1 /*- 2 * Copyright (c) 2000 Dag-Erling Co�dan Sm�rgrav 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer 10 * in this position and unchanged. 11 * 2. Redistributions in binary form must reproduce the above copyright 12 * notice, this list of conditions and the following disclaimer in the 13 * documentation and/or other materials provided with the distribution. 14 * 3. The name of the author may not be used to endorse or promote products 15 * derived from this software without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 22 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 23 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 24 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 25 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 26 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 27 * 28 * $FreeBSD$ 29 */ 30 31 /* 32 * The following copyright applies to the base64 code: 33 * 34 *- 35 * Copyright 1997 Massachusetts Institute of Technology 36 * 37 * Permission to use, copy, modify, and distribute this software and 38 * its documentation for any purpose and without fee is hereby 39 * granted, provided that both the above copyright notice and this 40 * permission notice appear in all copies, that both the above 41 * copyright notice and this permission notice appear in all 42 * supporting documentation, and that the name of M.I.T. not be used 43 * in advertising or publicity pertaining to distribution of the 44 * software without specific, written prior permission. M.I.T. makes 45 * no representations about the suitability of this software for any 46 * purpose. It is provided "as is" without express or implied 47 * warranty. 48 * 49 * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''. M.I.T. DISCLAIMS 50 * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE, 51 * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF 52 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT 53 * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 54 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 55 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF 56 * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND 57 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 58 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 59 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 60 * SUCH DAMAGE. 61 */ 62 63 #include <sys/param.h> 64 #include <sys/socket.h> 65 66 #include <ctype.h> 67 #include <err.h> 68 #include <errno.h> 69 #include <locale.h> 70 #include <netdb.h> 71 #include <stdarg.h> 72 #include <stdio.h> 73 #include <stdlib.h> 74 #include <string.h> 75 #include <time.h> 76 #include <unistd.h> 77 78 #include "fetch.h" 79 #include "common.h" 80 #include "httperr.h" 81 82 extern char *__progname; /* XXX not portable */ 83 84 /* Maximum number of redirects to follow */ 85 #define MAX_REDIRECT 5 86 87 /* Symbolic names for reply codes we care about */ 88 #define HTTP_OK 200 89 #define HTTP_PARTIAL 206 90 #define HTTP_MOVED_PERM 301 91 #define HTTP_MOVED_TEMP 302 92 #define HTTP_SEE_OTHER 303 93 #define HTTP_NEED_AUTH 401 94 #define HTTP_NEED_PROXY_AUTH 403 95 #define HTTP_PROTOCOL_ERROR 999 96 97 #define HTTP_REDIRECT(xyz) ((xyz) == HTTP_MOVED_PERM \ 98 || (xyz) == HTTP_MOVED_TEMP \ 99 || (xyz) == HTTP_SEE_OTHER) 100 101 102 103 /***************************************************************************** 104 * I/O