?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
fast-import: only allow cat-blob requests where it makes sense
[git.git] / fast-import.c
blob338db61e6e7bbb2ce41cf2057bf1c2df2496a09c
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
214 /* Signal handling */
215 static volatile sig_atomic_t checkpoint_requested;
216
217 /* Where to write output of cat-blob commands */
218 static int cat_blob_fd = STDOUT_FILENO;
219
220 static void parse_argv(void);
221 static void parse_get_mark(const char *p);
222 static void parse_cat_blob(const char *p);
223 static void parse_ls(const char *p, struct branch *b);
224
225 static void write_branch_report(FILE *rpt, struct branch *b)
226 {
227 fprintf(rpt, "%s:\n", b->name);
228
229 fprintf(rpt, " status :");
230 if (b->active)
231 fputs(" active", rpt);
232 if (b->branch_tree.tree)
233 fputs(" loaded", rpt);
234 if (is_null_oid(&b->branch_tree.versions[1].oid))
235 fputs(" dirty", rpt);
236 fputc('\n', rpt);
237
238 fprintf(rpt, " tip commit : %s\n", oid_to_hex(&b->oid));
239 fprintf(rpt, " old tree : %s\n",
240 oid_to_hex(&b->branch_tree.versions[0].oid));
241 fprintf(rpt, " cur tree : %s\n",
242 oid_to_hex(&b->branch_tree.versions[1].oid));
243 fprintf(rpt, " commit clock: %" PRIuMAX "\n", b->last_commit);
244
245 fputs(" last pack : ", rpt);
246 if (b->pack_id < MAX_PACK_ID)
247 fprintf(rpt, "%u", b->pack_id);
248 fputc('\n', rpt);
249
250 fputc('\n', rpt);
251 }
252
253 static void dump_marks_helper(FILE *, uintmax_t, struct mark_set *);
254
255 static void write_crash_report(const char *err)
256 {
257 char *loc = git_pathdup("fast_import_crash_%"PRIuMAX, (uintmax_t) getpid());
258 FILE *rpt = fopen(loc, "w");
259 struct branch *b;
260 unsigned long lu;
261 struct recent_command *rc;
262
263 if (!rpt) {
264 error_errno("can't write crash report %s", loc);
265 free(loc);
266 return;
267 }
268
269 fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
270
271 fprintf(rpt, "fast-import crash report:\n");
272 fprintf(rpt, " fast-import process: %"PRIuMAX"\n", (uintmax_t) getpid());
273 fprintf(rpt, " parent process : %"PRIuMAX"\n", (uintmax_t) getppid());
274 fprintf(rpt, " at %s\n", show_date(time(NULL), 0, DATE_MODE(ISO8601)));
275 fputc('\n', rpt);
276
277 fputs("fatal: ", rpt);
278 fputs(err, rpt);
279 fputc('\n', rpt);
280
281 fputc('\n', rpt);
282 fputs("Most Recent Commands Before Crash\n", rpt);
283 fputs("---------------------------------\n", rpt);
284 for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
285 if (rc->next == &cmd_hist)
286 fputs("* ", rpt);
287 else
288 fputs(" ", rpt);
289 fputs(rc->buf, rpt);
290 fputc('\n', rpt);
291 }
292
293 fputc('\n', rpt);
294 fputs("Active Branch LRU\n", rpt);
295 fputs("-----------------\n", rpt);
296 fprintf(rpt, " active_branches = %lu cur, %lu max\n",
297 cur_active_branches,
298 max_active_branches);
299 fputc('\n', rpt);
300 fputs(" pos clock name\n", rpt);
301 fputs(" ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
302 for (b = active_branches, lu = 0; b; b = b->active_next_branch)
303 fprintf(rpt, " %2lu) %6" PRIuMAX" %s\n",
304 ++lu, b->last_commit, b->name);
305
306 fputc('\n', rpt);
307 fputs("Inactive Branches\n", rpt);
308 fputs("-----------------\n", rpt);
309 for (lu = 0; lu < branch_table_sz; lu++) {
310 for (b = branch_table[lu]; b; b = b->table_next_branch)
311 write_branch_report(rpt, b);
312 }
313
314 if (first_tag) {
315 struct tag *tg;
316 fputc('\n', rpt);
317 fputs("Annotated Tags\n", rpt);
318 fputs("--------------\n", rpt);
319 for (tg = first_tag; tg; tg = tg->next_tag) {
320 fputs(oid_to_hex(&tg->oid), rpt);
321 fputc(' ', rpt);
322 fputs(tg->name, rpt);
323 fputc('\n', rpt);
324 }
325 }
326
327 fputc('\n', rpt);
328 fputs("Marks\n", rpt);
329 fputs("-----\n", rpt);
330 if (export_marks_file)
331 fprintf(rpt, " exported to %s\n", export_marks_file);
332 else
333 dump_marks_helper(rpt, 0, marks);
334
335 fputc('\n', rpt);
336 fputs("-------------------\n", rpt);
337 fputs("END OF CRASH REPORT\n", rpt);
338 fclose(rpt);
339 free(loc);
340 }
341
342 static void end_packfile(void);
343 static void unkeep_all_packs(void);
344 static void dump_marks(void);
345
346 static NORETURN void die_nicely(const char *err, va_list params)
347 {
348 static int zombie;
349 char message[2 * PATH_MAX];
350
351 vsnprintf(message, sizeof(message), err, params);
352 fputs("fatal: ", stderr);
353 fputs(message, stderr);
354 fputc('\n', stderr);
355
356 if (!zombie) {
357 zombie = 1;
358 write_crash_report(message);
359 end_packfile();
360 unkeep_all_packs();
361 dump_marks();
362 }
363 exit(128);
364 }
365
366 #ifndef SIGUSR1 /* Windows, for example */
367
368 static void set_checkpoint_signal(void)
369 {
370 }
371
372 #else
373
374 static void checkpoint_signal(int signo)
375 {
376 checkpoint_requested = 1;
377 }
378
379 static void set_checkpoint_signal(void)
380 {
381 struct sigaction sa;
382
383 memset(&sa, 0, sizeof(sa));
384 sa.sa_handler = checkpoint_signal;
385 sigemptyset(&sa.sa_mask);
386 sa.sa_flags = SA_RESTART;
387 sigaction(SIGUSR1, &sa, NULL);
388 }
389
390 #endif
391
392 static void alloc_objects(unsigned int cnt)
393 {
394 struct object_entry_pool *b;
395
396 b = xmalloc(sizeof(struct object_entry_pool)
397 + cnt * sizeof(struct object_entry));
398 b->next_pool = blocks;
399 b->next_free = b->entries;
400 b->end = b->entries + cnt;
401 blocks = b;
402 alloc_count += cnt;
403 }
404
405 static struct object_entry *new_object(struct object_id *oid)
406 {
407 struct object_entry *e;
408
409 if (blocks->next_free == blocks->end)
410 alloc_objects(object_entry_alloc);
411
412 e = blocks->next_free++;
413 oidcpy(&e->idx.oid, oid);
414 return e;
415 }
416
417 static struct object_entry *find_object(struct object_id *oid)
418 {
419 unsigned int h = oid->hash[0] << 8 | oid->hash[1];
420 struct object_entry *e;
421 for (e = object_table[h]; e; e = e->next)
422 if (oideq(oid, &e->idx.oid))
423 return e;
424 return NULL;
425 }
426
427 static struct object_entry *insert_object(struct object_id *oid)
428 {
429 unsigned int h = oid->hash[0] << 8 | oid->hash[1];
430 struct object_entry *e = object_table[h];
431
432 while (e) {
433 if (oideq(oid, &e->idx.oid))
434 return e;
435 e = e->next;
436 }
437
438 e = new_object(oid);
439 e->next = object_table[h];
440 e->idx.offset = 0;
441 object_table[h] = e;
442 return e;
443 }
444
445 static void invalidate_pack_id(unsigned int id)
446 {
447 unsigned int h;
448 unsigned long lu;
449 struct tag *t;
450
451 for (h = 0; h < ARRAY_SIZE(object_table); h++) {
452 struct object_entry *e;
453
454 for (e = object_table[h]; e; e = e->next)
455 if (e->pack_id == id)
456 e->pack_id = MAX_PACK_ID;
457 }
458
459 for (lu = 0; lu < branch_table_sz; lu++) {
460 struct branch *b;
461
462 for (b = branch_table[lu]; b; b = b->table_next_branch)
463 if (b->pack_id == id)
464 b->pack_id = MAX_PACK_ID;
465 }
466
467 for (t = first_tag; t; t = t->next_tag)
468 if (t->pack_id == id)
469 t->pack_id = MAX_PACK_ID;
470 }
471
472 static unsigned int hc_str(const char *s, size_t len)
473 {
474 unsigned int r = 0;
475 while (len-- > 0)
476 r = r * 31 + *s++;
477 return r;
478 }
479
480 static char *pool_strdup(const char *s)
481 {
482 size_t len = strlen(s) + 1;
483 char *r = mem_pool_alloc(&fi_mem_pool, len);
484 memcpy(r, s, len);
485 return r;
486 }
487
488 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
489 {
490 struct mark_set *s = marks;
491 while ((idnum >> s->shift) >= 1024) {
492 s = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
493 s->shift = marks->shift + 10;
494 s->data.sets[0] = marks;
495 marks = s;
496 }
497 while (s->shift) {
498 uintmax_t i = idnum >> s->shift;
499 idnum -= i << s->shift;
500 if (!s->data.sets[i]) {
501 s->data.sets[i] = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
502 s->data.sets[i]->shift = s->shift - 10;
503 }
504 s = s->data.sets[i];
505 }
506 if (!s->data.marked[idnum])
507 marks_set_count++;
508 s->data.marked[idnum] = oe;
509 }
510
511 static struct object_entry *find_mark(uintmax_t idnum)
512 {
513 uintmax_t orig_idnum = idnum;
514 struct mark_set *s = marks;
515 struct object_entry *oe = NULL;
516 if ((idnum >> s->shift) < 1024) {
517 while (s && s->shift) {
518 uintmax_t i = idnum >> s->shift;
519 idnum -= i << s->shift;
520 s = s->data.sets[i];
521 }
522 if (s)
523 oe = s->data.marked[idnum];
524 }
525 if (!oe)
526 die("mark :%" PRIuMAX " not declared", orig_idnum);
527 return oe;
528 }
529
530 static struct atom_str *to_atom(const char *s, unsigned short len)
531 {
532 unsigned int hc = hc_str(s, len) % atom_table_sz;
533 struct atom_str *c;
534
535 for (c = atom_table[hc]; c; c = c->next_atom)
536 if (c->str_len == len && !strncmp(s, c->str_dat, len))
537 return c;
538
539 c = mem_pool_alloc(&fi_mem_pool, sizeof(struct atom_str) + len + 1);
540 c->str_len = len;
541 memcpy(c->str_dat, s, len);
542 c->str_dat[len] = 0;
543 c->next_atom = atom_table[hc];
544 atom_table[hc] = c;
545 atom_cnt++;
546 return c;
547 }
548
549 static struct branch *lookup_branch(const char *name)
550 {
551 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
552 struct branch *b;
553
554 for (b = branch_table[hc]; b; b = b->table_next_branch)
555 if (!strcmp(name, b->name))
556 return b;
557 return NULL;
558 }
559
560 static struct branch *new_branch(const char *name)
561 {
562 unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
563 struct branch *b = lookup_branch(name);
564
565 if (b)
566 die("Invalid attempt to create duplicate branch: %s", name);
567 if (check_refname_format(name, REFNAME_ALLOW_ONELEVEL))
568 die("Branch name doesn't conform to GIT standards: %s", name);
569
570 b = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct branch));
571 b->name = pool_strdup(name);
572 b->table_next_branch = branch_table[hc];
573 b->branch_tree.versions[0].mode = S_IFDIR;
574 b->branch_tree.versions[1].mode = S_IFDIR;
575 b->num_notes = 0;
576 b->active = 0;
577 b->pack_id = MAX_PACK_ID;
578 branch_table[hc] = b;
579 branch_count++;
580 return b;
581 }
582
583 static unsigned int hc_entries(unsigned int cnt)
584 {
585 cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
586 return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
587 }
588
589 static struct tree_content *new_tree_content(unsigned int cnt)
590 {
591 struct avail_tree_content *f, *l = NULL;
592 struct tree_content *t;
593 unsigned int hc = hc_entries(cnt);
594
595 for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
596 if (f->entry_capacity >= cnt)
597 break;
598
599 if (f) {
600 if (l)
601 l->next_avail = f->next_avail;
602 else
603 avail_tree_table[hc] = f->next_avail;
604 } else {
605 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
606 f = mem_pool_alloc(&fi_mem_pool, sizeof(*t) + sizeof(t->entries[0]) * cnt);
607 f->entry_capacity = cnt;
608 }
609
610 t = (struct tree_content*)f;
611 t->entry_count = 0;
612 t->delta_depth = 0;
613 return t;
614 }
615
616 static void release_tree_entry(struct tree_entry *e);
617 static void release_tree_content(struct tree_content *t)
618 {
619 struct avail_tree_content *f = (struct avail_tree_content*)t;
620 unsigned int hc = hc_entries(f->entry_capacity);
621 f->next_avail = avail_tree_table[hc];
622 avail_tree_table[hc] = f;
623 }
624
625 static void release_tree_content_recursive(struct tree_content *t)
626 {
627 unsigned int i;
628 for (i = 0; i < t->entry_count; i++)
629 release_tree_entry(t->entries[i]);
630 release_tree_content(t);
631 }
632
633 static struct tree_content *grow_tree_content(
634 struct tree_content *t,
635 int amt)
636 {
637 struct tree_content *r = new_tree_content(t->entry_count + amt);
638 r->entry_count = t->entry_count;
639 r->delta_depth = t->delta_depth;
640 memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
641 release_tree_content(t);
642 return r;
643 }
644
645 static struct tree_entry *new_tree_entry(void)
646 {
647 struct tree_entry *e;
648
649 if (!avail_tree_entry) {
650 unsigned int n = tree_entry_alloc;
651 tree_entry_allocd += n * sizeof(struct tree_entry);
652 ALLOC_ARRAY(e, n);
653 avail_tree_entry = e;
654 while (n-- > 1) {
655 *((void**)e) = e + 1;
656 e++;
657 }
658 *((void**)e) = NULL;
659 }
660
661 e = avail_tree_entry;
662 avail_tree_entry = *((void**)e);
663 return e;
664 }
665
666 static void release_tree_entry(struct tree_entry *e)
667 {
668 if (e->tree)
669 release_tree_content_recursive(e->tree);
670 *((void**)e) = avail_tree_entry;
671 avail_tree_entry = e;
672 }
673
674 static struct tree_content *dup_tree_content(struct tree_content *s)
675 {
676 struct tree_content *d;
677 struct tree_entry *a, *b;
678 unsigned int i;
679
