Merge pull request #1911 from SamantazFox/fix-long-playlists

Fix long playlists (100+ videos)
2021-03-23 18:07:01 +00:00 · 2021-03-23 18:07:01 +00:00 · c481ca924b
commit c481ca924b
parent 7566243151 23e5b6ba72
6 changed files with 119 additions and 81 deletions
--- a/spec/helpers_spec.cr
+++ b/spec/helpers_spec.cr
@ -47,19 +47,13 @@ describe "Helper" do
    end
  end

-  describe "#produce_playlist_url" do
-    it "correctly produces url for requesting index `x` of a playlist" do
-      produce_playlist_url("UUCla9fZca4I7KagBtgRGnOw", 0).should eq("/browse_ajax?continuation=4qmFsgIqEhpWTFVVQ2xhOWZaY2E0STdLYWdCdGdSR25PdxoMZWdaUVZEcERRVUU9&gl=US&hl=en")
+  describe "#produce_playlist_continuation" do
+    it "correctly produces ctoken for requesting index `x` of a playlist" do
+      produce_playlist_continuation("UUCla9fZca4I7KagBtgRGnOw", 100).should eq("4qmFsgJNEhpWTFVVQ2xhOWZaY2E0STdLYWdCdGdSR25PdxoUQ0FGNkJsQlVPa05IVVElM0QlM0SaAhhVVUNsYTlmWmNhNEk3S2FnQnRnUkduT3c%3D")

-      produce_playlist_url("UCCla9fZca4I7KagBtgRGnOw", 0).should eq("/browse_ajax?continuation=4qmFsgIqEhpWTFVVQ2xhOWZaY2E0STdLYWdCdGdSR25PdxoMZWdaUVZEcERRVUU9&gl=US&hl=en")
+      produce_playlist_continuation("UCCla9fZca4I7KagBtgRGnOw", 200).should eq("4qmFsgJLEhpWTFVVQ2xhOWZaY2E0STdLYWdCdGdSR25PdxoSQ0FKNkIxQlVPa05OWjBJJTNEmgIYVVVDbGE5ZlpjYTRJN0thZ0J0Z1JHbk93")

-      produce_playlist_url("PLt5AfwLFPxWLNVixpe1w3fi6lE2OTq0ET", 0).should eq("/browse_ajax?continuation=4qmFsgI0EiRWTFBMdDVBZndMRlB4V0xOVml4cGUxdzNmaTZsRTJPVHEwRVQaDGVnWlFWRHBEUVVFPQ%3D%3D&gl=US&hl=en")
-
-      produce_playlist_url("PLt5AfwLFPxWLNVixpe1w3fi6lE2OTq0ET", 10000).should eq("/browse_ajax?continuation=4qmFsgI0EiRWTFBMdDVBZndMRlB4V0xOVml4cGUxdzNmaTZsRTJPVHEwRVQaDGVnZFFWRHBEU2tKUA%3D%3D&gl=US&hl=en")
-
-      produce_playlist_url("PL55713C70BA91BD6E", 0).should eq("/browse_ajax?continuation=4qmFsgIkEhRWTFBMNTU3MTNDNzBCQTkxQkQ2RRoMZWdaUVZEcERRVUU9&gl=US&hl=en")
-
-      produce_playlist_url("PL55713C70BA91BD6E", 10000).should eq("/browse_ajax?continuation=4qmFsgIkEhRWTFBMNTU3MTNDNzBCQTkxQkQ2RRoMZWdkUVZEcERTa0pQ&gl=US&hl=en")
+      produce_playlist_continuation("PL55713C70BA91BD6E", 100).should eq("4qmFsgJBEhRWTFBMNTU3MTNDNzBCQTkxQkQ2RRoUQ0FGNkJsQlVPa05IVVElM0QlM0SaAhJQTDU1NzEzQzcwQkE5MUJENkU%3D")
    end
  end

--- a/src/invidious/channels.cr
+++ b/src/invidious/channels.cr
@ -229,18 +229,18 @@ def fetch_channel(ucid, db, pull_all_videos = true, locale = nil)
  page = 1

  LOGGER.trace("fetch_channel: #{ucid} : Downloading channel videos page")
-  response = get_channel_videos_response(ucid, page, auto_generated: auto_generated)
+  response_body = get_channel_videos_response(ucid, page, auto_generated: auto_generated)

  videos = [] of SearchVideo
  begin
-    initial_data = JSON.parse(response.body)
+    initial_data = JSON.parse(response_body)
    raise InfoException.new("Could not extract channel JSON") if !initial_data