functions for decoding chunked streams 105 */ 106 107 struct cookie 108 { 109 int fd; 110 char *buf; 111 size_t b_size; 112 size_t b_len; 113 int b_pos; 114 int eof; 115 int error; 116 long chunksize; 117 #ifndef NDEBUG 118 long total; 119 #endif 120 }; 121 122 /* 123 * Get next chunk header 124 */ 125 static int 126 _http_new_chunk(struct cookie *c) 127 { 128 char *p; 129 130 if (_fetch_getln(c->fd, &c->buf, &c->b_size, &c->b_len) == -1) 131 return -1; 132 133 if (c->b_len < 2 || !ishexnumber(*c->buf)) 134 return -1; 135 136 for (p = c->buf; !isspace(*p) && *p != ';' && p < c->buf + c->b_len; ++p) 137 if (!ishexnumber(*p)) 138 return -1; 139 else if (isdigit(*p)) 140 c->chunksize = c->chunksize * 16 + *p - '0'; 141 else 142 c->chunksize = c->chunksize * 16 + 10 + tolower(*p) - 'a'; 143 144 #ifndef NDEBUG 145 c->total += c->chunksize; 146 if (c->chunksize == 0) 147 fprintf(stderr, "\033[1m_http_fillbuf(): " 148 "end of last chunk\033[m\n"); 149 else 150 fprintf(stderr, "\033[1m_http_fillbuf(): " 151 "new chunk: %ld (%ld)\033[m\n", c->chunksize, c->total); 152 #endif 153 154 return c->chunksize; 155 } 156 157 /* 158 * Fill the input buffer, do chunk decoding on the fly 159 */ 160 static int 161 _http_fillbuf(struct cookie *c) 162 { 163 if (c->error) 164 return -1; 165 if (c->eof) 166 return 0; 167 168 if (c->chunksize == 0) { 169 switch (_http_new_chunk(c)) { 170 case -1: 171 c->error = 1; 172 return -1; 173 case 0: 174 c->eof = 1; 175 return 0; 176 } 177 } 178 179 if (c->b_size < c->chunksize) { 180 char *tmp; 181 182 if ((tmp = realloc(c->buf, c->chunksize)) == NULL) 183 return -1; 184 c->buf = tmp; 185 c->b_size = c->chunksize; 186 } 187 188 if ((c->b_len = read(c->fd, c->buf, c->chunksize)) == -1) 189 return -1; 190 c->chunksize -= c->b_len; 191 192 if (c->chunksize == 0) { 193 char endl[2]; 194 read(c->fd, endl, 2); 195 } 196 197 c->b_pos = 0; 198 199 return c->b_len; 200 } 201 202 /* 203 * Read function 204 */ 205 static int 206 _http_readfn(void *v, char *buf, int len) 207 { 208 struct cookie *c = (struct cookie *)v; 209 int l, pos; 210 211 if (c->error) 212 return -1; 213 if (c->eof) 214 return 0; 215 216 for (pos = 0; len > 0; pos += l, len -= l) { 217 /* empty buffer */ 218 if (!c->buf || c->b_pos == c->b_len) 219 if (_http_fillbuf(c) < 1) 220 break; 221 l = c->b_len - c->b_pos; 222 if (len < l) 223 l = len; 224 bcopy(c->buf + c->b_pos, buf + pos, l); 225 c->b_pos += l; 226 } 227 228 if (!pos && c->error) 229 return -1; 230 return pos; 231 } 232 233 /* 234 * Write function 235 */ 236 static int 237 _http_writefn(void *v, const char *buf, int len) 238 { 239 struct cookie *c = (struct cookie *)v; 240 241 return write(c->fd, buf, len); 242 } 243 244 /* 245 * Close function 246 */ 247 static int 248 _http_closefn(void *v) 249 { 250 struct cookie *c = (struct cookie *)v; 251 int r; 252 253 r = close(c->fd); 254 if (c->buf) 255 free(c->buf); 256 free(c); 257 return r; 258 } 259 260 /* 261 * Wrap a file descriptor up 262 */ 263 static FILE * 264 _http_funopen(int fd) 265 { 266 struct cookie *c; 267 FILE *f; 268 269 if ((c = calloc(1, sizeof *c)) == NULL) { 270 _fetch_syserr(); 271 return NULL; 272 } 273 c->fd = fd; 274 if (!