?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
xsize_t: check whether we lose bits
[git/jnareb-git/bp-gitweb.git] / fast-import.c
blobfda50188eaf96d6450e802f54945f86a0776cf6f
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/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 void fixup_header_footer(void)
655 {
656 static const int buf_sz = 128 * 1024;
657 int pack_fd = pack_data->pack_fd;
658 SHA_CTX c;
659 struct pack_header hdr;
660 char *buf;
661
662 if (lseek(pack_fd, 0, SEEK_SET) != 0)
663 die("Failed seeking to start: %s", strerror(errno));
664 if (read_in_full(pack_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
665 die("Unable to reread header of %s", pack_data->pack_name);
666 if (lseek(pack_fd, 0, SEEK_SET) != 0)
667 die("Failed seeking to start: %s", strerror(errno));
668 hdr.hdr_entries = htonl(object_count);
669 write_or_die(pack_fd, &hdr, sizeof(hdr));
670
671 SHA1_Init(&c);
672 SHA1_Update(&c, &hdr, sizeof(hdr));
673
674 buf = xmalloc(buf_sz);
675 for (;;) {
676 size_t n = xread(pack_fd, buf, buf_sz);
677 if (!n)
678 break;
679 if (n < 0)
680 die("Failed to checksum %s", pack_data->pack_name);
681 SHA1_Update(&c, buf, n);
682 }
683 free(buf);
684
685 SHA1_Final(pack_data->sha1, &c);
686 write_or_die(pack_fd, pack_data->sha1, sizeof(pack_data->sha1));
687 close(pack_fd);
688 }
689
690 static int oecmp (const void *a_, const void *b_)
691 {
692 struct object_entry *a = *((struct object_entry**)a_);
693 struct object_entry *b = *((struct object_entry**)b_);
694 return hashcmp(a->sha1, b->sha1);
695 }
696
697 static char *create_index(void)
698 {
699 static char tmpfile[PATH_MAX];
700 SHA_CTX ctx;
701 struct sha1file *f;
702 struct object_entry **idx, **c, **last, *e;
703 struct object_entry_pool *o;
704 uint32_t array[256];
705 int i, idx_fd;
706
707 /* Build the sorted table of object IDs. */
708 idx = xmalloc(object_count * sizeof(struct object_entry*));
709 c = idx;
710 for (o = blocks; o; o = o->next_pool)
711 for (e = o->next_free; e-- != o->entries;)
712 if (pack_id == e->pack_id)
713 *c++ = e;
714 last = idx + object_count;
715 if (c != last)
716 die("internal consistency error creating the index");
717 qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
718
719 /* Generate the fan-out array. */
720 c = idx;
721 for (i = 0; i < 256; i++) {
722 struct object_entry **next = c;;
723 while (next < last) {
724 if ((*next)->sha1[0] != i)
725 break;
726 next++;
727 }
728 array[i] = htonl(next - idx);
729 c = next;
730 }
731
732 snprintf(tmpfile, sizeof(tmpfile),
733 "%s/index_XXXXXX", get_object_directory());
734 idx_fd = mkstemp(tmpfile);
735 if (idx_fd < 0)
736 die("Can't create %s: %s", tmpfile, strerror(errno));
737 f = sha1fd(idx_fd, tmpfile);
738 sha1write(f, array, 256 * sizeof(int));
739 SHA1_Init(&ctx);
740 for (c = idx; c != last; c++) {
741 uint32_t offset = htonl((*c)->offset);
742 sha1write(f, &offset, 4);
743 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
744 SHA1_Update(&ctx, (*c)->sha1, 20);
745 }
746 sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
747 sha1close(f, NULL, 1);
748 free(idx);
749 SHA1_Final(pack_data->sha1, &ctx);
750 return tmpfile;
751 }
752
753 static char *keep_pack(char *curr_index_name)
754 {
755 static char name[PATH_MAX];
756 static const char *keep_msg = "fast-import";
757 int keep_fd;
758
759 chmod(pack_data->pack_name, 0444);
760 chmod(curr_index_name, 0444);
761
762 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
763 get_object_directory(), sha1_to_hex(pack_data->sha1));
764 keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
765 if (keep_fd < 0)
766 die("cannot create keep file");
767 write(keep_fd, keep_msg, strlen(keep_msg));
768 close(keep_fd);
769
770 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
771 get_object_directory(), sha1_to_hex(pack_data->sha1));
772 if (move_temp_to_file(pack_data->pack_name, name))
773 die("cannot store pack file");
774
775 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
776 get_object_directory(), sha1_to_hex(pack_data->sha1));
777 if (move_temp_to_file(curr_index_name, name))
778 die("cannot store index file");
779 return name;
780 }
781
782 static void unkeep_all_packs(void)
783 {
784 static char name[PATH_MAX];
785 int k;
786
787 for (k = 0; k < pack_id; k++) {
788 struct packed_git *p = all_packs[k];
789 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
790 get_object_directory(), sha1_to_hex(p->sha1));
791 unlink(name);
792 }
793 }
794
795 static void end_packfile(void)
796 {
797 struct packed_git *old_p = pack_data, *new_p;
798
799 if (object_count) {
800 char *idx_name;
801 int i;
802 struct branch *b;
803 struct tag *t;
804
805 fixup_header_footer();
806 idx_name = keep_pack(create_index());
807
808 /* Register the packfile with core git's machinary. */
809 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
810 if (!new_p)
811 die("core git rejected index %s", idx_name);
812 new_p->windows = old_p->windows;
813 all_packs[pack_id] = new_p;
814 install_packed_git(new_p);
