git-rm: Fix to properly handle files with spaces, tabs, newlines, etc.
[git/dkf.git] / sha1_name.c
blobd67de18ba5f5aa1f5b3dd1c5c81e75618d12e330
1 #include "cache.h"
2 #include "tag.h"
3 #include "commit.h"
4 #include "tree.h"
5 #include "blob.h"
7 static int find_short_object_filename(int len, const char *name, unsigned char *sha1)
9 struct alternate_object_database *alt;
10 char hex[40];
11 int found = 0;
12 static struct alternate_object_database *fakeent;
14 if (!fakeent) {
15 const char *objdir = get_object_directory();
16 int objdir_len = strlen(objdir);
17 int entlen = objdir_len + 43;
18 fakeent = xmalloc(sizeof(*fakeent) + entlen);
19 memcpy(fakeent->base, objdir, objdir_len);
20 fakeent->name = fakeent->base + objdir_len + 1;
21 fakeent->name[-1] = '/';
23 fakeent->next = alt_odb_list;
25 sprintf(hex, "%.2s", name);
26 for (alt = fakeent; alt && found < 2; alt = alt->next) {
27 struct dirent *de;
28 DIR *dir;
29 sprintf(alt->name, "%.2s/", name);
30 dir = opendir(alt->base);
31 if (!dir)
32 continue;
33 while ((de = readdir(dir)) != NULL) {
34 if (strlen(de->d_name) != 38)
35 continue;
36 if (memcmp(de->d_name, name + 2, len - 2))
37 continue;
38 if (!found) {
39 memcpy(hex + 2, de->d_name, 38);
40 found++;
42 else if (memcmp(hex + 2, de->d_name, 38)) {
43 found = 2;
44 break;
47 closedir(dir);
49 if (found == 1)
50 return get_sha1_hex(hex, sha1) == 0;
51 return found;
54 static int match_sha(unsigned len, const unsigned char *a, const unsigned char *b)
56 do {
57 if (*a != *b)
58 return 0;
59 a++;
60 b++;
61 len -= 2;
62 } while (len > 1);
63 if (len)
64 if ((*a ^ *b) & 0xf0)
65 return 0;
66 return 1;
69 static int find_short_packed_object(int len, const unsigned char *match, unsigned char *sha1)
71 struct packed_git *p;
72 unsigned char found_sha1[20];
73 int found = 0;
75 prepare_packed_git();
76 for (p = packed_git; p && found < 2; p = p->next) {
77 unsigned num = num_packed_objects(p);
78 unsigned first = 0, last = num;
79 while (first < last) {
80 unsigned mid = (first + last) / 2;
81 unsigned char now[20];
82 int cmp;
84 nth_packed_object_sha1(p, mid, now);
85 cmp = memcmp(match, now, 20);
86 if (!cmp) {
87 first = mid;
88 break;
90 if (cmp > 0) {
91 first = mid+1;
92 continue;
94 last = mid;
96 if (first < num) {
97 unsigned char now[20], next[20];
98 nth_packed_object_sha1(p, first, now);
99 if (match_sha(len, match, now)) {
100 if (nth_packed_object_sha1(p, first+1, next) ||
101 !match_sha(len, match, next)) {
102 /* unique within this pack */
103 if (!found) {
104 memcpy(found_sha1, now, 20);
105 found++;
107 else if (memcmp(found_sha1, now, 20)) {
108 found = 2;
109 break;
112 else {
113 /* not even unique within this pack */
114 found = 2;
115 break;
120 if (found == 1)
121 memcpy(sha1, found_sha1, 20);
122 return found;
125 #define SHORT_NAME_NOT_FOUND (-1)
126 #define SHORT_NAME_AMBIGUOUS (-2)
128 static int find_unique_short_object(int len, char *canonical,
129 unsigned char *res, unsigned char *sha1)
131 int has_unpacked, has_packed;
132 unsigned char unpacked_sha1[20], packed_sha1[20];
134 has_unpacked = find_short_object_filename(len, canonical, unpacked_sha1);
135 has_packed = find_short_packed_object(len, res, packed_sha1);
136 if (!has_unpacked && !has_packed)
137 return SHORT_NAME_NOT_FOUND;
138 if (1 < has_unpacked || 1 < has_packed)
139 return SHORT_NAME_AMBIGUOUS;
140 if (has_unpacked != has_packed) {
141 memcpy(sha1, (has_packed ? packed_sha1 : unpacked_sha1), 20);
142 return 0;
144 /* Both have unique ones -- do they match? */
145 if (memcmp(packed_sha1, unpacked_sha1, 20))
146 return SHORT_NAME_AMBIGUOUS;
147 memcpy(sha1, packed_sha1, 20);
148 return 0;
151 static int get_short_sha1(const char *name, int len, unsigned char *sha1,
152 int quietly)
154 int i, status;
