curl/lib/file.c

545 строки
15 KiB
C
Исходник Обычный вид История

2002-09-03 15:52:59 +04:00
/***************************************************************************
* _ _ ____ _
* Project ___| | | | _ \| |
* / __| | | | |_) | |
* | (__| |_| | _ <| |___
1999-12-29 17:20:26 +03:00
* \___|\___/|_| \_\_____|
*
* Copyright (C) 1998 - 2019, Daniel Stenberg, <daniel@haxx.se>, et al.
1999-12-29 17:20:26 +03:00
*
2002-09-03 15:52:59 +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 https://curl.haxx.se/docs/copyright.html.
*
2001-01-03 12:29:33 +03:00
* 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 15:52:59 +04:00
* furnished to do so, under the terms of the COPYING file.
1999-12-29 17:20:26 +03:00
*
2001-01-03 12:29:33 +03:00
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
1999-12-29 17:20:26 +03:00
*
2002-09-03 15:52:59 +04:00
***************************************************************************/
1999-12-29 17:20:26 +03:00
build: fix circular header inclusion with other packages This commit renames lib/setup.h to lib/curl_setup.h and renames lib/setup_once.h to lib/curl_setup_once.h. Removes the need and usage of a header inclusion guard foreign to libcurl. [1] Removes the need and presence of an alarming notice we carried in old setup_once.h [2] ---------------------------------------- 1 - lib/setup_once.h used __SETUP_ONCE_H macro as header inclusion guard up to commit ec691ca3 which changed this to HEADER_CURL_SETUP_ONCE_H, this single inclusion guard is enough to ensure that inclusion of lib/setup_once.h done from lib/setup.h is only done once. Additionally lib/setup.h has always used __SETUP_ONCE_H macro to protect inclusion of setup_once.h even after commit ec691ca3, this was to avoid a circular header inclusion triggered when building a c-ares enabled version with c-ares sources available which also has a setup_once.h header. Commit ec691ca3 exposes the real nature of __SETUP_ONCE_H usage in lib/setup.h, it is a header inclusion guard foreign to libcurl belonging to c-ares's setup_once.h The renaming this commit does, fixes the circular header inclusion, and as such removes the need and usage of a header inclusion guard foreign to libcurl. Macro __SETUP_ONCE_H no longer used in libcurl. 2 - Due to the circular interdependency of old lib/setup_once.h and the c-ares setup_once.h header, old file lib/setup_once.h has carried back from 2006 up to now days an alarming and prominent notice about the need of keeping libcurl's and c-ares's setup_once.h in sync. Given that this commit fixes the circular interdependency, the need and presence of mentioned notice is removed. All mentioned interdependencies come back from now old days when the c-ares project lived inside a curl subdirectory. This commit removes last traces of such fact.
2013-01-06 22:06:49 +04:00
#include "curl_setup.h"
#ifndef CURL_DISABLE_FILE
1999-12-29 17:20:26 +03:00
2000-08-23 11:23:42 +04:00
#ifdef HAVE_NETINET_IN_H
1999-12-29 17:20:26 +03:00
#include <netinet/in.h>
2000-08-23 11:23:42 +04:00
#endif
#ifdef HAVE_NETDB_H
1999-12-29 17:20:26 +03:00
#include <netdb.h>
2000-08-23 11:23:42 +04:00
#endif
1999-12-29 17:20:26 +03:00
#ifdef HAVE_ARPA_INET_H
#include <arpa/inet.h>
#endif
#ifdef HAVE_NET_IF_H
#include <net/if.h>
#endif
#ifdef HAVE_SYS_IOCTL_H
1999-12-29 17:20:26 +03:00
#include <sys/ioctl.h>
#endif
1999-12-29 17:20:26 +03:00
#ifdef HAVE_SYS_PARAM_H
#include <sys/param.h>
#endif
#ifdef HAVE_FCNTL_H
#include <fcntl.h>
#endif
#include "strtoofft.h"
#include "urldata.h"
1999-12-29 17:20:26 +03:00
#include <curl/curl.h>
#include "progress.h"
#include "sendf.h"
#include "escape.h"
#include "file.h"
#include "speedcheck.h"
#include "getinfo.h"
#include "transfer.h"
#include "url.h"
#include "parsedate.h" /* for the week day and month names */
#include "warnless.h"
#include "curl_range.h"
/* The last 3 #include files should be in this order */
#include "curl_printf.h"
#include "curl_memory.h"
#include "memdebug.h"
1999-12-29 17:20:26 +03:00
#if defined(WIN32) || defined(MSDOS) || defined(__EMX__) || \
defined(__SYMBIAN32__)
#define DOS_FILESYSTEM 1
2008-04-17 04:45:33 +04:00
#endif
2009-06-04 23:11:11 +04:00
#ifdef OPEN_NEEDS_ARG3
# define open_readonly(p,f) open((p),(f),(0))
#else
# define open_readonly(p,f) open((p),(f))
#endif
/*
* Forward declarations.
