1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 2000-2014 Dag-Erling Smørgrav 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer 12 * in this position and unchanged. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. The name of the author may not be used to endorse or promote products 17 * derived from this software without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 20 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 21 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 22 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 23 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 24 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 25 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 26 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 27 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 28 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 29 */ 30 31 #include <sys/cdefs.h> 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 #include <sys/time.h> 67 68 #include <ctype.h> 69 #include <err.h> 70 #include <errno.h> 71 #include <locale.h> 72 #include <netdb.h> 73 #include <stdarg.h> 74 #include <stdbool.h> 75 #include <stdio.h> 76 #include <stdlib.h> 77 #include <string.h> 78 #include <time.h> 79 #include <unistd.h> 80 81 #ifdef WITH_SSL 82 #include <openssl/md5.h> 83 #define MD5Init(c) MD5_Init(c) 84 #define MD5Update(c, data, len) MD5_Update(c, data, len) 85 #define MD5Final(md, c) MD5_Final(md, c) 86 #else 87 #include <md5.h> 88 #endif 89 90 #include <netinet/in.h> 91 #include <netinet/tcp.h> 92 93 #include "fetch.h" 94 #include "common.h" 95 #include "httperr.h" 96 97 /* Maximum number of redirects to follow */ 98 #define MAX_REDIRECT 20 99 100 /* Symbolic names for reply codes we care about */ 101 #define HTTP_OK 200 102 #define HTTP_PARTIAL 206 103 #define HTTP_MOVED_PERM 301 104 #define HTTP_MOVED_TEMP 302 105 #define HTTP_SEE_OTHER 303 106 #define HTTP_NOT_MODIFIED 304 107 #define HTTP_USE_PROXY 305 108 #define HTTP_TEMP_REDIRECT 307 109 #define HTTP_PERM_REDIRECT 308 110 #define HTTP_NEED_AUTH 401 111 #define HTTP_NEED_PROXY_AUTH 407 112 #define HTTP_BAD_RANGE 416 113 #define HTTP_PROTOCOL_ERROR 999 114 115 #define HTTP_REDIRECT(xyz) ((xyz) == HTTP_MOVED_PERM \ 116 || (xyz) == HTTP_MOVED_TEMP \ 117 || (xyz) == HTTP_TEMP_REDIRECT \ 118 || (xyz) == HTTP_PERM_REDIRECT \ 119 || (xyz) == HTTP_USE_PROXY \ 120 || (xyz) == HTTP_SEE_OTHER) 121 122 #define HTTP_ERROR(xyz) ((xyz) >= 400 && (xyz) <= 599) 123 124 125 /***************************************************************************** 126 * I/O functions for decoding chunked streams 127 */ 128 129 struct httpio { 130 conn_t *conn; /* connection */ 131 int chunked; /* chunked mode */ 132 char *buf; /* chunk buffer */ 133 size_t bufsize; /* size of chunk buffer */ 134 size_t buflen; /* amount of data currently in buffer */ 135 size_t bufpos; /* current read offset in buffer */ 136 int eof; /* end-of-file flag */ 137 int error; /* error flag */ 138 size_t chunksize; /* remaining size of current chunk */ 139 #ifndef NDEBUG 140 size_t total; 141 #endif 142 }; 143 144 /* 145 * Get next chunk header 146 */ 147 static int 148 http_new_chunk(struct httpio *io) 149 { 150 char *p; 151 152 if (fetch_getln(io->conn) == -1) 153 return (-1); 154 155 for (p = io->conn->line; 156 *p != '\0' && !isspace((unsigned char)*p) && *p != ';'; 157 p++) { 158 io->chunksize <<= 4; 159 if (*p >= '0' && *p <= '9') 160 io->chunksize += *p - '0'; 161 else if (*p >= 'A' && *p <= 'F') 162 io->chunksize += 10 + *p - 'A'; 163 else if (*p >= 'a' && *p <= 'f') 164 io->chunksize += 10 + *p - 'a'; 165 else 166 return (-1); 167 } 168 169 #ifndef NDEBUG 170 if (fetchDebug) { 171 io->total += io->chunksize; 172 if (io->chunksize == 0) 173 fprintf(stderr, "%s(): end of last chunk\n", __func__); 174 else 175 fprintf(stderr, "%s(): new chunk: %lu (%lu)\n", 176 __func__, (unsigned long)io->chunksize, 177 (unsigned long)io->total); 178 } 179 #endif 180 181 return (io->chunksize); 182 } 183 184 /* 185 * Grow the input buffer to at least len bytes 186 */ 187 static inline int 188 http_growbuf(struct httpio *io, size_t len) 189 { 190 char *tmp; 191 192 if (io->bufsize >= len) 193 return (0); 194 195 if ((tmp = realloc(io->buf, len)) == NULL) 196 return (-1); 197 io->buf = tmp; 198 io->bufsize = len; 199 return (0); 200 } 201 202 /* 203 * Fill the input buffer, do chunk decoding on the fly 204 */ 205 static ssize_t 206 http_fillbuf(struct httpio *io, size_t len) 207 { 208 ssize_t nbytes; 209 char ch; 210 211 if (io->error) 212 return (-1); 213 if (io->eof) 214 return (0); 215 216 /* not chunked: just fetch the requested amount */ 217 if (io->chunked == 0) { 218 if (http_growbuf(io, len) == -1) 219 return (-1); 220 if ((nbytes = fetch_bufread(io->conn, io->buf, len)) == -1) { 221 io->error = errno; 222 return (-1); 223 } 224 io->buflen = nbytes; 225 io->bufpos = 0; 226 return (io->buflen); 227 } 228 229 /* chunked, but we ran out: get the next chunk header */ 230 if (io->chunksize == 0) { 231 switch (http_new_chunk(io)) { 232 case -1: 233 io->error = EPROTO; 234 return (-1); 235 case 0: 236 io->eof = 1; 237 return (0); 238 } 239 } 240 241 /* fetch the requested amount, but no more than the current chunk */ 242 if (len > io->chunksize) 243 len = io->chunksize; 244 if (http_growbuf(io, len) == -1) 245 return (-1); 246 if ((nbytes = fetch_bufread(io->conn, io->buf, len)) == -1) { 247 io->error = errno; 248 return (-1); 249 } 250 io->bufpos = 0; 251 io->buflen = nbytes; 252 io->chunksize -= nbytes; 253 254 if (io->chunksize == 0) { 255 if (fetch_bufread(io->conn, &ch, 1) != 1 || ch != '\r' || 256 fetch_bufread(io->conn, &ch, 1) != 1 || ch != '\n') 257 return (-1); 258 } 259 260 return (io->buflen); 261 } 262 263 /* 264 * Read function 265 */ 266 static int 267 http_readfn(void *v, char *buf, int len) 268 { 269 struct httpio *io = (struct httpio *)v; 270 int rlen; 271 272 if (io->error) 273 return (-1); 274 if (io->eof) 275 return (0); 276 277 /* empty buffer */ 278 if (!io->buf || io->bufpos == io->buflen) { 279 if ((rlen = http_fillbuf(io, len)) < 0) { 280 if ((errno = io->error) == EINTR) 281 io->error = 0; 282 return (-1); 283 } else if (rlen == 0) { 284 return (0); 285 } 286 } 287 288 rlen = io->buflen - io->bufpos; 289 if (len < rlen) 290 rlen = len; 291 memcpy(buf, io->buf + io->bufpos, rlen); 292 io->bufpos += rlen; 293 return (rlen); 294 } 295 296 /* 297 * Write function 298 */ 299 static int 300 http_writefn(void *v, const char *buf, int len) 301 { 302 struct httpio *io = (struct httpio *)v; 303 304 return (fetch_write(io->conn, buf, len)); 305 } 306 307 /* 308 * Close function 309 */ 310 static int 311 http_closefn(void *v) 312 { 313 struct httpio *io = (struct httpio *)v; 314 int r; 315 316 r = fetch_close(io->conn); 317 if (io->buf) 318 free(io->buf); 319 free(io); 320 return (r); 321 } 322 323 /* 324 * Wrap a file descriptor up 325 */ 326 static