?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Trim leading / off of paths in git-svn prop_walk
[git/gitweb.git] / fast-import.c
blob74597c901c4e7bff7ed5344719c3162b4e096c3a
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 | progress
12 ;
13
14 new_blob ::= 'blob' lf
15 mark?
16 file_content;
17 file_content ::= data;
18
19 new_commit ::= 'commit' sp ref_str lf
20 mark?
21 ('author' sp name '<' email '>' when lf)?
22 'committer' sp name '<' email '>' when lf
23 commit_msg
24 ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
25 ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
26 file_change*
27 lf?;
28 commit_msg ::= data;
29
30 file_change ::= file_clr
31 | file_del
32 | file_rnm
33 | file_cpy
34 | file_obm
35 | file_inm;
36 file_clr ::= 'deleteall' lf;
37 file_del ::= 'D' sp path_str lf;
38 file_rnm ::= 'R' sp path_str sp path_str lf;
39 file_cpy ::= 'C' sp path_str sp path_str lf;
40 file_obm ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf;
41 file_inm ::= 'M' sp mode sp 'inline' sp path_str lf
42 data;
43
44 new_tag ::= 'tag' sp tag_str lf
45 'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
46 'tagger' sp name '<' email '>' when lf
47 tag_msg;
48 tag_msg ::= data;
49
50 reset_branch ::= 'reset' sp ref_str lf
51 ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
52 lf?;
53
54 checkpoint ::= 'checkpoint' lf
55 lf?;
56
57 progress ::= 'progress' sp not_lf* lf
58 lf?;
59
60 # note: the first idnum in a stream should be 1 and subsequent
61 # idnums should not have gaps between values as this will cause
62 # the stream parser to reserve space for the gapped values. An
63 # idnum can be updated in the future to a new object by issuing
64 # a new mark directive with the old idnum.
65 #
66 mark ::= 'mark' sp idnum lf;
67 data ::= (delimited_data | exact_data)
68 lf?;
69
70 # note: delim may be any string but must not contain lf.
71 # data_line may contain any data but must not be exactly
72 # delim.
73 delimited_data ::= 'data' sp '<<' delim lf
74 (data_line lf)*
75 delim lf;
76
77 # note: declen indicates the length of binary_data in bytes.
78 # declen does not include the lf preceeding the binary data.
79 #
80 exact_data ::= 'data' sp declen lf
81 binary_data;
82
83 # note: quoted strings are C-style quoting supporting \c for
84 # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
85 # is the signed byte value in octal. Note that the only
86 # characters which must actually be escaped to protect the
87 # stream formatting is: \, " and LF. Otherwise these values
88 # are UTF8.
89 #
90 ref_str ::= ref;
91 sha1exp_str ::= sha1exp;
92 tag_str ::= tag;
93 path_str ::= path | '"' quoted(path) '"' ;
94 mode ::= '100644' | '644'
95 | '100755' | '755'
96 | '120000'
97 ;
98
99 declen ::= # unsigned 32 bit value, ascii base10 notation;
100 bigint ::= # unsigned integer value, ascii base10 notation;
101 binary_data ::= # file content, not interpreted;
102
103 when ::= raw_when | rfc2822_when;
104 raw_when ::= ts sp tz;
105 rfc2822_when ::= # Valid RFC 2822 date and time;
106
107 sp ::= # ASCII space character;
108 lf ::= # ASCII newline (LF) character;
109
110 # note: a colon (':') must precede the numerical value assigned to
111 # an idnum. This is to distinguish it from a ref or tag name as
112 # GIT does not permit ':' in ref or tag strings.
113 #
114 idnum ::= ':' bigint;
115 path ::= # GIT style file path, e.g. "a/b/c";
116 ref ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
117 tag ::= # GIT tag name, e.g. "FIREFOX_1_5";
118 sha1exp ::= # Any valid GIT SHA1 expression;
119 hexsha1 ::= # SHA1 in hexadecimal format;
120
121 # note: name and email are UTF8 strings, however name must not
122 # contain '<' or lf and email must not contain any of the
123 # following: '<', '>', lf.
124 #
125 name ::= # valid GIT author/committer name;
126 email ::= # valid GIT author/committer email;
127 ts ::= # time since the epoch in seconds, ascii base10 notation;
128 tz ::= # GIT style timezone;
129
130 # note: comments may appear anywhere in the input, except
131 # within a data command. Any form of the data command
132 # always escapes the related input from comment processing.
133 #
134 # In case it is not clear, the '#' that starts the comment
135 # must be the first character on that the line (an lf have
136 # preceeded it).
137 #
138 comment ::= '#' not_lf* lf;
139 not_lf ::= # Any byte that is not ASCII newline (LF);
140 */
141
142 #include "builtin.h"
143 #include "cache.h"
144 #include "object.h"
145 #include "blob.h"
146 #include "tree.h"
147 #include "commit.h"
148 #include "delta.h"
149 #include "pack.h"
150 #include "refs.h"
151 #include "csum-file.h"
152 #include "quote.h"
153
154 #define PACK_ID_BITS 16
155 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
156 #define DEPTH_BITS 13
157 #define MAX_DEPTH ((1<<DEPTH_BITS)-1)
158
159 struct object_entry
160 {
161 struct object_entry *next;
162 uint32_t offset;
163 uint32_t type : TYPE_BITS,
164 pack_id : PACK_ID_BITS,
165 depth : DEPTH_BITS;
166 unsigned char sha1[20];
167 };
168
169 struct object_entry_pool
170 {
171 struct object_entry_pool *next_pool;
172 struct object_entry *next_free;
173 struct object_entry *end;
174 struct object_entry entries[FLEX_ARRAY]; /* more */
175 };
176
177 struct mark_set
178 {
179 union {
180 struct object_entry *marked[1024];
181 struct mark_set *sets[1024];
182 } data;
183 unsigned int shift;
184 };
185
186 struct last_object
187 {
188 struct strbuf data;
189 uint32_t offset;
190 unsigned int depth;
191 unsigned no_swap : 1;
192 };
193
194 struct mem_pool
195 {
196 struct mem_pool *next_pool;
197 char *next_free;
198 char *end;
199 uintmax_t space[FLEX_ARRAY]; /* more */
200 };
201
202 struct atom_str
203 {
204 struct atom_str *next_atom;
205 unsigned short str_len;
206 char str_dat[FLEX_ARRAY]; /* more */
207 };
208
209 struct tree_content;
210 struct tree_entry
211 {
212 struct tree_content *tree;
213 struct atom_str* name;
214 struct tree_entry_ms
215 {
216 uint16_t mode;
217 unsigned char sha1[20];
218 } versions[2];
219 };
220
221 struct tree_content
222 {
223 unsigned int entry_capacity; /* must match avail_tree_content */
224 unsigned int entry_count;
225 unsigned int delta_depth;
226 struct tree_entry *entries[FLEX_ARRAY]; /* more */
227 };
228
229 struct avail_tree_content
230 {
231 unsigned int entry_capacity; /* must match tree_content */
232 struct avail_tree_content *next_avail;
233 };
234
235 struct branch
236 {
237 struct branch *table_next_branch;
238 struct branch *active_next_branch;
239 const char *name;
240 struct tree_entry branch_tree;
241 uintmax_t last_commit;
242 unsigned active : 1;
243 unsigned pack_id : PACK_ID_BITS;
244 unsigned char sha1[20];
245 };
246
247 struct tag
248 {
249 struct tag *next_tag;
250 const char *name;
251 unsigned int pack_id;
252 unsigned char sha1[20];
253 };
254
255 struct hash_list
256 {
257 struct hash_list *next;
258 unsigned char sha1[20];
259 };
260
261 typedef enum {
262 WHENSPEC_RAW = 1,
263 WHENSPEC_RFC2822,
264 WHENSPEC_NOW,
265 } whenspec_type;
266
267 struct recent_command
268 {
269 struct recent_command *prev;
270 struct recent_command *next;
271 char *buf;
272 };
273
274 /* Configured limits on output */
275 static unsigned long max_depth = 10;
276 static off_t max_packsize = (1LL << 32) - 1;
277 static int force_update;
278
279 /* Stats and misc. counters */
280 static uintmax_t alloc_count;
281 static uintmax_t marks_set_count;
282 static uintmax_t object_count_by_type[1 << TYPE_BITS];
283 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
284 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
285 static unsigned long object_count;
286 static unsigned long branch_count;
287 static unsigned long branch_load_count;
288 static int failure;
289 static FILE *pack_edges;
290
291 /* Memory pools */
292 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
293 static size_t total_allocd;
294 static struct mem_pool *mem_pool;
295
296 /* Atom management */
297 static unsigned int atom_table_sz = 4451;
298 static unsigned int atom_cnt;
299 static struct atom_str **atom_table;
300
301 /* The .pack file being generated */
302 static unsigned int pack_id;
303 static struct packed_git *pack_data;
