attr: skip UTF8 BOM at the beginning of the input file
[git.git] / attr.c
blob7c530f4d0c4c70666d49a97ed45e16461ccce702
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.
6 * One basic design decision here is that we are not going to support
7 * an insanely large number of attributes.
8 */
10 #define NO_THE_INDEX_COMPATIBILITY_MACROS
11 #include "cache.h"
12 #include "exec_cmd.h"
13 #include "attr.h"
14 #include "dir.h"
15 #include "utf8.h"
17 const char git_attr__true[] = "(builtin)true";
18 const char git_attr__false[] = "\0(builtin)false";
19 static const char git_attr__unknown[] = "(builtin)unknown";
20 #define ATTR__TRUE git_attr__true
21 #define ATTR__FALSE git_attr__false
22 #define ATTR__UNSET NULL
23 #define ATTR__UNKNOWN git_attr__unknown
25 /* This is a randomly chosen prime. */
26 #define HASHSIZE 257
28 #ifndef DEBUG_ATTR
29 #define DEBUG_ATTR 0
30 #endif
32 struct git_attr {
33 struct git_attr *next;
34 unsigned h;
35 int attr_nr;
36 char name[FLEX_ARRAY];
38 static int attr_nr;
40 static struct git_attr_check *check_all_attr;
41 static struct git_attr *(git_attr_hash[HASHSIZE]);
43 char *git_attr_name(struct git_attr *attr)
45 return attr->name;
48 static unsigned hash_name(const char *name, int namelen)
50 unsigned val = 0, c;
52 while (namelen--) {
53 c = *name++;
54 val = ((val << 7) | (val >> 22)) ^ c;
56 return val;
59 static int invalid_attr_name(const char *name, int namelen)
62 * Attribute name cannot begin with '-' and must consist of
63 * characters from [-A-Za-z0-9_.].
65 if (namelen <= 0 || *name == '-')
66 return -1;
67 while (namelen--) {
68 char ch = *name++;
69 if (! (ch == '-' || ch == '.' || ch == '_' ||
70 ('0' <= ch && ch <= '9') ||
71 ('a' <= ch && ch <= 'z') ||
72 ('A' <= ch && ch <= 'Z')) )
73 return -1;
75 return 0;
78 static struct git_attr *git_attr_internal(const char *name, int len)
80 unsigned hval = hash_name(name, len);
81 unsigned pos = hval % HASHSIZE;
82 struct git_attr *a;
84 for (a = git_attr_hash[pos]; a; a = a->next) {
85 if (a->h == hval &&
86 !memcmp(a->name, name, len) && !a->name[len])
87 return a;
90 if (invalid_attr_name(name, len))
91 return NULL;
93 a = xmalloc(sizeof(*a) + len + 1);
94 memcpy(a->name, name, len);
95 a->name[len] = 0;
96 a->h = hval;
97 a->next = git_attr_hash[pos];
98 a->attr_nr = attr_nr++;
99 git_attr_hash[pos] = a;
101 REALLOC_ARRAY(check_all_attr, attr_nr);
102 check_all_attr[a->attr_nr].attr = a;
103 check_all_attr[a->attr_nr].value = ATTR__UNKNOWN;
104 return a;
107 struct git_attr *git_attr(const char *name)
109 return git_attr_internal(name, strlen(name));
112 /* What does a matched pattern decide? */
113 struct attr_state {
114 struct git_attr *attr;
115 const char *setto;
118 struct pattern {
119 const char *pattern;
120 int patternlen;
121 int nowildcardlen;
122 int flags; /* EXC_FLAG_* */
126 * One rule, as from a .gitattributes file.
128 * If is_macro is true, then u.attr is a pointer to the git_attr being
129 * defined.
131 * If is_macro is false, then u.pattern points at the filename pattern
132 * to which the rule applies. (The memory pointed to is part of the
133 * memory block allocated for the match_attr instance.)
135 * In either case, num_attr is the number of attributes affected by
136 * this rule, and state is an array listing them. The attributes are
137 * listed as they appear in the file (macros unexpanded).
