1 /*
2  * Copyright (c) 2017 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/queue.h>
18 #include <sys/stat.h>
19 
20 #include <stdio.h>
21 #include <stdlib.h>
22 #include <string.h>
23 #include <limits.h>
24 #include <sha1.h>
25 #include <zlib.h>
26 
27 #include "got_repository.h"
28 #include "got_object.h"
29 #include "got_error.h"
30 #include "got_diff.h"
31 #include "got_opentemp.h"
32 
33 #include "got_lib_diff.h"
34 #include "got_lib_path.h"
35 #include "got_lib_delta.h"
36 #include "got_lib_inflate.h"
37 #include "got_lib_object.h"
38 
39 static const struct got_error *
40 diff_blobs(struct got_blob_object *blob1, struct got_blob_object *blob2,
41     const char *label1, const char *label2, int diff_context, FILE *outfile,
42     struct got_diff_changes *changes)
43 {
44 	struct got_diff_state ds;
45 	struct got_diff_args args;
46 	const struct got_error *err = NULL;
47 	FILE *f1 = NULL, *f2 = NULL;
48 	char hex1[SHA1_DIGEST_STRING_LENGTH];
49 	char hex2[SHA1_DIGEST_STRING_LENGTH];
50 	char *idstr1 = NULL, *idstr2 = NULL;
51 	size_t size1, size2;
52 	int res, flags = 0;
53 
54 	if (blob1) {
55 		f1 = got_opentemp();
56 		if (f1 == NULL)
57 			return got_error(GOT_ERR_FILE_OPEN);
58 	} else
59 		flags |= D_EMPTY1;
60 
61 	if (blob2) {
62 		f2 = got_opentemp();
63 		if (f2 == NULL) {
64 			fclose(f1);
65 			return got_error(GOT_ERR_FILE_OPEN);
66 		}
67 	} else
68 		flags |= D_EMPTY2;
69 
70 	size1 = 0;
71 	if (blob1) {
72 		idstr1 = got_object_blob_id_str(blob1, hex1, sizeof(hex1));
73 		err = got_object_blob_dump_to_file(&size1, NULL, f1, blob1);
74 		if (err)
75 			goto done;
76 	} else
77 		idstr1 = "/dev/null";
78 
79 	size2 = 0;
80 	if (blob2) {
81 		idstr2 = got_object_blob_id_str(blob2, hex2, sizeof(hex2));
82 		err = got_object_blob_dump_to_file(&size2, NULL, f2, blob2);
83 		if (err)
84 			goto done;
85 	} else
86 		idstr2 = "/dev/null";
87 
88 	memset(&ds, 0, sizeof(ds));
89 	/* XXX should stat buffers be passed in args instead of ds? */
90 	ds.stb1.st_mode = S_IFREG;
91 	if (blob1)
92 		ds.stb1.st_size = size1;
93 	ds.stb1.st_mtime = 0; /* XXX */
94 
95 	ds.stb2.st_mode = S_IFREG;
96 	if (blob2)
97 		ds.stb2.st_size = size2;
98 	ds.stb2.st_mtime = 0; /* XXX */
99 
100 	memset(&args, 0, sizeof(args));
101 	args.diff_format = D_UNIFIED;
102 	args.label[0] = label1 ? label1 : idstr1;
103 	args.label[1] = label2 ? label2 : idstr2;
104 	args.diff_context = diff_context;
105 	flags |= D_PROTOTYPE;
106 
107 	if (label1 && strcmp(label1, idstr1) != 0)
108 		fprintf(outfile, "blob - %s\n", idstr1);
109 	if (label2 && strcmp(label2, idstr2) != 0)
110 		fprintf(outfile, "blob + %s\n", idstr2);
111 
112 	err = got_diffreg(&res, f1, f2, flags, &args, &ds, outfile, changes);
113 done:
114 	if (f1)
115 		fclose(f1);
116 	if (f2)
117 		fclose(f2);
118 	return err;
119 }
120 
121 const struct got_error *
122 got_diff_blob(struct got_blob_object *blob1, struct got_blob_object *blob2,
123     const char *label1, const char *label2, int diff_context, FILE *outfile)
124 {
125 	return diff_blobs(blob1, blob2, label1, label2, diff_context, outfile,
126 	    NULL);
127 }
128 
129 const struct got_error *
130 got_diff_blob_lines_changed(struct got_diff_changes **changes,
131     struct got_blob_object *blob1, struct got_blob_object *blob2)
132 {
133 	const struct got_error *err = NULL;
134 
135 	*changes = calloc(1, sizeof(**changes));
136 	if (*changes == NULL)
137 		return got_error_from_errno();
