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 <stdint.h>
47 #include <stdio.h>
48 #include <stdlib.h>
49 #include <string.h>
50 #include <unistd.h>
51 #include <imsg.h>
52 
53 #include "got_error.h"
54 #include "got_object.h"
55 
56 #include "got_lib_delta.h"
57 #include "got_lib_object.h"
58 #include "got_lib_privsep.h"
59 #include "got_lib_sha1.h"
60 
61 struct imsgbuf ibuf;
62 
63 static const struct got_error *
64 send_patch(const char *oldname, const char *newname, const char *commitid,
65     const char *blob, const int xbit, int git)
66 {
67 	struct got_imsg_patch p;
68 
69 	memset(&p, 0, sizeof(p));
70 
71 	if (oldname != NULL)
72 		strlcpy(p.old, oldname, sizeof(p.old));
73 
74 	if (newname != NULL)
75 		strlcpy(p.new, newname, sizeof(p.new));
76 
77 	if (commitid != NULL)
78 		strlcpy(p.cid, commitid, sizeof(p.cid));
79 
80 	if (blob != NULL)
81 		strlcpy(p.blob, blob, sizeof(p.blob));
82 
83 	p.xbit = xbit;
84 	p.git = git;
85 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH, 0, 0, -1, &p, sizeof(p)) == -1)
86 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH");
87 	return NULL;
88 }
89 
90 static const struct got_error *
91 send_patch_done(void)
92 {
93 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_DONE, 0, 0, -1,
94 	    NULL, 0) == -1)
95 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
96 	return got_privsep_flush_imsg(&ibuf);
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 int
132 filexbit(const char *line)
133 {
134 	char *m;
135 
136 	m = strchr(line, '(');
137 	if (m && !strncmp(m + 1, "mode ", 5))
138 		return strncmp(m + 6, "755", 3) == 0;
139 
140 	return 0;
141 }
142 
143 static const struct got_error *
144 blobid(const char *line, char **blob, int git)
145 {
146 	uint8_t digest[SHA1_DIGEST_LENGTH];
147 	size_t len;
148 
149 	*blob = NULL;
150 
151 	len = strspn(line, "0123456789abcdefABCDEF");
152 	if ((*blob = strndup(line, len)) == NULL)
153 		return got_error_from_errno("strndup");
154 
155 	if (!git && !got_parse_sha1_digest(digest, *blob)) {
156 		/* silently ignore invalid blob ids */
157 		free(*blob);
158 		*blob = NULL;
159 	}
160 	return NULL;
161 }
162 
163 static const struct got_error *
164 patch_start(int *git, char **cid, FILE *fp)
165 {
166 	const struct got_error *err = NULL;
167 	char	*line = NULL;
168 	size_t	 linesize = 0;
169 	ssize_t	 linelen;
170 
171 	*git = 0;
172 
173 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
174 		if (!strncmp(line, "diff --git ", 11)) {
175 			*git = 1;
176 			free(*cid);
177 			*cid = NULL;
178 			break;
179 		} else if (!strncmp(line, "diff ", 5)) {
180 			*git = 0;
181 			free(*cid);
182 			*cid = NULL;
183 		} else if (!strncmp(line, "commit - ", 9)) {
184 			free(*cid);
185 			err = blobid(line + 9, cid, *git);
186 			if (err)
187 				break;
188 		} else if (!strncmp(line, "--- ", 4) ||
189 		    !strncmp(line, "+++ ", 4) ||
190 		    !strncmp(line, "blob - ", 7)) {
191 			/* rewind to previous line */
192 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
193 				err = got_error_from_errno("fseeko");
194 			break;
195 		}
196 	}
197 
198 	free(line);
199 	if (ferror(fp) && err == NULL)
200 		err = got_error_from_errno("getline");
201 	if (feof(fp) && err == NULL)
202 		err = got_error(GOT_ERR_NO_PATCH);
203 	return err;
204 }
205 
206 static const struct got_error *
207 find_diff(int *done, int *next, FILE *fp, int git, const char *commitid)
208 {
209 	const struct got_error *err = NULL;
210 	char	*old = NULL, *new = NULL;
211 	char	*blob = NULL;
212 	char	*line = NULL;
213 	size_t	 linesize = 0;
214 	ssize_t	 linelen;
215 	int	 create, rename = 0, xbit = 0;
216 
217 	*done = 0;
218 	*next = 0;
219 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
220 		/*
221 		 * Ignore the Index name like GNU and larry' patch,
222 		 * we don't have to follow POSIX.
