Chunk videoplayback response to avoid throttling

src/invidious.cr

@@ -54,6 +54,7 @@ MAX_ITEMS_PER_PAGE = 1500
 
 REQUEST_HEADERS_WHITELIST  = {"Accept", "Accept-Encoding", "Cache-Control", "Connection", "Content-Length", "If-None-Match", "Range"}
 RESPONSE_HEADERS_BLACKLIST = {"Access-Control-Allow-Origin", "Alt-Svc", "Server"}
+HTTP_CHUNK_SIZE            = 10485760 # ~10MB
 
 CURRENT_BRANCH  = {{ "#{`git branch | sed -n '/\* /s///p'`.strip}" }}
 CURRENT_COMMIT  = {{ "#{`git rev-list HEAD --max-count=1 --abbrev-commit`.strip}" }}
@@ -4648,39 +4649,56 @@ get "/videoplayback" do |env|
     next
   end
 
-  client = make_client(URI.parse(host), proxies, region)
   begin
-    client.get(url, headers) do |response|
-      env.response.status_code = response.status_code
+    range_begin, range_end = parse_range(env.request.headers["Range"]?)
 
-      response.headers.each do |key, value|
-        if !RESPONSE_HEADERS_BLACKLIST.includes? key
-          env.response.headers[key] = value
-        end
-      end
+    client = make_client(URI.parse(host), proxies, region)
+    (range_begin...range_end).each_slice(HTTP_CHUNK_SIZE) do |slice|
+      headers["Range"] = "bytes=#{slice[0]}-#{slice[-1]}"
+      begin
+        client.get(url, headers) do |response|
+          content_range = response.headers["Content-Range"].lchop("bytes ")
+          content_size = content_range.split("/")[-1].to_i
+
+          # Write headers for first chunk
+          if content_range.split("-")[0].to_i == range_begin
+            if !env.request.headers["Range"]? && response.status_code == 206
+              env.response.status_code = 200
+            else
+              env.response.status_code = response.status_code
+            end
 
-      env.response.headers["Access-Control-Allow-Origin"] = "*"
+            response.headers.each do |key, value|
+              if !RESPONSE_HEADERS_BLACKLIST.includes?(key) && key != "Content-Range"
+                env.response.headers[key] = value
+              end
+            end
 
-      if response.headers["Location"]?
-        url = URI.parse(response.headers["Location"])
-        host = url.host
+            reported_end = range_end ? range_end : content_size
+            env.response.content_length = reported_end - range_begin
 
-        url = url.full_path
-        url += "&host=#{host}"
+            if env.request.headers["Range"]?
+              env.response.headers["Content-Range"] = "bytes #{range_begin}-#{reported_end - 1}/#{content_size}"
+            end
 
-        if region
-          url += "&region=#{region}"
-        end
+            env.response.headers["Access-Control-Allow-Origin"] = "*"
 
-        next env.redirect url
-      end
+            if title = query_params["title"]?
+              # https://blog.fastmail.com/2011/06/24/download-non-english-filenames/
+              env.response.headers["Content-Disposition"] = "attachment; filename=\"#{URI.escape(title)}\"; filename*=UTF-8''#{URI.escape(title)}"
+            end
+          end
 
-      if title = query_params["title"]?
-        # https://blog.fastmail.com/2011/06/24/download-non-english-filenames/
-        env.response.headers["Content-Disposition"] = "attachment; filename=\"#{URI.escape(title)}\"; filename*=UTF-8''#{URI.escape(title)}"
+          proxy_file(response, env)
+        end
+      rescue ex
+        # FIXME: Potential bug in HTTP::Client
+        if ex.message == "Error reading socket: Connection reset by peer"
+          next
+        else
+          raise ex
+        end
       end
-
-      proxy_file(response, env)
     end
   rescue ex
   end

src/invidious/helpers/helpers.cr

@@ -656,7 +656,7 @@ def proxy_file(response, env)
 end
 
 # https://stackoverflow.com/a/44802810 <3
-def copy_in_chunks(input, output, chunk_size = 4096)
+def copy_in_chunks(input, output, chunk_size = 8192)
   size = 1
   while size > 0
     size = IO.copy(input, output, chunk_size)

src/invidious/helpers/utils.cr

@@ -358,3 +358,21 @@ def subscribe_pubsub(topic, key, config)
 
   return client.post("/subscribe", form: body)
 end
+
+def parse_range(range)
+  if !range
+    return 0, nil
+  end
+
+  ranges = range.lchop("bytes=").split(',')
+  ranges.each do |range|
+    start_range, end_range = range.split('-')
+
+    start_range = start_range.to_i? || 0
+    end_range = end_range.to_i?.try &.+ 1
+
+    return start_range, end_range
+  end
+
+  return 0, nil
+end