xref: /freebsd/lib/libfetch/http.c (revision e6182307e80c9b90c02bff8161456abbb6e46d01)
14ca1ab94SDag-Erling Smørgrav /*-
24ca1ab94SDag-Erling Smørgrav  * Copyright (c) 1998 Dag-Erling Co�dan Sm�rgrav
34ca1ab94SDag-Erling Smørgrav  * All rights reserved.
44ca1ab94SDag-Erling Smørgrav  *
54ca1ab94SDag-Erling Smørgrav  * Redistribution and use in source and binary forms, with or without
64ca1ab94SDag-Erling Smørgrav  * modification, are permitted provided that the following conditions
74ca1ab94SDag-Erling Smørgrav  * are met:
84ca1ab94SDag-Erling Smørgrav  * 1. Redistributions of source code must retain the above copyright
94ca1ab94SDag-Erling Smørgrav  *    notice, this list of conditions and the following disclaimer
104ca1ab94SDag-Erling Smørgrav  *    in this position and unchanged.
114ca1ab94SDag-Erling Smørgrav  * 2. Redistributions in binary form must reproduce the above copyright
124ca1ab94SDag-Erling Smørgrav  *    notice, this list of conditions and the following disclaimer in the
134ca1ab94SDag-Erling Smørgrav  *    documentation and/or other materials provided with the distribution.
144ca1ab94SDag-Erling Smørgrav  * 3. The name of the author may not be used to endorse or promote products
154ca1ab94SDag-Erling Smørgrav  *    derived from this software without specific prior written permission
164ca1ab94SDag-Erling Smørgrav  *
174ca1ab94SDag-Erling Smørgrav  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
184ca1ab94SDag-Erling Smørgrav  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
194ca1ab94SDag-Erling Smørgrav  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
204ca1ab94SDag-Erling Smørgrav  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
214ca1ab94SDag-Erling Smørgrav  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
224ca1ab94SDag-Erling Smørgrav  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
234ca1ab94SDag-Erling Smørgrav  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
244ca1ab94SDag-Erling Smørgrav  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
254ca1ab94SDag-Erling Smørgrav  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
264ca1ab94SDag-Erling Smørgrav  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
274ca1ab94SDag-Erling Smørgrav  *
287f3dea24SPeter Wemm  * $FreeBSD$
294ca1ab94SDag-Erling Smørgrav  */
304ca1ab94SDag-Erling Smørgrav 
31f62e5228SDag-Erling Smørgrav /*
32f62e5228SDag-Erling Smørgrav  * The base64 code in this file is based on code from MIT fetch, which
33f62e5228SDag-Erling Smørgrav  * has the following copyright and license:
34f62e5228SDag-Erling Smørgrav  *
35f62e5228SDag-Erling Smørgrav  *-
36f62e5228SDag-Erling Smørgrav  * Copyright 1997 Massachusetts Institute of Technology
37f62e5228SDag-Erling Smørgrav  *
38f62e5228SDag-Erling Smørgrav  * Permission to use, copy, modify, and distribute this software and
39f62e5228SDag-Erling Smørgrav  * its documentation for any purpose and without fee is hereby
40f62e5228SDag-Erling Smørgrav  * granted, provided that both the above copyright notice and this
41f62e5228SDag-Erling Smørgrav  * permission notice appear in all copies, that both the above
42f62e5228SDag-Erling Smørgrav  * copyright notice and this permission notice appear in all
43f62e5228SDag-Erling Smørgrav  * supporting documentation, and that the name of M.I.T. not be used
44f62e5228SDag-Erling Smørgrav  * in advertising or publicity pertaining to distribution of the
45f62e5228SDag-Erling Smørgrav  * software without specific, written prior permission.	 M.I.T. makes
46f62e5228SDag-Erling Smørgrav  * no representations about the suitability of this software for any
47f62e5228SDag-Erling Smørgrav  * purpose.  It is provided "as is" without express or implied
48f62e5228SDag-Erling Smørgrav  * warranty.