223 		 */
224 
225 		if (!strncmp(line, "--- ", 4)) {
226 			free(old);
227 			err = filename(line+4, &old);
228 		} else if (rename && !strncmp(line, "rename from ", 12)) {
229 			free(old);
230 			err = filename(line+12, &old);
231 		} else if (!strncmp(line, "+++ ", 4)) {
232 			free(new);
233 			err = filename(line+4, &new);
234 		} else if (!strncmp(line, "blob + ", 7)) {
235 			xbit = filexbit(line);
236 		} else if (!git && !strncmp(line, "blob - ", 7)) {
237 			free(blob);
238 			err = blobid(line + 7, &blob, git);
239 		} else if (rename && !strncmp(line, "rename to ", 10)) {
240 			free(new);
241 			err = filename(line + 10, &new);
242 		} else if (git && !strncmp(line, "similarity index 100%", 21))
243 			rename = 1;
244 		else if (git && !strncmp(line, "new file mode 100", 17))
245 			xbit = strncmp(line + 17, "755", 3) == 0;
246 		else if (git && !strncmp(line, "index ", 6)) {
247 			free(blob);
248 			err = blobid(line + 6, &blob, git);
249 		} else if (!strncmp(line, "diff ", 5)) {
250 			/* rewind to previous line */
251 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
252 				err = got_error_from_errno("fseeko");
253 			*next = 1;
254 			break;
255 		}
256 
257 		if (err)
258 			break;
259 
260 		/*
261 		 * Git-style diffs with "similarity index 100%" don't
262 		 * have any hunks and ends with the "rename to foobar"
263 		 * line.
264 		 */
265 		if (rename && old != NULL && new != NULL) {
266 			*done = 1;
267 			err = send_patch(old, new, commitid,
268 			    blob, xbit, git);
269 			break;
270 		}
271 
272 		if (!strncmp(line, "@@ -", 4)) {
273 			create = !strncmp(line+4, "0,0", 3);
274 			if ((old == NULL && new == NULL) ||
275 			    (!create && old == NULL))
276 				err = got_error(GOT_ERR_PATCH_MALFORMED);
277 			else
278 				err = send_patch(old, new, commitid,
279 				    blob, xbit, git);
280 
281 			if (err)
282 				break;
283 
284 			/* rewind to previous line */
285 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
286 				err = got_error_from_errno("fseeko");
287 			break;
288 		}
289 	}
290 
291 	free(old);
292 	free(new);
293 	free(blob);
294 	free(line);
295 	if (ferror(fp) && err == NULL)
296 		err = got_error_from_errno("getline");
297 	if (feof(fp) && err == NULL)
298 		err = got_error(GOT_ERR_NO_PATCH);
299 	return err;
300 }
301 
302 static const struct got_error *
303 strtolnum(char **str, int *n)
304 {
305 	char		*p, c;
306 	const char	*errstr;
307 
308 	for (p = *str; isdigit((unsigned char)*p); ++p)
309 		/* nop */;
310 
311 	c = *p;
312 	*p = '\0';
313 
314 	*n = strtonum(*str, 0, INT_MAX, &errstr);
315 	if (errstr != NULL)
316 		return got_error(GOT_ERR_PATCH_MALFORMED);
317 
318 	*p = c;
319 	*str = p;
320 	return NULL;
321 }
322 
323 static const struct got_error *
324 parse_hdr(char *s, int *done, struct got_imsg_patch_hunk *hdr)
325 {
326 	static const struct got_error *err = NULL;
327 
328 	if (strncmp(s, "@@ -", 4)) {
329 		*done = 1;
330 		return NULL;
331 	}
332 
333 	s += 4;
334 	if (!*s)
335 		return NULL;
336 	err = strtolnum(&s, &hdr->oldfrom);
337 	if (err)
338 		return err;
339 	if (*s == ',') {
340 		s++;
341 		err = strtolnum(&s, &hdr->oldlines);
342 		if (err)
343 			return err;
344 	} else
345 		hdr->oldlines = 1;
346 
347 	if (*s == ' ')
348 		s++;
349 
350 	if (*s != '+' || !*++s)
