bol, p->ignore_case, pmatch);
79212772
RS
403 else
404 hit = !regexec(&p->regexp, bol, 1, pmatch, eflags);
83b5d2f5 405
d7eb527d 406 if (hit && p->word_regexp) {
83b5d2f5 407 if ((pmatch[0].rm_so < 0) ||
84201eae 408 (eol - bol) < pmatch[0].rm_so ||
83b5d2f5
JH
409 (pmatch[0].rm_eo < 0) ||
410 (eol - bol) < pmatch[0].rm_eo)
411 die("regexp returned nonsense");
412
413 /* Match beginning must be either beginning of the
414 * line, or at word boundary (i.e. the last char must
415 * not be a word char). Similarly, match end must be
416 * either end of the line, or at word boundary
417 * (i.e. the next char must not be a word char).
418 */
fb62eb7f 419 if ( ((pmatch[0].rm_so == 0) ||
83b5d2f5
JH
420 !word_char(bol[pmatch[0].rm_so-1])) &&
421 ((pmatch[0].rm_eo == (eol-bol)) ||
422 !word_char(bol[pmatch[0].rm_eo])) )
423 ;
424 else
425 hit = 0;
426
84201eae
RS
427 /* Words consist of at least one character. */
428 if (pmatch->rm_so == pmatch->rm_eo)
429 hit = 0;
430
83b5d2f5
JH
431 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
432 /* There could be more than one match on the
433 * line, and the first match might not be
434 * strict word match. But later ones could be!
fb62eb7f
RS
435 * Forward to the next possible start, i.e. the
436 * next position following a non-word char.
83b5d2f5
JH
437 */
438 bol = pmatch[0].rm_so + bol + 1;
fb62eb7f
RS
439 while (word_char(bol[-1]) && bol < eol)
440 bol++;
dbb6a4ad 441 eflags |= REG_NOTBOL;
fb62eb7f
RS
442 if (bol < eol)
443 goto again;
83b5d2f5
JH
444 }
445 }
a4d7d2c6
JH
446 if (p->token == GREP_PATTERN_HEAD && saved_ch)
447 *eol = saved_ch;
e701fadb
RS
448 if (hit) {
449 pmatch[0].rm_so += bol - start;
450 pmatch[0].rm_eo += bol - start;
451 }
83b5d2f5
JH
452 return hit;
453}
454
d7eb527d
RS
455static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
456 enum grep_context ctx, int collect_hits)
83b5d2f5 457{
0ab7befa 458 int h = 0;
79212772 459 regmatch_t match;
0ab7befa 460
c922b01f
LT
461 if (!x)
462 die("Not a valid grep expression");
83b5d2f5
JH
463 switch (x->node) {
464 case GREP_NODE_ATOM:
79212772 465 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
83b5d2f5
JH
466 break;
467 case GREP_NODE_NOT:
d7eb527d 468 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
0ab7befa 469 break;
83b5d2f5 470 case GREP_NODE_AND:
d7eb527d 471 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
252d560d 472 return 0;
d7eb527d 473 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
0ab7befa 474 break;
83b5d2f5 475 case GREP_NODE_OR:
0ab7befa 476 if (!collect_hits)
d7eb527d 477 return (match_expr_eval(x->u.binary.left,
0ab7befa 478 bol, eol, ctx, 0) ||
d7eb527d 479 match_expr_eval(x->u.binary.right,
0ab7befa 480 bol, eol, ctx, 0));
d7eb527d 481 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
0ab7befa 482 x->u.binary.left->hit |= h;
d7eb527d 483 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
0ab7befa
JH
484 break;
485 default:
d7530708 486 die("Unexpected node type (internal error) %d", x->node);
83b5d2f5 487 }
0ab7befa
JH
488 if (collect_hits)
489 x->hit |= h;
490 return h;
83b5d2f5
JH
491}
492
480c1ca6 493static int match_expr(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 494 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
495{
496 struct grep_expr *x = opt->pattern_expression;
d7eb527d 497 return match_expr_eval(x, bol, eol, ctx, collect_hits);
83b5d2f5
JH
498}
499
480c1ca6 500static int match_line(struct grep_opt *opt, char *bol, char *eol,
0ab7befa 501 enum grep_context ctx, int collect_hits)
83b5d2f5
JH
502{
503 struct grep_pat *p;
79212772
RS
504 regmatch_t match;
