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/uio.h>
40 
41 #include <ctype.h>
42 #include <limits.h>
43 #include <paths.h>
44 #include <stdint.h>
45 #include <stdio.h>
46 #include <stdlib.h>
47 #include <string.h>
48 #include <unistd.h>
49 
50 #include "got_error.h"
51 #include "got_object.h"
52 
53 #include "got_compat.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_object.h"
57 #include "got_lib_privsep.h"
58 #include "got_lib_sha1.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, 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.git = git;
83 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH, 0, 0, -1, &p, sizeof(p)) == -1)
84 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH");
85 	return NULL;
86 }
87 
88 static const struct got_error *
89 send_patch_done(void)
90 {
91 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_DONE, 0, 0, -1,
92 	    NULL, 0) == -1)
93 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
94 	if (imsg_flush(&ibuf) == -1)
95 		return got_error_from_errno("imsg_flush");
96 	return NULL;
97 }
98 
99 /* based on fetchname from usr.bin/patch/util.c */
100 static const struct got_error *
101 filename(const char *at, char **name)
102 {
103 	char	*tmp, *t;
104 
105 	*name = NULL;
106 	if (*at == '\0')
107 		return NULL;
108 
109 	while (isspace((unsigned char)*at))
110 		at++;
111 
112 	/* files can be created or removed by diffing against /dev/null */
113 	if (!strncmp(at, _PATH_DEVNULL, sizeof(_PATH_DEVNULL) - 1))
114 		return NULL;
115 
116 	tmp = strdup(at);
117 	if (tmp == NULL)
118 		return got_error_from_errno("strdup");
119 	if ((t = strchr(tmp, '\t')) != NULL)
120 		*t = '\0';
121 	if ((t = strchr(tmp, '\n')) != NULL)
122 		*t = '\0';
123 
124 	*name = strdup(tmp);
125 	free(tmp);
126 	if (*name == NULL)
127 		return got_error_from_errno("strdup");
128 	return NULL;
129 }
130 
131 static const struct got_error *
132 blobid(const char *line, char **blob, int git)
133 {
134 	uint8_t digest[SHA1_DIGEST_LENGTH];
135 	size_t len;
136 
137 	*blob = NULL;
138 
139 	len = strspn(line, "0123456789abcdefABCDEF");
140 	if ((*blob = strndup(line, len)) == NULL)
141 		return got_error_from_errno("strndup");
142 
143 	if (!git && !got_parse_sha1_digest(digest, *blob)) {
144 		/* silently ignore invalid blob ids */
145 		free(*blob);
146 		*blob = NULL;
147 	}
148 	return NULL;
149 }
150 
151 static const struct got_error *
152 patch_start(int *git, char **cid, FILE *fp)
153 {
154 	const struct got_error *err = NULL;
155 	char	*line = NULL;
156 	size_t	 linesize = 0;
157 	ssize_t	 linelen;
158 
159 	*git = 0;
160 
161 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
162 		if (!strncmp(line, "diff --git ", 11)) {
163 			*git = 1;
164 			free(*cid);
165 			*cid = NULL;
166 			break;
167 		} else if (!strncmp(line, "diff ", 5)) {
168 			*git = 0;
169 			free(*cid);
170 			*cid = NULL;
171 		} else if (!strncmp(line, "commit - ", 9)) {
172 			free(*cid);
173 			err = blobid(line + 9, cid, *git);
174 			if (err)
175 				break;
176 		} else if (!strncmp(line, "--- ", 4) ||
177 		    !strncmp(line, "+++ ", 4) ||
178 		    !strncmp(line, "blob - ", 7)) {
179 			/* rewind to previous line */
180 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
181 				err = got_error_from_errno("fseeko");
182 			break;
183 		}
184 	}
185 
186 	free(line);
187 	if (ferror(fp) && err == NULL)
188 		err = got_error_from_errno("getline");
189 	if (feof(fp) && err == NULL)
190 		err = got_error(GOT_ERR_NO_PATCH);
191 	return err;
192 }
193 
194 static const struct got_error *
195 find_diff(int *done, int *next, FILE *fp, int git, const char *commitid)
196 {
197 	const struct got_error *err = NULL;
198 	char	*old = NULL, *new = NULL;
199 	char	*blob = NULL;
200 	char	*line = NULL;
201 	size_t	 linesize = 0;
202 	ssize_t	 linelen;
203 	int	 create, rename = 0;
204 
205 	*done = 0;
206 	*next = 0;
207 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
208 		/*
209 		 * Ignore the Index name like GNU and larry' patch,
210 		 * we don't have to follow POSIX.
