?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
diff -M: release the preimage candidate blobs after rename detection.
[git/git-bigfiles.git] / fast-import.c
blob3a2d5ed8e667af06f97d2559b3cffb8a01f94565
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 active : 1;
220 unsigned pack_id : PACK_ID_BITS;
221 unsigned char sha1[20];
222 };
223
224 struct tag
225 {
226 struct tag *next_tag;
227 const char *name;
228 unsigned int pack_id;
229 unsigned char sha1[20];
230 };
231
232 struct dbuf
233 {
234 void *buffer;
235 size_t capacity;
236 };
237
238 struct hash_list
239 {
240 struct hash_list *next;
241 unsigned char sha1[20];
242 };
243
244 typedef enum {
245 WHENSPEC_RAW = 1,
246 WHENSPEC_RFC2822,
247 WHENSPEC_NOW,
248 } whenspec_type;
249
250 /* Configured limits on output */
251 static unsigned long max_depth = 10;
252 static off_t max_packsize = (1LL << 32) - 1;
253 static int force_update;
254
255 /* Stats and misc. counters */
256 static uintmax_t alloc_count;
257 static uintmax_t marks_set_count;
258 static uintmax_t object_count_by_type[1 << TYPE_BITS];
259 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
260 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
261 static unsigned long object_count;
262 static unsigned long branch_count;
263 static unsigned long branch_load_count;
264 static int failure;
265 static FILE *pack_edges;
266
267 /* Memory pools */
268 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
269 static size_t total_allocd;
270 static struct mem_pool *mem_pool;
271
272 /* Atom management */
273 static unsigned int atom_table_sz = 4451;
274 static unsigned int atom_cnt;
275 static struct atom_str **atom_table;
276
277 /* The .pack file being generated */
278 static unsigned int pack_id;
279 static struct packed_git *pack_data;
280 static struct packed_git **all_packs;
281 static unsigned long pack_size;
282
283 /* Table of objects we've written. */
284 static unsigned int object_entry_alloc = 5000;
285 static struct object_entry_pool *blocks;
286 static struct object_entry *object_table[1 << 16];
287 static struct mark_set *marks;
288 static const char* mark_file;
289
290 /* Our last blob */
291 static struct last_object last_blob;
292
293 /* Tree management */
294 static unsigned int tree_entry_alloc = 1000;
295 static void *avail_tree_entry;
296 static unsigned int avail_tree_table_sz = 100;
297 static struct avail_tree_content **avail_tree_table;
298 static struct dbuf old_tree;
299 static struct dbuf new_tree;
300
301 /* Branch data */
302 static unsigned long max_active_branches = 5;
303 static unsigned long cur_active_branches;
304 static unsigned long branch_table_sz = 1039;
305 static struct branch **branch_table;
306 static struct branch *active_branches;
307
308 /* Tag data */
309 static struct tag *first_tag;
310 static struct tag *last_tag;
311
312 /* Input stream parsing */
313 static whenspec_type whenspec = WHENSPEC_RAW;
314 static struct strbuf command_buf;
315 static uintmax_t next_mark;
316 static struct dbuf new_data;
317
318
319 static void alloc_objects(unsigned int cnt)
320 {
321 struct object_entry_pool *b;
322
323 b = xmalloc(sizeof(struct object_entry_pool)
324 + cnt * sizeof(struct object_entry));
325 b->next_pool = blocks;
326 b->next_free = b->entries;
327 b->end = b->entries + cnt;
328 blocks = b;
329 alloc_count += cnt;
330 }
331
332 static struct object_entry *new_object(unsigned char *sha1)
333 {
334 struct object_entry *e;
335
336 if (blocks->next_free == blocks->end)
337 alloc_objects(object_entry_alloc);
338
339 e = blocks->next_free++;
340 hashcpy(e->sha1, sha1);
341 return e;
342 }
343
344 static struct object_entry *find_object(unsigned char *sha1)
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)
349 if (!hashcmp(sha1, e->sha1))
350 return e;
351 return NULL;
352 }
353
354 static struct object_entry *insert_object(unsigned char *sha1)
355 {
356 unsigned int h = sha1[0] << 8 | sha1[1];
357 struct object_entry *e = object_table[h];
358 struct object_entry *p = NULL;
359
360 while (e) {
361 if (!hashcmp(sha1, e->sha1))
362 return e;
363 p = e;
364 e = e->next;
365 }
366
367 e = new_object(sha1);
368 e->next = NULL;
369 e->offset = 0;
370 if (p)
371 p->next = e;
372 else
373 object_table[h] = e;
374 return e;
375 }
376
377 static 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
385 static void *pool_alloc(size_t len)
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;
408 /* round out to a pointer alignment */
409 if (len & (sizeof(void*) - 1))
410 len += sizeof(void*) - (len & (sizeof(void*) - 1));
411 p->next_free += len;
412 return r;
413 }
414
415 static void *pool_calloc(size_t count, size_t size)
416 {
417 size_t len = count * size;
418 void *r = pool_alloc(len);
419 memset(r, 0, len);
420 return r;
421 }
422
423 static char *pool_strdup(const char *s)
424 {
425 char *r = pool_alloc(strlen(s) + 1);
426 strcpy(r, s);
427 return r;
428 }
429
430 static 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
441 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
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) {
451 uintmax_t i = idnum >> s->shift;
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
464 static struct object_entry *find_mark(uintmax_t idnum)
465 {
466 uintmax_t orig_idnum = idnum;
467 struct mark_set *s = marks;
468 struct object_entry *oe = NULL;
469 if ((idnum >> s->shift) < 1024) {
470 while (s && s->shift) {
471 uintmax_t i = idnum >> s->shift;
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)
479 die("mark :%" PRIuMAX " not declared", orig_idnum);
480 return oe;
481 }
482
483 static struct atom_str *to_atom(const char *s, unsigned short len)
