From eb111a44c4392f982d120ce6158067101c27a636 Mon Sep 17 00:00:00 2001
From: Lomanic <lomanic@hotmail.fr>
Date: Sat, 22 May 2021 02:32:43 +0200
Subject: [PATCH] Proxy captions via new /proxy route

We can add more authorized paths to authorizedPaths if we need
more resources to be pulled from the NewLeaf/Invidious backend
on the same domain.

This route forwards to the client a few headers like Bibliogram
https://git.sr.ht/~cadence/bibliogram/tree/ea7cd5d5/item/src/site/api/proxy.js#L28-29
so we can also use this route to possibly proxy videos in the
future.

We are strict about the url parameter not overriding the
NewLeaf/Invidious backend (instanceOrigin) by ensuring
fetchURL has instanceOrigin as prefix.
---
 api/captions.js | 27 ---------------------------
 api/proxy.js    | 27 +++++++++++++++++++++++++++
 api/video.js    |  5 +++++
 3 files changed, 32 insertions(+), 27 deletions(-)
 delete mode 100644 api/captions.js
 create mode 100644 api/proxy.js

diff --git a/api/captions.js b/api/captions.js
deleted file mode 100644
index f44b921..0000000
--- a/api/captions.js
+++ /dev/null
@@ -1,27 +0,0 @@
-/** @type {import("node-fetch").default} */
-// @ts-ignore
-const fetch = require("node-fetch")
-const {getUser} = require("../utils/getuser")
-const constants = require("../utils/constants.js")
-
-module.exports = [
-	{
-		route: `/api/v1/captions/(${constants.regex.video_id})`, methods: ["GET"], code: async ({req, fill, url}) => {
-			const instanceOrigin = getUser(req).getSettingsOrDefaults().instance
-			const fetchURL = new URL(`${url.pathname}${url.search}`, instanceOrigin)
-			return fetch(fetchURL.toString()).then(res => {
-				return res.text().then(text => {
-					if (res.status === 200) {
-						// Remove the position annotations that youtube unhelpfully provides
-						text = text.replace(/(--> \S+).*/g, "$1")
-					}
-					return {
-						statusCode: res.status,
-						contentType: res.headers.get("content-type"),
-						content: text
-					}
-				})
-			})
-		}
-	}
-]
diff --git a/api/proxy.js b/api/proxy.js
new file mode 100644
index 0000000..a6fecbe
--- /dev/null
+++ b/api/proxy.js
@@ -0,0 +1,27 @@
+const {proxy} = require("pinski/plugins")
+const {getUser} = require("../utils/getuser")
+const constants = require("../utils/constants.js")
+
+// list of paths relative to the backend this route is authorized to serve
+const authorizedPaths = [`/api/v1/captions/(${constants.regex.video_id})`]
+
+// headers relayed as-is from the proxied backend to the client
+const proxiedHeaders = ["content-type", "date", "last-modified", "expires", "cache-control", "accept-ranges", "content-range", "origin", "etag", "content-length", "transfer-encoding"]
+
+module.exports = [
+	{
+		route: `/proxy`, methods: ["GET"], code: async ({req, fill, url}) => {
+			const instanceOrigin = getUser(req).getSettingsOrDefaults().instance
+			const remotePath = url.searchParams.get("url")
+			const fetchURL = new URL(remotePath, instanceOrigin)
+			if (!fetchURL.toString().startsWith(instanceOrigin) || !authorizedPaths.some(element => fetchURL.pathname.match(new RegExp(`^${element}$`)))) {
+				return {
+					statusCode: 401,
+					content: "CloudTube: Unauthorized",
+					contentType: "text/plain"
+				}
+			}
+			return proxy(fetchURL, {}, (h) => Object.keys(h).filter(key => proxiedHeaders.includes(key)).reduce((res, key) => (res[key] = h[key], res), {}))
+		}
+	}
+]
diff --git a/api/video.js b/api/video.js
index 1e0445b..d60f52d 100644
--- a/api/video.js
+++ b/api/video.js
@@ -174,6 +174,11 @@ module.exports = [
 				// rewrite description
 				video.descriptionHtml = converters.rewriteVideoDescription(video.descriptionHtml, id)
 
+				// rewrite captions urls so they are served on the same domain via the /proxy route
+				for (const caption of video.captions) {
+					caption.url = `/proxy?${new URLSearchParams({"url": caption.url})}`
+				}
+
 				return render(200, "pug/video.pug", {
 					url, video, formats, subscribed, instanceOrigin, mediaFragment, autoplay, continuous,
 					sessionWatched, sessionWatchedNext