    LOGGER.trace("fetch_channel: #{ucid} : Extracting videos from channel videos page initial_data")
    videos = extract_videos(initial_data.as_h, author, ucid)
  rescue ex
-    if response.body.includes?("To continue with your YouTube experience, please fill out the form below.") ||
-       response.body.includes?("https://www.google.com/sorry/index")
+    if response_body.includes?("To continue with your YouTube experience, please fill out the form below.") ||
+       response_body.includes?("https://www.google.com/sorry/index")
      raise InfoException.new("Could not extract channel info. Instance is likely blocked.")
    end
    raise ex
@ -304,8 +304,8 @@ def fetch_channel(ucid, db, pull_all_videos = true, locale = nil)
    ids = [] of String

    loop do
-      response = get_channel_videos_response(ucid, page, auto_generated: auto_generated)
-      initial_data = JSON.parse(response.body)
+      response_body = get_channel_videos_response(ucid, page, auto_generated: auto_generated)
+      initial_data = JSON.parse(response_body)
      raise InfoException.new("Could not extract channel JSON") if !initial_data
      videos = extract_videos(initial_data.as_h, author, ucid)

@ -447,6 +447,7 @@ def produce_channel_videos_continuation(ucid, page = 1, auto_generated = nil, so
  return continuation
 end

+# Used in bypass_captcha_job.cr
 def produce_channel_videos_url(ucid, page = 1, auto_generated = nil, sort_by = "newest", v2 = false)
  continuation = produce_channel_videos_continuation(ucid, page, auto_generated, sort_by, v2)
  return "/browse_ajax?continuation=#{continuation}&gl=US&hl=en"
@ -937,35 +938,19 @@ def get_about_info(ucid, locale)
  })
 end

-def get_channel_videos_response(ucid, page = 1, auto_generated = nil, sort_by = "newest", youtubei_browse = true)
-  if youtubei_browse
-    continuation = produce_channel_videos_continuation(ucid, page, auto_generated: auto_generated, sort_by: sort_by, v2: true)
-    data = {
-      "context": {
-        "client": {
-          "clientName":    "WEB",
-          "clientVersion": "2.20201021.03.00",
-        },
-      },
-      "continuation": continuation,
-    }.to_json
-    return YT_POOL.client &.post(
-      "/youtubei/v1/browse?key=AIzaSyAO_FJ2SlqU8Q4STEHLGCilw_Y9_11qcW8",
-      headers: HTTP::Headers{"content-type" => "application/json"},
-      body: data
-    )
-  else
-    url = produce_channel_videos_url(ucid, page, auto_generated: auto_generated, sort_by: sort_by, v2: true)
-    return YT_POOL.client &.get(url)
-  end
+def get_channel_videos_response(ucid, page = 1, auto_generated = nil, sort_by = "newest")
+  continuation = produce_channel_videos_continuation(ucid, page,
+    auto_generated: auto_generated, sort_by: sort_by, v2: true)
+
+  return request_youtube_api_browse(continuation)
 end

 def get_60_videos(ucid, author, page, auto_generated, sort_by = "newest")
  videos = [] of SearchVideo

  2.times do |i|
-    response = get_channel_videos_response(ucid, page * 2 + (i - 1), auto_generated: auto_generated, sort_by: sort_by)
-    initial_data = JSON.parse(response.body)
+    response_json = get_channel_videos_response(ucid, page * 2 + (i - 1), auto_generated: auto_generated, sort_by: sort_by)
+    initial_data = JSON.parse(response_json)
    break if !initial_data
    videos.concat extract_videos(initial_data.as_h, author, ucid)
  end
@ -974,8 +959,8 @@ def get_60_videos(ucid, author, page, auto_generated, sort_by = "newest")
 end