155 char canonical[40];
156 unsigned char res[20];
158 if (len < MINIMUM_ABBREV)
159 return -1;
160 memset(res, 0, 20);
161 memset(canonical, 'x', 40);
162 for (i = 0; i < len ;i++) {
163 unsigned char c = name[i];
164 unsigned char val;
165 if (c >= '0' && c <= '9')
166 val = c - '0';
167 else if (c >= 'a' && c <= 'f')
168 val = c - 'a' + 10;
169 else if (c >= 'A' && c <='F') {
170 val = c - 'A' + 10;
171 c -= 'A' - 'a';
173 else
174 return -1;
175 canonical[i] = c;
176 if (!(i & 1))
177 val <<= 4;
178 res[i >> 1] |= val;
181 status = find_unique_short_object(i, canonical, res, sha1);
182 if (!quietly && (status == SHORT_NAME_AMBIGUOUS))
183 return error("short SHA1 %.*s is ambiguous.", len, canonical);
184 return status;
187 const char *find_unique_abbrev(const unsigned char *sha1, int len)
189 int status, is_null;
190 static char hex[41];
192 is_null = !memcmp(sha1, null_sha1, 20);
193 memcpy(hex, sha1_to_hex(sha1), 40);
194 if (len == 40)
195 return hex;
196 while (len < 40) {
197 unsigned char sha1_ret[20];
198 status = get_short_sha1(hex, len, sha1_ret, 1);
199 if (!status ||
200 (is_null && status != SHORT_NAME_AMBIGUOUS)) {
201 hex[len] = 0;
202 return hex;
204 if (status != SHORT_NAME_AMBIGUOUS)
205 return NULL;
206 len++;
208 return NULL;
211 static int ambiguous_path(const char *path, int len)
213 int slash = 1;
214 int cnt;
216 for (cnt = 0; cnt < len; cnt++) {
217 switch (*path++) {
218 case '\0':
219 break;
220 case '/':
221 if (slash)
222 break;
223 slash = 1;
224 continue;
225 case '.':
226 continue;
227 default:
228 slash = 0;
229 continue;
231 break;
233 return slash;
236 static int get_sha1_basic(const char *str, int len, unsigned char *sha1)
238 static const char *prefix[] = {
240 "refs",
241 "refs/tags",
242 "refs/heads",
243 NULL
245 const char **p;
247 if (len == 40 && !get_sha1_hex(str, sha1))
248 return 0;
250 /* Accept only unambiguous ref paths. */
251 if (ambiguous_path(str, len))
252 return -1;
254 for (p = prefix; *p; p++) {
255 char *pathname = git_path("%s/%.*s", *p, len, str);
256 if (!read_ref(pathname, sha1))
257 return 0;
259 return -1;
262 static int get_sha1_1(const char *name, int len, unsigned char *sha1);
264 static int get_parent(const char *name, int len,
265 unsigned char *result, int idx)
267 unsigned char sha1[20];
268 int ret = get_sha1_1(name, len, sha1);
269 struct commit *commit;
270 struct commit_list *p;
272 if (ret)
273 return ret;
274 commit = lookup_commit_reference(sha1);
275 if (!commit)
276 return -1;
277 if (parse_commit(commit))
278 return -1;
279 if (!idx) {
280 memcpy(result, commit->object.sha1, 20);
281 return 0;
283 p = commit->parents;
284 while (p) {
285 if (!--idx) {
286 memcpy(result, p->item->object.sha1, 20);
287 return 0;
289 p = p->next;
291 return -1;
294 static int get_nth_ancestor(const char *name, int len,
295 unsigned char *result, int generation)
297 unsigned char sha1[20];
298 int ret = get_sha1_1(name, len, sha1);
299 if (ret)
300 return ret;
302 while (generation--) {
303 struct commit *commit = lookup_commit_reference(sha1);
305 if (!commit || parse_commit(commit) || !commit->parents)
306 return -1;
307 memcpy(sha1, commit->parents->item->object.sha1, 20);
309 memcpy(result, sha1, 20);
310 return 0;
313 static int peel_onion(const char *name, int len, unsigned char *sha1)
315 unsigned char outer[20];
316 const char *sp;
317 const char *type_string = NULL;
318 struct object *o;
321 * "ref^{type}" dereferences ref repeatedly until you cannot
322 * dereference anymore, or you get an object of given type,
323 * whichever comes first. "ref^{}" means just dereference
324 * tags until you get a non-tag. "ref^0" is a shorthand for
325 * "ref^{commit}". "commit^{tree}" could be used to find the
326 * top-level tree of the given commit.