138 	SIMPLEQ_INIT(&(*changes)->entries);
139 
140 	err = diff_blobs(blob1, blob2, NULL, NULL, 3, NULL, *changes);
141 	if (err) {
142 		got_diff_free_changes(*changes);
143 		*changes = NULL;
144 	}
145 	return err;
146 }
147 
148 void
149 got_diff_free_changes(struct got_diff_changes *changes)
150 {
151 	struct got_diff_change *change;
152 	while (!SIMPLEQ_EMPTY(&changes->entries)) {
153 		change = SIMPLEQ_FIRST(&changes->entries);
154 		SIMPLEQ_REMOVE_HEAD(&changes->entries, entry);
155 		free(change);
156 	}
157 	free(changes);
158 }
159 
160 struct got_tree_entry *
161 match_entry_by_name(struct got_tree_entry *te1, struct got_tree_object *tree2)
162 {
163 	struct got_tree_entry *te2;
164 	const struct got_tree_entries *entries2;
165 
166 	entries2 = got_object_tree_get_entries(tree2); 
167 	SIMPLEQ_FOREACH(te2, &entries2->head, entry) {
168 		if (strcmp(te1->name, te2->name) == 0)
169 			return te2;
170 	}
171 	return NULL;
172 }
173 
174 static const struct got_error *
175 diff_added_blob(struct got_object_id *id, const char *label,
176     int diff_context, struct got_repository *repo, FILE *outfile)
177 {
178 	const struct got_error *err;
179 	struct got_blob_object  *blob = NULL;
180 	struct got_object *obj = NULL;
181 
182 	err = got_object_open(&obj, repo, id);
183 	if (err)
184 		return err;
185 
186 	err = got_object_blob_open(&blob, repo, obj, 8192);
187 	if (err)
188 		goto done;
189 	err = got_diff_blob(NULL, blob, NULL, label, diff_context, outfile);
190 done:
191 	got_object_close(obj);
192 	if (blob)
193 		got_object_blob_close(blob);
194 	return err;
195 }
196 
197 static const struct got_error *
198 diff_modified_blob(struct got_object_id *id1, struct got_object_id *id2,
199     const char *label1, const char *label2, int diff_context,
200     struct got_repository *repo, FILE *outfile)
201 {
202 	const struct got_error *err;
203 	struct got_object *obj1 = NULL;
204 	struct got_object *obj2 = NULL;
205 	struct got_blob_object *blob1 = NULL;
206 	struct got_blob_object *blob2 = NULL;
207 
208 	err = got_object_open(&obj1, repo, id1);
209 	if (err)
210 		return err;
211 	if (obj1->type != GOT_OBJ_TYPE_BLOB) {
212 		err = got_error(GOT_ERR_OBJ_TYPE);
213 		goto done;
214 	}
215 
216 	err = got_object_open(&obj2, repo, id2);
217 	if (err)
218 		goto done;
219 	if (obj2->type != GOT_OBJ_TYPE_BLOB) {
220 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
221 		goto done;
222 	}
223 
224 	err = got_object_blob_open(&blob1, repo, obj1, 8192);
225 	if (err)
226 		goto done;
227 
228 	err = got_object_blob_open(&blob2, repo, obj2, 8192);
229 	if (err)
230 		goto done;
231 
232 	err = got_diff_blob(blob1, blob2, label1, label2, diff_context,
233 	    outfile);
234 
235 done:
236 	if (obj1)
237 		got_object_close(obj1);
238 	if (obj2)
239 		got_object_close(obj2);
240 	if (blob1)
241 		got_object_blob_close(blob1);
242 	if (blob2)
243 		got_object_blob_close(blob2);
244 	return err;
245 }
246 
247 static const struct got_error *
248 diff_deleted_blob(struct got_object_id *id, const char *label,
249     int diff_context, struct got_repository *repo, FILE *outfile)
250 {
251 	const struct got_error *err;
252 	struct got_blob_object  *blob = NULL;
253 	struct got_object *obj = NULL;
254 
255 	err = got_object_open(&obj, repo, id);
256 	if (err)
257 		return err;
258 
259 	err = got_object_blob_open(&blob, repo, obj, 8192);
260 	if (err)
261 		goto done;
262 	err = got_diff_blob(blob, NULL, label, NULL, diff_context, outfile);
