Merge branch 'db/doc-custom-xmlto' 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         "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 if (strchr(path, ' '))
189                 printf("\"%s\"", path);
190         else
191                 printf("%s", path);
192 }
193
194 static void show_filemodify(struct diff_queue_struct *q,
195                             struct diff_options *options, void *data)
196 {
197         int i;
198
199         /*
200          * Handle files below a directory first, in case they are all deleted
201          * and the directory changes to a file or symlink.
202          */
203         qsort(q->queue, q->nr, sizeof(q->queue[0]), depth_first);
204
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;
208
209                 switch (q->queue[i]->status) {
210                 case DIFF_STATUS_DELETED:
211                         printf("D ");
212                         print_path(spec->path);
213                         putchar('\n');
214                         break;
215
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');
223
224                         if (!hashcmp(ospec->sha1, spec->sha1) &&
225                             ospec->mode == spec->mode)
226                                 break;
227                         /* fallthrough */
228
229                 case DIFF_STATUS_TYPE_CHANGED:
230                 case DIFF_STATUS_MODIFIED:
231                 case DIFF_STATUS_ADDED:
232                         /*
233                          * Links refer to objects in another repositories;
234                          * output the SHA-1 verbatim.
235                          */
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));
243                         }
244                         print_path(spec->path);
245                         putchar('\n');
246                         break;
247
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");
253                 }
254         }
255 }
256
257 static const char *find_encoding(const char *begin, const char *end)
258 {
259         const char *needle = "\nencoding ";
260         char *bol, *eol;
261
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;
270 }
271
272 static void handle_commit(struct commit *commit, struct rev_info *rev)
273 {
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;
280
281         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
282
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;
300
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);
307         }
308         else
309                 diff_root_tree_sha1(commit->tree->object.sha1,
310                                     "", &rev->diffopt);
311
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);
316
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);
331
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++;
341         }
342
343         if (full_tree)
344                 printf("deleteall\n");
345         log_tree_diff_flush(rev);
346         rev->diffopt.output_format = saved_output_format;
347
348         printf("\n");
349
350         show_progress();
351 }
352
353 static void handle_tail(struct object_array *commits, struct rev_info *revs)
354 {
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--;
362         }
363 }
364
365 static void handle_tag(const char *name, struct tag *tag)
366 {
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;
375
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.
379          */
380         tagged = tag->tagged;
381         while (tagged->type == OBJ_TAG) {
382                 tagged = ((struct tag *)tagged)->tagged;
383         }
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;
388         }
389
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);
397         }
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');
409         }
410
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;
430                         }
431         }
432
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));
450                         }
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;
463                         }
464                         tagged_mark = get_object_mark(&p->object);
465                 }
466         }
467
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 : "");
475 }
476
477 static void get_tags_and_duplicates(struct object_array *pending,
478                                     struct string_list *extra_refs)
479 {
480         struct tag *tag;
481         int i;
482
483         for (i = 0; i < pending->nr; i++) {
484                 struct object_array_entry *e = pending->objects + i;
485                 unsigned char sha1[20];
486                 struct commit *commit = commit;
487                 char *full_name;
488
489                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
490                         continue;
491
492                 switch (e->item->type) {
493                 case OBJ_COMMIT:
494                         commit = (struct commit *)e->item;
495                         break;
496                 case OBJ_TAG:
497                         tag = (struct tag *)e->item;
498
499                         /* handle nested tags */
500                         while (tag && tag->object.type == OBJ_TAG) {
501                                 parse_object(tag->object.sha1);
502                                 string_list_append(extra_refs, full_name)->util = tag;
503                                 tag = (struct tag *)tag->tagged;
504                         }
505                         if (!tag)
506                                 die ("Tag %s points nowhere?", e->name);
507                         switch(tag->object.type) {
508                         case OBJ_COMMIT:
509                                 commit = (struct commit *)tag;
510                                 break;
511                         case OBJ_BLOB:
512                                 handle_object(tag->object.sha1);
513                                 continue;
514                         default: /* OBJ_TAG (nested tags) is already handled */
515                                 warning("Tag points to object of unexpected type %s, skipping.",
516                                         typename(tag->object.type));
517                                 continue;
518                         }
519                         break;
520                 default:
521                         warning("%s: Unexpected object of type %s, skipping.",
522                                 e->name,
523                                 typename(e->item->type));
524                         continue;
525                 }
526                 if (commit->util)
527                         /* more than one name for the same object */
528                         string_list_append(extra_refs, full_name)->util = commit;
529                 else
530                         commit->util = full_name;
531         }
532 }