FILE * 327 http_funopen(conn_t *conn, int chunked) 328 { 329 struct httpio *io; 330 FILE *f; 331 332 if ((io = calloc(1, sizeof(*io))) == NULL) { 333 fetch_syserr(); 334 return (NULL); 335 } 336 io->conn = conn; 337 io->chunked = chunked; 338 f = funopen(io, http_readfn, http_writefn, NULL, http_closefn); 339 if (f == NULL) { 340 fetch_syserr(); 341 free(io); 342 return (NULL); 343 } 344 return (f); 345 } 346 347 348 /***************************************************************************** 349 * Helper functions for talking to the server and parsing its replies 350 */ 351 352 /* Header types */ 353 typedef enum { 354 hdr_syserror = -2, 355 hdr_error = -1, 356 hdr_end = 0, 357 hdr_unknown = 1, 358 hdr_content_length, 359 hdr_content_range, 360 hdr_last_modified, 361 hdr_location, 362 hdr_transfer_encoding, 363 hdr_www_authenticate, 364 hdr_proxy_authenticate, 365 } hdr_t; 366 367 /* Names of interesting headers */ 368 static struct { 369 hdr_t num; 370 const char *name; 371 } hdr_names[] = { 372 { hdr_content_length, "Content-Length" }, 373 { hdr_content_range, "Content-Range" }, 374 { hdr_last_modified, "Last-Modified" }, 375 { hdr_location, "Location" }, 376 { hdr_transfer_encoding, "Transfer-Encoding" }, 377 { hdr_www_authenticate, "WWW-Authenticate" }, 378 { hdr_proxy_authenticate, "Proxy-Authenticate" }, 379 { hdr_unknown, NULL }, 380 }; 381 382 /* 383 * Send a formatted line; optionally echo to terminal 384 */ 385 static int 386 http_cmd(conn_t *conn, const char *fmt, ...) 387 { 388 va_list ap; 389 size_t len; 390 char *msg; 391 int r; 392 393 va_start(ap, fmt); 394 len = vasprintf(&msg, fmt, ap); 395 va_end(ap); 396 397 if (msg == NULL) { 398 errno = ENOMEM; 399 fetch_syserr(); 400 return (-1); 401 } 402 403 r = fetch_putln(conn, msg, len); 404 free(msg); 405 406 if (r == -1) { 407 fetch_syserr(); 408 return (-1); 409 } 410 411 return (0); 412 } 413 414 /* 415 * Get and parse status line 416 */ 417 static int 418 http_get_reply(conn_t *conn) 419 { 420 char *p; 421 422 if (fetch_getln(conn) == -1) 423 return (-1); 424 /* 425 * A valid status line looks like "HTTP/m.n xyz reason" where m 426 * and n are the major and minor protocol version numbers and xyz 427 * is the reply code. 428 * Unfortunately, there are servers out there (NCSA 1.5.1, to name 429 * just one) that do not send a version number, so we can't rely 430 * on finding one, but if we do, insist on it being 1.0 or 1.1. 431 * We don't care about the reason phrase. 432 */ 433 if (conn->linelen < 4 || strncmp(conn->line, "HTTP", 4) != 0) 434 return (HTTP_PROTOCOL_ERROR); 435 p = conn->line + 4; 436 if (*p == '/') { 437 if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1')) 438 return (HTTP_PROTOCOL_ERROR); 439 p += 4; 440 } 441 if (*p != ' ' || 442 !isdigit((unsigned char)p[1]) || 443 !isdigit((unsigned char)p[2]) || 444 !isdigit((unsigned char)p[3])) 445 return (HTTP_PROTOCOL_ERROR); 446 447 conn->err = (p[1] - '0') * 100 + (p[2] - '0') * 10 + (p[3] - '0'); 448 return (conn->err); 449 } 450 451 /* 452 * Check a header; if the type matches the given string, return a pointer 453 * to the beginning of the value. 454 */ 455 static const char * 456 http_match(const char *str, const char *hdr) 457 { 458 while (*str && *hdr && 459 tolower((unsigned char)*str++) == tolower((unsigned char)*hdr++)) 460 /* nothing */; 461 if (*str || *hdr != ':') 462 return (NULL); 463 while (*hdr && isspace((unsigned char)*++hdr)) 464 /* nothing */; 465 return (hdr); 466 } 467 468 469 /* 470 * Get the next header and return the appropriate symbolic code. We 471 * need to read one line ahead for checking for a continuation line 472 * belonging to the current header (continuation lines start with 473 * white space). 474 * 475 * We get called with a fresh line already in the conn buffer, either 476 * from the previous http_next_header() invocation, or, the first 477 * time, from a fetch_getln() performed by our caller. 478 * 479 * This stops when we encounter an empty line (we dont read beyond the header 480 * area). 481 * 482 * Note that the "headerbuf" is just a place to return the result. Its 483 * contents are not used for the next call. This means that no cleanup 484 * is needed when ie doing another connection, just call the cleanup when 485 * fully done to deallocate memory. 486 */ 487 488 /* Limit the max number of continuation lines to some reasonable value */ 489 #define HTTP_MAX_CONT_LINES 10 490 491 /* Place into which to build a header from one or several lines */ 492 typedef struct { 493 char *buf; /* buffer */ 494 size_t bufsize; /* buffer size */ 495 size_t buflen; /* length of buffer contents */ 496 } http_headerbuf_t; 497 498 static void 499 init_http_headerbuf(http_headerbuf_t *buf) 500 { 501 buf->buf = NULL; 502 buf->bufsize = 0; 503 buf->buflen = 0; 504 } 505 506 static void 507 clean_http_headerbuf(http_headerbuf_t *buf) 508 { 509 if (buf->buf) 510 free(buf->buf); 511 init_http_headerbuf(buf); 512 } 513 514 static hdr_t 515 http_next_header(conn_t *conn, http_headerbuf_t *hbuf, const char **p) 516 { 517 unsigned int i, len; 518 519 if (conn->linelen == 0) 520 return (hdr_end); 521 522 /* Copy the line to the headerbuf */ 523 if (hbuf->bufsize <= conn->linelen) { 524 if ((hbuf->buf = realloc(hbuf->buf, conn->linelen + 1)) == NULL) 525 return (hdr_syserror); 526 hbuf->bufsize = conn->linelen + 1; 527 } 528 memcpy(hbuf->buf, conn->line, conn->linelen + 1); 529 hbuf->buflen = conn->linelen; 530 531 /* 532 * Fetch possible continuation lines. Stop at 1st non-continuation 533 * and leave it in the conn buffer 534 */ 535 for (i = 0; i < HTTP_MAX_CONT_LINES; i++) { 536 if (fetch_getln(conn) == -1) 537 return (hdr_syserror); 538 539 /* 540 * Note: we previously considered a pure whitespace line 541 * equivalent to an empty one. This was incorrect. 542 */ 543 if (conn->line[0] != ' ' && conn->line[0] != '\t') 544 break; 545 546 /* Got a continuation line. Concatenate to previous */ 547 len = hbuf->buflen + conn->linelen; 548 if (hbuf->bufsize <= len) { 549 len *= 2; 550 if ((hbuf->buf = realloc(hbuf->buf, len + 1)) == NULL) 551 return (hdr_syserror); 552 hbuf->bufsize = len + 1; 553 } 554 memcpy(hbuf->buf + hbuf->buflen, conn->line, conn->linelen + 1); 555 hbuf->buflen += conn->linelen; 556 } 557 558 /* 559 * We could check for malformed headers but we don't really care. 560 * A valid header starts with a token immediately followed by a 561 * colon; a token is any sequence of non-control, non-whitespace 562 * characters except "()<>@,;:\\\"{}". 563 */ 564 for (i = 0; hdr_names[i].num != hdr_unknown; i++) 565 if ((*p = http_match(hdr_names[i].name, hbuf->buf)) != NULL) 566 return (hdr_names[i].num); 567 568 return (hdr_unknown); 569 } 570 571 /************************** 572 * [Proxy-]Authenticate header parsing 573 */ 574 575 /* 576 * Read doublequote-delimited string into output buffer obuf (allocated 577 * by caller, whose responsibility it is to ensure that it's big enough) 578 * cp points to the first char after the initial '"' 579 * Handles \ quoting 580 * Returns pointer to the first char after the terminating double quote, or 581 * NULL for error. 