Whole document tree

Whole document tree

gHTTP library



The HTTP protocol is the protocol that is the underlying transport mechanism for the modern world wide web. The protocol is well documented and widely implemented. While the use of the protocol is generally associated with the web, its uses can be extended to many other areas where a stateless, transactional protocol may be appropriate.

This library is fully compliant with HTTP 1.1 as defined in the draft 5 update of RFC 2068.

The gHTTP library is designed to be simple and easy to use while still allowing you to get your feet wet at the protocol layer if you have to. It is also designed with graphical, non-threaded applications in mind. You should be able to use the library in your application and never block waiting to send or recieve data to a remote server. The main thread of execution should always be available to refresh its display.

A simple example

Here's a simple example to get you started. This snippit of code shows how you can go out and get a file off of an HTTP server. Please note that this code contains no error checking at all.

/* This is the http request object */
ghttp_request *request = NULL;
/* Allocate a new empty request object */
request = ghttp_request_new();
/* Set the URI for the request object */
ghttp_set_uri(request, "http://localhost:8080/index.html");
/* Close the connection after you are done. */
ghttp_set_header(request, http_hdr_Connection, "close");
/* Prepare the connection */
/* Process the request */
/* Write out the body. Note that the body of the request may not be null terminated so we have to be careful of the length. */
fwrite(ghttp_get_body(request), ghttp_get_body_len(request), 1, stdout);
/* Destroy the request. This closes any file descriptors that may be open and will free any memory associated with the request. */

As you can see, you can get a lot of work done with very few function calls.

The flow of a HTTP request

HTTP requests, as implemented in gHTTP follow a very simple pattern. First, you must allocate a new HTTP request object using the call ghttp_request_new(). The request object is what you will use to set properties for your request and how you will retrieve the response to your requests.

Once you have allocated the request object you will need to set certian properties before actually making the request.

At a minimum, you must set the URI that the request is for. You can do this using the call ghttp_set_uri(). This function takes the request object and the uri that is the target of the request. You can use any http URI as defined by RFC 2396. gHTTP does not do any other protocols other than HTTP.

For example, the following are examples of valid http URI's:


There are a lot of other properties that you can set on a request. You can set specific headers in the request using ghttp_set_header(), you can set the http URI of your proxy host using ghttp_set_proxy(), or you can set the body of the HTTP request that you are making using ghttp_set_body(). Other properties that you can set on a transaction are documented below in the section Request Properties.

Once you have set all of the properties on your request, you must call ghttp_prepare() on the request. This will do a hostname lookup on the host that the resource is for and make some other internal preperations.

Now, you are ready to make the request. Do complete the request, simply call ghttp_process(). This will connect to the remote server and try to complete your transaction. If you have set the library to use asynchronous operation ( see Request Properties below ) you may need to call this more than once until the transaction is complete.

Now that you have completed this transaction you can do several things. If you want to access another resource that is on the same server you can change the URI and use the same request object again. For example if you have just completed a transaction getting the URI http://localhost/index.html and now you want to get http://localhost/image.gif you should call ghttp_set_uri() to set the request object to the new URI, call ghttp_prepare() to propagate any changes, ghttp_clean() to clean out any old buffers, and then call ghttp_process() to run your request. This has the advantage of using an open connection if it is available and DNS information if it has already been looked up.

Once you have completed a request you can use several functions to get the data from the response. You can examine a specific header in the response using ghttp_get_header(). You can get a handle to the body of the data using ghttp_get_body() and the associated length of the body using ghttp_get_body_len(). Remember that the data in the body of a response is not null-terminated as it can be data of any type.

Additionally, you can get the actual code that was given in the response using ghttp_status_code() and ghttp_reason_phrase().

Request Properties

Christopher Blizzard
Last modified: Sat Nov 14 04:07:56 XXX 1998