49f62e5228SDag-Erling Smørgrav  *
50f62e5228SDag-Erling Smørgrav  * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''.  M.I.T. DISCLAIMS
51f62e5228SDag-Erling Smørgrav  * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE,
52f62e5228SDag-Erling Smørgrav  * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
53f62e5228SDag-Erling Smørgrav  * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT
54f62e5228SDag-Erling Smørgrav  * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
55f62e5228SDag-Erling Smørgrav  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
56f62e5228SDag-Erling Smørgrav  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
57f62e5228SDag-Erling Smørgrav  * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
58f62e5228SDag-Erling Smørgrav  * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
59f62e5228SDag-Erling Smørgrav  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
60f62e5228SDag-Erling Smørgrav  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
61f62e5228SDag-Erling Smørgrav  * SUCH DAMAGE. */
62f62e5228SDag-Erling Smørgrav 
634ca1ab94SDag-Erling Smørgrav #include <sys/param.h>
644ca1ab94SDag-Erling Smørgrav 
654ca1ab94SDag-Erling Smørgrav #include <err.h>
664ca1ab94SDag-Erling Smørgrav #include <ctype.h>
67e6182307SDag-Erling Smørgrav #include <netdb.h>
68f62e5228SDag-Erling Smørgrav #include <stdarg.h>
694ca1ab94SDag-Erling Smørgrav #include <stdio.h>
704ca1ab94SDag-Erling Smørgrav #include <stdlib.h>
714ca1ab94SDag-Erling Smørgrav #include <string.h>
724ca1ab94SDag-Erling Smørgrav #include <unistd.h>
734ca1ab94SDag-Erling Smørgrav 
744ca1ab94SDag-Erling Smørgrav #include "fetch.h"
75842a95ccSDag-Erling Smørgrav #include "common.h"
760fba3a00SDag-Erling Smørgrav #include "httperr.h"
774ca1ab94SDag-Erling Smørgrav 
784ca1ab94SDag-Erling Smørgrav extern char *__progname;
794ca1ab94SDag-Erling Smørgrav 
804ca1ab94SDag-Erling Smørgrav #define ENDL "\r\n"
814ca1ab94SDag-Erling Smørgrav 
824ca1ab94SDag-Erling Smørgrav struct cookie
834ca1ab94SDag-Erling Smørgrav {
844ca1ab94SDag-Erling Smørgrav     FILE *real_f;
854ca1ab94SDag-Erling Smørgrav #define ENC_NONE 0
864ca1ab94SDag-Erling Smørgrav #define ENC_CHUNKED 1
874ca1ab94SDag-Erling Smørgrav     int encoding;			/* 1 = chunked, 0 = none */
884ca1ab94SDag-Erling Smørgrav #define HTTPCTYPELEN 59
894ca1ab94SDag-Erling Smørgrav     char content_type[HTTPCTYPELEN+1];
904ca1ab94SDag-Erling Smørgrav     char *buf;
914ca1ab94SDag-Erling Smørgrav     int b_cur, eof;
924ca1ab94SDag-Erling Smørgrav     unsigned b_len, chunksize;
934ca1ab94SDag-Erling Smørgrav };
944ca1ab94SDag-Erling Smørgrav 
95f62e5228SDag-Erling Smørgrav /*
96f62e5228SDag-Erling Smørgrav  * Send a formatted line; optionally echo to terminal
97f62e5228SDag-Erling Smørgrav  */
98f62e5228SDag-Erling Smørgrav static int
99f62e5228SDag-Erling Smørgrav _http_cmd(FILE *f, char *fmt, ...)
100f62e5228SDag-Erling Smørgrav {
101f62e5228SDag-Erling Smørgrav     va_list ap;
102f62e5228SDag-Erling Smørgrav 
103f62e5228SDag-Erling Smørgrav     va_start(ap, fmt);
104f62e5228SDag-Erling Smørgrav     vfprintf(f, fmt, ap);
105f62e5228SDag-Erling Smørgrav #ifndef NDEBUG
106f62e5228SDag-Erling Smørgrav     fprintf(stderr, "\033[1m>>> ");
107f62e5228SDag-Erling Smørgrav     vfprintf(stderr, fmt, ap);
108f62e5228SDag-Erling Smørgrav     fprintf(stderr, "\033[m");
109f62e5228SDag-Erling Smørgrav #endif
110f62e5228SDag-Erling Smørgrav     va_end(ap);
111f62e5228SDag-Erling Smørgrav 
112f62e5228SDag-Erling Smørgrav     return 0; /* XXX */
113f62e5228SDag-Erling Smørgrav }
114f62e5228SDag-Erling Smørgrav 
115f62e5228SDag-Erling Smørgrav /*
116f62e5228SDag-Erling Smørgrav  * Fill the input buffer, do chunk decoding on the fly
117f62e5228SDag-Erling Smørgrav  */
1184ca1ab94SDag-Erling Smørgrav static char *
1194ca1ab94SDag-Erling Smørgrav _http_fillbuf(struct cookie *c)
1204ca1ab94SDag-Erling Smørgrav {
1214ca1ab94SDag-Erling Smørgrav     char *ln;
1224ca1ab94SDag-Erling Smørgrav     unsigned int len;
1234ca1ab94SDag-Erling Smørgrav 
1244ca1ab94SDag-Erling Smørgrav     if (c->eof)
1254ca1ab94SDag-Erling Smørgrav 	return NULL;
1264ca1ab94SDag-Erling Smørgrav 
1274ca1ab94SDag-Erling Smørgrav     if (c->encoding == ENC_NONE) {
1284ca1ab94SDag-Erling Smørgrav 	c->buf = fgetln(c->real_f, &(c->b_len));
