Merge branch 'maint'
[git/git.git] / fast-import.c
CommitLineData
463acbe1
SP
1/*
2Format of STDIN stream:
3
4 stream ::= cmd*;
5
6 cmd ::= new_blob
c44cdc7e 7 | new_commit
463acbe1 8 | new_tag
5fced8dc 9 | reset_branch
8c1f22da 10 | checkpoint
463acbe1
SP
11 ;
12
c44cdc7e
SP
13 new_blob ::= 'blob' lf
14 mark?
15 file_content;
16 file_content ::= data;
463acbe1 17
c44cdc7e 18 new_commit ::= 'commit' sp ref_str lf
00e2b884 19 mark?
63e0c8b3
SP
20 ('author' sp name '<' email '>' when lf)?
21 'committer' sp name '<' email '>' when lf
00e2b884 22 commit_msg
02f3389d 23 ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
62b6f483 24 ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
c44cdc7e
SP
25 file_change*
26 lf;
27 commit_msg ::= data;
463acbe1 28
825769a8
SP
29 file_change ::= file_clr | file_del | file_obm | file_inm;
30 file_clr ::= 'deleteall' lf;
b715cfbb
SP
31 file_del ::= 'D' sp path_str lf;
32 file_obm ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf;
33 file_inm ::= 'M' sp mode sp 'inline' sp path_str lf
34 data;
c44cdc7e
SP
35
36 new_tag ::= 'tag' sp tag_str lf
37 'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
63e0c8b3 38 'tagger' sp name '<' email '>' when lf
c44cdc7e
SP
39 tag_msg;
40 tag_msg ::= data;
41
9938ffc5
SP
42 reset_branch ::= 'reset' sp ref_str lf
43 ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
44 lf;
5fced8dc 45
7bfe6e26
SP
46 checkpoint ::= 'checkpoint' lf
47 lf;
48
c44cdc7e
SP
49 # note: the first idnum in a stream should be 1 and subsequent
50 # idnums should not have gaps between values as this will cause
51 # the stream parser to reserve space for the gapped values. An
52 # idnum can be updated in the future to a new object by issuing
53 # a new mark directive with the old idnum.
54 #
55 mark ::= 'mark' sp idnum lf;
3b4dce02
SP
56 data ::= (delimited_data | exact_data)
57 lf;
58
59 # note: delim may be any string but must not contain lf.
60 # data_line may contain any data but must not be exactly
61 # delim.
62 delimited_data ::= 'data' sp '<<' delim lf
63 (data_line lf)*
64 delim lf;
c44cdc7e
SP
65
66 # note: declen indicates the length of binary_data in bytes.
3b4dce02 67 # declen does not include the lf preceeding the binary data.
c44cdc7e 68 #
3b4dce02
SP
69 exact_data ::= 'data' sp declen lf
70 binary_data;
c44cdc7e
SP
71
72 # note: quoted strings are C-style quoting supporting \c for
73 # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
74 # is the signed byte value in octal. Note that the only
75 # characters which must actually be escaped to protect the
76 # stream formatting is: \, " and LF. Otherwise these values
77 # are UTF8.
78 #
6c3aac1c
SP
79 ref_str ::= ref;
80 sha1exp_str ::= sha1exp;
81 tag_str ::= tag;
c44cdc7e 82 path_str ::= path | '"' quoted(path) '"' ;
b715cfbb
SP
83 mode ::= '100644' | '644'
84 | '100755' | '755'
9981b6d9 85 | '120000'
b715cfbb 86 ;
c44cdc7e
SP
87
88 declen ::= # unsigned 32 bit value, ascii base10 notation;
2104838b 89 bigint ::= # unsigned integer value, ascii base10 notation;
463acbe1 90 binary_data ::= # file content, not interpreted;
c44cdc7e 91
63e0c8b3
SP
92 when ::= raw_when | rfc2822_when;
93 raw_when ::= ts sp tz;
94 rfc2822_when ::= # Valid RFC 2822 date and time;
95
463acbe1
SP
96 sp ::= # ASCII space character;
97 lf ::= # ASCII newline (LF) character;
c44cdc7e
SP
98
99 # note: a colon (':') must precede the numerical value assigned to
100 # an idnum. This is to distinguish it from a ref or tag name as
101 # GIT does not permit ':' in ref or tag strings.
102 #
2104838b 103 idnum ::= ':' bigint;
c44cdc7e
SP
104 path ::= # GIT style file path, e.g. "a/b/c";
105 ref ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
106 tag ::= # GIT tag name, e.g. "FIREFOX_1_5";
463acbe1
SP
107 sha1exp ::= # Any valid GIT SHA1 expression;
108 hexsha1 ::= # SHA1 in hexadecimal format;
c44cdc7e
SP
109
110 # note: name and email are UTF8 strings, however name must not
111 # contain '<' or lf and email must not contain any of the
112 # following: '<', '>', lf.
113 #
114 name ::= # valid GIT author/committer name;
463acbe1 115 email ::= # valid GIT author/committer email;
c44cdc7e
SP
116 ts ::= # time since the epoch in seconds, ascii base10 notation;
117 tz ::= # GIT style timezone;
463acbe1
SP
118*/
119
db5e523f
SP
120#include "builtin.h"
121#include "cache.h"
122#include "object.h"
123#include "blob.h"
463acbe1 124#include "tree.h"
7073e69e 125#include "commit.h"
db5e523f
SP
126#include "delta.h"
127#include "pack.h"
463acbe1 128#include "refs.h"
db5e523f 129#include "csum-file.h"
c44cdc7e
SP
130#include "strbuf.h"
131#include "quote.h"
db5e523f 132
69e74e74
SP
133#define PACK_ID_BITS 16
134#define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
135
27d6d290
SP
136struct object_entry
137{
138 struct object_entry *next;
10831c55 139 uint32_t offset;
03842d8e 140 unsigned type : TYPE_BITS;
69e74e74 141 unsigned pack_id : PACK_ID_BITS;
27d6d290
SP
142 unsigned char sha1[20];
143};
144
463acbe1 145struct object_entry_pool
27d6d290 146{
463acbe1 147 struct object_entry_pool *next_pool;
27d6d290
SP
148 struct object_entry *next_free;
149 struct object_entry *end;
ac47a738 150 struct object_entry entries[FLEX_ARRAY]; /* more */
27d6d290
SP
151};
152
d8397168
SP
153struct mark_set
154{
d8397168
SP
155 union {
156 struct object_entry *marked[1024];
157 struct mark_set *sets[1024];
158 } data;
6f64f6d9 159 unsigned int shift;
d8397168
SP
160};
161
ac47a738
SP
162struct last_object
163{
164 void *data;
4cabf858 165 unsigned long len;
10831c55 166 uint32_t offset;
6bb5b329 167 unsigned int depth;
d489bc14 168 unsigned no_free:1;
ac47a738
SP
169};
170
463acbe1
SP
171struct mem_pool
172{
173 struct mem_pool *next_pool;
174 char *next_free;
175 char *end;
176 char space[FLEX_ARRAY]; /* more */
177};
178
179struct atom_str
180{
181 struct atom_str *next_atom;
10831c55 182 unsigned short str_len;
463acbe1
SP
183 char str_dat[FLEX_ARRAY]; /* more */
184};
185
186struct tree_content;
6bb5b329
SP
187struct tree_entry
188{
463acbe1
SP
189 struct tree_content *tree;
190 struct atom_str* name;
4cabf858
SP
191 struct tree_entry_ms
192 {
10831c55 193 uint16_t mode;
4cabf858
SP
194 unsigned char sha1[20];
195 } versions[2];
6bb5b329
SP
196};
197
463acbe1 198struct tree_content
6bb5b329 199{
463acbe1
SP
200 unsigned int entry_capacity; /* must match avail_tree_content */
201 unsigned int entry_count;
4cabf858 202 unsigned int delta_depth;
463acbe1
SP
203 struct tree_entry *entries[FLEX_ARRAY]; /* more */
204};
205
206struct avail_tree_content
207{
208 unsigned int entry_capacity; /* must match tree_content */
209 struct avail_tree_content *next_avail;
6bb5b329
SP
210};
211
212struct branch
213{
463acbe1
SP
214 struct branch *table_next_branch;
215 struct branch *active_next_branch;
6bb5b329 216 const char *name;
463acbe1 217 struct tree_entry branch_tree;
69e74e74 218 uintmax_t last_commit;
734c91f9
SP
219 unsigned active : 1;
220 unsigned pack_id : PACK_ID_BITS;
463acbe1 221 unsigned char sha1[20];
6bb5b329
SP
222};
223
72303d44
SP
224struct tag
225{
226 struct tag *next_tag;
227 const char *name;
2369ed79 228 unsigned int pack_id;
72303d44
SP
229 unsigned char sha1[20];
230};
231
e2eb469d
SP
232struct dbuf
233{
234 void *buffer;
235 size_t capacity;
236};
237
62b6f483
SP
238struct hash_list
239{
240 struct hash_list *next;
241 unsigned char sha1[20];
242};
463acbe1 243
63e0c8b3
SP
244typedef enum {
245 WHENSPEC_RAW = 1,
246 WHENSPEC_RFC2822,
247 WHENSPEC_NOW,
248} whenspec_type;
249
0ea9f045 250/* Configured limits on output */
d5c57b28 251static unsigned long max_depth = 10;
6777a59f 252static off_t max_packsize = (1LL << 32) - 1;
7073e69e 253static int force_update;
0ea9f045
SP
254
255/* Stats and misc. counters */
256static uintmax_t alloc_count;
0ea9f045
SP
257static uintmax_t marks_set_count;
258static uintmax_t object_count_by_type[1 << TYPE_BITS];
259static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
260static uintmax_t delta_count_by_type[1 << TYPE_BITS];
a7ddc487 261static unsigned long object_count;
6bb5b329 262static unsigned long branch_count;
d6c7eb2c 263static unsigned long branch_load_count;
7073e69e 264static int failure;
bdf1c06d 265static FILE *pack_edges;
ac47a738 266
463acbe1
SP
267/* Memory pools */
268static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
