Merge branch 'maint-1.8.2' into maint
[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         N_("git fast-export [rev-list-opts]"),
23         NULL
24 };
25
26 static int progress;
27 static enum { ABORT, VERBATIM, WARN, WARN_STRIP, STRIP } signed_tag_mode = ABORT;
28 static enum { ERROR, DROP, REWRITE } tag_of_filtered_mode = ERROR;
29 static int fake_missing_tagger;
30 static int use_done_feature;
31 static int no_data;
32 static int full_tree;
33
34 static int parse_opt_signed_tag_mode(const struct option *opt,
35                                      const char *arg, int unset)
36 {
37         if (unset || !strcmp(arg, "abort"))
38                 signed_tag_mode = ABORT;
39         else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
40                 signed_tag_mode = VERBATIM;
41         else if (!strcmp(arg, "warn"))
42                 signed_tag_mode = WARN;
43         else if (!strcmp(arg, "warn-strip"))
44                 signed_tag_mode = WARN_STRIP;
45         else if (!strcmp(arg, "strip"))
46                 signed_tag_mode = STRIP;
47         else
48                 return error("Unknown signed-tags mode: %s", arg);
49         return 0;
50 }
51
52 static int parse_opt_tag_of_filtered_mode(const struct option *opt,
53                                           const char *arg, int unset)
54 {
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;
64 }
65
66 static struct decoration idnums;
67 static uint32_t last_idnum;
68
69 static int has_unshown_parent(struct commit *commit)
70 {
71         struct commit_list *parent;
72
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;
78 }
79
80 /* Since intptr_t is C99, we do not use it here */
81 static inline uint32_t *mark_to_ptr(uint32_t mark)
82 {
83         return ((uint32_t *)NULL) + mark;
84 }
85
86 static inline uint32_t ptr_to_mark(void * mark)
87 {
88         return (uint32_t *)mark - (uint32_t *)NULL;
89 }
90
91 static inline void mark_object(struct object *object, uint32_t mark)
92 {
93         add_decoration(&idnums, object, mark_to_ptr(mark));
94 }
95
96 static inline void mark_next_object(struct object *object)
97 {
98         mark_object(object, ++last_idnum);
99 }
100
101 static int get_object_mark(struct object *object)
102 {
103         void *decoration = lookup_decoration(&idnums, object);
104         if (!decoration)
105                 return 0;
106         return ptr_to_mark(decoration);
107 }
108
109 static void show_progress(void)
110 {
111         static int counter = 0;
112         if (!progress)
113                 return;
114         if ((++counter % progress) == 0)
115                 printf("progress %d objects\n", counter);
116 }
117
118 static void export_blob(const unsigned char *sha1)
119 {
120         unsigned long size;
121         enum object_type type;
122         char *buf;
123         struct object *object;
124         int eaten;
125
126         if (no_data)
127                 return;
128
129         if (is_null_sha1(sha1))
130                 return;
131
132         object = lookup_object(sha1);
133         if (object && object->flags & SHOWN)
134                 return;
135
136         buf = read_sha1_file(sha1, &type, &size);
137         if (!buf)
138                 die ("Could not read blob %s", sha1_to_hex(sha1));
139         if (check_sha1_signature(sha1, buf, size, typename(type)) < 0)
140                 die("sha1 mismatch in blob %s", sha1_to_hex(sha1));
141         object = parse_object_buffer(sha1, type, size, buf, &eaten);
142         if (!object)
143                 die("Could not read blob %s", sha1_to_hex(sha1));
144
145         mark_next_object(object);
146
147         printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
148         if (size && fwrite(buf, size, 1, stdout) != 1)
149                 die_errno ("Could not write blob '%s'", sha1_to_hex(sha1));
150         printf("\n");
151
152         show_progress();
153
154         object->flags |= SHOWN;
155         if (!eaten)
156                 free(buf);
157 }
158
159 static int depth_first(const void *a_, const void *b_)
160 {
161         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
162         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
163         const char *name_a, *name_b;
164         int len_a, len_b, len;
165         int cmp;
166
167         name_a = a->one ? a->one->path : a->two->path;
168         name_b = b->one ? b->one->path : b->two->path;
169
170         len_a = strlen(name_a);
171         len_b = strlen(name_b);
172         len = (len_a < len_b) ? len_a : len_b;
173
174         /* strcmp will sort 'd' before 'd/e', we want 'd/e' before 'd' */
175         cmp = memcmp(name_a, name_b, len);
176         if (cmp)
177                 return cmp;
178         cmp = len_b - len_a;
179         if (cmp)
180                 return cmp;
181         /*
182          * Move 'R'ename entries last so that all references of the file
183          * appear in the output before it is renamed (e.g., when a file
184          * was copied and renamed in the same commit).
