limit feeds and delete materialized views

.github/workflows/build-nightly-container.yml

@@ -86,7 +86,7 @@ jobs:
 
       # https://github.com/marketplace/actions/docker-manifest-create-action
       - name: Create and push manifest
-        uses: int128/docker-manifest-create-action@v2.22.0
+        uses: int128/docker-manifest-create-action@v2.21.0
         with:
           push: true
           tags: quay.io/invidious/invidious:master

.github/workflows/build-stable-container.yml

@@ -78,7 +78,7 @@ jobs:
 
       # https://github.com/marketplace/actions/docker-manifest-create-action
       - name: Create and push manifest
-        uses: int128/docker-manifest-create-action@v2.22.0
+        uses: int128/docker-manifest-create-action@v2.21.0
         with:
           push: true
           tags: quay.io/invidious/invidious:latest

config/migrate-scripts/migrate-db-8bc91ce.sh

@@ -0,0 +1,6 @@
+CREATE INDEX channel_videos_ucid_published_idx
+  ON public.channel_videos
+  USING btree
+  (ucid COLLATE pg_catalog."default", published);
+
+DROP INDEX channel_videos_ucid_idx;

config/sql/channel_videos.sql

@@ -19,12 +19,12 @@ CREATE TABLE IF NOT EXISTS public.channel_videos
 
 GRANT ALL ON TABLE public.channel_videos TO current_user;
 
--- Index: public.channel_videos_ucid_idx
+-- Index: public.channel_videos_ucid_published_idx
 
--- DROP INDEX public.channel_videos_ucid_idx;
+-- DROP INDEX public.channel_videos_ucid_published_idx;
 
-CREATE INDEX IF NOT EXISTS channel_videos_ucid_idx
+CREATE INDEX IF NOT EXISTS channel_videos_ucid_published_idx
   ON public.channel_videos
   USING btree
-  (ucid COLLATE pg_catalog."default");
+  (ucid COLLATE pg_catalog."default", published);
 

kubernetes/values.yaml

@@ -0,0 +1,60 @@
+name: invidious
+
+image:
+  repository: quay.io/invidious/invidious
+  tag: latest
+  pullPolicy: Always
+
+replicaCount: 1
+
+autoscaling:
+  enabled: false
+  minReplicas: 1
+  maxReplicas: 16
+  targetCPUUtilizationPercentage: 50
+
+service:
+  type: ClusterIP
+  port: 3000
+  #loadBalancerIP:
+
+resources: {}
+  #requests:
+  #  cpu: 100m
+  #  memory: 64Mi
+  #limits:
+  #  cpu: 800m
+  #  memory: 512Mi
+
+securityContext:
+  allowPrivilegeEscalation: false
+  runAsUser: 1000
+  runAsGroup: 1000
+  fsGroup: 1000
+
+# See https://github.com/bitnami/charts/tree/master/bitnami/postgresql
+postgresql:
+  image:
+    tag: 13
+  auth:
+    username: kemal
+    password: kemal
+    database: invidious
+  primary:
+    initdb:
+      username: kemal
+      password: kemal
+      scriptsConfigMap: invidious-postgresql-init
+
+# Adapted from ../config/config.yml
+config:
+  channel_threads: 1
+  db:
+    user: kemal
+    password: kemal
+    host: invidious-postgresql
+    port: 5432
+    dbname: invidious
+  full_refresh: false
+  https_only: false
+  domain:

src/invidious.cr

@@ -107,14 +107,6 @@ Kemal.config.extra_options do |parser|
       exit
     end
   end
-  parser.on("-f THREADS", "--feed-threads=THREADS", "Number of threads for refreshing feeds (default: #{CONFIG.feed_threads})") do |number|
-    begin
-      CONFIG.feed_threads = number.to_i
-    rescue ex
-      puts "THREADS must be integer"
-      exit
-    end
-  end
   parser.on("-o OUTPUT", "--output=OUTPUT", "Redirect output (default: #{CONFIG.output})") do |output|
     CONFIG.output = output
   end
@@ -166,10 +158,6 @@ if CONFIG.channel_threads > 0
   Invidious::Jobs.register Invidious::Jobs::RefreshChannelsJob.new(PG_DB)
 end
 
