remote-hg: handle another funny author line from http://scelenic.com/hg
[git/dscho.git] / builtin / fast-export.c
blob1213155acf5246fb5ab04c6729db4abdb3432646
1 /*
2 * "git fast-export" builtin command
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"
21 #define REF_HANDLED (ALL_REV_FLAGS + 1)
23 static const char *fast_export_usage[] = {
24 "git fast-export [rev-list-opts]",
25 NULL
28 static int progress;
29 static enum { ABORT, VERBATIM, WARN, STRIP } signed_tag_mode = ABORT;
30 static enum { ERROR, DROP, REWRITE } tag_of_filtered_mode = ERROR;
31 static int fake_missing_tagger;
32 static int use_done_feature;
33 static int no_data;
34 static int full_tree;
36 static int parse_opt_signed_tag_mode(const struct option *opt,
37 const char *arg, int unset)
39 if (unset || !strcmp(arg, "abort"))
40 signed_tag_mode = ABORT;
41 else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
42 signed_tag_mode = VERBATIM;
43 else if (!strcmp(arg, "warn"))
44 signed_tag_mode = WARN;
45 else if (!strcmp(arg, "strip"))
46 signed_tag_mode = STRIP;
47 else
48 return error("Unknown signed-tag mode: %s", arg);
49 return 0;
52 static int parse_opt_tag_of_filtered_mode(const struct option *opt,
53 const char *arg, int unset)
55 if (unset || !strcmp(arg, "abort"))
56 tag_of_filtered_mode = ERROR;
57 else if (!strcmp(arg, "drop"))
58 tag_of_filtered_mode = DROP;
59 else if (!strcmp(arg, "rewrite"))
60 tag_of_filtered_mode = REWRITE;
61 else
62 return error("Unknown tag-of-filtered mode: %s", arg);
63 return 0;
66 static struct decoration idnums;
67 static uint32_t last_idnum;
69 static int has_unshown_parent(struct commit *commit)
71 struct commit_list *parent;
73 for (parent = commit->parents; parent; parent = parent->next)
74 if (!(parent->item->object.flags & SHOWN) &&
75 !(parent->item->object.flags & UNINTERESTING))
76 return 1;
77 return 0;
80 /* Since intptr_t is C99, we do not use it here */
81 static inline uint32_t *mark_to_ptr(uint32_t mark)
83 return ((uint32_t *)NULL) + mark;
86 static inline uint32_t ptr_to_mark(void * mark)
88 return (uint32_t *)mark - (uint32_t *)NULL;
91 static inline void mark_object(struct object *object, uint32_t mark)
93 add_decoration(&idnums, object, mark_to_ptr(mark));
96 static inline void mark_next_object(struct object *object)
98 mark_object(object, ++last_idnum);
101 static int get_object_mark(struct object *object)
103 void *decoration = lookup_decoration(&idnums, object);
104 if (!decoration)
105 return 0;
106 return ptr_to_mark(decoration);
109 static void show_progress(void)
111 static int counter = 0;
112 if (!progress)
113 return;
114 if ((++counter % progress) == 0)
115 printf("progress %d objects\n", counter);
118 static void handle_object(const unsigned char *sha1)
120 unsigned long size;
121 enum object_type type;
122 char *buf;
123 struct object *object;
125 if (no_data)
126 return;
128 if (is_null_sha1(sha1))
129 return;
131 object = parse_object(sha1);
132 if (!object)
133 die ("Could not read blob %s", sha1_to_hex(sha1));
135 if (object->flags & SHOWN)
136 return;
138 buf = read_sha1_file(sha1, &type, &size);
139 if (!buf)
140 die ("Could not read blob %s", sha1_to_hex(sha1));
142 mark_next_object(object);
144 printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
145 if (size && fwrite(buf, size, 1, stdout) != 1)
146 die_errno ("Could not write blob '%s'", sha1_to_hex(sha1));
147 printf("\n");
149 show_progress();
151 object->flags |= SHOWN;
152 free(buf);
155 static int depth_first(const void *a_, const void *b_)
157 const struct diff_filepair *a = *((const struct diff_filepair **)a_);
158 const struct diff_filepair *b = *((const struct diff_filepair **)b_);
159 const char *name_a, *name_b;
160 int len_a, len_b, len;
161 int cmp;
163 name_a = a->one ? a->one->path : a->two->path;
164 name_b = b->one ? b->one->path : b->two->path;
166 len_a = strlen(name_a);
167 len_b = strlen(name_b);
168 len = (len_a < len_b) ? len_a : len_b;
170 /* strcmp will sort 'd' before 'd/e', we want 'd/e' before 'd' */
171 cmp = memcmp(name_a, name_b, len);
172 if (cmp)
173 return cmp;
174 cmp = len_b - len_a;
175 if (cmp)
176 return cmp;
178 * Move 'R'ename entries last so that all references of the file
179 * appear in the output before it is renamed (e.g., when a file
180 * was copied and renamed in the same commit).
