2 Format of STDIN stream:
14 new_blob ::= 'blob' lf
17 file_content ::= data;
19 new_commit ::= 'commit' sp ref_str lf
21 ('author' sp name '<' email '>' when lf)?
22 'committer' sp name '<' email '>' when lf
24 ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
25 ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
30 file_change ::= file_clr
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
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
50 reset_branch ::= 'reset' sp ref_str lf
51 ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
54 checkpoint ::= 'checkpoint' lf
57 progress ::= 'progress' sp not_lf* lf
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.
66 mark ::= 'mark' sp idnum lf;
67 data ::= (delimited_data | exact_data)
70 # note: delim may be any string but must not contain lf.
71 # data_line may contain any data but must not be exactly
73 delimited_data ::= 'data' sp '<<' delim lf
77 # note: declen indicates the length of binary_data in bytes.
78 # declen does not include the lf preceeding the binary data.
80 exact_data ::= 'data' sp declen lf
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
91 sha1exp_str ::= sha1exp;
93 path_str ::= path | '"' quoted(path) '"' ;
94 mode ::= '100644' | '644'
99 declen ::= # unsigned 32 bit value, ascii base10 notation;
100 bigint ::= # unsigned integer value, ascii base10 notation;
101 binary_data ::= # file content, not interpreted;
103 when ::= raw_when | rfc2822_when;
104 raw_when ::= ts sp tz;
105 rfc2822_when ::= # Valid RFC 2822 date and time;
107 sp ::= # ASCII space character;
108 lf ::= # ASCII newline (LF) character;
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.
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;
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.
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;
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.
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
138 comment ::= '#' not_lf* lf;
139 not_lf ::= # Any byte that is not ASCII newline (LF);
151 #include "csum-file.h"
155 #define PACK_ID_BITS 16
156 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
160 struct object_entry
*next
;
162 unsigned type
: TYPE_BITS
;
163 unsigned pack_id
: PACK_ID_BITS
;
164 unsigned char sha1
[20];
167 struct object_entry_pool
169 struct object_entry_pool
*next_pool
;
170 struct object_entry
*next_free
;
171 struct object_entry
*end
;
172 struct object_entry entries
[FLEX_ARRAY
]; /* more */
178 struct object_entry
*marked
[1024];
179 struct mark_set
*sets
[1024];
195 struct mem_pool
*next_pool
;
198 char space
[FLEX_ARRAY
]; /* more */
203 struct atom_str
*next_atom
;
204 unsigned short str_len
;
205 char str_dat
[FLEX_ARRAY
]; /* more */
211 struct tree_content
*tree
;
212 struct atom_str
* name
;
216 unsigned char sha1
[20];
222 unsigned int entry_capacity
; /* must match avail_tree_content */
223 unsigned int entry_count
;
224 unsigned int delta_depth
;
225 struct tree_entry
*entries
[FLEX_ARRAY
]; /* more */
228 struct avail_tree_content
230 unsigned int entry_capacity
; /* must match tree_content */
231 struct avail_tree_content
*next_avail
;
236 struct branch
*table_next_branch
;
237 struct branch
*active_next_branch
;
239 struct tree_entry branch_tree
;
240 uintmax_t last_commit
;
242 unsigned pack_id
: PACK_ID_BITS
;
243 unsigned char sha1
[20];
248 struct tag
*next_tag
;
250 unsigned int pack_id
;
251 unsigned char sha1
[20];
262 struct hash_list
*next
;
263 unsigned char sha1
[20];
272 struct recent_command
274 struct recent_command
*prev
;
275 struct recent_command
*next
;
279 /* Configured limits on output */
280 static unsigned long max_depth
= 10;
281 static off_t max_packsize
= (1LL << 32) - 1;
282 static int force_update
;
284 /* Stats and misc. counters */
285 static uintmax_t alloc_count
;
286 static uintmax_t marks_set_count
;
287 static uintmax_t object_count_by_type
[1 << TYPE_BITS
];
288 static uintmax_t duplicate_count_by_type
[1 << TYPE_BITS
];
289 static uintmax_t delta_count_by_type
[1 << TYPE_BITS
];
290 static unsigned long object_count
;
291 static unsigned long branch_count
;
292 static unsigned long branch_load_count
;
294 static FILE *pack_edges
;
297 static size_t mem_pool_alloc
= 2*1024*1024 - sizeof(struct mem_pool
);
298 static size_t total_allocd
;
299 static struct mem_pool
*mem_pool
;
301 /* Atom management */
302 static unsigned int atom_table_sz
= 4451;
303 static unsigned int atom_cnt
;
304 static struct atom_str
**atom_table
;
306 /* The .pack file being generated */
307 static unsigned int pack_id
;
308 static struct packed_git
*pack_data
;
309 static struct packed_git
**all_packs
;
310 static unsigned long pack_size
;
312 /* Table of objects we've written. */
313 static unsigned int object_entry_alloc
= 5000;
314 static struct object_entry_pool
*blocks
;
315 static struct object_entry
*object_table
[1 << 16];
316 static struct mark_set
*marks
;
317 static const char* mark_file
;
320 static struct last_object last_blob
;
322 /* Tree management */
323 static unsigned int tree_entry_alloc
= 1000;
324 static void *avail_tree_entry
;
325 static unsigned int avail_tree_table_sz
= 100;
326 static struct avail_tree_content
**avail_tree_table
;
327 static struct dbuf old_tree
;
328 static struct dbuf new_tree
;
331 static unsigned long max_active_branches
= 5;
332 static unsigned long cur_active_branches
;
333 static unsigned long branch_table_sz
= 1039;
334 static struct branch
**branch_table
;
335 static struct branch
*active_branches
;
338 static struct tag
*first_tag
;
339 static struct tag
*last_tag
;
341 /* Input stream parsing */
342 static whenspec_type whenspec
= WHENSPEC_RAW
;
343 static struct strbuf command_buf
;
344 static int unread_command_buf
;
345 static struct recent_command cmd_hist
= {&cmd_hist
, &cmd_hist
, NULL
};
346 static struct recent_command
*cmd_tail
= &cmd_hist
;
347 static struct recent_command
*rc_free
;
348 static unsigned int cmd_save
= 100;
349 static uintmax_t next_mark
;
350 static struct dbuf new_data
;
352 static void write_branch_report(FILE *rpt
, struct branch
*b
)
354 fprintf(rpt
, "%s:\n", b
->name
);
356 fprintf(rpt
, " status :");
358 fputs(" active", rpt
);
359 if (b
->branch_tree
.tree
)
360 fputs(" loaded", rpt
);
361 if (is_null_sha1(b
->branch_tree
.versions
[1].sha1
))
362 fputs(" dirty", rpt
);
365 fprintf(rpt
, " tip commit : %s\n", sha1_to_hex(b
->sha1
));
366 fprintf(rpt
, " old tree : %s\n", sha1_to_hex(b
->branch_tree
