job-search/internal/server/server.go

// Package server assembles the HTTP server for the serve command.
package server

import (
	"context"
	"crypto/tls"
	"errors"
	"fmt"
	"log"
	"log/slog"
	"net"
	"net/http"
	"net/netip"
	"strconv"
	"strings"
	"time"

	"github.com/pires/go-proxyproto"

	"git.avsdev.uk/Go/job-search/internal/api"
	"git.avsdev.uk/Go/job-search/internal/config"
	"git.avsdev.uk/Go/job-search/internal/events"
	"git.avsdev.uk/Go/job-search/internal/logging"
	"git.avsdev.uk/Go/job-search/internal/middleware"
	"git.avsdev.uk/Go/job-search/internal/store"
	"git.avsdev.uk/Go/job-search/internal/util"
	"git.avsdev.uk/Go/job-search/internal/websocket"
)

// Group identifies where hook middleware is inserted.
type Group int

// The route groups hook middleware can be added to (see WithMiddleware).
const (
	// GroupGlobal is every request, after real-IP, request-ID and logging.
	GroupGlobal Group = iota
	// GroupAPI is /api/*, before CSRF protection.
	GroupAPI
	// GroupWebSocket is /ws upgrade requests.
	GroupWebSocket
)

// Middleware is the standard net/http middleware shape.
type Middleware = func(http.Handler) http.Handler

// Option customises a Server.
type Option func(*Server)

// WithLogger makes the server, its middleware and its WebSocket hub log to
// logger instead of slog.Default().
func WithLogger(logger *slog.Logger) Option {
	return func(s *Server) {
		s.log = logger
	}
}

// WithAccessLog makes the server write a line to logger for every request, in
// the Combined Log Format (see logging.AccessLog), as well as the request
// lines it logs with slog. The line is written when the request ends, and
// covers requests that middleware refuses.
func WithAccessLog(logger *log.Logger) Option {
	return func(s *Server) {
		s.accessLog = logger
	}
}

// WithStore makes the server's API serve the opportunities in `store` under
// /api/opportunities (see api.WithStore). Without it there are no
// opportunity routes.
func WithStore(store *store.Store) Option {
	return func(s *Server) {
		s.store = store
	}
}

// WithEvents connects the server to bus: every event published to it is
// pushed to the server's WebSocket clients (as websocket.TypeEvent
// messages), and the server's API publishes its changes to it (see
// api.WithEvents).
func WithEvents(bus *events.Bus) Option {
	return func(s *Server) {
		s.events = bus
	}
}

// WithMiddleware appends middleware to a group. This is the generic hook
// (auth, metrics, tracing, custom header rules, ...).
func WithMiddleware(group Group, middlewares ...Middleware) Option {
	return func(server *Server) {
		server.hooks[group] = append(server.hooks[group], middlewares...)
	}
}

// WithRateLimiter replaces the configured rate limiter of a group (for
// example with a Redis-backed implementation). Pass a nil Limiter to
// disable limiting for the group. A nil key defaults to the client IP.
func WithRateLimiter(
	group Group, limiter middleware.Limiter, key middleware.KeyFunc,
) Option {
	return func(server *Server) {
		server.limiters[group] = limiterHook{
			limiter: limiter,
			key:     key,
			set:     true,
		}
	}
}

// limiterHook is a group's rate limiter set by WithRateLimiter, replacing the
// configured one. set distinguishes a nil limiter (no limiting) from none.
type limiterHook struct {
	limiter middleware.Limiter
	key     middleware.KeyFunc
	set     bool
}

// Server is the application server.
type Server struct {
	conf *config.JSCServer
	log  *slog.Logger
	// accessLog is where the access log is written, or nil for none.
	accessLog *log.Logger
	hooks     map[Group][]Middleware
	limiters  map[Group]limiterHook

	// trusted are the proxies whose forwarding headers are believed.
	trusted []netip.Prefix
	// proxyProtocolTrusted are the sources allowed to send a PROXY header.
	proxyProtocolTrusted []netip.Prefix
	csrf                 *middleware.CSRF
	hub                  *websocket.Hub
	// store is the opportunities' database, or nil for none.
	store *store.Store
	// events is where events are pushed from and the API publishes to, or
	// nil for none.
	events *events.Bus
	// api is the REST API mounted under /api, built from the options.
	api     *api.API
	started time.Time
	handler http.Handler
	// backgroundCtx is for background work, such as the rate limiters'
	// janitors; backgroundCancel stops it.
	backgroundCtx    context.Context
	backgroundCancel context.CancelFunc
}

