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 <zlib.h>
25 
26 #include "got_compat.h"
27 
28 #include "got_object.h"
29 #include "got_repository.h"
30 #include "got_error.h"
31 #include "got_diff.h"
32 #include "got_path.h"
33 #include "got_cancel.h"
34 #include "got_worktree.h"
35 #include "got_opentemp.h"
36 
37 #include "got_lib_diff.h"
38 #include "got_lib_delta.h"
39 #include "got_lib_inflate.h"
40 #include "got_lib_object.h"
41 
42 static const struct got_error *
43 add_line_offset(off_t **line_offsets, size_t *nlines, off_t off)
44 {
45 	off_t *p;
46 
47 	p = reallocarray(*line_offsets, *nlines + 1, sizeof(off_t));
48 	if (p == NULL)
49 		return got_error_from_errno("reallocarray");
50 	*line_offsets = p;
51 	(*line_offsets)[*nlines] = off;
52 	(*nlines)++;
53 	return NULL;
54 }
55 
56 static const struct got_error *
57 diff_blobs(off_t **line_offsets, size_t *nlines,
58     struct got_diffreg_result **resultp, struct got_blob_object *blob1,
59     struct got_blob_object *blob2, FILE *f1, FILE *f2,
60     const char *label1, const char *label2, mode_t mode1, mode_t mode2,
61     int diff_context, int ignore_whitespace, int force_text_diff, FILE *outfile,
62     enum got_diff_algorithm diff_algo)
63 {
64 	const struct got_error *err = NULL, *free_err;
65 	char hex1[SHA1_DIGEST_STRING_LENGTH];
66 	char hex2[SHA1_DIGEST_STRING_LENGTH];
67 	const char *idstr1 = NULL, *idstr2 = NULL;
68 	off_t size1, size2;
69 	struct got_diffreg_result *result;
70 	off_t outoff = 0;
71 	int n;
72 
73 	if (line_offsets && *line_offsets && *nlines > 0)
74 		outoff = (*line_offsets)[*nlines - 1];
75 	else if (line_offsets) {
76 		err = add_line_offset(line_offsets, nlines, 0);
77 		if (err)
78 			goto done;
79 	}
80 
81 	if (resultp)
82 		*resultp = NULL;
83 
84 	if (f1) {
85 		err = got_opentemp_truncate(f1);
86 		if (err)
87 			goto done;
88 	}
89 	if (f2) {
90 		err = got_opentemp_truncate(f2);
91 		if (err)
92 			goto done;
93 	}
94 
95 	size1 = 0;
96 	if (blob1) {
97 		idstr1 = got_object_blob_id_str(blob1, hex1, sizeof(hex1));
98 		err = got_object_blob_dump_to_file(&size1, NULL, NULL, f1,
99 		    blob1);
100 		if (err)
101 			goto done;
102 	} else
103 		idstr1 = "/dev/null";
104 
105 	size2 = 0;
106 	if (blob2) {
107 		idstr2 = got_object_blob_id_str(blob2, hex2, sizeof(hex2));
108 		err = got_object_blob_dump_to_file(&size2, NULL, NULL, f2,
109 		    blob2);
110 		if (err)
111 			goto done;
112 	} else
113 		idstr2 = "/dev/null";
114 
115 	if (outfile) {
116 		char *modestr1 = NULL, *modestr2 = NULL;
117 		int modebits;
118 		if (mode1 && mode1 != mode2) {
119 			if (S_ISLNK(mode1))
120 				modebits = S_IFLNK;
121 			else
122 				modebits = (S_IRWXU | S_IRWXG | S_IRWXO);
123 			if (asprintf(&modestr1, " (mode %o)",
124 			    mode1 & modebits) == -1) {
125 				err = got_error_from_errno("asprintf");
126 				goto done;
127 			}
128 		}
129 		if (mode2 && mode1 != mode2) {
130 			if (S_ISLNK(mode2))
131 				modebits = S_IFLNK;
132 			else
133 				modebits = (S_IRWXU | S_IRWXG | S_IRWXO);
134 			if (asprintf(&modestr2, " (mode %o)",
135 			    mode2 & modebits) == -1) {
136 				err = got_error_from_errno("asprintf");
137 				goto done;
138 			}
139 		}
140 		n = fprintf(outfile, "blob - %s%s\n", idstr1,
141 		    modestr1 ? modestr1 : "");
142 		if (n < 0)
143 			goto done;
144 		outoff += n;
145 		if (line_offsets) {
146 			err = add_line_offset(line_offsets, nlines, outoff);
147 			if (err)
148 				goto done;
149 		}
150 
151 		n = fprintf(outfile, "blob + %s%s\n", idstr2,
152 		    modestr2 ? modestr2 : "");
153 		if (n < 0)
154 			goto done;
155 		outoff += n;
156 		if (line_offsets) {
157 			err = add_line_offset(line_offsets, nlines, outoff);
158 			if (err)
159 				goto done;
160 		}
161 
162 		free(modestr1);
163 		free(modestr2);
164 	}
165 	err = got_diffreg(&result, f1, f2, diff_algo, ignore_whitespace,
