Refactor instance fetching logic into separate job

src/invidious.cr

@@ -185,6 +185,8 @@ Invidious::Jobs.register Invidious::Jobs::NotificationJob.new(CONNECTION_CHANNEL
 
 Invidious::Jobs.register Invidious::Jobs::ClearExpiredItemsJob.new
 
+Invidious::Jobs.register Invidious::Jobs::InstanceListRefreshJob.new
+
 Invidious::Jobs.start_all
 
 def popular_videos

src/invidious/helpers/utils.cr

@@ -323,68 +323,6 @@ def parse_range(range)
   return 0_i64, nil
 end
 
-def fetch_random_instance
-  begin
-    instance_api_client = make_client(URI.parse("https://api.invidious.io"))
-
-    # Timeouts
-    instance_api_client.connect_timeout = 10.seconds
-    instance_api_client.dns_timeout = 10.seconds
-
-    instance_list = JSON.parse(instance_api_client.get("/instances.json").body).as_a
-    instance_api_client.close
-  rescue Socket::ConnectError | IO::TimeoutError | JSON::ParseException
-    instance_list = [] of JSON::Any
-  end
-
-  filtered_instance_list = [] of String
-
-  instance_list.each do |data|
-    # TODO Check if current URL is onion instance and use .onion types if so.
-    if data[1]["type"] == "https"
-      # Instances can have statistics disabled, which is an requirement of version validation.
-      # as_nil? doesn't exist. Thus we'll have to handle the error raised if as_nil fails.
-      begin
-        data[1]["stats"].as_nil
-        next
-      rescue TypeCastError
-      end
-
-      # stats endpoint could also lack the software dict.
-      next if data[1]["stats"]["software"]?.nil?
-
-      # Makes sure the instance isn't too outdated.
-      if remote_version = data[1]["stats"]?.try &.["software"]?.try &.["version"]
-        remote_commit_date = remote_version.as_s.match(/\d{4}\.\d{2}\.\d{2}/)
-        next if !remote_commit_date
-
-        remote_commit_date = Time.parse(remote_commit_date[0], "%Y.%m.%d", Time::Location::UTC)
-        local_commit_date = Time.parse(CURRENT_VERSION, "%Y.%m.%d", Time::Location::UTC)
-
-        next if (remote_commit_date - local_commit_date).abs.days > 30
-
-        begin
-          data[1]["monitor"].as_nil
-          health = data[1]["monitor"].as_h["dailyRatios"][0].as_h["ratio"]
-          filtered_instance_list << data[0].as_s if health.to_s.to_f > 90
-        rescue TypeCastError
-          # We can't check the health if the monitoring is broken. Thus we'll just add it to the list
-          # and move on. Ideally we'll ignore any instance that has broken health monitoring but due to the fact that
-          # it's an error that often occurs with all the instances at the same time, we have to just skip the check.
-          filtered_instance_list << data[0].as_s
-        end
-      end
-    end
-  end
-
-  # If for some reason no instances managed to get fetched successfully then we'll just redirect to redirect.invidious.io
-  if filtered_instance_list.size == 0
-    return "redirect.invidious.io"
-  end
-
-  return filtered_instance_list.sample(1)[0]
-end
-
 def reduce_uri(uri : URI | String, max_length : Int32 = 50, suffix : String = "…") : String
   str = uri.to_s.sub(/^https?:\/\//, "")
   if str.size > max_length

