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 #include "got_opentemp.h"
35 
36 #include "got_lib_diff.h"
37 #include "got_lib_delta.h"
38 #include "got_lib_inflate.h"
39 #include "got_lib_object.h"
40 
41 static const struct got_error *
42 add_line_offset(off_t **line_offsets, size_t *nlines, off_t off)
43 {
44 	off_t *p;
45 
46 	p = reallocarray(*line_offsets, *nlines + 1, sizeof(off_t));
47 	if (p == NULL)
48 		return got_error_from_errno("reallocarray");
49 	*line_offsets = p;
50 	(*line_offsets)[*nlines] = off;
51 	(*nlines)++;
52 	return NULL;
53 }
54 
55 static const struct got_error *
56 diff_blobs(off_t **line_offsets, size_t *nlines,
57     struct got_diffreg_result **resultp, struct got_blob_object *blob1,
58     struct got_blob_object *blob2, FILE *f1, FILE *f2,
59     const char *label1, const char *label2, mode_t mode1, mode_t mode2,
60     int diff_context, int ignore_whitespace, int force_text_diff, FILE *outfile)
61 {
62 	const struct got_error *err = NULL, *free_err;
63 	char hex1[SHA1_DIGEST_STRING_LENGTH];
64 	char hex2[SHA1_DIGEST_STRING_LENGTH];
65 	const char *idstr1 = NULL, *idstr2 = NULL;
66 	off_t size1, size2;
67 	struct got_diffreg_result *result;
68 	off_t outoff = 0;
69 	int n;
70 
71 	if (line_offsets && *line_offsets && *nlines > 0)
72 		outoff = (*line_offsets)[*nlines - 1];
73 	else if (line_offsets) {
74 		err = add_line_offset(line_offsets, nlines, 0);
75 		if (err)
76 			goto done;
77 	}
78 
79 	if (resultp)
80 		*resultp = NULL;
81 
82 	if (f1) {
83 		err = got_opentemp_truncate(f1);
84 		if (err)
85 			goto done;
86 	}
87 	if (f2) {
88 		err = got_opentemp_truncate(f2);
89 		if (err)
90 			goto done;
91 	}
92 
93 	size1 = 0;
94 	if (blob1) {
95 		idstr1 = got_object_blob_id_str(blob1, hex1, sizeof(hex1));
96 		err = got_object_blob_dump_to_file(&size1, NULL, NULL, f1,
97 		    blob1);
98 		if (err)
99 			goto done;
100 	} else
101 		idstr1 = "/dev/null";
102 
103 	size2 = 0;
104 	if (blob2) {
105 		idstr2 = got_object_blob_id_str(blob2, hex2, sizeof(hex2));
106 		err = got_object_blob_dump_to_file(&size2, NULL, NULL, f2,
107 		    blob2);
108 		if (err)
109 			goto done;
110 	} else
111 		idstr2 = "/dev/null";
112 
113 	if (outfile) {
114 		char *modestr1 = NULL, *modestr2 = NULL;
115 		int modebits;
116 		if (mode1 && mode1 != mode2) {
117 			if (S_ISLNK(mode1))
118 				modebits = S_IFLNK;
119 			else
120 				modebits = (S_IRWXU | S_IRWXG | S_IRWXO);
121 			if (asprintf(&modestr1, " (mode %o)",
122 			    mode1 & modebits) == -1) {
123 				err = got_error_from_errno("asprintf");
124 				goto done;
125 			}
126 		}
127 		if (mode2 && mode1 != mode2) {
128 			if (S_ISLNK(mode2))
129 				modebits = S_IFLNK;
130 			else
131 				modebits = (S_IRWXU | S_IRWXG | S_IRWXO);
132 			if (asprintf(&modestr2, " (mode %o)",
133 			    mode2 & modebits) == -1) {
134 				err = got_error_from_errno("asprintf");
135 				goto done;
136 			}
137 		}
138 		n = fprintf(outfile, "blob - %s%s\n", idstr1,
139 		    modestr1 ? modestr1 : "");
140 		if (n < 0)
141 			goto done;
142 		outoff += n;
143 		if (line_offsets) {
144 			err = add_line_offset(line_offsets, nlines, outoff);
145 			if (err)
146 				goto done;
147 		}
148 
149 		n = fprintf(outfile, "blob + %s%s\n", idstr2,
150 		    modestr2 ? modestr2 : "");
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 		free(modestr1);
161 		free(modestr2);
162 	}
163 	err = got_diffreg(&result, f1, f2, GOT_DIFF_ALGORITHM_PATIENCE,
164 	    ignore_whitespace, force_text_diff);
165 	if (err)
166 		goto done;
167 
168 	if (outfile) {
169 		err = got_diffreg_output(line_offsets, nlines, result,
170 		    blob1 != NULL, blob2 != NULL,
171 		    label1 ? label1 : idstr1,
172 		    label2 ? label2 : idstr2,
173 		    GOT_DIFF_OUTPUT_UNIDIFF, diff_context, outfile);
174 		if (err)
175 			goto done;
176 	}
177 
178 	if (resultp && err == NULL)
179 		*resultp = result;
180 	else {
181 		free_err = got_diffreg_result_free(result);
182 		if (free_err && err == NULL)
183 			err = free_err;
184 	}
185 done:
186 	return err;
187 }
188 
189 const struct got_error *
190 got_diff_blob_output_unidiff(void *arg, struct got_blob_object *blob1,
191     struct got_blob_object *blob2, FILE *f1, FILE *f2,
192     struct got_object_id *id1, struct got_object_id *id2,
