restart always

.gitignore

@@ -1 +1,5 @@
-.vscode
+.vscode
+__pycache__
+test.mp4
+env
+errors

Dockerfile

@@ -0,0 +1,9 @@
+FROM python:3.11
+RUN useradd -r -m u2b
+RUN apt update && apt install -y ffmpeg
+RUN pip install --no-cache-dir python-ffmpeg yt-dlp
+#COPY . /app
+#WORKDIR /app
+USER u2b
+#ENV PORT=8080
+#CMD ["python", "server.py"]

README.md

@@ -1,79 +0,0 @@
-# u2b.cx
-
-A raw YouTube video server/resolver via search query, made to let you watch what you want in Quest VRChat where you don't really have any other options to do so.
-
-
-## Usage
-
-### GET `https://u2b.cx/<query>`
-
-The server will search YouTube for `<query>`, pick the first result, pick the best quality all-in-one MP4 format available, and respond with a 302 redirect to the proxied raw MP4 file.
-
-Example: [`https://u2b.cx/nyan cat`](https://u2b.cx/nyan%20cat) will give you the raw mp4 file for [https://www.youtube.com/watch?v=QH2-TGUlwu4](https://www.youtube.com/watch?v=QH2-TGUlwu4), the first result on [https://www.youtube.com/results?search_query=nyan+cat](https://www.youtube.com/results?search_query=nyan+cat).
-
-
-## Installation
-
-Only python module needed is `yt-dlp`. It must be kept up to date, recommend auto update with cron.
-
-### caddy file
-```Caddyfile
-u2b.cx {
-        import common
-        route {
-                @www path / /favicon.ico
-                redir @www https://www.u2b.cx{uri} permanent
-
-                respond /robots.txt "User-agent: *
-Disallow: /
-"
-                respond /.* 403
-                @notget not method GET
-                respond @notget 403
-
-                reverse_proxy http://127.29.151.200:52482
-        }
-}
-
-proxy.u2b.cx {
-        import common
-        @gv path_regexp gvurl ^\/([a-z0-9-]+\.googlevideo\.com)
-        handle @gv {
-                uri strip_prefix /{re.gvurl.1}
-                #uri strip_prefix {re.gvurl.0} doesn't work for some reason despite that value being equal to /{re.gvurl.1}
-                reverse_proxy {
-                        to {re.gvurl.1}:443
-                        header_up Host {re.gvurl.1}
-                        transport http {
-                                tls
-                        }
-                }
-        }
-}
-
-www.u2b.cx {
-        import common
-        root * /srv/u2b.cx/www/
-        file_server
-}
-```
-
-### service
-youtubedl writes cache to disk, make user with home directory!
-
-```systemd
-[Unit]
-Description=u2b.cx youtube search resolver service thingy
-After=network.target
-
-[Service]
-User=u2b
-Group=u2b
-WorkingDirectory=/srv/u2b.cx/
-Environment=ADDRESS=127.29.151.200 PORT=52482 PROXY=https://proxy.u2b.cx/
-ExecStart=/usr/bin/python3.9 server.py
-MemoryMax=1G
-
-[Install]
-WantedBy=multi-user.target
-```

caddy/Caddyfile

@@ -0,0 +1,56 @@
+{
+	log http.log.access {
+		include http.log.access
+		output stdout
+		format formatted "[{ts}] {request>remote_ip} {request>headers>X-Forwarded-For} {request>method} {request>host}{request>uri} {status}	{request>headers>User-Agent} {request>headers>Referer}" {
+			time_format "02/Jan/2006:15:04:05-0700"
+		}
+	}
+	log default {
+		exclude http.log.access
+		output stderr
+		format console
+	}
+}
+
+{$CADDY_SITE:":80"} {
+	log
+	tls {
+		dns porkbun {
+			api_key {env.PORKBUN_API_KEY}
+ 			api_secret_key {env.PORKBUN_API_SECRET}
+		}
+	}
+	handle_path /proxy/* {
+		@gv path_regexp gvurl ^\/([a-z0-9-]+\.googlevideo\.com)
+		handle @gv {
+			uri strip_prefix /{re.gvurl.1}
+			reverse_proxy {
+				to {re.gvurl.1}:443
+				header_up Host {re.gvurl.1}
+				header_down Location "^https://(.*)" "/proxy/$1"
+				transport http {
+					tls
+				}
+			}
+		}
+		handle {
+			respond 403
+		}
+	}
+	handle {
+		route {
+			@www path / /favicon.ico
+			redir @www https://www.u2b.cx{uri} permanent
+		
+			respond /robots.txt "User-agent: *
+Disallow: /
+"
+			respond /.* 403
+			@notget not method GET
+			respond @notget 403
+			
+			reverse_proxy http://127.0.0.1:8080
+		}
+	}
+}