351 		return got_error(GOT_ERR_PATCH_MALFORMED);
352 	err = strtolnum(&s, &hdr->newfrom);
353 	if (err)
354 		return err;
355 	if (*s == ',') {
356 		s++;
357 		err = strtolnum(&s, &hdr->newlines);
358 		if (err)
359 			return err;
360 	} else
361 		hdr->newlines = 1;
362 
363 	if (*s == ' ')
364 		s++;
365 
366 	if (*s != '@')
367 		return got_error(GOT_ERR_PATCH_MALFORMED);
368 
369 	if (hdr->oldfrom >= INT_MAX - hdr->oldlines ||
370 	    hdr->newfrom >= INT_MAX - hdr->newlines ||
371 	    /* not so sure about this one */
372 	    hdr->oldlines >= INT_MAX - hdr->newlines - 1 ||
373 	    (hdr->oldlines == 0 && hdr->newlines == 0))
374 		return got_error(GOT_ERR_PATCH_MALFORMED);
375 
376 	if (hdr->oldlines == 0) {
377 		/* larry says to "do append rather than insert"; I don't
378 		 * quite get it, but i trust him.
379 		 */
380 		hdr->oldfrom++;
381 	}
382 
383 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_HUNK, 0, 0, -1,
384 	    hdr, sizeof(*hdr)) == -1)
385 		return got_error_from_errno(
386 		    "imsg_compose GOT_IMSG_PATCH_HUNK");
387 	return NULL;
388 }
389 
390 static const struct got_error *
391 send_line(const char *line)
392 {
393 	static const struct got_error *err = NULL;
394 	char *p = NULL;
395 
396 	if (*line != '+' && *line != '-' && *line != ' ' && *line != '\\') {
397 		if (asprintf(&p, " %s", line) == -1)
398 			return got_error_from_errno("asprintf");
399 		line = p;
400 	}
401 
402 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_LINE, 0, 0, -1,
403 	    line, strlen(line) + 1) == -1)
404 		err = got_error_from_errno(
405 		    "imsg_compose GOT_IMSG_PATCH_LINE");
406 
407 	free(p);
408 	return err;
409 }
410 
411 static const struct got_error *
412 peek_special_line(FILE *fp)
413 {
414 	const struct got_error *err;
415 	int ch;
416 
417 	ch = fgetc(fp);
418 	if (ch != EOF && ch != '\\') {
419 		ungetc(ch, fp);
420 		return NULL;
421 	}
422 
423 	if (ch == '\\') {
424 		err = send_line("\\");
425 		if (err)
426 			return err;
427 	}
428 
429 	while (ch != EOF && ch != '\n')
430 		ch = fgetc(fp);
431 
432 	if (ch != EOF || feof(fp))
433 		return NULL;
434 	return got_error(GOT_ERR_IO);
435 }
436 
437 static const struct got_error *
438 parse_hunk(FILE *fp, int *done)
439 {
440 	static const struct got_error *err = NULL;
441 	struct got_imsg_patch_hunk hdr;
442 	char	*line = NULL, ch;
443 	size_t	 linesize = 0;
444 	ssize_t	 linelen;
445 	int	 leftold, leftnew;
446 
447 	linelen = getline(&line, &linesize, fp);
448 	if (linelen == -1) {
449 		*done = 1;
450 		goto done;
451 	}
452 
453 	err = parse_hdr(line, done, &hdr);
454 	if (err)
455 		goto done;
456 	if (*done) {
457 		if (fseeko(fp, -linelen, SEEK_CUR) == -1)
458 			err = got_error_from_errno("fseeko");
459 		goto done;
460 	}
461 
462 	leftold = hdr.oldlines;
463 	leftnew = hdr.newlines;
464 
465 	while (leftold > 0 || leftnew > 0) {
466 		linelen = getline(&line, &linesize, fp);
467 		if (linelen == -1) {
468 			if (ferror(fp)) {
469 				err = got_error_from_errno("getline");
470 				goto done;
471 			}
472 
473 			/* trailing newlines may be chopped */
474 			if (leftold < 3 && leftnew < 3) {
475 				*done = 1;
476 				break;
477 			}
478 
479 			err = got_error(GOT_ERR_PATCH_TRUNCATED);
480 			goto done;
481 		}
482 		if (line[linelen - 1] == '\n')