582 */ 583 static const char * 584 http_parse_headerstring(const char *cp, char *obuf) 585 { 586 for (;;) { 587 switch (*cp) { 588 case 0: /* Unterminated string */ 589 *obuf = 0; 590 return (NULL); 591 case '"': /* Ending quote */ 592 *obuf = 0; 593 return (++cp); 594 case '\\': 595 if (*++cp == 0) { 596 *obuf = 0; 597 return (NULL); 598 } 599 /* FALLTHROUGH */ 600 default: 601 *obuf++ = *cp++; 602 } 603 } 604 } 605 606 /* Http auth challenge schemes */ 607 typedef enum {HTTPAS_UNKNOWN, HTTPAS_BASIC,HTTPAS_DIGEST} http_auth_schemes_t; 608 609 /* Data holder for a Basic or Digest challenge. */ 610 typedef struct { 611 http_auth_schemes_t scheme; 612 char *realm; 613 char *qop; 614 char *nonce; 615 char *opaque; 616 char *algo; 617 int stale; 618 int nc; /* Nonce count */ 619 } http_auth_challenge_t; 620 621 static void 622 init_http_auth_challenge(http_auth_challenge_t *b) 623 { 624 b->scheme = HTTPAS_UNKNOWN; 625 b->realm = b->qop = b->nonce = b->opaque = b->algo = NULL; 626 b->stale = b->nc = 0; 627 } 628 629 static void 630 clean_http_auth_challenge(http_auth_challenge_t *b) 631 { 632 if (b->realm) 633 free(b->realm); 634 if (b->qop) 635 free(b->qop); 636 if (b->nonce) 637 free(b->nonce); 638 if (b->opaque) 639 free(b->opaque); 640 if (b->algo) 641 free(b->algo); 642 init_http_auth_challenge(b); 643 } 644 645 /* Data holder for an array of challenges offered in an http response. */ 646 #define MAX_CHALLENGES 10 647 typedef struct { 648 http_auth_challenge_t *challenges[MAX_CHALLENGES]; 649 int count; /* Number of parsed challenges in the array */ 650 int valid; /* We did parse an authenticate header */ 651 } http_auth_challenges_t; 652 653 static void 654 init_http_auth_challenges(http_auth_challenges_t *cs) 655 { 656 int i; 657 for (i = 0; i < MAX_CHALLENGES; i++) 658 cs->challenges[i] = NULL; 659 cs->count = cs->valid = 0; 660 } 661 662 static void 663 clean_http_auth_challenges(http_auth_challenges_t *cs) 664 { 665 int i; 666 /* We rely on non-zero pointers being allocated, not on the count */ 667 for (i = 0; i < MAX_CHALLENGES; i++) { 668 if (cs->challenges[i] != NULL) { 669 clean_http_auth_challenge(cs->challenges[i]); 670 free(cs->challenges[i]); 671 } 672 } 673 init_http_auth_challenges(cs); 674 } 675 676 /* 677 * Enumeration for lexical elements. Separators will be returned as their own 678 * ascii value 679 */ 680 typedef enum {HTTPHL_WORD=256, HTTPHL_STRING=257, HTTPHL_END=258, 681 HTTPHL_ERROR = 259} http_header_lex_t; 682 683 /* 684 * Determine what kind of token comes next and return possible value 685 * in buf, which is supposed to have been allocated big enough by 686 * caller. Advance input pointer and return element type. 687 */ 688 static int 689 http_header_lex(const char **cpp, char *buf) 690 { 691 size_t l; 692 /* Eat initial whitespace */ 693 *cpp += strspn(*cpp, " \t"); 694 if (**cpp == 0) 695 return (HTTPHL_END); 696 697 /* Separator ? */ 698 if (**cpp == ',' || **cpp == '=') 699 return (*((*cpp)++)); 700 701 /* String ? */ 702 if (**cpp == '"') { 703 *cpp = http_parse_headerstring(++*cpp, buf); 704 if (*cpp == NULL) 705 return (HTTPHL_ERROR); 706 return (HTTPHL_STRING); 707 } 708 709 /* Read other token, until separator or whitespace */ 710 l = strcspn(*cpp, " \t,="); 711 memcpy(buf, *cpp, l); 712 buf[l] = 0; 713 *cpp += l; 714 return (HTTPHL_WORD); 715 } 716 717 /* 718 * Read challenges from http xxx-authenticate header and accumulate them 719 * in the challenges list structure. 720 * 721 * Headers with multiple challenges are specified by rfc2617, but 722 * servers (ie: squid) often send them in separate headers instead, 723 * which in turn is forbidden by the http spec (multiple headers with 724 * the same name are only allowed for pure comma-separated lists, see 725 * rfc2616 sec 4.2). 726 * 727 * We support both approaches anyway 728 */ 729 static int 730 http_parse_authenticate(const char *cp, http_auth_challenges_t *cs) 731 { 732 int ret = -1; 733 http_header_lex_t lex; 734 char *key = malloc(strlen(cp) + 1); 735 char *value = malloc(strlen(cp) + 1); 736 char *buf = malloc(strlen(cp) + 1); 737 738 if (key == NULL || value == NULL || buf == NULL) { 739 fetch_syserr(); 740 goto out; 741 } 742 743 /* In any case we've seen the header and we set the valid bit */ 744 cs->valid = 1; 745 746 /* Need word first */ 747 lex = http_header_lex(&cp, key); 748 if (lex != HTTPHL_WORD) 749 goto out; 750 751 /* Loop on challenges */ 752 for (; cs->count < MAX_CHALLENGES; cs->count++) { 753 cs->challenges[cs->count] = 754 malloc(sizeof(http_auth_challenge_t)); 755 if (cs->challenges[cs->count] == NULL) { 756 fetch_syserr(); 757 goto out; 758 } 759 init_http_auth_challenge(cs->challenges[cs->count]); 760 if (strcasecmp(key, "basic") == 0) { 761 cs->challenges[cs->count]->scheme = HTTPAS_BASIC; 762 } else if (strcasecmp(key, "digest") == 0) { 763 cs->challenges[cs->count]->scheme = HTTPAS_DIGEST; 764 } else { 765 cs->challenges[cs->count]->scheme = HTTPAS_UNKNOWN; 766 /* 767 * Continue parsing as basic or digest may 768 * follow, and the syntax is the same for 769 * all. We'll just ignore this one when 770 * looking at the list 771 */ 772 } 773 774 /* Loop on attributes */ 775 for (;;) { 776 /* Key */ 777 lex = http_header_lex(&cp, key); 778 if (lex != HTTPHL_WORD) 779 goto out; 780 781 /* Equal sign */ 782 lex = http_header_lex(&cp, buf); 783 if (lex != '=') 784 goto out; 785 786 /* Value */ 787 lex = http_header_lex(&cp, value); 788 if (lex != HTTPHL_WORD && lex != HTTPHL_STRING) 789 goto out; 790 791 if (strcasecmp(key, "realm") == 0) { 792 cs->challenges[cs->count]->realm = 793 strdup(value); 794 } else if (strcasecmp(key, "qop") == 0) { 795 cs->challenges[cs->count]->qop = 796 strdup(value); 797 } else if (strcasecmp(key, "nonce") == 0) { 798 cs->challenges[cs->count]->nonce = 799 strdup(value); 800 } else if (strcasecmp(key, "opaque") == 0) { 801 cs->challenges[cs->count]->opaque = 802 strdup(value); 803 } else if (strcasecmp(key, "algorithm") == 0) { 804 cs->challenges[cs->count]->algo = 805 strdup(value); 806 } else if (strcasecmp(key, "stale") == 0) { 807 cs->challenges[cs->count]->stale = 808 strcasecmp(value, "no"); 809 } else { 810 /* ignore unknown attributes */ 811 } 812 813 /* Comma or Next challenge or End */ 814 lex = http_header_lex(&cp, key); 815 /* 816 * If we get a word here, this is the beginning of the 817 * next challenge. Break the attributes loop 818 */ 819 if (lex == HTTPHL_WORD) 820 break; 821 822 if (lex == HTTPHL_END) { 823 /* End while looking for ',' is normal exit */ 824 cs->count++; 825 ret = 0; 826 goto out; 827 } 828 /* Anything else is an error */ 829 if (lex != ',') 830 goto out; 831 832 } /* End attributes loop */ 833 } /* End challenge loop */ 834 835 /* 836 * Challenges max count exceeded. This really can't happen 837 * with normal data, something's fishy -> error 838 */ 839 840 out: 841 if (key) 842 free(key); 843 if (value) 844 free(value); 845 if (buf) 846 free(buf); 847 return (ret); 848 } 849 850 851 /* 852 * Parse a last-modified header 853 */ 854 static int 855 http_parse_mtime(const char *p, time_t *mtime) 856 { 857 char locale[64], *r; 858 struct tm tm; 859 860 strlcpy(locale, setlocale(LC_TIME, NULL), sizeof(locale)); 861 setlocale(LC_TIME, "C"); 862 r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm); 863 /* 864 * Some proxies use UTC in response, but it should still be 865 * parsed. RFC2616 states GMT and UTC are exactly equal for HTTP. 866 */ 867 if (r == NULL) 868 r = strptime(p, "%a, %d %b %Y %H:%M:%S UTC", &tm); 869 /* XXX should add support for date-2 and date-3 */ 870 setlocale(LC_TIME, locale); 871 if (r == NULL) 872 return (-1); 873 DEBUGF("last modified: [%04d-%02d-%02d %02d:%02d:%02d]\n", 874 tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday, 875 tm.tm_hour, tm.tm_min, tm.tm_sec); 876 *mtime = timegm(&tm); 877 return (0); 878 } 879 880 /* 881 * Parse a content-length header 882 */ 883 static int 884 http_parse_length(const char *p, off_t *length) 885 { 886 off_t len; 887 888 for (len = 0; *p && isdigit((unsigned char)*p); ++p) 889 len = len * 10 + (*p - '0'); 890 if (*p) 891 return (-1); 892 DEBUGF("content length: [%lld]\n", (long long)len); 893 *length = len; 894 return (0); 895 } 896 897 /* 898 * Parse a content-range header 899 */ 900 static int 901 http_parse_range(const char *p, off_t *offset, off_t *length, off_t *size) 902 { 903 off_t first, last, len; 904 905 if (strncasecmp(p, "bytes ", 6) != 0) 906 return (-1); 907 p += 6; 908 if (*p == '*') { 909 first = last = -1; 910 ++p; 911 } else { 912 for (first = 0; *p && isdigit((unsigned char)*p); ++p) 913 first = first * 10 + *p - '0'; 914 if (*p != '-') 915 return (-1); 916 for (last = 0, ++p; *p && isdigit((unsigned char)*p); ++p) 917 last = last * 10 + *p - '0'; 918 } 919 if (first > last || *p != '/') 920 return (-1); 921 for (len = 0, ++p; *p && isdigit((unsigned char)*p); ++p) 922 len = len * 10 + *p - '0'; 923 if (*p || len < last - first + 1) 924 return (-1); 925 if (first == -1) { 926 DEBUGF("content range: [*/%lld]\n", (long long)len); 927 *length = 0; 928 } else { 929 DEBUGF("content range: [%lld-%lld/%lld]\n", 930 (long long)first, (long long)last, (long long)len); 931 *length = last - first + 1; 932 } 933 *offset = first; 934 *size = len; 935 return (0); 936 } 937 938 939 /***************************************************************************** 940 * Helper functions for authorization 941 */ 942 943 /* 944 * Base64 encoding 945 */ 946 static char * 947 http_base64(const char *src) 948 { 949 static const char base64[] = 950 "ABCDEFGHIJKLMNOPQRSTUVWXYZ" 951 "abcdefghijklmnopqrstuvwxyz" 952 "0123456789+/"; 953 char *str, *dst; 954 size_t l; 955 int t; 956 957 l = strlen(src); 958 if ((str = malloc(((l + 2) / 3) * 4 + 1)) == NULL) 959 return (NULL); 960 dst = str; 961 962 while (l >= 3) { 963 t = (src[0] << 16) | (src[1] << 8) | src[2]; 964 dst[0] = base64[(t >> 18) & 0x3f]; 965 dst[1] = base64[(t >> 12) & 0x3f]; 966 dst[2] = base64[(t >> 6) & 0x3f]; 967 dst[3] = base64[(t >> 0) & 0x3f]; 968 src += 3; l -= 3; 969 dst += 4; 970 } 971 972 switch (l) { 973 case 2: 974 t = (src[0] << 16) | (src[1] << 8); 975 dst[0] = base64[(t >> 18) & 0x3f]; 976 dst[1] = base64[(t >> 12) & 0x3f]; 977 dst[2] = base64[(t >> 6) & 0x3f]; 978 dst[3] = '='; 979 dst += 4; 980 break; 981 case 1: 982 t = src[0] << 16; 983 dst[0] = base64[(t >> 18) & 0x3f]; 984 dst[1] = base64[(t >> 12) & 0x3f]; 985 dst[2] = dst[3] = '='; 986 dst += 4; 987 break; 988 case 0: 989 break; 990 } 991 992 *dst = 0; 993 return (str); 994 } 995 996 997 /* 998 * Extract authorization parameters from environment value. 999 * The value is like scheme:realm:user:pass 1000 */ 1001 typedef struct { 1002 char *scheme; 1003 char *realm; 1004 char *user; 1005 char *password; 1006 } http_auth_params_t; 1007 1008 static void 1009 init_http_auth_params(http_auth_params_t *s) 1010 { 1011 s->scheme = s->realm = s->user = s->password = NULL; 1012 } 1013 1014 static void 1015 clean_http_auth_params(http_auth_params_t *s) 1016 { 1017 if (s->scheme) 1018 free(s->scheme); 1019 if (s->realm) 1020 free(s->realm); 1021 if (s->user) 1022 free(s->user); 1023 if (s->password) 1024 free(s->password); 1025 init_http_auth_params(s); 1026 } 1027 1028 static int 1029 http_authfromenv(const char *p, http_auth_params_t *parms) 1030 { 1031 int ret = -1; 1032 char *v, *ve; 1033 char *str = strdup(p); 1034 1035 if (str == NULL) { 1036 fetch_syserr(); 1037 return (-1); 1038 } 1039 v = str; 1040 1041 if ((ve = strchr(v, ':')) == NULL) 1042 goto out; 1043 1044 *ve = 0; 1045 if ((parms->scheme = strdup(v)) == NULL) { 1046 fetch_syserr(); 1047 goto out; 1048 } 1049 v = ve + 1; 1050 1051 if ((ve = strchr(v, ':')) == NULL) 1052 goto out; 1053 1054 *ve = 0; 1055 if ((parms->realm = strdup(v)) == NULL) { 1056 fetch_syserr(); 1057 goto out; 1058 } 1059 v = ve + 1; 1060 1061 if ((ve = strchr(v, ':')) == NULL) 1062 goto out; 1063 1064 *ve = 0; 1065 if ((parms->user = strdup(v)) == NULL) { 1066 fetch_syserr(); 1067 goto out; 1068 } 1069 v = ve + 1; 1070 1071 1072 if ((parms->password = strdup(v)) == NULL) { 1073 fetch_syserr(); 1074 goto out; 1075 } 1076 ret = 0; 1077 out: 1078 if (ret == -1) 1079 clean_http_auth_params(parms); 1080 if (str) 1081 free(str); 1082 return (ret); 1083 } 1084 1085 1086 /* 1087 * Digest response: the code to compute the digest is taken from the 1088 * sample implementation in RFC2616 1089 */ 1090 #define IN const 1091 #define OUT 1092 1093 #define HASHLEN 16 1094 typedef char HASH[HASHLEN]; 1095 #define HASHHEXLEN 32 1096 typedef char HASHHEX[HASHHEXLEN+1]; 1097 1098 static const char *hexchars = "0123456789abcdef"; 1099 static void 1100 CvtHex(IN HASH Bin, OUT HASHHEX Hex) 1101 { 1102 unsigned short i; 1103 unsigned char j; 1104 1105 for (i = 0; i < HASHLEN; i++) { 1106 j = (Bin[i] >> 4) & 0xf; 1107 Hex[i*2] = hexchars[j]; 1108 j = Bin[i] & 0xf; 1109 Hex[i*2+1] = hexchars[j]; 1110 } 1111 Hex[HASHHEXLEN] = '\0'; 1112 }; 1113 1114 /* calculate H(A1) as per spec */ 1115 static void 1116 DigestCalcHA1( 1117 IN char * pszAlg, 1118 IN char * pszUserName, 1119 IN char * pszRealm, 1120 IN char * pszPassword, 1121 IN char * pszNonce, 1122 IN char * pszCNonce, 1123 OUT HASHHEX SessionKey 1124 ) 1125 { 1126 MD5_CTX Md5Ctx; 1127 HASH HA1; 1128 1129 MD5Init(&Md5Ctx); 1130 MD5Update(&Md5Ctx, pszUserName, strlen(pszUserName)); 1131 MD5Update(&Md5Ctx, ":", 1); 1132 MD5Update(&Md5Ctx, pszRealm, strlen(pszRealm)); 1133 MD5Update(&Md5Ctx, ":", 1); 1134 MD5Update(&Md5Ctx, pszPassword, strlen(pszPassword)); 1135 MD5Final(HA1, &Md5Ctx); 1136 if (strcasecmp(pszAlg, "md5-sess") == 0) { 1137 1138 MD5Init(&Md5Ctx); 1139 MD5Update(&Md5Ctx, HA1, HASHLEN); 