269static size_t total_allocd;
270static struct mem_pool *mem_pool;
271
c44cdc7e 272/* Atom management */
463acbe1
SP
273static unsigned int atom_table_sz = 4451;
274static unsigned int atom_cnt;
275static struct atom_str **atom_table;
276
277/* The .pack file being generated */
7bfe6e26 278static unsigned int pack_id;
d489bc14 279static struct packed_git *pack_data;
7bfe6e26 280static struct packed_git **all_packs;
41e5257f 281static unsigned long pack_size;
ac47a738
SP
282
283/* Table of objects we've written. */
4cabf858 284static unsigned int object_entry_alloc = 5000;
463acbe1
SP
285static struct object_entry_pool *blocks;
286static struct object_entry *object_table[1 << 16];
d8397168 287static struct mark_set *marks;
a6a1a831 288static const char* mark_file;
ac47a738
SP
289
290/* Our last blob */
463acbe1
SP
291static struct last_object last_blob;
292
293/* Tree management */
294static unsigned int tree_entry_alloc = 1000;
295static void *avail_tree_entry;
296static unsigned int avail_tree_table_sz = 100;
297static struct avail_tree_content **avail_tree_table;
e2eb469d
SP
298static struct dbuf old_tree;
299static struct dbuf new_tree;
8bcce301 300
6bb5b329 301/* Branch data */
d5c57b28
SP
302static unsigned long max_active_branches = 5;
303static unsigned long cur_active_branches;
304static unsigned long branch_table_sz = 1039;
463acbe1
SP
305static struct branch **branch_table;
306static struct branch *active_branches;
307
72303d44
SP
308/* Tag data */
309static struct tag *first_tag;
310static struct tag *last_tag;
311
c44cdc7e 312/* Input stream parsing */
63e0c8b3 313static whenspec_type whenspec = WHENSPEC_RAW;
c44cdc7e 314static struct strbuf command_buf;
0ea9f045 315static uintmax_t next_mark;
243f801d 316static struct dbuf new_data;
c44cdc7e 317
6bb5b329 318
03842d8e 319static void alloc_objects(unsigned int cnt)
8bcce301 320{
463acbe1 321 struct object_entry_pool *b;
27d6d290 322
463acbe1 323 b = xmalloc(sizeof(struct object_entry_pool)
27d6d290 324 + cnt * sizeof(struct object_entry));
463acbe1 325 b->next_pool = blocks;
27d6d290
SP
326 b->next_free = b->entries;
327 b->end = b->entries + cnt;
328 blocks = b;
329 alloc_count += cnt;
330}
8bcce301 331
e5b1444b 332static struct object_entry *new_object(unsigned char *sha1)
8bcce301 333{
27d6d290 334 struct object_entry *e;
8bcce301 335
27d6d290 336 if (blocks->next_free == blocks->end)
463acbe1 337 alloc_objects(object_entry_alloc);
8bcce301 338
27d6d290 339 e = blocks->next_free++;
445b8599 340 hashcpy(e->sha1, sha1);
27d6d290 341 return e;
8bcce301
SP
342}
343
e5b1444b 344static struct object_entry *find_object(unsigned char *sha1)
463acbe1
SP
345{
346 unsigned int h = sha1[0] << 8 | sha1[1];
347 struct object_entry *e;
348 for (e = object_table[h]; e; e = e->next)
445b8599 349 if (!hashcmp(sha1, e->sha1))
463acbe1
SP
350 return e;
351 return NULL;
352}
353
e5b1444b 354static struct object_entry *insert_object(unsigned char *sha1)
8bcce301
SP
355{
356 unsigned int h = sha1[0] << 8 | sha1[1];
27d6d290 357 struct object_entry *e = object_table[h];
463acbe1 358 struct object_entry *p = NULL;
8bcce301
SP
359
360 while (e) {
445b8599 361 if (!hashcmp(sha1, e->sha1))
8bcce301
SP
362 return e;
363 p = e;
364 e = e->next;
365 }
366
367 e = new_object(sha1);
463acbe1 368 e->next = NULL;
8bcce301
SP
369 e->offset = 0;
370 if (p)
371 p->next = e;
372 else
27d6d290 373 object_table[h] = e;
8bcce301
SP
374 return e;
375}
db5e523f 376
463acbe1
SP
377static unsigned int hc_str(const char *s, size_t len)
378{
379 unsigned int r = 0;
380 while (len-- > 0)
381 r = r * 31 + *s++;
382 return r;
383}
384
e5b1444b 385static void *pool_alloc(size_t len)
463acbe1
SP
386{
387 struct mem_pool *p;
388 void *r;
389
390 for (p = mem_pool; p; p = p->next_pool)
391 if ((p->end - p->next_free >= len))
392 break;
393
394 if (!p) {
395 if (len >= (mem_pool_alloc/2)) {
396 total_allocd += len;
397 return xmalloc(len);
398 }
399 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
400 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
401 p->next_pool = mem_pool;
402 p->next_free = p->space;
403 p->end = p->next_free + mem_pool_alloc;
404 mem_pool = p;
405 }
406
407 r = p->next_free;
8d8928b0
SP
408 /* round out to a pointer alignment */
409 if (len & (sizeof(void*) - 1))
410 len += sizeof(void*) - (len & (sizeof(void*) - 1));
463acbe1
SP
411 p->next_free += len;
412 return r;
413}
414
e5b1444b 415static void *pool_calloc(size_t count, size_t size)
463acbe1
SP
416{
417 size_t len = count * size;
418 void *r = pool_alloc(len);
419 memset(r, 0, len);
420 return r;
421}
422
e5b1444b 423static char *pool_strdup(const char *s)
463acbe1
SP
424{
425 char *r = pool_alloc(strlen(s) + 1);
426 strcpy(r, s);
427 return r;
428}
429
243f801d
SP
430static void size_dbuf(struct dbuf *b, size_t maxlen)
431{
432 if (b->buffer) {
433 if (b->capacity >= maxlen)
434 return;
435 free(b->buffer);
436 }
437 b->capacity = ((maxlen / 1024) + 1) * 1024;
438 b->buffer = xmalloc(b->capacity);
439}
440
0ea9f045 441static void insert_mark(uintmax_t idnum, struct object_entry *oe)
d8397168
SP
442{
443 struct mark_set *s = marks;
444 while ((idnum >> s->shift) >= 1024) {
445 s = pool_calloc(1, sizeof(struct mark_set));
446 s->shift = marks->shift + 10;
447 s->data.sets[0] = marks;
448 marks = s;
449 }
450 while (s->shift) {
0ea9f045 451 uintmax_t i = idnum >> s->shift;
d8397168
SP
452 idnum -= i << s->shift;
453 if (!s->data.sets[i]) {
454 s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
455 s->data.sets[i]->shift = s->shift - 10;
456 }
457 s = s->data.sets[i];
458 }
459 if (!s->data.marked[idnum])
460 marks_set_count++;
461 s->data.marked[idnum] = oe;
462}
463
e5b1444b 464static struct object_entry *find_mark(uintmax_t idnum)
d8397168 465{
0ea9f045 466 uintmax_t orig_idnum = idnum;
d8397168
SP
467 struct mark_set *s = marks;
468 struct object_entry *oe = NULL;
469 if ((idnum >> s->shift) < 1024) {
470 while (s && s->shift) {
0ea9f045 471 uintmax_t i = idnum >> s->shift;
d8397168
SP
472 idnum -= i << s->shift;
473 s = s->data.sets[i];
474 }
475 if (s)
476 oe = s->data.marked[idnum];
477 }
478 if (!oe)
3efb1f34 479 die("mark :%" PRIuMAX " not declared", orig_idnum);
d8397168
SP
480 return oe;
481}
482
e5b1444b 483static struct atom_str *to_atom(const char *s, unsigned short len)
463acbe1
SP
484{
485 unsigned int hc = hc_str(s, len) % atom_table_sz;
486 struct atom_str *c;
487
488 for (c = atom_table[hc]; c; c = c->next_atom)
489 if (c->str_len == len && !strncmp(s, c->str_dat, len))
490 return c;
491
492 c = pool_alloc(sizeof(struct atom_str) + len + 1);
493 c->str_len = len;
494 strncpy(c->str_dat, s, len);
495 c->str_dat[len] = 0;
496 c->next_atom = atom_table[hc];
497 atom_table[hc] = c;
498 atom_cnt++;
499 return c;
500}
501
e5b1444b 502static struct branch *lookup_branch(const char *name)
463acbe1
SP
503{
504 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
505 struct branch *b;
506
507 for (b = branch_table[hc]; b; b = b->table_next_branch)
508 if (!strcmp(name, b->name))
509 return b;
510 return NULL;
511}
512
e5b1444b 513static struct branch *new_branch(const char *name)
463acbe1
SP
514{
515 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
516 struct branch* b = lookup_branch(name);
517
518 if (b)
519 die("Invalid attempt to create duplicate branch: %s", name);
c44cdc7e
SP
520 if (check_ref_format(name))
521 die("Branch name doesn't conform to GIT standards: %s", name);
463acbe1
SP
522
523 b = pool_calloc(1, sizeof(struct branch));
524 b->name = pool_strdup(name);
525 b->table_next_branch = branch_table[hc];
8a8c55ea
SP
526 b->branch_tree.versions[0].mode = S_IFDIR;
527 b->branch_tree.versions[1].mode = S_IFDIR;
734c91f9 528 b->active = 0;
69e74e74 529 b->pack_id = MAX_PACK_ID;
463acbe1
SP
530 branch_table[hc] = b;
531 branch_count++;
532 return b;
533}
534
535static unsigned int hc_entries(unsigned int cnt)
536{
537 cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
538 return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
539}
540
e5b1444b 541static struct tree_content *new_tree_content(unsigned int cnt)
463acbe1
SP
542{
543 struct avail_tree_content *f, *l = NULL;
544 struct tree_content *t;
545 unsigned int hc = hc_entries(cnt);
546
547 for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
