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 29 #include <sys/cdefs.h> 30 __FBSDID("$FreeBSD$"); 31 32 /* 33 * The following copyright applies to the base64 code: 34 * 35 *- 36 * Copyright 1997 Massachusetts Institute of Technology 37 * 38 * Permission to use, copy, modify, and distribute this software and 39 * its documentation for any purpose and without fee is hereby 40 * granted, provided that both the above copyright notice and this 41 * permission notice appear in all copies, that both the above 42 * copyright notice and this permission notice appear in all 43 * supporting documentation, and that the name of M.I.T. not be used 44 * in advertising or publicity pertaining to distribution of the 45 * software without specific, written prior permission. M.I.T. makes 46 * no representations about the suitability of this software for any 47 * purpose. It is provided "as is" without express or implied 48 * warranty. 49 * 50 * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''. M.I.T. DISCLAIMS 51 * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE, 52 * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF 53 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT 54 * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 55 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 56 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF 57 * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND 58 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 59 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 60 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 61 * SUCH DAMAGE. 62 */ 63 64 #include <sys/param.h> 65 #include <sys/socket.h> 66 67 #include <ctype.h> 68 #include <err.h> 69 #include <errno.h> 70 #include <locale.h> 71 #include <netdb.h> 72 #include <stdarg.h> 73 #include <stdio.h> 74 #include <stdlib.h> 75 #include <string.h> 76 #include <time.h> 77 #include <unistd.h> 78 79 #include "fetch.h" 80 #include "common.h" 81 #include "httperr.h" 82 83 extern char *__progname; /* XXX not portable */ 84 85 /* Maximum number of redirects to follow */ 86 #define MAX_REDIRECT 5 87 88 /* Symbolic names for reply codes we care about */ 89 #define HTTP_OK 200 90 #define HTTP_PARTIAL 206 91 #define HTTP_MOVED_PERM 301 92 #define HTTP_MOVED_TEMP 302 93 #define HTTP_SEE_OTHER 303 94 #define HTTP_NEED_AUTH 401 95 #define HTTP_NEED_PROXY_AUTH 403 96 #define HTTP_PROTOCOL_ERROR 999 97 98 #define HTTP_REDIRECT(xyz) ((xyz) == HTTP_MOVED_PERM \ 99 || (xyz) == HTTP_MOVED_TEMP \ 100 || (xyz) == HTTP_SEE_OTHER) 101 102 103 104 /***************************************************************************** 105 * I/O functions for decoding chunked streams 106 */ 107 108 struct cookie 109 { 110 int fd; 111 char *buf; 112 size_t b_size; 113 size_t b_len; 114 int b_pos; 115 int eof; 116 int error; 117 long chunksize; 118 #ifndef NDEBUG 119 long total; 120 #endif 121 }; 122 123 /* 124 * Get next chunk header 125 */ 126 static int 127 _http_new_chunk(struct cookie *c) 128 { 129 char *p; 130 131 if (_fetch_getln(c->fd, &c->buf, &c->b_size, &c->b_len) == -1) 132 return -1; 133 134 if (c->b_len < 2 || !ishexnumber(*c->buf)) 135 return -1; 136 137 for (p = c->buf; !isspace(*p) && *p != ';' && p < c->buf + c->b_len; ++p) 138 if (!ishexnumber(*p)) 139 return -1; 140 else if (isdigit(*p)) 141 c->chunksize = c->chunksize * 16 + *p - '0'; 142 else 143 c->chunksize = c->chunksize * 16 + 10 + tolower(*p) - 