1140 MD5Update(&Md5Ctx, ":", 1); 1141 MD5Update(&Md5Ctx, pszNonce, strlen(pszNonce)); 1142 MD5Update(&Md5Ctx, ":", 1); 1143 MD5Update(&Md5Ctx, pszCNonce, strlen(pszCNonce)); 1144 MD5Final(HA1, &Md5Ctx); 1145 } 1146 CvtHex(HA1, SessionKey); 1147 } 1148 1149 /* calculate request-digest/response-digest as per HTTP Digest spec */ 1150 static void 1151 DigestCalcResponse( 1152 IN HASHHEX HA1, /* H(A1) */ 1153 IN char * pszNonce, /* nonce from server */ 1154 IN char * pszNonceCount, /* 8 hex digits */ 1155 IN char * pszCNonce, /* client nonce */ 1156 IN char * pszQop, /* qop-value: "", "auth", "auth-int" */ 1157 IN char * pszMethod, /* method from the request */ 1158 IN char * pszDigestUri, /* requested URL */ 1159 IN HASHHEX HEntity, /* H(entity body) if qop="auth-int" */ 1160 OUT HASHHEX Response /* request-digest or response-digest */ 1161 ) 1162 { 1163 #if 0 1164 DEBUGF("Calc: HA1[%s] Nonce[%s] qop[%s] method[%s] URI[%s]\n", 1165 HA1, pszNonce, pszQop, pszMethod, pszDigestUri); 1166 #endif 1167 MD5_CTX Md5Ctx; 1168 HASH HA2; 1169 HASH RespHash; 1170 HASHHEX HA2Hex; 1171 1172 // calculate H(A2) 1173 MD5Init(&Md5Ctx); 1174 MD5Update(&Md5Ctx, pszMethod, strlen(pszMethod)); 1175 MD5Update(&Md5Ctx, ":", 1); 1176 MD5Update(&Md5Ctx, pszDigestUri, strlen(pszDigestUri)); 1177 if (strcasecmp(pszQop, "auth-int") == 0) { 1178 MD5Update(&Md5Ctx, ":", 1); 1179 MD5Update(&Md5Ctx, HEntity, HASHHEXLEN); 1180 } 1181 MD5Final(HA2, &Md5Ctx); 1182 CvtHex(HA2, HA2Hex); 1183 1184 // calculate response 1185 MD5Init(&Md5Ctx); 1186 MD5Update(&Md5Ctx, HA1, HASHHEXLEN); 1187 MD5Update(&Md5Ctx, ":", 1); 1188 MD5Update(&Md5Ctx, pszNonce, strlen(pszNonce)); 1189 MD5Update(&Md5Ctx, ":", 1); 1190 if (*pszQop) { 1191 MD5Update(&Md5Ctx, pszNonceCount, strlen(pszNonceCount)); 1192 MD5Update(&Md5Ctx, ":", 1); 1193 MD5Update(&Md5Ctx, pszCNonce, strlen(pszCNonce)); 1194 MD5Update(&Md5Ctx, ":", 1); 1195 MD5Update(&Md5Ctx, pszQop, strlen(pszQop)); 1196 MD5Update(&Md5Ctx, ":", 1); 1197 } 1198 MD5Update(&Md5Ctx, HA2Hex, HASHHEXLEN); 1199 MD5Final(RespHash, &Md5Ctx); 1200 CvtHex(RespHash, Response); 1201 } 1202 1203 /* 1204 * Generate/Send a Digest authorization header 1205 * This looks like: [Proxy-]Authorization: credentials 1206 * 1207 * credentials = "Digest" digest-response 1208 * digest-response = 1#( username | realm | nonce | digest-uri 1209 * | response | [ algorithm ] | [cnonce] | 1210 * [opaque] | [message-qop] | 1211 * [nonce-count] | [auth-param] ) 1212 * username = "username" "=" username-value 1213 * username-value = quoted-string 1214 * digest-uri = "uri" "=" digest-uri-value 1215 * digest-uri-value = request-uri ; As specified by HTTP/1.1 1216 * message-qop = "qop" "=" qop-value 1217 * cnonce = "cnonce" "=" cnonce-value 1218 * cnonce-value = nonce-value 1219 * nonce-count = "nc" "=" nc-value 1220 * nc-value = 8LHEX 1221 * response = "response" "=" request-digest 1222 * request-digest = <"> 32LHEX <"> 1223 */ 1224 static int 1225 http_digest_auth(conn_t *conn, const char *hdr, http_auth_challenge_t *c, 1226 http_auth_params_t *parms, struct url *url) 1227 { 1228 int r; 1229 char noncecount[10]; 1230 char cnonce[40]; 1231 char *options = NULL; 1232 1233 if (!c->realm || !c->nonce) { 1234 DEBUGF("realm/nonce not set in challenge\n"); 1235 return(-1); 1236 } 1237 if (!c->algo) 1238 c->algo = strdup(""); 1239 1240 if (asprintf(&options, "%s%s%s%s", 1241 *c->algo? ",algorithm=" : "", c->algo, 1242 c->opaque? ",opaque=" : "", c->opaque?c->opaque:"") < 0) 1243 return (-1); 1244 1245 if (!c->qop) { 1246 c->qop = strdup(""); 1247 *noncecount = 0; 1248 *cnonce = 0; 1249 } else { 1250 c->nc++; 1251 sprintf(noncecount, "%08x", c->nc); 1252 /* We don't try very hard with the cnonce ... */ 1253 sprintf(cnonce, "%x%lx", getpid(), (unsigned long)time(0)); 1254 } 1255 1256 HASHHEX HA1; 1257 DigestCalcHA1(c->algo, parms->user, c->realm, 1258 parms->password, c->nonce, cnonce, HA1); 1259 DEBUGF("HA1: [%s]\n", HA1); 1260 HASHHEX digest, null; 1261 memset(null, 0, sizeof(null)); 1262 DigestCalcResponse(HA1, c->nonce, noncecount, cnonce, c->qop, 1263 "GET", url->doc, null, digest); 1264 1265 if (c->qop[0]) { 1266 r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\"," 1267 "nonce=\"%s\",uri=\"%s\",response=\"%s\"," 1268 "qop=\"auth\", cnonce=\"%s\", nc=%s%s", 1269 hdr, parms->user, c->realm, 1270 c->nonce, url->doc, digest, 1271 cnonce, noncecount, options); 1272 } else { 1273 r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\"," 1274 "nonce=\"%s\",uri=\"%s\",response=\"%s\"%s", 1275 hdr, parms->user, c->realm, 1276 c->nonce, url->doc, digest, options); 1277 } 1278 if (options) 1279 free(options); 1280 return (r); 1281 } 1282 1283 /* 1284 * Encode username and password 1285 */ 1286 static int 1287 http_basic_auth(conn_t *conn, const char *hdr, const char *usr, const char *pwd) 1288 { 1289 char *upw, *auth; 1290 int r; 1291 1292 DEBUGF("basic: usr: [%s]\n", usr); 1293 DEBUGF("basic: pwd: [%s]\n", pwd); 1294 if (asprintf(&upw, "%s:%s", usr, pwd) == -1) 1295 return (-1); 1296 auth = http_base64(upw); 1297 free(upw); 1298 if (auth == NULL) 1299 return (-1); 1300 r = http_cmd(conn, "%s: Basic %s", hdr, auth); 1301 free(auth); 1302 return (r); 1303 } 1304 1305 /* 1306 * Chose the challenge to answer and call the appropriate routine to 1307 * produce the header. 1308 */ 1309 static int 1310 http_authorize(conn_t *conn, const char *hdr, http_auth_challenges_t *cs, 1311 http_auth_params_t *parms, struct url *url) 1312 { 1313 http_auth_challenge_t *digest = NULL; 1314 int i; 1315 1316 /* If user or pass are null we're not happy */ 1317 if (!parms->user || !parms->password) { 1318 DEBUGF("NULL usr or pass\n"); 1319 return (-1); 1320 } 1321 1322 /* Look for a Digest */ 1323 for (i = 0; i < cs->count; i++) { 1324 if (cs->challenges[i]->scheme == HTTPAS_DIGEST) 1325 digest = cs->challenges[i]; 1326 } 1327 1328 /* Error if "Digest" was specified and there is no Digest challenge */ 1329 if (!digest && 1330 (parms->scheme && strcasecmp(parms->scheme, "digest") == 0)) { 1331 DEBUGF("Digest auth in env, not supported by peer\n"); 1332 return (-1); 1333 } 1334 /* 1335 * If "basic" was specified in the environment, or there is no Digest 1336 * challenge, do the basic thing. Don't need a challenge for this, 1337 * so no need to check basic!=NULL 1338 */ 1339 if (!digest || 1340 (parms->scheme && strcasecmp(parms->scheme, "basic") == 0)) 1341 return (http_basic_auth(conn,hdr,parms->user,parms->password)); 1342 1343 /* Else, prefer digest. We just checked that it's not NULL */ 1344 return (http_digest_auth(conn, hdr, digest, parms, url)); 1345 } 1346 1347 /***************************************************************************** 1348 * Helper functions for connecting to a server or proxy 1349 */ 1350 1351 /* 1352 * Connect to the correct HTTP server or proxy. 