?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-bundle: fix pack generation.
[git/jrn.git] / fast-import.c
blobcc3347b23d0c7dba1bc05c65cb58215c92bb8547
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 #ifndef PRIuMAX
137 #define PRIuMAX "llu"
138 #endif
139
140 struct object_entry
141 {
142 struct object_entry *next;
143 uint32_t offset;
144 unsigned type : TYPE_BITS;
145 unsigned pack_id : PACK_ID_BITS;
146 unsigned char sha1[20];
147 };
148
149 struct object_entry_pool
150 {
151 struct object_entry_pool *next_pool;
152 struct object_entry *next_free;
153 struct object_entry *end;
154 struct object_entry entries[FLEX_ARRAY]; /* more */
155 };
156
157 struct mark_set
158 {
159 union {
160 struct object_entry *marked[1024];
161 struct mark_set *sets[1024];
162 } data;
163 unsigned int shift;
164 };
165
166 struct last_object
167 {
168 void *data;
169 unsigned long len;
170 uint32_t offset;
171 unsigned int depth;
172 unsigned no_free:1;
173 };
174
175 struct mem_pool
176 {
177 struct mem_pool *next_pool;
178 char *next_free;
179 char *end;
180 char space[FLEX_ARRAY]; /* more */
181 };
182
183 struct atom_str
184 {
185 struct atom_str *next_atom;
186 unsigned short str_len;
187 char str_dat[FLEX_ARRAY]; /* more */
188 };
189
190 struct tree_content;
191 struct tree_entry
192 {
193 struct tree_content *tree;
194 struct atom_str* name;
195 struct tree_entry_ms
196 {
197 uint16_t mode;
198 unsigned char sha1[20];
199 } versions[2];
200 };
201
202 struct tree_content
203 {
204 unsigned int entry_capacity; /* must match avail_tree_content */
205 unsigned int entry_count;
206 unsigned int delta_depth;
207 struct tree_entry *entries[FLEX_ARRAY]; /* more */
208 };
209
210 struct avail_tree_content
211 {
212 unsigned int entry_capacity; /* must match tree_content */
213 struct avail_tree_content *next_avail;
214 };
215
216 struct branch
217 {
218 struct branch *table_next_branch;
219 struct branch *active_next_branch;
220 const char *name;
221 struct tree_entry branch_tree;
222 uintmax_t last_commit;
223 unsigned active : 1;
224 unsigned pack_id : PACK_ID_BITS;
225 unsigned char sha1[20];
226 };
227
228 struct tag
229 {
230 struct tag *next_tag;
231 const char *name;
232 unsigned int pack_id;
233 unsigned char sha1[20];
234 };
235
236 struct dbuf
237 {
238 void *buffer;
239 size_t capacity;
240 };
241
242 struct hash_list
243 {
244 struct hash_list *next;
245 unsigned char sha1[20];
246 };
247
248 typedef enum {
249 WHENSPEC_RAW = 1,
250 WHENSPEC_RFC2822,
251 WHENSPEC_NOW,
252 } whenspec_type;
253
254 /* Configured limits on output */
255 static unsigned long max_depth = 10;
256 static unsigned long max_packsize = (1LL << 32) - 1;
257 static int force_update;
258
259 /* Stats and misc. counters */
260 static uintmax_t alloc_count;
261 static uintmax_t marks_set_count;
262 static uintmax_t object_count_by_type[1 << TYPE_BITS];
263 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
264 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
265 static unsigned long object_count;
266 static unsigned long branch_count;
267 static unsigned long branch_load_count;
268 static int failure;
269 static FILE *pack_edges;
270
271 /* Memory pools */
272 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
273 static size_t total_allocd;
274 static struct mem_pool *mem_pool;
275
276 /* Atom management */
277 static unsigned int atom_table_sz = 4451;
278 static unsigned int atom_cnt;
279 static struct atom_str **atom_table;
280
281 /* The .pack file being generated */
282 static unsigned int pack_id;
283 static struct packed_git *pack_data;
284 static struct packed_git **all_packs;
285 static unsigned long pack_size;
286
287 /* Table of objects we've written. */
288 static unsigned int object_entry_alloc = 5000;
289 static struct object_entry_pool *blocks;
290 static struct object_entry *object_table[1 << 16];
291 static struct mark_set *marks;
292 static const char* mark_file;
293
294 /* Our last blob */
295 static struct last_object last_blob;
296
297 /* Tree management */
298 static unsigned int tree_entry_alloc = 1000;
299 static void *avail_tree_entry;
300 static unsigned int avail_tree_table_sz = 100;
301 static struct avail_tree_content **avail_tree_table;
302 static struct dbuf old_tree;
303 static struct dbuf new_tree;
304
305 /* Branch data */
306 static unsigned long max_active_branches = 5;
307 static unsigned long cur_active_branches;
308 static unsigned long branch_table_sz = 1039;
309 static struct branch **branch_table;
310 static struct branch *active_branches;
311
312 /* Tag data */
313 static struct tag *first_tag;
314 static struct tag *last_tag;
315
316 /* Input stream parsing */
317 static whenspec_type whenspec = WHENSPEC_RAW;
318 static struct strbuf command_buf;
319 static uintmax_t next_mark;
320 static struct dbuf new_data;
321
322
323 static void alloc_objects(unsigned int cnt)
324 {
325 struct object_entry_pool *b;
326
327 b = xmalloc(sizeof(struct object_entry_pool)
328 + cnt * sizeof(struct object_entry));
329 b->next_pool = blocks;
330 b->next_free = b->entries;
331 b->end = b->entries + cnt;
332 blocks = b;
333 alloc_count += cnt;
334 }
335
336 static struct object_entry *new_object(unsigned char *sha1)
337 {
338 struct object_entry *e;
339
340 if (blocks->next_free == blocks->end)
341 alloc_objects(object_entry_alloc);
342
343 e = blocks->next_free++;
344 hashcpy(e->sha1, sha1);
345 return e;
346 }
347
348 static struct object_entry *find_object(unsigned char *sha1)
349 {
350 unsigned int h = sha1[0] << 8 | sha1[1];
351 struct object_entry *e;
352 for (e = object_table[h]; e; e = e->next)
353 if (!hashcmp(sha1, e->sha1))
354 return e;
355 return NULL;
356 }
357
358 static struct object_entry *insert_object(unsigned char *sha1)
359 {
360 unsigned int h = sha1[0] << 8 | sha1[1];
361 struct object_entry *e = object_table[h];
362 struct object_entry *p = NULL;
363
364 while (e) {
365 if (!hashcmp(sha1, e->sha1))
366 return e;
367 p = e;
368 e = e->next;
369 }
370
371 e = new_object(sha1);
372 e->next = NULL;
373 e->offset = 0;
374 if (p)
375 p->next = e;
376 else
377 object_table[h] = e;
378 return e;
379 }
380
381 static unsigned int hc_str(const char *s, size_t len)
382 {
383 unsigned int r = 0;
384 while (len-- > 0)
385 r = r * 31 + *s++;
386 return r;
387 }
388
389 static void *pool_alloc(size_t len)
390 {
391 struct mem_pool *p;
392 void *r;
393
394 for (p = mem_pool; p; p = p->next_pool)
395 if ((p->end - p->next_free >= len))
396 break;
397
398 if (!p) {
399 if (len >= (mem_pool_alloc/2)) {
400 total_allocd += len;
401 return xmalloc(len);
402 }
403 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
