discourse/app/models/topic_link_click.rb
Guo Xiang Tan 142571bba0 Remove use of rescue nil.
* `rescue nil` is a really bad pattern to use in our code base.
  We should rescue errors that we expect the code to throw and
  not rescue everything because we're unsure of what errors the
  code would throw. This would reduce the amount of pain we face
  when debugging why something isn't working as expexted. I've
  been bitten countless of times by errors being swallowed as a
  result during debugging sessions.
2018-04-02 13:52:51 +08:00

134 lines
3.7 KiB
Ruby

require_dependency 'discourse'
require 'ipaddr'
require 'url_helper'
class TopicLinkClick < ActiveRecord::Base
belongs_to :topic_link, counter_cache: :clicks
belongs_to :user
validates_presence_of :topic_link_id
validates_presence_of :ip_address
WHITELISTED_REDIRECT_HOSTNAMES = Set.new(%W{www.youtube.com youtu.be})
# Create a click from a URL and post_id
def self.create_from(args = {})
url = args[:url][0...TopicLink.max_url_length]
return nil if url.blank?
uri = begin
URI.parse(url)
rescue URI::InvalidURIError
end
urls = Set.new
urls << url
if url =~ /^http/
urls << url.sub(/^https/, 'http')
urls << url.sub(/^http:/, 'https:')
urls << UrlHelper.schemaless(url)
end
urls << UrlHelper.absolute_without_cdn(url)
urls << uri.path if uri.try(:host) == Discourse.current_hostname
query = url.index('?')
unless query.nil?
endpos = url.index('#') || url.size
urls << url[0..query - 1] + url[endpos..-1]
end
# link can have query params, and analytics can add more to the end:
i = url.length
while i = url.rindex('&', i - 1)
urls << url[0...i]
end
# add a cdn link
if uri
if Discourse.asset_host.present?
cdn_uri = begin
URI.parse(Discourse.asset_host)
rescue URI::InvalidURIError
end
if cdn_uri && cdn_uri.hostname == uri.hostname && uri.path.starts_with?(cdn_uri.path)
is_cdn_link = true
urls << uri.path[cdn_uri.path.length..-1]
end
end
if SiteSetting.Upload.s3_cdn_url.present?
cdn_uri = begin
URI.parse(SiteSetting.Upload.s3_cdn_url)
rescue URI::InvalidURIError
end
if cdn_uri && cdn_uri.hostname == uri.hostname && uri.path.starts_with?(cdn_uri.path)
is_cdn_link = true
path = uri.path[cdn_uri.path.length..-1]
urls << path
urls << "#{Discourse.store.absolute_base_url}#{path}"
end
end
end
link = TopicLink.select([:id, :user_id])
# test for all possible URLs
link = link.where(Array.new(urls.count, "url = ?").join(" OR "), *urls)
# Find the forum topic link
link = link.where(post_id: args[:post_id]) if args[:post_id].present?
# If we don't have a post, just find the first occurance of the link
link = link.where(topic_id: args[:topic_id]) if args[:topic_id].present?
link = link.first
# If no link is found...
unless link.present?
# ... return the url for relative links or when using the same host
return url if url =~ /^\/[^\/]/ || uri.try(:host) == Discourse.current_hostname
# If we have it somewhere else on the site, just allow the redirect.
# This is likely due to a onebox of another topic.
link = TopicLink.find_by(url: url)
return link.url if link.present?
return nil unless uri
# Only redirect to whitelisted hostnames
return url if WHITELISTED_REDIRECT_HOSTNAMES.include?(uri.hostname) || is_cdn_link
return nil
end
return url if args[:user_id] && link.user_id == args[:user_id]
# Rate limit the click counts to once in 24 hours
rate_key = "link-clicks:#{link.id}:#{args[:user_id] || args[:ip]}"
if $redis.setnx(rate_key, "1")
$redis.expire(rate_key, 1.day.to_i)
create!(topic_link_id: link.id, user_id: args[:user_id], ip_address: args[:ip])
end
url
end
end
# == Schema Information
#
# Table name: topic_link_clicks
#
# id :integer not null, primary key
# topic_link_id :integer not null
# user_id :integer
# created_at :datetime not null
# updated_at :datetime not null
# ip_address :inet not null
#
# Indexes
#
# by_link (topic_link_id)
#