?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'master' of git://git.kernel.org/pub/scm/git/git
[git/jnareb-git.git] / fast-import.c
blob404d911390bdb4700d0ee0df46bd7849c1890016
1 /*
2 Format of STDIN stream:
3
4 stream ::= cmd*;
5
6 cmd ::= new_blob
7 | new_commit
8 | new_tag
9 | reset_branch
10 | checkpoint
11 ;
12
13 new_blob ::= 'blob' lf
14 mark?
15 file_content;
16 file_content ::= data;
17
18 new_commit ::= 'commit' sp ref_str lf
19 mark?
20 ('author' sp name '<' email '>' when lf)?
21 'committer' sp name '<' email '>' when lf
22 commit_msg
23 ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
24 ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
25 file_change*
26 lf;
27 commit_msg ::= data;
28
29 file_change ::= file_clr | file_del | file_obm | file_inm;
30 file_clr ::= 'deleteall' lf;
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;
35
36 new_tag ::= 'tag' sp tag_str lf
37 'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
38 'tagger' sp name '<' email '>' when lf
39 tag_msg;
40 tag_msg ::= data;
41
42 reset_branch ::= 'reset' sp ref_str lf
43 ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
44 lf;
45
46 checkpoint ::= 'checkpoint' lf
47 lf;
48
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;
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;
65
66 # note: declen indicates the length of binary_data in bytes.
67 # declen does not include the lf preceeding the binary data.
68 #
69 exact_data ::= 'data' sp declen lf
70 binary_data;
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 #
79 ref_str ::= ref;
80 sha1exp_str ::= sha1exp;
81 tag_str ::= tag;
82 path_str ::= path | '"' quoted(path) '"' ;
83 mode ::= '100644' | '644'
84 | '100755' | '755'
85 | '120000'
86 ;
87
88 declen ::= # unsigned 32 bit value, ascii base10 notation;
89 bigint ::= # unsigned integer value, ascii base10 notation;
90 binary_data ::= # file content, not interpreted;
91
92 when ::= raw_when | rfc2822_when;
93 raw_when ::= ts sp tz;
94 rfc2822_when ::= # Valid RFC 2822 date and time;
95
96 sp ::= # ASCII space character;
97 lf ::= # ASCII newline (LF) character;
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 #
103 idnum ::= ':' bigint;
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";
107 sha1exp ::= # Any valid GIT SHA1 expression;
108 hexsha1 ::= # SHA1 in hexadecimal format;
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;
115 email ::= # valid GIT author/committer email;
116 ts ::= # time since the epoch in seconds, ascii base10 notation;
117 tz ::= # GIT style timezone;
118 */
119
120 #include "builtin.h"
121 #include "cache.h"
122 #include "object.h"
123 #include "blob.h"
124 #include "tree.h"
125 #include "commit.h"
126 #include "delta.h"
127 #include "pack.h"
128 #include "refs.h"
129 #include "csum-file.h"
130 #include "strbuf.h"
131 #include "quote.h"
132
133 #define PACK_ID_BITS 16
134 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
135
136 struct object_entry
137 {
138 struct object_entry *next;
139 uint32_t offset;
140 unsigned type : TYPE_BITS;
141 unsigned pack_id : PACK_ID_BITS;
142 unsigned char sha1[20];
143 };
144
145 struct object_entry_pool
146 {
147 struct object_entry_pool *next_pool;
148 struct object_entry *next_free;
149 struct object_entry *end;
150 struct object_entry entries[FLEX_ARRAY]; /* more */
151 };
152
153 struct mark_set
154 {
155 union {
156 struct object_entry *marked[1024];
157 struct mark_set *sets[1024];
158 } data;
159 unsigned int shift;
160 };
161
162 struct last_object
163 {
164 void *data;
165 unsigned long len;
166 uint32_t offset;
167 unsigned int depth;
168 unsigned no_free:1;
169 };
170
171 struct mem_pool
172 {
173 struct mem_pool *next_pool;
174 char *next_free;
175 char *end;
176 char space[FLEX_ARRAY]; /* more */
177 };
178
179 struct atom_str
180 {
181 struct atom_str *next_atom;
182 unsigned short str_len;
183 char str_dat[FLEX_ARRAY]; /* more */
184 };
185
186 struct tree_content;
187 struct tree_entry
188 {
189 struct tree_content *tree;
190 struct atom_str* name;
191 struct tree_entry_ms
192 {
193 uint16_t mode;
194 unsigned char sha1[20];
195 } versions[2];
196 };
197
198 struct tree_content
199 {
200 unsigned int entry_capacity; /* must match avail_tree_content */
201 unsigned int entry_count;
202 unsigned int delta_depth;
203 struct tree_entry *entries[FLEX_ARRAY]; /* more */
204 };
205
206 struct avail_tree_content
207 {
208 unsigned int entry_capacity; /* must match tree_content */
209 struct avail_tree_content *next_avail;
210 };
211
212 struct branch
213 {
214 struct branch *table_next_branch;
215 struct branch *active_next_branch;
216 const char *name;
217 struct tree_entry branch_tree;
218 uintmax_t last_commit;
219 unsigned int pack_id;
220 unsigned char sha1[20];
221 };
222
223 struct tag
224 {
225 struct tag *next_tag;
226 const char *name;
227 unsigned int pack_id;
228 unsigned char sha1[20];
229 };
230
231 struct dbuf
232 {
233 void *buffer;
234 size_t capacity;
235 };
236
237 struct hash_list
238 {
239 struct hash_list *next;
240 unsigned char sha1[20];
241 };
242
243 typedef enum {
244 WHENSPEC_RAW = 1,
245 WHENSPEC_RFC2822,
246 WHENSPEC_NOW,
247 } whenspec_type;
248
249 /* Configured limits on output */
250 static unsigned long max_depth = 10;
251 static unsigned long max_packsize = (1LL << 32) - 1;
252 static int force_update;
253
254 /* Stats and misc. counters */
255 static uintmax_t alloc_count;
256 static uintmax_t marks_set_count;
257 static uintmax_t object_count_by_type[1 << TYPE_BITS];
258 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
259 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
260 static unsigned long object_count;
261 static unsigned long branch_count;
262 static unsigned long branch_load_count;
263 static int failure;
264 static FILE *pack_edges;
265
266 /* Memory pools */
267 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
268 static size_t total_allocd;
269 static struct mem_pool *mem_pool;
270
271 /* Atom management */
272 static unsigned int atom_table_sz = 4451;
273 static unsigned int atom_cnt;
274 static struct atom_str **atom_table;
275
276 /* The .pack file being generated */
277 static unsigned int pack_id;
278 static struct packed_git *pack_data;
279 static struct packed_git **all_packs;
280 static unsigned long pack_size;
281
282 /* Table of objects we've written. */
283 static unsigned int object_entry_alloc = 5000;
284 static struct object_entry_pool *blocks;
285 static struct object_entry *object_table[1 << 16];
286 static struct mark_set *marks;
287 static const char* mark_file;
288
289 /* Our last blob */
290 static struct last_object last_blob;
291
292 /* Tree management */
293 static unsigned int tree_entry_alloc = 1000;
294 static void *avail_tree_entry;
295 static unsigned int avail_tree_table_sz = 100;
296 static struct avail_tree_content **avail_tree_table;
297 static struct dbuf old_tree;
298 static struct dbuf new_tree;
299
300 /* Branch data */