-if CONFIG.feed_threads > 0
-  Invidious::Jobs.register Invidious::Jobs::RefreshFeedsJob.new(PG_DB)
-end
-
 if CONFIG.statistics_enabled
   Invidious::Jobs.register Invidious::Jobs::StatisticsRefreshJob.new(PG_DB, SOFTWARE)
 end

src/invidious/config.cr

@@ -95,8 +95,6 @@ class Config
   # Time interval between two executions of the job that crawls channel videos (subscriptions update).
   @[YAML::Field(converter: Preferences::TimeSpanConverter)]
   property channel_refresh_interval : Time::Span = 30.minutes
-  # Number of threads to use for updating feeds
-  property feed_threads : Int32 = 1
   # Log file path or STDOUT
   property output : String = "STDOUT"
   # Default log level, valid YAML values are ints and strings, see src/invidious/helpers/logger.cr

src/invidious/helpers/errors.cr

@@ -201,7 +201,7 @@ def error_redirect_helper(env : HTTP::Server::Context)
           <a href="/redirect?referer=#{env.get("current_page")}">#{switch_instance}</a>
         </li>
         <li>
-          <a rel="noreferrer noopener" href="https://www.youtube.com#{env.request.resource}">#{go_to_youtube}</a>
+          <a rel="noreferrer noopener" href="https://youtube.com#{env.request.resource}">#{go_to_youtube}</a>
         </li>
       </ul>
     END_HTML

src/invidious/jobs/refresh_feeds_job.cr

@@ -1,75 +0,0 @@
-class Invidious::Jobs::RefreshFeedsJob < Invidious::Jobs::BaseJob
-  private getter db : DB::Database
-
-  def initialize(@db)
-  end
-
-  def begin
-    max_fibers = CONFIG.feed_threads
-    active_fibers = 0
-    active_channel = ::Channel(Bool).new
-
-    loop do
-      db.query("SELECT email FROM users WHERE feed_needs_update = true OR feed_needs_update IS NULL") do |rs|
-        rs.each do
-          email = rs.read(String)
-          view_name = "subscriptions_#{sha256(email)}"
-
-          if active_fibers >= max_fibers
-            if active_channel.receive
-              active_fibers -= 1
-            end
-          end
-
-          active_fibers += 1
-          spawn do
-            begin
-              # Drop outdated views
-              column_array = Invidious::Database.get_column_array(db, view_name)
-              ChannelVideo.type_array.each_with_index do |name, i|
-                if name != column_array[i]?
-                  LOGGER.info("RefreshFeedsJob: DROP MATERIALIZED VIEW #{view_name}")
-                  db.exec("DROP MATERIALIZED VIEW #{view_name}")
-                  raise "view does not exist"
-                end
-              end
-
-              if !db.query_one("SELECT pg_get_viewdef('#{view_name}')", as: String).includes? "WHERE ((cv.ucid = ANY (u.subscriptions))"
-                LOGGER.info("RefreshFeedsJob: Materialized view #{view_name} is out-of-date, recreating...")
-                db.exec("DROP MATERIALIZED VIEW #{view_name}")
-              end
-
-              db.exec("REFRESH MATERIALIZED VIEW #{view_name}")
-              db.exec("UPDATE users SET feed_needs_update = false WHERE email = $1", email)
-            rescue ex
-              # Rename old views
-              begin
-                legacy_view_name = "subscriptions_#{sha256(email)[0..7]}"
-
-                db.exec("SELECT * FROM #{legacy_view_name} LIMIT 0")
-                LOGGER.info("RefreshFeedsJob: RENAME MATERIALIZED VIEW #{legacy_view_name}")
-                db.exec("ALTER MATERIALIZED VIEW #{legacy_view_name} RENAME TO #{view_name}")
-              rescue ex
-                begin
-                  # While iterating through, we may have an email stored from a deleted account
-                  if db.query_one?("SELECT true FROM users WHERE email = $1", email, as: Bool)
-                    LOGGER.info("RefreshFeedsJob: CREATE #{view_name}")
-                    db.exec("CREATE MATERIALIZED VIEW #{view_name} AS #{MATERIALIZED_VIEW_SQL.call(email)}")
-                    db.exec("UPDATE users SET feed_needs_update = false WHERE email = $1", email)
-                  end
-                rescue ex
-                  LOGGER.error("RefreshFeedJobs: REFRESH #{email} : #{ex.message}")
-                end
-              end
-            end
-
-            active_channel.send(true)
-          end
-        end
-      end
-
-      sleep 5.seconds
-      Fiber.yield
-    end
-  end
-end

src/invidious/routes/account.cr

@@ -123,10 +123,8 @@ module Invidious::Routes::Account
       return error_template(400, ex)
     end
 
