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 #include "got_path.h"
33 
34 #include "got_lib_diff.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_prefix_errno("got_opentemp");
58 	} else
59 		flags |= D_EMPTY1;
60 
61 	if (blob2) {
62 		f2 = got_opentemp();
63 		if (f2 == NULL) {
64 			err = got_error_prefix_errno("got_opentemp");
65 			fclose(f1);
66 			return err;
67 		}
68 	} else
69 		flags |= D_EMPTY2;
70 
71 	size1 = 0;
72 	if (blob1) {
73 		idstr1 = got_object_blob_id_str(blob1, hex1, sizeof(hex1));
74 		err = got_object_blob_dump_to_file(&size1, NULL, f1, blob1);
75 		if (err)
76 			goto done;
77 	} else
78 		idstr1 = "/dev/null";
79 
80 	size2 = 0;
81 	if (blob2) {
82 		idstr2 = got_object_blob_id_str(blob2, hex2, sizeof(hex2));
83 		err = got_object_blob_dump_to_file(&size2, NULL, f2, blob2);
84 		if (err)
85 			goto done;
86 	} else
87 		idstr2 = "/dev/null";
88 
89 	memset(&ds, 0, sizeof(ds));
90 	/* XXX should stat buffers be passed in args instead of ds? */
91 	ds.stb1.st_mode = S_IFREG;
92 	if (blob1)
93 		ds.stb1.st_size = size1;
94 	ds.stb1.st_mtime = 0; /* XXX */
95 
96 	ds.stb2.st_mode = S_IFREG;
97 	if (blob2)
98 		ds.stb2.st_size = size2;
99 	ds.stb2.st_mtime = 0; /* XXX */
100 
101 	memset(&args, 0, sizeof(args));
102 	args.diff_format = D_UNIFIED;
103 	args.label[0] = label1 ? label1 : idstr1;
104 	args.label[1] = label2 ? label2 : idstr2;
105 	args.diff_context = diff_context;
106 	flags |= D_PROTOTYPE;
107 
108 	if (outfile) {
109 		fprintf(outfile, "blob - %s\n", idstr1);
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 && fclose(f1) != 0 && err == NULL)
115 		err = got_error_prefix_errno("fclose");
116 	if (f2 && fclose(f2) != 0 && err == NULL)
117 		err = got_error_prefix_errno("fclose");
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_file(struct got_blob_object *blob1, FILE *f2, size_t size2,
131     const char *label2, int diff_context, FILE *outfile)
132 {
133 	struct got_diff_state ds;
134 	struct got_diff_args args;
135 	const struct got_error *err = NULL;
136 	FILE *f1 = NULL;
137 	char hex1[SHA1_DIGEST_STRING_LENGTH];
138 	char *idstr1 = NULL;
139 	size_t size1;
140 	int res, flags = 0;
141 
142 	size1 = 0;
143 	if (blob1) {
144 		f1 = got_opentemp();
145 		if (f1 == NULL)
146 			return got_error_prefix_errno("got_opentemp");
147 		idstr1 = got_object_blob_id_str(blob1, hex1, sizeof(hex1));
148 		err = got_object_blob_dump_to_file(&size1, NULL, f1, blob1);
149 		if (err)
150 			goto done;
151 	} else {
152 		flags |= D_EMPTY1;
153 		idstr1 = "/dev/null";
154 	}
155 
156 	if (f2 == NULL)
157 		flags |= D_EMPTY2;
158 
159 	memset(&ds, 0, sizeof(ds));
160 	/* XXX should stat buffers be passed in args instead of ds? */
161 	ds.stb1.st_mode = S_IFREG;
162 	if (blob1)
163 		ds.stb1.st_size = size1;
164 	ds.stb1.st_mtime = 0; /* XXX */
165 
166 	ds.stb2.st_mode = S_IFREG;
167 	ds.stb2.st_size = size2;
168 	ds.stb2.st_mtime = 0; /* XXX */
169 
170 	memset(&args, 0, sizeof(args));
171 	args.diff_format = D_UNIFIED;
172 	args.label[0] = label2;
173 	args.label[1] = label2;
174 	args.diff_context = diff_context;
175 	flags |= D_PROTOTYPE;
176 
177 	fprintf(outfile, "blob - %s\n", idstr1);
178 	fprintf(outfile, "file + %s\n", f2 == NULL ? "/dev/null" : label2);
179 	err = got_diffreg(&res, f1, f2, flags, &args, &ds, outfile, NULL);
180 done:
181 	if (f1 && fclose(f1) != 0 && err == NULL)
182 		err = got_error_prefix_errno("fclose");
183 	return err;
184 }
185 
186 const struct got_error *
187 got_diff_blob_lines_changed(struct got_diff_changes **changes,
