An HTTP server is a complicated program. It must handle multiple simultaneous connections, parse a complex text-based protocol, handle malformed requests with the proper errors, and serve files. The example we develop in this chapter is greatly simplified from a production-ready server, but it is still a few hundred lines of code. We benefit from breaking the program down into separate functions and data structures.
At the global level, our program stores a linked list of data structures. This linked list contains one separate data structure for each connected client. This data structure stores information about each client such as their address, their socket, and their data received so far. We implement many helper functions that work on this global linked list. These functions are used to add new clients, drop clients, wait on client data, look up clients by their socket (as sockets are returned by select()), serve files to clients, and send error messages to clients.
Our server's main loop can then be simplified. It waits for new connections or new data. When new data is received, it checks whether the data consists of a complete HTTP request. If a complete HTTP request is received, the server attempts to send the requested resource. If the HTTP request is malformed or the resource cannot be found, then the server sends an error message to the connected client instead.
Most of the server complexity lies in handling multiple connections, parsing the HTTP request, and handling error conditions.
The server is also responsible for telling the client the content type of each resource it sends. There are a few ways to accomplish this; let's consider them next.