http.c revision 242293
137535Sdes/*- 2236103Sdes * Copyright (c) 2000-2011 Dag-Erling Sm��rgrav 337535Sdes * All rights reserved. 437535Sdes * 537535Sdes * Redistribution and use in source and binary forms, with or without 637535Sdes * modification, are permitted provided that the following conditions 737535Sdes * are met: 837535Sdes * 1. Redistributions of source code must retain the above copyright 937535Sdes * notice, this list of conditions and the following disclaimer 1037535Sdes * in this position and unchanged. 1137535Sdes * 2. Redistributions in binary form must reproduce the above copyright 1237535Sdes * notice, this list of conditions and the following disclaimer in the 1337535Sdes * documentation and/or other materials provided with the distribution. 1437535Sdes * 3. The name of the author may not be used to endorse or promote products 1563012Sdes * derived from this software without specific prior written permission. 1637535Sdes * 1737535Sdes * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 1837535Sdes * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 1937535Sdes * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 2037535Sdes * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 2137535Sdes * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 2237535Sdes * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 2337535Sdes * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 2437535Sdes * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 2537535Sdes * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 2637535Sdes * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 2737535Sdes */ 2837535Sdes 2984203Sdillon#include <sys/cdefs.h> 3084203Sdillon__FBSDID("$FreeBSD: stable/9/lib/libfetch/http.c 242293 2012-10-29 04:18:34Z eadler $"); 3184203Sdillon 3263236Sdes/* 3363236Sdes * The following copyright applies to the base64 code: 3463236Sdes * 3563236Sdes *- 3663236Sdes * Copyright 1997 Massachusetts Institute of Technology 3763236Sdes * 3863236Sdes * Permission to use, copy, modify, and distribute this software and 3963236Sdes * its documentation for any purpose and without fee is hereby 4063236Sdes * granted, provided that both the above copyright notice and this 4163236Sdes * permission notice appear in all copies, that both the above 4263236Sdes * copyright notice and this permission notice appear in all 4363236Sdes * supporting documentation, and that the name of M.I.T. not be used 4463236Sdes * in advertising or publicity pertaining to distribution of the 4563236Sdes * software without specific, written prior permission. M.I.T. makes 4663236Sdes * no representations about the suitability of this software for any 4763236Sdes * purpose. It is provided "as is" without express or implied 4863236Sdes * warranty. 4990267Sdes * 5063236Sdes * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''. M.I.T. DISCLAIMS 5163236Sdes * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE, 5263236Sdes * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF 5363236Sdes * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT 5463236Sdes * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 5563236Sdes * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 5663236Sdes * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF 5763236Sdes * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND 5863236Sdes * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 5963236Sdes * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 6063236Sdes * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 6163236Sdes * SUCH DAMAGE. 6263236Sdes */ 6363236Sdes 6437535Sdes#include <sys/param.h> 6560737Sume#include <sys/socket.h> 66186124Smurray#include <sys/time.h> 6737535Sdes 6863012Sdes#include <ctype.h> 6937535Sdes#include <err.h> 7063012Sdes#include <errno.h> 7160376Sdes#include <locale.h> 7260189Sdes#include <netdb.h> 7337608Sdes#include <stdarg.h> 7437535Sdes#include <stdio.h> 7537535Sdes#include <stdlib.h> 7637535Sdes#include <string.h> 7760376Sdes#include <time.h> 7837535Sdes#include <unistd.h> 79202613Sdes#include <md5.h> 8037535Sdes 81141958Skbyanc#include <netinet/in.h> 82141958Skbyanc#include <netinet/tcp.h> 83141958Skbyanc 8437535Sdes#include "fetch.h" 8540939Sdes#include "common.h" 8641862Sdes#include "httperr.h" 8737535Sdes 8863012Sdes/* Maximum number of redirects to follow */ 89242032Seadler#define MAX_REDIRECT 20 9037535Sdes 9163012Sdes/* Symbolic names for reply codes we care about */ 9263012Sdes#define HTTP_OK 200 9363012Sdes#define HTTP_PARTIAL 206 9463012Sdes#define HTTP_MOVED_PERM 301 9563012Sdes#define HTTP_MOVED_TEMP 302 9663012Sdes#define HTTP_SEE_OTHER 303 97186124Smurray#define HTTP_NOT_MODIFIED 304 98242293Seadler#define HTTP_USE_PROXY 305 99169386Sdes#define HTTP_TEMP_REDIRECT 307 100242289Seadler#define HTTP_PERM_REDIRECT 308 10163012Sdes#define HTTP_NEED_AUTH 401 10287317Sdes#define HTTP_NEED_PROXY_AUTH 407 103125696Sdes#define HTTP_BAD_RANGE 416 10463012Sdes#define HTTP_PROTOCOL_ERROR 999 10560196Sdes 10663012Sdes#define HTTP_REDIRECT(xyz) ((xyz) == HTTP_MOVED_PERM \ 10790267Sdes || (xyz) == HTTP_MOVED_TEMP \ 108169386Sdes || (xyz) == HTTP_TEMP_REDIRECT \ 109242293Seadler || (xyz) == HTTP_USE_PROXY \ 11090267Sdes || (xyz) == HTTP_SEE_OTHER) 11163012Sdes 11288771Sdes#define HTTP_ERROR(xyz) ((xyz) > 400 && (xyz) < 599) 11363012Sdes 11490267Sdes 11563012Sdes/***************************************************************************** 11663012Sdes * I/O functions for decoding chunked streams 11763012Sdes */ 11863012Sdes 11997859Sdesstruct httpio 12037535Sdes{ 12197858Sdes conn_t *conn; /* connection */ 12297866Sdes int chunked; /* chunked mode */ 12397858Sdes char *buf; /* chunk buffer */ 12497866Sdes size_t bufsize; /* size of chunk buffer */ 12597866Sdes ssize_t buflen; /* amount of data currently in buffer */ 12697866Sdes int bufpos; /* current read offset in buffer */ 12797858Sdes int eof; /* end-of-file flag */ 12897858Sdes int error; /* error flag */ 12997858Sdes size_t chunksize; /* remaining size of current chunk */ 13063281Sdes#ifndef NDEBUG 13190267Sdes size_t total; 13263012Sdes#endif 13337535Sdes}; 13437535Sdes 13537608Sdes/* 13663012Sdes * Get next chunk header 13737608Sdes */ 13837608Sdesstatic int 139174588Sdeshttp_new_chunk(struct httpio *io) 14037608Sdes{ 14190267Sdes char *p; 14290267Sdes 143174588Sdes if (fetch_getln(io->conn) == -1) 14490267Sdes return (-1); 14590267Sdes 146174761Sdes if (io->conn->buflen < 2 || !isxdigit((unsigned char)*io->conn->buf)) 14790267Sdes return (-1); 14890267Sdes 149174761Sdes for (p = io->conn->buf; *p && !isspace((unsigned char)*p); ++p) { 15090267Sdes if (*p == ';') 15190267Sdes break; 152174761Sdes if (!isxdigit((unsigned char)*p)) 15390267Sdes return (-1); 154174761Sdes if (isdigit((unsigned char)*p)) { 15597859Sdes io->chunksize = io->chunksize * 16 + 15690267Sdes *p - '0'; 15790267Sdes } else { 15897859Sdes io->chunksize = io->chunksize * 16 + 159176036Sdes 10 + tolower((unsigned char)*p) - 'a'; 16090267Sdes } 16190267Sdes } 16290267Sdes 16363281Sdes#ifndef NDEBUG 16490267Sdes if (fetchDebug) { 16597859Sdes io->total += io->chunksize; 16697859Sdes if (io->chunksize == 0) 167106207Sdes fprintf(stderr, "%s(): end of last chunk\n", __func__); 16890267Sdes else 169106207Sdes fprintf(stderr, "%s(): new chunk: %lu (%lu)\n", 170106207Sdes __func__, (unsigned long)io->chunksize, 171106207Sdes (unsigned long)io->total); 17290267Sdes } 17363012Sdes#endif 17490267Sdes 17597859Sdes return (io->chunksize); 17637608Sdes} 17737608Sdes 17837608Sdes/* 17997866Sdes * Grow the input buffer to at least len bytes 18097866Sdes */ 18197866Sdesstatic inline int 182174588Sdeshttp_growbuf(struct httpio *io, size_t len) 18397866Sdes{ 18497866Sdes char *tmp; 18597866Sdes 18697866Sdes if (io->bufsize >= len) 18797866Sdes return (0); 18897866Sdes 18997866Sdes if ((tmp = realloc(io->buf, len)) == NULL) 19097866Sdes return (-1); 19197866Sdes io->buf = tmp; 19297866Sdes io->bufsize = len; 193106044Sdes return (0); 19497866Sdes} 19597866Sdes 19697866Sdes/* 19737608Sdes * Fill the input buffer, do chunk decoding on the fly 19837608Sdes */ 19963012Sdesstatic int 200174588Sdeshttp_fillbuf(struct httpio *io, size_t len) 20137535Sdes{ 202231247Sbapt ssize_t nbytes; 203231247Sbapt 20497859Sdes if (io->error) 20590267Sdes return (-1); 20697859Sdes if (io->eof) 20790267Sdes return (0); 20890267Sdes 20997866Sdes if (io->chunked == 0) { 210174588Sdes if (http_growbuf(io, len) == -1) 21197866Sdes return (-1); 212231247Sbapt if ((nbytes = fetch_read(io->conn, io->buf, len)) == -1) { 213231247Sbapt io->error = errno; 21497866Sdes return (-1); 215106185Sdes } 216231247Sbapt io->buflen = nbytes; 21797866Sdes io->bufpos = 0; 21897866Sdes return (io->buflen); 21997866Sdes } 22097866Sdes 22197859Sdes if (io->chunksize == 0) { 222174588Sdes switch (http_new_chunk(io)) { 22390267Sdes case -1: 22497859Sdes io->error = 1; 22590267Sdes return (-1); 22690267Sdes case 0: 22797859Sdes io->eof = 1; 22890267Sdes return (0); 22990267Sdes } 23037535Sdes } 23163012Sdes 23297866Sdes if (len > io->chunksize) 23397866Sdes len = io->chunksize; 234174588Sdes if (http_growbuf(io, len) == -1) 23590267Sdes