1 /*
2  * Copyright 1986, Larry Wall
3  *
4  * Redistribution and use in source and binary forms, with or without
5  * modification, are permitted provided that the following condition is met:
6  * 1. Redistributions of source code must retain the above copyright notice,
7  * this condition and the following disclaimer.
8  *
9  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND ANY
10  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
11  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
12  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
13  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
14  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
15  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
16  * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
17  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
18  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
19  * SUCH DAMAGE.
20  */
21 
22 /*
23  * Copyright (c) 2022 Omar Polo <op@openbsd.org>
24  *
25  * Permission to use, copy, modify, and distribute this software for any
26  * purpose with or without fee is hereby granted, provided that the above
27  * copyright notice and this permission notice appear in all copies.
28  *
29  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
30  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
31  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
32  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
33  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
34  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
35  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
36  */
37 #include "got_compat.h"
38 
39 #include <sys/types.h>
40 #include <sys/queue.h>
41 #include <sys/uio.h>
42 
43 #include <ctype.h>
44 #include <limits.h>
45 #include <paths.h>
46 #include <stdint.h>
47 #include <stdio.h>
48 #include <stdlib.h>
49 #include <string.h>
50 #include <unistd.h>
51 
52 #include "got_error.h"
53 #include "got_object.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_object.h"
57 #include "got_lib_privsep.h"
58 #include "got_lib_hash.h"
59 
60 struct imsgbuf ibuf;
61 
62 static const struct got_error *
63 send_patch(const char *oldname, const char *newname, const char *commitid,
64     const char *blob, const int xbit, int git)
65 {
66 	struct got_imsg_patch p;
67 
68 	memset(&p, 0, sizeof(p));
69 
70 	if (oldname != NULL)
71 		strlcpy(p.old, oldname, sizeof(p.old));
72 
73 	if (newname != NULL)
74 		strlcpy(p.new, newname, sizeof(p.new));
75 
76 	if (commitid != NULL)
77 		strlcpy(p.cid, commitid, sizeof(p.cid));
78 
79 	if (blob != NULL)
80 		strlcpy(p.blob, blob, sizeof(p.blob));
81 
82 	p.xbit = xbit;
83 	p.git = git;
84 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH, 0, 0, -1, &p, sizeof(p)) == -1)
85 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH");
86 	return NULL;
87 }
88 
89 static const struct got_error *
90 send_patch_done(void)
91 {
92 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_DONE, 0, 0, -1,
93 	    NULL, 0) == -1)
94 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
95 	return got_privsep_flush_imsg(&ibuf);
96 }
97 
98 /* based on fetchname from usr.bin/patch/util.c */
99 static const struct got_error *
100 filename(const char *at, char **name)
101 {
102 	char	*tmp, *t;
103 
104 	*name = NULL;
105 	if (*at == '\0')
106 		return NULL;
107 
108 	while (isspace((unsigned char)*at))
109 		at++;
110 
111 	/* files can be created or removed by diffing against /dev/null */
112 	if (!strncmp(at, _PATH_DEVNULL, sizeof(_PATH_DEVNULL) - 1))
113 		return NULL;
114 
115 	tmp = strdup(at);
116 	if (tmp == NULL)
117 		return got_error_from_errno("strdup");