680 if (!s)
681 return NULL;
682 d = new_tree_content(s->entry_count);
683 for (i = 0; i < s->entry_count; i++) {
684 a = s->entries[i];
685 b = new_tree_entry();
686 memcpy(b, a, sizeof(*a));
687 if (a->tree && is_null_oid(&b->versions[1].oid))
688 b->tree = dup_tree_content(a->tree);
689 else
690 b->tree = NULL;
691 d->entries[i] = b;
692 }
693 d->entry_count = s->entry_count;
694 d->delta_depth = s->delta_depth;
695
696 return d;
697 }
698
699 static void start_packfile(void)
700 {
701 struct strbuf tmp_file = STRBUF_INIT;
702 struct packed_git *p;
703 struct pack_header hdr;
704 int pack_fd;
705
706 pack_fd = odb_mkstemp(&tmp_file, "pack/tmp_pack_XXXXXX");
707 FLEX_ALLOC_STR(p, pack_name, tmp_file.buf);
708 strbuf_release(&tmp_file);
709
710 p->pack_fd = pack_fd;
711 p->do_not_close = 1;
712 pack_file = hashfd(pack_fd, p->pack_name);
713
714 hdr.hdr_signature = htonl(PACK_SIGNATURE);
715 hdr.hdr_version = htonl(2);
716 hdr.hdr_entries = 0;
717 hashwrite(pack_file, &hdr, sizeof(hdr));
718
719 pack_data = p;
720 pack_size = sizeof(hdr);
721 object_count = 0;
722
723 REALLOC_ARRAY(all_packs, pack_id + 1);
724 all_packs[pack_id] = p;
725 }
726
727 static const char *create_index(void)
728 {
729 const char *tmpfile;
730 struct pack_idx_entry **idx, **c, **last;
731 struct object_entry *e;
732 struct object_entry_pool *o;
733
734 /* Build the table of object IDs. */
735 ALLOC_ARRAY(idx, object_count);
736 c = idx;
737 for (o = blocks; o; o = o->next_pool)
738 for (e = o->next_free; e-- != o->entries;)
739 if (pack_id == e->pack_id)
740 *c++ = &e->idx;
741 last = idx + object_count;
742 if (c != last)
743 die("internal consistency error creating the index");
744
745 tmpfile = write_idx_file(NULL, idx, object_count, &pack_idx_opts, pack_data->sha1);
746 free(idx);
747 return tmpfile;
748 }
749
750 static char *keep_pack(const char *curr_index_name)
751 {
752 static const char *keep_msg = "fast-import";
753 struct strbuf name = STRBUF_INIT;
754 int keep_fd;
755
756 odb_pack_name(&name, pack_data->sha1, "keep");
757 keep_fd = odb_pack_keep(name.buf);
758 if (keep_fd < 0)
759 die_errno("cannot create keep file");
760 write_or_die(keep_fd, keep_msg, strlen(keep_msg));
761 if (close(keep_fd))
762 die_errno("failed to write keep file");
763
764 odb_pack_name(&name, pack_data->sha1, "pack");
765 if (finalize_object_file(pack_data->pack_name, name.buf))
766 die("cannot store pack file");
767
768 odb_pack_name(&name, pack_data->sha1, "idx");
769 if (finalize_object_file(curr_index_name, name.buf))
770 die("cannot store index file");
771 free((void *)curr_index_name);
772 return strbuf_detach(&name, NULL);
773 }
774
775 static void unkeep_all_packs(void)
776 {
777 struct strbuf name = STRBUF_INIT;
778 int k;
779
780 for (k = 0; k < pack_id; k++) {
781 struct packed_git *p = all_packs[k];
782 odb_pack_name(&name, p->sha1, "keep");
783 unlink_or_warn(name.buf);
784 }
785 strbuf_release(&name);
786 }
787
788 static int loosen_small_pack(const struct packed_git *p)
789 {
790 struct child_process unpack = CHILD_PROCESS_INIT;
791
792 if (lseek(p->pack_fd, 0, SEEK_SET) < 0)
793 die_errno("Failed seeking to start of '%s'", p->pack_name);
794
795 unpack.in = p->pack_fd;
796 unpack.git_cmd = 1;
797 unpack.stdout_to_stderr = 1;
798 argv_array_push(&unpack.args, "unpack-objects");
799 if (!show_stats)
800 argv_array_push(&unpack.args, "-q");
801
802 return run_command(&unpack);
803 }
804
805 static void end_packfile(void)
806 {
807 static int running;
808
809 if (running || !pack_data)
810 return;
811
812 running = 1;
813 clear_delta_base_cache();
814 if (object_count) {
815 struct packed_git *new_p;
816 struct object_id cur_pack_oid;
817 char *idx_name;
818 int i;
819 struct branch *b;
820 struct tag *t;
821
822 close_pack_windows(pack_data);
823 finalize_hashfile(pack_file, cur_pack_oid.hash, 0);
824 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
825 pack_data->pack_name, object_count,
826 cur_pack_oid.hash, pack_size);
827
828 if (object_count <= unpack_limit) {
829 if (!loosen_small_pack(pack_data)) {
830 invalidate_pack_id(pack_id);
831 goto discard_pack;
832 }
833 }
834
835 close(pack_data->pack_fd);
836 idx_name = keep_pack(create_index());
837
838 /* Register the packfile with core git's machinery. */
839 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
840 if (!new_p)
841 die("core git rejected index %s", idx_name);
842 all_packs[pack_id] = new_p;
843 install_packed_git(the_repository, new_p);
844 free(idx_name);
845
846 /* Print the boundary */
847 if (pack_edges) {
848 fprintf(pack_edges, "%s:", new_p->pack_name);
849 for (i = 0; i < branch_table_sz; i++) {
850 for (b = branch_table[i]; b; b = b->table_next_branch) {
851 if (b->pack_id == pack_id)
852 fprintf(pack_edges, " %s",
853 oid_to_hex(&b->oid));
854 }
855 }
856 for (t = first_tag; t; t = t->next_tag) {
857 if (t->pack_id == pack_id)
858 fprintf(pack_edges, " %s",
859 oid_to_hex(&t->oid));
860 }
861 fputc('\n', pack_edges);
862 fflush(pack_edges);
863 }
864
865 pack_id++;
866 }
867 else {
868 discard_pack:
869 close(pack_data->pack_fd);
870 unlink_or_warn(pack_data->pack_name);
871 }
872 FREE_AND_NULL(pack_data);
873 running = 0;
874
875 /* We can't carry a delta across packfiles. */
876 strbuf_release(&last_blob.data);
877 last_blob.offset = 0;
878 last_blob.depth = 0;
879 }
880
881 static void cycle_packfile(void)
882 {
883 end_packfile();
884 start_packfile();
885 }
886
887 static int store_object(
888 enum object_type type,
889 struct strbuf *dat,
890 struct last_object *last,
891 struct object_id *oidout,
892 uintmax_t mark)
893 {
894 void *out, *delta;
895 struct object_entry *e;
896 unsigned char hdr[96];
897 struct object_id oid;
898 unsigned long hdrlen, deltalen;
899 git_hash_ctx c;
900 git_zstream s;
901
902 hdrlen = xsnprintf((char *)hdr, sizeof(hdr), "%s %lu",
903 type_name(type), (unsigned long)dat->len) + 1;
904 the_hash_algo->init_fn(&c);
905 the_hash_algo->update_fn(&c, hdr, hdrlen);
906 the_hash_algo->update_fn(&c, dat->buf, dat->len);
907 the_hash_algo->final_fn(oid.hash, &c);
908 if (oidout)
909 oidcpy(oidout, &oid);
910
911 e = insert_object(&oid);
912 if (mark)
913 insert_mark(mark, e);
914 if (e->idx.offset) {
915 duplicate_count_by_type[type]++;
916 return 1;
917 } else if (find_sha1_pack(oid.hash,
918 get_all_packs(the_repository))) {
919 e->type = type;
920 e->pack_id = MAX_PACK_ID;
921 e->idx.offset = 1; /* just not zero! */
922 duplicate_count_by_type[type]++;
923 return 1;
924 }
925
926 if (last && last->data.len && last->data.buf && last->depth < max_depth
927 && dat->len > the_hash_algo->rawsz) {
928
929 delta_count_attempts_by_type[type]++;
930 delta = diff_delta(last->data.buf, last->data.len,
931 dat->buf, dat->len,
932 &deltalen, dat->len - the_hash_algo->rawsz);
933 } else
934 delta = NULL;
935
936 git_deflate_init(&s, pack_compression_level);
937 if (delta) {
938 s.next_in = delta;
939 s.avail_in = deltalen;
940 } else {
941 s.next_in = (void *)dat->buf;
942 s.avail_in = dat->len;
943 }
944 s.avail_out = git_deflate_bound(&s, s.avail_in);
945 s.next_out = out = xmalloc(s.avail_out);
946 while (git_deflate(&s, Z_FINISH) == Z_OK)
947 ; /* nothing */
948 git_deflate_end(&s);
949
950 /* Determine if we should auto-checkpoint. */
951 if ((max_packsize && (pack_size + 60 + s.total_out) > max_packsize)
952 || (pack_size + 60 + s.total_out) < pack_size) {
953
954 /* This new object needs to *not* have the current pack_id. */
955 e->pack_id = pack_id + 1;
956 cycle_packfile();
957
958 /* We cannot carry a delta into the new pack. */
959 if (delta) {
960 FREE_AND_NULL(delta);
961
962 git_deflate_init(&s, pack_compression_level);
963 s.next_in = (void *)dat->buf;
964 s.avail_in = dat->len;
965 s.avail_out = git_deflate_bound(&s, s.avail_in);
966 s.next_out = out = xrealloc(out, s.avail_out);
967 while (git_deflate(&s, Z_FINISH) == Z_OK)
968 ; /* nothing */
969 git_deflate_end(&s);
970 }
971 }
972
973 e->type = type;
974 e->pack_id = pack_id;
975 e->idx.offset = pack_size;
976 object_count++;
977 object_count_by_type[type]++;
978
979 crc32_begin(pack_file);
980
981 if (delta) {
982 off_t ofs = e->idx.offset - last->offset;
983 unsigned pos = sizeof(hdr) - 1;
984
985 delta_count_by_type[type]++;
986 e->depth = last->depth + 1;
987
988 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
989 OBJ_OFS_DELTA, deltalen);
990 hashwrite(pack_file, hdr, hdrlen);
991 pack_size += hdrlen;
992
993 hdr[pos] = ofs & 127;
994 while (ofs >>= 7)
995 hdr[--pos] = 128 | (--ofs & 127);
996 hashwrite(pack_file, hdr + pos, sizeof(hdr) - pos);
997 pack_size += sizeof(hdr) - pos;
998 } else {
999 e->depth = 0;
1000 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
1001 type, dat->len);
1002 hashwrite(pack_file, hdr, hdrlen);
1003 pack_size += hdrlen;
1004 }
1005
1006 hashwrite(pack_file, out, s.total_out);
1007 pack_size += s.total_out;
1008
1009 e->idx.crc32 = crc32_end(pack_file);
1010
1011 free(out);
1012 free(delta);
1013 if (last) {
1014 if (last->no_swap) {
1015 last->data = *dat;
1016 } else {
1017 strbuf_swap(&last->data, dat);
1018 }
1019 last->offset = e->idx.offset;
1020 last->depth = e->depth;
1021 }
1022 return 0;
1023 }
1024
1025 static void truncate_pack(struct hashfile_checkpoint *checkpoint)
1026 {
1027 if (hashfile_truncate(pack_file, checkpoint))
1028 die_errno("cannot truncate pack to skip duplicate");
1029 pack_size = checkpoint->offset;
1030 }
1031
1032 static void stream_blob(uintmax_t len, struct object_id *oidout, uintmax_t mark)
1033 {
1034 size_t in_sz = 64 * 1024, out_sz = 64 * 1024;
1035 unsigned char *in_buf = xmalloc(in_sz);
1036 unsigned char *out_buf = xmalloc(out_sz);
1037 struct object_entry *e;
1038 struct object_id oid;
1039 unsigned long hdrlen;
1040 off_t offset;
1041 git_hash_ctx c;
1042 git_zstream s;
1043 struct hashfile_checkpoint checkpoint;
1044 int status = Z_OK;
1045
1046 /* Determine if we should auto-checkpoint. */
1047 if ((max_packsize && (pack_size + 60 + len) > max_packsize)
1048 || (pack_size + 60 + len) < pack_size)
1049 cycle_packfile();
1050
1051 hashfile_checkpoint(pack_file, &checkpoint);
1052 offset = checkpoint.offset;
1053
1054 hdrlen = xsnprintf((char *)out_buf, out_sz, "blob %" PRIuMAX, len) + 1;
1055
1056 the_hash_algo->init_fn(&c);
1057 the_hash_algo->update_fn(&c, out_buf, hdrlen);
1058
1059 crc32_begin(pack_file);
1060
1061 git_deflate_init(&s, pack_compression_level);
1062
1063 hdrlen = encode_in_pack_object_header(out_buf, out_sz, OBJ_BLOB, len);
1064
1065 s.next_out = out_buf + hdrlen;
1066 s.avail_out = out_sz - hdrlen;
1067
1068 while (status != Z_STREAM_END) {
1069 if (0 < len && !s.avail_in) {
1070 size_t cnt = in_sz < len ? in_sz : (size_t)len;
1071 size_t n = fread(in_buf, 1, cnt, stdin);
1072 if (!n && feof(stdin))
1073 die("EOF in data (%" PRIuMAX " bytes remaining)", len);
1074
1075 the_hash_algo->update_fn(&c, in_buf, n);
1076 s.next_in = in_buf;
1077 s.avail_in = n;
1078 len -= n;
1079 }
1080
1081 status = git_deflate(&s, len ? 0 : Z_FINISH);
1082
1083 if (!s.avail_out || status == Z_STREAM_END) {
1084 size_t n = s.next_out - out_buf;
1085 hashwrite(pack_file, out_buf, n);
1086 pack_size += n;
1087 s.next_out = out_buf;
1088 s.avail_out = out_sz;
1089 }
1090
1091 switch (status) {
1092 case Z_OK:
1093 case Z_BUF_ERROR:
1094 case Z_STREAM_END:
1095 continue;
1096 default:
1097 die("unexpected deflate failure: %d", status);
1098 }
1099 }
1100 git_deflate_end(&s);
1101 the_hash_algo->final_fn(oid.hash, &c);
1102
1103 if (oidout)
1104 oidcpy(oidout, &oid);
1105
1106 e = insert_object(&oid);
1107
1108 if (mark)
1109 insert_mark(mark, e);
1110
1111 if (e->idx.offset) {
1112 duplicate_count_by_type[OBJ_BLOB]++;
1113 truncate_pack(&checkpoint);
1114
1115 } else if (find_sha1_pack(oid.hash,
1116 get_all_packs(the_repository))) {
1117 e->type = OBJ_BLOB;
1118 e->pack_id = MAX_PACK_ID;
1119 e->idx.offset = 1; /* just not zero! */
1120 duplicate_count_by_type[OBJ_BLOB]++;
1121 truncate_pack(&checkpoint);
1122
1123 } else {
1124 e->depth = 0;
1125 e->type = OBJ_BLOB;
1126 e->pack_id = pack_id;
1127 e->idx.offset = offset;
1128 e->idx.crc32 = crc32_end(pack_file);
1129 object_count++;
1130 object_count_by_type[OBJ_BLOB]++;
1131 }
1132
1133 free(in_buf);
1134 free(out_buf);
1135 }
1136
1137 /* All calls must be guarded by find_object() or find_mark() to
1138 * ensure the 'struct object_entry' passed was written by this
1139 * process instance. We unpack the entry by the offset, avoiding
1140 * the need for the corresponding .idx file. This unpacking rule
1141 * works because we only use OBJ_REF_DELTA within the packfiles
1142 * created by fast-import.
