curl_easy_setopt(3) libcurl Manual curl_easy_setopt(3)
NAME
curl_easy_setopt - set options for a curl easy handle
SYNOPSIS
#include <curl/curl.h>
CURLcode curl_easy_setopt(CURL *handle, CURLoption option, parameter);
DESCRIPTION
curl_easy_setopt() is used to tell libcurl how to behave. By using the appropriate options to curl_easy_setopt, you can change libcurl's
behavior. All options are set with the option followed by a parameter. That parameter can be a long, a function pointer, an object pointer
or a curl_off_t, depending on what the specific option expects. Read this manual carefully as bad input values may cause libcurl to behave
badly! You can only set one option in each function call. A typical application uses many curl_easy_setopt() calls in the setup phase.
Options set with this function call are valid for all forthcoming transfers performed using this handle. The options are not in any way
reset between transfers, so if you want subsequent transfers with different options, you must change them between the transfers. You can
optionally reset all options back to internal default with curl_easy_reset(3).
Strings passed to libcurl as 'char *' arguments, are copied by the library; thus the string storage associated to the pointer argument may
be overwritten after curl_easy_setopt() returns. Exceptions to this rule are described in the option details below.
Before version 7.17.0, strings were not copied. Instead the user was forced keep them available until libcurl no longer needed them.
The handle is the return code from a curl_easy_init(3) or curl_easy_duphandle(3) call.
BEHAVIOR OPTIONS
CURLOPT_VERBOSE
Set the parameter to 1 to get the library to display a lot of verbose information about its operations. Very useful for libcurl
and/or protocol debugging and understanding. The verbose information will be sent to stderr, or the stream set with CURLOPT_STDERR.
You hardly ever want this set in production use, you will almost always want this when you debug/report problems. Another neat
option for debugging is the CURLOPT_DEBUGFUNCTION.
CURLOPT_HEADER
A parameter set to 1 tells the library to include the header in the body output. This is only relevant for protocols that actually
have headers preceding the data (like HTTP).
CURLOPT_NOPROGRESS
Pass a long. If set to 1, it tells the library to shut off the progress meter completely. It will also prevent the CURLOPT_PROGRESS-
FUNCTION from getting called.
Future versions of libcurl are likely to not have any built-in progress meter at all.
CURLOPT_NOSIGNAL
Pass a long. If it is 1, libcurl will not use any functions that install signal handlers or any functions that cause signals to be
sent to the process. This option is mainly here to allow multi-threaded unix applications to still set/use all timeout options etc,
without risking getting signals. (Added in 7.10)
If this option is set and libcurl has been built with the standard name resolver, timeouts will not occur while the name resolve
takes place. Consider building libcurl with c-ares support to enable asynchronous DNS lookups, which enables nice timeouts for name
resolves without signals.
Setting CURLOPT_NOSIGNAL to 1 makes libcurl NOT ask the system to ignore SIGPIPE signals, which otherwise are sent by the system
when trying to send data to a socket which is closed in the other end. libcurl makes an effort to never cause such SIGPIPEs to trig-
ger, but some operating systems have no way to avoid them and even on those that have there are some corner cases when they may
still happen, contrary to our desire. In addition, using CURLAUTH_NTLM_WB authentication could cause a SIGCHLD signal to be raised.
CURLOPT_WILDCARDMATCH
Set this option to 1 if you want to transfer multiple files according to a file name pattern. The pattern can be specified as part
of the CURLOPT_URL option, using an fnmatch-like pattern (Shell Pattern Matching) in the last part of URL (file name).
By default, libcurl uses its internal wildcard matching implementation. You can provide your own matching function by the CUR-
LOPT_FNMATCH_FUNCTION option.
This feature is only supported by the FTP download for now.
A brief introduction of its syntax follows:
* - ASTERISK
ftp://example.com/some/path/*.txt (for all txt's from the root directory)
? - QUESTION MARK
Question mark matches any (exactly one) character.
ftp://example.com/some/path/photo?.jpeg
[ - BRACKET EXPRESSION
The left bracket opens a bracket expression. The question mark and asterisk have no special meaning in a bracket expression.
Each bracket expression ends by the right bracket and matches exactly one character. Some examples follow:
[a-zA-Z0-9] or [f-gF-G] - character interval
[abc] - character enumeration
[^abc] or [!abc] - negation
[[:name:]] class expression. Supported classes are alnum,lower, space, alpha, digit, print, upper, blank, graph, xdigit.
[][-!^] - special case - matches only '-', ']', '[', '!' or '^'. These characters have no special purpose.
[[]\] - escape syntax. Matches '[', ']' or ''.
Using the rules above, a file name pattern can be constructed:
ftp://example.com/some/path/[a-z[:upper:]\].jpeg
(This was added in 7.21.0)
CALLBACK OPTIONS
CURLOPT_WRITEFUNCTION
Pass a pointer to a function that matches the following prototype: size_t function( char *ptr, size_t size, size_t nmemb, void
*userdata); This function gets called by libcurl as soon as there is data received that needs to be saved. The size of the data
pointed to by ptr is size multiplied with nmemb, it will not be zero terminated. Return the number of bytes actually taken care of.
If that amount differs from the amount passed to your function, it'll signal an error to the library. This will abort the transfer
and return CURLE_WRITE_ERROR.
From 7.18.0, the function can return CURL_WRITEFUNC_PAUSE which then will cause writing to this connection to become paused. See
curl_easy_pause(3) for further details.
This function may be called with zero bytes data if the transferred file is empty.
Set this option to NULL to get the internal default function. The internal default function will write the data to the FILE * given
with CURLOPT_WRITEDATA.
Set the userdata argument with the CURLOPT_WRITEDATA option.
The callback function will be passed as much data as possible in all invokes, but you cannot possibly make any assumptions. It may
be one byte, it may be thousands. The maximum amount of body data that can be passed to the write callback is defined in the curl.h
header file: CURL_MAX_WRITE_SIZE (the usual default is 16K). If you however have CURLOPT_HEADER set, which sends header data to the
write callback, you can get up to CURL_MAX_HTTP_HEADER bytes of header data passed into it. This usually means 100K.
CURLOPT_WRITEDATA
Data pointer to pass to the file write function. If you use the CURLOPT_WRITEFUNCTION option, this is the pointer you'll get as
input. If you don't use a callback, you must pass a 'FILE *' as libcurl will pass this to fwrite() when writing data.
The internal CURLOPT_WRITEFUNCTION will write the data to the FILE * given with this option, or to stdout if this option hasn't been
set.
If you're using libcurl as a win32 DLL, you MUST use the CURLOPT_WRITEFUNCTION if you set this option or you will experience
crashes.
This option is also known with the older name CURLOPT_FILE, the name CURLOPT_WRITEDATA was introduced in 7.9.7.
CURLOPT_READFUNCTION
Pass a pointer to a function that matches the following prototype: size_t function( void *ptr, size_t size, size_t nmemb, void
*userdata); This function gets called by libcurl as soon as it needs to read data in order to send it to the peer. The data area
pointed at by the pointer ptr may be filled with at most size multiplied with nmemb number of bytes. Your function must return the
actual number of bytes that you stored in that memory area. Returning 0 will signal end-of-file to the library and cause it to stop
the current transfer.
If you stop the current transfer by returning 0 "pre-maturely" (i.e before the server expected it, like when you've said you will
upload N bytes and you upload less than N bytes), you may experience that the server "hangs" waiting for the rest of the data that
won't come.
The read callback may return CURL_READFUNC_ABORT to stop the current operation immediately, resulting in a CURLE_ABORTED_BY_CALLBACK
error code from the transfer (Added in 7.12.1)
From 7.18.0, the function can return CURL_READFUNC_PAUSE which then will cause reading from this connection to become paused. See
curl_easy_pause(3) for further details.
Bugs: when doing TFTP uploads, you must return the exact amount of data that the callback wants, or it will be considered the final
packet by the server end and the transfer will end there.
If you set this callback pointer to NULL, or don't set it at all, the default internal read function will be used. It is doing an
fread() on the FILE * userdata set with CURLOPT_READDATA.
CURLOPT_READDATA
Data pointer to pass to the file read function. If you use the CURLOPT_READFUNCTION option, this is the pointer you'll get as input.
If you don't specify a read callback but instead rely on the default internal read function, this data must be a valid readable FILE
*.
If you're using libcurl as a win32 DLL, you MUST use a CURLOPT_READFUNCTION if you set this option.
This option was also known by the older name CURLOPT_INFILE, the name CURLOPT_READDATA was introduced in 7.9.7.
CURLOPT_IOCTLFUNCTION
Pass a pointer to a function that matches the following prototype: curlioerr function(CURL *handle, int cmd, void *clientp);. This
function gets called by libcurl when something special I/O-related needs to be done that the library can't do by itself. For now,
rewinding the read data stream is the only action it can request. The rewinding of the read data stream may be necessary when doing
a HTTP PUT or POST with a multi-pass authentication method. (Option added in 7.12.3).
Use CURLOPT_SEEKFUNCTION instead to provide seeking!
CURLOPT_IOCTLDATA
Pass a pointer that will be untouched by libcurl and passed as the 3rd argument in the ioctl callback set with CURLOPT_IOCTLFUNC-
TION. (Option added in 7.12.3)
CURLOPT_SEEKFUNCTION
Pass a pointer to a function that matches the following prototype: int function(void *instream, curl_off_t offset, int origin); This
function gets called by libcurl to seek to a certain position in the input stream and can be used to fast forward a file in a
resumed upload (instead of reading all uploaded bytes with the normal read function/callback). It is also called to rewind a stream
when doing a HTTP PUT or POST with a multi-pass authentication method. The function shall work like "fseek" or "lseek" and accepted
SEEK_SET, SEEK_CUR and SEEK_END as argument for origin, although (in 7.18.0) libcurl only passes SEEK_SET. The callback must return
0 (CURL_SEEKFUNC_OK) on success, 1 (CURL_SEEKFUNC_FAIL) to cause the upload operation to fail or 2 (CURL_SEEKFUNC_CANTSEEK) to indi-
cate that while the seek failed, libcurl is free to work around the problem if possible. The latter can sometimes be done by instead
reading from the input or similar.
If you forward the input arguments directly to "fseek" or "lseek", note that the data type for offset is not the same as defined for
curl_off_t on many systems! (Option added in 7.18.0)
CURLOPT_SEEKDATA
Data pointer to pass to the file seek function. If you use the CURLOPT_SEEKFUNCTION option, this is the pointer you'll get as input.
If you don't specify a seek callback, NULL is passed. (Option added in 7.18.0)
CURLOPT_SOCKOPTFUNCTION
Pass a pointer to a function that matches the following prototype: int function(void *clientp, curl_socket_t curlfd, curlsocktype
purpose);. This function gets called by libcurl after the socket() call but before the connect() call. The callback's purpose argu-
ment identifies the exact purpose for this particular socket:
CURLSOCKTYPE_IPCXN for actively created connections or since 7.28.0 CURLSOCKTYPE_ACCEPT for FTP when the connection was setup with
PORT/EPSV (in earlier versions these sockets weren't passed to this callback).
Future versions of libcurl may support more purposes. It passes the newly created socket descriptor so additional setsockopt() calls
can be done at the user's discretion. Return 0 (zero) from the callback on success. Return 1 from the callback function to signal
an unrecoverable error to the library and it will close the socket and return CURLE_COULDNT_CONNECT. (Option added in 7.16.0)
Added in 7.21.5, the callback function may return CURL_SOCKOPT_ALREADY_CONNECTED, which tells libcurl that the socket is in fact
already connected and then libcurl will not attempt to connect it.
CURLOPT_SOCKOPTDATA
Pass a pointer that will be untouched by libcurl and passed as the first argument in the sockopt callback set with CURLOPT_SOCKOPT-
FUNCTION. (Option added in 7.16.0)
CURLOPT_OPENSOCKETFUNCTION
Pass a pointer to a function that matches the following prototype: curl_socket_t function(void *clientp, curlsocktype purpose,
struct curl_sockaddr *address);. This function gets called by libcurl instead of the socket(2) call. The callback's purpose argument
identifies the exact purpose for this particular socket: CURLSOCKTYPE_IPCXN is for IP based connections. Future versions of libcurl
may support more purposes. It passes the resolved peer address as a address argument so the callback can modify the address or
refuse to connect at all. The callback function should return the socket or CURL_SOCKET_BAD in case no connection could be estab-
lished or another error was detected. Any additional setsockopt(2) calls can be done on the socket at the user's discretion.
CURL_SOCKET_BAD return value from the callback function will signal an unrecoverable error to the library and it will return
CURLE_COULDNT_CONNECT. This return code can be used for IP address blacklisting. The default behavior is:
return socket(addr->family, addr->socktype, addr->protocol);
(Option added in 7.17.1.)
CURLOPT_OPENSOCKETDATA
Pass a pointer that will be untouched by libcurl and passed as the first argument in the opensocket callback set with CUR-
LOPT_OPENSOCKETFUNCTION. (Option added in 7.17.1.)
CURLOPT_CLOSESOCKETFUNCTION
Pass a pointer to a function that matches the following prototype: int function(void *clientp, curl_socket_t item);. This function
gets called by libcurl instead of the close(3) or closesocket(3) call when sockets are closed (not for any other file descriptors).
This is pretty much the reverse to the CURLOPT_OPENSOCKETFUNCTION option. Return 0 to signal success and 1 if there was an error.
(Option added in 7.21.7)
CURLOPT_CLOSESOCKETDATA
Pass a pointer that will be untouched by libcurl and passed as the first argument in the closesocket callback set with CURLOPT_CLOS-
ESOCKETFUNCTION. (Option added in 7.21.7)
CURLOPT_PROGRESSFUNCTION
Pass a pointer to a function that matches the following prototype: int function(void *clientp, double dltotal, double dlnow, double
ultotal, double ulnow); . This function gets called by libcurl instead of its internal equivalent with a frequent interval during
operation (roughly once per second or sooner) no matter if data is being transferred or not. Unknown/unused argument values passed
to the callback will be set to zero (like if you only download data, the upload size will remain 0). Returning a non-zero value from
this callback will cause libcurl to abort the transfer and return CURLE_ABORTED_BY_CALLBACK.
If you transfer data with the multi interface, this function will not be called during periods of idleness unless you call the
appropriate libcurl function that performs transfers.
CURLOPT_NOPROGRESS must be set to 0 to make this function actually get called.
CURLOPT_PROGRESSDATA
Pass a pointer that will be untouched by libcurl and passed as the first argument in the progress callback set with CURLOPT_PRO-
GRESSFUNCTION.