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
502 static struct branch *lookup_branch(const char *name)
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
513 static struct branch *new_branch(const char *name)
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);
520 if (check_ref_format(name))
521 die("Branch name doesn't conform to GIT standards: %s", name);
522
523 b = pool_calloc(1, sizeof(struct branch));
524 b->name = pool_strdup(name);
525 b->table_next_branch = branch_table[hc];
526 b->branch_tree.versions[0].mode = S_IFDIR;
527 b->branch_tree.versions[1].mode = S_IFDIR;
528 b->active = 0;
529 b->pack_id = MAX_PACK_ID;
530 branch_table[hc] = b;
531 branch_count++;
532 return b;
533 }
534
535 static 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
541 static struct tree_content *new_tree_content(unsigned int cnt)
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;
564 t->delta_depth = 0;
565 return t;
566 }
567
568 static void release_tree_entry(struct tree_entry *e);
569 static 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);
573 f->next_avail = avail_tree_table[hc];
574 avail_tree_table[hc] = f;
575 }
576
577 static void release_tree_content_recursive(struct tree_content *t)
578 {
579 unsigned int i;
580 for (i = 0; i < t->entry_count; i++)
581 release_tree_entry(t->entries[i]);
582 release_tree_content(t);
583 }
584
585 static struct tree_content *grow_tree_content(
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;
591 r->delta_depth = t->delta_depth;
592 memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
593 release_tree_content(t);
594 return r;
595 }
596
597 static struct tree_entry *new_tree_entry(void)
598 {
599 struct tree_entry *e;
600
601 if (!avail_tree_entry) {
602 unsigned int n = tree_entry_alloc;
603 total_allocd += n * sizeof(struct tree_entry);
604 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
605 while (n-- > 1) {
606 *((void**)e) = e + 1;
607 e++;
608 }
609 *((void**)e) = NULL;
610 }
611
612 e = avail_tree_entry;
613 avail_tree_entry = *((void**)e);
614 return e;
615 }
616
617 static void release_tree_entry(struct tree_entry *e)
618 {
619 if (e->tree)
620 release_tree_content_recursive(e->tree);
621 *((void**)e) = avail_tree_entry;
622 avail_tree_entry = e;
623 }
624
625 static void start_packfile(void)
626 {
627 static char tmpfile[PATH_MAX];
628 struct packed_git *p;
629 struct pack_header hdr;
630 int pack_fd;
631
632 snprintf(tmpfile, sizeof(tmpfile),
633 "%s/tmp_pack_XXXXXX", get_object_directory());
634 pack_fd = mkstemp(tmpfile);
635 if (pack_fd < 0)
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);
639 p->pack_fd = pack_fd;
640
641 hdr.hdr_signature = htonl(PACK_SIGNATURE);
642 hdr.hdr_version = htonl(2);
643 hdr.hdr_entries = 0;
644 write_or_die(p->pack_fd, &hdr, sizeof(hdr));
645
646 pack_data = p;
647 pack_size = sizeof(hdr);
648 object_count = 0;
649
650 all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
651 all_packs[pack_id] = p;
652 }
653
654 static 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
661 static char *create_index(void)
662 {
663 static char tmpfile[PATH_MAX];
664 SHA_CTX ctx;
665 struct sha1file *f;
666 struct object_entry **idx, **c, **last, *e;
667 struct object_entry_pool *o;
668 uint32_t array[256];
669 int i, idx_fd;
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)
675 for (e = o->next_free; e-- != o->entries;)
676 if (pack_id == e->pack_id)
677 *c++ = e;
678 last = idx + object_count;
679 if (c != last)
680 die("internal consistency error creating the index");
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
696 snprintf(tmpfile, sizeof(tmpfile),
697 "%s/tmp_idx_XXXXXX", get_object_directory());
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);
702 sha1write(f, array, 256 * sizeof(int));
703 SHA1_Init(&ctx);
704 for (c = idx; c != last; c++) {
705 uint32_t offset = htonl((*c)->offset);
706 sha1write(f, &offset, 4);
707 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
708 SHA1_Update(&ctx, (*c)->sha1, 20);
709 }
710 sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
711 sha1close(f, NULL, 1);
712 free(idx);
713 SHA1_Final(pack_data->sha1, &ctx);
714 return tmpfile;
715 }
716
717 static char *keep_pack(char *curr_index_name)
718 {
719 static char name[PATH_MAX];
720 static const char *keep_msg = "fast-import";
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");
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
746 static void unkeep_all_packs(void)
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 }
757 }
758
759 static void end_packfile(void)
760 {
761 struct packed_git *old_p = pack_data, *new_p;
762
763 if (object_count) {
764 char *idx_name;
765 int i;
766 struct branch *b;
767 struct tag *t;
768
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);
772 idx_name = keep_pack(create_index());
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;
779 all_packs[pack_id] = new_p;
780 install_packed_git(new_p);
781
782 /* Print the boundary */
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 }
790 }
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);
797 }
798
799 pack_id++;
800 }
801 else
802 unlink(old_p->pack_name);
803 free(old_p);
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;
811 }
812
813 static void cycle_packfile(void)
814 {
815 end_packfile();
816 start_packfile();
817 }
818
819 static size_t encode_header(
820 enum object_type type,
821 size_t size,
822 unsigned char *hdr)
823 {
824 int n = 1;
825 unsigned char c;
826
827 if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
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
842 static int store_object(
