1 #include <u.h>
2 #include <libc.h>
3 #include <ctype.h>
4 #include <bio.h>
5 #include <ip.h>
6 #include <libsec.h>
7 #include <auth.h>
8 #include <thread.h>
9 
10 typedef struct URL URL;
11 struct URL
12 {
13 	int	method;
14 	char	*host;
15 	char	*port;
16 	char	*page;
17 	char	*etag;
18 	char	*redirect;
19 	char	*postbody;
20 	char	*cred;
21 	long	mtime;
22 };
23 
24 typedef struct Range Range;
25 struct Range
26 {
27 	long	start;	/* only 2 gig supported, tdb */
28 	long	end;
29 };
30 
31 typedef struct Out Out;
32 struct Out
33 {
34 	int fd;
35 	int offset;				/* notional current offset in output */
36 	int written;			/* number of bytes successfully transferred to output */
37 	DigestState *curr;		/* digest state up to offset (if known) */
38 	DigestState *hiwat;		/* digest state of all bytes written */
39 };
40 
41 enum
42 {
43 	Http,
44 	Https,
45 	Ftp,
46 	Other
47 };
48 
49 enum
50 {
51 	Eof = 0,
52 	Error = -1,
53 	Server = -2,
54 	Changed = -3
55 };
56 
57 int debug;
58 char *ofile;
59 
60 
61 int	doftp(URL*, URL*, Range*, Out*, long);
62 int	dohttp(URL*, URL*,  Range*, Out*, long);
63 int	crackurl(URL*, char*);
64 Range*	crackrange(char*);
65 int	getheader(int, char*, int);
66 int	httpheaders(int, int, URL*, Range*);
67 int	httprcode(int);
68 int	cistrncmp(char*, char*, int);
69 int	cistrcmp(char*, char*);
70 void	initibuf(void);
71 int	readline(int, char*, int);
72 int	readibuf(int, char*, int);
73 int	dfprint(int, char*, ...);
74 void	unreadline(char*);
75 int	output(Out*, char*, int);
76 void	setoffset(Out*, int);
77 
78 int	verbose;
79 char	*net;
80 char	tcpdir[NETPATHLEN];
81 int	headerprint;
82 
83 struct {
84 	char	*name;
85 	int	(*f)(URL*, URL*, Range*, Out*, long);
86 } method[] = {
87 	{ "http",	dohttp },
88 	{ "https",	dohttp },
89 	{ "ftp",	doftp },
90 	{ "_______",	nil },
91 };
92 
93 void
94 usage(void)
95 {
96 	fprint(2, "usage: %s [-hv] [-o outfile] [-p body] [-x netmtpt] url\n", argv0);
97 	threadexitsall("usage");
98 }
99 
100 void
101 threadmain(int argc, char **argv)
102 {
103 	URL u;
104 	Range r;
105 	int errs, n;
106 	ulong mtime;
107 	Dir *d;
108 	char postbody[4096], *p, *e, *t, *hpx;
109 	URL px; /* Proxy */
110 	Out out;
111 
112 	ofile = nil;
113 	p = postbody;
114 	e = p + sizeof(postbody);
115 	r.start = 0;
116 	r.end = -1;
117 	mtime = 0;
118 	memset(&u, 0, sizeof(u));
119 	memset(&px, 0, sizeof(px));
120 	hpx = getenv("httpproxy");
121 
122 	ARGBEGIN {
123 	case 'o':
124 		ofile = ARGF();
125 		break;
126 	case 'd':
127 		debug = 1;
128 		break;
129 	case 'h':
130 		headerprint = 1;
131 		break;
132 	case 'v':
133 		verbose = 1;
134 		break;
135 	case 'x':
136 		net = ARGF();
137 		if(net == nil)
138 			usage();
139 		break;
140 	case 'p':
141 		t = ARGF();
142 		if(t == nil)
143 			usage();
144 		if(p != postbody)
145 			p = seprint(p, e, "&%s", t);
146 		else
147 			p = seprint(p, e, "%s", t);
148 		u.postbody = postbody;
149 		
150 		break;
151 	default:
152 		usage();
153 	} ARGEND;
154 
155 	if(net != nil){
156 		if(strlen(net) > sizeof(tcpdir)-5)
157 			sysfatal("network mount point too long");
158 		snprint(tcpdir, sizeof(tcpdir), "%s/tcp", net);
159 	} else
160 		snprint(tcpdir, sizeof(tcpdir), "tcp");
161 
162 	if(argc != 1)
163 		usage();
164 
165 	
166 	out.fd = 1;
167 	out.written = 0;
168 	out.offset = 0;
169 	out.curr = nil;
170 	out.hiwat = nil;
171 	if(ofile != nil){
172 		d = dirstat(ofile);
173 		if(d == nil){
174 			out.fd = create(ofile, OWRITE, 0664);
175 			if(out.fd < 0)
176 				sysfatal("creating %s: %r", ofile);
177 		} else {
178 			out.fd = open(ofile, OWRITE);
179 			if(out.fd < 0)
180 				sysfatal("can't open %s: %r", ofile);
181 			r.start = d->length;
182 			mtime = d->mtime;
183 			free(d);
184 		}
185 	}
186 
187 	errs = 0;
188 
189 	if(crackurl(&u, argv[0]) < 0)
190 		sysfatal("%r");
191 	if(hpx && crackurl(&px, hpx) < 0)
192 		sysfatal("%r");
193 
194 	for(;;){
195 		setoffset(&out, 0);
196 		/* transfer data */
197 		werrstr("");
198 		n = (*method[u.method].f)(&u, &px, &r, &out, mtime);
199 
200 		switch(n){
201 		case Eof:
202 			threadexitsall(0);