CURLOPT_HEADERFUNCTION
Pass a pointer to a function that matches the following prototype: size_t function( void *ptr, size_t size, size_t nmemb, void
*userdata);. This function gets called by libcurl as soon as it has received header data. The header callback will be called once
for each header and only complete header lines are passed on to the callback. Parsing headers is very easy using this. The size of
the data pointed to by ptr is size multiplied with nmemb. Do not assume that the header line is zero terminated! The pointer named
userdata is the one you set with the CURLOPT_WRITEHEADER option. The callback function must return the number of bytes actually
taken care of. If that amount differs from the amount passed to your function, it'll signal an error to the library. This will abort
the transfer and return CURL_WRITE_ERROR.
A complete HTTP header that is passed to this function can be up to CURL_MAX_HTTP_HEADER (100K) bytes.
If this option is not set, or if it is set to NULL, but CURLOPT_HEADERDATA (CURLOPT_WRITEHEADER) is set to anything but NULL, the
function used to accept response data will be used instead. That is, it will be the function specified with CURLOPT_WRITEFUNCTION,
or if it is not specified or NULL - the default, stream-writing function.
It's important to note that the callback will be invoked for the headers of all responses received after initiating a request and
not just the final response. This includes all responses which occur during authentication negotiation. If you need to operate on
only the headers from the final response, you will need to collect headers in the callback yourself and use HTTP status lines, for
example, to delimit response boundaries.
When a server sends a chunked encoded transfer, it may contain a trailer. That trailer is identical to a HTTP header and if such a
trailer is received it is passed to the application using this callback as well. There are several ways to detect it being a trailer
and not an ordinary header: 1) it comes after the response-body. 2) it comes after the final header line (CR LF) 3) a Trailer:
header among the regular response-headers mention what header(s) to expect in the trailer.
For non-HTTP protocols like FTP, POP3, IMAP and SMTP this function will get called with the server responses to the commands that
libcurl sends.
CURLOPT_WRITEHEADER
(This option is also known as CURLOPT_HEADERDATA) Pass a pointer to be used to write the header part of the received data to. If you
don't use CURLOPT_WRITEFUNCTION or CURLOPT_HEADERFUNCTION to take care of the writing, this must be a valid FILE * as the internal
default will then be a plain fwrite(). See also the CURLOPT_HEADERFUNCTION option above on how to set a custom get-all-headers call-
back.
CURLOPT_DEBUGFUNCTION
Pass a pointer to a function that matches the following prototype: int curl_debug_callback (CURL *, curl_infotype, char *, size_t,
void *); CURLOPT_DEBUGFUNCTION replaces the standard debug function used when CURLOPT_VERBOSE is in effect. This callback receives
debug information, as specified with the curl_infotype argument. This function must return 0. The data pointed to by the char *
passed to this function WILL NOT be zero terminated, but will be exactly of the size as told by the size_t argument.
Available curl_infotype values:
CURLINFO_TEXT
The data is informational text.
CURLINFO_HEADER_IN
The data is header (or header-like) data received from the peer.
CURLINFO_HEADER_OUT
The data is header (or header-like) data sent to the peer.
CURLINFO_DATA_IN
The data is protocol data received from the peer.
CURLINFO_DATA_OUT
The data is protocol data sent to the peer.
CURLOPT_DEBUGDATA
Pass a pointer to whatever you want passed in to your CURLOPT_DEBUGFUNCTION in the last void * argument. This pointer is not used by
libcurl, it is only passed to the callback.
CURLOPT_SSL_CTX_FUNCTION
This option does only function for libcurl powered by OpenSSL. If libcurl was built against another SSL library, this functionality
is absent.
Pass a pointer to a function that matches the following prototype: CURLcode sslctxfun(CURL *curl, void *sslctx, void *parm); This
function gets called by libcurl just before the initialization of a SSL connection after having processed all other SSL related
options to give a last chance to an application to modify the behaviour of openssl's ssl initialization. The sslctx parameter is
actually a pointer to an openssl SSL_CTX. If an error is returned no attempt to establish a connection is made and the perform oper-
ation will return the error code from this callback function. Set the parm argument with the CURLOPT_SSL_CTX_DATA option. This
option was introduced in 7.11.0.
This function will get called on all new connections made to a server, during the SSL negotiation. The SSL_CTX pointer will be a new
one every time.
To use this properly, a non-trivial amount of knowledge of the openssl libraries is necessary. For example, using this function
allows you to use openssl callbacks to add additional validation code for certificates, and even to change the actual URI of a HTTPS
request (example used in the lib509 test case). See also the example section for a replacement of the key, certificate and trust
file settings.
CURLOPT_SSL_CTX_DATA
Data pointer to pass to the ssl context callback set by the option CURLOPT_SSL_CTX_FUNCTION, this is the pointer you'll get as third
parameter, otherwise NULL. (Added in 7.11.0)
CURLOPT_CONV_TO_NETWORK_FUNCTION
CURLOPT_CONV_FROM_NETWORK_FUNCTION
CURLOPT_CONV_FROM_UTF8_FUNCTION
Pass a pointer to a function that matches the following prototype: CURLcode function(char *ptr, size_t length);
These three options apply to non-ASCII platforms only. They are available only if CURL_DOES_CONVERSIONS was defined when libcurl
was built. When this is the case, curl_version_info(3) will return the CURL_VERSION_CONV feature bit set.
The data to be converted is in a buffer pointed to by the ptr parameter. The amount of data to convert is indicated by the length
parameter. The converted data overlays the input data in the buffer pointed to by the ptr parameter. CURLE_OK should be returned
upon successful conversion. A CURLcode return value defined by curl.h, such as CURLE_CONV_FAILED, should be returned if an error
was encountered.
CURLOPT_CONV_TO_NETWORK_FUNCTION and CURLOPT_CONV_FROM_NETWORK_FUNCTION convert between the host encoding and the network encoding.
They are used when commands or ASCII data are sent/received over the network.
CURLOPT_CONV_FROM_UTF8_FUNCTION is called to convert from UTF8 into the host encoding. It is required only for SSL processing.
If you set a callback pointer to NULL, or don't set it at all, the built-in libcurl iconv functions will be used. If HAVE_ICONV was
not defined when libcurl was built, and no callback has been established, conversion will return the CURLE_CONV_REQD error code.
If HAVE_ICONV is defined, CURL_ICONV_CODESET_OF_HOST must also be defined. For example:
#define CURL_ICONV_CODESET_OF_HOST "IBM-1047"
The iconv code in libcurl will default the network and UTF8 codeset names as follows:
#define CURL_ICONV_CODESET_OF_NETWORK "ISO8859-1"
#define CURL_ICONV_CODESET_FOR_UTF8 "UTF-8"
You will need to override these definitions if they are different on your system.
CURLOPT_INTERLEAVEFUNCTION
Pass a pointer to a function that matches the following prototype: size_t function( void *ptr, size_t size, size_t nmemb, void
*userdata). This function gets called by libcurl as soon as it has received interleaved RTP data. This function gets called for each
$ block and therefore contains exactly one upper-layer protocol unit (e.g. one RTP packet). Curl writes the interleaved header as
well as the included data for each call. The first byte is always an ASCII dollar sign. The dollar sign is followed by a one byte
channel identifier and then a 2 byte integer length in network byte order. See RFC2326 Section 10.12 for more information on how RTP
interleaving behaves. If unset or set to NULL, curl will use the default write function.
Interleaved RTP poses some challenges for the client application. Since the stream data is sharing the RTSP control connection, it
is critical to service the RTP in a timely fashion. If the RTP data is not handled quickly, subsequent response processing may
become unreasonably delayed and the connection may close. The application may use CURL_RTSPREQ_RECEIVE to service RTP data when no
requests are desired. If the application makes a request, (e.g. CURL_RTSPREQ_PAUSE) then the response handler will process any
pending RTP data before marking the request as finished. (Added in 7.20.0)
CURLOPT_INTERLEAVEDATA
This is the userdata pointer that will be passed to CURLOPT_INTERLEAVEFUNCTION when interleaved RTP data is received. (Added in
7.20.0)
CURLOPT_CHUNK_BGN_FUNCTION
Pass a pointer to a function that matches the following prototype: long function (const void *transfer_info, void *ptr, int
remains). This function gets called by libcurl before a part of the stream is going to be transferred (if the transfer supports
chunks).
This callback makes sense only when using the CURLOPT_WILDCARDMATCH option for now.
The target of transfer_info parameter is a "feature depended" structure. For the FTP wildcard download, the target is curl_fileinfo
structure (see curl/curl.h). The parameter ptr is a pointer given by CURLOPT_CHUNK_DATA. The parameter remains contains number of
chunks remaining per the transfer. If the feature is not available, the parameter has zero value.
Return CURL_CHUNK_BGN_FUNC_OK if everything is fine, CURL_CHUNK_BGN_FUNC_SKIP if you want to skip the concrete chunk or
CURL_CHUNK_BGN_FUNC_FAIL to tell libcurl to stop if some error occurred. (This was added in 7.21.0)
CURLOPT_CHUNK_END_FUNCTION
Pass a pointer to a function that matches the following prototype: long function(void *ptr). This function gets called by libcurl as
soon as a part of the stream has been transferred (or skipped).
Return CURL_CHUNK_END_FUNC_OK if everything is fine or CURL_CHUNK_END_FUNC_FAIL to tell the lib to stop if some error occurred.
(This was added in 7.21.0)
CURLOPT_CHUNK_DATA
Pass a pointer that will be untouched by libcurl and passed as the ptr argument to the CURL_CHUNK_BGN_FUNTION and
CURL_CHUNK_END_FUNTION. (This was added in 7.21.0)
CURLOPT_FNMATCH_FUNCTION
Pass a pointer to a function that matches the following prototype: int function(void *ptr, const char *pattern, const char *string)
prototype (see curl/curl.h). It is used internally for the wildcard matching feature.
Return CURL_FNMATCHFUNC_MATCH if pattern matches the string, CURL_FNMATCHFUNC_NOMATCH if not or CURL_FNMATCHFUNC_FAIL if an error
occurred. (This was added in 7.21.0)
CURLOPT_FNMATCH_DATA
Pass a pointer that will be untouched by libcurl and passed as the ptr argument to the CURL_FNMATCH_FUNCTION. (This was added in
7.21.0)
ERROR OPTIONS
CURLOPT_ERRORBUFFER
Pass a char * to a buffer that the libcurl may store human readable error messages in. This may be more helpful than just the return
code from curl_easy_perform. The buffer must be at least CURL_ERROR_SIZE big. Although this argument is a 'char *', it does not
describe an input string. Therefore the (probably undefined) contents of the buffer is NOT copied by the library. You must keep the
associated storage available until libcurl no longer needs it. Failing to do so will cause very odd behavior or even crashes.
libcurl will need it until you call curl_easy_cleanup(3) or you set the same option again to use a different pointer.
Use CURLOPT_VERBOSE and CURLOPT_DEBUGFUNCTION to better debug/trace why errors happen.
If the library does not return an error, the buffer may not have been touched. Do not rely on the contents in those cases.
CURLOPT_STDERR
Pass a FILE * as parameter. Tell libcurl to use this stream instead of stderr when showing the progress meter and displaying CUR-
LOPT_VERBOSE data.
CURLOPT_FAILONERROR
A parameter set to 1 tells the library to fail silently if the HTTP code returned is equal to or larger than 400. The default action
would be to return the page normally, ignoring that code.
This method is not fail-safe and there are occasions where non-successful response codes will slip through, especially when authen-
tication is involved (response codes 401 and 407).
You might get some amounts of headers transferred before this situation is detected, like when a "100-continue" is received as a
response to a POST/PUT and a 401 or 407 is received immediately afterwards.
NETWORK OPTIONS
CURLOPT_URL
Pass in a pointer to the actual URL to deal with. The parameter should be a char * to a zero terminated string which must be URL-
encoded in the following format:
scheme://host:port/path
For a greater explanation of the format please see RFC3986.
If the given URL lacks the scheme, or protocol, part ("http://" or "ftp://" etc), libcurl will attempt to resolve which protocol to
use based on the given host mame. If the protocol is not supported, libcurl will return (CURLE_UNSUPPORTED_PROTOCOL) when you call
curl_easy_perform(3) or curl_multi_perform(3). Use curl_version_info(3) for detailed information on which protocols are supported.
The host part of the URL contains the address of the server that you want to connect to. This can be the fully qualified domain name
of the server, the local network name of the machine on your network or the IP address of the server or machine represented by
either an IPv4 or IPv6 address. For example:
http://www.example.com/
http://hostname/
http://192.168.0.1/
http://[2001:1890:1112:1::20]/
It is also possible to specify the user name and password as part of the host, for some protocols, when connecting to servers that
require authentication.
For example the following types of authentication support this:
http://user:password@www.example.com
ftp://user:password@ftp.example.com
pop3://user:password@mail.example.com
The port is optional and when not specified libcurl will use the default port based on the determined or specified protocol: 80 for
HTTP, 21 for FTP and 25 for SMTP, etc. The following examples show how to specify the port:
http://www.example.com:8080/ - This will connect to a web server using port 8080 rather than 80.
smtp://mail.example.com:587/ - This will connect to a SMTP server on the alternative mail port.
The path part of the URL is protocol specific and whilst some examples are given below this list is not conclusive:
HTTP
The path part of a HTTP request specifies the file to retrieve and from what directory. If the directory is not specified then the
web server's root directory is used. If the file is omitted then the default document will be retrieved for either the directory
specified or the root directory. The exact resource returned for each URL is entirely dependent on the server's configuration.
http://www.example.com - This gets the main page from the web server.
http://www.example.com/index.html - This returns the main page by explicitly requesting it.
http://www.example.com/contactus/ - This returns the default document from the contactus directory.
FTP
The path part of an FTP request specifies the file to retrieve and from what directory. If the file part is omitted then libcurl
downloads the directory listing for the directory specified. If the directory is omitted then the directory listing for the root /
home directory will be returned.
ftp://ftp.example.com - This retrieves the directory listing for the root directory.
ftp://ftp.example.com/readme.txt - This downloads the file readme.txt from the root directory.
ftp://ftp.example.com/libcurl/readme.txt - This downloads readme.txt from the libcurl directory.
ftp://user:password@ftp.example.com/readme.txt - This retrieves the readme.txt file from the user's home directory. When a username
and password is specified, everything that is specified in the path part is relative to the user's home directory. To retrieve files
from the root directory or a directory underneath the root directory then the absolute path must be specified by prepending an addi-
tional forward slash to the beginning of the path.
ftp://user:password@ftp.example.com//readme.txt - This retrieves the readme.txt from the root directory when logging in as a speci-
fied user.
