http.c revision 64129
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: head/lib/libfetch/http.c 64129 2000-08-02 11:25:21Z des $ 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 82extern 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 107struct 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 */ 125static 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 */ 160static 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 */ 205static 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 */ 236static 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 */ 247static 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 */ 263static 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 */ 288typedef 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 */ 301static 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 313static char *reply_buf; 314static size_t reply_size; 315static size_t reply_length; 316 317/* 318 * Send a formatted line; optionally echo to terminal 319 */ 320static 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 */ 352static int 353_http_get_reply(int fd) 354{ 355 if (_fetch_getln(fd, &reply_buf, &reply_size, &reply_length) == -1) 356 return -1; 357 /* 358 * A valid status line looks like "HTTP/m.n xyz reason" where m 359 * and n are the major and minor protocol version numbers and xyz 360 * is the reply code. 361 * We grok HTTP 1.0 and 1.1, so m must be 1 and n must be 0 or 1. 362 * We don't care about the reason phrase. 363 */ 364 if (strncmp(reply_buf, "HTTP/1.", 7) != 0 365 || (reply_buf[7] != '0' && reply_buf[7] != '1') || reply_buf[8] != ' ' 366 || !isdigit(reply_buf[9]) 367 || !isdigit(reply_buf[10]) 368 || !isdigit(reply_buf[11])) 369 return HTTP_PROTOCOL_ERROR; 370 371 return ((reply_buf[9] - '0') * 100 372 + (reply_buf[10] - '0') * 10 373 + (reply_buf[11] - '0')); 374} 375 376/* 377 * Check a header; if the type matches the given string, return a 378 * pointer to the beginning of the value. 379 */ 380static char * 381_http_match(char *str, char *hdr) 382{ 383 while (*str && *hdr && tolower(*str++) == tolower(*hdr++)) 384 /* nothing */; 385 if (*str || *hdr != ':') 386 return NULL; 387 while (*hdr && isspace(*++hdr)) 388 /* nothing */; 389 return hdr; 390} 391 392/* 393 * Get the next header and return the appropriate symbolic code. 394 */ 395static hdr 396_http_next_header(int fd, char **p) 397{ 398 int i; 399 400 if (_fetch_getln(fd, &reply_buf, &reply_size, &reply_length) == -1) 401 return hdr_syserror; 402 while (reply_length && isspace(reply_buf[reply_length-1])) 403 reply_length--; 404 reply_buf[reply_length] = 0; 405 if (reply_length == 0) 406 return hdr_end; 407 /* 408 * We could check for malformed headers but we don't really care. 409 * A valid header starts with a token immediately followed by a 410 * colon; a token is any sequence of non-control, non-whitespace 411 * characters except "()<>@,;:\\\"{}". 412 */ 413 for (i = 0; hdr_names[i].num != hdr_unknown; i++) 414 if ((*p = _http_match(hdr_names[i].name, reply_buf)) != NULL) 415 return hdr_names[i].num; 416 return hdr_unknown; 417} 418 419/* 420 * Parse a last-modified header 421 */ 422static int 423_http_parse_mtime(char *p, time_t *mtime) 424{ 425 char locale[64], *r; 426 struct tm tm; 427 428 strncpy(locale, setlocale(LC_TIME, NULL), sizeof locale); 429 setlocale(LC_TIME, "C"); 430 r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm); 431 /* XXX should add support for date-2 and date-3 */ 432 setlocale(LC_TIME, locale); 433 if (r == NULL) 434 return -1; 435 DEBUG(fprintf(stderr, "last modified: [\033[1m%04d-%02d-%02d " 436 "%02d:%02d:%02d\033[m]\n", 437 tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday, 438 