.versions
[0].sha1
));
367 fprintf(rpt
, " cur tree : %s\n", sha1_to_hex(b
->branch_tree
.versions
[1].sha1
));
368 fprintf(rpt
, " commit clock: %" PRIuMAX
"\n", b
->last_commit
);
370 fputs(" last pack : ", rpt
);
371 if (b
->pack_id
< MAX_PACK_ID
)
372 fprintf(rpt
, "%u", b
->pack_id
);
378 static void write_crash_report(const char *err
, va_list params
)
380 char *loc
= git_path("fast_import_crash_%d", getpid());
381 FILE *rpt
= fopen(loc
, "w");
384 struct recent_command
*rc
;
387 error("can't write crash report %s: %s", loc
, strerror(errno
));
391 fprintf(stderr
, "fast-import: dumping crash report to %s\n", loc
);
393 fprintf(rpt
, "fast-import crash report:\n");
394 fprintf(rpt
, " fast-import process: %d\n", getpid());
395 fprintf(rpt
, " parent process : %d\n", getppid());
396 fprintf(rpt
, " at %s\n", show_date(time(NULL
), 0, DATE_LOCAL
));
399 fputs("fatal: ", rpt
);
400 vfprintf(rpt
, err
, params
);
404 fputs("Most Recent Commands Before Crash\n", rpt
);
405 fputs("---------------------------------\n", rpt
);
406 for (rc
= cmd_hist
.next
; rc
!= &cmd_hist
; rc
= rc
->next
) {
407 if (rc
->next
== &cmd_hist
)
416 fputs("Active Branch LRU\n", rpt
);
417 fputs("-----------------\n", rpt
);
418 fprintf(rpt
, " active_branches = %lu cur, %lu max\n",
420 max_active_branches
);
422 fputs(" pos clock name\n", rpt
);
423 fputs(" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt
);
424 for (b
= active_branches
, lu
= 0; b
; b
= b
->active_next_branch
)
425 fprintf(rpt
, " %2lu) %6" PRIuMAX
" %s\n",
426 ++lu
, b
->last_commit
, b
->name
);
429 fputs("Inactive Branches\n", rpt
);
430 fputs("-----------------\n", rpt
);
431 for (lu
= 0; lu
< branch_table_sz
; lu
++) {
432 for (b
= branch_table
[lu
]; b
; b
= b
->table_next_branch
)
433 write_branch_report(rpt
, b
);
437 fputs("-------------------\n", rpt
);
438 fputs("END OF CRASH REPORT\n", rpt
);
442 static NORETURN
void die_nicely(const char *err
, va_list params
)
447 va_copy(x_params
, params
);
448 fputs("fatal: ", stderr
);
449 vfprintf(stderr
, err
, params
);
454 write_crash_report(err
, x_params
);
460 static void alloc_objects(unsigned int cnt
)
462 struct object_entry_pool
*b
;
464 b
= xmalloc(sizeof(struct object_entry_pool
)
465 + cnt
* sizeof(struct object_entry
));
466 b
->next_pool
= blocks
;
467 b
->next_free
= b
->entries
;
468 b
->end
= b
->entries
+ cnt
;
473 static struct object_entry
*new_object(unsigned char *sha1
)
475 struct object_entry
*e
;
477 if (blocks
->next_free
== blocks
->end
)
478 alloc_objects(object_entry_alloc
);
480 e
= blocks
->next_free
++;
481 hashcpy(e
->sha1
, sha1
);
485 static struct object_entry
*find_object(unsigned char *sha1
)
487 unsigned int h
= sha1
[0] << 8 | sha1
[1];
488 struct object_entry
*e
;
489 for (e
= object_table
[h
]; e
; e
= e
->next
)
490 if (!hashcmp(sha1
, e
->sha1
))
495 static struct object_entry
*insert_object(unsigned char *sha1
)
497 unsigned int h
= sha1
[0] << 8 | sha1
[1];
498 struct object_entry
*e
= object_table
[h
];
499 struct object_entry
*p
= NULL
;
502 if (!hashcmp(sha1
, e
->sha1
))
508 e
= new_object(sha1
);
518 static unsigned int hc_str(const char *s
, size_t len
)
526 static void *pool_alloc(size_t len
)
531 for (p
= mem_pool
; p
; p
= p
->next_pool
)
532 if ((p
->end
- p
->next_free
>= len
))
536 if (len
>= (mem_pool_alloc
/2)) {
540 total_allocd
+= sizeof(struct mem_pool
) + mem_pool_alloc
;
541 p
= xmalloc(sizeof(struct mem_pool
) + mem_pool_alloc
);
542 p
->next_pool
= mem_pool
;
543 p
->next_free
= p
->space
;
544 p
->end
= p
->next_free
+ mem_pool_alloc
;
549 /* round out to a pointer alignment */
550 if (len
& (sizeof(void*) - 1))
551 len
+= sizeof(void*) - (len
& (sizeof(void*) - 1));
556 static void *pool_calloc(size_t count
, size_t size
)
558 size_t len
= count
* size
;
559 void *r
= pool_alloc(len
);
564 static char *pool_strdup(const char *s
)
566 char *r
= pool_alloc(strlen(s
) + 1);
571 static void size_dbuf(struct dbuf
*b
, size_t maxlen
)
574 if (b
->capacity
>= maxlen
)
578 b
->capacity
= ((maxlen
/ 1024) + 1) * 1024;
579 b
->buffer
= xmalloc(b
->capacity
);
582 static void insert_mark(uintmax_t idnum
, struct object_entry
*oe
)
584 struct mark_set
*s
= marks
;
585 while ((idnum
>> s
->shift
) >= 1024) {
586 s
= pool_calloc(1, sizeof(struct mark_set
));
587 s
->shift
= marks
->shift
+ 10;
588 s
->data
.sets
[0] = marks
;
592 uintmax_t i
= idnum
>> s
->shift
;
593 idnum
-= i
<< s
->shift
;
594 if (!s
->data
.sets
[i
]) {
595 s
->data
.sets
[i
] = pool_calloc(1, sizeof(struct mark_set
));
596 s
->data
.sets
[i
]->shift
= s
->shift
- 10;
600 if (!s
->data
.marked
[idnum
])
602 s
->data
.marked
[idnum
] = oe
;
605 static struct object_entry
*find_mark(uintmax_t idnum
)
607 uintmax_t orig_idnum
= idnum
;
608 struct mark_set
*s
= marks
;
609 struct object_entry
*oe
= NULL
;
610 if ((idnum
>> s
->shift
) < 1024) {
611 while (s
&& s
->shift
) {
612 uintmax_t i
= idnum
>> s
->shift
;
613 idnum
-= i
<< s
->shift
;
617 oe
= s
->data
.marked
[idnum
];
620 die("mark :%" PRIuMAX
" not declared", orig_idnum
);
624 static struct atom_str
*to_atom(const char *s
, unsigned short len
)
626 unsigned int hc
= hc_str(s
, len
) % atom_table_sz
;
629 for (c
= atom_table
[hc
]; c
; c
= c
->next_atom
)
630 if (c
->str_len
== len
&& !strncmp(s
, c
->str_dat
, len
))
633 c
= pool_alloc(sizeof(struct atom_str
) + len
+ 1);
635 strncpy(c
->str_dat
, s
, len
);
637 c
->next_atom
= atom_table
[hc
];
643 static struct branch
*lookup_branch(const char *name
)
645 unsigned int hc
= hc_str(name
, strlen(name
)) % branch_table_sz
;
648 for (b
= branch_table
[hc
]; b
; b
= b
->table_next_branch
)
649 if (!strcmp(name
, b
->name
))
654 static struct branch
*new_branch(const char *name
)
656 unsigned int hc
= hc_str(name
, strlen(name
)) % branch_table_sz
;
657 struct branch
* b
= lookup_branch(name
);
660 die("Invalid attempt to create duplicate branch: %s", name
);
661 switch (check_ref_format(name
)) {
662 case 0: break; /* its valid */
663 case -2: break; /* valid, but too few '/', allow anyway */
665 die("Branch name doesn't conform to GIT standards: %s", name
);
668 b
= pool_calloc(1, sizeof(struct branch
));
669 b
->name
= pool_strdup(name
);
670 b
->table_next_branch
= branch_table
[hc
];
671 b
->branch_tree
.versions
[0].mode
= S_IFDIR
;
672 b
->branch_tree
.versions
[1].mode
= S_IFDIR
;
674 b
->pack_id
= MAX_PACK_ID
;
675 branch_table
[hc
] = b
;
680 static unsigned int hc_entries(unsigned int cnt
)
682 cnt
= cnt
& 7 ? (cnt
/ 8) + 1 : cnt