 def get_latest_videos(ucid)
-  response = get_channel_videos_response(ucid)
-  initial_data = JSON.parse(response.body)
+  response_json = get_channel_videos_response(ucid)
+  initial_data = JSON.parse(response_json)
  return [] of SearchVideo if !initial_data
  author = initial_data["metadata"]?.try &.["channelMetadataRenderer"]?.try &.["title"]?.try &.as_s
  items = extract_videos(initial_data.as_h, author, ucid)
--- a/src/invidious/helpers/youtube_api.cr
+++ b/src/invidious/helpers/youtube_api.cr
@ -0,0 +1,31 @@
+#
+# This file contains youtube API wrappers
+#
+
+# Hard-coded constants required by the API
+HARDCODED_API_KEY     = "AIzaSyAO_FJ2SlqU8Q4STEHLGCilw_Y9_11qcW8"
+HARDCODED_CLIENT_VERS = "2.20210318.08.00"
+
+def request_youtube_api_browse(continuation)
+  # JSON Request data, required by the API
+  data = {
+    "context": {
+      "client": {
+        "hl":            "en",
+        "gl":            "US",
+        "clientName":    "WEB",
+        "clientVersion": HARDCODED_CLIENT_VERS,
+      },
+    },
+    "continuation": continuation,
+  }
+
+  # Send the POST request and return result
+  response = YT_POOL.client &.post(
+    "/youtubei/v1/browse?key=#{HARDCODED_API_KEY}",
+    headers: HTTP::Headers{"content-type" => "application/json"},
+    body: data.to_json
+  )
+
+  return response.body
+end
--- a/src/invidious/playlists.cr
+++ b/src/invidious/playlists.cr
@ -307,23 +307,32 @@ def subscribe_playlist(db, user, playlist)
  return playlist
 end

-def produce_playlist_url(id, index)
+def produce_playlist_continuation(id, index)
  if id.starts_with? "UC"
    id = "UU" + id.lchop("UC")
  end
  plid = "VL" + id

+  # Emulate a "request counter" increment, to make perfectly valid
+  # ctokens, even if at the time of writing, it's ignored by youtube.
+  request_count = (index / 100).to_i64 || 1_i64
+
  data = {"1:varint" => index.to_i64}
    .try { |i| Protodec::Any.cast_json(i) }
    .try { |i| Protodec::Any.from_json(i) }
    .try { |i| Base64.urlsafe_encode(i, padding: false) }

+  data_wrapper = {"1:varint" => request_count, "15:string" => "PT:#{data}"}
+    .try { |i| Protodec::Any.cast_json(i) }
+    .try { |i| Protodec::Any.from_json(i) }
+    .try { |i| Base64.urlsafe_encode(i) }
+    .try { |i| URI.encode_www_form(i) }
+
  object = {
    "80226972:embedded" => {
      "2:string"  => plid,
-      "3:base64" => {
-        "15:string" => "PT:#{data}",
-      },
+      "3:string"  => data_wrapper,
+      "35:string" => id,
    },
  }

@ -332,7 +341,7 @@ def produce_playlist_url(id, index)
    .try { |i| Base64.urlsafe_encode(i) }
    .try { |i| URI.encode_www_form(i) }

-  return "/browse_ajax?continuation=#{continuation}&gl=US&hl=en"
+  return continuation
 end

 def get_playlist(db, plid, locale, refresh = true, force_refresh = false)
@ -427,47 +436,59 @@ def fetch_playlist(plid, locale)
 end

 def get_playlist_videos(db, playlist, offset, locale = nil, continuation = nil)
-  if playlist.is_a? InvidiousPlaylist
-    db.query_all("SELECT * FROM playlist_videos WHERE plid = $1 ORDER BY array_position($2, index) LIMIT 100 OFFSET $3", playlist.id, playlist.index, offset, as: PlaylistVideo)
-  else
-    fetch_playlist_videos(playlist.id, playlist.video_count, offset, locale, continuation)
-  end
-end
-
-def fetch_playlist_videos(plid, video_count, offset = 0, locale = nil, continuation = nil)
-  if continuation
-    response = YT_POOL.client &.get("/watch?v=#{continuation}&list=#{plid}&gl=US&hl=en")
-    initial_data = extract_initial_data(response.body)
-    offset = initial_data["currentVideoEndpoint"]?.try &.["watchEndpoint"]?.try &.["index"]?.try &.as_i64 || offset
-  end
-
-  if video_count > 100
-    url = produce_playlist_url(plid, offset)
-
-    response = YT_POOL.client &.get(url)
-    initial_data = JSON.parse(response.body).as_a.find(&.as_h.["response"]?).try &.as_h
-  elsif offset > 100
+  # Show empy playlist if requested page is out of range
+  if offset >= playlist.video_count
    return [] of PlaylistVideo
-  else # Extract first page of videos
-    response = YT_POOL.client &.get("/playlist?list=#{plid}&gl=US&hl=en")
+  end
+
+  if playlist.is_a? InvidiousPlaylist
+    db.query_all("SELECT * FROM playlist_videos WHERE plid = $1 ORDER BY array_position($2, index) LIMIT 100 OFFSET $3",
+      playlist.id, playlist.index, offset, as: PlaylistVideo)
+  else
+    if offset >= 100
+      # Normalize offset to match youtube's behavior (100 videos chunck per request)
+      offset = (offset / 100).to_i64 * 100_i64
+
+      ctoken = produce_playlist_continuation(playlist.id, offset)
+      initial_data = JSON.parse(request_youtube_api_browse(ctoken)).as_h
+    else
+      response = YT_POOL.client &.get("/playlist?list=#{playlist.id}&gl=US&hl=en")
      initial_data = extract_initial_data(response.body)
    end