166 	     force_text_diff);
167 	if (err)
168 		goto done;
169 
170 	if (outfile) {
171 		err = got_diffreg_output(line_offsets, nlines, result,
172 		    blob1 != NULL, blob2 != NULL,
173 		    label1 ? label1 : idstr1,
174 		    label2 ? label2 : idstr2,
175 		    GOT_DIFF_OUTPUT_UNIDIFF, diff_context, outfile);
176 		if (err)
177 			goto done;
178 	}
179 
180 	if (resultp && err == NULL)
181 		*resultp = result;
182 	else {
183 		free_err = got_diffreg_result_free(result);
184 		if (free_err && err == NULL)
185 			err = free_err;
186 	}
187 done:
188 	return err;
189 }
190 
191 const struct got_error *
192 got_diff_blob_output_unidiff(void *arg, struct got_blob_object *blob1,
193     struct got_blob_object *blob2, FILE *f1, FILE *f2,
194     struct got_object_id *id1, struct got_object_id *id2,
195     const char *label1, const char *label2, mode_t mode1, mode_t mode2,
196     struct got_repository *repo)
197 {
198 	struct got_diff_blob_output_unidiff_arg *a = arg;
199 
200 	return diff_blobs(&a->line_offsets, &a->nlines, NULL,
201 	    blob1, blob2, f1, f2, label1, label2, mode1, mode2, a->diff_context,
202 	    a->ignore_whitespace, a->force_text_diff, a->outfile, a->diff_algo);
203 }
204 
205 const struct got_error *
206 got_diff_blob(off_t **line_offsets, size_t *nlines,
207     struct got_blob_object *blob1, struct got_blob_object *blob2,
208     FILE *f1, FILE *f2, const char *label1, const char *label2,
209     enum got_diff_algorithm diff_algo, int diff_context,
210     int ignore_whitespace, int force_text_diff, FILE *outfile)
211 {
212 	return diff_blobs(line_offsets, nlines, NULL, blob1, blob2, f1, f2,
213 	    label1, label2, 0, 0, diff_context, ignore_whitespace,
214 	    force_text_diff, outfile, diff_algo);
215 }
216 
217 static const struct got_error *
218 diff_blob_file(struct got_diffreg_result **resultp,
219     struct got_blob_object *blob1, FILE *f1, off_t size1, const char *label1,
220     FILE *f2, int f2_exists, size_t size2, const char *label2,
221     enum got_diff_algorithm diff_algo, int diff_context, int ignore_whitespace,
222     int force_text_diff, FILE *outfile)
223 {
224 	const struct got_error *err = NULL, *free_err;
225 	char hex1[SHA1_DIGEST_STRING_LENGTH];
226 	const char *idstr1 = NULL;
227 	struct got_diffreg_result *result = NULL;
228 
229 	if (resultp)
230 		*resultp = NULL;
231 
232 	if (blob1)
233 		idstr1 = got_object_blob_id_str(blob1, hex1, sizeof(hex1));
234 	else
235 		idstr1 = "/dev/null";
236 
237 	if (outfile) {
238 		fprintf(outfile, "blob - %s\n", label1 ? label1 : idstr1);
239 		fprintf(outfile, "file + %s\n",
240 		    f2_exists ? label2 : "/dev/null");
241 	}
242 
243 	err = got_diffreg(&result, f1, f2, diff_algo, ignore_whitespace,
244 	    force_text_diff);
245 	if (err)
246 		goto done;
247 
248 	if (outfile) {
249 		err = got_diffreg_output(NULL, NULL, result,
250 		    blob1 != NULL, f2_exists,
251 		    label2, /* show local file's path, not a blob ID */
252 		    label2, GOT_DIFF_OUTPUT_UNIDIFF,
253 		    diff_context, outfile);
254 		if (err)
255 			goto done;
256 	}
257 
258 	if (resultp && err == NULL)
259 		*resultp = result;
260 	else if (result) {
261 		free_err = got_diffreg_result_free(result);
262 		if (free_err && err == NULL)
263 			err = free_err;
264 	}
265 done:
266 	return err;
267 }
268 
269 const struct got_error *
270 got_diff_blob_file(struct got_blob_object *blob1, FILE *f1, off_t size1,
271     const char *label1, FILE *f2, int f2_exists, size_t size2,
272     const char *label2, enum got_diff_algorithm diff_algo, int diff_context,
273     int ignore_whitespace, int force_text_diff, FILE *outfile)
274 {
275 	return diff_blob_file(NULL, blob1, f1, size1, label1, f2, f2_exists,
276 	    size2, label2, diff_algo, diff_context, ignore_whitespace,
277 	    force_text_diff, outfile );
278 }
279 
280 static const struct got_error *
