1 /*
2  * Copyright (c) 2022 Omar Polo <op@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  * Apply patches.
17  *
18  * Things that are still missing:
19  *     + "No final newline" handling
20  *
21  * Things that we may want to support:
22  *     + support indented patches?
23  *     + support other kinds of patches?
24  */
25 
26 #include <sys/types.h>
27 #include <sys/queue.h>
28 #include <sys/socket.h>
29 #include <sys/uio.h>
30 
31 #include <limits.h>
32 #include <sha1.h>
33 #include <stdint.h>
34 #include <stdio.h>
35 #include <stdlib.h>
36 #include <string.h>
37 #include <unistd.h>
38 #include <imsg.h>
39 
40 #include "got_error.h"
41 #include "got_object.h"
42 #include "got_path.h"
43 #include "got_reference.h"
44 #include "got_cancel.h"
45 #include "got_worktree.h"
46 #include "got_opentemp.h"
47 #include "got_patch.h"
48 
49 #include "got_lib_delta.h"
50 #include "got_lib_object.h"
51 #include "got_lib_privsep.h"
52 
53 #define MIN(a, b) ((a) < (b) ? (a) : (b))
54 
55 struct got_patch_hunk {
56 	STAILQ_ENTRY(got_patch_hunk) entries;
57 	long	old_from;
58 	long	old_lines;
59 	long	new_from;
60 	long	new_lines;
61 	size_t	len;
62 	size_t	cap;
63 	char	**lines;
64 };
65 
66 struct got_patch {
67 	char	*old;
68 	char	*new;
69 	STAILQ_HEAD(, got_patch_hunk) head;
70 };
71 
72 static const struct got_error *
73 send_patch(struct imsgbuf *ibuf, int fd)
74 {
75 	const struct got_error *err = NULL;
76 
77 	if (imsg_compose(ibuf, GOT_IMSG_PATCH_FILE, 0, 0, fd,
78 	    NULL, 0) == -1) {
79 		err = got_error_from_errno(
80 		    "imsg_compose GOT_IMSG_PATCH_FILE");
81 		close(fd);
82 		return err;
83 	}
84 
85 	if (imsg_flush(ibuf) == -1) {
86 		err = got_error_from_errno("imsg_flush");
87 		imsg_clear(ibuf);
88 	}
89 
90 	return err;
91 }
92 
93 static void
94 patch_free(struct got_patch *p)
95 {
96 	struct got_patch_hunk *h;
97 	size_t i;
98 
99 	while (!STAILQ_EMPTY(&p->head)) {
100 		h = STAILQ_FIRST(&p->head);
101 		STAILQ_REMOVE_HEAD(&p->head, entries);
102 
103 		for (i = 0; i < h->len; ++i)
104 			free(h->lines[i]);
105 		free(h->lines);
106 		free(h);
107 	}
108 
109 	free(p->new);
110 	free(p->old);
111 }
112 
113 static const struct got_error *
114 pushline(struct got_patch_hunk *h, const char *line)
115 {
116 	void 	*t;
117 	size_t	 newcap;
118 
119 	if (h->len == h->cap) {
120 		if ((newcap = h->cap * 1.5) == 0)
121 			newcap = 16;
122 		t = recallocarray(h->lines, h->cap, newcap,
123 		    sizeof(h->lines[0]));
124 		if (t == NULL)
125 			return got_error_from_errno("recallocarray");
126 		h->lines = t;
127 		h->cap = newcap;
128 	}
129 
130 	if ((t = strdup(line)) == NULL)
131 		return got_error_from_errno("strdup");
132 
133 	h->lines[h->len++] = t;
134 	return NULL;
135 }
136 
137 static const struct got_error *