/ 8;
683 return cnt
< avail_tree_table_sz
? cnt
: avail_tree_table_sz
- 1;
686 static struct tree_content
*new_tree_content(unsigned int cnt
)
688 struct avail_tree_content
*f
, *l
= NULL
;
689 struct tree_content
*t
;
690 unsigned int hc
= hc_entries(cnt
);
692 for (f
= avail_tree_table
[hc
]; f
; l
= f
, f
= f
->next_avail
)
693 if (f
->entry_capacity
>= cnt
)
698 l
->next_avail
= f
->next_avail
;
700 avail_tree_table
[hc
] = f
->next_avail
;
702 cnt
= cnt
& 7 ? ((cnt
/ 8) + 1) * 8 : cnt
;
703 f
= pool_alloc(sizeof(*t
) + sizeof(t
->entries
[0]) * cnt
);
704 f
->entry_capacity
= cnt
;
707 t
= (struct tree_content
*)f
;
713 static void release_tree_entry(struct tree_entry
*e
);
714 static void release_tree_content(struct tree_content
*t
)
716 struct avail_tree_content
*f
= (struct avail_tree_content
*)t
;
717 unsigned int hc
= hc_entries(f
->entry_capacity
);
718 f
->next_avail
= avail_tree_table
[hc
];
719 avail_tree_table
[hc
] = f
;
722 static void release_tree_content_recursive(struct tree_content
*t
)
725 for (i
= 0; i
< t
->entry_count
; i
++)
726 release_tree_entry(t
->entries
[i
]);
727 release_tree_content(t
);
730 static struct tree_content
*grow_tree_content(
731 struct tree_content
*t
,
734 struct tree_content
*r
= new_tree_content(t
->entry_count
+ amt
);
735 r
->entry_count
= t
->entry_count
;
736 r
->delta_depth
= t
->delta_depth
;
737 memcpy(r
->entries
,t
->entries
,t
->entry_count
*sizeof(t
->entries
[0]));
738 release_tree_content(t
);
742 static struct tree_entry
*new_tree_entry(void)
744 struct tree_entry
*e
;
746 if (!avail_tree_entry
) {
747 unsigned int n
= tree_entry_alloc
;
748 total_allocd
+= n
* sizeof(struct tree_entry
);
749 avail_tree_entry
= e
= xmalloc(n
* sizeof(struct tree_entry
));
751 *((void**)e
) = e
+ 1;
757 e
= avail_tree_entry
;
758 avail_tree_entry
= *((void**)e
);
762 static void release_tree_entry(struct tree_entry
*e
)
765 release_tree_content_recursive(e
->tree
);
766 *((void**)e
) = avail_tree_entry
;
767 avail_tree_entry
= e
;
770 static struct tree_content
*dup_tree_content(struct tree_content
*s
)
772 struct tree_content
*d
;
773 struct tree_entry
*a
, *b
;
778 d
= new_tree_content(s
->entry_count
);
779 for (i
= 0; i
< s
->entry_count
; i
++) {
781 b
= new_tree_entry();
782 memcpy(b
, a
, sizeof(*a
));
783 if (a
->tree
&& is_null_sha1(b
->versions
[1].sha1
))
784 b
->tree
= dup_tree_content(a
->tree
);
789 d
->entry_count
= s
->entry_count
;
790 d
->delta_depth
= s
->delta_depth
;
795 static void start_packfile(void)
797 static char tmpfile
[PATH_MAX
];
798 struct packed_git
*p
;
799 struct pack_header hdr
;
802 snprintf(tmpfile
, sizeof(tmpfile
),
803 "%s/tmp_pack_XXXXXX", get_object_directory());
804 pack_fd
= xmkstemp(tmpfile
);
805 p
= xcalloc(1, sizeof(*p
) + strlen(tmpfile
) + 2);
806 strcpy(p
->pack_name
, tmpfile
);
807 p
->pack_fd
= pack_fd
;
809 hdr
.hdr_signature
= htonl(PACK_SIGNATURE
);
810 hdr
.hdr_version
= htonl(2);
812 write_or_die(p
->pack_fd
, &hdr
, sizeof(hdr
));
815 pack_size
= sizeof(hdr
);
818 all_packs
= xrealloc(all_packs
, sizeof(*all_packs
) * (pack_id
+ 1));
819 all_packs
[pack_id
] = p
;
822 static int oecmp (const void *a_
, const void *b_
)
824 struct object_entry
*a
= *((struct object_entry
**)a_
);
825 struct object_entry
*b
= *((struct object_entry
**)b_
);
826 return hashcmp(a
->sha1
, b
->sha1
);
829 static char *create_index(void)
831 static char tmpfile
[PATH_MAX
];
834 struct object_entry
**idx
, **c
, **last
, *e
;
835 struct object_entry_pool
*o
;
839 /* Build the sorted table of object IDs. */
840 idx
= xmalloc(object_count
* sizeof(struct object_entry
*));
842 for (o
= blocks
; o
; o
= o
->next_pool
)
843 for (e
= o
->next_free
; e
-- != o
->entries
;)
844 if (pack_id
== e
->pack_id
)
846 last
= idx
+ object_count
;
848 die("internal consistency error creating the index");
849 qsort(idx
, object_count
, sizeof(struct object_entry
*), oecmp
);
851 /* Generate the fan-out array. */
853 for (i
= 0; i
< 256; i
++) {
854 struct object_entry
**next
= c
;;
855 while (next
< last
) {
856 if ((*next
)->sha1
[0] != i
)
860 array
[i
] = htonl(next
- idx
);
864 snprintf(tmpfile
, sizeof(tmpfile
),
865 "%s/tmp_idx_XXXXXX", get_object_directory());
866 idx_fd
= xmkstemp(tmpfile
);
867 f
= sha1fd(idx_fd
, tmpfile
);
868 sha1write(f
, array
, 256 * sizeof(int));
870 for (c
= idx
; c
!= last
; c
++) {
871 uint32_t offset
= htonl((*c
)->offset
);
872 sha1write(f
, &offset
, 4);
873 sha1write(f
, (*c
)->sha1
, sizeof((*c
)->sha1
));
874 SHA1_Update(&ctx
, (*c
)->sha1
, 20);
876 sha1write(f
, pack_data
->sha1
, sizeof(pack_data
->sha1
));
877 sha1close(f
, NULL
, 1);
879 SHA1_Final(pack_data
->sha1
, &ctx
);
883 static char *keep_pack(char *curr_index_name
)
885 static char name
[PATH_MAX
];
886 static const char *keep_msg
= "fast-import";
889 chmod(pack_data
->pack_name
, 0444);
890 chmod(curr_index_name
, 0444);
892 snprintf(name
, sizeof(name
), "%s/pack/pack-%s.keep",
893 get_object_directory(), sha1_to_hex(pack_data
->sha1
));
894 keep_fd
= open(name
, O_RDWR
|O_CREAT
|O_EXCL
, 0600);
896 die("cannot create keep file");
897 write(keep_fd
, keep_msg
, strlen(keep_msg
));
900 snprintf(name
, sizeof(name
), "%s/pack/pack-%s.pack",
901 get_object_directory(), sha1_to_hex(pack_data
->sha1
));
902 if (move_temp_to_file(pack_data
->pack_name
, name
))
903 die("cannot store pack file");
905 snprintf(name
, sizeof(name
), "%s/pack/pack-%s.idx",
906 get_object_directory(), sha1_to_hex(pack_data
->sha1
));
907 if (move_temp_to_file(curr_index_name
, name
))
908 die("cannot store index file");
912 static void unkeep_all_packs(void)
914 static char name
[PATH_MAX
];
917 for (k
= 0; k
< pack_id
; k
++) {
918 struct packed_git
*p
= all_packs
[k
];
919 snprintf(name
, sizeof(name
), "%s/pack/pack-%s.keep",
920 get_object_directory(), sha1_to_hex(p
->sha1
));
925 static void end_packfile(void)
927 struct packed_git
*old_p
= pack_data
, *new_p
;
935 fixup_pack_header_footer(pack_data
->pack_fd
, pack_data
->sha1
,
936 pack_data
->pack_name
, object_count
);
937 close(pack_data
->pack_fd
);
938 idx_name
= keep_pack(create_index());
940 /* Register the packfile with core git's machinary. */
941 new_p
= add_packed_git(idx_name
, strlen(idx_name
), 1);
943 die("core git rejected index %s", idx_name
);
944 new_p
->windows
= old_p
->windows
;
945 all_packs
[pack_id
] = new_p
;
946 install_packed_git(new_p
);
948 /* Print the boundary */
950 fprintf(pack_edges
, "%s:", new_p
->pack_name
);
951 for (i