return (-1); 236231247Sbapt if ((nbytes = fetch_read(io->conn, io->buf, len)) == -1) { 237231247Sbapt io->error = errno; 23897866Sdes return (-1); 239106185Sdes } 240231247Sbapt io->buflen = nbytes; 24197866Sdes io->chunksize -= io->buflen; 24290267Sdes 24397859Sdes if (io->chunksize == 0) { 24497856Sdes char endl[2]; 24597856Sdes 246174588Sdes if (fetch_read(io->conn, endl, 2) != 2 || 24797856Sdes endl[0] != '\r' || endl[1] != '\n') 24890267Sdes return (-1); 24990267Sdes } 25090267Sdes 25197866Sdes io->bufpos = 0; 25290267Sdes 25397866Sdes return (io->buflen); 25437535Sdes} 25537535Sdes 25637608Sdes/* 25737608Sdes * Read function 25837608Sdes */ 25937535Sdesstatic int 260174588Sdeshttp_readfn(void *v, char *buf, int len) 26137535Sdes{ 26297859Sdes struct httpio *io = (struct httpio *)v; 26390267Sdes int l, pos; 26463012Sdes 26597859Sdes if (io->error) 26690267Sdes return (-1); 26797859Sdes if (io->eof) 26890267Sdes return (0); 26963012Sdes 27090267Sdes for (pos = 0; len > 0; pos += l, len -= l) { 27190267Sdes /* empty buffer */ 27297866Sdes if (!io->buf || io->bufpos == io->buflen) 273174588Sdes if (http_fillbuf(io, len) < 1) 27490267Sdes break; 27597866Sdes l = io->buflen - io->bufpos; 27690267Sdes if (len < l) 27790267Sdes l = len; 278176105Sdes memcpy(buf + pos, io->buf + io->bufpos, l); 27997866Sdes io->bufpos += l; 28090267Sdes } 28137535Sdes 282231247Sbapt if (!pos && io->error) { 283231247Sbapt if (io->error == EINTR) 284231247Sbapt io->error = 0; 28590267Sdes return (-1); 286231247Sbapt } 28790267Sdes return (pos); 28837535Sdes} 28937535Sdes 29037608Sdes/* 29137608Sdes * Write function 29237608Sdes */ 29337535Sdesstatic int 294174588Sdeshttp_writefn(void *v, const char *buf, int len) 29537535Sdes{ 29697859Sdes struct httpio *io = (struct httpio *)v; 29790267Sdes 298174588Sdes return (fetch_write(io->conn, buf, len)); 29937535Sdes} 30037535Sdes 30137608Sdes/* 30237608Sdes * Close function 30337608Sdes */ 30437535Sdesstatic int 305174588Sdeshttp_closefn(void *v) 30637535Sdes{ 30797859Sdes struct httpio *io = (struct httpio *)v; 30890267Sdes int r; 30963012Sdes 310174588Sdes r = fetch_close(io->conn); 31197859Sdes if (io->buf) 31297859Sdes free(io->buf); 31397859Sdes free(io); 31490267Sdes return (r); 31537535Sdes} 31637535Sdes 31737608Sdes/* 31863012Sdes * Wrap a file descriptor up 31937608Sdes */ 32063012Sdesstatic FILE * 321174588Sdeshttp_funopen(conn_t *conn, int chunked) 32237535Sdes{ 32397859Sdes struct httpio *io; 32490267Sdes FILE *f; 32563012Sdes 326109967Sdes if ((io = calloc(1, sizeof(*io))) == NULL) { 327174588Sdes fetch_syserr(); 32890267Sdes return (NULL); 32990267Sdes } 33097859Sdes io->conn = conn; 33197866Sdes io->chunked = chunked; 332174588Sdes f = funopen(io, http_readfn, http_writefn, NULL, http_closefn); 33390267Sdes if (f == NULL) { 334174588Sdes fetch_syserr(); 33597859Sdes free(io); 33690267Sdes return (NULL); 33790267Sdes } 33890267Sdes return (f); 33963012Sdes} 34063012Sdes 34190267Sdes 34263012Sdes/***************************************************************************** 34363012Sdes * Helper functions for talking to the server and parsing its replies 34463012Sdes */ 34563012Sdes 34663012Sdes/* Header types */ 34763012Sdestypedef enum { 34890267Sdes hdr_syserror = -2, 34990267Sdes hdr_error = -1, 35090267Sdes hdr_end = 0, 35190267Sdes hdr_unknown = 1, 35290267Sdes hdr_content_length, 35390267Sdes hdr_content_range, 35490267Sdes hdr_last_modified, 35590267Sdes hdr_location, 35690267Sdes hdr_transfer_encoding, 357202613Sdes hdr_www_authenticate, 358202613Sdes hdr_proxy_authenticate, 35985093Sdes} hdr_t; 36063012Sdes 36163012Sdes/* Names of interesting headers */ 36263012Sdesstatic struct { 36390267Sdes hdr_t num; 36490267Sdes const char *name; 36563012Sdes} hdr_names[] = { 36690267Sdes { hdr_content_length, "Content-Length" }, 36790267Sdes { hdr_content_range, "Content-Range" }, 36890267Sdes { hdr_last_modified, "Last-Modified" }, 36990267Sdes { hdr_location, "Location" }, 37090267Sdes { hdr_transfer_encoding, "Transfer-Encoding" }, 37190267Sdes { hdr_www_authenticate, "WWW-Authenticate" }, 372202613Sdes { hdr_proxy_authenticate, "Proxy-Authenticate" }, 37390267Sdes { hdr_unknown, NULL }, 37463012Sdes}; 37563012Sdes 37663012Sdes/* 37763012Sdes * Send a formatted line; optionally echo to terminal 37863012Sdes */ 37963012Sdesstatic int 380174588Sdeshttp_cmd(conn_t *conn, const char *fmt, ...) 38163012Sdes{ 38290267Sdes va_list ap; 38390267Sdes size_t len; 38490267Sdes char *msg; 38590267Sdes int r; 38663012Sdes 38790267Sdes va_start(ap, fmt); 38890267Sdes len = vasprintf(&msg, fmt, ap); 38990267Sdes va_end(ap); 39090267Sdes 39190267Sdes if (msg == NULL) { 39290267Sdes errno = ENOMEM; 393174588Sdes fetch_syserr(); 39490267Sdes return (-1); 39590267Sdes } 39690267Sdes 397174588Sdes r = fetch_putln(conn, msg, len); 39890267Sdes free(msg); 39990267Sdes 40090267Sdes if (r == -1) { 401174588Sdes fetch_syserr(); 40290267Sdes return (-1); 40390267Sdes } 40490267Sdes 40590267Sdes return (0); 40663012Sdes} 40763012Sdes 40863012Sdes/* 40963012Sdes * Get and parse status line 41063012Sdes */ 41163012Sdesstatic int 412174588Sdeshttp_get_reply(conn_t *conn) 41363012Sdes{ 41490267Sdes char *p; 41590267Sdes 416174588Sdes if (fetch_getln(conn) == -1) 41790267Sdes return (-1); 41890267Sdes /* 41990267Sdes * A valid status line looks like "HTTP/m.n xyz reason" where m 42090267Sdes * and n are the major and minor protocol version numbers and xyz 42190267Sdes * is the reply code. 42290267Sdes * Unfortunately, there are servers out there (NCSA 1.5.1, to name 42390267Sdes * just one) that do not send a version number, so we can't rely 42490267Sdes * on finding one, but if we do, insist on it being 1.0 or 1.1. 42590267Sdes * We don't care about the reason phrase. 42690267Sdes */ 42797856Sdes if (strncmp(conn->buf, "HTTP", 4) != 0) 42890267Sdes return (HTTP_PROTOCOL_ERROR); 42997856Sdes p = conn->buf + 4; 43090267Sdes if (*p == '/') { 43190267Sdes if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1')) 43290267Sdes return (HTTP_PROTOCOL_ERROR); 43390267Sdes p += 4; 43490267Sdes } 435174761Sdes if (*p != ' ' || 436174761Sdes !isdigit((unsigned char)p[1]) || 437174761Sdes !isdigit((unsigned char)p[2]) || 438174761Sdes !isdigit((unsigned char)p[3])) 43990267Sdes return (HTTP_PROTOCOL_ERROR); 44090267Sdes 44197856Sdes conn->err = (p[1] - '0') * 100 + (p[2] - '0') * 10 + (p[3] - '0'); 44297856Sdes return (conn->err); 44337535Sdes} 44437535Sdes 44537608Sdes/* 44690267Sdes * Check a header; if the type matches the given string, return a pointer 44790267Sdes * to the beginning of the value. 44863012Sdes */ 44975891Sarchiestatic const char * 450174588Sdeshttp_match(const char *str, const char *hdr) 45163012Sdes{ 452176036Sdes while (*str && *hdr && 453176036Sdes tolower((unsigned char)*str++) == tolower((unsigned char)*hdr++)) 45490267Sdes /* nothing */; 45590267Sdes if (*str || *hdr != ':') 45690267Sdes return (NULL); 457174761Sdes while (*hdr && isspace((unsigned char)*++hdr)) 45890267Sdes /* nothing */; 45990267Sdes return (hdr); 46063012Sdes} 46163012Sdes 462202613Sdes 46363012Sdes/* 464202613Sdes * Get the next header and return the appropriate symbolic code. We 465202613Sdes * need to read one line ahead for checking for a continuation line 466202613Sdes * belonging to the current header (continuation lines start with 467221821Sdes * white space). 468202613Sdes * 469202613Sdes * We get called with a fresh line already in the conn buffer, either 470202613Sdes * from the previous http_next_header() invocation, or, the first 471202613Sdes * time, from a fetch_getln() performed by our caller. 472202613Sdes * 473202613Sdes * This stops when we encounter an empty line (we dont read beyond the header 474202613Sdes * area). 475221821Sdes * 476202613Sdes * Note that the "headerbuf" is just a place to return the result. Its 477202613Sdes * contents are not used for the next call. This means that no cleanup 478202613Sdes * is needed when ie doing another connection, just call the cleanup when 479202613Sdes * fully done to deallocate memory. 