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_object.h"
28 #include "got_repository.h"
29 #include "got_error.h"
30 #include "got_diff.h"
31 #include "got_path.h"
32 #include "got_cancel.h"
33 #include "got_worktree.h"
34 
35 #include "got_lib_diff.h"
36 #include "got_lib_delta.h"
37 #include "got_lib_inflate.h"
38 #include "got_lib_object.h"
39 
40 static const struct got_error *
41 add_line_offset(off_t **line_offsets, size_t *nlines, off_t off)
42 {
43 	off_t *p;
44 
45 	p = reallocarray(*line_offsets, *nlines + 1, sizeof(off_t));
46 	if (p == NULL)
47 		return got_error_from_errno("reallocarray");
48 	*line_offsets = p;
49 	(*line_offsets)[*nlines] = off;
50 	(*nlines)++;
51 	return NULL;
52 }
53 
54 static const struct got_error *
55 reset_file(FILE *f)
56 {
57 	if (fpurge(f) == EOF)
58 		return got_error_from_errno("fpurge");
59 	if (ftruncate(fileno(f), 0L) == -1)
60 		return got_error_from_errno("ftruncate");
61 	if (fseeko(f, 0L, SEEK_SET) == -1)
62 		return got_error_from_errno("fseeko");
63 	return NULL;
64 }
65 
66 static const struct got_error *
67 diff_blobs(off_t **line_offsets, size_t *nlines,
68     struct got_diffreg_result **resultp, struct got_blob_object *blob1,
69     struct got_blob_object *blob2, FILE *f1, FILE *f2,
70     const char *label1, const char *label2, mode_t mode1, mode_t mode2,
71     int diff_context, int ignore_whitespace, int force_text_diff, FILE *outfile)
72 {
73 	const struct got_error *err = NULL, *free_err;
74 	char hex1[SHA1_DIGEST_STRING_LENGTH];
75 	char hex2[SHA1_DIGEST_STRING_LENGTH];
76 	char *idstr1 = NULL, *idstr2 = NULL;
77 	off_t size1, size2;
78 	struct got_diffreg_result *result;
79 	off_t outoff = 0;
80 	int n;
81 
82 	if (line_offsets && *line_offsets && *nlines > 0)
83 		outoff = (*line_offsets)[*nlines - 1];
84 	else if (line_offsets) {
85 		err = add_line_offset(line_offsets, nlines, 0);
86 		if (err)
87 			goto done;
88 	}
89 
90 	if (resultp)
91 		*resultp = NULL;
92 
93 	if (f1) {
94 		err = reset_file(f1);
95 		if (err)
96 			goto done;
97 	}
98 	if (f2) {
99 		err = reset_file(f2);
100 		if (err)
101 			goto done;
102 	}
103 
104 	size1 = 0;
105 	if (blob1) {
106 		idstr1 = got_object_blob_id_str(blob1, hex1, sizeof(hex1));
107 		err = got_object_blob_dump_to_file(&size1, NULL, NULL, f1,
108 		    blob1);
109 		if (err)
110 			goto done;
111 	} else
112 		idstr1 = "/dev/null";
113 
114 	size2 = 0;
115 	if (blob2) {
116 		idstr2 = got_object_blob_id_str(blob2, hex2, sizeof(hex2));
117 		err = got_object_blob_dump_to_file(&size2, NULL, NULL, f2,
118 		    blob2);
119 		if (err)
120 			goto done;
121 	} else
122 		idstr2 = "/dev/null";
123 
124 	if (outfile) {
125 		char *modestr1 = NULL, *modestr2 = NULL;
126 		int modebits;
127 		if (mode1 && mode1 != mode2) {
128 			if (S_ISLNK(mode1))
129 				modebits = S_IFLNK;
130 			else
131 				modebits = (S_IRWXU | S_IRWXG | S_IRWXO);
132 			if (asprintf(&modestr1, " (mode %o)",
133 			    mode1 & modebits) == -1) {
134 				err = got_error_from_errno("asprintf");
135 				goto done;
136 			}
137 		}
138 		if (mode2 && mode1 != mode2) {
139 			if (S_ISLNK(mode2))
140 				modebits = S_IFLNK;
141 			else
142 				modebits = (S_IRWXU | S_IRWXG | S_IRWXO);
143 			if (asprintf(&modestr2, " (mode %o)",
144 			    mode2 & modebits) == -1) {
145 				err = got_error_from_errno("asprintf");
146 				goto done;
147 			}
148 		}
149 		n = fprintf(outfile, "blob - %s%s\n", idstr1,
150 		    modestr1 ? modestr1 : "");
151 		if (n < 0)
152 			goto done;
153 		outoff += n;
154 		if (line_offsets) {
155 			err = add_line_offset(line_offsets, nlines, outoff);
156 			if (err)
157 				goto done;
158 		}
159 
160 		n = fprintf(outfile, "blob + %s%s\n", idstr2,
