This project has been created as part of the 42 curriculum by yingzhan, tthajan.

webserv

Description

webserv is a fully compliant HTTP/1.1 web server written in C++98, built from scratch as part of the 42 school curriculum. The goal is to deepen understanding of how web servers work at the network level — from raw TCP socket management to HTTP request parsing, response building, and CGI execution.

The server is designed around a non-blocking I/O event loop using poll(), enabling it to handle multiple simultaneous client connections without threads. It supports virtual hosting (multiple server {} blocks), configurable routes, file uploads, directory listing, custom error pages, HTTP redirects, and a multi-type CGI interface.

Key Features

Feature	Details
HTTP/1.1	GET, POST, DELETE methods; persistent connections
Non-blocking I/O	Single-thread poll()-based event loop
Virtual hosts	Multiple server {} blocks, matched by port and Host header
Static files	Serve files, directory autoindex, configurable root
File uploads	POST to upload, DELETE to remove, per-route body size limits
Custom error pages	Per-server error_page directives (400, 403, 404, 405, 413, 500, 501, 504)
Redirects	301/302/307 via return directive
CGI	Supports .py, .sh, .php, .pl — extensible via config
Chunked transfer	Chunked request body decoding
Config file	NGINX-inspired syntax with full validation

---

Instructions

Requirements

• Linux / macOS (poll() — POSIX-compliant)
• g++ / c++ with C++98 support
• python3 (for .py CGI scripts and tests)
• php-cgi (for .php CGI support, optional)
• perl (for .pl CGI support, optional)

Compilation

make

Produces the ./webserv binary. Use make clean / make fclean / make re as usual.

Running the Server

./webserv config/webserv.conf

The included config/webserv.conf starts two virtual hosts:

Host	Port	Notes
localhost	8080	Main site — 10 MB body limit, full CGI, uploads, redirects
api.localhost	8081	Minimal API — 512 KB body limit

Configuration File Format

server {
    listen 8080;
    server_name localhost;
    client_max_body_size 10M;

    error_page 404 ./www/errors/404.html;

    location / {
        methods GET;
        root ./www;
        index index.html;
        autoindex off;
    }

    location /cgi-bin/ {
        methods GET POST;
        root ./www/cgi-bin;
        cgi .py /usr/bin/python3;
        cgi .php /usr/bin/php-cgi;
    }

    location /old-page {
        return 301 /index.html;
    }
}

Running the Tests

# Full integration test suite (error pages, stress, scalability, redirects, CGI)
bash tests/full_test.sh

# CGI-focused unit test suite
python3 tests/cgi_test.py

See tests/TEST_GUIDE.md for a detailed description of every test case.

---

Resources

HTTP & Networking

• RFC 7230 — HTTP/1.1 Message Syntax
• RFC 3875 — The CGI/1.1 Specification
• Beej's Guide to Network Programming
• poll(2) Linux man page
• MDN HTTP documentation

CGI References

• php-cgi environment variables
• CGI tutorial (W3C)

AI Usage

AI tools (GitHub Copilot / ChatGPT) were used in this project for the following tasks:

• Understanding Concepts: Explaining network I/O, CGI environment variables, and HTTP protocol details.
• Debugging: Identifying misconfigurations and tracing logic errors in CGI handling.
• Test Scripting: Assisting in writing test scripts and fixing false-positive test cases.
• Documentation: Helping to draft and structure this README.

All AI-generated code and suggestions were reviewed, tested, and understood before being included in the project.