182 return (a->status == 'R') - (b->status == 'R');
185 static void print_path(const char *path)
187 int need_quote = quote_c_style(path, NULL, NULL, 0);
188 if (need_quote)
189 quote_c_style(path, NULL, stdout, 0);
190 else
191 printf("%s", path);
194 static void show_filemodify(struct diff_queue_struct *q,
195 struct diff_options *options, void *data)
197 int i;
200 * Handle files below a directory first, in case they are all deleted
201 * and the directory changes to a file or symlink.
203 qsort(q->queue, q->nr, sizeof(q->queue[0]), depth_first);
205 for (i = 0; i < q->nr; i++) {
206 struct diff_filespec *ospec = q->queue[i]->one;
207 struct diff_filespec *spec = q->queue[i]->two;
209 switch (q->queue[i]->status) {
210 case DIFF_STATUS_DELETED:
211 printf("D ");
212 print_path(spec->path);
213 putchar('\n');
214 break;
216 case DIFF_STATUS_COPIED:
217 case DIFF_STATUS_RENAMED:
218 printf("%c ", q->queue[i]->status);
219 print_path(ospec->path);
220 putchar(' ');
221 print_path(spec->path);
222 putchar('\n');
224 if (!hashcmp(ospec->sha1, spec->sha1) &&
225 ospec->mode == spec->mode)
226 break;
227 /* fallthrough */
229 case DIFF_STATUS_TYPE_CHANGED:
230 case DIFF_STATUS_MODIFIED:
231 case DIFF_STATUS_ADDED:
233 * Links refer to objects in another repositories;
234 * output the SHA-1 verbatim.
236 if (no_data || S_ISGITLINK(spec->mode))
237 printf("M %06o %s ", spec->mode,
238 sha1_to_hex(spec->sha1));
239 else {
240 struct object *object = lookup_object(spec->sha1);
241 printf("M %06o :%d ", spec->mode,
242 get_object_mark(object));
244 print_path(spec->path);
245 putchar('\n');
246 break;
248 default:
249 die("Unexpected comparison status '%c' for %s, %s",
250 q->queue[i]->status,
251 ospec->path ? ospec->path : "none",
252 spec->path ? spec->path : "none");
257 static const char *find_encoding(const char *begin, const char *end)
259 const char *needle = "\nencoding ";
260 char *bol, *eol;
262 bol = memmem(begin, end ? end - begin : strlen(begin),
263 needle, strlen(needle));
264 if (!bol)
265 return git_commit_encoding;
266 bol += strlen(needle);
267 eol = strchrnul(bol, '\n');
268 *eol = '\0';
269 return bol;
272 static void handle_commit(struct commit *commit, struct rev_info *rev)
274 int saved_output_format = rev->diffopt.output_format;
275 const char *author, *author_end, *committer, *committer_end;
276 const char *encoding, *message;
277 char *reencoded = NULL;
278 struct commit_list *p;
279 int i;
281 rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
283 parse_commit(commit);
284 author = strstr(commit->buffer, "\nauthor ");
285 if (!author)
286 die ("Could not find author in commit %s",
287 sha1_to_hex(commit->object.sha1));
288 author++;
289 author_end = strchrnul(author, '\n');
290 committer = strstr(author_end, "\ncommitter ");
291 if (!committer)
292 die ("Could not find committer in commit %s",
293 sha1_to_hex(commit->object.sha1));
294 committer++;
295 committer_end = strchrnul(committer, '\n');
296 message = strstr(committer_end, "\n\n");
297 encoding = find_encoding(committer_end, message);
298 if (message)
299 message += 2;
301 if (commit->parents &&
302 get_object_mark(&commit->parents->item->object) != 0 &&
303 !full_tree) {
304 parse_commit(commit->parents->item);
305 diff_tree_sha1(commit->parents->item->tree->object.sha1,
306 commit->tree->object.sha1, "", &rev->diffopt);
308 else
309 diff_root_tree_sha1(commit->tree->object.sha1,
310 "", &rev->diffopt);
312 /* Export the referenced blobs, and remember the marks. */
313 for (i = 0; i < diff_queued_diff.nr; i++)
314 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
315 handle_object(diff_queued_diff.queue[i]->two->sha1);
317 mark_next_object(&commit->object);
318 if (!is_encoding_utf8(encoding))
319 reencoded = reencode_string(message, "UTF-8", encoding);
