1/*
2 * Handle git attributes. See gitattributes(5) for a description of
3 * the file syntax, and Documentation/technical/api-gitattributes.txt
4 * for a description of the API.
5 *
6 * One basic design decision here is that we are not going to support
7 * an insanely large number of attributes.
8 */
9
10#define NO_THE_INDEX_COMPATIBILITY_MACROS
11#include "cache.h"
12#include "exec_cmd.h"
13#include "attr.h"
14
15const char git_attr__true[] = "(builtin)true";
16const char git_attr__false[] = "\0(builtin)false";
17static const char git_attr__unknown[] = "(builtin)unknown";
18#define ATTR__TRUE git_attr__true
19#define ATTR__FALSE git_attr__false
20#define ATTR__UNSET NULL
21#define ATTR__UNKNOWN git_attr__unknown
22
23static const char *attributes_file;
24
25/* This is a randomly chosen prime. */
26#define HASHSIZE 257
27
28#ifndef DEBUG_ATTR
29#define DEBUG_ATTR 0
30#endif
31
32struct git_attr {
33 struct git_attr *next;
34 unsigned h;
35 int attr_nr;
36 char name[FLEX_ARRAY];
37};
38static int attr_nr;
39
40static struct git_attr_check *check_all_attr;
41static struct git_attr *(git_attr_hash[HASHSIZE]);
42
43static unsigned hash_name(const char *name, int namelen)
44{
45 unsigned val = 0, c;
46
47 while (namelen--) {
48 c = *name++;
49 val = ((val << 7) | (val >> 22)) ^ c;
50 }
51 return val;
52}
53
54static int invalid_attr_name(const char *name, int namelen)
55{
56 /*
57 * Attribute name cannot begin with '-' and from
58 * [-A-Za-z0-9_.]. We'd specifically exclude '=' for now,
59 * as we might later want to allow non-binary value for
60 * attributes, e.g. "*.svg merge=special-merge-program-for-svg"
61 */
62 if (*name == '-')
63 return -1;
64 while (namelen--) {
65 char ch = *name++;
66 if (! (ch == '-' || ch == '.' || ch == '_' ||
67 ('0' <= ch && ch <= '9') ||
68 ('a' <= ch && ch <= 'z') ||
69 ('A' <= ch && ch <= 'Z')) )
70 return -1;
71 }
72 return 0;
73}
74
75static struct git_attr *git_attr_internal(const char *name, int len)
76{
77 unsigned hval = hash_name(name, len);
78 unsigned pos = hval % HASHSIZE;
79 struct git_attr *a;
80
81 for (a = git_attr_hash[pos]; a; a = a->next) {
82 if (a->h == hval &&
83 !memcmp(a->name, name, len) && !a->name[len])
84 return a;
85 }
86
87 if (invalid_attr_name(name, len))
88 return NULL;
89
90 a = xmalloc(sizeof(*a) + len + 1);
91 memcpy(a->name, name, len);
92 a->name[len] = 0;
93 a->h = hval;
94 a->next = git_attr_hash[pos];
95 a->attr_nr = attr_nr++;
96 git_attr_hash[pos] = a;
97
98 check_all_attr = xrealloc(check_all_attr,
99 sizeof(*check_all_attr) * attr_nr);
100 check_all_attr[a->attr_nr].attr = a;
101 check_all_attr[a->attr_nr].value = ATTR__UNKNOWN;
102 return a;
103}
104
105struct git_attr *git_attr(const char *name)
106{
107 return git_attr_internal(name, strlen(name));
108}
109
110/* What does a matched pattern decide? */
111struct attr_state {
112 struct git_attr *attr;
113 const char *setto;
114};
115
116struct match_attr {
117 union {
118 char *pattern;
119 struct git_attr *attr;
120 } u;
121 char is_macro;
122 unsigned num_attr;
123 struct attr_state state[FLEX_ARRAY];
124};
125