301 static unsigned long max_active_branches = 5;
302 static unsigned long cur_active_branches;
303 static unsigned long branch_table_sz = 1039;
304 static struct branch **branch_table;
305 static struct branch *active_branches;
306
307 /* Tag data */
308 static struct tag *first_tag;
309 static struct tag *last_tag;
310
311 /* Input stream parsing */
312 static whenspec_type whenspec = WHENSPEC_RAW;
313 static struct strbuf command_buf;
314 static uintmax_t next_mark;
315 static struct dbuf new_data;
316
317
318 static void alloc_objects(unsigned int cnt)
319 {
320 struct object_entry_pool *b;
321
322 b = xmalloc(sizeof(struct object_entry_pool)
323 + cnt * sizeof(struct object_entry));
324 b->next_pool = blocks;
325 b->next_free = b->entries;
326 b->end = b->entries + cnt;
327 blocks = b;
328 alloc_count += cnt;
329 }
330
331 static struct object_entry *new_object(unsigned char *sha1)
332 {
333 struct object_entry *e;
334
335 if (blocks->next_free == blocks->end)
336 alloc_objects(object_entry_alloc);
337
338 e = blocks->next_free++;
339 hashcpy(e->sha1, sha1);
340 return e;
341 }
342
343 static struct object_entry *find_object(unsigned char *sha1)
344 {
345 unsigned int h = sha1[0] << 8 | sha1[1];
346 struct object_entry *e;
347 for (e = object_table[h]; e; e = e->next)
348 if (!hashcmp(sha1, e->sha1))
349 return e;
350 return NULL;
351 }
352
353 static struct object_entry *insert_object(unsigned char *sha1)
354 {
355 unsigned int h = sha1[0] << 8 | sha1[1];
356 struct object_entry *e = object_table[h];
357 struct object_entry *p = NULL;
358
359 while (e) {
360 if (!hashcmp(sha1, e->sha1))
361 return e;
362 p = e;
363 e = e->next;
364 }
365
366 e = new_object(sha1);
367 e->next = NULL;
368 e->offset = 0;
369 if (p)
370 p->next = e;
371 else
372 object_table[h] = e;
373 return e;
374 }
375
376 static unsigned int hc_str(const char *s, size_t len)
377 {
378 unsigned int r = 0;
379 while (len-- > 0)
380 r = r * 31 + *s++;
381 return r;
382 }
383
384 static void *pool_alloc(size_t len)
385 {
386 struct mem_pool *p;
387 void *r;
388
389 for (p = mem_pool; p; p = p->next_pool)
390 if ((p->end - p->next_free >= len))
391 break;
392
393 if (!p) {
394 if (len >= (mem_pool_alloc/2)) {
395 total_allocd += len;
396 return xmalloc(len);
397 }
398 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
399 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
400 p->next_pool = mem_pool;
401 p->next_free = p->space;
402 p->end = p->next_free + mem_pool_alloc;
403 mem_pool = p;
404 }
405
406 r = p->next_free;
407 /* round out to a pointer alignment */
408 if (len & (sizeof(void*) - 1))
409 len += sizeof(void*) - (len & (sizeof(void*) - 1));
410 p->next_free += len;
411 return r;
412 }
413
414 static void *pool_calloc(size_t count, size_t size)
415 {
416 size_t len = count * size;
417 void *r = pool_alloc(len);
418 memset(r, 0, len);
419 return r;
420 }
421
422 static char *pool_strdup(const char *s)
423 {
424 char *r = pool_alloc(strlen(s) + 1);
425 strcpy(r, s);
426 return r;
427 }
428
429 static void size_dbuf(struct dbuf *b, size_t maxlen)
430 {
431 if (b->buffer) {
432 if (b->capacity >= maxlen)
433 return;
434 free(b->buffer);
435 }
436 b->capacity = ((maxlen / 1024) + 1) * 1024;
437 b->buffer = xmalloc(b->capacity);
438 }
439
440 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
441 {
442 struct mark_set *s = marks;
443 while ((idnum >> s->shift) >= 1024) {
444 s = pool_calloc(1, sizeof(struct mark_set));
445 s->shift = marks->shift + 10;
446 s->data.sets[0] = marks;
447 marks = s;
448 }
449 while (s->shift) {
450 uintmax_t i = idnum >> s->shift;
451 idnum -= i << s->shift;
452 if (!s->data.sets[i]) {
453 s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
454 s->data.sets[i]->shift = s->shift - 10;
455 }
456 s = s->data.sets[i];
457 }
458 if (!s->data.marked[idnum])
459 marks_set_count++;
460 s->data.marked[idnum] = oe;
461 }
462
463 static struct object_entry *find_mark(uintmax_t idnum)
464 {
465 uintmax_t orig_idnum = idnum;
466 struct mark_set *s = marks;
467 struct object_entry *oe = NULL;
468 if ((idnum >> s->shift) < 1024) {
469 while (s && s->shift) {
470 uintmax_t i = idnum >> s->shift;
471 idnum -= i << s->shift;
472 s = s->data.sets[i];
473 }
474 if (s)
475 oe = s->data.marked[idnum];
476 }
477 if (!oe)
478 die("mark :%ju not declared", orig_idnum);
479 return oe;
480 }
481
482 static struct atom_str *to_atom(const char *s, unsigned short len)
483 {
484 unsigned int hc = hc_str(s, len) % atom_table_sz;
485 struct atom_str *c;
486
487 for (c = atom_table[hc]; c; c = c->next_atom)
488 if (c->str_len == len && !strncmp(s, c->str_dat, len))
489 return c;
490
491 c = pool_alloc(sizeof(struct atom_str) + len + 1);
492 c->str_len = len;
493 strncpy(c->str_dat, s, len);
494 c->str_dat[len] = 0;
495 c->next_atom = atom_table[hc];
496 atom_table[hc] = c;
497 atom_cnt++;
498 return c;
499 }
500
501 static struct branch *lookup_branch(const char *name)
502 {
503 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
504 struct branch *b;
505
506 for (b = branch_table[hc]; b; b = b->table_next_branch)
507 if (!strcmp(name, b->name))
508 return b;
509 return NULL;
510 }
511
512 static struct branch *new_branch(const char *name)
513 {
514 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
515 struct branch* b = lookup_branch(name);
516
517 if (b)
518 die("Invalid attempt to create duplicate branch: %s", name);
519 if (check_ref_format(name))
520 die("Branch name doesn't conform to GIT standards: %s", name);
521
522 b = pool_calloc(1, sizeof(struct branch));
523 b->name = pool_strdup(name);
524 b->table_next_branch = branch_table[hc];
525 b->branch_tree.versions[0].mode = S_IFDIR;
526 b->branch_tree.versions[1].mode = S_IFDIR;
527 b->pack_id = MAX_PACK_ID;
528 branch_table[hc] = b;
529 branch_count++;
530 return b;
531 }
532
533 static unsigned int hc_entries(unsigned int cnt)
534 {
535 cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
536 return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
537 }
538
539 static struct tree_content *new_tree_content(unsigned int cnt)
540 {
541 struct avail_tree_content *f, *l = NULL;
542 struct tree_content *t;
543 unsigned int hc = hc_entries(cnt);
544
545 for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
546 if (f->entry_capacity >= cnt)
547 break;
548
549 if (f) {
550 if (l)
551 l->next_avail = f->next_avail;
552 else
553 avail_tree_table[hc] = f->next_avail;
554 } else {
555 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
556 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
557 f->entry_capacity = cnt;
558 }
559
560 t = (struct tree_content*)f;
561 t->entry_count = 0;
562 t->delta_depth = 0;
563 return t;
564 }
565
566 static void release_tree_entry(struct tree_entry *e);
567 static void release_tree_content(struct tree_content *t)
568 {
569 struct avail_tree_content *f = (struct avail_tree_content*)t;
570 unsigned int hc = hc_entries(f->entry_capacity);
