?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
hg-to-git: improve popen calls
[git/jnareb-git.git] / fast-import.c
blob82e916133982800420e77d1824a2714dd120550e
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_or_die(keep_fd, keep_msg, strlen(keep_msg));
882 if (close(keep_fd))
883 die("failed to write keep file");
884
885 snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
886 get_object_directory(), sha1_to_hex(pack_data->sha1));
887 if (move_temp_to_file(pack_data->pack_name, name))
888 die("cannot store pack file");
889
890 snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
891 get_object_directory(), sha1_to_hex(pack_data->sha1));
892 if (move_temp_to_file(curr_index_name, name))
893 die("cannot store index file");
894 return name;
895 }
896
897 static void unkeep_all_packs(void)
898 {
899 static char name[PATH_MAX];
900 int k;
901
902 for (k = 0; k < pack_id; k++) {
903 struct packed_git *p = all_packs[k];
904 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
905 get_object_directory(), sha1_to_hex(p->sha1));
906 unlink(name);
907 }
908 }
909
910 static void end_packfile(void)
911 {
912 struct packed_git *old_p = pack_data, *new_p;
913
914 if (object_count) {
915 char *idx_name;
916 int i;
917 struct branch *b;
918 struct tag *t;
919
920 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
921 pack_data->pack_name, object_count);
922 close(pack_data->pack_fd);
923 idx_name = keep_pack(create_index());
924
925 /* Register the packfile with core git's machinary. */
926 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
927 if (!new_p)
928 die("core git rejected index %s", idx_name);
929 new_p->windows = old_p->windows;
930 all_packs[pack_id] = new_p;
931 install_packed_git(new_p);
932
933 /* Print the boundary */
934 if (pack_edges) {
935 fprintf(pack_edges, "%s:", new_p->pack_name);
936 for (i = 0; i < branch_table_sz; i++) {
937 for (b = branch_table[i]; b; b = b->table_next_branch) {
938 if (b->pack_id == pack_id)
939 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
940 }
941 }
942 for (t = first_tag; t; t = t->next_tag) {
943 if (t->pack_id == pack_id)
944 fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
945 }
946 fputc('\n', pack_edges);
947 fflush(pack_edges);
948 }
949
950 pack_id++;
951 }
952 else
953 unlink(old_p->pack_name);
954 free(old_p);
955
956 /* We can't carry a delta across packfiles. */
957 strbuf_release(&last_blob.data);
958 last_blob.offset = 0;
959 last_blob.depth = 0;
960 }
961
962 static void cycle_packfile(void)
963 {
964 end_packfile();
965 start_packfile();
966 }
967
968 static size_t encode_header(
969 enum object_type type,
970 size_t size,
971 unsigned char *hdr)
972 {
973 int n = 1;
974 unsigned char c;
975
976 if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
977 die("bad type %d", type);
978
979 c = (type << 4) | (size & 15);
980 size >>= 4;
981 while (size) {
982 *hdr++ = c | 0x80;
983 c = size & 0x7f;
984 size >>= 7;
985 n++;
986 }
987 *hdr = c;
988 return n;
989 }
990
991 static int store_object(
992 enum object_type type,
993 struct strbuf *dat,
994 struct last_object *last,
995 unsigned char *sha1out,
996 uintmax_t mark)
997 {
998 void *out, *delta;
999 struct object_entry *e;
1000 unsigned char hdr[96];
1001 unsigned char sha1[20];
1002 unsigned long hdrlen, deltalen;
1003 SHA_CTX c;
1004 z_stream s;
1005
1006 hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
1007 (unsigned long)dat->len) + 1;
1008 SHA1_Init(&c);
1009 SHA1_Update(&c, hdr, hdrlen);
1010 SHA1_Update(&c, dat->buf, dat->len);
1011 SHA1_Final(sha1, &c);
1012 if (sha1out)
1013 hashcpy(sha1out, sha1);
1014
1015 e = insert_object(sha1);
1016 if (mark)
1017 insert_mark(mark, e);
1018 if (e->offset) {
1019 duplicate_count_by_type[type]++;
1020 return 1;
1021 } else if (find_sha1_pack(sha1, packed_git)) {
1022 e->type = type;
1023 e->pack_id = MAX_PACK_ID;
1024 e->offset = 1; /* just not zero! */
1025 duplicate_count_by_type[type]++;
1026 return 1;
1027 }
1028
1029 if (last && last->data.buf && last->depth < max_depth) {
1030 delta = diff_delta(last->data.buf, last->data.len,
1031 dat->buf, dat->len,
1032 &deltalen, 0);
1033 if (delta && deltalen >= dat->len) {
1034 free(delta);
1035 delta = NULL;
1036 }
1037 } else
1038 delta = NULL;
1039
1040 memset(&s, 0, sizeof(s));
1041 deflateInit(&s, zlib_compression_level);
1042 if (delta) {
1043 s.next_in = delta;
1044 s.avail_in = deltalen;
1045 } else {
1046 s.next_in = (void *)dat->buf;
1047 s.avail_in = dat->len;
1048 }
1049 s.avail_out = deflateBound(&s, s.avail_in);
1050 s.next_out = out = xmalloc(s.avail_out);
1051 while (deflate(&s, Z_FINISH) == Z_OK)
1052 /* nothing */;
1053 deflateEnd(&s);
1054
1055 /* Determine if we should auto-checkpoint. */
1056 if ((pack_size + 60 + s.total_out) > max_packsize
1057 || (pack_size + 60 + s.total_out) < pack_size) {
1058
1059 /* This new object needs to *not* have the current pack_id. */
1060 e->pack_id = pack_id + 1;
1061 cycle_packfile();
1062
1063 /* We cannot carry a delta into the new pack. */
1064 if (delta) {
1065 free(delta);
1066 delta = NULL;
1067
1068 memset(&s, 0, sizeof(s));
1069 deflateInit(&s, zlib_compression_level);
1070 s.next_in = (void *)dat->buf;
1071 s.avail_in = dat->len;
1072 s.avail_out = deflateBound(&s, s.avail_in);