815
816 /* Print the boundary */
817 if (pack_edges) {
818 fprintf(pack_edges, "%s:", new_p->pack_name);
819 for (i = 0; i < branch_table_sz; i++) {
820 for (b = branch_table[i]; b; b = b->table_next_branch) {
821 if (b->pack_id == pack_id)
822 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
823 }
824 }
825 for (t = first_tag; t; t = t->next_tag) {
826 if (t->pack_id == pack_id)
827 fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
828 }
829 fputc('\n', pack_edges);
830 fflush(pack_edges);
831 }
832
833 pack_id++;
834 }
835 else
836 unlink(old_p->pack_name);
837 free(old_p);
838
839 /* We can't carry a delta across packfiles. */
840 free(last_blob.data);
841 last_blob.data = NULL;
842 last_blob.len = 0;
843 last_blob.offset = 0;
844 last_blob.depth = 0;
845 }
846
847 static void cycle_packfile(void)
848 {
849 end_packfile();
850 start_packfile();
851 }
852
853 static size_t encode_header(
854 enum object_type type,
855 size_t size,
856 unsigned char *hdr)
857 {
858 int n = 1;
859 unsigned char c;
860
861 if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
862 die("bad type %d", type);
863
864 c = (type << 4) | (size & 15);
865 size >>= 4;
866 while (size) {
867 *hdr++ = c | 0x80;
868 c = size & 0x7f;
869 size >>= 7;
870 n++;
871 }
872 *hdr = c;
873 return n;
874 }
875
876 static int store_object(
877 enum object_type type,
878 void *dat,
879 size_t datlen,
880 struct last_object *last,
881 unsigned char *sha1out,
882 uintmax_t mark)
883 {
884 void *out, *delta;
885 struct object_entry *e;
886 unsigned char hdr[96];
887 unsigned char sha1[20];
888 unsigned long hdrlen, deltalen;
889 SHA_CTX c;
890 z_stream s;
891
892 hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
893 (unsigned long)datlen) + 1;
894 SHA1_Init(&c);
895 SHA1_Update(&c, hdr, hdrlen);
896 SHA1_Update(&c, dat, datlen);
897 SHA1_Final(sha1, &c);
898 if (sha1out)
899 hashcpy(sha1out, sha1);
900
901 e = insert_object(sha1);
902 if (mark)
903 insert_mark(mark, e);
904 if (e->offset) {
905 duplicate_count_by_type[type]++;
906 return 1;
907 }
908
909 if (last && last->data && last->depth < max_depth) {
910 delta = diff_delta(last->data, last->len,
911 dat, datlen,
912 &deltalen, 0);
913 if (delta && deltalen >= datlen) {
914 free(delta);
915 delta = NULL;
916 }
917 } else
918 delta = NULL;
919
920 memset(&s, 0, sizeof(s));
921 deflateInit(&s, zlib_compression_level);
922 if (delta) {
923 s.next_in = delta;
924 s.avail_in = deltalen;
925 } else {
926 s.next_in = dat;
927 s.avail_in = datlen;
928 }
929 s.avail_out = deflateBound(&s, s.avail_in);
930 s.next_out = out = xmalloc(s.avail_out);
931 while (deflate(&s, Z_FINISH) == Z_OK)
932 /* nothing */;
933 deflateEnd(&s);
934
935 /* Determine if we should auto-checkpoint. */
936 if ((pack_size + 60 + s.total_out) > max_packsize
937 || (pack_size + 60 + s.total_out) < pack_size) {
938
939 /* This new object needs to *not* have the current pack_id. */
940 e->pack_id = pack_id + 1;
941 cycle_packfile();
942
943 /* We cannot carry a delta into the new pack. */
944 if (delta) {
945 free(delta);
946 delta = NULL;
947
948 memset(&s, 0, sizeof(s));
949 deflateInit(&s, zlib_compression_level);
950 s.next_in = dat;
951 s.avail_in = datlen;
952 s.avail_out = deflateBound(&s, s.avail_in);
953 s.next_out = out = xrealloc(out, s.avail_out);
954 while (deflate(&s, Z_FINISH) == Z_OK)
955 /* nothing */;
956 deflateEnd(&s);
957 }
958 }
959
960 e->type = type;
961 e->pack_id = pack_id;
962 e->offset = pack_size;
963 object_count++;
964 object_count_by_type[type]++;
965
966 if (delta) {
967 unsigned long ofs = e->offset - last->offset;
968 unsigned pos = sizeof(hdr) - 1;
969
970 delta_count_by_type[type]++;
971 last->depth++;
972
973 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
974 write_or_die(pack_data->pack_fd, hdr, hdrlen);
975 pack_size += hdrlen;
976
977 hdr[pos] = ofs & 127;
978 while (ofs >>= 7)
979 hdr[--pos] = 128 | (--ofs & 127);
980 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
981 pack_size += sizeof(hdr) - pos;
982 } else {
983 if (last)
984 last->depth = 0;
985 hdrlen = encode_header(type, datlen, hdr);
986 write_or_die(pack_data->pack_fd, hdr, hdrlen);
987 pack_size += hdrlen;
988 }
989
990 write_or_die(pack_data->pack_fd, out, s.total_out);
991 pack_size += s.total_out;
992
993 free(out);
994 free(delta);
995 if (last) {
996 if (!last->no_free)
997 free(last->data);
998 last->data = dat;
999 last->offset = e->offset;
1000 last->len = datlen;
1001 }
1002 return 0;
1003 }
1004
1005 static void *gfi_unpack_entry(
1006 struct object_entry *oe,
1007 unsigned long *sizep)
1008 {
1009 enum object_type type;
1010 struct packed_git *p = all_packs[oe->pack_id];
1011 if (p == pack_data)
1012 p->pack_size = pack_size + 20;
1013 return unpack_entry(p, oe->offset, &type, sizep);
1014 }
1015
1016 static const char *get_mode(const char *str, uint16_t *modep)
1017 {
1018 unsigned char c;
1019 uint16_t mode = 0;
1020
1021 while ((c = *str++) != ' ') {
1022 if (c < '0' || c > '7')
1023 return NULL;
