builtin-fast-export: Only output a single parent per line
[git/git.git] / builtin-fast-export.c
CommitLineData
f2dc849e
JS
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 "path-list.h"
17#include "utf8.h"
18#include "parse-options.h"
19
20static const char *fast_export_usage[] = {
21 "git-fast-export [rev-list-opts]",
22 NULL
23};
24
25static int progress;
ee4bc371 26static enum { VERBATIM, WARN, STRIP, ABORT } signed_tag_mode = ABORT;
f2dc849e
JS
27
28static int parse_opt_signed_tag_mode(const struct option *opt,
29 const char *arg, int unset)
30{
31 if (unset || !strcmp(arg, "abort"))
32 signed_tag_mode = ABORT;
ee4bc371
JS
33 else if (!strcmp(arg, "verbatim") || !strcmp(arg, "ignore"))
34 signed_tag_mode = VERBATIM;
f2dc849e
JS
35 else if (!strcmp(arg, "warn"))
36 signed_tag_mode = WARN;
37 else if (!strcmp(arg, "strip"))
38 signed_tag_mode = STRIP;
39 else
40 return error("Unknown signed-tag mode: %s", arg);
41 return 0;
42}
43
44static struct decoration idnums;
45static uint32_t last_idnum;
46
47static int has_unshown_parent(struct commit *commit)
48{
49 struct commit_list *parent;
50
51 for (parent = commit->parents; parent; parent = parent->next)
52 if (!(parent->item->object.flags & SHOWN) &&
53 !(parent->item->object.flags & UNINTERESTING))
54 return 1;
55 return 0;
56}
57
58/* Since intptr_t is C99, we do not use it here */
59static void mark_object(struct object *object)
60{
61 last_idnum++;
62 add_decoration(&idnums, object, ((uint32_t *)NULL) + last_idnum);
63}
64
65static int get_object_mark(struct object *object)
66{
67 void *decoration = lookup_decoration(&idnums, object);
68 if (!decoration)
69 return 0;
70 return (uint32_t *)decoration - (uint32_t *)NULL;
71}
72
73static void show_progress(void)
74{
75 static int counter = 0;
76 if (!progress)
77 return;
78 if ((++counter % progress) == 0)
79 printf("progress %d objects\n", counter);
80}
81
82static void handle_object(const unsigned char *sha1)
83{
84 unsigned long size;
85 enum object_type type;
86 char *buf;
87 struct object *object;
88
89 if (is_null_sha1(sha1))
90 return;
91
92 object = parse_object(sha1);
93 if (!object)
94 die ("Could not read blob %s", sha1_to_hex(sha1));
95
96 if (object->flags & SHOWN)
97 return;
98
99 buf = read_sha1_file(sha1, &type, &size);
100 if (!buf)
101 die ("Could not read blob %s", sha1_to_hex(sha1));
102
103 mark_object(object);
104
105 printf("blob\nmark :%d\ndata %lu\n", last_idnum, size);
b0fe0d72 106 if (size && fwrite(buf, size, 1, stdout) != 1)
f2dc849e
JS
107 die ("Could not write blob %s", sha1_to_hex(sha1));
108 printf("\n");
109
110 show_progress();
111
112 object->flags |= SHOWN;
113 free(buf);
114}
115
116static void show_filemodify(struct diff_queue_struct *q,
117 struct diff_options *options, void *data)
118{
119 int i;
120 for (i = 0; i < q->nr; i++) {
121 struct diff_filespec *spec = q->queue[i]->two;
122 if (is_null_sha1(spec->sha1))
123 printf("D %s\n", spec->path);
124 else {
125 struct object *object = lookup_object(spec->sha1);
ce4a7bff 126 printf("M %06o :%d %s\n", spec->mode,
f2dc849e
JS
127 get_object_mark(object), spec->path);
128 }
129 }
130}
131
132static const char *find_encoding(const char *begin, const char *end)
133{
134 const char *needle = "\nencoding ";
135 char *bol, *eol;
136
137 bol = memmem(begin, end ? end - begin : strlen(begin),
138 needle, strlen(needle));
139 if (!bol)
140 return git_commit_encoding;
141 bol += strlen(needle);
142 eol = strchrnul(bol, '\n');
143 *eol = '\0';
144 return bol;
145}
146
147static void handle_commit(struct commit *commit, struct rev_info *rev)
148{
149 int saved_output_format = rev->diffopt.output_format;
