This project has been created as part of the 42 curriculum by pjarnac, kporceil.

Webserv

Description

Webserv is an HTTP/1.0 web server with some features of 1.1 written in C++98, developed as part of the 42 school curriculum. The goal of this project is to implement a fully functional web server from scratch, capable of serving static websites, handling file uploads, executing CGI scripts, and managing multiple servers through a configuration file inspired by NGINX.

The server uses a single epoll instance for all I/O multiplexing, ensuring non-blocking operation across all client connections, listener sockets, and CGI pipes. It supports the GET, POST, and DELETE HTTP methods.

Key Features

• Non-blocking I/O — All socket and pipe operations are driven by a single epoll event loop.
• Static file serving — Serve HTML, CSS, JS, images, and other static content.
• Directory listing — Optional autoindex for browsing directory contents.
• File uploads — Clients can upload files to configurable storage locations.
• CGI execution — Execute scripts (e.g., Python) based on file extension, with full environment variable support.
• Session management — Cookie-based session tracking with configurable timeouts (bonus).
• Multiple servers — Listen on multiple interface:port pairs, each with independent configuration.
• Custom error pages — Define custom error pages per server.
• NGINX-inspired configuration — Familiar syntax with server blocks, location blocks, and directives.

Instructions

Prerequisites

• A Linux operating system (the server uses epoll).
• A C++98 compatible compiler (c++ / g++ / clang++).
• make build tool.

Compilation

# Clone the repository
git clone https://forgejo.asventi.net/webserv/webserv.git
cd webserv

# Build the server
make

The Makefile supports the following targets:

Target	Description
all	Build the server (default)
clean	Remove object files
fclean	Remove object files and the binary
re	Full rebuild
debug	Build with debug symbols
fsanitize	Build with AddressSanitizer
test	Build and run the test suite

Running

# Run with a configuration file
./webserv webserv.conf

# Run with the default configuration
./webserv

Configuration

The server is configured through a configuration file with NGINX-inspired syntax. See docs/CONFIG_FORMAT.md for the complete reference.

A minimal configuration example:

server {
    listen 0.0.0.0:8080;

    location / {
        root ./site;
        autoindex on;
        allow_methods GET POST DELETE;
    }
}

Main Directives

Directive	Context	Description
listen	server	Interface and port to listen on (required)
server_name	server	Virtual host name(s)
root	server, location	Document root directory
client_max_body_size	server, location	Maximum request body size (e.g., 10M)
error_page	server	Custom error page for HTTP status codes
autoindex	location	Enable/disable directory listing
allow_methods	location	Allowed HTTP methods (GET, POST, DELETE)
index	location	Default file for directory requests
cgi_pass	location	CGI interpreter mapping (e.g., .py /usr/bin/python3)
enable_upload	location	Enable file upload support
upload_dir	location	Directory for uploaded files
enable_session	location	Enable cookie-based session management

Testing

# Run the unit test suite
make test
./webserv

# Or test manually with curl
curl -v http://localhost:8080/
curl -X POST -F "file=@myfile.txt" http://localhost:8080/test
curl -X DELETE http://localhost:8080/test/myfile.txt

The project can also be tested with any standard web browser by navigating to the configured address (e.g., http://localhost:8080).

Architecture

The project is organized into the following modules:

src/
├── main.cpp                 # Entry point
├── ServerManager.cpp        # Main event loop and server orchestration
├── http/                    # HTTP request/response handling
│   ├── ARequest.cpp         #   Abstract request base
│   ├── Get.cpp              #   GET method handler
│   ├── Post.cpp             #   POST method handler
│   ├── Delete.cpp           #   DELETE method handler
│   ├── Response.cpp         #   HTTP response builder
│   ├── RequestFactory.cpp   #   Request creation factory
│   ├── CGIResponseFactory.cpp # CGI response handling
│   ├── RequestValidator.cpp #   Request validation
│   └── SessionManager.cpp   #   Cookie/session management
├── io/                      # I/O abstraction layer
│   ├── EpollManager.cpp     #   epoll wrapper
│   ├── AIOFile.cpp          #   Abstract I/O file descriptor
│   ├── IOSocket.cpp         #   Socket I/O wrapper
│   └── CGISocket.cpp        #   CGI pipe I/O wrapper
├── network/                 # Network socket management
│   ├── ANetworkSocket.cpp   #   Abstract network socket
│   ├── ListenerSocket.cpp   #   Server listener socket
│   └── ClientSocket.cpp     #   Client connection socket
├── parsing/                 # Configuration file parsing
│   ├── ConfigParser.cpp     #   Orchestrator
│   ├── Tokenizer.cpp        #   Lexical analysis
│   ├── Parser.cpp           #   Syntactic analysis (AST)
│   └── validators/          #   Semantic validation
├── server/
│   └── Server.cpp           # Server instance (binds config to sockets)
├── locations/
│   └── Location.cpp         # Route/location configuration
└── logger/
    └── Logger.cpp           # Logging utility

For details on the configuration parsing system, see docs/PARSING_ARCHITECTURE.md.

Resources

References

• RFC 2616 — HTTP/1.1 — The HTTP/1.1 specification.
• RFC 1945 — HTTP/1.0 — The HTTP/1.0 specification.
• RFC 3875 — The Common Gateway Interface (CGI) Version 1.1 — CGI specification.
• NGINX Documentation — Used as a reference for configuration syntax and HTTP behaviour.
• Beej's Guide to Network Programming — Socket programming fundamentals.
• The Linux epoll man page — epoll I/O event notification reference.

AI Usage

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

• Boilerplate reduction — Generating repetitive code patterns (e.g., getters/setters, canonical form implementations).
• Documentation — Assisting with the writing and formatting of documentation files (CONFIG_FORMAT.md, README.md).

All AI-generated content was reviewed, tested, before being integrated into the project.