(f = funopen(c, _http_readfn, _http_writefn, NULL, _http_closefn))) { 275 _fetch_syserr(); 276 free(c); 277 return NULL; 278 } 279 return f; 280 } 281 282 283 /***************************************************************************** 284 * Helper functions for talking to the server and parsing its replies 285 */ 286 287 /* Header types */ 288 typedef enum { 289 hdr_syserror = -2, 290 hdr_error = -1, 291 hdr_end = 0, 292 hdr_unknown = 1, 293 hdr_content_length, 294 hdr_content_range, 295 hdr_last_modified, 296 hdr_location, 297 hdr_transfer_encoding 298 } hdr; 299 300 /* Names of interesting headers */ 301 static struct { 302 hdr num; 303 char *name; 304 } hdr_names[] = { 305 { hdr_content_length, "Content-Length" }, 306 { hdr_content_range, "Content-Range" }, 307 { hdr_last_modified, "Last-Modified" }, 308 { hdr_location, "Location" }, 309 { hdr_transfer_encoding, "Transfer-Encoding" }, 310 { hdr_unknown, NULL }, 311 }; 312 313 static char *reply_buf; 314 static size_t reply_size; 315 static size_t reply_length; 316 317 /* 318 * Send a formatted line; optionally echo to terminal 319 */ 320 static int 321 _http_cmd(int fd, char *fmt, ...) 322 { 323 va_list ap; 324 size_t len; 325 char *msg; 326 int r; 327 328 va_start(ap, fmt); 329 len = vasprintf(&msg, fmt, ap); 330 va_end(ap); 331 332 if (msg == NULL) { 333 errno = ENOMEM; 334 _fetch_syserr(); 335 return -1; 336 } 337 338 r = _fetch_putln(fd, msg, len); 339 free(msg); 340 341 if (r == -1) { 342 _fetch_syserr(); 343 return -1; 344 } 345 346 return 0; 347 } 348 349 /* 350 * Get and parse status line 351 */ 352 static int 353 _http_get_reply(int fd) 354 { 355 char *p; 356 357 if (_fetch_getln(fd, &reply_buf, &reply_size, &reply_length) == -1) 358 return -1; 359 /* 360 * A valid status line looks like "HTTP/m.n xyz reason" where m 361 * and n are the major and minor protocol version numbers and xyz 362 * is the reply code. 363 * Unfortunately, there are servers out there (NCSA 1.5.1, to name 364 * just one) that do not send a version number, so we can't rely 365 * on finding one, but if we do, insist on it being 1.0 or 1.1. 366 * We don't care about the reason phrase. 367 */ 368 if (strncmp(reply_buf, "HTTP", 4) != 0) 369 return HTTP_PROTOCOL_ERROR; 370 p = reply_buf + 4; 371 if (*p == '/') { 372 if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1')) 373 return HTTP_PROTOCOL_ERROR; 374 p += 4; 375 } 376 if (*p != ' ' 377 || !isdigit(p[1]) 378 || !isdigit(p[2]) 379 || !isdigit(p[3])) 380 return HTTP_PROTOCOL_ERROR; 381 382 return ((p[1] - '0') * 100 + (p[2] - '0') * 10 + (p[3] - '0')); 383 } 384 385 /* 386 * Check a header; if the type matches the given string, return a 387 * pointer to the beginning of the value. 388 */ 389 static char * 390 _http_match(char *str, char *hdr) 391 { 392 while (*str && *hdr && tolower(*str++) == tolower(*hdr++)) 393 /* nothing */; 394 if (*str || *hdr != ':') 395 return NULL; 396 while (*hdr && isspace(*++hdr)) 397 /* nothing */; 398 return hdr; 399 } 400 401 /* 402 * Get the next header and return the appropriate symbolic code. 403 */ 404 static hdr 405 _http_next_header(int fd, char **p) 406 { 407 int i; 408 409 if (_fetch_getln(fd, &reply_buf, &reply_size, &reply_length) == -1) 410 return hdr_syserror; 411 while (reply_length && isspace(reply_buf[reply_length-1])) 412 reply_length--; 413 reply_buf[reply_length] = 0; 414 if (reply_length == 0) 415 return hdr_end; 416 /* 417 * We could check for malformed headers but we don't really care. 418 * A valid header starts with a token immediately followed by a 419 * colon; a token is any sequence of non-control, non-whitespace 420 * characters except "()<>@,;:\\\"{}". 