483 			line[linelen - 1] = '\0';
484 
485 		/* usr.bin/patch allows '=' as context char */
486 		if (*line == '=')
487 			*line = ' ';
488 
489 		ch = *line;
490 		if (ch == '\t' || ch == '\0')
491 			ch = ' ';	/* the space got eaten */
492 
493 		switch (ch) {
494 		case '-':
495 			leftold--;
496 			break;
497 		case ' ':
498 			leftold--;
499 			leftnew--;
500 			break;
501 		case '+':
502 			leftnew--;
503 			break;
504 		default:
505 			err = got_error(GOT_ERR_PATCH_MALFORMED);
506 			goto done;
507 		}
508 
509 		if (leftold < 0 || leftnew < 0) {
510 			err = got_error(GOT_ERR_PATCH_MALFORMED);
511 			goto done;
512 		}
513 
514 		err = send_line(line);
515 		if (err)
516 			goto done;
517 
518 		if ((ch == '-' && leftold == 0) ||
519 		    (ch == '+' && leftnew == 0)) {
520 			err = peek_special_line(fp);
521 			if (err)
522 				goto done;
523 		}
524 	}
525 
526 done:
527 	free(line);
528 	return err;
529 }
530 
531 static const struct got_error *
532 read_patch(struct imsgbuf *ibuf, int fd)
533 {
534 	const struct got_error *err = NULL;
535 	FILE *fp;
536 	int git, patch_found = 0;
537 	char *cid = NULL;
538 
539 	if ((fp = fdopen(fd, "r")) == NULL) {
540 		err = got_error_from_errno("fdopen");
541 		close(fd);
542 		return err;
543 	}
544 
545 	while ((err = patch_start(&git, &cid, fp)) == NULL) {
546 		int done, next;
547 
548 		err = find_diff(&done, &next, fp, git, cid);
549 		if (err)
550 			goto done;
551 		if (next)
552 			continue;
553 
554 		patch_found = 1;
555 
556 		while (!done) {
557 			err = parse_hunk(fp, &done);
558 			if (err)
559 				goto done;
560 		}
561 
562 		err = send_patch_done();
563 		if (err)
564 			goto done;
565 	}
566 
567 done:
568 	fclose(fp);
569 	free(cid);
570 
571 	/* ignore trailing gibberish */
572 	if (err != NULL && err->code == GOT_ERR_NO_PATCH && patch_found)
573 		err = NULL;
574 
575 	return err;
576 }
577 
578 int
579 main(int argc, char **argv)
580 {
581 	const struct got_error *err = NULL;
582 	struct imsg imsg;
583 #if 0
584 	static int attached;
585 	while (!attached)
586 		sleep(1);
587 #endif
588 
589 	imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
590 #ifndef PROFILE
591 	/* revoke access to most system calls */
592 	if (pledge("stdio recvfd", NULL) == -1) {
593 		err = got_error_from_errno("pledge");
594 		got_privsep_send_error(&ibuf, err);
595 		return 1;
596 	}
597 #endif
598 
599 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
600 	if (err)
601 		goto done;
602 	if (imsg.hdr.type != GOT_IMSG_PATCH_FILE || imsg.fd == -1) {
603 		err = got_error(GOT_ERR_PRIVSEP_MSG);
604 		goto done;
605 	}
606 
607 	err = read_patch(&ibuf, imsg.fd);
608 	if (err)
609 		goto done;
610 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_EOF, 0, 0, -1,
611 	    NULL, 0) == -1) {
612 		err = got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
613 		goto done;
614 	}
615 	err = got_privsep_flush_imsg(&ibuf);
616 done:
617 	imsg_free(&imsg);
618 	if (err != NULL) {
619 		got_privsep_send_error(&ibuf, err);
620 		err = NULL;
621 	}
622 	if (close(GOT_IMSG_FD_CHILD) == -1 && err == NULL)
623 		err = got_error_from_errno("close");
624 	if (err && err->code != GOT_ERR_PRIVSEP_PIPE)
625 		fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
626 	return err ? 1 : 0;
627 }