281 diff_added_blob(struct got_object_id *id, FILE *f1, FILE *f2, int fd2,
282     const char *label, mode_t mode, struct got_repository *repo,
283     got_diff_blob_cb cb, void *cb_arg)
284 {
285 	const struct got_error *err;
286 	struct got_blob_object  *blob = NULL;
287 	struct got_object *obj = NULL;
288 
289 	err = got_object_open(&obj, repo, id);
290 	if (err)
291 		return err;
292 
293 	err = got_object_blob_open(&blob, repo, obj, 8192, fd2);
294 	if (err)
295 		goto done;
296 	err = cb(cb_arg, NULL, blob, f1, f2, NULL, id,
297 	    NULL, label, 0, mode, repo);
298 done:
299 	got_object_close(obj);
300 	if (blob)
301 		got_object_blob_close(blob);
302 	return err;
303 }
304 
305 static const struct got_error *
306 diff_modified_blob(struct got_object_id *id1, struct got_object_id *id2,
307     FILE *f1, FILE *f2, int fd1, int fd2,
308     const char *label1, const char *label2,
309     mode_t mode1, mode_t mode2, struct got_repository *repo,
310     got_diff_blob_cb cb, void *cb_arg)
311 {
312 	const struct got_error *err;
313 	struct got_object *obj1 = NULL;
314 	struct got_object *obj2 = NULL;
315 	struct got_blob_object *blob1 = NULL;
316 	struct got_blob_object *blob2 = NULL;
317 
318 	err = got_object_open(&obj1, repo, id1);
319 	if (err)
320 		return err;
321 
322 	if (obj1->type != GOT_OBJ_TYPE_BLOB) {
323 		err = got_error(GOT_ERR_OBJ_TYPE);
324 		goto done;
325 	}
326 
327 	err = got_object_open(&obj2, repo, id2);
328 	if (err)
329 		goto done;
330 	if (obj2->type != GOT_OBJ_TYPE_BLOB) {
331 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
332 		goto done;
333 	}
334 
335 	err = got_object_blob_open(&blob1, repo, obj1, 8192, fd1);
336 	if (err)
337 		goto done;
338 
339 	err = got_object_blob_open(&blob2, repo, obj2, 8192, fd2);
340 	if (err)
341 		goto done;
342 
343 	err = cb(cb_arg, blob1, blob2, f1, f2, id1, id2, label1, label2,
344 	    mode1, mode2, repo);
345 done:
346 	if (obj1)
347 		got_object_close(obj1);
348 	if (obj2)
349 		got_object_close(obj2);
350 	if (blob1)
351 		got_object_blob_close(blob1);
352 	if (blob2)
353 		got_object_blob_close(blob2);
354 	return err;
355 }
356 
357 static const struct got_error *
358 diff_deleted_blob(struct got_object_id *id, FILE *f1, int fd1,
359    FILE *f2, const char *label, mode_t mode, struct got_repository *repo,
360    got_diff_blob_cb cb, void *cb_arg)
361 {
362 	const struct got_error *err;
363 	struct got_blob_object  *blob = NULL;
364 	struct got_object *obj = NULL;
365 
366 	err = got_object_open(&obj, repo, id);
367 	if (err)
368 		return err;
369 
370 	err = got_object_blob_open(&blob, repo, obj, 8192, fd1);
371 	if (err)
372 		goto done;
373 	err = cb(cb_arg, blob, NULL, f1, f2, id, NULL, label, NULL,
374 	    mode, 0, repo);
375 done:
376 	got_object_close(obj);
377 	if (blob)
378 		got_object_blob_close(blob);
379 	return err;
380 }
381 
382 static const struct got_error *
383 diff_added_tree(struct got_object_id *id, FILE *f1, FILE *f2, int fd2,
384     const char *label, struct got_repository *repo, got_diff_blob_cb cb,
385     void *cb_arg, int diff_content)
386 {
387 	const struct got_error *err = NULL;
388 	struct got_object *treeobj = NULL;
389 	struct got_tree_object *tree = NULL;
390 
391 	err = got_object_open(&treeobj, repo, id);
392 	if (err)
393 		goto done;
394 
395 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
396 		err = got_error(GOT_ERR_OBJ_TYPE);
397 		goto done;
398 	}
399 
400 	err = got_object_tree_open(&tree, repo, treeobj);
401 	if (err)
402 		goto done;
403 
404 	err = got_diff_tree(NULL, tree, f1, f2, -1, fd2, NULL, label,
405 	    repo, cb, cb_arg, diff_content);
406 done:
407 	if (tree)
408 		got_object_tree_close(tree);
409 	if (treeobj)
410 		got_object_close(treeobj);
411 	return err;
412 }
413 
414 static const struct got_error *
415 diff_modified_tree(struct got_object_id *id1, struct got_object_id *id2,