304 static struct packed_git **all_packs;
305 static unsigned long pack_size;
306
307 /* Table of objects we've written. */
308 static unsigned int object_entry_alloc = 5000;
309 static struct object_entry_pool *blocks;
310 static struct object_entry *object_table[1 << 16];
311 static struct mark_set *marks;
312 static const char* mark_file;
313
314 /* Our last blob */
315 static struct last_object last_blob = { STRBUF_INIT, 0, 0, 0 };
316
317 /* Tree management */
318 static unsigned int tree_entry_alloc = 1000;
319 static void *avail_tree_entry;
320 static unsigned int avail_tree_table_sz = 100;
321 static struct avail_tree_content **avail_tree_table;
322 static struct strbuf old_tree = STRBUF_INIT;
323 static struct strbuf new_tree = STRBUF_INIT;
324
325 /* Branch data */
326 static unsigned long max_active_branches = 5;
327 static unsigned long cur_active_branches;
328 static unsigned long branch_table_sz = 1039;
329 static struct branch **branch_table;
330 static struct branch *active_branches;
331
332 /* Tag data */
333 static struct tag *first_tag;
334 static struct tag *last_tag;
335
336 /* Input stream parsing */
337 static whenspec_type whenspec = WHENSPEC_RAW;
338 static struct strbuf command_buf = STRBUF_INIT;
339 static int unread_command_buf;
340 static struct recent_command cmd_hist = {&cmd_hist, &cmd_hist, NULL};
341 static struct recent_command *cmd_tail = &cmd_hist;
342 static struct recent_command *rc_free;
343 static unsigned int cmd_save = 100;
344 static uintmax_t next_mark;
345 static struct strbuf new_data = STRBUF_INIT;
346
347 static void write_branch_report(FILE *rpt, struct branch *b)
348 {
349 fprintf(rpt, "%s:\n", b->name);
350
351 fprintf(rpt, " status :");
352 if (b->active)
353 fputs(" active", rpt);
354 if (b->branch_tree.tree)
355 fputs(" loaded", rpt);
356 if (is_null_sha1(b->branch_tree.versions[1].sha1))
357 fputs(" dirty", rpt);
358 fputc('\n', rpt);
359
360 fprintf(rpt, " tip commit : %s\n", sha1_to_hex(b->sha1));
361 fprintf(rpt, " old tree : %s\n", sha1_to_hex(b->branch_tree.versions[0].sha1));
362 fprintf(rpt, " cur tree : %s\n", sha1_to_hex(b->branch_tree.versions[1].sha1));
363 fprintf(rpt, " commit clock: %" PRIuMAX "\n", b->last_commit);
364
365 fputs(" last pack : ", rpt);
366 if (b->pack_id < MAX_PACK_ID)
367 fprintf(rpt, "%u", b->pack_id);
368 fputc('\n', rpt);
369
370 fputc('\n', rpt);
371 }
372
373 static void write_crash_report(const char *err)
374 {
375 char *loc = git_path("fast_import_crash_%d", getpid());
376 FILE *rpt = fopen(loc, "w");
377 struct branch *b;
378 unsigned long lu;
379 struct recent_command *rc;
380
381 if (!rpt) {
382 error("can't write crash report %s: %s", loc, strerror(errno));
383 return;
384 }
385
386 fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
387
388 fprintf(rpt, "fast-import crash report:\n");
389 fprintf(rpt, " fast-import process: %d\n", getpid());
390 fprintf(rpt, " parent process : %d\n", getppid());
391 fprintf(rpt, " at %s\n", show_date(time(NULL), 0, DATE_LOCAL));
392 fputc('\n', rpt);
393
394 fputs("fatal: ", rpt);
395 fputs(err, rpt);
396 fputc('\n', rpt);
397
398 fputc('\n', rpt);
399 fputs("Most Recent Commands Before Crash\n", rpt);
400 fputs("---------------------------------\n", rpt);
401 for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
402 if (rc->next == &cmd_hist)
403 fputs("* ", rpt);
404 else
405 fputs(" ", rpt);
406 fputs(rc->buf, rpt);
407 fputc('\n', rpt);
408 }
409
410 fputc('\n', rpt);
411 fputs("Active Branch LRU\n", rpt);
412 fputs("-----------------\n", rpt);
413 fprintf(rpt, " active_branches = %lu cur, %lu max\n",
414 cur_active_branches,
415 max_active_branches);
416 fputc('\n', rpt);
417 fputs(" pos clock name\n", rpt);
418 fputs(" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
419 for (b = active_branches, lu = 0; b; b = b->active_next_branch)
420 fprintf(rpt, " %2lu) %6" PRIuMAX" %s\n",
421 ++lu, b->last_commit, b->name);
422
423 fputc('\n', rpt);
424 fputs("Inactive Branches\n", rpt);
425 fputs("-----------------\n", rpt);
426 for (lu = 0; lu < branch_table_sz; lu++) {
427 for (b = branch_table[lu]; b; b = b->table_next_branch)
428 write_branch_report(rpt, b);
429 }
430
431 fputc('\n', rpt);
432 fputs("-------------------\n", rpt);
433 fputs("END OF CRASH REPORT\n", rpt);
434 fclose(rpt);
435 }
436
437 static NORETURN void die_nicely(const char *err, va_list params)
438 {
439 static int zombie;
440 char message[2 * PATH_MAX];
441
442 vsnprintf(message, sizeof(message), err, params);
443 fputs("fatal: ", stderr);
444 fputs(message, stderr);
445 fputc('\n', stderr);
446
447 if (!zombie) {
448 zombie = 1;
449 write_crash_report(message);
450 }
451 exit(128);
452 }
453
454 static void alloc_objects(unsigned int cnt)
455 {
456 struct object_entry_pool *b;
457
458 b = xmalloc(sizeof(struct object_entry_pool)
459 + cnt * sizeof(struct object_entry));
460 b->next_pool = blocks;
461 b->next_free = b->entries;
462 b->end = b->entries + cnt;
463 blocks = b;
464 alloc_count += cnt;
465 }
466
467 static struct object_entry *new_object(unsigned char *sha1)
468 {
469 struct object_entry *e;
470
471 if (blocks->next_free == blocks->end)
472 alloc_objects(object_entry_alloc);
473
474 e = blocks->next_free++;
475 hashcpy(e->sha1, sha1);
476 return e;
477 }
478
479 static struct object_entry *find_object(unsigned char *sha1)
480 {
481 unsigned int h = sha1[0] << 8 | sha1[1];
482 struct object_entry *e;
483 for (e = object_table[h]; e; e = e->next)
484 if (!hashcmp(sha1, e->sha1))
485 return e;
486 return NULL;
487 }
488
489 static struct object_entry *insert_object(unsigned char *sha1)
490 {
491 unsigned int h = sha1[0] << 8 | sha1[1];
492 struct object_entry *e = object_table[h];
493 struct object_entry *p = NULL;
494
495 while (e) {
496 if (!hashcmp(sha1, e->sha1))
497 return e;
498 p = e;
499 e = e->next;
500 }
501
502 e = new_object(sha1);
503 e->next = NULL;
504 e->offset = 0;
505 if (p)
506 p->next = e;
507 else
508 object_table[h] = e;
509 return e;
510 }
511
512 static unsigned int hc_str(const char *s, size_t len)
513 {
514 unsigned int r = 0;
515 while (len-- > 0)
516 r = r * 31 + *s++;
517 return r;
518 }
519
520 static void *pool_alloc(size_t len)
521 {
522 struct mem_pool *p;
523 void *r;
524
525 for (p = mem_pool; p; p = p->next_pool)
526 if ((p->end - p->next_free >= len))
527 break;
528
529 if (!p) {
530 if (len >= (mem_pool_alloc/2)) {
531 total_allocd += len;
532 return xmalloc(len);
533 }
534 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
535 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
536 p->next_pool = mem_pool;
537 p->next_free = (char *) p->space;
538 p->end = p->next_free + mem_pool_alloc;
539 mem_pool = p;
540 }
541
542 r = p->next_free;
543 /* round out to a 'uintmax_t' alignment */
544 if (len & (sizeof(uintmax_t) - 1))
545 len += sizeof(uintmax_t) - (len & (sizeof(uintmax_t) - 1));
546 p->next_free += len;
547 return r;
548 }
549
550 static void *pool_calloc(size_t count, size_t size)
551 {
552 size_t len = count * size;
553 void *r = pool_alloc(len);
554 memset(r, 0, len);
555 return r;
556 }
557
558 static char *pool_strdup(const char *s)
559 {
560 char *r = pool_alloc(strlen(s) + 1);
561 strcpy(r, s);
562 return r;
563 }
564
565 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
566 {
567 struct mark_set *s = marks;
568 while ((idnum >> s->shift) >= 1024) {
569 s = pool_calloc(1, sizeof(struct mark_set));
570 s->shift = marks->shift + 10;
571 s->data.sets[0] = marks;
572 marks = s;
573 }
574 while (s->shift) {
575 uintmax_t i = idnum >> s->shift;
576 idnum -= i << s->shift;
577 if (!s->data.sets[i]) {
578 s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
579 s->data.sets[i]->shift = s->shift - 10;
580 }
581 s = s->data.sets[i];
582 }
583 if (!s->data.marked[idnum])
584 marks_set_count++;
585 s->data.marked[idnum] = oe;
586 }
587
588 static struct object_entry *find_mark(uintmax_t idnum)
589 {
590 uintmax_t orig_idnum = idnum;
591 struct mark_set *s = marks;
592 struct object_entry *oe = NULL;
593 if ((idnum >> s->shift) < 1024) {
594 while (s && s->shift) {
595 uintmax_t i = idnum >> s->shift;
596 idnum -= i << s->shift;
597 s = s->data.sets[i];
598 }
599 if (s)
600 oe = s->data.marked[idnum];
601 }
602 if (!oe)
