14ca1ab94SDag-Erling Smørgrav /*- 2e4878e39SDag-Erling Smørgrav * Copyright (c) 2000 Dag-Erling Co�dan Sm�rgrav 34ca1ab94SDag-Erling Smørgrav * All rights reserved. 44ca1ab94SDag-Erling Smørgrav * 54ca1ab94SDag-Erling Smørgrav * Redistribution and use in source and binary forms, with or without 64ca1ab94SDag-Erling Smørgrav * modification, are permitted provided that the following conditions 74ca1ab94SDag-Erling Smørgrav * are met: 84ca1ab94SDag-Erling Smørgrav * 1. Redistributions of source code must retain the above copyright 94ca1ab94SDag-Erling Smørgrav * notice, this list of conditions and the following disclaimer 104ca1ab94SDag-Erling Smørgrav * in this position and unchanged. 114ca1ab94SDag-Erling Smørgrav * 2. Redistributions in binary form must reproduce the above copyright 124ca1ab94SDag-Erling Smørgrav * notice, this list of conditions and the following disclaimer in the 134ca1ab94SDag-Erling Smørgrav * documentation and/or other materials provided with the distribution. 144ca1ab94SDag-Erling Smørgrav * 3. The name of the author may not be used to endorse or promote products 15e4878e39SDag-Erling Smørgrav * derived from this software without specific prior written permission. 164ca1ab94SDag-Erling Smørgrav * 174ca1ab94SDag-Erling Smørgrav * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 184ca1ab94SDag-Erling Smørgrav * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 194ca1ab94SDag-Erling Smørgrav * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 204ca1ab94SDag-Erling Smørgrav * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 214ca1ab94SDag-Erling Smørgrav * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 224ca1ab94SDag-Erling Smørgrav * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 234ca1ab94SDag-Erling Smørgrav * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 244ca1ab94SDag-Erling Smørgrav * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 254ca1ab94SDag-Erling Smørgrav * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 264ca1ab94SDag-Erling Smørgrav * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 274ca1ab94SDag-Erling Smørgrav */ 284ca1ab94SDag-Erling Smørgrav 29cecb889fSMatthew Dillon #include <sys/cdefs.h> 30cecb889fSMatthew Dillon __FBSDID("$FreeBSD$"); 31cecb889fSMatthew Dillon 326290ee73SDag-Erling Smørgrav /* 336290ee73SDag-Erling Smørgrav * The following copyright applies to the base64 code: 346290ee73SDag-Erling Smørgrav * 356290ee73SDag-Erling Smørgrav *- 366290ee73SDag-Erling Smørgrav * Copyright 1997 Massachusetts Institute of Technology 376290ee73SDag-Erling Smørgrav * 386290ee73SDag-Erling Smørgrav * Permission to use, copy, modify, and distribute this software and 396290ee73SDag-Erling Smørgrav * its documentation for any purpose and without fee is hereby 406290ee73SDag-Erling Smørgrav * granted, provided that both the above copyright notice and this 416290ee73SDag-Erling Smørgrav * permission notice appear in all copies, that both the above 426290ee73SDag-Erling Smørgrav * copyright notice and this permission notice appear in all 436290ee73SDag-Erling Smørgrav * supporting documentation, and that the name of M.I.T. not be used 446290ee73SDag-Erling Smørgrav * in advertising or publicity pertaining to distribution of the 456290ee73SDag-Erling Smørgrav * software without specific, written prior permission. M.I.T. makes 466290ee73SDag-Erling Smørgrav * no representations about the suitability of this software for any 476290ee73SDag-Erling Smørgrav * purpose. It is provided "as is" without express or implied 486290ee73SDag-Erling Smørgrav * warranty. 496290ee73SDag-Erling Smørgrav * 506290ee73SDag-Erling Smørgrav * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''. M.I.T. DISCLAIMS 516290ee73SDag-Erling Smørgrav * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE, 526290ee73SDag-Erling Smørgrav * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF 536290ee73SDag-Erling Smørgrav * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT 546290ee73SDag-Erling Smørgrav * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 556290ee73SDag-Erling Smørgrav * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 566290ee73SDag-Erling Smørgrav * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF 576290ee73SDag-Erling Smørgrav * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND 586290ee73SDag-Erling Smørgrav * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 596290ee73SDag-Erling Smørgrav * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 606290ee73SDag-Erling Smørgrav * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 616290ee73SDag-Erling Smørgrav * SUCH DAMAGE. 626290ee73SDag-Erling Smørgrav */ 636290ee73SDag-Erling Smørgrav 644ca1ab94SDag-Erling Smørgrav #include <sys/param.h> 6528c645cfSHajimu UMEMOTO #include <sys/socket.h> 664ca1ab94SDag-Erling Smørgrav 674ca1ab94SDag-Erling Smørgrav #include <ctype.h> 68e4878e39SDag-Erling Smørgrav #include <err.h> 69e4878e39SDag-Erling Smørgrav #include <errno.h> 7060245e42SDag-Erling Smørgrav #include <locale.h> 71e6182307SDag-Erling Smørgrav #include <netdb.h> 72f62e5228SDag-Erling Smørgrav #include <stdarg.h> 734ca1ab94SDag-Erling Smørgrav #include <stdio.h> 744ca1ab94SDag-Erling Smørgrav #include <stdlib.h> 754ca1ab94SDag-Erling Smørgrav #include <string.h> 7660245e42SDag-Erling Smørgrav #include <time.h> 774ca1ab94SDag-Erling Smørgrav #include <unistd.h> 784ca1ab94SDag-Erling Smørgrav 794ca1ab94SDag-Erling Smørgrav #include "fetch.h" 80842a95ccSDag-Erling Smørgrav #include "common.h" 810fba3a00SDag-Erling Smørgrav #include "httperr.h" 824ca1ab94SDag-Erling Smørgrav 83e4878e39SDag-Erling Smørgrav /* Maximum number of redirects to follow */ 84e4878e39SDag-Erling Smørgrav #define MAX_REDIRECT 5 854ca1ab94SDag-Erling Smørgrav 86e4878e39SDag-Erling Smørgrav /* Symbolic names for reply codes we care about */ 873d2a8471SDag-Erling Smørgrav #define HTTP_OK 200 883d2a8471SDag-Erling Smørgrav #define HTTP_PARTIAL 206 89e4878e39SDag-Erling Smørgrav #define HTTP_MOVED_PERM 301 90e4878e39SDag-Erling Smørgrav #define HTTP_MOVED_TEMP 302 91e4878e39SDag-Erling Smørgrav #define HTTP_SEE_OTHER 303 92e4878e39SDag-Erling Smørgrav #define HTTP_NEED_AUTH 401 9365986545SDag-Erling Smørgrav #define HTTP_NEED_PROXY_AUTH 407 94e4878e39SDag-Erling Smørgrav #define HTTP_PROTOCOL_ERROR 999 95e4878e39SDag-Erling Smørgrav 96e4878e39SDag-Erling Smørgrav #define HTTP_REDIRECT(xyz) ((xyz) == HTTP_MOVED_PERM \ 97e4878e39SDag-Erling Smørgrav || (xyz) == HTTP_MOVED_TEMP \ 98e4878e39SDag-Erling Smørgrav || (xyz) == HTTP_SEE_OTHER) 99e4878e39SDag-Erling Smørgrav 100a8e9bd87SDag-Erling Smørgrav #define HTTP_ERROR(xyz) ((xyz) > 400 && (xyz) < 599) 101e4878e39SDag-Erling Smørgrav 102e19e6098SDag-Erling Smørgrav 103e4878e39SDag-Erling Smørgrav /***************************************************************************** 104e4878e39SDag-Erling Smørgrav * I/O functions for decoding chunked streams 105e4878e39SDag-Erling Smørgrav */ 1063d2a8471SDag-Erling Smørgrav 107ccdd94bdSDag-Erling Smørgrav struct httpio 1084ca1ab94SDag-Erling Smørgrav { 109d3b03a90SDag-Erling Smørgrav conn_t *conn; /* connection */ 1109601e333SDag-Erling Smørgrav int chunked; /* chunked mode */ 111d3b03a90SDag-Erling Smørgrav char *buf; /* chunk buffer */ 1129601e333SDag-Erling Smørgrav size_t bufsize; /* size of chunk buffer */ 1139601e333SDag-Erling Smørgrav ssize_t buflen; /* amount of data currently in buffer */ 1149601e333SDag-Erling Smørgrav int bufpos; /* current read offset in buffer */ 115d3b03a90SDag-Erling