138 recv_patch(struct imsgbuf *ibuf, int *done, struct got_patch *p)
139 {
140 	const struct got_error *err = NULL;
141 	struct imsg imsg;
142 	struct got_imsg_patch_hunk hdr;
143 	struct got_imsg_patch patch;
144 	struct got_patch_hunk *h = NULL;
145 	size_t datalen;
146 
147 	memset(p, 0, sizeof(*p));
148 	STAILQ_INIT(&p->head);
149 
150 	err = got_privsep_recv_imsg(&imsg, ibuf, 0);
151 	if (err)
152 		return err;
153 	if (imsg.hdr.type == GOT_IMSG_PATCH_EOF) {
154 		*done = 1;
155 		goto done;
156 	}
157 	if (imsg.hdr.type != GOT_IMSG_PATCH) {
158 		err = got_error(GOT_ERR_PRIVSEP_MSG);
159 		goto done;
160 	}
161 	datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
162 	if (datalen != sizeof(patch)) {
163 		err = got_error(GOT_ERR_PRIVSEP_LEN);
164 		goto done;
165 	}
166 	memcpy(&patch, imsg.data, sizeof(patch));
167 	if (*patch.old != '\0' && (p->old = strdup(patch.old)) == NULL) {
168 		err = got_error_from_errno("strdup");
169 		goto done;
170 	}
171 	if (*patch.new != '\0' && (p->new = strdup(patch.new)) == NULL) {
172 		err = got_error_from_errno("strdup");
173 		goto done;
174 	}
175 
176 	imsg_free(&imsg);
177 
178 	for (;;) {
179 		char *t;
180 
181 		err = got_privsep_recv_imsg(&imsg, ibuf, 0);
182 		if (err)
183 			return err;
184 
185 		switch (imsg.hdr.type) {
186 		case GOT_IMSG_PATCH_DONE:
187 			goto done;
188 		case GOT_IMSG_PATCH_HUNK:
189 			datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
190 			if (datalen != sizeof(hdr)) {
191 				err = got_error(GOT_ERR_PRIVSEP_LEN);
192 				goto done;
193 			}
194 			memcpy(&hdr, imsg.data, sizeof(hdr));
195 			if ((h = calloc(1, sizeof(*h))) == NULL) {
196 				err = got_error_from_errno("calloc");
197 				goto done;
198 			}
199 			h->old_from = hdr.oldfrom;
200 			h->old_lines = hdr.oldlines;
201 			h->new_from = hdr.newfrom;
202 			h->new_lines = hdr.newlines;
203 			STAILQ_INSERT_TAIL(&p->head, h, entries);
204 			break;
205 		case GOT_IMSG_PATCH_LINE:
206 			if (h == NULL) {
207 				err = got_error(GOT_ERR_PRIVSEP_MSG);
208 				goto done;
209 			}
210 			datalen = imsg.hdr.len - IMSG_HEADER_SIZE;
211 			t = imsg.data;
212 			/* at least one char plus newline */
213 			if (datalen < 2 || t[datalen-1] != '\0') {
214 				err = got_error(GOT_ERR_PRIVSEP_MSG);
215 				goto done;
216 			}
217 			if (*t != ' ' && *t != '-' && *t != '+') {
218 				err = got_error(GOT_ERR_PRIVSEP_MSG);
219 				goto done;
220 			}
221 			err = pushline(h, t);
222 			if (err)
223 				goto done;
224 			break;
225 		default:
226 			err = got_error(GOT_ERR_PRIVSEP_MSG);
227 			goto done;
228 		}
229 
230 		imsg_free(&imsg);
231 	}
232 
233 done:
234 	imsg_free(&imsg);
235 	return err;
236 }
237 
238 /*
239  * Copy data from orig starting at copypos until pos into tmp.
240  * If pos is -1, copy until EOF.