126static const char blank[] = " \t\r\n";
127
128static const char *parse_attr(const char *src, int lineno, const char *cp,
129 int *num_attr, struct match_attr *res)
130{
131 const char *ep, *equals;
132 int len;
133
134 ep = cp + strcspn(cp, blank);
135 equals = strchr(cp, '=');
136 if (equals && ep < equals)
137 equals = NULL;
138 if (equals)
139 len = equals - cp;
140 else
141 len = ep - cp;
142 if (!res) {
143 if (*cp == '-' || *cp == '!') {
144 cp++;
145 len--;
146 }
147 if (invalid_attr_name(cp, len)) {
148 fprintf(stderr,
149 "%.*s is not a valid attribute name: %s:%d\n",
150 len, cp, src, lineno);
151 return NULL;
152 }
153 } else {
154 struct attr_state *e;
155
156 e = &(res->state[*num_attr]);
157 if (*cp == '-' || *cp == '!') {
158 e->setto = (*cp == '-') ? ATTR__FALSE : ATTR__UNSET;
159 cp++;
160 len--;
161 }
162 else if (!equals)
163 e->setto = ATTR__TRUE;
164 else {
165 e->setto = xmemdupz(equals + 1, ep - equals - 1);
166 }
167 e->attr = git_attr_internal(cp, len);
168 }
169 (*num_attr)++;
170 return ep + strspn(ep, blank);
171}
172
173static struct match_attr *parse_attr_line(const char *line, const char *src,
174 int lineno, int macro_ok)
175{
176 int namelen;
177 int num_attr;
178 const char *cp, *name;
179 struct match_attr *res = NULL;
180 int pass;
181 int is_macro;
182
183 cp = line + strspn(line, blank);
184 if (!*cp || *cp == '#')
185 return NULL;
186 name = cp;
187 namelen = strcspn(name, blank);
188 if (strlen(ATTRIBUTE_MACRO_PREFIX) < namelen &&
189 !prefixcmp(name, ATTRIBUTE_MACRO_PREFIX)) {
190 if (!macro_ok) {
191 fprintf(stderr, "%s not allowed: %s:%d\n",
192 name, src, lineno);
193 return NULL;
194 }
195 is_macro = 1;
196 name += strlen(ATTRIBUTE_MACRO_PREFIX);
197 name += strspn(name, blank);
198 namelen = strcspn(name, blank);
199 if (invalid_attr_name(name, namelen)) {
200 fprintf(stderr,
201 "%.*s is not a valid attribute name: %s:%d\n",
202 namelen, name, src, lineno);
203 return NULL;
204 }
205 }
206 else
207 is_macro = 0;
208
209 for (pass = 0; pass < 2; pass++) {
210 /* pass 0 counts and allocates, pass 1 fills */
211 num_attr = 0;
212 cp = name + namelen;
213 cp = cp + strspn(cp, blank);
214 while (*cp) {
215 cp = parse_attr(src, lineno, cp, &num_attr, res);
216 if (!cp)
217 return NULL;
218 }
219 if (pass)
220 break;
221 res = xcalloc(1,
222 sizeof(*res) +
223 sizeof(struct attr_state) * num_attr +
224 (is_macro ? 0 : namelen + 1));
225 if (is_macro)
226 res->u.attr = git_attr_internal(name, namelen);
227 else {
228 res->u.pattern = (char *)&(res->state[num_attr]);
229 memcpy(res->u.pattern, name, namelen);
230 res->u.pattern[namelen] = 0;
231 }
232 res->is_macro = is_macro;
233 res->num_attr = num_attr;
234 }
235 return res;
236}
237
238/*
239 * Like info/exclude and .gitignore, the attribute information can
240 * come from many places.
241 *
242 * (1) .gitattribute file of the same directory;
243 * (2) .gitattribute file of the parent directory if (1) does not have
244 * any match; this goes recursively upwards, just like .gitignore.
245 * (3) $GIT_DIR/info/attributes, which overrides both of the above.