193     const char *label1, const char *label2, mode_t mode1, mode_t mode2,
194     struct got_repository *repo)
195 {
196 	struct got_diff_blob_output_unidiff_arg *a = arg;
197 
198 	return diff_blobs(&a->line_offsets, &a->nlines, NULL,
199 	    blob1, blob2, f1, f2, label1, label2, mode1, mode2, a->diff_context,
200 	    a->ignore_whitespace, a->force_text_diff, a->outfile);
201 }
202 
203 const struct got_error *
204 got_diff_blob(off_t **line_offsets, size_t *nlines,
205     struct got_blob_object *blob1, struct got_blob_object *blob2,
206     FILE *f1, FILE *f2, const char *label1, const char *label2,
207     int diff_context, int ignore_whitespace, int force_text_diff,
208     FILE *outfile)
209 {
210 	return diff_blobs(line_offsets, nlines, NULL, blob1, blob2, f1, f2,
211 	    label1, label2, 0, 0, diff_context, ignore_whitespace,
212 	    force_text_diff, outfile);
213 }
214 
215 static const struct got_error *
216 diff_blob_file(struct got_diffreg_result **resultp,
217     struct got_blob_object *blob1, FILE *f1, off_t size1, const char *label1,
218     FILE *f2, size_t size2, const char *label2, int diff_context,
219     int ignore_whitespace, int force_text_diff, FILE *outfile)
220 {
221 	const struct got_error *err = NULL, *free_err;
222 	char hex1[SHA1_DIGEST_STRING_LENGTH];
223 	const char *idstr1 = NULL;
224 	struct got_diffreg_result *result = NULL;
225 
226 	if (resultp)
227 		*resultp = NULL;
228 
229 	if (blob1)
230 		idstr1 = got_object_blob_id_str(blob1, hex1, sizeof(hex1));
231 	else
232 		idstr1 = "/dev/null";
233 
234 	if (outfile) {
235 		fprintf(outfile, "blob - %s\n", label1 ? label1 : idstr1);
236 		fprintf(outfile, "file + %s\n",
237 		    f2 == NULL ? "/dev/null" : label2);
238 	}
239 
240 	err = got_diffreg(&result, f1, f2, GOT_DIFF_ALGORITHM_PATIENCE, 
241 	    ignore_whitespace, force_text_diff);
242 	if (err)
243 		goto done;
244 
245 	if (outfile) {
246 		err = got_diffreg_output(NULL, NULL, result,
247 		    f1 != NULL, f2 != NULL,
248 		    label2, /* show local file's path, not a blob ID */
249 		    label2, GOT_DIFF_OUTPUT_UNIDIFF,
250 		    diff_context, outfile);
251 		if (err)
252 			goto done;
253 	}
254 
255 	if (resultp && err == NULL)
256 		*resultp = result;
257 	else if (result) {
258 		free_err = got_diffreg_result_free(result);
259 		if (free_err && err == NULL)
260 			err = free_err;
261 	}
262 done:
263 	return err;
264 }
265 
266 const struct got_error *
267 got_diff_blob_file(struct got_blob_object *blob1, FILE *f1, off_t size1,
268     const char *label1, FILE *f2, size_t size2, const char *label2,
269     int diff_context, int ignore_whitespace, int force_text_diff, FILE *outfile)
270 {
271 	return diff_blob_file(NULL, blob1, f1, size1, label1, f2, size2, label2,
272 	    diff_context, ignore_whitespace, force_text_diff, outfile);
273 }
274 
275 static const struct got_error *
276 diff_added_blob(struct got_object_id *id, FILE *f, const char *label,
277     mode_t mode, struct got_repository *repo,
278     got_diff_blob_cb cb, void *cb_arg)
279 {
280 	const struct got_error *err;
281 	struct got_blob_object  *blob = NULL;
282 	struct got_object *obj = NULL;
283 	int fd = -1;
284 
285 	err = got_object_open(&obj, repo, id);
286 	if (err)
287 		return err;
288 
289 	fd = got_opentempfd();
290 	if (fd == -1) {
291 		err = got_error_from_errno("got_opentempfd");
292 		goto done;
293 	}
294 
295 	err = got_object_blob_open(&blob, repo, obj, 8192, fd);
296 	if (err)
297 		goto done;
298 	err = cb(cb_arg, NULL, blob, NULL, f, NULL, id,
299 	    NULL, label, 0, mode, repo);
300 done:
301 	got_object_close(obj);
302 	if (fd != -1 && close(fd) == -1 && err == NULL)
303 		err = got_error_from_errno("close");
304 	if (blob)
305 		got_object_blob_close(blob);
306 	return err;
307 }
308 
309 static const struct got_error *
310 diff_modified_blob(struct got_object_id *id1, struct got_object_id *id2,
311     FILE *f1, FILE *f2, const char *label1, const char *label2,
312     mode_t mode1, mode_t mode2, struct got_repository *repo,
313     got_diff_blob_cb cb, void *cb_arg)
314 {
315 	const struct got_error *err;
316 	struct got_object *obj1 = NULL;
317 	struct got_object *obj2 = NULL;
318 	struct got_blob_object *blob1 = NULL;
319 	struct got_blob_object *blob2 = NULL;