843 enum object_type type,
844 void *dat,
845 size_t datlen,
846 struct last_object *last,
847 unsigned char *sha1out,
848 uintmax_t mark)
849 {
850 void *out, *delta;
851 struct object_entry *e;
852 unsigned char hdr[96];
853 unsigned char sha1[20];
854 unsigned long hdrlen, deltalen;
855 SHA_CTX c;
856 z_stream s;
857
858 hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
859 (unsigned long)datlen) + 1;
860 SHA1_Init(&c);
861 SHA1_Update(&c, hdr, hdrlen);
862 SHA1_Update(&c, dat, datlen);
863 SHA1_Final(sha1, &c);
864 if (sha1out)
865 hashcpy(sha1out, sha1);
866
867 e = insert_object(sha1);
868 if (mark)
869 insert_mark(mark, e);
870 if (e->offset) {
871 duplicate_count_by_type[type]++;
872 return 1;
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;
879 }
880
881 if (last && last->data && last->depth < max_depth) {
882 delta = diff_delta(last->data, last->len,
883 dat, datlen,
884 &deltalen, 0);
885 if (delta && deltalen >= datlen) {
886 free(delta);
887 delta = NULL;
888 }
889 } else
890 delta = NULL;
891
892 memset(&s, 0, sizeof(s));
893 deflateInit(&s, zlib_compression_level);
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. */
908 if ((pack_size + 60 + s.total_out) > max_packsize
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;
913 cycle_packfile();
914
915 /* We cannot carry a delta into the new pack. */
916 if (delta) {
917 free(delta);
918 delta = NULL;
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);
929 }
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]++;
937
938 if (delta) {
939 unsigned long ofs = e->offset - last->offset;
940 unsigned pos = sizeof(hdr) - 1;
941
942 delta_count_by_type[type]++;
943 last->depth++;
944
945 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
946 write_or_die(pack_data->pack_fd, hdr, hdrlen);
947 pack_size += hdrlen;
948
949 hdr[pos] = ofs & 127;
950 while (ofs >>= 7)
951 hdr[--pos] = 128 | (--ofs & 127);
952 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
953 pack_size += sizeof(hdr) - pos;
954 } else {
955 if (last)
956 last->depth = 0;
957 hdrlen = encode_header(type, datlen, hdr);
958 write_or_die(pack_data->pack_fd, hdr, hdrlen);
959 pack_size += hdrlen;
960 }
961
962 write_or_die(pack_data->pack_fd, out, s.total_out);
963 pack_size += s.total_out;
964
965 free(out);
966 free(delta);
967 if (last) {
968 if (!last->no_free)
969 free(last->data);
970 last->data = dat;
971 last->offset = e->offset;
972 last->len = datlen;
973 }
974 return 0;
975 }
976
977 static void *gfi_unpack_entry(
978 struct object_entry *oe,
979 unsigned long *sizep)
980 {
981 enum object_type type;
982 struct packed_git *p = all_packs[oe->pack_id];
983 if (p == pack_data)
984 p->pack_size = pack_size + 20;
985 return unpack_entry(p, oe->offset, &type, sizep);
986 }
987
988 static const char *get_mode(const char *str, uint16_t *modep)
989 {
990 unsigned char c;
991 uint16_t mode = 0;
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
1002 static void load_tree(struct tree_entry *root)
1003 {
1004 unsigned char* sha1 = root->versions[1].sha1;
1005 struct object_entry *myoe;
1006 struct tree_content *t;
1007 unsigned long size;
1008 char *buf;
1009 const char *c;
1010
1011 root->tree = t = new_tree_content(8);
1012 if (is_null_sha1(sha1))
1013 return;
1014
1015 myoe = find_object(sha1);
1016 if (myoe) {
1017 if (myoe->type != OBJ_TREE)
1018 die("Not a tree: %s", sha1_to_hex(sha1));
1019 t->delta_depth = 0;
1020 buf = gfi_unpack_entry(myoe, &size);
1021 } else {
1022 enum object_type type;
1023 buf = read_sha1_file(sha1, &type, &size);
1024 if (!buf || type != OBJ_TREE)
1025 die("Can't load tree %s", sha1_to_hex(sha1));
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)
1033 root->tree = t = grow_tree_content(t, t->entry_count);
1034 t->entries[t->entry_count++] = e;
1035
1036 e->tree = NULL;
1037 c = get_mode(c, &e->versions[1].mode);
1038 if (!c)
1039 die("Corrupt mode in %s", sha1_to_hex(sha1));
1040 e->versions[0].mode = e->versions[1].mode;
1041 e->name = to_atom(c, strlen(c));
1042 c += e->name->str_len + 1;
1043 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1044 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1045 c += 20;
1046 }
1047 free(buf);
1048 }
1049
1050 static int tecmp0 (const void *_a, const void *_b)
1051 {
1052 struct tree_entry *a = *((struct tree_entry**)_a);
1053 struct tree_entry *b = *((struct tree_entry**)_b);
1054 return base_name_compare(
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);
1057 }
1058
1059 static int tecmp1 (const void *_a, const void *_b)
1060 {
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
1068 static void mktree(struct tree_content *t,
1069 int v,
1070 unsigned long *szp,
1071 struct dbuf *b)
1072 {
1073 size_t maxlen = 0;
1074 unsigned int i;
1075 char *c;
1076
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);
1081
1082 for (i = 0; i < t->entry_count; i++) {
1083 if (t->entries[i]->versions[v].mode)
1084 maxlen += t->entries[i]->name->str_len + 34;
1085 }
1086
1087 size_dbuf(b, maxlen);
1088 c = b->buffer;
1089 for (i = 0; i < t->entry_count; i++) {
1090 struct tree_entry *e = t->entries[i];
1091 if (!e->versions[v].mode)
1092 continue;
1093 c += sprintf(c, "%o", (unsigned int)e->versions[v].mode);
1094 *c++ = ' ';
1095 strcpy(c, e->name->str_dat);
1096 c += e->name->str_len + 1;
1097 hashcpy((unsigned char*)c, e->versions[v].sha1);
1098 c += 20;
1099 }
1100 *szp = c - (char*)b->buffer;
1101 }
1102
1103 static void store_tree(struct tree_entry *root)
1104 {
1105 struct tree_content *t = root->tree;
1106 unsigned int i, j, del;
1107 unsigned long new_len;
1108 struct last_object lo;
1109 struct object_entry *le;