1024 mode = (mode << 3) + (c - '0');
1025 }
1026 *modep = mode;
1027 return str;
1028 }
1029
1030 static void load_tree(struct tree_entry *root)
1031 {
1032 unsigned char* sha1 = root->versions[1].sha1;
1033 struct object_entry *myoe;
1034 struct tree_content *t;
1035 unsigned long size;
1036 char *buf;
1037 const char *c;
1038
1039 root->tree = t = new_tree_content(8);
1040 if (is_null_sha1(sha1))
1041 return;
1042
1043 myoe = find_object(sha1);
1044 if (myoe) {
1045 if (myoe->type != OBJ_TREE)
1046 die("Not a tree: %s", sha1_to_hex(sha1));
1047 t->delta_depth = 0;
1048 buf = gfi_unpack_entry(myoe, &size);
1049 } else {
1050 enum object_type type;
1051 buf = read_sha1_file(sha1, &type, &size);
1052 if (!buf || type != OBJ_TREE)
1053 die("Can't load tree %s", sha1_to_hex(sha1));
1054 }
1055
1056 c = buf;
1057 while (c != (buf + size)) {
1058 struct tree_entry *e = new_tree_entry();
1059
1060 if (t->entry_count == t->entry_capacity)
1061 root->tree = t = grow_tree_content(t, 8);
1062 t->entries[t->entry_count++] = e;
1063
1064 e->tree = NULL;
1065 c = get_mode(c, &e->versions[1].mode);
1066 if (!c)
1067 die("Corrupt mode in %s", sha1_to_hex(sha1));
1068 e->versions[0].mode = e->versions[1].mode;
1069 e->name = to_atom(c, (unsigned short)strlen(c));
1070 c += e->name->str_len + 1;
1071 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1072 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1073 c += 20;
1074 }
1075 free(buf);
1076 }
1077
1078 static int tecmp0 (const void *_a, const void *_b)
1079 {
1080 struct tree_entry *a = *((struct tree_entry**)_a);
1081 struct tree_entry *b = *((struct tree_entry**)_b);
1082 return base_name_compare(
1083 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1084 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1085 }
1086
1087 static int tecmp1 (const void *_a, const void *_b)
1088 {
1089 struct tree_entry *a = *((struct tree_entry**)_a);
1090 struct tree_entry *b = *((struct tree_entry**)_b);
1091 return base_name_compare(
1092 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1093 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1094 }
1095
1096 static void mktree(struct tree_content *t,
1097 int v,
1098 unsigned long *szp,
1099 struct dbuf *b)
1100 {
1101 size_t maxlen = 0;
1102 unsigned int i;
1103 char *c;
1104
1105 if (!v)
1106 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1107 else
1108 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1109
1110 for (i = 0; i < t->entry_count; i++) {
1111 if (t->entries[i]->versions[v].mode)
1112 maxlen += t->entries[i]->name->str_len + 34;
1113 }
1114
1115 size_dbuf(b, maxlen);
1116 c = b->buffer;
1117 for (i = 0; i < t->entry_count; i++) {
1118 struct tree_entry *e = t->entries[i];
1119 if (!e->versions[v].mode)
1120 continue;
1121 c += sprintf(c, "%o", (unsigned int)e->versions[v].mode);
1122 *c++ = ' ';
1123 strcpy(c, e->name->str_dat);
1124 c += e->name->str_len + 1;
1125 hashcpy((unsigned char*)c, e->versions[v].sha1);
1126 c += 20;
1127 }
1128 *szp = c - (char*)b->buffer;
1129 }
1130
1131 static void store_tree(struct tree_entry *root)
1132 {
1133 struct tree_content *t = root->tree;
1134 unsigned int i, j, del;
1135 unsigned long new_len;
1136 struct last_object lo;
1137 struct object_entry *le;
1138
1139 if (!is_null_sha1(root->versions[1].sha1))
1140 return;
1141
1142 for (i = 0; i < t->entry_count; i++) {
1143 if (t->entries[i]->tree)
1144 store_tree(t->entries[i]);
1145 }
1146
1147 le = find_object(root->versions[0].sha1);
1148 if (!S_ISDIR(root->versions[0].mode)
1149 || !le
1150 || le->pack_id != pack_id) {
1151 lo.data = NULL;
1152 lo.depth = 0;
1153 } else {
1154 mktree(t, 0, &lo.len, &old_tree);
1155 lo.data = old_tree.buffer;
1156 lo.offset = le->offset;
1157 lo.depth = t->delta_depth;
1158 lo.no_free = 1;
1159 }
1160
1161 mktree(t, 1, &new_len, &new_tree);
1162 store_object(OBJ_TREE, new_tree.buffer, new_len,
1163 &lo, root->versions[1].sha1, 0);
1164
1165 t->delta_depth = lo.depth;
1166 for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1167 struct tree_entry *e = t->entries[i];
1168 if (e->versions[1].mode) {
1169 e->versions[0].mode = e->versions[1].mode;
1170 hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1171 t->entries[j++] = e;
1172 } else {
1173 release_tree_entry(e);
1174 del++;
1175 }
1176 }
1177 t->entry_count -= del;
1178 }
1179
1180 static int tree_content_set(
1181 struct tree_entry *root,
1182 const char *p,
1183 const unsigned char *sha1,
1184 const uint16_t mode)
1185 {
1186 struct tree_content *t = root->tree;
1187 const char *slash1;
1188 unsigned int i, n;
1189 struct tree_entry *e;
1190
1191 slash1 = strchr(p, '/');
1192 if (slash1)
1193 n = slash1 - p;
1194 else
1195 n = strlen(p);
1196
1197 for (i = 0; i < t->entry_count; i++) {
1198 e = t->entries[i];
1199 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1200 if (!slash1) {
1201 if (e->versions[1].mode == mode
1202 && !hashcmp(e->versions[1].sha1, sha1))
1203 return 0;
1204 e->versions[1].mode = mode;
