Merge branch 'mh/notes-duplicate-entries'
[git/git.git] / pathspec.c
1 #include "cache.h"
2 #include "config.h"
3 #include "dir.h"
4 #include "pathspec.h"
5 #include "attr.h"
6
7 /*
8 * Finds which of the given pathspecs match items in the index.
9 *
10 * For each pathspec, sets the corresponding entry in the seen[] array
11 * (which should be specs items long, i.e. the same size as pathspec)
12 * to the nature of the "closest" (i.e. most specific) match found for
13 * that pathspec in the index, if it was a closer type of match than
14 * the existing entry. As an optimization, matching is skipped
15 * altogether if seen[] already only contains non-zero entries.
16 *
17 * If seen[] has not already been written to, it may make sense
18 * to use find_pathspecs_matching_against_index() instead.
19 */
20 void add_pathspec_matches_against_index(const struct pathspec *pathspec,
21 const struct index_state *istate,
22 char *seen)
23 {
24 int num_unmatched = 0, i;
25
26 /*
27 * Since we are walking the index as if we were walking the directory,
28 * we have to mark the matched pathspec as seen; otherwise we will
29 * mistakenly think that the user gave a pathspec that did not match
30 * anything.
31 */
32 for (i = 0; i < pathspec->nr; i++)
33 if (!seen[i])
34 num_unmatched++;
35 if (!num_unmatched)
36 return;
37 for (i = 0; i < istate->cache_nr; i++) {
38 const struct cache_entry *ce = istate->cache[i];
39 ce_path_match(istate, ce, pathspec, seen);
40 }
41 }
42
43 /*
44 * Finds which of the given pathspecs match items in the index.
45 *
46 * This is a one-shot wrapper around add_pathspec_matches_against_index()
47 * which allocates, populates, and returns a seen[] array indicating the
48 * nature of the "closest" (i.e. most specific) matches which each of the
49 * given pathspecs achieves against all items in the index.
50 */
51 char *find_pathspecs_matching_against_index(const struct pathspec *pathspec,
52 const struct index_state *istate)
53 {
54 char *seen = xcalloc(pathspec->nr, 1);
55 add_pathspec_matches_against_index(pathspec, istate, seen);
56 return seen;
57 }
58
59 /*
60 * Magic pathspec
61 *
62 * Possible future magic semantics include stuff like:
63 *
64 * { PATHSPEC_RECURSIVE, '*', "recursive" },
65 * { PATHSPEC_REGEXP, '\0', "regexp" },
66 *
67 */
68
69 static struct pathspec_magic {
70 unsigned bit;
71 char mnemonic; /* this cannot be ':'! */
72 const char *name;
73 } pathspec_magic[] = {
74 { PATHSPEC_FROMTOP, '/', "top" },
75 { PATHSPEC_LITERAL, '\0', "literal" },
76 { PATHSPEC_GLOB, '\0', "glob" },
77 { PATHSPEC_ICASE, '\0', "icase" },
78 { PATHSPEC_EXCLUDE, '!', "exclude" },
79 { PATHSPEC_ATTR, '\0', "attr" },
80 };
81
82 static void prefix_magic(struct strbuf *sb, int prefixlen, unsigned magic)
83 {
84 int i;
85 strbuf_addstr(sb, ":(");
86 for (i = 0; i < ARRAY_SIZE(pathspec_magic); i++)
87 if (magic & pathspec_magic[i].bit) {
88 if (sb->buf[sb->len - 1] != '(')
89 strbuf_addch(sb, ',');
90 strbuf_addstr(sb, pathspec_magic[i].name);
91 }
92 strbuf_addf(sb, ",prefix:%d)", prefixlen);
93 }
94
95 static size_t strcspn_escaped(const char *s, const char *stop)
96 {
97 const char *i;
98
99 for (i = s; *i; i++) {
100 /* skip the escaped character */
101 if (i[0] == '\\' && i[1]) {
102 i++;
103 continue;
104 }
105
106 if (strchr(stop, *i))
107 break;
108 }
109 return i - s;
110 }
111
112 static inline int invalid_value_char(const char ch)
113 {
114 if (isalnum(ch) || strchr(",-_", ch))
115 return 0;
