Harden /next fallback: fix null subreason, escaping, caching

Review fixes on top of the rebased /next-fallback work:

- Fix TypeCastError when subreason is stored as JSON null: accessors use
  as_s?, and error returns go through a helper that omits nil subreason.
- Restore graceful no-companion path by reverting player() to return nil,
  and drop the now-dead info.nil? branch in fetch_video.
- Re-enable video caching when Invidious Companion is not configured.
- Detect genuinely unavailable videos via playabilityStatus == "ERROR"
  instead of the localized "Video unavailable" string.
- HTML-escape reason/subreason coming from YouTube to avoid HTML injection,
  and guard against a missing "runs" key.
- Tidy published-date fallback and add a regression test.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Emilien 2026-06-26 18:35:51 +02:00
parent 24182a28d5
commit 8864b41b0f
4 changed files with 75 additions and 55 deletions

View File

@ -205,3 +205,19 @@ Spectator.describe "parse_video_info" do
expect(info["relatedVideos"].as_a.size).to eq(20)
end
end
Spectator.describe "Video error accessors" do
it "returns nil for a subreason stored as JSON null" do
# Some error paths store `subreason` as a JSON `null`; the accessor must
# not raise a TypeCastError on it.
info = {
"reason" => JSON::Any.new("Video unavailable"),
"subreason" => JSON::Any.new(nil),
} of String => JSON::Any
video = Video.new({id: "dQw4w9WgXcQ", info: info, updated: Time.utc})
expect(video.reason).to eq("Video unavailable")
expect(video.subreason).to be_nil
end
end

View File

@ -179,11 +179,13 @@ struct Video
end
def reason : String?
info["reason"]?.try &.as_s
# Use `as_s?` so a JSON `null` (stored by some error paths) yields `nil`
# instead of raising a TypeCastError.
info["reason"]?.try &.as_s?
end
def subreason : String?
info["subreason"]?.try &.as_s
info["subreason"]?.try &.as_s?
end
def music : Array(VideoMusic)
@ -310,7 +312,7 @@ def get_video(id, refresh = true, region = nil, force_refresh = false)
video.schema_version != Video::SCHEMA_VERSION # cache control
begin
video = fetch_video(id, region)
Invidious::Database::Videos.update(video) if CONFIG.invidious_companion.present?
Invidious::Database::Videos.update(video)
rescue ex
Invidious::Database::Videos.delete(id)
raise ex
@ -318,7 +320,7 @@ def get_video(id, refresh = true, region = nil, force_refresh = false)
end
else
video = fetch_video(id, region)
Invidious::Database::Videos.insert(video) if !region && CONFIG.invidious_companion.present?
Invidious::Database::Videos.insert(video) if !region
end
return video
@ -329,22 +331,10 @@ rescue DB::Error
end
def fetch_video(id, region)
# `extract_video_info` always returns a hash (it raises `NotFoundException`
# for genuinely unavailable videos), so there is no nil case to handle here.
info = Invidious::Videos::Parser.extract_video_info(video_id: id)
if info.nil?
raise InfoException.new("Invidious companion is not available. \
Video playback cannot continue. \
If you are the administrator of this instance, install Invidious companion \
following the installation instructions \
<a href=\"https://docs.invidious.io/installation/\">https://docs.invidious.io/installation/</a>")
end
if reason = info["reason"]?
if reason == "Video unavailable" && !info["title"]?
raise NotFoundException.new(reason.as_s || "")
end
end
video = Video.new({
id: id,
info: info,

