Make on-disk index representation separate from in-core one
[git/git.git] / cache-tree.c
CommitLineData
74986462
JH
1#include "cache.h"
2#include "tree.h"
3#include "cache-tree.h"
4
4903161f 5#ifndef DEBUG
74986462 6#define DEBUG 0
4903161f 7#endif
74986462
JH
8
9struct cache_tree *cache_tree(void)
10{
11 struct cache_tree *it = xcalloc(1, sizeof(struct cache_tree));
12 it->entry_count = -1;
13 return it;
14}
15
bad68ec9 16void cache_tree_free(struct cache_tree **it_p)
74986462
JH
17{
18 int i;
bad68ec9 19 struct cache_tree *it = *it_p;
74986462
JH
20
21 if (!it)
22 return;
23 for (i = 0; i < it->subtree_nr; i++)
61fa3097
JH
24 if (it->down[i])
25 cache_tree_free(&it->down[i]->cache_tree);
74986462
JH
26 free(it->down);
27 free(it);
bad68ec9 28 *it_p = NULL;
74986462
JH
29}
30
61fa3097
JH
31static int subtree_name_cmp(const char *one, int onelen,
32 const char *two, int twolen)
33{
34 if (onelen < twolen)
35 return -1;
36 if (twolen < onelen)
37 return 1;
38 return memcmp(one, two, onelen);
39}
40
41static int subtree_pos(struct cache_tree *it, const char *path, int pathlen)
42{
43 struct cache_tree_sub **down = it->down;
44 int lo, hi;
45 lo = 0;
46 hi = it->subtree_nr;
47 while (lo < hi) {
48 int mi = (lo + hi) / 2;
49 struct cache_tree_sub *mdl = down[mi];
50 int cmp = subtree_name_cmp(path, pathlen,
51 mdl->name, mdl->namelen);
52 if (!cmp)
53 return mi;
54 if (cmp < 0)
55 hi = mi;
56 else
57 lo = mi + 1;
58 }
59 return -lo-1;
60}
61
74986462
JH
62static struct cache_tree_sub *find_subtree(struct cache_tree *it,
63 const char *path,
64 int pathlen,
65 int create)
66{
74986462 67 struct cache_tree_sub *down;
61fa3097
JH
68 int pos = subtree_pos(it, path, pathlen);
69 if (0 <= pos)
70 return it->down[pos];
74986462
JH
71 if (!create)
72 return NULL;
61fa3097
JH
73
74 pos = -pos-1;
74986462
JH
75 if (it->subtree_alloc <= it->subtree_nr) {
76 it->subtree_alloc = alloc_nr(it->subtree_alloc);
77 it->down = xrealloc(it->down, it->subtree_alloc *
78 sizeof(*it->down));
79 }
61fa3097
JH
80 it->subtree_nr++;
81
74986462 82 down = xmalloc(sizeof(*down) + pathlen + 1);
61fa3097 83 down->cache_tree = NULL;
74986462
JH
84 down->namelen = pathlen;
85 memcpy(down->name, path, pathlen);
61fa3097
JH
86 down->name[pathlen] = 0;
87
88 if (pos < it->subtree_nr)
89 memmove(it->down + pos + 1,
90 it->down + pos,
91 sizeof(down) * (it->subtree_nr - pos - 1));
92 it->down[pos] = down;
74986462
JH
93 return down;
94}
95
7927a55d
JH
96struct cache_tree_sub *cache_tree_sub(struct cache_tree *it, const char *path)
97{
98 int pathlen = strlen(path);
99 return find_subtree(it, path, pathlen, 1);
100}
101
74986462
JH
102void cache_tree_invalidate_path(struct cache_tree *it, const char *path)
103{
104 /* a/b/c
105 * ==> invalidate self
106 * ==> find "a", have it invalidate "b/c"
107 * a
108 * ==> invalidate self
109 * ==> if "a" exists as a subtree, remove it.