'a'; 144 145 #ifndef NDEBUG 146 c->total += c->chunksize; 147 if (c->chunksize == 0) 148 fprintf(stderr, "\033[1m_http_fillbuf(): " 149 "end of last chunk\033[m\n"); 150 else 151 fprintf(stderr, "\033[1m_http_fillbuf(): " 152 "new chunk: %ld (%ld)\033[m\n", c->chunksize, c->total); 153 #endif 154 155 return c->chunksize; 156 } 157 158 /* 159 * Fill the input buffer, do chunk decoding on the fly 160 */ 161 static int 162 _http_fillbuf(struct cookie *c) 163 { 164 if (c->error) 165 return -1; 166 if (c->eof) 167 return 0; 168 169 if (c->chunksize == 0) { 170 switch (_http_new_chunk(c)) { 171 case -1: 172 c->error = 1; 173 return -1; 174 case 0: 175 c->eof = 1; 176 return 0; 177 } 178 } 179 180 if (c->b_size < c->chunksize) { 181 char *tmp; 182 183 if ((tmp = realloc(c->buf, c->chunksize)) == NULL) 184 return -1; 185 c->buf = tmp; 186 c->b_size = c->chunksize; 187 } 188 189 if ((c->b_len = read(c->fd, c->buf, c->chunksize)) == -1) 190 return -1; 191 c->chunksize -= c->b_len; 192 193 if (c->chunksize == 0) { 194 char endl[2]; 195 read(c->fd, endl, 2); 196 } 197 198 c->b_pos = 0; 199 200 return c->b_len; 201 } 202 203 /* 204 * Read function 205 */ 206 static int 207 _http_readfn(void *v, char *buf, int len) 208 { 209 struct cookie *c = (struct cookie *)v; 210 int l, pos; 211 212 if (c->error) 213 return -1; 214 if (c->eof) 215 return 0; 216 217 for (pos = 0; len > 0; pos += l, len -= l) { 218 /* empty buffer */ 219 if (!c->buf || c->b_pos == c->b_len) 220 if (_http_fillbuf(c) < 1) 221 break; 222 l = c->b_len - c->b_pos; 223 if (len < l) 224 l = len; 225 bcopy(c->buf + c->b_pos, buf + pos, l); 226 c->b_pos += l; 227 } 228 229 if (!pos && c->error) 230 return -1; 231 return pos; 232 } 233 234 /* 235 * Write function 236 */ 237 static int 238 _http_writefn(void *v, const char *buf, int len) 239 { 240 struct cookie *c = (struct cookie *)v; 241 242 return write(c->fd, buf, len); 243 } 244 245 /* 246 * Close function 247 */ 248 static int 249 _http_closefn(void *v) 250 { 251 struct cookie *c = (struct cookie *)v; 252 int r; 253 254 r = close(c->fd); 255 if (c->buf) 256 free(c->buf); 257 free(c); 258 return r; 259 } 260 261 /* 262 * Wrap a file descriptor up 263 */ 264 static FILE * 265 _http_funopen(int fd) 266 { 267 struct cookie *c; 268 FILE *f; 269 270 if ((c = calloc(1, sizeof *c)) == NULL) { 271 _fetch_syserr(); 272 return NULL; 273 } 274 c->fd = fd; 275 if (!(f = funopen(c, _http_readfn, _http_writefn, NULL, _http_closefn))) { 276 _fetch_syserr(); 277 free(c); 278 return NULL; 279 } 280 return f; 281 } 282 283 284 /***************************************************************************** 285 * Helper functions for talking to the server and parsing its replies 286 */ 287 288 /* Header types */ 289 typedef enum { 290 hdr_syserror = -2, 291 hdr_error = -1, 292 hdr_end = 0, 293 hdr_unknown = 1, 294 hdr_content_length, 295 hdr_content_range, 296 hdr_last_modified, 297 hdr_location, 298 hdr_transfer_encoding, 299 hdr_www_authenticate 300 } hdr; 301 302 /* Names of interesting headers */ 303 static struct { 304 hdr num; 305 const char *name; 306 } hdr_names[] = { 307 { hdr_content_length, "Content-Length" }, 308 { hdr_content_range, "Content-Range" }, 309 { hdr_last_modified, "Last-Modified" }, 310 { hdr_location, "Location" }, 311 { hdr_transfer_encoding, "Transfer-Encoding" }, 312 { hdr_www_authenticate, "WWW-Authenticate" }, 313 { hdr_unknown, NULL }, 314 }; 315 316 static char *reply_buf; 317 static size_t reply_size; 318 static size_t reply_length; 319 320 /* 321 * Send a formatted line; optionally echo to terminal 322 */ 323 static int 324 _http_cmd(int fd, const char *fmt, ...) 