118 	if ((t = strchr(tmp, '\t')) != NULL)
119 		*t = '\0';
120 	if ((t = strchr(tmp, '\n')) != NULL)
121 		*t = '\0';
122 
123 	*name = strdup(tmp);
124 	free(tmp);
125 	if (*name == NULL)
126 		return got_error_from_errno("strdup");
127 	return NULL;
128 }
129 
130 static int
131 binary_deleted(const char *line)
132 {
133 	const char *prefix = "Binary files ";
134 	const char *suffix = " and /dev/null differ\n";
135 	size_t len, d;
136 
137 	if (strncmp(line, prefix, strlen(prefix)) != 0)
138 		return 0;
139 	line += strlen(prefix);
140 
141 	len = strlen(line);
142 	if (len <= strlen(suffix))
143 		return 0;
144 	d = len - strlen(suffix);
145 	return (strcmp(line + d, suffix) == 0);
146 }
147 
148 static const struct got_error *
149 binaryfilename(const char *at, char **name)
150 {
151 	const char *suffix = " and /dev/null differ\n";
152 	size_t len, d;
153 
154 	*name = NULL;
155 
156 	len = strlen(at);
157 	if (len <= strlen(suffix))
158 		return NULL;
159 
160 	d = len - strlen(suffix);
161 	if (strcmp(at + d, suffix) != 0)
162 		return NULL;
163 
164 	*name = strndup(at, d);
165 	if (*name == NULL)
166 		return got_error_from_errno("strndup");
167 	return NULL;
168 }
169 
170 static int
171 filexbit(const char *line)
172 {
173 	char *m;
174 
175 	m = strchr(line, '(');
176 	if (m && !strncmp(m + 1, "mode ", 5))
177 		return strncmp(m + 6, "755", 3) == 0;
178 
179 	return 0;
180 }
181 
182 static const struct got_error *
183 blobid(const char *line, char **blob, int git)
184 {
185 	uint8_t digest[SHA1_DIGEST_LENGTH];
186 	size_t len;
187 
188 	*blob = NULL;
189 
190 	len = strspn(line, "0123456789abcdefABCDEF");
191 	if ((*blob = strndup(line, len)) == NULL)
192 		return got_error_from_errno("strndup");
193 
194 	if (!git && !got_parse_hash_digest(digest, *blob, GOT_HASH_SHA1)) {
195 		/* silently ignore invalid blob ids */
196 		free(*blob);
197 		*blob = NULL;
198 	}
199 	return NULL;
200 }
201 
202 static const struct got_error *
203 patch_start(int *git, char **cid, FILE *fp)
204 {
205 	const struct got_error *err = NULL;
206 	char	*line = NULL;
207 	size_t	 linesize = 0;
208 	ssize_t	 linelen;
209 
210 	*git = 0;
211 
212 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
213 		if (!strncmp(line, "diff --git ", 11)) {
214 			*git = 1;
215 			free(*cid);
216 			*cid = NULL;
217 			break;
218 		} else if (!strncmp(line, "diff ", 5)) {
219 			*git = 0;
220 			free(*cid);
221 			*cid = NULL;
222 		} else if (!strncmp(line, "commit - ", 9)) {
223 			free(*cid);
224 			err = blobid(line + 9, cid, *git);
225 			if (err)
226 				break;
227 		} else if (!strncmp(line, "--- ", 4) ||
228 		    !strncmp(line, "+++ ", 4) ||
229 		    !strncmp(line, "blob - ", 7) ||
230 		    binary_deleted(line)) {
231 			/* rewind to previous line */
232 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
233 				err = got_error_from_errno("fseeko");
234 			break;
235 		}
236 	}
237 
238 	free(line);
239 	if (ferror(fp) && err == NULL)
240 		err = got_error_from_errno("getline");
241 	if (feof(fp) && err == NULL)
242 		err = got_error(GOT_ERR_NO_PATCH);
243 	return err;
244 }
245 
246 static const struct got_error *
247 find_diff(int *done, int *next, FILE *fp, int git, const char *commitid)
248 {
249 	const struct got_error *err = NULL;
250 	char	*old = NULL, *new = NULL;
251 	char	*blob = NULL;
252 	char	*line = NULL;
253 	size_t	 linesize = 0;
254 	ssize_t	 linelen;
255 	int	 create, delete_binary = 0, rename = 0, xbit = 0;
256 
257 	*done = 0;
258 	*next = 0;
259 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
260 		/*
261 		 * Ignore the Index name like GNU and larry' patch,
262 		 * we don't have to follow POSIX.