*/
static CURLcode file_do(struct connectdata *, bool *done);
static CURLcode file_done(struct connectdata *conn,
CURLcode status, bool premature);
static CURLcode file_connect(struct connectdata *conn, bool *done);
static CURLcode file_disconnect(struct connectdata *conn,
bool dead_connection);
static CURLcode file_setup_connection(struct connectdata *conn);
/*
* FILE scheme handler.
*/
const struct Curl_handler Curl_handler_file = {
"FILE", /* scheme */
file_setup_connection, /* setup_connection */
file_do, /* do_it */
file_done, /* done */
ZERO_NULL, /* do_more */
file_connect, /* connect_it */
ZERO_NULL, /* connecting */
ZERO_NULL, /* doing */
ZERO_NULL, /* proto_getsock */
ZERO_NULL, /* doing_getsock */
ZERO_NULL, /* domore_getsock */
ZERO_NULL, /* perform_getsock */
file_disconnect, /* disconnect */
ZERO_NULL, /* readwrite */
ZERO_NULL, /* connection_check */
0, /* defport */
CURLPROTO_FILE, /* protocol */
PROTOPT_NONETWORK | PROTOPT_NOURLQUERY /* flags */
};
static CURLcode file_setup_connection(struct connectdata *conn)
{
/* allocate the FILE specific struct */
conn->data->req.protop = calloc(1, sizeof(struct FILEPROTO));
if(!conn->data->req.protop)
return CURLE_OUT_OF_MEMORY;
return CURLE_OK;
}
2004-04-26 11:11:39 +04:00
/*
* file_connect() gets called from Curl_protocol_connect() to allow us to
2004-04-26 11:11:39 +04:00
* do protocol-specific actions at connect-time. We emulate a
* connect-then-transfer protocol and "connect" to the file here
*/
static CURLcode file_connect(struct connectdata *conn, bool *done)
1999-12-29 17:20:26 +03:00
{
struct Curl_easy *data = conn->data;
char *real_path;
struct FILEPROTO *file = data->req.protop;
1999-12-29 17:20:26 +03:00
int fd;
#ifdef DOS_FILESYSTEM
size_t i;
char *actual_path;
2001-03-14 11:28:19 +03:00
#endif
size_t real_path_len;
CURLcode result = Curl_urldecode(data, data->state.up.path, 0, &real_path,
&real_path_len, FALSE);
if(result)
return result;
#ifdef DOS_FILESYSTEM
/* If the first character is a slash, and there's
something that looks like a drive at the beginning of
the path, skip the slash. If we remove the initial
slash in all cases, paths without drive letters end up
relative to the current directory which isn't how
browsers work.
Some browsers accept | instead of : as the drive letter
separator, so we do too.
On other platforms, we need the slash to indicate an
absolute pathname. On Windows, absolute paths start
with a drive letter.
