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 { 131 conn_t *conn; /* connection */ 132 int chunked; /* chunked mode */ 133 char *buf; /* chunk buffer */ 134 size_t bufsize; /* size of chunk buffer */ 135 size_t buflen; /* amount of data currently in buffer */ 136 size_t bufpos; /* current read offset in buffer */ 137 int eof; /* end-of-file flag */ 138 int error; /* error flag */ 139 size_t chunksize; /* remaining size of current chunk */ 140 #ifndef NDEBUG 141 size_t total; 142 #endif 143 }; 144 145 /* 146 * Get next chunk header 147 */ 148 static int 149 http_new_chunk(struct httpio *io) 150 { 151 unsigned char *p, *eol; 152 153 if (fetch_getln(io->conn) == -1) 154 return (-1); 155 156 p = (unsigned char *)io->conn->buf; 157 if (io->conn->pos < 2 || !isxdigit(*p)) 158 return (-1); 159 160 eol = (unsigned char *)io->conn->buf + io->conn->pos; 161 while (p < eol && !isspace(*p)) { 162 if (*p == ';') { 163 break; 164 } else if (*p >= '0' && *p <= '9') { 165 io->chunksize = io->chunksize * 16 + 166 *p - '0'; 167 } else if (*p >= 'A' && *p <= 'F') { 168 io->chunksize = io->chunksize * 16 + 169 10 + tolower(*p) - 'A'; 170 } else if (*p >= 'a' && *p <= 'f') { 171 io->chunksize = io->chunksize * 16 + 172 10 + tolower(*p) - 'a'; 173 } else { 174 return (-1); 175 } 176 } 177 178 #ifndef NDEBUG 179 if (fetchDebug) { 180 io->total += io->chunksize; 181 if (io->chunksize == 0) 182 fprintf(stderr, "%s(): end of last chunk\n", __func__); 183 else 184 fprintf(stderr, "%s(): new chunk: %lu (%lu)\n", 185 __func__, (unsigned long)io->chunksize, 186 (unsigned long)io->total); 187 } 188 #endif 189 190 return (io->chunksize); 191 } 192 193 /* 194 * Grow the input buffer to at least len bytes 195 */ 196 static inline int 197 http_growbuf(struct httpio *io, size_t len) 198 { 199 char *tmp; 200 201 if (io->bufsize >= len) 202 return (0); 203 204 if ((tmp = realloc(io->buf, len)) == NULL) 205 return (-1); 206 io->buf = tmp; 207 io->bufsize = len; 208 return (0); 209 } 210 211 /* 212 * Fill the input buffer, do chunk decoding on the fly 213 */ 214 static ssize_t 215 http_fillbuf(struct httpio *io, size_t len) 216 { 217 ssize_t nbytes; 218 char ch; 219 220 if (io->error) 221 return (-1); 222 if (io->eof) 223 return (0); 224 225 /* not chunked: just fetch the requested amount */ 226 if (io->chunked == 0) { 227 if (http_growbuf(io, len) == -1) 228 return (-1); 229 if ((nbytes = fetch_bufread(io->conn, io->buf, len)) == -1) { 230 io->error = errno; 231 return (-1); 232 } 233 io->buflen = nbytes; 234 io->bufpos = 0; 235 return (io->buflen); 236 } 237 238 /* chunked, but we ran out: get the next chunk header */ 239 if (io->chunksize == 0) { 240 switch (http_new_chunk(io)) { 241 case -1: 242 io->error = EPROTO; 243 return (-1); 244 case 0: 245 io->eof = 1; 246 return (0); 247 } 248 } 249 250 /* fetch the requested amount, but no more than the current chunk */ 251 if (len > io->chunksize) 252 len = io->chunksize; 253 if (http_growbuf(io, len) == -1) 254 return (-1); 255 if ((nbytes = fetch_bufread(io->conn, io->buf, len)) == -1) { 256 io->error = errno; 257 return (-1); 258 } 259 io->bufpos = 0; 260 io->buflen = nbytes; 261 io->chunksize -= nbytes; 262 263 if (io->chunksize == 0) { 264 if (fetch_bufread(io->conn, &ch, 1) != 1 || ch != '\r' || 265 fetch_bufread(io->conn, &ch, 1) != 1 || ch != '\n') 266 return (-1); 267 } 268 269 return (io->buflen); 270 } 271 272 /* 273 * Read function 274 */ 275 static int 276 http_readfn(void *v, char *buf, int len) 277 { 278 struct httpio *io = (struct httpio *)v; 279 int rlen; 280 281 if (io->error) 282 return (-1); 283 if (io->eof) 284 return (0); 285 286 /* empty buffer */ 287 if (!io->buf || io->bufpos == io->buflen) { 288 if ((rlen = http_fillbuf(io, len)) < 0) { 289 if ((errno = io->error) == EINTR) 290 io->error = 0; 291 return (-1); 292 } else if (rlen == 0) { 293 return (0); 294 } 295 } 296 297 rlen = io->buflen - io->bufpos; 298 if (len < rlen) 299 rlen = len; 300 memcpy(buf, io->buf + io->bufpos, rlen); 301 io->bufpos += rlen; 302 return (rlen); 303 } 304 305 /* 306 * Write function 307 */ 308 static int 309 http_writefn(void *v, const char *buf, int len) 310 { 311 struct httpio *io = (struct httpio *)v; 312 313 return (fetch_write(io->conn, buf, len)); 314 } 315 316 /* 317 * Close function 318 */ 319 static int 320 http_closefn(void *v) 321 { 322 struct httpio *io = (struct httpio *)v; 323 int r; 324 325 r = fetch_close(io->conn); 326 if (io->buf) 327 free(io->buf); 328 free(io); 329 return (r); 330 } 331 332 /* 333 * Wrap a file descriptor up 334 */ 335 static FILE * 336 http_funopen(conn_t *conn, int chunked) 337 { 338 struct httpio *io; 339 FILE *f; 340 341 if ((io = calloc(1, sizeof(*io))) == NULL) { 342 fetch_syserr(); 343 return (NULL); 344 } 345 io->conn = conn; 346 io->chunked = chunked; 347 f = funopen(io, http_readfn, http_writefn, NULL, http_closefn); 348 if (f == NULL) { 349 fetch_syserr(); 350 free(io); 351 return (NULL); 352 } 353 return (f); 354 } 355 356 357 /***************************************************************************** 358 * Helper functions for talking to the server and parsing its replies 359 */ 360 361 /* Header types */ 362 typedef enum { 363 hdr_syserror = -2, 364 hdr_error = -1, 365 hdr_end = 0, 366 hdr_unknown = 1, 367 hdr_content_length, 368 hdr_content_range, 369 hdr_last_modified, 370 hdr_location, 371 hdr_transfer_encoding, 372 hdr_www_authenticate, 373 hdr_proxy_authenticate, 374 } hdr_t; 375 376 /* Names of interesting headers */ 377 static struct { 378 hdr_t num; 379 const char *name; 380 } hdr_names[] = { 381 { hdr_content_length, "Content-Length" }, 382 { hdr_content_range, "Content-Range" }, 383 { hdr_last_modified, "Last-Modified" }, 384 { hdr_location, "Location" }, 385 { hdr_transfer_encoding, "Transfer-Encoding" }, 386 { hdr_www_authenticate, "WWW-Authenticate" }, 387 { hdr_proxy_authenticate, "Proxy-Authenticate" }, 388 { hdr_unknown, NULL }, 389 }; 390 391 /* 392 * Send a formatted line; optionally echo to terminal 393 */ 394 static int 395 http_cmd(conn_t *conn, const char *fmt, ...) 396 { 397 va_list ap; 398 size_t len; 399 char *msg; 400 int r; 401 402 va_start(ap, fmt); 403 len = vasprintf(&msg, fmt, ap); 404 va_end(ap); 405 406 if (msg == NULL) { 407 errno = ENOMEM; 408 fetch_syserr(); 409 return (-1); 410 } 411 412 r = fetch_putln(conn, msg, len); 413 free(msg); 414 415 if (r == -1) { 416 fetch_syserr(); 417 return (-1); 418 } 419 420 return (0); 421 } 422 423 /* 424 * Get and parse status line 425 */ 426 static int 427 http_get_reply(conn_t *conn) 428 { 429 char *p; 430 431 if (fetch_getln(conn) == -1) 432 return (-1); 433 /* 434 * A valid status line looks like "HTTP/m.n xyz reason" where m 435 * and n are the major and minor protocol version numbers and xyz 436 * is the reply code. 437 * Unfortunately, there are servers out there (NCSA 1.5.1, to name 438 * just one) that do not send a version number, so we can't rely 439 * on finding one, but if we do, insist on it being 1.0 or 1.1. 