571 f->next_avail = avail_tree_table[hc];
572 avail_tree_table[hc] = f;
573 }
574
575 static void release_tree_content_recursive(struct tree_content *t)
576 {
577 unsigned int i;
578 for (i = 0; i < t->entry_count; i++)
579 release_tree_entry(t->entries[i]);
580 release_tree_content(t);
581 }
582
583 static struct tree_content *grow_tree_content(
584 struct tree_content *t,
585 int amt)
586 {
587 struct tree_content *r = new_tree_content(t->entry_count + amt);
588 r->entry_count = t->entry_count;
589 r->delta_depth = t->delta_depth;
590 memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
591 release_tree_content(t);
592 return r;
593 }
594
595 static struct tree_entry *new_tree_entry(void)
596 {
597 struct tree_entry *e;
598
599 if (!avail_tree_entry) {
600 unsigned int n = tree_entry_alloc;
601 total_allocd += n * sizeof(struct tree_entry);
602 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
603 while (n-- > 1) {
604 *((void**)e) = e + 1;
605 e++;
606 }
607 *((void**)e) = NULL;
608 }
609
610 e = avail_tree_entry;
611 avail_tree_entry = *((void**)e);
612 return e;
613 }
614
615 static void release_tree_entry(struct tree_entry *e)
616 {
617 if (e->tree)
618 release_tree_content_recursive(e->tree);
619 *((void**)e) = avail_tree_entry;
620 avail_tree_entry = e;
621 }
622
623 static void start_packfile(void)
624 {
625 static char tmpfile[PATH_MAX];
626 struct packed_git *p;
627 struct pack_header hdr;
628 int pack_fd;
629
630 snprintf(tmpfile, sizeof(tmpfile),
631 "%s/pack_XXXXXX", get_object_directory());
632 pack_fd = mkstemp(tmpfile);
633 if (pack_fd < 0)
634 die("Can't create %s: %s", tmpfile, strerror(errno));
635 p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
636 strcpy(p->pack_name, tmpfile);
637 p->pack_fd = pack_fd;
638
639 hdr.hdr_signature = htonl(PACK_SIGNATURE);
640 hdr.hdr_version = htonl(2);
641 hdr.hdr_entries = 0;
642 write_or_die(p->pack_fd, &hdr, sizeof(hdr));
643
644 pack_data = p;
645 pack_size = sizeof(hdr);
646 object_count = 0;
647
648 all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
649 all_packs[pack_id] = p;
650 }
651
652 static void fixup_header_footer(void)
653 {
654 static const int buf_sz = 128 * 1024;
655 int pack_fd = pack_data->pack_fd;
656 SHA_CTX c;
657 struct pack_header hdr;
658 char *buf;
659
660 if (lseek(pack_fd, 0, SEEK_SET) != 0)
661 die("Failed seeking to start: %s", strerror(errno));
662 if (read_in_full(pack_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
663 die("Unable to reread header of %s", pack_data->pack_name);
664 if (lseek(pack_fd, 0, SEEK_SET) != 0)
665 die("Failed seeking to start: %s", strerror(errno));
666 hdr.hdr_entries = htonl(object_count);
667 write_or_die(pack_fd, &hdr, sizeof(hdr));
668
669 SHA1_Init(&c);
670 SHA1_Update(&c, &hdr, sizeof(hdr));
671
672 buf = xmalloc(buf_sz);
673 for (;;) {
674 size_t n = xread(pack_fd, buf, buf_sz);
675 if (!n)
676 break;
677 if (n < 0)
678 die("Failed to checksum %s", pack_data->pack_name);
679 SHA1_Update(&c, buf, n);
680 }
681 free(buf);
682
683 SHA1_Final(pack_data->sha1, &c);
684 write_or_die(pack_fd, pack_data->sha1, sizeof(pack_data->sha1));
685 close(pack_fd);
686 }
687
688 static int oecmp (const void *a_, const void *b_)
689 {
690 struct object_entry *a = *((struct object_entry**)a_);
691 struct object_entry *b = *((struct object_entry**)b_);
692 return hashcmp(a->sha1, b->sha1);
693 }
694
695 static char *create_index(void)
696 {
697 static char tmpfile[PATH_MAX];
698 SHA_CTX ctx;
699 struct sha1file *f;
700 struct object_entry **idx, **c, **last, *e;
701 struct object_entry_pool *o;
702 uint32_t array[256];
703 int i, idx_fd;
704
705 /* Build the sorted table of object IDs. */
706 idx = xmalloc(object_count * sizeof(struct object_entry*));
707 c = idx;
708 for (o = blocks; o; o = o->next_pool)
709 for (e = o->next_free; e-- != o->entries;)
710 if (pack_id == e->pack_id)
711 *c++ = e;
712 last = idx + object_count;
713 if (c != last)
714 die("internal consistency error creating the index");
715 qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
716
717 /* Generate the fan-out array. */
718 c = idx;
719 for (i = 0; i < 256; i++) {
720 struct object_entry **next = c;;
721 while (next < last) {
722 if ((*next)->sha1[0] != i)
723 break;
724 next++;
725 }
726 array[i] = htonl(next - idx);
727 c = next;
728 }
729
730 snprintf(tmpfile, sizeof(tmpfile),
731 "%s/index_XXXXXX", get_object_directory());
732 idx_fd = mkstemp(tmpfile);
733 if (idx_fd < 0)
734 die("Can't create %s: %s", tmpfile, strerror(errno));
735 f = sha1fd(idx_fd, tmpfile);
736 sha1write(f, array, 256 * sizeof(int));
737 SHA1_Init(&ctx);
738 for (c = idx; c != last; c++) {
739 uint32_t offset = htonl((*c)->offset);
740 sha1write(f, &offset, 4);
741 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
742 SHA1_Update(&ctx, (*c)->sha1, 20);
743 }
744 sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
745 sha1close(f, NULL, 1);
746 free(idx);
747 SHA1_Final(pack_data->sha1, &ctx);
748 return tmpfile;
749 }
750
751 static char *keep_pack(char *curr_index_name)
752 {
753 static char name[PATH_MAX];
754 static char *keep_msg = "fast-import";
755 int keep_fd;
756
757 chmod(pack_data->pack_name, 0444);
758 chmod(curr_index_name, 0444);
759
760 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
761 get_object_directory(), sha1_to_hex(pack_data->sha1));
762 keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
763 if (keep_fd < 0)
764 die("cannot create keep file");
765 write(keep_fd, keep_msg, strlen(keep_msg));
766 close(keep_fd);
767
768 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
769 get_object_directory(), sha1_to_hex(pack_data->sha1));
770 if (move_temp_to_file(pack_data->pack_name, name))
771 die("cannot store pack file");
772
773 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
774 get_object_directory(), sha1_to_hex(pack_data->sha1));
775 if (move_temp_to_file(curr_index_name, name))
776 die("cannot store index file");
777 return name;
778 }
779
780 static void unkeep_all_packs(void)
781 {
782 static char name[PATH_MAX];
783 int k;
784
785 for (k = 0; k < pack_id; k++) {
786 struct packed_git *p = all_packs[k];
787 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
788 get_object_directory(), sha1_to_hex(p->sha1));
789 unlink(name);
790 }
791 }
792
793 static void end_packfile(void)
794 {
795 struct packed_git *old_p = pack_data, *new_p;
796
797 if (object_count) {
798 char *idx_name;
799 int i;
800 struct branch *b;
801 struct tag *t;
802
803 fixup_header_footer();
804 idx_name = keep_pack(create_index());
805
806 /* Register the packfile with core git's machinary. */
807 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
808 if (!new_p)
809 die("core git rejected index %s", idx_name);
810 new_p->windows = old_p->windows;
811 all_packs[pack_id] = new_p;
812 install_packed_git(new_p);
813
814 /* Print the boundary */
815 if (pack_edges) {
816 fprintf(pack_edges, "%s:", new_p->pack_name);
