lib/libfetch/http.c

   1 /*-
   2  * Copyright (c) 2000-2011 Dag-Erling Smørgrav
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer
  10  *    in this position and unchanged.
  11  * 2. Redistributions in binary form must reproduce the above copyright
  12  *    notice, this list of conditions and the following disclaimer in the
  13  *    documentation and/or other materials provided with the distribution.
  14  * 3. The name of the author may not be used to endorse or promote products
  15  *    derived from this software without specific prior written permission.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
  18  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
  19  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
  20  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
  21  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
  22  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  23  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  24  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  25  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
  26  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  27  */
  28
  29 #include <sys/cdefs.h>
  30 __FBSDID("$FreeBSD$");
  31
  32 /*
  33  * The following copyright applies to the base64 code:
  34  *
  35  *-
  36  * Copyright 1997 Massachusetts Institute of Technology
  37  *
  38  * Permission to use, copy, modify, and distribute this software and
  39  * its documentation for any purpose and without fee is hereby
  40  * granted, provided that both the above copyright notice and this
  41  * permission notice appear in all copies, that both the above
  42  * copyright notice and this permission notice appear in all
  43  * supporting documentation, and that the name of M.I.T. not be used
  44  * in advertising or publicity pertaining to distribution of the
  45  * software without specific, written prior permission.  M.I.T. makes
  46  * no representations about the suitability of this software for any
  47  * purpose.  It is provided "as is" without express or implied
  48  * warranty.
  49  *
  50  * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''.  M.I.T. DISCLAIMS
  51  * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE,
  52  * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
  53  * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT
  54  * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  55  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
  56  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
  57  * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
  58  * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
  59  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
  60  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  61  * SUCH DAMAGE.
  62  */
  63
  64 #include <sys/param.h>
  65 #include <sys/socket.h>
  66 #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 <stdio.h>
  75 #include <stdlib.h>
  76 #include <string.h>
  77 #include <time.h>
  78 #include <unistd.h>
  79 #include <md5.h>
  80
  81 #include <netinet/in.h>
  82 #include <netinet/tcp.h>
  83
  84 #include "fetch.h"
  85 #include "common.h"
  86 #include "httperr.h"
  87
  88 /* Maximum number of redirects to follow */
  89 #define MAX_REDIRECT 20
  90
  91 /* Symbolic names for reply codes we care about */
  92 #define HTTP_OK                 200
  93 #define HTTP_PARTIAL            206
  94 #define HTTP_MOVED_PERM         301
  95 #define HTTP_MOVED_TEMP         302
  96 #define HTTP_SEE_OTHER          303
  97 #define HTTP_NOT_MODIFIED       304
  98 #define HTTP_USE_PROXY          305
  99 #define HTTP_TEMP_REDIRECT      307
 100 #define HTTP_PERM_REDIRECT      308
 101 #define HTTP_NEED_AUTH          401
 102 #define HTTP_NEED_PROXY_AUTH    407
 103 #define HTTP_BAD_RANGE          416
 104 #define HTTP_PROTOCOL_ERROR     999
 105
 106 #define HTTP_REDIRECT(xyz) ((xyz) == HTTP_MOVED_PERM \
 107                             || (xyz) == HTTP_MOVED_TEMP \
 108                             || (xyz) == HTTP_TEMP_REDIRECT \
 109                             || (xyz) == HTTP_USE_PROXY \
 110                             || (xyz) == HTTP_SEE_OTHER)
 111
 112 #define HTTP_ERROR(xyz) ((xyz) > 400 && (xyz) < 599)
 113
 114
 115 /*****************************************************************************
 116  * I/O functions for decoding chunked streams
 117  */
 118
 119 struct httpio
 120 {
 121         conn_t          *conn;          /* connection */
 122         int              chunked;       /* chunked mode */
 123         char            *buf;           /* chunk buffer */
 124         size_t           bufsize;       /* size of chunk buffer */
 125         ssize_t          buflen;        /* amount of data currently in buffer */
 126         int              bufpos;        /* current read offset in buffer */
 127         int              eof;           /* end-of-file flag */
 128         int              error;         /* error flag */
 129         size_t           chunksize;     /* remaining size of current chunk */
 130 #ifndef NDEBUG
 131         size_t           total;
 132 #endif
 133 };
 134
 135 /*
 136  * Get next chunk header
 137  */
 138 static int
 139 http_new_chunk(struct httpio *io)
 140 {
 141         char *p;
 142
 143         if (fetch_getln(io->conn) == -1)
 144                 return (-1);
 145
 146         if (io->conn->buflen < 2 || !isxdigit((unsigned char)*io->conn->buf))
 147                 return (-1);
 148
 149         for (p = io->conn->buf; *p && !isspace((unsigned char)*p); ++p) {
 150                 if (*p == ';')
 151                         break;
 152                 if (!isxdigit((unsigned char)*p))
 153                         return (-1);
 154                 if (isdigit((unsigned char)*p)) {
 155                         io->chunksize = io->chunksize * 16 +
 156                             *p - '0';
 157                 } else {
 158                         io->chunksize = io->chunksize * 16 +
 159                             10 + tolower((unsigned char)*p) - 'a';
 160                 }
 161         }
 162
 163 #ifndef NDEBUG
 164         if (fetchDebug) {
 165                 io->total += io->chunksize;
 166                 if (io->chunksize == 0)
 167                         fprintf(stderr, "%s(): end of last chunk\n", __func__);
 168                 else
 169                         fprintf(stderr, "%s(): new chunk: %lu (%lu)\n",
 170                             __func__, (unsigned long)io->chunksize,
 171                             (unsigned long)io->total);
 172         }
 173 #endif
 174
 175         return (io->chunksize);
 176 }
 177
 178 /*
 179  * Grow the input buffer to at least len bytes
 180  */
 181 static inline int
 182 http_growbuf(struct httpio *io, size_t len)
 183 {
 184         char *tmp;
 185
 186         if (io->bufsize >= len)
 187                 return (0);
 188
 189         if ((tmp = realloc(io->buf, len)) == NULL)
 190                 return (-1);
 191         io->buf = tmp;
 192         io->bufsize = len;
 193         return (0);
 194 }
 195
 196 /*
 197  * Fill the input buffer, do chunk decoding on the fly
 198  */
 199 static int
 200 http_fillbuf(struct httpio *io, size_t len)
 201 {
 202         ssize_t nbytes;
 203
 204         if (io->error)
 205                 return (-1);
 206         if (io->eof)
 207                 return (0);
 208
 209         if (io->chunked == 0) {
 210                 if (http_growbuf(io, len) == -1)
 211                         return (-1);
 212                 if ((nbytes = fetch_read(io->conn, io->buf, len)) == -1) {
 213                         io->error = errno;
 214                         return (-1);
 215                 }
 216                 io->buflen = nbytes;
 217                 io->bufpos = 0;
 218                 return (io->buflen);
 219         }
 220
 221         if (io->chunksize == 0) {
 222                 switch (http_new_chunk(io)) {
 223                 case -1:
 224                         io->error = 1;
 225                         return (-1);
 226                 case 0:
 227                         io->eof = 1;
 228                         return (0);
 229                 }
 230         }
 231
 232         if (len > io->chunksize)
 233                 len = io->chunksize;
 234         if (http_growbuf(io, len) == -1)
 235                 return (-1);
 236         if ((nbytes = fetch_read(io->conn, io->buf, len)) == -1) {
 237                 io->error = errno;
 238                 return (-1);
 239         }
 240         io->buflen = nbytes;
 241         io->chunksize -= io->buflen;
 242
 243         if (io->chunksize == 0) {
 244                 char endl[2];
 245
 246                 if (fetch_read(io->conn, endl, 2) != 2 ||
 247                     endl[0] != '\r' || endl[1] != '\n')
 248                         return (-1);
 249         }
 250
 251         io->bufpos = 0;
 252
 253         return (io->buflen);
 254 }
 255
 256 /*
 257  * Read function
 258  */
 259 static int
 260 http_readfn(void *v, char *buf, int len)
 261 {
 262         struct httpio *io = (struct httpio *)v;
 263         int l, pos;
 264
 265         if (io->error)
 266                 return (-1);
 267         if (io->eof)
 268                 return (0);
 269
 270         for (pos = 0; len > 0; pos += l, len -= l) {
 271                 /* empty buffer */
 272                 if (!io->buf || io->bufpos == io->buflen)
 273                         if (http_fillbuf(io, len) < 1)
