mirror of
				https://github.com/iv-org/invidious.git
				synced 2025-10-31 04:32:02 +00:00 
			
		
		
		
	Add support for timedtext captions (#3533)
This commit is contained in:
		| @@ -93,45 +93,50 @@ module Invidious::Routes::API::V1::Videos | ||||
|     # as well as some other markup that makes it cumbersome, so we try to fix that here | ||||
|     if caption.name.includes? "auto-generated" | ||||
|       caption_xml = YT_POOL.client &.get(url).body | ||||
|       caption_xml = XML.parse(caption_xml) | ||||
|  | ||||
|       webvtt = String.build do |str| | ||||
|         str << <<-END_VTT | ||||
|         WEBVTT | ||||
|         Kind: captions | ||||
|         Language: #{tlang || caption.language_code} | ||||
|       if caption_xml.starts_with?("<?xml") | ||||
|         webvtt = caption.timedtext_to_vtt(caption_xml, tlang) | ||||
|       else | ||||
|         caption_xml = XML.parse(caption_xml) | ||||
|  | ||||
|         webvtt = String.build do |str| | ||||
|           str << <<-END_VTT | ||||
|           WEBVTT | ||||
|           Kind: captions | ||||
|           Language: #{tlang || caption.language_code} | ||||
|  | ||||
|  | ||||
|         END_VTT | ||||
|           END_VTT | ||||
|  | ||||
|         caption_nodes = caption_xml.xpath_nodes("//transcript/text") | ||||
|         caption_nodes.each_with_index do |node, i| | ||||
|           start_time = node["start"].to_f.seconds | ||||
|           duration = node["dur"]?.try &.to_f.seconds | ||||
|           duration ||= start_time | ||||
|           caption_nodes = caption_xml.xpath_nodes("//transcript/text") | ||||
|           caption_nodes.each_with_index do |node, i| | ||||
|             start_time = node["start"].to_f.seconds | ||||
|             duration = node["dur"]?.try &.to_f.seconds | ||||
|             duration ||= start_time | ||||
|  | ||||
|           if caption_nodes.size > i + 1 | ||||
|             end_time = caption_nodes[i + 1]["start"].to_f.seconds | ||||
|           else | ||||
|             end_time = start_time + duration | ||||
|             if caption_nodes.size > i + 1 | ||||
|               end_time = caption_nodes[i + 1]["start"].to_f.seconds | ||||
|             else | ||||
|               end_time = start_time + duration | ||||
|             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')}" | ||||
|             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 = text.gsub(/<font color="#[a-fA-F0-9]{6}">/, "") | ||||
|             text = text.gsub(/<\/font>/, "") | ||||
|             if md = text.match(/(?<name>.*) : (?<text>.*)/) | ||||
|               text = "<v #{md["name"]}>#{md["text"]}</v>" | ||||
|             end | ||||
|  | ||||
|             str << <<-END_CUE | ||||
|             #{start_time} --> #{end_time} | ||||
|             #{text} | ||||
|  | ||||
|  | ||||
|             END_CUE | ||||
|           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')}" | ||||
|           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 = text.gsub(/<font color="#[a-fA-F0-9]{6}">/, "") | ||||
|           text = text.gsub(/<\/font>/, "") | ||||
|           if md = text.match(/(?<name>.*) : (?<text>.*)/) | ||||
|             text = "<v #{md["name"]}>#{md["text"]}</v>" | ||||
|           end | ||||
|  | ||||
|           str << <<-END_CUE | ||||
|           #{start_time} --> #{end_time} | ||||
|           #{text} | ||||
|  | ||||
|  | ||||
|           END_CUE | ||||
|         end | ||||
|       end | ||||
|     else | ||||
| @@ -141,7 +146,12 @@ module Invidious::Routes::API::V1::Videos | ||||
|       # | ||||
|       # See: https://github.com/iv-org/invidious/issues/2391 | ||||
|       webvtt = YT_POOL.client &.get("#{url}&format=vtt").body | ||||
|         .gsub(/([0-9:.]{12} --> [0-9:.]{12}).+/, "\\1") | ||||
|       if webvtt.starts_with?("<?xml") | ||||
|         webvtt = caption.timedtext_to_vtt(webvtt) | ||||
|       else | ||||
|         webvtt = YT_POOL.client &.get("#{url}&format=vtt").body | ||||
|           .gsub(/([0-9:.]{12} --> [0-9:.]{12}).+/, "\\1") | ||||
|       end | ||||
|     end | ||||
|  | ||||
|     if title = env.params.query["title"]? | ||||
|   | ||||
| @@ -31,6 +31,72 @@ module Invidious::Videos | ||||
|       return captions_list | ||||
|     end | ||||
|  | ||||
|     def timedtext_to_vtt(timedtext : String, tlang = nil) : String | ||||
|       # In the future, we could just directly work with the url. This is more of a POC | ||||
|       cues = [] of XML::Node | ||||
|       tree = XML.parse(timedtext) | ||||
|       tree = tree.children.first | ||||
|  | ||||
|       tree.children.each do |item| | ||||
|         if item.name == "body" | ||||
|           item.children.each do |cue| | ||||
|             if cue.name == "p" && !(cue.children.size == 1 && cue.children[0].content == "\n") | ||||
|               cues << cue | ||||
|             end | ||||
|           end | ||||
|           break | ||||
|         end | ||||
|       end | ||||
|       result = String.build do |result| | ||||
|         result << <<-END_VTT | ||||
|         WEBVTT | ||||
|         Kind: captions | ||||
|         Language: #{tlang || @language_code} | ||||
|  | ||||
|  | ||||
|         END_VTT | ||||
|  | ||||
|         result << "\n\n" | ||||
|  | ||||
|         cues.each_with_index do |node, i| | ||||
|           start_time = node["t"].to_f.milliseconds | ||||
|  | ||||
|           duration = node["d"]?.try &.to_f.milliseconds | ||||
|  | ||||
|           duration ||= start_time | ||||
|  | ||||
|           if cues.size > i + 1 | ||||
|             end_time = cues[i + 1]["t"].to_f.milliseconds | ||||
|           else | ||||
|             end_time = start_time + duration | ||||
|           end | ||||
|  | ||||
|           # start_time | ||||
|           result << start_time.hours.to_s.rjust(2, '0') | ||||
|           result << ':' << start_time.minutes.to_s.rjust(2, '0') | ||||
|           result << ':' << start_time.seconds.to_s.rjust(2, '0') | ||||
|           result << '.' << start_time.milliseconds.to_s.rjust(3, '0') | ||||
|  | ||||
|           result << " --> " | ||||
|  | ||||
|           # end_time | ||||
|           result << end_time.hours.to_s.rjust(2, '0') | ||||
|           result << ':' << end_time.minutes.to_s.rjust(2, '0') | ||||
|           result << ':' << end_time.seconds.to_s.rjust(2, '0') | ||||
|           result << '.' << end_time.milliseconds.to_s.rjust(3, '0') | ||||
|  | ||||
|           result << "\n" | ||||
|  | ||||
|           node.children.each do |s| | ||||
|             result << s.content | ||||
|           end | ||||
|           result << "\n" | ||||
|           result << "\n" | ||||
|         end | ||||
|       end | ||||
|       return result | ||||
|     end | ||||
|  | ||||
|     # List of all caption languages available on Youtube. | ||||
|     LANGUAGES = { | ||||
|       "", | ||||
|   | ||||
		Reference in New Issue
	
	Block a user
	 Samantaz Fox
					Samantaz Fox