Merge branch '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
20 static const char *fast_export_usage[] = {
21         "git fast-export [rev-list-opts]",
22         NULL
23 };
24
25 static int progress;
26 static enum { VERBATIM, WARN, STRIP, ABORT } signed_tag_mode = ABORT;
27 static int fake_missing_tagger;
28
29 static int parse_opt_signed_tag_mode(const struct option *opt,
30                                      const char *arg, int unset)
31 {
32         if (unset || !strcmp(arg, "abort"))
33                 signed_tag_mode = ABORT;
34         else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
35                 signed_tag_mode = VERBATIM;
36         else if (!strcmp(arg, "warn"))
37                 signed_tag_mode = WARN;
38         else if (!strcmp(arg, "strip"))
39                 signed_tag_mode = STRIP;
40         else
41                 return error("Unknown signed-tag mode: %s", arg);
42         return 0;
43 }
44
45 static struct decoration idnums;
46 static uint32_t last_idnum;
47
48 static int has_unshown_parent(struct commit *commit)
49 {
50         struct commit_list *parent;
51
52         for (parent = commit->parents; parent; parent = parent->next)
53                 if (!(parent->item->object.flags & SHOWN) &&
54                     !(parent->item->object.flags & UNINTERESTING))
55                         return 1;
56         return 0;
57 }
58
59 /* Since intptr_t is C99, we do not use it here */
60 static inline uint32_t *mark_to_ptr(uint32_t mark)
61 {
62         return ((uint32_t *)NULL) + mark;
63 }
64
65 static inline uint32_t ptr_to_mark(void * mark)
66 {
67         return (uint32_t *)mark - (uint32_t *)NULL;
68 }
69
70 static inline void mark_object(struct object *object, uint32_t mark)
71 {
72         add_decoration(&idnums, object, mark_to_ptr(mark));
73 }
74
75 static inline void mark_next_object(struct object *object)
76 {
77         mark_object(object, ++last_idnum);
78 }
79
80 static int get_object_mark(struct object *object)
81 {
82         void *decoration = lookup_decoration(&idnums, object);
83         if (!decoration)
84                 return 0;
85         return ptr_to_mark(decoration);
86 }
87
88 static void show_progress(void)
89 {
90         static int counter = 0;
91         if (!progress)
92                 return;
93         if ((++counter % progress) == 0)
94                 printf("progress %d objects\n", counter);
95 }
96
97 static void handle_object(const unsigned char *sha1)
98 {
99         unsigned long size;
100         enum object_type type;
101         char *buf;
102         struct object *object;
103
104         if (is_null_sha1(sha1))
105                 return;
106
107         object = parse_object(sha1);
108         if (!object)
109                 die ("Could not read blob %s", sha1_to_hex(sha1));
110
111         if (object->flags & SHOWN)
112                 return;
113
114         buf = read_sha1_file(sha1, &type, &size);
115         if (!buf)
116                 die ("Could not read blob %s", sha1_to_hex(sha1));
117
118         mark_next_object(object);
119
120         printf("blob\nmark :%"PRIu32"\ndata %lu\n", last_idnum, size);
121         if (size && fwrite(buf, size, 1, stdout) != 1)
122                 die ("Could not write blob %s", sha1_to_hex(sha1));
123         printf("\n");
124
125         show_progress();
126
127         object->flags |= SHOWN;
128         free(buf);
129 }
130
131 static void show_filemodify(struct diff_queue_struct *q,
132                             struct diff_options *options, void *data)
133 {
134         int i;
135         for (i = 0; i < q->nr; i++) {
136                 struct diff_filespec *ospec = q->queue[i]->one;
137                 struct diff_filespec *spec = q->queue[i]->two;
138
139                 switch (q->queue[i]->status) {
140                 case DIFF_STATUS_DELETED:
141                         printf("D %s\n", spec->path);
142                         break;
143
144                 case DIFF_STATUS_COPIED:
145                 case DIFF_STATUS_RENAMED:
146                         printf("%c \"%s\" \"%s\"\n", q->queue[i]->status,
147                                ospec->path, spec->path);
148
149                         if (!hashcmp(ospec->sha1, spec->sha1) &&
150                             ospec->mode == spec->mode)
151                                 break;
152                         /* fallthrough */
153
154                 case DIFF_STATUS_TYPE_CHANGED:
155                 case DIFF_STATUS_MODIFIED:
156                 case DIFF_STATUS_ADDED:
157                         /*
158                          * Links refer to objects in another repositories;
159                          * output the SHA-1 verbatim.