161 		    modestr2 ? modestr2 : "");
162 		if (n < 0)
163 			goto done;
164 		outoff += n;
165 		if (line_offsets) {
166 			err = add_line_offset(line_offsets, nlines, outoff);
167 			if (err)
168 				goto done;
169 		}
170 
171 		free(modestr1);
172 		free(modestr2);
173 	}
174 	err = got_diffreg(&result, f1, f2, GOT_DIFF_ALGORITHM_PATIENCE,
175 	    ignore_whitespace, force_text_diff);
176 	if (err)
177 		goto done;
178 
179 	if (outfile) {
180 		err = got_diffreg_output(line_offsets, nlines, result,
181 		    blob1 != NULL, blob2 != NULL,
182 		    label1 ? label1 : idstr1,
183 		    label2 ? label2 : idstr2,
184 		    GOT_DIFF_OUTPUT_UNIDIFF, diff_context, outfile);
185 		if (err)
186 			goto done;
187 	}
188 
189 	if (resultp && err == NULL)
190 		*resultp = result;
191 	else {
192 		free_err = got_diffreg_result_free(result);
193 		if (free_err && err == NULL)
194 			err = free_err;
195 	}
196 done:
197 	return err;
198 }
199 
200 const struct got_error *
201 got_diff_blob_output_unidiff(void *arg, struct got_blob_object *blob1,
202     struct got_blob_object *blob2, FILE *f1, FILE *f2,
203     struct got_object_id *id1, struct got_object_id *id2,
204     const char *label1, const char *label2, mode_t mode1, mode_t mode2,
205     struct got_repository *repo)
206 {
207 	struct got_diff_blob_output_unidiff_arg *a = arg;
208 
209 	return diff_blobs(&a->line_offsets, &a->nlines, NULL,
210 	    blob1, blob2, f1, f2, label1, label2, mode1, mode2, a->diff_context,
211 	    a->ignore_whitespace, a->force_text_diff, a->outfile);
212 }
213 
214 const struct got_error *
215 got_diff_blob(off_t **line_offsets, size_t *nlines,
216     struct got_blob_object *blob1, struct got_blob_object *blob2,
217     FILE *f1, FILE *f2, const char *label1, const char *label2,
218     int diff_context, int ignore_whitespace, int force_text_diff,
219     FILE *outfile)
220 {
221 	return diff_blobs(line_offsets, nlines, NULL, blob1, blob2, f1, f2,
222 	    label1, label2, 0, 0, diff_context, ignore_whitespace,
223 	    force_text_diff, outfile);
224 }
225 
226 static const struct got_error *
227 diff_blob_file(struct got_diffreg_result **resultp,
228     struct got_blob_object *blob1, FILE *f1, off_t size1, const char *label1,
229     FILE *f2, size_t size2, const char *label2, int diff_context,
230     int ignore_whitespace, int force_text_diff, FILE *outfile)
231 {
232 	const struct got_error *err = NULL, *free_err;
233 	char hex1[SHA1_DIGEST_STRING_LENGTH];
234 	char *idstr1 = NULL;
235 	struct got_diffreg_result *result = NULL;
236 
237 	if (resultp)
238 		*resultp = NULL;
239 
240 	if (blob1)
241 		idstr1 = got_object_blob_id_str(blob1, hex1, sizeof(hex1));
242 	else
243 		idstr1 = "/dev/null";
244 
245 	if (outfile) {
246 		fprintf(outfile, "blob - %s\n", label1 ? label1 : idstr1);
247 		fprintf(outfile, "file + %s\n",
248 		    f2 == NULL ? "/dev/null" : label2);
249 	}
250 
251 	err = got_diffreg(&result, f1, f2, GOT_DIFF_ALGORITHM_PATIENCE, 
252 	    ignore_whitespace, force_text_diff);
253 	if (err)
254 		goto done;
255 
256 	if (outfile) {
257 		err = got_diffreg_output(NULL, NULL, result,
258 		    f1 != NULL, f2 != NULL,
259 		    label2, /* show local file's path, not a blob ID */
260 		    label2, GOT_DIFF_OUTPUT_UNIDIFF,
261 		    diff_context, outfile);
262 		if (err)
263 			goto done;
264 	}
265 
266 	if (resultp && err == NULL)
267 		*resultp = result;
268 	else if (result) {
269 		free_err = got_diffreg_result_free(result);
270 		if (free_err && err == NULL)
271 			err = free_err;
272 	}
273 done:
274 	return err;
275 }
276 
277 const struct got_error *
278 got_diff_blob_file(struct got_blob_object *blob1, FILE *f1, off_t size1,
