Merge branch 'jc/install-no-hardlinks'
[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 = 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, "strip"))
44                 signed_tag_mode = STRIP;
45         else
46                 return error("Unknown signed-tag mode: %s", arg);
47         return 0;
48 }
49
50 static int parse_opt_tag_of_filtered_mode(const struct option *opt,
51                                           const char *arg, int unset)
52 {
53         if (unset || !strcmp(arg, "abort"))
54                 tag_of_filtered_mode = ERROR;
55         else if (!strcmp(arg, "drop"))
56                 tag_of_filtered_mode = DROP;
57         else if (!strcmp(arg, "rewrite"))
58                 tag_of_filtered_mode = REWRITE;
59         else
60                 return error("Unknown tag-of-filtered mode: %s", arg);
61         return 0;
62 }
63
64 static struct decoration idnums;
65 static uint32_t last_idnum;
66
67 static int has_unshown_parent(struct commit *commit)
68 {
69         struct commit_list *parent;
70
71         for (parent = commit->parents; parent; parent = parent->next)
72                 if (!(parent->item->object.flags & SHOWN) &&
73                     !(parent->item->object.flags & UNINTERESTING))
74                         return 1;
75         return 0;
76 }
77
78 /* Since intptr_t is C99, we do not use it here */
79 static inline uint32_t *mark_to_ptr(uint32_t mark)
80 {
81         return ((uint32_t *)NULL) + mark;
82 }
83
84 static inline uint32_t ptr_to_mark(void * mark)
85 {
86         return (uint32_t *)mark - (uint32_t *)NULL;
87 }
88
89 static inline void mark_object(struct object *object, uint32_t mark)
90 {
91         add_decoration(&idnums, object, mark_to_ptr(mark));
92 }
93
94 static inline void mark_next_object(struct object *object)
95 {
96         mark_object(object, ++last_idnum);
97 }
98
99 static int get_object_mark(struct object *object)
100 {
101         void *decoration = lookup_decoration(&idnums, object);
102         if (!decoration)
103                 return 0;
104         return ptr_to_mark(decoration);
105 }
106
107 static void show_progress(void)
108 {
109         static int counter = 0;
110         if (!progress)
111                 return;
112         if ((++counter % progress) == 0)
113                 printf("progress %d objects\n", counter);
114 }
115
116 static void handle_object(const unsigned char *sha1)
117 {
118         unsigned long size;
119         enum object_type type;
120         char *buf;
121         struct object *object;
122
123         if (no_data)
124                 return;
125
126         if (is_null_sha1(sha1))
127                 return;
128
129         object = parse_object(sha1);
130         if (!object)
131                 die ("Could not read blob %s", sha1_to_hex(sha1));
132
133         if (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
140         mark_next_object(object);
141
142         printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
143         if (size && fwrite(buf, size, 1, stdout) != 1)
144                 die_errno ("Could not write blob '%s'", sha1_to_hex(sha1));
145         printf("\n");
146
147         show_progress();
148
149         object->flags |= SHOWN;
150         free(buf);
151 }
152
153 static int depth_first(const void *a_, const void *b_)
154 {
155         const struct diff_filepair *a = *((const struct diff_filepair **)a_);
156         const struct diff_filepair *b = *((const struct diff_filepair **)b_);
157         const char *name_a, *name_b;
158         int len_a, len_b, len;
159         int cmp;
160
161         name_a = a->one ? a->one->path : a->two->path;
162         name_b = b->one ? b->one->path : b->two->path;
163
164         len_a = strlen(name_a);
165         len_b = strlen(name_b);
166         len = (len_a < len_b) ? len_a : len_b;
167
168         /* strcmp will sort 'd' before 'd/e', we want 'd/e' before 'd' */
169         cmp = memcmp(name_a, name_b, len);
170         if (cmp)
171                 return cmp;
172         cmp = len_b - len_a;
173         if (cmp)
174                 return cmp;
175         /*
176          * Move 'R'ename entries last so that all references of the file
177          * appear in the output before it is renamed (e.g., when a file
178          * was copied and renamed in the same commit).