440 * We don't care about the reason phrase. 441 */ 442 if (strncmp(conn->buf, "HTTP", 4) != 0) 443 return (HTTP_PROTOCOL_ERROR); 444 p = conn->buf + 4; 445 if (*p == '/') { 446 if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1')) 447 return (HTTP_PROTOCOL_ERROR); 448 p += 4; 449 } 450 if (*p != ' ' || 451 !isdigit((unsigned char)p[1]) || 452 !isdigit((unsigned char)p[2]) || 453 !isdigit((unsigned char)p[3])) 454 return (HTTP_PROTOCOL_ERROR); 455 456 conn->err = (p[1] - '0') * 100 + (p[2] - '0') * 10 + (p[3] - '0'); 457 return (conn->err); 458 } 459 460 /* 461 * Check a header; if the type matches the given string, return a pointer 462 * to the beginning of the value. 463 */ 464 static const char * 465 http_match(const char *str, const char *hdr) 466 { 467 while (*str && *hdr && 468 tolower((unsigned char)*str++) == tolower((unsigned char)*hdr++)) 469 /* nothing */; 470 if (*str || *hdr != ':') 471 return (NULL); 472 while (*hdr && isspace((unsigned char)*++hdr)) 473 /* nothing */; 474 return (hdr); 475 } 476 477 478 /* 479 * Get the next header and return the appropriate symbolic code. We 480 * need to read one line ahead for checking for a continuation line 481 * belonging to the current header (continuation lines start with 482 * white space). 483 * 484 * We get called with a fresh line already in the conn buffer, either 485 * from the previous http_next_header() invocation, or, the first 486 * time, from a fetch_getln() performed by our caller. 487 * 488 * This stops when we encounter an empty line (we dont read beyond the header 489 * area). 490 * 491 * Note that the "headerbuf" is just a place to return the result. Its 492 * contents are not used for the next call. This means that no cleanup 493 * is needed when ie doing another connection, just call the cleanup when 494 * fully done to deallocate memory. 495 */ 496 497 /* Limit the max number of continuation lines to some reasonable value */ 498 #define HTTP_MAX_CONT_LINES 10 499 500 /* Place into which to build a header from one or several lines */ 501 typedef struct { 502 char *buf; /* buffer */ 503 size_t bufsize; /* buffer size */ 504 size_t buflen; /* length of buffer contents */ 505 } http_headerbuf_t; 506 507 static void 508 init_http_headerbuf(http_headerbuf_t *buf) 509 { 510 buf->buf = NULL; 511 buf->bufsize = 0; 512 buf->buflen = 0; 513 } 514 515 static void 516 clean_http_headerbuf(http_headerbuf_t *buf) 517 { 518 if (buf->buf) 519 free(buf->buf); 520 init_http_headerbuf(buf); 521 } 522 523 static hdr_t 524 http_next_header(conn_t *conn, http_headerbuf_t *hbuf, const char **p) 525 { 526 unsigned int i, len; 527 528 if (conn->pos == 0 || conn->buf[0] == '\0') 529 return (hdr_end); 530 531 /* Copy the line to the headerbuf */ 532 if (hbuf->bufsize < conn->pos + 1) { 533 if ((hbuf->buf = realloc(hbuf->buf, conn->pos + 1)) == NULL) 534 return (hdr_syserror); 535 hbuf->bufsize = conn->pos + 1; 536 } 537 strcpy(hbuf->buf, conn->buf); 538 hbuf->buflen = conn->buflen; 539 540 /* 541 * Fetch possible continuation lines. Stop at 1st non-continuation 542 * and leave it in the conn buffer 543 */ 544 for (i = 0; i < HTTP_MAX_CONT_LINES; i++) { 545 if (fetch_getln(conn) == -1) 546 return (hdr_syserror); 547 548 /* 549 * Note: we previously considered a pure whitespace line 550 * equivalent to an empty one. This was incorrect. 551 */ 552 if (conn->buf[0] != ' ' && conn->buf[0] != "\t"[0]) 553 break; 554 555 /* Got a continuation line. Concatenate to previous */ 556 len = hbuf->buflen + conn->buflen; 557 if (hbuf->bufsize < len + 1) { 558 len *= 2; 559 if ((hbuf->buf = realloc(hbuf->buf, len + 1)) == NULL) 560 return (hdr_syserror); 561 hbuf->bufsize = len + 1; 562 } 563 strcpy(hbuf->buf + hbuf->buflen, conn->buf); 564 hbuf->buflen += conn->buflen; 565 } 566 567 /* 568 * We could check for malformed headers but we don't really care. 569 * A valid header starts with a token immediately followed by a 570 * colon; a token is any sequence of non-control, non-whitespace 571 * characters except "()<>@,;:\\\"{}". 572 */ 573 for (i = 0; hdr_names[i].num != hdr_unknown; i++) 574 if ((*p = http_match(hdr_names[i].name, hbuf->buf)) != NULL) 575 return (hdr_names[i].num); 576 577 return (hdr_unknown); 578 } 579 580 /************************** 581 * [Proxy-]Authenticate header parsing 582 */ 583 584 /* 585 * Read doublequote-delimited string into output buffer obuf (allocated 586 * by caller, whose responsibility it is to ensure that it's big enough) 587 * cp points to the first char after the initial '"' 588 * Handles \ quoting 589 * Returns pointer to the first char after the terminating double quote, or 590 * NULL for error. 591 */ 592 static const char * 593 http_parse_headerstring(const char *cp, char *obuf) 594 { 595 for (;;) { 596 switch (*cp) { 597 case 0: /* Unterminated string */ 598 *obuf = 0; 599 return (NULL); 600 case '"': /* Ending quote */ 601 *obuf = 0; 602 return (++cp); 603 case '\\': 604 if (*++cp == 0) { 605 *obuf = 0; 606 return (NULL); 607 } 608 /* FALLTHROUGH */ 609 default: 610 *obuf++ = *cp++; 611 } 612 } 613 } 614 615 /* Http auth challenge schemes */ 616 typedef enum {HTTPAS_UNKNOWN, HTTPAS_BASIC,HTTPAS_DIGEST} http_auth_schemes_t; 617 618 /* Data holder for a Basic or Digest challenge. */ 619 typedef struct { 620 http_auth_schemes_t scheme; 621 char *realm; 622 char *qop; 623 char *nonce; 624 char *opaque; 625 char *algo; 626 int stale; 627 int nc; /* Nonce count */ 628 } http_auth_challenge_t; 629 630 static void 631 init_http_auth_challenge(http_auth_challenge_t *b) 632 { 633 b->scheme = HTTPAS_UNKNOWN; 634 b->realm = b->qop = b->nonce = b->opaque = b->algo = NULL; 635 b->stale = b->nc = 0; 636 } 637 638 static void 639 clean_http_auth_challenge(http_auth_challenge_t *b) 640 { 641 if (b->realm) 642 free(b->realm); 643 if (b->qop) 644 free(b->qop); 645 if (b->nonce) 646 free(b->nonce); 647 if (b->opaque) 648 free(b->opaque); 649 if (b->algo) 650 free(b->algo); 651 init_http_auth_challenge(b); 652 } 653 654 /* Data holder for an array of challenges offered in an http response. */ 655 #define MAX_CHALLENGES 10 656 typedef struct { 657 http_auth_challenge_t *challenges[MAX_CHALLENGES]; 658 int count; /* Number of parsed challenges in the array */ 659 int valid; /* We did parse an