404 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
405 p->next_pool = mem_pool;
406 p->next_free = p->space;
407 p->end = p->next_free + mem_pool_alloc;
408 mem_pool = p;
409 }
410
411 r = p->next_free;
412 /* round out to a pointer alignment */
413 if (len & (sizeof(void*) - 1))
414 len += sizeof(void*) - (len & (sizeof(void*) - 1));
415 p->next_free += len;
416 return r;
417 }
418
419 static void *pool_calloc(size_t count, size_t size)
420 {
421 size_t len = count * size;
422 void *r = pool_alloc(len);
423 memset(r, 0, len);
424 return r;
425 }
426
427 static char *pool_strdup(const char *s)
428 {
429 char *r = pool_alloc(strlen(s) + 1);
430 strcpy(r, s);
431 return r;
432 }
433
434 static void size_dbuf(struct dbuf *b, size_t maxlen)
435 {
436 if (b->buffer) {
437 if (b->capacity >= maxlen)
438 return;
439 free(b->buffer);
440 }
441 b->capacity = ((maxlen / 1024) + 1) * 1024;
442 b->buffer = xmalloc(b->capacity);
443 }
444
445 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
446 {
447 struct mark_set *s = marks;
448 while ((idnum >> s->shift) >= 1024) {
449 s = pool_calloc(1, sizeof(struct mark_set));
450 s->shift = marks->shift + 10;
451 s->data.sets[0] = marks;
452 marks = s;
453 }
454 while (s->shift) {
455 uintmax_t i = idnum >> s->shift;
456 idnum -= i << s->shift;
457 if (!s->data.sets[i]) {
458 s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
459 s->data.sets[i]->shift = s->shift - 10;
460 }
461 s = s->data.sets[i];
462 }
463 if (!s->data.marked[idnum])
464 marks_set_count++;
465 s->data.marked[idnum] = oe;
466 }
467
468 static struct object_entry *find_mark(uintmax_t idnum)
469 {
470 uintmax_t orig_idnum = idnum;
471 struct mark_set *s = marks;
472 struct object_entry *oe = NULL;
473 if ((idnum >> s->shift) < 1024) {
474 while (s && s->shift) {
475 uintmax_t i = idnum >> s->shift;
476 idnum -= i << s->shift;
477 s = s->data.sets[i];
478 }
479 if (s)
480 oe = s->data.marked[idnum];
481 }
482 if (!oe)
483 die("mark :%" PRIuMAX " not declared", orig_idnum);
484 return oe;
485 }
486
487 static struct atom_str *to_atom(const char *s, unsigned short len)
488 {
489 unsigned int hc = hc_str(s, len) % atom_table_sz;
490 struct atom_str *c;
491
492 for (c = atom_table[hc]; c; c = c->next_atom)
493 if (c->str_len == len && !strncmp(s, c->str_dat, len))
494 return c;
495
496 c = pool_alloc(sizeof(struct atom_str) + len + 1);
497 c->str_len = len;
498 strncpy(c->str_dat, s, len);
499 c->str_dat[len] = 0;
500 c->next_atom = atom_table[hc];
501 atom_table[hc] = c;
502 atom_cnt++;
503 return c;
504 }
505
506 static struct branch *lookup_branch(const char *name)
507 {
508 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
509 struct branch *b;
510
511 for (b = branch_table[hc]; b; b = b->table_next_branch)
512 if (!strcmp(name, b->name))
513 return b;
514 return NULL;
515 }
516
517 static struct branch *new_branch(const char *name)
518 {
519 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
520 struct branch* b = lookup_branch(name);
521
522 if (b)
523 die("Invalid attempt to create duplicate branch: %s", name);
524 if (check_ref_format(name))
525 die("Branch name doesn't conform to GIT standards: %s", name);
526
527 b = pool_calloc(1, sizeof(struct branch));
528 b->name = pool_strdup(name);
529 b->table_next_branch = branch_table[hc];
530 b->branch_tree.versions[0].mode = S_IFDIR;
531 b->branch_tree.versions[1].mode = S_IFDIR;
532 b->active = 0;
533 b->pack_id = MAX_PACK_ID;
534 branch_table[hc] = b;
535 branch_count++;
536 return b;
537 }
538
539 static unsigned int hc_entries(unsigned int cnt)
540 {
541 cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
542 return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
543 }
544
545 static struct tree_content *new_tree_content(unsigned int cnt)
546 {
547 struct avail_tree_content *f, *l = NULL;
548 struct tree_content *t;
549 unsigned int hc = hc_entries(cnt);
550
551 for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
552 if (f->entry_capacity >= cnt)
553 break;
554
555 if (f) {
556 if (l)
557 l->next_avail = f->next_avail;
558 else
559 avail_tree_table[hc] = f->next_avail;
560 } else {
561 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
562 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
563 f->entry_capacity = cnt;
564 }
565
566 t = (struct tree_content*)f;
567 t->entry_count = 0;
568 t->delta_depth = 0;
569 return t;
570 }
571
572 static void release_tree_entry(struct tree_entry *e);
573 static void release_tree_content(struct tree_content *t)
574 {
575 struct avail_tree_content *f = (struct avail_tree_content*)t;
576 unsigned int hc = hc_entries(f->entry_capacity);
577 f->next_avail = avail_tree_table[hc];
578 avail_tree_table[hc] = f;
579 }
580
581 static void release_tree_content_recursive(struct tree_content *t)
582 {
583 unsigned int i;
584 for (i = 0; i < t->entry_count; i++)
585 release_tree_entry(t->entries[i]);
586 release_tree_content(t);
587 }
588
589 static struct tree_content *grow_tree_content(
590 struct tree_content *t,
591 int amt)
592 {
593 struct tree_content *r = new_tree_content(t->entry_count + amt);
594 r->entry_count = t->entry_count;
595 r->delta_depth = t->delta_depth;
596 memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
597 release_tree_content(t);
598 return r;
599 }
600
601 static struct tree_entry *new_tree_entry(void)
602 {
603 struct tree_entry *e;
604
605 if (!avail_tree_entry) {
606 unsigned int n = tree_entry_alloc;
607 total_allocd += n * sizeof(struct tree_entry);
608 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
609 while (n-- > 1) {
610 *((void**)e) = e + 1;
611 e++;
612 }
613 *((void**)e) = NULL;
614 }
615
616 e = avail_tree_entry;
617 avail_tree_entry = *((void**)e);
618 return e;
619 }
620
621 static void release_tree_entry(struct tree_entry *e)
622 {
623 if (e->tree)
624 release_tree_content_recursive(e->tree);
625 *((void**)e) = avail_tree_entry;
626 avail_tree_entry = e;
627 }
628
629 static void start_packfile(void)
630 {
631 static char tmpfile[PATH_MAX];
632 struct packed_git *p;
633 struct pack_header hdr;
634 int pack_fd;
635
636 snprintf(tmpfile, sizeof(tmpfile),
637 "%s/pack_XXXXXX", get_object_directory());
638 pack_fd = mkstemp(tmpfile);
639 if (pack_fd < 0)
640 die("Can't create %s: %s", tmpfile, strerror(errno));
641 p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
642 strcpy(p->pack_name, tmpfile);
643 p->pack_fd = pack_fd;
644
645 hdr.hdr_signature = htonl(PACK_SIGNATURE);
646 hdr.hdr_version = htonl(2);
647 hdr.hdr_entries = 0;
648 write_or_die(p->pack_fd, &hdr, sizeof(hdr));
649
650 pack_data = p;
651 pack_size = sizeof(hdr);
652 object_count = 0;
653
