soup_server_new ()

SoupServer *
soup_server_new (const char *optname1,
                 ...);

Creates a new SoupServer. This is exactly equivalent to calling g_object_new() and specifying SOUP_TYPE_SERVER as the type.

soup_server_set_ssl_cert_file ()

gboolean
soup_server_set_ssl_cert_file (SoupServer *server,
                               const char *ssl_cert_file,
                               const char *ssl_key_file,
                               GError **error);

Sets server up to do https, using the SSL/TLS certificate specified by ssl_cert_file and ssl_key_file (which may point to the same file).

Alternatively, you can set the “tls-certificate” property at construction time, if you already have a GTlsCertificate.

Since: 2.48

soup_server_listen ()

gboolean
soup_server_listen (SoupServer *server,
                    GSocketAddress *address,
                    SoupServerListenOptions options,
                    GError **error);

This attempts to set up server to listen for connections on address .

If options includes SOUP_SERVER_LISTEN_HTTPS, and server has been configured for TLS, then server will listen for https connections on this port. Otherwise it will listen for plain http.

You may call this method (along with the other "listen" methods) any number of times on a server, if you want to listen on multiple ports, or set up both http and https service.

After calling this method, server will begin accepting and processing connections as soon as the appropriate GMainContext is run.

Note that SoupServer never makes use of dual IPv4/IPv6 sockets; if address is an IPv6 address, it will only accept IPv6 connections. You must configure IPv4 listening separately.

Since: 2.48

soup_server_listen_all ()

gboolean
soup_server_listen_all (SoupServer *server,
                        guint port,
                        SoupServerListenOptions options,
                        GError **error);

This attempts to set up server to listen for connections on all interfaces on the system. (That is, it listens on the addresses 0.0.0.0 and/or ::, depending on whether options includes SOUP_SERVER_LISTEN_IPV4_ONLY, SOUP_SERVER_LISTEN_IPV6_ONLY, or neither.) If port is specified, server will listen on that port. If it is 0, server will find an unused port to listen on. (In that case, you can use soup_server_get_uris() to find out what port it ended up choosing.)

See soup_server_listen() for more details.

Since: 2.48

soup_server_listen_local ()

gboolean
soup_server_listen_local (SoupServer *server,
                          guint port,
                          SoupServerListenOptions options,
                          GError **error);

This attempts to set up server to listen for connections on "localhost" (that is, 127.0.0.1 and/or ::1, depending on whether options includes SOUP_SERVER_LISTEN_IPV4_ONLY, SOUP_SERVER_LISTEN_IPV6_ONLY, or neither). If port is specified, server will listen on that port. If it is 0, server will find an unused port to listen on. (In that case, you can use soup_server_get_uris() to find out what port it ended up choosing.)

See soup_server_listen() for more details.

Since: 2.48

soup_server_listen_socket ()

gboolean
soup_server_listen_socket (SoupServer *server,
                           GSocket *socket,
                           SoupServerListenOptions options,
                           GError **error);

This attempts to set up server to listen for connections on socket .

See soup_server_listen() for more details.

Since: 2.48

soup_server_listen_fd ()

gboolean
soup_server_listen_fd (SoupServer *server,
                       int fd,
                       SoupServerListenOptions options,
                       GError **error);

This attempts to set up server to listen for connections on fd .

See soup_server_listen() for more details.

Note that server will close fd when you free it or call soup_server_disconnect().

Since: 2.48

soup_server_get_listeners ()

GSList *
soup_server_get_listeners (SoupServer *server);

Gets server 's list of listening sockets.

You should treat these sockets as read-only; writing to or modifiying any of these sockets may cause server to malfunction.

(Beware that in contrast to the old soup_server_get_listener(), this function returns GSockets, not SoupSockets.)

soup_server_get_uris ()

GSList *
soup_server_get_uris (SoupServer *server);

Gets a list of URIs corresponding to the interfaces server is listening on. These will contain IP addresses, not hostnames, and will also indicate whether the given listener is http or https.

Note that if you used soup_server_listen_all(), the returned URIs will use the addresses 0.0.0.0 and ::, rather than actually returning separate URIs for each interface on the system.

Since: 2.48

soup_server_disconnect ()

void
soup_server_disconnect (SoupServer *server);

Closes and frees server 's listening sockets. If you are using the old SoupServer APIs, this also includes the effect of soup_server_quit().