320 	int fd1 = -1, fd2 = -1;
321 
322 	err = got_object_open(&obj1, repo, id1);
323 	if (err)
324 		return err;
325 
326 	fd1 = got_opentempfd();
327 	if (fd1 == -1) {
328 		err = got_error_from_errno("got_opentempfd");
329 		goto done;
330 	}
331 	fd2 = got_opentempfd();
332 	if (fd2 == -1) {
333 		err = got_error_from_errno("got_opentempfd");
334 		goto done;
335 	}
336 
337 	if (obj1->type != GOT_OBJ_TYPE_BLOB) {
338 		err = got_error(GOT_ERR_OBJ_TYPE);
339 		goto done;
340 	}
341 
342 	err = got_object_open(&obj2, repo, id2);
343 	if (err)
344 		goto done;
345 	if (obj2->type != GOT_OBJ_TYPE_BLOB) {
346 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
347 		goto done;
348 	}
349 
350 	err = got_object_blob_open(&blob1, repo, obj1, 8192, fd1);
351 	if (err)
352 		goto done;
353 
354 	err = got_object_blob_open(&blob2, repo, obj2, 8192, fd2);
355 	if (err)
356 		goto done;
357 
358 	err = cb(cb_arg, blob1, blob2, f1, f2, id1, id2, label1, label2,
359 	    mode1, mode2, repo);
360 done:
361 	if (obj1)
362 		got_object_close(obj1);
363 	if (obj2)
364 		got_object_close(obj2);
365 	if (fd1 != -1 && close(fd1) == -1 && err == NULL)
366 		err = got_error_from_errno("close");
367 	if (blob1)
368 		got_object_blob_close(blob1);
369 	if (fd2 != -1 && close(fd2) == -1 && err == NULL)
370 		err = got_error_from_errno("close");
371 	if (blob2)
372 		got_object_blob_close(blob2);
373 	return err;
374 }
375 
376 static const struct got_error *
377 diff_deleted_blob(struct got_object_id *id, FILE *f, const char *label,
378     mode_t mode, struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg)
379 {
380 	const struct got_error *err;
381 	struct got_blob_object  *blob = NULL;
382 	struct got_object *obj = NULL;
383 	int fd = -1;
384 
385 	fd = got_opentempfd();
386 	if (fd == -1)
387 		return got_error_from_errno("got_opentempfd");
388 
389 	err = got_object_open(&obj, repo, id);
390 	if (err)
391 		return err;
392 
393 	err = got_object_blob_open(&blob, repo, obj, 8192, fd);
394 	if (err)
395 		goto done;
396 	err = cb(cb_arg, blob, NULL, f, NULL, id, NULL, label, NULL,
397 	    mode, 0, repo);
398 done:
399 	got_object_close(obj);
400 	if (fd != -1 && close(fd) == -1 && err == NULL)
401 		err = got_error_from_errno("close");
402 	if (blob)
403 		got_object_blob_close(blob);
404 	return err;
405 }
406 
407 static const struct got_error *
408 diff_added_tree(struct got_object_id *id, FILE *f, const char *label,
409     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
410     int diff_content)
411 {
412 	const struct got_error *err = NULL;
413 	struct got_object *treeobj = NULL;
414 	struct got_tree_object *tree = NULL;
415 
416 	err = got_object_open(&treeobj, repo, id);
417 	if (err)
418 		goto done;
419 
420 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
421 		err = got_error(GOT_ERR_OBJ_TYPE);
422 		goto done;
423 	}
424 
425 	err = got_object_tree_open(&tree, repo, treeobj);
426 	if (err)
427 		goto done;
428 
429 	err = got_diff_tree(NULL, tree, NULL, f, NULL, label,
430 	    repo, cb, cb_arg, diff_content);
431 done:
432 	if (tree)
433 		got_object_tree_close(tree);
434 	if (treeobj)
435 		got_object_close(treeobj);
436 	return err;
437 }
438 
439 static const struct got_error *
440 diff_modified_tree(struct got_object_id *id1, struct got_object_id *id2,
441     FILE *f1, FILE *f2, const char *label1, const char *label2,
442     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
443     int diff_content)
444 {
445 	const struct got_error *err;
446 	struct got_object *treeobj1 = NULL;
447 	struct got_object *treeobj2 = NULL;
448 	struct got_tree_object *tree1 = NULL;
449 	struct got_tree_object *tree2 = NULL;
450 
451 	err = got_object_open(&treeobj1, repo, id1);
452 	if (err)
453 		goto done;
454 
455 	if (treeobj1->type != GOT_OBJ_TYPE_TREE) {
456 		err = got_error(GOT_ERR_OBJ_TYPE);
457 		goto done;
458 	}
459 
460 	err = got_object_open(&treeobj2, repo, id2);
461 	if (err)
462 		goto done;
463 
464 	if (treeobj2->type != GOT_OBJ_TYPE_TREE) {
465 		err = got_error(GOT_ERR_OBJ_TYPE);
466 		goto done;
467 	}
468 
469 	err = got_object_tree_open(&tree1, repo, treeobj1);
470 	if (err)
471 		goto done;
472 
473 	err = got_object_tree_open(&tree2, repo, treeobj2);