1073 s.next_out = out = xrealloc(out, s.avail_out);
1074 while (deflate(&s, Z_FINISH) == Z_OK)
1075 /* nothing */;
1076 deflateEnd(&s);
1077 }
1078 }
1079
1080 e->type = type;
1081 e->pack_id = pack_id;
1082 e->offset = pack_size;
1083 object_count++;
1084 object_count_by_type[type]++;
1085
1086 if (delta) {
1087 unsigned long ofs = e->offset - last->offset;
1088 unsigned pos = sizeof(hdr) - 1;
1089
1090 delta_count_by_type[type]++;
1091 e->depth = last->depth + 1;
1092
1093 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
1094 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1095 pack_size += hdrlen;
1096
1097 hdr[pos] = ofs & 127;
1098 while (ofs >>= 7)
1099 hdr[--pos] = 128 | (--ofs & 127);
1100 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
1101 pack_size += sizeof(hdr) - pos;
1102 } else {
1103 e->depth = 0;
1104 hdrlen = encode_header(type, dat->len, hdr);
1105 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1106 pack_size += hdrlen;
1107 }
1108
1109 write_or_die(pack_data->pack_fd, out, s.total_out);
1110 pack_size += s.total_out;
1111
1112 free(out);
1113 free(delta);
1114 if (last) {
1115 if (last->no_swap) {
1116 last->data = *dat;
1117 } else {
1118 strbuf_swap(&last->data, dat);
1119 }
1120 last->offset = e->offset;
1121 last->depth = e->depth;
1122 }
1123 return 0;
1124 }
1125
1126 static void *gfi_unpack_entry(
1127 struct object_entry *oe,
1128 unsigned long *sizep)
1129 {
1130 enum object_type type;
1131 struct packed_git *p = all_packs[oe->pack_id];
1132 if (p == pack_data)
1133 p->pack_size = pack_size + 20;
1134 return unpack_entry(p, oe->offset, &type, sizep);
1135 }
1136
1137 static const char *get_mode(const char *str, uint16_t *modep)
1138 {
1139 unsigned char c;
1140 uint16_t mode = 0;
1141
1142 while ((c = *str++) != ' ') {
1143 if (c < '0' || c > '7')
1144 return NULL;
1145 mode = (mode << 3) + (c - '0');
1146 }
1147 *modep = mode;
1148 return str;
1149 }
1150
1151 static void load_tree(struct tree_entry *root)
1152 {
1153 unsigned char* sha1 = root->versions[1].sha1;
1154 struct object_entry *myoe;
1155 struct tree_content *t;
1156 unsigned long size;
1157 char *buf;
1158 const char *c;
1159
1160 root->tree = t = new_tree_content(8);
1161 if (is_null_sha1(sha1))
1162 return;
1163
1164 myoe = find_object(sha1);
1165 if (myoe && myoe->pack_id != MAX_PACK_ID) {
1166 if (myoe->type != OBJ_TREE)
1167 die("Not a tree: %s", sha1_to_hex(sha1));
1168 t->delta_depth = myoe->depth;
1169 buf = gfi_unpack_entry(myoe, &size);
1170 } else {
1171 enum object_type type;
1172 buf = read_sha1_file(sha1, &type, &size);
1173 if (!buf || type != OBJ_TREE)
1174 die("Can't load tree %s", sha1_to_hex(sha1));
1175 }
1176
1177 c = buf;
1178 while (c != (buf + size)) {
1179 struct tree_entry *e = new_tree_entry();
1180
1181 if (t->entry_count == t->entry_capacity)
1182 root->tree = t = grow_tree_content(t, t->entry_count);
1183 t->entries[t->entry_count++] = e;
1184
1185 e->tree = NULL;
1186 c = get_mode(c, &e->versions[1].mode);
1187 if (!c)
1188 die("Corrupt mode in %s", sha1_to_hex(sha1));
1189 e->versions[0].mode = e->versions[1].mode;
1190 e->name = to_atom(c, strlen(c));
1191 c += e->name->str_len + 1;
1192 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1193 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1194 c += 20;
1195 }
1196 free(buf);
1197 }
1198
1199 static int tecmp0 (const void *_a, const void *_b)
1200 {
1201 struct tree_entry *a = *((struct tree_entry**)_a);
1202 struct tree_entry *b = *((struct tree_entry**)_b);
1203 return base_name_compare(
1204 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1205 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1206 }
1207
1208 static int tecmp1 (const void *_a, const void *_b)
1209 {
1210 struct tree_entry *a = *((struct tree_entry**)_a);
1211 struct tree_entry *b = *((struct tree_entry**)_b);
1212 return base_name_compare(
1213 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1214 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1215 }
1216
1217 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1218 {
1219 size_t maxlen = 0;
1220 unsigned int i;
1221
1222 if (!v)
1223 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1224 else
1225 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1226
1227 for (i = 0; i < t->entry_count; i++) {
1228 if (t->entries[i]->versions[v].mode)
1229 maxlen += t->entries[i]->name->str_len + 34;
1230 }
1231
1232 strbuf_reset(b);
1233 strbuf_grow(b, maxlen);
1234 for (i = 0; i < t->entry_count; i++) {
1235 struct tree_entry *e = t->entries[i];
1236 if (!e->versions[v].mode)
1237 continue;
1238 strbuf_addf(b, "%o %s%c", (unsigned int)e->versions[v].mode,
1239 e->name->str_dat, '\0');
1240 strbuf_add(b, e->versions[v].sha1, 20);
1241 }
1242 }
1243
1244 static void store_tree(struct tree_entry *root)
1245 {
1246 struct tree_content *t = root->tree;
1247 unsigned int i, j, del;
1248 struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1249 struct object_entry *le;
1250
1251 if (!is_null_sha1(root->versions[1].sha1))
1252 return;
1253
1254 for (i = 0; i < t->entry_count; i++) {
1255 if (t->entries[i]->tree)
1256 store_tree(t->entries[i]);
1257 }
1258
1259 le = find_object(root->versions[0].sha1);