263 		 */
264 
265 		if (!strncmp(line, "--- ", 4)) {
266 			free(old);
267 			err = filename(line+4, &old);
268 		} else if (rename && !strncmp(line, "rename from ", 12)) {
269 			free(old);
270 			err = filename(line+12, &old);
271 		} else if (!strncmp(line, "+++ ", 4)) {
272 			free(new);
273 			err = filename(line+4, &new);
274 		} else if (!strncmp(line, "blob + ", 7) ||
275 		    !strncmp(line, "file + ", 7)) {
276 			xbit = filexbit(line);
277 		} else if (!git && !strncmp(line, "blob - ", 7)) {
278 			free(blob);
279 			err = blobid(line + 7, &blob, git);
280 		} else if (!strncmp(line, "Binary files ", 13)) {
281 			delete_binary = 1;
282 			free(old);
283 			err = binaryfilename(line + 13, &old);
284 		} else if (rename && !strncmp(line, "rename to ", 10)) {
285 			free(new);
286 			err = filename(line + 10, &new);
287 		} else if (git && !strncmp(line, "similarity index 100%", 21))
288 			rename = 1;
289 		else if (git && !strncmp(line, "new file mode 100", 17))
290 			xbit = strncmp(line + 17, "755", 3) == 0;
291 		else if (git && !strncmp(line, "index ", 6)) {
292 			free(blob);
293 			err = blobid(line + 6, &blob, git);
294 		} else if (!strncmp(line, "diff ", 5)) {
295 			/* rewind to previous line */
296 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
297 				err = got_error_from_errno("fseeko");
298 			*next = 1;
299 			break;
300 		}
301 
302 		if (err)
303 			break;
304 
305 		/*
306 		 * Git-style diffs with "similarity index 100%" don't
307 		 * have any hunks and ends with the "rename to foobar"
308 		 * line.
309 		 */
310 		if (rename && old != NULL && new != NULL) {
311 			*done = 1;
312 			err = send_patch(old, new, commitid,
313 			    blob, xbit, git);
314 			break;
315 		}
316 
317 		/*
318 		 * Diffs that remove binary files have no hunks.
319 		 */
320 		if (delete_binary && old != NULL) {
321 			*done = 1;
322 			err = send_patch(old, new, commitid,
323 			    blob, xbit, git);
324 			break;
325 		}
326 
327 		if (!strncmp(line, "@@ -", 4)) {
328 			create = !strncmp(line+4, "0,0", 3);
329 			if ((old == NULL && new == NULL) ||
330 			    (!create && old == NULL))
331 				err = got_error(GOT_ERR_PATCH_MALFORMED);
332 			else
333 				err = send_patch(old, new, commitid,
334 				    blob, xbit, git);
335 
336 			if (err)
337 				break;
338 
339 			/* rewind to previous line */
340 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
341 				err = got_error_from_errno("fseeko");
342 			break;
343 		}
344 	}
345 
346 	free(old);
347 	free(new);
348 	free(blob);
349 	free(line);
350 	if (ferror(fp) && err == NULL)
351 		err = got_error_from_errno("getline");
352 	if (feof(fp) && err == NULL)
353 		err = got_error(GOT_ERR_NO_PATCH);
354 	return err;
355 }
356 
357 static const struct got_error *
358 strtolnum(char **str, int *n)
359 {
360 	char		*p, c;
361 	const char	*errstr;
362 
363 	for (p = *str; isdigit((unsigned char)*p); ++p)
364 		/* nop */;
365 
366 	c = *p;
367 	*p = '\0';
368 
369 	*n = strtonum(*str, 0, INT_MAX, &errstr);
370 	if (errstr != NULL)
371 		return got_error(GOT_ERR_PATCH_MALFORMED);
372 
373 	*p = c;
374 	*str = p;
375 	return NULL;
376 }
377 
378 static const struct got_error *
379 parse_hdr(char *s, int *done, struct got_imsg_patch_hunk *hdr)
380 {
381 	static const struct got_error *err = NULL;
382 
383 	if (strncmp(s, "@@ -", 4)) {
384 		*done = 1;
385 		return NULL;
386 	}
387 
388 	s += 4;
389 	if (!*s)
390 		return NULL;
391 	err = strtolnum(&s, &hdr->oldfrom);
392 	if (err)
393 		return err;
394 	if (*s == ',') {
395 		s++;
396 		err = strtolnum(&s, &hdr->oldlines);
397 		if (err)
398 			return err;
399 	} else
400 		hdr->oldlines = 1;
401 
402 	if (*s == ' ')
403 		s++;
404 
405 	if (*s != '+' || !*++s)
406 		return got_error(GOT_ERR_PATCH_MALFORMED);
407 	err = strtolnum(&s, &hdr->newfrom);
408 	if (err)
409 		return err;
410 	if (*s == ',') {
411 		s++;
412 		err = strtolnum(&s, &hdr->newlines);
413 		if (err)
414 			return err;
415 	} else
416 		hdr->newlines = 1;
417 
418 	if (*s == ' ')
419 		s++;
420 
421 	if (*s != '@')
422 		return got_error(GOT_ERR_PATCH_MALFORMED);
423 
424 	if (hdr->oldfrom >= INT_MAX - hdr->oldlines ||
425 	    hdr->newfrom >= INT_MAX - hdr->newlines ||
426 	    /* not so sure about this one */
427 	    hdr->oldlines >= INT_MAX - hdr->newlines - 1 ||
428 	    (hdr->oldlines == 0 && hdr->newlines == 0))
429 		return got_error(GOT_ERR_PATCH_MALFORMED);
430 
431 	if (hdr->oldlines == 0) {
432 		/* larry says to "do append rather than insert"; I don't
433 		 * quite get it, but i trust him.