1294ca1ab94SDag-Erling Smørgrav 	c->b_cur = 0;
1304ca1ab94SDag-Erling Smørgrav     } else if (c->encoding == ENC_CHUNKED) {
1314ca1ab94SDag-Erling Smørgrav 	if (c->chunksize == 0) {
1324ca1ab94SDag-Erling Smørgrav 	    ln = fgetln(c->real_f, &len);
1334ca1ab94SDag-Erling Smørgrav 	    DEBUG(fprintf(stderr, "\033[1m_http_fillbuf(): new chunk: "
1344ca1ab94SDag-Erling Smørgrav 			  "%*.*s\033[m\n", (int)len-2, (int)len-2, ln));
1354ca1ab94SDag-Erling Smørgrav 	    sscanf(ln, "%x", &(c->chunksize));
1364ca1ab94SDag-Erling Smørgrav 	    if (!c->chunksize) {
1374ca1ab94SDag-Erling Smørgrav 		DEBUG(fprintf(stderr, "\033[1m_http_fillbuf(): "
1384ca1ab94SDag-Erling Smørgrav 			      "end of last chunk\033[m\n"));
1394ca1ab94SDag-Erling Smørgrav 		c->eof = 1;
1404ca1ab94SDag-Erling Smørgrav 		return NULL;
1414ca1ab94SDag-Erling Smørgrav 	    }
1424ca1ab94SDag-Erling Smørgrav 	    DEBUG(fprintf(stderr, "\033[1m_http_fillbuf(): "
1434ca1ab94SDag-Erling Smørgrav 			  "new chunk: %X\033[m\n", c->chunksize));
1444ca1ab94SDag-Erling Smørgrav 	}
1454ca1ab94SDag-Erling Smørgrav 	c->buf = fgetln(c->real_f, &(c->b_len));
1464ca1ab94SDag-Erling Smørgrav 	if (c->b_len > c->chunksize)
1474ca1ab94SDag-Erling Smørgrav 	    c->b_len = c->chunksize;
1484ca1ab94SDag-Erling Smørgrav 	c->chunksize -= c->b_len;
1494ca1ab94SDag-Erling Smørgrav 	c->b_cur = 0;
1504ca1ab94SDag-Erling Smørgrav     }
1514ca1ab94SDag-Erling Smørgrav     else return NULL; /* unknown encoding */
1524ca1ab94SDag-Erling Smørgrav     return c->buf;
1534ca1ab94SDag-Erling Smørgrav }
1544ca1ab94SDag-Erling Smørgrav 
155f62e5228SDag-Erling Smørgrav /*
156f62e5228SDag-Erling Smørgrav  * Read function
157f62e5228SDag-Erling Smørgrav  */
1584ca1ab94SDag-Erling Smørgrav static int
1594ca1ab94SDag-Erling Smørgrav _http_readfn(struct cookie *c, char *buf, int len)
1604ca1ab94SDag-Erling Smørgrav {
1614ca1ab94SDag-Erling Smørgrav     int l, pos = 0;
1624ca1ab94SDag-Erling Smørgrav     while (len) {
1634ca1ab94SDag-Erling Smørgrav 	/* empty buffer */
1644ca1ab94SDag-Erling Smørgrav 	if (!c->buf || (c->b_cur == c->b_len))
1654ca1ab94SDag-Erling Smørgrav 	    if (!_http_fillbuf(c))
1664ca1ab94SDag-Erling Smørgrav 		break;
1674ca1ab94SDag-Erling Smørgrav 
1684ca1ab94SDag-Erling Smørgrav 	l = c->b_len - c->b_cur;
1694ca1ab94SDag-Erling Smørgrav 	if (len < l) l = len;
1704ca1ab94SDag-Erling Smørgrav 	memcpy(buf + pos, c->buf + c->b_cur, l);
1714ca1ab94SDag-Erling Smørgrav 	c->b_cur += l;
1724ca1ab94SDag-Erling Smørgrav 	pos += l;
1734ca1ab94SDag-Erling Smørgrav 	len -= l;
1744ca1ab94SDag-Erling Smørgrav     }
1754ca1ab94SDag-Erling Smørgrav 
1764ca1ab94SDag-Erling Smørgrav     if (ferror(c->real_f))
1774ca1ab94SDag-Erling Smørgrav 	return -1;
1784ca1ab94SDag-Erling Smørgrav     else return pos;
1794ca1ab94SDag-Erling Smørgrav }
1804ca1ab94SDag-Erling Smørgrav 
181f62e5228SDag-Erling Smørgrav /*
182f62e5228SDag-Erling Smørgrav  * Write function
183f62e5228SDag-Erling Smørgrav  */
1844ca1ab94SDag-Erling Smørgrav static int
1854ca1ab94SDag-Erling Smørgrav _http_writefn(struct cookie *c, const char *buf, int len)
1864ca1ab94SDag-Erling Smørgrav {
1874ca1ab94SDag-Erling Smørgrav     size_t r = fwrite(buf, 1, (size_t)len, c->real_f);
1884ca1ab94SDag-Erling Smørgrav     return r ? r : -1;
1894ca1ab94SDag-Erling Smørgrav }
1904ca1ab94SDag-Erling Smørgrav 
191f62e5228SDag-Erling Smørgrav /*
192f62e5228SDag-Erling Smørgrav  * Close function
193f62e5228SDag-Erling Smørgrav  */
1944ca1ab94SDag-Erling Smørgrav static int
1954ca1ab94SDag-Erling Smørgrav _http_closefn(struct cookie *c)
1964ca1ab94SDag-Erling Smørgrav {
1974ca1ab94SDag-Erling Smørgrav     int r = fclose(c->real_f);
1984ca1ab94SDag-Erling Smørgrav     free(c);
1994ca1ab94SDag-Erling Smørgrav     return (r == EOF) ? -1 : 0;
2004ca1ab94SDag-Erling Smørgrav }
2014ca1ab94SDag-Erling Smørgrav 
202f62e5228SDag-Erling Smørgrav /*
203f62e5228SDag-Erling Smørgrav  * Extract content type from cookie
204f62e5228SDag-Erling Smørgrav  */
2054ca1ab94SDag-Erling Smørgrav char *
2064ca1ab94SDag-Erling Smørgrav fetchContentType(FILE *f)
2074ca1ab94SDag-Erling Smørgrav {
2084ca1ab94SDag-Erling Smørgrav     /*
2094ca1ab94SDag-Erling Smørgrav      * We have no way of making sure this really *is* one of our cookies,
2104ca1ab94SDag-Erling Smørgrav      * so just check for a null pointer and hope for the best.