*/
actual_path = real_path;
if((actual_path[0] == '/') &&
2003-08-08 21:56:47 +04:00
actual_path[1] &&
(actual_path[2] == ':' || actual_path[2] == '|')) {
actual_path[2] = ':';
actual_path++;
real_path_len--;
}
2004-11-09 17:57:11 +03:00
/* change path separators from '/' to '\\' for DOS, Windows and OS/2 */
for(i = 0; i < real_path_len; ++i)
if(actual_path[i] == '/')
1999-12-29 17:20:26 +03:00
actual_path[i] = '\\';
else if(!actual_path[i]) { /* binary zero */
Curl_safefree(real_path);
return CURLE_URL_MALFORMAT;
}
1999-12-29 17:20:26 +03:00
fd = open_readonly(actual_path, O_RDONLY|O_BINARY);
file->path = actual_path;
1999-12-29 17:20:26 +03:00
#else
if(memchr(real_path, 0, real_path_len)) {
/* binary zeroes indicate foul play */
Curl_safefree(real_path);
return CURLE_URL_MALFORMAT;
}
2009-06-04 23:11:11 +04:00
fd = open_readonly(real_path, O_RDONLY);
file->path = real_path;
1999-12-29 17:20:26 +03:00
#endif
file->freepath = real_path; /* free this when done */
1999-12-29 17:20:26 +03:00
file->fd = fd;
if(!data->set.upload && (fd == -1)) {
failf(data, "Couldn't open file %s", data->state.up.path);
file_done(conn, CURLE_FILE_COULDNT_READ_FILE, FALSE);
2000-05-22 18:12:12 +04:00
return CURLE_FILE_COULDNT_READ_FILE;
1999-12-29 17:20:26 +03:00
}
*done = TRUE;
return CURLE_OK;
}
static CURLcode file_done(struct connectdata *conn,
CURLcode status, bool premature)
{
struct FILEPROTO *file = conn->data->req.protop;
(void)status; /* not used */
(void)premature; /* not used */
if(file) {
Curl_safefree(file->freepath);
file->path = NULL;
if(file->fd != -1)
close(file->fd);
file->fd = -1;
}
return CURLE_OK;
}
static CURLcode file_disconnect(struct connectdata *conn,
bool dead_connection)
{
struct FILEPROTO *file = conn->data->req.protop;
(void)dead_connection; /* not used */
if(file) {
Curl_safefree(file->freepath);
file->path = NULL;
if(file->fd != -1)
close(file->fd);
file->fd = -1;
}
return CURLE_OK;
}
#ifdef DOS_FILESYSTEM
#define DIRSEP '\\'
#else
#define DIRSEP '/'
#endif
static CURLcode file_upload(struct connectdata *conn)
{
struct FILEPROTO *file = conn->data->req.protop;
2007-08-29 09:36:53 +04:00
const char *dir = strchr(file->path, DIRSEP);
int fd;
int mode;
CURLcode result = CURLE_OK;
struct Curl_easy *data = conn->data;
char *buf = data->state.buffer;
curl_off_t bytecount = 0;
struct_stat file_stat;
const char *buf2;
/*
* Since FILE: doesn't do the full init, we need to provide some extra
* assignments here.