Smørgrav int eof; /* end-of-file flag */ 116d3b03a90SDag-Erling Smørgrav int error; /* error flag */ 117d3b03a90SDag-Erling Smørgrav size_t chunksize; /* remaining size of current chunk */ 118e66b3802SDag-Erling Smørgrav #ifndef NDEBUG 119f573a5fcSDag-Erling Smørgrav size_t total; 120e4878e39SDag-Erling Smørgrav #endif 1214ca1ab94SDag-Erling Smørgrav }; 1224ca1ab94SDag-Erling Smørgrav 123f62e5228SDag-Erling Smørgrav /* 124e4878e39SDag-Erling Smørgrav * Get next chunk header 125f62e5228SDag-Erling Smørgrav */ 126f62e5228SDag-Erling Smørgrav static int 127ccdd94bdSDag-Erling Smørgrav _http_new_chunk(struct httpio *io) 128f62e5228SDag-Erling Smørgrav { 129e4878e39SDag-Erling Smørgrav char *p; 130f62e5228SDag-Erling Smørgrav 131ccdd94bdSDag-Erling Smørgrav if (_fetch_getln(io->conn) == -1) 132e19e6098SDag-Erling Smørgrav return (-1); 133f62e5228SDag-Erling Smørgrav 134ccdd94bdSDag-Erling Smørgrav if (io->conn->buflen < 2 || !ishexnumber(*io->conn->buf)) 135e19e6098SDag-Erling Smørgrav return (-1); 136e4878e39SDag-Erling Smørgrav 137ccdd94bdSDag-Erling Smørgrav for (p = io->conn->buf; *p && !isspace(*p); ++p) { 138e19e6098SDag-Erling Smørgrav if (*p == ';') 139e19e6098SDag-Erling Smørgrav break; 140e4878e39SDag-Erling Smørgrav if (!ishexnumber(*p)) 141e19e6098SDag-Erling Smørgrav return (-1); 142e19e6098SDag-Erling Smørgrav if (isdigit(*p)) { 143ccdd94bdSDag-Erling Smørgrav io->chunksize = io->chunksize * 16 + 144e19e6098SDag-Erling Smørgrav *p - '0'; 145e19e6098SDag-Erling Smørgrav } else { 146ccdd94bdSDag-Erling Smørgrav io->chunksize = io->chunksize * 16 + 147e19e6098SDag-Erling Smørgrav 10 + tolower(*p) - 'a'; 148e19e6098SDag-Erling Smørgrav } 149e19e6098SDag-Erling Smørgrav } 150e4878e39SDag-Erling Smørgrav 151e66b3802SDag-Erling Smørgrav #ifndef NDEBUG 1527f807cb8SDag-Erling Smørgrav if (fetchDebug) { 153ccdd94bdSDag-Erling Smørgrav io->total += io->chunksize; 154ccdd94bdSDag-Erling Smørgrav if (io->chunksize == 0) 155f67efa37SDag-Erling Smørgrav fprintf(stderr, "_http_fillbuf(): " 156f67efa37SDag-Erling Smørgrav "end of last chunk\n"); 157e4878e39SDag-Erling Smørgrav else 158f67efa37SDag-Erling Smørgrav fprintf(stderr, "_http_fillbuf(): " 159f67efa37SDag-Erling Smørgrav "new chunk: %lu (%lu)\n", 160ccdd94bdSDag-Erling Smørgrav (unsigned long)io->chunksize, (unsigned long)io->total); 1617f807cb8SDag-Erling Smørgrav } 162e4878e39SDag-Erling Smørgrav #endif 163e4878e39SDag-Erling Smørgrav 164ccdd94bdSDag-Erling Smørgrav return (io->chunksize); 165f62e5228SDag-Erling Smørgrav } 166f62e5228SDag-Erling Smørgrav 167f62e5228SDag-Erling Smørgrav /* 1689601e333SDag-Erling Smørgrav * Grow the input buffer to at least len bytes 1699601e333SDag-Erling Smørgrav */ 1709601e333SDag-Erling Smørgrav static inline int 1719601e333SDag-Erling Smørgrav _http_growbuf(struct httpio *io, size_t len) 1729601e333SDag-Erling Smørgrav { 1739601e333SDag-Erling Smørgrav char *tmp; 1749601e333SDag-Erling Smørgrav 1759601e333SDag-Erling Smørgrav if (io->bufsize >= len) 1769601e333SDag-Erling Smørgrav return (0); 1779601e333SDag-Erling Smørgrav 1789601e333SDag-Erling Smørgrav if ((tmp = realloc(io->buf, len)) == NULL) 1799601e333SDag-Erling Smørgrav return (-1); 1809601e333SDag-Erling Smørgrav io->buf = tmp; 1819601e333SDag-Erling Smørgrav io->bufsize = len; 1829601e333SDag-Erling Smørgrav } 1839601e333SDag-Erling Smørgrav 1849601e333SDag-Erling Smørgrav /* 185f62e5228SDag-Erling Smørgrav * Fill the input buffer, do chunk decoding on the fly 186f62e5228SDag-Erling Smørgrav */ 187e4878e39SDag-Erling Smørgrav static int 1889601e333SDag-Erling Smørgrav _http_fillbuf(struct httpio *io, size_t len) 1894ca1ab94SDag-Erling Smørgrav { 190ccdd94bdSDag-Erling Smørgrav if (io->error) 191e19e6098SDag-Erling Smørgrav return (-1); 192ccdd94bdSDag-Erling Smørgrav if (io->eof) 193e19e6098SDag-Erling Smørgrav return (0); 1944ca1ab94SDag-Erling Smørgrav 1959601e333SDag-Erling Smørgrav if (io->chunked == 0) { 1969601e333SDag-Erling Smørgrav if (_http_growbuf(io, len) == -1) 1979601e333SDag-Erling Smørgrav return (-1); 1989601e333SDag-Erling Smørgrav if ((io->buflen = _fetch_read(io->conn, io->buf, len)) == -1) 1999601e333SDag-Erling Smørgrav return (-1); 2009601e333SDag-Erling Smørgrav io->bufpos = 0; 2019601e333SDag-Erling Smørgrav return (io->buflen); 2029601e333SDag-Erling Smørgrav } 2039601e333SDag-Erling Smørgrav 204ccdd94bdSDag-Erling Smørgrav if (io->chunksize == 0) { 205ccdd94bdSDag-Erling Smørgrav switch (_http_new_chunk(io)) { 206e4878e39SDag-Erling Smørgrav case -1: 207ccdd94bdSDag-Erling Smørgrav io->error = 1; 208e19e6098SDag-Erling Smørgrav return (-1); 209e4878e39SDag-Erling Smørgrav case 0: 210ccdd94bdSDag-Erling Smørgrav io->eof = 1; 211e19e6098SDag-Erling Smørgrav return (0); 2124ca1ab94SDag-Erling Smørgrav } 2134ca1ab94SDag-Erling Smørgrav } 214e4878e39SDag-Erling Smørgrav 2159601e333SDag-Erling Smørgrav if (len > io->chunksize) 2169601e333SDag-Erling Smørgrav len = io->chunksize; 2179601e333SDag-Erling Smørgrav if (_http_growbuf(io, len) == -1) 218e19e6098SDag-Erling Smørgrav return (-1); 2199601e333SDag-Erling Smørgrav if ((io->buflen = _fetch_read(io->conn, io->buf, len)) == -1) 220e19e6098SDag-Erling Smørgrav return (-1); 2219601e333SDag-Erling Smørgrav io->chunksize -= io->buflen; 222e4878e39SDag-Erling Smørgrav 223ccdd94bdSDag-Erling Smørgrav if (io->chunksize == 0) { 224dea29ca1SDag-Erling Smørgrav char endl[2]; 225dea29ca1SDag-Erling Smørgrav 2269601e333SDag-Erling Smørgrav if (_fetch_read(io->conn, endl, 2) != 2 || 227dea29ca1SDag-Erling Smørgrav endl[0] != '\r' || endl[1] != '\n') 228e19e6098SDag-Erling Smørgrav return (-1); 2294ca1ab94SDag-Erling Smørgrav } 230e4878e39SDag-Erling Smørgrav 2319601e333SDag-Erling Smørgrav io->bufpos = 0; 232e4878e39SDag-Erling Smørgrav 2339601e333SDag-Erling Smørgrav return (io->buflen); 2344ca1ab94SDag-Erling Smørgrav } 2354ca1ab94SDag-Erling Smørgrav 236f62e5228SDag-Erling Smørgrav /* 237f62e5228SDag-Erling Smørgrav * Read function 238f62e5228SDag-Erling Smørgrav */ 2394ca1ab94SDag-Erling Smørgrav static int 240e4878e39SDag-Erling Smørgrav _http_readfn(void *v, char *buf, int len) 2414ca1ab94SDag-Erling Smørgrav { 242ccdd94bdSDag-Erling Smørgrav struct httpio *io = (struct httpio *)v; 243e4878e39SDag-Erling Smørgrav int l, pos; 2444ca1ab94SDag-Erling Smørgrav 245ccdd94bdSDag-Erling Smørgrav if (io->error) 246e19e6098SDag-Erling Smørgrav return (-1); 247ccdd94bdSDag-Erling Smørgrav if (io->eof) 248e19e6098SDag-Erling Smørgrav return (0); 249e4878e39SDag-Erling Smørgrav 250e4878e39SDag-Erling Smørgrav for (pos = 0; len > 0; pos += l, len -= l) { 251e4878e39SDag-Erling Smørgrav /* empty buffer */ 2529601e333SDag-Erling Smørgrav if (!io->buf || io->bufpos == io->buflen) 2539601e333SDag-Erling Smørgrav if (_http_fillbuf(io, len) < 1) 254e4878e39SDag-Erling Smørgrav break; 2559601e333SDag-Erling Smørgrav l = io->buflen - io->bufpos; 256e4878e39SDag-Erling Smørgrav if (len < l) 257e4878e39SDag-Erling Smørgrav l = len; 2589601e333SDag-Erling Smørgrav bcopy(io->buf + io->bufpos, buf + pos, l); 2599601e333SDag-Erling Smørgrav io->bufpos += l; 2604ca1ab94SDag-Erling Smørgrav } 2614ca1ab94SDag-Erling Smørgrav 262ccdd94bdSDag-Erling Smørgrav if (!pos && io->error) 263e19e6098SDag-Erling Smørgrav return (-1); 264e19e6098SDag-Erling Smørgrav return (pos); 2654ca1ab94SDag-Erling Smørgrav } 2664ca1ab94SDag-Erling Smørgrav 267f62e5228SDag-Erling Smørgrav /* 268f62e5228SDag-Erling Smørgrav * Write function 269f62e5228SDag-Erling Smørgrav */ 2704ca1ab94SDag-Erling Smørgrav static