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 5
  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_TEMP_REDIRECT      307
  99 #define HTTP_NEED_AUTH          401
 100 #define HTTP_NEED_PROXY_AUTH    407
 101 #define HTTP_BAD_RANGE          416
 102 #define HTTP_PROTOCOL_ERROR     999
 103
 104 #define HTTP_REDIRECT(xyz) ((xyz) == HTTP_MOVED_PERM \
 105                             || (xyz) == HTTP_MOVED_TEMP \
 106                             || (xyz) == HTTP_TEMP_REDIRECT \
 107                             || (xyz) == HTTP_SEE_OTHER)
 108
 109 #define HTTP_ERROR(xyz) ((xyz) > 400 && (xyz) < 599)
 110
 111
 112 /*****************************************************************************
 113  * I/O functions for decoding chunked streams
 114  */
 115
 116 struct httpio
 117 {
 118         conn_t          *conn;          /* connection */
 119         int              chunked;       /* chunked mode */
 120         char            *buf;           /* chunk buffer */
 121         size_t           bufsize;       /* size of chunk buffer */
 122         ssize_t          buflen;        /* amount of data currently in buffer */
 123         int              bufpos;        /* current read offset in buffer */
 124         int              eof;           /* end-of-file flag */
 125         int              error;         /* error flag */
 126         size_t           chunksize;     /* remaining size of current chunk */
 127 #ifndef NDEBUG
 128         size_t           total;
 129 #endif
 130 };
 131
 132 /*
 133  * Get next chunk header
 134  */
 135 static int
 136 http_new_chunk(struct httpio *io)
 137 {
 138         char *p;
 139
 140         if (fetch_getln(io->conn) == -1)
 141                 return (-1);
 142
 143         if (io->conn->buflen < 2 || !isxdigit((unsigned char)*io->conn->buf))
 144                 return (-1);
 145
 146         for (p = io->conn->buf; *p && !isspace((unsigned char)*p); ++p) {
 147                 if (*p == ';')
 148                         break;
 149                 if (!isxdigit((unsigned char)*p))
 150                         return (-1);
 151                 if (isdigit((unsigned char)*p)) {
 152                         io->chunksize = io->chunksize * 16 +
 153                             *p - '0';
 154                 } else {
 155                         io->chunksize = io->chunksize * 16 +
 156                             10 + tolower((unsigned char)*p) - 'a';
 157                 }
 158         }
 159
 160 #ifndef NDEBUG
 161         if (fetchDebug) {
 162                 io->total += io->chunksize;
 163                 if (io->chunksize == 0)
 164                         fprintf(stderr, "%s(): end of last chunk\n", __func__);
 165                 else
 166                         fprintf(stderr, "%s(): new chunk: %lu (%lu)\n",
 167                             __func__, (unsigned long)io->chunksize,
 168                             (unsigned long)io->total);
 169         }
 170 #endif
 171
 172         return (io->chunksize);
 173 }
 174
 175 /*
 176  * Grow the input buffer to at least len bytes
 177  */
 178 static inline int
 179 http_growbuf(struct httpio *io, size_t len)
 180 {
 181         char *tmp;
 182
 183         if (io->bufsize >= len)
 184                 return (0);
 185
 186         if ((tmp = realloc(io->buf, len)) == NULL)
 187                 return (-1);
 188         io->buf = tmp;
 189         io->bufsize = len;
 190         return (0);
 191 }
 192
 193 /*
 194  * Fill the input buffer, do chunk decoding on the fly
 195  */
 196 static int
 197 http_fillbuf(struct httpio *io, size_t len)
 198 {
 199         ssize_t nbytes;
 200
 201         if (io->error)
 202                 return (-1);
 203         if (io->eof)
 204                 return (0);
 205
 206         if (io->chunked == 0) {
 207                 if (http_growbuf(io, len) == -1)
 208                         return (-1);
 209                 if ((nbytes = fetch_read(io->conn, io->buf, len)) == -1) {
 210                         io->error = errno;
 211                         return (-1);
 212                 }
 213                 io->buflen = nbytes;
 214                 io->bufpos = 0;
 215                 return (io->buflen);
 216         }
 217
 218         if (io->chunksize == 0) {
 219                 switch (http_new_chunk(io)) {
 220                 case -1:
 221                         io->error = 1;
 222                         return (-1);
 223                 case 0:
 224                         io->eof = 1;
 225                         return (0);
 226                 }
 227         }
 228
 229         if (len > io->chunksize)
 230                 len = io->chunksize;
 231         if (http_growbuf(io, len) == -1)
 232                 return (-1);
 233         if ((nbytes = fetch_read(io->conn, io->buf, len)) == -1) {
 234                 io->error = errno;
 235                 return (-1);
 236         }
 237         io->buflen = nbytes;
 238         io->chunksize -= io->buflen;
 239
 240         if (io->chunksize == 0) {
 241                 char endl[2];
 242
 243                 if (fetch_read(io->conn, endl, 2) != 2 ||
 244                     endl[0] != '\r' || endl[1] != '\n')
 245                         return (-1);
 246         }
 247
 248         io->bufpos = 0;
 249
 250         return (io->buflen);
 251 }
 252
 253 /*
 254  * Read function
 255  */
 256 static int
 257 http_readfn(void *v, char *buf, int len)
 258 {
 259         struct httpio *io = (struct httpio *)v;
 260         int l, pos;
 261
 262         if (io->error)
 263                 return (-1);
 264         if (io->eof)
 265                 return (0);
 266
 267         for (pos = 0; len > 0; pos += l, len -= l) {
 268                 /* empty buffer */
 269                 if (!io->buf || io->bufpos == io->buflen)
 270                         if (http_fillbuf(io, len) < 1)
 271                                 break;
 272                 l = io->buflen - io->bufpos;
 273                 if (len < l)
 274                         l = len;
 275                 memcpy(buf + pos, io->buf + io->bufpos, l);
 276                 io->bufpos += l;
 277         }
 278
 279         if (!pos && io->error) {
 280                 if (io->error == EINTR)
 281                         io->error = 0;
 282                 return (-1);
 283         }
 284         return (pos);
 285 }
 286
 287 /*
 288  * Write function
 289  */
 290 static int
 291 http_writefn(void *v, const char *buf, int len)
 292 {
 293         struct httpio *io = (struct httpio *)v;
 294
 295         return (fetch_write(io->conn, buf, len));
 296 }
 297
 298 /*
 299  * Close function
 300  */
 301 static int
 302 http_closefn(void *v)
 303 {
 304         struct httpio *io = (struct httpio *)v;
 305         int r;
 306
 307         r = fetch_close(io->conn);
 308         if (io->buf)
 309                 free(io->buf);
 310         free(io);
 311         return (r);
 312 }
 313
 314 /*
 315  * Wrap a file descriptor up
 316  */
 317 static FILE *
 318 http_funopen(conn_t *conn, int chunked)
 319 {
 320         struct httpio *io;
 321         FILE *f;
 322
 323         if ((io = calloc(1, sizeof(*io))) == NULL) {
 324                 fetch_syserr();
 325                 return (NULL);
 326         }
 327         io->conn = conn;
 328         io->chunked = chunked;
 329         f = funopen(io, http_readfn, http_writefn, NULL, http_closefn);
 330         if (f == NULL) {
 331                 fetch_syserr();
 332                 free(io);
 333                 return (NULL);
 334         }
 335         return (f);
 336 }
 337
 338
 339 /*****************************************************************************
 340  * Helper functions for talking to the server and parsing its replies
 341  */
 342
 343 /* Header types */
 344 typedef enum {
 345         hdr_syserror = -2,
 346         hdr_error = -1,
 347         hdr_end = 0,
 348         hdr_unknown = 1,
 349         hdr_content_length,
 350         hdr_content_range,
 351         hdr_last_modified,
 352         hdr_location,
 353         hdr_transfer_encoding,
 354         hdr_www_authenticate,
 355         hdr_proxy_authenticate,
 356 } hdr_t;
 357
 358 /* Names of interesting headers */
 359 static struct {
 360         hdr_t            num;
 361         const char      *name;
 362 } hdr_names[] = {
 363         { hdr_content_length,           "Content-Length" },
 364         { hdr_content_range,            "Content-Range" },
 365         { hdr_last_modified,            "Last-Modified" },
 366         { hdr_location,                 "Location" },
 367         { hdr_transfer_encoding,        "Transfer-Encoding" },
 368         { hdr_www_authenticate,         "WWW-Authenticate" },
 369         { hdr_proxy_authenticate,       "Proxy-Authenticate" },
 370         { hdr_unknown,                  NULL },
 371 };
 372
 373 /*
 374  * Send a formatted line; optionally echo to terminal
 375  */
 376 static int
 377 http_cmd(conn_t *conn, const char *fmt, ...)
