1 /*
2  * Copyright (c) 2024 Omar Polo <op@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/time.h>
18 #include <sys/types.h>
19 #include <sys/socket.h>
20 
21 #include <err.h>
22 #include <errno.h>
23 #include <fcntl.h>
24 #include <limits.h>
25 #include <netdb.h>
26 #include <poll.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
30 #include <unistd.h>
31 
32 #include "got_opentemp.h"
33 #include "got_version.h"
34 
35 #include "bufio.h"
36 #include "utf8d.h"
37 
38 #define USERAGENT	 "got-notify-http/" GOT_VERSION_STR
39 
40 static int		 http_timeout = 300;	/* 5 minutes in seconds */
41 
42 __dead static void
43 usage(void)
44 {
45 	fprintf(stderr, "usage: %s [-c] -h host -p port path\n",
46 	    getprogname());
47 	exit(1);
48 }
49 
50 static int
51 dial(const char *host, const char *port)
52 {
53 	struct addrinfo	 hints, *res, *res0;
54 	const char	*cause = NULL;
55 	int		 s, error, save_errno;
56 
57 	memset(&hints, 0, sizeof(hints));
58 	hints.ai_family = AF_UNSPEC;
59 	hints.ai_socktype = SOCK_STREAM;
60 	error = getaddrinfo(host, port, &hints, &res0);
61 	if (error)
62 		errx(1, "failed to resolve %s:%s: %s", host, port,
63 		    gai_strerror(error));
64 
65 	s = -1;
66 	for (res = res0; res; res = res->ai_next) {
67 		s = socket(res->ai_family, res->ai_socktype,
68 		    res->ai_protocol);
69 		if (s == -1) {
70 			cause = "socket";
71 			continue;
72 		}
73 
74 		if (connect(s, res->ai_addr, res->ai_addrlen) == -1) {
75 			cause = "connect";
76 			save_errno = errno;
77 			close(s);
78 			errno = save_errno;
79 			s = -1;
80 			continue;
81 		}
82 
83 		break;
84 	}
85 
86 	freeaddrinfo(res0);
87 	if (s == -1)
88 		err(1, "%s", cause);
89 	return s;
90 }
91 
92 static void
93 escape(FILE *fp, const uint8_t *s)
94 {
95 	uint32_t codepoint, state;
96 	const uint8_t *start = s;
97 
98 	state = 0;
99 	for (; *s; ++s) {
100 		switch (decode(&state, &codepoint, *s)) {
101 		case UTF8_ACCEPT:
102 			switch (codepoint) {
103 			case '"':
104 			case '\\':
105 				fprintf(fp, "\\%c", *s);
106 				break;
107 			case '\b':
108 				fprintf(fp, "\\b");
109 				break;
110 			case '\f':
111 				fprintf(fp, "\\f");
112 				break;
113 			case '\n':
114 				fprintf(fp, "\\n");
115 				break;
116 			case '\r':
117 				fprintf(fp, "\\r");
118 				break;
119 			case '\t':
120 				fprintf(fp, "\\t");
121 				break;
122 			default:
123 				/* other control characters */
124 				if (codepoint < ' ' || codepoint == 0x7F) {
125 					fprintf(fp, "\\u%04x", codepoint);
126 					break;
127 				}
128 				fwrite(start, 1, s - start + 1, fp);
129 				break;
130 			}
131 			start = s + 1;
132 			break;
133 
134 		case UTF8_REJECT:
135 			/* bad UTF-8 sequence; try to recover */
136 			fputs("\\uFFFD", fp);
137 			state = UTF8_ACCEPT;
138 			start = s + 1;
139 			break;
140 		}
141         }
142 }
143 
144 static void
145 json_field(FILE *fp, const char *key, const char *val, int comma)
146 {
147 	fprintf(fp, "\"%s\":\"", key);
148 	escape(fp, val);
149 	fprintf(fp, "\"%s", comma ? "," : "");
150 }
151 
152 static void
153 json_author(FILE *fp, const char *type, char *address, int comma)
154 {
155 	char	*gt, *lt, *at, *email, *endname;
156 
157 	fprintf(fp, "\"%s\":{", type);
158 
159 	gt = strchr(address, '<');
160 	if (gt != NULL) {
161 		/* long format, e.g. "Omar Polo <op@openbsd.org>" */
162 
163 		json_field(fp, "full", address, 1);
164 
165 		endname = gt;
166 		while (endname > address && endname[-1] == ' ')
167 			endname--;
168 
169 		*endname = '\0';
170 		json_field(fp, "name", address, 1);
171 
172 		email = gt + 1;
173 		lt = strchr(email, '>');
174 		if (lt)
175 			*lt = '\0';
176 
177 		json_field(fp, "mail", email, 1);
178 
179 		at = strchr(email, '@');
180 		if (at)
181 			*at = '\0';
182 
183 		json_field(fp, "user", email, 0);