279     const char *label1, FILE *f2, size_t size2, const char *label2,
280     int diff_context, int ignore_whitespace, int force_text_diff, FILE *outfile)
281 {
282 	return diff_blob_file(NULL, blob1, f1, size1, label1, f2, size2, label2,
283 	    diff_context, ignore_whitespace, force_text_diff, outfile);
284 }
285 
286 static const struct got_error *
287 diff_added_blob(struct got_object_id *id, FILE *f, const char *label,
288     mode_t mode, struct got_repository *repo,
289     got_diff_blob_cb cb, void *cb_arg)
290 {
291 	const struct got_error *err;
292 	struct got_blob_object  *blob = NULL;
293 	struct got_object *obj = NULL;
294 
295 	err = got_object_open(&obj, repo, id);
296 	if (err)
297 		return err;
298 
299 	err = got_object_blob_open(&blob, repo, obj, 8192);
300 	if (err)
301 		goto done;
302 	err = cb(cb_arg, NULL, blob, NULL, f, NULL, id,
303 	    NULL, label, 0, mode, repo);
304 done:
305 	got_object_close(obj);
306 	if (blob)
307 		got_object_blob_close(blob);
308 	return err;
309 }
310 
311 static const struct got_error *
312 diff_modified_blob(struct got_object_id *id1, struct got_object_id *id2,
313     FILE *f1, FILE *f2, const char *label1, const char *label2,
314     mode_t mode1, mode_t mode2, struct got_repository *repo,
315     got_diff_blob_cb cb, void *cb_arg)
316 {
317 	const struct got_error *err;
318 	struct got_object *obj1 = NULL;
319 	struct got_object *obj2 = NULL;
320 	struct got_blob_object *blob1 = NULL;
321 	struct got_blob_object *blob2 = NULL;
322 
323 	err = got_object_open(&obj1, repo, id1);
324 	if (err)
325 		return err;
326 	if (obj1->type != GOT_OBJ_TYPE_BLOB) {
327 		err = got_error(GOT_ERR_OBJ_TYPE);
328 		goto done;
329 	}
330 
331 	err = got_object_open(&obj2, repo, id2);
332 	if (err)
333 		goto done;
334 	if (obj2->type != GOT_OBJ_TYPE_BLOB) {
335 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
336 		goto done;
337 	}
338 
339 	err = got_object_blob_open(&blob1, repo, obj1, 8192);
340 	if (err)
341 		goto done;
342 
343 	err = got_object_blob_open(&blob2, repo, obj2, 8192);
344 	if (err)
345 		goto done;
346 
347 	err = cb(cb_arg, blob1, blob2, f1, f2, id1, id2, label1, label2,
348 	    mode1, mode2, repo);
349 done:
350 	if (obj1)
351 		got_object_close(obj1);
352 	if (obj2)
353 		got_object_close(obj2);
354 	if (blob1)
355 		got_object_blob_close(blob1);
356 	if (blob2)
357 		got_object_blob_close(blob2);
358 	return err;
359 }
360 
361 static const struct got_error *
362 diff_deleted_blob(struct got_object_id *id, FILE *f, const char *label,
363     mode_t mode, struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg)
364 {
365 	const struct got_error *err;
366 	struct got_blob_object  *blob = NULL;
367 	struct got_object *obj = NULL;
368 
369 	err = got_object_open(&obj, repo, id);
370 	if (err)
371 		return err;
372 
373 	err = got_object_blob_open(&blob, repo, obj, 8192);
374 	if (err)
375 		goto done;
376 	err = cb(cb_arg, blob, NULL, f, NULL, id, NULL, label, NULL,
377 	    mode, 0, repo);
378 done:
379 	got_object_close(obj);
380 	if (blob)
381 		got_object_blob_close(blob);
382 	return err;
383 }
384 
385 static const struct got_error *
386 diff_added_tree(struct got_object_id *id, FILE *f, const char *label,
387     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
388     int diff_content)
389 {
390 	const struct got_error *err = NULL;
391 	struct got_object *treeobj = NULL;
392 	struct got_tree_object *tree = NULL;
393 
394 	err = got_object_open(&treeobj, repo, id);
395 	if (err)
396 		goto done;
397 
398 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
399 		err = got_error(GOT_ERR_OBJ_TYPE);
400 		goto done;
401 	}
402 
403 	err = got_object_tree_open(&tree, repo, treeobj);