SMTP
The path part of a SMTP request specifies the host name to present during communication with the mail server. If the path is omitted
then libcurl will attempt to resolve the local computer's host name. However, this may not return the fully qualified domain name
that is required by some mail servers and specifying this path allows you to set an alternative name, such as your machine's fully
qualified domain name, which you might have obtained from an external function such as gethostname or getaddrinfo.
smtp://mail.example.com - This connects to the mail server at example.com and sends your local computer's host name in the HELO /
EHLO command.
smtp://mail.example.com/client.example.com - This will send client.example.com in the HELO / EHLO command to the mail server at
example.com.
POP3
The path part of a POP3 request specifies the mailbox (message) to retrieve. If the mailbox is not specified then a list of waiting
messages is returned instead.
pop3://user:password@mail.example.com - This lists the available messages pop3://user:password@mail.example.com/1 - This retrieves
the first message
SCP
The path part of a SCP request specifies the file to retrieve and from what directory. The file part may not be omitted. The file is
taken as an absolute path from the root directory on the server. To specify a path relative to the user's home directory on the
server, prepend ~/ to the path portion. If the user name is not embedded in the URL, it can be set with the CURLOPT_USERPWD or CUR-
LOPT_USERNAME option.
scp://user@example.com/etc/issue - This specifies the file /etc/issue
scp://example.com/~/my-file - This specifies the file my-file in the user's home directory on the server
SFTP
The path part of a SFTP request specifies the file to retrieve and from what directory. If the file part is omitted then libcurl
downloads the directory listing for the directory specified. If the path ends in a / then a directory listing is returned instead
of a file. If the path is omitted entirely then the directory listing for the root / home directory will be returned. If the user
name is not embedded in the URL, it can be set with the CURLOPT_USERPWD or CURLOPT_USERNAME option.
sftp://user:password@example.com/etc/issue - This specifies the file /etc/issue
sftp://user@example.com/~/my-file - This specifies the file my-file in the user's home directory
sftp://ssh.example.com/~/Documents/ - This requests a directory listing of the Documents directory under the user's home directory
LDAP
The path part of a LDAP request can be used to specify the: Distinguished Name, Attributes, Scope, Filter and Extension for a LDAP
search. Each field is separated by a question mark and when that field is not required an empty string with the question mark sepa-
rator should be included.
ldap://ldap.example.com/o=My%20Organisation - This will perform a LDAP search with the DN as My Organisation.
ldap://ldap.example.com/o=My%20Organisation?postalAddress - This will perform the same search but will only return postalAddress
attributes.
ldap://ldap.example.com/?rootDomainNamingContext - This specifies an empty DN and requests information about the rootDomainNaming-
Context attribute for an Active Directory server.
For more information about the individual components of a LDAP URL please see RFC4516.
NOTES
Starting with version 7.20.0, the fragment part of the URI will not be sent as part of the path, which was previously the case.
CURLOPT_URL is the only option that must be set before curl_easy_perform(3) is called.
CURLOPT_PROTOCOLS can be used to limit what protocols libcurl will use for this transfer, independent of what libcurl has been com-
piled to support. That may be useful if you accept the URL from an external source and want to limit the accessibility.
CURLOPT_PROTOCOLS
Pass a long that holds a bitmask of CURLPROTO_* defines. If used, this bitmask limits what protocols libcurl may use in the trans-
fer. This allows you to have a libcurl built to support a wide range of protocols but still limit specific transfers to only be
allowed to use a subset of them. By default libcurl will accept all protocols it supports. See also CURLOPT_REDIR_PROTOCOLS. (Added
in 7.19.4)
CURLOPT_REDIR_PROTOCOLS
Pass a long that holds a bitmask of CURLPROTO_* defines. If used, this bitmask limits what protocols libcurl may use in a transfer
that it follows to in a redirect when CURLOPT_FOLLOWLOCATION is enabled. This allows you to limit specific transfers to only be
allowed to use a subset of protocols in redirections. By default libcurl will allow all protocols except for FILE and SCP. This is a
difference compared to pre-7.19.4 versions which unconditionally would follow to all protocols supported. (Added in 7.19.4)
CURLOPT_PROXY
Set HTTP proxy to use. The parameter should be a char * to a zero terminated string holding the host name or dotted IP address. To
specify port number in this string, append :[port] to the end of the host name. The proxy string may be prefixed with [protocol]://
since any such prefix will be ignored. The proxy's port number may optionally be specified with the separate option. If not speci-
fied, libcurl will default to using port 1080 for proxies. CURLOPT_PROXYPORT.
When you tell the library to use a HTTP proxy, libcurl will transparently convert operations to HTTP even if you specify an FTP URL
etc. This may have an impact on what other features of the library you can use, such as CURLOPT_QUOTE and similar FTP specifics that
don't work unless you tunnel through the HTTP proxy. Such tunneling is activated with CURLOPT_HTTPPROXYTUNNEL.
libcurl respects the environment variables http_proxy, ftp_proxy, all_proxy etc, if any of those are set. The CURLOPT_PROXY option
does however override any possibly set environment variables.
Setting the proxy string to "" (an empty string) will explicitly disable the use of a proxy, even if there is an environment vari-
able set for it.
Since 7.14.1, the proxy host string given in environment variables can be specified the exact same way as the proxy can be set with
CURLOPT_PROXY, include protocol prefix (http://) and embedded user + password.
Since 7.21.7, the proxy string may be specified with a protocol:// prefix to specify alternative proxy protocols. Use socks4://,
socks4a://, socks5:// or socks5h:// (the last one to enable socks5 and asking the proxy to do the resolving, also known as CURL-
PROXY_SOCKS5_HOSTNAME type) to request the specific SOCKS version to be used. No protocol specified, http:// and all others will be
treated as HTTP proxies.
CURLOPT_PROXYPORT
Pass a long with this option to set the proxy port to connect to unless it is specified in the proxy string CURLOPT_PROXY.
CURLOPT_PROXYTYPE
Pass a long with this option to set type of the proxy. Available options for this are CURLPROXY_HTTP, CURLPROXY_HTTP_1_0 (added in
7.19.4), CURLPROXY_SOCKS4 (added in 7.10), CURLPROXY_SOCKS5, CURLPROXY_SOCKS4A (added in 7.18.0) and CURLPROXY_SOCKS5_HOSTNAME
(added in 7.18.0). The HTTP type is default. (Added in 7.10)
If you set CURLOPT_PROXYTYPE to CURLPROXY_HTTP_1_0, it will only affect how libcurl speaks to a proxy when CONNECT is used. The HTTP
version used for "regular" HTTP requests is instead controlled with CURLOPT_HTTP_VERSION.
CURLOPT_NOPROXY
Pass a pointer to a zero terminated string. The string consists of a comma separated list of host names that do not require a proxy
to get reached, even if one is specified. The only wildcard available is a single * character, which matches all hosts, and effec-
tively disables the proxy. Each name in this list is matched as either a domain which contains the hostname, or the hostname itself.
For example, example.com would match example.com, example.com:80, and www.example.com, but not www.notanexample.com. (Added in
7.19.4)
CURLOPT_HTTPPROXYTUNNEL
Set the parameter to 1 to make the library tunnel all operations through a given HTTP proxy. There is a big difference between using
a proxy and to tunnel through it. If you don't know what this means, you probably don't want this tunneling option.
CURLOPT_SOCKS5_GSSAPI_SERVICE
Pass a char * as parameter to a string holding the name of the service. The default service name for a SOCKS5 server is rcmd/server-
fqdn. This option allows you to change it. (Added in 7.19.4)
CURLOPT_SOCKS5_GSSAPI_NEC
Pass a long set to 1 to enable or 0 to disable. As part of the gssapi negotiation a protection mode is negotiated. The RFC1961 says
in section 4.3/4.4 it should be protected, but the NEC reference implementation does not. If enabled, this option allows the unpro-
tected exchange of the protection mode negotiation. (Added in 7.19.4).
CURLOPT_INTERFACE
Pass a char * as parameter. This sets the interface name to use as outgoing network interface. The name can be an interface name, an
IP address, or a host name.
Starting with 7.24.0: If the parameter starts with "if!" then it is treated as only as interface name and no attempt will ever be
named to do treat it as an IP address or to do name resolution on it. If the parameter starts with "host!" it is treated as either
an IP address or a hostname. Hostnames are resolved synchronously. Using the if! format is highly recommended when using the multi
interfaces to avoid allowing the code to block. If "if!" is specified but the parameter does not match an existing interface,
CURLE_INTERFACE_FAILED is returned.
CURLOPT_LOCALPORT
Pass a long. This sets the local port number of the socket used for connection. This can be used in combination with CURLOPT_INTER-
FACE and you are recommended to use CURLOPT_LOCALPORTRANGE as well when this is set. Valid port numbers are 1 - 65535. (Added in
7.15.2)
CURLOPT_LOCALPORTRANGE
Pass a long. This is the number of attempts libcurl will make to find a working local port number. It starts with the given CUR-
LOPT_LOCALPORT and adds one to the number for each retry. Setting this to 1 or below will make libcurl do only one try for the exact
port number. Port numbers by nature are scarce resources that will be busy at times so setting this value to something too low might
cause unnecessary connection setup failures. (Added in 7.15.2)
CURLOPT_DNS_CACHE_TIMEOUT
Pass a long, this sets the timeout in seconds. Name resolves will be kept in memory for this number of seconds. Set to zero to com-
pletely disable caching, or set to -1 to make the cached entries remain forever. By default, libcurl caches this info for 60 sec-
onds.
The name resolve functions of various libc implementations don't re-read name server information unless explicitly told so (for
example, by calling res_init(3)). This may cause libcurl to keep using the older server even if DHCP has updated the server info,
and this may look like a DNS cache issue to the casual libcurl-app user.
CURLOPT_DNS_USE_GLOBAL_CACHE
Pass a long. If the value is 1, it tells curl to use a global DNS cache that will survive between easy handle creations and dele-
tions. This is not thread-safe and this will use a global variable.
WARNING: this option is considered obsolete. Stop using it. Switch over to using the share interface instead! See CURLOPT_SHARE and
curl_share_init(3).
CURLOPT_BUFFERSIZE
Pass a long specifying your preferred size (in bytes) for the receive buffer in libcurl. The main point of this would be that the
write callback gets called more often and with smaller chunks. This is just treated as a request, not an order. You cannot be guar-
anteed to actually get the given size. (Added in 7.10)
This size is by default set as big as possible (CURL_MAX_WRITE_SIZE), so it only makes sense to use this option if you want it
smaller.
CURLOPT_PORT
Pass a long specifying what remote port number to connect to, instead of the one specified in the URL or the default port for the
used protocol.
CURLOPT_TCP_NODELAY
Pass a long specifying whether the TCP_NODELAY option is to be set or cleared (1 = set, 0 = clear). The option is cleared by
default. This will have no effect after the connection has been established.
Setting this option will disable TCP's Nagle algorithm. The purpose of this algorithm is to try to minimize the number of small
packets on the network (where "small packets" means TCP segments less than the Maximum Segment Size (MSS) for the network).
Maximizing the amount of data sent per TCP segment is good because it amortizes the overhead of the send. However, in some cases
(most notably telnet or rlogin) small segments may need to be sent without delay. This is less efficient than sending larger amounts
of data at a time, and can contribute to congestion on the network if overdone.
CURLOPT_ADDRESS_SCOPE
Pass a long specifying the scope_id value to use when connecting to IPv6 link-local or site-local addresses. (Added in 7.19.0)
CURLOPT_TCP_KEEPALIVE
Pass a long. If set to 1, TCP keepalive probes will be sent. The delay and frequency of these probes can be controlled by the CUR-
LOPT_TCP_KEEPIDLE and CURLOPT_TCP_KEEPINTVL options, provided the operating system supports them. Set to 0 (default behavior) to
disable keepalive probes (Added in 7.25.0).
CURLOPT_TCP_KEEPIDLE
Pass a long. Sets the delay, in seconds, that the operating system will wait while the connection is idle before sending keepalive
probes. Not all operating systems support this option. (Added in 7.25.0)
CURLOPT_TCP_KEEPINTVL
Pass a long. Sets the interval, in seconds, that the operating system will wait between sending keepalive probes. Not all operating
systems support this option. (Added in 7.25.0)
NAMES and PASSWORDS OPTIONS (Authentication)
CURLOPT_NETRC
This parameter controls the preference of libcurl between using user names and passwords from your ~/.netrc file, relative to user
names and passwords in the URL supplied with CURLOPT_URL.
libcurl uses a user name (and supplied or prompted password) supplied with CURLOPT_USERPWD in preference to any of the options con-
trolled by this parameter.
Pass a long, set to one of the values described below.
CURL_NETRC_OPTIONAL
The use of your ~/.netrc file is optional, and information in the URL is to be preferred. The file will be scanned for the
host and user name (to find the password only) or for the host only, to find the first user name and password after that
machine, which ever information is not specified in the URL.
Undefined values of the option will have this effect.
CURL_NETRC_IGNORED
The library will ignore the file and use only the information in the URL.
This is the default.
CURL_NETRC_REQUIRED
This value tells the library that use of the file is required, to ignore the information in the URL, and to search the file
for the host only.
Only machine name, user name and password are taken into account (init macros and similar things aren't supported).
libcurl does not verify that the file has the correct properties set (as the standard Unix ftp client does). It should only be readable by
user.
CURLOPT_NETRC_FILE
Pass a char * as parameter, pointing to a zero terminated string containing the full path name to the file you want libcurl to use
as .netrc file. If this option is omitted, and CURLOPT_NETRC is set, libcurl will attempt to find a .netrc file in the current
user's home directory. (Added in 7.10.9)
CURLOPT_USERPWD
Pass a char * as parameter, which should be [user name]:[password] to use for the connection. Use CURLOPT_HTTPAUTH to decide the
authentication method.
When using NTLM, you can set the domain by prepending it to the user name and separating the domain and name with a forward (/) or
backward slash (). Like this: "domain/user:password" or "domainuser:password". Some HTTP servers (on Windows) support this style
even for Basic authentication.
When using HTTP and CURLOPT_FOLLOWLOCATION, libcurl might perform several requests to possibly different hosts. libcurl will only
send this user and password information to hosts using the initial host name (unless CURLOPT_UNRESTRICTED_AUTH is set), so if
libcurl follows locations to other hosts it will not send the user and password to those. This is enforced to prevent accidental
information leakage.
