From 32c87fbc39bec28cfc0d7f49677ef06fc34d17f9 Mon Sep 17 00:00:00 2001
From: Erick Ruiz de Chavez <erick@fastmail.net>
Date: Fri, 10 Jan 2025 07:24:45 -0500
Subject: [PATCH] Add new flags for robots.txt and URL rewrites

---
 README.md |  20 ++++++-
 main.go   | 170 ++++++++++++++++++++++++++++++++++--------------------
 2 files changed, 124 insertions(+), 66 deletions(-)

diff --git a/README.md b/README.md
index 977abe5..4c474f4 100644
--- a/README.md
+++ b/README.md
@@ -6,15 +6,24 @@ can proxy to any domain in order to facilitate linking to the rest of
 Geminispace, but it defaults to a specific domain.
 
 [http]: https://en.wikipedia.org/wiki/Hypertext_Transfer_Protocol
-[gemini]: https://gemini.circumlunar.space/
+[gemini]: https://geminiprotocol.net/
 
 ## Usage
 
 ```
 $ go build
-$ ./kineto [-b 127.0.0.1:8080] [-s style.css] [-e style.css] [-p https://portal.mozz.us] gemini://example.org
+$ ./kineto [-P] [-R] [-b 127.0.0.1:8080] [-r /learn-more-about-gemini] [-s style.css] [-e style.css] [-p https://portal.mozz.us] gemini://example.org
 ```
 
+The -P argument is optional and allows you to completely disable proxying
+and URL rewriting external gemini URLs. The user visiting your site will need
+to have a browser or browser extension that understands how to handle
+gemini:// links.
+
+The -R argument is optional and allows you to completely disables serving a
+custom robots.txt response. When enabled the request will be proxied to your
+gemini site.
+
 The -b argument is optional and allows you to bind to an arbitrary address; by
 default kineto will bind to `:8080`. You should set up some external reverse
 proxy like nginx to forward traffic to this port and add TLS.
@@ -30,8 +39,13 @@ in a `<style>` block like with the -s flag. The given stylesheet can be a
 relative link, for instance `-e /main.css` will serve `main.css` from the root
 of the proxied Gemini capsule.
 
+The -r argument is optional and allows you rewrite external gemini links to a
+custom URL. Use this if you have external gemini links but instead of proxying
+them you prefer to invite your visitors to learn more about the Gemini protocol
+and download a Gemini browser. This can be a relative or absolute URL.
+
 The -p argument is optional and allows you to specify a custom Portal to proxy
-external gemini requests. This option will effectively disable the use of `/x/`
+external Gemini requests. This option will effectively disable the use of `/x/`
 for all external URLs, that is al URLs whose host does not match the proxied site.
 
 ## "kineto"?
diff --git a/main.go b/main.go
index 283d126..dfac8f7 100644
--- a/main.go
+++ b/main.go
@@ -108,28 +108,45 @@ var gemtextPage = template.Must(template.
 
 			u = ctx.URL.ResolveReference(u)
 
-			if u.Scheme == "" || u.Scheme == "gemini" {
-				if u.Host != ctx.Root.Host {
-					if ctx.UsePortal {
-						p, err := url.Parse(ctx.Portal)
-						if err != nil {
-							return template.URL("error")
-						}
-						p = ctx.URL.ResolveReference(p)
-						u.Path = fmt.Sprintf("%s/gemini/%s%s", p.Path, u.Host, u.Path)
-						u.Scheme = p.Scheme
-						u.Host = p.Host
-					} else {
-						u.Path = fmt.Sprintf("/x/%s%s", u.Host, u.Path)
-						u.Scheme = ""
-						u.Host = ""		
-					}
-				} else {
-					u.Scheme = ""
-					u.Host = ""	
-				}
+			if u.Scheme != "" && u.Scheme != "gemini" {
+				return template.URL(u.String())
 			}
 
+			if u.Host == ctx.Root.Host {
+				u.Scheme = ""
+				u.Host = ""
+				return template.URL(u.String())
+			}
+
+			if ctx.DisableExternalProxy {
+				return template.URL(u.String())
+			}
+
+			if ctx.UseRewrite {
+				r, err := url.Parse(ctx.Rewrite)
+				if err != nil {
+					return template.URL("error")
+				}
+				return template.URL(r.String())
+			}
+
+			if ctx.UsePortal {
+				p, err := url.Parse(ctx.Portal)
+				if err != nil {
+					return template.URL("error")
+				}
+				p = ctx.URL.ResolveReference(p)
+				u.Path = fmt.Sprintf("%s/%s/%s%s", p.Path, u.Scheme, u.Host, u.Path)
+				u.Scheme = p.Scheme
+				u.Host = p.Host
+
+				return template.URL(u.String())
+			}
+
+			u.Path = fmt.Sprintf("/x/%s%s", u.Host, u.Path)
+			u.Scheme = ""
+			u.Host = ""
+
 			return template.URL(u.String())
 		},
 		"safeCSS": func(s string) template.CSS {
@@ -383,19 +400,27 @@ input:focus {
 }
 `
 
+const defaultRobots = `
+User-agent: *
+Disallow: /
+`
+
 type GemtextContext struct {
-	CSS         string
-	ExternalCSS bool
-	External    bool
-	Lines       []gemini.Line
-	Pre         int
-	Resp        *gemini.Response
-	Title       string
-	Lang        string
-	URL         *url.URL
-	Root        *url.URL
-	UsePortal   bool
-	Portal      string
+	CSS                  string
+	ExternalCSS          bool
+	External             bool
+	Lines                []gemini.Line
+	Pre                  int
+	Resp                 *gemini.Response
+	Title                string
+	Lang                 string
+	URL                  *url.URL
+	Root                 *url.URL
+	UsePortal            bool
+	Portal               string
+	DisableExternalProxy bool
+	UseRewrite           bool
+	Rewrite              string
 }
 
 type InputContext struct {
@@ -427,8 +452,9 @@ func createAnchor(heading string) string {
 	return strings.ToLower(anchor.String())
 }
 
