pack-check.c::verify_packfile(): don't run SHA-1 update on huge data
[git/mingw/4msysgit/wingit-dll.git] / sha1_name.c
blob6d7cd78381414aa2fef31d31d46fbb24b0aaab1d
1 #include "cache.h"
2 #include "tag.h"
3 #include "commit.h"
4 #include "tree.h"
5 #include "blob.h"
6 #include "tree-walk.h"
7 #include "refs.h"
9 static int find_short_object_filename(int len, const char *name, unsigned char *sha1)
11 struct alternate_object_database *alt;
12 char hex[40];
13 int found = 0;
14 static struct alternate_object_database *fakeent;
16 if (!fakeent) {
17 const char *objdir = get_object_directory();
18 int objdir_len = strlen(objdir);
19 int entlen = objdir_len + 43;
20 fakeent = xmalloc(sizeof(*fakeent) + entlen);
21 memcpy(fakeent->base, objdir, objdir_len);
22 fakeent->name = fakeent->base + objdir_len + 1;
23 fakeent->name[-1] = '/';
25 fakeent->next = alt_odb_list;
27 sprintf(hex, "%.2s", name);
28 for (alt = fakeent; alt && found < 2; alt = alt->next) {
29 struct dirent *de;
30 DIR *dir;
31 sprintf(alt->name, "%.2s/", name);
32 dir = opendir(alt->base);
33 if (!dir)
34 continue;
35 while ((de = readdir(dir)) != NULL) {
36 if (strlen(de->d_name) != 38)
37 continue;
38 if (memcmp(de->d_name, name + 2, len - 2))
39 continue;
40 if (!found) {
41 memcpy(hex + 2, de->d_name, 38);
42 found++;
44 else if (memcmp(hex + 2, de->d_name, 38)) {
45 found = 2;
46 break;
49 closedir(dir);
51 if (found == 1)
52 return get_sha1_hex(hex, sha1) == 0;
53 return found;
56 static int match_sha(unsigned len, const unsigned char *a, const unsigned char *b)
58 do {
59 if (*a != *b)
60 return 0;
61 a++;
62 b++;
63 len -= 2;
64 } while (len > 1);
65 if (len)
66 if ((*a ^ *b) & 0xf0)
67 return 0;
68 return 1;
71 static int find_short_packed_object(int len, const unsigned char *match, unsigned char *sha1)
73 struct packed_git *p;
74 unsigned char found_sha1[20];
75 int found = 0;
77 prepare_packed_git();
78 for (p = packed_git; p && found < 2; p = p->next) {
79 unsigned num = num_packed_objects(p);
80 unsigned first = 0, last = num;
81 while (first < last) {
82 unsigned mid = (first + last) / 2;
83 unsigned char now[20];
84 int cmp;
86 nth_packed_object_sha1(p, mid, now);
87 cmp = hashcmp(match, now);
88 if (!cmp) {
89 first = mid;
90 break;
92 if (cmp > 0) {
93 first = mid+1;
94 continue;
96 last = mid;
98 if (first < num) {
99 unsigned char now[20], next[20];
100 nth_packed_object_sha1(p, first, now);
101 if (match_sha(len, match, now)) {
102 if (nth_packed_object_sha1(p, first+1, next) ||
103 !match_sha(len, match, next)) {
104 /* unique within this pack */
105 if (!found) {
106 hashcpy(found_sha1, now);
107 found++;
109 else if (hashcmp(found_sha1, now)) {
110 found = 2;
111 break;
114 else {
115 /* not even unique within this pack */
116 found = 2;
117 break;
122 if (found == 1)
123 hashcpy(sha1, found_sha1);
124 return found;
127 #define SHORT_NAME_NOT_FOUND (-1)
128 #define SHORT_NAME_AMBIGUOUS (-2)
130 static int find_unique_short_object(int len, char *canonical,
131 unsigned char *res, unsigned char *sha1)
133 int has_unpacked, has_packed;
134 unsigned char unpacked_sha1[20], packed_sha1[20];
136 has_unpacked = find_short_object_filename(len, canonical, unpacked_sha1);
137 has_packed = find_short_packed_object(len, res, packed_sha1);
138 if (!has_unpacked && !has_packed)