authenticate header */ 660 } http_auth_challenges_t; 661 662 static void 663 init_http_auth_challenges(http_auth_challenges_t *cs) 664 { 665 int i; 666 for (i = 0; i < MAX_CHALLENGES; i++) 667 cs->challenges[i] = NULL; 668 cs->count = cs->valid = 0; 669 } 670 671 static void 672 clean_http_auth_challenges(http_auth_challenges_t *cs) 673 { 674 int i; 675 /* We rely on non-zero pointers being allocated, not on the count */ 676 for (i = 0; i < MAX_CHALLENGES; i++) { 677 if (cs->challenges[i] != NULL) { 678 clean_http_auth_challenge(cs->challenges[i]); 679 free(cs->challenges[i]); 680 } 681 } 682 init_http_auth_challenges(cs); 683 } 684 685 /* 686 * Enumeration for lexical elements. Separators will be returned as their own 687 * ascii value 688 */ 689 typedef enum {HTTPHL_WORD=256, HTTPHL_STRING=257, HTTPHL_END=258, 690 HTTPHL_ERROR = 259} http_header_lex_t; 691 692 /* 693 * Determine what kind of token comes next and return possible value 694 * in buf, which is supposed to have been allocated big enough by 695 * caller. Advance input pointer and return element type. 696 */ 697 static int 698 http_header_lex(const char **cpp, char *buf) 699 { 700 size_t l; 701 /* Eat initial whitespace */ 702 *cpp += strspn(*cpp, " \t"); 703 if (**cpp == 0) 704 return (HTTPHL_END); 705 706 /* Separator ? */ 707 if (**cpp == ',' || **cpp == '=') 708 return (*((*cpp)++)); 709 710 /* String ? */ 711 if (**cpp == '"') { 712 *cpp = http_parse_headerstring(++*cpp, buf); 713 if (*cpp == NULL) 714 return (HTTPHL_ERROR); 715 return (HTTPHL_STRING); 716 } 717 718 /* Read other token, until separator or whitespace */ 719 l = strcspn(*cpp, " \t,="); 720 memcpy(buf, *cpp, l); 721 buf[l] = 0; 722 *cpp += l; 723 return (HTTPHL_WORD); 724 } 725 726 /* 727 * Read challenges from http xxx-authenticate header and accumulate them 728 * in the challenges list structure. 729 * 730 * Headers with multiple challenges are specified by rfc2617, but 731 * servers (ie: squid) often send them in separate headers instead, 732 * which in turn is forbidden by the http spec (multiple headers with 733 * the same name are only allowed for pure comma-separated lists, see 734 * rfc2616 sec 4.2). 735 * 736 * We support both approaches anyway 737 */ 738 static int 739 http_parse_authenticate(const char *cp, http_auth_challenges_t *cs) 740 { 741 int ret = -1; 742 http_header_lex_t lex; 743 char *key = malloc(strlen(cp) + 1); 744 char *value = malloc(strlen(cp) + 1); 745 char *buf = malloc(strlen(cp) + 1); 746 747 if (key == NULL || value == NULL || buf == NULL) { 748 fetch_syserr(); 749 goto out; 750 } 751 752 /* In any case we've seen the header and we set the valid bit */ 753 cs->valid = 1; 754 755 /* Need word first */ 756 lex = http_header_lex(&cp, key); 757 if (lex != HTTPHL_WORD) 758 goto out; 759 760 /* Loop on challenges */ 761 for (; cs->count < MAX_CHALLENGES; cs->count++) { 762 cs->challenges[cs->count] = 763 malloc(sizeof(http_auth_challenge_t)); 764 if (cs->challenges[cs->count] == NULL) { 765 fetch_syserr(); 766 goto out; 767 } 768 init_http_auth_challenge(cs->challenges[cs->count]); 769 if (strcasecmp(key, "basic") == 0) { 770 cs->challenges[cs->count]->scheme = HTTPAS_BASIC; 771 } else if (strcasecmp(key, "digest") == 0) { 772 cs->challenges[cs->count]->scheme = HTTPAS_DIGEST; 773 } else { 774 cs->challenges[cs->count]->scheme = HTTPAS_UNKNOWN; 775 /* 776 * Continue parsing as basic or digest may 777 * follow, and the syntax is the same for 778 * all. We'll just ignore this one when 779 * looking at the list 780 */ 781 } 782 783 /* Loop on attributes */ 784 for (;;) { 785 /* Key */ 786 lex = http_header_lex(&cp, key); 787 if (lex != HTTPHL_WORD) 788 goto out; 789 790 /* Equal sign */ 791 lex = http_header_lex(&cp, buf); 792 if (lex != '=') 793 goto out; 794 795 /* Value */ 796 lex = http_header_lex(&cp, value); 797 if (lex != HTTPHL_WORD && lex != HTTPHL_STRING) 798 goto out; 799 800 if (strcasecmp(key, "realm") == 0) { 801 cs->challenges[cs->count]->realm = 802 strdup(value); 803 } else if (strcasecmp(key, "qop") == 0) { 804 cs->challenges[cs->count]->qop = 805 strdup(value); 806 } else if (strcasecmp(key, "nonce") == 0) { 807 cs->challenges[cs->count]->nonce = 808 strdup(value); 809 } else if (strcasecmp(key, "opaque") == 0) { 810 cs->challenges[cs->count]->opaque = 811 strdup(value); 812 } else if (strcasecmp(key, "algorithm") == 0) { 813 cs->challenges[cs->count]->algo = 814 strdup(value); 815 } else if (strcasecmp(key, "stale") == 0) { 816 cs->challenges[cs->count]->stale = 817 strcasecmp(value, "no"); 818 } else { 819 /* ignore unknown attributes */ 820 } 821 822 /* Comma or Next challenge or End */ 823 lex = http_header_lex(&cp, key); 824 /* 825 * If we get a word here, this is the beginning of the 826 * next challenge. Break the attributes loop 827 */ 828 if (lex == HTTPHL_WORD) 829 break; 830 831 if (lex == HTTPHL_END) { 832 /* End while looking for ',' is normal exit */ 833 cs->count++; 834 ret = 0; 835 goto out; 836 } 837 /* Anything else is an error */ 838 if (lex != ',') 839 goto out; 840 841 } /* End attributes loop */ 842 } /* End challenge loop */ 843 844 /* 845 * Challenges max count exceeded. This really can't happen 846 * with normal data, something's fishy -> error 847 */ 848 849 out: 850 if (key) 851 free(key); 852 if (value) 853 free(value); 854 if (buf) 855 free(buf); 856 return (ret); 857 } 858 859 860 /* 861 * Parse a last-modified header 862 */ 863 static int 864 http_parse_mtime(const char *p, time_t *mtime) 865 { 866 char locale[64], *r; 867 struct tm tm; 868 869 strlcpy(locale, setlocale(LC_TIME, NULL), sizeof(locale)); 870 setlocale(LC_TIME, "C"); 871 r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm); 872 /* 873 * Some proxies use UTC in response, but it should still be 874 * parsed. RFC2616 states GMT and UTC are exactly equal for HTTP. 875 */ 876 if (r == NULL) 877 r = strptime(p, "%a, %d %b %Y %H:%M:%S UTC", &tm); 878 /* XXX should add support for date-2 and date-3 */ 879 setlocale(LC_TIME, locale); 880 if (r == NULL) 881 return (-1); 882 DEBUGF("last modified: [%04d-%02d-%02d %02d:%02d:%02d]\n", 883 tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday, 884 tm.tm_hour, tm.tm_min, tm.tm_sec); 885 *mtime = timegm(&tm); 886 return (0); 887 } 888 889 /* 890 * Parse a content-length header 891 */ 892 static int 893 http_parse_length(const char *p, off_t *length) 894 { 895 off_t len; 896 897 for (len = 0; *p && isdigit((unsigned char)*p); ++p) 898 len = len * 10 + (*p - '0'); 899 if (*p) 900 return (-1); 901 DEBUGF("content length: [%lld]\n", (long long)len); 902 *length = len; 903 return (0); 904 } 905 906 /* 907 * Parse