404 	if (err)
405 		goto done;
406 
407 	err = got_diff_tree(NULL, tree, NULL, f, NULL, label,
408 	    repo, cb, cb_arg, diff_content);
409 done:
410 	if (tree)
411 		got_object_tree_close(tree);
412 	if (treeobj)
413 		got_object_close(treeobj);
414 	return err;
415 }
416 
417 static const struct got_error *
418 diff_modified_tree(struct got_object_id *id1, struct got_object_id *id2,
419     FILE *f1, FILE *f2, const char *label1, const char *label2,
420     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
421     int diff_content)
422 {
423 	const struct got_error *err;
424 	struct got_object *treeobj1 = NULL;
425 	struct got_object *treeobj2 = NULL;
426 	struct got_tree_object *tree1 = NULL;
427 	struct got_tree_object *tree2 = NULL;
428 
429 	err = got_object_open(&treeobj1, repo, id1);
430 	if (err)
431 		goto done;
432 
433 	if (treeobj1->type != GOT_OBJ_TYPE_TREE) {
434 		err = got_error(GOT_ERR_OBJ_TYPE);
435 		goto done;
436 	}
437 
438 	err = got_object_open(&treeobj2, repo, id2);
439 	if (err)
440 		goto done;
441 
442 	if (treeobj2->type != GOT_OBJ_TYPE_TREE) {
443 		err = got_error(GOT_ERR_OBJ_TYPE);
444 		goto done;
445 	}
446 
447 	err = got_object_tree_open(&tree1, repo, treeobj1);
448 	if (err)
449 		goto done;
450 
451 	err = got_object_tree_open(&tree2, repo, treeobj2);
452 	if (err)
453 		goto done;
454 
455 	err = got_diff_tree(tree1, tree2, f1, f2, label1, label2,
456 	    repo, cb, cb_arg, diff_content);
457 
458 done:
459 	if (tree1)
460 		got_object_tree_close(tree1);
461 	if (tree2)
462 		got_object_tree_close(tree2);
463 	if (treeobj1)
464 		got_object_close(treeobj1);
465 	if (treeobj2)
466 		got_object_close(treeobj2);
467 	return err;
468 }
469 
470 static const struct got_error *
471 diff_deleted_tree(struct got_object_id *id, FILE *f, const char *label,
472     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
473     int diff_content)
474 {
475 	const struct got_error *err;
476 	struct got_object *treeobj = NULL;
477 	struct got_tree_object *tree = NULL;
478 
479 	err = got_object_open(&treeobj, repo, id);
480 	if (err)
481 		goto done;
482 
483 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
484 		err = got_error(GOT_ERR_OBJ_TYPE);
485 		goto done;
486 	}
487 
488 	err = got_object_tree_open(&tree, repo, treeobj);
489 	if (err)
490 		goto done;
491 
492 	err = got_diff_tree(tree, NULL, f, NULL, label, NULL,
493 	    repo, cb, cb_arg, diff_content);
494 done:
495 	if (tree)
496 		got_object_tree_close(tree);
497 	if (treeobj)
498 		got_object_close(treeobj);
499 	return err;
500 }
501 
502 static const struct got_error *
503 diff_kind_mismatch(struct got_object_id *id1, struct got_object_id *id2,
504     const char *label1, const char *label2, struct got_repository *repo,
505     got_diff_blob_cb cb, void *cb_arg)
506 {
507 	/* XXX TODO */
508 	return NULL;
509 }
510 
511 static const struct got_error *
512 diff_entry_old_new(struct got_tree_entry *te1, struct got_tree_entry *te2,
513     FILE *f1, FILE *f2, const char *label1, const char *label2,
514     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
515     int diff_content)
516 {
517 	const struct got_error *err = NULL;
518 	int id_match;
519 
520 	if (got_object_tree_entry_is_submodule(te1))
521 		return NULL;
522 
523 	if (te2 == NULL) {
524 		if (S_ISDIR(te1->mode))
525 			err = diff_deleted_tree(&te1->id, f1, label1, repo,
526 			    cb, cb_arg, diff_content);
527 		else {
528 			if (diff_content)
529 				err = diff_deleted_blob(&te1->id, f1, label1,
530 				    te1->mode, repo, cb, cb_arg);
531 			else
532 				err = cb(cb_arg, NULL, NULL, NULL, NULL,
533 				    &te1->id, NULL, label1, NULL,
534 				    te1->mode, 0, repo);
