2002-09-03 07:52:59 -04:00
|
|
|
/***************************************************************************
|
2000-05-22 10:09:31 -04:00
|
|
|
* _ _ ____ _
|
|
|
|
* Project ___| | | | _ \| |
|
|
|
|
* / __| | | | |_) | |
|
|
|
|
* | (__| |_| | _ <| |___
|
|
|
|
* \___|\___/|_| \_\_____|
|
|
|
|
*
|
2004-01-07 04:19:33 -05:00
|
|
|
* Copyright (C) 1998 - 2004, Daniel Stenberg, <daniel@haxx.se>, et al.
|
2000-05-22 10:09:31 -04:00
|
|
|
*
|
2002-09-03 07:52:59 -04:00
|
|
|
* This software is licensed as described in the file COPYING, which
|
|
|
|
* you should have received as part of this distribution. The terms
|
|
|
|
* are also available at http://curl.haxx.se/docs/copyright.html.
|
|
|
|
*
|
2001-01-03 04:29:33 -05:00
|
|
|
* You may opt to use, copy, modify, merge, publish, distribute and/or sell
|
|
|
|
* copies of the Software, and permit persons to whom the Software is
|
2002-09-03 07:52:59 -04:00
|
|
|
* furnished to do so, under the terms of the COPYING file.
|
2000-05-22 10:09:31 -04:00
|
|
|
*
|
2001-01-03 04:29:33 -05:00
|
|
|
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
|
|
|
|
* KIND, either express or implied.
|
2000-05-22 10:09:31 -04:00
|
|
|
*
|
2001-01-03 04:29:33 -05:00
|
|
|
* $Id$
|
2002-09-03 07:52:59 -04:00
|
|
|
***************************************************************************/
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2000-08-24 10:26:33 -04:00
|
|
|
#include "setup.h"
|
|
|
|
|
2000-05-22 10:09:31 -04:00
|
|
|
/* -- WIN32 approved -- */
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <stdarg.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <ctype.h>
|
|
|
|
#include <sys/types.h>
|
|
|
|
#include <sys/stat.h>
|
|
|
|
|
|
|
|
#include <errno.h>
|
|
|
|
|
2004-01-05 17:29:29 -05:00
|
|
|
#include "strtoofft.h"
|
2000-05-22 10:09:31 -04:00
|
|
|
#include "strequal.h"
|
|
|
|
|
|
|
|
#if defined(WIN32) && !defined(__GNUC__) || defined(__MINGW32__)
|
|
|
|
#include <time.h>
|
|
|
|
#include <io.h>
|
|
|
|
#else
|
|
|
|
#ifdef HAVE_SYS_SOCKET_H
|
|
|
|
#include <sys/socket.h>
|
|
|
|
#endif
|
|
|
|
#include <netinet/in.h>
|
|
|
|
#include <sys/time.h>
|
|
|
|
#ifdef HAVE_UNISTD_H
|
|
|
|
#include <unistd.h>
|
|
|
|
#endif
|
|
|
|
#include <netdb.h>
|
|
|
|
#ifdef HAVE_ARPA_INET_H
|
|
|
|
#include <arpa/inet.h>
|
|
|
|
#endif
|
|
|
|
#ifdef HAVE_NET_IF_H
|
|
|
|
#include <net/if.h>
|
|
|
|
#endif
|
|
|
|
#include <sys/ioctl.h>
|
|
|
|
#include <signal.h>
|
|
|
|
|
|
|
|
#ifdef HAVE_SYS_PARAM_H
|
|
|
|
#include <sys/param.h>
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifdef HAVE_SYS_SELECT_H
|
|
|
|
#include <sys/select.h>
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifndef HAVE_SELECT
|
|
|
|
#error "We can't compile without select() support!"
|
|
|
|
#endif
|
|
|
|
#ifndef HAVE_SOCKET
|
|
|
|
#error "We can't compile without socket() support!"
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#include "urldata.h"
|
|
|
|
#include <curl/curl.h>
|
|
|
|
#include <curl/types.h>
|
|
|
|
#include "netrc.h"
|
|
|
|
|
2002-09-02 18:31:18 -04:00
|
|
|
#include "content_encoding.h" /* content encoding support. 08/27/02 jhrg */
|
|
|
|
|
2000-05-22 10:09:31 -04:00
|
|
|
#include "hostip.h"
|
2001-01-17 08:19:01 -05:00
|
|
|
#include "transfer.h"
|
2000-05-22 10:09:31 -04:00
|
|
|
#include "sendf.h"
|
|
|
|
#include "speedcheck.h"
|
|
|
|
#include "progress.h"
|
|
|
|
#include "getdate.h"
|
2001-03-07 18:51:41 -05:00
|
|
|
#include "http.h"
|
2001-08-15 02:53:10 -04:00
|
|
|
#include "url.h"
|
|
|
|
#include "getinfo.h"
|
2001-08-28 04:37:54 -04:00
|
|
|
#include "ssluse.h"
|
2003-05-22 12:09:54 -04:00
|
|
|
#include "http_digest.h"
|
2003-06-11 09:38:55 -04:00
|
|
|
#include "http_ntlm.h"
|
2003-06-10 08:22:19 -04:00
|
|
|
#include "http_negotiate.h"
|
2003-08-06 11:26:24 -04:00
|
|
|
#include "share.h"
|
2000-05-22 10:09:31 -04:00
|
|
|
|
|
|
|
#define _MPRINTF_REPLACE /* use our functions only */
|
|
|
|
#include <curl/mprintf.h>
|
|
|
|
|
2000-10-09 07:12:34 -04:00
|
|
|
/* The last #include file should be: */
|
2003-06-26 07:30:26 -04:00
|
|
|
#ifdef CURLDEBUG
|
2000-10-09 07:12:34 -04:00
|
|
|
#include "memdebug.h"
|
|
|
|
#endif
|
|
|
|
|
2000-10-03 07:02:52 -04:00
|
|
|
#ifndef min
|
|
|
|
#define min(a, b) ((a) < (b) ? (a) : (b))
|
|
|
|
#endif
|
|
|
|
|
2003-02-24 11:53:53 -05:00
|
|
|
#define CURL_TIMEOUT_EXPECT_100 1000 /* counting ms here */
|
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
enum {
|
|
|
|
KEEP_NONE,
|
|
|
|
KEEP_READ,
|
|
|
|
KEEP_WRITE
|
|
|
|
};
|
|
|
|
|
2002-05-28 10:18:36 -04:00
|
|
|
/* We keep this static and global since this is read-only and NEVER
|
|
|
|
changed. It should just remain a blanked-out timeout value. */
|
|
|
|
static struct timeval notimeout={0,0};
|
|
|
|
|
2002-12-10 08:10:00 -05:00
|
|
|
/*
|
|
|
|
* This function will call the read callback to fill our buffer with data
|
|
|
|
* to upload.
|
|
|
|
*/
|
|
|
|
static int fillbuffer(struct connectdata *conn,
|
|
|
|
int bytes)
|
|
|
|
{
|
|
|
|
int buffersize = bytes;
|
|
|
|
int nread;
|
|
|
|
|
|
|
|
if(conn->bits.upload_chunky) {
|
|
|
|
/* if chunked Transfer-Encoding */
|
|
|
|
buffersize -= (8 + 2 + 2); /* 32bit hex + CRLF + CRLF */
|
|
|
|
conn->upload_fromhere += 10; /* 32bit hex + CRLF */
|
|
|
|
}
|
|
|
|
|
|
|
|
nread = conn->fread(conn->upload_fromhere, 1,
|
|
|
|
buffersize, conn->fread_in);
|
|
|
|
|
|
|
|
if(!conn->bits.forbidchunk && conn->bits.upload_chunky) {
|
|
|
|
/* if chunked Transfer-Encoding */
|
|
|
|
char hexbuffer[11];
|
|
|
|
int hexlen = snprintf(hexbuffer, sizeof(hexbuffer),
|
|
|
|
"%x\r\n", nread);
|
|
|
|
/* move buffer pointer */
|
|
|
|
conn->upload_fromhere -= hexlen;
|
|
|
|
nread += hexlen;
|
|
|
|
|
|
|
|
/* copy the prefix to the buffer */
|
|
|
|
memcpy(conn->upload_fromhere, hexbuffer, hexlen);
|
|
|
|
if(nread>hexlen) {
|
|
|
|
/* append CRLF to the data */
|
|
|
|
memcpy(conn->upload_fromhere +
|
|
|
|
nread, "\r\n", 2);
|
|
|
|
nread+=2;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
/* mark this as done once this chunk is transfered */
|
|
|
|
conn->keep.upload_done = TRUE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return nread;
|
|
|
|
}
|
|
|
|
|
2003-01-07 11:15:53 -05:00
|
|
|
/*
|
|
|
|
* checkhttpprefix()
|
|
|
|
*
|
|
|
|
* Returns TRUE if member of the list matches prefix of string
|
|
|
|
*/
|
|
|
|
static bool
|
|
|
|
checkhttpprefix(struct SessionHandle *data,
|
|
|
|
const char *s)
|
|
|
|
{
|
|
|
|
struct curl_slist *head = data->set.http200aliases;
|
|
|
|
|
|
|
|
while (head) {
|
|
|
|
if (checkprefix(head->data, s))
|
|
|
|
return TRUE;
|
|
|
|
head = head->next;
|
|
|
|
}
|
|
|
|
|
|
|
|
if(checkprefix("HTTP/", s))
|
|
|
|
return TRUE;
|
|
|
|
|
|
|
|
return FALSE;
|
|
|
|
}
|
2002-12-10 08:10:00 -05:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
CURLcode Curl_readwrite(struct connectdata *conn,
|
|
|
|
bool *done)
|
2000-05-22 10:09:31 -04:00
|
|
|
{
|
2002-01-03 10:01:22 -05:00
|
|
|
struct Curl_transfer_keeper *k = &conn->keep;
|
|
|
|
struct SessionHandle *data = conn->data;
|
2002-01-14 18:14:59 -05:00
|
|
|
int result;
|
2002-01-03 10:01:22 -05:00
|
|
|
ssize_t nread; /* number of bytes read */
|
|
|
|
int didwhat=0;
|
2002-05-28 10:18:36 -04:00
|
|
|
|
|
|
|
/* These two are used only if no other select() or _fdset() have been
|
|
|
|
invoked before this. This typicly happens if you use the multi interface
|
|
|
|
and call curl_multi_perform() without calling curl_multi_fdset()
|
|
|
|
first. */
|
|
|
|
fd_set extrareadfd;
|
|
|
|
fd_set extrawritefd;
|
|
|
|
|
|
|
|
fd_set *readfdp = k->readfdp;
|
|
|
|
fd_set *writefdp = k->writefdp;
|
2003-12-03 02:52:00 -05:00
|
|
|
long contentlength;
|
2002-05-28 10:18:36 -04:00
|
|
|
|
|
|
|
if((k->keepon & KEEP_READ) && !readfdp) {
|
|
|
|
/* reading is requested, but no socket descriptor pointer was set */
|
|
|
|
FD_ZERO(&extrareadfd);
|
|
|
|
FD_SET(conn->sockfd, &extrareadfd);
|
|
|
|
readfdp = &extrareadfd;
|
|
|
|
|
|
|
|
/* no write, no exceptions, no timeout */
|
|
|
|
select(conn->sockfd+1, readfdp, NULL, NULL, ¬imeout);
|
|
|
|
}
|
|
|
|
if((k->keepon & KEEP_WRITE) && !writefdp) {
|
|
|
|
/* writing is requested, but no socket descriptor pointer was set */
|
|
|
|
FD_ZERO(&extrawritefd);
|
|
|
|
FD_SET(conn->writesockfd, &extrawritefd);
|
|
|
|
writefdp = &extrawritefd;
|
|
|
|
|
|
|
|
/* no read, no exceptions, no timeout */
|
|
|
|
select(conn->writesockfd+1, NULL, writefdp, NULL, ¬imeout);
|
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
do {
|
2002-05-05 08:11:03 -04:00
|
|
|
/* If we still have reading to do, we check if we have a readable
|
2003-08-20 11:39:49 -04:00
|
|
|
socket. Sometimes the reafdp is NULL, if no fd_set was done using
|
2002-05-05 08:11:03 -04:00
|
|
|
the multi interface and then we can do nothing but to attempt a
|
|
|
|
read to be sure. */
|
2002-01-03 10:01:22 -05:00
|
|
|
if((k->keepon & KEEP_READ) &&
|
2003-08-20 11:39:49 -04:00
|
|
|
(!readfdp || FD_ISSET(conn->sockfd, readfdp))) {
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2003-02-26 07:42:25 -05:00
|
|
|
bool readdone = TRUE;
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/* This is where we loop until we have read everything there is to
|
|
|
|
read or we get a EWOULDBLOCK */
|
|
|
|
do {
|
2003-02-26 07:42:25 -05:00
|
|
|
int buffersize = data->set.buffer_size?