139 return SHORT_NAME_NOT_FOUND;
140 if (1 < has_unpacked || 1 < has_packed)
141 return SHORT_NAME_AMBIGUOUS;
142 if (has_unpacked != has_packed) {
143 hashcpy(sha1, (has_packed ? packed_sha1 : unpacked_sha1));
144 return 0;
146 /* Both have unique ones -- do they match? */
147 if (hashcmp(packed_sha1, unpacked_sha1))
148 return SHORT_NAME_AMBIGUOUS;
149 hashcpy(sha1, packed_sha1);
150 return 0;
153 static int get_short_sha1(const char *name, int len, unsigned char *sha1,
154 int quietly)
156 int i, status;
157 char canonical[40];
158 unsigned char res[20];
160 if (len < MINIMUM_ABBREV || len > 40)
161 return -1;
162 hashclr(res);
163 memset(canonical, 'x', 40);
164 for (i = 0; i < len ;i++) {
165 unsigned char c = name[i];
166 unsigned char val;
167 if (c >= '0' && c <= '9')
168 val = c - '0';
169 else if (c >= 'a' && c <= 'f')
170 val = c - 'a' + 10;
171 else if (c >= 'A' && c <='F') {
172 val = c - 'A' + 10;
173 c -= 'A' - 'a';
175 else
176 return -1;
177 canonical[i] = c;
178 if (!(i & 1))
179 val <<= 4;
180 res[i >> 1] |= val;
183 status = find_unique_short_object(i, canonical, res, sha1);
184 if (!quietly && (status == SHORT_NAME_AMBIGUOUS))
185 return error("short SHA1 %.*s is ambiguous.", len, canonical);
186 return status;
189 const char *find_unique_abbrev(const unsigned char *sha1, int len)
191 int status, is_null;
192 static char hex[41];
194 is_null = is_null_sha1(sha1);
195 memcpy(hex, sha1_to_hex(sha1), 40);
196 if (len == 40 || !len)
197 return hex;
198 while (len < 40) {
199 unsigned char sha1_ret[20];
200 status = get_short_sha1(hex, len, sha1_ret, 1);
201 if (!status ||
202 (is_null && status != SHORT_NAME_AMBIGUOUS)) {
203 hex[len] = 0;
204 return hex;
206 if (status != SHORT_NAME_AMBIGUOUS)
207 return NULL;
208 len++;
210 return NULL;
213 static int ambiguous_path(const char *path, int len)
215 int slash = 1;
216 int cnt;
218 for (cnt = 0; cnt < len; cnt++) {
219 switch (*path++) {
220 case '\0':
221 break;
222 case '/':
223 if (slash)
224 break;
225 slash = 1;
226 continue;
227 case '.':
228 continue;
229 default:
230 slash = 0;
231 continue;
233 break;
235 return slash;
238 static int get_sha1_basic(const char *str, int len, unsigned char *sha1)
240 static const char *fmt[] = {
241 "%.*s",
242 "refs/%.*s",
243 "refs/tags/%.*s",
244 "refs/heads/%.*s",
245 "refs/remotes/%.*s",
246 "refs/remotes/%.*s/HEAD",
247 NULL
249 static const char *warning = "warning: refname '%.*s' is ambiguous.\n";
250 const char **p, *ref;
251 char *real_ref = NULL;
252 int refs_found = 0;
253 int at, reflog_len;
254 unsigned char *this_result;
255 unsigned char sha1_from_ref[20];
257 if (len == 40 && !get_sha1_hex(str, sha1))
258 return 0;
260 /* basic@{time or number} format to query ref-log */
261 reflog_len = at = 0;
262 if (str[len-1] == '}') {
263 for (at = 1; at < len - 1; at++) {
264 if (str[at] == '@' && str[at+1] == '{') {
265 reflog_len = (len-1) - (at+2);
266 len = at;
267 break;
272 /* Accept only unambiguous ref paths. */
273 if (ambiguous_path(str, len))
274 return -1;
276 for (p = fmt; *p; p++) {
277 this_result = refs_found ? sha1_from_ref : sha1;
278 ref = resolve_ref(mkpath(*p, len, str), this_result, 1, NULL);
279 if (ref) {
280 if (!refs_found++)
281 real_ref = xstrdup(ref);
282 if (!warn_ambiguous_refs)