185          */
186         return (a->status == 'R') - (b->status == 'R');
187 }
188
189 static void print_path(const char *path)
190 {
191         int need_quote = quote_c_style(path, NULL, NULL, 0);
192         if (need_quote)
193                 quote_c_style(path, NULL, stdout, 0);
194         else if (strchr(path, ' '))
195                 printf("\"%s\"", path);
196         else
197                 printf("%s", path);
198 }
199
200 static void show_filemodify(struct diff_queue_struct *q,
201                             struct diff_options *options, void *data)
202 {
203         int i;
204
205         /*
206          * Handle files below a directory first, in case they are all deleted
207          * and the directory changes to a file or symlink.
208          */
209         qsort(q->queue, q->nr, sizeof(q->queue[0]), depth_first);
210
211         for (i = 0; i < q->nr; i++) {
212                 struct diff_filespec *ospec = q->queue[i]->one;
213                 struct diff_filespec *spec = q->queue[i]->two;
214
215                 switch (q->queue[i]->status) {
216                 case DIFF_STATUS_DELETED:
217                         printf("D ");
218                         print_path(spec->path);
219                         putchar('\n');
220                         break;
221
222                 case DIFF_STATUS_COPIED:
223                 case DIFF_STATUS_RENAMED:
224                         printf("%c ", q->queue[i]->status);
225                         print_path(ospec->path);
226                         putchar(' ');
227                         print_path(spec->path);
228                         putchar('\n');
229
230                         if (!hashcmp(ospec->sha1, spec->sha1) &&
231                             ospec->mode == spec->mode)
232                                 break;
233                         /* fallthrough */
234
235                 case DIFF_STATUS_TYPE_CHANGED:
236                 case DIFF_STATUS_MODIFIED:
237                 case DIFF_STATUS_ADDED:
238                         /*
239                          * Links refer to objects in another repositories;
240                          * output the SHA-1 verbatim.
241                          */
242                         if (no_data || S_ISGITLINK(spec->mode))
243                                 printf("M %06o %s ", spec->mode,
244                                        sha1_to_hex(spec->sha1));
245                         else {
246                                 struct object *object = lookup_object(spec->sha1);
247                                 printf("M %06o :%d ", spec->mode,
248                                        get_object_mark(object));
249                         }
250                         print_path(spec->path);
251                         putchar('\n');
252                         break;
253
254                 default:
255                         die("Unexpected comparison status '%c' for %s, %s",
256                                 q->queue[i]->status,
257                                 ospec->path ? ospec->path : "none",
258                                 spec->path ? spec->path : "none");
259                 }
260         }
261 }
262
263 static const char *find_encoding(const char *begin, const char *end)
264 {
265         const char *needle = "\nencoding ";
266         char *bol, *eol;
267
268         bol = memmem(begin, end ? end - begin : strlen(begin),
269                      needle, strlen(needle));
270         if (!bol)
271                 return git_commit_encoding;
272         bol += strlen(needle);
273         eol = strchrnul(bol, '\n');
274         *eol = '\0';
275         return bol;
276 }
277
278 static void handle_commit(struct commit *commit, struct rev_info *rev)
279 {
280         int saved_output_format = rev->diffopt.output_format;
281         const char *author, *author_end, *committer, *committer_end;
282         const char *encoding, *message;
283         char *reencoded = NULL;
284         struct commit_list *p;
285         int i;
286
287         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
288
289         parse_commit(commit);
290         author = strstr(commit->buffer, "\nauthor ");
291         if (!author)
292                 die ("Could not find author in commit %s",
293                      sha1_to_hex(commit->object.sha1));
294         author++;
295         author_end = strchrnul(author, '\n');
296         committer = strstr(author_end, "\ncommitter ");
297         if (!committer)
298                 die ("Could not find committer in commit %s",
299                      sha1_to_hex(commit->object.sha1));
300         committer++;
301         committer_end = strchrnul(committer, '\n');
302         message = strstr(committer_end, "\n\n");
303         encoding = find_encoding(committer_end, message);
304         if (message)
305                 message += 2;
306
307         if (commit->parents &&
308             get_object_mark(&commit->parents->item->object) != 0 &&
309             !full_tree) {
310                 parse_commit(commit->parents->item);
311                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
312                                commit->tree->object.sha1, "", &rev->diffopt);
313         }
314         else
315                 diff_root_tree_sha1(commit->tree->object.sha1,
316                                     "", &rev->diffopt);
317
318         /* Export the referenced blobs, and remember the marks. */
319         for (i = 0; i < diff_queued_diff.nr; i++)
320                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
321                         export_blob(diff_queued_diff.queue[i]->two->sha1);
322
323         mark_next_object(&commit->object);
324         if (!is_encoding_utf8(encoding))