320 if (!commit->parents)
321 printf("reset %s\n", (const char*)commit->util);
322 printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
323 (const char *)commit->util, last_idnum,
324 (int)(author_end - author), author,
325 (int)(committer_end - committer), committer,
326 (unsigned)(reencoded
327 ? strlen(reencoded) : message
328 ? strlen(message) : 0),
329 reencoded ? reencoded : message ? message : "");
330 free(reencoded);
332 for (i = 0, p = commit->parents; p; p = p->next) {
333 int mark = get_object_mark(&p->item->object);
334 if (!mark)
335 continue;
336 if (i == 0)
337 printf("from :%d\n", mark);
338 else
339 printf("merge :%d\n", mark);
340 i++;
343 if (full_tree)
344 printf("deleteall\n");
345 log_tree_diff_flush(rev);
346 rev->diffopt.output_format = saved_output_format;
348 printf("\n");
350 show_progress();
353 static void handle_tail(struct object_array *commits, struct rev_info *revs)
355 struct commit *commit;
356 while (commits->nr) {
357 commit = (struct commit *)commits->objects[commits->nr - 1].item;
358 if (has_unshown_parent(commit))
359 return;
360 handle_commit(commit, revs);
361 commits->nr--;
365 static void handle_tag(const char *name, struct tag *tag)
367 unsigned long size;
368 enum object_type type;
369 char *buf;
370 const char *tagger, *tagger_end, *message;
371 size_t message_size = 0;
372 struct object *tagged;
373 int tagged_mark;
374 struct commit *p;
376 /* Trees have no identifer in fast-export output, thus we have no way
377 * to output tags of trees, tags of tags of trees, etc. Simply omit
378 * such tags.
380 tagged = tag->tagged;
381 while (tagged->type == OBJ_TAG) {
382 tagged = ((struct tag *)tagged)->tagged;
384 if (tagged->type == OBJ_TREE) {
385 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
386 sha1_to_hex(tag->object.sha1));
387 return;
390 buf = read_sha1_file(tag->object.sha1, &type, &size);
391 if (!buf)
392 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
393 message = memmem(buf, size, "\n\n", 2);
394 if (message) {
395 message += 2;
396 message_size = strlen(message);
398 tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
399 if (!tagger) {
400 if (fake_missing_tagger)
401 tagger = "tagger Unspecified Tagger "
402 "<unspecified-tagger> 0 +0000";
403 else
404 tagger = "";
405 tagger_end = tagger + strlen(tagger);
406 } else {
407 tagger++;
408 tagger_end = strchrnul(tagger, '\n');
411 /* handle signed tags */
412 if (message) {
413 const char *signature = strstr(message,
414 "\n-----BEGIN PGP SIGNATURE-----\n");
415 if (signature)
416 switch(signed_tag_mode) {
417 case ABORT:
418 die ("Encountered signed tag %s; use "
419 "--signed-tag=<mode> to handle it.",
420 sha1_to_hex(tag->object.sha1));
421 case WARN:
422 warning ("Exporting signed tag %s",
423 sha1_to_hex(tag->object.sha1));
424 /* fallthru */
425 case VERBATIM:
426 break;
427 case STRIP:
428 message_size = signature + 1 - message;
429 break;
433 /* handle tag->tagged having been filtered out due to paths specified */
434 tagged = tag->tagged;
435 tagged_mark = get_object_mark(tagged);
436 if (!tagged_mark) {
437 switch(tag_of_filtered_mode) {
438 case ABORT:
439 die ("Tag %s tags unexported object; use "
440 "--tag-of-filtered-object=<mode> to handle it.",
441 sha1_to_hex(tag->object.sha1));
442 case DROP:
443 /* Ignore this tag altogether */
444 return;
445 case REWRITE:
446 if (tagged->type != OBJ_COMMIT) {
447 die ("Tag %s tags unexported %s!",
448 sha1_to_hex(tag->object.sha1),
449 typename(tagged->type));
451 p = (struct commit *)tagged;
452 for (;;) {
453 if (p->parents && p->parents->next)
454 break;
455 if (p->object.flags & UNINTERESTING)
456 break;
457 if (!(p->object.flags & TREESAME))
458 break;
459 if (!p->parents)
460 die ("Can't find replacement commit for tag %s\n",
