?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Sync usage with man page.
[netbsd-mini2440.git] / external / bsd / fetch / dist / libfetch / http.c
blobf2b73c5243ab19e28402f9aa72d39d21edbae1b5
1 /* $NetBSD: http.c,v 1.25 2009/10/15 12:36:57 joerg Exp $ */
2 /*-
3 * Copyright (c) 2000-2004 Dag-Erling Coïdan Smørgrav
4 * Copyright (c) 2003 Thomas Klausner <wiz@NetBSD.org>
5 * Copyright (c) 2008, 2009 Joerg Sonnenberger <joerg@NetBSD.org>
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer
13 * in this position and unchanged.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. The name of the author may not be used to endorse or promote products
18 * derived from this software without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
21 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
22 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
23 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
24 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
25 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
29 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30 *
31 * $FreeBSD: http.c,v 1.83 2008/02/06 11:39:55 des Exp $
32 */
33
34 /*
35 * The following copyright applies to the base64 code:
36 *
37 *-
38 * Copyright 1997 Massachusetts Institute of Technology
39 *
40 * Permission to use, copy, modify, and distribute this software and
41 * its documentation for any purpose and without fee is hereby
42 * granted, provided that both the above copyright notice and this
43 * permission notice appear in all copies, that both the above
44 * copyright notice and this permission notice appear in all
45 * supporting documentation, and that the name of M.I.T. not be used
46 * in advertising or publicity pertaining to distribution of the
47 * software without specific, written prior permission. M.I.T. makes
48 * no representations about the suitability of this software for any
49 * purpose. It is provided "as is" without express or implied
50 * warranty.
51 *
52 * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''. M.I.T. DISCLAIMS
53 * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE,
54 * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
55 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT
56 * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
57 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
58 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
59 * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
60 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
61 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
62 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63 * SUCH DAMAGE.
64 */
65
66 #if defined(__linux__) || defined(__MINT__)
67 /* Keep this down to Linux or MiNT, it can create surprises elsewhere. */
68 #define _GNU_SOURCE
69 #endif
70
71 /* Needed for gmtime_r on Interix */
72 #define _REENTRANT
73
74 #if HAVE_CONFIG_H
75 #include "config.h"
76 #endif
77 #ifndef NETBSD
78 #include <nbcompat.h>
79 #endif
80
81 #include <sys/types.h>
82 #include <sys/socket.h>
83
84 #include <ctype.h>
85 #include <errno.h>
86 #include <locale.h>
87 #include <stdarg.h>
88 #ifndef NETBSD
89 #include <nbcompat/stdio.h>
90 #else
91 #include <stdio.h>
92 #endif
93 #include <stdlib.h>
94 #include <string.h>
95 #include <time.h>
96 #include <unistd.h>
97
98 #include <netinet/in.h>
99 #include <netinet/tcp.h>
100
101 #ifndef NETBSD
102 #include <nbcompat/netdb.h>
103 #else
104 #include <netdb.h>
105 #endif
106
107 #include <arpa/inet.h>
108
109 #include "fetch.h"
110 #include "common.h"
111 #include "httperr.h"
112
113 /* Maximum number of redirects to follow */
114 #define MAX_REDIRECT 5
115
116 /* Symbolic names for reply codes we care about */
117 #define HTTP_OK 200
118 #define HTTP_PARTIAL 206
119 #define HTTP_MOVED_PERM 301
120 #define HTTP_MOVED_TEMP 302
121 #define HTTP_SEE_OTHER 303
122 #define HTTP_NOT_MODIFIED 304
123 #define HTTP_TEMP_REDIRECT 307
124 #define HTTP_NEED_AUTH 401
125 #define HTTP_NEED_PROXY_AUTH 407
126 #define HTTP_BAD_RANGE 416
127 #define HTTP_PROTOCOL_ERROR 999
128
129 #define HTTP_REDIRECT(xyz) ((xyz) == HTTP_MOVED_PERM \
130 || (xyz) == HTTP_MOVED_TEMP \
131 || (xyz) == HTTP_TEMP_REDIRECT \
132 || (xyz) == HTTP_SEE_OTHER)
133
134 #define HTTP_ERROR(xyz) ((xyz) > 400 && (xyz) < 599)
135
136
137 /*****************************************************************************
138 * I/O functions for decoding chunked streams
139 */
140
141 struct httpio
142 {
143 conn_t *conn; /* connection */
144 int chunked; /* chunked mode */
145 char *buf; /* chunk buffer */
146 size_t bufsize; /* size of chunk buffer */
147 ssize_t buflen; /* amount of data currently in buffer */
148 int bufpos; /* current read offset in buffer */
149 int eof; /* end-of-file flag */
150 int error; /* error flag */
151 size_t chunksize; /* remaining size of current chunk */
152 };
153
154 /*
155 * Get next chunk header
156 */
157 static int
158 http_new_chunk(struct httpio *io)
159 {
160 char *p;
161
162 if (fetch_getln(io->conn) == -1)
163 return (-1);
164
165 if (io->conn->buflen < 2 || !isxdigit((unsigned char)*io->conn->buf))
166 return (-1);
167