241  */
242 static const struct got_error *
243 copy(FILE *tmp, FILE *orig, off_t copypos, off_t pos)
244 {
245 	char buf[BUFSIZ];
246 	size_t len, r, w;
247 
248 	if (fseek(orig, copypos, SEEK_SET) == -1)
249 		return got_error_from_errno("fseek");
250 
251 	while (pos == -1 || copypos < pos) {
252 		len = sizeof(buf);
253 		if (pos > 0)
254 			len = MIN(len, (size_t)pos - copypos);
255 		r = fread(buf, 1, len, orig);
256 		if (r != len && ferror(orig))
257 			return got_error_from_errno("fread");
258 		w = fwrite(buf, 1, r, tmp);
259 		if (w != r)
260 			return got_error_from_errno("fwrite");
261 		copypos += len;
262 		if (r != len && feof(orig)) {
263 			if (pos == -1)
264 				return NULL;
265 			return got_error(GOT_ERR_PATCH_DONT_APPLY);
266 		}
267 	}
268 	return NULL;
269 }
270 
271 static const struct got_error *
272 locate_hunk(FILE *orig, struct got_patch_hunk *h, long *lineno)
273 {
274 	const struct got_error *err = NULL;
275 	char *line = NULL;
276 	char mode = *h->lines[0];
277 	size_t linesize = 0;
278 	ssize_t linelen;
279 	off_t match = -1;
280 	long match_lineno = -1;
281 
282 	for (;;) {
283 		linelen = getline(&line, &linesize, orig);
284 		if (linelen == -1) {
285 			if (ferror(orig))
286 				err = got_error_from_errno("getline");
287 			else if (match == -1)
288 				err = got_error(GOT_ERR_PATCH_DONT_APPLY);
289 			break;
290 		}
291 		(*lineno)++;
292 
293 		if ((mode == ' ' && !strcmp(h->lines[0]+1, line)) ||
294 		    (mode == '-' && !strcmp(h->lines[0]+1, line)) ||
295 		    (mode == '+' && *lineno == h->old_from)) {
296 			match = ftello(orig);
297 			if (match == -1) {
298 				err = got_error_from_errno("ftello");
299 				break;
300 			}
301 			match -= linelen;
302 			match_lineno = (*lineno)-1;
303 		}
304 
305 		if (*lineno >= h->old_from && match != -1)
306 			break;
307 	}
308 
309 	if (err == NULL) {
310 		*lineno = match_lineno;
311 		if (fseek(orig, match, SEEK_SET) == -1)
312 			err = got_error_from_errno("fseek");
313 	}
314 
315 	free(line);
316 	return err;
317 }
318 
319 static const struct got_error *
320 test_hunk(FILE *orig, struct got_patch_hunk *h)
321 {
322 	const struct got_error *err = NULL;
323 	char *line = NULL;
324 	size_t linesize = 0, i = 0;
325 	ssize_t linelen;
326 
327 	for (i = 0; i < h->len; ++i) {
328 		switch (*h->lines[i]) {
329 		case '+':
330 			continue;
331 		case ' ':
332 		case '-':
333 			linelen = getline(&line, &linesize, orig);
334 			if (linelen == -1) {
335 				if (ferror(orig))
336 					err = got_error_from_errno("getline");
337 				else
338 					err = got_error(
339 					    GOT_ERR_PATCH_DONT_APPLY);
340 				goto done;
341 			}
342 			if (strcmp(h->lines[i]+1, line)) {
343 				err = got_error(GOT_ERR_PATCH_DONT_APPLY);
344 				goto done;
345 			}
346 			break;
347 		}
348 	}
349 
350 done:
351 	free(line);
352 	return err;
353 }
354 
355 static const struct got_error *
356 apply_hunk(FILE *tmp, struct got_patch_hunk *h, long *lineno)
357 {
358 	size_t i = 0;
359 
360 	for (i = 0; i < h->len; ++i) {
361 		switch (*h->lines[i]) {
362 		case ' ':
363 			if (fprintf(tmp, "%s", h->lines[i]+1) < 0)