139 struct match_attr {
140 union {
141 struct pattern pat;
142 struct git_attr *attr;
143 } u;
144 char is_macro;
145 unsigned num_attr;
146 struct attr_state state[FLEX_ARRAY];
149 static const char blank[] = " \t\r\n";
152 * Parse a whitespace-delimited attribute state (i.e., "attr",
153 * "-attr", "!attr", or "attr=value") from the string starting at src.
154 * If e is not NULL, write the results to *e. Return a pointer to the
155 * remainder of the string (with leading whitespace removed), or NULL
156 * if there was an error.
158 static const char *parse_attr(const char *src, int lineno, const char *cp,
159 struct attr_state *e)
161 const char *ep, *equals;
162 int len;
164 ep = cp + strcspn(cp, blank);
165 equals = strchr(cp, '=');
166 if (equals && ep < equals)
167 equals = NULL;
168 if (equals)
169 len = equals - cp;
170 else
171 len = ep - cp;
172 if (!e) {
173 if (*cp == '-' || *cp == '!') {
174 cp++;
175 len--;
177 if (invalid_attr_name(cp, len)) {
178 fprintf(stderr,
179 "%.*s is not a valid attribute name: %s:%d\n",
180 len, cp, src, lineno);
181 return NULL;
183 } else {
184 if (*cp == '-' || *cp == '!') {
185 e->setto = (*cp == '-') ? ATTR__FALSE : ATTR__UNSET;
186 cp++;
187 len--;
189 else if (!equals)
190 e->setto = ATTR__TRUE;
191 else {
192 e->setto = xmemdupz(equals + 1, ep - equals - 1);
194 e->attr = git_attr_internal(cp, len);
196 return ep + strspn(ep, blank);
199 static struct match_attr *parse_attr_line(const char *line, const char *src,
200 int lineno, int macro_ok)
202 int namelen;
203 int num_attr, i;
204 const char *cp, *name, *states;
205 struct match_attr *res = NULL;
206 int is_macro;
208 cp = line + strspn(line, blank);
209 if (!*cp || *cp == '#')
210 return NULL;
211 name = cp;
212 namelen = strcspn(name, blank);
213 if (strlen(ATTRIBUTE_MACRO_PREFIX) < namelen &&
214 starts_with(name, ATTRIBUTE_MACRO_PREFIX)) {
215 if (!macro_ok) {
216 fprintf(stderr, "%s not allowed: %s:%d\n",
217 name, src, lineno);
218 return NULL;
220 is_macro = 1;
221 name += strlen(ATTRIBUTE_MACRO_PREFIX);
222 name += strspn(name, blank);
223 namelen = strcspn(name, blank);
224 if (invalid_attr_name(name, namelen)) {
225 fprintf(stderr,
226 "%.*s is not a valid attribute name: %s:%d\n",
227 namelen, name, src, lineno);
228 return NULL;
231 else
232 is_macro = 0;
234 states = name + namelen;
235 states += strspn(states, blank);
237 /* First pass to count the attr_states */
238 for (cp = states, num_attr = 0; *cp; num_attr++) {
239 cp = parse_attr(src, lineno, cp, NULL);
240 if (!cp)
241 return NULL;
244 res = xcalloc(1,
245 sizeof(*res) +
246 sizeof(struct attr_state) * num_attr +
247 (is_macro ? 0 : namelen + 1));
248 if (is_macro)
249 res->u.attr = git_attr_internal(name, namelen);
250 else {
251 char *p = (char *)&(res->state[num_attr]);
252 memcpy(p, name, namelen);
253 res->u.pat.pattern = p;
254 parse_exclude_pattern(&res->u.pat.pattern,
255 &res->u.pat.patternlen,
256 &res->u.pat.flags,
257 &res->u.pat.nowildcardlen);
258 if (res->u.pat.flags & EXC_FLAG_NEGATIVE) {
259 warning(_("Negative patterns are ignored in git attributes\n"
260 "Use '\\!' for literal leading exclamation."));
261 return NULL;
264 res->is_macro = is_macro;
265 res->num_attr = num_attr;
267 /* Second pass to fill the attr_states */
268 for (cp = states, i = 0; *cp; i++) {
269 cp = parse_attr(src, lineno, cp, &(res->state[i]));
272 return res;
276 * Like info/exclude and .gitignore, the attribute information can
277 * come from many places.
