Merge pull request #1490 from tirz/fix-regex_parsing

fix: ytInitialData parsing with regex

src/invidious/channels.cr

@@ -775,41 +775,34 @@ def extract_channel_community_cursor(continuation)
   cursor
 end
 
-INITDATA_PREQUERY = "window[\"ytInitialData\"] = {"
-
 def get_about_info(ucid, locale)
-  about = YT_POOL.client &.get("/channel/#{ucid}/about?gl=US&hl=en")
+  result = YT_POOL.client &.get("/channel/#{ucid}/about?gl=US&hl=en")
-  if about.status_code != 200
+  if result.status_code != 200
-    about = YT_POOL.client &.get("/user/#{ucid}/about?gl=US&hl=en")
+    result = YT_POOL.client &.get("/user/#{ucid}/about?gl=US&hl=en")
   end
 
-  if md = about.headers["location"]?.try &.match(/\/channel\/(?<ucid>UC[a-zA-Z0-9_-]{22})/)
+  if md = result.headers["location"]?.try &.match(/\/channel\/(?<ucid>UC[a-zA-Z0-9_-]{22})/)
     raise ChannelRedirect.new(channel_id: md["ucid"])
   end
 
-  if about.status_code != 200
+  if result.status_code != 200
     error_message = translate(locale, "This channel does not exist.")
     raise error_message
   end
 
-  initdata_pre = about.body.index(INITDATA_PREQUERY)
+  about = XML.parse_html(result.body)
-  initdata_post = initdata_pre.nil? ? nil : about.body.index("};", initdata_pre)
-  if initdata_post.nil?
-    about = XML.parse_html(about.body)
-    error_message = about.xpath_node(%q(//div[@class="yt-alert-content"])).try &.content.strip
-    error_message ||= translate(locale, "Could not get channel info.")
-    raise error_message
-  end
-  initdata_pre = initdata_pre.not_nil! + INITDATA_PREQUERY.size - 1
-
-  initdata = JSON.parse(about.body[initdata_pre, initdata_post - initdata_pre + 1])
-  about = XML.parse_html(about.body)
-
   if about.xpath_node(%q(//div[contains(@class, "channel-empty-message")]))
     error_message = translate(locale, "This channel does not exist.")
     raise error_message
   end
 
+  initdata = extract_initial_data(result.body)
+  if initdata.empty?
+    error_message = about.xpath_node(%q(//div[@class="yt-alert-content"])).try &.content.strip
+    error_message ||= translate(locale, "Could not get channel info.")
+    raise error_message
+  end
+
   author = about.xpath_node(%q(//meta[@name="title"])).not_nil!["content"]
   author_url = about.xpath_node(%q(//link[@rel="canonical"])).not_nil!["href"]
   author_thumbnail = about.xpath_node(%q(//link[@rel="image_src"])).not_nil!["href"]

src/invidious/helpers/helpers.cr

@@ -598,12 +598,7 @@ def create_notification_stream(env, topics, connection_channel)
 end
 
 def extract_initial_data(body) : Hash(String, JSON::Any)
-  initial_data = body.match(/(window\["ytInitialData"\]|var\s+ytInitialData)\s*=\s*(?<info>.*?);+\s*\n/).try &.["info"] || "{}"
+  return JSON.parse(body.match(/(window\["ytInitialData"\]|var\s*ytInitialData)\s*=\s*(JSON\.parse\(")?(?<info>\{.*?\})("\))?;/m).try &.["info"] || "{}").as_h
-  if initial_data.starts_with?("JSON.parse(\"")
-    return JSON.parse(JSON.parse(%({"initial_data":"#{initial_data[12..-3]}"}))["initial_data"].as_s).as_h
-  else
-    return JSON.parse(initial_data).as_h
-  end
 end
 
 def proxy_file(response, env)

src/invidious/videos.cr

@@ -839,8 +839,7 @@ def extract_polymer_config(body)
     params[f] = player_response[f] if player_response[f]?
   end
 
-  yt_initial_data = body.match(/(window\["ytInitialData"\]|var\s+ytInitialData)\s*=\s*(?<info>.*?);\s*\n/)
+  yt_initial_data = extract_initial_data(body)
-    .try { |r| JSON.parse(r["info"]).as_h }
 
   params["relatedVideos"] = yt_initial_data.try &.["playerOverlays"]?.try &.["playerOverlayRenderer"]?
     .try &.["endScreen"]?.try &.["watchNextEndScreenRenderer"]?.try &.["results"]?.try &.as_a.compact_map { |r|