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 
38 #include <sys/types.h>
39 #include <sys/queue.h>
40 #include <sys/uio.h>
41 
42 #include <ctype.h>
43 #include <limits.h>
44 #include <paths.h>
45 #include <sha1.h>
46 #include <sha2.h>
47 #include <stdint.h>
48 #include <stdio.h>
49 #include <stdlib.h>
50 #include <string.h>
51 #include <unistd.h>
52 #include <imsg.h>
53 
54 #include "got_error.h"
55 #include "got_object.h"
56 
57 #include "got_lib_delta.h"
58 #include "got_lib_object.h"
59 #include "got_lib_privsep.h"
60 #include "got_lib_hash.h"
61 
62 struct imsgbuf ibuf;
63 
64 static const struct got_error *
65 send_patch(const char *oldname, const char *newname, const char *commitid,
66     const char *blob, const int xbit, int git)
67 {
68 	struct got_imsg_patch p;
69 
70 	memset(&p, 0, sizeof(p));
71 
72 	if (oldname != NULL)
73 		strlcpy(p.old, oldname, sizeof(p.old));
74 
75 	if (newname != NULL)
76 		strlcpy(p.new, newname, sizeof(p.new));
77 
78 	if (commitid != NULL)
79 		strlcpy(p.cid, commitid, sizeof(p.cid));
80 
81 	if (blob != NULL)
82 		strlcpy(p.blob, blob, sizeof(p.blob));
83 
84 	p.xbit = xbit;
85 	p.git = git;
86 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH, 0, 0, -1, &p, sizeof(p)) == -1)
87 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH");
88 	return NULL;
89 }
90 
91 static const struct got_error *
92 send_patch_done(void)
93 {
94 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_DONE, 0, 0, -1,
95 	    NULL, 0) == -1)
96 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
97 	return got_privsep_flush_imsg(&ibuf);
98 }
99 
100 /* based on fetchname from usr.bin/patch/util.c */
101 static const struct got_error *
102 filename(const char *at, char **name)
103 {
104 	char	*tmp, *t;
105 
106 	*name = NULL;
107 	if (*at == '\0')
108 		return NULL;
109 
110 	while (isspace((unsigned char)*at))
111 		at++;
112 
113 	/* files can be created or removed by diffing against /dev/null */
114 	if (!strncmp(at, _PATH_DEVNULL, sizeof(_PATH_DEVNULL) - 1))
115 		return NULL;
116 
117 	tmp = strdup(at);
118 	if (tmp == NULL)
119 		return got_error_from_errno("strdup");
120 	if ((t = strchr(tmp, '\t')) != NULL)
121 		*t = '\0';
122 	if ((t = strchr(tmp, '\n')) != NULL)
123 		*t = '\0';
124 
125 	*name = strdup(tmp);
126 	free(tmp);
127 	if (*name == NULL)
128 		return got_error_from_errno("strdup");
129 	return NULL;
130 }
131 
132 static int
133 binary_deleted(const char *line)
134 {
135 	const char *prefix = "Binary files ";
136 	const char *suffix = " and /dev/null differ\n";
137 	size_t len, d;
138 
139 	if (strncmp(line, prefix, strlen(prefix)) != 0)
140 		return 0;
141 	line += strlen(prefix);
142 
143 	len = strlen(line);
144 	if (len <= strlen(suffix))
145 		return 0;
146 	d = len - strlen(suffix);
147 	return (strcmp(line + d, suffix) == 0);
148 }
149 
150 static const struct got_error *
151 binaryfilename(const char *at, char **name)
152 {
153 	const char *suffix = " and /dev/null differ\n";
154 	size_t len, d;
155 
156 	*name = NULL;
157 
158 	len = strlen(at);
159 	if (len <= strlen(suffix))
160 		return NULL;
161 
162 	d = len - strlen(suffix);
163 	if (strcmp(at + d, suffix) != 0)
164 		return NULL;
165 
166 	*name = strndup(at, d);
167 	if (*name == NULL)
168 		return got_error_from_errno("strndup");
169 	return NULL;
170 }
171 
172 static int
173 filexbit(const char *line)
174 {
175 	char *m;
176 
177 	m = strchr(line, '(');
178 	if (m && !strncmp(m + 1, "mode ", 5))
179 		return strncmp(m + 6, "755", 3) == 0;
180 
181 	return 0;
182 }
183 
184 static const struct got_error *
185 blobid(const char *line, char **blob, int git)
186 {
187 	uint8_t digest[SHA1_DIGEST_LENGTH];