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
1119 le = find_object(root->versions[0].sha1);
1120 if (!S_ISDIR(root->versions[0].mode)
1121 || !le
1122 || le->pack_id != pack_id) {
1123 lo.data = NULL;
1124 lo.depth = 0;
1125 } else {
1126 mktree(t, 0, &lo.len, &old_tree);
1127 lo.data = old_tree.buffer;
1128 lo.offset = le->offset;
1129 lo.depth = t->delta_depth;
1130 lo.no_free = 1;
1131 }
1132
1133 mktree(t, 1, &new_len, &new_tree);
1134 store_object(OBJ_TREE, new_tree.buffer, new_len,
1135 &lo, root->versions[1].sha1, 0);
1136
1137 t->delta_depth = lo.depth;
1138 for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1139 struct tree_entry *e = t->entries[i];
1140 if (e->versions[1].mode) {
1141 e->versions[0].mode = e->versions[1].mode;
1142 hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1143 t->entries[j++] = e;
1144 } else {
1145 release_tree_entry(e);
1146 del++;
1147 }
1148 }
1149 t->entry_count -= del;
1150 }
1151
1152 static int tree_content_set(
1153 struct tree_entry *root,
1154 const char *p,
1155 const unsigned char *sha1,
1156 const uint16_t mode)
1157 {
1158 struct tree_content *t = root->tree;
1159 const char *slash1;
1160 unsigned int i, n;
1161 struct tree_entry *e;
1162
1163 slash1 = strchr(p, '/');
1164 if (slash1)
1165 n = slash1 - p;
1166 else
1167 n = strlen(p);
1168 if (!n)
1169 die("Empty path component found in input");
1170
1171 for (i = 0; i < t->entry_count; i++) {
1172 e = t->entries[i];
1173 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1174 if (!slash1) {
1175 if (e->versions[1].mode == mode
1176 && !hashcmp(e->versions[1].sha1, sha1))
1177 return 0;
1178 e->versions[1].mode = mode;
1179 hashcpy(e->versions[1].sha1, sha1);
1180 if (e->tree) {
1181 release_tree_content_recursive(e->tree);
1182 e->tree = NULL;
1183 }
1184 hashclr(root->versions[1].sha1);
1185 return 1;
1186 }
1187 if (!S_ISDIR(e->versions[1].mode)) {
1188 e->tree = new_tree_content(8);
1189 e->versions[1].mode = S_IFDIR;
1190 }
1191 if (!e->tree)
1192 load_tree(e);
1193 if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1194 hashclr(root->versions[1].sha1);
1195 return 1;
1196 }
1197 return 0;
1198 }
1199 }
1200
1201 if (t->entry_count == t->entry_capacity)
1202 root->tree = t = grow_tree_content(t, t->entry_count);
1203 e = new_tree_entry();
1204 e->name = to_atom(p, n);
1205 e->versions[0].mode = 0;
1206 hashclr(e->versions[0].sha1);
1207 t->entries[t->entry_count++] = e;
1208 if (slash1) {
1209 e->tree = new_tree_content(8);
1210 e->versions[1].mode = S_IFDIR;
1211 tree_content_set(e, slash1 + 1, sha1, mode);
1212 } else {
1213 e->tree = NULL;
1214 e->versions[1].mode = mode;
1215 hashcpy(e->versions[1].sha1, sha1);
1216 }
1217 hashclr(root->versions[1].sha1);
1218 return 1;
1219 }
1220
1221 static int tree_content_remove(struct tree_entry *root, const char *p)
1222 {
1223 struct tree_content *t = root->tree;
1224 const char *slash1;
1225 unsigned int i, n;
1226 struct tree_entry *e;
1227
1228 slash1 = strchr(p, '/');
1229 if (slash1)
1230 n = slash1 - p;
1231 else
1232 n = strlen(p);
1233
1234 for (i = 0; i < t->entry_count; i++) {
1235 e = t->entries[i];
1236 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1237 if (!slash1 || !S_ISDIR(e->versions[1].mode))
1238 goto del_entry;
1239 if (!e->tree)
1240 load_tree(e);
1241 if (tree_content_remove(e, slash1 + 1)) {
1242 for (n = 0; n < e->tree->entry_count; n++) {
1243 if (e->tree->entries[n]->versions[1].mode) {
1244 hashclr(root->versions[1].sha1);
1245 return 1;
1246 }
1247 }
1248 goto del_entry;
1249 }
1250 return 0;
1251 }
1252 }
1253 return 0;
1254
1255 del_entry:
1256 if (e->tree) {
1257 release_tree_content_recursive(e->tree);
1258 e->tree = NULL;
1259 }
1260 e->versions[1].mode = 0;
1261 hashclr(e->versions[1].sha1);
1262 hashclr(root->versions[1].sha1);
1263 return 1;
1264 }
1265
1266 static int update_branch(struct branch *b)
1267 {
1268 static const char *msg = "fast-import";
1269 struct ref_lock *lock;
1270 unsigned char old_sha1[20];
1271
1272 if (read_ref(b->name, old_sha1))
1273 hashclr(old_sha1);
1274 lock = lock_any_ref_for_update(b->name, old_sha1);
1275 if (!lock)
1276 return error("Unable to lock %s", b->name);
1277 if (!force_update && !is_null_sha1(old_sha1)) {
1278 struct commit *old_cmit, *new_cmit;
1279
1280 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1281 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1282 if (!old_cmit || !new_cmit) {
1283 unlock_ref(lock);
1284 return error("Branch %s is missing commits.", b->name);
1285 }
1286
1287 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1288 unlock_ref(lock);
1289 warning("Not updating %s"
1290 " (new tip %s does not contain %s)",
1291 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1292 return -1;
1293 }
1294 }
1295 if (write_ref_sha1(lock, b->sha1, msg) < 0)
1296 return error("Unable to update %s", b->name);
1297 return 0;
1298 }
1299
1300 static void dump_branches(void)
1301 {
1302 unsigned int i;
1303 struct branch *b;
1304
1305 for (i = 0; i < branch_table_sz; i++) {
1306 for (b = branch_table[i]; b; b = b->table_next_branch)
1307 failure |= update_branch(b);
1308 }
1309 }
1310
1311 static void dump_tags(void)
1312 {
1313 static const char *msg = "fast-import";
1314 struct tag *t;
1315 struct ref_lock *lock;
1316 char ref_name[PATH_MAX];
1317
1318 for (t = first_tag; t; t = t->next_tag) {
1319 sprintf(ref_name, "tags/%s", t->name);
1320 lock = lock_ref_sha1(ref_name, NULL);
1321 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1322 failure |= error("Unable to update %s", ref_name);
1323 }
1324 }
1325
1326 static void dump_marks_helper(FILE *f,
1327 uintmax_t base,
