14ca1ab94SDag-Erling Smørgrav /*- 2e4878e39SDag-Erling Smørgrav * Copyright (c) 2000 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 15e4878e39SDag-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 314ca1ab94SDag-Erling Smørgrav #include <sys/param.h> 3228c645cfSHajimu UMEMOTO #include <sys/socket.h> 334ca1ab94SDag-Erling Smørgrav 344ca1ab94SDag-Erling Smørgrav #include <ctype.h> 35e4878e39SDag-Erling Smørgrav #include <err.h> 36e4878e39SDag-Erling Smørgrav #include <errno.h> 3760245e42SDag-Erling Smørgrav #include <locale.h> 38e6182307SDag-Erling Smørgrav #include <netdb.h> 39f62e5228SDag-Erling Smørgrav #include <stdarg.h> 404ca1ab94SDag-Erling Smørgrav #include <stdio.h> 414ca1ab94SDag-Erling Smørgrav #include <stdlib.h> 424ca1ab94SDag-Erling Smørgrav #include <string.h> 4360245e42SDag-Erling Smørgrav #include <time.h> 444ca1ab94SDag-Erling Smørgrav #include <unistd.h> 454ca1ab94SDag-Erling Smørgrav 464ca1ab94SDag-Erling Smørgrav #include "fetch.h" 47842a95ccSDag-Erling Smørgrav #include "common.h" 480fba3a00SDag-Erling Smørgrav #include "httperr.h" 494ca1ab94SDag-Erling Smørgrav 50e4878e39SDag-Erling Smørgrav extern char *__progname; /* XXX not portable */ 514ca1ab94SDag-Erling Smørgrav 52e4878e39SDag-Erling Smørgrav /* Maximum number of redirects to follow */ 53e4878e39SDag-Erling Smørgrav #define MAX_REDIRECT 5 544ca1ab94SDag-Erling Smørgrav 55e4878e39SDag-Erling Smørgrav /* Symbolic names for reply codes we care about */ 563d2a8471SDag-Erling Smørgrav #define HTTP_OK 200 573d2a8471SDag-Erling Smørgrav #define HTTP_PARTIAL 206 58e4878e39SDag-Erling Smørgrav #define HTTP_MOVED_PERM 301 59e4878e39SDag-Erling Smørgrav #define HTTP_MOVED_TEMP 302 60e4878e39SDag-Erling Smørgrav #define HTTP_SEE_OTHER 303 61e4878e39SDag-Erling Smørgrav #define HTTP_NEED_AUTH 401 62e4878e39SDag-Erling Smørgrav #define HTTP_NEED_PROXY_AUTH 403 63e4878e39SDag-Erling Smørgrav #define HTTP_PROTOCOL_ERROR 999 64e4878e39SDag-Erling Smørgrav 65e4878e39SDag-Erling Smørgrav #define HTTP_REDIRECT(xyz) ((xyz) == HTTP_MOVED_PERM \ 66e4878e39SDag-Erling Smørgrav || (xyz) == HTTP_MOVED_TEMP \ 67e4878e39SDag-Erling Smørgrav || (xyz) == HTTP_SEE_OTHER) 68e4878e39SDag-Erling Smørgrav 69e4878e39SDag-Erling Smørgrav 70e4878e39SDag-Erling Smørgrav 71e4878e39SDag-Erling Smørgrav /***************************************************************************** 72e4878e39SDag-Erling Smørgrav * I/O functions for decoding chunked streams 73e4878e39SDag-Erling Smørgrav */ 743d2a8471SDag-Erling Smørgrav 754ca1ab94SDag-Erling Smørgrav struct cookie 764ca1ab94SDag-Erling Smørgrav { 77e4878e39SDag-Erling Smørgrav int fd; 784ca1ab94SDag-Erling Smørgrav char *buf; 79e4878e39SDag-Erling Smørgrav size_t b_size; 80e4878e39SDag-Erling Smørgrav size_t b_len; 81e4878e39SDag-Erling Smørgrav int b_pos; 82e4878e39SDag-Erling Smørgrav int eof; 83e4878e39SDag-Erling Smørgrav int error; 84e4878e39SDag-Erling Smørgrav long chunksize; 85e4878e39SDag-Erling Smørgrav #ifdef DEBUG 86e4878e39SDag-Erling Smørgrav long total; 87e4878e39SDag-Erling Smørgrav #endif 884ca1ab94SDag-Erling Smørgrav }; 894ca1ab94SDag-Erling Smørgrav 90f62e5228SDag-Erling Smørgrav /* 91e4878e39SDag-Erling Smørgrav * Get next chunk header 92f62e5228SDag-Erling Smørgrav */ 93f62e5228SDag-Erling Smørgrav static int 94e4878e39SDag-Erling Smørgrav _http_new_chunk(struct cookie *c) 95f62e5228SDag-Erling Smørgrav { 96e4878e39SDag-Erling Smørgrav char *p; 97f62e5228SDag-Erling Smørgrav 98e4878e39SDag-Erling Smørgrav if (_fetch_getln(c->fd, &c->buf, &c->b_size, &c->b_len) == -1) 99e4878e39SDag-Erling Smørgrav return -1; 100f62e5228SDag-Erling Smørgrav 101e4878e39SDag-Erling Smørgrav if (c->b_len < 2 || !ishexnumber(*c->buf)) 102e4878e39SDag-Erling Smørgrav return -1; 103e4878e39SDag-Erling Smørgrav 104e4878e39SDag-Erling Smørgrav for (p = c->buf; !isspace(*p) && *p != ';' && p < c->buf + c->b_len; ++p) 105e4878e39SDag-Erling Smørgrav if (!ishexnumber(*p)) 106e4878e39SDag-Erling Smørgrav return -1; 107e4878e39SDag-Erling Smørgrav else if (isdigit(*p)) 108e4878e39SDag-Erling Smørgrav c->chunksize = c->chunksize * 16 + *p - '0'; 109e4878e39SDag-Erling Smørgrav else 110e4878e39SDag-Erling Smørgrav c->chunksize = c->chunksize * 16 + 10 + tolower(*p) - 'a'; 111e4878e39SDag-Erling Smørgrav 112e4878e39SDag-Erling Smørgrav #ifdef DEBUG 113e4878e39SDag-Erling Smørgrav c->total += c->chunksize; 114e4878e39SDag-Erling Smørgrav if (c->chunksize == 0) 115e4878e39SDag-Erling Smørgrav fprintf(stderr, "\033[1m_http_fillbuf(): " 116e4878e39SDag-Erling Smørgrav "end of last chunk\033[m\n"); 117e4878e39SDag-Erling Smørgrav else 118e4878e39SDag-Erling Smørgrav fprintf(stderr, "\033[1m_http_fillbuf(): " 119e4878e39SDag-Erling Smørgrav "new chunk: %ld (%ld)\033[m\n", c->chunksize, c->total); 120e4878e39SDag-Erling Smørgrav #endif 121e4878e39SDag-Erling Smørgrav 122e4878e39SDag-Erling Smørgrav return c->chunksize; 123f62e5228SDag-Erling Smørgrav } 124f62e5228SDag-Erling Smørgrav 125f62e5228SDag-Erling Smørgrav /* 126f62e5228SDag-Erling Smørgrav * Fill the input buffer, do chunk decoding on the fly 127f62e5228SDag-Erling Smørgrav */ 128e4878e39SDag-Erling Smørgrav static int 1294ca1ab94SDag-Erling Smørgrav _http_fillbuf(struct cookie *c) 1304ca1ab94SDag-Erling Smørgrav { 131e4878e39SDag-Erling Smørgrav if (c->error) 132e4878e39SDag-Erling Smørgrav return -1; 1334ca1ab94SDag-Erling Smørgrav if (c->eof) 134e4878e39SDag-Erling Smørgrav return 0; 1354ca1ab94SDag-Erling Smørgrav 1364ca1ab94SDag-Erling Smørgrav if (c->chunksize == 0) { 137e4878e39SDag-Erling Smørgrav switch (_http_new_chunk(c)) { 138e4878e39SDag-Erling Smørgrav case -1: 139e4878e39SDag-Erling Smørgrav c->error = 1; 140e4878e39SDag-Erling Smørgrav return -1; 141e4878e39SDag-Erling Smørgrav case 0: 1424ca1ab94SDag-Erling Smørgrav c->eof = 1; 143e4878e39SDag-Erling Smørgrav return 0; 1444ca1ab94SDag-Erling Smørgrav } 1454ca1ab94SDag-Erling Smørgrav } 146e4878e39SDag-Erling Smørgrav 147e4878e39SDag-Erling Smørgrav if (c->b_size < c->chunksize) { 