*/
conn->data->req.upload_fromhere = buf;
if(!dir)
return CURLE_FILE_COULDNT_READ_FILE; /* fix: better error code */
if(!dir[1])
return CURLE_FILE_COULDNT_READ_FILE; /* fix: better error code */
#ifdef O_BINARY
#define MODE_DEFAULT O_WRONLY|O_CREAT|O_BINARY
#else
#define MODE_DEFAULT O_WRONLY|O_CREAT
#endif
if(data->state.resume_from)
mode = MODE_DEFAULT|O_APPEND;
else
mode = MODE_DEFAULT|O_TRUNC;
fd = open(file->path, mode, conn->data->set.new_file_perms);
if(fd < 0) {
failf(data, "Can't open %s for writing", file->path);
return CURLE_WRITE_ERROR;
}
if(-1 != data->state.infilesize)
/* known size of data to "upload" */
Curl_pgrsSetUploadSize(data, data->state.infilesize);
/* treat the negative resume offset value as the case of "-" */
if(data->state.resume_from < 0) {
if(fstat(fd, &file_stat)) {
close(fd);
failf(data, "Can't get the size of %s", file->path);
return CURLE_WRITE_ERROR;
}
data->state.resume_from = (curl_off_t)file_stat.st_size;
}
while(!result) {
size_t nread;
size_t nwrite;
size_t readcount;
result = Curl_fillreadbuffer(conn, data->set.buffer_size, &readcount);
if(result)
break;
if(!readcount)
break;
nread = readcount;
2005-12-30 03:07:25 +03:00
/*skip bytes before resume point*/
if(data->state.resume_from) {
2016-04-03 21:28:34 +03:00
if((curl_off_t)nread <= data->state.resume_from) {
data->state.resume_from -= nread;
nread = 0;
buf2 = buf;
}
else {
buf2 = buf + data->state.resume_from;
2008-01-11 19:49:35 +03:00
nread -= (size_t)data->state.resume_from;
data->state.resume_from = 0;
}
}
else
buf2 = buf;
/* write the data to the target */
nwrite = write(fd, buf2, nread);
if(nwrite != nread) {
result = CURLE_SEND_ERROR;
break;
}
bytecount += nread;
Curl_pgrsSetUploadCounter(data, bytecount);
if(Curl_pgrsUpdate(conn))
result = CURLE_ABORTED_BY_CALLBACK;
else
result = Curl_speedcheck(data, Curl_now());
}
if(!result && Curl_pgrsUpdate(conn))
result = CURLE_ABORTED_BY_CALLBACK;
close(fd);
return result;
}
2004-04-26 11:11:39 +04:00
/*
* file_do() is the protocol-specific function for the do-phase, separated
2004-04-26 11:11:39 +04:00
* from the connect-phase above. Other protocols merely setup the transfer in
* the do-phase, to have it done in the main transfer loop but since some
* platforms we support don't allow select()ing etc on file handles (as
* opposed to sockets) we instead perform the whole do-operation in this
* function.
*/
static CURLcode file_do(struct connectdata *conn, bool *done)
{
/* This implementation ignores the host name in conformance with
RFC 1738. Only local files (reachable via the standard file system)
are supported. This means that files on remotely mounted directories
(via NFS, Samba, NT sharing) can be accessed through a file:// URL
*/
CURLcode result = CURLE_OK;
struct_stat statbuf; /* struct_stat instead of struct stat just to allow the
Windows version to have a different struct without
having to redefine the simple word 'stat' */
curl_off_t expected_size = 0;
bool size_known;
bool fstated = FALSE;
struct Curl_easy *data = conn->data;
char *buf = data->state.buffer;
2004-01-22 15:45:50 +03:00
curl_off_t bytecount = 0;
int fd;
struct FILEPROTO *file;
*done = TRUE; /* unconditionally */
Curl_pgrsStartNow(data);
if(data->set.upload)
return file_upload(conn);
file = conn->data->req.protop;
/* get the fd from the connection phase */
fd = file->fd;
/* VMS: This only works reliable for STREAMLF files */
if(-1 != fstat(fd, &statbuf)) {
1999-12-29 17:20:26 +03:00
/* we could stat it, then read out the size */
expected_size = statbuf.st_size;
/* and store the modification time */
data->info.filetime = statbuf.st_mtime;
fstated = TRUE;
}