CURLOPT_PROXYUSERPWD
Pass a char * as parameter, which should be [user name]:[password] to use for the connection to the HTTP proxy. Use CURLOPT_PROX-
YAUTH to decide the authentication method.
CURLOPT_USERNAME
Pass a char * as parameter, which should be pointing to the zero terminated user name to use for the transfer.
CURLOPT_USERNAME sets the user name to be used in protocol authentication. You should not use this option together with the (older)
CURLOPT_USERPWD option.
In order to specify the password to be used in conjunction with the user name use the CURLOPT_PASSWORD option. (Added in 7.19.1)
CURLOPT_PASSWORD
Pass a char * as parameter, which should be pointing to the zero terminated password to use for the transfer.
The CURLOPT_PASSWORD option should be used in conjunction with the CURLOPT_USERNAME option. (Added in 7.19.1)
CURLOPT_PROXYUSERNAME
Pass a char * as parameter, which should be pointing to the zero terminated user name to use for the transfer while connecting to
Proxy.
The CURLOPT_PROXYUSERNAME option should be used in same way as the CURLOPT_PROXYUSERPWD is used. In comparison to CURLOPT_PROX-
YUSERPWD the CURLOPT_PROXYUSERNAME allows the username to contain a colon, like in the following example: "sip:user@example.com".
The CURLOPT_PROXYUSERNAME option is an alternative way to set the user name while connecting to Proxy. There is no meaning to use
it together with the CURLOPT_PROXYUSERPWD option.
In order to specify the password to be used in conjunction with the user name use the CURLOPT_PROXYPASSWORD option. (Added in
7.19.1)
CURLOPT_PROXYPASSWORD
Pass a char * as parameter, which should be pointing to the zero terminated password to use for the transfer while connecting to
Proxy.
The CURLOPT_PROXYPASSWORD option should be used in conjunction with the CURLOPT_PROXYUSERNAME option. (Added in 7.19.1)
CURLOPT_HTTPAUTH
Pass a long as parameter, which is set to a bitmask, to tell libcurl which authentication method(s) you want it to use. The avail-
able bits are listed below. If more than one bit is set, libcurl will first query the site to see which authentication methods it
supports and then pick the best one you allow it to use. For some methods, this will induce an extra network round-trip. Set the
actual name and password with the CURLOPT_USERPWD option or with the CURLOPT_USERNAME and the CURLOPT_PASSWORD options. (Added in
7.10.6)
CURLAUTH_BASIC
HTTP Basic authentication. This is the default choice, and the only method that is in wide-spread use and supported virtually
everywhere. This sends the user name and password over the network in plain text, easily captured by others.
CURLAUTH_DIGEST
HTTP Digest authentication. Digest authentication is defined in RFC2617 and is a more secure way to do authentication over
public networks than the regular old-fashioned Basic method.
CURLAUTH_DIGEST_IE
HTTP Digest authentication with an IE flavor. Digest authentication is defined in RFC2617 and is a more secure way to do
authentication over public networks than the regular old-fashioned Basic method. The IE flavor is simply that libcurl will
use a special "quirk" that IE is known to have used before version 7 and that some servers require the client to use. (This
define was added in 7.19.3)
CURLAUTH_GSSNEGOTIATE
HTTP GSS-Negotiate authentication. The GSS-Negotiate (also known as plain "Negotiate") method was designed by Microsoft and
is used in their web applications. It is primarily meant as a support for Kerberos5 authentication but may also be used along
with other authentication methods. For more information see IETF draft draft-brezak-spnego-http-04.txt.
You need to build libcurl with a suitable GSS-API library for this to work.
CURLAUTH_NTLM
HTTP NTLM authentication. A proprietary protocol invented and used by Microsoft. It uses a challenge-response and hash con-
cept similar to Digest, to prevent the password from being eavesdropped.
You need to build libcurl with either OpenSSL, GnuTLS or NSS support for this option to work, or build libcurl on Windows.
CURLAUTH_NTLM_WB
NTLM delegating to winbind helper. Authentication is performed by a separate binary application that is executed when needed.
The name of the application is specified at compile time but is typically /usr/bin/ntlm_auth (Added in 7.22.0)
Note that libcurl will fork when necessary to run the winbind application and kill it when complete, calling waitpid() to
await its exit when done. On POSIX operating systems, killing the process will cause a SIGCHLD signal to be raised (regard-
less of whether CURLOPT_NOSIGNAL is set), which must be handled intelligently by the application. In particular, the applica-
tion must not unconditionally call wait() in its SIGCHLD signal handler to avoid being subject to a race condition. This
behavior is subject to change in future versions of libcurl.
CURLAUTH_ANY
This is a convenience macro that sets all bits and thus makes libcurl pick any it finds suitable. libcurl will automatically
select the one it finds most secure.
CURLAUTH_ANYSAFE
This is a convenience macro that sets all bits except Basic and thus makes libcurl pick any it finds suitable. libcurl will
automatically select the one it finds most secure.
CURLAUTH_ONLY
This is a meta symbol. Or this value together with a single specific auth value to force libcurl to probe for un-restricted
auth and if not, only that single auth algorithm is acceptable. (Added in 7.21.3)
CURLOPT_TLSAUTH_TYPE
Pass a long as parameter, which is set to a bitmask, to tell libcurl which authentication method(s) you want it to use for TLS
authentication.
CURLOPT_TLSAUTH_SRP
TLS-SRP authentication. Secure Remote Password authentication for TLS is defined in RFC5054 and provides mutual authentica-
tion if both sides have a shared secret. To use TLS-SRP, you must also set the CURLOPT_TLSAUTH_USERNAME and CUR-
LOPT_TLSAUTH_PASSWORD options.
You need to build libcurl with GnuTLS or OpenSSL with TLS-SRP support for this to work. (Added in 7.21.4)
CURLOPT_TLSAUTH_USERNAME
Pass a char * as parameter, which should point to the zero terminated username to use for the TLS authentication method specified
with the CURLOPT_TLSAUTH_TYPE option. Requires that the CURLOPT_TLS_PASSWORD option also be set. (Added in 7.21.4)
CURLOPT_TLSAUTH_PASSWORD
Pass a char * as parameter, which should point to the zero terminated password to use for the TLS authentication method specified
with the CURLOPT_TLSAUTH_TYPE option. Requires that the CURLOPT_TLS_USERNAME option also be set. (Added in 7.21.4)
CURLOPT_PROXYAUTH
Pass a long as parameter, which is set to a bitmask, to tell libcurl which authentication method(s) you want it to use for your
proxy authentication. If more than one bit is set, libcurl will first query the site to see what authentication methods it supports
and then pick the best one you allow it to use. For some methods, this will induce an extra network round-trip. Set the actual name
and password with the CURLOPT_PROXYUSERPWD option. The bitmask can be constructed by or'ing together the bits listed above for the
CURLOPT_HTTPAUTH option. As of this writing, only Basic, Digest and NTLM work. (Added in 7.10.7)
HTTP OPTIONS
CURLOPT_AUTOREFERER
Pass a parameter set to 1 to enable this. When enabled, libcurl will automatically set the Referer: field in requests where it fol-
lows a Location: redirect.
CURLOPT_ACCEPT_ENCODING
Sets the contents of the Accept-Encoding: header sent in a HTTP request, and enables decoding of a response when a Content-Encoding:
header is received. Three encodings are supported: identity, which does nothing, deflate which requests the server to compress its
response using the zlib algorithm, and gzip which requests the gzip algorithm. If a zero-length string is set, then an Accept-
Encoding: header containing all supported encodings is sent.
This is a request, not an order; the server may or may not do it. This option must be set (to any non-NULL value) or else any unso-
licited encoding done by the server is ignored. See the special file lib/README.encoding for details.
(This option was called CURLOPT_ENCODING before 7.21.6)
CURLOPT_TRANSFER_ENCODING
Adds a request for compressed Transfer Encoding in the outgoing HTTP request. If the server supports this and so desires, it can
respond with the HTTP response sent using a compressed Transfer-Encoding that will be automatically uncompressed by libcurl on
reception.
Transfer-Encoding differs slightly from the Content-Encoding you ask for with CURLOPT_ACCEPT_ENCODING in that a Transfer-Encoding is
strictly meant to be for the transfer and thus MUST be decoded before the data arrives in the client. Traditionally, Transfer-Encod-
ing has been much less used and supported by both HTTP clients and HTTP servers.
(Added in 7.21.6)
CURLOPT_FOLLOWLOCATION
A parameter set to 1 tells the library to follow any Location: header that the server sends as part of a HTTP header.
This means that the library will re-send the same request on the new location and follow new Location: headers all the way until no
more such headers are returned. CURLOPT_MAXREDIRS can be used to limit the number of redirects libcurl will follow.
Since 7.19.4, libcurl can limit what protocols it will automatically follow. The accepted protocols are set with CURLOPT_REDIR_PRO-
TOCOLS and it excludes the FILE protocol by default.
CURLOPT_UNRESTRICTED_AUTH
A parameter set to 1 tells the library it can continue to send authentication (user+password) when following locations, even when
hostname changed. This option is meaningful only when setting CURLOPT_FOLLOWLOCATION.
CURLOPT_MAXREDIRS
Pass a long. The set number will be the redirection limit. If that many redirections have been followed, the next redirect will
cause an error (CURLE_TOO_MANY_REDIRECTS). This option only makes sense if the CURLOPT_FOLLOWLOCATION is used at the same time.
Added in 7.15.1: Setting the limit to 0 will make libcurl refuse any redirect. Set it to -1 for an infinite number of redirects
(which is the default)
CURLOPT_POSTREDIR
Pass a bitmask to control how libcurl acts on redirects after POSTs that get a 301, 302 or 303 response back. A parameter with bit
0 set (value CURL_REDIR_POST_301) tells the library to respect RFC2616/10.3.2 and not convert POST requests into GET requests when
following a 301 redirection. Setting bit 1 (value CURL_REDIR_POST_302) makes libcurl maintain the request method after a 302 redi-
rect whilst setting bit 2 (value CURL_REDIR_POST_303) makes libcurl maintain the request method after a 303 redirect. The value
CURL_REDIR_POST_ALL is a convenience define that sets all three bits.
The non-RFC behaviour is ubiquitous in web browsers, so the library does the conversion by default to maintain consistency. However,
a server may require a POST to remain a POST after such a redirection. This option is meaningful only when setting CURLOPT_FOLLOWLO-
CATION. (Added in 7.17.1) (This option was known as CURLOPT_POST301 up to 7.19.0 as it only supported the 301 then)
CURLOPT_PUT
A parameter set to 1 tells the library to use HTTP PUT to transfer data. The data should be set with CURLOPT_READDATA and CUR-
LOPT_INFILESIZE.
This option is deprecated and starting with version 7.12.1 you should instead use CURLOPT_UPLOAD.
CURLOPT_POST
A parameter set to 1 tells the library to do a regular HTTP post. This will also make the library use a "Content-Type: applica-
tion/x-www-form-urlencoded" header. (This is by far the most commonly used POST method).
Use one of CURLOPT_POSTFIELDS or CURLOPT_COPYPOSTFIELDS options to specify what data to post and CURLOPT_POSTFIELDSIZE or CUR-
LOPT_POSTFIELDSIZE_LARGE to set the data size.
Optionally, you can provide data to POST using the CURLOPT_READFUNCTION and CURLOPT_READDATA options but then you must make sure to
not set CURLOPT_POSTFIELDS to anything but NULL. When providing data with a callback, you must transmit it using chunked transfer-
encoding or you must set the size of the data with the CURLOPT_POSTFIELDSIZE or CURLOPT_POSTFIELDSIZE_LARGE option. To enable chun-
ked encoding, you simply pass in the appropriate Transfer-Encoding header, see the post-callback.c example.
You can override the default POST Content-Type: header by setting your own with CURLOPT_HTTPHEADER.
Using POST with HTTP 1.1 implies the use of a "Expect: 100-continue" header. You can disable this header with CURLOPT_HTTPHEADER as
usual.
If you use POST to a HTTP 1.1 server, you can send data without knowing the size before starting the POST if you use chunked encod-
ing. You enable this by adding a header like "Transfer-Encoding: chunked" with CURLOPT_HTTPHEADER. With HTTP 1.0 or without chunked
transfer, you must specify the size in the request.
When setting CURLOPT_POST to 1, it will automatically set CURLOPT_NOBODY to 0 (since 7.14.1).
If you issue a POST request and then want to make a HEAD or GET using the same re-used handle, you must explicitly set the new
request type using CURLOPT_NOBODY or CURLOPT_HTTPGET or similar.
CURLOPT_POSTFIELDS
Pass a void * as parameter, which should be the full data to post in a HTTP POST operation. You must make sure that the data is for-
matted the way you want the server to receive it. libcurl will not convert or encode it for you. Most web servers will assume this
data to be url-encoded.
The pointed data are NOT copied by the library: as a consequence, they must be preserved by the calling application until the trans-
fer finishes.
This POST is a normal application/x-www-form-urlencoded kind (and libcurl will set that Content-Type by default when this option is
used), which is the most commonly used one by HTML forms. See also the CURLOPT_POST. Using CURLOPT_POSTFIELDS implies CURLOPT_POST.
If you want to do a zero-byte POST, you need to set CURLOPT_POSTFIELDSIZE explicitly to zero, as simply setting CURLOPT_POSTFIELDS
to NULL or "" just effectively disables the sending of the specified string. libcurl will instead assume that you'll send the POST
data using the read callback!
Using POST with HTTP 1.1 implies the use of a "Expect: 100-continue" header. You can disable this header with CURLOPT_HTTPHEADER as
usual.
To make multipart/formdata posts (aka RFC2388-posts), check out the CURLOPT_HTTPPOST option.
CURLOPT_POSTFIELDSIZE
If you want to post data to the server without letting libcurl do a strlen() to measure the data size, this option must be used.
When this option is used you can post fully binary data, which otherwise is likely to fail. If this size is set to -1, the library
will use strlen() to get the size.
CURLOPT_POSTFIELDSIZE_LARGE
Pass a curl_off_t as parameter. Use this to set the size of the CURLOPT_POSTFIELDS data to prevent libcurl from doing strlen() on
the data to figure out the size. This is the large file version of the CURLOPT_POSTFIELDSIZE option. (Added in 7.11.1)
CURLOPT_COPYPOSTFIELDS
Pass a char * as parameter, which should be the full data to post in a HTTP POST operation. It behaves as the CURLOPT_POSTFIELDS
option, but the original data are copied by the library, allowing the application to overwrite the original data after setting this
option.