148e4878e39SDag-Erling Smørgrav char *tmp; 149e4878e39SDag-Erling Smørgrav 150e4878e39SDag-Erling Smørgrav if ((tmp = realloc(c->buf, c->chunksize)) == NULL) 151e4878e39SDag-Erling Smørgrav return -1; 152e4878e39SDag-Erling Smørgrav c->buf = tmp; 153e4878e39SDag-Erling Smørgrav c->b_size = c->chunksize; 154e4878e39SDag-Erling Smørgrav } 155e4878e39SDag-Erling Smørgrav 156e4878e39SDag-Erling Smørgrav if ((c->b_len = read(c->fd, c->buf, c->chunksize)) == -1) 157e4878e39SDag-Erling Smørgrav return -1; 1584ca1ab94SDag-Erling Smørgrav c->chunksize -= c->b_len; 159e4878e39SDag-Erling Smørgrav 160e4878e39SDag-Erling Smørgrav if (c->chunksize == 0) { 161e4878e39SDag-Erling Smørgrav char endl[2]; 162e4878e39SDag-Erling Smørgrav read(c->fd, endl, 2); 1634ca1ab94SDag-Erling Smørgrav } 164e4878e39SDag-Erling Smørgrav 165e4878e39SDag-Erling Smørgrav c->b_pos = 0; 166e4878e39SDag-Erling Smørgrav 167e4878e39SDag-Erling Smørgrav return c->b_len; 1684ca1ab94SDag-Erling Smørgrav } 1694ca1ab94SDag-Erling Smørgrav 170f62e5228SDag-Erling Smørgrav /* 171f62e5228SDag-Erling Smørgrav * Read function 172f62e5228SDag-Erling Smørgrav */ 1734ca1ab94SDag-Erling Smørgrav static int 174e4878e39SDag-Erling Smørgrav _http_readfn(void *v, char *buf, int len) 1754ca1ab94SDag-Erling Smørgrav { 176e4878e39SDag-Erling Smørgrav struct cookie *c = (struct cookie *)v; 177e4878e39SDag-Erling Smørgrav int l, pos; 1784ca1ab94SDag-Erling Smørgrav 179e4878e39SDag-Erling Smørgrav if (c->error) 180e4878e39SDag-Erling Smørgrav return -1; 181e4878e39SDag-Erling Smørgrav if (c->eof) 182e4878e39SDag-Erling Smørgrav return 0; 183e4878e39SDag-Erling Smørgrav 184e4878e39SDag-Erling Smørgrav for (pos = 0; len > 0; pos += l, len -= l) { 185e4878e39SDag-Erling Smørgrav /* empty buffer */ 186e4878e39SDag-Erling Smørgrav if (!c->buf || c->b_pos == c->b_len) 187e4878e39SDag-Erling Smørgrav if (_http_fillbuf(c) < 1) 188e4878e39SDag-Erling Smørgrav break; 189e4878e39SDag-Erling Smørgrav l = c->b_len - c->b_pos; 190e4878e39SDag-Erling Smørgrav if (len < l) 191e4878e39SDag-Erling Smørgrav l = len; 192e4878e39SDag-Erling Smørgrav bcopy(c->buf + c->b_pos, buf + pos, l); 193e4878e39SDag-Erling Smørgrav c->b_pos += l; 1944ca1ab94SDag-Erling Smørgrav } 1954ca1ab94SDag-Erling Smørgrav 196e4878e39SDag-Erling Smørgrav if (!pos && c->error) 1974ca1ab94SDag-Erling Smørgrav return -1; 198e4878e39SDag-Erling Smørgrav return pos; 1994ca1ab94SDag-Erling Smørgrav } 2004ca1ab94SDag-Erling Smørgrav 201f62e5228SDag-Erling Smørgrav /* 202f62e5228SDag-Erling Smørgrav * Write function 203f62e5228SDag-Erling Smørgrav */ 2044ca1ab94SDag-Erling Smørgrav static int 205e4878e39SDag-Erling Smørgrav _http_writefn(void *v, const char *buf, int len) 2064ca1ab94SDag-Erling Smørgrav { 207e4878e39SDag-Erling Smørgrav struct cookie *c = (struct cookie *)v; 208e4878e39SDag-Erling Smørgrav 209e4878e39SDag-Erling Smørgrav return write(c->fd, buf, len); 2104ca1ab94SDag-Erling Smørgrav } 2114ca1ab94SDag-Erling Smørgrav 212f62e5228SDag-Erling Smørgrav /* 213f62e5228SDag-Erling Smørgrav * Close function 214f62e5228SDag-Erling Smørgrav */ 2154ca1ab94SDag-Erling Smørgrav static int 216e4878e39SDag-Erling Smørgrav _http_closefn(void *v) 2174ca1ab94SDag-Erling Smørgrav { 218e4878e39SDag-Erling Smørgrav struct cookie *c = (struct cookie *)v; 219e4878e39SDag-Erling Smørgrav int r; 220e4878e39SDag-Erling Smørgrav 221e4878e39SDag-Erling Smørgrav r = close(c->fd); 222e4878e39SDag-Erling Smørgrav if (c->buf) 223e4878e39SDag-Erling Smørgrav free(c->buf); 2244ca1ab94SDag-Erling Smørgrav free(c); 225e4878e39SDag-Erling Smørgrav return r; 2264ca1ab94SDag-Erling Smørgrav } 2274ca1ab94SDag-Erling Smørgrav 228f62e5228SDag-Erling Smørgrav /* 229e4878e39SDag-Erling Smørgrav * Wrap a file descriptor up 230f62e5228SDag-Erling Smørgrav */ 231e4878e39SDag-Erling Smørgrav static FILE * 232e4878e39SDag-Erling Smørgrav _http_funopen(int fd) 2334ca1ab94SDag-Erling Smørgrav { 234e4878e39SDag-Erling Smørgrav struct cookie *c; 235e4878e39SDag-Erling Smørgrav FILE *f; 236e4878e39SDag-Erling Smørgrav 237e4878e39SDag-Erling Smørgrav if ((c = calloc(1, sizeof *c)) == NULL) { 238e4878e39SDag-Erling Smørgrav _fetch_syserr(); 239e4878e39SDag-Erling Smørgrav return NULL; 2404ca1ab94SDag-Erling Smørgrav } 241e4878e39SDag-Erling Smørgrav c->fd = fd; 242e4878e39SDag-Erling Smørgrav if (!(f = funopen(c, _http_readfn, _http_writefn, NULL, _http_closefn))) { 243e4878e39SDag-Erling Smørgrav _fetch_syserr(); 244e4878e39SDag-Erling Smørgrav free(c); 245e4878e39SDag-Erling Smørgrav return NULL; 246e4878e39SDag-Erling Smørgrav } 247e4878e39SDag-Erling Smørgrav return f; 248e4878e39SDag-Erling Smørgrav } 249e4878e39SDag-Erling Smørgrav 250e4878e39SDag-Erling Smørgrav 251e4878e39SDag-Erling Smørgrav /***************************************************************************** 252e4878e39SDag-Erling Smørgrav * Helper functions for talking to the server and parsing its replies 253e4878e39SDag-Erling Smørgrav */ 254e4878e39SDag-Erling Smørgrav 255e4878e39SDag-Erling Smørgrav /* Header types */ 256e4878e39SDag-Erling Smørgrav typedef enum { 257e4878e39SDag-Erling Smørgrav hdr_syserror = -2, 258e4878e39SDag-Erling Smørgrav hdr_error = -1, 259e4878e39SDag-Erling Smørgrav hdr_end = 0, 260e4878e39SDag-Erling Smørgrav hdr_unknown = 1, 261e4878e39SDag-Erling Smørgrav hdr_content_length, 262e4878e39SDag-Erling Smørgrav hdr_content_range, 263e4878e39SDag-Erling Smørgrav hdr_last_modified, 264e4878e39SDag-Erling Smørgrav hdr_location, 265e4878e39SDag-Erling Smørgrav hdr_transfer_encoding 266e4878e39SDag-Erling Smørgrav } hdr; 267e4878e39SDag-Erling Smørgrav 268e4878e39SDag-Erling Smørgrav /* Names of interesting headers */ 269e4878e39SDag-Erling Smørgrav static struct { 270e4878e39SDag-Erling Smørgrav hdr num; 271e4878e39SDag-Erling Smørgrav char *name; 272e4878e39SDag-Erling Smørgrav } hdr_names[] = { 273e4878e39SDag-Erling Smørgrav { hdr_content_length, "Content-Length" }, 274e4878e39SDag-Erling Smørgrav { hdr_content_range, "Content-Range" }, 275e4878e39SDag-Erling Smørgrav { hdr_last_modified, "Last-Modified" }, 276e4878e39SDag-Erling Smørgrav { hdr_location, "Location" }, 277e4878e39SDag-Erling Smørgrav { hdr_transfer_encoding, "Transfer-Encoding" }, 278e4878e39SDag-Erling Smørgrav { hdr_unknown, NULL }, 279e4878e39SDag-Erling Smørgrav }; 280e4878e39SDag-Erling Smørgrav 281e4878e39SDag-Erling Smørgrav static char *reply_buf; 282e4878e39SDag-Erling Smørgrav static size_t reply_size; 283e4878e39SDag-Erling Smørgrav static size_t reply_length; 284e4878e39SDag-Erling Smørgrav 285e4878e39SDag-Erling Smørgrav /* 286e4878e39SDag-Erling Smørgrav * Send a formatted line; optionally echo to terminal 287e4878e39SDag-Erling Smørgrav */ 288e4878e39SDag-Erling Smørgrav static int 289e4878e39SDag-Erling Smørgrav _http_cmd(int fd, char *fmt, ...) 