devtools/hackdoc/main.go

package main

import (
	"context"
	"flag"
	"fmt"
	"net/http"
	"path/filepath"
	"regexp"
	"strings"
	"time"

	"code.hackerspace.pl/hscloud/go/mirko"
	"code.hackerspace.pl/hscloud/go/pki"
	"github.com/golang/glog"
	"google.golang.org/grpc"

	dvpb "code.hackerspace.pl/hscloud/devtools/depotview/proto"
	"code.hackerspace.pl/hscloud/devtools/hackdoc/config"
	"code.hackerspace.pl/hscloud/devtools/hackdoc/source"
)

var (
	flagListen              = "127.0.0.1:8080"
	flagDocRoot             = ""
	flagDepotViewAddress    = ""
	flagHackdocURL          = ""
	flagGitwebDefaultBranch = "master"

	rePagePath = regexp.MustCompile(`^/([A-Za-z0-9_\-/\. ]*)$`)
)

func init() {
	flag.Set("logtostderr", "true")
}

func main() {
	flag.StringVar(&flagListen, "pub_listen", flagListen, "Address to listen on for HTTP traffic")
	flag.StringVar(&flagDocRoot, "docroot", flagDocRoot, "Path from which to serve documents. Either this or depotview must be set")
	flag.StringVar(&flagDepotViewAddress, "depotview", flagDepotViewAddress, "gRPC endpoint of depotview to serve from Git. Either this or docroot must be set")
	flag.StringVar(&flagHackdocURL, "hackdoc_url", flagHackdocURL, "Public URL of hackdoc. If not given, autogenerate from listen path for dev purposes")
	flag.StringVar(&source.FlagGitwebURLPattern, "gitweb_url_pattern", source.FlagGitwebURLPattern, "Pattern to sprintf to for URL for viewing a file in Git. First string is ref/rev, second is bare file path (sans //)")
	flag.StringVar(&flagGitwebDefaultBranch, "gitweb_default_rev", flagGitwebDefaultBranch, "Default Git rev to render/link to")
	flag.Parse()

	if flagHackdocURL == "" {
		flagHackdocURL = fmt.Sprintf("http://%s", flagListen)
	}

	if flagDocRoot == "" && flagDepotViewAddress == "" {
		glog.Errorf("Either -docroot or -depotview must be set")
	}
	if flagDocRoot != "" && flagDepotViewAddress != "" {
		glog.Errorf("Only one of -docroot or -depotview must be set")
	}

	m := mirko.New()
	if err := m.Listen(); err != nil {
		glog.Exitf("Listen(): %v", err)
	}

	var s *service
	if flagDocRoot != "" {
		path, err := filepath.Abs(flagDocRoot)
		if err != nil {
			glog.Exitf("Could not dereference path %q: %w", path, err)
		}
		glog.Infof("Starting in docroot mode for %q -> %q", flagDocRoot, path)

		s = &service{
			source: source.NewSingleRefProvider(source.NewLocal(path)),
		}
	} else {
		glog.Infof("Starting in depotview mode (server %q)", flagDepotViewAddress)
		conn, err := grpc.Dial(flagDepotViewAddress, pki.WithClientHSPKI())
		if err != nil {
			glog.Exitf("grpc.Dial(%q): %v", flagDepotViewAddress, err)
		}
		stub := dvpb.NewDepotViewClient(conn)
		s = &service{
			source: source.NewDepotView(stub),
		}
	}

	mux := http.NewServeMux()
	mux.HandleFunc("/", s.handler)
	srv := &http.Server{Addr: flagListen, Handler: mux}

	glog.Infof("Listening on %q...", flagListen)
	go func() {
		if err := srv.ListenAndServe(); err != nil {
			glog.Error(err)
		}
	}()

	<-m.Done()
	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
	defer cancel()
	srv.Shutdown(ctx)