246 *
247 * In the same file, later entries override the earlier match, so in the
248 * global list, we would have entries from info/attributes the earliest
249 * (reading the file from top to bottom), .gitattribute of the root
250 * directory (again, reading the file from top to bottom) down to the
251 * current directory, and then scan the list backwards to find the first match.
252 * This is exactly the same as what excluded() does in dir.c to deal with
253 * .gitignore
254 */
255
256static struct attr_stack {
257 struct attr_stack *prev;
258 char *origin;
259 unsigned num_matches;
260 unsigned alloc;
261 struct match_attr **attrs;
262} *attr_stack;
263
264static void free_attr_elem(struct attr_stack *e)
265{
266 int i;
267 free(e->origin);
268 for (i = 0; i < e->num_matches; i++) {
269 struct match_attr *a = e->attrs[i];
270 int j;
271 for (j = 0; j < a->num_attr; j++) {
272 const char *setto = a->state[j].setto;
273 if (setto == ATTR__TRUE ||
274 setto == ATTR__FALSE ||
275 setto == ATTR__UNSET ||
276 setto == ATTR__UNKNOWN)
277 ;
278 else
279 free((char *) setto);
280 }
281 free(a);
282 }
283 free(e);
284}
285
286static const char *builtin_attr[] = {
287 "[attr]binary -diff -text",
288 NULL,
289};
290
291static void handle_attr_line(struct attr_stack *res,
292 const char *line,
293 const char *src,
294 int lineno,
295 int macro_ok)
296{
297 struct match_attr *a;
298
299 a = parse_attr_line(line, src, lineno, macro_ok);
300 if (!a)
301 return;
302 if (res->alloc <= res->num_matches) {
303 res->alloc = alloc_nr(res->num_matches);
304 res->attrs = xrealloc(res->attrs,
305 sizeof(struct match_attr *) *
306 res->alloc);
307 }
308 res->attrs[res->num_matches++] = a;
309}
310
311static struct attr_stack *read_attr_from_array(const char **list)
312{
313 struct attr_stack *res;
314 const char *line;
315 int lineno = 0;
316
317 res = xcalloc(1, sizeof(*res));
318 while ((line = *(list++)) != NULL)
319 handle_attr_line(res, line, "[builtin]", ++lineno, 1);
320 return res;
321}
322
323static enum git_attr_direction direction;
324static struct index_state *use_index;
325
326static struct attr_stack *read_attr_from_file(const char *path, int macro_ok)
327{
328 FILE *fp = fopen(path, "r");
329 struct attr_stack *res;
330 char buf[2048];
331 int lineno = 0;
332
333 if (!fp)
334 return NULL;
335 res = xcalloc(1, sizeof(*res));
336 while (fgets(buf, sizeof(buf), fp))
337 handle_attr_line(res, buf, path, ++lineno, macro_ok);
338 fclose(fp);
339 return res;
340}
341
342static void *read_index_data(const char *path)
343{
344 int pos, len;
345 unsigned long sz;
346 enum object_type type;
347 void *data;
348 struct index_state *istate = use_index ? use_index : &the_index;
349
350 len = strlen(path);
351 pos = index_name_pos(istate, path, len);
352 if (pos < 0) {
353 /*
354 * We might be in the middle of a merge, in which
355 * case we would read stage #2 (ours).