1260 if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1261 mktree(t, 0, &old_tree);
1262 lo.data = old_tree;
1263 lo.offset = le->offset;
1264 lo.depth = t->delta_depth;
1265 }
1266
1267 mktree(t, 1, &new_tree);
1268 store_object(OBJ_TREE, &new_tree, &lo, root->versions[1].sha1, 0);
1269
1270 t->delta_depth = lo.depth;
1271 for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1272 struct tree_entry *e = t->entries[i];
1273 if (e->versions[1].mode) {
1274 e->versions[0].mode = e->versions[1].mode;
1275 hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1276 t->entries[j++] = e;
1277 } else {
1278 release_tree_entry(e);
1279 del++;
1280 }
1281 }
1282 t->entry_count -= del;
1283 }
1284
1285 static int tree_content_set(
1286 struct tree_entry *root,
1287 const char *p,
1288 const unsigned char *sha1,
1289 const uint16_t mode,
1290 struct tree_content *subtree)
1291 {
1292 struct tree_content *t = root->tree;
1293 const char *slash1;
1294 unsigned int i, n;
1295 struct tree_entry *e;
1296
1297 slash1 = strchr(p, '/');
1298 if (slash1)
1299 n = slash1 - p;
1300 else
1301 n = strlen(p);
1302 if (!n)
1303 die("Empty path component found in input");
1304 if (!slash1 && !S_ISDIR(mode) && subtree)
1305 die("Non-directories cannot have subtrees");
1306
1307 for (i = 0; i < t->entry_count; i++) {
1308 e = t->entries[i];
1309 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1310 if (!slash1) {
1311 if (!S_ISDIR(mode)
1312 && e->versions[1].mode == mode
1313 && !hashcmp(e->versions[1].sha1, sha1))
1314 return 0;
1315 e->versions[1].mode = mode;
1316 hashcpy(e->versions[1].sha1, sha1);
1317 if (e->tree)
1318 release_tree_content_recursive(e->tree);
1319 e->tree = subtree;
1320 hashclr(root->versions[1].sha1);
1321 return 1;
1322 }
1323 if (!S_ISDIR(e->versions[1].mode)) {
1324 e->tree = new_tree_content(8);
1325 e->versions[1].mode = S_IFDIR;
1326 }
1327 if (!e->tree)
1328 load_tree(e);
1329 if (tree_content_set(e, slash1 + 1, sha1, mode, subtree)) {
1330 hashclr(root->versions[1].sha1);
1331 return 1;
1332 }
1333 return 0;
1334 }
1335 }
1336
1337 if (t->entry_count == t->entry_capacity)
1338 root->tree = t = grow_tree_content(t, t->entry_count);
1339 e = new_tree_entry();
1340 e->name = to_atom(p, n);
1341 e->versions[0].mode = 0;
1342 hashclr(e->versions[0].sha1);
1343 t->entries[t->entry_count++] = e;
1344 if (slash1) {
1345 e->tree = new_tree_content(8);
1346 e->versions[1].mode = S_IFDIR;
1347 tree_content_set(e, slash1 + 1, sha1, mode, subtree);
1348 } else {
1349 e->tree = subtree;
1350 e->versions[1].mode = mode;
1351 hashcpy(e->versions[1].sha1, sha1);
1352 }
1353 hashclr(root->versions[1].sha1);
1354 return 1;
1355 }
1356
1357 static int tree_content_remove(
1358 struct tree_entry *root,
1359 const char *p,
1360 struct tree_entry *backup_leaf)
1361 {
1362 struct tree_content *t = root->tree;
1363 const char *slash1;
1364 unsigned int i, n;
1365 struct tree_entry *e;
1366
1367 slash1 = strchr(p, '/');
1368 if (slash1)
1369 n = slash1 - p;
1370 else
1371 n = strlen(p);
1372
1373 for (i = 0; i < t->entry_count; i++) {
1374 e = t->entries[i];
1375 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1376 if (!slash1 || !S_ISDIR(e->versions[1].mode))
1377 goto del_entry;
1378 if (!e->tree)
1379 load_tree(e);
1380 if (tree_content_remove(e, slash1 + 1, backup_leaf)) {
1381 for (n = 0; n < e->tree->entry_count; n++) {
1382 if (e->tree->entries[n]->versions[1].mode) {
1383 hashclr(root->versions[1].sha1);
1384 return 1;
1385 }
1386 }
1387 backup_leaf = NULL;
1388 goto del_entry;
1389 }
1390 return 0;
1391 }
1392 }
1393 return 0;
1394
1395 del_entry:
1396 if (backup_leaf)
1397 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1398 else if (e->tree)
1399 release_tree_content_recursive(e->tree);
1400 e->tree = NULL;
1401 e->versions[1].mode = 0;
1402 hashclr(e->versions[1].sha1);
1403 hashclr(root->versions[1].sha1);
1404 return 1;
1405 }
1406
1407 static int tree_content_get(
1408 struct tree_entry *root,
1409 const char *p,
1410 struct tree_entry *leaf)
1411 {
1412 struct tree_content *t = root->tree;
1413 const char *slash1;
1414 unsigned int i, n;
1415 struct tree_entry *e;
1416
1417 slash1 = strchr(p, '/');
1418 if (slash1)
1419 n = slash1 - p;
1420 else
1421 n = strlen(p);
1422
1423 for (i = 0; i < t->entry_count; i++) {
1424 e = t->entries[i];
1425 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1426 if (!slash1) {
1427 memcpy(leaf, e, sizeof(*leaf));
1428 if (e->tree && is_null_sha1(e->versions[1].sha1))
1429 leaf->tree = dup_tree_content(e->tree);
1430 else
1431 leaf->tree = NULL;
1432 return 1;
1433 }
1434 if (!S_ISDIR(e->versions[1].mode))
1435 return 0;
1436 if (!e->tree)
1437 load_tree(e);
1438 return tree_content_get(e, slash1 + 1, leaf);
1439 }
1440 }
1441 return 0;
1442 }
1443
1444 static int update_branch(struct branch *b)
1445 {
1446 static const char *msg = "fast-import";
1447 struct ref_lock *lock;
1448 unsigned char old_sha1[20];
1449
1450 if (read_ref(b->name, old_sha1))
1451 hashclr(old_sha1);
1452 lock = lock_any_ref_for_update(b->name, old_sha1, 0);
1453 if (!lock)
1454 return error("Unable to lock %s", b->name);
1455 if (!force_update && !is_null_sha1(old_sha1)) {
1456 struct commit *old_cmit, *new_cmit;
1457