505
83b5d2f5 506 if (opt->extended)
0ab7befa
JH
507 return match_expr(opt, bol, eol, ctx, collect_hits);
508
509 /* we do not call with collect_hits without being extended */
83b5d2f5 510 for (p = opt->pattern_list; p; p = p->next) {
79212772 511 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
83b5d2f5
JH
512 return 1;
513 }
514 return 0;
515}
516
7e8f59d5
RS
517static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
518 enum grep_context ctx,
519 regmatch_t *pmatch, int eflags)
520{
521 regmatch_t match;
522
523 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
524 return 0;
525 if (match.rm_so < 0 || match.rm_eo < 0)
526 return 0;
527 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
528 if (match.rm_so > pmatch->rm_so)
529 return 1;
530 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
531 return 1;
532 }
533 pmatch->rm_so = match.rm_so;
534 pmatch->rm_eo = match.rm_eo;
535 return 1;
536}
537
538static int next_match(struct grep_opt *opt, char *bol, char *eol,
539 enum grep_context ctx, regmatch_t *pmatch, int eflags)
540{
541 struct grep_pat *p;
542 int hit = 0;
543
544 pmatch->rm_so = pmatch->rm_eo = -1;
545 if (bol < eol) {
546 for (p = opt->pattern_list; p; p = p->next) {
547 switch (p->token) {
548 case GREP_PATTERN: /* atom */
549 case GREP_PATTERN_HEAD:
550 case GREP_PATTERN_BODY:
551 hit |= match_next_pattern(p, bol, eol, ctx,
552 pmatch, eflags);
553 break;
554 default:
555 break;
556 }
557 }
558 }
559 return hit;
560}
561
562static void show_line(struct grep_opt *opt, char *bol, char *eol,
563 const char *name, unsigned lno, char sign)
564{
565 int rest = eol - bol;
00588bb5 566 char *line_color = NULL;
7e8f59d5 567
ed24e401 568 if (opt->pre_context || opt->post_context) {
046802d0 569 if (opt->last_shown == 0) {
55f638bd
ML
570 if (opt->show_hunk_mark) {
571 output_color(opt, "--", 2, opt->color_sep);
572 opt->output(opt, "\n", 1);
07b838f0 573 }
55f638bd
ML
574 } else if (lno > opt->last_shown + 1) {
575 output_color(opt, "--", 2, opt->color_sep);
576 opt->output(opt, "\n", 1);
577 }
5dd06d38
RS
578 }
579 opt->last_shown = lno;
580
5b594f45 581 if (opt->pathname) {
55f638bd
ML
582 output_color(opt, name, strlen(name), opt->color_filename);
583 output_sep(opt, sign);
5b594f45
FK
584 }
585 if (opt->linenum) {
586 char buf[32];
587 snprintf(buf, sizeof(buf), "%d", lno);
55f638bd
ML
588 output_color(opt, buf, strlen(buf), opt->color_lineno);
589 output_sep(opt, sign);
5b594f45 590 }
7e8f59d5
RS
591 if (opt->color) {
592 regmatch_t match;
593 enum grep_context ctx = GREP_CONTEXT_BODY;
594 int ch = *eol;
595 int eflags = 0;
596
00588bb5
ML
597 if (sign == ':')
598 line_color = opt->color_selected;
599 else if (sign == '-')
600 line_color = opt->color_context;
601 else if (sign == '=')
602 line_color = opt->color_function;
7e8f59d5
RS
603 *eol = '\0';
604 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
1f5b9cc4
RS
605 if (match.rm_so == match.rm_eo)
606 break;
5b594f45 607
00588bb5 608 output_color(opt, bol, match.rm_so, line_color);
55f638bd
ML
609 output_color(opt, bol + match.rm_so,
610 match.rm_eo - match.rm_so,
611 opt->color_match);
7e8f59d5
RS
612 bol += match.rm_eo;
613 rest -= match.rm_eo;
614 eflags = REG_NOTBOL;
615 }
616 *eol = ch;
617 }
00588bb5 618 output_color(opt, bol, rest, line_color);
5b594f45 619 opt->output(opt, "\n", 1);
7e8f59d5
RS
620}
621
60ecac98 622static int match_funcname(struct grep_opt *opt, char *bol, char *eol)
2944e4e6 623{
60ecac98
RS
624 xdemitconf_t *xecfg = opt->priv;
625 if (xecfg && xecfg->find_func) {
626 char buf[1];
627 return xecfg->find_func(bol, eol - bol, buf, 1,