548 if (f->entry_capacity >= cnt)
549 break;
550
551 if (f) {
552 if (l)
553 l->next_avail = f->next_avail;
554 else
555 avail_tree_table[hc] = f->next_avail;
556 } else {
557 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
558 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
559 f->entry_capacity = cnt;
560 }
561
562 t = (struct tree_content*)f;
563 t->entry_count = 0;
4cabf858 564 t->delta_depth = 0;
463acbe1
SP
565 return t;
566}
567
568static void release_tree_entry(struct tree_entry *e);
569static void release_tree_content(struct tree_content *t)
570{
571 struct avail_tree_content *f = (struct avail_tree_content*)t;
572 unsigned int hc = hc_entries(f->entry_capacity);
afde8dd9
SP
573 f->next_avail = avail_tree_table[hc];
574 avail_tree_table[hc] = f;
575}
576
577static void release_tree_content_recursive(struct tree_content *t)
578{
463acbe1
SP
579 unsigned int i;
580 for (i = 0; i < t->entry_count; i++)
581 release_tree_entry(t->entries[i]);
afde8dd9 582 release_tree_content(t);
463acbe1
SP
583}
584
e5b1444b 585static struct tree_content *grow_tree_content(
463acbe1
SP
586 struct tree_content *t,
587 int amt)
588{
589 struct tree_content *r = new_tree_content(t->entry_count + amt);
590 r->entry_count = t->entry_count;
4cabf858 591 r->delta_depth = t->delta_depth;
463acbe1
SP
592 memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
593 release_tree_content(t);
594 return r;
595}
596
e5b1444b 597static struct tree_entry *new_tree_entry(void)
463acbe1
SP
598{
599 struct tree_entry *e;
600
601 if (!avail_tree_entry) {
602 unsigned int n = tree_entry_alloc;
8435a9cb 603 total_allocd += n * sizeof(struct tree_entry);
463acbe1 604 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
2eb26d84 605 while (n-- > 1) {
463acbe1
SP
606 *((void**)e) = e + 1;
607 e++;
608 }
35ef237c 609 *((void**)e) = NULL;
463acbe1
SP
610 }
611
612 e = avail_tree_entry;
613 avail_tree_entry = *((void**)e);
614 return e;
615}
616
617static void release_tree_entry(struct tree_entry *e)
618{
619 if (e->tree)
afde8dd9 620 release_tree_content_recursive(e->tree);
463acbe1
SP
621 *((void**)e) = avail_tree_entry;
622 avail_tree_entry = e;
623}
624
fd99224e 625static void start_packfile(void)
f70b6534 626{
8455e484 627 static char tmpfile[PATH_MAX];
7bfe6e26 628 struct packed_git *p;
f70b6534 629 struct pack_header hdr;
0fcbcae7 630 int pack_fd;
f70b6534 631
8455e484 632 snprintf(tmpfile, sizeof(tmpfile),
0e55181f 633 "%s/tmp_pack_XXXXXX", get_object_directory());
8455e484 634 pack_fd = mkstemp(tmpfile);
f70b6534 635 if (pack_fd < 0)
8455e484
SP
636 die("Can't create %s: %s", tmpfile, strerror(errno));
637 p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
638 strcpy(p->pack_name, tmpfile);
7bfe6e26 639 p->pack_fd = pack_fd;
f70b6534
SP
640
641 hdr.hdr_signature = htonl(PACK_SIGNATURE);
642 hdr.hdr_version = htonl(2);
643 hdr.hdr_entries = 0;
0fcbcae7 644 write_or_die(p->pack_fd, &hdr, sizeof(hdr));
7bfe6e26
SP
645
646 pack_data = p;
f70b6534
SP
647 pack_size = sizeof(hdr);
648 object_count = 0;
7bfe6e26
SP
649
650 all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
651 all_packs[pack_id] = p;
f70b6534
SP
652}
653
f70b6534
SP
654static int oecmp (const void *a_, const void *b_)
655{
656 struct object_entry *a = *((struct object_entry**)a_);
657 struct object_entry *b = *((struct object_entry**)b_);
658 return hashcmp(a->sha1, b->sha1);
659}
660
e5b1444b 661static char *create_index(void)
f70b6534 662{
8455e484
SP
663 static char tmpfile[PATH_MAX];
664 SHA_CTX ctx;
f70b6534
SP
665 struct sha1file *f;
666 struct object_entry **idx, **c, **last, *e;
667 struct object_entry_pool *o;
ebea9dd4 668 uint32_t array[256];
8455e484 669 int i, idx_fd;
f70b6534
SP
670
671 /* Build the sorted table of object IDs. */
672 idx = xmalloc(object_count * sizeof(struct object_entry*));
673 c = idx;
674 for (o = blocks; o; o = o->next_pool)
d9ee53ce
SP
675 for (e = o->next_free; e-- != o->entries;)
676 if (pack_id == e->pack_id)
677 *c++ = e;
f70b6534 678 last = idx + object_count;
2fce1f3c
SP
679 if (c != last)
680 die("internal consistency error creating the index");
f70b6534
SP
681 qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
682
683 /* Generate the fan-out array. */
684 c = idx;
685 for (i = 0; i < 256; i++) {
686 struct object_entry **next = c;;
687 while (next < last) {
688 if ((*next)->sha1[0] != i)
689 break;
690 next++;
691 }
692 array[i] = htonl(next - idx);
693 c = next;
694 }
695
8455e484 696 snprintf(tmpfile, sizeof(tmpfile),
0e55181f 697 "%s/tmp_idx_XXXXXX", get_object_directory());
8455e484
SP
698 idx_fd = mkstemp(tmpfile);
699 if (idx_fd < 0)
700 die("Can't create %s: %s", tmpfile, strerror(errno));
701 f = sha1fd(idx_fd, tmpfile);
f70b6534 702 sha1write(f, array, 256 * sizeof(int));
8455e484 703 SHA1_Init(&ctx);
f70b6534 704 for (c = idx; c != last; c++) {
ebea9dd4 705 uint32_t offset = htonl((*c)->offset);
f70b6534
SP
706 sha1write(f, &offset, 4);
707 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
8455e484 708 SHA1_Update(&ctx, (*c)->sha1, 20);
f70b6534 709 }
09543c96 710 sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
8455e484 711 sha1close(f, NULL, 1);
f70b6534 712 free(idx);
8455e484
SP
713 SHA1_Final(pack_data->sha1, &ctx);
714 return tmpfile;
715}
716
e5b1444b 717static char *keep_pack(char *curr_index_name)
8455e484
SP
718{
719 static char name[PATH_MAX];
3a55602e 720 static const char *keep_msg = "fast-import";
8455e484
SP
721 int keep_fd;
722
723 chmod(pack_data->pack_name, 0444);
724 chmod(curr_index_name, 0444);
725
726 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
727 get_object_directory(), sha1_to_hex(pack_data->sha1));
728 keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
729 if (keep_fd < 0)
730 die("cannot create keep file");
731 write(keep_fd, keep_msg, strlen(keep_msg));
732 close(keep_fd);
733
734 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
735 get_object_directory(), sha1_to_hex(pack_data->sha1));
736 if (move_temp_to_file(pack_data->pack_name, name))
737 die("cannot store pack file");
8455e484
SP
738
739 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
740 get_object_directory(), sha1_to_hex(pack_data->sha1));
741 if (move_temp_to_file(curr_index_name, name))
742 die("cannot store index file");
743 return name;
744}
745
fd99224e 746static void unkeep_all_packs(void)
8455e484
SP
747{
748 static char name[PATH_MAX];
749 int k;
750
751 for (k = 0; k < pack_id; k++) {
752 struct packed_git *p = all_packs[k];
753 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
754 get_object_directory(), sha1_to_hex(p->sha1));
755 unlink(name);
756 }
f70b6534
SP
757}
758
fd99224e 759static void end_packfile(void)
f70b6534 760{
7bfe6e26
SP
761 struct packed_git *old_p = pack_data, *new_p;
762
3e005baf 763 if (object_count) {
8455e484 764 char *idx_name;
2369ed79
SP
765 int i;
766 struct branch *b;
767 struct tag *t;
8455e484 768
8b0eca7c
DH
769 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
770 pack_data->pack_name, object_count);
771 close(pack_data->pack_fd);
8455e484 772 idx_name = keep_pack(create_index());
3e005baf
SP
773
774 /* Register the packfile with core git's machinary. */
775 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
776 if (!new_p)
777 die("core git rejected index %s", idx_name);
778 new_p->windows = old_p->windows;
2369ed79 779 all_packs[pack_id] = new_p;
3e005baf 780 install_packed_git(new_p);
2369ed79
SP
781
782 /* Print the boundary */
bdf1c06d
SP
783 if (pack_edges) {
784 fprintf(pack_edges, "%s:", new_p->pack_name);
785 for (i = 0; i < branch_table_sz; i++) {
786 for (b = branch_table[i]; b; b = b->table_next_branch) {
787 if (b->pack_id == pack_id)
788 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
789 }
2369ed79 790 }
bdf1c06d
SP
791 for (t = first_tag; t; t = t->next_tag) {
792 if (t->pack_id == pack_id)
793 fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
794 }
795 fputc('\n', pack_edges);
796 fflush(pack_edges);
2369ed79 797 }
2369ed79
SP
798
799 pack_id++;
3e005baf 800 }
12801587 801 else
3e005baf 802 unlink(old_p->pack_name);
7bfe6e26 803 free(old_p);
7bfe6e26
SP
804
805 /* We can't carry a delta across packfiles. */
806 free(last_blob.data);
807 last_blob.data = NULL;
808 last_blob.len = 0;
809 last_blob.offset = 0;
810 last_blob.depth = 0;
f70b6534