203 			break;
204 		case Error:
205 			if(errs++ < 10)
206 				continue;
207 			sysfatal("too many errors with no progress %r");
208 			break;
209 		case Server:
210 			sysfatal("server returned: %r");
211 			break;
212 		}
213 
214 		/* forward progress */
215 		errs = 0;
216 		r.start += n;
217 		if(r.start >= r.end)
218 			break;
219 	}
220 
221 	threadexitsall(0);
222 }
223 
224 int
225 crackurl(URL *u, char *s)
226 {
227 	char *p;
228 	int i;
229 
230 	if(u->host != nil){
231 		free(u->host);
232 		u->host = nil;
233 	}
234 	if(u->page != nil){
235 		free(u->page);
236 		u->page = nil;
237 	}
238 
239 	/* get type */
240 	u->method = Other;
241 	for(p = s; *p; p++){
242 		if(*p == '/'){
243 			u->method = Http;
244 			p = s;
245 			break;
246 		}
247 		if(*p == ':' && *(p+1)=='/' && *(p+2)=='/'){
248 			*p = 0;
249 			p += 3;
250 			for(i = 0; i < nelem(method); i++){
251 				if(cistrcmp(s, method[i].name) == 0){
252 					u->method = i;
253 					break;
254 				}
255 			}
256 			break;
257 		}
258 	}
259 
260 	if(u->method == Other){
261 		werrstr("unsupported URL type %s", s);
262 		return -1;
263 	}
264 
265 	/* get system */
266 	s = p;
267 	p = strchr(s, '/');
268 	if(p == nil){
269 		u->host = strdup(s);
270 		u->page = strdup("/");
271 	} else {
272 		u->page = strdup(p);
273 		*p = 0;
274 		u->host = strdup(s);
275 		*p = '/';
276 	}
277 
278 	if(p = strchr(u->host, ':')) {
279 		*p++ = 0;
280 		u->port = p;
281 	} else 
282 		u->port = method[u->method].name;
283 
284 	if(*(u->host) == 0){
285 		werrstr("bad url, null host");
286 		return -1;
287 	}
288 
289 	return 0;
290 }
291 
292 char *day[] = {
293 	"Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"
294 };
295 
296 char *month[] = {
297 	"Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
298 };
299 
300 struct
301 {
302 	int	fd;
303 	long	mtime;
304 } note;
305 
306 void
307 catch(void *v, char *s)
308 {
309 	Dir d;
310 
311 	USED(v);
312 	USED(s);
313 
314 	nulldir(&d);
315 	d.mtime = note.mtime;
316 	if(dirfwstat(note.fd, &d) < 0)
317 		sysfatal("catch: can't dirfwstat: %r");
318 	noted(NDFLT);
319 }
320 
321 int
322 dohttp(URL *u, URL *px, Range *r, Out *out, long mtime)
323 {
324 	int fd, cfd;
325 	int redirect, auth, loop;
326 	int n, rv, code;
327 	long tot, vtime;
328 	Tm *tm;
329 	char buf[1024];
330 	char err[ERRMAX];
331 
332 
333 	/*  always move back to a previous 512 byte bound because some
334 	 *  servers can't seem to deal with requests that start at the
335 	 *  end of the file
336 	 */
337 	if(r->start)
338 		r->start = ((r->start-1)/512)*512;
339 
340 	/* loop for redirects, requires reading both response code and headers */
341 	fd = -1;
342 	for(loop = 0; loop < 32; loop++){
343 		if(px->host == nil){
344 			fd = dial(netmkaddr(u->host, tcpdir, u->port), 0, 0, 0);
345 		} else {
346 			fd = dial(netmkaddr(px->host, tcpdir, px->port), 0, 0, 0);
347 		}
348 		if(fd < 0)
349 			return Error;
350 
351 		if(u->method == Https){
352 			int tfd;
353 			TLSconn conn;
354 
355 			memset(&conn, 0, sizeof conn);
356 			tfd = tlsClient(fd, &conn);
357 			if(tfd < 0){
358 				fprint(2, "tlsClient: %r\n");
359 				close(fd);
360 				return Error;
361 			}
362 			/* BUG: check cert here? */
363 			if(conn.cert)
364 				free(conn.cert);
365 			close(fd);
366 			fd = tfd;
367 		}
368 
369 		/* write request, use range if not start of file */
370 		if(u->postbody == nil){
371 			if(px->host == nil){
372 				dfprint(fd,	"GET %s HTTP/1.0\r\n"
373 						"Host: %s\r\n"
374 						"User-agent: Plan9/hget\r\n"
375 						"Cache-Control: no-cache\r\n"
376 						"Pragma: no-cache\r\n",
377 						u->page, u->host);
378 			} else {
379 				dfprint(fd,	"GET http://%s%s HTTP/1.0\r\n"
380 						"Host: %s\r\n"
381 						"User-agent: Plan9/hget\r\n"
382 						"Cache-Control: no-cache\r\n"
383 						"Pragma: no-cache\r\n",
384 						u->host, u->page, u->host);
385 			}
386 			if(u->cred)
387 				dfprint(fd,	"Authorization: Basic %s\r\n",
388 						u->cred);
389 		} else {
390 			dfprint(fd,	"POST %s HTTP/1.0\r\n"
391 					"Host: %s\r\n"
392 					"Content-type: application/x-www-form-urlencoded\r\n"