110 */
111 const char *slash;
112 int namelen;
113 struct cache_tree_sub *down;
114
00703e6d
JH
115#if DEBUG
116 fprintf(stderr, "cache-tree invalidate <%s>\n", path);
117#endif
118
74986462
JH
119 if (!it)
120 return;
121 slash = strchr(path, '/');
122 it->entry_count = -1;
123 if (!slash) {
61fa3097 124 int pos;
74986462 125 namelen = strlen(path);
61fa3097
JH
126 pos = subtree_pos(it, path, namelen);
127 if (0 <= pos) {
128 cache_tree_free(&it->down[pos]->cache_tree);
129 free(it->down[pos]);
74986462
JH
130 /* 0 1 2 3 4 5
131 * ^ ^subtree_nr = 6
61fa3097 132 * pos
74986462 133 * move 4 and 5 up one place (2 entries)
61fa3097 134 * 2 = 6 - 3 - 1 = subtree_nr - pos - 1
74986462 135 */
61fa3097 136 memmove(it->down+pos, it->down+pos+1,
74986462 137 sizeof(struct cache_tree_sub *) *
61fa3097 138 (it->subtree_nr - pos - 1));
74986462
JH
139 it->subtree_nr--;
140 }
141 return;
142 }
143 namelen = slash - path;
144 down = find_subtree(it, path, namelen, 0);
145 if (down)
146 cache_tree_invalidate_path(down->cache_tree, slash + 1);
147}
148
149static int verify_cache(struct cache_entry **cache,
150 int entries)
151{
152 int i, funny;
153
154 /* Verify that the tree is merged */
155 funny = 0;
156 for (i = 0; i < entries; i++) {
157 struct cache_entry *ce = cache[i];
158 if (ce_stage(ce)) {
159 if (10 < ++funny) {
160 fprintf(stderr, "...\n");
161 break;
162 }
163 fprintf(stderr, "%s: unmerged (%s)\n",
164 ce->name, sha1_to_hex(ce->sha1));
165 }
166 }
167 if (funny)
168 return -1;
169
170 /* Also verify that the cache does not have path and path/file
171 * at the same time. At this point we know the cache has only
172 * stage 0 entries.
173 */
174 funny = 0;
175 for (i = 0; i < entries - 1; i++) {
176 /* path/file always comes after path because of the way
177 * the cache is sorted. Also path can appear only once,
178 * which means conflicting one would immediately follow.
179 */
180 const char *this_name = cache[i]->name;
181 const char *next_name = cache[i+1]->name;
182 int this_len = strlen(this_name);
183 if (this_len < strlen(next_name) &&
184 strncmp(this_name, next_name, this_len) == 0 &&
185 next_name[this_len] == '/') {
186 if (10 < ++funny) {
187 fprintf(stderr, "...\n");
188 break;
189 }
190 fprintf(stderr, "You have both %s and %s\n",
191 this_name, next_name);
192 }
193 }
194 if (funny)
195 return -1;
196 return 0;
197}
198
199static void discard_unused_subtrees(struct cache_tree *it)
200{
201 struct cache_tree_sub **down = it->down;
202 int nr = it->subtree_nr;
203 int dst, src;
204 for (dst = src = 0; src < nr; src++) {
205 struct cache_tree_sub *s = down[src];
206 if (s->used)
207 down[dst++] = s;
208 else {
bad68ec9 209 cache_tree_free(&s->cache_tree);
74986462
JH
210 free(s);
211 it->subtree_nr--;
212 }
213 }
214}
215
bad68ec9
JH
216int cache_tree_fully_valid(struct cache_tree *it)
217{
218 int i;
219 if (!it)
220 return 0;
221 if (it->entry_count < 0 || !has_sha1_file(it->sha1))
222 return 0;
223 for (i = 0; i < it->subtree_nr; i++) {
224 if (!cache_tree_fully_valid(it->down[i]->cache_tree))
225 return 0;
226 }
227 return 1;
228}
229
74986462
JH
230static int update_one(struct cache_tree *it,
231 struct cache_entry **cache,
232 int entries,
233 const char *base,
234 int baselen,
2956dd3b
JH
235 int missing_ok,
236 int dryrun)
74986462 237{
5242bcbb 238 struct strbuf buffer;
74986462
JH
239 int i;
240
dd0c34c4 241 if (0 <= it->entry_count && has_sha1_file(it->sha1))
74986462
JH
242 return it->entry_count;
243
244 /*
245 * We first scan for subtrees and update them; we start by
246 * marking existing subtrees -- the ones that are unmarked
247 * should not be in the result.
248 */
249 for (i = 0; i < it->subtree_nr; i++)
250 it->down[i]->used = 0;
251
252 /*
253 * Find the subtrees and update them.
254 */
255 for (i = 0; i < entries; i++) {
256 struct cache_entry *ce = cache[i];
257 struct cache_tree_sub *sub;
258 const char *path, *slash;
259 int pathlen, sublen, subcnt;
260
261 path = ce->name;
262 pathlen = ce_namelen(ce);
263 if (pathlen <= baselen || memcmp(base, path, baselen))
264 break; /* at the end of this level */
265
266 slash = strchr(path + baselen, '/');
267 if (!slash)
268 continue;
269 /*
270 * a/bbb/c (base = a/, slash = /c)
271 * ==>
272 * path+baselen = bbb/c, sublen = 3
273 */
274 sublen = slash - (path + baselen);
275 sub = find_subtree(it, path + baselen, sublen, 1);
276 if (!sub->cache_tree)
277 sub->cache_tree = cache_tree();
278 subcnt = update_one(sub->cache_tree,
279 cache + i, entries - i,
280 path,
281 baselen + sublen + 1,
2956dd3b
JH
282 missing_ok,
283 dryrun);
3d12d0cf
JS
284 if (subcnt < 0)
285 return subcnt;
74986462
JH
286 i += subcnt - 1;
287 sub->used = 1;
288 }
289
290 discard_unused_subtrees(it);
291
292 /*
293 * Then write out the tree object for this level.