-    view_name = "subscriptions_#{sha256(user.email)}"
     Invidious::Database::Users.delete(user)
     Invidious::Database::SessionIDs.delete(email: user.email)
-    PG_DB.exec("DROP MATERIALIZED VIEW #{view_name}")
 
     env.request.cookies.each do |cookie|
       cookie.expires = Time.utc(1990, 1, 1)

src/invidious/routes/channels.cr

@@ -351,7 +351,7 @@ module Invidious::Routes::Channels
     invidious_url_params.delete_all("user")
 
     begin
-      resolved_url = YoutubeAPI.resolve_url("https://www.youtube.com#{env.request.path}#{yt_url_params.size > 0 ? "?#{yt_url_params}" : ""}")
+      resolved_url = YoutubeAPI.resolve_url("https://youtube.com#{env.request.path}#{yt_url_params.size > 0 ? "?#{yt_url_params}" : ""}")
       ucid = resolved_url["endpoint"]["browseEndpoint"]["browseId"]
     rescue ex : InfoException | KeyError
       return error_template(404, I18n.translate(locale, "This channel does not exist."))

src/invidious/routes/errors.cr

@@ -8,7 +8,7 @@ module Invidious::Routes::ErrorRoutes
     if md = env.request.path.match(/^\/(?<id>([a-zA-Z0-9_-]{11})|(\w+))$/)
       item = md["id"]
 
-      # Check if item is branding URL e.g. https://www.youtube.com/gaming
+      # Check if item is branding URL e.g. https://youtube.com/gaming
       response = YT_POOL.client &.get("/#{item}")
 
       if response.status_code == 301

src/invidious/routes/login.cr

@@ -125,9 +125,6 @@ module Invidious::Routes::Login
         Invidious::Database::Users.insert(user)
         Invidious::Database::SessionIDs.insert(sid, email)
 
-        view_name = "subscriptions_#{sha256(user.email)}"
-        PG_DB.exec("CREATE MATERIALIZED VIEW #{view_name} AS #{MATERIALIZED_VIEW_SQL.call(user.email)}")
-
         if alt = CONFIG.alternative_domains.index(host)
           env.response.cookies["SID"] = Invidious::User::Cookies.sid(CONFIG.alternative_domains[alt], sid)
         else

src/invidious/search/processors.cr

@@ -37,18 +37,18 @@ module Invidious::Search
 
     # Search inside of user subscriptions
     def subscriptions(query : Query, user : Invidious::User) : Array(ChannelVideo)
-      view_name = "subscriptions_#{sha256(user.email)}"
-
       return PG_DB.query_all("
         SELECT id,title,published,updated,ucid,author,length_seconds
         FROM (
-          SELECT *,
-          to_tsvector(#{view_name}.title) ||
-          to_tsvector(#{view_name}.author)
-          as document
-          FROM #{view_name}
-        ) v_search WHERE v_search.document @@ plainto_tsquery($1) LIMIT 20 OFFSET $2;",
-        query.text, (query.page - 1) * 20,
+          SELECT cv.*,
+          to_tsvector(cv.title) ||
+          to_tsvector(cv.author) AS document
+          FROM channel_videos cv
+          JOIN users ON cv.ucid = any(users.subscriptions)
+          WHERE users.email = $1 AND published > now() - interval '1 month'
+          ORDER BY published
+        ) v_search WHERE v_search.document @@ plainto_tsquery($2) LIMIT 20 OFFSET $3;",
+        user.email, query.text, (query.page - 1) * 20,
         as: ChannelVideo
       )
     end

src/invidious/users.cr

@@ -27,7 +27,6 @@ def get_subscription_feed(user, max_results = 40, page = 1)
   offset = (page - 1) * limit
 
   notifications = Invidious::Database::Users.select_notifications(user)
-  view_name = "subscriptions_#{sha256(user.email)}"
 
   if user.preferences.notifications_only && !notifications.empty?
     # Only show notifications
@@ -53,33 +52,39 @@ def get_subscription_feed(user, max_results = 40, page = 1)
         # Show latest video from a channel that a user hasn't watched
         # "unseen_only" isn't really correct here, more accurate would be "unwatched_only"
 