393 					"Content-length: %d\r\n"
394 					"User-agent: Plan9/hget\r\n"
395 					"\r\n",
396 					u->page, u->host, strlen(u->postbody));
397 			dfprint(fd,	"%s", u->postbody);
398 		}
399 		if(r->start != 0){
400 			dfprint(fd, "Range: bytes=%d-\n", r->start);
401 			if(u->etag != nil){
402 				dfprint(fd, "If-range: %s\n", u->etag);
403 			} else {
404 				tm = gmtime(mtime);
405 				dfprint(fd, "If-range: %s, %d %s %d %2d:%2.2d:%2.2d GMT\n",
406 					day[tm->wday], tm->mday, month[tm->mon],
407 					tm->year+1900, tm->hour, tm->min, tm->sec);
408 			}
409 		}
410 		if((cfd = open("/mnt/webcookies/http", ORDWR)) >= 0){
411 			if(fprint(cfd, "http://%s%s", u->host, u->page) > 0){
412 				while((n = read(cfd, buf, sizeof buf)) > 0){
413 					if(debug)
414 						write(2, buf, n);
415 					write(fd, buf, n);
416 				}
417 			}else{
418 				close(cfd);
419 				cfd = -1;
420 			}
421 		}
422 			
423 		dfprint(fd, "\r\n", u->host);
424 
425 		auth = 0;
426 		redirect = 0;
427 		initibuf();
428 		code = httprcode(fd);
429 		switch(code){
430 		case Error:	/* connection timed out */
431 		case Eof:
432 			close(fd);
433 			close(cfd);
434 			return code;
435 
436 		case 200:	/* OK */
437 		case 201:	/* Created */
438 		case 202:	/* Accepted */
439 			if(ofile == nil && r->start != 0)
440 				sysfatal("page changed underfoot");
441 			break;
442 
443 		case 204:	/* No Content */
444 			sysfatal("No Content");
445 
446 		case 206:	/* Partial Content */
447 			setoffset(out, r->start);
448 			break;
449 
450 		case 301:	/* Moved Permanently */
451 		case 302:	/* Moved Temporarily */
452 			redirect = 1;
453 			u->postbody = nil;
454 			break;
455 
456 		case 304:	/* Not Modified */
457 			break;
458 
459 		case 400:	/* Bad Request */
460 			sysfatal("Bad Request");
461 
462 		case 401:	/* Unauthorized */
463 			if (auth)
464 				sysfatal("Authentication failed");
465 			auth = 1;
466 			break;
467 
468 		case 402:	/* ??? */
469 			sysfatal("Unauthorized");
470 
471 		case 403:	/* Forbidden */
472 			sysfatal("Forbidden by server");
473 
474 		case 404:	/* Not Found */
475 			sysfatal("Not found on server");
476 
477 		case 407:	/* Proxy Authentication */
478 			sysfatal("Proxy authentication required");
479 
480 		case 500:	/* Internal server error */
481 			sysfatal("Server choked");
482 
483 		case 501:	/* Not implemented */
484 			sysfatal("Server can't do it!");
485 
486 		case 502:	/* Bad gateway */
487 			sysfatal("Bad gateway");
488 
489 		case 503:	/* Service unavailable */
490 			sysfatal("Service unavailable");
491 		
492 		default:
493 			sysfatal("Unknown response code %d", code);
494 		}
495 
496 		if(u->redirect != nil){
497 			free(u->redirect);
498 			u->redirect = nil;
499 		}
500 
501 		rv = httpheaders(fd, cfd, u, r);
502 		close(cfd);
503 		if(rv != 0){
504 			close(fd);
505 			return rv;
506 		}
507 
508 		if(!redirect && !auth)
509 			break;
510 
511 		if (redirect){
512 			if(u->redirect == nil)
513 				sysfatal("redirect: no URL");
514 			if(crackurl(u, u->redirect) < 0)
515 				sysfatal("redirect: %r");
516 		}
517 	}
518 
519 	/* transfer whatever you get */
520 	if(ofile != nil && u->mtime != 0){
521 		note.fd = out->fd;
522 		note.mtime = u->mtime;
523 		notify(catch);
524 	}
525 
526 	tot = 0;
527 	vtime = 0;
528 	for(;;){
529 		n = readibuf(fd, buf, sizeof(buf));
530 		if(n <= 0)
531 			break;
532 		if(output(out, buf, n) != n)
533 			break;
534 		tot += n;
535 		if(verbose && (vtime != time(0) || r->start == r->end)) {
536 			vtime = time(0);
537 			fprint(2, "%ld %ld\n", r->start+tot, r->end);
538 		}
539 	}
540 	notify(nil);
541 	close(fd);
542 
543 	if(ofile != nil && u->mtime != 0){
544 		Dir d;
545 
546 		rerrstr(err, sizeof err);
547 		nulldir(&d);
548 		d.mtime = u->mtime;
549 		if(dirfwstat(out->fd, &d) < 0)
550 			fprint(2, "couldn't set mtime: %r\n");
551 		errstr(err, sizeof err);
552 	}
553 
554 	return tot;
555 }
556 
557 /* get the http response code */
558 int
559 httprcode(int fd)
560 {
561 	int n;
562 	char *p;
563 	char buf[256];
564 
565 	n = readline(fd, buf, sizeof(buf)-1);