279 * (1) .gitattribute file of the same directory;
280 * (2) .gitattribute file of the parent directory if (1) does not have
281 * any match; this goes recursively upwards, just like .gitignore.
282 * (3) $GIT_DIR/info/attributes, which overrides both of the above.
284 * In the same file, later entries override the earlier match, so in the
285 * global list, we would have entries from info/attributes the earliest
286 * (reading the file from top to bottom), .gitattribute of the root
287 * directory (again, reading the file from top to bottom) down to the
288 * current directory, and then scan the list backwards to find the first match.
289 * This is exactly the same as what is_excluded() does in dir.c to deal with
290 * .gitignore
293 static struct attr_stack {
294 struct attr_stack *prev;
295 char *origin;
296 size_t originlen;
297 unsigned num_matches;
298 unsigned alloc;
299 struct match_attr **attrs;
300 } *attr_stack;
302 static void free_attr_elem(struct attr_stack *e)
304 int i;
305 free(e->origin);
306 for (i = 0; i < e->num_matches; i++) {
307 struct match_attr *a = e->attrs[i];
308 int j;
309 for (j = 0; j < a->num_attr; j++) {
310 const char *setto = a->state[j].setto;
311 if (setto == ATTR__TRUE ||
312 setto == ATTR__FALSE ||
313 setto == ATTR__UNSET ||
314 setto == ATTR__UNKNOWN)
316 else
317 free((char *) setto);
319 free(a);
321 free(e->attrs);
322 free(e);
325 static const char *builtin_attr[] = {
326 "[attr]binary -diff -merge -text",
327 NULL,
330 static void handle_attr_line(struct attr_stack *res,
331 const char *line,
332 const char *src,
333 int lineno,
334 int macro_ok)
336 struct match_attr *a;
338 a = parse_attr_line(line, src, lineno, macro_ok);
339 if (!a)
340 return;
341 ALLOC_GROW(res->attrs, res->num_matches + 1, res->alloc);
342 res->attrs[res->num_matches++] = a;
345 static struct attr_stack *read_attr_from_array(const char **list)
347 struct attr_stack *res;
348 const char *line;
349 int lineno = 0;
351 res = xcalloc(1, sizeof(*res));
352 while ((line = *(list++)) != NULL)
353 handle_attr_line(res, line, "[builtin]", ++lineno, 1);
354 return res;
357 static enum git_attr_direction direction;
358 static struct index_state *use_index;
360 static struct attr_stack *read_attr_from_file(const char *path, int macro_ok)
362 FILE *fp = fopen(path, "r");
363 struct attr_stack *res;
364 char buf[2048];
365 int lineno = 0;
367 if (!fp) {
368 if (errno != ENOENT && errno != ENOTDIR)
369 warn_on_inaccessible(path);
370 return NULL;
372 res = xcalloc(1, sizeof(*res));
373 while (fgets(buf, sizeof(buf), fp)) {
374 char *bufp = buf;
375 if (!lineno)
376 skip_utf8_bom(&bufp, strlen(bufp));
377 handle_attr_line(res, bufp, path, ++lineno, macro_ok);
379 fclose(fp);
380 return res;
383 static struct attr_stack *read_attr_from_index(const char *path, int macro_ok)
385 struct attr_stack *res;
386 char *buf, *sp;
387 int lineno = 0;
389 buf = read_blob_data_from_index(use_index ? use_index : &the_index, path, NULL);
390 if (!buf)
391 return NULL;
393 res = xcalloc(1, sizeof(*res));
394 for (sp = buf; *sp; ) {
395 char *ep;
396 int more;
397 for (ep = sp; *ep && *ep != '\n'; ep++)
399 more = (*ep == '\n');
400 *ep = '\0';
401 handle_attr_line(res, sp, path, ++lineno, macro_ok);
402 sp = ep + more;
404 free(buf);
405 return res;
408 static struct attr_stack *read_attr(const char *path, int macro_ok)
410 struct attr_stack *res;
412 if (direction == GIT_ATTR_CHECKOUT) {
413 res = read_attr_from_index(path, macro_ok);
414 if (!res)
415 res = read_attr_from_file(path, macro_ok);
417 else if (direction == GIT_ATTR_CHECKIN) {
418 res = read_attr_from_file(path, macro_ok);
419 if (!res)
421 * There is no checked out .gitattributes file there, but
422 * we might have it in the index. We allow operation in a
423 * sparsely checked out work tree, so read from it.