188 	size_t len;
189 
190 	*blob = NULL;
191 
192 	len = strspn(line, "0123456789abcdefABCDEF");
193 	if ((*blob = strndup(line, len)) == NULL)
194 		return got_error_from_errno("strndup");
195 
196 	if (!git && !got_parse_hash_digest(digest, *blob, GOT_HASH_SHA1)) {
197 		/* silently ignore invalid blob ids */
198 		free(*blob);
199 		*blob = NULL;
200 	}
201 	return NULL;
202 }
203 
204 static const struct got_error *
205 patch_start(int *git, char **cid, FILE *fp)
206 {
207 	const struct got_error *err = NULL;
208 	char	*line = NULL;
209 	size_t	 linesize = 0;
210 	ssize_t	 linelen;
211 
212 	*git = 0;
213 
214 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
215 		if (!strncmp(line, "diff --git ", 11)) {
216 			*git = 1;
217 			free(*cid);
218 			*cid = NULL;
219 			break;
220 		} else if (!strncmp(line, "diff ", 5)) {
221 			*git = 0;
222 			free(*cid);
223 			*cid = NULL;
224 		} else if (!strncmp(line, "commit - ", 9)) {
225 			free(*cid);
226 			err = blobid(line + 9, cid, *git);
227 			if (err)
228 				break;
229 		} else if (!strncmp(line, "--- ", 4) ||
230 		    !strncmp(line, "+++ ", 4) ||
231 		    !strncmp(line, "blob - ", 7) ||
232 		    binary_deleted(line)) {
233 			/* rewind to previous line */
234 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
235 				err = got_error_from_errno("fseeko");
236 			break;
237 		}
238 	}
239 
240 	free(line);
241 	if (ferror(fp) && err == NULL)
242 		err = got_error_from_errno("getline");
243 	if (feof(fp) && err == NULL)
244 		err = got_error(GOT_ERR_NO_PATCH);
245 	return err;
246 }
247 
248 static const struct got_error *
249 find_diff(int *done, int *next, FILE *fp, int git, const char *commitid)
250 {
251 	const struct got_error *err = NULL;
252 	char	*old = NULL, *new = NULL;
253 	char	*blob = NULL;
254 	char	*line = NULL;
255 	size_t	 linesize = 0;
256 	ssize_t	 linelen;
257 	int	 create, delete_binary = 0, rename = 0, xbit = 0;
258 
259 	*done = 0;
260 	*next = 0;
261 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
262 		/*
263 		 * Ignore the Index name like GNU and larry' patch,
264 		 * we don't have to follow POSIX.
265 		 */
266 
267 		if (!strncmp(line, "--- ", 4)) {
268 			free(old);
269 			err = filename(line+4, &old);
270 		} else if (rename && !strncmp(line, "rename from ", 12)) {
271 			free(old);
272 			err = filename(line+12, &old);
273 		} else if (!strncmp(line, "+++ ", 4)) {
274 			free(new);
275 			err = filename(line+4, &new);
276 		} else if (!strncmp(line, "blob + ", 7) ||
277 		    !strncmp(line, "file + ", 7)) {
278 			xbit = filexbit(line);
279 		} else if (!git && !strncmp(line, "blob - ", 7)) {
280 			free(blob);
281 			err = blobid(line + 7, &blob, git);
282 		} else if (!strncmp(line, "Binary files ", 13)) {
283 			delete_binary = 1;
284 			free(old);
285 			err = binaryfilename(line + 13, &old);
286 		} else if (rename && !strncmp(line, "rename to ", 10)) {
287 			free(new);
288 			err = filename(line + 10, &new);
289 		} else if (git && !strncmp(line, "similarity index 100%", 21))
290 			rename = 1;
291 		else if (git && !strncmp(line, "new file mode 100", 17))
292 			xbit = strncmp(line + 17, "755", 3) == 0;
293 		else if (git && !strncmp(line, "index ", 6)) {
294 			free(blob);
295 			err = blobid(line + 6, &blob, git);
296 		} else if (!strncmp(line, "diff ", 5)) {
297 			/* rewind to previous line */
298 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
299 				err = got_error_from_errno("fseeko");
300 			*next = 1;
301 			break;
302 		}
303 
304 		if (err)
305 			break;
306 
307 		/*
308 		 * Git-style diffs with "similarity index 100%" don't
309 		 * have any hunks and ends with the "rename to foobar"
310 		 * line.
