Merge branch 'maint'
[git/vmiklos.git] / grep.c
blobbdadf2c0ccfafc18011beee3bc05754860392523
1 #include "cache.h"
2 #include "grep.h"
3 #include "userdiff.h"
4 #include "xdiff-interface.h"
6 void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat)
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;
14 *opt->pattern_tail = p;
15 opt->pattern_tail = &p->next;
16 p->next = NULL;
19 void append_grep_pattern(struct grep_opt *opt, const char *pat,
20 const char *origin, int no, enum grep_pat_token t)
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;
32 static int is_fixed(const char *s)
34 while (*s && !is_regex_special(*s))
35 s++;
36 return !*s;
39 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
41 int err;
43 p->word_regexp = opt->word_regexp;
44 p->ignore_case = opt->ignore_case;
46 if (opt->fixed || is_fixed(p->pattern))
47 p->fixed = 1;
48 if (opt->regflags & REG_ICASE)
49 p->fixed = 0;
50 if (p->fixed)
51 return;
53 err = regcomp(&p->regexp, p->pattern, opt->regflags);
54 if (err) {
55 char errbuf[1024];
56 char where[1024];
57 if (p->no)
58 sprintf(where, "In '%s' at %d, ",
59 p->origin, p->no);
60 else if (p->origin)
61 sprintf(where, "%s, ", p->origin);
62 else
63 where[0] = 0;
64 regerror(err, &p->regexp, errbuf, 1024);
65 regfree(&p->regexp);
66 die("%s'%s': %s", where, p->pattern, errbuf);
70 static struct grep_expr *compile_pattern_or(struct grep_pat **);
71 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
73 struct grep_pat *p;
74 struct grep_expr *x;
76 p = *list;
77 if (!p)
78 return NULL;
79 switch (p->token) {
80 case GREP_PATTERN: /* atom */
81 case GREP_PATTERN_HEAD:
82 case GREP_PATTERN_BODY:
83 x = xcalloc(1, sizeof (struct grep_expr));
84 x->node = GREP_NODE_ATOM;
85 x->u.atom = p;
86 *list = p->next;
87 return x;
88 case GREP_OPEN_PAREN:
89 *list = p->next;
90 x = compile_pattern_or(list);
91 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
92 die("unmatched parenthesis");
93 *list = (*list)->next;
94 return x;
95 default:
96 return NULL;
100 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
102 struct grep_pat *p;
103 struct grep_expr *x;
105 p = *list;
106 if (!p)
107 return NULL;
108 switch (p->token) {
109 case GREP_NOT:
110 if (!p->next)
111 die("--not not followed by pattern expression");
112 *list = p->next;
113 x = xcalloc(1, sizeof (struct grep_expr));
114 x->node = GREP_NODE_NOT;
115 x->u.unary = compile_pattern_not(list);
116 if (!x->u.unary)
117 die("--not followed by non pattern expression");
118 return x;
119 default:
120 return compile_pattern_atom(list);
124 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
126 struct grep_pat *p;
127 struct grep_expr *x, *y, *z;
129 x = compile_pattern_not(list);
130 p = *list;
131 if (p && p->token == GREP_AND) {
132 if (!p->next)
133 die("--and not followed by pattern expression");
134 *list = p->next;
135 y = compile_pattern_and(list);
136 if (!y)
137 die("--and not followed by pattern expression");
138 z = xcalloc(1, sizeof (struct grep_expr));
139 z->node = GREP_NODE_AND;
140 z->u.binary.left = x;
141 z->u.binary.right = y;
142 return z;
144 return x;
147 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
149 struct grep_pat *p;
150 struct grep_expr *x, *y, *z;
152 x = compile_pattern_and(list);
153 p = *list;
154 if (x && p && p->token != GREP_CLOSE_PAREN) {
155 y = compile_pattern_or(list);
156 if (!y)
157 die("not a pattern expression %s", p->pattern);
158 z = xcalloc(1, sizeof (struct grep_expr));
159 z->node = GREP_NODE_OR;
160 z->u.binary.left = x;
161 z->u.binary.right = y;
162 return z;
164 return x;
167 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
169 return compile_pattern_or(list);
172 void compile_grep_patterns(struct grep_opt *opt)
174 struct grep_pat *p;
176 if (opt->all_match)
177 opt->extended = 1;
179 for (p = opt->pattern_list; p; p = p->next) {
180 switch (p->token) {
181 case GREP_PATTERN: /* atom */
182 case GREP_PATTERN_HEAD:
183 case GREP_PATTERN_BODY:
184 compile_regexp(p, opt);
185 break;
186 default:
187 opt->extended = 1;
188 break;
192 if (!opt->extended)
193 return;
195 /* Then bundle them up in an expression.
