Merge branch 'rs/grep-p' into next
[git/dscho.git] / grep.c
blobc47785a2f0df79fd5ee8bc77a165fb0be3811091
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;
45 if (opt->fixed || is_fixed(p->pattern))
46 p->fixed = 1;
47 if (opt->regflags & REG_ICASE)
48 p->fixed = 0;
49 if (p->fixed)
50 return;
52 err = regcomp(&p->regexp, p->pattern, opt->regflags);
53 if (err) {
54 char errbuf[1024];
55 char where[1024];
56 if (p->no)
57 sprintf(where, "In '%s' at %d, ",
58 p->origin, p->no);
59 else if (p->origin)
60 sprintf(where, "%s, ", p->origin);
61 else
62 where[0] = 0;
63 regerror(err, &p->regexp, errbuf, 1024);
64 regfree(&p->regexp);
65 die("%s'%s': %s", where, p->pattern, errbuf);
69 static struct grep_expr *compile_pattern_or(struct grep_pat **);
70 static struct grep_expr *compile_pattern_atom(struct grep_pat **list)
72 struct grep_pat *p;
73 struct grep_expr *x;
75 p = *list;
76 if (!p)
77 return NULL;
78 switch (p->token) {
79 case GREP_PATTERN: /* atom */
80 case GREP_PATTERN_HEAD:
81 case GREP_PATTERN_BODY:
82 x = xcalloc(1, sizeof (struct grep_expr));
83 x->node = GREP_NODE_ATOM;
84 x->u.atom = p;
85 *list = p->next;
86 return x;
87 case GREP_OPEN_PAREN:
88 *list = p->next;
89 x = compile_pattern_or(list);
90 if (!*list || (*list)->token != GREP_CLOSE_PAREN)
91 die("unmatched parenthesis");
92 *list = (*list)->next;
93 return x;
94 default:
95 return NULL;
99 static struct grep_expr *compile_pattern_not(struct grep_pat **list)
101 struct grep_pat *p;
102 struct grep_expr *x;
104 p = *list;
105 if (!p)
106 return NULL;
107 switch (p->token) {
108 case GREP_NOT:
109 if (!p->next)
110 die("--not not followed by pattern expression");
111 *list = p->next;
112 x = xcalloc(1, sizeof (struct grep_expr));
113 x->node = GREP_NODE_NOT;
114 x->u.unary = compile_pattern_not(list);
115 if (!x->u.unary)
116 die("--not followed by non pattern expression");
117 return x;
118 default:
119 return compile_pattern_atom(list);
123 static struct grep_expr *compile_pattern_and(struct grep_pat **list)
125 struct grep_pat *p;
126 struct grep_expr *x, *y, *z;
128 x = compile_pattern_not(list);
129 p = *list;
130 if (p && p->token == GREP_AND) {
131 if (!p->next)
132 die("--and not followed by pattern expression");
133 *list = p->next;
134 y = compile_pattern_and(list);
135 if (!y)
136 die("--and not followed by pattern expression");
137 z = xcalloc(1, sizeof (struct grep_expr));
138 z->node = GREP_NODE_AND;
139 z->u.binary.left = x;
140 z->u.binary.right = y;
141 return z;
143 return x;
146 static struct grep_expr *compile_pattern_or(struct grep_pat **list)
148 struct grep_pat *p;
149 struct grep_expr *x, *y, *z;
151 x = compile_pattern_and(list);
152 p = *list;
153 if (x && p && p->token != GREP_CLOSE_PAREN) {
154 y = compile_pattern_or(list);
155 if (!y)
156 die("not a pattern expression %s", p->pattern);
157 z = xcalloc(1, sizeof (struct grep_expr));
158 z->node = GREP_NODE_OR;
159 z->u.binary.left = x;
160 z->u.binary.right = y;
161 return z;
163 return x;
166 static struct grep_expr *compile_pattern_expr(struct grep_pat **list)
168 return compile_pattern_or(list);
171 void compile_grep_patterns(struct grep_opt *opt)
173 struct grep_pat *p;
175 if (opt->all_match)
176 opt->extended = 1;
178 for (p = opt->pattern_list; p; p = p->next) {
179 switch (p->token) {
180 case GREP_PATTERN: /* atom */
181 case GREP_PATTERN_HEAD:
182 case GREP_PATTERN_BODY:
183 compile_regexp(p, opt);
184 break;
185 default:
186 opt->extended = 1;
187 break;
191 if (!opt->extended)
192 return;
194 /* Then bundle them up in an expression.