 274                                 break;
 275                 l = io->buflen - io->bufpos;
 276                 if (len < l)
 277                         l = len;
 278                 memcpy(buf + pos, io->buf + io->bufpos, l);
 279                 io->bufpos += l;
 280         }
 281
 282         if (!pos && io->error) {
 283                 if (io->error == EINTR)
 284                         io->error = 0;
 285                 return (-1);
 286         }
 287         return (pos);
 288 }
 289
 290 /*
 291  * Write function
 292  */
 293 static int
 294 http_writefn(void *v, const char *buf, int len)
 295 {
 296         struct httpio *io = (struct httpio *)v;
 297
 298         return (fetch_write(io->conn, buf, len));
 299 }
 300
 301 /*
 302  * Close function
 303  */
 304 static int
 305 http_closefn(void *v)
 306 {
 307         struct httpio *io = (struct httpio *)v;
 308         int r;
 309
 310         r = fetch_close(io->conn);
 311         if (io->buf)
 312                 free(io->buf);
 313         free(io);
 314         return (r);
 315 }
 316
 317 /*
 318  * Wrap a file descriptor up
 319  */
 320 static FILE *
 321 http_funopen(conn_t *conn, int chunked)
 322 {
 323         struct httpio *io;
 324         FILE *f;
 325
 326         if ((io = calloc(1, sizeof(*io))) == NULL) {
 327                 fetch_syserr();
 328                 return (NULL);
 329         }
 330         io->conn = conn;
 331         io->chunked = chunked;
 332         f = funopen(io, http_readfn, http_writefn, NULL, http_closefn);
 333         if (f == NULL) {
 334                 fetch_syserr();
 335                 free(io);
 336                 return (NULL);
 337         }
 338         return (f);
 339 }
 340
 341
 342 /*****************************************************************************
 343  * Helper functions for talking to the server and parsing its replies
 344  */
 345
 346 /* Header types */
 347 typedef enum {
 348         hdr_syserror = -2,
 349         hdr_error = -1,
 350         hdr_end = 0,
 351         hdr_unknown = 1,
 352         hdr_content_length,
 353         hdr_content_range,
 354         hdr_last_modified,
 355         hdr_location,
 356         hdr_transfer_encoding,
 357         hdr_www_authenticate,
 358         hdr_proxy_authenticate,
 359 } hdr_t;
 360
 361 /* Names of interesting headers */
 362 static struct {
 363         hdr_t            num;
 364         const char      *name;
 365 } hdr_names[] = {
 366         { hdr_content_length,           "Content-Length" },
 367         { hdr_content_range,            "Content-Range" },
 368         { hdr_last_modified,            "Last-Modified" },
 369         { hdr_location,                 "Location" },
 370         { hdr_transfer_encoding,        "Transfer-Encoding" },
 371         { hdr_www_authenticate,         "WWW-Authenticate" },
 372         { hdr_proxy_authenticate,       "Proxy-Authenticate" },
 373         { hdr_unknown,                  NULL },
 374 };
 375
 376 /*
 377  * Send a formatted line; optionally echo to terminal
 378  */
 379 static int
 380 http_cmd(conn_t *conn, const char *fmt, ...)
 381 {
 382         va_list ap;
 383         size_t len;
 384         char *msg;
 385         int r;
 386
 387         va_start(ap, fmt);
 388         len = vasprintf(&msg, fmt, ap);
 389         va_end(ap);
 390
 391         if (msg == NULL) {
 392                 errno = ENOMEM;
 393                 fetch_syserr();
 394                 return (-1);
 395         }
 396
 397         r = fetch_putln(conn, msg, len);
 398         free(msg);
 399
 400         if (r == -1) {
 401                 fetch_syserr();
 402                 return (-1);
 403         }
 404
 405         return (0);
 406 }
 407
 408 /*
 409  * Get and parse status line
 410  */
 411 static int
 412 http_get_reply(conn_t *conn)
 413 {
 414         char *p;
 415
 416         if (fetch_getln(conn) == -1)
 417                 return (-1);
 418         /*
 419          * A valid status line looks like "HTTP/m.n xyz reason" where m
 420          * and n are the major and minor protocol version numbers and xyz
 421          * is the reply code.
 422          * Unfortunately, there are servers out there (NCSA 1.5.1, to name
 423          * just one) that do not send a version number, so we can't rely
 424          * on finding one, but if we do, insist on it being 1.0 or 1.1.
 425          * We don't care about the reason phrase.
 426          */
 427         if (strncmp(conn->buf, "HTTP", 4) != 0)
 428                 return (HTTP_PROTOCOL_ERROR);
 429         p = conn->buf + 4;
 430         if (*p == '/') {
 431                 if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1'))
 432                         return (HTTP_PROTOCOL_ERROR);
 433                 p += 4;
 434         }
 435         if (*p != ' ' ||
 436             !isdigit((unsigned char)p[1]) ||
 437             !isdigit((unsigned char)p[2]) ||
 438             !isdigit((unsigned char)p[3]))
 439                 return (HTTP_PROTOCOL_ERROR);
 440
 441         conn->err = (p[1] - '0') * 100 + (p[2] - '0') * 10 + (p[3] - '0');
 442         return (conn->err);
 443 }
 444
 445 /*
 446  * Check a header; if the type matches the given string, return a pointer
 447  * to the beginning of the value.
 448  */
 449 static const char *
 450 http_match(const char *str, const char *hdr)
 451 {
 452         while (*str && *hdr &&
 453             tolower((unsigned char)*str++) == tolower((unsigned char)*hdr++))
 454                 /* nothing */;
 455         if (*str || *hdr != ':')
 456                 return (NULL);
 457         while (*hdr && isspace((unsigned char)*++hdr))
 458                 /* nothing */;
 459         return (hdr);
 460 }
 461
 462
 463 /*
 464  * Get the next header and return the appropriate symbolic code.  We
 465  * need to read one line ahead for checking for a continuation line
 466  * belonging to the current header (continuation lines start with
 467  * white space).
 468  *
 469  * We get called with a fresh line already in the conn buffer, either
 470  * from the previous http_next_header() invocation, or, the first
 471  * time, from a fetch_getln() performed by our caller.
 472  *
 473  * This stops when we encounter an empty line (we dont read beyond the header
 474  * area).
 475  *
 476  * Note that the "headerbuf" is just a place to return the result. Its
 477  * contents are not used for the next call. This means that no cleanup
 478  * is needed when ie doing another connection, just call the cleanup when
 479  * fully done to deallocate memory.
 480  */
 481
 482 /* Limit the max number of continuation lines to some reasonable value */
 483 #define HTTP_MAX_CONT_LINES 10
 484
 485 /* Place into which to build a header from one or several lines */
 486 typedef struct {
 487         char    *buf;           /* buffer */
 488         size_t   bufsize;       /* buffer size */
 489         size_t   buflen;        /* length of buffer contents */
 490 } http_headerbuf_t;
 491
 492 static void
 493 init_http_headerbuf(http_headerbuf_t *buf)
 494 {
 495         buf->buf = NULL;
 496         buf->bufsize = 0;
 497         buf->buflen = 0;
 498 }
 499
 500 static void
 501 clean_http_headerbuf(http_headerbuf_t *buf)
 502 {
 503         if (buf->buf)
 504                 free(buf->buf);
 505         init_http_headerbuf(buf);
 506 }
 507
 508 /* Remove whitespace at the end of the buffer */
 509 static void
 510 http_conn_trimright(conn_t *conn)
 511 {
 512         while (conn->buflen &&
 513                isspace((unsigned char)conn->buf[conn->buflen - 1]))
 514                 conn->buflen--;
 515         conn->buf[conn->buflen] = '\0';
 516 }
 517
 518 static hdr_t
 519 http_next_header(conn_t *conn, http_headerbuf_t *hbuf, const char **p)
 520 {
 521         unsigned int i, len;
 522
 523         /*
 524          * Have to do the stripping here because of the first line. So
 525          * it's done twice for the subsequent lines. No big deal
 526          */
 527         http_conn_trimright(conn);
 528         if (conn->buflen == 0)
 529                 return (hdr_end);
 530
 531         /* Copy the line to the headerbuf */
 532         if (hbuf->bufsize < conn->buflen + 1) {
 533                 if ((hbuf->buf = realloc(hbuf->buf, conn->buflen + 1)) == NULL)
 534                         return (hdr_syserror);
 535                 hbuf->bufsize = conn->buflen + 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 carry on the idea from the previous version
 550                  * that a pure whitespace line is equivalent to an empty
 551                  * one (so it's not continuation and will be handled when
 552                  * we are called next)
 553                  */
 554                 http_conn_trimright(conn);
 555                 if (conn->buf[0] != ' ' && conn->buf[0] != "\t"[0])
 556                         break;
 557
 558                 /* Got a continuation line. Concatenate to previous */
 559                 len = hbuf->buflen + conn->buflen;
 560                 if (hbuf->bufsize < len + 1) {
 561                         len *= 2;
 562                         if ((hbuf->buf = realloc(hbuf->buf, len + 1)) == NULL)
 563                                 return (hdr_syserror);
 564                         hbuf->bufsize = len + 1;
 565                 }
 566                 strcpy(hbuf->buf + hbuf->buflen, conn->buf);
 567                 hbuf->buflen += conn->buflen;
 568         }
 569
 570         /*
 571          * We could check for malformed headers but we don't really care.