Note that if there are currently requests in progress on server , that they will continue to be processed if server 's GMainContext is still running.

You can call soup_server_listen(), etc, after calling this function if you want to start listening again.

soup_server_is_https ()

gboolean
soup_server_is_https (SoupServer *server);

Checks whether server is capable of https.

In order for a server to run https, you must call soup_server_set_ssl_cert_file(), or set the “tls-certificate” property, to provide it with a certificate to use.

If you are using the deprecated single-listener APIs, then a return value of TRUE indicates that the SoupServer serves https exclusively. If you are using soup_server_listen(), etc, then a TRUE return value merely indicates that the server is able to do https, regardless of whether it actually currently is or not. Use soup_server_get_uris() to see if it currently has any https listeners.

soup_server_accept_iostream ()

gboolean
soup_server_accept_iostream (SoupServer *server,
                             GIOStream *stream,
                             GSocketAddress *local_addr,
                             GSocketAddress *remote_addr,
                             GError **error);

Add a new client stream to the server .

Since: 2.50

SoupServerCallback ()

void
(*SoupServerCallback) (SoupServer *server,
                       SoupMessage *msg,
                       const char *path,
                       GHashTable *query,
                       SoupClientContext *client,
                       gpointer user_data);

A callback used to handle requests to a SoupServer.

path and query contain the likewise-named components of the Request-URI, subject to certain assumptions. By default, SoupServer decodes all percent-encoding in the URI path, such that "/foo%2Fbar" is treated the same as "/foo/bar". If your server is serving resources in some non-POSIX-filesystem namespace, you may want to distinguish those as two distinct paths. In that case, you can set the SOUP_SERVER_RAW_PATHS property when creating the SoupServer, and it will leave those characters undecoded. (You may want to call soup_uri_normalize() to decode any percent-encoded characters that you aren't handling specially.)

query contains the query component of the Request-URI parsed according to the rules for HTML form handling. Although this is the only commonly-used query string format in HTTP, there is nothing that actually requires that HTTP URIs use that format; if your server needs to use some other format, you can just ignore query , and call soup_message_get_uri() and parse the URI's query field yourself.

See soup_server_add_handler() and soup_server_add_early_handler() for details of what handlers can/should do.

soup_server_add_handler ()

void
soup_server_add_handler (SoupServer *server,
                         const char *path,
                         SoupServerCallback callback,
                         gpointer user_data,
                         GDestroyNotify destroy);

Adds a handler to server for requests under path . If path is NULL or "/", then this will be the default handler for all requests that don't have a more specific handler. (Note though that if you want to handle requests to the special "*" URI, you must explicitly register a handler for "*"; the default handler will not be used for that case.)

For requests under path (that have not already been assigned a status code by a SoupAuthDomain, an early SoupServerHandler, or a signal handler), callback will be invoked after receiving the request body; the message's “method”, “request-headers”, and “request-body” fields will be filled in.

After determining what to do with the request, the callback must at a minimum call soup_message_set_status() (or soup_message_set_status_full()) on the message to set the response status code. Additionally, it may set response headers and/or fill in the response body.

If the callback cannot fully fill in the response before returning (eg, if it needs to wait for information from a database, or another network server), it should call soup_server_pause_message() to tell server to not send the response right away. When the response is ready, call soup_server_unpause_message() to cause it to be sent.

To send the response body a bit at a time using "chunked" encoding, first call soup_message_headers_set_encoding() to set SOUP_ENCODING_CHUNKED on the “response-headers”. Then call soup_message_body_append() (or soup_message_body_append_buffer()) to append each chunk as it becomes ready, and soup_server_unpause_message() to make sure it's running. (The server will automatically pause the message if it is using chunked encoding but no more chunks are available.) When you are done, call soup_message_body_complete() to indicate that no more chunks are coming.

soup_server_add_early_handler ()

void
soup_server_add_early_handler (SoupServer *server,
                               const char *path,
                               SoupServerCallback callback,
                               gpointer user_data,
                               GDestroyNotify destroy);

Adds an "early" handler to server for requests under path . Note that "normal" and "early" handlers are matched up together, so if you add a normal handler for "/foo" and an early handler for "/foo/bar", then a request to "/foo/bar" (or any path below it) will run only the early handler. (But if you add both handlers at the same path, then both will get run.)