1353 */ 1354 static conn_t * 1355 http_connect(struct url *URL, struct url *purl, const char *flags) 1356 { 1357 struct url *curl; 1358 conn_t *conn; 1359 hdr_t h; 1360 http_headerbuf_t headerbuf; 1361 const char *p; 1362 int verbose; 1363 int af, val; 1364 int serrno; 1365 bool isproxyauth = false; 1366 http_auth_challenges_t proxy_challenges; 1367 1368 #ifdef INET6 1369 af = AF_UNSPEC; 1370 #else 1371 af = AF_INET; 1372 #endif 1373 1374 verbose = CHECK_FLAG('v'); 1375 if (CHECK_FLAG('4')) 1376 af = AF_INET; 1377 #ifdef INET6 1378 else if (CHECK_FLAG('6')) 1379 af = AF_INET6; 1380 #endif 1381 1382 curl = (purl != NULL) ? purl : URL; 1383 1384 retry: 1385 if ((conn = fetch_connect(curl->host, curl->port, af, verbose)) == NULL) 1386 /* fetch_connect() has already set an error code */ 1387 return (NULL); 1388 init_http_headerbuf(&headerbuf); 1389 if (strcmp(URL->scheme, SCHEME_HTTPS) == 0 && purl) { 1390 init_http_auth_challenges(&proxy_challenges); 1391 http_cmd(conn, "CONNECT %s:%d HTTP/1.1", URL->host, URL->port); 1392 http_cmd(conn, "Host: %s:%d", URL->host, URL->port); 1393 if (isproxyauth) { 1394 http_auth_params_t aparams; 1395 init_http_auth_params(&aparams); 1396 if (*purl->user || *purl->pwd) { 1397 aparams.user = strdup(purl->user); 1398 aparams.password = strdup(purl->pwd); 1399 } else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL && 1400 *p != '\0') { 1401 if (http_authfromenv(p, &aparams) < 0) { 1402 http_seterr(HTTP_NEED_PROXY_AUTH); 1403 fetch_syserr(); 1404 goto ouch; 1405 } 1406 } else if (fetch_netrc_auth(purl) == 0) { 1407 aparams.user = strdup(purl->user); 1408 aparams.password = strdup(purl->pwd); 1409 } else { 1410 /* 1411 * No auth information found in system - exiting 1412 * with warning. 1413 */ 1414 warnx("Missing username and/or password set"); 1415 fetch_syserr(); 1416 goto ouch; 1417 } 1418 http_authorize(conn, "Proxy-Authorization", 1419 &proxy_challenges, &aparams, purl); 1420 clean_http_auth_params(&aparams); 1421 } 1422 http_cmd(conn, ""); 1423 /* Get reply from CONNECT Tunnel attempt */ 1424 int httpreply = http_get_reply(conn); 1425 if (httpreply != HTTP_OK) { 1426 http_seterr(httpreply); 1427 /* If the error is a 407/HTTP_NEED_PROXY_AUTH */ 1428 if (httpreply == HTTP_NEED_PROXY_AUTH && 1429 ! isproxyauth) { 1430 /* Try again with authentication. */ 1431 clean_http_headerbuf(&headerbuf); 1432 fetch_close(conn); 1433 isproxyauth = true; 1434 goto retry; 1435 } 1436 goto ouch; 1437 } 1438 /* Read and discard the rest of the proxy response */ 1439 if (fetch_getln(conn) < 0) { 1440 fetch_syserr(); 1441 goto ouch; 1442 } 1443 do { 1444 switch ((h = http_next_header(conn, &headerbuf, &p))) { 1445 case hdr_syserror: 1446 fetch_syserr(); 1447 goto ouch; 1448 case hdr_error: 1449 http_seterr(HTTP_PROTOCOL_ERROR); 1450 goto ouch; 1451 default: 1452 /* ignore */ ; 1453 } 1454 } while (h > hdr_end); 1455 } 1456 if (strcmp(URL->scheme, SCHEME_HTTPS) == 0 && 1457 fetch_ssl(conn, URL, verbose) == -1) { 1458 /* grrr */ 1459 errno = EAUTH; 1460 fetch_syserr(); 1461 goto ouch; 1462 } 1463 1464 val = 1; 1465 setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val, sizeof(val)); 1466 1467 clean_http_headerbuf(&headerbuf); 1468 return (conn); 1469 ouch: 1470 serrno = errno; 1471 clean_http_headerbuf(&headerbuf); 1472 fetch_close(conn); 1473 errno = serrno; 1474 return (NULL); 1475 } 1476 1477 static struct url * 1478 http_get_proxy(struct url * url, const char *flags) 1479 { 1480 struct url *purl; 1481 char *p; 1482 1483 if (flags != NULL && strchr(flags, 'd') != NULL) 1484 return (NULL); 1485 if (fetch_no_proxy_match(url->host)) 1486 return (NULL); 1487 if (((p = getenv("HTTP_PROXY")) || (p = getenv("http_proxy"))) && 1488 *p && (purl = fetchParseURL(p))) { 1489 if (!*purl->scheme) 1490 strcpy(purl->scheme, SCHEME_HTTP); 1491 if (!purl->port) 1492 purl->port = fetch_default_proxy_port(purl->scheme); 1493 if (strcmp(purl->scheme, SCHEME_HTTP) == 0) 1494 return (purl); 1495 fetchFreeURL(purl); 1496 } 1497 return (NULL); 1498 } 1499 1500 static void 1501 http_print_html(FILE *out, FILE *in) 1502 { 1503 ssize_t len = 0; 1504 size_t cap; 1505 char *line = NULL, *p, *q; 1506 int comment, tag; 1507 1508 comment = tag = 0; 1509 while ((len = getline(&line, &cap, in)) >= 0) { 1510 while (len && isspace((unsigned char)line[len - 1])) 1511 --len; 1512 for (p = q = line; q < line + len; ++q) { 1513 if (comment && *q == '-') { 1514 if (q + 2 < line + len && 1515 strcmp(q, "-->") == 0) { 1516 tag = comment = 0; 1517 q += 2; 1518 } 1519 } else if (tag && !comment && *q == '>') { 1520 p = q + 1; 1521 tag = 0; 1522 } else if (!tag && *q == '<') { 1523 if (q > p) 1524 fwrite(p, q - p, 1, out); 1525 tag = 1; 1526 if (q + 3 < line + len && 1527 strcmp(q, "<!--") == 0) { 1528 comment = 1; 1529 q += 3; 1530 } 1531 } 1532 } 1533 if (!tag && q > p) 1534 fwrite(p, q - p, 1, out); 1535 fputc('\n', out); 1536 } 1537 1538 free(line); 1539 } 1540 1541 1542 /***************************************************************************** 1543 * Core 1544 */ 1545 1546 FILE * 1547 http_request(struct url *URL, const char *op, struct url_stat *us, 1548 struct url *purl, const char *flags) 1549 { 1550 1551 return (http_request_body(URL, op, us, purl, flags, NULL, NULL)); 1552 } 1553 1554 /* 1555 * Send a request and process the reply 1556 * 1557 * XXX This function is way too long, the do..while loop should be split 1558 * XXX off into a separate function. 1559 */ 1560 FILE * 1561 http_request_body(struct url *URL, const char *op, struct url_stat *us, 1562 struct url *purl, const char *flags, const char *content_type, 1563 const char *body) 1564 { 1565 char timebuf[80]; 1566 char hbuf[MAXHOSTNAMELEN + 7], *host; 1567 conn_t *conn; 1568 struct url *url, *new; 1569 int chunked, direct, ims, noredirect, verbose; 1570 int e, i, n, val; 1571 off_t offset, clength, length, size; 1572 time_t mtime; 1573 const char *p; 1574 FILE *f; 1575 hdr_t h; 1576 struct tm *timestruct; 1577 http_headerbuf_t headerbuf; 1578 http_auth_challenges_t server_challenges; 1579 http_auth_challenges_t proxy_challenges; 1580 size_t body_len; 1581 1582 /* The following calls don't allocate anything */ 1583 init_http_headerbuf(&headerbuf); 1584 init_http_auth_challenges(&server_challenges); 1585 init_http_auth_challenges(&proxy_challenges); 1586 1587 direct = CHECK_FLAG('d'); 1588 noredirect = CHECK_FLAG('A'); 1589 verbose = CHECK_FLAG('v'); 1590 ims = CHECK_FLAG('i'); 1591 1592 if (direct && purl) { 1593 fetchFreeURL(purl); 1594 purl = NULL; 1595 } 1596 1597 /* try the provided URL first */ 1598 url = URL; 1599 1600 n = MAX_REDIRECT; 1601 i = 0; 1602 1603 e = HTTP_PROTOCOL_ERROR; 1604 do { 1605 new = NULL; 1606 chunked = 0; 1607 offset = 0; 1608 clength = -1; 1609 length = -1; 1610 size = -1; 1611 mtime = 0; 1612 1613 /* check port */ 1614 if (!url->port) 1615 url->port = fetch_default_port(url->scheme); 1616 1617 /* were we redirected to an FTP URL? */ 1618 if (purl == NULL && strcmp(url->scheme, SCHEME_FTP) == 0) { 1619 if (strcmp(op, "GET") == 0) 1620 return (ftp_request(url, "RETR", us, purl, flags)); 1621 else if (strcmp(op, "HEAD") == 0) 1622 return (ftp_request(url, "STAT", us, purl, flags)); 1623 } 1624 1625 /* connect to server or proxy */ 1626 if ((conn = http_connect(url, purl, flags)) == NULL) 1627 goto ouch; 1628 1629 /* append port number only if necessary */ 1630 host = url->host; 1631 if (url->port != fetch_default_port(url->scheme)) { 1632 snprintf(hbuf, sizeof(hbuf), "%s:%d", host, url->port); 1633 host = hbuf; 1634 } 1635 1636 /* send request */ 1637 if (verbose) 1638 fetch_info("requesting %s://%s%s", 1639 url->scheme, host, url->doc); 1640 if (purl && strcmp(url->scheme, SCHEME_HTTPS) != 0) { 1641 http_cmd(conn, "%s %s://%s%s HTTP/1.1", 1642 op, url->scheme, host, url->doc); 1643 } else { 1644 http_cmd(conn, "%s %s HTTP/1.1", 1645 op, url->doc); 1646 } 1647 1648 if (ims && url->ims_time) { 1649 timestruct = gmtime((time_t *)&url->ims_time); 1650 (void)strftime(timebuf, 80, "%a, %d %b %Y %T GMT", 1651 timestruct); 1652 if (verbose) 1653 fetch_info("If-Modified-Since: %s", timebuf); 1654 http_cmd(conn, "If-Modified-Since: %s", timebuf); 1655 } 1656 /* virtual host */ 1657 http_cmd(conn, "Host: %s", host); 1658 1659 /* 1660 * Proxy authorization: we only send auth after we received 1661 * a 407 error. We do not first try basic anyway (changed 1662 * when support was added for digest-auth) 1663 */ 1664 if (purl && proxy_challenges.valid) { 1665 http_auth_params_t aparams; 1666 init_http_auth_params(&aparams); 1667 if (*purl->user || *purl->pwd) { 1668 aparams.user = strdup(purl->user); 1669 aparams.password = strdup(purl->pwd); 1670 } else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL && 1671 *p != '\0') { 1672 if (http_authfromenv(p, &aparams) < 0) { 1673 http_seterr(HTTP_NEED_PROXY_AUTH); 1674 goto ouch; 1675 } 1676 } else if (fetch_netrc_auth(purl) == 0) { 1677 aparams.user = strdup(purl->user); 1678 aparams.password = strdup(purl->pwd); 1679 } 1680 http_authorize(conn, "Proxy-Authorization", 1681 &proxy_challenges, &aparams, url); 1682 clean_http_auth_params(&aparams); 1683 } 1684 1685 /* 1686 * Server authorization: we never send "a priori" 1687 * Basic auth, which used to be done if user/pass were 1688 * set in the url. This would be weird because we'd send the 1689 * password in the clear even if Digest is finally to be 1690 * used (it would have made more sense for the 1691 * pre-digest version to do this when Basic was specified 1692 * in the environment) 1693 */ 1694 if (server_challenges.valid) { 1695 http_auth_params_t aparams; 1696 init_http_auth_params(&aparams); 1697 if (*url->user || *url->pwd) { 1698 aparams.user = strdup(url->user); 1699 aparams.password = strdup(url->pwd); 1700 } else if ((p = getenv("HTTP_AUTH")) != NULL && 1701 *p != '\0') { 1702 if (http_authfromenv(p, &aparams) < 0) { 1703 http_seterr(HTTP_NEED_AUTH); 1704 goto ouch; 1705 } 1706 } else if (fetch_netrc_auth(url) == 0) { 1707 aparams.user = strdup(url->user); 1708 aparams.password = strdup(url->pwd); 1709 } else if (fetchAuthMethod && 1710 fetchAuthMethod(url) == 0) { 1711 aparams.user = strdup(url->user); 1712 aparams.password = strdup(url->pwd); 1713 } else { 1714 http_seterr(HTTP_NEED_AUTH); 1715 goto ouch; 1716 } 1717 http_authorize(conn, "Authorization", 1718 &server_challenges, &aparams, url); 1719 clean_http_auth_params(&aparams); 1720 } 1721 1722 /* other headers */ 1723 if ((p = getenv("HTTP_ACCEPT")) != NULL) { 1724 if (*p != '\0') 1725 http_cmd(conn, "Accept: %s", p); 1726 } else { 1727 http_cmd(conn, "Accept: */*"); 1728 } 1729 if ((p = getenv("HTTP_REFERER")) != NULL && *p != '\0') { 1730 if (strcasecmp(p, "auto") == 0) 1731 http_cmd(conn, "Referer: %s://%s%s", 1732 url->scheme, host, url->doc); 1733 else 1734 http_cmd(conn, "Referer: %s", p); 1735 } 1736 if ((p = getenv("HTTP_USER_AGENT")) != NULL) { 1737 /* no User-Agent if defined but empty */ 1738 if (*p != '\0') 1739 http_cmd(conn, "User-Agent: %s", p); 1740 } else { 1741 /* default User-Agent */ 1742 http_cmd(conn, "User-Agent: %s " _LIBFETCH_VER, 1743 getprogname()); 1744 } 1745 if (url->offset > 0) 1746 http_cmd(conn, "Range: bytes=%lld-", (long long)url->offset); 1747 http_cmd(conn, "Connection: close"); 1748 1749 if (body) { 1750 body_len = strlen(body); 1751 http_cmd(conn, "Content-Length: %zu", body_len); 1752 if (content_type != NULL) 1753 http_cmd(conn, "Content-Type: %s", content_type); 1754 } 1755 1756 http_cmd(conn, ""); 1757 1758 if (body) 1759 fetch_write(conn, body, body_len); 1760 1761 /* 1762 * Force the queued request to be dispatched. Normally, one 1763 * would do this with shutdown(2) but squid proxies can be 1764 * configured to disallow such half-closed connections. To 1765 * be compatible with such configurations, fiddle with socket 1766 * options to force the pending data to be written. 1767 */ 1768 val = 0; 1769 setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val, 1770 sizeof(val)); 1771 val = 1; 1772 setsockopt(conn->sd, IPPROTO_TCP, TCP_NODELAY, &val, 1773 sizeof(val)); 1774 1775 /* get reply */ 1776 switch (http_get_reply(conn)) { 1777 case HTTP_OK: 1778 case HTTP_PARTIAL: 1779 case HTTP_NOT_MODIFIED: 1780 /* fine */ 1781 break; 1782 case HTTP_MOVED_PERM: 1783 case HTTP_MOVED_TEMP: 1784 case HTTP_TEMP_REDIRECT: 1785 case HTTP_PERM_REDIRECT: 1786 case HTTP_SEE_OTHER: 1787 case HTTP_USE_PROXY: 1788 /* 1789 * Not so fine, but we still have to read the 1790 * headers to get the new location. 1791 */ 1792 break; 1793 case HTTP_NEED_AUTH: 1794 if (server_challenges.valid) { 1795 /* 1796 * We already sent out authorization code, 1797 * so there's nothing more we can do. 1798 */ 1799 http_seterr(conn->err); 1800 goto ouch; 1801 } 1802 /* try again, but send the password this time */ 1803 if (verbose) 1804 fetch_info("server requires authorization"); 1805 break; 1806 case HTTP_NEED_PROXY_AUTH: 1807 if (proxy_challenges.valid) { 1808 /* 1809 * We already sent our proxy 1810 * authorization code, so there's 1811 * nothing more we can do. */ 1812 http_seterr(conn->err); 1813 goto ouch; 1814 } 1815 /* try again, but send the password this time */ 1816 if (verbose) 1817 fetch_info("proxy requires authorization"); 1818 break; 1819 case HTTP_BAD_RANGE: 1820 /* 1821 * This can happen if we ask for 0 bytes because 1822 * we already have the whole file. Consider this 1823 * a success for now, and check sizes later. 1824 */ 1825 break; 1826 case HTTP_PROTOCOL_ERROR: 1827 /* fall through */ 1828 case -1: 1829 fetch_syserr(); 1830 goto ouch; 1831 default: 1832 http_seterr(conn->err); 1833 if (!verbose) 1834 goto ouch; 1835 /* fall through so we can get the full error message */ 1836 } 1837 1838 /* get headers. http_next_header expects one line readahead */ 1839 if (fetch_getln(conn) == -1) { 1840 fetch_syserr(); 1841 goto ouch; 1842 } 1843 do { 1844 switch ((h = http_next_header(conn, &headerbuf, &p))) { 1845 case hdr_syserror: 1846 fetch_syserr(); 1847 goto ouch; 1848 case hdr_error: 1849 http_seterr(HTTP_PROTOCOL_ERROR); 1850 goto ouch; 1851 case hdr_content_length: 1852 http_parse_length(p, &clength); 1853 break; 1854 case hdr_content_range: 1855 http_parse_range(p, &offset, &length, &size); 1856 break; 1857 case hdr_last_modified: 1858 http_parse_mtime(p, &mtime); 1859 break; 1860 case hdr_location: 1861 if (!HTTP_REDIRECT(conn->err)) 1862 break; 1863 /* 1864 * if the A flag is set, we don't follow 1865 * temporary redirects. 