invidious/src/invidious.cr

274 lines
6.6 KiB
Crystal
Raw Normal View History

require "http/client"
require "json"
2017-11-23 07:48:55 +00:00
require "kemal"
require "pg"
2017-11-30 03:49:26 +00:00
require "time"
2018-01-16 20:02:35 +00:00
require "xml"
2017-11-29 21:33:46 +00:00
PG_DB = DB.open "postgres://kemal:kemal@localhost:5432/invidious"
2018-01-17 03:42:48 +00:00
URL = URI.parse("https://www.youtube.com")
CONTEXT = OpenSSL::SSL::Context::Client.new
CONTEXT.verify_mode = OpenSSL::SSL::VerifyMode::NONE
CONTEXT.add_options(
OpenSSL::SSL::Options::ALL |
OpenSSL::SSL::Options::NO_SSL_V2 |
OpenSSL::SSL::Options::NO_SSL_V3
)
POOL = Deque.new(30) do
HTTP::Client.new(URL, CONTEXT)
end
# Refresh all the connections in the pool by crawling recommended
spawn do
# Arbitrary start value
2018-01-19 03:46:29 +00:00
id = Deque.new(50,"0xjKNDMgE54")
while true
2018-01-17 03:42:48 +00:00
client = get_client
time = Time.now
begin
2018-01-19 03:46:29 +00:00
video = get_video(id[rand(id.size)], false)
rvs = [] of Hash(String, String)
video.info["rvs"].split(",").each do |rv|
rvs << HTTP::Params.parse(rv).to_h
end
rvs.each do |rv|
id << rv["id"]
end
puts "#{Time.now} 200 GET #{elapsed_text(Time.now - time)}"
2018-01-17 03:42:48 +00:00
rescue ex
next
2018-01-19 03:46:29 +00:00
ensure
POOL << client
2018-01-17 03:42:48 +00:00
end
end
2018-01-07 23:18:24 +00:00
end
2017-11-23 07:48:55 +00:00
macro templated(filename)
2017-11-29 21:33:46 +00:00
render "src/views/#{{{filename}}}.ecr", "src/views/layout.ecr"
2017-11-23 07:48:55 +00:00
end
class Video
module HTTPParamConverter
def self.from_rs(rs)
HTTP::Params.parse(rs.read(String))
end
end
module XMLConverter
def self.from_rs(rs)
2018-01-15 03:16:09 +00:00
XML.parse_html(rs.read(String))
end
2017-11-30 03:49:26 +00:00
end
def initialize(id, info, html, updated)
@id = id
@info = info
@html = html
@updated = updated
2017-11-30 03:49:26 +00:00
end
def to_a
return [@id, @info, @html, @updated]
end
DB.mapping({
id: String,
info: {
type: HTTP::Params,
default: HTTP::Params.parse(""),
converter: Video::HTTPParamConverter,
},
html: {
type: XML::Node,
2018-01-15 03:16:09 +00:00
default: XML.parse_html(""),
converter: Video::XMLConverter,
},
updated: Time,
})
2017-11-30 03:49:26 +00:00
end
# See http://www.evanmiller.org/how-not-to-sort-by-average-rating.html
def ci_lower_bound(pos, n)
if n == 0
return 0
end
2017-11-23 07:48:55 +00:00
# z value here represents a confidence level of 0.95
z = 1.96
phat = 1.0*pos/n
return (phat + z*z/(2*n) - z * Math.sqrt((phat*(1 - phat) + z*z/(4*n))/n))/(1 + z*z/n)
end
2018-01-17 03:42:48 +00:00
def elapsed_text(elapsed)
millis = elapsed.total_milliseconds
return "#{millis.round(2)}ms" if millis >= 1
"#{(millis * 1000).round(2)}µs"
end
2018-01-15 03:16:09 +00:00
def get_client
2018-01-07 23:18:24 +00:00
while POOL.empty?
sleep rand(0..10).milliseconds
end
2018-01-17 03:42:48 +00:00
return POOL.shift
2018-01-15 03:16:09 +00:00
end
def fetch_video(id)
# Grab connection from pool
client = get_client
2018-01-15 03:16:09 +00:00
info = client.get("/get_video_info?video_id=#{id}&el=detailpage&ps=default&eurl=&gl=US&hl=en").body
info = HTTP::Params.parse(info)
html = client.get("/watch?v=#{id}").body
2018-01-15 03:16:09 +00:00
html = XML.parse_html(html)
if info["reason"]?