116 return -1;
117 }
118
119 static char *attr_value_unescape(const char *value)
120 {
121 const char *src;
122 char *dst, *ret;
123
124 ret = xmallocz(strlen(value));
125 for (src = value, dst = ret; *src; src++, dst++) {
126 if (*src == '\\') {
127 if (!src[1])
128 die(_("Escape character '\\' not allowed as "
129 "last character in attr value"));
130 src++;
131 }
132 if (invalid_value_char(*src))
133 die("cannot use '%c' for value matching", *src);
134 *dst = *src;
135 }
136 *dst = '\0';
137 return ret;
138 }
139
140 static void parse_pathspec_attr_match(struct pathspec_item *item, const char *value)
141 {
142 struct string_list_item *si;
143 struct string_list list = STRING_LIST_INIT_DUP;
144
145 if (item->attr_check || item->attr_match)
146 die(_("Only one 'attr:' specification is allowed."));
147
148 if (!value || !*value)
149 die(_("attr spec must not be empty"));
150
151 string_list_split(&list, value, ' ', -1);
152 string_list_remove_empty_items(&list, 0);
153
154 item->attr_check = attr_check_alloc();
155 item->attr_match = xcalloc(list.nr, sizeof(struct attr_match));
156
157 for_each_string_list_item(si, &list) {
158 size_t attr_len;
159 char *attr_name;
160 const struct git_attr *a;
161
162 int j = item->attr_match_nr++;
163 const char *attr = si->string;
164 struct attr_match *am = &item->attr_match[j];
165
166 switch (*attr) {
167 case '!':
168 am->match_mode = MATCH_UNSPECIFIED;
169 attr++;
170 attr_len = strlen(attr);
171 break;
172 case '-':
173 am->match_mode = MATCH_UNSET;
174 attr++;
175 attr_len = strlen(attr);
176 break;
177 default:
178 attr_len = strcspn(attr, "=");
179 if (attr[attr_len] != '=')
180 am->match_mode = MATCH_SET;
181 else {
182 const char *v = &attr[attr_len + 1];
183 am->match_mode = MATCH_VALUE;
184 am->value = attr_value_unescape(v);
185 }
186 break;
187 }
188
189 attr_name = xmemdupz(attr, attr_len);
190 a = git_attr(attr_name);
191 if (!a)
192 die(_("invalid attribute name %s"), attr_name);
193
194 attr_check_append(item->attr_check, a);
195
196 free(attr_name);
197 }
198
199 if (item->attr_check->nr != item->attr_match_nr)
200 BUG("should have same number of entries");
201
202 string_list_clear(&list, 0);
203 }
204
205 static inline int get_literal_global(void)
206 {
207 static int literal = -1;
208
209 if (literal < 0)
210 literal = git_env_bool(GIT_LITERAL_PATHSPECS_ENVIRONMENT, 0);
211
212 return literal;
213 }
214
215 static inline int get_glob_global(void)
216 {
217 static int glob = -1;
218
219 if (glob < 0)
220 glob = git_env_bool(GIT_GLOB_PATHSPECS_ENVIRONMENT, 0);
221
222 return glob;
223 }
224
225 static inline int get_noglob_global(void)
226 {
227 static int noglob = -1;
228
229 if (noglob < 0)
230 noglob = git_env_bool(GIT_NOGLOB_PATHSPECS_ENVIRONMENT, 0);
231
232 return noglob;
233 }
234
235 static inline int get_icase_global(void)
236 {
237 static int icase = -1;
238
239 if (icase < 0)
240 icase = git_env_bool(GIT_ICASE_PATHSPECS_ENVIRONMENT, 0);
241
242 return icase;
243 }
244
245 static int get_global_magic(int element_magic)
246 {
247 int global_magic = 0;
248
249 if (get_literal_global())
250 global_magic |= PATHSPEC_LITERAL;
251
252 /* --glob-pathspec is overridden by :(literal) */
253 if (get_glob_global() && !(element_magic & PATHSPEC_LITERAL))
254 global_magic |= PATHSPEC_GLOB;
255
256 if (get_glob_global() && get_noglob_global())
257 die(_("global 'glob' and 'noglob' pathspec settings are incompatible"));
258
259 if (get_icase_global())