 378 {
 379         va_list ap;
 380         size_t len;
 381         char *msg;
 382         int r;
 383
 384         va_start(ap, fmt);
 385         len = vasprintf(&msg, fmt, ap);
 386         va_end(ap);
 387
 388         if (msg == NULL) {
 389                 errno = ENOMEM;
 390                 fetch_syserr();
 391                 return (-1);
 392         }
 393
 394         r = fetch_putln(conn, msg, len);
 395         free(msg);
 396
 397         if (r == -1) {
 398                 fetch_syserr();
 399                 return (-1);
 400         }
 401
 402         return (0);
 403 }
 404
 405 /*
 406  * Get and parse status line
 407  */
 408 static int
 409 http_get_reply(conn_t *conn)
 410 {
 411         char *p;
 412
 413         if (fetch_getln(conn) == -1)
 414                 return (-1);
 415         /*
 416          * A valid status line looks like "HTTP/m.n xyz reason" where m
 417          * and n are the major and minor protocol version numbers and xyz
 418          * is the reply code.
 419          * Unfortunately, there are servers out there (NCSA 1.5.1, to name
 420          * just one) that do not send a version number, so we can't rely
 421          * on finding one, but if we do, insist on it being 1.0 or 1.1.
 422          * We don't care about the reason phrase.
 423          */
 424         if (strncmp(conn->buf, "HTTP", 4) != 0)
 425                 return (HTTP_PROTOCOL_ERROR);
 426         p = conn->buf + 4;
 427         if (*p == '/') {
 428                 if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1'))
 429                         return (HTTP_PROTOCOL_ERROR);
 430                 p += 4;
 431         }
 432         if (*p != ' ' ||
 433             !isdigit((unsigned char)p[1]) ||
 434             !isdigit((unsigned char)p[2]) ||
 435             !isdigit((unsigned char)p[3]))
 436                 return (HTTP_PROTOCOL_ERROR);
 437
 438         conn->err = (p[1] - '0') * 100 + (p[2] - '0') * 10 + (p[3] - '0');
 439         return (conn->err);
 440 }
 441
 442 /*
 443  * Check a header; if the type matches the given string, return a pointer
 444  * to the beginning of the value.
 445  */
 446 static const char *
 447 http_match(const char *str, const char *hdr)
 448 {
 449         while (*str && *hdr &&
 450             tolower((unsigned char)*str++) == tolower((unsigned char)*hdr++))
 451                 /* nothing */;
 452         if (*str || *hdr != ':')
 453                 return (NULL);
 454         while (*hdr && isspace((unsigned char)*++hdr))
 455                 /* nothing */;
 456         return (hdr);
 457 }
 458
 459
 460 /*
 461  * Get the next header and return the appropriate symbolic code.  We
 462  * need to read one line ahead for checking for a continuation line
 463  * belonging to the current header (continuation lines start with
 464  * white space).
 465  *
 466  * We get called with a fresh line already in the conn buffer, either
 467  * from the previous http_next_header() invocation, or, the first
 468  * time, from a fetch_getln() performed by our caller.
 469  *
 470  * This stops when we encounter an empty line (we dont read beyond the header
 471  * area).
 472  *
 473  * Note that the "headerbuf" is just a place to return the result. Its
 474  * contents are not used for the next call. This means that no cleanup
 475  * is needed when ie doing another connection, just call the cleanup when
 476  * fully done to deallocate memory.
 477  */
 478
 479 /* Limit the max number of continuation lines to some reasonable value */
 480 #define HTTP_MAX_CONT_LINES 10
 481
 482 /* Place into which to build a header from one or several lines */
 483 typedef struct {
 484         char    *buf;           /* buffer */
 485         size_t   bufsize;       /* buffer size */
 486         size_t   buflen;        /* length of buffer contents */
 487 } http_headerbuf_t;
 488
 489 static void
 490 init_http_headerbuf(http_headerbuf_t *buf)
 491 {
 492         buf->buf = NULL;
 493         buf->bufsize = 0;
 494         buf->buflen = 0;
 495 }
 496
 497 static void
 498 clean_http_headerbuf(http_headerbuf_t *buf)
 499 {
 500         if (buf->buf)
 501                 free(buf->buf);
 502         init_http_headerbuf(buf);
 503 }
 504
 505 /* Remove whitespace at the end of the buffer */
 506 static void
 507 http_conn_trimright(conn_t *conn)
 508 {
 509         while (conn->buflen &&
 510                isspace((unsigned char)conn->buf[conn->buflen - 1]))
 511                 conn->buflen--;
 512         conn->buf[conn->buflen] = '\0';
 513 }
 514
 515 static hdr_t
 516 http_next_header(conn_t *conn, http_headerbuf_t *hbuf, const char **p)
 517 {
 518         unsigned int i, len;
 519
 520         /*
 521          * Have to do the stripping here because of the first line. So
 522          * it's done twice for the subsequent lines. No big deal
 523          */
 524         http_conn_trimright(conn);
 525         if (conn->buflen == 0)
 526                 return (hdr_end);
 527
 528         /* Copy the line to the headerbuf */
 529         if (hbuf->bufsize < conn->buflen + 1) {
 530                 if ((hbuf->buf = realloc(hbuf->buf, conn->buflen + 1)) == NULL)
 531                         return (hdr_syserror);
 532                 hbuf->bufsize = conn->buflen + 1;
 533         }
 534         strcpy(hbuf->buf, conn->buf);
 535         hbuf->buflen = conn->buflen;
 536
 537         /*
 538          * Fetch possible continuation lines. Stop at 1st non-continuation
 539          * and leave it in the conn buffer
 540          */
 541         for (i = 0; i < HTTP_MAX_CONT_LINES; i++) {
 542                 if (fetch_getln(conn) == -1)
 543                         return (hdr_syserror);
 544
 545                 /*
 546                  * Note: we carry on the idea from the previous version
 547                  * that a pure whitespace line is equivalent to an empty
 548                  * one (so it's not continuation and will be handled when
 549                  * we are called next)
 550                  */
 551                 http_conn_trimright(conn);
 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")) {
 770                         cs->challenges[cs->count]->scheme = HTTPAS_BASIC;
 771                 } else if (!strcasecmp(key, "digest")) {
 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"))
 801                                 cs->challenges[cs->count]->realm =
 802                                         strdup(value);
 803                         else if (!strcasecmp(key, "qop"))
 804                                 cs->challenges[cs->count]->qop =
 805                                         strdup(value);
 806                         else if (!strcasecmp(key, "nonce"))
 807                                 cs->challenges[cs->count]->nonce =
 808                                         strdup(value);
 809                         else if (!strcasecmp(key, "opaque"))