434 		 */
435 		hdr->oldfrom++;
436 	}
437 
438 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_HUNK, 0, 0, -1,
439 	    hdr, sizeof(*hdr)) == -1)
440 		return got_error_from_errno(
441 		    "imsg_compose GOT_IMSG_PATCH_HUNK");
442 	return NULL;
443 }
444 
445 static const struct got_error *
446 send_line(const char *line)
447 {
448 	static const struct got_error *err = NULL;
449 	char *p = NULL;
450 
451 	if (*line != '+' && *line != '-' && *line != ' ' && *line != '\\') {
452 		if (asprintf(&p, " %s", line) == -1)
453 			return got_error_from_errno("asprintf");
454 		line = p;
455 	}
456 
457 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_LINE, 0, 0, -1,
458 	    line, strlen(line) + 1) == -1)
459 		err = got_error_from_errno(
460 		    "imsg_compose GOT_IMSG_PATCH_LINE");
461 
462 	free(p);
463 	return err;
464 }
465 
466 static const struct got_error *
467 peek_special_line(FILE *fp)
468 {
469 	const struct got_error *err;
470 	int ch;
471 
472 	ch = fgetc(fp);
473 	if (ch != EOF && ch != '\\') {
474 		ungetc(ch, fp);
475 		return NULL;
476 	}
477 
478 	if (ch == '\\') {
479 		err = send_line("\\");
480 		if (err)
481 			return err;
482 	}
483 
484 	while (ch != EOF && ch != '\n')
485 		ch = fgetc(fp);
486 
487 	if (ch != EOF || feof(fp))
488 		return NULL;
489 	return got_error(GOT_ERR_IO);
490 }
491 
492 static const struct got_error *
493 parse_hunk(FILE *fp, int *done)
494 {
495 	static const struct got_error *err = NULL;
496 	struct got_imsg_patch_hunk hdr;
497 	char	*line = NULL, ch;
498 	size_t	 linesize = 0;
499 	ssize_t	 linelen;
500 	int	 leftold, leftnew;
501 
502 	linelen = getline(&line, &linesize, fp);
503 	if (linelen == -1) {
504 		*done = 1;
505 		goto done;
506 	}
507 
508 	err = parse_hdr(line, done, &hdr);
509 	if (err)
510 		goto done;
511 	if (*done) {
512 		if (fseeko(fp, -linelen, SEEK_CUR) == -1)
513 			err = got_error_from_errno("fseeko");
514 		goto done;
515 	}
516 
517 	leftold = hdr.oldlines;
518 	leftnew = hdr.newlines;
519 
520 	while (leftold > 0 || leftnew > 0) {
521 		linelen = getline(&line, &linesize, fp);
522 		if (linelen == -1) {
523 			if (ferror(fp)) {
524 				err = got_error_from_errno("getline");
525 				goto done;
526 			}
527 
528 			/* trailing newlines may be chopped */
529 			if (leftold < 3 && leftnew < 3) {
530 				*done = 1;
531 				break;
532 			}
533 
534 			err = got_error(GOT_ERR_PATCH_TRUNCATED);
535 			goto done;
536 		}
537 		if (line[linelen - 1] == '\n')
538 			line[linelen - 1] = '\0';
539 
540 		/* usr.bin/patch allows '=' as context char */
541 		if (*line == '=')
542 			*line = ' ';
543 
544 		ch = *line;
545 		if (ch == '\t' || ch == '\0')
546 			ch = ' ';	/* the space got eaten */
547 
548 		switch (ch) {
549 		case '-':
550 			leftold--;
551 			break;
552 		case ' ':
553 			leftold--;
554 			leftnew--;
555 			break;
556 		case '+':
557 			leftnew--;
558 			break;
559 		default:
560 			err = got_error(GOT_ERR_PATCH_MALFORMED);
561 			goto done;
562 		}
563 
564 		if (leftold < 0 || leftnew < 0) {