 572          * A valid header starts with a token immediately followed by a
 573          * colon; a token is any sequence of non-control, non-whitespace
 574          * characters except "()<>@,;:\\\"{}".
 575          */
 576         for (i = 0; hdr_names[i].num != hdr_unknown; i++)
 577                 if ((*p = http_match(hdr_names[i].name, hbuf->buf)) != NULL)
 578                         return (hdr_names[i].num);
 579
 580         return (hdr_unknown);
 581 }
 582
 583 /**************************
 584  * [Proxy-]Authenticate header parsing
 585  */
 586
 587 /*
 588  * Read doublequote-delimited string into output buffer obuf (allocated
 589  * by caller, whose responsibility it is to ensure that it's big enough)
 590  * cp points to the first char after the initial '"'
 591  * Handles \ quoting
 592  * Returns pointer to the first char after the terminating double quote, or
 593  * NULL for error.
 594  */
 595 static const char *
 596 http_parse_headerstring(const char *cp, char *obuf)
 597 {
 598         for (;;) {
 599                 switch (*cp) {
 600                 case 0: /* Unterminated string */
 601                         *obuf = 0;
 602                         return (NULL);
 603                 case '"': /* Ending quote */
 604                         *obuf = 0;
 605                         return (++cp);
 606                 case '\\':
 607                         if (*++cp == 0) {
 608                                 *obuf = 0;
 609                                 return (NULL);
 610                         }
 611                         /* FALLTHROUGH */
 612                 default:
 613                         *obuf++ = *cp++;
 614                 }
 615         }
 616 }
 617
 618 /* Http auth challenge schemes */
 619 typedef enum {HTTPAS_UNKNOWN, HTTPAS_BASIC,HTTPAS_DIGEST} http_auth_schemes_t;
 620
 621 /* Data holder for a Basic or Digest challenge. */
 622 typedef struct {
 623         http_auth_schemes_t scheme;
 624         char    *realm;
 625         char    *qop;
 626         char    *nonce;
 627         char    *opaque;
 628         char    *algo;
 629         int      stale;
 630         int      nc; /* Nonce count */
 631 } http_auth_challenge_t;
 632
 633 static void
 634 init_http_auth_challenge(http_auth_challenge_t *b)
 635 {
 636         b->scheme = HTTPAS_UNKNOWN;
 637         b->realm = b->qop = b->nonce = b->opaque = b->algo = NULL;
 638         b->stale = b->nc = 0;
 639 }
 640
 641 static void
 642 clean_http_auth_challenge(http_auth_challenge_t *b)
 643 {
 644         if (b->realm)
 645                 free(b->realm);
 646         if (b->qop)
 647                 free(b->qop);
 648         if (b->nonce)
 649                 free(b->nonce);
 650         if (b->opaque)
 651                 free(b->opaque);
 652         if (b->algo)
 653                 free(b->algo);
 654         init_http_auth_challenge(b);
 655 }
 656
 657 /* Data holder for an array of challenges offered in an http response. */
 658 #define MAX_CHALLENGES 10
 659 typedef struct {
 660         http_auth_challenge_t *challenges[MAX_CHALLENGES];
 661         int     count; /* Number of parsed challenges in the array */
 662         int     valid; /* We did parse an authenticate header */
 663 } http_auth_challenges_t;
 664
 665 static void
 666 init_http_auth_challenges(http_auth_challenges_t *cs)
 667 {
 668         int i;
 669         for (i = 0; i < MAX_CHALLENGES; i++)
 670                 cs->challenges[i] = NULL;
 671         cs->count = cs->valid = 0;
 672 }
 673
 674 static void
 675 clean_http_auth_challenges(http_auth_challenges_t *cs)
 676 {
 677         int i;
 678         /* We rely on non-zero pointers being allocated, not on the count */
 679         for (i = 0; i < MAX_CHALLENGES; i++) {
 680                 if (cs->challenges[i] != NULL) {
 681                         clean_http_auth_challenge(cs->challenges[i]);
 682                         free(cs->challenges[i]);
 683                 }
 684         }
 685         init_http_auth_challenges(cs);
 686 }
 687
 688 /*
 689  * Enumeration for lexical elements. Separators will be returned as their own
 690  * ascii value
 691  */
 692 typedef enum {HTTPHL_WORD=256, HTTPHL_STRING=257, HTTPHL_END=258,
 693               HTTPHL_ERROR = 259} http_header_lex_t;
 694
 695 /*
 696  * Determine what kind of token comes next and return possible value
 697  * in buf, which is supposed to have been allocated big enough by
 698  * caller. Advance input pointer and return element type.
 699  */
 700 static int
 701 http_header_lex(const char **cpp, char *buf)
 702 {
 703         size_t l;
 704         /* Eat initial whitespace */
 705         *cpp += strspn(*cpp, " \t");
 706         if (**cpp == 0)
 707                 return (HTTPHL_END);
 708
 709         /* Separator ? */
 710         if (**cpp == ',' || **cpp == '=')
 711                 return (*((*cpp)++));
 712
 713         /* String ? */
 714         if (**cpp == '"') {
 715                 *cpp = http_parse_headerstring(++*cpp, buf);
 716                 if (*cpp == NULL)
 717                         return (HTTPHL_ERROR);
 718                 return (HTTPHL_STRING);
 719         }
 720
 721         /* Read other token, until separator or whitespace */
 722         l = strcspn(*cpp, " \t,=");
 723         memcpy(buf, *cpp, l);
 724         buf[l] = 0;
 725         *cpp += l;
 726         return (HTTPHL_WORD);
 727 }
 728
 729 /*
 730  * Read challenges from http xxx-authenticate header and accumulate them
 731  * in the challenges list structure.
 732  *
 733  * Headers with multiple challenges are specified by rfc2617, but
 734  * servers (ie: squid) often send them in separate headers instead,
 735  * which in turn is forbidden by the http spec (multiple headers with
 736  * the same name are only allowed for pure comma-separated lists, see
 737  * rfc2616 sec 4.2).