628 xecfg->find_func_priv) >= 0;
629 }
630
2944e4e6
RS
631 if (bol == eol)
632 return 0;
633 if (isalpha(*bol) || *bol == '_' || *bol == '$')
634 return 1;
635 return 0;
636}
637
638static void show_funcname_line(struct grep_opt *opt, const char *name,
639 char *buf, char *bol, unsigned lno)
640{
641 while (bol > buf) {
642 char *eol = --bol;
643
644 while (bol > buf && bol[-1] != '\n')
645 bol--;
646 lno--;
647
648 if (lno <= opt->last_shown)
649 break;
650
60ecac98 651 if (match_funcname(opt, bol, eol)) {
2944e4e6
RS
652 show_line(opt, bol, eol, name, lno, '=');
653 break;
654 }
655 }
656}
657
49de3216
RS
658static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
659 char *bol, unsigned lno)
660{
2944e4e6
RS
661 unsigned cur = lno, from = 1, funcname_lno = 0;
662 int funcname_needed = opt->funcname;
49de3216
RS
663
664 if (opt->pre_context < lno)
665 from = lno - opt->pre_context;
666 if (from <= opt->last_shown)
667 from = opt->last_shown + 1;
668
669 /* Rewind. */
670 while (bol > buf && cur > from) {
2944e4e6
RS
671 char *eol = --bol;
672
49de3216
RS
673 while (bol > buf && bol[-1] != '\n')
674 bol--;
675 cur--;
60ecac98 676 if (funcname_needed && match_funcname(opt, bol, eol)) {
2944e4e6
RS
677 funcname_lno = cur;
678 funcname_needed = 0;
679 }
49de3216
RS
680 }
681
2944e4e6
RS
682 /* We need to look even further back to find a function signature. */
683 if (opt->funcname && funcname_needed)
684 show_funcname_line(opt, name, buf, bol, cur);
685
49de3216
RS
686 /* Back forward. */
687 while (cur < lno) {
2944e4e6 688 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
49de3216
RS
689
690 while (*eol != '\n')
691 eol++;
2944e4e6 692 show_line(opt, bol, eol, name, cur, sign);
49de3216
RS
693 bol = eol + 1;
694 cur++;
695 }
696}
697
a26345b6
JH
698static int should_lookahead(struct grep_opt *opt)
699{
700 struct grep_pat *p;
701
702 if (opt->extended)
703 return 0; /* punt for too complex stuff */
704 if (opt->invert)
705 return 0;
706 for (p = opt->pattern_list; p; p = p->next) {
707 if (p->token != GREP_PATTERN)
708 return 0; /* punt for "header only" and stuff */
709 }
710 return 1;
711}
712
713static int look_ahead(struct grep_opt *opt,
714 unsigned long *left_p,
715 unsigned *lno_p,
716 char **bol_p)
717{
718 unsigned lno = *lno_p;
719 char *bol = *bol_p;
720 struct grep_pat *p;
721 char *sp, *last_bol;
722 regoff_t earliest = -1;
723
724 for (p = opt->pattern_list; p; p = p->next) {
725 int hit;
726 regmatch_t m;
727
728 if (p->fixed)
e2d2e383 729 hit = !fixmatch(p->pattern, bol, p->ignore_case, &m);
24072c02
BK
730 else {
731#ifdef REG_STARTEND
732 m.rm_so = 0;
733 m.rm_eo = *left_p;
734 hit = !regexec(&p->regexp, bol, 1, &m, REG_STARTEND);
735#else
a26345b6 736 hit = !regexec(&p->regexp, bol, 1, &m, 0);
24072c02
BK
737#endif
738 }
a26345b6
JH
739 if (!hit || m.rm_so < 0 || m.rm_eo < 0)
740 continue;
741 if (earliest < 0 || m.rm_so < earliest)
742 earliest = m.rm_so;
743 }
744
745 if (earliest < 0) {
746 *bol_p = bol + *left_p;
747 *left_p = 0;
748 return 1;
749 }
750 for (sp = bol + earliest; bol < sp && sp[-1] != '\n'; sp--)
751 ; /* find the beginning of the line */
752 last_bol = sp;
753
754 for (sp = bol; sp < last_bol; sp++) {
755 if (*sp == '\n')
756 lno++;
757 }
758 *left_p -= last_bol - bol;
759 *bol_p = last_bol;
760 *lno_p = lno;
761 return 0;
762}
763
5b594f45
FK
764int grep_threads_ok(const struct grep_opt *opt)
765{
766 /* If this condition is true, then we may use the attribute
767 * machinery in grep_buffer_1. The attribute code is not
768 * thread safe, so we disable the use of threads.