SP
811}
812
820b9310 813static void cycle_packfile(void)
d9ee53ce
SP
814{
815 end_packfile();
816 start_packfile();
817}
818
c44cdc7e 819static size_t encode_header(
ac47a738 820 enum object_type type,
c44cdc7e 821 size_t size,
ac47a738 822 unsigned char *hdr)
db5e523f
SP
823{
824 int n = 1;
825 unsigned char c;
826
1fcdd62a 827 if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
db5e523f
SP
828 die("bad type %d", type);
829
830 c = (type << 4) | (size & 15);
831 size >>= 4;
832 while (size) {
833 *hdr++ = c | 0x80;
834 c = size & 0x7f;
835 size >>= 7;
836 n++;
837 }
838 *hdr = c;
839 return n;
840}
841
ac47a738
SP
842static int store_object(
843 enum object_type type,
844 void *dat,
c44cdc7e 845 size_t datlen,
6bb5b329 846 struct last_object *last,
d8397168 847 unsigned char *sha1out,
0ea9f045 848 uintmax_t mark)
db5e523f 849{
db5e523f 850 void *out, *delta;
ac47a738
SP
851 struct object_entry *e;
852 unsigned char hdr[96];
853 unsigned char sha1[20];
db5e523f 854 unsigned long hdrlen, deltalen;
ac47a738
SP
855 SHA_CTX c;
856 z_stream s;
857
df843662 858 hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
40db58b8 859 (unsigned long)datlen) + 1;
ac47a738
SP
860 SHA1_Init(&c);
861 SHA1_Update(&c, hdr, hdrlen);
862 SHA1_Update(&c, dat, datlen);
863 SHA1_Final(sha1, &c);
6bb5b329 864 if (sha1out)
445b8599 865 hashcpy(sha1out, sha1);
ac47a738
SP
866
867 e = insert_object(sha1);
d8397168
SP
868 if (mark)
869 insert_mark(mark, e);
ac47a738 870 if (e->offset) {
6143f064 871 duplicate_count_by_type[type]++;
463acbe1 872 return 1;
a5c1780a
SP
873 } else if (find_sha1_pack(sha1, packed_git)) {
874 e->type = type;
875 e->pack_id = MAX_PACK_ID;
876 e->offset = 1; /* just not zero! */
877 duplicate_count_by_type[type]++;
878 return 1;
ac47a738 879 }
db5e523f 880
d9ee53ce 881 if (last && last->data && last->depth < max_depth) {
ac47a738 882 delta = diff_delta(last->data, last->len,
db5e523f
SP
883 dat, datlen,
884 &deltalen, 0);
d9ee53ce
SP
885 if (delta && deltalen >= datlen) {
886 free(delta);
887 delta = NULL;
888 }
889 } else
890 delta = NULL;
db5e523f
SP
891
892 memset(&s, 0, sizeof(s));
893 deflateInit(&s, zlib_compression_level);
d9ee53ce
SP
894 if (delta) {
895 s.next_in = delta;
896 s.avail_in = deltalen;
897 } else {
898 s.next_in = dat;
899 s.avail_in = datlen;
900 }
901 s.avail_out = deflateBound(&s, s.avail_in);
902 s.next_out = out = xmalloc(s.avail_out);
903 while (deflate(&s, Z_FINISH) == Z_OK)
904 /* nothing */;
905 deflateEnd(&s);
906
907 /* Determine if we should auto-checkpoint. */
e5808826 908 if ((pack_size + 60 + s.total_out) > max_packsize
d9ee53ce
SP
909 || (pack_size + 60 + s.total_out) < pack_size) {
910
911 /* This new object needs to *not* have the current pack_id. */
912 e->pack_id = pack_id + 1;
820b9310 913 cycle_packfile();
d9ee53ce
SP
914
915 /* We cannot carry a delta into the new pack. */
916 if (delta) {
917 free(delta);
918 delta = NULL;
5d6f3ef6
SP
919
920 memset(&s, 0, sizeof(s));
921 deflateInit(&s, zlib_compression_level);
922 s.next_in = dat;
923 s.avail_in = datlen;
924 s.avail_out = deflateBound(&s, s.avail_in);
925 s.next_out = out = xrealloc(out, s.avail_out);
926 while (deflate(&s, Z_FINISH) == Z_OK)
927 /* nothing */;
928 deflateEnd(&s);
d9ee53ce 929 }
d9ee53ce
SP
930 }
931
932 e->type = type;
933 e->pack_id = pack_id;
934 e->offset = pack_size;
935 object_count++;
936 object_count_by_type[type]++;
db5e523f
SP
937
938 if (delta) {
d489bc14
SP
939 unsigned long ofs = e->offset - last->offset;
940 unsigned pos = sizeof(hdr) - 1;
941
4cabf858 942 delta_count_by_type[type]++;
ac47a738 943 last->depth++;
d489bc14
SP
944
945 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
0fcbcae7 946 write_or_die(pack_data->pack_fd, hdr, hdrlen);
d489bc14
SP
947 pack_size += hdrlen;
948
949 hdr[pos] = ofs & 127;
950 while (ofs >>= 7)
951 hdr[--pos] = 128 | (--ofs & 127);
0fcbcae7 952 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
d489bc14 953 pack_size += sizeof(hdr) - pos;
db5e523f 954 } else {
463acbe1
SP
955 if (last)
956 last->depth = 0;
ac47a738 957 hdrlen = encode_header(type, datlen, hdr);
0fcbcae7 958 write_or_die(pack_data->pack_fd, hdr, hdrlen);
41e5257f 959 pack_size += hdrlen;
db5e523f
SP
960 }
961
0fcbcae7 962 write_or_die(pack_data->pack_fd, out, s.total_out);
41e5257f 963 pack_size += s.total_out;
db5e523f
SP
964
965 free(out);
e7d06a4b 966 free(delta);
463acbe1 967 if (last) {
e7d06a4b 968 if (!last->no_free)
463acbe1
SP
969 free(last->data);
970 last->data = dat;
d489bc14 971 last->offset = e->offset;
463acbe1 972 last->len = datlen;
463acbe1
SP
973 }
974 return 0;
975}
976
7bfe6e26
SP
977static void *gfi_unpack_entry(
978 struct object_entry *oe,
979 unsigned long *sizep)
41e5257f 980{
21666f1a 981 enum object_type type;
7bfe6e26
SP
982 struct packed_git *p = all_packs[oe->pack_id];
983 if (p == pack_data)
984 p->pack_size = pack_size + 20;
21666f1a 985 return unpack_entry(p, oe->offset, &type, sizep);
41e5257f
SP
986}
987
10831c55 988static const char *get_mode(const char *str, uint16_t *modep)
463acbe1
SP
989{
990 unsigned char c;
10831c55 991 uint16_t mode = 0;
463acbe1
SP
992
993 while ((c = *str++) != ' ') {
994 if (c < '0' || c > '7')
995 return NULL;
996 mode = (mode << 3) + (c - '0');
997 }
998 *modep = mode;
999 return str;
1000}
1001
1002static void load_tree(struct tree_entry *root)
1003{
4cabf858 1004 unsigned char* sha1 = root->versions[1].sha1;
463acbe1
SP
1005 struct object_entry *myoe;
1006 struct tree_content *t;
1007 unsigned long size;
1008 char *buf;
1009 const char *c;
463acbe1
SP
1010
1011 root->tree = t = new_tree_content(8);
4cabf858 1012 if (is_null_sha1(sha1))
463acbe1
SP
1013 return;
1014
4cabf858 1015 myoe = find_object(sha1);
20f546a8 1016 if (myoe && myoe->pack_id != MAX_PACK_ID) {
41e5257f 1017 if (myoe->type != OBJ_TREE)
4cabf858 1018 die("Not a tree: %s", sha1_to_hex(sha1));
d489bc14 1019 t->delta_depth = 0;
7bfe6e26 1020 buf = gfi_unpack_entry(myoe, &size);
463acbe1 1021 } else {
21666f1a
NP
1022 enum object_type type;
1023 buf = read_sha1_file(sha1, &type, &size);
1024 if (!buf || type != OBJ_TREE)
4cabf858 1025 die("Can't load tree %s", sha1_to_hex(sha1));
463acbe1
SP
1026 }
1027
1028 c = buf;
1029 while (c != (buf + size)) {
1030 struct tree_entry *e = new_tree_entry();
1031
1032 if (t->entry_count == t->entry_capacity)
f022f85f 1033 root->tree = t = grow_tree_content(t, t->entry_count);
463acbe1
SP
1034 t->entries[t->entry_count++] = e;
1035
1036 e->tree = NULL;
4cabf858 1037 c = get_mode(c, &e->versions[1].mode);
463acbe1 1038 if (!c)
4cabf858
SP
1039 die("Corrupt mode in %s", sha1_to_hex(sha1));
1040 e->versions[0].mode = e->versions[1].mode;
061e35c5 1041 e->name = to_atom(c, strlen(c));
463acbe1 1042 c += e->name->str_len + 1;
4cabf858
SP
1043 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1044 hashcpy(e->versions[1].sha1, (unsigned char*)c);
463acbe1
SP
1045 c += 20;
1046 }
1047 free(buf);
1048}
1049
4cabf858 1050static int tecmp0 (const void *_a, const void *_b)
463acbe1
SP
1051{
1052 struct tree_entry *a = *((struct tree_entry**)_a);
1053 struct tree_entry *b = *((struct tree_entry**)_b);
1054 return base_name_compare(
4cabf858
SP
1055 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1056 b->name->str_dat, b->name->str_len, b->versions[0].mode);
463acbe1
SP
1057}
1058
4cabf858 1059static int tecmp1 (const void *_a, const void *_b)
463acbe1 1060{
4cabf858
SP
1061 struct tree_entry *a = *((struct tree_entry**)_a);
1062 struct tree_entry *b = *((struct tree_entry**)_b);
1063 return base_name_compare(
1064 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1065 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1066}
1067
e2eb469d
SP
1068static void mktree(struct tree_content *t,
1069 int v,
1070 unsigned long *szp,
1071 struct dbuf *b)
4cabf858
SP
1072{
1073 size_t maxlen = 0;
463acbe1 1074 unsigned int i;
e2eb469d 1075 char *c;
463acbe1 1076
4cabf858
SP
1077 if (!v)
1078 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1079 else
1080 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
463acbe1 1081
463acbe1 1082 for (i = 0; i < t->entry_count; i++) {
4cabf858
SP
1083 if (t->entries[i]->versions[v].mode)
1084 maxlen += t->entries[i]->name->str_len + 34;
463acbe1
SP
1085 }
1086
243f801d 1087 size_dbuf(b, maxlen);
e2eb469d 1088 c = b->buffer;