421 */ 422 for (i = 0; hdr_names[i].num != hdr_unknown; i++) 423 if ((*p = _http_match(hdr_names[i].name, reply_buf)) != NULL) 424 return hdr_names[i].num; 425 return hdr_unknown; 426 } 427 428 /* 429 * Parse a last-modified header 430 */ 431 static int 432 _http_parse_mtime(char *p, time_t *mtime) 433 { 434 char locale[64], *r; 435 struct tm tm; 436 437 strncpy(locale, setlocale(LC_TIME, NULL), sizeof locale); 438 setlocale(LC_TIME, "C"); 439 r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm); 440 /* XXX should add support for date-2 and date-3 */ 441 setlocale(LC_TIME, locale); 442 if (r == NULL) 443 return -1; 444 DEBUG(fprintf(stderr, "last modified: [\033[1m%04d-%02d-%02d " 445 "%02d:%02d:%02d\033[m]\n", 446 tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday, 447 tm.tm_hour, tm.tm_min, tm.tm_sec)); 448 *mtime = timegm(&tm); 449 return 0; 450 } 451 452 /* 453 * Parse a content-length header 454 */ 455 static int 456 _http_parse_length(char *p, off_t *length) 457 { 458 off_t len; 459 460 for (len = 0; *p && isdigit(*p); ++p) 461 len = len * 10 + (*p - '0'); 462 DEBUG(fprintf(stderr, "content length: [\033[1m%lld\033[m]\n", len)); 463 *length = len; 464 return 0; 465 } 466 467 /* 468 * Parse a content-range header 469 */ 470 static int 471 _http_parse_range(char *p, off_t *offset, off_t *length, off_t *size) 472 { 473 int first, last, len; 474 475 if (strncasecmp(p, "bytes ", 6) != 0) 476 return -1; 477 for (first = 0, p += 6; *p && isdigit(*p); ++p) 478 first = first * 10 + *p - '0'; 479 if (*p != '-') 480 return -1; 481 for (last = 0, ++p; *p && isdigit(*p); ++p) 482 last = last * 10 + *p - '0'; 483 if (first > last || *p != '/') 484 return -1; 485 for (len = 0, ++p; *p && isdigit(*p); ++p) 486 len = len * 10 + *p - '0'; 487 if (len < last - first + 1) 488 return -1; 489 DEBUG(fprintf(stderr, "content range: [\033[1m%d-%d/%d\033[m]\n", 490 first, last, len)); 491 *offset = first; 492 *length = last - first + 1; 493 *size = len; 494 return 0; 495 } 496 497 498 /***************************************************************************** 499 * Helper functions for authorization 500 */ 501 502 /* 503 * Base64 encoding 504 */ 505 static char * 506 _http_base64(char *src) 507 { 508 static const char base64[] = 509 "ABCDEFGHIJKLMNOPQRSTUVWXYZ" 510 "abcdefghijklmnopqrstuvwxyz" 511 "0123456789+/"; 512 char *str, *dst; 513 size_t l; 514 int t, r; 515 516 l = strlen(src); 517 if ((str = malloc(((l + 2) / 3) * 4)) == NULL) 518 return NULL; 519 dst = str; 520 r = 0; 521 522 while (l >= 3) { 523 t = (src[0] << 16) | (src[1] << 8) | src[2]; 524 dst[0] = base64[(t >> 18) & 0x3f]; 525 dst[1] = base64[(t >> 12) & 0x3f]; 526 dst[2] = base64[(t >> 6) & 0x3f]; 527 dst[3] = base64[(t >> 0) & 0x3f]; 528 src += 3; l -= 3; 529 dst += 4; r += 4; 530 } 531 532 switch (l) { 533 case 2: 534 t = (src[0] << 16) | (src[1] << 8); 535 dst[0] = base64[(t >> 18) & 0x3f]; 536 dst[1] = base64[(t >> 12) & 0x3f]; 537 dst[2] = base64[(t >> 6) & 0x3f]; 538 dst[3] = '='; 539 dst += 4; 540 r += 4; 541 break; 542 case 1: 543 t = src[0] << 16; 544 dst[0] = base64[(t >> 18) & 0x3f]; 545 dst[1] = base64[(t >> 12) & 0x3f]; 546 dst[2] = dst[3] = '='; 547 dst += 4; 548 r += 4; 549 break; 550 case 0: 551 break; 552 } 553 554 *dst = 0; 555 return str; 556 } 557 558 /* 559 * Encode username and password 560 */ 561 static int 562 _http_basic_auth(int fd, char *hdr, char *usr, char *pwd) 563 { 564 char *upw, *auth; 565 int r; 566 567 if (asprintf(&upw, "%s:%s", usr, pwd) == -1) 568 return -1; 569 auth = _http_base64(upw); 570 free(upw); 571 if (auth == NULL) 572 return -1; 573 r = _http_cmd(fd, "%s: Basic %s", hdr, auth); 574 free(auth); 575 return r; 576 } 577 578 /* 579 * Send an authorization header 580 */ 581 static int 582 _http_authorize(int fd, char *hdr, char *p) 583 { 584 /* basic authorization */ 585 if (strncasecmp(p, "basic:", 6) == 0) { 586 char *user, *pwd, *str; 587 int r; 588 589 /* skip realm */ 590 for (p += 6; *p && *p != ':'; ++p) 591 /* nothing */ ; 592 if (!*p || strchr(++p, ':') == NULL) 593 return -1; 594 if ((str = strdup(p)) == NULL) 595 return -1; /* XXX */ 596 user = str; 597 pwd = strchr(str, ':'); 598 *pwd++ = '\0'; 599 r = _http_basic_auth(fd, hdr, user, pwd); 600 free(str); 601 return r; 602 } 603 return -1; 604 } 605 606 607 /***************************************************************************** 608 * Helper functions for connecting to a server or proxy 609 */ 610 611 /* 612 * Connect to the correct HTTP server or proxy. 613 */ 614 static int 615 _http_connect(struct url *URL, struct url *purl, char *flags) 616 { 617 int verbose; 618 int af, fd; 619 620 #ifdef INET6 621 af = AF_UNSPEC; 622 #else 623 af = AF_INET; 624 #endif 625 626 verbose = CHECK_FLAG('v'); 627 if (CHECK_FLAG('4')) 628 af = AF_INET; 629 #ifdef INET6 630 else if (CHECK_FLAG('6')) 631 af = AF_INET6; 632 #endif 633 634 if (purl) { 635 URL = purl; 636 } else if (strcasecmp(URL->scheme, SCHEME_FTP) == 0) { 637 /* can't talk http to an ftp server */ 638 /* XXX should set an error code */ 639 return -1; 640 } 641 642 if ((fd = _fetch_connect(URL->host, URL->port, af, verbose)) == -1) 643 /* _fetch_connect() has already set an error code */ 644 return -1; 645 return fd; 646 } 647 648 static struct url * 649 _http_get_proxy() 650 { 651 struct url *purl; 652 char *p; 653 654 if (((p = getenv("HTTP_PROXY")) || (p = getenv("http_proxy"))) && 655 (purl = fetchParseURL(p))) { 656 if (!*purl->scheme) 657 strcpy(purl->scheme, SCHEME_HTTP); 658 if (!purl->port) 659 purl->port = _fetch_default_proxy_port(purl->scheme); 660 if (strcasecmp(purl->scheme, SCHEME_HTTP) == 0) 661 return purl; 662 fetchFreeURL(purl); 663 } 664 return NULL; 665 } 666 667 668 /***************************************************************************** 669 * Core 670 */ 671 672 /* 673 * Send a request and process the reply 674 */ 675 FILE * 676 _http_request(struct url *URL, char *op, struct url_stat *us, 677 struct url *purl, char *flags) 678 { 679 struct url *url, *new; 680 int chunked, direct, need_auth, noredirect, verbose; 681 int code, fd, i, n; 682 off_t offset, clength, length, size; 683 time_t mtime; 684 char *p; 685 FILE *f; 686 hdr h; 687 char *host; 688 #ifdef INET6 689 char hbuf[MAXHOSTNAMELEN + 1]; 690 #endif 691 692 direct = CHECK_FLAG('d'); 693 noredirect = CHECK_FLAG('A'); 694 verbose = CHECK_FLAG('v'); 695 696 if (direct && purl) { 697 fetchFreeURL(purl); 698 purl = NULL; 699 } 700 701 /* try the provided URL first */ 702 url = URL; 703 704 /* if the A flag is set, we only get one try */ 705 n = noredirect ? 