Pull 'extract_videos' out into seperate function

This commit is contained in:
Omar Roth
2018-08-10 09:44:19 -05:00
parent 2f8716d97f
commit 15c26d022b
4 changed files with 157 additions and 249 deletions

View File

@@ -286,3 +286,91 @@ def html_to_description(description_html)
return description, description_html
end
def extract_videos(nodeset, ucid = nil)
# TODO: Make this a 'common', so it makes more sense to be used here
videos = [] of SearchVideo
nodeset.each do |node|
anchor = node.xpath_node(%q(.//h3[contains(@class,"yt-lockup-title")]/a))
if !anchor
next
end
if anchor["href"].starts_with? "https://www.googleadservices.com"
next
end
title = anchor.content.strip
id = anchor["href"].lchop("/watch?v=")
if ucid
author = ""
author_id = ""
else
anchor = node.xpath_node(%q(.//div[contains(@class, "yt-lockup-byline")]/a))
if !anchor
next
end
author = anchor.content
author_id = anchor["href"].split("/")[-1]
end
# Skip playlists
if node.xpath_node(%q(.//div[contains(@class, "yt-playlist-renderer")]))
next
end
# Skip movies
if node.xpath_node(%q(.//div[contains(@class, "yt-lockup-movie-top-content")]))
next
end
metadata = node.xpath_nodes(%q(.//div[contains(@class,"yt-lockup-meta")]/ul/li))
if metadata.size == 0
next
elsif metadata.size == 1
if metadata[0].content.starts_with? "Starts"
view_count = 0_i64
published = Time.epoch(metadata[0].xpath_node(%q(.//span)).not_nil!["data-timestamp"].to_i64)
else
view_count = metadata[0].content.lchop("Streamed ").split(" ")[0].delete(",").to_i64
published = Time.now
end
else
published = decode_date(metadata[0].content)
view_count = metadata[1].content.split(" ")[0]
if view_count == "No"
view_count = 0_i64
else
view_count = view_count.delete(",").to_i64
end
end
description_html = node.xpath_node(%q(.//div[contains(@class, "yt-lockup-description")]))
description, description_html = html_to_description(description_html)
length_seconds = node.xpath_node(%q(.//span[@class="video-time"]))
if length_seconds
length_seconds = decode_length_seconds(length_seconds.content)
else
length_seconds = -1
end
videos << SearchVideo.new(
title,
id,
author,
author_id,
published,
view_count,
description,
description_html,
length_seconds,
)
end
return videos
end