654 all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
655 all_packs[pack_id] = p;
656 }
657
658 static void fixup_header_footer(void)
659 {
660 static const int buf_sz = 128 * 1024;
661 int pack_fd = pack_data->pack_fd;
662 SHA_CTX c;
663 struct pack_header hdr;
664 char *buf;
665
666 if (lseek(pack_fd, 0, SEEK_SET) != 0)
667 die("Failed seeking to start: %s", strerror(errno));
668 if (read_in_full(pack_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
669 die("Unable to reread header of %s", pack_data->pack_name);
670 if (lseek(pack_fd, 0, SEEK_SET) != 0)
671 die("Failed seeking to start: %s", strerror(errno));
672 hdr.hdr_entries = htonl(object_count);
673 write_or_die(pack_fd, &hdr, sizeof(hdr));
674
675 SHA1_Init(&c);
676 SHA1_Update(&c, &hdr, sizeof(hdr));
677
678 buf = xmalloc(buf_sz);
679 for (;;) {
680 size_t n = xread(pack_fd, buf, buf_sz);
681 if (!n)
682 break;
683 if (n < 0)
684 die("Failed to checksum %s", pack_data->pack_name);
685 SHA1_Update(&c, buf, n);
686 }
687 free(buf);
688
689 SHA1_Final(pack_data->sha1, &c);
690 write_or_die(pack_fd, pack_data->sha1, sizeof(pack_data->sha1));
691 close(pack_fd);
692 }
693
694 static int oecmp (const void *a_, const void *b_)
695 {
696 struct object_entry *a = *((struct object_entry**)a_);
697 struct object_entry *b = *((struct object_entry**)b_);
698 return hashcmp(a->sha1, b->sha1);
699 }
700
701 static char *create_index(void)
702 {
703 static char tmpfile[PATH_MAX];
704 SHA_CTX ctx;
705 struct sha1file *f;
706 struct object_entry **idx, **c, **last, *e;
707 struct object_entry_pool *o;
708 uint32_t array[256];
709 int i, idx_fd;
710
711 /* Build the sorted table of object IDs. */
712 idx = xmalloc(object_count * sizeof(struct object_entry*));
713 c = idx;
714 for (o = blocks; o; o = o->next_pool)
715 for (e = o->next_free; e-- != o->entries;)
716 if (pack_id == e->pack_id)
717 *c++ = e;
718 last = idx + object_count;
719 if (c != last)
720 die("internal consistency error creating the index");
721 qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
722
723 /* Generate the fan-out array. */
724 c = idx;
725 for (i = 0; i < 256; i++) {
726 struct object_entry **next = c;;
727 while (next < last) {
728 if ((*next)->sha1[0] != i)
729 break;
730 next++;
731 }
732 array[i] = htonl(next - idx);
733 c = next;
734 }
735
736 snprintf(tmpfile, sizeof(tmpfile),
737 "%s/index_XXXXXX", get_object_directory());
738 idx_fd = mkstemp(tmpfile);
739 if (idx_fd < 0)
740 die("Can't create %s: %s", tmpfile, strerror(errno));
741 f = sha1fd(idx_fd, tmpfile);
742 sha1write(f, array, 256 * sizeof(int));
743 SHA1_Init(&ctx);
744 for (c = idx; c != last; c++) {
745 uint32_t offset = htonl((*c)->offset);
746 sha1write(f, &offset, 4);
747 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
748 SHA1_Update(&ctx, (*c)->sha1, 20);
749 }
750 sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
751 sha1close(f, NULL, 1);
752 free(idx);
753 SHA1_Final(pack_data->sha1, &ctx);
754 return tmpfile;
755 }
756
757 static char *keep_pack(char *curr_index_name)
758 {
759 static char name[PATH_MAX];
760 static char *keep_msg = "fast-import";
761 int keep_fd;
762
763 chmod(pack_data->pack_name, 0444);
764 chmod(curr_index_name, 0444);
765
766 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
767 get_object_directory(), sha1_to_hex(pack_data->sha1));
768 keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
769 if (keep_fd < 0)
770 die("cannot create keep file");
771 write(keep_fd, keep_msg, strlen(keep_msg));
772 close(keep_fd);
773
774 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
775 get_object_directory(), sha1_to_hex(pack_data->sha1));
776 if (move_temp_to_file(pack_data->pack_name, name))
777 die("cannot store pack file");
778
779 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
780 get_object_directory(), sha1_to_hex(pack_data->sha1));
781 if (move_temp_to_file(curr_index_name, name))
782 die("cannot store index file");
783 return name;
784 }
785
786 static void unkeep_all_packs(void)
787 {
788 static char name[PATH_MAX];
789 int k;
790
791 for (k = 0; k < pack_id; k++) {
792 struct packed_git *p = all_packs[k];
793 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
794 get_object_directory(), sha1_to_hex(p->sha1));
795 unlink(name);
796 }
797 }
798
799 static void end_packfile(void)
800 {
801 struct packed_git *old_p = pack_data, *new_p;
802
803 if (object_count) {
804 char *idx_name;
805 int i;
806 struct branch *b;
807 struct tag *t;
808
809 fixup_header_footer();
810 idx_name = keep_pack(create_index());
811
812 /* Register the packfile with core git's machinary. */
813 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
814 if (!new_p)
815 die("core git rejected index %s", idx_name);
816 new_p->windows = old_p->windows;
817 all_packs[pack_id] = new_p;
818 install_packed_git(new_p);
819
820 /* Print the boundary */
821 if (pack_edges) {
822 fprintf(pack_edges, "%s:", new_p->pack_name);
823 for (i = 0; i < branch_table_sz; i++) {
824 for (b = branch_table[i]; b; b = b->table_next_branch) {
825 if (b->pack_id == pack_id)
826 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
827 }
828 }
829 for (t = first_tag; t; t = t->next_tag) {
830 if (t->pack_id == pack_id)
831 fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
832 }
833 fputc('\n', pack_edges);
834 fflush(pack_edges);
835 }
836
837 pack_id++;
838 }
839 else
840 unlink(old_p->pack_name);
841 free(old_p);
842
843 /* We can't carry a delta across packfiles. */
844 free(last_blob.data);
845 last_blob.data = NULL;
846 last_blob.len = 0;
847 last_blob.offset = 0;
848 last_blob.depth = 0;
849 }
850
851 static void cycle_packfile(void)
852 {
853 end_packfile();
854 start_packfile();
855 }
856
857 static size_t encode_header(
858 enum object_type type,
859 size_t size,
860 unsigned char *hdr)
861 {
862 int n = 1;
863 unsigned char c;
864
865 if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
866 die("bad type %d", type);
867
868 c = (type << 4) | (size & 15);
869 size >>= 4;
870 while (size) {
871 *hdr++ = c | 0x80;
872 c = size & 0x7f;
873 size >>= 7;
874 n++;
875 }
876 *hdr = c;
877 return n;
878 }
879
880 static int store_object(
881 enum object_type type,
882 void *dat,
883 size_t datlen,
884 struct last_object *last,
885 unsigned char *sha1out,
886 uintmax_t mark)
887 {
888 void *out, *delta;
889 struct object_entry *e;
890 unsigned char hdr[96];
891 unsigned char sha1[20];
892 unsigned long hdrlen, deltalen;
893 SHA_CTX c;
894 z_stream s;
895
896 hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
897 (unsigned long)datlen) + 1;
898 SHA1_Init(&c);
899 SHA1_Update(&c, hdr, hdrlen);
900 SHA1_Update(&c, dat, datlen);
