Merge branch 'jk/fast-export-quote-path' into maint
[git/git.git] / builtin / fast-export.c
1 /*
2 * "git fast-export" builtin command
3 *
4 * Copyright (C) 2007 Johannes E. Schindelin
5 */
6 #include "builtin.h"
7 #include "cache.h"
8 #include "commit.h"
9 #include "object.h"
10 #include "tag.h"
11 #include "diff.h"
12 #include "diffcore.h"
13 #include "log-tree.h"
14 #include "revision.h"
15 #include "decorate.h"
16 #include "string-list.h"
17 #include "utf8.h"
18 #include "parse-options.h"
19 #include "quote.h"
20
21 static const char *fast_export_usage[] = {
22 "git fast-export [rev-list-opts]",
23 NULL
24 };
25
26 static int progress;
27 static enum { ABORT, VERBATIM, WARN, STRIP } signed_tag_mode = ABORT;
28 static enum { ERROR, DROP, REWRITE } tag_of_filtered_mode = ABORT;
29 static int fake_missing_tagger;
30 static int no_data;
31 static int full_tree;
32
33 static int parse_opt_signed_tag_mode(const struct option *opt,
34 const char *arg, int unset)
35 {
36 if (unset || !strcmp(arg, "abort"))
37 signed_tag_mode = ABORT;
38 else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
39 signed_tag_mode = VERBATIM;
40 else if (!strcmp(arg, "warn"))
41 signed_tag_mode = WARN;
42 else if (!strcmp(arg, "strip"))
43 signed_tag_mode = STRIP;
44 else
45 return error("Unknown signed-tag mode: %s", arg);
46 return 0;
47 }
48
49 static int parse_opt_tag_of_filtered_mode(const struct option *opt,
50 const char *arg, int unset)
51 {
52 if (unset || !strcmp(arg, "abort"))
53 tag_of_filtered_mode = ABORT;
54 else if (!strcmp(arg, "drop"))
55 tag_of_filtered_mode = DROP;
56 else if (!strcmp(arg, "rewrite"))
57 tag_of_filtered_mode = REWRITE;
58 else
59 return error("Unknown tag-of-filtered mode: %s", arg);
60 return 0;
61 }
62
63 static struct decoration idnums;
64 static uint32_t last_idnum;
65
66 static int has_unshown_parent(struct commit *commit)
67 {
68 struct commit_list *parent;
69
70 for (parent = commit->parents; parent; parent = parent->next)
71 if (!(parent->item->object.flags & SHOWN) &&
72 !(parent->item->object.flags & UNINTERESTING))
73 return 1;
74 return 0;
75 }
76
77 /* Since intptr_t is C99, we do not use it here */
78 static inline uint32_t *mark_to_ptr(uint32_t mark)
79 {
80 return ((uint32_t *)NULL) + mark;
81 }
82
83 static inline uint32_t ptr_to_mark(void * mark)
84 {
85 return (uint32_t *)mark - (uint32_t *)NULL;
86 }
87
88 static inline void mark_object(struct object *object, uint32_t mark)
89 {
90 add_decoration(&idnums, object, mark_to_ptr(mark));
91 }
92
93 static inline void mark_next_object(struct object *object)
94 {
95 mark_object(object, ++last_idnum);
96 }
97
98 static int get_object_mark(struct object *object)
99 {
100 void *decoration = lookup_decoration(&idnums, object);
101 if (!decoration)
102 return 0;
103 return ptr_to_mark(decoration);
104 }
105
106 static void show_progress(void)
107 {
108 static int counter = 0;
109 if (!progress)
110 return;
111 if ((++counter % progress) == 0)
112 printf("progress %d objects\n", counter);
113 }
114
115 static void handle_object(const unsigned char *sha1)
116 {
117 unsigned long size;
118 enum object_type type;
119 char *buf;
120 struct object *object;
121
122 if (no_data)
123 return;
124
125 if (is_null_sha1(sha1))
126 return;
127
128 object = parse_object(sha1);
129 if (!object)
130 die ("Could not read blob %s", sha1_to_hex(sha1));
131
132 if (object->flags & SHOWN)
133 return;
134
135 buf = read_sha1_file(sha1, &type, &size);
136 if (!buf)
137 die ("Could not read blob %s", sha1_to_hex(sha1));
138
139 mark_next_object(object);
140
141 printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
142 if (size && fwrite(buf, size, 1, stdout) != 1)
143 die_errno ("Could not write blob '%s'", sha1_to_hex(sha1));
144 printf("\n");
145
146 show_progress();
147
148 object->flags |= SHOWN;
149 free(buf);
150 }
151
152 static int depth_first(const void *a_, const void *b_)
153 {
154 const struct diff_filepair *a = *((const struct diff_filepair **)a_);
155 const struct diff_filepair *b = *((const struct diff_filepair **)b_);
156 const char *name_a, *name_b;
157 int len_a, len_b, len;
158 int cmp;
159
160 name_a = a->one ? a->one->path : a->two->path;
161 name_b = b->one ? b->one->path : b->two->path;
162
163 len_a = strlen(name_a);
164 len_b = strlen(name_b);
165 len = (len_a < len_b) ? len_a : len_b;
166
167 /* strcmp will sort 'd' before 'd/e', we want 'd/e' before 'd' */
168 cmp = memcmp(name_a, name_b, len);
169 if (cmp)
170 return cmp;
171 cmp = len_b - len_a;
172 if (cmp)
173 return cmp;
174 /*
175 * Move 'R'ename entries last so that all references of the file
176 * appear in the output before it is renamed (e.g., when a file
177 * was copied and renamed in the same commit).