1458 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1459 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1460 if (!old_cmit || !new_cmit) {
1461 unlock_ref(lock);
1462 return error("Branch %s is missing commits.", b->name);
1463 }
1464
1465 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1466 unlock_ref(lock);
1467 warning("Not updating %s"
1468 " (new tip %s does not contain %s)",
1469 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1470 return -1;
1471 }
1472 }
1473 if (write_ref_sha1(lock, b->sha1, msg) < 0)
1474 return error("Unable to update %s", b->name);
1475 return 0;
1476 }
1477
1478 static void dump_branches(void)
1479 {
1480 unsigned int i;
1481 struct branch *b;
1482
1483 for (i = 0; i < branch_table_sz; i++) {
1484 for (b = branch_table[i]; b; b = b->table_next_branch)
1485 failure |= update_branch(b);
1486 }
1487 }
1488
1489 static void dump_tags(void)
1490 {
1491 static const char *msg = "fast-import";
1492 struct tag *t;
1493 struct ref_lock *lock;
1494 char ref_name[PATH_MAX];
1495
1496 for (t = first_tag; t; t = t->next_tag) {
1497 sprintf(ref_name, "tags/%s", t->name);
1498 lock = lock_ref_sha1(ref_name, NULL);
1499 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1500 failure |= error("Unable to update %s", ref_name);
1501 }
1502 }
1503
1504 static void dump_marks_helper(FILE *f,
1505 uintmax_t base,
1506 struct mark_set *m)
1507 {
1508 uintmax_t k;
1509 if (m->shift) {
1510 for (k = 0; k < 1024; k++) {
1511 if (m->data.sets[k])
1512 dump_marks_helper(f, (base + k) << m->shift,
1513 m->data.sets[k]);
1514 }
1515 } else {
1516 for (k = 0; k < 1024; k++) {
1517 if (m->data.marked[k])
1518 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1519 sha1_to_hex(m->data.marked[k]->sha1));
1520 }
1521 }
1522 }
1523
1524 static void dump_marks(void)
1525 {
1526 static struct lock_file mark_lock;
1527 int mark_fd;
1528 FILE *f;
1529
1530 if (!mark_file)
1531 return;
1532
1533 mark_fd = hold_lock_file_for_update(&mark_lock, mark_file, 0);
1534 if (mark_fd < 0) {
1535 failure |= error("Unable to write marks file %s: %s",
1536 mark_file, strerror(errno));
1537 return;
1538 }
1539
1540 f = fdopen(mark_fd, "w");
1541 if (!f) {
1542 rollback_lock_file(&mark_lock);
1543 failure |= error("Unable to write marks file %s: %s",
1544 mark_file, strerror(errno));
1545 return;
1546 }
1547
1548 dump_marks_helper(f, 0, marks);
1549 fclose(f);
1550 if (commit_lock_file(&mark_lock))
1551 failure |= error("Unable to write marks file %s: %s",
1552 mark_file, strerror(errno));
1553 }
1554
1555 static int read_next_command(void)
1556 {
1557 static int stdin_eof = 0;
1558
1559 if (stdin_eof) {
1560 unread_command_buf = 0;
1561 return EOF;
1562 }
1563
1564 do {
1565 if (unread_command_buf) {
1566 unread_command_buf = 0;
1567 } else {
1568 struct recent_command *rc;
1569
1570 strbuf_detach(&command_buf, NULL);
1571 stdin_eof = strbuf_getline(&command_buf, stdin, '\n');
1572 if (stdin_eof)
1573 return EOF;
1574
1575 rc = rc_free;
1576 if (rc)
1577 rc_free = rc->next;
1578 else {
1579 rc = cmd_hist.next;
1580 cmd_hist.next = rc->next;
1581 cmd_hist.next->prev = &cmd_hist;
1582 free(rc->buf);
1583 }
1584
1585 rc->buf = command_buf.buf;
1586 rc->prev = cmd_tail;
1587 rc->next = cmd_hist.prev;
1588 rc->prev->next = rc;
1589 cmd_tail = rc;
1590 }
1591 } while (command_buf.buf[0] == '#');
1592
1593 return 0;
1594 }
1595
1596 static void skip_optional_lf(void)
1597 {
1598 int term_char = fgetc(stdin);
1599 if (term_char != '\n' && term_char != EOF)
1600 ungetc(term_char, stdin);
1601 }
1602
1603 static void cmd_mark(void)
1604 {
1605 if (!prefixcmp(command_buf.buf, "mark :")) {
1606 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1607 read_next_command();
1608 }
1609 else
1610 next_mark = 0;
1611 }
1612
1613 static void cmd_data(struct strbuf *sb)
1614 {
1615 strbuf_reset(sb);
1616
1617 if (prefixcmp(command_buf.buf, "data "))
1618 die("Expected 'data n' command, found: %s", command_buf.buf);
1619
1620 if (!prefixcmp(command_buf.buf + 5, "<<")) {
1621 char *term = xstrdup(command_buf.buf + 5 + 2);
1622 size_t term_len = command_buf.len - 5 - 2;
1623
1624 strbuf_detach(&command_buf, NULL);
1625 for (;;) {
1626 if (strbuf_getline(&command_buf, stdin, '\n') == EOF)
1627 die("EOF in data (terminator '%s' not found)", term);
1628 if (term_len == command_buf.len
1629 && !strcmp(term, command_buf.buf))
1630 break;
1631 strbuf_addbuf(sb, &command_buf);
1632 strbuf_addch(sb, '\n');
1633 }
1634 free(term);
1635 }
1636 else {
1637 size_t n = 0, length;
1638
1639 length = strtoul(command_buf.buf + 5, NULL, 10);
1640
1641 while (n < length) {
1642 size_t s = strbuf_fread(sb, length - n, stdin);
1643 if (!s && feof(stdin))
1644 die("EOF in data (%lu bytes remaining)",
1645 (unsigned long)(length - n));
1646 n += s;
1647 }
1648 }
1649
1650 skip_optional_lf();
1651 }
1652
1653 static int validate_raw_date(const char *src, char *result, int maxlen)
1654 {
1655 const char *orig_src = src;
1656 char *endp, sign;
1657
1658 strtoul(src, &endp, 10);
1659 if (endp == src || *endp != ' ')
1660 return -1;
1661
1662 src = endp + 1;
1663 if (*src != '-' && *src != '+')
1664 return -1;
1665 sign = *src;
1666
1667 strtoul(src + 1, &endp, 10);
1668 if (endp == src || *endp || (endp - orig_src) >= maxlen)