535 		}
536 		return err;
537 	} else if (got_object_tree_entry_is_submodule(te2))
538 		return NULL;
539 
540 	id_match = (got_object_id_cmp(&te1->id, &te2->id) == 0);
541 	if (S_ISDIR(te1->mode) && S_ISDIR(te2->mode)) {
542 		if (!id_match)
543 			return diff_modified_tree(&te1->id, &te2->id, f1, f2,
544 			    label1, label2, repo, cb, cb_arg, diff_content);
545 	} else if ((S_ISREG(te1->mode) || S_ISLNK(te1->mode)) &&
546 	    (S_ISREG(te2->mode) || S_ISLNK(te2->mode))) {
547 		if (!id_match ||
548 		    ((te1->mode & (S_IFLNK | S_IXUSR))) !=
549 		    (te2->mode & (S_IFLNK | S_IXUSR))) {
550 			if (diff_content)
551 				return diff_modified_blob(&te1->id, &te2->id,
552 				    f1, f2, label1, label2, te1->mode,
553 				    te2->mode, repo, cb, cb_arg);
554 			else
555 				return cb(cb_arg, NULL, NULL, NULL, NULL,
556 				    &te1->id, &te2->id, label1, label2,
557 				    te1->mode, te2->mode, repo);
558 		}
559 	}
560 
561 	if (id_match)
562 		return NULL;
563 
564 	return diff_kind_mismatch(&te1->id, &te2->id, label1, label2, repo,
565 	    cb, cb_arg);
566 }
567 
568 static const struct got_error *
569 diff_entry_new_old(struct got_tree_entry *te2,
570     struct got_tree_entry *te1, FILE *f2, const char *label2,
571     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
572     int diff_content)
573 {
574 	if (te1 != NULL) /* handled by diff_entry_old_new() */
575 		return NULL;
576 
577 	if (got_object_tree_entry_is_submodule(te2))
578 		return NULL;
579 
580 	if (S_ISDIR(te2->mode))
581 		return diff_added_tree(&te2->id, f2, label2,
582 		    repo, cb, cb_arg, diff_content);
583 
584 	if (diff_content)
585 		return diff_added_blob(&te2->id, f2, label2, te2->mode,
586 		    repo, cb, cb_arg);
587 
588 	return cb(cb_arg, NULL, NULL, NULL, NULL, NULL, &te2->id,
589 	    NULL, label2, 0, te2->mode, repo);
590 }
591 
592 const struct got_error *
593 got_diff_tree_collect_changed_paths(void *arg, struct got_blob_object *blob1,
594     struct got_blob_object *blob2, FILE *f1, FILE *f2,
595     struct got_object_id *id1, struct got_object_id *id2,
596     const char *label1, const char *label2,
597     mode_t mode1, mode_t mode2, struct got_repository *repo)
598 {
599 	const struct got_error *err = NULL;
600 	struct got_pathlist_head *paths = arg;
601 	struct got_diff_changed_path *change = NULL;
602 	char *path = NULL;
603 
604 	path = strdup(label2 ? label2 : label1);
605 	if (path == NULL)
606 		return got_error_from_errno("malloc");
607 
608 	change = malloc(sizeof(*change));
609 	if (change == NULL) {
610 		err = got_error_from_errno("malloc");
611 		goto done;
612 	}
613 
614 	change->status = GOT_STATUS_NO_CHANGE;
615 	if (id1 == NULL)
616 		change->status = GOT_STATUS_ADD;
617 	else if (id2 == NULL)
618 		change->status = GOT_STATUS_DELETE;
619 	else {
620 		if (got_object_id_cmp(id1, id2) != 0)
621 			change->status = GOT_STATUS_MODIFY;
622 		else if (mode1 != mode2)
623 			change->status = GOT_STATUS_MODE_CHANGE;
624 	}
625 
626 	err = got_pathlist_append(paths, path, change);
627 done:
628 	if (err) {
629 		free(path);
630 		free(change);
631 	}
632 	return err;
633 }
634 
635 const struct got_error *
636 got_diff_tree(struct got_tree_object *tree1, struct got_tree_object *tree2,
637     FILE *f1, FILE *f2, const char *label1, const char *label2,
638     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
639     int diff_content)
640 {
641 	const struct got_error *err = NULL;
642 	struct got_tree_entry *te1 = NULL;
643 	struct got_tree_entry *te2 = NULL;
644 	char *l1 = NULL, *l2 = NULL;
645 	int tidx1 = 0, tidx2 = 0;
646 
647 	if (tree1) {
648 		te1 = got_object_tree_get_entry(tree1, 0);