int 271e4878e39SDag-Erling Smørgrav _http_writefn(void *v, const char *buf, int len) 2724ca1ab94SDag-Erling Smørgrav { 273ccdd94bdSDag-Erling Smørgrav struct httpio *io = (struct httpio *)v; 274e4878e39SDag-Erling Smørgrav 2759601e333SDag-Erling Smørgrav return (_fetch_write(io->conn, buf, len)); 2764ca1ab94SDag-Erling Smørgrav } 2774ca1ab94SDag-Erling Smørgrav 278f62e5228SDag-Erling Smørgrav /* 279f62e5228SDag-Erling Smørgrav * Close function 280f62e5228SDag-Erling Smørgrav */ 2814ca1ab94SDag-Erling Smørgrav static int 282e4878e39SDag-Erling Smørgrav _http_closefn(void *v) 2834ca1ab94SDag-Erling Smørgrav { 284ccdd94bdSDag-Erling Smørgrav struct httpio *io = (struct httpio *)v; 285e4878e39SDag-Erling Smørgrav int r; 286e4878e39SDag-Erling Smørgrav 287ccdd94bdSDag-Erling Smørgrav r = _fetch_close(io->conn); 288ccdd94bdSDag-Erling Smørgrav if (io->buf) 289ccdd94bdSDag-Erling Smørgrav free(io->buf); 290ccdd94bdSDag-Erling Smørgrav free(io); 291e19e6098SDag-Erling Smørgrav return (r); 2924ca1ab94SDag-Erling Smørgrav } 2934ca1ab94SDag-Erling Smørgrav 294f62e5228SDag-Erling Smørgrav /* 295e4878e39SDag-Erling Smørgrav * Wrap a file descriptor up 296f62e5228SDag-Erling Smørgrav */ 297e4878e39SDag-Erling Smørgrav static FILE * 2989601e333SDag-Erling Smørgrav _http_funopen(conn_t *conn, int chunked) 2994ca1ab94SDag-Erling Smørgrav { 300ccdd94bdSDag-Erling Smørgrav struct httpio *io; 301e4878e39SDag-Erling Smørgrav FILE *f; 302e4878e39SDag-Erling Smørgrav 303ccdd94bdSDag-Erling Smørgrav if ((io = calloc(1, sizeof *io)) == NULL) { 304e4878e39SDag-Erling Smørgrav _fetch_syserr(); 305e19e6098SDag-Erling Smørgrav return (NULL); 3064ca1ab94SDag-Erling Smørgrav } 307ccdd94bdSDag-Erling Smørgrav io->conn = conn; 3089601e333SDag-Erling Smørgrav io->chunked = chunked; 309ccdd94bdSDag-Erling Smørgrav f = funopen(io, _http_readfn, _http_writefn, NULL, _http_closefn); 310e19e6098SDag-Erling Smørgrav if (f == NULL) { 311e4878e39SDag-Erling Smørgrav _fetch_syserr(); 312ccdd94bdSDag-Erling Smørgrav free(io); 313e19e6098SDag-Erling Smørgrav return (NULL); 314e4878e39SDag-Erling Smørgrav } 315e19e6098SDag-Erling Smørgrav return (f); 316e4878e39SDag-Erling Smørgrav } 317e4878e39SDag-Erling Smørgrav 318e19e6098SDag-Erling Smørgrav 319e4878e39SDag-Erling Smørgrav /***************************************************************************** 320e4878e39SDag-Erling Smørgrav * Helper functions for talking to the server and parsing its replies 321e4878e39SDag-Erling Smørgrav */ 322e4878e39SDag-Erling Smørgrav 323e4878e39SDag-Erling Smørgrav /* Header types */ 324e4878e39SDag-Erling Smørgrav typedef enum { 325e4878e39SDag-Erling Smørgrav hdr_syserror = -2, 326e4878e39SDag-Erling Smørgrav hdr_error = -1, 327e4878e39SDag-Erling Smørgrav hdr_end = 0, 328e4878e39SDag-Erling Smørgrav hdr_unknown = 1, 329e4878e39SDag-Erling Smørgrav hdr_content_length, 330e4878e39SDag-Erling Smørgrav hdr_content_range, 331e4878e39SDag-Erling Smørgrav hdr_last_modified, 332e4878e39SDag-Erling Smørgrav hdr_location, 3336490b215SDag-Erling Smørgrav hdr_transfer_encoding, 3346490b215SDag-Erling Smørgrav hdr_www_authenticate 335f573a5fcSDag-Erling Smørgrav } hdr_t; 336e4878e39SDag-Erling Smørgrav 337e4878e39SDag-Erling Smørgrav /* Names of interesting headers */ 338e4878e39SDag-Erling Smørgrav static struct { 339f573a5fcSDag-Erling Smørgrav hdr_t num; 34038c7e4a6SArchie Cobbs const char *name; 341e4878e39SDag-Erling Smørgrav } hdr_names[] = { 342e4878e39SDag-Erling Smørgrav { hdr_content_length, "Content-Length" }, 343e4878e39SDag-Erling Smørgrav { hdr_content_range, "Content-Range" }, 344e4878e39SDag-Erling Smørgrav { hdr_last_modified, "Last-Modified" }, 345e4878e39SDag-Erling Smørgrav { hdr_location, "Location" }, 346e4878e39SDag-Erling Smørgrav { hdr_transfer_encoding, "Transfer-Encoding" }, 3476490b215SDag-Erling Smørgrav { hdr_www_authenticate, "WWW-Authenticate" }, 348e4878e39SDag-Erling Smørgrav { hdr_unknown, NULL }, 349e4878e39SDag-Erling Smørgrav }; 350e4878e39SDag-Erling Smørgrav 351e4878e39SDag-Erling Smørgrav /* 352e4878e39SDag-Erling Smørgrav * Send a formatted line; optionally echo to terminal 353e4878e39SDag-Erling Smørgrav */ 354e4878e39SDag-Erling Smørgrav static int 355dea29ca1SDag-Erling Smørgrav _http_cmd(conn_t *conn, const char *fmt, ...) 356e4878e39SDag-Erling Smørgrav { 357e4878e39SDag-Erling Smørgrav va_list ap; 358e4878e39SDag-Erling Smørgrav size_t len; 359e4878e39SDag-Erling Smørgrav char *msg; 360e4878e39SDag-Erling Smørgrav int r; 361e4878e39SDag-Erling Smørgrav 362e4878e39SDag-Erling Smørgrav va_start(ap, fmt); 363e4878e39SDag-Erling Smørgrav len = vasprintf(&msg, fmt, ap); 364e4878e39SDag-Erling Smørgrav va_end(ap); 365e4878e39SDag-Erling Smørgrav 366e4878e39SDag-Erling Smørgrav if (msg == NULL) { 367e4878e39SDag-Erling Smørgrav errno = ENOMEM; 368e4878e39SDag-Erling Smørgrav _fetch_syserr(); 369e19e6098SDag-Erling Smørgrav return (-1); 370e4878e39SDag-Erling Smørgrav } 371e4878e39SDag-Erling Smørgrav 372dea29ca1SDag-Erling Smørgrav r = _fetch_putln(conn, msg, len); 373e4878e39SDag-Erling Smørgrav free(msg); 374e4878e39SDag-Erling Smørgrav 375e4878e39SDag-Erling Smørgrav if (r == -1) { 376e4878e39SDag-Erling Smørgrav _fetch_syserr(); 377e19e6098SDag-Erling Smørgrav return (-1); 378e4878e39SDag-Erling Smørgrav } 379e4878e39SDag-Erling Smørgrav 380e19e6098SDag-Erling Smørgrav return (0); 381e4878e39SDag-Erling Smørgrav } 382e4878e39SDag-Erling Smørgrav 383e4878e39SDag-Erling Smørgrav /* 384e4878e39SDag-Erling Smørgrav * Get and parse status line 385e4878e39SDag-Erling Smørgrav */ 386e4878e39SDag-Erling Smørgrav static int 387dea29ca1SDag-Erling Smørgrav _http_get_reply(conn_t *conn) 388e4878e39SDag-Erling Smørgrav { 389a898bb8dSDag-Erling Smørgrav char *p; 390a898bb8dSDag-Erling Smørgrav 391dea29ca1SDag-Erling Smørgrav if (_fetch_getln(conn) == -1) 392e19e6098SDag-Erling Smørgrav return (-1); 393e4878e39SDag-Erling Smørgrav /* 394e4878e39SDag-Erling Smørgrav * A valid status line looks like "HTTP/m.n xyz reason" where m 395e4878e39SDag-Erling Smørgrav * and n are the major and minor protocol version numbers and xyz 396e4878e39SDag-Erling Smørgrav * is the reply code. 397a898bb8dSDag-Erling Smørgrav * Unfortunately, there are servers out there (NCSA 1.5.1, to name 398a898bb8dSDag-Erling Smørgrav * just one) that do not send a version number, so we can't rely 399a898bb8dSDag-Erling Smørgrav * on finding one, but if we do, insist on it being 1.0 or 1.1. 400e4878e39SDag-Erling Smørgrav * We don't care about the reason phrase. 401e4878e39SDag-Erling Smørgrav */ 402dea29ca1SDag-Erling Smørgrav if (strncmp(conn->buf, "HTTP", 4) != 0) 403e19e6098SDag-Erling Smørgrav return (HTTP_PROTOCOL_ERROR); 404dea29ca1SDag-Erling Smørgrav p = conn->buf + 4; 405a898bb8dSDag-Erling Smørgrav if (*p == '/') { 406a898bb8dSDag-Erling Smørgrav if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1')) 407e19e6098SDag-Erling Smørgrav return (HTTP_PROTOCOL_ERROR); 408a898bb8dSDag-Erling Smørgrav p += 4; 409a898bb8dSDag-Erling Smørgrav } 410e19e6098SDag-Erling Smørgrav if (*p != ' ' || !isdigit(p[1]) || !isdigit(p[2]) || !isdigit(p[3])) 411e19e6098SDag-Erling Smørgrav return (HTTP_PROTOCOL_ERROR); 412e4878e39SDag-Erling Smørgrav 413dea29ca1SDag-Erling Smørgrav conn->err = (p[1] - '0') * 100 + (p[2] - '0') * 10 + (p[3] - '0'); 414dea29ca1SDag-Erling Smørgrav return (conn->err); 415e4878e39SDag-Erling Smørgrav } 416e4878e39SDag-Erling Smørgrav 417e4878e39SDag-Erling Smørgrav /* 418e19e6098SDag-Erling Smørgrav * Check a header; if the type matches the given string, return a pointer 419e19e6098SDag-Erling Smørgrav * to the beginning of the value. 