For requests under path (that have not already been assigned a status code by a SoupAuthDomain or a signal handler), callback will be invoked after receiving the request headers, but before receiving the request body; the message's “method” and “request-headers” fields will be filled in.

Early handlers are generally used for processing requests with request bodies in a streaming fashion. If you determine that the request will contain a message body, normally you would call soup_message_body_set_accumulate() on the message's “request-body” to turn off request-body accumulation, and connect to the message's “got-chunk” signal to process each chunk as it comes in.

To complete the message processing after the full message body has been read, you can either also connect to “got-body”, or else you can register a non-early handler for path as well. As long as you have not set the “status-code” by the time “got-body” is emitted, the non-early handler will be run as well.

Since: 2.50

soup_server_remove_handler ()

void
soup_server_remove_handler (SoupServer *server,
                            const char *path);

Removes all handlers (early and normal) registered at path .

SoupServerWebsocketCallback ()

void
(*SoupServerWebsocketCallback) (SoupServer *server,
                                SoupWebsocketConnection *connection,
                                const char *path,
                                SoupClientContext *client,
                                gpointer user_data);

A callback used to handle WebSocket requests to a SoupServer. The callback will be invoked after sending the handshake response back to the client (and is only invoked if the handshake was successful).

path contains the path of the Request-URI, subject to the same rules as SoupServerCallback (qv).

soup_server_add_websocket_handler ()

void
soup_server_add_websocket_handler (SoupServer *server,
                                   const char *path,
                                   const char *origin,
                                   char **protocols,
                                   SoupServerWebsocketCallback callback,
                                   gpointer user_data,
                                   GDestroyNotify destroy);

Adds a WebSocket handler to server for requests under path . (If path is NULL or "/", then this will be the default handler for all requests that don't have a more specific handler.)

When a path has a WebSocket handler registered, server will check incoming requests for WebSocket handshakes after all other handlers have run (unless some earlier handler has already set a status code on the message), and update the request's status, response headers, and response body accordingly.

If origin is non-NULL, then only requests containing a matching "Origin" header will be accepted. If protocols is non-NULL, then only requests containing a compatible "Sec-WebSocket-Protocols" header will be accepted. More complicated requirements can be handled by adding a normal handler to path , and having it perform whatever checks are needed (possibly calling soup_server_check_websocket_handshake() one or more times), and setting a failure status code if the handshake should be rejected.

soup_client_context_get_local_address ()

GSocketAddress *
soup_client_context_get_local_address (SoupClientContext *client);

Retrieves the GSocketAddress associated with the local end of a connection.

Since: 2.48

soup_client_context_get_remote_address ()

GSocketAddress *
soup_client_context_get_remote_address
                               (SoupClientContext *client);

Retrieves the GSocketAddress associated with the remote end of a connection.

Since: 2.48

soup_client_context_get_host ()

const char *
soup_client_context_get_host (SoupClientContext *client);

Retrieves the IP address associated with the remote end of a connection.

soup_client_context_get_auth_domain ()

SoupAuthDomain *
soup_client_context_get_auth_domain (SoupClientContext *client);

Checks whether the request associated with client has been authenticated, and if so returns the SoupAuthDomain that authenticated it.

soup_client_context_get_auth_user ()

const char *
soup_client_context_get_auth_user (SoupClientContext *client);

Checks whether the request associated with client has been authenticated, and if so returns the username that the client authenticated as.

soup_client_context_get_gsocket ()

GSocket *
soup_client_context_get_gsocket (SoupClientContext *client);

Retrieves the GSocket that client is associated with.

If you are using this method to observe when multiple requests are made on the same persistent HTTP connection (eg, as the ntlm-test test program does), you will need to pay attention to socket destruction as well (eg, by using weak references), so that you do not get fooled when the allocator reuses the memory address of a previously-destroyed socket to represent a new socket.

Since: 2.48

soup_client_context_steal_connection ()

GIOStream *
soup_client_context_steal_connection (SoupClientContext *client);

"Steals" the HTTP connection associated with client from its SoupServer. This happens immediately, regardless of the current state of the connection; if the response to the current SoupMessage has not yet finished being sent, then it will be discarded; you can steal the connection from a “wrote-informational” or “wrote-body” signal handler if you need to wait for part or all of the response to be sent.

Note that when calling this function from C, client will most likely be freed as a side effect.