416     FILE *f1, FILE *f2, int fd1, int fd2,
417     const char *label1, const char *label2,
418     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
419     int diff_content)
420 {
421 	const struct got_error *err;
422 	struct got_object *treeobj1 = NULL;
423 	struct got_object *treeobj2 = NULL;
424 	struct got_tree_object *tree1 = NULL;
425 	struct got_tree_object *tree2 = NULL;
426 
427 	err = got_object_open(&treeobj1, repo, id1);
428 	if (err)
429 		goto done;
430 
431 	if (treeobj1->type != GOT_OBJ_TYPE_TREE) {
432 		err = got_error(GOT_ERR_OBJ_TYPE);
433 		goto done;
434 	}
435 
436 	err = got_object_open(&treeobj2, repo, id2);
437 	if (err)
438 		goto done;
439 
440 	if (treeobj2->type != GOT_OBJ_TYPE_TREE) {
441 		err = got_error(GOT_ERR_OBJ_TYPE);
442 		goto done;
443 	}
444 
445 	err = got_object_tree_open(&tree1, repo, treeobj1);
446 	if (err)
447 		goto done;
448 
449 	err = got_object_tree_open(&tree2, repo, treeobj2);
450 	if (err)
451 		goto done;
452 
453 	err = got_diff_tree(tree1, tree2, f1, f2, fd1, fd2,
454 	    label1, label2, repo, cb, cb_arg, diff_content);
455 
456 done:
457 	if (tree1)
458 		got_object_tree_close(tree1);
459 	if (tree2)
460 		got_object_tree_close(tree2);
461 	if (treeobj1)
462 		got_object_close(treeobj1);
463 	if (treeobj2)
464 		got_object_close(treeobj2);
465 	return err;
466 }
467 
468 static const struct got_error *
469 diff_deleted_tree(struct got_object_id *id, FILE *f1, int fd1,
470     FILE *f2, const char *label, struct got_repository *repo,
471     got_diff_blob_cb cb, void *cb_arg, int diff_content)
472 {
473 	const struct got_error *err;
474 	struct got_object *treeobj = NULL;
475 	struct got_tree_object *tree = NULL;
476 
477 	err = got_object_open(&treeobj, repo, id);
478 	if (err)
479 		goto done;
480 
481 	if (treeobj->type != GOT_OBJ_TYPE_TREE) {
482 		err = got_error(GOT_ERR_OBJ_TYPE);
483 		goto done;
484 	}
485 
486 	err = got_object_tree_open(&tree, repo, treeobj);
487 	if (err)
488 		goto done;
489 
490 	err = got_diff_tree(tree, NULL, f1, f2, fd1, -1, label, NULL,
491 	    repo, cb, cb_arg, diff_content);
492 done:
493 	if (tree)
494 		got_object_tree_close(tree);
495 	if (treeobj)
496 		got_object_close(treeobj);
497 	return err;
498 }
499 
500 static const struct got_error *
501 diff_kind_mismatch(struct got_object_id *id1, struct got_object_id *id2,
502     const char *label1, const char *label2, struct got_repository *repo,
503     got_diff_blob_cb cb, void *cb_arg)
504 {
505 	/* XXX TODO */
506 	return NULL;
507 }
508 
509 static const struct got_error *
510 diff_entry_old_new(struct got_tree_entry *te1, struct got_tree_entry *te2,
511     FILE *f1, FILE *f2, int fd1, int fd2,
512     const char *label1, const char *label2,
513     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
514     int diff_content)
515 {
516 	const struct got_error *err = NULL;
517 	int id_match;
518 
519 	if (got_object_tree_entry_is_submodule(te1))
520 		return NULL;
521 
522 	if (te2 == NULL) {
523 		if (S_ISDIR(te1->mode))
524 			err = diff_deleted_tree(&te1->id, f1, fd1, f2,
525 			    label1, repo, cb, cb_arg, diff_content);
526 		else {
527 			if (diff_content)
528 				err = diff_deleted_blob(&te1->id, f1, fd1,
529 				    f2, label1, te1->mode, repo, cb, cb_arg);
530 			else
531 				err = cb(cb_arg, NULL, NULL, NULL, NULL,
532 				    &te1->id, NULL, label1, NULL,
533 				    te1->mode, 0, repo);
534 		}
535 		return err;
536 	} else if (got_object_tree_entry_is_submodule(te2))
537 		return NULL;
538 
539 	id_match = (got_object_id_cmp(&te1->id, &te2->id) == 0);
540 	if (S_ISDIR(te1->mode) && S_ISDIR(te2->mode)) {
541 		if (!id_match)
542 			return diff_modified_tree(&te1->id, &te2->id, f1, f2,
543 			    fd1, fd2, label1, label2, repo, cb, cb_arg,
544 			    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, fd1, fd2, label1, label2,