1205 hashcpy(e->versions[1].sha1, sha1);
1206 if (e->tree) {
1207 release_tree_content_recursive(e->tree);
1208 e->tree = NULL;
1209 }
1210 hashclr(root->versions[1].sha1);
1211 return 1;
1212 }
1213 if (!S_ISDIR(e->versions[1].mode)) {
1214 e->tree = new_tree_content(8);
1215 e->versions[1].mode = S_IFDIR;
1216 }
1217 if (!e->tree)
1218 load_tree(e);
1219 if (tree_content_set(e, slash1 + 1, sha1, mode)) {
1220 hashclr(root->versions[1].sha1);
1221 return 1;
1222 }
1223 return 0;
1224 }
1225 }
1226
1227 if (t->entry_count == t->entry_capacity)
1228 root->tree = t = grow_tree_content(t, 8);
1229 e = new_tree_entry();
1230 e->name = to_atom(p, (unsigned short)n);
1231 e->versions[0].mode = 0;
1232 hashclr(e->versions[0].sha1);
1233 t->entries[t->entry_count++] = e;
1234 if (slash1) {
1235 e->tree = new_tree_content(8);
1236 e->versions[1].mode = S_IFDIR;
1237 tree_content_set(e, slash1 + 1, sha1, mode);
1238 } else {
1239 e->tree = NULL;
1240 e->versions[1].mode = mode;
1241 hashcpy(e->versions[1].sha1, sha1);
1242 }
1243 hashclr(root->versions[1].sha1);
1244 return 1;
1245 }
1246
1247 static int tree_content_remove(struct tree_entry *root, const char *p)
1248 {
1249 struct tree_content *t = root->tree;
1250 const char *slash1;
1251 unsigned int i, n;
1252 struct tree_entry *e;
1253
1254 slash1 = strchr(p, '/');
1255 if (slash1)
1256 n = slash1 - p;
1257 else
1258 n = strlen(p);
1259
1260 for (i = 0; i < t->entry_count; i++) {
1261 e = t->entries[i];
1262 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1263 if (!slash1 || !S_ISDIR(e->versions[1].mode))
1264 goto del_entry;
1265 if (!e->tree)
1266 load_tree(e);
1267 if (tree_content_remove(e, slash1 + 1)) {
1268 for (n = 0; n < e->tree->entry_count; n++) {
1269 if (e->tree->entries[n]->versions[1].mode) {
1270 hashclr(root->versions[1].sha1);
1271 return 1;
1272 }
1273 }
1274 goto del_entry;
1275 }
1276 return 0;
1277 }
1278 }
1279 return 0;
1280
1281 del_entry:
1282 if (e->tree) {
1283 release_tree_content_recursive(e->tree);
1284 e->tree = NULL;
1285 }
1286 e->versions[1].mode = 0;
1287 hashclr(e->versions[1].sha1);
1288 hashclr(root->versions[1].sha1);
1289 return 1;
1290 }
1291
1292 static int update_branch(struct branch *b)
1293 {
1294 static const char *msg = "fast-import";
1295 struct ref_lock *lock;
1296 unsigned char old_sha1[20];
1297
1298 if (read_ref(b->name, old_sha1))
1299 hashclr(old_sha1);
1300 lock = lock_any_ref_for_update(b->name, old_sha1);
1301 if (!lock)
1302 return error("Unable to lock %s", b->name);
1303 if (!force_update && !is_null_sha1(old_sha1)) {
1304 struct commit *old_cmit, *new_cmit;
1305
1306 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1307 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1308 if (!old_cmit || !new_cmit) {
1309 unlock_ref(lock);
1310 return error("Branch %s is missing commits.", b->name);
1311 }
1312
1313 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1314 unlock_ref(lock);
1315 warn("Not updating %s"
1316 " (new tip %s does not contain %s)",
1317 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1318 return -1;
1319 }
1320 }
1321 if (write_ref_sha1(lock, b->sha1, msg) < 0)
1322 return error("Unable to update %s", b->name);
1323 return 0;
1324 }
1325
1326 static void dump_branches(void)
1327 {
1328 unsigned int i;
1329 struct branch *b;
1330
1331 for (i = 0; i < branch_table_sz; i++) {
1332 for (b = branch_table[i]; b; b = b->table_next_branch)
1333 failure |= update_branch(b);
1334 }
1335 }
1336
1337 static void dump_tags(void)
1338 {
1339 static const char *msg = "fast-import";
1340 struct tag *t;
1341 struct ref_lock *lock;
1342 char ref_name[PATH_MAX];
1343
1344 for (t = first_tag; t; t = t->next_tag) {
1345 sprintf(ref_name, "tags/%s", t->name);
1346 lock = lock_ref_sha1(ref_name, NULL);
1347 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1348 failure |= error("Unable to update %s", ref_name);
1349 }
1350 }
1351
1352 static void dump_marks_helper(FILE *f,
1353 uintmax_t base,
1354 struct mark_set *m)
1355 {
1356 uintmax_t k;
1357 if (m->shift) {
1358 for (k = 0; k < 1024; k++) {
1359 if (m->data.sets[k])
1360 dump_marks_helper(f, (base + k) << m->shift,
1361 m->data.sets[k]);
1362 }
1363 } else {
1364 for (k = 0; k < 1024; k++) {
1365 if (m->data.marked[k])
1366 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1367 sha1_to_hex(m->data.marked[k]->sha1));
1368 }
1369 }
1370 }
1371
1372 static void dump_marks(void)
1373 {
1374 if (mark_file)
1375 {
1376 FILE *f = fopen(mark_file, "w");
1377 if (f) {
1378 dump_marks_helper(f, 0, marks);
1379 fclose(f);
1380 } else
1381 failure |= error("Unable to write marks file %s: %s",
1382 mark_file, strerror(errno));
1383 }
1384 }
1385
1386 static void read_next_command(void)
1387 {
1388 read_line(&command_buf, stdin, '\n');
1389 }
1390
1391 static void cmd_mark(void)
1392 {
1393 if (!prefixcmp(command_buf.buf, "mark :")) {
1394 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1395 read_next_command();
1396 }
1397 else
1398 next_mark = 0;