a content-range header 908 */ 909 static int 910 http_parse_range(const char *p, off_t *offset, off_t *length, off_t *size) 911 { 912 off_t first, last, len; 913 914 if (strncasecmp(p, "bytes ", 6) != 0) 915 return (-1); 916 p += 6; 917 if (*p == '*') { 918 first = last = -1; 919 ++p; 920 } else { 921 for (first = 0; *p && isdigit((unsigned char)*p); ++p) 922 first = first * 10 + *p - '0'; 923 if (*p != '-') 924 return (-1); 925 for (last = 0, ++p; *p && isdigit((unsigned char)*p); ++p) 926 last = last * 10 + *p - '0'; 927 } 928 if (first > last || *p != '/') 929 return (-1); 930 for (len = 0, ++p; *p && isdigit((unsigned char)*p); ++p) 931 len = len * 10 + *p - '0'; 932 if (*p || len < last - first + 1) 933 return (-1); 934 if (first == -1) { 935 DEBUGF("content range: [*/%lld]\n", (long long)len); 936 *length = 0; 937 } else { 938 DEBUGF("content range: [%lld-%lld/%lld]\n", 939 (long long)first, (long long)last, (long long)len); 940 *length = last - first + 1; 941 } 942 *offset = first; 943 *size = len; 944 return (0); 945 } 946 947 948 /***************************************************************************** 949 * Helper functions for authorization 950 */ 951 952 /* 953 * Base64 encoding 954 */ 955 static char * 956 http_base64(const char *src) 957 { 958 static const char base64[] = 959 "ABCDEFGHIJKLMNOPQRSTUVWXYZ" 960 "abcdefghijklmnopqrstuvwxyz" 961 "0123456789+/"; 962 char *str, *dst; 963 size_t l; 964 int t; 965 966 l = strlen(src); 967 if ((str = malloc(((l + 2) / 3) * 4 + 1)) == NULL) 968 return (NULL); 969 dst = str; 970 971 while (l >= 3) { 972 t = (src[0] << 16) | (src[1] << 8) | src[2]; 973 dst[0] = base64[(t >> 18) & 0x3f]; 974 dst[1] = base64[(t >> 12) & 0x3f]; 975 dst[2] = base64[(t >> 6) & 0x3f]; 976 dst[3] = base64[(t >> 0) & 0x3f]; 977 src += 3; l -= 3; 978 dst += 4; 979 } 980 981 switch (l) { 982 case 2: 983 t = (src[0] << 16) | (src[1] << 8); 984 dst[0] = base64[(t >> 18) & 0x3f]; 985 dst[1] = base64[(t >> 12) & 0x3f]; 986 dst[2] = base64[(t >> 6) & 0x3f]; 987 dst[3] = '='; 988 dst += 4; 989 break; 990 case 1: 991 t = src[0] << 16; 992 dst[0] = base64[(t >> 18) & 0x3f]; 993 dst[1] = base64[(t >> 12) & 0x3f]; 994 dst[2] = dst[3] = '='; 995 dst += 4; 996 break; 997 case 0: 998 break; 999 } 1000 1001 *dst = 0; 1002 return (str); 1003 } 1004 1005 1006 /* 1007 * Extract authorization parameters from environment value. 1008 * The value is like scheme:realm:user:pass 1009 */ 1010 typedef struct { 1011 char *scheme; 1012 char *realm; 1013 char *user; 1014 char *password; 1015 } http_auth_params_t; 1016 1017 static void 1018 init_http_auth_params(http_auth_params_t *s) 1019 { 1020 s->scheme = s->realm = s->user = s->password = NULL; 1021 } 1022 1023 static void 1024 clean_http_auth_params(http_auth_params_t *s) 1025 { 1026 if (s->scheme) 1027 free(s->scheme); 1028 if (s->realm) 1029 free(s->realm); 1030 if (s->user) 1031 free(s->user); 1032 if (s->password) 1033 free(s->password); 1034 init_http_auth_params(s); 1035 } 1036 1037 static int 1038 http_authfromenv(const char *p, http_auth_params_t *parms) 1039 { 1040 int ret = -1; 1041 char *v, *ve; 1042 char *str = strdup(p); 1043 1044 if (str == NULL) { 1045 fetch_syserr(); 1046 return (-1); 1047 } 1048 v = str; 1049 1050 if ((ve = strchr(v, ':')) == NULL) 1051 goto out; 1052 1053 *ve = 0; 1054 if ((parms->scheme = strdup(v)) == NULL) { 1055 fetch_syserr(); 1056 goto out; 1057 } 1058 v = ve + 1; 1059 1060 if ((ve = strchr(v, ':')) == NULL) 1061 goto out; 1062 1063 *ve = 0; 1064 if ((parms->realm = strdup(v)) == NULL) { 1065 fetch_syserr(); 1066 goto out; 1067 } 1068 v = ve + 1; 1069 1070 if ((ve = strchr(v, ':')) == NULL) 1071 goto out; 1072 1073 *ve = 0; 1074 if ((parms->user = strdup(v)) == NULL) { 1075 fetch_syserr(); 1076 goto out; 1077 } 1078 v = ve + 1; 1079 1080 1081 if ((parms->password = strdup(v)) == NULL) { 1082 fetch_syserr(); 1083 goto out; 1084 } 1085 ret = 0; 1086 out: 1087 if (ret == -1) 1088 clean_http_auth_params(parms); 1089 if (str) 1090 free(str); 1091 return (ret); 1092 } 1093 1094 1095 /* 1096 * Digest response: the code to compute the digest is taken from the 1097 * sample implementation in RFC2616 1098 */ 1099 #define IN const 1100 #define OUT 1101 1102 #define HASHLEN 16 1103 typedef char HASH[HASHLEN]; 1104 #define HASHHEXLEN 32 1105 typedef char HASHHEX[HASHHEXLEN+1]; 1106 1107 static const char *hexchars = "0123456789abcdef"; 1108 static void 1109 CvtHex(IN HASH Bin, OUT HASHHEX Hex) 1110 { 1111 unsigned short i; 1112 unsigned char j; 1113 1114 for (i = 0; i < HASHLEN; i++) { 1115 j = (Bin[i] >> 4) & 0xf; 1116 Hex[i*2] = hexchars[j]; 1117 j = Bin[i] & 0xf; 1118 Hex[i*2+1] = hexchars[j]; 1119 } 1120 Hex[HASHHEXLEN] = '\0'; 1121 }; 1122 1123 /* calculate H(A1) as per spec */ 1124 static void 1125 DigestCalcHA1( 1126 IN char * pszAlg, 1127 IN char * pszUserName, 1128 IN char * pszRealm, 1129 IN char * pszPassword, 1130 IN char * pszNonce, 1131 IN char * pszCNonce, 1132 OUT HASHHEX SessionKey 1133 ) 1134 { 1135 MD5_CTX Md5Ctx; 1136 HASH HA1; 1137 1138 MD5Init(&Md5Ctx); 1139 MD5Update(&Md5Ctx, pszUserName, strlen(pszUserName)); 1140 MD5Update(&Md5Ctx, ":", 1); 1141 MD5Update(&Md5Ctx, pszRealm, strlen(pszRealm)); 1142 MD5Update(&Md5Ctx, ":", 1); 1143 MD5Update(&Md5Ctx, pszPassword, strlen(pszPassword)); 1144 MD5Final(HA1, &Md5Ctx); 1145 if (strcasecmp(pszAlg, "md5-sess") == 0) { 1146 1147 MD5Init(&Md5Ctx); 1148 MD5Update(&Md5Ctx, HA1, HASHLEN); 1149 MD5Update(&Md5Ctx, ":", 1); 1150 MD5Update(&Md5Ctx, pszNonce, strlen(pszNonce)); 1151 MD5Update(&Md5Ctx, ":", 1); 1152 MD5Update(&Md5Ctx, pszCNonce, strlen(pszCNonce)); 1153 MD5Final(HA1, &Md5Ctx); 1154 } 1155 CvtHex(HA1, SessionKey); 1156 } 1157 1158 /* calculate request-digest/response-digest as per HTTP Digest spec */ 1159 static void 1160 DigestCalcResponse( 1161 IN HASHHEX HA1, /* H(A1) */ 1162 IN char * pszNonce, /* nonce from server */ 1163 IN char * pszNonceCount, /* 8 hex digits */ 1164 IN char * pszCNonce, /* client nonce */ 1165 IN char * pszQop, /* qop-value: "", "auth", "auth-int" */ 1166 IN char * pszMethod, /* method from the request */ 1167 IN char * pszDigestUri, /* requested URL */ 1168 IN HASHHEX HEntity, /* H(entity body) if qop="auth-int" */ 1169 OUT HASHHEX Response /* request-digest or response-digest */ 1170 ) 1171 { 1172 #if 0 1173 DEBUGF("Calc: HA1[%s] Nonce[%s] qop[%s] method[%s] URI[%s]\n", 1174 HA1, pszNonce, pszQop, pszMethod, pszDigestUri); 1175 #endif 1176 MD5_CTX Md5Ctx; 1177 HASH HA2; 1178 HASH RespHash; 1179 HASHHEX HA2Hex; 1180 1181 // calculate H(A2) 1182 MD5Init(&Md5Ctx); 1183 MD5Update(&Md5Ctx, pszMethod, strlen(pszMethod)); 1184 