294 */
f1696ee3 295 strbuf_init(&buffer, 8192);
74986462
JH
296
297 for (i = 0; i < entries; i++) {
298 struct cache_entry *ce = cache[i];
299 struct cache_tree_sub *sub;
300 const char *path, *slash;
301 int pathlen, entlen;
302 const unsigned char *sha1;
303 unsigned mode;
304
305 path = ce->name;
306 pathlen = ce_namelen(ce);
307 if (pathlen <= baselen || memcmp(base, path, baselen))
308 break; /* at the end of this level */
309
310 slash = strchr(path + baselen, '/');
311 if (slash) {
312 entlen = slash - (path + baselen);
313 sub = find_subtree(it, path + baselen, entlen, 0);
314 if (!sub)
315 die("cache-tree.c: '%.*s' in '%s' not found",
316 entlen, path + baselen, path);
317 i += sub->cache_tree->entry_count - 1;
318 sha1 = sub->cache_tree->sha1;
319 mode = S_IFDIR;
320 }
321 else {
322 sha1 = ce->sha1;
7a51ed66 323 mode = ce->ce_mode;
74986462
JH
324 entlen = pathlen - baselen;
325 }
302b9282 326 if (mode != S_IFGITLINK && !missing_ok && !has_sha1_file(sha1))
74986462
JH
327 return error("invalid object %s", sha1_to_hex(sha1));
328
7a51ed66 329 if (ce->ce_flags & CE_REMOVE)
74986462
JH
330 continue; /* entry being removed */
331
5242bcbb
PH
332 strbuf_grow(&buffer, entlen + 100);
333 strbuf_addf(&buffer, "%o %.*s%c", mode, entlen, path + baselen, '\0');
334 strbuf_add(&buffer, sha1, 20);
74986462
JH
335
336#if DEBUG
00703e6d 337 fprintf(stderr, "cache-tree update-one %o %.*s\n",
74986462
JH
338 mode, entlen, path + baselen);
339#endif
340 }
341
abdc3fc8 342 if (dryrun)
5242bcbb 343 hash_sha1_file(buffer.buf, buffer.len, tree_type, it->sha1);
2956dd3b 344 else
5242bcbb
PH
345 write_sha1_file(buffer.buf, buffer.len, tree_type, it->sha1);
346 strbuf_release(&buffer);
74986462
JH
347 it->entry_count = i;
348#if DEBUG
00703e6d 349 fprintf(stderr, "cache-tree update-one (%d ent, %d subtree) %s\n",
74986462
JH
350 it->entry_count, it->subtree_nr,
351 sha1_to_hex(it->sha1));
352#endif
353 return i;
354}
355
356int cache_tree_update(struct cache_tree *it,
357 struct cache_entry **cache,
358 int entries,
2956dd3b
JH
359 int missing_ok,
360 int dryrun)
74986462
JH
361{
362 int i;
363 i = verify_cache(cache, entries);
364 if (i)
365 return i;
2956dd3b 366 i = update_one(it, cache, entries, "", 0, missing_ok, dryrun);
74986462
JH
367 if (i < 0)
368 return i;
369 return 0;
370}
371
1dffb8fa
PH
372static void write_one(struct strbuf *buffer, struct cache_tree *it,
373 const char *path, int pathlen)
74986462
JH
374{
375 int i;
376
377 /* One "cache-tree" entry consists of the following:
378 * path (NUL terminated)
379 * entry_count, subtree_nr ("%d %d\n")
380 * tree-sha1 (missing if invalid)
381 * subtree_nr "cache-tree" entries for subtrees.