188     struct got_blob_object *blob1, struct got_blob_object *blob2)
189 {
190 	const struct got_error *err = NULL;
191 
192 	*changes = calloc(1, sizeof(**changes));
193 	if (*changes == NULL)
194 		return got_error_prefix_errno("calloc");
195 	SIMPLEQ_INIT(&(*changes)->entries);
196 
197 	err = diff_blobs(blob1, blob2, NULL, NULL, 3, NULL, *changes);
198 	if (err) {
199 		got_diff_free_changes(*changes);
200 		*changes = NULL;
201 	}
202 	return err;
203 }
204 
205 void
206 got_diff_free_changes(struct got_diff_changes *changes)
207 {
208 	struct got_diff_change *change;
209 	while (!SIMPLEQ_EMPTY(&changes->entries)) {
210 		change = SIMPLEQ_FIRST(&changes->entries);
211 		SIMPLEQ_REMOVE_HEAD(&changes->entries, entry);
212 		free(change);
213 	}
214 	free(changes);
215 }
216 
217 static struct got_tree_entry *
218 match_entry_by_name(struct got_tree_entry *te1, struct got_tree_object *tree2)
219 {
220 	struct got_tree_entry *te2;
221 	const struct got_tree_entries *entries2;
222 
223 	entries2 = got_object_tree_get_entries(tree2); 
224 	SIMPLEQ_FOREACH(te2, &entries2->head, entry) {
225 		if (strcmp(te1->name, te2->name) == 0)
226 			return te2;
227 	}
228 	return NULL;
229 }
230 
231 static const struct got_error *
232 diff_added_blob(struct got_object_id *id, const char *label,
233     int diff_context, struct got_repository *repo, FILE *outfile)
234 {
235 	const struct got_error *err;
236 	struct got_blob_object  *blob = NULL;
237 	struct got_object *obj = NULL;
238 
239 	err = got_object_open(&obj, repo, id);
240 	if (err)
241 		return err;
242 
243 	err = got_object_blob_open(&blob, repo, obj, 8192);
244 	if (err)
245 		goto done;
246 	err = got_diff_blob(NULL, blob, NULL, label, diff_context, outfile);
247 done:
248 	got_object_close(obj);
249 	if (blob)
250 		got_object_blob_close(blob);
251 	return err;
252 }
253 
254 static const struct got_error *
255 diff_modified_blob(struct got_object_id *id1, struct got_object_id *id2,
256     const char *label1, const char *label2, int diff_context,
257     struct got_repository *repo, FILE *outfile)
258 {
259 	const struct got_error *err;
260 	struct got_object *obj1 = NULL;
261 	struct got_object *obj2 = NULL;
262 	struct got_blob_object *blob1 = NULL;
263 	struct got_blob_object *blob2 = NULL;
264 
265 	err = got_object_open(&obj1, repo, id1);
266 	if (err)
267 		return err;
268 	if (obj1->type != GOT_OBJ_TYPE_BLOB) {
269 		err = got_error(GOT_ERR_OBJ_TYPE);
270 		goto done;
271 	}
272 
273 	err = got_object_open(&obj2, repo, id2);
274 	if (err)
275 		goto done;
276 	if (obj2->type != GOT_OBJ_TYPE_BLOB) {
277 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
278 		goto done;
279 	}
280 
281 	err = got_object_blob_open(&blob1, repo, obj1, 8192);
282 	if (err)
283 		goto done;
284 
285 	err = got_object_blob_open(&blob2, repo, obj2, 8192);
286 	if (err)
287 		goto done;
288 
289 	err = got_diff_blob(blob1, blob2, label1, label2, diff_context,
290 	    outfile);
291 
292 done:
293 	if (obj1)
294 		got_object_close(obj1);
295 	if (obj2)
296 		got_object_close(obj2);
297 	if (blob1)
298 		got_object_blob_close(blob1);
299 	if (blob2)
300 		got_object_blob_close(blob2);
301 	return err;
302 }
303 
304 static const struct got_error *
305 diff_deleted_blob(struct got_object_id *id, const char *label,
306     int diff_context, struct got_repository *repo, FILE *outfile)
307 {
308 	const struct got_error *err;
309 	struct got_blob_object  *blob = NULL;
310 	struct got_object *obj = NULL;
311 
312 	err = got_object_open(&obj, repo, id);