356 */
357 int i;
358 for (i = -pos - 1;
359 (pos < 0 && i < istate->cache_nr &&
360 !strcmp(istate->cache[i]->name, path));
361 i++)
362 if (ce_stage(istate->cache[i]) == 2)
363 pos = i;
364 }
365 if (pos < 0)
366 return NULL;
367 data = read_sha1_file(istate->cache[pos]->sha1, &type, &sz);
368 if (!data || type != OBJ_BLOB) {
369 free(data);
370 return NULL;
371 }
372 return data;
373}
374
375static struct attr_stack *read_attr_from_index(const char *path, int macro_ok)
376{
377 struct attr_stack *res;
378 char *buf, *sp;
379 int lineno = 0;
380
381 buf = read_index_data(path);
382 if (!buf)
383 return NULL;
384
385 res = xcalloc(1, sizeof(*res));
386 for (sp = buf; *sp; ) {
387 char *ep;
388 int more;
389 for (ep = sp; *ep && *ep != '\n'; ep++)
390 ;
391 more = (*ep == '\n');
392 *ep = '\0';
393 handle_attr_line(res, sp, path, ++lineno, macro_ok);
394 sp = ep + more;
395 }
396 free(buf);
397 return res;
398}
399
400static struct attr_stack *read_attr(const char *path, int macro_ok)
401{
402 struct attr_stack *res;
403
404 if (direction == GIT_ATTR_CHECKOUT) {
405 res = read_attr_from_index(path, macro_ok);
406 if (!res)
407 res = read_attr_from_file(path, macro_ok);
408 }
409 else if (direction == GIT_ATTR_CHECKIN) {
410 res = read_attr_from_file(path, macro_ok);
411 if (!res)
412 /*
413 * There is no checked out .gitattributes file there, but
414 * we might have it in the index. We allow operation in a
415 * sparsely checked out work tree, so read from it.
416 */
417 res = read_attr_from_index(path, macro_ok);
418 }
419 else
420 res = read_attr_from_index(path, macro_ok);
421 if (!res)
422 res = xcalloc(1, sizeof(*res));
423 return res;
424}
425
426#if DEBUG_ATTR
427static void debug_info(const char *what, struct attr_stack *elem)
428{
429 fprintf(stderr, "%s: %s\n", what, elem->origin ? elem->origin : "()");
430}
431static void debug_set(const char *what, const char *match, struct git_attr *attr, const void *v)
432{
433 const char *value = v;
434
435 if (ATTR_TRUE(value))
436 value = "set";
437 else if (ATTR_FALSE(value))
438 value = "unset";
439 else if (ATTR_UNSET(value))
440 value = "unspecified";
441
442 fprintf(stderr, "%s: %s => %s (%s)\n",
443 what, attr->name, (char *) value, match);
444}
445#define debug_push(a) debug_info("push", (a))
446#define debug_pop(a) debug_info("pop", (a))
447#else
448#define debug_push(a) do { ; } while (0)
449#define debug_pop(a) do { ; } while (0)
450#define debug_set(a,b,c,d) do { ; } while (0)
451#endif
452
453static void drop_attr_stack(void)
454{
455 while (attr_stack) {
456 struct attr_stack *elem = attr_stack;
457 attr_stack = elem->prev;
458 free_attr_elem(elem);
459 }
460}
461
462static const char *git_etc_gitattributes(void)
463{
464 static const char *system_wide;
465 if (!system_wide)
466 system_wide = system_path(ETC_GITATTRIBUTES);
467 return system_wide;
468}
469
470static int git_attr_system(void)
471{
472 return !git_env_bool("GIT_ATTR_NOSYSTEM", 0);
473}
474
475static int git_attr_config(const char *var, const char *value, void *dummy)
476{
477 if (!strcmp(var, "core.attributesfile"))
478 return git_config_pathname(&attributes_file, var, value);
479
480 return 0;
481}
482
483static void bootstrap_attr_stack(void)
484{
485 if (!attr_stack) {
486 struct attr_stack *elem;
487
488 elem = read_attr_from_array(builtin_attr);
489 elem->origin = NULL;
490 elem->prev = attr_stack;
491 attr_stack = elem;
492
493 if (git_attr_system()) {
494 elem = read_attr_from_file(git_etc_gitattributes(), 1);