553 				    te1->mode, 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 *f1, FILE *f2, int fd2, 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, f1, f2, fd2, label2,
582 		    repo, cb, cb_arg, diff_content);
583 
584 	if (diff_content)
585 		return diff_added_blob(&te2->id, f1, f2, fd2,
586 		    label2, te2->mode, 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, int fd1, int fd2,
638     const char *label1, const char *label2,
639     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg,
640     int diff_content)
641 {
642 	const struct got_error *err = NULL;
643 	struct got_tree_entry *te1 = NULL;
644 	struct got_tree_entry *te2 = NULL;
645 	char *l1 = NULL, *l2 = NULL;
646 	int tidx1 = 0, tidx2 = 0;
647 
648 	if (tree1) {
649 		te1 = got_object_tree_get_entry(tree1, 0);
650 		if (te1 && asprintf(&l1, "%s%s%s", label1, label1[0] ? "/" : "",
651 		    te1->name) == -1)
652 			return got_error_from_errno("asprintf");
653 	}
654 	if (tree2) {
655 		te2 = got_object_tree_get_entry(tree2, 0);
656 		if (te2 && asprintf(&l2, "%s%s%s", label2, label2[0] ? "/" : "",
657 		    te2->name) == -1)
658 			return got_error_from_errno("asprintf");
659 	}
660 
661 	do {
662 		if (te1) {
663 			struct got_tree_entry *te = NULL;
664 			if (tree2)
665 				te = got_object_tree_find_entry(tree2,
666 				    te1->name);
667 			if (te) {
668 				free(l2);
669 				l2 = NULL;
670 				if (te && asprintf(&l2, "%s%s%s", label2,
671 				    label2[0] ? "/" : "", te->name) == -1)
672 					return
673 					    got_error_from_errno("asprintf");
674 			}
675 			err = diff_entry_old_new(te1, te, f1, f2, fd1, fd2,
676 			    l1, l2, repo, cb, cb_arg, diff_content);
677 			if (err)
678 				break;
679 		}
680 
681 		if (te2) {
682 			struct got_tree_entry *te = NULL;
683 			if (tree1)
684 				te = got_object_tree_find_entry(tree1,
685 				    te2->name);
686 			free(l2);
687 			if (te) {
688 				if (asprintf(&l2, "%s%s%s", label2,
689 				    label2[0] ? "/" : "", te->name) == -1)
690 					return
691 					    got_error_from_errno("asprintf");
692 			} else {
693 				if (asprintf(&l2, "%s%s%s", label2,
694 				    label2[0] ? "/" : "", te2->name) == -1)
695 					return
696 					    got_error_from_errno("asprintf");
697 			}
698 			err = diff_entry_new_old(te2, te, f1, f2, fd2, l2,
699 			    repo, cb, cb_arg, diff_content);
700 			if (err)
701 				break;
702 		}
703 
704 		free(l1);
705 		l1 = NULL;
706 		if (te1) {
707 			tidx1++;
708 			te1 = got_object_tree_get_entry(tree1, tidx1);
709 			if (te1 &&
710 			    asprintf(&l1, "%s%s%s", label1,
711 			    label1[0] ? "/" : "", te1->name) == -1)
712 				return got_error_from_errno("asprintf");
713 		}
714 		free(l2);
715 		l2 = NULL;
716 		if (te2) {
717 			tidx2++;
718 			te2 = got_object_tree_get_entry(tree2, tidx2);
719 			if (te2 &&
720 			    asprintf(&l2, "%s%s%s", label2,
721 			        label2[0] ? "/" : "", te2->name) == -1)
722 				return got_error_from_errno("asprintf");
723 		}
724 	} while (te1 || te2);
725 
726 	return err;
727 }
728 
729 const struct got_error *
730 got_diff_objects_as_blobs(off_t **line_offsets, size_t *nlines,
731     FILE *f1, FILE *f2, int fd1, int fd2,
732     struct got_object_id *id1, struct got_object_id *id2,
733     const char *label1, const char *label2,
734     enum got_diff_algorithm diff_algo, int diff_context,
735     int ignore_whitespace, int force_text_diff,
736     struct got_repository *repo, FILE *outfile)
737 {
738 	const struct got_error *err;
739 	struct got_blob_object *blob1 = NULL, *blob2 = NULL;
740 
741 	if (id1 == NULL && id2 == NULL)
742 		return got_error(GOT_ERR_NO_OBJ);
743 
744 	if (id1) {
745 		err = got_object_open_as_blob(&blob1, repo, id1, 8192, fd1);
746 		if (err)
747 			goto done;
748 	}
749 	if (id2) {
750 		err = got_object_open_as_blob(&blob2, repo, id2, 8192, fd2);
751 		if (err)
752 			goto done;