290e4878e39SDag-Erling Smørgrav { 291e4878e39SDag-Erling Smørgrav va_list ap; 292e4878e39SDag-Erling Smørgrav size_t len; 293e4878e39SDag-Erling Smørgrav char *msg; 294e4878e39SDag-Erling Smørgrav int r; 295e4878e39SDag-Erling Smørgrav 296e4878e39SDag-Erling Smørgrav va_start(ap, fmt); 297e4878e39SDag-Erling Smørgrav len = vasprintf(&msg, fmt, ap); 298e4878e39SDag-Erling Smørgrav va_end(ap); 299e4878e39SDag-Erling Smørgrav 300e4878e39SDag-Erling Smørgrav if (msg == NULL) { 301e4878e39SDag-Erling Smørgrav errno = ENOMEM; 302e4878e39SDag-Erling Smørgrav _fetch_syserr(); 303e4878e39SDag-Erling Smørgrav return -1; 304e4878e39SDag-Erling Smørgrav } 305e4878e39SDag-Erling Smørgrav 306e4878e39SDag-Erling Smørgrav r = _fetch_putln(fd, msg, len); 307e4878e39SDag-Erling Smørgrav free(msg); 308e4878e39SDag-Erling Smørgrav 309e4878e39SDag-Erling Smørgrav if (r == -1) { 310e4878e39SDag-Erling Smørgrav _fetch_syserr(); 311e4878e39SDag-Erling Smørgrav return -1; 312e4878e39SDag-Erling Smørgrav } 313e4878e39SDag-Erling Smørgrav 314e4878e39SDag-Erling Smørgrav return 0; 315e4878e39SDag-Erling Smørgrav } 316e4878e39SDag-Erling Smørgrav 317e4878e39SDag-Erling Smørgrav /* 318e4878e39SDag-Erling Smørgrav * Get and parse status line 319e4878e39SDag-Erling Smørgrav */ 320e4878e39SDag-Erling Smørgrav static int 321e4878e39SDag-Erling Smørgrav _http_get_reply(int fd) 322e4878e39SDag-Erling Smørgrav { 323e4878e39SDag-Erling Smørgrav if (_fetch_getln(fd, &reply_buf, &reply_size, &reply_length) == -1) 324e4878e39SDag-Erling Smørgrav return -1; 325e4878e39SDag-Erling Smørgrav /* 326e4878e39SDag-Erling Smørgrav * A valid status line looks like "HTTP/m.n xyz reason" where m 327e4878e39SDag-Erling Smørgrav * and n are the major and minor protocol version numbers and xyz 328e4878e39SDag-Erling Smørgrav * is the reply code. 329e4878e39SDag-Erling Smørgrav * We grok HTTP 1.0 and 1.1, so m must be 1 and n must be 0 or 1. 330e4878e39SDag-Erling Smørgrav * We don't care about the reason phrase. 331e4878e39SDag-Erling Smørgrav */ 332e4878e39SDag-Erling Smørgrav if (strncmp(reply_buf, "HTTP/1.", 7) != 0 333e4878e39SDag-Erling Smørgrav || (reply_buf[7] != '0' && reply_buf[7] != '1') || reply_buf[8] != ' ' 334e4878e39SDag-Erling Smørgrav || !isdigit(reply_buf[9]) 335e4878e39SDag-Erling Smørgrav || !isdigit(reply_buf[10]) 336e4878e39SDag-Erling Smørgrav || !isdigit(reply_buf[11])) 337e4878e39SDag-Erling Smørgrav return HTTP_PROTOCOL_ERROR; 338e4878e39SDag-Erling Smørgrav 339e4878e39SDag-Erling Smørgrav return ((reply_buf[9] - '0') * 100 340e4878e39SDag-Erling Smørgrav + (reply_buf[10] - '0') * 10 341e4878e39SDag-Erling Smørgrav + (reply_buf[11] - '0')); 342e4878e39SDag-Erling Smørgrav } 343e4878e39SDag-Erling Smørgrav 344e4878e39SDag-Erling Smørgrav /* 345e4878e39SDag-Erling Smørgrav * Check a header; if the type matches the given string, return a 346e4878e39SDag-Erling Smørgrav * pointer to the beginning of the value. 347e4878e39SDag-Erling Smørgrav */ 348e4878e39SDag-Erling Smørgrav static char * 349e4878e39SDag-Erling Smørgrav _http_match(char *str, char *hdr) 350e4878e39SDag-Erling Smørgrav { 351e4878e39SDag-Erling Smørgrav while (*str && *hdr && tolower(*str++) == tolower(*hdr++)) 352e4878e39SDag-Erling Smørgrav /* nothing */; 353e4878e39SDag-Erling Smørgrav if (*str || *hdr != ':') 354e4878e39SDag-Erling Smørgrav return NULL; 355e4878e39SDag-Erling Smørgrav while (*hdr && isspace(*++hdr)) 356e4878e39SDag-Erling Smørgrav /* nothing */; 357e4878e39SDag-Erling Smørgrav return hdr; 358e4878e39SDag-Erling Smørgrav } 359e4878e39SDag-Erling Smørgrav 360e4878e39SDag-Erling Smørgrav /* 361e4878e39SDag-Erling Smørgrav * Get the next header and return the appropriate symbolic code. 362e4878e39SDag-Erling Smørgrav */ 363e4878e39SDag-Erling Smørgrav static hdr 364e4878e39SDag-Erling Smørgrav _http_next_header(int fd, char **p) 365e4878e39SDag-Erling Smørgrav { 366e4878e39SDag-Erling Smørgrav int i; 367e4878e39SDag-Erling Smørgrav 368e4878e39SDag-Erling Smørgrav if (_fetch_getln(fd, &reply_buf, &reply_size, &reply_length) == -1) 369e4878e39SDag-Erling Smørgrav return hdr_syserror; 370e4878e39SDag-Erling Smørgrav while (reply_length && isspace(reply_buf[reply_length-1])) 371e4878e39SDag-Erling Smørgrav reply_length--; 372e4878e39SDag-Erling Smørgrav reply_buf[reply_length] = 0; 373e4878e39SDag-Erling Smørgrav if (reply_length == 0) 374e4878e39SDag-Erling Smørgrav return hdr_end; 375e4878e39SDag-Erling Smørgrav /* 376e4878e39SDag-Erling Smørgrav * We could check for malformed headers but we don't really care. 377e4878e39SDag-Erling Smørgrav * A valid header starts with a token immediately followed by a 378e4878e39SDag-Erling Smørgrav * colon; a token is any sequence of non-control, non-whitespace 379e4878e39SDag-Erling Smørgrav * characters except "()<>@,;:\\\"{}". 