caddy/Dockerfile

@@ -0,0 +1,6 @@
+FROM caddy:2.6-builder AS builder
+RUN xcaddy build \
+	 --with github.com/caddyserver/transform-encoder \
+	 --with github.com/caddy-dns/porkbun
+FROM caddy:2.6
+COPY --from=builder /usr/bin/caddy /usr/bin/caddy

docker-compose.yml

@@ -0,0 +1,27 @@
+version: "3.8"
+name: "u2bcx"
+services:
+  app:
+    build: .
+    restart: always
+    volumes:
+      - ./:/app/
+    working_dir: /app/
+    environment:
+      - ADDRESS=127.0.0.1
+      - PORT=8080
+      - PROXY=/proxy/
+    network_mode: host
+    command: python server.py
+  caddy:
+    build: caddy
+    restart: always
+    network_mode: host
+    volumes:
+      - ./caddy/Caddyfile:/etc/caddy/Caddyfile
+      - caddy_data:/data
+      - caddy_config:/config
+    env_file: env
+volumes:
+  caddy_data:
+  caddy_config:

server.py

@@ -1,80 +1,217 @@
 from yt_dlp import YoutubeDL
+from importlib.metadata import version
+print("yt-dlp version", version("yt_dlp"))
 from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
 from urllib.parse import unquote, urlparse, parse_qs
-from threading import Event
+from threading import Event, Thread
 from datetime import datetime, timedelta
-from os import environ
+from time import sleep
+from os import environ, makedirs, stat
 import logging
+import re
+from ffmpeg import FFmpeg
+import textwrap
+from pathlib import Path
+from hashlib import sha256
+from shutil import copyfileobj
+from math import ceil
 
-def getBestMp4UrlFromInfo(info):
-	entries = info.get('entries')
-	if not entries: return None
-	formats = entries[0].get('formats')
-	if not formats: return None
-	valid = list(filter(lambda x: x['ext'] == "mp4" and x['vcodec'] != 'none' and x['acodec'] != 'none', formats))
-	if not valid: return None
-	best = max(valid, key=lambda x: x['height'])
-	return best.get('url')
-
-def getExpireDatetime(url):
-	alt_expire = datetime.now() + timedelta(hours=5)
-	if not url: return alt_expire
-	q = parse_qs(urlparse(url).query)
-	expire = q.get('expire')
-	if not expire: return alt_expire
-	expire = datetime.fromtimestamp(int(expire[0])) #this seems to always be +6 hours
-	return min([expire, alt_expire])
-
-query_dicts = {}
-ips_running_ytdl = []
+if 'DEBUG' in environ:
+	logging.basicConfig(level=logging.DEBUG)
 
+class Ratelimit(Exception): pass
+class CachedException(Exception): pass
 class Handler(BaseHTTPRequestHandler):
 	def address_string(self):
 		return getattr(self, 'headers', {}).get('X-Forwarded-For', '').split(',')[0] or self.client_address[0]
 
-	def do_GET(self):
-		# block other bot junk in reverse proxy
-		if self.path in ["/", "/favicon.ico"]:
-			self.send_error(404)
-			return
-		
-		query = unquote(self.path[1:])
-		dict = query_dicts.get(query)
+	def send_error(self, code, message=""):
+		body = bytes(message, "utf-8")
+		self.send_response(code)
+		self.send_header("Content-Type", "text/plain")
+		self.send_header("Content-Length", str(len(body)))
+		self.end_headers()
+		self.wfile.write(body)
 
-		if not dict or 'expire' in dict and datetime.now() >= dict['expire']:
-			client_ip = self.address_string()
-			if client_ip in ips_running_ytdl:
-				self.send_error(429)
-				return
-			try:
-				ips_running_ytdl.append(client_ip)
-				query_dicts[query] = dict = {'event': Event()}
-				with YoutubeDL() as ydl:
-					info = ydl.extract_info("ytsearch:" + query, download=False)
-					url = getBestMp4UrlFromInfo(info)
-					dict['url'] = url
-					dict['expire'] = getExpireDatetime(url)
-			except Exception as e:
-				logging.exception(e)
-				dict['exception'] = e
-			finally:
-				ips_running_ytdl.remove(client_ip)
-				dict['event'].set()
-		elif 'url' not in dict:
-			dict['event'].wait(60)
+	def send_error_video(self, text: str):
+		makedirs("errors", exist_ok=True)
+		hash = sha256(bytes(text, "utf8")).hexdigest()
+		file = Path(f"errors/{hash}.mp4")
 