184 	} else {
185 		/* short format only shows the username */
186 		json_field(fp, "user", address, 0);
187 	}
188 
189 	fprintf(fp, "}%s", comma ? "," : "");
190 }
191 
192 static int
193 jsonify_commit_short(FILE *fp, char *line)
194 {
195 	char	*t, *date, *id, *author, *message;
196 
197 	t = line;
198 	date = t;
199 	if ((t = strchr(t, ' ')) == NULL)
200       			errx(1, "malformed line");
201 	*t++ = '\0';
202 
203 	id = t;
204 	if ((t = strchr(t, ' ')) == NULL)
205       			errx(1, "malformed line");
206 	*t++ = '\0';
207 
208 	author = t;
209 	if ((t = strchr(t, ' ')) == NULL)
210       			errx(1, "malformed line");
211 	*t++ = '\0';
212 
213 	message = t;
214 
215 	fprintf(fp, "{\"type\":\"commit\",\"short\":true,");
216 	json_field(fp, "id", id, 1);
217 	json_author(fp, "committer", author, 1);
218 	json_field(fp, "date", date, 1);
219 	json_field(fp, "short_message", message, 0);
220 	fprintf(fp, "}");
221 
222 	return 0;
223 }
224 
225 static int
226 jsonify_commit(FILE *fp, char **line, ssize_t *linesize)
227 {
228 	const char	*errstr;
229 	char		*author = NULL;
230 	char		*l;
231 	ssize_t		 linelen;
232 	int		 parent = 0;
233 	int		 msglen = 0, msgwrote = 0;
234 	int		 done = 0;
235 	enum {
236 		P_FROM,
237 		P_VIA,
238 		P_DATE,
239 		P_PARENT,
240 		P_MSGLEN,
241 		P_MSG,
242 		P_DST,
243 		P_SUM,
244 	} phase = P_FROM;
245 
246 	l = *line;
247 	if (strncmp(l, "commit ", 7) != 0)
248 		errx(1, "%s: unexpected line: %s", __func__, l);
249 	l += 7;
250 	fprintf(fp, "{\"type\":\"commit\",\"short\":false,");
251 	json_field(fp, "id", l, 1);
252 
253 	while (!done) {
254 		if ((linelen = getline(line, linesize, stdin)) == -1)
255 			break;
256 
257 		if ((*line)[linelen - 1] == '\n')
258 			(*line)[--linelen] = '\0';
259 
260 		l = *line;
261 		switch (phase) {
262 		case P_FROM:
263 			if (strncmp(l, "from: ", 6) != 0)
264 				errx(1, "unexpected from line");
265 			l += 6;
266 
267 			author = strdup(l);
268 			if (author == NULL)
269 				err(1, "strdup");
270 
271 			json_author(fp, "author", l, 1);
272 
273 			phase = P_VIA;
274 			break;
275 
276 		case P_VIA:
277 			/* optional */
278 			if (!strncmp(l, "via: ", 5)) {
279 				l += 5;
280 				json_author(fp, "committer", l, 1);
281 				phase = P_DATE;
282 				break;
283 			}
284 
285 			if (author == NULL) /* impossible */
286 				err(1, "from not specified");
287 			json_author(fp, "committer", author, 1);
288 			free(author);
289 			author = NULL;
290 
291 			phase = P_DATE;
292 			/* fallthrough */
293 
294 		case P_DATE:
295 			/* optional */
296 			if (!strncmp(l, "date: ", 6)) {
297 				l += 6;
298 				json_field(fp, "date", l, 1);
299 				phase = P_PARENT;
300 				break;
301 			}
302 			phase = P_PARENT;
303 			/* fallthough */
304 
305 		case P_PARENT:
306 			/* optional - more than one */
307 			if (!strncmp(l, "parent ", 7)) {
308 				l += 7;
309 				l += strcspn(l, ":");
310 				l += strspn(l, " ");
311 
312 				if (parent == 0) {
313 					parent = 1;
314 					fprintf(fp, "\"parents\":[");
315 				}
316 
317 				fputc('"', fp);
318 				escape(fp, l);
319 				fputc('"', fp);
320 
321 				break;
322 			}
323 			if (parent != 0) {
324 				fprintf(fp, "],");
325 				parent = 0;
326 			}
327 			phase = P_MSGLEN;
328 			/* fallthrough */
329 
330 		case P_MSGLEN:
331 			if (strncmp(l, "messagelen: ", 12) != 0)
332 				errx(1, "unexpected messagelen line");
333 			l += 12;
334 			msglen = strtonum(l, 1, INT_MAX, &errstr);
335 			if (errstr)
336 				errx(1, "message len is %s: %s", errstr, l);
337 
338 			phase = P_MSG;
339 			break;
340 
341 		case P_MSG:
342 			/*
343 			 * The commit message is indented with one extra
344 			 * space which is not accounted for in messagelen,
345 			 * but we also strip the trailing \n so that
346 			 * accounts for it.