48063012Sdes */ 481202613Sdes 482202613Sdes/* Limit the max number of continuation lines to some reasonable value */ 483202613Sdes#define HTTP_MAX_CONT_LINES 10 484202613Sdes 485202613Sdes/* Place into which to build a header from one or several lines */ 486202613Sdestypedef struct { 487202613Sdes char *buf; /* buffer */ 488202613Sdes size_t bufsize; /* buffer size */ 489202613Sdes size_t buflen; /* length of buffer contents */ 490202613Sdes} http_headerbuf_t; 491202613Sdes 492202613Sdesstatic void 493202613Sdesinit_http_headerbuf(http_headerbuf_t *buf) 49463012Sdes{ 495202613Sdes buf->buf = NULL; 496202613Sdes buf->bufsize = 0; 497202613Sdes buf->buflen = 0; 498202613Sdes} 49990267Sdes 500221821Sdesstatic void 501202613Sdesclean_http_headerbuf(http_headerbuf_t *buf) 502202613Sdes{ 503202613Sdes if (buf->buf) 504202613Sdes free(buf->buf); 505202613Sdes init_http_headerbuf(buf); 506202613Sdes} 507202613Sdes 508202613Sdes/* Remove whitespace at the end of the buffer */ 509221821Sdesstatic void 510202613Sdeshttp_conn_trimright(conn_t *conn) 511202613Sdes{ 512221821Sdes while (conn->buflen && 513202613Sdes isspace((unsigned char)conn->buf[conn->buflen - 1])) 51497856Sdes conn->buflen--; 51597856Sdes conn->buf[conn->buflen] = '\0'; 516202613Sdes} 517202613Sdes 518202613Sdesstatic hdr_t 519202613Sdeshttp_next_header(conn_t *conn, http_headerbuf_t *hbuf, const char **p) 520202613Sdes{ 521221820Sdes unsigned int i, len; 522202613Sdes 523221821Sdes /* 524202613Sdes * Have to do the stripping here because of the first line. So 525221821Sdes * it's done twice for the subsequent lines. No big deal 526202613Sdes */ 527202613Sdes http_conn_trimright(conn); 52897856Sdes if (conn->buflen == 0) 52997856Sdes return (hdr_end); 530202613Sdes 531202613Sdes /* Copy the line to the headerbuf */ 532202613Sdes if (hbuf->bufsize < conn->buflen + 1) { 533202613Sdes if ((hbuf->buf = realloc(hbuf->buf, conn->buflen + 1)) == NULL) 534202613Sdes return (hdr_syserror); 535202613Sdes hbuf->bufsize = conn->buflen + 1; 536202613Sdes } 537202613Sdes strcpy(hbuf->buf, conn->buf); 538202613Sdes hbuf->buflen = conn->buflen; 539202613Sdes 540221821Sdes /* 541202613Sdes * Fetch possible continuation lines. Stop at 1st non-continuation 542221821Sdes * and leave it in the conn buffer 543221821Sdes */ 544202613Sdes for (i = 0; i < HTTP_MAX_CONT_LINES; i++) { 545202613Sdes if (fetch_getln(conn) == -1) 546202613Sdes return (hdr_syserror); 547202613Sdes 548221821Sdes /* 549202613Sdes * Note: we carry on the idea from the previous version 550202613Sdes * that a pure whitespace line is equivalent to an empty 551202613Sdes * one (so it's not continuation and will be handled when 552221821Sdes * we are called next) 553202613Sdes */ 554202613Sdes http_conn_trimright(conn); 555202613Sdes if (conn->buf[0] != ' ' && conn->buf[0] != "\t"[0]) 556202613Sdes break; 557202613Sdes 558202613Sdes /* Got a continuation line. Concatenate to previous */ 559202613Sdes len = hbuf->buflen + conn->buflen; 560202613Sdes if (hbuf->bufsize < len + 1) { 561202613Sdes len *= 2; 562202613Sdes if ((hbuf->buf = realloc(hbuf->buf, len + 1)) == NULL) 563202613Sdes return (hdr_syserror); 564202613Sdes hbuf->bufsize = len + 1; 565202613Sdes } 566202613Sdes strcpy(hbuf->buf + hbuf->buflen, conn->buf); 567202613Sdes hbuf->buflen += conn->buflen; 568221821Sdes } 569202613Sdes 57090267Sdes /* 57190267Sdes * We could check for malformed headers but we don't really care. 57290267Sdes * A valid header starts with a token immediately followed by a 57390267Sdes * colon; a token is any sequence of non-control, non-whitespace 57490267Sdes * characters except "()<>@,;:\\\"{}". 57590267Sdes */ 57690267Sdes for (i = 0; hdr_names[i].num != hdr_unknown; i++) 577202613Sdes if ((*p = http_match(hdr_names[i].name, hbuf->buf)) != NULL) 57890267Sdes return (hdr_names[i].num); 579202613Sdes 58090267Sdes return (hdr_unknown); 58163012Sdes} 58263012Sdes 583202613Sdes/************************** 584202613Sdes * [Proxy-]Authenticate header parsing 585202613Sdes */ 586202613Sdes 587221821Sdes/* 588221821Sdes * Read doublequote-delimited string into output buffer obuf (allocated 589202613Sdes * by caller, whose responsibility it is to ensure that it's big enough) 590202613Sdes * cp points to the first char after the initial '"' 591221821Sdes * Handles \ quoting 592221821Sdes * Returns pointer to the first char after the terminating double quote, or 593202613Sdes * NULL for error. 594202613Sdes */ 595202613Sdesstatic const char * 596202613Sdeshttp_parse_headerstring(const char *cp, char *obuf) 597202613Sdes{ 598202613Sdes for (;;) { 599202613Sdes switch (*cp) { 600202613Sdes case 0: /* Unterminated string */ 601202613Sdes *obuf = 0; 602202613Sdes return (NULL); 603202613Sdes case '"': /* Ending quote */ 604202613Sdes *obuf = 0; 605202613Sdes return (++cp); 606202613Sdes case '\\': 607202613Sdes if (*++cp == 0) { 608202613Sdes *obuf = 0; 609202613Sdes return (NULL); 610202613Sdes } 611202613Sdes /* FALLTHROUGH */ 612202613Sdes default: 613202613Sdes *obuf++ = *cp++; 614202613Sdes } 615202613Sdes } 616202613Sdes} 617202613Sdes 618202613Sdes/* Http auth challenge schemes */ 619202613Sdestypedef enum {HTTPAS_UNKNOWN, HTTPAS_BASIC,HTTPAS_DIGEST} http_auth_schemes_t; 620202613Sdes 621202613Sdes/* Data holder for a Basic or Digest challenge. */ 622202613Sdestypedef struct { 623202613Sdes http_auth_schemes_t scheme; 624202613Sdes char *realm; 625202613Sdes char *qop; 626202613Sdes char *nonce; 627202613Sdes char *opaque; 628202613Sdes char *algo; 629202613Sdes int stale; 630202613Sdes int nc; /* Nonce count */ 631202613Sdes} http_auth_challenge_t; 632202613Sdes 633221821Sdesstatic void 634202613Sdesinit_http_auth_challenge(http_auth_challenge_t *b) 635202613Sdes{ 636202613Sdes b->scheme = HTTPAS_UNKNOWN; 637202613Sdes b->realm = b->qop = b->nonce = b->opaque = b->algo = NULL; 638202613Sdes b->stale = b->nc = 0; 639202613Sdes} 640202613Sdes 641221821Sdesstatic void 642202613Sdesclean_http_auth_challenge(http_auth_challenge_t *b) 643202613Sdes{ 644221821Sdes if (b->realm) 645202613Sdes free(b->realm); 646221821Sdes if (b->qop) 647202613Sdes free(b->qop); 648221821Sdes if (b->nonce) 649202613Sdes free(b->nonce); 650221821Sdes if (b->opaque) 651202613Sdes free(b->opaque); 652221821Sdes if (b->algo) 653202613Sdes free(b->algo); 654202613Sdes init_http_auth_challenge(b); 655202613Sdes} 656202613Sdes 657202613Sdes/* Data holder for an array of challenges offered in an http response. */ 658202613Sdes#define MAX_CHALLENGES 10 659202613Sdestypedef struct { 660202613Sdes http_auth_challenge_t *challenges[MAX_CHALLENGES]; 661202613Sdes int count; /* Number of parsed challenges in the array */ 662202613Sdes int valid; /* We did parse an authenticate header */ 663202613Sdes} http_auth_challenges_t; 664202613Sdes 665221821Sdesstatic void 666202613Sdesinit_http_auth_challenges(http_auth_challenges_t *cs) 667202613Sdes{ 668202613Sdes int i; 669202613Sdes for (i = 0; i < MAX_CHALLENGES; i++) 670202613Sdes cs->challenges[i] = NULL; 671202613Sdes cs->count = cs->valid = 0; 672202613Sdes} 673202613Sdes 674221821Sdesstatic void 675202613Sdesclean_http_auth_challenges(http_auth_challenges_t *cs) 676202613Sdes{ 677202613Sdes int i; 678202613Sdes /* We rely on non-zero pointers being allocated, not on the count */ 679202613Sdes for (i = 0; i < MAX_CHALLENGES; i++) { 680202613Sdes if (cs->challenges[i] != NULL) { 681202613Sdes clean_http_auth_challenge(cs->challenges[i]); 682202613Sdes free(cs->challenges[i]); 683202613Sdes } 684202613Sdes } 685202613Sdes init_http_auth_challenges(cs); 686202613Sdes} 687202613Sdes 688221821Sdes/* 689202613Sdes * Enumeration for lexical elements. Separators will be returned as their own 690202613Sdes * ascii value 691202613Sdes */ 692202613Sdestypedef enum {HTTPHL_WORD=256, HTTPHL_STRING=257, HTTPHL_END=258, 693202613Sdes HTTPHL_ERROR = 259} http_header_lex_t; 694202613Sdes 695221821Sdes/* 696202613Sdes * Determine what kind of token comes next and return possible value 697202613Sdes * in buf, which is supposed to have been allocated big enough by 698221821Sdes * caller. Advance input pointer and return element type. 699202613Sdes */ 700221821Sdesstatic int 701202613Sdeshttp_header_lex(const char **cpp, char *buf) 702202613Sdes{ 703202613Sdes size_t l; 704202613Sdes /* Eat initial whitespace */ 705202613Sdes *cpp += strspn(*cpp, " \t"); 706202613Sdes if (**cpp == 0) 707202613Sdes return (HTTPHL_END); 708202613Sdes 709202613Sdes /* Separator ? */ 710202613Sdes if (**cpp == ',' || **cpp == '=') 711202613Sdes return (*((*cpp)++)); 712202613Sdes 713202613Sdes /* String ? */ 714202613Sdes if (**cpp == '"') { 715202613Sdes *cpp = http_parse_headerstring(++*cpp, buf); 716202613Sdes if (*cpp == NULL) 717202613Sdes return (HTTPHL_ERROR); 718202613Sdes return (HTTPHL_STRING); 719202613Sdes } 720202613Sdes 721202613Sdes /* Read other token, until separator or whitespace */ 722202613Sdes l = strcspn(*cpp, " \t,="); 723202613Sdes memcpy(buf, *cpp, l); 724202613Sdes buf[l] = 0; 725202613Sdes *cpp += l; 726202613Sdes return (HTTPHL_WORD); 727202613Sdes} 728202613Sdes 729221821Sdes/* 730202613Sdes * Read challenges from http xxx-authenticate header and accumulate them 731202613Sdes * in the challenges list structure. 732202613Sdes * 733202613Sdes * Headers with multiple challenges are specified by rfc2617, but 734202613Sdes * servers (ie: squid) often send them in separate headers instead, 735202613Sdes * which in turn is forbidden by the http spec (multiple headers with 736202613Sdes * the same name are only allowed for pure comma-separated lists, see 737202613Sdes * rfc2616 sec 4.2). 