I'd recommend that you take a look at: A Practical Guide to Writing Clients and Servers
What you have to implement in incremental steps is:
- Get your basic TCP sockets layer running (listen on port/ports, accept client connections and send/receive data).
- Implement a buffered reader so that you can read requests one line (delimited by CRLF) at a time.
- Read the very first line. Parse out the method, the request version and the path.
- Implement header parsing for the "Header: value" syntax. Don't forget unfolding folded headers.
- Check the request method, content type and content size to determine how/if the body will be read.
- Implement decoding of content based on content type.
- If you're going to support HTTP 1.1, implement things like "100 Continue", keep-alive, chunked transfer.
- Add robustness/security measures like detecting incomplete requests, limiting max number of clients etc.
- Shrink wrap your code and open-source it :)
I suggest you take a look at tiny httpd. If you want to write it from scratch, then you'll want to thoroughly read RFC 2616. Use BSD sockets to access the network at a really low level.
An HTTP server is conceptually simple:
- Open port 80 for listening
- When contact is made, gather a little information (get mainly - you can ignore the rest for now)
- Translate the request into a file request
- Open the file and spit it back at the client
It gets more difficult depending on how much of HTTP you want to support - POST is a little more complicated, scripts, handling multiple requests, etc.
But the base is very simple.
Mongoose (Formerly Simple HTTP Daemon) is pretty good. In particular, it's embeddable and compiles under Windows, Windows CE, and UNIX.
Menu