|
|
|
|
data->set.buffer_size:BUFSIZE -1;
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2003-02-26 07:42:25 -05:00
|
|
|
/* receive data from the network! */
|
|
|
|
result = Curl_read(conn, conn->sockfd, k->buf, buffersize, &nread);
|
2002-12-05 09:26:30 -05:00
|
|
|
|
|
|
|
if(0>result)
|
|
|
|
break; /* get out of loop */
|
|
|
|
if(result>0)
|
|
|
|
return result;
|
|
|
|
|
2003-02-24 11:53:53 -05:00
|
|
|
if ((k->bytecount == 0) && (k->writebytecount == 0)) {
|
2002-12-05 09:26:30 -05:00
|
|
|
Curl_pgrsTime(data, TIMER_STARTTRANSFER);
|
2003-02-24 11:53:53 -05:00
|
|
|
if(k->wait100_after_headers)
|
|
|
|
/* set time stamp to compare with when waiting for the 100 */
|
|
|
|
k->start100 = Curl_tvnow();
|
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
|
|
|
|
didwhat |= KEEP_READ;
|
|
|
|
|
|
|
|
/* NULL terminate, allowing string ops to be used */
|
|
|
|
if (0 < nread)
|
|
|
|
k->buf[nread] = 0;
|
|
|
|
|
|
|
|
/* if we receive 0 or less here, the server closed the connection and
|
|
|
|
we bail out from this! */
|
|
|
|
else if (0 >= nread) {
|
|
|
|
k->keepon &= ~KEEP_READ;
|
|
|
|
FD_ZERO(&k->rkeepfd);
|
|
|
|
readdone = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Default buffer to use when we write the buffer, it may be changed
|
|
|
|
in the flow below before the actual storing is done. */
|
|
|
|
k->str = k->buf;
|
|
|
|
|
|
|
|
/* Since this is a two-state thing, we check if we are parsing
|
|
|
|
headers at the moment or not. */
|
|
|
|
if (k->header) {
|
|
|
|
/* we are in parse-the-header-mode */
|
|
|
|
bool stop_reading = FALSE;
|
2002-01-14 18:14:59 -05:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/* header line within buffer loop */
|
|
|
|
do {
|
|
|
|
int hbufp_index;
|
2003-05-23 05:47:57 -04:00
|
|
|
int rest_length;
|
|
|
|
int full_length;
|
2003-06-12 19:03:08 -04:00
|
|
|
int writetype;
|
2002-01-03 10:01:22 -05:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/* str_start is start of line within buf */
|
|
|
|
k->str_start = k->str;
|
2002-01-03 10:01:22 -05:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
k->end_ptr = strchr (k->str_start, '\n');
|
2002-01-03 10:01:22 -05:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
if (!k->end_ptr) {
|
|
|
|
/* Not a complete header line within buffer, append the data to
|
|
|
|
the end of the headerbuff. */
|
|
|
|
|
|
|
|
if (k->hbuflen + nread >= data->state.headersize) {
|
|
|
|
/* We enlarge the header buffer as it is too small */
|
|
|
|
char *newbuff;
|
|
|
|
long newsize=MAX((k->hbuflen+nread)*3/2,
|
|
|
|
data->state.headersize*2);
|
|
|
|
hbufp_index = k->hbufp - data->state.headerbuff;
|
|
|
|
newbuff = (char *)realloc(data->state.headerbuff, newsize);
|
|
|
|
if(!newbuff) {
|
|
|
|
failf (data, "Failed to alloc memory for big header!");
|
|
|
|
return CURLE_OUT_OF_MEMORY;
|
|
|
|
}
|
|
|
|
data->state.headersize=newsize;
|
|
|
|
data->state.headerbuff = newbuff;
|
|
|
|
k->hbufp = data->state.headerbuff + hbufp_index;
|
|
|
|
}
|
|
|
|
memcpy(k->hbufp, k->str, nread);
|
|
|
|
k->hbufp += nread;
|
|
|
|
k->hbuflen += nread;
|
|
|
|
if (!k->headerline && (k->hbuflen>5)) {
|
|
|
|
/* make a first check that this looks like a HTTP header */
|
2003-01-07 11:15:53 -05:00
|
|
|
if(!checkhttpprefix(data, data->state.headerbuff)) {
|
2002-12-05 09:26:30 -05:00
|
|
|
/* this is not the beginning of a HTTP first header line */
|
|
|
|
k->header = FALSE;
|
|
|
|
k->badheader = HEADER_ALLBAD;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2003-01-15 06:44:33 -05:00
|
|
|
break; /* read more and try again */
|
2002-12-05 09:26:30 -05:00
|
|
|
}
|
|
|
|
|
2003-05-23 05:47:57 -04:00
|
|
|
/* decrease the size of the remaining (supposed) header line */
|
|
|
|
rest_length = (k->end_ptr - k->str)+1;
|
|
|
|
nread -= rest_length;
|
2002-09-16 10:02:08 -04:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
k->str = k->end_ptr + 1; /* move past new line */
|
|
|
|
|
2003-05-23 05:47:57 -04:00
|
|
|
full_length = k->str - k->str_start;
|
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/*
|
|
|
|
* We're about to copy a chunk of data to the end of the
|
|
|
|
* already received header. We make sure that the full string
|
|
|
|
* fit in the allocated header buffer, or else we enlarge
|
|
|
|
* it.
|
|
|
|
*/
|
2003-05-23 05:47:57 -04:00
|
|
|
if (k->hbuflen + full_length >=
|
2002-12-05 09:26:30 -05:00
|
|
|
data->state.headersize) {
|
2002-01-03 10:01:22 -05:00
|
|
|
char *newbuff;
|
2003-05-23 05:47:57 -04:00
|
|
|
long newsize=MAX((k->hbuflen+full_length)*3/2,
|
2002-01-03 10:01:22 -05:00
|
|
|
data->state.headersize*2);
|
|
|
|
hbufp_index = k->hbufp - data->state.headerbuff;
|
|
|
|
newbuff = (char *)realloc(data->state.headerbuff, newsize);
|
|
|
|
if(!newbuff) {
|
|
|
|
failf (data, "Failed to alloc memory for big header!");
|
2002-04-10 09:44:42 -04:00
|
|
|
return CURLE_OUT_OF_MEMORY;
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
data->state.headersize= newsize;
|
2002-01-03 10:01:22 -05:00
|
|
|
data->state.headerbuff = newbuff;
|
|
|
|
k->hbufp = data->state.headerbuff + hbufp_index;
|
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
|
|
|
|
/* copy to end of line */
|
2003-05-23 05:47:57 -04:00
|
|
|
strncpy (k->hbufp, k->str_start, full_length);
|
|
|
|
k->hbufp += full_length;
|
|
|
|
k->hbuflen += full_length;
|
2002-12-05 09:26:30 -05:00
|
|
|
*k->hbufp = 0;
|
2003-06-26 07:30:26 -04:00
|
|
|
k->end_ptr = k->hbufp;
|
2002-12-05 09:26:30 -05:00
|
|
|
|
|
|
|
k->p = data->state.headerbuff;
|
|
|
|
|
|
|
|
/****
|
|
|
|
* We now have a FULL header line that p points to
|
|
|
|
*****/
|
|
|
|
|
|
|
|
if(!k->headerline) {
|
|
|
|
/* the first read header */
|
|
|
|
if((k->hbuflen>5) &&
|
2003-01-07 11:15:53 -05:00
|
|
|
!checkhttpprefix(data, data->state.headerbuff)) {
|
2002-09-13 08:40:36 -04:00
|
|
|
/* this is not the beginning of a HTTP first header line */
|
|
|
|
k->header = FALSE;
|
2003-05-23 05:47:57 -04:00
|
|
|
if(nread)
|
|
|
|
/* since there's more, this is a partial bad header */
|
|
|
|
k->badheader = HEADER_PARTHEADER;
|
|
|
|
else {
|
|
|
|
/* this was all we read so its all a bad header */
|
|
|
|
k->badheader = HEADER_ALLBAD;
|
|
|
|
nread = rest_length;
|
|
|
|
}
|
2002-09-13 08:40:36 -04:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
if (('\n' == *k->p) || ('\r' == *k->p)) {
|
|
|
|
int headerlen;
|
|
|
|
/* Zero-length header line means end of headers! */
|
|
|
|
|
|
|
|
if ('\r' == *k->p)
|
|
|
|
k->p++; /* pass the \r byte */
|
|
|
|
if ('\n' == *k->p)
|
|
|
|
k->p++; /* pass the \n byte */
|
|
|
|
|
|
|
|
if(100 == k->httpcode) {
|
|
|
|
/*
|
2003-02-24 11:53:53 -05:00
|
|
|
* We have made a HTTP PUT or POST and this is 1.1-lingo
|
2002-12-05 09:26:30 -05:00
|
|
|
* that tells us that the server is OK with this and ready
|
2003-02-24 11:53:53 -05:00
|
|
|
* to receive the data.
|
2002-12-05 09:26:30 -05:00
|
|
|
* However, we'll get more headers now so we must get
|
|
|
|
* back into the header-parsing state!
|
|
|
|
*/
|
|
|
|
k->header = TRUE;
|
|
|
|
k->headerline = 0; /* restart the header line counter */
|
|
|
|
/* if we did wait for this do enable write now! */
|
|
|
|
if (k->write_after_100_header) {
|
|
|
|
|
|
|
|
k->write_after_100_header = FALSE;
|
|
|
|
FD_SET (conn->writesockfd, &k->writefd); /* write */
|
|
|
|
k->keepon |= KEEP_WRITE;
|
|
|
|
k->wkeepfd = k->writefd;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
k->header = FALSE; /* no more header to parse! */
|
|
|
|
|
|
|
|
if (417 == k->httpcode) {
|
|
|
|
/*
|
|
|
|
* we got: "417 Expectation Failed" this means:
|
|
|
|
* we have made a HTTP call and our Expect Header
|
|
|
|
* seems to cause a problem => abort the write operations
|
|
|
|
* (or prevent them from starting).
|
|
|
|
*/
|
|
|
|
k->write_after_100_header = FALSE;
|
|
|
|
k->keepon &= ~KEEP_WRITE;
|
|
|
|
FD_ZERO(&k->wkeepfd);
|
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/* now, only output this if the header AND body are requested:
|
|
|
|
*/
|
2003-06-12 19:03:08 -04:00
|
|
|
writetype = CLIENTWRITE_HEADER;
|
2002-12-05 09:26:30 -05:00
|
|
|
if (data->set.http_include_header)
|
2003-06-12 19:03:08 -04:00
|
|
|
writetype |= CLIENTWRITE_BODY;
|
2002-12-05 09:26:30 -05:00
|
|
|
|
|
|
|
headerlen = k->p - data->state.headerbuff;
|
|
|
|
|
2003-06-12 19:03:08 -04:00
|
|
|
result = Curl_client_write(data, writetype,
|
2002-12-05 09:26:30 -05:00
|
|
|
data->state.headerbuff,
|
|
|
|
headerlen);
|
|
|
|
if(result)
|
|
|
|
return result;
|
|
|
|
|
|
|
|
data->info.header_size += headerlen;
|
|
|
|
conn->headerbytecount += headerlen;
|
|
|
|
|
2003-10-22 07:15:48 -04:00
|
|
|
/* *auth_act() checks what authentication methods that are
|
|
|
|
available and decides which one (if any) to use. It will
|
|
|
|
set 'newurl' if an auth metod was picked. */
|
|
|
|
Curl_http_auth_act(conn);
|
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
if(!k->header) {
|
|
|
|
/*
|
|
|
|
* really end-of-headers.
|
|
|
|
*
|
|
|
|
* If we requested a "no body", this is a good time to get
|
|
|
|
* out and return home.
|
|
|
|
*/
|
|
|
|
if(data->set.no_body)
|
|
|
|
stop_reading = TRUE;
|
2003-06-02 10:57:08 -04:00
|
|
|
else {
|
|
|
|
/* If we know the expected size of this document, we set the
|
|
|
|
maximum download size to the size of the expected
|
|
|
|
document or else, we won't know when to stop reading!
|
|
|
|
|
|
|
|
Note that we set the download maximum even if we read a
|
|
|
|
"Connection: close" header, to make sure that
|
|
|
|
"Content-Length: 0" still prevents us from attempting to
|
|
|
|
read the (missing) response-body.
|
|
|
|
*/
|
2003-12-03 02:52:00 -05:00
|
|
|
/* According to RFC2616 section 4.4, we MUST ignore
|
|
|
|
Content-Length: headers if we are now receiving data
|
|
|
|
using chunked Transfer-Encoding.
|
|
|
|
*/
|
|
|
|
if(conn->bits.chunk)
|
|
|
|
conn->size=-1;
|
|
|
|
|
|
|
|
if(-1 != conn->size) {
|
|
|
|
Curl_pgrsSetDownloadSize(data, conn->size);
|
2002-12-05 09:26:30 -05:00
|
|
|
conn->maxdownload = conn->size;
|
2003-12-03 02:52:00 -05:00
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
}
|
|
|
|
/* If max download size is *zero* (nothing) we already
|
|
|
|
have nothing and can safely return ok now! */
|
|
|
|
if(0 == conn->maxdownload)
|
|
|
|
stop_reading = TRUE;
|
|
|
|
|
|
|
|
if(stop_reading) {
|
|
|
|
/* we make sure that this socket isn't read more now */
|
|
|
|
k->keepon &= ~KEEP_READ;
|
|
|
|
FD_ZERO(&k->rkeepfd);
|
|
|
|
}
|
2002-10-09 09:03:51 -04:00
|
|
|
|
2003-01-15 06:44:33 -05:00
|
|
|
break; /* exit header line loop */
|
2002-12-05 09:26:30 -05:00
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/* We continue reading headers, so reset the line-based
|
|
|
|
header parsing variables hbufp && hbuflen */
|
|
|
|
k->hbufp = data->state.headerbuff;
|
|
|
|
k->hbuflen = 0;
|
|
|
|
continue;
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/*
|
|
|
|
* Checks for special headers coming up.
