?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
mingw: detect when MSYS2's sh is to be spawned more robustly
[alt-git.git] / fast-import.c
bloba84354524c1690b73302dff3e30bfe85a7e7ebc3
1 #include "builtin.h"
2 #include "cache.h"
3 #include "repository.h"
4 #include "config.h"
5 #include "lockfile.h"
6 #include "object.h"
7 #include "blob.h"
8 #include "tree.h"
9 #include "commit.h"
10 #include "delta.h"
11 #include "pack.h"
12 #include "refs.h"
13 #include "csum-file.h"
14 #include "quote.h"
15 #include "dir.h"
16 #include "run-command.h"
17 #include "packfile.h"
18 #include "object-store.h"
19 #include "mem-pool.h"
20 #include "commit-reach.h"
21
22 #define PACK_ID_BITS 16
23 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
24 #define DEPTH_BITS 13
25 #define MAX_DEPTH ((1<<DEPTH_BITS)-1)
26
27 /*
28 * We abuse the setuid bit on directories to mean "do not delta".
29 */
30 #define NO_DELTA S_ISUID
31
32 struct object_entry {
33 struct pack_idx_entry idx;
34 struct object_entry *next;
35 uint32_t type : TYPE_BITS,
36 pack_id : PACK_ID_BITS,
37 depth : DEPTH_BITS;
38 };
39
40 struct object_entry_pool {
41 struct object_entry_pool *next_pool;
42 struct object_entry *next_free;
43 struct object_entry *end;
44 struct object_entry entries[FLEX_ARRAY]; /* more */
45 };
46
47 struct mark_set {
48 union {
49 struct object_entry *marked[1024];
50 struct mark_set *sets[1024];
51 } data;
52 unsigned int shift;
53 };
54
55 struct last_object {
56 struct strbuf data;
57 off_t offset;
58 unsigned int depth;
59 unsigned no_swap : 1;
60 };
61
62 struct atom_str {
63 struct atom_str *next_atom;
64 unsigned short str_len;
65 char str_dat[FLEX_ARRAY]; /* more */
66 };
67
68 struct tree_content;
69 struct tree_entry {
70 struct tree_content *tree;
71 struct atom_str *name;
72 struct tree_entry_ms {
73 uint16_t mode;
74 struct object_id oid;
75 } versions[2];
76 };
77
78 struct tree_content {
79 unsigned int entry_capacity; /* must match avail_tree_content */
80 unsigned int entry_count;
81 unsigned int delta_depth;
82 struct tree_entry *entries[FLEX_ARRAY]; /* more */
83 };
84
85 struct avail_tree_content {
86 unsigned int entry_capacity; /* must match tree_content */
87 struct avail_tree_content *next_avail;
88 };
89
90 struct branch {
91 struct branch *table_next_branch;
92 struct branch *active_next_branch;
93 const char *name;
94 struct tree_entry branch_tree;
95 uintmax_t last_commit;
96 uintmax_t num_notes;
97 unsigned active : 1;
98 unsigned delete : 1;
99 unsigned pack_id : PACK_ID_BITS;
100 struct object_id oid;
101 };
102
103 struct tag {
104 struct tag *next_tag;
105 const char *name;
106 unsigned int pack_id;
107 struct object_id oid;
108 };
109
110 struct hash_list {
111 struct hash_list *next;
112 struct object_id oid;
113 };
114
115 typedef enum {
116 WHENSPEC_RAW = 1,
117 WHENSPEC_RFC2822,
118 WHENSPEC_NOW
119 } whenspec_type;
120
121 struct recent_command {
122 struct recent_command *prev;
123 struct recent_command *next;
124 char *buf;
125 };
126
127 /* Configured limits on output */
128 static unsigned long max_depth = 50;
129 static off_t max_packsize;
130 static int unpack_limit = 100;
131 static int force_update;
132
133 /* Stats and misc. counters */
134 static uintmax_t alloc_count;
135 static uintmax_t marks_set_count;
136 static uintmax_t object_count_by_type[1 << TYPE_BITS];
137 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
138 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
139 static uintmax_t delta_count_attempts_by_type[1 << TYPE_BITS];
140 static unsigned long object_count;
141 static unsigned long branch_count;
142 static unsigned long branch_load_count;
143 static int failure;
144 static FILE *pack_edges;
145 static unsigned int show_stats = 1;
146 static int global_argc;
147 static const char **global_argv;
148
149 /* Memory pools */
150 static struct mem_pool fi_mem_pool = {NULL, 2*1024*1024 -
151 sizeof(struct mp_block), 0 };
152
153 /* Atom management */
154 static unsigned int atom_table_sz = 4451;
155 static unsigned int atom_cnt;
156 static struct atom_str **atom_table;
157
158 /* The .pack file being generated */
159 static struct pack_idx_option pack_idx_opts;
160 static unsigned int pack_id;
161 static struct hashfile *pack_file;
162 static struct packed_git *pack_data;
163 static struct packed_git **all_packs;
164 static off_t pack_size;
165
166 /* Table of objects we've written. */
167 static unsigned int object_entry_alloc = 5000;
168 static struct object_entry_pool *blocks;
169 static struct object_entry *object_table[1 << 16];
170 static struct mark_set *marks;
171 static const char *export_marks_file;
172 static const char *import_marks_file;
173 static int import_marks_file_from_stream;
174 static int import_marks_file_ignore_missing;
175 static int import_marks_file_done;
176 static int relative_marks_paths;
177
178 /* Our last blob */
179 static struct last_object last_blob = { STRBUF_INIT, 0, 0, 0 };
180
181 /* Tree management */
182 static unsigned int tree_entry_alloc = 1000;
183 static void *avail_tree_entry;
184 static unsigned int avail_tree_table_sz = 100;
185 static struct avail_tree_content **avail_tree_table;
186 static size_t tree_entry_allocd;
187 static struct strbuf old_tree = STRBUF_INIT;
188 static struct strbuf new_tree = STRBUF_INIT;
189
190 /* Branch data */
191 static unsigned long max_active_branches = 5;
192 static unsigned long cur_active_branches;
193 static unsigned long branch_table_sz = 1039;
194 static struct branch **branch_table;
195 static struct branch *active_branches;
196
197 /* Tag data */
198 static struct tag *first_tag;
199 static struct tag *last_tag;
200
201 /* Input stream parsing */
202 static whenspec_type whenspec = WHENSPEC_RAW;
203 static struct strbuf command_buf = STRBUF_INIT;
204 static int unread_command_buf;
205 static struct recent_command cmd_hist = {&cmd_hist, &cmd_hist, NULL};
206 static struct recent_command *cmd_tail = &cmd_hist;
207 static struct recent_command *rc_free;
208 static unsigned int cmd_save = 100;
209 static uintmax_t next_mark;
210 static struct strbuf new_data = STRBUF_INIT;
211 static int seen_data_command;
212 static int require_explicit_termination;
213 static int allow_unsafe_features;
214
215 /* Signal handling */
216 static volatile sig_atomic_t checkpoint_requested;
217
218 /* Where to write output of cat-blob commands */
219 static int cat_blob_fd = STDOUT_FILENO;
220
221 static void parse_argv(void);
222 static void parse_get_mark(const char *p);
223 static void parse_cat_blob(const char *p);
224 static void parse_ls(const char *p, struct branch *b);
225
226 static void write_branch_report(FILE *rpt, struct branch *b)
227 {
228 fprintf(rpt, "%s:\n", b->name);
229
230 fprintf(rpt, " status :");
231 if (b->active)
232 fputs(" active", rpt);
233 if (b->branch_tree.tree)
234 fputs(" loaded", rpt);
235 if (is_null_oid(&b->branch_tree.versions[1].oid))
236 fputs(" dirty", rpt);
237 fputc('\n', rpt);
238
239 fprintf(rpt, " tip commit : %s\n", oid_to_hex(&b->oid));
240 fprintf(rpt, " old tree : %s\n",
241 oid_to_hex(&b->branch_tree.versions[0].oid));
242 fprintf(rpt, " cur tree : %s\n",
243 oid_to_hex(&b->branch_tree.versions[1].oid));
244 fprintf(rpt, " commit clock: %" PRIuMAX "\n", b->last_commit);
245
246 fputs(" last pack : ", rpt);
247 if (b->pack_id < MAX_PACK_ID)
248 fprintf(rpt, "%u", b->pack_id);
249 fputc('\n', rpt);
250
251 fputc('\n', rpt);
252 }
253
254 static void dump_marks_helper(FILE *, uintmax_t, struct mark_set *);
255
256 static void write_crash_report(const char *err)
257 {
258 char *loc = git_pathdup("fast_import_crash_%"PRIuMAX, (uintmax_t) getpid());
259 FILE *rpt = fopen(loc, "w");
260 struct branch *b;
261 unsigned long lu;
262 struct recent_command *rc;
263
264 if (!rpt) {
265 error_errno("can't write crash report %s", loc);
266 free(loc);
267 return;
268 }
269
270 fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
271
272 fprintf(rpt, "fast-import crash report:\n");
273 fprintf(rpt, " fast-import process: %"PRIuMAX"\n", (uintmax_t) getpid());
274 fprintf(rpt, " parent process : %"PRIuMAX"\n", (uintmax_t) getppid());
275 fprintf(rpt, " at %s\n", show_date(time(NULL), 0, DATE_MODE(ISO8601)));
276 fputc('\n', rpt);
277
278 fputs("fatal: ", rpt);
279 fputs(err, rpt);
280 fputc('\n', rpt);
281
282 fputc('\n', rpt);
283 fputs("Most Recent Commands Before Crash\n", rpt);
284 fputs("---------------------------------\n", rpt);
285 for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
286 if (rc->next == &cmd_hist)
287 fputs("* ", rpt);
288 else
289 fputs(" ", rpt);
290 fputs(rc->buf, rpt);
291 fputc('\n', rpt);
292 }
293
294 fputc('\n', rpt);
295 fputs("Active Branch LRU\n", rpt);
296 fputs("-----------------\n", rpt);
297 fprintf(rpt, " active_branches = %lu cur, %lu max\n",
298 cur_active_branches,
299 max_active_branches);
300 fputc('\n', rpt);
301 fputs(" pos clock name\n", rpt);
302 fputs(" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
303 for (b = active_branches, lu = 0; b; b = b->active_next_branch)
304 fprintf(rpt, " %2lu) %6" PRIuMAX" %s\n",
305 ++lu, b->last_commit, b->name);
306
307 fputc('\n', rpt);
308 fputs("Inactive Branches\n", rpt);
309 fputs("-----------------\n", rpt);
310 for (lu = 0; lu < branch_table_sz; lu++) {
311 for (b = branch_table[lu]; b; b = b->table_next_branch)
312 write_branch_report(rpt, b);
313 }
314
315 if (first_tag) {
316 struct tag *tg;
317 fputc('\n', rpt);
318 fputs("Annotated Tags\n", rpt);
319 fputs("--------------\n", rpt);
320 for (tg = first_tag; tg; tg = tg->next_tag) {
321 fputs(oid_to_hex(&tg->oid), rpt);
322 fputc(' ', rpt);
323 fputs(tg->name, rpt);
324 fputc('\n', rpt);
325 }
326 }
327
328 fputc('\n', rpt);
329 fputs("Marks\n", rpt);
330 fputs("-----\n", rpt);
331 if (export_marks_file)
332 fprintf(rpt, " exported to %s\n", export_marks_file);
333 else
334 dump_marks_helper(rpt, 0, marks);
335
336 fputc('\n', rpt);
337 fputs("-------------------\n", rpt);
338 fputs("END OF CRASH REPORT\n", rpt);
339 fclose(rpt);
340 free(loc);
341 }
342
343 static void end_packfile(void);
344 static void unkeep_all_packs(void);
345 static void dump_marks(void);
346
347 static NORETURN void die_nicely(const char *err, va_list params)
348 {
349 static int zombie;
350 char message[2 * PATH_MAX];
351
352 vsnprintf(message, sizeof(message), err, params);
353 fputs("fatal: ", stderr);
354 fputs(message, stderr);
355 fputc('\n', stderr);
356
357 if (!zombie) {
358 zombie = 1;
359 write_crash_report(message);
360 end_packfile();
361 unkeep_all_packs();
362 dump_marks();
363 }
364 exit(128);
365 }
366
367 #ifndef SIGUSR1 /* Windows, for example */
368
369 static void set_checkpoint_signal(void)
370 {
371 }
372
373 #else
374
375 static void checkpoint_signal(int signo)
376 {
377 checkpoint_requested = 1;
378 }
379
380 static void set_checkpoint_signal(void)
381 {
382 struct sigaction sa;
383
384 memset(&sa, 0, sizeof(sa));
385 sa.sa_handler = checkpoint_signal;
386 sigemptyset(&sa.sa_mask);
387 sa.sa_flags = SA_RESTART;
388 sigaction(SIGUSR1, &sa, NULL);
389 }
390
391 #endif
392
393 static void alloc_objects(unsigned int cnt)
394 {
395 struct object_entry_pool *b;
396
397 b = xmalloc(sizeof(struct object_entry_pool)
398 + cnt * sizeof(struct object_entry));
399 b->next_pool = blocks;
400 b->next_free = b->entries;
401 b->end = b->entries + cnt;
402 blocks = b;
403 alloc_count += cnt;
404 }
405
406 static struct object_entry *new_object(struct object_id *oid)
407 {
408 struct object_entry *e;
409
410 if (blocks->next_free == blocks->end)
411 alloc_objects(object_entry_alloc);
412
413 e = blocks->next_free++;
414 oidcpy(&e->idx.oid, oid);
415 return e;
416 }
417
418 static struct object_entry *find_object(struct object_id *oid)
419 {
420 unsigned int h = oid->hash[0] << 8 | oid->hash[1];
421 struct object_entry *e;
422 for (e = object_table[h]; e; e = e->next)
423 if (oideq(oid, &e->idx.oid))
424 return e;
425 return NULL;
426 }
427
428 static struct object_entry *insert_object(struct object_id *oid)
429 {
430 unsigned int h = oid->hash[0] << 8 | oid->hash[1];
431 struct object_entry *e = object_table[h];
432
433 while (e) {
434 if (oideq(oid, &e->idx.oid))
435 return e;
436 e = e->next;
437 }
438
439 e = new_object(oid);
440 e->next = object_table[h];
441 e->idx.offset = 0;
442 object_table[h] = e;
443 return e;
444 }
445
446 static void invalidate_pack_id(unsigned int id)
447 {
448 unsigned int h;
449 unsigned long lu;
450 struct tag *t;
451
452 for (h = 0; h < ARRAY_SIZE(object_table); h++) {
453 struct object_entry *e;
454
455 for (e = object_table[h]; e; e = e->next)
456 if (e->pack_id == id)
457 e->pack_id = MAX_PACK_ID;
458 }
459
460 for (lu = 0; lu < branch_table_sz; lu++) {
461 struct branch *b;
462
463 for (b = branch_table[lu]; b; b = b->table_next_branch)
464 if (b->pack_id == id)
465 b->pack_id = MAX_PACK_ID;
466 }
467
468 for (t = first_tag; t; t = t->next_tag)
469 if (t->pack_id == id)
470 t->pack_id = MAX_PACK_ID;
471 }
472
473 static unsigned int hc_str(const char *s, size_t len)