1328 struct mark_set *m)
1329 {
1330 uintmax_t k;
1331 if (m->shift) {
1332 for (k = 0; k < 1024; k++) {
1333 if (m->data.sets[k])
1334 dump_marks_helper(f, (base + k) << m->shift,
1335 m->data.sets[k]);
1336 }
1337 } else {
1338 for (k = 0; k < 1024; k++) {
1339 if (m->data.marked[k])
1340 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1341 sha1_to_hex(m->data.marked[k]->sha1));
1342 }
1343 }
1344 }
1345
1346 static void dump_marks(void)
1347 {
1348 static struct lock_file mark_lock;
1349 int mark_fd;
1350 FILE *f;
1351
1352 if (!mark_file)
1353 return;
1354
1355 mark_fd = hold_lock_file_for_update(&mark_lock, mark_file, 0);
1356 if (mark_fd < 0) {
1357 failure |= error("Unable to write marks file %s: %s",
1358 mark_file, strerror(errno));
1359 return;
1360 }
1361
1362 f = fdopen(mark_fd, "w");
1363 if (!f) {
1364 rollback_lock_file(&mark_lock);
1365 failure |= error("Unable to write marks file %s: %s",
1366 mark_file, strerror(errno));
1367 return;
1368 }
1369
1370 dump_marks_helper(f, 0, marks);
1371 fclose(f);
1372 if (commit_lock_file(&mark_lock))
1373 failure |= error("Unable to write marks file %s: %s",
1374 mark_file, strerror(errno));
1375 }
1376
1377 static void read_next_command(void)
1378 {
1379 read_line(&command_buf, stdin, '\n');
1380 }
1381
1382 static void cmd_mark(void)
1383 {
1384 if (!prefixcmp(command_buf.buf, "mark :")) {
1385 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1386 read_next_command();
1387 }
1388 else
1389 next_mark = 0;
1390 }
1391
1392 static void *cmd_data (size_t *size)
1393 {
1394 size_t length;
1395 char *buffer;
1396
1397 if (prefixcmp(command_buf.buf, "data "))
1398 die("Expected 'data n' command, found: %s", command_buf.buf);
1399
1400 if (!prefixcmp(command_buf.buf + 5, "<<")) {
1401 char *term = xstrdup(command_buf.buf + 5 + 2);
1402 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1403 length = 0;
1404 buffer = xmalloc(sz);
1405 for (;;) {
1406 read_next_command();
1407 if (command_buf.eof)
1408 die("EOF in data (terminator '%s' not found)", term);
1409 if (term_len == command_buf.len
1410 && !strcmp(term, command_buf.buf))
1411 break;
1412 if (sz < (length + command_buf.len)) {
1413 sz = sz * 3 / 2 + 16;
1414 if (sz < (length + command_buf.len))
1415 sz = length + command_buf.len;
1416 buffer = xrealloc(buffer, sz);
1417 }
1418 memcpy(buffer + length,
1419 command_buf.buf,
1420 command_buf.len - 1);
1421 length += command_buf.len - 1;
1422 buffer[length++] = '\n';
1423 }
1424 free(term);
1425 }
1426 else {
1427 size_t n = 0;
1428 length = strtoul(command_buf.buf + 5, NULL, 10);
1429 buffer = xmalloc(length);
1430 while (n < length) {
1431 size_t s = fread(buffer + n, 1, length - n, stdin);
1432 if (!s && feof(stdin))
1433 die("EOF in data (%lu bytes remaining)",
1434 (unsigned long)(length - n));
1435 n += s;
1436 }
1437 }
1438
1439 if (fgetc(stdin) != '\n')
1440 die("An lf did not trail the binary data as expected.");
1441
1442 *size = length;
1443 return buffer;
1444 }
1445
1446 static int validate_raw_date(const char *src, char *result, int maxlen)
1447 {
1448 const char *orig_src = src;
1449 char *endp, sign;
1450
1451 strtoul(src, &endp, 10);
1452 if (endp == src || *endp != ' ')
1453 return -1;
1454
1455 src = endp + 1;
1456 if (*src != '-' && *src != '+')
1457 return -1;
1458 sign = *src;
1459
1460 strtoul(src + 1, &endp, 10);
1461 if (endp == src || *endp || (endp - orig_src) >= maxlen)
1462 return -1;
1463
1464 strcpy(result, orig_src);
1465 return 0;
1466 }
1467
1468 static char *parse_ident(const char *buf)
1469 {
1470 const char *gt;
1471 size_t name_len;
1472 char *ident;
1473
1474 gt = strrchr(buf, '>');
1475 if (!gt)
1476 die("Missing > in ident string: %s", buf);
1477 gt++;
1478 if (*gt != ' ')
1479 die("Missing space after > in ident string: %s", buf);
1480 gt++;
1481 name_len = gt - buf;
1482 ident = xmalloc(name_len + 24);
1483 strncpy(ident, buf, name_len);
1484
1485 switch (whenspec) {
1486 case WHENSPEC_RAW:
1487 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1488 die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1489 break;
1490 case WHENSPEC_RFC2822:
1491 if (parse_date(gt, ident + name_len, 24) < 0)
1492 die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1493 break;
1494 case WHENSPEC_NOW:
1495 if (strcmp("now", gt))
1496 die("Date in ident must be 'now': %s", buf);
1497 datestamp(ident + name_len, 24);
1498 break;
1499 }
1500
1501 return ident;
1502 }
1503
1504 static void cmd_new_blob(void)
1505 {
1506 size_t l;
1507 void *d;
1508
1509 read_next_command();
1510 cmd_mark();
1511 d = cmd_data(&l);
1512
1513 if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1514 free(d);
1515 }
1516
1517 static void unload_one_branch(void)
1518 {
1519 while (cur_active_branches
1520 && cur_active_branches >= max_active_branches) {
1521 uintmax_t min_commit = ULONG_MAX;
1522 struct branch *e, *l = NULL, *p = NULL;
1523
1524 for (e = active_branches; e; e = e->active_next_branch) {
1525 if (e->last_commit < min_commit) {
1526 p = l;
1527 min_commit = e->last_commit;
1528 }
1529 l = e;
1530 }
1531
1532 if (p) {
1533 e = p->active_next_branch;
1534 p->active_next_branch = e->active_next_branch;
1535 } else {
1536 e = active_branches;
1537 active_branches = e->active_next_branch;
1538 }
1539 e->active = 0;
1540 e->active_next_branch = NULL;
1541 if (e->branch_tree.tree) {
1542 release_tree_content_recursive(e->branch_tree.tree);
1543 e->branch_tree.tree = NULL;
1544 }
1545 cur_active_branches--;
1546 }
1547 }
1548
1549 static void load_branch(struct branch *b)
1550 {
1551 load_tree(&b->branch_tree);
1552 if (!b->active) {
1553 b->active = 1;
1554 b->active_next_branch = active_branches;