817 for (i = 0; i < branch_table_sz; i++) {
818 for (b = branch_table[i]; b; b = b->table_next_branch) {
819 if (b->pack_id == pack_id)
820 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
821 }
822 }
823 for (t = first_tag; t; t = t->next_tag) {
824 if (t->pack_id == pack_id)
825 fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
826 }
827 fputc('\n', pack_edges);
828 fflush(pack_edges);
829 }
830
831 pack_id++;
832 }
833 else
834 unlink(old_p->pack_name);
835 free(old_p);
836
837 /* We can't carry a delta across packfiles. */
838 free(last_blob.data);
839 last_blob.data = NULL;
840 last_blob.len = 0;
841 last_blob.offset = 0;
842 last_blob.depth = 0;
843 }
844
845 static void cycle_packfile(void)
846 {
847 end_packfile();
848 start_packfile();
849 }
850
851 static size_t encode_header(
852 enum object_type type,
853 size_t size,
854 unsigned char *hdr)
855 {
856 int n = 1;
857 unsigned char c;
858
859 if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
860 die("bad type %d", type);
861
862 c = (type << 4) | (size & 15);
863 size >>= 4;
864 while (size) {
865 *hdr++ = c | 0x80;
866 c = size & 0x7f;
867 size >>= 7;
868 n++;
869 }
870 *hdr = c;
871 return n;
872 }
873
874 static int store_object(
875 enum object_type type,
876 void *dat,
877 size_t datlen,
878 struct last_object *last,
879 unsigned char *sha1out,
880 uintmax_t mark)
881 {
882 void *out, *delta;
883 struct object_entry *e;
884 unsigned char hdr[96];
885 unsigned char sha1[20];
886 unsigned long hdrlen, deltalen;
887 SHA_CTX c;
888 z_stream s;
889
890 hdrlen = sprintf((char*)hdr,"%s %lu", type_names[type],
891 (unsigned long)datlen) + 1;
892 SHA1_Init(&c);
893 SHA1_Update(&c, hdr, hdrlen);
894 SHA1_Update(&c, dat, datlen);
895 SHA1_Final(sha1, &c);
896 if (sha1out)
897 hashcpy(sha1out, sha1);
898
899 e = insert_object(sha1);
900 if (mark)
901 insert_mark(mark, e);
902 if (e->offset) {
903 duplicate_count_by_type[type]++;
904 return 1;
905 }
906
907 if (last && last->data && last->depth < max_depth) {
908 delta = diff_delta(last->data, last->len,
909 dat, datlen,
910 &deltalen, 0);
911 if (delta && deltalen >= datlen) {
912 free(delta);
913 delta = NULL;
914 }
915 } else
916 delta = NULL;
917
918 memset(&s, 0, sizeof(s));
919 deflateInit(&s, zlib_compression_level);
920 if (delta) {
921 s.next_in = delta;
922 s.avail_in = deltalen;
923 } else {
924 s.next_in = dat;
925 s.avail_in = datlen;
926 }
927 s.avail_out = deflateBound(&s, s.avail_in);
928 s.next_out = out = xmalloc(s.avail_out);
929 while (deflate(&s, Z_FINISH) == Z_OK)
930 /* nothing */;
931 deflateEnd(&s);
932
933 /* Determine if we should auto-checkpoint. */
934 if ((pack_size + 60 + s.total_out) > max_packsize
935 || (pack_size + 60 + s.total_out) < pack_size) {
936
937 /* This new object needs to *not* have the current pack_id. */
938 e->pack_id = pack_id + 1;
939 cycle_packfile();
940
941 /* We cannot carry a delta into the new pack. */
942 if (delta) {
943 free(delta);
944 delta = NULL;
945
946 memset(&s, 0, sizeof(s));
947 deflateInit(&s, zlib_compression_level);
948 s.next_in = dat;
949 s.avail_in = datlen;
950 s.avail_out = deflateBound(&s, s.avail_in);
951 s.next_out = out = xrealloc(out, s.avail_out);
952 while (deflate(&s, Z_FINISH) == Z_OK)
953 /* nothing */;
954 deflateEnd(&s);
955 }
956 }
957
958 e->type = type;
959 e->pack_id = pack_id;
960 e->offset = pack_size;
961 object_count++;
962 object_count_by_type[type]++;
963
964 if (delta) {
965 unsigned long ofs = e->offset - last->offset;
966 unsigned pos = sizeof(hdr) - 1;
967
968 delta_count_by_type[type]++;
969 last->depth++;
970
971 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
972 write_or_die(pack_data->pack_fd, hdr, hdrlen);
973 pack_size += hdrlen;
974
975 hdr[pos] = ofs & 127;
976 while (ofs >>= 7)
977 hdr[--pos] = 128 | (--ofs & 127);
978 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
979 pack_size += sizeof(hdr) - pos;
980 } else {
981 if (last)
982 last->depth = 0;
983 hdrlen = encode_header(type, datlen, hdr);
984 write_or_die(pack_data->pack_fd, hdr, hdrlen);
985 pack_size += hdrlen;
986 }
987
988 write_or_die(pack_data->pack_fd, out, s.total_out);
989 pack_size += s.total_out;
990
991 free(out);
992 free(delta);
993 if (last) {
994 if (!last->no_free)
995 free(last->data);
996 last->data = dat;
997 last->offset = e->offset;
998 last->len = datlen;
999 }
1000 return 0;
1001 }
1002
1003 static void *gfi_unpack_entry(
1004 struct object_entry *oe,
1005 unsigned long *sizep)
1006 {
1007 static char type[20];
1008 struct packed_git *p = all_packs[oe->pack_id];
1009 if (p == pack_data)
1010 p->pack_size = pack_size + 20;
1011 return unpack_entry(p, oe->offset, type, sizep);
1012 }
1013
1014 static const char *get_mode(const char *str, uint16_t *modep)
1015 {
1016 unsigned char c;
1017 uint16_t mode = 0;
1018
1019 while ((c = *str++) != ' ') {
1020 if (c < '0' || c > '7')
1021 return NULL;
1022 mode = (mode << 3) + (c - '0');
1023 }
1024 *modep = mode;
1025 return str;
1026 }
1027
1028 static void load_tree(struct tree_entry *root)
1029 {
1030 unsigned char* sha1 = root->versions[1].sha1;
1031 struct object_entry *myoe;
1032 struct tree_content *t;
1033 unsigned long size;
1034 char *buf;
1035 const char *c;
1036
1037 root->tree = t = new_tree_content(8);
1038 if (is_null_sha1(sha1))
1039 return;
1040
1041 myoe = find_object(sha1);
1042 if (myoe) {
1043 if (myoe->type != OBJ_TREE)
1044 die("Not a tree: %s", sha1_to_hex(sha1));
1045 t->delta_depth = 0;
1046 buf = gfi_unpack_entry(myoe, &size);
1047 } else {
1048 char type[20];
1049 buf = read_sha1_file(sha1, type, &size);
1050 if (!buf || strcmp(type, tree_type))
1051 die("Can't load tree %s", sha1_to_hex(sha1));
1052 }
1053
1054 c = buf;
1055 while (c != (buf + size)) {
1056 struct tree_entry *e = new_tree_entry();
1057
1058 if (t->entry_count == t->entry_capacity)
1059 root->tree = t = grow_tree_content(t, 8);
1060 t->entries[t->entry_count++] = e;
1061
1062 e->tree = NULL;
1063 c = get_mode(c, &e->versions[1].mode);
1064 if (!c)
1065 die("Corrupt mode in %s", sha1_to_hex(sha1));
1066 e->versions[0].mode = e->versions[1].mode;
1067 e->name = to_atom(c, (unsigned short)strlen(c));
1068 c += e->name->str_len + 1;
1069 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1070 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1071 c += 20;
1072 }
1073 free(buf);
1074 }
1075
1076 static int tecmp0 (const void *_a, const void *_b)
1077 {
1078 struct tree_entry *a = *((struct tree_entry**)_a);
1079 struct tree_entry *b = *((struct tree_entry**)_b);
1080 return base_name_compare(
1081 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1082 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1083 }
1084
1085 static int tecmp1 (const void *_a, const void *_b)
1086 {
1087 struct tree_entry *a = *((struct tree_entry**)_a);
1088 struct tree_entry *b = *((struct tree_entry**)_b);
1089 return base_name_compare(
1090 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1091 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1092 }