= 0; i
< branch_table_sz
; i
++) {
952 for (b
= branch_table
[i
]; b
; b
= b
->table_next_branch
) {
953 if (b
->pack_id
== pack_id
)
954 fprintf(pack_edges
, " %s", sha1_to_hex(b
->sha1
));
957 for (t
= first_tag
; t
; t
= t
->next_tag
) {
958 if (t
->pack_id
== pack_id
)
959 fprintf(pack_edges
, " %s", sha1_to_hex(t
->sha1
));
961 fputc('\n', pack_edges
);
968 unlink(old_p
->pack_name
);
971 /* We can't carry a delta across packfiles. */
972 free(last_blob
.data
);
973 last_blob
.data
= NULL
;
975 last_blob
.offset
= 0;
979 static void cycle_packfile(void)
985 static size_t encode_header(
986 enum object_type type
,
993 if (type
< OBJ_COMMIT
|| type
> OBJ_REF_DELTA
)
994 die("bad type %d", type
);
996 c
= (type
<< 4) | (size
& 15);
1008 static int store_object(
1009 enum object_type type
,
1012 struct last_object
*last
,
1013 unsigned char *sha1out
,
1017 struct object_entry
*e
;
1018 unsigned char hdr
[96];
1019 unsigned char sha1
[20];
1020 unsigned long hdrlen
, deltalen
;
1024 hdrlen
= sprintf((char*)hdr
,"%s %lu", typename(type
),
1025 (unsigned long)datlen
) + 1;
1027 SHA1_Update(&c
, hdr
, hdrlen
);
1028 SHA1_Update(&c
, dat
, datlen
);
1029 SHA1_Final(sha1
, &c
);
1031 hashcpy(sha1out
, sha1
);
1033 e
= insert_object(sha1
);
1035 insert_mark(mark
, e
);
1037 duplicate_count_by_type
[type
]++;
1039 } else if (find_sha1_pack(sha1
, packed_git
)) {
1041 e
->pack_id
= MAX_PACK_ID
;
1042 e
->offset
= 1; /* just not zero! */
1043 duplicate_count_by_type
[type
]++;
1047 if (last
&& last
->data
&& last
->depth
< max_depth
) {
1048 delta
= diff_delta(last
->data
, last
->len
,
1051 if (delta
&& deltalen
>= datlen
) {
1058 memset(&s
, 0, sizeof(s
));
1059 deflateInit(&s
, zlib_compression_level
);
1062 s
.avail_in
= deltalen
;
1065 s
.avail_in
= datlen
;
1067 s
.avail_out
= deflateBound(&s
, s
.avail_in
);
1068 s
.next_out
= out
= xmalloc(s
.avail_out
);
1069 while (deflate(&s
, Z_FINISH
) == Z_OK
)
1073 /* Determine if we should auto-checkpoint. */
1074 if ((pack_size
+ 60 + s
.total_out
) > max_packsize
1075 || (pack_size
+ 60 + s
.total_out
) < pack_size
) {
1077 /* This new object needs to *not* have the current pack_id. */
1078 e
->pack_id
= pack_id
+ 1;
1081 /* We cannot carry a delta into the new pack. */
1086 memset(&s
, 0, sizeof(s
));
1087 deflateInit(&s
, zlib_compression_level
);
1089 s
.avail_in
= datlen
;
1090 s
.avail_out
= deflateBound(&s
, s
.avail_in
);
1091 s
.next_out
= out
= xrealloc(out
, s
.avail_out
);
1092 while (deflate(&s
, Z_FINISH
) == Z_OK
)
1099 e
->pack_id
= pack_id
;
1100 e
->offset
= pack_size
;
1102 object_count_by_type
[type
]++;
1105 unsigned long ofs
= e
->offset
- last
->offset
;
1106 unsigned pos
= sizeof(hdr
) - 1;
1108 delta_count_by_type
[type
]++;
1111 hdrlen
= encode_header(OBJ_OFS_DELTA
, deltalen
, hdr
);
1112 write_or_die(pack_data
->pack_fd
, hdr
, hdrlen
);
1113 pack_size
+= hdrlen
;
1115 hdr
[pos
] = ofs
& 127;
1117 hdr
[--pos
] = 128 | (--ofs
& 127);
1118 write_or_die(pack_data
->pack_fd
, hdr
+ pos
, sizeof(hdr
) - pos
);
1119 pack_size
+= sizeof(hdr
) - pos
;
1123 hdrlen
= encode_header(type
, datlen
, hdr
);
1124 write_or_die(pack_data
->pack_fd
, hdr
, hdrlen
);
1125 pack_size
+= hdrlen
;
1128 write_or_die(pack_data
->pack_fd
, out
, s
.total_out
);
1129 pack_size
+= s
.total_out
;
1137 last
->offset
= e
->offset
;
1143 static void *gfi_unpack_entry(
1144 struct object_entry
*oe
,
1145 unsigned long *sizep
)
1147 enum object_type type
;
1148 struct packed_git
*p
= all_packs
[oe
->pack_id
];
1150 p
->pack_size
= pack_size
+ 20;
1151 return unpack_entry(p
, oe
->offset
, &type
, sizep
);
1154 static const char *get_mode(const char *str
, uint16_t *modep
)
1159 while ((c
= *str
++) != ' ') {
1160 if (c
< '0' || c
> '7')
1162 mode
= (mode
<< 3) + (c
- '0');
1168 static void load_tree(struct tree_entry
*root
)
1170 unsigned char* sha1
= root
->versions
[1].sha1
;
1171 struct object_entry
*myoe
;
1172 struct tree_content
*t
;
1177 root
->tree
= t
= new_tree_content(8);
1178 if (is_null_sha1(sha1
))
1181 myoe
= find_object(sha1
);
1182 if (myoe
&& myoe
->pack_id
!= MAX_PACK_ID
) {
1183 if (myoe
->type
!= OBJ_TREE
)
1184 die("Not a tree: %s", sha1_to_hex(sha1
));
1186 buf
= gfi_unpack_entry(myoe
, &size
);
1188 enum object_type type
;
1189 buf
= read_sha1_file(sha1
, &type
, &size
);
1190 if (!buf
|| type
!= OBJ_TREE
)
1191 die("Can't load tree %s", sha1_to_hex(sha1
));
1195 while (c
!= (buf
+ size
)) {
1196 struct tree_entry
*e
= new_tree_entry();
1198 if (t
->entry_count
== t
->entry_capacity
)
1199 root
->tree
= t
= grow_tree_content(t
, t
->entry_count
);
1200 t
->entries
[t
->entry_count
++] = e
;
1203 c
= get_mode(c
, &e
->versions
[1].mode
);
1205 die("Corrupt mode in %s", sha1_to_hex(sha1
));
1206 e
->versions
[0].mode
= e
->versions
[1].mode
;
1207 e
->name
= to_atom(c
, strlen(c
));
1208 c
+= e
->name
->str_len
+ 1;
1209 hashcpy(e
->versions
[0].sha1
, (unsigned char*)c
);
1210 hashcpy(e
->versions
[1].sha1
, (unsigned char*)c
);
1216 static int tecmp0 (const void *_a
, const void *_b
)
1218 struct tree_entry
*a
= *((struct tree_entry
**)_a
);
1219 struct tree_entry
*b
= *((struct tree_entry
**)_b
);
1220 return base_name_compare(
1221 a
->name
->str_dat
, a
->name
->str_len
, a
->versions
[0].mode
,
1222 b
->name
->str_dat
, b
->name
->str_len
, b
->versions
[0].mode
);
1225 static int tecmp1 (const void *_a
, const void *_b
)
1227 struct tree_entry
*a
= *((struct tree_entry
**)_a
);
1228 struct tree_entry
*b
= *((struct tree_entry
**)_b
);
1229 return base_name_compare(
1230 a
->name
->str_dat
, a
->name
->str_len
, a
->versions
[1].mode
,
1231 b
->name
->str_dat
, b
->name
->str_len
, b
->versions
[1].mode
);
1234 static void mktree(struct tree_content
*t
,
1244 qsort(t
->entries
,t
->entry_count
,sizeof(t
->entries
[0]),tecmp0
);
1246 qsort(t
->entries
,t
->entry_count
,sizeof(t
->entries
[0]),tecmp1
);
1248 for (i
= 0; i
< t
->entry_count
; i
++) {
1249 if (t
->entries
[i
]->versions
[v
].mode
)
1250 maxlen
+= t
->entries
[i
]->name
->str_len
+ 34;
1253 size_dbuf(b
, maxlen
);
1255 for (i
= 0; i
< t
->entry_count
; i
++) {
1256 struct tree_entry
*e
= t
->entries
[i
];
1257 if (!e
->versions
[v
].mode
)
1259 c
+= sprintf(c
, "%o", (unsigned int)e
->versions
[v
].mode
);
1261 strcpy(c
, e
->name
->str_dat
);
1262 c
+= e
->name
->str_len
+ 1;
1263 hashcpy((unsigned char*)c
, e
->versions
[v
].sha1
);
1266 *szp
= c
- (char*)b
->buffer
;
1269 static void store_tree(struct tree_entry
*root
)
1271 struct tree_content