260 global_magic |= PATHSPEC_ICASE;
261
262 if ((global_magic & PATHSPEC_LITERAL) &&
263 (global_magic & ~PATHSPEC_LITERAL))
264 die(_("global 'literal' pathspec setting is incompatible "
265 "with all other global pathspec settings"));
266
267 /* --noglob-pathspec adds :(literal) _unless_ :(glob) is specified */
268 if (get_noglob_global() && !(element_magic & PATHSPEC_GLOB))
269 global_magic |= PATHSPEC_LITERAL;
270
271 return global_magic;
272 }
273
274 /*
275 * Parse the pathspec element looking for long magic
276 *
277 * saves all magic in 'magic'
278 * if prefix magic is used, save the prefix length in 'prefix_len'
279 * returns the position in 'elem' after all magic has been parsed
280 */
281 static const char *parse_long_magic(unsigned *magic, int *prefix_len,
282 struct pathspec_item *item,
283 const char *elem)
284 {
285 const char *pos;
286 const char *nextat;
287
288 for (pos = elem + 2; *pos && *pos != ')'; pos = nextat) {
289 size_t len = strcspn_escaped(pos, ",)");
290 int i;
291
292 if (pos[len] == ',')
293 nextat = pos + len + 1; /* handle ',' */
294 else
295 nextat = pos + len; /* handle ')' and '\0' */
296
297 if (!len)
298 continue;
299
300 if (starts_with(pos, "prefix:")) {
301 char *endptr;
302 *prefix_len = strtol(pos + 7, &endptr, 10);
303 if (endptr - pos != len)
304 die(_("invalid parameter for pathspec magic 'prefix'"));
305 continue;
306 }
307
308 if (starts_with(pos, "attr:")) {
309 char *attr_body = xmemdupz(pos + 5, len - 5);
310 parse_pathspec_attr_match(item, attr_body);
311 *magic |= PATHSPEC_ATTR;
312 free(attr_body);
313 continue;
314 }
315
316 for (i = 0; i < ARRAY_SIZE(pathspec_magic); i++) {
317 if (strlen(pathspec_magic[i].name) == len &&
318 !strncmp(pathspec_magic[i].name, pos, len)) {
319 *magic |= pathspec_magic[i].bit;
320 break;
321 }
322 }
323
324 if (ARRAY_SIZE(pathspec_magic) <= i)
325 die(_("Invalid pathspec magic '%.*s' in '%s'"),
326 (int) len, pos, elem);
327 }
328
329 if (*pos != ')')
330 die(_("Missing ')' at the end of pathspec magic in '%s'"),
331 elem);
332 pos++;
333
334 return pos;
335 }
336
337 /*
338 * Parse the pathspec element looking for short magic
339 *
340 * saves all magic in 'magic'
341 * returns the position in 'elem' after all magic has been parsed
342 */
343 static const char *parse_short_magic(unsigned *magic, const char *elem)
344 {
345 const char *pos;
346
347 for (pos = elem + 1; *pos && *pos != ':'; pos++) {
348 char ch = *pos;
349 int i;
350
351 /* Special case alias for '!' */
352 if (ch == '^') {
353 *magic |= PATHSPEC_EXCLUDE;
354 continue;
355 }
356
357 if (!is_pathspec_magic(ch))
358 break;
359
360 for (i = 0; i < ARRAY_SIZE(pathspec_magic); i++) {
361 if (pathspec_magic[i].mnemonic == ch) {
362 *magic |= pathspec_magic[i].bit;
363 break;
364 }
365 }
366
367 if (ARRAY_SIZE(pathspec_magic) <= i)
368 die(_("Unimplemented pathspec magic '%c' in '%s'"),
369 ch, elem);
370 }
371
372 if (*pos == ':')
373 pos++;
374
375 return pos;
376 }
377
378 static const char *parse_element_magic(unsigned *magic, int *prefix_len,
379 struct pathspec_item *item,
380 const char *elem)
381 {
382 if (elem[0] != ':' || get_literal_global())
383 return elem; /* nothing to do */
384 else if (elem[1] == '(')
385 /* longhand */
386 return parse_long_magic(magic, prefix_len, item, elem);
387 else
388 /* shorthand */
389 return parse_short_magic(magic, elem);
390 }
391
392 /*
393 * Perform the initialization of a pathspec_item based on a pathspec element.