311 		 */
312 		if (rename && old != NULL && new != NULL) {
313 			*done = 1;
314 			err = send_patch(old, new, commitid,
315 			    blob, xbit, git);
316 			break;
317 		}
318 
319 		/*
320 		 * Diffs that remove binary files have no hunks.
321 		 */
322 		if (delete_binary && old != NULL) {
323 			*done = 1;
324 			err = send_patch(old, new, commitid,
325 			    blob, xbit, git);
326 			break;
327 		}
328 
329 		if (!strncmp(line, "@@ -", 4)) {
330 			create = !strncmp(line+4, "0,0", 3);
331 			if ((old == NULL && new == NULL) ||
332 			    (!create && old == NULL))
333 				err = got_error(GOT_ERR_PATCH_MALFORMED);
334 			else
335 				err = send_patch(old, new, commitid,
336 				    blob, xbit, git);
337 
338 			if (err)
339 				break;
340 
341 			/* rewind to previous line */
342 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
343 				err = got_error_from_errno("fseeko");
344 			break;
345 		}
346 	}
347 
348 	free(old);
349 	free(new);
350 	free(blob);
351 	free(line);
352 	if (ferror(fp) && err == NULL)
353 		err = got_error_from_errno("getline");
354 	if (feof(fp) && err == NULL)
355 		err = got_error(GOT_ERR_NO_PATCH);
356 	return err;
357 }
358 
359 static const struct got_error *
360 strtolnum(char **str, int *n)
361 {
362 	char		*p, c;
363 	const char	*errstr;
364 
365 	for (p = *str; isdigit((unsigned char)*p); ++p)
366 		/* nop */;
367 
368 	c = *p;
369 	*p = '\0';
370 
371 	*n = strtonum(*str, 0, INT_MAX, &errstr);
372 	if (errstr != NULL)
373 		return got_error(GOT_ERR_PATCH_MALFORMED);
374 
375 	*p = c;
376 	*str = p;
377 	return NULL;
378 }
379 
380 static const struct got_error *
381 parse_hdr(char *s, int *done, struct got_imsg_patch_hunk *hdr)
382 {
383 	static const struct got_error *err = NULL;
384 
385 	if (strncmp(s, "@@ -", 4)) {
386 		*done = 1;
387 		return NULL;
388 	}
389 
390 	s += 4;
391 	if (!*s)
392 		return NULL;
393 	err = strtolnum(&s, &hdr->oldfrom);
394 	if (err)
395 		return err;
396 	if (*s == ',') {
397 		s++;
398 		err = strtolnum(&s, &hdr->oldlines);
399 		if (err)
400 			return err;
401 	} else
402 		hdr->oldlines = 1;
403 
404 	if (*s == ' ')
405 		s++;
406 
407 	if (*s != '+' || !*++s)
408 		return got_error(GOT_ERR_PATCH_MALFORMED);
409 	err = strtolnum(&s, &hdr->newfrom);
410 	if (err)
411 		return err;
412 	if (*s == ',') {
413 		s++;
414 		err = strtolnum(&s, &hdr->newlines);
415 		if (err)
416 			return err;
417 	} else
418 		hdr->newlines = 1;
419 
420 	if (*s == ' ')
421 		s++;
422 
423 	if (*s != '@')
424 		return got_error(GOT_ERR_PATCH_MALFORMED);
425 
426 	if (hdr->oldfrom >= INT_MAX - hdr->oldlines ||
427 	    hdr->newfrom >= INT_MAX - hdr->newlines ||
428 	    /* not so sure about this one */
429 	    hdr->oldlines >= INT_MAX - hdr->newlines - 1 ||
430 	    (hdr->oldlines == 0 && hdr->newlines == 0))
431 		return got_error(GOT_ERR_PATCH_MALFORMED);
432 
433 	if (hdr->oldlines == 0) {
434 		/* larry says to "do append rather than insert"; I don't
435 		 * quite get it, but i trust him.