649 		if (te1 && asprintf(&l1, "%s%s%s", label1, label1[0] ? "/" : "",
650 		    te1->name) == -1)
651 			return got_error_from_errno("asprintf");
652 	}
653 	if (tree2) {
654 		te2 = got_object_tree_get_entry(tree2, 0);
655 		if (te2 && asprintf(&l2, "%s%s%s", label2, label2[0] ? "/" : "",
656 		    te2->name) == -1)
657 			return got_error_from_errno("asprintf");
658 	}
659 
660 	do {
661 		if (te1) {
662 			struct got_tree_entry *te = NULL;
663 			if (tree2)
664 				te = got_object_tree_find_entry(tree2,
665 				    te1->name);
666 			if (te) {
667 				free(l2);
668 				l2 = NULL;
669 				if (te && asprintf(&l2, "%s%s%s", label2,
670 				    label2[0] ? "/" : "", te->name) == -1)
671 					return
672 					    got_error_from_errno("asprintf");
673 			}
674 			err = diff_entry_old_new(te1, te, f1, f2, l1, l2,
675 			    repo, cb, cb_arg, diff_content);
676 			if (err)
677 				break;
678 		}
679 
680 		if (te2) {
681 			struct got_tree_entry *te = NULL;
682 			if (tree1)
683 				te = got_object_tree_find_entry(tree1,
684 				    te2->name);
685 			free(l2);
686 			if (te) {
687 				if (asprintf(&l2, "%s%s%s", label2,
688 				    label2[0] ? "/" : "", te->name) == -1)
689 					return
690 					    got_error_from_errno("asprintf");
691 			} else {
692 				if (asprintf(&l2, "%s%s%s", label2,
693 				    label2[0] ? "/" : "", te2->name) == -1)
694 					return
695 					    got_error_from_errno("asprintf");
696 			}
697 			err = diff_entry_new_old(te2, te, f2, l2, repo,
698 			    cb, cb_arg, diff_content);
699 			if (err)
700 				break;
701 		}
702 
703 		free(l1);
704 		l1 = NULL;
705 		if (te1) {
706 			tidx1++;
707 			te1 = got_object_tree_get_entry(tree1, tidx1);
708 			if (te1 &&
709 			    asprintf(&l1, "%s%s%s", label1,
710 			    label1[0] ? "/" : "", te1->name) == -1)
711 				return got_error_from_errno("asprintf");
712 		}
713 		free(l2);
714 		l2 = NULL;
715 		if (te2) {
716 			tidx2++;
717 			te2 = got_object_tree_get_entry(tree2, tidx2);
718 			if (te2 &&
719 			    asprintf(&l2, "%s%s%s", label2,
720 			        label2[0] ? "/" : "", te2->name) == -1)
721 				return got_error_from_errno("asprintf");
722 		}
723 	} while (te1 || te2);
724 
725 	return err;
726 }
727 
728 const struct got_error *
729 got_diff_objects_as_blobs(off_t **line_offsets, size_t *nlines,
730     FILE *f1, FILE *f2, struct got_object_id *id1, struct got_object_id *id2,
731     const char *label1, const char *label2, int diff_context,
732     int ignore_whitespace, int force_text_diff,
733     struct got_repository *repo, FILE *outfile)
734 {
735 	const struct got_error *err;
736 	struct got_blob_object *blob1 = NULL, *blob2 = NULL;
737 
738 	if (id1 == NULL && id2 == NULL)
739 		return got_error(GOT_ERR_NO_OBJ);
740 
741 	if (id1) {
742 		err = got_object_open_as_blob(&blob1, repo, id1, 8192);
743 		if (err)
744 			goto done;
745 	}
746 	if (id2) {
747 		err = got_object_open_as_blob(&blob2, repo, id2, 8192);
748 		if (err)
749 			goto done;
750 	}
751 	err = got_diff_blob(line_offsets, nlines, blob1, blob2, f1, f2,
752 	    label1, label2, diff_context, ignore_whitespace, force_text_diff,
753 	    outfile);
754 done:
755 	if (blob1)
756 		got_object_blob_close(blob1);
757 	if (blob2)
758 		got_object_blob_close(blob2);
759 	return err;
760 }
761 
762 static const struct got_error *
763 diff_paths(struct got_tree_object *tree1, struct got_tree_object *tree2,
764     FILE *f1, FILE *f2, struct got_pathlist_head *paths,
765     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg)
766 {
767 	const struct got_error *err = NULL;
768 	struct got_pathlist_entry *pe;
769 	struct got_object_id *id1 = NULL, *id2 = NULL;
770 	struct got_tree_object *subtree1 = NULL, *subtree2 = NULL;