425 res = read_attr_from_index(path, macro_ok);
427 else
428 res = read_attr_from_index(path, macro_ok);
429 if (!res)
430 res = xcalloc(1, sizeof(*res));
431 return res;
434 #if DEBUG_ATTR
435 static void debug_info(const char *what, struct attr_stack *elem)
437 fprintf(stderr, "%s: %s\n", what, elem->origin ? elem->origin : "()");
439 static void debug_set(const char *what, const char *match, struct git_attr *attr, const void *v)
441 const char *value = v;
443 if (ATTR_TRUE(value))
444 value = "set";
445 else if (ATTR_FALSE(value))
446 value = "unset";
447 else if (ATTR_UNSET(value))
448 value = "unspecified";
450 fprintf(stderr, "%s: %s => %s (%s)\n",
451 what, attr->name, (char *) value, match);
453 #define debug_push(a) debug_info("push", (a))
454 #define debug_pop(a) debug_info("pop", (a))
455 #else
456 #define debug_push(a) do { ; } while (0)
457 #define debug_pop(a) do { ; } while (0)
458 #define debug_set(a,b,c,d) do { ; } while (0)
459 #endif
461 static void drop_attr_stack(void)
463 while (attr_stack) {
464 struct attr_stack *elem = attr_stack;
465 attr_stack = elem->prev;
466 free_attr_elem(elem);
470 static const char *git_etc_gitattributes(void)
472 static const char *system_wide;
473 if (!system_wide)
474 system_wide = system_path(ETC_GITATTRIBUTES);
475 return system_wide;
478 static int git_attr_system(void)
480 return !git_env_bool("GIT_ATTR_NOSYSTEM", 0);
483 static void bootstrap_attr_stack(void)
485 struct attr_stack *elem;
486 char *xdg_attributes_file;
488 if (attr_stack)
489 return;
491 elem = read_attr_from_array(builtin_attr);
492 elem->origin = NULL;
493 elem->prev = attr_stack;
494 attr_stack = elem;
496 if (git_attr_system()) {
497 elem = read_attr_from_file(git_etc_gitattributes(), 1);
498 if (elem) {
499 elem->origin = NULL;
500 elem->prev = attr_stack;
501 attr_stack = elem;
505 if (!git_attributes_file) {
506 home_config_paths(NULL, &xdg_attributes_file, "attributes");
507 git_attributes_file = xdg_attributes_file;
509 if (git_attributes_file) {
510 elem = read_attr_from_file(git_attributes_file, 1);
511 if (elem) {
512 elem->origin = NULL;
513 elem->prev = attr_stack;
514 attr_stack = elem;
518 if (!is_bare_repository() || direction == GIT_ATTR_INDEX) {
519 elem = read_attr(GITATTRIBUTES_FILE, 1);
520 elem->origin = xstrdup("");
521 elem->originlen = 0;
522 elem->prev = attr_stack;
523 attr_stack = elem;
524 debug_push(elem);
527 elem = read_attr_from_file(git_path(INFOATTRIBUTES_FILE), 1);
528 if (!elem)
529 elem = xcalloc(1, sizeof(*elem));
530 elem->origin = NULL;
531 elem->prev = attr_stack;
532 attr_stack = elem;
535 static void prepare_attr_stack(const char *path, int dirlen)
537 struct attr_stack *elem, *info;
538 int len;
539 const char *cp;
542 * At the bottom of the attribute stack is the built-in
543 * set of attribute definitions, followed by the contents
544 * of $(prefix)/etc/gitattributes and a file specified by
545 * core.attributesfile. Then, contents from
546 * .gitattribute files from directories closer to the
547 * root to the ones in deeper directories are pushed
548 * to the stack. Finally, at the very top of the stack
549 * we always keep the contents of $GIT_DIR/info/attributes.