901 SHA1_Final(sha1, &c);
902 if (sha1out)
903 hashcpy(sha1out, sha1);
904
905 e = insert_object(sha1);
906 if (mark)
907 insert_mark(mark, e);
908 if (e->offset) {
909 duplicate_count_by_type[type]++;
910 return 1;
911 }
912
913 if (last && last->data && last->depth < max_depth) {
914 delta = diff_delta(last->data, last->len,
915 dat, datlen,
916 &deltalen, 0);
917 if (delta && deltalen >= datlen) {
918 free(delta);
919 delta = NULL;
920 }
921 } else
922 delta = NULL;
923
924 memset(&s, 0, sizeof(s));
925 deflateInit(&s, zlib_compression_level);
926 if (delta) {
927 s.next_in = delta;
928 s.avail_in = deltalen;
929 } else {
930 s.next_in = dat;
931 s.avail_in = datlen;
932 }
933 s.avail_out = deflateBound(&s, s.avail_in);
934 s.next_out = out = xmalloc(s.avail_out);
935 while (deflate(&s, Z_FINISH) == Z_OK)
936 /* nothing */;
937 deflateEnd(&s);
938
939 /* Determine if we should auto-checkpoint. */
940 if ((pack_size + 60 + s.total_out) > max_packsize
941 || (pack_size + 60 + s.total_out) < pack_size) {
942
943 /* This new object needs to *not* have the current pack_id. */
944 e->pack_id = pack_id + 1;
945 cycle_packfile();
946
947 /* We cannot carry a delta into the new pack. */
948 if (delta) {
949 free(delta);
950 delta = NULL;
951
952 memset(&s, 0, sizeof(s));
953 deflateInit(&s, zlib_compression_level);
954 s.next_in = dat;
955 s.avail_in = datlen;
956 s.avail_out = deflateBound(&s, s.avail_in);
957 s.next_out = out = xrealloc(out, s.avail_out);
958 while (deflate(&s, Z_FINISH) == Z_OK)
959 /* nothing */;
960 deflateEnd(&s);
961 }
962 }
963
964 e->type = type;
965 e->pack_id = pack_id;
966 e->offset = pack_size;
967 object_count++;
968 object_count_by_type[type]++;
969
970 if (delta) {
971 unsigned long ofs = e->offset - last->offset;
972 unsigned pos = sizeof(hdr) - 1;
973
974 delta_count_by_type[type]++;
975 last->depth++;
976
977 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
978 write_or_die(pack_data->pack_fd, hdr, hdrlen);
979 pack_size += hdrlen;
980
981 hdr[pos] = ofs & 127;
982 while (ofs >>= 7)
983 hdr[--pos] = 128 | (--ofs & 127);
984 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
985 pack_size += sizeof(hdr) - pos;
986 } else {
987 if (last)
988 last->depth = 0;
989 hdrlen = encode_header(type, datlen, hdr);
990 write_or_die(pack_data->pack_fd, hdr, hdrlen);
991 pack_size += hdrlen;
992 }
993
994 write_or_die(pack_data->pack_fd, out, s.total_out);
995 pack_size += s.total_out;
996
997 free(out);
998 free(delta);
999 if (last) {
1000 if (!last->no_free)
1001 free(last->data);
1002 last->data = dat;
1003 last->offset = e->offset;
1004 last->len = datlen;
1005 }
1006 return 0;
1007 }
1008
1009 static void *gfi_unpack_entry(
1010 struct object_entry *oe,
1011 unsigned long *sizep)
1012 {
1013 enum object_type type;
1014 struct packed_git *p = all_packs[oe->pack_id];
1015 if (p == pack_data)
1016 p->pack_size = pack_size + 20;
1017 return unpack_entry(p, oe->offset, &type, sizep);
1018 }
1019
1020 static const char *get_mode(const char *str, uint16_t *modep)
1021 {
1022 unsigned char c;
1023 uint16_t mode = 0;
1024
1025 while ((c = *str++) != ' ') {
1026 if (c < '0' || c > '7')
1027 return NULL;
1028 mode = (mode << 3) + (c - '0');
1029 }
1030 *modep = mode;
1031 return str;
1032 }
1033
1034 static void load_tree(struct tree_entry *root)
1035 {
1036 unsigned char* sha1 = root->versions[1].sha1;
1037 struct object_entry *myoe;
1038 struct tree_content *t;
1039 unsigned long size;
1040 char *buf;
1041 const char *c;
1042
1043 root->tree = t = new_tree_content(8);
1044 if (is_null_sha1(sha1))
1045 return;
1046
1047 myoe = find_object(sha1);
1048 if (myoe) {
1049 if (myoe->type != OBJ_TREE)
1050 die("Not a tree: %s", sha1_to_hex(sha1));
1051 t->delta_depth = 0;
1052 buf = gfi_unpack_entry(myoe, &size);
1053 } else {
1054 enum object_type type;
1055 buf = read_sha1_file(sha1, &type, &size);
1056 if (!buf || type != OBJ_TREE)
1057 die("Can't load tree %s", sha1_to_hex(sha1));
1058 }
1059
1060 c = buf;
1061 while (c != (buf + size)) {
1062 struct tree_entry *e = new_tree_entry();
1063
1064 if (t->entry_count == t->entry_capacity)
1065 root->tree = t = grow_tree_content(t, 8);
1066 t->entries[t->entry_count++] = e;
1067
1068 e->tree = NULL;
1069 c = get_mode(c, &e->versions[1].mode);
1070 if (!c)
1071 die("Corrupt mode in %s", sha1_to_hex(sha1));
1072 e->versions[0].mode = e->versions[1].mode;
1073 e->name = to_atom(c, (unsigned short)strlen(c));
1074 c += e->name->str_len + 1;
1075 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1076 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1077 c += 20;
1078 }
1079 free(buf);
1080 }
1081
1082 static int tecmp0 (const void *_a, const void *_b)
1083 {
1084 struct tree_entry *a = *((struct tree_entry**)_a);
1085 struct tree_entry *b = *((struct tree_entry**)_b);
1086 return base_name_compare(
1087 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1088 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1089 }
1090
1091 static int tecmp1 (const void *_a, const void *_b)
1092 {
1093 struct tree_entry *a = *((struct tree_entry**)_a);
1094 struct tree_entry *b = *((struct tree_entry**)_b);
1095 return base_name_compare(
1096 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1097 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1098 }
1099
1100 static void mktree(struct tree_content *t,
1101 int v,
1102 unsigned long *szp,
1103 struct dbuf *b)
1104 {
1105 size_t maxlen = 0;
1106 unsigned int i;
1107 char *c;
1108
1109 if (!v)
1110 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1111 else
1112 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1113
1114 for (i = 0; i < t->entry_count; i++) {
1115 if (t->entries[i]->versions[v].mode)
1116 maxlen += t->entries[i]->name->str_len + 34;
1117 }
1118
1119 size_dbuf(b, maxlen);
1120 c = b->buffer;
1121 for (i = 0; i < t->entry_count; i++) {
1122 struct tree_entry *e = t->entries[i];
1123 if (!e->versions[v].mode)
1124 continue;
1125 c += sprintf(c, "%o", (unsigned int)e->versions[v].mode);
1126 *c++ = ' ';
1127 strcpy(c, e->name->str_dat);
1128 c += e->name->str_len + 1;
1129 hashcpy((unsigned char*)c, e->versions[v].sha1);
1130 c += 20;
1131 }
1132 *szp = c - (char*)b->buffer;
1133 }
1134
1135 static void store_tree(struct tree_entry *root)
1136 {
1137 struct tree_content *t = root->tree;
1138 unsigned int i, j, del;
1139 unsigned long new_len;
1140 struct last_object lo;
1141 struct object_entry *le;
1142
1143 if (!is_null_sha1(root->versions[1].sha1))
1144 return;
1145
1146 for (i = 0; i < t->entry_count; i++) {
1147 if (t->entries[i]->tree)
1148 store_tree(t->entries[i]);
1149 }
1150
1151 le = find_object(root->versions[0].sha1);