2114ca1ab94SDag-Erling Smørgrav      */
2124ca1ab94SDag-Erling Smørgrav     return f->_cookie ? (((struct cookie *)f->_cookie)->content_type) : NULL;
2134ca1ab94SDag-Erling Smørgrav }
2144ca1ab94SDag-Erling Smørgrav 
215f62e5228SDag-Erling Smørgrav /*
216f62e5228SDag-Erling Smørgrav  * Base64 encoding
217f62e5228SDag-Erling Smørgrav  */
218f62e5228SDag-Erling Smørgrav int
219f62e5228SDag-Erling Smørgrav _http_base64(char *dst, char *src, int l)
220f62e5228SDag-Erling Smørgrav {
221f62e5228SDag-Erling Smørgrav     static const char base64[] =
222f62e5228SDag-Erling Smørgrav 	"ABCDEFGHIJKLMNOPQRSTUVWXYZ"
223f62e5228SDag-Erling Smørgrav 	"abcdefghijklmnopqrstuvwxyz"
224f62e5228SDag-Erling Smørgrav 	"0123456789+/";
225f62e5228SDag-Erling Smørgrav     int t, r = 0;
226f62e5228SDag-Erling Smørgrav 
227f62e5228SDag-Erling Smørgrav     while (l >= 3) {
228f62e5228SDag-Erling Smørgrav 	t = (src[0] << 16) | (src[1] << 8) | src[2];
229f62e5228SDag-Erling Smørgrav 	dst[0] = base64[(t >> 18) & 0x3f];
230f62e5228SDag-Erling Smørgrav 	dst[1] = base64[(t >> 12) & 0x3f];
231f62e5228SDag-Erling Smørgrav 	dst[2] = base64[(t >> 6) & 0x3f];
232f62e5228SDag-Erling Smørgrav 	dst[3] = base64[(t >> 0) & 0x3f];
233f62e5228SDag-Erling Smørgrav 	src += 3; l -= 3;
234f62e5228SDag-Erling Smørgrav 	dst += 4; r += 4;
235f62e5228SDag-Erling Smørgrav     }
236f62e5228SDag-Erling Smørgrav 
237f62e5228SDag-Erling Smørgrav     switch (l) {
238f62e5228SDag-Erling Smørgrav     case 2:
239f62e5228SDag-Erling Smørgrav 	t = (src[0] << 16) | (src[1] << 8);
240f62e5228SDag-Erling Smørgrav 	dst[0] = base64[(t >> 18) & 0x3f];
241f62e5228SDag-Erling Smørgrav 	dst[1] = base64[(t >> 12) & 0x3f];
242f62e5228SDag-Erling Smørgrav 	dst[2] = base64[(t >> 6) & 0x3f];
243f62e5228SDag-Erling Smørgrav 	dst[3] = '=';
244f62e5228SDag-Erling Smørgrav 	dst += 4;
245f62e5228SDag-Erling Smørgrav 	r += 4;
246f62e5228SDag-Erling Smørgrav 	break;
247f62e5228SDag-Erling Smørgrav     case 1:
248f62e5228SDag-Erling Smørgrav 	t = src[0] << 16;
249f62e5228SDag-Erling Smørgrav 	dst[0] = base64[(t >> 18) & 0x3f];
250f62e5228SDag-Erling Smørgrav 	dst[1] = base64[(t >> 12) & 0x3f];
251f62e5228SDag-Erling Smørgrav 	dst[2] = dst[3] = '=';
252f62e5228SDag-Erling Smørgrav 	dst += 4;
253f62e5228SDag-Erling Smørgrav 	r += 4;
254f62e5228SDag-Erling Smørgrav 	break;
255f62e5228SDag-Erling Smørgrav     case 0:
256f62e5228SDag-Erling Smørgrav 	break;
257f62e5228SDag-Erling Smørgrav     }
258f62e5228SDag-Erling Smørgrav 
259f62e5228SDag-Erling Smørgrav     *dst = 0;
260f62e5228SDag-Erling Smørgrav     return r;
261f62e5228SDag-Erling Smørgrav }
262f62e5228SDag-Erling Smørgrav 
263f62e5228SDag-Erling Smørgrav /*
264f62e5228SDag-Erling Smørgrav  * Encode username and password
265f62e5228SDag-Erling Smørgrav  */
266f62e5228SDag-Erling Smørgrav char *
267f62e5228SDag-Erling Smørgrav _http_auth(char *usr, char *pwd)
268f62e5228SDag-Erling Smørgrav {
269f62e5228SDag-Erling Smørgrav     int len, lu, lp;
270f62e5228SDag-Erling Smørgrav     char *str, *s;
271f62e5228SDag-Erling Smørgrav 
272f62e5228SDag-Erling Smørgrav     lu = strlen(usr);
273f62e5228SDag-Erling Smørgrav     lp = strlen(pwd);
274f62e5228SDag-Erling Smørgrav 