1399 }
1400
1401 static void *cmd_data (size_t *size)
1402 {
1403 size_t length;
1404 char *buffer;
1405
1406 if (prefixcmp(command_buf.buf, "data "))
1407 die("Expected 'data n' command, found: %s", command_buf.buf);
1408
1409 if (!prefixcmp(command_buf.buf + 5, "<<")) {
1410 char *term = xstrdup(command_buf.buf + 5 + 2);
1411 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1412 length = 0;
1413 buffer = xmalloc(sz);
1414 for (;;) {
1415 read_next_command();
1416 if (command_buf.eof)
1417 die("EOF in data (terminator '%s' not found)", term);
1418 if (term_len == command_buf.len
1419 && !strcmp(term, command_buf.buf))
1420 break;
1421 if (sz < (length + command_buf.len)) {
1422 sz = sz * 3 / 2 + 16;
1423 if (sz < (length + command_buf.len))
1424 sz = length + command_buf.len;
1425 buffer = xrealloc(buffer, sz);
1426 }
1427 memcpy(buffer + length,
1428 command_buf.buf,
1429 command_buf.len - 1);
1430 length += command_buf.len - 1;
1431 buffer[length++] = '\n';
1432 }
1433 free(term);
1434 }
1435 else {
1436 size_t n = 0;
1437 length = strtoul(command_buf.buf + 5, NULL, 10);
1438 buffer = xmalloc(length);
1439 while (n < length) {
1440 size_t s = fread(buffer + n, 1, length - n, stdin);
1441 if (!s && feof(stdin))
1442 die("EOF in data (%lu bytes remaining)",
1443 (unsigned long)(length - n));
1444 n += s;
1445 }
1446 }
1447
1448 if (fgetc(stdin) != '\n')
1449 die("An lf did not trail the binary data as expected.");
1450
1451 *size = length;
1452 return buffer;
1453 }
1454
1455 static int validate_raw_date(const char *src, char *result, int maxlen)
1456 {
1457 const char *orig_src = src;
1458 char *endp, sign;
1459
1460 strtoul(src, &endp, 10);
1461 if (endp == src || *endp != ' ')
1462 return -1;
1463
1464 src = endp + 1;
1465 if (*src != '-' && *src != '+')
1466 return -1;
1467 sign = *src;
1468
1469 strtoul(src + 1, &endp, 10);
1470 if (endp == src || *endp || (endp - orig_src) >= maxlen)
1471 return -1;
1472
1473 strcpy(result, orig_src);
1474 return 0;
1475 }
1476
1477 static char *parse_ident(const char *buf)
1478 {
1479 const char *gt;
1480 size_t name_len;
1481 char *ident;
1482
1483 gt = strrchr(buf, '>');
1484 if (!gt)
1485 die("Missing > in ident string: %s", buf);
1486 gt++;
1487 if (*gt != ' ')
1488 die("Missing space after > in ident string: %s", buf);
1489 gt++;
1490 name_len = gt - buf;
1491 ident = xmalloc(name_len + 24);
1492 strncpy(ident, buf, name_len);
1493
1494 switch (whenspec) {
1495 case WHENSPEC_RAW:
1496 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1497 die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1498 break;
1499 case WHENSPEC_RFC2822:
1500 if (parse_date(gt, ident + name_len, 24) < 0)
1501 die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1502 break;
1503 case WHENSPEC_NOW:
1504 if (strcmp("now", gt))
1505 die("Date in ident must be 'now': %s", buf);
1506 datestamp(ident + name_len, 24);
1507 break;
1508 }
1509
1510 return ident;
1511 }
1512
1513 static void cmd_new_blob(void)
1514 {
1515 size_t l;
1516 void *d;
1517
1518 read_next_command();
1519 cmd_mark();
1520 d = cmd_data(&l);
1521
1522 if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1523 free(d);
1524 }
1525
1526 static void unload_one_branch(void)
1527 {
1528 while (cur_active_branches
1529 && cur_active_branches >= max_active_branches) {
1530 uintmax_t min_commit = ULONG_MAX;
1531 struct branch *e, *l = NULL, *p = NULL;
1532
1533 for (e = active_branches; e; e = e->active_next_branch) {
1534 if (e->last_commit < min_commit) {
1535 p = l;
1536 min_commit = e->last_commit;
1537 }
1538 l = e;
1539 }
1540
1541 if (p) {
1542 e = p->active_next_branch;
1543 p->active_next_branch = e->active_next_branch;
1544 } else {
1545 e = active_branches;
1546 active_branches = e->active_next_branch;
1547 }
1548 e->active = 0;
1549 e->active_next_branch = NULL;
1550 if (e->branch_tree.tree) {
1551 release_tree_content_recursive(e->branch_tree.tree);
1552 e->branch_tree.tree = NULL;
1553 }
1554 cur_active_branches--;
1555 }
1556 }
1557
1558 static void load_branch(struct branch *b)
1559 {
1560 load_tree(&b->branch_tree);
1561 if (!b->active) {
1562 b->active = 1;
1563 b->active_next_branch = active_branches;
1564 active_branches = b;
1565 cur_active_branches++;
1566 branch_load_count++;
1567 }
1568 }
1569
1570 static void file_change_m(struct branch *b)
1571 {
1572 const char *p = command_buf.buf + 2;
1573 char *p_uq;
1574 const char *endp;
1575 struct object_entry *oe = oe;
1576 unsigned char sha1[20];
1577 uint16_t mode, inline_data = 0;
1578
1579 p = get_mode(p, &mode);
1580 if (!p)
1581 die("Corrupt mode: %s", command_buf.buf);
1582 switch (mode) {
1583 case S_IFREG | 0644:
1584 case S_IFREG | 0755:
1585 case S_IFLNK:
1586 case 0644:
1587 case 0755:
1588 /* ok */
1589 break;
1590 default:
1591 die("Corrupt mode: %s", command_buf.buf);
1592 }
1593
1594 if (*p == ':') {
1595 char *x;
1596 oe = find_mark(strtoumax(p + 1, &x, 10));
1597 hashcpy(sha1, oe->sha1);
1598 p = x;
1599 } else if (!prefixcmp(p, "inline")) {