195 * A classic recursive descent parser would do.
197 p = opt->pattern_list;
198 if (p)
199 opt->pattern_expression = compile_pattern_expr(&p);
200 if (p)
201 die("incomplete pattern expression: %s", p->pattern);
204 static void free_pattern_expr(struct grep_expr *x)
206 switch (x->node) {
207 case GREP_NODE_ATOM:
208 break;
209 case GREP_NODE_NOT:
210 free_pattern_expr(x->u.unary);
211 break;
212 case GREP_NODE_AND:
213 case GREP_NODE_OR:
214 free_pattern_expr(x->u.binary.left);
215 free_pattern_expr(x->u.binary.right);
216 break;
218 free(x);
221 void free_grep_patterns(struct grep_opt *opt)
223 struct grep_pat *p, *n;
225 for (p = opt->pattern_list; p; p = n) {
226 n = p->next;
227 switch (p->token) {
228 case GREP_PATTERN: /* atom */
229 case GREP_PATTERN_HEAD:
230 case GREP_PATTERN_BODY:
231 regfree(&p->regexp);
232 break;
233 default:
234 break;
236 free(p);
239 if (!opt->extended)
240 return;
241 free_pattern_expr(opt->pattern_expression);
244 static char *end_of_line(char *cp, unsigned long *left)
246 unsigned long l = *left;
247 while (l && *cp != '\n') {
248 l--;
249 cp++;
251 *left = l;
252 return cp;
255 static int word_char(char ch)
257 return isalnum(ch) || ch == '_';
260 static void show_name(struct grep_opt *opt, const char *name)
262 printf("%s%c", name, opt->null_following_name ? '\0' : '\n');
265 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
267 char *hit = strstr(line, pattern);
268 if (!hit) {
269 match->rm_so = match->rm_eo = -1;
270 return REG_NOMATCH;
272 else {
273 match->rm_so = hit - line;
274 match->rm_eo = match->rm_so + strlen(pattern);
275 return 0;
279 static int strip_timestamp(char *bol, char **eol_p)
281 char *eol = *eol_p;
282 int ch;
284 while (bol < --eol) {
285 if (*eol != '>')
286 continue;
287 *eol_p = ++eol;
288 ch = *eol;
289 *eol = '\0';
290 return ch;
292 return 0;
295 static struct {
296 const char *field;
297 size_t len;
298 } header_field[] = {
299 { "author ", 7 },
300 { "committer ", 10 },
303 static int match_one_pattern(struct grep_pat *p, char *bol, char *eol,
304 enum grep_context ctx,
305 regmatch_t *pmatch, int eflags)
307 int hit = 0;
308 int saved_ch = 0;
309 const char *start = bol;
311 if ((p->token != GREP_PATTERN) &&
312 ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD)))
313 return 0;
315 if (p->token == GREP_PATTERN_HEAD) {
316 const char *field;
317 size_t len;
318 assert(p->field < ARRAY_SIZE(header_field));
319 field = header_field[p->field].field;
320 len = header_field[p->field].len;
321 if (strncmp(bol, field, len))
322 return 0;
323 bol += len;
324 saved_ch = strip_timestamp(bol, &eol);
327 again:
328 if (p->fixed)
329 hit = !fixmatch(p->pattern, bol, pmatch);
330 else
331 hit = !regexec(&p->regexp, bol, 1, pmatch, eflags);
333 if (hit && p->word_regexp) {
334 if ((pmatch[0].rm_so < 0) ||
335 (eol - bol) < pmatch[0].rm_so ||
336 (pmatch[0].rm_eo < 0) ||
337 (eol - bol) < pmatch[0].rm_eo)
338 die("regexp returned nonsense");
340 /* Match beginning must be either beginning of the
341 * line, or at word boundary (i.e. the last char must
342 * not be a word char). Similarly, match end must be
343 * either end of the line, or at word boundary
344 * (i.e. the next char must not be a word char).
346 if ( ((pmatch[0].rm_so == 0) ||
347 !word_char(bol[pmatch[0].rm_so-1])) &&
348 ((pmatch[0].rm_eo == (eol-bol)) ||
349 !word_char(bol[pmatch[0].rm_eo])) )
351 else
352 hit = 0;
354 /* Words consist of at least one character. */
355 if (pmatch->rm_so == pmatch->rm_eo)
356 hit = 0;
358 if (!hit && pmatch[0].rm_so + bol + 1 < eol) {
359 /* There could be more than one match on the
360 * line, and the first match might not be
361 * strict word match. But later ones could be!