566 	if(n <= 0)
567 		return n;
568 	if(debug)
569 		fprint(2, "%d <- %s\n", fd, buf);
570 	p = strchr(buf, ' ');
571 	if(strncmp(buf, "HTTP/", 5) != 0 || p == nil){
572 		werrstr("bad response from server");
573 		return -1;
574 	}
575 	buf[n] = 0;
576 	return atoi(p+1);
577 }
578 
579 /* read in and crack the http headers, update u and r */
580 void	hhetag(char*, URL*, Range*);
581 void	hhmtime(char*, URL*, Range*);
582 void	hhclen(char*, URL*, Range*);
583 void	hhcrange(char*, URL*, Range*);
584 void	hhuri(char*, URL*, Range*);
585 void	hhlocation(char*, URL*, Range*);
586 void	hhauth(char*, URL*, Range*);
587 
588 struct {
589 	char *name;
590 	void (*f)(char*, URL*, Range*);
591 } headers[] = {
592 	{ "etag:", hhetag },
593 	{ "last-modified:", hhmtime },
594 	{ "content-length:", hhclen },
595 	{ "content-range:", hhcrange },
596 	{ "uri:", hhuri },
597 	{ "location:", hhlocation },
598 	{ "WWW-Authenticate:", hhauth },
599 };
600 int
601 httpheaders(int fd, int cfd, URL *u, Range *r)
602 {
603 	char buf[2048];
604 	char *p;
605 	int i, n;
606 
607 	for(;;){
608 		n = getheader(fd, buf, sizeof(buf));
609 		if(n <= 0)
610 			break;
611 		if(cfd >= 0)
612 			fprint(cfd, "%s\n", buf);
613 		for(i = 0; i < nelem(headers); i++){
614 			n = strlen(headers[i].name);
615 			if(cistrncmp(buf, headers[i].name, n) == 0){
616 				/* skip field name and leading white */
617 				p = buf + n;
618 				while(*p == ' ' || *p == '\t')
619 					p++;
620 
621 				(*headers[i].f)(p, u, r);
622 				break;
623 			}
624 		}
625 	}
626 	return n;
627 }
628 
629 /*
630  *  read a single mime header, collect continuations.
631  *
632  *  this routine assumes that there is a blank line twixt
633  *  the header and the message body, otherwise bytes will
634  *  be lost.
635  */
636 int
637 getheader(int fd, char *buf, int n)
638 {
639 	char *p, *e;
640 	int i;
641 
642 	n--;
643 	p = buf;
644 	for(e = p + n; ; p += i){
645 		i = readline(fd, p, e-p);
646 		if(i < 0)
647 			return i;
648 
649 		if(p == buf){
650 			/* first line */
651 			if(strchr(buf, ':') == nil)
652 				break;		/* end of headers */
653 		} else {
654 			/* continuation line */
655 			if(*p != ' ' && *p != '\t'){
656 				unreadline(p);
657 				*p = 0;
658 				break;		/* end of this header */
659 			}
660 		}
661 	}
662 	if(headerprint)
663 		print("%s\n", buf);
664 
665 	if(debug)
666 		fprint(2, "%d <- %s\n", fd, buf);
667 	return p-buf;
668 }
669 
670 void
671 hhetag(char *p, URL *u, Range *r)
672 {
673 	USED(r);
674 	
675 	if(u->etag != nil){
676 		if(strcmp(u->etag, p) != 0)
677 			sysfatal("file changed underfoot");
678 	} else
679 		u->etag = strdup(p);
680 }
681 
682 char*	monthchars = "janfebmaraprmayjunjulaugsepoctnovdec";
683 
684 void
685 hhmtime(char *p, URL *u, Range *r)
686 {
687 	char *month, *day, *yr, *hms;
688 	char *fields[6];
689 	Tm tm, now;
690 	int i;
691 
692 	USED(r);
693 	
694 	i = getfields(p, fields, 6, 1, " \t");
695 	if(i < 5)
696 		return;
697 
698 	day = fields[1];
699 	month = fields[2];
700 	yr = fields[3];
701 	hms = fields[4];
702 
703 	/* default time */
704 	now = *gmtime(time(0));
705 	tm = now;
706 	tm.yday = 0;
707 
708 	/* convert ascii month to a number twixt 1 and 12 */
709 	if(*month >= '0' && *month <= '9'){
710 		tm.mon = atoi(month) - 1;
711 		if(tm.mon < 0 || tm.mon > 11)
712 			tm.mon = 5;
713 	} else {
714 		for(p = month; *p; p++)
715 			*p = tolower((uchar)*p);
716 		for(i = 0; i < 12; i++)
717 			if(strncmp(&monthchars[i*3], month, 3) == 0){
718 				tm.mon = i;
719 				break;
720 			}
721 	}
722 
723 	tm.mday = atoi(day);
724 
725 	if(hms) {
726 		tm.hour = strtoul(hms, &p, 10);
727 		if(*p == ':') {
728 			p++;
729 			tm.min = strtoul(p, &p, 10);
730 			if(*p == ':') {
731 				p++;
732 				tm.sec = strtoul(p, &p, 10);
733 			}
734 		}
735 		if(tolower((uchar)*p) == 'p')
736 			tm.hour += 12;
737 	}
738 
739 	if(yr) {
740 		tm.year = atoi(yr);
741 		if(tm.year >= 1900)
742 			tm.year -= 1900;
743 	} else {
744 		if(tm.mon > now.mon || (tm.mon == now.mon && tm.mday > now.mday+1))
745 			tm.year--;
746 	}
747 
748 	strcpy(tm.zone, "GMT");