1093
1094 static void mktree(struct tree_content *t,
1095 int v,
1096 unsigned long *szp,
1097 struct dbuf *b)
1098 {
1099 size_t maxlen = 0;
1100 unsigned int i;
1101 char *c;
1102
1103 if (!v)
1104 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1105 else
1106 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1107
1108 for (i = 0; i < t->entry_count; i++) {
1109 if (t->entries[i]->versions[v].mode)
1110 maxlen += t->entries[i]->name->str_len + 34;
1111 }
1112
1113 size_dbuf(b, maxlen);
1114 c = b->buffer;
1115 for (i = 0; i < t->entry_count; i++) {
1116 struct tree_entry *e = t->entries[i];
1117 if (!e->versions[v].mode)
1118 continue;
1119 c += sprintf(c, "%o", (unsigned int)e->versions[v].mode);
1120 *c++ = ' ';
1121 strcpy(c, e->name->str_dat);
1122 c += e->name->str_len + 1;
1123 hashcpy((unsigned char*)c, e->versions[v].sha1);
1124 c += 20;
1125 }
1126 *szp = c - (char*)b->buffer;
1127 }
1128
1129 static void store_tree(struct tree_entry *root)
1130 {
1131 struct tree_content *t = root->tree;
1132 unsigned int i, j, del;
1133 unsigned long new_len;
1134 struct last_object lo;
1135 struct object_entry *le;
1136
1137 if (!is_null_sha1(root->versions[1].sha1))
1138 return;
1139
1140 for (i = 0; i < t->entry_count; i++) {
1141 if (t->entries[i]->tree)
1142 store_tree(t->entries[i]);
1143 }
1144
1145 le = find_object(root->versions[0].sha1);
1146 if (!S_ISDIR(root->versions[0].mode)
1147 || !le
1148 || le->pack_id != pack_id) {
1149 lo.data = NULL;
1150 lo.depth = 0;
1151 } else {
1152 mktree(t, 0, &lo.len, &old_tree);
1153 lo.data = old_tree.buffer;
1154 lo.offset = le->offset;
1155 lo.depth = t->delta_depth;
1156 lo.no_free = 1;
1157 }
1158
1159 mktree(t, 1, &new_len, &new_tree);
1160 store_object(OBJ_TREE, new_tree.buffer, new_len,
1161 &lo, root->versions[1].sha1, 0);
1162
1163 t->delta_depth = lo.depth;
1164 for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1165 struct tree_entry *e = t->entries[i];
1166 if (e->versions[1].mode) {
1167 e->versions[0].mode = e->versions[1].mode;
1168 hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1169 t->entries[j++] = e;
1170 } else {
1171 release_tree_entry(e);
1172 del++;
1173 }
1174 }
1175 t->entry_count -= del;
1176 }
1177
1178 static int tree_content_set(
1179 struct tree_entry *root,
1180 const char *p,
1181 const unsigned char *sha1,
1182 const uint16_t mode)
1183 {
1184 struct tree_content *t = root->tree;
1185 const char *slash1;
1186 unsigned int i, n;
1187 struct tree_entry *e;
1188
1189 slash1 = strchr(p, '/');
1190 if (slash1)
1191 n = slash1 - p;
1192 else
1193 n = strlen(p);
1194
1195 for (i = 0; i < t->entry_count; i++) {
1196 e = t->entries[i];
1197 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1198 if (!slash1) {
1199 if (e->versions[1].mode == mode
1200 && !hashcmp(e->versions[1].sha1, sha1))
1201 return 0;
1202 e->versions[1].mode = mode;
1203 hashcpy(e->versions[1].sha1, sha1);
1204 if (e->tree) {
1205 release_tree_content_recursive(e->tree);
1206 e->tree = NULL;
1207 }
1208 hashclr(root->versions[1].sha1);
1209 return 1;
1210 }
1211 if (!S_ISDIR(e->versions[1].mode)) {
1212 e->tree = new_tree_content(8);
1213 e->versions[1].mode = S_IFDIR;
1214 }
1215 if (!e->tree)
1216 load_tree(e);
1217 if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1218 hashclr(root->versions[1].sha1);
1219 return 1;
1220 }
1221 return 0;
1222 }
1223 }
1224
1225 if (t->entry_count == t->entry_capacity)
1226 root->tree = t = grow_tree_content(t, 8);
1227 e = new_tree_entry();
1228 e->name = to_atom(p, (unsigned short)n);
1229 e->versions[0].mode = 0;
1230 hashclr(e->versions[0].sha1);
1231 t->entries[t->entry_count++] = e;
1232 if (slash1) {
1233 e->tree = new_tree_content(8);
1234 e->versions[1].mode = S_IFDIR;
1235 tree_content_set(e, slash1 + 1, sha1, mode);
1236 } else {
1237 e->tree = NULL;
1238 e->versions[1].mode = mode;
1239 hashcpy(e->versions[1].sha1, sha1);
1240 }
1241 hashclr(root->versions[1].sha1);
1242 return 1;
1243 }
1244
1245 static int tree_content_remove(struct tree_entry *root, const char *p)
1246 {
1247 struct tree_content *t = root->tree;
1248 const char *slash1;
1249 unsigned int i, n;
1250 struct tree_entry *e;
1251
1252 slash1 = strchr(p, '/');
1253 if (slash1)
1254 n = slash1 - p;
1255 else
1256 n = strlen(p);
1257
1258 for (i = 0; i < t->entry_count; i++) {
1259 e = t->entries[i];
1260 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1261 if (!slash1 || !S_ISDIR(e->versions[1].mode))
1262 goto del_entry;
1263 if (!e->tree)
1264 load_tree(e);
1265 if (tree_content_remove(e, slash1 + 1)) {
1266 for (n = 0; n < e->tree->entry_count; n++) {
1267 if (e->tree->entries[n]->versions[1].mode) {
1268 hashclr(root->versions[1].sha1);
1269 return 1;
1270 }
1271 }
1272 goto del_entry;
1273 }
1274 return 0;
1275 }
1276 }
1277 return 0;
1278
1279 del_entry:
1280 if (e->tree) {
1281 release_tree_content_recursive(e->tree);
1282 e->tree = NULL;
1283 }
1284 e->versions[1].mode = 0;
1285 hashclr(e->versions[1].sha1);
1286 hashclr(root->versions[1].sha1);
1287 return 1;
1288 }
1289
1290 static int update_branch(struct branch *b)
1291 {
1292 static const char *msg = "fast-import";
1293 struct ref_lock *lock;
1294 unsigned char old_sha1[20];
1295
1296 if (read_ref(b->name, old_sha1))
1297 hashclr(old_sha1);
1298 lock = lock_any_ref_for_update(b->name, old_sha1);
1299 if (!lock)
1300 return error("Unable to lock %s", b->name);
1301 if (!force_update && !is_null_sha1(old_sha1)) {
1302 struct commit *old_cmit, *new_cmit;
1303
1304 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1305 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1306 if (!old_cmit || !new_cmit) {
1307 unlock_ref(lock);
1308 return error("Branch %s is missing commits.", b->name);
1309 }
1310
1311 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1312 unlock_ref(lock);
1313 warn("Not updating %s"
1314 " (new tip %s does not contain %s)",
1315 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1316 return -1;
1317 }
1318 }
1319 if (write_ref_sha1(lock, b->sha1, msg) < 0)
1320 return error("Unable to update %s", b->name);
1321 return 0;
1322 }
1323
1324 static void dump_branches(void)
1325 {
1326 unsigned int i;
1327 struct branch *b;
1328
1329 for (i = 0; i < branch_table_sz; i++) {
1330 for (b = branch_table[i]; b; b = b->table_next_branch)
1331 failure |= update_branch(b);
1332 }
1333 }
1334
1335 static void dump_tags(void)
1336 {
1337 static const char *msg = "fast-import";
1338 struct tag *t;
1339 struct ref_lock *lock;
1340 char ref_name[PATH_MAX];
1341
1342 for (t = first_tag; t; t = t->next_tag) {
1343 sprintf(ref_name, "tags/%s", t->name);
1344 lock = lock_ref_sha1(ref_name, NULL);
1345 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1346 failure |= error("Unable to update %s", ref_name);
1347 }
1348 }
1349
1350 static void dump_marks_helper(FILE *f,
1351 uintmax_t base,
1352 struct mark_set *m)
1353 {