603 die("mark :%" PRIuMAX " not declared", orig_idnum);
604 return oe;
605 }
606
607 static struct atom_str *to_atom(const char *s, unsigned short len)
608 {
609 unsigned int hc = hc_str(s, len) % atom_table_sz;
610 struct atom_str *c;
611
612 for (c = atom_table[hc]; c; c = c->next_atom)
613 if (c->str_len == len && !strncmp(s, c->str_dat, len))
614 return c;
615
616 c = pool_alloc(sizeof(struct atom_str) + len + 1);
617 c->str_len = len;
618 strncpy(c->str_dat, s, len);
619 c->str_dat[len] = 0;
620 c->next_atom = atom_table[hc];
621 atom_table[hc] = c;
622 atom_cnt++;
623 return c;
624 }
625
626 static struct branch *lookup_branch(const char *name)
627 {
628 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
629 struct branch *b;
630
631 for (b = branch_table[hc]; b; b = b->table_next_branch)
632 if (!strcmp(name, b->name))
633 return b;
634 return NULL;
635 }
636
637 static struct branch *new_branch(const char *name)
638 {
639 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
640 struct branch* b = lookup_branch(name);
641
642 if (b)
643 die("Invalid attempt to create duplicate branch: %s", name);
644 switch (check_ref_format(name)) {
645 case 0: break; /* its valid */
646 case CHECK_REF_FORMAT_ONELEVEL:
647 break; /* valid, but too few '/', allow anyway */
648 default:
649 die("Branch name doesn't conform to GIT standards: %s", name);
650 }
651
652 b = pool_calloc(1, sizeof(struct branch));
653 b->name = pool_strdup(name);
654 b->table_next_branch = branch_table[hc];
655 b->branch_tree.versions[0].mode = S_IFDIR;
656 b->branch_tree.versions[1].mode = S_IFDIR;
657 b->active = 0;
658 b->pack_id = MAX_PACK_ID;
659 branch_table[hc] = b;
660 branch_count++;
661 return b;
662 }
663
664 static unsigned int hc_entries(unsigned int cnt)
665 {
666 cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
667 return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
668 }
669
670 static struct tree_content *new_tree_content(unsigned int cnt)
671 {
672 struct avail_tree_content *f, *l = NULL;
673 struct tree_content *t;
674 unsigned int hc = hc_entries(cnt);
675
676 for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
677 if (f->entry_capacity >= cnt)
678 break;
679
680 if (f) {
681 if (l)
682 l->next_avail = f->next_avail;
683 else
684 avail_tree_table[hc] = f->next_avail;
685 } else {
686 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
687 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
688 f->entry_capacity = cnt;
689 }
690
691 t = (struct tree_content*)f;
692 t->entry_count = 0;
693 t->delta_depth = 0;
694 return t;
695 }
696
697 static void release_tree_entry(struct tree_entry *e);
698 static void release_tree_content(struct tree_content *t)
699 {
700 struct avail_tree_content *f = (struct avail_tree_content*)t;
701 unsigned int hc = hc_entries(f->entry_capacity);
702 f->next_avail = avail_tree_table[hc];
703 avail_tree_table[hc] = f;
704 }
705
706 static void release_tree_content_recursive(struct tree_content *t)
707 {
708 unsigned int i;
709 for (i = 0; i < t->entry_count; i++)
710 release_tree_entry(t->entries[i]);
711 release_tree_content(t);
712 }
713
714 static struct tree_content *grow_tree_content(
715 struct tree_content *t,
716 int amt)
717 {
718 struct tree_content *r = new_tree_content(t->entry_count + amt);
719 r->entry_count = t->entry_count;
720 r->delta_depth = t->delta_depth;
721 memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
722 release_tree_content(t);
723 return r;
724 }
725
726 static struct tree_entry *new_tree_entry(void)
727 {
728 struct tree_entry *e;
729
730 if (!avail_tree_entry) {
731 unsigned int n = tree_entry_alloc;
732 total_allocd += n * sizeof(struct tree_entry);
733 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
734 while (n-- > 1) {
735 *((void**)e) = e + 1;
736 e++;
737 }
738 *((void**)e) = NULL;
739 }
740
741 e = avail_tree_entry;
742 avail_tree_entry = *((void**)e);
743 return e;
744 }
745
746 static void release_tree_entry(struct tree_entry *e)
747 {
748 if (e->tree)
749 release_tree_content_recursive(e->tree);
750 *((void**)e) = avail_tree_entry;
751 avail_tree_entry = e;
752 }
753
754 static struct tree_content *dup_tree_content(struct tree_content *s)
755 {
756 struct tree_content *d;
757 struct tree_entry *a, *b;
758 unsigned int i;
759
760 if (!s)
761 return NULL;
762 d = new_tree_content(s->entry_count);
763 for (i = 0; i < s->entry_count; i++) {
764 a = s->entries[i];
765 b = new_tree_entry();
766 memcpy(b, a, sizeof(*a));
767 if (a->tree && is_null_sha1(b->versions[1].sha1))
768 b->tree = dup_tree_content(a->tree);
769 else
770 b->tree = NULL;
771 d->entries[i] = b;
772 }
773 d->entry_count = s->entry_count;
774 d->delta_depth = s->delta_depth;
775
776 return d;
777 }
778
779 static void start_packfile(void)
780 {
781 static char tmpfile[PATH_MAX];
782 struct packed_git *p;
783 struct pack_header hdr;
784 int pack_fd;
785
786 snprintf(tmpfile, sizeof(tmpfile),
787 "%s/tmp_pack_XXXXXX", get_object_directory());
788 pack_fd = xmkstemp(tmpfile);
789 p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
790 strcpy(p->pack_name, tmpfile);
791 p->pack_fd = pack_fd;
792
793 hdr.hdr_signature = htonl(PACK_SIGNATURE);
794 hdr.hdr_version = htonl(2);
795 hdr.hdr_entries = 0;
796 write_or_die(p->pack_fd, &hdr, sizeof(hdr));
797
798 pack_data = p;
799 pack_size = sizeof(hdr);
800 object_count = 0;
801
802 all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
803 all_packs[pack_id] = p;
804 }
805
806 static int oecmp (const void *a_, const void *b_)
807 {
808 struct object_entry *a = *((struct object_entry**)a_);
809 struct object_entry *b = *((struct object_entry**)b_);
810 return hashcmp(a->sha1, b->sha1);
811 }
812
813 static char *create_index(void)
814 {
815 static char tmpfile[PATH_MAX];
816 SHA_CTX ctx;
817 struct sha1file *f;
818 struct object_entry **idx, **c, **last, *e;
819 struct object_entry_pool *o;
820 uint32_t array[256];
821 int i, idx_fd;
822
823 /* Build the sorted table of object IDs. */
824 idx = xmalloc(object_count * sizeof(struct object_entry*));
825 c = idx;
826 for (o = blocks; o; o = o->next_pool)
827 for (e = o->next_free; e-- != o->entries;)
828 if (pack_id == e->pack_id)
829 *c++ = e;
830 last = idx + object_count;
831 if (c != last)
832 die("internal consistency error creating the index");
833 qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
834
835 /* Generate the fan-out array. */
836 c = idx;
837 for (i = 0; i < 256; i++) {
838 struct object_entry **next = c;;
839 while (next < last) {
840 if ((*next)->sha1[0] != i)
841 break;
842 next++;
843 }
844 array[i] = htonl(next - idx);
845 c = next;
846 }
847
848 snprintf(tmpfile, sizeof(tmpfile),
849 "%s/tmp_idx_XXXXXX", get_object_directory());
850 idx_fd = xmkstemp(tmpfile);
851 f = sha1fd(idx_fd, tmpfile);
852 sha1write(f, array, 256 * sizeof(int));
853 SHA1_Init(&ctx);
854 for (c = idx; c != last; c++) {
855 uint32_t offset = htonl((*c)->offset);
856 sha1write(f, &offset, 4);
857 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
858 SHA1_Update(&ctx, (*c)->sha1, 20);
859 }
860 sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
861 sha1close(f, NULL, 1);
862 free(idx);
863 SHA1_Final(pack_data->sha1, &ctx);
864 return tmpfile;
865 }
866
867 static char *keep_pack(char *curr_index_name)
868 {
869 static char name[PATH_MAX];
870 static const char *keep_msg = "fast-import";
871 int keep_fd;
872
873 chmod(pack_data->pack_name, 0444);
874 chmod(curr_index_name, 0444);
875
876 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
877 get_object_directory(), sha1_to_hex(pack_data->sha1));
878 keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
879 if (keep_fd < 0)
880 die("cannot create keep file");
881 write(keep_fd, keep_msg, strlen(keep_msg));
882 close(keep_fd);
883
884 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
885 get_object_directory(), sha1_to_hex(pack_data->sha1));
886 if (move_temp_to_file(pack_data->pack_name, name))
887 die("cannot store pack file");
888
889 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
890 get_object_directory(), sha1_to_hex(pack_data->sha1));
891 if (move_temp_to_file(curr_index_name, name))
892 die("cannot store index file");
893 return name;
894 }
895
896 static void unkeep_all_packs(void)