}

type service struct {
	source source.SourceProvider
}

func (s *service) handler(w http.ResponseWriter, r *http.Request) {
	if r.Method != "GET" && r.Method != "HEAD" {
		w.WriteHeader(http.StatusMethodNotAllowed)
		fmt.Fprintf(w, "method not allowed")
		return
	}

	ref := r.URL.Query().Get("ref")
	if ref == "" {
		ref = flagGitwebDefaultBranch
	}

	ctx := r.Context()
	source, err := s.source.Source(ctx, ref)
	switch {
	case err != nil:
		glog.Errorf("Source(%q): %v", ref, err)
		handle500(w, r)
		return
	case source == nil:
		handle404(w, r)
		return
	}

	path := r.URL.Path

	if match := rePagePath.FindStringSubmatch(path); match != nil {
		req := &request{
			w:      w,
			r:      r,
			ctx:    r.Context(),
			ref:    ref,
			source: source,
		}
		req.handlePage(match[1])
		return
	}
	handle404(w, r)
}

type request struct {
	w   http.ResponseWriter
	r   *http.Request
	ctx context.Context

	ref    string
	source source.Source
	// rpath is the path requested by the client
	rpath string
}

func (r *request) handle500() {
	handle500(r.w, r.r)
}

func (r *request) handle404() {
	handle404(r.w, r.r)
}

func (r *request) logRequest(format string, args ...interface{}) {
	logRequest(r.w, r.r, format, args...)
}

func urlPathToDepotPath(url string) string {
	// Sanitize request.
	parts := strings.Split(url, "/")
	for i, p := range parts {
		// Allow last part to be "", ie, for a path to end in /
		if p == "" {
			if i != len(parts)-1 {
				return ""
			}
		}

		// net/http sanitizes this anyway, but we better be sure.
		if p == "." || p == ".." {
			return ""
		}
	}
	path := "//" + strings.Join(parts, "/")

	return path
}

func (r *request) handlePageAuto(dirpath string) {
	cfg, err := config.ForPath(r.ctx, r.source, dirpath)
	if err != nil {
		glog.Errorf("could not get config for path %q: %w", dirpath, err)
		r.handle500()
		return
	}
	for _, f := range cfg.DefaultIndex {
		fpath := dirpath + f
		file, err := r.source.IsFile(r.ctx, fpath)
		if err != nil {
			glog.Errorf("IsFile(%q): %w", fpath, err)
			r.handle500()
			return
		}

		if file {
			ref := r.ref
			if ref == flagGitwebDefaultBranch {
				ref = ""
			}
			path := "/" + fpath
			if ref != "" {
				path += "?ref=" + ref
			}
			http.Redirect(r.w, r.r, path, 302)
			return
		}
	}
	r.handle404()
}

func (r *request) handlePage(page string) {
	r.rpath = urlPathToDepotPath(page)

	if strings.HasSuffix(r.rpath, "/") {
		// Directory path given, autoresolve.
		dirpath := r.rpath
		if r.rpath != "//" {
			dirpath = strings.TrimSuffix(r.rpath, "/") + "/"
		}
		r.handlePageAuto(dirpath)
		return
	}

	// Otherwise, try loading the file.
	file, err := r.source.IsFile(r.ctx, r.rpath)
	if err != nil {
		glog.Errorf("IsFile(%q): %w", r.rpath, err)
		r.handle500()
		return
	}

	// File exists, render that.
	if file {
		parts := strings.Split(r.rpath, "/")
		dirpath := strings.Join(parts[:(len(parts)-1)], "/")
		// TODO(q3k): figure out this hack, hopefully by implementing a real path type
		if dirpath == "/" {
			dirpath = "//"
		}

		cfg, err := config.ForPath(r.ctx, r.source, dirpath)
		if err != nil {
			glog.Errorf("could not get config for path %q: %w", dirpath, err)
			r.handle500()
			return
		}
		r.handleFile(r.rpath, cfg)
		return
	}

	// Otherwise assume directory, try all posibilities.
	dirpath := r.rpath
	if r.rpath != "//" {
		dirpath = strings.TrimSuffix(r.rpath, "/") + "/"
	}
	r.handlePageAuto(dirpath)
}