179          */
180         return (a->status == 'R') - (b->status == 'R');
181 }
182
183 static void print_path(const char *path)
184 {
185         int need_quote = quote_c_style(path, NULL, NULL, 0);
186         if (need_quote)
187                 quote_c_style(path, NULL, stdout, 0);
188         else
189                 printf("%s", path);
190 }
191
192 static void show_filemodify(struct diff_queue_struct *q,
193                             struct diff_options *options, void *data)
194 {
195         int i;
196
197         /*
198          * Handle files below a directory first, in case they are all deleted
199          * and the directory changes to a file or symlink.
200          */
201         qsort(q->queue, q->nr, sizeof(q->queue[0]), depth_first);
202
203         for (i = 0; i < q->nr; i++) {
204                 struct diff_filespec *ospec = q->queue[i]->one;
205                 struct diff_filespec *spec = q->queue[i]->two;
206
207                 switch (q->queue[i]->status) {
208                 case DIFF_STATUS_DELETED:
209                         printf("D ");
210                         print_path(spec->path);
211                         putchar('\n');
212                         break;
213
214                 case DIFF_STATUS_COPIED:
215                 case DIFF_STATUS_RENAMED:
216                         printf("%c ", q->queue[i]->status);
217                         print_path(ospec->path);
218                         putchar(' ');
219                         print_path(spec->path);
220                         putchar('\n');
221
222                         if (!hashcmp(ospec->sha1, spec->sha1) &&
223                             ospec->mode == spec->mode)
224                                 break;
225                         /* fallthrough */
226
227                 case DIFF_STATUS_TYPE_CHANGED:
228                 case DIFF_STATUS_MODIFIED:
229                 case DIFF_STATUS_ADDED:
230                         /*
231                          * Links refer to objects in another repositories;
232                          * output the SHA-1 verbatim.
233                          */
234                         if (no_data || S_ISGITLINK(spec->mode))
235                                 printf("M %06o %s ", spec->mode,
236                                        sha1_to_hex(spec->sha1));
237                         else {
238                                 struct object *object = lookup_object(spec->sha1);
239                                 printf("M %06o :%d ", spec->mode,
240                                        get_object_mark(object));
241                         }
242                         print_path(spec->path);
243                         putchar('\n');
244                         break;
245
246                 default:
247                         die("Unexpected comparison status '%c' for %s, %s",
248                                 q->queue[i]->status,
249                                 ospec->path ? ospec->path : "none",
250                                 spec->path ? spec->path : "none");
251                 }
252         }
253 }
254
255 static const char *find_encoding(const char *begin, const char *end)
256 {
257         const char *needle = "\nencoding ";
258         char *bol, *eol;
259
260         bol = memmem(begin, end ? end - begin : strlen(begin),
261                      needle, strlen(needle));
262         if (!bol)
263                 return git_commit_encoding;
264         bol += strlen(needle);
265         eol = strchrnul(bol, '\n');
266         *eol = '\0';
267         return bol;
268 }
269
270 static void handle_commit(struct commit *commit, struct rev_info *rev)
271 {
272         int saved_output_format = rev->diffopt.output_format;
273         const char *author, *author_end, *committer, *committer_end;
274         const char *encoding, *message;
275         char *reencoded = NULL;
276         struct commit_list *p;
277         int i;
278
279         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
280
281         parse_commit(commit);
282         author = strstr(commit->buffer, "\nauthor ");
283         if (!author)
284                 die ("Could not find author in commit %s",
285                      sha1_to_hex(commit->object.sha1));
286         author++;
287         author_end = strchrnul(author, '\n');
288         committer = strstr(author_end, "\ncommitter ");
289         if (!committer)
290                 die ("Could not find committer in commit %s",
291                      sha1_to_hex(commit->object.sha1));
292         committer++;
293         committer_end = strchrnul(committer, '\n');
294         message = strstr(committer_end, "\n\n");
295         encoding = find_encoding(committer_end, message);
296         if (message)
297                 message += 2;
298
299         if (commit->parents &&
300             get_object_mark(&commit->parents->item->object) != 0 &&
301             !full_tree) {
302                 parse_commit(commit->parents->item);
303                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
304                                commit->tree->object.sha1, "", &rev->diffopt);
305         }
306         else
307                 diff_root_tree_sha1(commit->tree->object.sha1,
308                                     "", &rev->diffopt);
309
310         /* Export the referenced blobs, and remember the marks. */
311         for (i = 0; i < diff_queued_diff.nr; i++)
312                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
313                         handle_object(diff_queued_diff.queue[i]->two->sha1);