*t
= root
->tree
;
1272 unsigned int i
, j
, del
;
1273 unsigned long new_len
;
1274 struct last_object lo
;
1275 struct object_entry
*le
;
1277 if (!is_null_sha1(root
->versions
[1].sha1
))
1280 for (i
= 0; i
< t
->entry_count
; i
++) {
1281 if (t
->entries
[i
]->tree
)
1282 store_tree(t
->entries
[i
]);
1285 le
= find_object(root
->versions
[0].sha1
);
1286 if (!S_ISDIR(root
->versions
[0].mode
)
1288 || le
->pack_id
!= pack_id
) {
1293 mktree(t
, 0, &lo
.len
, &old_tree
);
1294 lo
.data
= old_tree
.buffer
;
1295 lo
.offset
= le
->offset
;
1296 lo
.depth
= t
->delta_depth
;
1300 mktree(t
, 1, &new_len
, &new_tree
);
1301 store_object(OBJ_TREE
, new_tree
.buffer
, new_len
,
1302 &lo
, root
->versions
[1].sha1
, 0);
1304 t
->delta_depth
= lo
.depth
;
1305 for (i
= 0, j
= 0, del
= 0; i
< t
->entry_count
; i
++) {
1306 struct tree_entry
*e
= t
->entries
[i
];
1307 if (e
->versions
[1].mode
) {
1308 e
->versions
[0].mode
= e
->versions
[1].mode
;
1309 hashcpy(e
->versions
[0].sha1
, e
->versions
[1].sha1
);
1310 t
->entries
[j
++] = e
;
1312 release_tree_entry(e
);
1316 t
->entry_count
-= del
;
1319 static int tree_content_set(
1320 struct tree_entry
*root
,
1322 const unsigned char *sha1
,
1323 const uint16_t mode
,
1324 struct tree_content
*subtree
)
1326 struct tree_content
*t
= root
->tree
;
1329 struct tree_entry
*e
;
1331 slash1
= strchr(p
, '/');
1337 die("Empty path component found in input");
1338 if (!slash1
&& !S_ISDIR(mode
) && subtree
)
1339 die("Non-directories cannot have subtrees");
1341 for (i
= 0; i
< t
->entry_count
; i
++) {
1343 if (e
->name
->str_len
== n
&& !strncmp(p
, e
->name
->str_dat
, n
)) {
1346 && e
->versions
[1].mode
== mode
1347 && !hashcmp(e
->versions
[1].sha1
, sha1
))
1349 e
->versions
[1].mode
= mode
;
1350 hashcpy(e
->versions
[1].sha1
, sha1
);
1352 release_tree_content_recursive(e
->tree
);
1354 hashclr(root
->versions
[1].sha1
);
1357 if (!S_ISDIR(e
->versions
[1].mode
)) {
1358 e
->tree
= new_tree_content(8);
1359 e
->versions
[1].mode
= S_IFDIR
;
1363 if (tree_content_set(e
, slash1
+ 1, sha1
, mode
, subtree
)) {
1364 hashclr(root
->versions
[1].sha1
);
1371 if (t
->entry_count
== t
->entry_capacity
)
1372 root
->tree
= t
= grow_tree_content(t
, t
->entry_count
);
1373 e
= new_tree_entry();
1374 e
->name
= to_atom(p
, n
);
1375 e
->versions
[0].mode
= 0;
1376 hashclr(e
->versions
[0].sha1
);
1377 t
->entries
[t
->entry_count
++] = e
;
1379 e
->tree
= new_tree_content(8);
1380 e
->versions
[1].mode
= S_IFDIR
;
1381 tree_content_set(e
, slash1
+ 1, sha1
, mode
, subtree
);
1384 e
->versions
[1].mode
= mode
;
1385 hashcpy(e
->versions
[1].sha1
, sha1
);
1387 hashclr(root
->versions
[1].sha1
);
1391 static int tree_content_remove(
1392 struct tree_entry
*root
,
1394 struct tree_entry
*backup_leaf
)
1396 struct tree_content
*t
= root
->tree
;
1399 struct tree_entry
*e
;
1401 slash1
= strchr(p
, '/');
1407 for (i
= 0; i
< t
->entry_count
; i
++) {
1409 if (e
->name
->str_len
== n
&& !strncmp(p
, e
->name
->str_dat
, n
)) {
1410 if (!slash1
|| !S_ISDIR(e
->versions
[1].mode
))
1414 if (tree_content_remove(e
, slash1
+ 1, backup_leaf
)) {
1415 for (n
= 0; n
< e
->tree
->entry_count
; n
++) {
1416 if (e
->tree
->entries
[n
]->versions
[1].mode
) {
1417 hashclr(root
->versions
[1].sha1
);
1431 memcpy(backup_leaf
, e
, sizeof(*backup_leaf
));
1433 release_tree_content_recursive(e
->tree
);
1435 e
->versions
[1].mode
= 0;
1436 hashclr(e
->versions
[1].sha1
);
1437 hashclr(root
->versions
[1].sha1
);
1441 static int tree_content_get(
1442 struct tree_entry
*root
,
1444 struct tree_entry
*leaf
)
1446 struct tree_content
*t
= root
->tree
;
1449 struct tree_entry
*e
;
1451 slash1
= strchr(p
, '/');
1457 for (i
= 0; i
< t
->entry_count
; i
++) {
1459 if (e
->name
->str_len
== n
&& !strncmp(p
, e
->name
->str_dat
, n
)) {
1461 memcpy(leaf
, e
, sizeof(*leaf
));
1462 if (e
->tree
&& is_null_sha1(e
->versions
[1].sha1
))
1463 leaf
->tree
= dup_tree_content(e
->tree
);
1468 if (!S_ISDIR(e
->versions
[1].mode
))
1472 return tree_content_get(e
, slash1
+ 1, leaf
);
1478 static int update_branch(struct branch
*b
)
1480 static const char *msg
= "fast-import";
1481 struct ref_lock
*lock
;
1482 unsigned char old_sha1
[20];
1484 if (read_ref(b
->name
, old_sha1
))
1486 lock
= lock_any_ref_for_update(b
->name
, old_sha1
, 0);
1488 return error("Unable to lock %s", b
->name
);
1489 if (!force_update
&& !is_null_sha1(old_sha1
)) {
1490 struct commit
*old_cmit
, *new_cmit
;
1492 old_cmit
= lookup_commit_reference_gently(old_sha1
, 0);
1493 new_cmit
= lookup_commit_reference_gently(b
->sha1
, 0);
1494 if (!old_cmit
|| !new_cmit
) {
1496 return error("Branch %s is missing commits.", b
->name
);
1499 if (!in_merge_bases(old_cmit
, &new_cmit
, 1)) {
1501 warning("Not updating %s"
1502 " (new tip %s does not contain %s)",
1503 b
->name
, sha1_to_hex(b
->sha1
), sha1_to_hex(old_sha1
));
1507 if (write_ref_sha1(lock
, b
->sha1
, msg
) < 0)
1508 return error("Unable to update %s", b
->name
);
1512 static void dump_branches(void)
1517 for (i
= 0; i
< branch_table_sz
; i
++) {
1518 for (b
= branch_table
[i
]; b
; b
= b
->table_next_branch
)
1519 failure
|= update_branch(b
);
1523 static void dump_tags(void)
1525 static const char *msg
= "fast-import";
1527 struct ref_lock
*lock
;
1528 char ref_name
[PATH_MAX
];
1530 for (t
= first_tag
; t
; t
= t
->next_tag
) {
1531 sprintf(ref_name
, "tags/%s", t
->name
);
1532 lock
= lock_ref_sha1(ref_name
, NULL
);
1533 if (!lock
|| write_ref_sha1(lock
, t
->sha1
, msg
) < 0)
1534 failure
|= error("Unable to update %s", ref_name
);
1538 static void dump_marks_helper(FILE *f
,
1544 for (k
= 0; k
< 1024; k
++) {
1545 if (m
->data
.sets
[k
])
1546 dump_marks_helper(f
, (base
+ k
) << m
->shift
,
1550 for (k
= 0; k
< 1024; k
++) {
1551 if (m
->data
.marked
[k
])
1552 fprintf(f
, ":%" PRIuMAX
" %s\n", base
+ k
,
1553 sha1_to_hex(m
->data
.marked
[k
]->sha1
));
1558 static void dump_marks(void)
1560 static struct lock_file mark_lock
;
1567 mark_fd
= hold_lock_file_for_update(&mark_lock
, mark_file
, 0);
1569 failure
|= error("Unable to write marks file %s: %s",
1570 mark_file
, strerror(errno
));
1574 f
= fdopen(mark_fd
, "w");
1576 rollback_lock_file(&mark_lock
);
1577 failure
|= error("Unable to write marks file %s: %s",
1578 mark_file
, strerror(errno
));
1582 dump_marks_helper(f
, 0, marks
);
1584 if (commit_lock_file(&mark_lock
))
1585 failure
|= error("Unable to write marks file %s: %s",
1586 mark_file
, strerror(errno