1669 return -1;
1670
1671 strcpy(result, orig_src);
1672 return 0;
1673 }
1674
1675 static char *parse_ident(const char *buf)
1676 {
1677 const char *gt;
1678 size_t name_len;
1679 char *ident;
1680
1681 gt = strrchr(buf, '>');
1682 if (!gt)
1683 die("Missing > in ident string: %s", buf);
1684 gt++;
1685 if (*gt != ' ')
1686 die("Missing space after > in ident string: %s", buf);
1687 gt++;
1688 name_len = gt - buf;
1689 ident = xmalloc(name_len + 24);
1690 strncpy(ident, buf, name_len);
1691
1692 switch (whenspec) {
1693 case WHENSPEC_RAW:
1694 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1695 die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1696 break;
1697 case WHENSPEC_RFC2822:
1698 if (parse_date(gt, ident + name_len, 24) < 0)
1699 die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1700 break;
1701 case WHENSPEC_NOW:
1702 if (strcmp("now", gt))
1703 die("Date in ident must be 'now': %s", buf);
1704 datestamp(ident + name_len, 24);
1705 break;
1706 }
1707
1708 return ident;
1709 }
1710
1711 static void cmd_new_blob(void)
1712 {
1713 static struct strbuf buf = STRBUF_INIT;
1714
1715 read_next_command();
1716 cmd_mark();
1717 cmd_data(&buf);
1718 store_object(OBJ_BLOB, &buf, &last_blob, NULL, next_mark);
1719 }
1720
1721 static void unload_one_branch(void)
1722 {
1723 while (cur_active_branches
1724 && cur_active_branches >= max_active_branches) {
1725 uintmax_t min_commit = ULONG_MAX;
1726 struct branch *e, *l = NULL, *p = NULL;
1727
1728 for (e = active_branches; e; e = e->active_next_branch) {
1729 if (e->last_commit < min_commit) {
1730 p = l;
1731 min_commit = e->last_commit;
1732 }
1733 l = e;
1734 }
1735
1736 if (p) {
1737 e = p->active_next_branch;
1738 p->active_next_branch = e->active_next_branch;
1739 } else {
1740 e = active_branches;
1741 active_branches = e->active_next_branch;
1742 }
1743 e->active = 0;
1744 e->active_next_branch = NULL;
1745 if (e->branch_tree.tree) {
1746 release_tree_content_recursive(e->branch_tree.tree);
1747 e->branch_tree.tree = NULL;
1748 }
1749 cur_active_branches--;
1750 }
1751 }
1752
1753 static void load_branch(struct branch *b)
1754 {
1755 load_tree(&b->branch_tree);
1756 if (!b->active) {
1757 b->active = 1;
1758 b->active_next_branch = active_branches;
1759 active_branches = b;
1760 cur_active_branches++;
1761 branch_load_count++;
1762 }
1763 }
1764
1765 static void file_change_m(struct branch *b)
1766 {
1767 const char *p = command_buf.buf + 2;
1768 static struct strbuf uq = STRBUF_INIT;
1769 const char *endp;
1770 struct object_entry *oe = oe;
1771 unsigned char sha1[20];
1772 uint16_t mode, inline_data = 0;
1773
1774 p = get_mode(p, &mode);
1775 if (!p)
1776 die("Corrupt mode: %s", command_buf.buf);
1777 switch (mode) {
1778 case S_IFREG | 0644:
1779 case S_IFREG | 0755:
1780 case S_IFLNK:
1781 case 0644:
1782 case 0755:
1783 /* ok */
1784 break;
1785 default:
1786 die("Corrupt mode: %s", command_buf.buf);
1787 }
1788
1789 if (*p == ':') {
1790 char *x;
1791 oe = find_mark(strtoumax(p + 1, &x, 10));
1792 hashcpy(sha1, oe->sha1);
1793 p = x;
1794 } else if (!prefixcmp(p, "inline")) {
1795 inline_data = 1;
1796 p += 6;
1797 } else {
1798 if (get_sha1_hex(p, sha1))
1799 die("Invalid SHA1: %s", command_buf.buf);
1800 oe = find_object(sha1);
1801 p += 40;
1802 }
1803 if (*p++ != ' ')
1804 die("Missing space after SHA1: %s", command_buf.buf);
1805
1806 strbuf_reset(&uq);
1807 if (!unquote_c_style(&uq, p, &endp)) {
1808 if (*endp)
1809 die("Garbage after path in: %s", command_buf.buf);
1810 p = uq.buf;
1811 }
1812
1813 if (inline_data) {
1814 static struct strbuf buf = STRBUF_INIT;
1815
1816 if (p != uq.buf) {
1817 strbuf_addstr(&uq, p);
1818 p = uq.buf;
1819 }
1820 read_next_command();
1821 cmd_data(&buf);
1822 store_object(OBJ_BLOB, &buf, &last_blob, sha1, 0);
1823 } else if (oe) {
1824 if (oe->type != OBJ_BLOB)
1825 die("Not a blob (actually a %s): %s",
1826 typename(oe->type), command_buf.buf);
1827 } else {
1828 enum object_type type = sha1_object_info(sha1, NULL);
1829 if (type < 0)
1830 die("Blob not found: %s", command_buf.buf);
1831 if (type != OBJ_BLOB)
1832 die("Not a blob (actually a %s): %s",
1833 typename(type), command_buf.buf);
1834 }
1835
1836 tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode, NULL);
1837 }
1838
1839 static void file_change_d(struct branch *b)
1840 {
1841 const char *p = command_buf.buf + 2;
1842 static struct strbuf uq = STRBUF_INIT;
1843 const char *endp;
1844
1845 strbuf_reset(&uq);
1846 if (!unquote_c_style(&uq, p, &endp)) {
1847 if (*endp)
1848 die("Garbage after path in: %s", command_buf.buf);
1849 p = uq.buf;
1850 }
1851 tree_content_remove(&b->branch_tree, p, NULL);
1852 }
1853
1854 static void file_change_cr(struct branch *b, int rename)
1855 {
1856 const char *s, *d;
1857 static struct strbuf s_uq = STRBUF_INIT;
1858 static struct strbuf d_uq = STRBUF_INIT;
1859 const char *endp;
1860 struct tree_entry leaf;
1861
1862 s = command_buf.buf + 2;
1863 strbuf_reset(&s_uq);
1864 if (!unquote_c_style(&s_uq, s, &endp)) {
1865 if (*endp != ' ')
1866 die("Missing space after source: %s", command_buf.buf);
1867 } else {
1868 endp = strchr(s, ' ');
1869 if (!endp)
1870 die("Missing space after source: %s", command_buf.buf);
1871 strbuf_add(&s_uq, s, endp - s);