1354 uintmax_t k;
1355 if (m->shift) {
1356 for (k = 0; k < 1024; k++) {
1357 if (m->data.sets[k])
1358 dump_marks_helper(f, (base + k) << m->shift,
1359 m->data.sets[k]);
1360 }
1361 } else {
1362 for (k = 0; k < 1024; k++) {
1363 if (m->data.marked[k])
1364 fprintf(f, ":%ju %s\n", base + k,
1365 sha1_to_hex(m->data.marked[k]->sha1));
1366 }
1367 }
1368 }
1369
1370 static void dump_marks(void)
1371 {
1372 if (mark_file)
1373 {
1374 FILE *f = fopen(mark_file, "w");
1375 if (f) {
1376 dump_marks_helper(f, 0, marks);
1377 fclose(f);
1378 } else
1379 failure |= error("Unable to write marks file %s: %s",
1380 mark_file, strerror(errno));
1381 }
1382 }
1383
1384 static void read_next_command(void)
1385 {
1386 read_line(&command_buf, stdin, '\n');
1387 }
1388
1389 static void cmd_mark(void)
1390 {
1391 if (!strncmp("mark :", command_buf.buf, 6)) {
1392 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1393 read_next_command();
1394 }
1395 else
1396 next_mark = 0;
1397 }
1398
1399 static void *cmd_data (size_t *size)
1400 {
1401 size_t length;
1402 char *buffer;
1403
1404 if (strncmp("data ", command_buf.buf, 5))
1405 die("Expected 'data n' command, found: %s", command_buf.buf);
1406
1407 if (!strncmp("<<", command_buf.buf + 5, 2)) {
1408 char *term = xstrdup(command_buf.buf + 5 + 2);
1409 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1410 length = 0;
1411 buffer = xmalloc(sz);
1412 for (;;) {
1413 read_next_command();
1414 if (command_buf.eof)
1415 die("EOF in data (terminator '%s' not found)", term);
1416 if (term_len == command_buf.len
1417 && !strcmp(term, command_buf.buf))
1418 break;
1419 if (sz < (length + command_buf.len)) {
1420 sz = sz * 3 / 2 + 16;
1421 if (sz < (length + command_buf.len))
1422 sz = length + command_buf.len;
1423 buffer = xrealloc(buffer, sz);
1424 }
1425 memcpy(buffer + length,
1426 command_buf.buf,
1427 command_buf.len - 1);
1428 length += command_buf.len - 1;
1429 buffer[length++] = '\n';
1430 }
1431 free(term);
1432 }
1433 else {
1434 size_t n = 0;
1435 length = strtoul(command_buf.buf + 5, NULL, 10);
1436 buffer = xmalloc(length);
1437 while (n < length) {
1438 size_t s = fread(buffer + n, 1, length - n, stdin);
1439 if (!s && feof(stdin))
1440 die("EOF in data (%lu bytes remaining)",
1441 (unsigned long)(length - n));
1442 n += s;
1443 }
1444 }
1445
1446 if (fgetc(stdin) != '\n')
1447 die("An lf did not trail the binary data as expected.");
1448
1449 *size = length;
1450 return buffer;
1451 }
1452
1453 static int validate_raw_date(const char *src, char *result, int maxlen)
1454 {
1455 const char *orig_src = src;
1456 char *endp, sign;
1457
1458 strtoul(src, &endp, 10);
1459 if (endp == src || *endp != ' ')
1460 return -1;
1461
1462 src = endp + 1;
1463 if (*src != '-' && *src != '+')
1464 return -1;
1465 sign = *src;
1466
1467 strtoul(src + 1, &endp, 10);
1468 if (endp == src || *endp || (endp - orig_src) >= maxlen)
1469 return -1;
1470
1471 strcpy(result, orig_src);
1472 return 0;
1473 }
1474
1475 static char *parse_ident(const char *buf)
1476 {
1477 const char *gt;
1478 size_t name_len;
1479 char *ident;
1480
1481 gt = strrchr(buf, '>');
1482 if (!gt)
1483 die("Missing > in ident string: %s", buf);
1484 gt++;
1485 if (*gt != ' ')
1486 die("Missing space after > in ident string: %s", buf);
1487 gt++;
1488 name_len = gt - buf;
1489 ident = xmalloc(name_len + 24);
1490 strncpy(ident, buf, name_len);
1491
1492 switch (whenspec) {
1493 case WHENSPEC_RAW:
1494 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1495 die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1496 break;
1497 case WHENSPEC_RFC2822:
1498 if (parse_date(gt, ident + name_len, 24) < 0)
1499 die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1500 break;
1501 case WHENSPEC_NOW:
1502 if (strcmp("now", gt))
1503 die("Date in ident must be 'now': %s", buf);
1504 datestamp(ident + name_len, 24);
1505 break;
1506 }
1507
1508 return ident;
1509 }
1510
1511 static void cmd_new_blob(void)
1512 {
1513 size_t l;
1514 void *d;
1515
1516 read_next_command();
1517 cmd_mark();
1518 d = cmd_data(&l);
1519
1520 if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1521 free(d);
1522 }
1523
1524 static void unload_one_branch(void)
1525 {
1526 while (cur_active_branches
1527 && cur_active_branches >= max_active_branches) {
1528 unsigned long min_commit = ULONG_MAX;
1529 struct branch *e, *l = NULL, *p = NULL;
1530
1531 for (e = active_branches; e; e = e->active_next_branch) {
1532 if (e->last_commit < min_commit) {
1533 p = l;
1534 min_commit = e->last_commit;
1535 }
1536 l = e;
1537 }
1538
1539 if (p) {
1540 e = p->active_next_branch;
1541 p->active_next_branch = e->active_next_branch;
1542 } else {
1543 e = active_branches;
1544 active_branches = e->active_next_branch;
1545 }
1546 e->active_next_branch = NULL;
1547 if (e->branch_tree.tree) {
1548 release_tree_content_recursive(e->branch_tree.tree);
1549 e->branch_tree.tree = NULL;
1550 }
1551 cur_active_branches--;
1552 }
1553 }
1554
1555 static void load_branch(struct branch *b)
1556 {
1557 load_tree(&b->branch_tree);
1558 b->active_next_branch = active_branches;
1559 active_branches = b;
1560 cur_active_branches++;
1561 branch_load_count++;
1562 }
1563
1564 static void file_change_m(struct branch *b)
1565 {
1566 const char *p = command_buf.buf + 2;
1567 char *p_uq;
1568 const char *endp;
1569 struct object_entry *oe = oe;
1570 unsigned char sha1[20];
1571 uint16_t mode, inline_data = 0;
1572 char type[20];
1573
1574 p = get_mode(p, &mode);
1575 if (!p)
1576 die("Corrupt mode: %s", command_buf.buf);
1577 switch (mode) {
1578 case S_IFREG | 0644:
1579 case S_IFREG | 0755:
1580 case S_IFLNK:
1581 case 0644:
1582 case 0755:
1583 /* ok */
1584 break;
1585 default:
1586 die("Corrupt mode: %s", command_buf.buf);
1587 }
1588
1589 if (*p == ':') {
1590 char *x;
1591 oe = find_mark(strtoumax(p + 1, &x, 10));
1592 hashcpy(sha1, oe->sha1);
1593 p = x;
1594 } else if (!strncmp("inline", p, 6)) {
1595 inline_data = 1;
1596 p += 6;
1597 } else {
1598 if (get_sha1_hex(p, sha1))
1599 die("Invalid SHA1: %s", command_buf.buf);
1600 oe = find_object(sha1);
1601 p += 40;
1602 }
1603 if (*p++ != ' ')
1604 die("Missing space after SHA1: %s", command_buf.buf);
1605
1606 p_uq = unquote_c_style(p, &endp);
1607 if (p_uq) {
1608 if (*endp)
1609 die("Garbage after path in: %s", command_buf.buf);
1610 p = p_uq;
1611 }
1612
1613 if (inline_data) {
1614 size_t l;
1615 void *d;
1616 if (!p_uq)
1617 p = p_uq = xstrdup(p);
1618 read_next_command();
1619 d = cmd_data(&l);
1620 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1621 free(d);
1622 } else if (oe) {
1623 if (oe->type != OBJ_BLOB)
1624 die("Not a blob (actually a %s): %s",
1625 command_buf.buf, type_names[oe->type]);
1626 } else {
1627 if (sha1_object_info(sha1, type, NULL))
1628 die("Blob not found: %s", command_buf.buf);
1629 if (strcmp(blob_type, type))
1630 die("Not a blob (actually a %s): %s",
1631 command_buf.buf, type);
1632 }
1633