749 	/* convert to epoch seconds */
750 	u->mtime = tm2sec(&tm);
751 }
752 
753 void
754 hhclen(char *p, URL *u, Range *r)
755 {
756 	USED(u);
757 	
758 	r->end = atoi(p);
759 }
760 
761 void
762 hhcrange(char *p, URL *u, Range *r)
763 {
764 	char *x;
765 	vlong l;
766 
767 	USED(u);
768 	l = 0;
769 	x = strchr(p, '/');
770 	if(x)
771 		l = atoll(x+1);
772 	if(l == 0)
773 	x = strchr(p, '-');
774 	if(x)
775 		l = atoll(x+1);
776 	if(l)
777 		r->end = l;
778 }
779 
780 void
781 hhuri(char *p, URL *u, Range *r)
782 {
783 	USED(r);
784 	
785 	if(*p != '<')
786 		return;
787 	u->redirect = strdup(p+1);
788 	p = strchr(u->redirect, '>');
789 	if(p != nil)
790 		*p = 0;
791 }
792 
793 void
794 hhlocation(char *p, URL *u, Range *r)
795 {
796 	USED(r);
797 	
798 	u->redirect = strdup(p);
799 }
800 
801 void
802 hhauth(char *p, URL *u, Range *r)
803 {
804 	char *f[4];
805 	UserPasswd *up;
806 	char *s, cred[64];
807 	
808 	USED(r);
809 
810 	if (cistrncmp(p, "basic ", 6) != 0)
811 		sysfatal("only Basic authentication supported");
812 
813 	if (gettokens(p, f, nelem(f), "\"") < 2)
814 		sysfatal("garbled auth data");
815 
816 	if ((up = auth_getuserpasswd(auth_getkey, "proto=pass service=http dom=%q relm=%q",
817 	    	u->host, f[1])) == nil)
818 			sysfatal("cannot authenticate");
819 
820 	s = smprint("%s:%s", up->user, up->passwd);
821 	if(enc64(cred, sizeof(cred), (uchar *)s, strlen(s)) == -1)
822 		sysfatal("enc64");
823   		free(s);
824 
825 	assert(u->cred = strdup(cred));
826 }
827 
828 enum
829 {
830 	/* ftp return codes */
831 	Extra=		1,
832 	Success=	2,
833 	Incomplete=	3,
834 	TempFail=	4,
835 	PermFail=	5,
836 
837 	Nnetdir=	64,	/* max length of network directory paths */
838 	Ndialstr=	64		/* max length of dial strings */
839 };
840 
841 int ftpcmd(int, char*, ...);
842 int ftprcode(int, char*, int);
843 int hello(int);
844 int logon(int);
845 int xfertype(int, char*);
846 int passive(int, URL*);
847 int active(int, URL*);
848 int ftpxfer(int, Out*, Range*);
849 int terminateftp(int, int);
850 int getaddrport(char*, uchar*, uchar*);
851 int ftprestart(int, Out*, URL*, Range*, long);
852 
853 int
854 doftp(URL *u, URL *px, Range *r, Out *out, long mtime)
855 {
856 	int pid, ctl, data, rv;
857 	Waitmsg *w;
858 	char msg[64];
859 
860 	/* untested, proxy dosn't work with ftp (I think) */
861 	if(px->host == nil){
862 		ctl = dial(netmkaddr(u->host, tcpdir, u->port), 0, 0, 0);
863 	} else {
864 		ctl = dial(netmkaddr(px->host, tcpdir, px->port), 0, 0, 0);
865 	}
866 
867 	if(ctl < 0)
868 		return Error;
869 	if(net == nil)
870 		strcpy(tcpdir, "tcp");
871 
872 	initibuf();
873 
874 	rv = hello(ctl);
875 	if(rv < 0)
876 		return terminateftp(ctl, rv);
877 
878 	rv = logon(ctl);
879 	if(rv < 0)
880 		return terminateftp(ctl, rv);
881 
882 	rv = xfertype(ctl, "I");
883 	if(rv < 0)
884 		return terminateftp(ctl, rv);
885 
886 	/* if file is up to date and the right size, stop */
887 	if(ftprestart(ctl, out, u, r, mtime) > 0){
888 		close(ctl);
889 		return Eof;
890 	}
891 		
892 	/* first try passive mode, then active */
893 	data = passive(ctl, u);
894 	if(data < 0){
895 		data = active(ctl, u);
896 		if(data < 0)
897 			return Error;
898 	}
899 
900 	/* fork */
901 	switch(pid = fork()){
902 	case -1:
903 		close(data);
904 		return terminateftp(ctl, Error);
905 	case 0:
906 		ftpxfer(data, out, r);
907 		close(data);
908 		#undef _exits
909 		_exits(0);
910 	default:
911 		close(data);
912 		break;
913 	}
914 
915 	/* wait for reply message */
916 	rv = ftprcode(ctl, msg, sizeof(msg));
917 	close(ctl);
918 
919 	/* wait for process to terminate */
920 	w = nil;
921 	for(;;){
922 		free(w);
923 		w = wait();
924 		if(w == nil)
925 			return Error;
926 		if(w->pid == pid){
927 			if(w->msg[0] == 0){
928 				free(w);
929 				break;
930 			}
931 			werrstr("xfer: %s", w->msg);
932 			free(w);
933 			return Error;
934 		}
935 	}
936 
937 	switch(rv){
938 	case Success:
939 		return Eof;
940 	case TempFail:
941 		return Server;
942 	default:
943 		return Error;
944 	}
945 }
946 
947 int
948 ftpcmd(int ctl, char *fmt, ...)