1600 inline_data = 1;
1601 p += 6;
1602 } else {
1603 if (get_sha1_hex(p, sha1))
1604 die("Invalid SHA1: %s", command_buf.buf);
1605 oe = find_object(sha1);
1606 p += 40;
1607 }
1608 if (*p++ != ' ')
1609 die("Missing space after SHA1: %s", command_buf.buf);
1610
1611 p_uq = unquote_c_style(p, &endp);
1612 if (p_uq) {
1613 if (*endp)
1614 die("Garbage after path in: %s", command_buf.buf);
1615 p = p_uq;
1616 }
1617
1618 if (inline_data) {
1619 size_t l;
1620 void *d;
1621 if (!p_uq)
1622 p = p_uq = xstrdup(p);
1623 read_next_command();
1624 d = cmd_data(&l);
1625 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1626 free(d);
1627 } else if (oe) {
1628 if (oe->type != OBJ_BLOB)
1629 die("Not a blob (actually a %s): %s",
1630 command_buf.buf, typename(oe->type));
1631 } else {
1632 enum object_type type = sha1_object_info(sha1, NULL);
1633 if (type < 0)
1634 die("Blob not found: %s", command_buf.buf);
1635 if (type != OBJ_BLOB)
1636 die("Not a blob (actually a %s): %s",
1637 typename(type), command_buf.buf);
1638 }
1639
1640 tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
1641 free(p_uq);
1642 }
1643
1644 static void file_change_d(struct branch *b)
1645 {
1646 const char *p = command_buf.buf + 2;
1647 char *p_uq;
1648 const char *endp;
1649
1650 p_uq = unquote_c_style(p, &endp);
1651 if (p_uq) {
1652 if (*endp)
1653 die("Garbage after path in: %s", command_buf.buf);
1654 p = p_uq;
1655 }
1656 tree_content_remove(&b->branch_tree, p);
1657 free(p_uq);
1658 }
1659
1660 static void file_change_deleteall(struct branch *b)
1661 {
1662 release_tree_content_recursive(b->branch_tree.tree);
1663 hashclr(b->branch_tree.versions[0].sha1);
1664 hashclr(b->branch_tree.versions[1].sha1);
1665 load_tree(&b->branch_tree);
1666 }
1667
1668 static void cmd_from(struct branch *b)
1669 {
1670 const char *from;
1671 struct branch *s;
1672
1673 if (prefixcmp(command_buf.buf, "from "))
1674 return;
1675
1676 if (b->branch_tree.tree) {
1677 release_tree_content_recursive(b->branch_tree.tree);
1678 b->branch_tree.tree = NULL;
1679 }
1680
1681 from = strchr(command_buf.buf, ' ') + 1;
1682 s = lookup_branch(from);
1683 if (b == s)
1684 die("Can't create a branch from itself: %s", b->name);
1685 else if (s) {
1686 unsigned char *t = s->branch_tree.versions[1].sha1;
1687 hashcpy(b->sha1, s->sha1);
1688 hashcpy(b->branch_tree.versions[0].sha1, t);
1689 hashcpy(b->branch_tree.versions[1].sha1, t);
1690 } else if (*from == ':') {
1691 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1692 struct object_entry *oe = find_mark(idnum);
1693 unsigned long size;
1694 char *buf;
1695 if (oe->type != OBJ_COMMIT)
1696 die("Mark :%" PRIuMAX " not a commit", idnum);
1697 hashcpy(b->sha1, oe->sha1);
1698 buf = gfi_unpack_entry(oe, &size);
1699 if (!buf || size < 46)
1700 die("Not a valid commit: %s", from);
1701 if (memcmp("tree ", buf, 5)
1702 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1703 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1704 free(buf);
1705 hashcpy(b->branch_tree.versions[0].sha1,
1706 b->branch_tree.versions[1].sha1);
1707 } else if (!get_sha1(from, b->sha1)) {
1708 if (is_null_sha1(b->sha1)) {
1709 hashclr(b->branch_tree.versions[0].sha1);
1710 hashclr(b->branch_tree.versions[1].sha1);
1711 } else {
1712 unsigned long size;
1713 char *buf;
1714
1715 buf = read_object_with_reference(b->sha1,
1716 commit_type, &size, b->sha1);
1717 if (!buf || size < 46)
1718 die("Not a valid commit: %s", from);
1719 if (memcmp("tree ", buf, 5)
1720 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1721 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1722 free(buf);
1723 hashcpy(b->branch_tree.versions[0].sha1,
1724 b->branch_tree.versions[1].sha1);
1725 }
1726 } else
1727 die("Invalid ref name or SHA1 expression: %s", from);
1728
1729 read_next_command();
1730 }
1731
1732 static struct hash_list *cmd_merge(unsigned int *count)
1733 {
1734 struct hash_list *list = NULL, *n, *e = e;
1735 const char *from;
1736 struct branch *s;
1737
1738 *count = 0;
1739 while (!prefixcmp(command_buf.buf, "merge ")) {
1740 from = strchr(command_buf.buf, ' ') + 1;
1741 n = xmalloc(sizeof(*n));
1742 s = lookup_branch(from);
1743 if (s)
1744 hashcpy(n->sha1, s->sha1);
1745 else if (*from == ':') {
1746 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1747 struct object_entry *oe = find_mark(idnum);
1748 if (oe->type != OBJ_COMMIT)
1749 die("Mark :%" PRIuMAX " not a commit", idnum);
1750 hashcpy(n->sha1, oe->sha1);
1751 } else if (!get_sha1(from, n->sha1)) {
1752 unsigned long size;
1753 char *buf = read_object_with_reference(n->sha1,
1754 commit_type, &size, n->sha1);
1755 if (!buf || size < 46)
1756 die("Not a valid commit: %s", from);
1757 free(buf);
1758 } else
1759 die("Invalid ref name or SHA1 expression: %s", from);
1760
1761 n->next = NULL;
1762 if (list)
1763 e->next = n;
1764 else
1765 list = n;
1766 e = n;
1767 (*count)++;
1768 read_next_command();
1769 }
1770 return list;
1771 }
1772
1773 static void cmd_new_commit(void)
1774 {