738202613Sdes * 739202613Sdes * We support both approaches anyway 740202613Sdes */ 741221821Sdesstatic int 742202613Sdeshttp_parse_authenticate(const char *cp, http_auth_challenges_t *cs) 743202613Sdes{ 744202613Sdes int ret = -1; 745202613Sdes http_header_lex_t lex; 746202613Sdes char *key = malloc(strlen(cp) + 1); 747202613Sdes char *value = malloc(strlen(cp) + 1); 748202613Sdes char *buf = malloc(strlen(cp) + 1); 749202613Sdes 750202613Sdes if (key == NULL || value == NULL || buf == NULL) { 751202613Sdes fetch_syserr(); 752202613Sdes goto out; 753202613Sdes } 754202613Sdes 755202613Sdes /* In any case we've seen the header and we set the valid bit */ 756202613Sdes cs->valid = 1; 757202613Sdes 758202613Sdes /* Need word first */ 759202613Sdes lex = http_header_lex(&cp, key); 760202613Sdes if (lex != HTTPHL_WORD) 761202613Sdes goto out; 762202613Sdes 763202613Sdes /* Loop on challenges */ 764202613Sdes for (; cs->count < MAX_CHALLENGES; cs->count++) { 765221821Sdes cs->challenges[cs->count] = 766202613Sdes malloc(sizeof(http_auth_challenge_t)); 767202613Sdes if (cs->challenges[cs->count] == NULL) { 768202613Sdes fetch_syserr(); 769202613Sdes goto out; 770202613Sdes } 771202613Sdes init_http_auth_challenge(cs->challenges[cs->count]); 772202613Sdes if (!strcasecmp(key, "basic")) { 773202613Sdes cs->challenges[cs->count]->scheme = HTTPAS_BASIC; 774202613Sdes } else if (!strcasecmp(key, "digest")) { 775202613Sdes cs->challenges[cs->count]->scheme = HTTPAS_DIGEST; 776202613Sdes } else { 777202613Sdes cs->challenges[cs->count]->scheme = HTTPAS_UNKNOWN; 778221821Sdes /* 779221821Sdes * Continue parsing as basic or digest may 780202613Sdes * follow, and the syntax is the same for 781202613Sdes * all. We'll just ignore this one when 782202613Sdes * looking at the list 783202613Sdes */ 784202613Sdes } 785221821Sdes 786202613Sdes /* Loop on attributes */ 787202613Sdes for (;;) { 788202613Sdes /* Key */ 789202613Sdes lex = http_header_lex(&cp, key); 790202613Sdes if (lex != HTTPHL_WORD) 791202613Sdes goto out; 792202613Sdes 793202613Sdes /* Equal sign */ 794202613Sdes lex = http_header_lex(&cp, buf); 795202613Sdes if (lex != '=') 796202613Sdes goto out; 797202613Sdes 798202613Sdes /* Value */ 799202613Sdes lex = http_header_lex(&cp, value); 800202613Sdes if (lex != HTTPHL_WORD && lex != HTTPHL_STRING) 801202613Sdes goto out; 802202613Sdes 803202613Sdes if (!strcasecmp(key, "realm")) 804221821Sdes cs->challenges[cs->count]->realm = 805202613Sdes strdup(value); 806202613Sdes else if (!strcasecmp(key, "qop")) 807221821Sdes cs->challenges[cs->count]->qop = 808202613Sdes strdup(value); 809202613Sdes else if (!strcasecmp(key, "nonce")) 810221821Sdes cs->challenges[cs->count]->nonce = 811202613Sdes strdup(value); 812202613Sdes else if (!strcasecmp(key, "opaque")) 813221821Sdes cs->challenges[cs->count]->opaque = 814202613Sdes strdup(value); 815202613Sdes else if (!strcasecmp(key, "algorithm")) 816221821Sdes cs->challenges[cs->count]->algo = 817202613Sdes strdup(value); 818202613Sdes else if (!strcasecmp(key, "stale")) 819221821Sdes cs->challenges[cs->count]->stale = 820202613Sdes strcasecmp(value, "no"); 821202613Sdes /* Else ignore unknown attributes */ 822202613Sdes 823202613Sdes /* Comma or Next challenge or End */ 824202613Sdes lex = http_header_lex(&cp, key); 825221821Sdes /* 826221821Sdes * If we get a word here, this is the beginning of the 827221821Sdes * next challenge. Break the attributes loop 828221821Sdes */ 829202613Sdes if (lex == HTTPHL_WORD) 830202613Sdes break; 831202613Sdes 832202613Sdes if (lex == HTTPHL_END) { 833202613Sdes /* End while looking for ',' is normal exit */ 834202613Sdes cs->count++; 835202613Sdes ret = 0; 836202613Sdes goto out; 837202613Sdes } 838202613Sdes /* Anything else is an error */ 839202613Sdes if (lex != ',') 840202613Sdes goto out; 841202613Sdes 842202613Sdes } /* End attributes loop */ 843202613Sdes } /* End challenge loop */ 844202613Sdes 845221821Sdes /* 846221821Sdes * Challenges max count exceeded. This really can't happen 847221821Sdes * with normal data, something's fishy -> error 848221821Sdes */ 849202613Sdes 850202613Sdesout: 851202613Sdes if (key) 852202613Sdes free(key); 853202613Sdes if (value) 854202613Sdes free(value); 855202613Sdes if (buf) 856202613Sdes free(buf); 857202613Sdes return (ret); 858202613Sdes} 859202613Sdes 860202613Sdes 86163012Sdes/* 86263012Sdes * Parse a last-modified header 86363012Sdes */ 86463716Sdesstatic int 865174588Sdeshttp_parse_mtime(const char *p, time_t *mtime) 86663012Sdes{ 86790267Sdes char locale[64], *r; 86890267Sdes struct tm tm; 86963012Sdes 870109967Sdes strncpy(locale, setlocale(LC_TIME, NULL), sizeof(locale)); 87190267Sdes setlocale(LC_TIME, "C"); 87290267Sdes r = strptime(p, "%a, %d %b %Y %H:%M:%S GMT", &tm); 87390267Sdes /* XXX should add support for date-2 and date-3 */ 87490267Sdes setlocale(LC_TIME, locale); 87590267Sdes if (r == NULL) 87690267Sdes return (-1); 87790267Sdes DEBUG(fprintf(stderr, "last modified: [%04d-%02d-%02d " 87888769Sdes "%02d:%02d:%02d]\n", 87963012Sdes tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday, 88063012Sdes tm.tm_hour, tm.tm_min, tm.tm_sec)); 88190267Sdes *mtime = timegm(&tm); 88290267Sdes return (0); 88363012Sdes} 88463012Sdes 88563012Sdes/* 88663012Sdes * Parse a content-length header 88763012Sdes */ 88863716Sdesstatic int 889174588Sdeshttp_parse_length(const char *p, off_t *length) 89063012Sdes{ 89190267Sdes off_t len; 89290267Sdes 893174761Sdes for (len = 0; *p && isdigit((unsigned char)*p); ++p) 89490267Sdes len = len * 10 + (*p - '0'); 89590267Sdes if (*p) 89690267Sdes return (-1); 89790267Sdes DEBUG(fprintf(stderr, "content length: [%lld]\n", 89890267Sdes (long long)len)); 89990267Sdes *length = len; 90090267Sdes return (0); 90163012Sdes} 90263012Sdes 90363012Sdes/* 90463012Sdes * Parse a content-range header 90563012Sdes */ 90663716Sdesstatic int 907174588Sdeshttp_parse_range(const char *p, off_t *offset, off_t *length, off_t *size) 90863012Sdes{ 90990267Sdes off_t first, last, len; 91063716Sdes 91190267Sdes if (strncasecmp(p, "bytes ", 6) != 0) 91290267Sdes return (-1); 913125696Sdes p += 6; 914125696Sdes if (*p == '*') { 915125696Sdes first = last = -1; 916125696Sdes ++p; 917125696Sdes } else { 918174761Sdes for (first = 0; *p && isdigit((unsigned char)*p); ++p) 919125696Sdes first = first * 10 + *p - '0'; 920125696Sdes if (*p != '-') 921125696Sdes return (-1); 922174761Sdes for (last = 0, ++p; *p && isdigit((unsigned char)*p); ++p) 923125696Sdes last = last * 10 + *p - '0'; 924125696Sdes } 92590267Sdes if (first > last || *p != '/') 92690267Sdes return (-1); 927174761Sdes for (len = 0, ++p; *p && isdigit((unsigned char)*p); ++p) 92890267Sdes len = len * 10 + *p - '0'; 92990267Sdes if (*p || len < last - first + 1) 93090267Sdes return (-1); 931125696Sdes if (first == -1) { 932125696Sdes DEBUG(fprintf(stderr, "content range: [*/%lld]\n", 933125696Sdes (long long)len)); 934125696Sdes *length = 0; 935125696Sdes } else { 936125696Sdes DEBUG(fprintf(stderr, "content range: [%lld-%lld/%lld]\n", 937125696Sdes (long long)first, (long long)last, (long long)len)); 938125696Sdes *length = last - first + 1; 939125696Sdes } 94090267Sdes *offset = first; 94190267Sdes *size = len; 94290267Sdes return (0); 94363012Sdes} 94463012Sdes 94590267Sdes 94663012Sdes/***************************************************************************** 94763012Sdes * Helper functions for authorization 94863012Sdes */ 94963012Sdes 95063012Sdes/* 95137608Sdes * Base64 encoding 95237608Sdes */ 95362965Sdesstatic char * 954174588Sdeshttp_base64(const char *src) 95537608Sdes{ 95690267Sdes static const char base64[] = 95790267Sdes "ABCDEFGHIJKLMNOPQRSTUVWXYZ" 95890267Sdes "abcdefghijklmnopqrstuvwxyz" 95990267Sdes "0123456789+/"; 96090267Sdes char *str, *dst; 96190267Sdes size_t l; 96290267Sdes int t, r; 96362965Sdes 96490267Sdes l = strlen(src); 965133280Sdes if ((str = malloc(((l + 2) / 3) * 4 + 1)) == NULL) 96690267Sdes return (NULL); 96790267Sdes dst = str; 96890267Sdes r = 0; 96937608Sdes 97090267Sdes while (l >= 3) { 97190267Sdes t = (src[0] << 16) | (src[1] << 8) | src[2]; 97290267Sdes dst[0] = base64[(t >> 18) & 0x3f]; 97390267Sdes dst[1] = base64[(t >> 12) & 0x3f]; 97490267Sdes dst[2] = base64[(t >> 6) & 0x3f]; 97590267Sdes dst[3] = base64[(t >> 0) & 0x3f]; 97690267Sdes src += 3; l -= 3; 97790267Sdes dst += 4; r += 4; 97890267Sdes } 97937608Sdes 98090267Sdes switch (l) { 98190267Sdes case 2: 98290267Sdes t = (src[0] << 16) | (src[1] << 8); 98390267Sdes dst[0] = base64[(t >> 18) & 0x3f]; 98490267Sdes dst[1] = base64[(t >> 12) & 0x3f]; 98590267Sdes dst[2] = base64[(t >> 6) & 0x3f]; 98690267Sdes dst[3] = '='; 98790267Sdes dst += 4; 98890267Sdes r += 4; 98990267Sdes break; 99090267Sdes case 1: 99190267Sdes t = src[0] << 16; 99290267Sdes dst[0] = base64[(t >> 18) & 0x3f]; 99390267Sdes dst[1] = base64[(t >> 12) & 0x3f]; 99490267Sdes dst[2] = dst[3] = '='; 99590267Sdes dst += 4; 99690267Sdes r += 4; 99790267Sdes break; 99890267Sdes case 0: 99990267Sdes break; 100090267Sdes } 100190267Sdes 100290267Sdes *dst = 0; 100390267Sdes return (str); 100437608Sdes} 100537608Sdes 1006202613Sdes 100737608Sdes/* 1008202613Sdes * Extract authorization parameters from environment value. 