Use fibers to try to speed up importing of channels

This commit is contained in:
Omar Roth 2019-01-02 19:28:01 -06:00
parent 7951d4c8aa
commit c2c224b16f
3 changed files with 39 additions and 42 deletions

View file

@ -21,6 +21,33 @@ class ChannelVideo
})
end
def get_batch_channels(channels, db, refresh = false, pull_all_videos = true, max_threads = 10)
active_threads = 0
active_channel = Channel(String | Nil).new
final = [] of String
channels.map do |ucid|
if active_threads >= max_threads
if response = active_channel.receive
active_threads -= 1
final << response
end
end
active_threads += 1
spawn do
begin
get_channel(ucid, db, refresh, pull_all_videos)
active_channel.send(ucid)
rescue ex
active_channel.send(nil)
end
end
end
return final
end
def get_channel(id, db, refresh = true, pull_all_videos = true)
client = make_client(YT_URL)

View file

@ -177,19 +177,16 @@ def fetch_user(sid, headers, db)
feed = XML.parse_html(feed.body)
channels = [] of String
feed.xpath_nodes(%q(//ul[@id="guide-channels"]/li/a)).each do |channel|
if !{"Popular on YouTube", "Music", "Sports", "Gaming"}.includes? channel["title"]
channel_id = channel["href"].lstrip("/channel/")
begin
channel = get_channel(channel_id, db, false, false)
channels << channel.id
rescue ex
next
end
channels = feed.xpath_nodes(%q(//ul[@id="guide-channels"]/li/a)).compact_map do |channel|
if {"Popular on YouTube", "Music", "Sports", "Gaming"}.includes? channel["title"]
nil
else
channel["href"].lstrip("/channel/")
end
end
channels = get_batch_channels(channels, db, false, false)
email = feed.xpath_node(%q(//a[@class="yt-masthead-picker-header yt-masthead-picker-active-account"]))
if email
email = email.content.strip