474 {
475 unsigned int r = 0;
476 while (len-- > 0)
477 r = r * 31 + *s++;
478 return r;
479 }
480
481 static char *pool_strdup(const char *s)
482 {
483 size_t len = strlen(s) + 1;
484 char *r = mem_pool_alloc(&fi_mem_pool, len);
485 memcpy(r, s, len);
486 return r;
487 }
488
489 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
490 {
491 struct mark_set *s = marks;
492 while ((idnum >> s->shift) >= 1024) {
493 s = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
494 s->shift = marks->shift + 10;
495 s->data.sets[0] = marks;
496 marks = s;
497 }
498 while (s->shift) {
499 uintmax_t i = idnum >> s->shift;
500 idnum -= i << s->shift;
501 if (!s->data.sets[i]) {
502 s->data.sets[i] = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
503 s->data.sets[i]->shift = s->shift - 10;
504 }
505 s = s->data.sets[i];
506 }
507 if (!s->data.marked[idnum])
508 marks_set_count++;
509 s->data.marked[idnum] = oe;
510 }
511
512 static struct object_entry *find_mark(uintmax_t idnum)
513 {
514 uintmax_t orig_idnum = idnum;
515 struct mark_set *s = marks;
516 struct object_entry *oe = NULL;
517 if ((idnum >> s->shift) < 1024) {
518 while (s && s->shift) {
519 uintmax_t i = idnum >> s->shift;
520 idnum -= i << s->shift;
521 s = s->data.sets[i];
522 }
523 if (s)
524 oe = s->data.marked[idnum];
525 }
526 if (!oe)
527 die("mark :%" PRIuMAX " not declared", orig_idnum);
528 return oe;
529 }
530
531 static struct atom_str *to_atom(const char *s, unsigned short len)
532 {
533 unsigned int hc = hc_str(s, len) % atom_table_sz;
534 struct atom_str *c;
535
536 for (c = atom_table[hc]; c; c = c->next_atom)
537 if (c->str_len == len && !strncmp(s, c->str_dat, len))
538 return c;
539
540 c = mem_pool_alloc(&fi_mem_pool, sizeof(struct atom_str) + len + 1);
541 c->str_len = len;
542 memcpy(c->str_dat, s, len);
543 c->str_dat[len] = 0;
544 c->next_atom = atom_table[hc];
545 atom_table[hc] = c;
546 atom_cnt++;
547 return c;
548 }
549
550 static struct branch *lookup_branch(const char *name)
551 {
552 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
553 struct branch *b;
554
555 for (b = branch_table[hc]; b; b = b->table_next_branch)
556 if (!strcmp(name, b->name))
557 return b;
558 return NULL;
559 }
560
561 static struct branch *new_branch(const char *name)
562 {
563 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
564 struct branch *b = lookup_branch(name);
565
566 if (b)
567 die("Invalid attempt to create duplicate branch: %s", name);
568 if (check_refname_format(name, REFNAME_ALLOW_ONELEVEL))
569 die("Branch name doesn't conform to GIT standards: %s", name);
570
571 b = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct branch));
572 b->name = pool_strdup(name);
573 b->table_next_branch = branch_table[hc];
574 b->branch_tree.versions[0].mode = S_IFDIR;
575 b->branch_tree.versions[1].mode = S_IFDIR;
576 b->num_notes = 0;
577 b->active = 0;
578 b->pack_id = MAX_PACK_ID;
579 branch_table[hc] = b;
580 branch_count++;
581 return b;
582 }
583
584 static unsigned int hc_entries(unsigned int cnt)
585 {
586 cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
587 return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
588 }
589
590 static struct tree_content *new_tree_content(unsigned int cnt)
591 {
592 struct avail_tree_content *f, *l = NULL;
593 struct tree_content *t;
594 unsigned int hc = hc_entries(cnt);
595
596 for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
597 if (f->entry_capacity >= cnt)
598 break;
599
600 if (f) {
601 if (l)
602 l->next_avail = f->next_avail;
603 else
604 avail_tree_table[hc] = f->next_avail;
605 } else {
606 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
607 f = mem_pool_alloc(&fi_mem_pool, sizeof(*t) + sizeof(t->entries[0]) * cnt);
608 f->entry_capacity = cnt;
609 }
610
611 t = (struct tree_content*)f;
612 t->entry_count = 0;
613 t->delta_depth = 0;
614 return t;
615 }
616
617 static void release_tree_entry(struct tree_entry *e);
618 static void release_tree_content(struct tree_content *t)
619 {
620 struct avail_tree_content *f = (struct avail_tree_content*)t;
621 unsigned int hc = hc_entries(f->entry_capacity);
622 f->next_avail = avail_tree_table[hc];
623 avail_tree_table[hc] = f;
624 }
625
626 static void release_tree_content_recursive(struct tree_content *t)
627 {
628 unsigned int i;
629 for (i = 0; i < t->entry_count; i++)
630 release_tree_entry(t->entries[i]);
631 release_tree_content(t);
632 }
633
634 static struct tree_content *grow_tree_content(
635 struct tree_content *t,
636 int amt)
637 {
638 struct tree_content *r = new_tree_content(t->entry_count + amt);
639 r->entry_count = t->entry_count;
640 r->delta_depth = t->delta_depth;
641 memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
642 release_tree_content(t);
643 return r;
644 }
645
646 static struct tree_entry *new_tree_entry(void)
647 {
648 struct tree_entry *e;
649
650 if (!avail_tree_entry) {
651 unsigned int n = tree_entry_alloc;
652 tree_entry_allocd += n * sizeof(struct tree_entry);
653 ALLOC_ARRAY(e, n);
654 avail_tree_entry = e;
655 while (n-- > 1) {
656 *((void**)e) = e + 1;
657 e++;
658 }
659 *((void**)e) = NULL;
660 }
661
662 e = avail_tree_entry;
663 avail_tree_entry = *((void**)e);
664 return e;
665 }
666
667 static void release_tree_entry(struct tree_entry *e)
668 {
669 if (e->tree)
670 release_tree_content_recursive(e->tree);
671 *((void**)e) = avail_tree_entry;
672 avail_tree_entry = e;
673 }
674
675 static struct tree_content *dup_tree_content(struct tree_content *s)
676 {
677 struct tree_content *d;
678 struct tree_entry *a, *b;
679 unsigned int i;
680
681 if (!s)
682 return NULL;
683 d = new_tree_content(s->entry_count);
684 for (i = 0; i < s->entry_count; i++) {
685 a = s->entries[i];
686 b = new_tree_entry();
687 memcpy(b, a, sizeof(*a));
688 if (a->tree && is_null_oid(&b->versions[1].oid))
689 b->tree = dup_tree_content(a->tree);
690 else
691 b->tree = NULL;
692 d->entries[i] = b;
693 }
694 d->entry_count = s->entry_count;
695 d->delta_depth = s->delta_depth;
696
697 return d;
698 }
699
700 static void start_packfile(void)
701 {
702 struct strbuf tmp_file = STRBUF_INIT;
703 struct packed_git *p;
704 struct pack_header hdr;
705 int pack_fd;
706
707 pack_fd = odb_mkstemp(&tmp_file, "pack/tmp_pack_XXXXXX");
708 FLEX_ALLOC_STR(p, pack_name, tmp_file.buf);
709 strbuf_release(&tmp_file);
710
711 p->pack_fd = pack_fd;
712 p->do_not_close = 1;
713 pack_file = hashfd(pack_fd, p->pack_name);
714
715 hdr.hdr_signature = htonl(PACK_SIGNATURE);
716 hdr.hdr_version = htonl(2);
717 hdr.hdr_entries = 0;
718 hashwrite(pack_file, &hdr, sizeof(hdr));
719
720 pack_data = p;
721 pack_size = sizeof(hdr);
722 object_count = 0;
723
724 REALLOC_ARRAY(all_packs, pack_id + 1);
725 all_packs[pack_id] = p;
726 }
727
728 static const char *create_index(void)
729 {
730 const char *tmpfile;
731 struct pack_idx_entry **idx, **c, **last;
732 struct object_entry *e;
733 struct object_entry_pool *o;
734
735 /* Build the table of object IDs. */
736 ALLOC_ARRAY(idx, object_count);
737 c = idx;
738 for (o = blocks; o; o = o->next_pool)
739 for (e = o->next_free; e-- != o->entries;)
740 if (pack_id == e->pack_id)
741 *c++ = &e->idx;
742 last = idx + object_count;
743 if (c != last)
744 die("internal consistency error creating the index");
745
746 tmpfile = write_idx_file(NULL, idx, object_count, &pack_idx_opts, pack_data->sha1);
747 free(idx);
748 return tmpfile;
749 }
750
751 static char *keep_pack(const char *curr_index_name)
752 {
753 static const char *keep_msg = "fast-import";
754 struct strbuf name = STRBUF_INIT;
755 int keep_fd;
756
757 odb_pack_name(&name, pack_data->sha1, "keep");
758 keep_fd = odb_pack_keep(name.buf);
759 if (keep_fd < 0)
760 die_errno("cannot create keep file");
761 write_or_die(keep_fd, keep_msg, strlen(keep_msg));
762 if (close(keep_fd))
763 die_errno("failed to write keep file");
764
765 odb_pack_name(&name, pack_data->sha1, "pack");
766 if (finalize_object_file(pack_data->pack_name, name.buf))
767 die("cannot store pack file");
768
769 odb_pack_name(&name, pack_data->sha1, "idx");
770 if (finalize_object_file(curr_index_name, name.buf))
771 die("cannot store index file");
772 free((void *)curr_index_name);
773 return strbuf_detach(&name, NULL);
774 }
775
776 static void unkeep_all_packs(void)
777 {
778 struct strbuf name = STRBUF_INIT;
779 int k;
780
781 for (k = 0; k < pack_id; k++) {
782 struct packed_git *p = all_packs[k];
783 odb_pack_name(&name, p->sha1, "keep");
784 unlink_or_warn(name.buf);
785 }
786 strbuf_release(&name);
787 }
788
789 static int loosen_small_pack(const struct packed_git *p)
790 {
791 struct child_process unpack = CHILD_PROCESS_INIT;
792
793 if (lseek(p->pack_fd, 0, SEEK_SET) < 0)
794 die_errno("Failed seeking to start of '%s'", p->pack_name);
795
796 unpack.in = p->pack_fd;
797 unpack.git_cmd = 1;
798 unpack.stdout_to_stderr = 1;
799 argv_array_push(&unpack.args, "unpack-objects");
800 if (!show_stats)
801 argv_array_push(&unpack.args, "-q");
802
803 return run_command(&unpack);
804 }
805
806 static void end_packfile(void)
807 {
808 static int running;
809
810 if (running || !pack_data)
811 return;
812
813 running = 1;
814 clear_delta_base_cache();
815 if (object_count) {
816 struct packed_git *new_p;
817 struct object_id cur_pack_oid;
818 char *idx_name;
819 int i;
820 struct branch *b;
821 struct tag *t;
822
823 close_pack_windows(pack_data);
824 finalize_hashfile(pack_file, cur_pack_oid.hash, 0);
825 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
826 pack_data->pack_name, object_count,
827 cur_pack_oid.hash, pack_size);
828
829 if (object_count <= unpack_limit) {
830 if (!loosen_small_pack(pack_data)) {
831 invalidate_pack_id(pack_id);
832 goto discard_pack;
833 }
834 }
835
836 close(pack_data->pack_fd);
837 idx_name = keep_pack(create_index());
838
839 /* Register the packfile with core git's machinery. */
840 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
841 if (!new_p)
842 die("core git rejected index %s", idx_name);
843 all_packs[pack_id] = new_p;
844 install_packed_git(the_repository, new_p);
845 free(idx_name);
846
847 /* Print the boundary */
848 if (pack_edges) {
849 fprintf(pack_edges, "%s:", new_p->pack_name);
850 for (i = 0; i < branch_table_sz; i++) {
851 for (b = branch_table[i]; b; b = b->table_next_branch) {
852 if (b->pack_id == pack_id)
853 fprintf(pack_edges, " %s",
854 oid_to_hex(&b->oid));
855 }
856 }
857 for (t = first_tag; t; t = t->next_tag) {
858 if (t->pack_id == pack_id)
859 fprintf(pack_edges, " %s",
860 oid_to_hex(&t->oid));
861 }
862 fputc('\n', pack_edges);
863 fflush(pack_edges);
864 }
865
866 pack_id++;
867 }
868 else {
869 discard_pack:
870 close(pack_data->pack_fd);
871 unlink_or_warn(pack_data->pack_name);
872 }
873 FREE_AND_NULL(pack_data);
874 running = 0;
875
876 /* We can't carry a delta across packfiles. */
877 strbuf_release(&last_blob.data);
878 last_blob.offset = 0;
879 last_blob.depth = 0;
880 }
881
882 static void cycle_packfile(void)
883 {
884 end_packfile();
885 start_packfile();
886 }
887
888 static int store_object(
889 enum object_type type,
890 struct strbuf *dat,
891 struct last_object *last,
892 struct object_id *oidout,
893 uintmax_t mark)
894 {
895 void *out, *delta;
896 struct object_entry *e;
897 unsigned char hdr[96];
898 struct object_id oid;
899 unsigned long hdrlen, deltalen;
900 git_hash_ctx c;
901 git_zstream s;
902
903 hdrlen = xsnprintf((char *)hdr, sizeof(hdr), "%s %lu",
904 type_name(type), (unsigned long)dat->len) + 1;
905 the_hash_algo->init_fn(&c);
906 the_hash_algo->update_fn(&c, hdr, hdrlen);
907 the_hash_algo->update_fn(&c, dat->buf, dat->len);
908 the_hash_algo->final_fn(oid.hash, &c);
909 if (oidout)
910 oidcpy(oidout, &oid);
911
912 e = insert_object(&oid);
913 if (mark)
914 insert_mark(mark, e);
915 if (e->idx.offset) {
916 duplicate_count_by_type[type]++;
917 return 1;
918 } else if (find_sha1_pack(oid.hash,
919 get_all_packs(the_repository))) {
920 e->type = type;
921 e->pack_id = MAX_PACK_ID;
922 e->idx.offset = 1; /* just not zero! */
923 duplicate_count_by_type[type]++;
924 return 1;
925 }
926
927 if (last && last->data.len && last->data.buf && last->depth < max_depth
928 && dat->len > the_hash_algo->rawsz) {
929
930 delta_count_attempts_by_type[type]++;
931 delta = diff_delta(last->data.buf, last->data.len,
932 dat->buf, dat->len,
933 &deltalen, dat->len - the_hash_algo->rawsz);
934 } else
935 delta = NULL;
936
937 git_deflate_init(&s, pack_compression_level);
938 if (delta) {
939 s.next_in = delta;
940 s.avail_in = deltalen;
941 } else {
942 s.next_in = (void *)dat->buf;
943 s.avail_in = dat->len;
944 }
945 s.avail_out = git_deflate_bound(&s, s.avail_in);
946 s.next_out = out = xmalloc(s.avail_out);
947 while (git_deflate(&s, Z_FINISH) == Z_OK)
948 ; /* nothing */
949 git_deflate_end(&s);
950
951 /* Determine if we should auto-checkpoint. */
952 if ((max_packsize && (pack_size + 60 + s.total_out) > max_packsize)
953 || (pack_size + 60 + s.total_out) < pack_size) {
954
955 /* This new object needs to *not* have the current pack_id. */