314
315         mark_next_object(&commit->object);
316         if (!is_encoding_utf8(encoding))
317                 reencoded = reencode_string(message, "UTF-8", encoding);
318         if (!commit->parents)
319                 printf("reset %s\n", (const char*)commit->util);
320         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
321                (const char *)commit->util, last_idnum,
322                (int)(author_end - author), author,
323                (int)(committer_end - committer), committer,
324                (unsigned)(reencoded
325                           ? strlen(reencoded) : message
326                           ? strlen(message) : 0),
327                reencoded ? reencoded : message ? message : "");
328         free(reencoded);
329
330         for (i = 0, p = commit->parents; p; p = p->next) {
331                 int mark = get_object_mark(&p->item->object);
332                 if (!mark)
333                         continue;
334                 if (i == 0)
335                         printf("from :%d\n", mark);
336                 else
337                         printf("merge :%d\n", mark);
338                 i++;
339         }
340
341         if (full_tree)
342                 printf("deleteall\n");
343         log_tree_diff_flush(rev);
344         rev->diffopt.output_format = saved_output_format;
345
346         printf("\n");
347
348         show_progress();
349 }
350
351 static void handle_tail(struct object_array *commits, struct rev_info *revs)
352 {
353         struct commit *commit;
354         while (commits->nr) {
355                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
356                 if (has_unshown_parent(commit))
357                         return;
358                 handle_commit(commit, revs);
359                 commits->nr--;
360         }
361 }
362
363 static void handle_tag(const char *name, struct tag *tag)
364 {
365         unsigned long size;
366         enum object_type type;
367         char *buf;
368         const char *tagger, *tagger_end, *message;
369         size_t message_size = 0;
370         struct object *tagged;
371         int tagged_mark;
372         struct commit *p;
373
374         /* Trees have no identifer in fast-export output, thus we have no way
375          * to output tags of trees, tags of tags of trees, etc.  Simply omit
376          * such tags.
377          */
378         tagged = tag->tagged;
379         while (tagged->type == OBJ_TAG) {
380                 tagged = ((struct tag *)tagged)->tagged;
381         }
382         if (tagged->type == OBJ_TREE) {
383                 warning("Omitting tag %s,\nsince tags of trees (or tags of tags of trees, etc.) are not supported.",
384                         sha1_to_hex(tag->object.sha1));
385                 return;
386         }
387
388         buf = read_sha1_file(tag->object.sha1, &type, &size);
389         if (!buf)
390                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
391         message = memmem(buf, size, "\n\n", 2);
392         if (message) {
393                 message += 2;
394                 message_size = strlen(message);
395         }
396         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
397         if (!tagger) {
398                 if (fake_missing_tagger)
399                         tagger = "tagger Unspecified Tagger "
400                                 "<unspecified-tagger> 0 +0000";
401                 else
402                         tagger = "";
403                 tagger_end = tagger + strlen(tagger);
404         } else {
405                 tagger++;
406                 tagger_end = strchrnul(tagger, '\n');
407         }
408
409         /* handle signed tags */
410         if (message) {
411                 const char *signature = strstr(message,
412                                                "\n-----BEGIN PGP SIGNATURE-----\n");
413                 if (signature)
414                         switch(signed_tag_mode) {
415                         case ABORT:
416                                 die ("Encountered signed tag %s; use "
417                                      "--signed-tag=<mode> to handle it.",
418                                      sha1_to_hex(tag->object.sha1));
419                         case WARN:
420                                 warning ("Exporting signed tag %s",
421                                          sha1_to_hex(tag->object.sha1));
422                                 /* fallthru */
423                         case VERBATIM:
424                                 break;
425                         case STRIP:
426                                 message_size = signature + 1 - message;
427                                 break;
428                         }
429         }
430
431         /* handle tag->tagged having been filtered out due to paths specified */
432         tagged = tag->tagged;
433         tagged_mark = get_object_mark(tagged);
434         if (!tagged_mark) {
435                 switch(tag_of_filtered_mode) {
436                 case ABORT:
437                         die ("Tag %s tags unexported object; use "
438                              "--tag-of-filtered-object=<mode> to handle it.",
439                              sha1_to_hex(tag->object.sha1));
440                 case DROP:
441                         /* Ignore this tag altogether */
442                         return;
443                 case REWRITE:
444                         if (tagged->type != OBJ_COMMIT) {