168 for (p = io->conn->buf; *p && !isspace((unsigned char)*p); ++p) {
169 if (*p == ';')
170 break;
171 if (!isxdigit((unsigned char)*p))
172 return (-1);
173 if (isdigit((unsigned char)*p)) {
174 io->chunksize = io->chunksize * 16 +
175 *p - '0';
176 } else {
177 io->chunksize = io->chunksize * 16 +
178 10 + tolower((unsigned char)*p) - 'a';
179 }
180 }
181
182 return (io->chunksize);
183 }
184
185 /*
186 * Grow the input buffer to at least len bytes
187 */
188 static int
189 http_growbuf(struct httpio *io, size_t len)
190 {
191 char *tmp;
192
193 if (io->bufsize >= len)
194 return (0);
195
196 if ((tmp = realloc(io->buf, len)) == NULL)
197 return (-1);
198 io->buf = tmp;
199 io->bufsize = len;
200 return (0);
201 }
202
203 /*
204 * Fill the input buffer, do chunk decoding on the fly
205 */
206 static int
207 http_fillbuf(struct httpio *io, size_t len)
208 {
209 if (io->error)
210 return (-1);
211 if (io->eof)
212 return (0);
213
214 if (io->chunked == 0) {
215 if (http_growbuf(io, len) == -1)
216 return (-1);
217 if ((io->buflen = fetch_read(io->conn, io->buf, len)) == -1) {
218 io->error = 1;
219 return (-1);
220 }
221 io->bufpos = 0;
222 return (io->buflen);
223 }
224
225 if (io->chunksize == 0) {
226 switch (http_new_chunk(io)) {
227 case -1:
228 io->error = 1;
229 return (-1);
230 case 0:
231 io->eof = 1;
232 return (0);
233 }
234 }
235
236 if (len > io->chunksize)
237 len = io->chunksize;
238 if (http_growbuf(io, len) == -1)
239 return (-1);
240 if ((io->buflen = fetch_read(io->conn, io->buf, len)) == -1) {
241 io->error = 1;
242 return (-1);
243 }
244 io->chunksize -= io->buflen;
245
246 if (io->chunksize == 0) {
247 char endl[2];
248 ssize_t len2;
249
250 len2 = fetch_read(io->conn, endl, 2);
251 if (len2 == 1 && fetch_read(io->conn, endl + 1, 1) != 1)
252 return (-1);
253 if (len2 == -1 || endl[0] != '\r' || endl[1] != '\n')
254 return (-1);
255 }
256
257 io->bufpos = 0;
258
259 return (io->buflen);
260 }
261
262 /*
263 * Read function
264 */
265 static ssize_t
266 http_readfn(void *v, void *buf, size_t len)
267 {
268 struct httpio *io = (struct httpio *)v;
269 size_t l, pos;
270
271 if (io->error)
272 return (-1);
273 if (io->eof)
274 return (0);
275
276 for (pos = 0; len > 0; pos += l, len -= l) {
277 /* empty buffer */
278 if (!io->buf || io->bufpos == io->buflen)
279 if (http_fillbuf(io, len) < 1)
280 break;
281 l = io->buflen - io->bufpos;
282 if (len < l)
283 l = len;
284 memcpy((char *)buf + pos, io->buf + io->bufpos, l);
285 io->bufpos += l;
286 }
287
288 if (!pos && io->error)
289 return (-1);
290 return (pos);
291 }
292
293 /*
294 * Write function
295 */
296 static ssize_t
297 http_writefn(void *v, const void *buf, size_t len)
298 {
299 struct httpio *io = (struct httpio *)v;
300
301 return (fetch_write(io->conn, buf, len));
302 }
303
304 /*
305 * Close function
306 */
307 static void
308 http_closefn(void *v)
309 {
310 struct httpio *io = (struct httpio *)v;
311
312 fetch_close(io->conn);
313 if (io->buf)
314 free(io->buf);
315 free(io);
316 }
317
318 /*
319 * Wrap a file descriptor up
320 */
321 static fetchIO *
322 http_funopen(conn_t *conn, int chunked)
323 {
324 struct httpio *io;
325 fetchIO *f;
326
327 if ((io = calloc(1, sizeof(*io))) == NULL) {
328 fetch_syserr();
329 return (NULL);
330 }
331 io->conn = conn;
332 io->chunked = chunked;
333 f = fetchIO_unopen(io, http_readfn, http_writefn, http_closefn);
334 if (f == NULL) {
335 fetch_syserr();
336 free(io);
337 return (NULL);
338 }
339 return (f);
340 }
341
342
343 /*****************************************************************************
344 * Helper functions for talking to the server and parsing its replies
345 */
346
347 /* Header types */
348 typedef enum {
349 hdr_syserror = -2,
350 hdr_error = -1,
351 hdr_end = 0,
352 hdr_unknown = 1,
353 hdr_content_length,
354 hdr_content_range,
355 hdr_last_modified,
356 hdr_location,
357 hdr_transfer_encoding,
358 hdr_www_authenticate
359 } hdr_t;
360
361 /* Names of interesting headers */
362 static struct {
363 hdr_t num;
364 const char *name;
365 } hdr_names[] = {
366 { hdr_content_length, "Content-Length" },
367 { hdr_content_range, "Content-Range" },
368 { hdr_last_modified, "Last-Modified" },
369 { hdr_location, "Location" },
370 { hdr_transfer_encoding, "Transfer-Encoding" },
371 { hdr_www_authenticate, "WWW-Authenticate" },
372 { hdr_unknown, NULL },
373 };
374
375 /*
376 * Send a formatted line; optionally echo to terminal
377 */
378 static int
379 http_cmd(conn_t *conn, const char *fmt, ...)
380 {
381 va_list ap;
382 size_t len;
383 char *msg;
384 int r;
385
386 va_start(ap, fmt);
387 len = vasprintf(&msg, fmt, ap);
388 va_end(ap);
389
390 if (msg == NULL) {
391 errno = ENOMEM;
392 fetch_syserr();
393 return (-1);
394 }
395
396 r = fetch_putln(conn, msg, len);
397 free(msg);
398
399 if (r == -1) {
400 fetch_syserr();
401 return (-1);
402 }
403
404 return (0);
405 }
406
407 /*
408 * Get and parse status line
409 */
410 static int
411 http_get_reply(conn_t *conn)
412 {
413 char *p;
414
415 if (fetch_getln(conn) == -1)
416 return (-1);
417 /*
418 * A valid status line looks like "HTTP/m.n xyz reason" where m
419 * and n are the major and minor protocol version numbers and xyz
420 * is the reply code.