 738  *
 739  * We support both approaches anyway
 740  */
 741 static int
 742 http_parse_authenticate(const char *cp, http_auth_challenges_t *cs)
 743 {
 744         int ret = -1;
 745         http_header_lex_t lex;
 746         char *key = malloc(strlen(cp) + 1);
 747         char *value = malloc(strlen(cp) + 1);
 748         char *buf = malloc(strlen(cp) + 1);
 749
 750         if (key == NULL || value == NULL || buf == NULL) {
 751                 fetch_syserr();
 752                 goto out;
 753         }
 754
 755         /* In any case we've seen the header and we set the valid bit */
 756         cs->valid = 1;
 757
 758         /* Need word first */
 759         lex = http_header_lex(&cp, key);
 760         if (lex != HTTPHL_WORD)
 761                 goto out;
 762
 763         /* Loop on challenges */
 764         for (; cs->count < MAX_CHALLENGES; cs->count++) {
 765                 cs->challenges[cs->count] =
 766                         malloc(sizeof(http_auth_challenge_t));
 767                 if (cs->challenges[cs->count] == NULL) {
 768                         fetch_syserr();
 769                         goto out;
 770                 }
 771                 init_http_auth_challenge(cs->challenges[cs->count]);
 772                 if (!strcasecmp(key, "basic")) {
 773                         cs->challenges[cs->count]->scheme = HTTPAS_BASIC;
 774                 } else if (!strcasecmp(key, "digest")) {
 775                         cs->challenges[cs->count]->scheme = HTTPAS_DIGEST;
 776                 } else {
 777                         cs->challenges[cs->count]->scheme = HTTPAS_UNKNOWN;
 778                         /*
 779                          * Continue parsing as basic or digest may
 780                          * follow, and the syntax is the same for
 781                          * all. We'll just ignore this one when
 782                          * looking at the list
 783                          */
 784                 }
 785
 786                 /* Loop on attributes */
 787                 for (;;) {
 788                         /* Key */
 789                         lex = http_header_lex(&cp, key);
 790                         if (lex != HTTPHL_WORD)
 791                                 goto out;
 792
 793                         /* Equal sign */
 794                         lex = http_header_lex(&cp, buf);
 795                         if (lex != '=')
 796                                 goto out;
 797
 798                         /* Value */
 799                         lex = http_header_lex(&cp, value);
 800                         if (lex != HTTPHL_WORD && lex != HTTPHL_STRING)
 801                                 goto out;
 802
 803                         if (!strcasecmp(key, "realm"))
 804                                 cs->challenges[cs->count]->realm =
 805                                         strdup(value);
 806                         else if (!strcasecmp(key, "qop"))
 807                                 cs->challenges[cs->count]->qop =
 808                                         strdup(value);
 809                         else if (!strcasecmp(key, "nonce"))
 810                                 cs->challenges[cs->count]->nonce =
 811                                         strdup(value);
 812                         else if (!strcasecmp(key, "opaque"))
 813                                 cs->challenges[cs->count]->opaque =
 814                                         strdup(value);
 815                         else if (!strcasecmp(key, "algorithm"))
 816                                 cs->challenges[cs->count]->algo =
 817                                         strdup(value);
 818                         else if (!strcasecmp(key, "stale"))
 819                                 cs->challenges[cs->count]->stale =
 820                                         strcasecmp(value, "no");
 821                         /* Else ignore unknown attributes */
 822
 823                         /* Comma or Next challenge or End */
 824                         lex = http_header_lex(&cp, key);
 825                         /*
 826                          * If we get a word here, this is the beginning of the
 827                          * next challenge. Break the attributes loop
 828                          */
 829                         if (lex == HTTPHL_WORD)
 830                                 break;
 831
 832                         if (lex == HTTPHL_END) {
 833                                 /* End while looking for ',' is normal exit */
 834                                 cs->count++;
 835                                 ret = 0;
 836                                 goto out;
 837                         }
 838                         /* Anything else is an error */
 839                         if (lex != ',')
 840                                 goto out;
 841
 842                 } /* End attributes loop */
 843         } /* End challenge loop */
 844
 845         /*
 846          * Challenges max count exceeded. This really can't happen
 847          * with normal data, something's fishy -> error
 848          */
 849
 850 out:
 851         if (key)
 852                 free(key);
 853         if (value)
 854                 free(value);
 855         if (buf)
 856                 free(buf);
 857         return (ret);
 858 }
 859
 860
 861 /*
 862  * Parse a last-modified header
 863  */
 864 static int
 865 http_parse_mtime(const char *p, time_t *mtime)
 866 {
 867         char locale[64], *r;
 868         struct tm tm;
 869
 870         strncpy(locale, setlocale(LC_TIME, NULL), sizeof(locale));
 871         setlocale(LC_TIME, "C");
 872         r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm);
 873         /* XXX should add support for date-2 and date-3 */
 874         setlocale(LC_TIME, locale);
 875         if (r == NULL)
 876                 return (-1);
 877         DEBUG(fprintf(stderr, "last modified: [%04d-%02d-%02d "
 878                   "%02d:%02d:%02d]\n",
 879                   tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday,
 880                   tm.tm_hour, tm.tm_min, tm.tm_sec));
 881         *mtime = timegm(&tm);
 882         return (0);
 883 }
 884
 885 /*
 886  * Parse a content-length header
 887  */
 888 static int
 889 http_parse_length(const char *p, off_t *length)
 890 {
 891         off_t len;
 892
 893         for (len = 0; *p && isdigit((unsigned char)*p); ++p)
 894                 len = len * 10 + (*p - '0');
 895         if (*p)
 896                 return (-1);
 897         DEBUG(fprintf(stderr, "content length: [%lld]\n",
 898             (long long)len));
 899         *length = len;
 900         return (0);
 901 }
 902
 903 /*
 904  * Parse a content-range header
 905  */
 906 static int
 907 http_parse_range(const char *p, off_t *offset, off_t *length, off_t *size)
 908 {
 909         off_t first, last, len;
 910
 911         if (strncasecmp(p, "bytes ", 6) != 0)
 912                 return (-1);
 913         p += 6;
 914         if (*p == '*') {
 915                 first = last = -1;
 916                 ++p;
 917         } else {
 918                 for (first = 0; *p && isdigit((unsigned char)*p); ++p)
 919                         first = first * 10 + *p - '0';
 920                 if (*p != '-')
 921                         return (-1);
 922                 for (last = 0, ++p; *p && isdigit((unsigned char)*p); ++p)
 923                         last = last * 10 + *p - '0';
 924         }
 925         if (first > last || *p != '/')
 926                 return (-1);
 927         for (len = 0, ++p; *p && isdigit((unsigned char)*p); ++p)
 928                 len = len * 10 + *p - '0';
 929         if (*p || len < last - first + 1)
 930                 return (-1);
 931         if (first == -1) {
 932                 DEBUG(fprintf(stderr, "content range: [*/%lld]\n",
 933                     (long long)len));
 934                 *length = 0;
 935         } else {
 936                 DEBUG(fprintf(stderr, "content range: [%lld-%lld/%lld]\n",
 937                     (long long)first, (long long)last, (long long)len));
 938                 *length = last - first + 1;
 939         }
 940         *offset = first;
 941         *size = len;
 942         return (0);
 943 }
 944
 945
 946 /*****************************************************************************
 947  * Helper functions for authorization
 948  */
 949
 950 /*
 951  * Base64 encoding
 952  */
 953 static char *
 954 http_base64(const char *src)
 955 {
 956         static const char base64[] =
 957             "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 958             "abcdefghijklmnopqrstuvwxyz"
 959             "0123456789+/";
 960         char *str, *dst;
 961         size_t l;
 962         int t, r;
 963
 964         l = strlen(src);
 965         if ((str = malloc(((l + 2) / 3) * 4 + 1)) == NULL)
 966                 return (NULL);
 967         dst = str;
 968         r = 0;
 969
 970         while (l >= 3) {
 971                 t = (src[0] << 16) | (src[1] << 8) | src[2];
 972                 dst[0] = base64[(t >> 18) & 0x3f];
 973                 dst[1] = base64[(t >> 12) & 0x3f];
 974                 dst[2] = base64[(t >> 6) & 0x3f];
 975                 dst[3] = base64[(t >> 0) & 0x3f];
 976                 src += 3; l -= 3;
 977                 dst += 4; r += 4;
 978         }
 979
 980         switch (l) {
 981         case 2:
 982                 t = (src[0] << 16) | (src[1] << 8);
 983                 dst[0] = base64[(t >> 18) & 0x3f];
 984                 dst[1] = base64[(t >> 12) & 0x3f];
 985                 dst[2] = base64[(t >> 6) & 0x3f];
 986                 dst[3] = '=';
 987                 dst += 4;
 988                 r += 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                 r += 4;
 997                 break;
 998         case 0:
 999                 break;
1000         }
1001
1002         *dst = 0;
1003         return (str);
1004 }
1005
1006
1007 /*
1008  * Extract authorization parameters from environment value.
