lib/libfetch/http.c

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