753 	}
754 	err = got_diff_blob(line_offsets, nlines, blob1, blob2, f1, f2,
755 	    label1, label2, diff_algo, diff_context, ignore_whitespace,
756 	    force_text_diff, outfile);
757 done:
758 	if (blob1)
759 		got_object_blob_close(blob1);
760 	if (blob2)
761 		got_object_blob_close(blob2);
762 	return err;
763 }
764 
765 static const struct got_error *
766 diff_paths(struct got_tree_object *tree1, struct got_tree_object *tree2,
767     FILE *f1, FILE *f2, int fd1, int fd2, struct got_pathlist_head *paths,
768     struct got_repository *repo, got_diff_blob_cb cb, void *cb_arg)
769 {
770 	const struct got_error *err = NULL;
771 	struct got_pathlist_entry *pe;
772 	struct got_object_id *id1 = NULL, *id2 = NULL;
773 	struct got_tree_object *subtree1 = NULL, *subtree2 = NULL;
774 	struct got_blob_object *blob1 = NULL, *blob2 = NULL;
775 
776 	TAILQ_FOREACH(pe, paths, entry) {
777 		int type1 = GOT_OBJ_TYPE_ANY, type2 = GOT_OBJ_TYPE_ANY;
778 		mode_t mode1 = 0, mode2 = 0;
779 
780 		free(id1);
781 		id1 = NULL;
782 		free(id2);
783 		id2 = NULL;
784 		if (subtree1) {
785 			got_object_tree_close(subtree1);
786 			subtree1 = NULL;
787 		}
788 		if (subtree2) {
789 			got_object_tree_close(subtree2);
790 			subtree2 = NULL;
791 		}
792 		if (blob1) {
793 			got_object_blob_close(blob1);
794 			blob1 = NULL;
795 		}
796 		if (blob2) {
797 			got_object_blob_close(blob2);
798 			blob2 = NULL;
799 		}
800 
801 		err = got_object_tree_find_path(&id1, &mode1, repo, tree1,
802 		    pe->path);
803 		if (err && err->code != GOT_ERR_NO_TREE_ENTRY)
804 			goto done;
805 		err = got_object_tree_find_path(&id2, &mode2, repo, tree2,
806 		    pe->path);
807 		if (err && err->code != GOT_ERR_NO_TREE_ENTRY)
808 			goto done;
809 		if (id1 == NULL && id2 == NULL) {
810 			err = got_error_path(pe->path, GOT_ERR_NO_TREE_ENTRY);
811 			goto done;
812 		}
813 		if (id1) {
814 			err = got_object_get_type(&type1, repo, id1);
815 			if (err)
816 				goto done;
817 		}
818 		if (id2) {
819 			err = got_object_get_type(&type2, repo, id2);
820 			if (err)
821 				goto done;
822 		}
823 		if (type1 == GOT_OBJ_TYPE_ANY &&
824 		    type2 == GOT_OBJ_TYPE_ANY) {
825 			err = got_error_path(pe->path, GOT_ERR_NO_OBJ);
826 			goto done;
827 		} else if (type1 != GOT_OBJ_TYPE_ANY &&
828 		    type2 != GOT_OBJ_TYPE_ANY && type1 != type2) {
829 			err = got_error(GOT_ERR_OBJ_TYPE);
830 			goto done;
831 		}
832 
833 		if (type1 == GOT_OBJ_TYPE_BLOB ||
834 		    type2 == GOT_OBJ_TYPE_BLOB) {
835 			if (id1) {
836 				err = got_object_open_as_blob(&blob1, repo,
837 				    id1, 8192, fd1);
838 				if (err)
839 					goto done;
840 			}
841 			if (id2) {
842 				err = got_object_open_as_blob(&blob2, repo,
843 				    id2, 8192, fd2);
844 				if (err)
845 					goto done;
846 			}
847 			err = cb(cb_arg, blob1, blob2, f1, f2, id1, id2,
848 			    id1 ? pe->path : "/dev/null",
849 			    id2 ? pe->path : "/dev/null",
850 			    mode1, mode2, repo);
851 			if (err)
852 				goto done;
853 		} else if (type1 == GOT_OBJ_TYPE_TREE ||
854 		    type2 == GOT_OBJ_TYPE_TREE) {
855 			if (id1) {
856 				err = got_object_open_as_tree(&subtree1, repo,
857 				    id1);
858 				if (err)
859 					goto done;
860 			}
861 			if (id2) {
862 				err = got_object_open_as_tree(&subtree2, repo,
863 				    id2);
864 				if (err)
865 					goto done;
866 			}
867 			err = got_diff_tree(subtree1, subtree2, f1, f2,
868 			    fd1, fd2,
869 			    id1 ? pe->path : "/dev/null",
870 			    id2 ? pe->path : "/dev/null",
871 			    repo, cb, cb_arg, 1);
872 			if (err)
873 				goto done;
874 		} else {
875 			err = got_error(GOT_ERR_OBJ_TYPE);
876 			goto done;
877 		}
878 	}
879 done:
880 	free(id1);
881 	free(id2);
882 	if (subtree1)
883 		got_object_tree_close(subtree1);
884 	if (subtree2)
885 		got_object_tree_close(subtree2);
886 	if (blob1)
887 		got_object_blob_close(blob1);