1634 tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1635 free(p_uq);
1636 }
1637
1638 static void file_change_d(struct branch *b)
1639 {
1640 const char *p = command_buf.buf + 2;
1641 char *p_uq;
1642 const char *endp;
1643
1644 p_uq = unquote_c_style(p, &endp);
1645 if (p_uq) {
1646 if (*endp)
1647 die("Garbage after path in: %s", command_buf.buf);
1648 p = p_uq;
1649 }
1650 tree_content_remove(&b->branch_tree, p);
1651 free(p_uq);
1652 }
1653
1654 static void file_change_deleteall(struct branch *b)
1655 {
1656 release_tree_content_recursive(b->branch_tree.tree);
1657 hashclr(b->branch_tree.versions[0].sha1);
1658 hashclr(b->branch_tree.versions[1].sha1);
1659 load_tree(&b->branch_tree);
1660 }
1661
1662 static void cmd_from(struct branch *b)
1663 {
1664 const char *from;
1665 struct branch *s;
1666
1667 if (strncmp("from ", command_buf.buf, 5))
1668 return;
1669
1670 if (b->branch_tree.tree) {
1671 release_tree_content_recursive(b->branch_tree.tree);
1672 b->branch_tree.tree = NULL;
1673 }
1674
1675 from = strchr(command_buf.buf, ' ') + 1;
1676 s = lookup_branch(from);
1677 if (b == s)
1678 die("Can't create a branch from itself: %s", b->name);
1679 else if (s) {
1680 unsigned char *t = s->branch_tree.versions[1].sha1;
1681 hashcpy(b->sha1, s->sha1);
1682 hashcpy(b->branch_tree.versions[0].sha1, t);
1683 hashcpy(b->branch_tree.versions[1].sha1, t);
1684 } else if (*from == ':') {
1685 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1686 struct object_entry *oe = find_mark(idnum);
1687 unsigned long size;
1688 char *buf;
1689 if (oe->type != OBJ_COMMIT)
1690 die("Mark :%ju not a commit", idnum);
1691 hashcpy(b->sha1, oe->sha1);
1692 buf = gfi_unpack_entry(oe, &size);
1693 if (!buf || size < 46)
1694 die("Not a valid commit: %s", from);
1695 if (memcmp("tree ", buf, 5)
1696 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1697 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1698 free(buf);
1699 hashcpy(b->branch_tree.versions[0].sha1,
1700 b->branch_tree.versions[1].sha1);
1701 } else if (!get_sha1(from, b->sha1)) {
1702 if (is_null_sha1(b->sha1)) {
1703 hashclr(b->branch_tree.versions[0].sha1);
1704 hashclr(b->branch_tree.versions[1].sha1);
1705 } else {
1706 unsigned long size;
1707 char *buf;
1708
1709 buf = read_object_with_reference(b->sha1,
1710 type_names[OBJ_COMMIT], &size, b->sha1);
1711 if (!buf || size < 46)
1712 die("Not a valid commit: %s", from);
1713 if (memcmp("tree ", buf, 5)
1714 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1715 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1716 free(buf);
1717 hashcpy(b->branch_tree.versions[0].sha1,
1718 b->branch_tree.versions[1].sha1);
1719 }
1720 } else
1721 die("Invalid ref name or SHA1 expression: %s", from);
1722
1723 read_next_command();
1724 }
1725
1726 static struct hash_list *cmd_merge(unsigned int *count)
1727 {
1728 struct hash_list *list = NULL, *n, *e = e;
1729 const char *from;
1730 struct branch *s;
1731
1732 *count = 0;
1733 while (!strncmp("merge ", command_buf.buf, 6)) {
1734 from = strchr(command_buf.buf, ' ') + 1;
1735 n = xmalloc(sizeof(*n));
1736 s = lookup_branch(from);
1737 if (s)
1738 hashcpy(n->sha1, s->sha1);
1739 else if (*from == ':') {
1740 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1741 struct object_entry *oe = find_mark(idnum);
1742 if (oe->type != OBJ_COMMIT)
1743 die("Mark :%ju not a commit", idnum);
1744 hashcpy(n->sha1, oe->sha1);
1745 } else if (get_sha1(from, n->sha1))
1746 die("Invalid ref name or SHA1 expression: %s", from);
1747
1748 n->next = NULL;
1749 if (list)
1750 e->next = n;
1751 else
1752 list = n;
1753 e = n;
1754 (*count)++;
1755 read_next_command();
1756 }
1757 return list;
1758 }
1759
1760 static void cmd_new_commit(void)
1761 {
1762 struct branch *b;
1763 void *msg;
1764 size_t msglen;
1765 char *sp;
1766 char *author = NULL;
1767 char *committer = NULL;
1768 struct hash_list *merge_list = NULL;
1769 unsigned int merge_count;
1770
1771 /* Obtain the branch name from the rest of our command */
1772 sp = strchr(command_buf.buf, ' ') + 1;
1773 b = lookup_branch(sp);
1774 if (!b)
1775 b = new_branch(sp);
1776
1777 read_next_command();
1778 cmd_mark();
1779 if (!strncmp("author ", command_buf.buf, 7)) {
1780 author = parse_ident(command_buf.buf + 7);
1781 read_next_command();
1782 }
1783 if (!strncmp("committer ", command_buf.buf, 10)) {
1784 committer = parse_ident(command_buf.buf + 10);
1785 read_next_command();
1786 }
1787 if (!committer)
1788 die("Expected committer but didn't get one");
1789 msg = cmd_data(&msglen);
1790 read_next_command();
1791 cmd_from(b);
1792 merge_list = cmd_merge(&merge_count);
1793
1794 /* ensure the branch is active/loaded */
1795 if (!b->branch_tree.tree || !max_active_branches) {
1796 unload_one_branch();
1797 load_branch(b);
1798 }
1799
1800 /* file_change* */
1801 for (;;) {
1802 if (1 == command_buf.len)
1803 break;
1804 else if (!strncmp("M ", command_buf.buf, 2))
1805 file_change_m(b);
1806 else if (!strncmp("D ", command_buf.buf, 2))
1807 file_change_d(b);
1808 else if (!strcmp("deleteall", command_buf.buf))
1809 file_change_deleteall(b);
1810 else
1811 die("Unsupported file_change: %s", command_buf.buf);
1812 read_next_command();
1813 }
1814
1815 /* build the tree and the commit */
1816 store_tree(&b->branch_tree);
1817 hashcpy(b->branch_tree.versions[0].sha1,
1818 b->branch_tree.versions[1].sha1);
1819 size_dbuf(&new_data, 114 + msglen
1820 + merge_count * 49
1821 + (author
1822 ? strlen(author) + strlen(committer)
1823 : 2 * strlen(committer)));
1824 sp = new_data.buffer;
1825 sp += sprintf(sp, "tree %s\n",
1826 sha1_to_hex(b->branch_tree.versions[1].sha1));
1827 if (!is_null_sha1(b->sha1))
1828 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1829 while (merge_list) {
1830 struct hash_list *next = merge_list->next;
1831 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1832 free(merge_list);
1833 merge_list = next;
1834 }
1835 sp += sprintf(sp, "author %s\n", author ? author : committer);
1836 sp += sprintf(sp, "committer %s\n", committer);
1837 *sp++ = '\n';
1838 memcpy(sp, msg, msglen);
1839 sp += msglen;
1840 free(author);
1841 free(committer);
1842 free(msg);
1843
1844 if (!store_object(OBJ_COMMIT,
1845 new_data.buffer, sp - (char*)new_data.buffer,
1846 NULL, b->sha1, next_mark))
1847 b->pack_id = pack_id;
1848 b->last_commit = object_count_by_type[OBJ_COMMIT];
1849 }
1850
1851 static void cmd_new_tag(void)
1852 {
1853 char *sp;
1854 const char *from;
1855 char *tagger;
1856 struct branch *s;
1857 void *msg;
1858 size_t msglen;
1859 struct tag *t;
1860 uintmax_t from_mark = 0;
1861 unsigned char sha1[20];
1862
1863 /* Obtain the new tag name from the rest of our command */
1864 sp = strchr(command_buf.buf, ' ') + 1;
1865 t = pool_alloc(sizeof(struct tag));
1866 t->next_tag = NULL;
1867 t->name = pool_strdup(sp);
1868 if (last_tag)
1869 last_tag->next_tag = t;
1870 else
1871 first_tag = t;
1872 last_tag = t;
1873 read_next_command();
1874