1143 *
1144 * oe must not be NULL. Such an oe usually comes from giving
1145 * an unknown SHA-1 to find_object() or an undefined mark to
1146 * find_mark(). Callers must test for this condition and use
1147 * the standard read_sha1_file() when it happens.
1148 *
1149 * oe->pack_id must not be MAX_PACK_ID. Such an oe is usually from
1150 * find_mark(), where the mark was reloaded from an existing marks
1151 * file and is referencing an object that this fast-import process
1152 * instance did not write out to a packfile. Callers must test for
1153 * this condition and use read_sha1_file() instead.
1154 */
1155 static void *gfi_unpack_entry(
1156 struct object_entry *oe,
1157 unsigned long *sizep)
1158 {
1159 enum object_type type;
1160 struct packed_git *p = all_packs[oe->pack_id];
1161 if (p == pack_data && p->pack_size < (pack_size + the_hash_algo->rawsz)) {
1162 /* The object is stored in the packfile we are writing to
1163 * and we have modified it since the last time we scanned
1164 * back to read a previously written object. If an old
1165 * window covered [p->pack_size, p->pack_size + rawsz) its
1166 * data is stale and is not valid. Closing all windows
1167 * and updating the packfile length ensures we can read
1168 * the newly written data.
1169 */
1170 close_pack_windows(p);
1171 hashflush(pack_file);
1172
1173 /* We have to offer rawsz bytes additional on the end of
1174 * the packfile as the core unpacker code assumes the
1175 * footer is present at the file end and must promise
1176 * at least rawsz bytes within any window it maps. But
1177 * we don't actually create the footer here.
1178 */
1179 p->pack_size = pack_size + the_hash_algo->rawsz;
1180 }
1181 return unpack_entry(the_repository, p, oe->idx.offset, &type, sizep);
1182 }
1183
1184 static const char *get_mode(const char *str, uint16_t *modep)
1185 {
1186 unsigned char c;
1187 uint16_t mode = 0;
1188
1189 while ((c = *str++) != ' ') {
1190 if (c < '0' || c > '7')
1191 return NULL;
1192 mode = (mode << 3) + (c - '0');
1193 }
1194 *modep = mode;
1195 return str;
1196 }
1197
1198 static void load_tree(struct tree_entry *root)
1199 {
1200 struct object_id *oid = &root->versions[1].oid;
1201 struct object_entry *myoe;
1202 struct tree_content *t;
1203 unsigned long size;
1204 char *buf;
1205 const char *c;
1206
1207 root->tree = t = new_tree_content(8);
1208 if (is_null_oid(oid))
1209 return;
1210
1211 myoe = find_object(oid);
1212 if (myoe && myoe->pack_id != MAX_PACK_ID) {
1213 if (myoe->type != OBJ_TREE)
1214 die("Not a tree: %s", oid_to_hex(oid));
1215 t->delta_depth = myoe->depth;
1216 buf = gfi_unpack_entry(myoe, &size);
1217 if (!buf)
1218 die("Can't load tree %s", oid_to_hex(oid));
1219 } else {
1220 enum object_type type;
1221 buf = read_object_file(oid, &type, &size);
1222 if (!buf || type != OBJ_TREE)
1223 die("Can't load tree %s", oid_to_hex(oid));
1224 }
1225
1226 c = buf;
1227 while (c != (buf + size)) {
1228 struct tree_entry *e = new_tree_entry();
1229
1230 if (t->entry_count == t->entry_capacity)
1231 root->tree = t = grow_tree_content(t, t->entry_count);
1232 t->entries[t->entry_count++] = e;
1233
1234 e->tree = NULL;
1235 c = get_mode(c, &e->versions[1].mode);
1236 if (!c)
1237 die("Corrupt mode in %s", oid_to_hex(oid));
1238 e->versions[0].mode = e->versions[1].mode;
1239 e->name = to_atom(c, strlen(c));
1240 c += e->name->str_len + 1;
1241 hashcpy(e->versions[0].oid.hash, (unsigned char *)c);
1242 hashcpy(e->versions[1].oid.hash, (unsigned char *)c);
1243 c += GIT_SHA1_RAWSZ;
1244 }
1245 free(buf);
1246 }
1247
1248 static int tecmp0 (const void *_a, const void *_b)
1249 {
1250 struct tree_entry *a = *((struct tree_entry**)_a);
1251 struct tree_entry *b = *((struct tree_entry**)_b);
1252 return base_name_compare(
1253 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1254 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1255 }
1256
1257 static int tecmp1 (const void *_a, const void *_b)
1258 {
1259 struct tree_entry *a = *((struct tree_entry**)_a);
1260 struct tree_entry *b = *((struct tree_entry**)_b);
1261 return base_name_compare(
1262 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1263 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1264 }
1265
1266 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1267 {
1268 size_t maxlen = 0;
1269 unsigned int i;
1270
1271 if (!v)
1272 QSORT(t->entries, t->entry_count, tecmp0);
1273 else
1274 QSORT(t->entries, t->entry_count, tecmp1);
1275
1276 for (i = 0; i < t->entry_count; i++) {
1277 if (t->entries[i]->versions[v].mode)
1278 maxlen += t->entries[i]->name->str_len + 34;
1279 }
1280
1281 strbuf_reset(b);
1282 strbuf_grow(b, maxlen);
1283 for (i = 0; i < t->entry_count; i++) {
1284 struct tree_entry *e = t->entries[i];
1285 if (!e->versions[v].mode)
1286 continue;
1287 strbuf_addf(b, "%o %s%c",
1288 (unsigned int)(e->versions[v].mode & ~NO_DELTA),
1289 e->name->str_dat, '\0');
1290 strbuf_add(b, e->versions[v].oid.hash, GIT_SHA1_RAWSZ);
1291 }
1292 }
1293
1294 static void store_tree(struct tree_entry *root)
1295 {
1296 struct tree_content *t;
1297 unsigned int i, j, del;
1298 struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1299 struct object_entry *le = NULL;
1300
1301 if (!is_null_oid(&root->versions[1].oid))
1302 return;
1303
1304 if (!root->tree)
1305 load_tree(root);
1306 t = root->tree;
1307
1308 for (i = 0; i < t->entry_count; i++) {
1309 if (t->entries[i]->tree)
1310 store_tree(t->entries[i]);
1311 }
1312
1313 if (!(root->versions[0].mode & NO_DELTA))
1314 le = find_object(&root->versions[0].oid);
1315 if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1316 mktree(t, 0, &old_tree);
1317 lo.data = old_tree;
1318 lo.offset = le->idx.offset;
1319 lo.depth = t->delta_depth;
1320 }
1321
1322 mktree(t, 1, &new_tree);
1323 store_object(OBJ_TREE, &new_tree, &lo, &root->versions[1].oid, 0);
1324
1325 t->delta_depth = lo.depth;
1326 for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1327 struct tree_entry *e = t->entries[i];
1328 if (e->versions[1].mode) {
1329 e->versions[0].mode = e->versions[1].mode;
1330 oidcpy(&e->versions[0].oid, &e->versions[1].oid);
1331 t->entries[j++] = e;
1332 } else {
1333 release_tree_entry(e);
1334 del++;
1335 }
1336 }
1337 t->entry_count -= del;
1338 }
1339
1340 static void tree_content_replace(
1341 struct tree_entry *root,
1342 const struct object_id *oid,
1343 const uint16_t mode,
1344 struct tree_content *newtree)
1345 {
1346 if (!S_ISDIR(mode))
1347 die("Root cannot be a non-directory");
1348 oidclr(&root->versions[0].oid);
1349 oidcpy(&root->versions[1].oid, oid);
1350 if (root->tree)
1351 release_tree_content_recursive(root->tree);
1352 root->tree = newtree;
1353 }
1354
1355 static int tree_content_set(
1356 struct tree_entry *root,
1357 const char *p,
1358 const struct object_id *oid,
1359 const uint16_t mode,
1360 struct tree_content *subtree)
1361 {
1362 struct tree_content *t;
1363 const char *slash1;
1364 unsigned int i, n;
1365 struct tree_entry *e;
1366
1367 slash1 = strchrnul(p, '/');
1368 n = slash1 - p;
1369 if (!n)
1370 die("Empty path component found in input");
1371 if (!*slash1 && !S_ISDIR(mode) && subtree)
1372 die("Non-directories cannot have subtrees");
1373
1374 if (!root->tree)
1375 load_tree(root);
1376 t = root->tree;
1377 for (i = 0; i < t->entry_count; i++) {
1378 e = t->entries[i];
1379 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1380 if (!*slash1) {
1381 if (!S_ISDIR(mode)
1382 && e->versions[1].mode == mode
1383 && oideq(&e->versions[1].oid, oid))
1384 return 0;
1385 e->versions[1].mode = mode;
1386 oidcpy(&e->versions[1].oid, oid);
1387 if (e->tree)
1388 release_tree_content_recursive(e->tree);
1389 e->tree = subtree;
1390
1391 /*
1392 * We need to leave e->versions[0].sha1 alone
1393 * to avoid modifying the preimage tree used
1394 * when writing out the parent directory.
1395 * But after replacing the subdir with a
1396 * completely different one, it's not a good
1397 * delta base any more, and besides, we've
1398 * thrown away the tree entries needed to
1399 * make a delta against it.
1400 *
1401 * So let's just explicitly disable deltas
1402 * for the subtree.