tm.tm_hour, tm.tm_min, tm.tm_sec)); 439 *mtime = timegm(&tm); 440 return 0; 441} 442 443/* 444 * Parse a content-length header 445 */ 446static int 447_http_parse_length(char *p, off_t *length) 448{ 449 off_t len; 450 451 for (len = 0; *p && isdigit(*p); ++p) 452 len = len * 10 + (*p - '0'); 453 DEBUG(fprintf(stderr, "content length: [\033[1m%lld\033[m]\n", len)); 454 *length = len; 455 return 0; 456} 457 458/* 459 * Parse a content-range header 460 */ 461static int 462_http_parse_range(char *p, off_t *offset, off_t *length, off_t *size) 463{ 464 int first, last, len; 465 466 if (strncasecmp(p, "bytes ", 6) != 0) 467 return -1; 468 for (first = 0, p += 6; *p && isdigit(*p); ++p) 469 first = first * 10 + *p - '0'; 470 if (*p != '-') 471 return -1; 472 for (last = 0, ++p; *p && isdigit(*p); ++p) 473 last = last * 10 + *p - '0'; 474 if (first > last || *p != '/') 475 return -1; 476 for (len = 0, ++p; *p && isdigit(*p); ++p) 477 len = len * 10 + *p - '0'; 478 if (len < last - first + 1) 479 return -1; 480 DEBUG(fprintf(stderr, "content range: [\033[1m%d-%d/%d\033[m]\n", 481 first, last, len)); 482 *offset = first; 483 *length = last - first + 1; 484 *size = len; 485 return 0; 486} 487 488 489/***************************************************************************** 490 * Helper functions for authorization 491 */ 492 493/* 494 * Base64 encoding 495 */ 496static char * 497_http_base64(char *src) 498{ 499 static const char base64[] = 500 "ABCDEFGHIJKLMNOPQRSTUVWXYZ" 501 "abcdefghijklmnopqrstuvwxyz" 502 "0123456789+/"; 503 char *str, *dst; 504 size_t l; 505 int t, r; 506 507 l = strlen(src); 508 if ((str = malloc(((l + 2) / 3) * 4)) == NULL) 509 return NULL; 510 dst = str; 511 r = 0; 512 513 while (l >= 3) { 514 t = (src[0] << 16) | (src[1] << 8) | src[2]; 515 dst[0] = base64[(t >> 18) & 0x3f]; 516 dst[1] = base64[(t >> 12) & 0x3f]; 517 dst[2] = base64[(t >> 6) & 0x3f]; 518 dst[3] = base64[(t >> 0) & 0x3f]; 519 src += 3; l -= 3; 520 dst += 4; r += 4; 521 } 522 523 switch (l) { 524 case 2: 525 t = (src[0] << 16) | (src[1] << 8); 526 dst[0] = base64[(t >> 18) & 0x3f]; 527 dst[1] = base64[(t >> 12) & 0x3f]; 528 dst[2] = base64[(t >> 6) & 0x3f]; 529 dst[3] = '='; 530 dst += 4; 531 r += 4; 532 break; 533 case 1: 534 t = src[0] << 16; 535 dst[0] = base64[(t >> 18) & 0x3f]; 536 dst[1] = base64[(t >> 12) & 0x3f]; 537 dst[2] = dst[3] = '='; 538 dst += 4; 539 r += 4; 540 break; 541 case 0: 542 break; 543 } 544 545 *dst = 0; 546 return str; 547} 548 549/* 550 * Encode username and password 551 */ 552static int 553_http_basic_auth(int fd, char *hdr, char *usr, char *pwd) 554{ 555 char *upw, *auth; 556 int r; 557 558 if (asprintf(&upw, "%s:%s", usr, pwd) == -1) 559 return -1; 560 auth = _http_base64(upw); 561 free(upw); 562 if (auth == NULL) 563 return -1; 564 r = _http_cmd(fd, "%s: Basic %s", hdr, auth); 565 free(auth); 566 return r; 567} 568 569/* 570 * Send an authorization header 571 */ 572static int 573_http_authorize(int fd, char *hdr, char *p) 574{ 575 /* basic authorization */ 576 if (strncasecmp(p, "basic:", 6) == 0) { 577 char *user, *pwd, *str; 578 int r; 579 580 /* skip realm */ 581 for (p += 6; *p && *p != ':'; ++p) 582 /* nothing */ ; 583 if (!*p || strchr(++p, ':') == NULL) 584 return -1; 585 if ((str = strdup(p)) == NULL) 586 return -1; /* XXX */ 587 user = str; 588 pwd = strchr(str, ':'); 589 *pwd++ = '\0'; 590 r = _http_basic_auth(fd, hdr, user, pwd); 591 free(str); 592 return r; 593 } 594 return -1; 595} 596 597 598/***************************************************************************** 599 * Helper functions for connecting to a server or proxy 600 */ 601 602/* 603 * Return the default port for this scheme 604 */ 605static int 606_http_default_port(char *scheme) 607{ 608 struct servent *se; 609 610 if ((se = getservbyname(scheme, "tcp")) != NULL) 611 return ntohs(se->s_port); 612 if (strcasecmp(scheme, "ftp") == 0) 613 return FTP_DEFAULT_PORT; 614 if (strcasecmp(scheme, "http") == 0) 615 return HTTP_DEFAULT_PORT; 616 return 0; 617} 618 619/* 620 * Connect to the specified HTTP proxy server. 