mirror/curl
2
0
Fork 0
mirror of https://github.com/curl/curl.git synced 2024-12-27 06:59:43 +08:00
Code Issues Packages Projects Releases Wiki Activity
d7b6ce64ce
curl/lib/file.c
Stefan Eissing d7b6ce64ce
lib: replace readwrite with write_resp 
This clarifies the handling of server responses by folding the code for
the complicated protocols into their protocol handlers. This concerns
mainly HTTP and its bastard sibling RTSP.

The terms "read" and "write" are often used without clear context if
they refer to the connect or the client/application side of a
transfer. This PR uses "read/write" for operations on the client side
and "send/receive" for the connection, e.g. server side. If this is
considered useful, we can revisit renaming of further methods in another
PR.

Curl's protocol handler `readwrite()` method been changed:

```diff
-  CURLcode (*readwrite)(struct Curl_easy *data, struct connectdata *conn,
-                        const char *buf, size_t blen,
-                        size_t *pconsumed, bool *readmore);
+  CURLcode (*write_resp)(struct Curl_easy *data, const char *buf, size_t blen,
+                         bool is_eos, bool *done);
```

The name was changed to clarify that this writes reponse data to the
client side. The parameter changes are:

* `conn` removed as it always operates on `data->conn`
* `pconsumed` removed as the method needs to handle all data on success
* `readmore` removed as no longer necessary
* `is_eos` as indicator that this is the last call for the transfer
  response (end-of-stream).
* `done` TRUE on return iff the transfer response is to be treated as
  finished

This change affects many files only because of updated comments in
handlers that provide no implementation. The real change is that the
HTTP protocol handlers now provide an implementation.

The HTTP protocol handlers `write_resp()` implementation will get passed
**all** raw data of a server response for the transfer. The HTTP/1.x
formatted status and headers, as well as the undecoded response
body. `Curl_http_write_resp_hds()` is used internally to parse the
response headers and pass them on. This method is public as the RTSP
protocol handler also uses it.

HTTP/1.1 "chunked" transport encoding is now part of the general
*content encoding* writer stack, just like other encodings. A new flag
`CLIENTWRITE_EOS` was added for the last client write. This allows
writers to verify that they are in a valid end state. The chunked
decoder will check if it indeed has seen the last chunk.

The general response handling in `transfer.c:466` happens in function
`readwrite_data()`. This mainly operates now like:

```
static CURLcode readwrite_data(data, ...)
{
  do {
    Curl_xfer_recv_resp(data, buf)
    ...
    Curl_xfer_write_resp(data, buf)
    ...
  } while(interested);
  ...
}
```

All the response data handling is implemented in
`Curl_xfer_write_resp()`. It calls the protocol handler's `write_resp()`
implementation if available, or does the default behaviour.

All raw response data needs to pass through this function. Which also
means that anyone in possession of such data may call
`Curl_xfer_write_resp()`.