Because data are copied, care must be taken when using this option in conjunction with CURLOPT_POSTFIELDSIZE or CURLOPT_POSTFIELD-
SIZE_LARGE: If the size has not been set prior to CURLOPT_COPYPOSTFIELDS, the data are assumed to be a NUL-terminated string; else
the stored size informs the library about the data byte count to copy. In any case, the size must not be changed after CURLOPT_COPY-
POSTFIELDS, unless another CURLOPT_POSTFIELDS or CURLOPT_COPYPOSTFIELDS option is issued. (Added in 7.17.1)
CURLOPT_HTTPPOST
Tells libcurl you want a multipart/formdata HTTP POST to be made and you instruct what data to pass on to the server. Pass a
pointer to a linked list of curl_httppost structs as parameter. The easiest way to create such a list, is to use curl_formadd(3) as
documented. The data in this list must remain intact until you close this curl handle again with curl_easy_cleanup(3).
Using POST with HTTP 1.1 implies the use of a "Expect: 100-continue" header. You can disable this header with CURLOPT_HTTPHEADER as
usual.
When setting CURLOPT_HTTPPOST, it will automatically set CURLOPT_NOBODY to 0 (since 7.14.1).
CURLOPT_REFERER
Pass a pointer to a zero terminated string as parameter. It will be used to set the Referer: header in the http request sent to the
remote server. This can be used to fool servers or scripts. You can also set any custom header with CURLOPT_HTTPHEADER.
CURLOPT_USERAGENT
Pass a pointer to a zero terminated string as parameter. It will be used to set the User-Agent: header in the http request sent to
the remote server. This can be used to fool servers or scripts. You can also set any custom header with CURLOPT_HTTPHEADER.
CURLOPT_HTTPHEADER
Pass a pointer to a linked list of HTTP headers to pass to the server in your HTTP request. The linked list should be a fully valid
list of struct curl_slist structs properly filled in. Use curl_slist_append(3) to create the list and curl_slist_free_all(3) to
clean up an entire list. If you add a header that is otherwise generated and used by libcurl internally, your added one will be used
instead. If you add a header with no content as in 'Accept:' (no data on the right side of the colon), the internally used header
will get disabled. Thus, using this option you can add new headers, replace internal headers and remove internal headers. To add a
header with no content, make the content be two quotes: "". The headers included in the linked list must not be CRLF-terminated,
because curl adds CRLF after each header item. Failure to comply with this will result in strange bugs because the server will most
likely ignore part of the headers you specified.
The first line in a request (containing the method, usually a GET or POST) is not a header and cannot be replaced using this option.
Only the lines following the request-line are headers. Adding this method line in this list of headers will only cause your request
to send an invalid header.
Pass a NULL to this to reset back to no custom headers.
The most commonly replaced headers have "shortcuts" in the options CURLOPT_COOKIE, CURLOPT_USERAGENT and CURLOPT_REFERER.
CURLOPT_HTTP200ALIASES
Pass a pointer to a linked list of aliases to be treated as valid HTTP 200 responses. Some servers respond with a custom header
response line. For example, IceCast servers respond with "ICY 200 OK". By including this string in your list of aliases, the
response will be treated as a valid HTTP header line such as "HTTP/1.0 200 OK". (Added in 7.10.3)
The linked list should be a fully valid list of struct curl_slist structs, and be properly filled in. Use curl_slist_append(3) to
create the list and curl_slist_free_all(3) to clean up an entire list.
The alias itself is not parsed for any version strings. Before libcurl 7.16.3, Libcurl used the value set by option CUR-
LOPT_HTTP_VERSION, but starting with 7.16.3 the protocol is assumed to match HTTP 1.0 when an alias matched.
CURLOPT_COOKIE
Pass a pointer to a zero terminated string as parameter. It will be used to set a cookie in the http request. The format of the
string should be NAME=CONTENTS, where NAME is the cookie name and CONTENTS is what the cookie should contain.
If you need to set multiple cookies, you need to set them all using a single option and thus you need to concatenate them all in one
single string. Set multiple cookies in one string like this: "name1=content1; name2=content2;" etc.
This option sets the cookie header explicitly in the outgoing request(s). If multiple requests are done due to authentication, fol-
lowed redirections or similar, they will all get this cookie passed on.
Using this option multiple times will only make the latest string override the previous ones.
CURLOPT_COOKIEFILE
Pass a pointer to a zero terminated string as parameter. It should contain the name of your file holding cookie data to read. The
cookie data may be in Netscape / Mozilla cookie data format or just regular HTTP-style headers dumped to a file.
Given an empty or non-existing file or by passing the empty string (""), this option will enable cookies for this curl handle, mak-
ing it understand and parse received cookies and then use matching cookies in future requests.
If you use this option multiple times, you just add more files to read. Subsequent files will add more cookies.
CURLOPT_COOKIEJAR
Pass a file name as char *, zero terminated. This will make libcurl write all internally known cookies to the specified file when
curl_easy_cleanup(3) is called. If no cookies are known, no file will be created. Specify "-" to instead have the cookies written to
stdout. Using this option also enables cookies for this session, so if you for example follow a location it will make matching cook-
ies get sent accordingly.
If the cookie jar file can't be created or written to (when the curl_easy_cleanup(3) is called), libcurl will not and cannot report
an error for this. Using CURLOPT_VERBOSE or CURLOPT_DEBUGFUNCTION will get a warning to display, but that is the only visible feed-
back you get about this possibly lethal situation.
CURLOPT_COOKIESESSION
Pass a long set to 1 to mark this as a new cookie "session". It will force libcurl to ignore all cookies it is about to load that
are "session cookies" from the previous session. By default, libcurl always stores and loads all cookies, independent if they are
session cookies or not. Session cookies are cookies without expiry date and they are meant to be alive and existing for this "ses-
sion" only.
CURLOPT_COOKIELIST
Pass a char * to a cookie string. Cookie can be either in Netscape / Mozilla format or just regular HTTP-style header (Set-Cookie:
...) format. If cURL cookie engine was not enabled it will enable its cookie engine. Passing a magic string "ALL" will erase all
cookies known by cURL. (Added in 7.14.1) Passing the special string "SESS" will only erase all session cookies known by cURL. (Added
in 7.15.4) Passing the special string "FLUSH" will write all cookies known by cURL to the file specified by CURLOPT_COOKIEJAR.
(Added in 7.17.1)
CURLOPT_HTTPGET
Pass a long. If the long is 1, this forces the HTTP request to get back to GET. Usable if a POST, HEAD, PUT, or a custom request has
been used previously using the same curl handle.
When setting CURLOPT_HTTPGET to 1, it will automatically set CURLOPT_NOBODY to 0 (since 7.14.1).
CURLOPT_HTTP_VERSION
Pass a long, set to one of the values described below. They force libcurl to use the specific HTTP versions. This is not sensible to
do unless you have a good reason.
CURL_HTTP_VERSION_NONE
We don't care about what version the library uses. libcurl will use whatever it thinks fit.
CURL_HTTP_VERSION_1_0
Enforce HTTP 1.0 requests.
CURL_HTTP_VERSION_1_1
Enforce HTTP 1.1 requests.
CURLOPT_IGNORE_CONTENT_LENGTH
Ignore the Content-Length header. This is useful for Apache 1.x (and similar servers) which will report incorrect content length for
files over 2 gigabytes. If this option is used, curl will not be able to accurately report progress, and will simply stop the down-
load when the server ends the connection. (added in 7.14.1)
CURLOPT_HTTP_CONTENT_DECODING
Pass a long to tell libcurl how to act on content decoding. If set to zero, content decoding will be disabled. If set to 1 it is
enabled. Libcurl has no default content decoding but requires you to use CURLOPT_ENCODING for that. (added in 7.16.2)
CURLOPT_HTTP_TRANSFER_DECODING
Pass a long to tell libcurl how to act on transfer decoding. If set to zero, transfer decoding will be disabled, if set to 1 it is
enabled (default). libcurl does chunked transfer decoding by default unless this option is set to zero. (added in 7.16.2)
SMTP OPTIONS
CURLOPT_MAIL_FROM
Pass a pointer to a zero terminated string as parameter. This should be used to specify the sender's email address when sending SMTP
mail with libcurl.
An originator email address should be specified with angled brackets (<>) around it, which if not specified, will be added by
libcurl from version 7.21.4 onwards. Failing to provide such brackets may cause the server to reject the email.
If this parameter is not specified then an empty address will be sent to the mail server which may or may not cause the email to be
rejected.
(Added in 7.20.0)
CURLOPT_MAIL_RCPT
Pass a pointer to a linked list of recipients to pass to the server in your SMTP mail request. The linked list should be a fully
valid list of struct curl_slist structs properly filled in. Use curl_slist_append(3) to create the list and curl_slist_free_all(3)
to clean up an entire list.
Each recipient should be specified within a pair of angled brackets (<>), however, should you not use an angled bracket as the first
character libcurl will assume you provided a single email address and enclose that address within brackets for you.
(Added in 7.20.0)
CURLOPT_MAIL_AUTH
Pass a pointer to a zero terminated string as parameter. This will be used to specify the authentication address (identity) of a
submitted message that is being relayed to another server.
This optional parameter allows co-operating agents in a trusted environment to communicate the authentication of individual messages
and should only be used by the application program, using libcurl, if the application is itself a mail server acting in such an
environment. If the application is operating as such and the AUTH address is not known or is invalid, then an empty string should be
used for this parameter.
Unlike CURLOPT_MAIL_FROM and CURLOPT_MAIL_RCPT, the address should not be specified within a pair of angled brackets (<>). However,
if an empty string is used then a pair of brackets will be sent by libcurl as required by RFC2554.
(Added in 7.25.0)
TFTP OPTIONS
CURLOPT_TFTP_BLKSIZE
Specify block size to use for TFTP data transmission. Valid range as per RFC2348 is 8-65464 bytes. The default of 512 bytes will be
used if this option is not specified. The specified block size will only be used pending support by the remote server. If the server
does not return an option acknowledgement or returns an option acknowledgement with no blksize, the default of 512 bytes will be
used. (added in 7.19.4)
FTP OPTIONS
CURLOPT_FTPPORT
Pass a pointer to a zero terminated string as parameter. It will be used to get the IP address to use for the FTP PORT instruction.
The PORT instruction tells the remote server to connect to our specified IP address. The string may be a plain IP address, a host
name, a network interface name (under Unix) or just a '-' symbol to let the library use your system's default IP address. Default
FTP operations are passive, and thus won't use PORT.
The address can be followed by a ':' to specify a port, optionally followed by a '-' to specify a port range. If the port specified
is 0, the operating system will pick a free port. If a range is provided and all ports in the range are not available, libcurl will
report CURLE_FTP_PORT_FAILED for the handle. Invalid port/range settings are ignored. IPv6 addresses followed by a port or por-
trange have to be in brackets. IPv6 addresses without port/range specifier can be in brackets. (added in 7.19.5)
Examples with specified ports:
eth0:0
192.168.1.2:32000-33000
curl.se:32123
[::1]:1234-4567
You disable PORT again and go back to using the passive version by setting this option to NULL.
CURLOPT_QUOTE
Pass a pointer to a linked list of FTP or SFTP commands to pass to the server prior to your FTP request. This will be done before
any other commands are issued (even before the CWD command for FTP). The linked list should be a fully valid list of 'struct
curl_slist' structs properly filled in with text strings. Use curl_slist_append(3) to append strings (commands) to the list, and
clear the entire list afterwards with curl_slist_free_all(3). Disable this operation again by setting a NULL to this option. When
speaking to a FTP (or SFTP since 7.24.0) server, prefix the command with an asterisk (*) to make libcurl continue even if the com-
mand fails as by default libcurl will stop at first failure.
The set of valid FTP commands depends on the server (see RFC959 for a list of mandatory commands).
The valid SFTP commands are: chgrp, chmod, chown, ln, mkdir, pwd, rename, rm, rmdir, symlink (see curl(1)) (SFTP support added in
7.16.3)
CURLOPT_POSTQUOTE
Pass a pointer to a linked list of FTP or SFTP commands to pass to the server after your FTP transfer request. The commands will
only be run if no error occurred. The linked list should be a fully valid list of struct curl_slist structs properly filled in as
described for CURLOPT_QUOTE. Disable this operation again by setting a NULL to this option.
CURLOPT_PREQUOTE
Pass a pointer to a linked list of FTP commands to pass to the server after the transfer type is set. The linked list should be a
fully valid list of struct curl_slist structs properly filled in as described for CURLOPT_QUOTE. Disable this operation again by
setting a NULL to this option. Before version 7.16.0, if you also set CURLOPT_NOBODY to 1, this option didn't work.
CURLOPT_DIRLISTONLY
A parameter set to 1 tells the library to just list the names of files in a directory, instead of doing a full directory listing
that would include file sizes, dates etc. This works for FTP and SFTP URLs.
This causes an FTP NLST command to be sent on an FTP server. Beware that some FTP servers list only files in their response to
NLST; they might not include subdirectories and symbolic links.
Setting this option to 1 also implies a directory listing even if the URL doesn't end with a slash, which otherwise is necessary.
Do NOT use this option if you also use CURLOPT_WILDCARDMATCH as it will effectively break that feature then.
(This option was known as CURLOPT_FTPLISTONLY up to 7.16.4)
CURLOPT_APPEND
A parameter set to 1 tells the library to append to the remote file instead of overwrite it. This is only useful when uploading to
an FTP site.
(This option was known as CURLOPT_FTPAPPEND up to 7.16.4)
CURLOPT_FTP_USE_EPRT
Pass a long. If the value is 1, it tells curl to use the EPRT (and LPRT) command when doing active FTP downloads (which is enabled
by CURLOPT_FTPPORT). Using EPRT means that it will first attempt to use EPRT and then LPRT before using PORT, but if you pass zero
to this option, it will not try using EPRT or LPRT, only plain PORT. (Added in 7.10.5)
If the server is an IPv6 host, this option will have no effect as of 7.12.3.
CURLOPT_FTP_USE_EPSV
Pass a long. If the value is 1, it tells curl to use the EPSV command when doing passive FTP downloads (which it always does by
default). Using EPSV means that it will first attempt to use EPSV before using PASV, but if you pass zero to this option, it will
not try using EPSV, only plain PASV.
If the server is an IPv6 host, this option will have no effect as of 7.12.3.