362 * Forward to the next possible start, i.e. the
363 * next position following a non-word char.
365 bol = pmatch[0].rm_so + bol + 1;
366 while (word_char(bol[-1]) && bol < eol)
367 bol++;
368 eflags |= REG_NOTBOL;
369 if (bol < eol)
370 goto again;
373 if (p->token == GREP_PATTERN_HEAD && saved_ch)
374 *eol = saved_ch;
375 if (hit) {
376 pmatch[0].rm_so += bol - start;
377 pmatch[0].rm_eo += bol - start;
379 return hit;
382 static int match_expr_eval(struct grep_expr *x, char *bol, char *eol,
383 enum grep_context ctx, int collect_hits)
385 int h = 0;
386 regmatch_t match;
388 if (!x)
389 die("Not a valid grep expression");
390 switch (x->node) {
391 case GREP_NODE_ATOM:
392 h = match_one_pattern(x->u.atom, bol, eol, ctx, &match, 0);
393 break;
394 case GREP_NODE_NOT:
395 h = !match_expr_eval(x->u.unary, bol, eol, ctx, 0);
396 break;
397 case GREP_NODE_AND:
398 if (!match_expr_eval(x->u.binary.left, bol, eol, ctx, 0))
399 return 0;
400 h = match_expr_eval(x->u.binary.right, bol, eol, ctx, 0);
401 break;
402 case GREP_NODE_OR:
403 if (!collect_hits)
404 return (match_expr_eval(x->u.binary.left,
405 bol, eol, ctx, 0) ||
406 match_expr_eval(x->u.binary.right,
407 bol, eol, ctx, 0));
408 h = match_expr_eval(x->u.binary.left, bol, eol, ctx, 0);
409 x->u.binary.left->hit |= h;
410 h |= match_expr_eval(x->u.binary.right, bol, eol, ctx, 1);
411 break;
412 default:
413 die("Unexpected node type (internal error) %d", x->node);
415 if (collect_hits)
416 x->hit |= h;
417 return h;
420 static int match_expr(struct grep_opt *opt, char *bol, char *eol,
421 enum grep_context ctx, int collect_hits)
423 struct grep_expr *x = opt->pattern_expression;
424 return match_expr_eval(x, bol, eol, ctx, collect_hits);
427 static int match_line(struct grep_opt *opt, char *bol, char *eol,
428 enum grep_context ctx, int collect_hits)
430 struct grep_pat *p;
431 regmatch_t match;
433 if (opt->extended)
434 return match_expr(opt, bol, eol, ctx, collect_hits);
436 /* we do not call with collect_hits without being extended */
437 for (p = opt->pattern_list; p; p = p->next) {
438 if (match_one_pattern(p, bol, eol, ctx, &match, 0))
439 return 1;
441 return 0;
444 static int match_next_pattern(struct grep_pat *p, char *bol, char *eol,
445 enum grep_context ctx,
446 regmatch_t *pmatch, int eflags)
448 regmatch_t match;
450 if (!match_one_pattern(p, bol, eol, ctx, &match, eflags))
451 return 0;
452 if (match.rm_so < 0 || match.rm_eo < 0)
453 return 0;
454 if (pmatch->rm_so >= 0 && pmatch->rm_eo >= 0) {
455 if (match.rm_so > pmatch->rm_so)
456 return 1;
457 if (match.rm_so == pmatch->rm_so && match.rm_eo < pmatch->rm_eo)
458 return 1;
460 pmatch->rm_so = match.rm_so;
461 pmatch->rm_eo = match.rm_eo;
462 return 1;
465 static int next_match(struct grep_opt *opt, char *bol, char *eol,
466 enum grep_context ctx, regmatch_t *pmatch, int eflags)
468 struct grep_pat *p;
469 int hit = 0;
471 pmatch->rm_so = pmatch->rm_eo = -1;
472 if (bol < eol) {
473 for (p = opt->pattern_list; p; p = p->next) {
474 switch (p->token) {
475 case GREP_PATTERN: /* atom */
476 case GREP_PATTERN_HEAD:
477 case GREP_PATTERN_BODY:
478 hit |= match_next_pattern(p, bol, eol, ctx,
479 pmatch, eflags);
480 break;
481 default:
482 break;
486 return hit;
489 static void show_line(struct grep_opt *opt, char *bol, char *eol,
490 const char *name, unsigned lno, char sign)
492 int rest = eol - bol;
494 if (opt->pre_context || opt->post_context || opt->funcname) {
495 if (opt->last_shown == 0) {
496 if (opt->show_hunk_mark)
497 fputs(opt->funcname ? "==\n" : "--\n", stdout);
498 else
499 opt->show_hunk_mark = 1;
500 } else if (lno > opt->last_shown + 1) {
501 if (opt->pre_context || opt->post_context)
502 fputs((sign == '=') ? "==\n" : "--\n", stdout);
503 else if (sign == '=')
504 fputs("==\n", stdout);
507 opt->last_shown = lno;
509 if (opt->null_following_name)