1009  * The value is like scheme:realm:user:pass
1010  */
1011 typedef struct {
1012         char    *scheme;
1013         char    *realm;
1014         char    *user;
1015         char    *password;
1016 } http_auth_params_t;
1017
1018 static void
1019 init_http_auth_params(http_auth_params_t *s)
1020 {
1021         s->scheme = s->realm = s->user = s->password = 0;
1022 }
1023
1024 static void
1025 clean_http_auth_params(http_auth_params_t *s)
1026 {
1027         if (s->scheme)
1028                 free(s->scheme);
1029         if (s->realm)
1030                 free(s->realm);
1031         if (s->user)
1032                 free(s->user);
1033         if (s->password)
1034                 free(s->password);
1035         init_http_auth_params(s);
1036 }
1037
1038 static int
1039 http_authfromenv(const char *p, http_auth_params_t *parms)
1040 {
1041         int ret = -1;
1042         char *v, *ve;
1043         char *str = strdup(p);
1044
1045         if (str == NULL) {
1046                 fetch_syserr();
1047                 return (-1);
1048         }
1049         v = str;
1050
1051         if ((ve = strchr(v, ':')) == NULL)
1052                 goto out;
1053
1054         *ve = 0;
1055         if ((parms->scheme = 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->realm = strdup(v)) == NULL) {
1066                 fetch_syserr();
1067                 goto out;
1068         }
1069         v = ve + 1;
1070
1071         if ((ve = strchr(v, ':')) == NULL)
1072                 goto out;
1073
1074         *ve = 0;
1075         if ((parms->user = strdup(v)) == NULL) {
1076                 fetch_syserr();
1077                 goto out;
1078         }
1079         v = ve + 1;
1080
1081
1082         if ((parms->password = strdup(v)) == NULL) {
1083                 fetch_syserr();
1084                 goto out;
1085         }
1086         ret = 0;
1087 out:
1088         if (ret == -1)
1089                 clean_http_auth_params(parms);
1090         if (str)
1091                 free(str);
1092         return (ret);
1093 }
1094
1095
1096 /*
1097  * Digest response: the code to compute the digest is taken from the
1098  * sample implementation in RFC2616
1099  */
1100 #define IN const
1101 #define OUT
1102
1103 #define HASHLEN 16
1104 typedef char HASH[HASHLEN];
1105 #define HASHHEXLEN 32
1106 typedef char HASHHEX[HASHHEXLEN+1];
1107
1108 static const char *hexchars = "0123456789abcdef";
1109 static void
1110 CvtHex(IN HASH Bin, OUT HASHHEX Hex)
1111 {
1112         unsigned short i;
1113         unsigned char j;
1114
1115         for (i = 0; i < HASHLEN; i++) {
1116                 j = (Bin[i] >> 4) & 0xf;
1117                 Hex[i*2] = hexchars[j];
1118                 j = Bin[i] & 0xf;
1119                 Hex[i*2+1] = hexchars[j];
1120         };
1121         Hex[HASHHEXLEN] = '\0';
1122 };
1123
1124 /* calculate H(A1) as per spec */
1125 static void
1126 DigestCalcHA1(
1127         IN char * pszAlg,
1128         IN char * pszUserName,
1129         IN char * pszRealm,
1130         IN char * pszPassword,
1131         IN char * pszNonce,
1132         IN char * pszCNonce,
1133         OUT HASHHEX SessionKey
1134         )
1135 {
1136         MD5_CTX Md5Ctx;
1137         HASH HA1;
1138
1139         MD5Init(&Md5Ctx);
1140         MD5Update(&Md5Ctx, pszUserName, strlen(pszUserName));
1141         MD5Update(&Md5Ctx, ":", 1);
1142         MD5Update(&Md5Ctx, pszRealm, strlen(pszRealm));
1143         MD5Update(&Md5Ctx, ":", 1);
1144         MD5Update(&Md5Ctx, pszPassword, strlen(pszPassword));
1145         MD5Final(HA1, &Md5Ctx);
1146         if (strcasecmp(pszAlg, "md5-sess") == 0) {
1147
1148                 MD5Init(&Md5Ctx);
1149                 MD5Update(&Md5Ctx, HA1, HASHLEN);
1150                 MD5Update(&Md5Ctx, ":", 1);
1151                 MD5Update(&Md5Ctx, pszNonce, strlen(pszNonce));
1152                 MD5Update(&Md5Ctx, ":", 1);
1153                 MD5Update(&Md5Ctx, pszCNonce, strlen(pszCNonce));
1154                 MD5Final(HA1, &Md5Ctx);
1155         };
1156         CvtHex(HA1, SessionKey);
1157 }
1158
1159 /* calculate request-digest/response-digest as per HTTP Digest spec */
1160 static void
1161 DigestCalcResponse(
1162         IN HASHHEX HA1,           /* H(A1) */
1163         IN char * pszNonce,       /* nonce from server */
1164         IN char * pszNonceCount,  /* 8 hex digits */
1165         IN char * pszCNonce,      /* client nonce */
1166         IN char * pszQop,         /* qop-value: "", "auth", "auth-int" */
1167         IN char * pszMethod,      /* method from the request */
1168         IN char * pszDigestUri,   /* requested URL */
1169         IN HASHHEX HEntity,       /* H(entity body) if qop="auth-int" */
1170         OUT HASHHEX Response      /* request-digest or response-digest */
1171         )
1172 {
1173 /*      DEBUG(fprintf(stderr,
1174                       "Calc: HA1[%s] Nonce[%s] qop[%s] method[%s] URI[%s]\n",
1175                       HA1, pszNonce, pszQop, pszMethod, pszDigestUri));*/
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 = 0;
1241
1242         if (!c->realm || !c->nonce) {
1243                 DEBUG(fprintf(stderr, "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:"")== -1)
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         DEBUG(fprintf(stderr, "HA1: [%s]\n", HA1));
1269         HASHHEX digest;
1270         DigestCalcResponse(HA1, c->nonce, noncecount, cnonce, c->qop,
1271                            "GET", url->doc, "", digest);
1272
1273         if (c->qop[0]) {
1274                 r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\","
1275                              "nonce=\"%s\",uri=\"%s\",response=\"%s\","
1276                              "qop=\"auth\", cnonce=\"%s\", nc=%s%s",
1277                              hdr, parms->user, c->realm,
1278                              c->nonce, url->doc, digest,
1279                              cnonce, noncecount, options);
1280         } else {
1281                 r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\","
1282                              "nonce=\"%s\",uri=\"%s\",response=\"%s\"%s",
1283                              hdr, parms->user, c->realm,
1284                              c->nonce, url->doc, digest, options);
1285         }
1286         if (options)
1287                 free(options);
1288         return (r);
1289 }
1290
1291 /*
1292  * Encode username and password
1293  */
1294 static int
1295 http_basic_auth(conn_t *conn, const char *hdr, const char *usr, const char *pwd)
1296 {
1297         char *upw, *auth;
1298         int r;
1299
1300         DEBUG(fprintf(stderr, "basic: usr: [%s]\n", usr));
1301         DEBUG(fprintf(stderr, "basic: pwd: [%s]\n", pwd));
1302         if (asprintf(&upw, "%s:%s", usr, pwd) == -1)
1303                 return (-1);
1304         auth = http_base64(upw);
1305         free(upw);
1306         if (auth == NULL)
1307                 return (-1);
1308         r = http_cmd(conn, "%s: Basic %s", hdr, auth);
1309         free(auth);
1310         return (r);
1311 }
1312
1313 /*
1314  * Chose the challenge to answer and call the appropriate routine to
1315  * produce the header.