275f62e5228SDag-Erling Smørgrav     len = (lu * 4 + 2) / 3	/* user name, round up */
276f62e5228SDag-Erling Smørgrav 	+ 1			/* colon */
277f62e5228SDag-Erling Smørgrav 	+ (lp * 4 + 2) / 3	/* password, round up */
278f62e5228SDag-Erling Smørgrav 	+ 1;			/* null */
279f62e5228SDag-Erling Smørgrav 
280f62e5228SDag-Erling Smørgrav     if ((s = str = (char *)malloc(len)) == NULL)
281f62e5228SDag-Erling Smørgrav 	return NULL;
282f62e5228SDag-Erling Smørgrav 
283f62e5228SDag-Erling Smørgrav     s += _http_base64(s, usr, lu);
284f62e5228SDag-Erling Smørgrav     *s++ = ':';
285f62e5228SDag-Erling Smørgrav     s += _http_base64(s, pwd, lp);
286f62e5228SDag-Erling Smørgrav     *s = 0;
287f62e5228SDag-Erling Smørgrav 
288f62e5228SDag-Erling Smørgrav     return str;
289f62e5228SDag-Erling Smørgrav }
290f62e5228SDag-Erling Smørgrav 
291f62e5228SDag-Erling Smørgrav /*
292d8acd8dcSDag-Erling Smørgrav  * Retrieve a file by HTTP
293f62e5228SDag-Erling Smørgrav  */
2944ca1ab94SDag-Erling Smørgrav FILE *
295d8acd8dcSDag-Erling Smørgrav fetchGetHTTP(struct url *URL, char *flags)
2964ca1ab94SDag-Erling Smørgrav {
297f5f109a0SDag-Erling Smørgrav     int sd = -1, e, i, enc = ENC_NONE, direct, verbose;
2984ca1ab94SDag-Erling Smørgrav     struct cookie *c;
2993b7a6740SDag-Erling Smørgrav     char *ln, *p, *px, *q;
3004ca1ab94SDag-Erling Smørgrav     FILE *f, *cf;
3014ca1ab94SDag-Erling Smørgrav     size_t len;
3024ca1ab94SDag-Erling Smørgrav 
303f5f109a0SDag-Erling Smørgrav     direct = (flags && strchr(flags, 'd'));
304f5f109a0SDag-Erling Smørgrav     verbose = (flags && strchr(flags, 'v'));
3050fba3a00SDag-Erling Smørgrav 
3064ca1ab94SDag-Erling Smørgrav     /* allocate cookie */
307e6182307SDag-Erling Smørgrav     if ((c = calloc(1, sizeof *c)) == NULL)
3084ca1ab94SDag-Erling Smørgrav 	return NULL;
3094ca1ab94SDag-Erling Smørgrav 
3104ca1ab94SDag-Erling Smørgrav     /* check port */
311e6182307SDag-Erling Smørgrav     if (!URL->port) {
312e6182307SDag-Erling Smørgrav 	struct servent *se;
313e6182307SDag-Erling Smørgrav 
314e6182307SDag-Erling Smørgrav 	if ((se = getservbyname("http", "tcp")) != NULL)
315e6182307SDag-Erling Smørgrav 	    URL->port = ntohs(se->s_port);
316e6182307SDag-Erling Smørgrav 	else
317e6182307SDag-Erling Smørgrav 	    URL->port = 80;
318e6182307SDag-Erling Smørgrav     }
3194ca1ab94SDag-Erling Smørgrav 
3204ca1ab94SDag-Erling Smørgrav     /* attempt to connect to proxy server */
321f5f109a0SDag-Erling Smørgrav     if (!direct && (px = getenv("HTTP_PROXY")) != NULL) {
3223b7a6740SDag-Erling Smørgrav 	char host[MAXHOSTNAMELEN];
323e6182307SDag-Erling Smørgrav 	int port = 0;
3244ca1ab94SDag-Erling Smørgrav 
3254ca1ab94SDag-Erling Smørgrav 	/* measure length */
3264ca1ab94SDag-Erling Smørgrav 	len = strcspn(px, ":");
3274ca1ab94SDag-Erling Smørgrav 
328f5f109a0SDag-Erling Smørgrav 	/* get port (XXX atoi is a little too tolerant perhaps?) */
329e6182307SDag-Erling Smørgrav 	if (px[len] == ':') {
330e6182307SDag-Erling Smørgrav 	    if (strspn(px+len+1, "0123456789") != strlen(px+len+1)
331e6182307SDag-Erling Smørgrav 		|| strlen(px+len+1) > 5) {
332e6182307SDag-Erling Smørgrav 		/* XXX we should emit some kind of warning */
333e6182307SDag-Erling Smørgrav 	    }