211 		 */
212 
213 		if (!strncmp(line, "--- ", 4)) {
214 			free(old);
215 			err = filename(line+4, &old);
216 		} else if (rename && !strncmp(line, "rename from ", 12)) {
217 			free(old);
218 			err = filename(line+12, &old);
219 		} else if (!strncmp(line, "+++ ", 4)) {
220 			free(new);
221 			err = filename(line+4, &new);
222 		} else if (!git && !strncmp(line, "blob - ", 7)) {
223 			free(blob);
224 			err = blobid(line + 7, &blob, git);
225 		} else if (rename && !strncmp(line, "rename to ", 10)) {
226 			free(new);
227 			err = filename(line + 10, &new);
228 		} else if (git && !strncmp(line, "similarity index 100%", 21))
229 			rename = 1;
230 		else if (git && !strncmp(line, "index ", 6)) {
231 			free(blob);
232 			err = blobid(line + 6, &blob, git);
233 		} else if (!strncmp(line, "diff ", 5)) {
234 			/* rewind to previous line */
235 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
236 				err = got_error_from_errno("fseeko");
237 			*next = 1;
238 			break;
239 		}
240 
241 		if (err)
242 			break;
243 
244 		/*
245 		 * Git-style diffs with "similarity index 100%" don't
246 		 * have any hunks and ends with the "rename to foobar"
247 		 * line.
248 		 */
249 		if (rename && old != NULL && new != NULL) {
250 			*done = 1;
251 			err = send_patch(old, new, commitid,
252 			    blob, git);
253 			break;
254 		}
255 
256 		if (!strncmp(line, "@@ -", 4)) {
257 			create = !strncmp(line+4, "0,0", 3);
258 			if ((old == NULL && new == NULL) ||
259 			    (!create && old == NULL))
260 				err = got_error(GOT_ERR_PATCH_MALFORMED);
261 			else
262 				err = send_patch(old, new, commitid,
263 				    blob, git);
264 
265 			if (err)
266 				break;
267 
268 			/* rewind to previous line */
269 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
270 				err = got_error_from_errno("fseeko");
271 			break;
272 		}
273 	}
274 
275 	free(old);
276 	free(new);
277 	free(blob);
278 	free(line);
279 	if (ferror(fp) && err == NULL)
280 		err = got_error_from_errno("getline");
281 	if (feof(fp) && err == NULL)
282 		err = got_error(GOT_ERR_NO_PATCH);
283 	return err;
284 }
285 
286 static const struct got_error *
287 strtolnum(char **str, int *n)
288 {
289 	char		*p, c;
290 	const char	*errstr;
291 
292 	for (p = *str; isdigit((unsigned char)*p); ++p)
293 		/* nop */;
294 
295 	c = *p;
296 	*p = '\0';
297 
298 	*n = strtonum(*str, 0, INT_MAX, &errstr);
299 	if (errstr != NULL)
300 		return got_error(GOT_ERR_PATCH_MALFORMED);
301 
302 	*p = c;
303 	*str = p;
304 	return NULL;
305 }
306 
307 static const struct got_error *
308 parse_hdr(char *s, int *done, struct got_imsg_patch_hunk *hdr)
309 {
310 	static const struct got_error *err = NULL;
311 
312 	if (strncmp(s, "@@ -", 4)) {
313 		*done = 1;
314 		return NULL;
315 	}
316 
317 	s += 4;
318 	if (!*s)
319 		return NULL;
320 	err = strtolnum(&s, &hdr->oldfrom);
321 	if (err)
322 		return err;
323 	if (*s == ',') {
324 		s++;
325 		err = strtolnum(&s, &hdr->oldlines);
326 		if (err)
327 			return err;
328 	} else
329 		hdr->oldlines = 1;
330 
331 	if (*s == ' ')
332 		s++;
333 
334 	if (*s != '+' || !*++s)
335 		return got_error(GOT_ERR_PATCH_MALFORMED);
336 	err = strtolnum(&s, &hdr->newfrom);
337 	if (err)
338 		return err;
339 	if (*s == ',') {
340 		s++;
341 		err = strtolnum(&s, &hdr->newlines);
342 		if (err)
343 			return err;
344 	} else
345 		hdr->newlines = 1;
346 
347 	if (*s == ' ')
348 		s++;
349 
350 	if (*s != '@')
351 		return got_error(GOT_ERR_PATCH_MALFORMED);
352 
353 	if (hdr->oldfrom >= INT_MAX - hdr->oldlines ||
354 	    hdr->newfrom >= INT_MAX - hdr->newlines ||
355 	    /* not so sure about this one */
356 	    hdr->oldlines >= INT_MAX - hdr->newlines - 1 ||
357 	    (hdr->oldlines == 0 && hdr->newlines == 0))
358 		return got_error(GOT_ERR_PATCH_MALFORMED);
359 
360 	if (hdr->oldlines == 0) {
361 		/* larry says to "do append rather than insert"; I don't
362 		 * quite get it, but i trust him.