263 done:
264 	got_object_close(obj);
265 	if (blob)
266 		got_object_blob_close(blob);
267 	return err;
268 }
269 
270 static const struct got_error *
271 diff_added_tree(struct got_object_id *id, const char *label,
272     int diff_context, struct got_repository *repo, FILE *outfile)
273 {
274 	const struct got_error *err = NULL;
275 	struct got_object *treeobj = NULL;
276 	struct got_tree_object *tree = NULL;
277 
278 	err = got_object_open(&treeobj, repo, id);
279 	if (err)
280 		goto done;
281 
282 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
283 		err = got_error(GOT_ERR_OBJ_TYPE);
284 		goto done;
285 	}
286 
287 	err = got_object_tree_open(&tree, repo, treeobj);
288 	if (err)
289 		goto done;
290 
291 	err = got_diff_tree(NULL, tree, NULL, label, diff_context, repo,
292 	    outfile);
293 
294 done:
295 	if (tree)
296 		got_object_tree_close(tree);
297 	if (treeobj)
298 		got_object_close(treeobj);
299 	return err;
300 }
301 
302 static const struct got_error *
303 diff_modified_tree(struct got_object_id *id1, struct got_object_id *id2,
304     const char *label1, const char *label2, int diff_context,
305     struct got_repository *repo, FILE *outfile)
306 {
307 	const struct got_error *err;
308 	struct got_object *treeobj1 = NULL;
309 	struct got_object *treeobj2 = NULL;
310 	struct got_tree_object *tree1 = NULL;
311 	struct got_tree_object *tree2 = NULL;
312 
313 	err = got_object_open(&treeobj1, repo, id1);
314 	if (err)
315 		goto done;
316 
317 	if (treeobj1->type != GOT_OBJ_TYPE_TREE) {
318 		err = got_error(GOT_ERR_OBJ_TYPE);
319 		goto done;
320 	}
321 
322 	err = got_object_open(&treeobj2, repo, id2);
323 	if (err)
324 		goto done;
325 
326 	if (treeobj2->type != GOT_OBJ_TYPE_TREE) {
327 		err = got_error(GOT_ERR_OBJ_TYPE);
328 		goto done;
329 	}
330 
331 	err = got_object_tree_open(&tree1, repo, treeobj1);
332 	if (err)
333 		goto done;
334 
335 	err = got_object_tree_open(&tree2, repo, treeobj2);
336 	if (err)
337 		goto done;
338 
339 	err = got_diff_tree(tree1, tree2, label1, label2, diff_context, repo,
340 	    outfile);
341 
342 done:
343 	if (tree1)
344 		got_object_tree_close(tree1);
345 	if (tree2)
346 		got_object_tree_close(tree2);
347 	if (treeobj1)
348 		got_object_close(treeobj1);
349 	if (treeobj2)
350 		got_object_close(treeobj2);
351 	return err;
352 }
353 
354 static const struct got_error *
355 diff_deleted_tree(struct got_object_id *id, const char *label,
356     int diff_context, struct got_repository *repo, FILE *outfile)
357 {
358 	const struct got_error *err;
359 	struct got_object *treeobj = NULL;
360 	struct got_tree_object *tree = NULL;
361 
362 	err = got_object_open(&treeobj, repo, id);
363 	if (err)
364 		goto done;
365 
366 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
367 		err = got_error(GOT_ERR_OBJ_TYPE);
368 		goto done;
369 	}
370 
371 	err = got_object_tree_open(&tree, repo, treeobj);
372 	if (err)
373 		goto done;
374 
375 	err = got_diff_tree(tree, NULL, label, NULL, diff_context, repo,
376 	    outfile);
377 done:
378 	if (tree)
379 		got_object_tree_close(tree);
380 	if (treeobj)
381 		got_object_close(treeobj);
382 	return err;
383 }
384 
385 static const struct got_error *
386 diff_kind_mismatch(struct got_object_id *id1, struct got_object_id *id2,
387     const char *label1, const char *label2, FILE *outfile)
388 {
389 	/* XXX TODO */
390 	return NULL;
391 }
392 
393 static const struct got_error *