956 e->pack_id = pack_id + 1;
957 cycle_packfile();
958
959 /* We cannot carry a delta into the new pack. */
960 if (delta) {
961 FREE_AND_NULL(delta);
962
963 git_deflate_init(&s, pack_compression_level);
964 s.next_in = (void *)dat->buf;
965 s.avail_in = dat->len;
966 s.avail_out = git_deflate_bound(&s, s.avail_in);
967 s.next_out = out = xrealloc(out, s.avail_out);
968 while (git_deflate(&s, Z_FINISH) == Z_OK)
969 ; /* nothing */
970 git_deflate_end(&s);
971 }
972 }
973
974 e->type = type;
975 e->pack_id = pack_id;
976 e->idx.offset = pack_size;
977 object_count++;
978 object_count_by_type[type]++;
979
980 crc32_begin(pack_file);
981
982 if (delta) {
983 off_t ofs = e->idx.offset - last->offset;
984 unsigned pos = sizeof(hdr) - 1;
985
986 delta_count_by_type[type]++;
987 e->depth = last->depth + 1;
988
989 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
990 OBJ_OFS_DELTA, deltalen);
991 hashwrite(pack_file, hdr, hdrlen);
992 pack_size += hdrlen;
993
994 hdr[pos] = ofs & 127;
995 while (ofs >>= 7)
996 hdr[--pos] = 128 | (--ofs & 127);
997 hashwrite(pack_file, hdr + pos, sizeof(hdr) - pos);
998 pack_size += sizeof(hdr) - pos;
999 } else {
1000 e->depth = 0;
1001 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
1002 type, dat->len);
1003 hashwrite(pack_file, hdr, hdrlen);
1004 pack_size += hdrlen;
1005 }
1006
1007 hashwrite(pack_file, out, s.total_out);
1008 pack_size += s.total_out;
1009
1010 e->idx.crc32 = crc32_end(pack_file);
1011
1012 free(out);
1013 free(delta);
1014 if (last) {
1015 if (last->no_swap) {
1016 last->data = *dat;
1017 } else {
1018 strbuf_swap(&last->data, dat);
1019 }
1020 last->offset = e->idx.offset;
1021 last->depth = e->depth;
1022 }
1023 return 0;
1024 }
1025
1026 static void truncate_pack(struct hashfile_checkpoint *checkpoint)
1027 {
1028 if (hashfile_truncate(pack_file, checkpoint))
1029 die_errno("cannot truncate pack to skip duplicate");
1030 pack_size = checkpoint->offset;
1031 }
1032
1033 static void stream_blob(uintmax_t len, struct object_id *oidout, uintmax_t mark)
1034 {
1035 size_t in_sz = 64 * 1024, out_sz = 64 * 1024;
1036 unsigned char *in_buf = xmalloc(in_sz);
1037 unsigned char *out_buf = xmalloc(out_sz);
1038 struct object_entry *e;
1039 struct object_id oid;
1040 unsigned long hdrlen;
1041 off_t offset;
1042 git_hash_ctx c;
1043 git_zstream s;
1044 struct hashfile_checkpoint checkpoint;
1045 int status = Z_OK;
1046
1047 /* Determine if we should auto-checkpoint. */
1048 if ((max_packsize && (pack_size + 60 + len) > max_packsize)
1049 || (pack_size + 60 + len) < pack_size)
1050 cycle_packfile();
1051
1052 hashfile_checkpoint(pack_file, &checkpoint);
1053 offset = checkpoint.offset;
1054
1055 hdrlen = xsnprintf((char *)out_buf, out_sz, "blob %" PRIuMAX, len) + 1;
1056
1057 the_hash_algo->init_fn(&c);
1058 the_hash_algo->update_fn(&c, out_buf, hdrlen);
1059
1060 crc32_begin(pack_file);
1061
1062 git_deflate_init(&s, pack_compression_level);
1063
1064 hdrlen = encode_in_pack_object_header(out_buf, out_sz, OBJ_BLOB, len);
1065
1066 s.next_out = out_buf + hdrlen;
1067 s.avail_out = out_sz - hdrlen;
1068
1069 while (status != Z_STREAM_END) {
1070 if (0 < len && !s.avail_in) {
1071 size_t cnt = in_sz < len ? in_sz : (size_t)len;
1072 size_t n = fread(in_buf, 1, cnt, stdin);
1073 if (!n && feof(stdin))
1074 die("EOF in data (%" PRIuMAX " bytes remaining)", len);
1075
1076 the_hash_algo->update_fn(&c, in_buf, n);
1077 s.next_in = in_buf;
1078 s.avail_in = n;
1079 len -= n;
1080 }
1081
1082 status = git_deflate(&s, len ? 0 : Z_FINISH);
1083
1084 if (!s.avail_out || status == Z_STREAM_END) {
1085 size_t n = s.next_out - out_buf;
1086 hashwrite(pack_file, out_buf, n);
1087 pack_size += n;
1088 s.next_out = out_buf;
1089 s.avail_out = out_sz;
1090 }
1091
1092 switch (status) {
1093 case Z_OK:
1094 case Z_BUF_ERROR:
1095 case Z_STREAM_END:
1096 continue;
1097 default:
1098 die("unexpected deflate failure: %d", status);
1099 }
1100 }
1101 git_deflate_end(&s);
1102 the_hash_algo->final_fn(oid.hash, &c);
1103
1104 if (oidout)
1105 oidcpy(oidout, &oid);
1106
1107 e = insert_object(&oid);
1108
1109 if (mark)
1110 insert_mark(mark, e);
1111
1112 if (e->idx.offset) {
1113 duplicate_count_by_type[OBJ_BLOB]++;
1114 truncate_pack(&checkpoint);
1115
1116 } else if (find_sha1_pack(oid.hash,
1117 get_all_packs(the_repository))) {
1118 e->type = OBJ_BLOB;
1119 e->pack_id = MAX_PACK_ID;
1120 e->idx.offset = 1; /* just not zero! */
1121 duplicate_count_by_type[OBJ_BLOB]++;
1122 truncate_pack(&checkpoint);
1123
1124 } else {
1125 e->depth = 0;
1126 e->type = OBJ_BLOB;
1127 e->pack_id = pack_id;
1128 e->idx.offset = offset;
1129 e->idx.crc32 = crc32_end(pack_file);
1130 object_count++;
1131 object_count_by_type[OBJ_BLOB]++;
1132 }
1133
1134 free(in_buf);
1135 free(out_buf);
1136 }
1137
1138 /* All calls must be guarded by find_object() or find_mark() to
1139 * ensure the 'struct object_entry' passed was written by this
1140 * process instance. We unpack the entry by the offset, avoiding
1141 * the need for the corresponding .idx file. This unpacking rule
1142 * works because we only use OBJ_REF_DELTA within the packfiles
1143 * created by fast-import.
1144 *
1145 * oe must not be NULL. Such an oe usually comes from giving
1146 * an unknown SHA-1 to find_object() or an undefined mark to
1147 * find_mark(). Callers must test for this condition and use
1148 * the standard read_sha1_file() when it happens.
1149 *
1150 * oe->pack_id must not be MAX_PACK_ID. Such an oe is usually from
1151 * find_mark(), where the mark was reloaded from an existing marks
1152 * file and is referencing an object that this fast-import process
1153 * instance did not write out to a packfile. Callers must test for
1154 * this condition and use read_sha1_file() instead.
1155 */
1156 static void *gfi_unpack_entry(
1157 struct object_entry *oe,
1158 unsigned long *sizep)
1159 {
1160 enum object_type type;
1161 struct packed_git *p = all_packs[oe->pack_id];
1162 if (p == pack_data && p->pack_size < (pack_size + the_hash_algo->rawsz)) {
1163 /* The object is stored in the packfile we are writing to
1164 * and we have modified it since the last time we scanned
1165 * back to read a previously written object. If an old
1166 * window covered [p->pack_size, p->pack_size + rawsz) its
1167 * data is stale and is not valid. Closing all windows
1168 * and updating the packfile length ensures we can read
1169 * the newly written data.
1170 */
1171 close_pack_windows(p);
1172 hashflush(pack_file);
1173
1174 /* We have to offer rawsz bytes additional on the end of
1175 * the packfile as the core unpacker code assumes the
1176 * footer is present at the file end and must promise
1177 * at least rawsz bytes within any window it maps. But
1178 * we don't actually create the footer here.
1179 */
1180 p->pack_size = pack_size + the_hash_algo->rawsz;
1181 }
1182 return unpack_entry(the_repository, p, oe->idx.offset, &type, sizep);
1183 }
1184
1185 static const char *get_mode(const char *str, uint16_t *modep)
1186 {
1187 unsigned char c;
1188 uint16_t mode = 0;
1189
1190 while ((c = *str++) != ' ') {
1191 if (c < '0' || c > '7')
1192 return NULL;
1193 mode = (mode << 3) + (c - '0');
1194 }
1195 *modep = mode;
1196 return str;
1197 }
1198
1199 static void load_tree(struct tree_entry *root)
1200 {
1201 struct object_id *oid = &root->versions[1].oid;
1202 struct object_entry *myoe;
1203 struct tree_content *t;
1204 unsigned long size;
1205 char *buf;
1206 const char *c;
1207
1208 root->tree = t = new_tree_content(8);
1209 if (is_null_oid(oid))
1210 return;
1211
1212 myoe = find_object(oid);
1213 if (myoe && myoe->pack_id != MAX_PACK_ID) {
1214 if (myoe->type != OBJ_TREE)
1215 die("Not a tree: %s", oid_to_hex(oid));
1216 t->delta_depth = myoe->depth;
1217 buf = gfi_unpack_entry(myoe, &size);
1218 if (!buf)
1219 die("Can't load tree %s", oid_to_hex(oid));
1220 } else {
1221 enum object_type type;
1222 buf = read_object_file(oid, &type, &size);
1223 if (!buf || type != OBJ_TREE)
1224 die("Can't load tree %s", oid_to_hex(oid));
1225 }
1226
1227 c = buf;
1228 while (c != (buf + size)) {
1229 struct tree_entry *e = new_tree_entry();
1230
1231 if (t->entry_count == t->entry_capacity)
1232 root->tree = t = grow_tree_content(t, t->entry_count);
1233 t->entries[t->entry_count++] = e;
1234
1235 e->tree = NULL;
1236 c = get_mode(c, &e->versions[1].mode);
1237 if (!c)
1238 die("Corrupt mode in %s", oid_to_hex(oid));
1239 e->versions[0].mode = e->versions[1].mode;
1240 e->name = to_atom(c, strlen(c));
1241 c += e->name->str_len + 1;
1242 hashcpy(e->versions[0].oid.hash, (unsigned char *)c);
1243 hashcpy(e->versions[1].oid.hash, (unsigned char *)c);
1244 c += GIT_SHA1_RAWSZ;
1245 }
1246 free(buf);
1247 }
1248
1249 static int tecmp0 (const void *_a, const void *_b)
1250 {
1251 struct tree_entry *a = *((struct tree_entry**)_a);
1252 struct tree_entry *b = *((struct tree_entry**)_b);
1253 return base_name_compare(
1254 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1255 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1256 }
1257
1258 static int tecmp1 (const void *_a, const void *_b)
1259 {
1260 struct tree_entry *a = *((struct tree_entry**)_a);
1261 struct tree_entry *b = *((struct tree_entry**)_b);
1262 return base_name_compare(
1263 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1264 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1265 }
1266
1267 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1268 {
1269 size_t maxlen = 0;
1270 unsigned int i;
1271
1272 if (!v)
1273 QSORT(t->entries, t->entry_count, tecmp0);
1274 else
1275 QSORT(t->entries, t->entry_count, tecmp1);
1276
1277 for (i = 0; i < t->entry_count; i++) {
1278 if (t->entries[i]->versions[v].mode)
1279 maxlen += t->entries[i]->name->str_len + 34;
1280 }
1281
1282 strbuf_reset(b);
1283 strbuf_grow(b, maxlen);
1284 for (i = 0; i < t->entry_count; i++) {
1285 struct tree_entry *e = t->entries[i];
1286 if (!e->versions[v].mode)
1287 continue;
1288 strbuf_addf(b, "%o %s%c",
1289 (unsigned int)(e->versions[v].mode & ~NO_DELTA),
1290 e->name->str_dat, '\0');
1291 strbuf_add(b, e->versions[v].oid.hash, GIT_SHA1_RAWSZ);
1292 }
1293 }
1294
1295 static void store_tree(struct tree_entry *root)
1296 {
1297 struct tree_content *t;
1298 unsigned int i, j, del;
1299 struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1300 struct object_entry *le = NULL;
1301
1302 if (!is_null_oid(&root->versions[1].oid))
1303 return;
1304
1305 if (!root->tree)
1306 load_tree(root);
1307 t = root->tree;
1308
1309 for (i = 0; i < t->entry_count; i++) {
1310 if (t->entries[i]->tree)
1311 store_tree(t->entries[i]);
1312 }
1313
1314 if (!(root->versions[0].mode & NO_DELTA))
1315 le = find_object(&root->versions[0].oid);
1316 if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1317 mktree(t, 0, &old_tree);
1318 lo.data = old_tree;
1319 lo.offset = le->idx.offset;
1320 lo.depth = t->delta_depth;
1321 }
1322
1323 mktree(t, 1, &new_tree);
1324 store_object(OBJ_TREE, &new_tree, &lo, &root->versions[1].oid, 0);
1325
1326 t->delta_depth = lo.depth;
1327 for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1328 struct tree_entry *e = t->entries[i];
1329 if (e->versions[1].mode) {
1330 e->versions[0].mode = e->versions[1].mode;
1331 oidcpy(&e->versions[0].oid, &e->versions[1].oid);
1332 t->entries[j++] = e;
1333 } else {
1334 release_tree_entry(e);
1335 del++;
1336 }
1337 }
1338 t->entry_count -= del;
1339 }
1340
1341 static void tree_content_replace(
1342 struct tree_entry *root,
1343 const struct object_id *oid,
1344 const uint16_t mode,
1345 struct tree_content *newtree)
1346 {
1347 if (!S_ISDIR(mode))
1348 die("Root cannot be a non-directory");
1349 oidclr(&root->versions[0].oid);
1350 oidcpy(&root->versions[1].oid, oid);
1351 if (root->tree)
1352 release_tree_content_recursive(root->tree);
1353 root->tree = newtree;
1354 }
1355
1356 static int tree_content_set(
1357 struct tree_entry *root,
1358 const char *p,
1359 const struct object_id *oid,
1360 const uint16_t mode,
1361 struct tree_content *subtree)
1362 {
1363 struct tree_content *t;
1364 const char *slash1;
1365 unsigned int i, n;
1366 struct tree_entry *e;
1367
1368 slash1 = strchrnul(p, '/');
1369 n = slash1 - p;
1370 if (!n)
1371 die("Empty path component found in input");
1372 if (!*slash1 && !S_ISDIR(mode) && subtree)
1373 die("Non-directories cannot have subtrees");
1374
1375 if (!root->tree)
1376 load_tree(root);
1377 t = root->tree;
1378 for (i = 0; i < t->entry_count; i++) {
1379 e = t->entries[i];
1380 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1381 if (!*slash1) {
1382 if (!S_ISDIR(mode)
1383 && e->versions[1].mode == mode
1384 && oideq(&e->versions[1].oid, oid))
1385 return 0;
1386 e->versions[1].mode = mode;
1387 oidcpy(&e->versions[1].oid, oid);
1388 if (e->tree)
1389 release_tree_content_recursive(e->tree);
1390 e->tree = subtree;
1391
1392 /*
1393 * We need to leave e->versions[0].sha1 alone
1394 * to avoid modifying the preimage tree used
1395 * when writing out the parent directory.