// New builds the server and its router. It logs to slog.Default() unless
// given WithLogger.
func New(conf *config.JSCServer, opts ...Option) (*Server, error) {
	server := &Server{
		conf:     conf,
		hooks:    make(map[Group][]Middleware),
		limiters: make(map[Group]limiterHook),
		started:  time.Now(),
	}
	for _, option := range opts {
		option(server)
	}
	server.api = api.New(
		api.WithStore(server.store),
		api.WithEvents(server.events),
	)

	var err error
	server.trusted, err = util.ParseNetPrefixes(conf.Proxy.TrustedProxies)
	if err != nil {
		return nil, fmt.Errorf("parsing proxy.trusted: %w", err)
	}

	proxyProtocolSources := conf.Proxy.ProxyProtocol.TrustedSources
	if len(proxyProtocolSources) == 0 {
		proxyProtocolSources = conf.Proxy.TrustedProxies
	}
	server.proxyProtocolTrusted, err = util.ParseNetPrefixes(
		proxyProtocolSources,
	)
	if err != nil {
		return nil, fmt.Errorf(
			"parsing proxy.proxy_protocol.trusted_sources: %w", err,
		)
	}

	server.csrf, err = middleware.NewCSRF(
		conf.CSRF, middleware.WithLogger(server.untaggedLogger()),
	)
	if err != nil {
		return nil, err
	}

	server.hub = websocket.NewHub(
		conf.WebSocket,
		websocket.WithLogger(server.untaggedLogger()),
		websocket.WithTrustedOrigins(
			conf.CSRF.CrossOriginProtection.TrustedOrigins,
		),
	)

	server.backgroundCtx, server.backgroundCancel = context.WithCancel(
		context.Background(),
	)
	server.handler, err = server.routes()
	if err != nil {
		server.backgroundCancel()
		return nil, err
	}
	if server.accessLog != nil {
		server.handler = logging.AccessLog(server.accessLog, true, server.handler)
	}

	// Subscribed last, so a server that failed to build is never pushed to.
	if server.events != nil {
		server.events.Subscribe(func(event events.Event) {
			server.hub.PushEvent(event.Type, event.Payload)
		})
	}

	return server, nil
}

// untaggedLogger returns the logger given WithLogger, or slog.Default(), for
// the server's middleware and hub to tag with their own components.
func (s *Server) untaggedLogger() *slog.Logger {
	if s.log == nil {
		return slog.Default()
	}
	return s.log
}

// logger returns the logger the server logs to, tagged with its component.
func (s *Server) logger() *slog.Logger {
	return s.untaggedLogger().With("component", "server")
}

// Handler exposes the root handler (useful for httptest).
func (s *Server) Handler() http.Handler { return s.handler }

// Run listens and serves until ctx is cancelled, then shuts down gracefully.
// With TLS, it also serves an HTTP to HTTPS redirect if tls.redirect_http_port
// is set.
func (s *Server) Run(ctx context.Context) error {
	defer s.backgroundCancel()
	conf := s.conf

	addr := net.JoinHostPort(conf.Host, strconv.Itoa(conf.Port))

	listen, err := net.Listen("tcp", addr)
	if err != nil {
		return fmt.Errorf("listen %s: %w", addr, err)
	}
	if conf.Proxy.ProxyProtocol.Enabled {
		listen = &proxyproto.Listener{
			Listener:   listen,
			ConnPolicy: s.proxyPolicy,
			ReadHeaderTimeout: noLimitIfUnset(
				conf.Proxy.ProxyProtocol.ReadHeaderTimeout,
			),
		}
	}

	httpServer := &http.Server{
		Handler:           s.handler,
		ReadHeaderTimeout: noLimitIfUnset(conf.Timeouts.ReadHeaderTimeout),
		ReadTimeout:       conf.Timeouts.ReadTimeout,
		WriteTimeout:      conf.Timeouts.WriteTimeout,
		IdleTimeout:       noLimitIfUnset(conf.Timeouts.IdleTimeout),
		MaxHeaderBytes:    maxHeaderBytes(conf),
		ErrorLog: slog.NewLogLogger(
			s.logger().Handler(), slog.LevelWarn,
		),
	}

	if conf.TLS.Enabled {
		minVersion := uint16(tls.VersionTLS12)
		if conf.TLS.MinVersion == "1.3" {
			minVersion = tls.VersionTLS13
		}
		httpServer.TLSConfig = &tls.Config{
			MinVersion: minVersion,
		}
	}

	serveErrors := make(chan error, 2)
	go func() {
		s.logger().Info(
			"listening",
			"addr", listen.Addr().String(),
			"tls", conf.TLS.Enabled,
			"proxy_protocol", conf.Proxy.ProxyProtocol.Enabled,
		)
		var err error
		if conf.TLS.Enabled {
			err = httpServer.ServeTLS(listen, conf.TLS.CertFile, conf.TLS.KeyFile)
		} else {
			err = httpServer.Serve(listen)
		}
		serveErrors <- err
	}()

	var redirect *http.Server
	if conf.TLS.Enabled && conf.TLS.RedirectHTTPPort != 0 {
		httpAddr := net.JoinHostPort(
			conf.Host, strconv.Itoa(conf.TLS.RedirectHTTPPort),
		)