463acbe1
SP
1089 for (i = 0; i < t->entry_count; i++) {
1090 struct tree_entry *e = t->entries[i];
4cabf858
SP
1091 if (!e->versions[v].mode)
1092 continue;
10831c55 1093 c += sprintf(c, "%o", (unsigned int)e->versions[v].mode);
463acbe1
SP
1094 *c++ = ' ';
1095 strcpy(c, e->name->str_dat);
1096 c += e->name->str_len + 1;
4cabf858 1097 hashcpy((unsigned char*)c, e->versions[v].sha1);
463acbe1
SP
1098 c += 20;
1099 }
e2eb469d 1100 *szp = c - (char*)b->buffer;
4cabf858
SP
1101}
1102
1103static void store_tree(struct tree_entry *root)
1104{
1105 struct tree_content *t = root->tree;
1106 unsigned int i, j, del;
e2eb469d 1107 unsigned long new_len;
4cabf858 1108 struct last_object lo;
d489bc14 1109 struct object_entry *le;
4cabf858
SP
1110
1111 if (!is_null_sha1(root->versions[1].sha1))
1112 return;
1113
1114 for (i = 0; i < t->entry_count; i++) {
1115 if (t->entries[i]->tree)
1116 store_tree(t->entries[i]);
1117 }
1118
d489bc14 1119 le = find_object(root->versions[0].sha1);
7bfe6e26
SP
1120 if (!S_ISDIR(root->versions[0].mode)
1121 || !le
1122 || le->pack_id != pack_id) {
4cabf858
SP
1123 lo.data = NULL;
1124 lo.depth = 0;
b259157f 1125 lo.no_free = 0;
4cabf858 1126 } else {
e2eb469d
SP
1127 mktree(t, 0, &lo.len, &old_tree);
1128 lo.data = old_tree.buffer;
d489bc14 1129 lo.offset = le->offset;
4cabf858 1130 lo.depth = t->delta_depth;
e2eb469d 1131 lo.no_free = 1;
4cabf858 1132 }
4cabf858 1133
8a8c55ea 1134 mktree(t, 1, &new_len, &new_tree);
e2eb469d 1135 store_object(OBJ_TREE, new_tree.buffer, new_len,
4cabf858 1136 &lo, root->versions[1].sha1, 0);
4cabf858
SP
1137
1138 t->delta_depth = lo.depth;
4cabf858
SP
1139 for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1140 struct tree_entry *e = t->entries[i];
1141 if (e->versions[1].mode) {
1142 e->versions[0].mode = e->versions[1].mode;
1143 hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1144 t->entries[j++] = e;
1145 } else {
1146 release_tree_entry(e);
1147 del++;
1148 }
1149 }
1150 t->entry_count -= del;
463acbe1
SP
1151}
1152
1153static int tree_content_set(
1154 struct tree_entry *root,
1155 const char *p,
1156 const unsigned char *sha1,
10831c55 1157 const uint16_t mode)
463acbe1
SP
1158{
1159 struct tree_content *t = root->tree;
1160 const char *slash1;
1161 unsigned int i, n;
1162 struct tree_entry *e;
1163
1164 slash1 = strchr(p, '/');
1165 if (slash1)
1166 n = slash1 - p;
1167 else
1168 n = strlen(p);
475d1b33
SP
1169 if (!n)
1170 die("Empty path component found in input");
463acbe1
SP
1171
1172 for (i = 0; i < t->entry_count; i++) {
1173 e = t->entries[i];
1174 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1175 if (!slash1) {
4cabf858
SP
1176 if (e->versions[1].mode == mode
1177 && !hashcmp(e->versions[1].sha1, sha1))
463acbe1 1178 return 0;
4cabf858
SP
1179 e->versions[1].mode = mode;
1180 hashcpy(e->versions[1].sha1, sha1);
463acbe1 1181 if (e->tree) {
afde8dd9 1182 release_tree_content_recursive(e->tree);
463acbe1
SP
1183 e->tree = NULL;
1184 }
4cabf858 1185 hashclr(root->versions[1].sha1);
463acbe1
SP
1186 return 1;
1187 }
4cabf858 1188 if (!S_ISDIR(e->versions[1].mode)) {
463acbe1 1189 e->tree = new_tree_content(8);
4cabf858 1190 e->versions[1].mode = S_IFDIR;
463acbe1
SP
1191 }
1192 if (!e->tree)
1193 load_tree(e);
1194 if (tree_content_set(e, slash1 + 1, sha1, mode)) {
4cabf858 1195 hashclr(root->versions[1].sha1);
463acbe1
SP
1196 return 1;
1197 }
1198 return 0;
1199 }
1200 }
1201
1202 if (t->entry_count == t->entry_capacity)
f022f85f 1203 root->tree = t = grow_tree_content(t, t->entry_count);
463acbe1 1204 e = new_tree_entry();
061e35c5 1205 e->name = to_atom(p, n);
4cabf858
SP
1206 e->versions[0].mode = 0;
1207 hashclr(e->versions[0].sha1);
463acbe1
SP
1208 t->entries[t->entry_count++] = e;
1209 if (slash1) {
1210 e->tree = new_tree_content(8);
4cabf858 1211 e->versions[1].mode = S_IFDIR;
463acbe1
SP
1212 tree_content_set(e, slash1 + 1, sha1, mode);
1213 } else {
1214 e->tree = NULL;
4cabf858
SP
1215 e->versions[1].mode = mode;
1216 hashcpy(e->versions[1].sha1, sha1);
463acbe1 1217 }
4cabf858 1218 hashclr(root->versions[1].sha1);
463acbe1
SP
1219 return 1;
1220}
1221
1222static int tree_content_remove(struct tree_entry *root, const char *p)
1223{
1224 struct tree_content *t = root->tree;
1225 const char *slash1;
1226 unsigned int i, n;
1227 struct tree_entry *e;
1228
1229 slash1 = strchr(p, '/');
1230 if (slash1)
1231 n = slash1 - p;
1232 else
1233 n = strlen(p);
1234
1235 for (i = 0; i < t->entry_count; i++) {
1236 e = t->entries[i];
1237 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
4cabf858 1238 if (!slash1 || !S_ISDIR(e->versions[1].mode))
463acbe1
SP
1239 goto del_entry;
1240 if (!e->tree)
1241 load_tree(e);
1242 if (tree_content_remove(e, slash1 + 1)) {
b54d6422
SP
1243 for (n = 0; n < e->tree->entry_count; n++) {
1244 if (e->tree->entries[n]->versions[1].mode) {
1245 hashclr(root->versions[1].sha1);
1246 return 1;
1247 }
1248 }
1249 goto del_entry;
463acbe1
SP
1250 }
1251 return 0;
1252 }
1253 }
1254 return 0;
1255
1256del_entry:
4cabf858
SP
1257 if (e->tree) {
1258 release_tree_content_recursive(e->tree);
1259 e->tree = NULL;
1260 }
1261 e->versions[1].mode = 0;
1262 hashclr(e->versions[1].sha1);
1263 hashclr(root->versions[1].sha1);
ac47a738 1264 return 1;
db5e523f
SP
1265}
1266
7073e69e 1267static int update_branch(struct branch *b)
463acbe1
SP
1268{
1269 static const char *msg = "fast-import";
7073e69e
SP
1270 struct ref_lock *lock;
1271 unsigned char old_sha1[20];
1272
1273 if (read_ref(b->name, old_sha1))
1274 hashclr(old_sha1);
68db31cc 1275 lock = lock_any_ref_for_update(b->name, old_sha1, 0);
7073e69e
SP
1276 if (!lock)
1277 return error("Unable to lock %s", b->name);
1278 if (!force_update && !is_null_sha1(old_sha1)) {
1279 struct commit *old_cmit, *new_cmit;
1280
1281 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1282 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1283 if (!old_cmit || !new_cmit) {
1284 unlock_ref(lock);
1285 return error("Branch %s is missing commits.", b->name);
1286 }
1287
4a164d48 1288 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
7073e69e 1289 unlock_ref(lock);
46efd2d9 1290 warning("Not updating %s"
7073e69e
SP
1291 " (new tip %s does not contain %s)",
1292 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1293 return -1;
1294 }
1295 }
1296 if (write_ref_sha1(lock, b->sha1, msg) < 0)
1297 return error("Unable to update %s", b->name);
1298 return 0;
1299}
1300
1301static void dump_branches(void)
1302{
463acbe1
SP
1303 unsigned int i;
1304 struct branch *b;
463acbe1
SP
1305
1306 for (i = 0; i < branch_table_sz; i++) {
7073e69e
SP
1307 for (b = branch_table[i]; b; b = b->table_next_branch)
1308 failure |= update_branch(b);
463acbe1
SP
1309 }
1310}
1311
fd99224e 1312static void dump_tags(void)
72303d44
SP
1313{
1314 static const char *msg = "fast-import";
1315 struct tag *t;
1316 struct ref_lock *lock;
7073e69e 1317 char ref_name[PATH_MAX];
72303d44
SP
1318
1319 for (t = first_tag; t; t = t->next_tag) {
7073e69e
SP
1320 sprintf(ref_name, "tags/%s", t->name);
1321 lock = lock_ref_sha1(ref_name, NULL);
72303d44 1322 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
7073e69e 1323 failure |= error("Unable to update %s", ref_name);
72303d44
SP
1324 }
1325}
1326
a6a1a831 1327static void dump_marks_helper(FILE *f,
0ea9f045 1328 uintmax_t base,
a6a1a831
SP
1329 struct mark_set *m)
1330{
0ea9f045 1331 uintmax_t k;
a6a1a831
SP
1332 if (m->shift) {
1333 for (k = 0; k < 1024; k++) {
1334 if (m->data.sets[k])
1335 dump_marks_helper(f, (base + k) << m->shift,
1336 m->data.sets[k]);
1337 }
1338 } else {
1339 for (k = 0; k < 1024; k++) {
1340 if (m->data.marked[k])
3efb1f34 1341 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
a6a1a831
SP
1342 sha1_to_hex(m->data.marked[k]->sha1));
1343 }
1344 }
1345}
1346
fd99224e 1347static void dump_marks(void)
a6a1a831 1348{
60b9004c
SP
1349 static struct lock_file mark_lock;
1350 int mark_fd;
1351 FILE *f;
1352
1353 if (!mark_file)
1354 return;
1355
1356 mark_fd = hold_lock_file_for_update(&mark_lock, mark_file, 0);
1357 if (mark_fd < 0) {
1358 failure |= error("Unable to write marks file %s: %s",
1359 mark_file, strerror(errno));
1360 return;
a6a1a831 1361 }
60b9004c
SP
1362
1363 f = fdopen(mark_fd, "w");
1364 if (!f) {
1365 rollback_lock_file(&mark_lock);