533
534 static void handle_tags_and_duplicates(struct string_list *extra_refs)
535 {
536         struct commit *commit;
537         int i;
538
539         for (i = extra_refs->nr - 1; i >= 0; i--) {
540                 const char *name = extra_refs->items[i].string;
541                 struct object *object = extra_refs->items[i].util;
542                 switch (object->type) {
543                 case OBJ_TAG:
544                         handle_tag(name, (struct tag *)object);
545                         break;
546                 case OBJ_COMMIT:
547                         /* create refs pointing to already seen commits */
548                         commit = (struct commit *)object;
549                         printf("reset %s\nfrom :%d\n\n", name,
550                                get_object_mark(&commit->object));
551                         show_progress();
552                         break;
553                 }
554         }
555 }
556
557 static void export_marks(char *file)
558 {
559         unsigned int i;
560         uint32_t mark;
561         struct object_decoration *deco = idnums.hash;
562         FILE *f;
563         int e = 0;
564
565         f = fopen(file, "w");
566         if (!f)
567                 die_errno("Unable to open marks file %s for writing.", file);
568
569         for (i = 0; i < idnums.size; i++) {
570                 if (deco->base && deco->base->type == 1) {
571                         mark = ptr_to_mark(deco->decoration);
572                         if (fprintf(f, ":%"PRIu32" %s\n", mark,
573                                 sha1_to_hex(deco->base->sha1)) < 0) {
574                             e = 1;
575                             break;
576                         }
577                 }
578                 deco++;
579         }
580
581         e |= ferror(f);
582         e |= fclose(f);
583         if (e)
584                 error("Unable to write marks file %s.", file);
585 }
586
587 static void import_marks(char *input_file)
588 {
589         char line[512];
590         FILE *f = fopen(input_file, "r");
591         if (!f)
592                 die_errno("cannot read '%s'", input_file);
593
594         while (fgets(line, sizeof(line), f)) {
595                 uint32_t mark;
596                 char *line_end, *mark_end;
597                 unsigned char sha1[20];
598                 struct object *object;
599
600                 line_end = strchr(line, '\n');
601                 if (line[0] != ':' || !line_end)
602                         die("corrupt mark line: %s", line);
603                 *line_end = '\0';
604
605                 mark = strtoumax(line + 1, &mark_end, 10);
606                 if (!mark || mark_end == line + 1
607                         || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
608                         die("corrupt mark line: %s", line);
609
610                 object = parse_object(sha1);
611                 if (!object)
612                         die ("Could not read blob %s", sha1_to_hex(sha1));
613
614                 if (object->flags & SHOWN)
615                         error("Object %s already has a mark", sha1_to_hex(sha1));
616
617                 mark_object(object, mark);
618                 if (last_idnum < mark)
619                         last_idnum = mark;
620
621                 object->flags |= SHOWN;
622         }
623         fclose(f);
624 }
625
626 int cmd_fast_export(int argc, const char **argv, const char *prefix)
627 {
628         struct rev_info revs;
629         struct object_array commits = OBJECT_ARRAY_INIT;
630         struct string_list extra_refs = STRING_LIST_INIT_NODUP;
631         struct commit *commit;
632         char *export_filename = NULL, *import_filename = NULL;
633         struct option options[] = {
634                 OPT_INTEGER(0, "progress", &progress,
635                             "show progress after <n> objects"),
636                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
637                              "select handling of signed tags",
638                              parse_opt_signed_tag_mode),
639                 OPT_CALLBACK(0, "tag-of-filtered-object", &tag_of_filtered_mode, "mode",
640                              "select handling of tags that tag filtered objects",
641                              parse_opt_tag_of_filtered_mode),
642                 OPT_STRING(0, "export-marks", &export_filename, "file",
643                              "Dump marks to this file"),
644                 OPT_STRING(0, "import-marks", &import_filename, "file",
645                              "Import marks from this file"),
646                 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
647                              "Fake a tagger when tags lack one"),
648                 OPT_BOOLEAN(0, "full-tree", &full_tree,
649                              "Output full tree for each commit"),
650                 OPT_BOOLEAN(0, "use-done-feature", &use_done_feature,
651                              "Use the done feature to terminate the stream"),
652                 OPT_BOOL(0, "no-data", &no_data, "Skip output of blob data"),
653                 OPT_END()
654         };
655
656         if (argc == 1)
657                 usage_with_options (fast_export_usage, options);
658
659         /* we handle encodings */
660         git_config(git_default_config, NULL);
661
662         init_revisions(&revs, prefix);
663         revs.topo_order = 1;
664         revs.show_source = 1;
665         revs.rewrite_parents = 1;
666         argc = setup_revisions(argc, argv, &revs, NULL);
667         argc = parse_options(argc, argv, prefix, options, fast_export_usage, 0);
668         if (argc > 1)
669                 usage_with_options (fast_export_usage, options);
670
671         if (use_done_feature)
672                 printf("feature done\n");
673
674         if (import_filename)
675                 import_marks(import_filename);
676
677         if (import_filename && revs.prune_data.nr)
678                 full_tree = 1;
679
680         get_tags_and_duplicates(&revs.pending, &extra_refs);
681
682         if (prepare_revision_walk(&revs))
683                 die("revision walk setup failed");
684         revs.diffopt.format_callback = show_filemodify;
685         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
686         while ((commit = get_revision(&revs))) {
687                 if (has_unshown_parent(commit)) {
688                         add_object_array(&commit->object, NULL, &commits);
689                 }
690                 else {
691                         handle_commit(commit, &revs);
692                         handle_tail(&commits, &revs);
693                 }
694         }
695
696         handle_tags_and_duplicates(&extra_refs);
697
698         if (export_filename)
699                 export_marks(export_filename);
700
701         if (use_done_feature)
702                 printf("done\n");
703
704         return 0;
705 }