1396 * But after replacing the subdir with a
1397 * completely different one, it's not a good
1398 * delta base any more, and besides, we've
1399 * thrown away the tree entries needed to
1400 * make a delta against it.
1401 *
1402 * So let's just explicitly disable deltas
1403 * for the subtree.
1404 */
1405 if (S_ISDIR(e->versions[0].mode))
1406 e->versions[0].mode |= NO_DELTA;
1407
1408 oidclr(&root->versions[1].oid);
1409 return 1;
1410 }
1411 if (!S_ISDIR(e->versions[1].mode)) {
1412 e->tree = new_tree_content(8);
1413 e->versions[1].mode = S_IFDIR;
1414 }
1415 if (!e->tree)
1416 load_tree(e);
1417 if (tree_content_set(e, slash1 + 1, oid, mode, subtree)) {
1418 oidclr(&root->versions[1].oid);
1419 return 1;
1420 }
1421 return 0;
1422 }
1423 }
1424
1425 if (t->entry_count == t->entry_capacity)
1426 root->tree = t = grow_tree_content(t, t->entry_count);
1427 e = new_tree_entry();
1428 e->name = to_atom(p, n);
1429 e->versions[0].mode = 0;
1430 oidclr(&e->versions[0].oid);
1431 t->entries[t->entry_count++] = e;
1432 if (*slash1) {
1433 e->tree = new_tree_content(8);
1434 e->versions[1].mode = S_IFDIR;
1435 tree_content_set(e, slash1 + 1, oid, mode, subtree);
1436 } else {
1437 e->tree = subtree;
1438 e->versions[1].mode = mode;
1439 oidcpy(&e->versions[1].oid, oid);
1440 }
1441 oidclr(&root->versions[1].oid);
1442 return 1;
1443 }
1444
1445 static int tree_content_remove(
1446 struct tree_entry *root,
1447 const char *p,
1448 struct tree_entry *backup_leaf,
1449 int allow_root)
1450 {
1451 struct tree_content *t;
1452 const char *slash1;
1453 unsigned int i, n;
1454 struct tree_entry *e;
1455
1456 slash1 = strchrnul(p, '/');
1457 n = slash1 - p;
1458
1459 if (!root->tree)
1460 load_tree(root);
1461
1462 if (!*p && allow_root) {
1463 e = root;
1464 goto del_entry;
1465 }
1466
1467 t = root->tree;
1468 for (i = 0; i < t->entry_count; i++) {
1469 e = t->entries[i];
1470 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1471 if (*slash1 && !S_ISDIR(e->versions[1].mode))
1472 /*
1473 * If p names a file in some subdirectory, and a
1474 * file or symlink matching the name of the
1475 * parent directory of p exists, then p cannot
1476 * exist and need not be deleted.
1477 */
1478 return 1;
1479 if (!*slash1 || !S_ISDIR(e->versions[1].mode))
1480 goto del_entry;
1481 if (!e->tree)
1482 load_tree(e);
1483 if (tree_content_remove(e, slash1 + 1, backup_leaf, 0)) {
1484 for (n = 0; n < e->tree->entry_count; n++) {
1485 if (e->tree->entries[n]->versions[1].mode) {
1486 oidclr(&root->versions[1].oid);
1487 return 1;
1488 }
1489 }
1490 backup_leaf = NULL;
1491 goto del_entry;
1492 }
1493 return 0;
1494 }
1495 }
1496 return 0;
1497
1498 del_entry:
1499 if (backup_leaf)
1500 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1501 else if (e->tree)
1502 release_tree_content_recursive(e->tree);
1503 e->tree = NULL;
1504 e->versions[1].mode = 0;
1505 oidclr(&e->versions[1].oid);
1506 oidclr(&root->versions[1].oid);
1507 return 1;
1508 }
1509
1510 static int tree_content_get(
1511 struct tree_entry *root,
1512 const char *p,
1513 struct tree_entry *leaf,
1514 int allow_root)
1515 {
1516 struct tree_content *t;
1517 const char *slash1;
1518 unsigned int i, n;
1519 struct tree_entry *e;
1520
1521 slash1 = strchrnul(p, '/');
1522 n = slash1 - p;
1523 if (!n && !allow_root)
1524 die("Empty path component found in input");
1525
1526 if (!root->tree)
1527 load_tree(root);
1528
1529 if (!n) {
1530 e = root;
1531 goto found_entry;
1532 }
1533
1534 t = root->tree;
1535 for (i = 0; i < t->entry_count; i++) {
1536 e = t->entries[i];
1537 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1538 if (!*slash1)
1539 goto found_entry;
1540 if (!S_ISDIR(e->versions[1].mode))
1541 return 0;
1542 if (!e->tree)
1543 load_tree(e);
1544 return tree_content_get(e, slash1 + 1, leaf, 0);
1545 }
1546 }
1547 return 0;
1548
1549 found_entry:
1550 memcpy(leaf, e, sizeof(*leaf));
1551 if (e->tree && is_null_oid(&e->versions[1].oid))
1552 leaf->tree = dup_tree_content(e->tree);
1553 else
1554 leaf->tree = NULL;
1555 return 1;
1556 }
1557
1558 static int update_branch(struct branch *b)
1559 {
1560 static const char *msg = "fast-import";
1561 struct ref_transaction *transaction;
1562 struct object_id old_oid;
1563 struct strbuf err = STRBUF_INIT;
1564
1565 if (is_null_oid(&b->oid)) {
1566 if (b->delete)
1567 delete_ref(NULL, b->name, NULL, 0);
1568 return 0;
1569 }
1570 if (read_ref(b->name, &old_oid))
1571 oidclr(&old_oid);
1572 if (!force_update && !is_null_oid(&old_oid)) {
1573 struct commit *old_cmit, *new_cmit;
1574
1575 old_cmit = lookup_commit_reference_gently(the_repository,
1576 &old_oid, 0);
1577 new_cmit = lookup_commit_reference_gently(the_repository,
1578 &b->oid, 0);
1579 if (!old_cmit || !new_cmit)
1580 return error("Branch %s is missing commits.", b->name);
1581
1582 if (!in_merge_bases(old_cmit, new_cmit)) {
1583 warning("Not updating %s"
1584 " (new tip %s does not contain %s)",
1585 b->name, oid_to_hex(&b->oid),
1586 oid_to_hex(&old_oid));
1587 return -1;
1588 }
1589 }
1590 transaction = ref_transaction_begin(&err);
1591 if (!transaction ||
1592 ref_transaction_update(transaction, b->name, &b->oid, &old_oid,
1593 0, msg, &err) ||
1594 ref_transaction_commit(transaction, &err)) {
1595 ref_transaction_free(transaction);
1596 error("%s", err.buf);
1597 strbuf_release(&err);
1598 return -1;
1599 }
1600 ref_transaction_free(transaction);
1601 strbuf_release(&err);
1602 return 0;
1603 }
1604
1605 static void dump_branches(void)
1606 {
1607 unsigned int i;
1608 struct branch *b;
1609
1610 for (i = 0; i < branch_table_sz; i++) {
1611 for (b = branch_table[i]; b; b = b->table_next_branch)
1612 failure |= update_branch(b);
1613 }
1614 }
1615
1616 static void dump_tags(void)
1617 {
1618 static const char *msg = "fast-import";
1619 struct tag *t;
1620 struct strbuf ref_name = STRBUF_INIT;
1621 struct strbuf err = STRBUF_INIT;
1622 struct ref_transaction *transaction;
1623
1624 transaction = ref_transaction_begin(&err);
1625 if (!transaction) {
1626 failure |= error("%s", err.buf);
1627 goto cleanup;
1628 }
1629 for (t = first_tag; t; t = t->next_tag) {
1630 strbuf_reset(&ref_name);
1631 strbuf_addf(&ref_name, "refs/tags/%s", t->name);
1632
1633 if (ref_transaction_update(transaction, ref_name.buf,
1634 &t->oid, NULL, 0, msg, &err)) {
1635 failure |= error("%s", err.buf);
1636 goto cleanup;
1637 }
1638 }
1639 if (ref_transaction_commit(transaction, &err))
1640 failure |= error("%s", err.buf);
1641
1642 cleanup:
1643 ref_transaction_free(transaction);
1644 strbuf_release(&ref_name);
1645 strbuf_release(&err);
1646 }
1647
1648 static void dump_marks_helper(FILE *f,
1649 uintmax_t base,
1650 struct mark_set *m)
1651 {
1652 uintmax_t k;
1653 if (m->shift) {
1654 for (k = 0; k < 1024; k++) {
1655 if (m->data.sets[k])
1656 dump_marks_helper(f, base + (k << m->shift),
1657 m->data.sets[k]);
1658 }
1659 } else {
1660 for (k = 0; k < 1024; k++) {
1661 if (m->data.marked[k])
1662 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1663 oid_to_hex(&m->data.marked[k]->idx.oid));
1664 }
1665 }
1666 }
1667
1668 static void dump_marks(void)
1669 {
1670 struct lock_file mark_lock = LOCK_INIT;
1671 FILE *f;
1672
1673 if (!export_marks_file || (import_marks_file && !import_marks_file_done))
1674 return;
1675
1676 if (safe_create_leading_directories_const(export_marks_file)) {
1677 failure |= error_errno("unable to create leading directories of %s",
1678 export_marks_file);
1679 return;
1680 }
1681
1682 if (hold_lock_file_for_update(&mark_lock, export_marks_file, 0) < 0) {
1683 failure |= error_errno("Unable to write marks file %s",
1684 export_marks_file);
1685 return;
1686 }
1687
1688 f = fdopen_lock_file(&mark_lock, "w");
1689 if (!f) {
1690 int saved_errno = errno;
1691 rollback_lock_file(&mark_lock);
1692 failure |= error("Unable to write marks file %s: %s",
1693 export_marks_file, strerror(saved_errno));
1694 return;
1695 }
1696
1697 dump_marks_helper(f, 0, marks);
1698 if (commit_lock_file(&mark_lock)) {
1699 failure |= error_errno("Unable to write file %s",
1700 export_marks_file);
1701 return;
1702 }
1703 }
1704
1705 static void read_marks(void)
1706 {
1707 char line[512];
1708 FILE *f = fopen(import_marks_file, "r");
1709 if (f)
1710 ;
1711 else if (import_marks_file_ignore_missing && errno == ENOENT)
1712 goto done; /* Marks file does not exist */
1713 else
1714 die_errno("cannot read '%s'", import_marks_file);
1715 while (fgets(line, sizeof(line), f)) {
1716 uintmax_t mark;
1717 char *end;
1718 struct object_id oid;
1719 struct object_entry *e;
1720
1721 end = strchr(line, '\n');
1722 if (line[0] != ':' || !end)
1723 die("corrupt mark line: %s", line);
1724 *end = 0;
1725 mark = strtoumax(line + 1, &end, 10);
1726 if (!mark || end == line + 1
1727 || *end != ' ' || get_oid_hex(end + 1, &oid))
1728 die("corrupt mark line: %s", line);
1729 e = find_object(&oid);
1730 if (!e) {
1731 enum object_type type = oid_object_info(the_repository,
1732 &oid, NULL);
1733 if (type < 0)
1734 die("object not found: %s", oid_to_hex(&oid));
1735 e = insert_object(&oid);
1736 e->type = type;
1737 e->pack_id = MAX_PACK_ID;
1738 e->idx.offset = 1; /* just not zero! */
1739 }
1740 insert_mark(mark, e);
1741 }
1742 fclose(f);
1743 done:
1744 import_marks_file_done = 1;
1745 }
1746
1747
1748 static int read_next_command(void)
1749 {
1750 static int stdin_eof = 0;
1751
1752 if (stdin_eof) {
1753 unread_command_buf = 0;
1754 return EOF;
1755 }
1756
1757 for (;;) {
1758 const char *p;
1759
1760 if (unread_command_buf) {
1761 unread_command_buf = 0;
1762 } else {
1763 struct recent_command *rc;
1764
1765 strbuf_detach(&command_buf, NULL);
1766 stdin_eof = strbuf_getline_lf(&command_buf, stdin);
1767 if (stdin_eof)
1768 return EOF;
1769
1770 if (!seen_data_command
1771 && !starts_with(command_buf.buf, "feature ")
1772 && !starts_with(command_buf.buf, "option ")) {
1773 parse_argv();
1774 }
1775
1776 rc = rc_free;
1777 if (rc)
1778 rc_free = rc->next;
1779 else {
1780 rc = cmd_hist.next;
1781 cmd_hist.next = rc->next;
1782 cmd_hist.next->prev = &cmd_hist;
1783 free(rc->buf);
1784 }
1785
1786 rc->buf = command_buf.buf;
1787 rc->prev = cmd_tail;
1788 rc->next = cmd_hist.prev;
1789 rc->prev->next = rc;
1790 cmd_tail = rc;
1791 }
1792 if (skip_prefix(command_buf.buf, "get-mark ", &p)) {
1793 parse_get_mark(p);
1794 continue;
1795 }
1796 if (skip_prefix(command_buf.buf, "cat-blob ", &p)) {
1797 parse_cat_blob(p);
1798 continue;
1799 }
1800 if (command_buf.buf[0] == '#')
1801 continue;
1802 return 0;
1803 }
1804 }
1805
1806 static void skip_optional_lf(void)
1807 {
1808 int term_char = fgetc(stdin);
1809 if (term_char != '\n' && term_char != EOF)
1810 ungetc(term_char, stdin);
1811 }
1812
1813 static void parse_mark(void)
1814 {
1815 const char *v;
1816 if (skip_prefix(command_buf.buf, "mark :", &v)) {
1817 next_mark = strtoumax(v, NULL, 10);
1818 read_next_command();
1819 }
1820 else
1821 next_mark = 0;
1822 }
1823
1824 static void parse_original_identifier(void)
1825 {
1826 const char *v;
1827 if (skip_prefix(command_buf.buf, "original-oid ", &v))
1828 read_next_command();
1829 }
1830
1831 static int parse_data(struct strbuf *sb, uintmax_t limit, uintmax_t *len_res)
1832 {
1833 const char *data;
1834 strbuf_reset(sb);
1835
1836 if (!skip_prefix(command_buf.buf, "data ", &data))
1837 die("Expected 'data n' command, found: %s", command_buf.buf);
1838
1839 if (skip_prefix(data, "<<", &data)) {
1840 char *term = xstrdup(data);
1841 size_t term_len = command_buf.len - (data - command_buf.buf);