Closes #12480
2024-01-13 17:23:42 +01:00

582 lines
16 KiB
C
Raw Blame History

/***************************************************************************
* _ _ ____ _
* Project ___| | | | _ \| |
* / __| | | | |_) | |
* | (__| |_| | _ <| |___
* \___|\___/|_| \_\_____|
*
* Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
*
* 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.se/docs/copyright.html.
*
* 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
* furnished to do so, under the terms of the COPYING file.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
* SPDX-License-Identifier: curl
*
***************************************************************************/
#include "curl_setup.h"
#ifndef CURL_DISABLE_FILE
#ifdef HAVE_NETINET_IN_H
#include <netinet/in.h>
#endif
#ifdef HAVE_NETDB_H
#include <netdb.h>
#endif
#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
#include <sys/ioctl.h>
#endif
#ifdef HAVE_SYS_PARAM_H
#include <sys/param.h>
#endif
#ifdef HAVE_FCNTL_H
#include <fcntl.h>
#endif
#include "strtoofft.h"
#include "urldata.h"
#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"
#if defined(_WIN32) || defined(MSDOS) || defined(__EMX__)
#define DOS_FILESYSTEM 1
#elif defined(__amigaos4__)
#define AMIGA_FILESYSTEM 1
#endif
#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 Curl_easy *data, bool *done);
static CURLcode file_done(struct Curl_easy *data,
CURLcode status, bool premature);
static CURLcode file_connect(struct Curl_easy *data, bool *done);
static CURLcode file_disconnect(struct Curl_easy *data,
struct connectdata *conn,
bool dead_connection);
static CURLcode file_setup_connection(struct Curl_easy *data,
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, /* write_resp */
ZERO_NULL, /* connection_check */
ZERO_NULL, /* attach connection */
0, /* defport */
CURLPROTO_FILE, /* protocol */
CURLPROTO_FILE, /* family */
PROTOPT_NONETWORK | PROTOPT_NOURLQUERY /* flags */
};
static CURLcode file_setup_connection(struct Curl_easy *data,
struct connectdata *conn)
{
(void)conn;
/* allocate the FILE specific struct */
data->req.p.file = calloc(1, sizeof(struct FILEPROTO));
if(!data->req.p.file)
return CURLE_OUT_OF_MEMORY;
return CURLE_OK;
}
/*
* file_connect() gets called from Curl_protocol_connect() to allow us to
* 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 Curl_easy *data, bool *done)
{
char *real_path;
struct FILEPROTO *file = data->req.p.file;
int fd;
#ifdef DOS_FILESYSTEM
size_t i;
char *actual_path;
#endif
size_t real_path_len;
CURLcode result;
if(file->path) {
/* already connected.
* the handler->connect_it() is normally only called once, but
* FILE does a special check on setting up the connection which
* calls this explicitly. */
*done = TRUE;
return CURLE_OK;
}
result = Curl_urldecode(data->state.up.path, 0, &real_path,
&real_path_len, REJECT_ZERO);
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] == '/') &&
actual_path[1] &&
(actual_path[2] == ':' || actual_path[2] == '|')) {
actual_path[2] = ':';
actual_path++;
real_path_len--;
}
/* change path separators from '/' to '\\' for DOS, Windows and OS/2 */
for(i = 0; i < real_path_len; ++i)
if(actual_path[i] == '/')
actual_path[i] = '\\';
else if(!actual_path[i]) { /* binary zero */
Curl_safefree(real_path);
return CURLE_URL_MALFORMAT;
}
fd = open_readonly(actual_path, O_RDONLY|O_BINARY);
file->path = actual_path;
#else
if(memchr(real_path, 0, real_path_len)) {
/* binary zeroes indicate foul play */
Curl_safefree(real_path);
return CURLE_URL_MALFORMAT;
}
#ifdef AMIGA_FILESYSTEM
/*
* A leading slash in an AmigaDOS path denotes the parent
* directory, and hence we block this as it is relative.
* Absolute paths start with 'volumename:', so we check for
* this first. Failing that, we treat the path as a real unix
* path, but only if the application was compiled with -lunix.
*/
fd = -1;
file->path = real_path;
if(real_path[0] == '/') {
extern int __unix_path_semantics;
if(strchr(real_path + 1, ':')) {
/* Amiga absolute path */
fd = open_readonly(real_path + 1, O_RDONLY);
file->path++;
}
else if(__unix_path_semantics) {
/* -lunix fallback */
fd = open_readonly(real_path, O_RDONLY);
}
}
#else
fd = open_readonly(real_path, O_RDONLY);
file->path = real_path;
#endif
#endif