|
|
|
|
*/
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
if (!k->headerline++) {
|
|
|
|
/* This is the first header, it MUST be the error code line
|
|
|
|
or else we consiser this to be the body right away! */
|
|
|
|
int httpversion_major;
|
|
|
|
int nc=sscanf (k->p, " HTTP/%d.%d %3d",
|
|
|
|
&httpversion_major,
|
|
|
|
&k->httpversion,
|
|
|
|
&k->httpcode);
|
|
|
|
if (nc==3) {
|
|
|
|
k->httpversion += 10 * httpversion_major;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
/* this is the real world, not a Nirvana
|
|
|
|
NCSA 1.5.x returns this crap when asked for HTTP/1.1
|
|
|
|
*/
|
|
|
|
nc=sscanf (k->p, " HTTP %3d", &k->httpcode);
|
|
|
|
k->httpversion = 10;
|
2003-01-07 11:15:53 -05:00
|
|
|
|
|
|
|
/* If user has set option HTTP200ALIASES,
|
|
|
|
compare header line against list of aliases
|
|
|
|
*/
|
|
|
|
if (!nc) {
|
|
|
|
if (checkhttpprefix(data, k->p)) {
|
|
|
|
nc = 1;
|
|
|
|
k->httpcode = 200;
|
|
|
|
k->httpversion =
|
|
|
|
(data->set.httpversion==CURL_HTTP_VERSION_1_0)? 10 : 11;
|
|
|
|
}
|
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
}
|
2002-01-29 15:28:59 -05:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
if (nc) {
|
|
|
|
data->info.httpcode = k->httpcode;
|
|
|
|
data->info.httpversion = k->httpversion;
|
|
|
|
|
|
|
|
/* 404 -> URL not found! */
|
|
|
|
if (data->set.http_fail_on_error &&
|
|
|
|
(k->httpcode >= 400)) {
|
|
|
|
/* If we have been told to fail hard on HTTP-errors,
|
|
|
|
here is the check for that: */
|
|
|
|
/* serious error, go home! */
|
2003-06-02 09:14:57 -04:00
|
|
|
failf (data, "The requested URL returned error: %d",
|
|
|
|
k->httpcode);
|
2002-12-18 11:51:02 -05:00
|
|
|
return CURLE_HTTP_RETURNED_ERROR;
|
2002-12-05 09:26:30 -05:00
|
|
|
}
|
|
|
|
|
|
|
|
if(k->httpversion == 10)
|
|
|
|
/* Default action for HTTP/1.0 must be to close, unless
|
|
|
|
we get one of those fancy headers that tell us the
|
|
|
|
server keeps it open for us! */
|
|
|
|
conn->bits.close = TRUE;
|
|
|
|
|
|
|
|
switch(k->httpcode) {
|
|
|
|
case 204:
|
|
|
|
/* (quote from RFC2616, section 10.2.5): The server has
|
|
|
|
* fulfilled the request but does not need to return an
|
|
|
|
* entity-body ... The 204 response MUST NOT include a
|
|
|
|
* message-body, and thus is always terminated by the first
|
|
|
|
* empty line after the header fields. */
|
|
|
|
/* FALLTHROUGH */
|
2003-10-24 17:54:34 -04:00
|
|
|
case 416: /* Requested Range Not Satisfiable, it has the
|
|
|
|
Content-Length: set as the "real" document but no
|
|
|
|
actual response is sent. */
|
2002-12-05 09:26:30 -05:00
|
|
|
case 304:
|
2003-10-24 17:54:34 -04:00
|
|
|
/* (quote from RFC2616, section 10.3.5): The 304 response
|
|
|
|
* MUST NOT contain a message-body, and thus is always
|
|
|
|
* terminated by the first empty line after the header
|
|
|
|
* fields. */
|
2002-12-05 09:26:30 -05:00
|
|
|
conn->size=0;
|
|
|
|
conn->maxdownload=0;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
/* nothing */
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
k->header = FALSE; /* this is not a header line */
|
|
|
|
break;
|
2000-05-22 10:09:31 -04:00
|
|
|
}
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2003-10-24 17:54:34 -04:00
|
|
|
/* Check for Content-Length: header lines to get size. Ignore
|
|
|
|
the header completely if we get a 416 response as then we're
|
|
|
|
resuming a document that we don't get, and this header contains
|
|
|
|
info about the true size of the document we didn't get now. */
|
|
|
|
if ((k->httpcode != 416) &&
|
2004-01-05 17:29:29 -05:00
|
|
|
checkprefix("Content-Length:", k->p)) {
|
|
|
|
contentlength = strtoofft(k->p+15, NULL, 10);
|
2003-12-03 02:52:00 -05:00
|
|
|
if (data->set.max_filesize && contentlength >
|
2003-10-17 09:11:00 -04:00
|
|
|
data->set.max_filesize) {
|
|
|
|
failf(data, "Maximum file size exceeded");
|
|
|
|
return CURLE_FILESIZE_EXCEEDED;
|
|
|
|
}
|
2003-12-03 02:52:00 -05:00
|
|
|
conn->size = contentlength;
|
2002-12-05 09:26:30 -05:00
|
|
|
}
|
|
|
|
/* check for Content-Type: header lines to get the mime-type */
|
|
|
|
else if (checkprefix("Content-Type:", k->p)) {
|
|
|
|
char *start;
|
|
|
|
char *end;
|
|
|
|
int len;
|
|
|
|
|
|
|
|
/* Find the first non-space letter */
|
2003-04-08 10:48:38 -04:00
|
|
|
for(start=k->p+13;
|
2002-12-05 09:26:30 -05:00
|
|
|
*start && isspace((int)*start);
|
|
|
|
start++);
|
|
|
|
|
2003-01-30 00:04:02 -05:00
|
|
|
end = strchr(start, '\r');
|
|
|
|
if(!end)
|
|
|
|
end = strchr(start, '\n');
|
|
|
|
|
|
|
|
if(end) {
|
|
|
|
/* skip all trailing space letters */
|
2003-01-30 01:06:24 -05:00
|
|
|
for(; isspace((int)*end) && (end > start); end--);
|
2003-01-30 00:04:02 -05:00
|
|
|
|
|
|
|
/* get length of the type */
|
|
|
|
len = end-start+1;
|
2003-01-24 06:13:59 -05:00
|
|
|
|
2003-01-30 00:04:02 -05:00
|
|
|
/* allocate memory of a cloned copy */
|
2003-08-05 10:40:59 -04:00
|
|
|
Curl_safefree(data->info.contenttype);
|
2003-01-30 00:04:02 -05:00
|
|
|
|
|
|
|
data->info.contenttype = malloc(len + 1);
|
|
|
|
if (NULL == data->info.contenttype)
|
|
|
|
return CURLE_OUT_OF_MEMORY;
|
2002-12-05 09:26:30 -05:00
|
|
|
|
2003-01-30 00:04:02 -05:00
|
|
|
/* copy the content-type string */
|
|
|
|
memcpy(data->info.contenttype, start, len);
|
|
|
|
data->info.contenttype[len] = 0; /* zero terminate */
|
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
}
|
|
|
|
else if((k->httpversion == 10) &&
|
|
|
|
conn->bits.httpproxy &&
|
|
|
|
Curl_compareheader(k->p,
|
|
|
|
"Proxy-Connection:", "keep-alive")) {
|
2002-01-03 10:01:22 -05:00
|
|
|
/*
|
2002-12-05 09:26:30 -05:00
|
|
|
* When a HTTP/1.0 reply comes when using a proxy, the
|
|
|
|
* 'Proxy-Connection: keep-alive' line tells us the
|
|
|
|
* connection will be kept alive for our pleasure.
|
|
|
|
* Default action for 1.0 is to close.
|
2002-01-03 10:01:22 -05:00
|
|
|
*/
|
2002-12-05 09:26:30 -05:00
|
|
|
conn->bits.close = FALSE; /* don't close when done */
|
|
|
|
infof(data, "HTTP/1.0 proxy connection set to keep alive!\n");
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
else if((k->httpversion == 10) &&
|
|
|
|
Curl_compareheader(k->p, "Connection:", "keep-alive")) {
|
2001-05-12 12:11:14 -04:00
|
|
|
/*
|
2002-12-05 09:26:30 -05:00
|
|
|
* A HTTP/1.0 reply with the 'Connection: keep-alive' line
|
|
|
|
* tells us the connection will be kept alive for our
|
|
|
|
* pleasure. Default action for 1.0 is to close.
|
2002-01-03 10:01:22 -05:00
|
|
|
*
|
2002-12-05 09:26:30 -05:00
|
|
|
* [RFC2068, section 19.7.1] */
|
|
|
|
conn->bits.close = FALSE; /* don't close when done */
|
|
|
|
infof(data, "HTTP/1.0 connection set to keep alive!\n");
|
|
|
|
}
|
|
|
|
else if (Curl_compareheader(k->p, "Connection:", "close")) {
|
|
|
|
/*
|
|
|
|
* [RFC 2616, section 8.1.2.1]
|
|
|
|
* "Connection: close" is HTTP/1.1 language and means that
|
|
|
|
* the connection will close when this request has been
|
|
|
|
* served.
|
2001-05-12 12:11:14 -04:00
|
|
|
*/
|
2002-12-05 09:26:30 -05:00
|
|
|
conn->bits.close = TRUE; /* close when done */
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
else if (Curl_compareheader(k->p,
|
|
|
|
"Transfer-Encoding:", "chunked")) {
|
|
|
|
/*
|
|
|
|
* [RFC 2616, section 3.6.1] A 'chunked' transfer encoding
|
|
|
|
* means that the server will send a series of "chunks". Each
|
|
|
|
* chunk starts with line with info (including size of the
|
|
|
|
* coming block) (terminated with CRLF), then a block of data
|
|
|
|
* with the previously mentioned size. There can be any amount
|
|
|
|
* of chunks, and a chunk-data set to zero signals the
|
|
|
|
* end-of-chunks. */
|
|
|
|
conn->bits.chunk = TRUE; /* chunks coming our way */
|
|
|
|
|
|
|
|
/* init our chunky engine */
|
|
|
|
Curl_httpchunk_init(conn);
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
else if (checkprefix("Content-Encoding:", k->p) &&
|
|
|
|
data->set.encoding) {
|
|
|
|
/*
|
2003-05-01 13:49:47 -04:00
|
|
|
* Process Content-Encoding. Look for the values: identity,
|
|
|
|
* gzip, deflate, compress, x-gzip and x-compress. x-gzip and
|
2002-12-05 09:26:30 -05:00
|
|
|
* x-compress are the same as gzip and compress. (Sec 3.5 RFC
|
2003-05-01 13:49:47 -04:00
|
|
|
* 2616). zlib cannot handle compress. However, errors are
|
|
|
|
* handled further down when the response body is processed
|
|
|
|
*/
|
2002-12-05 09:26:30 -05:00
|
|
|
char *start;
|
|
|
|
|
|
|
|
/* Find the first non-space letter */
|
|
|
|
for(start=k->p+17;
|
|
|
|
*start && isspace((int)*start);
|
|
|
|
start++);
|
|
|
|
|
|
|
|
/* Record the content-encoding for later use. 08/27/02 jhrg */
|
|
|
|
if (checkprefix("identity", start))
|
|
|
|
k->content_encoding = IDENTITY;
|
|
|
|
else if (checkprefix("deflate", start))
|
|
|
|
k->content_encoding = DEFLATE;
|
|
|
|
else if (checkprefix("gzip", start)
|
|
|
|
|| checkprefix("x-gzip", start))
|
|
|
|
k->content_encoding = GZIP;
|
|
|
|
else if (checkprefix("compress", start)
|
|
|
|
|| checkprefix("x-compress", start))
|
|
|
|
k->content_encoding = COMPRESS;
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
2004-01-05 17:29:29 -05:00
|
|
|
else if (Curl_compareheader(k->p, "Content-Range:", "bytes")) {
|
|
|
|
/* Content-Range: bytes [num]-
|
|
|
|
Content-Range: bytes: [num]-
|
|
|
|
|
|
|
|
The second format was added August 1st 2000 by Igor
|
|
|
|
Khristophorov since Sun's webserver JavaWebServer/1.1.1
|
|
|
|
obviously sends the header this way! :-( */
|
|
|
|
|
|
|
|
char *ptr = strstr(k->p, "bytes");
|
|
|
|
ptr+=5;
|
|
|
|
|
|
|
|
if(*ptr == ':')
|
|
|
|
/* stupid colon skip */
|
|
|
|
ptr++;
|
|
|
|
|
|
|
|
k->offset = strtoofft(ptr, NULL, 10);
|
|
|
|
|
|
|
|
if (conn->resume_from == k->offset)
|
|
|
|
/* we asked for a resume and we got it */
|
|
|
|
k->content_range = TRUE;
|
2002-09-13 08:40:36 -04:00
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
else if(data->cookies &&
|
|
|
|
checkprefix("Set-Cookie:", k->p)) {
|
2003-08-06 11:26:24 -04:00
|
|
|
Curl_share_lock(data, CURL_LOCK_DATA_COOKIE,
|
|
|
|
CURL_LOCK_ACCESS_SINGLE);
|
2003-08-11 05:56:06 -04:00
|
|
|
Curl_cookie_add(data,
|
|
|
|
data->cookies, TRUE, k->p+11,
|
2003-04-30 13:05:19 -04:00
|
|
|
/* If there is a custom-set Host: name, use it
|
|
|
|
here, or else use real peer host name. */
|
|
|
|
conn->allocptr.cookiehost?