1009202613Sdes * The value is like scheme:realm:user:pass 1010202613Sdes */ 1011202613Sdestypedef struct { 1012202613Sdes char *scheme; 1013202613Sdes char *realm; 1014202613Sdes char *user; 1015202613Sdes char *password; 1016202613Sdes} http_auth_params_t; 1017202613Sdes 1018202613Sdesstatic void 1019202613Sdesinit_http_auth_params(http_auth_params_t *s) 1020202613Sdes{ 1021202613Sdes s->scheme = s->realm = s->user = s->password = 0; 1022202613Sdes} 1023202613Sdes 1024221821Sdesstatic void 1025202613Sdesclean_http_auth_params(http_auth_params_t *s) 1026202613Sdes{ 1027221821Sdes if (s->scheme) 1028202613Sdes free(s->scheme); 1029221821Sdes if (s->realm) 1030202613Sdes free(s->realm); 1031221821Sdes if (s->user) 1032202613Sdes free(s->user); 1033221821Sdes if (s->password) 1034202613Sdes free(s->password); 1035202613Sdes init_http_auth_params(s); 1036202613Sdes} 1037202613Sdes 1038202613Sdesstatic int 1039202613Sdeshttp_authfromenv(const char *p, http_auth_params_t *parms) 1040202613Sdes{ 1041202613Sdes int ret = -1; 1042202613Sdes char *v, *ve; 1043202613Sdes char *str = strdup(p); 1044202613Sdes 1045202613Sdes if (str == NULL) { 1046202613Sdes fetch_syserr(); 1047202613Sdes return (-1); 1048202613Sdes } 1049202613Sdes v = str; 1050202613Sdes 1051202613Sdes if ((ve = strchr(v, ':')) == NULL) 1052202613Sdes goto out; 1053202613Sdes 1054202613Sdes *ve = 0; 1055202613Sdes if ((parms->scheme = strdup(v)) == NULL) { 1056202613Sdes fetch_syserr(); 1057202613Sdes goto out; 1058202613Sdes } 1059202613Sdes v = ve + 1; 1060202613Sdes 1061202613Sdes if ((ve = strchr(v, ':')) == NULL) 1062202613Sdes goto out; 1063202613Sdes 1064202613Sdes *ve = 0; 1065202613Sdes if ((parms->realm = strdup(v)) == NULL) { 1066202613Sdes fetch_syserr(); 1067202613Sdes goto out; 1068202613Sdes } 1069202613Sdes v = ve + 1; 1070202613Sdes 1071202613Sdes if ((ve = strchr(v, ':')) == NULL) 1072202613Sdes goto out; 1073202613Sdes 1074202613Sdes *ve = 0; 1075202613Sdes if ((parms->user = strdup(v)) == NULL) { 1076202613Sdes fetch_syserr(); 1077202613Sdes goto out; 1078202613Sdes } 1079202613Sdes v = ve + 1; 1080202613Sdes 1081202613Sdes 1082202613Sdes if ((parms->password = strdup(v)) == NULL) { 1083202613Sdes fetch_syserr(); 1084202613Sdes goto out; 1085202613Sdes } 1086202613Sdes ret = 0; 1087202613Sdesout: 1088221821Sdes if (ret == -1) 1089202613Sdes clean_http_auth_params(parms); 1090202613Sdes if (str) 1091202613Sdes free(str); 1092202613Sdes return (ret); 1093202613Sdes} 1094202613Sdes 1095202613Sdes 1096221821Sdes/* 1097202613Sdes * Digest response: the code to compute the digest is taken from the 1098221821Sdes * sample implementation in RFC2616 1099202613Sdes */ 1100221822Sdes#define IN const 1101202613Sdes#define OUT 1102202613Sdes 1103202613Sdes#define HASHLEN 16 1104202613Sdestypedef char HASH[HASHLEN]; 1105202613Sdes#define HASHHEXLEN 32 1106202613Sdestypedef char HASHHEX[HASHHEXLEN+1]; 1107202613Sdes 1108202613Sdesstatic const char *hexchars = "0123456789abcdef"; 1109221821Sdesstatic void 1110202613SdesCvtHex(IN HASH Bin, OUT HASHHEX Hex) 1111202613Sdes{ 1112202613Sdes unsigned short i; 1113202613Sdes unsigned char j; 1114202613Sdes 1115202613Sdes for (i = 0; i < HASHLEN; i++) { 1116202613Sdes j = (Bin[i] >> 4) & 0xf; 1117202613Sdes Hex[i*2] = hexchars[j]; 1118202613Sdes j = Bin[i] & 0xf; 1119202613Sdes Hex[i*2+1] = hexchars[j]; 1120202613Sdes }; 1121202613Sdes Hex[HASHHEXLEN] = '\0'; 1122202613Sdes}; 1123202613Sdes 1124202613Sdes/* calculate H(A1) as per spec */ 1125221821Sdesstatic void 1126202613SdesDigestCalcHA1( 1127202613Sdes IN char * pszAlg, 1128202613Sdes IN char * pszUserName, 1129202613Sdes IN char * pszRealm, 1130202613Sdes IN char * pszPassword, 1131202613Sdes IN char * pszNonce, 1132202613Sdes IN char * pszCNonce, 1133202613Sdes OUT HASHHEX SessionKey 1134202613Sdes ) 1135202613Sdes{ 1136202613Sdes MD5_CTX Md5Ctx; 1137202613Sdes HASH HA1; 1138202613Sdes 1139202613Sdes MD5Init(&Md5Ctx); 1140202613Sdes MD5Update(&Md5Ctx, pszUserName, strlen(pszUserName)); 1141202613Sdes MD5Update(&Md5Ctx, ":", 1); 1142202613Sdes MD5Update(&Md5Ctx, pszRealm, strlen(pszRealm)); 1143202613Sdes MD5Update(&Md5Ctx, ":", 1); 1144202613Sdes MD5Update(&Md5Ctx, pszPassword, strlen(pszPassword)); 1145202613Sdes MD5Final(HA1, &Md5Ctx); 1146202613Sdes if (strcasecmp(pszAlg, "md5-sess") == 0) { 1147202613Sdes 1148202613Sdes MD5Init(&Md5Ctx); 1149202613Sdes MD5Update(&Md5Ctx, HA1, HASHLEN); 1150202613Sdes MD5Update(&Md5Ctx, ":", 1); 1151202613Sdes MD5Update(&Md5Ctx, pszNonce, strlen(pszNonce)); 1152202613Sdes MD5Update(&Md5Ctx, ":", 1); 1153202613Sdes MD5Update(&Md5Ctx, pszCNonce, strlen(pszCNonce)); 1154202613Sdes MD5Final(HA1, &Md5Ctx); 1155202613Sdes }; 1156202613Sdes CvtHex(HA1, SessionKey); 1157202613Sdes} 1158202613Sdes 1159202613Sdes/* calculate request-digest/response-digest as per HTTP Digest spec */ 1160221821Sdesstatic void 1161202613SdesDigestCalcResponse( 1162202613Sdes IN HASHHEX HA1, /* H(A1) */ 1163202613Sdes IN char * pszNonce, /* nonce from server */ 1164202613Sdes IN char * pszNonceCount, /* 8 hex digits */ 1165202613Sdes IN char * pszCNonce, /* client nonce */ 1166202613Sdes IN char * pszQop, /* qop-value: "", "auth", "auth-int" */ 1167202613Sdes IN char * pszMethod, /* method from the request */ 1168202613Sdes IN char * pszDigestUri, /* requested URL */ 1169202613Sdes IN HASHHEX HEntity, /* H(entity body) if qop="auth-int" */ 1170202613Sdes OUT HASHHEX Response /* request-digest or response-digest */ 1171202613Sdes ) 1172202613Sdes{ 1173221821Sdes/* DEBUG(fprintf(stderr, 1174202613Sdes "Calc: HA1[%s] Nonce[%s] qop[%s] method[%s] URI[%s]\n", 1175202613Sdes HA1, pszNonce, pszQop, pszMethod, pszDigestUri));*/ 1176202613Sdes MD5_CTX Md5Ctx; 1177202613Sdes HASH HA2; 1178202613Sdes HASH RespHash; 1179202613Sdes HASHHEX HA2Hex; 1180202613Sdes 1181202613Sdes // calculate H(A2) 1182202613Sdes MD5Init(&Md5Ctx); 1183202613Sdes MD5Update(&Md5Ctx, pszMethod, strlen(pszMethod)); 1184202613Sdes MD5Update(&Md5Ctx, ":", 1); 1185202613Sdes MD5Update(&Md5Ctx, pszDigestUri, strlen(pszDigestUri)); 1186202613Sdes if (strcasecmp(pszQop, "auth-int") == 0) { 1187202613Sdes MD5Update(&Md5Ctx, ":", 1); 1188202613Sdes MD5Update(&Md5Ctx, HEntity, HASHHEXLEN); 1189202613Sdes }; 1190202613Sdes MD5Final(HA2, &Md5Ctx); 1191202613Sdes CvtHex(HA2, HA2Hex); 1192202613Sdes 1193202613Sdes // calculate response 1194202613Sdes MD5Init(&Md5Ctx); 1195202613Sdes MD5Update(&Md5Ctx, HA1, HASHHEXLEN); 1196202613Sdes MD5Update(&Md5Ctx, ":", 1); 1197202613Sdes MD5Update(&Md5Ctx, pszNonce, strlen(pszNonce)); 1198202613Sdes MD5Update(&Md5Ctx, ":", 1); 1199202613Sdes if (*pszQop) { 1200202613Sdes MD5Update(&Md5Ctx, pszNonceCount, strlen(pszNonceCount)); 1201202613Sdes MD5Update(&Md5Ctx, ":", 1); 1202202613Sdes MD5Update(&Md5Ctx, pszCNonce, strlen(pszCNonce)); 1203202613Sdes MD5Update(&Md5Ctx, ":", 1); 1204202613Sdes MD5Update(&Md5Ctx, pszQop, strlen(pszQop)); 1205202613Sdes MD5Update(&Md5Ctx, ":", 1); 1206202613Sdes }; 1207202613Sdes MD5Update(&Md5Ctx, HA2Hex, HASHHEXLEN); 1208202613Sdes MD5Final(RespHash, &Md5Ctx); 1209202613Sdes CvtHex(RespHash, Response); 1210202613Sdes} 1211202613Sdes 1212221821Sdes/* 1213221821Sdes * Generate/Send a Digest authorization header 1214202613Sdes * This looks like: [Proxy-]Authorization: credentials 1215202613Sdes * 1216202613Sdes * credentials = "Digest" digest-response 1217202613Sdes * digest-response = 1#( username | realm | nonce | digest-uri 1218202613Sdes * | response | [ algorithm ] | [cnonce] | 1219202613Sdes * [opaque] | [message-qop] | 1220202613Sdes * [nonce-count] | [auth-param] ) 1221202613Sdes * username = "username" "=" username-value 1222202613Sdes * username-value = quoted-string 1223202613Sdes * digest-uri = "uri" "=" digest-uri-value 1224202613Sdes * digest-uri-value = request-uri ; As specified by HTTP/1.1 1225202613Sdes * message-qop = "qop" "=" qop-value 1226202613Sdes * cnonce = "cnonce" "=" cnonce-value 1227202613Sdes * cnonce-value = nonce-value 1228202613Sdes * nonce-count = "nc" "=" nc-value 1229202613Sdes * nc-value = 8LHEX 1230202613Sdes * response = "response" "=" request-digest 1231202613Sdes * request-digest = <"> 32LHEX <"> 1232202613Sdes */ 1233202613Sdesstatic int 1234202613Sdeshttp_digest_auth(conn_t *conn, const char *hdr, http_auth_challenge_t *c, 1235202613Sdes http_auth_params_t *parms, struct url *url) 1236202613Sdes{ 1237202613Sdes int r; 1238202613Sdes char noncecount[10]; 1239202613Sdes char cnonce[40]; 1240202613Sdes char *options = 0; 1241202613Sdes 1242202613Sdes if (!c->realm || !c->nonce) { 1243202613Sdes DEBUG(fprintf(stderr, "realm/nonce not set in challenge\n")); 1244202613Sdes return(-1); 1245202613Sdes } 1246221821Sdes if (!c->algo) 1247202613Sdes c->algo = strdup(""); 1248202613Sdes 1249221821Sdes if (asprintf(&options, "%s%s%s%s", 1250202613Sdes *c->algo? ",algorithm=" : "", c->algo, 1251202613Sdes c->opaque? ",opaque=" : "", c->opaque?c->opaque:"")== -1) 1252202613Sdes return (-1); 1253202613Sdes 1254202613Sdes if (!c->qop) { 1255202613Sdes c->qop = strdup(""); 1256202613Sdes *noncecount = 0; 1257202613Sdes *cnonce = 0; 1258202613Sdes } else { 1259202613Sdes c->nc++; 1260202613Sdes sprintf(noncecount, "%08x", c->nc); 1261202613Sdes /* We don't try very hard with the cnonce ... */ 1262202613Sdes sprintf(cnonce, "%x%lx", getpid(), (unsigned long)time(0)); 1263202613Sdes } 1264202613Sdes 1265202613Sdes HASHHEX HA1; 1266202613Sdes DigestCalcHA1(c->algo, parms->user, c->realm, 1267202613Sdes parms->password, c->nonce, cnonce, HA1); 1268202613Sdes DEBUG(fprintf(stderr, "HA1: [%s]\n", HA1)); 1269202613Sdes HASHHEX digest; 1270202613Sdes DigestCalcResponse(HA1, c->nonce, noncecount, cnonce, c->qop, 1271202613Sdes "GET", url->doc, "", digest); 1272202613Sdes 1273202613Sdes