));
1589 static void read_next_command(void)
1592 if (unread_command_buf
) {
1593 unread_command_buf
= 0;
1594 if (command_buf
.eof
)
1597 struct recent_command
*rc
;
1599 command_buf
.buf
= NULL
;
1600 read_line(&command_buf
, stdin
, '\n');
1601 if (command_buf
.eof
)
1609 cmd_hist
.next
= rc
->next
;
1610 cmd_hist
.next
->prev
= &cmd_hist
;
1614 rc
->buf
= command_buf
.buf
;
1615 rc
->prev
= cmd_tail
;
1616 rc
->next
= cmd_hist
.prev
;
1617 rc
->prev
->next
= rc
;
1620 } while (command_buf
.buf
[0] == '#');
1623 static void skip_optional_lf(void)
1625 int term_char
= fgetc(stdin
);
1626 if (term_char
!= '\n' && term_char
!= EOF
)
1627 ungetc(term_char
, stdin
);
1630 static void cmd_mark(void)
1632 if (!prefixcmp(command_buf
.buf
, "mark :")) {
1633 next_mark
= strtoumax(command_buf
.buf
+ 6, NULL
, 10);
1634 read_next_command();
1640 static void *cmd_data (size_t *size
)
1645 if (prefixcmp(command_buf
.buf
, "data "))
1646 die("Expected 'data n' command, found: %s", command_buf
.buf
);
1648 if (!prefixcmp(command_buf
.buf
+ 5, "<<")) {
1649 char *term
= xstrdup(command_buf
.buf
+ 5 + 2);
1650 size_t sz
= 8192, term_len
= command_buf
.len
- 5 - 2;
1652 buffer
= xmalloc(sz
);
1653 command_buf
.buf
= NULL
;
1655 read_line(&command_buf
, stdin
, '\n');
1656 if (command_buf
.eof
)
1657 die("EOF in data (terminator '%s' not found)", term
);
1658 if (term_len
== command_buf
.len
1659 && !strcmp(term
, command_buf
.buf
))
1661 ALLOC_GROW(buffer
, length
+ command_buf
.len
, sz
);
1662 memcpy(buffer
+ length
,
1664 command_buf
.len
- 1);
1665 length
+= command_buf
.len
- 1;
1666 buffer
[length
++] = '\n';
1672 length
= strtoul(command_buf
.buf
+ 5, NULL
, 10);
1673 buffer
= xmalloc(length
);
1674 while (n
< length
) {
1675 size_t s
= fread(buffer
+ n
, 1, length
- n
, stdin
);
1676 if (!s
&& feof(stdin
))
1677 die("EOF in data (%lu bytes remaining)",
1678 (unsigned long)(length
- n
));
1688 static int validate_raw_date(const char *src
, char *result
, int maxlen
)
1690 const char *orig_src
= src
;
1693 strtoul(src
, &endp
, 10);
1694 if (endp
== src
|| *endp
!= ' ')
1698 if (*src
!= '-' && *src
!= '+')
1702 strtoul(src
+ 1, &endp
, 10);
1703 if (endp
== src
|| *endp
|| (endp
- orig_src
) >= maxlen
)
1706 strcpy(result
, orig_src
);
1710 static char *parse_ident(const char *buf
)
1716 gt
= strrchr(buf
, '>');
1718 die("Missing > in ident string: %s", buf
);
1721 die("Missing space after > in ident string: %s", buf
);
1723 name_len
= gt
- buf
;
1724 ident
= xmalloc(name_len
+ 24);
1725 strncpy(ident
, buf
, name_len
);
1729 if (validate_raw_date(gt
, ident
+ name_len
, 24) < 0)
1730 die("Invalid raw date \"%s\" in ident: %s", gt
, buf
);
1732 case WHENSPEC_RFC2822
:
1733 if (parse_date(gt
, ident
+ name_len
, 24) < 0)
1734 die("Invalid rfc2822 date \"%s\" in ident: %s", gt
, buf
);
1737 if (strcmp("now", gt
))
1738 die("Date in ident must be 'now': %s", buf
);
1739 datestamp(ident
+ name_len
, 24);
1746 static void cmd_new_blob(void)
1751 read_next_command();
1755 if (store_object(OBJ_BLOB
, d
, l
, &last_blob
, NULL
, next_mark
))
1759 static void unload_one_branch(void)
1761 while (cur_active_branches
1762 && cur_active_branches
>= max_active_branches
) {
1763 uintmax_t min_commit
= ULONG_MAX
;
1764 struct branch
*e
, *l
= NULL
, *p
= NULL
;
1766 for (e
= active_branches
; e
; e
= e
->active_next_branch
) {
1767 if (e
->last_commit
< min_commit
) {
1769 min_commit
= e
->last_commit
;
1775 e
= p
->active_next_branch
;
1776 p
->active_next_branch
= e
->active_next_branch
;
1778 e
= active_branches
;
1779 active_branches
= e
->active_next_branch
;
1782 e
->active_next_branch
= NULL
;
1783 if (e
->branch_tree
.tree
) {
1784 release_tree_content_recursive(e
->branch_tree
.tree
);
1785 e
->branch_tree
.tree
= NULL
;
1787 cur_active_branches
--;
1791 static void load_branch(struct branch
*b
)
1793 load_tree(&b
->branch_tree
);
1796 b
->active_next_branch
= active_branches
;
1797 active_branches
= b
;
1798 cur_active_branches
++;
1799 branch_load_count
++;
1803 static void file_change_m(struct branch
*b
)
1805 const char *p
= command_buf
.buf
+ 2;
1808 struct object_entry
*oe
= oe
;
1809 unsigned char sha1
[20];
1810 uint16_t mode
, inline_data
= 0;
1812 p
= get_mode(p
, &mode
);
1814 die("Corrupt mode: %s", command_buf
.buf
);
1816 case S_IFREG
| 0644:
1817 case S_IFREG
| 0755:
1824 die("Corrupt mode: %s", command_buf
.buf
);
1829 oe
= find_mark(strtoumax(p
+ 1, &x
, 10));
1830 hashcpy(sha1
, oe
->sha1
);
1832 } else if (!prefixcmp(p
, "inline")) {
1836 if (get_sha1_hex(p
, sha1
))
1837 die("Invalid SHA1: %s", command_buf
.buf
);
1838 oe
= find_object(sha1
);
1842 die("Missing space after SHA1: %s", command_buf
.buf
);
1844 p_uq
= unquote_c_style(p
, &endp
);
1847 die("Garbage after path in: %s", command_buf
.buf
);
1855 p
= p_uq
= xstrdup(p
);
1856 read_next_command();
1858 if (store_object(OBJ_BLOB
, d
, l
, &last_blob
, sha1
, 0))
1861 if (oe
->type
!= OBJ_BLOB
)
1862 die("Not a blob (actually a %s): %s",
1863 command_buf
.buf
, typename(oe
->type
));
1865 enum object_type type
= sha1_object_info(sha1
, NULL
);
1867 die("Blob not found: %s", command_buf
.buf
);
1868 if (type
!= OBJ_BLOB
)
1869 die("Not a blob (actually a %s): %s",
1870 typename(type
), command_buf
.buf
);
1873 tree_content_set(&b
->branch_tree
, p
, sha1
, S_IFREG
| mode
, NULL
);
1877 static void file_change_d(struct branch
*b
)
1879 const char *p
= command_buf
.buf
+ 2;
1883 p_uq
= unquote_c_style(p
, &endp
);
1886 die("Garbage after path in: %s", command_buf
.buf
);
1889 tree_content_remove(&b
->branch_tree
, p
, NULL
);
1893 static void file_change_cr(struct branch
*b
, int rename
)
1898 struct tree_entry leaf
;
1900 s
= command_buf
.buf
+ 2;
1901 s_uq
= unquote_c_style(s
, &endp
);
1904 die("Missing space after source: %s", command_buf
.buf
);
1907 endp
= strchr(s
, ' ');
1909 die("Missing space after source: %s", command_buf
.buf
);
1910 s_uq
= xmalloc(endp
- s
+ 1);
1911 memcpy(s_uq
, s
, endp
- s
);
1918 die("Missing dest: %s", command_buf
.buf
);
1921 d_uq
= unquote_c_style(d
, &endp
);
1924 die("Garbage after dest in: %s", command_buf
.buf
);
1928 memset(&leaf
, 0, sizeof(leaf
));
1930 tree_content_remove(&b
->branch_tree
, s
, &leaf
);
1932 tree_content_get(&b
->branch_tree
, s
, &leaf
);
1933 if (!leaf
.versions
[1].mode
)
1934 die("Path %s not in branch", s
);
1935 tree_content_set(&b
->branch_tree
, d
,
1936 leaf
.versions
[1].sha1
,
1937 leaf
.versions
[1].mode
,
1944 static void file_change_deleteall(struct branch