1555 active_branches = b;
1556 cur_active_branches++;
1557 branch_load_count++;
1558 }
1559 }
1560
1561 static void file_change_m(struct branch *b)
1562 {
1563 const char *p = command_buf.buf + 2;
1564 char *p_uq;
1565 const char *endp;
1566 struct object_entry *oe = oe;
1567 unsigned char sha1[20];
1568 uint16_t mode, inline_data = 0;
1569
1570 p = get_mode(p, &mode);
1571 if (!p)
1572 die("Corrupt mode: %s", command_buf.buf);
1573 switch (mode) {
1574 case S_IFREG | 0644:
1575 case S_IFREG | 0755:
1576 case S_IFLNK:
1577 case 0644:
1578 case 0755:
1579 /* ok */
1580 break;
1581 default:
1582 die("Corrupt mode: %s", command_buf.buf);
1583 }
1584
1585 if (*p == ':') {
1586 char *x;
1587 oe = find_mark(strtoumax(p + 1, &x, 10));
1588 hashcpy(sha1, oe->sha1);
1589 p = x;
1590 } else if (!prefixcmp(p, "inline")) {
1591 inline_data = 1;
1592 p += 6;
1593 } else {
1594 if (get_sha1_hex(p, sha1))
1595 die("Invalid SHA1: %s", command_buf.buf);
1596 oe = find_object(sha1);
1597 p += 40;
1598 }
1599 if (*p++ != ' ')
1600 die("Missing space after SHA1: %s", command_buf.buf);
1601
1602 p_uq = unquote_c_style(p, &endp);
1603 if (p_uq) {
1604 if (*endp)
1605 die("Garbage after path in: %s", command_buf.buf);
1606 p = p_uq;
1607 }
1608
1609 if (inline_data) {
1610 size_t l;
1611 void *d;
1612 if (!p_uq)
1613 p = p_uq = xstrdup(p);
1614 read_next_command();
1615 d = cmd_data(&l);
1616 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1617 free(d);
1618 } else if (oe) {
1619 if (oe->type != OBJ_BLOB)
1620 die("Not a blob (actually a %s): %s",
1621 command_buf.buf, typename(oe->type));
1622 } else {
1623 enum object_type type = sha1_object_info(sha1, NULL);
1624 if (type < 0)
1625 die("Blob not found: %s", command_buf.buf);
1626 if (type != OBJ_BLOB)
1627 die("Not a blob (actually a %s): %s",
1628 typename(type), command_buf.buf);
1629 }
1630
1631 tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1632 free(p_uq);
1633 }
1634
1635 static void file_change_d(struct branch *b)
1636 {
1637 const char *p = command_buf.buf + 2;
1638 char *p_uq;
1639 const char *endp;
1640
1641 p_uq = unquote_c_style(p, &endp);
1642 if (p_uq) {
1643 if (*endp)
1644 die("Garbage after path in: %s", command_buf.buf);
1645 p = p_uq;
1646 }
1647 tree_content_remove(&b->branch_tree, p);
1648 free(p_uq);
1649 }
1650
1651 static void file_change_deleteall(struct branch *b)
1652 {
1653 release_tree_content_recursive(b->branch_tree.tree);
1654 hashclr(b->branch_tree.versions[0].sha1);
1655 hashclr(b->branch_tree.versions[1].sha1);
1656 load_tree(&b->branch_tree);
1657 }
1658
1659 static void cmd_from(struct branch *b)
1660 {
1661 const char *from;
1662 struct branch *s;
1663
1664 if (prefixcmp(command_buf.buf, "from "))
1665 return;
1666
1667 if (b->branch_tree.tree) {
1668 release_tree_content_recursive(b->branch_tree.tree);
1669 b->branch_tree.tree = NULL;
1670 }
1671
1672 from = strchr(command_buf.buf, ' ') + 1;
1673 s = lookup_branch(from);
1674 if (b == s)
1675 die("Can't create a branch from itself: %s", b->name);
1676 else if (s) {
1677 unsigned char *t = s->branch_tree.versions[1].sha1;
1678 hashcpy(b->sha1, s->sha1);
1679 hashcpy(b->branch_tree.versions[0].sha1, t);
1680 hashcpy(b->branch_tree.versions[1].sha1, t);
1681 } else if (*from == ':') {
1682 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1683 struct object_entry *oe = find_mark(idnum);
1684 unsigned long size;
1685 char *buf;
1686 if (oe->type != OBJ_COMMIT)
1687 die("Mark :%" PRIuMAX " not a commit", idnum);
1688 hashcpy(b->sha1, oe->sha1);
1689 buf = gfi_unpack_entry(oe, &size);
1690 if (!buf || size < 46)
1691 die("Not a valid commit: %s", from);
1692 if (memcmp("tree ", buf, 5)
1693 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1694 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1695 free(buf);
1696 hashcpy(b->branch_tree.versions[0].sha1,
1697 b->branch_tree.versions[1].sha1);
1698 } else if (!get_sha1(from, b->sha1)) {
1699 if (is_null_sha1(b->sha1)) {
1700 hashclr(b->branch_tree.versions[0].sha1);
1701 hashclr(b->branch_tree.versions[1].sha1);
1702 } else {
1703 unsigned long size;
1704 char *buf;
1705
1706 buf = read_object_with_reference(b->sha1,
1707 commit_type, &size, b->sha1);
1708 if (!buf || size < 46)
1709 die("Not a valid commit: %s", from);
1710 if (memcmp("tree ", buf, 5)
1711 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1712 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1713 free(buf);
1714 hashcpy(b->branch_tree.versions[0].sha1,
1715 b->branch_tree.versions[1].sha1);
1716 }
1717 } else
1718 die("Invalid ref name or SHA1 expression: %s", from);
1719
1720 read_next_command();
1721 }
1722
1723 static struct hash_list *cmd_merge(unsigned int *count)
1724 {
1725 struct hash_list *list = NULL, *n, *e = e;
1726 const char *from;
1727 struct branch *s;
1728
1729 *count = 0;
1730 while (!prefixcmp(command_buf.buf, "merge ")) {
1731 from = strchr(command_buf.buf, ' ') + 1;
1732 n = xmalloc(sizeof(*n));
1733 s = lookup_branch(from);
1734 if (s)
1735 hashcpy(n->sha1, s->sha1);
1736 else if (*from == ':') {
1737 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1738 struct object_entry *oe = find_mark(idnum);
1739 if (oe->type != OBJ_COMMIT)
1740 die("Mark :%" PRIuMAX " not a commit", idnum);
1741 hashcpy(n->sha1, oe->sha1);
1742 } else if (!get_sha1(from, n->sha1)) {
1743 unsigned long size;
1744 char *buf = read_object_with_reference(n->sha1,
1745 commit_type, &size, n->sha1);
1746 if (!buf || size < 46)
1747 die("Not a valid commit: %s", from);
1748 free(buf);
1749 } else
1750 die("Invalid ref name or SHA1 expression: %s", from);
1751
1752 n->next = NULL;