 810                                 cs->challenges[cs->count]->opaque =
 811                                         strdup(value);
 812                         else if (!strcasecmp(key, "algorithm"))
 813                                 cs->challenges[cs->count]->algo =
 814                                         strdup(value);
 815                         else if (!strcasecmp(key, "stale"))
 816                                 cs->challenges[cs->count]->stale =
 817                                         strcasecmp(value, "no");
 818                         /* Else ignore unknown attributes */
 819
 820                         /* Comma or Next challenge or End */
 821                         lex = http_header_lex(&cp, key);
 822                         /*
 823                          * If we get a word here, this is the beginning of the
 824                          * next challenge. Break the attributes loop
 825                          */
 826                         if (lex == HTTPHL_WORD)
 827                                 break;
 828
 829                         if (lex == HTTPHL_END) {
 830                                 /* End while looking for ',' is normal exit */
 831                                 cs->count++;
 832                                 ret = 0;
 833                                 goto out;
 834                         }
 835                         /* Anything else is an error */
 836                         if (lex != ',')
 837                                 goto out;
 838
 839                 } /* End attributes loop */
 840         } /* End challenge loop */
 841
 842         /*
 843          * Challenges max count exceeded. This really can't happen
 844          * with normal data, something's fishy -> error
 845          */
 846
 847 out:
 848         if (key)
 849                 free(key);
 850         if (value)
 851                 free(value);
 852         if (buf)
 853                 free(buf);
 854         return (ret);
 855 }
 856
 857
 858 /*
 859  * Parse a last-modified header
 860  */
 861 static int
 862 http_parse_mtime(const char *p, time_t *mtime)
 863 {
 864         char locale[64], *r;
 865         struct tm tm;
 866
 867         strncpy(locale, setlocale(LC_TIME, NULL), sizeof(locale));
 868         setlocale(LC_TIME, "C");
 869         r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm);
 870         /* XXX should add support for date-2 and date-3 */
 871         setlocale(LC_TIME, locale);
 872         if (r == NULL)
 873                 return (-1);
 874         DEBUG(fprintf(stderr, "last modified: [%04d-%02d-%02d "
 875                   "%02d:%02d:%02d]\n",
 876                   tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday,
 877                   tm.tm_hour, tm.tm_min, tm.tm_sec));
 878         *mtime = timegm(&tm);
 879         return (0);
 880 }
 881
 882 /*
 883  * Parse a content-length header
 884  */
 885 static int
 886 http_parse_length(const char *p, off_t *length)
 887 {
 888         off_t len;
 889
 890         for (len = 0; *p && isdigit((unsigned char)*p); ++p)
 891                 len = len * 10 + (*p - '0');
 892         if (*p)
 893                 return (-1);
 894         DEBUG(fprintf(stderr, "content length: [%lld]\n",
 895             (long long)len));
 896         *length = len;
 897         return (0);
 898 }
 899
 900 /*
 901  * Parse a content-range header
 902  */
 903 static int
 904 http_parse_range(const char *p, off_t *offset, off_t *length, off_t *size)
 905 {
 906         off_t first, last, len;
 907
 908         if (strncasecmp(p, "bytes ", 6) != 0)
 909                 return (-1);
 910         p += 6;
 911         if (*p == '*') {
 912                 first = last = -1;
 913                 ++p;
 914         } else {
 915                 for (first = 0; *p && isdigit((unsigned char)*p); ++p)
 916                         first = first * 10 + *p - '0';
 917                 if (*p != '-')
 918                         return (-1);
 919                 for (last = 0, ++p; *p && isdigit((unsigned char)*p); ++p)
 920                         last = last * 10 + *p - '0';
 921         }
 922         if (first > last || *p != '/')
 923                 return (-1);
 924         for (len = 0, ++p; *p && isdigit((unsigned char)*p); ++p)
 925                 len = len * 10 + *p - '0';
 926         if (*p || len < last - first + 1)
 927                 return (-1);
 928         if (first == -1) {
 929                 DEBUG(fprintf(stderr, "content range: [*/%lld]\n",
 930                     (long long)len));
 931                 *length = 0;
 932         } else {
 933                 DEBUG(fprintf(stderr, "content range: [%lld-%lld/%lld]\n",
 934                     (long long)first, (long long)last, (long long)len));
 935                 *length = last - first + 1;
 936         }
 937         *offset = first;
 938         *size = len;
 939         return (0);
 940 }
 941
 942
 943 /*****************************************************************************
 944  * Helper functions for authorization
 945  */
 946
 947 /*
 948  * Base64 encoding
 949  */
 950 static char *
 951 http_base64(const char *src)
 952 {
 953         static const char base64[] =
 954             "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
 955             "abcdefghijklmnopqrstuvwxyz"
 956             "0123456789+/";
 957         char *str, *dst;
 958         size_t l;
 959         int t, r;
 960
 961         l = strlen(src);
 962         if ((str = malloc(((l + 2) / 3) * 4 + 1)) == NULL)
 963                 return (NULL);
 964         dst = str;
 965         r = 0;
 966
 967         while (l >= 3) {
 968                 t = (src[0] << 16) | (src[1] << 8) | src[2];
 969                 dst[0] = base64[(t >> 18) & 0x3f];
 970                 dst[1] = base64[(t >> 12) & 0x3f];
 971                 dst[2] = base64[(t >> 6) & 0x3f];
 972                 dst[3] = base64[(t >> 0) & 0x3f];
 973                 src += 3; l -= 3;
 974                 dst += 4; r += 4;
 975         }
 976
 977         switch (l) {
 978         case 2:
 979                 t = (src[0] << 16) | (src[1] << 8);
 980                 dst[0] = base64[(t >> 18) & 0x3f];
 981                 dst[1] = base64[(t >> 12) & 0x3f];
 982                 dst[2] = base64[(t >> 6) & 0x3f];
 983                 dst[3] = '=';
 984                 dst += 4;
 985                 r += 4;
 986                 break;
 987         case 1:
 988                 t = src[0] << 16;
 989                 dst[0] = base64[(t >> 18) & 0x3f];
 990                 dst[1] = base64[(t >> 12) & 0x3f];
 991                 dst[2] = dst[3] = '=';
 992                 dst += 4;
 993                 r += 4;
 994                 break;
 995         case 0:
 996                 break;
 997         }
 998
 999         *dst = 0;
1000         return (str);
1001 }
1002
1003
1004 /*
1005  * Extract authorization parameters from environment value.