3344ca1ab94SDag-Erling Smørgrav 	    port = atoi(px+len+1);
335e6182307SDag-Erling Smørgrav 	    if (port < 1 || port > 65535) {
336e6182307SDag-Erling Smørgrav 		/* XXX we should emit some kind of warning */
337e6182307SDag-Erling Smørgrav 	    }
338e6182307SDag-Erling Smørgrav 	}
339e6182307SDag-Erling Smørgrav 	if (!port) {
340e6182307SDag-Erling Smørgrav #if 0
341e6182307SDag-Erling Smørgrav 	    /*
342e6182307SDag-Erling Smørgrav 	     * commented out, since there is currently no service name
343e6182307SDag-Erling Smørgrav 	     * for HTTP proxies
344e6182307SDag-Erling Smørgrav 	     */
345e6182307SDag-Erling Smørgrav 	    struct servent *se;
346e6182307SDag-Erling Smørgrav 
347e6182307SDag-Erling Smørgrav 	    if ((se = getservbyname("xxxx", "tcp")) != NULL)
348e6182307SDag-Erling Smørgrav 		port = ntohs(se->s_port);
349e6182307SDag-Erling Smørgrav 	    else
350e6182307SDag-Erling Smørgrav #endif
351e6182307SDag-Erling Smørgrav 		port = 3128;
352e6182307SDag-Erling Smørgrav 	}
3534ca1ab94SDag-Erling Smørgrav 
3544ca1ab94SDag-Erling Smørgrav 	/* get host name */
3554ca1ab94SDag-Erling Smørgrav 	if (len >= MAXHOSTNAMELEN)
3564ca1ab94SDag-Erling Smørgrav 	    len = MAXHOSTNAMELEN - 1;
3574ca1ab94SDag-Erling Smørgrav 	strncpy(host, px, len);
3584ca1ab94SDag-Erling Smørgrav 	host[len] = 0;
3594ca1ab94SDag-Erling Smørgrav 
3604ca1ab94SDag-Erling Smørgrav 	/* connect */
361d941fd2dSDag-Erling Smørgrav 	sd = _fetch_connect(host, port, verbose);
3624ca1ab94SDag-Erling Smørgrav     }
3634ca1ab94SDag-Erling Smørgrav 
3644ca1ab94SDag-Erling Smørgrav     /* if no proxy is configured or could be contacted, try direct */
365ecc91352SDag-Erling Smørgrav     if (sd == -1) {
366d941fd2dSDag-Erling Smørgrav 	if ((sd = _fetch_connect(URL->host, URL->port, verbose)) == -1)
3674ca1ab94SDag-Erling Smørgrav 	    goto ouch;
3684ca1ab94SDag-Erling Smørgrav     }
3694ca1ab94SDag-Erling Smørgrav 
3704ca1ab94SDag-Erling Smørgrav     /* reopen as stream */
3718e3986eaSDag-Erling Smørgrav     if ((f = fdopen(sd, "r+")) == NULL)
3724ca1ab94SDag-Erling Smørgrav 	goto ouch;
3734ca1ab94SDag-Erling Smørgrav     c->real_f = f;
3744ca1ab94SDag-Erling Smørgrav 
3754ca1ab94SDag-Erling Smørgrav     /* send request (proxies require absolute form, so use that) */
3760fba3a00SDag-Erling Smørgrav     if (verbose)
3770fba3a00SDag-Erling Smørgrav 	_fetch_info("requesting http://%s:%d%s",
3780fba3a00SDag-Erling Smørgrav 		    URL->host, URL->port, URL->doc);
379f62e5228SDag-Erling Smørgrav     _http_cmd(f, "GET http://%s:%d%s HTTP/1.1" ENDL,
3804ca1ab94SDag-Erling Smørgrav 	      URL->host, URL->port, URL->doc);
3814ca1ab94SDag-Erling Smørgrav 
3824ca1ab94SDag-Erling Smørgrav     /* start sending headers away */
3834ca1ab94SDag-Erling Smørgrav     if (URL->user[0] || URL->pwd[0]) {
384f62e5228SDag-Erling Smørgrav 	char *auth_str = _http_auth(URL->user, URL->pwd);
385f62e5228SDag-Erling Smørgrav 	if (!auth_str)
386f62e5228SDag-Erling Smørgrav 	    goto fouch;
387f62e5228SDag-Erling Smørgrav 	_http_cmd(f, "Authorization: Basic %s" ENDL, auth_str);
388f62e5228SDag-Erling Smørgrav 	free(auth_str);
3894ca1ab94SDag-Erling Smørgrav     }
390f62e5228SDag-Erling Smørgrav     _http_cmd(f, "Host: %s:%d" ENDL, URL->host, URL->port);