551 * When checking, we use entries from near the top of the
552 * stack, preferring $GIT_DIR/info/attributes, then
553 * .gitattributes in deeper directories to shallower ones,
554 * and finally use the built-in set as the default.
556 bootstrap_attr_stack();
559 * Pop the "info" one that is always at the top of the stack.
561 info = attr_stack;
562 attr_stack = info->prev;
565 * Pop the ones from directories that are not the prefix of
566 * the path we are checking. Break out of the loop when we see
567 * the root one (whose origin is an empty string "") or the builtin
568 * one (whose origin is NULL) without popping it.
570 while (attr_stack->origin) {
571 int namelen = strlen(attr_stack->origin);
573 elem = attr_stack;
574 if (namelen <= dirlen &&
575 !strncmp(elem->origin, path, namelen) &&
576 (!namelen || path[namelen] == '/'))
577 break;
579 debug_pop(elem);
580 attr_stack = elem->prev;
581 free_attr_elem(elem);
585 * Read from parent directories and push them down
587 if (!is_bare_repository() || direction == GIT_ATTR_INDEX) {
589 * bootstrap_attr_stack() should have added, and the
590 * above loop should have stopped before popping, the
591 * root element whose attr_stack->origin is set to an
592 * empty string.
594 struct strbuf pathbuf = STRBUF_INIT;
596 assert(attr_stack->origin);
597 while (1) {
598 len = strlen(attr_stack->origin);
599 if (dirlen <= len)
600 break;
601 cp = memchr(path + len + 1, '/', dirlen - len - 1);
602 if (!cp)
603 cp = path + dirlen;
604 strbuf_add(&pathbuf, path, cp - path);
605 strbuf_addch(&pathbuf, '/');
606 strbuf_addstr(&pathbuf, GITATTRIBUTES_FILE);
607 elem = read_attr(pathbuf.buf, 0);
608 strbuf_setlen(&pathbuf, cp - path);
609 elem->origin = strbuf_detach(&pathbuf, &elem->originlen);
610 elem->prev = attr_stack;
611 attr_stack = elem;
612 debug_push(elem);
615 strbuf_release(&pathbuf);
619 * Finally push the "info" one at the top of the stack.
621 info->prev = attr_stack;
622 attr_stack = info;
625 static int path_matches(const char *pathname, int pathlen,
626 int basename_offset,
627 const struct pattern *pat,
628 const char *base, int baselen)
630 const char *pattern = pat->pattern;
631 int prefix = pat->nowildcardlen;
632 int isdir = (pathlen && pathname[pathlen - 1] == '/');
634 if ((pat->flags & EXC_FLAG_MUSTBEDIR) && !isdir)
635 return 0;
637 if (pat->flags & EXC_FLAG_NODIR) {
638 return match_basename(pathname + basename_offset,
639 pathlen - basename_offset - isdir,
640 pattern, prefix,
641 pat->patternlen, pat->flags);
643 return match_pathname(pathname, pathlen - isdir,
644 base, baselen,
645 pattern, prefix, pat->patternlen, pat->flags);
648 static int macroexpand_one(int attr_nr, int rem);
650 static int fill_one(const char *what, struct match_attr *a, int rem)
652 struct git_attr_check *check = check_all_attr;
653 int i;
655 for (i = a->num_attr - 1; 0 < rem && 0 <= i; i--) {
656 struct git_attr *attr = a->state[i].attr;
657 const char **n = &(check[attr->attr_nr].value);
658 const char *v = a->state[i].setto;
660 if (*n == ATTR__UNKNOWN) {
661 debug_set(what,
662 a->is_macro ? a->u.attr->name : a->u.pat.pattern,
663 attr, v);
