git-remote-mediawiki: fix encoding issue for UTF-8 media files
[alt-git.git] / object.c
blob57a0890a87b66ea493337409606e57fd44d29f90
1 #include "cache.h"
2 #include "object.h"
3 #include "blob.h"
4 #include "tree.h"
5 #include "commit.h"
6 #include "tag.h"
8 static struct object **obj_hash;
9 static int nr_objs, obj_hash_size;
11 unsigned int get_max_object_index(void)
13 return obj_hash_size;
16 struct object *get_indexed_object(unsigned int idx)
18 return obj_hash[idx];
21 static const char *object_type_strings[] = {
22 NULL, /* OBJ_NONE = 0 */
23 "commit", /* OBJ_COMMIT = 1 */
24 "tree", /* OBJ_TREE = 2 */
25 "blob", /* OBJ_BLOB = 3 */
26 "tag", /* OBJ_TAG = 4 */
29 const char *typename(unsigned int type)
31 if (type >= ARRAY_SIZE(object_type_strings))
32 return NULL;
33 return object_type_strings[type];
36 int type_from_string(const char *str)
38 int i;
40 for (i = 1; i < ARRAY_SIZE(object_type_strings); i++)
41 if (!strcmp(str, object_type_strings[i]))
42 return i;
43 die("invalid object type \"%s\"", str);
47 * Return a numerical hash value between 0 and n-1 for the object with
48 * the specified sha1. n must be a power of 2. Please note that the
49 * return value is *not* consistent across computer architectures.
51 static unsigned int hash_obj(const unsigned char *sha1, unsigned int n)
53 unsigned int hash;
56 * Since the sha1 is essentially random, we just take the
57 * required number of bits directly from the first
58 * sizeof(unsigned int) bytes of sha1. First we have to copy
59 * the bytes into a properly aligned integer. If we cared
60 * about getting consistent results across architectures, we
61 * would have to call ntohl() here, too.
63 memcpy(&hash, sha1, sizeof(unsigned int));
64 return hash & (n - 1);
68 * Insert obj into the hash table hash, which has length size (which
69 * must be a power of 2). On collisions, simply overflow to the next
70 * empty bucket.
72 static void insert_obj_hash(struct object *obj, struct object **hash, unsigned int size)
74 unsigned int j = hash_obj(obj->sha1, size);
76 while (hash[j]) {
77 j++;
78 if (j >= size)
79 j = 0;
81 hash[j] = obj;
85 * Look up the record for the given sha1 in the hash map stored in
86 * obj_hash. Return NULL if it was not found.
88 struct object *lookup_object(const unsigned char *sha1)
90 unsigned int i, first;
91 struct object *obj;
93 if (!obj_hash)
94 return NULL;
96 first = i = hash_obj(sha1, obj_hash_size);
97 while ((obj = obj_hash[i]) != NULL) {
98 if (!hashcmp(sha1, obj->sha1))
99 break;
100 i++;
101 if (i == obj_hash_size)
102 i = 0;
104 if (obj && i != first) {
106 * Move object to where we started to look for it so
107 * that we do not need to walk the hash table the next
108 * time we look for it.
110 struct object *tmp = obj_hash[i];
111 obj_hash[i] = obj_hash[first];
112 obj_hash[first] = tmp;
114 return obj;
118 * Increase the size of the hash map stored in obj_hash to the next
119 * power of 2 (but at least 32). Copy the existing values to the new
120 * hash map.
122 static void grow_object_hash(void)
124 int i;
126 * Note that this size must always be power-of-2 to match hash_obj
127 * above.