391f62e5228SDag-Erling Smørgrav     _http_cmd(f, "User-Agent: %s " _LIBFETCH_VER ENDL, __progname);
392f62e5228SDag-Erling Smørgrav     _http_cmd(f, "Connection: close" ENDL ENDL);
3934ca1ab94SDag-Erling Smørgrav 
3944ca1ab94SDag-Erling Smørgrav     /* get response */
3954ca1ab94SDag-Erling Smørgrav     if ((ln = fgetln(f, &len)) == NULL)
3964ca1ab94SDag-Erling Smørgrav 	goto fouch;
3974ca1ab94SDag-Erling Smørgrav     DEBUG(fprintf(stderr, "response: [\033[1m%*.*s\033[m]\n",
3984ca1ab94SDag-Erling Smørgrav 		  (int)len-2, (int)len-2, ln));
3994ca1ab94SDag-Erling Smørgrav 
4004ca1ab94SDag-Erling Smørgrav     /* we can't use strchr() and friends since ln isn't NUL-terminated */
4014ca1ab94SDag-Erling Smørgrav     p = ln;
4024ca1ab94SDag-Erling Smørgrav     while ((p < ln + len) && !isspace(*p))
4034ca1ab94SDag-Erling Smørgrav 	p++;
4044ca1ab94SDag-Erling Smørgrav     while ((p < ln + len) && !isdigit(*p))
4054ca1ab94SDag-Erling Smørgrav 	p++;
4064ca1ab94SDag-Erling Smørgrav     if (!isdigit(*p))
4074ca1ab94SDag-Erling Smørgrav 	goto fouch;
4083b7a6740SDag-Erling Smørgrav     e = atoi(p);
4093b7a6740SDag-Erling Smørgrav     DEBUG(fprintf(stderr, "code:     [\033[1m%d\033[m]\n", e));
4104ca1ab94SDag-Erling Smørgrav 
4114ca1ab94SDag-Erling Smørgrav     /* add code to handle redirects later */
4123b7a6740SDag-Erling Smørgrav     if (e != 200) {
4133b7a6740SDag-Erling Smørgrav 	_http_seterr(e);
4144ca1ab94SDag-Erling Smørgrav 	goto fouch;
4158e3986eaSDag-Erling Smørgrav     }
4164ca1ab94SDag-Erling Smørgrav 
4174ca1ab94SDag-Erling Smørgrav     /* browse through header */
4184ca1ab94SDag-Erling Smørgrav     while (1) {
4194ca1ab94SDag-Erling Smørgrav 	if ((ln = fgetln(f, &len)) == NULL)
4204ca1ab94SDag-Erling Smørgrav 	    goto fouch;
4214ca1ab94SDag-Erling Smørgrav 	if ((ln[0] == '\r') || (ln[0] == '\n'))
4224ca1ab94SDag-Erling Smørgrav 	    break;
4234ca1ab94SDag-Erling Smørgrav 	DEBUG(fprintf(stderr, "header:	 [\033[1m%*.*s\033[m]\n",
4244ca1ab94SDag-Erling Smørgrav 		      (int)len-2, (int)len-2, ln));
4254ca1ab94SDag-Erling Smørgrav #define XFERENC "Transfer-Encoding:"
426e6182307SDag-Erling Smørgrav 	if (strncasecmp(ln, XFERENC, sizeof XFERENC - 1) == 0) {
427e6182307SDag-Erling Smørgrav 	    p = ln + sizeof XFERENC - 1;
4284ca1ab94SDag-Erling Smørgrav 	    while ((p < ln + len) && isspace(*p))
4294ca1ab94SDag-Erling Smørgrav 		p++;
4304ca1ab94SDag-Erling Smørgrav 	    for (q = p; (q < ln + len) && !isspace(*q); q++)
4314ca1ab94SDag-Erling Smørgrav 		/* VOID */ ;
4324ca1ab94SDag-Erling Smørgrav 	    *q = 0;
4334ca1ab94SDag-Erling Smørgrav 	    if (strcasecmp(p, "chunked") == 0)
4344ca1ab94SDag-Erling Smørgrav 		enc = ENC_CHUNKED;
4354ca1ab94SDag-Erling Smørgrav 	    DEBUG(fprintf(stderr, "xferenc:  [\033[1m%s\033[m]\n", p));
4364ca1ab94SDag-Erling Smørgrav #undef XFERENC
4374ca1ab94SDag-Erling Smørgrav #define CONTTYPE "Content-Type:"
438e6182307SDag-Erling Smørgrav 	} else if (strncasecmp(ln, CONTTYPE, sizeof CONTTYPE - 1) == 0) {
439e6182307SDag-Erling Smørgrav 	    p = ln + sizeof CONTTYPE - 1;
4404ca1ab94SDag-Erling Smørgrav 	    while ((p < ln + len) && isspace(*p))
4414ca1ab94SDag-Erling Smørgrav 		p++;