160                          */
161                         if (S_ISGITLINK(spec->mode))
162                                 printf("M %06o %s %s\n", spec->mode,
163                                        sha1_to_hex(spec->sha1), spec->path);
164                         else {
165                                 struct object *object = lookup_object(spec->sha1);
166                                 printf("M %06o :%d %s\n", spec->mode,
167                                        get_object_mark(object), spec->path);
168                         }
169                         break;
170
171                 default:
172                         die("Unexpected comparison status '%c' for %s, %s",
173                                 q->queue[i]->status,
174                                 ospec->path ? ospec->path : "none",
175                                 spec->path ? spec->path : "none");
176                 }
177         }
178 }
179
180 static const char *find_encoding(const char *begin, const char *end)
181 {
182         const char *needle = "\nencoding ";
183         char *bol, *eol;
184
185         bol = memmem(begin, end ? end - begin : strlen(begin),
186                      needle, strlen(needle));
187         if (!bol)
188                 return git_commit_encoding;
189         bol += strlen(needle);
190         eol = strchrnul(bol, '\n');
191         *eol = '\0';
192         return bol;
193 }
194
195 static void handle_commit(struct commit *commit, struct rev_info *rev)
196 {
197         int saved_output_format = rev->diffopt.output_format;
198         const char *author, *author_end, *committer, *committer_end;
199         const char *encoding, *message;
200         char *reencoded = NULL;
201         struct commit_list *p;
202         int i;
203
204         rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
205
206         parse_commit(commit);
207         author = strstr(commit->buffer, "\nauthor ");
208         if (!author)
209                 die ("Could not find author in commit %s",
210                      sha1_to_hex(commit->object.sha1));
211         author++;
212         author_end = strchrnul(author, '\n');
213         committer = strstr(author_end, "\ncommitter ");
214         if (!committer)
215                 die ("Could not find committer in commit %s",
216                      sha1_to_hex(commit->object.sha1));
217         committer++;
218         committer_end = strchrnul(committer, '\n');
219         message = strstr(committer_end, "\n\n");
220         encoding = find_encoding(committer_end, message);
221         if (message)
222                 message += 2;
223
224         if (commit->parents &&
225             get_object_mark(&commit->parents->item->object) != 0) {
226                 parse_commit(commit->parents->item);
227                 diff_tree_sha1(commit->parents->item->tree->object.sha1,
228                                commit->tree->object.sha1, "", &rev->diffopt);
229         }
230         else
231                 diff_root_tree_sha1(commit->tree->object.sha1,
232                                     "", &rev->diffopt);
233
234         /* Export the referenced blobs, and remember the marks. */
235         for (i = 0; i < diff_queued_diff.nr; i++)
236                 if (!S_ISGITLINK(diff_queued_diff.queue[i]->two->mode))
237                         handle_object(diff_queued_diff.queue[i]->two->sha1);
238
239         mark_next_object(&commit->object);
240         if (!is_encoding_utf8(encoding))
241                 reencoded = reencode_string(message, "UTF-8", encoding);
242         if (!commit->parents)
243                 printf("reset %s\n", (const char*)commit->util);
244         printf("commit %s\nmark :%"PRIu32"\n%.*s\n%.*s\ndata %u\n%s",
245                (const char *)commit->util, last_idnum,
246                (int)(author_end - author), author,
247                (int)(committer_end - committer), committer,
248                (unsigned)(reencoded
249                           ? strlen(reencoded) : message
250                           ? strlen(message) : 0),
251                reencoded ? reencoded : message ? message : "");
252         free(reencoded);
253
254         for (i = 0, p = commit->parents; p; p = p->next) {
255                 int mark = get_object_mark(&p->item->object);
256                 if (!mark)
257                         continue;
258                 if (i == 0)
259                         printf("from :%d\n", mark);
260                 else
261                         printf("merge :%d\n", mark);
262                 i++;
263         }
264
265         log_tree_diff_flush(rev);
266         rev->diffopt.output_format = saved_output_format;
267
268         printf("\n");
269
270         show_progress();
271 }
272
273 static void handle_tail(struct object_array *commits, struct rev_info *revs)
274 {
275         struct commit *commit;
276         while (commits->nr) {
277                 commit = (struct commit *)commits->objects[commits->nr - 1].item;
278                 if (has_unshown_parent(commit))
279                         return;
280                 handle_commit(commit, revs);
281                 commits->nr--;
282         }
283 }
284
285 static void handle_tag(const char *name, struct tag *tag)
286 {
287         unsigned long size;
288         enum object_type type;
289         char *buf;