1 : MAX_REDIRECT; 706 i = 0; 707 708 do { 709 new = NULL; 710 chunked = 0; 711 need_auth = 0; 712 offset = 0; 713 clength = -1; 714 length = -1; 715 size = -1; 716 mtime = 0; 717 retry: 718 /* check port */ 719 if (!url->port) 720 url->port = _fetch_default_port(url->scheme); 721 722 /* connect to server or proxy */ 723 if ((fd = _http_connect(url, purl, flags)) == -1) 724 goto ouch; 725 726 host = url->host; 727 #ifdef INET6 728 if (strchr(url->host, ':')) { 729 snprintf(hbuf, sizeof(hbuf), "[%s]", url->host); 730 host = hbuf; 731 } 732 #endif 733 734 /* send request */ 735 if (verbose) 736 _fetch_info("requesting %s://%s:%d%s", 737 url->scheme, host, url->port, url->doc); 738 if (purl) { 739 _http_cmd(fd, "%s %s://%s:%d%s HTTP/1.1", 740 op, url->scheme, host, url->port, url->doc); 741 } else { 742 _http_cmd(fd, "%s %s HTTP/1.1", 743 op, url->doc); 744 } 745 746 /* proxy authorization */ 747 if (purl) { 748 if (*purl->user || *purl->pwd) 749 _http_basic_auth(fd, "Proxy-Authorization", 750 purl->user, purl->pwd); 751 else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL && *p != '\0') 752 _http_authorize(fd, "Proxy-Authorization", p); 753 } 754 755 /* server authorization */ 756 if (need_auth) { 757 if (*url->user || *url->pwd) 758 _http_basic_auth(fd, "Authorization", url->user, url->pwd); 759 else if ((p = getenv("HTTP_AUTH")) != NULL && *p != '\0') 760 _http_authorize(fd, "Authorization", p); 761 else { 762 _http_seterr(HTTP_NEED_AUTH); 763 goto ouch; 764 } 765 } 766 767 /* other headers */ 768 if (url->port == _fetch_default_port(url->scheme)) 769 _http_cmd(fd, "Host: %s", host); 770 else 771 _http_cmd(fd, "Host: %s:%d", host, url->port); 772 _http_cmd(fd, "User-Agent: %s " _LIBFETCH_VER, __progname); 773 if (url->offset) 774 _http_cmd(fd, "Range: bytes=%lld-", url->offset); 775 _http_cmd(fd, "Connection: close"); 776 _http_cmd(fd, ""); 777 778 /* get reply */ 779 switch ((code = _http_get_reply(fd))) { 780 case HTTP_OK: 781 case HTTP_PARTIAL: 782 /* fine */ 783 break; 784 case HTTP_MOVED_PERM: 785 case HTTP_MOVED_TEMP: 786 /* 787 * Not so fine, but we still have to read the headers to 788 * get the new location. 789 */ 790 break; 791 case HTTP_NEED_AUTH: 792 if (need_auth) { 793 /* 794 * We already sent out authorization code, so there's 795 * nothing more we can do. 796 */ 797 _http_seterr(code); 798 goto ouch; 799 } 800 /* try again, but send the password this time */ 801 if (verbose) 802 _fetch_info("server requires authorization"); 803 need_auth = 1; 804 close(fd); 805 goto retry; 806 case HTTP_NEED_PROXY_AUTH: 807 /* 808 * If we're talking to a proxy, we already sent our proxy 809 * authorization code, so there's nothing more we can do. 810 */ 811 _http_seterr(code); 812 goto ouch; 813 case HTTP_PROTOCOL_ERROR: 814 /* fall through */ 815 case -1: 816 _fetch_syserr(); 817 goto ouch; 818 default: 819 _http_seterr(code); 820 goto ouch; 821 } 822 823 /* get headers */ 824 do { 825 switch ((h = _http_next_header(fd, &p))) { 826 case hdr_syserror: 827 _fetch_syserr(); 828 goto ouch; 829 case hdr_error: 830 _http_seterr(HTTP_PROTOCOL_ERROR); 831 goto ouch; 832 case hdr_content_length: 833 _http_parse_length(p, &clength); 834 break; 835 case hdr_content_range: 836 _http_parse_range(p, &offset, &length, &size); 837 break; 838 case hdr_last_modified: 839 _http_parse_mtime(p, &mtime); 840 break; 841 case hdr_location: 842 if (!HTTP_REDIRECT(code)) 843 break; 844 if (new) 845 free(new); 846 if (verbose) 847 _fetch_info("%d redirect to %s", code, p); 848 if (*p == '/') 849 /* absolute path */ 850 new = fetchMakeURL(url->scheme, url->host, url->port, p, 851 url->user, url->pwd); 852 else 853 new = fetchParseURL(p); 854 if (new == NULL) { 855 /* XXX should set an error code */ 856 DEBUG(fprintf(stderr, "failed to parse new URL\n")); 857 goto ouch; 858 } 859 if (!