420e4878e39SDag-Erling Smørgrav */ 42138c7e4a6SArchie Cobbs static const char * 42238c7e4a6SArchie Cobbs _http_match(const char *str, const char *hdr) 423e4878e39SDag-Erling Smørgrav { 424e4878e39SDag-Erling Smørgrav while (*str && *hdr && tolower(*str++) == tolower(*hdr++)) 425e4878e39SDag-Erling Smørgrav /* nothing */; 426e4878e39SDag-Erling Smørgrav if (*str || *hdr != ':') 427e19e6098SDag-Erling Smørgrav return (NULL); 428e4878e39SDag-Erling Smørgrav while (*hdr && isspace(*++hdr)) 429e4878e39SDag-Erling Smørgrav /* nothing */; 430e19e6098SDag-Erling Smørgrav return (hdr); 431e4878e39SDag-Erling Smørgrav } 432e4878e39SDag-Erling Smørgrav 433e4878e39SDag-Erling Smørgrav /* 434e4878e39SDag-Erling Smørgrav * Get the next header and return the appropriate symbolic code. 435e4878e39SDag-Erling Smørgrav */ 436f573a5fcSDag-Erling Smørgrav static hdr_t 437dea29ca1SDag-Erling Smørgrav _http_next_header(conn_t *conn, const char **p) 438e4878e39SDag-Erling Smørgrav { 439e4878e39SDag-Erling Smørgrav int i; 440e4878e39SDag-Erling Smørgrav 441dea29ca1SDag-Erling Smørgrav if (_fetch_getln(conn) == -1) 442e19e6098SDag-Erling Smørgrav return (hdr_syserror); 443dea29ca1SDag-Erling Smørgrav while (conn->buflen && isspace(conn->buf[conn->buflen - 1])) 444dea29ca1SDag-Erling Smørgrav conn->buflen--; 445dea29ca1SDag-Erling Smørgrav conn->buf[conn->buflen] = '\0'; 446dea29ca1SDag-Erling Smørgrav if (conn->buflen == 0) 447e19e6098SDag-Erling Smørgrav return (hdr_end); 448e4878e39SDag-Erling Smørgrav /* 449e4878e39SDag-Erling Smørgrav * We could check for malformed headers but we don't really care. 450e4878e39SDag-Erling Smørgrav * A valid header starts with a token immediately followed by a 451e4878e39SDag-Erling Smørgrav * colon; a token is any sequence of non-control, non-whitespace 452e4878e39SDag-Erling Smørgrav * characters except "()<>@,;:\\\"{}". 453e4878e39SDag-Erling Smørgrav */ 454e4878e39SDag-Erling Smørgrav for (i = 0; hdr_names[i].num != hdr_unknown; i++) 455dea29ca1SDag-Erling Smørgrav if ((*p = _http_match(hdr_names[i].name, conn->buf)) != NULL) 456e19e6098SDag-Erling Smørgrav return (hdr_names[i].num); 457e19e6098SDag-Erling Smørgrav return (hdr_unknown); 458e4878e39SDag-Erling Smørgrav } 459e4878e39SDag-Erling Smørgrav 460e4878e39SDag-Erling Smørgrav /* 461e4878e39SDag-Erling Smørgrav * Parse a last-modified header 462e4878e39SDag-Erling Smørgrav */ 463c78f1cc9SDag-Erling Smørgrav static int 46438c7e4a6SArchie Cobbs _http_parse_mtime(const char *p, time_t *mtime) 465e4878e39SDag-Erling Smørgrav { 466c78f1cc9SDag-Erling Smørgrav char locale[64], *r; 467e4878e39SDag-Erling Smørgrav struct tm tm; 468e4878e39SDag-Erling Smørgrav 469e4878e39SDag-Erling Smørgrav strncpy(locale, setlocale(LC_TIME, NULL), sizeof locale); 470e4878e39SDag-Erling Smørgrav setlocale(LC_TIME, "C"); 471c78f1cc9SDag-Erling Smørgrav r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm); 472e4878e39SDag-Erling Smørgrav /* XXX should add support for date-2 and date-3 */ 473e4878e39SDag-Erling Smørgrav setlocale(LC_TIME, locale); 474c78f1cc9SDag-Erling Smørgrav if (r == NULL) 475e19e6098SDag-Erling Smørgrav return (-1); 476f67efa37SDag-Erling Smørgrav DEBUG(fprintf(stderr, "last modified: [%04d-%02d-%02d " 477f67efa37SDag-Erling Smørgrav "%02d:%02d:%02d]\n", 478e4878e39SDag-Erling Smørgrav tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday, 479e4878e39SDag-Erling Smørgrav tm.tm_hour, tm.tm_min, tm.tm_sec)); 480c78f1cc9SDag-Erling Smørgrav *mtime = timegm(&tm); 481e19e6098SDag-Erling Smørgrav return (0); 482e4878e39SDag-Erling Smørgrav } 483e4878e39SDag-Erling Smørgrav 484e4878e39SDag-Erling Smørgrav /* 485e4878e39SDag-Erling Smørgrav * Parse a content-length header 486e4878e39SDag-Erling Smørgrav */ 487c78f1cc9SDag-Erling Smørgrav static int 48838c7e4a6SArchie Cobbs _http_parse_length(const char *p, off_t *length) 489e4878e39SDag-Erling Smørgrav { 490be6aff99SDag-Erling Smørgrav off_t len; 491e4878e39SDag-Erling Smørgrav 492e4878e39SDag-Erling Smørgrav for (len = 0; *p && isdigit(*p); ++p) 493e4878e39SDag-Erling Smørgrav len = len * 10 + (*p - '0'); 494f573a5fcSDag-Erling Smørgrav if (*p) 495e19e6098SDag-Erling Smørgrav return (-1); 496f67efa37SDag-Erling Smørgrav DEBUG(fprintf(stderr, "content length: [%lld]\n", 497f573a5fcSDag-Erling Smørgrav (long long)len)); 498c78f1cc9SDag-Erling Smørgrav *length = len; 499e19e6098SDag-Erling Smørgrav return (0); 500e4878e39SDag-Erling Smørgrav } 501e4878e39SDag-Erling Smørgrav 502e4878e39SDag-Erling Smørgrav /* 503e4878e39SDag-Erling Smørgrav * Parse a content-range header 504e4878e39SDag-Erling Smørgrav */ 505c78f1cc9SDag-Erling Smørgrav static int 50638c7e4a6SArchie Cobbs _http_parse_range(const char *p, off_t *offset, off_t *length, off_t *size) 507e4878e39SDag-Erling Smørgrav { 508f573a5fcSDag-Erling Smørgrav off_t first, last, len; 509e4878e39SDag-Erling Smørgrav 510e4878e39SDag-Erling Smørgrav if (strncasecmp(p, "bytes ", 6) != 0) 511e19e6098SDag-Erling Smørgrav return (-1); 512c78f1cc9SDag-Erling Smørgrav for (first = 0, p += 6; *p && isdigit(*p); ++p) 513c78f1cc9SDag-Erling Smørgrav first = first * 10 + *p - '0'; 514e4878e39SDag-Erling Smørgrav if (*p != '-') 515e19e6098SDag-Erling Smørgrav return (-1); 516c78f1cc9SDag-Erling Smørgrav for (last = 0, ++p; *p && isdigit(*p); ++p) 517c78f1cc9SDag-Erling Smørgrav last = last * 10 + *p - '0'; 518c78f1cc9SDag-Erling Smørgrav if (first > last || *p != '/') 519e19e6098SDag-Erling Smørgrav return (-1); 520c78f1cc9SDag-Erling Smørgrav for (len = 0, ++p; *p && isdigit(*p); ++p) 521c78f1cc9SDag-Erling Smørgrav len = len * 10 + *p - '0'; 522f573a5fcSDag-Erling Smørgrav if (*p || len < last - first + 1) 523e19e6098SDag-Erling Smørgrav return (-1); 524f67efa37SDag-Erling Smørgrav DEBUG(fprintf(stderr, "content range: [%lld-%lld/%lld]\n", 525f573a5fcSDag-Erling Smørgrav (long long)first, (long long)last, (long long)len)); 526c78f1cc9SDag-Erling Smørgrav *offset = first; 527c78f1cc9SDag-Erling Smørgrav *length = last - first + 1; 528c78f1cc9SDag-Erling Smørgrav *size = len; 529e19e6098SDag-Erling Smørgrav return (0); 530e4878e39SDag-Erling Smørgrav } 531e4878e39SDag-Erling Smørgrav 532e19e6098SDag-Erling Smørgrav 533e4878e39SDag-Erling Smørgrav /***************************************************************************** 534e4878e39SDag-Erling Smørgrav * Helper functions for authorization 535e4878e39SDag-Erling Smørgrav */ 5364ca1ab94SDag-Erling Smørgrav 537f62e5228SDag-Erling Smørgrav /* 538f62e5228SDag-Erling Smørgrav * Base64 encoding 539f62e5228SDag-Erling Smørgrav */ 54035f723dbSDag-Erling Smørgrav static char * 541e19e6098SDag-Erling Smørgrav _http_base64(const char *src) 542f62e5228SDag-Erling Smørgrav { 543f62e5228SDag-Erling Smørgrav static const char base64[] = 544f62e5228SDag-Erling Smørgrav "ABCDEFGHIJKLMNOPQRSTUVWXYZ" 545f62e5228SDag-Erling Smørgrav "abcdefghijklmnopqrstuvwxyz" 546f62e5228SDag-Erling Smørgrav "0123456789+/"; 54735f723dbSDag-Erling Smørgrav char *str, *dst; 54835f723dbSDag-Erling Smørgrav size_t l; 54935f723dbSDag-Erling Smørgrav int t, r; 55035f723dbSDag-Erling Smørgrav 55135f723dbSDag-Erling Smørgrav l = strlen(src); 55235f723dbSDag-Erling Smørgrav if ((str = malloc(((l + 2) / 3) * 4)) == NULL) 553e19e6098SDag-Erling Smørgrav return (NULL); 55435f723dbSDag-Erling Smørgrav dst = str; 55535f723dbSDag-Erling Smørgrav r = 0; 556f62e5228SDag-Erling Smørgrav 557f62e5228SDag-Erling Smørgrav while (l >= 