1403 */
1404 if (S_ISDIR(e->versions[0].mode))
1405 e->versions[0].mode |= NO_DELTA;
1406
1407 oidclr(&root->versions[1].oid);
1408 return 1;
1409 }
1410 if (!S_ISDIR(e->versions[1].mode)) {
1411 e->tree = new_tree_content(8);
1412 e->versions[1].mode = S_IFDIR;
1413 }
1414 if (!e->tree)
1415 load_tree(e);
1416 if (tree_content_set(e, slash1 + 1, oid, mode, subtree)) {
1417 oidclr(&root->versions[1].oid);
1418 return 1;
1419 }
1420 return 0;
1421 }
1422 }
1423
1424 if (t->entry_count == t->entry_capacity)
1425 root->tree = t = grow_tree_content(t, t->entry_count);
1426 e = new_tree_entry();
1427 e->name = to_atom(p, n);
1428 e->versions[0].mode = 0;
1429 oidclr(&e->versions[0].oid);
1430 t->entries[t->entry_count++] = e;
1431 if (*slash1) {
1432 e->tree = new_tree_content(8);
1433 e->versions[1].mode = S_IFDIR;
1434 tree_content_set(e, slash1 + 1, oid, mode, subtree);
1435 } else {
1436 e->tree = subtree;
1437 e->versions[1].mode = mode;
1438 oidcpy(&e->versions[1].oid, oid);
1439 }
1440 oidclr(&root->versions[1].oid);
1441 return 1;
1442 }
1443
1444 static int tree_content_remove(
1445 struct tree_entry *root,
1446 const char *p,
1447 struct tree_entry *backup_leaf,
1448 int allow_root)
1449 {
1450 struct tree_content *t;
1451 const char *slash1;
1452 unsigned int i, n;
1453 struct tree_entry *e;
1454
1455 slash1 = strchrnul(p, '/');
1456 n = slash1 - p;
1457
1458 if (!root->tree)
1459 load_tree(root);
1460
1461 if (!*p && allow_root) {
1462 e = root;
1463 goto del_entry;
1464 }
1465
1466 t = root->tree;
1467 for (i = 0; i < t->entry_count; i++) {
1468 e = t->entries[i];
1469 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1470 if (*slash1 && !S_ISDIR(e->versions[1].mode))
1471 /*
1472 * If p names a file in some subdirectory, and a
1473 * file or symlink matching the name of the
1474 * parent directory of p exists, then p cannot
1475 * exist and need not be deleted.
1476 */
1477 return 1;
1478 if (!*slash1 || !S_ISDIR(e->versions[1].mode))
1479 goto del_entry;
1480 if (!e->tree)
1481 load_tree(e);
1482 if (tree_content_remove(e, slash1 + 1, backup_leaf, 0)) {
1483 for (n = 0; n < e->tree->entry_count; n++) {
1484 if (e->tree->entries[n]->versions[1].mode) {
1485 oidclr(&root->versions[1].oid);
1486 return 1;
1487 }
1488 }
1489 backup_leaf = NULL;
1490 goto del_entry;
1491 }
1492 return 0;
1493 }
1494 }
1495 return 0;
1496
1497 del_entry:
1498 if (backup_leaf)
1499 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1500 else if (e->tree)
1501 release_tree_content_recursive(e->tree);
1502 e->tree = NULL;
1503 e->versions[1].mode = 0;
1504 oidclr(&e->versions[1].oid);
1505 oidclr(&root->versions[1].oid);
1506 return 1;
1507 }
1508
1509 static int tree_content_get(
1510 struct tree_entry *root,
1511 const char *p,
1512 struct tree_entry *leaf,
1513 int allow_root)
1514 {
1515 struct tree_content *t;
1516 const char *slash1;
1517 unsigned int i, n;
1518 struct tree_entry *e;
1519
1520 slash1 = strchrnul(p, '/');
1521 n = slash1 - p;
1522 if (!n && !allow_root)
1523 die("Empty path component found in input");
1524
1525 if (!root->tree)
1526 load_tree(root);
1527
1528 if (!n) {
1529 e = root;
1530 goto found_entry;
1531 }
1532
1533 t = root->tree;
1534 for (i = 0; i < t->entry_count; i++) {
1535 e = t->entries[i];
1536 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1537 if (!*slash1)
1538 goto found_entry;
1539 if (!S_ISDIR(e->versions[1].mode))
1540 return 0;
1541 if (!e->tree)
1542 load_tree(e);
1543 return tree_content_get(e, slash1 + 1, leaf, 0);
1544 }
1545 }
1546 return 0;
1547
1548 found_entry:
1549 memcpy(leaf, e, sizeof(*leaf));
1550 if (e->tree && is_null_oid(&e->versions[1].oid))
1551 leaf->tree = dup_tree_content(e->tree);
1552 else
1553 leaf->tree = NULL;
1554 return 1;
1555 }
1556
1557 static int update_branch(struct branch *b)
1558 {
1559 static const char *msg = "fast-import";
1560 struct ref_transaction *transaction;
1561 struct object_id old_oid;
1562 struct strbuf err = STRBUF_INIT;
1563
1564 if (is_null_oid(&b->oid)) {
1565 if (b->delete)
1566 delete_ref(NULL, b->name, NULL, 0);
1567 return 0;
1568 }
1569 if (read_ref(b->name, &old_oid))
1570 oidclr(&old_oid);
1571 if (!force_update && !is_null_oid(&old_oid)) {
1572 struct commit *old_cmit, *new_cmit;
1573
1574 old_cmit = lookup_commit_reference_gently(the_repository,
1575 &old_oid, 0);
1576 new_cmit = lookup_commit_reference_gently(the_repository,
1577 &b->oid, 0);
1578 if (!old_cmit || !new_cmit)
1579 return error("Branch %s is missing commits.", b->name);
1580
1581 if (!in_merge_bases(old_cmit, new_cmit)) {
1582 warning("Not updating %s"
1583 " (new tip %s does not contain %s)",
1584 b->name, oid_to_hex(&b->oid),
1585 oid_to_hex(&old_oid));
1586 return -1;
1587 }
1588 }
1589 transaction = ref_transaction_begin(&err);
1590 if (!transaction ||
1591 ref_transaction_update(transaction, b->name, &b->oid, &old_oid,
1592 0, msg, &err) ||
1593 ref_transaction_commit(transaction, &err)) {
1594 ref_transaction_free(transaction);
1595 error("%s", err.buf);
1596 strbuf_release(&err);
1597 return -1;
1598 }
1599 ref_transaction_free(transaction);
1600 strbuf_release(&err);
1601 return 0;
1602 }
1603
1604 static void dump_branches(void)
1605 {
1606 unsigned int i;
1607 struct branch *b;
1608
1609 for (i = 0; i < branch_table_sz; i++) {
1610 for (b = branch_table[i]; b; b = b->table_next_branch)
1611 failure |= update_branch(b);
1612 }
1613 }
1614
1615 static void dump_tags(void)
1616 {
1617 static const char *msg = "fast-import";
1618 struct tag *t;
1619 struct strbuf ref_name = STRBUF_INIT;
1620 struct strbuf err = STRBUF_INIT;
1621 struct ref_transaction *transaction;
1622
1623 transaction = ref_transaction_begin(&err);
1624 if (!transaction) {
1625 failure |= error("%s", err.buf);
1626 goto cleanup;
1627 }
1628 for (t = first_tag; t; t = t->next_tag) {
1629 strbuf_reset(&ref_name);
1630 strbuf_addf(&ref_name, "refs/tags/%s", t->name);
1631
1632 if (ref_transaction_update(transaction, ref_name.buf,
1633 &t->oid, NULL, 0, msg, &err)) {
1634 failure |= error("%s", err.buf);
1635 goto cleanup;
1636 }
1637 }
1638 if (ref_transaction_commit(transaction, &err))
1639 failure |= error("%s", err.buf);
1640
1641 cleanup:
1642 ref_transaction_free(transaction);
1643 strbuf_release(&ref_name);
1644 strbuf_release(&err);
1645 }
1646
1647 static void dump_marks_helper(FILE *f,
1648 uintmax_t base,
1649 struct mark_set *m)
1650 {
1651 uintmax_t k;
1652 if (m->shift) {
1653 for (k = 0; k < 1024; k++) {
1654 if (m->data.sets[k])
1655 dump_marks_helper(f, base + (k << m->shift),
1656 m->data.sets[k]);
1657 }
1658 } else {
1659 for (k = 0; k < 1024; k++) {
1660 if (m->data.marked[k])
1661 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1662 oid_to_hex(&m->data.marked[k]->idx.oid));
1663 }
1664 }
1665 }
1666
1667 static void dump_marks(void)
1668 {
1669 struct lock_file mark_lock = LOCK_INIT;
1670 FILE *f;
1671
1672 if (!export_marks_file || (import_marks_file && !import_marks_file_done))
1673 return;
1674
1675 if (hold_lock_file_for_update(&mark_lock, export_marks_file, 0) < 0) {
1676 failure |= error_errno("Unable to write marks file %s",
1677 export_marks_file);
1678 return;
1679 }
1680
1681 f = fdopen_lock_file(&mark_lock, "w");
1682 if (!f) {
1683 int saved_errno = errno;
1684 rollback_lock_file(&mark_lock);
1685 failure |= error("Unable to write marks file %s: %s",
1686 export_marks_file, strerror(saved_errno));
1687 return;
1688 }
1689
1690 dump_marks_helper(f, 0, marks);
1691 if (commit_lock_file(&mark_lock)) {
1692 failure |= error_errno("Unable to write file %s",
1693 export_marks_file);
1694 return;
1695 }
1696 }
1697
1698 static void read_marks(void)
1699 {
1700 char line[512];
1701 FILE *f = fopen(import_marks_file, "r");
1702 if (f)
1703 ;
1704 else if (import_marks_file_ignore_missing && errno == ENOENT)
1705 goto done; /* Marks file does not exist */
1706 else
1707 die_errno("cannot read '%s'", import_marks_file);
1708 while (fgets(line, sizeof(line), f)) {
1709 uintmax_t mark;
1710 char *end;
1711 struct object_id oid;
1712 struct object_entry *e;
1713
1714 end = strchr(line, '\n');
1715 if (line[0] != ':' || !end)
1716 die("corrupt mark line: %s", line);
1717 *end = 0;
1718 mark = strtoumax(line + 1, &end, 10);
1719 if (!mark || end == line + 1
1720 || *end != ' ' || get_oid_hex(end + 1, &oid))
1721 die("corrupt mark line: %s", line);
1722 e = find_object(&oid);
1723 if (!e) {
1724 enum object_type type = oid_object_info(the_repository,
1725 &oid, NULL);
1726 if (type < 0)
1727 die("object not found: %s", oid_to_hex(&oid));
1728 e = insert_object(&oid);
1729 e->type = type;
1730 e->pack_id = MAX_PACK_ID;
1731 e->idx.offset = 1; /* just not zero! */
1732 }
1733 insert_mark(mark, e);
1734 }
1735 fclose(f);
1736 done:
1737 import_marks_file_done = 1;
1738 }
1739
1740
1741 static int read_next_command(void)
1742 {
1743 static int stdin_eof = 0;
1744
1745 if (stdin_eof) {
1746 unread_command_buf = 0;
1747 return EOF;
1748 }
1749
1750 for (;;) {
1751 const char *p;
1752
1753 if (unread_command_buf) {
1754 unread_command_buf = 0;
1755 } else {
1756 struct recent_command *rc;
1757
1758 strbuf_detach(&command_buf, NULL);
1759 stdin_eof = strbuf_getline_lf(&command_buf, stdin);
1760 if (stdin_eof)
1761 return EOF;
1762
1763 if (!seen_data_command
1764 && !starts_with(command_buf.buf, "feature ")
1765 && !starts_with(command_buf.buf, "option ")) {
1766 parse_argv();
1767 }
1768
1769 rc = rc_free;
1770 if (rc)
1771 rc_free = rc->next;
1772 else {
1773 rc = cmd_hist.next;
1774 cmd_hist.next = rc->next;
1775 cmd_hist.next->prev = &cmd_hist;
1776 free(rc->buf);
1777 }
1778
1779 rc->buf = command_buf.buf;
1780 rc->prev = cmd_tail;
1781 rc->next = cmd_hist.prev;
1782 rc->prev->next = rc;
1783 cmd_tail = rc;
1784 }
1785 if (skip_prefix(command_buf.buf, "get-mark ", &p)) {
1786 parse_get_mark(p);
1787 continue;
1788 }
1789 if (command_buf.buf[0] == '#')
1790 continue;
1791 return 0;
1792 }
1793 }
1794
1795 static void skip_optional_lf(void)
1796 {
1797 int term_char = fgetc(stdin);
1798 if (term_char != '\n' && term_char != EOF)
1799 ungetc(term_char, stdin);
1800 }
1801
1802 static void parse_mark(void)
1803 {
1804 const char *v;
1805 if (skip_prefix(command_buf.buf, "mark :", &v)) {
1806 next_mark = strtoumax(v, NULL, 10);
1807 read_next_command();
1808 }
1809 else
1810 next_mark = 0;
1811 }
1812
1813 static void parse_original_identifier(void)
1814 {
1815 const char *v;
1816 if (skip_prefix(command_buf.buf, "original-oid ", &v))
1817 read_next_command();
1818 }
1819
1820 static int parse_data(struct strbuf *sb, uintmax_t limit, uintmax_t *len_res)
1821 {
1822 const char *data;
1823 strbuf_reset(sb);
1824
1825 if (!skip_prefix(command_buf.buf, "data ", &data))
1826 die("Expected 'data n' command, found: %s", command_buf.buf);
1827
1828 if (skip_prefix(data, "<<", &data)) {
1829 char *term = xstrdup(data);
1830 size_t term_len = command_buf.len - (data - command_buf.buf);
1831
1832 strbuf_detach(&command_buf, NULL);
1833 for (;;) {
1834 if (strbuf_getline_lf(&command_buf, stdin) == EOF)
1835 die("EOF in data (terminator '%s' not found)", term);
1836 if (term_len == command_buf.len
1837 && !strcmp(term, command_buf.buf))
1838 break;
1839 strbuf_addbuf(sb, &command_buf);