Since: 2.50

soup_server_add_auth_domain ()

void
soup_server_add_auth_domain (SoupServer *server,
                             SoupAuthDomain *auth_domain);

Adds an authentication domain to server . Each auth domain will have the chance to require authentication for each request that comes in; normally auth domains will require authentication for requests on certain paths that they have been set up to watch, or that meet other criteria set by the caller. If an auth domain determines that a request requires authentication (and the request doesn't contain authentication), server will automatically reject the request with an appropriate status (401 Unauthorized or 407 Proxy Authentication Required). If the request used the "100-continue" Expectation, server will reject it before the request body is sent.

soup_server_remove_auth_domain ()

void
soup_server_remove_auth_domain (SoupServer *server,
                                SoupAuthDomain *auth_domain);

Removes auth_domain from server .

soup_server_pause_message ()

void
soup_server_pause_message (SoupServer *server,
                           SoupMessage *msg);

Pauses I/O on msg . This can be used when you need to return from the server handler without having the full response ready yet. Use soup_server_unpause_message() to resume I/O.

This must only be called on SoupMessages which were created by the SoupServer and are currently doing I/O, such as those passed into a SoupServerCallback or emitted in a “request-read” signal.

soup_server_unpause_message ()

void
soup_server_unpause_message (SoupServer *server,
                             SoupMessage *msg);

Resumes I/O on msg . Use this to resume after calling soup_server_pause_message(), or after adding a new chunk to a chunked response.

I/O won't actually resume until you return to the main loop.

This must only be called on SoupMessages which were created by the SoupServer and are currently doing I/O, such as those passed into a SoupServerCallback or emitted in a “request-read” signal.

SoupServer

typedef struct _SoupServer SoupServer;

enum SoupServerListenOptions

Options to pass to soup_server_listen(), etc.

SOUP_SERVER_LISTEN_IPV4_ONLY and SOUP_SERVER_LISTEN_IPV6_ONLY only make sense with soup_server_listen_all() and soup_server_listen_local(), not plain soup_server_listen() (which simply listens on whatever kind of socket you give it). And you cannot specify both of them in a single call.

Since: 2.48

SoupClientContext

typedef struct SoupClientContext SoupClientContext;

A SoupClientContext provides additional information about the client making a particular request. In particular, you can use soup_client_context_get_auth_domain() and soup_client_context_get_auth_user() to determine if HTTP authentication was used successfully.

soup_client_context_get_remote_address() and/or soup_client_context_get_host() can be used to get information for logging or debugging purposes. soup_client_context_get_gsocket() may also be of use in some situations (eg, tracking when multiple requests are made on the same connection).

SOUP_SERVER_TLS_CERTIFICATE

#define SOUP_SERVER_TLS_CERTIFICATE "tls-certificate"

Alias for the “tls-certificate” property, qv.

Since: 2.38

SOUP_SERVER_RAW_PATHS

#define SOUP_SERVER_RAW_PATHS       "raw-paths"

Alias for the “raw-paths” property. (If TRUE, percent-encoding in the Request-URI path will not be automatically decoded.)

SOUP_SERVER_SERVER_HEADER

#define SOUP_SERVER_SERVER_HEADER   "server-header"

Alias for the “server-header” property, qv.

SOUP_SERVER_HTTP_ALIASES

#define SOUP_SERVER_HTTP_ALIASES    "http-aliases"

Alias for the “http-aliases” property, qv.

Since: 2.44

SOUP_SERVER_HTTPS_ALIASES

#define SOUP_SERVER_HTTPS_ALIASES   "https-aliases"

Alias for the “https-aliases” property, qv.

Since: 2.44

The “async-context” property

  “async-context”            gpointer

The server's GMainContext, if you are using the old API. Servers created using soup_server_listen() will listen on the GMainContext that was the thread-default context at the time soup_server_listen() was called.

Flags: Read / Write / Construct Only

The “http-aliases” property

  “http-aliases”             GStrv

A NULL-terminated array of URI schemes that should be considered to be aliases for "http". Eg, if this included "dav", than a URI of dav://example.com/path would be treated identically to http://example.com/path. In particular, this is needed in cases where a client sends requests with absolute URIs, where those URIs do not use "http:".

The default value is an array containing the single element "*", a special value which means that any scheme except "https" is considered to be an alias for "http".

See also “https-aliases”.

Flags: Read / Write

Since: 2.44

The “https-aliases” property

  “https-aliases”            GStrv

A comma-delimited list of URI schemes that should be considered to be aliases for "https". See “http-aliases” for more information.