421 * Unfortunately, there are servers out there (NCSA 1.5.1, to name
422 * just one) that do not send a version number, so we can't rely
423 * on finding one, but if we do, insist on it being 1.0 or 1.1.
424 * We don't care about the reason phrase.
425 */
426 if (strncmp(conn->buf, "HTTP", 4) != 0)
427 return (HTTP_PROTOCOL_ERROR);
428 p = conn->buf + 4;
429 if (*p == '/') {
430 if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1'))
431 return (HTTP_PROTOCOL_ERROR);
432 p += 4;
433 }
434 if (*p != ' ' ||
435 !isdigit((unsigned char)p[1]) ||
436 !isdigit((unsigned char)p[2]) ||
437 !isdigit((unsigned char)p[3]))
438 return (HTTP_PROTOCOL_ERROR);
439
440 conn->err = (p[1] - '0') * 100 + (p[2] - '0') * 10 + (p[3] - '0');
441 return (conn->err);
442 }
443
444 /*
445 * Check a header; if the type matches the given string, return a pointer
446 * to the beginning of the value.
447 */
448 static const char *
449 http_match(const char *str, const char *hdr)
450 {
451 while (*str && *hdr &&
452 tolower((unsigned char)*str++) == tolower((unsigned char)*hdr++))
453 /* nothing */;
454 if (*str || *hdr != ':')
455 return (NULL);
456 while (*hdr && isspace((unsigned char)*++hdr))
457 /* nothing */;
458 return (hdr);
459 }
460
461 /*
462 * Get the next header and return the appropriate symbolic code.
463 */
464 static hdr_t
465 http_next_header(conn_t *conn, const char **p)
466 {
467 int i;
468
469 if (fetch_getln(conn) == -1)
470 return (hdr_syserror);
471 while (conn->buflen && isspace((unsigned char)conn->buf[conn->buflen - 1]))
472 conn->buflen--;
473 conn->buf[conn->buflen] = '\0';
474 if (conn->buflen == 0)
475 return (hdr_end);
476 /*
477 * We could check for malformed headers but we don't really care.
478 * A valid header starts with a token immediately followed by a
479 * colon; a token is any sequence of non-control, non-whitespace
480 * characters except "()<>@,;:\\\"{}".
481 */
482 for (i = 0; hdr_names[i].num != hdr_unknown; i++)
483 if ((*p = http_match(hdr_names[i].name, conn->buf)) != NULL)
484 return (hdr_names[i].num);
485 return (hdr_unknown);
486 }
487
488 /*
489 * Parse a last-modified header
490 */
491 static int
492 http_parse_mtime(const char *p, time_t *mtime)
493 {
494 char locale[64], *r;
495 struct tm tm;
496
497 strncpy(locale, setlocale(LC_TIME, NULL), sizeof(locale));
498 setlocale(LC_TIME, "C");
499 r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm);
500 /* XXX should add support for date-2 and date-3 */
501 setlocale(LC_TIME, locale);
502 if (r == NULL)
503 return (-1);
504 *mtime = timegm(&tm);
505 return (0);
506 }
507
508 /*
509 * Parse a content-length header
510 */
511 static int
512 http_parse_length(const char *p, off_t *length)
513 {
514 off_t len;
515
516 for (len = 0; *p && isdigit((unsigned char)*p); ++p)
517 len = len * 10 + (*p - '0');
518 if (*p)
519 return (-1);
520 *length = len;
521 return (0);
522 }
523
524 /*
525 * Parse a content-range header
526 */
527 static int
528 http_parse_range(const char *p, off_t *offset, off_t *length, off_t *size)
529 {
530 off_t first, last, len;
531
532 if (strncasecmp(p, "bytes ", 6) != 0)
533 return (-1);
534 p += 6;
535 if (*p == '*') {
536 first = last = -1;
537 ++p;
538 } else {
539 for (first = 0; *p && isdigit((unsigned char)*p); ++p)
540 first = first * 10 + *p - '0';
541 if (*p != '-')
542 return (-1);
543 for (last = 0, ++p; *p && isdigit((unsigned char)*p); ++p)
544 last = last * 10 + *p - '0';
545 }
546 if (first > last || *p != '/')
547 return (-1);
548 for (len = 0, ++p; *p && isdigit((unsigned char)*p); ++p)
549 len = len * 10 + *p - '0';
550 if (*p || len < last - first + 1)
551 return (-1);
552 if (first == -1)
553 *length = 0;
554 else
555 *length = last - first + 1;
556 *offset = first;
557 *size = len;
558 return (0);
559 }
560
561
562 /*****************************************************************************
563 * Helper functions for authorization
564 */
565
566 /*
567 * Base64 encoding
568 */
569 static char *
570 http_base64(const char *src)
571 {
572 static const char base64[] =
573 "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
574 "abcdefghijklmnopqrstuvwxyz"
575 "0123456789+/";
576 char *str, *dst;
577 size_t l;
578 int t, r;
579
580 l = strlen(src);
581 if ((str = malloc(((l + 2) / 3) * 4 + 1)) == NULL)
582 return (NULL);
583 dst = str;
584 r = 0;
585
586 while (l >= 3) {
587 t = (src[0] << 16) | (src[1] << 8) | src[2];
588 dst[0] = base64[(t >> 18) & 0x3f];
589 dst[1] = base64[(t >> 12) & 0x3f];
590 dst[2] = base64[(t >> 6) & 0x3f];