325 { 326 va_list ap; 327 size_t len; 328 char *msg; 329 int r; 330 331 va_start(ap, fmt); 332 len = vasprintf(&msg, fmt, ap); 333 va_end(ap); 334 335 if (msg == NULL) { 336 errno = ENOMEM; 337 _fetch_syserr(); 338 return -1; 339 } 340 341 r = _fetch_putln(fd, msg, len); 342 free(msg); 343 344 if (r == -1) { 345 _fetch_syserr(); 346 return -1; 347 } 348 349 return 0; 350 } 351 352 /* 353 * Get and parse status line 354 */ 355 static int 356 _http_get_reply(int fd) 357 { 358 char *p; 359 360 if (_fetch_getln(fd, &reply_buf, &reply_size, &reply_length) == -1) 361 return -1; 362 /* 363 * A valid status line looks like "HTTP/m.n xyz reason" where m 364 * and n are the major and minor protocol version numbers and xyz 365 * is the reply code. 366 * Unfortunately, there are servers out there (NCSA 1.5.1, to name 367 * just one) that do not send a version number, so we can't rely 368 * on finding one, but if we do, insist on it being 1.0 or 1.1. 369 * We don't care about the reason phrase. 370 */ 371 if (strncmp(reply_buf, "HTTP", 4) != 0) 372 return HTTP_PROTOCOL_ERROR; 373 p = reply_buf + 4; 374 if (*p == '/') { 375 if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1')) 376 return HTTP_PROTOCOL_ERROR; 377 p += 4; 378 } 379 if (*p != ' ' 380 || !isdigit(p[1]) 381 || !isdigit(p[2]) 382 || !isdigit(p[3])) 383 return HTTP_PROTOCOL_ERROR; 384 385 return ((p[1] - '0') * 100 + (p[2] - '0') * 10 + (p[3] - '0')); 386 } 387 388 /* 389 * Check a header; if the type matches the given string, return a 390 * pointer to the beginning of the value. 391 */ 392 static const char * 393 _http_match(const char *str, const char *hdr) 394 { 395 while (*str && *hdr && tolower(*str++) == tolower(*hdr++)) 396 /* nothing */; 397 if (*str || *hdr != ':') 398 return NULL; 399 while (*hdr && isspace(*++hdr)) 400 /* nothing */; 401 return hdr; 402 } 403 404 /* 405 * Get the next header and return the appropriate symbolic code. 406 */ 407 static hdr 408 _http_next_header(int fd, const char **p) 409 { 410 int i; 411 412 if (_fetch_getln(fd, &reply_buf, &reply_size, &reply_length) == -1) 413 return hdr_syserror; 414 while (reply_length && isspace(reply_buf[reply_length-1])) 415 reply_length--; 416 reply_buf[reply_length] = 0; 417 if (reply_length == 0) 418 return hdr_end; 419 /* 420 * We could check for malformed headers but we don't really care. 421 * A valid header starts with a token immediately followed by a 422 * colon; a token is any sequence of non-control, non-whitespace 423 * characters except "()<>@,;:\\\"{}". 424 */ 425 for (i = 0; hdr_names[i].num != hdr_unknown; i++) 426 if ((*p = _http_match(hdr_names[i].name, reply_buf)) != NULL) 427 return hdr_names[i].num; 428 return hdr_unknown; 429 } 430 431 /* 432 * Parse a last-modified header 433 */ 434 static int 435 _http_parse_mtime(const char *p, time_t *mtime) 436 { 437 char locale[64], *r; 438 struct tm tm; 439 440 strncpy(locale, setlocale(LC_TIME, NULL), sizeof locale); 441 setlocale(LC_TIME, "C"); 442 r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm); 443 /* XXX should add support for date-2 and date-3 */ 444 setlocale(LC_TIME, locale); 445 if (r == NULL) 446 return -1; 447 DEBUG(fprintf(stderr, "last modified: [\033[1m%04d-%02d-%02d " 448 "%02d:%02d:%02d\033[m]\n", 