380e4878e39SDag-Erling Smørgrav */ 381e4878e39SDag-Erling Smørgrav for (i = 0; hdr_names[i].num != hdr_unknown; i++) 382e4878e39SDag-Erling Smørgrav if ((*p = _http_match(hdr_names[i].name, reply_buf)) != NULL) 383e4878e39SDag-Erling Smørgrav return hdr_names[i].num; 384e4878e39SDag-Erling Smørgrav return hdr_unknown; 385e4878e39SDag-Erling Smørgrav } 386e4878e39SDag-Erling Smørgrav 387e4878e39SDag-Erling Smørgrav /* 388e4878e39SDag-Erling Smørgrav * Parse a last-modified header 389e4878e39SDag-Erling Smørgrav */ 390e4878e39SDag-Erling Smørgrav static time_t 391e4878e39SDag-Erling Smørgrav _http_parse_mtime(char *p) 392e4878e39SDag-Erling Smørgrav { 393e4878e39SDag-Erling Smørgrav char locale[64]; 394e4878e39SDag-Erling Smørgrav struct tm tm; 395e4878e39SDag-Erling Smørgrav 396e4878e39SDag-Erling Smørgrav strncpy(locale, setlocale(LC_TIME, NULL), sizeof locale); 397e4878e39SDag-Erling Smørgrav setlocale(LC_TIME, "C"); 398e4878e39SDag-Erling Smørgrav strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm); 399e4878e39SDag-Erling Smørgrav /* XXX should add support for date-2 and date-3 */ 400e4878e39SDag-Erling Smørgrav setlocale(LC_TIME, locale); 401e4878e39SDag-Erling Smørgrav DEBUG(fprintf(stderr, "last modified: [\033[1m%04d-%02d-%02d " 402e4878e39SDag-Erling Smørgrav "%02d:%02d:%02d\033[m]\n", 403e4878e39SDag-Erling Smørgrav tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday, 404e4878e39SDag-Erling Smørgrav tm.tm_hour, tm.tm_min, tm.tm_sec)); 405e4878e39SDag-Erling Smørgrav return timegm(&tm); 406e4878e39SDag-Erling Smørgrav } 407e4878e39SDag-Erling Smørgrav 408e4878e39SDag-Erling Smørgrav /* 409e4878e39SDag-Erling Smørgrav * Parse a content-length header 410e4878e39SDag-Erling Smørgrav */ 411e4878e39SDag-Erling Smørgrav static off_t 412e4878e39SDag-Erling Smørgrav _http_parse_length(char *p) 413e4878e39SDag-Erling Smørgrav { 414e4878e39SDag-Erling Smørgrav off_t len; 415e4878e39SDag-Erling Smørgrav 416e4878e39SDag-Erling Smørgrav for (len = 0; *p && isdigit(*p); ++p) 417e4878e39SDag-Erling Smørgrav len = len * 10 + (*p - '0'); 418e4878e39SDag-Erling Smørgrav DEBUG(fprintf(stderr, "content length: [\033[1m%lld\033[m]\n", len)); 419e4878e39SDag-Erling Smørgrav return len; 420e4878e39SDag-Erling Smørgrav } 421e4878e39SDag-Erling Smørgrav 422e4878e39SDag-Erling Smørgrav /* 423e4878e39SDag-Erling Smørgrav * Parse a content-range header 424e4878e39SDag-Erling Smørgrav */ 425e4878e39SDag-Erling Smørgrav static off_t 426e4878e39SDag-Erling Smørgrav _http_parse_range(char *p) 427e4878e39SDag-Erling Smørgrav { 428e4878e39SDag-Erling Smørgrav off_t off; 429e4878e39SDag-Erling Smørgrav 430e4878e39SDag-Erling Smørgrav if (strncasecmp(p, "bytes ", 6) != 0) 431e4878e39SDag-Erling Smørgrav return -1; 432e4878e39SDag-Erling Smørgrav for (p += 6, off = 0; *p && isdigit(*p); ++p) 433e4878e39SDag-Erling Smørgrav off = off * 10 + *p - '0'; 434e4878e39SDag-Erling Smørgrav if (*p != '-') 435e4878e39SDag-Erling Smørgrav return -1; 436e4878e39SDag-Erling Smørgrav DEBUG(fprintf(stderr, "content range: [\033[1m%lld-\033[m]\n", off)); 437e4878e39SDag-Erling Smørgrav return off; 438e4878e39SDag-Erling Smørgrav } 439e4878e39SDag-Erling Smørgrav 440e4878e39SDag-Erling Smørgrav 441e4878e39SDag-Erling Smørgrav /***************************************************************************** 442e4878e39SDag-Erling Smørgrav * Helper functions for authorization 443e4878e39SDag-Erling Smørgrav */ 4444ca1ab94SDag-Erling Smørgrav 445f62e5228SDag-Erling Smørgrav /* 446f62e5228SDag-Erling Smørgrav * Base64 encoding 447f62e5228SDag-Erling Smørgrav */ 44835f723dbSDag-Erling Smørgrav static char * 44935f723dbSDag-Erling Smørgrav _http_base64(char *src) 450f62e5228SDag-Erling Smørgrav { 451f62e5228SDag-Erling Smørgrav static const char base64[] = 452f62e5228SDag-Erling Smørgrav "ABCDEFGHIJKLMNOPQRSTUVWXYZ" 453f62e5228SDag-Erling Smørgrav "abcdefghijklmnopqrstuvwxyz" 454f62e5228SDag-Erling Smørgrav "0123456789+/"; 45535f723dbSDag-Erling Smørgrav char *str, *dst; 45635f723dbSDag-Erling Smørgrav size_t l; 45735f723dbSDag-Erling Smørgrav int t, r; 45835f723dbSDag-Erling Smørgrav 45935f723dbSDag-Erling Smørgrav l = strlen(src); 46035f723dbSDag-Erling Smørgrav if ((str = malloc(((l + 2) / 3) * 4)) == NULL) 46135f723dbSDag-Erling Smørgrav return NULL; 46235f723dbSDag-Erling Smørgrav dst = str; 46335f723dbSDag-Erling Smørgrav r = 0; 464f62e5228SDag-Erling Smørgrav 465f62e5228SDag-Erling Smørgrav while (l >= 3) { 466f62e5228SDag-Erling Smørgrav t = (src[0] << 16) | (src[1] << 8) | src[2]; 467f62e5228SDag-Erling Smørgrav dst[0] = base64[(t >> 18) & 0x3f]; 468f62e5228SDag-Erling Smørgrav dst[1] = base64[(t >> 12) & 0x3f]; 469f62e5228SDag-Erling Smørgrav dst[2] = base64[(t >> 6) & 0x3f]; 470f62e5228SDag-Erling Smørgrav dst[3] = base64[(t >> 0) & 0x3f]; 471f62e5228SDag-Erling Smørgrav src += 3; l -= 3; 472f62e5228SDag-Erling Smørgrav dst += 4; r += 4; 473f62e5228SDag-Erling Smørgrav } 474f62e5228SDag-Erling Smørgrav 475f62e5228SDag-Erling Smørgrav switch (l) { 476f62e5228SDag-Erling Smørgrav case 2: 477f62e5228SDag-Erling Smørgrav t = (src[0] << 16) | (src[1] << 8); 478f62e5228SDag-Erling Smørgrav dst[0] = base64[(t >> 18) & 0x3f]; 479f62e5228SDag-Erling Smørgrav dst[1] = base64[(t >> 12) & 0x3f]; 480f62e5228SDag-Erling Smørgrav dst[2] = base64[(t >> 6) & 0x3f]; 481f62e5228SDag-Erling Smørgrav dst[3] = '='; 482f62e5228SDag-Erling Smørgrav dst += 4; 483f62e5228SDag-Erling Smørgrav r += 4; 484f62e5228SDag-Erling Smørgrav break; 485f62e5228SDag-Erling Smørgrav case 1: 486f62e5228SDag-Erling Smørgrav t = src[0] << 16; 487f62e5228SDag-Erling Smørgrav dst[0] = base64[(t >> 18) & 0x3f]; 488f62e5228SDag-Erling Smørgrav dst[1] = base64[(t >> 12) & 0x3f]; 489f62e5228SDag-Erling Smørgrav dst[2] = dst[3] = '='; 490f62e5228SDag-Erling Smørgrav dst += 4; 491f62e5228SDag-Erling Smørgrav r += 4; 492f62e5228SDag-Erling Smørgrav break; 493f62e5228SDag-Erling Smørgrav case 0: 494f62e5228SDag-Erling Smørgrav break; 495f62e5228SDag-Erling Smørgrav } 496f62e5228SDag-Erling Smørgrav 497f62e5228SDag-Erling Smørgrav *dst = 0; 49835f723dbSDag-Erling Smørgrav return str; 499f62e5228SDag-Erling Smørgrav } 500f62e5228SDag-Erling Smørgrav 501f62e5228SDag-Erling Smørgrav /* 502f62e5228SDag-Erling Smørgrav * Encode username and password 503f62e5228SDag-Erling Smørgrav */ 50435f723dbSDag-Erling Smørgrav static int 