1753 if (list)
1754 e->next = n;
1755 else
1756 list = n;
1757 e = n;
1758 (*count)++;
1759 read_next_command();
1760 }
1761 return list;
1762 }
1763
1764 static void cmd_new_commit(void)
1765 {
1766 struct branch *b;
1767 void *msg;
1768 size_t msglen;
1769 char *sp;
1770 char *author = NULL;
1771 char *committer = NULL;
1772 struct hash_list *merge_list = NULL;
1773 unsigned int merge_count;
1774
1775 /* Obtain the branch name from the rest of our command */
1776 sp = strchr(command_buf.buf, ' ') + 1;
1777 b = lookup_branch(sp);
1778 if (!b)
1779 b = new_branch(sp);
1780
1781 read_next_command();
1782 cmd_mark();
1783 if (!prefixcmp(command_buf.buf, "author ")) {
1784 author = parse_ident(command_buf.buf + 7);
1785 read_next_command();
1786 }
1787 if (!prefixcmp(command_buf.buf, "committer ")) {
1788 committer = parse_ident(command_buf.buf + 10);
1789 read_next_command();
1790 }
1791 if (!committer)
1792 die("Expected committer but didn't get one");
1793 msg = cmd_data(&msglen);
1794 read_next_command();
1795 cmd_from(b);
1796 merge_list = cmd_merge(&merge_count);
1797
1798 /* ensure the branch is active/loaded */
1799 if (!b->branch_tree.tree || !max_active_branches) {
1800 unload_one_branch();
1801 load_branch(b);
1802 }
1803
1804 /* file_change* */
1805 for (;;) {
1806 if (1 == command_buf.len)
1807 break;
1808 else if (!prefixcmp(command_buf.buf, "M "))
1809 file_change_m(b);
1810 else if (!prefixcmp(command_buf.buf, "D "))
1811 file_change_d(b);
1812 else if (!strcmp("deleteall", command_buf.buf))
1813 file_change_deleteall(b);
1814 else
1815 die("Unsupported file_change: %s", command_buf.buf);
1816 read_next_command();
1817 }
1818
1819 /* build the tree and the commit */
1820 store_tree(&b->branch_tree);
1821 hashcpy(b->branch_tree.versions[0].sha1,
1822 b->branch_tree.versions[1].sha1);
1823 size_dbuf(&new_data, 114 + msglen
1824 + merge_count * 49
1825 + (author
1826 ? strlen(author) + strlen(committer)
1827 : 2 * strlen(committer)));
1828 sp = new_data.buffer;
1829 sp += sprintf(sp, "tree %s\n",
1830 sha1_to_hex(b->branch_tree.versions[1].sha1));
1831 if (!is_null_sha1(b->sha1))
1832 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1833 while (merge_list) {
1834 struct hash_list *next = merge_list->next;
1835 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1836 free(merge_list);
1837 merge_list = next;
1838 }
1839 sp += sprintf(sp, "author %s\n", author ? author : committer);
1840 sp += sprintf(sp, "committer %s\n", committer);
1841 *sp++ = '\n';
1842 memcpy(sp, msg, msglen);
1843 sp += msglen;
1844 free(author);
1845 free(committer);
1846 free(msg);
1847
1848 if (!store_object(OBJ_COMMIT,
1849 new_data.buffer, sp - (char*)new_data.buffer,
1850 NULL, b->sha1, next_mark))
1851 b->pack_id = pack_id;
1852 b->last_commit = object_count_by_type[OBJ_COMMIT];
1853 }
1854
1855 static void cmd_new_tag(void)
1856 {
1857 char *sp;
1858 const char *from;
1859 char *tagger;
1860 struct branch *s;
1861 void *msg;
1862 size_t msglen;
1863 struct tag *t;
1864 uintmax_t from_mark = 0;
1865 unsigned char sha1[20];
1866
1867 /* Obtain the new tag name from the rest of our command */
1868 sp = strchr(command_buf.buf, ' ') + 1;
1869 t = pool_alloc(sizeof(struct tag));
1870 t->next_tag = NULL;
1871 t->name = pool_strdup(sp);
1872 if (last_tag)
1873 last_tag->next_tag = t;
1874 else
1875 first_tag = t;
1876 last_tag = t;
1877 read_next_command();
1878
1879 /* from ... */
1880 if (prefixcmp(command_buf.buf, "from "))
1881 die("Expected from command, got %s", command_buf.buf);
1882 from = strchr(command_buf.buf, ' ') + 1;
1883 s = lookup_branch(from);
1884 if (s) {
1885 hashcpy(sha1, s->sha1);
1886 } else if (*from == ':') {
1887 struct object_entry *oe;
1888 from_mark = strtoumax(from + 1, NULL, 10);
1889 oe = find_mark(from_mark);
1890 if (oe->type != OBJ_COMMIT)
1891 die("Mark :%" PRIuMAX " not a commit", from_mark);
1892 hashcpy(sha1, oe->sha1);
1893 } else if (!get_sha1(from, sha1)) {
1894 unsigned long size;
1895 char *buf;
1896
1897 buf = read_object_with_reference(sha1,
1898 commit_type, &size, sha1);
1899 if (!buf || size < 46)
1900 die("Not a valid commit: %s", from);
1901 free(buf);
1902 } else
1903 die("Invalid ref name or SHA1 expression: %s", from);
1904 read_next_command();
1905
1906 /* tagger ... */
1907 if (prefixcmp(command_buf.buf, "tagger "))
1908 die("Expected tagger command, got %s", command_buf.buf);
1909 tagger = parse_ident(command_buf.buf + 7);
1910
1911 /* tag payload/message */
1912 read_next_command();
1913 msg = cmd_data(&msglen);
1914
1915 /* build the tag object */
1916 size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1917 sp = new_data.buffer;
1918 sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1919 sp += sprintf(sp, "type %s\n", commit_type);
1920 sp += sprintf(sp, "tag %s\n", t->name);
1921 sp += sprintf(sp, "tagger %s\n", tagger);
1922 *sp++ = '\n';
1923 memcpy(sp, msg, msglen);
1924 sp += msglen;
1925 free(tagger);
1926 free(msg);
1927
1928 if (store_object(OBJ_TAG, new_data.buffer,
1929 sp - (char*)new_data.buffer,
1930 NULL, t->sha1, 0))
1931 t->pack_id = MAX_PACK_ID;
1932 else
1933 t->pack_id = pack_id;
1934 }
1935
1936 static void cmd_reset_branch(void)
1937 {
1938 struct branch *b;
1939 char *sp;
1940
1941 /* Obtain the branch name from the rest of our command */
1942 sp = strchr(command_buf.buf, ' ') + 1;
1943 b = lookup_branch(sp);
1944 if (b) {
1945 hashclr(b->sha1);
1946 hashclr(b->branch_tree.versions[0].sha1);
1947 hashclr(b->branch_tree.versions[1].sha1);
1948 if (b->branch_tree.tree) {
1949 release_tree_content_recursive(b->branch_tree.tree);
1950 b->branch_tree.tree = NULL;
1951 }
1952 }
1953 else