449 tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday, 450 tm.tm_hour, tm.tm_min, tm.tm_sec)); 451 *mtime = timegm(&tm); 452 return 0; 453 } 454 455 /* 456 * Parse a content-length header 457 */ 458 static int 459 _http_parse_length(const char *p, off_t *length) 460 { 461 off_t len; 462 463 for (len = 0; *p && isdigit(*p); ++p) 464 len = len * 10 + (*p - '0'); 465 DEBUG(fprintf(stderr, "content length: [\033[1m%lld\033[m]\n", len)); 466 *length = len; 467 return 0; 468 } 469 470 /* 471 * Parse a content-range header 472 */ 473 static int 474 _http_parse_range(const char *p, off_t *offset, off_t *length, off_t *size) 475 { 476 int first, last, len; 477 478 if (strncasecmp(p, "bytes ", 6) != 0) 479 return -1; 480 for (first = 0, p += 6; *p && isdigit(*p); ++p) 481 first = first * 10 + *p - '0'; 482 if (*p != '-') 483 return -1; 484 for (last = 0, ++p; *p && isdigit(*p); ++p) 485 last = last * 10 + *p - '0'; 486 if (first > last || *p != '/') 487 return -1; 488 for (len = 0, ++p; *p && isdigit(*p); ++p) 489 len = len * 10 + *p - '0'; 490 if (len < last - first + 1) 491 return -1; 492 DEBUG(fprintf(stderr, "content range: [\033[1m%d-%d/%d\033[m]\n", 493 first, last, len)); 494 *offset = first; 495 *length = last - first + 1; 496 *size = len; 497 return 0; 498 } 499 500 501 /***************************************************************************** 502 * Helper functions for authorization 503 */ 504 505 /* 506 * Base64 encoding 507 */ 508 static char * 509 _http_base64(char *src) 510 { 511 static const char base64[] = 512 "ABCDEFGHIJKLMNOPQRSTUVWXYZ" 513 "abcdefghijklmnopqrstuvwxyz" 514 "0123456789+/"; 515 char *str, *dst; 516 size_t l; 517 int t, r; 518 519 l = strlen(src); 520 if ((str = malloc(((l + 2) / 3) * 4)) == NULL) 521 return NULL; 522 dst = str; 523 r = 0; 524 525 while (l >= 3) { 526 t = (src[0] << 16) | (src[1] << 8) | src[2]; 527 dst[0] = base64[(t >> 18) & 0x3f]; 528 dst[1] = base64[(t >> 12) & 0x3f]; 529 dst[2] = base64[(t >> 6) & 0x3f]; 530 dst[3] = base64[(t >> 0) & 0x3f]; 531 src += 3; l -= 3; 532 dst += 4; r += 4; 533 } 534 535 switch (l) { 536 case 2: 537 t = (src[0] << 16) | (src[1] << 8); 538 dst[0] = base64[(t >> 18) & 0x3f]; 539 dst[1] = base64[(t >> 12) & 0x3f]; 540 dst[2] = base64[(t >> 6) & 0x3f]; 541 dst[3] = '='; 542 dst += 4; 543 r += 4; 544 break; 545 case 1: 546 t = src[0] << 16; 547 dst[0] = base64[(t >> 18) & 0x3f]; 548 dst[1] = base64[(t >> 12) & 0x3f]; 549 dst[2] = dst[3] = '='; 550 dst += 4; 551 r += 4; 552 break; 553 case 0: 554 break; 555 } 556 557 *dst = 0; 558 return str; 559 } 560 561 /* 562 * Encode username and password 563 */ 564 static int 565 _http_basic_auth(int fd, const char *hdr, const char *usr, const char *pwd) 566 { 567 char *upw, *auth; 568 int r; 569 570 DEBUG(fprintf(stderr, "usr: [\033[1m%s\033[m]\n", usr)); 571 DEBUG(fprintf(stderr, "pwd: [\033[1m%s\033[m]\n", pwd)); 572 if (asprintf(&upw, "%s:%s", usr, pwd) == -1) 573 return -1; 574 auth = _http_base64(upw); 575 free(upw); 576 if (auth == NULL) 577 return -1; 578 r = _http_cmd(fd, "%s: Basic %s", hdr, auth); 579 free(auth); 580 return r; 581 } 582 583 /* 584 * Send an authorization header 585 */ 586 static int 587 _http_authorize(int fd, const char *hdr, const char *p) 588 { 589 /* basic authorization */ 590 if (strncasecmp(p, "basic:", 6) == 0) { 591 char *user, *pwd, *str; 592 int r; 593 594 /* skip realm */ 595 for (p += 6; *p && *p != ':'; ++p) 596 /* nothing */ ; 597 if (!