897 {
898 static char name[PATH_MAX];
899 int k;
900
901 for (k = 0; k < pack_id; k++) {
902 struct packed_git *p = all_packs[k];
903 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
904 get_object_directory(), sha1_to_hex(p->sha1));
905 unlink(name);
906 }
907 }
908
909 static void end_packfile(void)
910 {
911 struct packed_git *old_p = pack_data, *new_p;
912
913 if (object_count) {
914 char *idx_name;
915 int i;
916 struct branch *b;
917 struct tag *t;
918
919 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
920 pack_data->pack_name, object_count);
921 close(pack_data->pack_fd);
922 idx_name = keep_pack(create_index());
923
924 /* Register the packfile with core git's machinary. */
925 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
926 if (!new_p)
927 die("core git rejected index %s", idx_name);
928 new_p->windows = old_p->windows;
929 all_packs[pack_id] = new_p;
930 install_packed_git(new_p);
931
932 /* Print the boundary */
933 if (pack_edges) {
934 fprintf(pack_edges, "%s:", new_p->pack_name);
935 for (i = 0; i < branch_table_sz; i++) {
936 for (b = branch_table[i]; b; b = b->table_next_branch) {
937 if (b->pack_id == pack_id)
938 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
939 }
940 }
941 for (t = first_tag; t; t = t->next_tag) {
942 if (t->pack_id == pack_id)
943 fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
944 }
945 fputc('\n', pack_edges);
946 fflush(pack_edges);
947 }
948
949 pack_id++;
950 }
951 else
952 unlink(old_p->pack_name);
953 free(old_p);
954
955 /* We can't carry a delta across packfiles. */
956 strbuf_release(&last_blob.data);
957 last_blob.offset = 0;
958 last_blob.depth = 0;
959 }
960
961 static void cycle_packfile(void)
962 {
963 end_packfile();
964 start_packfile();
965 }
966
967 static size_t encode_header(
968 enum object_type type,
969 size_t size,
970 unsigned char *hdr)
971 {
972 int n = 1;
973 unsigned char c;
974
975 if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
976 die("bad type %d", type);
977
978 c = (type << 4) | (size & 15);
979 size >>= 4;
980 while (size) {
981 *hdr++ = c | 0x80;
982 c = size & 0x7f;
983 size >>= 7;
984 n++;
985 }
986 *hdr = c;
987 return n;
988 }
989
990 static int store_object(
991 enum object_type type,
992 struct strbuf *dat,
993 struct last_object *last,
994 unsigned char *sha1out,
995 uintmax_t mark)
996 {
997 void *out, *delta;
998 struct object_entry *e;
999 unsigned char hdr[96];
1000 unsigned char sha1[20];
1001 unsigned long hdrlen, deltalen;
1002 SHA_CTX c;
1003 z_stream s;
1004
1005 hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
1006 (unsigned long)dat->len) + 1;
1007 SHA1_Init(&c);
1008 SHA1_Update(&c, hdr, hdrlen);
1009 SHA1_Update(&c, dat->buf, dat->len);
1010 SHA1_Final(sha1, &c);
1011 if (sha1out)
1012 hashcpy(sha1out, sha1);
1013
1014 e = insert_object(sha1);
1015 if (mark)
1016 insert_mark(mark, e);
1017 if (e->offset) {
1018 duplicate_count_by_type[type]++;
1019 return 1;
1020 } else if (find_sha1_pack(sha1, packed_git)) {
1021 e->type = type;
1022 e->pack_id = MAX_PACK_ID;
1023 e->offset = 1; /* just not zero! */
1024 duplicate_count_by_type[type]++;
1025 return 1;
1026 }
1027
1028 if (last && last->data.buf && last->depth < max_depth) {
1029 delta = diff_delta(last->data.buf, last->data.len,
1030 dat->buf, dat->len,
1031 &deltalen, 0);
1032 if (delta && deltalen >= dat->len) {
1033 free(delta);
1034 delta = NULL;
1035 }
1036 } else
1037 delta = NULL;
1038
1039 memset(&s, 0, sizeof(s));
1040 deflateInit(&s, zlib_compression_level);
1041 if (delta) {
1042 s.next_in = delta;
1043 s.avail_in = deltalen;
1044 } else {
1045 s.next_in = (void *)dat->buf;
1046 s.avail_in = dat->len;
1047 }
1048 s.avail_out = deflateBound(&s, s.avail_in);
1049 s.next_out = out = xmalloc(s.avail_out);
1050 while (deflate(&s, Z_FINISH) == Z_OK)
1051 /* nothing */;
1052 deflateEnd(&s);
1053
1054 /* Determine if we should auto-checkpoint. */
1055 if ((pack_size + 60 + s.total_out) > max_packsize
1056 || (pack_size + 60 + s.total_out) < pack_size) {
1057
1058 /* This new object needs to *not* have the current pack_id. */
1059 e->pack_id = pack_id + 1;
1060 cycle_packfile();
1061
1062 /* We cannot carry a delta into the new pack. */
1063 if (delta) {
1064 free(delta);
1065 delta = NULL;
1066
1067 memset(&s, 0, sizeof(s));
1068 deflateInit(&s, zlib_compression_level);
1069 s.next_in = (void *)dat->buf;
1070 s.avail_in = dat->len;
1071 s.avail_out = deflateBound(&s, s.avail_in);
1072 s.next_out = out = xrealloc(out, s.avail_out);
1073 while (deflate(&s, Z_FINISH) == Z_OK)
1074 /* nothing */;
1075 deflateEnd(&s);
1076 }
1077 }
1078
1079 e->type = type;
1080 e->pack_id = pack_id;
1081 e->offset = pack_size;
1082 object_count++;
1083 object_count_by_type[type]++;
1084
1085 if (delta) {
1086 unsigned long ofs = e->offset - last->offset;
1087 unsigned pos = sizeof(hdr) - 1;
1088
1089 delta_count_by_type[type]++;
1090 e->depth = last->depth + 1;
1091
1092 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
1093 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1094 pack_size += hdrlen;
1095
1096 hdr[pos] = ofs & 127;
1097 while (ofs >>= 7)
1098 hdr[--pos] = 128 | (--ofs & 127);
1099 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
1100 pack_size += sizeof(hdr) - pos;
1101 } else {
1102 e->depth = 0;
1103 hdrlen = encode_header(type, dat->len, hdr);
1104 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1105 pack_size += hdrlen;
1106 }
1107
1108 write_or_die(pack_data->pack_fd, out, s.total_out);
1109 pack_size += s.total_out;
1110
1111 free(out);
1112 free(delta);
1113 if (last) {
1114 if (last->no_swap) {
1115 last->data = *dat;
1116 } else {
1117 strbuf_swap(&last->data, dat);
1118 }
1119 last->offset = e->offset;
1120 last->depth = e->depth;
1121 }
1122 return 0;
1123 }
1124
1125 static void *gfi_unpack_entry(
1126 struct object_entry *oe,
1127 unsigned long *sizep)
1128 {
1129 enum object_type type;
1130 struct packed_git *p = all_packs[oe->pack_id];
1131 if (p == pack_data)
1132 p->pack_size = pack_size + 20;
1133 return unpack_entry(p, oe->offset, &type, sizep);
1134 }
1135
1136 static const char *get_mode(const char *str, uint16_t *modep)
1137 {
1138 unsigned char c;
1139 uint16_t mode = 0;
1140
1141 while ((c = *str++) != ' ') {
1142 if (c < '0' || c > '7')
1143 return NULL;
1144 mode = (mode << 3) + (c - '0');
1145 }
1146 *modep = mode;
1147 return str;
1148 }
1149
1150 static void load_tree(struct tree_entry *root)
1151 {
1152 unsigned char* sha1 = root->versions[1].sha1;
1153 struct object_entry *myoe;
1154 struct tree_content *t;
1155 unsigned long size;
1156 char *buf;
1157 const char *c;
1158
1159 root->tree = t = new_tree_content(8);
1160 if (is_null_sha1(sha1))
1161 return;
1162
1163 myoe = find_object(sha1);
1164 if (myoe && myoe->pack_id != MAX_PACK_ID) {
1165 if (myoe->type != OBJ_TREE)
1166 die("Not a tree: %s", sha1_to_hex(sha1));
1167 t->delta_depth = myoe->depth;
1168 buf = gfi_unpack_entry(myoe, &size);
1169 } else {
1170 enum object_type type;
1171 buf = read_sha1_file(sha1, &type, &size);
1172 if (!buf || type != OBJ_TREE)
1173 die("Can't load tree %s", sha1_to_hex(sha1));
1174 }
1175
1176 c = buf;
1177 while (c != (buf + size)) {
1178 struct tree_entry *e = new_tree_entry();
1179
1180 if (t->entry_count == t->entry_capacity)
1181 root->tree = t = grow_tree_content(t, t->entry_count);
1182 t->entries[t->entry_count++] = e;
1183
1184 e->tree = NULL;
1185 c = get_mode(c, &e->versions[1].mode);
1186 if (!c)
1187 die("Corrupt mode in %s", sha1_to_hex(sha1));
1188 e->versions[0].mode = e->versions[1].mode;
1189 e->name = to_atom(c, strlen(c));
1190 c += e->name->str_len + 1;
1191 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1192 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1193 c += 20;
1194 }
1195 free(buf);
1196 }
1197
1198 static int tecmp0 (const void *_a, const void *_b)
1199 {
1200 struct tree_entry *a = *((struct tree_entry**)_a);
1201 struct tree_entry *b = *((struct tree_entry**)_b);
1202 return base_name_compare(
1203 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1204 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1205 }
1206
1207 static int tecmp1 (const void *_a, const void *_b)
1208 {