|
|
|
|
conn->allocptr.cookiehost:conn->name,
|
|
|
|
conn->ppath);
|
2003-08-04 11:02:42 -04:00
|
|
|
Curl_share_unlock(data, CURL_LOCK_DATA_COOKIE);
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
else if(checkprefix("Last-Modified:", k->p) &&
|
|
|
|
(data->set.timecondition || data->set.get_filetime) ) {
|
|
|
|
time_t secs=time(NULL);
|
2002-12-09 10:37:54 -05:00
|
|
|
k->timeofdoc = curl_getdate(k->p+strlen("Last-Modified:"),
|
|
|
|
&secs);
|
|
|
|
if(data->set.get_filetime)
|
|
|
|
data->info.filetime = k->timeofdoc;
|
|
|
|
}
|
2003-08-11 07:47:45 -04:00
|
|
|
else if((checkprefix("WWW-Authenticate:", k->p) &&
|
2003-09-03 17:51:28 -04:00
|
|
|
(401 == k->httpcode)) ||
|
2003-08-11 07:47:45 -04:00
|
|
|
(checkprefix("Proxy-authenticate:", k->p) &&
|
2003-09-03 17:51:28 -04:00
|
|
|
(407 == k->httpcode))) {
|
|
|
|
result = Curl_http_auth(conn, k->httpcode, k->p);
|
|
|
|
if(result)
|
|
|
|
return result;
|
2003-05-22 12:09:54 -04:00
|
|
|
}
|
2002-12-09 10:37:54 -05:00
|
|
|
else if ((k->httpcode >= 300 && k->httpcode < 400) &&
|
|
|
|
checkprefix("Location:", k->p)) {
|
2003-05-22 12:09:54 -04:00
|
|
|
if(data->set.http_follow_location) {
|
|
|
|
/* this is the URL that the server advices us to get instead */
|
|
|
|
char *ptr;
|
|
|
|
char *start=k->p;
|
|
|
|
char backup;
|
|
|
|
|
|
|
|
start += 9; /* pass "Location:" */
|
|
|
|
|
|
|
|
/* Skip spaces and tabs. We do this to support multiple
|
|
|
|
white spaces after the "Location:" keyword. */
|
|
|
|
while(*start && isspace((int)*start ))
|
|
|
|
start++;
|
2003-06-26 07:30:26 -04:00
|
|
|
|
|
|
|
/* Scan through the string from the end to find the last
|
|
|
|
non-space. k->end_ptr points to the actual terminating zero
|
|
|
|
letter, move pointer one letter back and start from
|
|
|
|
there. This logic strips off trailing whitespace, but keeps
|
|
|
|
any embedded whitespace. */
|
|
|
|
ptr = k->end_ptr-1;
|
|
|
|
while((ptr>=start) && isspace((int)*ptr))
|
|
|
|
ptr--;
|
|
|
|
ptr++;
|
2003-05-22 12:09:54 -04:00
|
|
|
|
|
|
|
backup = *ptr; /* store the ending letter */
|
|
|
|
if(ptr != start) {
|
|
|
|
*ptr = '\0'; /* zero terminate */
|
|
|
|
conn->newurl = strdup(start); /* clone string */
|
|
|
|
*ptr = backup; /* restore ending letter */
|
|
|
|
}
|
2002-12-09 10:37:54 -05:00
|
|
|
}
|
2003-05-22 12:09:54 -04:00
|
|
|
#if 0 /* for consideration */
|
|
|
|
else {
|
|
|
|
/* This is a Location: but we have not been instructed to
|
|
|
|
follow it */
|
|
|
|
infof(data, "We ignore this location header as instructed\n");
|
|
|
|
}
|
|
|
|
#endif
|
2002-10-18 09:51:00 -04:00
|
|
|
}
|
2000-11-22 07:54:48 -05:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
/*
|
|
|
|
* End of header-checks. Write them to the client.
|
|
|
|
*/
|
2001-05-12 12:11:14 -04:00
|
|
|
|
2003-06-12 19:03:08 -04:00
|
|
|
writetype = CLIENTWRITE_HEADER;
|
2002-12-09 10:37:54 -05:00
|
|
|
if (data->set.http_include_header)
|
2003-06-12 19:03:08 -04:00
|
|
|
writetype |= CLIENTWRITE_BODY;
|
2000-11-22 07:54:48 -05:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
if(data->set.verbose)
|
|
|
|
Curl_debug(data, CURLINFO_HEADER_IN,
|
|
|
|
k->p, k->hbuflen);
|
2002-05-03 08:07:32 -04:00
|
|
|
|
2003-06-12 19:03:08 -04:00
|
|
|
result = Curl_client_write(data, writetype, k->p, k->hbuflen);
|
2002-12-09 10:37:54 -05:00
|
|
|
if(result)
|
|
|
|
return result;
|
2000-11-22 07:54:48 -05:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
data->info.header_size += k->hbuflen;
|
|
|
|
conn->headerbytecount += k->hbuflen;
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
/* reset hbufp pointer && hbuflen */
|
|
|
|
k->hbufp = data->state.headerbuff;
|
|
|
|
k->hbuflen = 0;
|
|
|
|
}
|
|
|
|
while (!stop_reading && *k->str); /* header line within buffer */
|
2002-04-04 07:19:56 -05:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
if(stop_reading)
|
|
|
|
/* We've stopped dealing with input, get out of the do-while loop */
|
|
|
|
break;
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
/* We might have reached the end of the header part here, but
|
|
|
|
there might be a non-header part left in the end of the read
|
|
|
|
buffer. */
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2003-01-15 06:44:33 -05:00
|
|
|
} /* end if header mode */
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
/* This is not an 'else if' since it may be a rest from the header
|
|
|
|
parsing, where the beginning of the buffer is headers and the end
|
|
|
|
is non-headers. */
|
|
|
|
if (k->str && !k->header && (nread > 0)) {
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
if(0 == k->bodywrites) {
|
|
|
|
/* These checks are only made the first time we are about to
|
|
|
|
write a piece of the body */
|
|
|
|
if(conn->protocol&PROT_HTTP) {
|
|
|
|
/* HTTP-only checks */
|
2003-06-26 07:30:26 -04:00
|
|
|
|
2003-06-12 19:03:08 -04:00
|
|
|
if (conn->newurl) {
|
|
|
|
if(conn->bits.close) {
|
|
|
|
/* Abort after the headers if "follow Location" is set
|
|
|
|
and we're set to close anyway. */
|
|
|
|
k->keepon &= ~KEEP_READ;
|
|
|
|
FD_ZERO(&k->rkeepfd);
|
|
|
|
*done = TRUE;
|
|
|
|
return CURLE_OK;
|
|
|
|
}
|
|
|
|
/* We have a new url to load, but since we want to be able
|
|
|
|
to re-use this connection properly, we read the full
|
|
|
|
response in "ignore more" */
|
|
|
|
k->ignorebody = TRUE;
|
|
|
|
infof(data, "Ignoring the response-body\n");
|
2002-12-09 10:37:54 -05:00
|
|
|
}
|
2003-06-12 19:03:08 -04:00
|
|
|
if (conn->resume_from &&
|
2002-12-09 10:37:54 -05:00
|
|
|
!k->content_range &&
|
|
|
|
(data->set.httpreq==HTTPREQ_GET)) {
|
|
|
|
/* we wanted to resume a download, although the server
|
|
|
|
doesn't seem to support this and we did this with a GET
|
|
|
|
(if it wasn't a GET we did a POST or PUT resume) */
|
|
|
|
failf (data, "HTTP server doesn't seem to support "
|
|
|
|
"byte ranges. Cannot resume.");
|
|
|
|
return CURLE_HTTP_RANGE_ERROR;
|
|
|
|
}
|
|
|
|
else if(data->set.timecondition && !conn->range) {
|
|
|
|
/* A time condition has been set AND no ranges have been
|
|
|
|
requested. This seems to be what chapter 13.3.4 of
|
|
|
|
RFC 2616 defines to be the correct action for a
|
|
|
|
HTTP/1.1 client */
|
|
|
|
if((k->timeofdoc > 0) && (data->set.timevalue > 0)) {
|
|
|
|
switch(data->set.timecondition) {
|
|
|
|
case TIMECOND_IFMODSINCE:
|
|
|
|
default:
|
|
|
|
if(k->timeofdoc < data->set.timevalue) {
|
|
|
|
infof(data,
|
|
|
|
"The requested document is not new enough\n");
|
|
|
|
*done = TRUE;
|
|
|
|
return CURLE_OK;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case TIMECOND_IFUNMODSINCE:
|
|
|
|
if(k->timeofdoc > data->set.timevalue) {
|
|
|
|
infof(data,
|
|
|
|
"The requested document is not old enough\n");
|
|
|
|
*done = TRUE;
|
|
|
|
return CURLE_OK;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
} /* switch */
|
|
|
|
} /* two valid time strings */
|
|
|
|
} /* we have a time condition */
|
|
|
|
|
|
|
|
} /* this is HTTP */
|
|
|
|
} /* this is the first time we write a body part */
|
|
|
|
k->bodywrites++;
|
|
|
|
|
|
|
|
/* pass data to the debug function before it gets "dechunked" */
|
|
|
|
if(data->set.verbose) {
|
|
|
|
if(k->badheader) {
|
|
|
|
Curl_debug(data, CURLINFO_DATA_IN, data->state.headerbuff,
|
|
|
|
k->hbuflen);
|
|
|
|
if(k->badheader == HEADER_PARTHEADER)
|
|
|
|
Curl_debug(data, CURLINFO_DATA_IN, k->str, nread);
|
2001-03-07 12:12:12 -05:00
|
|
|
}
|
2002-12-09 10:37:54 -05:00
|
|
|
else
|
2002-10-23 09:48:37 -04:00
|
|
|
Curl_debug(data, CURLINFO_DATA_IN, k->str, nread);
|
|
|
|
}
|
2002-05-03 08:07:32 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
if(conn->bits.chunk) {
|
|
|
|
/*
|
|
|
|
* Bless me father for I have sinned. Here comes a chunked
|
|
|
|
* transfer flying and we need to decode this properly. While
|
|
|
|
* the name says read, this function both reads and writes away
|
|
|
|
* the data. The returned 'nread' holds the number of actual
|
|
|
|
* data it wrote to the client. */
|
|
|
|
CHUNKcode res =
|
|
|
|
Curl_httpchunk_read(conn, k->str, nread, &nread);
|
|
|
|
|
|
|
|
if(CHUNKE_OK < res) {
|
|
|
|
if(CHUNKE_WRITE_ERROR == res) {
|
|
|
|
failf(data, "Failed writing data");
|
|
|
|
return CURLE_WRITE_ERROR;
|
|
|
|
}
|
2003-01-21 11:03:38 -05:00
|
|
|
failf(data, "Received problem %d in the chunky parser", res);
|
2002-12-09 10:37:54 -05:00
|
|
|
return CURLE_RECV_ERROR;
|
2002-03-15 07:42:41 -05:00
|
|
|
}
|
2002-12-09 10:37:54 -05:00
|
|
|
else if(CHUNKE_STOP == res) {
|
|
|
|
/* we're done reading chunks! */
|
|
|
|
k->keepon &= ~KEEP_READ; /* read no more */
|
|
|
|
FD_ZERO(&k->rkeepfd);
|
2002-01-03 10:01:22 -05:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
/* There are now possibly N number of bytes at the end of the
|
|
|
|
str buffer that weren't written to the client, but we don't
|
|
|
|
care about them right now. */
|
|
|
|
}
|
|
|
|
/* If it returned OK, we just keep going */
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
if((-1 != conn->maxdownload) &&
|
|
|
|
(k->bytecount + nread >= conn->maxdownload)) {
|
2004-01-05 17:29:29 -05:00
|
|
|
nread = (ssize_t) (conn->maxdownload - k->bytecount);
|
2002-12-09 10:37:54 -05:00
|
|
|
if(nread < 0 ) /* this should be unusual */
|
|
|
|
nread = 0;
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
k->keepon &= ~KEEP_READ; /* we're done reading */
|
|
|
|
FD_ZERO(&k->rkeepfd);
|
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
k->bytecount += nread;
|
2002-01-03 10:01:22 -05:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
Curl_pgrsSetDownloadCounter(data, (double)k->bytecount);
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
if(!conn->bits.chunk && (nread || k->badheader)) {
|
|
|
|
/* If this is chunky transfer, it was already written */
|
|
|
|
|
2003-06-12 19:03:08 -04:00
|
|
|
if(k->badheader && !k->ignorebody) {
|
2002-12-09 10:37:54 -05:00
|
|
|
/* we parsed a piece of data wrongly assuming it was a header
|
|
|
|
and now we output it as body instead */
|
|
|
|
result = Curl_client_write(data, CLIENTWRITE_BODY,
|
|
|
|
data->state.headerbuff,
|
|
|
|
k->hbuflen);
|
|
|
|
}
|
|
|
|
if(k->badheader < HEADER_ALLBAD) {
|
|
|
|
/* This switch handles various content encodings. If there's an
|
|
|
|
error here, be sure to check over the almost identical code
|
2003-05-12 08:45:14 -04:00
|
|
|
in http_chunks.c. 08/29/02 jhrg
|
|
|
|
Make sure that ALL_CONTENT_ENCODINGS contains all the
|
|
|
|
encodings handled here. */
|
2002-09-02 18:31:18 -04:00
|
|
|
#ifdef HAVE_LIBZ
|
2002-12-09 10:37:54 -05:00
|
|
|
switch (k->content_encoding) {
|
|
|
|
case IDENTITY:
|
2002-09-02 18:31:18 -04:00
|
|
|
#endif
|
2002-12-09 10:37:54 -05:00
|
|
|
/* This is the default when the server sends no
|
|
|
|
Content-Encoding header. See Curl_readwrite_init; the
|
|
|
|
memset() call initializes k->content_encoding to zero.