*p || strchr(++p, ':') == NULL) 598 return -1; 599 if ((str = strdup(p)) == NULL) 600 return -1; /* XXX */ 601 user = str; 602 pwd = strchr(str, ':'); 603 *pwd++ = '\0'; 604 r = _http_basic_auth(fd, hdr, user, pwd); 605 free(str); 606 return r; 607 } 608 return -1; 609 } 610 611 612 /***************************************************************************** 613 * Helper functions for connecting to a server or proxy 614 */ 615 616 /* 617 * Connect to the correct HTTP server or proxy. 618 */ 619 static int 620 _http_connect(struct url *URL, struct url *purl, const char *flags) 621 { 622 int verbose; 623 int af, fd; 624 625 #ifdef INET6 626 af = AF_UNSPEC; 627 #else 628 af = AF_INET; 629 #endif 630 631 verbose = CHECK_FLAG('v'); 632 if (CHECK_FLAG('4')) 633 af = AF_INET; 634 #ifdef INET6 635 else if (CHECK_FLAG('6')) 636 af = AF_INET6; 637 #endif 638 639 if (purl) { 640 URL = purl; 641 } else if (strcasecmp(URL->scheme, SCHEME_FTP) == 0) { 642 /* can't talk http to an ftp server */ 643 /* XXX should set an error code */ 644 return -1; 645 } 646 647 if ((fd = _fetch_connect(URL->host, URL->port, af, verbose)) == -1) 648 /* _fetch_connect() has already set an error code */ 649 return -1; 650 return fd; 651 } 652 653 static struct url * 654 _http_get_proxy(void) 655 { 656 struct url *purl; 657 char *p; 658 659 if (((p = getenv("HTTP_PROXY")) || (p = getenv("http_proxy"))) && 660 (purl = fetchParseURL(p))) { 661 if (!*purl->scheme) 662 strcpy(purl->scheme, SCHEME_HTTP); 663 if (!purl->port) 664 purl->port = _fetch_default_proxy_port(purl->scheme); 665 if (strcasecmp(purl->scheme, SCHEME_HTTP) == 0) 666 return purl; 667 fetchFreeURL(purl); 668 } 669 return NULL; 670 } 671 672 673 /***************************************************************************** 674 * Core 675 */ 676 677 /* 678 * Send a request and process the reply 679 */ 680 FILE * 681 _http_request(struct url *URL, const char *op, struct url_stat *us, 682 struct url *purl, const char *flags) 683 { 684 struct url *url, *new; 685 int chunked, direct, need_auth, noredirect, verbose; 686 int code, fd, i, n; 687 off_t offset, clength, length, size; 688 time_t mtime; 689 const char *p; 690 FILE *f; 691 hdr h; 692 char *host; 693 #ifdef INET6 694 char hbuf[MAXHOSTNAMELEN + 1]; 695 #endif 696 697 direct = CHECK_FLAG('d'); 698 noredirect = CHECK_FLAG('A'); 699 verbose = CHECK_FLAG('v'); 700 701 if (direct && purl) { 702 fetchFreeURL(purl); 703 purl = NULL; 704 } 705 706 /* try the provided URL first */ 707 url = URL; 708 709 /* if the A flag is set, we only get one try */ 710 n = noredirect ? 1 : MAX_REDIRECT; 711 i = 0; 712 713 need_auth = 0; 714 do { 715 new = NULL; 716 chunked = 0; 717 offset = 0; 718 clength = -1; 719 length = -1; 720 size = -1; 721 mtime = 0; 722 723 /* check port */ 724 if (!url->port) 725 url->port = _fetch_default_port(url->scheme); 726 727 /* connect to server or proxy */ 728 if ((fd = _http_connect(url, purl, flags)) == -1) 729 goto ouch; 730 731 host = url->host; 732 #ifdef INET6 733 if (strchr(url->host, ':')) { 734 snprintf(hbuf, sizeof(hbuf), "[%s]", url->host); 735 host = hbuf; 736 } 737 #endif 738 739 /* send request */ 740 if (verbose) 741 _fetch_info("requesting %s://%s:%d%s", 742 url->scheme, host, url->port, url->doc); 743 if (purl) { 744 _http_cmd(fd, "%s %s://%s:%d%s HTTP/1.1", 745 op, url->scheme, host, url->port, url->doc); 746 } else { 747 _http_cmd(fd, "%s %s HTTP/1.1", 748 