394 */
395 static void init_pathspec_item(struct pathspec_item *item, unsigned flags,
396 const char *prefix, int prefixlen,
397 const char *elt)
398 {
399 unsigned magic = 0, element_magic = 0;
400 const char *copyfrom = elt;
401 char *match;
402 int pathspec_prefix = -1;
403
404 item->attr_check = NULL;
405 item->attr_match = NULL;
406 item->attr_match_nr = 0;
407
408 /* PATHSPEC_LITERAL_PATH ignores magic */
409 if (flags & PATHSPEC_LITERAL_PATH) {
410 magic = PATHSPEC_LITERAL;
411 } else {
412 copyfrom = parse_element_magic(&element_magic,
413 &pathspec_prefix,
414 item,
415 elt);
416 magic |= element_magic;
417 magic |= get_global_magic(element_magic);
418 }
419
420 item->magic = magic;
421
422 if (pathspec_prefix >= 0 &&
423 (prefixlen || (prefix && *prefix)))
424 BUG("'prefix' magic is supposed to be used at worktree's root");
425
426 if ((magic & PATHSPEC_LITERAL) && (magic & PATHSPEC_GLOB))
427 die(_("%s: 'literal' and 'glob' are incompatible"), elt);
428
429 /* Create match string which will be used for pathspec matching */
430 if (pathspec_prefix >= 0) {
431 match = xstrdup(copyfrom);
432 prefixlen = pathspec_prefix;
433 } else if (magic & PATHSPEC_FROMTOP) {
434 match = xstrdup(copyfrom);
435 prefixlen = 0;
436 } else {
437 match = prefix_path_gently(prefix, prefixlen,
438 &prefixlen, copyfrom);
439 if (!match)
440 die(_("%s: '%s' is outside repository"), elt, copyfrom);
441 }
442
443 item->match = match;
444 item->len = strlen(item->match);
445 item->prefix = prefixlen;
446
447 /*
448 * Prefix the pathspec (keep all magic) and assign to
449 * original. Useful for passing to another command.
450 */
451 if ((flags & PATHSPEC_PREFIX_ORIGIN) &&
452 !get_literal_global()) {
453 struct strbuf sb = STRBUF_INIT;
454
455 /* Preserve the actual prefix length of each pattern */
456 prefix_magic(&sb, prefixlen, element_magic);
457
458 strbuf_addstr(&sb, match);
459 item->original = strbuf_detach(&sb, NULL);
460 } else {
461 item->original = xstrdup(elt);
462 }
463
464 if (magic & PATHSPEC_LITERAL) {
465 item->nowildcard_len = item->len;
466 } else {
467 item->nowildcard_len = simple_length(item->match);
468 if (item->nowildcard_len < prefixlen)
469 item->nowildcard_len = prefixlen;
470 }
471
472 item->flags = 0;
473 if (magic & PATHSPEC_GLOB) {
474 /*
475 * FIXME: should we enable ONESTAR in _GLOB for
476 * pattern "* * / * . c"?