4424ca1ab94SDag-Erling Smørgrav 	    for (i = 0; p < ln + len; p++)
4434ca1ab94SDag-Erling Smørgrav 		if (i < HTTPCTYPELEN)
4444ca1ab94SDag-Erling Smørgrav 		    c->content_type[i++] = *p;
4454ca1ab94SDag-Erling Smørgrav 	    do c->content_type[i--] = 0; while (isspace(c->content_type[i]));
4464ca1ab94SDag-Erling Smørgrav 	    DEBUG(fprintf(stderr, "conttype: [\033[1m%s\033[m]\n",
4474ca1ab94SDag-Erling Smørgrav 			  c->content_type));
4484ca1ab94SDag-Erling Smørgrav #undef CONTTYPE
4494ca1ab94SDag-Erling Smørgrav 	}
4504ca1ab94SDag-Erling Smørgrav     }
4514ca1ab94SDag-Erling Smørgrav 
4524ca1ab94SDag-Erling Smørgrav     /* only body remains */
4534ca1ab94SDag-Erling Smørgrav     c->encoding = enc;
4544ca1ab94SDag-Erling Smørgrav     cf = funopen(c,
4554ca1ab94SDag-Erling Smørgrav 		 (int (*)(void *, char *, int))_http_readfn,
4564ca1ab94SDag-Erling Smørgrav 		 (int (*)(void *, const char *, int))_http_writefn,
4574ca1ab94SDag-Erling Smørgrav 		 (fpos_t (*)(void *, fpos_t, int))NULL,
4584ca1ab94SDag-Erling Smørgrav 		 (int (*)(void *))_http_closefn);
4594ca1ab94SDag-Erling Smørgrav     if (cf == NULL)
4604ca1ab94SDag-Erling Smørgrav 	goto fouch;
461e6182307SDag-Erling Smørgrav 
4624ca1ab94SDag-Erling Smørgrav     return cf;
4634ca1ab94SDag-Erling Smørgrav 
4644ca1ab94SDag-Erling Smørgrav ouch:
4658e3986eaSDag-Erling Smørgrav     if (sd >= 0)
4668e3986eaSDag-Erling Smørgrav 	close(sd);
4674ca1ab94SDag-Erling Smørgrav     free(c);
4680fba3a00SDag-Erling Smørgrav     _http_seterr(999); /* XXX do this properly RSN */
4694ca1ab94SDag-Erling Smørgrav     return NULL;
4704ca1ab94SDag-Erling Smørgrav fouch:
4714ca1ab94SDag-Erling Smørgrav     fclose(f);
4724ca1ab94SDag-Erling Smørgrav     free(c);
4730fba3a00SDag-Erling Smørgrav     _http_seterr(999); /* XXX do this properly RSN */
4744ca1ab94SDag-Erling Smørgrav     return NULL;
4754ca1ab94SDag-Erling Smørgrav }
4764ca1ab94SDag-Erling Smørgrav 
4774ca1ab94SDag-Erling Smørgrav FILE *
478d8acd8dcSDag-Erling Smørgrav fetchPutHTTP(struct url *URL, char *flags)
4794ca1ab94SDag-Erling Smørgrav {
4804ca1ab94SDag-Erling Smørgrav     warnx("fetchPutHTTP(): not implemented");
4814ca1ab94SDag-Erling Smørgrav     return NULL;
4824ca1ab94SDag-Erling Smørgrav }
483d8acd8dcSDag-Erling Smørgrav 
484d8acd8dcSDag-Erling Smørgrav /*
485d8acd8dcSDag-Erling Smørgrav  * Get an HTTP document's metadata
486d8acd8dcSDag-Erling Smørgrav  */
487d8acd8dcSDag-Erling Smørgrav int
488d8acd8dcSDag-Erling Smørgrav fetchStatHTTP(struct url *url, struct url_stat *us, char *flags)
489d8acd8dcSDag-Erling Smørgrav {
490d8acd8dcSDag-Erling Smørgrav     warnx("fetchStatHTTP(): not implemented");
491d8acd8dcSDag-Erling Smørgrav     return -1;
492d8acd8dcSDag-Erling Smørgrav }
493ce71b736SDag-Erling Smørgrav 
494ce71b736SDag-Erling Smørgrav /*
495ce71b736SDag-Erling Smørgrav  * List a directory
496ce71b736SDag-Erling Smørgrav  */
497ce71b736SDag-Erling Smørgrav struct url_ent *
498ce71b736SDag-Erling Smørgrav fetchListHTTP(struct url *url, char *flags)
499ce71b736SDag-Erling Smørgrav {
500ce71b736SDag-Erling Smørgrav     warnx("fetchListHTTP(): not implemented");
501ce71b736SDag-Erling Smørgrav     return NULL;
502ce71b736SDag-Erling Smørgrav }
503