-  return [] of PlaylistVideo if !initial_data
-  videos = extract_playlist_videos(initial_data)
-
-  until videos.empty? || videos[0].index == offset
-    videos.shift
+    if initial_data
+      return extract_playlist_videos(initial_data)
+    else
+      return [] of PlaylistVideo
+    end
  end
-
-  return videos
 end

 def extract_playlist_videos(initial_data : Hash(String, JSON::Any))
  videos = [] of PlaylistVideo

-  (initial_data["contents"]?.try &.["twoColumnBrowseResultsRenderer"]["tabs"].as_a.select(&.["tabRenderer"]["selected"]?.try &.as_bool)[0]["tabRenderer"]["content"]["sectionListRenderer"]["contents"][0]["itemSectionRenderer"]["contents"][0]["playlistVideoListRenderer"]["contents"].as_a ||
-    initial_data["response"]?.try &.["continuationContents"]["playlistVideoListContinuation"]["contents"].as_a).try &.each do |item|
+  if initial_data["contents"]?
+    tabs = initial_data["contents"]["twoColumnBrowseResultsRenderer"]["tabs"]
+    tabs_renderer = tabs.as_a.select(&.["tabRenderer"]["selected"]?.try &.as_bool)[0]["tabRenderer"]
+
+    # Watch out the two versions, with and without "s"
+    if tabs_renderer["contents"]? || tabs_renderer["content"]?
+      # Initial playlist data
+      tabs_contents = tabs_renderer.["contents"]? || tabs_renderer.["content"]
+
+      list_renderer = tabs_contents.["sectionListRenderer"]["contents"][0]
+      item_renderer = list_renderer.["itemSectionRenderer"]["contents"][0]
+      contents = item_renderer.["playlistVideoListRenderer"]["contents"].as_a
+    else
+      # Continuation data
+      contents = initial_data["onResponseReceivedActions"][0]?
+        .try &.["appendContinuationItemsAction"]["continuationItems"].as_a
+    end
+  else
+    contents = initial_data["response"]?.try &.["continuationContents"]["playlistVideoListContinuation"]["contents"].as_a
+  end
+
+  contents.try &.each do |item|
    if i = item["playlistVideoRenderer"]?
      video_id = i["navigationEndpoint"]["watchEndpoint"]["videoId"].as_s
      plid = i["navigationEndpoint"]["watchEndpoint"]["playlistId"].as_s
--- a/src/invidious/routes/playlists.cr
+++ b/src/invidious/routes/playlists.cr
@ -433,6 +433,13 @@ class Invidious::Routes::Playlists < Invidious::Routes::BaseRoute
      return error_template(500, ex)
    end

+    page_count = (playlist.video_count / 100).to_i
+    page_count = 1 if page_count == 0
+
+    if page > page_count
+      return env.redirect "/playlist?list=#{plid}&page=#{page_count}"
+    end
+
    if playlist.privacy == PlaylistPrivacy::Private && playlist.author != user.try &.email
      return error_template(403, "This playlist is private.")
    end
@ -440,7 +447,7 @@ class Invidious::Routes::Playlists < Invidious::Routes::BaseRoute
    begin
      videos = get_playlist_videos(PG_DB, playlist, offset: (page - 1) * 100, locale: locale)
    rescue ex
-      videos = [] of PlaylistVideo
+      return error_template(500, "Error encountered while retrieving playlist videos.<br>#{ex.message}")
    end

    if playlist.author == user.try &.email
--- a/src/invidious/views/playlist.ecr
+++ b/src/invidious/views/playlist.ecr
@ -103,7 +103,7 @@
    </div>
    <div class="pure-u-1 pure-u-lg-3-5"></div>
    <div class="pure-u-1 pure-u-lg-1-5" style="text-align:right">
-        <% if videos.size == 100 %>
+        <% if page_count != 1 && page < page_count %>
            <a href="/playlist?list=<%= playlist.id %>&page=<%= page + 1 %>">
                <%= translate(locale, "Next page") %>
            </a>