1366 failure |= error("Unable to write marks file %s: %s",
1367 mark_file, strerror(errno));
1368 return;
a6a1a831 1369 }
60b9004c
SP
1370
1371 dump_marks_helper(f, 0, marks);
1372 fclose(f);
1373 if (commit_lock_file(&mark_lock))
1374 failure |= error("Unable to write marks file %s: %s",
1375 mark_file, strerror(errno));
a6a1a831
SP
1376}
1377
fd99224e 1378static void read_next_command(void)
c44cdc7e
SP
1379{
1380 read_line(&command_buf, stdin, '\n');
1381}
1382
fd99224e 1383static void cmd_mark(void)
c44cdc7e 1384{
599065a3 1385 if (!prefixcmp(command_buf.buf, "mark :")) {
0ea9f045 1386 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
c44cdc7e
SP
1387 read_next_command();
1388 }
1389 else
d8397168 1390 next_mark = 0;
c44cdc7e
SP
1391}
1392
e5b1444b 1393static void *cmd_data (size_t *size)
c44cdc7e 1394{
c44cdc7e 1395 size_t length;
3b4dce02 1396 char *buffer;
c44cdc7e 1397
599065a3 1398 if (prefixcmp(command_buf.buf, "data "))
c44cdc7e
SP
1399 die("Expected 'data n' command, found: %s", command_buf.buf);
1400
599065a3 1401 if (!prefixcmp(command_buf.buf + 5, "<<")) {
3b4dce02
SP
1402 char *term = xstrdup(command_buf.buf + 5 + 2);
1403 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1404 length = 0;
1405 buffer = xmalloc(sz);
1406 for (;;) {
1407 read_next_command();
1408 if (command_buf.eof)
1409 die("EOF in data (terminator '%s' not found)", term);
1410 if (term_len == command_buf.len
1411 && !strcmp(term, command_buf.buf))
1412 break;
1413 if (sz < (length + command_buf.len)) {
1414 sz = sz * 3 / 2 + 16;
1415 if (sz < (length + command_buf.len))
1416 sz = length + command_buf.len;
1417 buffer = xrealloc(buffer, sz);
1418 }
1419 memcpy(buffer + length,
1420 command_buf.buf,
1421 command_buf.len - 1);
1422 length += command_buf.len - 1;
1423 buffer[length++] = '\n';
1424 }
1425 free(term);
1426 }
1427 else {
1428 size_t n = 0;
1429 length = strtoul(command_buf.buf + 5, NULL, 10);
1430 buffer = xmalloc(length);
1431 while (n < length) {
1432 size_t s = fread(buffer + n, 1, length - n, stdin);
1433 if (!s && feof(stdin))
40db58b8
JS
1434 die("EOF in data (%lu bytes remaining)",
1435 (unsigned long)(length - n));
3b4dce02
SP
1436 n += s;
1437 }
c44cdc7e
SP
1438 }
1439
1440 if (fgetc(stdin) != '\n')
1441 die("An lf did not trail the binary data as expected.");
1442
1443 *size = length;
1444 return buffer;
1445}
1446
63e0c8b3
SP
1447static int validate_raw_date(const char *src, char *result, int maxlen)
1448{
1449 const char *orig_src = src;
1450 char *endp, sign;
1451
1452 strtoul(src, &endp, 10);
1453 if (endp == src || *endp != ' ')
1454 return -1;
1455
1456 src = endp + 1;
1457 if (*src != '-' && *src != '+')
1458 return -1;
1459 sign = *src;
1460
1461 strtoul(src + 1, &endp, 10);
1462 if (endp == src || *endp || (endp - orig_src) >= maxlen)
1463 return -1;
1464
1465 strcpy(result, orig_src);
1466 return 0;
1467}
1468
1469static char *parse_ident(const char *buf)
1470{
1471 const char *gt;
1472 size_t name_len;
1473 char *ident;
1474
1475 gt = strrchr(buf, '>');
1476 if (!gt)
1477 die("Missing > in ident string: %s", buf);
1478 gt++;
1479 if (*gt != ' ')
1480 die("Missing space after > in ident string: %s", buf);
1481 gt++;
1482 name_len = gt - buf;
1483 ident = xmalloc(name_len + 24);
1484 strncpy(ident, buf, name_len);
1485
1486 switch (whenspec) {
1487 case WHENSPEC_RAW:
1488 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1489 die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1490 break;
1491 case WHENSPEC_RFC2822:
1492 if (parse_date(gt, ident + name_len, 24) < 0)
1493 die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1494 break;
1495 case WHENSPEC_NOW:
1496 if (strcmp("now", gt))
1497 die("Date in ident must be 'now': %s", buf);
1498 datestamp(ident + name_len, 24);
1499 break;
1500 }
1501
1502 return ident;
1503}
1504
fd99224e 1505static void cmd_new_blob(void)
6143f064 1506{
d8397168
SP
1507 size_t l;
1508 void *d;
c44cdc7e
SP
1509
1510 read_next_command();
1511 cmd_mark();
d8397168 1512 d = cmd_data(&l);
6143f064 1513
d8397168
SP
1514 if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1515 free(d);
6143f064
SP
1516}
1517
fd99224e 1518static void unload_one_branch(void)
6bb5b329 1519{
41e5257f
SP
1520 while (cur_active_branches
1521 && cur_active_branches >= max_active_branches) {
6777a59f 1522 uintmax_t min_commit = ULONG_MAX;
463acbe1
SP
1523 struct branch *e, *l = NULL, *p = NULL;
1524
1525 for (e = active_branches; e; e = e->active_next_branch) {
1526 if (e->last_commit < min_commit) {
1527 p = l;
1528 min_commit = e->last_commit;
1529 }
1530 l = e;
1531 }
1532
1533 if (p) {
1534 e = p->active_next_branch;
1535 p->active_next_branch = e->active_next_branch;
1536 } else {
1537 e = active_branches;
1538 active_branches = e->active_next_branch;
1539 }
734c91f9 1540 e->active = 0;
463acbe1
SP
1541 e->active_next_branch = NULL;
1542 if (e->branch_tree.tree) {
afde8dd9 1543 release_tree_content_recursive(e->branch_tree.tree);
463acbe1
SP
1544 e->branch_tree.tree = NULL;
1545 }
1546 cur_active_branches--;
6bb5b329 1547 }
6bb5b329
SP
1548}
1549
463acbe1 1550static void load_branch(struct branch *b)
6bb5b329 1551{
463acbe1 1552 load_tree(&b->branch_tree);
734c91f9
SP
1553 if (!b->active) {
1554 b->active = 1;
1555 b->active_next_branch = active_branches;
1556 active_branches = b;
1557 cur_active_branches++;
1558 branch_load_count++;
1559 }
6bb5b329
SP
1560}
1561
463acbe1 1562static void file_change_m(struct branch *b)
6bb5b329 1563{
c44cdc7e
SP
1564 const char *p = command_buf.buf + 2;
1565 char *p_uq;
1566 const char *endp;
10e8d688 1567 struct object_entry *oe = oe;
463acbe1 1568 unsigned char sha1[20];
10831c55 1569 uint16_t mode, inline_data = 0;
6bb5b329 1570
c44cdc7e
SP
1571 p = get_mode(p, &mode);
1572 if (!p)
1573 die("Corrupt mode: %s", command_buf.buf);
1574 switch (mode) {
1575 case S_IFREG | 0644:
1576 case S_IFREG | 0755:
ace4a9d1 1577 case S_IFLNK:
c44cdc7e
SP
1578 case 0644:
1579 case 0755:
1580 /* ok */
1581 break;
1582 default:
1583 die("Corrupt mode: %s", command_buf.buf);
1584 }
1585
d8397168
SP
1586 if (*p == ':') {
1587 char *x;
0ea9f045 1588 oe = find_mark(strtoumax(p + 1, &x, 10));
cacbdd0a 1589 hashcpy(sha1, oe->sha1);
d8397168 1590 p = x;
599065a3 1591 } else if (!prefixcmp(p, "inline")) {
b715cfbb
SP
1592 inline_data = 1;
1593 p += 6;
d8397168
SP
1594 } else {
1595 if (get_sha1_hex(p, sha1))
1596 die("Invalid SHA1: %s", command_buf.buf);
1597 oe = find_object(sha1);
1598 p += 40;
1599 }
c44cdc7e
SP
1600 if (*p++ != ' ')
1601 die("Missing space after SHA1: %s", command_buf.buf);
1602
1603 p_uq = unquote_c_style(p, &endp);
1604 if (p_uq) {
1605 if (*endp)
1606 die("Garbage after path in: %s", command_buf.buf);
1607 p = p_uq;
1608 }
6bb5b329 1609
b715cfbb
SP
1610 if (inline_data) {
1611 size_t l;
1612 void *d;
1613 if (!p_uq)
1614 p = p_uq = xstrdup(p);
1615 read_next_command();
1616 d = cmd_data(&l);
1617 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1618 free(d);
1619 } else if (oe) {
7111feed 1620 if (oe->type != OBJ_BLOB)
c44cdc7e 1621 die("Not a blob (actually a %s): %s",
df843662 1622 command_buf.buf, typename(oe->type));
7111feed 1623 } else {
21666f1a
NP
1624 enum object_type type = sha1_object_info(sha1, NULL);
1625 if (type < 0)
c44cdc7e 1626 die("Blob not found: %s", command_buf.buf);
21666f1a 1627 if (type != OBJ_BLOB)
c44cdc7e 1628 die("Not a blob (actually a %s): %s",
21666f1a 1629 typename(type), command_buf.buf);
7111feed 1630 }
6bb5b329 1631
c44cdc7e 1632 tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
e7d06a4b 1633 free(p_uq);
463acbe1 1634}
6bb5b329 1635
463acbe1
SP
1636static void file_change_d(struct branch *b)
1637{
c44cdc7e
SP
1638 const char *p = command_buf.buf + 2;
1639 char *p_uq;
1640 const char *endp;
1641
1642 p_uq = unquote_c_style(p, &endp);
1643 if (p_uq) {
1644 if (*endp)
1645 die("Garbage after path in: %s", command_buf.buf);
1646 p = p_uq;
1647 }
1648 tree_content_remove(&b->branch_tree, p);
e7d06a4b 1649 free(p_uq);
6bb5b329
SP
1650}
1651
825769a8
SP
1652static void file_change_deleteall(struct branch *b)
1653{
1654 release_tree_content_recursive(b->branch_tree.tree);
1655 hashclr(b->branch_tree.versions[0].sha1);
1656 hashclr(b->branch_tree.versions[1].sha1);
1657 load_tree(&b->branch_tree);
1658}
1659
654aaa37
SP