1152 if (!S_ISDIR(root->versions[0].mode)
1153 || !le
1154 || le->pack_id != pack_id) {
1155 lo.data = NULL;
1156 lo.depth = 0;
1157 } else {
1158 mktree(t, 0, &lo.len, &old_tree);
1159 lo.data = old_tree.buffer;
1160 lo.offset = le->offset;
1161 lo.depth = t->delta_depth;
1162 lo.no_free = 1;
1163 }
1164
1165 mktree(t, 1, &new_len, &new_tree);
1166 store_object(OBJ_TREE, new_tree.buffer, new_len,
1167 &lo, root->versions[1].sha1, 0);
1168
1169 t->delta_depth = lo.depth;
1170 for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1171 struct tree_entry *e = t->entries[i];
1172 if (e->versions[1].mode) {
1173 e->versions[0].mode = e->versions[1].mode;
1174 hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1175 t->entries[j++] = e;
1176 } else {
1177 release_tree_entry(e);
1178 del++;
1179 }
1180 }
1181 t->entry_count -= del;
1182 }
1183
1184 static int tree_content_set(
1185 struct tree_entry *root,
1186 const char *p,
1187 const unsigned char *sha1,
1188 const uint16_t mode)
1189 {
1190 struct tree_content *t = root->tree;
1191 const char *slash1;
1192 unsigned int i, n;
1193 struct tree_entry *e;
1194
1195 slash1 = strchr(p, '/');
1196 if (slash1)
1197 n = slash1 - p;
1198 else
1199 n = strlen(p);
1200
1201 for (i = 0; i < t->entry_count; i++) {
1202 e = t->entries[i];
1203 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1204 if (!slash1) {
1205 if (e->versions[1].mode == mode
1206 && !hashcmp(e->versions[1].sha1, sha1))
1207 return 0;
1208 e->versions[1].mode = mode;
1209 hashcpy(e->versions[1].sha1, sha1);
1210 if (e->tree) {
1211 release_tree_content_recursive(e->tree);
1212 e->tree = NULL;
1213 }
1214 hashclr(root->versions[1].sha1);
1215 return 1;
1216 }
1217 if (!S_ISDIR(e->versions[1].mode)) {
1218 e->tree = new_tree_content(8);
1219 e->versions[1].mode = S_IFDIR;
1220 }
1221 if (!e->tree)
1222 load_tree(e);
1223 if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1224 hashclr(root->versions[1].sha1);
1225 return 1;
1226 }
1227 return 0;
1228 }
1229 }
1230
1231 if (t->entry_count == t->entry_capacity)
1232 root->tree = t = grow_tree_content(t, 8);
1233 e = new_tree_entry();
1234 e->name = to_atom(p, (unsigned short)n);
1235 e->versions[0].mode = 0;
1236 hashclr(e->versions[0].sha1);
1237 t->entries[t->entry_count++] = e;
1238 if (slash1) {
1239 e->tree = new_tree_content(8);
1240 e->versions[1].mode = S_IFDIR;
1241 tree_content_set(e, slash1 + 1, sha1, mode);
1242 } else {
1243 e->tree = NULL;
1244 e->versions[1].mode = mode;
1245 hashcpy(e->versions[1].sha1, sha1);
1246 }
1247 hashclr(root->versions[1].sha1);
1248 return 1;
1249 }
1250
1251 static int tree_content_remove(struct tree_entry *root, const char *p)
1252 {
1253 struct tree_content *t = root->tree;
1254 const char *slash1;
1255 unsigned int i, n;
1256 struct tree_entry *e;
1257
1258 slash1 = strchr(p, '/');
1259 if (slash1)
1260 n = slash1 - p;
1261 else
1262 n = strlen(p);
1263
1264 for (i = 0; i < t->entry_count; i++) {
1265 e = t->entries[i];
1266 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1267 if (!slash1 || !S_ISDIR(e->versions[1].mode))
1268 goto del_entry;
1269 if (!e->tree)
1270 load_tree(e);
1271 if (tree_content_remove(e, slash1 + 1)) {
1272 for (n = 0; n < e->tree->entry_count; n++) {
1273 if (e->tree->entries[n]->versions[1].mode) {
1274 hashclr(root->versions[1].sha1);
1275 return 1;
1276 }
1277 }
1278 goto del_entry;
1279 }
1280 return 0;
1281 }
1282 }
1283 return 0;
1284
1285 del_entry:
1286 if (e->tree) {
1287 release_tree_content_recursive(e->tree);
1288 e->tree = NULL;
1289 }
1290 e->versions[1].mode = 0;
1291 hashclr(e->versions[1].sha1);
1292 hashclr(root->versions[1].sha1);
1293 return 1;
1294 }
1295
1296 static int update_branch(struct branch *b)
1297 {
1298 static const char *msg = "fast-import";
1299 struct ref_lock *lock;
1300 unsigned char old_sha1[20];
1301
1302 if (read_ref(b->name, old_sha1))
1303 hashclr(old_sha1);
1304 lock = lock_any_ref_for_update(b->name, old_sha1);
1305 if (!lock)
1306 return error("Unable to lock %s", b->name);
1307 if (!force_update && !is_null_sha1(old_sha1)) {
1308 struct commit *old_cmit, *new_cmit;
1309
1310 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1311 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1312 if (!old_cmit || !new_cmit) {
1313 unlock_ref(lock);
1314 return error("Branch %s is missing commits.", b->name);
1315 }
1316
1317 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1318 unlock_ref(lock);
1319 warn("Not updating %s"
1320 " (new tip %s does not contain %s)",
1321 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1322 return -1;
1323 }
1324 }
1325 if (write_ref_sha1(lock, b->sha1, msg) < 0)
1326 return error("Unable to update %s", b->name);
1327 return 0;
1328 }
1329
1330 static void dump_branches(void)
1331 {
1332 unsigned int i;
1333 struct branch *b;
1334
1335 for (i = 0; i < branch_table_sz; i++) {
1336 for (b = branch_table[i]; b; b = b->table_next_branch)
1337 failure |= update_branch(b);
1338 }
1339 }
1340
1341 static void dump_tags(void)
1342 {
1343 static const char *msg = "fast-import";
1344 struct tag *t;
1345 struct ref_lock *lock;
1346 char ref_name[PATH_MAX];
1347
1348 for (t = first_tag; t; t = t->next_tag) {
1349 sprintf(ref_name, "tags/%s", t->name);
1350 lock = lock_ref_sha1(ref_name, NULL);
1351 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1352 failure |= error("Unable to update %s", ref_name);
1353 }
1354 }
1355
1356 static void dump_marks_helper(FILE *f,
1357 uintmax_t base,
1358 struct mark_set *m)
1359 {
1360 uintmax_t k;
1361 if (m->shift) {
1362 for (k = 0; k < 1024; k++) {
1363 if (m->data.sets[k])
1364 dump_marks_helper(f, (base + k) << m->shift,
1365 m->data.sets[k]);
1366 }
1367 } else {
1368 for (k = 0; k < 1024; k++) {
1369 if (m->data.marked[k])
1370 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1371 sha1_to_hex(m->data.marked[k]->sha1));
1372 }
1373 }
1374 }
1375
1376 static void dump_marks(void)
1377 {
1378 if (mark_file)
1379 {
1380 FILE *f = fopen(mark_file, "w");
1381 if (f) {
1382 dump_marks_helper(f, 0, marks);
1383 fclose(f);
1384 } else
1385 failure |= error("Unable to write marks file %s: %s",
1386 mark_file, strerror(errno));
1387 }
1388 }
1389
1390 static void read_next_command(void)
1391 {
1392 read_line(&command_buf, stdin, '\n');
1393 }
1394
1395 static void cmd_mark(void)
1396 {
1397 if (!prefixcmp(command_buf.buf, "mark :")) {
1398 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1399 read_next_command();
1400 }
1401 else
1402 next_mark = 0;
1403 }
1404
1405 static void *cmd_data (size_t *size)
1406 {
1407 size_t length;
1408 char *buffer;
1409