*b
)
1946 release_tree_content_recursive(b
->branch_tree
.tree
);
1947 hashclr(b
->branch_tree
.versions
[0].sha1
);
1948 hashclr(b
->branch_tree
.versions
[1].sha1
);
1949 load_tree(&b
->branch_tree
);
1952 static void cmd_from_commit(struct branch
*b
, char *buf
, unsigned long size
)
1954 if (!buf
|| size
< 46)
1955 die("Not a valid commit: %s", sha1_to_hex(b
->sha1
));
1956 if (memcmp("tree ", buf
, 5)
1957 || get_sha1_hex(buf
+ 5, b
->branch_tree
.versions
[1].sha1
))
1958 die("The commit %s is corrupt", sha1_to_hex(b
->sha1
));
1959 hashcpy(b
->branch_tree
.versions
[0].sha1
,
1960 b
->branch_tree
.versions
[1].sha1
);
1963 static void cmd_from_existing(struct branch
*b
)
1965 if (is_null_sha1(b
->sha1
)) {
1966 hashclr(b
->branch_tree
.versions
[0].sha1
);
1967 hashclr(b
->branch_tree
.versions
[1].sha1
);
1972 buf
= read_object_with_reference(b
->sha1
,
1973 commit_type
, &size
, b
->sha1
);
1974 cmd_from_commit(b
, buf
, size
);
1979 static int cmd_from(struct branch
*b
)
1984 if (prefixcmp(command_buf
.buf
, "from "))
1987 if (b
->branch_tree
.tree
) {
1988 release_tree_content_recursive(b
->branch_tree
.tree
);
1989 b
->branch_tree
.tree
= NULL
;
1992 from
= strchr(command_buf
.buf
, ' ') + 1;
1993 s
= lookup_branch(from
);
1995 die("Can't create a branch from itself: %s", b
->name
);
1997 unsigned char *t
= s
->branch_tree
.versions
[1].sha1
;
1998 hashcpy(b
->sha1
, s
->sha1
);
1999 hashcpy(b
->branch_tree
.versions
[0].sha1
, t
);
2000 hashcpy(b
->branch_tree
.versions
[1].sha1
, t
);
2001 } else if (*from
== ':') {
2002 uintmax_t idnum
= strtoumax(from
+ 1, NULL
, 10);
2003 struct object_entry
*oe
= find_mark(idnum
);
2004 if (oe
->type
!= OBJ_COMMIT
)
2005 die("Mark :%" PRIuMAX
" not a commit", idnum
);
2006 hashcpy(b
->sha1
, oe
->sha1
);
2007 if (oe
->pack_id
!= MAX_PACK_ID
) {
2009 char *buf
= gfi_unpack_entry(oe
, &size
);
2010 cmd_from_commit(b
, buf
, size
);
2013 cmd_from_existing(b
);
2014 } else if (!get_sha1(from
, b
->sha1
))
2015 cmd_from_existing(b
);
2017 die("Invalid ref name or SHA1 expression: %s", from
);
2019 read_next_command();
2023 static struct hash_list
*cmd_merge(unsigned int *count
)
2025 struct hash_list
*list
= NULL
, *n
, *e
= e
;
2030 while (!prefixcmp(command_buf
.buf
, "merge ")) {
2031 from
= strchr(command_buf
.buf
, ' ') + 1;
2032 n
= xmalloc(sizeof(*n
));
2033 s
= lookup_branch(from
);
2035 hashcpy(n
->sha1
, s
->sha1
);
2036 else if (*from
== ':') {
2037 uintmax_t idnum
= strtoumax(from
+ 1, NULL
, 10);
2038 struct object_entry
*oe
= find_mark(idnum
);
2039 if (oe
->type
!= OBJ_COMMIT
)
2040 die("Mark :%" PRIuMAX
" not a commit", idnum
);
2041 hashcpy(n
->sha1
, oe
->sha1
);
2042 } else if (!get_sha1(from
, n
->sha1
)) {
2044 char *buf
= read_object_with_reference(n
->sha1
,
2045 commit_type
, &size
, n
->sha1
);
2046 if (!buf
|| size
< 46)
2047 die("Not a valid commit: %s", from
);
2050 die("Invalid ref name or SHA1 expression: %s", from
);
2059 read_next_command();
2064 static void cmd_new_commit(void)
2070 char *author
= NULL
;
2071 char *committer
= NULL
;
2072 struct hash_list
*merge_list
= NULL
;
2073 unsigned int merge_count
;
2075 /* Obtain the branch name from the rest of our command */
2076 sp
= strchr(command_buf
.buf
, ' ') + 1;
2077 b
= lookup_branch(sp
);
2081 read_next_command();
2083 if (!prefixcmp(command_buf
.buf
, "author ")) {
2084 author
= parse_ident(command_buf
.buf
+ 7);
2085 read_next_command();
2087 if (!prefixcmp(command_buf
.buf
, "committer ")) {
2088 committer
= parse_ident(command_buf
.buf
+ 10);
2089 read_next_command();
2092 die("Expected committer but didn't get one");
2093 msg
= cmd_data(&msglen
);
2094 read_next_command();
2096 merge_list
= cmd_merge(&merge_count
);
2098 /* ensure the branch is active/loaded */
2099 if (!b
->branch_tree
.tree
|| !max_active_branches
) {
2100 unload_one_branch();
2105 while (!command_buf
.eof
&& command_buf
.len
> 1) {
2106 if (!prefixcmp(command_buf
.buf
, "M "))
2108 else if (!prefixcmp(command_buf
.buf
, "D "))
2110 else if (!prefixcmp(command_buf
.buf
, "R "))
2111 file_change_cr(b
, 1);
2112 else if (!prefixcmp(command_buf
.buf
, "C "))
2113 file_change_cr(b
, 0);
2114 else if (!strcmp("deleteall", command_buf
.buf
))
2115 file_change_deleteall(b
);
2117 unread_command_buf
= 1;
2120 read_next_command();
2123 /* build the tree and the commit */
2124 store_tree(&b
->branch_tree
);
2125 hashcpy(b
->branch_tree
.versions
[0].sha1
,
2126 b
->branch_tree
.versions
[1].sha1
);
2127 size_dbuf(&new_data
, 114 + msglen
2130 ? strlen(author
) + strlen(committer
)
2131 : 2 * strlen(committer
)));
2132 sp
= new_data
.buffer
;
2133 sp
+= sprintf(sp
, "tree %s\n",
2134 sha1_to_hex(b
->branch_tree
.versions
[1].sha1
));
2135 if (!is_null_sha1(b
->sha1
))
2136 sp
+= sprintf(sp
, "parent %s\n", sha1_to_hex(b
->sha1
));
2137 while (merge_list
) {
2138 struct hash_list
*next
= merge_list
->next
;
2139 sp
+= sprintf(sp
, "parent %s\n", sha1_to_hex(merge_list
->sha1
));
2143 sp
+= sprintf(sp
, "author %s\n", author
? author
: committer
);
2144 sp
+= sprintf(sp
, "committer %s\n", committer
);
2146 memcpy(sp
, msg
, msglen
);
2152 if (!store_object(OBJ_COMMIT
,
2153 new_data
.buffer
, sp
- (char*)new_data
.buffer
,
2154 NULL
, b
->sha1
, next_mark
))
2155 b
->pack_id
= pack_id
;
2156 b
->last_commit
= object_count_by_type
[OBJ_COMMIT
];
2159 static void cmd_new_tag(void)
2168 uintmax_t from_mark
= 0;
2169 unsigned char sha1
[20];
2171 /* Obtain the new tag name from the rest of our command */
2172 sp
= strchr(command_buf
.buf
, ' ') + 1;
2173 t
= pool_alloc(sizeof(struct tag
));
2175 t
->name
= pool_strdup(sp
);
2177 last_tag
->next_tag
= t
;
2181 read_next_command();
2184 if (prefixcmp(command_buf
.buf
, "from "))
2185 die("Expected from command, got %s", command_buf
.buf
);
2186 from
= strchr(command_buf
.buf
, ' ') + 1;
2187 s
= lookup_branch(from
);
2189 hashcpy(sha1
, s
->sha1
);
2190 } else if (*from
== ':') {
2191 struct object_entry
*oe
;
2192 from_mark
= strtoumax(from
+ 1, NULL
, 10);
2193 oe
= find_mark(from_mark
);
2194 if (oe
->type
!= OBJ_COMMIT
)
2195 die("Mark :%" PRIuMAX
" not a commit", from_mark
);
2196 hashcpy(sha1
, oe
->sha1
);
2197 } else if (!get_sha1(from
, sha1
)) {
2201 buf
= read_object_with_reference(sha1
,
2202 commit_type
, &size
, sha1
);
2203 if (!buf
|| size
< 46)
2204 die("Not a valid commit: %s", from
);
2207 die("Invalid ref name or SHA1 expression: %s", from
);
2208 read_next_command();