1954 b = new_branch(sp);
1955 read_next_command();
1956 cmd_from(b);
1957 }
1958
1959 static void cmd_checkpoint(void)
1960 {
1961 if (object_count) {
1962 cycle_packfile();
1963 dump_branches();
1964 dump_tags();
1965 dump_marks();
1966 }
1967 read_next_command();
1968 }
1969
1970 static void import_marks(const char *input_file)
1971 {
1972 char line[512];
1973 FILE *f = fopen(input_file, "r");
1974 if (!f)
1975 die("cannot read %s: %s", input_file, strerror(errno));
1976 while (fgets(line, sizeof(line), f)) {
1977 uintmax_t mark;
1978 char *end;
1979 unsigned char sha1[20];
1980 struct object_entry *e;
1981
1982 end = strchr(line, '\n');
1983 if (line[0] != ':' || !end)
1984 die("corrupt mark line: %s", line);
1985 *end = 0;
1986 mark = strtoumax(line + 1, &end, 10);
1987 if (!mark || end == line + 1
1988 || *end != ' ' || get_sha1(end + 1, sha1))
1989 die("corrupt mark line: %s", line);
1990 e = find_object(sha1);
1991 if (!e) {
1992 enum object_type type = sha1_object_info(sha1, NULL);
1993 if (type < 0)
1994 die("object not found: %s", sha1_to_hex(sha1));
1995 e = insert_object(sha1);
1996 e->type = type;
1997 e->pack_id = MAX_PACK_ID;
1998 e->offset = 1; /* just not zero! */
1999 }
2000 insert_mark(mark, e);
2001 }
2002 fclose(f);
2003 }
2004
2005 static const char fast_import_usage[] =
2006 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
2007
2008 int main(int argc, const char **argv)
2009 {
2010 int i, show_stats = 1;
2011
2012 git_config(git_default_config);
2013 alloc_objects(object_entry_alloc);
2014 strbuf_init(&command_buf);
2015 atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2016 branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2017 avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2018 marks = pool_calloc(1, sizeof(struct mark_set));
2019
2020 for (i = 1; i < argc; i++) {
2021 const char *a = argv[i];
2022
2023 if (*a != '-' || !strcmp(a, "--"))
2024 break;
2025 else if (!prefixcmp(a, "--date-format=")) {
2026 const char *fmt = a + 14;
2027 if (!strcmp(fmt, "raw"))
2028 whenspec = WHENSPEC_RAW;
2029 else if (!strcmp(fmt, "rfc2822"))
2030 whenspec = WHENSPEC_RFC2822;
2031 else if (!strcmp(fmt, "now"))
2032 whenspec = WHENSPEC_NOW;
2033 else
2034 die("unknown --date-format argument %s", fmt);
2035 }
2036 else if (!prefixcmp(a, "--max-pack-size="))
2037 max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2038 else if (!prefixcmp(a, "--depth="))
2039 max_depth = strtoul(a + 8, NULL, 0);
2040 else if (!prefixcmp(a, "--active-branches="))
2041 max_active_branches = strtoul(a + 18, NULL, 0);
2042 else if (!prefixcmp(a, "--import-marks="))
2043 import_marks(a + 15);
2044 else if (!prefixcmp(a, "--export-marks="))
2045 mark_file = a + 15;
2046 else if (!prefixcmp(a, "--export-pack-edges=")) {
2047 if (pack_edges)
2048 fclose(pack_edges);
2049 pack_edges = fopen(a + 20, "a");
2050 if (!pack_edges)
2051 die("Cannot open %s: %s", a + 20, strerror(errno));
2052 } else if (!strcmp(a, "--force"))
2053 force_update = 1;
2054 else if (!strcmp(a, "--quiet"))
2055 show_stats = 0;
2056 else if (!strcmp(a, "--stats"))
2057 show_stats = 1;
2058 else
2059 die("unknown option %s", a);
2060 }
2061 if (i != argc)
2062 usage(fast_import_usage);
2063
2064 prepare_packed_git();
2065 start_packfile();
2066 for (;;) {
2067 read_next_command();
2068 if (command_buf.eof)
2069 break;
2070 else if (!strcmp("blob", command_buf.buf))
2071 cmd_new_blob();
2072 else if (!prefixcmp(command_buf.buf, "commit "))
2073 cmd_new_commit();
2074 else if (!prefixcmp(command_buf.buf, "tag "))
2075 cmd_new_tag();
2076 else if (!prefixcmp(command_buf.buf, "reset "))
2077 cmd_reset_branch();
2078 else if (!strcmp("checkpoint", command_buf.buf))
2079 cmd_checkpoint();
2080 else
2081 die("Unsupported command: %s", command_buf.buf);
2082 }
2083 end_packfile();
2084
2085 dump_branches();
2086 dump_tags();
2087 unkeep_all_packs();
2088 dump_marks();
2089
2090 if (pack_edges)
2091 fclose(pack_edges);
2092
2093 if (show_stats) {
2094 uintmax_t total_count = 0, duplicate_count = 0;
2095 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2096 total_count += object_count_by_type[i];
2097 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2098 duplicate_count += duplicate_count_by_type[i];
2099
2100 fprintf(stderr, "%s statistics:\n", argv[0]);
2101 fprintf(stderr, "---------------------------------------------------------------------\n");
2102 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2103 fprintf(stderr, "Total objects: %10" PRIuMAX " (%10" PRIuMAX " duplicates )\n", total_count, duplicate_count);
2104 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]);
2105 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]);
2106 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]);
2107 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]);
2108 fprintf(stderr, "Total branches: %10lu (%10lu loads )\n", branch_count, branch_load_count);
2109 fprintf(stderr, " marks: %10" PRIuMAX " (%10" PRIuMAX " unique )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2110 fprintf(stderr, " atoms: %10u\n", atom_cnt);
2111 fprintf(stderr, "Memory total: %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2112 fprintf(stderr, " pools: %10lu KiB\n", (unsigned long)(total_allocd/1024));
2113 fprintf(stderr, " objects: %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2114 fprintf(stderr, "---------------------------------------------------------------------\n");
2115 pack_report();
2116 fprintf(stderr, "---------------------------------------------------------------------\n");
2117 fprintf(stderr, "\n");
2118 }
2119
2120 return failure ? 1 : 0;
2121 }
Git for large files