1840 strbuf_addch(sb, '\n');
1841 }
1842 free(term);
1843 }
1844 else {
1845 uintmax_t len = strtoumax(data, NULL, 10);
1846 size_t n = 0, length = (size_t)len;
1847
1848 if (limit && limit < len) {
1849 *len_res = len;
1850 return 0;
1851 }
1852 if (length < len)
1853 die("data is too large to use in this context");
1854
1855 while (n < length) {
1856 size_t s = strbuf_fread(sb, length - n, stdin);
1857 if (!s && feof(stdin))
1858 die("EOF in data (%lu bytes remaining)",
1859 (unsigned long)(length - n));
1860 n += s;
1861 }
1862 }
1863
1864 skip_optional_lf();
1865 return 1;
1866 }
1867
1868 static int validate_raw_date(const char *src, struct strbuf *result)
1869 {
1870 const char *orig_src = src;
1871 char *endp;
1872 unsigned long num;
1873
1874 errno = 0;
1875
1876 num = strtoul(src, &endp, 10);
1877 /* NEEDSWORK: perhaps check for reasonable values? */
1878 if (errno || endp == src || *endp != ' ')
1879 return -1;
1880
1881 src = endp + 1;
1882 if (*src != '-' && *src != '+')
1883 return -1;
1884
1885 num = strtoul(src + 1, &endp, 10);
1886 if (errno || endp == src + 1 || *endp || 1400 < num)
1887 return -1;
1888
1889 strbuf_addstr(result, orig_src);
1890 return 0;
1891 }
1892
1893 static char *parse_ident(const char *buf)
1894 {
1895 const char *ltgt;
1896 size_t name_len;
1897 struct strbuf ident = STRBUF_INIT;
1898
1899 /* ensure there is a space delimiter even if there is no name */
1900 if (*buf == '<')
1901 --buf;
1902
1903 ltgt = buf + strcspn(buf, "<>");
1904 if (*ltgt != '<')
1905 die("Missing < in ident string: %s", buf);
1906 if (ltgt != buf && ltgt[-1] != ' ')
1907 die("Missing space before < in ident string: %s", buf);
1908 ltgt = ltgt + 1 + strcspn(ltgt + 1, "<>");
1909 if (*ltgt != '>')
1910 die("Missing > in ident string: %s", buf);
1911 ltgt++;
1912 if (*ltgt != ' ')
1913 die("Missing space after > in ident string: %s", buf);
1914 ltgt++;
1915 name_len = ltgt - buf;
1916 strbuf_add(&ident, buf, name_len);
1917
1918 switch (whenspec) {
1919 case WHENSPEC_RAW:
1920 if (validate_raw_date(ltgt, &ident) < 0)
1921 die("Invalid raw date \"%s\" in ident: %s", ltgt, buf);
1922 break;
1923 case WHENSPEC_RFC2822:
1924 if (parse_date(ltgt, &ident) < 0)
1925 die("Invalid rfc2822 date \"%s\" in ident: %s", ltgt, buf);
1926 break;
1927 case WHENSPEC_NOW:
1928 if (strcmp("now", ltgt))
1929 die("Date in ident must be 'now': %s", buf);
1930 datestamp(&ident);
1931 break;
1932 }
1933
1934 return strbuf_detach(&ident, NULL);
1935 }
1936
1937 static void parse_and_store_blob(
1938 struct last_object *last,
1939 struct object_id *oidout,
1940 uintmax_t mark)
1941 {
1942 static struct strbuf buf = STRBUF_INIT;
1943 uintmax_t len;
1944
1945 if (parse_data(&buf, big_file_threshold, &len))
1946 store_object(OBJ_BLOB, &buf, last, oidout, mark);
1947 else {
1948 if (last) {
1949 strbuf_release(&last->data);
1950 last->offset = 0;
1951 last->depth = 0;
1952 }
1953 stream_blob(len, oidout, mark);
1954 skip_optional_lf();
1955 }
1956 }
1957
1958 static void parse_new_blob(void)
1959 {
1960 read_next_command();
1961 parse_mark();
1962 parse_original_identifier();
1963 parse_and_store_blob(&last_blob, NULL, next_mark);
1964 }
1965
1966 static void unload_one_branch(void)
1967 {
1968 while (cur_active_branches
1969 && cur_active_branches >= max_active_branches) {
1970 uintmax_t min_commit = ULONG_MAX;
1971 struct branch *e, *l = NULL, *p = NULL;
1972
1973 for (e = active_branches; e; e = e->active_next_branch) {
1974 if (e->last_commit < min_commit) {
1975 p = l;
1976 min_commit = e->last_commit;
1977 }
1978 l = e;
1979 }
1980
1981 if (p) {
1982 e = p->active_next_branch;
1983 p->active_next_branch = e->active_next_branch;
1984 } else {
1985 e = active_branches;
1986 active_branches = e->active_next_branch;
1987 }
1988 e->active = 0;
1989 e->active_next_branch = NULL;
1990 if (e->branch_tree.tree) {
1991 release_tree_content_recursive(e->branch_tree.tree);
1992 e->branch_tree.tree = NULL;
1993 }
1994 cur_active_branches--;
1995 }
1996 }
1997
1998 static void load_branch(struct branch *b)
1999 {
2000 load_tree(&b->branch_tree);
2001 if (!b->active) {
2002 b->active = 1;
2003 b->active_next_branch = active_branches;
2004 active_branches = b;
2005 cur_active_branches++;
2006 branch_load_count++;
2007 }
2008 }
2009
2010 static unsigned char convert_num_notes_to_fanout(uintmax_t num_notes)
2011 {
2012 unsigned char fanout = 0;
2013 while ((num_notes >>= 8))
2014 fanout++;
2015 return fanout;
2016 }
2017
2018 static void construct_path_with_fanout(const char *hex_sha1,
2019 unsigned char fanout, char *path)
2020 {
2021 unsigned int i = 0, j = 0;
2022 if (fanout >= the_hash_algo->rawsz)
2023 die("Too large fanout (%u)", fanout);
2024 while (fanout) {
2025 path[i++] = hex_sha1[j++];
2026 path[i++] = hex_sha1[j++];
2027 path[i++] = '/';
2028 fanout--;
2029 }
2030 memcpy(path + i, hex_sha1 + j, the_hash_algo->hexsz - j);
2031 path[i + the_hash_algo->hexsz - j] = '\0';
2032 }
2033
2034 static uintmax_t do_change_note_fanout(
2035 struct tree_entry *orig_root, struct tree_entry *root,
2036 char *hex_oid, unsigned int hex_oid_len,
2037 char *fullpath, unsigned int fullpath_len,
2038 unsigned char fanout)
2039 {
2040 struct tree_content *t;
2041 struct tree_entry *e, leaf;
2042 unsigned int i, tmp_hex_oid_len, tmp_fullpath_len;
2043 uintmax_t num_notes = 0;
2044 struct object_id oid;
2045 char realpath[60];
2046
2047 if (!root->tree)
2048 load_tree(root);
2049 t = root->tree;
2050
2051 for (i = 0; t && i < t->entry_count; i++) {
2052 e = t->entries[i];
2053 tmp_hex_oid_len = hex_oid_len + e->name->str_len;
2054 tmp_fullpath_len = fullpath_len;
2055
2056 /*
2057 * We're interested in EITHER existing note entries (entries
2058 * with exactly 40 hex chars in path, not including directory
2059 * separators), OR directory entries that may contain note
2060 * entries (with < 40 hex chars in path).
2061 * Also, each path component in a note entry must be a multiple
2062 * of 2 chars.
2063 */
2064 if (!e->versions[1].mode ||
2065 tmp_hex_oid_len > GIT_SHA1_HEXSZ ||
2066 e->name->str_len % 2)
2067 continue;
2068
2069 /* This _may_ be a note entry, or a subdir containing notes */
2070 memcpy(hex_oid + hex_oid_len, e->name->str_dat,
2071 e->name->str_len);
2072 if (tmp_fullpath_len)
2073 fullpath[tmp_fullpath_len++] = '/';
2074 memcpy(fullpath + tmp_fullpath_len, e->name->str_dat,
2075 e->name->str_len);
2076 tmp_fullpath_len += e->name->str_len;
2077 fullpath[tmp_fullpath_len] = '\0';
2078
2079 if (tmp_hex_oid_len == GIT_SHA1_HEXSZ && !get_oid_hex(hex_oid, &oid)) {
2080 /* This is a note entry */
2081 if (fanout == 0xff) {
2082 /* Counting mode, no rename */
2083 num_notes++;
2084 continue;
2085 }
2086 construct_path_with_fanout(hex_oid, fanout, realpath);
2087 if (!strcmp(fullpath, realpath)) {
2088 /* Note entry is in correct location */
2089 num_notes++;
2090 continue;
2091 }
2092
2093 /* Rename fullpath to realpath */
2094 if (!tree_content_remove(orig_root, fullpath, &leaf, 0))
2095 die("Failed to remove path %s", fullpath);
2096 tree_content_set(orig_root, realpath,
2097 &leaf.versions[1].oid,
2098 leaf.versions[1].mode,
2099 leaf.tree);
2100 } else if (S_ISDIR(e->versions[1].mode)) {
2101 /* This is a subdir that may contain note entries */
2102 num_notes += do_change_note_fanout(orig_root, e,
2103 hex_oid, tmp_hex_oid_len,
2104 fullpath, tmp_fullpath_len, fanout);
2105 }
2106
2107 /* The above may have reallocated the current tree_content */
2108 t = root->tree;
2109 }
2110 return num_notes;
2111 }
2112
2113 static uintmax_t change_note_fanout(struct tree_entry *root,
2114 unsigned char fanout)
2115 {
2116 /*
2117 * The size of path is due to one slash between every two hex digits,
2118 * plus the terminating NUL. Note that there is no slash at the end, so
2119 * the number of slashes is one less than half the number of hex
2120 * characters.
2121 */
2122 char hex_oid[GIT_MAX_HEXSZ], path[GIT_MAX_HEXSZ + (GIT_MAX_HEXSZ / 2) - 1 + 1];
2123 return do_change_note_fanout(root, root, hex_oid, 0, path, 0, fanout);
2124 }
2125
2126 /*
2127 * Given a pointer into a string, parse a mark reference:
2128 *
2129 * idnum ::= ':' bigint;
2130 *
2131 * Return the first character after the value in *endptr.
2132 *
2133 * Complain if the following character is not what is expected,
2134 * either a space or end of the string.
2135 */
2136 static uintmax_t parse_mark_ref(const char *p, char **endptr)
2137 {
2138 uintmax_t mark;
2139
2140 assert(*p == ':');
2141 p++;
2142 mark = strtoumax(p, endptr, 10);
2143 if (*endptr == p)
2144 die("No value after ':' in mark: %s", command_buf.buf);
2145 return mark;
2146 }
2147
2148 /*
2149 * Parse the mark reference, and complain if this is not the end of
2150 * the string.
2151 */
2152 static uintmax_t parse_mark_ref_eol(const char *p)
2153 {
2154 char *end;
2155 uintmax_t mark;
2156
2157 mark = parse_mark_ref(p, &end);
2158 if (*end != '\0')
2159 die("Garbage after mark: %s", command_buf.buf);
2160 return mark;
2161 }
2162
2163 /*
2164 * Parse the mark reference, demanding a trailing space. Return a
2165 * pointer to the space.
2166 */
2167 static uintmax_t parse_mark_ref_space(const char **p)
2168 {
2169 uintmax_t mark;
2170 char *end;
2171
2172 mark = parse_mark_ref(*p, &end);
2173 if (*end++ != ' ')
2174 die("Missing space after mark: %s", command_buf.buf);
2175 *p = end;
2176 return mark;
2177 }
2178
2179 static void file_change_m(const char *p, struct branch *b)
2180 {
2181 static struct strbuf uq = STRBUF_INIT;
2182 const char *endp;
2183 struct object_entry *oe;
2184 struct object_id oid;
2185 uint16_t mode, inline_data = 0;
2186
2187 p = get_mode(p, &mode);
2188 if (!p)
2189 die("Corrupt mode: %s", command_buf.buf);
2190 switch (mode) {
2191 case 0644:
2192 case 0755:
2193 mode |= S_IFREG;
2194 case S_IFREG | 0644:
2195 case S_IFREG | 0755:
2196 case S_IFLNK:
2197 case S_IFDIR:
2198 case S_IFGITLINK:
2199 /* ok */
2200 break;
2201 default:
2202 die("Corrupt mode: %s", command_buf.buf);
2203 }
2204
2205 if (*p == ':') {
2206 oe = find_mark(parse_mark_ref_space(&p));
2207 oidcpy(&oid, &oe->idx.oid);
2208 } else if (skip_prefix(p, "inline ", &p)) {
2209 inline_data = 1;
2210 oe = NULL; /* not used with inline_data, but makes gcc happy */
2211 } else {
2212 if (parse_oid_hex(p, &oid, &p))
2213 die("Invalid dataref: %s", command_buf.buf);
2214 oe = find_object(&oid);
2215 if (*p++ != ' ')
2216 die("Missing space after SHA1: %s", command_buf.buf);
2217 }
2218
2219 strbuf_reset(&uq);
2220 if (!unquote_c_style(&uq, p, &endp)) {
2221 if (*endp)
2222 die("Garbage after path in: %s", command_buf.buf);
2223 p = uq.buf;
2224 }
2225
2226 /* Git does not track empty, non-toplevel directories. */
2227 if (S_ISDIR(mode) && is_empty_tree_oid(&oid) && *p) {
2228 tree_content_remove(&b->branch_tree, p, NULL, 0);
2229 return;
2230 }
2231
2232 if (S_ISGITLINK(mode)) {
2233 if (inline_data)
2234 die("Git links cannot be specified 'inline': %s",
2235 command_buf.buf);
2236 else if (oe) {
2237 if (oe->type != OBJ_COMMIT)
2238 die("Not a commit (actually a %s): %s",
2239 type_name(oe->type), command_buf.buf);
2240 }
2241 /*
2242 * Accept the sha1 without checking; it expected to be in
2243 * another repository.