283 break;
287 if (!refs_found)
288 return -1;
290 if (warn_ambiguous_refs && refs_found > 1)
291 fprintf(stderr, warning, len, str);
293 if (reflog_len) {
294 /* Is it asking for N-th entry, or approxidate? */
295 int nth, i;
296 unsigned long at_time;
297 for (i = nth = 0; 0 <= nth && i < reflog_len; i++) {
298 char ch = str[at+2+i];
299 if ('0' <= ch && ch <= '9')
300 nth = nth * 10 + ch - '0';
301 else
302 nth = -1;
304 if (0 <= nth)
305 at_time = 0;
306 else
307 at_time = approxidate(str + at + 2);
308 read_ref_at(real_ref, at_time, nth, sha1);
311 free(real_ref);
312 return 0;
315 static int get_sha1_1(const char *name, int len, unsigned char *sha1);
317 static int get_parent(const char *name, int len,
318 unsigned char *result, int idx)
320 unsigned char sha1[20];
321 int ret = get_sha1_1(name, len, sha1);
322 struct commit *commit;
323 struct commit_list *p;
325 if (ret)
326 return ret;
327 commit = lookup_commit_reference(sha1);
328 if (!commit)
329 return -1;
330 if (parse_commit(commit))
331 return -1;
332 if (!idx) {
333 hashcpy(result, commit->object.sha1);
334 return 0;
336 p = commit->parents;
337 while (p) {
338 if (!--idx) {
339 hashcpy(result, p->item->object.sha1);
340 return 0;
342 p = p->next;
344 return -1;
347 static int get_nth_ancestor(const char *name, int len,
348 unsigned char *result, int generation)
350 unsigned char sha1[20];
351 int ret = get_sha1_1(name, len, sha1);
352 if (ret)
353 return ret;
355 while (generation--) {
356 struct commit *commit = lookup_commit_reference(sha1);
358 if (!commit || parse_commit(commit) || !commit->parents)
359 return -1;
360 hashcpy(sha1, commit->parents->item->object.sha1);
362 hashcpy(result, sha1);
363 return 0;
366 static int peel_onion(const char *name, int len, unsigned char *sha1)
368 unsigned char outer[20];
369 const char *sp;
370 unsigned int expected_type = 0;
371 struct object *o;
374 * "ref^{type}" dereferences ref repeatedly until you cannot
375 * dereference anymore, or you get an object of given type,
376 * whichever comes first. "ref^{}" means just dereference
377 * tags until you get a non-tag. "ref^0" is a shorthand for
378 * "ref^{commit}". "commit^{tree}" could be used to find the
379 * top-level tree of the given commit.
381 if (len < 4 || name[len-1] != '}')
382 return -1;
384 for (sp = name + len - 1; name <= sp; sp--) {
385 int ch = *sp;
386 if (ch == '{' && name < sp && sp[-1] == '^')
387 break;
389 if (sp <= name)
390 return -1;
392 sp++; /* beginning of type name, or closing brace for empty */
393 if (!strncmp(commit_type, sp, 6) && sp[6] == '}')
394 expected_type = OBJ_COMMIT;
395 else if (!strncmp(tree_type, sp, 4) && sp[4] == '}')
396 expected_type = OBJ_TREE;
397 else if (!strncmp(blob_type, sp, 4) && sp[4] == '}')
398 expected_type = OBJ_BLOB;
399 else if (sp[0] == '}')
400 expected_type = OBJ_NONE;
401 else
402 return -1;
404 if (get_sha1_1(name, sp - name - 2, outer))
405 return -1;
407 o = parse_object(outer);
408 if (!o)
409 return -1;
410 if (!expected_type) {
411 o = deref_tag(o, name, sp - name - 2);
412 if (!o || (!o->parsed && !parse_object(o->sha1)))
413 return -1;
414 hashcpy(sha1, o->sha1);
416 else {
417 /* At this point, the syntax look correct, so
418 * if we do not get the needed object, we should
419 * barf.
