pollux/AGENTS.md
Jeena 0468781a69 Add configurable global concurrent request limiting
- Add max_concurrent_requests config option (default: 1000)
- Implement global AtomicUsize counter for concurrent request tracking
- Return status 41 'Server unavailable' when limit exceeded
- Proper counter management with decrements on all exit paths
- Add comprehensive config validation (1-1,000,000 range)
- Update documentation with rate limiting details
- Add unit tests for config parsing
- Thread-safe implementation using Ordering::Relaxed

This provides effective DDoS protection by limiting concurrent
connections to prevent server overload while maintaining
configurability for different deployment scenarios.
2026-01-16 02:26:59 +00:00

128 lines
5.3 KiB
Markdown

# Overview
This project is a very simple gemini server which only serves static files,
nothing else. It is meant to be generic so other people can use it.
# Build/Test/Lint Commands
## Core Commands
- `cargo build` - Build the project
- `cargo build --release` - Build optimized release version
- `cargo run` - Run the server with default config
- `cargo test` - Run all unit tests
- `cargo test <test_name>` - Run a specific test
- `cargo test <module>::tests` - Run tests in a specific module
- `cargo clippy` - Run linter checks for code quality
- `cargo clippy --fix` - Automatically fix clippy suggestions
- `cargo clippy --bin <name>` - Check specific binary
- `cargo fmt` - Format code according to Rust standards
- `cargo check` - Quick compile check without building
## Common Test Patterns
- `cargo test config::tests` - Run config module tests
- `cargo test request::tests` - Run request handling tests
- `cargo test -- --nocapture` - Show println output in tests
# Code Style Guidelines
## Imports
- Group imports: std libs first, then external crates, then local modules
- Use `use crate::module::function` for internal imports
- Prefer specific imports over `use std::prelude::*`
- Keep imports at module level, not inside functions
## Code Structure
- Use `#[tokio::main]` for async main function
- Keep functions small and focused (single responsibility)
- Use `const` for configuration values that don't change
- Error handling with `Result<T, E>` and `?` operator
- Use `tracing` for logging, not `println!` in production code
## Naming Conventions
- `PascalCase` for types, structs, enums
- `snake_case` for functions, variables, modules
- `SCREAMING_SNAKE_CASE` for constants
- Use descriptive names that indicate purpose
## Error Handling
- Use `io::Result<()>` for I/O operations
- Convert errors to appropriate types with `map_err` when needed
- Use `unwrap()` only in tests and main() for unrecoverable errors
- Use `expect()` with meaningful messages for debugging
- Return early with `Err()` for validation failures
## Security Requirements
- **Critical**: Always validate file paths with `path_security::validate_path`
- Never construct paths from user input without validation
- Use timeouts for network operations (`tokio::time::timeout`)
- Limit request sizes (see `MAX_REQUEST_SIZE` constant)
- Validate TLS certificates properly
- Never expose directory listings
## Testing Guidelines
- Use `tempfile::TempDir` for temporary directories in tests
- Test both success and error paths
- Use `#[cfg(test)]` for test modules
- Create temporary test files in `tmp/` directory
- Test security boundaries (path traversal, invalid inputs)
- Use `assert_eq!` and `assert!` for validations
## Lint Checking
- `cargo clippy` - Run linter checks for code quality
- `cargo clippy --fix` - Automatically fix clippy suggestions
- `cargo clippy --bin <name>` - Check specific binary
- `cargo fmt` - Format code to match Rust standards
- **Run clippy before every commit** - Address all warnings before pushing code
- Current clippy warnings (2025-01-15):
- src/server.rs:16-17 - Unnecessary borrows on file_path
- src/logging.rs:31 - Match could be simplified to let statement
## Async Patterns
- Use `.await` on async calls
- Prefer `tokio::fs` over `std::fs` in async contexts
- Handle timeouts for network operations
- Use `Arc<Clone>` for shared data across tasks
## Gemini Protocol Specific
- Response format: "STATUS META\r\n"
- Status 20: Success (follow with MIME type)
- Status 41: Server unavailable (timeout, overload)
- Status 51: Not found (resource doesn't exist)
- Status 59: Bad request (malformed URL, protocol violation)
- Default MIME: "text/gemini" for .gmi files
- Default file: "index.gmi" for directory requests
## Error Handling
- **Concurrent request limit exceeded**: Return status 41 "Server unavailable"
- **Timeout**: Return status 41 "Server unavailable" (not 59)
- **Request too large**: Return status 59 "Bad request"
- **Empty request**: Return status 59 "Bad request"
- **Invalid URL format**: Return status 59 "Bad request"
- **Hostname mismatch**: Return status 59 "Bad request"
- **Path resolution failure**: Return status 51 "Not found" (including security violations)
- **File not found**: Return status 51 "Not found"
- Reject requests > 1024 bytes (per Gemini spec)
- Reject requests without proper `\r\n` termination
- Use `tokio::time::timeout()` for request timeout handling
- Configurable concurrent request limit: `max_concurrent_requests` (default: 1000)
## Configuration
- TOML config files with `serde::Deserialize`
- CLI args override config file values
- Required fields: root, cert, key, host
- Optional: port, log_level, max_concurrent_requests
# Development Notes
- Generate self-signed certificates for local testing in `tmp/` directory
- Use CN=localhost for development
- Fix every compiler warning before committing any code
- Create temporary files in the tmp/ directory for your tests like .gem files
or images, etc., so they are gitignored
- Use `path-security` crate for path validation
- Default port: 1965 (standard Gemini port)
- Default host: 0.0.0.0 for listening
- Log level defaults to "info"
## Environment Setup
- Install clippy: `rustup component add clippy`
- Ensure `~/.cargo/bin` is in PATH (add `source "$HOME/.cargo/env"` to `~/.bashrc`)
- Verify setup: `cargo clippy --version`