3) { 558f62e5228SDag-Erling Smørgrav t = (src[0] << 16) | (src[1] << 8) | src[2]; 559f62e5228SDag-Erling Smørgrav dst[0] = base64[(t >> 18) & 0x3f]; 560f62e5228SDag-Erling Smørgrav dst[1] = base64[(t >> 12) & 0x3f]; 561f62e5228SDag-Erling Smørgrav dst[2] = base64[(t >> 6) & 0x3f]; 562f62e5228SDag-Erling Smørgrav dst[3] = base64[(t >> 0) & 0x3f]; 563f62e5228SDag-Erling Smørgrav src += 3; l -= 3; 564f62e5228SDag-Erling Smørgrav dst += 4; r += 4; 565f62e5228SDag-Erling Smørgrav } 566f62e5228SDag-Erling Smørgrav 567f62e5228SDag-Erling Smørgrav switch (l) { 568f62e5228SDag-Erling Smørgrav case 2: 569f62e5228SDag-Erling Smørgrav t = (src[0] << 16) | (src[1] << 8); 570f62e5228SDag-Erling Smørgrav dst[0] = base64[(t >> 18) & 0x3f]; 571f62e5228SDag-Erling Smørgrav dst[1] = base64[(t >> 12) & 0x3f]; 572f62e5228SDag-Erling Smørgrav dst[2] = base64[(t >> 6) & 0x3f]; 573f62e5228SDag-Erling Smørgrav dst[3] = '='; 574f62e5228SDag-Erling Smørgrav dst += 4; 575f62e5228SDag-Erling Smørgrav r += 4; 576f62e5228SDag-Erling Smørgrav break; 577f62e5228SDag-Erling Smørgrav case 1: 578f62e5228SDag-Erling Smørgrav t = src[0] << 16; 579f62e5228SDag-Erling Smørgrav dst[0] = base64[(t >> 18) & 0x3f]; 580f62e5228SDag-Erling Smørgrav dst[1] = base64[(t >> 12) & 0x3f]; 581f62e5228SDag-Erling Smørgrav dst[2] = dst[3] = '='; 582f62e5228SDag-Erling Smørgrav dst += 4; 583f62e5228SDag-Erling Smørgrav r += 4; 584f62e5228SDag-Erling Smørgrav break; 585f62e5228SDag-Erling Smørgrav case 0: 586f62e5228SDag-Erling Smørgrav break; 587f62e5228SDag-Erling Smørgrav } 588f62e5228SDag-Erling Smørgrav 589f62e5228SDag-Erling Smørgrav *dst = 0; 590e19e6098SDag-Erling Smørgrav return (str); 591f62e5228SDag-Erling Smørgrav } 592f62e5228SDag-Erling Smørgrav 593f62e5228SDag-Erling Smørgrav /* 594f62e5228SDag-Erling Smørgrav * Encode username and password 595f62e5228SDag-Erling Smørgrav */ 59635f723dbSDag-Erling Smørgrav static int 597dea29ca1SDag-Erling Smørgrav _http_basic_auth(conn_t *conn, const char *hdr, const char *usr, const char *pwd) 598f62e5228SDag-Erling Smørgrav { 59935f723dbSDag-Erling Smørgrav char *upw, *auth; 60035f723dbSDag-Erling Smørgrav int r; 601f62e5228SDag-Erling Smørgrav 602f67efa37SDag-Erling Smørgrav DEBUG(fprintf(stderr, "usr: [%s]\n", usr)); 603f67efa37SDag-Erling Smørgrav DEBUG(fprintf(stderr, "pwd: [%s]\n", pwd)); 60435f723dbSDag-Erling Smørgrav if (asprintf(&upw, "%s:%s", usr, pwd) == -1) 605e19e6098SDag-Erling Smørgrav return (-1); 60635f723dbSDag-Erling Smørgrav auth = _http_base64(upw); 60735f723dbSDag-Erling Smørgrav free(upw); 60835f723dbSDag-Erling Smørgrav if (auth == NULL) 609e19e6098SDag-Erling Smørgrav return (-1); 610dea29ca1SDag-Erling Smørgrav r = _http_cmd(conn, "%s: Basic %s", hdr, auth); 61135f723dbSDag-Erling Smørgrav free(auth); 612e19e6098SDag-Erling Smørgrav return (r); 6134d029f13SDag-Erling Smørgrav } 61435f723dbSDag-Erling Smørgrav 61535f723dbSDag-Erling Smørgrav /* 61635f723dbSDag-Erling Smørgrav * Send an authorization header 61735f723dbSDag-Erling Smørgrav */ 61835f723dbSDag-Erling Smørgrav static int 619dea29ca1SDag-Erling Smørgrav _http_authorize(conn_t *conn, const char *hdr, const char *p) 62035f723dbSDag-Erling Smørgrav { 62135f723dbSDag-Erling Smørgrav /* basic authorization */ 62235f723dbSDag-Erling Smørgrav if (strncasecmp(p, "basic:", 6) == 0) { 62335f723dbSDag-Erling Smørgrav char *user, *pwd, *str; 62435f723dbSDag-Erling Smørgrav int r; 62535f723dbSDag-Erling Smørgrav 62635f723dbSDag-Erling Smørgrav /* skip realm */ 62735f723dbSDag-Erling Smørgrav for (p += 6; *p && *p != ':'; ++p) 62835f723dbSDag-Erling Smørgrav /* nothing */ ; 62935f723dbSDag-Erling Smørgrav if (!*p || strchr(++p, ':') == NULL) 630e19e6098SDag-Erling Smørgrav return (-1); 63135f723dbSDag-Erling Smørgrav if ((str = strdup(p)) == NULL) 632e19e6098SDag-Erling Smørgrav return (-1); /* XXX */ 63335f723dbSDag-Erling Smørgrav user = str; 63435f723dbSDag-Erling Smørgrav pwd = strchr(str, ':'); 63535f723dbSDag-Erling Smørgrav *pwd++ = '\0'; 636dea29ca1SDag-Erling Smørgrav r = _http_basic_auth(conn, hdr, user, pwd); 63735f723dbSDag-Erling Smørgrav free(str); 638e19e6098SDag-Erling Smørgrav return (r); 6394d029f13SDag-Erling Smørgrav } 640e19e6098SDag-Erling Smørgrav return (-1); 641f62e5228SDag-Erling Smørgrav } 642f62e5228SDag-Erling Smørgrav 643e19e6098SDag-Erling Smørgrav 644e4878e39SDag-Erling Smørgrav /***************************************************************************** 645e4878e39SDag-Erling Smørgrav * Helper functions for connecting to a server or proxy 646f62e5228SDag-Erling Smørgrav */ 647e4878e39SDag-Erling Smørgrav 648e4878e39SDag-Erling Smørgrav /* 649e4878e39SDag-Erling Smørgrav * Connect to the correct HTTP server or proxy. 65060245e42SDag-Erling Smørgrav */ 651dea29ca1SDag-Erling Smørgrav static conn_t * 65238c7e4a6SArchie Cobbs _http_connect(struct url *URL, struct url *purl, const char *flags) 65360245e42SDag-Erling Smørgrav { 654dea29ca1SDag-Erling Smørgrav conn_t *conn; 6551a16ed4cSDag-Erling Smørgrav int verbose; 656dea29ca1SDag-Erling Smørgrav int af; 657e4878e39SDag-Erling Smørgrav 658e4878e39SDag-Erling Smørgrav #ifdef INET6 659e4878e39SDag-Erling Smørgrav af = AF_UNSPEC; 660e4878e39SDag-Erling Smørgrav #else 661e4878e39SDag-Erling Smørgrav af = AF_INET; 662e4878e39SDag-Erling Smørgrav #endif 663e4878e39SDag-Erling Smørgrav 664d74a913bSDag-Erling Smørgrav verbose = CHECK_FLAG('v'); 665d74a913bSDag-Erling Smørgrav if (CHECK_FLAG('4')) 666e4878e39SDag-Erling Smørgrav af = AF_INET; 6671a16ed4cSDag-Erling Smørgrav #ifdef INET6 668d74a913bSDag-Erling Smørgrav else if (CHECK_FLAG('6')) 669e4878e39SDag-Erling Smørgrav af = AF_INET6; 6701a16ed4cSDag-Erling Smørgrav #endif 671e4878e39SDag-Erling Smørgrav 672111e2510SDag-Erling Smørgrav if (purl && strcasecmp(URL->scheme, SCHEME_HTTPS) != 0) { 6731a16ed4cSDag-Erling Smørgrav URL = purl; 6741a16ed4cSDag-Erling Smørgrav } else if (strcasecmp(URL->scheme, SCHEME_FTP) == 0) { 675e4878e39SDag-Erling Smørgrav /* can't talk http to an ftp server */ 676e4878e39SDag-Erling Smørgrav /* XXX should set an error code */ 677dea29ca1SDag-Erling Smørgrav return (NULL); 678e4878e39SDag-Erling Smørgrav } 6791a16ed4cSDag-Erling Smørgrav 680dea29ca1SDag-Erling Smørgrav if ((conn = _fetch_connect(URL->host, URL->port, af, verbose)) == NULL) 681e4878e39SDag-Erling Smørgrav /* _fetch_connect() has already set an error code */ 682dea29ca1SDag-Erling Smørgrav return (NULL); 683111e2510SDag-Erling Smørgrav if (strcasecmp(URL->scheme, SCHEME_HTTPS) == 0 && 684111e2510SDag-Erling Smørgrav _fetch_ssl(conn, verbose) == -1) { 685111e2510SDag-Erling Smørgrav _fetch_close(conn); 686111e2510SDag-Erling Smørgrav return (NULL); 687111e2510SDag-Erling Smørgrav } 688dea29ca1SDag-Erling Smørgrav return (conn); 689e4878e39SDag-Erling Smørgrav } 690e4878e39SDag-Erling Smørgrav 6911a16ed4cSDag-Erling Smørgrav static struct url * 69238c7e4a6SArchie Cobbs _http_get_proxy(void) 6931a16ed4cSDag-Erling Smørgrav { 6941a16ed4cSDag-Erling Smørgrav struct url *purl; 6951a16ed4cSDag-Erling Smørgrav char *p; 6961a16ed4cSDag-Erling Smørgrav 6974cee73c8SDag-Erling Smørgrav if (((p = getenv("HTTP_PROXY")) || (p = getenv("http_proxy"))) && 6984cee73c8SDag-Erling Smørgrav (purl = fetchParseURL(p))) { 6991a16ed4cSDag-Erling Smørgrav if (!