474 	if (err)
475 		goto done;
476 
477 	err = got_diff_tree(tree1, tree2, f1, f2, label1, label2,
478 	    repo, cb, cb_arg, diff_content);
479 
480 done:
481 	if (tree1)
482 		got_object_tree_close(tree1);
483 	if (tree2)
484 		got_object_tree_close(tree2);
485 	if (treeobj1)
486 		got_object_close(treeobj1);
487 	if (treeobj2)
488 		got_object_close(treeobj2);
489 	return err;
490 }
491 
492 static const struct got_error *
493 diff_deleted_tree(struct got_object_id *id, FILE *f, const char *label,
494     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
495     int diff_content)
496 {
497 	const struct got_error *err;
498 	struct got_object *treeobj = NULL;
499 	struct got_tree_object *tree = NULL;
500 
501 	err = got_object_open(&treeobj, repo, id);
502 	if (err)
503 		goto done;
504 
505 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
506 		err = got_error(GOT_ERR_OBJ_TYPE);
507 		goto done;
508 	}
509 
510 	err = got_object_tree_open(&tree, repo, treeobj);
511 	if (err)
512 		goto done;
513 
514 	err = got_diff_tree(tree, NULL, f, NULL, label, NULL,
515 	    repo, cb, cb_arg, diff_content);
516 done:
517 	if (tree)
518 		got_object_tree_close(tree);
519 	if (treeobj)
520 		got_object_close(treeobj);
521 	return err;
522 }
523 
524 static const struct got_error *
525 diff_kind_mismatch(struct got_object_id *id1, struct got_object_id *id2,
526     const char *label1, const char *label2, struct got_repository *repo,
527     got_diff_blob_cb cb, void *cb_arg)
528 {
529 	/* XXX TODO */
530 	return NULL;
531 }
532 
533 static const struct got_error *
534 diff_entry_old_new(struct got_tree_entry *te1, struct got_tree_entry *te2,
535     FILE *f1, FILE *f2, const char *label1, const char *label2,
536     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
537     int diff_content)
538 {
539 	const struct got_error *err = NULL;
540 	int id_match;
541 
542 	if (got_object_tree_entry_is_submodule(te1))
543 		return NULL;
544 
545 	if (te2 == NULL) {
546 		if (S_ISDIR(te1->mode))
547 			err = diff_deleted_tree(&te1->id, f1, label1, repo,
548 			    cb, cb_arg, diff_content);
549 		else {
550 			if (diff_content)
551 				err = diff_deleted_blob(&te1->id, f1, label1,
552 				    te1->mode, repo, cb, cb_arg);
553 			else
554 				err = cb(cb_arg, NULL, NULL, NULL, NULL,
555 				    &te1->id, NULL, label1, NULL,
556 				    te1->mode, 0, repo);
557 		}
558 		return err;
559 	} else if (got_object_tree_entry_is_submodule(te2))
560 		return NULL;
561 
562 	id_match = (got_object_id_cmp(&te1->id, &te2->id) == 0);
563 	if (S_ISDIR(te1->mode) && S_ISDIR(te2->mode)) {
564 		if (!id_match)
565 			return diff_modified_tree(&te1->id, &te2->id, f1, f2,
566 			    label1, label2, repo, cb, cb_arg, diff_content);
567 	} else if ((S_ISREG(te1->mode) || S_ISLNK(te1->mode)) &&
568 	    (S_ISREG(te2->mode) || S_ISLNK(te2->mode))) {
569 		if (!id_match ||
570 		    ((te1->mode & (S_IFLNK | S_IXUSR))) !=
571 		    (te2->mode & (S_IFLNK | S_IXUSR))) {
572 			if (diff_content)
573 				return diff_modified_blob(&te1->id, &te2->id,
574 				    f1, f2, label1, label2, te1->mode,
575 				    te2->mode, repo, cb, cb_arg);
576 			else
577 				return cb(cb_arg, NULL, NULL, NULL, NULL,
578 				    &te1->id, &te2->id, label1, label2,
579 				    te1->mode, te2->mode, repo);
580 		}
581 	}
582 
583 	if (id_match)
584 		return NULL;
585 
586 	return diff_kind_mismatch(&te1->id, &te2->id, label1, label2, repo,
587 	    cb, cb_arg);
588 }
589 
590 static const struct got_error *
591 diff_entry_new_old(struct got_tree_entry *te2,
592     struct got_tree_entry *te1, FILE *f2, const char *label2,
593     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
594     int diff_content)
595 {
596 	if (te1 != NULL) /* handled by diff_entry_old_new() */
597 		return NULL;
598 
599 	if (got_object_tree_entry_is_submodule(te2))
600 		return NULL;
601 
602 	if (S_ISDIR(te2->mode))
603 		return diff_added_tree(&te2->id, f2, label2,
604 		    repo, cb, cb_arg, diff_content);
605 
606 	if (diff_content)
607 		return diff_added_blob(&te2->id, f2, label2, te2->mode,
608 		    repo, cb, cb_arg);
609 
610 	return cb(cb_arg, NULL, NULL, NULL, NULL, NULL, &te2->id,
611 	    NULL, label2, 0, te2->mode, repo);
612 }
613 
614 const struct got_error *