-func proxyGemini(req gemini.Request, external bool, root *url.URL,
-	w http.ResponseWriter, r *http.Request, css string, externalCSS bool, usePortal bool, portal string) {
+func proxyGemini(req gemini.Request, external bool, root *url.URL, w http.ResponseWriter,
+	r *http.Request, css string, externalCSS bool, usePortal bool, portal string, disableExternalProxy bool,
+	useRewrite bool, rewrite string) {
 
 	ctx, cancel := context.WithTimeout(r.Context(), 20*time.Second)
 	defer cancel()
@@ -524,16 +550,19 @@ func proxyGemini(req gemini.Request, external bool, root *url.URL,
 
 	w.Header().Add("Content-Type", "text/html")
 	gemctx := &GemtextContext{
-		CSS:         css,
-		ExternalCSS: externalCSS,
-		External:    external,
-		Resp:        resp,
-		Title:       req.URL.Host + " " + req.URL.Path,
-		Lang:        lang,
-		URL:         req.URL,
-		Root:        root,
-		UsePortal:   usePortal,
-		Portal:      portal,
+		CSS:                  css,
+		ExternalCSS:          externalCSS,
+		External:             external,
+		Resp:                 resp,
+		Title:                req.URL.Host + " " + req.URL.Path,
+		Lang:                 lang,
+		URL:                  req.URL,
+		Root:                 root,
+		UsePortal:            usePortal,
+		Portal:               portal,
+		DisableExternalProxy: disableExternalProxy,
+		UseRewrite:           useRewrite,
+		Rewrite:              rewrite,
 	}
 
 	var title bool
@@ -557,21 +586,40 @@ func proxyGemini(req gemini.Request, external bool, root *url.URL,
 
 func main() {
 	var (
-		bind      string = ":8080"
-		css       string = defaultCSS
-		external  bool   = false
-		usePortal bool   = false
-		portal    string = ""
+		bind                 string = ":8080"
+		css                  string = defaultCSS
+		external             bool   = false
+		usePortal            bool   = false
+		portal               string = ""
+		useRewrite           bool   = false
+		rewrite              string = ""
+		disableExternalProxy bool   = false
+		disableRobots        bool   = false
 	)
 
-	opts, optind, err := getopt.Getopts(os.Args, "b:c:s:e:p:")
+	opts, optind, err := getopt.Getopts(os.Args, "PRb:c:e:p:r:s:")
+
 	if err != nil {
 		log.Fatal(err)
 	}
+
 	for _, opt := range opts {
 		switch opt.Option {
+		case 'P':
+			disableExternalProxy = true
+		case 'R':
+			disableRobots = true
 		case 'b':
 			bind = opt.Value
+		case 'e':
+			external = true
+			css = opt.Value
+		case 'p':
+			usePortal = true
+			portal = opt.Value
+		case 'r':
+			useRewrite = true
+			rewrite = opt.Value
 		case 's':
 			external = false
 			cssContent, err := os.ReadFile(opt.Value)
@@ -580,12 +628,6 @@ func main() {
 			} else {
 				log.Fatalf("Error opening custom CSS from '%s': %v", opt.Value, err)
 			}
-		case 'e':
-			external = true
-			css = opt.Value
-		case 'p':
-			usePortal = true
-			portal = opt.Value
 		}
 	}
 
@@ -625,10 +667,12 @@ func main() {
 			return
 		}
 
-		if r.URL.Path == "/robots.txt" {
-			w.WriteHeader(http.StatusOK)
-			w.Write([]byte("User-agent: *\nDisallow: /\n"))
-			return
+		if !disableRobots {
+			if r.URL.Path == "/robots.txt" {
+				w.WriteHeader(http.StatusOK)
+				w.Write([]byte(defaultRobots))
+				return
+			}
 		}
 
 		req := gemini.Request{}
@@ -637,7 +681,7 @@ func main() {
 		req.URL.Host = root.Host
 		req.URL.Path = r.URL.Path
 		req.URL.RawQuery = r.URL.RawQuery
-		proxyGemini(req, false, root, w, r, css, external, usePortal, portal)
+		proxyGemini(req, false, root, w, r, css, external, usePortal, portal, disableExternalProxy, useRewrite, rewrite)
 	}))
 
 	http.Handle("/x/", http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
@@ -646,7 +690,7 @@ func main() {
 			path = append(path, "")
 		}
 
-		if usePortal && path[2] != root.Host {
+		if path[2] != root.Host && (usePortal || disableExternalProxy) {
 			w.WriteHeader(http.StatusMethodNotAllowed)
 			w.Write([]byte("404 Not found"))
 			return
@@ -678,7 +722,7 @@ func main() {
 		req.URL.Path = "/" + path[3]
 		req.URL.RawQuery = r.URL.RawQuery
 		log.Printf("%s (external) %s%s", r.Method, r.URL.Host, r.URL.Path)
-		proxyGemini(req, true, root, w, r, css, external, usePortal, portal)
+		proxyGemini(req, true, root, w, r, css, external, usePortal, portal, disableExternalProxy, useRewrite, rewrite)
 	}))
 
 	log.Printf("HTTP server listening on %s", bind)