NewLeaf: Partially fix #18 proxy non-adaptative format videos through new /videoplayback endpoint v1 PROPOSED

As discussed a while ago, this is my current progress for #18, I use
this version of the code since the date of this commit and this is
functional. It's especially useful for the " - Topic" channels auto-
generated music videos.
Though, I didn't manage to have DASH streams to work proxied, so this MR
only proxies the regular non-adaptative streams.

As this is very much WIP, don't merge probably. I'm also not happy with
the verbose `for k, v in r.headers.items():` loop to set headers in
server-side requests.

Lomanic (1):
  Partially fix #18 proxy non-adaptative format videos through new
    /videoplayback endpoint

 configuration.sample.py |  3 +++
 extractors/video.py     |  3 ++-
 index.py                | 18 ++++++++++++++++++
 3 files changed, 23 insertions(+), 1 deletion(-)

-- 
2.34.4

[PATCH NewLeaf 1/1] Partially fix #18 proxy non-adaptative format videos through new /videoplayback endpoint Export this patch

From: Lomanic <lomanic@hotmail.fr>

Configured instance-wide with a boolean in configuration.py
Disabled by default
---
 configuration.sample.py |  3 +++
 extractors/video.py     |  3 ++-
 index.py                | 18 ++++++++++++++++++
 3 files changed, 23 insertions(+), 1 deletion(-)

diff --git a/configuration.sample.py b/configuration.sample.py
index 470892f..63156e3 100644
--- a/configuration.sample.py
+++ b/configuration.sample.py
@@ -16,3 +16,6 @@ website_origin = "http://example.com:3000"
 
 # The port to bind to.
 #bind_port = 3000
+
+# Tell NewLeaf to proxy video streams
+# proxy_videos = False
diff --git a/extractors/video.py b/extractors/video.py
index 68fc86a..4e8ec64 100644
--- a/extractors/video.py
+++ b/extractors/video.py
@@ -14,6 +14,7 @@ from urllib.parse import parse_qs, urlparse, urlencode
 from cachetools import TTLCache
 
 video_cache = TTLCache(maxsize=50, ttl=300)
+proxy_videos = getattr(configuration, "proxy_videos", False)
 
 ytdl_opts = {
 	"quiet": True,
@@ -157,7 +158,7 @@ def extract_video(id):
 				})
 			else: # format is not adaptive
 				result["formatStreams"].append({
-					"url": format["url"],
+					"url": "{}/videoplayback?{}".format(configuration.website_origin, urlencode({"url": format["url"]})) if proxy_videos else format["url"],
 					"itag": format["format_id"],
 					"type": result_type,
 					"second__mime": mime,
diff --git a/index.py b/index.py
index 482c6d5..a5d7da6 100644
--- a/index.py
+++ b/index.py
@@ -3,6 +3,7 @@ import json
 import pathlib
 import requests
 import yt_dlp
+from urllib.parse import urlparse
 from extractors.video import extract_video
 from extractors.channel import extract_channel, extract_channel_videos, extract_channel_latest
 from extractors.manifest import extract_manifest
@@ -141,6 +142,23 @@ class NewLeaf(object):
 		cherrypy.response.headers["content-type"] = r.headers["content-type"]
 		return next(r.iter_content(chunk_size=None))
 
+	@cherrypy.expose
+	def videoplayback(self, url):
+		proxy_videos = getattr(configuration, "proxy_videos", False)
+		if not proxy_videos:
+			raise cherrypy.HTTPError(401, "Unauthorized")
+		u = urlparse(url)
+		if not (u.hostname.endswith(".googlevideo.com") and u.path == "/videoplayback"):
+			raise cherrypy.HTTPError(401, "Unauthorized")
+		headers = {k: v for k, v in cherrypy.request.headers.items() if k.lower() not in ["host", "remote-addr"] and not k.lower().startswith("x-")}
+		with requests.get(url, headers=headers, stream=True) as r:
+			r.raise_for_status()
+			for k, v in r.headers.items():
+				cherrypy.response.headers[k] = v
+			cherrypy.response.status = r.status_code
+			# return 5MB at a time, the browser will ask for the next chunk thanks to HTTP 206 Partial Content streaming magic
+			return next(r.iter_content(chunk_size=5*1024*1024))
+
 bind_port = getattr(configuration, "bind_port", 3000)
 bind_host = getattr(configuration, "bind_host", "0.0.0.0")
 server_root = pathlib.Path(__file__).parent.joinpath("root")
-- 
2.34.4