|
|
|
|
08/28/02 jhrg */
|
2003-06-12 19:03:08 -04:00
|
|
|
if(!k->ignorebody)
|
|
|
|
result = Curl_client_write(data, CLIENTWRITE_BODY, k->str,
|
|
|
|
nread);
|
2002-09-02 18:31:18 -04:00
|
|
|
#ifdef HAVE_LIBZ
|
2002-12-09 10:37:54 -05:00
|
|
|
break;
|
|
|
|
|
|
|
|
case DEFLATE:
|
|
|
|
/* Assume CLIENTWRITE_BODY; headers are not encoded. */
|
|
|
|
result = Curl_unencode_deflate_write(data, k, nread);
|
|
|
|
break;
|
|
|
|
|
2003-04-11 04:49:20 -04:00
|
|
|
case GZIP:
|
|
|
|
/* Assume CLIENTWRITE_BODY; headers are not encoded. */
|
|
|
|
result = Curl_unencode_gzip_write(data, k, nread);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case COMPRESS: /* FIXME 08/27/02 jhrg */
|
2002-12-09 10:37:54 -05:00
|
|
|
default:
|
|
|
|
failf (data, "Unrecognized content encoding type. "
|
2003-04-22 18:33:39 -04:00
|
|
|
"libcurl understands `identity', `deflate' and `gzip' "
|
2002-12-09 10:37:54 -05:00
|
|
|
"content encodings.");
|
|
|
|
result = CURLE_BAD_CONTENT_ENCODING;
|
|
|
|
break;
|
|
|
|
}
|
2002-09-02 18:31:18 -04:00
|
|
|
#endif
|
2002-12-09 10:37:54 -05:00
|
|
|
}
|
|
|
|
k->badheader = HEADER_NORMAL; /* taken care of now */
|
2002-09-02 18:31:18 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
if(result)
|
|
|
|
return result;
|
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
} /* if (! header and data to read ) */
|
2002-12-05 09:26:30 -05:00
|
|
|
|
|
|
|
} while(!readdone);
|
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
} /* if( read from socket ) */
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-05-05 08:11:03 -04:00
|
|
|
/* If we still have writing to do, we check if we have a writable
|
2003-08-20 11:39:49 -04:00
|
|
|
socket. Sometimes the writefdp is NULL, if no fd_set was done using
|
2002-05-05 08:11:03 -04:00
|
|
|
the multi interface and then we can do nothing but to attempt a
|
|
|
|
write to be sure. */
|
2002-01-03 10:01:22 -05:00
|
|
|
if((k->keepon & KEEP_WRITE) &&
|
2003-08-20 11:39:49 -04:00
|
|
|
(!writefdp || FD_ISSET(conn->writesockfd, writefdp)) ) {
|
2002-01-03 10:01:22 -05:00
|
|
|
/* write */
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
int i, si;
|
2002-01-16 09:50:53 -05:00
|
|
|
ssize_t bytes_written;
|
2003-05-12 08:37:05 -04:00
|
|
|
bool writedone=TRUE;
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
if ((k->bytecount == 0) && (k->writebytecount == 0))
|
|
|
|
Curl_pgrsTime(data, TIMER_STARTTRANSFER);
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
didwhat |= KEEP_WRITE;
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/*
|
|
|
|
* We loop here to do the READ and SEND loop until we run out of
|
|
|
|
* data to send or until we get EWOULDBLOCK back
|
|
|
|
*/
|
|
|
|
do {
|
|
|
|
|
|
|
|
/* only read more data if there's no upload data already
|
|
|
|
present in the upload buffer */
|
|
|
|
if(0 == conn->upload_present) {
|
|
|
|
/* init the "upload from here" pointer */
|
|
|
|
conn->upload_fromhere = k->uploadbuf;
|
|
|
|
|
2003-02-24 11:53:53 -05:00
|
|
|
if(!k->upload_done) {
|
|
|
|
/* HTTP pollution, this should be written nicer to become more
|
|
|
|
protocol agnostic. */
|
|
|
|
|
|
|
|
if(k->wait100_after_headers &&
|
|
|
|
(conn->proto.http->sending == HTTPSEND_BODY)) {
|
|
|
|
/* If this call is to send body data, we must take some action:
|
|
|
|
We have sent off the full HTTP 1.1 request, and we shall now
|
|
|
|
go into the Expect: 100 state and await such a header */
|
|
|
|
k->wait100_after_headers = FALSE; /* headers sent */
|
|
|
|
k->write_after_100_header = TRUE; /* wait for the header */
|
|
|
|
FD_ZERO (&k->writefd); /* clear it */
|
|
|
|
k->wkeepfd = k->writefd; /* set the keeper variable */
|
|
|
|
k->keepon &= ~KEEP_WRITE; /* disable writing */
|
|
|
|
k->start100 = Curl_tvnow(); /* timeout count starts now */
|
|
|
|
didwhat &= ~KEEP_WRITE; /* we didn't write anything actually */
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2002-12-10 08:10:00 -05:00
|
|
|
nread = fillbuffer(conn, BUFSIZE);
|
2003-02-24 11:53:53 -05:00
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
else
|
|
|
|
nread = 0; /* we're done uploading/reading */
|
|
|
|
|
|
|
|
/* the signed int typecase of nread of for systems that has
|
|
|
|
unsigned size_t */
|
|
|
|
if (nread<=0) {
|
|
|
|
/* done */
|
|
|
|
k->keepon &= ~KEEP_WRITE; /* we're done writing */
|
|
|
|
FD_ZERO(&k->wkeepfd);
|
|
|
|
writedone = TRUE;
|
|
|
|
break;
|
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/* store number of bytes available for upload */
|
|
|
|
conn->upload_present = nread;
|
2002-01-03 10:01:22 -05:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/* convert LF to CRLF if so asked */
|
|
|
|
if (data->set.crlf) {
|
2003-01-20 07:52:34 -05:00
|
|
|
if(data->state.scratch == NULL)
|
|
|
|
data->state.scratch = malloc(2*BUFSIZE);
|
|
|
|
if(data->state.scratch == NULL) {
|
|
|
|
failf (data, "Failed to alloc scratch buffer!");
|
|
|
|
return CURLE_OUT_OF_MEMORY;
|
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
for(i = 0, si = 0; i < nread; i++, si++) {
|
|
|
|
if (conn->upload_fromhere[i] == 0x0a) {
|
|
|
|
data->state.scratch[si++] = 0x0d;
|
|
|
|
data->state.scratch[si] = 0x0a;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
data->state.scratch[si] = conn->upload_fromhere[i];
|
2002-01-16 09:50:53 -05:00
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
if(si != nread) {
|
|
|
|
/* only perform the special operation if we really did replace
|
|
|
|
anything */
|
|
|
|
nread = si;
|
2002-03-20 05:53:24 -05:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/* upload from the new (replaced) buffer instead */
|
|
|
|
conn->upload_fromhere = data->state.scratch;
|
2002-03-20 05:53:24 -05:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
/* set the new amount too */
|
|
|
|
conn->upload_present = nread;
|
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
}
|
|
|
|
}
|
2002-12-05 09:26:30 -05:00
|
|
|
else {
|
|
|
|
/* We have a partial buffer left from a previous "round". Use
|
|
|
|
that instead of reading more data */
|
|
|
|
}
|
|
|
|
|
2002-12-09 10:37:54 -05:00
|
|
|
/* write to socket (send away data) */
|
2002-12-05 09:26:30 -05:00
|
|
|
result = Curl_write(conn,
|
2002-12-09 10:37:54 -05:00
|
|
|
conn->writesockfd, /* socket to send to */
|
|
|
|
conn->upload_fromhere, /* buffer pointer */
|
|
|
|
conn->upload_present, /* buffer size */
|
|
|
|
&bytes_written); /* actually send away */
|
2002-12-05 09:26:30 -05:00
|
|
|
if(result)
|
|
|
|
return result;
|
2003-01-29 07:52:45 -05:00
|
|
|
|
|
|
|
if(data->set.verbose)
|
|
|
|
/* show the data before we change the pointer upload_fromhere */
|
|
|
|
Curl_debug(data, CURLINFO_DATA_OUT, conn->upload_fromhere,
|
|
|
|
bytes_written);
|
|
|
|
|
|
|
|
if(conn->upload_present != bytes_written) {
|
2002-12-05 09:26:30 -05:00
|
|
|
/* we only wrote a part of the buffer (if anything), deal with it! */
|
|
|
|
|
|
|
|
/* store the amount of bytes left in the buffer to write */
|
|
|
|
conn->upload_present -= bytes_written;
|
|
|
|
|
|
|
|
/* advance the pointer where to find the buffer when the next send
|
|
|
|
is to happen */
|
|
|
|
conn->upload_fromhere += bytes_written;
|
|
|
|
|
|
|
|
writedone = TRUE; /* we are done, stop the loop */
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
/* we've uploaded that buffer now */
|
|
|
|
conn->upload_fromhere = k->uploadbuf;
|
|
|
|
conn->upload_present = 0; /* no more bytes left */
|
|
|
|
|
|
|
|
if(k->upload_done) {
|
|
|
|
/* switch off writing, we're done! */
|
|
|
|
k->keepon &= ~KEEP_WRITE; /* we're done writing */
|
|
|
|
FD_ZERO(&k->wkeepfd);
|
|
|
|
writedone = TRUE;
|
|
|
|
}
|
2002-11-11 03:40:37 -05:00
|
|
|
}
|
2002-01-16 09:50:53 -05:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
k->writebytecount += bytes_written;
|
|
|
|
Curl_pgrsSetUploadCounter(data, (double)k->writebytecount);
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-12-05 09:26:30 -05:00
|
|
|
} while(!writedone); /* loop until we're done writing! */
|
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
2001-11-01 07:18:53 -05:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
} while(0); /* just to break out from! */
|
2000-10-03 07:02:52 -04:00
|
|
|
|
2003-02-24 11:53:53 -05:00
|
|
|
k->now = Curl_tvnow();
|
2002-01-03 10:01:22 -05:00
|
|
|
if(didwhat) {
|
|
|
|
/* Update read/write counters */
|
|
|
|
if(conn->bytecountp)
|
|
|
|
*conn->bytecountp = k->bytecount; /* read count */
|
|
|
|
if(conn->writebytecountp)
|
|
|
|
*conn->writebytecountp = k->writebytecount; /* write count */
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
/* no read no write, this is a timeout? */
|
|
|
|
if (k->write_after_100_header) {
|
|
|
|
/* This should allow some time for the header to arrive, but only a
|
|
|
|
very short time as otherwise it'll be too much wasted times too
|
|
|
|
often. */
|
2003-02-24 11:53:53 -05:00
|
|
|
|
|
|
|
/* Quoting RFC2616, section "8.2.3 Use of the 100 (Continue) Status":
|
|
|
|
|
|
|
|
Therefore, when a client sends this header field to an origin server
|
|
|
|
(possibly via a proxy) from which it has never seen a 100 (Continue)
|
|
|
|
status, the client SHOULD NOT wait for an indefinite period before
|
|
|
|
sending the request body.
|
|
|
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
int ms = Curl_tvdiff(k->now, k->start100);
|
|
|
|
if(ms > CURL_TIMEOUT_EXPECT_100) {
|
|
|
|
/* we've waited long enough, continue anyway */
|
|
|
|
k->write_after_100_header = FALSE;
|
|
|
|
FD_SET (conn->writesockfd, &k->writefd); /* write socket */
|
|
|
|
k->keepon |= KEEP_WRITE;
|
|
|
|
k->wkeepfd = k->writefd;
|
|
|
|
}
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
|
|
|
}
|
2001-11-01 07:18:53 -05:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
if(Curl_pgrsUpdate(conn))
|
|
|
|
result = CURLE_ABORTED_BY_CALLBACK;
|
|
|
|
else
|
|
|
|
result = Curl_speedcheck (data, k->now);
|
|
|
|
if (result)
|
|
|
|
return result;
|
|
|
|
|
|
|
|
if (data->set.timeout &&
|
|
|
|
((Curl_tvdiff(k->now, k->start)/1000) >= data->set.timeout)) {
|
2004-01-05 17:29:29 -05:00
|
|
|
failf (data, "Operation timed out with %Od out of %Od bytes received",
|
2002-01-03 10:01:22 -05:00
|
|
|
k->bytecount, conn->size);
|
|
|
|
return CURLE_OPERATION_TIMEOUTED;
|
|
|
|
}
|
|
|
|
|
|
|
|
if(!k->keepon) {
|
|
|
|
/*
|
|
|
|
* The transfer has been performed. Just make some general checks before
|
|
|
|
* returning.
