discourse/app/models/topic_link_click.rb

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

133 lines
3.9 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
2013-02-05 14:16:51 -05:00
require 'ipaddr'
require 'url_helper'
2013-02-05 14:16:51 -05:00
class TopicLinkClick < ActiveRecord::Base
belongs_to :topic_link, counter_cache: :clicks
belongs_to :user
validates_presence_of :topic_link_id
ALLOWED_REDIRECT_HOSTNAMES = Set.new(%W{www.youtube.com youtu.be})
include ActiveSupport::Deprecation::DeprecatedConstantAccessor
deprecate_constant 'WHITELISTED_REDIRECT_HOSTNAMES', 'TopicLinkClick::ALLOWED_REDIRECT_HOSTNAMES'
2013-02-05 14:16:51 -05:00
# Create a click from a URL and post_id
def self.create_from(args = {})
2015-09-25 14:07:04 -04:00
url = args[:url][0...TopicLink.max_url_length]
return nil if url.blank?
uri = UrlHelper.relaxed_parse(url)
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
2017-12-13 15:47:42 -05:00
while i = url.rindex('&', i - 1)
urls << url[0...i]
end
# add a cdn link
2016-06-28 15:52:38 -04:00
if uri
if Discourse.asset_host.present?
cdn_uri = begin
URI.parse(Discourse.asset_host)
rescue URI::Error
end
2016-06-28 15:52:38 -04:00
if cdn_uri && cdn_uri.hostname == uri.hostname && uri.path.starts_with?(cdn_uri.path)
is_cdn_link = true
2016-06-30 10:55:01 -04:00
urls << uri.path[cdn_uri.path.length..-1]
2016-06-28 15:52:38 -04:00
end
end
if SiteSetting.Upload.s3_cdn_url.present?
cdn_uri = begin
URI.parse(SiteSetting.Upload.s3_cdn_url)
rescue URI::Error
end
2016-06-28 15:52:38 -04:00
if cdn_uri && cdn_uri.hostname == uri.hostname && uri.path.starts_with?(cdn_uri.path)
is_cdn_link = true
2016-06-30 10:55:01 -04:00
path = uri.path[cdn_uri.path.length..-1]
2016-06-28 15:52:38 -04:00
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)
2013-02-05 14:16:51 -05:00
# 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...
2013-07-27 13:18:37 -04:00
unless link.present?
# ... return the url for relative links or when using the same host
2016-06-30 10:55:01 -04:00
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 allowlisted hostnames
return url if ALLOWED_REDIRECT_HOSTNAMES.include?(uri.hostname) || is_cdn_link
return nil
2013-07-27 13:18:37 -04:00
end
return url if args[:user_id] && link.user_id == args[:user_id]
2013-07-26 17:29:43 -04:00
2013-02-05 14:16:51 -05:00
# Rate limit the click counts to once in 24 hours
rate_key = "link-clicks:#{link.id}:#{args[:user_id] || args[:ip]}"
if Discourse.redis.setnx(rate_key, "1")
Discourse.redis.expire(rate_key, 1.day.to_i)
args[:ip] = nil if args[:user_id]
create!(topic_link_id: link.id, user_id: args[:user_id], ip_address: args[:ip])
2013-02-05 14:16:51 -05:00
end
url
2013-02-05 14:16:51 -05:00
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
#
# Indexes
#
2020-10-27 14:12:33 -04:00
# index_forum_thread_link_clicks_on_forum_thread_link_id (topic_link_id)
#