1006  * The value is like scheme:realm:user:pass
1007  */
1008 typedef struct {
1009         char    *scheme;
1010         char    *realm;
1011         char    *user;
1012         char    *password;
1013 } http_auth_params_t;
1014
1015 static void
1016 init_http_auth_params(http_auth_params_t *s)
1017 {
1018         s->scheme = s->realm = s->user = s->password = 0;
1019 }
1020
1021 static void
1022 clean_http_auth_params(http_auth_params_t *s)
1023 {
1024         if (s->scheme)
1025                 free(s->scheme);
1026         if (s->realm)
1027                 free(s->realm);
1028         if (s->user)
1029                 free(s->user);
1030         if (s->password)
1031                 free(s->password);
1032         init_http_auth_params(s);
1033 }
1034
1035 static int
1036 http_authfromenv(const char *p, http_auth_params_t *parms)
1037 {
1038         int ret = -1;
1039         char *v, *ve;
1040         char *str = strdup(p);
1041
1042         if (str == NULL) {
1043                 fetch_syserr();
1044                 return (-1);
1045         }
1046         v = str;
1047
1048         if ((ve = strchr(v, ':')) == NULL)
1049                 goto out;
1050
1051         *ve = 0;
1052         if ((parms->scheme = strdup(v)) == NULL) {
1053                 fetch_syserr();
1054                 goto out;
1055         }
1056         v = ve + 1;
1057
1058         if ((ve = strchr(v, ':')) == NULL)
1059                 goto out;
1060
1061         *ve = 0;
1062         if ((parms->realm = strdup(v)) == NULL) {
1063                 fetch_syserr();
1064                 goto out;
1065         }
1066         v = ve + 1;
1067
1068         if ((ve = strchr(v, ':')) == NULL)
1069                 goto out;
1070
1071         *ve = 0;
1072         if ((parms->user = strdup(v)) == NULL) {
1073                 fetch_syserr();
1074                 goto out;
1075         }
1076         v = ve + 1;
1077
1078
1079         if ((parms->password = strdup(v)) == NULL) {
1080                 fetch_syserr();
1081                 goto out;
1082         }
1083         ret = 0;
1084 out:
1085         if (ret == -1)
1086                 clean_http_auth_params(parms);
1087         if (str)
1088                 free(str);
1089         return (ret);
1090 }
1091
1092
1093 /*
1094  * Digest response: the code to compute the digest is taken from the
1095  * sample implementation in RFC2616
1096  */
1097 #define IN const
1098 #define OUT
1099
1100 #define HASHLEN 16
1101 typedef char HASH[HASHLEN];
1102 #define HASHHEXLEN 32
1103 typedef char HASHHEX[HASHHEXLEN+1];
1104
1105 static const char *hexchars = "0123456789abcdef";
1106 static void
1107 CvtHex(IN HASH Bin, OUT HASHHEX Hex)
1108 {
1109         unsigned short i;
1110         unsigned char j;
1111
1112         for (i = 0; i < HASHLEN; i++) {
1113                 j = (Bin[i] >> 4) & 0xf;
1114                 Hex[i*2] = hexchars[j];
1115                 j = Bin[i] & 0xf;
1116                 Hex[i*2+1] = hexchars[j];
1117         };
1118         Hex[HASHHEXLEN] = '\0';
1119 };
1120
1121 /* calculate H(A1) as per spec */
1122 static void
1123 DigestCalcHA1(
1124         IN char * pszAlg,
1125         IN char * pszUserName,
1126         IN char * pszRealm,
1127         IN char * pszPassword,
1128         IN char * pszNonce,
1129         IN char * pszCNonce,
1130         OUT HASHHEX SessionKey
1131         )
1132 {
1133         MD5_CTX Md5Ctx;
1134         HASH HA1;
1135
1136         MD5Init(&Md5Ctx);
1137         MD5Update(&Md5Ctx, pszUserName, strlen(pszUserName));
1138         MD5Update(&Md5Ctx, ":", 1);
1139         MD5Update(&Md5Ctx, pszRealm, strlen(pszRealm));
1140         MD5Update(&Md5Ctx, ":", 1);
1141         MD5Update(&Md5Ctx, pszPassword, strlen(pszPassword));
1142         MD5Final(HA1, &Md5Ctx);
1143         if (strcasecmp(pszAlg, "md5-sess") == 0) {
1144
1145                 MD5Init(&Md5Ctx);
1146                 MD5Update(&Md5Ctx, HA1, HASHLEN);
1147                 MD5Update(&Md5Ctx, ":", 1);
1148                 MD5Update(&Md5Ctx, pszNonce, strlen(pszNonce));
1149                 MD5Update(&Md5Ctx, ":", 1);
1150                 MD5Update(&Md5Ctx, pszCNonce, strlen(pszCNonce));
1151                 MD5Final(HA1, &Md5Ctx);
1152         };
1153         CvtHex(HA1, SessionKey);
1154 }
1155
1156 /* calculate request-digest/response-digest as per HTTP Digest spec */
1157 static void
1158 DigestCalcResponse(
1159         IN HASHHEX HA1,           /* H(A1) */
1160         IN char * pszNonce,       /* nonce from server */
1161         IN char * pszNonceCount,  /* 8 hex digits */
1162         IN char * pszCNonce,      /* client nonce */
1163         IN char * pszQop,         /* qop-value: "", "auth", "auth-int" */
1164         IN char * pszMethod,      /* method from the request */
1165         IN char * pszDigestUri,   /* requested URL */
1166         IN HASHHEX HEntity,       /* H(entity body) if qop="auth-int" */
1167         OUT HASHHEX Response      /* request-digest or response-digest */
1168         )
1169 {
1170 /*      DEBUG(fprintf(stderr,
1171                       "Calc: HA1[%s] Nonce[%s] qop[%s] method[%s] URI[%s]\n",
1172                       HA1, pszNonce, pszQop, pszMethod, pszDigestUri));*/
1173         MD5_CTX Md5Ctx;
1174         HASH HA2;
1175         HASH RespHash;
1176         HASHHEX HA2Hex;
1177
1178         // calculate H(A2)
1179         MD5Init(&Md5Ctx);
1180         MD5Update(&Md5Ctx, pszMethod, strlen(pszMethod));
1181         MD5Update(&Md5Ctx, ":", 1);
1182         MD5Update(&Md5Ctx, pszDigestUri, strlen(pszDigestUri));
1183         if (strcasecmp(pszQop, "auth-int") == 0) {
1184                 MD5Update(&Md5Ctx, ":", 1);
1185                 MD5Update(&Md5Ctx, HEntity, HASHHEXLEN);
1186         };
1187         MD5Final(HA2, &Md5Ctx);
1188         CvtHex(HA2, HA2Hex);
1189
1190         // calculate response
1191         MD5Init(&Md5Ctx);
1192         MD5Update(&Md5Ctx, HA1, HASHHEXLEN);
1193         MD5Update(&Md5Ctx, ":", 1);
1194         MD5Update(&Md5Ctx, pszNonce, strlen(pszNonce));
1195         MD5Update(&Md5Ctx, ":", 1);
1196         if (*pszQop) {
1197                 MD5Update(&Md5Ctx, pszNonceCount, strlen(pszNonceCount));
1198                 MD5Update(&Md5Ctx, ":", 1);
1199                 MD5Update(&Md5Ctx, pszCNonce, strlen(pszCNonce));