196 * A classic recursive descent parser would do.
198 p = opt->pattern_list;
199 if (p)
200 opt->pattern_expression = compile_pattern_expr(&p);
201 if (p)
202 die("incomplete pattern expression: %s", p->pattern);
205 static void free_pattern_expr(struct grep_expr *x)
207 switch (x->node) {
208 case GREP_NODE_ATOM:
209 break;
210 case GREP_NODE_NOT:
211 free_pattern_expr(x->u.unary);
212 break;
213 case GREP_NODE_AND:
214 case GREP_NODE_OR:
215 free_pattern_expr(x->u.binary.left);
216 free_pattern_expr(x->u.binary.right);
217 break;
219 free(x);
222 void free_grep_patterns(struct grep_opt *opt)
224 struct grep_pat *p, *n;
226 for (p = opt->pattern_list; p; p = n) {
227 n = p->next;
228 switch (p->token) {
229 case GREP_PATTERN: /* atom */
230 case GREP_PATTERN_HEAD:
231 case GREP_PATTERN_BODY:
232 regfree(&p->regexp);
233 break;
234 default:
235 break;
237 free(p);
240 if (!opt->extended)
241 return;
242 free_pattern_expr(opt->pattern_expression);
245 static char *end_of_line(char *cp, unsigned long *left)
247 unsigned long l = *left;
248 while (l && *cp != '\n') {
249 l--;
250 cp++;
252 *left = l;
253 return cp;
256 static int word_char(char ch)
258 return isalnum(ch) || ch == '_';
261 static void show_name(struct grep_opt *opt, const char *name)
263 printf("%s%c", name, opt->null_following_name ? '\0' : '\n');
267 static int fixmatch(const char *pattern, char *line, int ignore_case, regmatch_t *match)
269 char *hit;
270 if (ignore_case)
271 hit = strcasestr(line, pattern);
272 else
273 hit = strstr(line, pattern);
275 if (!hit) {
276 match->rm_so = match->rm_eo = -1;
277 return REG_NOMATCH;
279 else {
280 match->rm_so = hit - line;
281 match->rm_eo = match->rm_so + strlen(pattern);
282 return 0;
286 static int strip_timestamp(char *bol, char **eol_p)
288 char *eol = *eol_p;
289 int ch;
291 while (bol < --eol) {
292 if (*eol != '>')
293 continue;
294 *eol_p = ++eol;
295 ch = *eol;
296 *eol = '\0';
297 return ch;
299 return 0;
302 static struct {
303 const char *field;
304 size_t len;
305 } header_field[] = {
306 { "author ", 7 },
307 { "committer ", 10 },
310 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
311 enum grep_context ctx,
312 regmatch_t *pmatch, int eflags)
314 int hit = 0;
315 int saved_ch = 0;
316 const char *start = bol;
318 if ((p->token != GREP_PATTERN) &&
319 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
320 return 0;
322 if (p->token == GREP_PATTERN_HEAD) {
323 const char *field;
324 size_t len;
325 assert(p->field < ARRAY_SIZE(header_field));
326 field = header_field[p->field].field;
327 len = header_field[p->field].len;
328 if (strncmp(bol, field, len))
329 return 0;
330 bol += len;
331 saved_ch = strip_timestamp(bol, &eol);
334 again:
335 if (p->fixed)
336 hit = !fixmatch(p->pattern, bol, p->ignore_case, pmatch);
337 else
338 hit = !regexec(&p->regexp, bol, 1, pmatch, eflags);
340 if (hit && p->word_regexp) {
341 if ((pmatch[0].rm_so < 0) ||
342 (eol - bol) < pmatch[0].rm_so ||
343 (pmatch[0].rm_eo < 0) ||
344 (eol - bol) < pmatch[0].rm_eo)
345 die("regexp returned nonsense");
347 /* Match beginning must be either beginning of the
348 * line, or at word boundary (i.e. the last char must
349 * not be a word char). Similarly, match end must be
350 * either end of the line, or at word boundary
351 * (i.e. the next char must not be a word char).