591 dst[3] = base64[(t >> 0) & 0x3f];
592 src += 3; l -= 3;
593 dst += 4; r += 4;
594 }
595
596 switch (l) {
597 case 2:
598 t = (src[0] << 16) | (src[1] << 8);
599 dst[0] = base64[(t >> 18) & 0x3f];
600 dst[1] = base64[(t >> 12) & 0x3f];
601 dst[2] = base64[(t >> 6) & 0x3f];
602 dst[3] = '=';
603 dst += 4;
604 r += 4;
605 break;
606 case 1:
607 t = src[0] << 16;
608 dst[0] = base64[(t >> 18) & 0x3f];
609 dst[1] = base64[(t >> 12) & 0x3f];
610 dst[2] = dst[3] = '=';
611 dst += 4;
612 r += 4;
613 break;
614 case 0:
615 break;
616 }
617
618 *dst = 0;
619 return (str);
620 }
621
622 /*
623 * Encode username and password
624 */
625 static int
626 http_basic_auth(conn_t *conn, const char *hdr, const char *usr, const char *pwd)
627 {
628 char *upw, *auth;
629 int r;
630
631 if (asprintf(&upw, "%s:%s", usr, pwd) == -1)
632 return (-1);
633 auth = http_base64(upw);
634 free(upw);
635 if (auth == NULL)
636 return (-1);
637 r = http_cmd(conn, "%s: Basic %s", hdr, auth);
638 free(auth);
639 return (r);
640 }
641
642 /*
643 * Send an authorization header
644 */
645 static int
646 http_authorize(conn_t *conn, const char *hdr, const char *p)
647 {
648 /* basic authorization */
649 if (strncasecmp(p, "basic:", 6) == 0) {
650 char *user, *pwd, *str;
651 int r;
652
653 /* skip realm */
654 for (p += 6; *p && *p != ':'; ++p)
655 /* nothing */ ;
656 if (!*p || strchr(++p, ':') == NULL)
657 return (-1);
658 if ((str = strdup(p)) == NULL)
659 return (-1); /* XXX */
660 user = str;
661 pwd = strchr(str, ':');
662 *pwd++ = '\0';
663 r = http_basic_auth(conn, hdr, user, pwd);
664 free(str);
665 return (r);
666 }
667 return (-1);
668 }
669
670
671 /*****************************************************************************
672 * Helper functions for connecting to a server or proxy
673 */
674
675 /*
676 * Connect to the correct HTTP server or proxy.
677 */
678 static conn_t *
679 http_connect(struct url *URL, struct url *purl, const char *flags)
680 {
681 conn_t *conn;
682 int af, verbose;
683 #ifdef TCP_NOPUSH
684 int val;
685 #endif
686
687 #ifdef INET6
688 af = AF_UNSPEC;
689 #else
690 af = AF_INET;
691 #endif
692
693 verbose = CHECK_FLAG('v');
694 if (CHECK_FLAG('4'))
695 af = AF_INET;
696 #ifdef INET6
697 else if (CHECK_FLAG('6'))
698 af = AF_INET6;
699 #endif
700
701 if (purl && strcasecmp(URL->scheme, SCHEME_HTTPS) != 0) {
702 URL = purl;
703 } else if (strcasecmp(URL->scheme, SCHEME_FTP) == 0) {
704 /* can't talk http to an ftp server */
705 /* XXX should set an error code */
706 return (NULL);
707 }
708
709 if ((conn = fetch_connect(URL->host, URL->port, af, verbose)) == NULL)
710 /* fetch_connect() has already set an error code */
711 return (NULL);
712 if (strcasecmp(URL->scheme, SCHEME_HTTPS) == 0 &&
713 fetch_ssl(conn, verbose) == -1) {
714 fetch_close(conn);
715 /* grrr */
716 #ifdef EAUTH
717 errno = EAUTH;
718 #else
719 errno = EPERM;
720 #endif
721 fetch_syserr();
722 return (NULL);
723 }
724
725 #ifdef TCP_NOPUSH
726 val = 1;
727 setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val, sizeof(val));
728 #endif
729
730 return (conn);
731 }
732
733 static struct url *
734 http_get_proxy(struct url * url, const char *flags)
735 {
736 struct url *purl;
737 char *p;
738
739 if (flags != NULL && strchr(flags, 'd') != NULL)
740 return (NULL);
741 if (fetch_no_proxy_match(url->host))
742 return (NULL);
743 if (((p = getenv("HTTP_PROXY")) || (p = getenv("http_proxy"))) &&
744 *p && (purl = fetchParseURL(p))) {
745 if (!*purl->scheme)
746 strcpy(purl->scheme, SCHEME_HTTP);
747 if (!purl->port)
748 purl->port = fetch_default_proxy_port(purl->scheme);
749 if (strcasecmp(purl->scheme, SCHEME_HTTP) == 0)
750 return (purl);
751 fetchFreeURL(purl);
752 }
753 return (NULL);
754 }
755
756 static void
757 set_if_modified_since(conn_t *conn, time_t last_modified)
758 {
759 static const char weekdays[] = "SunMonTueWedThuFriSat";
760 static const char months[] = "JanFebMarAprMayJunJulAugSepOctNovDec";
761 struct tm tm;
762 char buf[80];
763 gmtime_r(&last_modified, &tm);
764 snprintf(buf, sizeof(buf), "%.3s, %02d %.3s %4d %02d:%02d:%02d GMT",
765 weekdays + tm.tm_wday * 3, tm.tm_mday, months + tm.tm_mon * 3,
766 tm.tm_year + 1900, tm.tm_hour, tm.tm_min, tm.tm_sec);
767 http_cmd(conn, "If-Modified-Since: %s", buf);
768 }
769
770
771 /*****************************************************************************
772 * Core
773 */
774
775 /*
776 * Send a request and process the reply
777 *
778 * XXX This function is way too long, the do..while loop should be split
779 * XXX off into a separate function.