1316  */
1317 static int
1318 http_authorize(conn_t *conn, const char *hdr, http_auth_challenges_t *cs,
1319                http_auth_params_t *parms, struct url *url)
1320 {
1321         http_auth_challenge_t *basic = NULL;
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                 DEBUG(fprintf(stderr, "NULL usr or pass\n"));
1328                 return (-1);
1329         }
1330
1331         /* Look for a Digest and a Basic challenge */
1332         for (i = 0; i < cs->count; i++) {
1333                 if (cs->challenges[i]->scheme == HTTPAS_BASIC)
1334                         basic = cs->challenges[i];
1335                 if (cs->challenges[i]->scheme == HTTPAS_DIGEST)
1336                         digest = cs->challenges[i];
1337         }
1338
1339         /* Error if "Digest" was specified and there is no Digest challenge */
1340         if (!digest && (parms->scheme &&
1341                         !strcasecmp(parms->scheme, "digest"))) {
1342                 DEBUG(fprintf(stderr,
1343                               "Digest auth in env, not supported by peer\n"));
1344                 return (-1);
1345         }
1346         /*
1347          * If "basic" was specified in the environment, or there is no Digest
1348          * challenge, do the basic thing. Don't need a challenge for this,
1349          * so no need to check basic!=NULL
1350          */
1351         if (!digest || (parms->scheme && !strcasecmp(parms->scheme,"basic")))
1352                 return (http_basic_auth(conn,hdr,parms->user,parms->password));
1353
1354         /* Else, prefer digest. We just checked that it's not NULL */
1355         return (http_digest_auth(conn, hdr, digest, parms, url));
1356 }
1357
1358 /*****************************************************************************
1359  * Helper functions for connecting to a server or proxy
1360  */
1361
1362 /*
1363  * Connect to the correct HTTP server or proxy.
1364  */
1365 static conn_t *
1366 http_connect(struct url *URL, struct url *purl, const char *flags)
1367 {
1368         conn_t *conn;
1369         int verbose;
1370         int af, val;
1371
1372 #ifdef INET6
1373         af = AF_UNSPEC;
1374 #else
1375         af = AF_INET;
1376 #endif
1377
1378         verbose = CHECK_FLAG('v');
1379         if (CHECK_FLAG('4'))
1380                 af = AF_INET;
1381 #ifdef INET6
1382         else if (CHECK_FLAG('6'))
1383                 af = AF_INET6;
1384 #endif
1385
1386         if (purl && strcasecmp(URL->scheme, SCHEME_HTTPS) != 0) {
1387                 URL = purl;
1388         } else if (strcasecmp(URL->scheme, SCHEME_FTP) == 0) {
1389                 /* can't talk http to an ftp server */
1390                 /* XXX should set an error code */
1391                 return (NULL);
1392         }
1393
1394         if ((conn = fetch_connect(URL->host, URL->port, af, verbose)) == NULL)
1395                 /* fetch_connect() has already set an error code */
1396                 return (NULL);
1397         if (strcasecmp(URL->scheme, SCHEME_HTTPS) == 0 &&
1398             fetch_ssl(conn, verbose) == -1) {
1399                 fetch_close(conn);
1400                 /* grrr */
1401                 errno = EAUTH;
1402                 fetch_syserr();
1403                 return (NULL);
1404         }
1405
1406         val = 1;
1407         setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val, sizeof(val));
1408
1409         return (conn);
1410 }
1411
1412 static struct url *
1413 http_get_proxy(struct url * url, const char *flags)
1414 {
1415         struct url *purl;
1416         char *p;
1417
1418         if (flags != NULL && strchr(flags, 'd') != NULL)
1419                 return (NULL);
1420         if (fetch_no_proxy_match(url->host))
1421                 return (NULL);
1422         if (((p = getenv("HTTP_PROXY")) || (p = getenv("http_proxy"))) &&
1423             *p && (purl = fetchParseURL(p))) {
1424                 if (!*purl->scheme)
1425                         strcpy(purl->scheme, SCHEME_HTTP);
1426                 if (!purl->port)
1427                         purl->port = fetch_default_proxy_port(purl->scheme);
1428                 if (strcasecmp(purl->scheme, SCHEME_HTTP) == 0)
1429                         return (purl);
1430                 fetchFreeURL(purl);
1431         }
1432         return (NULL);
1433 }
1434
1435 static void
1436 http_print_html(FILE *out, FILE *in)
1437 {
1438         size_t len;
1439         char *line, *p, *q;
1440         int comment, tag;
1441
1442         comment = tag = 0;
1443         while ((line = fgetln(in, &len)) != NULL) {
1444                 while (len && isspace((unsigned char)line[len - 1]))
1445                         --len;
1446                 for (p = q = line; q < line + len; ++q) {
1447                         if (comment && *q == '-') {
1448                                 if (q + 2 < line + len &&
1449                                     strcmp(q, "-->") == 0) {
1450                                         tag = comment = 0;
1451                                         q += 2;
1452                                 }
1453                         } else if (tag && !comment && *q == '>') {
1454                                 p = q + 1;
1455                                 tag = 0;
1456                         } else if (!tag && *q == '<') {
1457                                 if (q > p)
1458                                         fwrite(p, q - p, 1, out);
1459                                 tag = 1;
1460                                 if (q + 3 < line + len &&
1461                                     strcmp(q, "<!--") == 0) {
1462                                         comment = 1;
1463                                         q += 3;
1464                                 }
1465                         }
1466                 }
1467                 if (!tag && q > p)
1468                         fwrite(p, q - p, 1, out);
1469                 fputc('\n', out);
1470         }
1471 }
1472
1473
1474 /*****************************************************************************
1475  * Core
1476  */
1477
1478 /*
1479  * Send a request and process the reply
1480  *
1481  * XXX This function is way too long, the do..while loop should be split
1482  * XXX off into a separate function.
1483  */
1484 FILE *
1485 http_request(struct url *URL, const char *op, struct url_stat *us,
1486         struct url *purl, const char *flags)
1487 {
1488         char timebuf[80];
1489         char hbuf[MAXHOSTNAMELEN + 7], *host;
1490         conn_t *conn;
1491         struct url *url, *new;
1492         int chunked, direct, ims, noredirect, verbose;
1493         int e, i, n, val;
1494         off_t offset, clength, length, size;
1495         time_t mtime;
1496         const char *p;
1497         FILE *f;
1498         hdr_t h;
1499         struct tm *timestruct;
1500         http_headerbuf_t headerbuf;
1501         http_auth_challenges_t server_challenges;
1502         http_auth_challenges_t proxy_challenges;
1503
1504         /* The following calls don't allocate anything */
1505         init_http_headerbuf(&headerbuf);
1506         init_http_auth_challenges(&server_challenges);
1507         init_http_auth_challenges(&proxy_challenges);
1508
1509         direct = CHECK_FLAG('d');
1510         noredirect = CHECK_FLAG('A');
1511         verbose = CHECK_FLAG('v');
1512         ims = CHECK_FLAG('i');
1513
1514         if (direct && purl) {
1515                 fetchFreeURL(purl);
1516                 purl = NULL;
1517         }
1518
1519         /* try the provided URL first */
1520         url = URL;
1521
1522         n = MAX_REDIRECT;
1523         i = 0;
1524
1525         e = HTTP_PROTOCOL_ERROR;
1526         do {
1527                 new = NULL;
1528                 chunked = 0;
1529                 offset = 0;
1530                 clength = -1;
1531                 length = -1;
1532                 size = -1;
1533                 mtime = 0;
1534
1535                 /* check port */
1536                 if (!url->port)
1537                         url->port = fetch_default_port(url->scheme);
1538
1539                 /* were we redirected to an FTP URL? */
1540                 if (purl == NULL && strcmp(url->scheme, SCHEME_FTP) == 0) {
1541                         if (strcmp(op, "GET") == 0)
1542                                 return (ftp_request(url, "RETR", us, purl, flags));
1543                         else if (strcmp(op, "HEAD") == 0)
1544                                 return (ftp_request(url, "STAT", us, purl, flags));
1545                 }
1546
1547                 /* connect to server or proxy */
1548                 if ((conn = http_connect(url, purl, flags)) == NULL)
1549                         goto ouch;
1550
1551                 host = url->host;
1552 #ifdef INET6
1553                 if (strchr(url->host, ':')) {
1554                         snprintf(hbuf, sizeof(hbuf), "[%s]", url->host);
1555                         host = hbuf;
1556                 }
1557 #endif
1558                 if (url->port != fetch_default_port(url->scheme)) {
1559                         if (host != hbuf) {
1560                                 strcpy(hbuf, host);
1561                                 host = hbuf;
1562                         }
1563                         snprintf(hbuf + strlen(hbuf),
1564                             sizeof(hbuf) - strlen(hbuf), ":%d", url->port);
1565                 }
1566
1567                 /* send request */
1568                 if (verbose)
1569                         fetch_info("requesting %s://%s%s",
1570                             url->scheme, host, url->doc);
1571                 if (purl) {
1572                         http_cmd(conn, "%s %s://%s%s HTTP/1.1",
1573                             op, url->scheme, host, url->doc);
1574                 } else {
1575                         http_cmd(conn, "%s %s HTTP/1.1",
1576                             op, url->doc);
1577                 }
1578
1579                 if (ims && url->ims_time) {
1580                         timestruct = gmtime((time_t *)&url->ims_time);
1581                         (void)strftime(timebuf, 80, "%a, %d %b %Y %T GMT",
1582                             timestruct);
1583                         if (verbose)
1584                                 fetch_info("If-Modified-Since: %s", timebuf);
1585                         http_cmd(conn, "If-Modified-Since: %s", timebuf);
1586                 }
1587                 /* virtual host */
1588                 http_cmd(conn, "Host: %s", host);
1589
1590                 /*
1591                  * Proxy authorization: we only send auth after we received
1592                  * a 407 error. We do not first try basic anyway (changed
1593                  * when support was added for digest-auth)
1594                  */
1595                 if (purl && proxy_challenges.valid) {
1596                         http_auth_params_t aparams;
1597                         init_http_auth_params(&aparams);
1598                         if (*purl->user || *purl->pwd) {
1599                                 aparams.user = purl->user ?