353 if ( ((pmatch[0].rm_so == 0) ||
354 !word_char(bol[pmatch[0].rm_so-1])) &&
355 ((pmatch[0].rm_eo == (eol-bol)) ||
356 !word_char(bol[pmatch[0].rm_eo])) )
358 else
359 hit = 0;
361 /* Words consist of at least one character. */
362 if (pmatch->rm_so == pmatch->rm_eo)
363 hit = 0;
365 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
366 /* There could be more than one match on the
367 * line, and the first match might not be
368 * strict word match. But later ones could be!
369 * Forward to the next possible start, i.e. the
370 * next position following a non-word char.
372 bol = pmatch[0].rm_so + bol + 1;
373 while (word_char(bol[-1]) && bol < eol)
374 bol++;
375 eflags |= REG_NOTBOL;
376 if (bol < eol)
377 goto again;
380 if (p->token == GREP_PATTERN_HEAD && saved_ch)
381 *eol = saved_ch;
382 if (hit) {
383 pmatch[0].rm_so += bol - start;
384 pmatch[0].rm_eo += bol - start;
386 return hit;
389 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
390 enum grep_context ctx, int collect_hits)
392 int h = 0;
393 regmatch_t match;
395 if (!x)
396 die("Not a valid grep expression");
397 switch (x->node) {
398 case GREP_NODE_ATOM:
399 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
400 break;
401 case GREP_NODE_NOT:
402 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
403 break;
404 case GREP_NODE_AND:
405 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
406 return 0;
407 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
408 break;
409 case GREP_NODE_OR:
410 if (!collect_hits)
411 return (match_expr_eval(x->u.binary.left,
412 bol, eol, ctx, 0) ||
413 match_expr_eval(x->u.binary.right,
414 bol, eol, ctx, 0));
415 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
416 x->u.binary.left->hit |= h;
417 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
418 break;
419 default:
420 die("Unexpected node type (internal error) %d", x->node);
422 if (collect_hits)
423 x->hit |= h;
424 return h;
427 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
428 enum grep_context ctx, int collect_hits)
430 struct grep_expr *x = opt->pattern_expression;
431 return match_expr_eval(x, bol, eol, ctx, collect_hits);
434 static int match_line(struct grep_opt *opt, char *bol, char *eol,
435 enum grep_context ctx, int collect_hits)
437 struct grep_pat *p;
438 regmatch_t match;
440 if (opt->extended)
441 return match_expr(opt, bol, eol, ctx, collect_hits);
443 /* we do not call with collect_hits without being extended */
444 for (p = opt->pattern_list; p; p = p->next) {
445 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
446 return 1;
448 return 0;
451 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
452 enum grep_context ctx,
453 regmatch_t *pmatch, int eflags)
455 regmatch_t match;
457 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
458 return 0;
459 if (match.rm_so < 0 || match.rm_eo < 0)
460 return 0;
461 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
462 if (match.rm_so > pmatch->rm_so)
463 return 1;
464 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
465 return 1;
467 pmatch->rm_so = match.rm_so;
468 pmatch->rm_eo = match.rm_eo;
469 return 1;
472 static int next_match(struct grep_opt *opt, char *bol, char *eol,
473 enum grep_context ctx, regmatch_t *pmatch, int eflags)
475 struct grep_pat *p;
476 int hit = 0;
478 pmatch->rm_so = pmatch->rm_eo = -1;
479 if (bol < eol) {
480 for (p = opt->pattern_list; p; p = p->next) {
481 switch (p->token) {
482 case GREP_PATTERN: /* atom */
483 case GREP_PATTERN_HEAD:
484 case GREP_PATTERN_BODY:
485 hit |= match_next_pattern(p, bol, eol, ctx,
486 pmatch, eflags);
487 break;
488 default:
489 break;
493 return hit;
496 static void show_line(struct grep_opt *opt, char *bol, char *eol,
497 const char *name, unsigned lno, char sign)
499 int rest = eol - bol;
501 if (opt->pre_context || opt->post_context) {
502 if (opt->last_shown == 0) {
503 if (opt->show_hunk_mark)
504 fputs("--\n", stdout);
505 else