325                 reencoded = reencode_string(message, "UTF-8", encoding);
326         if (!commit->parents)
327                 printf("reset %s\n", (const char*)commit->util);
328         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
329                (const char *)commit->util, last_idnum,
330                (int)(author_end - author), author,
331                (int)(committer_end - committer), committer,
332                (unsigned)(reencoded
333                           ? strlen(reencoded) : message
334                           ? strlen(message) : 0),
335                reencoded ? reencoded : message ? message : "");
336         free(reencoded);
337
338         for (i = 0, p = commit->parents; p; p = p->next) {
339                 int mark = get_object_mark(&p->item->object);
340                 if (!mark)
341                         continue;
342                 if (i == 0)
343                         printf("from :%d\n", mark);
344                 else
345                         printf("merge :%d\n", mark);
346                 i++;
347         }
348
349         if (full_tree)
350                 printf("deleteall\n");
351         log_tree_diff_flush(rev);
352         rev->diffopt.output_format = saved_output_format;
353
354         printf("\n");
355
356         show_progress();
357 }
358
359 static void handle_tail(struct object_array *commits, struct rev_info *revs)
360 {
361         struct commit *commit;
362         while (commits->nr) {
363                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
364                 if (has_unshown_parent(commit))
365                         return;
366                 handle_commit(commit, revs);
367                 commits->nr--;
368         }
369 }
370
371 static void handle_tag(const char *name, struct tag *tag)
372 {
373         unsigned long size;
374         enum object_type type;
375         char *buf;
376         const char *tagger, *tagger_end, *message;
377         size_t message_size = 0;
378         struct object *tagged;
379         int tagged_mark;
380         struct commit *p;
381
382         /* Trees have no identifer in fast-export output, thus we have no way
383          * to output tags of trees, tags of tags of trees, etc.  Simply omit
384          * such tags.
385          */
386         tagged = tag->tagged;
387         while (tagged->type == OBJ_TAG) {
388                 tagged = ((struct tag *)tagged)->tagged;
389         }
390         if (tagged->type == OBJ_TREE) {
391                 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
392                         sha1_to_hex(tag->object.sha1));
393                 return;
394         }
395
396         buf = read_sha1_file(tag->object.sha1, &type, &size);
397         if (!buf)
398                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
399         message = memmem(buf, size, "\n\n", 2);
400         if (message) {
401                 message += 2;
402                 message_size = strlen(message);
403         }
404         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
405         if (!tagger) {
406                 if (fake_missing_tagger)
407                         tagger = "tagger Unspecified Tagger "
408                                 "<unspecified-tagger> 0 +0000";
409                 else
410                         tagger = "";
411                 tagger_end = tagger + strlen(tagger);
412         } else {
413                 tagger++;
414                 tagger_end = strchrnul(tagger, '\n');
415         }
416
417         /* handle signed tags */
418         if (message) {
419                 const char *signature = strstr(message,
420                                                "\n-----BEGIN PGP SIGNATURE-----\n");
421                 if (signature)
422                         switch(signed_tag_mode) {
423                         case ABORT:
424                                 die ("Encountered signed tag %s; use "
425                                      "--signed-tags=<mode> to handle it.",
426                                      sha1_to_hex(tag->object.sha1));
427                         case WARN:
428                                 warning ("Exporting signed tag %s",
429                                          sha1_to_hex(tag->object.sha1));
430                                 /* fallthru */
431                         case VERBATIM:
432                                 break;
433                         case WARN_STRIP:
434                                 warning ("Stripping signature from tag %s",
435                                          sha1_to_hex(tag->object.sha1));
436                                 /* fallthru */
437                         case STRIP:
438                                 message_size = signature + 1 - message;
439                                 break;
440                         }
441         }
442
443         /* handle tag->tagged having been filtered out due to paths specified */
444         tagged = tag->tagged;
445         tagged_mark = get_object_mark(tagged);
446         if (!tagged_mark) {
447                 switch(tag_of_filtered_mode) {
448                 case ABORT:
449                         die ("Tag %s tags unexported object; use "
450                              "--tag-of-filtered-object=<mode> to handle it.",
451                              sha1_to_hex(tag->object.sha1));
452                 case DROP:
453                         /* Ignore this tag altogether */
454                         return;
455                 case REWRITE:
456                         if (tagged->type != OBJ_COMMIT) {
457                                 die ("Tag %s tags unexported %s!",
458                                      sha1_to_hex(tag->object.sha1),