1660static void cmd_from_commit(struct branch *b, char *buf, unsigned long size)
1661{
1662 if (!buf || size < 46)
1663 die("Not a valid commit: %s", sha1_to_hex(b->sha1));
1664 if (memcmp("tree ", buf, 5)
1665 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1666 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1667 hashcpy(b->branch_tree.versions[0].sha1,
1668 b->branch_tree.versions[1].sha1);
1669}
1670
1671static void cmd_from_existing(struct branch *b)
1672{
1673 if (is_null_sha1(b->sha1)) {
1674 hashclr(b->branch_tree.versions[0].sha1);
1675 hashclr(b->branch_tree.versions[1].sha1);
1676 } else {
1677 unsigned long size;
1678 char *buf;
1679
1680 buf = read_object_with_reference(b->sha1,
1681 commit_type, &size, b->sha1);
1682 cmd_from_commit(b, buf, size);
1683 free(buf);
1684 }
1685}
1686
00e2b884
SP
1687static void cmd_from(struct branch *b)
1688{
6c3aac1c 1689 const char *from;
00e2b884
SP
1690 struct branch *s;
1691
599065a3 1692 if (prefixcmp(command_buf.buf, "from "))
00e2b884
SP
1693 return;
1694
ea5e370a
SP
1695 if (b->branch_tree.tree) {
1696 release_tree_content_recursive(b->branch_tree.tree);
1697 b->branch_tree.tree = NULL;
1698 }
00e2b884
SP
1699
1700 from = strchr(command_buf.buf, ' ') + 1;
00e2b884
SP
1701 s = lookup_branch(from);
1702 if (b == s)
1703 die("Can't create a branch from itself: %s", b->name);
1704 else if (s) {
4cabf858 1705 unsigned char *t = s->branch_tree.versions[1].sha1;
445b8599 1706 hashcpy(b->sha1, s->sha1);
4cabf858
SP
1707 hashcpy(b->branch_tree.versions[0].sha1, t);
1708 hashcpy(b->branch_tree.versions[1].sha1, t);
00e2b884 1709 } else if (*from == ':') {
0ea9f045 1710 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
00e2b884 1711 struct object_entry *oe = find_mark(idnum);
00e2b884 1712 if (oe->type != OBJ_COMMIT)
3efb1f34 1713 die("Mark :%" PRIuMAX " not a commit", idnum);
445b8599 1714 hashcpy(b->sha1, oe->sha1);
aac65ed1 1715 if (oe->pack_id != MAX_PACK_ID) {
00e2b884 1716 unsigned long size;
aac65ed1
SP
1717 char *buf = gfi_unpack_entry(oe, &size);
1718 cmd_from_commit(b, buf, size);
00e2b884 1719 free(buf);
aac65ed1
SP
1720 } else
1721 cmd_from_existing(b);
654aaa37
SP
1722 } else if (!get_sha1(from, b->sha1))
1723 cmd_from_existing(b);
1724 else
00e2b884
SP
1725 die("Invalid ref name or SHA1 expression: %s", from);
1726
1727 read_next_command();
1728}
1729
e5b1444b 1730static struct hash_list *cmd_merge(unsigned int *count)
62b6f483 1731{
10e8d688 1732 struct hash_list *list = NULL, *n, *e = e;
6c3aac1c 1733 const char *from;
62b6f483
SP
1734 struct branch *s;
1735
1736 *count = 0;
599065a3 1737 while (!prefixcmp(command_buf.buf, "merge ")) {
62b6f483 1738 from = strchr(command_buf.buf, ' ') + 1;
62b6f483
SP
1739 n = xmalloc(sizeof(*n));
1740 s = lookup_branch(from);
1741 if (s)
1742 hashcpy(n->sha1, s->sha1);
1743 else if (*from == ':') {
0ea9f045 1744 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
62b6f483
SP
1745 struct object_entry *oe = find_mark(idnum);
1746 if (oe->type != OBJ_COMMIT)
3efb1f34 1747 die("Mark :%" PRIuMAX " not a commit", idnum);
62b6f483 1748 hashcpy(n->sha1, oe->sha1);
2f6dc35d
SP
1749 } else if (!get_sha1(from, n->sha1)) {
1750 unsigned long size;
1751 char *buf = read_object_with_reference(n->sha1,
6b4318e6 1752 commit_type, &size, n->sha1);
2f6dc35d
SP
1753 if (!buf || size < 46)
1754 die("Not a valid commit: %s", from);
1755 free(buf);
1756 } else
62b6f483
SP
1757 die("Invalid ref name or SHA1 expression: %s", from);
1758
1759 n->next = NULL;
1760 if (list)
1761 e->next = n;
1762 else
1763 list = n;
1764 e = n;
10e8d688 1765 (*count)++;
62b6f483
SP
1766 read_next_command();
1767 }
1768 return list;
1769}
1770
fd99224e 1771static void cmd_new_commit(void)
6bb5b329 1772{
c44cdc7e
SP
1773 struct branch *b;
1774 void *msg;
1775 size_t msglen;
c44cdc7e
SP
1776 char *sp;
1777 char *author = NULL;
1778 char *committer = NULL;
62b6f483
SP
1779 struct hash_list *merge_list = NULL;
1780 unsigned int merge_count;
c44cdc7e
SP
1781
1782 /* Obtain the branch name from the rest of our command */
1783 sp = strchr(command_buf.buf, ' ') + 1;
c44cdc7e 1784 b = lookup_branch(sp);
463acbe1 1785 if (!b)
00e2b884 1786 b = new_branch(sp);
c44cdc7e
SP
1787
1788 read_next_command();
1789 cmd_mark();
599065a3 1790 if (!prefixcmp(command_buf.buf, "author ")) {
63e0c8b3 1791 author = parse_ident(command_buf.buf + 7);
c44cdc7e
SP
1792 read_next_command();
1793 }
599065a3 1794 if (!prefixcmp(command_buf.buf, "committer ")) {
63e0c8b3 1795 committer = parse_ident(command_buf.buf + 10);
c44cdc7e
SP
1796 read_next_command();
1797 }
1798 if (!committer)
1799 die("Expected committer but didn't get one");
1800 msg = cmd_data(&msglen);
02f3389d
SP
1801 read_next_command();
1802 cmd_from(b);
62b6f483 1803 merge_list = cmd_merge(&merge_count);
c44cdc7e
SP
1804
1805 /* ensure the branch is active/loaded */
41e5257f 1806 if (!b->branch_tree.tree || !max_active_branches) {
463acbe1
SP
1807 unload_one_branch();
1808 load_branch(b);
1809 }
6bb5b329 1810
463acbe1
SP
1811 /* file_change* */
1812 for (;;) {
c44cdc7e 1813 if (1 == command_buf.len)
463acbe1 1814 break;
599065a3 1815 else if (!prefixcmp(command_buf.buf, "M "))
463acbe1 1816 file_change_m(b);
599065a3 1817 else if (!prefixcmp(command_buf.buf, "D "))
463acbe1 1818 file_change_d(b);
825769a8
SP
1819 else if (!strcmp("deleteall", command_buf.buf))
1820 file_change_deleteall(b);
463acbe1 1821 else
c44cdc7e 1822 die("Unsupported file_change: %s", command_buf.buf);
02f3389d 1823 read_next_command();
6bb5b329 1824 }
6bb5b329 1825
c44cdc7e 1826 /* build the tree and the commit */
463acbe1 1827 store_tree(&b->branch_tree);
8a8c55ea
SP
1828 hashcpy(b->branch_tree.versions[0].sha1,
1829 b->branch_tree.versions[1].sha1);
63e0c8b3 1830 size_dbuf(&new_data, 114 + msglen
62b6f483 1831 + merge_count * 49
c44cdc7e
SP
1832 + (author
1833 ? strlen(author) + strlen(committer)
1834 : 2 * strlen(committer)));
243f801d 1835 sp = new_data.buffer;
4cabf858
SP
1836 sp += sprintf(sp, "tree %s\n",
1837 sha1_to_hex(b->branch_tree.versions[1].sha1));
445b8599 1838 if (!is_null_sha1(b->sha1))
c44cdc7e 1839 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
62b6f483
SP
1840 while (merge_list) {
1841 struct hash_list *next = merge_list->next;
1842 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1843 free(merge_list);
1844 merge_list = next;
1845 }
63e0c8b3
SP
1846 sp += sprintf(sp, "author %s\n", author ? author : committer);
1847 sp += sprintf(sp, "committer %s\n", committer);
1848 *sp++ = '\n';
c44cdc7e
SP
1849 memcpy(sp, msg, msglen);
1850 sp += msglen;
e7d06a4b 1851 free(author);
c44cdc7e
SP
1852 free(committer);
1853 free(msg);
1854
69e74e74 1855 if (!store_object(OBJ_COMMIT,
243f801d 1856 new_data.buffer, sp - (char*)new_data.buffer,
69e74e74
SP
1857 NULL, b->sha1, next_mark))
1858 b->pack_id = pack_id;
463acbe1 1859 b->last_commit = object_count_by_type[OBJ_COMMIT];
6bb5b329
SP
1860}
1861
fd99224e 1862static void cmd_new_tag(void)
72303d44 1863{
72303d44
SP
1864 char *sp;
1865 const char *from;
1866 char *tagger;
1867 struct branch *s;
1868 void *msg;
1869 size_t msglen;
72303d44 1870 struct tag *t;
0ea9f045 1871 uintmax_t from_mark = 0;
72303d44
SP
1872 unsigned char sha1[20];
1873
1874 /* Obtain the new tag name from the rest of our command */
1875 sp = strchr(command_buf.buf, ' ') + 1;
72303d44
SP
1876 t = pool_alloc(sizeof(struct tag));
1877 t->next_tag = NULL;
1878 t->name = pool_strdup(sp);
1879 if (last_tag)
1880 last_tag->next_tag = t;
1881 else
1882 first_tag = t;
1883 last_tag = t;
72303d44
SP
1884 read_next_command();
1885
1886 /* from ... */
599065a3 1887 if (prefixcmp(command_buf.buf, "from "))
72303d44 1888 die("Expected from command, got %s", command_buf.buf);
72303d44 1889 from = strchr(command_buf.buf, ' ') + 1;
72303d44
SP
1890 s = lookup_branch(from);
1891 if (s) {
445b8599 1892 hashcpy(sha1, s->sha1);
72303d44 1893 } else if (*from == ':') {
10e8d688 1894 struct object_entry *oe;
0ea9f045 1895 from_mark = strtoumax(from + 1, NULL, 10);
10e8d688 1896 oe = find_mark(from_mark);
72303d44 1897 if (oe->type != OBJ_COMMIT)
3efb1f34 1898 die("Mark :%" PRIuMAX " not a commit", from_mark);
445b8599 1899 hashcpy(sha1, oe->sha1);
72303d44
SP
1900 } else if (!get_sha1(from, sha1)) {
1901 unsigned long size;
1902 char *buf;
1903
1904 buf = read_object_with_reference(sha1,