1600                                         strdup(purl->user) : strdup("");
1601                                 aparams.password = purl->pwd?
1602                                         strdup(purl->pwd) : strdup("");
1603                         } else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL &&
1604                                    *p != '\0') {
1605                                 if (http_authfromenv(p, &aparams) < 0) {
1606                                         http_seterr(HTTP_NEED_PROXY_AUTH);
1607                                         goto ouch;
1608                                 }
1609                         }
1610                         http_authorize(conn, "Proxy-Authorization",
1611                                        &proxy_challenges, &aparams, url);
1612                         clean_http_auth_params(&aparams);
1613                 }
1614
1615                 /*
1616                  * Server authorization: we never send "a priori"
1617                  * Basic auth, which used to be done if user/pass were
1618                  * set in the url. This would be weird because we'd send the
1619                  * password in the clear even if Digest is finally to be
1620                  * used (it would have made more sense for the
1621                  * pre-digest version to do this when Basic was specified
1622                  * in the environment)
1623                  */
1624                 if (server_challenges.valid) {
1625                         http_auth_params_t aparams;
1626                         init_http_auth_params(&aparams);
1627                         if (*url->user || *url->pwd) {
1628                                 aparams.user = url->user ?
1629                                         strdup(url->user) : strdup("");
1630                                 aparams.password = url->pwd ?
1631                                         strdup(url->pwd) : strdup("");
1632                         } else if ((p = getenv("HTTP_AUTH")) != NULL &&
1633                                    *p != '\0') {
1634                                 if (http_authfromenv(p, &aparams) < 0) {
1635                                         http_seterr(HTTP_NEED_AUTH);
1636                                         goto ouch;
1637                                 }
1638                         } else if (fetchAuthMethod &&
1639                                    fetchAuthMethod(url) == 0) {
1640                                 aparams.user = url->user ?
1641                                         strdup(url->user) : strdup("");
1642                                 aparams.password = url->pwd ?
1643                                         strdup(url->pwd) : strdup("");
1644                         } else {
1645                                 http_seterr(HTTP_NEED_AUTH);
1646                                 goto ouch;
1647                         }
1648                         http_authorize(conn, "Authorization",
1649                                        &server_challenges, &aparams, url);
1650                         clean_http_auth_params(&aparams);
1651                 }
1652
1653                 /* other headers */
1654                 if ((p = getenv("HTTP_REFERER")) != NULL && *p != '\0') {
1655                         if (strcasecmp(p, "auto") == 0)
1656                                 http_cmd(conn, "Referer: %s://%s%s",
1657                                     url->scheme, host, url->doc);
1658                         else
1659                                 http_cmd(conn, "Referer: %s", p);
1660                 }
1661                 if ((p = getenv("HTTP_USER_AGENT")) != NULL && *p != '\0')
1662                         http_cmd(conn, "User-Agent: %s", p);
1663                 else
1664                         http_cmd(conn, "User-Agent: %s " _LIBFETCH_VER, getprogname());
1665                 if (url->offset > 0)
1666                         http_cmd(conn, "Range: bytes=%lld-", (long long)url->offset);
1667                 http_cmd(conn, "Connection: close");
1668                 http_cmd(conn, "");
1669
1670                 /*
1671                  * Force the queued request to be dispatched.  Normally, one
1672                  * would do this with shutdown(2) but squid proxies can be
1673                  * configured to disallow such half-closed connections.  To
1674                  * be compatible with such configurations, fiddle with socket
1675                  * options to force the pending data to be written.
1676                  */
1677                 val = 0;
1678                 setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val,
1679                            sizeof(val));
1680                 val = 1;
1681                 setsockopt(conn->sd, IPPROTO_TCP, TCP_NODELAY, &val,
1682                            sizeof(val));
1683
1684                 /* get reply */
1685                 switch (http_get_reply(conn)) {
1686                 case HTTP_OK:
1687                 case HTTP_PARTIAL:
1688                 case HTTP_NOT_MODIFIED:
1689                         /* fine */
1690                         break;
1691                 case HTTP_MOVED_PERM:
1692                 case HTTP_MOVED_TEMP:
1693                 case HTTP_SEE_OTHER:
1694                 case HTTP_USE_PROXY:
1695                         /*
1696                          * Not so fine, but we still have to read the
1697                          * headers to get the new location.
1698                          */
1699                         break;
1700                 case HTTP_NEED_AUTH:
1701                         if (server_challenges.valid) {
1702                                 /*
1703                                  * We already sent out authorization code,
1704                                  * so there's nothing more we can do.
1705                                  */
1706                                 http_seterr(conn->err);
1707                                 goto ouch;
1708                         }
1709                         /* try again, but send the password this time */
1710                         if (verbose)
1711                                 fetch_info("server requires authorization");
1712                         break;
1713                 case HTTP_NEED_PROXY_AUTH:
1714                         if (proxy_challenges.valid) {
1715                                 /*
1716                                  * We already sent our proxy
1717                                  * authorization code, so there's
1718                                  * nothing more we can do. */
1719                                 http_seterr(conn->err);
1720                                 goto ouch;
1721                         }
1722                         /* try again, but send the password this time */
1723                         if (verbose)
1724                                 fetch_info("proxy requires authorization");
1725                         break;
1726                 case HTTP_BAD_RANGE:
1727                         /*
1728                          * This can happen if we ask for 0 bytes because
1729                          * we already have the whole file.  Consider this
1730                          * a success for now, and check sizes later.
1731                          */
1732                         break;
1733                 case HTTP_PROTOCOL_ERROR:
1734                         /* fall through */
1735                 case -1:
1736                         fetch_syserr();
1737                         goto ouch;
1738                 default:
1739                         http_seterr(conn->err);
1740                         if (!verbose)
1741                                 goto ouch;
1742                         /* fall through so we can get the full error message */
1743                 }
1744
1745                 /* get headers. http_next_header expects one line readahead */
1746                 if (fetch_getln(conn) == -1) {
1747                     fetch_syserr();
1748                     goto ouch;
1749                 }
1750                 do {
1751                     switch ((h = http_next_header(conn, &headerbuf, &p))) {
1752                         case hdr_syserror:
1753                                 fetch_syserr();
1754                                 goto ouch;
1755                         case hdr_error:
1756                                 http_seterr(HTTP_PROTOCOL_ERROR);
1757                                 goto ouch;
1758                         case hdr_content_length:
1759                                 http_parse_length(p, &clength);
1760                                 break;
1761                         case hdr_content_range:
1762                                 http_parse_range(p, &offset, &length, &size);
1763                                 break;
1764                         case hdr_last_modified:
1765                                 http_parse_mtime(p, &mtime);
1766                                 break;
1767                         case hdr_location:
1768                                 if (!HTTP_REDIRECT(conn->err))
1769                                         break;
1770                                 /*
1771                                  * if the A flag is set, we don't follow
1772                                  * temporary redirects.