445                                 die ("Tag %s tags unexported %s!",
446                                      sha1_to_hex(tag->object.sha1),
447                                      typename(tagged->type));
448                         }
449                         p = (struct commit *)tagged;
450                         for (;;) {
451                                 if (p->parents && p->parents->next)
452                                         break;
453                                 if (p->object.flags & UNINTERESTING)
454                                         break;
455                                 if (!(p->object.flags & TREESAME))
456                                         break;
457                                 if (!p->parents)
458                                         die ("Can't find replacement commit for tag %s\n",
459                                              sha1_to_hex(tag->object.sha1));
460                                 p = p->parents->item;
461                         }
462                         tagged_mark = get_object_mark(&p->object);
463                 }
464         }
465
466         if (!prefixcmp(name, "refs/tags/"))
467                 name += 10;
468         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
469                name, tagged_mark,
470                (int)(tagger_end - tagger), tagger,
471                tagger == tagger_end ? "" : "\n",
472                (int)message_size, (int)message_size, message ? message : "");
473 }
474
475 static void get_tags_and_duplicates(struct object_array *pending,
476                                     struct string_list *extra_refs)
477 {
478         struct tag *tag;
479         int i;
480
481         for (i = 0; i < pending->nr; i++) {
482                 struct object_array_entry *e = pending->objects + i;
483                 unsigned char sha1[20];
484                 struct commit *commit = commit;
485                 char *full_name;
486
487                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
488                         continue;
489
490                 switch (e->item->type) {
491                 case OBJ_COMMIT:
492                         commit = (struct commit *)e->item;
493                         break;
494                 case OBJ_TAG:
495                         tag = (struct tag *)e->item;
496
497                         /* handle nested tags */
498                         while (tag && tag->object.type == OBJ_TAG) {
499                                 parse_object(tag->object.sha1);
500                                 string_list_append(extra_refs, full_name)->util = tag;
501                                 tag = (struct tag *)tag->tagged;
502                         }
503                         if (!tag)
504                                 die ("Tag %s points nowhere?", e->name);
505                         switch(tag->object.type) {
506                         case OBJ_COMMIT:
507                                 commit = (struct commit *)tag;
508                                 break;
509                         case OBJ_BLOB:
510                                 handle_object(tag->object.sha1);
511                                 continue;
512                         default: /* OBJ_TAG (nested tags) is already handled */
513                                 warning("Tag points to object of unexpected type %s, skipping.",
514                                         typename(tag->object.type));
515                                 continue;
516                         }
517                         break;
518                 default:
519                         warning("%s: Unexpected object of type %s, skipping.",
520                                 e->name,
521                                 typename(e->item->type));
522                         continue;
523                 }
524                 if (commit->util)
525                         /* more than one name for the same object */
526                         string_list_append(extra_refs, full_name)->util = commit;
527                 else
528                         commit->util = full_name;
529         }
530 }
531
532 static void handle_tags_and_duplicates(struct string_list *extra_refs)
533 {
534         struct commit *commit;
535         int i;
536
537         for (i = extra_refs->nr - 1; i >= 0; i--) {
538                 const char *name = extra_refs->items[i].string;
539                 struct object *object = extra_refs->items[i].util;
540                 switch (object->type) {
541                 case OBJ_TAG:
542                         handle_tag(name, (struct tag *)object);
543                         break;
544                 case OBJ_COMMIT:
545                         /* create refs pointing to already seen commits */
546                         commit = (struct commit *)object;
547                         printf("reset %s\nfrom :%d\n\n", name,
548                                get_object_mark(&commit->object));
549                         show_progress();
550                         break;
551                 }
552         }
553 }
554
555 static void export_marks(char *file)
556 {
557         unsigned int i;
558         uint32_t mark;
559         struct object_decoration *deco = idnums.hash;
560         FILE *f;
561         int e = 0;
562
563         f = fopen(file, "w");
564         if (!f)
565                 die_errno("Unable to open marks file %s for writing.", file);
566
567         for (i = 0; i < idnums.size; i++) {
568                 if (deco->base && deco->base->type == 1) {