*new->user && !*new->pwd) { 860 strcpy(new->user, url->user); 861 strcpy(new->pwd, url->pwd); 862 } 863 new->offset = url->offset; 864 new->length = url->length; 865 break; 866 case hdr_transfer_encoding: 867 /* XXX weak test*/ 868 chunked = (strcasecmp(p, "chunked") == 0); 869 break; 870 case hdr_end: 871 /* fall through */ 872 case hdr_unknown: 873 /* ignore */ 874 break; 875 } 876 } while (h > hdr_end); 877 878 /* we either have a hit, or a redirect with no Location: header */ 879 if (code == HTTP_OK || code == HTTP_PARTIAL || !new) 880 break; 881 882 /* we have a redirect */ 883 close(fd); 884 fd = -1; 885 if (url != URL) 886 fetchFreeURL(url); 887 url = new; 888 } while (++i < n); 889 890 /* no success */ 891 if (fd == -1) { 892 _http_seterr(code); 893 goto ouch; 894 } 895 896 DEBUG(fprintf(stderr, "offset %lld, length %lld, size %lld, clength %lld\n", 897 offset, length, size, clength)); 898 899 /* check for inconsistencies */ 900 if (clength != -1 && length != -1 && clength != length) { 901 _http_seterr(HTTP_PROTOCOL_ERROR); 902 goto ouch; 903 } 904 if (clength == -1) 905 clength = length; 906 if (clength != -1) 907 length = offset + clength; 908 if (length != -1 && size != -1 && length != size) { 909 _http_seterr(HTTP_PROTOCOL_ERROR); 910 goto ouch; 911 } 912 if (size == -1) 913 size = length; 914 915 /* fill in stats */ 916 if (us) { 917 us->size = size; 918 us->atime = us->mtime = mtime; 919 } 920 921 /* too far? */ 922 if (offset > URL->offset) { 923 _http_seterr(HTTP_PROTOCOL_ERROR); 924 goto ouch; 925 } 926 927 /* report back real offset and size */ 928 URL->offset = offset; 929 URL->length = clength; 930 931 /* wrap it up in a FILE */ 932 if ((f = chunked ? _http_funopen(fd) : fdopen(fd, "r")) == NULL) { 933 _fetch_syserr(); 934 goto ouch; 935 } 936 937 if (url != URL) 938 fetchFreeURL(url); 939 if (purl) 940 fetchFreeURL(purl); 941 942 return f; 943 944 ouch: 945 if (url != URL) 946 fetchFreeURL(url); 947 if (purl) 948 fetchFreeURL(purl); 949 if (fd != -1) 950 close(fd); 951 return NULL; 952 } 953 954 955 /***************************************************************************** 956 * Entry points 957 */ 958 959 /* 960 * Retrieve and stat a file by HTTP 961 */ 962 FILE * 963 fetchXGetHTTP(struct url *URL, struct url_stat *us, char *flags) 964 { 965 return _http_request(URL, "GET", us, _http_get_proxy(), flags); 966 } 967 968 /* 969 * Retrieve a file by HTTP 970 */ 971 FILE * 972 fetchGetHTTP(struct url *URL, char *flags) 973 { 974 return fetchXGetHTTP(URL, NULL, flags); 975 } 976 977 /* 978 * Store a file by HTTP 979 */ 980 FILE * 981 fetchPutHTTP(struct url *URL, char *flags) 982 { 983 warnx("fetchPutHTTP(): not implemented"); 984 return NULL; 985 } 986 987 /* 988 * Get an HTTP document's metadata 989 */ 990 int 991 fetchStatHTTP(struct url *URL, struct url_stat *us, char *flags) 992 { 993 FILE *f; 994 995 if ((f = _http_request(URL, "HEAD", us, _http_get_proxy(), flags)) == NULL) 996 return -1; 997 fclose(f); 998 return 0; 999 } 1000 1001 /* 1002 * List a directory 1003 */ 1004 struct url_ent * 1005 fetchListHTTP(struct url *url, char *flags) 1006 { 1007 warnx("fetchListHTTP(): not implemented"); 1008 return NULL; 1009 } 1010