461 sha1_to_hex(tag->object.sha1));
462 p = p->parents->item;
464 tagged_mark = get_object_mark(&p->object);
468 if (!prefixcmp(name, "refs/tags/"))
469 name += 10;
470 printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
471 name, tagged_mark,
472 (int)(tagger_end - tagger), tagger,
473 tagger == tagger_end ? "" : "\n",
474 (int)message_size, (int)message_size, message ? message : "");
477 static void get_tags_and_duplicates(struct object_array *pending,
478 struct string_list *refs,
479 struct string_list *extra_refs)
481 struct tag *tag;
482 int i;
484 for (i = 0; i < pending->nr; i++) {
485 struct object_array_entry *e = pending->objects + i;
486 unsigned char sha1[20];
487 struct commit *commit = commit;
488 char *full_name;
490 if ((e->flags & UNINTERESTING) || dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
491 continue;
493 switch (e->item->type) {
494 case OBJ_COMMIT:
495 commit = (struct commit *)e->item;
496 break;
497 case OBJ_TAG:
498 tag = (struct tag *)e->item;
500 /* handle nested tags */
501 while (tag && tag->object.type == OBJ_TAG) {
502 parse_object(tag->object.sha1);
503 string_list_append(extra_refs, full_name)->util = tag;
504 tag = (struct tag *)tag->tagged;
506 if (!tag)
507 die ("Tag %s points nowhere?", e->name);
508 switch(tag->object.type) {
509 case OBJ_COMMIT:
510 commit = (struct commit *)tag;
511 break;
512 case OBJ_BLOB:
513 handle_object(tag->object.sha1);
514 continue;
515 default: /* OBJ_TAG (nested tags) is already handled */
516 warning("Tag points to object of unexpected type %s, skipping.",
517 typename(tag->object.type));
518 continue;
520 break;
521 default:
522 warning("%s: Unexpected object of type %s, skipping.",
523 e->name,
524 typename(e->item->type));
525 continue;
527 if (commit->util)
528 /* more than one name for the same object */
529 string_list_append(extra_refs, full_name)->util = commit;
530 else {
531 commit->util = full_name;
532 /* we might need to set this ref explicitly */
533 string_list_append(refs, full_name)->util = commit;
538 static void handle_reset(const char *name, struct object *object)
540 int mark = get_object_mark(object);
542 if (mark)
543 printf("reset %s\nfrom :%d\n\n", name,
544 get_object_mark(object));
545 else
546 printf("reset %s\nfrom %s\n\n", name,
547 sha1_to_hex(object->sha1));
550 static void handle_tags_and_duplicates(struct string_list *refs, struct string_list *extra_refs)
552 int i;
554 /* even if no commits were exported, we need to export the ref */
555 for (i = refs->nr - 1; i >= 0; i--) {
556 const char *name = refs->items[i].string;
557 struct object *object = refs->items[i].util;
559 if (!(object->flags & REF_HANDLED)) {
560 if (object->type & OBJ_TAG)
561 handle_tag(name, (struct tag *)object);
562 else {
563 if (!prefixcmp(name, "refs/tags/") &&
564 (tag_of_filtered_mode != REWRITE ||
565 !get_object_mark(object)))
566 continue;
567 handle_reset(name, object);
568 object->flags |= REF_HANDLED;
573 for (i = extra_refs->nr - 1; i >= 0; i--) {
574 const char *name = extra_refs->items[i].string;
575 struct object *object = extra_refs->items[i].util;
576 switch (object->type) {
577 case OBJ_TAG:
578 handle_tag(name, (struct tag *)object);
579 break;
580 case OBJ_COMMIT:
581 /* create refs pointing to already seen commits */
582 handle_reset(name, object);
583 show_progress();
584 break;
589 static void export_marks(char *file)
591 unsigned int i;
592 uint32_t mark;
593 struct object_decoration *deco = idnums.hash;
594 FILE *f;
595 int e = 0;
597 f = fopen(file, "w");
598 if (!f)
599 die_errno("Unable to open marks file %s for writing.", file);
601 for (i = 0; i < idnums.size; i++) {
602 if (deco->base && deco->base->type == 1) {
603 mark = ptr_to_mark(deco->decoration);