363 		 */
364 		hdr->oldfrom++;
365 	}
366 
367 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_HUNK, 0, 0, -1,
368 	    hdr, sizeof(*hdr)) == -1)
369 		return got_error_from_errno(
370 		    "imsg_compose GOT_IMSG_PATCH_HUNK");
371 	return NULL;
372 }
373 
374 static const struct got_error *
375 send_line(const char *line)
376 {
377 	static const struct got_error *err = NULL;
378 	char *p = NULL;
379 
380 	if (*line != '+' && *line != '-' && *line != ' ' && *line != '\\') {
381 		if (asprintf(&p, " %s", line) == -1)
382 			return got_error_from_errno("asprintf");
383 		line = p;
384 	}
385 
386 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_LINE, 0, 0, -1,
387 	    line, strlen(line) + 1) == -1)
388 		err = got_error_from_errno(
389 		    "imsg_compose GOT_IMSG_PATCH_LINE");
390 
391 	free(p);
392 	return err;
393 }
394 
395 static const struct got_error *
396 peek_special_line(FILE *fp)
397 {
398 	const struct got_error *err;
399 	int ch;
400 
401 	ch = fgetc(fp);
402 	if (ch != EOF && ch != '\\') {
403 		ungetc(ch, fp);
404 		return NULL;
405 	}
406 
407 	if (ch == '\\') {
408 		err = send_line("\\");
409 		if (err)
410 			return err;
411 	}
412 
413 	while (ch != EOF && ch != '\n')
414 		ch = fgetc(fp);
415 
416 	if (ch != EOF || feof(fp))
417 		return NULL;
418 	return got_error(GOT_ERR_IO);
419 }
420 
421 static const struct got_error *
422 parse_hunk(FILE *fp, int *done)
423 {
424 	static const struct got_error *err = NULL;
425 	struct got_imsg_patch_hunk hdr;
426 	char	*line = NULL, ch;
427 	size_t	 linesize = 0;
428 	ssize_t	 linelen;
429 	int	 leftold, leftnew;
430 
431 	linelen = getline(&line, &linesize, fp);
432 	if (linelen == -1) {
433 		*done = 1;
434 		goto done;
435 	}
436 
437 	err = parse_hdr(line, done, &hdr);
438 	if (err)
439 		goto done;
440 	if (*done) {
441 		if (fseeko(fp, -linelen, SEEK_CUR) == -1)
442 			err = got_error_from_errno("fseeko");
443 		goto done;
444 	}
445 
446 	leftold = hdr.oldlines;
447 	leftnew = hdr.newlines;
448 
449 	while (leftold > 0 || leftnew > 0) {
450 		linelen = getline(&line, &linesize, fp);
451 		if (linelen == -1) {
452 			if (ferror(fp)) {
453 				err = got_error_from_errno("getline");
454 				goto done;
455 			}
456 
457 			/* trailing newlines may be chopped */
458 			if (leftold < 3 && leftnew < 3) {
459 				*done = 1;
460 				break;
461 			}
462 
463 			err = got_error(GOT_ERR_PATCH_TRUNCATED);
464 			goto done;
465 		}
466 		if (line[linelen - 1] == '\n')
467 			line[linelen - 1] = '\0';
468 
469 		/* usr.bin/patch allows '=' as context char */
470 		if (*line == '=')
471 			*line = ' ';
472 
473 		ch = *line;
474 		if (ch == '\t' || ch == '\0')
475 			ch = ' ';	/* the space got eaten */
476 
477 		switch (ch) {
478 		case '-':
479 			leftold--;
480 			break;
481 		case ' ':
482 			leftold--;
483 			leftnew--;
484 			break;
485 		case '+':
486 			leftnew--;
487 			break;
488 		default:
489 			err = got_error(GOT_ERR_PATCH_MALFORMED);
490 			goto done;
491 		}
492 
493 		if (leftold < 0 || leftnew < 0) {