436 		 */
437 		hdr->oldfrom++;
438 	}
439 
440 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_HUNK, 0, 0, -1,
441 	    hdr, sizeof(*hdr)) == -1)
442 		return got_error_from_errno(
443 		    "imsg_compose GOT_IMSG_PATCH_HUNK");
444 	return NULL;
445 }
446 
447 static const struct got_error *
448 send_line(const char *line, size_t len)
449 {
450 	static const struct got_error *err = NULL;
451 	struct iovec iov[2];
452 	int iovcnt = 0;
453 
454 	memset(&iov, 0, sizeof(iov));
455 
456 	if (*line != '+' && *line != '-' && *line != ' ' && *line != '\\') {
457 		iov[iovcnt].iov_base = (void *)" ";
458 		iov[iovcnt].iov_len = 1;
459 		iovcnt++;
460 	}
461 
462 	iov[iovcnt].iov_base = (void *)line;
463 	iov[iovcnt].iov_len = len;
464 	iovcnt++;
465 
466 	if (imsg_composev(&ibuf, GOT_IMSG_PATCH_LINE, 0, 0, -1,
467 	    iov, iovcnt) == -1)
468 		err = got_error_from_errno(
469 		    "imsg_compose GOT_IMSG_PATCH_LINE");
470 
471 	return err;
472 }
473 
474 static const struct got_error *
475 peek_special_line(FILE *fp)
476 {
477 	const struct got_error *err;
478 	int ch;
479 
480 	ch = fgetc(fp);
481 	if (ch != EOF && ch != '\\') {
482 		ungetc(ch, fp);
483 		return NULL;
484 	}
485 
486 	if (ch == '\\') {
487 		err = send_line("\\", 2);
488 		if (err)
489 			return err;
490 	}
491 
492 	while (ch != EOF && ch != '\n')
493 		ch = fgetc(fp);
494 
495 	if (ch != EOF || feof(fp))
496 		return NULL;
497 	return got_error(GOT_ERR_IO);
498 }
499 
500 static const struct got_error *
501 parse_hunk(FILE *fp, int *done)
502 {
503 	static const struct got_error *err = NULL;
504 	struct got_imsg_patch_hunk hdr;
505 	char	*line = NULL, ch;
506 	size_t	 linesize = 0;
507 	ssize_t	 linelen;
508 	int	 leftold, leftnew;
509 
510 	linelen = getline(&line, &linesize, fp);
511 	if (linelen == -1) {
512 		*done = 1;
513 		goto done;
514 	}
515 
516 	err = parse_hdr(line, done, &hdr);
517 	if (err)
518 		goto done;
519 	if (*done) {
520 		if (fseeko(fp, -linelen, SEEK_CUR) == -1)
521 			err = got_error_from_errno("fseeko");
522 		goto done;
523 	}
524 
525 	leftold = hdr.oldlines;
526 	leftnew = hdr.newlines;
527 
528 	while (leftold > 0 || leftnew > 0) {
529 		linelen = getline(&line, &linesize, fp);
530 		if (linelen == -1) {
531 			if (ferror(fp)) {
532 				err = got_error_from_errno("getline");
533 				goto done;
534 			}
535 
536 			/* trailing newlines may be chopped */
537 			if (leftold < 3 && leftnew < 3) {
538 				*done = 1;
539 				break;
540 			}
541 
542 			err = got_error(GOT_ERR_PATCH_TRUNCATED);
543 			goto done;
544 		}
545 		if (line[linelen - 1] == '\n')
546 			line[linelen - 1] = '\0';
547 
548 		/* usr.bin/patch allows '=' as context char */
549 		if (*line == '=')
550 			*line = ' ';
551 
552 		ch = *line;
553 		if (ch == '\t' || ch == '\0')
554 			ch = ' ';	/* the space got eaten */
555 
556 		switch (ch) {
557 		case '-':
558 			leftold--;
559 			break;
560 		case ' ':
561 			leftold--;
562 			leftnew--;
563 			break;
564 		case '+':