if(fstated && !data->state.range && data->set.timecondition) {
if(!Curl_meets_timecondition(data, data->info.filetime)) {
*done = TRUE;
return CURLE_OK;
}
}
if(fstated) {
time_t filetime;
struct tm buffer;
const struct tm *tm = &buffer;
char header[80];
msnprintf(header, sizeof(header),
"Content-Length: %" CURL_FORMAT_CURL_OFF_T "\r\n",
expected_size);
result = Curl_client_write(conn, CLIENTWRITE_HEADER, header, 0);
if(result)
return result;
result = Curl_client_write(conn, CLIENTWRITE_HEADER,
2004-06-24 19:06:25 +04:00
(char *)"Accept-ranges: bytes\r\n", 0);
if(result)
return result;
filetime = (time_t)statbuf.st_mtime;
result = Curl_gmtime(filetime, &buffer);
if(result)
return result;
/* format: "Tue, 15 Nov 1994 12:45:26 GMT" */
msnprintf(header, sizeof(header),
"Last-Modified: %s, %02d %s %4d %02d:%02d:%02d GMT\r\n%s",
Curl_wkday[tm->tm_wday?tm->tm_wday-1:6],
tm->tm_mday,
Curl_month[tm->tm_mon],
tm->tm_year + 1900,
tm->tm_hour,
tm->tm_min,
tm->tm_sec,
data->set.opt_no_body ? "": "\r\n");
result = Curl_client_write(conn, CLIENTWRITE_HEADER, header, 0);
if(result)
return result;
/* set the file size to make it available post transfer */
Curl_pgrsSetDownloadSize(data, expected_size);
if(data->set.opt_no_body)
return result;
1999-12-29 17:20:26 +03:00
}
/* Check whether file range has been specified */
result = Curl_range(conn);
if(result)
return result;
/* Adjust the start offset in case we want to get the N last bytes
* of the stream if the filesize could be determined */
if(data->state.resume_from < 0) {
if(!fstated) {
failf(data, "Can't get the size of file.");
return CURLE_READ_ERROR;
}
data->state.resume_from += (curl_off_t)statbuf.st_size;
}
if(data->state.resume_from <= expected_size)
expected_size -= data->state.resume_from;
else {
failf(data, "failed to resume file:// transfer");
return CURLE_BAD_DOWNLOAD_RESUME;
}
/* A high water mark has been specified so we obey... */
if(data->req.maxdownload > 0)
expected_size = data->req.maxdownload;
if(!fstated || (expected_size == 0))
size_known = FALSE;
else
size_known = TRUE;
1999-12-29 17:20:26 +03:00
/* The following is a shortcut implementation of file reading
this is both more efficient than the former call to download() and
it avoids problems with select() and recv() on file descriptors
in Winsock */
if(fstated)
Curl_pgrsSetDownloadSize(data, expected_size);
if(data->state.resume_from) {
if(data->state.resume_from !=
lseek(fd, data->state.resume_from, SEEK_SET))
return CURLE_BAD_DOWNLOAD_RESUME;
}
Curl_pgrsTime(data, TIMER_STARTTRANSFER);
while(!result) {
ssize_t nread;
/* Don't fill a whole buffer if we want less than all data */
size_t bytestoread;
if(size_known) {
bytestoread = (expected_size < data->set.buffer_size) ?
curlx_sotouz(expected_size) : (size_t)data->set.buffer_size;
}
else
bytestoread = data->set.buffer_size-1;
2011-10-14 20:23:16 +04:00
2008-01-11 19:49:35 +03:00
nread = read(fd, buf, bytestoread);
1999-12-29 17:20:26 +03:00
if(nread > 0)
1999-12-29 17:20:26 +03:00
buf[nread] = 0;
if(nread <= 0 || (size_known && (expected_size == 0)))
1999-12-29 17:20:26 +03:00
break;
1999-12-29 17:20:26 +03:00
bytecount += nread;
if(size_known)
expected_size -= nread;
2000-11-22 15:51:18 +03:00
result = Curl_client_write(conn, CLIENTWRITE_BODY, buf, nread);
if(result)
return result;
2000-11-22 15:51:18 +03:00
Curl_pgrsSetDownloadCounter(data, bytecount);
if(Curl_pgrsUpdate(conn))
result = CURLE_ABORTED_BY_CALLBACK;
else
result = Curl_speedcheck(data, Curl_now());
1999-12-29 17:20:26 +03:00
}
if(Curl_pgrsUpdate(conn))
result = CURLE_ABORTED_BY_CALLBACK;
1999-12-29 17:20:26 +03:00
return result;
1999-12-29 17:20:26 +03:00
}
#endif