CURLOPT_FTP_USE_PRET
Pass a long. If the value is 1, it tells curl to send a PRET command before PASV (and EPSV). Certain FTP servers, mainly drftpd,
require this non-standard command for directory listings as well as up and downloads in PASV mode. Has no effect when using the
active FTP transfers mode. (Added in 7.20.0)
CURLOPT_FTP_CREATE_MISSING_DIRS
Pass a long. If the value is 1, curl will attempt to create any remote directory that it fails to CWD into. CWD is the command that
changes working directory. (Added in 7.10.7)
This setting also applies to SFTP-connections. curl will attempt to create the remote directory if it can't obtain a handle to the
target-location. The creation will fail if a file of the same name as the directory to create already exists or lack of permissions
prevents creation. (Added in 7.16.3)
Starting with 7.19.4, you can also set this value to 2, which will make libcurl retry the CWD command again if the subsequent MKD
command fails. This is especially useful if you're doing many simultaneous connections against the same server and they all have
this option enabled, as then CWD may first fail but then another connection does MKD before this connection and thus MKD fails but
trying CWD works! 7.19.4 also introduced the CURLFTP_CREATE_DIR and CURLFTP_CREATE_DIR_RETRY enum names for these arguments.
Before version 7.19.4, libcurl will simply ignore arguments set to 2 and act as if 1 was selected.
CURLOPT_FTP_RESPONSE_TIMEOUT
Pass a long. Causes curl to set a timeout period (in seconds) on the amount of time that the server is allowed to take in order to
generate a response message for a command before the session is considered hung. While curl is waiting for a response, this value
overrides CURLOPT_TIMEOUT. It is recommended that if used in conjunction with CURLOPT_TIMEOUT, you set CURLOPT_FTP_RESPONSE_TIMEOUT
to a value smaller than CURLOPT_TIMEOUT. (Added in 7.10.8)
CURLOPT_FTP_ALTERNATIVE_TO_USER
Pass a char * as parameter, pointing to a string which will be used to authenticate if the usual FTP "USER user" and "PASS password"
negotiation fails. This is currently only known to be required when connecting to Tumbleweed's Secure Transport FTPS server using
client certificates for authentication. (Added in 7.15.5)
CURLOPT_FTP_SKIP_PASV_IP
Pass a long. If set to 1, it instructs libcurl to not use the IP address the server suggests in its 227-response to libcurl's PASV
command when libcurl connects the data connection. Instead libcurl will re-use the same IP address it already uses for the control
connection. But it will use the port number from the 227-response. (Added in 7.14.2)
This option has no effect if PORT, EPRT or EPSV is used instead of PASV.
CURLOPT_FTPSSLAUTH
Pass a long using one of the values from below, to alter how libcurl issues "AUTH TLS" or "AUTH SSL" when FTP over SSL is activated
(see CURLOPT_USE_SSL). (Added in 7.12.2)
CURLFTPAUTH_DEFAULT
Allow libcurl to decide.
CURLFTPAUTH_SSL
Try "AUTH SSL" first, and only if that fails try "AUTH TLS".
CURLFTPAUTH_TLS
Try "AUTH TLS" first, and only if that fails try "AUTH SSL".
CURLOPT_FTP_SSL_CCC
If enabled, this option makes libcurl use CCC (Clear Command Channel). It shuts down the SSL/TLS layer after authenticating. The
rest of the control channel communication will be unencrypted. This allows NAT routers to follow the FTP transaction. Pass a long
using one of the values below. (Added in 7.16.1)
CURLFTPSSL_CCC_NONE
Don't attempt to use CCC.
CURLFTPSSL_CCC_PASSIVE
Do not initiate the shutdown, but wait for the server to do it. Do not send a reply.
CURLFTPSSL_CCC_ACTIVE
Initiate the shutdown and wait for a reply.
CURLOPT_FTP_ACCOUNT
Pass a pointer to a zero terminated string (or NULL to disable). When an FTP server asks for "account data" after user name and
password has been provided, this data is sent off using the ACCT command. (Added in 7.13.0)
CURLOPT_FTP_FILEMETHOD
Pass a long that should have one of the following values. This option controls what method libcurl should use to reach a file on a
FTP(S) server. The argument should be one of the following alternatives:
CURLFTPMETHOD_MULTICWD
libcurl does a single CWD operation for each path part in the given URL. For deep hierarchies this means many commands. This
is how RFC1738 says it should be done. This is the default but the slowest behavior.
CURLFTPMETHOD_NOCWD
libcurl does no CWD at all. libcurl will do SIZE, RETR, STOR etc and give a full path to the server for all these commands.
This is the fastest behavior.
CURLFTPMETHOD_SINGLECWD
libcurl does one CWD with the full target directory and then operates on the file "normally" (like in the multicwd case).
This is somewhat more standards compliant than 'nocwd' but without the full penalty of 'multicwd'.
(Added in 7.15.1)
RTSP OPTIONS
CURLOPT_RTSP_REQUEST
Tell libcurl what kind of RTSP request to make. Pass one of the following RTSP enum values. Unless noted otherwise, commands require
the Session ID to be initialized. (Added in 7.20.0)
CURL_RTSPREQ_OPTIONS
Used to retrieve the available methods of the server. The application is responsible for parsing and obeying the response.
(The session ID is not needed for this method.) (Added in 7.20.0)
CURL_RTSPREQ_DESCRIBE
Used to get the low level description of a stream. The application should note what formats it understands in the 'Accept:'
header. Unless set manually, libcurl will automatically fill in 'Accept: application/sdp'. Time-condition headers will be
added to Describe requests if the CURLOPT_TIMECONDITION option is active. (The session ID is not needed for this method)
(Added in 7.20.0)
CURL_RTSPREQ_ANNOUNCE
When sent by a client, this method changes the description of the session. For example, if a client is using the server to
record a meeting, the client can use Announce to inform the server of all the meta-information about the session. ANNOUNCE
acts like a HTTP PUT or POST just like CURL_RTSPREQ_SET_PARAMETER (Added in 7.20.0)
CURL_RTSPREQ_SETUP
Setup is used to initialize the transport layer for the session. The application must set the desired Transport options for a
session by using the CURLOPT_RTSP_TRANSPORT option prior to calling setup. If no session ID is currently set with CUR-
LOPT_RTSP_SESSION_ID, libcurl will extract and use the session ID in the response to this request. (The session ID is not
needed for this method). (Added in 7.20.0)
CURL_RTSPREQ_PLAY
Send a Play command to the server. Use the CURLOPT_RANGE option to modify the playback time (e.g. 'npt=10-15'). (Added in
7.20.0)
CURL_RTSPREQ_PAUSE
Send a Pause command to the server. Use the CURLOPT_RANGE option with a single value to indicate when the stream should be
halted. (e.g. npt='25') (Added in 7.20.0)
CURL_RTSPREQ_TEARDOWN
This command terminates an RTSP session. Simply closing a connection does not terminate the RTSP session since it is valid to
control an RTSP session over different connections. (Added in 7.20.0)
CURL_RTSPREQ_GET_PARAMETER
Retrieve a parameter from the server. By default, libcurl will automatically include a Content-Type: text/parameters header
on all non-empty requests unless a custom one is set. GET_PARAMETER acts just like a HTTP PUT or POST (see
CURL_RTSPREQ_SET_PARAMETER). Applications wishing to send a heartbeat message (e.g. in the presence of a server-specified
timeout) should send use an empty GET_PARAMETER request. (Added in 7.20.0)
CURL_RTSPREQ_SET_PARAMETER
Set a parameter on the server. By default, libcurl will automatically include a Content-Type: text/parameters header unless a
custom one is set. The interaction with SET_PARAMTER is much like a HTTP PUT or POST. An application may either use CUR-
LOPT_UPLOAD with CURLOPT_READDATA like a HTTP PUT, or it may use CURLOPT_POSTFIELDS like a HTTP POST. No chunked transfers
are allowed, so the application must set the CURLOPT_INFILESIZE in the former and CURLOPT_POSTFIELDSIZE in the latter. Also,
there is no use of multi-part POSTs within RTSP. (Added in 7.20.0)
CURL_RTSPREQ_RECORD
Used to tell the server to record a session. Use the CURLOPT_RANGE option to modify the record time. (Added in 7.20.0)
CURL_RTSPREQ_RECEIVE
This is a special request because it does not send any data to the server. The application may call this function in order to
receive interleaved RTP data. It will return after processing one read buffer of data in order to give the application a
chance to run. (Added in 7.20.0)
CURLOPT_RTSP_SESSION_ID
Pass a char * as a parameter to set the value of the current RTSP Session ID for the handle. Useful for resuming an in-progress ses-
sion. Once this value is set to any non-NULL value, libcurl will return CURLE_RTSP_SESSION_ERROR if ID received from the server does
not match. If unset (or set to NULL), libcurl will automatically set the ID the first time the server sets it in a response. (Added
in 7.20.0)
CURLOPT_RTSP_STREAM_URI
Set the stream URI to operate on by passing a char * . For example, a single session may be controlling rtsp://foo/twister/audio and
rtsp://foo/twister/video and the application can switch to the appropriate stream using this option. If unset, libcurl will default
to operating on generic server options by passing '*' in the place of the RTSP Stream URI. This option is distinct from CURLOPT_URL.
When working with RTSP, the CURLOPT_STREAM_URI indicates what URL to send to the server in the request header while the CURLOPT_URL
indicates where to make the connection to. (e.g. the CURLOPT_URL for the above examples might be set to rtsp://foo/twister (Added
in 7.20.0)
CURLOPT_RTSP_TRANSPORT
Pass a char * to tell libcurl what to pass for the Transport: header for this RTSP session. This is mainly a convenience method to
avoid needing to set a custom Transport: header for every SETUP request. The application must set a Transport: header before issuing
a SETUP request. (Added in 7.20.0)
CURLOPT_RTSP_HEADER
This option is simply an alias for CURLOPT_HTTP_HEADER. Use this to replace the standard headers that RTSP and HTTP share. It is
also valid to use the shortcuts such as CURLOPT_USERAGENT. (Added in 7.20.0)
CURLOPT_RTSP_CLIENT_CSEQ
Manually set the the CSEQ number to issue for the next RTSP request. Useful if the application is resuming a previously broken con-
nection. The CSEQ will increment from this new number henceforth. (Added in 7.20.0)
CURLOPT_RTSP_SERVER_CSEQ
Manually set the CSEQ number to expect for the next RTSP Server->Client request. At the moment, this feature (listening for Server
requests) is unimplemented. (Added in 7.20.0)
PROTOCOL OPTIONS
CURLOPT_TRANSFERTEXT
A parameter set to 1 tells the library to use ASCII mode for FTP transfers, instead of the default binary transfer. For win32 sys-
tems it does not set the stdout to binary mode. This option can be usable when transferring text data between systems with different
views on certain characters, such as newlines or similar.
libcurl does not do a complete ASCII conversion when doing ASCII transfers over FTP. This is a known limitation/flaw that nobody has
rectified. libcurl simply sets the mode to ASCII and performs a standard transfer.
CURLOPT_PROXY_TRANSFER_MODE
Pass a long. If the value is set to 1 (one), it tells libcurl to set the transfer mode (binary or ASCII) for FTP transfers done via
a HTTP proxy, by appending ;type=a or ;type=i to the URL. Without this setting, or it being set to 0 (zero, the default), CUR-
LOPT_TRANSFERTEXT has no effect when doing FTP via a proxy. Beware that not all proxies support this feature. (Added in 7.18.0)
CURLOPT_CRLF
Pass a long. If the value is set to 1 (one), libcurl converts Unix newlines to CRLF newlines on transfers. Disable this option again
by setting the value to 0 (zero).
CURLOPT_RANGE
Pass a char * as parameter, which should contain the specified range you want. It should be in the format "X-Y", where X or Y may be
left out. HTTP transfers also support several intervals, separated with commas as in "X-Y,N-M". Using this kind of multiple inter-
vals will cause the HTTP server to send the response document in pieces (using standard MIME separation techniques). For RTSP, the
formatting of a range should follow RFC2326 Section 12.29. For RTSP, byte ranges are not permitted. Instead, ranges should be given
in npt, utc, or smpte formats.
Pass a NULL to this option to disable the use of ranges.
Ranges work on HTTP, FTP, FILE (since 7.18.0), and RTSP (since 7.20.0) transfers only.
CURLOPT_RESUME_FROM
Pass a long as parameter. It contains the offset in number of bytes that you want the transfer to start from. Set this option to 0
to make the transfer start from the beginning (effectively disabling resume). For FTP, set this option to -1 to make the transfer
start from the end of the target file (useful to continue an interrupted upload).
When doing uploads with FTP, the resume position is where in the local/source file libcurl should try to resume the upload from and
it will then append the source file to the remote target file.
CURLOPT_RESUME_FROM_LARGE
Pass a curl_off_t as parameter. It contains the offset in number of bytes that you want the transfer to start from. (Added in
7.11.0)
CURLOPT_CUSTOMREQUEST
Pass a pointer to a zero terminated string as parameter. It can be used to specify the request instead of GET or HEAD when perform-
ing HTTP based requests, instead of LIST and NLST when performing FTP directory listings and instead of LIST and RETR when issuing
POP3 based commands. This is particularly useful, for example, for performing a HTTP DELETE request or a POP3 DELE command.
Please don't perform this at will, on HTTP based requests, by making sure your server supports the command you are sending first.
When you change the request method by setting CURLOPT_CUSTOMREQUEST to something, you don't actually change how libcurl behaves or
acts in regards to the particular request method, it will only change the actual string sent in the request.
For example:
With the HTTP protocol when you tell libcurl to do a HEAD request, but then specify a GET though a custom request libcurl will still
act as if it sent a HEAD. To switch to a proper HEAD use CURLOPT_NOBODY, to switch to a proper POST use CURLOPT_POST or CUR-
LOPT_POSTFIELDS and to switch to a proper GET use CURLOPT_HTTPGET.
With the POP3 protocol when you tell libcurl to use a custom request it will behave like a LIST or RETR command was sent where it
expects data to be returned by the server. As such CURLOPT_NOBODY should be used when specifying commands such as DELE and NOOP for
example.
Restore to the internal default by setting this to NULL.
Many people have wrongly used this option to replace the entire request with their own, including multiple headers and POST con-
tents. While that might work in many cases, it will cause libcurl to send invalid requests and it could possibly confuse the remote
server badly. Use CURLOPT_POST and CURLOPT_POSTFIELDS to set POST data. Use CURLOPT_HTTPHEADER to replace or extend the set of head-
ers sent by libcurl. Use CURLOPT_HTTP_VERSION to change HTTP version.
(Support for POP3 added in 7.26.0)
CURLOPT_FILETIME
Pass a long. If it is 1, libcurl will attempt to get the modification date of the remote document in this operation. This requires
that the remote server sends the time or replies to a time querying command. The curl_easy_getinfo(3) function with the
CURLINFO_FILETIME argument can be used after a transfer to extract the received time (if any).