313 	if (err)
314 		return err;
315 
316 	err = got_object_blob_open(&blob, repo, obj, 8192);
317 	if (err)
318 		goto done;
319 	err = got_diff_blob(blob, NULL, label, NULL, diff_context, outfile);
320 done:
321 	got_object_close(obj);
322 	if (blob)
323 		got_object_blob_close(blob);
324 	return err;
325 }
326 
327 static const struct got_error *
328 diff_added_tree(struct got_object_id *id, const char *label,
329     int diff_context, struct got_repository *repo, FILE *outfile)
330 {
331 	const struct got_error *err = NULL;
332 	struct got_object *treeobj = NULL;
333 	struct got_tree_object *tree = NULL;
334 
335 	err = got_object_open(&treeobj, repo, id);
336 	if (err)
337 		goto done;
338 
339 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
340 		err = got_error(GOT_ERR_OBJ_TYPE);
341 		goto done;
342 	}
343 
344 	err = got_object_tree_open(&tree, repo, treeobj);
345 	if (err)
346 		goto done;
347 
348 	err = got_diff_tree(NULL, tree, NULL, label, diff_context, repo,
349 	    outfile);
350 
351 done:
352 	if (tree)
353 		got_object_tree_close(tree);
354 	if (treeobj)
355 		got_object_close(treeobj);
356 	return err;
357 }
358 
359 static const struct got_error *
360 diff_modified_tree(struct got_object_id *id1, struct got_object_id *id2,
361     const char *label1, const char *label2, int diff_context,
362     struct got_repository *repo, FILE *outfile)
363 {
364 	const struct got_error *err;
365 	struct got_object *treeobj1 = NULL;
366 	struct got_object *treeobj2 = NULL;
367 	struct got_tree_object *tree1 = NULL;
368 	struct got_tree_object *tree2 = NULL;
369 
370 	err = got_object_open(&treeobj1, repo, id1);
371 	if (err)
372 		goto done;
373 
374 	if (treeobj1->type != GOT_OBJ_TYPE_TREE) {
375 		err = got_error(GOT_ERR_OBJ_TYPE);
376 		goto done;
377 	}
378 
379 	err = got_object_open(&treeobj2, repo, id2);
380 	if (err)
381 		goto done;
382 
383 	if (treeobj2->type != GOT_OBJ_TYPE_TREE) {
384 		err = got_error(GOT_ERR_OBJ_TYPE);
385 		goto done;
386 	}
387 
388 	err = got_object_tree_open(&tree1, repo, treeobj1);
389 	if (err)
390 		goto done;
391 
392 	err = got_object_tree_open(&tree2, repo, treeobj2);
393 	if (err)
394 		goto done;
395 
396 	err = got_diff_tree(tree1, tree2, label1, label2, diff_context, repo,
397 	    outfile);
398 
399 done:
400 	if (tree1)
401 		got_object_tree_close(tree1);
402 	if (tree2)
403 		got_object_tree_close(tree2);
404 	if (treeobj1)
405 		got_object_close(treeobj1);
406 	if (treeobj2)
407 		got_object_close(treeobj2);
408 	return err;
409 }
410 
411 static const struct got_error *
412 diff_deleted_tree(struct got_object_id *id, const char *label,
413     int diff_context, struct got_repository *repo, FILE *outfile)
414 {
415 	const struct got_error *err;
416 	struct got_object *treeobj = NULL;
417 	struct got_tree_object *tree = NULL;
418 
419 	err = got_object_open(&treeobj, repo, id);
420 	if (err)
421 		goto done;
422 
423 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
424 		err = got_error(GOT_ERR_OBJ_TYPE);
425 		goto done;
426 	}
427 
428 	err = got_object_tree_open(&tree, repo, treeobj);
429 	if (err)
430 		goto done;
431 
432 	err = got_diff_tree(tree, NULL, label, NULL, diff_context, repo,
433 	    outfile);
434 done:
435 	if (tree)
436 		got_object_tree_close(tree);
437 	if (treeobj)
438 		got_object_close(treeobj);
439 	return err;
440 }
441 
442 static const struct got_error *
443 diff_kind_mismatch(struct got_object_id *id1, struct got_object_id *id2,
444     const char *label1, const char *label2, FILE *outfile)
445 {