1842
1843 strbuf_detach(&command_buf, NULL);
1844 for (;;) {
1845 if (strbuf_getline_lf(&command_buf, stdin) == EOF)
1846 die("EOF in data (terminator '%s' not found)", term);
1847 if (term_len == command_buf.len
1848 && !strcmp(term, command_buf.buf))
1849 break;
1850 strbuf_addbuf(sb, &command_buf);
1851 strbuf_addch(sb, '\n');
1852 }
1853 free(term);
1854 }
1855 else {
1856 uintmax_t len = strtoumax(data, NULL, 10);
1857 size_t n = 0, length = (size_t)len;
1858
1859 if (limit && limit < len) {
1860 *len_res = len;
1861 return 0;
1862 }
1863 if (length < len)
1864 die("data is too large to use in this context");
1865
1866 while (n < length) {
1867 size_t s = strbuf_fread(sb, length - n, stdin);
1868 if (!s && feof(stdin))
1869 die("EOF in data (%lu bytes remaining)",
1870 (unsigned long)(length - n));
1871 n += s;
1872 }
1873 }
1874
1875 skip_optional_lf();
1876 return 1;
1877 }
1878
1879 static int validate_raw_date(const char *src, struct strbuf *result)
1880 {
1881 const char *orig_src = src;
1882 char *endp;
1883 unsigned long num;
1884
1885 errno = 0;
1886
1887 num = strtoul(src, &endp, 10);
1888 /* NEEDSWORK: perhaps check for reasonable values? */
1889 if (errno || endp == src || *endp != ' ')
1890 return -1;
1891
1892 src = endp + 1;
1893 if (*src != '-' && *src != '+')
1894 return -1;
1895
1896 num = strtoul(src + 1, &endp, 10);
1897 if (errno || endp == src + 1 || *endp || 1400 < num)
1898 return -1;
1899
1900 strbuf_addstr(result, orig_src);
1901 return 0;
1902 }
1903
1904 static char *parse_ident(const char *buf)
1905 {
1906 const char *ltgt;
1907 size_t name_len;
1908 struct strbuf ident = STRBUF_INIT;
1909
1910 /* ensure there is a space delimiter even if there is no name */
1911 if (*buf == '<')
1912 --buf;
1913
1914 ltgt = buf + strcspn(buf, "<>");
1915 if (*ltgt != '<')
1916 die("Missing < in ident string: %s", buf);
1917 if (ltgt != buf && ltgt[-1] != ' ')
1918 die("Missing space before < in ident string: %s", buf);
1919 ltgt = ltgt + 1 + strcspn(ltgt + 1, "<>");
1920 if (*ltgt != '>')
1921 die("Missing > in ident string: %s", buf);
1922 ltgt++;
1923 if (*ltgt != ' ')
1924 die("Missing space after > in ident string: %s", buf);
1925 ltgt++;
1926 name_len = ltgt - buf;
1927 strbuf_add(&ident, buf, name_len);
1928
1929 switch (whenspec) {
1930 case WHENSPEC_RAW:
1931 if (validate_raw_date(ltgt, &ident) < 0)
1932 die("Invalid raw date \"%s\" in ident: %s", ltgt, buf);
1933 break;
1934 case WHENSPEC_RFC2822:
1935 if (parse_date(ltgt, &ident) < 0)
1936 die("Invalid rfc2822 date \"%s\" in ident: %s", ltgt, buf);
1937 break;
1938 case WHENSPEC_NOW:
1939 if (strcmp("now", ltgt))
1940 die("Date in ident must be 'now': %s", buf);
1941 datestamp(&ident);
1942 break;
1943 }
1944
1945 return strbuf_detach(&ident, NULL);
1946 }
1947
1948 static void parse_and_store_blob(
1949 struct last_object *last,
1950 struct object_id *oidout,
1951 uintmax_t mark)
1952 {
1953 static struct strbuf buf = STRBUF_INIT;
1954 uintmax_t len;
1955
1956 if (parse_data(&buf, big_file_threshold, &len))
1957 store_object(OBJ_BLOB, &buf, last, oidout, mark);
1958 else {
1959 if (last) {
1960 strbuf_release(&last->data);
1961 last->offset = 0;
1962 last->depth = 0;
1963 }
1964 stream_blob(len, oidout, mark);
1965 skip_optional_lf();
1966 }
1967 }
1968
1969 static void parse_new_blob(void)
1970 {
1971 read_next_command();
1972 parse_mark();
1973 parse_original_identifier();
1974 parse_and_store_blob(&last_blob, NULL, next_mark);
1975 }
1976
1977 static void unload_one_branch(void)
1978 {
1979 while (cur_active_branches
1980 && cur_active_branches >= max_active_branches) {
1981 uintmax_t min_commit = ULONG_MAX;
1982 struct branch *e, *l = NULL, *p = NULL;
1983
1984 for (e = active_branches; e; e = e->active_next_branch) {
1985 if (e->last_commit < min_commit) {
1986 p = l;
1987 min_commit = e->last_commit;
1988 }
1989 l = e;
1990 }
1991
1992 if (p) {
1993 e = p->active_next_branch;
1994 p->active_next_branch = e->active_next_branch;
1995 } else {
1996 e = active_branches;
1997 active_branches = e->active_next_branch;
1998 }
1999 e->active = 0;
2000 e->active_next_branch = NULL;
2001 if (e->branch_tree.tree) {
2002 release_tree_content_recursive(e->branch_tree.tree);
2003 e->branch_tree.tree = NULL;
2004 }
2005 cur_active_branches--;
2006 }
2007 }
2008
2009 static void load_branch(struct branch *b)
2010 {
2011 load_tree(&b->branch_tree);
2012 if (!b->active) {
2013 b->active = 1;
2014 b->active_next_branch = active_branches;
2015 active_branches = b;
2016 cur_active_branches++;
2017 branch_load_count++;
2018 }
2019 }
2020
2021 static unsigned char convert_num_notes_to_fanout(uintmax_t num_notes)
2022 {
2023 unsigned char fanout = 0;
2024 while ((num_notes >>= 8))
2025 fanout++;
2026 return fanout;
2027 }
2028
2029 static void construct_path_with_fanout(const char *hex_sha1,
2030 unsigned char fanout, char *path)
2031 {
2032 unsigned int i = 0, j = 0;
2033 if (fanout >= the_hash_algo->rawsz)
2034 die("Too large fanout (%u)", fanout);
2035 while (fanout) {
2036 path[i++] = hex_sha1[j++];
2037 path[i++] = hex_sha1[j++];
2038 path[i++] = '/';
2039 fanout--;
2040 }
2041 memcpy(path + i, hex_sha1 + j, the_hash_algo->hexsz - j);
2042 path[i + the_hash_algo->hexsz - j] = '\0';
2043 }
2044
2045 static uintmax_t do_change_note_fanout(
2046 struct tree_entry *orig_root, struct tree_entry *root,
2047 char *hex_oid, unsigned int hex_oid_len,
2048 char *fullpath, unsigned int fullpath_len,
2049 unsigned char fanout)
2050 {
2051 struct tree_content *t;
2052 struct tree_entry *e, leaf;
2053 unsigned int i, tmp_hex_oid_len, tmp_fullpath_len;
2054 uintmax_t num_notes = 0;
2055 struct object_id oid;
2056 char realpath[60];
2057
2058 if (!root->tree)
2059 load_tree(root);
2060 t = root->tree;
2061
2062 for (i = 0; t && i < t->entry_count; i++) {
2063 e = t->entries[i];
2064 tmp_hex_oid_len = hex_oid_len + e->name->str_len;
2065 tmp_fullpath_len = fullpath_len;
2066
2067 /*
2068 * We're interested in EITHER existing note entries (entries
2069 * with exactly 40 hex chars in path, not including directory
2070 * separators), OR directory entries that may contain note
2071 * entries (with < 40 hex chars in path).
2072 * Also, each path component in a note entry must be a multiple
2073 * of 2 chars.
2074 */
2075 if (!e->versions[1].mode ||
2076 tmp_hex_oid_len > GIT_SHA1_HEXSZ ||
2077 e->name->str_len % 2)
2078 continue;
2079
2080 /* This _may_ be a note entry, or a subdir containing notes */
2081 memcpy(hex_oid + hex_oid_len, e->name->str_dat,
2082 e->name->str_len);
2083 if (tmp_fullpath_len)
2084 fullpath[tmp_fullpath_len++] = '/';
2085 memcpy(fullpath + tmp_fullpath_len, e->name->str_dat,
2086 e->name->str_len);
2087 tmp_fullpath_len += e->name->str_len;
2088 fullpath[tmp_fullpath_len] = '\0';
2089
2090 if (tmp_hex_oid_len == GIT_SHA1_HEXSZ && !get_oid_hex(hex_oid, &oid)) {
2091 /* This is a note entry */
2092 if (fanout == 0xff) {
2093 /* Counting mode, no rename */
2094 num_notes++;
2095 continue;
2096 }
2097 construct_path_with_fanout(hex_oid, fanout, realpath);
2098 if (!strcmp(fullpath, realpath)) {
2099 /* Note entry is in correct location */
2100 num_notes++;
2101 continue;
2102 }
2103
2104 /* Rename fullpath to realpath */
2105 if (!tree_content_remove(orig_root, fullpath, &leaf, 0))
2106 die("Failed to remove path %s", fullpath);
2107 tree_content_set(orig_root, realpath,
2108 &leaf.versions[1].oid,
2109 leaf.versions[1].mode,
2110 leaf.tree);
2111 } else if (S_ISDIR(e->versions[1].mode)) {
2112 /* This is a subdir that may contain note entries */
2113 num_notes += do_change_note_fanout(orig_root, e,
2114 hex_oid, tmp_hex_oid_len,
2115 fullpath, tmp_fullpath_len, fanout);
2116 }
2117
2118 /* The above may have reallocated the current tree_content */
2119 t = root->tree;
2120 }
2121 return num_notes;
2122 }
2123
2124 static uintmax_t change_note_fanout(struct tree_entry *root,
2125 unsigned char fanout)
2126 {
2127 /*
2128 * The size of path is due to one slash between every two hex digits,
2129 * plus the terminating NUL. Note that there is no slash at the end, so
2130 * the number of slashes is one less than half the number of hex
2131 * characters.
2132 */
2133 char hex_oid[GIT_MAX_HEXSZ], path[GIT_MAX_HEXSZ + (GIT_MAX_HEXSZ / 2) - 1 + 1];
2134 return do_change_note_fanout(root, root, hex_oid, 0, path, 0, fanout);
2135 }
2136
2137 /*
2138 * Given a pointer into a string, parse a mark reference:
2139 *
2140 * idnum ::= ':' bigint;
2141 *
2142 * Return the first character after the value in *endptr.
2143 *
2144 * Complain if the following character is not what is expected,
2145 * either a space or end of the string.
2146 */
2147 static uintmax_t parse_mark_ref(const char *p, char **endptr)
2148 {
2149 uintmax_t mark;
2150
2151 assert(*p == ':');
2152 p++;
2153 mark = strtoumax(p, endptr, 10);
2154 if (*endptr == p)
2155 die("No value after ':' in mark: %s", command_buf.buf);
2156 return mark;
2157 }
2158
2159 /*
2160 * Parse the mark reference, and complain if this is not the end of
2161 * the string.
2162 */
2163 static uintmax_t parse_mark_ref_eol(const char *p)
2164 {
2165 char *end;
2166 uintmax_t mark;
2167
2168 mark = parse_mark_ref(p, &end);
2169 if (*end != '\0')
2170 die("Garbage after mark: %s", command_buf.buf);
2171 return mark;
2172 }
2173
2174 /*
2175 * Parse the mark reference, demanding a trailing space. Return a
2176 * pointer to the space.
2177 */
2178 static uintmax_t parse_mark_ref_space(const char **p)
2179 {
2180 uintmax_t mark;
2181 char *end;
2182
2183 mark = parse_mark_ref(*p, &end);
2184 if (*end++ != ' ')
2185 die("Missing space after mark: %s", command_buf.buf);
2186 *p = end;
2187 return mark;
2188 }
2189
2190 static void file_change_m(const char *p, struct branch *b)
2191 {
2192 static struct strbuf uq = STRBUF_INIT;
2193 const char *endp;
2194 struct object_entry *oe;
2195 struct object_id oid;
2196 uint16_t mode, inline_data = 0;
2197
2198 p = get_mode(p, &mode);
2199 if (!p)
2200 die("Corrupt mode: %s", command_buf.buf);
2201 switch (mode) {
2202 case 0644:
2203 case 0755:
2204 mode |= S_IFREG;
2205 case S_IFREG | 0644:
2206 case S_IFREG | 0755:
2207 case S_IFLNK:
2208 case S_IFDIR:
2209 case S_IFGITLINK:
2210 /* ok */
2211 break;
2212 default:
2213 die("Corrupt mode: %s", command_buf.buf);
2214 }
2215
2216 if (*p == ':') {
2217 oe = find_mark(parse_mark_ref_space(&p));
2218 oidcpy(&oid, &oe->idx.oid);
2219 } else if (skip_prefix(p, "inline ", &p)) {
2220 inline_data = 1;
2221 oe = NULL; /* not used with inline_data, but makes gcc happy */
2222 } else {
2223 if (parse_oid_hex(p, &oid, &p))
2224 die("Invalid dataref: %s", command_buf.buf);
2225 oe = find_object(&oid);
2226 if (*p++ != ' ')
2227 die("Missing space after SHA1: %s", command_buf.buf);
2228 }
2229
2230 strbuf_reset(&uq);
2231 if (!unquote_c_style(&uq, p, &endp)) {
2232 if (*endp)
2233 die("Garbage after path in: %s", command_buf.buf);
2234 p = uq.buf;
2235 }
2236
2237 /* Git does not track empty, non-toplevel directories. */
2238 if (S_ISDIR(mode) && is_empty_tree_oid(&oid) && *p) {
2239 tree_content_remove(&b->branch_tree, p, NULL, 0);
2240 return;
2241 }
2242
2243 if (S_ISGITLINK(mode)) {
2244 if (inline_data)
2245 die("Git links cannot be specified 'inline': %s",
2246 command_buf.buf);
2247 else if (oe) {
2248 if (oe->type != OBJ_COMMIT)
2249 die("Not a commit (actually a %s): %s",
2250 type_name(oe->type), command_buf.buf);
2251 }
2252 /*
2253 * Accept the sha1 without checking; it expected to be in
2254 * another repository.