*purl->scheme) 7001a16ed4cSDag-Erling Smørgrav strcpy(purl->scheme, SCHEME_HTTP); 7011a16ed4cSDag-Erling Smørgrav if (!purl->port) 702e828ada7SDag-Erling Smørgrav purl->port = _fetch_default_proxy_port(purl->scheme); 7031a16ed4cSDag-Erling Smørgrav if (strcasecmp(purl->scheme, SCHEME_HTTP) == 0) 704e19e6098SDag-Erling Smørgrav return (purl); 7051a16ed4cSDag-Erling Smørgrav fetchFreeURL(purl); 7061a16ed4cSDag-Erling Smørgrav } 707e19e6098SDag-Erling Smørgrav return (NULL); 708e4878e39SDag-Erling Smørgrav } 709e4878e39SDag-Erling Smørgrav 710a8e9bd87SDag-Erling Smørgrav static void 711a8e9bd87SDag-Erling Smørgrav _http_print_html(FILE *out, FILE *in) 712a8e9bd87SDag-Erling Smørgrav { 713a8e9bd87SDag-Erling Smørgrav size_t len; 714a8e9bd87SDag-Erling Smørgrav char *line, *p, *q; 715a8e9bd87SDag-Erling Smørgrav int comment, tag; 716a8e9bd87SDag-Erling Smørgrav 717a8e9bd87SDag-Erling Smørgrav comment = tag = 0; 718a8e9bd87SDag-Erling Smørgrav while ((line = fgetln(in, &len)) != NULL) { 719a8e9bd87SDag-Erling Smørgrav while (len && isspace(line[len - 1])) 720a8e9bd87SDag-Erling Smørgrav --len; 721a8e9bd87SDag-Erling Smørgrav for (p = q = line; q < line + len; ++q) { 722a8e9bd87SDag-Erling Smørgrav if (comment && *q == '-') { 723e19e6098SDag-Erling Smørgrav if (q + 2 < line + len && 724e19e6098SDag-Erling Smørgrav strcmp(q, "-->") == 0) { 725a8e9bd87SDag-Erling Smørgrav tag = comment = 0; 726a8e9bd87SDag-Erling Smørgrav q += 2; 727a8e9bd87SDag-Erling Smørgrav } 728e19e6098SDag-Erling Smørgrav } else if (tag && !comment && *q == '>') { 729a8e9bd87SDag-Erling Smørgrav p = q + 1; 730a8e9bd87SDag-Erling Smørgrav tag = 0; 731a8e9bd87SDag-Erling Smørgrav } else if (!tag && *q == '<') { 732a8e9bd87SDag-Erling Smørgrav if (q > p) 733a8e9bd87SDag-Erling Smørgrav fwrite(p, q - p, 1, out); 734a8e9bd87SDag-Erling Smørgrav tag = 1; 735e19e6098SDag-Erling Smørgrav if (q + 3 < line + len && 736e19e6098SDag-Erling Smørgrav strcmp(q, "<!--") == 0) { 737a8e9bd87SDag-Erling Smørgrav comment = 1; 738a8e9bd87SDag-Erling Smørgrav q += 3; 739a8e9bd87SDag-Erling Smørgrav } 740a8e9bd87SDag-Erling Smørgrav } 741a8e9bd87SDag-Erling Smørgrav } 742a8e9bd87SDag-Erling Smørgrav if (!tag && q > p) 743a8e9bd87SDag-Erling Smørgrav fwrite(p, q - p, 1, out); 744a8e9bd87SDag-Erling Smørgrav fputc('\n', out); 745a8e9bd87SDag-Erling Smørgrav } 746a8e9bd87SDag-Erling Smørgrav } 747a8e9bd87SDag-Erling Smørgrav 748e19e6098SDag-Erling Smørgrav 749e4878e39SDag-Erling Smørgrav /***************************************************************************** 750e4878e39SDag-Erling Smørgrav * Core 751e4878e39SDag-Erling Smørgrav */ 752e4878e39SDag-Erling Smørgrav 753e4878e39SDag-Erling Smørgrav /* 754e4878e39SDag-Erling Smørgrav * Send a request and process the reply 7559601e333SDag-Erling Smørgrav * 7569601e333SDag-Erling Smørgrav * XXX This function is way too long, the do..while loop should be split 7579601e333SDag-Erling Smørgrav * XXX off into a separate function. 758e4878e39SDag-Erling Smørgrav */ 7591a16ed4cSDag-Erling Smørgrav FILE * 76038c7e4a6SArchie Cobbs _http_request(struct url *URL, const char *op, struct url_stat *us, 76138c7e4a6SArchie Cobbs struct url *purl, const char *flags) 762e4878e39SDag-Erling Smørgrav { 763dea29ca1SDag-Erling Smørgrav conn_t *conn; 764e4878e39SDag-Erling Smørgrav struct url *url, *new; 7651a16ed4cSDag-Erling Smørgrav int chunked, direct, need_auth, noredirect, verbose; 766dea29ca1SDag-Erling Smørgrav int i, n; 767be6aff99SDag-Erling Smørgrav off_t offset, clength, length, size; 768c78f1cc9SDag-Erling Smørgrav time_t mtime; 76938c7e4a6SArchie Cobbs const char *p; 770e4878e39SDag-Erling Smørgrav FILE *f; 771f573a5fcSDag-Erling Smørgrav hdr_t h; 77228c645cfSHajimu UMEMOTO char *host; 77328c645cfSHajimu UMEMOTO #ifdef INET6 77428c645cfSHajimu UMEMOTO char hbuf[MAXHOSTNAMELEN + 1]; 77528c645cfSHajimu UMEMOTO #endif 77660245e42SDag-Erling Smørgrav 777d74a913bSDag-Erling Smørgrav direct = CHECK_FLAG('d'); 778d74a913bSDag-Erling Smørgrav noredirect = CHECK_FLAG('A'); 779d74a913bSDag-Erling Smørgrav verbose = CHECK_FLAG('v'); 7804ca1ab94SDag-Erling Smørgrav 7811a16ed4cSDag-Erling Smørgrav if (direct && purl) { 7821a16ed4cSDag-Erling Smørgrav fetchFreeURL(purl); 7831a16ed4cSDag-Erling Smørgrav purl = NULL; 7841a16ed4cSDag-Erling Smørgrav } 7851a16ed4cSDag-Erling Smørgrav 786c78f1cc9SDag-Erling Smørgrav /* try the provided URL first */ 787c78f1cc9SDag-Erling Smørgrav url = URL; 788c78f1cc9SDag-Erling Smørgrav 789c78f1cc9SDag-Erling Smørgrav /* if the A flag is set, we only get one try */ 790e4878e39SDag-Erling Smørgrav n = noredirect ? 1 : MAX_REDIRECT; 791c78f1cc9SDag-Erling Smørgrav i = 0; 792e4878e39SDag-Erling Smørgrav 7936490b215SDag-Erling Smørgrav need_auth = 0; 794c78f1cc9SDag-Erling Smørgrav do { 795f4683775SDag-Erling Smørgrav new = NULL; 796f4683775SDag-Erling Smørgrav chunked = 0; 797f4683775SDag-Erling Smørgrav offset = 0; 798c78f1cc9SDag-Erling Smørgrav clength = -1; 799c78f1cc9SDag-Erling Smørgrav length = -1; 800c78f1cc9SDag-Erling Smørgrav size = -1; 801c78f1cc9SDag-Erling Smørgrav mtime = 0; 8026490b215SDag-Erling Smørgrav 8031a16ed4cSDag-Erling Smørgrav /* check port */ 8041a16ed4cSDag-Erling Smørgrav if (!url->port) 805e828ada7SDag-Erling Smørgrav url->port = _fetch_default_port(url->scheme); 8061a16ed4cSDag-Erling Smørgrav 80765986545SDag-Erling Smørgrav /* were we redirected to an FTP URL? */ 80865986545SDag-Erling Smørgrav if (purl == NULL && strcmp(url->scheme, SCHEME_FTP) == 0) { 80965986545SDag-Erling Smørgrav if (strcmp(op, "GET") == 0) 810e19e6098SDag-Erling Smørgrav return (_ftp_request(url, "RETR", us, purl, flags)); 81165986545SDag-Erling Smørgrav else if (strcmp(op, "HEAD") == 0) 812e19e6098SDag-Erling Smørgrav return (_ftp_request(url, "STAT", us, purl, flags)); 81365986545SDag-Erling Smørgrav } 81465986545SDag-Erling Smørgrav 815e4878e39SDag-Erling Smørgrav /* connect to server or proxy */ 816dea29ca1SDag-Erling Smørgrav if ((conn = _http_connect(url, purl, flags)) == NULL) 817e4878e39SDag-Erling Smørgrav goto ouch; 818e4878e39SDag-Erling Smørgrav 819e4878e39SDag-Erling Smørgrav host = url->host; 82028c645cfSHajimu UMEMOTO #ifdef INET6 821e4878e39SDag-Erling Smørgrav if (strchr(url->host, ':')) { 822e4878e39SDag-Erling Smørgrav snprintf(hbuf, sizeof(hbuf), "[%s]", url->host); 82328c645cfSHajimu UMEMOTO host = hbuf; 82428c645cfSHajimu UMEMOTO } 82528c645cfSHajimu UMEMOTO #endif 82628c645cfSHajimu UMEMOTO 827e4878e39SDag-Erling Smørgrav /* send request */ 8280fba3a00SDag-Erling Smørgrav if (verbose) 829c97925adSHajimu UMEMOTO _fetch_info("requesting %s://%s:%d%s", 830e4878e39SDag-Erling Smørgrav url->scheme, host, url->port, url->doc); 8311a16ed4cSDag-Erling Smørgrav if (purl) { 832dea29ca1SDag-Erling Smørgrav _http_cmd(conn, "%s %s://%s:%d%s HTTP/1.1", 833e4878e39SDag-Erling Smørgrav op, url->scheme, host, url->port, url->doc); 834e4878e39SDag-Erling Smørgrav } else { 835dea29ca1SDag-Erling Smørgrav _http_cmd(conn, "%s %s HTTP/1.1", 836e4878e39SDag-Erling Smørgrav op, url->doc); 83760245e42SDag-Erling Smørgrav } 83860245e42SDag-Erling Smørgrav 8396490b215SDag-Erling Smørgrav /* virtual host */ 8406490b215SDag-Erling Smørgrav if (url->port == _fetch_default_port(url->scheme)) 841dea29ca1SDag-Erling Smørgrav _http_cmd(conn, "Host: %s", host); 8426490b215SDag-Erling Smørgrav else 843dea29ca1SDag-Erling Smørgrav _http_cmd(conn, "Host: %s:%d", host, url->port); 8446490b215SDag-Erling Smørgrav 845e4878e39SDag-Erling Smørgrav /* proxy authorization */ 8461a16ed4cSDag-Erling Smørgrav if (purl) { 8471a16ed4cSDag-Erling Smørgrav if (*purl->user || *purl->pwd) 848dea29ca1SDag-Erling Smørgrav _http_basic_auth(conn, "Proxy-Authorization", 8491a16ed4cSDag-Erling Smørgrav purl->user, purl->pwd); 8501a16ed4cSDag-Erling Smørgrav