510 sign = '\0';
511 if (opt->pathname)
512 printf("%s%c", name, sign);
513 if (opt->linenum)
514 printf("%d%c", lno, sign);
515 if (opt->color) {
516 regmatch_t match;
517 enum grep_context ctx = GREP_CONTEXT_BODY;
518 int ch = *eol;
519 int eflags = 0;
521 *eol = '\0';
522 while (next_match(opt, bol, eol, ctx, &match, eflags)) {
523 if (match.rm_so == match.rm_eo)
524 break;
525 printf("%.*s%s%.*s%s",
526 (int)match.rm_so, bol,
527 opt->color_match,
528 (int)(match.rm_eo - match.rm_so), bol + match.rm_so,
529 GIT_COLOR_RESET);
530 bol += match.rm_eo;
531 rest -= match.rm_eo;
532 eflags = REG_NOTBOL;
534 *eol = ch;
536 printf("%.*s\n", rest, bol);
539 static int match_funcname(struct grep_opt *opt, char *bol, char *eol)
541 xdemitconf_t *xecfg = opt->priv;
542 if (xecfg && xecfg->find_func) {
543 char buf[1];
544 return xecfg->find_func(bol, eol - bol, buf, 1,
545 xecfg->find_func_priv) >= 0;
548 if (bol == eol)
549 return 0;
550 if (isalpha(*bol) || *bol == '_' || *bol == '$')
551 return 1;
552 return 0;
555 static void show_funcname_line(struct grep_opt *opt, const char *name,
556 char *buf, char *bol, unsigned lno)
558 while (bol > buf) {
559 char *eol = --bol;
561 while (bol > buf && bol[-1] != '\n')
562 bol--;
563 lno--;
565 if (lno <= opt->last_shown)
566 break;
568 if (match_funcname(opt, bol, eol)) {
569 show_line(opt, bol, eol, name, lno, '=');
570 break;
575 static void show_pre_context(struct grep_opt *opt, const char *name, char *buf,
576 char *bol, unsigned lno)
578 unsigned cur = lno, from = 1, funcname_lno = 0;
579 int funcname_needed = opt->funcname;
581 if (opt->pre_context < lno)
582 from = lno - opt->pre_context;
583 if (from <= opt->last_shown)
584 from = opt->last_shown + 1;
586 /* Rewind. */
587 while (bol > buf && cur > from) {
588 char *eol = --bol;
590 while (bol > buf && bol[-1] != '\n')
591 bol--;
592 cur--;
593 if (funcname_needed && match_funcname(opt, bol, eol)) {
594 funcname_lno = cur;
595 funcname_needed = 0;
599 /* We need to look even further back to find a function signature. */
600 if (opt->funcname && funcname_needed)
601 show_funcname_line(opt, name, buf, bol, cur);
603 /* Back forward. */
604 while (cur < lno) {
605 char *eol = bol, sign = (cur == funcname_lno) ? '=' : '-';
607 while (*eol != '\n')
608 eol++;
609 show_line(opt, bol, eol, name, cur, sign);
610 bol = eol + 1;
611 cur++;
615 static int grep_buffer_1(struct grep_opt *opt, const char *name,
616 char *buf, unsigned long size, int collect_hits)
618 char *bol = buf;
619 unsigned long left = size;
620 unsigned lno = 1;
621 unsigned last_hit = 0;
622 int binary_match_only = 0;
623 unsigned count = 0;
624 enum grep_context ctx = GREP_CONTEXT_HEAD;
625 xdemitconf_t xecfg;
627 opt->last_shown = 0;
629 if (buffer_is_binary(buf, size)) {
630 switch (opt->binary) {
631 case GREP_BINARY_DEFAULT:
632 binary_match_only = 1;
633 break;
634 case GREP_BINARY_NOMATCH:
635 return 0; /* Assume unmatch */
636 break;
637 default:
638 break;
642 memset(&xecfg, 0, sizeof(xecfg));
643 if (opt->funcname && !opt->unmatch_name_only && !opt->status_only &&
644 !opt->name_only && !binary_match_only && !collect_hits) {
645 struct userdiff_driver *drv = userdiff_find_by_path(name);
646 if (drv && drv->funcname.pattern) {
647 const struct userdiff_funcname *pe = &drv->funcname;
648 xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
649 opt->priv = &xecfg;
653 while (left) {
654 char *eol, ch;
655 int hit;
657 eol = end_of_line(bol, &left);
658 ch = *eol;
659 *eol = 0;
661 if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol))
662 ctx = GREP_CONTEXT_BODY;
664 hit = match_line(opt, bol, eol, ctx, collect_hits);
665 *eol = ch;
667 if (collect_hits)
668 goto next_line;
670 /* "grep -v -e foo -e bla" should list lines
671 * that do not have either, so inversion should
672 * be done outside.