664 *n = v;
665 rem--;
666 rem = macroexpand_one(attr->attr_nr, rem);
669 return rem;
672 static int fill(const char *path, int pathlen, int basename_offset,
673 struct attr_stack *stk, int rem)
675 int i;
676 const char *base = stk->origin ? stk->origin : "";
678 for (i = stk->num_matches - 1; 0 < rem && 0 <= i; i--) {
679 struct match_attr *a = stk->attrs[i];
680 if (a->is_macro)
681 continue;
682 if (path_matches(path, pathlen, basename_offset,
683 &a->u.pat, base, stk->originlen))
684 rem = fill_one("fill", a, rem);
686 return rem;
689 static int macroexpand_one(int attr_nr, int rem)
691 struct attr_stack *stk;
692 struct match_attr *a = NULL;
693 int i;
695 if (check_all_attr[attr_nr].value != ATTR__TRUE)
696 return rem;
698 for (stk = attr_stack; !a && stk; stk = stk->prev)
699 for (i = stk->num_matches - 1; !a && 0 <= i; i--) {
700 struct match_attr *ma = stk->attrs[i];
701 if (!ma->is_macro)
702 continue;
703 if (ma->u.attr->attr_nr == attr_nr)
704 a = ma;
707 if (a)
708 rem = fill_one("expand", a, rem);
710 return rem;
714 * Collect all attributes for path into the array pointed to by
715 * check_all_attr.
717 static void collect_all_attrs(const char *path)
719 struct attr_stack *stk;
720 int i, pathlen, rem, dirlen;
721 const char *cp, *last_slash = NULL;
722 int basename_offset;
724 for (cp = path; *cp; cp++) {
725 if (*cp == '/' && cp[1])
726 last_slash = cp;
728 pathlen = cp - path;
729 if (last_slash) {
730 basename_offset = last_slash + 1 - path;
731 dirlen = last_slash - path;
732 } else {
733 basename_offset = 0;
734 dirlen = 0;
737 prepare_attr_stack(path, dirlen);
738 for (i = 0; i < attr_nr; i++)
739 check_all_attr[i].value = ATTR__UNKNOWN;
741 rem = attr_nr;
742 for (stk = attr_stack; 0 < rem && stk; stk = stk->prev)
743 rem = fill(path, pathlen, basename_offset, stk, rem);
746 int git_check_attr(const char *path, int num, struct git_attr_check *check)
748 int i;
750 collect_all_attrs(path);
752 for (i = 0; i < num; i++) {
753 const char *value = check_all_attr[check[i].attr->attr_nr].value;
754 if (value == ATTR__UNKNOWN)
755 value = ATTR__UNSET;
756 check[i].value = value;
759 return 0;
762 int git_all_attrs(const char *path, int *num, struct git_attr_check **check)
764 int i, count, j;
766 collect_all_attrs(path);
768 /* Count the number of attributes that are set. */
769 count = 0;
770 for (i = 0; i < attr_nr; i++) {
771 const char *value = check_all_attr[i].value;
772 if (value != ATTR__UNSET && value != ATTR__UNKNOWN)
773 ++count;
775 *num = count;
776 *check = xmalloc(sizeof(**check) * count);
777 j = 0;
778 for (i = 0; i < attr_nr; i++) {
779 const char *value = check_all_attr[i].value;
780 if (value != ATTR__UNSET && value != ATTR__UNKNOWN) {
781 (*check)[j].attr = check_all_attr[i].attr;
782 (*check)[j].value = value;
783 ++j;
787 return 0;
790 void git_attr_set_direction(enum git_attr_direction new, struct index_state *istate)
792 enum git_attr_direction old = direction;
794 if (is_bare_repository() && new != GIT_ATTR_INDEX)
795 die("BUG: non-INDEX attr direction in a bare repo");
797 direction = new;
798 if (new != old)
799 drop_attr_stack();
800 use_index = istate;