477 */
478 } else {
479 if (item->nowildcard_len < item->len &&
480 item->match[item->nowildcard_len] == '*' &&
481 no_wildcard(item->match + item->nowildcard_len + 1))
482 item->flags |= PATHSPEC_ONESTAR;
483 }
484
485 /* sanity checks, pathspec matchers assume these are sane */
486 if (item->nowildcard_len > item->len ||
487 item->prefix > item->len) {
488 BUG("error initializing pathspec_item");
489 }
490 }
491
492 static int pathspec_item_cmp(const void *a_, const void *b_)
493 {
494 struct pathspec_item *a, *b;
495
496 a = (struct pathspec_item *)a_;
497 b = (struct pathspec_item *)b_;
498 return strcmp(a->match, b->match);
499 }
500
501 static void NORETURN unsupported_magic(const char *pattern,
502 unsigned magic)
503 {
504 struct strbuf sb = STRBUF_INIT;
505 int i;
506 for (i = 0; i < ARRAY_SIZE(pathspec_magic); i++) {
507 const struct pathspec_magic *m = pathspec_magic + i;
508 if (!(magic & m->bit))
509 continue;
510 if (sb.len)
511 strbuf_addstr(&sb, ", ");
512
513 if (m->mnemonic)
514 strbuf_addf(&sb, _("'%s' (mnemonic: '%c')"),
515 m->name, m->mnemonic);
516 else
517 strbuf_addf(&sb, "'%s'", m->name);
518 }
519 /*
520 * We may want to substitute "this command" with a command
521 * name. E.g. when add--interactive dies when running
522 * "checkout -p"
523 */
524 die(_("%s: pathspec magic not supported by this command: %s"),
525 pattern, sb.buf);
526 }
527
528 void parse_pathspec(struct pathspec *pathspec,
529 unsigned magic_mask, unsigned flags,
530 const char *prefix, const char **argv)
531 {
532 struct pathspec_item *item;
533 const char *entry = argv ? *argv : NULL;
534 int i, n, prefixlen, nr_exclude = 0;
535
536 memset(pathspec, 0, sizeof(*pathspec));
537
538 if (flags & PATHSPEC_MAXDEPTH_VALID)
539 pathspec->magic |= PATHSPEC_MAXDEPTH;
540
541 /* No arguments, no prefix -> no pathspec */
542 if (!entry && !prefix)
543 return;
544
545 if ((flags & PATHSPEC_PREFER_CWD) &&
546 (flags & PATHSPEC_PREFER_FULL))
547 BUG("PATHSPEC_PREFER_CWD and PATHSPEC_PREFER_FULL are incompatible");
548
549 /* No arguments with prefix -> prefix pathspec */
550 if (!entry) {
551 if (flags & PATHSPEC_PREFER_FULL)
552 return;
553
554 if (!(flags & PATHSPEC_PREFER_CWD))
555 BUG("PATHSPEC_PREFER_CWD requires arguments");
556
557 pathspec->items = item = xcalloc(1, sizeof(*item));
558 item->match = xstrdup(prefix);
559 item->original = xstrdup(prefix);
560 item->nowildcard_len = item->len = strlen(prefix);
561 item->prefix = item->len;
562 pathspec->nr = 1;
563 return;
564 }
565
566 n = 0;
567 while (argv[n]) {
568 if (*argv[n] == '\0')
569 die("empty string is not a valid pathspec. "
570 "please use . instead if you meant to match all paths");
571 n++;
572 }
573
574 pathspec->nr = n;
575 ALLOC_ARRAY(pathspec->items, n + 1);
576 item = pathspec->items;
577 prefixlen = prefix ? strlen(prefix) : 0;
578
579 for (i = 0; i < n; i++) {
580 entry = argv[i];
581
582 init_pathspec_item(item + i, flags, prefix, prefixlen, entry);
583
584 if (item[i].magic & PATHSPEC_EXCLUDE)
585 nr_exclude++;
586 if (item[i].magic & magic_mask)
587 unsupported_magic(entry, item[i].magic & magic_mask);
588
589 if ((flags & PATHSPEC_SYMLINK_LEADING_PATH) &&
590 has_symlink_leading_path(item[i].match, item[i].len)) {
591 die(_("pathspec '%s' is beyond a symbolic link"), entry);
592 }
593
594 if (item[i].nowildcard_len < item[i].len)
595 pathspec->has_wildcard = 1;
596 pathspec->magic |= item[i].magic;
597 }
598
599 /*
600 * If everything is an exclude pattern, add one positive pattern
601 * that matches everything. We allocated an extra one for this.