364 				return got_error_from_errno("fprintf");
365 			/* fallthrough */
366 		case '-':
367 			(*lineno)++;
368 			break;
369 		case '+':
370 			if (fprintf(tmp, "%s", h->lines[i]+1) < 0)
371 				return got_error_from_errno("fprintf");
372 			break;
373 		}
374 	}
375 	return NULL;
376 }
377 
378 static const struct got_error *
379 apply_patch(struct got_worktree *worktree, struct got_repository *repo,
380     struct got_patch *p, got_worktree_delete_cb delete_cb, void *delete_arg,
381     got_worktree_checkout_cb add_cb, void *add_arg)
382 {
383 	const struct got_error *err = NULL;
384 	struct got_pathlist_head paths;
385 	struct got_pathlist_entry *pe;
386 	char *path = NULL, *tmppath = NULL, *template = NULL;
387 	FILE *orig = NULL, *tmp = NULL;
388 	struct got_patch_hunk *h;
389 	size_t i;
390 	long lineno = 0;
391 	off_t copypos, pos;
392 	char *line = NULL;
393 	size_t linesize = 0;
394 	ssize_t linelen;
395 
396 	TAILQ_INIT(&paths);
397 
398 	if (p->old == NULL && p->new == NULL)
399 		return got_error(GOT_ERR_PATCH_MALFORMED);
400 
401 	err = got_worktree_resolve_path(&path, worktree,
402 	    p->new != NULL ? p->new : p->old);
403 	if (err)
404 		return err;
405 	err = got_pathlist_insert(&pe, &paths, path, NULL);
406 	if (err)
407 		goto done;
408 
409 	if (p->old != NULL && p->new == NULL) {
410 		/*
411 		 * special case: delete a file.  don't try to match
412 		 * the lines but just schedule the removal.
413 		 */
414 		err = got_worktree_schedule_delete(worktree, &paths,
415 		    0, NULL, delete_cb, delete_arg, repo, 0, 0);
416 		goto done;
417 	} else if (p->old != NULL && strcmp(p->old, p->new)) {
418 		err = got_error(GOT_ERR_PATCH_PATHS_DIFFER);
419 		goto done;
420 	}
421 
422 	if (asprintf(&template, "%s/got-patch",
423 	    got_worktree_get_root_path(worktree)) == -1) {
424 		err = got_error_from_errno(template);
425 		goto done;
426 	}
427 
428 	err = got_opentemp_named(&tmppath, &tmp, template);
429 	if (err)
430 		goto done;
431 
432 	if (p->old == NULL) {				/* create */
433 		h = STAILQ_FIRST(&p->head);
434 		if (h == NULL || STAILQ_NEXT(h, entries) != NULL) {
435 			err = got_error(GOT_ERR_PATCH_MALFORMED);
436 			goto done;
437 		}
438 		for (i = 0; i < h->len; ++i) {
439 			if (fprintf(tmp, "%s", h->lines[i]+1) < 0) {
440 				err = got_error_from_errno("fprintf");
441 				goto done;
442 			}
443 		}
444 		goto rename;
445 	}
446 
447 	if ((orig = fopen(path, "r")) == NULL) {
448 		err = got_error_from_errno2("fopen", path);
449 		goto done;
450 	}
451 
452 	copypos = 0;
453 	STAILQ_FOREACH(h, &p->head, entries) {
454 	tryagain:
455 		err = locate_hunk(orig, h, &lineno);
456 		if (err != NULL)
457 			goto done;
458 		if ((pos = ftello(orig)) == -1) {
459 			err = got_error_from_errno("ftello");
460 			goto done;
461 		}
462 		err = copy(tmp, orig, copypos, pos);
463 		if (err != NULL)
464 			goto done;
465 		copypos = pos;
466 
467 		err = test_hunk(orig, h);
468 		if (err != NULL && err->code == GOT_ERR_PATCH_DONT_APPLY) {
469 			/*
470 			 * try to apply the hunk again starting the search
471 			 * after the previous partial match.