949 {
950 	va_list arg;
951 	char buf[2*1024], *s;
952 
953 	va_start(arg, fmt);
954 	s = vseprint(buf, buf + (sizeof(buf)-4) / sizeof(*buf), fmt, arg);
955 	va_end(arg);
956 	if(debug)
957 		fprint(2, "%d -> %s\n", ctl, buf);
958 	*s++ = '\r';
959 	*s++ = '\n';
960 	if(write(ctl, buf, s - buf) != s - buf)
961 		return -1;
962 	return 0;
963 }
964 
965 int
966 ftprcode(int ctl, char *msg, int len)
967 {
968 	int rv;
969 	int i;
970 	char *p;
971 
972 	len--;	/* room for terminating null */
973 	for(;;){
974 		*msg = 0;
975 		i = readline(ctl, msg, len);
976 		if(i < 0)
977 			break;
978 		if(debug)
979 			fprint(2, "%d <- %s\n", ctl, msg);
980 
981 		/* stop if not a continuation */
982 		rv = strtol(msg, &p, 10);
983 		if(rv >= 100 && rv < 600 && p==msg+3 && *p == ' ')
984 			return rv/100;
985 	}
986 	*msg = 0;
987 
988 	return -1;
989 }
990 
991 int
992 hello(int ctl)
993 {
994 	char msg[1024];
995 
996 	/* wait for hello from other side */
997 	if(ftprcode(ctl, msg, sizeof(msg)) != Success){
998 		werrstr("HELLO: %s", msg);
999 		return Server;
1000 	}
1001 	return 0;
1002 }
1003 
1004 int
1005 getdec(char *p, int n)
1006 {
1007 	int x = 0;
1008 	int i;
1009 
1010 	for(i = 0; i < n; i++)
1011 		x = x*10 + (*p++ - '0');
1012 	return x;
1013 }
1014 
1015 int
1016 ftprestart(int ctl, Out *out, URL *u, Range *r, long mtime)
1017 {
1018 	Tm tm;
1019 	char msg[1024];
1020 	long x, rmtime;
1021 
1022 	ftpcmd(ctl, "MDTM %s", u->page);
1023 	if(ftprcode(ctl, msg, sizeof(msg)) != Success){
1024 		r->start = 0;
1025 		return 0;		/* need to do something */
1026 	}
1027 
1028 	/* decode modification time */
1029 	if(strlen(msg) < 4 + 4 + 2 + 2 + 2 + 2 + 2){
1030 		r->start = 0;
1031 		return 0;		/* need to do something */
1032 	}
1033 	memset(&tm, 0, sizeof(tm));
1034 	tm.year = getdec(msg+4, 4) - 1900;
1035 	tm.mon = getdec(msg+4+4, 2) - 1;
1036 	tm.mday = getdec(msg+4+4+2, 2);
1037 	tm.hour = getdec(msg+4+4+2+2, 2);
1038 	tm.min = getdec(msg+4+4+2+2+2, 2);
1039 	tm.sec = getdec(msg+4+4+2+2+2+2, 2);
1040 	strcpy(tm.zone, "GMT");
1041 	rmtime = tm2sec(&tm);
1042 	if(rmtime > mtime)
1043 		r->start = 0;
1044 
1045 	/* get size */
1046 	ftpcmd(ctl, "SIZE %s", u->page);
1047 	if(ftprcode(ctl, msg, sizeof(msg)) == Success){
1048 		x = atol(msg+4);
1049 		if(r->start == x)
1050 			return 1;	/* we're up to date */
1051 		r->end = x;
1052 	}
1053 
1054 	/* seek to restart point */
1055 	if(r->start > 0){
1056 		ftpcmd(ctl, "REST %lud", r->start);
1057 		if(ftprcode(ctl, msg, sizeof(msg)) == Incomplete){
1058 			setoffset(out, r->start);
1059 		}else
1060 			r->start = 0;
1061 	}
1062 
1063 	return 0;	/* need to do something */
1064 }
1065 
1066 int
1067 logon(int ctl)
1068 {
1069 	char msg[1024];
1070 
1071 	/* login anonymous */
1072 	ftpcmd(ctl, "USER anonymous");
1073 	switch(ftprcode(ctl, msg, sizeof(msg))){
1074 	case Success:
1075 		return 0;
1076 	case Incomplete:
1077 		break;	/* need password */
1078 	default:
1079 		werrstr("USER: %s", msg);
1080 		return Server;
1081 	}
1082 
1083 	/* send user id as password */
1084 	sprint(msg, "%s@closedmind.org", getuser());
1085 	ftpcmd(ctl, "PASS %s", msg);
1086 	if(ftprcode(ctl, msg, sizeof(msg)) != Success){
1087 		werrstr("PASS: %s", msg);
1088 		return Server;
1089 	}
1090 
1091 	return 0;
1092 }
1093 
1094 int
1095 xfertype(int ctl, char *t)
1096 {
1097 	char msg[1024];
1098 
1099 	ftpcmd(ctl, "TYPE %s", t);