178 */
179 return (a->status == 'R') - (b->status == 'R');
180 }
181
182 static void print_path(const char *path)
183 {
184 int need_quote = quote_c_style(path, NULL, NULL, 0);
185 if (need_quote)
186 quote_c_style(path, NULL, stdout, 0);
187 else
188 printf("%s", path);
189 }
190
191 static void show_filemodify(struct diff_queue_struct *q,
192 struct diff_options *options, void *data)
193 {
194 int i;
195
196 /*
197 * Handle files below a directory first, in case they are all deleted
198 * and the directory changes to a file or symlink.
199 */
200 qsort(q->queue, q->nr, sizeof(q->queue[0]), depth_first);
201
202 for (i = 0; i < q->nr; i++) {
203 struct diff_filespec *ospec = q->queue[i]->one;
204 struct diff_filespec *spec = q->queue[i]->two;
205
206 switch (q->queue[i]->status) {
207 case DIFF_STATUS_DELETED:
208 printf("D ");
209 print_path(spec->path);
210 putchar('\n');
211 break;
212
213 case DIFF_STATUS_COPIED:
214 case DIFF_STATUS_RENAMED:
215 printf("%c ", q->queue[i]->status);
216 print_path(ospec->path);
217 putchar(' ');
218 print_path(spec->path);
219 putchar('\n');
220
221 if (!hashcmp(ospec->sha1, spec->sha1) &&
222 ospec->mode == spec->mode)
223 break;
224 /* fallthrough */
225
226 case DIFF_STATUS_TYPE_CHANGED:
227 case DIFF_STATUS_MODIFIED:
228 case DIFF_STATUS_ADDED:
229 /*
230 * Links refer to objects in another repositories;
231 * output the SHA-1 verbatim.
232 */
233 if (no_data || S_ISGITLINK(spec->mode))
234 printf("M %06o %s ", spec->mode,
235 sha1_to_hex(spec->sha1));
236 else {
237 struct object *object = lookup_object(spec->sha1);
238 printf("M %06o :%d ", spec->mode,
239 get_object_mark(object));
240 }
241 print_path(spec->path);
242 putchar('\n');
243 break;
244
245 default:
246 die("Unexpected comparison status '%c' for %s, %s",
247 q->queue[i]->status,
248 ospec->path ? ospec->path : "none",
249 spec->path ? spec->path : "none");
250 }
251 }
252 }
253
254 static const char *find_encoding(const char *begin, const char *end)
255 {
256 const char *needle = "\nencoding ";
257 char *bol, *eol;
258
259 bol = memmem(begin, end ? end - begin : strlen(begin),
260 needle, strlen(needle));
261 if (!bol)
262 return git_commit_encoding;
263 bol += strlen(needle);
264 eol = strchrnul(bol, '\n');
265 *eol = '\0';
266 return bol;
267 }
268
269 static void handle_commit(struct commit *commit, struct rev_info *rev)
270 {
271 int saved_output_format = rev->diffopt.output_format;
272 const char *author, *author_end, *committer, *committer_end;
273 const char *encoding, *message;
274 char *reencoded = NULL;
275 struct commit_list *p;
276 int i;
277
278 rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
279
280 parse_commit(commit);
281 author = strstr(commit->buffer, "\nauthor ");
282 if (!author)
283 die ("Could not find author in commit %s",