615 got_diff_tree_collect_changed_paths(void *arg, struct got_blob_object *blob1,
616     struct got_blob_object *blob2, FILE *f1, FILE *f2,
617     struct got_object_id *id1, struct got_object_id *id2,
618     const char *label1, const char *label2,
619     mode_t mode1, mode_t mode2, struct got_repository *repo)
620 {
621 	const struct got_error *err = NULL;
622 	struct got_pathlist_head *paths = arg;
623 	struct got_diff_changed_path *change = NULL;
624 	char *path = NULL;
625 
626 	path = strdup(label2 ? label2 : label1);
627 	if (path == NULL)
628 		return got_error_from_errno("malloc");
629 
630 	change = malloc(sizeof(*change));
631 	if (change == NULL) {
632 		err = got_error_from_errno("malloc");
633 		goto done;
634 	}
635 
636 	change->status = GOT_STATUS_NO_CHANGE;
637 	if (id1 == NULL)
638 		change->status = GOT_STATUS_ADD;
639 	else if (id2 == NULL)
640 		change->status = GOT_STATUS_DELETE;
641 	else {
642 		if (got_object_id_cmp(id1, id2) != 0)
643 			change->status = GOT_STATUS_MODIFY;
644 		else if (mode1 != mode2)
645 			change->status = GOT_STATUS_MODE_CHANGE;
646 	}
647 
648 	err = got_pathlist_append(paths, path, change);
649 done:
650 	if (err) {
651 		free(path);
652 		free(change);
653 	}
654 	return err;
655 }
656 
657 const struct got_error *
658 got_diff_tree(struct got_tree_object *tree1, struct got_tree_object *tree2,
659     FILE *f1, FILE *f2, const char *label1, const char *label2,
660     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
661     int diff_content)
662 {
663 	const struct got_error *err = NULL;
664 	struct got_tree_entry *te1 = NULL;
665 	struct got_tree_entry *te2 = NULL;
666 	char *l1 = NULL, *l2 = NULL;
667 	int tidx1 = 0, tidx2 = 0;
668 
669 	if (tree1) {
670 		te1 = got_object_tree_get_entry(tree1, 0);
671 		if (te1 && asprintf(&l1, "%s%s%s", label1, label1[0] ? "/" : "",
672 		    te1->name) == -1)
673 			return got_error_from_errno("asprintf");
674 	}
675 	if (tree2) {
676 		te2 = got_object_tree_get_entry(tree2, 0);
677 		if (te2 && asprintf(&l2, "%s%s%s", label2, label2[0] ? "/" : "",
678 		    te2->name) == -1)
679 			return got_error_from_errno("asprintf");
680 	}
681 
682 	do {
683 		if (te1) {
684 			struct got_tree_entry *te = NULL;
685 			if (tree2)
686 				te = got_object_tree_find_entry(tree2,
687 				    te1->name);
688 			if (te) {
689 				free(l2);
690 				l2 = NULL;
691 				if (te && asprintf(&l2, "%s%s%s", label2,
692 				    label2[0] ? "/" : "", te->name) == -1)
693 					return
694 					    got_error_from_errno("asprintf");
695 			}
696 			err = diff_entry_old_new(te1, te, f1, f2, l1, l2,
697 			    repo, cb, cb_arg, diff_content);
698 			if (err)
699 				break;
700 		}
701 
702 		if (te2) {
703 			struct got_tree_entry *te = NULL;
704 			if (tree1)
705 				te = got_object_tree_find_entry(tree1,
706 				    te2->name);
707 			free(l2);
708 			if (te) {
709 				if (asprintf(&l2, "%s%s%s", label2,
710 				    label2[0] ? "/" : "", te->name) == -1)
711 					return
712 					    got_error_from_errno("asprintf");
713 			} else {
714 				if (asprintf(&l2, "%s%s%s", label2,
715 				    label2[0] ? "/" : "", te2->name) == -1)
716 					return
717 					    got_error_from_errno("asprintf");
718 			}
719 			err = diff_entry_new_old(te2, te, f2, l2, repo,
720 			    cb, cb_arg, diff_content);
721 			if (err)
722 				break;
723 		}
724 
725 		free(l1);
726 		l1 = NULL;
727 		if (te1) {
728 			tidx1++;
729 			te1 = got_object_tree_get_entry(tree1, tidx1);
730 			if (te1 &&
731 			    asprintf(&l1, "%s%s%s", label1,
732 			    label1[0] ? "/" : "", te1->name) == -1)
733 				return got_error_from_errno("asprintf");
734 		}
735 		free(l2);
736 		l2 = NULL;
737 		if (te2) {
738 			tidx2++;
739 			te2 = got_object_tree_get_entry(tree2, tidx2);
740 			if (te2 &&
741 			    asprintf(&l2, "%s%s%s", label2,
742 			        label2[0] ? "/" : "", te2->name) == -1)
743 				return got_error_from_errno("asprintf");
744 		}
745 	} while (te1 || te2);
746 
747 	return err;
748 }
749 
750 const struct got_error *
751 got_diff_objects_as_blobs(off_t **line_offsets, size_t *nlines,
752     FILE *f1, FILE *f2, struct got_object_id *id1, struct got_object_id *id2,
753     const char *label1, const char *label2, int diff_context,
754     int ignore_whitespace, int force_text_diff,
