Rewrite transcript logic to be more generic

The transcript logic in Invidious was written specifically
as a workaround for captions, and not transcripts as a feature.

This commit genericises the logic a bit as so it can be used for
implementing transcripts within Invidious' API and UI as well.

The most notable change is the added parsing of section headings
when it was previously skipped over in favor of regular lines.
pull/4747/head
syeopite 2024-06-11 17:57:33 -07:00
rodzic eda7444ca4
commit 0224162ad2
Nie znaleziono w bazie danych klucza dla tego podpisu
ID klucza GPG: A73C186DA3955A1A
2 zmienionych plików z 63 dodań i 36 usunięć

Wyświetl plik

@ -89,9 +89,14 @@ module Invidious::Routes::API::V1::Videos
if CONFIG.use_innertube_for_captions
params = Invidious::Videos::Transcript.generate_param(id, caption.language_code, caption.auto_generated)
initial_data = YoutubeAPI.get_transcript(params)
webvtt = Invidious::Videos::Transcript.convert_transcripts_to_vtt(initial_data, caption.language_code)
transcript = Invidious::Videos::Transcript.from_raw(
YoutubeAPI.get_transcript(params),
caption.language_code,
caption.auto_generated
)
webvtt = transcript.to_vtt
else
# Timedtext API handling
url = URI.parse("#{caption.base_url}&tlang=#{tlang}").request_target

Wyświetl plik

@ -1,8 +1,21 @@
module Invidious::Videos
# Namespace for methods primarily relating to Transcripts
module Transcript
record TranscriptLine, start_ms : Time::Span, end_ms : Time::Span, line : String
# A `Transcripts` struct encapsulates a sequence of lines that together forms the whole transcript for a given YouTube video.
# These lines can be categorized into two types: section headings and regular lines representing content from the video.
struct Transcript
# Types
record HeadingLine, start_ms : Time::Span, end_ms : Time::Span, line : String
record RegularLine, start_ms : Time::Span, end_ms : Time::Span, line : String
alias TranscriptLine = HeadingLine | RegularLine
property lines : Array(TranscriptLine)
property language_code : String
property auto_generated : Bool
# Initializes a new Transcript struct with the contents and associated metadata describing it
def initialize(@lines : Array(TranscriptLine), @language_code : String, @auto_generated : Bool)
end
# Generates a protobuf string to fetch the requested transcript from YouTube
def self.generate_param(video_id : String, language_code : String, auto_generated : Bool) : String
kind = auto_generated ? "asr" : ""
@ -30,48 +43,57 @@ module Invidious::Videos
return params
end
def self.convert_transcripts_to_vtt(initial_data : Hash(String, JSON::Any), target_language : String) : String
# Convert into array of TranscriptLine
lines = self.parse(initial_data)
# Constructs a Transcripts struct from the initial YouTube response
def self.from_raw(initial_data : Hash(String, JSON::Any), language_code : String, auto_generated : Bool)
body = initial_data.dig("actions", 0, "updateEngagementPanelAction", "content", "transcriptRenderer",
"content", "transcriptSearchPanelRenderer", "body", "transcriptSegmentListRenderer",
"initialSegments").as_a
lines = [] of TranscriptLine
body.each do |line|
if unpacked_line = line["transcriptSectionHeaderRenderer"]?
line_type = HeadingLine
else
unpacked_line = line["transcriptSegmentRenderer"]
line_type = RegularLine
end
start_ms = unpacked_line["startMs"].as_s.to_i.millisecond
end_ms = unpacked_line["endMs"].as_s.to_i.millisecond
text = extract_text(unpacked_line["snippet"]) || ""
lines << line_type.new(start_ms, end_ms, text)
end
return Transcript.new(
lines: lines,
language_code: language_code,
auto_generated: auto_generated,
)
end
# Converts transcript lines to a WebVTT file
#
# This is used within Invidious to replace subtitles
# as to workaround YouTube's rate-limited timedtext endpoint.
def to_vtt
settings_field = {
"Kind" => "captions",
"Language" => target_language,
"Language" => @language_code,
}
# Taken from Invidious::Videos::Captions::Metadata.timedtext_to_vtt()
vtt = WebVTT.build(settings_field) do |vtt|
lines.each do |line|
@lines.each do |line|
# Section headers are excluded from the VTT conversion as to
# match the regular captions returned from YouTube as much as possible
next if line.is_a? HeadingLine
vtt.cue(line.start_ms, line.end_ms, line.line)
end
end
return vtt
end
private def self.parse(initial_data : Hash(String, JSON::Any))
body = initial_data.dig("actions", 0, "updateEngagementPanelAction", "content", "transcriptRenderer",
"content", "transcriptSearchPanelRenderer", "body", "transcriptSegmentListRenderer",
"initialSegments").as_a
lines = [] of TranscriptLine
body.each do |line|
# Transcript section headers. They are not apart of the captions and as such we can safely skip them.
if line.as_h.has_key?("transcriptSectionHeaderRenderer")
next
end
line = line["transcriptSegmentRenderer"]
start_ms = line["startMs"].as_s.to_i.millisecond
end_ms = line["endMs"].as_s.to_i.millisecond
text = extract_text(line["snippet"]) || ""
lines << TranscriptLine.new(start_ms, end_ms, text)
end
return lines
end
end
end