if (c->qop[0]) { 1274202613Sdes r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\"," 1275202613Sdes "nonce=\"%s\",uri=\"%s\",response=\"%s\"," 1276202613Sdes "qop=\"auth\", cnonce=\"%s\", nc=%s%s", 1277221821Sdes hdr, parms->user, c->realm, 1278202613Sdes c->nonce, url->doc, digest, 1279202613Sdes cnonce, noncecount, options); 1280202613Sdes } else { 1281202613Sdes r = http_cmd(conn, "%s: Digest username=\"%s\",realm=\"%s\"," 1282202613Sdes "nonce=\"%s\",uri=\"%s\",response=\"%s\"%s", 1283221821Sdes hdr, parms->user, c->realm, 1284202613Sdes c->nonce, url->doc, digest, options); 1285202613Sdes } 1286202613Sdes if (options) 1287202613Sdes free(options); 1288202613Sdes return (r); 1289202613Sdes} 1290202613Sdes 1291202613Sdes/* 129237608Sdes * Encode username and password 129337608Sdes */ 129462965Sdesstatic int 1295174588Sdeshttp_basic_auth(conn_t *conn, const char *hdr, const char *usr, const char *pwd) 129637608Sdes{ 129790267Sdes char *upw, *auth; 129890267Sdes int r; 129937608Sdes 1300202613Sdes DEBUG(fprintf(stderr, "basic: usr: [%s]\n", usr)); 1301202613Sdes DEBUG(fprintf(stderr, "basic: pwd: [%s]\n", pwd)); 130290267Sdes if (asprintf(&upw, "%s:%s", usr, pwd) == -1) 130390267Sdes return (-1); 1304174588Sdes auth = http_base64(upw); 130590267Sdes free(upw); 130690267Sdes if (auth == NULL) 130790267Sdes return (-1); 1308174588Sdes r = http_cmd(conn, "%s: Basic %s", hdr, auth); 130990267Sdes free(auth); 131090267Sdes return (r); 131162965Sdes} 131262965Sdes 131362965Sdes/* 1314221821Sdes * Chose the challenge to answer and call the appropriate routine to 1315202613Sdes * produce the header. 131662965Sdes */ 131762965Sdesstatic int 1318202613Sdeshttp_authorize(conn_t *conn, const char *hdr, http_auth_challenges_t *cs, 1319202613Sdes http_auth_params_t *parms, struct url *url) 132062965Sdes{ 1321202613Sdes http_auth_challenge_t *basic = NULL; 1322202613Sdes http_auth_challenge_t *digest = NULL; 1323202613Sdes int i; 132462965Sdes 1325202613Sdes /* If user or pass are null we're not happy */ 1326202613Sdes if (!parms->user || !parms->password) { 1327202613Sdes DEBUG(fprintf(stderr, "NULL usr or pass\n")); 1328202613Sdes return (-1); 132990267Sdes } 1330202613Sdes 1331202613Sdes /* Look for a Digest and a Basic challenge */ 1332202613Sdes for (i = 0; i < cs->count; i++) { 1333202613Sdes if (cs->challenges[i]->scheme == HTTPAS_BASIC) 1334202613Sdes basic = cs->challenges[i]; 1335202613Sdes if (cs->challenges[i]->scheme == HTTPAS_DIGEST) 1336202613Sdes digest = cs->challenges[i]; 1337202613Sdes } 1338202613Sdes 1339202613Sdes /* Error if "Digest" was specified and there is no Digest challenge */ 1340221821Sdes if (!digest && (parms->scheme && 1341202613Sdes !strcasecmp(parms->scheme, "digest"))) { 1342221821Sdes DEBUG(fprintf(stderr, 1343202613Sdes "Digest auth in env, not supported by peer\n")); 1344202613Sdes return (-1); 1345202613Sdes } 1346221821Sdes /* 1347221821Sdes * If "basic" was specified in the environment, or there is no Digest 1348202613Sdes * challenge, do the basic thing. Don't need a challenge for this, 1349221821Sdes * so no need to check basic!=NULL 1350202613Sdes */ 1351202613Sdes if (!digest || (parms->scheme && !strcasecmp(parms->scheme,"basic"))) 1352202613Sdes return (http_basic_auth(conn,hdr,parms->user,parms->password)); 1353202613Sdes 1354202613Sdes /* Else, prefer digest. We just checked that it's not NULL */ 1355202613Sdes return (http_digest_auth(conn, hdr, digest, parms, url)); 135637608Sdes} 135737608Sdes 135863012Sdes/***************************************************************************** 135963012Sdes * Helper functions for connecting to a server or proxy 136063012Sdes */ 136163012Sdes 136237608Sdes/* 136390267Sdes * Connect to the correct HTTP server or proxy. 136463012Sdes */ 136597856Sdesstatic conn_t * 1366174588Sdeshttp_connect(struct url *URL, struct url *purl, const char *flags) 136763012Sdes{ 136897856Sdes conn_t *conn; 136990267Sdes int verbose; 1370141958Skbyanc int af, val; 137190267Sdes 137263012Sdes#ifdef INET6 137390267Sdes af = AF_UNSPEC; 137460737Sume#else 137590267Sdes af = AF_INET; 137660737Sume#endif 137790267Sdes 137890267Sdes verbose = CHECK_FLAG('v'); 137990267Sdes if (CHECK_FLAG('4')) 138090267Sdes af = AF_INET; 138167043Sdes#ifdef INET6 138290267Sdes else if (CHECK_FLAG('6')) 138390267Sdes af = AF_INET6; 138467043Sdes#endif 138567043Sdes 138697868Sdes if (purl && strcasecmp(URL->scheme, SCHEME_HTTPS) != 0) { 138790267Sdes URL = purl; 138890267Sdes } else if (strcasecmp(URL->scheme, SCHEME_FTP) == 0) { 138990267Sdes /* can't talk http to an ftp server */ 139090267Sdes /* XXX should set an error code */ 139197856Sdes return (NULL); 139290267Sdes } 139390267Sdes 1394174588Sdes if ((conn = fetch_connect(URL->host, URL->port, af, verbose)) == NULL) 1395174588Sdes /* fetch_connect() has already set an error code */ 139697856Sdes return (NULL); 139797868Sdes if (strcasecmp(URL->scheme, SCHEME_HTTPS) == 0 && 1398174588Sdes fetch_ssl(conn, verbose) == -1) { 1399174588Sdes fetch_close(conn); 140097891Sdes /* grrr */ 140197891Sdes errno = EAUTH; 1402174588Sdes fetch_syserr(); 140397868Sdes return (NULL); 140497868Sdes } 1405141958Skbyanc 1406141958Skbyanc val = 1; 1407141958Skbyanc setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val, sizeof(val)); 1408141958Skbyanc 140997856Sdes return (conn); 141067043Sdes} 141167043Sdes 141267043Sdesstatic struct url * 1413174752Sdeshttp_get_proxy(struct url * url, const char *flags) 141467043Sdes{ 141590267Sdes struct url *purl; 141690267Sdes char *p; 141790267Sdes 1418112797Sdes if (flags != NULL && strchr(flags, 'd') != NULL) 1419112081Sdes return (NULL); 1420174752Sdes if (fetch_no_proxy_match(url->host)) 1421174752Sdes return (NULL); 142290267Sdes if (((p = getenv("HTTP_PROXY")) || (p = getenv("http_proxy"))) && 1423149414Sdes *p && (purl = fetchParseURL(p))) { 142490267Sdes if (!*purl->scheme) 142590267Sdes strcpy(purl->scheme, SCHEME_HTTP); 142690267Sdes if (!purl->port) 1427174588Sdes purl->port = fetch_default_proxy_port(purl->scheme); 142890267Sdes if (strcasecmp(purl->scheme, SCHEME_HTTP) == 0) 142990267Sdes return (purl); 143090267Sdes fetchFreeURL(purl); 143190267Sdes } 143290267Sdes return (NULL); 143360376Sdes} 143460376Sdes 143588771Sdesstatic void 1436174588Sdeshttp_print_html(FILE *out, FILE *in) 143788771Sdes{ 143890267Sdes size_t len; 143990267Sdes char *line, *p, *q; 144090267Sdes int comment, tag; 144188771Sdes 144290267Sdes comment = tag = 0; 144390267Sdes while ((line = fgetln(in, &len)) != NULL) { 1444174761Sdes while (len && isspace((unsigned char)line[len - 1])) 144590267Sdes --len; 144690267Sdes for (p = q = line; q < line + len; ++q) { 144790267Sdes if (comment && *q == '-') { 144890267Sdes if (q + 2 < line + len && 144990267Sdes strcmp(q, "-->") == 0) { 145090267Sdes tag = comment = 0; 145190267Sdes q += 2; 145290267Sdes } 145390267Sdes } else if (tag && !comment && *q == '>') { 145490267Sdes p = q + 1; 145590267Sdes tag = 0; 145690267Sdes } else if (!tag && *q == '<') { 145790267Sdes if (q > p) 145890267Sdes fwrite(p, q - p, 1, out); 145990267Sdes tag = 1; 146090267Sdes if (q + 3 < line + len && 146190267Sdes strcmp(q, "<!--") == 0) { 146290267Sdes comment = 1; 146390267Sdes q += 3; 146490267Sdes } 146590267Sdes } 146688771Sdes } 146790267Sdes if (!tag && q > p) 146890267Sdes fwrite(p, q - p, 1, out); 146990267Sdes fputc('\n', out); 147088771Sdes } 147188771Sdes} 147288771Sdes 147390267Sdes 147463012Sdes/***************************************************************************** 147563012Sdes * Core 147660954Sdes */ 147760954Sdes 147860954Sdes/* 147963012Sdes * Send a request and process the reply 148097866Sdes * 148197866Sdes * XXX This function is way too long, the do..while loop should be split 148297866Sdes * XXX off into a separate function. 