755     struct got_repository *repo, FILE *outfile)
756 {
757 	const struct got_error *err;
758 	struct got_blob_object *blob1 = NULL, *blob2 = NULL;
759 	int fd1 = -1, fd2 = -1;
760 
761 	if (id1 == NULL && id2 == NULL)
762 		return got_error(GOT_ERR_NO_OBJ);
763 
764 	fd1 = got_opentempfd();
765 	if (fd1 == -1)
766 		return got_error_from_errno("got_opentempfd");
767 	fd2 = got_opentempfd();
768 	if (fd2 == -1) {
769 		err = got_error_from_errno("got_opentempfd");
770 		goto done;
771 	}
772 
773 	if (id1) {
774 		err = got_object_open_as_blob(&blob1, repo, id1, 8192, fd1);
775 		if (err)
776 			goto done;
777 	}
778 	if (id2) {
779 		err = got_object_open_as_blob(&blob2, repo, id2, 8192, fd2);
780 		if (err)
781 			goto done;
782 	}
783 	err = got_diff_blob(line_offsets, nlines, blob1, blob2, f1, f2,
784 	    label1, label2, diff_context, ignore_whitespace, force_text_diff,
785 	    outfile);
786 done:
787 	if (fd1 != -1 && close(fd1) == -1 && err == NULL)
788 		err = got_error_from_errno("close");
789 	if (blob1)
790 		got_object_blob_close(blob1);
791 	if (fd2 != -1 && close(fd2) == -1 && err == NULL)
792 		err = got_error_from_errno("close");
793 	if (blob2)
794 		got_object_blob_close(blob2);
795 	return err;
796 }
797 
798 static const struct got_error *
799 diff_paths(struct got_tree_object *tree1, struct got_tree_object *tree2,
800     FILE *f1, FILE *f2, struct got_pathlist_head *paths,
801     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg)
802 {
803 	const struct got_error *err = NULL;
804 	struct got_pathlist_entry *pe;
805 	struct got_object_id *id1 = NULL, *id2 = NULL;
806 	struct got_tree_object *subtree1 = NULL, *subtree2 = NULL;
807 	struct got_blob_object *blob1 = NULL, *blob2 = NULL;
808 	int fd1 = -1, fd2 = -1;
809 
810 	fd1 = got_opentempfd();
811 	if (fd1 == -1)
812 		return got_error_from_errno("got_opentempfd");
813 	fd2 = got_opentempfd();
814 	if (fd2 == -1) {
815 		err = got_error_from_errno("got_opentempfd");
816 		goto done;
817 	}
818 	TAILQ_FOREACH(pe, paths, entry) {
819 		int type1 = GOT_OBJ_TYPE_ANY, type2 = GOT_OBJ_TYPE_ANY;
820 		mode_t mode1 = 0, mode2 = 0;
821 
822 		free(id1);
823 		id1 = NULL;
824 		free(id2);
825 		id2 = NULL;
826 		if (subtree1) {
827 			got_object_tree_close(subtree1);
828 			subtree1 = NULL;
829 		}
830 		if (subtree2) {
831 			got_object_tree_close(subtree2);
832 			subtree2 = NULL;
833 		}
834 		if (blob1) {
835 			got_object_blob_close(blob1);
836 			blob1 = NULL;
837 		}
838 		if (blob2) {
839 			got_object_blob_close(blob2);
840 			blob2 = NULL;
841 		}
842 
843 		err = got_object_tree_find_path(&id1, &mode1, repo, tree1,
844 		    pe->path);
845 		if (err && err->code != GOT_ERR_NO_TREE_ENTRY)
846 			goto done;
847 		err = got_object_tree_find_path(&id2, &mode2, repo, tree2,
848 		    pe->path);
849 		if (err && err->code != GOT_ERR_NO_TREE_ENTRY)
850 			goto done;
851 		if (id1 == NULL && id2 == NULL) {
852 			err = got_error_path(pe->path, GOT_ERR_NO_TREE_ENTRY);
853 			goto done;
854 		}
855 		if (id1) {
856 			err = got_object_get_type(&type1, repo, id1);
857 			if (err)
858 				goto done;
859 		}
860 		if (id2) {
861 			err = got_object_get_type(&type2, repo, id2);
862 			if (err)
863 				goto done;
864 		}
865 		if (type1 == GOT_OBJ_TYPE_ANY &&
866 		    type2 == GOT_OBJ_TYPE_ANY) {
867 			err = got_error_path(pe->path, GOT_ERR_NO_OBJ);
868 			goto done;
869 		} else if (type1 != GOT_OBJ_TYPE_ANY &&
870 		    type2 != GOT_OBJ_TYPE_ANY && type1 != type2) {
871 			err = got_error(GOT_ERR_OBJ_TYPE);
872 			goto done;
873 		}
874 
875 		if (type1 == GOT_OBJ_TYPE_BLOB ||
876 		    type2 == GOT_OBJ_TYPE_BLOB) {
877 			if (id1) {
878 				err = got_object_open_as_blob(&blob1, repo,
879 				    id1, 8192, fd1);
880 				if (err)
881 					goto done;
882 			}
883 			if (id2) {
884 				err = got_object_open_as_blob(&blob2, repo,
885 				    id2, 8192, fd2);
886 				if (err)
887 					goto done;
888 			}
889 			err = cb(cb_arg, blob1, blob2, f1, f2, id1, id2,
890 			    id1 ? pe->path : "/dev/null",
891 			    id2 ? pe->path : "/dev/null",
892 			    mode1, mode2, repo);
893 			if (err)
894 				goto done;
895 		} else if (type1 == GOT_OBJ_TYPE_TREE ||
896 		    type2 == GOT_OBJ_TYPE_TREE) {
897 			if (id1) {