780 */
781 fetchIO *
782 http_request(struct url *URL, const char *op, struct url_stat *us,
783 struct url *purl, const char *flags)
784 {
785 conn_t *conn;
786 struct url *url, *new;
787 int chunked, direct, if_modified_since, need_auth, noredirect, verbose;
788 int e, i, n, val;
789 off_t offset, clength, length, size;
790 time_t mtime;
791 const char *p;
792 fetchIO *f;
793 hdr_t h;
794 char hbuf[URL_HOSTLEN + 7], *host;
795
796 direct = CHECK_FLAG('d');
797 noredirect = CHECK_FLAG('A');
798 verbose = CHECK_FLAG('v');
799 if_modified_since = CHECK_FLAG('i');
800
801 if (direct && purl) {
802 fetchFreeURL(purl);
803 purl = NULL;
804 }
805
806 /* try the provided URL first */
807 url = URL;
808
809 /* if the A flag is set, we only get one try */
810 n = noredirect ? 1 : MAX_REDIRECT;
811 i = 0;
812
813 e = HTTP_PROTOCOL_ERROR;
814 need_auth = 0;
815 do {
816 new = NULL;
817 chunked = 0;
818 offset = 0;
819 clength = -1;
820 length = -1;
821 size = -1;
822 mtime = 0;
823
824 /* check port */
825 if (!url->port)
826 url->port = fetch_default_port(url->scheme);
827
828 /* were we redirected to an FTP URL? */
829 if (purl == NULL && strcmp(url->scheme, SCHEME_FTP) == 0) {
830 if (strcmp(op, "GET") == 0)
831 return (ftp_request(url, "RETR", NULL, us, purl, flags));
832 else if (strcmp(op, "HEAD") == 0)
833 return (ftp_request(url, "STAT", NULL, us, purl, flags));
834 }
835
836 /* connect to server or proxy */
837 if ((conn = http_connect(url, purl, flags)) == NULL)
838 goto ouch;
839
840 host = url->host;
841 #ifdef INET6
842 if (strchr(url->host, ':')) {
843 snprintf(hbuf, sizeof(hbuf), "[%s]", url->host);
844 host = hbuf;
845 }
846 #endif
847 if (url->port != fetch_default_port(url->scheme)) {
848 if (host != hbuf) {
849 strcpy(hbuf, host);
850 host = hbuf;
851 }
852 snprintf(hbuf + strlen(hbuf),
853 sizeof(hbuf) - strlen(hbuf), ":%d", url->port);
854 }
855
856 /* send request */
857 if (verbose)
858 fetch_info("requesting %s://%s%s",
859 url->scheme, host, url->doc);
860 if (purl) {
861 http_cmd(conn, "%s %s://%s%s HTTP/1.1",
862 op, url->scheme, host, url->doc);
863 } else {
864 http_cmd(conn, "%s %s HTTP/1.1",
865 op, url->doc);
866 }
867
868 if (if_modified_since && url->last_modified > 0)
869 set_if_modified_since(conn, url->last_modified);
870
871 /* virtual host */
872 http_cmd(conn, "Host: %s", host);
873
874 /* proxy authorization */
875 if (purl) {
876 if (*purl->user || *purl->pwd)
877 http_basic_auth(conn, "Proxy-Authorization",
878 purl->user, purl->pwd);
879 else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL && *p != '\0')
880 http_authorize(conn, "Proxy-Authorization", p);
881 }
882
883 /* server authorization */
884 if (need_auth || *url->user || *url->pwd) {
885 if (*url->user || *url->pwd)
886 http_basic_auth(conn, "Authorization", url->user, url->pwd);
887 else if ((p = getenv("HTTP_AUTH")) != NULL && *p != '\0')
888 http_authorize(conn, "Authorization", p);
889 else if (fetchAuthMethod && fetchAuthMethod(url) == 0) {
890 http_basic_auth(conn, "Authorization", url->user, url->pwd);
891 } else {
892 http_seterr(HTTP_NEED_AUTH);
893 goto ouch;
894 }
895 }
896
897 /* other headers */
898 if ((p = getenv("HTTP_REFERER")) != NULL && *p != '\0') {
899 if (strcasecmp(p, "auto") == 0)
900 http_cmd(conn, "Referer: %s://%s%s",
901 url->scheme, host, url->doc);
902 else
903 http_cmd(conn, "Referer: %s", p);
904 }
905 if ((p = getenv("HTTP_USER_AGENT")) != NULL && *p != '\0')
906 http_cmd(conn, "User-Agent: %s", p);
907 else
908 http_cmd(conn, "User-Agent: %s ", _LIBFETCH_VER);
909 if (url->offset > 0)
910 http_cmd(conn, "Range: bytes=%lld-", (long long)url->offset);
911 http_cmd(conn, "Connection: close");
912 http_cmd(conn, "");
913
914 /*
915 * Force the queued request to be dispatched. Normally, one
916 * would do this with shutdown(2) but squid proxies can be
917 * configured to disallow such half-closed connections. To
918 * be compatible with such configurations, fiddle with socket
919 * options to force the pending data to be written.