505e4878e39SDag-Erling Smørgrav _http_basic_auth(int fd, char *hdr, char *usr, char *pwd) 506f62e5228SDag-Erling Smørgrav { 50735f723dbSDag-Erling Smørgrav char *upw, *auth; 50835f723dbSDag-Erling Smørgrav int r; 509f62e5228SDag-Erling Smørgrav 51035f723dbSDag-Erling Smørgrav if (asprintf(&upw, "%s:%s", usr, pwd) == -1) 51135f723dbSDag-Erling Smørgrav return -1; 51235f723dbSDag-Erling Smørgrav auth = _http_base64(upw); 51335f723dbSDag-Erling Smørgrav free(upw); 51435f723dbSDag-Erling Smørgrav if (auth == NULL) 51535f723dbSDag-Erling Smørgrav return -1; 516e4878e39SDag-Erling Smørgrav r = _http_cmd(fd, "%s: Basic %s", hdr, auth); 51735f723dbSDag-Erling Smørgrav free(auth); 51835f723dbSDag-Erling Smørgrav return r; 5194d029f13SDag-Erling Smørgrav } 52035f723dbSDag-Erling Smørgrav 52135f723dbSDag-Erling Smørgrav /* 52235f723dbSDag-Erling Smørgrav * Send an authorization header 52335f723dbSDag-Erling Smørgrav */ 52435f723dbSDag-Erling Smørgrav static int 525e4878e39SDag-Erling Smørgrav _http_authorize(int fd, char *hdr, char *p) 52635f723dbSDag-Erling Smørgrav { 52735f723dbSDag-Erling Smørgrav /* basic authorization */ 52835f723dbSDag-Erling Smørgrav if (strncasecmp(p, "basic:", 6) == 0) { 52935f723dbSDag-Erling Smørgrav char *user, *pwd, *str; 53035f723dbSDag-Erling Smørgrav int r; 53135f723dbSDag-Erling Smørgrav 53235f723dbSDag-Erling Smørgrav /* skip realm */ 53335f723dbSDag-Erling Smørgrav for (p += 6; *p && *p != ':'; ++p) 53435f723dbSDag-Erling Smørgrav /* nothing */ ; 53535f723dbSDag-Erling Smørgrav if (!*p || strchr(++p, ':') == NULL) 53635f723dbSDag-Erling Smørgrav return -1; 53735f723dbSDag-Erling Smørgrav if ((str = strdup(p)) == NULL) 53835f723dbSDag-Erling Smørgrav return -1; /* XXX */ 53935f723dbSDag-Erling Smørgrav user = str; 54035f723dbSDag-Erling Smørgrav pwd = strchr(str, ':'); 54135f723dbSDag-Erling Smørgrav *pwd++ = '\0'; 542e4878e39SDag-Erling Smørgrav r = _http_basic_auth(fd, hdr, user, pwd); 54335f723dbSDag-Erling Smørgrav free(str); 54435f723dbSDag-Erling Smørgrav return r; 5454d029f13SDag-Erling Smørgrav } 54635f723dbSDag-Erling Smørgrav return -1; 547f62e5228SDag-Erling Smørgrav } 548f62e5228SDag-Erling Smørgrav 549e4878e39SDag-Erling Smørgrav 550e4878e39SDag-Erling Smørgrav /***************************************************************************** 551e4878e39SDag-Erling Smørgrav * Helper functions for connecting to a server or proxy 552f62e5228SDag-Erling Smørgrav */ 553e4878e39SDag-Erling Smørgrav 554e4878e39SDag-Erling Smørgrav /* 555e4878e39SDag-Erling Smørgrav * Connect to the specified HTTP proxy server. 556e4878e39SDag-Erling Smørgrav */ 557e4878e39SDag-Erling Smørgrav static int 558e4878e39SDag-Erling Smørgrav _http_proxy_connect(char *proxy, int af, int verbose) 5594ca1ab94SDag-Erling Smørgrav { 560e4878e39SDag-Erling Smørgrav char *hostname, *p; 561e4878e39SDag-Erling Smørgrav int fd, port; 562e4878e39SDag-Erling Smørgrav 563e4878e39SDag-Erling Smørgrav /* get hostname */ 564e4878e39SDag-Erling Smørgrav hostname = NULL; 56528c645cfSHajimu UMEMOTO #ifdef INET6 566e4878e39SDag-Erling Smørgrav /* host part can be an IPv6 address enclosed in square brackets */ 567e4878e39SDag-Erling Smørgrav if (*proxy == '[') { 568e4878e39SDag-Erling Smørgrav if ((p = strchr(proxy, ']')) == NULL) { 569e4878e39SDag-Erling Smørgrav /* no terminating bracket */ 570e4878e39SDag-Erling Smørgrav /* XXX should set an error code */ 571e4878e39SDag-Erling Smørgrav goto ouch; 572e6182307SDag-Erling Smørgrav } 573e4878e39SDag-Erling Smørgrav if (p[1] != '\0' && p[1] != ':') { 574e4878e39SDag-Erling Smørgrav /* garbage after address */ 575e4878e39SDag-Erling Smørgrav /* XXX should set an error code */ 576e4878e39SDag-Erling Smørgrav goto ouch; 577e4878e39SDag-Erling Smørgrav } 578e4878e39SDag-Erling Smørgrav if ((hostname = malloc(p - proxy)) == NULL) { 579e4878e39SDag-Erling Smørgrav errno = ENOMEM; 580e4878e39SDag-Erling Smørgrav _fetch_syserr(); 581e4878e39SDag-Erling Smørgrav goto ouch; 582e4878e39SDag-Erling Smørgrav } 583e4878e39SDag-Erling Smørgrav strncpy(hostname, proxy + 1, p - proxy - 1); 584e4878e39SDag-Erling Smørgrav hostname[p - proxy - 1] = '\0'; 585e4878e39SDag-Erling Smørgrav ++p; 586e4878e39SDag-Erling Smørgrav } else { 587e4878e39SDag-Erling Smørgrav #endif /* INET6 */ 588e4878e39SDag-Erling Smørgrav if ((p = strchr(proxy, ':')) == NULL) 589e4878e39SDag-Erling Smørgrav p = strchr(proxy, '\0'); 590e4878e39SDag-Erling Smørgrav if ((hostname = malloc(p - proxy + 1)) == NULL) { 591e4878e39SDag-Erling Smørgrav errno = ENOMEM; 592e4878e39SDag-Erling Smørgrav _fetch_syserr(); 593e4878e39SDag-Erling Smørgrav goto ouch; 594e4878e39SDag-Erling Smørgrav } 595e4878e39SDag-Erling Smørgrav strncpy(hostname, proxy, p - proxy); 596e4878e39SDag-Erling Smørgrav hostname[p - proxy] = '\0'; 59728c645cfSHajimu UMEMOTO #ifdef INET6 598e6182307SDag-Erling Smørgrav } 599e4878e39SDag-Erling Smørgrav #endif /* INET6 */ 600e4878e39SDag-Erling Smørgrav DEBUG(fprintf(stderr, "proxy name: [%s]\n", hostname)); 601e4878e39SDag-Erling Smørgrav 602e4878e39SDag-Erling Smørgrav /* get port number */ 603e4878e39SDag-Erling Smørgrav port = 0; 604e4878e39SDag-Erling Smørgrav if (*p == ':') { 605e4878e39SDag-Erling Smørgrav ++p; 606e4878e39SDag-Erling Smørgrav if (strspn(p, "0123456789") != strlen(p) || strlen(p) > 5) { 607e4878e39SDag-Erling Smørgrav /* port number is non-numeric or too long */ 608e4878e39SDag-Erling Smørgrav /* XXX should set an error code */ 609e4878e39SDag-Erling Smørgrav goto ouch; 610e4878e39SDag-Erling Smørgrav } 611e4878e39SDag-Erling Smørgrav port = atoi(p); 612e6182307SDag-Erling Smørgrav if (port < 1 || port > 65535) { 613e4878e39SDag-Erling Smørgrav /* port number is out of range */ 614e4878e39SDag-Erling Smørgrav /* XXX should set an error code */ 615e4878e39SDag-Erling Smørgrav goto ouch; 616e6182307SDag-Erling Smørgrav } 617e6182307SDag-Erling Smørgrav } 618e4878e39SDag-Erling Smørgrav 619e6182307SDag-Erling Smørgrav if (!port) { 620e6182307SDag-Erling Smørgrav #if 0 