1200                 MD5Update(&Md5Ctx, ":", 1);
1201                 MD5Update(&Md5Ctx, pszQop, strlen(pszQop));
1202                 MD5Update(&Md5Ctx, ":", 1);
1203         };
1204         MD5Update(&Md5Ctx, HA2Hex, HASHHEXLEN);
1205         MD5Final(RespHash, &Md5Ctx);
1206         CvtHex(RespHash, Response);
1207 }
1208
1209 /*
1210  * Generate/Send a Digest authorization header
1211  * This looks like: [Proxy-]Authorization: credentials
1212  *
1213  *  credentials      = "Digest" digest-response
1214  *  digest-response  = 1#( username | realm | nonce | digest-uri
1215  *                      | response | [ algorithm ] | [cnonce] |
1216  *                      [opaque] | [message-qop] |
1217  *                          [nonce-count]  | [auth-param] )
1218  *  username         = "username" "=" username-value
1219  *  username-value   = quoted-string
1220  *  digest-uri       = "uri" "=" digest-uri-value
1221  *  digest-uri-value = request-uri   ; As specified by HTTP/1.1
1222  *  message-qop      = "qop" "=" qop-value
1223  *  cnonce           = "cnonce" "=" cnonce-value
1224  *  cnonce-value     = nonce-value
1225  *  nonce-count      = "nc" "=" nc-value
1226  *  nc-value         = 8LHEX
1227  *  response         = "response" "=" request-digest
1228  *  request-digest = <"> 32LHEX <">
1229  */
1230 static int
1231 http_digest_auth(conn_t *conn, const char *hdr, http_auth_challenge_t *c,
1232                  http_auth_params_t *parms, struct url *url)
1233 {
1234         int r;
1235         char noncecount[10];
1236         char cnonce[40];
1237         char *options = 0;
1238
1239         if (!c->realm || !c->nonce) {
1240                 DEBUG(fprintf(stderr, "realm/nonce not set in challenge\n"));
1241                 return(-1);
1242         }
1243         if (!c->algo)
1244                 c->algo = strdup("");
1245
1246         if (asprintf(&options, "%s%s%s%s",
1247                      *c->algo? ",algorithm=" : "", c->algo,
1248                      c->opaque? ",opaque=" : "", c->opaque?c->opaque:"")== -1)
1249                 return (-1);
1250
1251         if (!c->qop) {
1252                 c->qop = strdup("");
1253                 *noncecount = 0;
1254                 *cnonce = 0;
1255         } else {
1256                 c->nc++;
1257                 sprintf(noncecount, "%08x", c->nc);
1258                 /* We don't try very hard with the cnonce ... */
1259                 sprintf(cnonce, "%x%lx", getpid(), (unsigned long)time(0));
1260         }
1261
1262         HASHHEX HA1;
1263         DigestCalcHA1(c->algo, parms->user, c->realm,
1264                       parms->password, c->nonce, cnonce, HA1);
1265         DEBUG(fprintf(stderr, "HA1: [%s]\n", HA1));
1266         HASHHEX digest;
1267         DigestCalcResponse(HA1, c->nonce, noncecount, cnonce, c->qop,
1268                            "GET", url->doc, "", digest);
1269
1270         if (c->qop[0]) {
1271                 r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\","
1272                              "nonce=\"%s\",uri=\"%s\",response=\"%s\","
1273                              "qop=\"auth\", cnonce=\"%s\", nc=%s%s",
1274                              hdr, parms->user, c->realm,
1275                              c->nonce, url->doc, digest,
1276                              cnonce, noncecount, options);
1277         } else {
1278                 r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\","
1279                              "nonce=\"%s\",uri=\"%s\",response=\"%s\"%s",
1280                              hdr, parms->user, c->realm,
1281                              c->nonce, url->doc, digest, options);
1282         }
1283         if (options)
1284                 free(options);
1285         return (r);
1286 }
1287
1288 /*
1289  * Encode username and password
1290  */
1291 static int
1292 http_basic_auth(conn_t *conn, const char *hdr, const char *usr, const char *pwd)
1293 {
1294         char *upw, *auth;
1295         int r;
1296
1297         DEBUG(fprintf(stderr, "basic: usr: [%s]\n", usr));
1298         DEBUG(fprintf(stderr, "basic: pwd: [%s]\n", pwd));
1299         if (asprintf(&upw, "%s:%s", usr, pwd) == -1)
1300                 return (-1);
1301         auth = http_base64(upw);
1302         free(upw);
1303         if (auth == NULL)
1304                 return (-1);
1305         r = http_cmd(conn, "%s: Basic %s", hdr, auth);
1306         free(auth);
1307         return (r);
1308 }
1309
1310 /*
1311  * Chose the challenge to answer and call the appropriate routine to
1312  * produce the header.
1313  */
1314 static int
1315 http_authorize(conn_t *conn, const char *hdr, http_auth_challenges_t *cs,
1316                http_auth_params_t *parms, struct url *url)
1317 {
1318         http_auth_challenge_t *basic = NULL;
1319         http_auth_challenge_t *digest = NULL;
1320         int i;
1321
1322         /* If user or pass are null we're not happy */
1323         if (!parms->user || !parms->password) {
1324                 DEBUG(fprintf(stderr, "NULL usr or pass\n"));
1325                 return (-1);
1326         }
1327
1328         /* Look for a Digest and a Basic challenge */
1329         for (i = 0; i < cs->count; i++) {
1330                 if (cs->challenges[i]->scheme == HTTPAS_BASIC)
1331                         basic = cs->challenges[i];
1332                 if (cs->challenges[i]->scheme == HTTPAS_DIGEST)
1333                         digest = cs->challenges[i];
1334         }
1335
1336         /* Error if "Digest" was specified and there is no Digest challenge */
1337         if (!digest && (parms->scheme &&
1338                         !strcasecmp(parms->scheme, "digest"))) {
1339                 DEBUG(fprintf(stderr,
1340                               "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 || (parms->scheme && !strcasecmp(parms->scheme,"basic")))
1349                 return (http_basic_auth(conn,hdr,parms->user,parms->password));
1350
1351         /* Else, prefer digest. We just checked that it's not NULL */
1352         return (http_digest_auth(conn, hdr, digest, parms, url));
1353 }
1354
1355 /*****************************************************************************
1356  * Helper functions for connecting to a server or proxy
1357  */
1358
1359 /*
1360  * Connect to the correct HTTP server or proxy.