op, url->doc); 749 } 750 751 /* virtual host */ 752 if (url->port == _fetch_default_port(url->scheme)) 753 _http_cmd(fd, "Host: %s", host); 754 else 755 _http_cmd(fd, "Host: %s:%d", host, url->port); 756 757 /* proxy authorization */ 758 if (purl) { 759 if (*purl->user || *purl->pwd) 760 _http_basic_auth(fd, "Proxy-Authorization", 761 purl->user, purl->pwd); 762 else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL && *p != '\0') 763 _http_authorize(fd, "Proxy-Authorization", p); 764 } 765 766 /* server authorization */ 767 if (need_auth || *url->user || *url->pwd) { 768 if (*url->user || *url->pwd) 769 _http_basic_auth(fd, "Authorization", url->user, url->pwd); 770 else if ((p = getenv("HTTP_AUTH")) != NULL && *p != '\0') 771 _http_authorize(fd, "Authorization", p); 772 else if (fetchAuthMethod && fetchAuthMethod(url) == 0) { 773 _http_basic_auth(fd, "Authorization", url->user, url->pwd); 774 } else { 775 _http_seterr(HTTP_NEED_AUTH); 776 goto ouch; 777 } 778 } 779 780 /* other headers */ 781 if ((p = getenv("HTTP_USER_AGENT")) != NULL && *p != '\0') 782 _http_cmd(fd, "User-Agent: %s", p); 783 else 784 _http_cmd(fd, "User-Agent: %s " _LIBFETCH_VER, __progname); 785 if (url->offset) 786 _http_cmd(fd, "Range: bytes=%lld-", url->offset); 787 _http_cmd(fd, "Connection: close"); 788 _http_cmd(fd, ""); 789 790 /* get reply */ 791 switch ((code = _http_get_reply(fd))) { 792 case HTTP_OK: 793 case HTTP_PARTIAL: 794 /* fine */ 795 break; 796 case HTTP_MOVED_PERM: 797 case HTTP_MOVED_TEMP: 798 /* 799 * Not so fine, but we still have to read the headers to 800 * get the new location. 801 */ 802 break; 803 case HTTP_NEED_AUTH: 804 if (need_auth) { 805 /* 806 * We already sent out authorization code, so there's 807 * nothing more we can do. 808 */ 809 _http_seterr(code); 810 goto ouch; 811 } 812 /* try again, but send the password this time */ 813 if (verbose) 814 _fetch_info("server requires authorization"); 815 break; 816 case HTTP_NEED_PROXY_AUTH: 817 /* 818 * If we're talking to a proxy, we already sent our proxy 819 * authorization code, so there's nothing more we can do. 820 */ 821 _http_seterr(code); 822 goto ouch; 823 case HTTP_PROTOCOL_ERROR: 824 /* fall through */ 825 case -1: 826 _fetch_syserr(); 827 goto ouch; 828 default: 829 _http_seterr(code); 830 goto ouch; 831 } 832 833 /* get headers */ 834 do { 835 switch ((h = _http_next_header(fd, &p))) { 836 case hdr_syserror: 837 _fetch_syserr(); 838 goto ouch; 839 case hdr_error: 840 _http_seterr(HTTP_PROTOCOL_ERROR); 841 goto ouch; 842 case hdr_content_length: 843 _http_parse_length(p, &clength); 844 break; 845 case hdr_content_range: 846 _http_parse_range(p, &offset, &length, &size); 847 break; 848 case hdr_last_modified: 849 _http_parse_mtime(p, &mtime); 850 break; 851 case hdr_location: 852 if (!HTTP_REDIRECT(code)) 853 break; 854 if (new) 855 free(new); 856 if (verbose) 857 _fetch_info("%d redirect to %s", code, p); 858 if (*p == '/') 859 /* absolute path */ 860 new = fetchMakeURL(url->scheme, url->host, url->port, p, 861 url->user, url->pwd); 862 else 863 new = fetchParseURL(p); 864 if (new == NULL) { 865 /* XXX should set an error code */ 866 DEBUG(fprintf(stderr, "failed to parse new URL\n")); 867 goto ouch; 868 } 869 if (!*new->user && !