621e6182307SDag-Erling Smørgrav /* 622e6182307SDag-Erling Smørgrav * commented out, since there is currently no service name 623e6182307SDag-Erling Smørgrav * for HTTP proxies 624e6182307SDag-Erling Smørgrav */ 625e6182307SDag-Erling Smørgrav struct servent *se; 626e6182307SDag-Erling Smørgrav 627e6182307SDag-Erling Smørgrav if ((se = getservbyname("xxxx", "tcp")) != NULL) 628e6182307SDag-Erling Smørgrav port = ntohs(se->s_port); 629e6182307SDag-Erling Smørgrav else 630e6182307SDag-Erling Smørgrav #endif 631e6182307SDag-Erling Smørgrav port = 3128; 632e6182307SDag-Erling Smørgrav } 633e4878e39SDag-Erling Smørgrav DEBUG(fprintf(stderr, "proxy port: %d\n", port)); 6344ca1ab94SDag-Erling Smørgrav 6354ca1ab94SDag-Erling Smørgrav /* connect */ 636e4878e39SDag-Erling Smørgrav if ((fd = _fetch_connect(hostname, port, af, verbose)) == -1) 637e4878e39SDag-Erling Smørgrav _fetch_syserr(); 638e4878e39SDag-Erling Smørgrav return fd; 63960245e42SDag-Erling Smørgrav 64060245e42SDag-Erling Smørgrav ouch: 641e4878e39SDag-Erling Smørgrav if (hostname) 642e4878e39SDag-Erling Smørgrav free(hostname); 643e4878e39SDag-Erling Smørgrav return -1; 64460245e42SDag-Erling Smørgrav } 64560245e42SDag-Erling Smørgrav 64660245e42SDag-Erling Smørgrav /* 647e4878e39SDag-Erling Smørgrav * Connect to the correct HTTP server or proxy. 64860245e42SDag-Erling Smørgrav */ 64935f723dbSDag-Erling Smørgrav static int 650e4878e39SDag-Erling Smørgrav _http_connect(struct url *URL, int *proxy, char *flags) 65160245e42SDag-Erling Smørgrav { 652e4878e39SDag-Erling Smørgrav int direct, verbose; 653e4878e39SDag-Erling Smørgrav int af, fd; 654e4878e39SDag-Erling Smørgrav char *p; 655e4878e39SDag-Erling Smørgrav 656e4878e39SDag-Erling Smørgrav #ifdef INET6 657e4878e39SDag-Erling Smørgrav af = AF_UNSPEC; 658e4878e39SDag-Erling Smørgrav #else 659e4878e39SDag-Erling Smørgrav af = AF_INET; 660e4878e39SDag-Erling Smørgrav #endif 661e4878e39SDag-Erling Smørgrav 662e4878e39SDag-Erling Smørgrav direct = (flags && strchr(flags, 'd')); 663e4878e39SDag-Erling Smørgrav verbose = (flags && strchr(flags, 'v')); 664e4878e39SDag-Erling Smørgrav if (flags && strchr(flags, '4')) 665e4878e39SDag-Erling Smørgrav af = AF_INET; 666e4878e39SDag-Erling Smørgrav else if (flags && strchr(flags, '6')) 667e4878e39SDag-Erling Smørgrav af = AF_INET6; 668e4878e39SDag-Erling Smørgrav 669e4878e39SDag-Erling Smørgrav /* check port */ 670e4878e39SDag-Erling Smørgrav if (!URL->port) { 671e4878e39SDag-Erling Smørgrav struct servent *se; 672e4878e39SDag-Erling Smørgrav 673e4878e39SDag-Erling Smørgrav /* Scheme can be ftp if we're using a proxy */ 674e4878e39SDag-Erling Smørgrav if (strcasecmp(URL->scheme, "ftp") == 0) 675e4878e39SDag-Erling Smørgrav if ((se = getservbyname("ftp", "tcp")) != NULL) 676e4878e39SDag-Erling Smørgrav URL->port = ntohs(se->s_port); 677e4878e39SDag-Erling Smørgrav else 678e4878e39SDag-Erling Smørgrav URL->port = 21; 679e4878e39SDag-Erling Smørgrav else 680e4878e39SDag-Erling Smørgrav if ((se = getservbyname("http", "tcp")) != NULL) 681e4878e39SDag-Erling Smørgrav URL->port = ntohs(se->s_port); 682e4878e39SDag-Erling Smørgrav else 683e4878e39SDag-Erling Smørgrav URL->port = 80; 684e4878e39SDag-Erling Smørgrav } 685e4878e39SDag-Erling Smørgrav 686e4878e39SDag-Erling Smørgrav if (!direct && (p = getenv("HTTP_PROXY")) != NULL) { 687e4878e39SDag-Erling Smørgrav /* attempt to connect to proxy server */ 688e4878e39SDag-Erling Smørgrav if ((fd = _http_proxy_connect(p, af, verbose)) == -1) 689e4878e39SDag-Erling Smørgrav return -1; 690e4878e39SDag-Erling Smørgrav *proxy = 1; 691e4878e39SDag-Erling Smørgrav } else { 692e4878e39SDag-Erling Smørgrav /* if no proxy is configured, try direct */ 693e4878e39SDag-Erling Smørgrav if (strcasecmp(URL->scheme, "ftp") == 0) { 694e4878e39SDag-Erling Smørgrav /* can't talk http to an ftp server */ 695e4878e39SDag-Erling Smørgrav /* XXX should set an error code */ 696e4878e39SDag-Erling Smørgrav return -1; 697e4878e39SDag-Erling Smørgrav } 698e4878e39SDag-Erling Smørgrav if ((fd = _fetch_connect(URL->host, URL->port, af, verbose)) == -1) 699e4878e39SDag-Erling Smørgrav /* _fetch_connect() has already set an error code */ 700e4878e39SDag-Erling Smørgrav return -1; 701e4878e39SDag-Erling Smørgrav *proxy = 0; 702e4878e39SDag-Erling Smørgrav } 703e4878e39SDag-Erling Smørgrav 704e4878e39SDag-Erling Smørgrav return fd; 705e4878e39SDag-Erling Smørgrav } 706e4878e39SDag-Erling Smørgrav 707e4878e39SDag-Erling Smørgrav 708e4878e39SDag-Erling Smørgrav /***************************************************************************** 709e4878e39SDag-Erling Smørgrav * Core 710e4878e39SDag-Erling Smørgrav */ 711e4878e39SDag-Erling Smørgrav 712e4878e39SDag-Erling Smørgrav /* 713e4878e39SDag-Erling Smørgrav * Send a request and process the reply 714e4878e39SDag-Erling Smørgrav */ 715e4878e39SDag-Erling Smørgrav static FILE * 716e4878e39SDag-Erling Smørgrav _http_request(struct url *URL, char *op, struct url_stat *us, char *flags) 717e4878e39SDag-Erling Smørgrav { 718e4878e39SDag-Erling Smørgrav struct url *url, *new; 719e4878e39SDag-Erling Smørgrav int chunked, need_auth, noredirect, proxy, verbose; 720e4878e39SDag-Erling Smørgrav int code, fd, i, n; 721e4878e39SDag-Erling Smørgrav off_t offset; 722e4878e39SDag-Erling Smørgrav char *p; 723e4878e39SDag-Erling Smørgrav FILE *f; 724e4878e39SDag-Erling Smørgrav hdr h; 72528c645cfSHajimu UMEMOTO char *host; 72628c645cfSHajimu UMEMOTO #ifdef INET6 72728c645cfSHajimu UMEMOTO char hbuf[MAXHOSTNAMELEN + 1]; 72828c645cfSHajimu UMEMOTO #endif 72960245e42SDag-Erling Smørgrav 730e4878e39SDag-Erling Smørgrav noredirect = (flags && strchr(flags, 'A')); 73160245e42SDag-Erling Smørgrav verbose = (flags && strchr(flags, 'v')); 7324ca1ab94SDag-Erling Smørgrav 733e4878e39SDag-Erling Smørgrav n = noredirect ? 