674 if (opt->invert)
675 hit = !hit;
676 if (opt->unmatch_name_only) {
677 if (hit)
678 return 0;
679 goto next_line;
681 if (hit) {
682 count++;
683 if (opt->status_only)
684 return 1;
685 if (binary_match_only) {
686 printf("Binary file %s matches\n", name);
687 return 1;
689 if (opt->name_only) {
690 show_name(opt, name);
691 return 1;
693 /* Hit at this line. If we haven't shown the
694 * pre-context lines, we would need to show them.
695 * When asked to do "count", this still show
696 * the context which is nonsense, but the user
697 * deserves to get that ;-).
699 if (opt->pre_context)
700 show_pre_context(opt, name, buf, bol, lno);
701 else if (opt->funcname)
702 show_funcname_line(opt, name, buf, bol, lno);
703 if (!opt->count)
704 show_line(opt, bol, eol, name, lno, ':');
705 last_hit = lno;
707 else if (last_hit &&
708 lno <= last_hit + opt->post_context) {
709 /* If the last hit is within the post context,
710 * we need to show this line.
712 show_line(opt, bol, eol, name, lno, '-');
715 next_line:
716 bol = eol + 1;
717 if (!left)
718 break;
719 left--;
720 lno++;
723 if (collect_hits)
724 return 0;
726 if (opt->status_only)
727 return 0;
728 if (opt->unmatch_name_only) {
729 /* We did not see any hit, so we want to show this */
730 show_name(opt, name);
731 return 1;
734 xdiff_clear_find_func(&xecfg);
735 opt->priv = NULL;
737 /* NEEDSWORK:
738 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
739 * which feels mostly useless but sometimes useful. Maybe
740 * make it another option? For now suppress them.
742 if (opt->count && count)
743 printf("%s%c%u\n", name,
744 opt->null_following_name ? '\0' : ':', count);
745 return !!last_hit;
748 static void clr_hit_marker(struct grep_expr *x)
750 /* All-hit markers are meaningful only at the very top level
751 * OR node.
753 while (1) {
754 x->hit = 0;
755 if (x->node != GREP_NODE_OR)
756 return;
757 x->u.binary.left->hit = 0;
758 x = x->u.binary.right;
762 static int chk_hit_marker(struct grep_expr *x)
764 /* Top level nodes have hit markers. See if they all are hits */
765 while (1) {
766 if (x->node != GREP_NODE_OR)
767 return x->hit;
768 if (!x->u.binary.left->hit)
769 return 0;
770 x = x->u.binary.right;
774 int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size)
777 * we do not have to do the two-pass grep when we do not check
778 * buffer-wide "all-match".
780 if (!opt->all_match)
781 return grep_buffer_1(opt, name, buf, size, 0);
783 /* Otherwise the toplevel "or" terms hit a bit differently.
784 * We first clear hit markers from them.
786 clr_hit_marker(opt->pattern_expression);
787 grep_buffer_1(opt, name, buf, size, 1);
789 if (!chk_hit_marker(opt->pattern_expression))
790 return 0;
792 return grep_buffer_1(opt, name, buf, size, 0);