1410 if (prefixcmp(command_buf.buf, "data "))
1411 die("Expected 'data n' command, found: %s", command_buf.buf);
1412
1413 if (!prefixcmp(command_buf.buf + 5, "<<")) {
1414 char *term = xstrdup(command_buf.buf + 5 + 2);
1415 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1416 length = 0;
1417 buffer = xmalloc(sz);
1418 for (;;) {
1419 read_next_command();
1420 if (command_buf.eof)
1421 die("EOF in data (terminator '%s' not found)", term);
1422 if (term_len == command_buf.len
1423 && !strcmp(term, command_buf.buf))
1424 break;
1425 if (sz < (length + command_buf.len)) {
1426 sz = sz * 3 / 2 + 16;
1427 if (sz < (length + command_buf.len))
1428 sz = length + command_buf.len;
1429 buffer = xrealloc(buffer, sz);
1430 }
1431 memcpy(buffer + length,
1432 command_buf.buf,
1433 command_buf.len - 1);
1434 length += command_buf.len - 1;
1435 buffer[length++] = '\n';
1436 }
1437 free(term);
1438 }
1439 else {
1440 size_t n = 0;
1441 length = strtoul(command_buf.buf + 5, NULL, 10);
1442 buffer = xmalloc(length);
1443 while (n < length) {
1444 size_t s = fread(buffer + n, 1, length - n, stdin);
1445 if (!s && feof(stdin))
1446 die("EOF in data (%lu bytes remaining)",
1447 (unsigned long)(length - n));
1448 n += s;
1449 }
1450 }
1451
1452 if (fgetc(stdin) != '\n')
1453 die("An lf did not trail the binary data as expected.");
1454
1455 *size = length;
1456 return buffer;
1457 }
1458
1459 static int validate_raw_date(const char *src, char *result, int maxlen)
1460 {
1461 const char *orig_src = src;
1462 char *endp, sign;
1463
1464 strtoul(src, &endp, 10);
1465 if (endp == src || *endp != ' ')
1466 return -1;
1467
1468 src = endp + 1;
1469 if (*src != '-' && *src != '+')
1470 return -1;
1471 sign = *src;
1472
1473 strtoul(src + 1, &endp, 10);
1474 if (endp == src || *endp || (endp - orig_src) >= maxlen)
1475 return -1;
1476
1477 strcpy(result, orig_src);
1478 return 0;
1479 }
1480
1481 static char *parse_ident(const char *buf)
1482 {
1483 const char *gt;
1484 size_t name_len;
1485 char *ident;
1486
1487 gt = strrchr(buf, '>');
1488 if (!gt)
1489 die("Missing > in ident string: %s", buf);
1490 gt++;
1491 if (*gt != ' ')
1492 die("Missing space after > in ident string: %s", buf);
1493 gt++;
1494 name_len = gt - buf;
1495 ident = xmalloc(name_len + 24);
1496 strncpy(ident, buf, name_len);
1497
1498 switch (whenspec) {
1499 case WHENSPEC_RAW:
1500 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1501 die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1502 break;
1503 case WHENSPEC_RFC2822:
1504 if (parse_date(gt, ident + name_len, 24) < 0)
1505 die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1506 break;
1507 case WHENSPEC_NOW:
1508 if (strcmp("now", gt))
1509 die("Date in ident must be 'now': %s", buf);
1510 datestamp(ident + name_len, 24);
1511 break;
1512 }
1513
1514 return ident;
1515 }
1516
1517 static void cmd_new_blob(void)
1518 {
1519 size_t l;
1520 void *d;
1521
1522 read_next_command();
1523 cmd_mark();
1524 d = cmd_data(&l);
1525
1526 if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1527 free(d);
1528 }
1529
1530 static void unload_one_branch(void)
1531 {
1532 while (cur_active_branches
1533 && cur_active_branches >= max_active_branches) {
1534 unsigned long min_commit = ULONG_MAX;
1535 struct branch *e, *l = NULL, *p = NULL;
1536
1537 for (e = active_branches; e; e = e->active_next_branch) {
1538 if (e->last_commit < min_commit) {
1539 p = l;
1540 min_commit = e->last_commit;
1541 }
1542 l = e;
1543 }
1544
1545 if (p) {
1546 e = p->active_next_branch;
1547 p->active_next_branch = e->active_next_branch;
1548 } else {
1549 e = active_branches;
1550 active_branches = e->active_next_branch;
1551 }
1552 e->active = 0;
1553 e->active_next_branch = NULL;
1554 if (e->branch_tree.tree) {
1555 release_tree_content_recursive(e->branch_tree.tree);
1556 e->branch_tree.tree = NULL;
1557 }
1558 cur_active_branches--;
1559 }
1560 }
1561
1562 static void load_branch(struct branch *b)
1563 {
1564 load_tree(&b->branch_tree);
1565 if (!b->active) {
1566 b->active = 1;
1567 b->active_next_branch = active_branches;
1568 active_branches = b;
1569 cur_active_branches++;
1570 branch_load_count++;
1571 }
1572 }
1573
1574 static void file_change_m(struct branch *b)
1575 {
1576 const char *p = command_buf.buf + 2;
1577 char *p_uq;
1578 const char *endp;
1579 struct object_entry *oe = oe;
1580 unsigned char sha1[20];
1581 uint16_t mode, inline_data = 0;
1582
1583 p = get_mode(p, &mode);
1584 if (!p)
1585 die("Corrupt mode: %s", command_buf.buf);
1586 switch (mode) {
1587 case S_IFREG | 0644:
1588 case S_IFREG | 0755:
1589 case S_IFLNK:
1590 case 0644:
1591 case 0755:
1592 /* ok */
1593 break;
1594 default:
1595 die("Corrupt mode: %s", command_buf.buf);
1596 }
1597
1598 if (*p == ':') {
1599 char *x;
1600 oe = find_mark(strtoumax(p + 1, &x, 10));
1601 hashcpy(sha1, oe->sha1);
1602 p = x;
1603 } else if (!prefixcmp(p, "inline")) {
1604 inline_data = 1;
1605 p += 6;
1606 } else {
1607 if (get_sha1_hex(p, sha1))
1608 die("Invalid SHA1: %s", command_buf.buf);
1609 oe = find_object(sha1);
1610 p += 40;
1611 }
1612 if (*p++ != ' ')
1613 die("Missing space after SHA1: %s", command_buf.buf);
1614
1615 p_uq = unquote_c_style(p, &endp);
1616 if (p_uq) {
1617 if (*endp)
1618 die("Garbage after path in: %s", command_buf.buf);
1619 p = p_uq;
1620 }
1621
1622 if (inline_data) {
1623 size_t l;
1624 void *d;
1625 if (!p_uq)
1626 p = p_uq = xstrdup(p);
1627 read_next_command();
1628 d = cmd_data(&l);
1629 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1630 free(d);
1631 } else if (oe) {
1632 if (oe->type != OBJ_BLOB)
1633 die("Not a blob (actually a %s): %s",
1634 command_buf.buf, typename(oe->type));
1635 } else {
1636 enum object_type type = sha1_object_info(sha1, NULL);
1637 if (type < 0)
1638 die("Blob not found: %s", command_buf.buf);
1639 if (type != OBJ_BLOB)
1640 die("Not a blob (actually a %s): %s",
1641 typename(type), command_buf.buf);
1642 }
1643
1644 tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1645 free(p_uq);
1646 }
1647
1648 static void file_change_d(struct branch *b)
1649 {
1650 const char *p = command_buf.buf + 2;
1651 char *p_uq;
1652 const char *endp;
1653
1654 p_uq = unquote_c_style(p, &endp);
1655 if (p_uq) {
1656 if (*endp)
1657 die("Garbage after path in: %s", command_buf.buf);
1658 p = p_uq;
1659 }
1660 tree_content_remove(&b->branch_tree, p);
1661 free(p_uq);
1662 }
1663
1664 static void file_change_deleteall(struct branch *b)
1665 {
1666 release_tree_content_recursive(b->branch_tree.tree);
1667 hashclr(b->branch_tree.versions[0].sha1);
1668 hashclr(b->branch_tree.versions[1].sha1);
1669 load_tree(&b->branch_tree);
1670 }
1671