-		if not dict.get('url'):
-			if 'exception' in dict:
-				self.send_error(500, message=str(dict['exception']), explain="An unexpected exception was encountered while resolving this query")
-			else:
-				self.send_error(404)
-		else:
-			url = dict['url']
-			if 'PROXY' in environ:
-				url = environ['PROXY'] + url.replace("https://",'')
-			self.send_response(302)
-			self.send_header("Location", url)
+		if not file.exists():
+			text = re.sub("(\x9B|\x1B\[)[0-?]*[ -\/]*[@-~]", '', text)
+			text = text.replace("\\", "\\\\").replace('"', '""').replace("'", "''").replace("%", "\\%").replace(":", "\\:")
+			text = textwrap.fill(text, 90)
+			peg = FFmpeg().option("y").input("bg.png", {
+				"framerate": "0.1"
+			}).output(str(file), {
+				"f": "mp4",
+				"t": "10",
+				"c:v": "libx264",
+				"pix_fmt": "yuv420p",
+				"vf": "drawtext=font=monospace:fontsize=24:x=10:y=10:text='"+text+"':"
+			})
+			@peg.on("start")
+			def on_start(arguments):
+				logging.debug("cmd:" + ' '.join(arguments))
+			@peg.on("stderr")
+			def on_stderr(line):
+				logging.debug("stderr:" + line)
+			peg.execute()
+
+		with file.open('rb') as f:
+			fs = stat(f.fileno())
+			self.send_response(200)
+			self.send_header("Content-Type", "video/mp4")
+			self.send_header("Content-Length", str(fs.st_size))
 			self.end_headers()
+			copyfileobj(f, self.wfile)
+
+	def do_GET(self):
+		try:
+			if self.path in ["/", "/favicon.ico"] or self.path.startswith("/."):
+				self.send_error(404)
+				return
+			
+			path = unquote(self.path)
+			id_match = re.match("\/(?:id\/|(?:https?:\/\/)?(?:(?:www\.|music\.|m\.)?youtube\.com\/(?:watch\?v=|shorts\/|live\/)|youtu\.be\/))([A-Za-z0-9_-]{11})", path)
+			
+			if id_match:
+				video_id = id_match[1]
+			else:
+				search_match = re.match("^\/(.+?)(?:\/(\d*))?$", path)
+				if not search_match:
+					self.send_error(404)
+					return
+				search_query = search_match[1]
+				search_index = int(search_match[2]) if search_match[2] and search_match[2].isdigit() else 1
+				search_index = max(min(search_index, 100), 1)
+				video_id = ytdl_search_to_id(self, search_query, search_index)
+
+			self.send_response(302)
+			self.send_header("Location", f"https://www.youtube.com/watch?v={video_id}")
+			self.end_headers()
+			#half this code now defunct
+			return
+
+			video_url = ytdl_resolve_mp4_url(self, video_id)
+
+			if 'PROXY' in environ:
+				video_url = environ['PROXY'] + video_url.replace("https://",'')
+			self.send_response(302)
+			self.send_header("Location", video_url)
+			self.end_headers()
+		except Ratelimit:
+			self.send_error(429)
+		except CachedException as e:
+			self.send_error_video(str(e))
+		except Exception as e:
+			logging.exception(e)
+			self.send_error_video(str(e))
+
+
+ips_running_ytdl = []
+def invoke_youtubedl(self: Handler, input: str) -> dict:
+	ip = self.address_string()
+	if ip in ips_running_ytdl:
+		raise Ratelimit()
+	ips_running_ytdl.append(ip)
+	try:
+		with YoutubeDL({'extractor_args': {'youtube': {'skip': ['dash', 'hls']}}}) as ydl:
+			return ydl.extract_info(input, download=False, process=False)
+	finally:
+		ips_running_ytdl.remove(ip)
+
+
+search_cache = {}
+def ytdl_search_to_id(self: Handler, query: str, index: int) -> str:
+	ctx = search_cache.get(query)
+	if ctx:
+		ctx['event'].wait(60)
+		if 'error' in ctx:
+			raise CachedException(ctx['error'])
+		results = ctx.get('results')
+	else:
+		results = None
+
+	if results == None or results['count'] < index or datetime.now() >= ctx['expires_at']:
+		search_cache[query] = ctx = {
+			'event': Event(),
+			'expires_at': datetime.now() + timedelta(hours=5)
+		}
+		try:
+			count = ceil(index/10)*10
+			info = invoke_youtubedl(self, f"ytsearch{count}:{query}")
+			entries = list(info['entries'])
+			if not entries:
+				raise Exception("ERROR: No results!")
+			ids = [video['id'] for video in entries]
+			ctx['results'] = results = {'ids': ids, 'count': count}
+		except Exception as e:
+			ctx['error'] = str(e)
+			raise
+		finally:
+			ctx['event'].set()
+	
+	return results['ids'][min(index, len(results['ids'])) - 1]
+
+
+resolve_cache = {}
+def ytdl_resolve_mp4_url(self: Handler, input: str) -> str:
+	ctx = resolve_cache.get(input)
+	if ctx and datetime.now() <= ctx['expires_at']:
+		ctx['event'].wait(60)
+		if 'error' in ctx:
+			raise CachedException(ctx['error'])
+		return ctx['url']
+	
+	resolve_cache[input] = ctx = {
+		'event': Event(),
+		'expires_at': datetime.now() + timedelta(hours=5)
+	}
+	try:
+		info = invoke_youtubedl(self, input)
+		selection = info
+		if "entries" in info:
+			if not info["entries"]:
+				raise Exception("ERROR: No video found!")
+			else:
+				selection = info["entries"][0]
+
+		suitable_formats = list(filter(lambda x: x['ext'] == "mp4" and x['vcodec'] != 'none' and x['acodec'] != 'none', selection["formats"]))
+		if not suitable_formats:
+			raise Exception(f"ERROR: {selection['id']}: No suitable formats of this video available!")
+		best_format = max(suitable_formats, key=lambda x: x['height'])
+
+		ctx['url'] = url = best_format['url']
+
+		try:
+			expire = parse_qs(urlparse(url).query).get('expire', [])[0]
+			if expire:
+				expire = datetime.fromtimestamp(int(expire))
+				if expire < ctx['expires_at']:
+					ctx['expires_at'] = expire
+		except Exception as e:
+			logging.exception("failed parsing expire", e)
+	except Exception as e:
+		ctx['error'] = str(e)
+		raise
+	finally:
+		ctx['event'].set()
+
+	return url
+
+
+def cache_prune_loop():
+	while True:
+		sleep(3600)
+		for key in list(search_cache.keys()):
+			if datetime.now() >= search_cache[key]['expires_at']:
+				del search_cache[key]
+		for key in list(resolve_cache.keys()):
+			if datetime.now() >= resolve_cache[key]['expires_at']:
+				del resolve_cache[key]
+Thread(target=cache_prune_loop, daemon=True).start()
 