129 int new_hash_size = obj_hash_size < 32 ? 32 : 2 * obj_hash_size;
130 struct object **new_hash;
132 new_hash = xcalloc(new_hash_size, sizeof(struct object *));
133 for (i = 0; i < obj_hash_size; i++) {
134 struct object *obj = obj_hash[i];
135 if (!obj)
136 continue;
137 insert_obj_hash(obj, new_hash, new_hash_size);
139 free(obj_hash);
140 obj_hash = new_hash;
141 obj_hash_size = new_hash_size;
144 void *create_object(const unsigned char *sha1, int type, void *o)
146 struct object *obj = o;
148 obj->parsed = 0;
149 obj->used = 0;
150 obj->type = type;
151 obj->flags = 0;
152 hashcpy(obj->sha1, sha1);
154 if (obj_hash_size - 1 <= nr_objs * 2)
155 grow_object_hash();
157 insert_obj_hash(obj, obj_hash, obj_hash_size);
158 nr_objs++;
159 return obj;
162 struct object *lookup_unknown_object(const unsigned char *sha1)
164 struct object *obj = lookup_object(sha1);
165 if (!obj)
166 obj = create_object(sha1, OBJ_NONE, alloc_object_node());
167 return obj;
170 struct object *parse_object_buffer(const unsigned char *sha1, enum object_type type, unsigned long size, void *buffer, int *eaten_p)
172 struct object *obj;
173 *eaten_p = 0;
175 obj = NULL;
176 if (type == OBJ_BLOB) {
177 struct blob *blob = lookup_blob(sha1);
178 if (blob) {
179 if (parse_blob_buffer(blob, buffer, size))
180 return NULL;
181 obj = &blob->object;
183 } else if (type == OBJ_TREE) {
184 struct tree *tree = lookup_tree(sha1);
185 if (tree) {
186 obj = &tree->object;
187 if (!tree->buffer)
188 tree->object.parsed = 0;
189 if (!tree->object.parsed) {
190 if (parse_tree_buffer(tree, buffer, size))
191 return NULL;
192 *eaten_p = 1;
195 } else if (type == OBJ_COMMIT) {
196 struct commit *commit = lookup_commit(sha1);
197 if (commit) {
198 if (parse_commit_buffer(commit, buffer, size))
199 return NULL;
200 if (!commit->buffer) {
201 commit->buffer = buffer;
202 *eaten_p = 1;
204 obj = &commit->object;
206 } else if (type == OBJ_TAG) {
207 struct tag *tag = lookup_tag(sha1);
208 if (tag) {
209 if (parse_tag_buffer(tag, buffer, size))
210 return NULL;
211 obj = &tag->object;
213 } else {
214 warning("object %s has unknown type id %d", sha1_to_hex(sha1), type);
215 obj = NULL;
217 if (obj && obj->type == OBJ_NONE)
218 obj->type = type;
219 return obj;
222 struct object *parse_object_or_die(const unsigned char *sha1,
223 const char *name)
225 struct object *o = parse_object(sha1);
226 if (o)
227 return o;
229 die(_("unable to parse object: %s"), name ? name : sha1_to_hex(sha1));
232 struct object *parse_object(const unsigned char *sha1)
234 unsigned long size;
235 enum object_type type;
236 int eaten;
237 const unsigned char *repl = lookup_replace_object(sha1);
238 void *buffer;
239 struct object *obj;
241 obj = lookup_object(sha1);
242 if (obj && obj->parsed)
243 return obj;
245 if ((obj && obj->type == OBJ_BLOB) ||
246 (!obj && has_sha1_file(sha1) &&
247 sha1_object_info(sha1, NULL) == OBJ_BLOB)) {
248 if (check_sha1_signature(repl, NULL, 0, NULL) < 0) {
249 error("sha1 mismatch %s", sha1_to_hex(repl));
250 return NULL;
252 parse_blob_buffer(lookup_blob(sha1), NULL, 0);
253 return lookup_object(sha1);
256 buffer = read_sha1_file(sha1, &type, &size);
257 if (buffer) {
258 if (check_sha1_signature(repl, buffer, size, typename(type)) < 0) {
259 free(buffer);
260 error("sha1 mismatch %s", sha1_to_hex(repl));
261 return NULL;
264 obj = parse_object_buffer(sha1, type, size, buffer, &eaten);
265 if (!eaten)
266 free(buffer);
267 return obj;
269 return NULL;
272 struct object_list *object_list_insert(struct object *item,
273 struct object_list **list_p)
275 struct object_list *new_list = xmalloc(sizeof(struct object_list));
276 new_list->item = item;
277 new_list->next = *list_p;
278 *list_p = new_list;
279 return new_list;
282 int object_list_contains(struct object_list *list, struct object *obj)
284 while (list) {
285 if (list->item == obj)
286 return 1;
287 list = list->next;
289 return 0;
293 * A zero-length string to which object_array_entry::name can be
294 * initialized without requiring a malloc/free.