MD5Update(&Md5Ctx, ":", 1); 1185 MD5Update(&Md5Ctx, pszDigestUri, strlen(pszDigestUri)); 1186 if (strcasecmp(pszQop, "auth-int") == 0) { 1187 MD5Update(&Md5Ctx, ":", 1); 1188 MD5Update(&Md5Ctx, HEntity, HASHHEXLEN); 1189 } 1190 MD5Final(HA2, &Md5Ctx); 1191 CvtHex(HA2, HA2Hex); 1192 1193 // calculate response 1194 MD5Init(&Md5Ctx); 1195 MD5Update(&Md5Ctx, HA1, HASHHEXLEN); 1196 MD5Update(&Md5Ctx, ":", 1); 1197 MD5Update(&Md5Ctx, pszNonce, strlen(pszNonce)); 1198 MD5Update(&Md5Ctx, ":", 1); 1199 if (*pszQop) { 1200 MD5Update(&Md5Ctx, pszNonceCount, strlen(pszNonceCount)); 1201 MD5Update(&Md5Ctx, ":", 1); 1202 MD5Update(&Md5Ctx, pszCNonce, strlen(pszCNonce)); 1203 MD5Update(&Md5Ctx, ":", 1); 1204 MD5Update(&Md5Ctx, pszQop, strlen(pszQop)); 1205 MD5Update(&Md5Ctx, ":", 1); 1206 } 1207 MD5Update(&Md5Ctx, HA2Hex, HASHHEXLEN); 1208 MD5Final(RespHash, &Md5Ctx); 1209 CvtHex(RespHash, Response); 1210 } 1211 1212 /* 1213 * Generate/Send a Digest authorization header 1214 * This looks like: [Proxy-]Authorization: credentials 1215 * 1216 * credentials = "Digest" digest-response 1217 * digest-response = 1#( username | realm | nonce | digest-uri 1218 * | response | [ algorithm ] | [cnonce] | 1219 * [opaque] | [message-qop] | 1220 * [nonce-count] | [auth-param] ) 1221 * username = "username" "=" username-value 1222 * username-value = quoted-string 1223 * digest-uri = "uri" "=" digest-uri-value 1224 * digest-uri-value = request-uri ; As specified by HTTP/1.1 1225 * message-qop = "qop" "=" qop-value 1226 * cnonce = "cnonce" "=" cnonce-value 1227 * cnonce-value = nonce-value 1228 * nonce-count = "nc" "=" nc-value 1229 * nc-value = 8LHEX 1230 * response = "response" "=" request-digest 1231 * request-digest = <"> 32LHEX <"> 1232 */ 1233 static int 1234 http_digest_auth(conn_t *conn, const char *hdr, http_auth_challenge_t *c, 1235 http_auth_params_t *parms, struct url *url) 1236 { 1237 int r; 1238 char noncecount[10]; 1239 char cnonce[40]; 1240 char *options = NULL; 1241 1242 if (!c->realm || !c->nonce) { 1243 DEBUGF("realm/nonce not set in challenge\n"); 1244 return(-1); 1245 } 1246 if (!c->algo) 1247 c->algo = strdup(""); 1248 1249 if (asprintf(&options, "%s%s%s%s", 1250 *c->algo? ",algorithm=" : "", c->algo, 1251 c->opaque? ",opaque=" : "", c->opaque?c->opaque:"") < 0) 1252 return (-1); 1253 1254 if (!c->qop) { 1255 c->qop = strdup(""); 1256 *noncecount = 0; 1257 *cnonce = 0; 1258 } else { 1259 c->nc++; 1260 sprintf(noncecount, "%08x", c->nc); 1261 /* We don't try very hard with the cnonce ... */ 1262 sprintf(cnonce, "%x%lx", getpid(), (unsigned long)time(0)); 1263 } 1264 1265 HASHHEX HA1; 1266 DigestCalcHA1(c->algo, parms->user, c->realm, 1267 parms->password, c->nonce, cnonce, HA1); 1268 DEBUGF("HA1: [%s]\n", HA1); 1269 HASHHEX digest, null; 1270 memset(null, 0, sizeof(null)); 1271 DigestCalcResponse(HA1, c->nonce, noncecount, cnonce, c->qop, 1272 "GET", url->doc, null, digest); 1273 1274 if (c->qop[0]) { 1275 r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\"," 1276 "nonce=\"%s\",uri=\"%s\",response=\"%s\"," 1277 "qop=\"auth\", cnonce=\"%s\", nc=%s%s", 1278 hdr, parms->user, c->realm, 1279 c->nonce, url->doc, digest, 1280 cnonce, noncecount, options); 1281 } else { 1282 r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\"," 1283 "nonce=\"%s\",uri=\"%s\",response=\"%s\"%s", 1284 hdr, parms->user, c->realm, 1285 c->nonce, url->doc, digest, options); 1286 } 1287 if (options) 1288 free(options); 1289 return (r); 1290 } 1291 1292 /* 1293 * Encode username and password 1294 */ 1295 static int 1296 http_basic_auth(conn_t *conn, const char *hdr, const char *usr, const char *pwd) 1297 { 1298 char *upw, *auth; 1299 int r; 1300 1301 DEBUGF("basic: usr: [%s]\n", usr); 1302 DEBUGF("basic: pwd: [%s]\n", pwd); 1303 if (asprintf(&upw, "%s:%s", usr, pwd) == -1) 1304 return (-1); 1305 auth = http_base64(upw); 1306 free(upw); 1307 if (auth == NULL) 1308 return (-1); 1309 r = http_cmd(conn, "%s: Basic %s", hdr, auth); 1310 free(auth); 1311 return (r); 1312 } 1313 1314 /* 1315 * Chose the challenge to answer and call the appropriate routine to 1316 * produce the header. 1317 */ 1318 static int 1319 http_authorize(conn_t *conn, const char *hdr, http_auth_challenges_t *cs, 1320 http_auth_params_t *parms, struct url *url) 1321 { 1322 http_auth_challenge_t *digest = NULL; 1323 int i; 1324 1325 /* If user or pass are null we're not happy */ 1326 if (!parms->user || !parms->password) { 1327 DEBUGF("NULL usr or pass\n"); 1328 return (-1); 1329 } 1330 1331 /* Look for a Digest */ 1332 for (i = 0; i < cs->count; i++) { 1333 if (cs->challenges[i]->scheme == HTTPAS_DIGEST) 1334 digest = cs->challenges[i]; 1335 } 1336 1337 /* Error if "Digest" was specified and there is no Digest challenge */ 1338 if (!digest && 1339 (parms->scheme && strcasecmp(parms->scheme, "digest") == 0)) { 1340 DEBUGF("Digest auth in env, not supported by peer\n"); 1341 return (-1); 1342 } 1343 /* 1344 * If "basic" was specified in the environment, or there is no Digest 1345 * challenge, do the basic thing. Don't need a challenge for this, 1346 * so no need to check basic!=NULL 1347 */ 1348 if (!digest || 1349 (parms->scheme && strcasecmp(parms->scheme, "basic") == 0)) 1350 return (http_basic_auth(conn,hdr,parms->user,parms->password)); 1351 1352 /* Else, prefer digest. We just checked that it's not NULL */ 1353 return (http_digest_auth(conn, hdr, digest, parms, url)); 1354 } 1355 1356 /***************************************************************************** 1357 * Helper functions for connecting to a server or proxy 1358 */ 1359 1360 /* 1361 * Connect to the correct HTTP server or proxy. 1362 */ 1363 static conn_t * 1364 http_connect(struct url *URL, struct url *purl, const char *flags) 1365 { 1366 struct url *curl; 1367 conn_t *conn; 1368 hdr_t h; 1369 http_headerbuf_t headerbuf; 1370 const char *p; 1371 int verbose; 1372 int af, val; 1373 int serrno; 1374 bool isproxyauth = false; 1375 http_auth_challenges_t proxy_challenges; 1376 1377 #ifdef INET6 1378 af = AF_UNSPEC; 1379 #else 1380 af = AF_INET; 1381 #endif 1382 1383 verbose = CHECK_FLAG('v'); 1384 if (CHECK_FLAG('4')) 1385 af = AF_INET; 1386 #ifdef INET6 1387 else if (CHECK_FLAG('6')) 1388 af = AF_INET6; 1389 #endif 1390 1391 curl = (purl != NULL) ? purl : URL; 1392 1393 retry: 1394 if ((conn = fetch_connect(curl->host, curl->port, af, verbose)) == NULL) 1395 /* fetch_connect() has already set an error code */ 1396 return (NULL); 1397 init_http_headerbuf(&headerbuf); 1398 if (strcmp(URL->scheme, SCHEME_HTTPS) == 0 && purl) { 1399 init_http_auth_challenges(&proxy_challenges); 1400 http_cmd(conn, "CONNECT %s:%d HTTP/1.1", URL->host, URL->port); 1401 http_cmd(conn, "Host: %s:%d", URL->host, URL->port); 1402 if (isproxyauth) { 1403 http_auth_params_t aparams; 1404 init_http_auth_params(&aparams); 1405 if (*purl->user || *purl->pwd) { 1406 aparams.user = strdup(purl->user); 1407 aparams.password = strdup(purl->pwd); 1408 } else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL && 1409 *p != '\0') { 1410 if (http_authfromenv(p, &aparams) < 0) { 1411 http_seterr(HTTP_NEED_PROXY_AUTH); 1412 fetch_syserr(); 1413 goto ouch; 1414 } 1415 } else if (fetch_netrc_auth(purl) == 0) { 1416 aparams.user = strdup(purl->user); 1417 aparams.password = strdup(purl->pwd); 1418 } else { 1419 /* 1420 * No auth information found in system - exiting 1421 * with warning. 