1361  */
1362 static conn_t *
1363 http_connect(struct url *URL, struct url *purl, const char *flags)
1364 {
1365         conn_t *conn;
1366         int verbose;
1367         int af, val;
1368
1369 #ifdef INET6
1370         af = AF_UNSPEC;
1371 #else
1372         af = AF_INET;
1373 #endif
1374
1375         verbose = CHECK_FLAG('v');
1376         if (CHECK_FLAG('4'))
1377                 af = AF_INET;
1378 #ifdef INET6
1379         else if (CHECK_FLAG('6'))
1380                 af = AF_INET6;
1381 #endif
1382
1383         if (purl && strcasecmp(URL->scheme, SCHEME_HTTPS) != 0) {
1384                 URL = purl;
1385         } else if (strcasecmp(URL->scheme, SCHEME_FTP) == 0) {
1386                 /* can't talk http to an ftp server */
1387                 /* XXX should set an error code */
1388                 return (NULL);
1389         }
1390
1391         if ((conn = fetch_connect(URL->host, URL->port, af, verbose)) == NULL)
1392                 /* fetch_connect() has already set an error code */
1393                 return (NULL);
1394         if (strcasecmp(URL->scheme, SCHEME_HTTPS) == 0 &&
1395             fetch_ssl(conn, verbose) == -1) {
1396                 fetch_close(conn);
1397                 /* grrr */
1398                 errno = EAUTH;
1399                 fetch_syserr();
1400                 return (NULL);
1401         }
1402
1403         val = 1;
1404         setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val, sizeof(val));
1405
1406         return (conn);
1407 }
1408
1409 static struct url *
1410 http_get_proxy(struct url * url, const char *flags)
1411 {
1412         struct url *purl;
1413         char *p;
1414
1415         if (flags != NULL && strchr(flags, 'd') != NULL)
1416                 return (NULL);
1417         if (fetch_no_proxy_match(url->host))
1418                 return (NULL);
1419         if (((p = getenv("HTTP_PROXY")) || (p = getenv("http_proxy"))) &&
1420             *p && (purl = fetchParseURL(p))) {
1421                 if (!*purl->scheme)
1422                         strcpy(purl->scheme, SCHEME_HTTP);
1423                 if (!purl->port)
1424                         purl->port = fetch_default_proxy_port(purl->scheme);
1425                 if (strcasecmp(purl->scheme, SCHEME_HTTP) == 0)
1426                         return (purl);
1427                 fetchFreeURL(purl);
1428         }
1429         return (NULL);
1430 }
1431
1432 static void
1433 http_print_html(FILE *out, FILE *in)
1434 {
1435         size_t len;
1436         char *line, *p, *q;
1437         int comment, tag;
1438
1439         comment = tag = 0;
1440         while ((line = fgetln(in, &len)) != NULL) {
1441                 while (len && isspace((unsigned char)line[len - 1]))
1442                         --len;
1443                 for (p = q = line; q < line + len; ++q) {
1444                         if (comment && *q == '-') {
1445                                 if (q + 2 < line + len &&
1446                                     strcmp(q, "-->") == 0) {
1447                                         tag = comment = 0;
1448                                         q += 2;
1449                                 }
1450                         } else if (tag && !comment && *q == '>') {
1451                                 p = q + 1;
1452                                 tag = 0;
1453                         } else if (!tag && *q == '<') {
1454                                 if (q > p)
1455                                         fwrite(p, q - p, 1, out);
1456                                 tag = 1;
1457                                 if (q + 3 < line + len &&
1458                                     strcmp(q, "<!--") == 0) {
1459                                         comment = 1;
1460                                         q += 3;
1461                                 }
1462                         }
1463                 }
1464                 if (!tag && q > p)
1465                         fwrite(p, q - p, 1, out);
1466                 fputc('\n', out);
1467         }
1468 }
1469
1470
1471 /*****************************************************************************
1472  * Core
1473  */
1474
1475 /*
1476  * Send a request and process the reply
1477  *
1478  * XXX This function is way too long, the do..while loop should be split
1479  * XXX off into a separate function.
1480  */
1481 FILE *
1482 http_request(struct url *URL, const char *op, struct url_stat *us,
1483         struct url *purl, const char *flags)
1484 {
1485         char timebuf[80];
1486         char hbuf[MAXHOSTNAMELEN + 7], *host;
1487         conn_t *conn;
1488         struct url *url, *new;
1489         int chunked, direct, ims, noredirect, verbose;
1490         int e, i, n, val;
1491         off_t offset, clength, length, size;
1492         time_t mtime;
1493         const char *p;
1494         FILE *f;
1495         hdr_t h;
1496         struct tm *timestruct;
1497         http_headerbuf_t headerbuf;
1498         http_auth_challenges_t server_challenges;
1499         http_auth_challenges_t proxy_challenges;
1500
1501         /* The following calls don't allocate anything */
1502         init_http_headerbuf(&headerbuf);
1503         init_http_auth_challenges(&server_challenges);
1504         init_http_auth_challenges(&proxy_challenges);
1505
1506         direct = CHECK_FLAG('d');
1507         noredirect = CHECK_FLAG('A');
1508         verbose = CHECK_FLAG('v');
1509         ims = CHECK_FLAG('i');
1510
1511         if (direct && purl) {
1512                 fetchFreeURL(purl);
1513                 purl = NULL;
1514         }
1515
1516         /* try the provided URL first */
1517         url = URL;
1518
1519         /* if the A flag is set, we only get one try */
1520         n = noredirect ? 1 : MAX_REDIRECT;
1521         i = 0;
1522
1523         e = HTTP_PROTOCOL_ERROR;
1524         do {
1525                 new = NULL;
1526                 chunked = 0;
1527                 offset = 0;
1528                 clength = -1;
1529                 length = -1;
1530                 size = -1;
1531                 mtime = 0;
1532
1533                 /* check port */
1534                 if (!url->port)
1535                         url->port = fetch_default_port(url->scheme);
1536
1537                 /* were we redirected to an FTP URL? */
1538                 if (purl == NULL && strcmp(url->scheme, SCHEME_FTP) == 0) {
1539                         if (strcmp(op, "GET") == 0)
1540                                 return (ftp_request(url, "RETR", us, purl, flags));
1541                         else if (strcmp(op, "HEAD") == 0)
1542                                 return (ftp_request(url, "STAT", us, purl, flags));
1543                 }
1544
1545                 /* connect to server or proxy */
1546                 if ((conn = http_connect(url, purl, flags)) == NULL)
1547                         goto ouch;
1548
1549                 host = url->host;
1550 #ifdef INET6
1551                 if (strchr(url->host, ':')) {
1552                         snprintf(hbuf, sizeof(hbuf), "[%s]", url->host);
1553                         host = hbuf;
1554                 }
1555 #endif
1556                 if (url->port != fetch_default_port(url->scheme)) {
1557                         if (host != hbuf) {
1558                                 strcpy(hbuf, host);
1559                                 host = hbuf;
1560                         }
1561                         snprintf(hbuf + strlen(hbuf),
1562                             sizeof(hbuf) - strlen(hbuf), ":%d", url->port);
1563                 }
1564
1565                 /* send request */
1566                 if (verbose)
1567                         fetch_info("requesting %s://%s%s",
1568                             url->scheme, host, url->doc);
1569                 if (purl) {
1570                         http_cmd(conn, "%s %s://%s%s HTTP/1.1",
1571                             op, url->scheme, host, url->doc);
1572                 } else {
1573                         http_cmd(conn, "%s %s HTTP/1.1",
1574                             op, url->doc);
1575                 }
1576
1577                 if (ims && url->ims_time) {
1578                         timestruct = gmtime((time_t *)&url->ims_time);
1579                         (void)strftime(timebuf, 80, "%a, %d %b %Y %T GMT",
1580                             timestruct);
1581                         if (verbose)
1582                                 fetch_info("If-Modified-Since: %s", timebuf);
1583                         http_cmd(conn, "If-Modified-Since: %s", timebuf);
1584                 }
1585                 /* virtual host */
1586                 http_cmd(conn, "Host: %s", host);
1587
1588                 /*
1589                  * Proxy authorization: we only send auth after we received
1590                  * a 407 error. We do not first try basic anyway (changed
1591                  * when support was added for digest-auth)
1592                  */
1593                 if (purl && proxy_challenges.valid) {
1594                         http_auth_params_t aparams;
1595                         init_http_auth_params(&aparams);
1596                         if (*purl->user || *purl->pwd) {
1597                                 aparams.user = purl->user ?
1598                                         strdup(purl->user) : strdup("");
1599                                 aparams.password = purl->pwd?