382 */
5242bcbb
PH
383 strbuf_grow(buffer, pathlen + 100);
384 strbuf_add(buffer, path, pathlen);
385 strbuf_addf(buffer, "%c%d %d\n", 0, it->entry_count, it->subtree_nr);
74986462
JH
386
387#if DEBUG
388 if (0 <= it->entry_count)
389 fprintf(stderr, "cache-tree <%.*s> (%d ent, %d subtree) %s\n",
390 pathlen, path, it->entry_count, it->subtree_nr,
391 sha1_to_hex(it->sha1));
392 else
393 fprintf(stderr, "cache-tree <%.*s> (%d subtree) invalid\n",
394 pathlen, path, it->subtree_nr);
395#endif
396
397 if (0 <= it->entry_count) {
5242bcbb 398 strbuf_add(buffer, it->sha1, 20);
74986462
JH
399 }
400 for (i = 0; i < it->subtree_nr; i++) {
401 struct cache_tree_sub *down = it->down[i];
61fa3097
JH
402 if (i) {
403 struct cache_tree_sub *prev = it->down[i-1];
404 if (subtree_name_cmp(down->name, down->namelen,
405 prev->name, prev->namelen) <= 0)
406 die("fatal - unsorted cache subtree");
407 }
1dffb8fa 408 write_one(buffer, down->cache_tree, down->name, down->namelen);
74986462 409 }
74986462
JH
410}
411
1dffb8fa 412void cache_tree_write(struct strbuf *sb, struct cache_tree *root)
74986462 413{
1dffb8fa 414 write_one(sb, root, "", 0);
74986462
JH
415}
416
417static struct cache_tree *read_one(const char **buffer, unsigned long *size_p)
418{
419 const char *buf = *buffer;
420 unsigned long size = *size_p;
0111ea38
JS
421 const char *cp;
422 char *ep;
74986462
JH
423 struct cache_tree *it;
424 int i, subtree_nr;
425
426 it = NULL;
427 /* skip name, but make sure name exists */
428 while (size && *buf) {
429 size--;
430 buf++;
431 }
432 if (!size)
433 goto free_return;
434 buf++; size--;
435 it = cache_tree();
0111ea38
JS
436
437 cp = buf;
438 it->entry_count = strtol(cp, &ep, 10);
439 if (cp == ep)
440 goto free_return;
441 cp = ep;
442 subtree_nr = strtol(cp, &ep, 10);
443 if (cp == ep)
74986462
JH
444 goto free_return;
445 while (size && *buf && *buf != '\n') {
446 size--;
447 buf++;
448 }
449 if (!size)
450 goto free_return;
451 buf++; size--;
452 if (0 <= it->entry_count) {
453 if (size < 20)
454 goto free_return;
63daae47 455 hashcpy(it->sha1, (const unsigned char*)buf);
74986462
JH
456 buf += 20;
457 size -= 20;
458 }
459
460#if DEBUG
461 if (0 <= it->entry_count)
462 fprintf(stderr, "cache-tree <%s> (%d ent, %d subtree) %s\n",
463 *buffer, it->entry_count, subtree_nr,
464 sha1_to_hex(it->sha1));
465 else
466 fprintf(stderr, "cache-tree <%s> (%d subtrees) invalid\n",
467 *buffer, subtree_nr);
468#endif
469
470 /*
471 * Just a heuristic -- we do not add directories that often but
472 * we do not want to have to extend it immediately when we do,
473 * hence +2.
474 */
475 it->subtree_alloc = subtree_nr + 2;
476 it->down = xcalloc(it->subtree_alloc, sizeof(struct cache_tree_sub *));
477 for (i = 0; i < subtree_nr; i++) {
478 /* read each subtree */
479 struct cache_tree *sub;
61fa3097 480 struct cache_tree_sub *subtree;
74986462 481 const char *name = buf;
7927a55d 482
74986462
JH
483 sub = read_one(&buf, &size);
484 if (!sub)
485 goto free_return;
7927a55d 486 subtree = cache_tree_sub(it, name);
61fa3097 487 subtree->cache_tree = sub;
74986462
JH
488 }
489 if (subtree_nr != it->subtree_nr)
490 die("cache-tree: internal error");
491 *buffer = buf;
492 *size_p = size;
493 return it;
494
495 free_return:
bad68ec9 496 cache_tree_free(&it);
74986462
JH
497 return NULL;
498}
499
bad68ec9 500struct cache_tree *cache_tree_read(const char *buffer, unsigned long size)
74986462 501{
bad68ec9 502 if (buffer[0])
74986462 503 return NULL; /* not the whole tree */
74986462
JH
504 return read_one(&buffer, &size);
505}
6bd20358
JH
506
507struct cache_tree *cache_tree_find(struct cache_tree *it, const char *path)
508{
509 while (*path) {
510 const char *slash;
511 struct cache_tree_sub *sub;
512
513 slash = strchr(path, '/');
514 if (!slash)
515 slash = path + strlen(path);
516 /* between path and slash is the name of the
517 * subtree to look for.
518 */
519 sub = find_subtree(it, path, slash - path, 0);
520 if (!sub)
521 return NULL;
522 it = sub->cache_tree;
523 if (slash)
524 while (*slash && *slash == '/')
525 slash++;
526 if (!slash || !*slash)
527 return it; /* prefix ended with slashes */
528 path = slash;
529 }
530 return it;
531}