-        if user.watched.empty?
-          values = "'{}'"
-        else
-          values = "VALUES #{user.watched.map { |id| %(('#{id}')) }.join(",")}"
-        end
-        videos = PG_DB.query_all("SELECT DISTINCT ON (ucid) * FROM #{view_name} WHERE NOT id = ANY (#{values}) ORDER BY ucid, published DESC", as: ChannelVideo)
+        # "SELECT cv.* FROM channel_videos cv JOIN users ON cv.ucid = any(users.subscriptions) WHERE users.email = $1 AND published > now() - interval '1 month' ORDER BY published DESC"
+        # "SELECT DISTINCT ON (cv.ucid) cv.* FROM channel_videos cv JOIN users ON cv.ucid = any(users.subscriptions) WHERE users.email = ? AND NOT cv.id = any(users.watched) AND published > now() - interval '1 month' ORDER BY ucid, published DESC"
+        videos = PG_DB.query_all("SELECT DISTINCT ON (cv.ucid) cv.* " \
+                                 "FROM channel_videos cv " \
+                                 "JOIN users ON cv.ucid = any(users.subscriptions) " \
+                                 "WHERE users.email = $1 AND NOT cv.id = any(users.watched) AND published > now() - interval '1 month' " \
+                                 "ORDER BY ucid, published DESC", user.email, as: ChannelVideo)
       else
         # Show latest video from each channel
 
-        videos = PG_DB.query_all("SELECT DISTINCT ON (ucid) * FROM #{view_name} ORDER BY ucid, published DESC", as: ChannelVideo)
+        videos = PG_DB.query_all("SELECT DISTINCT ON (cv.ucid) cv.* " \
+                                 "FROM channel_videos cv " \
+                                 "JOIN users ON cv.ucid = any(users.subscriptions) " \
+                                 "WHERE users.email = $1 AND published > now() - interval '1 month' " \
+                                 "ORDER BY ucid, published DESC", user.email, as: ChannelVideo)
       end
 
       videos.sort_by!(&.published).reverse!
     else
       if user.preferences.unseen_only
         # Only show unwatched
-
-        if user.watched.empty?
-          values = "'{}'"
-        else
-          values = "VALUES #{user.watched.map { |id| %(('#{id}')) }.join(",")}"
-        end
-        videos = PG_DB.query_all("SELECT * FROM #{view_name} WHERE NOT id = ANY (#{values}) ORDER BY published DESC LIMIT $1 OFFSET $2", limit, offset, as: ChannelVideo)
+        videos = PG_DB.query_all("SELECT cv.* " \
+                                 "FROM channel_videos cv " \
+                                 "JOIN users ON cv.ucid = any(users.subscriptions) " \
+                                 "WHERE users.email = $1 AND NOT cv.id = any(users.watched) AND published > now() - interval '1 month' " \
+                                 "ORDER BY published DESC LIMIT $2 OFFSET $3", user.email, limit, offset, as: ChannelVideo)
       else
         # Sort subscriptions as normal
-
-        videos = PG_DB.query_all("SELECT * FROM #{view_name} ORDER BY published DESC LIMIT $1 OFFSET $2", limit, offset, as: ChannelVideo)
+        videos = PG_DB.query_all("SELECT cv.* " \
+                                 "FROM channel_videos cv " \
+                                 "JOIN users ON cv.ucid = any(users.subscriptions) " \
+                                 "WHERE users.email = $1 AND published > now() - interval '1 month' " \
+                                 "ORDER BY published DESC LIMIT $2 OFFSET $3", user.email, limit, offset, as: ChannelVideo)
       end
     end
 

src/invidious/yt_backend/youtube_api.cr

@@ -480,7 +480,7 @@ module YoutubeAPI
   #
   # ```
   # # Valid channel "brand URL" gives the related UCID and browse ID
-  # channel_a = YoutubeAPI.resolve_url("https://www.youtube.com/c/google")
+  # channel_a = YoutubeAPI.resolve_url("https://youtube.com/c/google")
   # channel_a # => {
   #   "endpoint": {
   #     "browseEndpoint": {
@@ -492,7 +492,7 @@ module YoutubeAPI
   # }
   #
   # # Invalid URL returns throws an InfoException
-  # channel_b = YoutubeAPI.resolve_url("https://www.youtube.com/c/invalid")
+  # channel_b = YoutubeAPI.resolve_url("https://youtube.com/c/invalid")
   # ```
   #
   def resolve_url(url : String, client_config : ClientConfig | Nil = nil)