920 */
921 #ifdef TCP_NOPUSH
922 val = 0;
923 setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val,
924 sizeof(val));
925 #endif
926 val = 1;
927 setsockopt(conn->sd, IPPROTO_TCP, TCP_NODELAY, &val,
928 sizeof(val));
929
930 /* get reply */
931 switch (http_get_reply(conn)) {
932 case HTTP_OK:
933 case HTTP_PARTIAL:
934 case HTTP_NOT_MODIFIED:
935 /* fine */
936 break;
937 case HTTP_MOVED_PERM:
938 case HTTP_MOVED_TEMP:
939 case HTTP_SEE_OTHER:
940 /*
941 * Not so fine, but we still have to read the
942 * headers to get the new location.
943 */
944 break;
945 case HTTP_NEED_AUTH:
946 if (need_auth) {
947 /*
948 * We already sent out authorization code,
949 * so there's nothing more we can do.
950 */
951 http_seterr(conn->err);
952 goto ouch;
953 }
954 /* try again, but send the password this time */
955 if (verbose)
956 fetch_info("server requires authorization");
957 break;
958 case HTTP_NEED_PROXY_AUTH:
959 /*
960 * If we're talking to a proxy, we already sent
961 * our proxy authorization code, so there's
962 * nothing more we can do.
963 */
964 http_seterr(conn->err);
965 goto ouch;
966 case HTTP_BAD_RANGE:
967 /*
968 * This can happen if we ask for 0 bytes because
969 * we already have the whole file. Consider this
970 * a success for now, and check sizes later.
971 */
972 break;
973 case HTTP_PROTOCOL_ERROR:
974 /* fall through */
975 case -1:
976 fetch_syserr();
977 goto ouch;
978 default:
979 http_seterr(conn->err);
980 if (!verbose)
981 goto ouch;
982 /* fall through so we can get the full error message */
983 }
984
985 /* get headers */
986 do {
987 switch ((h = http_next_header(conn, &p))) {
988 case hdr_syserror:
989 fetch_syserr();
990 goto ouch;
991 case hdr_error:
992 http_seterr(HTTP_PROTOCOL_ERROR);
993 goto ouch;
994 case hdr_content_length:
995 http_parse_length(p, &clength);
996 break;
997 case hdr_content_range:
998 http_parse_range(p, &offset, &length, &size);
999 break;
1000 case hdr_last_modified:
1001 http_parse_mtime(p, &mtime);
1002 break;
1003 case hdr_location:
1004 if (!HTTP_REDIRECT(conn->err))
1005 break;
1006 if (new)
1007 free(new);
1008 if (verbose)
1009 fetch_info("%d redirect to %s", conn->err, p);
1010 if (*p == '/')
1011 /* absolute path */
1012 new = fetchMakeURL(url->scheme, url->host, url->port, p,
1013 url->user, url->pwd);
1014 else
1015 new = fetchParseURL(p);
1016 if (new == NULL) {
1017 /* XXX should set an error code */
1018 goto ouch;
1019 }
1020 if (!*new->user && !*new->pwd) {
1021 strcpy(new->user, url->user);
1022 strcpy(new->pwd, url->pwd);
1023 }
1024 new->offset = url->offset;
1025 new->length = url->length;
1026 break;
1027 case hdr_transfer_encoding:
1028 /* XXX weak test*/
1029 chunked = (strcasecmp(p, "chunked") == 0);
1030 break;
1031 case hdr_www_authenticate:
1032 if (conn->err != HTTP_NEED_AUTH)
1033 break;
1034 /* if we were smarter, we'd check the method and realm */
1035 break;
1036 case hdr_end:
1037 /* fall through */
1038 case hdr_unknown:
1039 /* ignore */
1040 break;
1041 }
1042 } while (h > hdr_end);
1043
1044 /* we need to provide authentication */
1045 if (conn->err == HTTP_NEED_AUTH) {
1046 e = conn->err;
1047 need_auth = 1;
1048 fetch_close(conn);
1049 conn = NULL;
1050 continue;
1051 }
1052
1053 /* requested range not satisfiable */
1054 if (conn->err == HTTP_BAD_RANGE) {
1055 if (url->offset == size && url->length == 0) {
1056 /* asked for 0 bytes; fake it */
1057 offset = url->offset;
1058 conn->err = HTTP_OK;
1059 break;
1060 } else {
1061 http_seterr(conn->err);
1062 goto ouch;
1063 }
1064 }
1065
1066 /* we have a hit or an error */
1067 if (conn->err == HTTP_OK ||
1068 conn->err == HTTP_PARTIAL ||
1069 conn->err == HTTP_NOT_MODIFIED ||
1070 HTTP_ERROR(conn->err))
1071 break;
1072
1073 /* all other cases: we got a redirect */
1074 e = conn->err;
1075 need_auth = 0;
1076 fetch_close(conn);
1077 conn = NULL;
1078 if (!new)
1079 break;
1080 if (url != URL)
1081 fetchFreeURL(url);
1082 url = new;
1083 } while (++i < n);
1084
1085 /* we failed, or ran out of retries */
1086 if (conn == NULL) {
1087 http_seterr(e);
1088 goto ouch;
1089 }
1090
1091 /* check for inconsistencies */
1092 if (clength != -1 && length != -1 && clength != length) {
1093 http_seterr(HTTP_PROTOCOL_ERROR);
1094 goto ouch;
1095 }
1096 if (clength == -1)
1097 clength = length;
1098 if (clength != -1)