1209 struct tree_entry *a = *((struct tree_entry**)_a);
1210 struct tree_entry *b = *((struct tree_entry**)_b);
1211 return base_name_compare(
1212 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1213 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1214 }
1215
1216 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1217 {
1218 size_t maxlen = 0;
1219 unsigned int i;
1220
1221 if (!v)
1222 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1223 else
1224 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1225
1226 for (i = 0; i < t->entry_count; i++) {
1227 if (t->entries[i]->versions[v].mode)
1228 maxlen += t->entries[i]->name->str_len + 34;
1229 }
1230
1231 strbuf_reset(b);
1232 strbuf_grow(b, maxlen);
1233 for (i = 0; i < t->entry_count; i++) {
1234 struct tree_entry *e = t->entries[i];
1235 if (!e->versions[v].mode)
1236 continue;
1237 strbuf_addf(b, "%o %s%c", (unsigned int)e->versions[v].mode,
1238 e->name->str_dat, '\0');
1239 strbuf_add(b, e->versions[v].sha1, 20);
1240 }
1241 }
1242
1243 static void store_tree(struct tree_entry *root)
1244 {
1245 struct tree_content *t = root->tree;
1246 unsigned int i, j, del;
1247 struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1248 struct object_entry *le;
1249
1250 if (!is_null_sha1(root->versions[1].sha1))
1251 return;
1252
1253 for (i = 0; i < t->entry_count; i++) {
1254 if (t->entries[i]->tree)
1255 store_tree(t->entries[i]);
1256 }
1257
1258 le = find_object(root->versions[0].sha1);
1259 if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1260 mktree(t, 0, &old_tree);
1261 lo.data = old_tree;
1262 lo.offset = le->offset;
1263 lo.depth = t->delta_depth;
1264 }
1265
1266 mktree(t, 1, &new_tree);
1267 store_object(OBJ_TREE, &new_tree, &lo, root->versions[1].sha1, 0);
1268
1269 t->delta_depth = lo.depth;
1270 for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1271 struct tree_entry *e = t->entries[i];
1272 if (e->versions[1].mode) {
1273 e->versions[0].mode = e->versions[1].mode;
1274 hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1275 t->entries[j++] = e;
1276 } else {
1277 release_tree_entry(e);
1278 del++;
1279 }
1280 }
1281 t->entry_count -= del;
1282 }
1283
1284 static int tree_content_set(
1285 struct tree_entry *root,
1286 const char *p,
1287 const unsigned char *sha1,
1288 const uint16_t mode,
1289 struct tree_content *subtree)
1290 {
1291 struct tree_content *t = root->tree;
1292 const char *slash1;
1293 unsigned int i, n;
1294 struct tree_entry *e;
1295
1296 slash1 = strchr(p, '/');
1297 if (slash1)
1298 n = slash1 - p;
1299 else
1300 n = strlen(p);
1301 if (!n)
1302 die("Empty path component found in input");
1303 if (!slash1 && !S_ISDIR(mode) && subtree)
1304 die("Non-directories cannot have subtrees");
1305
1306 for (i = 0; i < t->entry_count; i++) {
1307 e = t->entries[i];
1308 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1309 if (!slash1) {
1310 if (!S_ISDIR(mode)
1311 && e->versions[1].mode == mode
1312 && !hashcmp(e->versions[1].sha1, sha1))
1313 return 0;
1314 e->versions[1].mode = mode;
1315 hashcpy(e->versions[1].sha1, sha1);
1316 if (e->tree)
1317 release_tree_content_recursive(e->tree);
1318 e->tree = subtree;
1319 hashclr(root->versions[1].sha1);
1320 return 1;
1321 }
1322 if (!S_ISDIR(e->versions[1].mode)) {
1323 e->tree = new_tree_content(8);
1324 e->versions[1].mode = S_IFDIR;
1325 }
1326 if (!e->tree)
1327 load_tree(e);
1328 if (tree_content_set(e, slash1 + 1, sha1, mode, subtree)) {
1329 hashclr(root->versions[1].sha1);
1330 return 1;
1331 }
1332 return 0;
1333 }
1334 }
1335
1336 if (t->entry_count == t->entry_capacity)
1337 root->tree = t = grow_tree_content(t, t->entry_count);
1338 e = new_tree_entry();
1339 e->name = to_atom(p, n);
1340 e->versions[0].mode = 0;
1341 hashclr(e->versions[0].sha1);
1342 t->entries[t->entry_count++] = e;
1343 if (slash1) {
1344 e->tree = new_tree_content(8);
1345 e->versions[1].mode = S_IFDIR;
1346 tree_content_set(e, slash1 + 1, sha1, mode, subtree);
1347 } else {
1348 e->tree = subtree;
1349 e->versions[1].mode = mode;
1350 hashcpy(e->versions[1].sha1, sha1);
1351 }
1352 hashclr(root->versions[1].sha1);
1353 return 1;
1354 }
1355
1356 static int tree_content_remove(
1357 struct tree_entry *root,
1358 const char *p,
1359 struct tree_entry *backup_leaf)
1360 {
1361 struct tree_content *t = root->tree;
1362 const char *slash1;
1363 unsigned int i, n;
1364 struct tree_entry *e;
1365
1366 slash1 = strchr(p, '/');
1367 if (slash1)
1368 n = slash1 - p;
1369 else
1370 n = strlen(p);
1371
1372 for (i = 0; i < t->entry_count; i++) {
1373 e = t->entries[i];
1374 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1375 if (!slash1 || !S_ISDIR(e->versions[1].mode))
1376 goto del_entry;
1377 if (!e->tree)
1378 load_tree(e);
1379 if (tree_content_remove(e, slash1 + 1, backup_leaf)) {
1380 for (n = 0; n < e->tree->entry_count; n++) {
1381 if (e->tree->entries[n]->versions[1].mode) {
1382 hashclr(root->versions[1].sha1);
1383 return 1;
1384 }
1385 }
1386 backup_leaf = NULL;
1387 goto del_entry;
1388 }
1389 return 0;
1390 }
1391 }
1392 return 0;
1393
1394 del_entry:
1395 if (backup_leaf)
1396 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1397 else if (e->tree)
1398 release_tree_content_recursive(e->tree);
1399 e->tree = NULL;
1400 e->versions[1].mode = 0;
1401 hashclr(e->versions[1].sha1);
1402 hashclr(root->versions[1].sha1);
1403 return 1;
1404 }
1405
1406 static int tree_content_get(
1407 struct tree_entry *root,
1408 const char *p,
1409 struct tree_entry *leaf)
1410 {
1411 struct tree_content *t = root->tree;
1412 const char *slash1;
1413 unsigned int i, n;
1414 struct tree_entry *e;
1415
1416 slash1 = strchr(p, '/');
1417 if (slash1)
1418 n = slash1 - p;
1419 else
1420 n = strlen(p);
1421
1422 for (i = 0; i < t->entry_count; i++) {
1423 e = t->entries[i];
1424 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1425 if (!slash1) {
1426 memcpy(leaf, e, sizeof(*leaf));
1427 if (e->tree && is_null_sha1(e->versions[1].sha1))
1428 leaf->tree = dup_tree_content(e->tree);
1429 else
1430 leaf->tree = NULL;
1431 return 1;
1432 }
1433 if (!S_ISDIR(e->versions[1].mode))
1434 return 0;
1435 if (!e->tree)
1436 load_tree(e);
1437 return tree_content_get(e, slash1 + 1, leaf);
1438 }
1439 }
1440 return 0;
1441 }
1442
1443 static int update_branch(struct branch *b)
1444 {
1445 static const char *msg = "fast-import";
1446 struct ref_lock *lock;
1447 unsigned char old_sha1[20];
1448
1449 if (read_ref(b->name, old_sha1))
1450 hashclr(old_sha1);
1451 lock = lock_any_ref_for_update(b->name, old_sha1, 0);
1452 if (!lock)
1453 return error("Unable to lock %s", b->name);
1454 if (!force_update && !is_null_sha1(old_sha1)) {
1455 struct commit *old_cmit, *new_cmit;
1456
1457 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1458 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1459 if (!old_cmit || !new_cmit) {
1460 unlock_ref(lock);
1461 return error("Branch %s is missing commits.", b->name);
1462 }
1463
1464 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1465 unlock_ref(lock);
1466 warning("Not updating %s"
1467 " (new tip %s does not contain %s)",
1468 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1469 return -1;
1470 }
1471 }
1472 if (write_ref_sha1(lock, b->sha1, msg) < 0)
1473 return error("Unable to update %s", b->name);
1474 return 0;
1475 }
1476
1477 static void dump_branches(void)
1478 {
1479 unsigned int i;
1480 struct branch *b;
1481
1482 for (i = 0; i < branch_table_sz; i++) {
1483 for (b = branch_table[i]; b; b = b->table_next_branch)
1484 failure |= update_branch(b);
1485 }
1486 }
1487
1488 static void dump_tags(void)
1489 {
1490 static const char *msg = "fast-import";
1491 struct tag *t;
1492 struct ref_lock *lock;
1493 char ref_name[PATH_MAX];
1494
1495 for (t = first_tag; t; t = t->next_tag) {
1496 sprintf(ref_name, "tags/%s", t->name);
1497 lock = lock_ref_sha1(ref_name, NULL);
1498 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1499 failure |= error("Unable to update %s", ref_name);
1500 }
1501 }
1502
1503 static void dump_marks_helper(FILE *f,
1504 uintmax_t base,
1505 struct mark_set *m)
1506 {
1507 uintmax_t k;
1508 if (m->shift) {
1509 for (k = 0; k < 1024; k++) {
1510 if (m->data.sets[k])
1511 dump_marks_helper(f, (base + k) << m->shift,