1773                                  */
1774                                 if (noredirect &&
1775                                     conn->err != HTTP_MOVED_PERM &&
1776                                     conn->err != HTTP_PERM_REDIRECT &&
1777                                     conn->err != HTTP_USE_PROXY) {
1778                                         n = 1;
1779                                         break;
1780                                 }
1781                                 if (new)
1782                                         free(new);
1783                                 if (verbose)
1784                                         fetch_info("%d redirect to %s", conn->err, p);
1785                                 if (*p == '/')
1786                                         /* absolute path */
1787                                         new = fetchMakeURL(url->scheme, url->host, url->port, p,
1788                                             url->user, url->pwd);
1789                                 else
1790                                         new = fetchParseURL(p);
1791                                 if (new == NULL) {
1792                                         /* XXX should set an error code */
1793                                         DEBUG(fprintf(stderr, "failed to parse new URL\n"));
1794                                         goto ouch;
1795                                 }
1796
1797                                 /* Only copy credentials if the host matches */
1798                                 if (!strcmp(new->host, url->host) && !*new->user && !*new->pwd) {
1799                                         strcpy(new->user, url->user);
1800                                         strcpy(new->pwd, url->pwd);
1801                                 }
1802                                 new->offset = url->offset;
1803                                 new->length = url->length;
1804                                 break;
1805                         case hdr_transfer_encoding:
1806                                 /* XXX weak test*/
1807                                 chunked = (strcasecmp(p, "chunked") == 0);
1808                                 break;
1809                         case hdr_www_authenticate:
1810                                 if (conn->err != HTTP_NEED_AUTH)
1811                                         break;
1812                                 if (http_parse_authenticate(p, &server_challenges) == 0)
1813                                         ++n;
1814                                 break;
1815                         case hdr_proxy_authenticate:
1816                                 if (conn->err != HTTP_NEED_PROXY_AUTH)
1817                                         break;
1818                                 if (http_parse_authenticate(p, &proxy_challenges) == 0)
1819                                         ++n;
1820                                 break;
1821                         case hdr_end:
1822                                 /* fall through */
1823                         case hdr_unknown:
1824                                 /* ignore */
1825                                 break;
1826                         }
1827                 } while (h > hdr_end);
1828
1829                 /* we need to provide authentication */
1830                 if (conn->err == HTTP_NEED_AUTH ||
1831                     conn->err == HTTP_NEED_PROXY_AUTH) {
1832                         e = conn->err;
1833                         if ((conn->err == HTTP_NEED_AUTH &&
1834                              !server_challenges.valid) ||
1835                             (conn->err == HTTP_NEED_PROXY_AUTH &&
1836                              !proxy_challenges.valid)) {
1837                                 /* 401/7 but no www/proxy-authenticate ?? */
1838                                 DEBUG(fprintf(stderr, "401/7 and no auth header\n"));
1839                                 goto ouch;
1840                         }
1841                         fetch_close(conn);
1842                         conn = NULL;
1843                         continue;
1844                 }
1845
1846                 /* requested range not satisfiable */
1847                 if (conn->err == HTTP_BAD_RANGE) {
1848                         if (url->offset == size && url->length == 0) {
1849                                 /* asked for 0 bytes; fake it */
1850                                 offset = url->offset;
1851                                 clength = -1;
1852                                 conn->err = HTTP_OK;
1853                                 break;
1854                         } else {
1855                                 http_seterr(conn->err);
1856                                 goto ouch;
1857                         }
1858                 }
1859
1860                 /* we have a hit or an error */
1861                 if (conn->err == HTTP_OK
1862                     || conn->err == HTTP_NOT_MODIFIED
1863                     || conn->err == HTTP_PARTIAL
1864                     || HTTP_ERROR(conn->err))
1865                         break;
1866
1867                 /* all other cases: we got a redirect */
1868                 e = conn->err;
1869                 clean_http_auth_challenges(&server_challenges);
1870                 fetch_close(conn);
1871                 conn = NULL;
1872                 if (!new) {
1873                         DEBUG(fprintf(stderr, "redirect with no new location\n"));
1874                         break;
1875                 }
1876                 if (url != URL)
1877                         fetchFreeURL(url);
1878                 url = new;
1879         } while (++i < n);
1880
1881         /* we failed, or ran out of retries */
1882         if (conn == NULL) {
1883                 http_seterr(e);
1884                 goto ouch;
1885         }
1886
1887         DEBUG(fprintf(stderr, "offset %lld, length %lld,"
1888                   " size %lld, clength %lld\n",
1889                   (long long)offset, (long long)length,
1890                   (long long)size, (long long)clength));
1891
1892         if (conn->err == HTTP_NOT_MODIFIED) {
1893                 http_seterr(HTTP_NOT_MODIFIED);
1894                 return (NULL);
1895         }
1896
1897         /* check for inconsistencies */
1898         if (clength != -1 && length != -1 && clength != length) {
1899                 http_seterr(HTTP_PROTOCOL_ERROR);
1900                 goto ouch;
1901         }
1902         if (clength == -1)
1903                 clength = length;
1904         if (clength != -1)
1905                 length = offset + clength;
1906         if (length != -1 && size != -1 && length != size) {
1907                 http_seterr(HTTP_PROTOCOL_ERROR);
1908                 goto ouch;
1909         }
1910         if (size == -1)
1911                 size = length;
1912
1913         /* fill in stats */
1914         if (us) {
1915                 us->size = size;
1916                 us->atime = us->mtime = mtime;
1917         }
1918
1919         /* too far? */
1920         if (URL->offset > 0 && offset > URL->offset) {
1921                 http_seterr(HTTP_PROTOCOL_ERROR);
1922                 goto ouch;
1923         }
1924
1925         /* report back real offset and size */
1926         URL->offset = offset;
1927         URL->length = clength;
1928
1929         /* wrap it up in a FILE */
1930         if ((f = http_funopen(conn, chunked)) == NULL) {
1931                 fetch_syserr();
1932                 goto ouch;
1933         }
1934
1935         if (url != URL)
1936                 fetchFreeURL(url);
1937         if (purl)
1938                 fetchFreeURL(purl);
1939
1940         if (HTTP_ERROR(conn->err)) {
1941                 http_print_html(stderr, f);
1942                 fclose(f);
1943                 f = NULL;
1944         }
1945         clean_http_headerbuf(&headerbuf);
1946         clean_http_auth_challenges(&server_challenges);
1947         clean_http_auth_challenges(&proxy_challenges);
1948         return (f);
1949
1950 ouch:
1951         if (url != URL)
1952                 fetchFreeURL(url);
1953         if (purl)
1954                 fetchFreeURL(purl);
1955         if (conn != NULL)
1956                 fetch_close(conn);
1957         clean_http_headerbuf(&headerbuf);
1958         clean_http_auth_challenges(&server_challenges);
1959         clean_http_auth_challenges(&proxy_challenges);
1960         return (NULL);
1961 }
1962
1963
1964 /*****************************************************************************
1965  * Entry points
1966  */
1967
1968 /*
1969  * Retrieve and stat a file by HTTP
1970  */
1971 FILE *
1972 fetchXGetHTTP(struct url *URL, struct url_stat *us, const char *flags)
1973 {
1974         return (http_request(URL, "GET", us, http_get_proxy(URL, flags), flags));
1975 }
1976
1977 /*
1978  * Retrieve a file by HTTP
1979  */
1980 FILE *
1981 fetchGetHTTP(struct url *URL, const char *flags)
1982 {
1983         return (fetchXGetHTTP(URL, NULL, flags));
1984 }
1985
1986 /*
1987  * Store a file by HTTP
1988  */
1989 FILE *
1990 fetchPutHTTP(struct url *URL __unused, const char *flags __unused)
1991 {
1992         warnx("fetchPutHTTP(): not implemented");
1993         return (NULL);
1994 }
1995
1996 /*
1997  * Get an HTTP document's metadata
1998  */
1999 int
2000 fetchStatHTTP(struct url *URL, struct url_stat *us, const char *flags)
2001 {
2002         FILE *f;
2003
2004         f = http_request(URL, "HEAD", us, http_get_proxy(URL, flags), flags);
2005         if (f == NULL)
2006                 return (-1);
2007         fclose(f);
2008         return (0);
2009 }
2010
2011 /*
2012  * List a directory
2013  */
2014 struct url_ent *
2015 fetchListHTTP(struct url *url __unused, const char *flags __unused)
2016 {
2017         warnx("fetchListHTTP(): not implemented");
2018         return (NULL);
2019 }