		redirect = &http.Server{
			Addr:              httpAddr,
			Handler:           middleware.RequestChecks(*conf)(httpsRedirect(addr)),
			ReadHeaderTimeout: noLimitIfUnset(conf.Timeouts.ReadHeaderTimeout),
			IdleTimeout:       noLimitIfUnset(conf.Timeouts.IdleTimeout),
		}
		go func() {
			s.logger().Info("http->https redirect listening", "addr", httpAddr)
			serveErrors <- redirect.ListenAndServe()
		}()
	}

	select {
	case err := <-serveErrors:
		if !errors.Is(err, http.ErrServerClosed) {
			return err
		}
	case <-ctx.Done():
	}

	s.logger().Info("shutting down")
	shutdownCtx, cancel := util.WithTimeout(
		context.Background(), conf.Timeouts.ShutdownTimeout,
	)
	defer cancel()

	// Stop accepting first, then close WebSockets (hijacked connections are
	// not tracked by http.Server.Shutdown), then drain in-flight requests.
	var errs []error
	shutdownErr := make(chan error, 1)
	go func() {
		shutdownErr <- httpServer.Shutdown(shutdownCtx)
	}()
	if err := s.hub.Shutdown(shutdownCtx); err != nil {
		errs = append(errs, fmt.Errorf("websocket shutdown: %w", err))
	}
	if err := <-shutdownErr; err != nil {
		errs = append(errs, fmt.Errorf("http shutdown: %w", err))
	}
	if redirect != nil {
		redirect.Shutdown(shutdownCtx) //nolint:errcheck,gosec // only redirects; nothing to drain
	}

	return errors.Join(errs...)
}

// noLimitIfUnset returns timeout for a field that reads 0 as some other
// timeout, such as http.Server's ReadHeaderTimeout and IdleTimeout (which
// use ReadTimeout) or the PROXY listener's ReadHeaderTimeout (10s): -1,
// which those fields read as no limit, if timeout is 0 or less, as the
// configuration's timeouts mean.
func noLimitIfUnset(timeout time.Duration) time.Duration {
	if timeout <= 0 {
		return -1
	}
	return timeout
}

// defaultMaxHeaderBytes is the largest request header block when
// max_header_bytes is 0 or less: 8KB, as the configuration's default. The
// configuration rejects -1 (no limit), as http.Server can't have one.
const defaultMaxHeaderBytes = 8_000

// maxHeaderBytes is the largest request header block: max_header_bytes, or
// defaultMaxHeaderBytes if that is 0 or less.
func maxHeaderBytes(conf *config.JSCServer) int {
	if conf.MaxHeaderBytes <= 0 {
		return defaultMaxHeaderBytes
	}
	return int(conf.MaxHeaderBytes)
}

// proxyPolicy decides per connection how to treat a PROXY header. Only
// trusted sources may supply one. Untrusted peers are SKIPped (the header is
// not parsed; a spoofed one becomes garbage to the HTTP parser and fails).
// Never return a non-ErrInvalidUpstream error here: it would stop Accept.
func (s *Server) proxyPolicy(
	options proxyproto.ConnPolicyOptions,
) (proxyproto.Policy, error) {
	ip := addrOf(options.Upstream)
	for _, prefix := range s.proxyProtocolTrusted {
		if ip.IsValid() && prefix.Contains(ip) {
			if s.conf.Proxy.ProxyProtocol.Required {
				return proxyproto.REQUIRE, nil
			}
			return proxyproto.USE, nil
		}
	}
	return proxyproto.SKIP, nil
}

// addrOf returns the IP address of a connection's address (unmapped to IPv4
// if it is an IPv4-mapped IPv6 address), or the zero Addr if it has none.
func addrOf(netAddr net.Addr) netip.Addr {
	switch tcpAddr := netAddr.(type) {
	case *net.TCPAddr:
		ip, _ := netip.AddrFromSlice(tcpAddr.IP)
		return ip.Unmap()
	default:
		addrPort, err := netip.ParseAddrPort(netAddr.String())
		if err != nil {
			return netip.Addr{}
		}
		return addrPort.Addr().Unmap()
	}
}

// httpsRedirect redirects every request permanently to the same host and
// path over HTTPS, on the port of tlsAddr (the TLS listener's address),
// which is left out if it is 443.
func httpsRedirect(tlsAddr string) http.Handler {
	_, port, _ := net.SplitHostPort(tlsAddr)
	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		host := r.Host
		if hostname, _, err := net.SplitHostPort(host); err == nil {
			host = hostname
		}
		if strings.Contains(host, ":") { // bare IPv6
			host = "[" + host + "]"
		}
		if port != "" && port != "443" {
			host = net.JoinHostPort(strings.Trim(host, "[]"), port)
		}
		// The redirect only switches the scheme (and port) of the URL the
		// client asked for, on the host it asked for, so it can't send the
		// client anywhere else; RequestChecks has checked that host.
		http.Redirect( //nolint:gosec // G710: see above
			w, r, "https://"+host+r.URL.RequestURI(), http.StatusPermanentRedirect,
		)
	})
}