The default value is NULL, meaning that no URI schemes are considered aliases for "https".

Flags: Read / Write

Since: 2.44

The “interface” property

  “interface”                SoupAddress *

The address of the network interface the server is listening on, if you are using the old SoupServer API. (This will not be set if you use soup_server_listen(), etc.)

Flags: Read / Write / Construct Only

The “port” property

  “port”                     guint

The port the server is listening on, if you are using the old SoupServer API. (This will not be set if you use soup_server_listen(), etc.)

Flags: Read / Write / Construct Only

Allowed values: <= 65536

Default value: 0

The “raw-paths” property

  “raw-paths”                gboolean

If %TRUE, percent-encoding in the Request-URI path will not be automatically decoded.

Flags: Read / Write / Construct Only

Default value: FALSE

The “server-header” property

  “server-header”            gchar *

If non-NULL, the value to use for the "Server" header on SoupMessages processed by this server.

The Server header is the server equivalent of the User-Agent header, and provides information about the server and its components. It contains a list of one or more product tokens, separated by whitespace, with the most significant product token coming first. The tokens must be brief, ASCII, and mostly alphanumeric (although "-", "_", and "." are also allowed), and may optionally include a "/" followed by a version string. You may also put comments, enclosed in parentheses, between or after the tokens.

Some HTTP server implementations intentionally do not use version numbers in their Server header, so that installations running older versions of the server don't end up advertising their vulnerability to specific security holes.

As with “user_agent”, if you set a “server_header” property that has trailing whitespace, SoupServer will append its own product token (eg, "libsoup/2.3.2") to the end of the header for you.

Flags: Read / Write / Construct

Default value: NULL

The “ssl-cert-file” property

  “ssl-cert-file”            gchar *

Path to a file containing a PEM-encoded certificate.

If you set this property and “ssl-key-file” at construct time, then soup_server_new() will try to read the files; if it cannot, it will return NULL, with no explicit indication of what went wrong (and logging a warning with newer versions of glib, since returning NULL from a constructor is illegal).

Flags: Read / Write / Construct Only

Default value: NULL

The “ssl-key-file” property

  “ssl-key-file”             gchar *

Path to a file containing a PEM-encoded private key. See “ssl-cert-file” for more information about how this is used.

Flags: Read / Write / Construct Only

Default value: NULL

The “tls-certificate” property

  “tls-certificate”          GTlsCertificate *

A GTlsCertificate that has a “private-key” set. If this is set, then the server will be able to speak https in addition to (or instead of) plain http.

Alternatively, you can call soup_server_set_ssl_cert_file() to have SoupServer read in a a certificate from a file.

Flags: Read / Write / Construct Only

Since: 2.38

The “request-aborted” signal

void
user_function (SoupServer        *server,
               SoupMessage       *message,
               SoupClientContext *client,
               gpointer           user_data)

Emitted when processing has failed for a message; this could mean either that it could not be read (if “request_read” has not been emitted for it yet), or that the response could not be written back (if “request_read” has been emitted but “request_finished” has not been).

message is in an undefined state when this signal is emitted; the signal exists primarily to allow the server to free any state that it may have allocated in “request_started”.

Flags: Run First

The “request-finished” signal

void
user_function (SoupServer        *server,
               SoupMessage       *message,
               SoupClientContext *client,
               gpointer           user_data)

Emitted when the server has finished writing a response to a request.

Flags: Run First

The “request-read” signal

void
user_function (SoupServer        *server,
               SoupMessage       *message,
               SoupClientContext *client,
               gpointer           user_data)

Emitted when the server has successfully read a request. message will have all of its request-side information filled in, and if the message was authenticated, client will have information about that. This signal is emitted before any (non-early) handlers are called for the message, and if it sets the message's status_code, then normal handler processing will be skipped.

Flags: Run First

The “request-started” signal

void
user_function (SoupServer        *server,
               SoupMessage       *message,
               SoupClientContext *client,
               gpointer           user_data)

Emitted when the server has started reading a new request. message will be completely blank; not even the Request-Line will have been read yet. About the only thing you can usefully do with it is connect to its signals.

If the request is read successfully, this will eventually be followed by a “request_read” signal. If a response is then sent, the request processing will end with a “request_finished” signal. If a network error occurs, the processing will instead end with “request_aborted”.

Flags: Run First