602 */
603 if (nr_exclude == n) {
604 int plen = (!(flags & PATHSPEC_PREFER_CWD)) ? 0 : prefixlen;
605 init_pathspec_item(item + n, 0, prefix, plen, "");
606 pathspec->nr++;
607 }
608
609 if (pathspec->magic & PATHSPEC_MAXDEPTH) {
610 if (flags & PATHSPEC_KEEP_ORDER)
611 BUG("PATHSPEC_MAXDEPTH_VALID and PATHSPEC_KEEP_ORDER are incompatible");
612 QSORT(pathspec->items, pathspec->nr, pathspec_item_cmp);
613 }
614 }
615
616 void copy_pathspec(struct pathspec *dst, const struct pathspec *src)
617 {
618 int i, j;
619
620 *dst = *src;
621 ALLOC_ARRAY(dst->items, dst->nr);
622 COPY_ARRAY(dst->items, src->items, dst->nr);
623
624 for (i = 0; i < dst->nr; i++) {
625 struct pathspec_item *d = &dst->items[i];
626 struct pathspec_item *s = &src->items[i];
627
628 d->match = xstrdup(s->match);
629 d->original = xstrdup(s->original);
630
631 ALLOC_ARRAY(d->attr_match, d->attr_match_nr);
632 COPY_ARRAY(d->attr_match, s->attr_match, d->attr_match_nr);
633 for (j = 0; j < d->attr_match_nr; j++) {
634 const char *value = s->attr_match[j].value;
635 d->attr_match[j].value = xstrdup_or_null(value);
636 }
637
638 d->attr_check = attr_check_dup(s->attr_check);
639 }
640 }
641
642 void clear_pathspec(struct pathspec *pathspec)
643 {
644 int i, j;
645
646 for (i = 0; i < pathspec->nr; i++) {
647 free(pathspec->items[i].match);
648 free(pathspec->items[i].original);
649
650 for (j = 0; j < pathspec->items[i].attr_match_nr; j++)
651 free(pathspec->items[i].attr_match[j].value);
652 free(pathspec->items[i].attr_match);
653
654 if (pathspec->items[i].attr_check)
655 attr_check_free(pathspec->items[i].attr_check);
656 }
657
658 FREE_AND_NULL(pathspec->items);
659 pathspec->nr = 0;
660 }
661
662 int match_pathspec_attrs(const struct index_state *istate,
663 const char *name, int namelen,
664 const struct pathspec_item *item)
665 {
666 int i;
667 char *to_free = NULL;
668
669 if (name[namelen])
670 name = to_free = xmemdupz(name, namelen);
671
672 git_check_attr(istate, name, item->attr_check);
673
674 free(to_free);
675
676 for (i = 0; i < item->attr_match_nr; i++) {
677 const char *value;
678 int matched;
679 enum attr_match_mode match_mode;
680
681 value = item->attr_check->items[i].value;
682 match_mode = item->attr_match[i].match_mode;
683
684 if (ATTR_TRUE(value))
685 matched = (match_mode == MATCH_SET);
686 else if (ATTR_FALSE(value))
687 matched = (match_mode == MATCH_UNSET);
688 else if (ATTR_UNSET(value))
689 matched = (match_mode == MATCH_UNSPECIFIED);
690 else
691 matched = (match_mode == MATCH_VALUE &&
692 !strcmp(item->attr_match[i].value, value));
693 if (!matched)
694 return 0;
695 }
696
697 return 1;
698 }