1512 m->data.sets[k]);
1513 }
1514 } else {
1515 for (k = 0; k < 1024; k++) {
1516 if (m->data.marked[k])
1517 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1518 sha1_to_hex(m->data.marked[k]->sha1));
1519 }
1520 }
1521 }
1522
1523 static void dump_marks(void)
1524 {
1525 static struct lock_file mark_lock;
1526 int mark_fd;
1527 FILE *f;
1528
1529 if (!mark_file)
1530 return;
1531
1532 mark_fd = hold_lock_file_for_update(&mark_lock, mark_file, 0);
1533 if (mark_fd < 0) {
1534 failure |= error("Unable to write marks file %s: %s",
1535 mark_file, strerror(errno));
1536 return;
1537 }
1538
1539 f = fdopen(mark_fd, "w");
1540 if (!f) {
1541 rollback_lock_file(&mark_lock);
1542 failure |= error("Unable to write marks file %s: %s",
1543 mark_file, strerror(errno));
1544 return;
1545 }
1546
1547 dump_marks_helper(f, 0, marks);
1548 fclose(f);
1549 if (commit_lock_file(&mark_lock))
1550 failure |= error("Unable to write marks file %s: %s",
1551 mark_file, strerror(errno));
1552 }
1553
1554 static int read_next_command(void)
1555 {
1556 static int stdin_eof = 0;
1557
1558 if (stdin_eof) {
1559 unread_command_buf = 0;
1560 return EOF;
1561 }
1562
1563 do {
1564 if (unread_command_buf) {
1565 unread_command_buf = 0;
1566 } else {
1567 struct recent_command *rc;
1568
1569 strbuf_detach(&command_buf, NULL);
1570 stdin_eof = strbuf_getline(&command_buf, stdin, '\n');
1571 if (stdin_eof)
1572 return EOF;
1573
1574 rc = rc_free;
1575 if (rc)
1576 rc_free = rc->next;
1577 else {
1578 rc = cmd_hist.next;
1579 cmd_hist.next = rc->next;
1580 cmd_hist.next->prev = &cmd_hist;
1581 free(rc->buf);
1582 }
1583
1584 rc->buf = command_buf.buf;
1585 rc->prev = cmd_tail;
1586 rc->next = cmd_hist.prev;
1587 rc->prev->next = rc;
1588 cmd_tail = rc;
1589 }
1590 } while (command_buf.buf[0] == '#');
1591
1592 return 0;
1593 }
1594
1595 static void skip_optional_lf(void)
1596 {
1597 int term_char = fgetc(stdin);
1598 if (term_char != '\n' && term_char != EOF)
1599 ungetc(term_char, stdin);
1600 }
1601
1602 static void cmd_mark(void)
1603 {
1604 if (!prefixcmp(command_buf.buf, "mark :")) {
1605 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1606 read_next_command();
1607 }
1608 else
1609 next_mark = 0;
1610 }
1611
1612 static void cmd_data(struct strbuf *sb)
1613 {
1614 strbuf_reset(sb);
1615
1616 if (prefixcmp(command_buf.buf, "data "))
1617 die("Expected 'data n' command, found: %s", command_buf.buf);
1618
1619 if (!prefixcmp(command_buf.buf + 5, "<<")) {
1620 char *term = xstrdup(command_buf.buf + 5 + 2);
1621 size_t term_len = command_buf.len - 5 - 2;
1622
1623 strbuf_detach(&command_buf, NULL);
1624 for (;;) {
1625 if (strbuf_getline(&command_buf, stdin, '\n') == EOF)
1626 die("EOF in data (terminator '%s' not found)", term);
1627 if (term_len == command_buf.len
1628 && !strcmp(term, command_buf.buf))
1629 break;
1630 strbuf_addbuf(sb, &command_buf);
1631 strbuf_addch(sb, '\n');
1632 }
1633 free(term);
1634 }
1635 else {
1636 size_t n = 0, length;
1637
1638 length = strtoul(command_buf.buf + 5, NULL, 10);
1639
1640 while (n < length) {
1641 size_t s = strbuf_fread(sb, length - n, stdin);
1642 if (!s && feof(stdin))
1643 die("EOF in data (%lu bytes remaining)",
1644 (unsigned long)(length - n));
1645 n += s;
1646 }
1647 }
1648
1649 skip_optional_lf();
1650 }
1651
1652 static int validate_raw_date(const char *src, char *result, int maxlen)
1653 {
1654 const char *orig_src = src;
1655 char *endp, sign;
1656
1657 strtoul(src, &endp, 10);
1658 if (endp == src || *endp != ' ')
1659 return -1;
1660
1661 src = endp + 1;
1662 if (*src != '-' && *src != '+')
1663 return -1;
1664 sign = *src;
1665
1666 strtoul(src + 1, &endp, 10);
1667 if (endp == src || *endp || (endp - orig_src) >= maxlen)
1668 return -1;
1669
1670 strcpy(result, orig_src);
1671 return 0;
1672 }
1673
1674 static char *parse_ident(const char *buf)
1675 {
1676 const char *gt;
1677 size_t name_len;
1678 char *ident;
1679
1680 gt = strrchr(buf, '>');
1681 if (!gt)
1682 die("Missing > in ident string: %s", buf);
1683 gt++;
1684 if (*gt != ' ')
1685 die("Missing space after > in ident string: %s", buf);
1686 gt++;
1687 name_len = gt - buf;
1688 ident = xmalloc(name_len + 24);
1689 strncpy(ident, buf, name_len);
1690
1691 switch (whenspec) {
1692 case WHENSPEC_RAW:
1693 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1694 die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1695 break;
1696 case WHENSPEC_RFC2822:
1697 if (parse_date(gt, ident + name_len, 24) < 0)
1698 die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1699 break;
1700 case WHENSPEC_NOW:
1701 if (strcmp("now", gt))
1702 die("Date in ident must be 'now': %s", buf);
1703 datestamp(ident + name_len, 24);
1704 break;
1705 }
1706
1707 return ident;
1708 }
1709
1710 static void cmd_new_blob(void)
1711 {
1712 static struct strbuf buf = STRBUF_INIT;
1713
1714 read_next_command();
1715 cmd_mark();
1716 cmd_data(&buf);
1717 store_object(OBJ_BLOB, &buf, &last_blob, NULL, next_mark);
1718 }
1719
1720 static void unload_one_branch(void)
1721 {
1722 while (cur_active_branches
1723 && cur_active_branches >= max_active_branches) {
1724 uintmax_t min_commit = ULONG_MAX;
1725 struct branch *e, *l = NULL, *p = NULL;
1726
1727 for (e = active_branches; e; e = e->active_next_branch) {
1728 if (e->last_commit < min_commit) {
1729 p = l;
1730 min_commit = e->last_commit;
1731 }
1732 l = e;
1733 }
1734
1735 if (p) {
1736 e = p->active_next_branch;
1737 p->active_next_branch = e->active_next_branch;
1738 } else {
1739 e = active_branches;
1740 active_branches = e->active_next_branch;
1741 }
1742 e->active = 0;
1743 e->active_next_branch = NULL;
1744 if (e->branch_tree.tree) {
1745 release_tree_content_recursive(e->branch_tree.tree);
1746 e->branch_tree.tree = NULL;
1747 }
1748 cur_active_branches--;
1749 }
1750 }
1751
1752 static void load_branch(struct branch *b)
1753 {
1754 load_tree(&b->branch_tree);
1755 if (!b->active) {
1756 b->active = 1;
1757 b->active_next_branch = active_branches;
1758 active_branches = b;
1759 cur_active_branches++;
1760 branch_load_count++;
1761 }
1762 }
1763
1764 static void file_change_m(struct branch *b)
1765 {
1766 const char *p = command_buf.buf + 2;
1767 static struct strbuf uq = STRBUF_INIT;
1768 const char *endp;
1769 struct object_entry *oe = oe;
1770 unsigned char sha1[20];
1771 uint16_t mode, inline_data = 0;
1772
1773 p = get_mode(p, &mode);
1774 if (!p)
1775 die("Corrupt mode: %s", command_buf.buf);
1776 switch (mode) {
1777 case S_IFREG | 0644:
1778 case S_IFREG | 0755:
1779 case S_IFLNK:
1780 case 0644:
1781 case 0755:
1782 /* ok */
1783 break;
1784 default:
1785 die("Corrupt mode: %s", command_buf.buf);
1786 }
1787
1788 if (*p == ':') {
1789 char *x;
1790 oe = find_mark(strtoumax(p + 1, &x, 10));
1791 hashcpy(sha1, oe->sha1);
1792 p = x;
1793 } else if (!prefixcmp(p, "inline")) {
1794 inline_data = 1;
1795 p += 6;
1796 } else {
1797 if (get_sha1_hex(p, sha1))
1798 die("Invalid SHA1: %s", command_buf.buf);
1799 oe = find_object(sha1);
1800 p += 40;
1801 }
1802 if (*p++ != ' ')
1803 die("Missing space after SHA1: %s", command_buf.buf);
1804
1805 strbuf_reset(&uq);
1806 if (!unquote_c_style(&uq, p, &endp)) {
1807 if (*endp)
1808 die("Garbage after path in: %s", command_buf.buf);
1809 p = uq.buf;
1810 }
1811
1812 if (inline_data) {
1813 static struct strbuf buf = STRBUF_INIT;
1814
1815 if (p != uq.buf) {
1816 strbuf_addstr(&uq, p);
1817 p = uq.buf;
1818 }
1819 read_next_command();
1820 cmd_data(&buf);
1821 store_object(OBJ_BLOB, &buf, &last_blob, sha1, 0);
1822 } else if (oe) {
1823 if (oe->type != OBJ_BLOB)
1824 die("Not a blob (actually a %s): %s",
1825 typename(oe->type), command_buf.buf);
1826 } else {
1827 enum object_type type = sha1_object_info(sha1, NULL);
1828 if (type < 0)
1829 die("Blob not found: %s", command_buf.buf);
1830 if (type != OBJ_BLOB)
1831 die("Not a blob (actually a %s): %s",
1832 typename(type), command_buf.buf);
1833 }
1834
1835 tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode, NULL);
1836 }
1837
1838 static void file_change_d(struct branch *b)
1839 {
1840 const char *p = command_buf.buf + 2;
1841 static struct strbuf uq = STRBUF_INIT;
1842 const char *endp;
1843
1844 strbuf_reset(&uq);
1845 if (!unquote_c_style(&uq, p, &endp)) {