1 : MAX_REDIRECT; 734e4878e39SDag-Erling Smørgrav 735e4878e39SDag-Erling Smørgrav us->size = -1; 736e4878e39SDag-Erling Smørgrav us->atime = us->mtime = 0; 737e4878e39SDag-Erling Smørgrav chunked = 0; 738e4878e39SDag-Erling Smørgrav offset = 0; 739e4878e39SDag-Erling Smørgrav fd = -1; 740e4878e39SDag-Erling Smørgrav code = HTTP_PROTOCOL_ERROR; /* just to appease a compiler warning */ 741e4878e39SDag-Erling Smørgrav 742e4878e39SDag-Erling Smørgrav for (url = URL, i = 0; i < n; ++i) { 743e4878e39SDag-Erling Smørgrav need_auth = 0; 744e4878e39SDag-Erling Smørgrav retry: 745e4878e39SDag-Erling Smørgrav /* connect to server or proxy */ 746e4878e39SDag-Erling Smørgrav if ((fd = _http_connect(url, &proxy, flags)) == -1) 747e4878e39SDag-Erling Smørgrav goto ouch; 748e4878e39SDag-Erling Smørgrav 749e4878e39SDag-Erling Smørgrav host = url->host; 75028c645cfSHajimu UMEMOTO #ifdef INET6 751e4878e39SDag-Erling Smørgrav if (strchr(url->host, ':')) { 752e4878e39SDag-Erling Smørgrav snprintf(hbuf, sizeof(hbuf), "[%s]", url->host); 75328c645cfSHajimu UMEMOTO host = hbuf; 75428c645cfSHajimu UMEMOTO } 75528c645cfSHajimu UMEMOTO #endif 75628c645cfSHajimu UMEMOTO 757e4878e39SDag-Erling Smørgrav /* send request */ 7580fba3a00SDag-Erling Smørgrav if (verbose) 759c97925adSHajimu UMEMOTO _fetch_info("requesting %s://%s:%d%s", 760e4878e39SDag-Erling Smørgrav url->scheme, host, url->port, url->doc); 761e4878e39SDag-Erling Smørgrav if (proxy) { 762e4878e39SDag-Erling Smørgrav _http_cmd(fd, "%s %s://%s:%d%s HTTP/1.1", 763e4878e39SDag-Erling Smørgrav op, url->scheme, host, url->port, url->doc); 764e4878e39SDag-Erling Smørgrav } else { 765e4878e39SDag-Erling Smørgrav _http_cmd(fd, "%s %s HTTP/1.1", 766e4878e39SDag-Erling Smørgrav op, url->doc); 76760245e42SDag-Erling Smørgrav } 76860245e42SDag-Erling Smørgrav 769e4878e39SDag-Erling Smørgrav /* proxy authorization */ 770e4878e39SDag-Erling Smørgrav if (proxy && (p = getenv("HTTP_PROXY_AUTH")) != NULL) 771e4878e39SDag-Erling Smørgrav _http_authorize(fd, "Proxy-Authorization", p); 772e4878e39SDag-Erling Smørgrav 773e4878e39SDag-Erling Smørgrav /* server authorization */ 774e4878e39SDag-Erling Smørgrav if (need_auth) { 775e4878e39SDag-Erling Smørgrav if (*url->user || *url->pwd) 776e4878e39SDag-Erling Smørgrav _http_basic_auth(fd, "Authorization", 777e4878e39SDag-Erling Smørgrav url->user ? url->user : "", 778e4878e39SDag-Erling Smørgrav url->pwd ? url->pwd : ""); 779e4878e39SDag-Erling Smørgrav else if ((p = getenv("HTTP_AUTH")) != NULL) 780e4878e39SDag-Erling Smørgrav _http_authorize(fd, "Authorization", p); 781e4878e39SDag-Erling Smørgrav else { 782e4878e39SDag-Erling Smørgrav _http_seterr(HTTP_NEED_AUTH); 783e4878e39SDag-Erling Smørgrav goto ouch; 784e4878e39SDag-Erling Smørgrav } 785e4878e39SDag-Erling Smørgrav } 786e4878e39SDag-Erling Smørgrav 787e4878e39SDag-Erling Smørgrav /* other headers */ 788e4878e39SDag-Erling Smørgrav _http_cmd(fd, "Host: %s:%d", host, url->port); 789e4878e39SDag-Erling Smørgrav _http_cmd(fd, "User-Agent: %s " _LIBFETCH_VER, __progname); 790e4878e39SDag-Erling Smørgrav if (URL->offset) 791e4878e39SDag-Erling Smørgrav _http_cmd(fd, "Range: bytes=%lld-", url->offset); 792e4878e39SDag-Erling Smørgrav _http_cmd(fd, "Connection: close"); 793e4878e39SDag-Erling Smørgrav _http_cmd(fd, ""); 794e4878e39SDag-Erling Smørgrav 795e4878e39SDag-Erling Smørgrav /* get reply */ 796e4878e39SDag-Erling Smørgrav switch ((code = _http_get_reply(fd))) { 797e4878e39SDag-Erling Smørgrav case HTTP_OK: 798e4878e39SDag-Erling Smørgrav case HTTP_PARTIAL: 799e4878e39SDag-Erling Smørgrav /* fine */ 800e4878e39SDag-Erling Smørgrav break; 801e4878e39SDag-Erling Smørgrav case HTTP_MOVED_PERM: 802e4878e39SDag-Erling Smørgrav case HTTP_MOVED_TEMP: 803e4878e39SDag-Erling Smørgrav /* 804e4878e39SDag-Erling Smørgrav * Not so fine, but we still have to read the headers to 805e4878e39SDag-Erling Smørgrav * get the new location. 806e4878e39SDag-Erling Smørgrav */ 807e4878e39SDag-Erling Smørgrav break; 808e4878e39SDag-Erling Smørgrav case HTTP_NEED_AUTH: 809e4878e39SDag-Erling Smørgrav if (need_auth) { 810e4878e39SDag-Erling Smørgrav /* 811e4878e39SDag-Erling Smørgrav * We already sent out authorization code, so there's 812e4878e39SDag-Erling Smørgrav * nothing more we can do. 813e4878e39SDag-Erling Smørgrav */ 814e4878e39SDag-Erling Smørgrav _http_seterr(code); 815e4878e39SDag-Erling Smørgrav goto ouch; 816e4878e39SDag-Erling Smørgrav } 817e4878e39SDag-Erling Smørgrav /* try again, but send the password this time */ 818e4878e39SDag-Erling Smørgrav if (verbose) 819e4878e39SDag-Erling Smørgrav _fetch_info("server requires authorization"); 820e4878e39SDag-Erling Smørgrav need_auth = 1; 821e4878e39SDag-Erling Smørgrav close(fd); 822e4878e39SDag-Erling Smørgrav goto retry; 823e4878e39SDag-Erling Smørgrav case HTTP_NEED_PROXY_AUTH: 824e4878e39SDag-Erling Smørgrav /* 825e4878e39SDag-Erling Smørgrav * If we're talking to a proxy, we already sent our proxy 826e4878e39SDag-Erling Smørgrav * authorization code, so there's nothing more we can do. 827e4878e39SDag-Erling Smørgrav */ 828e4878e39SDag-Erling Smørgrav _http_seterr(code); 829e4878e39SDag-Erling Smørgrav goto ouch; 830e4878e39SDag-Erling Smørgrav case HTTP_PROTOCOL_ERROR: 831e4878e39SDag-Erling Smørgrav /* fall through */ 832e4878e39SDag-Erling Smørgrav case -1: 833e4878e39SDag-Erling Smørgrav _fetch_syserr(); 834e4878e39SDag-Erling Smørgrav goto ouch; 835e4878e39SDag-Erling Smørgrav default: 836e4878e39SDag-Erling Smørgrav _http_seterr(code); 837e4878e39SDag-Erling Smørgrav goto ouch; 838e4878e39SDag-Erling Smørgrav } 839e4878e39SDag-Erling Smørgrav 840e4878e39SDag-Erling Smørgrav /* get headers */ 841e4878e39SDag-Erling Smørgrav do { 842e4878e39SDag-Erling Smørgrav switch ((h = _http_next_header(fd, &p))) { 843e4878e39SDag-Erling Smørgrav case hdr_syserror: 844e4878e39SDag-Erling Smørgrav _fetch_syserr(); 845e4878e39SDag-Erling Smørgrav goto ouch; 846e4878e39SDag-Erling Smørgrav case hdr_error: 847e4878e39SDag-Erling Smørgrav _http_seterr(HTTP_PROTOCOL_ERROR); 848e4878e39SDag-Erling Smørgrav goto ouch; 849e4878e39SDag-Erling Smørgrav case hdr_content_length: 850e4878e39SDag-Erling Smørgrav us->size = _http_parse_length(p); 851e4878e39SDag-Erling Smørgrav break; 852e4878e39SDag-Erling Smørgrav case hdr_content_range: 853e4878e39SDag-Erling Smørgrav offset = _http_parse_range(p); 854e4878e39SDag-Erling Smørgrav break; 855e4878e39SDag-Erling Smørgrav case hdr_last_modified: 856e4878e39SDag-Erling Smørgrav us->atime = us->mtime = _http_parse_mtime(p); 857e4878e39SDag-Erling Smørgrav break; 858e4878e39SDag-Erling Smørgrav case hdr_location: 859e4878e39SDag-Erling Smørgrav if (!HTTP_REDIRECT(code)) 860e4878e39SDag-Erling Smørgrav break; 861e4878e39SDag-Erling Smørgrav if (verbose) 862e4878e39SDag-Erling Smørgrav _fetch_info("%d redirect to %s", code, p); 863e4878e39SDag-Erling Smørgrav if ((new = fetchParseURL(p)) == NULL) 864e4878e39SDag-Erling Smørgrav /* invalid location */ 865e4878e39SDag-Erling Smørgrav goto ouch; 866e4878e39SDag-Erling Smørgrav if (!