CURLOPT_NOBODY
A parameter set to 1 tells the library to not include the body-part in the output. This is only relevant for protocols that have
separate header and body parts. On HTTP(S) servers, this will make libcurl do a HEAD request.
To change request to GET, you should use CURLOPT_HTTPGET. Change request to POST with CURLOPT_POST etc.
CURLOPT_INFILESIZE
When uploading a file to a remote site, this option should be used to tell libcurl what the expected size of the infile is. This
value should be passed as a long. See also CURLOPT_INFILESIZE_LARGE.
For uploading using SCP, this option or CURLOPT_INFILESIZE_LARGE is mandatory.
When sending emails using SMTP, this command can be used to specify the optional SIZE parameter for the MAIL FROM command. (Added in
7.23.0)
This option does not limit how much data libcurl will actually send, as that is controlled entirely by what the read callback
returns.
CURLOPT_INFILESIZE_LARGE
When uploading a file to a remote site, this option should be used to tell libcurl what the expected size of the infile is. This
value should be passed as a curl_off_t. (Added in 7.11.0)
For uploading using SCP, this option or CURLOPT_INFILESIZE is mandatory.
This option does not limit how much data libcurl will actually send, as that is controlled entirely by what the read callback
returns.
CURLOPT_UPLOAD
A parameter set to 1 tells the library to prepare for an upload. The CURLOPT_READDATA and CURLOPT_INFILESIZE or CURLOPT_INFILE-
SIZE_LARGE options are also interesting for uploads. If the protocol is HTTP, uploading means using the PUT request unless you tell
libcurl otherwise.
Using PUT with HTTP 1.1 implies the use of a "Expect: 100-continue" header. You can disable this header with CURLOPT_HTTPHEADER as
usual.
If you use PUT to a HTTP 1.1 server, you can upload data without knowing the size before starting the transfer if you use chunked
encoding. You enable this by adding a header like "Transfer-Encoding: chunked" with CURLOPT_HTTPHEADER. With HTTP 1.0 or without
chunked transfer, you must specify the size.
CURLOPT_MAXFILESIZE
Pass a long as parameter. This allows you to specify the maximum size (in bytes) of a file to download. If the file requested is
larger than this value, the transfer will not start and CURLE_FILESIZE_EXCEEDED will be returned.
The file size is not always known prior to download, and for such files this option has no effect even if the file transfer ends up
being larger than this given limit. This concerns both FTP and HTTP transfers.
CURLOPT_MAXFILESIZE_LARGE
Pass a curl_off_t as parameter. This allows you to specify the maximum size (in bytes) of a file to download. If the file requested
is larger than this value, the transfer will not start and CURLE_FILESIZE_EXCEEDED will be returned. (Added in 7.11.0)
The file size is not always known prior to download, and for such files this option has no effect even if the file transfer ends up
being larger than this given limit. This concerns both FTP and HTTP transfers.
CURLOPT_TIMECONDITION
Pass a long as parameter. This defines how the CURLOPT_TIMEVALUE time value is treated. You can set this parameter to CURL_TIME-
COND_IFMODSINCE or CURL_TIMECOND_IFUNMODSINCE. This feature applies to HTTP, FTP, RTSP, and FILE.
The last modification time of a file is not always known and in such instances this feature will have no effect even if the given
time condition would not have been met. curl_easy_getinfo(3) with the CURLINFO_CONDITION_UNMET option can be used after a transfer
to learn if a zero-byte successful "transfer" was due to this condition not matching.
CURLOPT_TIMEVALUE
Pass a long as parameter. This should be the time in seconds since 1 Jan 1970, and the time will be used in a condition as specified
with CURLOPT_TIMECONDITION.
CONNECTION OPTIONS
CURLOPT_TIMEOUT
Pass a long as parameter containing the maximum time in seconds that you allow the libcurl transfer operation to take. Normally,
name lookups can take a considerable time and limiting operations to less than a few minutes risk aborting perfectly normal opera-
tions. This option will cause curl to use the SIGALRM to enable time-outing system calls.
In unix-like systems, this might cause signals to be used unless CURLOPT_NOSIGNAL is set.
Default timeout is 0 (zero) which means it never times out.
CURLOPT_TIMEOUT_MS
Like CURLOPT_TIMEOUT but takes number of milliseconds instead. If libcurl is built to use the standard system name resolver, that
portion of the transfer will still use full-second resolution for timeouts with a minimum timeout allowed of one second. (Added in
7.16.2)
CURLOPT_LOW_SPEED_LIMIT
Pass a long as parameter. It contains the transfer speed in bytes per second that the transfer should be below during CUR-
LOPT_LOW_SPEED_TIME seconds for the library to consider it too slow and abort.
CURLOPT_LOW_SPEED_TIME
Pass a long as parameter. It contains the time in seconds that the transfer should be below the CURLOPT_LOW_SPEED_LIMIT for the
library to consider it too slow and abort.
CURLOPT_MAX_SEND_SPEED_LARGE
Pass a curl_off_t as parameter. If an upload exceeds this speed (counted in bytes per second) on cumulative average during the
transfer, the transfer will pause to keep the average rate less than or equal to the parameter value. Defaults to unlimited speed.
(Added in 7.15.5)
CURLOPT_MAX_RECV_SPEED_LARGE
Pass a curl_off_t as parameter. If a download exceeds this speed (counted in bytes per second) on cumulative average during the
transfer, the transfer will pause to keep the average rate less than or equal to the parameter value. Defaults to unlimited speed.
(Added in 7.15.5)
CURLOPT_MAXCONNECTS
Pass a long. The set number will be the persistent connection cache size. The set amount will be the maximum amount of simultane-
ously open connections that libcurl may cache in this easy handle. Default is 5, and there isn't much point in changing this value
unless you are perfectly aware of how this works and changes libcurl's behaviour. This concerns connections using any of the proto-
cols that support persistent connections.
When reaching the maximum limit, curl closes the oldest one in the cache to prevent increasing the number of open connections.
If you already have performed transfers with this curl handle, setting a smaller MAXCONNECTS than before may cause open connections
to get closed unnecessarily.
If you add this easy handle to a multi handle, this setting is not acknowledged, and you must instead use curl_multi_setopt(3) and
the CURLMOPT_MAXCONNECTS option.
CURLOPT_CLOSEPOLICY
(Obsolete) This option does nothing.
CURLOPT_FRESH_CONNECT
Pass a long. Set to 1 to make the next transfer use a new (fresh) connection by force. If the connection cache is full before this
connection, one of the existing connections will be closed as according to the selected or default policy. This option should be
used with caution and only if you understand what it does. Set this to 0 to have libcurl attempt re-using an existing connection
(default behavior).
CURLOPT_FORBID_REUSE
Pass a long. Set to 1 to make the next transfer explicitly close the connection when done. Normally, libcurl keeps all connections
alive when done with one transfer in case a succeeding one follows that can re-use them. This option should be used with caution
and only if you understand what it does. Set to 0 to have libcurl keep the connection open for possible later re-use (default behav-
ior).
CURLOPT_CONNECTTIMEOUT
Pass a long. It should contain the maximum time in seconds that you allow the connection to the server to take. This only limits
the connection phase, once it has connected, this option is of no more use. Set to zero to switch to the default built-in connection
timeout - 300 seconds. See also the CURLOPT_TIMEOUT option.
In unix-like systems, this might cause signals to be used unless CURLOPT_NOSIGNAL is set.
CURLOPT_CONNECTTIMEOUT_MS
Like CURLOPT_CONNECTTIMEOUT but takes the number of milliseconds instead. If libcurl is built to use the standard system name
resolver, that portion of the connect will still use full-second resolution for timeouts with a minimum timeout allowed of one sec-
ond. (Added in 7.16.2)
CURLOPT_IPRESOLVE
Allows an application to select what kind of IP addresses to use when resolving host names. This is only interesting when using host
names that resolve addresses using more than one version of IP. The allowed values are:
CURL_IPRESOLVE_WHATEVER
Default, resolves addresses to all IP versions that your system allows.
CURL_IPRESOLVE_V4
Resolve to IPv4 addresses.
CURL_IPRESOLVE_V6
Resolve to IPv6 addresses.
CURLOPT_CONNECT_ONLY
Pass a long. If the parameter equals 1, it tells the library to perform all the required proxy authentication and connection setup,
but no data transfer. This option is implemented for HTTP, SMTP and POP3.
The option can be used to simply test a connection to a server, but is more useful when used with the CURLINFO_LASTSOCKET option to
curl_easy_getinfo(3) as the library can set up the connection and then the application can obtain the most recently used socket for
special data transfers. (Added in 7.15.2)
CURLOPT_USE_SSL
Pass a long using one of the values from below, to make libcurl use your desired level of SSL for the transfer. (Added in 7.11.0)
This is for enabling SSL/TLS when you use FTP, SMTP, POP3, IMAP etc.
(This option was known as CURLOPT_FTP_SSL up to 7.16.4, and the constants were known as CURLFTPSSL_*)
CURLUSESSL_NONE
Don't attempt to use SSL.
CURLUSESSL_TRY
Try using SSL, proceed as normal otherwise.
CURLUSESSL_CONTROL
Require SSL for the control connection or fail with CURLE_USE_SSL_FAILED.
CURLUSESSL_ALL
Require SSL for all communication or fail with CURLE_USE_SSL_FAILED.
CURLOPT_RESOLVE
Pass a pointer to a linked list of strings with host name resolve information to use for requests with this handle. The linked list
should be a fully valid list of struct curl_slist structs properly filled in. Use curl_slist_append(3) to create the list and
curl_slist_free_all(3) to clean up an entire list.
Each single name resolve string should be written using the format HOST:PORT:ADDRESS where HOST is the name libcurl will try to
resolve, PORT is the port number of the service where libcurl wants to connect to the HOST and ADDRESS is the numerical IP address.
If libcurl is built to support IPv6, ADDRESS can of course be either IPv4 or IPv6 style addressing.
This option effectively pre-populates the DNS cache with entries for the host+port pair so redirects and everything that operations
against the HOST+PORT will instead use your provided ADDRESS.
You can remove names from the DNS cache again, to stop providing these fake resolves, by including a string in the linked list that
uses the format "-HOST:PORT". The host name must be prefixed with a dash, and the host name and port number must exactly match what
was already added previously.
(Added in 7.21.3)
CURLOPT_DNS_SERVERS
Set the list of DNS servers to be used instead of the system default. The format of the dns servers option is:
host[:port][,host[:port]]...
For example:
192.168.1.100,192.168.1.101,3.4.5.6
This option requires that libcurl was built with a resolver backend that supports this operation. The c-ares backend is the only
such one.
(Added in 7.24.0)
CURLOPT_ACCEPTTIMEOUT_MS
Pass a long telling libcurl the maximum number of milliseconds to wait for a server to connect back to libcurl when an active FTP
connection is used. If no timeout is set, the internal default of 60000 will be used. (Added in 7.24.0)
SSL and SECURITY OPTIONS
CURLOPT_SSLCERT
Pass a pointer to a zero terminated string as parameter. The string should be the file name of your certificate. The default format
is "PEM" and can be changed with CURLOPT_SSLCERTTYPE.
With NSS this can also be the nickname of the certificate you wish to authenticate with. If you want to use a file from the current
directory, please precede it with "./" prefix, in order to avoid confusion with a nickname.
CURLOPT_SSLCERTTYPE
Pass a pointer to a zero terminated string as parameter. The string should be the format of your certificate. Supported formats are
"PEM" and "DER". (Added in 7.9.3)
CURLOPT_SSLKEY
Pass a pointer to a zero terminated string as parameter. The string should be the file name of your private key. The default format
is "PEM" and can be changed with CURLOPT_SSLKEYTYPE.
CURLOPT_SSLKEYTYPE
Pass a pointer to a zero terminated string as parameter. The string should be the format of your private key. Supported formats are
"PEM", "DER" and "ENG".
The format "ENG" enables you to load the private key from a crypto engine. In this case CURLOPT_SSLKEY is used as an identifier
passed to the engine. You have to set the crypto engine with CURLOPT_SSLENGINE. "DER" format key file currently does not work
because of a bug in OpenSSL.
CURLOPT_KEYPASSWD
Pass a pointer to a zero terminated string as parameter. It will be used as the password required to use the CURLOPT_SSLKEY or CUR-
LOPT_SSH_PRIVATE_KEYFILE private key. You never needed a pass phrase to load a certificate but you need one to load your private
key.
(This option was known as CURLOPT_SSLKEYPASSWD up to 7.16.4 and CURLOPT_SSLCERTPASSWD up to 7.9.2)
CURLOPT_SSLENGINE
Pass a pointer to a zero terminated string as parameter. It will be used as the identifier for the crypto engine you want to use for
your private key.
If the crypto device cannot be loaded, CURLE_SSL_ENGINE_NOTFOUND is returned.
CURLOPT_SSLENGINE_DEFAULT
Sets the actual crypto engine as the default for (asymmetric) crypto operations.
If the crypto device cannot be set, CURLE_SSL_ENGINE_SETFAILED is returned.
Even though this option doesn't need any parameter, in some configurations curl_easy_setopt might be defined as a macro taking
exactly three arguments. Therefore, it's recommended to pass 1 as parameter to this option.
CURLOPT_SSLVERSION
Pass a long as parameter to control what version of SSL/TLS to attempt to use. (Added in 7.9.2)
The available options are:
CURL_SSLVERSION_DEFAULT
The default action. This will attempt to figure out the remote SSL protocol version, i.e. either SSLv3 or TLSv1 (but not
SSLv2, which became disabled by default with 7.18.1).
CURL_SSLVERSION_TLSv1
Force TLSv1.x
CURL_SSLVERSION_SSLv2
Force SSLv2
CURL_SSLVERSION_SSLv3
Force SSLv3
CURL_SSLVERSION_TLSv1_0
Force TLSv1.0 (Added in 7.34.0)
CURL_SSLVERSION_TLSv1_1
Force TLSv1.1 (Added in 7.34.0)
CURL_SSLVERSION_TLSv1_2
Force TLSv1.2 (Added in 7.34.0)
CURLOPT_SSL_VERIFYPEER
Pass a long as parameter. By default, curl assumes a value of 1.
This option determines whether curl verifies the authenticity of the peer's certificate. A value of 1 means curl verifies; 0 (zero)
means it doesn't.