290         const char *tagger, *tagger_end, *message;
291         size_t message_size = 0;
292
293         buf = read_sha1_file(tag->object.sha1, &type, &size);
294         if (!buf)
295                 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
296         message = memmem(buf, size, "\n\n", 2);
297         if (message) {
298                 message += 2;
299                 message_size = strlen(message);
300         }
301         tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
302         if (!tagger) {
303                 if (fake_missing_tagger)
304                         tagger = "tagger Unspecified Tagger "
305                                 "<unspecified-tagger> 0 +0000";
306                 else
307                         tagger = "";
308                 tagger_end = tagger + strlen(tagger);
309         } else {
310                 tagger++;
311                 tagger_end = strchrnul(tagger, '\n');
312         }
313
314         /* handle signed tags */
315         if (message) {
316                 const char *signature = strstr(message,
317                                                "\n-----BEGIN PGP SIGNATURE-----\n");
318                 if (signature)
319                         switch(signed_tag_mode) {
320                         case ABORT:
321                                 die ("Encountered signed tag %s; use "
322                                      "--signed-tag=<mode> to handle it.",
323                                      sha1_to_hex(tag->object.sha1));
324                         case WARN:
325                                 warning ("Exporting signed tag %s",
326                                          sha1_to_hex(tag->object.sha1));
327                                 /* fallthru */
328                         case VERBATIM:
329                                 break;
330                         case STRIP:
331                                 message_size = signature + 1 - message;
332                                 break;
333                         }
334         }
335
336         if (!prefixcmp(name, "refs/tags/"))
337                 name += 10;
338         printf("tag %s\nfrom :%d\n%.*s%sdata %d\n%.*s\n",
339                name, get_object_mark(tag->tagged),
340                (int)(tagger_end - tagger), tagger,
341                tagger == tagger_end ? "" : "\n",
342                (int)message_size, (int)message_size, message ? message : "");
343 }
344
345 static void get_tags_and_duplicates(struct object_array *pending,
346                                     struct string_list *extra_refs)
347 {
348         struct tag *tag;
349         int i;
350
351         for (i = 0; i < pending->nr; i++) {
352                 struct object_array_entry *e = pending->objects + i;
353                 unsigned char sha1[20];
354                 struct commit *commit = commit;
355                 char *full_name;
356
357                 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
358                         continue;
359
360                 switch (e->item->type) {
361                 case OBJ_COMMIT:
362                         commit = (struct commit *)e->item;
363                         break;
364                 case OBJ_TAG:
365                         tag = (struct tag *)e->item;
366                         while (tag && tag->object.type == OBJ_TAG) {
367                                 string_list_append(full_name, extra_refs)->util = tag;
368                                 tag = (struct tag *)tag->tagged;
369                         }
370                         if (!tag)
371                                 die ("Tag %s points nowhere?", e->name);
372                         switch(tag->object.type) {
373                         case OBJ_COMMIT:
374                                 commit = (struct commit *)tag;
375                                 break;
376                         case OBJ_BLOB:
377                                 handle_object(tag->object.sha1);
378                                 continue;
379                         }
380                         break;
381                 default:
382                         die ("Unexpected object of type %s",
383                              typename(e->item->type));
384                 }
385                 if (commit->util)
386                         /* more than one name for the same object */
387                         string_list_append(full_name, extra_refs)->util = commit;
388                 else
389                         commit->util = full_name;
390         }
391 }
392
393 static void handle_tags_and_duplicates(struct string_list *extra_refs)
394 {
395         struct commit *commit;
396         int i;
397
398         for (i = extra_refs->nr - 1; i >= 0; i--) {
399                 const char *name = extra_refs->items[i].string;
400                 struct object *object = extra_refs->items[i].util;
401                 switch (object->type) {
402                 case OBJ_TAG:
403                         handle_tag(name, (struct tag *)object);
404                         break;
405                 case OBJ_COMMIT:
406                         /* create refs pointing to already seen commits */
407                         commit = (struct commit *)object;
408                         printf("reset %s\nfrom :%d\n\n", name,
