feat: block any requests that are disallowed by robots txt

package.json

@@ -47,6 +47,7 @@
     "nanoid": "^5.0.9",
     "rehype-autolink-headings": "^7.1.0",
     "rehype-slug": "^6.0.0",
+    "robots-parser": "^3.0.1",
     "steamgriddb": "^2.2.0",
     "typescript-svelte-plugin": "^0.3.43"
   },

src/lib/lastfm.ts

@@ -15,7 +15,7 @@ export const lastFmGetNowPlaying: () => Promise<LastTrack | null> = async () =>
     try {
         var resp = await (await fetch(GET_RECENT_TRACKS_ENDPOINT)).json()
         var track = resp.recenttracks.track[0] ?? null
-        if (!(track['@attr'].nowplaying ?? null)) {
+        if (!((track['@attr'] ?? {}).nowplaying ?? null)) {
             throw "no nowplaying track found"
         }
         var data = {

src/lib/robots.ts

@@ -0,0 +1,53 @@
+import { env } from '$env/dynamic/private'
+import { get, writable } from 'svelte/store'
+import { type Robot } from 'robots-parser'
+import robotsParser from 'robots-parser'
+import { PUBLIC_BASE_URL } from '$env/static/public'
+
+const cachedParsedRobots = writable<Robot | null>(null)
+const cachedRobots = writable<string>("")
+const lastFetched = writable<number>(Date.now())
+
+const fetchRobotsTxt = async () => {
+    const robotsTxtResp = await fetch(
+        "https://api.darkvisitors.com/robots-txts",
+        {
+            method: "POST",
+            headers: {
+                "Authorization": `Bearer ${env.DARK_VISITORS_TOKEN}`,
+                "Content-Type": "application/json"
+            },
+            body: JSON.stringify({
+                agent_types: [
+                    "AI Assistant",
+                    "AI Data Scraper",
+                    "AI Search Crawler",
+                    "Undocumented AI Agent",
+                ],
+                disallow: "/"
+            })
+        }
+    )
+    const robotsTxt = await robotsTxtResp.text()
+    lastFetched.set(Date.now())
+    return robotsTxt
+}
+
+export const getRobotsTxt = async () => {
+    let robotsTxt = get(cachedRobots)
+    if (robotsTxt.length === 0 || Date.now() - get(lastFetched) > 1000 * 60 * 60 * 24) {
+        robotsTxt = await fetchRobotsTxt()
+        cachedRobots.set(robotsTxt)
+        cachedParsedRobots.set(robotsParser(`${PUBLIC_BASE_URL}/robots.txt`, robotsTxt))
+    }
+    return robotsTxt
+}
+
+export const testUa = async (url: string, ua: string) => {
+    let parsedRobots = get(cachedParsedRobots)
+    if (parsedRobots === null) {
+        parsedRobots = robotsParser(`${PUBLIC_BASE_URL}/robots.txt`, await getRobotsTxt())
+        cachedParsedRobots.set(parsedRobots)
+    }
+    return parsedRobots.isAllowed(url, ua)
+}

src/routes/+layout.server.ts

@@ -1,13 +1,20 @@
+import { testUa } from '$lib/robots.js';
 import { incrementVisitCount, notifyDarkVisitors } from '$lib/visits.js';
+import { error } from '@sveltejs/kit';
 
 export const csr = true;
 export const ssr = true;
 export const prerender = false;
 export const trailingSlash = 'always'; 
 
-export async function load({ request, cookies, url, setHeaders }) {
+export async function load({ request, cookies, url }) {
     notifyDarkVisitors(url, request) // no await so it doesnt block load
 
+    // block any requests if the user agent is disallowed by our robots txt
+    if (await testUa(url.toString(), request.headers.get('user-agent') ?? "unknown user agent") === false) {
+        throw error(403, "get a better user agent silly")
+    }
+
     return {
         route: url.pathname,
         visitCount: incrementVisitCount(request, cookies),

src/routes/robots.txt/+server.ts

@@ -1,39 +1,5 @@
-import { env } from '$env/dynamic/private';
-import { get, writable } from 'svelte/store';
-
-const cachedRobots = writable<string>("")
-const lastFetched = writable<number>(Date.now())
-
-const fetchRobotsTxt = async () => {
-    const robotsTxtResp = await fetch(
-        "https://api.darkvisitors.com/robots-txts",
-        {
-            method: "POST",
-            headers: {
-                "Authorization": `Bearer ${env.DARK_VISITORS_TOKEN}`,
-                "Content-Type": "application/json"
-            },
-            body: JSON.stringify({
-                agent_types: [
-                    "AI Assistant",
-                    "AI Data Scraper",
-                    "AI Search Crawler",
-                    "Undocumented AI Agent",
-                ],
-                disallow: "/"
-            })
-        }
-    )
-    const robotsTxt = await robotsTxtResp.text()
-    lastFetched.set(Date.now())
-    return robotsTxt
-}
+import { getRobotsTxt } from "$lib/robots"
 
 export const GET = async ({ }) => {
-    let robotsTxt = get(cachedRobots)
-    if (robotsTxt.length === 0 || Date.now() - get(lastFetched) > 1000 * 60 * 60 * 24) {
-        robotsTxt = await fetchRobotsTxt()
-        cachedRobots.set(robotsTxt)
-    }
-    return new Response(robotsTxt)
+    return new Response(await getRobotsTxt())
 }