495 if (elem) {
496 elem->origin = NULL;
497 elem->prev = attr_stack;
498 attr_stack = elem;
499 }
500 }
501
502 git_config(git_attr_config, NULL);
503 if (attributes_file) {
504 elem = read_attr_from_file(attributes_file, 1);
505 if (elem) {
506 elem->origin = NULL;
507 elem->prev = attr_stack;
508 attr_stack = elem;
509 }
510 }
511
512 if (!is_bare_repository() || direction == GIT_ATTR_INDEX) {
513 elem = read_attr(GITATTRIBUTES_FILE, 1);
514 elem->origin = strdup("");
515 elem->prev = attr_stack;
516 attr_stack = elem;
517 debug_push(elem);
518 }
519
520 elem = read_attr_from_file(git_path(INFOATTRIBUTES_FILE), 1);
521 if (!elem)
522 elem = xcalloc(1, sizeof(*elem));
523 elem->origin = NULL;
524 elem->prev = attr_stack;
525 attr_stack = elem;
526 }
527}
528
529static void prepare_attr_stack(const char *path, int dirlen)
530{
531 struct attr_stack *elem, *info;
532 int len;
533 struct strbuf pathbuf;
534
535 strbuf_init(&pathbuf, dirlen+2+strlen(GITATTRIBUTES_FILE));
536
537 /*
538 * At the bottom of the attribute stack is the built-in
539 * set of attribute definitions, followed by the contents
540 * of $(prefix)/etc/gitattributes and a file specified by
541 * core.attributesfile. Then, contents from
542 * .gitattribute files from directories closer to the
543 * root to the ones in deeper directories are pushed
544 * to the stack. Finally, at the very top of the stack
545 * we always keep the contents of $GIT_DIR/info/attributes.
546 *
547 * When checking, we use entries from near the top of the
548 * stack, preferring $GIT_DIR/info/attributes, then
549 * .gitattributes in deeper directories to shallower ones,
550 * and finally use the built-in set as the default.
551 */
552 if (!attr_stack)
553 bootstrap_attr_stack();
554
555 /*
556 * Pop the "info" one that is always at the top of the stack.
557 */
558 info = attr_stack;
559 attr_stack = info->prev;
560
561 /*
562 * Pop the ones from directories that are not the prefix of
563 * the path we are checking.
564 */
565 while (attr_stack && attr_stack->origin) {
566 int namelen = strlen(attr_stack->origin);
567
568 elem = attr_stack;
569 if (namelen <= dirlen &&
570 !strncmp(elem->origin, path, namelen))
571 break;
572
573 debug_pop(elem);
574 attr_stack = elem->prev;
575 free_attr_elem(elem);
576 }
577
578 /*
579 * Read from parent directories and push them down
580 */
581 if (!is_bare_repository() || direction == GIT_ATTR_INDEX) {
582 while (1) {
583 char *cp;
584
585 len = strlen(attr_stack->origin);
586 if (dirlen <= len)
587 break;
588 strbuf_reset(&pathbuf);
589 strbuf_add(&pathbuf, path, dirlen);
590 strbuf_addch(&pathbuf, '/');
591 cp = strchr(pathbuf.buf + len + 1, '/');
592 strcpy(cp + 1, GITATTRIBUTES_FILE);
593 elem = read_attr(pathbuf.buf, 0);
594 *cp = '\0';
595 elem->origin = strdup(pathbuf.buf);
596 elem->prev = attr_stack;
597 attr_stack = elem;
598 debug_push(elem);
599 }
600 }
601
602 strbuf_release(&pathbuf);
603
604 /*
605 * Finally push the "info" one at the top of the stack.
606 */
607 info->prev = attr_stack;
608 attr_stack = info;
609}
610
611static int path_matches(const char *pathname, int pathlen,
612 const char *pattern,
613 const char *base, int baselen)
614{
615 if (!strchr(pattern, '/')) {
616 /* match basename */
617 const char *basename = strrchr(pathname, '/');
618 basename = basename ? basename + 1 : pathname;
619 return (fnmatch(pattern, basename, 0) == 0);
620 }
621 /*
622 * match with FNM_PATHNAME; the pattern has base implicitly
623 * in front of it.