1600                                         strdup(purl->pwd) : strdup("");
1601                         } else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL &&
1602                                    *p != '\0') {
1603                                 if (http_authfromenv(p, &aparams) < 0) {
1604                                         http_seterr(HTTP_NEED_PROXY_AUTH);
1605                                         goto ouch;
1606                                 }
1607                         }
1608                         http_authorize(conn, "Proxy-Authorization",
1609                                        &proxy_challenges, &aparams, url);
1610                         clean_http_auth_params(&aparams);
1611                 }
1612
1613                 /*
1614                  * Server authorization: we never send "a priori"
1615                  * Basic auth, which used to be done if user/pass were
1616                  * set in the url. This would be weird because we'd send the
1617                  * password in the clear even if Digest is finally to be
1618                  * used (it would have made more sense for the
1619                  * pre-digest version to do this when Basic was specified
1620                  * in the environment)
1621                  */
1622                 if (server_challenges.valid) {
1623                         http_auth_params_t aparams;
1624                         init_http_auth_params(&aparams);
1625                         if (*url->user || *url->pwd) {
1626                                 aparams.user = url->user ?
1627                                         strdup(url->user) : strdup("");
1628                                 aparams.password = url->pwd ?
1629                                         strdup(url->pwd) : strdup("");
1630                         } else if ((p = getenv("HTTP_AUTH")) != NULL &&
1631                                    *p != '\0') {
1632                                 if (http_authfromenv(p, &aparams) < 0) {
1633                                         http_seterr(HTTP_NEED_AUTH);
1634                                         goto ouch;
1635                                 }
1636                         } else if (fetchAuthMethod &&
1637                                    fetchAuthMethod(url) == 0) {
1638                                 aparams.user = url->user ?
1639                                         strdup(url->user) : strdup("");
1640                                 aparams.password = url->pwd ?
1641                                         strdup(url->pwd) : strdup("");
1642                         } else {
1643                                 http_seterr(HTTP_NEED_AUTH);
1644                                 goto ouch;
1645                         }
1646                         http_authorize(conn, "Authorization",
1647                                        &server_challenges, &aparams, url);
1648                         clean_http_auth_params(&aparams);
1649                 }
1650
1651                 /* other headers */
1652                 if ((p = getenv("HTTP_REFERER")) != NULL && *p != '\0') {
1653                         if (strcasecmp(p, "auto") == 0)
1654                                 http_cmd(conn, "Referer: %s://%s%s",
1655                                     url->scheme, host, url->doc);
1656                         else
1657                                 http_cmd(conn, "Referer: %s", p);
1658                 }
1659                 if ((p = getenv("HTTP_USER_AGENT")) != NULL && *p != '\0')
1660                         http_cmd(conn, "User-Agent: %s", p);
1661                 else
1662                         http_cmd(conn, "User-Agent: %s " _LIBFETCH_VER, getprogname());
1663                 if (url->offset > 0)
1664                         http_cmd(conn, "Range: bytes=%lld-", (long long)url->offset);
1665                 http_cmd(conn, "Connection: close");
1666                 http_cmd(conn, "");
1667
1668                 /*
1669                  * Force the queued request to be dispatched.  Normally, one
1670                  * would do this with shutdown(2) but squid proxies can be
1671                  * configured to disallow such half-closed connections.  To
1672                  * be compatible with such configurations, fiddle with socket
1673                  * options to force the pending data to be written.
1674                  */
1675                 val = 0;
1676                 setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val,
1677                            sizeof(val));
1678                 val = 1;
1679                 setsockopt(conn->sd, IPPROTO_TCP, TCP_NODELAY, &val,
1680                            sizeof(val));
1681
1682                 /* get reply */
1683                 switch (http_get_reply(conn)) {
1684                 case HTTP_OK:
1685                 case HTTP_PARTIAL:
1686                 case HTTP_NOT_MODIFIED:
1687                         /* fine */
1688                         break;
1689                 case HTTP_MOVED_PERM:
1690                 case HTTP_MOVED_TEMP:
1691                 case HTTP_SEE_OTHER:
1692                         /*
1693                          * Not so fine, but we still have to read the
1694                          * headers to get the new location.
1695                          */
1696                         break;
1697                 case HTTP_NEED_AUTH:
1698                         if (server_challenges.valid) {
1699                                 /*
1700                                  * We already sent out authorization code,
1701                                  * so there's nothing more we can do.
1702                                  */
1703                                 http_seterr(conn->err);
1704                                 goto ouch;
1705                         }
1706                         /* try again, but send the password this time */
1707                         if (verbose)
1708                                 fetch_info("server requires authorization");
1709                         break;
1710                 case HTTP_NEED_PROXY_AUTH:
1711                         if (proxy_challenges.valid) {
1712                                 /*
1713                                  * We already sent our proxy
1714                                  * authorization code, so there's
1715                                  * nothing more we can do. */
1716                                 http_seterr(conn->err);
1717                                 goto ouch;
1718                         }
1719                         /* try again, but send the password this time */
1720                         if (verbose)
1721                                 fetch_info("proxy requires authorization");
1722                         break;
1723                 case HTTP_BAD_RANGE:
1724                         /*
1725                          * This can happen if we ask for 0 bytes because
1726                          * we already have the whole file.  Consider this
1727                          * a success for now, and check sizes later.