472 			 */
473 			if (fseek(orig, pos, SEEK_SET) == -1) {
474 				err = got_error_from_errno("fseek");
475 				goto done;
476 			}
477 			linelen = getline(&line, &linesize, orig);
478 			if (linelen == -1) {
479 				err = got_error_from_errno("getline");
480 				goto done;
481 			}
482 			lineno++;
483 			goto tryagain;
484 		}
485 		if (err != NULL)
486 			goto done;
487 
488 		err = apply_hunk(tmp, h, &lineno);
489 		if (err != NULL)
490 			goto done;
491 		
492 		copypos = ftello(orig);
493 		if (copypos == -1) {
494 			err = got_error_from_errno("ftello");
495 			goto done;
496 		}
497 	}
498 
499 	if (!feof(orig)) {
500 		err = copy(tmp, orig, copypos, -1);
501 		if (err)
502 			goto done;
503 	}
504 
505 rename:
506 	if (rename(tmppath, path) == -1) {
507 		err = got_error_from_errno3("rename", tmppath, path);
508 		goto done;
509 	}
510 
511 	if (p->old == NULL)
512 		err = got_worktree_schedule_add(worktree, &paths,
513 		    add_cb, add_arg, repo, 1);
514 	else
515 		printf("M  %s\n", path); /* XXX */
516 done:
517 	free(template);
518 	if (err != NULL && p->old == NULL && path != NULL)
519 		unlink(path);
520 	if (tmp != NULL)
521 		fclose(tmp);
522 	if (tmppath != NULL)
523 		unlink(tmppath);
524 	free(tmppath);
525 	if (orig != NULL) {
526 		if (p->old == NULL && err != NULL)
527 			unlink(path);
528 		fclose(orig);
529 	}
530 	free(path);
531 	free(line);
532 	got_pathlist_free(&paths);
533 	return err;
534 }
535 
536 const struct got_error *
537 got_patch(int fd, struct got_worktree *worktree, struct got_repository *repo,
538     got_worktree_delete_cb delete_cb, void *delete_arg,
539     got_worktree_checkout_cb add_cb, void *add_arg)
540 {
541 	const struct got_error *err = NULL;
542 	struct imsgbuf *ibuf;
543 	int imsg_fds[2] = {-1, -1};
544 	int done = 0;
545 	pid_t pid;
546 
547 	ibuf = calloc(1, sizeof(*ibuf));
548 	if (ibuf == NULL) {
549 		err = got_error_from_errno("calloc");
550 		goto done;
551 	}
552 
553 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
554 		err = got_error_from_errno("socketpair");
555 		goto done;
556 	}
557 
558 	pid = fork();
559 	if (pid == -1) {
560 		err = got_error_from_errno("fork");
561 		goto done;
562 	} else if (pid == 0) {
563 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PATCH,
564 		    NULL);
565 		/* not reached */
566 	}
567 
568 	if (close(imsg_fds[1]) == -1) {
569 		err = got_error_from_errno("close");
570 		goto done;
571 	}
572 	imsg_fds[1] = -1;
573 	imsg_init(ibuf, imsg_fds[0]);
574 
575 	err = send_patch(ibuf, fd);
576 	fd = -1;
577 	if (err)
578 		goto done;
579 
580 	while (!done && err == NULL) {
581 		struct got_patch p;
582 
583 		err = recv_patch(ibuf, &done, &p);
584 		if (err || done)
585 			break;
586 
587 		err = apply_patch(worktree, repo, &p, delete_cb, delete_arg,
588 		    add_cb, add_arg);
589 		patch_free(&p);
590 		if (err)
591 			break;
592 	}
593 
594 done:
595 	if (fd != -1 && close(fd) == -1 && err == NULL)
596 		err = got_error_from_errno("close");
597 	if (ibuf != NULL)
598 		imsg_clear(ibuf);
599 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
600 		err = got_error_from_errno("close");
601 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
602 		err = got_error_from_errno("close");
603 	return err;
604 }