src/invidious/jobs/instance_refresh_job.cr

@@ -0,0 +1,94 @@
+class Invidious::Jobs::InstanceListRefreshJob < Invidious::Jobs::BaseJob
+  # We update the internals of a constant as so it can be accessed from anywhere
+  # within the codebase
+  #
+  # "INSTANCES" => Array(Tuple(String, String))  # region, instance
+
+  INSTANCES = {"INSTANCES" => [] of Tuple(String, String)}
+
+  def initialize
+  end
+
+  def begin
+    loop do
+      refresh_instances
+      LOGGER.info("InstanceListRefreshJob: Done, sleeping for 30 minutes")
+      sleep 30.minute
+      Fiber.yield
+    end
+  end
+
+  # Refreshes the list of instances used for redirects.
+  #
+  # Does the following three checks for each instance
+  # -  Is it a clear-net instance?
+  # -  Is it an instance with a good uptime?
+  # -  Is it an updated instance?
+  private def refresh_instances
+    raw_instance_list = self.fetch_instances
+    filtered_instance_list = [] of Tuple(String, String)
+
+    raw_instance_list.each do |instance_data|
+      # TODO allow Tor hidden service instances when the current instance
+      # is also a hidden service. Same for i2p and any other non-clearnet instances.
+      begin
+        domain = instance_data[0]
+        info = instance_data[1]
+        stats = info["stats"]
+
+        next unless info["type"] == "https"
+        next if bad_uptime?(info["monitor"])
+        next if outdated?(stats["software"]["version"])
+
+        filtered_instance_list << {info["region"].as_s, domain.as_s}
+      rescue ex
+        if domain
+          LOGGER.info("InstanceListRefreshJob: failed to parse information from '#{domain}' because \"#{ex}\"\n\"#{ex.backtrace.join('\n')}\"  ")
+        else
+          LOGGER.info("InstanceListRefreshJob: failed to parse information from an instance because \"#{ex}\"\n\"#{ex.backtrace.join('\n')}\"  ")
+        end
+      end
+    end
+
+    if !filtered_instance_list.empty?
+      INSTANCES["INSTANCES"] = filtered_instance_list
+    end
+  end
+
+  # Fetches information regarding instances from api.invidious.io or an otherwise configured URL
+  private def fetch_instances : Array(JSON::Any)
+    begin
+      instance_api_client = make_client(URI.parse("https://api.invidious.io"))
+
+      # Timeouts
+      instance_api_client.connect_timeout = 10.seconds
+      instance_api_client.dns_timeout = 10.seconds
+
+      raw_instance_list = JSON.parse(instance_api_client.get("/instances.json").body).as_a
+      instance_api_client.close
+    rescue Socket::ConnectError | IO::TimeoutError | JSON::ParseException
+      raw_instance_list = [] of JSON::Any
+    end
+
+    return raw_instance_list
+  end
+
+  # Checks if the given target instance is outdated
+  private def outdated?(target_instance_version) : Bool
+    remote_commit_date = target_instance_version.as_s.match(/\d{4}\.\d{2}\.\d{2}/)
+    return false if !remote_commit_date
+
+    remote_commit_date = Time.parse(remote_commit_date[0], "%Y.%m.%d", Time::Location::UTC)
+    local_commit_date = Time.parse(CURRENT_VERSION, "%Y.%m.%d", Time::Location::UTC)
+
+    return (remote_commit_date - local_commit_date).abs.days > 30
+  end
+
+  # Checks if the uptime of the target instance is greater than 90% over a 30 day period
+  private def bad_uptime?(target_instance_health_monitor) : Bool
+    return false if !target_instance_health_monitor["statusClass"] == "success"
+    return false if target_instance_health_monitor["30dRatio"]["ratio"].as_s.to_f < 90
+
+    return true
+  end
+end

src/invidious/routes/misc.cr

@@ -40,7 +40,16 @@ module Invidious::Routes::Misc
 
   def self.cross_instance_redirect(env)
     referer = get_referer(env)
-    instance_url = fetch_random_instance
+
+    instance_list = Invidious::Jobs::InstanceListRefreshJob::INSTANCES["INSTANCES"]
+    if instance_list.empty?
+      instance_url = "redirect.invidious.io"
+    else
+      # Sample returns an array
+      # Instances are packaged as {region, domain} in the instance list
+      instance_url = instance_list.sample(1)[0][1]
+    end
+
     env.redirect "https://#{instance_url}#{referer}"
   end
 end