569                         mark = ptr_to_mark(deco->decoration);
570                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
571                                 sha1_to_hex(deco->base->sha1)) < 0) {
572                             e = 1;
573                             break;
574                         }
575                 }
576                 deco++;
577         }
578
579         e |= ferror(f);
580         e |= fclose(f);
581         if (e)
582                 error("Unable to write marks file %s.", file);
583 }
584
585 static void import_marks(char *input_file)
586 {
587         char line[512];
588         FILE *f = fopen(input_file, "r");
589         if (!f)
590                 die_errno("cannot read '%s'", input_file);
591
592         while (fgets(line, sizeof(line), f)) {
593                 uint32_t mark;
594                 char *line_end, *mark_end;
595                 unsigned char sha1[20];
596                 struct object *object;
597
598                 line_end = strchr(line, '\n');
599                 if (line[0] != ':' || !line_end)
600                         die("corrupt mark line: %s", line);
601                 *line_end = '\0';
602
603                 mark = strtoumax(line + 1, &mark_end, 10);
604                 if (!mark || mark_end == line + 1
605                         || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
606                         die("corrupt mark line: %s", line);
607
608                 object = parse_object(sha1);
609                 if (!object)
610                         die ("Could not read blob %s", sha1_to_hex(sha1));
611
612                 if (object->flags & SHOWN)
613                         error("Object %s already has a mark", sha1);
614
615                 mark_object(object, mark);
616                 if (last_idnum < mark)
617                         last_idnum = mark;
618
619                 object->flags |= SHOWN;
620         }
621         fclose(f);
622 }
623
624 int cmd_fast_export(int argc, const char **argv, const char *prefix)
625 {
626         struct rev_info revs;
627         struct object_array commits = OBJECT_ARRAY_INIT;
628         struct string_list extra_refs = STRING_LIST_INIT_NODUP;
629         struct commit *commit;
630         char *export_filename = NULL, *import_filename = NULL;
631         struct option options[] = {
632                 OPT_INTEGER(0, "progress", &progress,
633                             "show progress after <n> objects"),
634                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
635                              "select handling of signed tags",
636                              parse_opt_signed_tag_mode),
637                 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, "mode",
638                              "select handling of tags that tag filtered objects",
639                              parse_opt_tag_of_filtered_mode),
640                 OPT_STRING(0, "export-marks", &export_filename, "file",
641                              "Dump marks to this file"),
642                 OPT_STRING(0, "import-marks", &import_filename, "file",
643                              "Import marks from this file"),
644                 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
645                              "Fake a tagger when tags lack one"),
646                 OPT_BOOLEAN(0, "full-tree", &full_tree,
647                              "Output full tree for each commit"),
648                 OPT_BOOLEAN(0, "use-done-feature", &use_done_feature,
649                              "Use the done feature to terminate the stream"),
650                 OPT_BOOL(0, "no-data", &no_data, "Skip output of blob data"),
651                 OPT_END()
652         };
653
654         if (argc == 1)
655                 usage_with_options (fast_export_usage, options);
656
657         /* we handle encodings */
658         git_config(git_default_config, NULL);
659
660         init_revisions(&revs, prefix);
661         revs.topo_order = 1;
662         revs.show_source = 1;
663         revs.rewrite_parents = 1;
664         argc = setup_revisions(argc, argv, &revs, NULL);
665         argc = parse_options(argc, argv, prefix, options, fast_export_usage, 0);
666         if (argc > 1)
667                 usage_with_options (fast_export_usage, options);
668
669         if (use_done_feature)
670                 printf("feature done\n");
671
672         if (import_filename)
673                 import_marks(import_filename);
674
675         if (import_filename && revs.prune_data.nr)
676                 full_tree = 1;
677
678         get_tags_and_duplicates(&revs.pending, &extra_refs);
679
680         if (prepare_revision_walk(&revs))
681                 die("revision walk setup failed");
682         revs.diffopt.format_callback = show_filemodify;
683         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
684         while ((commit = get_revision(&revs))) {
685                 if (has_unshown_parent(commit)) {
686                         add_object_array(&commit->object, NULL, &commits);
687                 }
688                 else {
689                         handle_commit(commit, &revs);
690                         handle_tail(&commits, &revs);
691                 }
692         }
693
694         handle_tags_and_duplicates(&extra_refs);
695
696         if (export_filename)
697                 export_marks(export_filename);
698
699         if (use_done_feature)
700                 printf("done\n");
701
702         return 0;
703 }