150 const char *author, *author_end, *committer, *committer_end;
151 const char *encoding, *message;
152 char *reencoded = NULL;
153 struct commit_list *p;
154 int i;
155
156 rev->diffopt.output_format = DIFF_FORMAT_CALLBACK;
157
158 parse_commit(commit);
159 author = strstr(commit->buffer, "\nauthor ");
160 if (!author)
161 die ("Could not find author in commit %s",
162 sha1_to_hex(commit->object.sha1));
163 author++;
164 author_end = strchrnul(author, '\n');
165 committer = strstr(author_end, "\ncommitter ");
166 if (!committer)
167 die ("Could not find committer in commit %s",
168 sha1_to_hex(commit->object.sha1));
169 committer++;
170 committer_end = strchrnul(committer, '\n');
171 message = strstr(committer_end, "\n\n");
172 encoding = find_encoding(committer_end, message);
173 if (message)
174 message += 2;
175
176 if (commit->parents) {
177 parse_commit(commit->parents->item);
178 diff_tree_sha1(commit->parents->item->tree->object.sha1,
179 commit->tree->object.sha1, "", &rev->diffopt);
180 }
181 else
182 diff_root_tree_sha1(commit->tree->object.sha1,
183 "", &rev->diffopt);
184
185 for (i = 0; i < diff_queued_diff.nr; i++)
186 handle_object(diff_queued_diff.queue[i]->two->sha1);
187
188 mark_object(&commit->object);
189 if (!is_encoding_utf8(encoding))
190 reencoded = reencode_string(message, "UTF-8", encoding);
191 printf("commit %s\nmark :%d\n%.*s\n%.*s\ndata %u\n%s",
192 (const char *)commit->util, last_idnum,
193 (int)(author_end - author), author,
194 (int)(committer_end - committer), committer,
195 (unsigned)(reencoded
196 ? strlen(reencoded) : message
197 ? strlen(message) : 0),
198 reencoded ? reencoded : message ? message : "");
199 if (reencoded)
200 free(reencoded);
201
202 for (i = 0, p = commit->parents; p; p = p->next) {
203 int mark = get_object_mark(&p->item->object);
204 if (!mark)
205 continue;
206 if (i == 0)
207 printf("from :%d\n", mark);
f2dc849e 208 else
5070b49e 209 printf("merge :%d\n", mark);
f2dc849e
JS
210 i++;
211 }
f2dc849e
JS
212
213 log_tree_diff_flush(rev);
214 rev->diffopt.output_format = saved_output_format;
215
216 printf("\n");
217
218 show_progress();
219}
220
221static void handle_tail(struct object_array *commits, struct rev_info *revs)
222{
223 struct commit *commit;
224 while (commits->nr) {
225 commit = (struct commit *)commits->objects[commits->nr - 1].item;
226 if (has_unshown_parent(commit))
227 return;
228 handle_commit(commit, revs);
229 commits->nr--;
230 }
231}
232
233static void handle_tag(const char *name, struct tag *tag)
234{
235 unsigned long size;
236 enum object_type type;
237 char *buf;
238 const char *tagger, *tagger_end, *message;
239 size_t message_size = 0;
240
241 buf = read_sha1_file(tag->object.sha1, &type, &size);
242 if (!buf)
243 die ("Could not read tag %s", sha1_to_hex(tag->object.sha1));
244 message = memmem(buf, size, "\n\n", 2);
245 if (message) {
246 message += 2;
247 message_size = strlen(message);
248 }
249 tagger = memmem(buf, message ? message - buf : size, "\ntagger ", 8);
250 if (!tagger)
251 die ("No tagger for tag %s", sha1_to_hex(tag->object.sha1));
252 tagger++;
253 tagger_end = strchrnul(tagger, '\n');
254
255 /* handle signed tags */
256 if (message) {
257 const char *signature = strstr(message,
258 "\n-----BEGIN PGP SIGNATURE-----\n");
259 if (signature)
260 switch(signed_tag_mode) {
261 case ABORT:
262 die ("Encountered signed tag %s; use "
263 "--signed-tag=<mode> to handle it.",
264 sha1_to_hex(tag->object.sha1));
265 case WARN:
266 warning ("Exporting signed tag %s",
267 sha1_to_hex(tag->object.sha1));
268 /* fallthru */
ee4bc371 269 case VERBATIM:
f2dc849e
JS
270 break;
271 case STRIP:
