Only modify cues for auto-generated captions
This commit is contained in:
parent
ab4df7e078
commit
5730280325
1 changed files with 35 additions and 26 deletions
|
@ -3203,45 +3203,54 @@ get "/api/v1/captions/:id" do |env|
|
||||||
caption = caption[0]
|
caption = caption[0]
|
||||||
end
|
end
|
||||||
|
|
||||||
caption_xml = client.get(caption.baseUrl + "&tlang=#{tlang}").body
|
url = caption.baseUrl + "&tlang=#{tlang}"
|
||||||
caption_xml = XML.parse(caption_xml)
|
|
||||||
|
|
||||||
webvtt = <<-END_VTT
|
# Auto-generated captions often have cues that aren't aligned properly with the video,
|
||||||
WEBVTT
|
# as well as some other markup that makes it cumbersome, so we try to fix that here
|
||||||
Kind: captions
|
if caption.name.simpleText.includes? "auto-generated"
|
||||||
Language: #{tlang || caption.languageCode}
|
caption_xml = client.get(url).body
|
||||||
|
caption_xml = XML.parse(caption_xml)
|
||||||
|
|
||||||
|
webvtt = <<-END_VTT
|
||||||
|
WEBVTT
|
||||||
|
Kind: captions
|
||||||
|
Language: #{tlang || caption.languageCode}
|
||||||
|
|
||||||
|
|
||||||
END_VTT
|
END_VTT
|
||||||
|
|
||||||
caption_nodes = caption_xml.xpath_nodes("//transcript/text")
|
caption_nodes = caption_xml.xpath_nodes("//transcript/text")
|
||||||
caption_nodes.each_with_index do |node, i|
|
caption_nodes.each_with_index do |node, i|
|
||||||
start_time = node["start"].to_f.seconds
|
start_time = node["start"].to_f.seconds
|
||||||
duration = node["dur"]?.try &.to_f.seconds
|
duration = node["dur"]?.try &.to_f.seconds
|
||||||
duration ||= start_time
|
duration ||= start_time
|
||||||
|
|
||||||
if caption_nodes.size > i + 1
|
if caption_nodes.size > i + 1
|
||||||
end_time = caption_nodes[i + 1]["start"].to_f.seconds
|
end_time = caption_nodes[i + 1]["start"].to_f.seconds
|
||||||
else
|
else
|
||||||
end_time = start_time + duration
|
end_time = start_time + duration
|
||||||
end
|
end
|
||||||
|
|
||||||
start_time = "#{start_time.hours.to_s.rjust(2, '0')}:#{start_time.minutes.to_s.rjust(2, '0')}:#{start_time.seconds.to_s.rjust(2, '0')}.#{start_time.milliseconds.to_s.rjust(3, '0')}"
|
start_time = "#{start_time.hours.to_s.rjust(2, '0')}:#{start_time.minutes.to_s.rjust(2, '0')}:#{start_time.seconds.to_s.rjust(2, '0')}.#{start_time.milliseconds.to_s.rjust(3, '0')}"
|
||||||
end_time = "#{end_time.hours.to_s.rjust(2, '0')}:#{end_time.minutes.to_s.rjust(2, '0')}:#{end_time.seconds.to_s.rjust(2, '0')}.#{end_time.milliseconds.to_s.rjust(3, '0')}"
|
end_time = "#{end_time.hours.to_s.rjust(2, '0')}:#{end_time.minutes.to_s.rjust(2, '0')}:#{end_time.seconds.to_s.rjust(2, '0')}.#{end_time.milliseconds.to_s.rjust(3, '0')}"
|
||||||
|
|
||||||
text = HTML.unescape(node.content)
|
text = HTML.unescape(node.content)
|
||||||
text = text.gsub(/<font color="#[a-fA-F0-9]{6}">/, "")
|
text = text.gsub(/<font color="#[a-fA-F0-9]{6}">/, "")
|
||||||
text = text.gsub(/<\/font>/, "")
|
text = text.gsub(/<\/font>/, "")
|
||||||
if md = text.match(/(?<name>.*) : (?<text>.*)/)
|
if md = text.match(/(?<name>.*) : (?<text>.*)/)
|
||||||
text = "<v #{md["name"]}>#{md["text"]}</v>"
|
text = "<v #{md["name"]}>#{md["text"]}</v>"
|
||||||
end
|
end
|
||||||
|
|
||||||
webvtt += <<-END_CUE
|
webvtt += <<-END_CUE
|
||||||
#{start_time} --> #{end_time}
|
#{start_time} --> #{end_time}
|
||||||
#{text}
|
#{text}
|
||||||
|
|
||||||
|
|
||||||
END_CUE
|
END_CUE
|
||||||
|
end
|
||||||
|
else
|
||||||
|
url += "&format=vtt"
|
||||||
|
webvtt = client.get(url).body
|
||||||
end
|
end
|
||||||
|
|
||||||
if title = env.params.query["title"]?
|
if title = env.params.query["title"]?
|
||||||
|
|
Loading…
Reference in a new issue