else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL && *p != '\0') 851dea29ca1SDag-Erling Smørgrav _http_authorize(conn, "Proxy-Authorization", p); 8521a16ed4cSDag-Erling Smørgrav } 853e4878e39SDag-Erling Smørgrav 854e4878e39SDag-Erling Smørgrav /* server authorization */ 8556490b215SDag-Erling Smørgrav if (need_auth || *url->user || *url->pwd) { 856e4878e39SDag-Erling Smørgrav if (*url->user || *url->pwd) 857dea29ca1SDag-Erling Smørgrav _http_basic_auth(conn, "Authorization", url->user, url->pwd); 858c78f1cc9SDag-Erling Smørgrav else if ((p = getenv("HTTP_AUTH")) != NULL && *p != '\0') 859dea29ca1SDag-Erling Smørgrav _http_authorize(conn, "Authorization", p); 8606490b215SDag-Erling Smørgrav else if (fetchAuthMethod && fetchAuthMethod(url) == 0) { 861dea29ca1SDag-Erling Smørgrav _http_basic_auth(conn, "Authorization", url->user, url->pwd); 8626490b215SDag-Erling Smørgrav } else { 863e4878e39SDag-Erling Smørgrav _http_seterr(HTTP_NEED_AUTH); 864e4878e39SDag-Erling Smørgrav goto ouch; 865e4878e39SDag-Erling Smørgrav } 866e4878e39SDag-Erling Smørgrav } 867e4878e39SDag-Erling Smørgrav 868e4878e39SDag-Erling Smørgrav /* other headers */ 8696a0cf64bSDag-Erling Smørgrav if ((p = getenv("HTTP_USER_AGENT")) != NULL && *p != '\0') 870dea29ca1SDag-Erling Smørgrav _http_cmd(conn, "User-Agent: %s", p); 8716a0cf64bSDag-Erling Smørgrav else 872dea29ca1SDag-Erling Smørgrav _http_cmd(conn, "User-Agent: %s " _LIBFETCH_VER, getprogname()); 87387b41116SDag-Erling Smørgrav if (url->offset) 874dea29ca1SDag-Erling Smørgrav _http_cmd(conn, "Range: bytes=%lld-", (long long)url->offset); 875dea29ca1SDag-Erling Smørgrav _http_cmd(conn, "Connection: close"); 876dea29ca1SDag-Erling Smørgrav _http_cmd(conn, ""); 877e4878e39SDag-Erling Smørgrav 878e4878e39SDag-Erling Smørgrav /* get reply */ 879dea29ca1SDag-Erling Smørgrav switch (_http_get_reply(conn)) { 880e4878e39SDag-Erling Smørgrav case HTTP_OK: 881e4878e39SDag-Erling Smørgrav case HTTP_PARTIAL: 882e4878e39SDag-Erling Smørgrav /* fine */ 883e4878e39SDag-Erling Smørgrav break; 884e4878e39SDag-Erling Smørgrav case HTTP_MOVED_PERM: 885e4878e39SDag-Erling Smørgrav case HTTP_MOVED_TEMP: 88665986545SDag-Erling Smørgrav case HTTP_SEE_OTHER: 887e4878e39SDag-Erling Smørgrav /* 888e4878e39SDag-Erling Smørgrav * Not so fine, but we still have to read the headers to 889e4878e39SDag-Erling Smørgrav * get the new location. 890e4878e39SDag-Erling Smørgrav */ 891e4878e39SDag-Erling Smørgrav break; 892e4878e39SDag-Erling Smørgrav case HTTP_NEED_AUTH: 893e4878e39SDag-Erling Smørgrav if (need_auth) { 894e4878e39SDag-Erling Smørgrav /* 895e4878e39SDag-Erling Smørgrav * We already sent out authorization code, so there's 896e4878e39SDag-Erling Smørgrav * nothing more we can do. 897e4878e39SDag-Erling Smørgrav */ 898dea29ca1SDag-Erling Smørgrav _http_seterr(conn->err); 899e4878e39SDag-Erling Smørgrav goto ouch; 900e4878e39SDag-Erling Smørgrav } 901e4878e39SDag-Erling Smørgrav /* try again, but send the password this time */ 902e4878e39SDag-Erling Smørgrav if (verbose) 903e4878e39SDag-Erling Smørgrav _fetch_info("server requires authorization"); 9046490b215SDag-Erling Smørgrav break; 905e4878e39SDag-Erling Smørgrav case HTTP_NEED_PROXY_AUTH: 906e4878e39SDag-Erling Smørgrav /* 907e4878e39SDag-Erling Smørgrav * If we're talking to a proxy, we already sent our proxy 908e4878e39SDag-Erling Smørgrav * authorization code, so there's nothing more we can do. 909e4878e39SDag-Erling Smørgrav */ 910dea29ca1SDag-Erling Smørgrav _http_seterr(conn->err); 911e4878e39SDag-Erling Smørgrav goto ouch; 912e4878e39SDag-Erling Smørgrav case HTTP_PROTOCOL_ERROR: 913e4878e39SDag-Erling Smørgrav /* fall through */ 914e4878e39SDag-Erling Smørgrav case -1: 915e4878e39SDag-Erling Smørgrav _fetch_syserr(); 916e4878e39SDag-Erling Smørgrav goto ouch; 917e4878e39SDag-Erling Smørgrav default: 918dea29ca1SDag-Erling Smørgrav _http_seterr(conn->err); 919a8e9bd87SDag-Erling Smørgrav if (!verbose) 920e4878e39SDag-Erling Smørgrav goto ouch; 921a8e9bd87SDag-Erling Smørgrav /* fall through so we can get the full error message */ 922e4878e39SDag-Erling Smørgrav } 923e4878e39SDag-Erling Smørgrav 924e4878e39SDag-Erling Smørgrav /* get headers */ 925e4878e39SDag-Erling Smørgrav do { 926dea29ca1SDag-Erling Smørgrav switch ((h = _http_next_header(conn, &p))) { 927e4878e39SDag-Erling Smørgrav case hdr_syserror: 928e4878e39SDag-Erling Smørgrav _fetch_syserr(); 929e4878e39SDag-Erling Smørgrav goto ouch; 930e4878e39SDag-Erling Smørgrav case hdr_error: 931e4878e39SDag-Erling Smørgrav _http_seterr(HTTP_PROTOCOL_ERROR); 932e4878e39SDag-Erling Smørgrav goto ouch; 933e4878e39SDag-Erling Smørgrav case hdr_content_length: 934c78f1cc9SDag-Erling Smørgrav _http_parse_length(p, &clength); 935e4878e39SDag-Erling Smørgrav break; 936e4878e39SDag-Erling Smørgrav case hdr_content_range: 937c78f1cc9SDag-Erling Smørgrav _http_parse_range(p, &offset, &length, &size); 938e4878e39SDag-Erling Smørgrav break; 939e4878e39SDag-Erling Smørgrav case hdr_last_modified: 940c78f1cc9SDag-Erling Smørgrav _http_parse_mtime(p, &mtime); 941e4878e39SDag-Erling Smørgrav break; 942e4878e39SDag-Erling Smørgrav case hdr_location: 943dea29ca1SDag-Erling Smørgrav if (!HTTP_REDIRECT(conn->err)) 944e4878e39SDag-Erling Smørgrav break; 945f4683775SDag-Erling Smørgrav if (new) 946f4683775SDag-Erling Smørgrav free(new); 947e4878e39SDag-Erling Smørgrav if (verbose) 948dea29ca1SDag-Erling Smørgrav _fetch_info("%d redirect to %s", conn->err, p); 949f4683775SDag-Erling Smørgrav if (*p == '/') 950f4683775SDag-Erling Smørgrav /* absolute path */ 951f4683775SDag-Erling Smørgrav new = fetchMakeURL(url->scheme, url->host, url->port, p, 952f4683775SDag-Erling Smørgrav url->user, url->pwd); 953f4683775SDag-Erling Smørgrav else 954f4683775SDag-Erling Smørgrav new = fetchParseURL(p); 955f4683775SDag-Erling Smørgrav if (new == NULL) { 956f4683775SDag-Erling Smørgrav /* XXX should set an error code */ 957f4683775SDag-Erling Smørgrav DEBUG(fprintf(stderr, "failed to parse new URL\n")); 958e4878e39SDag-Erling Smørgrav goto ouch; 959f4683775SDag-Erling Smørgrav } 960e4878e39SDag-Erling Smørgrav if (!*new->user && !*new->pwd) { 961e4878e39SDag-Erling Smørgrav strcpy(new->user, url->user); 962e4878e39SDag-Erling Smørgrav strcpy(new->pwd, url->pwd); 963e4878e39SDag-Erling Smørgrav } 964e4878e39SDag-Erling Smørgrav new->offset = url->offset; 965e4878e39SDag-Erling Smørgrav new->length = url->length; 966f4683775SDag-Erling Smørgrav break; 967e4878e39SDag-Erling Smørgrav case hdr_transfer_encoding: 968e4878e39SDag-Erling Smørgrav /* XXX weak test*/ 969e4878e39SDag-Erling Smørgrav chunked = (strcasecmp(p, "chunked") == 0); 970e4878e39SDag-Erling Smørgrav break; 9716490b215SDag-Erling Smørgrav case hdr_www_authenticate: 972dea29ca1SDag-Erling Smørgrav if (conn->err != HTTP_NEED_AUTH) 9736490b215SDag-Erling Smørgrav break; 9746490b215SDag-Erling Smørgrav /* if we were smarter, we'd check the method and realm */ 9756490b215SDag-Erling Smørgrav break; 976e4878e39SDag-Erling Smørgrav case hdr_end: 977e4878e39SDag-Erling Smørgrav /* fall through */ 978e4878e39SDag-Erling Smørgrav case hdr_unknown: 979e4878e39SDag-Erling Smørgrav /* ignore */ 980e4878e39SDag-Erling Smørgrav break; 981e4878e39SDag-Erling Smørgrav } 982e4878e39SDag-Erling Smørgrav } while (h > hdr_end); 983e4878e39SDag-Erling Smørgrav 984a8e9bd87SDag-Erling Smørgrav /* we have a hit or an error */ 985dea29ca1SDag-Erling Smørgrav if (conn->err == HTTP_OK || conn->err == HTTP_PARTIAL || HTTP_ERROR(conn->err)) 986e4878e39SDag-Erling Smørgrav break; 987f4683775SDag-Erling