2244 */
2245 } else if (inline_data) {
2246 if (S_ISDIR(mode))
2247 die("Directories cannot be specified 'inline': %s",
2248 command_buf.buf);
2249 if (p != uq.buf) {
2250 strbuf_addstr(&uq, p);
2251 p = uq.buf;
2252 }
2253 while (read_next_command() != EOF) {
2254 const char *v;
2255 if (skip_prefix(command_buf.buf, "cat-blob ", &v))
2256 parse_cat_blob(v);
2257 else {
2258 parse_and_store_blob(&last_blob, &oid, 0);
2259 break;
2260 }
2261 }
2262 } else {
2263 enum object_type expected = S_ISDIR(mode) ?
2264 OBJ_TREE: OBJ_BLOB;
2265 enum object_type type = oe ? oe->type :
2266 oid_object_info(the_repository, &oid,
2267 NULL);
2268 if (type < 0)
2269 die("%s not found: %s",
2270 S_ISDIR(mode) ? "Tree" : "Blob",
2271 command_buf.buf);
2272 if (type != expected)
2273 die("Not a %s (actually a %s): %s",
2274 type_name(expected), type_name(type),
2275 command_buf.buf);
2276 }
2277
2278 if (!*p) {
2279 tree_content_replace(&b->branch_tree, &oid, mode, NULL);
2280 return;
2281 }
2282 tree_content_set(&b->branch_tree, p, &oid, mode, NULL);
2283 }
2284
2285 static void file_change_d(const char *p, struct branch *b)
2286 {
2287 static struct strbuf uq = STRBUF_INIT;
2288 const char *endp;
2289
2290 strbuf_reset(&uq);
2291 if (!unquote_c_style(&uq, p, &endp)) {
2292 if (*endp)
2293 die("Garbage after path in: %s", command_buf.buf);
2294 p = uq.buf;
2295 }
2296 tree_content_remove(&b->branch_tree, p, NULL, 1);
2297 }
2298
2299 static void file_change_cr(const char *s, struct branch *b, int rename)
2300 {
2301 const char *d;
2302 static struct strbuf s_uq = STRBUF_INIT;
2303 static struct strbuf d_uq = STRBUF_INIT;
2304 const char *endp;
2305 struct tree_entry leaf;
2306
2307 strbuf_reset(&s_uq);
2308 if (!unquote_c_style(&s_uq, s, &endp)) {
2309 if (*endp != ' ')
2310 die("Missing space after source: %s", command_buf.buf);
2311 } else {
2312 endp = strchr(s, ' ');
2313 if (!endp)
2314 die("Missing space after source: %s", command_buf.buf);
2315 strbuf_add(&s_uq, s, endp - s);
2316 }
2317 s = s_uq.buf;
2318
2319 endp++;
2320 if (!*endp)
2321 die("Missing dest: %s", command_buf.buf);
2322
2323 d = endp;
2324 strbuf_reset(&d_uq);
2325 if (!unquote_c_style(&d_uq, d, &endp)) {
2326 if (*endp)
2327 die("Garbage after dest in: %s", command_buf.buf);
2328 d = d_uq.buf;
2329 }
2330
2331 memset(&leaf, 0, sizeof(leaf));
2332 if (rename)
2333 tree_content_remove(&b->branch_tree, s, &leaf, 1);
2334 else
2335 tree_content_get(&b->branch_tree, s, &leaf, 1);
2336 if (!leaf.versions[1].mode)
2337 die("Path %s not in branch", s);
2338 if (!*d) { /* C "path/to/subdir" "" */
2339 tree_content_replace(&b->branch_tree,
2340 &leaf.versions[1].oid,
2341 leaf.versions[1].mode,
2342 leaf.tree);
2343 return;
2344 }
2345 tree_content_set(&b->branch_tree, d,
2346 &leaf.versions[1].oid,
2347 leaf.versions[1].mode,
2348 leaf.tree);
2349 }
2350
2351 static void note_change_n(const char *p, struct branch *b, unsigned char *old_fanout)
2352 {
2353 static struct strbuf uq = STRBUF_INIT;
2354 struct object_entry *oe;
2355 struct branch *s;
2356 struct object_id oid, commit_oid;
2357 char path[60];
2358 uint16_t inline_data = 0;
2359 unsigned char new_fanout;
2360
2361 /*
2362 * When loading a branch, we don't traverse its tree to count the real
2363 * number of notes (too expensive to do this for all non-note refs).
2364 * This means that recently loaded notes refs might incorrectly have
2365 * b->num_notes == 0, and consequently, old_fanout might be wrong.
2366 *
2367 * Fix this by traversing the tree and counting the number of notes
2368 * when b->num_notes == 0. If the notes tree is truly empty, the
2369 * calculation should not take long.
2370 */
2371 if (b->num_notes == 0 && *old_fanout == 0) {
2372 /* Invoke change_note_fanout() in "counting mode". */
2373 b->num_notes = change_note_fanout(&b->branch_tree, 0xff);
2374 *old_fanout = convert_num_notes_to_fanout(b->num_notes);
2375 }
2376
2377 /* Now parse the notemodify command. */
2378 /* <dataref> or 'inline' */
2379 if (*p == ':') {
2380 oe = find_mark(parse_mark_ref_space(&p));
2381 oidcpy(&oid, &oe->idx.oid);
2382 } else if (skip_prefix(p, "inline ", &p)) {
2383 inline_data = 1;
2384 oe = NULL; /* not used with inline_data, but makes gcc happy */
2385 } else {
2386 if (parse_oid_hex(p, &oid, &p))
2387 die("Invalid dataref: %s", command_buf.buf);
2388 oe = find_object(&oid);
2389 if (*p++ != ' ')
2390 die("Missing space after SHA1: %s", command_buf.buf);
2391 }
2392
2393 /* <commit-ish> */
2394 s = lookup_branch(p);
2395 if (s) {
2396 if (is_null_oid(&s->oid))
2397 die("Can't add a note on empty branch.");
2398 oidcpy(&commit_oid, &s->oid);
2399 } else if (*p == ':') {
2400 uintmax_t commit_mark = parse_mark_ref_eol(p);
2401 struct object_entry *commit_oe = find_mark(commit_mark);
2402 if (commit_oe->type != OBJ_COMMIT)
2403 die("Mark :%" PRIuMAX " not a commit", commit_mark);
2404 oidcpy(&commit_oid, &commit_oe->idx.oid);
2405 } else if (!get_oid(p, &commit_oid)) {
2406 unsigned long size;
2407 char *buf = read_object_with_reference(&commit_oid,
2408 commit_type, &size,
2409 &commit_oid);
2410 if (!buf || size < 46)
2411 die("Not a valid commit: %s", p);
2412 free(buf);
2413 } else
2414 die("Invalid ref name or SHA1 expression: %s", p);
2415
2416 if (inline_data) {
2417 if (p != uq.buf) {
2418 strbuf_addstr(&uq, p);
2419 p = uq.buf;
2420 }
2421 read_next_command();
2422 parse_and_store_blob(&last_blob, &oid, 0);
2423 } else if (oe) {
2424 if (oe->type != OBJ_BLOB)
2425 die("Not a blob (actually a %s): %s",
2426 type_name(oe->type), command_buf.buf);
2427 } else if (!is_null_oid(&oid)) {
2428 enum object_type type = oid_object_info(the_repository, &oid,
2429 NULL);
2430 if (type < 0)
2431 die("Blob not found: %s", command_buf.buf);
2432 if (type != OBJ_BLOB)
2433 die("Not a blob (actually a %s): %s",
2434 type_name(type), command_buf.buf);
2435 }
2436
2437 construct_path_with_fanout(oid_to_hex(&commit_oid), *old_fanout, path);
2438 if (tree_content_remove(&b->branch_tree, path, NULL, 0))
2439 b->num_notes--;
2440
2441 if (is_null_oid(&oid))
2442 return; /* nothing to insert */
2443
2444 b->num_notes++;
2445 new_fanout = convert_num_notes_to_fanout(b->num_notes);
2446 construct_path_with_fanout(oid_to_hex(&commit_oid), new_fanout, path);
2447 tree_content_set(&b->branch_tree, path, &oid, S_IFREG | 0644, NULL);
2448 }
2449
2450 static void file_change_deleteall(struct branch *b)
2451 {
2452 release_tree_content_recursive(b->branch_tree.tree);
2453 oidclr(&b->branch_tree.versions[0].oid);
2454 oidclr(&b->branch_tree.versions[1].oid);
2455 load_tree(&b->branch_tree);
2456 b->num_notes = 0;
2457 }
2458
2459 static void parse_from_commit(struct branch *b, char *buf, unsigned long size)
2460 {
2461 if (!buf || size < GIT_SHA1_HEXSZ + 6)
2462 die("Not a valid commit: %s", oid_to_hex(&b->oid));
2463 if (memcmp("tree ", buf, 5)
2464 || get_oid_hex(buf + 5, &b->branch_tree.versions[1].oid))
2465 die("The commit %s is corrupt", oid_to_hex(&b->oid));
2466 oidcpy(&b->branch_tree.versions[0].oid,
2467 &b->branch_tree.versions[1].oid);
2468 }
2469
2470 static void parse_from_existing(struct branch *b)
2471 {
2472 if (is_null_oid(&b->oid)) {
2473 oidclr(&b->branch_tree.versions[0].oid);
2474 oidclr(&b->branch_tree.versions[1].oid);
2475 } else {
2476 unsigned long size;
2477 char *buf;
2478
2479 buf = read_object_with_reference(&b->oid, commit_type, &size,
2480 &b->oid);
2481 parse_from_commit(b, buf, size);
2482 free(buf);
2483 }
2484 }
2485
2486 static int parse_from(struct branch *b)
2487 {
2488 const char *from;
2489 struct branch *s;
2490 struct object_id oid;
2491
2492 if (!skip_prefix(command_buf.buf, "from ", &from))
2493 return 0;
2494
2495 oidcpy(&oid, &b->branch_tree.versions[1].oid);
2496
2497 s = lookup_branch(from);
2498 if (b == s)
2499 die("Can't create a branch from itself: %s", b->name);
2500 else if (s) {
2501 struct object_id *t = &s->branch_tree.versions[1].oid;
2502 oidcpy(&b->oid, &s->oid);
2503 oidcpy(&b->branch_tree.versions[0].oid, t);
2504 oidcpy(&b->branch_tree.versions[1].oid, t);
2505 } else if (*from == ':') {
2506 uintmax_t idnum = parse_mark_ref_eol(from);
2507 struct object_entry *oe = find_mark(idnum);
2508 if (oe->type != OBJ_COMMIT)
2509 die("Mark :%" PRIuMAX " not a commit", idnum);
2510 if (!oideq(&b->oid, &oe->idx.oid)) {
2511 oidcpy(&b->oid, &oe->idx.oid);
2512 if (oe->pack_id != MAX_PACK_ID) {
2513 unsigned long size;
2514 char *buf = gfi_unpack_entry(oe, &size);
2515 parse_from_commit(b, buf, size);
2516 free(buf);
2517 } else
2518 parse_from_existing(b);
2519 }
2520 } else if (!get_oid(from, &b->oid)) {
2521 parse_from_existing(b);
2522 if (is_null_oid(&b->oid))
2523 b->delete = 1;
2524 }
2525 else
2526 die("Invalid ref name or SHA1 expression: %s", from);
2527
2528 if (b->branch_tree.tree && !oideq(&oid, &b->branch_tree.versions[1].oid)) {
2529 release_tree_content_recursive(b->branch_tree.tree);
2530 b->branch_tree.tree = NULL;
2531 }
2532
2533 read_next_command();
2534 return 1;
2535 }
2536
2537 static struct hash_list *parse_merge(unsigned int *count)
2538 {
2539 struct hash_list *list = NULL, **tail = &list, *n;
2540 const char *from;
2541 struct branch *s;
2542
2543 *count = 0;
2544 while (skip_prefix(command_buf.buf, "merge ", &from)) {
2545 n = xmalloc(sizeof(*n));
2546 s = lookup_branch(from);
2547 if (s)
2548 oidcpy(&n->oid, &s->oid);
2549 else if (*from == ':') {
2550 uintmax_t idnum = parse_mark_ref_eol(from);
2551 struct object_entry *oe = find_mark(idnum);
2552 if (oe->type != OBJ_COMMIT)
2553 die("Mark :%" PRIuMAX " not a commit", idnum);
2554 oidcpy(&n->oid, &oe->idx.oid);
2555 } else if (!get_oid(from, &n->oid)) {
2556 unsigned long size;
2557 char *buf = read_object_with_reference(&n->oid,
2558 commit_type,
2559 &size, &n->oid);
2560 if (!buf || size < 46)
2561 die("Not a valid commit: %s", from);
2562 free(buf);
2563 } else
2564 die("Invalid ref name or SHA1 expression: %s", from);
2565
2566 n->next = NULL;
2567 *tail = n;
2568 tail = &n->next;
2569
2570 (*count)++;
2571 read_next_command();
2572 }
2573 return list;
2574 }
2575
2576 static void parse_new_commit(const char *arg)
2577 {
2578 static struct strbuf msg = STRBUF_INIT;
2579 struct branch *b;
2580 char *author = NULL;
2581 char *committer = NULL;
2582 struct hash_list *merge_list = NULL;
2583 unsigned int merge_count;
2584 unsigned char prev_fanout, new_fanout;
2585 const char *v;
2586
2587 b = lookup_branch(arg);
2588 if (!b)
2589 b = new_branch(arg);
2590
2591 read_next_command();
2592 parse_mark();
2593 parse_original_identifier();
2594 if (skip_prefix(command_buf.buf, "author ", &v)) {
2595 author = parse_ident(v);
2596 read_next_command();
2597 }
2598 if (skip_prefix(command_buf.buf, "committer ", &v)) {
2599 committer = parse_ident(v);
2600 read_next_command();
2601 }
2602 if (!committer)
2603 die("Expected committer but didn't get one");
2604 parse_data(&msg, 0, NULL);
2605 read_next_command();
2606 parse_from(b);
2607 merge_list = parse_merge(&merge_count);
2608
2609 /* ensure the branch is active/loaded */
2610 if (!b->branch_tree.tree || !max_active_branches) {
2611 unload_one_branch();
2612 load_branch(b);
2613 }
2614
2615 prev_fanout = convert_num_notes_to_fanout(b->num_notes);