604 if (fprintf(f, ":%"PRIu32" %s\n", mark,
605 sha1_to_hex(deco->base->sha1)) < 0) {
606 e = 1;
607 break;
610 deco++;
613 e |= ferror(f);
614 e |= fclose(f);
615 if (e)
616 error("Unable to write marks file %s.", file);
619 static void import_marks(char *input_file)
621 char line[512];
622 FILE *f = fopen(input_file, "r");
623 if (!f)
624 die_errno("cannot read '%s'", input_file);
626 while (fgets(line, sizeof(line), f)) {
627 uint32_t mark;
628 char *line_end, *mark_end;
629 unsigned char sha1[20];
630 struct object *object;
632 line_end = strchr(line, '\n');
633 if (line[0] != ':' || !line_end)
634 die("corrupt mark line: %s", line);
635 *line_end = '\0';
637 mark = strtoumax(line + 1, &mark_end, 10);
638 if (!mark || mark_end == line + 1
639 || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
640 die("corrupt mark line: %s", line);
642 object = parse_object(sha1);
643 if (!object)
644 die ("Could not read blob %s", sha1_to_hex(sha1));
646 if (object->flags & SHOWN)
647 error("Object %s already has a mark", sha1_to_hex(sha1));
649 mark_object(object, mark);
650 if (last_idnum < mark)
651 last_idnum = mark;
653 object->flags |= SHOWN;
655 fclose(f);
658 int cmd_fast_export(int argc, const char **argv, const char *prefix)
660 struct rev_info revs;
661 struct object_array commits = OBJECT_ARRAY_INIT;
662 struct string_list refs = STRING_LIST_INIT_NODUP, extra_refs = STRING_LIST_INIT_NODUP;
663 struct commit *commit;
664 char *export_filename = NULL, *import_filename = NULL;
665 struct option options[] = {
666 OPT_INTEGER(0, "progress", &progress,
667 "show progress after <n> objects"),
668 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
669 "select handling of signed tags",
670 parse_opt_signed_tag_mode),
671 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, "mode",
672 "select handling of tags that tag filtered objects",
673 parse_opt_tag_of_filtered_mode),
674 OPT_STRING(0, "export-marks", &export_filename, "file",
675 "Dump marks to this file"),
676 OPT_STRING(0, "import-marks", &import_filename, "file",
677 "Import marks from this file"),
678 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
679 "Fake a tagger when tags lack one"),
680 OPT_BOOLEAN(0, "full-tree", &full_tree,
681 "Output full tree for each commit"),
682 OPT_BOOLEAN(0, "use-done-feature", &use_done_feature,
683 "Use the done feature to terminate the stream"),
684 OPT_BOOL(0, "no-data", &no_data, "Skip output of blob data"),
685 OPT_END()
688 if (argc == 1)
689 usage_with_options (fast_export_usage, options);
691 /* we handle encodings */
692 git_config(git_default_config, NULL);
694 init_revisions(&revs, prefix);
695 revs.topo_order = 1;
696 revs.show_source = 1;
697 revs.rewrite_parents = 1;
698 argc = setup_revisions(argc, argv, &revs, NULL);
699 argc = parse_options(argc, argv, prefix, options, fast_export_usage, 0);
700 if (argc > 1)
701 usage_with_options (fast_export_usage, options);
703 if (use_done_feature)
704 printf("feature done\n");
706 if (import_filename)
707 import_marks(import_filename);
709 if (import_filename && revs.prune_data.nr)
710 full_tree = 1;
712 get_tags_and_duplicates(&revs.pending, &refs, &extra_refs);
714 if (prepare_revision_walk(&revs))
715 die("revision walk setup failed");
716 revs.diffopt.format_callback = show_filemodify;
717 DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
718 while ((commit = get_revision(&revs))) {
719 if (has_unshown_parent(commit)) {
720 add_object_array(&commit->object, NULL, &commits);
722 else {
723 handle_commit(commit, &revs);
724 commit->object.flags |= REF_HANDLED;
725 handle_tail(&commits, &revs);
729 handle_tags_and_duplicates(&refs, &extra_refs);
731 if (export_filename)
732 export_marks(export_filename);
734 if (use_done_feature)
735 printf("done\n");
737 return 0;