|
|
|
|
*/
|
|
|
|
|
2003-12-03 02:52:00 -05:00
|
|
|
if(!(data->set.no_body) && (conn->size != -1) &&
|
|
|
|
(k->bytecount != conn->size) &&
|
2002-01-03 10:01:22 -05:00
|
|
|
!conn->newurl) {
|
2004-01-05 17:29:29 -05:00
|
|
|
failf(data, "transfer closed with %Od bytes remaining to read",
|
2003-12-03 02:52:00 -05:00
|
|
|
conn->size - k->bytecount);
|
2002-01-03 10:01:22 -05:00
|
|
|
return CURLE_PARTIAL_FILE;
|
|
|
|
}
|
|
|
|
else if(conn->bits.chunk && conn->proto.http->chunk.datasize) {
|
|
|
|
failf(data, "transfer closed with at least %d bytes remaining",
|
|
|
|
conn->proto.http->chunk.datasize);
|
|
|
|
return CURLE_PARTIAL_FILE;
|
2000-05-22 10:09:31 -04:00
|
|
|
}
|
2002-01-03 10:01:22 -05:00
|
|
|
if(Curl_pgrsUpdate(conn))
|
|
|
|
return CURLE_ABORTED_BY_CALLBACK;
|
2000-05-22 10:09:31 -04:00
|
|
|
}
|
2001-05-12 12:11:14 -04:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
/* Now update the "done" boolean we return */
|
|
|
|
*done = !k->keepon;
|
2001-05-12 12:11:14 -04:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
return CURLE_OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
CURLcode Curl_readwrite_init(struct connectdata *conn)
|
|
|
|
{
|
|
|
|
struct SessionHandle *data = conn->data;
|
|
|
|
struct Curl_transfer_keeper *k = &conn->keep;
|
|
|
|
|
2002-09-02 18:31:18 -04:00
|
|
|
/* NB: the content encoding software depends on this initialization of
|
|
|
|
Curl_transfer_keeper. 08/28/02 jhrg */
|
2002-01-03 10:01:22 -05:00
|
|
|
memset(k, 0, sizeof(struct Curl_transfer_keeper));
|
|
|
|
|
|
|
|
k->start = Curl_tvnow(); /* start time */
|
|
|
|
k->now = k->start; /* current time is now */
|
|
|
|
k->header = TRUE; /* assume header */
|
|
|
|
k->httpversion = -1; /* unknown at this point */
|
|
|
|
|
|
|
|
data = conn->data; /* there's the root struct */
|
|
|
|
k->buf = data->state.buffer;
|
2002-01-16 09:50:53 -05:00
|
|
|
k->uploadbuf = data->state.uploadbuffer;
|
2002-01-03 10:01:22 -05:00
|
|
|
k->maxfd = (conn->sockfd>conn->writesockfd?
|
|
|
|
conn->sockfd:conn->writesockfd)+1;
|
|
|
|
k->hbufp = data->state.headerbuff;
|
2003-06-12 19:03:08 -04:00
|
|
|
k->ignorebody=FALSE;
|
2002-01-03 10:01:22 -05:00
|
|
|
|
|
|
|
Curl_pgrsTime(data, TIMER_PRETRANSFER);
|
|
|
|
Curl_speedinit(data);
|
|
|
|
|
2002-05-02 04:52:09 -04:00
|
|
|
Curl_pgrsSetUploadCounter(data, 0);
|
|
|
|
Curl_pgrsSetDownloadCounter(data, 0);
|
|
|
|
|
2002-11-11 18:03:03 -05:00
|
|
|
if (!conn->bits.getheader) {
|
2002-01-03 10:01:22 -05:00
|
|
|
k->header = FALSE;
|
|
|
|
if(conn->size > 0)
|
2004-01-05 17:29:29 -05:00
|
|
|
Curl_pgrsSetDownloadSize(data, (double)conn->size);
|
2000-05-22 10:09:31 -04:00
|
|
|
}
|
2002-01-03 10:01:22 -05:00
|
|
|
/* we want header and/or body, if neither then don't do this! */
|
2002-11-11 18:03:03 -05:00
|
|
|
if(conn->bits.getheader || !data->set.no_body) {
|
2002-01-03 10:01:22 -05:00
|
|
|
|
2003-01-15 06:44:33 -05:00
|
|
|
FD_ZERO (&k->readfd); /* clear it */
|
2002-01-03 10:01:22 -05:00
|
|
|
if(conn->sockfd != -1) {
|
|
|
|
FD_SET (conn->sockfd, &k->readfd); /* read socket */
|
|
|
|
k->keepon |= KEEP_READ;
|
|
|
|
}
|
|
|
|
|
2003-01-15 06:44:33 -05:00
|
|
|
FD_ZERO (&k->writefd); /* clear it */
|
2002-01-03 10:01:22 -05:00
|
|
|
if(conn->writesockfd != -1) {
|
2003-02-24 11:53:53 -05:00
|
|
|
/* HTTP 1.1 magic:
|
|
|
|
|
|
|
|
Even if we require a 100-return code before uploading data, we might
|
|
|
|
need to write data before that since the REQUEST may not have been
|
|
|
|
finished sent off just yet.
|
|
|
|
|
|
|
|
Thus, we must check if the request has been sent before we set the
|
|
|
|
state info where we wait for the 100-return code
|
|
|
|
*/
|
|
|
|
if (data->set.expect100header &&
|
|
|
|
(conn->proto.http->sending == HTTPSEND_BODY)) {
|
2002-01-03 10:01:22 -05:00
|
|
|
/* wait with write until we either got 100-continue or a timeout */
|
|
|
|
k->write_after_100_header = TRUE;
|
2003-02-24 11:53:53 -05:00
|
|
|
k->start100 = k->start;
|
|
|
|
}
|
2002-01-03 10:01:22 -05:00
|
|
|
else {
|
2003-02-24 11:53:53 -05:00
|
|
|
if(data->set.expect100header)
|
|
|
|
/* when we've sent off the rest of the headers, we must await a
|
|
|
|
100-continue */
|
|
|
|
k->wait100_after_headers = TRUE;
|
2002-01-03 10:01:22 -05:00
|
|
|
FD_SET (conn->writesockfd, &k->writefd); /* write socket */
|
|
|
|
k->keepon |= KEEP_WRITE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* get these in backup variables to be able to restore them on each lap in
|
|
|
|
the select() loop */
|
|
|
|
k->rkeepfd = k->readfd;
|
|
|
|
k->wkeepfd = k->writefd;
|
|
|
|
|
2001-03-16 10:45:12 -05:00
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
|
|
|
return CURLE_OK;
|
|
|
|
}
|
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
void Curl_single_fdset(struct connectdata *conn,
|
|
|
|
fd_set *read_fd_set,
|
|
|
|
fd_set *write_fd_set,
|
|
|
|
fd_set *exc_fd_set,
|
|
|
|
int *max_fd)
|
2000-05-22 10:09:31 -04:00
|
|
|
{
|
2002-01-03 10:01:22 -05:00
|
|
|
*max_fd = -1; /* init */
|
|
|
|
if(conn->keep.keepon & KEEP_READ) {
|
|
|
|
FD_SET(conn->sockfd, read_fd_set);
|
|
|
|
*max_fd = conn->sockfd;
|
2002-05-02 18:14:31 -04:00
|
|
|
conn->keep.readfdp = read_fd_set; /* store the address of the set */
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
|
|
|
if(conn->keep.keepon & KEEP_WRITE) {
|
|
|
|
FD_SET(conn->writesockfd, write_fd_set);
|
|
|
|
if(conn->writesockfd > *max_fd)
|
|
|
|
*max_fd = conn->writesockfd;
|
2002-05-02 18:14:31 -04:00
|
|
|
conn->keep.writefdp = write_fd_set; /* store the address of the set */
|
2002-01-03 10:01:22 -05:00
|
|
|
}
|
|
|
|
/* we don't use exceptions, only touch that one to prevent compiler
|
|
|
|
warnings! */
|
|
|
|
*exc_fd_set = *exc_fd_set;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Transfer()
|
|
|
|
*
|
|
|
|
* This function is what performs the actual transfer. It is capable of
|
|
|
|
* doing both ways simultaneously.
|
|
|
|
* The transfer must already have been setup by a call to Curl_Transfer().
|
|
|
|
*
|
|
|
|
* Note that headers are created in a preallocated buffer of a default size.
|
|
|
|
* That buffer can be enlarged on demand, but it is never shrinken again.
|
|
|
|
*
|
|
|
|
* Parts of this function was once written by the friendly Mark Butler
|
|
|
|
* <butlerm@xmission.com>.
|
|
|
|
*/
|
|
|
|
|
|
|
|
static CURLcode
|
|
|
|
Transfer(struct connectdata *conn)
|
|
|
|
{
|
|
|
|
struct SessionHandle *data = conn->data;
|
|
|
|
CURLcode result;
|
|
|
|
struct Curl_transfer_keeper *k = &conn->keep;
|
|
|
|
bool done=FALSE;
|
|
|
|
|
|
|
|
Curl_readwrite_init(conn);
|
|
|
|
|
|
|
|
if((conn->sockfd == -1) && (conn->writesockfd == -1))
|
|
|
|
/* nothing to read, nothing to write, we're already OK! */
|
|
|
|
return CURLE_OK;
|
|
|
|
|
|
|
|
/* we want header and/or body, if neither then don't do this! */
|
2002-11-11 18:03:03 -05:00
|
|
|
if(!conn->bits.getheader && data->set.no_body)
|
2002-01-03 10:01:22 -05:00
|
|
|
return CURLE_OK;
|
|
|
|
|
2002-05-28 10:18:36 -04:00
|
|
|
k->writefdp = &k->writefd; /* store the address of the set */
|
|
|
|
k->readfdp = &k->readfd; /* store the address of the set */
|
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
while (!done) {
|
|
|
|
struct timeval interval;
|
|
|
|
k->readfd = k->rkeepfd; /* set these every lap in the loop */
|
|
|
|
k->writefd = k->wkeepfd;
|
|
|
|
interval.tv_sec = 1;
|
|
|
|
interval.tv_usec = 0;
|
2002-01-29 15:28:59 -05:00
|
|
|
|
2002-05-02 18:14:31 -04:00
|
|
|
switch (select (k->maxfd, k->readfdp, k->writefdp, NULL, &interval)) {
|
2002-01-03 10:01:22 -05:00
|
|
|
case -1: /* select() error, stop reading */
|
|
|
|
#ifdef EINTR
|
|
|
|
/* The EINTR is not serious, and it seems you might get this more
|
|
|
|
ofen when using the lib in a multi-threaded environment! */
|
|
|
|
if(errno == EINTR)
|
|
|
|
;
|
|
|
|
else
|
2001-11-07 09:13:29 -05:00
|
|
|
#endif
|
2002-01-03 10:01:22 -05:00
|
|
|
done = TRUE; /* no more read or write */
|
|
|
|
continue;
|
|
|
|
case 0: /* timeout */
|
|
|
|
default: /* readable descriptors */
|
|
|
|
result = Curl_readwrite(conn, &done);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if(result)
|
|
|
|
return result;
|
|
|
|
|
|
|
|
/* "done" signals to us if the transfer(s) are ready */
|
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
return CURLE_OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
CURLcode Curl_pretransfer(struct SessionHandle *data)
|
|
|
|
{
|
2003-08-14 11:06:36 -04:00
|
|
|
CURLcode res;
|
2001-08-30 18:48:34 -04:00
|
|
|
if(!data->change.url)
|
2001-04-11 10:14:28 -04:00
|
|
|
/* we can't do anything wihout URL */
|
|
|
|
return CURLE_URL_MALFORMAT;
|
|
|
|
|
2001-08-28 04:37:54 -04:00
|
|
|
#ifdef USE_SSLEAY
|
|
|
|
/* Init the SSL session ID cache here. We do it here since we want to
|
|
|
|
do it after the *_setopt() calls (that could change the size) but
|
|
|
|
before any transfer. */
|
2003-08-14 11:06:36 -04:00
|
|
|
res = Curl_SSL_InitSessions(data, data->set.ssl.numsessions);
|
|
|
|
if(res)
|
|
|
|
return res;
|
|
|
|
#else
|
|
|
|
(void)res;
|
2001-08-28 04:37:54 -04:00
|
|
|
#endif
|
|
|
|
|
2001-08-30 18:48:34 -04:00
|
|
|
data->set.followlocation=0; /* reset the location-follow counter */
|
|
|
|
data->state.this_is_a_follow = FALSE; /* reset this */
|
2001-11-02 17:30:34 -05:00
|
|
|
data->state.errorbuf = FALSE; /* no error has occurred */
|
2001-03-22 14:14:35 -05:00
|
|
|
|
2003-06-26 07:30:26 -04:00
|
|
|
/* set preferred authentication, default to basic */
|
2003-08-11 07:47:45 -04:00
|
|
|
|
|
|
|
data->state.authstage = 0; /* initialize authentication later */
|
2003-06-26 07:30:26 -04:00
|
|
|
|
2002-10-17 03:10:39 -04:00
|
|
|
/* If there was a list of cookie files to read and we haven't done it before,
|
|
|
|
do it now! */
|
|
|
|
if(data->change.cookielist) {
|
|
|
|
struct curl_slist *list = data->change.cookielist;
|
2003-08-04 11:02:42 -04:00
|
|
|
Curl_share_lock(data, CURL_LOCK_DATA_COOKIE, CURL_LOCK_ACCESS_SINGLE);
|
2002-10-17 03:10:39 -04:00
|
|
|
while(list) {
|
2003-08-11 05:56:06 -04:00
|
|
|
data->cookies = Curl_cookie_init(data,
|
|
|
|
list->data,
|
2002-10-17 03:10:39 -04:00
|
|
|
data->cookies,
|
|
|
|
data->set.cookiesession);
|
|
|
|
list = list->next;