506 opt->show_hunk_mark = 1;
507 } else if (lno > opt->last_shown + 1)
508 fputs("--\n", stdout);
510 opt->last_shown = lno;
512 if (opt->null_following_name)
513 sign = '\0';
514 if (opt->pathname)
515 printf("%s%c", name, sign);
516 if (opt->linenum)
517 printf("%d%c", lno, sign);
518 if (opt->color) {
519 regmatch_t match;
520 enum grep_context ctx = GREP_CONTEXT_BODY;
521 int ch = *eol;
522 int eflags = 0;
524 *eol = '\0';
525 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
526 if (match.rm_so == match.rm_eo)
527 break;
528 printf("%.*s%s%.*s%s",
529 (int)match.rm_so, bol,
530 opt->color_match,
531 (int)(match.rm_eo - match.rm_so), bol + match.rm_so,
532 GIT_COLOR_RESET);
533 bol += match.rm_eo;
534 rest -= match.rm_eo;
535 eflags = REG_NOTBOL;
537 *eol = ch;
539 printf("%.*s\n", rest, bol);
542 static int match_funcname(struct grep_opt *opt, char *bol, char *eol)
544 xdemitconf_t *xecfg = opt->priv;
545 if (xecfg && xecfg->find_func) {
546 char buf[1];
547 return xecfg->find_func(bol, eol - bol, buf, 1,
548 xecfg->find_func_priv) >= 0;
551 if (bol == eol)
552 return 0;
553 if (isalpha(*bol) || *bol == '_' || *bol == '$')
554 return 1;
555 return 0;
558 static void show_funcname_line(struct grep_opt *opt, const char *name,
559 char *buf, char *bol, unsigned lno)
561 while (bol > buf) {
562 char *eol = --bol;
564 while (bol > buf && bol[-1] != '\n')
565 bol--;
566 lno--;
568 if (lno <= opt->last_shown)
569 break;
571 if (match_funcname(opt, bol, eol)) {
572 show_line(opt, bol, eol, name, lno, '=');
573 break;
578 static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
579 char *bol, unsigned lno)
581 unsigned cur = lno, from = 1, funcname_lno = 0;
582 int funcname_needed = opt->funcname;
584 if (opt->pre_context < lno)
585 from = lno - opt->pre_context;
586 if (from <= opt->last_shown)
587 from = opt->last_shown + 1;
589 /* Rewind. */
590 while (bol > buf && cur > from) {
591 char *eol = --bol;
593 while (bol > buf && bol[-1] != '\n')
594 bol--;
595 cur--;
596 if (funcname_needed && match_funcname(opt, bol, eol)) {
597 funcname_lno = cur;
598 funcname_needed = 0;
602 /* We need to look even further back to find a function signature. */
603 if (opt->funcname && funcname_needed)
604 show_funcname_line(opt, name, buf, bol, cur);
606 /* Back forward. */
607 while (cur < lno) {
608 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
610 while (*eol != '\n')
611 eol++;
612 show_line(opt, bol, eol, name, cur, sign);
613 bol = eol + 1;
614 cur++;
618 static int grep_buffer_1(struct grep_opt *opt, const char *name,
619 char *buf, unsigned long size, int collect_hits)
621 char *bol = buf;
622 unsigned long left = size;
623 unsigned lno = 1;
624 unsigned last_hit = 0;
625 int binary_match_only = 0;
626 unsigned count = 0;
627 enum grep_context ctx = GREP_CONTEXT_HEAD;
628 xdemitconf_t xecfg;
630 opt->last_shown = 0;
632 if (buffer_is_binary(buf, size)) {
633 switch (opt->binary) {
634 case GREP_BINARY_DEFAULT:
635 binary_match_only = 1;
636 break;
637 case GREP_BINARY_NOMATCH:
638 return 0; /* Assume unmatch */
639 break;
640 default:
641 break;
645 memset(&xecfg, 0, sizeof(xecfg));
646 if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
647 !opt->name_only && !binary_match_only && !collect_hits) {
648 struct userdiff_driver *drv = userdiff_find_by_path(name);
649 if (drv && drv->funcname.pattern) {
650 const struct userdiff_funcname *pe = &drv->funcname;
651 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
652 opt->priv = &xecfg;
656 while (left) {
657 char *eol, ch;
658 int hit;
660 eol = end_of_line(bol, &left);
661 ch = *eol;
662 *eol = 0;
664 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
665 ctx = GREP_CONTEXT_BODY;
667 hit = match_line(opt, bol, eol, ctx, collect_hits);
668 *eol = ch;
670 if (collect_hits)
671 goto next_line;
673 /* "grep -v -e foo -e bla" should list lines
674 * that do not have either, so inversion should
675 * be done outside.