raise info["reason"]
end
# Return connection to pool
POOL << client
2018-01-07 23:18:24 +00:00
video = Video.new(id, info, html, Time.now)
2018-01-07 23:18:24 +00:00
return video
end
2018-01-07 17:42:24 +00:00
def get_video(id, refresh = true)
if PG_DB.query_one?("SELECT EXISTS (SELECT true FROM videos WHERE id = $1)", id, as: Bool)
video = PG_DB.query_one("SELECT * FROM videos WHERE id = $1", id, as: Video)
# If record was last updated more than 5 hours ago, refresh (expire param in response lasts for 6 hours)
2018-01-07 17:42:24 +00:00
if refresh && Time.now - video.updated > Time::Span.new(0, 5, 0, 0)
video = fetch_video(id)
2018-01-07 17:42:24 +00:00
PG_DB.exec("UPDATE videos SET info = $2, html = $3, updated = $4 WHERE id = $1", video.to_a)
end
else
video = fetch_video(id)
PG_DB.exec("INSERT INTO videos VALUES ($1, $2, $3, $4)", video.to_a)
end
return video
end
get "/" do |env|
templated "index"
end
get "/watch" do |env|
id = env.params.query["v"]
2018-01-15 03:16:09 +00:00
listen = env.params.query["listen"]? || "false"
2018-01-16 02:30:57 +00:00
env.params.query.delete_all("listen")
begin
video = get_video(id)
rescue ex
error_message = ex.message
next templated "error"
end
2018-01-15 03:16:09 +00:00
player_response = JSON.parse(video.info["player_response"])
2018-01-04 02:06:16 +00:00
fmt_stream = [] of HTTP::Params
video.info["url_encoded_fmt_stream_map"].split(",") do |string|
fmt_stream << HTTP::Params.parse(string)
end
2017-11-23 07:48:55 +00:00
fmt_stream.reverse! # We want lowest quality first
2018-01-04 02:06:16 +00:00
adaptive_fmts = [] of HTTP::Params
video.info["adaptive_fmts"].split(",") do |string|
2018-01-04 02:06:16 +00:00
adaptive_fmts << HTTP::Params.parse(string)
end
likes = video.html.xpath_node(%q(//button[@title="I like this"]/span))
2018-01-15 03:16:09 +00:00
likes = likes ? likes.content.delete(",").to_i : 1
dislikes = video.html.xpath_node(%q(//button[@title="I dislike this"]/span))
2018-01-15 03:16:09 +00:00
dislikes = dislikes ? dislikes.content.delete(",").to_i : 1
description = video.html.xpath_node(%q(//p[@id="eow-description"]))
2018-01-15 03:16:09 +00:00
description = description ? description.to_xml : "Could not load description"
views = video.info["view_count"].to_i64
rating = video.info["avg_rating"].to_f64
2018-01-15 03:16:09 +00:00
engagement = ((dislikes.to_f + likes.to_f)/views * 100)
calculated_rating = (likes.to_f/(likes.to_f + dislikes.to_f) * 4 + 1)
2018-01-15 03:16:09 +00:00
rvs = [] of Hash(String, String)
video.info["rvs"].split(",").each do |rv|
rvs << HTTP::Params.parse(rv).to_h
end
2017-11-23 07:48:55 +00:00
templated "watch"
end
2017-12-30 21:21:43 +00:00
get "/search" do |env|
2018-01-07 17:42:24 +00:00
query = env.params.query["q"]
2018-01-15 03:16:09 +00:00
page = env.params.query["page"]? && env.params.query["page"].to_i? ? env.params.query["page"].to_i : 1
2018-01-07 17:42:24 +00:00
2018-01-15 03:16:09 +00:00
client = get_client
2018-01-07 23:18:24 +00:00
2018-01-16 04:11:51 +00:00
html = client.get("https://www.youtube.com/results?q=#{URI.escape(query)}&page=#{page}&sp=EgIQAVAU").body
2018-01-15 03:16:09 +00:00
html = XML.parse_html(html)
2018-01-15 03:16:09 +00:00
videos = Array(Hash(String, String)).new
2017-12-30 21:21:43 +00:00
2018-01-16 04:11:51 +00:00
html.xpath_nodes(%q(//ol[@class="item-section"]/li)).each do |item|
root = item.xpath_node(%q(div[contains(@class,"yt-lockup-video")]/div))
if root
video = {} of String => String
2017-12-30 21:21:43 +00:00
2018-01-16 04:11:51 +00:00
link = root.xpath_node(%q(div[contains(@class,"yt-lockup-thumbnail")]/a/@href))
2018-01-15 03:16:09 +00:00
if link
video["link"] = link.content
2018-01-16 04:11:51 +00:00
else
video["link"] = "#"
2018-01-15 03:16:09 +00:00
end
2017-12-30 21:21:43 +00:00
2018-01-16 04:11:51 +00:00
title = root.xpath_node(%q(div[@class="yt-lockup-content"]/h3/a))
2018-01-15 03:16:09 +00:00
if title
video["title"] = title.content
2018-01-16 04:11:51 +00:00
else
video["title"] = "Something went wrong"
2018-01-15 03:16:09 +00:00
end
2018-01-16 04:11:51 +00:00
thumbnail = root.xpath_node(%q(div[contains(@class,"yt-lockup-thumbnail")]/a/div/span/img/@src))
if thumbnail && !thumbnail.content.ends_with?(".gif")
video["thumbnail"] = thumbnail.content
else
thumbnail = root.xpath_node(%q(div[contains(@class,"yt-lockup-thumbnail")]/a/div/span/img/@data-thumb))
if thumbnail
video["thumbnail"] = thumbnail.content
else
video["thumbnail"] = "http://via.placeholder.com/246x138"
end
end
videos << video
end
end
POOL << client
2017-12-30 21:21:43 +00:00
templated "search"
end
error 404 do |env|
templated "index"
end
error 500 do |env|
templated "index"
end
2017-11-23 07:48:55 +00:00
public_folder "assets"
Kemal.run