888 	if (blob2)
889 		got_object_blob_close(blob2);
890 	return err;
891 }
892 
893 static const struct got_error *
894 show_object_id(off_t **line_offsets, size_t *nlines, const char *obj_typestr,
895     int ch, const char *id_str, FILE *outfile)
896 {
897 	const struct got_error *err;
898 	int n;
899 	off_t outoff = 0;
900 
901 	n = fprintf(outfile, "%s %c %s\n", obj_typestr, ch, id_str);
902 	if (line_offsets != NULL && *line_offsets != NULL) {
903 		if (*nlines == 0) {
904 			err = add_line_offset(line_offsets, nlines, 0);
905 			if (err)
906 				return err;
907 		} else
908 			outoff = (*line_offsets)[*nlines - 1];
909 
910 		outoff += n;
911 		err = add_line_offset(line_offsets, nlines, outoff);
912 		if (err)
913 			return err;
914 	}
915 
916 	return NULL;
917 }
918 
919 static const struct got_error *
920 diff_objects_as_trees(off_t **line_offsets, size_t *nlines,
921     FILE *f1, FILE *f2, int fd1, int fd2,
922     struct got_object_id *id1, struct got_object_id *id2,
923     struct got_pathlist_head *paths, const char *label1, const char *label2,
924     int diff_context, int ignore_whitespace, int force_text_diff,
925     struct got_repository *repo, FILE *outfile,
926     enum got_diff_algorithm diff_algo)
927 {
928 	const struct got_error *err;
929 	struct got_tree_object *tree1 = NULL, *tree2 = NULL;
930 	struct got_diff_blob_output_unidiff_arg arg;
931 	int want_lineoffsets = (line_offsets != NULL && *line_offsets != NULL);
932 
933 	if (id1 == NULL && id2 == NULL)
934 		return got_error(GOT_ERR_NO_OBJ);
935 
936 	if (id1) {
937 		err = got_object_open_as_tree(&tree1, repo, id1);
938 		if (err)
939 			goto done;
940 	}
941 	if (id2) {
942 		err = got_object_open_as_tree(&tree2, repo, id2);
943 		if (err)
944 			goto done;
945 	}
946 
947 	arg.diff_algo = diff_algo;
948 	arg.diff_context = diff_context;
949 	arg.ignore_whitespace = ignore_whitespace;
950 	arg.force_text_diff = force_text_diff;
951 	arg.outfile = outfile;
952 	if (want_lineoffsets) {
953 		arg.line_offsets = *line_offsets;
954 		arg.nlines = *nlines;
955 	} else {
956 		arg.line_offsets = NULL;
957 		arg.nlines = 0;
958 	}
959 	if (paths == NULL || TAILQ_EMPTY(paths)) {
960 		err = got_diff_tree(tree1, tree2, f1, f2, fd1, fd2,
961 		    label1, label2, repo,
962 		    got_diff_blob_output_unidiff, &arg, 1);
963 	} else {
964 		err = diff_paths(tree1, tree2, f1, f2, fd1, fd2, paths, repo,
965 		    got_diff_blob_output_unidiff, &arg);
966 	}
967 	if (want_lineoffsets) {
968 		*line_offsets = arg.line_offsets; /* was likely re-allocated */
969 		*nlines = arg.nlines;
970 	}
971 done:
972 	if (tree1)
973 		got_object_tree_close(tree1);
974 	if (tree2)
975 		got_object_tree_close(tree2);
976 	return err;
977 }
978 
979 const struct got_error *
980 got_diff_objects_as_trees(off_t **line_offsets, size_t *nlines,
981     FILE *f1, FILE *f2, int fd1, int fd2,
982     struct got_object_id *id1, struct got_object_id *id2,
983     struct got_pathlist_head *paths, const char *label1, const char *label2,
984     enum got_diff_algorithm diff_algo, int diff_context, int ignore_whitespace,
985     int force_text_diff, struct got_repository *repo, FILE *outfile)
986 {
987 	const struct got_error *err;
988 	char *idstr = NULL;
989 
990 	if (id1 == NULL && id2 == NULL)
991 		return got_error(GOT_ERR_NO_OBJ);
992 
993 	if (id1) {
994 		err = got_object_id_str(&idstr, id1);
995 		if (err)
996 			goto done;
997 		err = show_object_id(line_offsets, nlines, "tree", '-',
998 		    idstr, outfile);
999 		if (err)
1000 			goto done;
1001 		free(idstr);
1002 		idstr = NULL;
1003 	} else {
1004 		err = show_object_id(line_offsets, nlines, "tree", '-',
1005 		    "/dev/null", outfile);
1006 		if (err)
1007 			goto done;
1008 	}
1009 
1010 	if (id2) {
1011 		err = got_object_id_str(&idstr, id2);
1012 		if (err)
1013 			goto done;