1875 /* from ... */
1876 if (strncmp("from ", command_buf.buf, 5))
1877 die("Expected from command, got %s", command_buf.buf);
1878 from = strchr(command_buf.buf, ' ') + 1;
1879 s = lookup_branch(from);
1880 if (s) {
1881 hashcpy(sha1, s->sha1);
1882 } else if (*from == ':') {
1883 struct object_entry *oe;
1884 from_mark = strtoumax(from + 1, NULL, 10);
1885 oe = find_mark(from_mark);
1886 if (oe->type != OBJ_COMMIT)
1887 die("Mark :%ju not a commit", from_mark);
1888 hashcpy(sha1, oe->sha1);
1889 } else if (!get_sha1(from, sha1)) {
1890 unsigned long size;
1891 char *buf;
1892
1893 buf = read_object_with_reference(sha1,
1894 type_names[OBJ_COMMIT], &size, sha1);
1895 if (!buf || size < 46)
1896 die("Not a valid commit: %s", from);
1897 free(buf);
1898 } else
1899 die("Invalid ref name or SHA1 expression: %s", from);
1900 read_next_command();
1901
1902 /* tagger ... */
1903 if (strncmp("tagger ", command_buf.buf, 7))
1904 die("Expected tagger command, got %s", command_buf.buf);
1905 tagger = parse_ident(command_buf.buf + 7);
1906
1907 /* tag payload/message */
1908 read_next_command();
1909 msg = cmd_data(&msglen);
1910
1911 /* build the tag object */
1912 size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1913 sp = new_data.buffer;
1914 sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1915 sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
1916 sp += sprintf(sp, "tag %s\n", t->name);
1917 sp += sprintf(sp, "tagger %s\n", tagger);
1918 *sp++ = '\n';
1919 memcpy(sp, msg, msglen);
1920 sp += msglen;
1921 free(tagger);
1922 free(msg);
1923
1924 if (store_object(OBJ_TAG, new_data.buffer,
1925 sp - (char*)new_data.buffer,
1926 NULL, t->sha1, 0))
1927 t->pack_id = MAX_PACK_ID;
1928 else
1929 t->pack_id = pack_id;
1930 }
1931
1932 static void cmd_reset_branch(void)
1933 {
1934 struct branch *b;
1935 char *sp;
1936
1937 /* Obtain the branch name from the rest of our command */
1938 sp = strchr(command_buf.buf, ' ') + 1;
1939 b = lookup_branch(sp);
1940 if (b) {
1941 hashclr(b->sha1);
1942 hashclr(b->branch_tree.versions[0].sha1);
1943 hashclr(b->branch_tree.versions[1].sha1);
1944 if (b->branch_tree.tree) {
1945 release_tree_content_recursive(b->branch_tree.tree);
1946 b->branch_tree.tree = NULL;
1947 }
1948 }
1949 else
1950 b = new_branch(sp);
1951 read_next_command();
1952 cmd_from(b);
1953 }
1954
1955 static void cmd_checkpoint(void)
1956 {
1957 if (object_count) {
1958 cycle_packfile();
1959 dump_branches();
1960 dump_tags();
1961 dump_marks();
1962 }
1963 read_next_command();
1964 }
1965
1966 static const char fast_import_usage[] =
1967 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
1968
1969 int main(int argc, const char **argv)
1970 {
1971 int i, show_stats = 1;
1972
1973 git_config(git_default_config);
1974
1975 for (i = 1; i < argc; i++) {
1976 const char *a = argv[i];
1977
1978 if (*a != '-' || !strcmp(a, "--"))
1979 break;
1980 else if (!strncmp(a, "--date-format=", 14)) {
1981 const char *fmt = a + 14;
1982 if (!strcmp(fmt, "raw"))
1983 whenspec = WHENSPEC_RAW;
1984 else if (!strcmp(fmt, "rfc2822"))
1985 whenspec = WHENSPEC_RFC2822;
1986 else if (!strcmp(fmt, "now"))
1987 whenspec = WHENSPEC_NOW;
1988 else
1989 die("unknown --date-format argument %s", fmt);
1990 }
1991 else if (!strncmp(a, "--max-pack-size=", 16))
1992 max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
1993 else if (!strncmp(a, "--depth=", 8))
1994 max_depth = strtoul(a + 8, NULL, 0);
1995 else if (!strncmp(a, "--active-branches=", 18))
1996 max_active_branches = strtoul(a + 18, NULL, 0);
1997 else if (!strncmp(a, "--export-marks=", 15))
1998 mark_file = a + 15;
1999 else if (!strncmp(a, "--export-pack-edges=", 20)) {
2000 if (pack_edges)
2001 fclose(pack_edges);
2002 pack_edges = fopen(a + 20, "a");
2003 if (!pack_edges)
2004 die("Cannot open %s: %s", a + 20, strerror(errno));
2005 } else if (!strcmp(a, "--force"))
2006 force_update = 1;
2007 else if (!strcmp(a, "--quiet"))
2008 show_stats = 0;
2009 else if (!strcmp(a, "--stats"))
2010 show_stats = 1;
2011 else
2012 die("unknown option %s", a);
2013 }
2014 if (i != argc)
2015 usage(fast_import_usage);
2016
2017 alloc_objects(object_entry_alloc);
2018 strbuf_init(&command_buf);
2019
2020 atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2021 branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2022 avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2023 marks = pool_calloc(1, sizeof(struct mark_set));
2024
2025 start_packfile();
2026 for (;;) {
2027 read_next_command();
2028 if (command_buf.eof)
2029 break;
2030 else if (!strcmp("blob", command_buf.buf))
2031 cmd_new_blob();
2032 else if (!strncmp("commit ", command_buf.buf, 7))
2033 cmd_new_commit();
2034 else if (!strncmp("tag ", command_buf.buf, 4))
2035 cmd_new_tag();
2036 else if (!strncmp("reset ", command_buf.buf, 6))
2037 cmd_reset_branch();
2038 else if (!strcmp("checkpoint", command_buf.buf))
2039 cmd_checkpoint();
2040 else
2041 die("Unsupported command: %s", command_buf.buf);
2042 }
2043 end_packfile();
2044
2045 dump_branches();
2046 dump_tags();
2047 unkeep_all_packs();
2048 dump_marks();
2049
2050 if (pack_edges)
2051 fclose(pack_edges);
2052
2053 if (show_stats) {
2054 uintmax_t total_count = 0, duplicate_count = 0;
2055 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2056 total_count += object_count_by_type[i];
2057 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2058 duplicate_count += duplicate_count_by_type[i];
2059
2060 fprintf(stderr, "%s statistics:\n", argv[0]);
2061 fprintf(stderr, "---------------------------------------------------------------------\n");
2062 fprintf(stderr, "Alloc'd objects: %10ju\n", alloc_count);
2063 fprintf(stderr, "Total objects: %10ju (%10ju duplicates )\n", total_count, duplicate_count);
2064 fprintf(stderr, " blobs : %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
2065 fprintf(stderr, " trees : %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
2066 fprintf(stderr, " commits: %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
2067 fprintf(stderr, " tags : %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
2068 fprintf(stderr, "Total branches: %10lu (%10lu loads )\n", branch_count, branch_load_count);
2069 fprintf(stderr, " marks: %10ju (%10ju unique )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2070 fprintf(stderr, " atoms: %10u\n", atom_cnt);
2071 fprintf(stderr, "Memory total: %10ju KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2072 fprintf(stderr, " pools: %10lu KiB\n", (unsigned long)(total_allocd/1024));
2073 fprintf(stderr, " objects: %10ju KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2074 fprintf(stderr, "---------------------------------------------------------------------\n");
2075 pack_report();
2076 fprintf(stderr, "---------------------------------------------------------------------\n");
2077 fprintf(stderr, "\n");
2078 }
2079
2080 return failure ? 1 : 0;
2081 }
git with Jakub Narębski modifications