2255 */
2256 } else if (inline_data) {
2257 if (S_ISDIR(mode))
2258 die("Directories cannot be specified 'inline': %s",
2259 command_buf.buf);
2260 if (p != uq.buf) {
2261 strbuf_addstr(&uq, p);
2262 p = uq.buf;
2263 }
2264 read_next_command();
2265 parse_and_store_blob(&last_blob, &oid, 0);
2266 } else {
2267 enum object_type expected = S_ISDIR(mode) ?
2268 OBJ_TREE: OBJ_BLOB;
2269 enum object_type type = oe ? oe->type :
2270 oid_object_info(the_repository, &oid,
2271 NULL);
2272 if (type < 0)
2273 die("%s not found: %s",
2274 S_ISDIR(mode) ? "Tree" : "Blob",
2275 command_buf.buf);
2276 if (type != expected)
2277 die("Not a %s (actually a %s): %s",
2278 type_name(expected), type_name(type),
2279 command_buf.buf);
2280 }
2281
2282 if (!*p) {
2283 tree_content_replace(&b->branch_tree, &oid, mode, NULL);
2284 return;
2285 }
2286 tree_content_set(&b->branch_tree, p, &oid, mode, NULL);
2287 }
2288
2289 static void file_change_d(const char *p, struct branch *b)
2290 {
2291 static struct strbuf uq = STRBUF_INIT;
2292 const char *endp;
2293
2294 strbuf_reset(&uq);
2295 if (!unquote_c_style(&uq, p, &endp)) {
2296 if (*endp)
2297 die("Garbage after path in: %s", command_buf.buf);
2298 p = uq.buf;
2299 }
2300 tree_content_remove(&b->branch_tree, p, NULL, 1);
2301 }
2302
2303 static void file_change_cr(const char *s, struct branch *b, int rename)
2304 {
2305 const char *d;
2306 static struct strbuf s_uq = STRBUF_INIT;
2307 static struct strbuf d_uq = STRBUF_INIT;
2308 const char *endp;
2309 struct tree_entry leaf;
2310
2311 strbuf_reset(&s_uq);
2312 if (!unquote_c_style(&s_uq, s, &endp)) {
2313 if (*endp != ' ')
2314 die("Missing space after source: %s", command_buf.buf);
2315 } else {
2316 endp = strchr(s, ' ');
2317 if (!endp)
2318 die("Missing space after source: %s", command_buf.buf);
2319 strbuf_add(&s_uq, s, endp - s);
2320 }
2321 s = s_uq.buf;
2322
2323 endp++;
2324 if (!*endp)
2325 die("Missing dest: %s", command_buf.buf);
2326
2327 d = endp;
2328 strbuf_reset(&d_uq);
2329 if (!unquote_c_style(&d_uq, d, &endp)) {
2330 if (*endp)
2331 die("Garbage after dest in: %s", command_buf.buf);
2332 d = d_uq.buf;
2333 }
2334
2335 memset(&leaf, 0, sizeof(leaf));
2336 if (rename)
2337 tree_content_remove(&b->branch_tree, s, &leaf, 1);
2338 else
2339 tree_content_get(&b->branch_tree, s, &leaf, 1);
2340 if (!leaf.versions[1].mode)
2341 die("Path %s not in branch", s);
2342 if (!*d) { /* C "path/to/subdir" "" */
2343 tree_content_replace(&b->branch_tree,
2344 &leaf.versions[1].oid,
2345 leaf.versions[1].mode,
2346 leaf.tree);
2347 return;
2348 }
2349 tree_content_set(&b->branch_tree, d,
2350 &leaf.versions[1].oid,
2351 leaf.versions[1].mode,
2352 leaf.tree);
2353 }
2354
2355 static void note_change_n(const char *p, struct branch *b, unsigned char *old_fanout)
2356 {
2357 static struct strbuf uq = STRBUF_INIT;
2358 struct object_entry *oe;
2359 struct branch *s;
2360 struct object_id oid, commit_oid;
2361 char path[60];
2362 uint16_t inline_data = 0;
2363 unsigned char new_fanout;
2364
2365 /*
2366 * When loading a branch, we don't traverse its tree to count the real
2367 * number of notes (too expensive to do this for all non-note refs).
2368 * This means that recently loaded notes refs might incorrectly have
2369 * b->num_notes == 0, and consequently, old_fanout might be wrong.
2370 *
2371 * Fix this by traversing the tree and counting the number of notes
2372 * when b->num_notes == 0. If the notes tree is truly empty, the
2373 * calculation should not take long.
2374 */
2375 if (b->num_notes == 0 && *old_fanout == 0) {
2376 /* Invoke change_note_fanout() in "counting mode". */
2377 b->num_notes = change_note_fanout(&b->branch_tree, 0xff);
2378 *old_fanout = convert_num_notes_to_fanout(b->num_notes);
2379 }
2380
2381 /* Now parse the notemodify command. */
2382 /* <dataref> or 'inline' */
2383 if (*p == ':') {
2384 oe = find_mark(parse_mark_ref_space(&p));
2385 oidcpy(&oid, &oe->idx.oid);
2386 } else if (skip_prefix(p, "inline ", &p)) {
2387 inline_data = 1;
2388 oe = NULL; /* not used with inline_data, but makes gcc happy */
2389 } else {
2390 if (parse_oid_hex(p, &oid, &p))
2391 die("Invalid dataref: %s", command_buf.buf);
2392 oe = find_object(&oid);
2393 if (*p++ != ' ')
2394 die("Missing space after SHA1: %s", command_buf.buf);
2395 }
2396
2397 /* <commit-ish> */
2398 s = lookup_branch(p);
2399 if (s) {
2400 if (is_null_oid(&s->oid))
2401 die("Can't add a note on empty branch.");
2402 oidcpy(&commit_oid, &s->oid);
2403 } else if (*p == ':') {
2404 uintmax_t commit_mark = parse_mark_ref_eol(p);
2405 struct object_entry *commit_oe = find_mark(commit_mark);
2406 if (commit_oe->type != OBJ_COMMIT)
2407 die("Mark :%" PRIuMAX " not a commit", commit_mark);
2408 oidcpy(&commit_oid, &commit_oe->idx.oid);
2409 } else if (!get_oid(p, &commit_oid)) {
2410 unsigned long size;
2411 char *buf = read_object_with_reference(&commit_oid,
2412 commit_type, &size,
2413 &commit_oid);
2414 if (!buf || size < 46)
2415 die("Not a valid commit: %s", p);
2416 free(buf);
2417 } else
2418 die("Invalid ref name or SHA1 expression: %s", p);
2419
2420 if (inline_data) {
2421 if (p != uq.buf) {
2422 strbuf_addstr(&uq, p);
2423 p = uq.buf;
2424 }
2425 read_next_command();
2426 parse_and_store_blob(&last_blob, &oid, 0);
2427 } else if (oe) {
2428 if (oe->type != OBJ_BLOB)
2429 die("Not a blob (actually a %s): %s",
2430 type_name(oe->type), command_buf.buf);
2431 } else if (!is_null_oid(&oid)) {
2432 enum object_type type = oid_object_info(the_repository, &oid,
2433 NULL);
2434 if (type < 0)
2435 die("Blob not found: %s", command_buf.buf);
2436 if (type != OBJ_BLOB)
2437 die("Not a blob (actually a %s): %s",
2438 type_name(type), command_buf.buf);
2439 }
2440
2441 construct_path_with_fanout(oid_to_hex(&commit_oid), *old_fanout, path);
2442 if (tree_content_remove(&b->branch_tree, path, NULL, 0))
2443 b->num_notes--;
2444
2445 if (is_null_oid(&oid))
2446 return; /* nothing to insert */
2447
2448 b->num_notes++;
2449 new_fanout = convert_num_notes_to_fanout(b->num_notes);
2450 construct_path_with_fanout(oid_to_hex(&commit_oid), new_fanout, path);
2451 tree_content_set(&b->branch_tree, path, &oid, S_IFREG | 0644, NULL);
2452 }
2453
2454 static void file_change_deleteall(struct branch *b)
2455 {
2456 release_tree_content_recursive(b->branch_tree.tree);
2457 oidclr(&b->branch_tree.versions[0].oid);
2458 oidclr(&b->branch_tree.versions[1].oid);
2459 load_tree(&b->branch_tree);
2460 b->num_notes = 0;
2461 }
2462
2463 static void parse_from_commit(struct branch *b, char *buf, unsigned long size)
2464 {
2465 if (!buf || size < GIT_SHA1_HEXSZ + 6)
2466 die("Not a valid commit: %s", oid_to_hex(&b->oid));
2467 if (memcmp("tree ", buf, 5)
2468 || get_oid_hex(buf + 5, &b->branch_tree.versions[1].oid))
2469 die("The commit %s is corrupt", oid_to_hex(&b->oid));
2470 oidcpy(&b->branch_tree.versions[0].oid,
2471 &b->branch_tree.versions[1].oid);
2472 }
2473
2474 static void parse_from_existing(struct branch *b)
2475 {
2476 if (is_null_oid(&b->oid)) {
2477 oidclr(&b->branch_tree.versions[0].oid);
2478 oidclr(&b->branch_tree.versions[1].oid);
2479 } else {
2480 unsigned long size;
2481 char *buf;
2482
2483 buf = read_object_with_reference(&b->oid, commit_type, &size,
2484 &b->oid);
2485 parse_from_commit(b, buf, size);
2486 free(buf);
2487 }
2488 }
2489
2490 static int parse_from(struct branch *b)
2491 {
2492 const char *from;
2493 struct branch *s;
2494 struct object_id oid;
2495
2496 if (!skip_prefix(command_buf.buf, "from ", &from))
2497 return 0;
2498
2499 oidcpy(&oid, &b->branch_tree.versions[1].oid);
2500
2501 s = lookup_branch(from);
2502 if (b == s)
2503 die("Can't create a branch from itself: %s", b->name);
2504 else if (s) {
2505 struct object_id *t = &s->branch_tree.versions[1].oid;
2506 oidcpy(&b->oid, &s->oid);
2507 oidcpy(&b->branch_tree.versions[0].oid, t);
2508 oidcpy(&b->branch_tree.versions[1].oid, t);
2509 } else if (*from == ':') {
2510 uintmax_t idnum = parse_mark_ref_eol(from);
2511 struct object_entry *oe = find_mark(idnum);
2512 if (oe->type != OBJ_COMMIT)
2513 die("Mark :%" PRIuMAX " not a commit", idnum);
2514 if (!oideq(&b->oid, &oe->idx.oid)) {
2515 oidcpy(&b->oid, &oe->idx.oid);
2516 if (oe->pack_id != MAX_PACK_ID) {
2517 unsigned long size;
2518 char *buf = gfi_unpack_entry(oe, &size);
2519 parse_from_commit(b, buf, size);
2520 free(buf);
2521 } else
2522 parse_from_existing(b);
2523 }
2524 } else if (!get_oid(from, &b->oid)) {
2525 parse_from_existing(b);
2526 if (is_null_oid(&b->oid))
2527 b->delete = 1;
2528 }
2529 else
2530 die("Invalid ref name or SHA1 expression: %s", from);
2531
2532 if (b->branch_tree.tree && !oideq(&oid, &b->branch_tree.versions[1].oid)) {
2533 release_tree_content_recursive(b->branch_tree.tree);
2534 b->branch_tree.tree = NULL;
2535 }
2536
2537 read_next_command();
2538 return 1;
2539 }
2540
2541 static struct hash_list *parse_merge(unsigned int *count)
2542 {
2543 struct hash_list *list = NULL, **tail = &list, *n;
2544 const char *from;
2545 struct branch *s;
2546
2547 *count = 0;
2548 while (skip_prefix(command_buf.buf, "merge ", &from)) {
2549 n = xmalloc(sizeof(*n));
2550 s = lookup_branch(from);
2551 if (s)
2552 oidcpy(&n->oid, &s->oid);
2553 else if (*from == ':') {
2554 uintmax_t idnum = parse_mark_ref_eol(from);
2555 struct object_entry *oe = find_mark(idnum);
2556 if (oe->type != OBJ_COMMIT)
2557 die("Mark :%" PRIuMAX " not a commit", idnum);
2558 oidcpy(&n->oid, &oe->idx.oid);
2559 } else if (!get_oid(from, &n->oid)) {
2560 unsigned long size;
2561 char *buf = read_object_with_reference(&n->oid,
2562 commit_type,
2563 &size, &n->oid);
2564 if (!buf || size < 46)
2565 die("Not a valid commit: %s", from);
2566 free(buf);
2567 } else
2568 die("Invalid ref name or SHA1 expression: %s", from);
2569
2570 n->next = NULL;
2571 *tail = n;
2572 tail = &n->next;
2573
2574 (*count)++;
2575 read_next_command();
2576 }
2577 return list;
2578 }
2579
2580 static void parse_new_commit(const char *arg)
2581 {
2582 static struct strbuf msg = STRBUF_INIT;
2583 struct branch *b;
2584 char *author = NULL;
2585 char *committer = NULL;
2586 struct hash_list *merge_list = NULL;
2587 unsigned int merge_count;
2588 unsigned char prev_fanout, new_fanout;
2589 const char *v;
2590
2591 b = lookup_branch(arg);
2592 if (!b)
2593 b = new_branch(arg);
2594
2595 read_next_command();
2596 parse_mark();
2597 parse_original_identifier();
2598 if (skip_prefix(command_buf.buf, "author ", &v)) {
2599 author = parse_ident(v);
2600 read_next_command();
2601 }
2602 if (skip_prefix(command_buf.buf, "committer ", &v)) {
2603 committer = parse_ident(v);
2604 read_next_command();
2605 }
2606 if (!committer)
2607 die("Expected committer but didn't get one");
2608 parse_data(&msg, 0, NULL);
2609 read_next_command();
2610 parse_from(b);
2611 merge_list = parse_merge(&merge_count);
2612
2613 /* ensure the branch is active/loaded */
2614 if (!b->branch_tree.tree || !max_active_branches) {
2615 unload_one_branch();
2616 load_branch(b);
2617 }
2618
2619 prev_fanout = convert_num_notes_to_fanout(b->num_notes);
2620