*new->pwd) { 870 strcpy(new->user, url->user); 871 strcpy(new->pwd, url->pwd); 872 } 873 new->offset = url->offset; 874 new->length = url->length; 875 break; 876 case hdr_transfer_encoding: 877 /* XXX weak test*/ 878 chunked = (strcasecmp(p, "chunked") == 0); 879 break; 880 case hdr_www_authenticate: 881 if (code != HTTP_NEED_AUTH) 882 break; 883 /* if we were smarter, we'd check the method and realm */ 884 break; 885 case hdr_end: 886 /* fall through */ 887 case hdr_unknown: 888 /* ignore */ 889 break; 890 } 891 } while (h > hdr_end); 892 893 /* we have a hit */ 894 if (code == HTTP_OK || code == HTTP_PARTIAL) 895 break; 896 897 /* we need to provide authentication */ 898 if (code == HTTP_NEED_AUTH) { 899 need_auth = 1; 900 close(fd); 901 fd = -1; 902 continue; 903 } 904 905 /* all other cases: we got a redirect */ 906 need_auth = 0; 907 close(fd); 908 fd = -1; 909 if (!new) { 910 DEBUG(fprintf(stderr, "redirect with no new location\n")); 911 break; 912 } 913 if (url != URL) 914 fetchFreeURL(url); 915 url = new; 916 } while (++i < n); 917 918 /* we failed, or ran out of retries */ 919 if (fd == -1) { 920 _http_seterr(code); 921 goto ouch; 922 } 923 924 DEBUG(fprintf(stderr, "offset %lld, length %lld, size %lld, clength %lld\n", 925 offset, length, size, clength)); 926 927 /* check for inconsistencies */ 928 if (clength != -1 && length != -1 && clength != length) { 929 _http_seterr(HTTP_PROTOCOL_ERROR); 930 goto ouch; 931 } 932 if (clength == -1) 933 clength = length; 934 if (clength != -1) 935 length = offset + clength; 936 if (length != -1 && size != -1 && length != size) { 937 _http_seterr(HTTP_PROTOCOL_ERROR); 938 goto ouch; 939 } 940 if (size == -1) 941 size = length; 942 943 /* fill in stats */ 944 if (us) { 945 us->size = size; 946 us->atime = us->mtime = mtime; 947 } 948 949 /* too far? */ 950 if (offset > URL->offset) { 951 _http_seterr(HTTP_PROTOCOL_ERROR); 952 goto ouch; 953 } 954 955 /* report back real offset and size */ 956 URL->offset = offset; 957 URL->length = clength; 958 959 /* wrap it up in a FILE */ 960 if ((f = chunked ? _http_funopen(fd) : fdopen(fd, "r")) == NULL) { 961 _fetch_syserr(); 962 goto ouch; 963 } 964 965 if (url != URL) 966 fetchFreeURL(url); 967 if (purl) 968 fetchFreeURL(purl); 969 970 return f; 971 972 ouch: 973 if (url != URL) 974 fetchFreeURL(url); 975 if (purl) 976 fetchFreeURL(purl); 977 if (fd != -1) 978 close(fd); 979 return NULL; 980 } 981 982 983 /***************************************************************************** 984 * Entry points 985 */ 986 987 /* 988 * Retrieve and stat a file by HTTP 989 */ 990 FILE * 991 fetchXGetHTTP(struct url *URL, struct url_stat *us, const char *flags) 992 { 993 return _http_request(URL, "GET", us, _http_get_proxy(), flags); 994 } 995 996 /* 997 * Retrieve a file by HTTP 998 */ 999 FILE * 1000 fetchGetHTTP(struct url *URL, const char *flags) 1001 { 1002 return fetchXGetHTTP(URL, NULL, flags); 1003 } 1004 1005 /* 1006 * Store a file by HTTP 1007 */ 1008 FILE * 1009 fetchPutHTTP(struct url *URL, const char *flags) 1010 { 1011 warnx("fetchPutHTTP(): not implemented"); 1012 return NULL; 1013 } 1014 1015 /* 1016 * Get an HTTP document's metadata 1017 */ 1018 int 1019 fetchStatHTTP(struct url *URL, struct url_stat *us, const char *flags) 1020 { 1021 FILE *f; 1022 1023 if ((f = _http_request(URL, "HEAD", us, _http_get_proxy(), flags)) == NULL) 1024 return -1; 1025 fclose(f); 1026 return 0; 1027 } 1028 1029 /* 1030 * List a directory 1031 */ 1032 struct url_ent * 1033 fetchListHTTP(struct url *url, const char *flags) 1034 { 1035 warnx("fetchListHTTP(): not implemented"); 1036 return NULL; 1037 } 1038