|
|
|
|
}
|
2003-08-04 11:02:42 -04:00
|
|
|
Curl_share_unlock(data, CURL_LOCK_DATA_COOKIE);
|
2002-10-17 03:10:39 -04:00
|
|
|
curl_slist_free_all(data->change.cookielist); /* clean up list */
|
|
|
|
data->change.cookielist = NULL; /* don't do this again! */
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
/* Allow data->set.use_port to set which port to use. This needs to be
|
|
|
|
* disabled for example when we follow Location: headers to URLs using
|
|
|
|
* different ports! */
|
|
|
|
data->state.allow_port = TRUE;
|
|
|
|
|
2003-12-02 05:12:44 -05:00
|
|
|
#if defined(HAVE_SIGNAL) && defined(SIGPIPE) && !defined(HAVE_MSG_NOSIGNAL)
|
2001-11-07 09:13:29 -05:00
|
|
|
/*************************************************************
|
|
|
|
* Tell signal handler to ignore SIGPIPE
|
|
|
|
*************************************************************/
|
2002-08-08 18:52:50 -04:00
|
|
|
if(!data->set.no_signal)
|
|
|
|
data->state.prev_signal = signal(SIGPIPE, SIG_IGN);
|
2001-11-07 09:13:29 -05:00
|
|
|
#endif
|
2001-08-15 02:53:10 -04:00
|
|
|
|
2001-11-07 09:13:29 -05:00
|
|
|
Curl_initinfo(data); /* reset session-specific information "variables" */
|
2001-01-05 05:11:41 -05:00
|
|
|
Curl_pgrsStartNow(data);
|
2000-06-16 09:14:27 -04:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
return CURLE_OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
CURLcode Curl_posttransfer(struct SessionHandle *data)
|
|
|
|
{
|
2003-12-02 05:12:44 -05:00
|
|
|
#if defined(HAVE_SIGNAL) && defined(SIGPIPE) && !defined(HAVE_MSG_NOSIGNAL)
|
2002-01-03 10:01:22 -05:00
|
|
|
/* restore the signal handler for SIGPIPE before we get back */
|
2002-08-08 18:52:50 -04:00
|
|
|
if(!data->set.no_signal)
|
|
|
|
signal(SIGPIPE, data->state.prev_signal);
|
2003-12-02 08:40:12 -05:00
|
|
|
#else
|
|
|
|
(void)data; /* unused parameter */
|
2002-01-03 10:01:22 -05:00
|
|
|
#endif
|
|
|
|
|
|
|
|
return CURLE_OK;
|
|
|
|
}
|
|
|
|
|
2003-06-26 07:30:26 -04:00
|
|
|
static int strlen_url(char *url)
|
|
|
|
{
|
|
|
|
char *ptr;
|
|
|
|
int newlen=0;
|
|
|
|
bool left=TRUE; /* left side of the ? */
|
|
|
|
|
|
|
|
for(ptr=url; *ptr; ptr++) {
|
|
|
|
switch(*ptr) {
|
|
|
|
case '?':
|
|
|
|
left=FALSE;
|
|
|
|
default:
|
|
|
|
newlen++;
|
|
|
|
break;
|
|
|
|
case ' ':
|
|
|
|
if(left)
|
|
|
|
newlen+=3;
|
|
|
|
else
|
|
|
|
newlen++;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return newlen;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void strcpy_url(char *output, char *url)
|
|
|
|
{
|
|
|
|
/* we must add this with whitespace-replacing */
|
|
|
|
bool left=TRUE;
|
|
|
|
char *iptr;
|
|
|
|
char *optr = output;
|
|
|
|
for(iptr = url; /* read from here */
|
|
|
|
*iptr; /* until zero byte */
|
|
|
|
iptr++) {
|
|
|
|
switch(*iptr) {
|
|
|
|
case '?':
|
|
|
|
left=FALSE;
|
|
|
|
default:
|
|
|
|
*optr++=*iptr;
|
|
|
|
break;
|
|
|
|
case ' ':
|
|
|
|
if(left) {
|
|
|
|
*optr++='%'; /* add a '%' */
|
|
|
|
*optr++='2'; /* add a '2' */
|
|
|
|
*optr++='0'; /* add a '0' */
|
|
|
|
}
|
|
|
|
else
|
|
|
|
*optr++='+'; /* add a '+' here */
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
*optr=0; /* zero terminate output buffer */
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2002-10-07 09:38:34 -04:00
|
|
|
CURLcode Curl_follow(struct SessionHandle *data,
|
|
|
|
char *newurl) /* this 'newurl' is the Location: string,
|
|
|
|
and it must be malloc()ed before passed
|
|
|
|
here */
|
|
|
|
{
|
|
|
|
/* Location: redirect */
|
|
|
|
char prot[16]; /* URL protocol string storage */
|
|
|
|
char letter; /* used for a silly sscanf */
|
2003-06-26 07:30:26 -04:00
|
|
|
int newlen;
|
|
|
|
char *newest;
|
2002-10-07 09:38:34 -04:00
|
|
|
|
|
|
|
if (data->set.maxredirs &&
|
|
|
|
(data->set.followlocation >= data->set.maxredirs)) {
|
|
|
|
failf(data,"Maximum (%d) redirects followed", data->set.maxredirs);
|
|
|
|
return CURLE_TOO_MANY_REDIRECTS;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* mark the next request as a followed location: */
|
|
|
|
data->state.this_is_a_follow = TRUE;
|
|
|
|
|
|
|
|
data->set.followlocation++; /* count location-followers */
|
|
|
|
|
|
|
|
if(data->set.http_auto_referer) {
|
|
|
|
/* We are asked to automatically set the previous URL as the
|
|
|
|
referer when we get the next URL. We pick the ->url field,
|
|
|
|
which may or may not be 100% correct */
|
|
|
|
|
|
|
|
if(data->change.referer_alloc)
|
|
|
|
/* If we already have an allocated referer, free this first */
|
|
|
|
free(data->change.referer);
|
|
|
|
|
|
|
|
data->change.referer = strdup(data->change.url);
|
|
|
|
data->change.referer_alloc = TRUE; /* yes, free this later */
|
|
|
|
}
|
|
|
|
|
|
|
|
if(2 != sscanf(newurl, "%15[^?&/:]://%c", prot, &letter)) {
|
|
|
|
/***
|
|
|
|
*DANG* this is an RFC 2068 violation. The URL is supposed
|
|
|
|
to be absolute and this doesn't seem to be that!
|
|
|
|
***
|
|
|
|
Instead, we have to TRY to append this new path to the old URL
|
|
|
|
to the right of the host part. Oh crap, this is doomed to cause
|
|
|
|
problems in the future...
|
|
|
|
*/
|
|
|
|
char *protsep;
|
|
|
|
char *pathsep;
|
|
|
|
|
|
|
|
char *useurl = newurl;
|
2003-06-26 07:30:26 -04:00
|
|
|
int urllen;
|
2002-10-07 09:38:34 -04:00
|
|
|
|
|
|
|
/* we must make our own copy of the URL to play with, as it may
|
|
|
|
point to read-only data */
|
|
|
|
char *url_clone=strdup(data->change.url);
|
|
|
|
|
|
|
|
if(!url_clone)
|
|
|
|
return CURLE_OUT_OF_MEMORY; /* skip out of this NOW */
|
|
|
|
|
|
|
|
/* protsep points to the start of the host name */
|
|
|
|
protsep=strstr(url_clone, "//");
|
|
|
|
if(!protsep)
|
|
|
|
protsep=url_clone;
|
|
|
|
else
|
|
|
|
protsep+=2; /* pass the slashes */
|
|
|
|
|
|
|
|
if('/' != newurl[0]) {
|
|
|
|
int level=0;
|
|
|
|
|
|
|
|
/* First we need to find out if there's a ?-letter in the URL,
|
|
|
|
and cut it and the right-side of that off */
|
|
|
|
pathsep = strrchr(protsep, '?');
|
|
|
|
if(pathsep)
|
|
|
|
*pathsep=0;
|
|
|
|
|
|
|
|
/* we have a relative path to append to the last slash if
|
|
|
|
there's one available */
|
|
|
|
pathsep = strrchr(protsep, '/');
|
|
|
|
if(pathsep)
|
|
|
|
*pathsep=0;
|
|
|
|
|
|
|
|
/* Check if there's any slash after the host name, and if so,
|
|
|
|
remember that position instead */
|
|
|
|
pathsep = strchr(protsep, '/');
|
|
|
|
if(pathsep)
|
|
|
|
protsep = pathsep+1;
|
|
|
|
else
|
|
|
|
protsep = NULL;
|
|
|
|
|
|
|
|
/* now deal with one "./" or any amount of "../" in the newurl
|
|
|
|
and act accordingly */
|
|
|
|
|
|
|
|
if((useurl[0] == '.') && (useurl[1] == '/'))
|
|
|
|
useurl+=2; /* just skip the "./" */
|
|
|
|
|
|
|
|
while((useurl[0] == '.') &&
|
|
|
|
(useurl[1] == '.') &&
|
|
|
|
(useurl[2] == '/')) {
|
|
|
|
level++;
|
|
|
|
useurl+=3; /* pass the "../" */
|
|
|
|
}
|
|
|
|
|
|
|
|
if(protsep) {
|
|
|
|
while(level--) {
|
|
|
|
/* cut off one more level from the right of the original URL */
|
|
|
|
pathsep = strrchr(protsep, '/');
|
|
|
|
if(pathsep)
|
|
|
|
*pathsep=0;
|
|
|
|
else {
|
|
|
|
*protsep=0;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
/* We got a new absolute path for this server, cut off from the
|
|
|
|
first slash */
|
|
|
|
pathsep = strchr(protsep, '/');
|
|
|
|
if(pathsep)
|
|
|
|
*pathsep=0;
|
2003-12-18 04:19:10 -05:00
|
|
|
else {
|
|
|
|
/* There was no slash. Now, since we might be operating on a badly
|
|
|
|
formatted URL, such as "http://www.url.com?id=2380" which doesn't
|
|
|
|
use a slash separator as it is supposed to, we need to check for a
|
|
|
|
?-letter as well! */
|
|
|
|
pathsep = strchr(protsep, '?');
|
|
|
|
if(pathsep)
|
|
|
|
*pathsep=0;
|
|
|
|
}
|
2002-10-07 09:38:34 -04:00
|
|
|
}
|
|
|
|
|
2003-06-26 07:30:26 -04:00
|
|
|
/* If the new part contains a space, this is a mighty stupid redirect
|
|
|
|
but we still make an effort to do "right". To the left of a '?'
|
|
|
|
letter we replace each space with %20 while it is replaced with '+'
|
|
|
|
on the right side of the '?' letter.
|
|
|
|
*/
|
|
|
|
newlen = strlen_url(useurl);
|
|
|
|
|
|
|
|
urllen = strlen(url_clone);
|
|
|
|
|
|
|
|
newest=(char *)malloc( urllen + 1 + /* possible slash */
|
|
|
|
newlen + 1 /* zero byte */);
|
2002-10-07 09:38:34 -04:00
|
|
|
|
|
|
|
if(!newest)
|
|
|
|
return CURLE_OUT_OF_MEMORY; /* go out from this */
|
|
|
|
|
2003-06-26 07:30:26 -04:00
|
|
|
/* copy over the root url part */
|
|
|
|
memcpy(newest, url_clone, urllen);
|
|
|
|
|
|
|
|
/* check if we need to append a slash */
|
|
|
|
if(('/' == useurl[0]) || (protsep && !*protsep))
|
|
|
|
;
|
|
|
|
else
|
|
|
|
newest[urllen++]='/';
|
|
|
|
|
|
|
|
/* then append the new piece on the right side */
|
|
|
|
strcpy_url(&newest[urllen], useurl);
|
|
|
|
|
2002-10-07 09:38:34 -04:00
|
|
|
free(newurl); /* newurl is the allocated pointer */
|
|
|
|
free(url_clone);
|
|
|
|
newurl = newest;
|
|
|
|
}
|
2003-06-26 07:30:26 -04:00
|
|
|
else {
|
2002-10-07 09:38:34 -04:00
|
|
|
/* This is an absolute URL, don't allow the custom port number */
|
|
|
|
data->state.allow_port = FALSE;
|
|
|
|
|
2003-06-26 07:30:26 -04:00
|
|
|
if(strchr(newurl, ' ')) {
|
|
|
|
/* This new URL contains at least one space, this is a mighty stupid
|
|
|
|
redirect but we still make an effort to do "right". */
|
|
|
|
newlen = strlen_url(newurl);
|
|
|
|
|
|
|
|
newest = malloc(newlen+1); /* get memory for this */
|
|
|
|
if(newest) {
|
|
|
|
strcpy_url(newest, newurl); /* create a space-free URL */
|
|
|
|
|
|
|
|
free(newurl); /* that was no good */
|
|
|
|
newurl = newest; /* use this instead now */
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
}
|
|
|
|
|
2002-10-07 09:38:34 -04:00
|
|
|
if(data->change.url_alloc)
|
|
|
|
free(data->change.url);
|
|
|
|
else
|
|
|
|
data->change.url_alloc = TRUE; /* the URL is allocated */
|
|
|
|
|
|
|
|
data->change.url = newurl;
|
|
|
|
newurl = NULL; /* don't free! */
|
|
|
|
|
2003-05-22 18:39:38 -04:00
|
|
|
infof(data, "Issue another request to this URL: '%s'\n", data->change.url);
|
2002-10-07 09:38:34 -04:00
|
|
|
|
|
|
|
/*
|
2003-06-12 13:40:56 -04:00
|
|
|
* We get here when the HTTP code is 300-399 (and 401). We need to perform
|
2002-10-07 09:38:34 -04:00
|
|
|
* differently based on exactly what return code there was.