1846 if (*endp)
1847 die("Garbage after path in: %s", command_buf.buf);
1848 p = uq.buf;
1849 }
1850 tree_content_remove(&b->branch_tree, p, NULL);
1851 }
1852
1853 static void file_change_cr(struct branch *b, int rename)
1854 {
1855 const char *s, *d;
1856 static struct strbuf s_uq = STRBUF_INIT;
1857 static struct strbuf d_uq = STRBUF_INIT;
1858 const char *endp;
1859 struct tree_entry leaf;
1860
1861 s = command_buf.buf + 2;
1862 strbuf_reset(&s_uq);
1863 if (!unquote_c_style(&s_uq, s, &endp)) {
1864 if (*endp != ' ')
1865 die("Missing space after source: %s", command_buf.buf);
1866 } else {
1867 endp = strchr(s, ' ');
1868 if (!endp)
1869 die("Missing space after source: %s", command_buf.buf);
1870 strbuf_add(&s_uq, s, endp - s);
1871 }
1872 s = s_uq.buf;
1873
1874 endp++;
1875 if (!*endp)
1876 die("Missing dest: %s", command_buf.buf);
1877
1878 d = endp;
1879 strbuf_reset(&d_uq);
1880 if (!unquote_c_style(&d_uq, d, &endp)) {
1881 if (*endp)
1882 die("Garbage after dest in: %s", command_buf.buf);
1883 d = d_uq.buf;
1884 }
1885
1886 memset(&leaf, 0, sizeof(leaf));
1887 if (rename)
1888 tree_content_remove(&b->branch_tree, s, &leaf);
1889 else
1890 tree_content_get(&b->branch_tree, s, &leaf);
1891 if (!leaf.versions[1].mode)
1892 die("Path %s not in branch", s);
1893 tree_content_set(&b->branch_tree, d,
1894 leaf.versions[1].sha1,
1895 leaf.versions[1].mode,
1896 leaf.tree);
1897 }
1898
1899 static void file_change_deleteall(struct branch *b)
1900 {
1901 release_tree_content_recursive(b->branch_tree.tree);
1902 hashclr(b->branch_tree.versions[0].sha1);
1903 hashclr(b->branch_tree.versions[1].sha1);
1904 load_tree(&b->branch_tree);
1905 }
1906
1907 static void cmd_from_commit(struct branch *b, char *buf, unsigned long size)
1908 {
1909 if (!buf || size < 46)
1910 die("Not a valid commit: %s", sha1_to_hex(b->sha1));
1911 if (memcmp("tree ", buf, 5)
1912 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1913 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1914 hashcpy(b->branch_tree.versions[0].sha1,
1915 b->branch_tree.versions[1].sha1);
1916 }
1917
1918 static void cmd_from_existing(struct branch *b)
1919 {
1920 if (is_null_sha1(b->sha1)) {
1921 hashclr(b->branch_tree.versions[0].sha1);
1922 hashclr(b->branch_tree.versions[1].sha1);
1923 } else {
1924 unsigned long size;
1925 char *buf;
1926
1927 buf = read_object_with_reference(b->sha1,
1928 commit_type, &size, b->sha1);
1929 cmd_from_commit(b, buf, size);
1930 free(buf);
1931 }
1932 }
1933
1934 static int cmd_from(struct branch *b)
1935 {
1936 const char *from;
1937 struct branch *s;
1938
1939 if (prefixcmp(command_buf.buf, "from "))
1940 return 0;
1941
1942 if (b->branch_tree.tree) {
1943 release_tree_content_recursive(b->branch_tree.tree);
1944 b->branch_tree.tree = NULL;
1945 }
1946
1947 from = strchr(command_buf.buf, ' ') + 1;
1948 s = lookup_branch(from);
1949 if (b == s)
1950 die("Can't create a branch from itself: %s", b->name);
1951 else if (s) {
1952 unsigned char *t = s->branch_tree.versions[1].sha1;
1953 hashcpy(b->sha1, s->sha1);
1954 hashcpy(b->branch_tree.versions[0].sha1, t);
1955 hashcpy(b->branch_tree.versions[1].sha1, t);
1956 } else if (*from == ':') {
1957 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1958 struct object_entry *oe = find_mark(idnum);
1959 if (oe->type != OBJ_COMMIT)
1960 die("Mark :%" PRIuMAX " not a commit", idnum);
1961 hashcpy(b->sha1, oe->sha1);
1962 if (oe->pack_id != MAX_PACK_ID) {
1963 unsigned long size;
1964 char *buf = gfi_unpack_entry(oe, &size);
1965 cmd_from_commit(b, buf, size);
1966 free(buf);
1967 } else
1968 cmd_from_existing(b);
1969 } else if (!get_sha1(from, b->sha1))
1970 cmd_from_existing(b);
1971 else
1972 die("Invalid ref name or SHA1 expression: %s", from);
1973
1974 read_next_command();
1975 return 1;
1976 }
1977
1978 static struct hash_list *cmd_merge(unsigned int *count)
1979 {
1980 struct hash_list *list = NULL, *n, *e = e;
1981 const char *from;
1982 struct branch *s;
1983
1984 *count = 0;
1985 while (!prefixcmp(command_buf.buf, "merge ")) {
1986 from = strchr(command_buf.buf, ' ') + 1;
1987 n = xmalloc(sizeof(*n));
1988 s = lookup_branch(from);
1989 if (s)
1990 hashcpy(n->sha1, s->sha1);
1991 else if (*from == ':') {
1992 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1993 struct object_entry *oe = find_mark(idnum);
1994 if (oe->type != OBJ_COMMIT)
1995 die("Mark :%" PRIuMAX " not a commit", idnum);
1996 hashcpy(n->sha1, oe->sha1);
1997 } else if (!get_sha1(from, n->sha1)) {
1998 unsigned long size;
1999 char *buf = read_object_with_reference(n->sha1,
2000 commit_type, &size, n->sha1);
2001 if (!buf || size < 46)
2002 die("Not a valid commit: %s", from);
2003 free(buf);
2004 } else
2005 die("Invalid ref name or SHA1 expression: %s", from);
2006
2007 n->next = NULL;
2008 if (list)
2009 e->next = n;
2010 else
2011 list = n;
2012 e = n;
2013 (*count)++;
2014 read_next_command();
2015 }
2016 return list;
2017 }
2018
2019 static void cmd_new_commit(void)
2020 {
2021 static struct strbuf msg = STRBUF_INIT;
2022 struct branch *b;
2023 char *sp;
2024 char *author = NULL;
2025 char *committer = NULL;
2026 struct hash_list *merge_list = NULL;
2027 unsigned int merge_count;
2028
2029 /* Obtain the branch name from the rest of our command */
2030 sp = strchr(command_buf.buf, ' ') + 1;
2031 b = lookup_branch(sp);
2032 if (!b)
2033 b = new_branch(sp);
2034
2035 read_next_command();
2036 cmd_mark();
2037 if (!prefixcmp(command_buf.buf, "author ")) {
2038 author = parse_ident(command_buf.buf + 7);
2039 read_next_command();
2040 }
2041 if (!prefixcmp(command_buf.buf, "committer ")) {
2042 committer = parse_ident(command_buf.buf + 10);
2043 read_next_command();
2044 }
2045 if (!committer)
2046 die("Expected committer but didn't get one");
2047 cmd_data(&msg);
2048 read_next_command();
2049 cmd_from(b);
2050 merge_list = cmd_merge(&merge_count);
2051
2052 /* ensure the branch is active/loaded */
2053 if (!b->branch_tree.tree || !max_active_branches) {
2054 unload_one_branch();
2055 load_branch(b);
2056 }
2057
2058 /* file_change* */
2059 while (command_buf.len > 0) {
2060 if (!prefixcmp(command_buf.buf, "M "))
2061 file_change_m(b);
2062 else if (!prefixcmp(command_buf.buf, "D "))
2063 file_change_d(b);
2064 else if (!prefixcmp(command_buf.buf, "R "))
2065 file_change_cr(b, 1);
2066 else if (!prefixcmp(command_buf.buf, "C "))
2067 file_change_cr(b, 0);
2068 else if (!strcmp("deleteall", command_buf.buf))
2069 file_change_deleteall(b);
2070 else {
2071 unread_command_buf = 1;
2072 break;
2073 }
2074 if (read_next_command() == EOF)
2075 break;
2076 }
2077
2078 /* build the tree and the commit */
2079 store_tree(&b->branch_tree);
2080 hashcpy(b->branch_tree.versions[0].sha1,
2081 b->branch_tree.versions[1].sha1);
2082
2083 strbuf_reset(&new_data);
2084 strbuf_addf(&new_data, "tree %s\n",
2085 sha1_to_hex(b->branch_tree.versions[1].sha1));
2086 if (!is_null_sha1(b->sha1))
2087 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(b->sha1));
2088 while (merge_list) {
2089 struct hash_list *next = merge_list->next;
2090 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(merge_list->sha1));
2091 free(merge_list);
2092 merge_list = next;
2093 }
2094 strbuf_addf(&new_data,
2095 "author %s\n"
2096 "committer %s\n"
2097 "\n",
2098 author ? author : committer, committer);
2099 strbuf_addbuf(&new_data, &msg);
2100 free(author);
2101 free(committer);
2102
2103 if (!store_object(OBJ_COMMIT, &new_data, NULL, b->sha1, next_mark))
2104 b->pack_id = pack_id;
2105 b->last_commit = object_count_by_type[OBJ_COMMIT];
2106 }
2107
2108 static void cmd_new_tag(void)
2109 {
2110 static struct strbuf msg = STRBUF_INIT;
2111 char *sp;
2112 const char *from;
2113 char *tagger;
2114 struct branch *s;
2115 struct tag *t;
2116 uintmax_t from_mark = 0;
2117 unsigned char sha1[20];
2118
2119 /* Obtain the new tag name from the rest of our command */
2120 sp = strchr(command_buf.buf, ' ') + 1;
2121 t = pool_alloc(sizeof(struct tag));
2122 t->next_tag = NULL;
2123 t->name = pool_strdup(sp);
2124 if (last_tag)
2125 last_tag->next_tag = t;
2126 else
2127 first_tag = t;
2128 last_tag = t;
2129 read_next_command();
2130
2131 /* from ... */
2132 if (prefixcmp(command_buf.buf, "from "))