347 			 *
348 			 * Since we read line-by-line and there is always
349 			 * a \n added at the end of the message,
350 			 * tolerate one byte less than advertised.
351 			 */
352 			if (*l == ' ') {
353 				l++; /* skip leading space */
354 				linelen--;
355 
356 				if (msgwrote == 0 && linelen != 0) {
357 					json_field(fp, "short_message", l, 1);
358 					fprintf(fp, "\"message\":\"");
359 					escape(fp, l);
360 					escape(fp, "\n");
361 					msgwrote += linelen;
362 				} else if (msgwrote != 0) {
363 					escape(fp, l);
364 					escape(fp, "\n");
365 				}
366 			}
367 			msglen -= linelen + 1;
368 			if (msglen <= 1) {
369 				fprintf(fp, "\",");
370 				msgwrote = 0;
371 				phase = P_DST;
372 			}
373 			break;
374 
375 		case P_DST:
376 			/* XXX: ignore the diffstat for now */
377 			if (*l == '\0') {
378 				fprintf(fp, "\"diffstat\":{},");
379 				phase = P_SUM;
380 				break;
381 			}
382 			break;
383 
384 		case P_SUM:
385 			/* XXX: ignore the sum of changes for now */
386 			fprintf(fp, "\"changes\":{}}");
387 			done = 1;
388 			break;
389 
390 		default:
391 			/* unreachable */
392 			errx(1, "unexpected line: %s", *line);
393 		}
394 	}
395 	if (ferror(stdin))
396 		err(1, "getline");
397 	if (!done)
398 		errx(1, "unexpected EOF");
399 
400 	return 0;
401 }
402 
403 static int
404 jsonify(FILE *fp)
405 {
406 	char		*line = NULL;
407 	size_t		 linesize = 0;
408 	ssize_t		 linelen;
409 	int		 needcomma = 0;
410 
411 	fprintf(fp, "{\"notifications\":[");
412 	while ((linelen = getline(&line, &linesize, stdin)) != -1) {
413 		if (line[linelen - 1] == '\n')
414 			line[--linelen] = '\0';
415 
416 		if (*line == '\0')
417 			continue;
418 
419 		if (needcomma)
420 			fputc(',', fp);
421 		needcomma = 1;
422 
423 		if (strncmp(line, "commit ", 7) == 0) {
424 			if (jsonify_commit(fp, &line, &linesize) == -1)
425 				err(1, "jsonify_commit");
426 			continue;
427 		}
428 
429 		if (*line >= '0' && *line <= '9') {
430 			if (jsonify_commit_short(fp, line) == -1)
431 				err(1, "jsonify_commit_short");
432 			continue;
433 		}
434 
435 		errx(1, "unexpected line: %s", line);
436 	}
437 	if (ferror(stdin))
438 		err(1, "getline");
439 	fprintf(fp, "]}");
440 
441 	return 0;
442 }
443 
444 static char *
445 basic_auth(const char *username, const char *password)
446 {
447 	char	*tmp;
448 	int	 len;
449 
450 	len = asprintf(&tmp, "%s:%s", username, password);
451 	if (len == -1)
452 		err(1, "asprintf");
453 
454 	/* XXX base64-ify */
455 	return tmp;
456 }
457 
458 static inline int
459 bufio2poll(struct bufio *bio)
460 {
461 	int f, ret = 0;
462 
463 	f = bufio_ev(bio);
464 	if (f & BUFIO_WANT_READ)
465 		ret |= POLLIN;
466 	if (f & BUFIO_WANT_WRITE)
467 		ret |= POLLOUT;
468 	return ret;
469 }
470 
471 int
472 main(int argc, char **argv)
473 {
474 	FILE		*tmpfp;
475 	struct bufio	 bio;
476 	struct pollfd	 pfd;
477 	struct timespec	 timeout;