1100 	if(ftprcode(ctl, msg, sizeof(msg)) != Success){
1101 		werrstr("TYPE %s: %s", t, msg);
1102 		return Server;
1103 	}
1104 
1105 	return 0;
1106 }
1107 
1108 int
1109 passive(int ctl, URL *u)
1110 {
1111 	char msg[1024];
1112 	char ipaddr[32];
1113 	char *f[6];
1114 	char *p;
1115 	int fd;
1116 	int port;
1117 	char aport[12];
1118 
1119 	ftpcmd(ctl, "PASV");
1120 	if(ftprcode(ctl, msg, sizeof(msg)) != Success)
1121 		return Error;
1122 
1123 	/* get address and port number from reply, this is AI */
1124 	p = strchr(msg, '(');
1125 	if(p == nil){
1126 		for(p = msg+3; *p; p++)
1127 			if(isdigit((uchar)*p))
1128 				break;
1129 	} else
1130 		p++;
1131 	if(getfields(p, f, 6, 0, ",)") < 6){
1132 		werrstr("ftp protocol botch");
1133 		return Server;
1134 	}
1135 	snprint(ipaddr, sizeof(ipaddr), "%s.%s.%s.%s",
1136 		f[0], f[1], f[2], f[3]);
1137 	port = ((atoi(f[4])&0xff)<<8) + (atoi(f[5])&0xff);
1138 	sprint(aport, "%d", port);
1139 
1140 	/* open data connection */
1141 	fd = dial(netmkaddr(ipaddr, tcpdir, aport), 0, 0, 0);
1142 	if(fd < 0){
1143 		werrstr("passive mode failed: %r");
1144 		return Error;
1145 	}
1146 
1147 	/* tell remote to send a file */
1148 	ftpcmd(ctl, "RETR %s", u->page);
1149 	if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
1150 		werrstr("RETR %s: %s", u->page, msg);
1151 		return Error;
1152 	}
1153 	return fd;
1154 }
1155 
1156 int
1157 active(int ctl, URL *u)
1158 {
1159 	char msg[1024];
1160 	char dir[40], ldir[40];
1161 	uchar ipaddr[4];
1162 	uchar port[2];
1163 	int lcfd, dfd, afd;
1164 
1165 	/* announce a port for the call back */
1166 	snprint(msg, sizeof(msg), "%s!*!0", tcpdir);
1167 	afd = announce(msg, dir);
1168 	if(afd < 0)
1169 		return Error;
1170 
1171 	/* get a local address/port of the annoucement */
1172 	if(getaddrport(dir, ipaddr, port) < 0){
1173 		close(afd);
1174 		return Error;
1175 	}
1176 
1177 	/* tell remote side address and port*/
1178 	ftpcmd(ctl, "PORT %d,%d,%d,%d,%d,%d", ipaddr[0], ipaddr[1], ipaddr[2],
1179 		ipaddr[3], port[0], port[1]);
1180 	if(ftprcode(ctl, msg, sizeof(msg)) != Success){
1181 		close(afd);
1182 		werrstr("active: %s", msg);
1183 		return Error;
1184 	}
1185 
1186 	/* tell remote to send a file */
1187 	ftpcmd(ctl, "RETR %s", u->page);
1188 	if(ftprcode(ctl, msg, sizeof(msg)) != Extra){
1189 		close(afd);
1190 		werrstr("RETR: %s", msg);
1191 		return Server;
1192 	}
1193 
1194 	/* wait for a connection */
1195 	lcfd = listen(dir, ldir);
1196 	if(lcfd < 0){
1197 		close(afd);
1198 		return Error;
1199 	}
1200 	dfd = accept(lcfd, ldir);
1201 	if(dfd < 0){
1202 		close(afd);
1203 		close(lcfd);
1204 		return Error;
1205 	}
1206 	close(afd);
1207 	close(lcfd);
1208 	
1209 	return dfd;
1210 }
1211 
1212 int
1213 ftpxfer(int in, Out *out, Range *r)
1214 {
1215 	char buf[1024];
1216 	long vtime;
1217 	int i, n;
1218 
1219 	vtime = 0;
1220 	for(n = 0;;n += i){
1221 		i = read(in, buf, sizeof(buf));
1222 		if(i == 0)
1223 			break;
1224 		if(i < 0)
1225 			return Error;
1226 		if(output(out, buf, i) != i)
1227 			return Error;
1228 		r->start += i;
1229 		if(verbose && (vtime != time(0) || r->start == r->end)) {
1230 			vtime = time(0);
1231 			fprint(2, "%ld %ld\n", r->start, r->end);
1232 		}
1233 	}
1234 	return n;
1235 }
1236 
1237 int
1238 terminateftp(int ctl, int rv)
1239 {
1240 	close(ctl);
1241 	return rv;
1242 }
1243 
1244 /*
1245  * case insensitive strcmp (why aren't these in libc?)