df843662 1905 commit_type, &size, sha1);
72303d44
SP
1906 if (!buf || size < 46)
1907 die("Not a valid commit: %s", from);
1908 free(buf);
1909 } else
1910 die("Invalid ref name or SHA1 expression: %s", from);
72303d44
SP
1911 read_next_command();
1912
1913 /* tagger ... */
599065a3 1914 if (prefixcmp(command_buf.buf, "tagger "))
72303d44 1915 die("Expected tagger command, got %s", command_buf.buf);
63e0c8b3 1916 tagger = parse_ident(command_buf.buf + 7);
72303d44
SP
1917
1918 /* tag payload/message */
1919 read_next_command();
1920 msg = cmd_data(&msglen);
1921
1922 /* build the tag object */
243f801d
SP
1923 size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1924 sp = new_data.buffer;
72303d44 1925 sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
df843662 1926 sp += sprintf(sp, "type %s\n", commit_type);
72303d44 1927 sp += sprintf(sp, "tag %s\n", t->name);
63e0c8b3
SP
1928 sp += sprintf(sp, "tagger %s\n", tagger);
1929 *sp++ = '\n';
72303d44
SP
1930 memcpy(sp, msg, msglen);
1931 sp += msglen;
1932 free(tagger);
1933 free(msg);
1934
69e74e74
SP
1935 if (store_object(OBJ_TAG, new_data.buffer,
1936 sp - (char*)new_data.buffer,
1937 NULL, t->sha1, 0))
1938 t->pack_id = MAX_PACK_ID;
1939 else
1940 t->pack_id = pack_id;
72303d44
SP
1941}
1942
fd99224e 1943static void cmd_reset_branch(void)
5fced8dc
SP
1944{
1945 struct branch *b;
5fced8dc
SP
1946 char *sp;
1947
1948 /* Obtain the branch name from the rest of our command */
1949 sp = strchr(command_buf.buf, ' ') + 1;
5fced8dc
SP
1950 b = lookup_branch(sp);
1951 if (b) {
ea5e370a
SP
1952 hashclr(b->sha1);
1953 hashclr(b->branch_tree.versions[0].sha1);
1954 hashclr(b->branch_tree.versions[1].sha1);
5fced8dc
SP
1955 if (b->branch_tree.tree) {
1956 release_tree_content_recursive(b->branch_tree.tree);
1957 b->branch_tree.tree = NULL;
1958 }
1959 }
9938ffc5
SP
1960 else
1961 b = new_branch(sp);
9938ffc5
SP
1962 read_next_command();
1963 cmd_from(b);
5fced8dc
SP
1964}
1965
fd99224e 1966static void cmd_checkpoint(void)
7bfe6e26 1967{
820b9310
SP
1968 if (object_count) {
1969 cycle_packfile();
1970 dump_branches();
1971 dump_tags();
1972 dump_marks();
1973 }
7bfe6e26
SP
1974 read_next_command();
1975}
1976
e8438420
SP
1977static void import_marks(const char *input_file)
1978{
1979 char line[512];
1980 FILE *f = fopen(input_file, "r");
1981 if (!f)
1982 die("cannot read %s: %s", input_file, strerror(errno));
1983 while (fgets(line, sizeof(line), f)) {
1984 uintmax_t mark;
1985 char *end;
1986 unsigned char sha1[20];
1987 struct object_entry *e;
1988
1989 end = strchr(line, '\n');
1990 if (line[0] != ':' || !end)
1991 die("corrupt mark line: %s", line);
1992 *end = 0;
1993 mark = strtoumax(line + 1, &end, 10);
1994 if (!mark || end == line + 1
1995 || *end != ' ' || get_sha1(end + 1, sha1))
1996 die("corrupt mark line: %s", line);
1997 e = find_object(sha1);
1998 if (!e) {
1999 enum object_type type = sha1_object_info(sha1, NULL);
2000 if (type < 0)
2001 die("object not found: %s", sha1_to_hex(sha1));
2002 e = insert_object(sha1);
2003 e->type = type;
2004 e->pack_id = MAX_PACK_ID;
a5c1780a 2005 e->offset = 1; /* just not zero! */
e8438420
SP
2006 }
2007 insert_mark(mark, e);
2008 }
2009 fclose(f);
2010}
2011
d5c57b28 2012static const char fast_import_usage[] =
63e0c8b3 2013"git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
d5c57b28 2014
8bcce301
SP
2015int main(int argc, const char **argv)
2016{
c499d768 2017 int i, show_stats = 1;
8bcce301 2018
463acbe1 2019 git_config(git_default_config);
93e72d8d
SP
2020 alloc_objects(object_entry_alloc);
2021 strbuf_init(&command_buf);
2022 atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2023 branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2024 avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2025 marks = pool_calloc(1, sizeof(struct mark_set));
463acbe1 2026
d5c57b28
SP
2027 for (i = 1; i < argc; i++) {
2028 const char *a = argv[i];
2029
2030 if (*a != '-' || !strcmp(a, "--"))
2031 break;
cc44c765 2032 else if (!prefixcmp(a, "--date-format=")) {
63e0c8b3
SP
2033 const char *fmt = a + 14;
2034 if (!strcmp(fmt, "raw"))
2035 whenspec = WHENSPEC_RAW;
2036 else if (!strcmp(fmt, "rfc2822"))
2037 whenspec = WHENSPEC_RFC2822;
2038 else if (!strcmp(fmt, "now"))
2039 whenspec = WHENSPEC_NOW;
2040 else
2041 die("unknown --date-format argument %s", fmt);
2042 }
cc44c765 2043 else if (!prefixcmp(a, "--max-pack-size="))
0ea9f045 2044 max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
cc44c765 2045 else if (!prefixcmp(a, "--depth="))
d5c57b28 2046 max_depth = strtoul(a + 8, NULL, 0);
cc44c765 2047 else if (!prefixcmp(a, "--active-branches="))
d5c57b28 2048 max_active_branches = strtoul(a + 18, NULL, 0);
e8438420
SP
2049 else if (!prefixcmp(a, "--import-marks="))
2050 import_marks(a + 15);
cc44c765 2051 else if (!prefixcmp(a, "--export-marks="))
a6a1a831 2052 mark_file = a + 15;
cc44c765 2053 else if (!prefixcmp(a, "--export-pack-edges=")) {
bdf1c06d
SP
2054 if (pack_edges)
2055 fclose(pack_edges);
2056 pack_edges = fopen(a + 20, "a");
2057 if (!pack_edges)
2058 die("Cannot open %s: %s", a + 20, strerror(errno));
2059 } else if (!strcmp(a, "--force"))
7073e69e 2060 force_update = 1;
c499d768
SP
2061 else if (!strcmp(a, "--quiet"))
2062 show_stats = 0;
2063 else if (!strcmp(a, "--stats"))
2064 show_stats = 1;
d5c57b28
SP
2065 else
2066 die("unknown option %s", a);
2067 }
8455e484 2068 if (i != argc)
d5c57b28 2069 usage(fast_import_usage);
d5c57b28 2070
a5c1780a 2071 prepare_packed_git();
f70b6534 2072 start_packfile();
db5e523f 2073 for (;;) {
c44cdc7e
SP
2074 read_next_command();
2075 if (command_buf.eof)
db5e523f 2076 break;
c44cdc7e
SP
2077 else if (!strcmp("blob", command_buf.buf))
2078 cmd_new_blob();
599065a3 2079 else if (!prefixcmp(command_buf.buf, "commit "))
c44cdc7e 2080 cmd_new_commit();
599065a3 2081 else if (!prefixcmp(command_buf.buf, "tag "))
72303d44 2082 cmd_new_tag();
599065a3 2083 else if (!prefixcmp(command_buf.buf, "reset "))
5fced8dc 2084 cmd_reset_branch();
7bfe6e26
SP
2085 else if (!strcmp("checkpoint", command_buf.buf))
2086 cmd_checkpoint();
c44cdc7e
SP
2087 else
2088 die("Unsupported command: %s", command_buf.buf);
db5e523f 2089 }
f70b6534 2090 end_packfile();
c44cdc7e 2091
463acbe1 2092 dump_branches();
72303d44 2093 dump_tags();
8455e484 2094 unkeep_all_packs();
a6a1a831 2095 dump_marks();
8bcce301 2096
bdf1c06d
SP
2097 if (pack_edges)
2098 fclose(pack_edges);
2099
c499d768
SP
2100 if (show_stats) {
2101 uintmax_t total_count = 0, duplicate_count = 0;
2102 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2103 total_count += object_count_by_type[i];
2104 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2105 duplicate_count += duplicate_count_by_type[i];
2106
2107 fprintf(stderr, "%s statistics:\n", argv[0]);
2108 fprintf(stderr, "---------------------------------------------------------------------\n");
3efb1f34
JR
2109 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2110 fprintf(stderr, "Total objects: %10" PRIuMAX " (%10" PRIuMAX " duplicates )\n", total_count, duplicate_count);
2111 fprintf(stderr, " blobs : %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
2112 fprintf(stderr, " trees : %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
2113 fprintf(stderr, " commits: %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
2114 fprintf(stderr, " tags : %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
c499d768 2115 fprintf(stderr, "Total branches: %10lu (%10lu loads )\n", branch_count, branch_load_count);
3efb1f34 2116 fprintf(stderr, " marks: %10" PRIuMAX " (%10" PRIuMAX " unique )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
c499d768 2117 fprintf(stderr, " atoms: %10u\n", atom_cnt);
3efb1f34 2118 fprintf(stderr, "Memory total: %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
40db58b8 2119 fprintf(stderr, " pools: %10lu KiB\n", (unsigned long)(total_allocd/1024));
3efb1f34 2120 fprintf(stderr, " objects: %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
c499d768
SP
2121 fprintf(stderr, "---------------------------------------------------------------------\n");
2122 pack_report();
2123 fprintf(stderr, "---------------------------------------------------------------------\n");
2124 fprintf(stderr, "\n");
2125 }
db5e523f 2126
7073e69e 2127 return failure ? 1 : 0;
db5e523f 2128}