478 	const char	*username;
479 	const char	*password;
480 	const char	*timeoutstr;
481 	const char	*errstr;
482 	const char	*host = NULL, *port = NULL, *path = NULL;
483 	char		*auth, *line, *spc;
484 	size_t		 len;
485 	ssize_t		 r;
486 	off_t		 paylen;
487 	int		 tls = 0;
488 	int		 response_code = 0, done = 0;
489 	int		 ch, flags, ret, nonstd = 0;
490 
491 #ifndef PROFILE
492 	if (pledge("stdio rpath tmppath dns inet", NULL) == -1)
493 		err(1, "pledge");
494 #endif
495 
496 	while ((ch = getopt(argc, argv, "ch:p:")) != -1) {
497 		switch (ch) {
498 		case 'c':
499 			tls = 1;
500 			break;
501 		case 'h':
502 			host = optarg;
503 			break;
504 		case 'p':
505 			port = optarg;
506 			break;
507 		default:
508 			usage();
509 		}
510 	}
511 	argc -= optind;
512 	argv += optind;
513 
514 	if (host == NULL || argc != 1)
515 		usage();
516 	if (tls && port == NULL)
517 		port = "443";
518 	path = argv[0];
519 
520 	username = getenv("GOT_NOTIFY_HTTP_USER");
521 	password = getenv("GOT_NOTIFY_HTTP_PASS");
522 	if ((username != NULL && password == NULL) ||
523 	    (username == NULL && password != NULL))
524 		errx(1, "username or password are not specified");
525 	if (username && *password == '\0')
526 		errx(1, "password can't be empty");
527 
528 	/* used by the regression test suite */
529 	timeoutstr = getenv("GOT_NOTIFY_TIMEOUT");
530 	if (timeoutstr) {
531 		http_timeout = strtonum(timeoutstr, 0, 600, &errstr);
532 		if (errstr != NULL)
533 			errx(1, "timeout in seconds is %s: %s",
534 			    errstr, timeoutstr);
535 	}
536 
537 	memset(&timeout, 0, sizeof(timeout));
538 	timeout.tv_sec = http_timeout;
539 
540 	tmpfp = got_opentemp();
541 	if (tmpfp == NULL)
542 		err(1, "opentemp");
543 
544 	jsonify(tmpfp);
545 
546 	paylen = ftello(tmpfp);
547 	if (paylen == -1)
548 		err(1, "ftello");
549 	if (fseeko(tmpfp, 0, SEEK_SET) == -1)
550 		err(1, "fseeko");
551 
552 #ifndef PROFILE
553 	/* drop tmppath */
554 	if (pledge("stdio rpath dns inet", NULL) == -1)
555 		err(1, "pledge");
556 #endif
557 
558 	memset(&pfd, 0, sizeof(pfd));
559 	pfd.fd = dial(host, port);
560 
561 	if ((flags = fcntl(pfd.fd, F_GETFL)) == -1)
562 		err(1, "fcntl(F_GETFL)");
563 	if (fcntl(pfd.fd, F_SETFL, flags | O_NONBLOCK) == -1)
564 		err(1, "fcntl(F_SETFL)");
565 
566 	if (bufio_init(&bio) == -1)
567 		err(1, "bufio_init");
568 	bufio_set_fd(&bio, pfd.fd);
569 	if (tls && bufio_starttls(&bio, host, 0, NULL, 0, NULL, 0) == -1)
570 		err(1, "bufio_starttls");
571 
572 #ifndef PROFILE
573 	/* drop rpath dns inet */
574 	if (pledge("stdio", NULL) == -1)
575 		err(1, "pledge");
576 #endif
577 
578 	if ((!tls && strcmp(port, "80") != 0) ||
579 	    (tls && strcmp(port, "443")) != 0)
580 		nonstd = 1;
581 
582 	ret = bufio_compose_fmt(&bio,
583 	    "POST %s HTTP/1.1\r\n"
584 	    "Host: %s%s%s\r\n"
585 	    "Content-Type: application/json\r\n"
586 	    "Content-Length: %lld\r\n"