494 			err = got_error(GOT_ERR_PATCH_MALFORMED);
495 			goto done;
496 		}
497 
498 		err = send_line(line);
499 		if (err)
500 			goto done;
501 
502 		if ((ch == '-' && leftold == 0) ||
503 		    (ch == '+' && leftnew == 0)) {
504 			err = peek_special_line(fp);
505 			if (err)
506 				goto done;
507 		}
508 	}
509 
510 done:
511 	free(line);
512 	return err;
513 }
514 
515 static const struct got_error *
516 read_patch(struct imsgbuf *ibuf, int fd)
517 {
518 	const struct got_error *err = NULL;
519 	FILE *fp;
520 	int git, patch_found = 0;
521 	char *cid = NULL;
522 
523 	if ((fp = fdopen(fd, "r")) == NULL) {
524 		err = got_error_from_errno("fdopen");
525 		close(fd);
526 		return err;
527 	}
528 
529 	while ((err = patch_start(&git, &cid, fp)) == NULL) {
530 		int done, next;
531 
532 		err = find_diff(&done, &next, fp, git, cid);
533 		if (err)
534 			goto done;
535 		if (next)
536 			continue;
537 
538 		patch_found = 1;
539 
540 		while (!done) {
541 			err = parse_hunk(fp, &done);
542 			if (err)
543 				goto done;
544 		}
545 
546 		err = send_patch_done();
547 		if (err)
548 			goto done;
549 	}
550 
551 done:
552 	fclose(fp);
553 	free(cid);
554 
555 	/* ignore trailing gibberish */
556 	if (err != NULL && err->code == GOT_ERR_NO_PATCH && patch_found)
557 		err = NULL;
558 
559 	return err;
560 }
561 
562 int
563 main(int argc, char **argv)
564 {
565 	const struct got_error *err = NULL;
566 	struct imsg imsg;
567 #if 0
568 	static int attached;
569 	while (!attached)
570 		sleep(1);
571 #endif
572 
573 	imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
574 #ifndef PROFILE
575 	/* revoke access to most system calls */
576 	if (pledge("stdio recvfd", NULL) == -1) {
577 		err = got_error_from_errno("pledge");
578 		got_privsep_send_error(&ibuf, err);
579 		return 1;
580 	}
581 
582 	/* revoke fs access */
583 	if (landlock_no_fs() == -1) {
584 		err = got_error_from_errno("landlock_no_fs");
585 		got_privsep_send_error(&ibuf, err);
586 		return 1;
587 	}
588 	if (cap_enter() == -1) {
589 		err = got_error_from_errno("cap_enter");
590 		got_privsep_send_error(&ibuf, err);
591 		return 1;
592 	}
593 #endif
594 
595 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
596 	if (err)
597 		goto done;
598 	if (imsg.hdr.type != GOT_IMSG_PATCH_FILE || imsg.fd == -1) {
599 		err = got_error(GOT_ERR_PRIVSEP_MSG);
600 		goto done;
601 	}
602 
603 	err = read_patch(&ibuf, imsg.fd);
604 	if (err)
605 		goto done;
606 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_EOF, 0, 0, -1,
607 	    NULL, 0) == -1) {
608 		err = got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
609 		goto done;
610 	}
611 	err = got_privsep_flush_imsg(&ibuf);
612 done:
613 	imsg_free(&imsg);
614 	if (err != NULL) {
615 		got_privsep_send_error(&ibuf, err);
616 		err = NULL;
617 	}
618 	if (close(GOT_IMSG_FD_CHILD) == -1 && err == NULL)
619 		err = got_error_from_errno("close");
620 	if (err && err->code != GOT_ERR_PRIVSEP_PIPE)
621 		fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
622 	return err ? 1 : 0;
623 }