677 if (opt->invert)
678 hit = !hit;
679 if (opt->unmatch_name_only) {
680 if (hit)
681 return 0;
682 goto next_line;
684 if (hit) {
685 count++;
686 if (opt->status_only)
687 return 1;
688 if (binary_match_only) {
689 printf("Binary file %s matches\n", name);
690 return 1;
692 if (opt->name_only) {
693 show_name(opt, name);
694 return 1;
696 /* Hit at this line. If we haven't shown the
697 * pre-context lines, we would need to show them.
698 * When asked to do "count", this still show
699 * the context which is nonsense, but the user
700 * deserves to get that ;-).
702 if (opt->pre_context)
703 show_pre_context(opt, name, buf, bol, lno);
704 else if (opt->funcname)
705 show_funcname_line(opt, name, buf, bol, lno);
706 if (!opt->count)
707 show_line(opt, bol, eol, name, lno, ':');
708 last_hit = lno;
710 else if (last_hit &&
711 lno <= last_hit + opt->post_context) {
712 /* If the last hit is within the post context,
713 * we need to show this line.
715 show_line(opt, bol, eol, name, lno, '-');
718 next_line:
719 bol = eol + 1;
720 if (!left)
721 break;
722 left--;
723 lno++;
726 if (collect_hits)
727 return 0;
729 if (opt->status_only)
730 return 0;
731 if (opt->unmatch_name_only) {
732 /* We did not see any hit, so we want to show this */
733 show_name(opt, name);
734 return 1;
737 xdiff_clear_find_func(&xecfg);
738 opt->priv = NULL;
740 /* NEEDSWORK:
741 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
742 * which feels mostly useless but sometimes useful. Maybe
743 * make it another option? For now suppress them.
745 if (opt->count && count)
746 printf("%s%c%u\n", name,
747 opt->null_following_name ? '\0' : ':', count);
748 return !!last_hit;
751 static void clr_hit_marker(struct grep_expr *x)
753 /* All-hit markers are meaningful only at the very top level
754 * OR node.
756 while (1) {
757 x->hit = 0;
758 if (x->node != GREP_NODE_OR)
759 return;
760 x->u.binary.left->hit = 0;
761 x = x->u.binary.right;
765 static int chk_hit_marker(struct grep_expr *x)
767 /* Top level nodes have hit markers. See if they all are hits */
768 while (1) {
769 if (x->node != GREP_NODE_OR)
770 return x->hit;
771 if (!x->u.binary.left->hit)
772 return 0;
773 x = x->u.binary.right;
777 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
780 * we do not have to do the two-pass grep when we do not check
781 * buffer-wide "all-match".
783 if (!opt->all_match)
784 return grep_buffer_1(opt, name, buf, size, 0);
786 /* Otherwise the toplevel "or" terms hit a bit differently.
787 * We first clear hit markers from them.
789 clr_hit_marker(opt->pattern_expression);
790 grep_buffer_1(opt, name, buf, size, 1);
792 if (!chk_hit_marker(opt->pattern_expression))
793 return 0;
795 return grep_buffer_1(opt, name, buf, size, 0);