328 if (len < 4 || name[len-1] != '}')
329 return -1;
331 for (sp = name + len - 1; name <= sp; sp--) {
332 int ch = *sp;
333 if (ch == '{' && name < sp && sp[-1] == '^')
334 break;
336 if (sp <= name)
337 return -1;
339 sp++; /* beginning of type name, or closing brace for empty */
340 if (!strncmp(commit_type, sp, 6) && sp[6] == '}')
341 type_string = commit_type;
342 else if (!strncmp(tree_type, sp, 4) && sp[4] == '}')
343 type_string = tree_type;
344 else if (!strncmp(blob_type, sp, 4) && sp[4] == '}')
345 type_string = blob_type;
346 else if (sp[0] == '}')
347 type_string = NULL;
348 else
349 return -1;
351 if (get_sha1_1(name, sp - name - 2, outer))
352 return -1;
354 o = parse_object(outer);
355 if (!o)
356 return -1;
357 if (!type_string) {
358 o = deref_tag(o, name, sp - name - 2);
359 if (!o || (!o->parsed && !parse_object(o->sha1)))
360 return -1;
361 memcpy(sha1, o->sha1, 20);
363 else {
364 /* At this point, the syntax look correct, so
365 * if we do not get the needed object, we should
366 * barf.
369 while (1) {
370 if (!o || (!o->parsed && !parse_object(o->sha1)))
371 return -1;
372 if (o->type == type_string) {
373 memcpy(sha1, o->sha1, 20);
374 return 0;
376 if (o->type == tag_type)
377 o = ((struct tag*) o)->tagged;
378 else if (o->type == commit_type)
379 o = &(((struct commit *) o)->tree->object);
380 else
381 return error("%.*s: expected %s type, but the object dereferences to %s type",
382 len, name, type_string,
383 o->type);
384 if (!o->parsed)
385 parse_object(o->sha1);
388 return 0;
391 static int get_sha1_1(const char *name, int len, unsigned char *sha1)
393 int ret, has_suffix;
394 const char *cp;
396 /* "name~3" is "name^^^",
397 * "name~" and "name~0" are name -- not "name^0"!
398 * "name^" is not "name^0"; it is "name^1".
400 has_suffix = 0;
401 for (cp = name + len - 1; name <= cp; cp--) {
402 int ch = *cp;
403 if ('0' <= ch && ch <= '9')
404 continue;
405 if (ch == '~' || ch == '^')
406 has_suffix = ch;
407 break;
410 if (has_suffix) {
411 int num = 0;
412 int len1 = cp - name;
413 cp++;
414 while (cp < name + len)
415 num = num * 10 + *cp++ - '0';
416 if (has_suffix == '^') {
417 if (!num && len1 == len - 1)
418 num = 1;
419 return get_parent(name, len1, sha1, num);
421 /* else if (has_suffix == '~') -- goes without saying */
422 return get_nth_ancestor(name, len1, sha1, num);
425 ret = peel_onion(name, len, sha1);
426 if (!ret)
427 return 0;
429 ret = get_sha1_basic(name, len, sha1);
430 if (!ret)
431 return 0;
432 return get_short_sha1(name, len, sha1, 0);
436 * This is like "get_sha1_basic()", except it allows "sha1 expressions",
437 * notably "xyz^" for "parent of xyz"
439 int get_sha1(const char *name, unsigned char *sha1)
441 prepare_alt_odb();
442 return get_sha1_1(name, strlen(name), sha1);