587 	    "User-Agent: %s\r\n"
588 	    "Connection: close\r\n",
589 	    path, host,
590 	    nonstd ? ":" : "", nonstd ? port : "",
591 	    (long long)paylen, USERAGENT);
592 	if (ret == -1)
593 		err(1, "bufio_compose_fmt");
594 
595 	if (username) {
596 		auth = basic_auth(username, password);
597 		ret = bufio_compose_fmt(&bio, "Authorization: basic %s\r\n",
598 		    auth);
599 		if (ret == -1)
600 			err(1, "bufio_compose_fmt");
601 		free(auth);
602 	}
603 
604 	if (bufio_compose(&bio, "\r\n", 2) == -1)
605 		err(1, "bufio_compose");
606 
607 	while (!done) {
608 		struct timespec	 elapsed, start, stop;
609 		char		 buf[BUFSIZ];
610 
611 		pfd.events = bufio2poll(&bio);
612 		clock_gettime(CLOCK_MONOTONIC, &start);
613 		ret = ppoll(&pfd, 1, &timeout, NULL);
614 		if (ret == -1)
615 			err(1, "poll");
616 		clock_gettime(CLOCK_MONOTONIC, &stop);
617 		timespecsub(&stop, &start, &elapsed);
618 		timespecsub(&timeout, &elapsed, &timeout);
619 		if (ret == 0 || timeout.tv_sec <= 0)
620 			errx(1, "timeout");
621 
622 		if (bio.wbuf.len > 0 && (pfd.revents & POLLOUT)) {
623 			if (bufio_write(&bio) == -1 && errno != EAGAIN)
624 				errx(1, "bufio_write: %s", bufio_io_err(&bio));
625 		}
626 		if (pfd.revents & POLLIN) {
627 			r = bufio_read(&bio);
628 			if (r == -1 && errno != EAGAIN)
629 				errx(1, "bufio_read: %s", bufio_io_err(&bio));
630 			if (r == 0)
631 				errx(1, "unexpected EOF");
632 
633 			for (;;) {
634 				line = buf_getdelim(&bio.rbuf, "\r\n", &len);
635 				if (line == NULL)
636 					break;
637 				if (response_code && *line == '\0') {
638 					/*
639 					 * end of headers, don't bother
640 					 * reading the body, if there is.
641 					 */
642 					done = 1;
643 					break;
644 				}
645 				if (response_code) {
646 					buf_drain(&bio.rbuf, len);
647 					continue;
648 				}
649 				spc = strchr(line, ' ');
650 				if (spc == NULL)
651 					errx(1, "bad reply");
652 				*spc++ = '\0';
653 				if (strcasecmp(line, "HTTP/1.1") != 0)
654 					errx(1, "unexpected protocol: %s",
655 					    line);
656 				line = spc;
657 
658 				spc = strchr(line, ' ');
659 				if (spc == NULL)
660 					errx(1, "bad reply");
661 				*spc++ = '\0';
662 
663 				response_code = strtonum(line, 100, 599,
664 				    &errstr);
665 				if (errstr != NULL)
666 					errx(1, "response code is %s: %s",
667 					    errstr, line);
668 
669 				buf_drain(&bio.rbuf, len);
670 			}
671 			if (done)
672 				break;
673 		}
674 
675 		if (!feof(tmpfp) && bio.wbuf.len < sizeof(buf)) {
676 			len = fread(buf, 1, sizeof(buf), tmpfp);
677 			if (len == 0) {
678 				if (ferror(tmpfp))
679 					err(1, "fread");
680 				continue;
681 			}
682 
683 			if (bufio_compose(&bio, buf, len) == -1)
684 				err(1, "buf_compose");
685 		}
686 	}
687 
688 	if (response_code >= 200 && response_code < 300)
689 		return 0;
690 	errx(1, "request failed with code %d", response_code);
691 }