624 */
625 if (*pattern == '/')
626 pattern++;
627 if (pathlen < baselen ||
628 (baselen && pathname[baselen] != '/') ||
629 strncmp(pathname, base, baselen))
630 return 0;
631 if (baselen != 0)
632 baselen++;
633 return fnmatch(pattern, pathname + baselen, FNM_PATHNAME) == 0;
634}
635
636static int macroexpand_one(int attr_nr, int rem);
637
638static int fill_one(const char *what, struct match_attr *a, int rem)
639{
640 struct git_attr_check *check = check_all_attr;
641 int i;
642
643 for (i = a->num_attr - 1; 0 < rem && 0 <= i; i--) {
644 struct git_attr *attr = a->state[i].attr;
645 const char **n = &(check[attr->attr_nr].value);
646 const char *v = a->state[i].setto;
647
648 if (*n == ATTR__UNKNOWN) {
649 debug_set(what,
650 a->is_macro ? a->u.attr->name : a->u.pattern,
651 attr, v);
652 *n = v;
653 rem--;
654 rem = macroexpand_one(attr->attr_nr, rem);
655 }
656 }
657 return rem;
658}
659
660static int fill(const char *path, int pathlen, struct attr_stack *stk, int rem)
661{
662 int i;
663 const char *base = stk->origin ? stk->origin : "";
664
665 for (i = stk->num_matches - 1; 0 < rem && 0 <= i; i--) {
666 struct match_attr *a = stk->attrs[i];
667 if (a->is_macro)
668 continue;
669 if (path_matches(path, pathlen,
670 a->u.pattern, base, strlen(base)))
671 rem = fill_one("fill", a, rem);
672 }
673 return rem;
674}
675
676static int macroexpand_one(int attr_nr, int rem)
677{
678 struct attr_stack *stk;
679 struct match_attr *a = NULL;
680 int i;
681
682 if (check_all_attr[attr_nr].value != ATTR__TRUE)
683 return rem;
684
685 for (stk = attr_stack; !a && stk; stk = stk->prev)
686 for (i = stk->num_matches - 1; !a && 0 <= i; i--) {
687 struct match_attr *ma = stk->attrs[i];
688 if (!ma->is_macro)
689 continue;
690 if (ma->u.attr->attr_nr == attr_nr)
691 a = ma;
692 }
693
694 if (a)
695 rem = fill_one("expand", a, rem);
696
697 return rem;
698}
699
700int git_checkattr(const char *path, int num, struct git_attr_check *check)
701{
702 struct attr_stack *stk;
703 const char *cp;
704 int dirlen, pathlen, i, rem;
705
706 bootstrap_attr_stack();
707 for (i = 0; i < attr_nr; i++)
708 check_all_attr[i].value = ATTR__UNKNOWN;
709
710 pathlen = strlen(path);
711 cp = strrchr(path, '/');
712 if (!cp)
713 dirlen = 0;
714 else
715 dirlen = cp - path;
716 prepare_attr_stack(path, dirlen);
717 rem = attr_nr;
718 for (stk = attr_stack; 0 < rem && stk; stk = stk->prev)
719 rem = fill(path, pathlen, stk, rem);
720
721 for (i = 0; i < num; i++) {
722 const char *value = check_all_attr[check[i].attr->attr_nr].value;
723 if (value == ATTR__UNKNOWN)
724 value = ATTR__UNSET;
725 check[i].value = value;
726 }
727
728 return 0;
729}
730
731void git_attr_set_direction(enum git_attr_direction new, struct index_state *istate)
732{
733 enum git_attr_direction old = direction;
734
735 if (is_bare_repository() && new != GIT_ATTR_INDEX)
736 die("BUG: non-INDEX attr direction in a bare repo");
737
738 direction = new;
739 if (new != old)
740 drop_attr_stack();
741 use_index = istate;
742}