1872 }
1873 s = s_uq.buf;
1874
1875 endp++;
1876 if (!*endp)
1877 die("Missing dest: %s", command_buf.buf);
1878
1879 d = endp;
1880 strbuf_reset(&d_uq);
1881 if (!unquote_c_style(&d_uq, d, &endp)) {
1882 if (*endp)
1883 die("Garbage after dest in: %s", command_buf.buf);
1884 d = d_uq.buf;
1885 }
1886
1887 memset(&leaf, 0, sizeof(leaf));
1888 if (rename)
1889 tree_content_remove(&b->branch_tree, s, &leaf);
1890 else
1891 tree_content_get(&b->branch_tree, s, &leaf);
1892 if (!leaf.versions[1].mode)
1893 die("Path %s not in branch", s);
1894 tree_content_set(&b->branch_tree, d,
1895 leaf.versions[1].sha1,
1896 leaf.versions[1].mode,
1897 leaf.tree);
1898 }
1899
1900 static void file_change_deleteall(struct branch *b)
1901 {
1902 release_tree_content_recursive(b->branch_tree.tree);
1903 hashclr(b->branch_tree.versions[0].sha1);
1904 hashclr(b->branch_tree.versions[1].sha1);
1905 load_tree(&b->branch_tree);
1906 }
1907
1908 static void cmd_from_commit(struct branch *b, char *buf, unsigned long size)
1909 {
1910 if (!buf || size < 46)
1911 die("Not a valid commit: %s", sha1_to_hex(b->sha1));
1912 if (memcmp("tree ", buf, 5)
1913 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1914 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1915 hashcpy(b->branch_tree.versions[0].sha1,
1916 b->branch_tree.versions[1].sha1);
1917 }
1918
1919 static void cmd_from_existing(struct branch *b)
1920 {
1921 if (is_null_sha1(b->sha1)) {
1922 hashclr(b->branch_tree.versions[0].sha1);
1923 hashclr(b->branch_tree.versions[1].sha1);
1924 } else {
1925 unsigned long size;
1926 char *buf;
1927
1928 buf = read_object_with_reference(b->sha1,
1929 commit_type, &size, b->sha1);
1930 cmd_from_commit(b, buf, size);
1931 free(buf);
1932 }
1933 }
1934
1935 static int cmd_from(struct branch *b)
1936 {
1937 const char *from;
1938 struct branch *s;
1939
1940 if (prefixcmp(command_buf.buf, "from "))
1941 return 0;
1942
1943 if (b->branch_tree.tree) {
1944 release_tree_content_recursive(b->branch_tree.tree);
1945 b->branch_tree.tree = NULL;
1946 }
1947
1948 from = strchr(command_buf.buf, ' ') + 1;
1949 s = lookup_branch(from);
1950 if (b == s)
1951 die("Can't create a branch from itself: %s", b->name);
1952 else if (s) {
1953 unsigned char *t = s->branch_tree.versions[1].sha1;
1954 hashcpy(b->sha1, s->sha1);
1955 hashcpy(b->branch_tree.versions[0].sha1, t);
1956 hashcpy(b->branch_tree.versions[1].sha1, t);
1957 } else if (*from == ':') {
1958 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1959 struct object_entry *oe = find_mark(idnum);
1960 if (oe->type != OBJ_COMMIT)
1961 die("Mark :%" PRIuMAX " not a commit", idnum);
1962 hashcpy(b->sha1, oe->sha1);
1963 if (oe->pack_id != MAX_PACK_ID) {
1964 unsigned long size;
1965 char *buf = gfi_unpack_entry(oe, &size);
1966 cmd_from_commit(b, buf, size);
1967 free(buf);
1968 } else
1969 cmd_from_existing(b);
1970 } else if (!get_sha1(from, b->sha1))
1971 cmd_from_existing(b);
1972 else
1973 die("Invalid ref name or SHA1 expression: %s", from);
1974
1975 read_next_command();
1976 return 1;
1977 }
1978
1979 static struct hash_list *cmd_merge(unsigned int *count)
1980 {
1981 struct hash_list *list = NULL, *n, *e = e;
1982 const char *from;
1983 struct branch *s;
1984
1985 *count = 0;
1986 while (!prefixcmp(command_buf.buf, "merge ")) {
1987 from = strchr(command_buf.buf, ' ') + 1;
1988 n = xmalloc(sizeof(*n));
1989 s = lookup_branch(from);
1990 if (s)
1991 hashcpy(n->sha1, s->sha1);
1992 else if (*from == ':') {
1993 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1994 struct object_entry *oe = find_mark(idnum);
1995 if (oe->type != OBJ_COMMIT)
1996 die("Mark :%" PRIuMAX " not a commit", idnum);
1997 hashcpy(n->sha1, oe->sha1);
1998 } else if (!get_sha1(from, n->sha1)) {
1999 unsigned long size;
2000 char *buf = read_object_with_reference(n->sha1,
2001 commit_type, &size, n->sha1);
2002 if (!buf || size < 46)
2003 die("Not a valid commit: %s", from);
2004 free(buf);
2005 } else
2006 die("Invalid ref name or SHA1 expression: %s", from);
2007
2008 n->next = NULL;
2009 if (list)
2010 e->next = n;
2011 else
2012 list = n;
2013 e = n;
2014 (*count)++;
2015 read_next_command();
2016 }
2017 return list;
2018 }
2019
2020 static void cmd_new_commit(void)
2021 {
2022 static struct strbuf msg = STRBUF_INIT;
2023 struct branch *b;
2024 char *sp;
2025 char *author = NULL;
2026 char *committer = NULL;
2027 struct hash_list *merge_list = NULL;
2028 unsigned int merge_count;
2029
2030 /* Obtain the branch name from the rest of our command */
2031 sp = strchr(command_buf.buf, ' ') + 1;
2032 b = lookup_branch(sp);
2033 if (!b)
2034 b = new_branch(sp);
2035
2036 read_next_command();
2037 cmd_mark();
2038 if (!prefixcmp(command_buf.buf, "author ")) {
2039 author = parse_ident(command_buf.buf + 7);
2040 read_next_command();
2041 }
2042 if (!prefixcmp(command_buf.buf, "committer ")) {
2043 committer = parse_ident(command_buf.buf + 10);
2044 read_next_command();
2045 }
2046 if (!committer)
2047 die("Expected committer but didn't get one");
2048 cmd_data(&msg);
2049 read_next_command();
2050 cmd_from(b);
2051 merge_list = cmd_merge(&merge_count);
2052
2053 /* ensure the branch is active/loaded */
2054 if (!b->branch_tree.tree || !max_active_branches) {