1728                          */
1729                         break;
1730                 case HTTP_PROTOCOL_ERROR:
1731                         /* fall through */
1732                 case -1:
1733                         fetch_syserr();
1734                         goto ouch;
1735                 default:
1736                         http_seterr(conn->err);
1737                         if (!verbose)
1738                                 goto ouch;
1739                         /* fall through so we can get the full error message */
1740                 }
1741
1742                 /* get headers. http_next_header expects one line readahead */
1743                 if (fetch_getln(conn) == -1) {
1744                     fetch_syserr();
1745                     goto ouch;
1746                 }
1747                 do {
1748                     switch ((h = http_next_header(conn, &headerbuf, &p))) {
1749                         case hdr_syserror:
1750                                 fetch_syserr();
1751                                 goto ouch;
1752                         case hdr_error:
1753                                 http_seterr(HTTP_PROTOCOL_ERROR);
1754                                 goto ouch;
1755                         case hdr_content_length:
1756                                 http_parse_length(p, &clength);
1757                                 break;
1758                         case hdr_content_range:
1759                                 http_parse_range(p, &offset, &length, &size);
1760                                 break;
1761                         case hdr_last_modified:
1762                                 http_parse_mtime(p, &mtime);
1763                                 break;
1764                         case hdr_location:
1765                                 if (!HTTP_REDIRECT(conn->err))
1766                                         break;
1767                                 if (new)
1768                                         free(new);
1769                                 if (verbose)
1770                                         fetch_info("%d redirect to %s", conn->err, p);
1771                                 if (*p == '/')
1772                                         /* absolute path */
1773                                         new = fetchMakeURL(url->scheme, url->host, url->port, p,
1774                                             url->user, url->pwd);
1775                                 else
1776                                         new = fetchParseURL(p);
1777                                 if (new == NULL) {
1778                                         /* XXX should set an error code */
1779                                         DEBUG(fprintf(stderr, "failed to parse new URL\n"));
1780                                         goto ouch;
1781                                 }
1782                                 if (!*new->user && !*new->pwd) {
1783                                         strcpy(new->user, url->user);
1784                                         strcpy(new->pwd, url->pwd);
1785                                 }
1786                                 new->offset = url->offset;
1787                                 new->length = url->length;
1788                                 break;
1789                         case hdr_transfer_encoding:
1790                                 /* XXX weak test*/
1791                                 chunked = (strcasecmp(p, "chunked") == 0);
1792                                 break;
1793                         case hdr_www_authenticate:
1794                                 if (conn->err != HTTP_NEED_AUTH)
1795                                         break;
1796                                 if (http_parse_authenticate(p, &server_challenges) == 0)
1797                                         ++n;
1798                                 break;
1799                         case hdr_proxy_authenticate:
1800                                 if (conn->err != HTTP_NEED_PROXY_AUTH)
1801                                         break;
1802                                 if (http_parse_authenticate(p, &proxy_challenges) == 0)
1803                                         ++n;
1804                                 break;
1805                         case hdr_end:
1806                                 /* fall through */
1807                         case hdr_unknown:
1808                                 /* ignore */
1809                                 break;
1810                         }
1811                 } while (h > hdr_end);
1812
1813                 /* we need to provide authentication */
1814                 if (conn->err == HTTP_NEED_AUTH ||
1815                     conn->err == HTTP_NEED_PROXY_AUTH) {
1816                         e = conn->err;
1817                         if ((conn->err == HTTP_NEED_AUTH &&
1818                              !server_challenges.valid) ||
1819                             (conn->err == HTTP_NEED_PROXY_AUTH &&
1820                              !proxy_challenges.valid)) {
1821                                 /* 401/7 but no www/proxy-authenticate ?? */
1822                                 DEBUG(fprintf(stderr, "401/7 and no auth header\n"));
1823                                 goto ouch;
1824                         }
1825                         fetch_close(conn);
1826                         conn = NULL;
1827                         continue;
1828                 }
1829
1830                 /* requested range not satisfiable */
1831                 if (conn->err == HTTP_BAD_RANGE) {
1832                         if (url->offset == size && url->length == 0) {
1833                                 /* asked for 0 bytes; fake it */
1834                                 offset = url->offset;
1835                                 clength = -1;
1836                                 conn->err = HTTP_OK;
1837                                 break;
1838                         } else {
1839                                 http_seterr(conn->err);
1840                                 goto ouch;
1841                         }
1842                 }
1843
1844                 /* we have a hit or an error */
1845                 if (conn->err == HTTP_OK
1846                     || conn->err == HTTP_NOT_MODIFIED
1847                     || conn->err == HTTP_PARTIAL
1848                     || HTTP_ERROR(conn->err))
1849                         break;
1850
1851                 /* all other cases: we got a redirect */
1852                 e = conn->err;
1853                 clean_http_auth_challenges(&server_challenges);
1854                 fetch_close(conn);
1855                 conn = NULL;
1856                 if (!new) {
1857                         DEBUG(fprintf(stderr, "redirect with no new location\n"));
1858                         break;
1859                 }
1860                 if (url != URL)
1861                         fetchFreeURL(url);
1862                 url = new;
1863         } while (++i < n);
1864
1865         /* we failed, or ran out of retries */
1866         if (conn == NULL) {
1867                 http_seterr(e);
1868                 goto ouch;
1869         }
1870
1871         DEBUG(fprintf(stderr, "offset %lld, length %lld,"
1872                   " size %lld, clength %lld\n",
1873                   (long long)offset, (long long)length,
1874                   (long long)size, (long long)clength));
1875
1876         if (conn->err == HTTP_NOT_MODIFIED) {
1877                 http_seterr(HTTP_NOT_MODIFIED);
1878                 return (NULL);
1879         }
1880
1881         /* check for inconsistencies */
1882         if (clength != -1 && length != -1 && clength != length) {
1883                 http_seterr(HTTP_PROTOCOL_ERROR);
1884                 goto ouch;
1885         }
1886         if (clength == -1)
1887                 clength = length;
1888         if (clength != -1)
1889                 length = offset + clength;
1890         if (length != -1 && size != -1 && length != size) {
1891                 http_seterr(HTTP_PROTOCOL_ERROR);
1892                 goto ouch;
1893         }
1894         if (size == -1)
1895                 size = length;
1896
1897         /* fill in stats */
1898         if (us) {
1899                 us->size = size;
1900                 us->atime = us->mtime = mtime;
1901         }
1902
1903         /* too far? */
1904         if (URL->offset > 0 && offset > URL->offset) {
1905                 http_seterr(HTTP_PROTOCOL_ERROR);
1906                 goto ouch;
1907         }
1908
1909         /* report back real offset and size */
1910         URL->offset = offset;
1911         URL->length = clength;
1912
1913         /* wrap it up in a FILE */
1914         if ((f = http_funopen(conn, chunked)) == NULL) {
1915                 fetch_syserr();
1916                 goto ouch;
1917         }
1918
1919         if (url != URL)
1920                 fetchFreeURL(url);
1921         if (purl)
1922                 fetchFreeURL(purl);
1923
1924         if (HTTP_ERROR(conn->err)) {
1925                 http_print_html(stderr, f);
1926                 fclose(f);
1927                 f = NULL;
1928         }
1929         clean_http_headerbuf(&headerbuf);
1930         clean_http_auth_challenges(&server_challenges);
1931         clean_http_auth_challenges(&proxy_challenges);
1932         return (f);
1933
1934 ouch:
1935         if (url != URL)
1936                 fetchFreeURL(url);
1937         if (purl)
1938                 fetchFreeURL(purl);
1939         if (conn != NULL)
1940                 fetch_close(conn);
1941         clean_http_headerbuf(&headerbuf);
1942         clean_http_auth_challenges(&server_challenges);
1943         clean_http_auth_challenges(&proxy_challenges);
1944         return (NULL);
1945 }
1946
1947
1948 /*****************************************************************************
1949  * Entry points
1950  */
1951
1952 /*
1953  * Retrieve and stat a file by HTTP
1954  */
1955 FILE *
1956 fetchXGetHTTP(struct url *URL, struct url_stat *us, const char *flags)
1957 {
1958         return (http_request(URL, "GET", us, http_get_proxy(URL, flags), flags));
1959 }
1960
1961 /*
1962  * Retrieve a file by HTTP
1963  */
1964 FILE *
1965 fetchGetHTTP(struct url *URL, const char *flags)
1966 {
1967         return (fetchXGetHTTP(URL, NULL, flags));
1968 }
1969
1970 /*
1971  * Store a file by HTTP
1972  */
1973 FILE *
1974 fetchPutHTTP(struct url *URL __unused, const char *flags __unused)
1975 {
1976         warnx("fetchPutHTTP(): not implemented");
1977         return (NULL);
1978 }
1979
1980 /*
1981  * Get an HTTP document's metadata
1982  */
1983 int
1984 fetchStatHTTP(struct url *URL, struct url_stat *us, const char *flags)
1985 {
1986         FILE *f;
1987
1988         f = http_request(URL, "HEAD", us, http_get_proxy(URL, flags), flags);
1989         if (f == NULL)
1990                 return (-1);
1991         fclose(f);
1992         return (0);
1993 }
1994
1995 /*
1996  * List a directory
1997  */
1998 struct url_ent *
1999 fetchListHTTP(struct url *url __unused, const char *flags __unused)
2000 {
2001         warnx("fetchListHTTP(): not implemented");
2002         return (NULL);
2003 }