565 			err = got_error(GOT_ERR_PATCH_MALFORMED);
566 			goto done;
567 		}
568 
569 		err = send_line(line);
570 		if (err)
571 			goto done;
572 
573 		if ((ch == '-' && leftold == 0) ||
574 		    (ch == '+' && leftnew == 0)) {
575 			err = peek_special_line(fp);
576 			if (err)
577 				goto done;
578 		}
579 	}
580 
581 done:
582 	free(line);
583 	return err;
584 }
585 
586 static const struct got_error *
587 read_patch(struct imsgbuf *ibuf, int fd)
588 {
589 	const struct got_error *err = NULL;
590 	FILE *fp;
591 	int git, patch_found = 0;
592 	char *cid = NULL;
593 
594 	if ((fp = fdopen(fd, "r")) == NULL) {
595 		err = got_error_from_errno("fdopen");
596 		close(fd);
597 		return err;
598 	}
599 
600 	while ((err = patch_start(&git, &cid, fp)) == NULL) {
601 		int done, next;
602 
603 		err = find_diff(&done, &next, fp, git, cid);
604 		if (err)
605 			goto done;
606 		if (next)
607 			continue;
608 
609 		patch_found = 1;
610 
611 		while (!done) {
612 			err = parse_hunk(fp, &done);
613 			if (err)
614 				goto done;
615 		}
616 
617 		err = send_patch_done();
618 		if (err)
619 			goto done;
620 	}
621 
622 done:
623 	fclose(fp);
624 	free(cid);
625 
626 	/* ignore trailing gibberish */
627 	if (err != NULL && err->code == GOT_ERR_NO_PATCH && patch_found)
628 		err = NULL;
629 
630 	return err;
631 }
632 
633 int
634 main(int argc, char **argv)
635 {
636 	const struct got_error *err = NULL;
637 	struct imsg imsg;
638 #if 0
639 	static int attached;
640 	while (!attached)
641 		sleep(1);
642 #endif
643 
644 	imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
645 #ifndef PROFILE
646 	/* revoke access to most system calls */
647 	if (pledge("stdio recvfd", NULL) == -1) {
648 		err = got_error_from_errno("pledge");
649 		got_privsep_send_error(&ibuf, err);
650 		return 1;
651 	}
652 
653 	/* revoke fs access */
654 	if (landlock_no_fs() == -1) {
655 		err = got_error_from_errno("landlock_no_fs");
656 		got_privsep_send_error(&ibuf, err);
657 		return 1;
658 	}
659 	if (cap_enter() == -1) {
660 		err = got_error_from_errno("cap_enter");
661 		got_privsep_send_error(&ibuf, err);
662 		return 1;
663 	}
664 #endif
665 
666 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
667 	if (err)
668 		goto done;
669 	if (imsg.hdr.type != GOT_IMSG_PATCH_FILE || imsg.fd == -1) {
670 		err = got_error(GOT_ERR_PRIVSEP_MSG);
671 		goto done;
672 	}
673 
674 	err = read_patch(&ibuf, imsg.fd);
675 	if (err)
676 		goto done;
677 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_EOF, 0, 0, -1,
678 	    NULL, 0) == -1) {
679 		err = got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
680 		goto done;
681 	}
682 	err = got_privsep_flush_imsg(&ibuf);
683 done:
684 	imsg_free(&imsg);
685 	if (err != NULL) {
686 		got_privsep_send_error(&ibuf, err);
687 		err = NULL;
688 	}
689 	if (close(GOT_IMSG_FD_CHILD) == -1 && err == NULL)
690 		err = got_error_from_errno("close");
691 	if (err && err->code != GOT_ERR_PRIVSEP_PIPE)
692 		fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
693 	return err ? 1 : 0;
694 }