394 diff_entry_old_new(struct got_tree_entry *te1, struct got_tree_entry *te2,
395     const char *label1, const char *label2, int diff_context,
396     struct got_repository *repo, FILE *outfile)
397 {
398 	const struct got_error *err = NULL;
399 	int id_match;
400 
401 	if (te2 == NULL) {
402 		if (S_ISDIR(te1->mode))
403 			err = diff_deleted_tree(te1->id, label1, diff_context,
404 			    repo, outfile);
405 		else
406 			err = diff_deleted_blob(te1->id, label1, diff_context,
407 			    repo, outfile);
408 		return err;
409 	}
410 
411 	id_match = (got_object_id_cmp(te1->id, te2->id) == 0);
412 	if (S_ISDIR(te1->mode) && S_ISDIR(te2->mode)) {
413 		if (!id_match)
414 			return diff_modified_tree(te1->id, te2->id,
415 			    label1, label2, diff_context, repo, outfile);
416 	} else if (S_ISREG(te1->mode) && S_ISREG(te2->mode)) {
417 		if (!id_match)
418 			return diff_modified_blob(te1->id, te2->id,
419 			    label1, label2, diff_context, repo, outfile);
420 	}
421 
422 	if (id_match)
423 		return NULL;
424 
425 	return diff_kind_mismatch(te1->id, te2->id, label1, label2, outfile);
426 }
427 
428 static const struct got_error *
429 diff_entry_new_old(struct got_tree_entry *te2, struct got_tree_entry *te1,
430     const char *label2, int diff_context, struct got_repository *repo,
431     FILE *outfile)
432 {
433 	if (te1 != NULL) /* handled by diff_entry_old_new() */
434 		return NULL;
435 
436 	if (S_ISDIR(te2->mode))
437 		return diff_added_tree(te2->id, label2, diff_context, repo,
438 		    outfile);
439 
440 	return diff_added_blob(te2->id, label2, diff_context, repo, outfile);
441 }
442 
443 const struct got_error *
444 got_diff_tree(struct got_tree_object *tree1, struct got_tree_object *tree2,
445     const char *label1, const char *label2, int diff_context,
446     struct got_repository *repo, FILE *outfile)
447 {
448 	const struct got_error *err = NULL;
449 	struct got_tree_entry *te1 = NULL;
450 	struct got_tree_entry *te2 = NULL;
451 	char *l1 = NULL, *l2 = NULL;
452 
453 	if (tree1) {
454 		const struct got_tree_entries *entries;
455 		entries = got_object_tree_get_entries(tree1);
456 		te1 = SIMPLEQ_FIRST(&entries->head);
457 		if (te1 && asprintf(&l1, "%s%s%s", label1, label1[0] ? "/" : "",
458 		    te1->name) == -1)
459 			return got_error_from_errno();
460 	}
461 	if (tree2) {
462 		const struct got_tree_entries *entries;
463 		entries = got_object_tree_get_entries(tree2);
464 		te2 = SIMPLEQ_FIRST(&entries->head);
465 		if (te2 && asprintf(&l2, "%s%s%s", label2, label2[0] ? "/" : "",
466 		    te2->name) == -1)
467 			return got_error_from_errno();
468 	}
469 
470 	do {
471 		if (te1) {
472 			struct got_tree_entry *te = NULL;
473 			if (tree2)
474 				te = match_entry_by_name(te1, tree2);
475 			if (te) {
476 				free(l2);
477 				l2 = NULL;
478 				if (te && asprintf(&l2, "%s%s%s", label2,
479 				    label2[0] ? "/" : "", te->name) == -1)
480 					return got_error_from_errno();
481 			}
482 			err = diff_entry_old_new(te1, te, l1, l2, diff_context,
483 			    repo, outfile);
484 			if (err)
485 				break;
486 		}
487 
488 		if (te2) {
489 			struct got_tree_entry *te = NULL;
490 			if (tree1)
491 				te = match_entry_by_name(te2, tree1);
492 			free(l2);
493 			if (te) {
494 				if (asprintf(&l2, "%s%s%s", label2,
495 				    label2[0] ? "/" : "", te->name) == -1)
496 					return got_error_from_errno();
497 			} else {
498 				if (asprintf(&l2, "%s%s%s", label2,