148360376Sdes */ 148467043SdesFILE * 1485174588Sdeshttp_request(struct url *URL, const char *op, struct url_stat *us, 1486202613Sdes struct url *purl, const char *flags) 148760376Sdes{ 1488186124Smurray char timebuf[80]; 1489186124Smurray char hbuf[MAXHOSTNAMELEN + 7], *host; 149097856Sdes conn_t *conn; 149190267Sdes struct url *url, *new; 1492202613Sdes int chunked, direct, ims, noredirect, verbose; 1493143049Skbyanc int e, i, n, val; 149490267Sdes off_t offset, clength, length, size; 149590267Sdes time_t mtime; 149690267Sdes const char *p; 149790267Sdes FILE *f; 149890267Sdes hdr_t h; 1499186124Smurray struct tm *timestruct; 1500202613Sdes http_headerbuf_t headerbuf; 1501202613Sdes http_auth_challenges_t server_challenges; 1502202613Sdes http_auth_challenges_t proxy_challenges; 150363012Sdes 1504202613Sdes /* The following calls don't allocate anything */ 1505221821Sdes init_http_headerbuf(&headerbuf); 1506202613Sdes init_http_auth_challenges(&server_challenges); 1507202613Sdes init_http_auth_challenges(&proxy_challenges); 1508202613Sdes 150990267Sdes direct = CHECK_FLAG('d'); 151090267Sdes noredirect = CHECK_FLAG('A'); 151190267Sdes verbose = CHECK_FLAG('v'); 1512186124Smurray ims = CHECK_FLAG('i'); 151360737Sume 151490267Sdes if (direct && purl) { 151590267Sdes fetchFreeURL(purl); 151690267Sdes purl = NULL; 151790267Sdes } 151863716Sdes 151990267Sdes /* try the provided URL first */ 152090267Sdes url = URL; 152163012Sdes 1522242289Seadler n = MAX_REDIRECT; 152390267Sdes i = 0; 152463012Sdes 152598422Sdes e = HTTP_PROTOCOL_ERROR; 152690267Sdes do { 152790267Sdes new = NULL; 152890267Sdes chunked = 0; 152990267Sdes offset = 0; 153090267Sdes clength = -1; 153190267Sdes length = -1; 153290267Sdes size = -1; 153390267Sdes mtime = 0; 153490267Sdes 153590267Sdes /* check port */ 153690267Sdes if (!url->port) 1537174588Sdes url->port = fetch_default_port(url->scheme); 153890267Sdes 153990267Sdes /* were we redirected to an FTP URL? */ 154090267Sdes if (purl == NULL && strcmp(url->scheme, SCHEME_FTP) == 0) { 154190267Sdes if (strcmp(op, "GET") == 0) 1542174588Sdes return (ftp_request(url, "RETR", us, purl, flags)); 154390267Sdes else if (strcmp(op, "HEAD") == 0) 1544174588Sdes return (ftp_request(url, "STAT", us, purl, flags)); 154590267Sdes } 154690267Sdes 154790267Sdes /* connect to server or proxy */ 1548174588Sdes if ((conn = http_connect(url, purl, flags)) == NULL) 154990267Sdes goto ouch; 155090267Sdes 155190267Sdes host = url->host; 155260737Sume#ifdef INET6 155390267Sdes if (strchr(url->host, ':')) { 155490267Sdes snprintf(hbuf, sizeof(hbuf), "[%s]", url->host); 155590267Sdes host = hbuf; 155690267Sdes } 155760737Sume#endif 1558174588Sdes if (url->port != fetch_default_port(url->scheme)) { 1559107372Sdes if (host != hbuf) { 1560107372Sdes strcpy(hbuf, host); 1561107372Sdes host = hbuf; 1562107372Sdes } 1563107372Sdes snprintf(hbuf + strlen(hbuf), 1564107372Sdes sizeof(hbuf) - strlen(hbuf), ":%d", url->port); 1565107372Sdes } 156637535Sdes 156790267Sdes /* send request */ 156890267Sdes if (verbose) 1569174588Sdes fetch_info("requesting %s://%s%s", 1570107372Sdes url->scheme, host, url->doc); 157190267Sdes if (purl) { 1572174588Sdes http_cmd(conn, "%s %s://%s%s HTTP/1.1", 1573107372Sdes op, url->scheme, host, url->doc); 157490267Sdes } else { 1575174588Sdes http_cmd(conn, "%s %s HTTP/1.1", 157690267Sdes op, url->doc); 157790267Sdes } 157837535Sdes 1579186124Smurray if (ims && url->ims_time) { 1580186124Smurray timestruct = gmtime((time_t *)&url->ims_time); 1581186124Smurray (void)strftime(timebuf, 80, "%a, %d %b %Y %T GMT", 1582186124Smurray timestruct); 1583186124Smurray if (verbose) 1584186124Smurray fetch_info("If-Modified-Since: %s", timebuf); 1585186124Smurray http_cmd(conn, "If-Modified-Since: %s", timebuf); 1586186124Smurray } 158790267Sdes /* virtual host */ 1588174588Sdes http_cmd(conn, "Host: %s", host); 158990267Sdes 1590221821Sdes /* 1591221821Sdes * Proxy authorization: we only send auth after we received 1592221821Sdes * a 407 error. We do not first try basic anyway (changed 1593221821Sdes * when support was added for digest-auth) 1594221821Sdes */ 1595202613Sdes if (purl && proxy_challenges.valid) { 1596202613Sdes http_auth_params_t aparams; 1597202613Sdes init_http_auth_params(&aparams); 1598202613Sdes if (*purl->user || *purl->pwd) { 1599221821Sdes aparams.user = purl->user ? 1600202613Sdes strdup(purl->user) : strdup(""); 1601202613Sdes aparams.password = purl->pwd? 1602202613Sdes strdup(purl->pwd) : strdup(""); 1603221821Sdes } else if ((p = getenv("HTTP_PROXY_AUTH")) != NULL && 1604202613Sdes *p != '\0') { 1605202613Sdes if (http_authfromenv(p, &aparams) < 0) { 1606202613Sdes http_seterr(HTTP_NEED_PROXY_AUTH); 1607202613Sdes goto ouch; 1608202613Sdes } 1609202613Sdes } 1610221821Sdes http_authorize(conn, "Proxy-Authorization", 1611202613Sdes &proxy_challenges, &aparams, url); 1612202613Sdes clean_http_auth_params(&aparams); 161390267Sdes } 161490267Sdes 1615221821Sdes /* 1616221821Sdes * Server authorization: we never send "a priori" 1617202613Sdes * Basic auth, which used to be done if user/pass were 1618202613Sdes * set in the url. This would be weird because we'd send the 1619221821Sdes * password in the clear even if Digest is finally to be 1620202613Sdes * used (it would have made more sense for the 1621221821Sdes * pre-digest version to do this when Basic was specified 1622221821Sdes * in the environment) 1623221821Sdes */ 1624202613Sdes if (server_challenges.valid) { 1625202613Sdes http_auth_params_t aparams; 1626202613Sdes init_http_auth_params(&aparams); 1627202613Sdes if (*url->user || *url->pwd) { 1628221821Sdes aparams.user = url->user ? 1629202613Sdes strdup(url->user) : strdup(""); 1630221821Sdes aparams.password = url->pwd ? 1631202613Sdes strdup(url->pwd) : strdup(""); 1632221821Sdes } else if ((p = getenv("HTTP_AUTH")) != NULL && 1633202613Sdes *p != '\0') { 1634202613Sdes if (http_authfromenv(p, &aparams) < 0) { 1635202613Sdes http_seterr(HTTP_NEED_AUTH); 1636202613Sdes goto ouch; 1637202613Sdes } 1638221821Sdes } else if (fetchAuthMethod && 1639202613Sdes fetchAuthMethod(url) == 0) { 1640221821Sdes aparams.user = url->user ? 1641202613Sdes strdup(url->user) : strdup(""); 1642221821Sdes aparams.password = url->pwd ? 1643202613Sdes strdup(url->pwd) : strdup(""); 164490267Sdes } else { 1645174588Sdes http_seterr(HTTP_NEED_AUTH); 164690267Sdes goto ouch; 164790267Sdes } 1648221821Sdes http_authorize(conn, "Authorization", 1649202613Sdes &server_challenges, &aparams, url); 1650202613Sdes clean_http_auth_params(&aparams); 165190267Sdes } 165290267Sdes 165390267Sdes /* other headers */ 1654107372Sdes if ((p = getenv("HTTP_REFERER")) != NULL && *p != '\0') { 1655107372Sdes if (strcasecmp(p, "auto") == 0) 1656174588Sdes http_cmd(conn, "Referer: %s://%s%s", 1657107372Sdes url->scheme, host, url->doc); 1658107372Sdes else 1659174588Sdes http_cmd(conn, "Referer: %s", p); 1660107372Sdes } 166190267Sdes if ((p = getenv("HTTP_USER_AGENT")) != NULL && *p != '\0') 1662174588Sdes http_cmd(conn, "User-Agent: %s", p); 166390267Sdes else 1664174588Sdes http_cmd(conn, "User-Agent: %s " _LIBFETCH_VER, getprogname()); 1665109693Sdes if (url->offset > 0) 1666174588Sdes http_cmd(conn, "Range: bytes=%lld-", (long long)url->offset); 1667174588Sdes http_cmd(conn, "Connection: close"); 1668174588Sdes http_cmd(conn, ""); 166990267Sdes 1670143049Skbyanc /* 1671143049Skbyanc * Force the queued request to be dispatched. Normally, one 1672143049Skbyanc * would do this with shutdown(2) but squid proxies can be 1673143049Skbyanc * configured to disallow such half-closed connections. To 1674143049Skbyanc * be compatible with such configurations, fiddle with socket 1675143049Skbyanc * options to force the pending data to be written. 1676143049Skbyanc */ 1677143049Skbyanc val = 0; 1678143049Skbyanc setsockopt(conn->sd, IPPROTO_TCP, TCP_NOPUSH, &val, 1679143049Skbyanc sizeof(val)); 1680143049Skbyanc val = 1; 1681143049Skbyanc setsockopt(conn->sd, IPPROTO_TCP, TCP_NODELAY, &val, 1682143049Skbyanc sizeof(val)); 1683143049Skbyanc 168490267Sdes /* get reply */ 1685174588Sdes switch (http_get_reply(conn)) { 168690267Sdes case HTTP_OK: 168790267Sdes case HTTP_PARTIAL: 1688186124Smurray case HTTP_NOT_MODIFIED: 168990267Sdes /* fine */ 169090267Sdes break; 169190267Sdes case HTTP_MOVED_PERM: 169290267Sdes case HTTP_MOVED_TEMP: 169390267Sdes case HTTP_SEE_OTHER: 1694242293Seadler case HTTP_USE_PROXY: 169590267Sdes /* 1696125695Sdes * Not so fine, but we still have to read the 1697125695Sdes * headers to get the new location. 169890267Sdes */ 169990267Sdes break; 170090267Sdes case HTTP_NEED_AUTH: 1701202613Sdes if (server_challenges.valid) { 170290267Sdes /* 1703125695Sdes * We already sent out authorization code, 1704125695Sdes * so there's nothing more we can do. 170590267Sdes */ 1706174588Sdes http_seterr(conn->err); 170790267Sdes goto ouch; 170890267Sdes } 170990267Sdes /* try again, but send the password this time */ 171090267Sdes if (verbose) 1711174588Sdes fetch_info("server requires authorization"); 171290267Sdes break; 171390267Sdes case HTTP_NEED_PROXY_AUTH: 1714202613Sdes if (proxy_challenges.valid) { 1715202613Sdes /* 1716202613Sdes * We already sent our proxy 1717202613Sdes * authorization code, so there's 1718202613Sdes * nothing more we can do. */ 1719202613Sdes http_seterr(conn->err); 1720202613Sdes goto ouch; 1721202613Sdes } 1722202613Sdes /* try again, but send the password this time */ 1723202613Sdes if (verbose) 1724202613Sdes fetch_info("proxy requires authorization"); 1725202613Sdes break; 1726125696Sdes case HTTP_BAD_RANGE: 1727125696Sdes /* 1728125696Sdes * This can happen if we ask for 0 bytes because 1729125696Sdes * we already have the whole file. Consider this 1730125696Sdes * a success for now, and check sizes later. 1731125696Sdes */ 1732125696Sdes break; 173390267Sdes case HTTP_PROTOCOL_ERROR: 173490267Sdes /* fall through */ 173590267Sdes case -1: 1736174588Sdes fetch_syserr(); 173790267Sdes goto ouch; 173890267Sdes default: 1739174588Sdes http_seterr(conn->err); 174090267Sdes if (!verbose) 174190267Sdes goto ouch; 174290267Sdes /* fall through so we can get the full error message */ 174390267Sdes } 174490267Sdes 1745202613Sdes /* get headers. http_next_header expects one line readahead */ 1746202613Sdes if (fetch_getln(conn) == -1) { 1747202613Sdes fetch_syserr(); 1748202613Sdes goto ouch; 1749202613Sdes } 175090267Sdes do { 1751202613Sdes switch ((h = http_next_header(conn, &headerbuf, &p))) { 175290267Sdes case hdr_syserror: 1753174588Sdes fetch_syserr(); 175490267Sdes goto ouch; 175590267Sdes case hdr_error: 1756174588Sdes http_seterr(HTTP_PROTOCOL_ERROR); 175790267Sdes goto ouch; 175890267Sdes case hdr_content_length: 1759174588Sdes http_parse_length(p, &clength); 176090267Sdes break; 176190267Sdes case hdr_content_range: 1762174588Sdes http_parse_range(p, &offset, &length, &size); 176390267Sdes break; 176490267Sdes case hdr_last_modified: 1765174588Sdes http_parse_mtime(p, &mtime); 176690267Sdes break; 176790267Sdes case hdr_location: 176897856Sdes if (!HTTP_REDIRECT(conn->err)) 176990267Sdes break; 1770242289Seadler /* 1771242289Seadler * if the A flag is set, we don't follow 1772242289Seadler * temporary redirects. 