459                                      typename(tagged->type));
460                         }
461                         p = (struct commit *)tagged;
462                         for (;;) {
463                                 if (p->parents && p->parents->next)
464                                         break;
465                                 if (p->object.flags & UNINTERESTING)
466                                         break;
467                                 if (!(p->object.flags & TREESAME))
468                                         break;
469                                 if (!p->parents)
470                                         die ("Can't find replacement commit for tag %s\n",
471                                              sha1_to_hex(tag->object.sha1));
472                                 p = p->parents->item;
473                         }
474                         tagged_mark = get_object_mark(&p->object);
475                 }
476         }
477
478         if (!prefixcmp(name, "refs/tags/"))
479                 name += 10;
480         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
481                name, tagged_mark,
482                (int)(tagger_end - tagger), tagger,
483                tagger == tagger_end ? "" : "\n",
484                (int)message_size, (int)message_size, message ? message : "");
485 }
486
487 static void get_tags_and_duplicates(struct rev_cmdline_info *info,
488                                     struct string_list *extra_refs)
489 {
490         struct tag *tag;
491         int i;
492
493         for (i = 0; i < info->nr; i++) {
494                 struct rev_cmdline_entry *e = info->rev + i;
495                 unsigned char sha1[20];
496                 struct commit *commit;
497                 char *full_name;
498
499                 if (e->flags & UNINTERESTING)
500                         continue;
501
502                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
503                         continue;
504
505                 switch (e->item->type) {
506                 case OBJ_COMMIT:
507                         commit = (struct commit *)e->item;
508                         break;
509                 case OBJ_TAG:
510                         tag = (struct tag *)e->item;
511
512                         /* handle nested tags */
513                         while (tag && tag->object.type == OBJ_TAG) {
514                                 parse_object(tag->object.sha1);
515                                 string_list_append(extra_refs, full_name)->util = tag;
516                                 tag = (struct tag *)tag->tagged;
517                         }
518                         if (!tag)
519                                 die ("Tag %s points nowhere?", e->name);
520                         switch(tag->object.type) {
521                         case OBJ_COMMIT:
522                                 commit = (struct commit *)tag;
523                                 break;
524                         case OBJ_BLOB:
525                                 export_blob(tag->object.sha1);
526                                 continue;
527                         default: /* OBJ_TAG (nested tags) is already handled */
528                                 warning("Tag points to object of unexpected type %s, skipping.",
529                                         typename(tag->object.type));
530                                 continue;
531                         }
532                         break;
533                 default:
534                         warning("%s: Unexpected object of type %s, skipping.",
535                                 e->name,
536                                 typename(e->item->type));
537                         continue;
538                 }
539
540                 /*
541                  * This ref will not be updated through a commit, lets make
542                  * sure it gets properly updated eventually.
543                  */
544                 if (commit->util || commit->object.flags & SHOWN)
545                         string_list_append(extra_refs, full_name)->util = commit;
546                 if (!commit->util)
547                         commit->util = full_name;
548         }
549 }
550
551 static void handle_tags_and_duplicates(struct string_list *extra_refs)
552 {
553         struct commit *commit;
554         int i;
555
556         for (i = extra_refs->nr - 1; i >= 0; i--) {
557                 const char *name = extra_refs->items[i].string;
558                 struct object *object = extra_refs->items[i].util;
559                 switch (object->type) {
560                 case OBJ_TAG:
561                         handle_tag(name, (struct tag *)object);
562                         break;
563                 case OBJ_COMMIT:
564                         /* create refs pointing to already seen commits */
565                         commit = (struct commit *)object;
566                         printf("reset %s\nfrom :%d\n\n", name,
567                                get_object_mark(&commit->object));
568                         show_progress();
569                         break;
570                 }
571         }
572 }
573
574 static void export_marks(char *file)
575 {
576         unsigned int i;
577         uint32_t mark;
578         struct object_decoration *deco = idnums.hash;
579         FILE *f;
580         int e = 0;
581
582         f = fopen(file, "w");
583         if (!f)
584                 die_errno("Unable to open marks file %s for writing.", file);
585
586         for (i = 0; i < idnums.size; i++) {
587                 if (deco->base && deco->base->type == 1) {
588                         mark = ptr_to_mark(deco->decoration);