284 sha1_to_hex(commit->object.sha1));
285 author++;
286 author_end = strchrnul(author, '\n');
287 committer = strstr(author_end, "\ncommitter ");
288 if (!committer)
289 die ("Could not find committer in commit %s",
290 sha1_to_hex(commit->object.sha1));
291 committer++;
292 committer_end = strchrnul(committer, '\n');
293 message = strstr(committer_end, "\n\n");
294 encoding = find_encoding(committer_end, message);
295 if (message)
296 message += 2;
297
298 if (commit->parents &&
299 get_object_mark(&commit->parents->item->object) != 0 &&
300 !full_tree) {
301 parse_commit(commit->parents->item);
302 diff_tree_sha1(commit->parents->item->tree->object.sha1,
303 commit->tree->object.sha1, "", &rev->diffopt);
304 }
305 else
306 diff_root_tree_sha1(commit->tree->object.sha1,
307 "", &rev->diffopt);
308
309 /* Export the referenced blobs, and remember the marks. */
310 for (i = 0; i < diff_queued_diff.nr; i++)
311 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
312 handle_object(diff_queued_diff.queue[i]->two->sha1);
313
314 mark_next_object(&commit->object);
315 if (!is_encoding_utf8(encoding))
316 reencoded = reencode_string(message, "UTF-8", encoding);
317 if (!commit->parents)
318 printf("reset %s\n", (const char*)commit->util);
319 printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
320 (const char *)commit->util, last_idnum,
321 (int)(author_end - author), author,
322 (int)(committer_end - committer), committer,
323 (unsigned)(reencoded
324 ? strlen(reencoded) : message
325 ? strlen(message) : 0),
326 reencoded ? reencoded : message ? message : "");
327 free(reencoded);
328
329 for (i = 0, p = commit->parents; p; p = p->next) {
330 int mark = get_object_mark(&p->item->object);
331 if (!mark)
332 continue;
333 if (i == 0)
334 printf("from :%d\n", mark);
335 else
336 printf("merge :%d\n", mark);
337 i++;
338 }
339
340 if (full_tree)
341 printf("deleteall\n");
342 log_tree_diff_flush(rev);
343 rev->diffopt.output_format = saved_output_format;
344
345 printf("\n");
346
347 show_progress();
348 }
349
350 static void handle_tail(struct object_array *commits, struct rev_info *revs)
351 {
352 struct commit *commit;
353 while (commits->nr) {
354 commit = (struct commit *)commits->objects[commits->nr - 1].item;
355 if (has_unshown_parent(commit))
356 return;
357 handle_commit(commit, revs);
358 commits->nr--;
359 }
360 }
361
362 static void handle_tag(const char *name, struct tag *tag)
363 {
364 unsigned long size;
365 enum object_type type;
366 char *buf;
367 const char *tagger, *tagger_end, *message;
368 size_t message_size = 0;
369 struct object *tagged;
370 int tagged_mark;
371 struct commit *p;
372
373 /* Trees have no identifer in fast-export output, thus we have no way
374 * to output tags of trees, tags of tags of trees, etc. Simply omit
375 * such tags.