296 static char object_array_slopbuf[1];
298 static void add_object_array_with_mode_context(struct object *obj, const char *name,
299 struct object_array *array,
300 unsigned mode,
301 struct object_context *context)
303 unsigned nr = array->nr;
304 unsigned alloc = array->alloc;
305 struct object_array_entry *objects = array->objects;
306 struct object_array_entry *entry;
308 if (nr >= alloc) {
309 alloc = (alloc + 32) * 2;
310 objects = xrealloc(objects, alloc * sizeof(*objects));
311 array->alloc = alloc;
312 array->objects = objects;
314 entry = &objects[nr];
315 entry->item = obj;
316 if (!name)
317 entry->name = NULL;
318 else if (!*name)
319 /* Use our own empty string instead of allocating one: */
320 entry->name = object_array_slopbuf;
321 else
322 entry->name = xstrdup(name);
323 entry->mode = mode;
324 entry->context = context;
325 array->nr = ++nr;
328 void add_object_array(struct object *obj, const char *name, struct object_array *array)
330 add_object_array_with_mode(obj, name, array, S_IFINVALID);
333 void add_object_array_with_mode(struct object *obj, const char *name, struct object_array *array, unsigned mode)
335 add_object_array_with_mode_context(obj, name, array, mode, NULL);
338 void add_object_array_with_context(struct object *obj, const char *name, struct object_array *array, struct object_context *context)
340 if (context)
341 add_object_array_with_mode_context(obj, name, array, context->mode, context);
342 else
343 add_object_array_with_mode_context(obj, name, array, S_IFINVALID, context);
346 void object_array_filter(struct object_array *array,
347 object_array_each_func_t want, void *cb_data)
349 unsigned nr = array->nr, src, dst;
350 struct object_array_entry *objects = array->objects;
352 for (src = dst = 0; src < nr; src++) {
353 if (want(&objects[src], cb_data)) {
354 if (src != dst)
355 objects[dst] = objects[src];
356 dst++;
357 } else {
358 if (objects[src].name != object_array_slopbuf)
359 free(objects[src].name);
362 array->nr = dst;
366 * Return true iff array already contains an entry with name.
368 static int contains_name(struct object_array *array, const char *name)
370 unsigned nr = array->nr, i;
371 struct object_array_entry *object = array->objects;
373 for (i = 0; i < nr; i++, object++)
374 if (!strcmp(object->name, name))
375 return 1;
376 return 0;
379 void object_array_remove_duplicates(struct object_array *array)
381 unsigned nr = array->nr, src;
382 struct object_array_entry *objects = array->objects;
384 array->nr = 0;
385 for (src = 0; src < nr; src++) {
386 if (!contains_name(array, objects[src].name)) {
387 if (src != array->nr)
388 objects[array->nr] = objects[src];
389 array->nr++;
390 } else {
391 if (objects[src].name != object_array_slopbuf)
392 free(objects[src].name);
397 void clear_object_flags(unsigned flags)
399 int i;
401 for (i=0; i < obj_hash_size; i++) {
402 struct object *obj = obj_hash[i];
403 if (obj)
404 obj->flags &= ~flags;