|
2003-05-22 12:09:54 -04:00
|
|
|
*
|
2003-06-12 13:40:56 -04:00
|
|
|
* News from 7.10.6: we can also get here on a 401, in case we act on a
|
|
|
|
* HTTP authentication scheme other than Basic.
|
2002-10-07 09:38:34 -04:00
|
|
|
*/
|
|
|
|
switch(data->info.httpcode) {
|
2003-05-22 12:09:54 -04:00
|
|
|
case 401:
|
2003-06-12 13:40:56 -04:00
|
|
|
/* Act on an authentication, we keep on moving and do the Authorization:
|
|
|
|
XXXX header in the HTTP request code snippet */
|
2003-05-22 12:09:54 -04:00
|
|
|
break;
|
2002-10-07 09:38:34 -04:00
|
|
|
case 300: /* Multiple Choices */
|
|
|
|
case 306: /* Not used */
|
|
|
|
case 307: /* Temporary Redirect */
|
|
|
|
default: /* for all unknown ones */
|
|
|
|
/* These are explicitly mention since I've checked RFC2616 and they
|
|
|
|
* seem to be OK to POST to.
|
|
|
|
*/
|
|
|
|
break;
|
|
|
|
case 301: /* Moved Permanently */
|
|
|
|
/* (quote from RFC2616, section 10.3.2):
|
|
|
|
*
|
|
|
|
* Note: When automatically redirecting a POST request after
|
|
|
|
* receiving a 301 status code, some existing HTTP/1.0 user agents
|
|
|
|
* will erroneously change it into a GET request.
|
|
|
|
*
|
|
|
|
* ----
|
|
|
|
* Warning: Because most of importants user agents do this clear
|
|
|
|
* RFC2616 violation, many webservers expect this misbehavior. So
|
|
|
|
* these servers often answers to a POST request with an error page.
|
|
|
|
* To be sure that libcurl gets the page that most user agents
|
|
|
|
* would get, libcurl has to force GET:
|
|
|
|
*/
|
|
|
|
if( data->set.httpreq == HTTPREQ_POST
|
|
|
|
|| data->set.httpreq == HTTPREQ_POST_FORM) {
|
|
|
|
infof(data,
|
|
|
|
"Violate RFC 2616/10.3.2 and switch from POST to GET\n");
|
|
|
|
data->set.httpreq = HTTPREQ_GET;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case 302: /* Found */
|
|
|
|
/* (From 10.3.3)
|
|
|
|
|
|
|
|
Note: RFC 1945 and RFC 2068 specify that the client is not allowed
|
|
|
|
to change the method on the redirected request. However, most
|
|
|
|
existing user agent implementations treat 302 as if it were a 303
|
|
|
|
response, performing a GET on the Location field-value regardless
|
|
|
|
of the original request method. The status codes 303 and 307 have
|
|
|
|
been added for servers that wish to make unambiguously clear which
|
|
|
|
kind of reaction is expected of the client.
|
|
|
|
|
|
|
|
(From 10.3.4)
|
|
|
|
|
|
|
|
Note: Many pre-HTTP/1.1 user agents do not understand the 303
|
|
|
|
status. When interoperability with such clients is a concern, the
|
|
|
|
302 status code may be used instead, since most user agents react
|
|
|
|
to a 302 response as described here for 303.
|
|
|
|
*/
|
|
|
|
case 303: /* See Other */
|
|
|
|
/* Disable both types of POSTs, since doing a second POST when
|
|
|
|
* following isn't what anyone would want! */
|
|
|
|
if(data->set.httpreq != HTTPREQ_GET) {
|
|
|
|
data->set.httpreq = HTTPREQ_GET; /* enforce GET request */
|
|
|
|
infof(data, "Disables POST, goes with %s\n",
|
|
|
|
data->set.no_body?"HEAD":"GET");
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case 304: /* Not Modified */
|
|
|
|
/* 304 means we did a conditional request and it was "Not modified".
|
|
|
|
* We shouldn't get any Location: header in this response!
|
|
|
|
*/
|
|
|
|
break;
|
|
|
|
case 305: /* Use Proxy */
|
|
|
|
/* (quote from RFC2616, section 10.3.6):
|
|
|
|
* "The requested resource MUST be accessed through the proxy given
|
|
|
|
* by the Location field. The Location field gives the URI of the
|
|
|
|
* proxy. The recipient is expected to repeat this single request
|
|
|
|
* via the proxy. 305 responses MUST only be generated by origin
|
|
|
|
* servers."
|
|
|
|
*/
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
Curl_pgrsTime(data, TIMER_REDIRECT);
|
|
|
|
Curl_pgrsResetTimes(data);
|
|
|
|
|
|
|
|
return CURLE_OK;
|
|
|
|
}
|
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
CURLcode Curl_perform(struct SessionHandle *data)
|
|
|
|
{
|
|
|
|
CURLcode res;
|
|
|
|
CURLcode res2;
|
|
|
|
struct connectdata *conn=NULL;
|
|
|
|
char *newurl = NULL; /* possibly a new URL to follow to! */
|
|
|
|
|
2002-08-05 13:04:39 -04:00
|
|
|
data->state.used_interface = Curl_if_easy;
|
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
res = Curl_pretransfer(data);
|
|
|
|
if(res)
|
|
|
|
return res;
|
|
|
|
|
2001-11-07 09:13:29 -05:00
|
|
|
/*
|
2002-09-23 08:55:36 -04:00
|
|
|
* It is important that there is NO 'return' from this function at any other
|
|
|
|
* place than falling down to the end of the function! This is because we
|
|
|
|
* have cleanup stuff that must be done before we get back, and that is only
|
|
|
|
* performed after this do-while loop.
|
2001-11-07 09:13:29 -05:00
|
|
|
*/
|
|
|
|
|
2003-07-04 12:29:23 -04:00
|
|
|
do {
|
|
|
|
int urlchanged = FALSE;
|
|
|
|
do {
|
2003-08-05 10:40:59 -04:00
|
|
|
bool async;
|
2003-07-04 12:29:23 -04:00
|
|
|
Curl_pgrsTime(data, TIMER_STARTSINGLE);
|
|
|
|
data->change.url_changed = FALSE;
|
2003-08-05 10:40:59 -04:00
|
|
|
res = Curl_connect(data, &conn, &async);
|
|
|
|
|
|
|
|
if((CURLE_OK == res) && async) {
|
|
|
|
/* Now, if async is TRUE here, we need to wait for the name
|
|
|
|
to resolve */
|
|
|
|
res = Curl_wait_for_resolv(conn, NULL);
|
|
|
|
if(CURLE_OK == res)
|
|
|
|
/* Resolved, continue with the connection */
|
|
|
|
res = Curl_async_resolved(conn);
|
|
|
|
}
|
|
|
|
if(res)
|
|
|
|
break;
|
|
|
|
|
2003-07-04 12:29:23 -04:00
|
|
|
/* If a callback (or something) has altered the URL we should use within
|
|
|
|
the Curl_connect(), we detect it here and act as if we are redirected
|
|
|
|
to the new URL */
|
|
|
|
urlchanged = data->change.url_changed;
|
|
|
|
if ((CURLE_OK == res) && urlchanged) {
|
|
|
|
res = Curl_done(conn);
|
|
|
|
if(CURLE_OK == res) {
|
2003-11-06 02:55:45 -05:00
|
|
|
char *gotourl = strdup(data->change.url);
|
2003-10-18 16:28:53 -04:00
|
|
|
res = Curl_follow(data, gotourl);
|
2003-07-04 12:29:23 -04:00
|
|
|
if(res)
|
2003-10-18 16:28:53 -04:00
|
|
|
free(gotourl);
|
2003-07-04 12:29:23 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
} while (urlchanged && res == CURLE_OK) ;
|
|
|
|
|
2000-05-22 10:09:31 -04:00
|
|
|
if(res == CURLE_OK) {
|
2002-01-03 10:01:22 -05:00
|
|
|
res = Curl_do(&conn);
|
2001-10-31 10:13:19 -05:00
|
|
|
|
2000-05-22 10:09:31 -04:00
|
|
|
if(res == CURLE_OK) {
|
2001-03-09 10:16:28 -05:00
|
|
|
res = Transfer(conn); /* now fetch that URL please */
|
2001-10-17 08:24:51 -04:00
|
|
|
if(res == CURLE_OK)
|
2001-04-03 09:37:53 -04:00
|
|
|
/*
|
|
|
|
* We must duplicate the new URL here as the connection data
|
|
|
|
* may be free()ed in the Curl_done() function.
|
|
|
|
*/
|
|
|
|
newurl = conn->newurl?strdup(conn->newurl):NULL;
|
2002-06-12 18:04:25 -04:00
|
|
|
else {
|
2001-10-22 02:34:14 -04:00
|
|
|
/* The transfer phase returned error, we mark the connection to get
|
|
|
|
* closed to prevent being re-used. This is becasue we can't
|
|
|
|
* possibly know if the connection is in a good shape or not now. */
|
|
|
|
conn->bits.close = TRUE;
|
2001-04-03 09:37:53 -04:00
|
|
|
|
2003-11-24 02:15:37 -05:00
|
|
|
if(-1 != conn->sock[SECONDARYSOCKET]) {
|
2002-06-12 18:04:25 -04:00
|
|
|
/* if we failed anywhere, we must clean up the secondary socket if
|
|
|
|
it was used */
|
2003-11-24 02:15:37 -05:00
|
|
|
sclose(conn->sock[SECONDARYSOCKET]);
|
|
|
|
conn->sock[SECONDARYSOCKET]=-1;
|
2002-06-12 18:04:25 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2001-10-17 08:24:51 -04:00
|
|
|
/* Always run Curl_done(), even if some of the previous calls
|
|
|
|
failed, but return the previous (original) error code */
|
|
|
|
res2 = Curl_done(conn);
|
|
|
|
|
|
|
|
if(CURLE_OK == res)
|
|
|
|
res = res2;
|
2000-05-22 10:09:31 -04:00
|
|
|
}
|
|
|
|
|
2001-04-03 09:37:53 -04:00
|
|
|
/*
|
|
|
|
* Important: 'conn' cannot be used here, since it may have been closed
|
|
|
|
* in 'Curl_done' or other functions.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if((res == CURLE_OK) && newurl) {
|
2002-10-07 09:38:34 -04:00
|
|
|
res = Curl_follow(data, newurl);
|
|
|
|
if(CURLE_OK == res) {
|
|
|
|
newurl = NULL;
|
|
|
|
continue;
|
2001-01-27 15:31:51 -05:00
|
|
|
}
|
2000-05-22 10:09:31 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
break; /* it only reaches here when this shouldn't loop */
|
|
|
|
|
|
|
|
} while(1); /* loop if Location: */
|
|
|
|
|
2001-04-03 09:37:53 -04:00
|
|
|
if(newurl)
|
|
|
|
free(newurl);
|
2000-05-22 10:09:31 -04:00
|
|
|
|
2002-01-03 10:01:22 -05:00
|
|
|
/* run post-transfer uncondionally, but don't clobber the return code if
|
|
|
|
we already have an error code recorder */
|
|
|
|
res2 = Curl_posttransfer(data);
|
|
|
|
if(!res && res2)
|
|
|
|
res = res2;
|
2001-11-07 09:13:29 -05:00
|
|
|
|
2000-05-22 10:09:31 -04:00
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
2001-01-17 08:19:01 -05:00
|
|
|
CURLcode
|
|
|
|
Curl_Transfer(struct connectdata *c_conn, /* connection data */
|
2003-11-24 02:15:37 -05:00
|
|
|
int sockindex, /* socket index to read from or -1 */
|
2004-01-05 17:29:29 -05:00
|
|
|
off_t size, /* -1 if unknown at this point */
|
2003-01-15 06:44:33 -05:00
|
|
|
bool getheader, /* TRUE if header parsing is wanted */
|
2004-01-05 17:29:29 -05:00
|
|
|
off_t *bytecountp, /* return number of bytes read or NULL */
|
2003-11-24 02:15:37 -05:00
|
|
|
int writesockindex, /* socket index to write to, it may very
|
|
|
|
well be the same we read from. -1
|
|
|
|
disables */
|
2004-01-05 17:29:29 -05:00
|
|
|
off_t *writebytecountp /* return number of bytes written or
|
2001-01-17 08:19:01 -05:00
|
|
|
NULL */
|
|
|
|
)
|
|
|
|
{
|
|
|
|
struct connectdata *conn = (struct connectdata *)c_conn;
|
|
|
|
if(!conn)
|
|
|
|
return CURLE_BAD_FUNCTION_ARGUMENT;
|
|
|
|
|
2003-12-10 10:27:27 -05:00
|
|
|
curlassert(sockindex <= 1);
|
|
|
|
|
2001-01-17 08:19:01 -05:00
|
|
|
/* now copy all input parameters */
|
2003-11-24 02:15:37 -05:00
|
|
|
conn->sockfd = sockindex==-1?-1:conn->sock[sockindex];
|
2001-01-17 08:19:01 -05:00
|
|
|
conn->size = size;
|
2002-11-11 18:03:03 -05:00
|
|
|
conn->bits.getheader = getheader;
|
2001-01-17 08:19:01 -05:00
|
|
|
conn->bytecountp = bytecountp;
|
2003-11-24 02:15:37 -05:00
|
|
|
conn->writesockfd = writesockindex==-1?-1:conn->sock[writesockindex];
|
2001-01-17 08:19:01 -05:00
|
|
|
conn->writebytecountp = writebytecountp;
|
|
|
|
|
|
|
|
return CURLE_OK;
|
|
|
|
|
|
|
|
}
|