2621 /* file_change* */
2622 while (command_buf.len > 0) {
2623 if (skip_prefix(command_buf.buf, "M ", &v))
2624 file_change_m(v, b);
2625 else if (skip_prefix(command_buf.buf, "D ", &v))
2626 file_change_d(v, b);
2627 else if (skip_prefix(command_buf.buf, "R ", &v))
2628 file_change_cr(v, b, 1);
2629 else if (skip_prefix(command_buf.buf, "C ", &v))
2630 file_change_cr(v, b, 0);
2631 else if (skip_prefix(command_buf.buf, "N ", &v))
2632 note_change_n(v, b, &prev_fanout);
2633 else if (!strcmp("deleteall", command_buf.buf))
2634 file_change_deleteall(b);
2635 else if (skip_prefix(command_buf.buf, "ls ", &v))
2636 parse_ls(v, b);
2637 else {
2638 unread_command_buf = 1;
2639 break;
2640 }
2641 if (read_next_command() == EOF)
2642 break;
2643 }
2644
2645 new_fanout = convert_num_notes_to_fanout(b->num_notes);
2646 if (new_fanout != prev_fanout)
2647 b->num_notes = change_note_fanout(&b->branch_tree, new_fanout);
2648
2649 /* build the tree and the commit */
2650 store_tree(&b->branch_tree);
2651 oidcpy(&b->branch_tree.versions[0].oid,
2652 &b->branch_tree.versions[1].oid);
2653
2654 strbuf_reset(&new_data);
2655 strbuf_addf(&new_data, "tree %s\n",
2656 oid_to_hex(&b->branch_tree.versions[1].oid));
2657 if (!is_null_oid(&b->oid))
2658 strbuf_addf(&new_data, "parent %s\n",
2659 oid_to_hex(&b->oid));
2660 while (merge_list) {
2661 struct hash_list *next = merge_list->next;
2662 strbuf_addf(&new_data, "parent %s\n",
2663 oid_to_hex(&merge_list->oid));
2664 free(merge_list);
2665 merge_list = next;
2666 }
2667 strbuf_addf(&new_data,
2668 "author %s\n"
2669 "committer %s\n"
2670 "\n",
2671 author ? author : committer, committer);
2672 strbuf_addbuf(&new_data, &msg);
2673 free(author);
2674 free(committer);
2675
2676 if (!store_object(OBJ_COMMIT, &new_data, NULL, &b->oid, next_mark))
2677 b->pack_id = pack_id;
2678 b->last_commit = object_count_by_type[OBJ_COMMIT];
2679 }
2680
2681 static void parse_new_tag(const char *arg)
2682 {
2683 static struct strbuf msg = STRBUF_INIT;
2684 const char *from;
2685 char *tagger;
2686 struct branch *s;
2687 struct tag *t;
2688 uintmax_t from_mark = 0;
2689 struct object_id oid;
2690 enum object_type type;
2691 const char *v;
2692
2693 t = mem_pool_alloc(&fi_mem_pool, sizeof(struct tag));
2694 memset(t, 0, sizeof(struct tag));
2695 t->name = pool_strdup(arg);
2696 if (last_tag)
2697 last_tag->next_tag = t;
2698 else
2699 first_tag = t;
2700 last_tag = t;
2701 read_next_command();
2702
2703 /* from ... */
2704 if (!skip_prefix(command_buf.buf, "from ", &from))
2705 die("Expected from command, got %s", command_buf.buf);
2706 s = lookup_branch(from);
2707 if (s) {
2708 if (is_null_oid(&s->oid))
2709 die("Can't tag an empty branch.");
2710 oidcpy(&oid, &s->oid);
2711 type = OBJ_COMMIT;
2712 } else if (*from == ':') {
2713 struct object_entry *oe;
2714 from_mark = parse_mark_ref_eol(from);
2715 oe = find_mark(from_mark);
2716 type = oe->type;
2717 oidcpy(&oid, &oe->idx.oid);
2718 } else if (!get_oid(from, &oid)) {
2719 struct object_entry *oe = find_object(&oid);
2720 if (!oe) {
2721 type = oid_object_info(the_repository, &oid, NULL);
2722 if (type < 0)
2723 die("Not a valid object: %s", from);
2724 } else
2725 type = oe->type;
2726 } else
2727 die("Invalid ref name or SHA1 expression: %s", from);
2728 read_next_command();
2729
2730 /* original-oid ... */
2731 parse_original_identifier();
2732
2733 /* tagger ... */
2734 if (skip_prefix(command_buf.buf, "tagger ", &v)) {
2735 tagger = parse_ident(v);
2736 read_next_command();
2737 } else
2738 tagger = NULL;
2739
2740 /* tag payload/message */
2741 parse_data(&msg, 0, NULL);
2742
2743 /* build the tag object */
2744 strbuf_reset(&new_data);
2745
2746 strbuf_addf(&new_data,
2747 "object %s\n"
2748 "type %s\n"
2749 "tag %s\n",
2750 oid_to_hex(&oid), type_name(type), t->name);
2751 if (tagger)
2752 strbuf_addf(&new_data,
2753 "tagger %s\n", tagger);
2754 strbuf_addch(&new_data, '\n');
2755 strbuf_addbuf(&new_data, &msg);
2756 free(tagger);
2757
2758 if (store_object(OBJ_TAG, &new_data, NULL, &t->oid, 0))
2759 t->pack_id = MAX_PACK_ID;
2760 else
2761 t->pack_id = pack_id;
2762 }
2763
2764 static void parse_reset_branch(const char *arg)
2765 {
2766 struct branch *b;
2767
2768 b = lookup_branch(arg);
2769 if (b) {
2770 oidclr(&b->oid);
2771 oidclr(&b->branch_tree.versions[0].oid);
2772 oidclr(&b->branch_tree.versions[1].oid);
2773 if (b->branch_tree.tree) {
2774 release_tree_content_recursive(b->branch_tree.tree);
2775 b->branch_tree.tree = NULL;
2776 }
2777 }
2778 else
2779 b = new_branch(arg);
2780 read_next_command();
2781 parse_from(b);
2782 if (command_buf.len > 0)
2783 unread_command_buf = 1;
2784 }
2785
2786 static void cat_blob_write(const char *buf, unsigned long size)
2787 {
2788 if (write_in_full(cat_blob_fd, buf, size) < 0)
2789 die_errno("Write to frontend failed");
2790 }
2791
2792 static void cat_blob(struct object_entry *oe, struct object_id *oid)
2793 {
2794 struct strbuf line = STRBUF_INIT;
2795 unsigned long size;
2796 enum object_type type = 0;
2797 char *buf;
2798
2799 if (!oe || oe->pack_id == MAX_PACK_ID) {
2800 buf = read_object_file(oid, &type, &size);
2801 } else {
2802 type = oe->type;
2803 buf = gfi_unpack_entry(oe, &size);
2804 }
2805
2806 /*
2807 * Output based on batch_one_object() from cat-file.c.
2808 */
2809 if (type <= 0) {
2810 strbuf_reset(&line);
2811 strbuf_addf(&line, "%s missing\n", oid_to_hex(oid));
2812 cat_blob_write(line.buf, line.len);
2813 strbuf_release(&line);
2814 free(buf);
2815 return;
2816 }
2817 if (!buf)
2818 die("Can't read object %s", oid_to_hex(oid));
2819 if (type != OBJ_BLOB)
2820 die("Object %s is a %s but a blob was expected.",
2821 oid_to_hex(oid), type_name(type));
2822 strbuf_reset(&line);
2823 strbuf_addf(&line, "%s %s %"PRIuMAX"\n", oid_to_hex(oid),
2824 type_name(type), (uintmax_t)size);
2825 cat_blob_write(line.buf, line.len);
2826 strbuf_release(&line);
2827 cat_blob_write(buf, size);
2828 cat_blob_write("\n", 1);
2829 if (oe && oe->pack_id == pack_id) {
2830 last_blob.offset = oe->idx.offset;
2831 strbuf_attach(&last_blob.data, buf, size, size);
2832 last_blob.depth = oe->depth;
2833 } else
2834 free(buf);
2835 }
2836
2837 static void parse_get_mark(const char *p)
2838 {
2839 struct object_entry *oe;
2840 char output[GIT_MAX_HEXSZ + 2];
2841
2842 /* get-mark SP <object> LF */
2843 if (*p != ':')
2844 die("Not a mark: %s", p);
2845
2846 oe = find_mark(parse_mark_ref_eol(p));
2847 if (!oe)
2848 die("Unknown mark: %s", command_buf.buf);
2849
2850 xsnprintf(output, sizeof(output), "%s\n", oid_to_hex(&oe->idx.oid));
2851 cat_blob_write(output, GIT_SHA1_HEXSZ + 1);
2852 }
2853
2854 static void parse_cat_blob(const char *p)
2855 {
2856 struct object_entry *oe;
2857 struct object_id oid;
2858
2859 /* cat-blob SP <object> LF */
2860 if (*p == ':') {
2861 oe = find_mark(parse_mark_ref_eol(p));
2862 if (!oe)
2863 die("Unknown mark: %s", command_buf.buf);
2864 oidcpy(&oid, &oe->idx.oid);
2865 } else {
2866 if (parse_oid_hex(p, &oid, &p))
2867 die("Invalid dataref: %s", command_buf.buf);
2868 if (*p)
2869 die("Garbage after SHA1: %s", command_buf.buf);
2870 oe = find_object(&oid);
2871 }
2872
2873 cat_blob(oe, &oid);
2874 }
2875
2876 static struct object_entry *dereference(struct object_entry *oe,
2877 struct object_id *oid)
2878 {
2879 unsigned long size;
2880 char *buf = NULL;
2881 if (!oe) {
2882 enum object_type type = oid_object_info(the_repository, oid,
2883 NULL);
2884 if (type < 0)
2885 die("object not found: %s", oid_to_hex(oid));
2886 /* cache it! */
2887 oe = insert_object(oid);
2888 oe->type = type;
2889 oe->pack_id = MAX_PACK_ID;
2890 oe->idx.offset = 1;
2891 }
2892 switch (oe->type) {
2893 case OBJ_TREE: /* easy case. */
2894 return oe;
2895 case OBJ_COMMIT:
2896 case OBJ_TAG:
2897 break;
2898 default:
2899 die("Not a tree-ish: %s", command_buf.buf);
2900 }
2901
2902 if (oe->pack_id != MAX_PACK_ID) { /* in a pack being written */
2903 buf = gfi_unpack_entry(oe, &size);
2904 } else {
2905 enum object_type unused;
2906 buf = read_object_file(oid, &unused, &size);
2907 }
2908 if (!buf)
2909 die("Can't load object %s", oid_to_hex(oid));
2910
2911 /* Peel one layer. */
2912 switch (oe->type) {
2913 case OBJ_TAG:
2914 if (size < GIT_SHA1_HEXSZ + strlen("object ") ||
2915 get_oid_hex(buf + strlen("object "), oid))
2916 die("Invalid SHA1 in tag: %s", command_buf.buf);
2917 break;
2918 case OBJ_COMMIT:
2919 if (size < GIT_SHA1_HEXSZ + strlen("tree ") ||
2920 get_oid_hex(buf + strlen("tree "), oid))
2921 die("Invalid SHA1 in commit: %s", command_buf.buf);
2922 }
2923
2924 free(buf);
2925 return find_object(oid);
2926 }
2927
2928 static struct object_entry *parse_treeish_dataref(const char **p)
2929 {
2930 struct object_id oid;
2931 struct object_entry *e;
2932
2933 if (**p == ':') { /* <mark> */
2934 e = find_mark(parse_mark_ref_space(p));
2935 if (!e)
2936 die("Unknown mark: %s", command_buf.buf);
2937 oidcpy(&oid, &e->idx.oid);
2938 } else { /* <sha1> */
2939 if (parse_oid_hex(*p, &oid, p))
2940 die("Invalid dataref: %s", command_buf.buf);
2941 e = find_object(&oid);
2942 if (*(*p)++ != ' ')
2943 die("Missing space after tree-ish: %s", command_buf.buf);
2944 }
2945
2946 while (!e || e->type != OBJ_TREE)
2947 e = dereference(e, &oid);
2948 return e;
2949 }
2950
2951 static void print_ls(int mode, const unsigned char *sha1, const char *path)
2952 {
2953 static struct strbuf line = STRBUF_INIT;
2954
2955 /* See show_tree(). */
2956 const char *type =
2957 S_ISGITLINK(mode) ? commit_type :
2958 S_ISDIR(mode) ? tree_type :
2959 blob_type;
2960
2961 if (!mode) {
2962 /* missing SP path LF */
2963 strbuf_reset(&line);
2964 strbuf_addstr(&line, "missing ");
2965 quote_c_style(path, &line, NULL, 0);
2966 strbuf_addch(&line, '\n');
2967 } else {
2968 /* mode SP type SP object_name TAB path LF */
2969 strbuf_reset(&line);
2970 strbuf_addf(&line, "%06o %s %s\t",
2971 mode & ~NO_DELTA, type, sha1_to_hex(sha1));
2972 quote_c_style(path, &line, NULL, 0);
2973 strbuf_addch(&line, '\n');
2974 }
2975 cat_blob_write(line.buf, line.len);
2976 }
2977
2978 static void parse_ls(const char *p, struct branch *b)
2979 {
2980 struct tree_entry *root = NULL;
2981 struct tree_entry leaf = {NULL};
2982
2983 /* ls SP (<tree-ish> SP)? <path> */
2984 if (*p == '"') {
2985 if (!b)
2986 die("Not in a commit: %s", command_buf.buf);
2987 root = &b->branch_tree;
2988 } else {
2989 struct object_entry *e = parse_treeish_dataref(&p);
2990 root = new_tree_entry();
2991 oidcpy(&root->versions[1].oid, &e->idx.oid);
2992 if (!is_null_oid(&root->versions[1].oid))
2993 root->versions[1].mode = S_IFDIR;
2994 load_tree(root);
2995 }
2996 if (*p == '"') {
2997 static struct strbuf uq = STRBUF_INIT;
2998 const char *endp;
2999 strbuf_reset(&uq);
3000 if (unquote_c_style(&uq, p, &endp))
3001 die("Invalid path: %s", command_buf.buf);
3002 if (*endp)
3003 die("Garbage after path in: %s", command_buf.buf);
3004 p = uq.buf;
3005 }
3006 tree_content_get(root, p, &leaf, 1);
3007 /*
3008 * A directory in preparation would have a sha1 of zero
3009 * until it is saved. Save, for simplicity.