 with ThreadingHTTPServer((environ.get('ADDRESS', ''), int(environ.get('PORT', 80))), Handler) as server:
 	server.serve_forever()

u2b.cx.service

@@ -0,0 +1,17 @@
+[Unit]
+Description=u2b.cx youtube search resolver service thingy
+After=network.target
+
+[Service]
+#youtubedl writes cache on disk so user should have home
+User=u2b
+Group=u2b
+WorkingDirectory=/srv/u2b.cx/
+Environment=ADDRESS=127.0.0.1 PORT=52482
+ExecStart=/usr/bin/python3.11 server.py
+MemoryMax=1G
+LimitNOFILE=262144
+Restart=always
+
+[Install]
+WantedBy=multi-user.target

www/index.html

@@ -1,30 +0,0 @@
-<!DOCTYPE html><html><head>
-	<meta property="og:title" content="Watch YouTube on Quest VRChat!!!" />
-	<meta property="og:description" content="Just type in https://u2b.cx/ followed by whatever you wanna watch!" />
-	<meta property="og:image" content="https://www.u2b.cx/promo.gif" />
-	<meta property="og:type" content="website" />
-	<meta name="twitter:card" content="summary_large_image" />
-	<meta name="keywords" content="youtube,quest,vrchat" />
-	<title>Watch YouTube on Quest VRChat with u2b.cx!!!</title>
-	<style>
-		img {
-			vertical-align: middle;
-		}
-	</style>
-</head><body>
-	<i><h2>Problem:</h2></i>
-	<img src="VRChat_2023-03-21_22-37-32.758_1920x1080.png" alt="A video player in VRChat" style="max-height: 50vh" />
-	<p>No video! No PC to put video! No URLs memorized! What do???</p>
-	<i><h2>SOLUTION!</h2></i>
-	<ol>
-		<li>Type the following exactly into the URL bar: <code>https://u2b.cx/</code></li>
-		<li>TIP: Press the <img alt="copy" src="data:image/png;base64,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" width="24" height="24" /> button on the lower left 😉</li>
-		<li>Now after that just type what you wanna watch. For example, if you wanna play nyan cat, type in like so: <code>https://u2b.cx/nyan cat</code></li>
-		<li>The first result from YouTube will start playing. YAAY!</li>
-		<li>TIP: When you wanna watch again, press the <img alt="paste" src="data:image/png;base64,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" width="24" height="24" /> button on the lower right so you don't have to type it all again 👌</li>
-	</ol>
-	<i><h3>Demo</h3></i>
-	<video controls style="max-height: 80vh"><source src="com.vrchat.oculus.quest-20230321-223838.mp4" type="video/mp4"/></video>
-	<i><h3>Problems?</h3></i>
-	<p>If you encounter any issues, please submit on <a href="https://gitea.moe/lamp/u2b.cx/issues">issue tracker</a>. You can use Discord or GitHub account.</p>
-</body></html>