1775 struct branch *b;
1776 void *msg;
1777 size_t msglen;
1778 char *sp;
1779 char *author = NULL;
1780 char *committer = NULL;
1781 struct hash_list *merge_list = NULL;
1782 unsigned int merge_count;
1783
1784 /* Obtain the branch name from the rest of our command */
1785 sp = strchr(command_buf.buf, ' ') + 1;
1786 b = lookup_branch(sp);
1787 if (!b)
1788 b = new_branch(sp);
1789
1790 read_next_command();
1791 cmd_mark();
1792 if (!prefixcmp(command_buf.buf, "author ")) {
1793 author = parse_ident(command_buf.buf + 7);
1794 read_next_command();
1795 }
1796 if (!prefixcmp(command_buf.buf, "committer ")) {
1797 committer = parse_ident(command_buf.buf + 10);
1798 read_next_command();
1799 }
1800 if (!committer)
1801 die("Expected committer but didn't get one");
1802 msg = cmd_data(&msglen);
1803 read_next_command();
1804 cmd_from(b);
1805 merge_list = cmd_merge(&merge_count);
1806
1807 /* ensure the branch is active/loaded */
1808 if (!b->branch_tree.tree || !max_active_branches) {
1809 unload_one_branch();
1810 load_branch(b);
1811 }
1812
1813 /* file_change* */
1814 for (;;) {
1815 if (1 == command_buf.len)
1816 break;
1817 else if (!prefixcmp(command_buf.buf, "M "))
1818 file_change_m(b);
1819 else if (!prefixcmp(command_buf.buf, "D "))
1820 file_change_d(b);
1821 else if (!strcmp("deleteall", command_buf.buf))
1822 file_change_deleteall(b);
1823 else
1824 die("Unsupported file_change: %s", command_buf.buf);
1825 read_next_command();
1826 }
1827
1828 /* build the tree and the commit */
1829 store_tree(&b->branch_tree);
1830 hashcpy(b->branch_tree.versions[0].sha1,
1831 b->branch_tree.versions[1].sha1);
1832 size_dbuf(&new_data, 114 + msglen
1833 + merge_count * 49
1834 + (author
1835 ? strlen(author) + strlen(committer)
1836 : 2 * strlen(committer)));
1837 sp = new_data.buffer;
1838 sp += sprintf(sp, "tree %s\n",
1839 sha1_to_hex(b->branch_tree.versions[1].sha1));
1840 if (!is_null_sha1(b->sha1))
1841 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1842 while (merge_list) {
1843 struct hash_list *next = merge_list->next;
1844 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1845 free(merge_list);
1846 merge_list = next;
1847 }
1848 sp += sprintf(sp, "author %s\n", author ? author : committer);
1849 sp += sprintf(sp, "committer %s\n", committer);
1850 *sp++ = '\n';
1851 memcpy(sp, msg, msglen);
1852 sp += msglen;
1853 free(author);
1854 free(committer);
1855 free(msg);
1856
1857 if (!store_object(OBJ_COMMIT,
1858 new_data.buffer, sp - (char*)new_data.buffer,
1859 NULL, b->sha1, next_mark))
1860 b->pack_id = pack_id;
1861 b->last_commit = object_count_by_type[OBJ_COMMIT];
1862 }
1863
1864 static void cmd_new_tag(void)
1865 {
1866 char *sp;
1867 const char *from;
1868 char *tagger;
1869 struct branch *s;
1870 void *msg;
1871 size_t msglen;
1872 struct tag *t;
1873 uintmax_t from_mark = 0;
1874 unsigned char sha1[20];
1875
1876 /* Obtain the new tag name from the rest of our command */
1877 sp = strchr(command_buf.buf, ' ') + 1;
1878 t = pool_alloc(sizeof(struct tag));
1879 t->next_tag = NULL;
1880 t->name = pool_strdup(sp);
1881 if (last_tag)
1882 last_tag->next_tag = t;
1883 else
1884 first_tag = t;
1885 last_tag = t;
1886 read_next_command();
1887
1888 /* from ... */
1889 if (prefixcmp(command_buf.buf, "from "))
1890 die("Expected from command, got %s", command_buf.buf);
1891 from = strchr(command_buf.buf, ' ') + 1;
1892 s = lookup_branch(from);
1893 if (s) {
1894 hashcpy(sha1, s->sha1);
1895 } else if (*from == ':') {
1896 struct object_entry *oe;
1897 from_mark = strtoumax(from + 1, NULL, 10);
1898 oe = find_mark(from_mark);
1899 if (oe->type != OBJ_COMMIT)
1900 die("Mark :%" PRIuMAX " not a commit", from_mark);
1901 hashcpy(sha1, oe->sha1);
1902 } else if (!get_sha1(from, sha1)) {
1903 unsigned long size;
1904 char *buf;
1905
1906 buf = read_object_with_reference(sha1,
1907 commit_type, &size, sha1);
1908 if (!buf || size < 46)
1909 die("Not a valid commit: %s", from);
1910 free(buf);
1911 } else
1912 die("Invalid ref name or SHA1 expression: %s", from);
1913 read_next_command();
1914
1915 /* tagger ... */
1916 if (prefixcmp(command_buf.buf, "tagger "))
1917 die("Expected tagger command, got %s", command_buf.buf);
1918 tagger = parse_ident(command_buf.buf + 7);
1919
1920 /* tag payload/message */
1921 read_next_command();
1922 msg = cmd_data(&msglen);
1923
1924 /* build the tag object */
1925 size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
1926 sp = new_data.buffer;
1927 sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
1928 sp += sprintf(sp, "type %s\n", commit_type);
1929 sp += sprintf(sp, "tag %s\n", t->name);
1930 sp += sprintf(sp, "tagger %s\n", tagger);
1931 *sp++ = '\n';
1932 memcpy(sp, msg, msglen);
1933 sp += msglen;
1934 free(tagger);
1935 free(msg);
1936
1937 if (store_object(OBJ_TAG, new_data.buffer,
1938 sp - (char*)new_data.buffer,
1939 NULL, t->sha1, 0))
1940 t->pack_id = MAX_PACK_ID;
1941 else
1942 t->pack_id = pack_id;
1943 }
1944
1945 static void cmd_reset_branch(void)
1946 {