1422 */ 1423 warnx("Missing username and/or password set"); 1424 fetch_syserr(); 1425 goto ouch; 1426 } 1427 http_authorize(conn, "Proxy-Authorization", 1428 &proxy_challenges, &aparams, purl); 1429 clean_http_auth_params(&aparams); 1430 } 1431 http_cmd(conn, ""); 1432 /* Get reply from CONNECT Tunnel attempt */ 1433 int httpreply = http_get_reply(conn); 1434 if (httpreply != HTTP_OK) { 1435 http_seterr(httpreply); 1436 /* If the error is a 407/HTTP_NEED_PROXY_AUTH */ 1437 if (httpreply == HTTP_NEED_PROXY_AUTH && 1438 ! isproxyauth) { 1439 /* Try again with authentication. */ 1440 clean_http_headerbuf(&headerbuf); 1441 fetch_close(conn); 1442 isproxyauth = true; 1443 goto retry; 1444 } 1445 goto ouch; 1446 } 1447 /* Read and discard the rest of the proxy response */ 1448 if (fetch_getln(conn) < 0) { 1449 fetch_syserr(); 1450 goto ouch; 1451 } 1452 do { 1453 switch ((h = http_next_header(conn, &headerbuf, &p))) { 1454 case hdr_syserror: 1455 fetch_syserr(); 1456 goto ouch; 1457 case hdr_error: 1458 http_seterr(HTTP_PROTOCOL_ERROR); 1459 goto ouch; 1460 default: 1461 /* ignore */ ; 1462 } 1463 } while (h > hdr_end); 1464 } 1465 if (strcmp(URL->scheme, SCHEME_HTTPS) == 0 && 1466 fetch_ssl(conn, URL, verbose) == -1) { 1467 /* grrr */ 1468 errno = EAUTH; 1469 fetch_syserr(); 1470 goto ouch; 1471 } 1472 1473 val = 1; 1474 setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val, sizeof(val)); 1475 1476 clean_http_headerbuf(&headerbuf); 1477 return (conn); 1478 ouch: 1479 serrno = errno; 1480 clean_http_headerbuf(&headerbuf); 1481 fetch_close(conn); 1482 errno = serrno; 1483 return (NULL); 1484 } 1485 1486 static struct url * 1487 http_get_proxy(struct url * url, const char *flags) 1488 { 1489 struct url *purl; 1490 char *p; 1491 1492 if (flags != NULL && strchr(flags, 'd') != NULL) 1493 return (NULL); 1494 if (fetch_no_proxy_match(url->host)) 1495 return (NULL); 1496 if (((p = getenv("HTTP_PROXY")) || (p = getenv("http_proxy"))) && 1497 *p && (purl = fetchParseURL(p))) { 1498 if (!*purl->scheme) 1499 strcpy(purl->scheme, SCHEME_HTTP); 1500 if (!purl->port) 1501 purl->port = fetch_default_proxy_port(purl->scheme); 1502 if (strcmp(purl->scheme, SCHEME_HTTP) == 0) 1503 return (purl); 1504 fetchFreeURL(purl); 1505 } 1506 return (NULL); 1507 } 1508 1509 static void 1510 http_print_html(FILE *out, FILE *in) 1511 { 1512 ssize_t len = 0; 1513 size_t cap; 1514 char *line = NULL, *p, *q; 1515 int comment, tag; 1516 1517 comment = tag = 0; 1518 while ((len = getline(&line, &cap, in)) >= 0) { 1519 while (len && isspace((unsigned char)line[len - 1])) 1520 --len; 1521 for (p = q = line; q < line + len; ++q) { 1522 if (comment && *q == '-') { 1523 if (q + 2 < line + len && 1524 strcmp(q, "-->") == 0) { 1525 tag = comment = 0; 1526 q += 2; 1527 } 1528 } else if (tag && !comment && *q == '>') { 1529 p = q + 1; 1530 tag = 0; 1531 } else if (!tag && *q == '<') { 1532 if (q > p) 1533 fwrite(p, q - p, 1, out); 1534 tag = 1; 1535 if (q + 3 < line + len && 1536 strcmp(q, "<!--") == 0) { 1537 comment = 1; 1538 q += 3; 1539 } 1540 } 1541 } 1542 if (!tag && q > p) 1543 fwrite(p, q - p, 1, out); 1544 fputc('\n', out); 1545 } 1546 1547 free(line); 1548 } 1549 1550 1551 /***************************************************************************** 1552 * Core 1553 */ 1554 1555 FILE * 1556 http_request(struct url *URL, const char *op, struct url_stat *us, 1557 struct url *purl, const char *flags) 1558 { 1559 1560 return (http_request_body(URL, op, us, purl, flags, NULL, NULL)); 1561 } 1562 1563 /* 1564 * Send a request and process the reply 1565 * 1566 * XXX This function is way too long, the do..while loop should be split 1567 * XXX off into a separate function. 1568 */ 1569 FILE * 1570 http_request_body(struct url *URL, const char *op, struct url_stat *us, 1571 struct url *purl, const char *flags, const char *content_type, 1572 const char *body) 1573 { 1574 char timebuf[80]; 1575 char hbuf[MAXHOSTNAMELEN + 7], *host; 1576 conn_t *conn; 1577 struct url *url, *new; 1578 int chunked, direct, ims, noredirect, verbose; 1579 int e, i, n, val; 1580 off_t offset, clength, length, size; 1581 time_t mtime; 1582 const char *p; 1583 FILE *f; 1584 hdr_t h; 1585 struct tm *timestruct; 1586 http_headerbuf_t headerbuf; 1587 http_auth_challenges_t server_challenges; 1588 http_auth_challenges_t proxy_challenges; 1589 size_t body_len; 1590 1591 /* The following calls don't allocate anything */ 1592 init_http_headerbuf(&headerbuf); 1593 init_http_auth_challenges(&server_challenges); 1594 init_http_auth_challenges(&proxy_challenges); 1595 1596 direct = CHECK_FLAG('d'); 1597 noredirect = CHECK_FLAG('A'); 1598 verbose = CHECK_FLAG('v'); 1599 ims = CHECK_FLAG('i'); 1600 1601 if (direct && purl) { 1602 fetchFreeURL(purl); 1603 purl = NULL; 1604 } 1605 1606 /* try the provided URL first */ 1607 url = URL; 1608 1609 n = MAX_REDIRECT; 1610 i = 0; 1611 1612 e = HTTP_PROTOCOL_ERROR; 1613 do { 1614 new = NULL; 1615 chunked = 0; 1616 offset = 0; 1617 clength = -1; 1618 length = -1; 1619 size = -1; 1620 mtime = 0; 1621 1622 /* check port */ 1623 if (!url->port) 1624 url->port = fetch_default_port(url->scheme); 1625 1626 /* were we redirected to an FTP URL? */ 1627 if (purl == NULL && strcmp(url->scheme, SCHEME_FTP) == 0) { 1628 if (strcmp(op, "GET") == 0) 1629 return (ftp_request(url, "RETR", us, purl, flags)); 1630 else if (strcmp(op, "HEAD") == 0) 1631 return (ftp_request(url, "STAT", us, purl, flags)); 1632 } 1633 1634 /* connect to server or proxy */ 1635 if ((conn = http_connect(url, purl, flags)) == NULL) 1636 goto ouch; 1637 1638 /* append port number only if necessary */ 1639 host = url->host; 1640 if (url->port != fetch_default_port(url->scheme)) { 1641 snprintf(hbuf, sizeof(hbuf), "%s:%d", host, url->port); 1642 host = hbuf; 1643 } 1644 1645 /* send request */ 1646 if (verbose) 1647 fetch_info("requesting %s://%s%s", 1648 url->scheme, host, url->doc); 