446 	/* XXX TODO */
447 	return NULL;
448 }
449 
450 static const struct got_error *
451 diff_entry_old_new(struct got_tree_entry *te1, struct got_tree_entry *te2,
452     const char *label1, const char *label2, int diff_context,
453     struct got_repository *repo, FILE *outfile)
454 {
455 	const struct got_error *err = NULL;
456 	int id_match;
457 
458 	if (te2 == NULL) {
459 		if (S_ISDIR(te1->mode))
460 			err = diff_deleted_tree(te1->id, label1, diff_context,
461 			    repo, outfile);
462 		else
463 			err = diff_deleted_blob(te1->id, label1, diff_context,
464 			    repo, outfile);
465 		return err;
466 	}
467 
468 	id_match = (got_object_id_cmp(te1->id, te2->id) == 0);
469 	if (S_ISDIR(te1->mode) && S_ISDIR(te2->mode)) {
470 		if (!id_match)
471 			return diff_modified_tree(te1->id, te2->id,
472 			    label1, label2, diff_context, repo, outfile);
473 	} else if (S_ISREG(te1->mode) && S_ISREG(te2->mode)) {
474 		if (!id_match)
475 			return diff_modified_blob(te1->id, te2->id,
476 			    label1, label2, diff_context, repo, outfile);
477 	}
478 
479 	if (id_match)
480 		return NULL;
481 
482 	return diff_kind_mismatch(te1->id, te2->id, label1, label2, outfile);
483 }
484 
485 static const struct got_error *
486 diff_entry_new_old(struct got_tree_entry *te2, struct got_tree_entry *te1,
487     const char *label2, int diff_context, struct got_repository *repo,
488     FILE *outfile)
489 {
490 	if (te1 != NULL) /* handled by diff_entry_old_new() */
491 		return NULL;
492 
493 	if (S_ISDIR(te2->mode))
494 		return diff_added_tree(te2->id, label2, diff_context, repo,
495 		    outfile);
496 
497 	return diff_added_blob(te2->id, label2, diff_context, repo, outfile);
498 }
499 
500 const struct got_error *
501 got_diff_tree(struct got_tree_object *tree1, struct got_tree_object *tree2,
502     const char *label1, const char *label2, int diff_context,
503     struct got_repository *repo, FILE *outfile)
504 {
505 	const struct got_error *err = NULL;
506 	struct got_tree_entry *te1 = NULL;
507 	struct got_tree_entry *te2 = NULL;
508 	char *l1 = NULL, *l2 = NULL;
509 
510 	if (tree1) {
511 		const struct got_tree_entries *entries;
512 		entries = got_object_tree_get_entries(tree1);
513 		te1 = SIMPLEQ_FIRST(&entries->head);
514 		if (te1 && asprintf(&l1, "%s%s%s", label1, label1[0] ? "/" : "",
515 		    te1->name) == -1)
516 			return got_error_prefix_errno("asprintf");
517 	}
518 	if (tree2) {
519 		const struct got_tree_entries *entries;
520 		entries = got_object_tree_get_entries(tree2);
521 		te2 = SIMPLEQ_FIRST(&entries->head);
522 		if (te2 && asprintf(&l2, "%s%s%s", label2, label2[0] ? "/" : "",
523 		    te2->name) == -1)
524 			return got_error_prefix_errno("asprintf");
525 	}
526 
527 	do {
528 		if (te1) {
529 			struct got_tree_entry *te = NULL;
530 			if (tree2)
531 				te = match_entry_by_name(te1, tree2);
532 			if (te) {
533 				free(l2);
534 				l2 = NULL;
535 				if (te && asprintf(&l2, "%s%s%s", label2,
536 				    label2[0] ? "/" : "", te->name) == -1)
537 					return
538 					    got_error_prefix_errno("asprintf");
539 			}
540 			err = diff_entry_old_new(te1, te, l1, l2, diff_context,
541 			    repo, outfile);
542 			if (err)
543 				break;
544 		}
545 
546 		if (te2) {
547 			struct got_tree_entry *te = NULL;
548 			if (tree1)
549 				te = match_entry_by_name(te2, tree1);
550 			free(l2);
551 			if (te) {
552 				if (asprintf(&l2, "%s%s%s", label2,
553 				    label2[0] ? "/" : "", te->name) == -1)
554 					return
555 					    got_error_prefix_errno("asprintf");
556 			} else {