565 			leftnew--;
566 			break;
567 		default:
568 			err = got_error(GOT_ERR_PATCH_MALFORMED);
569 			goto done;
570 		}
571 
572 		if (leftold < 0 || leftnew < 0) {
573 			err = got_error(GOT_ERR_PATCH_MALFORMED);
574 			goto done;
575 		}
576 
577 		err = send_line(line, linelen);
578 		if (err)
579 			goto done;
580 
581 		if ((ch == '-' && leftold == 0) ||
582 		    (ch == '+' && leftnew == 0)) {
583 			err = peek_special_line(fp);
584 			if (err)
585 				goto done;
586 		}
587 	}
588 
589 done:
590 	free(line);
591 	return err;
592 }
593 
594 static const struct got_error *
595 read_patch(struct imsgbuf *ibuf, FILE *fp)
596 {
597 	const struct got_error *err = NULL;
598 	int git, patch_found = 0;
599 	char *cid = NULL;
600 
601 	while ((err = patch_start(&git, &cid, fp)) == NULL) {
602 		int done, next;
603 
604 		err = find_diff(&done, &next, fp, git, cid);
605 		if (err)
606 			goto done;
607 		if (next)
608 			continue;
609 
610 		patch_found = 1;
611 
612 		while (!done) {
613 			err = parse_hunk(fp, &done);
614 			if (err)
615 				goto done;
616 		}
617 
618 		err = send_patch_done();
619 		if (err)
620 			goto done;
621 	}
622 
623 done:
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 	FILE *fp = NULL;
639 	int fd = -1;
640 #if 0
641 	static int attached;
642 	while (!attached)
643 		sleep(1);
644 #endif
645 
646 	imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
647 #ifndef PROFILE
648 	/* revoke access to most system calls */
649 	if (pledge("stdio recvfd", NULL) == -1) {
650 		err = got_error_from_errno("pledge");
651 		got_privsep_send_error(&ibuf, err);
652 		return 1;
653 	}
654 #endif
655 
656 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
657 	if (err)
658 		goto done;
659 	if (imsg.hdr.type != GOT_IMSG_PATCH_FILE) {
660 		err = got_error(GOT_ERR_PRIVSEP_MSG);
661 		goto done;
662 	}
663 	fd = imsg_get_fd(&imsg);
664 	if (fd == -1) {
665 		err = got_error(GOT_ERR_PRIVSEP_NO_FD);
666 		goto done;
667 	}
668 
669 	fp = fdopen(fd, "r");
670 	if (fp == NULL) {
671 		err = got_error_from_errno("fdopen");
672 		goto done;
673 	}
674 	fd = -1;
675 
676 	err = read_patch(&ibuf, fp);
677 	if (err)
678 		goto done;
679 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_EOF, 0, 0, -1,
680 	    NULL, 0) == -1) {
681 		err = got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
682 		goto done;
683 	}
684 	err = got_privsep_flush_imsg(&ibuf);
685 	imsg_free(&imsg);
686 done:
687 	if (fd != -1 && close(fd) == -1 && err == NULL)
688 		err = got_error_from_errno("close");
689 	if (fp != NULL && fclose(fp) == EOF && err == NULL)
690 		err = got_error_from_errno("fclose");
691 	if (err != NULL) {
692 		got_privsep_send_error(&ibuf, err);
693 		err = NULL;
694 	}
695 	if (close(GOT_IMSG_FD_CHILD) == -1 && err == NULL)
696 		err = got_error_from_errno("close");
697 	if (err && err->code != GOT_ERR_PRIVSEP_PIPE)
698 		fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
699 	return err ? 1 : 0;
700 }