2055 unload_one_branch();
2056 load_branch(b);
2057 }
2058
2059 /* file_change* */
2060 while (command_buf.len > 0) {
2061 if (!prefixcmp(command_buf.buf, "M "))
2062 file_change_m(b);
2063 else if (!prefixcmp(command_buf.buf, "D "))
2064 file_change_d(b);
2065 else if (!prefixcmp(command_buf.buf, "R "))
2066 file_change_cr(b, 1);
2067 else if (!prefixcmp(command_buf.buf, "C "))
2068 file_change_cr(b, 0);
2069 else if (!strcmp("deleteall", command_buf.buf))
2070 file_change_deleteall(b);
2071 else {
2072 unread_command_buf = 1;
2073 break;
2074 }
2075 if (read_next_command() == EOF)
2076 break;
2077 }
2078
2079 /* build the tree and the commit */
2080 store_tree(&b->branch_tree);
2081 hashcpy(b->branch_tree.versions[0].sha1,
2082 b->branch_tree.versions[1].sha1);
2083
2084 strbuf_reset(&new_data);
2085 strbuf_addf(&new_data, "tree %s\n",
2086 sha1_to_hex(b->branch_tree.versions[1].sha1));
2087 if (!is_null_sha1(b->sha1))
2088 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(b->sha1));
2089 while (merge_list) {
2090 struct hash_list *next = merge_list->next;
2091 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(merge_list->sha1));
2092 free(merge_list);
2093 merge_list = next;
2094 }
2095 strbuf_addf(&new_data,
2096 "author %s\n"
2097 "committer %s\n"
2098 "\n",
2099 author ? author : committer, committer);
2100 strbuf_addbuf(&new_data, &msg);
2101 free(author);
2102 free(committer);
2103
2104 if (!store_object(OBJ_COMMIT, &new_data, NULL, b->sha1, next_mark))
2105 b->pack_id = pack_id;
2106 b->last_commit = object_count_by_type[OBJ_COMMIT];
2107 }
2108
2109 static void cmd_new_tag(void)
2110 {
2111 static struct strbuf msg = STRBUF_INIT;
2112 char *sp;
2113 const char *from;
2114 char *tagger;
2115 struct branch *s;
2116 struct tag *t;
2117 uintmax_t from_mark = 0;
2118 unsigned char sha1[20];
2119
2120 /* Obtain the new tag name from the rest of our command */
2121 sp = strchr(command_buf.buf, ' ') + 1;
2122 t = pool_alloc(sizeof(struct tag));
2123 t->next_tag = NULL;
2124 t->name = pool_strdup(sp);
2125 if (last_tag)
2126 last_tag->next_tag = t;
2127 else
2128 first_tag = t;
2129 last_tag = t;
2130 read_next_command();
2131
2132 /* from ... */
2133 if (prefixcmp(command_buf.buf, "from "))
2134 die("Expected from command, got %s", command_buf.buf);
2135 from = strchr(command_buf.buf, ' ') + 1;
2136 s = lookup_branch(from);
2137 if (s) {
2138 hashcpy(sha1, s->sha1);
2139 } else if (*from == ':') {
2140 struct object_entry *oe;
2141 from_mark = strtoumax(from + 1, NULL, 10);
2142 oe = find_mark(from_mark);
2143 if (oe->type != OBJ_COMMIT)
2144 die("Mark :%" PRIuMAX " not a commit", from_mark);
2145 hashcpy(sha1, oe->sha1);
2146 } else if (!get_sha1(from, sha1)) {
2147 unsigned long size;
2148 char *buf;
2149
2150 buf = read_object_with_reference(sha1,
2151 commit_type, &size, sha1);
2152 if (!buf || size < 46)
2153 die("Not a valid commit: %s", from);
2154 free(buf);
2155 } else
2156 die("Invalid ref name or SHA1 expression: %s", from);
2157 read_next_command();
2158
2159 /* tagger ... */
2160 if (prefixcmp(command_buf.buf, "tagger "))
2161 die("Expected tagger command, got %s", command_buf.buf);
2162 tagger = parse_ident(command_buf.buf + 7);
2163
2164 /* tag payload/message */
2165 read_next_command();
2166 cmd_data(&msg);
2167
2168 /* build the tag object */
2169 strbuf_reset(&new_data);
2170 strbuf_addf(&new_data,
2171 "object %s\n"
2172 "type %s\n"
2173 "tag %s\n"
2174 "tagger %s\n"
2175 "\n",
2176 sha1_to_hex(sha1), commit_type, t->name, tagger);
2177 strbuf_addbuf(&new_data, &msg);
2178 free(tagger);
2179
2180 if (store_object(OBJ_TAG, &new_data, NULL, t->sha1, 0))
2181 t->pack_id = MAX_PACK_ID;
2182 else
2183 t->pack_id = pack_id;
2184 }
2185
2186 static void cmd_reset_branch(void)
2187 {
2188 struct branch *b;
2189 char *sp;
2190
2191 /* Obtain the branch name from the rest of our command */
2192 sp = strchr(command_buf.buf, ' ') + 1;
2193 b = lookup_branch(sp);
2194 if (b) {
2195 hashclr(b->sha1);
2196 hashclr(b->branch_tree.versions[0].sha1);
2197 hashclr(b->branch_tree.versions[1].sha1);
2198 if (b->branch_tree.tree) {
2199 release_tree_content_recursive(b->branch_tree.tree);
2200 b->branch_tree.tree = NULL;
2201 }
2202 }
2203 else
2204 b = new_branch(sp);
2205 read_next_command();
2206 if (!cmd_from(b) && command_buf.len > 0)
2207 unread_command_buf = 1;
2208 }
2209
2210 static void cmd_checkpoint(void)
2211 {
2212 if (object_count) {
2213 cycle_packfile();
2214 dump_branches();
2215 dump_tags();
2216 dump_marks();
2217 }
2218 skip_optional_lf();
2219 }
2220
2221 static void cmd_progress(void)
2222 {
2223 fwrite(command_buf.buf, 1, command_buf.len, stdout);
2224 fputc('\n', stdout);
2225 fflush(stdout);
2226 skip_optional_lf();
2227 }
2228
2229 static void import_marks(const char *input_file)
2230 {
2231 char line[512];
2232 FILE *f = fopen(input_file, "r");
2233 if (!f)
2234 die("cannot read %s: %s", input_file, strerror(errno));
2235 while (fgets(line, sizeof(line), f)) {
2236 uintmax_t mark;
2237 char *end;
2238 unsigned char sha1[20];
2239 struct object_entry *e;
2240
2241 end = strchr(line, '\n');
2242 if (line[0] != ':' || !end)
2243 die("corrupt mark line: %s", line);
2244 *end = 0;
2245 mark = strtoumax(line + 1, &end, 10);