409                                get_object_mark(&commit->object));
410                         show_progress();
411                         break;
412                 }
413         }
414 }
415
416 static void export_marks(char *file)
417 {
418         unsigned int i;
419         uint32_t mark;
420         struct object_decoration *deco = idnums.hash;
421         FILE *f;
422
423         f = fopen(file, "w");
424         if (!f)
425                 error("Unable to open marks file %s for writing", file);
426
427         for (i = 0; i < idnums.size; i++) {
428                 if (deco->base && deco->base->type == 1) {
429                         mark = ptr_to_mark(deco->decoration);
430                         fprintf(f, ":%"PRIu32" %s\n", mark,
431                                 sha1_to_hex(deco->base->sha1));
432                 }
433                 deco++;
434         }
435
436         if (ferror(f) || fclose(f))
437                 error("Unable to write marks file %s.", file);
438 }
439
440 static void import_marks(char *input_file)
441 {
442         char line[512];
443         FILE *f = fopen(input_file, "r");
444         if (!f)
445                 die("cannot read %s: %s", input_file, strerror(errno));
446
447         while (fgets(line, sizeof(line), f)) {
448                 uint32_t mark;
449                 char *line_end, *mark_end;
450                 unsigned char sha1[20];
451                 struct object *object;
452
453                 line_end = strchr(line, '\n');
454                 if (line[0] != ':' || !line_end)
455                         die("corrupt mark line: %s", line);
456                 *line_end = '\0';
457
458                 mark = strtoumax(line + 1, &mark_end, 10);
459                 if (!mark || mark_end == line + 1
460                         || *mark_end != ' ' || get_sha1(mark_end + 1, sha1))
461                         die("corrupt mark line: %s", line);
462
463                 object = parse_object(sha1);
464                 if (!object)
465                         die ("Could not read blob %s", sha1_to_hex(sha1));
466
467                 if (object->flags & SHOWN)
468                         error("Object %s already has a mark", sha1);
469
470                 mark_object(object, mark);
471                 if (last_idnum < mark)
472                         last_idnum = mark;
473
474                 object->flags |= SHOWN;
475         }
476         fclose(f);
477 }
478
479 int cmd_fast_export(int argc, const char **argv, const char *prefix)
480 {
481         struct rev_info revs;
482         struct object_array commits = { 0, 0, NULL };
483         struct string_list extra_refs = { NULL, 0, 0, 0 };
484         struct commit *commit;
485         char *export_filename = NULL, *import_filename = NULL;
486         struct option options[] = {
487                 OPT_INTEGER(0, "progress", &progress,
488                             "show progress after <n> objects"),
489                 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
490                              "select handling of signed tags",
491                              parse_opt_signed_tag_mode),
492                 OPT_STRING(0, "export-marks", &export_filename, "FILE",
493                              "Dump marks to this file"),
494                 OPT_STRING(0, "import-marks", &import_filename, "FILE",
495                              "Import marks from this file"),
496                 OPT_BOOLEAN(0, "fake-missing-tagger", &fake_missing_tagger,
497                              "Fake a tagger when tags lack one"),
498                 OPT_END()
499         };
500
501         if (argc == 1)
502                 usage_with_options (fast_export_usage, options);
503
504         /* we handle encodings */
505         git_config(git_default_config, NULL);
506
507         init_revisions(&revs, prefix);
508         argc = setup_revisions(argc, argv, &revs, NULL);
509         argc = parse_options(argc, argv, options, fast_export_usage, 0);
510         if (argc > 1)
511                 usage_with_options (fast_export_usage, options);
512
513         if (import_filename)
514                 import_marks(import_filename);
515
516         get_tags_and_duplicates(&revs.pending, &extra_refs);
517
518         revs.topo_order = 1;
519         if (prepare_revision_walk(&revs))
520                 die("revision walk setup failed");
521         revs.diffopt.format_callback = show_filemodify;
522         DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
523         while ((commit = get_revision(&revs))) {
524                 if (has_unshown_parent(commit)) {
525                         struct commit_list *parent = commit->parents;
526                         add_object_array(&commit->object, NULL, &commits);
527                         for (; parent; parent = parent->next)
528                                 if (!parent->item->util)
529                                         parent->item->util = commit->util;
530                 }
531                 else {
532                         handle_commit(commit, &revs);
533                         handle_tail(&commits, &revs);
534                 }
535         }
536
537         handle_tags_and_duplicates(&extra_refs);
538
539         if (export_filename)
540                 export_marks(export_filename);
541
542         return 0;
543 }