1773242289Seadler */ 1774242289Seadler if (noredirect && 1775242289Seadler conn->err != HTTP_MOVED_PERM && 1776242293Seadler conn->err != HTTP_PERM_REDIRECT && 1777242293Seadler conn->err != HTTP_USE_PROXY) { 1778242289Seadler n = 1; 1779242289Seadler break; 1780242289Seadler } 178190267Sdes if (new) 178290267Sdes free(new); 178390267Sdes if (verbose) 1784174588Sdes fetch_info("%d redirect to %s", conn->err, p); 178590267Sdes if (*p == '/') 178690267Sdes /* absolute path */ 178790267Sdes new = fetchMakeURL(url->scheme, url->host, url->port, p, 178890267Sdes url->user, url->pwd); 178990267Sdes else 179090267Sdes new = fetchParseURL(p); 179190267Sdes if (new == NULL) { 179290267Sdes /* XXX should set an error code */ 179390267Sdes DEBUG(fprintf(stderr, "failed to parse new URL\n")); 179490267Sdes goto ouch; 179590267Sdes } 1796236108Sdes 1797236108Sdes /* Only copy credentials if the host matches */ 1798236108Sdes if (!strcmp(new->host, url->host) && !*new->user && !*new->pwd) { 179990267Sdes strcpy(new->user, url->user); 180090267Sdes strcpy(new->pwd, url->pwd); 180190267Sdes } 180290267Sdes new->offset = url->offset; 180390267Sdes new->length = url->length; 180490267Sdes break; 180590267Sdes case hdr_transfer_encoding: 180690267Sdes /* XXX weak test*/ 180790267Sdes chunked = (strcasecmp(p, "chunked") == 0); 180890267Sdes break; 180990267Sdes case hdr_www_authenticate: 181097856Sdes if (conn->err != HTTP_NEED_AUTH) 181190267Sdes break; 1812210563Sdes if (http_parse_authenticate(p, &server_challenges) == 0) 1813209632Sdes ++n; 181490267Sdes break; 1815202613Sdes case hdr_proxy_authenticate: 1816202613Sdes if (conn->err != HTTP_NEED_PROXY_AUTH) 1817202613Sdes break; 1818210563Sdes if (http_parse_authenticate(p, &proxy_challenges) == 0) 1819209632Sdes ++n; 1820202613Sdes break; 182190267Sdes case hdr_end: 182290267Sdes /* fall through */ 182390267Sdes case hdr_unknown: 182490267Sdes /* ignore */ 182590267Sdes break; 182690267Sdes } 182790267Sdes } while (h > hdr_end); 182890267Sdes 182990267Sdes /* we need to provide authentication */ 1830221821Sdes if (conn->err == HTTP_NEED_AUTH || 1831202613Sdes conn->err == HTTP_NEED_PROXY_AUTH) { 183298422Sdes e = conn->err; 1833221821Sdes if ((conn->err == HTTP_NEED_AUTH && 1834221821Sdes !server_challenges.valid) || 1835221821Sdes (conn->err == HTTP_NEED_PROXY_AUTH && 1836202613Sdes !proxy_challenges.valid)) { 1837202613Sdes /* 401/7 but no www/proxy-authenticate ?? */ 1838202613Sdes DEBUG(fprintf(stderr, "401/7 and no auth header\n")); 1839202613Sdes goto ouch; 1840202613Sdes } 1841174588Sdes fetch_close(conn); 184297856Sdes conn = NULL; 184390267Sdes continue; 184490267Sdes } 184590267Sdes 1846125696Sdes /* requested range not satisfiable */ 1847125696Sdes if (conn->err == HTTP_BAD_RANGE) { 1848125696Sdes if (url->offset == size && url->length == 0) { 1849125696Sdes /* asked for 0 bytes; fake it */ 1850125696Sdes offset = url->offset; 1851184222Sru clength = -1; 1852125696Sdes conn->err = HTTP_OK; 1853125696Sdes break; 1854125696Sdes } else { 1855174588Sdes http_seterr(conn->err); 1856125696Sdes goto ouch; 1857125696Sdes } 1858125696Sdes } 1859125696Sdes 1860104404Sru /* we have a hit or an error */ 1861186124Smurray if (conn->err == HTTP_OK 1862186124Smurray || conn->err == HTTP_NOT_MODIFIED 1863186124Smurray || conn->err == HTTP_PARTIAL 1864186124Smurray || HTTP_ERROR(conn->err)) 1865104404Sru break; 1866104404Sru 186790267Sdes /* all other cases: we got a redirect */ 186898422Sdes e = conn->err; 1869202613Sdes clean_http_auth_challenges(&server_challenges); 1870174588Sdes fetch_close(conn); 187197856Sdes conn = NULL; 187290267Sdes if (!new) { 187390267Sdes DEBUG(fprintf(stderr, "redirect with no new location\n")); 187490267Sdes break; 187590267Sdes } 187690267Sdes if (url != URL) 187790267Sdes fetchFreeURL(url); 187890267Sdes url = new; 187990267Sdes } while (++i < n); 188090267Sdes 188190267Sdes /* we failed, or ran out of retries */ 188297856Sdes if (conn == NULL) { 1883174588Sdes http_seterr(e); 188463012Sdes goto ouch; 188563012Sdes } 188660376Sdes 188790267Sdes DEBUG(fprintf(stderr, "offset %lld, length %lld," 188890267Sdes " size %lld, clength %lld\n", 188990267Sdes (long long)offset, (long long)length, 189090267Sdes (long long)size, (long long)clength)); 189160376Sdes 1892186124Smurray if (conn->err == HTTP_NOT_MODIFIED) { 1893186124Smurray http_seterr(HTTP_NOT_MODIFIED); 1894186124Smurray return (NULL); 1895186124Smurray } 1896186124Smurray 189790267Sdes /* check for inconsistencies */ 189890267Sdes if (clength != -1 && length != -1 && clength != length) { 1899174588Sdes http_seterr(HTTP_PROTOCOL_ERROR); 190063012Sdes goto ouch; 190163012Sdes } 190290267Sdes if (clength == -1) 190390267Sdes clength = length; 190490267Sdes if (clength != -1) 190590267Sdes length = offset + clength; 190690267Sdes if (length != -1 && size != -1 && length != size) { 1907174588Sdes http_seterr(HTTP_PROTOCOL_ERROR); 190863012Sdes goto ouch; 190990267Sdes } 191090267Sdes if (size == -1) 191190267Sdes size = length; 191260376Sdes 191390267Sdes /* fill in stats */ 191490267Sdes if (us) { 191590267Sdes us->size = size; 191690267Sdes us->atime = us->mtime = mtime; 191790267Sdes } 191863069Sdes 191990267Sdes /* too far? */ 1920109693Sdes if (URL->offset > 0 && offset > URL->offset) { 1921174588Sdes http_seterr(HTTP_PROTOCOL_ERROR); 192290267Sdes goto ouch; 192377238Sdes } 192460376Sdes 192590267Sdes /* report back real offset and size */ 192690267Sdes URL->offset = offset; 192790267Sdes URL->length = clength; 192837535Sdes 192990267Sdes /* wrap it up in a FILE */ 1930174588Sdes if ((f = http_funopen(conn, chunked)) == NULL) { 1931174588Sdes fetch_syserr(); 193290267Sdes goto ouch; 193390267Sdes } 193463716Sdes 193590267Sdes if (url != URL) 193690267Sdes fetchFreeURL(url); 193790267Sdes if (purl) 193890267Sdes fetchFreeURL(purl); 193963567Sdes 194097856Sdes if (HTTP_ERROR(conn->err)) { 1941174588Sdes http_print_html(stderr, f); 194290267Sdes fclose(f); 194390267Sdes f = NULL; 194490267Sdes } 1945202613Sdes clean_http_headerbuf(&headerbuf); 1946202613Sdes clean_http_auth_challenges(&server_challenges); 1947202613Sdes clean_http_auth_challenges(&proxy_challenges); 194890267Sdes return (f); 194988771Sdes 195090267Sdesouch: 195190267Sdes if (url != URL) 195290267Sdes fetchFreeURL(url); 195390267Sdes if (purl) 195490267Sdes fetchFreeURL(purl); 195597856Sdes if (conn != NULL) 1956174588Sdes fetch_close(conn); 1957202613Sdes clean_http_headerbuf(&headerbuf); 1958202613Sdes clean_http_auth_challenges(&server_challenges); 1959202613Sdes clean_http_auth_challenges(&proxy_challenges); 196090267Sdes return (NULL); 196163012Sdes} 196260189Sdes 196390267Sdes 196463012Sdes/***************************************************************************** 196563012Sdes * Entry points 196663012Sdes */ 196763012Sdes 196863012Sdes/* 196963340Sdes * Retrieve and stat a file by HTTP 197063340Sdes */ 197163340SdesFILE * 197275891SarchiefetchXGetHTTP(struct url *URL, struct url_stat *us, const char *flags) 197363340Sdes{ 1974174752Sdes return (http_request(URL, "GET", us, http_get_proxy(URL, flags), flags)); 197563340Sdes} 197663340Sdes 197763340Sdes/* 197863012Sdes * Retrieve a file by HTTP 197963012Sdes */ 198063012SdesFILE * 198175891SarchiefetchGetHTTP(struct url *URL, const char *flags) 198263012Sdes{ 198390267Sdes return (fetchXGetHTTP(URL, NULL, flags)); 198437535Sdes} 198537535Sdes 198663340Sdes/* 198763340Sdes * Store a file by HTTP 198863340Sdes */ 198937535SdesFILE * 199085093SdesfetchPutHTTP(struct url *URL __unused, const char *flags __unused) 199137535Sdes{ 199290267Sdes warnx("fetchPutHTTP(): not implemented"); 199390267Sdes return (NULL); 199437535Sdes} 199540975Sdes 199640975Sdes/* 199740975Sdes * Get an HTTP document's metadata 199840975Sdes */ 199940975Sdesint 200075891SarchiefetchStatHTTP(struct url *URL, struct url_stat *us, const char *flags) 200140975Sdes{ 200290267Sdes FILE *f; 200390267Sdes 2004174752Sdes f = http_request(URL, "HEAD", us, http_get_proxy(URL, flags), flags); 2005112081Sdes if (f == NULL) 200690267Sdes return (-1); 200790267Sdes fclose(f); 200890267Sdes return (0); 200940975Sdes} 201041989Sdes 201141989Sdes/* 201241989Sdes * List a directory 201341989Sdes */ 201441989Sdesstruct url_ent * 201585093SdesfetchListHTTP(struct url *url __unused, const char *flags __unused) 201641989Sdes{ 201790267Sdes warnx("fetchListHTTP(): not implemented"); 201890267Sdes return (NULL); 201941989Sdes} 2020