1014 		err = show_object_id(line_offsets, nlines, "tree", '+',
1015 		    idstr, outfile);
1016 		if (err)
1017 			goto done;
1018 		free(idstr);
1019 		idstr = NULL;
1020 	} else {
1021 		err = show_object_id(line_offsets, nlines, "tree", '+',
1022 		    "/dev/null", outfile);
1023 		if (err)
1024 			goto done;
1025 	}
1026 
1027 	err = diff_objects_as_trees(line_offsets, nlines, f1, f2, fd1, fd2,
1028 	    id1, id2, paths, label1, label2, diff_context, ignore_whitespace,
1029 	    force_text_diff, repo, outfile, diff_algo);
1030 done:
1031 	free(idstr);
1032 	return err;
1033 }
1034 
1035 const struct got_error *
1036 got_diff_objects_as_commits(off_t **line_offsets, size_t *nlines,
1037     FILE *f1, FILE *f2, int fd1, int fd2,
1038     struct got_object_id *id1, struct got_object_id *id2,
1039     struct got_pathlist_head *paths, enum got_diff_algorithm diff_algo,
1040     int diff_context, int ignore_whitespace, int force_text_diff,
1041     struct got_repository *repo, FILE *outfile)
1042 {
1043 	const struct got_error *err;
1044 	struct got_commit_object *commit1 = NULL, *commit2 = NULL;
1045 	char *idstr = NULL;
1046 
1047 	if (id2 == NULL)
1048 		return got_error(GOT_ERR_NO_OBJ);
1049 
1050 	if (id1) {
1051 		err = got_object_open_as_commit(&commit1, repo, id1);
1052 		if (err)
1053 			goto done;
1054 		err = got_object_id_str(&idstr, id1);
1055 		if (err)
1056 			goto done;
1057 		err = show_object_id(line_offsets, nlines, "commit", '-',
1058 		    idstr, outfile);
1059 		if (err)
1060 			goto done;
1061 		free(idstr);
1062 		idstr = NULL;
1063 	} else {
1064 		err = show_object_id(line_offsets, nlines, "commit", '-',
1065 		    "/dev/null", outfile);
1066 		if (err)
1067 			goto done;
1068 	}
1069 
1070 	err = got_object_open_as_commit(&commit2, repo, id2);
1071 	if (err)
1072 		goto done;
1073 
1074 	err = got_object_id_str(&idstr, id2);
1075 	if (err)
1076 		goto done;
1077 	err = show_object_id(line_offsets, nlines, "commit", '+',
1078 	    idstr, outfile);
1079 	if (err)
1080 		goto done;
1081 
1082 	err = diff_objects_as_trees(line_offsets, nlines, f1, f2, fd1, fd2,
1083 	    commit1 ? got_object_commit_get_tree_id(commit1) : NULL,
1084 	    got_object_commit_get_tree_id(commit2), paths, "", "",
1085 	    diff_context, ignore_whitespace, force_text_diff, repo, outfile,
1086 	    diff_algo);
1087 done:
1088 	if (commit1)
1089 		got_object_commit_close(commit1);
1090 	if (commit2)
1091 		got_object_commit_close(commit2);
1092 	free(idstr);
1093 	return err;
1094 }
1095 
1096 const struct got_error *
1097 got_diff_files(struct got_diffreg_result **resultp,
1098     FILE *f1, int f1_exists, const char *label1, FILE *f2, int f2_exists,
1099     const char *label2, int diff_context, int ignore_whitespace,
1100     int force_text_diff, FILE *outfile, enum got_diff_algorithm diff_algo)
1101 {
1102 	const struct got_error *err = NULL;
1103 	struct got_diffreg_result *diffreg_result = NULL;
1104 
1105 	if (resultp)
1106 		*resultp = NULL;
1107 
1108 	if (outfile) {
1109 		fprintf(outfile, "file - %s\n",
1110 		    f1_exists ? label1 : "/dev/null");
1111 		fprintf(outfile, "file + %s\n",
1112 		    f2_exists ? label2 : "/dev/null");
1113 	}
1114 
1115 	err = got_diffreg(&diffreg_result, f1, f2, diff_algo,
1116 	    ignore_whitespace, force_text_diff);
1117 	if (err)
1118 		goto done;
1119 
1120 	if (outfile) {
1121 		err = got_diffreg_output(NULL, NULL, diffreg_result,
1122 		    f1_exists, f2_exists, label1, label2,
1123 		    GOT_DIFF_OUTPUT_UNIDIFF, diff_context, outfile);
1124 		if (err)
1125 			goto done;
1126 	}
1127 
1128 done:
1129 	if (resultp && err == NULL)
1130 		*resultp = diffreg_result;
1131 	else if (diffreg_result) {
1132 		const struct got_error *free_err;
1133 		free_err = got_diffreg_result_free(diffreg_result);
1134 		if (free_err && err == NULL)
1135 			err = free_err;
1136 	}
1137 
1138 	return err;
1139 }