1246  */
1247 int
1248 cistrncmp(char *a, char *b, int n)
1249 {
1250 	while(n-- > 0){
1251 		if(tolower((uchar)*a++) != tolower((uchar)*b++))
1252 			return -1;
1253 	}
1254 	return 0;
1255 }
1256 
1257 int
1258 cistrcmp(char *a, char *b)
1259 {
1260 	while(*a || *b)
1261 		if(tolower((uchar)*a++) != tolower((uchar)*b++))
1262 			return -1;
1263 
1264 	return 0;
1265 }
1266 
1267 /*
1268  *  buffered io
1269  */
1270 struct
1271 {
1272 	char *rp;
1273 	char *wp;
1274 	char buf[4*1024];
1275 } b;
1276 
1277 void
1278 initibuf(void)
1279 {
1280 	b.rp = b.wp = b.buf;
1281 }
1282 
1283 /*
1284  *  read a possibly buffered line, strip off trailing while
1285  */
1286 int
1287 readline(int fd, char *buf, int len)
1288 {
1289 	int n;
1290 	char *p;
1291 	int eof = 0;
1292 
1293 	len--;
1294 
1295 	for(p = buf;;){
1296 		if(b.rp >= b.wp){
1297 			n = read(fd, b.wp, sizeof(b.buf)/2);
1298 			if(n < 0)
1299 				return -1;
1300 			if(n == 0){
1301 				eof = 1;
1302 				break;
1303 			}
1304 			b.wp += n;
1305 		}
1306 		n = *b.rp++;
1307 		if(len > 0){
1308 			*p++ = n;
1309 			len--;
1310 		}
1311 		if(n == '\n')
1312 			break;
1313 	}
1314 
1315 	/* drop trailing white */
1316 	for(;;){
1317 		if(p <= buf)
1318 			break;
1319 		n = *(p-1);
1320 		if(n != ' ' && n != '\t' && n != '\r' && n != '\n')
1321 			break;
1322 		p--;
1323 	}
1324 	*p = 0;
1325 
1326 	if(eof && p == buf)
1327 		return -1;
1328 
1329 	return p-buf;
1330 }
1331 
1332 void
1333 unreadline(char *line)
1334 {
1335 	int i, n;
1336 
1337 	i = strlen(line);
1338 	n = b.wp-b.rp;
1339 	memmove(&b.buf[i+1], b.rp, n);
1340 	memmove(b.buf, line, i);
1341 	b.buf[i] = '\n';
1342 	b.rp = b.buf;
1343 	b.wp = b.rp + i + 1 + n;
1344 }
1345 
1346 int
1347 readibuf(int fd, char *buf, int len)
1348 {
1349 	int n;
1350 
1351 	n = b.wp-b.rp;
1352 	if(n > 0){
1353 		if(n > len)
1354 			n = len;
1355 		memmove(buf, b.rp, n);
1356 		b.rp += n;
1357 		return n;
1358 	}
1359 	return read(fd, buf, len);
1360 }
1361 
1362 int
1363 dfprint(int fd, char *fmt, ...)
1364 {
1365 	char buf[4*1024];
1366 	va_list arg;
1367 
1368 	va_start(arg, fmt);
1369 	vseprint(buf, buf+sizeof(buf), fmt, arg);
1370 	va_end(arg);
1371 	if(debug)
1372 		fprint(2, "%d -> %s", fd, buf);
1373 	return fprint(fd, "%s", buf);
1374 }
1375 
1376 int
1377 getaddrport(char *dir, uchar *ipaddr, uchar *port)
1378 {
1379 	char buf[256];
1380 	int fd, i;
1381 	char *p;
1382 
1383 	snprint(buf, sizeof(buf), "%s/local", dir);
1384 	fd = open(buf, OREAD);
1385 	if(fd < 0)
1386 		return -1;
1387 	i = read(fd, buf, sizeof(buf)-1);
1388 	close(fd);
1389 	if(i <= 0)
1390 		return -1;
1391 	buf[i] = 0;
1392 	p = strchr(buf, '!');
1393 	if(p != nil)
1394 		*p++ = 0;
1395 	v4parseip(ipaddr, buf);
1396 	i = atoi(p);
1397 	port[0] = i>>8;
1398 	port[1] = i;
1399 	return 0;
1400 }
1401 
1402 void
1403 md5free(DigestState *state)
1404 {
1405 	uchar x[MD5dlen];
1406 	md5(nil, 0, x, state);
1407 }
1408 
1409 DigestState*
1410 md5dup(DigestState *state)
1411 {
1412 	DigestState *s2;
1413 	
1414 	s2 = malloc(sizeof(DigestState));
1415 	if(s2 == nil)
1416 		sysfatal("malloc: %r");
1417 	*s2 = *state;
1418 	s2->malloced = 1;
1419 	return s2;
1420 }
1421 
1422 void
1423 setoffset(Out *out, int offset)
1424 {
1425 	md5free(out->curr);
1426 	if(offset == 0)
1427 		out->curr = md5(nil, 0, nil, nil);
1428 	else
1429 		out->curr = nil;
1430 	out->offset = offset;
1431 }
1432 
1433 /*
1434  * write some output, discarding it (but keeping track)
1435  * if we've already written it. if we've gone backwards,
1436  * verify that everything previously written matches
1437  * that which would have been written from the current
1438  * output.
1439  */
1440 int
1441 output(Out *out, char *buf, int nb)
1442 {
1443 	int n, d;
1444 	uchar m0[MD5dlen], m1[MD5dlen];
1445 
1446 	n = nb;
1447 	d = out->written - out->offset;
1448 	assert(d >= 0);
1449 	if(d > 0){
1450 		if(n < d){
1451 			if(out->curr != nil)
1452 				md5((uchar*)buf, n, nil, out->curr);
1453 			out->offset += n;
1454 			return n;
1455 		}
1456 		if(out->curr != nil){
1457 			md5((uchar*)buf, d, m0, out->curr);
1458 			out->curr = nil;
1459 			md5(nil, 0, m1, md5dup(out->hiwat));
1460 			if(memcmp(m0, m1, MD5dlen) != 0){
1461 				fprint(2, "integrity check failure at offset %d\n", out->written);
1462 				return -1;
1463 			}
1464 		}
1465 		buf += d;
1466 		n -= d;
1467 		out->offset += d;
1468 	}
1469 	if(n > 0){
1470 		out->hiwat = md5((uchar*)buf, n, nil, out->hiwat);
1471 		n = write(out->fd, buf, n);
1472 		if(n > 0){
1473 			out->offset += n;
1474 			out->written += n;
1475 		}
1476 	}
1477 	return n + d;
1478 }
1479