3010 */
3011 if (S_ISDIR(leaf.versions[1].mode))
3012 store_tree(&leaf);
3013
3014 print_ls(leaf.versions[1].mode, leaf.versions[1].oid.hash, p);
3015 if (leaf.tree)
3016 release_tree_content_recursive(leaf.tree);
3017 if (!b || root != &b->branch_tree)
3018 release_tree_entry(root);
3019 }
3020
3021 static void checkpoint(void)
3022 {
3023 checkpoint_requested = 0;
3024 if (object_count) {
3025 cycle_packfile();
3026 }
3027 dump_branches();
3028 dump_tags();
3029 dump_marks();
3030 }
3031
3032 static void parse_checkpoint(void)
3033 {
3034 checkpoint_requested = 1;
3035 skip_optional_lf();
3036 }
3037
3038 static void parse_progress(void)
3039 {
3040 fwrite(command_buf.buf, 1, command_buf.len, stdout);
3041 fputc('\n', stdout);
3042 fflush(stdout);
3043 skip_optional_lf();
3044 }
3045
3046 static char* make_fast_import_path(const char *path)
3047 {
3048 if (!relative_marks_paths || is_absolute_path(path))
3049 return xstrdup(path);
3050 return git_pathdup("info/fast-import/%s", path);
3051 }
3052
3053 static void option_import_marks(const char *marks,
3054 int from_stream, int ignore_missing)
3055 {
3056 if (import_marks_file) {
3057 if (from_stream)
3058 die("Only one import-marks command allowed per stream");
3059
3060 /* read previous mark file */
3061 if(!import_marks_file_from_stream)
3062 read_marks();
3063 }
3064
3065 import_marks_file = make_fast_import_path(marks);
3066 import_marks_file_from_stream = from_stream;
3067 import_marks_file_ignore_missing = ignore_missing;
3068 }
3069
3070 static void option_date_format(const char *fmt)
3071 {
3072 if (!strcmp(fmt, "raw"))
3073 whenspec = WHENSPEC_RAW;
3074 else if (!strcmp(fmt, "rfc2822"))
3075 whenspec = WHENSPEC_RFC2822;
3076 else if (!strcmp(fmt, "now"))
3077 whenspec = WHENSPEC_NOW;
3078 else
3079 die("unknown --date-format argument %s", fmt);
3080 }
3081
3082 static unsigned long ulong_arg(const char *option, const char *arg)
3083 {
3084 char *endptr;
3085 unsigned long rv = strtoul(arg, &endptr, 0);
3086 if (strchr(arg, '-') || endptr == arg || *endptr)
3087 die("%s: argument must be a non-negative integer", option);
3088 return rv;
3089 }
3090
3091 static void option_depth(const char *depth)
3092 {
3093 max_depth = ulong_arg("--depth", depth);
3094 if (max_depth > MAX_DEPTH)
3095 die("--depth cannot exceed %u", MAX_DEPTH);
3096 }
3097
3098 static void option_active_branches(const char *branches)
3099 {
3100 max_active_branches = ulong_arg("--active-branches", branches);
3101 }
3102
3103 static void option_export_marks(const char *marks)
3104 {
3105 export_marks_file = make_fast_import_path(marks);
3106 }
3107
3108 static void option_cat_blob_fd(const char *fd)
3109 {
3110 unsigned long n = ulong_arg("--cat-blob-fd", fd);
3111 if (n > (unsigned long) INT_MAX)
3112 die("--cat-blob-fd cannot exceed %d", INT_MAX);
3113 cat_blob_fd = (int) n;
3114 }
3115
3116 static void option_export_pack_edges(const char *edges)
3117 {
3118 if (pack_edges)
3119 fclose(pack_edges);
3120 pack_edges = xfopen(edges, "a");
3121 }
3122
3123 static int parse_one_option(const char *option)
3124 {
3125 if (skip_prefix(option, "max-pack-size=", &option)) {
3126 unsigned long v;
3127 if (!git_parse_ulong(option, &v))
3128 return 0;
3129 if (v < 8192) {
3130 warning("max-pack-size is now in bytes, assuming --max-pack-size=%lum", v);
3131 v *= 1024 * 1024;
3132 } else if (v < 1024 * 1024) {
3133 warning("minimum max-pack-size is 1 MiB");
3134 v = 1024 * 1024;
3135 }
3136 max_packsize = v;
3137 } else if (skip_prefix(option, "big-file-threshold=", &option)) {
3138 unsigned long v;
3139 if (!git_parse_ulong(option, &v))
3140 return 0;
3141 big_file_threshold = v;
3142 } else if (skip_prefix(option, "depth=", &option)) {
3143 option_depth(option);
3144 } else if (skip_prefix(option, "active-branches=", &option)) {
3145 option_active_branches(option);
3146 } else if (skip_prefix(option, "export-pack-edges=", &option)) {
3147 option_export_pack_edges(option);
3148 } else if (!strcmp(option, "quiet")) {
3149 show_stats = 0;
3150 } else if (!strcmp(option, "stats")) {
3151 show_stats = 1;
3152 } else if (!strcmp(option, "allow-unsafe-features")) {
3153 ; /* already handled during early option parsing */
3154 } else {
3155 return 0;
3156 }
3157
3158 return 1;
3159 }
3160
3161 static void check_unsafe_feature(const char *feature, int from_stream)
3162 {
3163 if (from_stream && !allow_unsafe_features)
3164 die(_("feature '%s' forbidden in input without --allow-unsafe-features"),
3165 feature);
3166 }
3167
3168 static int parse_one_feature(const char *feature, int from_stream)
3169 {
3170 const char *arg;
3171
3172 if (skip_prefix(feature, "date-format=", &arg)) {
3173 option_date_format(arg);
3174 } else if (skip_prefix(feature, "import-marks=", &arg)) {
3175 check_unsafe_feature("import-marks", from_stream);
3176 option_import_marks(arg, from_stream, 0);
3177 } else if (skip_prefix(feature, "import-marks-if-exists=", &arg)) {
3178 check_unsafe_feature("import-marks-if-exists", from_stream);
3179 option_import_marks(arg, from_stream, 1);
3180 } else if (skip_prefix(feature, "export-marks=", &arg)) {
3181 check_unsafe_feature(feature, from_stream);
3182 option_export_marks(arg);
3183 } else if (!strcmp(feature, "get-mark")) {
3184 ; /* Don't die - this feature is supported */
3185 } else if (!strcmp(feature, "cat-blob")) {
3186 ; /* Don't die - this feature is supported */
3187 } else if (!strcmp(feature, "relative-marks")) {
3188 relative_marks_paths = 1;
3189 } else if (!strcmp(feature, "no-relative-marks")) {
3190 relative_marks_paths = 0;
3191 } else if (!strcmp(feature, "done")) {
3192 require_explicit_termination = 1;
3193 } else if (!strcmp(feature, "force")) {
3194 force_update = 1;
3195 } else if (!strcmp(feature, "notes") || !strcmp(feature, "ls")) {
3196 ; /* do nothing; we have the feature */
3197 } else {
3198 return 0;
3199 }
3200
3201 return 1;
3202 }
3203
3204 static void parse_feature(const char *feature)
3205 {
3206 if (seen_data_command)
3207 die("Got feature command '%s' after data command", feature);
3208
3209 if (parse_one_feature(feature, 1))
3210 return;
3211
3212 die("This version of fast-import does not support feature %s.", feature);
3213 }
3214
3215 static void parse_option(const char *option)
3216 {
3217 if (seen_data_command)
3218 die("Got option command '%s' after data command", option);
3219
3220 if (parse_one_option(option))
3221 return;
3222
3223 die("This version of fast-import does not support option: %s", option);
3224 }
3225
3226 static void git_pack_config(void)
3227 {
3228 int indexversion_value;
3229 int limit;
3230 unsigned long packsizelimit_value;
3231
3232 if (!git_config_get_ulong("pack.depth", &max_depth)) {
3233 if (max_depth > MAX_DEPTH)
3234 max_depth = MAX_DEPTH;
3235 }
3236 if (!git_config_get_int("pack.indexversion", &indexversion_value)) {
3237 pack_idx_opts.version = indexversion_value;
3238 if (pack_idx_opts.version > 2)
3239 git_die_config("pack.indexversion",
3240 "bad pack.indexversion=%"PRIu32, pack_idx_opts.version);
3241 }
3242 if (!git_config_get_ulong("pack.packsizelimit", &packsizelimit_value))
3243 max_packsize = packsizelimit_value;
3244
3245 if (!git_config_get_int("fastimport.unpacklimit", &limit))
3246 unpack_limit = limit;
3247 else if (!git_config_get_int("transfer.unpacklimit", &limit))
3248 unpack_limit = limit;
3249
3250 git_config(git_default_config, NULL);
3251 }
3252
3253 static const char fast_import_usage[] =
3254 "git fast-import [--date-format=<f>] [--max-pack-size=<n>] [--big-file-threshold=<n>] [--depth=<n>] [--active-branches=<n>] [--export-marks=<marks.file>]";
3255
3256 static void parse_argv(void)
3257 {
3258 unsigned int i;
3259
3260 for (i = 1; i < global_argc; i++) {
3261 const char *a = global_argv[i];
3262
3263 if (*a != '-' || !strcmp(a, "--"))
3264 break;
3265
3266 if (!skip_prefix(a, "--", &a))
3267 die("unknown option %s", a);
3268
3269 if (parse_one_option(a))
3270 continue;
3271
3272 if (parse_one_feature(a, 0))
3273 continue;
3274
3275 if (skip_prefix(a, "cat-blob-fd=", &a)) {
3276 option_cat_blob_fd(a);
3277 continue;
3278 }
3279
3280 die("unknown option --%s", a);
3281 }
3282 if (i != global_argc)
3283 usage(fast_import_usage);
3284
3285 seen_data_command = 1;
3286 if (import_marks_file)
3287 read_marks();
3288 }
3289
3290 int cmd_main(int argc, const char **argv)
3291 {
3292 unsigned int i;
3293
3294 if (argc == 2 && !strcmp(argv[1], "-h"))
3295 usage(fast_import_usage);
3296
3297 setup_git_directory();
3298 reset_pack_idx_option(&pack_idx_opts);
3299 git_pack_config();
3300
3301 alloc_objects(object_entry_alloc);
3302 strbuf_init(&command_buf, 0);
3303 atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
3304 branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
3305 avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
3306 marks = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
3307
3308 /*
3309 * We don't parse most options until after we've seen the set of
3310 * "feature" lines at the start of the stream (which allows the command
3311 * line to override stream data). But we must do an early parse of any
3312 * command-line options that impact how we interpret the feature lines.
3313 */
3314 for (i = 1; i < argc; i++) {
3315 const char *arg = argv[i];
3316 if (*arg != '-' || !strcmp(arg, "--"))
3317 break;
3318 if (!strcmp(arg, "--allow-unsafe-features"))
3319 allow_unsafe_features = 1;
3320 }
3321
3322 global_argc = argc;
3323 global_argv = argv;
3324
3325 rc_free = mem_pool_alloc(&fi_mem_pool, cmd_save * sizeof(*rc_free));
3326 for (i = 0; i < (cmd_save - 1); i++)
3327 rc_free[i].next = &rc_free[i + 1];
3328 rc_free[cmd_save - 1].next = NULL;
3329
3330 start_packfile();
3331 set_die_routine(die_nicely);
3332 set_checkpoint_signal();
3333 while (read_next_command() != EOF) {
3334 const char *v;
3335 if (!strcmp("blob", command_buf.buf))
3336 parse_new_blob();
3337 else if (skip_prefix(command_buf.buf, "ls ", &v))
3338 parse_ls(v, NULL);
3339 else if (skip_prefix(command_buf.buf, "commit ", &v))
3340 parse_new_commit(v);
3341 else if (skip_prefix(command_buf.buf, "tag ", &v))
3342 parse_new_tag(v);
3343 else if (skip_prefix(command_buf.buf, "reset ", &v))
3344 parse_reset_branch(v);
3345 else if (!strcmp("checkpoint", command_buf.buf))
3346 parse_checkpoint();
3347 else if (!strcmp("done", command_buf.buf))
3348 break;
3349 else if (starts_with(command_buf.buf, "progress "))
3350 parse_progress();
3351 else if (skip_prefix(command_buf.buf, "feature ", &v))
3352 parse_feature(v);
3353 else if (skip_prefix(command_buf.buf, "option git ", &v))
3354 parse_option(v);
3355 else if (starts_with(command_buf.buf, "option "))
3356 /* ignore non-git options*/;
3357 else
3358 die("Unsupported command: %s", command_buf.buf);
3359
3360 if (checkpoint_requested)
3361 checkpoint();
3362 }
3363
3364 /* argv hasn't been parsed yet, do so */
3365 if (!seen_data_command)
3366 parse_argv();
3367
3368 if (require_explicit_termination && feof(stdin))
3369 die("stream ends early");
3370
3371 end_packfile();
3372
3373 dump_branches();
3374 dump_tags();
3375 unkeep_all_packs();
3376 dump_marks();
3377
3378 if (pack_edges)
3379 fclose(pack_edges);
3380
3381 if (show_stats) {
3382 uintmax_t total_count = 0, duplicate_count = 0;
3383 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
3384 total_count += object_count_by_type[i];
3385 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
3386 duplicate_count += duplicate_count_by_type[i];
3387
3388 fprintf(stderr, "%s statistics:\n", argv[0]);
3389 fprintf(stderr, "---------------------------------------------------------------------\n");
3390 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
3391 fprintf(stderr, "Total objects: %10" PRIuMAX " (%10" PRIuMAX " duplicates )\n", total_count, duplicate_count);
3392 fprintf(stderr, " blobs : %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB], delta_count_attempts_by_type[OBJ_BLOB]);
3393 fprintf(stderr, " trees : %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE], delta_count_attempts_by_type[OBJ_TREE]);
3394 fprintf(stderr, " commits: %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT], delta_count_attempts_by_type[OBJ_COMMIT]);
3395 fprintf(stderr, " tags : %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG], delta_count_attempts_by_type[OBJ_TAG]);
3396 fprintf(stderr, "Total branches: %10lu (%10lu loads )\n", branch_count, branch_load_count);
3397 fprintf(stderr, " marks: %10" PRIuMAX " (%10" PRIuMAX " unique )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
3398 fprintf(stderr, " atoms: %10u\n", atom_cnt);
3399 fprintf(stderr, "Memory total: %10" PRIuMAX " KiB\n", (tree_entry_allocd + fi_mem_pool.pool_alloc + alloc_count*sizeof(struct object_entry))/1024);
3400 fprintf(stderr, " pools: %10lu KiB\n", (unsigned long)((tree_entry_allocd + fi_mem_pool.pool_alloc) /1024));
3401 fprintf(stderr, " objects: %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
3402 fprintf(stderr, "---------------------------------------------------------------------\n");
3403 pack_report();
3404 fprintf(stderr, "---------------------------------------------------------------------\n");
3405 fprintf(stderr, "\n");
3406 }
3407
3408 return failure ? 1 : 0;
3409 }
Alternate source for your git.git needs