589                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
590                                 sha1_to_hex(deco->base->sha1)) < 0) {
591                             e = 1;
592                             break;
593                         }
594                 }
595                 deco++;
596         }
597
598         e |= ferror(f);
599         e |= fclose(f);
600         if (e)
601                 error("Unable to write marks file %s.", file);
602 }
603
604 static void import_marks(char *input_file)
605 {
606         char line[512];
607         FILE *f = fopen(input_file, "r");
608         if (!f)
609                 die_errno("cannot read '%s'", input_file);
610
611         while (fgets(line, sizeof(line), f)) {
612                 uint32_t mark;
613                 char *line_end, *mark_end;
614                 unsigned char sha1[20];
615                 struct object *object;
616
617                 line_end = strchr(line, '\n');
618                 if (line[0] != ':' || !line_end)
619                         die("corrupt mark line: %s", line);
620                 *line_end = '\0';
621
622                 mark = strtoumax(line + 1, &mark_end, 10);
623                 if (!mark || mark_end == line + 1
624                         || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
625                         die("corrupt mark line: %s", line);
626
627                 if (last_idnum < mark)
628                         last_idnum = mark;
629
630                 object = parse_object(sha1);
631                 if (!object)
632                         continue;
633
634                 if (object->flags & SHOWN)
635                         error("Object %s already has a mark", sha1_to_hex(sha1));
636
637                 if (object->type != OBJ_COMMIT)
638                         /* only commits */
639                         continue;
640
641                 mark_object(object, mark);
642
643                 object->flags |= SHOWN;
644         }
645         fclose(f);
646 }
647
648 int cmd_fast_export(int argc, const char **argv, const char *prefix)
649 {
650         struct rev_info revs;
651         struct object_array commits = OBJECT_ARRAY_INIT;
652         struct string_list extra_refs = STRING_LIST_INIT_NODUP;
653         struct commit *commit;
654         char *export_filename = NULL, *import_filename = NULL;
655         uint32_t lastimportid;
656         struct option options[] = {
657                 OPT_INTEGER(0, "progress", &progress,
658                             N_("show progress after <n> objects")),
659                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, N_("mode"),
660                              N_("select handling of signed tags"),
661                              parse_opt_signed_tag_mode),
662                 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, N_("mode"),
663                              N_("select handling of tags that tag filtered objects"),
664                              parse_opt_tag_of_filtered_mode),
665                 OPT_STRING(0, "export-marks", &export_filename, N_("file"),
666                              N_("Dump marks to this file")),
667                 OPT_STRING(0, "import-marks", &import_filename, N_("file"),
668                              N_("Import marks from this file")),
669                 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
670                              N_("Fake a tagger when tags lack one")),
671                 OPT_BOOLEAN(0, "full-tree", &full_tree,
672                              N_("Output full tree for each commit")),
673                 OPT_BOOLEAN(0, "use-done-feature", &use_done_feature,
674                              N_("Use the done feature to terminate the stream")),
675                 OPT_BOOL(0, "no-data", &no_data, N_("Skip output of blob data")),
676                 OPT_END()
677         };
678
679         if (argc == 1)
680                 usage_with_options (fast_export_usage, options);
681
682         /* we handle encodings */
683         git_config(git_default_config, NULL);
684
685         init_revisions(&revs, prefix);
686         revs.topo_order = 1;
687         revs.show_source = 1;
688         revs.rewrite_parents = 1;
689         argc = setup_revisions(argc, argv, &revs, NULL);
690         argc = parse_options(argc, argv, prefix, options, fast_export_usage, 0);
691         if (argc > 1)
692                 usage_with_options (fast_export_usage, options);
693
694         if (use_done_feature)
695                 printf("feature done\n");
696
697         if (import_filename)
698                 import_marks(import_filename);
699         lastimportid = last_idnum;
700
701         if (import_filename && revs.prune_data.nr)
702                 full_tree = 1;
703
704         get_tags_and_duplicates(&revs.cmdline, &extra_refs);
705
706         if (prepare_revision_walk(&revs))
707                 die("revision walk setup failed");
708         revs.diffopt.format_callback = show_filemodify;
709         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
710         while ((commit = get_revision(&revs))) {
711                 if (has_unshown_parent(commit)) {
712                         add_object_array(&commit->object, NULL, &commits);
713                 }
714                 else {
715                         handle_commit(commit, &revs);
716                         handle_tail(&commits, &revs);
717                 }
718         }
719
720         handle_tags_and_duplicates(&extra_refs);
721
722         if (export_filename && lastimportid != last_idnum)
723                 export_marks(export_filename);
724
725         if (use_done_feature)
726                 printf("done\n");
727
728         return 0;
729 }