*new->user && !*new->pwd) { 867e4878e39SDag-Erling Smørgrav strcpy(new->user, url->user); 868e4878e39SDag-Erling Smørgrav strcpy(new->pwd, url->pwd); 869e4878e39SDag-Erling Smørgrav } 870e4878e39SDag-Erling Smørgrav new->offset = url->offset; 871e4878e39SDag-Erling Smørgrav new->length = url->length; 872e4878e39SDag-Erling Smørgrav close(fd); 873e4878e39SDag-Erling Smørgrav us->size = -1; 874e4878e39SDag-Erling Smørgrav us->atime = us->mtime = 0; 875e4878e39SDag-Erling Smørgrav chunked = 0; 876e4878e39SDag-Erling Smørgrav offset = 0; 877e4878e39SDag-Erling Smørgrav fd = -1; 878e4878e39SDag-Erling Smørgrav if (url != URL) 879e4878e39SDag-Erling Smørgrav fetchFreeURL(url); 880e4878e39SDag-Erling Smørgrav url = new; 881e4878e39SDag-Erling Smørgrav continue; 882e4878e39SDag-Erling Smørgrav case hdr_transfer_encoding: 883e4878e39SDag-Erling Smørgrav /* XXX weak test*/ 884e4878e39SDag-Erling Smørgrav chunked = (strcasecmp(p, "chunked") == 0); 885e4878e39SDag-Erling Smørgrav break; 886e4878e39SDag-Erling Smørgrav case hdr_end: 887e4878e39SDag-Erling Smørgrav /* fall through */ 888e4878e39SDag-Erling Smørgrav case hdr_unknown: 889e4878e39SDag-Erling Smørgrav /* ignore */ 890e4878e39SDag-Erling Smørgrav break; 891e4878e39SDag-Erling Smørgrav } 892e4878e39SDag-Erling Smørgrav } while (h > hdr_end); 893e4878e39SDag-Erling Smørgrav 894e4878e39SDag-Erling Smørgrav if (code == HTTP_OK || code == HTTP_PARTIAL) 895e4878e39SDag-Erling Smørgrav break; 896e4878e39SDag-Erling Smørgrav } 897e4878e39SDag-Erling Smørgrav 898e4878e39SDag-Erling Smørgrav /* no success */ 899e4878e39SDag-Erling Smørgrav if (fd == -1) { 900e4878e39SDag-Erling Smørgrav _http_seterr(code); 901e4878e39SDag-Erling Smørgrav goto ouch; 902e4878e39SDag-Erling Smørgrav } 903e4878e39SDag-Erling Smørgrav 904e4878e39SDag-Erling Smørgrav /* wrap it up in a FILE */ 905e4878e39SDag-Erling Smørgrav if ((f = chunked ? _http_funopen(fd) : fdopen(fd, "r")) == NULL) { 906e4878e39SDag-Erling Smørgrav _fetch_syserr(); 907e4878e39SDag-Erling Smørgrav goto ouch; 908e4878e39SDag-Erling Smørgrav } 909e4878e39SDag-Erling Smørgrav 910e4878e39SDag-Erling Smørgrav while (offset++ < url->offset) 911e4878e39SDag-Erling Smørgrav if (fgetc(f) == EOF) { 912e4878e39SDag-Erling Smørgrav _fetch_syserr(); 913e4878e39SDag-Erling Smørgrav fclose(f); 914e4878e39SDag-Erling Smørgrav f = NULL; 915e4878e39SDag-Erling Smørgrav } 916e4878e39SDag-Erling Smørgrav 917e4878e39SDag-Erling Smørgrav if (url != URL) 918e4878e39SDag-Erling Smørgrav fetchFreeURL(url); 919e4878e39SDag-Erling Smørgrav 920e4878e39SDag-Erling Smørgrav return f; 921e4878e39SDag-Erling Smørgrav 922e4878e39SDag-Erling Smørgrav ouch: 923e4878e39SDag-Erling Smørgrav if (url != URL) 924e4878e39SDag-Erling Smørgrav fetchFreeURL(url); 925e4878e39SDag-Erling Smørgrav if (fd != -1) 926e4878e39SDag-Erling Smørgrav close(fd); 927e4878e39SDag-Erling Smørgrav return NULL; 928e4878e39SDag-Erling Smørgrav } 929e4878e39SDag-Erling Smørgrav 930e4878e39SDag-Erling Smørgrav 931e4878e39SDag-Erling Smørgrav /***************************************************************************** 932e4878e39SDag-Erling Smørgrav * Entry points 933e4878e39SDag-Erling Smørgrav */ 934e4878e39SDag-Erling Smørgrav 93560245e42SDag-Erling Smørgrav /* 93660245e42SDag-Erling Smørgrav * Retrieve a file by HTTP 93760245e42SDag-Erling Smørgrav */ 93860245e42SDag-Erling Smørgrav FILE * 93960245e42SDag-Erling Smørgrav fetchGetHTTP(struct url *URL, char *flags) 94060245e42SDag-Erling Smørgrav { 941e4878e39SDag-Erling Smørgrav struct url_stat us; 94260245e42SDag-Erling Smørgrav 943e4878e39SDag-Erling Smørgrav return _http_request(URL, "GET", &us, flags); 9444ca1ab94SDag-Erling Smørgrav } 9454ca1ab94SDag-Erling Smørgrav 9464ca1ab94SDag-Erling Smørgrav FILE * 947d8acd8dcSDag-Erling Smørgrav fetchPutHTTP(struct url *URL, char *flags) 9484ca1ab94SDag-Erling Smørgrav { 9494ca1ab94SDag-Erling Smørgrav warnx("fetchPutHTTP(): not implemented"); 9504ca1ab94SDag-Erling Smørgrav return NULL; 9514ca1ab94SDag-Erling Smørgrav } 952d8acd8dcSDag-Erling Smørgrav 953d8acd8dcSDag-Erling Smørgrav /* 954d8acd8dcSDag-Erling Smørgrav * Get an HTTP document's metadata 955d8acd8dcSDag-Erling Smørgrav */ 956d8acd8dcSDag-Erling Smørgrav int 95760245e42SDag-Erling Smørgrav fetchStatHTTP(struct url *URL, struct url_stat *us, char *flags) 958d8acd8dcSDag-Erling Smørgrav { 95960245e42SDag-Erling Smørgrav FILE *f; 96060245e42SDag-Erling Smørgrav 961e4878e39SDag-Erling Smørgrav if ((f = _http_request(URL, "HEAD", us, flags)) == NULL) 96260245e42SDag-Erling Smørgrav return -1; 963f8f4130bSDag-Erling Smørgrav fclose(f); 96460245e42SDag-Erling Smørgrav return 0; 965d8acd8dcSDag-Erling Smørgrav } 966ce71b736SDag-Erling Smørgrav 967ce71b736SDag-Erling Smørgrav /* 968ce71b736SDag-Erling Smørgrav * List a directory 969ce71b736SDag-Erling Smørgrav */ 970ce71b736SDag-Erling Smørgrav struct url_ent * 971ce71b736SDag-Erling Smørgrav fetchListHTTP(struct url *url, char *flags) 972ce71b736SDag-Erling Smørgrav { 973ce71b736SDag-Erling Smørgrav warnx("fetchListHTTP(): not implemented"); 974ce71b736SDag-Erling Smørgrav return NULL; 975ce71b736SDag-Erling Smørgrav } 976