621 */ 622static int 623_http_proxy_connect(char *proxy, int af, int verbose) 624{ 625 char *hostname, *p; 626 int fd, port; 627 628 /* get hostname */ 629 hostname = NULL; 630#ifdef INET6 631 /* host part can be an IPv6 address enclosed in square brackets */ 632 if (*proxy == '[') { 633 if ((p = strchr(proxy, ']')) == NULL) { 634 /* no terminating bracket */ 635 /* XXX should set an error code */ 636 goto ouch; 637 } 638 if (p[1] != '\0' && p[1] != ':') { 639 /* garbage after address */ 640 /* XXX should set an error code */ 641 goto ouch; 642 } 643 if ((hostname = malloc(p - proxy)) == NULL) { 644 errno = ENOMEM; 645 _fetch_syserr(); 646 goto ouch; 647 } 648 strncpy(hostname, proxy + 1, p - proxy - 1); 649 hostname[p - proxy - 1] = '\0'; 650 ++p; 651 } else { 652#endif /* INET6 */ 653 if ((p = strchr(proxy, ':')) == NULL) 654 p = strchr(proxy, '\0'); 655 if ((hostname = malloc(p - proxy + 1)) == NULL) { 656 errno = ENOMEM; 657 _fetch_syserr(); 658 goto ouch; 659 } 660 strncpy(hostname, proxy, p - proxy); 661 hostname[p - proxy] = '\0'; 662#ifdef INET6 663 } 664#endif /* INET6 */ 665 DEBUG(fprintf(stderr, "proxy name: [%s]\n", hostname)); 666 667 /* get port number */ 668 port = 0; 669 if (*p == ':') { 670 ++p; 671 if (strspn(p, "0123456789") != strlen(p) || strlen(p) > 5) { 672 /* port number is non-numeric or too long */ 673 /* XXX should set an error code */ 674 goto ouch; 675 } 676 port = atoi(p); 677 if (port < 1 || port > 65535) { 678 /* port number is out of range */ 679 /* XXX should set an error code */ 680 goto ouch; 681 } 682 } 683 684 if (!port) { 685#if 0 686 /* 687 * commented out, since there is currently no service name 688 * for HTTP proxies 689 */ 690 struct servent *se; 691 692 if ((se = getservbyname("xxxx", "tcp")) != NULL) 693 port = ntohs(se->s_port); 694 else 695#endif 696 port = 3128; 697 } 698 DEBUG(fprintf(stderr, "proxy port: %d\n", port)); 699 700 /* connect */ 701 if ((fd = _fetch_connect(hostname, port, af, verbose)) == -1) 702 _fetch_syserr(); 703 return fd; 704 705 ouch: 706 if (hostname) 707 free(hostname); 708 return -1; 709} 710 711/* 712 * Connect to the correct HTTP server or proxy. 713 */ 714static int 715_http_connect(struct url *URL, int *proxy, char *flags) 716{ 717 int direct, verbose; 718 int af, fd; 719 char *p; 720 721#ifdef INET6 722 af = AF_UNSPEC; 723#else 724 af = AF_INET; 725#endif 726 727 direct = (flags && strchr(flags, 'd')); 728 verbose = (flags && strchr(flags, 'v')); 729 if (flags && strchr(flags, '4')) 730 af = AF_INET; 731 else if (flags && strchr(flags, '6')) 732 af = AF_INET6; 733 734 /* check port */ 735 if (!URL->port) 736 URL->port = _http_default_port(URL->scheme); 737 738 if (!direct && (p = getenv("HTTP_PROXY")) != NULL && *p != '\0') { 739 /* attempt to connect to proxy server */ 740 if ((fd = _http_proxy_connect(p, af, verbose)) == -1) 741 return -1; 742 *proxy = 1; 743 } else { 744 /* if no proxy is configured, try direct */ 745 if (strcasecmp(URL->scheme, "ftp") == 0) { 746 /* can't talk http to an ftp server */ 747 /* XXX should set an error code */ 748 return -1; 749 } 750 if ((fd = _fetch_connect(URL->host, URL->port, af, verbose)) == -1) 751 /* _fetch_connect() has already set an error code */ 752 return -1; 753 *proxy = 0; 754 } 755 756 return fd; 757} 758 759 760/***************************************************************************** 761 * Core 762 */ 763 764/* 765 * Send a request and process the reply 766 */ 767static FILE * 768_http_request(struct url *URL, char *op, struct url_stat *us, char *flags) 769{ 770 struct url *url, *new; 771 int chunked, need_auth, noredirect, proxy, verbose; 772 int code, fd, i, n; 773 off_t offset, clength, length, size; 774 time_t mtime; 775 char *p; 776 FILE *f; 777 hdr h; 778 char *host; 779#ifdef INET6 780 char hbuf[MAXHOSTNAMELEN + 1]; 781#endif 782 783 noredirect = (flags && strchr(flags, 'A')); 784 verbose = (flags && strchr(flags, 'v')); 785 786 /* try the provided URL first */ 787 url = URL; 788 789 /* if the A flag is set, we only get one try */ 790 n = noredirect ? 1 : MAX_REDIRECT; 791 i = 0; 792 793 do { 794 new = NULL; 795 chunked = 0; 796 need_auth = 0; 797 offset = 0; 798 clength = -1; 799 length = -1; 800 size = -1; 801 mtime = 0; 802 retry: 803 /* connect to server or proxy */ 804 if ((fd = _http_connect(url, &proxy, flags)) == -1) 805 goto ouch; 806 807 host = url->host; 808#ifdef INET6 809 if (strchr(url->host, ':')) { 810 snprintf(hbuf, sizeof(hbuf), "[%s]", url->host); 811 host = hbuf; 812 } 813#endif 814 815 /* send request */ 816 if (verbose) 817 _fetch_info("requesting %s://%s:%d%s", 818 url->scheme, host, url->port, url->doc); 819 if (proxy) { 820 _http_cmd(fd, "%s %s://%s:%d%s HTTP/1.1", 821 op, url->scheme, host, url->port, url->doc); 822 } else { 823 _http_cmd(fd, "%s %s HTTP/1.1", 824 op, url->doc); 825 } 826 827 /* proxy authorization */ 828 if (proxy && (p = getenv("HTTP_PROXY_AUTH")) != NULL && *p != '\0') 829 _http_authorize(fd, "Proxy-Authorization", p); 830 831 /* server authorization */ 832 if (need_auth) { 833 if (*url->user || *url->pwd) 834 _http_basic_auth(fd, "Authorization", 835 url->user ? url->user : "", 836 url->pwd ? url->pwd : ""); 837 else if ((p = getenv("HTTP_AUTH")) != NULL && *p != '\0') 838 _http_authorize(fd, "Authorization", p); 839 else { 840 _http_seterr(HTTP_NEED_AUTH); 841 goto ouch; 842 } 843 } 844 845 /* other headers */ 846 if (url->port == _http_default_port(url->scheme)) 847 _http_cmd(fd, "Host: %s", host); 848 else 849 _http_cmd(fd, "Host: %s:%d", host, url->port); 850 _http_cmd(fd, "User-Agent: %s " _LIBFETCH_VER, __progname); 851 if (url->offset) 852 _http_cmd(fd, "Range: bytes=%lld-", url->offset); 853 _http_cmd(fd, "Connection: close"); 854 _http_cmd(fd, ""); 855 856 /* get reply */ 857 switch ((code = _http_get_reply(fd))) { 858 case HTTP_OK: 859 case HTTP_PARTIAL: 860 /* fine */ 861 break; 862 case HTTP_MOVED_PERM: 863 case HTTP_MOVED_TEMP: 864 /* 865 * Not so fine, but we still have to read the headers to 866 * get the new location. 867 */ 868 break; 869 case HTTP_NEED_AUTH: 870 if (need_auth) { 871 /* 872 * We already sent out authorization code, so there's 873 * nothing more we can do. 874 */ 875 _http_seterr(code); 876 goto ouch; 877 } 878 /* try again, but send the password this time */ 879 if (verbose) 880 _fetch_info("server requires authorization"); 881 need_auth = 1; 882 close(fd); 883 goto retry; 884 case HTTP_NEED_PROXY_AUTH: 885 /* 886 * If we're talking to a proxy, we already sent our proxy 887 * authorization code, so there's nothing more we can do. 888 */ 889 _http_seterr(code); 890 goto ouch; 891 case HTTP_PROTOCOL_ERROR: 892 /* fall through */ 893 case -1: 894 _fetch_syserr(); 895 goto ouch; 896 default: 897 _http_seterr(code); 898 goto ouch; 899 } 900 901 /* get headers */ 902 do { 903 switch ((h = _http_next_header(fd, &p))) { 904 case hdr_syserror: 905 _fetch_syserr(); 906 goto ouch; 907 case hdr_error: 908 _http_seterr(HTTP_PROTOCOL_ERROR); 909 goto ouch; 910 case hdr_content_length: 911 _http_parse_length(p, &clength); 912 break; 913 case hdr_content_range: 914 _http_parse_range(p, &offset, &length, &size); 915 break; 916 case hdr_last_modified: 917 _http_parse_mtime(p, &mtime); 918 break; 919 case hdr_location: 920 if (!HTTP_REDIRECT(code)) 921 break; 922 if (new) 923 free(new); 924 if (verbose) 925 _fetch_info("%d redirect to %s", code, p); 926 if (*p == '/') 927 /* absolute path */ 928 new = fetchMakeURL(url->scheme, url->host, url->port, p, 929 url->user, url->pwd); 930 else 931 new = fetchParseURL(p); 932 if (new == NULL) { 933 /* XXX should set an error code */ 934 DEBUG(fprintf(stderr, "failed to parse new URL\n")); 935 goto ouch; 936 } 937 if (!*new->user && !*new->pwd) { 938 strcpy(new->user, url->user); 939 strcpy(new->pwd, url->pwd); 940 } 941 new->offset = url->offset; 942 new->length = url->length; 943 break; 944 case hdr_transfer_encoding: 945 /* XXX weak test*/ 946 chunked = (strcasecmp(p, "chunked") == 0); 947 break; 948 case hdr_end: 949 /* fall through */ 950 case hdr_unknown: 951 /* ignore */ 952 break; 953 } 954 } while (h > hdr_end); 955 956 /* we either have a hit, or a redirect with no Location: header */ 957 if (code == HTTP_OK || code == HTTP_PARTIAL || !new) 958 break; 959 960 /* we have a redirect */ 961 close(fd); 962 fd = -1; 963 if (url != URL) 964 fetchFreeURL(url); 965 url = new; 966 } while (++i < n); 967 968 /* no success */ 969 if (fd == -1) { 970 _http_seterr(code); 971 goto ouch; 972 } 973 974 DEBUG(fprintf(stderr, "offset %lld, length %lld, size %lld, clength %lld\n", 975 offset, length, size, clength)); 976 977 /* check for inconsistencies */ 978 if (clength != -1 && length != -1 && clength != length) { 979 _http_seterr(HTTP_PROTOCOL_ERROR); 980 goto ouch; 981 } 982 if (clength == -1) 983 clength = length; 984 if (clength != -1) 985 length = offset + clength; 986 if (length != -1 && size != -1 && length != size) { 987 _http_seterr(HTTP_PROTOCOL_ERROR); 988 goto ouch; 989 } 990 if (size == -1) 991 size = length; 992 993 /* fill in stats */ 994 if (us) { 995 us->size = size; 996 us->atime = us->mtime = mtime; 997 } 998 999 /* too far? */ 1000 if (offset > URL->offset) { 1001 _http_seterr(HTTP_PROTOCOL_ERROR); 1002 goto ouch; 1003 } 1004 1005 /* report back real offset and size */ 1006 URL->offset = offset; 1007 URL->length = clength; 1008 1009 /* wrap it up in a FILE */ 1010 if ((f = chunked ? _http_funopen(fd) : fdopen(fd, "r")) == NULL) { 1011 _fetch_syserr(); 1012 goto ouch; 1013 } 1014 1015 if (url != URL) 1016 fetchFreeURL(url); 1017 1018 return f; 1019 1020 ouch: 1021 if (url != URL) 1022 fetchFreeURL(url); 1023 if (fd != -1) 1024 close(fd); 1025 return NULL; 1026} 1027 1028 1029/***************************************************************************** 1030 * Entry points 1031 */ 1032 1033/* 1034 * Retrieve and stat a file by HTTP 1035 */ 1036FILE * 1037fetchXGetHTTP(struct url *URL, struct url_stat *us, char *flags) 1038{ 1039 return _http_request(URL, "GET", us, flags); 1040} 1041 1042/* 1043 * Retrieve a file by HTTP 1044 */ 1045FILE * 1046fetchGetHTTP(struct url *URL, char *flags) 1047{ 1048 return fetchXGetHTTP(URL, NULL, flags); 1049} 1050 1051/* 1052 * Store a file by HTTP 1053 */ 1054FILE * 1055fetchPutHTTP(struct url *URL, char *flags) 1056{ 1057 warnx("fetchPutHTTP(): not implemented"); 1058 return NULL; 1059} 1060 1061/* 1062 * Get an HTTP document's metadata 1063 */ 1064int 1065fetchStatHTTP(struct url *URL, struct url_stat *us, char *flags) 1066{ 1067 FILE *f; 1068 1069 if ((f = _http_request(URL, "HEAD", us, flags)) == NULL) 1070 return -1; 1071 fclose(f); 1072 return 0; 1073} 1074 1075/* 1076 * List a directory 1077 */ 1078struct url_ent * 1079fetchListHTTP(struct url *url, char *flags) 1080{ 1081 warnx("fetchListHTTP(): not implemented"); 1082 return NULL; 1083} 1084