discourse/app/models/topic_link.rb

224 lines
7.4 KiB
Ruby
Raw Normal View History

2013-02-05 14:16:51 -05:00
require 'uri'
require_dependency 'slug'
class TopicLink < ActiveRecord::Base
belongs_to :topic
belongs_to :user
belongs_to :post
belongs_to :link_topic, class_name: 'Topic'
belongs_to :link_post, class_name: 'Post'
2013-02-05 14:16:51 -05:00
validates_presence_of :url
validates_length_of :url, maximum: 500
validates_uniqueness_of :url, scope: [:topic_id, :post_id]
has_many :topic_link_clicks, dependent: :destroy
2013-02-05 14:16:51 -05:00
validate :link_to_self
2014-04-05 14:47:25 -04:00
after_commit :crawl_link_title
2013-02-05 14:16:51 -05:00
# Make sure a topic can't link to itself
def link_to_self
errors.add(:base, "can't link to the same topic") if (topic_id == link_topic_id)
end
2013-11-15 12:15:46 -05:00
def self.topic_map(guardian, topic_id)
# Sam: complicated reports are really hard in AR
builder = SqlBuilder.new("SELECT ftl.url,
2014-04-05 14:47:25 -04:00
COALESCE(ft.title, ftl.title) AS title,
ftl.link_topic_id,
ftl.reflection,
ftl.internal,
2014-04-05 14:47:25 -04:00
ftl.domain,
MIN(ftl.user_id) AS user_id,
SUM(clicks) AS clicks
FROM topic_links AS ftl
LEFT JOIN topics AS ft ON ftl.link_topic_id = ft.id
LEFT JOIN categories AS c ON c.id = ft.category_id
/*where*/
2014-04-05 14:47:25 -04:00
GROUP BY ftl.url, ft.title, ftl.title, ftl.link_topic_id, ftl.reflection, ftl.internal, ftl.domain
ORDER BY clicks DESC")
builder.where('ftl.topic_id = :topic_id', topic_id: topic_id)
builder.where('ft.deleted_at IS NULL')
builder.secure_category(guardian.secure_category_ids)
builder.exec.to_a
end
def self.counts_for(guardian,topic, posts)
return {} if posts.blank?
# Sam: I don't know how to write this cleanly in AR,
# in particular the securing logic is tricky and would fallback to SQL anyway
builder = SqlBuilder.new("SELECT
l.post_id,
l.url,
l.clicks,
2014-04-05 14:47:25 -04:00
COALESCE(t.title, l.title) AS title,
l.internal,
2014-04-05 14:47:25 -04:00
l.reflection,
l.domain
FROM topic_links l
LEFT JOIN topics t ON t.id = l.link_topic_id
LEFT JOIN categories AS c ON c.id = t.category_id
/*where*/
ORDER BY reflection ASC, clicks DESC")
builder.where('t.deleted_at IS NULL')
# not certain if pluck is right, cause it may interfere with caching
builder.where('l.post_id IN (:post_ids)', post_ids: posts.map(&:id))
builder.secure_category(guardian.secure_category_ids)
builder.map_exec(OpenStruct).each_with_object({}) do |l,result|
result[l.post_id] ||= []
result[l.post_id] << {url: l.url,
clicks: l.clicks,
title: l.title,
internal: l.internal,
reflection: l.reflection}
end
end
2013-02-05 14:16:51 -05:00
# Extract any urls in body
def self.extract_from(post)
return unless post.present?