376 */
377 tagged = tag->tagged;
378 while (tagged->type == OBJ_TAG) {
379 tagged = ((struct tag *)tagged)->tagged;
380 }
381 if (tagged->type == OBJ_TREE) {
382 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
383 sha1_to_hex(tag->object.sha1));
384 return;
385 }
386
387 buf = read_sha1_file(tag->object.sha1, &type, &size);
388 if (!buf)
389 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
390 message = memmem(buf, size, "\n\n", 2);
391 if (message) {
392 message += 2;
393 message_size = strlen(message);
394 }
395 tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
396 if (!tagger) {
397 if (fake_missing_tagger)
398 tagger = "tagger Unspecified Tagger "
399 "<unspecified-tagger> 0 +0000";
400 else
401 tagger = "";
402 tagger_end = tagger + strlen(tagger);
403 } else {
404 tagger++;
405 tagger_end = strchrnul(tagger, '\n');
406 }
407
408 /* handle signed tags */
409 if (message) {
410 const char *signature = strstr(message,
411 "\n-----BEGIN PGP SIGNATURE-----\n");
412 if (signature)
413 switch(signed_tag_mode) {
414 case ABORT:
415 die ("Encountered signed tag %s; use "
416 "--signed-tag=<mode> to handle it.",
417 sha1_to_hex(tag->object.sha1));
418 case WARN:
419 warning ("Exporting signed tag %s",
420 sha1_to_hex(tag->object.sha1));
421 /* fallthru */
422 case VERBATIM:
423 break;
424 case STRIP:
425 message_size = signature + 1 - message;
426 break;
427 }
428 }
429
430 /* handle tag->tagged having been filtered out due to paths specified */
431 tagged = tag->tagged;
432 tagged_mark = get_object_mark(tagged);
433 if (!tagged_mark) {
434 switch(tag_of_filtered_mode) {
435 case ABORT:
436 die ("Tag %s tags unexported object; use "
437 "--tag-of-filtered-object=<mode> to handle it.",
438 sha1_to_hex(tag->object.sha1));
439 case DROP:
440 /* Ignore this tag altogether */
441 return;
442 case REWRITE:
443 if (tagged->type != OBJ_COMMIT) {
444 die ("Tag %s tags unexported %s!",
445 sha1_to_hex(tag->object.sha1),
446 typename(tagged->type));
447 }
448 p = (struct commit *)tagged;
449 for (;;) {
450 if (p->parents && p->parents->next)
451 break;
452 if (p->object.flags & UNINTERESTING)
453 break;
454 if (!(p->object.flags & TREESAME))
455 break;
456 if (!p->parents)
457 die ("Can't find replacement commit for tag %s\n",
458 sha1_to_hex(tag->object.sha1));
459 p = p->parents->item;
460 }
461 tagged_mark = get_object_mark(&p->object);
462 }
463 }
464
465 if (!prefixcmp(name, "refs/tags/"))
466 name += 10;
467 printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
468 name, tagged_mark,
469 (int)(tagger_end - tagger), tagger,
470 tagger == tagger_end ? "" : "\n",
471 (int)message_size, (int)message_size, message ? message : "");
472 }
473
474 static void get_tags_and_duplicates(struct object_array *pending,
475 struct string_list *extra_refs)
476 {
477 struct tag *tag;
478 int i;
479
480 for (i = 0; i < pending->nr; i++) {
481 struct object_array_entry *e = pending->objects + i;
482 unsigned char sha1[20];
483 struct commit *commit = commit;
484 char *full_name;
485
486 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
487 continue;
488
489 switch (e->item->type) {
490 case OBJ_COMMIT:
491 commit = (struct commit *)e->item;
492 break;
493 case OBJ_TAG:
494 tag = (struct tag *)e->item;
495
496 /* handle nested tags */
497 while (tag && tag->object.type == OBJ_TAG) {
498 parse_object(tag->object.sha1);
499 string_list_append(extra_refs, full_name)->util = tag;
500 tag = (struct tag *)tag->tagged;
501 }
502 if (!tag)
503 die ("Tag %s points nowhere?", e->name);
504 switch(tag->object.type) {
505 case OBJ_COMMIT:
506 commit = (struct commit *)tag;
507 break;
508 case OBJ_BLOB:
509 handle_object(tag->object.sha1);
510 continue;
511 default: /* OBJ_TAG (nested tags) is already handled */
512 warning("Tag points to object of unexpected type %s, skipping.",
513 typename(tag->object.type));
514 continue;
515 }
516 break;
517 default:
518 warning("%s: Unexpected object of type %s, skipping.",
519 e->name,
520 typename(e->item->type));
521 continue;
522 }
523 if (commit->util)
524 /* more than one name for the same object */
525 string_list_append(extra_refs, full_name)->util = commit;
526 else
527 commit->util = full_name;
528 }
529 }
530
531 static void handle_tags_and_duplicates(struct string_list *extra_refs)
532 {
533 struct commit *commit;
534 int i;
535
536 for (i = extra_refs->nr - 1; i >= 0; i--) {
537 const char *name = extra_refs->items[i].string;
538 struct object *object = extra_refs->items[i].util;