771 	struct got_blob_object *blob1 = NULL, *blob2 = NULL;
772 
773 	TAILQ_FOREACH(pe, paths, entry) {
774 		int type1 = GOT_OBJ_TYPE_ANY, type2 = GOT_OBJ_TYPE_ANY;
775 		mode_t mode1 = 0, mode2 = 0;
776 
777 		free(id1);
778 		id1 = NULL;
779 		free(id2);
780 		id2 = NULL;
781 		if (subtree1) {
782 			got_object_tree_close(subtree1);
783 			subtree1 = NULL;
784 		}
785 		if (subtree2) {
786 			got_object_tree_close(subtree2);
787 			subtree2 = NULL;
788 		}
789 		if (blob1) {
790 			got_object_blob_close(blob1);
791 			blob1 = NULL;
792 		}
793 		if (blob2) {
794 			got_object_blob_close(blob2);
795 			blob2 = NULL;
796 		}
797 
798 		err = got_object_tree_find_path(&id1, &mode1, repo, tree1,
799 		    pe->path);
800 		if (err && err->code != GOT_ERR_NO_TREE_ENTRY)
801 			goto done;
802 		err = got_object_tree_find_path(&id2, &mode2, repo, tree2,
803 		    pe->path);
804 		if (err && err->code != GOT_ERR_NO_TREE_ENTRY)
805 			goto done;
806 		if (id1 == NULL && id2 == NULL) {
807 			err = got_error_path(pe->path, GOT_ERR_NO_TREE_ENTRY);
808 			goto done;
809 		}
810 		if (id1) {
811 			err = got_object_get_type(&type1, repo, id1);
812 			if (err)
813 				goto done;
814 		}
815 		if (id2) {
816 			err = got_object_get_type(&type2, repo, id2);
817 			if (err)
818 				goto done;
819 		}
820 		if (type1 == GOT_OBJ_TYPE_ANY &&
821 		    type2 == GOT_OBJ_TYPE_ANY) {
822 			err = got_error_path(pe->path, GOT_ERR_NO_OBJ);
823 			goto done;
824 		} else if (type1 != GOT_OBJ_TYPE_ANY &&
825 		    type2 != GOT_OBJ_TYPE_ANY && type1 != type2) {
826 			err = got_error(GOT_ERR_OBJ_TYPE);
827 			goto done;
828 		}
829 
830 		if (type1 == GOT_OBJ_TYPE_BLOB ||
831 		    type2 == GOT_OBJ_TYPE_BLOB) {
832 			if (id1) {
833 				err = got_object_open_as_blob(&blob1, repo,
834 				    id1, 8192);
835 				if (err)
836 					goto done;
837 			}
838 			if (id2) {
839 				err = got_object_open_as_blob(&blob2, repo,
840 				    id2, 8192);
841 				if (err)
842 					goto done;
843 			}
844 			err = cb(cb_arg, blob1, blob2, f1, f2, id1, id2,
845 			    id1 ? pe->path : "/dev/null",
846 			    id2 ? pe->path : "/dev/null",
847 			    mode1, mode2, repo);
848 			if (err)
849 				goto done;
850 		} else if (type1 == GOT_OBJ_TYPE_TREE ||
851 		    type2 == GOT_OBJ_TYPE_TREE) {
852 			if (id1) {
853 				err = got_object_open_as_tree(&subtree1, repo,
854 				    id1);
855 				if (err)
856 					goto done;
857 			}
858 			if (id2) {
859 				err = got_object_open_as_tree(&subtree2, repo,
860 				    id2);
861 				if (err)
862 					goto done;
863 			}
864 			err = got_diff_tree(subtree1, subtree2, f1, f2,
865 			    id1 ? pe->path : "/dev/null",
866 			    id2 ? pe->path : "/dev/null",
867 			    repo, cb, cb_arg, 1);
868 			if (err)
869 				goto done;
870 		} else {
871 			err = got_error(GOT_ERR_OBJ_TYPE);
872 			goto done;
873 		}
874 	}
875 done:
876 	free(id1);
877 	free(id2);
878 	if (subtree1)
879 		got_object_tree_close(subtree1);
880 	if (subtree2)
881 		got_object_tree_close(subtree2);
882 	if (blob1)
883 		got_object_blob_close(blob1);
884 	if (blob2)
885 		got_object_blob_close(blob2);
886 	return err;
887 }
888 
889 const struct got_error *
890 got_diff_objects_as_trees(off_t **line_offsets, size_t *nlines,
891     FILE *f1, FILE *f2, struct got_object_id *id1, struct got_object_id *id2,
892     struct got_pathlist_head *paths,
893     char *label1, char *label2, int diff_context, int ignore_whitespace,
894     int force_text_diff, struct got_repository *repo, FILE *outfile)
895 {
896 	const struct got_error *err;
897 	struct got_tree_object *tree1 = NULL, *tree2 = NULL;
898 	struct got_diff_blob_output_unidiff_arg arg;
899 	int want_lineoffsets = (line_offsets != NULL && *line_offsets != NULL);