2133 die("Expected from command, got %s", command_buf.buf);
2134 from = strchr(command_buf.buf, ' ') + 1;
2135 s = lookup_branch(from);
2136 if (s) {
2137 hashcpy(sha1, s->sha1);
2138 } else if (*from == ':') {
2139 struct object_entry *oe;
2140 from_mark = strtoumax(from + 1, NULL, 10);
2141 oe = find_mark(from_mark);
2142 if (oe->type != OBJ_COMMIT)
2143 die("Mark :%" PRIuMAX " not a commit", from_mark);
2144 hashcpy(sha1, oe->sha1);
2145 } else if (!get_sha1(from, sha1)) {
2146 unsigned long size;
2147 char *buf;
2148
2149 buf = read_object_with_reference(sha1,
2150 commit_type, &size, sha1);
2151 if (!buf || size < 46)
2152 die("Not a valid commit: %s", from);
2153 free(buf);
2154 } else
2155 die("Invalid ref name or SHA1 expression: %s", from);
2156 read_next_command();
2157
2158 /* tagger ... */
2159 if (prefixcmp(command_buf.buf, "tagger "))
2160 die("Expected tagger command, got %s", command_buf.buf);
2161 tagger = parse_ident(command_buf.buf + 7);
2162
2163 /* tag payload/message */
2164 read_next_command();
2165 cmd_data(&msg);
2166
2167 /* build the tag object */
2168 strbuf_reset(&new_data);
2169 strbuf_addf(&new_data,
2170 "object %s\n"
2171 "type %s\n"
2172 "tag %s\n"
2173 "tagger %s\n"
2174 "\n",
2175 sha1_to_hex(sha1), commit_type, t->name, tagger);
2176 strbuf_addbuf(&new_data, &msg);
2177 free(tagger);
2178
2179 if (store_object(OBJ_TAG, &new_data, NULL, t->sha1, 0))
2180 t->pack_id = MAX_PACK_ID;
2181 else
2182 t->pack_id = pack_id;
2183 }
2184
2185 static void cmd_reset_branch(void)
2186 {
2187 struct branch *b;
2188 char *sp;
2189
2190 /* Obtain the branch name from the rest of our command */
2191 sp = strchr(command_buf.buf, ' ') + 1;
2192 b = lookup_branch(sp);
2193 if (b) {
2194 hashclr(b->sha1);
2195 hashclr(b->branch_tree.versions[0].sha1);
2196 hashclr(b->branch_tree.versions[1].sha1);
2197 if (b->branch_tree.tree) {
2198 release_tree_content_recursive(b->branch_tree.tree);
2199 b->branch_tree.tree = NULL;
2200 }
2201 }
2202 else
2203 b = new_branch(sp);
2204 read_next_command();
2205 if (!cmd_from(b) && command_buf.len > 0)
2206 unread_command_buf = 1;
2207 }
2208
2209 static void cmd_checkpoint(void)
2210 {
2211 if (object_count) {
2212 cycle_packfile();
2213 dump_branches();
2214 dump_tags();
2215 dump_marks();
2216 }
2217 skip_optional_lf();
2218 }
2219
2220 static void cmd_progress(void)
2221 {
2222 fwrite(command_buf.buf, 1, command_buf.len, stdout);
2223 fputc('\n', stdout);
2224 fflush(stdout);
2225 skip_optional_lf();
2226 }
2227
2228 static void import_marks(const char *input_file)
2229 {
2230 char line[512];
2231 FILE *f = fopen(input_file, "r");
2232 if (!f)
2233 die("cannot read %s: %s", input_file, strerror(errno));
2234 while (fgets(line, sizeof(line), f)) {
2235 uintmax_t mark;
2236 char *end;
2237 unsigned char sha1[20];
2238 struct object_entry *e;
2239
2240 end = strchr(line, '\n');
2241 if (line[0] != ':' || !end)
2242 die("corrupt mark line: %s", line);
2243 *end = 0;
2244 mark = strtoumax(line + 1, &end, 10);
2245 if (!mark || end == line + 1
2246 || *end != ' ' || get_sha1(end + 1, sha1))
2247 die("corrupt mark line: %s", line);
2248 e = find_object(sha1);
2249 if (!e) {
2250 enum object_type type = sha1_object_info(sha1, NULL);
2251 if (type < 0)
2252 die("object not found: %s", sha1_to_hex(sha1));
2253 e = insert_object(sha1);
2254 e->type = type;
2255 e->pack_id = MAX_PACK_ID;
2256 e->offset = 1; /* just not zero! */
2257 }
2258 insert_mark(mark, e);
2259 }
2260 fclose(f);
2261 }
2262
2263 static const char fast_import_usage[] =
2264 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
2265
2266 int main(int argc, const char **argv)
2267 {
2268 unsigned int i, show_stats = 1;
2269
2270 git_config(git_default_config);
2271 alloc_objects(object_entry_alloc);
2272 strbuf_init(&command_buf, 0);
2273 atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2274 branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2275 avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2276 marks = pool_calloc(1, sizeof(struct mark_set));
2277
2278 for (i = 1; i < argc; i++) {
2279 const char *a = argv[i];
2280
2281 if (*a != '-' || !strcmp(a, "--"))
2282 break;
2283 else if (!prefixcmp(a, "--date-format=")) {
2284 const char *fmt = a + 14;
2285 if (!strcmp(fmt, "raw"))
2286 whenspec = WHENSPEC_RAW;
2287 else if (!strcmp(fmt, "rfc2822"))
2288 whenspec = WHENSPEC_RFC2822;
2289 else if (!strcmp(fmt, "now"))
2290 whenspec = WHENSPEC_NOW;
2291 else
2292 die("unknown --date-format argument %s", fmt);
2293 }
2294 else if (!prefixcmp(a, "--max-pack-size="))
2295 max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2296 else if (!prefixcmp(a, "--depth=")) {
2297 max_depth = strtoul(a + 8, NULL, 0);
2298 if (max_depth > MAX_DEPTH)
2299 die("--depth cannot exceed %u", MAX_DEPTH);
2300 }
2301 else if (!prefixcmp(a, "--active-branches="))
2302 max_active_branches = strtoul(a + 18, NULL, 0);
2303 else if (!prefixcmp(a, "--import-marks="))
2304 import_marks(a + 15);
2305 else if (!prefixcmp(a, "--export-marks="))
2306 mark_file = a + 15;
2307 else if (!prefixcmp(a, "--export-pack-edges=")) {
2308 if (pack_edges)
2309 fclose(pack_edges);
2310 pack_edges = fopen(a + 20, "a");
2311 if (!pack_edges)
2312 die("Cannot open %s: %s", a + 20, strerror(errno));
2313 } else if (!strcmp(a, "--force"))
2314 force_update = 1;
2315 else if (!strcmp(a, "--quiet"))
2316 show_stats = 0;
2317 else if (!strcmp(a, "--stats"))
2318 show_stats = 1;
2319 else
2320 die("unknown option %s", a);
2321 }
2322 if (i != argc)
2323 usage(fast_import_usage);
2324
2325 rc_free = pool_alloc(cmd_save * sizeof(*rc_free));
2326 for (i = 0; i < (cmd_save - 1); i++)
2327 rc_free[i].next = &rc_free[i + 1];
2328 rc_free[cmd_save - 1].next = NULL;
2329
2330 prepare_packed_git();
2331 start_packfile();
2332 set_die_routine(die_nicely);
2333 while (read_next_command() != EOF) {
2334 if (!strcmp("blob", command_buf.buf))
2335 cmd_new_blob();
2336 else if (!prefixcmp(command_buf.buf, "commit "))
2337 cmd_new_commit();
2338 else if (!prefixcmp(command_buf.buf, "tag "))
2339 cmd_new_tag();
2340 else if (!prefixcmp(command_buf.buf, "reset "))
2341 cmd_reset_branch();
2342 else if (!strcmp("checkpoint", command_buf.buf))
2343 cmd_checkpoint();
2344 else if (!prefixcmp(command_buf.buf, "progress "))
2345 cmd_progress();
2346 else
2347 die("Unsupported command: %s", command_buf.buf);
2348 }
2349 end_packfile();
2350
2351 dump_branches();
2352 dump_tags();
2353 unkeep_all_packs();
2354 dump_marks();
2355
2356 if (pack_edges)
2357 fclose(pack_edges);
2358
2359 if (show_stats) {
2360 uintmax_t total_count = 0, duplicate_count = 0;
2361 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2362 total_count += object_count_by_type[i];
2363 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2364 duplicate_count += duplicate_count_by_type[i];
2365
2366 fprintf(stderr, "%s statistics:\n", argv[0]);
2367 fprintf(stderr, "---------------------------------------------------------------------\n");
2368 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2369 fprintf(stderr, "Total objects: %10" PRIuMAX " (%10" PRIuMAX " duplicates )\n", total_count, duplicate_count);
2370 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]);
2371 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]);
2372 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]);
2373 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]);
2374 fprintf(stderr, "Total branches: %10lu (%10lu loads )\n", branch_count, branch_load_count);
2375 fprintf(stderr, " marks: %10" PRIuMAX " (%10" PRIuMAX " unique )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2376 fprintf(stderr, " atoms: %10u\n", atom_cnt);
2377 fprintf(stderr, "Memory total: %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2378 fprintf(stderr, " pools: %10lu KiB\n", (unsigned long)(total_allocd/1024));
2379 fprintf(stderr, " objects: %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2380 fprintf(stderr, "---------------------------------------------------------------------\n");
2381 pack_report();
2382 fprintf(stderr, "---------------------------------------------------------------------\n");
2383 fprintf(stderr, "\n");
2384 }
2385
2386 return failure ? 1 : 0;
2387 }
Gitweb development nest