1866 */ 1867 if (noredirect && 1868 conn->err != HTTP_MOVED_PERM && 1869 conn->err != HTTP_PERM_REDIRECT && 1870 conn->err != HTTP_USE_PROXY) { 1871 n = 1; 1872 break; 1873 } 1874 if (new) 1875 free(new); 1876 if (verbose) 1877 fetch_info("%d redirect to %s", 1878 conn->err, p); 1879 if (*p == '/') 1880 /* absolute path */ 1881 new = fetchMakeURL(url->scheme, url->host, 1882 url->port, p, url->user, url->pwd); 1883 else 1884 new = fetchParseURL(p); 1885 if (new == NULL) { 1886 /* XXX should set an error code */ 1887 DEBUGF("failed to parse new URL\n"); 1888 goto ouch; 1889 } 1890 1891 /* Only copy credentials if the host matches */ 1892 if (strcmp(new->host, url->host) == 0 && 1893 !*new->user && !*new->pwd) { 1894 strcpy(new->user, url->user); 1895 strcpy(new->pwd, url->pwd); 1896 } 1897 new->offset = url->offset; 1898 new->length = url->length; 1899 new->ims_time = url->ims_time; 1900 break; 1901 case hdr_transfer_encoding: 1902 /* XXX weak test*/ 1903 chunked = (strcasecmp(p, "chunked") == 0); 1904 break; 1905 case hdr_www_authenticate: 1906 if (conn->err != HTTP_NEED_AUTH) 1907 break; 1908 if (http_parse_authenticate(p, &server_challenges) == 0) 1909 ++n; 1910 break; 1911 case hdr_proxy_authenticate: 1912 if (conn->err != HTTP_NEED_PROXY_AUTH) 1913 break; 1914 if (http_parse_authenticate(p, &proxy_challenges) == 0) 1915 ++n; 1916 break; 1917 case hdr_end: 1918 /* fall through */ 1919 case hdr_unknown: 1920 /* ignore */ 1921 break; 1922 } 1923 } while (h > hdr_end); 1924 1925 /* we need to provide authentication */ 1926 if (conn->err == HTTP_NEED_AUTH || 1927 conn->err == HTTP_NEED_PROXY_AUTH) { 1928 e = conn->err; 1929 if ((conn->err == HTTP_NEED_AUTH && 1930 !server_challenges.valid) || 1931 (conn->err == HTTP_NEED_PROXY_AUTH && 1932 !proxy_challenges.valid)) { 1933 /* 401/7 but no www/proxy-authenticate ?? */ 1934 DEBUGF("%03d without auth header\n", conn->err); 1935 goto ouch; 1936 } 1937 fetch_close(conn); 1938 conn = NULL; 1939 continue; 1940 } 1941 1942 /* requested range not satisfiable */ 1943 if (conn->err == HTTP_BAD_RANGE) { 1944 if (url->offset > 0 && url->length == 0) { 1945 /* asked for 0 bytes; fake it */ 1946 offset = url->offset; 1947 clength = -1; 1948 conn->err = HTTP_OK; 1949 break; 1950 } else { 1951 http_seterr(conn->err); 1952 goto ouch; 1953 } 1954 } 1955 1956 /* we have a hit or an error */ 1957 if (conn->err == HTTP_OK 1958 || conn->err == HTTP_NOT_MODIFIED 1959 || conn->err == HTTP_PARTIAL 1960 || HTTP_ERROR(conn->err)) 1961 break; 1962 1963 /* all other cases: we got a redirect */ 1964 e = conn->err; 1965 clean_http_auth_challenges(&server_challenges); 1966 fetch_close(conn); 1967 conn = NULL; 1968 if (!new) { 1969 DEBUGF("redirect with no new location\n"); 1970 break; 1971 } 1972 if (url != URL) 1973 fetchFreeURL(url); 1974 url = new; 1975 } while (++i < n); 1976 1977 /* we failed, or ran out of retries */ 1978 if (conn == NULL) { 1979 http_seterr(e); 1980 goto ouch; 1981 } 1982 1983 DEBUGF("offset %lld, length %lld, size %lld, clength %lld\n", 1984 (long long)offset, (long long)length, 1985 (long long)size, (long long)clength); 1986 1987 if (conn->err == HTTP_NOT_MODIFIED) { 1988 http_seterr(HTTP_NOT_MODIFIED); 1989 return (NULL); 1990 } 1991 1992 /* check for inconsistencies */ 1993 if (clength != -1 && length != -1 && clength != length) { 1994 http_seterr(HTTP_PROTOCOL_ERROR); 1995 goto ouch; 1996 } 1997 if (clength == -1) 1998 clength = length; 1999 if (clength != -1) 2000 length = offset + clength; 2001 if (length != -1 && size != -1 && length != size) { 2002 http_seterr(HTTP_PROTOCOL_ERROR); 2003 goto ouch; 2004 } 2005 if (size == -1) 2006 size = length; 2007 2008 /* fill in stats */ 2009 if (us) { 2010 us->size = size; 2011 us->atime = us->mtime = mtime; 2012 } 2013 2014 /* too far? */ 2015 if (URL->offset > 0 && offset > URL->offset) { 2016 http_seterr(HTTP_PROTOCOL_ERROR); 2017 goto ouch; 2018 } 2019 2020 /* report back real offset and size */ 2021 URL->offset = offset; 2022 URL->length = clength; 2023 2024 /* wrap it up in a FILE */ 2025 if ((f = http_funopen(conn, chunked)) == NULL) { 2026 fetch_syserr(); 2027 goto ouch; 2028 } 2029 2030 if (url != URL) 2031 fetchFreeURL(url); 2032 if (purl) 2033 fetchFreeURL(purl); 2034 2035 if (HTTP_ERROR(conn->err)) { 2036 http_print_html(stderr, f); 2037 fclose(f); 2038 f = NULL; 2039 } 2040 clean_http_headerbuf(&headerbuf); 2041 clean_http_auth_challenges(&server_challenges); 2042 clean_http_auth_challenges(&proxy_challenges); 2043 return (f); 2044 2045 ouch: 2046 if (url != URL) 2047 fetchFreeURL(url); 2048 if (purl) 2049 fetchFreeURL(purl); 2050 if (conn != NULL) 2051 fetch_close(conn); 2052 clean_http_headerbuf(&headerbuf); 2053 clean_http_auth_challenges(&server_challenges); 2054 clean_http_auth_challenges(&proxy_challenges); 2055 return (NULL); 2056 } 2057 2058 2059 /***************************************************************************** 2060 * Entry points 2061 */ 2062 2063 /* 2064 * Retrieve and stat a file by HTTP 2065 */ 2066 FILE * 2067 fetchXGetHTTP(struct url *URL, struct url_stat *us, const char *flags) 2068 { 2069 return (http_request(URL, "GET", us, http_get_proxy(URL, flags), flags)); 2070 } 2071 2072 /* 2073 * Retrieve a file by HTTP 2074 */ 2075 FILE * 2076 fetchGetHTTP(struct url *URL, const char *flags) 2077 { 2078 return (fetchXGetHTTP(URL, NULL, flags)); 2079 } 2080 2081 /* 2082 * Store a file by HTTP 2083 */ 2084 FILE * 2085 fetchPutHTTP(struct url *URL __unused, const char *flags __unused) 2086 { 2087 warnx("fetchPutHTTP(): not implemented"); 2088 return (NULL); 2089 } 2090 2091 /* 2092 * Get an HTTP document's metadata 2093 */ 2094 int 2095 fetchStatHTTP(struct url *URL, struct url_stat *us, const char *flags) 2096 { 2097 FILE *f; 2098 2099 f = http_request(URL, "HEAD", us, http_get_proxy(URL, flags), flags); 2100 if (f == NULL) 2101 return (-1); 2102 fclose(f); 2103 return (0); 2104 } 2105 2106 /* 2107 * List a directory 2108 */ 2109 struct url_ent * 2110 fetchListHTTP(struct url *url __unused, const char *flags __unused) 2111 { 2112 warnx("fetchListHTTP(): not implemented"); 2113 return (NULL); 2114 } 2115 2116 /* 2117 * Arbitrary HTTP verb and content requests 2118 */ 2119 FILE * 2120 fetchReqHTTP(struct url *URL, const char *method, const char *flags, 2121 const char *content_type, const char *body) 2122 { 2123 2124 return (http_request_body(URL, method, NULL, http_get_proxy(URL, flags), 2125 flags, content_type, body)); 2126 } 2127