557 				if (asprintf(&l2, "%s%s%s", label2,
558 				    label2[0] ? "/" : "", te2->name) == -1)
559 					return
560 					    got_error_prefix_errno("asprintf");
561 			}
562 			err = diff_entry_new_old(te2, te, l2, diff_context,
563 			    repo, outfile);
564 			if (err)
565 				break;
566 		}
567 
568 		free(l1);
569 		l1 = NULL;
570 		if (te1) {
571 			te1 = SIMPLEQ_NEXT(te1, entry);
572 			if (te1 &&
573 			    asprintf(&l1, "%s%s%s", label1,
574 			    label1[0] ? "/" : "", te1->name) == -1)
575 				return got_error_prefix_errno("asprintf");
576 		}
577 		free(l2);
578 		l2 = NULL;
579 		if (te2) {
580 			te2 = SIMPLEQ_NEXT(te2, entry);
581 			if (te2 &&
582 			    asprintf(&l2, "%s%s%s", label2,
583 			        label2[0] ? "/" : "", te2->name) == -1)
584 				return got_error_prefix_errno("asprintf");
585 		}
586 	} while (te1 || te2);
587 
588 	return err;
589 }
590 
591 const struct got_error *
592 got_diff_objects_as_blobs(struct got_object_id *id1, struct got_object_id *id2,
593     const char *label1, const char *label2, int diff_context,
594     struct got_repository *repo, FILE *outfile)
595 {
596 	const struct got_error *err;
597 	struct got_blob_object *blob1 = NULL, *blob2 = NULL;
598 
599 	if (id1 == NULL && id2 == NULL)
600 		return got_error(GOT_ERR_NO_OBJ);
601 
602 	if (id1) {
603 		err = got_object_open_as_blob(&blob1, repo, id1, 8192);
604 		if (err)
605 			goto done;
606 	}
607 	if (id2) {
608 		err = got_object_open_as_blob(&blob2, repo, id2, 8192);
609 		if (err)
610 			goto done;
611 	}
612 	err = got_diff_blob(blob1, blob2, label1, label2, diff_context,
613 	    outfile);
614 done:
615 	if (blob1)
616 		got_object_blob_close(blob1);
617 	if (blob2)
618 		got_object_blob_close(blob2);
619 	return err;
620 }
621 
622 const struct got_error *
623 got_diff_objects_as_trees(struct got_object_id *id1, struct got_object_id *id2,
624     char *label1, char *label2, int diff_context, struct got_repository *repo,
625     FILE *outfile)
626 {
627 	const struct got_error *err;
628 	struct got_tree_object *tree1 = NULL, *tree2 = NULL;
629 
630 	if (id1 == NULL && id2 == NULL)
631 		return got_error(GOT_ERR_NO_OBJ);
632 
633 	if (id1) {
634 		err = got_object_open_as_tree(&tree1, repo, id1);
635 		if (err)
636 			goto done;
637 	}
638 	if (id2) {
639 		err = got_object_open_as_tree(&tree2, repo, id2);
640 		if (err)
641 			goto done;
642 	}
643 	err = got_diff_tree(tree1, tree2, label1, label2, diff_context,
644 	   repo, outfile);
645 done:
646 	if (tree1)
647 		got_object_tree_close(tree1);
648 	if (tree2)
649 		got_object_tree_close(tree2);
650 	return err;
651 }
652 
653 const struct got_error *
654 got_diff_objects_as_commits(struct got_object_id *id1,
655     struct got_object_id *id2, int diff_context,
656     struct got_repository *repo, FILE *outfile)
657 {
658 	const struct got_error *err;
659 	struct got_commit_object *commit1 = NULL, *commit2 = NULL;
660 
661 	if (id2 == NULL)
662 		return got_error(GOT_ERR_NO_OBJ);
663 
664 	if (id1) {
665 		err = got_object_open_as_commit(&commit1, repo, id1);
666 		if (err)
667 			goto done;
668 	}
669 
670 	err = got_object_open_as_commit(&commit2, repo, id2);
671 	if (err)
672 		goto done;
673 
674 	err = got_diff_objects_as_trees(
675 	    commit1 ? got_object_commit_get_tree_id(commit1) : NULL,
676 	    got_object_commit_get_tree_id(commit2), "", "", diff_context, repo,
677 	    outfile);
678 done:
679 	if (commit1)
680 		got_object_commit_close(commit1);
681 	if (commit2)
682 		got_object_commit_close(commit2);
683 	return err;
684 }