2211 if (prefixcmp(command_buf
.buf
, "tagger "))
2212 die("Expected tagger command, got %s", command_buf
.buf
);
2213 tagger
= parse_ident(command_buf
.buf
+ 7);
2215 /* tag payload/message */
2216 read_next_command();
2217 msg
= cmd_data(&msglen
);
2219 /* build the tag object */
2220 size_dbuf(&new_data
, 67+strlen(t
->name
)+strlen(tagger
)+msglen
);
2221 sp
= new_data
.buffer
;
2222 sp
+= sprintf(sp
, "object %s\n", sha1_to_hex(sha1
));
2223 sp
+= sprintf(sp
, "type %s\n", commit_type
);
2224 sp
+= sprintf(sp
, "tag %s\n", t
->name
);
2225 sp
+= sprintf(sp
, "tagger %s\n", tagger
);
2227 memcpy(sp
, msg
, msglen
);
2232 if (store_object(OBJ_TAG
, new_data
.buffer
,
2233 sp
- (char*)new_data
.buffer
,
2235 t
->pack_id
= MAX_PACK_ID
;
2237 t
->pack_id
= pack_id
;
2240 static void cmd_reset_branch(void)
2245 /* Obtain the branch name from the rest of our command */
2246 sp
= strchr(command_buf
.buf
, ' ') + 1;
2247 b
= lookup_branch(sp
);
2250 hashclr(b
->branch_tree
.versions
[0].sha1
);
2251 hashclr(b
->branch_tree
.versions
[1].sha1
);
2252 if (b
->branch_tree
.tree
) {
2253 release_tree_content_recursive(b
->branch_tree
.tree
);
2254 b
->branch_tree
.tree
= NULL
;
2259 read_next_command();
2260 if (!cmd_from(b
) && command_buf
.len
> 1)
2261 unread_command_buf
= 1;
2264 static void cmd_checkpoint(void)
2275 static void cmd_progress(void)
2277 fwrite(command_buf
.buf
, 1, command_buf
.len
- 1, stdout
);
2278 fputc('\n', stdout
);
2283 static void import_marks(const char *input_file
)
2286 FILE *f
= fopen(input_file
, "r");
2288 die("cannot read %s: %s", input_file
, strerror(errno
));
2289 while (fgets(line
, sizeof(line
), f
)) {
2292 unsigned char sha1
[20];
2293 struct object_entry
*e
;
2295 end
= strchr(line
, '\n');
2296 if (line
[0] != ':' || !end
)
2297 die("corrupt mark line: %s", line
);
2299 mark
= strtoumax(line
+ 1, &end
, 10);
2300 if (!mark
|| end
== line
+ 1
2301 || *end
!= ' ' || get_sha1(end
+ 1, sha1
))
2302 die("corrupt mark line: %s", line
);
2303 e
= find_object(sha1
);
2305 enum object_type type
= sha1_object_info(sha1
, NULL
);
2307 die("object not found: %s", sha1_to_hex(sha1
));
2308 e
= insert_object(sha1
);
2310 e
->pack_id
= MAX_PACK_ID
;
2311 e
->offset
= 1; /* just not zero! */
2313 insert_mark(mark
, e
);
2318 static const char fast_import_usage
[] =
2319 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
2321 int main(int argc
, const char **argv
)
2323 unsigned int i
, show_stats
= 1;
2325 git_config(git_default_config
);
2326 alloc_objects(object_entry_alloc
);
2327 strbuf_init(&command_buf
);
2328 atom_table
= xcalloc(atom_table_sz
, sizeof(struct atom_str
*));
2329 branch_table
= xcalloc(branch_table_sz
, sizeof(struct branch
*));
2330 avail_tree_table
= xcalloc(avail_tree_table_sz
, sizeof(struct avail_tree_content
*));
2331 marks
= pool_calloc(1, sizeof(struct mark_set
));
2333 for (i
= 1; i
< argc
; i
++) {
2334 const char *a
= argv
[i
];
2336 if (*a
!= '-' || !strcmp(a
, "--"))
2338 else if (!prefixcmp(a
, "--date-format=")) {
2339 const char *fmt
= a
+ 14;
2340 if (!strcmp(fmt
, "raw"))
2341 whenspec
= WHENSPEC_RAW
;
2342 else if (!strcmp(fmt
, "rfc2822"))
2343 whenspec
= WHENSPEC_RFC2822
;
2344 else if (!strcmp(fmt
, "now"))
2345 whenspec
= WHENSPEC_NOW
;
2347 die("unknown --date-format argument %s", fmt
);
2349 else if (!prefixcmp(a
, "--max-pack-size="))
2350 max_packsize
= strtoumax(a
+ 16, NULL
, 0) * 1024 * 1024;
2351 else if (!prefixcmp(a
, "--depth="))
2352 max_depth
= strtoul(a
+ 8, NULL
, 0);
2353 else if (!prefixcmp(a
, "--active-branches="))
2354 max_active_branches
= strtoul(a
+ 18, NULL
, 0);
2355 else if (!prefixcmp(a
, "--import-marks="))
2356 import_marks(a
+ 15);
2357 else if (!prefixcmp(a
, "--export-marks="))
2359 else if (!prefixcmp(a
, "--export-pack-edges=")) {
2362 pack_edges
= fopen(a
+ 20, "a");
2364 die("Cannot open %s: %s", a
+ 20, strerror(errno
));
2365 } else if (!strcmp(a
, "--force"))
2367 else if (!strcmp(a
, "--quiet"))
2369 else if (!strcmp(a
, "--stats"))
2372 die("unknown option %s", a
);
2375 usage(fast_import_usage
);
2377 rc_free
= pool_alloc(cmd_save
* sizeof(*rc_free
));
2378 for (i
= 0; i
< (cmd_save
- 1); i
++)
2379 rc_free
[i
].next
= &rc_free
[i
+ 1];
2380 rc_free
[cmd_save
- 1].next
= NULL
;
2382 prepare_packed_git();
2384 set_die_routine(die_nicely
);
2386 read_next_command();
2387 if (command_buf
.eof
)
2389 else if (!strcmp("blob", command_buf
.buf
))
2391 else if (!prefixcmp(command_buf
.buf
, "commit "))
2393 else if (!prefixcmp(command_buf
.buf
, "tag "))
2395 else if (!prefixcmp(command_buf
.buf
, "reset "))
2397 else if (!strcmp("checkpoint", command_buf
.buf
))
2399 else if (!prefixcmp(command_buf
.buf
, "progress "))
2402 die("Unsupported command: %s", command_buf
.buf
);
2415 uintmax_t total_count
= 0, duplicate_count
= 0;
2416 for (i
= 0; i
< ARRAY_SIZE(object_count_by_type
); i
++)
2417 total_count
+= object_count_by_type
[i
];
2418 for (i
= 0; i
< ARRAY_SIZE(duplicate_count_by_type
); i
++)
2419 duplicate_count
+= duplicate_count_by_type
[i
];
2421 fprintf(stderr
, "%s statistics:\n", argv
[0]);
2422 fprintf(stderr
, "---------------------------------------------------------------------\n");
2423 fprintf(stderr
, "Alloc'd objects: %10" PRIuMAX
"\n", alloc_count
);
2424 fprintf(stderr
, "Total objects: %10" PRIuMAX
" (%10" PRIuMAX
" duplicates )\n", total_count
, duplicate_count
);
2425 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
]);
2426 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
]);
2427 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
]);
2428 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
]);
2429 fprintf(stderr
, "Total branches: %10lu (%10lu loads )\n", branch_count
, branch_load_count
);
2430 fprintf(stderr
, " marks: %10" PRIuMAX
" (%10" PRIuMAX
" unique )\n", (((uintmax_t)1) << marks
->shift
) * 1024, marks_set_count
);
2431 fprintf(stderr
, " atoms: %10u\n", atom_cnt
);
2432 fprintf(stderr
, "Memory total: %10" PRIuMAX
" KiB\n", (total_allocd
+ alloc_count
*sizeof(struct object_entry
))/1024);
2433 fprintf(stderr
, " pools: %10lu KiB\n", (unsigned long)(total_allocd
/1024));
2434 fprintf(stderr
, " objects: %10" PRIuMAX
" KiB\n", (alloc_count
*sizeof(struct object_entry
))/1024);
2435 fprintf(stderr
, "---------------------------------------------------------------------\n");
2437 fprintf(stderr
, "---------------------------------------------------------------------\n");
2438 fprintf(stderr
, "\n");
2441 return failure
? 1 : 0;