Curl_safefree(file->freepath);
file->freepath = real_path; /* free this when done */
file->fd = fd;
if(!data->state.upload && (fd == -1)) {
failf(data, "Couldn't open file %s", data->state.up.path);
file_done(data, CURLE_FILE_COULDNT_READ_FILE, FALSE);
return CURLE_FILE_COULDNT_READ_FILE;
}
*done = TRUE;
return CURLE_OK;
}
static CURLcode file_done(struct Curl_easy *data,
CURLcode status, bool premature)
{
struct FILEPROTO *file = data->req.p.file;
(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 Curl_easy *data,
struct connectdata *conn,
bool dead_connection)
{
(void)dead_connection; /* not used */
(void)conn;
return file_done(data, CURLE_OK, FALSE);
}
#ifdef DOS_FILESYSTEM
#define DIRSEP '\\'
#else
#define DIRSEP '/'
#endif
static CURLcode file_upload(struct Curl_easy *data)
{
struct FILEPROTO *file = data->req.p.file;
const char *dir = strchr(file->path, DIRSEP);
int fd;
int mode;
CURLcode result = CURLE_OK;
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.
*/
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, 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;
ssize_t nwrite;
size_t readcount;
result = Curl_fillreadbuffer(data, data->set.buffer_size, &readcount);
if(result)
break;
if(!readcount)
break;
nread = readcount;
/* skip bytes before resume point */
if(data->state.resume_from) {
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;
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((size_t)nwrite != nread) {
result = CURLE_SEND_ERROR;
break;
}
bytecount += nread;
Curl_pgrsSetUploadCounter(data, bytecount);
if(Curl_pgrsUpdate(data))
result = CURLE_ABORTED_BY_CALLBACK;
else
result = Curl_speedcheck(data, Curl_now());
}
if(!result && Curl_pgrsUpdate(data))
result = CURLE_ABORTED_BY_CALLBACK;
close(fd);
return result;
}
/*
* file_do() is the protocol-specific function for the do-phase, separated
* 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 Curl_easy *data, 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 = -1;
bool size_known;
bool fstated = FALSE;
char *buf = data->state.buffer;
int fd;
struct FILEPROTO *file;
*done = TRUE; /* unconditionally */
if(data->state.upload)
return file_upload(data);
file = data->req.p.file;
/* get the fd from the connection phase */
fd = file->fd;
/* VMS: This only works reliable for STREAMLF files */
if(-1 != fstat(fd, &statbuf)) {
if(!S_ISDIR(statbuf.st_mode))
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];
int headerlen;
char accept_ranges[24]= { "Accept-ranges: bytes\r\n" };
if(expected_size >= 0) {
headerlen = msnprintf(header, sizeof(header),
"Content-Length: %" CURL_FORMAT_CURL_OFF_T "\r\n",
expected_size);
result = Curl_client_write(data, CLIENTWRITE_HEADER, header, headerlen);
if(result)
return result;
result = Curl_client_write(data, CLIENTWRITE_HEADER,
accept_ranges, strlen(accept_ranges));
if(result != CURLE_OK)
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" */
headerlen = 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->req.no_body ? "": "\r\n");
result = Curl_client_write(data, CLIENTWRITE_HEADER, header, headerlen);
if(result)
return result;
/* set the file size to make it available post transfer */
Curl_pgrsSetDownloadSize(data, expected_size);
if(data->req.no_body)
return result;
}
/* Check whether file range has been specified */
result = Curl_range(data);
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 > 0) {
/* We check explicitly if we have a start offset, because
* expected_size may be -1 if we don't know how large the file is,
* in which case we should not adjust it. */
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;
/* 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(size_known)
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;
nread = read(fd, buf, bytestoread);
if(nread > 0)
buf[nread] = 0;
if(nread <= 0 || (size_known && (expected_size == 0)))
break;
if(size_known)
expected_size -= nread;
result = Curl_client_write(data, CLIENTWRITE_BODY, buf, nread);
if(result)
return result;
if(Curl_pgrsUpdate(data))
result = CURLE_ABORTED_BY_CALLBACK;
else
result = Curl_speedcheck(data, Curl_now());
}
if(Curl_pgrsUpdate(data))
result = CURLE_ABORTED_BY_CALLBACK;
return result;
}
#endif
Reference in New Issue View Git Blame Copy Permalink