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