1649 if (purl && strcmp(url->scheme, SCHEME_HTTPS) != 0) { 1650 http_cmd(conn, "%s %s://%s%s HTTP/1.1", 1651 op, url->scheme, host, url->doc); 1652 } else { 1653 http_cmd(conn, "%s %s HTTP/1.1", 1654 op, url->doc); 1655 } 1656 1657 if (ims && url->ims_time) { 1658 timestruct = gmtime((time_t *)&url->ims_time); 1659 (void)strftime(timebuf, 80, "%a, %d %b %Y %T GMT", 1660 timestruct); 1661 if (verbose) 1662 fetch_info("If-Modified-Since: %s", timebuf); 1663 http_cmd(conn, "If-Modified-Since: %s", timebuf); 1664 } 1665 /* virtual host */ 1666 http_cmd(conn, "Host: %s", host); 1667 1668 /* 1669 * Proxy authorization: we only send auth after we received 1670 * a 407 error. We do not first try basic anyway (changed 1671 * when support was added for digest-auth) 1672 */ 1673 if (purl && proxy_challenges.valid) { 1674 http_auth_params_t aparams; 1675 init_http_auth_params(&aparams); 1676 if (*purl->user || *purl->pwd) { 1677 aparams.user = strdup(purl->user); 1678 aparams.password = strdup(purl->pwd); 1679 } else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL && 1680 *p != '\0') { 1681 if (http_authfromenv(p, &aparams) < 0) { 1682 http_seterr(HTTP_NEED_PROXY_AUTH); 1683 goto ouch; 1684 } 1685 } else if (fetch_netrc_auth(purl) == 0) { 1686 aparams.user = strdup(purl->user); 1687 aparams.password = strdup(purl->pwd); 1688 } 1689 http_authorize(conn, "Proxy-Authorization", 1690 &proxy_challenges, &aparams, url); 1691 clean_http_auth_params(&aparams); 1692 } 1693 1694 /* 1695 * Server authorization: we never send "a priori" 1696 * Basic auth, which used to be done if user/pass were 1697 * set in the url. This would be weird because we'd send the 1698 * password in the clear even if Digest is finally to be 1699 * used (it would have made more sense for the 1700 * pre-digest version to do this when Basic was specified 1701 * in the environment) 1702 */ 1703 if (server_challenges.valid) { 1704 http_auth_params_t aparams; 1705 init_http_auth_params(&aparams); 1706 if (*url->user || *url->pwd) { 1707 aparams.user = strdup(url->user); 1708 aparams.password = strdup(url->pwd); 1709 } else if ((p = getenv("HTTP_AUTH")) != NULL && 1710 *p != '\0') { 1711 if (http_authfromenv(p, &aparams) < 0) { 1712 http_seterr(HTTP_NEED_AUTH); 1713 goto ouch; 1714 } 1715 } else if (fetch_netrc_auth(url) == 0) { 1716 aparams.user = strdup(url->user); 1717 aparams.password = strdup(url->pwd); 1718 } else if (fetchAuthMethod && 1719 fetchAuthMethod(url) == 0) { 1720 aparams.user = strdup(url->user); 1721 aparams.password = strdup(url->pwd); 1722 } else { 1723 http_seterr(HTTP_NEED_AUTH); 1724 goto ouch; 1725 } 1726 http_authorize(conn, "Authorization", 1727 &server_challenges, &aparams, url); 1728 clean_http_auth_params(&aparams); 1729 } 1730 1731 /* other headers */ 1732 if ((p = getenv("HTTP_ACCEPT")) != NULL) { 1733 if (*p != '\0') 1734 http_cmd(conn, "Accept: %s", p); 1735 } else { 1736 http_cmd(conn, "Accept: */*"); 1737 } 1738 if ((p = getenv("HTTP_REFERER")) != NULL && *p != '\0') { 1739 if (strcasecmp(p, "auto") == 0) 1740 http_cmd(conn, "Referer: %s://%s%s", 1741 url->scheme, host, url->doc); 1742 else 1743 http_cmd(conn, "Referer: %s", p); 1744 } 1745 if ((p = getenv("HTTP_USER_AGENT")) != NULL) { 1746 /* no User-Agent if defined but empty */ 1747 if (*p != '\0') 1748 http_cmd(conn, "User-Agent: %s", p); 1749 } else { 1750 /* default User-Agent */ 1751 http_cmd(conn, "User-Agent: %s " _LIBFETCH_VER, 1752 getprogname()); 1753 } 1754 if (url->offset > 0) 1755 http_cmd(conn, "Range: bytes=%lld-", (long long)url->offset); 1756 http_cmd(conn, "Connection: close"); 1757 1758 if (body) { 1759 body_len = strlen(body); 1760 http_cmd(conn, "Content-Length: %zu", body_len); 1761 if (content_type != NULL) 1762 http_cmd(conn, "Content-Type: %s", content_type); 1763 } 1764 1765 http_cmd(conn, ""); 1766 1767 if (body) 1768 fetch_write(conn, body, body_len); 1769 1770 /* 1771 * Force the queued request to be dispatched. Normally, one 1772 * would do this with shutdown(2) but squid proxies can be 1773 * configured to disallow such half-closed connections. To 1774 * be compatible with such configurations, fiddle with socket 1775 * options to force the pending data to be written. 1776 */ 1777 val = 0; 1778 setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val, 1779 sizeof(val)); 1780 val = 1; 1781 setsockopt(conn->sd, IPPROTO_TCP, TCP_NODELAY, &val, 1782 sizeof(val)); 1783 1784 /* get reply */ 1785 switch (http_get_reply(conn)) { 1786 case HTTP_OK: 1787 case HTTP_PARTIAL: 1788 case HTTP_NOT_MODIFIED: 1789 /* fine */ 1790 break; 1791 case HTTP_MOVED_PERM: 1792 case HTTP_MOVED_TEMP: 1793 case HTTP_TEMP_REDIRECT: 1794 case HTTP_PERM_REDIRECT: 1795 case HTTP_SEE_OTHER: 1796 case HTTP_USE_PROXY: 1797 /* 1798 * Not so fine, but we still have to read the 1799 * headers to get the new location. 1800 */ 1801 break; 1802 case HTTP_NEED_AUTH: 1803 if (server_challenges.valid) { 1804 /* 1805 * We already sent out authorization code, 1806 * so there's nothing more we can do. 1807 */ 1808 http_seterr(conn->err); 1809 goto ouch; 1810 } 1811 /* try again, but send the password this time */ 1812 if (verbose) 1813 fetch_info("server requires authorization"); 1814 break; 1815 case HTTP_NEED_PROXY_AUTH: 1816 if (proxy_challenges.valid) { 1817 /* 1818 * We already sent our proxy 1819 * authorization code, so there's 1820 * nothing more we can do. */ 1821 http_seterr(conn->err); 1822 goto ouch; 1823 } 1824 /* try again, but send the password this time */ 1825 if (verbose) 1826 fetch_info("proxy requires authorization"); 1827 break; 1828 case HTTP_BAD_RANGE: 1829 /* 1830 * This can happen if we ask for 0 bytes because 1831 * we already have the whole file. Consider this 1832 * a success for now, and check sizes later. 1833 */ 1834 break; 1835 case HTTP_PROTOCOL_ERROR: 1836 /* fall through */ 1837 case -1: 1838 fetch_syserr(); 1839 goto ouch; 1840 default: 1841 http_seterr(conn->err); 1842 if (!verbose) 1843 goto ouch; 1844 /* fall through so we can get the full error message */ 1845 } 1846 1847 /* get headers. http_next_header expects one line readahead */ 1848 if (fetch_getln(conn) == -1) { 1849 fetch_syserr(); 1850 goto ouch; 1851 } 1852 do { 1853 switch ((h = http_next_header(conn, &headerbuf, &p))) { 1854 case hdr_syserror: 1855 fetch_syserr(); 1856 goto ouch; 1857 case hdr_error: 1858 http_seterr(HTTP_PROTOCOL_ERROR); 1859 goto ouch; 1860 case hdr_content_length: 1861 http_parse_length(p, &clength); 1862 break; 1863 case hdr_content_range: 1864 http_parse_range(p, &offset, &length, &size); 1865 break; 1866 case hdr_last_modified: 1867 http_parse_mtime(p, &mtime); 1868 break; 1869 case hdr_location: 1870 if (!HTTP_REDIRECT(conn->err)) 1871 break; 1872 /* 1873 * if the A flag is set, we don't follow 1874 * temporary redirects. 