272 message_size = signature + 1 - message;
273 break;
274 }
275 }
276
277 if (!prefixcmp(name, "refs/tags/"))
278 name += 10;
279 printf("tag %s\nfrom :%d\n%.*s\ndata %d\n%.*s\n",
280 name, get_object_mark(tag->tagged),
281 (int)(tagger_end - tagger), tagger,
282 (int)message_size, (int)message_size, message ? message : "");
283}
284
285static void get_tags_and_duplicates(struct object_array *pending,
286 struct path_list *extra_refs)
287{
288 struct tag *tag;
289 int i;
290
291 for (i = 0; i < pending->nr; i++) {
292 struct object_array_entry *e = pending->objects + i;
293 unsigned char sha1[20];
294 struct commit *commit = commit;
295 char *full_name;
296
297 if (dwim_ref(e->name, strlen(e->name), sha1, &full_name) != 1)
298 continue;
299
300 switch (e->item->type) {
301 case OBJ_COMMIT:
302 commit = (struct commit *)e->item;
303 break;
304 case OBJ_TAG:
305 tag = (struct tag *)e->item;
306 while (tag && tag->object.type == OBJ_TAG) {
307 path_list_insert(full_name, extra_refs)->util = tag;
308 tag = (struct tag *)tag->tagged;
309 }
310 if (!tag)
311 die ("Tag %s points nowhere?", e->name);
312 switch(tag->object.type) {
313 case OBJ_COMMIT:
314 commit = (struct commit *)tag;
315 break;
316 case OBJ_BLOB:
317 handle_object(tag->object.sha1);
318 continue;
319 }
320 break;
321 default:
322 die ("Unexpected object of type %s",
323 typename(e->item->type));
324 }
325 if (commit->util)
326 /* more than one name for the same object */
327 path_list_insert(full_name, extra_refs)->util = commit;
328 else
329 commit->util = full_name;
330 }
331}
332
333static void handle_tags_and_duplicates(struct path_list *extra_refs)
334{
335 struct commit *commit;
336 int i;
337
338 for (i = extra_refs->nr - 1; i >= 0; i--) {
339 const char *name = extra_refs->items[i].path;
340 struct object *object = extra_refs->items[i].util;
341 switch (object->type) {
342 case OBJ_TAG:
343 handle_tag(name, (struct tag *)object);
344 break;
345 case OBJ_COMMIT:
346 /* create refs pointing to already seen commits */
347 commit = (struct commit *)object;
348 printf("reset %s\nfrom :%d\n\n", name,
349 get_object_mark(&commit->object));
350 show_progress();
351 break;
352 }
353 }
354}
355
356int cmd_fast_export(int argc, const char **argv, const char *prefix)
357{
358 struct rev_info revs;
359 struct object_array commits = { 0, 0, NULL };
360 struct path_list extra_refs = { NULL, 0, 0, 0 };
361 struct commit *commit;
362 struct option options[] = {
363 OPT_INTEGER(0, "progress", &progress,
364 "show progress after <n> objects"),
365 OPT_CALLBACK(0, "signed-tags", &signed_tag_mode, "mode",
366 "select handling of signed tags",
367 parse_opt_signed_tag_mode),
368 OPT_END()
369 };
370
371 /* we handle encodings */
372 git_config(git_default_config);
373
374 init_revisions(&revs, prefix);
375 argc = setup_revisions(argc, argv, &revs, NULL);
376 argc = parse_options(argc, argv, options, fast_export_usage, 0);
377 if (argc > 1)
378 usage_with_options (fast_export_usage, options);
379
380 get_tags_and_duplicates(&revs.pending, &extra_refs);
381
382 prepare_revision_walk(&revs);
383 revs.diffopt.format_callback = show_filemodify;
384 DIFF_OPT_SET(&revs.diffopt, RECURSIVE);
385 while ((commit = get_revision(&revs))) {
386 if (has_unshown_parent(commit)) {
387 struct commit_list *parent = commit->parents;
388 add_object_array(&commit->object, NULL, &commits);
389 for (; parent; parent = parent->next)
390 if (!parent->item->util)
391 parent->item->util = commit->util;
392 }
393 else {
394 handle_commit(commit, &revs);
395 handle_tail(&commits, &revs);
396 }
397 }
398
399 handle_tags_and_duplicates(&extra_refs);
400
401 return 0;
402}