1672 static void cmd_from(struct branch *b)
1673 {
1674 const char *from;
1675 struct branch *s;
1676
1677 if (prefixcmp(command_buf.buf, "from "))
1678 return;
1679
1680 if (b->branch_tree.tree) {
1681 release_tree_content_recursive(b->branch_tree.tree);
1682 b->branch_tree.tree = NULL;
1683 }
1684
1685 from = strchr(command_buf.buf, ' ') + 1;
1686 s = lookup_branch(from);
1687 if (b == s)
1688 die("Can't create a branch from itself: %s", b->name);
1689 else if (s) {
1690 unsigned char *t = s->branch_tree.versions[1].sha1;
1691 hashcpy(b->sha1, s->sha1);
1692 hashcpy(b->branch_tree.versions[0].sha1, t);
1693 hashcpy(b->branch_tree.versions[1].sha1, t);
1694 } else if (*from == ':') {
1695 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1696 struct object_entry *oe = find_mark(idnum);
1697 unsigned long size;
1698 char *buf;
1699 if (oe->type != OBJ_COMMIT)
1700 die("Mark :%" PRIuMAX " not a commit", idnum);
1701 hashcpy(b->sha1, oe->sha1);
1702 buf = gfi_unpack_entry(oe, &size);
1703 if (!buf || size < 46)
1704 die("Not a valid commit: %s", from);
1705 if (memcmp("tree ", buf, 5)
1706 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1707 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1708 free(buf);
1709 hashcpy(b->branch_tree.versions[0].sha1,
1710 b->branch_tree.versions[1].sha1);
1711 } else if (!get_sha1(from, b->sha1)) {
1712 if (is_null_sha1(b->sha1)) {
1713 hashclr(b->branch_tree.versions[0].sha1);
1714 hashclr(b->branch_tree.versions[1].sha1);
1715 } else {
1716 unsigned long size;
1717 char *buf;
1718
1719 buf = read_object_with_reference(b->sha1,
1720 commit_type, &size, b->sha1);
1721 if (!buf || size < 46)
1722 die("Not a valid commit: %s", from);
1723 if (memcmp("tree ", buf, 5)
1724 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1725 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1726 free(buf);
1727 hashcpy(b->branch_tree.versions[0].sha1,
1728 b->branch_tree.versions[1].sha1);
1729 }
1730 } else
1731 die("Invalid ref name or SHA1 expression: %s", from);
1732
1733 read_next_command();
1734 }
1735
1736 static struct hash_list *cmd_merge(unsigned int *count)
1737 {
1738 struct hash_list *list = NULL, *n, *e = e;
1739 const char *from;
1740 struct branch *s;
1741
1742 *count = 0;
1743 while (!prefixcmp(command_buf.buf, "merge ")) {
1744 from = strchr(command_buf.buf, ' ') + 1;
1745 n = xmalloc(sizeof(*n));
1746 s = lookup_branch(from);
1747 if (s)
1748 hashcpy(n->sha1, s->sha1);
1749 else if (*from == ':') {
1750 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1751 struct object_entry *oe = find_mark(idnum);
1752 if (oe->type != OBJ_COMMIT)
1753 die("Mark :%" PRIuMAX " not a commit", idnum);
1754 hashcpy(n->sha1, oe->sha1);
1755 } else if (!get_sha1(from, n->sha1)) {
1756 unsigned long size;
1757 char *buf = read_object_with_reference(n->sha1,
1758 commit_type, &size, n->sha1);
1759 if (!buf || size < 46)
1760 die("Not a valid commit: %s", from);
1761 free(buf);
1762 } else
1763 die("Invalid ref name or SHA1 expression: %s", from);
1764
1765 n->next = NULL;
1766 if (list)
1767 e->next = n;
1768 else
1769 list = n;
1770 e = n;
1771 (*count)++;
1772 read_next_command();
1773 }
1774 return list;
1775 }
1776
1777 static void cmd_new_commit(void)
1778 {
1779 struct branch *b;
1780 void *msg;
1781 size_t msglen;
1782 char *sp;
1783 char *author = NULL;
1784 char *committer = NULL;
1785 struct hash_list *merge_list = NULL;
1786 unsigned int merge_count;
1787
1788 /* Obtain the branch name from the rest of our command */
1789 sp = strchr(command_buf.buf, ' ') + 1;
1790 b = lookup_branch(sp);
1791 if (!b)
1792 b = new_branch(sp);
1793
1794 read_next_command();
1795 cmd_mark();
1796 if (!prefixcmp(command_buf.buf, "author ")) {
1797 author = parse_ident(command_buf.buf + 7);
1798 read_next_command();
1799 }
1800 if (!prefixcmp(command_buf.buf, "committer ")) {
1801 committer = parse_ident(command_buf.buf + 10);
1802 read_next_command();
1803 }
1804 if (!committer)
1805 die("Expected committer but didn't get one");
1806 msg = cmd_data(&msglen);
1807 read_next_command();
1808 cmd_from(b);
1809 merge_list = cmd_merge(&merge_count);
1810
1811 /* ensure the branch is active/loaded */
1812 if (!b->branch_tree.tree || !max_active_branches) {
1813 unload_one_branch();
1814 load_branch(b);
1815 }
1816
1817 /* file_change* */
1818 for (;;) {
1819 if (1 == command_buf.len)
1820 break;
1821 else if (!prefixcmp(command_buf.buf, "M "))
1822 file_change_m(b);
1823 else if (!prefixcmp(command_buf.buf, "D "))
1824 file_change_d(b);
1825 else if (!strcmp("deleteall", command_buf.buf))
1826 file_change_deleteall(b);
1827 else
1828 die("Unsupported file_change: %s", command_buf.buf);
1829 read_next_command();
1830 }
1831
1832 /* build the tree and the commit */
1833 store_tree(&b->branch_tree);
1834 hashcpy(b->branch_tree.versions[0].sha1,
1835 b->branch_tree.versions[1].sha1);
1836 size_dbuf(&new_data, 114 + msglen
1837 + merge_count * 49
1838 + (author
1839 ? strlen(author) + strlen(committer)
1840 : 2 * strlen(committer)));
1841 sp = new_data.buffer;
1842 sp += sprintf(sp, "tree %s\n",
1843 sha1_to_hex(b->branch_tree.versions[1].sha1));
1844 if (!is_null_sha1(b->sha1))
1845 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1846 while (merge_list) {
1847 struct hash_list *next = merge_list->next;
1848 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1849 free(merge_list);
1850 merge_list = next;
1851 }
1852 sp += sprintf(sp, "author %s\n", author ? author : committer);
1853 sp += sprintf(sp, "committer %s\n", committer);
1854 *sp++ = '\n';
1855 memcpy(sp, msg, msglen);
1856 sp += msglen;
1857 free(author);
1858 free(committer);
1859 free(msg);
1860
1861 if (!store_object(OBJ_COMMIT,
1862 new_data.buffer, sp - (char*)new_data.buffer,
1863 NULL, b->sha1, next_mark))
1864 b->pack_id = pack_id;
1865 b->last_commit = object_count_by_type[OBJ_COMMIT];
1866 }
1867
1868 static void cmd_new_tag(void)
1869 {
1870 char *sp;
1871 const char *from;
1872 char *tagger;
1873 struct branch *s;
1874 void *msg;
1875 size_t msglen;
1876 struct tag *t;
1877 uintmax_t from_mark = 0;
1878 unsigned char sha1[20];
1879
1880 /* Obtain the new tag name from the rest of our command */
1881 sp = strchr(command_buf.buf, ' ') + 1;
1882 t = pool_alloc(sizeof(struct tag));
1883 t->next_tag = NULL;
1884 t->name = pool_strdup(sp);
1885 if (last_tag)
1886 last_tag->next_tag = t;
1887 else
1888 first_tag = t;
1889 last_tag = t;
1890 read_next_command();
1891
1892 /* from ... */
1893 if (prefixcmp(command_buf.buf, "from "))
1894 die("Expected from command, got %s", command_buf.buf);
1895 from = strchr(command_buf.buf, ' ') + 1;
1896 s = lookup_branch(from);