900 
901 	if (id1 == NULL && id2 == NULL)
902 		return got_error(GOT_ERR_NO_OBJ);
903 
904 	if (id1) {
905 		err = got_object_open_as_tree(&tree1, repo, id1);
906 		if (err)
907 			goto done;
908 	}
909 	if (id2) {
910 		err = got_object_open_as_tree(&tree2, repo, id2);
911 		if (err)
912 			goto done;
913 	}
914 
915 	arg.diff_context = diff_context;
916 	arg.ignore_whitespace = ignore_whitespace;
917 	arg.force_text_diff = force_text_diff;
918 	arg.outfile = outfile;
919 	if (want_lineoffsets) {
920 		arg.line_offsets = *line_offsets;
921 		arg.nlines = *nlines;
922 	} else {
923 		arg.line_offsets = NULL;
924 		arg.nlines = 0;
925 	}
926 	if (paths == NULL || TAILQ_EMPTY(paths)) {
927 		err = got_diff_tree(tree1, tree2, f1, f2, label1, label2,
928 		    repo, got_diff_blob_output_unidiff, &arg, 1);
929 	} else {
930 		err = diff_paths(tree1, tree2, f1, f2, paths, repo,
931 		    got_diff_blob_output_unidiff, &arg);
932 	}
933 	if (want_lineoffsets) {
934 		*line_offsets = arg.line_offsets; /* was likely re-allocated */
935 		*nlines = arg.nlines;
936 	}
937 done:
938 	if (tree1)
939 		got_object_tree_close(tree1);
940 	if (tree2)
941 		got_object_tree_close(tree2);
942 	return err;
943 }
944 
945 const struct got_error *
946 got_diff_objects_as_commits(off_t **line_offsets, size_t *nlines,
947     FILE *f1, FILE *f2, struct got_object_id *id1, struct got_object_id *id2,
948     struct got_pathlist_head *paths,
949     int diff_context, int ignore_whitespace, int force_text_diff,
950     struct got_repository *repo, FILE *outfile)
951 {
952 	const struct got_error *err;
953 	struct got_commit_object *commit1 = NULL, *commit2 = NULL;
954 
955 	if (id2 == NULL)
956 		return got_error(GOT_ERR_NO_OBJ);
957 
958 	if (id1) {
959 		err = got_object_open_as_commit(&commit1, repo, id1);
960 		if (err)
961 			goto done;
962 	}
963 
964 	err = got_object_open_as_commit(&commit2, repo, id2);
965 	if (err)
966 		goto done;
967 
968 	err = got_diff_objects_as_trees(line_offsets, nlines, f1, f2,
969 	    commit1 ? got_object_commit_get_tree_id(commit1) : NULL,
970 	    got_object_commit_get_tree_id(commit2), paths, "", "",
971 	    diff_context, ignore_whitespace, force_text_diff, repo, outfile);
972 done:
973 	if (commit1)
974 		got_object_commit_close(commit1);
975 	if (commit2)
976 		got_object_commit_close(commit2);
977 	return err;
978 }
979 
980 const struct got_error *
981 got_diff_files(struct got_diffreg_result **resultp,
982     FILE *f1, const char *label1, FILE *f2, const char *label2,
983     int diff_context, int ignore_whitespace, int force_text_diff,
984     FILE *outfile)
985 {
986 	const struct got_error *err = NULL;
987 	struct got_diffreg_result *diffreg_result = NULL;
988 
989 	if (resultp)
990 		*resultp = NULL;
991 
992 	if (outfile) {
993 		fprintf(outfile, "file - %s\n",
994 		    f1 == NULL ? "/dev/null" : label1);
995 		fprintf(outfile, "file + %s\n",
996 		    f2 == NULL ? "/dev/null" : label2);
997 	}
998 
999 	err = got_diffreg(&diffreg_result, f1, f2, GOT_DIFF_ALGORITHM_PATIENCE,
1000 	    ignore_whitespace, force_text_diff);
1001 	if (err)
1002 		goto done;
1003 
1004 	if (outfile) {
1005 		err = got_diffreg_output(NULL, NULL, diffreg_result,
1006 		    f1 != NULL, f2 != NULL, label1, label2,
1007 		    GOT_DIFF_OUTPUT_UNIDIFF, diff_context, outfile);
1008 		if (err)
1009 			goto done;
1010 	}
1011 
1012 done:
1013 	if (resultp && err == NULL)
1014 		*resultp = diffreg_result;
1015 	else if (diffreg_result) {
1016 		const struct got_error *free_err;
1017 		free_err = got_diffreg_result_free(diffreg_result);
1018 		if (free_err && err == NULL)
1019 			err = free_err;
1020 	}
1021 
1022 	return err;
1023 }