2616
2617 /* file_change* */
2618 while (command_buf.len > 0) {
2619 if (skip_prefix(command_buf.buf, "M ", &v))
2620 file_change_m(v, b);
2621 else if (skip_prefix(command_buf.buf, "D ", &v))
2622 file_change_d(v, b);
2623 else if (skip_prefix(command_buf.buf, "R ", &v))
2624 file_change_cr(v, b, 1);
2625 else if (skip_prefix(command_buf.buf, "C ", &v))
2626 file_change_cr(v, b, 0);
2627 else if (skip_prefix(command_buf.buf, "N ", &v))
2628 note_change_n(v, b, &prev_fanout);
2629 else if (!strcmp("deleteall", command_buf.buf))
2630 file_change_deleteall(b);
2631 else if (skip_prefix(command_buf.buf, "ls ", &v))
2632 parse_ls(v, b);
2633 else if (skip_prefix(command_buf.buf, "cat-blob ", &v))
2634 parse_cat_blob(v);
2635 else {
2636 unread_command_buf = 1;
2637 break;
2638 }
2639 if (read_next_command() == EOF)
2640 break;
2641 }
2642
2643 new_fanout = convert_num_notes_to_fanout(b->num_notes);
2644 if (new_fanout != prev_fanout)
2645 b->num_notes = change_note_fanout(&b->branch_tree, new_fanout);
2646
2647 /* build the tree and the commit */
2648 store_tree(&b->branch_tree);
2649 oidcpy(&b->branch_tree.versions[0].oid,
2650 &b->branch_tree.versions[1].oid);
2651
2652 strbuf_reset(&new_data);
2653 strbuf_addf(&new_data, "tree %s\n",
2654 oid_to_hex(&b->branch_tree.versions[1].oid));
2655 if (!is_null_oid(&b->oid))
2656 strbuf_addf(&new_data, "parent %s\n",
2657 oid_to_hex(&b->oid));
2658 while (merge_list) {
2659 struct hash_list *next = merge_list->next;
2660 strbuf_addf(&new_data, "parent %s\n",
2661 oid_to_hex(&merge_list->oid));
2662 free(merge_list);
2663 merge_list = next;
2664 }
2665 strbuf_addf(&new_data,
2666 "author %s\n"
2667 "committer %s\n"
2668 "\n",
2669 author ? author : committer, committer);
2670 strbuf_addbuf(&new_data, &msg);
2671 free(author);
2672 free(committer);
2673
2674 if (!store_object(OBJ_COMMIT, &new_data, NULL, &b->oid, next_mark))
2675 b->pack_id = pack_id;
2676 b->last_commit = object_count_by_type[OBJ_COMMIT];
2677 }
2678
2679 static void parse_new_tag(const char *arg)
2680 {
2681 static struct strbuf msg = STRBUF_INIT;
2682 const char *from;
2683 char *tagger;
2684 struct branch *s;
2685 struct tag *t;
2686 uintmax_t from_mark = 0;
2687 struct object_id oid;
2688 enum object_type type;
2689 const char *v;
2690
2691 t = mem_pool_alloc(&fi_mem_pool, sizeof(struct tag));
2692 memset(t, 0, sizeof(struct tag));
2693 t->name = pool_strdup(arg);
2694 if (last_tag)
2695 last_tag->next_tag = t;
2696 else
2697 first_tag = t;
2698 last_tag = t;
2699 read_next_command();
2700
2701 /* from ... */
2702 if (!skip_prefix(command_buf.buf, "from ", &from))
2703 die("Expected from command, got %s", command_buf.buf);
2704 s = lookup_branch(from);
2705 if (s) {
2706 if (is_null_oid(&s->oid))
2707 die("Can't tag an empty branch.");
2708 oidcpy(&oid, &s->oid);
2709 type = OBJ_COMMIT;
2710 } else if (*from == ':') {
2711 struct object_entry *oe;
2712 from_mark = parse_mark_ref_eol(from);
2713 oe = find_mark(from_mark);
2714 type = oe->type;
2715 oidcpy(&oid, &oe->idx.oid);
2716 } else if (!get_oid(from, &oid)) {
2717 struct object_entry *oe = find_object(&oid);
2718 if (!oe) {
2719 type = oid_object_info(the_repository, &oid, NULL);
2720 if (type < 0)
2721 die("Not a valid object: %s", from);
2722 } else
2723 type = oe->type;
2724 } else
2725 die("Invalid ref name or SHA1 expression: %s", from);
2726 read_next_command();
2727
2728 /* original-oid ... */
2729 parse_original_identifier();
2730
2731 /* tagger ... */
2732 if (skip_prefix(command_buf.buf, "tagger ", &v)) {
2733 tagger = parse_ident(v);
2734 read_next_command();
2735 } else
2736 tagger = NULL;
2737
2738 /* tag payload/message */
2739 parse_data(&msg, 0, NULL);
2740
2741 /* build the tag object */
2742 strbuf_reset(&new_data);
2743
2744 strbuf_addf(&new_data,
2745 "object %s\n"
2746 "type %s\n"
2747 "tag %s\n",
2748 oid_to_hex(&oid), type_name(type), t->name);
2749 if (tagger)
2750 strbuf_addf(&new_data,
2751 "tagger %s\n", tagger);
2752 strbuf_addch(&new_data, '\n');
2753 strbuf_addbuf(&new_data, &msg);
2754 free(tagger);
2755
2756 if (store_object(OBJ_TAG, &new_data, NULL, &t->oid, 0))
2757 t->pack_id = MAX_PACK_ID;
2758 else
2759 t->pack_id = pack_id;
2760 }
2761
2762 static void parse_reset_branch(const char *arg)
2763 {
2764 struct branch *b;
2765
2766 b = lookup_branch(arg);
2767 if (b) {
2768 oidclr(&b->oid);
2769 oidclr(&b->branch_tree.versions[0].oid);
2770 oidclr(&b->branch_tree.versions[1].oid);
2771 if (b->branch_tree.tree) {
2772 release_tree_content_recursive(b->branch_tree.tree);
2773 b->branch_tree.tree = NULL;
2774 }
2775 }
2776 else
2777 b = new_branch(arg);
2778 read_next_command();
2779 parse_from(b);
2780 if (command_buf.len > 0)
2781 unread_command_buf = 1;
2782 }
2783
2784 static void cat_blob_write(const char *buf, unsigned long size)
2785 {
2786 if (write_in_full(cat_blob_fd, buf, size) < 0)
2787 die_errno("Write to frontend failed");
2788 }
2789
2790 static void cat_blob(struct object_entry *oe, struct object_id *oid)
2791 {
2792 struct strbuf line = STRBUF_INIT;
2793 unsigned long size;
2794 enum object_type type = 0;
2795 char *buf;
2796
2797 if (!oe || oe->pack_id == MAX_PACK_ID) {
2798 buf = read_object_file(oid, &type, &size);
2799 } else {
2800 type = oe->type;
2801 buf = gfi_unpack_entry(oe, &size);
2802 }
2803
2804 /*
2805 * Output based on batch_one_object() from cat-file.c.
2806 */
2807 if (type <= 0) {
2808 strbuf_reset(&line);
2809 strbuf_addf(&line, "%s missing\n", oid_to_hex(oid));
2810 cat_blob_write(line.buf, line.len);
2811 strbuf_release(&line);
2812 free(buf);
2813 return;
2814 }
2815 if (!buf)
2816 die("Can't read object %s", oid_to_hex(oid));
2817 if (type != OBJ_BLOB)
2818 die("Object %s is a %s but a blob was expected.",
2819 oid_to_hex(oid), type_name(type));
2820 strbuf_reset(&line);
2821 strbuf_addf(&line, "%s %s %"PRIuMAX"\n", oid_to_hex(oid),
2822 type_name(type), (uintmax_t)size);
2823 cat_blob_write(line.buf, line.len);
2824 strbuf_release(&line);
2825 cat_blob_write(buf, size);
2826 cat_blob_write("\n", 1);
2827 if (oe && oe->pack_id == pack_id) {
2828 last_blob.offset = oe->idx.offset;
2829 strbuf_attach(&last_blob.data, buf, size, size);
2830 last_blob.depth = oe->depth;
2831 } else
2832 free(buf);
2833 }
2834
2835 static void parse_get_mark(const char *p)
2836 {
2837 struct object_entry *oe;
2838 char output[GIT_MAX_HEXSZ + 2];
2839
2840 /* get-mark SP <object> LF */
2841 if (*p != ':')
2842 die("Not a mark: %s", p);
2843
2844 oe = find_mark(parse_mark_ref_eol(p));
2845 if (!oe)
2846 die("Unknown mark: %s", command_buf.buf);
2847
2848 xsnprintf(output, sizeof(output), "%s\n", oid_to_hex(&oe->idx.oid));
2849 cat_blob_write(output, GIT_SHA1_HEXSZ + 1);
2850 }
2851
2852 static void parse_cat_blob(const char *p)
2853 {
2854 struct object_entry *oe;
2855 struct object_id oid;
2856
2857 /* cat-blob SP <object> LF */
2858 if (*p == ':') {
2859 oe = find_mark(parse_mark_ref_eol(p));
2860 if (!oe)
2861 die("Unknown mark: %s", command_buf.buf);
2862 oidcpy(&oid, &oe->idx.oid);
2863 } else {
2864 if (parse_oid_hex(p, &oid, &p))
2865 die("Invalid dataref: %s", command_buf.buf);
2866 if (*p)
2867 die("Garbage after SHA1: %s", command_buf.buf);
2868 oe = find_object(&oid);
2869 }
2870
2871 cat_blob(oe, &oid);
2872 }
2873
2874 static struct object_entry *dereference(struct object_entry *oe,
2875 struct object_id *oid)
2876 {
2877 unsigned long size;
2878 char *buf = NULL;
2879 if (!oe) {
2880 enum object_type type = oid_object_info(the_repository, oid,
2881 NULL);
2882 if (type < 0)
2883 die("object not found: %s", oid_to_hex(oid));
2884 /* cache it! */
2885 oe = insert_object(oid);
2886 oe->type = type;
2887 oe->pack_id = MAX_PACK_ID;
2888 oe->idx.offset = 1;
2889 }
2890 switch (oe->type) {
2891 case OBJ_TREE: /* easy case. */
2892 return oe;
2893 case OBJ_COMMIT:
2894 case OBJ_TAG:
2895 break;
2896 default:
2897 die("Not a tree-ish: %s", command_buf.buf);
2898 }
2899
2900 if (oe->pack_id != MAX_PACK_ID) { /* in a pack being written */
2901 buf = gfi_unpack_entry(oe, &size);
2902 } else {
2903 enum object_type unused;
2904 buf = read_object_file(oid, &unused, &size);
2905 }
2906 if (!buf)
2907 die("Can't load object %s", oid_to_hex(oid));
2908
2909 /* Peel one layer. */
2910 switch (oe->type) {
2911 case OBJ_TAG:
2912 if (size < GIT_SHA1_HEXSZ + strlen("object ") ||
2913 get_oid_hex(buf + strlen("object "), oid))
2914 die("Invalid SHA1 in tag: %s", command_buf.buf);
2915 break;
2916 case OBJ_COMMIT:
2917 if (size < GIT_SHA1_HEXSZ + strlen("tree ") ||
2918 get_oid_hex(buf + strlen("tree "), oid))
2919 die("Invalid SHA1 in commit: %s", command_buf.buf);
2920 }
2921
2922 free(buf);
2923 return find_object(oid);
2924 }
2925
2926 static struct object_entry *parse_treeish_dataref(const char **p)
2927 {
2928 struct object_id oid;
2929 struct object_entry *e;
2930
2931 if (**p == ':') { /* <mark> */
2932 e = find_mark(parse_mark_ref_space(p));
2933 if (!e)
2934 die("Unknown mark: %s", command_buf.buf);
2935 oidcpy(&oid, &e->idx.oid);
2936 } else { /* <sha1> */
2937 if (parse_oid_hex(*p, &oid, p))
2938 die("Invalid dataref: %s", command_buf.buf);
2939 e = find_object(&oid);
2940 if (*(*p)++ != ' ')
2941 die("Missing space after tree-ish: %s", command_buf.buf);
2942 }
2943
2944 while (!e || e->type != OBJ_TREE)
2945 e = dereference(e, &oid);
2946 return e;
2947 }
2948
2949 static void print_ls(int mode, const unsigned char *sha1, const char *path)
2950 {
2951 static struct strbuf line = STRBUF_INIT;
2952
2953 /* See show_tree(). */
2954 const char *type =
2955 S_ISGITLINK(mode) ? commit_type :
2956 S_ISDIR(mode) ? tree_type :
2957 blob_type;
2958
2959 if (!mode) {
2960 /* missing SP path LF */
2961 strbuf_reset(&line);
2962 strbuf_addstr(&line, "missing ");
2963 quote_c_style(path, &line, NULL, 0);
2964 strbuf_addch(&line, '\n');
2965 } else {
2966 /* mode SP type SP object_name TAB path LF */
2967 strbuf_reset(&line);
2968 strbuf_addf(&line, "%06o %s %s\t",
2969 mode & ~NO_DELTA, type, sha1_to_hex(sha1));
2970 quote_c_style(path, &line, NULL, 0);
2971 strbuf_addch(&line, '\n');
2972 }
2973 cat_blob_write(line.buf, line.len);
2974 }
2975
2976 static void parse_ls(const char *p, struct branch *b)
2977 {
2978 struct tree_entry *root = NULL;
2979 struct tree_entry leaf = {NULL};
2980
2981 /* ls SP (<tree-ish> SP)? <path> */
2982 if (*p == '"') {
2983 if (!b)
2984 die("Not in a commit: %s", command_buf.buf);
2985 root = &b->branch_tree;
2986 } else {
2987 struct object_entry *e = parse_treeish_dataref(&p);
2988 root = new_tree_entry();
2989 oidcpy(&root->versions[1].oid, &e->idx.oid);
2990 if (!is_null_oid(&root->versions[1].oid))
2991 root->versions[1].mode = S_IFDIR;
2992 load_tree(root);
2993 }
2994 if (*p == '"') {
2995 static struct strbuf uq = STRBUF_INIT;
2996 const char *endp;
2997 strbuf_reset(&uq);
2998 if (unquote_c_style(&uq, p, &endp))
2999 die("Invalid path: %s", command_buf.buf);
3000 if (*endp)
3001 die("Garbage after path in: %s", command_buf.buf);
3002 p = uq.buf;
3003 }
3004 tree_content_get(root, p, &leaf, 1);
3005 /*
3006 * A directory in preparation would have a sha1 of zero
3007 * until it is saved. Save, for simplicity.