Smørgrav 9886490b215SDag-Erling Smørgrav /* we need to provide authentication */ 989dea29ca1SDag-Erling Smørgrav if (conn->err == HTTP_NEED_AUTH) { 9906490b215SDag-Erling Smørgrav need_auth = 1; 991dea29ca1SDag-Erling Smørgrav _fetch_close(conn); 992dea29ca1SDag-Erling Smørgrav conn = NULL; 9936490b215SDag-Erling Smørgrav continue; 9946490b215SDag-Erling Smørgrav } 9956490b215SDag-Erling Smørgrav 9966490b215SDag-Erling Smørgrav /* all other cases: we got a redirect */ 9976490b215SDag-Erling Smørgrav need_auth = 0; 998dea29ca1SDag-Erling Smørgrav _fetch_close(conn); 999dea29ca1SDag-Erling Smørgrav conn = NULL; 10006490b215SDag-Erling Smørgrav if (!new) { 10016490b215SDag-Erling Smørgrav DEBUG(fprintf(stderr, "redirect with no new location\n")); 10026490b215SDag-Erling Smørgrav break; 10036490b215SDag-Erling Smørgrav } 1004f4683775SDag-Erling Smørgrav if (url != URL) 1005f4683775SDag-Erling Smørgrav fetchFreeURL(url); 1006f4683775SDag-Erling Smørgrav url = new; 1007c78f1cc9SDag-Erling Smørgrav } while (++i < n); 1008e4878e39SDag-Erling Smørgrav 10096490b215SDag-Erling Smørgrav /* we failed, or ran out of retries */ 1010dea29ca1SDag-Erling Smørgrav if (conn == NULL) { 1011dea29ca1SDag-Erling Smørgrav _http_seterr(conn->err); 1012e4878e39SDag-Erling Smørgrav goto ouch; 1013e4878e39SDag-Erling Smørgrav } 1014e4878e39SDag-Erling Smørgrav 1015f573a5fcSDag-Erling Smørgrav DEBUG(fprintf(stderr, "offset %lld, length %lld," 1016f573a5fcSDag-Erling Smørgrav " size %lld, clength %lld\n", 1017f573a5fcSDag-Erling Smørgrav (long long)offset, (long long)length, 1018f573a5fcSDag-Erling Smørgrav (long long)size, (long long)clength)); 1019c78f1cc9SDag-Erling Smørgrav 1020c78f1cc9SDag-Erling Smørgrav /* check for inconsistencies */ 1021c78f1cc9SDag-Erling Smørgrav if (clength != -1 && length != -1 && clength != length) { 1022c78f1cc9SDag-Erling Smørgrav _http_seterr(HTTP_PROTOCOL_ERROR); 1023c78f1cc9SDag-Erling Smørgrav goto ouch; 1024c78f1cc9SDag-Erling Smørgrav } 1025c78f1cc9SDag-Erling Smørgrav if (clength == -1) 1026c78f1cc9SDag-Erling Smørgrav clength = length; 1027c78f1cc9SDag-Erling Smørgrav if (clength != -1) 1028c78f1cc9SDag-Erling Smørgrav length = offset + clength; 1029c78f1cc9SDag-Erling Smørgrav if (length != -1 && size != -1 && length != size) { 1030c78f1cc9SDag-Erling Smørgrav _http_seterr(HTTP_PROTOCOL_ERROR); 1031c78f1cc9SDag-Erling Smørgrav goto ouch; 1032c78f1cc9SDag-Erling Smørgrav } 1033c78f1cc9SDag-Erling Smørgrav if (size == -1) 1034c78f1cc9SDag-Erling Smørgrav size = length; 1035c78f1cc9SDag-Erling Smørgrav 1036c78f1cc9SDag-Erling Smørgrav /* fill in stats */ 1037c78f1cc9SDag-Erling Smørgrav if (us) { 1038c78f1cc9SDag-Erling Smørgrav us->size = size; 1039c78f1cc9SDag-Erling Smørgrav us->atime = us->mtime = mtime; 1040c78f1cc9SDag-Erling Smørgrav } 1041c78f1cc9SDag-Erling Smørgrav 104287b41116SDag-Erling Smørgrav /* too far? */ 1043c78f1cc9SDag-Erling Smørgrav if (offset > URL->offset) { 104487b41116SDag-Erling Smørgrav _http_seterr(HTTP_PROTOCOL_ERROR); 104587b41116SDag-Erling Smørgrav goto ouch; 104687b41116SDag-Erling Smørgrav } 104787b41116SDag-Erling Smørgrav 1048c78f1cc9SDag-Erling Smørgrav /* report back real offset and size */ 104987b41116SDag-Erling Smørgrav URL->offset = offset; 1050c78f1cc9SDag-Erling Smørgrav URL->length = clength; 105187b41116SDag-Erling Smørgrav 1052e4878e39SDag-Erling Smørgrav /* wrap it up in a FILE */ 10539601e333SDag-Erling Smørgrav if ((f = _http_funopen(conn, chunked)) == NULL) { 1054e4878e39SDag-Erling Smørgrav _fetch_syserr(); 1055e4878e39SDag-Erling Smørgrav goto ouch; 1056e4878e39SDag-Erling Smørgrav } 1057e4878e39SDag-Erling Smørgrav 1058e4878e39SDag-Erling Smørgrav if (url != URL) 1059e4878e39SDag-Erling Smørgrav fetchFreeURL(url); 10601a16ed4cSDag-Erling Smørgrav if (purl) 10611a16ed4cSDag-Erling Smørgrav fetchFreeURL(purl); 1062e4878e39SDag-Erling Smørgrav 1063dea29ca1SDag-Erling Smørgrav if (HTTP_ERROR(conn->err)) { 1064a8e9bd87SDag-Erling Smørgrav _http_print_html(stderr, f); 1065a8e9bd87SDag-Erling Smørgrav fclose(f); 1066a8e9bd87SDag-Erling Smørgrav f = NULL; 1067a8e9bd87SDag-Erling Smørgrav } 1068a8e9bd87SDag-Erling Smørgrav 1069e19e6098SDag-Erling Smørgrav return (f); 1070e4878e39SDag-Erling Smørgrav 1071e4878e39SDag-Erling Smørgrav ouch: 1072e4878e39SDag-Erling Smørgrav if (url != URL) 1073e4878e39SDag-Erling Smørgrav fetchFreeURL(url); 10741a16ed4cSDag-Erling Smørgrav if (purl) 10751a16ed4cSDag-Erling Smørgrav fetchFreeURL(purl); 1076dea29ca1SDag-Erling Smørgrav if (conn != NULL) 1077dea29ca1SDag-Erling Smørgrav _fetch_close(conn); 1078e19e6098SDag-Erling Smørgrav return (NULL); 1079e4878e39SDag-Erling Smørgrav } 1080e4878e39SDag-Erling Smørgrav 1081e19e6098SDag-Erling Smørgrav 1082e4878e39SDag-Erling Smørgrav /***************************************************************************** 1083e4878e39SDag-Erling Smørgrav * Entry points 1084e4878e39SDag-Erling Smørgrav */ 1085e4878e39SDag-Erling Smørgrav 108660245e42SDag-Erling Smørgrav /* 10871a5faa10SDag-Erling Smørgrav * Retrieve and stat a file by HTTP 10881a5faa10SDag-Erling Smørgrav */ 10891a5faa10SDag-Erling Smørgrav FILE * 109038c7e4a6SArchie Cobbs fetchXGetHTTP(struct url *URL, struct url_stat *us, const char *flags) 10911a5faa10SDag-Erling Smørgrav { 1092e19e6098SDag-Erling Smørgrav return (_http_request(URL, "GET", us, _http_get_proxy(), flags)); 10931a5faa10SDag-Erling Smørgrav } 10941a5faa10SDag-Erling Smørgrav 10951a5faa10SDag-Erling Smørgrav /* 109660245e42SDag-Erling Smørgrav * Retrieve a file by HTTP 109760245e42SDag-Erling Smørgrav */ 109860245e42SDag-Erling Smørgrav FILE * 109938c7e4a6SArchie Cobbs fetchGetHTTP(struct url *URL, const char *flags) 110060245e42SDag-Erling Smørgrav { 1101e19e6098SDag-Erling Smørgrav return (fetchXGetHTTP(URL, NULL, flags)); 11024ca1ab94SDag-Erling Smørgrav } 11034ca1ab94SDag-Erling Smørgrav 11041a5faa10SDag-Erling Smørgrav /* 11051a5faa10SDag-Erling Smørgrav * Store a file by HTTP 11061a5faa10SDag-Erling Smørgrav */ 11074ca1ab94SDag-Erling Smørgrav FILE * 1108f573a5fcSDag-Erling Smørgrav fetchPutHTTP(struct url *URL __unused, const char *flags __unused) 11094ca1ab94SDag-Erling Smørgrav { 11104ca1ab94SDag-Erling Smørgrav warnx("fetchPutHTTP(): not implemented"); 1111e19e6098SDag-Erling Smørgrav return (NULL); 11124ca1ab94SDag-Erling Smørgrav } 1113d8acd8dcSDag-Erling Smørgrav 1114d8acd8dcSDag-Erling Smørgrav /* 1115d8acd8dcSDag-Erling Smørgrav * Get an HTTP document's metadata 1116d8acd8dcSDag-Erling Smørgrav */ 1117d8acd8dcSDag-Erling Smørgrav int 111838c7e4a6SArchie Cobbs fetchStatHTTP(struct url *URL, struct url_stat *us, const char *flags) 1119d8acd8dcSDag-Erling Smørgrav { 112060245e42SDag-Erling Smørgrav FILE *f; 112160245e42SDag-Erling Smørgrav 11221a16ed4cSDag-Erling Smørgrav if ((f = _http_request(URL, "HEAD", us, _http_get_proxy(), flags)) == NULL) 1123e19e6098SDag-Erling Smørgrav return (-1); 1124f8f4130bSDag-Erling Smørgrav fclose(f); 1125e19e6098SDag-Erling Smørgrav return (0); 1126d8acd8dcSDag-Erling Smørgrav } 1127ce71b736SDag-Erling Smørgrav 1128ce71b736SDag-Erling Smørgrav /* 1129ce71b736SDag-Erling Smørgrav * List a directory 1130ce71b736SDag-Erling Smørgrav */ 1131ce71b736SDag-Erling Smørgrav struct url_ent * 1132f573a5fcSDag-Erling Smørgrav fetchListHTTP(struct url *url __unused, const char *flags __unused) 1133ce71b736SDag-Erling Smørgrav { 1134ce71b736SDag-Erling Smørgrav warnx("fetchListHTTP(): not implemented"); 1135e19e6098SDag-Erling Smørgrav return (NULL); 1136ce71b736SDag-Erling Smørgrav } 1137