2013-02-07 10:45:24 -05:00
2014-04-05 14:47:25 -04:00
added_urls = []
2013-02-05 14:16:51 -05:00
TopicLink.transaction do
added_urls = []
reflected_urls = []
PrettyText
.extract_links(post.cooked)
.map{|u| [u, URI.parse(u)] rescue nil}
.reject{|u,p| p.nil?}
.uniq{|u,p| u}
2013-02-07 10:45:24 -05:00
.each do |url, parsed|
2013-02-05 14:16:51 -05:00
begin
internal = false
topic_id = nil
post_number = nil
if Discourse.store.has_been_uploaded?(url)
internal = Discourse.store.internal?
elsif parsed.host == Discourse.current_hostname || !parsed.host
2013-02-05 14:16:51 -05:00
internal = true
2013-02-07 10:45:24 -05:00
2013-02-05 14:16:51 -05:00
route = Rails.application.routes.recognize_path(parsed.path)
# We aren't interested in tracking internal links to users
next if route[:controller] == 'users'
2013-02-05 14:16:51 -05:00
topic_id = route[:topic_id]
post_number = route[:post_number] || 1
# Store the canonical URL
topic = Topic.where(id: topic_id).first
if topic.present?
url = "#{Discourse.base_url}#{topic.relative_url}"
url << "/#{post_number}" if post_number.to_i > 1
end
2013-02-05 14:16:51 -05:00
end
# Skip linking to ourselves
next if topic_id == post.topic_id
reflected_post = nil
if post_number && topic_id
reflected_post = Post.where(topic_id: topic_id, post_number: post_number.to_i).first
end
2013-02-05 14:16:51 -05:00
added_urls << url
2013-02-07 10:45:24 -05:00
TopicLink.create(post_id: post.id,
user_id: post.user_id,
topic_id: post.topic_id,
url: url,
domain: parsed.host || Discourse.current_hostname,
internal: internal,
link_topic_id: topic_id,
link_post_id: reflected_post.try(:id))
2013-02-05 14:16:51 -05:00
# Create the reflection if we can
if topic_id.present?
topic = Topic.where(id: topic_id).first
if topic && post.topic && post.topic.archetype != 'private_message' && topic.archetype != 'private_message'
2013-02-05 14:16:51 -05:00
prefix = Discourse.base_url
reflected_url = "#{prefix}#{post.topic.relative_url(post.post_number)}"
reflected_urls << reflected_url
TopicLink.create(user_id: post.user_id,
topic_id: topic_id,
post_id: reflected_post.try(:id),
url: reflected_url,
domain: Discourse.current_hostname,
reflection: true,
internal: true,
link_topic_id: post.topic_id,
link_post_id: post.id)
end
end
rescue URI::InvalidURIError
# if the URI is invalid, don't store it.
rescue ActionController::RoutingError
2013-02-07 10:45:24 -05:00
# If we can't find the route, no big deal
2013-02-05 14:16:51 -05:00
end
2013-02-07 10:45:24 -05:00
end
2013-02-05 14:16:51 -05:00
# Remove links that aren't there anymore
if added_urls.present?
TopicLink.delete_all ["(url not in (:urls)) AND (post_id = :post_id)", urls: added_urls, post_id: post.id]
TopicLink.delete_all ["(url not in (:urls)) AND (link_post_id = :post_id)", urls: reflected_urls, post_id: post.id]
else
TopicLink.delete_all ["post_id = :post_id OR link_post_id = :post_id", post_id: post.id]
end
end
end
2014-04-05 14:47:25 -04:00
# Crawl a link's title after it's saved
def crawl_link_title
Jobs.enqueue(:crawl_topic_link, topic_link_id: id)
end
2013-02-05 14:16:51 -05:00
end
# == Schema Information
#
# Table name: topic_links
#
# id :integer not null, primary key
# topic_id :integer not null
# post_id :integer
# user_id :integer not null
# url :string(500) not null
# domain :string(100) not null
# internal :boolean default(FALSE), not null
# link_topic_id :integer
# created_at :datetime not null
# updated_at :datetime not null
# reflection :boolean default(FALSE)
# clicks :integer default(0), not null
# link_post_id :integer
2014-04-08 11:35:44 -04:00
# title :string(255)
# crawled_at :datetime
#
# Indexes
#
# index_forum_thread_links_on_forum_thread_id (topic_id)
# index_forum_thread_links_on_forum_thread_id_and_post_id_and_url (topic_id,post_id,url) UNIQUE
#