769 */
770 if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
771 !opt->name_only)
772 return 0;
773
5b594f45
FK
774 return 1;
775}
776
777static void std_output(struct grep_opt *opt, const void *buf, size_t size)
778{
779 fwrite(buf, size, 1, stdout);
780}
781
0ab7befa
JH
782static int grep_buffer_1(struct grep_opt *opt, const char *name,
783 char *buf, unsigned long size, int collect_hits)
83b5d2f5
JH
784{
785 char *bol = buf;
786 unsigned long left = size;
787 unsigned lno = 1;
83b5d2f5 788 unsigned last_hit = 0;
83b5d2f5 789 int binary_match_only = 0;
83b5d2f5 790 unsigned count = 0;
a26345b6 791 int try_lookahead = 0;
480c1ca6 792 enum grep_context ctx = GREP_CONTEXT_HEAD;
60ecac98 793 xdemitconf_t xecfg;
83b5d2f5 794
5b594f45
FK
795 if (!opt->output)
796 opt->output = std_output;
797
431d6e7b
RS
798 if (opt->last_shown && (opt->pre_context || opt->post_context) &&
799 opt->output == std_output)
800 opt->show_hunk_mark = 1;
801 opt->last_shown = 0;
802
83b5d2f5
JH
803 if (buffer_is_binary(buf, size)) {
804 switch (opt->binary) {
805 case GREP_BINARY_DEFAULT:
806 binary_match_only = 1;
807 break;
808 case GREP_BINARY_NOMATCH:
809 return 0; /* Assume unmatch */
810 break;
811 default:
812 break;
813 }
814 }
815
60ecac98
RS
816 memset(&xecfg, 0, sizeof(xecfg));
817 if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
818 !opt->name_only && !binary_match_only && !collect_hits) {
819 struct userdiff_driver *drv = userdiff_find_by_path(name);
820 if (drv && drv->funcname.pattern) {
821 const struct userdiff_funcname *pe = &drv->funcname;
822 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
823 opt->priv = &xecfg;
824 }
825 }
a26345b6 826 try_lookahead = should_lookahead(opt);
60ecac98 827
83b5d2f5
JH
828 while (left) {
829 char *eol, ch;
0ab7befa 830 int hit;
83b5d2f5 831
a26345b6
JH
832 /*
833 * look_ahead() skips quicly to the line that possibly
834 * has the next hit; don't call it if we need to do
835 * something more than just skipping the current line
836 * in response to an unmatch for the current line. E.g.
837 * inside a post-context window, we will show the current
838 * line as a context around the previous hit when it
839 * doesn't hit.
840 */
841 if (try_lookahead
842 && !(last_hit
843 && lno <= last_hit + opt->post_context)
844 && look_ahead(opt, &left, &lno, &bol))
845 break;
83b5d2f5
JH
846 eol = end_of_line(bol, &left);
847 ch = *eol;
848 *eol = 0;
849
480c1ca6
JH
850 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
851 ctx = GREP_CONTEXT_BODY;
852
0ab7befa 853 hit = match_line(opt, bol, eol, ctx, collect_hits);
83b5d2f5
JH
854 *eol = ch;
855
0ab7befa
JH
856 if (collect_hits)
857 goto next_line;
858
83b5d2f5
JH
859 /* "grep -v -e foo -e bla" should list lines
860 * that do not have either, so inversion should
861 * be done outside.