1099 length = offset + clength;
1100 if (length != -1 && size != -1 && length != size) {
1101 http_seterr(HTTP_PROTOCOL_ERROR);
1102 goto ouch;
1103 }
1104 if (size == -1)
1105 size = length;
1106
1107 /* fill in stats */
1108 if (us) {
1109 us->size = size;
1110 us->atime = us->mtime = mtime;
1111 }
1112
1113 /* too far? */
1114 if (URL->offset > 0 && offset > URL->offset) {
1115 http_seterr(HTTP_PROTOCOL_ERROR);
1116 goto ouch;
1117 }
1118
1119 /* report back real offset and size */
1120 URL->offset = offset;
1121 URL->length = clength;
1122
1123 if (conn->err == HTTP_NOT_MODIFIED) {
1124 http_seterr(HTTP_NOT_MODIFIED);
1125 goto ouch;
1126 }
1127
1128 /* wrap it up in a fetchIO */
1129 if ((f = http_funopen(conn, chunked)) == NULL) {
1130 fetch_syserr();
1131 goto ouch;
1132 }
1133
1134 if (url != URL)
1135 fetchFreeURL(url);
1136 if (purl)
1137 fetchFreeURL(purl);
1138
1139 if (HTTP_ERROR(conn->err)) {
1140 fetchIO_close(f);
1141 f = NULL;
1142 }
1143
1144 return (f);
1145
1146 ouch:
1147 if (url != URL)
1148 fetchFreeURL(url);
1149 if (purl)
1150 fetchFreeURL(purl);
1151 if (conn != NULL)
1152 fetch_close(conn);
1153 return (NULL);
1154 }
1155
1156
1157 /*****************************************************************************
1158 * Entry points
1159 */
1160
1161 /*
1162 * Retrieve and stat a file by HTTP
1163 */
1164 fetchIO *
1165 fetchXGetHTTP(struct url *URL, struct url_stat *us, const char *flags)
1166 {
1167 return (http_request(URL, "GET", us, http_get_proxy(URL, flags), flags));
1168 }
1169
1170 /*
1171 * Retrieve a file by HTTP
1172 */
1173 fetchIO *
1174 fetchGetHTTP(struct url *URL, const char *flags)
1175 {
1176 return (fetchXGetHTTP(URL, NULL, flags));
1177 }
1178
1179 /*
1180 * Store a file by HTTP
1181 */
1182 fetchIO *
1183 fetchPutHTTP(struct url *URL, const char *flags)
1184 {
1185 fprintf(stderr, "fetchPutHTTP(): not implemented\n");
1186 return (NULL);
1187 }
1188
1189 /*
1190 * Get an HTTP document's metadata
1191 */
1192 int
1193 fetchStatHTTP(struct url *URL, struct url_stat *us, const char *flags)
1194 {
1195 fetchIO *f;
1196
1197 f = http_request(URL, "HEAD", us, http_get_proxy(URL, flags), flags);
1198 if (f == NULL)
1199 return (-1);
1200 fetchIO_close(f);
1201 return (0);
1202 }
1203
1204 enum http_states {
1205 ST_NONE,
1206 ST_LT,
1207 ST_LTA,
1208 ST_TAGA,
1209 ST_H,
1210 ST_R,
1211 ST_E,
1212 ST_F,
1213 ST_HREF,
1214 ST_HREFQ,
1215 ST_TAG,
1216 ST_TAGAX,
1217 ST_TAGAQ
1218 };
1219
1220 struct index_parser {
1221 struct url_list *ue;
1222 struct url *url;
1223 enum http_states state;
1224 };
1225
1226 static ssize_t
1227 parse_index(struct index_parser *parser, const char *buf, size_t len)
1228 {
1229 char *end_attr, p = *buf;
1230
1231 switch (parser->state) {
1232 case ST_NONE:
1233 /* Plain text, not in markup */
1234 if (p == '<')
1235 parser->state = ST_LT;
1236 return 1;
1237 case ST_LT:
1238 /* In tag -- "<" already found */
1239 if (p == '>')
1240 parser->state = ST_NONE;
1241 else if (p == 'a' || p == 'A')
1242 parser->state = ST_LTA;
1243 else if (!isspace((unsigned char)p))
1244 parser->state = ST_TAG;
1245 return 1;
1246 case ST_LTA:
1247 /* In tag -- "<a" already found */
1248 if (p == '>')
1249 parser->state = ST_NONE;
1250 else if (p == '"')
1251 parser->state = ST_TAGAQ;
1252 else if (isspace((unsigned char)p))
1253 parser->state = ST_TAGA;
1254 else
1255 parser->state = ST_TAG;
1256 return 1;
1257 case ST_TAG:
1258 /* In tag, but not "<a" -- disregard */
1259 if (p == '>')
1260 parser->state = ST_NONE;
1261 return 1;
1262 case ST_TAGA:
1263 /* In a-tag -- "<a " already found */
1264 if (p == '>')
1265 parser->state = ST_NONE;
1266 else if (p == '"')
1267 parser->state = ST_TAGAQ;
1268 else if (p == 'h' || p == 'H')
1269 parser->state = ST_H;
1270 else if (!isspace((unsigned char)p))
1271 parser->state = ST_TAGAX;
1272 return 1;
1273 case ST_TAGAX:
1274 /* In unknown keyword in a-tag */
1275 if (p == '>')
1276 parser->state = ST_NONE;
1277 else if (p == '"')
1278 parser->state = ST_TAGAQ;
1279 else if (isspace((unsigned char)p))
1280 parser->state = ST_TAGA;
1281 return 1;
1282 case ST_TAGAQ:
1283 /* In a-tag, unknown argument for keys. */
1284 if (p == '>')
1285 parser->state = ST_NONE;
1286 else if (p == '"')
1287 parser->state = ST_TAGA;
1288 return 1;
1289 case ST_H:
1290 /* In a-tag -- "<a h" already found */