898 				err = got_object_open_as_tree(&subtree1, repo,
899 				    id1);
900 				if (err)
901 					goto done;
902 			}
903 			if (id2) {
904 				err = got_object_open_as_tree(&subtree2, repo,
905 				    id2);
906 				if (err)
907 					goto done;
908 			}
909 			err = got_diff_tree(subtree1, subtree2, f1, f2,
910 			    id1 ? pe->path : "/dev/null",
911 			    id2 ? pe->path : "/dev/null",
912 			    repo, cb, cb_arg, 1);
913 			if (err)
914 				goto done;
915 		} else {
916 			err = got_error(GOT_ERR_OBJ_TYPE);
917 			goto done;
918 		}
919 		if (ftruncate(fd1, 0L) == -1)
920 			return got_error_from_errno("ftruncate");
921 		if (ftruncate(fd2, 0L) == -1)
922 			return got_error_from_errno("ftruncate");
923 	}
924 done:
925 	free(id1);
926 	free(id2);
927 	if (subtree1)
928 		got_object_tree_close(subtree1);
929 	if (subtree2)
930 		got_object_tree_close(subtree2);
931 	if (fd1 != -1 && close(fd1) == -1 && err == NULL)
932 		err = got_error_from_errno("close");
933 	if (blob1)
934 		got_object_blob_close(blob1);
935 	if (fd2 != -1 && close(fd2) == -1 && err == NULL)
936 		err = got_error_from_errno("close");
937 	if (blob2)
938 		got_object_blob_close(blob2);
939 	return err;
940 }
941 
942 static const struct got_error *
943 show_object_id(off_t **line_offsets, size_t *nlines, const char *obj_typestr,
944     int ch, const char *id_str, FILE *outfile)
945 {
946 	const struct got_error *err;
947 	int n;
948 	off_t outoff = 0;
949 
950 	n = fprintf(outfile, "%s %c %s\n", obj_typestr, ch, id_str);
951 	if (line_offsets != NULL && *line_offsets != NULL) {
952 		if (*nlines == 0) {
953 			err = add_line_offset(line_offsets, nlines, 0);
954 			if (err)
955 				return err;
956 		} else
957 			outoff = (*line_offsets)[*nlines - 1];
958 
959 		outoff += n;
960 		err = add_line_offset(line_offsets, nlines, outoff);
961 		if (err)
962 			return err;
963 	}
964 
965 	return NULL;
966 }
967 
968 static const struct got_error *
969 diff_objects_as_trees(off_t **line_offsets, size_t *nlines,
970     FILE *f1, FILE *f2, struct got_object_id *id1, struct got_object_id *id2,
971     struct got_pathlist_head *paths, const char *label1, const char *label2,
972     int diff_context, int ignore_whitespace, int force_text_diff,
973     struct got_repository *repo, FILE *outfile)
974 {
975 	const struct got_error *err;
976 	struct got_tree_object *tree1 = NULL, *tree2 = NULL;
977 	struct got_diff_blob_output_unidiff_arg arg;
978 	int want_lineoffsets = (line_offsets != NULL && *line_offsets != NULL);
979 
980 	if (id1 == NULL && id2 == NULL)
981 		return got_error(GOT_ERR_NO_OBJ);
982 
983 	if (id1) {
984 		err = got_object_open_as_tree(&tree1, repo, id1);
985 		if (err)
986 			goto done;
987 	}
988 	if (id2) {
989 		err = got_object_open_as_tree(&tree2, repo, id2);
990 		if (err)
991 			goto done;
992 	}
993 
994 	arg.diff_context = diff_context;
995 	arg.ignore_whitespace = ignore_whitespace;
996 	arg.force_text_diff = force_text_diff;
997 	arg.outfile = outfile;
998 	if (want_lineoffsets) {
999 		arg.line_offsets = *line_offsets;
1000 		arg.nlines = *nlines;
1001 	} else {
1002 		arg.line_offsets = NULL;
1003 		arg.nlines = 0;
1004 	}
1005 	if (paths == NULL || TAILQ_EMPTY(paths)) {
1006 		err = got_diff_tree(tree1, tree2, f1, f2, label1, label2,
1007 		    repo, got_diff_blob_output_unidiff, &arg, 1);
1008 	} else {
1009 		err = diff_paths(tree1, tree2, f1, f2, paths, repo,
1010 		    got_diff_blob_output_unidiff, &arg);
1011 	}
1012 	if (want_lineoffsets) {
1013 		*line_offsets = arg.line_offsets; /* was likely re-allocated */
1014 		*nlines = arg.nlines;
1015 	}
1016 done:
1017 	if (tree1)
1018 		got_object_tree_close(tree1);
1019 	if (tree2)
1020 		got_object_tree_close(tree2);
1021 	return err;
1022 }
1023 
1024 const struct got_error *
1025 got_diff_objects_as_trees(off_t **line_offsets, size_t *nlines,
1026     FILE *f1, FILE *f2, struct got_object_id *id1, struct got_object_id *id2,
1027     struct got_pathlist_head *paths, const char *label1, const char *label2,
1028     int diff_context, int ignore_whitespace, int force_text_diff,
1029     struct got_repository *repo, FILE *outfile)
1030 {
1031 	const struct got_error *err;
1032 	char *idstr = NULL;
1033 
1034 	if (id1 == NULL && id2 == NULL)
1035 		return got_error(GOT_ERR_NO_OBJ);