2246 if (!mark || end == line + 1
2247 || *end != ' ' || get_sha1(end + 1, sha1))
2248 die("corrupt mark line: %s", line);
2249 e = find_object(sha1);
2250 if (!e) {
2251 enum object_type type = sha1_object_info(sha1, NULL);
2252 if (type < 0)
2253 die("object not found: %s", sha1_to_hex(sha1));
2254 e = insert_object(sha1);
2255 e->type = type;
2256 e->pack_id = MAX_PACK_ID;
2257 e->offset = 1; /* just not zero! */
2258 }
2259 insert_mark(mark, e);
2260 }
2261 fclose(f);
2262 }
2263
2264 static const char fast_import_usage[] =
2265 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
2266
2267 int main(int argc, const char **argv)
2268 {
2269 unsigned int i, show_stats = 1;
2270
2271 git_config(git_default_config);
2272 alloc_objects(object_entry_alloc);
2273 strbuf_init(&command_buf, 0);
2274 atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2275 branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2276 avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2277 marks = pool_calloc(1, sizeof(struct mark_set));
2278
2279 for (i = 1; i < argc; i++) {
2280 const char *a = argv[i];
2281
2282 if (*a != '-' || !strcmp(a, "--"))
2283 break;
2284 else if (!prefixcmp(a, "--date-format=")) {
2285 const char *fmt = a + 14;
2286 if (!strcmp(fmt, "raw"))
2287 whenspec = WHENSPEC_RAW;
2288 else if (!strcmp(fmt, "rfc2822"))
2289 whenspec = WHENSPEC_RFC2822;
2290 else if (!strcmp(fmt, "now"))
2291 whenspec = WHENSPEC_NOW;
2292 else
2293 die("unknown --date-format argument %s", fmt);
2294 }
2295 else if (!prefixcmp(a, "--max-pack-size="))
2296 max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2297 else if (!prefixcmp(a, "--depth=")) {
2298 max_depth = strtoul(a + 8, NULL, 0);
2299 if (max_depth > MAX_DEPTH)
2300 die("--depth cannot exceed %u", MAX_DEPTH);
2301 }
2302 else if (!prefixcmp(a, "--active-branches="))
2303 max_active_branches = strtoul(a + 18, NULL, 0);
2304 else if (!prefixcmp(a, "--import-marks="))
2305 import_marks(a + 15);
2306 else if (!prefixcmp(a, "--export-marks="))
2307 mark_file = a + 15;
2308 else if (!prefixcmp(a, "--export-pack-edges=")) {
2309 if (pack_edges)
2310 fclose(pack_edges);
2311 pack_edges = fopen(a + 20, "a");
2312 if (!pack_edges)
2313 die("Cannot open %s: %s", a + 20, strerror(errno));
2314 } else if (!strcmp(a, "--force"))
2315 force_update = 1;
2316 else if (!strcmp(a, "--quiet"))
2317 show_stats = 0;
2318 else if (!strcmp(a, "--stats"))
2319 show_stats = 1;
2320 else
2321 die("unknown option %s", a);
2322 }
2323 if (i != argc)
2324 usage(fast_import_usage);
2325
2326 rc_free = pool_alloc(cmd_save * sizeof(*rc_free));
2327 for (i = 0; i < (cmd_save - 1); i++)
2328 rc_free[i].next = &rc_free[i + 1];
2329 rc_free[cmd_save - 1].next = NULL;
2330
2331 prepare_packed_git();
2332 start_packfile();
2333 set_die_routine(die_nicely);
2334 while (read_next_command() != EOF) {
2335 if (!strcmp("blob", command_buf.buf))
2336 cmd_new_blob();
2337 else if (!prefixcmp(command_buf.buf, "commit "))
2338 cmd_new_commit();
2339 else if (!prefixcmp(command_buf.buf, "tag "))
2340 cmd_new_tag();
2341 else if (!prefixcmp(command_buf.buf, "reset "))
2342 cmd_reset_branch();
2343 else if (!strcmp("checkpoint", command_buf.buf))
2344 cmd_checkpoint();
2345 else if (!prefixcmp(command_buf.buf, "progress "))
2346 cmd_progress();
2347 else
2348 die("Unsupported command: %s", command_buf.buf);
2349 }
2350 end_packfile();
2351
2352 dump_branches();
2353 dump_tags();
2354 unkeep_all_packs();
2355 dump_marks();
2356
2357 if (pack_edges)
2358 fclose(pack_edges);
2359
2360 if (show_stats) {
2361 uintmax_t total_count = 0, duplicate_count = 0;
2362 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2363 total_count += object_count_by_type[i];
2364 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2365 duplicate_count += duplicate_count_by_type[i];
2366
2367 fprintf(stderr, "%s statistics:\n", argv[0]);
2368 fprintf(stderr, "---------------------------------------------------------------------\n");
2369 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2370 fprintf(stderr, "Total objects: %10" PRIuMAX " (%10" PRIuMAX " duplicates )\n", total_count, duplicate_count);
2371 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]);
2372 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]);
2373 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]);
2374 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]);
2375 fprintf(stderr, "Total branches: %10lu (%10lu loads )\n", branch_count, branch_load_count);
2376 fprintf(stderr, " marks: %10" PRIuMAX " (%10" PRIuMAX " unique )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2377 fprintf(stderr, " atoms: %10u\n", atom_cnt);
2378 fprintf(stderr, "Memory total: %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2379 fprintf(stderr, " pools: %10lu KiB\n", (unsigned long)(total_allocd/1024));
2380 fprintf(stderr, " objects: %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2381 fprintf(stderr, "---------------------------------------------------------------------\n");
2382 pack_report();
2383 fprintf(stderr, "---------------------------------------------------------------------\n");
2384 fprintf(stderr, "\n");
2385 }
2386
2387 return failure ? 1 : 0;
2388 }
git with Jakub Narębski modifications