862 */
863 if (opt->invert)
864 hit = !hit;
865 if (opt->unmatch_name_only) {
866 if (hit)
867 return 0;
868 goto next_line;
869 }
870 if (hit) {
871 count++;
872 if (opt->status_only)
873 return 1;
874 if (binary_match_only) {
5b594f45 875 opt->output(opt, "Binary file ", 12);
55f638bd
ML
876 output_color(opt, name, strlen(name),
877 opt->color_filename);
5b594f45 878 opt->output(opt, " matches\n", 9);
83b5d2f5
JH
879 return 1;
880 }
881 if (opt->name_only) {
83caecca 882 show_name(opt, name);
83b5d2f5
JH
883 return 1;
884 }
885 /* Hit at this line. If we haven't shown the
886 * pre-context lines, we would need to show them.
887 * When asked to do "count", this still show
888 * the context which is nonsense, but the user
889 * deserves to get that ;-).
890 */
49de3216
RS
891 if (opt->pre_context)
892 show_pre_context(opt, name, buf, bol, lno);
2944e4e6
RS
893 else if (opt->funcname)
894 show_funcname_line(opt, name, buf, bol, lno);
83b5d2f5
JH
895 if (!opt->count)
896 show_line(opt, bol, eol, name, lno, ':');
5dd06d38 897 last_hit = lno;
83b5d2f5
JH
898 }
899 else if (last_hit &&
900 lno <= last_hit + opt->post_context) {
901 /* If the last hit is within the post context,
902 * we need to show this line.
903 */
83b5d2f5 904 show_line(opt, bol, eol, name, lno, '-');
83b5d2f5 905 }
83b5d2f5
JH
906
907 next_line:
908 bol = eol + 1;
909 if (!left)
910 break;
911 left--;
912 lno++;
913 }
914
0ab7befa
JH
915 if (collect_hits)
916 return 0;
b48fb5b6 917
83b5d2f5
JH
918 if (opt->status_only)
919 return 0;
920 if (opt->unmatch_name_only) {
921 /* We did not see any hit, so we want to show this */
83caecca 922 show_name(opt, name);
83b5d2f5
JH
923 return 1;
924 }
925
60ecac98
RS
926 xdiff_clear_find_func(&xecfg);
927 opt->priv = NULL;
928
83b5d2f5
JH
929 /* NEEDSWORK:
930 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
931 * which feels mostly useless but sometimes useful. Maybe
932 * make it another option? For now suppress them.
933 */
5b594f45
FK
934 if (opt->count && count) {
935 char buf[32];
55f638bd
ML
936 output_color(opt, name, strlen(name), opt->color_filename);
937 output_sep(opt, ':');
938 snprintf(buf, sizeof(buf), "%u\n", count);
5b594f45
FK
939 opt->output(opt, buf, strlen(buf));
940 }
83b5d2f5
JH
941 return !!last_hit;
942}
943
0ab7befa
JH
944static void clr_hit_marker(struct grep_expr *x)
945{
946 /* All-hit markers are meaningful only at the very top level
947 * OR node.
948 */
949 while (1) {
950 x->hit = 0;
951 if (x->node != GREP_NODE_OR)
952 return;
953 x->u.binary.left->hit = 0;
954 x = x->u.binary.right;
955 }
956}
957
958static int chk_hit_marker(struct grep_expr *x)
959{
960 /* Top level nodes have hit markers. See if they all are hits */
961 while (1) {
962 if (x->node != GREP_NODE_OR)
963 return x->hit;
964 if (!x->u.binary.left->hit)
965 return 0;
966 x = x->u.binary.right;
967 }
968}
969
970int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
971{
972 /*
973 * we do not have to do the two-pass grep when we do not check
974 * buffer-wide "all-match".
975 */
976 if (!opt->all_match)
977 return grep_buffer_1(opt, name, buf, size, 0);
978
979 /* Otherwise the toplevel "or" terms hit a bit differently.
980 * We first clear hit markers from them.
981 */
982 clr_hit_marker(opt->pattern_expression);
983 grep_buffer_1(opt, name, buf, size, 1);
984
985 if (!chk_hit_marker(opt->pattern_expression))
986 return 0;
987
988 return grep_buffer_1(opt, name, buf, size, 0);
989}