1897 if (s) {
1898 hashcpy(sha1, s->sha1);
1899 } else if (*from == ':') {
1900 struct object_entry *oe;
1901 from_mark = strtoumax(from + 1, NULL, 10);
1902 oe = find_mark(from_mark);
1903 if (oe->type != OBJ_COMMIT)
1904 die("Mark :%" PRIuMAX " not a commit", from_mark);
1905 hashcpy(sha1, oe->sha1);
1906 } else if (!get_sha1(from, sha1)) {
1907 unsigned long size;
1908 char *buf;
1909
1910 buf = read_object_with_reference(sha1,
1911 commit_type, &size, sha1);
1912 if (!buf || size < 46)
1913 die("Not a valid commit: %s", from);
1914 free(buf);
1915 } else
1916 die("Invalid ref name or SHA1 expression: %s", from);
1917 read_next_command();
1918
1919 /* tagger ... */
1920 if (prefixcmp(command_buf.buf, "tagger "))
1921 die("Expected tagger command, got %s", command_buf.buf);
1922 tagger = parse_ident(command_buf.buf + 7);
1923
1924 /* tag payload/message */
1925 read_next_command();
1926 msg = cmd_data(&msglen);
1927
1928 /* build the tag object */
1929 size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1930 sp = new_data.buffer;
1931 sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1932 sp += sprintf(sp, "type %s\n", commit_type);
1933 sp += sprintf(sp, "tag %s\n", t->name);
1934 sp += sprintf(sp, "tagger %s\n", tagger);
1935 *sp++ = '\n';
1936 memcpy(sp, msg, msglen);
1937 sp += msglen;
1938 free(tagger);
1939 free(msg);
1940
1941 if (store_object(OBJ_TAG, new_data.buffer,
1942 sp - (char*)new_data.buffer,
1943 NULL, t->sha1, 0))
1944 t->pack_id = MAX_PACK_ID;
1945 else
1946 t->pack_id = pack_id;
1947 }
1948
1949 static void cmd_reset_branch(void)
1950 {
1951 struct branch *b;
1952 char *sp;
1953
1954 /* Obtain the branch name from the rest of our command */
1955 sp = strchr(command_buf.buf, ' ') + 1;
1956 b = lookup_branch(sp);
1957 if (b) {
1958 hashclr(b->sha1);
1959 hashclr(b->branch_tree.versions[0].sha1);
1960 hashclr(b->branch_tree.versions[1].sha1);
1961 if (b->branch_tree.tree) {
1962 release_tree_content_recursive(b->branch_tree.tree);
1963 b->branch_tree.tree = NULL;
1964 }
1965 }
1966 else
1967 b = new_branch(sp);
1968 read_next_command();
1969 cmd_from(b);
1970 }
1971
1972 static void cmd_checkpoint(void)
1973 {
1974 if (object_count) {
1975 cycle_packfile();
1976 dump_branches();
1977 dump_tags();
1978 dump_marks();
1979 }
1980 read_next_command();
1981 }
1982
1983 static const char fast_import_usage[] =
1984 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
1985
1986 int main(int argc, const char **argv)
1987 {
1988 int i, show_stats = 1;
1989
1990 git_config(git_default_config);
1991
1992 for (i = 1; i < argc; i++) {
1993 const char *a = argv[i];
1994
1995 if (*a != '-' || !strcmp(a, "--"))
1996 break;
1997 else if (!prefixcmp(a, "--date-format=")) {
1998 const char *fmt = a + 14;
1999 if (!strcmp(fmt, "raw"))
2000 whenspec = WHENSPEC_RAW;
2001 else if (!strcmp(fmt, "rfc2822"))
2002 whenspec = WHENSPEC_RFC2822;
2003 else if (!strcmp(fmt, "now"))
2004 whenspec = WHENSPEC_NOW;
2005 else
2006 die("unknown --date-format argument %s", fmt);
2007 }
2008 else if (!prefixcmp(a, "--max-pack-size="))
2009 max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2010 else if (!prefixcmp(a, "--depth="))
2011 max_depth = strtoul(a + 8, NULL, 0);
2012 else if (!prefixcmp(a, "--active-branches="))
2013 max_active_branches = strtoul(a + 18, NULL, 0);
2014 else if (!prefixcmp(a, "--export-marks="))
2015 mark_file = a + 15;
2016 else if (!prefixcmp(a, "--export-pack-edges=")) {
2017 if (pack_edges)
2018 fclose(pack_edges);
2019 pack_edges = fopen(a + 20, "a");
2020 if (!pack_edges)
2021 die("Cannot open %s: %s", a + 20, strerror(errno));
2022 } else if (!strcmp(a, "--force"))
2023 force_update = 1;
2024 else if (!strcmp(a, "--quiet"))
2025 show_stats = 0;
2026 else if (!strcmp(a, "--stats"))
2027 show_stats = 1;
2028 else
2029 die("unknown option %s", a);
2030 }
2031 if (i != argc)
2032 usage(fast_import_usage);
2033
2034 alloc_objects(object_entry_alloc);
2035 strbuf_init(&command_buf);
2036
2037 atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2038 branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2039 avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2040 marks = pool_calloc(1, sizeof(struct mark_set));
2041
2042 start_packfile();
2043 for (;;) {
2044 read_next_command();
2045 if (command_buf.eof)
2046 break;
2047 else if (!strcmp("blob", command_buf.buf))
2048 cmd_new_blob();
2049 else if (!prefixcmp(command_buf.buf, "commit "))
2050 cmd_new_commit();
2051 else if (!prefixcmp(command_buf.buf, "tag "))
2052 cmd_new_tag();
2053 else if (!prefixcmp(command_buf.buf, "reset "))
2054 cmd_reset_branch();
2055 else if (!strcmp("checkpoint", command_buf.buf))
2056 cmd_checkpoint();
2057 else
2058 die("Unsupported command: %s", command_buf.buf);
2059 }
2060 end_packfile();
2061
2062 dump_branches();
2063 dump_tags();
2064 unkeep_all_packs();
2065 dump_marks();
2066
2067 if (pack_edges)
2068 fclose(pack_edges);
2069
2070 if (show_stats) {
2071 uintmax_t total_count = 0, duplicate_count = 0;
2072 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2073 total_count += object_count_by_type[i];
2074 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2075 duplicate_count += duplicate_count_by_type[i];
2076
2077 fprintf(stderr, "%s statistics:\n", argv[0]);
2078 fprintf(stderr, "---------------------------------------------------------------------\n");
2079 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2080 fprintf(stderr, "Total objects: %10" PRIuMAX " (%10" PRIuMAX " duplicates )\n", total_count, duplicate_count);
2081 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]);
2082 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]);
2083 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]);
2084 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]);
2085 fprintf(stderr, "Total branches: %10lu (%10lu loads )\n", branch_count, branch_load_count);
2086 fprintf(stderr, " marks: %10" PRIuMAX " (%10" PRIuMAX " unique )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2087 fprintf(stderr, " atoms: %10u\n", atom_cnt);
2088 fprintf(stderr, "Memory total: %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2089 fprintf(stderr, " pools: %10lu KiB\n", (unsigned long)(total_allocd/1024));
2090 fprintf(stderr, " objects: %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2091 fprintf(stderr, "---------------------------------------------------------------------\n");
2092 pack_report();
2093 fprintf(stderr, "---------------------------------------------------------------------\n");
2094 fprintf(stderr, "\n");
2095 }
2096
2097 return failure ? 1 : 0;
2098 }
Jonathan Nieder’s miscellaneous git branches