1036 
1037 	if (id1) {
1038 		err = got_object_id_str(&idstr, id1);
1039 		if (err)
1040 			goto done;
1041 		err = show_object_id(line_offsets, nlines, "tree", '-',
1042 		    idstr, outfile);
1043 		if (err)
1044 			goto done;
1045 		free(idstr);
1046 		idstr = NULL;
1047 	} else {
1048 		err = show_object_id(line_offsets, nlines, "tree", '-',
1049 		    "/dev/null", outfile);
1050 		if (err)
1051 			goto done;
1052 	}
1053 
1054 	if (id2) {
1055 		err = got_object_id_str(&idstr, id2);
1056 		if (err)
1057 			goto done;
1058 		err = show_object_id(line_offsets, nlines, "tree", '+',
1059 		    idstr, outfile);
1060 		if (err)
1061 			goto done;
1062 		free(idstr);
1063 		idstr = NULL;
1064 	} else {
1065 		err = show_object_id(line_offsets, nlines, "tree", '+',
1066 		    "/dev/null", outfile);
1067 		if (err)
1068 			goto done;
1069 	}
1070 
1071 	err = diff_objects_as_trees(line_offsets, nlines, f1, f2, id1, id2,
1072 	    paths, label1, label2, diff_context, ignore_whitespace,
1073 	    force_text_diff, repo, outfile);
1074 done:
1075 	free(idstr);
1076 	return err;
1077 }
1078 
1079 const struct got_error *
1080 got_diff_objects_as_commits(off_t **line_offsets, size_t *nlines,
1081     FILE *f1, FILE *f2, struct got_object_id *id1, struct got_object_id *id2,
1082     struct got_pathlist_head *paths,
1083     int diff_context, int ignore_whitespace, int force_text_diff,
1084     struct got_repository *repo, FILE *outfile)
1085 {
1086 	const struct got_error *err;
1087 	struct got_commit_object *commit1 = NULL, *commit2 = NULL;
1088 	char *idstr = NULL;
1089 
1090 	if (id2 == NULL)
1091 		return got_error(GOT_ERR_NO_OBJ);
1092 
1093 	if (id1) {
1094 		err = got_object_open_as_commit(&commit1, repo, id1);
1095 		if (err)
1096 			goto done;
1097 		err = got_object_id_str(&idstr, id1);
1098 		if (err)
1099 			goto done;
1100 		err = show_object_id(line_offsets, nlines, "commit", '-',
1101 		    idstr, outfile);
1102 		if (err)
1103 			goto done;
1104 		free(idstr);
1105 		idstr = NULL;
1106 	} else {
1107 		err = show_object_id(line_offsets, nlines, "commit", '-',
1108 		    "/dev/null", outfile);
1109 		if (err)
1110 			goto done;
1111 	}
1112 
1113 	err = got_object_open_as_commit(&commit2, repo, id2);
1114 	if (err)
1115 		goto done;
1116 
1117 	err = got_object_id_str(&idstr, id2);
1118 	if (err)
1119 		goto done;
1120 	err = show_object_id(line_offsets, nlines, "commit", '+',
1121 	    idstr, outfile);
1122 	if (err)
1123 		goto done;
1124 
1125 	err = diff_objects_as_trees(line_offsets, nlines, f1, f2,
1126 	    commit1 ? got_object_commit_get_tree_id(commit1) : NULL,
1127 	    got_object_commit_get_tree_id(commit2), paths, "", "",
1128 	    diff_context, ignore_whitespace, force_text_diff, repo, outfile);
1129 done:
1130 	if (commit1)
1131 		got_object_commit_close(commit1);
1132 	if (commit2)
1133 		got_object_commit_close(commit2);
1134 	free(idstr);
1135 	return err;
1136 }
1137 
1138 const struct got_error *
1139 got_diff_files(struct got_diffreg_result **resultp,
1140     FILE *f1, const char *label1, FILE *f2, const char *label2,
1141     int diff_context, int ignore_whitespace, int force_text_diff,
1142     FILE *outfile)
1143 {
1144 	const struct got_error *err = NULL;
1145 	struct got_diffreg_result *diffreg_result = NULL;
1146 
1147 	if (resultp)
1148 		*resultp = NULL;
1149 
1150 	if (outfile) {
1151 		fprintf(outfile, "file - %s\n",
1152 		    f1 == NULL ? "/dev/null" : label1);
1153 		fprintf(outfile, "file + %s\n",
1154 		    f2 == NULL ? "/dev/null" : label2);
1155 	}
1156 
1157 	err = got_diffreg(&diffreg_result, f1, f2, GOT_DIFF_ALGORITHM_PATIENCE,
1158 	    ignore_whitespace, force_text_diff);
1159 	if (err)
1160 		goto done;
1161 
1162 	if (outfile) {
1163 		err = got_diffreg_output(NULL, NULL, diffreg_result,
1164 		    f1 != NULL, f2 != NULL, label1, label2,
1165 		    GOT_DIFF_OUTPUT_UNIDIFF, diff_context, outfile);
1166 		if (err)
1167 			goto done;
1168 	}
1169 
1170 done:
1171 	if (resultp && err == NULL)
1172 		*resultp = diffreg_result;
1173 	else if (diffreg_result) {
1174 		const struct got_error *free_err;
1175 		free_err = got_diffreg_result_free(diffreg_result);
1176 		if (free_err && err == NULL)
1177 			err = free_err;
1178 	}
1179 
1180 	return err;
1181 }