539 switch (object->type) {
540 case OBJ_TAG:
541 handle_tag(name, (struct tag *)object);
542 break;
543 case OBJ_COMMIT:
544 /* create refs pointing to already seen commits */
545 commit = (struct commit *)object;
546 printf("reset %s\nfrom :%d\n\n", name,
547 get_object_mark(&commit->object));
548 show_progress();
549 break;
550 }
551 }
552 }
553
554 static void export_marks(char *file)
555 {
556 unsigned int i;
557 uint32_t mark;
558 struct object_decoration *deco = idnums.hash;
559 FILE *f;
560 int e = 0;
561
562 f = fopen(file, "w");
563 if (!f)
564 die_errno("Unable to open marks file %s for writing.", file);
565
566 for (i = 0; i < idnums.size; i++) {
567 if (deco->base && deco->base->type == 1) {
568 mark = ptr_to_mark(deco->decoration);
569 if (fprintf(f, ":%"PRIu32" %s\n", mark,
570 sha1_to_hex(deco->base->sha1)) < 0) {
571 e = 1;
572 break;
573 }
574 }
575 deco++;
576 }
577
578 e |= ferror(f);
579 e |= fclose(f);
580 if (e)
581 error("Unable to write marks file %s.", file);
582 }
583
584 static void import_marks(char *input_file)
585 {
586 char line[512];
587 FILE *f = fopen(input_file, "r");
588 if (!f)
589 die_errno("cannot read '%s'", input_file);
590
591 while (fgets(line, sizeof(line), f)) {
592 uint32_t mark;
593 char *line_end, *mark_end;
594 unsigned char sha1[20];
595 struct object *object;
596
597 line_end = strchr(line, '\n');
598 if (line[0] != ':' || !line_end)
599 die("corrupt mark line: %s", line);
600 *line_end = '\0';
601
602 mark = strtoumax(line + 1, &mark_end, 10);
603 if (!mark || mark_end == line + 1
604 || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
605 die("corrupt mark line: %s", line);
606
607 object = parse_object(sha1);
608 if (!object)
609 die ("Could not read blob %s", sha1_to_hex(sha1));
610
611 if (object->flags & SHOWN)
612 error("Object %s already has a mark", sha1);
613
614 mark_object(object, mark);
615 if (last_idnum < mark)
616 last_idnum = mark;
617
618 object->flags |= SHOWN;
619 }
620 fclose(f);
621 }
622
623 int cmd_fast_export(int argc, const char **argv, const char *prefix)
624 {
625 struct rev_info revs;
626 struct object_array commits = OBJECT_ARRAY_INIT;
627 struct string_list extra_refs = STRING_LIST_INIT_NODUP;
628 struct commit *commit;
629 char *export_filename = NULL, *import_filename = NULL;
630 struct option options[] = {
631 OPT_INTEGER(0, "progress", &progress,
632 "show progress after <n> objects"),
633 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
634 "select handling of signed tags",
635 parse_opt_signed_tag_mode),
636 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, "mode",
637 "select handling of tags that tag filtered objects",
638 parse_opt_tag_of_filtered_mode),
639 OPT_STRING(0, "export-marks", &export_filename, "file",
640 "Dump marks to this file"),
641 OPT_STRING(0, "import-marks", &import_filename, "file",
642 "Import marks from this file"),
643 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
644 "Fake a tagger when tags lack one"),
645 OPT_BOOLEAN(0, "full-tree", &full_tree,
646 "Output full tree for each commit"),
647 { OPTION_NEGBIT, 0, "data", &no_data, NULL,
648 "Skip output of blob data",
649 PARSE_OPT_NOARG | PARSE_OPT_NEGHELP, NULL, 1 },
650 OPT_END()
651 };
652
653 if (argc == 1)
654 usage_with_options (fast_export_usage, options);
655
656 /* we handle encodings */
657 git_config(git_default_config, NULL);
658
659 init_revisions(&revs, prefix);
660 revs.topo_order = 1;
661 revs.show_source = 1;
662 revs.rewrite_parents = 1;
663 argc = setup_revisions(argc, argv, &revs, NULL);
664 argc = parse_options(argc, argv, prefix, options, fast_export_usage, 0);
665 if (argc > 1)
666 usage_with_options (fast_export_usage, options);
667
668 if (import_filename)
669 import_marks(import_filename);
670
671 if (import_filename && revs.prune_data.nr)
672 full_tree = 1;
673
674 get_tags_and_duplicates(&revs.pending, &extra_refs);
675
676 if (prepare_revision_walk(&revs))
677 die("revision walk setup failed");
678 revs.diffopt.format_callback = show_filemodify;
679 DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
680 while ((commit = get_revision(&revs))) {
681 if (has_unshown_parent(commit)) {
682 add_object_array(&commit->object, NULL, &commits);
683 }
684 else {
685 handle_commit(commit, &revs);
686 handle_tail(&commits, &revs);
687 }
688 }
689
690 handle_tags_and_duplicates(&extra_refs);
691
692 if (export_filename)
693 export_marks(export_filename);
694
695 return 0;
696 }