1291 if (p == '>')
1292 parser->state = ST_NONE;
1293 else if (p == '"')
1294 parser->state = ST_TAGAQ;
1295 else if (p == 'r' || p == 'R')
1296 parser->state = ST_R;
1297 else if (isspace((unsigned char)p))
1298 parser->state = ST_TAGA;
1299 else
1300 parser->state = ST_TAGAX;
1301 return 1;
1302 case ST_R:
1303 /* In a-tag -- "<a hr" already found */
1304 if (p == '>')
1305 parser->state = ST_NONE;
1306 else if (p == '"')
1307 parser->state = ST_TAGAQ;
1308 else if (p == 'e' || p == 'E')
1309 parser->state = ST_E;
1310 else if (isspace((unsigned char)p))
1311 parser->state = ST_TAGA;
1312 else
1313 parser->state = ST_TAGAX;
1314 return 1;
1315 case ST_E:
1316 /* In a-tag -- "<a hre" already found */
1317 if (p == '>')
1318 parser->state = ST_NONE;
1319 else if (p == '"')
1320 parser->state = ST_TAGAQ;
1321 else if (p == 'f' || p == 'F')
1322 parser->state = ST_F;
1323 else if (isspace((unsigned char)p))
1324 parser->state = ST_TAGA;
1325 else
1326 parser->state = ST_TAGAX;
1327 return 1;
1328 case ST_F:
1329 /* In a-tag -- "<a href" already found */
1330 if (p == '>')
1331 parser->state = ST_NONE;
1332 else if (p == '"')
1333 parser->state = ST_TAGAQ;
1334 else if (p == '=')
1335 parser->state = ST_HREF;
1336 else if (!isspace((unsigned char)p))
1337 parser->state = ST_TAGAX;
1338 return 1;
1339 case ST_HREF:
1340 /* In a-tag -- "<a href=" already found */
1341 if (p == '>')
1342 parser->state = ST_NONE;
1343 else if (p == '"')
1344 parser->state = ST_HREFQ;
1345 else if (!isspace((unsigned char)p))
1346 parser->state = ST_TAGA;
1347 return 1;
1348 case ST_HREFQ:
1349 /* In href of the a-tag */
1350 end_attr = memchr(buf, '"', len);
1351 if (end_attr == NULL)
1352 return 0;
1353 *end_attr = '\0';
1354 parser->state = ST_TAGA;
1355 if (fetch_add_entry(parser->ue, parser->url, buf, 1))
1356 return -1;
1357 return end_attr + 1 - buf;
1358 }
1359 abort();
1360 }
1361
1362 struct http_index_cache {
1363 struct http_index_cache *next;
1364 struct url *location;
1365 struct url_list ue;
1366 };
1367
1368 static struct http_index_cache *index_cache;
1369
1370 /*
1371 * List a directory
1372 */
1373 int
1374 fetchListHTTP(struct url_list *ue, struct url *url, const char *pattern, const char *flags)
1375 {
1376 fetchIO *f;
1377 char buf[2 * PATH_MAX];
1378 size_t buf_len, sum_processed;
1379 ssize_t read_len, processed;
1380 struct index_parser state;
1381 struct http_index_cache *cache = NULL;
1382 int do_cache, ret;
1383
1384 do_cache = CHECK_FLAG('c');
1385
1386 if (do_cache) {
1387 for (cache = index_cache; cache != NULL; cache = cache->next) {
1388 if (strcmp(cache->location->scheme, url->scheme))
1389 continue;
1390 if (strcmp(cache->location->user, url->user))
1391 continue;
1392 if (strcmp(cache->location->pwd, url->pwd))
1393 continue;
1394 if (strcmp(cache->location->host, url->host))
1395 continue;
1396 if (cache->location->port != url->port)
1397 continue;
1398 if (strcmp(cache->location->doc, url->doc))
1399 continue;
1400 return fetchAppendURLList(ue, &cache->ue);
1401 }
1402
1403 cache = malloc(sizeof(*cache));
1404 fetchInitURLList(&cache->ue);
1405 cache->location = fetchCopyURL(url);
1406 }
1407
1408 f = fetchGetHTTP(url, flags);
1409 if (f == NULL) {
1410 if (do_cache) {
1411 fetchFreeURLList(&cache->ue);
1412 fetchFreeURL(cache->location);
1413 free(cache);
1414 }
1415 return -1;
1416 }
1417
1418 state.url = url;
1419 state.state = ST_NONE;
1420 if (do_cache) {
1421 state.ue = &cache->ue;
1422 } else {
1423 state.ue = ue;
1424 }
1425
1426 buf_len = 0;
1427
1428 while ((read_len = fetchIO_read(f, buf + buf_len, sizeof(buf) - buf_len)) > 0) {
1429 buf_len += read_len;
1430 sum_processed = 0;
1431 do {
1432 processed = parse_index(&state, buf + sum_processed, buf_len);
1433 if (processed == -1)
1434 break;
1435 buf_len -= processed;
1436 sum_processed += processed;
1437 } while (processed != 0 && buf_len > 0);
1438 if (processed == -1) {
1439 read_len = -1;
1440 break;
1441 }
1442 memmove(buf, buf + sum_processed, buf_len);
1443 }
1444
1445 fetchIO_close(f);
1446
1447 ret = read_len < 0 ? -1 : 0;
1448
1449 if (do_cache) {
1450 if (ret == 0) {
1451 cache->next = index_cache;
1452 index_cache = cache;
1453 }
1454
1455 if (fetchAppendURLList(ue, &cache->ue))
1456 ret = -1;
1457 }
1458
1459 return ret;
1460 }
NetBSD on the FriendlyARM MINI2440