mirror of
https://github.com/discourse/discourse.git
synced 2024-11-24 04:31:56 +08:00
30990006a9
This reduces chances of errors where consumers of strings mutate inputs and reduces memory usage of the app. Test suite passes now, but there may be some stuff left, so we will run a few sites on a branch prior to merging
61 lines
1.7 KiB
Ruby
61 lines
1.7 KiB
Ruby
# frozen_string_literal: true
|
|
|
|
require_dependency 'screening_model'
|
|
|
|
# A ScreenedUrl record represents a URL that is being watched.
|
|
# If the URL is found in a post, some action can be performed.
|
|
|
|
# For now, nothing is done. We're just collecting the data and will decide
|
|
# what to do with it later.
|
|
class ScreenedUrl < ActiveRecord::Base
|
|
|
|
include ScreeningModel
|
|
|
|
default_action :do_nothing
|
|
|
|
before_validation :normalize
|
|
|
|
validates :url, presence: true, uniqueness: true
|
|
validates :domain, presence: true
|
|
|
|
def normalize
|
|
self.url = ScreenedUrl.normalize_url(self.url) if self.url
|
|
self.domain = self.domain.downcase.sub(/^www\./, '') if self.domain
|
|
end
|
|
|
|
def self.watch(url, domain, opts = {})
|
|
find_match(url) || create(opts.slice(:action_type, :ip_address).merge(url: url, domain: domain))
|
|
end
|
|
|
|
def self.find_match(url)
|
|
find_by_url normalize_url(url)
|
|
end
|
|
|
|
def self.normalize_url(url)
|
|
normalized = url.gsub(/http(s?):\/\//i, '')
|
|
normalized.gsub!(/(\/)+$/, '') # trim trailing slashes
|
|
normalized.gsub!(/^([^\/]+)(?:\/)?/) { |m| m.downcase } # downcase the domain part of the url
|
|
normalized
|
|
end
|
|
end
|
|
|
|
# == Schema Information
|
|
#
|
|
# Table name: screened_urls
|
|
#
|
|
# id :integer not null, primary key
|
|
# url :string not null
|
|
# domain :string not null
|
|
# action_type :integer not null
|
|
# match_count :integer default(0), not null
|
|
# last_match_at :datetime
|
|
# created_at :datetime not null
|
|
# updated_at :datetime not null
|
|
# ip_address :inet
|
|
#
|
|
# Indexes
|
|
#
|
|
# index_screened_urls_on_last_match_at (last_match_at)
|
|
# index_screened_urls_on_url (url) UNIQUE
|
|
#
|