View File

@ -55,6 +55,18 @@ module Invidious::Videos::Parser
}
end
# Builds the minimal "error" info hash returned when a video can't be
# played but we still want to render the watch page. `subreason` is only
# included when present, so it never gets stored as a JSON `null`.
private def self.error_video_info(reason : String?, subreason : String?) : Hash(String, JSON::Any)
params = {
"version" => JSON::Any.new(Video::SCHEMA_VERSION.to_i64),
} of String => JSON::Any
params["reason"] = JSON::Any.new(reason) if reason
params["subreason"] = JSON::Any.new(subreason) if subreason
params
end
def extract_video_info(video_id : String)
# Fetch data from the player endpoint
player_response = nil
@ -93,18 +105,21 @@ If you are the administrator, install Invidious Companion: \
reason ||= player_response.dig?("playabilityStatus", "errorScreen", "playerErrorMessageRenderer", "reason", "simpleText").try &.as_s
subreason_main = player_response.dig?("playabilityStatus", "errorScreen", "playerErrorMessageRenderer", "subreason")
subreason = subreason_main.try &.[]?("simpleText").try &.as_s
subreason ||= subreason_main.try &.[]("runs").as_a.map(&.[]("text")).join("")
subreason = subreason_main.try &.["simpleText"]?.try &.as_s
subreason ||= subreason_main.try &.["runs"]?.try &.as_a.map(&.["text"]).join("")
# Stop here if video is not found, or private with no further data.
# For other statuses (UNPLAYABLE, etc.), continue to extract as much
# data as possible from the /next endpoint.
if reason == "Video unavailable" && playability_status != "UNPLAYABLE"
return {
"version" => JSON::Any.new(Video::SCHEMA_VERSION.to_i64),
"reason" => JSON::Any.new(reason),
"subreason" => JSON::Any.new(subreason),
}
# These strings come straight from YouTube and are rendered as raw HTML
# on the watch/embed pages, so escape them to avoid HTML injection.
reason = HTML.escape(reason) if reason
subreason = HTML.escape(subreason) if subreason
# Stop here if the video is genuinely unavailable (deleted, removed or
# nonexistent): YouTube returns status "ERROR" for these regardless of
# the (localized) reason text, and there is nothing to extract from the
# /next endpoint. For other statuses (UNPLAYABLE, LOGIN_REQUIRED,
# LIVE_STREAM_OFFLINE, ...) we continue and extract as much as possible.
if playability_status == "ERROR"
raise NotFoundException.new(reason || "Video unavailable")
end
elsif video_id != player_response.dig?("videoDetails", "videoId")
# YouTube may return a different video player response than expected.
@ -115,10 +130,7 @@ If you are the administrator, install Invidious Companion: \
# wrong video means that we should count it as a failure.
Helpers.get_playback_statistic["totalRequests"] += 1
return {
"version" => JSON::Any.new(Video::SCHEMA_VERSION.to_i64),
"reason" => JSON::Any.new("Can't load the video on this Invidious instance. YouTube is currently trying to block Invidious instances. <a href=\"https://github.com/iv-org/invidious/issues/3822\">Click here for more info about the issue.</a>"),
}
return error_video_info("Can't load the video on this Invidious instance. YouTube is currently trying to block Invidious instances. <a href=\"https://github.com/iv-org/invidious/issues/3822\">Click here for more info about the issue.</a>", nil)
else
reason = nil
end
@ -134,11 +146,7 @@ If you are the administrator, install Invidious Companion: \
# If we're in an error state and /next fails, return what we have
if reason
LOGGER.debug("extract_video_info: /next failed for #{video_id}: #{ex.message}")
return {
"version" => JSON::Any.new(Video::SCHEMA_VERSION.to_i64),
"reason" => JSON::Any.new(reason),
"subreason" => JSON::Any.new(subreason),
}
return error_video_info(reason, subreason)
end
raise ex
end
@ -147,11 +155,7 @@ If you are the administrator, install Invidious Companion: \
params = self.parse_video_info(video_id, player_response)
rescue ex : BrokenTubeException
if reason
return {
"version" => JSON::Any.new(Video::SCHEMA_VERSION.to_i64),
"reason" => JSON::Any.new(reason),
"subreason" => JSON::Any.new(subreason),
}
return error_video_info(reason, subreason)
end
raise ex
end
@ -257,15 +261,22 @@ If you are the administrator, install Invidious Companion: \
.try { |t| Time.parse(t.as_s, "%Y-%m-%d", Time::Location::UTC) }
if published.nil?
# Fall back to the human-readable date shown on the watch page (e.g.
# "Started streaming 3 hours ago" or "Premiered Jan 1, 2024").
published_txt = video_primary_renderer
.try &.dig?("dateText", "simpleText")
.try &.dig?("dateText", "simpleText").try &.as_s
if published_txt.try &.as_s.includes?("ago") && !published_txt.nil?
published = decode_date(published_txt.as_s.lchop("Started streaming "))
elsif published_txt && published_txt.try &.as_s.matches?(/(\w{3} \d{1,2}, \d{4})$/)
published = Time.parse(published_txt.as_s.match!(/(\w{3} \d{1,2}, \d{4})$/)[0], "%b %-d, %Y", Time::Location::UTC)
date_regex = /(\w{3} \d{1,2}, \d{4})$/
published =
if published_txt.nil?
Time.utc
elsif published_txt.includes?("ago")
decode_date(published_txt.lchop("Started streaming "))
elsif match = published_txt.match(date_regex)
Time.parse(match[0], "%b %-d, %Y", Time::Location::UTC)
else
published = Time.utc
Time.utc
end
end
@ -295,12 +306,13 @@ If you are the administrator, install Invidious Companion: \
.try &.as_a.map &.as_s || [] of String
allow_ratings = video_details["allowRatings"]?.try &.as_bool
# When the microformat is missing (e.g. /next-only data) we can't know for
# sure, so optimistically assume the video is family safe.
family_friendly = microformat["isFamilySafe"]?.try &.as_bool
if family_friendly.nil?
family_friendly = true
end
family_friendly = true if family_friendly.nil?
is_listed = video_details["isCrawlable"]?.try &.as_bool
if is_listed.nil?
# Without videoDetails, infer from the "Unlisted" badge if present.
if video_badges = video_primary_renderer.try &.dig?("badges")
is_listed = !has_unlisted_badge?(video_badges)
else

View File

@ -463,7 +463,9 @@ module YoutubeAPI
if CONFIG.invidious_companion.present?
return self._post_invidious_companion("/youtubei/v1/player", data)
else
return self._post_json("/youtubei/v1/player", data, nil)
# Without Invidious Companion the player endpoint is unavailable.
# Returning nil lets the caller fall back to the /next endpoint.
return nil
end
end