When negotiating a SSL connection, the server sends a certificate indicating its identity. Curl verifies whether the certificate is
authentic, i.e. that you can trust that the server is who the certificate says it is. This trust is based on a chain of digital
signatures, rooted in certification authority (CA) certificates you supply. curl uses a default bundle of CA certificates (the path
for that is determined at build time) and you can specify alternate certificates with the CURLOPT_CAINFO option or the CURLOPT_CAP-
ATH option.
When CURLOPT_SSL_VERIFYPEER is nonzero, and the verification fails to prove that the certificate is authentic, the connection fails.
When the option is zero, the peer certificate verification succeeds regardless.
Authenticating the certificate is not by itself very useful. You typically want to ensure that the server, as authentically identi-
fied by its certificate, is the server you mean to be talking to. Use CURLOPT_SSL_VERIFYHOST to control that. The check that the
host name in the certificate is valid for the host name you're connecting to is done independently of the CURLOPT_SSL_VERIFYPEER
option.
CURLOPT_CAINFO
Pass a char * to a zero terminated string naming a file holding one or more certificates to verify the peer with. This makes sense
only when used in combination with the CURLOPT_SSL_VERIFYPEER option. If CURLOPT_SSL_VERIFYPEER is zero, CURLOPT_CAINFO need not
even indicate an accessible file.
This option is by default set to the system path where libcurl's cacert bundle is assumed to be stored, as established at build
time.
If curl is built against the NSS SSL library, the NSS PEM PKCS#11 module (libnsspem.so) needs to be available for this option to
work properly.
CURLOPT_ISSUERCERT
Pass a char * to a zero terminated string naming a file holding a CA certificate in PEM format. If the option is set, an additional
check against the peer certificate is performed to verify the issuer is indeed the one associated with the certificate provided by
the option. This additional check is useful in multi-level PKI where one needs to enforce that the peer certificate is from a spe-
cific branch of the tree.
This option makes sense only when used in combination with the CURLOPT_SSL_VERIFYPEER option. Otherwise, the result of the check is
not considered as failure.
A specific error code (CURLE_SSL_ISSUER_ERROR) is defined with the option, which is returned if the setup of the SSL/TLS session has
failed due to a mismatch with the issuer of peer certificate (CURLOPT_SSL_VERIFYPEER has to be set too for the check to fail).
(Added in 7.19.0)
CURLOPT_CAPATH
Pass a char * to a zero terminated string naming a directory holding multiple CA certificates to verify the peer with. If libcurl is
built against OpenSSL, the certificate directory must be prepared using the openssl c_rehash utility. This makes sense only when
used in combination with the CURLOPT_SSL_VERIFYPEER option. If CURLOPT_SSL_VERIFYPEER is zero, CURLOPT_CAPATH need not even indi-
cate an accessible path. The CURLOPT_CAPATH function apparently does not work in Windows due to some limitation in openssl. This
option is OpenSSL-specific and does nothing if libcurl is built to use GnuTLS. NSS-powered libcurl provides the option only for
backward compatibility.
CURLOPT_CRLFILE
Pass a char * to a zero terminated string naming a file with the concatenation of CRL (in PEM format) to use in the certificate val-
idation that occurs during the SSL exchange.
When curl is built to use NSS or GnuTLS, there is no way to influence the use of CRL passed to help in the verification process.
When libcurl is built with OpenSSL support, X509_V_FLAG_CRL_CHECK and X509_V_FLAG_CRL_CHECK_ALL are both set, requiring CRL check
against all the elements of the certificate chain if a CRL file is passed.
This option makes sense only when used in combination with the CURLOPT_SSL_VERIFYPEER option.
A specific error code (CURLE_SSL_CRL_BADFILE) is defined with the option. It is returned when the SSL exchange fails because the CRL
file cannot be loaded. A failure in certificate verification due to a revocation information found in the CRL does not trigger this
specific error. (Added in 7.19.0)
CURLOPT_SSL_VERIFYHOST
Pass a long as parameter.
This option determines whether libcurl verifies that the server cert is for the server it is known as.
When negotiating a SSL connection, the server sends a certificate indicating its identity.
When CURLOPT_SSL_VERIFYHOST is 2, that certificate must indicate that the server is the server to which you meant to connect, or the
connection fails.
Curl considers the server the intended one when the Common Name field or a Subject Alternate Name field in the certificate matches
the host name in the URL to which you told Curl to connect.
When the value is 1, libcurl will return a failure. It was previously (in 7.28.0 and earlier) a debug option of some sorts, but it
is no longer supported due to frequently leading to programmer mistakes.
When the value is 0, the connection succeeds regardless of the names in the certificate.
The default value for this option is 2.
This option controls checking the server's certificate's claimed identity. The server could be lying. To control lying, see CUR-
LOPT_SSL_VERIFYPEER. If libcurl is built against NSS and CURLOPT_SSL_VERIFYPEER is zero, CURLOPT_SSL_VERIFYHOST is ignored.
CURLOPT_CERTINFO
Pass a long set to 1 to enable libcurl's certificate chain info gatherer. With this enabled, libcurl (if built with OpenSSL) will
extract lots of information and data about the certificates in the certificate chain used in the SSL connection. This data is then
possible to extract after a transfer using curl_easy_getinfo(3) and its option CURLINFO_CERTINFO. (Added in 7.19.1)
CURLOPT_RANDOM_FILE
Pass a char * to a zero terminated file name. The file will be used to read from to seed the random engine for SSL. The more random
the specified file is, the more secure the SSL connection will become.
CURLOPT_EGDSOCKET
Pass a char * to the zero terminated path name to the Entropy Gathering Daemon socket. It will be used to seed the random engine for
SSL.
CURLOPT_SSL_CIPHER_LIST
Pass a char *, pointing to a zero terminated string holding the list of ciphers to use for the SSL connection. The list must be syn-
tactically correct, it consists of one or more cipher strings separated by colons. Commas or spaces are also acceptable separators
but colons are normally used, !, - and + can be used as operators.
For OpenSSL and GnuTLS valid examples of cipher lists include 'RC4-SHA', 'SHA1+DES', 'TLSv1' and 'DEFAULT'. The default list is nor-
mally set when you compile OpenSSL.
You'll find more details about cipher lists on this URL: http://www.openssl.org/docs/apps/ciphers.html
For NSS, valid examples of cipher lists include 'rsa_rc4_128_md5', 'rsa_aes_128_sha', etc. With NSS you don't add/remove ciphers. If
one uses this option then all known ciphers are disabled and only those passed in are enabled.
You'll find more details about the NSS cipher lists on this URL: http://git.fedora-
hosted.org/cgit/mod_nss.git/plain/docs/mod_nss.html#Directives
CURLOPT_SSL_SESSIONID_CACHE
Pass a long set to 0 to disable libcurl's use of SSL session-ID caching. Set this to 1 to enable it. By default all transfers are
done using the cache. While nothing ever should get hurt by attempting to reuse SSL session-IDs, there seem to be broken SSL imple-
mentations in the wild that may require you to disable this in order for you to succeed. (Added in 7.16.0)
CURLOPT_SSL_OPTIONS
Pass a long with a bitmask to tell libcurl about specific SSL behaviors.
CURLSSLOPT_ALLOW_BEAST is the only supported bit and by setting this the user will tell libcurl to not attempt to use any work-
arounds for a security flaw in the SSL3 and TLS1.0 protocols. If this option isn't used or this bit is set to 0, the SSL layer
libcurl uses may use a work-around for this flaw although it might cause interoperability problems with some (older) SSL implementa-
tions. WARNING: avoiding this work-around loosens the security, and by setting this option to 1 you ask for exactly that. (Added in
7.25.0)
CURLOPT_KRBLEVEL
Pass a char * as parameter. Set the kerberos security level for FTP; this also enables kerberos awareness. This is a string,
'clear', 'safe', 'confidential' or 'private'. If the string is set but doesn't match one of these, 'private' will be used. Set the
string to NULL to disable kerberos support for FTP.
(This option was known as CURLOPT_KRB4LEVEL up to 7.16.3)
CURLOPT_GSSAPI_DELEGATION
Set the parameter to CURLGSSAPI_DELEGATION_FLAG to allow unconditional GSSAPI credential delegation. The delegation is disabled by
default since 7.21.7. Set the parameter to CURLGSSAPI_DELEGATION_POLICY_FLAG to delegate only if the OK-AS-DELEGATE flag is set in
the service ticket in case this feature is supported by the GSSAPI implementation and the definition of GSS_C_DELEG_POLICY_FLAG was
available at compile-time. (Added in 7.22.0)
SSH OPTIONS
CURLOPT_SSH_AUTH_TYPES
Pass a long set to a bitmask consisting of one or more of CURLSSH_AUTH_PUBLICKEY, CURLSSH_AUTH_PASSWORD, CURLSSH_AUTH_HOST,
CURLSSH_AUTH_KEYBOARD and CURLSSH_AUTH_AGENT. Set CURLSSH_AUTH_ANY to let libcurl pick a suitable one. Currently CURLSSH_AUTH_HOST
has no effect. (Added in 7.16.1) If CURLSSH_AUTH_AGENT is used, libcurl attempts to connect to ssh-agent or pageant and let the
agent attempt the authentication. (Added in 7.28.0)
CURLOPT_SSH_HOST_PUBLIC_KEY_MD5
Pass a char * pointing to a string containing 32 hexadecimal digits. The string should be the 128 bit MD5 checksum of the remote
host's public key, and libcurl will reject the connection to the host unless the md5sums match. This option is only for SCP and SFTP
transfers. (Added in 7.17.1)
CURLOPT_SSH_PUBLIC_KEYFILE
Pass a char * pointing to a file name for your public key. If not used, libcurl defaults to $HOME/.ssh/id_dsa.pub if the HOME envi-
ronment variable is set, and just "id_dsa.pub" in the current directory if HOME is not set. (Added in 7.16.1) If an empty string is
passed, libcurl will pass no public key to libssh2 which then tries to compute it from the private key, this is known to work when
libssh2 1.4.0+ is linked against OpenSSL. (Added in 7.26.0)
CURLOPT_SSH_PRIVATE_KEYFILE
Pass a char * pointing to a file name for your private key. If not used, libcurl defaults to $HOME/.ssh/id_dsa if the HOME environ-
ment variable is set, and just "id_dsa" in the current directory if HOME is not set. If the file is password-protected, set the
password with CURLOPT_KEYPASSWD. (Added in 7.16.1)
CURLOPT_SSH_KNOWNHOSTS
Pass a pointer to a zero terminated string holding the file name of the known_host file to use. The known_hosts file should use the
OpenSSH file format as supported by libssh2. If this file is specified, libcurl will only accept connections with hosts that are
known and present in that file, with a matching public key. Use CURLOPT_SSH_KEYFUNCTION to alter the default behavior on host and
key (mis)matching. (Added in 7.19.6)
CURLOPT_SSH_KEYFUNCTION
Pass a pointer to a curl_sshkeycallback function. It gets called when the known_host matching has been done, to allow the applica-
tion to act and decide for libcurl how to proceed. The callback will only be called if CURLOPT_SSH_KNOWNHOSTS is also set.
The curl_sshkeycallback function gets passed the CURL handle, the key from the known_hosts file, the key from the remote site, info
from libcurl on the matching status and a custom pointer (set with CURLOPT_SSH_KEYDATA). It MUST return one of the following return
codes to tell libcurl how to act:
CURLKHSTAT_FINE_ADD_TO_FILE
The host+key is accepted and libcurl will append it to the known_hosts file before continuing with the connection. This will
also add the host+key combo to the known_host pool kept in memory if it wasn't already present there. The adding of data to
the file is done by completely replacing the file with a new copy, so the permissions of the file must allow this.
CURLKHSTAT_FINE
The host+key is accepted libcurl will continue with the connection. This will also add the host+key combo to the known_host
pool kept in memory if it wasn't already present there.
CURLKHSTAT_REJECT
The host+key is rejected. libcurl will deny the connection to continue and it will be closed.
CURLKHSTAT_DEFER
The host+key is rejected, but the SSH connection is asked to be kept alive. This feature could be used when the app wants to
somehow return back and act on the host+key situation and then retry without needing the overhead of setting it up from
scratch again.
(Added in 7.19.6)
CURLOPT_SSH_KEYDATA
Pass a void * as parameter. This pointer will be passed along verbatim to the callback set with CURLOPT_SSH_KEYFUNCTION. (Added in
7.19.6)
OTHER OPTIONS
CURLOPT_PRIVATE
Pass a void * as parameter, pointing to data that should be associated with this curl handle. The pointer can subsequently be
retrieved using curl_easy_getinfo(3) with the CURLINFO_PRIVATE option. libcurl itself does nothing with this data. (Added in 7.10.3)
CURLOPT_SHARE
Pass a share handle as a parameter. The share handle must have been created by a previous call to curl_share_init(3). Setting this
option, will make this curl handle use the data from the shared handle instead of keeping the data to itself. This enables several
curl handles to share data. If the curl handles are used simultaneously in multiple threads, you MUST use the locking methods in the
share handle. See curl_share_setopt(3) for details.
If you add a share that is set to share cookies, your easy handle will use that cookie cache and get the cookie engine enabled. If
you unshare an object that was using cookies (or change to another object that doesn't share cookies), the easy handle will get its
cookie engine disabled.
Data that the share object is not set to share will be dealt with the usual way, as if no share was used.
CURLOPT_NEW_FILE_PERMS
Pass a long as a parameter, containing the value of the permissions that will be assigned to newly created files on the remote
server. The default value is 0644, but any valid value can be used. The only protocols that can use this are sftp://, scp://, and
file://. (Added in 7.16.4)
CURLOPT_NEW_DIRECTORY_PERMS
Pass a long as a parameter, containing the value of the permissions that will be assigned to newly created directories on the remote
server. The default value is 0755, but any valid value can be used. The only protocols that can use this are sftp://, scp://, and
file://. (Added in 7.16.4)
TELNET OPTIONS
CURLOPT_TELNETOPTIONS
Provide a pointer to a curl_slist with variables to pass to the telnet negotiations. The variables should be in the format
<option=value>. libcurl supports the options 'TTYPE', 'XDISPLOC' and 'NEW_ENV'. See the TELNET standard for details.
RETURN VALUE
CURLE_OK (zero) means that the option was set properly, non-zero means an error occurred as <curl/curl.h> defines. See the libcurl-
errors(3) man page for the full list with descriptions.
If you try to set an option that libcurl doesn't know about, perhaps because the library is too old to support it or the option was removed
in a recent version, this function will return CURLE_FAILED_INIT.
SEE ALSO
curl_easy_init(3), curl_easy_cleanup(3), curl_easy_reset(3)
libcurl 7.20.0 1 Jan 2010 curl_easy_setopt(3)