1947 struct branch *b;
1948 char *sp;
1949
1950 /* Obtain the branch name from the rest of our command */
1951 sp = strchr(command_buf.buf, ' ') + 1;
1952 b = lookup_branch(sp);
1953 if (b) {
1954 hashclr(b->sha1);
1955 hashclr(b->branch_tree.versions[0].sha1);
1956 hashclr(b->branch_tree.versions[1].sha1);
1957 if (b->branch_tree.tree) {
1958 release_tree_content_recursive(b->branch_tree.tree);
1959 b->branch_tree.tree = NULL;
1960 }
1961 }
1962 else
1963 b = new_branch(sp);
1964 read_next_command();
1965 cmd_from(b);
1966 }
1967
1968 static void cmd_checkpoint(void)
1969 {
1970 if (object_count) {
1971 cycle_packfile();
1972 dump_branches();
1973 dump_tags();
1974 dump_marks();
1975 }
1976 read_next_command();
1977 }
1978
1979 static const char fast_import_usage[] =
1980 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
1981
1982 int main(int argc, const char **argv)
1983 {
1984 int i, show_stats = 1;
1985
1986 git_config(git_default_config);
1987
1988 for (i = 1; i < argc; i++) {
1989 const char *a = argv[i];
1990
1991 if (*a != '-' || !strcmp(a, "--"))
1992 break;
1993 else if (!prefixcmp(a, "--date-format=")) {
1994 const char *fmt = a + 14;
1995 if (!strcmp(fmt, "raw"))
1996 whenspec = WHENSPEC_RAW;
1997 else if (!strcmp(fmt, "rfc2822"))
1998 whenspec = WHENSPEC_RFC2822;
1999 else if (!strcmp(fmt, "now"))
2000 whenspec = WHENSPEC_NOW;
2001 else
2002 die("unknown --date-format argument %s", fmt);
2003 }
2004 else if (!prefixcmp(a, "--max-pack-size="))
2005 max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2006 else if (!prefixcmp(a, "--depth="))
2007 max_depth = strtoul(a + 8, NULL, 0);
2008 else if (!prefixcmp(a, "--active-branches="))
2009 max_active_branches = strtoul(a + 18, NULL, 0);
2010 else if (!prefixcmp(a, "--export-marks="))
2011 mark_file = a + 15;
2012 else if (!prefixcmp(a, "--export-pack-edges=")) {
2013 if (pack_edges)
2014 fclose(pack_edges);
2015 pack_edges = fopen(a + 20, "a");
2016 if (!pack_edges)
2017 die("Cannot open %s: %s", a + 20, strerror(errno));
2018 } else if (!strcmp(a, "--force"))
2019 force_update = 1;
2020 else if (!strcmp(a, "--quiet"))
2021 show_stats = 0;
2022 else if (!strcmp(a, "--stats"))
2023 show_stats = 1;
2024 else
2025 die("unknown option %s", a);
2026 }
2027 if (i != argc)
2028 usage(fast_import_usage);
2029
2030 alloc_objects(object_entry_alloc);
2031 strbuf_init(&command_buf);
2032
2033 atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2034 branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2035 avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2036 marks = pool_calloc(1, sizeof(struct mark_set));
2037
2038 start_packfile();
2039 for (;;) {
2040 read_next_command();
2041 if (command_buf.eof)
2042 break;
2043 else if (!strcmp("blob", command_buf.buf))
2044 cmd_new_blob();
2045 else if (!prefixcmp(command_buf.buf, "commit "))
2046 cmd_new_commit();
2047 else if (!prefixcmp(command_buf.buf, "tag "))
2048 cmd_new_tag();
2049 else if (!prefixcmp(command_buf.buf, "reset "))
2050 cmd_reset_branch();
2051 else if (!strcmp("checkpoint", command_buf.buf))
2052 cmd_checkpoint();
2053 else
2054 die("Unsupported command: %s", command_buf.buf);
2055 }
2056 end_packfile();
2057
2058 dump_branches();
2059 dump_tags();
2060 unkeep_all_packs();
2061 dump_marks();
2062
2063 if (pack_edges)
2064 fclose(pack_edges);
2065
2066 if (show_stats) {
2067 uintmax_t total_count = 0, duplicate_count = 0;
2068 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2069 total_count += object_count_by_type[i];
2070 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2071 duplicate_count += duplicate_count_by_type[i];
2072
2073 fprintf(stderr, "%s statistics:\n", argv[0]);
2074 fprintf(stderr, "---------------------------------------------------------------------\n");
2075 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2076 fprintf(stderr, "Total objects: %10" PRIuMAX " (%10" PRIuMAX " duplicates )\n", total_count, duplicate_count);
2077 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]);
2078 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]);
2079 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]);
2080 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]);
2081 fprintf(stderr, "Total branches: %10lu (%10lu loads )\n", branch_count, branch_load_count);
2082 fprintf(stderr, " marks: %10" PRIuMAX " (%10" PRIuMAX " unique )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2083 fprintf(stderr, " atoms: %10u\n", atom_cnt);
2084 fprintf(stderr, "Memory total: %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2085 fprintf(stderr, " pools: %10lu KiB\n", (unsigned long)(total_allocd/1024));
2086 fprintf(stderr, " objects: %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2087 fprintf(stderr, "---------------------------------------------------------------------\n");
2088 pack_report();
2089 fprintf(stderr, "---------------------------------------------------------------------\n");
2090 fprintf(stderr, "\n");
2091 }
2092
2093 return failure ? 1 : 0;
2094 }