1875 */ 1876 if (noredirect && 1877 conn->err != HTTP_MOVED_PERM && 1878 conn->err != HTTP_PERM_REDIRECT && 1879 conn->err != HTTP_USE_PROXY) { 1880 n = 1; 1881 break; 1882 } 1883 if (new) 1884 free(new); 1885 if (verbose) 1886 fetch_info("%d redirect to %s", 1887 conn->err, p); 1888 if (*p == '/') 1889 /* absolute path */ 1890 new = fetchMakeURL(url->scheme, url->host, 1891 url->port, p, url->user, url->pwd); 1892 else 1893 new = fetchParseURL(p); 1894 if (new == NULL) { 1895 /* XXX should set an error code */ 1896 DEBUGF("failed to parse new URL\n"); 1897 goto ouch; 1898 } 1899 1900 /* Only copy credentials if the host matches */ 1901 if (strcmp(new->host, url->host) == 0 && 1902 !*new->user && !*new->pwd) { 1903 strcpy(new->user, url->user); 1904 strcpy(new->pwd, url->pwd); 1905 } 1906 new->offset = url->offset; 1907 new->length = url->length; 1908 new->ims_time = url->ims_time; 1909 break; 1910 case hdr_transfer_encoding: 1911 /* XXX weak test*/ 1912 chunked = (strcasecmp(p, "chunked") == 0); 1913 break; 1914 case hdr_www_authenticate: 1915 if (conn->err != HTTP_NEED_AUTH) 1916 break; 1917 if (http_parse_authenticate(p, &server_challenges) == 0) 1918 ++n; 1919 break; 1920 case hdr_proxy_authenticate: 1921 if (conn->err != HTTP_NEED_PROXY_AUTH) 1922 break; 1923 if (http_parse_authenticate(p, &proxy_challenges) == 0) 1924 ++n; 1925 break; 1926 case hdr_end: 1927 /* fall through */ 1928 case hdr_unknown: 1929 /* ignore */ 1930 break; 1931 } 1932 } while (h > hdr_end); 1933 1934 /* we need to provide authentication */ 1935 if (conn->err == HTTP_NEED_AUTH || 1936 conn->err == HTTP_NEED_PROXY_AUTH) { 1937 e = conn->err; 1938 if ((conn->err == HTTP_NEED_AUTH && 1939 !server_challenges.valid) || 1940 (conn->err == HTTP_NEED_PROXY_AUTH && 1941 !proxy_challenges.valid)) { 1942 /* 401/7 but no www/proxy-authenticate ?? */ 1943 DEBUGF("%03d without auth header\n", conn->err); 1944 goto ouch; 1945 } 1946 fetch_close(conn); 1947 conn = NULL; 1948 continue; 1949 } 1950 1951 /* requested range not satisfiable */ 1952 if (conn->err == HTTP_BAD_RANGE) { 1953 if (url->offset > 0 && url->length == 0) { 1954 /* asked for 0 bytes; fake it */ 1955 offset = url->offset; 1956 clength = -1; 1957 conn->err = HTTP_OK; 1958 break; 1959 } else { 1960 http_seterr(conn->err); 1961 goto ouch; 1962 } 1963 } 1964 1965 /* we have a hit or an error */ 1966 if (conn->err == HTTP_OK 1967 || conn->err == HTTP_NOT_MODIFIED 1968 || conn->err == HTTP_PARTIAL 1969 || HTTP_ERROR(conn->err)) 1970 break; 1971 1972 /* all other cases: we got a redirect */ 1973 e = conn->err; 1974 clean_http_auth_challenges(&server_challenges); 1975 fetch_close(conn); 1976 conn = NULL; 1977 if (!new) { 1978 DEBUGF("redirect with no new location\n"); 1979 break; 1980 } 1981 if (url != URL) 1982 fetchFreeURL(url); 1983 url = new; 1984 } while (++i < n); 1985 1986 /* we failed, or ran out of retries */ 1987 if (conn == NULL) { 1988 http_seterr(e); 1989 goto ouch; 1990 } 1991 1992 DEBUGF("offset %lld, length %lld, size %lld, clength %lld\n", 1993 (long long)offset, (long long)length, 1994 (long long)size, (long long)clength); 1995 1996 if (conn->err == HTTP_NOT_MODIFIED) { 1997 http_seterr(HTTP_NOT_MODIFIED); 1998 return (NULL); 1999 } 2000 2001 /* check for inconsistencies */ 2002 if (clength != -1 && length != -1 && clength != length) { 2003 http_seterr(HTTP_PROTOCOL_ERROR); 2004 goto ouch; 2005 } 2006 if (clength == -1) 2007 clength = length; 2008 if (clength != -1) 2009 length = offset + clength; 2010 if (length != -1 && size != -1 && length != size) { 2011 http_seterr(HTTP_PROTOCOL_ERROR); 2012 goto ouch; 2013 } 2014 if (size == -1) 2015 size = length; 2016 2017 /* fill in stats */ 2018 if (us) { 2019 us->size = size; 2020 us->atime = us->mtime = mtime; 2021 } 2022 2023 /* too far? */ 2024 if (URL->offset > 0 && offset > URL->offset) { 2025 http_seterr(HTTP_PROTOCOL_ERROR); 2026 goto ouch; 2027 } 2028 2029 /* report back real offset and size */ 2030 URL->offset = offset; 2031 URL->length = clength; 2032 2033 /* wrap it up in a FILE */ 2034 if ((f = http_funopen(conn, chunked)) == NULL) { 2035 fetch_syserr(); 2036 goto ouch; 2037 } 2038 2039 if (url != URL) 2040 fetchFreeURL(url); 2041 if (purl) 2042 fetchFreeURL(purl); 2043 2044 if (HTTP_ERROR(conn->err)) { 2045 http_print_html(stderr, f); 2046 fclose(f); 2047 f = NULL; 2048 } 2049 clean_http_headerbuf(&headerbuf); 2050 clean_http_auth_challenges(&server_challenges); 2051 clean_http_auth_challenges(&proxy_challenges); 2052 return (f); 2053 2054 ouch: 2055 if (url != URL) 2056 fetchFreeURL(url); 2057 if (purl) 2058 fetchFreeURL(purl); 2059 if (conn != NULL) 2060 fetch_close(conn); 2061 clean_http_headerbuf(&headerbuf); 2062 clean_http_auth_challenges(&server_challenges); 2063 clean_http_auth_challenges(&proxy_challenges); 2064 return (NULL); 2065 } 2066 2067 2068 /***************************************************************************** 2069 * Entry points 2070 */ 2071 2072 /* 2073 * Retrieve and stat a file by HTTP 2074 */ 2075 FILE * 2076 fetchXGetHTTP(struct url *URL, struct url_stat *us, const char *flags) 2077 { 2078 return (http_request(URL, "GET", us, http_get_proxy(URL, flags), flags)); 2079 } 2080 2081 /* 2082 * Retrieve a file by HTTP 2083 */ 2084 FILE * 2085 fetchGetHTTP(struct url *URL, const char *flags) 2086 { 2087 return (fetchXGetHTTP(URL, NULL, flags)); 2088 } 2089 2090 /* 2091 * Store a file by HTTP 2092 */ 2093 FILE * 2094 fetchPutHTTP(struct url *URL __unused, const char *flags __unused) 2095 { 2096 warnx("fetchPutHTTP(): not implemented"); 2097 return (NULL); 2098 } 2099 2100 /* 2101 * Get an HTTP document's metadata 2102 */ 2103 int 2104 fetchStatHTTP(struct url *URL, struct url_stat *us, const char *flags) 2105 { 2106 FILE *f; 2107 2108 f = http_request(URL, "HEAD", us, http_get_proxy(URL, flags), flags); 2109 if (f == NULL) 2110 return (-1); 2111 fclose(f); 2112 return (0); 2113 } 2114 2115 /* 2116 * List a directory 2117 */ 2118 struct url_ent * 2119 fetchListHTTP(struct url *url __unused, const char *flags __unused) 2120 { 2121 warnx("fetchListHTTP(): not implemented"); 2122 return (NULL); 2123 } 2124 2125 /* 2126 * Arbitrary HTTP verb and content requests 2127 */ 2128 FILE * 2129 fetchReqHTTP(struct url *URL, const char *method, const char *flags, 2130 const char *content_type, const char *body) 2131 { 2132 2133 return (http_request_body(URL, method, NULL, http_get_proxy(URL, flags), 2134 flags, content_type, body)); 2135 } 2136