422 while (1) {
423 if (!o || (!o->parsed && !parse_object(o->sha1)))
424 return -1;
425 if (o->type == expected_type) {
426 hashcpy(sha1, o->sha1);
427 return 0;
429 if (o->type == OBJ_TAG)
430 o = ((struct tag*) o)->tagged;
431 else if (o->type == OBJ_COMMIT)
432 o = &(((struct commit *) o)->tree->object);
433 else
434 return error("%.*s: expected %s type, but the object dereferences to %s type",
435 len, name, typename(expected_type),
436 typename(o->type));
437 if (!o->parsed)
438 parse_object(o->sha1);
441 return 0;
444 static int get_describe_name(const char *name, int len, unsigned char *sha1)
446 const char *cp;
448 for (cp = name + len - 1; name + 2 <= cp; cp--) {
449 char ch = *cp;
450 if (hexval(ch) & ~0377) {
451 /* We must be looking at g in "SOMETHING-g"
452 * for it to be describe output.
454 if (ch == 'g' && cp[-1] == '-') {
455 cp++;
456 len -= cp - name;
457 return get_short_sha1(cp, len, sha1, 1);
461 return -1;
464 static int get_sha1_1(const char *name, int len, unsigned char *sha1)
466 int ret, has_suffix;
467 const char *cp;
469 /* "name~3" is "name^^^",
470 * "name~" and "name~0" are name -- not "name^0"!
471 * "name^" is not "name^0"; it is "name^1".
473 has_suffix = 0;
474 for (cp = name + len - 1; name <= cp; cp--) {
475 int ch = *cp;
476 if ('0' <= ch && ch <= '9')
477 continue;
478 if (ch == '~' || ch == '^')
479 has_suffix = ch;
480 break;
483 if (has_suffix) {
484 int num = 0;
485 int len1 = cp - name;
486 cp++;
487 while (cp < name + len)
488 num = num * 10 + *cp++ - '0';
489 if (has_suffix == '^') {
490 if (!num && len1 == len - 1)
491 num = 1;
492 return get_parent(name, len1, sha1, num);
494 /* else if (has_suffix == '~') -- goes without saying */
495 return get_nth_ancestor(name, len1, sha1, num);
498 ret = peel_onion(name, len, sha1);
499 if (!ret)
500 return 0;
502 ret = get_sha1_basic(name, len, sha1);
503 if (!ret)
504 return 0;
506 /* It could be describe output that is "SOMETHING-gXXXX" */
507 ret = get_describe_name(name, len, sha1);
508 if (!ret)
509 return 0;
511 return get_short_sha1(name, len, sha1, 0);
515 * This is like "get_sha1_basic()", except it allows "sha1 expressions",
516 * notably "xyz^" for "parent of xyz"
518 int get_sha1(const char *name, unsigned char *sha1)
520 int ret, bracket_depth;
521 unsigned unused;
522 int namelen = strlen(name);
523 const char *cp;
525 prepare_alt_odb();
526 ret = get_sha1_1(name, namelen, sha1);
527 if (!ret)
528 return ret;
529 /* sha1:path --> object name of path in ent sha1
530 * :path -> object name of path in index
531 * :[0-3]:path -> object name of path in index at stage
533 if (name[0] == ':') {
534 int stage = 0;
535 struct cache_entry *ce;
536 int pos;
537 if (namelen < 3 ||
538 name[2] != ':' ||
539 name[1] < '0' || '3' < name[1])
540 cp = name + 1;
541 else {
542 stage = name[1] - '0';
543 cp = name + 3;
545 namelen = namelen - (cp - name);
546 if (!active_cache)
547 read_cache();
548 if (active_nr < 0)
549 return -1;
550 pos = cache_name_pos(cp, namelen);
551 if (pos < 0)
552 pos = -pos - 1;
553 while (pos < active_nr) {
554 ce = active_cache[pos];
555 if (ce_namelen(ce) != namelen ||
556 memcmp(ce->name, cp, namelen))
557 break;
558 if (ce_stage(ce) == stage) {
559 hashcpy(sha1, ce->sha1);
560 return 0;
562 pos++;
564 return -1;
566 for (cp = name, bracket_depth = 0; *cp; cp++) {
567 if (*cp == '{')
568 bracket_depth++;
569 else if (bracket_depth && *cp == '}')
570 bracket_depth--;
571 else if (!bracket_depth && *cp == ':')
572 break;
574 if (*cp == ':') {
575 unsigned char tree_sha1[20];
576 if (!get_sha1_1(name, cp-name, tree_sha1))
577 return get_tree_entry(tree_sha1, cp+1, sha1,
578 &unused);
580 return ret;