3008 */
3009 if (S_ISDIR(leaf.versions[1].mode))
3010 store_tree(&leaf);
3011
3012 print_ls(leaf.versions[1].mode, leaf.versions[1].oid.hash, p);
3013 if (leaf.tree)
3014 release_tree_content_recursive(leaf.tree);
3015 if (!b || root != &b->branch_tree)
3016 release_tree_entry(root);
3017 }
3018
3019 static void checkpoint(void)
3020 {
3021 checkpoint_requested = 0;
3022 if (object_count) {
3023 cycle_packfile();
3024 }
3025 dump_branches();
3026 dump_tags();
3027 dump_marks();
3028 }
3029
3030 static void parse_checkpoint(void)
3031 {
3032 checkpoint_requested = 1;
3033 skip_optional_lf();
3034 }
3035
3036 static void parse_progress(void)
3037 {
3038 fwrite(command_buf.buf, 1, command_buf.len, stdout);
3039 fputc('\n', stdout);
3040 fflush(stdout);
3041 skip_optional_lf();
3042 }
3043
3044 static char* make_fast_import_path(const char *path)
3045 {
3046 if (!relative_marks_paths || is_absolute_path(path))
3047 return xstrdup(path);
3048 return git_pathdup("info/fast-import/%s", path);
3049 }
3050
3051 static void option_import_marks(const char *marks,
3052 int from_stream, int ignore_missing)
3053 {
3054 if (import_marks_file) {
3055 if (from_stream)
3056 die("Only one import-marks command allowed per stream");
3057
3058 /* read previous mark file */
3059 if(!import_marks_file_from_stream)
3060 read_marks();
3061 }
3062
3063 import_marks_file = make_fast_import_path(marks);
3064 safe_create_leading_directories_const(import_marks_file);
3065 import_marks_file_from_stream = from_stream;
3066 import_marks_file_ignore_missing = ignore_missing;
3067 }
3068
3069 static void option_date_format(const char *fmt)
3070 {
3071 if (!strcmp(fmt, "raw"))
3072 whenspec = WHENSPEC_RAW;
3073 else if (!strcmp(fmt, "rfc2822"))
3074 whenspec = WHENSPEC_RFC2822;
3075 else if (!strcmp(fmt, "now"))
3076 whenspec = WHENSPEC_NOW;
3077 else
3078 die("unknown --date-format argument %s", fmt);
3079 }
3080
3081 static unsigned long ulong_arg(const char *option, const char *arg)
3082 {
3083 char *endptr;
3084 unsigned long rv = strtoul(arg, &endptr, 0);
3085 if (strchr(arg, '-') || endptr == arg || *endptr)
3086 die("%s: argument must be a non-negative integer", option);
3087 return rv;
3088 }
3089
3090 static void option_depth(const char *depth)
3091 {
3092 max_depth = ulong_arg("--depth", depth);
3093 if (max_depth > MAX_DEPTH)
3094 die("--depth cannot exceed %u", MAX_DEPTH);
3095 }
3096
3097 static void option_active_branches(const char *branches)
3098 {
3099 max_active_branches = ulong_arg("--active-branches", branches);
3100 }
3101
3102 static void option_export_marks(const char *marks)
3103 {
3104 export_marks_file = make_fast_import_path(marks);
3105 safe_create_leading_directories_const(export_marks_file);
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 (starts_with(option, "quiet")) {
3149 show_stats = 0;
3150 } else if (starts_with(option, "stats")) {
3151 show_stats = 1;
3152 } else {
3153 return 0;
3154 }
3155
3156 return 1;
3157 }
3158
3159 static int parse_one_feature(const char *feature, int from_stream)
3160 {
3161 const char *arg;
3162
3163 if (skip_prefix(feature, "date-format=", &arg)) {
3164 option_date_format(arg);
3165 } else if (skip_prefix(feature, "import-marks=", &arg)) {
3166 option_import_marks(arg, from_stream, 0);
3167 } else if (skip_prefix(feature, "import-marks-if-exists=", &arg)) {
3168 option_import_marks(arg, from_stream, 1);
3169 } else if (skip_prefix(feature, "export-marks=", &arg)) {
3170 option_export_marks(arg);
3171 } else if (!strcmp(feature, "get-mark")) {
3172 ; /* Don't die - this feature is supported */
3173 } else if (!strcmp(feature, "cat-blob")) {
3174 ; /* Don't die - this feature is supported */
3175 } else if (!strcmp(feature, "relative-marks")) {
3176 relative_marks_paths = 1;
3177 } else if (!strcmp(feature, "no-relative-marks")) {
3178 relative_marks_paths = 0;
3179 } else if (!strcmp(feature, "done")) {
3180 require_explicit_termination = 1;
3181 } else if (!strcmp(feature, "force")) {
3182 force_update = 1;
3183 } else if (!strcmp(feature, "notes") || !strcmp(feature, "ls")) {
3184 ; /* do nothing; we have the feature */
3185 } else {
3186 return 0;
3187 }
3188
3189 return 1;
3190 }
3191
3192 static void parse_feature(const char *feature)
3193 {
3194 if (seen_data_command)
3195 die("Got feature command '%s' after data command", feature);
3196
3197 if (parse_one_feature(feature, 1))
3198 return;
3199
3200 die("This version of fast-import does not support feature %s.", feature);
3201 }
3202
3203 static void parse_option(const char *option)
3204 {
3205 if (seen_data_command)
3206 die("Got option command '%s' after data command", option);
3207
3208 if (parse_one_option(option))
3209 return;
3210
3211 die("This version of fast-import does not support option: %s", option);
3212 }
3213
3214 static void git_pack_config(void)
3215 {
3216 int indexversion_value;
3217 int limit;
3218 unsigned long packsizelimit_value;
3219
3220 if (!git_config_get_ulong("pack.depth", &max_depth)) {
3221 if (max_depth > MAX_DEPTH)
3222 max_depth = MAX_DEPTH;
3223 }
3224 if (!git_config_get_int("pack.indexversion", &indexversion_value)) {
3225 pack_idx_opts.version = indexversion_value;
3226 if (pack_idx_opts.version > 2)
3227 git_die_config("pack.indexversion",
3228 "bad pack.indexversion=%"PRIu32, pack_idx_opts.version);
3229 }
3230 if (!git_config_get_ulong("pack.packsizelimit", &packsizelimit_value))
3231 max_packsize = packsizelimit_value;
3232
3233 if (!git_config_get_int("fastimport.unpacklimit", &limit))
3234 unpack_limit = limit;
3235 else if (!git_config_get_int("transfer.unpacklimit", &limit))
3236 unpack_limit = limit;
3237
3238 git_config(git_default_config, NULL);
3239 }
3240
3241 static const char fast_import_usage[] =
3242 "git fast-import [--date-format=<f>] [--max-pack-size=<n>] [--big-file-threshold=<n>] [--depth=<n>] [--active-branches=<n>] [--export-marks=<marks.file>]";
3243
3244 static void parse_argv(void)
3245 {
3246 unsigned int i;
3247
3248 for (i = 1; i < global_argc; i++) {
3249 const char *a = global_argv[i];
3250
3251 if (*a != '-' || !strcmp(a, "--"))
3252 break;
3253
3254 if (!skip_prefix(a, "--", &a))
3255 die("unknown option %s", a);
3256
3257 if (parse_one_option(a))
3258 continue;
3259
3260 if (parse_one_feature(a, 0))
3261 continue;
3262
3263 if (skip_prefix(a, "cat-blob-fd=", &a)) {
3264 option_cat_blob_fd(a);
3265 continue;
3266 }
3267
3268 die("unknown option --%s", a);
3269 }
3270 if (i != global_argc)
3271 usage(fast_import_usage);
3272
3273 seen_data_command = 1;
3274 if (import_marks_file)
3275 read_marks();
3276 }
3277
3278 int cmd_main(int argc, const char **argv)
3279 {
3280 unsigned int i;
3281
3282 if (argc == 2 && !strcmp(argv[1], "-h"))
3283 usage(fast_import_usage);
3284
3285 setup_git_directory();
3286 reset_pack_idx_option(&pack_idx_opts);
3287 git_pack_config();
3288
3289 alloc_objects(object_entry_alloc);
3290 strbuf_init(&command_buf, 0);
3291 atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
3292 branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
3293 avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
3294 marks = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
3295
3296 global_argc = argc;
3297 global_argv = argv;
3298
3299 rc_free = mem_pool_alloc(&fi_mem_pool, cmd_save * sizeof(*rc_free));
3300 for (i = 0; i < (cmd_save - 1); i++)
3301 rc_free[i].next = &rc_free[i + 1];
3302 rc_free[cmd_save - 1].next = NULL;
3303
3304 start_packfile();
3305 set_die_routine(die_nicely);
3306 set_checkpoint_signal();
3307 while (read_next_command() != EOF) {
3308 const char *v;
3309 if (!strcmp("blob", command_buf.buf))
3310 parse_new_blob();
3311 else if (skip_prefix(command_buf.buf, "commit ", &v))
3312 parse_new_commit(v);
3313 else if (skip_prefix(command_buf.buf, "tag ", &v))
3314 parse_new_tag(v);
3315 else if (skip_prefix(command_buf.buf, "reset ", &v))
3316 parse_reset_branch(v);
3317 else if (skip_prefix(command_buf.buf, "ls ", &v))
3318 parse_ls(v, NULL);
3319 else if (skip_prefix(command_buf.buf, "cat-blob ", &v))
3320 parse_cat_blob(v);
3321 else if (!strcmp("checkpoint", command_buf.buf))
3322 parse_checkpoint();
3323 else if (!strcmp("done", command_buf.buf))
3324 break;
3325 else if (starts_with(command_buf.buf, "progress "))
3326 parse_progress();
3327 else if (skip_prefix(command_buf.buf, "feature ", &v))
3328 parse_feature(v);
3329 else if (skip_prefix(command_buf.buf, "option git ", &v))
3330 parse_option(v);
3331 else if (starts_with(command_buf.buf, "option "))
3332 /* ignore non-git options*/;
3333 else
3334 die("Unsupported command: %s", command_buf.buf);
3335
3336 if (checkpoint_requested)
3337 checkpoint();
3338 }
3339
3340 /* argv hasn't been parsed yet, do so */
3341 if (!seen_data_command)
3342 parse_argv();
3343
3344 if (require_explicit_termination && feof(stdin))
3345 die("stream ends early");
3346
3347 end_packfile();
3348
3349 dump_branches();
3350 dump_tags();
3351 unkeep_all_packs();
3352 dump_marks();
3353
3354 if (pack_edges)
3355 fclose(pack_edges);
3356
3357 if (show_stats) {
3358 uintmax_t total_count = 0, duplicate_count = 0;
3359 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
3360 total_count += object_count_by_type[i];
3361 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
3362 duplicate_count += duplicate_count_by_type[i];
3363
3364 fprintf(stderr, "%s statistics:\n", argv[0]);
3365 fprintf(stderr, "---------------------------------------------------------------------\n");
3366 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
3367 fprintf(stderr, "Total objects: %10" PRIuMAX " (%10" PRIuMAX " duplicates )\n", total_count, duplicate_count);
3368 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]);
3369 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]);
3370 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]);
3371 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]);
3372 fprintf(stderr, "Total branches: %10lu (%10lu loads )\n", branch_count, branch_load_count);
3373 fprintf(stderr, " marks: %10" PRIuMAX " (%10" PRIuMAX " unique )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
3374 fprintf(stderr, " atoms: %10u\n", atom_cnt);
3375 fprintf(stderr, "Memory total: %10" PRIuMAX " KiB\n", (tree_entry_allocd + fi_mem_pool.pool_alloc + alloc_count*sizeof(struct object_entry))/1024);
3376 fprintf(stderr, " pools: %10lu KiB\n", (unsigned long)((tree_entry_allocd + fi_mem_pool.pool_alloc) /1024));
3377 fprintf(stderr, " objects: %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
3378 fprintf(stderr, "---------------------------------------------------------------------\n");
3379 pack_report();
3380 fprintf(stderr, "---------------------------------------------------------------------\n");
3381 fprintf(stderr, "\n");
3382 }
3383
3384 return failure ? 1 : 0;
3385 }
The Best VCS Around