499 				    label2[0] ? "/" : "", te2->name) == -1)
500 					return got_error_from_errno();
501 			}
502 			err = diff_entry_new_old(te2, te, l2, diff_context,
503 			    repo, outfile);
504 			if (err)
505 				break;
506 		}
507 
508 		free(l1);
509 		l1 = NULL;
510 		if (te1) {
511 			te1 = SIMPLEQ_NEXT(te1, entry);
512 			if (te1 &&
513 			    asprintf(&l1, "%s%s%s", label1,
514 			    label1[0] ? "/" : "", te1->name) == -1)
515 				return got_error_from_errno();
516 		}
517 		free(l2);
518 		l2 = NULL;
519 		if (te2) {
520 			te2 = SIMPLEQ_NEXT(te2, entry);
521 			if (te2 &&
522 			    asprintf(&l2, "%s%s%s", label2,
523 			        label2[0] ? "/" : "", te2->name) == -1)
524 				return got_error_from_errno();
525 		}
526 	} while (te1 || te2);
527 
528 	return err;
529 }
530 
531 const struct got_error *
532 got_diff_objects_as_blobs(struct got_object_id *id1, struct got_object_id *id2,
533     const char *label1, const char *label2, int diff_context,
534     struct got_repository *repo, FILE *outfile)
535 {
536 	const struct got_error *err;
537 	struct got_blob_object *blob1 = NULL, *blob2 = NULL;
538 
539 	if (id1 == NULL && id2 == NULL)
540 		return got_error(GOT_ERR_NO_OBJ);
541 
542 	if (id1) {
543 		err = got_object_open_as_blob(&blob1, repo, id1, 8192);
544 		if (err)
545 			goto done;
546 	}
547 	if (id2) {
548 		err = got_object_open_as_blob(&blob2, repo, id2, 8192);
549 		if (err)
550 			goto done;
551 	}
552 	err = got_diff_blob(blob1, blob2, label1, label2, diff_context,
553 	    outfile);
554 done:
555 	if (blob1)
556 		got_object_blob_close(blob1);
557 	if (blob2)
558 		got_object_blob_close(blob2);
559 	return err;
560 }
561 
562 const struct got_error *
563 got_diff_objects_as_trees(struct got_object_id *id1, struct got_object_id *id2,
564     char *label1, char *label2, int diff_context, struct got_repository *repo,
565     FILE *outfile)
566 {
567 	const struct got_error *err;
568 	struct got_tree_object *tree1 = NULL, *tree2 = NULL;
569 
570 	if (id1 == NULL && id2 == NULL)
571 		return got_error(GOT_ERR_NO_OBJ);
572 
573 	if (id1) {
574 		err = got_object_open_as_tree(&tree1, repo, id1);
575 		if (err)
576 			goto done;
577 	}
578 	if (id2) {
579 		err = got_object_open_as_tree(&tree2, repo, id2);
580 		if (err)
581 			goto done;
582 	}
583 	err = got_diff_tree(tree1, tree2, label1, label2, diff_context,
584 	   repo, outfile);
585 done:
586 	if (tree1)
587 		got_object_tree_close(tree1);
588 	if (tree2)
589 		got_object_tree_close(tree2);
590 	return err;
591 }
592 
593 const struct got_error *
594 got_diff_objects_as_commits(struct got_object_id *id1,
595     struct got_object_id *id2, int diff_context,
596     struct got_repository *repo, FILE *outfile)
597 {
598 	const struct got_error *err;
599 	struct got_commit_object *commit1 = NULL, *commit2 = NULL;
600 
601 	if (id2 == NULL)
602 		return got_error(GOT_ERR_NO_OBJ);
603 
604 	if (id1) {
605 		err = got_object_open_as_commit(&commit1, repo, id1);
606 		if (err)
607 			goto done;
608 	}
609 
610 	err = got_object_open_as_commit(&commit2, repo, id2);
611 	if (err)
612 		goto done;
613 
614 	err = got_diff_objects_as_trees(
615 	    commit1 ? got_object_commit_get_tree_id(commit1) : NULL,
616 	    got_object_commit_get_tree_id(commit2), "", "", diff_context, repo,
617 	    outfile);
618 done:
619 	if (commit1)
620 		got_object_commit_close(commit1);
621 	if (commit2)
622 		got_object_commit_close(commit2);
623 	return err;
624 }